Professional Documents
Culture Documents
Purushotham Tukkaraja (Editor) - Underground Ventilation - Proceedings of The 19th North American Mine Ventilation Symp
Purushotham Tukkaraja (Editor) - Underground Ventilation - Proceedings of The 19th North American Mine Ventilation Symp
Ventilation
EDITED BY Purushotham Tukkaraja
UNDERGROUND VENTILATION
Underground Ventilation contains the proceedings of the 19th North American Mine Ventilation
Symposium held at the South Dakota School of Mines & Technology (South Dakota Mines) in Rapid
City, South Dakota, June 17-22, 2023. South Dakota Mines organized this symposium in collaboration
with the Underground Ventilation Committee (UVC) of the Society for Mining, Metallurgy & Exploration
(SME).
The Mine Ventilation Symposium series has always been a premier forum for ventilation experts,
practitioners, educators, students, regulators, and suppliers from around the world to exchange
knowledge, ideas, and opinions. Underground Ventilation features sixty-seven selected technical
papers in a wide range of ventilation topics including: auxiliary and primary systems, mine fans,
case studies, computational fluid dynamics applications, diesel particulate control, electric machinery,
mine cooling and refrigeration, mine dust monitoring and control, mine fires and explosion prevention,
mine gases, mine heat, mine ventilation and automation, occupational health and safety, renewable/
alternative energy, monitoring and measurement, network analysis and optimization, and planning and
design.
PROCEEDINGS OF THE 19TH NORTH AMERICAN MINE VENTILATION SYMPOSIUM
(NAMVS 2023), 17-22 JUNE 2023, RAPID CITY, SOUTH DAKOTA, USA
Underground Ventilation
Edited by
Purushotham Tukkaraja, Ph.D., QP
Mining Engineering & Management, South Dakota Mines, Rapid City, SD, USA
Front Cover Image: © Zitrón - Ventilation solutions for underground mines
CRC Press/Balkema is an imprint of the Taylor & Francis Group, an informa business
© 2023 selection and editorial matter, Purushotham Tukkaraja; individual chapters, the
contributors
The right of Purushotham Tukkaraja to be identified as the author of the editorial material,
and of the authors for their individual chapters, has been asserted in accordance with sections
77 and 78 of the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form
or by any electronic, mechanical, or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or retrieval system, without
permission in writing from the publishers.
Although all care is taken to ensure integrity and the quality of this publication and the
information herein, no responsibility is assumed by the publishers nor the author for any
damage to the property or persons as a result of operation or use of this publication and/or
the information contained herein.
Table of contents
Preface xi
Organizing committees xiii
Underground Ventilation Committee (UVC) xiii
Review committee xiii
Sponsors and Exhibitors xv
Auxiliary ventilation
An investigation of booster fan placements in a large opening underground stone mine
utilizing CFD 3
N. Gendrue, S. Liu & S. Bhattacharya
Quantifying assemblage losses in auxiliary ventilation systems 15
E. De Souza
v
Diesel particulate control
Comparing diesel and GDiesel® exhaust exposures in an underground mining laboratory 95
R.J. Reed, J.L. Burgess & E.A. Lutz
Improvement of size-selective sampling of diesel aerosols in underground mines 104
A.D. Bugarski, T.L. Barone, J.A. Hummer, T. Lee, S. Vanderslice & S. Friend
Results of diesel exhaust nanoparticle experimental sampling in a cabin of LHD loader
operating in an active ore heading area 115
S. Sabanov, N. Magauiya, A. Zeinulla, A. Abil, A. Qureshi, M. Torkmahalleh,
G. Nurshaiykova & D. Rakhimov
Importance of using real-time and microscopic analysis techniques to characterize DPM in
underground mines 120
A.A. Habibi, K.O. Homan & A.D. Bugarski
Development and evaluation of innovative diesel particulate filter technology 129
J. Stachulak, B. Rubeli, D. Young, K. Watson & B. McLean
DPM reduction through emission assisted maintenance - PT Freeport Indonesia program
update 135
A.A. Habibi, E. Pinto, M. Mardon, K. Wijayanto & C. Rose
vi
Effects of vertical air-blocking ring of drill shroud on dust control for surface mine drilling
operation using CFD 239
Y. Zheng, J.D. Potts & W.R. Reed
Investigation on the effect of water pressure on spray performance for removal of respirable dust 247
H. Jiang, S. Klima, T. Beck & Y. Zheng
Real-time measurements of respirable crystalline silica, kaolinite, coal, and calcite 256
W. Arnott, C. Kocsis, X. Wang, B. Osho, P. Nascimento, S. Taylor, B. Bingham, C. Murphy &
M. Sandink
Comparing respirable dust characteristics from full-scale cutting tests of three rock samples
with conical picks at three stages of wear 264
S. Slouka, E. Sidrow, C. Tsai & J. Brune
Accuracy of low-cost particulate matter sensor in measuring coal mine dust- a wind tunnel
evaluation 274
M.M. Zaid, G. Xu & N.A. Amoah
Parametric studies to maximize the dust protection performance of the two-level manifold
canopy air curtain and computational fluid dynamics modeling 285
N.A. Amoah, G. Xu & A.R. Kumar
Respirable coal mine dust research: Characterization and toxicity analysis based on dust
sources 296
V.P. Salinas, M.C. Das, G. Rubasinghege, P. Roghanchi & K. Zychowski
Comparison of respirable coal and silica dust monitoring systems for underground mining
applications 305
A. Medina, A. Vanegas, E. Madureira, P. Roghanchi, R. Rajapaksha, L. Uecker, T. Rawson &
C. Harb
Evaluation of different surfactants’ performance in varying coal dust concentration through
logistic regression analysis 313
Z. Zhao, A. Ghosh, P. Chang & Y. Liu
Process ventilation solutions for mitigation of combustible and non-combustible dust
hazards at mining operations 322
J. Finn
Development of VR-CFD-based training tool for dust control in gateroad development 330
M. Qiao, T. Ren, J. Roberts, J. Hines, C. Chow & A. Clayton
Mine fans
Stall impact on axial fans and testing of anti-stall rings 341
J. Fernandez
Practical values for the evaluation of fan system efficiencies 353
J. Bowling, G. Schult & J. Van Diest
vii
Evaluation of different suppression techniques for lithium-ion battery fires 384
L. Yuan, W. Tang, R.A. Thomas & J. Soles
Characterization and preliminary assessment of diesel fire prior to setting up large size
battery fire experiment 393
R.I. Pushparaj, G. Xu, A. Iqbal & O.B. Salami
Fire-induced temperature attenuation under the influence of a single ceiling smoke extraction
point in a bifurcated drift 399
O.B. Salami, G. Xu, A.R. Kumar, R.I. Pushparaj & A. Iqbal
Application of pressure balancing techniques at the West Elk coal mine 411
C. Kocsis, F. Calizaya, J. Johnson, T. Dias, N. Nunes, E. Lindgren, G. Atchley & J. Poulos
Mine gases
Underground coal methane gas forecasting using multivariate time series with one and two
auxiliary variables 423
J.C. Diaz, Z. Agioutantis, S. Schafrik & D.T. Hristopulos
Complete degasification of longwall panels in U.S. coal mines 431
P.C. Thakur
Study of in-situ coal seam gas content for Australian coal and gas outburst management:
A field data analysis and laboratory experiment 439
Z.B. Li, T. Ren, M. Qiao, D. Black & J. Juric
Airflow patterns and blast fume dispersion in different mining methods 451
S. Jayaraman Sridharan, A. Adhikari, P. Tukkaraja & J. Connot
The oxiperator for Ventilation Air Methane (VAM) 462
E. Prabhu, M. Prabhu & J.E. Fox
Mine heat
Findings and learnings from thermal parameter studies at four LKAB sites 469
F.K.R. Klose, A.L. Martikainen & T.H. Jones
Scenario-driven evaluation of heat sources in underground production scheduling 484
J.A. Buaba, E. Udofia, A. Newman & A.J. Brickey
viii
Occupational health and safety in mine ventilation
Ventilation research findings for enhanced worker safety when mining near unconventional
gas wells in longwall abutment pillars 529
S.J. Schatzel, K.M. Ajayi, Z. Khademian, R. Kimutis, M.L. Harris, M. Van Dyke, J.D. Addis,
H. Dougherty & E. Watkins
Optimization for fire evacuation applying maximum flow cost algorithm to improve the
time-response in underground coal mines 541
S. Lotero, H. Khaniani, V. Androulakis, M. Hassanalian, S. Shao & P. Roghanchi
Heat mitigation for underground coal mine refuge alternatives 550
D.S. Yantek & L. Yan
Occupancy derating for underground coal mine refuge alternatives 559
D.S. Yantek & L. Yan
ix
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Editor(s), ISBN 978-1-032-55146-3
Preface
This volume contains the proceedings of the 19th North American Mine Ventilation Sympo
sium held at the South Dakota School of Mines & Technology (South Dakota Mines) in
Rapid City, South Dakota, June 17-22, 2023. South Dakota Mines organized this symposium
in collaboration with the Underground Ventilation Committee (UVC) of the Society for
Mining, Metallurgy & Exploration (SME).
The North American Mine Ventilation Symposium series was initiated by the UVC in 1982. The
UVC is a Joint Technical Committee of the Coal and Energy and the Mining and Exploration
Divisions of the SME. The purpose of the UVC is to promote engineering interest and techno
logical progress in the ventilation of mines, tunnels, and other subsurface openings. The UVC
accomplishes its purpose by conducting technical sessions at SME-AIME meetings, sponsoring
the North American Mine Ventilation Symposium with host universities and other organizations,
and soliciting papers for publication in Mining, Metallurgy & Exploration Journal and SME con
ference proceedings. The UVC offers an affiliation home for SME members, and others engaged
in the practice of underground ventilation. In these ways, the UVC seeks to encourage research,
education, publications, and technology transfer in the field of underground ventilation.
The North American Mine Ventilation Symposium, held every two to three years since 1982,
provides a forum for practitioners, educators, and researchers to exchange the latest informa
tion on the ventilation of mines, tunnels, and other underground facilities.
xi
2004 – University of Alaska Fairbanks – Sukumar Bandopadhyay and Rajive Ganguli
With the help of the organizing committee, a solid 3-day program was assembled, with tech
nical papers, panel discussions, and keynote presentations organized in 21 sessions. A total
of 82 abstracts and 67 final papers were received. Session ventilation themes include case
studies, computational fluid dynamics applications, diesel particulate control, electric
machinery, mine cooling and refrigeration, mine dust monitoring and control, primary and
auxiliary systems, mine fans, mine fires and explosion prevention, mine gases, mine heat,
ventilation management, automation, occupational health and safety, renewable/alternative
energy, monitoring and measurement, network analysis and optimization, and planning and
design.
I would like to thank the UVC and Review Committee members for their help with peer-
reviewing papers, chairing technical sessions, and advice to make this symposium a success.
Finally, I would like to thank the Center for Alumni Relations and Advancement (CARA)
and the Office of Marketing and Communications at South Dakota Mines for assisting with
the symposium registration, advertisement, and website services.
xii
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Editor(s), ISBN 978-1-032-55146-3
Organizing committees
Review committee
xiii
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Editor(s), ISBN 978-1-032-55146-3
Sponsors
Platinum
Freeport-McMoRan
G+ Plastics
Hatch
Howden
Minetek
SRK Consulting, Inc.
Gold
Accutron Instruments
Turnstone – ABC VS & JENNMAR
Silver
BBE Consulting
Maestro Digital Mine
Mecanicad
Spendrup Fan Co.
Exhibitors
Accutron Instruments
BBE Consulting
CDC Dust Control & Air Blow Fans
CFT Compact Filter Technic
G+Plastics
Howden
Hyperflo
Maestro Digital Mine
Mecanicad
Minetek
Pinssar
Quick Supply Co
Spendrup Fan Co
SRK Consulting
TLT-Turbo
Turnstone – ABC VS & JENNMAR
Zitron
xv
Auxiliary ventilation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: The optimization and planning of the mine ventilation system is a key compo
nent of mine operation given that ventilation related costs can range between 20% and 50% of
the total operating cost of the mine. In large opening mines (LOMs) utilizing perimeter venti
lation schemes minimal guidance is available for determining optimal booster fan (BF) place
ments. A ventilation survey was conducted and published previously by the author which
established a CFD model for a section of an underground room and pillar large opening lime
stone mine. In this work the previously created CFD model was utilized to investigate a total
of 15 BF positions with a focus on recirculation patterns, overall airflow within the BF entry,
and airflow around face area. It was found that the maximum airflow around the face areas
can be achieved with fan positioned on the same side of the entry as face area; with the max
imum airflow through the BF entry being achieved when the fan is placed in the center of the
entry on the upstream side of the pillar line. The recirculation percentages were similar in all
cases reaching a maximum between 35%-40% of the total air movement. However, the highest
recirculation percentages also facilitated, via air entrainment, the highest airflow magnitudes
through the BF entry. The booster fan’s ability to stimulate airflow through adjacent entries
was found to be reduced by approximately 30% for each adjacent entry. Therefore, the recom
mendation was given to position the BF within 3 entries of the face to achieve adequate
airflow.
1 INTRODUCTION
Mine ventilation is an essential part of any underground mining system and accounts for
between 20% and 50% of the total operating cost of the mine (Carter, 2018, Leonida, 2019,
Babu et al., 2015). In large opening mines (LOMs, mines with cross-sections >~1000 ft2), par
ticularly in stone mines utilizing perimeter ventilation schemes, the placement of booster fans
is typically empirically determined by rule of thumb or trial and error until satisfactory condi
tions are achieved. Many LOMs do not utilize typical auxiliary ventilation systems with an
auxiliary fan and bag. Therefore, proper BF placement is imperative considering the lack of
other ventilation controls.
In place of typical auxiliary systems many LOMs utilize large diameter fans that produce
high-volume with low-pressure; and commonly referred to as box fans or propellor fans. Com
pared to jet fans these fans provide the increased airflow needed to maintain regulatory com
pliance; however, the low-pressure tradeoff creates other concerns over the increased role of
natural ventilation plays in the system due to the low system pressure, ventilation efficiency,
recirculation, and short circuiting of mining sections. Specifically for large opening stone
mines there is little quantitative guidance on how to combat these issues. This work will inves
tigate the effect of BF placement within the entry and its influence on recirculation and face
ventilation efficiencies.
DOI: 10.1201/9781003429241-1
3
In Gendrue et al. (Gendrue et al., 2023), a ventilation survey was conducted and used to
create a computational fluid dynamics (CFD) model of a section of a large opening stone
mine. The simulations resulted in a discussion of the mines selected BF placement and its
effectiveness as well as discussions on measurement locations for ventilation surveys. In this
work, the same model will be utilized to simulate different BF placements and discussions on
the BF effectiveness around the face area and airflow directions and quantities will be given.
This work aims to provide a foundation for future CFD BF placement studies on perimeter
ventilation systems and discuss important considerations when selecting a position for a BF.
2 BACKGROUND
Previous mine ventilation studies related to face ventilation focus on auxiliary ventilation of
drifts or dead-end headings with an emphasis on contaminant distributions and removal. Vari
ables typically considered when evaluating face ventilation are fan selection, curtain or auxiliary
tubing placement, airflow penetration depth, entrainment ratios, recirculation, and jet fan
angles being common among the studies listed below. However, large opening stone mine venti
lation schemes are unique and must be evaluated separately. Studies specifically in LOM envir
onments are limited however, the same quantitative variables are used to evaluate the systems.
Dunn et al. (Dunn et al., 1983) studied how BF inclination, fan elevation and ultimately air
flow around open entries and the face area in large and medium sized airways. However, their
results and discussions were derived from field data which inherently limits the analysis due to
the smaller number of sampling locations when compared to numerical simulations. Nonethe
less, their recommendations for optimizing face ventilation were to: (i) utilize a pushing system
rather than a pulling system, however, no pulling systems were investigated, (ii) place fans to
direct flow diagonally across the face, (iii) place fans one pillar inby the general flow to minimize
recirculation. They also noted that larger capacity fans ventilate more effectively and inclination
and elevations of the fans have little effect on face ventilation. Krog and Grau (Krog and Grau,
2006) discussed the use of vane axial or propellor fans in terms of their recirculation, overall
airflow, and entrainment ability for use in LOMs. Ultimately giving recommendations that pro
pellor fans be placed in the fresh air stream and work best for regional ventilation applications
while vane axial fans be placed 1 entry behind the airstream and work best for dead end ventila
tion i.e., headings. They also tested pushing and pulling systems for a new mine utilizing
a textbook example of a split mine ventilation scheme concluding that a BF positioned just
outby a portal is effective for mine airflow. Later, Grau and Krog (Grau III and Krog, 2009)
discussed BF placement and mine design criteria such as long pillars in terms of ventilation effi
ciency for LOMs. Recommendations were given to place the booster fans outby the last open
crosscut. Furthermore, it was noted that fans positioned in the middle of the entry will increase
recirculation and that “not all recirculation is detrimental.”
These three studies, however, do not cover all the common ventilation schemes used today in
LOMs. The study by Krog and Grau (Krog and Grau, 2006) studied only one adjacent entry
when three to five entries are common in perimeter schemes. They also investigated a split mine
ventilation scheme for a new mine. Grau and Krog (Grau and Krog, 2009) utilized a 3-entry
heading which operated under a pseudo split ventilation scheme with fresh air on one side and
return air on the other. The BF as directed perpendicular to the face opposite what is done in
perimeter ventilation where the fan is directed parallel with the face. Furthermore, the study by
Dunn et al. (Dunn et al., 1983) may initially seem like a perimeter ventilation scheme but all the
tested fan positions were at a 90° angle from the general airflow direction which is more analo
gous to a unit ventilation scheme. These studies also utilized smaller fans than are typically seen
today with the largest studies being an 8 ft. in Krog and Grau, 2006. Therefore, these studies are
fundamentally different not only in terms of fan technology but also in terms of their fundamen
tal ventilation system design as defined by Krog et al. (Krog et al., 2004) as split, unit, and per
imeter schemes.
BF placement and ventilation system designs have been noted in many studies as the most
or one of the most important factors in LOM ventilation systems, however, there are minimal
4
studies that discuss optimal BF placement (Thimons et al., 1985, Dunn et al., 1983, Grau and
Krog, 2009, Grau III et al., 2006). Thus, BF placement must be reexamined with current fan
capacities and in a perimeter ventilation context with the fan directing airflow parallel to the
face area rather than at a 90° angle to it.
3 METHOD
Figure 1. Plan view of the partner limestone mine with curtains, stoppings, and booster fans highlighted
in black, with the selected CFD simulation region highlighted in yellow.
5
Table 1. Measured airflow velocities and fan settings observed during the ventilation survey that were
used for the boundary conditions of the CFD simulation, measurement locations and names are shown
in Figure 2B.
6
some level of recirculation. The air flowing South-North (1N-8N) will be referenced as ‘entry’
airflow while the air flowing West-East (1W-12W) will be referenced as ‘crosscut’ airflow. Air
flow that is opposite the BF direction (flowing from north to south) in the entries and from
east to west in the crosscuts is labeled as negative airflow and is used to calculate
a recirculation ratio. The recirculation ratio is defined as the sum of negative airflow across
a pillar line divided by the total airflow across the pillar line.
Figure 2. A) Booster fan placements inside the model geometry, 15 different booster fan locations were
simulated with 9 arranged in a grid pattern between the same pillar marked original fan location, another
3 locations are 2 pillars inby the original location and the final group of 3 locations being 2 pillars outby
the original location B) Location of airflow measurements from the simulation files, entries parallel to the
booster fan direction are labeled 1N-8N while the crosscut entries are labeled 1W-12W with the original
booster fan placement shown as the blue fan symbol.
7
airflows and the traditional entrainment pattern at the first crosscut. Furthermore, the airflow
entering the model through inlets 1-8 in Figure 4A is mainly pulled though entry 8N while the
inlet airflow in Figure 4B is more uniformly distributed throughout all entries. This shows the
importance of BF placement within the context of pillar geometry. The pillar geometry was
created from the mine map, and when pillars are offset, or not perfectly aligned due to blast
ing, then BF airflow and effectiveness can be minimized due to the airstream disruption as
described above. In typical mine ventilation surveys that an operator would perform daily it is
common to measure airflow downstream of the BF and in an adjacent entry when time per
mits. With these minimal measurement locations, the region cannot be fully quantified. In
Figure 4A and 4B the red circled areas on the left upstream side of the region have minimal
airflow of under 0.5 m/s and are insignificantly affected by all fan positions in scenario one.
Figure 3. Airflow through entries 1N-8N for scenario one, fan locations are marked 1 through 9 in
a counterclockwise direction, positive airflow indicates flow in the booster fan direction (downstream)
and negative airflow indicates flow opposite the booster fan direction (upstream).
The right fan positions in the entry (locations 2, 3, and 9) show the lowest total air move
ment across the pillar lines and through entry 6N in Figure 3; these locations also have the
highest downstream positive airflow through entries 7N and 8N which is likely hindering air
entrainment as noted in Krog and Grau (Krog and Grau, 2006) that a more uniform distribu
tion of airflow will not induce entrainment. Since this uniform airflow does not occur when
the BF is on the right side of the entry it is likely due to the fan positioned on the same side of
the entry as the adjacent curtain line. Indicating that a fan positioned on the same side of the
entry as a curtain/stopping line in adjacent entries will minimize BF airflow due to the loss of
the booster fans entrainment abilities.
Table 2 shows the recirculation percentage across the pillar lines 1W-12W as defined in
Figure 2B in section 4.1. Recirculation percentages above 30% are highlighted in orange while
percentages below 10% are highlighted in blue. When no recirculation occurs, the cell is
marked with a dash (-). With the BF positioned in the center of the entry (positions 1, 4,
and 8) there are similar airflow magnitudes in Figure 3 as to when the BF is positioned on the
8
left side of the entry but with less recirculation as seen in Table 2. An increase in airflow and
recirculation can be seen in the first two pillars upstream from the BF when the fan is posi
tioned on the left-side of the entry (positions 5, 6, and 7) which is not seen with the center
positioning (Org 1). The left-side locations (Org 5, 6, and 7) also account for the highest recir
culation percentages in Table 2. With the fan positioned on the left side the cone effect of the
BF is maximized allowing for the airflow to be distributed across the entry without pillar/rib
interaction; however, with these fan positions the recirculation around the left adjacent pillars
are the highest compared to all other positions in Figure 3. Due to the large airflow (velocity)
gradient between the BF entry and the adjacent entries in the left positions it is unsurprising
these locations also have the highest airflow due to recirculation and entrainment properties.
Interestingly, in Figure 3 the position 8 graph is more similar to the left-side graphs in terms
of total airflow, recirculation amounts, and face airflow. Indicating that more total air move
ment can be achieved when the BF is positioned in either position 8, in the middle of the entry
in line with the upstream side of the pillar edge, or in the left positions, on the same side of the
entry as the open/face area.
Figure 5 shows the recirculation percentage through the entries as in Section 4.1. When the
BF is positioned in the center of the entry (positions 1, 4, and 8) there are similar airflow mag
nitudes in Figure 3 as to when the BF is positioned on the left side of the entry but with less
recirculation as seen in Figure 5. An increase in airflow and recirculation can be seen in the
first two pillars upstream from the BF when the fan is positioned on the left-side of the entry
(positions 5, 6, and 7) which is not seen with the center positioning. The left-side locations
also account for the highest recirculation percentages for all pillar lines (1W-12W) in Figure 5.
With the fan positioned on the left side the cone effect of the BF is maximized allowing for the
airflow to be distributed across the entry without pillar/rib interaction; however, with these
fan positions the recirculation around the left adjacent pillars is the highest compared to all
other positions in Figure 3. Due to the large airflow (velocity) gradient between the BF entry
and the adjacent entries in the left positions it is unsurprising these locations also have the
highest airflow due to recirculation and entrainment properties. Interestingly, in Figure 3 the
position 8 graph is more similar to the left-side graphs in terms of total airflow, recirculation
amounts, and face airflow. Indicating that more total air movement can be achieved when the
BF is positioned in either position 8, in the middle of the entry in line with the upstream side
of the pillar edge, or in the left positions, on the same side of the entry as the open/face area.
Figure 5 shows the airflow through the crosscuts labeled1W-12W with each bar representing
a north to south pillar lines labeled as right or left of the BF entry. Again, positive airflow in
Figure 5 represents air moving from west to east in the geometry in Figure 2B. The total bar
height represents the magnitude of airflow through the crosscuts or the exchange between the
9
Figure 4. Velocity contour of positions 6 and 8 of scenario one near the breathing plane (Z=2m), A)
location 6, B) location 8.
entries. The BF location in scenario one is between crosscuts 6W and 8W. Given the large
amounts of air entrainment seen in positions 1, 4, 6, 7, 8, in Figure 3 it is expected that the air
flow in the first and second crosscuts (7W and 8W) are high in magnitude and positive on the left
and negative on the right side of the BF entry indicating air flowing inwards towards the BF
entry i.e., entrainment. Furthermore, the maximum airflow at these locations in Figure 3 is
reached a maximum either one pillar (80 ft.) or two pillars (160 ft.) downstream the BF which is
consistent with the finding of Krog and Grau (Krog and Grau, 2006) as described by Figure 6 in
their work that maximum airflow for propeller fans being reached ~ 52 m (170 ft.) downstream.
Moving to ventilation effectiveness across the face area the ability to dilute contaminates
generated near the face is also related to crosscut flow due to the geometry, the only way to
transport containments out of the mine is towards the general airstream in the BF entry which
must be achieved through crosscut airflow. Therefore, it is important to maximize the crosscut
airflow near the face (5 and 4 pillars to the left of the BF entry). Locations 5, 6, and 8 in
Figure 5 shows similarly high values for airflow 5 entries to the left of the BF entry (between
entries 1N and 2N) compared to all other fan positions. This may be due to the higher overall
airflow generated by the BF across the region however, location 7 also has high absolute air
flow of ~804 m3/s which does not translate to higher face airflow. Interestingly at position 7 in
Figure 5, marked by the red box, the first left adjacent entry (1 left) has the highest positive
and overall airflow through it which does not translate to high face airflow. Thus, leading to
the conclusion that not only is high regional airflow necessary to ventilate further adjacent
entries but some amount of airstream disruption via air diversion from pillar corners is
needed to maximize crosscut exchange between entries.
To quantify the influence of the BF to adjacent entries the percent reduction in airflow
exchange (crosscut airflow) between the BF entry and adjacent entries can be calculated.
When face airflow is higher, as seen in positions 5, 6, and 8 a reduction in crosscut airflow can
be seen starting from 1 pillar to the left of the BF entry until approximately 4 pillars to the left
as shown by the red arrows in position 5 on Figure 5. When averaged across these positions
a total decrease in airflow of 91% on the positive side and 84% on the negative side is seen.
Indicating that crosscut airflow is reduced by approximately 46% to 55% for every consecutive
adjacent entry from the BF entry. Therefore, the BF effectiveness to facilitate airflow past
four adjacent entries is seen to be less than 15% of the BF entry and may not be sufficient to
be relied upon for mine ventilation planning.
4.3 Scenario two: Three pillars inby and outby the original fan location
Six more simulations were conducted three pillars inby and three pillars outby the original BF
location. Booster fan positions 1, 2, and 6 were selected for these inby and outby locations.
The calculated airflow values were obtained in the same way as the previous section.
10
Figure 5. Airflow through the crosscuts (1W-12W) from scenario one, fan locations marked 1 through 9
in a counterclockwise direction, positive airflow indicates flow from west to east in the crosscuts (1W to
12W) while negative airflow values indicate airflow in the opposite direction.
Figure 6 shows the calculated airflow values through entries 1N through 8N with the ori
ginal inby and outby BF locations marked on the y-axis. Inby location 1 shows the highest
airflow through entry 6N with a value of ~ 575 m3/s with inby position 6 showing the lowest
airflow through entry 6N of ~330 m3/s. Inby location 1 showed significant negative airflow
values in entries 3N to 5N throughout the middle portion of the geometry, which facilitates
the high airflow in entry 6N through air entrainment as discussed in Section 4.2. The outby
fan locations in Figure 6 shows minimal airflow through the majority of the geometry due
to the BF being in the outby position. However, similar maximum airflow values are still
reached downstream from the BF. All three of the outby fan positions did not have negative
airflow until 3 pillars upstream from the BF location which allows the majority of the sec
tion to have positive airflow. This leads to all three outby simulations (similar to original
location 2) having positive airflow throughout the face area (entries 1N-3N). However, the
magnitude of the positive airflow seen around the face area was not significantly larger than
any of the other simulations that saw negative airflow. Indicating that if avoiding negative
airflow is a necessity then a more outby BF location would be desired with a preference
towards a fan placement in the middle of the entry or on the same side as the adjacent cur
tain/stopping lines.
The recirculation percentages in Table 3 of the outby and inby fan locations are similar in
magnitude to that of the original locations. The recirculation at the original, and outby loca
tion 6 have the highest magnitude and occurrence while inby location 1 has the highest mag
nitude and occurrence. Recirculation across the BF pillar line can be thought of as the
‘overall sections recirculation’ given that the air is directly recirculating from downstream to
upstream of the BF. Location 6 has the largest magnitude in all positionings, 36% for outby,
30% for original, and 33% for inby positions. This is consistent with the results from Table 2
showing the highest recirculation percentages in original locations 5, 6, and 7 of scen
ario one.
Lastly, the crosscut airflow from the inby, outby, and original locations are shown
in Figure 7. A similar reduction in airflow in adjacent entries can be seen in both the
inby and outby locations. Inby location 6 shows a significantly different pattern than
11
the rest of the locations due the high airflow through entry 4W. Similar to Figure 4A,
the airflow in entry 6N in the inby location 6 simulation caught the corner of a pillar
and split the air stream between entry 6N and crosscut 4W. This air stream continued
towards the face and split resulting in positive airflow downstream and negative air
flow upstream through entry 1N as shown by the red circle in the inby location 6
graph in Figure 6. Even with this extreme example of airflow being directed towards
the face the average reduction in crosscut airflow still remains and is 79% and 74% for
positive and negative airflow respectably in the inby locations. The average reduction
for outby locations is 91% and 77% for positive and negative airflow respectably.
These reductions correspond to a 25-30% reduction per pillar line in all cases, indicat
ing that positioning the BF at an inby or outby locations does not have a significant
effect on crosscut airflow.
Table 3. Recirculation percentage of the inby, original, and outby fan locations,
fan locations 1, 2, and 6.
Figure 6. Airflow through entries 1N-8N for booster fan locations 1, 2, and 6 at the inby, original, and
outby fan locations, positive airflow indicates flow in the booster fan direction (downstream) and nega
tive airflow indicates flow opposite the booster fan direction (upstream).
12
Figure 7. Airflow through the crosscuts (1W-12W) from the inby, original, and outby fan locations, posi
tive airflow indicates flow from entries west to east in the crosscuts (1W to 12W) while negative airflow
values indicate airflow in the opposite direction.
A CFD model was created based on a mine ventilation survey of a typical perimeter ventilation
schemed large opening room and pillar underground mine. 15 BF positions were investigated in
two scenarios: the first scenario consisted of nine fan positions around the original BF placement
selected by the mine operator with the second scenario consisting of three fan positions three pil
lars inby and outby the original location. The airflow around the BF was investigated through
recirculation patterns and overall airflow within the BF entry and the room and pillar region with
a focus around the face velocity. The following conclusions were made based on the simulations:
(1) The highest total airflow in the section may not always be obtained when the BF is posi
tioned at the center of the entry but rather when the BF is positioned on the same side of
the entry as the open room and pillar areas.
(2) As noted in Krog and Grau (Krog and Grau, 2006) and confirmed in these simulations
air entrainment is maximized when high velocity gradients or non-uniform distribution
of airflow is seen through the adjacent BF entries. Furthermore, the maximum airflow
through the BF entry was confirmed through the simulations to be ~ 160 ft downstream
the BF which is consistent with their findings of ~170 ft.
(3) Not only is high regional airflow necessary to ventilate further adjacent entries from the
BF but some amount of airstream disruption via air diversion from pillar corners is
needed to maximize crosscut exchange between entries.
(4) The airflow through the crosscuts observes a 46%-55% reduction per entry indicating
that booster fans in this geometry have minimal effect past 4 adjacent entries. Leading
to a recommendation to keep the BF within 3-4 entries of the face.
(5) In situations where recirculation must be avoided a BF located more outby from the
center of the section is preferable. This outby fan location was shown to create ‘positive’
airflow around the face areas i.e., non-recirculating airflow.
While all mine geometries are unique and BF positioning should be investigated on a case-
by-case basis with the purpose of the BF in mind, these simulations and recommendations can
act as a starting point for LOM operators who utilize a perimeter ventilation scheme.
13
ACKNOWLEDGEMENT
This work was financially supported by The National Institute of Occupational Safety and
Health (NIOSH) under contract No. 75D30119C05743.
REFERENCES
Babu, V.R., Maity, T., Prasad, H., 2015. Energy saving techniques for ventilation fans used in under
ground coal mines—A survey. J. Min. Sci. 51, 1001–1008. https://doi.org/10.1134/S1062739115050198.
Carter, R., 2018. Focusing the Flow. Eng. Min. J. 219, 50–56.
Dunn, M., Kendorski, F., Rahim, M., Mukherjee, A., 1983. Testing Jet Fans in Metal/nonmetal Mines
With Large Cross-sectional Airways.pdf.
Gendrue, N., Liu, S., Bhattacharyya, S., Clister, R., 2023. An investigation of airflow distributions with
booster fan for a large opening mine through field study and CFD modeling. Tunn. Undergr. Sp.
Technol. 132.
Goodman, G., Taylor, C., Thimons, E., 1992. Jet Fan Ventilation in Very Deep Cuts-A Preliminary
Analysis.
Grau III, R.H., Krog, R.B., Robertson, S.B., 2006. Maximizing the ventilation of large-opening mines.
Proc. 11th U.S./North Am. Mine Vent. Symp. - 11th U.S./North Am. Mine Vent. Symp. 2006 53–59.
https://doi.org/10.1201/9781439833391.ch8.
Grau III, H., Krog, R., 2009. Using mine planning and other techniques to improve ventilation in
large-opening mines. Min. Eng. 61, 46–50.
Hargreaves, D.M., Lowndes, I.S., 2007. The computational modeling of the ventilation flows within
a rapid development drivage. Tunn. Undergr. Sp. Technol. 22, 150–160. https://doi.org/10.1016/j.
tust.2006.06.002.
Krog, R.B., Grau III, R.H., Mucho, T.P., Robertson, S.B., 2004. Ventilation planning layouts for large
opening mines. Soc. Mining, Metall. Explor. 1–9.
Krog, R.B., Grau, R.H., 2006. Fan selection for large-opening mines: Vane-axial or propeller fans -
Which to choose? Proc. 11th U.S./North Am. Mine Vent. Symp. - 11th U.S./North Am. Mine Vent.
Symp. 2006 535–542.
Leonida, C., 2019. Changing the Face of Mine Ventilation.
Nguyen, V.D., Heo, W.H., Kubuya, R., Lee, C.W., 2019. Pressurization ventilation technique for con
trolling gas leakage and dispersion at backfilled working faces in large-opening underground mines:
CFD analysis and experimental tests. Sustain. 11. https://doi.org/10.3390/SU11123313.
Parra, M.T., Villafruela, J.M., Castro, F., Méndez, C., 2006. Numerical and experimental analysis of
different ventilation systems in deep mines. Build. Environ. 41, 87–93. https://doi.org/10.1016/j.
buildenv.2005.01.002.
Thimons, E., Kohler, J., 1985. Measurement of air velocity in mines.
14
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
E. De Souza
AirFinders Inc., Kingston, Canada
ABSTRACT: While mine ventilation systems may account for 40% to 50% of the energy con
sumption of a mine operation, auxiliary ventilation alone may be accountable for half of this con
sumption. In effect, auxiliary ventilation systems comprise a significant portion of a mine
operation’s base energy demand and is consequently responsible for a large percentage of the total
mine operating costs. This paper presents how engineering design principles can be applied to
design efficient and reliable auxiliary ventilation systems, especially focusing on assemblage losses.
Case studies are presented to demonstrate the effect of design, installation and maintenance prac
tices on system reliability and operating costs. In particular, the effect of assemblage losses is
quantified in terms of operating efficiencies, energy consumption and costs.
1 INTRODUCTION
Increasing costs of electricity have resulted in emphasis on energy-efficient designs and oper
ation for all energy-consuming systems in mining. Since ventilation systems normally account
for 25-40% of the total energy costs and 40-50% of the electrical consumption of a mine oper
ation (De Souza, 2018, 2013), the optimization of ventilation systems is today a ventilation
engineer’s top priority.
While modern duct-fan systems require precise engineering design, meticulous attention to
installation and regular maintenance practices, many installations are often designed based on
outdated rules of thumb and with disregard to best installation practices. In the course of
many years of investigations of duct-fan systems, the author has found them to be, in general,
fairly energy inefficient, with many systems operating at efficiencies below 65% and with air
leakages ranging between 25% and 75% (De Souza, 2004).
Auxiliary ventilation systems comprise a significant portion of a mine operation’s base energy
demand and may be accountable for half of a mine ventilation system energy consumption.
Typical auxiliary ventilation systems are fairly energy inefficient; factors affecting mechanical
and ventilation efficiencies loss include design flaws, installation practices and air leakage.
General solutions and tactics for improving auxiliary ventilation systems as presented in
this paper come from multiple ventilation audits performed by the author. They target sub-
system components which affect shock losses and mechanical and ventilation efficiencies. By
increasing the efficiency of auxiliary ventilation system components and by correcting
inappropriate designs or system degradation caused by poor maintenance, the overall capacity
of the system in delivering air to the required active faces can often be improved.
2 CASE APPLICATION
A case application associated with extensive engineering work conducted by the author is pre
sented in this section to demonstrate how, by conducting detailed ventilation efficiency audits,
simple low-cost solutions can be devised to increase efficiency, reduce power consumption,
and lower operating costs.
DOI: 10.1201/9781003429241-2
15
The case study is based on the auxiliary ventilation system illustrated in Figure 1. A series
of analysis are performed to quantify the contribution of each component of the complete
system to energy consumption and costs: inlet bell, screen, silencers, system friction, couplings,
bends, duct exit. The effect of air leakage and of installation practices is also quantified. The
analysis was based on analytical procedures and scientific guidelines developed by the author
(De Souza & Dirige, 2022) and use of specialized software (AirFinders, 2022).
The auxiliary ventilation system, to be installed in a development drift to supply 18.88 m3/s
air for a diesel production fleet rated at 298.3 kW, has the following design characteristics:
– layflat duct - new, 1.2192 m diameter, provided in 15.24 m long sections, 10 sections of
ducting. Multi clip joints.
– spiral duct, new, 1.2192 m diameter, 1 section 15.24 m for bend. Multi clip joints.
– total column length - 167.64 m.
– fan: 1.219-0.80-1780 (fan diameter-hub diameter-rpm) with a 149.14 kW motor. 600 V.
Motor efficiency 95%. Power factor 0.84.
– inlet bell - 1.651 m diameter, 0.2159 m long.
– screen - wire mesh screen of 95% net free area.
– silencers - two podless flow through silencers.
– bend - right angle normal bend of 1.524 m radius.
– cost of power: $0.08/ kW.hr.
– operation - 24 hours/day, 365 days/year.
– standard density conditions.
Three cases, with operational features presented in Table 1, are considered. Case 1 represents
the system design to meet the flow requirements of at the face of 18.88 m3/s and considers
a design air leakage of 20%. The fan supply flow is 23.6 m3/s. An ‘installation quality factor’ of
10%, representing a ‘good’ installation, is used to adjust the system static resistance pressure.
Case 2 represents the system as installed. Attained flows at the face of 15.56 m3/s do not meet
requirements, and is based on an air leakage of 30%. The fan supply flow is 22.23 m3/s. An
‘installation quality factor’ of 30%, representing a ‘poor’ installation, is used. Case 3 represents
the system with the fan blade setting adjusted to achieve the required face flow of 18.88 m3/s.
Air leakage is not constrained, remaining at 30%. The fan supply flow is 26.97m3/s. The system
installation quality is not improved; an ‘installation quality factor’ of 30% is used.
The contribution of each duct system component to power consumption and operating
costs is presented in the following sections.
16
without an inlet bell. Table 2 presents a summary of results for the 3 cases. For cases 2 and 3,
substantial increases in operating cost of 723% and of 1,370% are noted when an inlet bell is
not used. Table 2 clearly shows that, when an inlet bell is installed, significant energy and cost
savings can be achieved.
Table 2. Head losses, power, and operating costs for inlet bell.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 14.73 0.59 412.28 –
2 108.90 4.84 3,393.06 723
3 160.27 8.64 6,058.39 1,370
2.2 Screen
The fan screen prevents debris from entering the fan. Case 1 has a screen of 95% net free area
installed and in Cases 2 and 3 the fan screen is partially blocked by the deposition of debris.
Table 3 presents a summary of results for the 3 cases. For cases 2 and 3, substantial increases
in operating cost of 340% and of 686% are noted when the screen is partly blocked with debris
and not well maintained.
2.3 Silencers
Silencers provide a level of noise reduction to meet specific needs and for compliancy with
regulations. Case 1 has silencers properly connected to the fan and in Cases 2 and 3 the silen
cers have a non-aerodynamic connection to fan. Table 4 presents a summary of results for the
3 cases. For cases 2 and 3, relatively large increases in operating cost of 19.7% and of 45% are
noted when the silencers are not properly installed.
17
Table 4. Head losses, power, and operating costs for silencers.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 49.77 1.99 1,392.95 –
2 53.50 2.39 1,666.93 19.67
3 53.50 2.89 2,022.36 45.19
Table 5. Head losses, power, and operating costs for layflat duct friction losses.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 604.66 21.60 15,136.12 –
2 469.54 17.47 12,240.01 –19.13
3 691.04 31.19 21,854.54 44.39
Table 6. Head losses, power, and operating costs for spiral duct friction losses.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 226.75 9.06 6,346.37 –
2 201.23 8.95 6,269.83 –1.21
3 296.16 15.97 11,195.12 76.40
Table 7. Head losses, power, and operating costs for layflat duct coupling losses.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 64.27 2.30 1,608.96 –
2 49.91 1.86 1,301.10 –19.13
3 73.46 3.31 2,323.12 44.39
18
3 cases. Because of the reduced fan flows, case 2 has a reduced operating cost of 1.2% and, for
case 3, because of the increased fan flow, an increase in operating cost of 76.4% is noted.
Table 8. Head losses, power, and operating costs for spiral duct coupling losses.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 71.42 2.85 1,998.86 –
2 63.38 2.82 1,974.75 –1.21
3 93.28 5.03 3,526.02 76.40
2.8 Bend
Case 1 has a properly designed bend and, in Cases 2 and 3, the installed bend is ‘kinked’,
resulting in a higher resistance pressure. Table 9 presents a summary of results for the 3 cases.
For cases 2 and 3, relatively large increases in operating cost of 97.6% and of 252.8% are
noted when the bend is not properly installed.
Table 9. Head losses, power, and operating costs for bend losses.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 45.40 1.81 1,270.75 –
2 80.59 3.58 2,510.85 97.59
3 118.60 6.40 4,483.27 252.80
Table 10. Head losses, power, and operating costs for exit losses.
Case Head Loss (Pa) Power (kW) Cost/Year ($/y) % Change in Cost
1 157.07 5.02 3,516.58 –
2 337.29 10.50 7,356.40 109.19
3 496.41 18.74 13,134.63 273.51
19
Table 11. Component percent contribution to annual operating costs.
Component Percent $/year change relative to case 1
Case 2 Case 3
Inlet bell 723.0 1,369.5
Screen 340.3 686.1
Silencers 19.7 45.2
Friction layflat –19.1 44.4
Friction spiral –1.2 76.4
Joints layflat –19.1 44.4
Joints spiral –1.2 76.4
Bend 97.6 252.8
Exit 101.2 273.5
Overall 19.1 109.6
the correcting duct outlet. Also, the duct column installation quality should be improved, and
air leakage controlled. These actions will bring the system operation close to the design
specifications.
Figure 2. Fan characteristics and operating points for the three cases.
20
2.12 Operational costs
Table 13 presents annual fan operating costs as a function of the number of duct installations.
Independent on the number of installations, fan operating costs increase by 10% and 93.7%
for cases 2 and 3, relative to case 1. Typical hard rock mines may have well over 40 auxiliary
ventilation systems installed to support development and production activities. For the case
study presented, annual cost savings reaching some $1.65M can be realized when a system is
installed and maintained according to design.
It is noted that the comparative analysis was based on direct costs only. It is recognized that
ventilation training, asset management and preventive maintenance produce significant reduc
tions in ventilation operating costs. While indirect costs are site dependent, the author has
observed that when workers acquire practical ventilation training prior to working under
ground, which normally comes at a very nominal cost, considerable improvements in ventila
tion system performance are readily realized.
Table 13. Annual operating costs for multiple duct system installations.
Number of Duct System Installations
Case 1 5 10 20 30 40
1 43,773 218,865 437,730 875,460 1,313,190 1,750,920
2 48,160 240,800 481,600 963,200 1,444,800 1,926,400
3 84,797 422,983 847,965 1,695,930 2,543,895 3,391,860
3 CONCLUSIONS
This paper has demonstrated how the use of proper engineering design for optimal duct
system installations would dramatically result in reduced energy consumption and in reduced
operating costs. When properly designed, installed, and maintained, an auxiliary ventilation
system can operate efficiently with substantial power and operating cost savings.
In a case study presented, it has been illustrated that savings in fan energy consumption
exceeding 94% can be achieved by correctly commissioning and maintaining a properly
designed auxiliary ventilation system.
REFERENCES
AirFinders, 2022. AirFinders Force Auxiliary Ventilation System Design. Engineering Design software.
De Souza, E. 2004. Auxiliary ventilation operation practices. Proceedings of 10th U.S./ North American
Mine Ventilation Symposium. Anchorage:Balkema. 341–348.
De Souza, E. 2013. Improving the energy efficiency of mine fan assemblages. Proceedings of the 23rd
World Mining Congress, Montreal, QC. pp 9.
De Souza, E. 2018. Cost saving strategies in mine ventilation. CIM Journal. Volume 9. Issue 2.
De Souza, E. & Dirige, P. 2022. Auxiliary mine ventilation manual. Version 2.0. Workplace Safety
North. Pp 261.
21
Case studies of mine ventilation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
A.A. Habibi
Technical Expert, PT Freeport Indonesia
I. Setiawan
Chief Ventilation Engineer, PT Freeport Indonesia
R. Prasojo
Mine Engineer, PT Freeport Indonesia
C. Stewart
Principal Engineer, Minware Consulting
ABSTRACT: A ventilation study was conducted on Kucing Liar (KL) mine. KL is the add
ition to PT Freeport Indonesia (PTFI) panel cave operations and is anticipated to operate in
parallel to Grasberg Block Cave (GBC). The study consisted of model construction, budget
preparation and investigating multiple scenarios to meet the minimum budget requirements.
Utilization of new engine technologies controlled partial air reuse, push-pull and utilization of
Battery Electric Vehicles (BEVs) were considered during this feasibility study. A main intake
and exhaust drift size optimization study was also conducted to determine the number of add
itional main fans and drift sizes required to support the KL mine. The results show the
requirement for two additional 5MW main fans, two intake drifts and two exhaust drifts
(7 m x7 m). The results also show the requirement for eight 6m ventilation raises to support
primary air delivery to KL mine. Proof of concept trials are deemed necessary to determine
the feasibility of applying some of the proposed concepts in underground operations (such as
BEVs). The trial results will be used to complete a detailed ventilation network design.
1 INTRODUCTION
GBC mine development started in 2004 with initial drifting and the first drawbell blast in 2018.
GBC mine is expected to achieve full production rates of 130 kt/d to 160 kt/d of ore by 2025.
Currently, the three active underground mining operations in the Grasberg Mining District are
the GBC block cave mine (target 130-160 kt/d ore), the Deep Mill Level Zone (DMLZ) panel
cave mine (target 80 kt/d), and the Big Gossan open stoping operation (7 kt/d). The Deep Ore
Zone (DOZ) panel cave mine (80 kt/d) has been closed and is no longer in operation.
The KL orebody is a new large caving operation where development started in 2021 and pro
duction is due to begin in 2027. The Grasberg Open Pit commenced production in 1990 with
active pit production ending in late 2019. Figure 1 is a schematic view of the Grasberg District
orebodies. The concentrating plant has a peak capacity of about 240 kt/d. Peak production targets
from the underground district will remain at 240 kt/d and will be entirely from underground. The
GBC will be the primary source of mill feed targeted at 130-160 kt/d (Brannon et al. 2020).
DOI: 10.1201/9781003429241-3
25
KL mine. Five Howden mixed flow rated at 5 MW each are currently installed in a parallel in an
exhaust configuration to support the GBC mine. Fresh air from the mill valley is downcast to the
mine through four parallel primary intakes (GVD 1-4) mined at 6.8 mW × 9.0 mH, at 2.6 km
from the portals to the footprint. Air is distributed to the working areas of the mine using ventila
tion drifts and raises mined at various dimensions. The return air reports back to the under
ground main exhaust fans and out into the valley. Figure 2 displays the main components of the
ventilation system. The system capacity is 3500 m3/s when all five fans are running.
Based on the recent LOM baseline estimate, the current GBC mine (which will be operating
in parallel with KL block cave) consumes approximately 25 m3/s/ktpd. Other block cave
mines typically have ratios from 17 to 40 m3/s/ktpd (Brannon et al. 2020). while the ratio cur
rently sits at the lower end of the scale, a Ventilation On Demand (VOD) system has been put
in place to improve the efficiency of air distribution.
Air velocity and diesel exhaust dilution requirements form the main design criteria. The air
flow velocity requirements air primarily for dust control in the ore flow system, and diesel
dilution for development and production activities.
2 DESIGN SCENARIOS
The GBC mine production is scheduled to increase to 136 ktpd by 2024. KL mine develop
ment has commenced, with steady-state full production planned for 90 ktpd by 2033. The
mine will supplement GBC mine production which reduces to 103 ktpd by 2033, giving
a combined total tonnage from both mines from 2033 of 193 ktpd. Five scenarios were con
sidered as case studies for this design.
27
2.2 Scenario 2 - KL mine air resuse
An air reuse option redistributes 600 m3/s of cleaned infrastructure exhaust to the extraction
level, effectively reducing the airflow drawn to and from the surface by 600 m3/s. This permits
a reduction in primary main fan infrastructure requirements, and a resultant reduction in cap
ital and operating costs.
Infrastructure level exhaust is proposed for reuse as it is lower (per unit volume) in diesel
activity (DPM is more difficult to scrub and reduce). Reintroduction of the scrubbed air to
the extraction horizon is a form of ‘series’ ventilation meaning gases will not continue to build
as may be the case in a recirculated design.
This proposal is speculative, in that no effective validated air-reuse scrubber arrangement
has been proven for this mine. To do this, the characterization of dust and contaminants
would need to be performed, and available technologies evaluated for cost and efficiency.
A proposed arrangement is to parallel 80 m3/s capacity scrubbing units together in at least
two locations on the infrastructure air return level, linked by raises into the extraction level.
Several proven wet and dry technologies exist for removing fine silica and nuisance dust
from the air. Removal of DPM from the air requires dry filter scrubbing which should be
avoided from a likely cost and maintenance perspective. The most likely robust fine dust
removal technology is ultra-fine wet spray chambers, using water atomized with the assistance
of air compressors. For design purposes, power totaling 400 kW is assumed to overcome 2
kPa of filter media or sprays with collectors, and 1.5 kPa air reinjection into the mine extrac
tion level.
28
ventilated diesel machine fire. A BEV LHD trial is underway (in 2023) as a follow-up to
hazard and risk assessment strategy identifying BEV-specific safety issues, controls and mine
design changes.
29
Figure 3. KL mine ventilation budget reduction.
30
Figure 4. Velocity optimization for primary ventilation drifts.
Where personnel are required to travel in high-velocity drifts, dust and coarse particle
entrainment in the air can create safety hazards, particularly with eye injuries. The KL
design for PTFI uses an economic velocity range that may exceed 10 m/s for infre
quently travelled airways while maintaining 10 m/s or less for common travel ways. Spe
cial safety precautions and travel restrictions will apply for travelling in airways
exceeding 10 m/s. Nonetheless, every effort is made to reduce air velocity to below the
maximum economic limit. For example, for Scenario 1 (fully diesel mine, no air re-use)
two GVD intakes have been added to the system as optimum velocity analysis indicates
that this can be done at a cost-neutral basis, the impact being higher capital costs being
offset by longer-term lower operating costs.
31
Figure 5. KL service level profile size optimization.
Figure 6 shows the results for selected scenarios. The base case diesel mine, utilizing existing
airflow concepts and designs, is unsurprisingly the highest cost case, requiring at least two full
intake and exhaust drifts and fans and higher power for the increased ventilation flow.
The full BEV option delivers the lowest overall capital and operating cost, reducing primary
infrastructure requirements and reducing power costs. A partial BEV and air reuse option
deliver the next lowest capital and operating costs, while an air reuse option with diesel equip
ment is the third lowest cost option.
32
4 CONCLUSION
No commitment to any specific mine ventilation design scenario could be made until further
research is completed on BEV equipment options, and air reuse strategies. An intermediate
KL Mine Development plan required for all scenarios is recommended for one additional
7.0m x 7.0m GVD intake, exhaust drift, primary fan and four 6.0m internal shafts each for
the intake and exhaust connections. In summary, the intermediate mining and infrastructure
plan will increase GVD exhaust capacity from 3750 m3/s to nearly 4500 m3/s.
Mining an additional GVD exhaust and intake access of 7.0m x 7.0m linear development
meters was recommended. While the optimum development size should technically be 6.8m
x 9.0m high, a more rapid single-pass mining option of 7.0m x 7.0m is acceptable which gives
near equivalent financial optimization at the loss of 100 m3/s of airflow.
The addition of a 5MW primary mixed flow fan was recommended and added to the plan
as well as four KL exhaust raises (360m each at 6.0m diameter) and four equivalent KL intake
raises.
The intermediate plan is required at a minimum for every conceivable scenario until the via
bility of options of BEV or air reuse is resolved. Further work will be undertaken to validate
BEV viability and air reuse viability. An additional intake and exhaust drift with a primary
fan and internal raise will only be required if the reuse of infrastructure air or BEV use cannot
be made viable.
The full BEV option provides the lowest-cost ventilation option and, subject to other viabil
ity analyses of BEVs, is recommended. Partial air re-use is viewed with some risk and contro
versy, particularly if diesel equipment is continued to be used, as fine DPM is difficult to
remove.
Electric vehicles (BEVs) are seen as an inevitable implementation in underground mines
given the demonstrable benefits in lower heat, reduced airflow requirement and the elimin
ation of gas and DPM emissions. As the mining ventilation plan for scenarios 1, 3 and 4 are
similar, the implementation of BEVs does not influence the intermediate mine ventilation
infrastructure plan, only the later size and type an air-re-use facility.
REFERENCES
Brannon, C. Brard, D. Pascoe, N. Priantna, A. 2020, “Development and production update for the Gras
berg Block Cave mine – PT Freeport Indonesia”, 8th International conference and exhibition on Mass
Mining virtual conference Proc, pp 747 to 760, 9-11 December 2020, Santiago Chile. ISBN 978-956-
19-1196-3
McGuire, C, Witow D, Mayhew M, Bowness, K. 2022, “Comparison of heat, noise and ore handling
capacity of battery-electric versus diesel LHD”, Australian Mine Ventilation Conference, Gold Coast,
QLD, Australia, 10-12 October 2022.
Stewart, C.M., 2022. “Ventilation considerations and modelling of lithium battery fires in underground
mining.” Australian Mine Ventilation Conference, Gold Coast, QLD, Australia, 10-12 October 2022.
33
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: This paper presents engineering design and equipment selection for a successful
temporary spot cooling installation to support underground shaft sinking that has been in oper
ation for the last 14 months: September 2021 to November 2023. The need for cooling was
driven by the depth of shaft sink (starting from >1,900m below surface), however this system
was subject to many of the common challenges preventing the widespread use of spot cooling,
including limited process water & dewatering capability, heat rejection equipment placement in
the path of blasting fumes, limited airflow quantity for heat rejection, and layout constraints due
to the existing and upcoming mine services installations and construction. Components were
selected for mine-duty with consideration for the dusty environment. Use of hybrid cooling
towers allow for increased heat rejection capacity from evaporative cooling while maintaining
a fully closed-loop condenser water circuit. Skid-mounting of all components allowed for easy
placement and relocation. Use of HDPE piping lashed to existing ground support allowed for
maximum layout flexibility and minimized installation time. Performance, operational features,
and additional lessons learned, including feedback from operations personnel, are shared.
1 INTRODUCTION
This paper concerns a case study project which is currently in execution phase. This project
required a small-scale underground “spot cooling” system deployment to support shaft sinking
operations in the construction of a new winze. Mechanical refrigeration at this relatively small
scale is widely understood and deployed in industrial surface applications; however, the ability to
effectively reject heat within the confines of the underground workplace requires careful design
due to a variety of factors including limited footprint and headroom availability, high air temper
atures, low available air flow volumes, and dirty/dusty air conditions. Additionally, where these
spot cooling installations are proposed for temporary applications, there is often a constraint on
available electrical services and water supply/removal capabilities, and little desire to invest in
reticulation systems to support the temporary facility.
None of these challenges are, on their own, sufficient to disqualify mechanical refrigeration
from spot cooling applications. Operational challenges can be attributed to multiple small issues
concurrently as opposed to a single, broad failure in design. Instead, careful attention needs to be
paid in the design and specification of equipment for underground service, to obtain fit-for-
purpose equipment which maximizes the chance of success given the known constraints. This
paper presents one such case study in which the selection of spot cooling equipment specifically
for the challenging underground service has resulted in a robust and reliable operation.
2 APPLICATION DESCRIPTION
The case study considers a project in the process of an expansion to a deep orebody which will be
accessed by an internal winze shaft from approximately 1,200m to 2,600m below surface. The
DOI: 10.1201/9781003429241-4
34
lower leg (1,900m to 2,600m) of the winze is being developed using conventional “blind” shaft
sinking methodology with a Galloway stage and equipping stage progressing in parallel. Prior to
starting development of the winze, there was no refrigeration capacity present at the mine.
Due to the significant depth of the sinking activity, refrigeration is required to manage
workplace temperature conditions and maintain safe, productive advance rates for the largely
manual mining workplace. The mine will operate a large refrigeration plant for future mech
anized development and production activities, but this plant will not be available until the
completion of the sinking process. Therefore, a temporary interim solution was required,
which was to be fit-for-purpose for the reduced ventilation demands of a shaft sinking applica
tion as opposed to mechanized drill & blast with multiple crews and diesel vehicles. The appli
cation is considered to be in-line with typical “spot cooling” applications in underground
mining due to the small air flow rate and dedicated cooling of a single workplace, as compared
to “bulk air cooling” of a mine’s entire intake air stream.
A simplified schematic of the mine’s ventilation system is presented in Figure 1. Custom
ductwork was installed to allow intake air from the downcast portion of the shaft, bypassing
the heat from diesel mobile equipment operating in the service decline to support development
and construction activities throughout the mine.
The cooling system is located at 1,900m below surface at a mid-shaft station. During the
plant’s period of operation, none of the mine’s permanent systems and infrastructure (e.g.,
process water supply, dewatering, and electrical distribution) are available at the plant loca
tion. Therefore, minimizing the duty with which these services were impacted by the cooling
plant was a key design consideration.
3 PROCESS CONDITIONS
35
The outlet temperature from the air-cooling plant was selected by the design engineers
based on heat modelling, reflecting conditions to mitigate workplace heat stress risk. The heat
stress index of choice at the site is wet bulb globe temperature (WBGT), and the maximum
workplace temperature target is 28°C WBGT, based on the ACGIH limit for 100% work with
a moderate work intensity.
Process conditions for the system are summarized in Table 1.
One critical and uncommon characteristic of the ventilation system is the use of fiberglass-
reinforced polymer (FRP) construction for the sinking ventilation duct. This material has better
thermal insulative properties than steel or flexible fabric duct, allowing for warmer chilled air tem
perature at the cooling coil outlet for a given temperature at the discharge of the duct. This selection,
in addition to equipment selections described in Section 4 below, was made with the primary object
ive of minimizing electrical loading required to achieve a safe workplace for the shaft sinking crews.
4 EQUIPMENT SELECTION
Equipment selection for the spot cooling system focused on ensuring robust performance at
the required process conditions while attempting to minimize impacts to other mine systems
(e.g., water and electrical) during the temporary installation period. A schematic of the final
system is shown in Figure 2.
The general configuration was largely dictated based on the overall process and mechanical con
ditions. Specifically, the air-cooling requirements fell well within typical applications for indirect
water-cooling coils (i.e., tube and fin heat exchangers). This equipment type also worked well to
facilitate tie-ins to the fresh air ductwork supporting the mine development. Although the cooling
coil is located in the mine’s fresh air stream, there is still anticipated to be dust present due to the
nature of underground operations. Thus, the design was specified to include a wider-than-standard
fin spacing of 7 fins per inch (vs. common surface applications as high as 12 fins per inch), which
reduces risk of dust accumulation plugging the coils and reducing both airflow and heat transfer
efficiency. 7 fins per inch was selected in consultation with the equipment supplier as a compromise
between risk of fouling and the reduced heat transfer performance offered by the reduced fin quan
tity. The increased spacing has the added benefit of reducing the coil air side pressure drop, thereby
minimizing the fan power required for the temporary ventilation system.
Water is transferred in a closed loop circuit between the air-cooling coil bank and a water-cooled
chiller. The large physical space and air-flow requirements associated with an air-cooled chiller were
deemed not feasible with an underground installation location. Additionally, presence of dust in
underground workplaces presents significant operational and maintenance challenges with air-cooled
chilling equipment. Frequent inspection and replacement of air side filters are required which can put
strain on owner or contractor maintenance resources. Thus, this equipment type was not considered
for the application. A screw compressor with dual compressor arrangement was selected to maximize
compressor lift available in the event of challenging heat rejection conditions (i.e., high condenser
temperature/pressure), and R-134a refrigerant is used due to its Class A1 (non-flammable and non-
toxic) properties.
36
Figure 2. Schematic of the Cooling System.
The last component of the spot cooling system is the heat rejection circuit. Design of this area
was of critical importance in ensuring consistent, robust performance of the overall system. Design
of heat rejection systems in underground applications requires careful consideration and can be
challenging due to a variety of factors including high air temperatures, low available air flow vol
umes, and dirty/dusty air conditions. For the spot cooling system, hybrid cooling towers were ultim
ately selected for heat rejection. The towers operate through indirect heat exchange between the air
and a closed-loop water coil connected to the chiller. They also utilize a small spray water stream
on the coil exterior to facilitate latent heat transfer, and notably this evaporative cooling water
source is completely independent of the chiller system (which prevents contamination of the conden
ser cooling water circuit).
Overall, this heat rejection configuration provided many benefits to the overall installation.
First, it enabled maximum thermal performance without introducing potential fouling and
water quality concerns with the chiller. The project did consider indirect heat transfer through
the use of tube and fin heat exchangers similar to the air-cooling coil; however, this configur
ation was eliminated. The lack of evaporative cooling and the high dry bulb temperature of the
mine air stream available for heat rejection (only approximately 5°C cooler than the condenser
37
water stream) resulted in prohibitive quantities of ventilation air required to achieve the design
heat rejection.
Secondly, the use of indirect heat exchange with only an external spray design simplified the
process water circuit by minimizing make-up and blow-down water requirements as compared
to direct evaporative cooling. The range of acceptable water chemistry is also greater in the
cooling towers, which are typical galvanized steel construction, in comparison to the special
ized material present in the chiller heat exchangers which likely requires chemical dosing and
water treatment consideration to prolong the life. This was an important consideration as the
mine water systems are limited during temporary development phases, both for supply of
fresh process water for make-up and for dewatering capacity to accommodate blow-down.
Finally, the tower design attempts to minimize concerns with dirty/dusty air conditions and
potential heat exchanger fouling (which will worsen heat transfer performance). Although the
tower does contain a coil, it is designed to reduce air-side pressure drop while still maximizing
area and heat transfer performance. The external spray water system also helps promote coil
cleaning and removal of potential dust build-up on the heat exchange surface.
It is also noted that the system was designed from a process and mechanical perspective to
minimize electrical system impacts, again an important consideration due to electrical system
constraints during early mine development periods. For example, reasonable airflow rates and
velocities were targeted to minimize pressure drops and resulting electrical motor sizes. In add
ition, the closed-loop water circuit temperatures were mindful of chiller lift considerations and
enabled a coefficient-of-performance of slightly below 4 at maximum design conditions
(impacting the compressor motor size and electrical consumption).
Because this spot cooling system is directly coupled to the shaft sinking ventilation system,
reliable operation is vital to prevent downtime on the project critical path. Improvements in
reliability were achieved by installing piped-in standby pumps on both closed-loop water cir
cuits. Additionally, both the cooling coils and the cooling towers consist of two units installed
in parallel. While this does not achieve full redundancy, it removes the risk of a “single point
of failure” in the cooling system causing a complete outage. Consultation with equipment sup
pliers indicated that the system could operate (at reduced capacity) in the event of failure of
any one compressor, cooling tower or cooling coil. Preventative maintenance and cleaning of
the system was identified as a key enabler of reliable performance and would be performed on
the system regularly via a service agreement with a local supplier.
5 INSTALLATION
Another key aspect of the equipment design and selection was the ease of installation in an
underground mining environment. The system was designed such that installation and connec
tions could be achieved by tradespeople within the workforce of typical underground mining
contractors, and not requiring any specific qualifications. Given the relatively small size and cap
acity of the chiller, it was shipped to site fully assembled and charged with refrigerant, eliminating
the need for any qualified refrigerant technicians to participate in the installation process.
All pipe connections were Victaulic grooved, consistent with the mine’s standard for air, water
and dewatering lines. Complex piping arrangements including hard-piped headers and tie points
were fabricated in the manufacturer’s shop, minimizing the number of field connections required
and ensuring that all field connections required identical couplings. An example is shown in
Figure 3, in which four parallel inlets and four parallel outlets for the cooling coils are piped
back to a single DN150 (6-inch) grooved inlet and single DN150 grooved discharge connection.
Due to the mine layout and ventilation system, the cooling towers were installed approximately
200 metres from the chiller and cooling coils, adjacent to the return air raise. The main condenser
water piping across this distance was installed using HDPE piping, which offered several
advantages:
• Pipes were sufficiently lightweight to be supported from existing ground support screen
using chains (see Figure 4), eliminating the need to drill any rock anchors for pipe supports
for a temporary application,
38
Figure 3. Photo of shop-installed pipe headers on air-cooling coils.
• Thermal expansion and contraction can be easily accommodated by allowing the pipe to
sag between subsequent supports,
• Flexibility to adjust layout to work around existing infrastructure and avoid interferences,
• Significant reduction in the number of couplings required, improving installation time.
Flexible hose was selected for all water lines except the 200m run from the chiller to the
cooling towers. This provided flexibility to field-locate equipment skids and adjust to condi
tions in the underground workplace.
Finally, in addition to local installation considerations, the system was specifically designed
to be easily transported from surface to underground. The system is operating in a shaft
Figure 4. Photos of HDPE Pipe Supported by Chain from Mine Ground Support Screen.
39
access mine, and therefore all components were selected and manufactured to be shipped
within cage size constraints. Of all the equipment included in the system, only the cooling
towers needed to be broken down for transport due to their overall height of 3.9m being
unsuitable for ramp transport on flatbed trucks available at the mine. A photo of the cooling
tower split in preparation for transport is shown in Figure 5.
Custom skids were engineered for each piece of equipment, which allowed for forklift trans
port and creating a stable base that allowed the equipment to be installed on compacted mine
roadway ballast. Skids with fork pockets are visible in Figure 3 and Figure 5. The level of
flexibility was so pronounced that the complete chiller, cooling coil and evaporator water
system was installed in a temporary location to provide interim cooling in the summer
months, and then relocated to the final location only three months later.
6 PERFORMANCE
The temporary refrigeration system has been in service for 14 months as of the writing of this
paper in October, 2022. The system has been operating reliably and has yet to experience any
unplanned downtime. The only unplanned maintenance activity to date resulted from a failed
seal on the operating condenser water pump; however, the standby pump was switched into
service and the repair was made without any appreciable downtime to the plant.
A ventilation fan upgrade in 2022 has increased the total airflow delivery to the sinking
crews from 12 m3/s to approximately 20 m3/s since the start of operation. Periodic measure
ments of the air side differential pressure have seen a negligible change over time, indicating
little dust build-up on the coils. The housing was opened after 12 months of service to clean
the coils, and no notable dust was observed. Differential pressure remained unchanged after
cleaning. A photo of the air-cooling equipment is presented in Figure 6.
Cooling tower performance has been stable throughout the operation. Dust filters on the
air intakes are cleaned periodically by spraying down with process water - a cleaning process
which can be done without shutdown of the units. The self-cleaning characteristics of the
40
Figure 6. Photo of sinking ventilation fan and air-cooling coil in service.
internal coils with water sprays have maintained consistent operation despite the dusty envir
onment. A photo of the cooling towers in service is shown in Figure 7. It can be seen that the
intakes on the right-hand unit have recently been cleaned at the time of the photo.
Operator feedback has been positive, noting that the complete system is operated in
a hands-off manner. The chiller automatically modulates duty by adjusting screw compressor
speed using its on-board variable speed drive (VSD), and all other components operate con
tinuously without need for any modulation or intervention.
System performance has been measured to align with the engineering design values for the
conditions to date. As sinking progresses and ambient temperatures increase, the system per
formance at the design maximum values will be measured and validated.
Figure 7. Photo of cooling towers in service with recently cleaned air intakes on the right hand unit.
41
Computational fluid dynamics applications in mine ventilation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: The ventilation of mines and underground works has been studied for more
than 100 years, with extensive and rigorous results that currently aid in safely performing
underground operations of different kinds and characteristics. However, the study of under
ground ventilation in large volumes such as caverns has seen relatively little open study, being
part of engineering studies without further disclosure. In general terms, it is the final use of
the cavern that determines the ventilation system to be used. In this regard, the present study
is generated from the need to define elementary ventilation systems for underground caverns
linked to the construction system, which allows for maximum pollutant drag with minimum
flow rate and reducing recirculating secondary flows. For this, Computational Fluid Mechan
ics is used as a simulation and analysis tool through the Ansys Fluent software, generating
various simulation scenarios of steady and transient flows. The results show different ways of
ventilating according to the established boundary conditions and the construction geometries
used, which could help to orientate future ventilation designs from an academic exercise.
1 INTRODUCTION
DOI: 10.1201/9781003429241-5
45
Figure 1). This also aids with fortification as the walls are lowered, leaving the cavern ceiling
and walls reinforced. Also, can be find three tunnels, one arrives at the upper and two at the
lower opposite ends.
Figure 1. Examples of mining caverns of Codelco: (a) Crusher cavern of Dacita, El Teniente (b)
Crusher cavern of Chuquicamata Underground; (c) Crusher cavern of Diablo Regimiento, El Teniente.
46
2 GEOMETRY DESIGN
This provides the results of factors shown in Table 1, from which measures with a factor
equal to the average of both (0.31) were chosen for this study. The value of 0.31 converges for
other case studies carried out, mainly those in health and social areas, including operating
rooms (Anuraghava et al., 2021), supermarkets (Mohamadi and Fazeli, 2022) and schools
(Jeong et al., 2022). Therefore, the dimensions chosen to carry out the base design according to
the geometry of the cavern are 30 m high, 21 m wide, 45 m long, and a size factor of 0.31.
47
Special design: a design that considers the air inlet and outlet from the same face or wall,
for cases where there is no possibility of creating ventilation paths on the opposite wall. This
is considered in a separate analysis because, constructively, the excavation would only be from
below, when access to the cavern construction is limited (for example, disassembling a TBM
in a dead end).
3 CFD SETUP
1 1,852,108
2 1,892,369
3 1,882,534
Boundary conditions established are velocity inlet (upper tunnel as inlet) and pressure outlet
(lower tunnel), as shown in Table 3. The same table includes initial conditions, velocities, and
time steps used in the unsteady cases. Time independence tests were made to validate conver
gence, based on Courant number less than 1. Unsteady cases were made to test flow instabilities
and some DPM cases to evaluate dust residence time, but most cases were a steady simulation.
48
3.2 Turbulence model
In order to choose the turbulence model to be used, the researchers’ experience was considered
and an exhaustive bibliographical analysis was carried out, searching in various areas such as
transportation, energy, mining, health, and others (17 studies overall). Additionally, for this
work it not needed heat transfer. So that, the chosen turbulence model is k-epsilon Realizable,
with second order upwind spatial discretization and Simplec pressure-velocity coupling scheme.
1 0.5 8
2 2.5 40
3 5.0 80
4 10 160
4 RESULTS
49
hand, the maximum speed of 10 m/s (case 4) gives good sweeping of contaminants, with the excep
tion of the special design, wherein even with high flow speed the sweeping is not entirely good.
The special design has poor performance, even with high speed, because the inlet and outlet air
ways are closely each other.
Table 6. Design-case combinations.
Air volume with effective Air volume with effective
Design Case circulation Design Case circulation
Figure 3. Designs and cases for lower (left side) and higher velocities (right side) tested.
50
4.2 Optimized designs-case
From the exposed cases, the ones with the best performance have been selected, i.e., with
very high or total effective circulation. These were considered with the potential to be opti
mized by applying the DPM model, for a turbulent flow with the presence of mineral par
ticles requiring a minimum fluidization speed of 0.04 m/s (Sakar, 2015), to ensure an
effective drag of larger particles. To optimize cases, an analysis of particle entrainment and
particle residence time is required to establish the optimal contaminant aeration model.
Dust particles with a minimum diameter of 1 μm and a maximum of 1 mm to be injected are
considered (Ormeño, 2014).
Additionally, a mixed circuit push-pull (push from the top and pull from the bottom) was
also applied to designs 2 case 4 and design 3 case 3, to analyze if the flow improves taking into
account an additional pressure output (exhaust pressure). Figures 4 and 5 show particle Iso
surfaces with lower velocity than 0.04 m/s (Re<4000) and particle residence time (seconds),
respectively, for Design 3-Case 4 (a) and Design 2-Case 4 mixed (b). Table 7 shows particle
drag is bad for most of cases, which means some particles staying permanently into the
vortex, except Design 3 case 4 and Design 2 case 4 Mixed. Design 3 case 4 is the best, since it
only requires air injection and not a mixed air circuit (Push-Pull).
High speeds are directly related to better ventilation and removal of polluting particles.
Although the geometries have been obtained from crushing caverns, the simulations carried
out do not have a specific focus on this use of caverns. This is a general approach for entrain
ing pollutants to keep indoor air clean. In particular, in the case of crushing caverns, both the
air speeds at the outlet must be adequate so as not to generate contaminant dispersion, so it is
recommended to carry out a specific CFD study for these cases.
2 4 78% Bad No
3 3 89% Bad No
3 4 95% High Yes
2 4-Mix 95% Medium No
3 3-Mix 89% Bad No
Special 4 62% Bad No
Figure 4. Particle Isosurfaces with lower velocity than 0.04 m/s for Design 3-Case 4 (a) and Design
2-Case 4 mixed (b).
51
Figure 5. Particle residence time (s) for Design 3-Case 4 (a) and Design 2-Case 4 mixed (b).
5 CONCLUSIONS
Cavern ventilation designs are established to take advantage of the constructive galleries used
for their development. The objective of these studies is to avoid dead spaces without air move
ment. These designs are studied through computational fluid mechanics, a k-epsilon turbu
lence model and a particle transport model (DPM), and a type of hexagonal structured mesh
being recommended for this type of simulation. Although the proposed models generally meet
the basic purpose of ventilation, optimized designs are determined. Optimized designs allow
for entraining all the air in the cavern, but need significant velocities of 5 to 10 m/s. Design 3
with a speed of 5 m/s shows particularly high study opportunities, as it can deliver satisfactory
results.
In general, cases with areas with less air renewal are identified, particularly in the upper sector
of the face containing the exits and the space corresponding to the lower part of the wall with
the entrance; these require the implementation of boosters or fans. Designs with cross ventila
tion are considered better, with an entrance and exit on opposite faces and positions, generating
greater air mobility inside the cavern. A good entrance speed is established with this method,
generally 5 m/s. If it is not possible to access sufficient ventilation in flow or crossover design,
the use of boosters is recommended to avoid leaving dead spaces, for which CFD tools should
be used. The special design (both accesses at the same face) has poor performance even at high
speed, so this design should be avoided from a ventilation point of view if possible.
6 FUTURE WORKS
Cavern ventilation design is a challenge to face case by case. As part of the research develop
ment, future works other variables will incorporated as local booster fans to improve stagna
tion zones, heat sources and heat dissipation, different shapes of cavern, others in and out
locations of the air and ascending ventilation (bottom to up). For these types of modelling or
similar, authors suggest to use DPM model to dust modeling, because is a very useful tool to
analyze it behavior according to velocities and shapes.
REFERENCES
Anuraghava, C., Abhiram, K., Sai Reddy, V. N. & Rajan, H. 2021. CFD modelling of airborne virus
diffusion characteristics in a negative pressure room with mixed mode ventilation. International Jour
nal for Simulation and Multidisciplinary Design Optimization 12: 8.
Bayatian, M., Ashra, K., Amiri, Z. & Jafari, E. 2021. Computational Fluid Dynamics Simulation of
Airow and Air Pattern in the Living Room for Reducing Coronavirus Exposure. Research Square.
Bhattacharyya, S., Dey, K., Paul, A. R. & Biswas, R. 2020. A novel CFD analysis to minimize the spread
of COVID-19 virus in hospital isolation room. Chaos, Solitons & Fractals 139.
52
Chang, P., Xu, G., Zhou, F., Mullins, B., Abishek, S. & Chalmers, D. 2019. Minimizing DPM pollution
in an underground mine by optimizing auxiliary ventilation systems using CFD. Tunnelling and Under
ground Space Technology 87: 112–121.
Guan, X., Jiang, Y. & Yao, Y. CFD simulation of a novel ventilation system of subway station in
Harbin. Building Simulation, 2007 Beijing, China. 432–346.
Hoek, E., Carranza-Torres, C., & Corkum, B. (2002). Hoek-Brown Failure Criterion. En North American
Rock Mechanics Symposium and The 17th Tunnelling Association of Canada Conference. Toronto:
Tunnelling Association of Canada. Recuperado el 25 de Abril de 2022
Jeong, D., Yi, H., Park, J.-H., Park, H. W. & Park, K. 2022. A vertical laminar airflow system to prevent
aerosol transmission of SARS-CoV-2 in building space: Computational fluid dynamics (CFD) and
experimental approach. Indoor and Built Environment 31(5): 1319–1338.
Mohamadi, F. & Fazeli, A. 2022. A Review on Applications of CFD Modeling in COVID-19 Pandemic.
Archives of Computational Methods in Engineering 29: 3567–3586.
Obeidat, B., Fawwaz Alrebei, O., Atef Abdallah, I., Darwish, E. F. & Amhamed, A. 2021. CFD Ana
lyses: The Effect of Pressure Suction and Airflow Velocity on Coronavirus Dispersal. Applied Sciences
11(6): 7450.
Ormeño, M. 2014. Estudio de un sistema de limpieza aire-polvo utilizando cámara de remanso y filtración
por agua. Universidad de Santiago de Chile.
Ren, T., Wang, Z. & Cooper, G. 2014. CFD modelling of ventilation and dust flow behaviour above an
underground bin and the design of an innovative dust mitigation system. Tunnelling and Underground
Space Technology 41: 241–254.
Sakar, D. K. 2015. Thermal power plant. Design and Operation, Amsterdam, Netherlands, Elsevier Inc.
Sharma, S. & Judd, W., 1991. Underground opening damage from earthquakes. Clemson, Engineering
Geology.
Villablanca, A. M. 2014. Guía de diseño de soporte de cavernas subterráneas en roca. Universidad de San
tiago de Chile.
Zhang, Heng, et al. 2018. Optimization Scheme for Construction Ventilation in Large-Scale Underground
Oil Storage Caverns. 10, Basilea: MDPI, 2018, Applied Sciences, Vol. 8.
Zhou, Gang, et al. 2020. Dust removal effect of negatively-pressured spraying collector for advancing sup
port in fully mechanized coal mining face: Numerical simulation and engineering application. Amster
dam: Elsevier, January 2020, Tunneling and Underground Space Technology, Vol. 95.
53
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
1 INTRODUCTION
ABB is a leading provider of mine ventilation control technology, in recognition of the need to
optimize mine ventilation given the high energy demands and costs associated with the technol
ogy. This paper describes an exploration to investigate the possibility of leveraging multiphy
sics-based models to take current mine ventilation control technology to the next level. The
modeling methodology specifically investigated here is computational fluid dynamics (CFD)
modeling which has the capability of evaluating time-varying complex fluid flow, gas mixing
and thermal events within three dimensional domains. Understanding such physical complex
ities and application of the same in improvising mine control strategies can be a significant step
in improving ABB’s lead in the competition. In modern terminology such physics-based models
are also called “digital twins”, given their capability of replicating behavour of actual systems.
While ABB, with its rich legacy in control and optimization, has developed state-of-the-art tech
nologies in mine ventilation control, it is ever desirable to continue upgrading offerings keeping in
mind the ever-growing competition and global trends in embracing modern technologies like digi
tal twins, machine learning, artificial intelligence etc. Certain locations in the mine, like freshly
excavated cavities and dead ends, are critical from the safety point of view, given that these are
possible sites of ventilation air starvation and consequently toxic gas accumulation. In the interest
of improving mine safety standards with optimum usage of resources, it is important to accurately
estimate critical parameters like ventilation air velocity, toxic gas concentration and temperature
distribution in critical mine locations where sensors are inaccessible. Therefore, it is desirable to
develop a tool that can perform a thorough physics-based evaluation of these locations evaluating
critical thermo-fluidic parameters with high reliability. CFD modeling technique using finite elem
ent or finite volume discretization methods, is by far the most popular physics-based technique in
resolving complex fluid dynamics, gas mixing and thermal events within enclosures. The technique
can generate three-dimensional holistic maps of the mentioned parameters and if required can per
form time varying analysis. Direct or indirect incorporation of CFD models would enable devel
opment of the next generation intelligent mine ventilation control strategies, ensuring a high
safety standard while catering to growing demands on saving ventilation costs and energy.
DOI: 10.1201/9781003429241-6
54
Several researchers in the past have used CFD model to simulate and study gas dynamics
within mines and investigate technologies to control pollution levels. Kurnia et al. (2014) per
formed CFD modeling studies to understand methane flow dispersion in underground mines and
evaluated a novel flow divertor to focus ventilation air specifically to zones with high methane
concentration for reduction in overall concentration below safety levels. Lihong et al. (2015) devel
oped an experimental test set-up to replicate a mine like enclosure, simulated air flow across the
domain and measured air velocities at selected points. The measurements were used in the present
study to validate the modeling methodology. More recently, Liu et al. (2019) developed CFD
model to simulate mine gas dynamics and investigated the possibility of mitigating pollution
levels, by optimizing the mine ventilation strategy, specifically the ventilation air velocity profiles.
While CFD modeling is a high-fidelity tool in analyzing complex physical events, given the
computational time associated with the models, it is desirable to develop reduced order
models or ROMs, to facilitate integration with control system platforms.
ROMs retains the essential features of a full-scale model and enables simulation of any given
set of inputs, to yield the output in acceptable time. In short, ROM offers the following benefits:
• Optimizing product design using several iterations in less time.
• Offering the opportunity to develop on-line physics-based digital twins.
• Allowing low or mid-level experts to use simulations.
The subsequent sections describe the ROM development procedure starting with development
of the full scale CFD model.
2 METHODOLOGY
Figure 1a. Computational domain replicating a typical mine tunnel with ventilation duct and vehicle as
the source of pollutants.
55
Figure 1b. Discretization of computational domain.
As usual in CFD calculations the mass and momentum conservation equations are first
solved. The mass conservation equation is written as:
where i stands for 1, 2, 3 (all three coordinates). We see that the mass conservation equations
is in terms of the time average velocity. The momentum conservation equation is written as
We see�that the
� momentum equation is not independent of the velocity fluctuations. The
0 0
last term ρui uj , also known as Reynolds stress term (or stress tensor), imposes an additional
stress on the fluid and manifests as the spread of momentum due to turbulence. The k-omega
SST turbulent model (a variety of RANS or Reynold’s average navier stokes equation) was
used to account for turbulence. A y+ value of less than 1 was ensured at all walls during the
simulations, as per the requirements, indicating proper resolution of the boundary layer
physics.
To solve the dispersion and distribution of carbon dioxide and other gases the species trans
port equation is solved.
Where C is the concentration at a given spatial point and time and D is the diffusivity of
carbon dioxide in air.
The CFD model described above, yielded comprehensive details of the velocity and gas con
centration distribution over the three dimensional domain and as a function of time.
Figure 2a shows the distribution of carbon dioxide within the domain. In the current settings
it is encouraging to see that the concentration of carbon dioxide is lower towards the vehicle
end and higher towards the middle of the tunnel and the entry. This ensures minimal exposure
of the gas to personnel manning the vehicle. Figure 2b further shows gas concentration con
tours across transverse planes at the middle of the tunnel, the tunnel entry and end.
56
Figure 2a. Distribution of carbon dioxide within mine tunnel.
Figure 2b. Carbon dioxide concentration distribution across key transverse planes.
Figure 3. Comparing experimentally measured and model computed air velocities within mine-like arti
ficial enclosure. Trends in approximate agreement.
mines is difficult owing to challenges like inaccessibility, errors in measurement devices and
limitations in the number of measurement devices that need to be used. In this paper, experi
mental velocity measurements from a previous publication (Lihong et al. 2015) was used for
validating the modeling methodology. Lihong et al. (2015) developed an artificial mine tunnel
set up or an enclosure, with air flowing through a convoluted tunnel. Air velocity was meas
ured at selected points within the enclosure. In this study, the set up was replicated by devel
oping a CFD model and the model was run using the parameters selected by Lihong et al.
(2015). Figure 3 below shows the comparison between the measured and computed velocities
at the selected points. It is seen that in terms of trends the experimental and computed values
are approximately in agreement. The mismatch between the absolute values is due to several
57
unknown parameters like the uncertainty in sensor location and details of additional features
in the experimental set up. However, it may be concluded that the modeling methodology
used in this study yields results almost close to reality, and is worthwhile to be used for the
subsequent development of the ROM.
Design of experiments or DOE was performed using Optimal Space-Filling Design (OSF)
method. Initially 10 design points (DPs) or variations of input parameters were considered.
These DP’s were simulated using the CFD model. The input parameters were mapped to the
output resulting in generation of the ROM using the genetic algorithm, working behind the
scenes. The output of the ROM is the response surface (Figure 4) which graphically depicts
the relation between the primary output results, in this case mass averaged carbon dioxide
within the tunnel volume and the input parameters, or the ventilation air and vehicle exhaust
velocities.
Figure 4. ROM relating mass averaged CO2 concentration with air and gas velocities.
The accuracy and reliability of the ROM is checked using the goodness of fit procedure the
results of which are shown in Figure 5. The procedure involves comparing the predicted
values from the response surface for some selected points or test points and with the ones
obtained from the full-scale CFD model for the same test points. Square symbols show the
design points used to develop the ROM, while the circular symbols show the additional verifi
cation or test points. It is required that all the verification points follow the regression line to
ensure proper goodness of fit. In this case the RMS error is 1 × 10-3, an acceptable error level.
This established credibility in the ROM as a reliable tool to analyze mine gas dynamics, a tool
that retains the accuracy of the full-scale model, while allowing a quick prediction of gas dis
tribution under a given set of conditions.
58
Figure 5. Goodness of fit procedure results to check accuracy of ROM prediction.
3 CONCLUSION
CFD modeling to predict gas dynamics within mine tunnels, yields useful information that
could guide improvisation of current ventilation protocols. However, a gap in the modeling
method is the high computational times required by the model under any given set of condi
tions, making it difficult to use the model as an online tool for ventilation optimization. How
ever, the response surface generated by the ROM, a result of CFD calculations, relating
exhaust gas concentration to the input parameters can be an effective tool in the online control
of mine pollution if integrated with ABB control platforms. The background of the ROM
being three dimensional high-end CFD calculations, it was seen during our investigations,
that prediction accuracy could be retained within acceptable limits. The ROM approach inves
tigated shows promise to add a new dimension to future ventilation control technology. This
study lays the foundation towards development of a new physics-based analytical technology
for mine ventilation control. As a future step, extensive experimental studies using state of the
art sensor systems in a controlled environment may be performed and the accuracy of the
CFD model can be improved beyond current standards.
REFERENCES
Kurnia, J.C. 2014. CFD simulation of methane dispersion and innovative methane management in
underground mining faces. Applied Mathematical Modeling. 3467–3484.
Lihong, Z. 2015. CFD modeling of methane distribution at a continuous miner face with various curtain
setback distances. Journal of Mining Science and Technology. 635–640.
Liu, H. 2019. A Case Study of an Optimized Intermittent Ventilation Strategy Based on CFD Modeling
and the Concept of FCT. 1–16.
59
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J.K. Shaw
BBE Consulting Canada
M. Kaufman
FLOWCARE Engineering Inc.
ABSTRACT: Vale’s Coleman mine, operating at depths reaching 1850m, could experience
underground summer temperatures >40°Cdb/28°Cwb without cooling. Through 2019-20,
BBE provided EPCM for a 10.3MWR vapour-compression cooling plant that allowed 15%
ambient air mixing. The footprint was restricted on three sides: two rock walls and the existing
surface intake fans/heater facility. The designed proximity of the new system, immediately
upstream, could not impact current performance.
CFD analyses, by Flowcare, were used to ensure the 6m set-back design of eight bulk air
coolers (BACs), 4 wide by 2 high, did not impact flow or pressure of the existing fan/heater,
or cooled air delivery efficiency. Seasonal operation requirements, wind effects, and mitiga
tions were considered. Configurations evaluated included: base case (no cooling plant), per
mutations of the BAC design, and cross winds up to 30km/h.
The CFD analyses showed negligible aerodynamic impacts and minor influence on the
fan’s 2kPa total pressure and 566m3/s flow; but wind effects potentially caused up to 40%
cooling loss. The addition of side-shields and BAC fan deflectors were shown to reduce the
loss to a maximum of 6%. This paper discusses the CFD modeling, incremental design
improvements, and system performance data gathered to date.
1 INTRODUCTION
The shockwaves of the industrial revolution that resulted in the mechanization of under
ground mining have resulted in most underground mines in the Sudbury basin reaching
depths exceeding 1800m. At this depth, auto-compression of air alone will result in work/rest
regimes being implemented during hot summer months, which are detrimental to productivity.
To minimize the impact of hot climatic conditions underground, refrigeration plants are
installed alongside heating plants, which are used to heat the air in the winter. Among these is
Vale’s Coleman mine, located in Levack Ontario.
Vale’s Coleman mine has been in operation for over 50 years and is critical to Vale’s Sud
bury operations. Current mining activity in the 170 OB is at depths of up 1850 m below sur
face which introduces challenges with respect to ventilation supply and acceptable climatic
conditions. Unlike Vale’s Creighton mine, Coleman does not have the benefit of natural cool
ing capacity. Consequently, the mine experiences hot working conditions, especially during
the summer, that have impacted employees and ultimately reduced productive face time.
DOI: 10.1201/9781003429241-7
60
Hot underground conditions result not only from auto-compression, which increase temper
atures by approximately 10°Cdb (dry-bulb) and 5°Cwb (wet-bulb) per every 1000m depth
below surface, but also from fan, equipment, and mobile machinery heat. Strata heat is less of
an issue as the surrounding rock can even moderate temperatures in the summer months.
However, freshly broken rock can represent a significant heat input at the workface.
With summer surface ambient conditions that are often more than 30°Cdb/20°Cwb, under
ground workplace temperatures will be in the range that require heat exposure management
through work-recovery regimens or stop work conditions above 40°Cdb or 32°Cwb. Thus, air
cooling is required to avoid production interruptions.
In the mine planning phase, Vale recognized the need for cooling and engaged BBE in 2016
to investigate novel cooling and traditional mechanical technologies assessing both under
ground and surface options. It was identified that although the cooling aspect was feasible,
providing the required services, electrical power, clean water supply, and locating the heat
rejection without impacting mine activities would be problematic. While focused cooling
requires less refrigeration, capacity challenges with underground infrastructure exist. Alterna
tively, surface cooling plants are more economical per unit of cooling but result in overall
higher cooling capacity as a larger volume of air is cooled. Ultimately, the findings of the
study recommended a surface cooling plant (Hardcastle, 2022).
BBE were contracted to design and construct the plant as a turnkey project in Decem
ber 2018. The initial design included investigating alternative mechanical options and oppor
tunistic coolth storage solutions, with the requirement of an operating turn-key system for the
summer of 2020. In light of the 18-month design-build timeline, the unproven coolth storage
options were discounted, and Vale opted for a 10.3 MWR conventional vapour compression
mechanical system. The project was fast-tracked to deliver as much cooling as possible for the
summer 2020 cooling season, which resulted in several innovative construction practices to
bring the plant online at half capacity in the middle of construction while work continued on
the remaining plant components.
Detailed design established a plant nominal duty of 10.3 MWR based on design input ambi
ent air conditions of 26°Cdb/19°Cwb. Ambient conditions for the mine were defined using
several years of historical climate data available publicly from the “Sudbury A” climate sta
tion (Government of Canada, 2022), as well as ASHRAE publicized data for the same climate
station. The design input ambient conditions align approximately with the yearly 2% cooling
design temperature in the ASHRAE climatic data (ASHRAE, 2017).
Surveying and groundbreaking began in October 2019, with the project adopting an
AGILE “just-in time” engineering approach to maintain project momentum. Primary engin
eering for permitting was complete by January 2020. Half of the cooling duty was provided by
July 2020 while construction continued around the operating plant; the entire plant cooling
capacity became available in September 2022. Full details on the timeline of construction are
available in a recorded format (Shaw, 2022) presented at the MeMO 2022 conference.
The project criteria, timeline, and budget dictated a fit-for-purpose, modular, off-the-shelf
design where pre-engineered and pre-assembled modules were used wherever possible. The
only site-specific structural designs were the Bulk Air Cooler (BAC) module supports and
cooling tower stairs. One of the many challenges to this project, in addition to the short deliv
ery timeline, included fitting the plant into the limited 20-metre by 30-metre space available
directly in front of the mine’s main intake fans without impacting fan performance or airflow
rate, and successful winter operation of the direct-fired natural gas mine air heaters positioned
upstream of the main fans.
The plant design resulted in the placement of the bulk air coolers shown on the right-hand
side of Figure 1. The BACs were situated with a 5-metre open-air gap between the outlet of
the coolers and the inlet wall of the fan house; this was intended to eliminate any airflow dis
ruption or added resistance to the main ventilation fans. This configuration also allows the
fans and heaters to operate in the winter without any added resistance from the BACs and
permits access in front of the fans for maintenance. However, this configuration introduces
the potential that not all of the cooled air leaving the BACs is drawn into the main fans
intake.
61
The BACs are film-packed crossflow modular cooling towers with fans mounted to one side
used to provide the required flow through the tower and to deliver it, across the open-air gap,
to the inside of the intake fan building envelope. The fans are shown on the left side of the
BACs in Figure 1.
There are a total of 8 BACs, arranged into 4 pairs of units, with 2 units installed on each
level as shown in Figure 2. The main fan intake system is designed to handle a wet basis mass
flow 662 kg/s. The BACs cool an engineered flow of 563 kg/s wet basis, resulting in a net
system mass underflow of 101 kg/s, or 15%.
The direct proximity of the plant to the primary mine intake fans was considered through
computational fluid dynamic modelling with the help of FLOWCARE Engineering Inc. utiliz
ing Ansys CFX-18.1 software. The solver high resolution advection scheme was used in con
junction with a standard K-epsilon (k-ϵ) turbulence model to compute steady state solutions
for the model configurations of interest (Kaufman, 2021). The heat transfer models were pri
marily isothermal, however thermal energy modelling was used to consider the effects of buoy
ancy of the cooled air.
The results presented were obtained using a tetrahedral mesh, whose sensitivity was minim
ized using nominal element sizes ranging from 2 to 20 inches, resulting in approximately
1 million nodes (including boundary inflation).
62
The essence of the CFD work was to compare varying configurations and arrangements to
the accepted base case flow. The base case flow considers the previously discussed engineered
mass-underflow with the air cooler delivering 564 kg/s of cooled air while the total mass flow
through the fans is 665 kg/s. The pressures at the main mine fans were used to indicate the
pressure loss of the system. Alternative designs were iteratively evaluated relative to this base
line performance.
Figure 3 confirms that without wind loading, most of the BAC outlet air is received at the
fan intake as indicated by the blue streamlines converging into the heater house. Excavation
for the pump pit, which recesses the pumps and pipework below the collection basins of the
BACs, revealed bedrock at an elevation higher than expected. This resulted in a 1.1m change
in the floor elevation, bringing the entire structure up with it. This introduced airflow impinge
ment on the top lip of the heater house which is later quantified in this analysis.
The mass flow of cooled air present at the main fan outlets for the final design of the cool
ing system was predicted to be 557 kg/s without wind present, and 537 kg/s with a 30 km/hr
crosswind detailed further in this paper. While these values are both lower than the 564 kg/s
design value, BBE accepted the losses considering the access gap required in the design.
The fan pressures predicted by the CFD for the final cooler design were marginally higher
in magnitude when compared with the absence of BACs. The reduction in flow, considered
negligible, was expected with the placement of the bulk air coolers. Fresh air fan operating
63
data over the last two years has validated the modelling, as there have been no perceptible
changes in flow rate, fan pressure, or motor power.
Elevation of the BAC structure by 1.1m, in response to the pump pit geotechnical constraints,
caused secondary losses through impingement of air on the upper lip of the heater intake
building, as shown in Figure 4. To resolve this, various degrees of airflow directionality con
trol were examined – from 18.5 to 10 degrees downwards from the primary flow axis.
Figure 5 presents the CFD predicted flow profile, shown as streamlines, exiting the coolers
and entering the existing heater house for the final design model. Crosswind flows normal to
the BAC airflow were modelled and the air from the top set of coolers was directed downward
at 10 degrees. Figure 5 also includes side walls to control the influence of local wind disturb
ances which are discussed in the next section of this paper.
With the addition of sidewalls and airflow deflection, the mass flow of cooled air at the fan
outlets was predicted to be 537 kg/s relative to the target of 564 kg/s which BBE indicated was
acceptable. Contrasted with the 518 kg/s with sidewalls only, directing the top row of air
cooler outlet flows downwards nets a 19 kg/s (or 3%) overall improvement.
The proposed airflow directionality control comes at a significant price and access penalty
since it would require repurposing and redesigning the existing BAC structure to support the
additional control infrastructure. Repurposing the structure in this way would result in access
restrictions to the front of the fans for maintenance and inspection and would necessitate
access via aerial work platform or the installation of new stair towers.
For these reasons, the addition of airflow deflectors has been postponed until existing
improvements are complete and the effect of enclosing the system on 5 sides can be evaluated.
Considering the engineered 101kg/s mass underflow through the BACs relative to the total
mass flow of 665 kg/s air through the fans, a trade-off study quantifying the losses can be
made empirically by measuring the in-situ system. This trade-off study, which will begin when
the construction of the sidewalls is complete, will evaluate the capital outlay required to justify
the forecast 3% gain in cooled air into the fan station.
64
Figure 5. Final design - Sidewalls and 10 degree decline.
4 WIND LOSSES
CFD analysis indicated that incident wind causes a flow disruption. Annual windspeed and
wind direction were reviewed from the “Sudbury A” climate station and the wind-rose shown
in Figure 6 was produced to weight the output of the CFD analysis relative to the hours
per year at various intensities and directions to focus on the summer months.
The worst-case scenario is a southeast wind which acts normal to the flow of air; this occurs
less than 10% of July. Air arriving from the south also negatively impacts the airflow but
rarely exceeds 30km/hr. The majority of July air arrives from the west and southwest which
benefits this installation as it is blocked by the fan and heater house.
CFD analysis of a south-easterly crossflow normal to the primary axis of airflow for the
system resulted in air losses shown in Figure 7 (155° from N), which assumed a worst-case
windspeed of 30km/hr. This figure shows a large divergence in the flow streamlines indicating
substantial losses to atmosphere. The analysis varied wind directions from 110° to 235° from
N to quantify losses. Table 1 shows a summary of the results from the variation in wind angle,
in terms of the proportion of cooled air arriving at the intake fans.
Based on the results of Table 1, the decision was made to start the engineering and construc
tion of the proposed sidewalls to limit the atmospheric interference on the cooled air leaving
the BACs shown in Figure 8. This would improve the theoretical performance of the cooled
air delivery during heavy wind from 338 kg/s to 518 kg/s.
65
Figure 6. Wind Rose for Sudbury A – July.
Table 1. Mass Flow of Cooled Air at Fan Inlet (Design = 564 kg/s).
Wind Angle No Shields North Shield South Shield Both Shields
kg/s kg/s kg/s kg/s
66
The sidewalls will be in place for the 2023 cooling season, and their efficacy will be quanti
fied by comparing the 2022 cooling performance without the side shields and published in an
upcoming paper. This paper will also attempt to correlate the high windspeeds with high
ambient wet bulb temperatures.
5 WINTER OPERATION
When ambient temperatures fall below the setpoint of the cooling plant, approximately 15°C,
the plant is brought offline. In addition to the CFD performed to confirm the pressure drop
across the main fans, the burner velocity profile was also verified. The burners require
a specific and uniform air velocity across their profile, and analysis indicated this would be
relatively unaffected and uniform with the installation of the bulk air coolers.
To further minimize the pressure drop associated with the BAC infrastructure by allowing
air to bypass the coolers, two louvres were installed in the sidewalls and are designed to be
opened during the winter. The louvres are sliding barn-door type and are shown in light blue
in Figure 8 of the previous section and provide access for equipment to the front of the fans
and burners.
6 CONCLUSION
CFD analysis was used to perform an aerodynamic design review of a new cooling system at
Vale’s Coleman mine in Sudbury for an existing fan intake and burner system. The results of
the CFD were used throughout the design process. The primary objectives were to confirm no
negative impacts to the existing fan and burner system, as well as maximize the amount of
cooled air delivered to the mine.
Crosswinds were determined to significantly impact the volume of cooled air delivered, and
thus it was decided to install side walls to limit interaction with incident wind. The installation
of the sidewalls began in summer of 2022 and their performance/efficacy will be studied
during the 2023 cooling season.
Field conditions discovered during construction required the BAC centerline elevation to be
raised by approximately 1 metre, causing a portion of the cooled air to impinge on the heater
house and be lost. Capitally intensive steps would be required to direct the air from the top set of
67
coolers at an angle of 10° to 18.5° toward the opening of the heater house to achieve an approxi
mate 3% gain in cooled air delivery. The current system with its engineered 101 kg/s mass under
flow will be re-evaluated after the installation of sidewalls to quantify the cost and benefit to
direct airflow downwards towards in the inlet.
The overall design, with the installation of sidewalls, was satisfactory for its intended pur
pose with acceptable losses to the environment, and with minimal impact to the fans and
burners. During winter, CFD-predicted pressure losses were found to be like those for the
summer operation.
REFERENCES
68
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
K.V. Raj
National Intitute for Occupational Safety and Health, Spokane Mining Research Division, Spokane,
WA, USA
Vasu Gangrade
National Intitute for Occupational Safety and Health, Pittsburgh Mining Research Division,
Pittbsburgh, PA, USA
ABSTRACT: Large-opening stone mines are characterized by entry sizes twice or thrice the
size of typical underground coal or metal/nonmetal mines. Due to the large openings, the
volume of air needed to ventilate these mines is significantly higher than in coal or metal/non
metal mines. This leads to low air velocities with low static pressure drop. Airflow in the mine
mostly relies on natural ventilation and auxiliary fans at the working face. The National Insti
tute for Occupational Safety and Health (NIOSH) is conducting research on the ventilation of
large-opening stone mines to reduce worker exposure to dust and other contaminants such as
diesel particulate matter (DPM). To understand airflow in a large-opening mine, we con
ducted a numerical modeling study using computational fluid dynamics (CFD). This paper
presents the results from the CFD modeling of airflow in a large-opening stone mine. The
CFD model is calibrated against the data previously collected at a large-opening stone mine.
The team ran different scenarios with fan placement along with the movement of truck to
simulate how effectively air is moving in the mine.
1 INTRODUCTION
Stone mines produce a wide range of raw material such as basalt, granite, limestone, marble, etc.
for the construction industry needed for infrastructure development. As of 2020, there are 4,248
stone mining operations with 110 underground operations in the United States (NMA, 2020). The
majority of the underground stone mines contain broken limestones according to the standard
industrial classification (SIC) system. These underground limestone mines are characterized by
openings which are larger than the openings compared to coal and metal/nonmetal mines. The
mining method for large-opening limestone mines is room and pillar which is similar to coal
mines. However, the dimensions of large-opening limestone mines are twice or thrice of the typical
coal mines. This larger dimension of the limestone mines creates a unique challenge of ventilation
in the mines. Due to these large dimensions, airflow in the mine is affected by natural ventilation
with little differential pressure change. There are limited studies done to address this unique venti
lation issue related to large-opening limestone mines.
In the early 2000s, the National Institute for Occupational Safety and Health (NIOSH)
started conducting research to improve air quality by reducing contaminants such as diesel
particulate matter (DPM) in large-opening mines. These studies looked into ventilation, fan
selection, stoppings, and mine planning to reduce the contaminants level in large-opening
mines (Grau III et al., 2002). The research suggested that large-opening mines face three pri
mary ventilation challenges: moving adequate volumes of ventilation air, controlling and
DOI: 10.1201/9781003429241-8
69
directing the airflow, and planning ventilation systems that work well with production require
ments (Grau III et al., 2006). Multiple studies were also conducted that showed increased suit
ability of propellor fans for low-resistance ventilation systems present in stone mines compared
to the vane axial fans commonly applied in coal mines (Grau III et al., 2004, Krog et al., 2004).
Studies on the effectiveness of different large-opening stone mine ventilation stoppings showed
steel stoppings were effective for permanent stopping locations, fabric stoppings were more
effective closer to blasting areas, and long stone pillars could perform the same role as a row of
stoppings (Grau III and Meighen, 2006, Grau III and Krog, 2009). Recently, researchers have
conducted a field study of large-opening underground mines looking at variations in pressure,
relative humidity, and temperature over time (Gendrue et al., 2021). Past research, however, has
not looked into numerical techniques to analyze the airflow patterns in large-opening mines.
A numerical technique such as computational fluid dynamics (CFD) has been used for solving
airflow problems in mine ventilation for quite some time. CFD modeling has been widely used
in many areas of airflow and particulate and gaseous contaminant modeling from underground
mine ventilation to open-pit ventilation (Raj et al., 2015, Raj and Bandopadhyay, 2017, Kumar
et al., 2021, Bhargava et al., 2021, Morla et al., 2021). Most recently, Watkins and Gangrade
conducted a study using ANSYS Fluent to optimize the auxiliary fan placement in a large-
opening stone mine (Watkins and Gangrade, 2022). There are many commercially off-the-shelf
and open-source CFD software programs, such as ANSYS Fluent, Cradle CFD, COMSOL
Multiphysics, and OpenFOAM, that are available and being used to solve airflow problems in
the mining industry. The majority of these CFD programs are based on Navier-Stokes equa
tions, the energy equation, the mass conversion, and transport equations. CFD models have the
potential to provide a pattern of airflow in large-opening underground stone mines.
This study will present a two-dimensional modeling of a large-opening stone mine using the
COMSOL Multiphysics® CFD modeling program. This paper focuses on airflow patterns
with and without a fan in the model domain and how the airflow pattern changes with move
ment of a truck in the model.
The geometry for this work was adopted from Grau III and Krog (2009) work to simulate the
airflow inside the mine with four scenarios. Figure 1 shows the model geometry. The entry of
the mine is 50 ft, and dimension of the pillars are 50 ft by 50 ft. The extent of the model in the
x-direction is 1,450 ft, and in the y-direction it is 1,000 ft. In the first scenario, we simulated
the airflow inside the mine without any mechanical ventilation such as a fan, then followed
the second scenario by adding a fan to the model. Grau III and Krog (2009) mentioned in
their work that truck movement inside the mine has an impact on the ventilation system. In
the third scenario, we simulated the movement of the truck along and against the airflow.
Finally, the fourth scenario presents the simulation of the airflow with movement of the truck
as well as presence of the fan in the model.
70
Figure 1. Geometry of the model domain (Grau III and Krog (2009)).
Figure 2. An expanded view of the mesh size (in m) near interior walls in the model geometry.
based on work of Grau III and Krog (2009) and outlet boundary condition at zero pressure.
Velocity profile along with streamlines are shown in Figure 3. It can be seen from the figure
that most of the air comes from the inlet boundary to passing in a straight path through
Return 1 and goes out without much movement in the top right corner of the model. Even
though the velocity streamlines suggest some movement and recirculation of air near the
Return 2 side of the geometry, the velocity is relatively lower compared to velocity at
Return 1.
2.2 Scenario 2: Model with fan in working area near Return 2 and Return 1
In this scenario, a fan was placed in the working area of the model near Return 2 to move the
air coming from the inlet to the top right area of the model. Figure 4 shows the placement of
the fan in the model geometry. The fan in the model represents an 8-ft diameter propeller fan
71
Figure 3. Velocity profile along with streamlines of the base airflow model.
with capacity of 55 m3/s at 0.127 kPa. Being a 2D model, model input for the fan was the
speed instead of volumetric flow rate. Both side of the fan was treated as inlet boundary con
dition with positive and negative fan speeds. Fan speed was 12 m/s based on the data collected
for the propeller fan. Figure 5 shows the mesh elements around the fan as very fine mesh is
needed around the fan geometry to resolve the flow. However, the inlet velocity was set to
4.8 m/s as per the work of (Grau III and Krog 2009). Figure 6 shows the velocity profile and
streamlines for the steady-state simulation. It can be seen from the figure that similar to the
model with no fan most of the air is following the path from the inlet to outlet, and there is air
movement in the top near Return 2 due to the fan. However, most of the air is being recircu
lated rather than going out through Return 2. This means that fresh air from the inlet is not
being utilized by the fan, and due to recirculation contaminants like diesel particulate matter
(DPM), dust and other engine exhaust concentrations might increase over time. Grau III et al.
Figure 4. Model geometry with fan (in red circle) in the working area near Return 2.
72
Figure 5. Mesh elements around the fan in the model.
Figure 6. Velocity profile along with streamlines of the model with fan near Return 2.
(2002) suggested that fan placement affect ventilation in large-opening mines. To this end, the
fan position was changed from working area near Return 2 to Return 1. This was done to
push fresh air coming from inlet toward the working area. Grau III and Krog (2009) in their
study used a 36-in diameter axial-vane fan with a 23-in reducer mounted at the outlet.
In the model with the fan near Return 1, the inlet velocity was 2.4 m/s with the fan speed of
19.4 m/s based on the data (Figure 6). Figure 7 shows the geometry with a fan in the model
domain. Figure 8 presents the velocity profile and streamlines of the simulation. It can be
inferred from the figure that this fan placement worked as suggested by Grau III and Krog
(2009). There is a clear indication that fresh air from the inlet is being pushed towards the
working area at the top of the model and that there is no direct path of air from the inlet to
outlet through Return 1. This scenario will lead to the removal of contaminants from the
working area through Return 2.
Figure 9 shows the velocity profiles and average velocities along the cross section taken in
the middle of Return 2 for the above models. The model with no fan has an inverted
73
Figure 7. Model geometry with fan (in red circle) in the working area near Return 1.
U-shaped velocity profile with a maximum velocity of 1.8 m/s reaching in the middle. The
model with an 8-ft diameter fan has the maximum velocity of 3.6 m/s between the three
models but a decline over the length of the Return 2 cross section. The model with a 36-in
diameter fan has the maximum velocity of around 2 m/s. However, if we look at the average
Figure 8. Velocity profile and streamlines of the model with fan near Return 1.
velocities of the three models, it was found that the model with a 36-in diameter fan has an
average velocity of 1.4 m/s followed by a model with no fan. The average velocity for the
model with no fan was 1.3 m/s, whereas the model with an 8-ft diameter fan had an average
velocity of 1.1 m/s. This shows that with an 8-ft diameter fan in the model domain, most of
the air was recirculating rather than passing through the Return 2. Grau and others also sug
gested that movement of a truck in the mine can affect the airflow in the mine. Modeling of
the movement of a truck along and against the airflow will be discussed in the following
sections.
74
Figure 9. Velocity profiles along the cross section of Return 2.
2.3 Scenario 3: Modeling with truck movement in the mine without fan
As mentioned in the previous section, movement of a truck has an impact on the airflow move
ment in the mine, and in order to simulate that we are first modeling the movement of a truck
without any fan in the model domain. Trucks in the mine move from working area towards the
entry and from an entry towards the working area of the mine. When the truck moves towards
the entry, it moves against the airflow whereas the truck moves along the airflow when it moves
towards Return 1. Figure 10 presents the model geometry with a truck near the Return 1. For the
first model, the truck will move from Return 1 towards the inlet against the airflow. The truck is
represented by a rectangle with a dimension of 30 × 12 ft. An expanded view of the truck can be
seen in Figure 11. The truck speed for the modeling was assumed to be 9 m/s, which is 20 miles
per hour. The important part of modeling a moving body in an airflow simulation is the domain
in which the truck will move. For modeling truck movement in both directions, the moving
domain was defined as a rectangle of a dimension of 1,250 × 30 ft as shown in Figure 10.
75
Modeling a moving body in CFD is challenging as the model needs to solve for both airflow and
movement of the body in the fluid domain. Apart from that, the most critical aspect is the meshing
of the moving domain as well as the rest of the geometry. Mesh elements for the geometry other
than in the moving domain and around the truck were triangular, whereas the mesh elements in the
moving domain were mapped, forming rectangular mesh elements. Figure 12 shows expanded view
of the mesh size distribution along with mesh elements in the model domain. Mesh elements around
the truck and moving domain is also visible in Figure 12. The model was run for 38 seconds during
this time when the truck moved from the Return 1 to the other side of the moving domain.
Figure 13 shows the velocity profile and streamlines at time t = 0. It can be seen from the figure that
there is air moving towards the working area near Return 2. Figure 14 shows the average airflow
passing through Return 2 over the entire simulation run. As the truck passes through Return 1,
average airflow velocity in Return 2 is increased by around 50% and starts to lower over time. This
shows that there is positive impact of movement of a truck against the airflow.
Figure 12. An expanded view of the mesh elements of the moving domain.
76
Similarly, the modeling of the truck moving from the inlet towards the Return 1 was con
ducted. In this scenario, the model was run for 39 seconds, and the velocity profile along with
streamlines are presented in Figure 15. Figure 15 shows that air is moving towards Return 2
at a low velocity. The average air velocity passing through Return 2 over time is presented in
Figure 16. It can be seen from Figure 16 that there is no positive impact on the air velocity at
Return 2 as seen in the previous scenario; instead, we see the opposite—that the average air
velocity decreases over time.
Figure 13. Velocity profile and streamlines at initial time step with truck moving towards the inlet.
Figure 14. Average air velocity passing through Return 2 over time with truck moving towards the inlet.
2.4 Scenario 4: Modeling with truck movement in the mine with fan
In this scenario, a combination of the above two scenarios were modeled: the truck’s movement
and a fan running near Return 1 (Figure 7), which is a combination of the above two scenarios.
Again, modeling was conducted considering the truck movement towards and away from the
inlet. Figure 17 shows the velocity profile and streamlines at the initial time step. It can be seen
77
from the figure that the fan is pushing fresh air towards the working area near Return 2.
Figure 18 shows the average air velocity passing through Return 2 over time. The profile looks
similar to the model where no fan was present; however, due to fan, the average air velocity is
higher. The decrease in the air velocity was observed after the truck passed the fan. Figure 19 pre
sents the velocity profile and streamlines when the truck is crossing paths with the fan flow. This
time is around 10 seconds. The truck movement disrupted the flow of the fan towards the working
area, restricting air coming from the inlet. This led to a decrease in the airflow at Return 2.
Figure 15. Velocity profile and streamlines at initial time step with truck moving away from the inlet.
Figure 16. Average air velocity passing through Return 2 over time with truck moving away from the inlet.
In the second model the truck was moving away from the inlet with a fan running near
Return 1. All the model settings were the same as in the scenario with truck moving and no fan
in the model domain. This model was run for 39 seconds. Figure 20 shows the velocity profile
and streamlines at initial time step. It can be seen from the figure that the flow profile near the
main entry and the truck looks similar to the velocity profile when there is no fan in the model
(Figure 15). The main change in this case can be seen as due to running of the fan near Return
78
1, which is pushing the fresh air towards the working area with the air passing through Return
2. This will help in the removal of pollutants generated at the working area. The model ran
without an issue until it crossed paths with the fan. The model crashed when the truck crossed
the path with a fan just after 27 seconds. Currently, this issue has not been resolved. In order to
look at the airflow with movement of the truck, the model was run by changing the position of
Figure 17. Velocity profile and streamlines at initial time step with fan and truck moving towards the inlet.
the truck just past the fan to see how the airflow changes. Figure 21 shows the velocity profile
and streamlines at t = 27 seconds. Figure 22 presents the average air velocity passing through
Return 2 over time. There is a brake in the line plot at 27 seconds to show the discontinuity
due to the model crash. However, it can be seen that after restarting the model with a truck
positioned just after the fan that it follows a similar trend as in Figure 15. The main difference
found out to be the average flow velocity due to the operating fan.
Figure 18. Average air velocity passing through Return 2 over time with fan near Return 1 and truck
moving towards the inlet.
79
Figure 19. Velocity profile and streamlines at t= 10 seconds with fan and truck moving towards the inlet.
Figure 20. Velocity profile and streamlines at initial time step with fan and truck moving towards the inlet.
3 SUMMARY
The ventilation of large-opening underground stone mines is challenging due to the larger
dimension of the mines. Ventilation issues can be resolved by use of the proper placement of
a fan in combination with a truck moving underground. We considered different scenarios
such as fan placement in the mine, movement of a truck, and a combination of fan and truck
movement in the mine to model. The CFD modeling results presented in this paper show that
placement of a fan can both positively and negatively affect the ventilation of fresh air near the
working area of the mine. Simulation results from movement of a truck has also some impact
on the airflow pattern in the mine affecting the ventilation; however, the movement of a truck
against the airflow affects the airflow more than the truck moving in the same direction of the
airflow. Modeling results from the combination of a fan and truck movement yielded similar
result as with movement of a truck alone, and the only difference was the magnitude of the
velocity due to the presence of a fan in the model domain. Finally, these are only the 2D
80
Figure 21. Velocity profile and streamlines at t = 27 seconds with fan and truck moving away from the
inlet.
Figure 22. Average air velocity passing through Return 2 over time with fan near Return 1 and truck
moving away from the inlet.
modeling results, which does not provide the airflow pattern in the third dimension, and a 3D
modeling is needed to fully understand the airflow pattern in large-opening stone mines.
DISCLAIMER
The findings and conclusions in this report are those of the author(s) and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health,
Centers for Disease Control and Prevention. Mention of any company or product does not
constitute endorsement by NIOSH.
81
REFERENCES
Bhargava, R., Tukkaraja, P., Adhikari, A., Sridharan, S. J. & Vytla, V. V. S. (2021) Airflow characteristic
curves for a mature block cave mine. In Tukkaraja, P. (Ed. Proceedings of the 18th North American
Mine Ventilation Symposium. Rapid City, SD, CRC Press.
Gendrue, N., Liu, S., Bhattacharyya, S. & Spellman, C. (2021) Field survey of mine ventilation system
for large opening underground mines: Pressure, relative humidity, and temperature. In Tukkaraja, P.
(Ed. Proceedings of the 18th North American Mine Ventilation Symposium. Rapid City, SD, CRC
Press.
Grau III, R. H. & Krog, R. B. (2009) Ventilating large opening mines. Journal of the Mine Ventilation
Society of South Africa, 62, 8–14.
Grau III, R. H., Krog, R. B. & Robertson, S. B. (2006) Maximizing The Ventilation Of Large-Opening
Mines. In JM, M. & RV., R. (Eds.) Proceedings of the 11th U.S./North American Mine Ventilation
Symposium. University Park, PA, Taylor & Francis Group.
Grau III, R. H. & Meighen, G. M. (2006) Novel stopping designs for large-opening metal/nonmetal
mines. In JM, M. & RV., R. (Eds.) Proceedings of 11th U.S./North American Mine Ventilation Sympo
sium. University Park, PA, Taylor & Francis Group.
Grau III, R. H., Robertson, S. B., Krog, R. B., Chekan, G. J. & Mucho, T. P. (2004) Raising the bar of
ventilation for large-opening stone mines. In Bandopadhyay, S. & Ganguli, R. (Eds.) Proceedings of
the 10th US/North American Mine Ventilation System. Anchorage, AK.
Grau III, R. H., Robertson, S. B., Mucho, T. P., Garcia, F. & Smith, A. C. (2002) NIOSH ventilation
research addressing diesel emissions and other air quality issues in nonmetal mines. SME Annual
Meeting. Phoenix, AZ.
Krog, R. B., Grau III, R. H., Mucho, T. P. & Robertson, S. B. (2004) Ventilation planning layouts for
large opening mines. SME Annual Meeting. Denver, CO.
Kumar, A. R., Henderson, K. M. & Schafrik, S. (2021) Scale modeling, PIV, and LES of blowing type
airflow in a deep cut continuous coal mining section. In Tukkaraja, P. (Ed. Proceedings of the 18th
North American Mine Ventilation Symposium. Rapid City, SD, CRC Press.
Morla, R., Karekal, S. & Godbole, A. (2021) Transient-flow modelling of DPM dispersion in unventil
ated dead-end crosscuts and control strategy using curtain. In Tukkaraja, P. (Ed. Proceedings of the
18th North American Mine Ventilation Symposium. Rapid City, SD, CRC Press.
NMA (2020) Mine Safety & Health Administration’s Number of Coal and Non-fuel Mineral Operations
in the U.S.
Raj, K. V. & Bandopadhyay, S. (2017) CFD Modeling Of Cloud Cover for Pollutants Dispersion in
Deep Open-Pit Mines Under Arctic Air Inversion. SME Annual Meeting. Denver, CO, SME.
Raj, K. V., Bandopadhyay, S. & Ramani, R. V. (2015) Turbulent Models for Pollutant Transport in
Openpit Mines under Stable Boundary Layer. Transactions of Society for Mining, Metallurgy, and
Exploration, Inc, 338, 476–486.
Watkins, E. & Gangrade, V. (2022) Optimization of auxiliary fan placement for large-opening under
ground stone mines. SME Annual Meeting. Salt Lake City, UT.
82
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Shock loss occurs when there is a change in the area of a drift or airflow direc
tion. These shock losses accumulate throughout the ventilation system and will increase the
mine resistance as underground development deepens or becomes more complex. Hence, shock
losses should be carefully considered when establishing the resistance through a series of
branches as they can be a significant contributor to the estimated mine resistance. Using com
puter software to simulate a ventilation system is common but applying reasonable shock loss
factors in the software is not a simple or direct process. Determining shock loss values includes
the consideration of historical measurements, application of factors and formulas from litera
ture based on the geometry and complexity of junctions, and the geometry variation at different
locations. For example, the shock loss at the bottom of a ventilation raise can be comprised of
a 90° bend and a sudden area change. This paper presents an approach for estimating shock
loss factors near raise junctions using computational fluid dynamics. Current and future work
will also be compared with field measurements. Results from this approach are compared with
software defaults as well as classical techniques in published literature and standards.
1 INTRODUCTION
The ventilation system is one of the most critical systems in underground mines. It provides
fresh air to personnel and clears contaminants out of mines. However, the efficacy and efficiency
of the ventilation system can be significantly impacted by the flow phenomena and energy losses
at junctions where there are changes in geometry or airflow direction. As these “shock losses”
accumulate through a ventilation system, they can impart a negative impact on fan pressure
requirements and energy consumption, as well reduced air quality and potential safety concerns.
Researchers (Jade & Sastry, 2008; Tien, 1987; Purushotham et al., 2010) have conducted labora
tory studies to estimate shock loss factors and explore the factors (e.g., geometry, wall roughness)
that contribute to shock losses. Although there is plenty of field data (e.g., pressure data collected
using trailing hose and manometer) in published literature, it does not contain accompanying
information about the junction geometry. Therefore, one cannot easily establish a catalogue of
shock losses for different cases. The motivation to take these measurements comes from available
textbooks (e.g., (McPherson, 2009)) and publications that do not have practically relevant shock
loss factors considering the combined effect from turns and sudden expansion or sudden contrac
tion. However, in operating mines, ventilation engineers and specialists encounter these questions
regularly. Hence, it is necessary to collect field data and calibrate models (e.g., computational fluid
dynamics (CFD) models, network models) to better predict shock loss factors and consequently
achieve higher fidelity ventilation designs and fan selections.
This paper presents two experiments conducted in two operating underground hard-rock
mines in Canada. This paper also presents a proposed method for isolating the shock loss
from overall pressure loss data in a CFD analysis. Three-dimensional (3D) CFD models were
created to represent the geometry of the experiment locations, and the CFD results are
DOI: 10.1201/9781003429241-9
83
compared with the field measurements. The findings of this paper will be of interest to mine
ventilation engineers and researchers who seek to conduct optimization and improve the effi
ciency of underground mine ventilation systems.
2 EXPERIMENT DETAILS
Two sets of field measurements were conducted at two different mines. The data from each
mine is summarized below.
2.1 Experiment 1
Experiment 1 was conducted in a fresh air transfer network as shown in Figure 1(a) and 1(b). Two
horizontal drifts are connected by a 4.27 m (14 ft)-diameter, 288 m (945 ft)-long unsupported
round raise. The intent of the experiment was to record trailing hose pressure readings every
30.48 m (100 ft) down the raise to determine the raise inlet and outlet shock loss and the friction
loss in the raise. Data was collected using 365.76 m (1200 ft) of 0.0048 m (3/16 inch) inside diam
eter vinyl tubing and a Model 5825 (TSI Incorporated, 2022) DP-Calc MicroManometer. Unfor
tunately, the hose hung up within the first ~60.96 m (200 ft) rendering the efforts unsuccessful.
Additional data collected included: temperature, relative humidity, and barometric pressure (using
a Kestrel environmental meter (Nielsen-Kellerman Co., 2020)) in both the upper and lower trans
fer drifts, refer to Table 1. The flow rate through the fresh air transfer was established with a pitot
tube survey of the two booster fans at the bottom level; the result was 216.2 m3/s (460 kcfm). The
pitot tube survey data are not presented in this paper due to the paper’s length limitation. Two
CFD models were created for Experiment 1 to investigate the shock loss that occurs entering and
exiting the raise. The classical calculation of the shock loss based on the field data is shown in
Table 2.
84
Table 2. The classical calculation to subtract raise friction loss and arrive at an estimated value of the
combined shock loss (i.e., 90° corner and sudden contraction) into the raise as shown in Figure 1(a).
Length of hose in the Measured pressure Friction loss1 Shock loss2 Velocity Shock
raise (m) loss (Pa) (Pa) (Pa) pressure3 (Pa) factor4
Notes:
1
The friction loss was calculated using a friction factor of 0.0053 kg/m3 at local air density (McPherson,
2009) based on Equation (1).
2
The shock loss is the difference between the measured pressure loss and the friction loss
3
Velocity pressure was calculated using velocity at the inlet (Figure 1(a)) based on Equation (2).
4
Shock factor is calculated using shock loss divided by velocity pressure (or using Equation (3)).
The friction pressure drop is calculated using Atkinson’s Equation (Atkinson, 1854).
Where P is the friction pressure drop, k1.2 is the Atkinson friction factor at standard air
density (1.2 kg/m3), L is the length of a given airway, per is the perimeter, A is the cross-
sectional area, Q is air volume, ρ is local air density.
The velocity pressure is calculated using Equation (2) (McPherson, 2009).
Where Pshock is the pressure loss caused by shock losses, X is the shock loss factor.
Based on Table 2, the average shock factor referred to the inlet velocity pressure is 1.70,
a combined factor considering sudden contraction and a sharp turn.
2.2 Experiment 2
Experiment 2 was conducted on two levels, as shown in Figure 2. There are two 1.57 m (5.17 ft)-
diameter, 41.16 m (135 ft) long, unsupported round raises between the levels. A 60.96 m (200 ft)-
long trailing hose was released from the upper level in both raises. Efforts to characterize internal
raise losses using trailing hose measurements were successful in this experiment because of the
shorter raise length and better planning based on the lessons learned from Experiment 1. Meas
ured temperature, relative humidity, and barometric pressure data are presented in Table 3. The
traverse method with a Testo 417 vane anemometer (Testo SE & Co. KGaA, 2023) was used to
measure the average air velocity in the upper and lower drifts. Due to safety constraints associated
with working near open holes, individual raise flow rates were not measured; the measured total
flow has been assumed to split evenly into the two raises. Air velocity in the drifts and raises was
relatively low at ~1.5 m/s and ~6 m/s, respectively. Air volume measurements in Experiment 2 are
shown in Table 4. Pressure data collected in the field for raise 1 and raise 2 are shown in Tables 5
and 7, respectively. The calculated shock and friction losses for raise 1 and raise 2 are shown in
Tables 6 and 8, respectively. One CFD model was created for Experiment 2 to investigate the over
all shock loss that occurred in the ventilation system.
85
Figure 2. 3D view of Experiment 2 (dimensions are in meters).
Table 5. Pressure loss measurements taken and calculated for raise 1 in Experiment 2.
Differential static Calculated Calculated Calculated
pressure (measured in velocity pressure2 static pressure3 total pressure4
Sample location the field1) (Pa) (Pa) (Pa) (Pa)
Notes:
1
Each sample was taken every 6.1 m (20 ft). Each sample data represents a 10-second average pressure
reading. A hose was dropped from the upper level to the lower level and retrieved from the lower level.
2
The velocity pressure is calculated based on 11.61 m3/s of air in the raise. It is assumed the raise has
a 3% overbreak (1.57 m in diameter as shown in Figure 2 while the design diameter is 1.52 m).
3
The static pressure is calculated based on the differential static pressure in column 2.
4
The total pressure is calculated as the sum of the velocity pressure and the static pressure.
5
This measurement was taken when the hose first entered into the raise from the lower level.
6
These total pressure data are outliers because they are greater than the upper drift total pressure.
Hence, they are not used to calculate the results shown in Table 6.
86
Table 6. Pressure loss results calculated for raise 1 based on the field data shown in Table 5.
Inlet shock loss (Pa) Raise friction loss (Pa) Outlet shock loss (Pa)
1 2
5.47 4.19 26.153
Notes:
1
5.47 = 33.55-28.08. This result represents the inlet shock loss and a small friction loss (along
24.4 m in the raise) because of the field data outliers collected from 6.1 m to 18.3 m in the
raise.
2
4.19 = 43.95x(28.08-26.33)/(42.7-24.4)
3
26.15 = 26.33 – 0.18
Table 7. Pressure loss measurements taken and calculated for raise 2 in Experiment 2.
Differential static Calculated Calculated Calculated
pressure (measured in velocity pressure static pressure total pressure
Sample location the field) (Pa) (Pa) (Pa) (Pa)
Notes:
1
This measurement was taken when the hose first entered into the raise from the lower level.
Table 8. Pressure loss results calculated for raise 2 based on the field data shown in Table 7.
Inlet shock loss (Pa) Raise friction loss (Pa) Outlet shock loss (Pa)
Notes:
1
0.74 = 33.8-33.06
2
5.73 = 33.06-27.33
3
27.9 = 28.08 – 0.18
Based on classical shock factor equations for sudden contraction and sudden enlargement
(McPherson, 2009), the shock factors for the upper bend (from the upper drift to the raise) and
the lower bend (from the raise to the lower drift) are calculated to be 0.42 and 0.84, respectively.
The shock losses for the upper and lower bends are calculated to be 9.95 Pa and 19.89 Pa, respect
ively. Compared to Tables 6 and 8, the lower bend shock losses (26.15 Pa and 27.9 Pa) calculated
based on the field pressure data are consistently higher than that (19.89 Pa) calculated using the
classical equations. In contrast, the upper bend shock losses (5.47 Pa and 0.74 Pa) calculated
based on the field pressure data are less than that (9.95 Pa) calculated using the classical equations.
It is apparent that the difference between the pressure losses calculated from the field data and the
classical equations is less at the lower bend than at the upper bend. The low air velocity in the
drift, actual raise diameter (e.g., overbreak), measurement uncertainty in collecting the field data
all contribute to the mentioned differences.
3D CFD models were created for each experiment matching the geometry and configuration
shown in Figures 1 and 2.
87
3.1 Assumptions
The geometry for the two experiments was simplified, including no arch on the back, no over
break on walls and back, and idealized overbreak (i.e. smooth fillets) at raise intersections. In
addition, uniform flow at the inlet was applied.
3.3 A CFD modelling approach for determining the shock loss and shock loss factor
One modelling approach to establish CFD estimated pressure losses associated with a shock
loss is to use three individual CFD models. For example, Figure 3 illustrates the CFD models
used to estimate a sudden expansion shock loss factor.
Figure 3. CFD models (plan view) required to estimate a sudden expansion shock loss factor.
In Figure 3, Model 1 and Model 2 are used to establish the pressure loss due to friction in
each geometry segment. Model 3 establishes the total pressure loss through the entire geom
etry; the shock loss associated with the expansion is determined using Equation (4).
Where dPshock is the pressure drop caused by shock loss, dP3 is the total pressure difference
between the inlet and outlet in Model 3 in Figure 3, dP1 and dP2 are the total pressure differ
ence between the inlet and outlet in Model 1 and Model 2, respectively.
88
4 RESULTS
CFD results for experiments 1 and 2 are shown in Tables 9 and 10, respectively. Pressure
result comparisons between the field data and the CFD model are shown in Table 11.
Figure 1(a) Both raise and drift 241.59 149.90 91.69 65.60
Raise only 176.36 150.28 26.09 26.09
Figure 1(b) Both raise and drift 206.56 11.01 195.54 167.67
Drift only 11.64 10.08 1.57 1.79
Raise only 176.36 150.28 26.09 26.09
Notes:
1
Area Average total pressure at the inlet based on the CFD model
2
Area Average total pressure at the outlet based on the CFD model
3
Pressure difference between the inlet and outlet
4
Friction loss is the total loss along the drift or raise
5
Shock loss is calculated using Equation (4)
Notes:
1
The friction loss in the raise is calculated based on the linear relationship between the fully-developed
airflow region (velocity contours apart from each other by 6.1 m shown in Figure 4) and the average
length of the two raises (42.94 m). So 14.80 is the result of 42.94x2.10/6.10.
Figure 4. Contour of velocity magnitude in the 1.57 m (5.17 ft)-diameter raise in Experiment 2.
The CFD results are all less than those measured in the field for Experiment 1. A reason for
the considerable disagreement between the results is the field data is point data, whereas the
CFD results are the area-average pressure at the monitoring planes. In addition, it is import
ant to pay attention when comparing the CFD-calculated pressure loss with the field-
measured value. The CFD results are useful to offer a relative measure of the geometry
change impacts, but it worth noting that the geometry modelled in the CFD model is not
89
Table 11. Pressure result comparisons (between CFD and field data) for experiments 1 and 2.
Experiment 1 (Figure 1(a)) Experiment 2 Raise 2 (Table 7)
Field data collected at Difference Field Difference
Pressure CFD results 60.96 m below the collar (Pa) CFD results data (Pa)
identical (e.g., arched drifts, uneven level) to the actual geometry and the numerical models
used in CFD representing the complex physics of turbulent airflow can also contribute to dis
crepancies between the results in the CFD and field measurements. As well, the raise condition
(e.g., smooth wall, broken wall) is unknown, and the hose continuously moved because of the
high air velocity in the raise while the pressure was taken. These difficulties in taking accurate
pressure readings affected the field data accuracy.
From Table 11, the CFD total pressure results agree with that measured in the field for
Experiment 2. Similar to Experiment 1, the shock loss result from the CFD model is less than
that from the field. The field data indicates less friction loss in the raise compared to that in
the CFD model.
In this paper, two field experiments were conducted at two different mine sites. The two
experiments used the trailing hose method to measure the pressure loss. An average shock
factor referred to the inlet is 1.70 in Experiment 1. Results from the CFD model of Experi
ment 1 do not agree well with the field data. Efforts to characterize internal raise losses using
trailing hose measurements were unsuccessful in Experiment 1 because the hose hung up in
the raise. However, it was successful in Experiment 2. There is a good agreement on the total
pressure loss between the results from the CFD model and the field data for Experiment 2
despite more complexity in the geometry of Experiment 2.
The authors identified the following factors that contribute to the total pressure losses:
• The overbreak of the raises during construction
• The connection type (e.g., sharp 90° turn, fillet) between the raise and horizontal drift
• Accuracy of the instrument (e.g., vane anemometers, manometers)
• Methods of air velocity measurement (e.g., fixed points, traverse)
• Wall roughness height
The authors also found that the Ansys CFX and Fluent solvers could generate differ
ent results based on a preliminary study on the total pressure losses obtained with the
same mesh, near identical solver settings (understanding that some settings are fundamen
tally different). A literature review (Acharya, 2016) has identified discrepancies between
Fluent and CFX results even when the model input parameters are as closely matched as
possible between the two solvers. Ongoing research into this topic and how it may
impact the use of CFD methods to estimate shock losses in mine ventilation systems will
be conducted by the authors.
In the future, a mesh independence study with various mesh elements or nodes will be con
ducted to ensure the CFD results are independent of the mesh quantity and quality. In add
ition, more field data will be collected at geometrically preferred (e.g., one raise between the
two levels instead of two raises) locations at different mines.
90
ACKNOWLEDGEMENTS
The authors want to thank Vale Canada Ltd. for the support of this work. The authors are
also grateful to their colleagues Fletcher Friesen, Paul Aho, Daniel Caton, Allen Jonas, Hugo
Ferrari, Stefano Girardo, Scott MacMillan, Peter McGirr, Brandan Randell, Deon Jenkins,
and Matthew Cheeseman for helping set up the field experiments and take measurements.
REFERENCES
Acharya, R. 2016. Investigation of differences in Ansys solvers CFX and fluent. Royal Institute of Technol
ogy, KTH, Stockholm, Sweden.
Ansys Inc. 2021. Discovery SpaceClaim. Canonsburg, PA, USA.
ANSYS Inc. 2022. Ansys CFD-Post User’s Guide. Canonsburg, PA, USA.
ANSYS Inc. 2009. ANSYS CFX-Solver Theory Guide. Canonsburg, PA, USA.
ANSYS Inc. 2010. ANSYS Meshing User’s Guide. Canonsburg, PA, USA.
Atkinson, J.J. 1854. On the Theory of the Ventilation of Mines. North of England Institute of Mining
Engineers, 3: 118.
Jade, R.K. & Sastry, B.S. 2008. An experimental and numerical study of two-way splits and junctions in
mine airways. In 12th US/North American mine ventilation symposium. Reno, Nevada, USA.
McPherson, M.J. 2009. Subsurface Ventilation and Environmental Engineering. Fresno, CA, USA: Mine
Ventilation Services, Inc.
Purushotham, T., Sastry, B.S. & Samanta, B. 2010. Estimation of shock loss factors at shaft bottom junc
tion using computational fluid dynamics and scale model studies. CIM Journal, 1(2): 130–139.
Testo SE & Co. KGaA. 2023. Instruction manual Testo 417 vane anemometer. West Chester, PA, USA.
Tien, J.C. 1987. Shock losses around underground overcasts. In J. Mutmansky, ed. 3rd Mine Ventilation
Symposium. State College, PA, USA.
TSI Incorporated. 2022. DP-CALC MicroManometer Model 5825 Operation and Service Manual. Shore
view, Minnesota, USA.
White, F.M. 2015. Fluid Mechanics. 8th ed. New York, NY, USA: McGraw-Hill Education.
91
Diesel particulate control
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
E.A. Lutz
School of Mining and Mineral Resources, College of Engineering, University of Arizona, Tucson, Arizona, USA
ABSTRACT: Diesel exhaust is a known carcinogen and has been linked to several negative
health outcomes, yet it is commonly used in industrial settings such as underground mining. Alter
native fuels such as GDiesel® (GD), a natural gas/diesel blend, have the potential to reduce health
exposures and associated health effects. Mirroring a previous study in which use of GD in a 2005
Wagner load-haul-dump (LHD) vehicle with oxidation catalyst demonstrated significant reduc
tions in diesel exhaust exposures, operator-location and area exposure samples were collected in
an underground mining laboratory with diesel (D) and then GD fuel while operating a JCI LHD.
Analytes of interest included total and respirable diesel particulate matter (tDPM and rDPM,
respectively), total and respirable elemental and organic carbon (tEC, rEC, tOC, rOC, respect
ively), as well as formaldehyde (CH2O), nitric oxide (NO), and nitrogen dioxide (NO2). Use of
GD resulted in non-significant reductions in median rOC, tDPM, tOC, and NO2 concentrations,
non-significant increases in rDPM, rEC, tEC, and NO, and identical median CH2O concentra
tions. A significant decrease in NO2 exposure concentrations (p=0.012) and increase tEC and
rEC exposure concentrations (p=0.023 and p=0.024, respectively) were observed. After controlling
for environmental confounders there was no difference observed in analyte concentrations
between the two fuels. Further research is needed to determine whether GD alone can significantly
reduce human exposures across vehicles and pollution configuration types.
1 INTRODUCTION
1.1 Background
Diesel fuel (D) is used in a variety of applications for several industries, including transportation,
agriculture, railroads, construction, and mining. In 2012 the International Agency for Research
on Cancer designated D exhaust as a Group 1 carcinogen in humans (WHO 2012). Large longitu
dinal studies performed with underground miners have linked their exposure to D with increased
risk of lung cancer (Attfield et al 2012, Silverman et al 2012). Exposure to D engine emissions is
associated with adverse health effects including: chronic bronchitis, respiratory tract infections,
asthma exacerbation, and increased cardiovascular morbidity and mortality (Bonauto et al 2007,
Culp et al 2011, Kagawa 2002, Taylor 2006). Given the nature of D exposures in the mining
industry and the many associated health effects (Attfield et al 2012, Silverman et al 2012, Sydbom
et al 2001), reducing engine emissions has become an industrial hygiene priority. In the US, there
are nearly 14,000 mine sites that employ over 210,000 individuals in the workforce (BLS 2017,
MSHA 2014). Both underground and surface mines predominantly use diesel-powered vehicles at
their operations to perform tasks such as drilling, mucking, and hauling. Miners are regularly
exposed to D during their work, especially in areas where there is limited ventilation.
DOI: 10.1201/9781003429241-10
95
1.2 Regulatory compliance
In 2008 the Mine Safety and Health Administration (MSHA) published updated standards for
the allowable concentration of respirable (<1.0 µm with impactor) diesel particulate matter
(rDPM) in an effort to control D exhaust exposures. The administration has also recently pub
lished a request for information (RFI) regarding diesel exhaust exposures and controls (MSHA
2016). The permissible exposure limit (PEL) of 160 µg/m3 as total (combined inorganic and
organic) carbon (CFR 2016) is monitored through air sampling and processing per the National
Institute for Occupational Safety and Health (NIOSH) analytical method 5040. Despite the
installation of controls such as mine ventilation, the underground rDPM exposure standard is
frequently exceeded. Since 2000, MSHA has made detailed information regarding the occupa
tional health exposure assessments performed during inspections available to the public. Accord
ing to data since the new diesel exhaust standard, diesel particulate matter (DPM) had a higher
median exposure concentration-to-PEL (ECP) ratio (0.47) than any other airborne contaminant
(for which 100 or more samples were taken) except silver fumes (0.88). This ratio indicates that
mine operators continue to struggle to control D exhaust exposures, and that governmental,
national, and industrial organizations should focus on developing better control options.
1.3 Biodiesel
Given the health hazards associated with and MSHA’s focus on D exposures, mine operators
seek effective control methods for rDPM. Some have explored the use of biodiesel (B) fuel
blends. Studies comparing D to B emissions and exposures report a decrease in total carbon
output but an increase in organic carbon, aldehydes, and nitrogen dioxide (Turrio-Baldassarri
et al 2004, Knothe, Sharp, & Ryan 2006, Li & McLaughlin 2005, Bugarski et al 2010, He et al
2009, Bugarski et al 2003). However, other research demonstrated an increase in aldehyde,
nitrogen dioxide (NO2), and organic carbon fraction concentrations with use of B mixtures
(Bugarski et al 2010, Liu et al 2009). Despite the increasing usage of B, there is a lack of infor
mation on the human health effects of exposure to these emissions. Recent studies also suggest
that exposure to B exhaust may be as or more toxic than D (Donoghue & Bates 2000, Bhavar
aju et al 2013, Brak & Bates 2003, Bunger et al 2000, Armstrong et al 1998, Mehus et al 2015).
1.4 GDiesel ®
GDiesel® (GD), an EPA-approved natural gas/diesel fuel mixture, became commercially avail
able in 2011 and provides a second alternative to D fuel. GD is prepared by combining diesel
with natural gas, using a proprietary charged-catalytic reaction. The end product is an American
Society for Testing and Materials (ASTM)-designated D fuel with purportedly reduced tailpipe
emissions of diesel particulate matter (DPM) and oxides of nitrogen (NOx) compounds (NDEP
2010, ARC 2014). Our independent, simulated study using a 2005 Wagner load-haul dump
(LHD) with a diesel oxidative catalyst (DOC) but no diesel particulate filter (DPF) at an under
ground mining laboratory showed a 66% reduction in rDPM and significant reductions for all
other analytes except carbon monoxide (CO) associated with use of GD (Lutz et al 2015). LHDs
are commonly used in underground mine settings and often associated with high exposures to
D (Bugarski et al 2004, Bugarski et al 2003). While promising, this GD study represents a single
vehicle, study design, and pollution configuration. Other studies have shown that D is influenced
by a variety of factors including type of equipment, load, and pollution control equipment
(Bugarski et al 2010, Liu et al 2009, Bugarski et al 2004, Durbin et al 2000, Tsai et al 2011).
1.5 Objective
The objective of this pilot study was to evaluate the exhaust exposures of GD, as compared to
D, using a JCI LHD with Deutz engine and DOC but no DPF. We hypothesized that use of
GD fuel in the JCI LHD would result in significant reductions for all analytes, comparable to
those observed in our previous study.
96
2 METHODS
Figure 1. Depiction of the decline at the University of Arizona San Xavier Underground Mining
Laboratory.
Note: Image not to scale.
97
A JCI LHD with inline, 4-cylinder, 55 horsepower at 2300 RPM Deutz engine (motor
number 7534-462) with an Engine Control Systems Purimuffler (part number A17-0313) DOC,
but no DPF, was borrowed from an underground mine in the region. It also had an open cabin.
This study was conducted at the SX decline (see Figure 1 below). Similar to the previous study
(Lutz et al 2015), the decline’s gate (position B) was covered with a vinyl tarp and the metal door
(position D) entering the ‘adit’ level was closed at the start of each exposure session to minimize
airflow into and out of the decline. The mine fan was not operated during the study sessions.
2.3 Procedure
Research staff operated the vehicle for five consecutive days using D. Two 200-minute exposure
sessions were completed each day, with one operator-location and one area sample taken during
each session. In order to change from D to GD fuel, the LHD fuel tank was emptied using
a fuel pump, filled with GD, and then driven for approximately one hour before beginning the
first GD exposure session. The LHD was then operated for a total of four days with GD fuel.
Due to water constraints in the mine and potential for significant dust dispersion, mucking was
not performed. The LHD bucket was filled once at the beginning of the study and the equip
ment simply driven back and forth on the 30-meter decline mucking path by the same operator.
The bucket was not emptied until the end of the study. Each day the vehicle was not allowed to
idle before the first session and, before beginning the second, the tarp was removed for 30 min
utes to allow natural ventilation to completely disperse carbon monoxide. CO dispersion was
confirmed with an MSA 4X Altair gas monitor (Cranberry Township, PA).
98
remaining environmental measures as inputs. An alpha error threshold of 0.05 was used
throughout.
Using fences at 150% of the interquartile range, seven outliers were identified, by fuel type,
and removed from the D (rOC, NO2) and GD (rDPM, rEC, rOC, NO2) datasets. Statistical
analysis was performed with and without outliers. In three of the four datasets there were no
changes in conclusions observed when outliers were included versus not included. However,
the rEC sample set changed when outliers were included. The rEC results with and without
outliers are provided in the results section.
3 RESULTS
3.1 Overview
There were a total of 20 D and 16 GD samples collected. Variations in samples collected were due
to LHD equipment malfunction during two of the GD sampling days. For the unadjusted rank-
sum difference, median rOC, tDPM, tOC, and NO2 TWA8 exposure concentrations were reduced
with use of GD (see Figure 2 below), but only NO2 was significantly so (p=0.012). Use of GD
resulted in increased median concentrations of rDPM, rEC, tEC, and NO, though only tEC and
rEC (after removing outliers) were significantly greater (p=0.023 and p=0.024, respectively).
Figure 2. Median time-weighted average exposure concentrations by analyte. Note that Diesel (D) has
been set as a reference and GDiesel (GD) marker placement represents a respective percentage compared
to D. Abbreviations: respirable diesel particulate matter (rDPM), respirable elemental carbon (rEC), res
pirable organic carbon (rOC), total diesel particulate matter (tDPM), total elemental carbon (tEC), total
organic carbon (tOC), formaldehyde (CH2O), nitric oxide (NO), and nitrogen dioxide (NO2).
99
Median formaldehyde concentrations were identical. For more detailed results, see Tables 1, 2
and 3 below. After controlling for confounders, the likelihood of a given TWA8 being associated
with either fuel was not statistically significant for any analytes.
Table 1. Descriptive statistics for gaseous analytes formaldehyde (CH2O), nitric oxide (NO), and nitro
gen dioxide (NO2), by fuel. Abbreviations: diesel fuel (D), and GDiesel fuel (GD). All units in ppm.
Analyte CH2O NO NO2
Fuel D GD D GD D GD
N 19 16 20 16 20 16
Mean 0.035 0.033 1.946 2.976 0.345 0.159
SD 0.013 0.010 0.783 2.322 0.589 0.098
Median 0.030 0.030 1.725 1.868 0.186 0.120*
Q1 0.027 0.025 1.469 1.470 0.156 0.102
Q3 0.041 0.040 2.619 4.100 0.293 0.171
Min 0.020 0.020 0.413 1.000 0.088 0.076
Max 0.058 0.052 3.559 7.877 2.813 0.417
Table 2. Descriptive statistics for respirable particulates: respirable diesel particulate matter (rDPM),
respirable elemental carbon (rEC), and respirable organic carbon (rOC), by fuel. Abbreviations: Diesel
fuel (D), and GDiesel fuel (GD). All units in µg/m3.
Analyte rDPM rEC rOC
Fuel D GD D GD GDa D GD
N 20 15 20 15 14 20 15
Mean 274.9 305.2 115.6 137.6 144.4 159.3 167.8
SD 112.2 84.3 36.4 32.6 19.5 85.4 100.8
Median 250.4 283.4 113.4 140.9 143.8* 150.9 138.2
Q1 203.5 245.0 90.5 126.9 130.7 92.7 114.3
Q3 347.8 336.7 146.6 160.7 160.7 200.8 172.5
Min 111.9 217.8 50.8 41.4 115.3 53.9 93.5
Max 533.8 549.8 177.6 183.6 183.6 377.4 507.5
a
With outliers removed,
* significantly different from D at p<0.05.
Table 3. Descriptive statistics for total particulates: total diesel particulate matter (tDPM), total elemen
tal carbon (tEC), and total organic carbon (tOC), by fuel. Abbreviations: Diesel fuel (D), and GDiesel fuel
(GD). All units in µg/m3.
Analyte tDPM tEC tOC
Fuel D GD D GD D GD
N 20 16 20 16 20 16
Mean 551.7 526.4 136.2 161.6 413.3 361.4
SD 141.8 75.4 34.0 20.0 120.6 52.8
Median 551.1 500.0 134.3 155.8* 405.9 343.6
Q1 464.1 466.4 118.4 149.1 330.4 321.4
Q3 620.5 566.7 159.4 176.3 484.1 396.7
Min 283.3 460.6 66.7 123.3 212.5 298.8
Max 843.3 683.3 195.5 195.5 690.0 469.8
100
4 DISCUSSION
4.1 Findings
This study demonstrated little statistical difference between D and GD exhaust exposures in
a JCI LHD with DOC at the SX mine. The results of this study are inconsistent with results
observed by Lutz et. al. (2015), while using the Wagner LHD with DOC. Previously rDPM,
rEC, rOC, tDPM, tEC, tOC, CH2O, NO, and NO2 were significantly reduced, whereas this
study observed a significant decrease in NO2 alone, with significant increases in rEC and tEC.
These findings are also inconsistent with those described by the manufacturer of GD (NDEP
2010, ARC 2014). However, the Nevada Division of Environmental Protection (2010)
reported that use of GD may cause small increases in total hydrocarbon emissions. This
research is consistent with previous work that has demonstrated variability in exposure reduc
tions based on the use of different equipment and pollution control configurations (Bugarski
et al 2004, Bugarski et al 2003, Bugarski et al 2010, Durbin et al 2000, Tsai et al 2011).
Compared to regular D fuel, this pilot study demonstrated a reduction in NO2 and increase in
tEC and rEC with use of GD, with no change in rDPM or other analytes. Given that the MSHA
standard is based on rDPM exposures, this work makes it unclear whether use of GD would
likely increase compliance with the federal regulation. Despite MSHA’s focus on rDPM, other
components of D contribute to adverse health effects. Insufficient data are available to indicate
whether reducing one or some of the contaminant exposures from D influences observed health
effects.
Use of compressed natural gas in D engines has been shown to reduce particulate emissions,
polycyclic aromatic hydrocarbons, formaldehyde, and genotoxic activity, compared to D fuel
Fritz & Egbuonu 1992, Greenwood et al 1996, Ayala et al 2002, Turrio-Baldassarri et al
2006). Despite this study’s findings, a reduction in exposures with GD fuel would seem reason
able given the apparently cleaner burning qualities of natural gas.
Differences observed between fuel types are likely a function of the fuels’ chemical compos
ition, vehicle engine conditions, such as temperature of combustion and concentration of
oxygen present, and emission source controls, such as DOCs and DPFs. For example, differ
ences in DPM could be due to more complete combustion from higher engine temperatures,
or impurities and hydrocarbon chain lengths found in each fuel. In addition, EC is released
directly from the incomplete combustion of fossil fuels while OC, an aggregate of hundreds of
individual compounds spanning a wide range of chemical and thermodynamic properties, is
formed by a variety of processes, suggesting perhaps more complete combustion with D in the
JCI LHD. Variation in NOx concentrations could be due to higher combustion temperatures,
as well as interaction with a DOC. Higher engine temperatures and increased oxygen content
may lead to conditions that increase the formation of short hydrocarbons, such as formalde
hyde and acetaldehyde. Increased combustion temperature may therefore explain the decrease
in DPM but increase in aldehydes associated with B in some studies.
4.2 Limitations
The limitations of our pilot study include use of a single vehicle and single pollution control
configuration (DOC, no DPF). In addition, this study occurred at a naturally ventilated
underground laboratory, rather than an operational mine where other pieces of equipment
and active, forced ventilation are typically present. During two of the scheduled GD exposure
sessions the LHD would not start due to heating of the starter solenoid. Heating of the engine
over time, as well as the chosen order of fuel use, may have biased these results. While efforts
were taken to minimize airflow into and out of the mine, outdoor environmental conditions
such as wind speed and barometric pressure may have influenced our observations.
4.3 Conclusions
The current lack of information regarding the alternative fuel’s emissions suggests that further
research is needed to determine whether GD can significantly reduce occupational and
101
environmental exposures across vehicle and pollution configuration types. We recommend
research that utilizes a crossover design, targets several of the most commonly used diesel-
fueled vehicles in the mining industry and monitors exposures and health effects over an
extended period. The current study, coupled with previous research [28], demonstrates how
changes to the types of equipment used, driving conditions and environmental conditions
make predicting exposure risks from diesel exhaust extremely difficult.
REFERENCES
Advanced Refining Concepts, LLC (ARC). 2014. “GDiesel® - The World’s Best Diesel Fuel.” Available
at http://www.clearrefining.com/newfactsheet1211.pdf
Armstrong, L.E., Soto, J.A.H. and Hacker, F.T., 1999. Urinary indices during dehydration, exercise, and
rehydration. Occupational Health and Industrial Medicine, 2(40), p. 97.
Attfield, M.D., Schleiff, P.L., Lubin, J.H., Blair, A., Stewart, P.A., Vermeulen, R., Coble, J.B. and
Silverman, D.T., 2012. The diesel exhaust in miners study: a cohort mortality study with emphasis on
lung cancer. Journal of the National Cancer Institute, 104(11), pp. 869–883.
Ayala, A., Kado, N.Y., Okamoto, R.A., Holmén, B.A., Kuzmicky, P.A., Kobayashi, R. and Stiglitz, K.
E., 2002. Diesel and CNG heavy-duty transit bus emissions over multiple driving schedules: Regulated
pollutants and project overview. SAE Transactions, pp. 735–747.
Bhavaraju, L., Shannahan, J., William, A., McCormick, R., McGee, J., Kodavanti, U. and Madden, M.,
2014. Diesel and biodiesel exhaust particle effects on rat alveolar macrophages with in vitro exposure.
Chemosphere, 104, pp. 126–133.
BLS: U.S. Department of Labor. 2017. “Industries at a Glance: Mining (except Oil and Gas): NCAIS
212. U.S. Bureau of Labor Statistics. Available at http://www.bls.gov/iag/tgs/iag212.htm.
Bonauto, D., Anderson, R., Rauser, E. and Burke, B., 2007. Occupational heat illness in Washington
State, 1995–2005. American journal of industrial medicine, 50(12), pp. 940–950.
Brake, D.J. and Bates, G.P., 2003. Fluid losses and hydration status of industrial workers under thermal
stress working extended shifts. Occupational and environmental medicine, 60(2), pp. 90–96.
Bugarski, A., Schnakenberg, G., Noll, J., Mischler, S., Patts, L., Hummer, J., Vanderslice, S., Crum, M.
and Anderson, R., 2003. ‘The Effectiveness of Selected Technologies in Controlling Diesel Emissions
in an Underground Mine—Isolated Zone Study at Stillwater Mining Company’s Nye Mine. Draft
Report,”NIOSH, Pittsburgh Research Laboratory.
Bugarski, A., Mischler, S., Noll, J., Schnakenberg, G., Crum, M. and Anderson, R., 2004. ‘An Evaluation
of the Effects of Diesel Particulate Filter Systems on Air Quality and Personal Exposure of Miners at
Stillwater Mine Case Study: Production Zone. Report to M/NM Diesel Partnership, April, 1.
Bugarski, A.D., Cauda, E.G., Janisko, S.J., Hummer, J.A. and Patts, L.D., 2010. Aerosols emitted in
underground mine air by diesel engine fueled with biodiesel. Journal of the Air & Waste Management
Association, 60(2), pp. 237–244.
Bünger, J., Müller, M.M., Krahl, J., Baum, K., Weigel, A., Hallier, E. and Schulz, T.G., 2000. Mutageni
city of diesel exhaust particles from two fossil and two plant oil fuels. Mutagenesis, 15(5), pp. 391–397.
Code of Federal Regulations (CFR). 2016. “Limit on exposure to diesel particulate matter.” Title 30.
Part 57.5060.
Culp, K., Tonelli, S., Ramey, S.L., Donham, K. and Fuortes, L., 2011. Preventing heat-related illness
among Hispanic farmworkers. AAOHN journal, 59(1), pp. 23–32.
Donoghue, A.M., and Bates, G.P. 2000. The risk of heat exhaustion at a deep underground metalliferous
mine in relation to body-mass index and predicted Vo(2)max. Occupational Medicine-Oxford 50(4):
259–263.
Durbin, T.D., Collins, J.R., Norbeck, J.M. and Smith, M.R., 2000. Effects of biodiesel, biodiesel blends,
and a synthetic diesel on emissions from light heavy-duty diesel vehicles. Environmental science &
technology, 34(3), pp. 349–355.
Fritz, S.G., and Egbuonu, R.I. 1992. EMISSIONS FROM HEAVY-DUTY TRUCKS CONVERTED
TO CNG. Journal of Engineering for Gas Turbines and Power-Transactions of the Asme 114(3):
561–567.
Greenwood, S.J., Coxon, J.E., Biddulph, T. and Bennett, J., 1996. An investigation to determine the
exhaust particulate size distributions for diesel, petrol, and compressed natural gas fuelled vehicles.
SAE transactions, pp. 640–646.
He, C., Ge, Y., Tan, J., You, K., Han, X., Wang, J., You, Q. and Shah, A.N., 2009. Comparison of
carbonyl compounds emissions from diesel engine fueled with biodiesel and diesel. Atmospheric Envir
onment, 43(24), pp. 3657–3661.
102
Kagawa, J., 2002. Health effects of diesel exhaust emissions—a mixture of air pollutants of worldwide
concern. Toxicology, 181, pp. 349–353.
Knothe, G., Sharp, C.A. and Ryan, T.W., 2006. Exhaust emissions of biodiesel, petrodiesel, neat methyl
esters, and alkanes in a new technology engine. Energy & fuels, 20(1), pp. 403–408.
Li, Y.X., McLaughlin, N.B., Patterson, B.S. and Burtt, S.D., 2006. Fuel efficiency and exhaust emissions
for biodiesel blends in an agricultural tractor. Canadian biosystems engineering, 48(2).
Liu, Y.Y., Lin, T.C., Wang, Y.J. and Ho, W.L., 2009. Carbonyl compounds and toxicity assessments of
emissions from a diesel engine running on biodiesels. Journal of the Air & Waste Management Associ
ation, 59(2), pp. 163–171.
Lutz, E.A., Reed, R.J., Lee, V.S. and Burgess, J.L., 2015. Occupational exposures to emissions from
combustion of diesel and alternative fuels in underground mining—a simulated pilot study. Journal of
occupational and environmental hygiene, 12(3), pp .18–25.
Mehus, A.A., Reed, R.J., Lee, V.S., Littau, S.R., Hu, C., Lutz, E.A. and Burgess, J.L., 2015. Comparison
of acute health effects from exposures to diesel and biodiesel fuel emissions. Journal of occupational
and environmental medicine, 57(7), p. 705.
Mine Safety and Health Administration (MSHA). 2014. “U.S. Department of Labor: “Mining Safety
and Health At a Glance.” Available at http://www.msha.gov/MSHAINFO/FactSheets/MSHAFCT10.
asp (accessed May 01, 2014).
Mine Safety and Health Administration (MSHA). 2016. Exposure of Underground Miners to Diesel
Exhaust: A Proposed Rule by the Mine Safety and Health Administration.
Nevada Division of Environmental Protection (NDEP). 2010. “Notice of Proposed Action.”
National Institute of Occupational Safety and Health (NIOSH). 1994. U.S. Department of Health and
Human Services. NITRIC OXIDE and NITROGEN DIOXIDE, 6014. NIOSH Manual of Analytical
Methods (NMAM), Fourth Edition. National Institute of Occupational Safety and Health.
National Institute of Occupational Safety and Health (NIOSH). 2003a. U.S. Department of Health and
Human Services. DIESEL PARTICULATE MATTER, 5040. NIOSH Manual of Analytical Methods
(NMAM), Fourth Edition. National Institute of Occupational Safety and Health.
National Institute of Occupational Safety and Health (NIOSH). 2003b. U.S. Department of Health and
Human Services. FORMALDEHYDE, 2016. NIOSH Manual of Analytical Methods (NMAM),
Fourth Edition. National Institute of Occupational Safety and Health.
Silverman, D. T., Samanic, C. M., Lubin, J. H., Blair, A. E., Stewart, P. A., Vermeulen, R., Coble, J. B.,
Rothman, N., Schleiff, P. L., Travis, W. D., Ziegler, R. G., Wacholder, S., & Attfield, M. D. 2012.
The Diesel Exhaust in Miners study: a nested case-control study of lung cancer and diesel exhaust.
Journal of the National Cancer Institute, 104(11),855–868. https://doi.org/10.1093/jnci/djs034.
Sydbom, A., Blomberg, A., Parnia, S., Stenfors, N., Sandström, T. and Dahlen, S.E., 2001. Health effects
of diesel exhaust emissions. European Respiratory Journal, 17(4), pp. 733–746.
Tsai, J.H., Chen, S.J., Huang, K.L., Lee, W.J., Kuo, W.C. and Lin, W.Y., 2011. Characteristics of par
ticulate emissions from a diesel generator fueled with varying blends of biodiesel and fossil diesel.
Journal of Environmental Science and Health, Part A, 46(2), pp. 204–213.
Turrio-Baldassarri, L., Battistelli, C.L., Conti, L., Crebelli, R., De Berardis, B., Iamiceli, A.L.,
Gambino, M. and Iannaccone, S., 2004. Emission comparison of urban bus engine fueled with diesel
oil and ‘biodiesel’blend. Science of the Total Environment, 327(1-3), pp. 147–162.
Turrio-Baldassarri, L., Battistelli, C.L., Conti, L., Crebelli, R., De Berardis, B., Iamiceli, A.L.,
Gambino, M. and Iannaccone, S., 2006. Evaluation of emission toxicity of urban bus engines: Com
pressed natural gas and comparison with liquid fuels. Science of the Total Environment, 355(1-3), pp.
64–77.
World Health Organization. 2012. “IARC: Diesel Engine Exhaust Carcinogenic.” International Agency
for Research on Cancer.
103
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
S. Friend
National Institute for Occupational Safety and Health, Health Effects Laboratory Division,
Morgantown, West Virginia, USA
1 INTRODUCTION
Due to widespread use of various types of diesel-powered vehicles in hard to cost-effectively ven
tilate confined spaces, underground miners around the world are exposed to relatively high con
centrations of submicron aerosols and gases (Cohen et al. 2002, Pronk et al. 2009, Peters et al.
2017, Susanto et al. 2018, MSHA 2022). Long-term exposure to aerosols and gases emitted by
“traditional” (Ruehl et al. 2015) diesel engines was found to have adverse effects on pulmonary
(Kim et al. 2018, Du et al. 2020), vascular (Mills et al. 2005, Nejad et al. 2013), and other systems
(Faherty et al. 2021). The International Agency on Research on Cancer (IARC) recognizes the
risks of lung cancer and categorized diesel engine exhaust as a carcinogen to humans (group 1)
(IARC 2012). Rapid development of engine, aftertreatment, fuel, and lubrication oil technologies
over the past couple of decades resulted in “clean” engines that produce substantially lower par
ticulate and gaseous emissions (Fiebig et al. 2014, Ruehl et al. 2015, Dallman and Menon 2016).
The implementation of various control strategies and technologies such as replacement of older
with advanced engines (Ruehl et al. 2015, Bugarski et al. 2020a), retrofitting older engines with
exhaust aftertreatment technologies (Bugarski et al. 2009, Bugarski et al. 2020b), the use of alter
native fuels (Bugarski et al. 2014, Bugarski et al. 2016), and replacement of diesel-powered
vehicles with battery-powered (Halim et al. 2022) electric vehicles have resulted in a gradual
reduction in the industry-wide average exposures of underground metal and nonmetal miners to
elemental carbon (EC) and total carbon (TC) concentrations (Bugarski and Potts 2018, Bugarski
DOI: 10.1201/9781003429241-11
104
et al. 2022, Gren et al. 2022). In addition, the exposure to exhaust from contemporary “clean”
diesel engines equipped with advanced exhaust aftertreatment systems were shown to have less
adverse health effects (McDonald et al. 2015). However, some occupations across the industry
are still exposed to relatively high concentrations of “traditional” diesel aerosols (Bugarski and
Potts 2018, Gren et al. 2022, MSHA 2022).
Monitoring ambient concentrations and personal exposures to diesel particulate matter
(DPM) is complicated by the fact that DPM describes very complex ever-changing products of
incomplete combustion of petroleum diesel and alternative fuels (Bugarski et al. 2014, Bugarski
et al. 2016). This is further complicated in many situations by the use of multiple diesel engines
of various generations and designs (Bugarski et al. 2020b) operated over a wide spectrum of
conditions at various locations in underground mines (Rubeli et al. 2004, Saarikoski et al.
2018, Saarikoski et al. 2019, Bugarski et al. 2020c, Bugarski et al. 2022). In addition, sampling,
analysis, and measurement of DPM is even further complicated by other factors including
small particle size, the dynamic nature of the formation and transformation of DPM, and the
interaction with other aerosols and environment (Bugarski et al. 2012, Giechaskiel et al. 2014).
The currently used methods for compliance monitoring of exposures of underground miners
to diesel aerosols at permissible exposure levels (PELs) of 160 µgTC/m³ (71 Fed. Reg. 33387 2006,
OHSA 2017) and 100 µgEC/m³ (AIOH 2013) are grounded in the use of the National Institute for
Occupational Safety and Health (NIOSH) Method 5040 (Birch 2002, NIOSH 2003, NIOSH
2016). These offline methods are founded on: (1) collecting representative samples of diesel aero
sols on quartz fiber filters (QFFs) using size-selective sampling, and (2) subsequent carbon ana
lysis of those filters using a thermo-optical transmittance (TOT) method (NIOSH 2003).
The sampling apparatus currently used for compliance sampling incorporates two consecutive
pre-selectors (MSHA 2020, SKC 2022). This apparatus was designed using parameters estab
lished during development of the personal diesel aerosol sampler by the U.S. Bureau of Mines
(Cantrell and Rubow 1991, McCartney and Cantrell 1992): (1) a respirable cyclone operated at
1.7 lpm with a cut-off size of approximately 4.0 µm, which is used to eliminate the majority of
coarse, mostly mechanically generated dust from respirable dust, and (2) a four-orifice sharp-cut
impactor with a cut-off size of approximately 0.8 µm when operated at 1.7 lpm (McCartney and
Cantrell 1992, Noll et al. 2005) incorporated with the cassette is used to eliminate the remaining
dust from diesel aerosols. The design criteria for the impactors (Huang 2005) used with the per
sonal diesel aerosol sampler (PDAS) (McCartney and Cantrell 1992) and consequently in the
SKC Inc. DPM cassette (Noll et al. 2005) were established using the results of the size-selective
aerosol measurements in four underground coal mines in the United States in the late 1980s
(Cantrell and Rubow 1991). Those measurements showed that contemporary diesel engines con
tributed aerosols with a mass median diameter larger than 150 nm. The sharp-cut impactor with
cut-off size of 0.8 ± 0.1 µm was found to provide acceptable separation of coal dust and diesel
aerosols. The average loss of diesel aerosols was estimated to be 0.2% to 8.4%, and contamin
ation with coarse aerosols was estimated to be 0.1% to 3.6% (Cantrell and Rubow 1991). It is
important to note that the concentrations of diesel aerosol in various areas in those mines where
diesel vehicles were used were relatively high (substantially above current PEL) and contributed
a substantial fraction of respirable aerosol mass (Cantrell et al. 1992).
The DPM collected on QFFs is analyzed for EC and organic carbon (OC) using TOT analysis
(NIOSH 2003). The results are used to determine total carbon (TC=EC+OC). The TOT analysis
was chosen over the traditionally used gravimetric analysis for its sensitivity and selectivity for
monitoring DPM at the Mine Safety and Health Administration (MSHA) targeted PEL levels
(Ramachandran and Watts 2003). The method was originally developed for EC analysis and later
adopted for TC analysis (MSHA 2016). EC was proposed as a more suitable surrogate than TC
for DPM monitoring (Birch and Noll 2004) because EC is: (1) selective of combustion byproducts
of diesel fuels, (2) not present in other submicron aerosols in underground mines such as cigarette
smoke, wood smoke, and oil mist, and is (3) present in small quantities in coal and other carbon
aceous dust. However, since TC makes up the majority of “traditional” DPM and includes
important carcinogenic organic compounds, it was selected by the MSHA as a surrogate for
monitoring exposures of underground miners to DPM (71 Fed. Reg. 33387 2006). The appropri
ateness of the methodology of using TC as a surrogate for assessing exposures of underground
105
miners to DPM have been challenged previously (Cohen et al. 2002, Noll et al. 2015). The defi
ciencies in the precision and accuracy of this methodology were primarily attributed to a number
of issues including: (1) potential presence of OC interferences in an underground mining environ
ment (Cohen et al. 2002, Birch and Noll 2004, Noll et al. 2015), (2) adsorption of gas phase (non-
DPM) OC on QFFs (Noll and Birch 2008, Bosch 2015), (3) contamination of QFFs in SKC
DPM cassettes (Noll et al. 2019), and (4) losses of DPM to coarse aerosols (Gaillard et al. 2019).
At present, MSHA does not require monitoring exposure to DPM in underground coal mines in
the United States. Monitoring DPM exposures of underground metal, nonmetal, and coal mines
in Australian states and Germany is performed using EC as a surrogate (AIOH 2013).
Since the presence of micron-sized particles in the DPM samples can interfere with TOT analysis,
the effective removal of micron-sized aerosols using size-selective sampling is critical to monitoring
DPM exposures in underground mining environments. The presence of coal dust that contains non
trivial amounts of OC (Yang and Yu 2002) and to some extent EC (Birch and Noll, 2004) can cause
overestimation of OC and EC and errors in determination of the EC/OC split. If present in the
DPM samples, alkali, alkaline-earth, and transition metal salts could interfere with TOT analysis by
changing EC oxidation temperature and causing more charring (Wang et al. 2010). Carbonates pre
sent in rock dusting material could also interfere with the DPM measurement (Cavalli et al. 2010).
The evolution of diesel technology resulted in lower emission rates, decrease in size, and changes
in chemical composition of emitted diesel aerosols. Measurement of size distributions of diesel aero
sols in underground mines (Skubacz et al. 2017, Saarikoski et al. 2019, Salo et al. 2021, Bugarski
et al. 2022) shows that the size distributions of diesel aerosols in many contemporary underground
mines, particularly in the mines with a high penetration of advanced diesel technology, are character
ized with smaller diameters from those used when designing the PDAS (McCartney and Cantrell
1992) and DPM cassette (SKC). In the late 1990s and early 2000s when EC and TC were selected as
surrogates for monitoring DPM exposures of underground metal and nonmetal mines in the United
States (71 Fed. Reg. 33387 2006, MSHA 2016), the contemporary diesel engines emitted DPM that
was primarily made of EC and OC (Lowenthal et al. 1994, McDonald et al. 2003). Although EC
and especially OC still contribute substantially to total particulate matter mass, the dominant con
stituents of particulate mass emitted by advanced diesel engines equipped with diesel particulate filter
(DPF) systems are noncarbonaceous compounds such as nitrites, sulfates, and ammonium (Herner
et al. 2011, Khalek et al. 2011, Khalek et al. 2015, Ruehl et al. 2015).
The changes in the DPM concentrations and physical properties necessitates improvements
in the precision and accuracy of the methodology used for monitoring DPM using TC as
a surrogate, at levels substantially below current MSHA PEL of 160 µgTC/m³. Although previ
ously proposed interventions (Noll et al. 2005, Noll et al. 2015, Noll et al. 2019) resulted in
measurable improvements, additional work is needed to minimize collection of micron-sized
particles that can interfere with TOT analysis. This evaluation was conducted to evaluate the
potential for an improvement in sampling methodology for the collection of diesel aerosols by
reducing and eventually optimizing cut-off size for the impactor in the DPM sampling cassette.
2 METHODOLOGY
Three impactors with different sized orifices were evaluated. The reference impactor (I-061) was
the one in the commercially available DPM cassettes (SKC 225-317). The additional two impact
ors were constructed by modifying impactor assemblies in the existing DPM cassettes to develop
the sharp-cut impactors with lower cut-off sizes. In the modified assemblies, the original 0.61-
mm synthetic sapphire orifice inserts (SSOIs) were replaced with 0.52-mm SSOIs (O’Keefe Con
trols Co., Monroe, CT, SAP-44-SA) to obtain I-052 and with 0.44-mm SSOIs (O’Keefe Controls
Co., Monroe, CT, SAP-44-SA) to obtain I-044. For the sake of simplicity, the distance between
nozzle and impaction plate was kept identical to the one in the existing DPM cassette.
The penetration efficiencies of the impactors were determined using polystyrene divinylben
zene (PS/DVB) beads with nominal diameters of 200, 300, 400, 500, 600, 700, 800, and 900 nm
in aqueous suspension (Thermo Fisher Scientific, Fremont, California, Nanosphere™ Poly
mer Microspheres). The PS/DVB beads were aerosolized using a 3-Jet Collison nebulizer (CH
106
Technologies, Westwood, New Jersey). The normalized cumulative size distributions of aero
solized beads measured with the scanning mobility particle sizer (SMPS) with the electrostatic
classifier (TSI, Shoreview, MN, Models 3080L) and ultrafine condensation particle counter
(TSI, Shoreview, MN, Models 3776), and Aerodynamic Particle Sizer® (APS™) (TSI, Shore
view, MN, Model 3321) are shown in Figure 1.
Figure 1. Normalized cumulative size distributions of aerosolized beads measured with SMPS and APS.
The impactors were field evaluated in the outby area of the underground coal mine during
two shifts (S2 and S3) to assess: (a) performance of the I-061, I-052, and I-044 impactors with
respect to penetration of micron-sized dust, and (b) differences in losses in diesel aerosols
between those three impactors. This evaluation was executed as a part of the study described
by Bugarski et al. (2022). The mine used “traditional” diesel engines in all categories of
vehicles, retrofit type DPFs on the nonpermissible equipment, and filtration systems with dis
posable filter elements on permissible equipment. The aerosol sampling and measurements
were performed at the continuous miner section. The size distribution of aerosols measured
with the fast mobility particle sizer (FMPS, TSI Model 3091) during S2 and S3 were found to
be bimodal with the agglomeration mode electrical mobility CMDs between 52 and 86 nm
(Bugarski et al. 2022). By introducing the assumption that the density of diesel aerosols in the
mine correlate with equations introduced by Maricq and Xu (2004) and Liu et al. (2009), the
mass median diameters of those aerosols could be estimated to be between 123 and 172 nm.
Figure 2. Sampling apparatuses: (a) used to collect DPM for SEM analysis, and (b) used to collect
DPM for carbon analysis.
107
The issue of penetration of micron-sized aerosols through the I-061, I-052, and I-044 impact
ors was investigated qualitatively on the results of scanning electron microscopy (SEM) analysis
performed on the samples collected with the sampling apparatus shown in Figure 2a. Four differ
ent kinds of samples were collected concurrently using: (a) open-face 5-piece cassette, (b) 5-piece
cassette preceded by cyclone and I-061, (c) 5-piece cassette preceded by cyclone and I-052, and
(d) 5-piece cassette preceded by I-044. The samples were collected on the 25-mm, 0.2-μm track-
etched polycarbonate membrane filters (Millipore Sigma, Whatman® Nuclepore™, Darstadt,
Germany). The samples collected at a 600-second sampling time were used for analysis.
Twenty-one sections of each of the filters, one in the center and four groups of five radially distrib
uted around the center at 90° spacing were analyzed by scanning electron microscopy (SEM) (Model
S-4800 by Hitachi, Tokyo, Japan) using 10K magnification and energy dispersive x-ray spectroscopy
using 20 to 70K magnification (by Bruker Quantax, Madison, Wisconsin). The particle imaging was
performed at 20 kV, and elemental analysis was performed using 20-kV incident beam energy.
The issue of the differences in diesel aerosol (EC and OC) losses in the three evaluated impactors
was studied using the results of the carbon analysis on sets of at least four filter samples of the same
kind collected concurrently using the custom sampling apparatus shown in Figure 2b. The respir
able cyclones (Zefon International, Ocala, FL, Zefon Nylon Dorr-Oliver Cyclone) were used ahead
of each of the impactors. The DPM samples were collected on the primary and secondary quartz
fiber filters (QFFs) (Millipore Sigma, Darmstadt, Germany, AQFA03700) housed inside of the
DPM cassettes. To minimize OC contamination, the QFFs were removed from the impactors and
preheated for two hours at a temperature of 800°C (Noll et al. 2019). The nominal sampling flow
rates of 2.0 lpm were maintained using critical flow orifices installed in two common manifolds.
A single vacuum pump (Oerlikon Leybold Vacuum, Cologne, Germany, Segovac SV25B) was used
to draw all samples. All DPM samples were analyzed at the NIOSH Pittsburgh Mining Research
Division using the NIOSH Method 5040 with modified temperature protocol. The analysis was
performed using an OC/EC Aerosol Analyzer (Sunset Laboratory Inc., Portland, OR).
3 RESULTS
The penetration curves for the assessed I-061, I-052, and I-044 impactors are shown in
Figure 3. The cut-off diameters (D50) for those impactors were estimated to be 716, 466, and
258 nm, respectively. The geometric standard deviations (GSDs) were calculated (Marple et al.
2001) as 1.359, 1.582, and 2.151. The D50 and GSD for I-061 are similar to those assessed for
the impactors in SKC DPM cassettes by Olson (2001) (746-770 nm and 1.391-1.420).
Twenty-one images of each of the filters were examined for the presence of micron-sized
particles. A total of twenty, eleven, seven, and three micron-sized particles were identified on
the open face, I-061, I-052, and I-044, respectively. The SEM images of the filters collected as
open face and with I-061, I-052, and I-044 impactors preceded with respirable cyclones are
shown in Figure 4. Elemental analysis of the selected micron-sized particles found at the same
108
locations showed that those particles were made primarily of the elements found in explosion
suppression rock dust (Man and Teacoach 2009) that likely were aerosolized by movement of
the diesel-powered equipment.
The results of carbon analysis performed on the samples collected with I-061, I-052, and I-044
impactors are shown in Figure 5. The mean concentrations of OC, EC, and TC were calculated
using results for at least 4 samples of the same kind. Certain samples (Figure 5) were deemed out
liers due to potential contamination during the manufacturing process and excluded from the cal
culations. The error bars represent one standard deviation of means for the specific sets of
samples. The OC, EC, and TC concentrations between the impactors were compared using ana
lysis of variance (ANOVA, OriginPro, OriginLab Corporation, Northampton, MA). All OC, EC,
and TC concentrations were significantly different (p<0.05) between the impactors except OC
concentrations in S2.
Figure 4. SEM images for the samples collected: (a) without pre-selectors, (b) with cyclone/I-061, (c)
with cyclone/I-052, and (d) with cyclone/I-044.
In the case of samples collected during S2, the OC, EC, and TC concentrations were 17.5%,
6.6%, and 11.0% lower for the samples collected with I-052 rather than I-061, and 34.3%, 14.1%,
and 22.2% lower for the samples collected with I-044 rather than I-061. In the case of samples col
lected during S3, the OC, EC, and TC concentrations were 24.5%, 7.4%, and 15.6% lower for the
samples collected with I-052 rather than I-061 and 30.4%, 13.9%, and 21.7% for the samples col
lected with I-044 rather than I-061.
Figure 5. OC, EC, and TC concentrations measured during: (a) S2, and (b) S3 (Number of samples: * -
four, ** - five, *** - six).
109
4 DISCUSSION
Reliable monitoring of exposures to TC, particularly the OC fraction of it, at the levels sub
stantially below the current MSHA PEL of 160 µgTC/m³ is challenging due to earlier identified
sampling and analytical issues. One possible improvement involves minimizing penetration of
the micron-sized non-diesel aerosols that could potentially adversely affect the results of the
TOT analysis (Yang and Yu 2002, Birch and Noll, 2004, Wang et al., 2010, Cavalli et al.
2010). We examined the potential of the SKC DPM cassettes with the modified impaction
assemblies to eliminate micron-sized aerosols more effectively from DPM samples.
The results of SEM analysis performed on the limited set of samples collected in
a contemporary coal mine showed that a substantial number of micron-sized aerosols present in
the mine atmosphere penetrated the impactor currently used in the SKC DPM cassette (I-061,
D50 = 716 nm). The I-052 (D50 = 466 nm) and particularly the I-044 (D50 = 258 nm) impactors
were much more effective in preventing micron-sized aerosols from reaching the sampling filters.
The QFFs collected with I-061, I-052, and I-044 contained comparable amounts of EC and
OC. The observed differences in EC and OC were at the level of limit of quantification (LOQ)
for EC and OC of the NIOSH Method 5040. The losses in EC for I-052, 6.6 and 7.4% for S2
and S3, respectively, were found to be statistically significant, but still relatively minor. The
losses in EC for I-044, 14.1 and 13.9% for S2 and S3, respectively, were somewhat more pro
nounced. The losses in OC for I-052, 17.5 and 24.5% for S2 and S3, respectively, and for
I-044, 34.3 and 30.4% for S2 and S3, respectively, were found to be more substantial than
those observed for EC. It is important to note that the uncertainty of OC measurements was
substantially higher for the majority of samples than that of EC measurements.
This limited set of data indicate that due to relatively large losses in TC, the evaluated I-044
might not be suitable for DPMs sampling in the mines with a large presence of equipment
powered by “traditional” diesel engines. Optimization of the preselector for specific applica
tions would be necessary. Those efforts would require good understanding of the physical
properties, particularly size distribution of diesel aerosols at the specific site. Unfortunately,
this kind of data is currently available for a very few sites.
Additionally, the design of the impactor needs to be optimized. In the case of the I-044
impactor, the relatively large GSD of 2.151 indicates that the design, in particular sharpness
of the impactor, can be potentially further improved by reducing the distance between nozzle
and impaction plate.
5 CONCLUSION
The changes in concentrations and physical properties of diesel aerosols in contemporary under
ground mines necessitates improvements in precision and accuracy of the methodology used for
sampling and analysis of DPM. This is particularly important if TC is used as a surrogate for moni
toring exposure to DPM. The results of this evaluation showed the potential for improvements in
the currently used size-selective sampling of diesel aerosols in underground mines. The SKC DPM
cassettes with the modified impaction assemblies with lower cut-off sizes were shown to effectively
eliminate micron-sized aerosols from DPM samples. The cut-off size of the impactor could be fur
ther optimized to minimize penetration of micro-sized aerosols and losses in EC or OC.
It is important to note that this evaluation was limited in scope and just proved the concept.
Measurements were done over relatively limited period(s) of time in a single section of an under
ground coal mine that is using equipment powered by “traditional” diesel engines. Additional
studies are needed to develop, optimize, and evaluate impactors for the specific applications, par
ticularly in the underground mines with the substantial use of “clean” diesel technologies.
6 DISCLAIMER
The findings and conclusions in this manuscript are those of the authors and do not necessar
ily represent the official position of the National Institute for Occupational Safety and Health
110
(NIOSH), Centers for Disease Control and Prevention (CDC). Mention of company names or
products does not constitute endorsement by NIOSH or CDC.
REFERENCES
71 Fed. Reg. 33387 2006. 30 CFR 57: Diesel particulate matter exposure of underground metal and non
metal miners. Limit on concentration of diesel particulate matter. Code of Federal Regulations. Wash
ington, DC: U.S. Government Printing Office, Office of the Federal Register.
AIOH 2013. Diesel Particulate Matter & Occupational Health Issues. Position Paper. Australian Institute
of Occupational Hygiene. AIOH Exposure Standards Committee.
Birch, E.M. 2002. Occupational monitoring of particulate diesel exhaust by NIOSH method 5040.
Applied Occupational and Environmental Hygiene, Vol. 17(6): 400–405. https://doi.org/10.1080/
10473220290035390.
Birch, M.E., Noll, J.D. 2004. Submicrometer elemental carbon as a selective measure of diesel particulate
matter in coal mines. Journal of Environmental Monitoring 6:799–806. https://doi.org/10.1039/b407507b.
Bosch, R. 2015. Review of sampling and evaluation techniques for diesel exhaust particulate matter in the
mining industry and the application of a vapor phase organic carbon correction factor. Montana Tech of the
University of Montana. Graduate Theses Non-Theses. 61. http://digitalcommons.mtech.edu/grad_rsch/61.
Bugarski, A.D. Hummer, J.A., Vanderslice, S. 2016. Effects of hydrotreated vegetable oil on emissions of
aerosols and gases from light-duty and medium duty older technology engines, Journal of Occupa
tional and Environmental Hygiene, 13:4, 293–302, http://dx.doi.org/10.1080/15459624.2015.1116695.
Bugarski, A.D., Cauda E., Janisko, S.J., Hummer, J.A., Patts, L.D. 2010. Aerosols emitted in under
ground mine air by diesel engine fueled with biodiesel. Journal of Air and Waste Management Associ
ation. 60, 237–244. https://doi.org/10.3155/1047-3289.60.2.237.
Bugarski, A.D., Hummer, J.A., Vanderslice, S., Barone, T. 2020a. Retrofitting and repowering as control
strategies for the curtailment of exposure of underground miners to diesel aerosols. Journal of Mining,
Metallurgy & Exploration, 37(2):791–802. https://doi.org/10.1007/s42461-019-00146-z.
Bugarski, A.D., Hummer, J.A., Vanderslice, S., Mischler, S. 2020b. Contribution of various types and
categories of diesel-powered vehicles to aerosols in an underground mine. Journal of Occupational
and Environmental Hygiene 17: 121–134. https://doi.org/10.1080/15459624.2020.1718157.
Bugarski, A.D., Hummer, J.A., Vanderslice, S., Shahan, M.R. 2020c. Characterization of aerosols in an
underground mine during a longwall move. Mining, Metallurgy & Exploration (MM&E)
37:1065–1078. https://doi.org/10.1007/s42461-020-00209-6.
Bugarski, A.D., Janisko S., Cauda E.G., Noll, J.D., Mischler, S.E. 2012. Controlling Exposure - Diesel
Emissions in Underground Mines. Society for Mining, Metallurgy, and Exploration. ISBN-13:
9780873353601.
Bugarski, A.D., Janisko, S.J., Cauda, E.G., Patts, L.D., Hummer, J.A., Westover, C., Terrillion, T. 2014.
Aerosols and criteria gases in an underground mine that uses FAME biodiesel blends. Ann. Occup.
Hyg. 58(8):971–982. http://dx.doi.org/10.1093/annhyg/meu049.
Bugarski, A.D., Potts, J.D. 2018. Exposures of Underground Miners to Diesel Particulate Matter in the
United States. 24th Annual Mining Diesel Emission Conference (MDEC) Conference. Toronto, Ontario,
Canada, October 2–4. Available from https://www.mdec.ca/2018/. (Accessed on January 14, 2021).
Bugarski, A.D., Schnakenberg, G.H., Jr., Hummer, J.A., Cauda E., Janisko, S.J., and Patts, L.D. 2009.
Effects of diesel exhaust aftertreatment devices on concentrations and size distribution of aerosols in
underground mine air. Environmental Science and Technology 43, 6737–6743. https://doi.org/10.3155/
1047-3289.60.2.237.
Bugarski, A.D., Vanderslice, S., Hummer, J.A., Barone, T.L., Mischler, S.E., Peters, S., Cochrane, S.,
Winkler, J. 2022. Diesel aerosols and gases in an underground coal mine. Mining, Metallurgy &
Exploration 39:937–945, https://doi.org/10.1007/s42461-022-00588-y.
Cantrell, B.K., Rubow K.L. 1991. Development of personal diesel aerosol sampler design and perform
ance criteria. Mining Engineering 43(2):232–236.
Cantrell, B.K., Rubow, K.L., Watts, W.F. Jr., Carlson, D.H. 1992. Pollutant levels in underground coal
mines using diesel equipment. Society for Mining Metallurgy and Exploration Transaction 290:1901–1907.
Cavalli, F., Viana, M., Yttri, K.E., Genberg, J. and Putaud, J.P., 2010. Toward a standardised
thermal-optical protocol for measuring atmospheric organic and elemental carbon: the EUSAAR
protocol. Atmospheric Measurement Techniques, 3(1):79–89. www.atmos-meas-tech.net/3/79/2010/.
Cohen, H.J., Borak, J., Hall, T., Sirianni, G., Chemerynski S. 2002. Exposure of miners to diesel exhaust
particulates in underground nonmetal mines, AIHA Journal, 63:5, 651–658, https://doi.org/10.1080/
15428110208984753.
111
Dallmann, T., Menon, A. 2016. Technology pathways for diesel engines used in non-road vehicle and
equipment. The International Council on Clean Transportation. White paper. September.
Du, M., Hall, G.l., Franklin, P., Musk, A.W. (Bill), Mullins, B.J., de Klerk, N., Elliott, N.S.J., Sodhi-
Berry, N., Brims, F., Reid, A. 2020. Association between diesel engine exhaust exposure and lung
function in Australian gold miners. International Journal of Hygiene and Environmental Health 226
113507. https://doi.org/10.1016/j.ijheh.2020.113507.
Gaillard, S., Sarver, E., Cauda, E. 2019. A field study on the possible attachment of DPM and respirable dust in
mining environments. Journal of Sustainable Mining 18: 100–108. https://doi.org/10.1016/j.jsm.2019.02.008.
Faherty, T., MacKenzie, A.R., Shapiro, K., Delgado-Saborit, J.M., McFiggans, G., Alfarra, R.,
Raymond, J. 2021. The impact of acute diesel exhaust exposure on executive brain function. Journal
of Vision. Volume 21 (9): 2562. https://doi.org/10.1167/jov.21.9.2562.
Fiebig, M., Wiartalla, A., Holdenbaum, B., Kiesow, S. 2014. Particulate emissions from diesel engines:
Correlation between engine technology and emissions. J Occup Medic Toxic 9(1):6. https://doi.org/
10.1186/1745-6673-9-6.
Giechaskiel, B., Maricq, M., Ntziachristos, L., Dardiotis C., Wang, X., Axmann, H., Bergmann, A.,
Schindler W. 2014. Review of motor vehicle particulate emissions sampling and measurement: From
smoke and filter mass to particle number. Journal of Aerosol Science 67: 48–86. https://doi.org/
10.1016/j.jaerosci.2013.09.003.
Gren, L., Krais, A.M., Assarsson, E., Broberg, K., Engfeld, M., Lindh, C., Strandberg, B., Pagels, J.,
Hedmer, M 2022. Underground emissions and miners’ personal exposure to diesel and renewable
diesel exhaust in a Swedish iron ore mine. International Archives of Occupational and Environmental
Health 95:1369–1388,. https://doi.org/10.1007/s00420-022-01843-x.
Halim, A., Lööw, J., Johansson, J., Gustafsson, J., van Wageningen, A., Kocsis, K. [2022]. Improvement
of working conditions and opinions of mine workers when battery electric vehicles (BEVs) are used
instead of diesel machines — Results of field trial at the Kittilä Mine, Finland. Mining, Metallurgy &
Exploration (2022) 39:203–219. https://doi.org/10.1007/s42461-021-00506-8.
Herner, J.D., Hu, S., Robertson, W.H., Huai, T., Chang, M.-C.O., Rieger, P., Ayala, A. 2011. Effect of
advanced aftertreatment for PM and NOx reduction on heavy-duty diesel engine ultrafine particle
emissions. Environ Sci Technol 45:2413–2419.
IARC 2012. IARC Press Release No. 213, June 12. Diesel engine exhaust carcinogenic. International
Agency for Research on Cancer, World Health Organization International Agency for Research on
Cancer, Lyon, France
Khalek, I.A., Blanks, M.G., Merritt, P.M., Zielinska, B. 2015. Regulated and unregulated emissions
from modern 2010 emissions-compliant heavy-duty on highway diesel engines. J. Air Waste Manage.
Assoc. 65(8):987–1001. https://doi.org/10.1080/10962247.2015.1051606.
Khalek, I.A., Bougher, T.L., Merritt, P.M., Zielinska, B. 2011. Regulated and unregulated emissions
from highway heavy-duty diesel engines complying with U.S. Environmental Protection Agency 2007
emissions standards. J. Air Waste Manage. Assoc. 61(4):427–442. doi:10.3155/1047-3289.61.4.427.
Kim, J., Peters, C.E., Arrandale, V.H., Labreche, F., Ge, C.B., McLeod, C.B., Song, C., Lavoué, J., Davies, H.
W., Nicol, A.-M., Pahwa, M., Demers, P.A. 2018. Burden of lung cancer attributable to occupational diesel
engine exhaust exposure in Canada. Occupational and Environmental Medicine 75(9):617–22.
Liu, G.Z., Vasys, V.N., Dettmann, M.E., Schauer, J.J., Kittelson, D.B., Swanson, J. 2009. Comparison
of strategies for the measurement of mass emissions from diesel engines emitting ultra-low levels of
particulate matter. Aerosol Sci. Technol. 43, 1142–1152.
Lowenthal, D.H., Zielinska, B., Chow, J.C., Watson, J.G., Gautam, M., Ferguson, D.H., Neuroth, G.R.,
Stevens, K.D. 1994. Characterization of heavy-duty diesel vehicle emissions. Atmos Environ 28(4): 731–743.
Man, C.K., Teacoach, K.A. 2009. How does limestone rock dust prevent coal dust explosions in coal
mines?. Mining Engineering 61(9):69–69.
Maricq, M.M., Xu, N. 2004. The effective density and fractal dimension of soot particle from premixed
flames and motor vehicle exhaust. Aerosol Science 35, 1251–1274.
McCartney, T.C., Cantrell, B.K. 1992. A cost-effective personal diesel exhaust aerosol sampler. United
States Department of the Interior, Bureau of Mines, IC 9324:24–30.
McDonald, J.D., Doyle-Eisele, M., Seagrave, J.C., Gigliotti, A.P., Chow, J., Zielinska, B., Mauderly, J.
L., Seilkop, K.S., Miller, R.A., 2015. Part 1. Assessment of carcinogenicity and biologic responses in
rats after lifetime inhalation of new-technology diesel exhaust in the ACES bioassay. In: Advanced
Collaborative Emissions Study (ACES): Lifetime Cancer and Non-Cancer Assessment in Rats
Exposed to New-Technology Diesel Exhaust. Research Report 184. Health Effects Institute, Boston.
McDonald, J.D., Zielinska, B., Sagebiel J.C., McDaniel, M.R., Mousset-Jones, P. 2003. Source appor
tionment of airborne fine particulate matter in an underground mine. Journal of the Air & Waste
Management Association, 53 (4): 386–395, https://doi.org/10.1080/10473289.2003.10466178.
112
Mills N.L., Törnqvist H., Robinson S.D., Gonzalez M., Darnley K., MacNee W., Boon, N.A.,
Donaldson, K., Blomberg, A., Sandstrom, T., Newby, D.E. 2005. Diesel exhaust inhalation causes
vascular dysfunction and impaired endogenous fibrinolysis. Circulation 112(25):3930–6. https://doi.
org/10.1161/CIRCULATIONAHA.105.588962.
MSHA 2016. Carbon (Diesel Particulate Matter) by Thermal/Optical Analysis. MSHA P-13. Revision
#003, U.S. Department of Labor. Mine Safety and Health Administration.
MSHA 2020. Particulate sampling methods. In: Health Inspection Procedures Handbook. PH20-V-4.
U.S. Department of Labor, Mine Safety and Health Administration. Available from: https://arlweb.
msha.gov/READROOM/HANDBOOK/PH20-V-4.pdf (Accessed on August 15, 2022.).
MSHA 2022. The DPM personal sampling compliance data. The United States Department of Labor.
Mine Safety and Health Administration (MSHA). https://arlweb.msha.gov/OpenGovernmentData/
OGIMSHA.asp. (Accessed on August 11, 2022).
Nejad S.H., Takechi, R., Mullins, B.G., Giles, C., Larcombe A.N., Bertolatti, D., Rumchev, K.,
Dhaliwal, S., Mamo J. 2013. The effect of diesel exhaust exposure on blood-brain barrier integrity and
function in a murine model. Journal of Applied Toxicology 35:41–47. https://doi.org/10.1002/jat.2985.
NIOSH 2003. Diesel Particulate Matter (as Elemental carbon). Method 5040. In: NIOSH Manual of
Analytical Methods. 4th ed. Cincinnati, OH: U.S. Department of Health and Human Services, Centers
for Disease Control and Prevention, National Institute for Occupational Safety and Health.
NIOSH 2016. Monitoring diesel exhaust in the workplace. In: NIOSH Manual of Analytical Methods
(NMAM), 5th Edition, Chapter DL, Cincinnati, OH: U.S. Department of Health and Human Services,
Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health. Avail
able from: https://www.cdc.gov/niosh/nmam/pdfs/NMAM_5thEd_EBook.pdf. Accessed on July 2, 2022.
Noll, J., Birch, M.E. 2008. Effects of sampling artifacts on occupational samples of diesel particulate
matter. Environ. Sci. Technol. 42:5223–5228. https://doi.org/10.1021/es702883k.
Noll, J., Cauda, E., Vanderslice, S.; Barone, T. 2019. Quantification of the effects of carbon on filter
media in SKC DPM cassettes on measurements of diesel particulate matter in underground mines.
SME Annual Meeting.
Noll, J., Gilles. S., Wu, H.W., Rubinstein. E. 2015. The relationship between elemental carbon and diesel
particulate matter in underground metal/nonmetal mines in the United States and coal mines. In Aus
tralia Journal of Occupational and Environmental Hygiene 12: 205–211. https://doi.org/10.1080/
15459624.2014.960577.
Noll, J.D., Timko, R.J., McWilliams, L., Hall, P., Haney, R. 2005. Sampling results of the improved
SKC diesel particulate matter cassette. Journal of Occupational and Environmental Hygiene, 2:1,
29–37, https://doi.org/10.1080/15459620590900320.
OHSA 2017. Occupational Health and Safety Act, Ontario Regulation 854. Mines and Mining Plants.
Ontario Ministry of Labor.
Olson, B. 2001. Particle calibration of the disposable personal diesel aerosol sampler (PDAS). Report
submitted to the National Institute for Occupational Safety and Health. Minneapolis, MN: University
of Minnesota, Particle Calibration Laboratory.
Peters S., de Klerk, N., Reid, A., Fritschi L., Musk, A., Vermeulen, R. 2017. Quantitative levels of diesel
exhaust exposure and the health impact in the contemporary Australian mining industry. Occup
Environ Med. 74(4):282–289. https://doi.org/10.1136/oemed-2016-103808.
Pronk, A., Coble, J., Stewart, P.A. 2009. Occupational exposure to diesel engine exhaust: A literature
review. J. Exp. Sci. Envir. Epid. 19:443–457 (2009).
Ramachandran, G., Watts, W. 2003. Statistical comparison of diesel particulate matter measurement
methods. AIHA Journal, 64 (3):329–337, https://doi.org/10.1080/1542811030898482.
Rubeli, B., Gangal, M., Butler, K., Alder, W. 2004. Evaluation of the contribution of light-duty vehicle to
the underground atmosphere diesel emissions burden. In: Ganguli R, Bandopadhyay S, eds. Proceed
ings of the 10th U.S./North American Mine Ventilation Symposium. Leiden, Netherlands: Balkema.
Ruehl, C., Herner, J.D., Yoon, S., Collins, J.F., Misra, C., Na, K., Robertson, W.H., Biswas, S.,
Chang, M.-C.O., Ayala, A. 2015. Similarities and differences between traditional and clean diesel PM.
Emiss Control Sci Technol 1:17–23.
Saarikoski, S., Salo, L., Bloss, M., Alanen, J., Teinilä, K., Reyes, F., Vázquez, Y., Keskinen, J.,
Oyola, P., Rönkkö, T., Timonen, H. 2019. Sources and characteristics of particulate matter at five
locations in an underground mine. Aerosol and Air Quality Research, 19:2613–2624, https://doi.org/
10.4209/aaqr.2019.03.0118.
Saarikoski, S., Teinilä, K., Timonen, H., Aurela, M., Laaksovirta, T., Reyes, F., Váasques, Y.,
Oyola, P., Artaxo, P., Pennanen, A.S., Junttila, S., Linnainmaa, M., Salonen, R. O., Hillamo, R.
2018. Particulate matter characteristics, dynamics and sources in an underground mine. Aerosol. Sci.
Technol. 52(1):114–122. doi:10.1080/02786826.2017.1384788.
113
Salo, L., Rönkkö, T., Saarikoski, S., Teinilä, K., Kuula, J., Alanen, J., Arffman, A., Timonen, H., Keskinen, J.
2021. Concentrations and size distributions of particle lung-deposited surface area (LDSA) in an under
ground mine. Aerosol and Air Quality Research 21, 200660. https://doi.org/10.4209/aaqr.200660.
SKC 2022. DPM Cassette with impactor Cat. No. 225–317 – Operating Instructions. Form 400081 Rev
1808. SKC Inc. Eighty Four, PA. https://www.skcltd.com/images/pdfs/40081.pdf. (Accessed on
August 10, 2022).
Skubacz, K., Wojtecki, Ł., Urban, P. 2017. Aerosol concentration and particle size distributions in
underground excavations of a hard coal mine, International Journal of Occupational Safety and Ergo
nomics, 23:3, 318–327, https://doi.org/10.1080/10803548.2016.1198553.
Susanto, A., Purwanto, P., Sunoko, H.R., Setiani, O. 2018. Assessment of diesel particulate matter
exposure of underground miners in Indonesia. Journal of Ecological Engineering 19(4): 34–42, https://
doi.org/10.12911/22998993/89671.
Wang, Y., Chung, A. Paulson, S.E. 2010. The effect of metal salts on quantification of elemental and
organic carbon in diesel exhaust particles using thermal-optical evolved gas analysis. Atmospheric
Chemistry and Physics, 10(23): 11447–11457. https://doi.org/10.5194/acp-10-11447-2010.
Yang, H. and Yu, J.Z. 2002. Uncertainties in charring correction in the analysis of elemental and organic
carbon in atmospheric particles by thermal/optical methods. Environmental Science & Technology, 36
(23): 5199–5204. https://doi.org/10.1021/es025672z.
114
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
M. Torkmahalleh
University of Illinois, Chicago, IL, USA
1 INTRODUCTION
Diesel-powered equipment is widely used in underground mining because of its high efficiency,
flexibility during developpment stage, reliability, and durability (Bugarski et al., 2012). Combustion
of diesel is the main source of the toxic gases and particles in underground metal and non-metal
mines. When underground miners, particularly truck drivers, continuously breathe contaminated
air for a prolonged length of time, severe occupational disorders may develop. US Federal M/NM
Mine Regulations have set a limit of 160 μg/m3 of total carbon (TC) for an eight-hour time weight
average (TWA) (OSHA-MSHA). TC value is obtained by summing the EC and OC numbers
together, and it generally accounts for 80 percent of the DPM (Kimbal et al, 2012). In addition to
toxic gases, diesel engines also release ultrafine particles (UFP) with a size of less than 0.1 µm. Their
accumulation in respiratory system is hazardous and can result in severe respiratory illnesses. When
evaluated in various environmental circumstances, the Lung Deposited Surface Area Concentration
(LDSA) approach was successful in recognizing UFP (Braakhuis et al. 2014, Kuuluvainen et al.
2016, Kuula et al. 2020, Tran et al. 2020, Afshar-Mohajer et al. 2020, Huynh et al. 2018). Ultrafine
particles (UFPs) can carry high concentration of toxic substances, and therefore, they can cause
inflammation through oxidative stress responses, atherosclerosis, increase of blood pressure, and
even myocardial infraction (Delfino et al. 2005).
The Lung Deposited Surface Area Concentration (LDSA) is a different way of describing
particle toxicity. The challenge with current mass-based exposure limits is that they fail to
account for ultrafine particles (less than 100 nanometres), which has negligible contributions
DOI: 10.1201/9781003429241-12
115
to particle mass but can penetrate deep into human airways and lodge in lung alveoli and
have high toxicity contributions. This is especially concerning because most diesel engine par
ticle emissions (measured by number) fall into the ultrafine range (Braakhuis et al., 2014). In
distinct metropolitan contexts, LDSA concentrations, size distributions, and height profiles
have been measured in ambient conditions (Kuuluvainen et al., 2016; Kuula et al., 2020; Tran
et al., 2020). LDSA concentrations have been connected to emissions from combustion
sources. Afshar-Mohajer et al. (2020) used a variety of measures, including LDSA, to investi
gate the variability of aerosol concentrations in different processing zones of a taconite mine.
Huynh et al. (2018) used numerous particle concentration measures, including LDSA, to
investigate fine particle concentrations in six taconite mines. LDSA concentrations in their
investigation ranged from 50 to 300 μm2/cm3 depending on the processing region. Although
LDSA level varied from station to station, Afshar-Mohajer et al. (2020) also noticed the
remarkable repeatability of concentration each day. The aim of this experimental sampling
was to determine the exposures of the cabin operator to diesel exhaust nanoparticles.
2 SAMPLING METHODOLOGY
The study was carried out in a polymetallic mine in the Eastern Kazakhstan. Diesel exhaust
samples were collected at the level +335m, which has 450 tons/day of total material moved.
The mine is operating 365 days a year and working in two 11-hour shifts each day, with two
mid-shift blasting breaks. The mine employs a push-pull ventilation system with two main
fans and several auxiliary fans within mine working faces.
Measurements were conducted in an area of the active ore heading, requiring the use of the
LHD loader CAT R1700 (engine model Cat@C11 ACERT, 241 kW) operated in a 70 meters
length drift which is 4.3 meters in height and 4.2 meters in wide (Figure 1). Average air vel
ocity in the drift was around 0.5 m/s. The auxiliary fans was on during the active ore haulage
process to pull fresh air through the two ventilation ducts. Figure 1 shows the positioning of
LHD loader during its operational process.
During measurements, the LHD diesel loader was operating in its operational cycle: load
(~18 seconds), haul (~41 seconds), and dump (~7 seconds). The main source of diesel fumes
was only an exhaust pipe of the LHD loader.
The sampling was conducted by the Naneos Partector 2 within 12 minutes in the operator
open cabin of the mine face LHD loader working in the active ore heading area. The loader
engine corresponds to the Tier 3 or Stage IIIA Equivalent Engine and exploits the high-
quality diesel fuel DT-L-K2 (standard GOST 305-82), which in the first approximation corres
ponds to the Euro 3 standard and differs only in the cetane number.
116
The Naneos Partector 2 multimeric nanoparticle detector was used to measure the Lung Depos
ited Surface Area (LDSA) and particle diameter. Prior to sampling, the collecting tape was
replaced, and all required calibrations were completed. During the sampling time temperature was
24° C, atmospheric pressure 964 hPa and humidity 53%. During ore haulage, the Naneos Partec
tor 2 was placed inside the cabin with the suction duct taped to the loader roof frame at the oper
ators face level (approximately 2.3 m from the floor). This was done for better understanding the
accurate exposure level of the workers. The LHD loader was not equipped with a sealed cab
(environmental cab) to protect the workers from the toxic gases or aerosols. Therefore, the expos
ure levels indicated by the Partector 2 assumed to be equal to the actual exposure of the driver.
As a result of Naneos Partector 2 produced from 700 samplings (every second), the averaged
data for Lung Deposited Surface Area (LDSA) was 7 470 µm²/cm³ with minimum of 1 657
µm²/cm³ and maximum of 12 792 µm²/cm³ (Figure 2). The LHD loader diesel engine was idle
running for the first 150 seconds. This can be observed from the graphs (Figure 2) where
LDSA are around 2000-4000 µm²/cm³. Then the LHD started its regular operational cycle.
Indicated peaks are basically related when the LHD engine increased revolutions per minute
for its operational processes and also the loader position in the drift.
117
Figure 3. Particles diameter over elapsed time.
4 CONCLUSIONS
Experimental sampling of diesel exhaust nanoparticles using a Naneos Partector 2 instrument was
conducted in the underground polymetallic mine. This sampling was conducted in the operator
open cabin of the mine face LHD loader R1700 (engine model Cat@C11 ACERT, 241 kW, Tier
3/Stage IIIA Equivalent Engine) working in the active ore heading area. The aim of this experi
mental sampling was to determine the exposures of the cabin operator to diesel exhaust nanopar
ticles. The LHD loader operated in a 70 meters length drift (4.3m x 4.2m), where the auxiliary
ventilation velocity was around 0.5 m/s. As a result, the averaged data for LDSA and particles
diameter were found 7 470 µm²/cm³ and 90 nm, re-spectively. The LDSA results were compared
to the sampling produced out of the cabin in the same drift for a longer time period, which were
almost twice lower than in the cabin. However average particle diameter left unchanged 90 nm.
This demonstrates that LHD loader open cabin operators are more vulnerable to diesel exhaust
nanoparticles. The results of this study suggest that the Naneos Partector 2 sampling instrument
can be employed in polymetallic mines for monitoring workplace surveillance. This study offers
important data on particle surface area. As the surface contact between the particles and human
cells is crucial, particle surface area might be a new approach to investigate particle toxicity.
ACKNOWLEDGMENT
This study was supported by Nazarbayev University Grant Program: FY2020-СRP-1 Collab
orative Research Project # 091019CRP2104.
REFERENCES
Afshar-Mohajer, N., Foos, R., Volckens, J., Ramachandran, G. 2020. Variability of aerosol mass and
number concentrations during taconite mining operations. Applied Occupational and Environmental
Hygiene. 17: 1–14. https://doi.org/10.1080/15459624.2019.1688823.
Black J, Reid D (2017) Delivering a healthy atmosphere underground in a cost-conscious environment,
Mine Ventilation Journal, ISBN 978 1 925100 617: 159–167.
118
Braakhuis, HM, Park, MV, Gosens, I, De Jong, WH, Cassee, FR, 2014. Physicochemical characteristics
of nanomaterials that affect pulmonary inflammation, Particle and Fibre Toxicology, vol. 11, no. 1, p.
18.
Cauda, E.G., Ku, B.K., Miller, A.L., Barone, T.L. 2012. Toward developing a new occupational expos
ure metric approach for characterization of diesel aerosols. Aerosol. Sci. Technol. 46 (12): 1370–1381.
Hennig, F., Quass, U., Hellack, B., Küpper, M., Kuhlbusch, T. A. J., Stafoggia, M., & Hoffmann, B.
2018. Ultrafine and Fine Particle Number and Surface Area Concentrations and Daily Cause-Specific
Mortality in the Ruhr Area, Germany, 2009–2014., Environmental Health Perspectives, 126 (2),
027008. doi:10.1289/ehp2054.
Hussain S, Boland S, Baeza-Squiban A, Hamel R, Thomassen LCJ, Martens JA, et al. 2009. Oxidative
stress and proinflammatory effects of carbon black and titanium dioxide nanoparticles: role of particle
surface area and internalized amount. Toxicology 260(1-3):142–149, PMID: 19464580, https://doi.org/
10.1016/j.tox.2009.04.001.
Huynh, T., Ramachandran, G., Quick, H., Hwang, J., Raynor, P. C., Alexander, B. H., & Mandel, J. H.
2018. Ambient fine aerosol concentrations in multiple metrics in taconite mining operations. Annals of
Work Exposures and Health, 63(1): 77–90. https://doi.org/10.1093/annweh/wxy086.
Khan, M. U., Gillies, A. D. 2019. Diesel particulate matter: Monitoring and control improves safety and
air quality. Advances in Productive, Safe, and Responsible Coal Mining: 199–213. https://doi.org/
10.1016/b978-0-08-101288-8.00009-2.
Kimbal, K C, Pahler, L, Larson, R and Vanderslice, J, 2012. Monitoring diesel particulate matter and
calculating diesel particulate densities using grimm model 1.109 real-time aerosol monitors in under
ground mines, Journal of Occupational and Environmental Hygiene, 9(6):353–361.
Kuula, J., Kuuluvainen, H., Niemi, J.V., Saukko, E., Portin, H., Kousa, A., Aurela, M., Rönkkö, T.,
Timonen, H. 2020. Long-term sensor measurements of lung deposited surface area of particulate
matter emitted from local vehicular and residential wood combustion sources, Aerosol Sci. Technol.
54: 190–202. https://doi.org/10.1080/02786826.2019.1668909.
Kuuluvainen, H., Rönkkö, T., Järvinen, A., Saari, S., Karjalainen, P., Lähde, T., Pirjola, L.,
Niemi, J. V., Hillamo, R., Keskinen, J. 2016. Lung deposited surface area size distributions of particu
late matter in different urban areas. Atmos. Environ. 136: 105–113. https://doi.org/10.1016/j.
atmosenv.2016.04.019.
OSHA-MSHA. Diesel Exhaust/Diesel Particulate Matter’, December 2013. https://www.osha.gov/Publi
cations/OSHA-3590.pdf.
Ramachandran, G., Paulsen, D., Watts, W., Kittelson, D., 2005 Mass, surface area and number metrics
in diesel occupational exposure assessment. Environmental Monitoring, 7 (7): 728–735. https://doi.org/
10.1039/b503854e.
Sabanov, S, Zeinulla, A, Brune, J & Amouei Torkmahalleh, M 2021, ‘The results of the evaluations of
the instrumentation used to monitor concentrations of DPM’, in Proceedings of 18th North American
Mine Ventilation Symposium: 133–138.
Sioutas, C., Delfino, R. J., & Singh, M. 2005. Exposure assessment for atmospheric ultrafine particles
(ufps) and implications in Epidemiologic Research. Environmental Health Perspectives, 113
(8),947–955. https://doi.org/10.1289/ehp.7939.
Tran, P.T.M., Ngoh, J.R., Balasubramanian, R. 2020. Assessment of the integrated personal exposure to
particulate emissions in urban micro-environments: A pilot study. Aerosol Air Qual. Res. 20: 341–357.
https://doi.org/10.4209/aaqr.2019.04.0201.
119
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
A.A. Habibi
Ventilation Expert, PT Freeport Indonesia
K.O. Homan
Associate Professor, Missouri University of Science and Technology, MO, USA
A.D. Bugarski
Pittsburgh Mining Research Division, National Institute for Occupational Safety and Health,
Pittsburgh, Pennsylvania, USA
1 INTRODUCTION
Diesel engines are the main source of carbon-rich submicron aerosol in underground mines.
Other sources of carbon-rich aerosols are oil mist from drilling operations, cigarette smoke,
and suspended carbonate dust in ore (Zielinska et al., 2002; McDonald et al., 2003; Noll et al.,
2006). Diesel aerosols are classified in three modes: nucleation mode (3-30 nm), accumulation
mode (30-500 nm), and coarse mode (>500 nm) (Kittelson et al. 1998). Due to small size, the
nucleation mode aerosols contribute substantially to number concentrations, but marginally
to the particle total mass (Bugarski et al., 2012). In the U.S., the Mine Safety and Health
Administration (MSHA) limits exposure of underground metal and nonmetal miners to diesel
particulate matter (DPM) to 160 µg/m³ of total carbon (TC) (30 CFR 57.5060). The TC con
centrations are determined using NIOSH Method 5040 (NIOSH 2016). The mining industry
implemented various engineering controls and technologies to reduce the exposure (Bugarski
et al. 2014, Bugarski et al. 2022). A detailed knowledge of the size, structure, morphology,
fractal dimension, and composition could be used to identify the sources and processes guid
ing formation, transformation of diesel aerosols. The same knowledge can be used for identi
fying effective control strategies and technologies, source apportionment, and to understand
the interaction between micron-size dust particles and DPM.
In order to achieve the objectives, size segregated sampling using a multi-stage impactor
with Transmission Electron Microscopy (TEM) grids was conducted. The sampling was
DOI: 10.1201/9781003429241-13
120
conducted during different mining operation phases to collect a range of diverse samples.
Additionally, interaction of dust and diesel agglomerates can be investigated using Energy
Dispersive Spectroscopy (EDS) and microscopic image analysis.
2 METHODOLOGY
2.1 Instrumentation
Real time instrumentation (Table 1) was used to measure the particles size distributions and
number concentrations based on aerodynamic mobility (ELPI). The Electrical Low Pressure
Impactor (ELPI+) was used to collect size-segregated samples for S/TEM analysis (Table 1).
Electrical Low Pressure Impactor (Dekati, ELPI), Continuous measurement of number concentra
(ELPI used for real time readings) tions and size distributions of aerosols with aero
dynamic diameters between 30 and 10,000 nm
Electrical Low Pressure Impactor (Dekati, ELPI +) Collection of size-segregated samples for S/TEM
(ELPI+ is a newer model and was used to collect
S/TEMgrid samples)
Figure 1. A) S/TEM sampling grids taped on the 25-mm aluminum foil. The soot traces are visible. b)
ELPI+ components disassembled for exchanging the substrates between the tests.
121
2.3 Sample collection
The measurements and sampling were conducted downwind of a 300m long active work zone
shown in Figure 2. The measurement and sampling station was located between of the point
where four ventilation drifts merged and the flow regulator.
The total flowrate through the zone was 26 – 33 m3/s (55,000 – 70,000 cfm). During the
measurement and sampling periods, the type of vehicles and corresponding activities were
logged. During that period, the relatively large number of light-duty (LD) vehicles was used to
deliver material and manpower . The heavy-duty (HD) equipment which was concurrently
used in the work zone is listed in Table 2.
The ELPI was used for continuous real-time monitoring of number concentrations and size
distribution of aerosols caried by ventilation air to the measurement/sampling station. Con
currently, the ELPI+ was used collect the size segregated samples for S/TEM and EDS ana
lyses. The sampling station set-up are shown in Figure 3 a and 3b.
The results from six tests were used to study the impact of equipment activities on the aerosols
number and size distributions. The vehicle log showed that the exhaust of the over 35 LD and
6 HD vehicles contributed to the aerosols concentrations. The filtration system consisted of
122
Figure 3. Sampling station set-up. a) ELPI and ELPI+, b) Sampling station.
DFEs, tier 3 engine with diesel oxidation catalyst convertor (DOC) and tier 3 engine with
active regeneration system (SMF-AR). Number and mass concentration measurements
The number and mass concentration of aerosols were measured by ELPI instrument. The
results are for all six tests are shown in Figure 4. The number and mass concentrations fluctuated
substantially during all periods. This appears to be the consequence of diversity in the tasks per
formed by the HD and LD equipment. During Test 1, two HD pieces of equipment retrofitted
with SMF after treatment systems and several LD equipment were used most heavily. As shown
in Table 2, the sixe HD vehicles included those with DFP filtration systems and SMF-AR. The
high mass concentrations of predominantly micron size aerosols is attributed to high activity of
the SMF-equipped vehicles that exhausted relatively low number concentrations of submicron
aerosols but entrained substantial quantities of micron-sized dust. The results for test 2 – 6 show
the relatively higher number concentrations of submicron aerosols along with proportionally high
mass concentration of submicron aerosols primarily attributed to the diesel emissions of the HD
and LD vehicles operated upwind and in the zone.
123
aerosols with count median diameters (CMD) ranging from 160 nm to 210 nm and freshly gen
erated aerosols mainly in nucleation modes and partially in agglomeration modes with CMD
ranging from 27 and 166 nm. The mechanically generated particulates (coarse particulate) con
tributed substantially less to number concentration compared to combustion generated aerosols.
The number concentrations were noted to be higher when the diesel equipment was operated
closer to the sampling station. During all six sampling periods, presence of aged aerosols trans
ported from the outby areas was noted. The number distributions of aerosols at the sampling
station represented a mixture of aged aerosols transported with ventilation air from outby areas
and fresh aerosols generated in the zone. These distributions were mostly bimodal.
ELPI+ was used to collect TEM grids to conduct microscopic morphological and image
analysis. Thirty-five samples were collected during six sampling periods. S/TEM analysis was
conducted to obtain Dprojected for the collected agglomerates. The results show that size distri
bution obtained from S/TEM analysis are single modal and normally distributed. Three
counts of coarse agglomerates (larger than 1um) were sampled.
The mass size distributions were dominated by entrained dust generated by movement of
diesel-powered and battery powered equipment. The mass median diameter (MMD) ranging
from 2.6 µm to 30 µm was measured. The Mass concentrations were mostly normally distrib
uted except in two cases. The diesel particulates contributed to the second modes. In the case
of coarse mode, the lognormal distribution was obtained by extrapolation and fitting the log-
normal curves (Chimera Technologies 2009). Figure 5 shows the normalized mass size distri
bution during Test 6. The high spike at 2056 s was noted to be the result of entrained dust by
an equipment operating close to the sampling station. Figure 5 a and b show size and mass
distribution ELPI measurement and Dprojected STEM results.
Figure 5. A) ELPI size distribution and Dprojected from STEM analysis. b) mass size distributions
measured at the different instances during the Test 6.
124
without DFE). These results therefore show that larger entrained particles and aged agglomer
ates have contributed to larger fractal dimensions.
Shape factor is used to describe the agglomerates elongation. If shape factor is close to 1.0,
it indicates that the agglomerates are spherical, where 0.1 shape factor is an indicative of long-
chained or elongated agglomerates (Nord et al., 2004; Mustafi and Raine, 2009). The results
show a majority of agglomerates with shape factor of higher than 0.5. Figure 6 (a) and (b)
show the results for fractal dimension and shape factor.
Figure 7. TEM BF image of primary particles (a) with graphitic structure with multiple nuclei centers
and (b) amorphous structure.
125
The diameter of carbon-rich primary particles dp are normally distributed for agglomerates
with graphitic structure with mean value of 23.6 nm and s = 6.3 nm and for the agglomerates
with catalyst with mean value of 24.2 nm and s = 6.2 nm. The dp agglomerates with amorph
ous structure were bi-modal.
Figure 8. S/TEM image analysis a) STEM BF of particles at 400K magnification b) SE image of par
ticles at 300K magnification c) STEM BF of volatile interface at time 0 second d) STEM BF of volatile
interface at time lag 20 seconds.
126
The EDS results show agglomerate #2, is carbon rich. Traces of Na and Si did not continue
to agglomerate #2 unlike S. The EDS result for agglomerate #1 showed the trace of S net
counts was insignificant, unlike agglomerate #2. Similarly, the Si, Al, Mg and Na net counts
were either similar or slightly lower in agglomerate #1. This suggests that agglomerate #1 was
carbon-rich, volatile, and from a different source than #2. The morphological S/TEM analysis
of the particles mentioned above showed a graphitic structure for the primary particle of
chain-like agglomerate #2. However, an amorphous structure was observed for primary par
ticles in agglomerate #1.
The TEM and SE images of agglomerate #2 show small particles on the primary particles.
The EDS results indicate Na and S elements surrounding the primary particles. The diameter
of primary particles for agglomerate #2 ranged from 23 nm to 50 nm. For agglomerate #1,
the primary particles appear to be smaller than 30 nm. Agglomerate #2 with the Lmax of
642.81 nm, width of 379.65 nm and calculated SF of 0.59 seemed to have similar characteris
tics to the samples collected in sampling zone. Agglomerate #2 therefore appears likely to be
a product of the diesel combustion process from equipment in the sampling zone. The
attached Na particles represent the dust in the uncontrolled zone whereas S is likely to be
from the engine. As for agglomerate #1, projected measurements were difficult to obtain as
some of the volatile primary particles evaporated while EDS mapping was underway. How
ever, the combined results and observations of morphological and EDS mapping make it
likely that agglomerate #2 did not have a combustion source.
4 CONCLUSION
The present study was designed to evaluate importance of monitoring and sampling the
aerosols emitted by diesel-powered equipment performing multiple tasks in the section of
underground mine. The study showed that entrainment of the micron sized dust due to
movement of diesel-powered equipment is the primary contributor to mass concentra
tions of particulates at the section. The mass median diameter of entrained dust was
found to range between10 µm to 30 µm. The dominant sources of the submicron aerosols
were various types of LD and HD diesel-powered equipment. Diesel aerosols were found
to be distributed between nucleation and agglomeration modes. We observed that fresh
aerosols were distributed in single mode and aged aerosols were distributed bimodaly.
The electron microscopy analysis of size segregated aerosols samples revealed that diesel aero
sols agglomerates consist of three types of carbon-rich primary particles. with the majority of
those are made of carbon-rich primary particles with graphitic structure. The smaller part is made
of carbon-rich primary particles with amorphous structure or carbon-rich primary particles
infused with catalyst. The diameter of primary particle with amorphous structure is bi-modally
distributed.
The EDS results along with morphological attributes were used to investigate chemical compos
ition of the fractal agglomerates attached to the micron size particles. EDS results show that the
diesel chain-like agglomerates with high carbon to oxygen ratio and significantly low net counts
of dust and mining environment elements such as Na, Si, Fe, K and S. Volatile chain-like agglom
erates with amorphous primary particles seemed susceptible to beam damage. EDS results for
these agglomerates showed high C net counts and equal or slightly lower Na, Si and K.
The EDS results show that the typical elements (Na, S either from fuel or not, Si) were
observed in lower stages of ELPI+. The elements were observed at very low net counts during
control zone test and higher net counts during regular mining activity due to the presence of
dust. High C/O net counts ratio for chain-like agglomerates varied from 6.5 to 9.6. Most of the
chain-like agglomerates appeared to have chain-like primary particles with graphitic structure.
The interaction of dust and volatile chain-like agglomerate was investigated at higher stages of
ELPI+ multi-stage impactor. It was observed that amorphous carbon rich particles tend to be
present at the interface of dust and chain-like agglomerates. The primary particles at the interface
with large particles were sensitive to beam damage and evaporated during EDS mapping.
127
The sampling during mining activity provided valuable data to compare the results with
controlled zone sampling and investigate: i. The interaction of carbon rich agglomerates at the
interface with large dust particles. ii. The morphological differences between the agglomerates.
iii. Investigate the volatile agglomerates observed during the sampling to determine the
sources of agglomerates.
5 DISCLAIMER
The findings and conclusions in this manuscript are those of the authors and do not necessar
ily represent the official position of the National Institute for Occupational Safety and Health
(NIOSH), Centers for Disease Control and Prevention (CDC). Mention of company names or
products does not constitute endorsement by NIOSH or CDC.
REFERENCES
Bugarski, A.D., Janisko, S.J., Cauda, E.G., Patts, L.D., Hummer, J.A., Westover, C., Terrillion, T. 2014.
Aerosols and criteria gases in an underground mine that uses FAME biodiesel blends. Ann. Occup.
Hyg. 58(8):971–982. http://dx.doi.org/10.1093/annhyg/meu049.
Bugarski, A.D., Hummer, J.A., Vanderslice, S., Mischler, S. 2020. Contribution of various types and cat
egories of diesel-powered vehicles to aerosols in an underground mine. Journal of Occupational and
Environmental Hygiene 17: 121–134. https://doi.org/10.1080/15459624.2020.1718157.
NIOSH 2016. Monitoring diesel exhaust in the workplace. In: NIOSH Manual of Analytical Methods
(NMAM), 5th Edition, Chapter DL, Cincinnati, OH: U.S. Department of Health and Human Services,
Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health. Avail
able from: https://www.cdc.gov/niosh/nmam/pdfs/NMAM_5thEd_EBook.pdf. Accessed on July 2, 2022.
Zielinska B, Sagebiel J, McDonald J, Rogers CF, Fujita E, Mousset-Jones P (2002). Measuring diesel emis
sions exposure in underground mines: a feasibility study. Research directions to improve estimates of
human exposure and risk from diesel exhaust, Special report. Boston: Health Effects Institute, pp. 181–232
Kittelson, 1998, D. B. Engines and nanoparticles: A review. J. Aerosol. Sci. 1998, 29 (5-6), 575–588.
Mathis, U., Kaegi, R., Mohr, M., Zenobi, R. (2004). TEM analysis of volatile nanoparticles from particle
trap equipped diesel and direct-injection spark-ignition vehicles. Atmospheric Environment 38 (2004)
4347–4355
Nord, K., Haupt, D., Ahlvik, P., and Egeback, K.-E. (2004). Particulate Emissions from an Ethanol
Fueled Heavy-Duty Diesel Engine Equipped with EGR, Catalyst and DPF, SAE Paper 2004-01-1987
128
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J. Stachulak
MIRARCO Mining Innovation, Sudbury, Ontario, USA
ABSTRACT: The introduction of diesel particulate filters (DPF) technology has been an
important step towards controlling diesel emission from heavy-duty diesel -powered under
ground mining vehicles. The Johnson Matthey CRT® (Continuously Regenerating Trap)
technology that combines a diesel oxidation catalyst (DOC) with a DPF has be extensively
used to trap harmful diesel aerosols, CO and hydrocarbons from on-road vehicles. In order to
address concerns over adverse effects of CRT® system on NO2 emissions and allow for use of
these systems in underground mining environments, Johnson Matthey (JM) developed, the
Mining CRT® system which integrates CRT® technology with NO2 abatement components.
The effects of the system on particulate and gaseous emissions were evaluated in the engine
dynamometer laboratory. The evaluation was done using Caterpillar C11, Tier 3 engine oper
ated over the ISO 8178, 8-mode test cycles. The engine was fueled with an ultra–low-sulfur
diesel fuel. The system was found to effectively reduce total particulate mass and number,
total hydrocarbon, and CO emissions. The novel fuel injection and NO2 decomposition cata
lyst were found to be effective in preventing NO2 slip out of the system.
1 INTRODUCTION
This study evaluated the emissions from a Caterpillar C11, Tier 3 engine, obtained with
and without the application of JM Diesel-Particulate-Filter Low NO2-CRT® emission con
trol device. The objective of this evaluation was to provide information on the emission
performance of the DPF for application in Underground mines where increase of NO2 in
the mine environment is undesirable. The testing was performed at the Diesel Emissions
Research Laboratory at CANMET- MMSL, Ottawa. The emission testing was performed
on an engine dynamometer under controlled conditions using the ISO 8178-C1, 8-mode test
cycle.
Measurements were made for the basic engine parameters and exhaust gas concentrations
were measured for carbon monoxide (CO), carbon dioxide (CO2), oxygen (O2), nitric oxide
(NO), oxides of nitrogen (NOX), total hydrocarbons (THC), and diesel particulate matter
(DPM). Measurements were also made for nano-particle distribution for each 8-mode cycle.
Several steady state and transient test cycles were performed to calibrate and fine tune the fuel
injection protocol for emission reduction device.
The device tested showed a significant reduction in exhaust CO (96%), THC (58%), NO2
(55%) and nanoparticle DPM (99%) while CO2 remained nearly unchanged.
DOI: 10.1201/9781003429241-14
129
This study was a prelude to an extensive field evaluation in underground operation at
Vale’s Copper Cliff Mine in Sudbury. The JM’s Mining CRT was installed on Caterpillar
1700 LHD, 263 kW C11 Tier 3. The Mining CRT acquired 7380 hours without the need for
cleaning and replacement while maintaining NO2 emission at engine-out level (Stachulak
et al., 2022). The lessons learned from this successful application of Mining CRT technology
is not part of this paper and will be a topic of the separate publication.
2 EXPERIMENTAL APPROACH
The engine used for the testing was a rebuilt Caterpillar C11, Table 1 provides some engine
specification data
The diesel fuel used for this study was an ultra-low-sulfur diesel fuel with a sulfur value of
11 ppm (CAN/CGSB-3.517-2000). Some of the laboratory analyzed fuel properties are given
in Table 2.
130
Figure 1. Diesel-Particulate-Filter Low NO2-CRT® emission control device and Caterpillar C11 diesel
engine installed in the CANMET-MMSL test cell.
The device was calibrated by the manufacturer’s representative according to both steady
state and transients’ tests.
131
2. Carbon dioxide (CO2) and high carbon monoxide (HCO) model CAI 300 IR, non- disper
sive infrared (NDIR) detection system, CO2 span ranges 0-5.0% and 0-20.0%, and CO
span range 0-1.0%.
3. Oxygen (O2), model CAI 300 Oxygen, paramagnetic (PMA) detection system, span ranges
0-5% and 0-25%.
4. Oxides of nitrogen (NO/NOx), model CAI 400 HCLD heated, chemiluminescence
(CLD) detection system, span ranges 0-1000 ppm and 0-3000 ppm. The gas cart has two
NO/NOx analyzers that provide simultaneous measurements of NO and NOx. The con
centration of NO2 is determined by the difference of NOx and NO concentrations.
5. Total hydrocarbons (THC), model CAI 300 HFID, heated flame ionization detector
(HFID), span ranges 0-100 ppm and 0-1000 ppm.
132
Figure 2. Diesel-particulate-filter low NO2-CRT®.
3 TEST RESULTS
Table 4 provides 8-Mode integrated average values of specific emission data for gases
and DPM.
The percent reduction in emissions compared to engine baseline data is calculated using the
Equation 1, and the results for all devices (Gangal et al., 2012) are shown in Table 4.
Table 4. Specific exhaust emissions for the integrated 8-Mode test cycle, test #7 and #8.
Integrated Specific emissions Baseline Test 8 CRT Test 7 Reduction
The overall DPM conversion of 47.7% reported over the Low NO2 CRT® system is lower
than expected. DPM removal over a ceramic wall flow filter is usually >90%. This lower than
expected conversion over the filter system could be due to Diesel fuel droplets, derived from
133
the fuel injection that takes place between the filter and the NO2 decomposition system, slip
ping from the system and then being measured as DPM. This theory is supported by the rela
tively poor reductions (58%) of THC found over the Low NO2 CRT® system. A THC
conversion of >90% is usually expected over a CRT® system operated at these temperatures
and the higher tailpipe levels found over this “Low NO2” system probably result from the
Diesel that is added after the filter to reduce tailpipe NO2 levels. The high reduction of nano
particles again suggests that the filter is working efficiently and that the measured mass of par
ticulate is largely due to condensed Diesel fuel rather than soot from the engine. It is expected
that this level of Diesel fuel slip can be reduced in future tests by system optimization.
4 SUMMARY
ACKNOWLEDGEMENTS
The authors would like to acknowledge the support and contributions of Vale, Toromont
CAT, Johnson Matthey Ltd. for their help and insight during this work.
REFERENCES
Gangal, M., Rubeli, B., Young, D., Leung, E. & Feres, V. 2012. Performance Evaluation of Diesel Par
ticulate-Filter Low NO2 CRT®.
ISO 8178-1. 1996. 1996(E) PART 1 AND ISO 8178-4:1996(E) PART 4: RECIPROCATING INTERNAL
COMBUSTION ENGINES - EXHAUST EMISSION MEASUREMENT.
Stachulak, J., Gangle, M., Young, D., Watson, K. & McLean, B. 2022. Performance Evaluation of
Diesel Particulate-Filter Low NO2 CRT. In MDEC Annual Conference. Toronto, Ontario, Canada.
134
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
A.A. Habibi
PT Freeport Indonesia, Papua, Indonesia
E. Pinto
Superintendent-Fire and Emission Maintenance, Papua, Indonesia
M. Mardon
Technical Expert-Operations Maintenance, Papua, Indonesia
K. Wijayanto
Maintenance Planner-Operations Maintenance, Papua, Indonesia
C. Rose
Director-Occupational Health and PTFI Health & Safety, Phoenix, Arizona, USA
ABSTRACT: Reducing Diesel Particulate Matter has a direct impact on improving air qual
ity and thus workforce health and safety. PT Freeport Indonesia has selected a multi-stage
reduction strategy to reduce DPM emissions in its underground operations. This paper pro
vides an update on challenges, success and lessons learned during the first two stages of the
Emission Assisted Maintenance (EAM) program, including mobile equipment selection for
trials and instrumentation used for tailpipe measurements. The first stage consists of various
Diesel Particulate Filter trials to measure the effectiveness and feasibility of a large scale roll
out. Use of new engine technologies was trialed in the second stage. Additionally, the PT Free
port Indonesia’s EAM program includes wet and traditional DPF cleaning stations and
a Digital Data Management dashboard to control, monitor and measure success. Lastly, the
paper summarizes the impact of high altitude in engine emissions and DPF life as well as pro
viding insight regarding the opportunity to scale up the EAM program.
1 INTRODUCTION
Emission Assisted Maintenance programs have been integral to the underground mining
industry efforts to reduce exposure of miners to gases and aerosols emitted by diesel-powered
equipment. The long-term data generated through EAM programs have been used to enhance
preventive and corrective maintenance and maintain diesel engine emissions at Original
Equipment Manufacturers specifications and certification levels. These programs were focus
ing not only on keeping in-use emissions under control, preferentially at the certification/
approval levels (30 CFR 75.1914, 30 CFR 57.5066) but also to assure effectiveness of OEM
and retrofit control technologies and strategies (Bugarski et al. 2012) over their useful service
time. The collected data from EAM can be used to help ensure engine health in each interim.
These programs are designed to identify significant increases in emissions due to inadequate
engine maintenance, engine and exhaust aftertreatment aging, system malfunctioning, poison
ing components of an exhaust aftertreatment system, and potentially tampering with engine
and exhaust aftertreatment systems (BHP Billiton 2005, McGinn 2010). The EAM programs
DOI: 10.1201/9781003429241-15
135
reduced the DPM level in the underground mines and proved to be useful in reducing poten
tial downtime needed to accommodate unscheduled repairs and improve availability of diesel-
powered equipment and viability of the underground mining operations.
The implementation of complex engine end exhaust aftertreatment systems in support of
low emissions requirements introduced with EPA Tier 4f/EU Stage IV and EU Stage
V standards resulted in an increased demand for advancements in maintenance practices and
embracing new technologies. Under circumstances where the underground mining diesel fleets
due to implementation of control strategies and technologies contribute less and less pollu
tants to the underground environments, a small fraction of the poorly maintained fleet may
contribute a growing and disproportionate share of the overall pollutants. Therefore, the role
of EAM programs is as important as ever.
136
Figure 1. PTFI mining district overview.
The EAM program is an entity under the PTFI Air Quality Improvement umbrella. The pro
gram is designed to tackle DPM challenges. The committee consists of all engaged stakeholders
as shown in Figure 2. The committee was formed in 2016, with initial tender for the Stage 1 trial
in 4Q 2020 and the official trial starting in 4Q 2022. Figure 2 shows the PTFI air quality matrix.
The DPF trial covered multiple stages. Stage 1 included trialing passive and active DPFs.
The approach allowed the researchers to understand the difficulties and challenges and
troubleshoot. Stage 2 consisted of trialing multiple Tier 4f.
137
Supply Chain, ventilation and support advisors from corporate. In addition, OEMs and DPF
vendors are included in the committee via dotted line.
In addition to the members shown in Figure 3, the committee regularly updated the mem
bers of PTFI up to Executive level on successes and failures. It is important to note that direct
support from senior management has been vital to overcoming major obstacles. The support
consists of Management of Change for field work, data processing and funding. Otherwise,
this project could have been viewed as a disruption.
138
Figure 4. PTFI UG fleet engines.
(such as Getman emulsion, flat bed and mixer trucks) showed lower temperatures due to high
idling time. The trial planned to run the DPFs until they reached their first cleaning cycle. The
goal for LD vehicles was set for 4,000 km and for HD equipment at 3,000 engine hours.
139
and NOx are main indicators of diesel combustion. Because most legacy engines are not
equipped with On Board Diagnostic systems, a regular tailpipe measurement was conducted
on all trial units. This data was used to ensure the health of the engine and DPF/DOC as well
as indicators of any physical damage to the DPF substrate. The low and high idle measure
ments were collected for LD vehicles, and low, high and stall measurements were taken for
HD equipment.
3.1.2 PM measurements
For DPM measurements, PTFI is using the Pacific Data System analyzer (DPM-RT) model 1
and 2. The instruments operates on light-scattering principles to estimate mass of particles.
DPM-RT 2 is equipped with a touch screen display and logging capabilities. PTFI EAM tail
pipe measurements records the DPM concentrations before and after the DPF/DOC package.
The procedure is used based on instrument manufacturer recommendation. The 60 seconds
procedure consists of 20 seconds low idle, 20 seconds high idle (3200 rpm for LV and
2300 rpm for HD) followed by the last 20 seconds low idle. The device provides and average
results for the 60 seconds, as well as individual readings.
In addition, DPM committee has acquired Nanoparticle Emission Tester model 3795
(NPET) manufactured by TSI. This instrument is to be used on new engines (T4F/Stage V).
NPET uses the principle of a condensation particle counter (CPC) to measure particle concen
tration in the tailpipe. Assuming a mean particle diameter, the particle number concentrations
can be converted to mass concentrations (TSI 2021). The decision was due to repeated zero
DPM-RT tailpipe measurements indicating low mass concentrations and high particle
number concentrations.
140
troubleshooting. Lastly, Figure 9 shows NOx ratio (NO2/NOX) to ensure catalysts are not
adding to NO2 levels in underground workings. The results show NO2 levels are either
decreasing or staying at pre-DPF levels, and the DPF installation has not affected the level
adversely. An example will be discussed in the next section.
141
• The deterioration of exhaust piping was found after two months on trial units. PTFI has
opted to use stainless steel for all DPF accessories and mufflers.
• It is vital for the engine to be tuned up prior to DPF installation.
Figure 6. Light Vehicle Toyota DPF photos, dirty and clean sides.
142
Figure 7. Comparison of LD vehicles’ gas emissions.
143
• DPF A was installed on AD55
• DPF B was trialed on AD30.
4.2.4 R1600H
DPF A failed prematurely repeatedly after only 50 to 100 hours. The investigation showed
high back pressure was recorded; however, temperature readings of over 350 degrees C were
reached more than 75 percent of the time.
Since no evident reason was found, the RCA investigated the impact of elevation on tailpipe
emissions. Tailpipe measurements were collected at sea level and site level (2,500 m to
2,900 m above sea level). The results show a significant increase of tailpipe emissions on CO,
NOx, and CO2 except AD60 and 1700k. Figure 9 shows the percent variance for each emis
sion. This issue was reported to the OEM engine manufacturer. The DPM committee is
exploring other possible solutions for R1600H in 2023.
144
Figure 10. PTFI cleaning station.
5 CLEANING STATION
Due to the remoteness of the PTFI jobsite, PTFI has established an inhouse cleaning station.
The cleaning station is in operation and equipped with traditional and washing solutions. The
committee has opted to trial both approaches individually and in tandem to determine the
impact of each approach to the life of DPFs. Figure 10 shows the cleaning station. A digital
dashboard is required to be built to track the life of DPFs for this purpose. The committee
believes this approach is to be completed by the end of 2023.
6 CONCLUSION
A combination of DPF retrofit installation and modern technology engines were trialed at the
PTFI jobsite. The results from LD vehicles were found to be positive with the DPF lasting up
to 4,000 km. This successful trial resulted in installation of an additional 150 DPFs across the
LD fleet. Additionally, the trial of new T4F Toyota LVs showed significant reduction in CO
and PM concentrations. The PTFI LD fleet is expected to be replaced or be equipped with
DPF by the end of 2025.
The HD DPF trial was more challenging. The trial was successful on AD30 and AD55
trucks. LHD 2900G resulted in positive results from two different DPF vendors. DPF
B worked without any issues on R1700G, and DPF C appears to be the solution for Getman
MEQ. The R1600H trial was not successful, and the DPM committee is exploring other
options as part of 2023 plan.
The trial of T4F HD LHDs shows significant reduction in emissions; however, the reliabil
ity of MEQ appears to be low, and maintaining the LHD in operational conditions has been
challenging. The T4F LHDs has not been able to achieve the target hours yet. PTFI has been
working with the OEM and is committed to resolve the reliability issues.
The PTFI EAM process is moving forward with a trial on five other types of MEQ. The
program is aiming to tackle more outby equipment, mainly used for service, transportation
and goods delivery. These efforts are aligned with the PTFI five-year DPM reduction forecast
to improve working conditions in the underground.
The PTFI DPM committee consists of members from all departments. It is crucially import
ant to note that direct support from the highest levels management has been and will be vital
to the success of this project.
REFERENCES
30 CFR 57.5066. Safety and Health Standards. Underground metal and nonmetal mines. Maintenance
standards. Code of Federal regulations. Washington, DC: U.S. Government Printing Office, Office of
the Federal Register.
145
30 CFR 75.1914. Mandatory Safety Standards. Underground coal mines. Maintenance of diesel-powered
equipment. Mine Safety and Health Administration. Code of Federal regulations. Washington, DC:
U.S. Government Printing Office, Office of the Federal Register.
BHP Billiton [2005]. Diesel Emissions Management. Rev.1. BHP Billliton Limited.
Bugarski, A.D., Janisko S., Cauda E.G., Noll, J.D., Mischler, S.E. [2012]. Controlling Exposure - Diesel
Emissions in Underground Mines. Society for Mining, Metallurgy, and Exploration. ISBN-13:
9780873353601. Available from: http://smemi.personifycloud.com/PersonifyEbusiness/Store/Product
Details.aspx?productId=116967 (accessed on February 1, 2021).
Cummins [2016]. Ready for more. Underground. Underground Mining Engines. Cummins Inc.
Demirgok B [2018]. Development of an emissions monitoring methodology using on-board NOx sensors
and revision to current in-use emissions regulatory protocols. Graduate Theses, Dissertations, and
Problem Reports. 5473. Available from https://researchrepository.wvu.edu/etd/5473 (Accessed on Feb
ruary 2, 2021).
Gallus, J., Kirchner, U., Vogt, R., Börensen, C., Benter, T. 2016. On-road particle number measure
ments using a portable emission measurement system (PEMS). Atmospheric Environment 124:
37–45, doi: 10.1016/j.atmosenv.2015.11.012.
Giechaskiel, B., Schwelberger, M., Delacroix, C., Marchetti, M., Feijene, M., Prieger, K., Andersson, S.,
Karlssong, H.L. 2018. Experimental assessment of solid particle number Portable Emissions Measure
ment Systems (PEMS) for heavy-duty vehicles applications. Journal of Aerosol Science 123: 161–170,
doi: 10.1016/j.jaerosci.2018.06.014.
Hallquist, A. M., M. Jersko, H. Fallgren, J. Westerlund and A. Sjodin (2012). “Particle and gaseous
emissions from individual diesel and CNG buses.” Atmos. Chem. Phys. Discuss. 12: 27737–27773.
Hines J [2019]. The role of emissions based maintenance to reduce diesel exhaust emissions, worker
exposure and fuel consumption. Doctor of Philosophy thesis, School of Health and Society, University
of Wollongong. Available from https://ro.uow.edu.au/theses1/854 (Accessed on February 2, 2021).
Kadijk G, Spreen LS, van der Mark PJ [2016]. Investigation into a periodic technical inspection test
method to check for presence and proper functioning of diesel particulate filters in light-duty diesel
vehicles. TNO 2016 R10735v2. Earth, Life & Social Sciences. The Netherlands.
McGinn S, Ellington R, Penney J, Graham C, Huzij B [2010]. Diesel emissions: mechanics’ maintenance
manual. DEEP: Diesel Emissions Evaluation Program.
Spears MW [1997]. An emissions-assisted maintenance procedure for diesel-powered equipment. Minne
apolis, MN: University of Minnesota, Center for Diesel Research. National Institute for Occupational
Safety and Health (NIOSH). Contract No. USDI/1432 CO36900.
Spears, M. (1997). An Emissions-Assisted Maintenance Procedure (EAMP) for Diesel-Powere Equip
ment. Minneapolis, MN, University of Minnesota, Center for Diesel Research.
TSI. (2021). “NANOPARTICLE EMISSION TESTER MODEL 3795.” Retrieved 10/ 21/2021,from
https://tsi.com/products/particle-counters-and-detectors/engine-exhaust-particle-counters/nanoparti
cle-emission-tester-model-3795/.
Waytulonis RW [1987]. An overview of the effects of diesel engine maintenance on emissions and
performance. In Proceedings: Bureau of Mines Technology Transfer Seminar, Louisville, KY, April 21,
1987, and Denver, CO, April 23, 1987. Information Circular 9141. Bureau of Mines. United States
Department of the Interior.
Yang, J., Durbin, T.D., Jiang, Y, Tange, T., Karavalakis, G., Cocker III, D.R., Johnson, K.C. 2018.
A comparison of a mini-PEMS and a 1065 compliant PEMS for on-road gaseous and particulate emis
sions from a light duty diesel truck. Science of the Total Environment 640–641:364–376, doi: 10.1016/
j.scitotenv.2018.04.383.
146
Electric machinery in mine ventilation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
C.A. Rawlins
Director - RME Consulting Canada Inc
ABSTRACT: Mine ventilation planning using trackless diesel equipment is well known in
the industry. Battery Electrical Vehicle (BEV’s) type equipment is increasingly replacing diesel
vehicles underground and different ventilation design methodologies need to be evaluated for
safety and health of workers. Planning further evaluate the heat loads of equipment to counter
their negative impact on the ventilation system. It is therefore important to evaluate the over
all ventilation system when comparing Diesel vs BEV’s thereby accounting for other aspects
such as blasting and re-entry periods for the mine including multi-blast development ends. In
general, is it anticipated that replacing diesel vehicles with BEV’s will imply reduced air quan
tity in a mine, therefore contributing positively towards overall mine power and an improved
return on an investment. BEV’s and other types such as trolley assisted vehicles all fall within
the heat and airflow reduction application philosophy and the ventilation planning needs to
be addressed to ensure designs don’t under-or-over-estimate their positive contribution in
underground mines as we always need to determine the in-mine heat load for shallow and
deep mines to ascertain compliance with design parameters. This paper elaborates on ventila
tion design parameters for comparison with diesel vs BEV application.
1 INTRODUCTION
DOI: 10.1201/9781003429241-16
149
Besides regulating the diesel fuel content with relation to the Sulphur content and other
contaminants such as DPM, etc., Ontario, Canada, and other countries further regulates the
air supply for diesel vehicles used underground. This is required to ensure the Threshold Limit
Value (TLV) of airborne pollutants are kept at or below their respective limit values. Diesel
engines are tested at CANMET in Canada and MSHA in the USA to determine the air
supply to the vehicle thus countering the tailpipe contaminants to acceptable levels. These
tested air supply values are available at the different internet sites as needed.
Furthermore, diesel engines are further known as Tier 1, 2, 3 up to the current Tier 4 type
engine. Table 1 (Metrovancouver, 2023) shows the diesel engine Tier and year of manufactur
ing for reference.
• Tier 1-3 Standards. The first federal standards (Tier 1) for new nonroad diesel engines were
adopted in 1994 for engines over 50 hp and phased-in from 1996 to 2000. Tier 1-3 standards
are met through advanced engine design, with no or only limited use of exhaust gas after
treatment (oxidation catalysts).
• Tier 4 Standards. On May 11, 2004, EPA signed the final rule introducing Tier 4 emission
standards, which were phased-in over the period of 2008-2015. The Tier 4 standards require
that emissions of PM and NOx be further reduced by about 90%. Such emission reductions
can be achieved using control technologies—including advanced exhaust gas aftertreatment.
• Tier 5 Standards. In November 2021, the California Air Resources Board held the first
public workshop development of Tier 5 emission standards that will seek to further reduce
NOx and PM emissions by 50-90%, depending on the engine power category, in the 2028-
2030 timeframe.
In review of the current diesel engine pollutant requirements for Tier 4 and later year
designs, the outlet tailpipe pollutants are reduced reasonably well. It is important to note that
in some provinces in Canada and in the USA, legislation is being adopted to apply the actual
tested air quantity value noted in the test results for the different engines and this is related to
the Tier type of the engine manufacturer.
Furthermore, using the legal (Ontario, Canada) value of 0.06 m3/s/kW (100 cfm/bhp) for
diesel engines underground, in conjunction with exhaust emission dilution methods and
advanced engine control technologies, this will maintain the tailpipe pollutants below the
respective TLV’s. This air quantity applied is further used to evaluate the heat load compo
nent onto the surrounding air from the diesel equipment. It is generally known that a diesel
machine typically supplies between 1.5 and 3 times the rated engine power as a heat compo
nent depending on the machines age and maintenance applied over time including the engine
technology. This is all related to the efficiency of an engine which is typically around
a maximum of about 40%.
Notwithstanding the fact that diesel engine manufacturers have reduced their engine tail
pipe pollutants to very low values after tailpipe advanced filtering methods and engine control
150
measures, the industry is moving towards more and more electrical based machinery which
totally avoid inducing pollutants such as carbon monoxide, sulphur content and nitrous
oxides and other harmful contaminants (DPM’s, etc.) into the underground air.
Some research indicated that electrical motors (excluding mechanical resistance items such
as wheel bearings, drive shafts, etc.) operate at about 38°C while diesel motors operate around
98°C (excluding mechanical items and exhaust pipe temperatures reaching 400 - 550°C at
times), thus indicating a motor heat load reduction of about 61% when compared to diesel
engines. Therefore, diesel engines need to be evaluated from two specific aspects namely (a)
a pollutant aspect, and (b) a heat load aspect, whereas the electrical machinery needs to be
evaluated from a heat load perspective only.
2 VENTILATION DESIGN
For actual mine ventilation design from conceptual to implementation phase, there are steps
to consider during each phase of the project fuel and equipment background. When consider
ing a mine doing ventilation design there are a couple of steps to follow as mentioned above
and depending on the project level of design, i.e., Conceptual, Pre-feasibility, Feasibility, or
finally the Implementation phase, certain aspects need to be investigated in detail (Rawlins,
2004). Aspects such as cooling requirements, heat loads, dust liberation, air quantity estimates
or actual detailed calculations.
In general, a project phase will be as follows.
As indicated in Figure 1, the ventilation design starts at a conceptual phase whereby identi
fying typing parameters such as production and development rock/ore production. This will
151
include the evaluation of the mining method to be applied for the project. The mining method
is important as this can influence the air quantity either more or less as needed. This further
will indicate either trackless or track-bound vehicle operation or both (hybrid type systems).
Using BEV’s, used as track-bound locomotives to move the broken rock from the produc
tion area of the mine to the rock-passes at the shaft, in mining is not a new concept as it has
been applied to shallow and deep mining operations in mines from the early mining oper
ations. Figure 2 illustrates a typical battery electrical locomotive in this type of application.
152
Table 2. Limited example of ventilation specifications or design criteria.
Parameters Value Unit
5. Sufficient air to dilute flammable gas encountered and to dilute dust that could be created
during normal mining operations, i.e., drilling, tramming, etc. To satisfy cooling of work
ers, a minimum of 0.5 m/s is advocated, and where tramming is carried out from the face
area, a dust and fume clearing velocity of 1.0 m/s is required.
6. The heat load of the mine with reflection on the diesel machinery, VRT, fans, motors,
blasting, etc.
153
As for any mine design and depending on the design stage, some limited specifications are
needed to ensure the design is done according to specific parameters and criteria with assumptions
included.
This information provides important parameters that influence the air requirements for an
underground mine. The amount of air required at the point of operation in any mine must
ensure that any gas/fumes caused by blasting and/or mechanical equipment is adequately
diluted to legally allowed Occupational Exposure Limits (OEL’s), and that any dust generated
by mining activities is directed away to return airways to ensure that workers are not exposed
to undue occupational risk. In the situation where a mine experience increased levels of heat
ingress, the ambient wet bulb temperature (i.e., WBT ≥ 27.5°C) and air velocities would dictate
the Air-Cooling Power (i.e., ACP- ~170 W/m2) of air (Rawlins, 2004 (3)).
The denominators indicate the Canadian threshold limit values applied for each contamin
ant at the time the EQI was developed (reference Canadian Standard - CSA-M424.2 (1990).
Once the EQI is determined the air density (kg/m3) and engine exhaust gas flow rate (kg/s) is
introduced to determine the air quantity certified for that machine usage underground.
The quantity is calculated in the following equation:
Therefore, the calculated quantity, using an exhaust gas mass flow rate of 456 kg/hr, is,
say 6.1 m3/s air required for this engine. Another parameter to be calculated is the heat pro
duced from diesel vehicles.
154
2.2.3 Diesel vehicle heat
An accurate means of calculating trackless diesel equipment heat load (McPherson, 1993) is
from a fuel consumption and fuel property point of view. As an example, the diesel consump
tion for the LHD is 0.254 L/h per rated power (kW). The diesel usage is 25.4 L/h and the fuel
calorific value is 35400 kJ/L. Therefore,
Heat produced = Diesel usage x Calorific value of the diesel/3600 seconds per hour
The heat produced from diesel machinery comprises of both sensible and latent heat. The
latent heat component could be calculated from the water produced after combustion (taken
at 100% in this example) at a mean of five (5) litre per litre of fuel consumed (water pro
duced = 5×25.4 = 126.9 L/h). Given that the latent heat of� evaporation of water is about
2450 kJ/kg then the latent heat produced would be 126:9�2450
3600 ¼ 86:4� kW. The sensible com
ponent is �therefore (249.6
� – 86.4) = 163.2 kW. The trackless equipment rated output power
251:4�3600 �
would be ð0:254�35400 Þ ¼ 100 kW. The heat load factor imposed onto the surroundings would
�
be 249:6
100 ¼ 2:5 (100% utilisation at full power rating). Depending on the variable parameters
used such as the diesel consumption for this specific LHD type, so will the heat load change
(McPherson, 1993). Figure 3 shows the heat load imposed on the surrounding air in relation to
the 100 kW diesel engine operated in conjunction with its full power output requirement at dif
ferent utilisation factors.
Figure 3. Heat load and heat load factor graph for a 100 kW diesel engine.
Furthermore, when a diesel machine operates underground, the heat load imposed on the
surroundings influences the ambient air temperature. Therefore, notwithstanding the air
quantity determined from a fume dilution point of view, the heat expelled from the machine
could mean that the quantity requirement can change to counter this effect. In mechanized
155
mines, trackless machine heat load could affect the ventilation system requirements and the
air quantity could be required to change.
The wet bulb reject temperature in most mines range between 26°C and 28.5°C. In our
example the reject wet bulb temperature WB is taken to be 27.5°C. It is therefore important to
know what the intake WB just before the LHD should be, in order not to allow the exit tem
perature (after the LHD) to exceed the design reject temperature of say 27.5°C. A simplistic
psychrometric calculation (Rawlins and Phillips, 2005) is done using the following input
parameters after the LHD.
a) Barometric pressure (P) = 103.6 kPa
b) Exit air temperature (Te) after the LHD: 27.5°C wb
c) Air mass flow (M2) = 20 kg/s (17.3 m3/s)
Calculations indicated that the energy content on the LHD intake side should not exceed
(1670.9 – 249.7) 1421.2 kJ/s. The intake energy content corresponds to a sigma (S2) heat value
of 71.06 kJ/kg. The WB temperature on the intake side of the LHD should therefore not
exceed 24.4°C taken that the surrounding rock and other heat loads are excluded at this point.
A WB above 24.4°C before the LHD (LHD operating at 100% power) would increase the tem
perature after the LHD to above the design WB temperature of 27.5°C.
Table 3. Various scenarios for both BEV and diesel vehicles comparisons for heat load.
Mass Heat
Temperature Quantity Velocity Quantity
flow Load
Heat Load
No. Description WB (C) m3/s m/s kg/s kW Factor m3/s/kW
As there is no specific or legal air quantity requirement for BEV operation in mines, the
only parameter to evaluate BEV’s is the imposed heat load on the environment. This is besides
other aspects such as versatility requirements, etc.
The different scenarios are as follows.
156
1. The airflow remained at 20 kg/s.
Comment:
BEV: Wet Bulb Temperature increased from 20°C to 21.6°C.
Diesel: Wet Bulb Temperature increased from 20°C to 23.6°C.
The BEV heat load (HL) was 106.1 kW and the Diesel HL was 248.3 kW.
2. The wet bulb temperature was increased to 27.5°C.
Comment:
Air quantity was reduced, i.e., BEV air quantity is 3.2 m3/s to reach 27.5°C, and Diesel
vehicle air quantity is 7.9 m3/s to reach 27.5°C.
Heat Loads were similar as before.
The haulage air velocity in both the BEV and the diesel vehicle haulage were below the
design air velocity of a minimum of 0.5 m/s, i.e., 0.13 m/s and 0.31 m/s respectively.
3. The haulage air velocity was 0.5 m/s as for the design requirements.
Comment:
Air quantity for both the BEV and the diesel vehicle situations are 12.3 m3/s and 12.6 m3/s
and therefore very similar.
Wet Bulb Temperature increased to 22.2°C and 24.9°C for the BEV and the diesel vehicles
respectively.
The BEV heat load (HL) was 107.0 kW, and the Diesel HL was 251.6 kW.
4. The haulage air quantity was set at 6.0 m3/s to coincide with the design value for diesel
equipment of 0.06 m3/s/kW.
Comment:
Wet Bulb Temperature increased to 24.3°C and 29.7°C for the BEV and the diesel vehicles
respectively. The diesel vehicle wet bulb temperature is above the design value requirement
of 27.5°C and this indicates that the minimum standard set for the diesel vehicle in this
situation is inadequate and the air quantity need to increase to at least 7.9 m3/s to maintain
the wet bulb design value as indicated in scenario No. 2. Furthermore, the minimum air
velocity is not attained with the 7.9 m3/s air quantity and needs further adjustment to at
least 12.6 m3/s to maintain the minimum air velocity in the haulage as a design require
ment.
The BEV heat load (HL) was 104.1 kW, and the Diesel HL was 249.2 kW.
Summary comment:
The minimum air velocity range of 0.5 m/s would be the more appropriate measure in the
BEV and diesel vehicle cases as this parameter maintains the heat load and therefore the wet
bulb temperature in both the BEV and the diesel vehicle scenarios evaluated.
The re-entry period scenario is also evaluated and given below.
The haulage length was calculated to be about 113.4 m length to maintain 30-minutes after the
blast with the minimum air quantity of 12.6 m3/s to further maintain the minimum air velocity in
the haulage. Therefore, the haulage length should not exceed the length of 113.4 m as this is the
maximum length to maintain a 30-minute rapid re-entry period. From the evaluations done it is
evident that the air velocity minimum of 0.5 m/s is the determined factor for this system where the
100 kW BEV or diesel vehicle operates. This minimum air velocity will counter the heat load in
157
the diesel vehicle as well. This further indicates that the heat load factor is about 1.1 for the BEV
and 2.5 for the diesel vehicle. The air ratio evaluation indicated that the BEV will have a value of
about 0.03 m3/s/kW and the diesel vehicle air ratio will need to increase from 0.06 m3/s/kW to
about 0.08 m3/s/kW to maintain the heat load aspect for the system. The scenarios as described
are illustrated in Figure 4.
It is notable that the design value of 0.06 m3/s/kW is not adequate for the diesel vehicle,
however it is in order when the BEV is applied. The design air velocity of 0.5 m/s is the param
eter to be maintained at this scenario evaluation for the 100 kW vehicle scenario for heat load.
This paper describes ventilation design parameters when diesel vehicles and BEV’s are compared.
When considering this shallow mine scenario described, it was evident that given the param
eters of the design such as wet bulb air temperature, minimum air velocity in a mine haulage,
etc., the heat load is mostly the important factor to evaluate. When normal project ventilation
design work is done, we determine the total mine air requirement from all diesel vehicles.
Once this value is calculated, then the air distribution is done preliminary and other factors
such as air leakage, auto-compression effects, etc., are made part of the design.
Evaluating a shallow or deep mine system, the heat load will be the most important factor
to be evaluated against the trackless fleet employed, i.e., either diesel or BEV type machines.
When doing this shallow mine comparison for BEV vs diesel vehicles, it was evident that
the general air to engine power ratio of 0.06 m3/s/kW is not adequate as a standalone factor to
evaluate. It is rarely the only factor employed when trackless diesel equipment is evaluated in
a ventilation design.
To maintain the given wet bulb design temperature of 27.5°C with the given 100 kW vehicle
engine, the BEV indicated a value of about 0.032 m3/s/kW and the diesel vehicle indicated
158
around 0.079 m3/s/kW. This indicates a 58.9% air requirement reduction when compared to
the diesel vehicle, although this air quantity determined indicated an air velocity below the
design value of 0.5 m/s. Therefore, the ai velocity design value is the prominent value to apply
in this design. Other evaluations and suggestions proposed by Halim and Kerai (2013) indi
cated values such as 0.04 m3/s/kW for deep mines and 0.025 – 0.037 m3/s/kW for shallow
mines where this shallow mine evaluated in this paper seems to be in-line with their initial
findings.
Other aspects such as equipment versatility is important in certain circumstances and there
fore, as with typical coal mines, there is always a diesel scoop running around in a mine to
assist in certain work to be done besides the primary cable electrical equipment applied in pro
duction sections.
The heat load of a mine is important to be evaluated when the vehicles are included. This
parameter will indicate what the appropriate air quantity is to be applied to the project. The
thoughts of implementing BEV’s into a project and reducing the amount of air substantially is
not true and could lead to inadequate air supply from a heat load perspective, also from
a dust dilution perspective when the minimum air velocity range is applied. A mines overall
re-entry period after a main blast (including rapid haulage development as needed) is further
aspects that need to be part of the overall mine design to ensure the re-entry period in
adequate as per mine design.
REFERENCES
159
Mine cooling and refrigeration
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
C. Allen
Vale, Copper Cliff, Ontario, USA
ABSTRACT: Vale’s Coleman mine has extracted minerals from 200-1,700 m depth for over 50
years and is currently expanding 1,850 m. At depth, autocompression effects superimposed on
summer climatic conditions, generate increasingly adverse thermal working conditions with tem
peratures exceeding 40°C DB and 28°C WB. These require heat exposure management through
work:recovery regimens and stop work conditions. Through mine planning, Vale recognized they
would need cooling, and in 2016 engaged BBE to investigate novel cooling and traditional tech
nologies, assessing both underground and surface options to provide focused to the 170 Orebody.
It was identified that although the cooling aspect was feasible, providing the required electrical
power, clean water supply, and locating the heat rejection without impacting mine activities
would be problematic. In 2019, Vale again engaged BBE to design and supply a surface cooling
plant, this included investigating alternative mechanical options, opportunistic coolth storage solu
tions, and the final provision of an operating turn-key system for the summer of 2020. Consider
ing the 18-month timeline, the capital costs, unproven technologies, and production loss risks,
most of the options reviewed were discounted and Vale opted for a conventional vapor compres
sion mechanical system. This paper discusses the challenges of surface and underground cooling,
novel options, the installation, and operational challenges of a surface air cooling plant in
a northern climate.
Vale’s Coleman mine located 50 km NW of Sudbury, Ontario, Canada, has a humid continental
climate with warm/hot humid summers and cold winters. Ambient temperatures can range sea
sonally from -32°C through to +32°C, and temperatures in excess of 25°C can be expected from
May through September. With mines reaching greater depths, such summer conditions will
required heat stress management protocols (from work:recovery regimes to stop work orders) to
be progressively introduced due to autocompression effects causing the temperatures to increase.
Coleman mine currently produces to depths of 1,700 m with development to 1,850 m, and due to
autocompression thermal working conditions could exceed 40°C DB and 32.0°C WB; these tem
peratures are Vale’s respective stop work limits. Heat stress protocols are required to ensure the
health and safety of the work force but reduce a mine’s productivity and thereby its profitability.
Through its mine planning, Vale recognized they would increasingly need mine cooling, and
from 2016 onward have been exploring options prior to the final selection of a surface plant in
2019, and requiring it being operational for the summer of 2020.
As will be described in the following sections, the need for cooling gradually progressed
from providing focused localized cooling, through to recognizing the need for a surface plant.
At both stages, various options and alternative technologies were evaluated prior to deciding
upon a conventional mechanical vapor compression refrigeration system.
DOI: 10.1201/9781003429241-17
163
2 COOLING THE 170 OREBODY (OB)
Figure 2. Underground bulk air conditioner spray chamber, laronde mine (courtesy of Agnico Eagle).
Two mechanical cooling options were considered, single and two stage direct contact cold
water open spray chambers, Bulk Air Coolers (BACs), and slipstream cooling coils circulating
cold water. Regardless of choice, they would be serviced from a 6 or 2 MWR refrigeration
164
plant room, and heat rejection would be through respectively a 8 or 3 MW direct contact
open hot water Condenser Spray Chamber (CSC). A slipstream coil solution with air also
being allowed to bypass the unit was chosen over a bulkheaded arrangement due to not want
ing to restrict the airflow when cooling was not required. Only a CSC option for heat rejection
was considered at this stage as the recipient would be air returning from mine workings con
taining dust, particulate, blast fumes etc. Figure 1 and Figure 2 respectively show a BAC that
was installed underground at Agnico Eagle’s Laronde mine, and a schematic drawing of
a coiling coil fan unit.
Some of the specific considerations to each cooling option and the heat rejection are:
• BAC have a larger footprint and cross-sectional area (a function of required air velocity),
need near level ground for a water dam, but has no specific air quality limitations.
• The cooling coils, have a small footprint, can accommodate a grade change but need a fan
and clean air.
• A horizontal CSC has similar requirements to the BAC, a vertical CSC requires an eleva
tion change and additional pumping power.
The general dimensions of each element and the required services for the various options to
supply 6.0 MWR and 2.0 MWR are given in Table 1; where possible the mine’s standard road
way dimensions were maintained with a view to repurpose existing openings.
Table 1. Conceptual Design details of 6.0 and 2.0 MWR cooling arrange
ments for an underground installation.
Although all concept options were to a certain degree feasible at the respective locations,
each had its limitations. These ranged from minor groundwork, essentially repurposing exist
ing spaces, through to major construction requiring new locations to accommodate the BAC,
Plant Room and CSC. The amount of construction could be limited by using cooling coils
over an open BAC spray chamber. However, the common issue to all options was the lack of
a suitable airstream, at a sufficient volume for heat rejection. The CSC process to reject heat
raises both the temperature and relative humidity of the host airstream, typical practice would
be to reject this air direct to a non-travelled return air route. For the 170OB, the primary
return air route was the main haulage ramp, and resultant conditions could be expected to
reach the order of 33°C DB with very high humidity nearing saturated conditions well in
excess of 27.5°C WB. These conditions would be prohibitive, even upon considering the
trucking is performed in AC operator cabs, the condensation on the ramp as the air ascends
would affect vehicle traction.
Considering this, alternate heat rejection options were explored. At some locations, the air
could be routed through a repurposed manway, but ultimately that air returned to the main
haulage ramp. In terms of air-based solutions, the most promising option was to use a drift
connecting through to an area of old mine workings which was already providing an inde
pendent exhaust to surface. This air route was a small dimension 900 m long section of track-
drift, nominally at 3,500 ft, relatively close in elevation to the 3770L, however a pipe-raise to
accommodate two 355 mm NB pipes would be required to the higher elevation, and a fan
165
would be needed to induce the higher reject airflow requirement. The use of this airway as
a heat sink was attractive as it did not compromise other activities within the mine. However,
its use similarly came with consequences, it was anticipated that the hot humid air would cool
as it travelled through the track-drift and condensation would occur producing an estimated
2.5 l/s of water needing to be managed by the mine’s dewatering system. This would require
upgrading the local dewatering pump capacity at the far end of the drift, which would necessi
tate the drift to be rehabilitated for its entire length at considerable cost.
At this point, other cooling options, the use of ice slurry delivered from surface, and heat
rejection, the use of mine water, were assessed. The ice option was discounted as a 25 kg/s
supply would add 25 l/s to the mine’s limited dewatering capacity. The use of mine water as
a heat sink was also considered impractical as it would require 150 l/s, and such volumes were
not available. Of all the underground options considered, the 6 MWR cooling of air using slip
stream coils located on the 3770L with heat rejection to the adjacent track drift was the pre
ferred solution. However, it would also require a major electrical supply upgrade within the
mine and on surface to accommodate peak demands of 2.5 MWE.
The FAR contained a 1.8 m diameter duct that was previously used as an exhaust. At
a velocity of 35 m/s the duct could supply 93 m3/s, this was only deemed sufficient to delivery
in the order of 2.0-3.0 MWR to the upper 3770L location. This option was quickly eliminated
based upon its limited cooling capacity, and because the current integrity of the duct, and its
long-term viability could not be assured.
The U-tube option would require both high pressure piping and coiling heat exchangers, nom
inally two 400 mm diameter 120 Schedule piping capable to withstand 2,500-4,000 psi. These
pipes would need to be installed within the existing #1 FAR or routed through an independent
pipe raise. The FAR concept was deemed impractical considering its was a decommissioned
stripped shaft and extensive work and mine downtime would be required for pipe installation.
The cost for directionally drilling a dedicated pipe raise was also considered prohibitive.
The remaining option was to cool the mines total 566 m3/s intake airflow with the first
order cost estimate being 2x that of selectively cooling only 212 m3/s. In addition, it would
have a larger power demand, greater than 3 MWE, and the mine already had power supply
limitations.
In late 2018, after considering the overall logistics of a 170OB solution and the loss potential
for productive time, Vale issued a request for the design and turn-key installation of a surface
166
cooling plant to deliver nominally 10.5 MWR. The specification was to be able to cool 566 m3/
s of air to a maximum temperature of 21°C DB/ 15.4 °C WB at the collar of the fresh air raise
downstream of the fans; plus, the requirement for the system to be operational for the
Summer of 2020. As part of the contract award to BBE, Vale required a stage-gate analysis
evaluating the potential to use alternate technologies from the typical mechanical vapor com
pression solution to leverage innovation funding for environmental initiatives. Within this
analysis specific attention was to be given to low water consumption and low energy use solu
tions respecting potential limitations at the mine. At the outset of this work, the cooling strat
egies to be reviewed are detailed in Figure 3.
Within the above ranking the wording “opportunistic” assumes the infrastructure required
for this technology is pre-existing. For example, Glencore’s Kidd Mine, Timmins, Ontario
(Counter, 2014), has near surface stopes within which ice forms during Winter, and in
Summer, it is then used to provide the equivalent of 8.5 MWR of cooling. Vale’s Stobie Mine
also has an ice-stope that was designed to offset winter heating needs (Stachulak, 1991); had it
been designed for the reverse process; it could provide in the order of 6.0 MWR of natural
cooling (Millar, 2016). The term NHEA refers to a seasonal Natural Heat Exchanger Area,
such as that employed at Vale’s Creighton mine, here air enters the mine through an extensive
167
broken rock mass in its former caved zone; it provides heating during the winter and cooling
during the summer (Stachulak, 1991). Consequently, Creighton mine does not have heating
costs and cooling is estimated to be equivalent to a 17.5 MWR plant (Millar, 2016).
A general description of each technology assessed follows:
1. Ice stopes are underground or surface-built modular thermal storage units that allow the pass-
through flow of cold winter air which contacts water directly, giving up energy to the water as
it freezes to form ice, in-turn heating the air to some temperature approaching the water being
supplied for freezing. In summer the process is reversed with warm air passing over the ice.
2. NHEAs are typically composed of previous excavations or pits. The rock surface and
broken rock mass maximize air-to-rock surface area and provides a degree of thermal iner
tia for the air being drawn through the excavation. Seasonally they provide heating and
cooling.
3. HAC refers to a hydraulic air compressor that function on the principle of air entrainment
as turbulent waterflows descend some vertical distance. The entrained air has the added
benefit of direct contact with the water as it is compressed. This compressed air can be cap
tured, transported, and decompressed to provide dry, sensible cooling.
4. Lake cooling takes advantage of large thermal masses and deep lakes to provide a source
of ~5-7°C water throughout the year. This can be of sufficient thermal quality to use dir
ectly in cooling applications, reducing interface inefficiencies, and reducing capital costs.
5. Mechanical chillers are a commercial vapor compression based system.
6. Ice slurries are produced with vapor compression technology with the cooling fluid frozen into
a slurry. The benefit here is that ice shipped underground doesn’t suffer autocompression.
7. Absorption chillers work on refrigeration cycles like mechanical vapor-compression except
they use a heat input directly instead of a compressor. Double effect absorption uses steam,
whereas single effect absorption uses hot water.
8. Ejectors convert static pressure into velocity pressure. A sudden decrease in pressure evap
orates a liquid-phase refrigerant (water) in a flash chamber, taking process heat with it,
and allowing it to be re-condensed elsewhere and cooled in contact with some heat rejec
tion infrastructure.
9. Liquid air can be used to provide point-source, dry, sensible cooling provided the infrastruc
ture exists to ship it. Air is liquefied in the same cycle as traditional vapor compression.
During the first stage-gate analysis ice stopes, an NHEA, a HAC, ice slurries, ejectors and
liquid air were eliminated due to either the lack of the “opportunistic” infrastructure, or the
inability to construct the delivery systems to depth.
Other options identified in Figure 3 were eliminated as they did not fit Vale’s requirement
for them to be a low energy, low water consumer, as per Table 3.
Considering the constraints discussed above and as listed specific to a technology, BBE and
Vale short-listed the analysis to the following alternative cooling technologies:
• Combined Heat and Power (CHP) Natural Gas Generator & Absorption Chiller
• Natural-Gas-Fired Absorption Chillers
• Glycol Thermal Storage & Distribution System
• Stratified Thermal Energy Storage
These were subsequently analyzed using a baseline requirement of 2 MWR. A natural gas
supply was available as it was used for winter heating.
168
Table 3. Other alternate cooling technologies eliminated.
Energy Analysis
Lake Water Lake water was removed due to the lack of a suitable deep lake close to the intake shaft.
Environmental permitting and related schedule implications were assumed to prohibitively
complicate the use of lake water in spray chambers or by heat pumps. A deep lake based
cooler sized to provide 2 MWR cooling at required design conditions would need a flow of
60 kg/s or more, with an estimated seasonal volume of 24,000 m3. The lake would need to
be significantly larger.
Process Process water flows of sufficient volume are not available near the intake fans.
Water
Solar Solar-thermal systems by their nature only function when sunlight is available. This tran
Thermal sient source of input energy could be used as a heat source for an absorption chiller. Local
conditions are assumed unfavorable for solar thermal systems; solar radiation data was
not verified. A solar collector with an area of approximately 70 m x 70 m would be
required to serve a 2 MWR absorption chiller. This and the absorption infrastructure’s
large capital cost and relatively high maintenance requirements were barriers to adoption.
Exhaust Air Exhaust air energy recovery cannot be applied feasibly to this project as the exhaust shaft
is not situated in close enough proximity to consider a glycol energy transfer system or
a heat pipe.
Ground A ground thermal energy source coupled with a heat pump could be used to provide heat
Source ing in winter and cooling in summer. However, in the absence of a natural geothermal
heat source, conventional ground source – heat pump systems are not suited to high
duties. A nominal 48 km of buried piping would be required to serve a 2 MWR heat
pump.
169
3.5 Seasonal thermal (coolth) storage systems
Two seasonal thermal storage systems were reviewed for Coleman. The first relied on a phase
change from water to ice to provide heating during winter and cooling during summer driven by
a glycol circuit. In winter, glycol is cooled, and the air heated via an air heating radiator. The cold
glycol is pumped in closed circuit to a bladder located in water reservoirs. Through heat transfer,
the water freezes to form the cold storage. The now warmed glycol returns to the heat radiator to
complete the circuit. In summer, cold glycol is pumped from the thermal storage to a water heat
exchanger with the chilled water then available to provide the mine cooling. This process is similar
to district heating and cooling systems. For its intended duty 2 MWR and duration, it was esti
mated a 0.55M gallon reservoir in excess of 10 m depth by 16 m diameter would be required.
The second approach uses water and an insulated ice thermal storage reservoir. In winter
the reservoir is filled with ice which could be produced by various methods. In summer the
melt water is circulated to cool air. Initial calculation of the sizing storage system shows the
reservoir to be in the order of 10 m depth by 45 m diameter.
Comparing these two systems the former would require a significant volume of glycol at an
outlay exceeding $0.5M, while the latter had a far more significant footprint.
A third novel system including water atomizing sprays (ultrasonic eductors) and MURL
foil dehumidification blades was also briefly reviewed but in the absence of detailed informa
tion from the proponent the system could not be technically assessed.
170
4 SURFACE COOLING OF COLEMAN MINE – SITE LIMITATIONS
The planned location for the cooling plant placed constraints on the final cooling plant selec
tion. The available space in front of the pre-existing fan/heater system was 29 m long by
20 m wide, see Figure 4, and had a pre-existing headframe infrastructure slab. The area was
also bounded by rock walls on the North and East sides, the mine air heater to the West and
an access road to the South. The water supply was constrained, being a shared resource from
a common pipeline also serving an adjacent mine property, and electrical supply was limited
with the mine being at the end of a transmission line. An additional constraint put on the
installation required it to not impact the existing intake fans which were already operating
near their maximum flow/pressure duty. After reviewing the costs, viability, logistics involved
in the various options and the 9-month remaining aggressive timeline for finalizing the design
and completion, Vale opted the selection of a full 10.3 MWR vapor compression system. The
prime driver was to guarantee the required amount of cooling be available throughout the
summer of 2020 to not impact production.
Figure 5. Conceptual and final design of BACs in relation to the mine heaters.
Table 5. Final design parameters for the Coleman surface cooling plant.
Design Parameter Value
171
4.1 Final design criteria
Figure 5 shows the initial conceptual and final arrangement of the BACs set back from the front
of the pre-existing heater/fan installation to minimize interference, and Table 5 provides the
design values. The design basis was to condition 88% of the required mine flow with the make-up
bypassing from the sides and above. Due to concerns of cooling loss under certain wind condi
tions, side walls with an access door have subsequently been added and the top remains open.
The placement of the BACs in relation to the fan-heater house and wind effects were stud
ied through CFD analyses, (Shaw and Falk, 2023). The mine opted to use rented natural gas
generators to meet the electrical needs of the cooling plant until power can be upgraded. To
address possible short-term water shortages, or interruptions, the design also included a water
storage capacity capable of providing a 3.5-hr operational buffer.
Figure 6. Final system installed at Coleman, from right to left: 8 BACs (4x2 arrangement), cooling
plant building containing 3 x 3.5 MWR chillers, and 3 CCTs, 1 per chiller machine.
This effort permitted the plant to be 50% operational for the summer of 2020; the final com
pletion of the system was delayed from that point on due to having to work around a part
commissioned plant. The full capacity of the plant was available October 2020. Post commis
sioning analyses have shown a potential loss of cooling due to certain wind conditions and
172
airflow characteristics. As shown by (Shaw and Falk, 2023) these have subsequently been
addressed with side walls being installed in 2022.
5 CONCLUSIONS
Vale and BBE went to considerable lengths to identify a workable cooling solution for the
Coleman Mine. The major hurdle to using an underground solution related to the lack of heat
rejection capacity, and possible solutions each had their drawbacks. Delivering chilled water
or ice to the required cooling horizon were also found to be impractical. The final option was
to cool the mine’s full intake flow at surface. With that choice, alternative cooling methods
and concepts were evaluated to leverage environmental initiative innovation funds. However,
none could provide the required guaranteed coolth, and capable of being constructed and
operational within the specified time frame. This time limitation was driven by the need to
avoid reduced work schedules or stop work conditions that would impact production in the
coming and subsequent summer.
Once the approval was given for a 10.3 MWR plant, BBE were responsible for its design,
construction, and commissioning within a 13-month period. Despite the numerous delays and
the major impact of COVID-19, the plant was operational at 50% capacity for the summer of
2020 and full commissioning completed October 2020. Further optimization work and con
struction started in the fall of 2022 to be ready for the 2023 cooling season.
REFERENCES
Counter, D.B., 2014. Kidd Mine – dealing with issues of deep and high stress mining – past, present, and
future, Proc. Seventh International Conference on Deep and High Stress Mining, Deep Mining 2014,
Australian Centre for Geomechanics.
Millar, D.L. 2014. A review of options for cooling ultra-deep mines in Northern Ontario and Quebec.
CIM MEMO Conference 2014.
Shaw, J. Falk, L. 2023. Minimizing the Aerodynamic Impact of a New Cooling Plant Installed Upstream
of an Existing Surface Fan and Heater Arrangement Through CFD Analyses. Proc. 19th North Ameri
can Mine Ventilation Symposium.
Stachulak, J.S., 1991. Ventilation and unique air conditioning at Inco Limited, CIM Bulletin, Vol. 84, No. 950.
Trapani, K., Romero, A. and Millar, D., 2016. Deep mine cooling, a case for Northern Ontario: Part II.
International Journal of Mining Science and Technology, 26 (2016) 1033–1042.
173
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
D.W. Durieux
Hatch, Ontario, Canada
ABSTRACT: With the ever-increasing drive towards decarbonization, the case for mine
refrigeration using geothermal means is gaining traction. Potential mines must however be
located in both favorable geo-political jurisdictions and areas with adequate geological condi
tions. One such area is Sudbury, Ontario, Canada where a conceptual evaluation found that
geothermal additions system can be cheaper and reduce CO2 emissions when compared to
a conventional mine refrigeration system.
1 INTRODUCTION
The largest mining companies including BHP, Rio Tinto, Glencore, and Vale have all targeted
“net zero” operations by 2050. In addition, they have collectively set medium-term carbon
reduction goals of 30% by 2030 (BHP), 50% by 2030 (Rio Tinto), 40% by 2035 (Glencore),
and 33% by 2030 (Vale) respectively. Therefore, given the relatively high energy consumption
of conventional mine refrigeration systems, and the resulting CO2 emissions associated with
power supply, the focus on mine refrigeration system efficiency is at an all-time high. The less
energy the mine refrigeration system requires to produce a given cooling duty, the lower the
associated carbon footprint.
The design process used to cool mines is well understood and has been successfully imple
mented for many decades. These cooling systems, and the associated equipment employed,
have typically been designed and fabricated for the harsh mining environment with a major
focus on robustness, often at the cost of refrigeration efficiency, or more accurately, coeffi
cient of performance (COP). System efficiency, of course, has featured in design decisions for
mine refrigeration systems, but the additional CAPEX and complexity required for very high
system efficiency has historically not been justifiable. That is until now, where the addition of
a CO2 emissions criterion into upfront option evaluation strongly favors highly efficient sys
tems. This paper attempts to compare a base case conventional mine refrigeration system,
located in the Greater Sudbury Area in Northern Ontario, Canada to several geothermal
modifications made to this base case system. This comparison can be seen as an option evalu
ation exercise similar to what is done during conceptual design.
2 ELECTRICITY EMMISSIONS
Greenhouse gases (GHG) emitted as a result of electricity generation and consumption largely
depends on the source of energy used. Common energy sources include coal, natural gas,
nuclear, hydro, and renewables such as wind and solar.
In Canada, the consumption intensity is the metric used to quantify the GHG emissions
from electricity consumption and Environment and Climate Change Canada recommended
a consumption intensity of 28 g CO2 eq./kWh for Ontario in 2020.
DOI: 10.1201/9781003429241-18
174
3 GROUND SOURCE HEAT PUMPS
Kavanaugh & Rafferty (2014) describes the term ground-source heat pumps (GSHP) as an
all-inclusive term for a variety of systems that use ground, groundwater, or surface water as
a heat source and/or heat sink. This includes various types and sub-types which is summarized
in Table 1. The choice of GSHP type will be both system- and site-dependent and will likely
require trade-off studies to select the optimal geothermal system or to reject geothermal
altogether. This paper, however, will focus on the vertical GCHP system (item 1.1 in Table 1)
as it is the most common and most applicable to a wide geography.
1 Ground- Closed-looped 1.1 Vertical GCHP using HDPE tubes in a vertical borehole.
coupled piping buried in 1.2 Horizontal GCHP including HDPE single pipe, multiple pipes,
heat pump the ground and coiled pipes.
(GCHP)
2 Ground- Opened-loop 2.1 Plate heat exchanger between heat pump and ground water loop.
water heat piping systems 2.2 Groundwater flowing through heat pump heat exchanger.
pump with water wells
(GWHP)
3 Surface- Systems con 3.1 Closed-loop similar to GCHP.
water heat nected to lakes, 3.2 Open-loop similar to GWHP.
pumps streams, or
(SWHP) other reservoirs.
175
required for successful implementation – both from the design engineers and installation
Contractors.
Hence, DN 65 mm (2 ½”) threaded carbon steel pipes (schedule 40) located inside
90 m (300 ft) deep, 160 mm (6 ¼”) diameter boreholes are used for the remaining analysis.
4 GROUND CONDITIONS
Typical Virgin Rock Temperature (VRT) equations used in the mining industry take the
form of Equation 1 which assumes VRT increases linearly with depth. This is generally not
the case as shown in Figure 2 for borehole temperatures in a hot region. It can be seen that
there is a relatively constant rock temperature until a depth of about 150 m. However, for
most mining applications, the main aim of the VRT equation is to provide reliable VRTs
at sufficient depth where heat from rock strata starts having a noticeable effect on the
total mine heat load. Therefore, the “surface temperature (indicated)” in Equation 1 does
not represent the actual surface rock temperature, but rather an intersection of the vertical
axis (temperature) at zero depth that results from optimizing the curve fit to the non-
surface regions of interest. In the case of the example shown in Figure 2, the actual surface
rock temperature is about 25.0°C while the VRT surface temperature (indicated) is 23.5°C.
The “surface temperatures (indicated)” documented for common mining regions are indi
cative only and need to be validated with site-specific near-to-ground thermal property
tests.
Sudbury is the location for this comparison and analysis and thus a VRT equation is
required. Cleland’s (1933) measurements found that the Frood Mine in Sudbury had
a geothermal gradient of about 1.14°C per 100m (1°F/160ft) with an indicated surface tem
perature of 7.6°C (45.7°F). Misner (1950) found that Creighton Mine in Sudbury had
a geothermal gradient of about 1.59°C per 100m (1°F/115ft) with an indicated surface tem
perature of 7.2°C (45.0°F), while more recent borehole tests in the Sudbury area suggest
a geothermal gradient of about 1.9°C per 100m (1°F/95ft) with an indicated surface tempera
ture of -2.5°C (27.5°F). The above variations for the Sudbury area suggest that there are great
variations in geothermal gradients as well as indicated surface temperatures; hence, site spe
cific borehole testing is imperative for accurate geothermal system assessment. For this ana
lysis, however, the VRT equation stated in Equation (2) will be used to represent the Sudbury
area as it represents an average of the data stated above.
5 REFRIGERATION SYSTEMS
176
COP is a metric used to define the “efficiency” of a refrigeration machine or an overall
refrigeration system and is defined as the ratio of cooling capacity provided to the electrical
energy input required to achieve this cooling. The higher the COP the better. For example,
a COP of 5 means that 1 kWe of electrical power is required to produce 5 kW of cooling.
“Refrigeration machine COP” is specific to the refrigeration machine only (evaporator duty
to compressor power) while “overall system COP” includes the whole system (air cooling duty
to total input power inclusive of pumping and ancillary systems). For the vapor-compression
refrigeration machines used in conventional mine refrigeration systems, refrigeration machine
COP is a function of the refrigerant evaporating temperature and the condensing temperature.
The closer these temperatures are to each other, the lower the compressor power requirement
and the higher the COP, for a given compressor isentropic efficiency. This is shown in
Figure 3 for typical mine refrigeration machine temperatures.
The hypothesis therefore is that the addition of a geothermal heat sink(s) to conventional
mine refrigeration systems will result in a lower refrigerant condensing temperature and hence
higher refrigeration machine COP for a given evaporating temperature. Three geothermal
Cases will be compared to the conventional Base Case.
Figure 1. Typical GSHP U-tube layout (left-unconsolidated & right-consolidated) [Kavanaugh et al.].
The design-day condition is taken at the ASHRAE yearly 98th percentile. The refrigeration
system is assumed on surface (Z = 0 m) close to the mine fresh air intake. All kW values refer
to thermal duty while kWe is used for electrical power. Air mass flowrates are for dry air and
temperatures are written as wet-bulb/dry-bulb.
Base Case (Figure 4) system comprises water-cooled R-134a centrifugal vapor-compression
refrigeration machine. Air cooling is done via a Bulk Air Cooler (BAC) and heat rejection via
CCT, both direct-contact type. Cold water is circulated from the BAC to the refrigeration
machine evaporator via the BAC pump while hot water is circulated from the CCT to the
refrigeration machine condenser. Refrigerant is circulated between the evaporator and con
denser via the compressor. Cold, fresh air is induced into the mine via existing primary fans
(not included in the analysis). Make-up water is required at the CCT to account for
177
Figure 2. Typical borehole temperature results: VRT (°C) = [0.01 (°C/m) x depth (m)] + 23.5 °C.
evaporation and water quality. The BAC condensate flows to the CCT make-up water circuit
to reduce the total make-up water requirement. The system COP is 5.6.
Case 1 (Figure 5) adds a closed-loop geothermal heat sink circuit to pre-cool the evaporator
water returning from the BAC. The vertical GCHP geothermal configuration described in Sec
tion 2 is employed. The required evaporator duty decreases due to the addition of the geother
mal heat sink. This results in a smaller overall refrigeration machine duty and compressor, as
well as a smaller CCT. But a larger BAC pump is required to overcome the borehole pipe
resistance. The system COP increases to 6.0.
Case 2 (Figure 6) replaces the CCT in the Base Case with a closed-loop geothermal heat
sink circuit for heat rejection. Again, the vertical GCHP geothermal configuration is used –
albeit for a much larger duty. The evaporator duty requirement remains the same as the Base
Case, but the condenser duty reduces because of the smaller compressor duty requirement.
The reduced compressor duty is driven by the superior heat transfer characteristics of the geo
thermal heat sink compared to the base case CCT resulting in much lower condenser tempera
ture. Thus, a lower compressor lift is required between the evaporator and condenser circuits
which improves compressor COP (Figure 3). But again, a larger CCT pump is required to
overcome the borehole pipe resistance. The system COP increases to 8.2.
Case 3 (Figure 7) combines both Case 1 and Case 2 with a resulting system COP of 7.3. The
decrease in system COP when compared to Case 2 is primarily owed to the large electrical
load of the two geothermal water circuit pumps.
The estimated savings for both cost and CO2 emissions, relative to the conventional base case,
are shown in Table 2. OPEX includes electrical power ($0.08 per kWh) and provision for an
Ontario carbon tax at $155 per tonne of CO2 equivalent (average forecast over the next 15
years). OPEX does not include items such as maintenance, stationary engineer staffing or
make-up water. CAPEX is an all-in cost and includes both the refrigeration systems and the
boreholes and pipes for the geothermal heat sinks. The all-in refrigeration rate used is
$1,800 per kW of evaporator duty which is based on historical project data for similar sys
tems. The borehole and pipe rate used is $80 per meter which is largely based on costs
178
Figure 3. Refrigeration Machine COP as a function of evaporator & condenser temperatures.
179
Figure 5. Case 1: Pre-cooled evaporator water via geothermal heat sink.
180
Figure 7. Case 3: Pre-cooled evaporator water and heat rejection via geothermal heat sink.
presented in Kavanaugh & Rafferty (2014) and adjusted for inflation as well as steel pipes
versus HDPE. Total Owning Cost (TOC) is used to evaluate the net effect of both CAPEX
and OPEX over a 15-year period with a 2% discount rate. The analysis makes provision for
4-months of continuous operation at peak duty which somewhat accounts for part-load oper
ation and the 2 excluded shoulder months typical in Ontario, Canada.
7 DISCUSSION
The results shown in Table 2 suggest that the most optimal configuration is that of Case 3,
since both TOC cost and CO2 emissions are lower than in the conventional Base Case system.
However, there are merits to the other configurations assessed as well. If the goal is maximum
CO2 reduction or minimal electrical consumption (e.g., remote site without access to grid
power), then Case 2 is the best configuration with a marginal TOC savings. Alternately, if the
project is CAPEX sensitive, Case 1 suggests the lowest CAPEX with the benefit of lower CO2
emissions and a lower overall TOC.
The total length of borehole pipes required for adequate heat transfer for the geothermal
systems is relatively high ranging from 2.2 km to 8.8 km. The boreholes must also be suffi
ciently spaced to reduce thermal interaction and thus relatively large land areas are required.
But most mines are in rural areas and therefore available land is likely. The boreholes must
also be located fairly close to the mine fresh air intake where the cooling system will be
located.
The geothermal configurations require less, or no make-up water compared to the conven
tional Base Case system. This may be a major factor where make-up water is not readily avail
able and/or handling of the CCT blowdown water is problematic. It also reduces the need for
sophisticated water treatment systems.
The geothermal configurations require relatively small refrigeration machine compressor
motor which may negate the need for higher voltage electrical equipment specifically for the
181
Table 2. Cost and CO2 savings versus the conventional base case system.
Base Case Case 1 Case 2 Case 3
refrigeration machine. There is therefore opportunity for a common low-voltage strategy for
the geothermal systems. However, special compressors may be required for some of the very
low lifts.
Although not explored in this paper, the application of geothermal to mine air heating sys
tems commonly employed in cold regions potentially offers greater savings than those
explored in this paper. For example, a system comprising one heat pump which provides both
cold air to the mine in the summer and warm air to the mine in the winter. This all-in-one
system may eliminate the need for conventional large air heaters which typically burn natural
gas, or at the very least, supplement the air heaters reducing the natural gas consumption and
GHG emissions.
8 CONCLUSION
It has been shown that geothermal additions, in the form of vertical GCHP heat sinks, to con
ventional mine refrigeration systems potentially offer both CO2 emission reduction and cost
savings. Multiple configurations were explored, yielding varying improvements on either cap
ital or operating cost, and offering flexibility for the mine to select the best technology to suit
their specific goals and constraints. Technological innovation is likely required to achieve the
ambitious net-zero goals set by some major mining companies, and geothermal is presented as
a relatively simple option to supplement conventional mine refrigeration systems.
REFERENCES
Cleland, R.H. 1933. Rock Temperatures and Some Ventilation Conditions in the Mines of Northern
Ontario, Canadian Institute of Mining, Metallurgy and Petroleum.
Environment and Climate Change Canada. 2022. Annex 13, Electricity in Canada: Summary and Inten
sity Tables, National Inventory Report 1990-2020: Greenhouse Gas Sources and Sinks in Canada –
Part 3.
Kavanaugh, S. and Rafferty, K. 2014. Geothermal Heating and Cooling, ASHRAE.
Misener, A.D. and Thompson, L.G.D. 1950. Temperature Gradients in Ontario and Quebec, Canadian
Institute of Mining, Metallurgy and Petroleum, Transactions, Volume LIII, pp. 380–383.
182
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
E. Pilkington
Sudbury Integrated Nickel Operations, a Glencore Company
ABSTRACT: The Onaping Depth Project is a nickel/copper underground mine project cur
rently in execution by Glencore’s Sudbury Integrated Nickel Operations. The Onaping Depth
orebodies will be accessed via the existing Craig Mine workings after completion of sinking
a winze from 1,200 to 2,635m below surface. Due to the final depth of the workplace, a 16
MW(R) installed capacity underground refrigeration plant will be operated at a depth of
1,915m below surface to provide conditioned air to the workplace throughout the project
development and life-of-mine production phases. This paper will provide an update on the
design, construction and commissioning of the underground refrigeration plant, including les
sons learned in the contracting and construction methodologies.
1 INTRODUCTION
Glencore’s Sudbury Integrated Nickel Operations (Sudbury INO) are in the progress of exe
cuting the Onaping Depth Project, an ultra-deep nickel/copper mining complex in Sudbury
Ontario Canada. Refrigeration has been a key consideration for the project since the study
phases due to the autocompression heat associated with the orebody depth (in excess of 2,500
metres below surface), high in situ rock temperature (45°C in deep workplaces) and large ven
tilation volume to support mechanized mining. Through a series of studies and trade-offs, an
underground refrigeration plant was identified as the preferred technology to ensure safe
working conditions in the deep zone. This selection was made feasible by the project’s decision
to mine with exclusively battery-electric equipment, allowing for the air volume circulated to
depth (and therefore required excavation size for bulk air cooling) to be reduced by approxi
mately 40% when compared to a typical diesel mining method (Wisniewski 2017).
Since 2016, feasibility-level and detailed engineering have advanced on the Onaping Depth
underground refrigeration plant. Major mechanical equipment was purchased in 2017 and
delivered to site in 2019-2020. Construction commenced at the 1915L (nominal metres below
surface) complex in early 2021 and commissioning of the refrigeration plant was substantially
completed in Q1 2023. The plant has been designed with the challenges of underground
refrigeration in mind. A unique layout was developed with a centrally-located plant room
between the two open spray chambers, minimizing excavation requirements and reducing
large bore pipe lengths, all while respecting the geotechnical constraints of designing large
excavations at a depth of 1,915 metres below surface.
2 PROJECT BACKGROUND
The Onaping Depth deposit has been known since the 1990s, when exploration drilling from
the existing Craig-Onaping mine complex identified the main orebody. The 18.4Mt deposit
DOI: 10.1201/9781003429241-19
183
resides in the vicinity of 2.6km below surface and will be mined at a target production rate of
1.2 million tonnes per annum with a design life of 20 years. The refrigeration plant will oper
ate for the complete life of mine in addition to two to three years of the Project period
required to access the orebody.
Multiple attempts to develop a mine design which would be economical with conventional
diesel mechanized mining methods were unsuccessful. Mechanical refrigeration is necessary due
to drill-indicated virgin rock temperatures in excess of 45°C and auto-compression across more
than 2,500m of vertical depth. These factors resulted in excessive workplace temperature condi
tions in the absence of cooling intervention. In addition, capital and operating costs associated
with the volume of air required to support a diesel fleet were found to be prohibitive.
In 2016, the Onaping Depth Project carried out a feasibility study on the basis of an entirely
battery-electric fleet for the deep workings. This allowed for a significant reduction in the ven
tilation volume (approximately 40%; Wisniewski 2017), and therefore both the ventilation and
refrigeration power consumption (approximately 45%; Wisniewski 2017). In addition, major
capital cost savings were realized from a reduction in excavation sizes and development quan
tities in the primary airways, in addition to the reduction in ventilation and refrigeration infra
structure cost. The project was given approval to commence execution in late 2016, with the
first development blasts fired in January 2017.
Lateral development for mine access construction within the first five years of the project
timeline has been executed with diesel equipment. All of this development is accessible from
the existing ramp system in Craig Mine, and is supported by existing diesel distribution and
fueling infrastructure. The lower workings of the mine, known as “The Depth”, will only be
accessible via a new internal winze excavated from 1150L to 2635L. All mobile equipment in
use in the Depth for development or production will be battery-electric vehicles (BEVs). Shaft
sinking is expected to be finished in 2023, at which time the first BEVs will be required on site.
This execution strategy allowed for five years of research and development from the
announcement of project execution until the first heavy vehicles (load-haul-dump (LHD)
machines and haul trucks) were required on site.
During early engineering phases, the Onaping Depth Project undertook detailed assessments
and trade-offs to determine the optimal cooling plant location. Overall, locating the cooling
plant underground at 1915L was deemed the most favourable from a technical and economic
standpoint when compared with a surface refrigeration installation. This was largely attrib
uted to the improved positional efficiency of an underground plant location, as there is
a reduced temperature differential with the rock strata upstream of the cooling plant for an
underground plant location and therefore less driving force for heat transfer from the strata.
The previous work has identified that powerful synergies arise due to the combination of
underground cooling with the reduction in primary ventilation volume attainable when regu
latory requirements for diesel exhaust dilution do not exist. This synergy results in reductions
in both capital and operating costs when compared to surface plants, and is explored in detail
by Brown et al.(2018).
Key process conditions forming the basis of the Onaping Depth underground cooling plant
design are summarized in Table 1.
Note that these parameters are specific to the life-of-mine (maximum) design condition for air
flow and refrigeration duty; the plant will be operated at a reduced air flow with a target chilled
air temperature of 10°C during the development and construction phases of the project.
The cooling plant consists of the following main components: bulk air cooler (BAC) for air-
cooling, refrigeration plant room (housing the refrigeration machines to provide chilled water to
the BAC), condenser spray chamber (CSC) for system heat rejection, water pumps and water
treatment for the recirculating water circuits between the refrigeration machines and spray cham
bers. Both the BAC and CSC consist of two-stage, unpacked horizontal spray chambers with
sloped basin floors to minimize slimes build-up and facilitate cleaning. Nominal dimensions of
184
Table 1. Summary of air side process parameters for refrigeration plant design.
Parameter Units Value
3
Air Flow (Volumetric) m /s 229
Mass Flow (Dry Air) kg/s 308
Inlet Air Temperature °C wb/db 27.6/30.5
Chilled Air Temperature °C wb/db 12.0/12.0
Barometric Pressure kPa 120.5
Nominal Air Cooling Duty kWT 14,600
Air Temperature Available for Heat Rejection °C wb/db 23.1/26.1
Maximum Air Flow Available for Heat Rejection m3/s 203
each spray chamber are 8.0m wide x 7.0m high inside shotcrete ground support, with the max
imum height of 8.0m high at the low point in the sloped floor.
The Plant Room is 9.0m wide x 8.2m high and contains three R-134a single-stage centrifu
gal compressor refrigeration machines (Johnson Controls Model YK), arranged in a series-
counterflow configuration. Water treatment systems consist of side-stream filters and chemical
dosing equipment including corrosion inhibitor, biocide, and alkali for pH control.
A summary of the water side process conditions is provided in Table 2.
185
• Plant room centrally located to minimize pump and piping requirements between refriger
ation machines and spray chambers,
• Chillers installed in series counterflow arrangement due to high lift and desire to avoid two-
stage centrifugal compressors,
• Sloped spray chamber basin floors to minimize excavation quantity,
• Equipment access and maintenance requirements including overhead crane,
• Geomechanical considerations for drift size and spacing.
The sloped floor of the BAC can be seen in Figure 2, which is a similar design for the CSC.
In addition to minimizing excavation volumes, the sloped concrete basins will facilitate wash-
down of collected dust residues during plant outages as needed. Stainless steel work platforms
are provided above the water level to allow for inspection and maintenance of the spray pipes
without having to drain the full system volume from the basin. The depressed centre section
visible is local to the pump suction (see section view in Figure 3). The spray chamber floor is
excavated approximately two metres above the adjacent pump room floor in order to provide
a suitable minimum suction head of 1.5 m at spray chamber basin low water level.
Design of the plant prioritized robust, reliable, and safe performance in an underground mining
environment. For example, the refrigeration machines were specified with high heat exchanger
fouling factors reflective of conditions in underground mining operations. Heat exchanger tubes
186
were specified to be thicker than standard and constructed of more robust copper-nickel alloy
material for long-term protection against both the corrosion and the erosion expected in an under
ground environment with the expected water quality. The chiller units were also specified to use
a Class A1 refrigerant with low toxicity and no flame propagation to minimize risk to workers in
the event of a leak in the confined underground environment. In addition, the airflow in the plant
room is ventilated directly to exhaust, reducing personnel risk from refrigerant leaks.
Another design aspect critical to the successful operation of the plant was inclusion of water
treatment systems. An appreciable portion of the dust entrained in the mine ventilation air
stream is expected to be captured in the spray chambers as the water sprays act like a wet par
ticulate scrubber. A side stream filter is included in the design of the CSC water circuit to
remove suspended solids and prevent clogging of the condenser tubes or CSC spray nozzles.
The BAC water circuit is expected to see less contamination as it is on the fresh air side of the
mine ventilation circuit; however, provision for installation of an identical side stream filter
has been made in the excavation and layout allowance in case of issues with suspended solids
on the evaporator water circuit.
In addition, chemical dosing to inhibit corrosion, scaling, and biofilm formation is included
in the design. This is critical to ensure proper water quality in the equipment, thus preventing
material degradation and ensuring proper heat transfer performance, as well as minimizing
personnel risk to biological contaminants such as legionella (Legionnaire’s Disease).
The refrigeration plant was standardized to operate at 4,160V, with motors for all chillers,
pumps, and the nearby primary ventilation fans and dewatering pumps all operating at this
specification. Power distribution to the level is accomplished by a pair of 25kV shaft feeders
from surface to the main substation near the entrance to the Plant Room, at which point it is
stepped down to 4,160V and 600V for ancillary loads.
4 CONTRACTING STRATEGY
Construction of underground refrigeration plants extends beyond the typical skill set of most
underground mining trade workforces. In particular, fabrication and construction of custom,
large bore pipes (DN450 and DN500) and work with refrigeration equipment are unusual in
the underground environment. However, some of the construction activities such as construc
tion of concrete bulkheads, drilling of rock anchors for pipe supports and structural steel tie-
backs, etc. are well-suited to mining contractors with experience in drilling and underground
construction.
The project team sought to develop a contracting model which would allow for both the high
degree of specialization associated with the cooling systems and the expertise in underground
construction. A lump sum contract model was selected, with an approved bidders list consisting
of underground mining contractors and contract language which ensured that dedicated piping
fabricators and tradespeople would be engaged from more traditional “surface contractors”.
This decision also intended to put the onus on the successful contractor to receive personnel and
materials at the shaft station(s) underground and provide transport to the refrigeration plant
workplace – approximately 3.5 km down ramp from the lowest shaft station.
Comprehensive tender evaluations revealed that the contract arrangement resulted in
a detrimental outcome primarily due to subcontracting markups included in all bid prices. In
response, an alternate strategy was proposed in which the total scope was divided into mul
tiple subcontracts, each consisting of a single specialized trade. Bidders lists were crafted such
that some contracts were only bid to mining contractors, while others were only bid to special
ist surface contractors. The division of scope is listed in Table 3 below.
In addition to these four construction contracts, an Underground Logistics contractor was
brought onto site who took responsibility for delivery of all construction materials and per
sonnel from the shaft station to the contract workplaces. Thus, the single lump-sum contract
became five specialized contracts which were all tendered competitively. Although an increase
in project management and contracts administration oversight is required, the reduction in
construction cost was approximately 27%.
187
Table 3. Construction contract summary.
Contract Scope Included Contractor Type
At the time of writing in November 2022, the refrigeration plant construction (aggregate of all
contracts) was substantially complete. Photos below show milestones at the completion of chiller
installation, with the overhead bridge crane used for installation and maintenance visible above
(Figure 4), and an end view of one chiller following piping tie-in and insulation (Figure 5).
188
Figure 5. View of insulated evaporator and evaporator piping.
Figure 6. View of CSC pipes, pipe supports and access platform following construction.
189
The BAC and CSC follow similar construction methodologies, with lower spray headers
supported from the platform structure and upper spray headers supported on beams spanning
the width of the spray chamber. This is visible in Figure 6.
A commissioning strategy was developed which separates the refrigeration plant into five
subsystems which were commissioned independently as shown in Figure 7. This approach
improves flexibility in the execution of the field work – where work on any subsystem can
start as soon as construction completion allows.
6 CONCLUSIONS
For the Onaping Depth Project, replacement of diesel equipment with BEV’s in the mining
plan allowed reduced airflow and heat loads in the mine.
An underground cooling plant was selected to provide bulk air cooling in close proximity to
the deep mining operations. This was shown to be more economical than a surface cooling
plant for the specific circumstances of the Onaping Depth mine design.
Equipment selection included single-stage, water cooled, centrifugal mechanical chillers
with synthetic refrigerant and shell and tube heat exchangers.
Both bulk air cooling and condenser heat rejection were designed to use direct contact open
spray chambers adjacent to the refrigeration plant room.
190
Contracting strategy for the underground construction works was found to be sub-economic
using a single lump sum contract, and cost savings were achieved by dividing work to specialist
contractors and intensively coordinating to minimize work clashes between these packages.
REFERENCES
Brown, M., Guse, T., Arsenault, S. & Rogers, B. Effects of an Electric Fleet on Mine Ventilation and
Refrigeration with a Comparison to an Equivalent Diesel Mine. Paper presented at the Roomvent &
Ventilation Conference; 2018 Jun 2–5; Epsoo:Finland.
Wisniewski, S. Onaping Depth Project. Presented at CIM Sudbury General Meeting; 2017 Mar 23; Sud
bury:Canada.
191
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
A. Pandey
Department of Mining Engineering, BIT Sindri, India
S. Jayaraman Sridharan
South Dakota Mines, Rapid City, SD, USA
B.S. Sastry
Department of Mining Engineering, IIT Kharagpur, India
Keywords: Sigma heat, Wet-bulb temperature, Mine ventilation, Underground mine envir
onment, Prediction model, Psychometric chart
1 INTRODUCTION
DOI: 10.1201/9781003429241-20
192
Hence, the enthalpy (H) of the air does not remain constant in an adiabatic humidification pro
cess. However, enthalpy minus the sensible heat of evaporated liquid water would be a property
value that remained genuinely constant and known as sigma heat (S) (Watson, 1981). The energy
contained within the air can be described by S (Ryan & Euler, 2017).
Since the concept of sigma heat was initially been discussed by Carrier (1911) in 1911, it has
not found wide use by air conditioning engineers until it was recognized by Whillier (1969)
and Bluhm (1981). Later, Bluhm (1990) shown that the sigma heat differences are more accur
ate than the traditional enthalpy differences in approximating total heat transfer from a wet
surface and, the use of S is more accurate than H in the typical approximate energy balance
relationships between water and air streams.
It should be noted that the sigma heat is a unique function of wet-bulb temperature (Tw) at
a given barometric pressure (P). Hence, thermal exchange in an airway can easily be quanti
fied from psychrometric observations using S. Moreover, as a directly measurable parameter,
a change in Tw indicates heat transfer from the machines, strata, or any other source, immedi
ately. Estimation of S is necessary for a variety of heat exchange problems in mines, such as
designing of spray chambers (Bluhm & Whillier, 1978) and in the calculation of the factor of
merit in cooling towers (McPherson, 1993).
Several researchers presented approximation equation for Tw as a function of dry-bulb tem
perature (Td) and either dew point temperature (Tdp) or relative humidity (RH). The standard
method of determining Tw from Td and Tdp is “one of the most involved and time consuming
calculations in psychrometrics” because it require iterations involving several psychrometric
equations (Chau, 1980). Al-Ismaili et al. (2016) presented an iterative approach to find Tw
from Td, RH, and actual station pressure values. An empirical equation for Tw as a function
of Td and RH is given by Estrada-Jaramillo et al. (2014) using a meteorological database. To
determine Tw for a given S and P, McPherson (1993) adopted an iterative method by iterating
between several psychrometric equations. Stull (2011) presented an empirical equation for Tw
by using Gene-expression programming (Ferreira, 2006) regression method. He used wet-bulb
depression (Td-Tw) and RH as input to the model for P of 1 atm. Mine ventilation software
such as VentSim and Vuma use iterative methods for the psychrometric calculations.
Although many equations are given by earlier research, in many ventilation problems, due
to statutory compliance or design requirements, Tw needed to calculated from given heat
values, and in many cases Td and RH values might not be known beforehand. Therefore, an
expression for Tw for a given S and P is desirable in such circumstances.
The aim of this paper is to provide an equation for sigma heat only as a function of and Tw
and P, which has the advantage of reducing the complexity of the computation. Also, an
expression for Tw as a function of S and P is also presented. Moreover, example of cooling
tower and spray chambers calculation is given to show the efficacy of the prediction equa
tions. Furthermore, a psychrometric chart based in-terms of S is also presented in this study
which will provide a more intuitive and visual understanding of the underground psychro
metric processes.
2 METHODOLOGY
193
Where Lw is latent heat of evaporation of water, X is moisture content of air, P is barometric
pressure in kPa, e is actual vapor pressure, Cw is specific heat of liquid water (4.187 kJ/kgK), Td
is dry bulb temperature of air in °C, Cpa is specific heat of dry air at constant pressure (1.005
kJ/kgK), Cpv is specific heat of water vapor at constant pressure (1.884 kJ/kgK), esw is saturated
vapor pressure of air at Tw, and Xs is the saturated moisture content of air at Tw.
Figure 1. Calculated sigma heat vs. wet-bulb temperature and barometric pressure.
194
Figure 2. Wet-bulb temperature vs. calculated sigma heat and barometric.
It can be seen from Figure 1 that S increases exponentially with an increase in Tw. Also,
Figure 3 shows that S decreases as P increases. Also, there is a higher rate of reduction in S as
P increases, at higher values of Tw.
Figure 3. Calculated sigma heat vs. barometric pressure at different wet-bulb temperature.
Based on the relationship observed in Figures 1, 2 and 3, prediction model for S (kJ/kgda)
and Tw (°C) are developed using regression analysis and expressed as:
195
0:73
where x ¼ S:P
1000
3 PSYCHROMETRIC CHART
Using the psychrometric equations to plot the psychrometric curves, a numerical code had
been developed in MATLAB R2016a to plot a psychrometric chart in-terms of S, from 5°C to
50°C dry bulb temperature, and atmospheric pressure range of 99 kPa to 120 kPa. The rele
vant thermodynamic psychrometric equations are used from published ASHARE handbook
(ASHARE, 2013). The psychrometric charts obtained from the current study at 1 atm or
101.325 kPa is shown Figure 4. Except constant Sigma heat lines, the chart presented in
Figure 4 resembles the presently available psychrometric charts, and is validated as described
later which establishes the credibility of the developed code and reliability of the present chart
for practical purposes.
Figure 4. Psychrometric chart for 1atm pressure and 5 to 50 °C in-terms of Sigma heat.
As the depth of the mines increases, P increases irrespective of the cooling system used in mines.
Generally, in non-artificially cooled mines, P and Tw both increase simultaneously with an
increase in depth. Taking the above-mentioned factors and the general climatic condition in an
underground mine into consideration, the ranges of P (99 kPa to 120 kPa) and Tw (5 °C to 50 °
C) are selected. In order to know the performance of the proposed models, 5000 randomly gen
erated data points of Tw and P are used to calculate S by Eq. 8 (calculated sigma heat) and
Eq. 9 (predicted sigma heat). These data points are plotted as a scatter plot and shown in
Figure 5. Also, same data set is used to calculate Tw as a function of P and S by Eq. 10.
196
Figure 5. Calculated vs. Predicted sigma heat for 5000 randomly generated values of sigma heat.
Figure 6. Calculated vs. Predicted sigma heat for 5000 randomly generated values of Wet-bulb
Temperature.
The relationship between the calculated and predicted sigma heat shown in Figure 5 & 6 is
easy to comprehend. An almost superimposed relationship between the data points and 1:1 or
reference line, suggests the good performance of the prediction model. Also, the Pearson’s cor
relation coefficient (r) of 0.999 shows the strong positive correlation between calculated and
predicted S as well as for calculated and predicted Tw.
197
Table 1. Maximum error, minimum error, and root mean square
error (RMSE) for the prediction models for S and Tw (Eq. 9 and 10
respectively).
Parameters Sigma heat (kJ/kgda) Tw (°C)
While calculating S for 5000 data points of P and Tw in MATLAB R2016a (MATLAB,
2016), the time taken for the calculation by the proposed equation (Eq. 9) is 25 % less than
that by using Eq. 8. The main reason for this is that the number of variables needed to be
addressed is less for the proposed equation. This saves the computation time and resources
required to calculate S, especially in the software tools applicable for modelling, designing,
and simulation of an underground mine environment.
Table 2. Sigma heat values from theoretical equation (Eq. 8), psychrometric chart (Figure 4) and predic
tion equation (Eq. 9), with respective absolute percentage deviation from theoretical sigma heat values
at P = 1 atm.
Wet-bulb Theoretical Predicted Deviation from Sigma heat from psy Deviation from
temperature Sigma heat Sigma heat Theoretical value chrometric chart Theoretical value
(°C) (Eq. 8) (Eq. 9) (%) (Figure 6) (%)
The maximum and minimum percentage deviation of S obtained from the presented psy
chrometric chart, from theoretical S values are found to be 0.10 % and 0.00 %, respectively.
Also, the percentage deviation of predicted S from theoretical S is between 0.32 % to 2.31 %.
This suggests that the values are acceptable for all kinds of underground calculation purposes.
5 CONCLUSION
Based on the fundamental formula given in literatures, a new prediction equation for S is derived.
The prediction results show a high level of agreement with the calculated values with acceptable
error limits. The prediction model also helps in faster calculation of S and consumes less computa
tional resources. The results suggest that the prediction model may be an acceptable substitute for
the theoretical equation for S in a variety of underground climate simulation software.
Using the available thermodynamic data and psychrometric relations, a chart with constant
Sigma heat lines is plotted in MATLAB for Td of 5 to 50 °C and pressure range of 99 to 120
198
kPa. Comparison with a previously available psychrometric equations shows that the chart
obtained in the current study is equally accurate with an absolute deviation less than 1.84 %.
Moreover, reliance on the data taken from available sources, which are equally accurate, con
firms the reliability of the developed chart. The success of this work could allow the investiga
tor to extend this to study at different pressures and at different water-vapor systems.
Although many iterative approaches have been in use for years, presently, there is no direct
equations to determine Tw for a given S and P. The most important part of the present work
is to provide a direct equation for Tw as a function of S and P to avoid this iterative exercise.
The model has an acceptable fit with R2 of 99% and RMSE of 0.11 °C. The prediction equa
tions presented in the study facilitates a reliable tool to ventilation engineer for the analysis
and faster calculations in various underground heat transfer problems.
REFERENCES
Al-Ismail, A.M. & Al-Azri, N.A. 2016. Simple iterative approach to calculate wet-bulb temperature for
estimating evaporative cooling efficiency. Stress (FITS). 12:13.
ASHARE. 2013. Psychrometrics. In ASHARE fundamental handbook. American Society of Heating
Refrigeration and Air-Conditioning Engineers, Georgia, USA.
ASHRAE, H.O.F. 2001. ASHRAE fundamentals handbook. American Society of Heating Refrigeration
and Air-Conditioning Engineers. Atlanta GA.
Bluhm, S. 1981. Performance of direct contact heat exchangers. Journal of Mine Ventilation Society of
South Africa. 34(Aug.):155–160.
Bluhm, S.J. 1990. Thermal performance of direct-contact water-air heat exchangers. University of the
Witwatersrand, Johannesburg.
Bluhm, S.J. & Whillier, A. 1978. Design of Spray Chambers for Bulk Cooling of the Air in Mines. Jour
nal of The South African Institute of Mining and Metallurgy. 79(1):1–9.
Carrier, W.H. 1911. Rational psychromatric formulae. Transactions ASME. 33:1309.
Chau, K. V. 1980. Some new empirical equations for properties of moist air. Transactions of the ASAE.
23(5):1266–1271.
Estrada-Jaramillo, M., Vera-Romero, I., Mart\’\inez-Reyes, J., Ort\’\iz-Soriano, A. & Barajas-Ledesma,
E. 2014. Empirical model to calculate the thermodynamic wet-bulb temperature of moist air. Engineer
ing. 2014.
Ferreira, C. 2006. Gene expression programming: mathematical modeling by an artificial intelligence.
V. 21. Springer.
Hemp, R. 1985. Air Temperature Increases in Airways. Journal of the Mine Ventilation Society of South
Africa. 38(2):13–20.
Jordan, D.W. 1965. The numerical solution of underground heat transfer problems—I Method relating
to dry roadways. International Journal of Rock Mechanics and Mining Sciences & Geomechanics
Abstracts. Pergamon. 2(3):247–270.
MATLAB. 2016.
Maurya, T., Karena, K., Vardhan, H., Aruna, M. & Raj, M.G. 2015. Potential Sources of Heat in
Underground Mines – A Review. Procedia Earth and Planetary Science. 11:463–468. DOI: 10.1016/j.
proeps.2015.06.046.
McPherson, M.J. 1993. Subsurface Ventilation and Environmental Engineering. Dordrecht: Springer Neth
erlands. DOI: 10.1007/978-94-011-1550-6.
Nelson, H.F. & Sauer, H.J. 2002. Formulation of high-temperature properties for moist air. HVAC&R
Research. 8(3):311–334.
Nie, X., Wei, X., Li, X. & Lu, C. 2018. Heat Treatment and Ventilation Optimization in a Deep Mine.
Advances in Civil Engineering. 2018. DOI: 10.1155/2018/1529490.
Ryan, A. & Euler, D.S. 2017. Heat stress management in underground mines. International Journal of
Mining Science and Technology. 27(4):651–655. DOI: 10.1016/j.ijmst.2017.05.020.
Stull, R. 2011. Wet-bulb temperature from relative humidity and air temperature. Journal of applied
meteorology and climatology. 50(11):2267–2269.
Watson, A.G. 1981. The contribution of conveyed coal to mine heat problems. University of
Nottingham.
Whillier, A. 1969. An improved analysis of counterflow cooling tower performance. Journal of South
African Mechanical Engineer. 18(9):226–232.
199
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
K. Tom
Howden, St-Bruno de Montarville, Canada
ABSTRACT: Deep mines often require cooling to comply with local heat stress regulations.
Attempts to mitigate heat strain via improved heat indices are well documented. However, the
impact of the heat-stress index on refrigeration capacity is less explored. The paper examines the
refrigeration capacity necessary to develop a deep mineshaft as a function of both wet bulb globe
temperature (WBGT) and wet bulb temperature (WB). Thermodynamic simulations in Ventsim
indicated chiller capacity sized entirely from a wet bulb or sigma-heat (σ) balance was insufficient
to prevent elevated heat stress, at working depth, according to the WBGT. The study showed
that the primary factor causing increased wet-bulb depressions (the difference between wet bulb
and dry bulb temperatures), was input heat containing an overwhelming percentage of sensible
heat (dry heat) to total heat. Furthermore, the work investigated active engineering controls like
increased airflow quantity, duct insulation, and secondary cooling to help support the refriger
ation equipment constraints, such as minimum collar WB, and ventilation design criteria. Add
itional research is necessary to develop empirical methods that could facilitate the prediction of
cooling requirements for mine sites that do not employ WB as the design heat index.
1 INTRODUCTION
1.1 Background
As the world’s reliance on essential metals and minerals increases it will become necessary to
mine deeper to extract the ore. Inherent challenges that workers face in deep underground
mines are autocompression and strata heat, which can result in elevated cases of heat stress
arising from increasing temperature and humidity. NIOSH (2016) defines the former as the
heat load that an individual is exposed to from the aggregate contributions of environmental
and physiological factors, which result in a net increase in heat storage in the body. Mines
employ heat stress indices, which is a single number that predicts the level of heat stress in
a hot environment, to comply with local legislations on heat criteria. However, there is a lack
of universal agreement on a one-size-fits-all index due to the following constraints:
• be accurate and feasible,
• account for both environmental and physiological cursors of heat stress,
• calculations or measurements that are easy to perform,
• threshold limits that function under a wide range of metabolic and environmental conditions
• a range that is physiologically or psychologically weighted to inform of an increased risk to
health and safety (Epstein and Moran, 2006; NIOSH, 2016).
State-of-the-art research on the choice of heat stress indices typically involve metabolic rate.
Brake and Bates (2001) demonstrated a method to compare indices on the basis of the particular
index’s maximum metabolic rate. The results indicated that an increase in air velocity permitted
more strenuous work (i.e. higher metabolic rates), however, the authors were unable to define
DOI: 10.1201/9781003429241-21
200
the optimal index as none of the considered indices spanned the full range of maximum meta
bolic rates. Moreover, the method becomes increasingly subjective with the inclusion of worker
acclimatization and clothing factors. Instead, Roghanchi, Sunkpal, Carpenter, and Kocsis
(2016) recommended the use of rational (i.e. physiologically based) or empirical indices, such as
wet bulb globe temperature (WBGT), during the design stage and direct indices, like the psy
chrometric wet bulb temperature (WB), for thermal management. Despite the study’s advances,
it failed to address the practical ramifications that the category of heat stress index would have
on engineering control measures.
2 CASE STUDY
Figure 1. Ventilation schematic with insets from VentsimTM model showing heat and cooling sources.
201
A dual-stage auxiliary fan supplies fresh air to the shaft bottom through a single duct
column. Downstream of the fan assembly and inline with the duct is a bulk air cooler (BAC)
that chills the air to a specified WB at the shaft collar. The cool air ventilates a number of
service and work decks situated on the cutting machine. The diagram’s inset of the cutter area
indicates the heat sources in the different airways. Notably, air exiting “Airway 03” splits two
ways: a small percentage intakes through another duct that connects to a muck filtration
system (“Airway 08” in the inset), while the remainder meanders through the upper decks.
The two air streams subsequently mix and return to the surface through a ducted, single-stage
exhaust fan. Furthermore, the duty point of this fan is such that ambient air makes up
a percentage of the total exhaust flow.
The heat loads in the model consist of autocompression, fan compression, strata, duct-
airway heat transfer, hydration, and electric machinery. All sources, with the exception of the
strata, are treated as sensible heat. The magnitude of evaporation from the rock depends on
the level of shaft wetness and its age. The shaft-cutter supplier provided the quantity and loca
tions of electric heat along the machine, while hydration heat calculates from Nykänen’s equa
tion (Al-Neshawy, 2015).
Figure 2. Productivity loss as a function of psychrometric wet-bulb temperature (Belle and Biffi, 2018).
Moreover, the WB can assess the air’s heat absorption capacity before requiring refriger
ation through the cooling power surplus (CPS) described in equation 1,
where ṁ is the mass flow of dry-air (kg/s); σ is the sigma heat (kJ/kg) with subscripts denoting
the reject and inlet WB, respectively. Sigma heat is calculated from equation 2 (McPherson, 1993)
where ω is the absolute humidity (kgv/kga) with the subscripts, “v” and “a”, signifying
water vapor and dry-air, respectively.
202
Some mines, in particular those located in South Africa and Australia, may impose add
itional constraints like a maximum DB (i.e. 37°C) or an air velocity (V) of at least 0.25 m/s
within a risk management plan that objectively categorizes an environment as being hot
(Webber, Franz, Marx, and Schutte, 2003). One distinct observation is that these conditions
can primarily be interpreted as working with the WB. Outside the two abovementioned coun
tries, the WBGT is widely preferred, instead, because it is easy to calculate and better accounts
for both environmental and physiological predictors of heat stress (Epstein and Moran, 2006).
In underground mines with air temperatures measured away from radiant heat sources, the
index can be expressed by equation 3,
where WBn = natural wet bulb (°C), which can be obtained from equation 4 (Carter,
Zaitchik, Gohlke, Wang, and Richardson, 2020),
where C = 0.85 for V < 0.03 m/s; C = 0.96 + 0.069 log10 V for 0.03 m/s ≤ V ≤ 3 m/s; C = 1
for V > 3 m/s.
The replacement of WB with WBn in the WBGTn calculation partially addresses the criti
cism of the index being insensitive to wind speed. Exposure limits on the WBGT scale can
vary slightly across industrial hygiene organizations due to discrepancies in the cut-off range
for metabolic rates. The limiting values adopted in this study are shown in Table 1,
The reject WB and WBGT exposure limits complement the model input variables specified
in Table 2,
203
3 RESULTS
One method to refine the ventilation and cooling strategy is to compute and understand the
heat load distribution like the chart displayed in Figure 3. It can be observed that the electric
equipment and duct heat transfer constitutes more than half of the total heat load. Duct insula
tion was proposed to reduce this component of heat, however, a review of the shaft fit-out infra
structure rendered this option impractical. Alternatively, the duct heat transfer can be reduced
indirectly through increased flow, which enhances the overall cooling capacity, at the expense of
a step change in fan motor size. The amount of supply air increase is limited by the performance
envelope of the BAC to maintain an identical collar WB without increasing surface cooling cap
acity. A secondary chiller, installed mid-shaft to depress the WB back to collar value, was sug
gested as a contingency plan to implement into the shaft design. These planned measures
prompted the roadheader manufacturer to consider retrofitting the machine with a water cooler
(utilizing the shaft’s existing service and return water pipes) to absorb radiant heat that would
subsequently be pumped back to surface.
Although the strata heat comprises the smallest percentage of the overall heat, the author
surmises, from a simulation perspective, that the wetness factor may play a sensitive role in
the WBGT. For example, a wetter shaft accelerates evaporation, which causes a reduction in
the return air’s DB profile. Consequently, the smaller DB differential, between the shaft’s
return air and the chilled fresh air, dampens the effect of duct heat transfer.
204
Figure 3. Baseline heat load distribution.
205
3.2.1 Heat balance
The combination chart in Figure 5 illustrates the heat balance for all scenarios. The stacked
column represents the heat sources, while the CPS (interpreted as a negative value) is calculated
prior to the cutter at the shaft bottom. The refrigeration quantity is specified to reach the design
WB at the collar and mid-shaft. Ideally, the predicted refrigeration quantity should be the differ
ence between the heat sources and CPS. However, the iterative solution behind duct heat causes
an imbalance in the calculations. Autocompression increases proportionally with mass flow, but
has an equal effect on temperature rise. Fan compression and power increases with flowrate to
the third power, and all three enhanced flow scenarios demonstrate a greater value than baseline.
An overlap between CPS and refrigeration quantity is observed in the case of the second chiller
and equipment water cooling. The larger overlap in the former, linked to nearly twice the refriger
ation, corroborates the above evidence of WBGT exposure limit compliance. On the contrary,
conclusions based on reject WB would argue an overly cooled climate due to excessive engineer
ing contingencies. Interestingly, the case with the mid-shaft chiller has the greatest quantity of
strata heat, attributed to a higher thermal driving potential between the lower DB and prevailing
virgin rock temperature.
4 CONCLUSIONS
206
REFERENCES
Al-Neshawy, F. (2015). Estimation of the heat evolution during the hydration of concrete, mix design of hot
concrete, drying and coating of concrete structures. Slideplayer. https://slideplayer.com/slide/8020402/
Belle, B., & Biffi, M. (2018). Cooling pathways for deep Australian longwall coal mines of the future.
International Journal of Mining Science and Technology. 28 (6): 865–875.
Brake, R. & Bates, G. (2001). A valid method for comparing rational and empirical heat stress indices.
Annals of Occupational Hygiene 46 (2): 165–174.
Brake, R. & Bates, G. (2002). Limiting metabolic rate (thermal work limit) as an index of thermal stress.
Applied Occupational and Environmental Hygiene 17 (3): 176–186.
Carter, A.W., Zaitchik, B.F., Gohlke, J. M., Wang, S., & Richardson, M.B. (2020). Methods for estimat
ing wet bulb globe temperature from remote and low-cost data: A comparative study in Central
Alabama. GeoHealth, 4: 1–16.
Epstein, Y. & Moran, D.S. (2006). Thermal comfort and the heat stress indices. Industrial Health 44:
388–398.
McPherson, M.J. (1993). Subsurface ventilation and environmental engineering. New York, NY: Springer.
NIOSH. (2016). Criteria for a recommended standard: Occupational exposure to heat and hot environments.
U.S. Department of Health & Human Services, Centers for Disease Control & Prevention, National
Institute for Occupational Safety & Health.
Roghanchi, P., Sunkpal, M., Carpenter, K., Kocsis, C. (2016). Application of heat stress indices in under
ground mines. [Poster session]. https://doi.org/10.13140/RG.2.2.18728.16643
Webber, R.C.W., Franz, R.M., Marx, W.M. & Schutte, P.C. (2003). A review of local and international
heat stress indices, standards and limits with reference to ultra-deep mining. The Journal of The South
African Institute of Mining and Metallurgy: 313–324.
207
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J.E. Fox
PhD., Senior Consultant, Mine Ventilation Services, a business unit of SRK Consulting, USA
ABSTRACT: As relatively hot mineral deposits are being developed to considerable depth,
the issue of mitigating heat will require monumental cooling capacity in some cases. For deep
deposits to be mined at increasing depths, mechanical cooling systems used to maintain an
acceptable underground climate will most certainly become increasingly complex and expen
sive. In most methods, the main source of potential cooling energy comes from cold water.
The cold water can come from ice (stowed or transported), in situ reservoirs, or refrigerated
water chillers. This paper investigates the use of in-situ ice as the source of cooling.
Air cooling power is available by passing intake airflow through stowed ice accumulated
during favorable atmospheric conditions. Planning for sustainable cooling potential requires
protected space to successfully maintain ice in storage until needed. Stored ice or cold water
should be judiciously cared for, safely kept from the ill effects of heat, in order to remain
useful for bulk air cooling. Maximizing the production of ice means that atmospheric condi
tions will be monitored closely during the formation phase. Storage facilities would maintain
restricted airflow while minimizing possibilities for heat transfer during the storage phase.
Successful installations will prioritize environmental management to favor long-timespan
usage of sustainable cooling systems. Sustainable cooling systems could reduce or eliminate
the need for mechanical cooling if installed with favorable conditions.
1.1 Background
The current market trend in profitable mining operations involves underground mining activ
ity below a viable mineral deposit which was initially developed with an open pit. Due to the
scale of this type of operation, a popular method from the surface mining perspective is to use
very large haul trucks bringing ore up a long underground ramp system which daylights in the
open pit. This type of ramp mine portal is driven into the most stable strata surrounding the
open pit. This mining arrangement is the proposed application for bulk air cooling via stowed
ice reservoirs.
Triple gallery entries would be driven approximately parallel to the pit wall ramp portal.
This energy conscious arrangement will cool intake ramp air with natural flow through venti
lation. Since many northern Nevada underground mines are accessed through large open pit,
the following ice accumulation gallery arrangement could work successfully in Carlin Trend
or other mining operations that experience cold winters. Triple snow storage galleries devel
oped in bedrock adjacent to the underground ramp would use construction materials to main
tain a sufficiently cool environment that maximizes the volume of ice preserved until the heat
of summer.
DOI: 10.1201/9781003429241-22
208
1.2 Site considerations
Construction of the underground ice reservoir will require diligence in planning and care during
excavation. Selection of the site where the ramp system is driven will be extremely important if
this type of system is to be employed. Pit wall in which the construction will be excavated needs
to be seismically sound and proven to be devoid of significant failure joints, weakness planes or
ground water draws that will decrease the geological structure rating. In other words finding the
most attractive site, rock mechanically speaking, will be of significant importance during the
planning and subsequently affect associated capital and operating costs.
Maintaining ice reservoir liner integrity during construction is one of the project’s highest
quality control priorities. Floors of the reservoir should be concrete that is substantial enough
to withstand freeze/thaw conditions. Walls of the reservoir should be sprayed with shotcrete
as the minimum. Walls and reservoir bottom should be lined with waterproof material.
A cast-in-place reinforced concrete dam will be constructed at the reservoirs’ low point. Add
itional sumps should be driven down grade of the reservoir galleries to ensure any melt water
from the reservoirs be captured before allowing overflow into mine workings.
This type of ice generation system would preferably be excavated into the sound and
extremely stable highwall of an open pit. Preference would be given, for the construction area,
to be in close proximity to the ramp system accessing the underground below. In order to con
tinually verify structural integrity and the host pit wall stability, the snow accumulation gal
lery system should include monitoring wells drilled into the strata surrounding the galleries.
Additionally, groundwater monitoring and seepage drainpipes would be drilled horizontally
below the cold water bulk air cooling service water reservoir.
Detailed slow slip analyses should be conducted in the immediate vicinity of the planned
site with numerous core drill data in the immediate area where the ice accumulation galleries
are to be excavated. Thorough rock mechanics analysis needs to be performed to determine
all possible failure planes in the construction zone. Careful attention should be given during
the core logging so that localized weakness planes and failure zones can be accurately identi
fied. This careful analysis will reveal if required structural load bearing exists to store pro
posed ice banks for 5 months until the ice would be used for cooling.
Snowmaking is a heat exchange process. Heat is removed from snowmaking water by evapora
tive and convective cooling and released into the surrounding environment. This heat creates
a micro-climate inside the snowmaking plume that is very different from ambient conditions.
Understanding this process can lead to practical benefits for the snowmaker. There are many
variables that affect snowmaking. Three of the most important operational factors are wet
bulb temperature, nucleation temperature and droplet size vs. nucleation reagent mixing.
209
2.1 Wet bulb temperature
The temperature of a water droplet exiting a snow gun is typically between 34 °F and 44 °F.
Once a water droplet passes the nozzle and is released into the air, its temperature falls rapidly
due to expansive and convective cooling and evaporative effects. The droplet’s temperature
will continue to fall until equilibrium is reached. This is the effective ambient wet bulb tem
perature and it is as important as dry bulb (ambient) temperature in predicting snowmaking
success. For example, snowmaking conditions of 28 °F and 10% humidity are basically
equivalent to those encountered at 20 °F and 90% humidity.
210
as long as it remains at or below 32 °F, but only after it has first cooled to its nucleation tem
perature. Any excess energy will be dissipated into the atmosphere.
211
latent heat of melting must be applied to the ventilation model as heat sinks in each ice accumu
lation gallery as the respective ice stopes melt when exposed to the warm fresh air supply.
Latent heat due to the moisture in air can be calculated with the following equation:
where hl = latent heat (kW); Q = volumetric airflow (m3/s); ρ = air density (kg/m3); and hwe =
latent heat of vaporization of water (kJ/kg); and dwkg = humidity ratio difference (kg H2O/kg
dry air).
212
Therefore, the assumed latent heat of fusion source added for each active snowmaking
nozzle is 94.3 kW (94.3 = (12.9)(1.186)(2465.56)(0.0025)).
213
Figures 3 through 5, describe the temperature characteristics for key snow formation steps.
Figure 2. Ventsim© model of ice accumulation galleries with no snowmaking or bulk air cooling; air
flow quantities shown in m3/s.
Figure 3. Ventsim© model of ice accumulation Gallery 1 when snowmaking begins; wet bulb/dry bulb
temperatures shown in °C.
Figure 4. Ventsim© model of ice accumulation gallery 2 during initial snowmaking phase; gallery 1 in
initial conditioning phase; wet bulb/dry bulb temperatures shown in °C.
214
Figure 5. Ventsim© model of ice accumulation galleries when snow has fully accumulated in gallery 3;
gallery 1 and 2 in initial conditioning phase; wet bulb/dry bulb temperatures shown in °C.
Figure 6. Ventsim© model of ice accumulation galleries using stowed ice for bulk air cooling; wet bulb/
dry bulb temperatures shown in °C.
This innovative air cooling system will help underground mines reduce the operational carbon
footprint by replacing electrical power with renewable cooling energy. Many factors contrib
ute to the core fundamentals of sustainability. Cultural responsibility to sustainability and
maintaining the environment in a safe and healthy condition lends credibility to mining oper
ations with extensive mineral reserves that could potentially lead to strategic economic viabil
ity. Positioning a mine operation with lower capital expenditure and lower operating expenses
will provide long-term economic sustainability through uncertain economic times. Innovative
use of the environment and responsible care of natural resources available at the mine site will
215
inevitably lead to a sustainable mine. Sustainable cooling systems could reduce or eliminate
the need for mechanical cooling if installed with favorable conditions.
Successful installations will prioritize environmental management to favor long time span
usage of sustainable cooling systems.The ice accumulation reservoirs supplying 855 kW of cool
ing power continuously over 5000+ hours would yield over 4.2 GWh. According to the United
States Environmental Protection Agency (EPA, 2021), 1 kWh is equivalent to 7.09e-4 metric tons
CO2 gas. Conservatively, this renewable cooling system could offset 3,031 tonnes CO2 annually.
Lowering the intake airflow requirement down the ramp will significantly improve the per
formance of the ice accumulation galleries’ contribution into the overall capacity of the mine
air conditioning system. Sensitivity analyses should be conducted to determine optimum air
flow regimes for given atmospheric conditions of relative humidity, temperature and baromet
ric pressure. Additional analyses should be conducted to determine optimum gallery length,
cross sectional area, grade, orientation, trajectory and total number of galleries simulated for
this sustainable mine air conditioning system.
Investigation of practical, operable air cooling systems for underground mining should
include maximum transfer distance, both laterally and vertically, for air cooled from the ice
accumulation galleries. Development of renewable energy air cooling systems for hot under
ground mines will ultimately benefit any operation that can take advantage of this seasonably
available energy resource. Correct sizing of the air cooling system will determine the success of
this type of air conditioning system. Supplying ice harvested near the surface to underground
air cooling systems is a potential source of immense savings for mines that would require bulk
cooling of the intake airflow of the identified mining scenario.
Dr. Jon Fox received his BS in Mining Engineering from Missouri School of Mines in 1987.
He was awarded degrees of MS in Mining Engineering and PhD in Mineral Resources Engin
eering from University of Nevada, Reno in 2012 and 2021 respectively. Jon returned to uni
versity after 25 years, beginning graduate studies in 2010 and the award of Mackay MS
student of the year was awarded Jon in 2012. Dr. Charles Kocsis acted as Jon’s doctoral
advisor since 2017. Jon has over 32 years industry experience in civil, electrical and mining
engineering and contracting services. Jon consults with Mine Ventilation Services of SRK as
216
a mine ventilation engineer, where dedication to the health and safety of the underground
workforce is his highest priority. Jon’s research topics on innovative renewable energy air
cooling systems are timely and well thought out, thanks to all the input received over the
years from fellow professionals in the mining industry.
REFERENCES
217
Mine dust monitoring and control
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Control of dust in underground coal mines is critical for mitigating both
explosibility and respiratory hazards. For decades, the National Institute of Occupational
Safety and Health (NIOSH) has led research to evaluate the effectiveness of various dust con
trol technologies in coal mines, including several recent studies related to auxiliary scrubbers.
While these have demonstrated scrubber efficiency to reduce respirable coal mine dust mass
concentrations, detailed dust characterization was not a focus. The current study aims to
expand on NIOSH’s prior work by evaluating the effect of two different scrubbers (one wet
and one dry) on respirable dust composition and particle size distribution. For this, preserved
filter samples were obtained from NIOSH and analyzed by multiple methods. Results indicate
that dust composition was not appreciably affected by either scrubber. However, as expected,
the wet scrubber appeared to decrease the overall respirable particle size distribution.
Although such an effect of the dry scrubber was not consistently observed, this can be attrib
uted to the fact that the sample collected downstream of the face area cleaned by the dry
scrubber contained mixed dust from the scrubber exhaust and other parts of the mine.
1 INTRODUCTION
Dust in coal mines presents both safety and health hazards. Control of airborne float dust is
a critical strategy for mitigating explosibility hazards (Shahan et al., 2017), and control of res
pirable coal mine dust is crucial for preventing occupational lung disease (Blackley et al., 2018;
Hall et al., 2019). Indeed, dust control has been a key focus of the mining research portfolio at
the National Institute of Occupational Safety and Health (NIOSH)—including work on scrub
bers (Colinet et al., 2013; Colinet & Jankowski, 2000; Janisko et al., 2015; Organiscak et al.,
2016; Organiscak & Beck, 2010; Patts et al., 2016). Scrubbers are commonly used on continu
ous miner machines (CM) to capture dust being generated at the cutting face, and thus prevent
its transport into the return airways and contamination of the ventilating air more broadly
(NIOSH, 2021). CM-mounted scrubbers, referred to as flooded-bed scrubbers, essentially cap
ture particles by forcing the dust-laden air through a wet filter where they either deposit or
attach to water droplets that are removed by a demister (NIOSH, 2021; Patts et al., 2016).
To improve dust control in other priority areas of underground coal mines (i.e., not neces
sarily coupled with the CM operation), auxiliary (stand-alone) scrubbers could be another
option. Recently, NIOSH has conducted field studies of a wet auxiliary scrubber (i.e., based
on the traditional flooded-bed type) (Janisko et al., 2015; Patts et al., 2016), as well as a dry
unit that can provide even more flexibility in terms of placement in the mine (Organiscak
et al., 2016). Since the primary objective of these studies was to evaluate the overall dust col
lection efficiency of each scrubber, detailed dust characterization was not a focus. However,
NIOSH did preserve the dust samples, which now enable further analysis. In the current
work, those samples were used to investigate the effect of the aforementioned wet and dry
auxiliary scrubbers on dust composition and particle size.
DOI: 10.1201/9781003429241-23
221
2 SUMMARY OF NIOSH STUDIES ON AUXILIARY SCRUBBERS
The sections below overview the relevant NIOSH studies and available dust samples that are
the subject of the current work.
222
ranged from 2720-4900 cfm (1.28 - 2.31 m3/s), while the initial airflow supplied to the face areas
(i.e., without the scrubber operating) was 1660-4890 cfm (0.784 - 2.31 m3/s). Organiscak et al.,
(2016) reported that during the tests in Section 1 the scrubber intermittently turned off and had
to be restarted. (Notably, they also collected noise measurements and concluded that, under the
conditions tested in Mine B, workers should not be overexposed to noise. However, under higher
airflows and as the scrubber filter becomes loaded with dust, Organiscak et al., (2016) noted that
noise over-exposure could be possible.)
During each test, Organiscak et al., (2016) sampled in three stationary locations as shown in
Figure 1b: just upstream of the scrubber, downstream of the face area cleaned by the scrubber
(i.e., the face in the return entry), and further downstream in the return. Samples were also col
lected in the RB intake and on the RB, though these are not included in the current work. The
sampling duration for each test was between about 340-400 minutes. In each location, respirable
dust samples were collected using the same equipment, flowrate, filters, and cassettes as used in
the wet scrubber study for respirable dust, and a real-time monitor was also used (pDR-1000,
Thermo Scientific, Waltham, MA). The pDR data had to be calibrated (i.e., using the filter sam
ples) to estimate respirable dust concentration, which could then be correlated back to time
periods of interest (e.g., to look at specific scenarios of RB and CM placement relative to the
scrubber). A pDR was also used over short periods to monitor right at the scrubber exhaust (i.e.,
before mixing with other air could occur in the face area that should be cleaned).
Comparing pDR-derived respirable dust concentrations just upstream of the dry scrubber
versus at its exhaust indicated an actual collection efficiency of more than 90%. That said,
across all tests in Mine B, the respirable concentration was only reduced by an average of
about 50% in the face area cleaned by the scrubber during times when the CM and/or RB
were operating upstream (Organiscak et al., 2016). The discrepancy was attributed to mixing
of the clean scrubber exhaust with more dusty air in face area being sampled.
Figure 1. Schematics of sampling locations in NIOSH studies of a) an auxiliary wet scrubber in Mine
A (modified from(Janisko et al., 2015)) and b) an auxiliary dry scrubber in Mine B (modified from (Organis
cak et al., 2016)). Only samples from the locations denoted by green circles were included in the current work.
223
a clean test tube; these filter pieces from respirable samples were later prepared for dust particle size
analysis by scanning electron microscope with energy dispersive X-ray (SEM-EDX). The rest of
each filter was placed into another clean test tube and prepared for thermogravimetric analysis
(TGA) to estimate the mass fractions of coal, non-carbonate minerals, and carbonates, to loosely
approximate the major sources of dust (i.e., coal and rock strata being mined, and rock dust prod
ucts being applied, respectively) (Agioutanti et al., 2020; Jaramillo et al., 2022).
224
be used as a crude comparison of these relative heights. For example, Cuts 1, 2 and 6 appear
to show a high coal to non-carbonate content ratio, suggesting more of the mining height was
in coal than in rock during these cuts (as compared to Cuts 4 and 5).
Likewise, the TGA data for the carbonates content, which indicates rock dusting activities
in the mine, show differences between cuts. For instance, Cuts 1, 4 and 5 show more carbon
ates than Cuts 2 and 6, as seen in Table 1. While rock dusting was halted during the dust sam
pling by NIOSH, Janisko et al., (2015) did report that operation of the auxiliary fan with the
scrubber appeared to re-entrain rock dust that had been applied earlier. This could explain the
relatively high carbonates content during Cuts 4 and 5, though high carbonates during Cut 1
(when the auxiliary fan and scrubber were not in place) shows that significant rock dust was
also in the -40 µm dust fractions during periods without the fan. This is consistent with obser
vations of CM section return samples in other mines (Sarver et al., 2019).
Table 1. Summary of dust samples from NIOSH’s wet scrubber study in Mine A that were selected for
analysis here. (NCB = non-carbonate minerals; CB = carbonates).
TGA analysis
225
In general, the TGA results also show that dust sourced from the rock dust application (i.e.,
carbonates) tends to be more concentrated in the larger size fractions, whereas the mine
strata-sourced dust (i.e., coal+non-carbonates) is more concentrated in the finer (including
respirable) fractions. However, no discernable trends could be observed in dust composition
moving from the 200 to 400 feet sampling locations.
With respect to particle size, Figure 2 shows the cumulative size distributions for the respir
able samples at the 200- and 400-feet locations for all five CM cuts. The distributions were
clearly coarser when the scrubber was not in use and finer when it was. This is consistent with
expectations as the scrubber should be more efficient in removing coarser particle sizes
(NIOSH, 2021). Size distributions did not generally change much between the 200- and 400-
feet locations, which is not surprising since respirable-sized particles should remain entrained
in the mine air over long distances. On the other hand, differences in size distributions between
the 200- and 400-feet locations might be expected for larger size fractions (e.g., 20 or 40 µm
samples), though these were not analyzed in the current work. It is well established that float
dust (containing the larger particle sizes) deposition varies with distance from the mine face
(Courtney et al., 1986; Janisko et al., 2015; Patts et al., 2016; Shahan et al., 2017).
Figure 2. Overall particle size distribution for respirable dust collected in the 200 and 400 ft locations
during all five CM cuts reported in NIOSH’s wet scrubber study.
226
found to be slightly finer than dust in the UDS location. This is consistent with expectations
(i.e., the scrubber should more efficient at removing coarser particles), and it must be reiter
ated that the DDS samples do not perfectly represent the dry scrubber exhaust, but rather the
face area that was supposed to be cleaned by the scrubber. Thus, DDS samples represent
a mix of the scrubber exhaust and other mine air in the return entry face area—and the true
particle size reduction yielded by the dry scrubber is likely muted in the DDS samples. On the
other hand, in Section 1, it appears that the dust size distribution was actually coarser in the
DDS location as compared to the UDS location (Figure 3a), which is seemingly inconsistent
with expectations. In this section, it is possible that any size reduction yielded by the scrubber
was completely overwhelmed by coarser particles in the return air that mixed with the scrub
ber exhaust in the DDS location. Moreover, Organiscak et al., (2016) noted an issue with the
scrubber intermittently turning off and needing to be restarted during the testing in Section 1.
It is also possible that these startup events dislodged coarse particles from the scrubber filter,
which might have affected the size distribution results in the DDS location.
Table 2. Summary of dust samples analyzed from NIOSH’s dry scrubber study in Mine B and the TGA
results. (NCB = non-carbonate minerals; CB = carbonates).
TGA Analysis
The current work demonstrates that detailed dust analysis can be performed on appropriate
samples preserved from prior studies. This approach has obvious merits: it enables further
insights to be gained about the effects of particular dust controls or sampling conditions on
respirable dust characteristics, without the significant time, cost and logistical efforts needed
to conduct new field work. With these benefits, however, come clear limitations too: follow-
up work on preserved samples, and interpretation of results, is constrained by the original
study design. Regarding NIOSH’s original studies that were revisited for the current work,
their sampling designs were geared toward specific research objectives, which did not include
the sort of dust sample analysis performed here. For example, samples were not collected at
the wet scrubber intake or dry scrubber exhaust, which somewhat limits the conclusions that
can be drawn about the direct effects of either scrubber on particle sizes or specific dust
components.
227
5 CONCLUSIONS
NIOSH previously evaluated a wet and a dry auxiliary scrubber and showed that both could
be effective for reducing respirable dust and/or airborne float dust mass concentrations in
underground coal mines. Here, preserved dust samples from those prior studies were further
analyzed to investigate the effect of the scrubbers on dust composition and particle size.
Results indicated that neither scrubber has an appreciable effect on dust composition. Both
scrubbers are expected to more efficiently remove coarser particles from the air that they treat
—meaning that air exiting the scrubber should have a finer size distribution than the air enter
ing the scrubber. This effect was evident for the wet scrubber, but results were less clear for
the dry scrubber due to mixing of the scrubber exhaust with return air and/or intermittent
issues with the scrubber power during testing.
While the current work implies that scrubbers are not necessarily more efficient on one gen
eral component of the respirable dust (i.e., dust source from the mine face versus from rock
dusting activities), future studies should focus on specific constituents of interest. For
example, crystalline silica is considered to be the most hazardous constituent and can have
a different size distribution than other dust constituents (Sarver et al., 2021). Moreover, the
effects of other controls, such as water sprays, on particle sizes or specific dust constituents, or
dust generated from specific sources, should be investigated.
Figure 3. Results of Sections 1 and 2 (i.e., the dry scrubber performance evaluation) overall particle size
distributions from SEM-EDX analysis and based on two locations (i.e., Upstream the dry
scrubber-UDS, and Downstream the face area cleaned by the dry scrubber-DDS). Plot a (Section 1) and
plot b (Section 2) represent the overall PSD for all the major dust constituents (i.e., coal, non-carbonate
and carbonates).
ACKNOWLEDGEMENT
The authors are grateful to CDC/NIOSH for funding this research (contract
75D30119C05529). Special thanks are extended to Jay Colinet, Milan Yekich, and George
Luxbacher for their support to identify and provide existing dust samples and related field
data. We also thank Justin Patts for his assistance with data interpretation for dust samples
228
collected in Mine A. We wish to acknowledge all of the other authors involved in both of the
NIOSH studies revisited here.
REFERENCES
Agioutanti, E., Keles, C., & Sarver, E. (2020). A thermogravimetric analysis application to determine
coal, carbonate, and non-carbonate minerals mass fractions in respirable mine dust. Journal of Occu
pational and Environmental Hygiene, 17(2–3), 47–58. https://doi.org/10.1080/15459624.2019.1695057
Blackley, D. J., Halldin, C. N., & Scott Laney, A. (2018). Continued Increase in Prevalence of Coal
Workers’ Pneumoconiosis in the United States, 1970-2017. American Journal of Public Health, 108(9),
1220–1222. https://doi.org/10.2105/AJPH.2018.304517
CFR. (2021). Code of Federal Regulations-Title 30: Mineral Resources. https://www.govinfo.gov/content/
pkg/CFR-2020-title30-vol2/pdf/CFR-2020-title30-vol2-chapII.pdf
Colinet, J. F., & Jankowski, R. A. (2000). Silica collection concerns when using flooded-bed scrubbers.
Mining Engineering (Littleton, Colorado), 52.
Colinet, J. F., Reed, W., & Potts, J. D. (2013). Impact on Respirable Dust Levels When Operating a
Flooded-bed Scrubber in 20-foot Cuts. DHHS (Niosh) Publication 2014–105, Ri 9693.
Courtney, W. G., Cheng, L., & Divers, E. F. (1986). Deposition of Respirable Coal Dust in an Airway.
Report of Investigations - United States, Bureau of Mines.
Hall, N. B., Blackley, D. J., Halldin, C. N., & Laney, A. S. (2019). Current Review of Pneumoconiosis
Among US Coal Miners. Current Environmental Health Reports, 6(3), 137–147. https://doi.org/
10.1007/S40572-019-00237-5
Janisko, S. J., Colinet, J. F., Patts, J. R., Barone, T., & Patts, L. D. (2015). Field evaluation of an inline
wet scrubber for reducing float coal dust on a continuous miner section. SME Annual Meeting, 1–6.
Jaramillo, L., Agioutanti, E., Ghaychi Afrouz, S., Keles, C., & Sarver, E. (2022). Thermogravimetric
analysis of respirable coal mine dust for simple source apportionment. Journal of Occupational and
Environmental Hygiene, 19(9), 568–579. https://doi.org/10.1080/15459624.2022.2100409
NIOSH. (2021). Best Practices for Dust Control in Coal Mining. In J. By Colinet, Jay F. Halldin, Cara
N. Schall (Ed.), U.S. Department of Health and Human Services, Centers for Disease Control and Pre
vention, National Institute for Occupational Safety and Health, DHHS (NIOSH) (Second Edi). https://
doi.org/https://doi.org/10.26616/NIOSHPUB2021119
Organiscak, J. A., Noll, J., Yantek, D., Kendall, B., & Fletcher, J. H. (2016). Examination of a newly
developed mobile dry scrubber (DS) for coal mine dust control applications. SME Annual Meeting,
1–7.
Organiscak, J., & Beck, T. (2010). Continuous Miner Spray Considerations for Optimizing Scrubber Per
formance in exhaust ventilation systems. Society for Mining, Metallurgy, and Exploration (U.S.).
Meeting (2010 : February 28-March 3 : Phoenix, AZ). https://stacks.cdc.gov/view/cdc/9654
Patts, J. R., Colinet, J. F., Janisko, S. J., Barone, T. L., & Patts, L. D. (2016). Reducing float coal dust:
Field evaluation of an inline auxiliary fan scrubber. Mining Engineering, 68(12), 63–68. https://doi.org/
10.19150/me.6883
Sarver, E., Keleş, Ç., & Afrouz, S. G. (2021). Particle size and mineralogy distributions in respirable dust
samples from 25 US underground coal mines. International Journal of Coal Geology, 247, 103851.
https://doi.org/10.1016/J.COAL.2021.103851
Sarver, E., Keles, C., & Rezaee, M. (2019). Beyond conventional metrics: Comprehensive characteriza
tion of respirable coal mine dust. International Journal of Coal Geology, 207(September 2018), 84–95.
https://doi.org/10.1016/j.coal.2019.03.015
Shahan, M. R., Seaman, C. E., Beck, T. W., Colinet, J. F., & Mischler, S. E. (2017). Characterization of
airborne : Float coal dust emitted during continuous mining, longwall mining and belt transport.
Mining Engineering, 69(9), 61–66. https://doi.org/10.19150/me.7746
229
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: The characterization of respirable dust on the basis of constituent fractions and
particle sizes is increasingly of concern for evaluating exposure hazards. For high-resolution par
ticle analysis, scanning electron microscopy with energy dispersive X-ray (SEM-EDX) can be an
effective tool. However, it requires particles to be deposited on a smooth, uniform substrate such
as a polycarbonate (PC) filter for optimal results. While direct sampling onto PC is possible, this
is not the standard approach in many situations. For example, in coal mines, respirable dust sam
ples have typically been collected onto polyvinyl chloride (PVC) filters because they are intended
for gravimetric and/or infrared spectroscopy analysis. Such fibrous substrates are not ideal for
SEM-EDX (or other microscopy), but an effective method to recover and redeposit the dust par
ticles could render such samples suitable for the additional analysis. Here, we present a simple
method and compare SEM-EDX results for paired samples analyzed (1) directly on PC and (2)
following recovery from PVC and redeposition on PC. The results indicate that dust recovery
from PVC is generally representative in terms of mineral constituents, although some changes in
particle size distributions can be observed. Careful control of particle loading density is import
ant for either direct-on-filter or recovered dust analysis.
1 INTRODUCTION
While routine monitoring of inhalable and respirable particulates typically involves measuring mass
concentration, there is increasingly a focus on particle characteristics to better understand exposure
hazards (Stanek et al. 2011, Shekarian et al. 2021, Kumar et al. 2021). For instance, particle size
can control lung deposition and relative surface area (e.g., Sioutas et al. 2005, Ramgolam et al.
2009, Oberdorster, 2005); and particle surface area and chemistry can control lung response (e.g.,
Dalal et al. 1995, Leung et al. 2012, Ross & Murray 2004, Harrington et al. 2012). In light of the
unexpected resurgence of occupational lung disease among US coal miners (Laney et al. 2010),
a consensus report from the National Academies of Science, Engineering and Medicine recently
called for more in-depth characterization of respirable coal mine dust (RCMD) to enable a clearer
understanding of both dust sources and health implications (NASEM 2018).
Although various instruments are available for generic particle counting and sizing, to con
duct chemical or mineralogical analysis at the individual particle level, microscopy has been
used (Abbasi et al., 2021). For particles in the respirable range (i.e., on the order of microns
and smaller), scanning electron microscopy with energy dispersive X-ray (SEM-EDX) can be
used to simultaneously acquire particle size and shape data (i.e., from the SEM images) along
with chemistry (i.e., from the EDX spectra); and the chemistry can be interpreted to infer min
eralogy. The particle data generated by the SEM-EDX can be accumulated to estimate distri
butions for the entire sample. Due to the time-consuming nature of manually analyzing
individual particles by SEM-EDX, computer algorithms can be used to automate the process
and more easily generate quantitative data on a per sample basis (i.e., which requires analysis
of relatively many particles in a sample). In fact, computer-controlled SEM-EDX routines
specific to RCMD analysis have previously been developed and used by the authors (e.g., see
DOI: 10.1201/9781003429241-24
230
Johann-Essex et al. 2017, Sarver et al. 2019, Sarver et al. 2021) as well as other researchers
(e.g., Pan et al. 2021, Salinas et al. 2022).
One important condition for automated SEM-EDX analysis is that the particles are deposited on
a suitable substrate—meaning one that allows for an image processing routine to accurately identify
particles in contrast to the background and also exhibits uniform composition so as not to interfere
with EDX analysis on the particles. Track-etched polycarbonate (PC) filters are commonly used for
this work because their smooth background, low impurity, and consistently sized pores are ideal for
both sample collection and analysis (Willis et al. 2002). If a dust sample is collected on PC or other
suitable filter media, SEM-EDX analysis can potentially be conducted directly on the sample filter.
However, in many cases, samples are collected for other types of analyses (Jensen & Schafer 1998,
Kulkarni et al. 2011). For example, polyvinyl chloride (PVC) filters are commonly used to sample
respirable dust when the primary aim is gravimetric analysis to determine the mass concentration in
the environment; the fibrous structure of PVC helps trap particles to avoid sample mass loss during
handling, and it is non-hygroscopic which minimizes effects of humidity on weight measurements.
For respirable mine dust, including RCMD, sampling on PVC filters also enables the determination
of quartz using infrared spectroscopy methods such as NIOSH Method 7603 (NIOSH 2003), MSHA
Method P7 (MSHA 2008), or the direct-on-filter Fourier Transform infrared method that NIOSH
has been working to develop (Miller et al. 2012, Miller et al. 2017, Cauda et al. 2016). To use such
PVC filter samples for microscopy analysis though, the dust particles should ideally be recovered and
redeposited on a suitable substrate like PC (Jensen & Schafer, 1998). This is because, even if the PVC
composition itself does not interfere with EDX analysis of the particles, the fibrous nature of the
material can hinder particle identification (versus background), especially for automated work, and
indeed particles embedded in the filter (rather than sitting on the surface) may not be seen at all.
Particulate recovery from air sampling filters has been demonstrated for applications such as
forensic analysis of dust and microbial particles from filters in heating, ventilation, and air condition
ing (HVAC) systems (Stanley et al. 2008, Mahdavi & Siegel 2020); extraction of particulate matter
mass (PM10 or PM2.5) from PTFE-coated filters for supporting toxicological analysis (Roper et al.
2018); and the removal of PM2.5 from quartz filters to determine the oxidative potential of particu
late to cause adverse health effects (Fang et al. 2016). Likewise, a simple method for recovery of
respirable dust from standard PVC filters for purposes of microscopy to understand dust sources
and potential health impacts would be quite valuable. However, to our knowledge, such a method
has not been demonstrated heretofore. The current work aims to evaluate the recovery and redepos
ition of RCMD from a PVC filter onto a PC filter for SEM-EDX analysis. For this, paired PVC
and PC filters were used for dust collection and then SEM-EDX results for particle characteristics
were compared for each pair of directly analyzed (D) and recovered dust samples (R).
2 METHODOLOGY
A total of 10 pairs of respirable dust samples were generated in the laboratory for this study,
and each pair consisted of 1 PC filter and 1 PVC filter. Four powdered source materials were
used in different ratios to create the sample pairs: coal, rock dust, kaolinite, and silica. These
materials were chosen to represent the major particle types observed in real RCMD samples
when using SEM-EDX to classify particles by their mineralogy (Johann-Essex et al. 2017,
Sarver et al. 2021, Pokhrel et al. 2021a). For the coal, a clean coal product was provided by
a mine partner and then pulverized to -230 mesh (<63 µm) before dust sample generation. The
rock dust, which is primarily calcium carbonate, was sourced from a real rock dust product
provided by a mine partner. It did not need to be pulverized as it was already in a powdered
form. To contribute silica dust, MIN-U-SIL® 5 (US Silica, Katy, TX, USA) was used; this is
a high-purity silica product with a topsize of 5 µm. As a surrogate for alumino-silicates, in
general, a high-purity kaolinite powder was used (Ward’s Science, Rochester, NY, USA).
The dust samples were generated in a small enclosure per the method described by (Agioutanti
et al. 2020). Briefly, a small mass (1-10 mg) of each source material was placed at the bottom of the
enclosure; the ratio between materials was varied for each of the 10 sampling events to achieve dif
ferent ratios of dust constituents in the samples from each event. The dust materials were
231
aerosolized using short pulses of compressed high-purity air. During each event, two respirable dust
samples (i.e., a pair) were collected in 2-piece styrene cassettes using standard equipment consisting
of a 10-mm Dorr-Oliver nylon cyclone, an ELF escort air pump (Zefon International, Ocala, FL)
operating at 2 liters/minute. One sample was collected onto a 37-mm PC filter with track-etched 0.4
µm pores (Zefon International, Ocala, FL, USA, and the other was collected onto 37-mm PVC
filters with a nominal 5.0 µm pore size (Zefon International, Ocala, FL, USA).
The SEM-EDX analysis (described below) was conducted directly on each of the PC filter
samples obtained during dust collection, and these are referred to from here as samples D1-
D10 (to denote the sample pair/sampling event number). For the PVC filter samples, the dust
was recovered and then redeposited on a PC filter for analysis (see Figure 1). Briefly, this pro
cedure was conducted as follows: (1) the PVC filter sample was submerged in a test tube con
taining isopropyl alcohol (IPA); (2) the test tube was sonicated for three minutes at 30 C to
dislodge the respirable dust from the PVC filter; (3) the PVC filter was removed from the test
tube; (4) the dust suspension in IPA was pulled through a 47-mm PC filter (0.4 µm pore size)
using a vacuum filtration; and (5) the PC filter was placed in clean cassette and left overnight
to dry completely. The recovered dust samples are referred to as R1-R10 from here.
Figure 1. Sequence for handling each respirable dust sample pair before SEM-EDX analysis.
The D (direct) and R (recovered) samples were prepared for SEM-EDX analysis by carefully
cutting a 9-mm circular subsection from each filter using a stainless-steel trephine. These subsections
were attached with double-sided tape to aluminum stubs and then sputter coated using Au/Pd to
make the samples conductive. SEM-EDX analysis was conducted using an FEI Quanta 600 FEG
environmental scanning electron microscope (Hillsboro, OR, USA) which was equipped with
a backscatter electron detector and a Bruker Quantax 400 EDX spectroscope (Ewing, NJ, USA)
(Sarver et al. 2021). Bruker’s Esprit software (Version 1.9.4) was used to run an automated routine
on each sample to find and analyze about 500 total particles in the range of about 1,000 nm -
10,000 nm. This “supramicron” routine and instrument settings were described in detail by Sarver
et al. (2021). Some other key parameters of the SEM-EDX analysis included a voltage of 15 kV,
1000x magnification (yielding a frame area of 14025 µm2), 5.5 spot size, and 12.5 mm working dis
tance. For each particle, the length and width were recorded, and the elemental spectra were cap
tured. Based on its normalized elemental ratios (i.e., for elements C, O, Al, Si, Ca, Mg, Fe, and Ti),
each particle was binned using the criteria reported by Sarver et al. (2021) into one of eight mineral
ogy classes: carbonaceous (C), mixed-carbonaceous (MC), aluminosilicates (AS), silica (S), other
silicates (SLO), carbonates (CB), heavy minerals (M), or other (O). Then, the per-particle size and
classification data were used to estimate the particle count % in each mineralogy class per sample,
again following the approach by Sarver et al. (2021).
232
3 RESULTS AND DISCUSSION
Table 1 shows a summary of the results for all 10 sample pairs analyzed in this study. In addition
to the mineralogy distributions (particle count % in each class), the D50 for the particles in each
class and all particles analyzed in each sample is shown, as is the particle loading density (PLD).
The D50 represents the size at which 50% of the particles are finer than this particular size. The
PLD represents the number of particles observed under the SEM per unit area (i.e., particles/µm2).
Table 1. Summary of results for each pair of lab-generated respirable dust samples included in this
study.
Estimated Particle Count (%)
D50 (µm)
Pair Sample PLD C MC AS S SLO CB M O All p-value
Figure 2 shows mineralogy distributions for all ten sample pairs. Visually, there appears to be
good agreement between the D and R samples in most pairs. This indicates the method used to
recover particles from the PVC filters and redeposit them for SEM-EDX analysis generally yields
233
a representative sample. To statistically test for the similarity between the D and R sample results
for each pair, the Freeman-Halton exact test of independence was conducted (Agresti 1992). Here,
the null hypothesis is that the sample mineralogy is independent of whether the dust is analyzed
directly or after being recovered and redeposited. P-values for each sample pair are shown in
Table 1 for 95% confidence, and confirm that mineralogy distributions are indeed similar for seven
of the 10 pairs (i.e., p > 0.05). For the other three pairs (Pairs 3, 5, and 7), a close inspection of the
mineralogy results in Table 1 and Figure 2 reveals that disagreements between the D and
R samples are primarily due to shifts in the particle count % reporting to the C, AS and CB classes.
For Pairs 3 and 5, the SEM-EDX analysis indicated the D sample had relatively more C (i.e., coal)
but less AS (i.e., kaolinite) content than its R counterpart—and this shift in mineralogy was concur
rent with a relatively large increase in the PLD from the D to R sample. On the other hand, for
Pair 7, the D sample had less C but more AS and CB (i.e., rock dust) than its R counterpart—and
the PLD decreased from the D to R sample. Another possibility for the disagreement for pairs 3
and 5 may have been the high amount of kaolinite powder, >50%, used to generate the dust sam
ples. Because of the SEM-EDX’s tendency to erroneously identify the C present in coal as AS pre
sent in the kaolinite, the high amount of respirable kaolinite particles surrounding coal particles
may have caused coal particles to be identified as kaolinite.
Figure 3 illustrates the apparent shifts in mineralogy from D to R samples as a function of
the difference in those samples’ PLDs for all 10 pairs. (Note that only the C, MC, AS, S, and
CB classes are shown since these accounted for nearly all particles in each sample.) The shift
Figure 2. Mineralogy distribution results (based on particle counts) for the 10 pairs of directly analyzed
(D) and recovered dust samples (R).
Figure 3. Difference in mineralogy class (i.e., C, MC, AS, S or CB) distributions based on particle
counts versus the difference in PLD (particles/µm2) for the 10 pairs of directly analyzed (D) and
recovered dust samples (R).
234
Figure 4. Cumulative particle size distributions for the directly-analyzed (D) and recovered dust sam
ples (R) in all 10 sample pairs. The y-axis in each plot shows the % of total particles finer than the respect
ive particle size shown on the x-axis.
between C and AS is correlated with changes in PLD and can be explained at least in part by
the increased likelihood of misclassifying C particles as PLD is increased. In essence, very
nearby mineral particles (e.g., rich in Al and Si) can interfere with the EDX spectrum collected
on a coal particle (i.e., which, itself, is presumably deficient in Al and Si), such that it might be
classified as mineral (e.g., AS) instead of C. The possibility of particle misclassification due to
high loading has been recently discussed in other studies (Keles et al., 2022, Pokhrel et al.,
2021b) and can potentially affect other particles too (e.g., silica). That said, Figure 3 indicates
changes in PLD between the D and R samples did not have much effect on the classification of
235
particles in the MC, S, or CB classes. Notably, the tendency for D samples to have more CB
than their R counterparts could mean that CB particles were somewhat more susceptible to
loss during the dust recovery procedure than other particle types.
Figure 4 allows a comparison of the D and R samples concerning their overall particle size
distributions. Across all 10 sample pairs, a slight reduction in the size distribution can be
observed for the recovered dust. This is likely due to some deagglomeration during the recov
ery procedure. A recent study by Gonzalez et al. (2022) showed that respirable-sized agglom
erates were present in dust samples generated in the laboratory under similar conditions to
those used here and that deagglomeration could be promoted by sonication. Aside from yield
ing a slightly finer size distribution, deagglomeration might also contribute to the small shift
toward more C and less AS in some of the R samples compared to their D counterparts (e.g.,
if coal-mineral agglomerates break up)—though it is impossible in the current work to separ
ate this effect from that of the PLD (Figure 3).
When comparing the overall particle size distributions seen in Figure 4 concerning mineral
ogy class, the observed trends do support that the recovery method used yields a representative
sample of the present particles. The largest observed shift in particle sizes can be seen for the
C and AS mineralogy classes as seen in samples 3, 5, and 7. In these pairs, the R samples are
always finer than the D samples. This is likely because of the deagglomeration which occurs as
a result of sonication when recovering dust from the PVC filters onto a PC filter. Sonication
would cause fine AS particles to separate from coarser C particles. The effects of the sonication
are also evident in the D50 values for C and AS when comparing samples 3, 5, and 7 as the
D50 values for the R samples are lower than the D samples. This reinforces the fact that deag
glomeration resulting from sonication can have an impact on particle size distribution.
4 CONCLUSION
A simple method for recovering respirable dust from fibrous sampling filters can enable post
hoc analysis of individual particles. This could be useful in a variety of situations including,
for instance, the case where preserved mine dust samples could be revisited to explore particle
characteristics—even if such analysis was not the original intent of the sample collection.
Overall, the methodology demonstrated here appears reliable for recovering a representative
sample of dust from fibrous PVC filters—especially if particle loading density (PLD) can be
controlled when the recovered dust is redeposited on the PC filter for SEM-EDX analysis. For
the current study, where dissimilarities were observed in mineralogy distributions between dir
ectly analyzed and recovered dust sample pairs, differences in PLD appear to be the primary
factor. The potential for deagglomeration of dust particles during the recovery procedure must
also be considered with regard to the analytical objectives and interpretation of results.
ACKNOWLEDGEMENTS
The authors would like to thank the Alpha Foundation for Improvement of Mine Safety and
Health (grant AFCTG20-104) and the National Institute of Occupational Safety & Health (con
tract 75D30119C05529) for funding this work. Additionally, we acknowledge the Institute for
Critical Technology and Applied Sciences’ Nanoscale Characterization and Fabrication Labora
tory (NCFL), which houses the SEM-EDX facilities used here. The NCFL is supported by the
Virginia Tech National Center for Earth and Environmental Nanotechnology Infrastructure
(NanoEarth), a member of the National Nanotechnology Coordinated Infrastructure (NNCI),
supported by NSF (ECCS 1542100 and ECCS 2025151). Views expressed in this study are those
of the authors and not necessarily the views of sponsors or research partners.
REFERENCES
Abbasi, B., Wang, X., Chow, J. C., Watson, J. G., Peik, B., Nasiri, V., Riemenschnitter, K., &
Elahifard, M. (2021). Review of respirable coal mine dust characterization for mass concentration,
size distribution, and chemical composition. Minerals, 11(4), 426.
236
Agioutanti, E., Keleş, Ç, & Sarver, E. (2020). A thermogravimetric analysis application to determine
coal, carbonate, and non-carbonate minerals mass fractions in respirable mine dust. Journal of Occu
pational and Environmental Hygiene, 17(2-3), 47–58.
Agresti, A (1992). A survey of exact inference for contingency tables. Statistical Science, 7, 131–153.
Cauda, E., Miller, A. & Drake, P. (2016) Promoting early exposure monitoring for respirable crystalline
silica: Taking the laboratory to the mine site. Journal of Occupational and Environmental Hygiene, 13,
D39–D45.
Dalal, N.S., Newman, J., Pack, D., Leonard, S., & Vallyathan, V. (1995). Hydroxyl radical generation by
coal mine dust: possible implication to coal workers’ pneumoconiosis (CWP). Free Radic. Biol. Med.
18 (1), 11–20.
Fang, T., Verma, V., Bates, J. T., Abrams, J., Klein, M., Strickland, M. J., Sarnat, S.E., Chang, H.H.,
Mulholland, J.A., Tolbert, P.E., Russell, A.G., & Weber, R. J. (2016). Oxidative potential of ambient
water-soluble PM 2.5 in the southeastern United States: contrasts in sources and health associations
between ascorbic acid (AA) and dithiothreitol (DTT) assays. Atmospheric Chemistry and Physics, 16
(6), 3865–3879.
Gonzalez, J., Keleş, Ç, & Sarver, E. (2022). On the occurrence and persistence of coal-mineral microag
glomerates in respirable coal mine dust. Mining, Metallurgy & Exploration, 39(2), 271–282.
Harrington, A.D., Hylton, S., Schoonen, M.A.A., 2012. Pyrite-driven reactive oxygen species formation
in simulated lung fluid: Implications for coal workers’ pneumoconiosis. Environ. Geochem. Health 34
(4), 527–538.
Jensen, P. A., & Schafer, M. P. (1998). Sampling and characterization of bioaerosols. NIOSH manual of
analytical methods, 1(15), 82–112.
Johann-Essex, V., Keleş, Ç, & Sarver, E. (2017). A computer-controlled SEM-EDX routine for charac
terizing respirable coal mine dust. Minerals 7 (1), 15. https://doi.org/10.3390/min7010015.
Keles, C., Pokhrel, N., & Sarver, E. (2022). A Study of Respirable Silica in Underground Coal Mines:
Sources. Minerals, 12(9), 1115.
Kulkarni, P., Baron, P. A., & Willeke, K. (Eds.). (2011). Aerosol measurement: principles, techniques, and
applications. John Wiley & Sons.
Kumar, P., Kalaiarasan, G., Porter, A.E., Pinna, A., Klosowski, M.M., Demokritou, P., Chung, K.F.,
Pain, C., Arvind, D.K., Arcucci, R., & Adcock, I.M. (2021). An overview of methods of fine and ultra
fine particle collection for physicochemical characterization and toxicity assessments. Science of the
Total Environment, 756, 143553.
Laney, A. S., Petsonk, E. L., and Attfield, M. D. 2010. Pneumoconiosis among underground bituminous
coal miners in the United States: Is silicosis becoming more frequent? Occupational and Environmental
Medicine 67(10):652–656.
Leung, C. C., Yu, I. T. S., & Chen, W. (2008). Silicosis. The Lancet, 379(9830), 2008–2018.
Mahdavi, A., & Siegel, J. A. (2020). Extraction of dust collected in HVAC filters for quantitative filter
forensics. Aerosol Science and Technology, 54(11), 1282–1292.
Miller, A. L., Drake, P. L., Murphy, N. C., Noll, J. D. & Volkwein, J. C. (2012) Evaluating portable
infrared spectrometers for measuring the silica content of coal dust. J. Environ. Monit., 14, 48–55.
Miller, A. L., Weakley, A. T., Griffiths, P. R., Cauda, E. G. & Bayman, S. (2017) Direct-on-Filter α-
Quartz Estimation in Respirable Coal Mine Dust Using Transmission Fourier Transform Infrared
Spectrometry. Applied Spectroscopy, 71(5), 1014–1024.
MSHA, 2008. Method No. MSHA P7: Infrared Determination of Quartz in Respirable Coal Mine Dust.
United States Department of Labor-MSHA-Pittsburgh Safety and Health Technology Center.
NASEM (2018). Monitoring and Sampling Approaches to Assess Underground Coal Mine Dust Expos
ures, Washington, DC, The National Academies Press.
National Institute for Occupational Safety and Health (NIOSH): QUARTZ in coal mine dust, by IR
(NIOSH 7603, NIOSH Manual of Analytical Methods 4th ed.). Cincinnati, OH: NIOSH, 2003.
Oberdörster, G., Oberdörster, E., & Oberdörster, J. (2005). Nanotoxicology: an emerging discipline
evolving from studies of ultrafine particles. Environmental Health Perspectives, 113(7), 823–839.
Pan, L., Golden, S., Assemi, S., Sime, M. F., Wang, X., Gao, Y., & Miller, J. (2021). Characterization of
particle size and composition of respirable coal mine dust. Minerals, 11(3), 276.
Pokhrel, N., Keles, C., Jaramillo, L., Agioutanti, E., & Sarver, E. (2021a). Direct-on-Filter FTIR Spec
troscopy to Estimate Calcite as A Proxy for Limestone ‘Rock Dust’in Respirable Coal Mine Dust
Samples. Minerals, 11(9), 922.
Pokhrel, N., Agioutanti, E., Keleş, Ç., Afrouz, S., & Sarver, E. A. (2021b). Comparison of mineral con
tent in respirable coal mine dust samples estimated using FTIR, TGA, and SEM-EDX. In Mine Venti
lation (pp. 255–267). CRC Press.
237
Ramgolam, K., Favez, O., Cachier, H., Gaudichet, A., Marano, F., Martinon, L., & Baeza-Squiban, A.
(2009). Size-partitioning of an urban aerosol to identify particle determinants involved in the proin
flammatory response induced in airway epithelial cells. Particle and Fibre Toxicology, 6(1), 1–12.
Roper, C., Delgado, L. S., Barrett, D., Massey Simonich, S. L., & Tanguay, R. L. (2018). PM2. 5 filter
extraction methods: implications for chemical and toxicological analyses. Environmental science &
technology, 53(1), 434–442.
Ross, M. H., & Murray, J. (2004). Occupational respiratory disease in mining. Occupational medicine, 54
(5), 304–310.
Salinas, V., Das, M., Jacquez, Q., Camacho, A., Zychowski, K., Hovingh, M., Medina, A.,
Rubasinghege, G., Rezaee, M., Baltrus. J., Fairley, N., & Roghanchi, P. (2022). Characterization and
Toxicity Analysis of Lab-Created Respirable Coal Mine Dust from the Appalachians and Rocky
Mountains Regions. Minerals, 12(7), 898.
Sarver, E., Keles, C., & Rezaee, M. (2019). Characteristics of respirable dust in eight appalachian coal
mines: A dataset including particle size and mineralogy distributions, and metal and trace element
mass concentrations. Data in brief, 25, 104032.
Sarver, E., Keleş, Ç., & Afrouz, S. G. (2021). Particle size and mineralogy distributions in respirable dust
samples from 25 US underground coal mines. International Journal of Coal Geology, 247, 103851.
Shekarian, Y., Rahimi, E., Rezaee, M., Su, W. C., & Roghanchi, P. (2021). Respirable coal mine dust:
a review of respiratory deposition, regulations, and characterization. Minerals, 11(7), 696.
Sioutas, C., Delfino, R. J., & Singh, M. (2005). Exposure assessment for atmospheric ultrafine particles
(UFPs) and implications in epidemiologic research. Environmental health perspectives, 113(8), 947–955.
Stanek, L. W., Sacks, J. D., Dutton, S. J., & Dubois, J. J. B. (2011). Attributing health effects to appor
tioned components and sources of particulate matter: an evaluation of collective results. Atmospheric
Environment, 45(32), 5655–5663.
Stanley, N. J., T. H. Kuehn, S. W. Kim, P. C. Raynor, S. Anantharaman, M. A. Ramakrishnan, and
S. M. Goyal. 2008. Background culturable bacteria aerosol in two large public buildings using HVAC
filters as long term, passive, high-volume air samplers. J. Environ. Monit. 10 (4): 474–481.
Willis, R.D. & Conner, T.L. Guidelines for the Application of SEM/EDX Analytical Techniques to Par
ticulate Matter Samples; US EPA: Washington, DC, USA, 2002.
238
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Dust control by a horizontal air-blocking shelf inside a drill shroud was effect
ive in a previous study by researchers at the National Institute for Occupational Safety and
Health (NIOSH). However, when a medium-sized drilling machine lowers its mast, material
deposited on the shelf can produce a significant dust cloud. To minimize this dust exposure
hazard, researchers applied a 208-L (55-gallon) drum-sized ring, installing it concentrically
with the drill stem under the deck table to act as a vertical air-blocking ring to control the drill
dust leakage. Simultaneously, gravity force was utilized to minimize material buildup. Com
puter simulation was used in the evaluation. Three groups of ring heights were studied: short
[5.1-cm (2-in), 7.6-cm (3-in), and 10.2-cm (4-in)], medium [15.2-cm (6-in), 30.5-cm (12-in), and
45.7-cm (18-in)], and long [61.0-cm (24-in) and 76.2-cm (30-in)]. The dust leakage was simu
lated using computational fluid dynamics (CFD) under the commonly encountered field con
ditions of a 2:1 dust-collector-to-bailing airflow ratio and a 5.1-cm (2-in) shroud-to-ground
gap. Based on this simulation study, the vertical air-blocking ring with medium heights has
the potential to effectively confine the dust inside the drill shroud. Other shorter or longer
height rings may not be as effective in preventing dust leakage from the shroud. The results of
this study can be used to guide future laboratory tests.
1 INTRODUCTION
Drilling operations are common sources of respirable dust generation and liberation at surface
mines. These dust emissions can cause high dust exposures for the drill operator and nearby
mineworkers downwind of the drill if windy conditions are present. When drilling through
silica-bearing materials, such as sandstone and shale, mineworkers can be exposed to high
concentrations of respirable silica dust. Silica dust overexposures can lead to silicosis, an occu
pational lung disease that has no cure and is often fatal [Lara, 2020].
To reduce or eliminate the likelihood of developing silicosis, many rotary drilling operations
use a dry collection system [Organiscak and Page, 2005]. Figure 1 shows a typical dry dust
collection system schematic for medium and large drills. Compressed bailing air is forced
through the inside of the hollow drill stem and released at the end of the drill bit to flush the
drill cuttings to the surface. On the surface, the bailing air with cuttings are contained by the
drill deck and deck shroud. The larger cuttings are deposited inside the shroud area while fine
particles are pulled into a dust collection system, where they are trapped in filters prior to the
air being exhausted back into the environment through the fan. To prolong filter life, the fil
ters are back-flushed regularly with compressed air to remove excess particulates, typically
known as the dust filter cake. This dust falls out into the collector chamber and then ultim
ately drops out of the bottom of the collector onto the mining bench. More detailed descrip
tions of this operation can be found in studies conducted by the U.S. Bureau of Mines and the
NIOSH [Maksimovic and Page, 1985; Organiscak and Page, 1995; Reed et al., 2008].
DOI: 10.1201/9781003429241-25
239
Figure 1. A basic dry dust collection system on a drill.
Previous laboratory tests and CFD studies have demonstrated that under a 5.1-cm (2-in)
deck shroud-to-ground gap condition, if the dust collector-to-bailing airflow ratios are 3:1,
the dust leakage from the drill shroud will be greatly reduced. If the ratios can be increased to
4:1, the respirable dust can be well-confined within the drill shroud. However, this higher ratio
is difficult to attain at mining sites. The ratio encountered during past NIOSH studies of oper
ating drills averaged about 2:1 [Zheng et al., 2016].
To better confine the dust cloud inside the drill shroud, the airflow pattern was further
investigated. The Coanda effect was observed [Potts and Reed, 2008]. Two common phenom
ena of the Coanda effect are: first, the tendency of an air jet approaching a curved surface to
remain attached to the surface even after the surface curves away from the initial jet direction;
and second, the tendency of an air jet to adhere to a nearby surface [Trancossi, 2011]. Within
a drill shroud, the airflow tends to follow the inner surfaces of the drill shroud and leaks out
at any shroud-to-ground gap. An air-blocking shelf was configured to reduce the Coanda
effect and was shown to be effective in both laboratory and field test scenarios [Potts and
Reed, 2008; 2011]. CFD simulations also evaluated the air-blocking shelf and estimated that
a dust-collector-to-bailing airflow ratio as low as 1.75:1 when using the air-blocking shelf may
effectively confine dust inside the drill shroud [Zheng et al., 2018].
During field studies, a considerable amount of debris was deposited on the air-blocking
shelf during the drilling process (Figure 2), and these deposits did not appear to affect the dust
control effects of the air-blocking shelf. However, when the drill mast was lowered to move
from one location to another, the buildup of materials on the air-blocking shelf liberated
a dust cloud when the material fell off the shelf, as shown in Figure 3.
To eliminate debris on top of the air-blocking shelf, NIOSH researchers proposed two types
of inner-shaped drill shrouds, U-shaped and V-shaped, along with a drill shroud with air-
blocking shelves installed at a 45° angle from the horizontal direction. Those three designs
provide a sloped shelf surface with an angle greater than the dust’s angle of repose. This
allows gravity to overcome the material’s internal friction to prevent material accumulations.
The research results revealed that the U-shaped drill shroud and the 45° air-blocking shelves
can potentially confine the dust inside the drill shroud [Zheng et al., 2021].
This study uses a vertical air-blocking ring to eliminate debris accumulation. To simplify
the application, commercially available 208-L (55-gallon) drums (Figure 4) can be utilized
accordingly and cut into different heights. The resulting drum section can then be installed
concentrically with the drill stem under the drill deck to act as a vertical air-blocking ring.
A total of eight cases were studied using CFD analysis with different heights of rings cut from
the commercially available 208-L (55-gallon) drum [0.58-m (23-in) in diameter and 0.86-m
(34-in) in total height]: 5.1-cm (2-in), 7.6-cm (3-in), 10.2-cm (4-in), 15.2-cm (6-in), 30.5-cm (12-
in), 45.7-cm (18-in), 61.0-cm (24-in), and 76.2-cm (30-in). The dust leakage was evaluated
under commonly encountered field conditions of a 2:1 dust-collector-to-bailing airflow ratio
and a 5.1-cm (2-in) shroud-to-ground gap. This current study, along with previously validated
CFD studies, was adopted to evaluate whether the dust-laden airflow can still be effectively
confined inside the drill shroud with minimal leakage when using the vertical air-blocking
ring.
240
Figure 2. The buildup of drill cuttings on the Figure 3. Dust generated during the lowering of the
air-blocking shelves. drill mast.
Figure 4. 208-L (55-gallon) drums manufactured with different materials with 0.58-m (23-in) in diam
eter and 0.86-m (34-in) in total height [EBK, 2022].
2 CFD MODELING
The ANSYS Fluent Version 18.2 program (ANSYS, Canonsburg, PA) was used to perform
the analysis of dust distribution within the drill shroud and surrounding space. The schematic
of the airflow domain inside the drill table simulator was built according to the geometry
measured from NIOSH’s full-sized test facility. Figure 5 (a) shows the size of the dust cham
ber: 3.66-m wide by 3.05-m deep by 2.44-m high (12-ft by 10-ft by 8-ft). The shroud, located
in the center of the dust chamber, measures 1.52-m wide by 1.22-m deep by 1.22-m high (5-ft
by 4-ft by 4-ft). More detailed setup of the full-scale facility can be found in previous literature
[Organiscak and Page, 2005; Potts and Reed, 2008, 2011].
A schematic of the computational domain is shown in Figure 5. Only two cases out of the
eight geometric models are shown here. In Figure 5 (a), the segment of the drum is a 5.1-cm
(2-in) height ring under the bottom of the drill deck with a common drum diameter of
58.4-cm (23-in). In Figure 5 (b), the height of the ring shown is 45.7-cm (18-in).
The locations of the boundary conditions for CFD are illustrated in Figure 5 (b). The
boundary conditions applied in the simulation include roof inlet, bailing air inlet, and dust
collector outlet. Intake airflow is pulled into the simulation domain through the three open
ings on the roof, shown as “Roof inlet” in Figure 5 (b). At the same time, the bailing air with
dust is injected into the simulation domain from a circular face inside the drill pipe, as indi
cated by the bailing air inlet. The dust collector pulls air from inside the drill shroud and dis
charges it at the dust collector outlet.
Previously validated CFD models were used in this study to evaluate the effect of these ver
tical air-blocking rings of the drill shroud on dust control [Zheng et al., 2016; 2018]. Dust was
treated as a gas (CO2), and a species transport model in ANSYS Fluent was used. The bound
ary conditions used to determine the dust distribution inside the domain can be referred to in
Zheng [2016]. The airflow patterns and dust control capabilities were simulated for the eight
241
Figure 5. Overview of the CFD model with boundary conditions for (a) 5.1-cm (2-in) vertical
air-blocking ring; (b) 45.7-cm (18-in) vertical air-blocking ring.
configurations and evaluated under the same ventilation conditions: 0.24 m3/s (500 cfm) of
bailing airflow, 0.24 m3/s (500 cfm) of intake roof airflow, and 0.48 m3/s (1,000 cfm) of dust
air mixture collected by the dust collector outlet, which equates to a 2:1 collector-to-bailing
airflow ratio.
The vertical air-blocking ring has a 90° slope angle in all eight cases, which is much larger
than the angles of repose of minerals commonly encountered during drilling, which range
from 0° to 50° [Dougherty and Schissler, 2020]. Under the vertical slope angle, the materials
on the ring’s inner walls cannot stay in place and will fall to the ground during the drilling
process. The inner surface of the ring should be smooth enough to not provide any rough sur
faces that allows any deposition of small dust particles. In addition, the material of the drum
should not offer any electrostatic force to attach small particles. Based on the simulation
results of the eight cases, the phenomena and dust control effects can be separated into three
groups and are described in the following sections:
3.1 Effect of the short vertical air-blocking ring from 5.1-cm (2-in) to 10.2-cm (4-in)
Similar airflow patterns are observed for the vertical air-blocking ring with 5.1-cm (2-in),
7.6-cm (3-in), and 10.2-cm (4-in) height cases. Overall, the ring can divert airflow toward the
ground, but cannot confine the dusty air inside the drill shroud.
Figure 6 shows that after the dust-laden air is released into the bailing air inlet, the air
travels down inside the hollow drill stem, then reverses direction up the gap between the drill
steel and drill hole. As the air flows up, it attaches to the outside surface of the drill steel and
encounters the underside of the drill deck, where it fans out in all directions and flows to
where the vertical air-blocking ring is located.
After the flow encounters the vertical air-blocking ring, the CFD simulation shows that the
flow is blocked by the ring and is diverted downward, as shown in Figure 6 (b). Then, on the
left rear corner of the drill shroud where the inlet of the dust collector is located, the dusty
bailing air is exhausted into the dust collector pipe after the ring. On the right side of the drill
shroud, the bailing air encounters additional air that is drawn into the drill shroud through
the 5.1-cm (2-in) shroud-to-ground gap. The combined air is then recirculated counterclock
wise toward the upper region of the drill shroud and flows toward the dust collector inlet.
Some bailing airflow at the front right corner of the drill shroud, after being deflected from
the vertical ring, flows outward and is attached to the vertical deck shroud. It then flows
down to the shroud-to-ground gap. Part of this airflow can leak out of the drill shroud after
striking the ground.
242
The airflow before encountering the vertical air-blocking ring shows a Coanda effect. For the
three cases with 5.1-cm (2-in), 7.6-cm (3-in), and 10.2-cm (4-in) ring heights, the diverted down
ward flow can still flow outward and attach to the vertical deck shroud wall. This is also
a Coanda effect and can cause dust leakage. It can be observed in Figure 6 that airflow path
lines are shown outside of the shroud and in Figure 7 that dust can leak out of the shroud and
potentially expose any workers around the drill to respirable dust. The amount of dust leakage
can result in potential relative dust concentrations ranging from 1.5 to 15.5 mg/m3. In addition,
the dust leakage occurs on the opposite side of the collector inlet, where the influence of the
dust collector is relatively weak. This side is generally where the drill operator’s cab is situated.
Figure 6. The pathlines of bailing airflow colored by velocity magnitude (0.0–10.0 m/s), 10.2-cm (4-in)
height vertical air-blocking ring: (a) fully developed pathlines; (b) bailing air as it passes by the ring.
Figure 7. Respirable dust concentration distributions under 10.2-cm (4-in) height vertical air-blocking
ring; legend shows the dust levels (1.5–36.5 mg/m3).
3.2 Effect of the medium vertical air-blocking ring from 15.2-cm (6-in) to 45.7-cm (18-in)
For the cases with 15.2-cm (6-in), 30.5-cm (12-in), and 45.7-cm (18-in) height rings, the
increased ring height can confine dusty bailing air more effectively at the center region inside
the drill shroud as shown in Figure 8. The dust-laden air travels down the inside of the hollow
drill stem, then goes up along the outside surface of the drill steel and fans out in all directions
on the underside of the drill deck before the vertical air-blocking ring, similarly to the airflows
depicted in Figure 6 (b).
As the airflow comes out below the ring, there are roughly three groups of flow: 1) the flow
that goes directly toward the dust collector inlet on the left rear corner of the drill shroud; 2)
the flow on the right side of the drill shroud, where the bailing air flows counterclockwise
243
toward the upper region of the drill shroud and then flows toward the dust collector inlet
through the upper front and upper rear regions outside of the ring; and 3) the flow that comes
down at the front and right corner of the drill shroud which generally does not reach the
ground, then flows toward the dust collector inlet. Compared to the flow in Figure 6, the dif
ference is the third group of bailing airflow which is diverted by the vertical ring closer to the
center of the drill shroud without attachment to the vertical drill shroud wall and therefore
has a smaller chance to leak out of the shroud. This can also be observed in Figure 9 which
demonstrates no respirable dust outside of the shroud.
Figure 8. The pathlines of bailing airflow col Figure 9. Respirable dust concentration distributions
ored by velocity magnitude (0.0–10.0 m/s), under 30.5-cm (12-in) height vertical air-blocking ring;
15.2-cm (6-in) height vertical air-blocking ring. the legend shows the dust levels (1.5–36.5 mg/m3).
3.3 Effect of the long vertical air-blocking ring from 61.0-cm (24-in) to 76.2-cm (30-in)
As the vertical air-blocking ring increases to 61.0-cm (24-in) and 76.2-cm (30-in), the airflow
pattern changes. There is a Coanda effect that shows up on the ground, which can be
a problem, as revealed by this study.
The results from the CFD simulation in Figure 10 show that the pathlines of bailing airflow
have the same flow patterns before they leave the bottom edge of the ring. At this point, the
airflow remains directed downward, and the airflow velocity can be in the range of 4.0–5.0 m/
s (787.4–984.3 fpm). Because the distance from the ground to the bottom of the ring is
52.8-cm (20.8-in) for the 61.0-cm (24-in) height ring and 42.7-cm (16.8-in) for the 76.2-cm (30-
in) height ring, the high-velocity downward flow can reach the ground.
Once the airflow reaches the ground, it fans out horizontally in all directions (Coanda effect)
toward the shroud-to-ground gap where it encounters the inflow from outside of the drill
shroud. Although there is incoming flow in the opposite direction, dusty bailing air can still leak
Figure 10. The pathlines of bailing airflow Figure 11. Respirable dust concentration distributions
colored by velocity magnitude (0.0–10.0 m/s), under 76.2-cm (30-in) height vertical air-blocking ring;
76.2-cm (30-in) height vertical air-blocking ring. legend shows the dust levels (1.5–36.5 mg/m3).
244
out of the drill shroud in regions where the inflow is weak. As shown in Figure 11, these regions
are located at the front right corner of the drill shroud, opposite of the dust collector inlet.
4 DISCUSSIONS
It was observed from the eight CFD simulations in this study that when the vertical air-
blocking ring is short (not higher than 10.2-cm or 4-in), it can divert the airflow away from the
bottom of the drill deck. However, it cannot alter the airflow direction completely from horizon
tal to vertical. The angled downward bailing airflow can still reattach to the vertical deck
shroud walls and flow to the shroud-to-ground gap. Leaking can occur at the front and right
corner where the influence of the dust collector is weak. This location is where the drill oper
ator’s cab is situated. Therefore, it is important for the drill operator’s cab to have a properly
designed and maintained enclosed cab filtration and pressurization system to prevent the drill
operator from respirable dust exposure during drilling [Organiscak et al., 2018].
When the height of the vertical air-blocking ring increases to the medium range, from
15.2-cm (6-in) to 45.7-cm (18-in), most of the bailing airflow is confined in the central region
of the drill shroud. Very little airflow reattaches to the vertical deck shroud walls contrary to
the short vertical air-blocking ring cases (as shown in Section 3.1). There may be some airflow
that reaches the ground as in longer vertical air-blocking ring cases (as shown in Section 3.3).
However, most of the airflow remains in the center region of the drill shroud and can be con
fined effectively by the inflow of air from outside the drill shroud.
In the long range, once the heights of the vertical air-blocking ring are increased to 61.0-cm
(24-in) and above, the bailing airflow is redirected vertically down to the ground and fans out
in all directions due to the Coanda effect. As a result of the relatively high speed of the flow
from the vertical ring, the outflow fanning at the ground surface can break through the
shroud-to-ground gap, overwhelm incoming airflow from the opposite direction and increase
the dust exposure to the surrounding environment.
For medium and large drill equipment with similar drill shroud size, this CFD simulation
provides a range of vertical air-blocking ring heights that may be effective for dust control.
However, it should be noted that as the drill cuttings accumulate on the ground inside the drill
shroud (as shown in Figures 1 and 2), the distance between the growing cuttings and the
bottom of the ring can dramatically decrease. The application of the vertical air-blocking ring
may need to be adjusted accordingly based on the field conditions. For example, if the cuttings
gradually accumulated to 25.4-cm (10-in) above the ground, the ring heights from 15.2-cm
(6-in) to 45.7-cm (18-in) that were shown to be effective in this simulation may be required to
be changed to between 15.2-cm (6-in) and 20.3-cm (8-in).
5 CONCLUSIONS
It has been previously demonstrated by NIOSH researchers that horizontally installed air-
blocking shelves can effectively control the respirable dust inside the drill shroud during the
drilling process. These shelf designs have no impact on drill operations or present any add
itional maintenance requirements. Once installed, the device is invisible, and drilling can con
tinue as before with no change in productivity. However, the deposits on these shelves can
create dust plumes when the machine changes drilling locations. In this study, various heights
of vertical air-blocking rings were investigated to utilize gravity force to remove the deposit
during drilling operations, preventing the buildup of material.
To simplify the application and minimize cost, the dimension of a commercially available
208-L (55-gallon) drum was built in the simulation and cut into a specific height of ring. The
ring was then put concentrically with the drill stem under the deck table to act as a vertical
air-blocking ring. The dust control capacities of eight rings’ heights were evaluated using
a previously validated CFD model.
Based on this study, the vertical air-blocking ring, categorized as medium, with a height
between 15.2-cm (6-in) and 45.7-cm (18-in), has the potential to effectively confine the dust
245
inside the drill shroud. The cases with shorter heights between 5.1-cm (2-in) and 10.2-cm (4-in)
and longer heights above 61.0-cm (24-in) may have dust-leaking problems at regions where
the dust collector’s influence is weak.
The results of this study can be used to guide future laboratory tests. The vertical air-blocking
ring combined with a different inner-shaped drill shroud design [Zheng et al., 2021] can provide
a solution to dust problems while drilling and when the surface drill location is changed.
ACKNOWLEDGEMENTS
One of the authors of this paper, J. Drew Potts, designed the vertical air-blocking ring in this
study. The authors of this paper would like to sincerely acknowledge Andy B. Cecala, Carin
L. Kosmoski, Liming Yuan, Lihong Zhou, Tim W. Beck, Hua Jiang, and Scott S. Klima for
their technical support.
DISCLAIMERS
The findings and conclusions in this manuscript are those of the authors and do not necessar
ily represent the official position of the National Institute for Occupational Safety and Health
(NIOSH). Mention of company names or products does not constitute an endorsement by
NIOSH.
REFERENCES
Dougherty, HN, and Schissler, AP. 2020. SME mining reference handbook. Englewood, Colorado: Soci
ety for Mining, Metallurgy & Exploration (SME).ISBN #: 978–0–87335–435–6.
EBK 2022. 55 Gallon Drums. EBK containers, https://ebkcontainers.com/drums-ibc-totes/drums/55-
gallon-drums/, accessed Oct 28, 2022.
Lara AR. 2020. Silicosis. The Merck Manual, Consumer Version. Kenilworth NJ: Merck & Co. Inc.
Maksimovic SD and Page SJ 1985. Quartz dust sources during overburden drilling at surface coal mines.
Department of the Interior, U.S. Bureau of Mines, Information Circular 9056.
Organiscak JA and Page SJ 1995. Assessment of airborne dust generated from small truck-mounted rock
drills. Department of the Interior, U.S. Bureau of Mines, Report of Investigations 9616.
Organiscak JA and Page SJ 2005. Development of a dust collector inlet hood for enhanced surface mine
drill dust capture. Int J Surf Min Reclamat. Environ 19(1): 12–28.
Organiscak JA, Cecala AB, and Hall RM 2018. Design, testing, and modeling of environmental enclos
ures for controlling worker exposure to airborne contaminants. Pittsburgh PA: U.S. Department of
Health and Human Services, Centers for Disease Control and Prevention, National Institute for Occu
pational Safety and Health, DHHS (NIOSH) Publication No. 2018–123, IC 9531. https://doi.org/
10.26616/NIOSHPUB2018123.
Potts JD and Reed WR 2008. Horizontal air-blocking shelf reduces dust leakage from surface drill
shroud. Transactions of the Society of Mining, Metallurgy, and Exploration 324: 55–60.
Potts JD, and Reed WR 2011. Field evaluation of air-blocking shelf for dust control on blasthole drills.
Int J Surf Min Reclamat. Environ 25(1): 32–40.
Reed WR, Listak J, Page S and Organiscak J 2008. Summary of NIOSH research completed on dust
control methods for surface and underground drilling. Transactions of the Society of Mining, Metal
lurgy, and Exploration 324: 32–40.
Trancossi M 2011. An Overview of Scientific and Technical Literature on Coanda Effect Applied to
Nozzles. SAE Technical Paper: 2011-01-2591.
Zheng Y, Reed WR, Zhou L, and Rider JP 2016. Computational fluid dynamics modeling of a medium-sized
surface mine blasthole drill shroud. Mining Engineering 68(11): 43–49.
Zheng Y, Reed WR, Potts JD, Li M, and Rider JP 2018. Dust control by air-blocking shelves and dust
collector-to-bailing airflow ratios for a surface mine drill shroud. Mining Engineering 70(5): 69–74.
Zheng Y, Reed WR, and Potts JD 2021. Different inner shapes of drill shroud on dust control for surface
mine drilling operation. Tukkaraja, P. (Ed.). 2021. Mine Ventilation: Proceedings of the 18th North
American Mine Ventilation Symposium, 12–17 June 2021, Rapid City, South Dakota, USA (1st ed.).
CRC Press. https://doi.org/10.1201/9781003188476.
246
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Dust control challenges exist in numerous mining applications and is espe
cially prevalent in confined production spaces like mining faces. Water sprays are the most
economical and technically feasible means of reducing dust concentrations in a wide range of
applications. This study investigated the respirable dust knockdown performance and submi
cron particle suppression behavior for a typical mining spray nozzle operating at pressures
ranging from 100 to 800 psi. A confined chamber dust removal evaluation approach and an
optical particle counter were used for this study. For respirable dust knockdown, high spray
pressures were more effective in lowering dust concentrations. However, submicron-sized par
ticles are difficult to remove; at 100-psi water pressure, the knockdown fraction for submicron
particles remained at only 40 % at the end of a 1-hr test. The lowest suppression effective size
boundary was identified from 0.465 to 0.897 µm under a tested pressure range. Resuspension
of submicron material was observed at all water pressure levels after the start of the spray
operation. The resuspended submicron dust remained airborne for different durations under
different spray pressures. Overall, results show a low capture efficiency and a high resuspen
sion possibility for micron- or nano-scale particles under testing conditions.
1 INTRODUCTION
Dust control problems exist in a variety of industrial sectors, from construction sites to mining
sites (Stacey et al., 2011). It is even more challenging when dealing with respirable dust in
a confined production space, such as the underground mining face. Mining activities can gener
ate a large amount of dust particles including respirable dust and even submicron dusts (Sarver
et al. 2019). Despite the Mine Safety and Health Administration (MSHA) setting regulations to
improve coal miner health and safety, there has been a dramatic increase in the prevalence of
progressive massive fibrosis among underground coal miners in central Appalachia over the
past decade (Doney et al., 2019). It has long been recognized that the inhalation and deposition
of respirable dust in the lung respiratory tract is the main cause for pneumoconiosis and silicosis
in coal miners (Cecala et al., 2019). The human respiratory tract can be divided into three main
regions—the head, tracheobronchial region, and the gas-exchange or alveolar region (Brown
et al., 2013). Respirable fraction specifically means the mass fraction of inhaled particles pene
trating into the gas-exchange region. In terms of particle size, the limit for entering the alveolar
region is between 10 and 15 µm (WHO, 1999; European Committee for Standardization, 2014).
These respirable fractions that cause the greatest hazard include quartz and other dusts contain
ing free crystalline silica produced by grinding or drilling.
Dust mitigation techniques commonly used in the underground mining workplace generally
include three distinct types: (1) ventilation systems to dilute and transport dust away from
workers, (2) isolation of workers by enclosures, and (3) capture of airborne dust using water
sprays. Among these techniques, water sprays are the most economical and technically feasible
DOI: 10.1201/9781003429241-26
247
means of reducing dust concentrations in a wide range of applications (Beck et al., 2018). Sev
eral water spray applications for controlling longwall face dust have been developed and
adopted in recent years. Drum-mounted, full-cone water sprays are very effective for dust sup
pression at the point of coal fracture and minimizing dust liberation during coal transport.
Shearer-body-mounted water sprays oriented downwind can promote movement of dust-
laden air along the face side of the shearer to prevent migration into the walkway. There are
also water sprays mounted on the longwall shield to address the dust generated during shield
movement. Underside shield canopy sprays can create a moving water curtain to contain the
dust cloud near the drum area and away from the walkway (Colinet et al. 2010). The atomiza
tion characteristics and dust removal efficiency of high-pressure sprays were investigated for
optimizing the performance. Cheng et al. (2011) conducted a field test showing that increasing
the longwall shearer and shield spray pressure from 290 to 1,160 psi decreased the dust con
centration from 536 mg/m3 to 97 mg/m3. This research also recommended the application of
1,160-psi spray pressure for better respirable dust capture efficiency in coal mines. An experi
mental study concluded that the spray pressure is directly proportional to spray water flow,
while inversely proportional to the water droplet size and spray angle (Han et al., 2020). The
overall dust suppression efficiency for total dust and respirable dust increased with the
increase of pressure from 290 to 1,160 psi.
Dust capture by water droplets can be accomplished by numerous mechanisms including
inertial impaction, interception, diffusion, thermophoresis, and electrostatic charge collection.
These processes usually occur simultaneously, although inertial impaction and interception
are predominant for the capture of coal-mine-generated dusts. The theoretical foundation for
high-pressure suppression performance improvements can be found from the droplet breakup
mode (Majithia et al., 2008), dust suppression efficiency by inertial impaction (Walton and
Woolcock, 1960), and dust suppression efficiency by interception (Fuchs, 1964), illustrated in
equations 1, 2, and 3 respectively:
where
Dw is the diameter of water droplet, m; Wec is the critical Weber number when the droplet
starts to breakup; ρg is the air density, kg/m3; ν is the relative velocity between the dust par
ticles and the droplets, m/s; σ is the surface tension, N/m; ηimp is the dust suppression efficiency
by inertial impaction; St is Stock number of particle; ηint is the dust suppression efficiency by
interception; Dp is the diameter of dust particle, m; Cf denotes the Cunningham correction
factor; ρp is the density of the dust particles, kg/m3; μ denotes the air dynamic viscosity, Pa·s.
As the spray pressure increases, the water flow rate increases and thereby results in higher
droplet velocity. According to Eq. (1) a higher relative velocity produces smaller water drop
lets. Eq. (2) and (4) show that a smaller water droplet and larger relative velocity can increase
the dust capture efficiency by inertial impaction. Eq. (3) demonstrates that the interception
efficiency for small dust particulates can be limited but reducing the size of water droplets can
increase the dust interception efficiency. In summary, high-pressure spray generates finer
water droplets and can yield a higher dust suppression efficiency.
248
There is still skepticism of the water suppression effect for respirable particulates by
researchers (Organiscak and Page 2005). The detailed size-specific behavior within the respir
able range under suppression was rarely studied. The spray knockdown efficiency for submi
cron particles was unknown for either low- or high-pressure spray in coal mining. In this
study, the respirable dust suppression characteristics were investigated in a confined chamber
under spray pressure that ranged from 100 to 800 psi. The spray respirable dust knockdown
was monitored and evaluated over a 1-hour testing period. The suppression behavior for sub
micron particles under different pressures were also investigated.
Figure 1. Dust chamber system for dust deposition and water spray dust suppression test.
This study is only interested in the respirable particles, over-sized particles in the raw coal
dust were removed by cyclone during the sampling process. In this test, three real-time aerosol
monitors (TSI Dust Trak@ II aerosol monitor model 8530, Shoreview, MN) were connected
to Dorr-Oliver 10-mm nylon cyclones for the purpose of measuring the respirable size fraction
of dust (≤ 10 µm). Sampling points were located 10 inches above the spray. During testing,
one aerosol monitor continuously sampled and recorded the dust concentration, while the
249
other two monitors began sampling and recording when dust concentration reached the start
criteria. The sampling flow rate for each monitor was set to 1.7 L/min, and the data logging
interval was 5 seconds. All reported dust concentrations are the relative mass concentration
measured by an aerosol monitor instrument unless expressed otherwise. One Optical Particle
Sizer (OPS) (TSI model 3330, Shoreview, MN) was used to obtain the particle count results
with a sampling rate of 1 L/min. The OPS operates on the principle of single particle counting.
Particle pulses are counted individually and binned into the 16 channels based on their pulse
heights. Sixteen (16) bin ranges were determined with a refractive index of 1.245 to ensure the
accuracy of the particle count result. A 10:1 aerosol diluter (TSI model 3332, Shoreview, MN)
was installed before the inlet on the OPS to ensure the accuracy of the measurement. It should
be noted that limited water-only testing was conducted to assess the impact of spray water
droplets on OPS particle counting and that the spray droplet had no impact.
Before each test, the chamber was sealed and the spray was operated to wet the interior
walls. Dust was injected into the chamber until the instantaneous concentration reached at
least 100 mg/m3. The spray was tested at designated water pressures and testing was started
once the dust concentration naturally decayed to approximately 100 mg/m3. Water pressures
of 100, 200, 400, 600, and 800 psi were evaluated. At the same time, the aerosol monitors and
particle counter were started and operated continuously throughout the whole test period.
Each test continued for 60 minutes or until the dust concentration reached below 1.0 mg/m3.
A randomized full factorial experimental design with three repetitions for each test condition
was performed.
The water flow rates and spray angles under different pressures operated in the chamber were
measured and are listed in Table 1. It is obvious as the water pressure increases, the water
volume sprayed out from the nozzle increased, but the spray angle narrowed down. With
a smaller spray angle, a smaller spray coverage (area or volume) in the chamber can be
expected as well. As can be observed from the spray pattern, the plume bottom converged
considerably starting from 400 psi. This convergence started closer to the nozzle tip under
higher pressures. This divergence effect can reduce the actual spray coverage from the spray
angle based on theoretical coverage calculations.
Figure 3 shows the respirable dust concentration (solid line) and cumulative water consump
tion (dashed line) results for tests under different pressures over the testing period. The dust
concentration dropped sharply for all pressures within the first 10 minutes, then it slowly
decreased to the ending concentration. Under spray pressure 100 and 200 psi, the final dust
concentration reached 8.70 and 2.77 mg/m3, respectively. While under higher pressure, the
final dust concentrations were able to reach the 1.0 mg/m3 stop criteria, and the time it takes
to reach this criterion for tests with spray pressures of 400, 600, and 800 psi are 30.8, 13.7, and
8.3 minutes, respectively.
250
Table 1. Spray specifications and characteristics of WhirlJet® BDM3 hollow cone spray under different
water pressure.
These results were anticipated because more water was sprayed into the chamber under
higher spray pressure. Evidence can be found from the dashed lines in Figure 3 that the cumu
lative water consumption for higher pressure is always higher than lower pressure tests. How
ever, since high-pressure spray is more effective on capturing dust, the total water
consumption to realize the same dust reduction for high pressure is much smaller. For
example, to realize 90 % dust reduction, the total water consumption under 100, 200, 400, 600,
and 800-psi spray pressures is 41.7, 20.1, 10.2, 7.4, and 5.9 gallon, respectively.
A previous study (Jiang et al., 2022) addressed the issue of low capture efficiency for submi
cron particulates by water sprays. The results revealed that under 100-psi spray pressure, par
ticles under 1 µm can remain airborne through the 60-minute test, and particles under 0.5 µm
can be considered irremovable particles under the same condition. Therefore, this section
251
analyzes the submicron particulate suppression performance under different spray pressures.
Since size channels for the particle counter were calculated with a refractive index, the cap size
was not rounded at 1 micron. It should also be noted that each size bin is represented by its
top size range, which means bin 0.374 µm indicates a particulate size in the 0.3 to 0.374 µm
range.
Figure 4 plots the sub-1.117-µm particle counts during the whole testing process under dif
ferent spray pressures. A particle count peak can be found under each pressure within the first
10 minutes of spray. These peaks were likely caused by resuspension of the residual dust in the
chamber by the spray action. Based on the ending counts from different pressures, it appears
that submicrons can be suppressed by higher spray pressures in a relatively short period of
time. It took around 6, 10, and 24 minutes for spray pressure of 800, 600, and 400 psi to
reduce 80 % of submicrons from their peak. The final reduction percentage of submicrons
under 200 and 100 psi are 69.4 and 41.6 %, respectively. These results agree with findings from
other studies that the wettability of submicrons is lower than larger sizes (Yu et al., 2020).
Considering the short interaction period of spray and particulate in the underground face,
and it can be expected that most submicrons are likely to escape from low-pressure open
sprays and enter the mine atmosphere.
In addition to the sub-1.117-µm particle count results, the detailed particle count results for
7 size bins at different time stamps (shown by various colors) under different pressures are
plotted in Figures 5, 6, and 7. The y-axis is the dust counter result percentage at the time com
pared to its initial count.
Figure 5. Sub-1.117-particle count results under 100-psi spray water pressure (mean ± SD, n=3).
Resuspension of particles were observed after 3 minutes of spray under all pressure levels.
The observed size range for particle resuspension is narrower for higher pressure spray than
252
Figure 6. Sub-1.117-particle count results under 200-psi spray water pressure (mean ± SD, n=3).
lower pressure spray. At 100-psi spray pressure, the size range for resuspension is 0 to 0.721
µm, while the range under 800 psi is below 0.374 µm. These figures also show that when the
peak of particle resuspension is reached at different times after spraying begins, it takes longer
to reach the peak under low-pressure spray—at 100 psi it took almost 30 minutes to fully
resuspend the sub-0.3-µm particles. It should be noted that under a 100-psi and 200-psi spray
pressure condition, there is a substantial amount of dust remaining above the initial level after
a 60-minute test. The knockdown efficiency for that size range under these pressures is so low
that after 60-minute test, the ending particle count is still higher than when the spray begins.
Because the spray water droplet and dust interaction period is less than 3 minutes in the
production face, the size boundary for effective suppression can be identified from the dust
counter results. The lower boundary can be found when the counter results at 3 minutes in
that size bin are lower than the starting level. Using this criterion, the lowest effective suppres
sion size boundary for this spray under different pressures can be identified as 0.897 µm for
both 100 and 200 psi, 0.721 µm for 400 psi, and 0.465 µm for both 600 and 800 psi.
The above results indicate a low capture efficiency and high resuspension possibility for par
ticles under micron scale or nano-scale. In recent years, nano-scale dust particles and their
aggregates or agglomerates have obtained significant attention due to their potential influence
on the health of workers. Based on the analysis of coal dust samples collected from several
underground coal mines, the presence of Coal Dust NanoParticles (CD-NPs, smaller than 500
um) have been confirmed (Keles, et al., 2022). Studies have examined and compared the differ
ences in physicochemical properties, the acute pulmonary toxicity, and the chronic pulmonary
Figure 7. Sub-1.117-particle count results under 800-psi spray water pressure (mean ± SD, n=3).
253
fibrosis-inducing ability of Coal Dust Micron Particles (CD-MPs, smaller than 5 µm) and
CD-NPs as well as their adverse effects of coal dust size on the lung (Zhang, et al., 2022).
Findings indicate that CD-NPs are more chronically toxic than CD-MPs due to a lower
oxygen-to-carbon ratio, smaller size, and larger surface area based on characterization ana
lysis of particles. Other researchers also conclude that CD-NPs pose a higher risk to human
health and cause a larger surface area with a different toxicology compared to the same mass
concentration of dust of a relatively large size (Fan and Liu, 2021). Though it has been proven
that submicron particles are difficult to capture by water sprays, their flow path can be easily
manipulated by air stream. This means that these particles can be easily isolated from workers
by ventilation design. In this case, there will be advantages to using the air-moving properties
of the spray rather than the suppression ability.
6 CONCLUSION
Laboratory tests were conducted in a 512-ft3 dust chamber with the WhirlJet® BDM3 hollow
cone spray to study its respirable dust removal performance at five different water pressure
levels ranging from 100 to 800 psi. For each test, the respirable dust concentration and particle
count were continuously monitored during the 60-min spray application or when the dust con
centration dropped below 1.0 mg/m3. Results show that the total water consumption to
achieve the same dust reduction under high pressure is smaller than a low-pressure spray. The
suppression effectiveness of submicron particulates was also evaluated in this study. The final
reduction percentages of submicrons under 100 and 200 psi are 41.6 % and 69.4 %, respect
ively. The lowest suppression effective size boundary from 100 to 800 psi is between 0.897 to
0.465 µm.
The resuspension of submicron particles was observed under all spray pressure levels, and
the cap size for particle resuspension is lower for a higher-pressure spray than for a lower-
pressure spray. There is a substantial amount of submicron particles that remain airborne
after a 60-minute spray test, especially for the low-pressure condition. Overall, results show
a low capture efficiency and high resuspension possibility for micron- or nano-scale particles
under the testing conditions. Considering the adverse effect of smaller particles on lung func
tion and the higher risk to human health compared to the same mass concentration of dust
with a relatively larger size, other mitigation methods or ventilation tools should be used to
address this problem.
DISCLAIMER
The findings and conclusions in this report are those of the author(s) and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health
(NIOSH), Centers for Disease Control and Prevention (CDC). Mention of any company or
product does not constitute endorsement by NIOSH.
REFERENCES
Stacey P., Andrew Thorpe, and Paul Roberts. 2011. Levels of respirable dust and respirable crystalline
silica at construction sites, RR878 research report, Health and Safety Executive.
Sarver E, Keles C, Rezaee M. 2019. Characteristics of respirable dust in eight Appalachian coal mines:
A dataset including particle size and mineralogy distributions, and metal and trace element mass
concentrations. Data Brief. 2019;25:104032.
Doney BC, Blackley D, Hale JM. 2019. Respirable coal mine dust in underground mines, United States,
1982–2017. Am J Ind Med. 62(6):478–485.
Cecala A. AD O’Brien, J Schall et al. 2019. Dust control handbook for industrial minerals mining and
processing, Second Edition. National Institute for Occupational Safety and Health, DHHS (NIOSH)
Publication No. 2019-124 (RI 9701)
254
Brown JS, Gordon T, Price O, Asgharian B. 2013. Thoracic and respirable particle definitions for human
health risk assessment. Part Fibre Toxicol. 2013 Apr 10;10:12. doi: 10.1186/1743-8977-10-12.
WHO, OEH. 1999. Hazard prevention and control in the work environment:Airborne dust. Geneva.
European Committee for Standardization. 2014. EN 13205 Part 1–6, Workplace exposure Assessment of
sampler performance for measurement of airborne particle concentrations. Brussels.
Cheng WM, Nie W, Zhou G and Zuo QM. 2011. Study of dust suppression by atomized water from
high-pressure sprays in mines. Journal of China University of Mining and Technology 40(2):185–189.
Han, H., Wang, P., Liu, R. et al. 2020. Experimental study on atomization characteristics and
dust-reduction performance of four common types of pressure nozzles in underground coal mines.
Int J Coal Sci Technol 7, 581–596. doi.org/10.1007/s40789-020-00329-w
Majithia A.K., Hall S., Harper L., Bowen P.J. 2008. Droplet Breakup Quantification and Processes in
Constant and Pulsed Air Flows, Paper ID ILASS08-4-4, Institute for Liquid Atomization and Spray
Systems, Sep. 8–10, Como Lake, Italy.
Walton W. and Woolcock A. 1960. The suppression of airborne dust by water sprays. Int. J. Air. Pollut.
3. 129-1.53.
Fuchs N.A. 1964. The Mechanics of Aerosols, p. 165. Macmillan, New York.
Organiscak JA, Page SJ. 2005. Development of a dust collector inlet hood for enhanced surface mine
drill dust capture. Int J Surf Min Reclam Environ. 19(1):12–28.
McCoy, J.F., Schroeder, W.E., Rajan, S.R., Ruggieri, S.K., and Kissell, F.N. 1985. New Laboratory
Measurement Method for Water Spray Dust Control Effectiveness, American Industrial Hygiene Asso
ciation Journal, 46:12, pp:735–740, DOI:10.1080/15298668591395652.
Yu H, Wang Y, Liu E, Zhang G, Huang L, Liu L, et al. 2020. Experimental study of the influences of
temperature on the properties of particles in a gasifier during coal-water slurry gasification. ACS
Omega. 2020;5:5698–703.
Keles C, Taborda MJ, Sarver E. 2022. Characteristics of respirable dust in eight Appalachian coal mines:
A dataset including particle size and mineralogy distributions, and metal and trace element mass con
centrations with expanded data to cover a total of 25 US mines. Data Brief. 2022;42:108125.
Zhang Yinci, Amin Li, Jiafeng Gao et al. 2022. Differences in the characteristics and pulmonary toxicity
of nano- and micron-sized respirable coal dust, Respiratory Research, 23: 197. doi.org/10.1186/s12931-
022-02120-8.
Fan, L., Liu, S. 2021. Respirable nano-particulate generations and their pathogenesis in mining work
places: A review. Int J Coal Sci Technol 8, 179–198. doi.org/10.1007/s40789-021-00412-w.
Colinet JF, Rider JP, Listak JM, Organiscak JA, Wolfe AL. 2010. Best Practices for Dust Control in
Coal Mining. U.S. Department of Health and Human Services, Centers for Disease Control and Pre
vention, National Institute for Occupational Safety and Health. 2010 Information Circular 9517,
DHHS (NIOSH) Publication No. 2010–110.
Beck T.W., C.E. Seaman, M.R. Shahan, and S.E. Mischler. 2018. Open-air sprays for capturing and con
trolling airborne float coal dust on longwall faces, Min Eng., 70(1): 42–48. doi:10.19150/me.7978.
Han, H., Wang, P., Liu, R. et al. 2020. Experimental study on atomization characteristics and
dust-reduction performance of four common types of pressure nozzles in underground coal mines.
Int J Coal Sci Technol 7, 581–596. doi.org/10.1007/s40789-020-00329-w.
Jiang H., . Zheng Y, Klima S., Seaman C.E. and Beck T.W. 2022. A Laboratory Study of the Dust
Deposition and Suppression Process for Respirable Coal Dust in a Confined Chamber, Preprint for
SME annual conference, Feb 27- Mar 02, Salt Lake City, UT.
255
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
William Arnott
Atmospheric Science Program, Department of Physics, University of Nevada, Reno
Charles Kocsis
Mining Engineering Department, University of Utah, Salt Lake City
Xiaoliang Wang
Desert Research Institute, Reno, Nevada
1 INTRODUCTION
Respirable (Vincent, 2005) coal dust is produced when coal is extracted from the seam by drill
ing, blasting, cutting, loading and transporting operations. In underground mines it consists
of aerosol with diameters (0.01 - 20 μm) of coal, silica, kaolinite, and other minerals which can
be inhaled and deposited into the lungs, to a worker’s alveoli where oxygen enters the blood
stream. This is a serious health concern because in this region of the lungs, the human body
has few means of removing hazardous particles. As a result, respirable mine dust, which
includes coal as well as carbonate (i.e. rock dust (Pokhrel, Keles, Jaramillo, Agioutanti, &
Sarver, 2021)) and non-carbonate material (i.e. RCS) can generate serious health hazards if
these elevated dust concentrations are not adequately controlled (Kachuri et al., 2014).
The long-term exposure of coal miners to high levels of respirable coal dust can cause
pneumoconiosis, emphysema, and other occupational lung diseases such as the chronic
obstructive pulmonary disease, which is collectively known as black lung disease. Overexpos
ure to respirable silica dust can lead to silicosis. Short-term, but elevated respirable silica dust
DOI: 10.1201/9781003429241-27
256
concentrations are considerably more hazardous than the same level of cumulative exposure
at lower concentrations (e.g. recommend10-hour average exposure to RCS of < 0.05 mg m-3)
(Howard, 2011). Consequently, it is important to design, manufacture, and commercialize
a new personal real-time monitoring instrument, which has the ability to measure and monitor
both respirable coal dust and respirable silica dust concentrations in the production workings
and throughout the mine.
Our goal is to develop, fabricate and test a new real-time personal coal and silica dust monitor
ing instrument, based on photoacoustic absorption spectroscopy, as opposed to time averaged
filter-based design principles such as end-of-shift FTIR analysis (Pampena, Cauda, Chubb, &
Meadows, 2020) or quantum cascade laser filter analysis (Wei, Kulkarni, Ashley, & Zheng, 2017)
or aerosol light scattering that is not sensitive to aerosol composition (Zhang, Nie, Liang, Chen,
& Peng, 2021). The instrument will have the ability to continuously measure respirable coal dust
and respirable silica dust concentrations in the production areas and throughout the mine with
no user intervention and minimal equipment maintenance needs. As opposed to the current
filter-based technology, the new instrument is based on photoacoustic principles; the effects of
the absorbed laser light by the carbon and silica particles are measured with an acoustic detection
system. Details are in the following section. The photoacoustic measurement theory has been
developed (Taylor, Nascimento, Arnott, & Kocsis, 2022) and experimental demonstration in the
laboratory has occurred for dust containing coal, RCS, and kaolinite (Nascimento et al., 2022).
This manuscript extends our previous work to include calcite dust (i.e. rock dust), and to
give the theory and results for measurements of the SDMC that are needed for autonomous
instrument operation.
2 METHODS
Laboratory measurements of SDMC are obtained using our Marple-like (Marple & Rubow,
1983) dust resuspension chamber shown schematically in Figure 1 (Nascimento et al., 2022).
Figure 1. Schematic of the aerosol resuspension chamber used for measuring SDMC and evaluating
and calibrating the photoacoustic instrument. The chamber is 2 meters long and the plexiglass window is
centered. The current PA instrument is on a 2ʹ x 4ʹ optical table and the instrument enclosure fits in a 19”
equipment rack.
Dust is dispensed from a container, measured at ¼ teaspoon, and placed in the metal bowl.
A five second duration burst of compressed air suspends the dust. The dust mixes to
a uniform state in about six minutes, its concentration diminishing slowly over about a three-
hour interval. The inlets for each instrument are close together and are approximately 20 cm
above the chamber floor.
The photoacoustic instrument measures aerosol light absorption by means of an acoustic detec
tion system. A schematic of the instrument is shown in Figure 2, taken from (Nascimento et al.,
2022) where details may be found. Briefly, sample air enters the inlet and is drawn through the
instrument with a vacuum pump. The aerosols interact with laser light in the resonator section,
257
Figure 2. Schematic of the photoacoustic instrument.
creating the sound that is measured to quantify light absorption. Aerosol heats during light absorp
tion and transfers heat to the surrounding air, thereby creating a very small pressure wave, sound,
at the resonant frequency of the chamber since the laser power is modulated at this frequency
(Arnott, Moosmuller, Rogers, Jin, & Bruch, 1999), then detected by a microphone. The instrument
gains spectroscopic capability by use of a tunable quantum cascade laser (QCL), as the wavelength
can be changed quickly. Laser power is measured with a mercury cadmium telluride (MCT)
detector salvaged from a Fourier Transform Infrared Spectrometer (FTIR). The piezoelectric trans
ducer is used to measure the resonance frequency, quality factor, and microphone calibration at
resonance. Measurements are reported for both 1 and 16 second integration time. Noise equivalent
light absorption measurements are obtained by integrating the microphone signal away from reson
ance (Lewis, Arnott, Moosmuller, & Wold, 2008).
The Aerodynamic Particle Sizer Spectrometer (APS, model 3321, TSI Incorporated, Shore
view, MN, USA) instrument optically detects the time of flight of accelerated aerosol to
obtain aerodynamic diameter for individual aerosol. A 20:1 dilution ratio for sample air
versus clean air was achieved using the TSI Aerosol Diluter model 3302A. The mass distribu
tion for laboratory generated RCS is shown in Figure 3. Distributions for other dust types
were similar. Most of the aerosol mass is below 4 μm, especially after the middle of the experi
ment, as the larger particles settle. The mass distributions are integrated to obtain real-time
respirable dust mass concentrations. Because the user must input the dust density for mass
concentration measurements, the APS is best suited for experiments involving homogeneous
dust. Average mass distribution near coal mining fronts have peak diameters of 3-6 μm, and
Figure 3. Mass distribution measurements from the APS for silica. Curves are for the indicated times.
The vertical line is at diameter of 4 μm.
258
have more coarse mode aerosol than our laboratory aerosol, indicating a need to test our
instruments in mines (Skubacz, Wojtecki, & Urban, 2016).
The SPS30 is a low cost air quality sensor manufactured by Sensirion (Sensirion, 2020). It is an
optical particle counter that reports aerosol number concentrations for 4 size ranges in addition to
PM1, PM2.5, PM4, and PM10. The recommended mass concentration range of PM2.5 is up to
1 mg m-3. We have found that SPS30 PM4 values correlate well with gravimetric mass measure
ments, even with concentrations up to around 50 mg m-3. Laboratory evaluation showed that the
SPS30 over (under) reported PM2.5 for woodsmoke (Arizona road dust) compared with filter
measurements, and that the SPS30 operated properly after prolonged sampling at concentrations
as high as 33 mg m-3 (Tryner, Mehaffy, Miller-Lionberg, & Volckens, 2020).
Gravimetric and FTIR sampling were performed using PVC filters to obtain total respirable
mass and silica concentrations using the NIOSH FAST technique. SKC aluminum cyclones (Zefon
International, Catalog NO. ZA0060) selected respirable dust that was deposited through use of per
sonal sampling cassettes (Zefon International, REF 745PVC-CF-FTIR) on 37 mm PVC filters
(Zefon International, REF FSP37R). A vacuum pump external to the aerosol chamber was con
nected to the filter samplers with tubing, and operated nominally at 2.5 LPM as measured with
a flow meter. The cyclone sampler operates at 2.5 LPM to obtain respirable dust. The exposed area
of the filter was 6.92 cm2. A custom filter holder was developed for FTIR measurements.
3 RESULTS
As previously mentioned, laboratory dust simulants may have different composition and aerosol
size than dust found in mining environments. Many grades and type of coal are found in nature,
though all absorb light throughout the 11-13 μm measurement range. Pre-dispersed silica dust can
coagulate into larger, spherical clumps, perhaps due to static generated by the plastic storage con
tainer. Our research has identified key wavelengths to use to measure absorption by four coal-mine
-relevant dust species. We have compared the SPS30 and APS measurements of total dust mass
concentration with gravimetric measurements, and have developed a method to calculate SDMC
from multispectral photoacoustic aerosol light absorption measurements.
FTIR analysis of individual filters collected during dust sampling procedures confirms the
use of unique wavelengths for each components’ absorption spectrum to obtain SDMC. The
PAS can also be used to find the absorption spectrum of any sampled air; however, FTIR
analysis is used as the industry’s standard method to compare the accuracy of the PAS, as
Figure 4. FTIR analysis of dust mixture and individual dusts, demonstrating the absorption spectra for
each component, and how they come together to form the absorption spectra for the mixture. Also illus
trated are laser lines for the wavelengths best suited for quantifying dust constituents.
259
well to provide supporting evidence for the wavelengths chosen to quantify SDMC. Figure 4
shows FTIR absorption spectra obtained as described in (Nascimento et al., 2022) of the indi
vidual dust types (kaolinite, calcite, coal and silica) along with the dust mixture of all four
dust types in equal parts. This figure nicely illustrates the dust mixture’s absorption as
a combination of its constituents, represented by “Mixture PVC9 Babs.”
Figure 4 also illustrates the wavelengths at which kaolinite, calcite, coal and silica absorb
strongly, with the least amount of interference from water vapor and other minerals, and were care
fully chosen to represent each mineral. Each dust type also absorbs a little laser light at the other
three wavelengths. The wavelengths that will most effectively represent and quantify each mineral
in question are 11.040 μm for kaolinite, 11.408 μm for calcite, 11.826 μm for coal, and 12.495 μm
for silica. These wavelengths will be used in the further development of the PAS control software to
automatically obtain concentrations of each mineral. The process of obtaining SDMC through ana
lysis is elaborated on in Section 4. Strong light absorption by water vapor at 12.53 μm, for example,
can be used along with relative humidity measurements to periodically evaluate the instrument cali
bration (Nascimento et al., 2022; Taylor et al., 2022).
Measurements of aerosolized kaolinite in the dust chamber are shown in Figure 5 to further
demonstrate the absorption theory for SDMC. We can see how kaolinite has the greatest
absorption at the ‘Kaolinite λ’=11.04 μm, the wavelength chosen to quantify it. We can also
see how kaolinite absorbs at the other wavelengths designated for calcite, silica and coal. The
behavior of the dust as it is aerosolized can also be observed in this figure, from 14:27 for
about 60 seconds. Once the dust is aerosolized, there is a large spike in absorption measure
ments from the PAS as well as PM4 measurements from the other instruments. Immediately
afterwards, there is a sharp dip, representative of the dust plume as it is initially lifted, and
then rises out of reach of the sensors and vacuum pumps in the chamber. For this reason, to
more accurately represent mine conditions, data analysis and filter sampling doesn’t com
mence until about six minutes after aerosolization. Data analysis for determining SDMC can
be normalized using either data from the SPS30 or the APS. The APS is the standard for
quantifying particulate matter concentrations however, our findings suggest the SPS30 might
be a more suitable instrument for this purpose.
Gravimetric and optically determined aerosol mass concentrations by the SPS30 and APS
are shown in Figure 6 for all four dust types. APS data like that shown in Figure 3 are inte
grated over diameter until 4 μm, and the SPS30 directly reports PM4. “PM4” is used to repre
sent respirable gravimetric aerosol mass concentration. Figure 3 shows that most of the
260
Figure 6. Scatter plot of filter mass concentration graphed against PM4 measurements for both the
SPS30 and APS for all four dust types measured separately. The SPS30 has a much better fit to the gravi
metric mass concentration, with an R-squared value of 0.92.
aerosol mass diameter is below 4 μm for our laboratory generated aerosol. The SPS30 data
most closely resembles the filter mass, with a coefficient of determination of around 0.92, com
pared to the APS data, with a coefficient of determination of 0.55. Both however also have
a large offset. These findings support the use of the SPS30 PM data for use in the matrix inver
sion calculations.
There are two main equations that allow for the PAS to measure aerosol concentrations from
light absorption. The first is an expression for aerosol light absorption, Babs. It is given by
where Pm is the acoustic pressure measured with the microphone, PL is the laser power used,
about 5 mW in our case, Ares is the cross-sectional area of the resonator, γ=1.4 is the is the
ratio between isobaric and isochoric dry air specific heats, f0 is the resonance frequency of the
chamber, about 1500Hz, and Q is the quality factor, which is about 60. Babs has units of Mm–1,
or inverse megameters. All of these values are either controlled by experimental parameters or
measured with the PAS. The other equation vital to the success of our method is given by
where PM has units of [μg m–3] and MAE, or mass absorption efficiency, has units of [m2g–1].
In determining the SDMC, which here are silica, kaolinite, black coal and as a new addition
calcite, we must turn PM and Babs into vectors, with dimensions defined by the number of
species, which would subsequently turn MAE into a square matrix also with dimensions
defined by species count. For this case, there are four species present, corresponding to a 4x4
MAE matrix, and 4x1 PM and Babs vectors. The MAE matrix is separated into both species
and wavelength used to measure a species to be able to quantify contributions from other
261
species at wavelengths where a given species was strongly absorbing. For clarity, below is
a more detailed look at Equation 2:
Of course, the end goal here is measuring PM for different species. So, multiplying the left
of both equations with the inverse of the MAE matrix gives MAE 1 � Babs ¼ PM. The inver
sion of the matrix is done after all elements are obtained.
To inspect the relationship between PM and Babs, values of SPS30 PM4, SPS30 PM2.5, and
APS PM4 were plotted with Babs measurements as time series on separate axes for a given species
at a given wavelength. To obtain MAE matrix elements, PM values were plotted against Babs
value as a scatter plot, and a linear regression was performed. For four dust species, each meas
ured at four separate wavelengths, there are 16 total charts like this for each PM type. The slopes
of the regressions for PM2.5 were used to obtain MAE matrix elements. The completed inverse
MAE matrix, along with the example of all species combined together, is included as Equation 4.
The inverse MAE matrix, Equation 4, has units of [g m–2]. This matrix uses SPS30 PM2.5
values instead of the APS PM4 because one: SPS30 has proven to function with high-fidelity
compared to the APS, as shown in Figure 6; and two: PM2.5 appears to have more accur
acy in its prediction than does PM4. This is due to a more steady and consistent flow of
PM2.5 for the PAS. These Babs entries were calculated by fitting the PM2.5 time series for
the filter sample of the dust mixture, containing equal parts by volume of coal, kaolinite,
calcite and silica, to a second-order polynomial, applying this curve fit to Babs to renormalize
Babs such that there is no drop in PM for a given time interval (30 minutes in this case),
mimicking measuring all four species at once with four different lasers, instead of in
sequence. It will be necessary to perform the measurements sequentially at each wavelength
as quickly as possible (1-16 seconds per measurement) for a fully automated PAS, and to
include a SPS30 to normalize the PM measurements.
5 CONCLUSION
We have developed a method for obtaining speciated dust mass concentration (SDMC)
through mass absorption efficiency matrix inversion, found unique wavelengths at which to
quantify absorption spectra of kaolinite, calcite, coal and silica dusts and provided evidence to
support the use of SPS30 measurements in lieu of APS data. This was achieved with the use of
the real-time personal coal and silica dust monitoring instrument, based on photoacoustic
absorption spectroscopy. The procedure for obtaining SDMC was developed through use of
measurements with the laser manually set to the four wavelengths used for each dust type.
The current instrument is mounted on a 2ʹ x 4ʹ optical table and the instrument enclosure
fits in a 19” equipment rack. Future development of the photoacoustic spectrometer involves
refinement and further testing in both laboratory and mine settings. The instrument will have
the ability to continuously measure respirable coal and silica dust concentrations in the pro
duction areas and throughout the mine with no user intervention and minimal equipment
262
maintenance needs. Areas in need of further development include: automatic calibration with
relative humidity, measurement of speciated dust movement in-flight with rapid (1-16 second)
measurements at the four laser wavelength, reducing the size of the acoustical resonator,
replacing the liquid-nitrogen-cooled integrating sphere laser power detector with a smaller
device—all to reduce instrument size and to improve commercial viability.
The relevance and importance of this research cannot be stressed enough—the health and
safety of the men and women procuring natural resources, the very backbone of the develop
ing and developed world, depends on our ability to monitor the hazards they are exposed to
during a normal work day. Without the health and safety of our miners, the industry risks
productivity and financial sustainability, not to mention loss of life.
REFERENCES
Arnott, W. P., Moosmuller, H., Rogers, C. F., Jin, T. F., & Bruch, R. (1999). Photoacoustic spectrometer
for measuring light absorption by aerosol: instrument description. Atmospheric Environment, 33(17),
2845–2852.
Howard, John. (2011). Coal Mine Dust Exposures and Associated Health Outcomes: A Review of Informa
tion Published Since 1995. Retrieved from https://arlweb.msha.gov/s&hinfo/blacklung/2011-
172NIOSH.pdf
Kachuri, L., Villeneuve, P. J., Parent, M. E., Johnson, K. C., Harris, S. A., & Canadian Canc, Registries.
(2014). Occupational exposure to crystalline silica and the risk of lung cancer in Canadian men. Inter
national Journal of Cancer, 135(1), 138–148. doi:10.1002/ijc.28629
Lewis, K., Arnott, W. P., Moosmuller, H., & Wold, C. E. (2008). Strong spectral variation of biomass
smoke light absorption and single scattering albedo observed with a novel dual-wavelength photoa
coustic instrument. Journal of Geophysical Research-Atmospheres, 113(D16), 14. doi:10.1029/
2007jd009699
Marple, V. A., & Rubow, K. L. (1983). AN AEROSOL CHAMBER FOR INSTRUMENT EVALU
ATION AND CALIBRATION. American Industrial Hygiene Association Journal, 44(5), 361–367.
doi:10.1080/15298668391404978
Nascimento, P., Taylor, S. J., Arnott, W. P., Kocsis, K. C., Wang, X. L., & Firouzkouhi, H. (2022).
Development of a Real-Time Respirable Coal Dust and Silica Dust Monitoring Instrument Based on
Photoacoustic Spectroscopy. Mining, Metallurgy & Exploration, 9. doi:https://doi.org/10.1007/s42461-
022-00653-6
Pampena, J. D., Cauda, E. G., Chubb, L. G., & Meadows, J. J. (2020). Use of the Field-Based Silica
Monitoring Technique in a Coal Mine: A Case Study. Mining Metallurgy & Exploration, 37(2),
717–726. doi:10.1007/s42461-019-00161-0
Pokhrel, Nishan, Keles, Cigdem, Jaramillo, Lizeth, Agioutanti, Eleftheria, & Sarver, Emily. (2021).
Direct-on-Filter FTIR Spectroscopy to Estimate Calcite as A Proxy for Limestone ‘Rock Dust’ in
Respirable Coal Mine Dust Samples. Minerals, 11(9), 922.
Sensirion. (2020). SPS30 Specification Statement. In: Sensirion.
Skubacz, K., Wojtecki, L., & Urban, P. (2016). The influence of particle size distribution on dose conver
sion factors for radon progeny in the underground excavations of hard coal mine. Journal of Environ
mental Radioactivity, 162, 68–79. doi:10.1016/j.jenvrad.2016.05.020
Taylor, S. J., Nascimento, P., Arnott, W. P., & Kocsis, C. (2022). Real-time Photoacoustic Measure
ments of the Mass Concentration of Respirable Crystal Silica Dust: Theory. Mining Metallurgy &
Exploration, 10. doi:10.1007/s42461-022-00657-2
Tryner, J., Mehaffy, J., Miller-Lionberg, D., & Volckens, J. (2020). Effects of aerosol type and simulated
aging on performance of low-cost PM sensors. Journal of Aerosol Science, 150, 17. doi:10.1016/j.
jaerosci.2020.105654
Vincent, J. H. (2005). Health-related aerosol measurement: a review of existing sampling criteria and pro
posals for new ones. Journal of Environmental Monitoring, 7(11), 1037–1053. doi:10.1039/b509617k
Wei, S. J., Kulkarni, P., Ashley, K., & Zheng, L. N. (2017). Measurement of Crystalline Silica Aerosol
Using Quantum Cascade Laser-Based Infrared Spectroscopy. Scientific Reports, 7, 8. doi:10.1038/
s41598-017-14363-3
Zhang, H., Nie, W., Liang, Y., Chen, J. G., & Peng, H. T. (2021). Development and performance detec
tion of higher precision optical sensor for coal dust concentration measurement based on Mie scatter
ing theory. Optics and Lasers in Engineering, 144, 15. doi:10.1016/j.optlaseng.2021.106642
263
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
S. Slouka
Colorado School of Mines, Golden, CO, USA
E. Sidrow
University of British Columbia, Vancouver, Canada
C. Tsai
University of California Los Angeles, Los Angeles, CA, USA
J. Brune
Colorado School of Mines, Golden, CO, USA
ABSTRACT: In environments where excavation is used to break rock, airborne rock dust
poses respiratory health threats to workers. This includes mining with mechanical cutting
machines, which are common in many mines and tunneling operations. The purpose of this
study is to compare characteristics of rock dust generated by pick cutters at different wear
conditions such that results could be used in the evaluation of proper bit management and
dust suppression controls in the underground environment. The study included full-scale
laboratory cutting of concrete, limestone, and sandstone samples. Each rock was cut with
three conical picks each at new, moderately worn, and fully worn stages of wear. Dust samples
were analyzed to reveal the concentrations, mineralogy, particle size distributions, and particle
shapes. The results show that for all rock types, the worn picks consistently generated the
highest concentration of dust, all picks consistently contained quartz, and all three picks con
sistently generated dust particles of similar shapes. Additionally, although there are slight
shifts in the particle size distributions between the various pick wears, all the particle sizes
reside in the respirable range of less than 10µm in aerodynamic diameter for all the rock types
cut in this experiment.
1 INTRODUCTION
Exposure to airborne rock dust particles containing silica, coal, or other minerals, can cause
irreversible diseases such as coal worker’s pneumoconiosis (Department of Health and
Human Services, 2011), silicosis (Ross & Murray, 2004), and other lung complications (Pope
III et al., 2002; Stansbury, 2018). However, even with the regulations implemented by the US
Mine Safety and Health Administration (MSHA) to limit the concentration exposure and
exposure to silica, there has been an unfortunate increase in lung disease cases in the United
States since the 1990s with a continued rise in numbers in recent years (Blackley et al., 2016;
Doney et al., 2019; Hall et al., 2019; Lee et al., 2014; Zosky et al., 2016).
It is not fully understood why modern miners’ lung diseases are increasing with the MSHA
concentration and silica exposure limits in place (Centers of Disease Control and Prevention,
1974; MSHA, 1977). Therefore, various current research investigations are analyzing the pres
ence of nanoparticles, particle size distributions, mineralization analysis, and how particle
DOI: 10.1201/9781003429241-28
264
shape affects deposition into the human lung (Fan & Liu, 2021; Khattak et al., 2019; Lab
ranche et al., 2021; Sarver et al., 2019; Zellnitz et al., 2019). Additionally, there is limited
understanding and implementation of dust suppression management in relation to the pick
cutter life.
With the most dominant machines used for soft and moderate strength rock excavation
being roadheaders, continuous miners, or drum shearers, the primary rock cutting tools are
conical picks. Rock dust is created and released at the interface between the conical pick and
rock surface where some of the deposited dust generated can be transported and re-
introduced into the air during the muck handling process. Conical picks create dust at the car
bide tip because the concentrated loads they apply penetrate the rock surface and fragment
the rock into smaller pieces and dust particles. The geometry of the pick tip will change over
time as they are used because they wear down (Plis et al., 1988; W. Roepke et al., 1976), but it
is uncertain if the changing of the pick tip changes dust characteristics. The limited evidence
connecting pick wear to dust characteristics is that coal dust concentration increases with the
wear of pick tips (Fowell & Ochei, 1984; Hanson & Roepke, 1979; Qayyum, 2003;
W. W. Roepke & Hanson, 1983; Zhou et al., 2020). There are limited further investigations on
other rock types, nor is there a clear quantitative relationship of the concentration of dust gen
erated to the geometry of the pick tip.
Therefore, the following research investigates the concentration, mineral composition, par
ticle size distributions, and particle shapes of suspended and deposited rock dust generated
from three different conical pick wears during laboratory full scale cutting of a concrete, lime
stone, and sandstone rock block. The conical pick is used for cutting because these picks are
regularly implemented in excavation operations of concrete, limestone, and sandstone in
a variety of applications (Roxborough et al., 1981; Shao, 2016; Su & Akkaş, 2020; Talbo &
Sevigné, 1986). The purpose of the work is to compare the dust particle characteristics gener
ated from rock cutting with three pick tip geometries as it wears out during operation. This
can eventually guide operators to strategize bit management in relation to production rate
and machine utilization, with consideration of dust suppression measures when cutting soft to
medium-strength rock.
The overarching conclusions reveal that as the pick wear increases, the suspended respirable
dust concentration increases in all rock types. Independent of pick wear, all three picks in all
the rock types generated dust containing silica. Additionally, there are slight trends acknow
ledged between the pick wear and the particle size distributions for all the rock types, however,
all particle size distributions reside in the respirable size range. Lastly, all the picks generated
suspended respirable particles with similar particle shapes in all rock types in terms of round
ness, aspect ratio, and roughness measures.
2 RESEARCH METHODS
For full-scale cutting tests, a fine-grained concrete, Indiana limestone, and sandstone rock
block were used with the Linear Cutting Machine (LCM) at the Earth Mechanics Institute
(EMI) at the Colorado School of Mines. Each of the three samples share similar rock strength
properties and provide uniform and homogenous material throughout the specimens as seen
in Figure 1. For example, none of the samples contained any apparent joint sets or discontinu
ities in structure, which made cutting as consistent as possible. The rock strength properties
measured in the rock mechanics lab at EMI include unconfined compressive strength (UCS),
Brazilian tensile strength (BTS), Cerchar abrasivity index (CAI), and the punch penetration
energy slope index. These properties determined for each sample are given in Table 1.
The cutting methodology, dust collection setup, pick wear quantification, and particle char
acterization parameters are detailed in a previous study of the limestone block experiments
(Slouka, Brune, Rostami, et al., 2022). The same testing equipment and parameters are used
for all three rock samples. Therefore, when cutting rock samples, 0.76 cm (0.3 in) penetration
and 3.81 cm (1.5 in) spacings at 250 mm/sec (10 in/sec) are used on all samples. When collect
ing dust, the same Dorr-Oliver cyclones, Tsai Diffusion Samplers (TDS), and vacuums are
265
Figure 1. Concrete (left), limestone (center), and sandstone (right) samples cast into concrete and cut
with multiple lines of cuts along the surface.
utilized to collect suspended and deposited dust. When quantifying pick wear, the same U92
4.0 conical picks were used with the new, moderately worn, and fully worn pick tip radii at
1.8 mm (0.07 in), 3.7 mm (0.15 in), and 5.2 mm (0.2 in), respectively. The concentration, min
eralogy, particle size distribution, and particle shape characteristics are analyzed with the
same equipment and methods, such as utilizing the NIOSH manual of analytical methods
(NMAM), field emission scanning electron microscope (FE-SEM), and laser diffraction.
3 RESULTS
266
Figure 2. Dust concentrations of suspended respirable dust generated with three pick wears and deter
mined from NMAM 0600 method. A) Concrete. B) Limestone. C) Sandstone. Note that the scale of the
y-axis changes between sub-figures.
generated in dust while cutting. Additionally, there is weak statistical evidence that the rock
type and the pick wear interact to affect the percent silica generated within the dust.
Figure 3. Percent silica determined in the suspended respirable dust generated with three pick wears
and determined from NMAM 7500 method. A) Concrete. B) Limestone. C) Sandstone. Note that the
scale of the y-axis changes between sub-figures.
267
3.2 Particle size distributions
Laser diffraction was used to determine the particle size distribution for the suspended and
deposited particles, as outlined in previous studies. The “percent channel” used for analysis
results refers to the bins during particle counting, where each particle is individually analyzed
and placed into bins dependent on their size. The physical diameters obtained were then cor
rected to aerodynamic diameters using the Cunningham correction factor and slip correction
factor (Hinds, 1999). Additionally, each colored curve displayed in Figures 4 and 5 is the aver
age of three duplicate sample runs.
Figure 4 reveals the suspended particle size distributions for the three pick wears generated
from the three rock types. Visually analyzing the graphs, the particle size single-mode distribu
tions generated from each pick wear in every rock type lay close to one another without any
shifts or trends, which is expected when cutting rock materials. However, the sandstone gener
ated slightly more particles in smaller sizes compared visually to the concrete and limestone
samples. Therefore, Kolmogorov-Smirnov (KS) statistical tests were performed between par
ticle size distributions generated by the three pick wears to test the null hypothesis that the
airborne distributions of particle sizes generated from each pick are the same. A significance
level of 0.05 was used in the statistical analysis. Therefore, p-values below 0.05 provide strong
statistical evidence to reject the null hypothesis. The results suggest that the particle size distri
butions are statistically different between different pick wears with the trend confirmed for
each rock type cut. For example, all p-values determined between particle size distributions
generated from the three different pick wears were 1.5E-10 or less.
Additionally, Figure 5 reveals the deposited particle size distributions for the three pick
wears generated from the three rock types. For concrete, limestone, and sandstone, the aver
age values for each distribution increase with the increase in wear. For example, concrete aver
age deposited particle aerodynamic diameter from the new, moderately worn, and fully worn
pick was 2.47, 2.74, and 2.80 respectively with standard deviations between 1.9 and 2.2 for
each of the pick wears.
KS statistical tests were also performed between the deposited particle size distributions
generated by the three pick wears. Of the about 10,000 particles analyzed for each pick wear
of each rock, the recorded sizes were input into the KS test to test the null hypothesis that the
distributions of particle sizes generated from each pick are the same. A significance level of
0.05 was used in the statistical analysis. Therefore, p-values below 0.05 provide strong statis
tical evidence to reject the null hypothesis. The results suggest that the particle size distribu
tions are statistically different between size distributions generated by different pick wears.
This trend is confirmed for each rock type cut. For example, all p-values determined between
particle size distributions generated from the three different pick wears were 9.1E-27 or less.
268
Figure 4. Suspended particle size distributions generated from three different pick wears. A) Concrete.
B) Limestone. C) Sandstone.
Figure 5. Deposited particle size distributions generated from three different pick wears. A) Concrete.
B) Limestone. C) Sandstone.
269
Figure 6. FE-SEM images of suspended dust generated from the moderate pick while cutting concrete
(left), limestone (center), and sandstone (right).
Figure 7. Suspended respirable particle shapes generated from the three rock types. A) Roundness
values. B) Aspect ratios. C) Roughness values. D) Two-way ANOVA p-value results of particle shapes.
particle shapes. On the other hand, with the rock type p-values all below the 0.05 threshold,
the statistical test provides evidence that the rock type causes an effect to the roundness,
aspect ratio, and roughness particle shapes.
4 DISCUSSION
270
dusts need to be mitigated and suppressed to limit silica exposures when cutting concrete,
limestone, or sandstone.
Table 2. Average particle shape values for the three various rock types cut.
Shape characteristic Concrete Limestone Sandstone Overall average
271
5 CONCLUSION
Suspended and deposited dust was generated from concrete, limestone, and sandstone and
collected while cutting full-scale rock samples with a new, moderately worn, and worn conical
pick. The concentrations, silica contents, particle size distributions, and particle shapes were
determined for each dust generated. The analyses reveal that suspended respirable dust con
centrations increased as the pick wear increased for all rock types. In terms of silica, all three
picks cutting all the rock types generated suspended dust containing quartz or cristobalite. In
addition, the suspended particle size distributions are similar for all pick types, whereas the
deposited particle sizes slightly increase as the pick wear increases. Although there are slight
trends between the pick wear and the deposited particle size distributions for all the rock
types, there are no significant differences as all particle size distributions reside in the respir
able size range. Lastly, the pick wear does not influence any change in the particle shapes. All
the picks consistently generated suspended respirable particles with similar particle shapes
that are slightly oval with mostly smooth edges.
REFERENCES
Blackley, D. J., Crum, J. B., Halldin, C. N., Storey, E., & Laney, A. S. (2016). Resurgence of progressive
massive fibrosis in coal miners — Eastern Kentucky, 2016. Morbidity and Mortality Weekly Report, 65
(49), 1385–1389. https://doi.org/10.15585/mmwr.mm6549a1
Centers of Disease Control and Prevention. (1974). Recommendations for an occupational exposure stand
ard for crystalline silica (pp. 1–23). https://www.cdc.gov/niosh/pdfs/75-120a.pdf
Department of Health and Human Services. (2011). Coal Mine Dust Exposures and Associated Health
Outcomes: A Review of Information Published Since 1995. Current Intelligence Bulletin, 64.
Dogruoz, C., & Bolukbasi, N. (2014). Effect of cutting tool blunting on the performances of various
mechanical excavators used in low- and medium-strength rocks. Bulletin of Engineering Geology and
the Environment, 73(3), 781–789. https://doi.org/10.1007/s10064-013-0551-y
Dogruoz, C., Bolukbasi, N., Rostami, J., & Acar, C. (2016). An Experimental Study of Cutting Perform
ances of Worn Picks. Rock Mechanics and Rock Engineering, 49(1), 213–224. https://doi.org/10.1007/
s00603-015-0734-x
Doney, B. C., Blackley, D., Hale, J. M., Halldin, C., Kurth, L., Syamlal, G., & Laney, A. S. (2019). Res
pirable coal mine dust in underground mines, United States, 1982-2017. American Journal of Industrial
Medicine, 62(6), 478–485. https://doi.org/10.1002/ajim.22974
Evans, I. (1965). The force required to cut coal with blunt wedges. International Journal of Rock Mechan
ics and Mining Sciences And, 2(1), 1–12. https://doi.org/10.1016/0148-9062(65)90018-5
Fan, L., & Liu, S. (2021). Respirable nano-particulate generations and their pathogenesis in mining
workplaces: a review. International Journal of Coal Science and Technology, 8(2), 179–198. https://doi.
org/10.1007/s40789-021-00412-w
Fowell, R. J., & Ochei, N. N. (1984). A comparison of dust make and energy requirements for rock cut
ting tools. International Journal of Mining Engineering, 2(1), 73–83. https://doi.org/10.1007/
BF00880859
Hall, N. B., Blackley, D. J., Halldin, C. N., & Laney, A. S. (2019). Current Review of Pneumoconiosis
Among US Coal Miners. Current Environmental Health Reports, 6(3), 1–11. https://doi.org/10.1007/
s40572-019-00237-5.Current
Hanson, B. D., & Roepke, W. W. (1979). Effect of Symmetric Bit Wear and Attack Angle on Airborne
Respirable Dust and Energy Consumption. Report of Investigations - United States, Bureau of Mines,
8395.
Hinds, W. C. (1999). Aerosol technology: Properties, behavior, and measurement of airborne particles.
In Library of Congress Cataloging in Publication Data. https://doi.org/10.1016/0048-9697(83)90099-2
Hurt, K. G., & Evans, I. (1981). Point Attack Tools: an Evaluation of Function and Use for Rock
Cutting. Mining Engineer, 140(234), 673–683.
Khattak, J., Theisen, D., & Tsai, C. S. J. (2019). Sampling Techniques on Collecting Fine Carbon Nano
tube Fibers for Exposure Assessment. Scientific Reports, 9(1), 1–8. https://doi.org/10.1038/s41598-019-
43661-1
Kleinstreuer, C., & Feng, Y. (2013). Computational analysis of non-spherical particle transport and
deposition in shear flow with application to lung aerosol dynamics - A review. Journal of Biomechan
ical Engineering, 135(2). https://doi.org/10.1115/1.4023236
272
Kori, J., & Pratibha. (2019). Simulation and modeling for aging and particle shape effect on airflow
dynamics and filtration efficiency of human lung. Journal of Applied Fluid Mechanics, 12(4),
1273–1285. https://doi.org/10.29252/jafm.12.04.29290
Labranche, N., Keles, C., Sarver, E., Johnstone, K., & Cliff, D. (2021). Characterization of particulates
from australian underground coal mines. Minerals, 11(5), 1–10. https://doi.org/10.3390/min11050447
Lee, K., Lee, E. J., Lee, S. Y., In, K. H., Kim, H.-K., & Kang, M.-S. (2014). Resurgence of
a Debilitating and Entirely Preventable Respiratory Disease among Working Coal Miners. American
Journal of Respiratory and Critical Care Medicine, 190(6), 708–709.
MSHA. (1977). Federal Mine Safety & Health Act of 1977 (Mine Act).
Ohsaki, S., Mitani, R., Fujiwara, S., Nakamura, H., & Watano, S. (2019). Effect of particle-wall inter
action and particle shape on particle deposition behavior in human respiratory system. Chemical and
Pharmaceutical Bulletin, 67(12), 1328–1336. https://doi.org/10.1248/cpb.c19-00693
Plis, M., Wingquist, C., & Roepke, W. (1988). Preliminary Evaluation of the Relationship of Bit Wear to
Cutting Distance, Forces, and Dust Using Selected Commercial and Experimental Coal- and
Rock-Cutting Tools. US Bureau of Mines: Report of Investigations, 9193.
Pope III, C. A., Burnett, R. T., Thun, M. J., Calle, E. E., Krewski, D., & Thurston, G. D. (2002). Lung
Cancer, Cardiopulmonary Mortality, and Long-term Exposure to Fine Particulate Air Pollution. The
Journal of the American Medical Association, 287(9), 1132–1141. http://jama.jamanetwork.com/article.
aspx?doi=10.1001/jama.287.9.1132
Qayyum, R. A. (2003). Effects of bit geometry in multiple bit-rock interaction.
Roepke, W., Lindroth, D. P., & Myren, T. A. (1976). Reduction of dust and energy during coal cutting
using point-attack bits. Bureau of Mines, 8185.
Roepke, W. W., & Hanson, B. D. (1983). Effect of asymmetric wear of point attack bits on coal-cutting
parameters and primary dust generation. US Department of the Interior, Bureau of Mines, 8761.
Ross, M. H., & Murray, J. (2004). Occupational respiratory disease in mining. Occupational Medicine, 54
(5), 304–310. https://doi.org/10.1093/occmed/kqh073
Roxborough, F. F., King, P., & Pedroncelli, E. J. (1981). Tests on the Cutting Performance of
a Continuous Miner. Journal of The South African Institute of Mining and Metallurgy, 81(1), 9–25.
https://doi.org/10.1016/0148-9062(81)90237-0
Sarver, E., Keles, C., & Rezaee, M. (2019). Characteristics of respirable dust in eight appalachian coal
mines: A dataset including particle size and mineralogy distributions, and metal and trace element
mass concentrations. Data in Brief, 25, 104032. https://doi.org/10.1016/j.dib.2019.104032
Shao, W. (2016). A study of rock cutting with point attack picks. The University of Queensland, School of
Mechanical and Mining Engineering, Australia.
Slouka, S., Brune, J., & Rostami, J. (2022). Characterization of Respirable Dust Generated from
Full-Scale Laboratory Igneous Rock Cutting Tests with Conical Picks at Two Stages of Wear.
Mining, Metallurgy & Exploration. https://doi.org/10.1007/s42461-022-00625-w
Slouka, S., Brune, J., Rostami, J., Tsai, C., & Sidrow, E. (2022). Characterization of Respirable Dust
Generated from Full Scale Cutting Tests in Limestone with Conical Picks at Three Stages of Wear.
Minerals, 12(8), 930. https://doi.org/10.3390/min12080930
Stansbury, R. C. (2018). Progressive massive fibrosis and coal mine dust lung disease: The continued
resurgence of a preventable disease. Annals of the American Thoracic Society, 15(12), 1394–1396.
https://doi.org/10.1513/AnnalsATS.201809-598ED
Sturm, R. (2010). Theoretical models for dynamic shape factors and lung deposition of small particle
aggregates originating from combustion processes. Zeitschrift Fur Medizinische Physik, 20(3),
226–234. https://doi.org/10.1016/j.zemedi.2010.04.001
Su, O., & Akkaş, M. (2020). Assessment of pick wear based on the field performance of two transverse
type roadheaders: a case study from Amasra coalfield. Bulletin of Engineering Geology and the Environ
ment, 79(5), 2499–2512. https://doi.org/10.1007/s10064-019-01712-x
Talbo, H., & Sevigné, C. (1986). Longwall mining system U.S. Patent No. 4,382,633.
Zellnitz, S., Zellnitz, L., Müller, M. T., Meindl, C., Schröttner, H., & Fröhlich, E. (2019). Impact of drug
particle shape on permeability and cellular uptake in the lung. European Journal of Pharmaceutical
Sciences, 139(September), 105065. https://doi.org/10.1016/j.ejps.2019.105065
Zhou, W., Wang, H., Wang, D., Zhang, K., Du, Y., & Yang, H. (2020). The effect of geometries and
cutting parameters of conical pick on the characteristics of dust generation: Experimental investigation
and theoretical exploration. Fuel Processing Technology, 198 (October2019), 106243. https://doi.org/
10.1016/j.fuproc.2019.106243
Zosky, G. R., Hoy, R. F., Silverstone, E. J., Brims, F. J., Miles, S., Johnson, A. R., Gibson, P. G., &
Yates, D. H. (2016). Coal workers’ pneumoconiosis: An Australian perspective. Medical Journal of
Australia, 204(11), 414–418.e2. https://doi.org/10.5694/mja16.00357
273
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Increased exposure to respirable coal mine dust leads to irreversible lung dis
eases in mine workers. MSHA mandates a maximum dust concentration of 1.5 mg/m3 in
underground coal operations. Current ways of monitoring concentrations using a filter-based
method or expensive PDM3700 monitors cannot provide real-time concentration. Low-cost
light-scattering particulate matter (PM) sensors have the potential to overcome these chal
lenges. Tests were performed using factorial design to evaluate the impact of wind velocity
and sensor direction on the sensor response. Three levels of wind velocity 3.0 m/s, 1.5 m/s, and
0.5 m/s, and sensor direction towards the stream, perpendicular to the stream and opposite to
the stream and were evaluated. The results showed that the direction of the sensor has
a significant impact on their response. For each input concentration of 0.0-1.0 mg/m3,
1.0-2.0 mg/m3, and 2.0-3.0 mg/m3, PSM 5003 sensor has a superior response than any other
sensor. This study will provide valuable information on the feasibility of using these sensors to
make real-time decisions.
1 INTRODUCTION
Coal mining generates dust whose exposure for longer periods results in various health dis
eases including coal workers’ pneumoconiosis (CWP) and silicosis (Ishtiaq et al., 2018).
Increased exposure leads to various lung diseases that increase miners’ mortality (Zhang et al.,
2007). The permissible coal mine dust exposure in underground and surface coal mines of the
USA is 1.5 mg/cm3 given by Mine Safety and Health Administration (MSHA). CWP has
steadily increased in underground coal miners since 2000, especially in young miners in the
Appalachian region (Blackley, David J. Laura E., Reynolds Connie Short, Ron Carson,
Eileen Storey, Cara N. Halldin, 2022; Potera, 2019), (Sellaro et al., 2015). Even with the
advancement in underground coal excavation methods, the coal workers in an underground
mine are at risk of inhalation of respirable coal mine dust (dos S Antao et al., 2005). There
fore, due to the threat posed by the coal mine dust to the miners, it is critical to constantly
monitor the dust concentration.
Current ways of monitoring coal mine dust concentration are either filter-based methods or
using the MSHA-certified PDM3700 monitor. Some instruments use filter-based gravimetric
methods to measure PM concentration. FLIR Airtec is a filter-based DPM monitor that
measures the elemental carbon (EC) to determine the exposure of workers in underground
mines to diesel particulate matter. One disadvantage associated with these monitors is that
EC’s standard varies in different countries. EC: TC ratios must be set on the instrument to get
the conversion of EC to TC and hence the DPM concentration. Due to drastic changes in
temperatures during experiments, vapor condensation can increase on the sampling cassette
which results in an overestimation of particle concentration (Janisko and Noll, 2010; Takiff
DOI: 10.1201/9781003429241-29
274
and Aiken, 2010). Additionally, the measured concentration is based on the average value of
high and low concentrations during the shift (Wang et al., 2016). A high concentration spike
for a short time can be misleading (Chung et al., 2001). The second type of instrument, like
the PDM3700, uses a micro-balanced-based method using a tapered element oscillating micro
balance (TEOM). The PDM measures dust concentration every minute. Research has shown
that transport losses and particle blow-off of the filter may result in underestimation of the
results (Halterman et al., 2018; Volkwein et al., 2006a, 2004). The PDM’s high cost also pro
hibits it to be used by each miner during the shift. Another disadvantage reported in PDM is
reporting negative mass concentration due to the loss of volatile material due to heating of the
air inlet and tapered element to ensure stability (Abbasi et al., 2021). Therefore, these instru
ments are economically expensive to be utilized frequently to collect the PM concentration.
Low-cost light-scattering PM sensors have the potential to overcome these drawbacks. The
particles are exposed to the light beam in the sensing area, and the intensity of the light scattered
by particles is detected by a photodetector. These sensosor does not directly measure the mass
concentration rather the microprocessor converts the intensity into mass or number concentra
tion (Masson et al., 2015; van den Bossche et al., 2017). Recently, three different low-cost PM
sensors (Shinyei PPD42 NS, Sharp GP2Y1010AU0F, and Lase SEN0177) were tested for coal
mine dust and their response was compared with a reference monitor DustTrack. These sensors
yielded a fair response when compared to the reference monitor. The mean square error (MSE)
was the least for the laser sensor (0.0206) and MSE was 0.0232 and 0.0311 for Shinyei and
Sharp sensors (Ghamari et al., 2022). In another study, an optical sensor was used for coal mine
dust concentration measurement. Results showed that this sensor can measure coal mine dust
concentration accurately and also has good stability. (Zhang et al., 2021a).
There are limited studies available on the performance of low-cost PM sensors when meas
uring coal mine dust. A study compared the low-cost sensor performance but their responses
were compared to DustTrack as a reference monitor (Ghamari et al., 2022). DustTrack is not
a certified reference monitor to measure coal mine dust unlike a PDM 3700. In another study,
two low-cost sensors, Plan tower PMS1003 and Alphasense CO–B4, were tested as an alterna
tive for long-term air quality monitoring. These sensors were located at four locations in Aus
tralia and China and one of the testing sites in China had coal mine dust as one of the
pollution sources (Liu et al., 2020). The author did not explain the response of these sensors
when challenged with the coal dust. Based on these studies, it is important to evaluate the per
formance of low-cost sensors in coal mine dust measurement. Recently, an optical sensor for
higher-precision coal mine dust measurement was developed. The results indicated that this
showed considerable precision and stability while measuring coal mine dust. The mean meas
ured error was found to be 3.0-4.0 % in short-term tests and 4.0-5.0% in long-term tests. the
coal concentration used in this research ranged between 200.0-800.0 mg/m3 (Zhang et al.,
2021b). This concentration is exceptionally high compared to the limit imposed by MSHA.
This requires more studies on its detection performance near exposure limit concentrations.
Wind velocity and sensor direction were shown to have an impact on the PM sensors’
response. However, their impact on coal mine dust measurement is unknown. A few studies
have reported the impact of low wind velocities on the performance of low-cost PM sensors.
A wind velocity of less than 0.5 m/s was used for low-cost sensor evaluation (Alfano et al.,
2020; Zikova et al., 2017). In underground mining, an airflow velocity ranges between
1.0-2.0 m/s to provide a comfortable environment for the workers (Roghanchi et al., 2016).
Although most pollutants can be diluted with an airflow velocity of 0.3 m/s. However, the vel
ocity usually varies between 1.0-3.0 m/s in the production working areas (Christensen et al.,
1984; Fobelets, 1987). A wind velocity of more than 4.0 m/s can enhance the amount of dust
produced (Fanger, 1977). Another study evaluated the low-cost PM sensor with a wind vel
ocity range of 2.0-7.0 m/s (Mukherjee et al., 2017). In a recent study, three low-cost plan
tower PSM3003 sensors were evaluated that concluded that changing wind direction has an
impact on the performance of these low-cost PM sensors (Bulot et al., 2019; Liu et al., 2020).
There is a need to assess the performance of low-cost PM sensors with the lowest and highest
range of air velocity in underground mining. The impact of the low-cost PM sensor inlet direc
tion on the accuracy of the measurements needs to be investigated as well.
275
Thus, this paper aims to evaluate the impact of wind velocity and sensor direction on the
performance of low-cost PM sensors. Evaluation will be performed using Keystone mineral
black 325. Factorial design of experiments will be performed at three dust concentration
levels. At each level, there will be two factors (wind velocity and sensor direction) with three
levels for each factor. Three levels of wind velocities are, 0.5 m/s, 1.5 m/s, and 3.0 m/s. The
sensor direction will be towards the stream, perpendicular to the stream, and opposite to the
stream. The response of each low-cost sensor will be compared with the reference monitor.
This study investigates the factors commonly encountered in an underground mining environ
ment, which provides valuable information on the feasibility of using these low-cost sensors in
the mining industry where coal mine dust is a concern.
2 EXPERIMENTAL METHODS
In this study, four low-cost PM sensors were evaluated. These sensors include the Gaslab,
CM-505 (Farrar, n.d.) referred to as the GLS multi-gas sensors, Plantower PMS5003 (LTD,
2023), referred to as PMS low-cost PM sensor SPS30 sensor (Sensiron, n.d.), and the Airtrek
sensor(“AerosolWorks LLC,” n.d.) . To observe a better precision response between them,
two units of each type of sensor, except Airtrek, were evaluated. The Gaslab sensors can meas
ure PM2.5 and PM10 together with other gases like carbon dioxide (CO2), carbon monoxide
(CO), and Oxygen (O2) concentration. The second type PMS is affordable, lightweight, and
gives real-time concentration. These PMS sensors characterize PM by size into PM1 PM2.5
and PM10. SPS 30 sensor can read PM 1.0, 2.5,4 and 10.
An MSHA-approved instrument called a personal dust monitor (PDM 3700) is used as
a reference monitor in this study (Volkwein et al., 2006b). This instrument uses the principle
of tapered element oscillation microbalance (TEOM). The cut-off diameter of the coal mine
dust particle which can enter this instrument is 4.0 µm. The wind tunnel used in this study is
a custom-built wind tunnel made with metal frames and acrylic glass panels. The wind tunnel
has a particle generator made up of an air blower connected to a venturi feeder. A vibratory
feeder is used to feed the coal to the venturi feeder. The particle generation setup is shown in
Figure 1.
The coal mine dust used for this experiment is the Keystone mineral black 325 with a density
of 1,220 kg/m3 and particle sizes in the range of 0.04 µm to 35.00 µm(Kumar, 2018). Before the
injection of dust particles PDM was placed outside the tunnel, and their inlet nozzles along with
two PMS, two SPS30, two GLS, and one Airtrek sensor were placed inside the tunnel.
A full factorial experimental design was used to determine the accuracy of the low-cost sen
sors for measuring coal mine dust. Three concentration levels of coal mine dust,0-1.0, 1.0-2.0,
and 2.0-3.0 mg/m3 were used. A total of 27 experiments were performed with a set of 9
276
experiments for each concentration of coal. Each concentration of dust was evaluated in
one day. On each concentration level, the effect of three levels of wind velocity 0.5, 1.5, and
3.0 m/s and three directions of sensors (towards the stream, perpendicular to the stream, and
opposite to the stream) were tested. The response of low-cost sensors will be compared to ta
reference monitors i.e., PDM 3700. It will also be evaluated how these sensors respond to coal
mine dust, and the sensor direction and wind velocity impact the linearity of these sensors
with the reference monitor.
3 RESULTS
The objective of this study is to evaluate the impact of sensor direction and wind velocity on
the performance of low-cost PM sensors. A pair of each low-cost PM sensor i.e., PMS5003,
SPS30, Gaslab, and one Airtrek sensor was tested during this study. The PDM utilizes the
BGI HD cyclone with a precise D50 cutoff point of 4.37 µm.
277
Figure 2. Correlation between PDM reference Monitor and Gaslab Sensor at (A) 2.0-3.0 concentration,
(B) 1.0-2.0 concentration, (C) 0-0.1 Concentration.
Sensor direction has a significant effect on the linearity of SPS30 sensor response within
their effective range of concentration (0-1 mg/m3). At a wind velocity of 3.0 m/sec inside the
tunnel, the sensor showed moderate linearity with the reference monitor which is evident from
its R-squared value of 0.71. The R-squared values decreased to 0.36 and 0.18 when the sensor
direction was changed to perpendicular and opposite respectively as shown in Figure 5(a).
A similar trend was observed for the sensor response for sensor direction at a wind velocity of
1.5 m/s and 0.5 m/s and it can be seen in Figure 5 b & c. A p-value of 0.01 shows the statistical
significance of sensor direction and sensor response.
Figure 3. Correlation between PDM reference Monitor and Airtrek Sensor at (A) 2.0-3.0 mg/m3 con
centration, (B) 1.0-2.0 mg/m3 concentration, (C) 0-0.1 mg/m3 concentration.
278
Figure 4. The correlation between the PDM reference monitor and SPS30 sensor at (a) Concentration
1.0-2.0 mg/m3 and (b) Concentration 2.0-3.0 mg/m3.
The effect of the sensor direction can be explained in two ways. A moderately high R-squared
value for sensor direction towards the stream shows that particles have more chances to enter
the sensor inlet and be detected. On contrary, when the direction of the sensor is perpendicular
or opposite to the stream, the coal particles have lesser chances to enter the inlet. This is espe
cially true when the direction is opposite to the stream. Inferior R-square values for perpendicu
lar and opposite sensor directions also indicate that the fan at the sensor inlet is not very
powerful to draw the same number of particles when facing the stream. This could be the result
of low to bad linearity of the sensor while changing the direction even when the concentration
in the tunnel is within the same range as the sensor direction toward the stream.
Figure 5. Correlation between PDM reference monitor and SPS30 sensor at concentration 0-1.0 mg/m3
(a) velocity 3.0 m/sec, (b) velocity 1.5 m/sec, (c) velocity 0.5 m/sec.
Sensor direction also has a significant impact on the linearity of PMS sensors. As can be
seen from the plots in Figure 6, PMS sensors showed a high correlation with the reference
monitors than any other sensor in this study when the inlet direction of the sensor was
towards the stream. At an actual concentration between 0-1.0 mg/cm3 and a wind velocity of
279
3.0 m/s, the PMS sensor showed very high linearity with an R-squared value of 0.91 Figure 6.
R-square values of 0.85 and 0.8 were obtained when the sensor direction was toward the
stream at wind velocities of 1.5 and 0.5 m/s in the tunnel respectively. These high correlation
values agree with previous studies which also obtained high linearity values for PMS sensors
(Amoah et al., 2023; Sayahi et al., 2019; Wang et al., 2015). When the direction of the sensor
is perpendicular to the stream, a considerable drop in the linearity is observed at all three
wind velocities. An R-square value of 0.31, 0.26, and 0.21 is observed in Figure 6 (a, b, & c).
Similarly, for the sensor direction opposite to the air stream, the PMS sensor showed the
lowest correlation. In Figure 6, R-square values of 0.008, 0.0123, and 0.08 were recorded for
wind velocities of 3.0, 1.5, and 0.5 m/s. Statistical analysis confirmed the significance of sensor
direction on sensor response with a P value of 0.01.
The same trend for sensor direction was observed in the response of PMS sensors with
a higher concentration between 1.0-2.0 mg/m3 as the previous concentration range which
can be seen in Figure 7 a, b & c. PMS sensors showed an interesting trend for the linearity at
higher input concentrations in the 2.0-3.0 mg/cm3 range. At high wind velocity at 3.0 m/s,
the PMS sensor showed a moderately high correlation with, an R-square value of 0.75, with
the reference monitor when the sensor inlet direction is toward the stream Figure 7(a).
Unlike the previous concentrations, the PMS sensor showed very low linearity with the ref
erence monitor at an inlet velocity of 1.5 and 0.5 m/s. At a wind velocity of 0.5 m/s, when
the sensor direction was toward the stream, an R-square value of 0.38 was recorded (see
Figure 7(c)). At a wind velocity of 1.5 m/s, an R-square value of 0.3 was recorded. Figure 7
(b). These low values of linearity when the inlet sensor direction is towards the stream are
due to the coincidence error. It can be observed in Figure 7 that the actual concentration is
more than 3.0.
Figure 6. Correlation between PDM reference monitor and PMS5003 sensor at concentration 0-1.0 mg/
m3 (a) velocity 3.0 m/sec, (b) velocity 1.5 m/sec, (c) velocity 0.5 m/sec.
At higher concentrations when there are many particles in the sensing area, the sensor reads
multiple particles as one large and heavy particle which results in an incorrect interpretation
of the mass concentration as compared to the reference monitor.
280
3.2 Wind velocity
Wind velocity showed a significant impact on the performance of the low-cost sensors used in this
study. In Figure 5 (a), (b) & (c) it can be seen that velocity does impact the sensor response. At
a concentration of 1.0 mg/cm3, an R-squared value of 0.71, 0.66, and 0.62 were recorded at
a velocity is 3.0, 1.5, and 0.5 m/s. Similarly, the same trend was observed in the perpendicular
direction. PMS sensors also followed the same trend, but their correlation was better than the
SPS30 sensor. A P-value of 0.1 shows that the impact of the velocity on the linearity of the sensor
is significant. Similarly, in Figure 7 and Figure 8, it can be seen that when the velocity is highest
at 3.0 m/s and the sensor direction toward the air stream, a better correlation was observed.
A better correlation at the highest velocities indicates that the fan inlet velocity of the sensor is
close to 3.0 m/. This can be explained by the concept of isokinetic sampling (Wilcox, 1956). For
example, if the wind tunnel velocity is as same as the fan inlet velocity (3 m/sec), then particles
will just follow their streamlines and enter the sensor inlet without much deviation. In the case of
a wind tunnel velocity of 1.5 and 0.5 m/sec, this is different than the velocity at the fan inlet. In
this case, the particles will not follow the streamlines and will deviate from their path at the fan
inlet which results in incorrect sampling entering the sensor and hence lower sensor response.
4 PRECISION
The results of the precision test are displayed in Figure 9 where the response of pair of each
kind of PMS and Sps30 was cpmpared to each other. It can be seen from the Figure 9 that
sps30 sensors have shown the highest degree of precision with an R-squared value of 0.94. The
manufacturer datasheet says that sps30 sensors can effectively read the concentration up to
1.0 mg/m3. These sensors can report higher concentrations, but these concentrations are
imprecise. After sps30, the PMS5003 has shown the second-highest precision.
Figure 7. Correlation between PDM reference monitor and PMS5003 sensor at concentration
1.0-2.0 mg/m3 (a) velocity 3.0 m/s, (b) velocity 1.5 m/s, (c) velocity 0.5 m/s.
Unlike sps30 sensors, these sensors did not experience excessive peak,which means that
sometimes these snosrs reported concetrtaion as high as 15 mg/m3,and reported the concentra
tion in good agreement with the reference monitors. For both SPS30 and PMS5003 sensors,
the precision is decreasing with increasing concentration.
281
Figure 8. Correlation between PDM reference monitor and PMS5003 sensor at concentration
2.0-3.0 mg/m3 (a) velocity 3.0 m/s, (b) velocity 1.5 m/s, (c) velocity 0.5 m/s.
5 CONCLUSION
The aim of this study was to evaluate the impact of sensor direction and wind velocity on the
performance of low-cost PM sensors. It was found that PM 5003 sensor showed the best lin
earity as compared to all other sensors evaluated in this study. The impact of three sensor dir
ections, with three wind velocities and three different concentration levels was evaluated.
Figure 9. Intra-mode correlation between two models of each sensors type (a) PMS5003 (b) SPS30.
Sensor direction has a significant impact on the sensor response. The R-squared values
drastically changed for different sensor directions. The sensors face towards, perpendicular,
and opposite to the wind direction. When the sensor direction was towards the stream, PMS
5003 sensor showed an excellent linearity which is evident from an R-squared value of as high
as 0.91, 0.85 and 0.8 at different wind velocity levels. SPS-30 sensors also showed a very good
response when the sensor direction is toward the stream with an R-squared value of 0.71, 0.66
and 0.62. when the sensor direction is changed to perpendicular and opposite to the stream,
a significant drop was observed in the R-squared value which showed the impact of sensor
direction. It is found that PMS sensor can be utilized to measure the coal mine dust accurately
upto 3 mg/m3.
The impact of the wind velocity is statistically significant on the linearity of these Low-cost
PM sensors. Apparently, R-squared value at different levels of wind velocity did not show
282
a huge difference. We performed a statistical analysis and did the effect test and it turned out
that wind velocity does have a significant effect on the linearity of these low-cost PM sensors
while measuring coal mine dust. It is evident from the F-ratio value <0.05.
During this study, it was found that SPS-30 sensors can only read the concentration effect
ively with 0-1.0 mg/m3 concentration range. Even though these sensors can measure high con
centration, but the accuracy of these sensors is compromised over a concentration of
1.0 mg/m3.
Airtrek and Gaslab sensors have very low linearity with the reference monitors at any con
centration level. Due this this reason the author decided not the evaluate these sensors further
for other factors.
It is concluded from this research that PMS 5003 sensor can be utilized for accurate dust
monitoring up to a concentration level of 3 mg/m3, and SPS-30 sensor can used for coal mine
dust monitoring at lower concentration level up to 1.0 mg/m3.
REFERENCES
Abbasi, B., Wang, X., Chow, J.C., Watson, J.G., Peik, B., Nasiri, V., Riemenschnitter, K.B.,
Elahifard, M., 2021. Review of respirable coal mine dust characterization for mass concentration, size
distribution and chemical composition. Minerals 11, 1–36.
AerosolWorks LLC [WWW Document], n.d. URL https://aerosolworks.com/ (accessed 7.3.23).
Alfano, B., Barretta, L., Giudice, A. Del, De Vito, S., Francia, G. Di, Esposito, E., Formisano, F.,
Massera, E., Miglietta, M.L., Polichetti, T., 2020. A review of low-cost particulate matter sensors
from the developers’ perspectives. Sensors (Switzerland) 20, 1–56.
Amoah, N.A., Xu, G., Kumar, A.R., Wang, Y., 2023. Calibration of low-cost particulate matter sensors
for coal dust monitoring. Sci. Total Environ. 859, 160336.
Blackley, David J. Laura E., Reynolds Connie Short, Ron Carson, Eileen Storey, Cara N. Halldin, A.S.
L., 2022. Progressive Massive Fibrosis in Coal Miners From 3 Clinics in Virginia 2017–2018.
Bulot, F.M.J., Johnston, S.J., Basford, P.J., Easton, N.H.C., Apetroaie-Cristea, M., Foster, G.L.,
Morris, A.K.R., Cox, S.J., Loxham, M., 2019. Long-term field comparison of multiple low-cost par
ticulate matter sensors in an outdoor urban environment. Sci. Rep. 9, 1–13.
Christensen, N.K., Albrechtsen, O., Fanger, P.O., Trzeciakiewicz, A., 1984. Air movement and draught.
In: Indoor Air. Proceedings of the 3rd International Conference on Indoor Air Quality and Climate.
pp. 301–308.
Chung, A., Chang, D.P.Y., Kleeman, M.J., Perry, K.D., Cahill, T.A., Dutcher, D., McDougall, E.M.,
Stroud, K., 2001. Comparison of real-time instruments used to monitor airborne particulate matter.
J. Air Waste Manag. Assoc. 51, 109–120.
dos S Antao, V.C., Petsonk, E.L., Sokolow, L.Z., Wolfe, A.L., Pinheiro, G.A., Hale, J.M., Attfield, M.
D., 2005. Rapidly progressive coal workers’ pneumoconiosis in the United States: geographic cluster
ing and other factors. Occup. Environ. Med. 62, 670–674.
Fanger, P.O., 1977. Discomfort due to air velocities in spaces. In: Proc. of Meeting of Commission B1,
B2, E1, of Instit. Refrig. pp. 289–296.
Farrar, P., n.d. No TitleGASLAB PLUS CM-500 MULTI-GAS DETECTOR SERIES [WWW Docu
ment]. URL https://www.co2meter.com/collections/cm-500-gaslab-plus-collection (accessed 7.3.23).
Fobelets, A.P.R., 1987. Subjective human response to low-level air currents and asymmetric radiation.
ASHRAE Trans. 93, 497–523.
Ghamari, M., Soltanpur, C., Rangel, P., Groves, W.A., Kecojevic, V., 2022. Laboratory and field evalu
ation of three low-cost particulate matter sensors. IET Wirel. Sens. Syst. 12, 21–32.
Halterman, A., Sousan, S., Peters, T.M., 2018. Comparison of respirable mass concentrations measured
by a personal dust monitor and a personal DataRAM to gravimetric measurements. Ann. Work
Expo. Heal. 62, 62–71.
Ishtiaq, M., Jehan, N., Khan, S.A., Muhammad, S., Saddique, U., Iftikhar, B., Zahidullah, 2018. Poten
tial harmful elements in coal dust and human health risk assessment near the mining areas in Cherat,
Pakistan. Environ. Sci. Pollut. Res. 25, 14666–14673.
Janisko, S., Noll, J.D., 2010. Field Evaluation of Diesel Particulate Matter Using Portable Elemental
Carbon Monitors. 13th United States/North Am. Mine Vent. Symp. 1–6.
Kumar, A.R., 2018. Dust Control Examination using Computational Fluid Dynamics Modeling and
Laboratory Testing of Vortecone and Impingement Screen Filters 49.
283
Liu, X., Jayaratne, R., Thai, P., Kuhn, T., Zing, I., Christensen, B., Lamont, R., Dunbabin, M., Zhu, S.,
Gao, J., Wainwright, D., Neale, D., Kan, R., Kirkwood, J., Morawska, L., 2020. Low-cost sensors as
an alternative for long-term air quality monitoring. Environ. Res. 185, 109438.
LTD, P., 2023. PMS5003 Particulate Matter Sensor with Cable [WWW Document]. URL https://shop.
pimoroni.com/products/pms5003-particulate-matter-sensor-with-cable (accessed 7.3.23).
Masson, N., Piedrahita, R., Hannigan, M., 2015. Approach for quantification of metal oxide type semi
conductor gas sensors used for ambient air quality monitoring. Sensors Actuators, B Chem.
Mukherjee, A., Stanton, L.G., Graham, A.R., Roberts, P.T., 2017. Assessing the utility of low-cost par
ticulate matter sensors over a 12-week period in the Cuyama valley of California. Sensors (Switzer
land) 17.
Potera, C., 2019. Black Lung Disease Resurges in Appalachian Coal Miners. Am. J. Nurs. 119, 14.
Roghanchi, P., Kocsis, K.C., Sunkpal, M., 2016. Sensitivity analysis of the effect of airflow velocity on
the thermal comfort in underground mines. J. Sustain. Min.
Sayahi, T., Kaufman, D., Becnel, T., Kaur, K., Butterfield, A.E., Collingwood, S., Zhang, Y.,
Gaillardon, P.E., Kelly, K.E., 2019. Development of a calibration chamber to evaluate the perform
ance of low-cost particulate matter sensors. Environ. Pollut. 255, 1–22.
Sellaro, R., Sarver, E., Baxter, D., 2015. A standard characterization methodology for respirable coal
mine dust using SEM-EDX. Resources 4, 939–957.
Sensiron, n.d. SPS30 [WWW Document]. URL https://sensirion.com/products/catalog/SPS30/ (accessed
7.3.23).
Takiff, L., Aiken, G., 2010. A Real-Time, Wearable Elemental Carbon Monitor for Use in Underground
Mines. 13th United States/North Am. Mine Vent. Symp. 137–141.
van den Bossche, M., Rose, N.T., De Wekker, S.F.J., 2017. Potential of a low-cost gas sensor for atmos
pheric methane monitoring. Sensors Actuators, B Chem.
Volkwein, J.C., Vinson, R.P., McWilliams, L.J., Tuchman, D.P., Mischler, S.E., 2004. Performance of
a New Personal Respirable Dust Monitor for Mine Use. Rep. Investig. 9663, Natl. Inst. Occup. Saf.
Heal. Pittsburgh Res. Lab.
Volkwein, J.C., Vinson, R.P., Page, S.J., McWilliams, L.J., Joy, G.J., Mischler, S.E., Tuchman, D.P.,
2006a. Laboratory and Field Performance of a Continuously Measuring Personal Respirable Dust
Monitor. Usbm Ri 9669.
Volkwein, J.C., Vinson, R.P., Page, S.J., McWilliams, L.J., Joy, G.J., Mischler, S.E., Tuchman, D.P.,
2006b. Laboratory and field performance of a continuously measuring personal respirable dust
monitor.
Wang, Y., Li, J., Jing, H., Zhang, Q., Jiang, J., Biswas, P., 2015. Laboratory Evaluation and Calibration
of Three Low-Cost Particle Sensors for Particulate Matter Measurement. Aerosol Sci. Technol. 49,
1063–1077.
Wang, Z., Calderón, L., Patton, A.P., Sorensen Allacci, M.A., Senick, J., Wener, R., Andrews, C.J.,
Mainelis, G., 2016. Comparison of real-time instruments and gravimetric method when measuring
particulate matter in a residential building. J. Air Waste Manag. Assoc. 66, 1109–1120.
Wilcox, J.D., 1956. Isokinetic flow and sampling. J. Air Pollut. Control Assoc. 5, 226–245.
Zhang, H., Nie, W., Liang, Y., Chen, J., Peng, H., 2021a. Development and performance detection of
higher precision optical sensor for coal dust concentration measurement based on Mie scattering
theory. Opt. Lasers Eng. 144.
Zhang, H., Nie, W., Liang, Y., Chen, J., Peng, H., 2021b. Development and performance detection of
higher precision optical sensor for coal dust concentration measurement based on Mie scattering
theory. Opt. Lasers Eng. 144.
Zhang, R., Li, G., Fan, J., Wu, D.L., Molina, M.J., 2007. Intensification of Pacific storm track linked to
Asian pollution. Proc. Natl. Acad. Sci. 104, 5295–5299.
Zikova, N., Masiol, M., Chalupa, D.C., Rich, D.Q., Ferro, A.R., Hopke, P.K., 2017. Estimating hourly
concentrations of PM2.5 across a metropolitan area using low-cost particle monitors. Sensors (Switz
erland) 17, 1–19.
284
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
A.R. Kumar
Department of Energy and Mineral Engineering, Pennsylvania State University, University Park,
Pennsylvania, USA
ABSTRACT: Prolonged exposure to high concentrations of respirable coal mine dust causes
coal workers’ pneumoconiosis and silicosis. Underground coal mine roof bolter operators are
more prone to these ailments due to their elevated exposure to dust. The canopy air curtain
(CAC) was developed by NIOSH to alleviate their exposure. The CAC supplies filtered air
over their breathing zone that dilutes the high coal dust concentrations and provides for an
impenetrable air curtain. Many studies attempted to improve the CAC design, but field tests
show variable dust control efficiencies. The non-uniform airflow distribution across the
plenum and ineffective perimeter flow are the major impediments. This study, therefore, sys
tematically redesigns the CAC for an optimized flow distribution. The Simplex Evolutionary
Operational (EVOP) optimization algorithm was applied to the geometrical parameters. Com
putational fluid dynamics (CFD) simulations were run at each iteration of the algorithm. The
combination of these methodologies led to an enhanced uniformity of the airflow distribution
across the plenum to achieve the best possible dust suppression efficiency. A lab experiment
using a physical model was used to validate the computer models and validate the efficacy of
this parametrically designed CAC.
1 INTRODUCTION
Extended exposure to elevated respirable coal mine dust concentrations causes coal workers’
pneumoconiosis (CWP), a potentially fatal lung disease with no known conclusive cure (Blackley
et al., 2018; Laney & Attfield, 2010; Mine Safety and Health Administration, 2020). This has
also been linked to the onset of silicosis if the silica content in the coal dust is high (Cohen et al.,
2016; Sato et al., 2018). Underground coal mine roof bolter operators are prone to a higher risk
of both excessive coal dust and silica exposure due to their proximity to the source (Mine Safety
and Health Administration, 2020; Pollock et al., 2009; Reed et al., 2020a). Miners’ position
downwind of the continuous miner, improper and inadequate ventilation, and dust escaping
from the drill shrouds due to faulty or unmaintained dust collectors add to the severity. Occa
sionally, roof bolting downwind to the continuous miner can substantially increase bolter
operators’ coal dust exposure level up to 7.0 mg/m3 (Listak & Beck, 2012a). These levels exceed
the maximum concentrations allowed by MSHA outlined in the New Dust Rules of 2014
(MSHA, 2014). The rule requires the mines to maintain a time-weighted average (TWA)
respirable coal dust concentration at or below 1.5 mg/m3 at the shift end. The permissible
limit is further reduced if the mine’s respirable dust contains more than 5% silica.
DOI: 10.1201/9781003429241-30
285
To reduce dust exposure levels of coal mine roof bolters, a canopy air curtain (CAC) was
developed by the National Institute of Occupational Safety and Health (NIOSH). The CAC is
mounted below the roof bolter’s protective canopy. The CAC uses an exhaust fan to draw air
from the mine atmosphere through a high-efficiency particulate air (HEPA) filter. The filtered air
is then delivered through tubing to the plenums which are located underneath the canopy of the
bolting machine. The plenums supply filtered air over the breathing zone of the operator during
the bolting operations. This offers a level of coal dust protection without affecting operator
safety, comfort, and mobility (Listak & Beck, 2012a). Laboratory assessment carried out by
NIOSH on their initial design yielded an efficiency of 62.0 % (Goodman & Organiscak, 2001).
Many improvements were made in the CAC since its invention. A new prototype with an
improved design covered the entire operating canopy area (Listak & Beck, 2012b). It was then
optimized to provide uniform airflow over the entire plenum, as proven by NIOSH for
enhanced efficiency, using computer simulations and a trial-and-error process. Its dust reduc
tion efficacy exceeded 67.0 % for entry velocities up to 0.61 m/s (120 fpm). Another design
was then made with a much thinner profile with slotted openings to provide perimeter airflow
(Reed et al., 2017). However, NIOSH gallery tests indicated its low efficiency of about 24 %.
Later, a 3rd generation CAC was designed with a higher perimeter air velocity to prevent con
taminated airflow from infiltrating the protected zone. Although its lab test dust reduction
has improved up to 49.3 % (Reed et al., 2019), the field test has shown variable efficiencies.
This necessitates further design improvements (Reed et al., 2020b). Therefore, it is important
to ensure an even distribution of airflow across the plenum at a velocity higher than 0.51 m/s
(100 fpm) to protect the entire canopy area.
The objective of this research is to optimize the design of the CAC using a two-level mani
fold. The manifold consists of the main flow header and several outlets. However, it is challen
ging for a typical manifold with a constant cross-sectional area header to achieve uniform flow
distribution. This is due to the static pressure build-up toward the end. Studies have shown
that the manifold with a tapered longitudinal section can achieve nearly uniform flow distribu
tion under the effects of the increased resistance but decreasing cross-section area (Hassan
et al., 2014). Air is firstly distributed horizontally evenly using the main manifold. Its outlet is
connected to a series of sub-manifolds that uniformly distribute air downward. Computational
Fluid Dynamics (CFD) simulations are carried out using the existing peripheral design used in
a previous NIOSH study but with a redesigned internal structure to incorporate the two-level
manifold system. A CFD-based parametric study was performed to determine the optimum
manifold design. The Simplex Evolutionary Operation (EVOP) optimization algorithm was
used to optimize the CAC geometrical features. Each iteration in the optimization process
required a CFD simulation. A CAC prototype was then built based on this optimum design
configuration and tested in laboratory experiments to validate the CFD models.
2 MODEL DESCRIPTION
An efficient CAC must discharge a high volumetric flow of air through the outlets toward the
operator’s breathing zone. The CAC structure must also allow for an elevated and uniform air
velocity at the outlets resulting in an air-curtain formation that protects the operator from
exposure to particulate matter. The following sections describe the geometry and associated
CFD models.
2.1 Geometry
This research builds on a J.H. Fletcher and Co. CAC model developed by NIOSH. Figure 1
illustrates the peripheral geometry used in this study (Listak & Beck, 2012b). Previous designs
have added various designs of baffles, inflow vanes, and flow straighteners to the CAC
internal structure. In the existing design, a single angled plate with an adjustable louver is
installed inside the plenum to regulate the airflow distribution. This current design has an
uneven flow of air beneath the plenum with approximately 30.0 % of the plenum area being
286
unprotected. Areas with low to no airflow are the trapezoidal region as well as the transition
zone between the square and the trapezoidal region. The currently used third-generation CAC
only reduces coal dust exposure by 34.6 % - 49.3 % (Reed et al., 2019).
The schematic of the manifold structure used in this study is demonstrated in Figure 2. The
internal design of our model consists of a two-level manifold system as shown in Figure 3. The
outlets of the first-level manifolds serve as the inlets to the second-level manifolds. The first level
manifold connected to the fan has a cross-sectional dimension of 10.2 cm × 4.0 cm at its entry
and is tapered at an angle at two points along its length. The fan duct measures 10.2 cm in diam
eter and is considered the inlet for CFD analysis. The CAC is required to have a height restriction
of 5.1 cm (2 in.) to provide head clearance for operators. This restricts the overall entry height of
the first level manifold of the CAC to approximately 5.0 cm. Air flows through the first level
manifold and distributes into the plenum area containing the second level manifolds. Internally,
the plenum area is made up of branches whose parameters are designed in this study. The heights
of these manifolds were fixed at 4.0 cm due to the height restrictions for the CAC. This also maxi
mized the use of the available space. The widths of these second-level manifolds are, however,
included in the optimization process to be determined. The design constraints restricted the
number of second-level manifolds to ten. The separation distance was also optimized in this
study. These second-level manifolds are linearly tapered at an angle along the header for con
trolled airflow distribution. Their ends’ dimensions were also optimized to generate the optimum
tapering angle. The use of the two-level manifold gives a significantly higher control over airflow
distribution across the plenum by controlling the quantity of air being sent into each branch.
Figure 1. Button-up view of the original geometry of the CAC showing the perforated plate.
287
Figure 2. Schematic of (a) first level manifold and (b) second-level manifold with labels of parameters
to be optimized.
288
Once the above seven models were developed, the dimensionless nonuniformity coefficient was
calculated for each case and are ranked in decreasing order. The objective of this exercise was to
obtain the case with the lowest non-uniformity (elaborated described in section 2.3). The coordin
ate of this worst response was assigned the ‘w’ variable and the next to worst as the ‘n’ variable.
These denote the coordinate matrix of all responses that are better than ‘w’ by ‘c’. The inflection
point ‘r’ is calculated using Equation 3. Point ‘r’ is the next model that needs to be simulated. This
is expected to move away in the opposite direction from the one with the worst non-uniformity.
This case will replace the previous worst case, and the procedure is repeated. If the inflection point
is outside the constraints of feasible dimensions of any parameter, it is treated as having the worst
non-uniformity. This will force the procedure to reflect cases with feasible parameters.
289
obtained by applying Newton’s law of motion applied to the discretized fluid element. The
equations represent the conservation of mass (Equation 7) and momentum (Equation 8). The
equations were solved iteratively using the Fluent solver until convergence was achieved.
A threshold residual value of 0.0001 in velocity components the and mass flow rate was set.
where ρ is the density of the simulation fluid (kg/m3), t is the time (seconds), and ~
v is the vel
ocity vector.
τ is the viscous stress tensor (N) given by Equation 4 below for a Newtonian fluid, ~
~
where ~ b is
body force and μ is molecular viscosity coefficient.
290
Table 4. Calculation of mesh convergence indices.
Parameters Coarse Mesh Medium Mesh Fine Mesh
3 RESULTS
There are six design factors covering both the primary and secondary level manifolds of the
CAC. The results obtained from the simulations were used in combination with the Simplex
EVOP optimization algorithm to determine the optimum configuration of these design fac
tors. The results are discussed in the sections that follow.
291
areas with lower air velocities below 1.0 m/s. As can be seen from Figure 7, airflow from the fan
forces its way into the first-level manifold of the CAC which is uniformly distributed among
the second-level manifolds. The optimized tapered design of the first-level manifold in this study
confirms that the optimum taper design generated nearly no static pressure build-up within any
part of the manifold causing a perfectly uniform efflux throughout the second-level manifolds.
Airflow contours of the CAC outlets shown in Figure 5 moreover indicate that there is a high
degree of improved outflow uniformity across the plenum. This shows a significant improve
ment in the existing third-generation CAC model. However, a small region of relatively lower
flow quantities can still be observed towards the center of the CAC. This, however, is of no sig
nificant influence on performance and the overall uniformity and performance of the model as
the results give a standard deviation of 1.03 m/s. Moreover, this will contribute to enforcing that
polluted airflows do not penetrate the perimeter airflow region. Analysis of the quantity vectors
of this simulation moreover showed that there was no backflow of air into the CAC.
292
Figure 7. The contour of internal airflow pattern in CAC.
4 CONCLUSION
Roof-bolters operators are exposed to elevated dust concentrations that make them prone to the
onset of irreversible diseases such as CWP and silicosis. To control their exposure levels, the
canopy air curtain was designed by NIOSH to deliver filtered air on the roof bolter operators
breathing zone to dilute coal dust within their working area. Additionally, this design and subse
quent improvements also had a provision for perimeter flow. A high-velocity air jet emancipating
from the slits downstream serves as the air curtain. This is impassable to dust particles present in
the ventilation air stream lowering his exposure to respirable dust. This paper summarizes the
parametric studies of the geometry of the CAC aimed at allowing near-equal airflow through all
the branches. The authors investigated the optimum design parameters for the proposed two-level
manifold-based CAC which will achieve the best possible uniformity and best efficiency.
The original perimeter profile of the CAC was used in this study. A two-level manifold system
was introduced into the CAC peripheral design to carry air toward the operator’s breathing zone.
Steady-state CFD models were developed to investigate the impact of four major parameters on
the airflows. This study focused on optimizing the uniformity of the CAC by tapering the two-
level manifolds to redistribute airflow across their outlets. Systematically adjusting the tapering
angle resulted in the desired airflow redistribution. A parameter called the non-uniformity index,
Φ, was calculated for all the models generated. A low Φ value indicates an even volumetric airflow
rate through all the outlets. The design which resulted in the least Φ was determined to be the
optimum design of the two-level manifold-based CAC. This design is critical for the effective dilu
tion of dust-laden air close to the breathing zone of the miner.
293
Analysis of all the parameters showed that an inlet size (D1) dimension of 5.0 cm × 5.0 cm,
tapered header end size (D2) dimension of 5.0 cm × 0.2 cm, outlet size (d) of 1.27 cm (1/2 in.)
in diameter, outlet separation of 6.35 mm (1/4 in.) and a header length of 56.52 cm was the
optimum configuration for the second level manifolds. Additional studies show that an inlet
size (D1) of 10.2 × 6.0 cm, tapered header end size (D2) of 2.0 × 6.0 cm, header length (L) of
101.6 cm, outlet size (d) of 5.0 × 5.0 cm and outlet separation (S) of 0.5 cm was the optimum
configuration for the first level manifold. The combination of these parameters yields the best
possible design to achieve the most uniform airflow distribution across the plenum of a CAC.
Further studies are focusing on building a physical model based on these parameters to be
evaluated in a mining setup to validate the CFD results and to determine the efficiency of this
model. Future studies will focus on optimizing additional parameters to investigate the pos
sible improvement in the uniformities in airflow distribution.
REFERENCES
Blackley, D. J., Halldin, C. N., & Scott Laney, A. (2018). Continued increase in prevalence of coal work
ers’ pneumoconiosis in the United States, 1970-2017. In American Journal of Public Health (Vol. 108,
Issue 9, pp. 1220–1222). American Public Health Association Inc. https://doi.org/10.2105/
AJPH.2018.304517
Cohen, R. A., Petsonk, E. L., Rose, C., Young, B., Regier, M., Najmuddin, A., Abraham, J. L.,
Churg, A., & Green, F. H. Y. (2016). Lung pathology in U.S. coal workers with rapidly progressive
pneumoconiosis implicates silica and silicates. American Journal of Respiratory and Critical Care Medi
cine, 193(6), 673–680. https://doi.org/10.1164/rccm.201505-1014OC
Goodman, G. V. R., & Organiscak, J. A. (2001). Laboratory Evaluation of a Canopy Air Curtain for
Controlling Occupational Exposures of Roof Bolters. Seventh International Mine Ventilation Congress,
Proceeding, 299–305.
Hassan, J. M., Mohamed, T. A., Mohammed, W. S., & Alawee, W. H. (2014). Modeling the Uniformity
of Manifold with Various Configurations. Journal of Fluids, 2014(April), 1–8. https://doi.org/10.1155/
2014/325259
Laney, A. S., & Attfield, M. D. (2010). Coal workers’ pneumoconiosis and progressive massive fibrosis
are increasingly more prevalent among workers in small underground coal mines in the United States.
Occupational and Environmental Medicine, 67(6), 428–431. https://doi.org/10.1136/oem.2009.050757
Listak, J. M., & Beck, T. W. (2012a). Development of a canopy air curtain to reduce roof bolters’ dust
exposure. Mining Engineering, 64(7), 72–79.
Mine Safety and Health Administration. (2020). Open Government Initiative Portal - MSHA. https://
arlweb.msha.gov/OpenGovernmentData/OGIMSHA.asp
MSHA. (2014). Lowering Miners’ Exposure to Respirable Coal Mine Dust, Including Continuous Personal
Dust Monitors (Vol. 79). http://www.cdc.gov/niosh/docs/2008-143/pdfs/
Pollock, D. E., Potts, J. D., & Joy, G. J. (2009). Investigation into dust and exposures and mining practices
in mines in the southern appalachian region. Society for Mining, Mettalurgy, and Exploration, Incorp
orated, 2009.
Reed, W. R., Joy, G. J., Kendall, B., Bailey, A., & Zheng, Y. (2017). Development of a roof bolter
canopy air curtain for respirable dust control. Mining Engineering, 69(1), 33–39. https://doi.org/
10.19150/me.7010
Reed, W. R., Joy, G. J., Shahan, M., Klima, S., & Ross, G. (2019). Laboratory results of a 3rd gener
ation roof bolter canopy air curtain for respirable coal mine dust control. International Journal of Coal
Science and Technology, 6(1), 15–26. https://doi.org/10.1007/s40789-019-0237-6
Reed, W. R., Shahan, M., Klima, S., Ross, G., Singh, K., Cross, R., & Grounds, T. (2020a). Field study
results of a 3rd generation roof bolter canopy air curtain for respirable coal mine dust control. Inter
national Journal of Coal Science and Technology, 7(1), 79–87. https://doi.org/10.1007/s40789-019-
00280-5
Reed, W. R., Shahan, M., Klima, S., Ross, G., Singh, K., Cross, R., & Grounds, T. (2020b). Field study
results of a 3rd generation roof bolter canopy air curtain for respirable coal mine dust control. Inter
national Journal of Coal Science and Technology, 7(1), 79–87. https://doi.org/10.1007/s40789-019-00280-5
Roache, P. J. (1997). Quantification of uncertainty in computational fluid dynamics. Annual Review of
Fluid Mechanics, 29, 123–160. https://doi.org/10.1146/annurev.fluid.29.1.123
Sato, T., Shimosato, T., & Klinman, D. M. (2018). Silicosis and lung cancer: Current perspectives. Lung
Cancer: Targets and Therapy, 9, 91–101. https://doi.org/10.2147/LCTT.S156376
294
Tong, J. C. K., Sparrow, E. M., & Abraham, J. P. (2009). Geometric strategies for attainment of identical
outflows through all of the exit ports of a distribution manifold in a manifold system. Applied Thermal
Engineering, 29(17–18), 3552–3560. https://doi.org/10.1016/j.applthermaleng.2009.06.010
Xu, G., Luxbacher, K. D., Ragab, S., Xu, J., & Ding, X. (2017). Computational fluid dynamics applied
to mining engineering: a review. International Journal of Mining, Reclamation and Environment, 31(4),
251–275. https://doi.org/10.1080/17480930.2016.1138570
Zheng, Y., & Reed, W. R. (2020). Effects of Roof Bolter Canopy Air Curtain on Airflow and Dust Disper
sion in an Entry Using Exhaust Curtain Ventilation. August. https://doi.org/10.1007/s42461-020-00294-7
295
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
K. Zychowski
College of Nursing, University of New Mexico-Health Sciences Center, Albuquerque, NM, USA
ABSTRACT: This study investigates the respirable coal mine dust (RCMD) characteristics
and toxicity based on the mine’s sources of dust particles. Dissolution experiments in simu
lated lung fluids (SLFs) and in-vitro response were conducted to determine the toxicity level
of samples collected from the coal seam, and the host rock (roof and floor) of an underground
coal mine. Dust characteristics were investigated using various dust characterization studies.
Significantly higher mineral and elemental contents were found in samples from the host rock,
suggesting higher toxicity. Si, Al, Fe, Cu, Ba, Sr, and Pb were found in dissolution experi
ments. Host rock displayed a much higher bio-accessibility than coal when exposed to artifi
cial lysosomal fluid (ALF), but not with Gamble’s solution (GS). Additionally, neutrophilic
and lung epithelial cells showed major affinity to activate inflammatory pathways when
exposed to host rock dust. Contrary, macrophage cells were more like to activate inflamma
tion from the coal, but with low expressions.
1 INTRODUCTION
Mineral coal continues nowadays to be important in industry and life because it still provides
a significant percentage of the energy consumed in the U.S. (EIA, 2022). RCMD toxicity
gained considerable relevance in coal mining in the last 50 years after encountering thousands
of coal mine workers suffering from lung diseases and even dying as a result of years of con
tinued exposure to RCMD.
Although efforts have been made to identify the characteristics and root causes of RCMD
toxicity, they remain unclear and are a subject of investigation (Rahimi et al, 2023).
The toxicity of RCMD has been related to several factors, such as particle size, shape, com
position, silica content, and mine practices, among others (Shekarian et al., 2021). The higher
rates of reported miners diagnosed with coal workers’ pneumoconiosis in the eastern region of
the U.S. have also suggested that the RCMD coming from this area may exhibit higher tox
icity, but until now, the reason why has not been clarified (Salinas et al., 2022). Some studies
have suggested that the seam thickness may influence due to the high silica in the air resulting
from cutting the rock strata of the floor and roof in the production areas (Sarver et al., 2019;
NAS, 2018).
Silica has been pointed out as an enhancer of toxicity in RCMD (Porter et al., 2002; Baron
et al., 2002). It is classified as a carcinogen material, so it is believed that higher silica content
in the RCMD may make it more toxic (Abbasi et al., 2021). Quartz is the most common crys
talline silica polymorph and the second most abundant mineral in the continental crust. It is
one of the most important minerals in forming rocks and soil and is often found in coal mines
and as a component of RCMD (Tarbuck & Lutgens, 2005).
DOI: 10.1201/9781003429241-31
296
Also, several studies have tried to determine whether the content of Fe in RCMD can
increase its toxicity, but until now, there have been findings indicating both that it is related
and not (Shangguan et al., 2022; Trechera et al., 2020)
Considering that different factors may influence the toxicity of RCMD, the primary research
focused on three approaches or variables to determine RCMD toxicity: geographic location,
source, and particle size. The geographic location influence was previously evaluated, and
results can be found in Salinas et al. (2022). In this work, samples taken directly from the coal
seam of five mines from the Appalachian and Rocky Mountains regions were characterized and
analyzed regarding their toxicity using dissolutions in simulated lung fluids and in-vitro experi
ments. Still, no trends were found indicating that the geographic location influenced the toxicity
based on these experiments (Salinas et al., 2022).
Regarding the source influence, the dust particles in the mine can come from different
sources, such as the host rock, rock dust, the coal seam, diesel particulate matter (DPM), and
inlet air, among others (Shekarian et al., 2022). All these particles coming from different
sources have different compositions, mineralogic and elemental wise. These differences may
make the dust particles display different toxicity when exposed to the lungs. The hypothesis is
that the RCMD generated by cutting host rock is more toxic than the dust coming from cut
ting coal seam. This variable is further analyzed in this document.
RCMD is composed of particles with a size below 10 µm (Brown et al., 2013). In this range,
particles of different sizes may expose different toxicities. Smaller particles may have a more
extensive specific area exposed to be in contact with the lungs when inhaled, which may help
the elements that compound the samples to dissolve more. So, the hypothesis regarding the
influence of the particle size is that smaller dust particles have higher metal/elemental dissol
ution than larger particles in simulated lung fluids. The evaluation of this last variable is pro
posed for future work.
Samples were collected from an underground coal mine located in the state of Pennsylvania,
belonging to the coal basin of the Appalachian region. Bulk samples were taken from the coal
seam and the host rock surrounding the seam, including one sample from the roof and
another from the floor of the production area. The information about the mine from where
the samples were collected and the sample IDs used in this document are shown in Table 1.
Location Pennsylvania
– CO: Coal seam
Samples collected – HR: Host roof
– HF: Host floor
Mine type Underground
Mine method Room and pillar
Coal rank Sub-bituminous
Seam height Low seam
297
10 µm fraction. The sample particle size distribution was verified using a scanning electron
microscope (SEM). The resulting samples were used for further analysis.
Characterization included X-ray diffraction (XRD) to identify the mineral components in
the samples, total microwave digestion to determine the elemental composition, and Fourier
Transformed Infrared spectrometry to identify the functional groups in the samples before
and after dissolution experiments.
The toxicity assessment was performed using dissolution in simulated lung fluids (SLFs) and
in-vitro analysis. The objective of the dissolution experiment was to analyze the elements dis
solving when the samples were in contact with two different SLFs after 24 hours. The SLFs
used were Gamble’s solution (GS) and artificial lysosomal fluid (ALF). GS and ALF are some
of the most used SLFs and have been widely used to assess the bio-accessibility of different
metals from urban and industrial airborne particulates. GS mimics the fluid present in the lung
interstitium and ALF mimics the acid fluid that engulfs particles in the lung macrophages.
When particles are inhaled and reach the alveolar region, can get in contact with these SLFs
giving them the chance of being totally or partially digested. To simulate the process and the
lung conditions, dust samples were put in contact with the SLFs for 24h, the experiment was
conducted in a dark room (to simulate the absence of light in the lungs) at a constant tempera
ture of 37°C (to simulate body temperature), and oxygen was added to the solution before start
ing. Additional SLFs models have been used before for other specific purposes like to model
interactions of particles with extracellular lung fluids (Mareno et al., 2019).
The in-vitro experiments looked at the inflammatory response of different human cells
involved in lung disease development (macrophages, epithelial, and neutrophilic) after dust
exposures. For each cell, 10 cytokines were analyzed. Cytokines were selected based on the
cell type exposed and their potential role in coal workers’ pneumoconiosis (CWP), occupa
tional lung injury, or inflammation, according to the literature (Kilpatrick & Harris, 2011;
Salinas et al., 2022).
3.1 Characterization
3.1.1 XRD
The XRD results are shown in Table 2 and Figure 1. The analysis showed that all the samples
have quartz in their mineral composition. As mentioned before, this mineral is very common
in rocks and coal mines. However, it is noticeable that the quartz peaks (e. g. the one at ~26.7°
in Figure 1) in the host rocks (HR and HF) are considerably more significant than the quartz
peak for the coal sample (CO), which means that the host rock has a much higher content of
quartz, behavior also observed in the peaks for muscovite and kaolinite, indicating a more sig
nificant amount of minerals in the host rock. These results are expected since the CO sample
might be compound mainly by organic material. This organic material is evidenced in the CO
pattern with the amorphous hump that can be observed at the beginning of the graph for this
sample. Regarding the rest of the mineral components, it was observed that the CO compos
ition was limited to kaolinite, pyrite, and calcite, while the HR and HF had in their compos
ition a variety of minerals that are usually found in the latest stages of rock disintegration
forming sedimentary rocks such as shales and limestone, which are typically found interlaying
the seams in coal deposits.
The following abbreviations represent the different minerals in Figure 1: Quartz (Qz), Kao
linite (Kln), Pyrite (Py), Muscovite (Mu), Siderite (Sd), and Calcite (Ca). Microcline and
chlorite are present in the samples in very small quantities, so they cannot be observed without
zooming in on the patterns.
298
Table 2. Minerals, compound names, and chemical formu
las of the crystalline phases present in CO. HR, and HF.
Sample ID
Mineral CO HR HF
Quartz X X X
Kaolinite X X
Pyrite X X
Muscovite X X
Microcline +/-
Chlorite Trace X
Calcite X X
+/- Under detection limit (2.5%). The sample may or may not
contain the mineral.
concentration of most of the elements analyzed in the host rock (HR and HF), especially for
Si, Ti, Ba, V, and Cr. The major elements found in the samples were aluminum (Al), silicon
(Si), iron (Fe), potassium (K), and titanium (Ti), and the trace elements were strontium (Sr),
magnesium (Mg), barium (Ba), lead (Pb), manganese (Mn), nickel (Ni), copper (Cu), zinc
(Zn), arsenic (As), vanadium (V), and chromium (Cr). Many of these elements have been asso
ciated with the production of reactive oxygen species which have been related to RCMD tox
icity (Moreno et al., 2019). Table 3 shows the concentration of each element analyzed.
299
Table 3. Elemental concentrations in CO, HR, and HF samples.
Sample Al Si Fe K Ti Sr Mg Ba Pb Mn Ni Cu Zn As V Cr
ID % % % % % µg/g µg/g µg/g µg/g µg/g µg/g µg/g µg/g µg/g µg/g µg/g
CO 0.90 3.53 0.9 N.A. 0.11 41.2 N.A. 56.6 16.6 24.4 21.8 20.4 52.7 27.8 36.9 26.3
HR 0.6 23.8 1.1 2.35 0.63 256.7 237 442.9 27.5 57.5 46.5 48.3 51.2 18.9 179.0 92.2
HF 0.5 25.0 1.7 2.04 0.72 53.4 125 341.7 1.1 160.2 42.9 35.1 138.7 3.3 170.1 94.9
3.1.3 FTIR
The FTIR analysis was carried out on the samples before and after dissolution to determine
the functional groups in the samples’ composition and their changes after exposure to SLFs.
Figure 3 and Table 4 show the FTIR spectra obtained and the peak assignments respectively.
In the first graph (a), which corresponds to the CO sample, it can be observed that on the left
side, numerous peaks are missing in the samples HR and HF. These peaks belong to
antisymmetric–CH3 deformation (1445 cm−1) and the benzene C=C stretching (1602 cm−1)
which are characteristics of the carbonaceous material that is present in CO. Also, it can be
observed that most of the peaks decrease after the dissolution experiment, indicating that
those functional groups are actually dissolving, especially the ones for quartz (779 and
799 cm−1) and kaolinite (913 cm−1). These peaks are pretty much gone after dissolution. This
behavior was also observed in the previous work where the influence of geographic location
was evaluated [4].
Figure 3. FTIR spectra of (a) CO, (b) HR, and (c) HF.
300
Table 4. Peak assignments from FT-IR (Lin et al., 2019).
Wavenumber (cm−1) Peak Assignment
Figure 4. Mass normalized dissolution of metals in GS vs time from (a) CO, (b) HR, (c) HF.
301
Figure 5. Mass normalized dissolution of metals in ALF vs time from (a) CO, (b) HR, (c) HF.
• HL-60 and A549 cells: IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12p70, IL-13, and TNF-α.
• THP-1 cells: GM-CSF, IL-1α, IL-5, IL-7, IL-12/IL-23p40, IL-15, IL-16, IL-17A, TNF-β,
and VEGF-A
Figure 6. Results of dust exposures to (a) HL-60 cells using low (5 µg/mL), medium (10 µg/mL), and
high (20 µg/mL) concentrations, and (b) THP-1 cell and (c) A549 cells using low (10 µg/mL) and high
(100 µg/mL) concentrations.
302
These cells play an important role in the body’s immunological response to external agents.
Neutrophilic cells are some of the first cells to react against harmful agents such as viruses and
bacteria, and their main role is to fight infection. Epithelial cells make up the tissue of the
internal and external surfaces of the body and are the first to come in contact with external
agents, such as dust. They contribute to the immunological response mainly with the produc
tion of cytokines. Macrophages are cells that act against infection and injury by engulfing and
eliminating foreign agents, damaged or dead cells, and other harmful organisms. These cells
are also involved in other immunologic processes.
The results of selected cytokines for cells HL-60, A549, and THP-1 are shown in Figure 6.
In general, cytokines are proteins released from cells that are involved in multiple processes,
such as signaling the immune system to act against foreign agents, cell growth, and pro-
inflammatory and anti-inflammatory responses. In the experiment, an upregulated response
(positive values) indicates a pro-inflammatory response, while a downregulated response
(negative values) may indicate an immunosuppressive response. Immunosuppression is still
important because an immunosuppressed lung may be more susceptible to infection (influ
enza, etc.), especially when considering occupational dust exposure.
In HL-60 cells (neutrophilic), 6/10 cytokines had at least one proinflammatory response
among the 3 samples and the different concentrations of exposure (IL-1β, IL-2, IL-4, IL-8, IL-
12p70, and TNF-α). In general, almost all the reactions were obtained from HR. In A549 cells
(lung epithelial), 8/10 cytokines had at least one proinflammatory response (IFN-γ, IL-1β, IL-
2, IL-4, IL-6, IL-8, IL-13, and TNF-α), and 2 of them had a reaction from all the samples
under all the concentrations tested (IL-4 and IL-6). In these last 2 cytokines, the reactions
were mostly higher from the samples HR and HF, and for the other 6 cytokines, the reaction
came from either HR or HF, displaying a major affinity to display a proinflammatory
response with exposures of the host rock (HR and HF). In THP-1 cells (macrophage), 8/10
cytokines had at least one proinflammatory response, from which 7 had a reaction from the
CO sample. This indicates, opposite to the other cells, that the coal sample (CO, even with less
element and mineral content) is the activator of inflammation in the macrophage cells,
although the expressions were low. Additionally, HF was the sample with the highest content
of Fe. It has been mentioned in previous research that Fe may enhance the toxicity of RCMD
(Trechera, et al., 2020), but in this case, only some of the lung epithelial cytokines displayed
an inflammatory response (IL-1β, IL-4, IL-6, IL-8, and TNF-α), but not with a remarkable
trend, and in the macrophage and neutrophilic cells, almost any cytokine reacted to HF.
Thus, this study does not demonstrate whether Fe influences RCMD toxicity.
In summary, neutrophilic and epithelial cells had a greater inflammatory response from the
host rock samples in general, while macrophages had activation of inflammation produced
mainly by the coal sample. This indicates that the host rock is more toxic than coal for the
neutrophilic and epithelial cells, but the opposite for macrophage cells.
4 CONCLUSIONS
This study compared the characteristics and toxicity of dust samples from different sources in
the mine: coal seam and host rock, with the objective of determining if the dust coming from
the host rock is more toxic. The following are the main conclusions found:
• When the composition was compared, as expected, it was found that the host rock has
a significantly higher content of elements (especially Si, Ti, Ba, V, and Cr) and minerals
(quartz, kaolinite, and muscovite), which may influence in its final toxicity.
• Dissolution experiments showed no big differences between the dissolutions in coal and
host rock samples when GS was used. The element concentrations remained similar regard
less of the initial bioavailability. However, when tested with ALF, more elements were dis
solved, and the dissolutions increased in an order of magnitude of 10 (from ×104 to ×105)
when comparing coal and host rock samples. This indicates that the elemental bio-
accessibility is higher in samples from the host rock, and thus, their toxicity.
303
• In the neutrophilic cells the inflammatory response was moderate across the cytokines ana
lyzed, and the responses were coming mainly from HR. Lung epithelial cells displayed
major affinity to respond to host rock samples, while the macrophage cells’ response was
associated with the CO samples, but low. This indicates that the host rock is more likely to
activate the cytokines from the neutrophilic and lung epithelial cells, but not much from the
macrophages. Macrophage cells were more likely to be activated by the coal samples that
were considered less toxic regarding composition and elemental dissolution in SLFs.
• Further research is needed using different sets of samples from different mines to determine
if the trends found in this research are consistent from mine to mine. Also, additional exam
ination of the implications of metal release in the body is required to better understand
their interactions with the body processes and their relationship with RCMD toxicity.
REFERENCES
304
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Respirable coal mine dust (RCMD) and respirable crystalline silica (RCS) are
hazardous materials that must be controlled in underground coal mines to reduce miners’ expos
ure. For this, monitoring systems are used to identify high dust concertation spots and apply crit
ical controls. Using RingIR’s patented rtCRDS technology, a portable rtCRD spectrometer was
built to use as a dust monitor in mining environments to measure coal and silica respirable dust
levels rapidly and accurately. This research compares a PDM3700, a TSI aerodynamic particle
sizer 3321, an SPS30 particulate matter sensor, and NIOSH methods for silica analysis with the
new dust monitor developed by Ring-IR to test its functionality in simulated underground coal
mine conditions. Simulated mine conditions are created inside a dust chamber using a Topas
solid aerosol generator for dust supply and a humidity control station for humidity conditions.
1 INTRODUCTION
Mine workers encounter several health and safety hazards during daily job activities, increas
ing the risk of illnesses and injuries (Akintoye & MacLeod, 1997; Chu & Muradian, 2016;
Domínguez et al., 2019; Mitchell et al., 1998; Sanmiquel et al., 2015; Stewart, 2020). One of
these dangerous situations is exposure to respirable dust particles in working areas. Dust is
generated daily by underground and surface mining activities, such as grinding, cutting, and
transporting materials. Nonetheless, underground mines’ conditions could be more dangerous
due to limited ventilation, confined spaces, and other environmental conditions (Jiskani et al.,
2020; Strzemecka et al., 2019; Sunkpal et al., 2018; Yang et al., 2021). Consequently, there are
more chances of finding high dust concentration areas in underground coal mines, which
leads to the need for more engineering and administrative controls to keep a safe workplace.
In underground coal mines, dust particles’ respirable fraction is called respirable coal mine dust
(RCMD), which mineral content depends on the specific geology of the deposit, but the most
common minerals associated include quartz, illite, calcite, pyrite, and kaolinite (National Institute
for Occupational Safety and Health (NIOSH), 1995). It can also include diesel particulate matter
(DPM), among other materials (Pan et al., 2021). Some of these materials, such as crystalline silica
and DPM, are carcinogens (Calas et al., 2017). In addition, respirable crystalline silica (RCS) is
a critical component of RCMD and its content should be considered in workers’ exposure limits.
Long-time exposure to RCMD and RCS can lead to multiple lung diseases, including coal
worker’s pneumoconiosis (CWP), mixed dust pneumoconiosis, silicosis, dust-related diffuse
fibrosis (DDF), and progressive massive fibrosis (PMF) (Centers for Disease Control and Pre
vention, 2002; Cullinan et al., 2017; NIOSH, 2021; Salinas et al., 2022; Shekarian et al., 2021).
Dust is controlled in mines through several mitigation techniques involving water spray systems,
filters, and equipment to reduce dust concentrations in working areas. OSHA and MSHA defined
the permissible exposure limits (PEL) to control exposure in work environments, also known as
DOI: 10.1201/9781003429241-32
305
dust standards (Centers for Disease Control and Prevention, 2022). If corrections action, engineer
ing, and administrative controls cannot eliminate the hazard or reduce concentrations below PEL,
PPEs are needed. Having all this in mines, continuous personal dust monitors (CPDM) will finally
alert miners to move from high concentration zones and avoid breathing these particles. Continu
ous dust monitoring across the mines can significantly help in the decision-making to reduce
miners’ exposure in working areas.
The approved CPDM device to measure RCMD in underground coal mines is the PDM3700,
which can provide dust concentration in near real-time. However, it cannot distinguish difference
components in RCMD, and therefore it does not give silica content in RCMD samples. Coal mine
dust personal sampling units or equivalent are approved devices commonly used to measure RCMD
and RCS. The process of measuring RCS can take more time and involve lab techniques that require
taking the sample off-site, which has led to the need to create a CPDM for silica analysis.
This research aims to create laboratory conditions to simulate underground coal mine
environments and test monitoring systems to measure RCMD and RCS. Available technology
in the market will be used to test the effectiveness of monitoring systems under different
humidity conditions, and a near real-time dust monitor developed by Ring-IR will be tested to
measure RCMD and RCS using cavity-enhanced direct absorption methods.
2 EQUIPMENT
The development of this project included two main parts before experimentation. The first
was to create a platform to simulate underground coal mine conditions, modifying humidity
and concentration inside a dust chamber. The second was to complete the Ring-IR dust moni
tor development for testing in the platform. The experimental part of this research will involve
using the following equipment.
306
2.2 PDM3700
The personal dust monitor developed by Thermo Fisher is an approved device to measure
RCMD in underground coal mines. It will be used as a reference for comparison when testing
the Ring–IR dust monitor.
3 DUST CHAMBER
A customized static dissipative PVC dust chamber was created in collaboration with Terra
Universal Inc. and will work as the testing body to simulate underground coal mine condi
tions. Dimensions are 35” L x 24” W x 25” H with a total volume of 12.5 ft3. The chamber
has a door to access it, openings to introduce tubing from equipment, and a removable sheet
to introduce more extensive equipment. The chamber was designed to include multiple acces
sories for the intended experiments, such as a transfer chamber or airlock to pass equipment
or instruments during testing, automatic RB valves to maintain positive pressure inside,
a humidity module that works with the HCS to keep the desired RH values, a differential
pressure gauge, and gloves to access the chamber during experiments. Also, it includes
a recirculating system with a filter for air cleaning. The final product can be seen in Figure 2.
307
Figure 2. Dust chamber.
The CRD signal from MCT detector is fed into a Field Programmable Gate Array (FPGA) based
lock-in amplifier and hence respective absorption spectrum will be generated. Preliminary tests were
done using a coal dust sample (New Mexico mine, <104μm), and a silica dust sample (Australian
mine, <10μm) for this analysis. The dust sample was measured in a dust sampling jar and the cap was
replaced. Cyclone was set up with inlet and outlet tubing as shown in Figure 4. The inlet tubing was
inserted into the sampling jar from the opening of the cap and this tube was set near the upper portion
of the jar without any contact with dust in the bottom of the jar. In this way, only small size particles
will get through the instrument. The outlet tubing was connected to the inlet of the optical cavity and
the cavity outlet was connected to a particle filter inlet. Then the particle filter outlet was connected to
the pump. The sampling jar was continuously shaken throughout the experiment.
308
Figure 4. Sampling jar attachment to the spectrometer.
After each dust sample, room air was pumped through the system for one minute and
checked for the existence of dust particles by capturing a spectrum. That spectrum was com
pared against the initial background spectrum. The next sample was carried out only when the
system was free from dust particles and the background spectrum was matched to the initial
background spectrum. Dust samples were tested from lowest to highest concentrations.
309
5 DISCUSSIONS
The preliminary dust results obtained for silica dust samples are shown in Figure 6. This
data demonstrates the potential of rtCRDS as a dust monitoring in mine environments.
Further testing is underway with the dust chamber simulating underground coal mine
conditions.
Simulated underground coal mine conditions were created inside a dust chamber to com
pare monitoring systems to measure RCMD and RCS. The humidity control station (HCS)
constantly supplied the dust chamber to produce low or high RH values. In addition, an aero
sol generator was used to keep airborne particle concentrations during experiments.
The HCS was initially tested using a model with a volume of 0.7 ft3 and a bigger chamber
with a volume of 18 ft3; this last one was different from the one shown in Figure 2. The model
was a smaller testing body and took about 1 to 5 minutes to reach very dry or humid values.
On the other hand, experiments in the chamber usually took more than an hour to get 80%
RH conditions. After reaching desired RH conditions, humidity maintains very similar for
more than 1 hour inside the chamber.
For dust concentration testing, dust was conducted from the aerosol generator to the
chamber with tubing to the back top middle (TM) area to disperse from top to bottom, and
an APS 3321 was used to measure concentration values. Area distributions are shown in
Figure 7. Experiments were performed in six internal areas to better understand concentra
tion conditions inside the chamber, measuring changes in particle concentration with differ
ent materials and various feed rates from the aerosol generator. After analyzing results and
comparing changes over different experimental scenarios, the bottom middle (BM) area
was defined as the more reliable internal area to do readings during experiments with moni
toring systems due to consistency in readings and similarity with average and medium
values. Figure 8 shows the concentration over time graph for each internal area. After
twelve initial tests to define the best experimental setup, a final test was completed with
three reading intervals of 10 minutes for each area using coal mine dust to measure the
total concentration inside the chamber in number of particles per cubic centimeter
(#pt /cm³).
310
Figure 7. Internal area distributions dust chamber.
6 FUTURE WORK
More research should be conducted under several scenarios inside the dust chamber, changing
humidity and concentration values to test the Ring-IR dust monitor response to typical
changes in underground coal mines. The next set of experiments will compare the PDM3700
and the SPS30 sensor with values from the Ring-IR device. Once consistent results have been
found measuring RCMD, the following experiments will involve using a coal mine dust per
sonal sampling unit to measure silica content using NIOSH 7500, NIOSH 7603, and NIOSH
FAST methods and compare results with the Ring-IR dust monitor.
REFERENCES
Akintoye, A. S., & MacLeod, M. J. (1997). Risk analysis and management in construction. International
Journal of Project Management, 15(1), 31–38. https://doi.org/10.1016/S0263-7863(96)00035-X
311
Calas, A., Uzu, G., Martins, J. M. F., Voisin, D., Spadini, L., Lacroix, T., & Jaffrezo, J.-L. (2017). The
importance of simulated lung fluid (SLF) extractions for a more relevant evaluation of the oxidative
potential of particulate matter. Scientific Reports, 7(1), 11617. https://doi.org/10.1038/s41598-017-11979-3
Centers for Disease Control and Prevention. (2022, October 30). Methods: Exposure. https://Wwwn.Cdc.
Gov/Eworld/Appendix/Exposure.
Centers for Disease Control and Prevention, N. I. for O. S. and Health. (2002). Health Effects of Occupa
tional Exposure to Respirable Crystalline Silica.. Retrieved from https://www.Cdc.Gov/Niosh/Docs/
2002-129/.
Chu, C., & Muradian, N. (2016). Safety and environmental implications of coal mining. International
Journal of Environment and Pollution, 59 (2/3/4),250. https://doi.org/10.1504/IJEP.2016.079899
Cullinan, P., Muñoz, X., Suojalehto, H., Agius, R., Jindal, S., Sigsgaard, T., Blomberg, A., Charpin, D.,
Annesi-Maesano, I., Gulati, M., Kim, Y., Frank, A. L., Akgün, M., Fishwick, D., de la Hoz, R. E., &
Moitra, S. (2017). Occupational lung diseases: from old and novel exposures to effective preventive
strategies. The Lancet Respiratory Medicine, 5(5), 445–455. https://doi.org/10.1016/S2213-2600(16)30424-6
Domínguez, C. R., Martínez, I. V., Piñón Peña, P. M., & Rodríguez Ochoa, A. (2019). Analysis and
evaluation of risks in underground mining using the decision matrix risk-assessment (DMRA) tech
nique, in Guanajuato, Mexico. Journal of Sustainable Mining, 18(1), 52–59. https://doi.org/10.1016/J.
JSM.2019.01.001
Jiskani, I. M., Han, S., Shahani, N. M., Ali, M., Dianat, I., & Chalgri, S. R. (2020). Evaluation of phys
ical and environmental working conditions of underground coal mines within the framework of
ergonomics. International Journal of Mining and Mineral Engineering, 11(3), 240. https://doi.org/
10.1504/IJMME.2020.109639
Mitchell, R. J., Driscoll, T. R., & Harrison, J. E. (1998). Traumatic work-related fatalities involving
mining in Australia. Safety Science, 29(2), 107–123. https://doi.org/10.1016/S0925-7535(98)00012-5
National Institute for Occupational Safety and Health (NIOSH). (1995). Criteria for a Recommended
Standard. Occupational Exposure to Respirable Coal Mine Dust.
NIOSH. (2021). Best practices for dust control in coal mining, second edition. https://doi.org/10.26616/
NIOSHPUB2021119
Pan, L., Golden, S., Assemi, S., Sime, M. F., Wang, X., Gao, Y., & Miller, J. (2021). Characterization of
Particle Size and Composition of Respirable Coal Mine Dust. Minerals, 11(3), 276. https://doi.org/
10.3390/min11030276
Salinas, V., Das, M., Jacquez, Q., Camacho, A., Zychowski, K., Hovingh, M., Medina, A.,
Rubasinghege, G., Rezaee, M., Baltrusaitis, J., Fairley, N., & Roghanchi, P. (2022). Characterization
and Toxicity Analysis of Lab-Created Respirable Coal Mine Dust from the Appalachians and Rocky
Mountains Regions. Minerals, 12(7), 898. https://doi.org/10.3390/min12070898
Sanmiquel, L., Rossell, J. M., & Vintró, C. (2015). Study of Spanish mining accidents using data mining
techniques. Safety Science, 75, 49–55. https://doi.org/10.1016/J.SSCI.2015.01.016
Shekarian, Y., Rahimi, E., Rezaee, M., Su, W.-C., & Roghanchi, P. (2021). Respirable Coal Mine Dust:
A Review of Respiratory Deposition, Regulations, and Characterization. Minerals, 11(7), 696. https://
doi.org/10.3390/min11070696
Stewart, A. G. (2020). Mining is bad for health: a voyage of discovery. Environmental Geochemistry and
Health, 42(4), 1153–1165. https://doi.org/10.1007/s10653-019-00367-7
Strzemecka, J., Goździewska, M., Skrodziuk, J., Galińska, E., & Lachowski, S. (2019). Factors of work
environment hazardous for health in opinions of employees working underground in the ‘Bogdanka’
coal mine. Annals of Agricultural and Environmental Medicine, 26(3), 409–414. https://doi.org/
10.26444/aaem/106224
Sunkpal, M., Roghanchi, P., & Kocsis, K. C. (2018). A Method to Protect Mine Workers in Hot and Humid
Environments. Safety and Health at Work, 9(2), 149–158. https://doi.org/10.1016/j.shaw.2017.06.011
Yang, L., Birhane, G. E., Zhu, J., & Geng, J. (2021). Mining Employees Safety and the Application of
Information Technology in Coal Mining: Review. Frontiers in Public Health, 9, 709987. https://doi.
org/10.3389/fpubh.2021.709987
312
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
Y. Liu
Henan Joint International Research Laboratory of Mine Safety and Occupational Health, Henan Polytechnic
University, Jiaozuo, Henan, China
ABSTRACT: Coal dust causes a range of diseases and health problems worldwide, such as
Coal Workers’ Pneumoconiosis. Numbers of static studies showed that surfactants could
increase the coal dust wettability effectively in water spraying. However, coal dust does not
remain in a static state in underground mines. Therefore, studies should be conducted in
a dynamic state where coal dust and surfactant droplets will get a shorter contact time. In this
study, wind tunnel tests have been conducted to capture the dynamic process. Logistic regres
sion modelling has been done to evaluate the suppression performance in terms of efficiency
and quantity of dust suppressed with respect to three dependent variables: surfactant types,
surfactant quantity, and coal dust concentration. The results showed that at low coal dust
concentrations, the suppression efficiency is greater compared to high dust concentrations.
Moreover, the surfactant quantity of 0.05 % to 0.20 % is more efficient than above. Although
the surfactant type did not emerge as significant when low and high dust concentrations were
combined, it becomes significant at low dust concentrations where SDS and CTAB are more
efficient than SDBS and TX100. The results will help in selecting surfactants with the appro
priate quantity in different coal dust concentrations.
1 INTRODUCTION
Coal dust is a fine powdered form of coal, which is created by the crushing, grinding, or pulver
izing of coal during mining, transportation, or by mechanically handling coal. It is a form of
fugitive dust and poses the acute hazard of forming an explosive mixture in the air and the
chronic hazard of causing pulmonary illness in people who inhale excessive quantities of it. Coal
workers’ pneumoconiosis (CWP), also known as the black lung disease, is one of the most
famous occupational diseases, which is usually caused by overexposure to respirable coal dust.
In 2012, there were 27,420 cases that were reported as occupational diseases globally, and the
CWP occupied 45.24 % of that (Organization, 2013). In 2013, CWP has resulted in 25,000
deaths globally (Abubakar et al., 2015). In the U.S., the prevalence of CWP decreased from 3.3 %
to 0.3 % from 1970 to 1998, then rose to 3.3 % as of 2012 (Blackley et al., 2014, Blackley et al.,
2016). In the United Kingdom, although the prevalence of pneumoconiosis decreased from 12 %
to 0.2 % during 1994-1997, it increased to 0.8 % from 1998 to 2000 (Scarisbrick and Quinlan,
2002). In Australia, 20 new cases of CWP had shown within only a fortnight in 2019, and 31 cases
were confirmed in 2020 (Hendry, 2019, Burt; and McGhee, 2020). All the data illustrated
above indicated that the CWP has shown a resurgence tendency. Hence, an effective method for
controlling the coal dust in coal mines is a necessity for the prevention of CWP occurrence.
DOI: 10.1201/9781003429241-33
313
Generally, there are three methods for controlling coal dust for underground mining,
including ventilation, foam technology, and water spray (Chang et al., 2019, Ren et al., 2012,
Rider and Colinet, 2007, Chang et al., 2020). Water spray is still a primary method that has
been undertaken in the past decades. This method is readily applicable to mining industries,
no customized equipment is required, and the cost of consumables and equipment is relatively
low. Although the efficiency of untreated water is usually lower than expected, surfactants can
dramatically increase the efficiency of coal dust suppression. Surfactants usually contain both
hydrophobic groups and hydrophilic groups. While surfactants diffuse in water, the hydro
philic groups remain in the water phase, and the hydrophobic groups extend out of the bulk
water phase for capturing (Xu et al., 2018, Zhao et al., 2022). In the water spray system, sur
factant solution will be sprayed out into the air then hydrophobic groups will catch dust par
ticles. Previous studies claimed that the surfactant-aided solution could dramatically reduce
the coal dust by up to 90 % (Chander et al., 1990, Kilau, 1996, Tien and Kim, 1997). For
example, in Zhou et al.’s study, it was shown that adding sodium dodecyl benzenesulfonate
(SDBS) into untreated water can increase the suppression efficiency for total dust from 55 %
to 76 % in a field test (Zhou et al., 2019). In another study by the same research group, SDBS
was evaluated with additives to improve performance (Zhou et al., 2017). It was found that
the suppression efficiency of both total dust and respirable dust can be improved from 53 %
and 38 % to 75 % and 70 %, respectively. Similarly, another compounding surfactant was also
evaluated by a field test in Wang et al.’s study (Wang et al., 2019). It was claimed that the
suppression efficiency of total dust could be increased from 50 % to 90 %, while the same of
respirable dust can be dramatically improved from 39 % to 88 %. Therefore, the evaluation of
surfactants is significant for improving the suppression efficiency.
The evaluation methods of surfactants can be divided into two categories, including static
tests and dynamic tests. In previous studies, static tests, such as the surface tension test and
the sink test, generally measure interactions between particles and surfactant solutions, such
as the surfactant adsorption ability and coal wettability (Xu et al., 2018, Zhao et al., 2021,
Chang et al., 2021). In contrast, dynamic tests, such as the wind tunnel test and the field test,
can directly test the results of coal dust suppression and then analyze the efficiency of suppres
sion. Moreover, compared to static tests, where the coal particles and the solution are firmly
contacted over the test period, dynamic tests take into consideration of particle contacts, colli
sions, and fluid dynamics (Xu et al., 2018). Although the result of dynamic tests is more reli
able, however, many studies conducted static tests the most when evaluating surfactants due
to time-saving and economic reasons. For instance, Chen et al. evaluated SDBS in static tests
such as the sink test, the surface tension test, and the zeta potential test (Chen et al., 2017).
The result showed that SDBS ranging from 0.2 % to 0.4 % had the best performance in static
tests. Another study by Wang et al. evaluated several surfactants by using static tests, such as
the surface tension test and the contact angle test, and the wetting agent A was optimized with
the concentration and thus had excellent performance in both tests (Wang et al., 2014). Add
itionally, in Chang et al.’s study, three surfactants, including SDBS, Triton X-100 (TX100),
and cetyl trimethylammonium bromide (CTAB), were tested by the surface tension test and
the sink test. It was found that TX100 performed the greatest in terms of coal dust wettability,
while CTAB performed the weakest.
Despite the fact that static tests usually are more convenient than conducting dynamic tests,
static tests and dynamic tests sometimes may provide discrepant results. For example, the per
formance of TX100 and sodium dodecyl sulphate (SDS) was evaluated and compared by
static tests in Glanville and Haley’s study (Glanville and Haley, 1982). It was concluded that
SDS performed better and showed a higher wettability than TX100. Nevertheless, another
study of dynamic tests illustrated a different result, that TX100 could improve the suppression
efficiency greater than SDS in dynamic tests (Tessum and Raynor, 2017). Moreover, the coal
dust concentration underground usually is fugitive and may differ with coal mines. Although
static tests can easily evaluate surfactants with different types or concentrations, the coal dust
concentration usually cannot be taken into consideration. Thus, the wind tunnel test was con
ducted in this study to evaluate surfactants with different coal dust concentrations. This paper
aims to evaluate the suppression efficiency of surfactants with different surfactant
314
concentrations and initial coal dust concentrations. The significance of surfactant type, surfac
tant concentration, and initial coal dust concentration will be analyzed and discussed by
a factorial experiment design. A wind tunnel apparatus was applied to conduct all experiments
in this study. Four surfactants were evaluated, including CTAB, SDS, SDBS, and TX100.
Seven concentrations of surfactants were tested, including 0.00 %, 0.05 %, 0.10 %, 0.15 %,
0.20 %, 0.25 %, and 0.30 %. Two initial coal dust concentrations were selected, including Low
(13.35 mg/m3 (95 % CI: 12.21-14.50)) and High (101.62 (95 % CI: 99.23-104.02)). The findings
of this study can help coal dust control for industries in terms of the determination and opti
mization of surfactants.
2 METHOD
Where the R represents the suppression rate, C (Initial) and C (After) represent the coal
dust concentration prior to and after the surfactant solution action, respectively.
315
Figure 1. Wind tunnel apparatus (1) dust generation system, (2) spray system, (3) monitor (4) dust col
lection system.
3 RESULTS
Table 1 and Figure 2 show the data summary of this experiment. Logistic regression is a kind
of data analysis model that can evaluate data from different factors. In this study, this method
can evaluate the relations between the suppression efficiency and surfactant type, surfactant
concentration, and coal dust concentration, respectively. The overall logistic regression results
are shown in Table 2.
316
Figure 2. Surfactant concentration type interaction.
Coal dust concentration 4.356 1.135 14.733 1 0.000 77.934 8.428 720.617
Surfactant type .928 .813 1.303 1 0.254 2.530 .514 12.447
Surfactant concentration 2.430 1.116 4.739 1 0.029 11.362 1.274 101.337
constant -3.857 1.256 9.425 1 0.002 .021
As resulted in Table 2, the significance of coal dust concentration ranked at the top among
the three factors, which was less than 0.001. The Odd Ratio (OR) was 77.93 with 95 % Confi
dence Interval (CI) between 8.43 and 720.62, which indicated that the suppression efficiency
can be improved significantly when the coal dust concentration decreased. In addition, the sig
nificance of the concentration of surfactants was 0.029, which was the second significant
factor. The OR was 11.362 with 95 % CI between 1.274 and 101.337, showing that the sup
pression efficiency can be easily impacted by the surfactant type. The surfactant concentra
tions were divided into two groups, including a lower group (0.05 %-0.15 %) and a higher
group (0.20 %-0.30 %). As a result of this analysis, the surfactant solution in the lower group
was more efficient than the higher group. Figure 2 presents the average suppression efficiency
of four surfactants with different concentrations. It was found that 0.05 % CTAB had the
highest average suppression efficiency, while 0.10 % TX100 had the lowest. SDS had the best
overall suppression efficiency among all concentrations. For a better result of the analysis, the
data can be divided and analyzed into two groups, including the low initial coal dust concen
tration group and the high initial coal dust concentration group. The statistical analysis of
both these two groups is shown in Table 3 and Table 4.
317
Table 3. Analysis results at low initial coal dust concentration.
95 % C.I. for EXP(B)
B S.E. Wald Df Sig. OR Lower Upper
Table 3 shows the logistic regression results of the low initial coal dust concentration group.
Surfactant types were classified by type A (CTAB and SDS) and type B (SDBS and TX100).
As observed, the surfactant type significantly impacted the result of suppression efficiency due
to the conspicuous significance of 0.007, which was less than 0.05. Type A (CTAB and SDS)
had a higher effect on suppression efficiency than type B (SDBS and TX100). However, the
surfactant concentration only had a less relevant coefficient against suppression efficiency,
although it had a more considerable significance towards the entire experiment. The signifi
cance was 0.067, and the 95 % CI was from 0.009 to 1.168, which indicated that there was no
obvious evidence to prove the relationship between the concentration of surfactants and the
suppression efficiency. Figure 3 showed the average suppression efficiency at the low initial
coal dust concentration. Similarly, SDS had the best overall suppression efficiency among
these concentrations. Interestingly, CTAB at 0.05 % and 0.10 % concentrations had great per
formance than at higher concentrations. For other surfactants, the average suppression effi
ciency was similar at each concentration. Hence, at lower initial coal dust concentrations, the
surfactant type had the dominant impact on the suppression efficiency of coal dust.
Table 4 shows the logistic regression results of the high initial coal dust concentration
group, and the interaction between surfactant concentration and type is illustrated in
Figure 4. It is noteworthy that both surfactant type and surfactant concentration had less cor
relation towards the suppression efficiency. The significances of surfactant type and surfactant
concentration were 1.000 and 0.133, while the 95 % CI of both makeups were from 0.213 to
4.7 and 0.693 to 16.022, respectively. This could occur mainly because of insufficient surfac
tant, which cannot suppress a higher volume of coal dust. Therefore, it is necessary to analyze
318
the impacts of the surfactant type and the surfactant concentration towards the quantity of
coal dust suppression instead of the efficiency rate, which is shown in Table 5.
Table 5 shows the impacts of the surfactant type and the surfactant concentration towards
the quantity of coal dust suppression. The significance of the surfactant concentration was
0.028, which indicated that the surfactant concentration can significantly impact the quantity
of coal dust suppression. Meanwhile, the expectation of the surfactant concentration was 6.6,
which means the surfactant concentration had a higher suppression efficiency from 0.05 % to
0.15 %.
319
From the above data analysis and discussion, the suppression efficiency of four common
surfactants was evaluated with different surfactant concentrations and different initial coal
dust concentrations. It was found that, increasing surfactant concentration generally can
improve the suppression efficiency, while the performance of surfactants would be weakened
if the initial coal dust concentration was high. As for the surfactant type, SDS and CTAB had
greater average suppression efficiency than TX100 and SDBS. It is noticeable that SDS
showed a great performance on overall surfactant concentrations, while CTAB performed
better at lower surfactant concentrations. It was suggested that, SDS ranging from 0.10% to
0.20% should be applied for controlling coal dust, because it can significantly increase the sup
pression efficiency of coal dust in both low and high initial coal dust concentration conditions.
Additionally, CTAB at lower surfactant concentrations could be an alternative option for
controlling coal dust, especially for suppressing coal dust with lower concentrations. Surpris
ingly, the concentration of surfactants was a crucial factor impacting the suppression effi
ciency at lower coal dust concentrations. Nevertheless, it was insignificant in improving the
suppression efficiency of the high initial coal dust concentrations. An inadequate amount of
surfactant spraying could be the main reason for this phenomenon. In future studies, the
impact of surfactant spraying rate on the suppression efficiency should be investigated.
4 CONCLUSION
The main target of this study is to evaluate how three factors, including the initial coal dust con
centration, the surfactant type, and the surfactant concentration, impact the coal dust suppression
efficiency. From the Logistic Regression modelling results, the initial coal dust concentration had
a significant impact overall towards the coal dust suppression efficiency among three factors.
Moreover, the surfactant concentration also showed a dramatic impact, whereas the surfactant
type was relatively less critical towards the suppression efficiency. However, for the lower initial
coal dust concentration, there was a different situation for the surfactant type. The surfactant type
had a higher effect than the surfactant concentration. The CTAB and SDS were more significant
for the coal dust suppression than that of SDBS and TX100. At the higher initial coal dust concen
tration, it is noteworthy that both the surfactant type and surfactant concentration were less sig
nificant towards the suppression efficiency. This is probably caused by insufficient surfactant
quantities for dust suppression. Thus, it is crucial to evaluate the quantities of coal dust suppres
sion rather than efficiency at the higher initial coal dust concentration. The result clearly showed
that the surfactant concentration can substantially affect the suppression quantities of the coal
dust, even though the surfactant type is still of unimportance. In conclusion, the initial coal dust
concentration had the highest impact on coal dust suppression efficiency. At the lower initial coal
dust concentration, the surfactant type showed a higher impact on the suppression efficiency than
that of the surfactant concentration. In contrast, at the higher initial coal dust concentration, the
surfactant concentration can substantially impact the suppression quantities of the coal dust.
REFERENCES
Abubakar, I., Tillmann, T. & Banerjee, A. (2015) Global, regional, and national age-sex specific all-cause
and cause-specific mortality for 240 causes of death, 1990-2013: a systematic analysis for the Global
Burden of Disease Study 2013. Lancet, 385, 117–171.
Blackley, D. J., Crum, J. B., Halldin, C. N., Storey, E. & Laney, A. S. (2016) Resurgence of progressive
massive fibrosis in coal miners—Eastern Kentucky, 2016. Morbidity and mortality weekly report, 65,
1385–1389.
Blackley, D. J., Halldin, C. N. & Laney, A. S. (2014) Resurgence of a debilitating and entirely prevent
able respiratory disease among working coal miners. American journal of respiratory and critical care
medicine, 190, 708–709.
Burt;, J. & Mcghee, R. (2020) Cases of mine-dust lung disease and silicosis increasingly found in Queens
land coal mine workers. Retrieved from https://www.abc.net.au/news/2020-02-25/silicosis-and-black-
lung-cases-rise-queensland-workers/11998404.
320
Chander, S., Alaboyun, A. & Aplan, F. (1990) On the mechanism of capture of coal dust particles by
sprays. Proceedings of the Third Symposium on Respirable Dust in the Mineral Industries.
Chang, P., Chen, Y., Xu, G., Huang, J., Ghosh, A. & Liu, W. V. (2019) Numerical study of coal dust
behaviours and experimental investigation on coal dust suppression efficiency of surfactant solution
by using wind tunnel tests. Energy Sources, Part A: Recovery, Utilization, and Environmental Effects,
1–16.
Chang, P., Xu, G., Chen, Y., Ghosh, A. & Moridi, M. A. (2021) Improving coal powder wettability
using electrolyte assisted surfactant solution. Colloids and Surfaces A: Physicochemical and Engineering
Aspects, 613, 126042.
Chang, P., Zhao, Z., Xu, G., Ghosh, A., Huang, J. & Yang, T. (2020) Evaluation of the coal dust sup
pression efficiency of different surfactants: A factorial experiment. Colloids and Surfaces A: Physico
chemical and Engineering Aspects, 124686.
Chen, Y., Xu, G. & Albijanic, B. (2017) Evaluation of SDBS surfactant on coal wetting performance
with static methods: Preliminary laboratory tests. Energy Sources, Part A: Recovery, Utilization, and
Environmental Effects, 39, 2140–2150.
Glanville, J. O. & Haley, L. H. (1982) Studies of coal dust wetting by surfactant solutions. Colloids and
Surfaces, 4, 209–212.
Hendry, M. (2019) Black lung advocates say 20 Queenslanders diagnosed with coal dust diseases in a
fortnight. Retrieved from https://www.abc.net.au/news/2019-02-26/dozens-of-new-black-lung-cases-
qld-advocates-say/10851482.
Kilau, H. W. (1996) Suppression of longwall respirable dust using conventional water sprays inoculated
with surfactants and polymers.
Organization (2013) National Health and Family Planning Commission of the People’s Republic of
China. 9th Global Conference on Health Promotion, Shanghai.
Ren, X. W., Wang, M. D., Kang, H. Z. & Lu, X. X. (2012) Engineering case report: A new method for
reducing the prevalence of pneumoconiosis among coal miners: Foam technology for dust control.
Journal of occupational and environmental hygiene, 9, D77–D83.
Rider, J. P. & Colinet, J. F. (2007) Current dust control practices on US longwalls.
Scarisbrick, D. & Quinlan, R. (2002) Health surveillance for coal workers’ pneumoconiosis in the United
Kingdom 1998–2000. Annals of Occupational Hygiene, 46, 254–256.
Tessum, M. W. & Raynor, P. C. (2017) Effects of Spray Surfactant and Particle Charge on Respirable
Coal Dust Capture. Safety and Health at Work, 8, 296–305.
Tien, J. C. & Kim, J. (1997) Respirable coal dust control using surfactants. Applied occupational and
environmental hygiene, 12, 957–963.
Wang, K., Jiang, S., Wu, Z., Shao, H. & Pei, X. (2014) Experimental study on complex wetting agent
enhancing the dusting effect and its application. Electron. J. Geotechnical Eng., 19, 9784–9798.
Wang, X., Yuan, S., Li, X. & Jiang, B. (2019) Synergistic effect of surfactant compounding on improving
dust suppression in a coal mine in Erdos, China. Powder Technology, 344, 561–569.
Xu, G., Chen, Y., Eksteen, J. & Xu, J. (2018) Surfactant-aided coal dust suppression: A review of evalu
ation methods and influencing factors. Science of the Total Environment, 639, 1060–1076.
Zhao, Z., Chang, P., Xu, G., Ghosh, A., Li, D., & Huang, J. (2021) Comparison of the coal dust suppres
sion performance of surfactants using static test and dynamic test. Journal of Cleaner Production,
129633.
Zhao, Z., Chang, P., Xu, G., Xie, Q. & Ghosh, A. (2022) Comparison of static tests and dynamic tests
for coal dust surfactants evaluation: A review. Fuel, 330, 125625.
Zhou, Q., Qin, B., Ma, D. & Jiang, N. (2017) Novel technology for synergetic dust suppression using
surfactant-magnetized water in underground coal mines. Process Safety and Environmental Protection,
109, 631–638.
Zhou, Q., Qin, B., Wang, F. & Wang, H. (2019) Experimental investigation on the performance of
a novel magnetized apparatus used to improve the dust suppression ability of surfactant-magnetized
water. Powder Technology, 354, 149–157.
321
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J. Finn
Englo Inc
ABSTRACT: Process ventilation involves using airflow to control dust generated in process
operations. Understanding of process ventilation design practice is important for mining engin
eers and operations personnel responsible for mitigating the risk of handling combustible and
non-combustible dust hazards. Applications such as crushing, screening, conveying, loading,
and other bulk material handling processes often require use of exhaust hoods and ductwork to
contain and then convey dust emissions to an air cleaning device. This paper is intended to be
a short summary of nine key steps to follow when developing a process dust control solution.
1 INTRODUCTION
Dust emission control at surface and underground mining operations is an important issue for
several reasons. Among these are respirable hazards such as pneumoconiosis (black lung) at
coal handling sites and silicosis where silica dust is present. In addition to these health hazards
there are issues such as accumulation of dust in electrical controls and premature failure of
operating equipment. Combustible dust at coal handling operations presents a huge hazard
that requires specific risk mitigation design practices.
Process dust control solutions include housekeeping cleanup, dust suppression misting
sprays and exhaust ventilation dust control systems. Operations design should also include
consideration for making process modifications to minimize dust generation from the
outset.
Process ventilation guidelines detailed in Industrial Ventilation: A Manual of Recom
mended Practice, published by ACGIH, are used for engineering of dust control sys
tems in the USA. NFPA National Fire Protection Association publishes standards for
design of combustible dust control systems. NFPA 652 Standard on the Fundamentals
of Combustible Dust 2016 provides basic principles of and requirements for identifying
and managing the fire and explosion hazards of combustible dusts and particulate
solids.
NFPA 652 requires completion of a Dust Hazard Analysis (DHA) as a necessary starting
point for defining the scope of combustible dust control needed when combustible dusts are
involved. The publication, Guidelines for Combustible Dust Hazard Analysis, by The Center
for Process Safety of the American Institute of Chemical Engineers, is an excellent resource
for this topic.
NIOSH Mining Program Information Circular IC 9532, Best Practices for Dust Control in
Coal Mining, is another important resource that covers a much broader view of mining dust
control than provided by the very limited scope of this paper.
Designing for mining dust control based on the resources and practices mentioned
above ensures that best practice is achieved for safety, maintenance, and effective system
operation.
DOI: 10.1201/9781003429241-34
322
2 DUST MANAGEMENT SOLUTIONS
Approaches for control of fugitive dust include process design change, housekeeping cleanup,
misting suppression, roadway dust containment, unducted exhaust ventilation, machine mounted
control equipment and exhausting with hoods through ductwork to an air control device.
Three general ventilation examples of unducted ventilation are as follows:
Figure 1 shows a wet dust extractor with a water recirculation system. The unducted extractor
was installed at an underground drift mine being developed by drill and blast. The equipment is
placed outby of the blast zone and left off for the blast. Following the blast the extractor is
started and quickly clears the dust and blasting odors by recirculation of ambient air.
Another version of unducted ventilation is this example of a wet dust extractor mounted on
a roadheader mining machine being readied for delivery to a tunneling project (Figure 2).
Lastly, Figure 3 shows one of two river diversion tunnels developed using roadheaders. In
this project an inadequate ventilation plan resulted in excessive dust and the tunnel develop
ment was shut down. Exhausting air at the face was increased with larger ductwork connected
to Englo wet dust extractors located at the tunnel entrance. The project went into compliance,
and hundreds of men resumed work resulting in completion in a tight deadline time frame.
Nine steps for developing a process dust control system using exhaust ventilation systems are
presented in the following sections. They are based in general on design procedure taught in
323
the Industrial Ventilation Manual, although not always in the same order. The examples
given are mostly from evaluation of existing systems. However, the same design approach
applies when developing solutions for new systems.
In addition to the design procedure, it is important to keep in mind that regulatory require
ments must be considered so that the completed project achieves these objectives.
The Industrial Ventilation manual has extensive information about exhaust design prin
ciples with comprehensive details about each of the design steps. The description provided
here is a very brief overview of the design approach.
324
Displaced air is the volume of air caused by the flow of ore thru a process location. Calcu
late the volume of bulk material being processed. Multiply density of crushed ore times the
flow rate to get the product displacement volume.
Calculation of displacement for transferring material from a belt conveyor can be deter
mined by multiplying the ore cross section times belt speed. Once the bulk material displace
ment volume is determined, a multiplication factor is then used to calculate a displaced air
volume for design. This factor may range from 1.2 to 2.0 depending on process conditions.
The equation, Q = VA, is used in many ways during the design process. Q equals volume of
air expressed as cubic feet per minute, cfm, or cubic meters per second, m3/s. V equals velocity
of the air in feet per minute, fpm, or meters per second, mps.
Figure 5 illustrates a conveyor skirt box exhausted with a tapered hood. Consider a 48”
conveyor running at 500 fpm. The air displacement, Q=VA is 4ʼ width x 1ʼ material average
height x 500 fpm belt speed equals 2000 cfm. The average height factor of one foot is conser
vative as the actual number will be less than that in most instances. This approach also covers
the minimal factor of Induced Air. Displaced Air is then determined by multiplying the cfm
by a factor ranging from 1.2 or more if needed. Displaced Air equals 3000 cfm.
Given this information we can determine the exhaust hood design. The first item to decide
is the area of the hood at the capture point. With a belt conveyor the general rule is to keep
the hood inlet face velocity below 500 fpm. Therefore, the hood inlet area is 3000 cfm/500 fpm
equals six sq ft.
Next, calculate the duct diameter for the hood connection. Key to this is determining min
imum duct transport velocity, V. Transport velocity depends on multiple factors and can
range from very low, below 2000 fpm for an HVAC clean air duct to above 10,000 fpm in
a high vacuum system. For combustible coal dust V equals 4000 to 5000 fpm. Using Q of 3000
cfm, you then calculate A = Q/V, or A = 3000/4500 = 0.67 sq ft, which lets you select the duct
diameter.
Engineering Toolbox has a wonderful duct design calculation tool at “Air Ducts - Major
Friction Head Loss, Online Calculator.” Enter 3000 cfm and iterate to an 11” diameter duct
with a velocity of 4546 fpm. There is also a metric tabulation in this calculator.
Note, in Figure 5, a bleed-in branch is used in combination with a butterfly damper to regu
late indraft at the conveyor hood. The purpose of this design approach is to enable setting the
draft at the capture zone while adjusting air volume going to the main ducting using the
bleed-in branch. This ensures that duct velocity is kept in balance with the hood draw. Main
tained air velocity is fundamental to preventing dust accumulation in the ductwork.
325
3.3 Provide plan and elevation views
Preparing layout drawings describing each hood location is the next step in the design process.
These will typically be preliminary sketches including plan views, elevation views and section
views when working with an existing process. Photos and video are a great help as are equip
ment and facility drawings.
Field sketches showing hood and equipment locations in reference to column centers are very
important. A plan view field sketch of exhaust hoods for the screening plant from Figure 4 is
shown in Figure 6. Plan, elevation and section views were made for creating the engineering
drawings.
The sketch includes hood and ducting diameters and a duct routing layout. Additional
details including elevation and section views as well as hood details are used to prepare
a SolidWorks 3D design model. Hood exhaust volumes are combined to determine duct and
branch sizing for use in Step Four.
326
A spreadsheet is an alternative to the Air Duct tool for sizing duct diameters as well as for
calculating system static pressure. The spreadsheet in Figure 7 calculated a system volume of
17,600 cfm at 8.61 inches of static pressure loss. This information is then used for selection of
the dust collector and fan.
327
Figure 9. Reverse air baghouse.
Figure 8. Cartridge dust collector.
subject to NFPA 652 guidelines since combustible dust is involved. Provisions for using dry
collectors may include explosion venting, fire suppression, backdraft dampers and other
safety protections.
The reverse air baghouse was eventually replaced by a wet dust extractor to mitigate fire
and explosion risk. The extractor, including integral fan, was installed where the original fan
had been located. Dust transported at 4500 fpm velocity over a 400 ft length enters the wet
zone of the extractor in less than six seconds. Wet dust extractors instantly mitigate the fire
and dust explosion potential in the dust collector and are recognized as BDT Best Determined
Technology for coal processing and handling operations.
328
Figure 11. Sample room design drawings.
4 ENDING COMMENTS
Readers will have seen an emphasis in this paper on installations where wet dust extractors were
used instead of baghouse or cartridge dust collectors. This is not intended to say that dry dust
collectors should not be considered in a system design. My experience with dust control systems
began in the 1970’s, as a member of an engineering team that developed cartridge dust collectors
for downhole drill rigs used in strip mining. I have worked with hundreds of installations based
on dry collection. However, wet dust extraction system design has been the primary approach
used in my work over the past twenty years and is the main source of my examples.
Designing a safe and effective process ventilation dust control system is a process involving
many considerations. Having team members who understand this process is important to
achieving a successful installation.
REFERENCES
329
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J. Roberts
School of Mechanical, Material, Mechatronics and Biomedical Engineering, University of Wollongong,
Wollongong, New South Wales
J. Hines
School of Health and Society, University of Wollongong, Wollongong, New South Wales
C. Chow
School of Computing and Information Technology, University of Wollongong, Wollongong, New South
Wales
A. Clayton
Coal Services, Wollongong, New South Wales
ABSTRACT: This paper details the development of an immersive simulation tool that can be
used to provide enhanced 3D visualization of monitoring and computational modeling datasets
of dust and ventilation flow in an interactive, virtual environment, and clearly demonstrate the
exposure impact of certain working condition changes and dust mitigation practices in gateroad
development panels. To achieve this aim, onsite dust monitoring with AM520i and PDM3700
dust monitors and computational fluid dynamics (CFD) modeling were conducted to gain an
improved understanding of dust and airflow characteristics around the continuous miner (CM)
and evaluate different practices for dust controls. An algorithm was then developed to process
large amounts of datasets exported from CFD into a form that can be visualized in a virtual
reality (VR) environment using standalone VR headsets. A VR-CFD-based training tool in the
form of VR headsets has been developed and used as an educational platform to create an
effective communication of WH&S awareness of dust exposure in the workplace for CM oper
ators and provide industry workforce (CM operators and development crews) training that
enhances knowledge learning and operational practices for dust exposure reduction.
1 INTRODUCTION
Long-term exposure to respirable coal dust and respirable silica dust in underground mines
continues to present an occupational health and safety hazard to mine workers in under
ground coal mines (Wang et al., 2019, Ren et al., 2018, Zosky et al., 2016). Airborne dust
monitoring data collected by Coal Services (CS) show that Development/Continuous Miner
(CM) operators have the second highest workplace exposures next to longwall (LW) workers
when compared to the workplace exposure standards (WES) (Boyne, 2022). A similar trend
has also been reported in Queensland underground coal mines. The solution to effective miti
gation of dust exposure depends not only on effective engineering controls, but also on on-going
DOI: 10.1201/9781003429241-35
330
innovative education and training programs for the employees on-the-job, to be able to identify
changing conditions, understand the potential impacts of operational changes, and know how
they should react to correct the situation in a timely manner.
Currently, the majority of mining machines for developing gateroad is 12CM30’s. Focusing
on the 12CM30 only will reduce the variability in the existing data while still maintaining
a large data set that can be interrogated. An initial review of the existing NSW data from
2018- 2021 shows that there are 910 (respirable coal dust) and 700 (respirable crystalline silica)
sample results taken from the breathing zone of bolting rig operators and miner drivers while
conducting operations on 12CM30 single pass miners (Coal Services, 2021). During 2021,
there were 2544 respirable crystalline silica (RCS) samples collected. Of these, 862 were taken
from the breathing zone of bolting rig operators and miner drivers while conducting oper
ations on 12CM30 single-pass miners. 5% of the total samples exceeded the workplace expos
ure standard (WES) of 0.05 mg/m3 and 39% of these were on the CM. Dust exposure within
the development panels where the CMs operate can be attributed to a number of different fac
tors. As shown in Figure 1, the top three contributing factors to high dust exposures are high
lighted in a red rectangle, including cutting stone, operator positioning and poor ventilation
standards, which will be used to develop the major dust scenarios of this study.
Figure 1. NSW Order 42 12CM30 Airborne dust exposure standard exceedance contributing factors
July 2018-June 2021 (Coal Services, 2021).
The evaluation of dust exposures within underground mines from historical data allows the def
inition of three overarching scenarios in development panels with due consideration of the top
three contributing factors to be extensively studied in mine site dust monitoring, computational
fluid dynamics (CFD) modeling, and virtual reality (VR) development process. A VR-CFD-based
simulation tool is finally developed and inbuilt into a VR headset that has been taken to mine
sites to train users on the hazards of dust emissions and the best practices for dust control.
2 DUST MONITORING
Mine site dust monitoring was conducted in an underground coal mine situated near Sydney,
Australia, where the development panels are tunneled by a 12CM30-type CM, and cutting
coal is transported out of the heading face by a shuttle car (SC), with the layout of develop
ment gateroad being shown in Figure 2. The gateroad is 5.2 m in width and 2.9 m in height.
The fresh air is provided from the travel road, with some streams of air flowing towards the
heading face of the travel road and some air passing through cut-through (CT) 15 and then
diverting into the belt road. The harmful dust produced in the process of coal cutting, trans
porting and dumping process is sucked into the exhausting ventilation tube.
331
Real-time monitors are specific types of direct-reading instruments that detect, react and
measure a specific contaminant. The measurements are usually data logged and are recorded in
short periods of time with the appropriate response time needs to be defined by the application
(Cauda 2021). The use of real-time monitors is becoming more common for specific monitoring
situations. Real-time monitoring is particularly useful in assessing more accurately where con
trols are needed and the effectiveness of implemented controls. TSI AM520i SidePak™ Personal
Aerosol Monitor (AM520i) and the PDM3700 were determined in this study, both of which are
characterized by battery-operated portable monitoring devices designed for personal monitoring
(to be worn on a worker). However, these two main instruments can also be used for area moni
toring (placed in a location) to gain a measurement of dust concentration.
Dust monitoring instruments were placed in the vicinity of the CM, with the exact locations
shown in Figure 3. Location 1 and 4 were in front of the miner left and right respectively, 2
and 5, mid miner left and right respectively, and 3 and 6 rear of the miner left and right
respectively. The alignment of each sample head was located to be in the same direction and
at the same height.
The continuous data collected from the AM520i was used directly in the CFD validation as
the data was in a format that was easier to manipulate, and more data was collected from this
instrument with seventeen (17) sets of sampling results compared to 7 for the PDM3700,
which allowed a more robust interpretation and modeling of the data. The PDM 3700 results
were limited in number and variation in location.
332
3 CFD MODELING
Based on the mine site condition, a three-dimensional (3D) computational model consisting of
CM, SC and ventilation tube was built, as shown in Figure 4. The fresh air intake marked
with a blue arrow is provided via a cut-through connecting with the heading of interest, while
the dusty air is marked with a red arrow on the return side (away from the CM and SC).
In order to reduce total computational elements and preserve high accuracy, the model was
then meshed using the polyhedral meshing method, and the mesh element size was progressively
reduced closer to the heading face such that the required CFD modeling could be completed with
sufficient accuracy and detail while maintaining computationally appropriate simulation times.
The simulations were optimized to be completed in less than 48 hrs so that future work can allow
further scenarios to be generated quickly with mine-specific setups or scenarios. A mesh independ
ence study was conducted to eliminate the influence of mesh size on simulation results, and the
total elements of the selected meshed model were approximately 2.4 million, as shown in Figure 5.
The minimum orthogonal quality of the meshed model was 0.20, which was of good quality
according to the standard released by the ANSYS Meshing User’s Guide (ANSYS, 2022).
To be able to model any arbitrary scenario for visualization in the VR system, it was neces
sary to verify the CFD model against the real-time dust monitoring data collected from the
mine site. The simulation results were validated against the site data based on a 1-min average
dust concentration. As illustrated in Figure 6, the CFD model correlated well with onsite dust
monitoring data, meaning that the CFD would be sufficient to simulate other scenarios with
confidence in its accuracy. A key aspect of the good correlation between site data and simula
tion data was the inclusion of multiple dust sources to match observations of dust generation
333
during face cutting and loading of the SC; this included from the face, the throat of the miner,
the conveyor, and at the loading point in the SC. This was a significant improvement on previ
ous simulations that have been conducted in the past for this purpose.
With the validated CFD model., twelve scenarios were simulated, including coal cutting at
different sections of the development gateroad (roof, middle and floor) under two different
ventilation conditions where the exhausting ventilation duct was advanced to a standard pos
ition or in a poor position, stone dust cutting at the roof section under standard or poor venti
lation condition, sudden ventilation fan tripping-off, more dust from outbye of the SC,
breaking-away the cut-through as well as cut-through holing into another development gate
road. Only two scenarios are introduced in this paper where the full study can be found in the
final ACARP report for project C33030 (Ren Ting et al., 2022).
Figure 7. Dust dispersion characteristics with coal or stone cutting at the roof section under standard
ventilation.
334
The dust dispersion characteristics with coal or stone cutting at the roof position under
standard ventilation conditions are shown in Figure 7. At t=5 s, it is noted that dust produced
by cutting becomes caught in one of the two recirculating zones around or below the cutting
drum and a portion of high-concentration dust at the left-hand-side (LHS) of the CM is
sucked into the ventilation duct. In addition, in the middle of the cutting drum with stone cut
ting, some dust falls onto the floor due to the entrainment effect of air coming off the drum
and the downward gravity force, which is different from the coal-cutting counterpart. At
t=10 s, dust from the shovel and conveyor belt passes through the gap between the top cover
and beam of the CM and is sucked into the duct with a mass concentration of approximately
35 mg/m3. In comparison to the scenario of coal cutting at the roof section under standard
ventilation, more stone dust at the right-hand side (RHS) of the CM drops underneath the
cutting drum at t=10 s, which is attributed to a larger density of stone dust than coal dust. At
t=15 s, dust from the SC mixes with dust from the shovel and conveyor belt at the top cover
of the CM and then is drawn into the ventilation duct. After 5 s without dust releasing
(t=25 s), it is noted that dust concentration at the LHS of the CM is still higher than 100 mg/m3
for both scenarios, which means more time is required to dilute dust concentration below the
statutory limit.
4 VR-CFD SYSTEM
In order for further visualization in the VR system, simulation data was required to be
exported from the CFD software and reprocessed into a form suitable for VR. To create
awareness of dust and ventilation flow from the viewpoint of mineworkers and to allow mine
workers to better understand their real surroundings and make critical decisions during
mining operations, the real-time visualization of actual dust particles and the ability to inter
act with the data and the virtual environment are required. However, the real-time visualiza
tion of dust particles from the CFD datasets generated for this project is a challenging
problem. This is due to the fact that the data produced by the CFD models for each respective
scenario is enormous. This makes it impossible to render the raw data in a VR environment in
real time. To overcome this problem, extensive research was required to examine context-
preserving sampling techniques to convert and filter CFD data into a form suitable to be
imported into a VR platform, along with intuitive visual representations of the data in an
interactive immersive VR system, with overall workflow shown in Figure 8.
An algorithm was therefore developed for processing large volumes of CFD datasets. The data
processing method included four steps, namely filtering raw data, applying clustering algorithms,
random sampling, and finally visualizing the data in the VR system. Firstly, to filter the raw data,
a dust density threshold was set to remove unneeded data; Secondly, the K-means method was
used to group the remaining data into different clusters by choosing a centroid that minimized the
within-cluster sum-of-squares. Thirdly, a pre-defined amount of data in each cluster was randomly
sampled, and data that was selected in the last frame is given a high probability of reselection to
help stabilize the visualization. Finally, the pre-processed data was installed into a headset and the
particle system of Unity was applied for visualization where each measurement in the data set
corresponds to a particle.
335
Outcomes were significant after data pre-processing, as illustrated in Figure 9. It was evi
dent that although the density of the dataset was reduced with the following processing, the
results still preserved the required fidelity for adequate visualization. Processing of the data
required approximately 10 hours to complete on an i9-10940X CPU where the number of
data points was reduced from the initial 3 million down to 5000 per frame which results in
a reduction in total file size from approximately 230GB to around 0.18GB, dropping by
approximately 99.92%. This process was required for all scenarios that were to be visualized
in the VR environment.
Figure 9. Examples comparing the filtered data with the original raw data.
The final purpose of the proposed VR tool was a self-guided learning experience where the
user could fit the headset and be taken through a predefined learning experience with moder
ate user interaction. The VR tools guided users through a number of scenarios that demon
strate safe and unsafe situations. The dust simulations clearly showed where the dust was
going under different circumstances. Voice-over was used to educate users about what they
were seeing and encouraged them to engage with the content. Users were able to teleport
around the room and view the miniature development panel from many different angles. The
user could also enter the mine at a 1:1 scale to see the simulation from the point of view of an
operator. Finally, the menu showed a quiz button that tested users on what they had learned.
The VR system developed is shown in Figure 10.
5 INDUSTRY SHOWCASES
As a significant task of this ACARP project, several industry showcases and technology trans
fers have been made to project meetings, industry workshops and conferences by using stan
dalone Meta Quest 2 headsets. This enables stakeholders or interested parties to gain a better
understanding of dust and ventilation behavior in various development operations and cutting
scenarios and provide feedback on the developed intuitive VR training tool. Significant
improvements have been made according to feedback and an improved VR training tool has
been developed for stakeholders to start to use in their operations.
336
Figure 11. Industry showcases made to key industry stakeholders and a wider community in the industry.
6 CONCLUSIONS
The aim of this study was to develop an immersive simulation tool for the 3D of dust and
ventilation flow in an interactive virtual environment to demonstrate the exposure impact of
certain working conditions. The advantage of the interactive VR simulation tool is that it can
be configured and customized to various site conditions and equipment, and provide the
industry workforce with training that enhances knowledge and learning of operational prac
tices for dust exposure reduction. To achieve this aim there were various key stages, enabling
technologies and research innovations that were required to produce the end product of the
VR tool; in addition to the VR tool there were various other important and interesting learn
ing outcomes that came from the project which may be useful to the industry.
Overall, the study has resulted in the development of a VR visualization and educational tool
that can more intuitively and comprehensively communicate the results of CFD simulation data
to mine workers. In doing so, a number of key technological achievements were made that can be
used in the future to more quickly translate real-world mine site particulate readings into a VR
visualizer. Along the way, we were able to ensure that we aligned the VR solution to known best
practices in the field, as well as also identify areas of future research where there is more to learn.
ACKNOWLEDGEMENT
This study was supported by ACARP project C33030. The authors acknowledge the support
from various mine sites in NSW.
REFERENCES
337
Mine fans
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J. Fernandez
Deputy Chief Executive Officer, Zitron S.A
1 INTRODUCTION
Energy consumption by mine ventilation systems is higher than ever due to mining operations
being conducted at ever deeper levels. Whilst the increased use of Ventilation On Demand sys
tems (VODs) will in many cases produce significant savings in energy consumption as
a consequence of an enhanced understanding of the ventilation equipment’s performance
under real-time underground conditions, the primary ventilation systems will be required to
deliver higher volumes of fresh air over longer distances, which will result in generating higher
pressures and more powerful fans being required.
To address these challenges, it is vital that mine ventilation practitioners carry out accurate
ventilation calculations, clearly defining all possible system resistances that the mine’s fans
will have to overcome during their design life. Providing an adequate pressure margin to the
fan stall point is a critical consideration during the design process. However, consideration
DOI: 10.1201/9781003429241-36
341
should be given to the fact that excessive pressure margins will impact fan efficiency and there
fore fan energy consumption.
Anti-stall rings (ASR) can severely impact fan efficiency and therefore the use of ASR has
not always been regarded as the preferred solution. However, if the ASR system is designed
correctly, it can enable the safe operation of a fan under severe pressure fluctuations, without
a significant impact on fan efficiency, thereby making the use of an ASR a good choice when
setting a mine’s fan selection strategy.
This paper provides a brief description of the axial fan stall phenomenon, commenting on
the alternatives available to control and manage the fan stall, and reviewing the data from
two axial fan performance tests:
1. Test #1: An aerodynamic fan test, where different blade pitch angles are tested with and
without an anti-stall ring fitted to the fan.
2. Test #2: A strain gauge test, measuring the mechanical and electrical impact of pressure
fluctuations on axial fans with and without an anti-stall ring fitted.
When a fan is operating in a stalled condition, severe separation of the flow lines along the
profile of the blade occurs, and this induces a strong decrease in the increasing pressure pro
duced by the fan, negatively impacting fan efficiency, increasing radiated noise and the mech
anical stress on the blades, which may lead to catastrophic fan failure. Some of the main
problems arising from the fan stall condition are described below:
• The fan cannot deliver design-duty performance and generates more noise.
• Fan efficiency is significantly impacted, which in turn increases the fan’s energy
consumption.
• Increased vibration levels could lead to a fan/motor premature failure.
• Stall operation may result in fan blade rupture due to metal fatigue.
The effects of fan stall are not the same for all fans. Stall severity depends on two main
factors: (i) pressure developing capacity of the fan; and (ii) impeller pitch angle.
Impellers with a higher hub-to-tip ratio and a larger number of blades can develop higher
pressure. However, they are subjected to a more severe stall operation, with a significant dip
in the stall region (Figure 1).
Low hub-to-tip diameter ratio fans with low pitch angle settings are the preferred option
for minimizing the impact stall in axial fans (Figure 2). However, it must be noted that select
ing fans with a low pitch angle will impact fan efficiency, which in turn increases energy con
sumption. Selecting these fans as a strategy to minimize the risks associated with fan stall may
342
result in oversized fan diameters and higher motor speeds to compensate for the lower pres
sure developing the capacity of the fan. Caution must be taken with these fan selections, as
they have higher centrifugal forces and tip speeds which may impact the life of the fan.
The impeller pitch angle is also an important contributing factor to stall severity. Figure 3
shows the aerodynamic tested performance of a medium-pressure axial fan. The tested model
has a 2.5-m impeller diameter, 1,250-mm-diameter hub, and 450 kW motor-rated power. As
reflected in Figure 3, the 2.5-m-diameter tested fan will show some kind of ‘non-stall’ charac
teristic with a -10° pitch setting.
A commonly used strategy to avoid fan stall is to select a fan with high-pressure developing
capacity, allowing for a significant pressure margin with the design duty point(s). Typically,
mine primary fans are medium to high-pressure types, which is relevant because higher-
pressure fans normally have their peak efficiency in areas closer to their stall limit. Therefore,
selecting a fan with a large pressure margin to the stall point may be less efficient at the design
duty point(s). Three (3) options available to avoid a fan’s operation in stall are described in
this paper.
343
3.1 Option 1: Using stall detectors
A simple and cost-effective system for fan stall detection is the Petermann probe system
described in Figure 4. The pressure variations in the stall zone of an axial flow fan can cause the
airflow to be blocked or flow in the opposite direction. The Petermann probe detects the stall
condition of the fan by measuring the pressure difference between the total air pressure in the
opposite direction of rotation of the fan impeller and a reference pressure (static pressure at the
wall of the air duct upstream of the fan blades). This pressure difference is negligible in the
stable range but increases significantly when the fan enters the stall condition. The graph in
Figure 4 shows the pressure difference between the probes at different air flow rates, indicating
a characteristic change at the junction between the stable and unstable ranges of the fan.
3.2 Option 2: Bringing the fan out of the stall while the fan is in operation
The mine ventilation system resistance may deviate from the expected due to miscalculations
or unforeseen mine developments. In Figure 5, a primary axial fan was designed for a duty
point of 300 m3/s and 4.2 kPa, corresponding to a system resistance of 0.046 Ns2/m8. If the
actual resistance is over twice as high, the fan would stall. By reducing the pitch angle, such as
to +2° or -4°, fan stall operation can be avoided.
It is important to note that it is not possible to take a fan out of its stalled operation just by
reducing the motor speed (Figure 5). Sample fan performance curves at various speeds for the
same blade angle are shown in Figure 6.
Reducing the system resistance or reducing the impeller pitch angle is the only solution. The use
of a stall detector system in combination with a variable pitch in impeller motion and a variable
speed drive to adjust the motor speed as required, provide a significant degree of flexibility to the
ventilation system and is a good solution to protect the fan against fan-stalled operation.
344
Figure 5. Excessive system resistance, fan stalled operation.
The effect of the ASR system on the fan performance curve is to eliminate the severe dip in
the stall region. Once the severe separation of the flow lines along the blade profile of the
blade occurs, the ASR guide vanes will re-inject the flow in the axial direction. The images in
Figure 8 show the flow lines at the blade tip for the 2.5-m-diameter impeller fan described in
this paper. Three different fan duty point scenarios have been studied, each with and without
the ASR fitted to the fan:
• Scenario 1: fan operating duty point within the normal fan operating range.
• Scenario 2: fan operating in an early stage of a stall.
• Scenario 3: fan operating under a very developed stalled condition.
345
Figure 7. Axial fan impeller with ASR.
From the figures below, it is possible to observe the area impacted by the stall phenomenon
(flow in the reverse direction) and how part of this flow is injected into the ASR chamber and
re-injected in the correct direction (left to right in the figures below).
In the case of a very developed stall (scenario 3), it is possible to observe that a big area is
impacted by the stall, while with the ASR fitted the stall-affected area is significantly
smaller.
Figure 8. CFD simulation axial fan in operation with and without ASR.
346
4 TESTING THE STALL IMPACT ON AXIAL FANS
In order to analyze the impact of the stall effect on axial fans, two (2) tests have been carried
out. The first test is aimed at understanding how the ASR impacts fan aerodynamic perform
ance and is carried out in the Zitrón test tunnel at its manufacturing operations in Gijon,
Spain.
Measuring 100 meters in length and 52 square meters in cross-section, this is the largest cer
tified test tunnel in the world, allowing axial fans up to 5.8 meters in diameter to be tested at
full power, including up to 2,000 kW, whilst reproducing any real ventilation circuit
resistance.
4.1 Test #1: Measuring the aerodynamic benefits of anti-stall rings (ASR)
Once the aerodynamic performance test of the 2.5-m impeller diameter fan described in this
paper is undertaken with the ASR casing fitted, a steel cover plate is used (Figure 9) to test the
performance of the axial fan without ASR.
The graph in Figure 10 below shows the fan aerodynamic test results when ASR is fitted.
Test results conducted without ASR (Figure 3) are now compared with the test results in
Figure 10. For the particular fan tested, the ASR system provides significant benefits when
the impeller is set to +10° pitch. However, the benefits are not so significant when the impeller
pitch is set to -10° (Figure 11).
It is also important to highlight that the impact on peak efficiency is higher at -10° pitch. It
must be noted that the ASR system is designed to operate at its maximum efficiency on
a specific blade pitch angle, therefore fan pitch adjustments on site (i.e., to accommodate for
future mine resistances) may impact the ASR effectiveness. The ASR efficiency impact in low-
pitch angles will be different from other ASR designs.
Other important considerations when designing ASR systems for mining primary fans are
the air speed through the ASR and its size so that potential blockages due to dust, and other
typical mine pollutants are avoided.
4.2 Test #2: Measuring the mechanical and electrical impact of pressure fluctuations on axial fans
This second set of tests is aimed to analyse the impact of pressure fluctuations on axial fans
with and without a fitted ASR. In a primary mine fan, there are several possible scenarios that
could impact the axial fan performance and its mechanical strength. Important natural
347
Figure 10. A 2.5-m-diameter fan with ASR.
Figure 11. A 2.5-m-diameter fan, +10 & -10 pitch with and without ASR.
348
ventilation effects, piston effects generated by the circulation of large vehicles through
declines, or the influence of other ventilation fans in the system, may induce dangerous pres
sure fluctuations for mine primary axial fans.
Fitting an anti-stall ring will minimize the impeller fatigue stress suffered by the impeller
blades during the fan stalled operation. However, it must be noted that although the ASR will
significantly reduce the fatigue stress on the blades, it does not eliminate it completely.
The fan pressure fluctuation test setup is described in Figure 12. It is composed of
a ventilation circuit including flow straighteners, a closing valve (guillotine), and fan perform
ance monitoring instrumentation (flow, pressure strain gauges on blades, etc.). The change in
the system resistance is achieved by the operation of an auxiliary fan and a control damper.
Figure 12. Test bench arrangement for pressure fluctuations on axial fans.
The main objective of the test is to measure the impact of pressure fluctuations on axial
fans. The axial fan chosen for the strain gauge test has a 2.2m impeller diameter and the test
ing facility is operated to analyze pressure fluctuations of +35%/- 70% of the fan design duty
point (Figure 15).
Figure 13. Tested pressure fluctuations on the 2.2-m impeller diameter axial fan.
The strain-stress is measured (Figure 13) by installing two (2) strain gauges on the blade
pressure side and another two on its opposite side (pull/push forces). The alternating stress
produced by the variations in pressure is measured (Figure 14) under four (4) possible fan
operation scenarios: fan running at full and half nominal speeds, with and without ASR
fitted.
349
Figure 14. Blade fatigue stress measurement.
The fan tested was originally designed to operate with a duty point of 100 m3/s and 1.5 kPa
total pressure. Several fan operating scenarios are tested, with system resistances of +500 Pa
and – 675 Pa over the system design resistance (Figure 15).
The fan performance and mechanical stress of the fan when running at full and half nom
inal speed and under higher and lower system resistances were analyzed. Below is a summary
of the analysis of the mechanical stress test results:
• The average stress on the blades is proportional to the pressure variation on the fan.
• The amplitude and fluctuations of stress are dependent on the fan operating duty point.
• The stress fluctuations frequency has two components (Figure 16): the fan rotation speed
(16,6 Hz – 993 rpm) and the blade natural frequency (112 Hz).
The speed change cycles impact blade fatigue life, therefore reducing the amplitude of the
stress fluctuations is very important to assure a long design life of the equipment. The test
results show that the anti-stall ring system reduces the amplitude of the stress fluctuations,
350
Figure 16. Frequency of stress fluctuations.
particularly when the fan is running in the unstable region of the fan performance curve.
Table 1 below shows the summary of the peak alternating stress measured during the tests.
When the fan is running at full speed (993 rpm) and with the ASR fitted, the peak alternat
ing stress is 6.1 MPa (Figure 17). In this test, the blade’s peak alternating stress increased to
8.1 MPa (33% increase) when the fan was tested without a fitted ASR.
Figure 17. Alternating stress - fan running at full speed 993 rpm with ASR.
According to fan laws, the pressure developed by the fan is proportional to the square of its
running speed. It is therefore evident that a +500 Pa pressure increase will have a higher
impact on the blades alternating stress when it is operated at lower speeds.
351
When the fan was operated at half of its nominal speed (500 rpm) and without a fitted ASR,
the increase in alternating stress ranged from 8.1 MPa (993 rpm) to 10.5 MPa (500 rpm). This
is a 30% increase in alternating stress produced by the system resistance increase.
When the fan with a fitted ASR was operated at 500 rpm, and under the +500 Pa increased
system resistance, the peak alternating stress was 6.2 MPa, which represents a 40% reduction
of maximum stress when compared to the test without ASR (10.5 MPa).
For this paper, only the alternating stress on the fan is analyzed. However, from the vari
able speed drive electrical readings, we can observe that significant negative pressure fluctu
ations, particularly when the fan is running at lower speeds, could make the fan operate in
a negative pressure part of its performance curve. This effect could damage the VSD (DC
link) and the use of regenerative VSDs may be needed.
5 CONCLUSION
The impact of a stall is not the same on all axial fans. Fan aerodynamic design is very relevant
for its behavior under stalled conditions. However, its impeller setting is also important. While
the higher-pressure fans, with higher hub-to-tip ratios and shorter blades, are typically sub
jected to more severe effects on a stall, any axial fan configured with a low blade pitch angle
will have a much better response to the stall than a fan with a higher blade pitch angle setting.
It must be noted that the results presented in this paper are only applicable to the specific
fans tested. However, they may be used to establish best practices when selecting and operat
ing axial fans.
There are several alternatives to avoid fan stalled operation. Selecting fans with a higher-
pressure developing capacity and therefore bigger pressure margins to the stall area, may be an
option. However, it must be noted that this selection strategy impacts the fan operating efficiency
and therefore the power consumed by the ventilation system. Sensors based on the Petermann
probe are very effective to detect the fan stalled operation, shutting down the fan as required.
However, for those cases when the fan cannot be stopped during its stalled operation or for
scenarios when the axial fans are subjected to severe pressure fluctuations, the anti-stall ring
may be a good solution.
From an aerodynamic point of view, ASR systems will allow the continuous rise of the fan
characteristic back toward zero flow. However, there are cases where the impeller design angle
is so low, that the aerodynamic benefits of the ASR are negligible. From a mechanical point
of view, the ASR systems do not eliminate the impact on the impeller blades. However, they
reduce it. ASR systems may be particularly important for those cases where the fans are sub
ject to high-pressure fluctuations while running at reduced speeds.
Fan factory testing, in combination with the manufacturer’s system calculations and fluid
dynamic simulations, results in a very accurate prediction of any possible fan operating condi
tion. Consideration must be given to the requirement for full fan testing, including the fan
unstable area of operation at the factory. Fan tests at various speeds and pitch blade angles in
a controlled environment, provide very useful information, allowing ventilation system practi
tioners to make any necessary improvement to the system on-site as required.
ACKNOWLEDGEMENTS
The author would like to acknowledge Roberto Arias, Susana Rodriguez, Jorge G. Norniella,
and Adrian Salvado, for carrying out the fan aerodynamic testing and strain gauge testing, as
well as finite element analysis that has allowed the author to prepare this paper.
REFERENCES
Eurovent 1/11 – 2007. Fans and System Stall: Problems and Solution.
Piston effect testing results 2021, Zitron S.A.
Factory acceptance test Model ZVN 1-25-450/6 (manufacturing order 23444), Zitron S.A.
352
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
John Bowling
SRK Consulting, USA
ABSTRACT: Fans purchased for mines are typically specified for the projected maximum
operating point that they are expected to perform at or close to it. It is common to use an
overall efficiency for the fan installation to predict power consumption. While the fan manu
facturer recommends a fan based upon an efficient operating range, there are, however, sev
eral factors that can affect the fan system efficiency such as poor installation practices,
operating the fan outside the predicted operating range, and poor maintenance. In this study,
a set of data collected from main and booster fan installations is used to provide additional
information with the intent of improving the understanding of how to determine the range of
fan system efficiencies. The study discusses how data were measured and sorts the data so that
one may compare the differences between fan installations at coal and metal operations.
Keywords: mine ventilation, planning, fans, efficiency, coal, metal/nonmetal, fan installa
tions, main fans, booster fans
1 INTRODUCTION
Primary fan installations are often among the most energy consuming fixed equipment at an
underground mine. Power costs are largely influenced by fan efficiency. That is why it is
important for planning purposes to be able to correctly estimate power consumption for fans.
In most cases, a fan is sized for the maximum operating duty over its life span, which is in
most cases based on life-of-mine (LOM) planning and matched to a fan that operates in an
efficient range at its LOM or peak duty. This operating point is generally developed by
a ventilation engineer using mine projections to predict the required airflow and the total pres
sure difference required to overcome the losses occurring in the airways and at the intake and
discharge. This pressure loss often doesn’t include any losses associated with components of
the fan installation itself such as ducting, dampers, screens, silencers, etc. Losses due to the
components are in fact part of the “fan system efficiency” (ASHRAE, 2020). Fan system effi
ciency is typically assumed to be 65% to 75% for planning purposes, or efficiency values from
manufacturers’ fan performance curves are used without considering the additional losses
associated with the fan installation. Using measured data from 84 unique fan installations,
this study examines fan system efficiency calculated from real-world examples.
2 BACKGROUND INFORMATION
DOI: 10.1201/9781003429241-37
353
This is done by measuring the airflow quantity through the fan and the fan total pressure.
The second part is determining the actual absorbed power of the fan. This is accomplished by
measuring the voltage, current, and, if only apparent current and voltage are measured, the
system power factor. Using these values, the operating power of the fan can be calculated using
Equation 2 (Feick, 2005). System efficiency is calculated by dividing the air/output power by
motor absorbed power, as in Equation 3.
Where:
PO = output power or air power (kW)
Q = quantity of airflow (m3/s)
PFT = fan total pressure (kPa)
Where:
PM = motor absorbed power (kW)
Vavg = average phase voltage (V)
Aavg = average phase current (A)
PF = power factor (unitless)
Where:
η = efficiency (unitless)
PO = output power or air power (kW)
PM = motor absorbed power (kW)
This measurement is system efficiency, which for the purpose of this analysis includes all
the components of the fan installation losses, not just the fan’s aerodynamic efficiency (which
is reported on the manufacturer’s performance curve). The components that make up the fan
system losses are illustrated in Figure 1. Component losses included in fan system efficiency
can be described as follows (ASHRAE, 2020):
• Inlet Losses: Shock losses due to duct inlet and other duct components upstream of the
impeller, e.g., silencers, dampers, changes in duct orientation, size, or shape.
• Motor Loss: Electrical inefficiency of the motor operating at the given speed and torque/
load.
• Drive Loss: Mechanical losses in the drive train, i.e., slippage and bearing friction.
• Mechanical/Aerodynamic Loss: Loss of useful fluid energy due to aerodynamic factors of
the fan impeller and guide vanes, including drag and vortex induction.
• Discharge Losses: Shock losses due to duct outlet and other duct components downstream
of the impeller, e.g., silencers, dampers, changes in duct orientation, size, or shape.
A detailed treatment of calculation or estimation of these losses, e.g., computation of shock
losses, explanation of electrical motor and drive efficiencies, can be found in many compre
hensive reference texts[1,3]. Estimation of the contributions of each of these loss types was not
considered in the scope of this study. Note: for any given flow rate, pressure loss and power
loss are used interchangeably as “loss”, as the two variables are linked by Equation 1.
Fan measurement ports are often unavailable or nonideally located for measuring the oper
ating pressure of an installed mine fan. Therefore, the actual fan total pressure in Equation 1
is often higher than indicated by typical in-situ measurement, but for the purpose of this ana
lysis, any measurable fan system pressure is treated as the “fan total pressure” for the purpose
of computing output/air power. For underground installations, applied pressure from the fan
354
Figure 1. Components of the fan system losses.
bulkhead is often used. In-situ measurements typically do not allow for discrete quantification
of the internal losses of the fan and other system components (e.g., silencers, diffusers). The
motor efficiency is also incorporated into the power reading which results in additional losses
being unaccounted for. The combined system efficiency, if plotted on a generalized fan per
formance curve, might look like that shown in in Figure 2.
Figure 2. Idealized axial fan performance curve (total pressure and efficiency vs. airflow).
355
were collected from a wide range of underground mine sites, with measurements representing
a variety of fan sizes and duties installed both on surface and underground. Mine sites repre
sented include coal, metal, and nonmetal mines primarily in North America with some in South
America and Europe. The range of operating points of these fans are shown in Figure 3.
For underground fans installed in bulkheads, typically the static pressure was measured
with a digital manometer by placing a tube through the bulkhead. This measures pressure
applied to the ventilation network and therefore the fan system applied pressure.
For fans installed on surface in ductwork, a digital manometer connected to a pitot tube
was used to measure pressure through ports in the fan housing. For an exhausting fan, the
total pressure was measured at (as near as practically measured) the inlet side of an exhausting
fan was measured with a digital manometer. For practical reasons, this measurement was trea
ted as the fan total pressure. A similar method was used to measure forcing fans, though the
measurement ports were located at the fan outlet. These measurements are not perfect read
ings but provide enough resolution to provide an estimation of the fan operating conditions.
Power measurements for volts and amps were typically collected from the fan motor PLC
display at the time of fan measurement or from site electricians during the site visit. The
power factor was often estimated for the entire mine site – not necessarily the actual power
factor of the fan or simply unavailable. For other fans, the PLC displays power factor and
only active amps and volts are displayed. For fans driven by variable frequency drives
(VFDs), drive input power is measured and VFD inefficiency is accounted for in the power
consumption. For fixed-speed fans (those not using a VFD), the total absorbed power may
include the reactive power if the fan current and voltage data recorded were corrected to
active current and voltage.
United States coal mines must locate their primary fans on surface, a requirement which
tends to concentrate the total fan power into fewer larger fans compared with metal/non-
metal mines.
356
3 RESULTS AND ANALYSIS
3.1 Overall
The overall results combine fans of all motor sizes and installation types and from both coal
and metal mines. A total of 84 unique fans were used in the total. A visual representation of
the results is shown in Figure 4. The results show that there is a very wide spread of fan effi
ciencies and lower values than the defaults provided by popular ventilation network simu
lators. The software values are meant to represent only typical aerodynamic efficiency, except
where labelled specifically. Simple linear regression shows a weak trend of increasing efficiency
with increasing fan output/air power (dotted lines on Figure 4) for both coal and metal fans,
though a typical efficiency of nearly 10% higher for coal mine fans.
Figure 4. Plot of fan system (static efficiency) vs. output/air power, by commodity type.
357
Figure 5. Comparative histogram of fan system efficiencies.
4 CONCLUSIONS
When planning for the power requirement of a fan installation it is important to understand the
contributors to fan system efficiency and what assumptions the value includes. For mining appli
cations, the fan system static efficiency is the relevant metric; except for jet fans, air momentum is
358
regarded as contributing no useful work to the ventilation system (Jorgensen, 1999). The fan
system (static) efficiency is comprised of duct and other system losses associated with the fan
installation as well as motor and drive efficiency and the aerodynamic efficiency of the fan (impel
ler). The fan system efficiency is calculated by dividing the resulting air power the fan applies to
the ventilation system divided by the total electrical power absorbed by the fan system.
Default simulation software values, published fan curves, and manufacturers’ quoted fan effi
ciencies all contribute to confusion regarding what these values represent. Fan efficiency as read
from the manufacturer’s performance curve typically represents only aerodynamic efficiency,
one component of overall fan system (static) efficiency. The fan system (static) efficiency is
affected by many factors such as the inlet and diffuser accessories, silencers, dampers, duct tran
sitions, and/or unfavorable velocity distribution at the fan inlet. Performance and efficiency can
also change over time due to blade wear, deformation of duct components, and wear or damage
to motors and drive linkages and vary as a function of the fan speed/percent of full load.
Operating points and power consumption data were gathered from 84 unique primary fan
installations as part of ventilation surveys to create or update ventilation system simulations.
These data were reviewed collated to provide a guide for planning for the power demand for
main fan installations. The results in Table 1 show that the actual system efficiency for main
fans is lower than what would traditionally be expected. Planning software has default ranges
between (65% and 76%) depending on how the fan was modeled. The fan system efficiency is
important in planning for the total power demand the ventilation system has on the power
grid and directly relates operating power costs.
A notable difference is observed between coal and Metal/Non- metal efficiencies with coal
averaging about 13% higher. The main contributing factors between the two is that typically
only one or two fans at coal mines are installed on surface with a well-engineered ducting
system that is easily accessible to maintain. Metal mines often install fan underground in bulk
heads where additional losses can be generated due to an under-engineered installation (e.g.,
due to lack of space for an inlet bell or diffuser) or subsequent damage to duct components.
Metal mines often install several fans.
Shortcomings recognized in this study include a relatively limited dataset, including lack of
exact description of fan measurement points and method for each fan. The authors acknow
ledge that further detailed descriptions of the measured fans would provide for more consist
ent comparison among fan installations.
The collected data suggest that many primary fans are operating at a lower efficiency than
expected. It is common for design purposes to expect a fan operating with a system efficiency
of 75% and fan manufacturers often quote expected operating efficiencies of 80% or higher.
These values account for the aerodynamic and perhaps motor/drive losses, but typically do
not account for other losses necessary in a fan system. Possible reasons for this discrepancy
include the following:
• Fans are not operating at the designed operating point. The fan may have been specified
during an early phase of the mine’s development and the expected total mine resistance
curve is not as expected.
• The fans may need maintenance. Blade and bearing wear can attribute to additional mech
anical and aerodynamic losses.
• Fans/fan system components are poorly installed. The fan system could be installed such
that excessive losses are experienced at the inlet or discharge.
359
Ventilation simulations may be hastily performed for the purpose of specifying fan duties
and may not include all these system components, which often reside in a “gray area” between
the modeler’s assumptions and the fan manufacturer’s proposal. Duct and other system losses
may be missed in the specification of fan systems if the system purchaser does not understand
the battery limits. The authors recommend that:
• Purchasing mine engineers review/commission a third-party review of assumed fan efficien
cies when these assumptions are used for motor sizing and power requirement estimation.
• If a less-detailed model is used (e.g., at an earlier study stage), a lower system efficiency
should be assumed for estimating power requirements to allow tolerance for poorly defined
inlet and discharge losses.
• As study detail progresses, the fan system efficiency value should increase to approach
the manufacturer’s aerodynamic efficiency, especially as the fan model and more of the
components comprising inlet and discharge losses are defined and represented separately
in simulations.
REFERENCES
ASHRAE. 2020. 2020 ASHRAE Handbook: HVAC Systems and Equipment, SI Edition. Atlanta, Geor
gia: ASHRAE.
Feick, T. 2005. Fan Primer, 2nd Edition. Zweibrücken. Bad Hersfeld, Germany: TLT-Turbo GmbH.
Jorgensen, R, Ed. 1999. Fan Engineering, Ninth Edition. Buffalo, New York: Howden Buffalo.
McPherson, M.J. 2009. Subsurface Ventilation Engineering. Fresno, CA: Mine Ventilation Services, Inc.
360
Mine fires and explosion prevention
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
B. Belle
Anglo American Steelmaking Coal, Australia
University of New South Wales, Sydney, Australia
University of Queensland, Brisbane, Australia
University of Pretoria, South Africa
R. Balusu
CSIRO Mineral Resources, Australia
ABSTRACT: The evolution of major coal oxidation and resulting sponcom incidents are
sudden and may result in catastrophic negative safety outcome or result in the withdrawal of
persons and closure of longwall panels/mines. Historically, gassy longwall workings in Austra
lian Goonyella Middle (GM) seam (late-1990’s to mid-2010’s), experienced increasing trend in
CO levels associated with coal oxidation and sponcom indicator gases and major safety inci
dents were due to oxygen ingress on the maingate side. The original Australian active longwall
goaf gas drainage system designs are based on the past work of the CSIRO, supported by the
operational experiences. The CSIRO based studies in gassy and hot coal mines had carried
out numerical and field data investigations on goaf hole gas flow mechanisms and proactive
inertisation strategies for preventative spontaneous combustion and goaf gas management.
This critical foundational knowledge work contributed to the original goaf gas drainage and
sponcom management strategies in other Australian longwall mines and potentially extended
to rest of the world. Considering the risks associated with sponcom, GM seam operations
were the first operations in Australia to introduce proactive N2 injection along the MG in
mid-to-late-2000’s, to manage sponcom fire and explosion risks in an active goaf, despite not
having field data readily available to calibrate the models prior to its implementation.
Over two decades ago, active goaf gas drainage flow rates were moderate (2,000 l/s to 3,000 l/s)
and the oxygen ingress on TG side was not a major concern. However, with increasing goaf gas
drainages rates and manual or automated mode operation of goaf wells to extreme flow rates to
address higher longwall goaf gas emissions, TG oxygen ingress and air wash zones became
a major issue recently, necessitating the introduction of TG inertisation strategies now to address
this emerging issue. Introduction of MG proactive inertisation strategy had ultimately reduced the
number of high CO or intensive oxidation incidents over two decades. This paper provides prac
tical safety benefits of longwall tail gate (TG) inertisation supported by the original computational
fluid dynamics (CFD) modelling studies carried out by the CSIRO. The field verification with
both MG and TG inertisation using proactive N2 injection during various phases of longwall pro
duction and stoppages in an active longwall provides reasonable technical and operational justifi
cations on gas and sponcom management strategy for worker’s safety.
1 INTRODUCTION
The original and modified longwall seam gas drainage system design evolutions are essentially
based on the joint industry led historic initiatives and work of the CSIRO, supported by the
DOI: 10.1201/9781003429241-38
363
GM seam operations (Balusu et al. 2001, 2002, 2004, 2005, 2011, 2017, 2019). The historic
review of goaf drainage introduction in Australia and the rest of the world, drainage design
and operational practices are summarised elsewhere (Belle, 2015, 2017). In addition, the
impact of longwall and TG hole positioning study by the CSIRO for the GM seam operations
is summarised elsewhere (Khanal et al, 2021). The science-based CSIRO studies in gassy,
sponcom prone hot coal mines had been carried out through numerical and field data investi
gations on goaf hole gas flow mechanisms and proactive inertisation strategies for preventa
tive spontaneous combustion management. This critical operational knowledge contributed
significantly to the original goaf gas drainage and sponcom management strategies in Austra
lian underground longwall mines and potentially extended to rest of the coal mining world.
The contradictory nature of the GM seam sponcom led fire and gassy mine requiring maxi
mised goaf drainage capacity systems and the need to reduce oxygen ingress into the MG and
TG active goaf required careful operational strategies. The term ‘sponcom’ used in this paper
is used to discuss the various stages of coal oxidation to the development of fully uncontrol
lable combustion resulting in fire when large quantities of coal left in the goaf due to geo-
technical and mining safety considerations. Considering the risks associated with sponcom,
GM seam mines were the first operations in Australia to introduce proactive N2 injection to
manage sponcom related fire and explosion risks in active or sealed goaf areas. In this context,
it is important to note that inappropriate use of ventilation driven controls for longwall tail
gate gas management that are practiced elsewhere in Australia or in the world, are not neces
sarily apt for gassy, steep geo-thermal gradient and known sponcom prone GM seam mines.
Gassy and known sponcom prone GM seam longwall operations require a greater under
standing the goaf gas behaviour, post gas drainage control strategies for TG gas management,
and well balanced gas and sponcom management strategies of high magnitude gas reservoir
mines. Following paragraphs below highlight the spectrum of scientific applied research based
engineering controls and monitoring systems developed and improved over the last two dec
ades in the Australian longwall operations.
364
At the tailgate (TG) return, an outflow boundary condition was specified in the modelling
simulations. The longwall panel width is 300 m and the roadway width on both maingate
(MG) and tailgate (TG) sides of the face is 5.4 m. The goaf height up to 80 m above the work
ing seam and the floor strata down to 10 m below the working seam is included in all the
CFD models. The CFD models incorporated MG and TG cut-throughs of 5 m in width and
cut-throughs spaced at 100 m intervals along the panel and goaf drainage holes replicated the
drainage conditions of the operating site. The total number of finite volume cells used for
meshing are around 2.0 million, for obtaining the grid independent solutions in simulations.
For visual understanding purposes, methane and oxygen gas distributions patterns in long
wall goafs under two different conditions using operational longwall panel gas emissions and
goaf gas drainage conditions with total gas emissions into the longwall goaf of around 9,000 l/s
with 98% methane (CH4) is shown Figure 2. The total goaf gas drainage rate was around
8,000 l/s, with gas concentration in different vertical goaf holes varying between 80% and
95% with adjacent sealed panel goaf drainage of 800 l/s for a typical U ventilation system in
a 2 gateroad panels. In the methane and oxygen gas distribution color contours below, the
red colour indicates higher gas concentration and the blue colour indicates lower methane or
oxygen gas concentration. As noted herein, the presence of oxygen and continued methane emis
sions are contradictory controls requiring finer balance and continued vigilance in goaf drainage
operations and highly reliable gas trend monitoring.
Figure 2. Methane and oxygen gas distributions patterns in longwall goafs under two different
conditions.
Various parametric studies by the CSIRO verified by the operational data from Australian
mines had indicated that the gradient of the seam, both across the face and along the panel,
had a major effect on the distribution of gasses within the goaf. Similarly, results showed that
gas emission rate and face airflow have a substantial effect on oxygen ingress into the goaf,
particularly into the deep goaf (> 1 to 2 km behind the longwall face). Results show that
intake airflow influenced airwash zone in the goaf with over 10% oxygen concentration levels
has extended further into the goaf with increase in intake airflow. In base case simulations the
high oxygen level zone extended up to 150 - 200 m behind the face, whereas in the case of high
intake airflow the high oxygen zone extended up to 300 m into the goaf (Figure 2).
In the recent years, there is often a preferential emphasis put on the “pressure differentials” or
loosely termed “pressure” in an active goaf for ventilation and gas management. Historic work
in relation to the static pressure distribution in the goaf with traditional practice of gas drainage
from two goaf holes near the longwall face is presented visually in Figure 3 for surface goaf holes
closest to the face operating at the total flow rate of about 1,500 l/s. Results show that with this
type of goaf hole drainage, gas static pressure in the goaf measured at the MG seal builds up to
365
180 Pa, that is sensitive various other mining and natural factors. The time based static pressure
distribution model indicates that all the goaf gas migrates towards the tailgate corner of the goaf
and a major proportion of the goaf gas may escape into the tailgate return airway, particularly
during low barometric pressure periods. Therefore, both the goaf gas drainage strategy and its
operation are paramount in addition to the goaf hole design.
In order to improve the gas drainage system efficiency, deep goaf holes gas drainage strat
egy was introduced into the modelling simulations. Deep goaf hole drainage in this paper
refer to those vertical goaf drainage holes from surface that drain the goaf gas located greater
than 1 to 2 km behind the operating longwall face and yet times even near the longwall start
up face area for very long panels. The static goaf gas pressure distribution in the goaf with
maximized gas drainage strategy is presented in Figure 3. Results showed that the static pres
sure development in the goaf with this strategy is only 80 Pa, compared with 180 Pa in the
traditional strategy scenario. Results indicated that goaf gas migrates to a wider area towards
the tailgate side and only a minor proportion of goaf gas escapes towards the tailgate return.
In addition, the low goaf gas pressure development in the goaf helps in reducing the effects of
changes in barometric pressure on return gas levels. These results indicate that the optimum
gas drainage strategy should incorporate goaf holes near the face as well as deep goaf holes
with optimized increased numbers in the panel in order to improve the gas drainage system
efficiency with large longwall block well retreated with large goaf gas reservoir size.
Figure 3. Goaf gas pressure distribution in the goaf with few operating wells behind the face (Left) to
maximized goaf drainage system (right).
366
would also reduce the explosive fringe gas distribution profile near the tailgate area in the
longwall goaf to manage the explosion risk. The results of the CSIRO CFD modelling simula
tions calibrated with field conditions indicated that there is a significant difference in the
spread of explosive fringe gas distribution profiles in the longwall goaf under 2 gateroad and 3
gateroad conditions, i.e. a significant increase in the spread of explosive fringe (or close to
explosive range) zone in the goaf under 3 gateroad conditions. Based on the results of these
investigations, appropriate strategies have been developed for gas control and minimization of
the spread of explosive fringe gas distribution in the longwall goaf (Balusu, Belle, Tangutiri,
2021). Therefore, based on the extensive operational gas management experience, it is the pre-
drainage and longwall goaf gas drainage management is the primary gas management control
rather than ventilation engineering controls to manage the major gas hazard.
One of the major difficulties in the ventilation and gas flow dynamics is our inability to visualize
the complex likely gas concentration profiles, i.e., methane or oxygen in the active goaf with time
and non-constant retreating longwall. The advances in the CFD numerical calculations have
enabled the industry by providing an understanding of gas management or the extent of ‘air wash
zone’ that is often used colloquially during risk assessment or emergency situations. In this paper,
air wash zone is typically referred to as the concentration of relative oxygen in the goaf atmos
phere aiding the left over coal oxidation, and typically referred to with oxygen concentration
values of 3 % to 21%. Fresh air is the concentration that is representative of longwall fresh air
intake. It is to be noted that due to almost no relative airflow movement that can be measured in
the goaf may also mean the presence of oxygen even at levels of 2 % to 3 %. The CSIRO studies
(Balusu, 2002) provided a visual scenario of potential oxygen distribution in an active LW goaf
and goaf well nearer to the LW face. The field studies have noted that (Balusu et al, 2006) the
oxygen concentration was above 19 % for up to 100 m behind the longwall face and reduced to 6
% at 250 m behind the face in the absence of any inertisation control. This air penetration dis
tances of 250 m to 350 m may be mainly attributed due to poor MG brattice control practices
along with the increased longwall airflow rates for gas dilution purposes and inadequate gas drain
age. The tracer gas studies have revealed that the goaf at 300 m behind the face is highly consoli
dated and does not allow direct travel of air fom the intake side to return side of the TG.
Similarly, methane gas concentration distribution profiles at the tailgate region of different
mines vary significantly depending on the geological, gas, mining and operational conditions. Gas
concentration distribution profiles (Balusu, 2020) at the tailgate area under three mining condi
tions are presented in Figure 4. The white box in the plots show the TG motor area and white line
replicating the Bretby across the face. In addition, the gas distribution profiles even at the same
mine can vary significantly depending on number of conditions, including changes in barometric
pressures, goaf falls, gas emission rates, goaf gas drainage efficiency, face location with respect to
goaf holes and cut-throughs, face creep, floor contours, caving conditions behind the face and in
gateroads, coal production rates, face ventilation, face cutting and chock advance sequences. Gas
concentration distribution profiles and potential flammable gas mixture zones near the tailgate
roadway area of the longwall face are dynamic and complex in nature and varies widely depend
ing on the changes in above parameters during mining operations. Thus the behavior of goaf gas
composition is complex and influenced by mining engineering and fluid dynamics and are to be
assessed by suitably skilled, qualified and experienced expert to provide appropriate guidance to
safe operation of coal mines.
An example comparison of the methane and oxygen gas concentration distribution patterns
in the longwall goaf near the tailgate area in a 3 gate longwall retreat scenario are presented in
Figure 5. Results of this simulations indicate that the methane gas distribution inbye of the
longwall face is close to the explosive range in both cases which reflects the goaf drainage hole
design and their operational effectiveness. The contour scale provides the methane and oxygen
distribution. For example, 0.207 % equal to 20.7 % oxygen and 0.40% equal to 40 % methane.
In the Figure 5, “closed at location 1” signifies the temporary roadway seal. The results of the
CFD modelling simulations indicate that there is a significant difference in the spread of explo
sive fringe gas distribution profiles in the longwall goaf under 2 gateroad and 3 gateroad con
ditions, i.e. a significant increase in the spread of explosive fringe (or close to explosive range)
zone in the goaf under 3 gateroad conditions or partial 3 gate road LW operations.
367
Figure 4. A snapshot of methane gas distributions profiles near the tailgate area of longwalls.
Figure 5. Close up views of methane and oxygen distribution inbye of the longwall face (Balusu, Belle,
Tangutiri, 2021).
368
• When MG proactive inertisation strategy developed by the CSIRO and the operators was
first introduced, there was no precedence in Australia and there was no field data to valid
ate its effectiveness (prior to its implementation). However, it is to be noted that during
longwall operations of GM seams, it’s the additional proactive N2 inertisation strategy that
was essential to successfully manage the sponcom and resulting major fire risks during long
periods of production stoppages due to geo-technical and mining related matters.
• The evolution of major coal oxidation and sponcom incidents are sudden and may result in
catastrophic negative safety outcome or result in the withdrawal of persons and closure of
panels/mines.
• In view of the recent incidents in a number of mines working in GM seam (irrespective of
the cause of the incidents), elevated oxidation reaction may potentially become an ignition
source and inadequate control may result in the undesirable safety outcome.
369
• Goaf gas drainage should be carried out from around maximised number of goaf holes
with the retreating longwall goaf in the panel (including deep holes), instead of the standard
practice of gas drainage from just few goaf holes closest to the face for gassy high produc
tion longwall mining.
• Application of increased suction pressure to drain more gas from goaf holes closest to the
face might result in increased air dilution, without any net increase in gas drainage flow rates.
• The ventilation system in the panel should be designed to minimise oxygen ingress into the
goaf, including immediate sealing-off all the cut-throughs behind the face, MG tight brat
tice control in order to improve overall gas drainage efficiency.
• Oxygen concentration level in the goaf hole flow should be less than 5% for extended
periods of time in goaf holes beyond 100 m from the LW face line to reduce sponcom risk
in the longwall goafs during normal production and long periods of stoppages.
• Gas drainage from adjacent old goafs should also be carried out wherever possible, depend
ing on the goaf gas emission flow rates and adjacent seal strengths (Belle, 2014).
In order to manage the elevated oxidation levels with increased goaf gas drainage in high pro
duction gassy mines to effectively manage the active longwall TG gas levels, following pro
active inertisation strategy was implemented in an operating mine. The optimum locations of
the TG and MG holes for both gas and sponcom management with proactive inertisation for
high gassy mines were based on the original and fundamental goaf gas and sponcom manage
ment work by the CSIRO, calibrated with the operational experiences over the last decade
(Balusu, 2021).
370
3.2 LW gas management and proactive MG and TG inertisation assessment
An extensive data analyses was carried out for the entire longwall panel for the gas and spon
com management effectiveness and are discussed hereafter. Figure 6 shows the profiles of
active longwall goaf gas composition with retreating longwall with likely less than effective
inertisation and the absence of TG inertisation.
Key observations from the analyses are as follows:
• The effectiveness of the N2 injection and the operation of goaf gas flow rate management is
reflected in the gas composition of the active longwall goaf gases, viz., CO, O2, CH4.
• The composition of higher levels of oxygen behind the longwall face in the TG region (upto
300 m behind the face) may be attributed to the very higher flow rates and goaf holes
coming online at the time of goaf formation.
• Similarly, higher daily average CO levels, demonstrate the presence of the excessive and
conducive oxygen rich environment as a result of deeper air wash zone in the TG area of
the goaf contributing towards early oxidation of left over coal in the goaf.
• It is equally noted that the methane concentration behind the active longwall in overall
terms increases in the purity as a result of continued desorption of overlaying undrained
coal seams with greater gas reservoir size. The failure to drain the deeper portion of the
active longwall goaf will eventually travel towards the tail gate region as a result of increas
ing goaf gas reservoir size with the longwall retreat and definitively contributes in higher
gas levels during the steep drop in barometric pressures.
• It is equally important to note that the uncontrolled deep goaf hole operation at very high
flow rates, may further exacerbate the oxygen rich environment deeper in the TG goaf area
in the absence of effective TG inertisation.
Figure 6. Goaf hole gas composition without the TG inertisation with retreating LW face- Oxygen
(Top Left), CH4 (Top right), CO (Bottom Left); Flow rate (Bottom right).
Similarly, Figure 7 shows the profiles of active longwall goaf gas composition with retreat
ing longwall with effective MG inertisation and with the introduction of TG inertisation for
the first time.
Key observations from the analyses are as follows:
371
Figure 7. Goaf hole gas composition with MG and TG inertisation with retreating LW face- Oxygen
(Top Left), CH4 (Top right), CO (Bottom Left); Flow rate (Bottom right).
• Combined MG (1500 L/s) and TG inertisation rate of 750 L/s with greater goaf gas man
agement is clearly reflected in the favourable gas composition of the active longwall goaf
gases, viz., CO, O2, CH4.
• The composition of reduced levels of oxygen behind the longwall face in the TG region
(upto 100 m behind the LW face) may be attributed to the N2 injection in the tailgate min
imising the oxygen ingress into the goaf. There is a significant difference in the oxygen
levels at less than 5 % beyond 100 m of the inbye goaf assisting in minimising the risk of
coal oxidation and oxygen presence in the deeper portion of the goaf.
• Similarly, daily average CO levels in the active goaf have significantly reduced, demonstrat
ing the effectiveness of the N2 injection and less than adequate conducive oxygen presence
in the TG airwash zone, thus potentially minimising early oxidation.
• As before, the methane concentration behind the active longwall in overall terms increases
in the purity as a result of continued desorption of overlaying undrained coal seams with
greater gas reservoir size and aided by the N2 presence. This environment enables the
deeper goaf drainage practices at low to moderate flow rate for maximised goaf gas man
agement and drainage efficiency, to reduce the gas reservoir as well as ventilation air
methane (VAM) and greenhouse gas (GHG) management.
372
Figure 8. Historic MG and TG inertisation conceptual strategy for longwalls (Balusu, 2005).
Figure 9. Latest MG and TG inertisation strategy for an operating gassy longwall mine (Balusu, 2021).
U ventilation system with varying goaf hole operational controls and designs without pro
active inertisation. The goaf drainage operation suggests that greater oxidation may be pos
sible in the TG, while there is minimal occurrence of it at MG area with greater ventilation
controls such as tight brattice controls and timely build up of seals.
Figure 10. A conceptual model of goaf gas environment under intensive goaf gas drainage impact.
The historic field studies (including tracer gas and goaf hole shut off studies) of CSIRO
have shown that reducing and increasing the flow capacity of goaf holes, including complete
shut-off of the holes and measuring corresponding changes in return gas levels and response
times have shown that 80 % to 90% of goaf gas migrated to longwall return within 1 to 3.0 hrs
when goaf holes were 100 to 400 m from the longwall face. Response times varied between
two minutes (150 m behind the face) to an hour (1000 m behind the face) along the MG and
on the TG goaf holes with few minutes to several hours depending on the goaf hole designs
373
and caving characteristics. Field studies have shown that even when goaf holes located more
than 1,000 m from the face, had a substantial effect on gas flow dynamics and on longwall
return gas levels. The recommended strategy was that the deep goaf holes are to be operated
at low to moderate flow rates continuously as long as the oxygen levels below 3 % and no
increased trend in the CO levels.
One of the additional challenges often faced during an unfortunate gas event leads to pleth
ora of suggestions without having appropriate science and data based studies or specialist
expertise in the field. One such view is that the goaf drainage flow rate has no impact on
oxygen in goaf well or active goaf. Figure 11 below shows the relationship between goaf hole
flow rates and goaf hole oxygen levels at various distances from the active longwall face.
Figure 11. A conceptual model of goaf gas environment under intensive goaf gas drainage impact.
It is observed from the above plot that when the goaf holes are deep, the O2 levels are typic
ally very low and lower flow rates help to minimize the goaf reservoir size with the retreating
longwall with major rate of barometric pressure drop. The general observation is that the
increase in goaf flow rate to the extreme regions causes significant increase in the O2 levels in
the goaf, which would require inertisation to minimize the potential oxidation. It’s the moderate
flow rate with increased number of operating wells against few number of wells operating at
extreme flow rates assists in managing the TG gas levels. While the sole intent is to minimize the
TG gas levels, the risk of operating few wells at maximum flow rates behind the longwall face
would certainly bring the fresh air or oxygen into the active goaf of a retreating longwall. It
must be noted that the goaf hole design including its distance from the working seam (historic
design of ideally 10 m or 40 ft as in the USA) will have the impact on TG gas levels as well as
oxygen levels in the active goaf. Therefore, it’s the maximized goaf drainage of active goaf with
moderate goaf flow rates and increased number of holes supported by the active inertisation
would assist in the effective management of gas and sponcom management risks.
Conjointly managing the gas and sponcom risk is fundamental to securing a safe underground
place of work at GM seam longwall operations. With the increasing gassy and known spon
com prone coal seams and a working depth with steep geothermal gradient is contributing
towards the step changes in controls required for gas and sponcom management in order to
be compliant with the safe TG gas limits as well as reduced ventilation air methane (VAM)
emissions. This strategy reinforces the fundamental importance of the pre-drainage systems
with long lead drainage time prior to longwall mining. Over two decades ago in Australia,
goaf gas drainage rates in Qld and NSW were low to moderate (1,000 l/s to 3,000 l/s) and the
ingress of airwash zone on longwall TG side was not a major concern. However, in the recent
374
years, with increasing goaf gas drainages rates upto 6,000 l/s 10,000 l/s and manual or auto
matic mode operation of goaf wells to the extreme flow rates to address higher goaf gas emis
sions, TG oxygen ingressing deeper in the goaf has become a major issue, necessitating the
introduction of both TG and MG inertisation strategies now to address this emerging coal
oxidation risk. Furthermore, contrary to the views in relation to elevated coal oxidation and
sponcom events related to goaf hole spacing and maximized goaf drainage practices, it is pru
dent to note that there have been historic cases of sponcom events with 400 m to 200 m goaf
hole spacing, and even with no goaf drainage practices.
Following safety benefits reasoned with adequate technical and operational justifications
for gas and sponcom management strategy aided with both MG and TG inertisation using
proactive N2 injection during various phases of longwall production and stoppages:
• In the absence of active and continuous proactive N2 inertisation, maintaining 5 % to 8 %
O2 levels in the active goaf would be very difficult and may exacerbate the oxidation in an
active goaf when the longwall retreat rates slows down or stops for weeks and months. On
the contrary, sub optimal goaf drainage to manage the O2 ingress in deep goaf area of TG
region will significantly increase the longwall TG gas levels. This approach would put oper
ations in a dangerous position from gas management perspective.
• Based on the data analyses evidence of manual daily data on O2, it is noted that there is no
clear trend in terms of goaf well spacing justifying the oxygen management, rather the
lower O2 levels in deep goaf or when the LW face is further away from the goaf well, are
conducive to operate at lower flow rates for steady deeper goaf drainage holes for gas man
agement for retreating longwall of growing goaf gas reservoir size.
• It is the operational management of goaf wells (not sudden or automatic operation of the
goaf wells to the extreme flow rates, rather stepwise increase) with proactive LW active
goaf inertisation for sponcom management will enable the appropriate maximised goaf
drainage gas management to manage the longwall return gas levels and oxidation risks.
• The introduction of TG inertisation assists in reducing the airwash zone. In addition, long
wall operations need to continue with the leapfrogging of well-established MG seal inertisa
tion strategy [including quicker and timely MG seal build, and tight MG brattice control],
along with the dedicated tube bundle monitoring points for goaf gas monitoring to under
stand the goaf flow dynamics and sponcom management.
• As a general long-term strategy, known sponcom prone longwall operations to ensure flexible
and contingency inertisation infrastructure is readily available and to remain in a state that it
is able to be recommissioned within a single shift at any time during the monitoring period.
• The association of slow retreat due to known or unknown geological structures in the long
wall hazard plan and known historic oxidation related incidents, have further reinforced
the active inertisation system maintaining at least 2,200 l/s of inert flow into the active
panel [around 1200 l/s on the MG side and 1000 l/s on the TG side using dedicated vertical
N2 holes drilled 5 to 10 m from the working seam height].
• Considering the various uncertainties associated with the LW operations in sponcom prone
seams, recommendation of inertisation holes at 200 m spacing on TG side is essential and
appropriate for long term risk management planning and design purposes. If the evidence
suggests otherwise, i.e., constant increased retreat or no stoppages, then TG proactive iner
tisation may be carried out through inertisation holes at increased spacing (i.e. alternate
inertisation holes) in those areas. This would mean that the intermediate dedicated holes
and other old/deep inertisation holes can be equally used for oxygen ingress or airwash
zone monitoring and sponcom monitoring on the TG side of the goaf as per the oxygen
limit recommendations of an active goaf.
• It is to be noted that these gas and sponcom management strategies are not merely based
on the LW retreat rate, but also includes the inherent nature of seam propensity for spon
com despite it being equally rated as “low” risk sometimes, geo-thermal gradient, existence
of faults and structures, amount of coal left behind, changes to the goaf hole designs, delay
in operational related building of MG seals, inadequate MG brattice leakage control for
long periods. other engineering related uncertainties associated with the LW equipment,
375
strata control uncertainties associated with moisture/water in TG roadways, and uncertain
ties associated with cavity control measures.
• The implemented TG and MG proactive N2 inertisation strategy of active longwall goaf at
the sponcom prone seam has resulted in the new technical/empirical data generation in veri
fying the fundamental understanding of goaf gas drainage maximization and sponcom
management.
• Finally, strengthening the proactive inertisation strategy on both MG and TG with flexible
inertisation capacity and responding to the up to date trigger response values of the oxidation
scenario rapidly developing into an advanced stage using appropriate early monitoring strat
egy is essential for the future proofing of underground sponcom risk.
REFERENCES
Balusu, R., Deguchi, G., Holland, R., Moreby, R., Xue, S., Wendt, M. & Mallett, C. (2001). Goaf gas
flow mechanics and development of gas and sponcom control strategies at a highly gassy coal mine.
Australia-Japan Technology Exchange Workshop, 3-4, December 2001. Hunter Valley, Australia. 18
pp.
Balusu R, Mallett C, Xue S, Wendt M, and Worrall R (2002). Mine Gas Control – JCOAL, CSIRO and
Dartbrook Collaborative project Final report. CSIRO Exploration and Mining Report No. 1012C,
March/October 2002, 183 pp.
Balusu, R., Tuffs, N., Peace, R., Harvey, T., Xue, S., Ishikawa, H. (2004). Optimisation of Goaf Gas
Drainage and Contro Strategies, ACARP Project C10017, CSIRO Exploration and Mining Report
1186F, 149 pp.
Rao Balusu, Ting X Ren and Patrick Humphries,(2005). Proactive Inertisation Strategies and Technol
ogy Development, ACARP Project C12020, CSIRO Exploration and Mining Report P2006/26, 114
pp.
Balusu, R., Schiefelbein, K., Ren, T, O’Grady, P, Harvey T. (2011). Prevention and control of fires and
explosions in underground coal mines, The Australian Mine Ventilation Conference, Sydney, Austra
lia, September 2011, pp 69–77.
Balusu, R., Belle, B., Tanguturi, K. (2017). Development of gas and spontaneous combustion control
strategies for 6.0km long longwall panels, 16th US Mine Ventilation Symposium, Golden, Colorado,
USA, June 2017, pp 18–11 – 18-18.
Balusu, R., Belle, B., Tanguturi, K. (2019). Development of Goaf Gas Drainage and Inertisation Strat
egies in 1.0-km- and 3.0km-Long Panels. Mining, Metallurgy and Exploration. 2019; 36(6):1127–1136.
https://doi.org/10.1007/s42461-019-0071-9.
Balusu, R., Tanguturi, K. (2019). Gas management and risk mitigation strategies for longwalls, ACARP
Project C25066, CSIRO Energy Report EP191657, March 2019, 210 pp.
Balusu, R, Tanguturi, K, Belle, B. (2021). Goaf gas distribution profiles near the longwall tailgate area,
Proceedings of the 18th North American Mine Ventilation Symposium, 12-17 June, 2021, Rapid City,
South Dakota, USA, Edited ByPurushotham Tukkaraja.
Balusu, R. (2020). Longwall Tailgate Gas Compositions, Internal Industry Note, Australia
Balusu, R. (2021). Unpublished CFD Model Study on MG and TG inertisation Strategies, CSIRO.
Belle, B, 2014. Evaluation of barometric pressure and cage effect on longwall tailgate gas levels for pre
vention of explosion in coalmines, IMVC, Johannesburg.
Belle, B. (2015). Innovative tailgate mobile goaf gas management in two gateroad longwall panels – con
cept to implementation”, The Australian Mine Ventilation Conference, Sydney, Australia.
Belle, B. (2017). Optimal goaf hole spacing in high production gassy Australian longwall mines – oper
ational experiences, The Australian Mine Ventilation Conference, Brisbane, Australia.
Khanal, M, Poulsen, B, Adhikary, D, Balusu, R, Wilkins, A, Belle, B. (2021). Numerical Study of Stabil
ity and Connectivity of Vertical Goaf Drainage Holes. Geotechnical and Geological Engineering.
2021; 39(3):2669–2679. https://doi.org/10.1007/s10706-020-01650-6.
Xiang, Z, Si, G, Wang, Y, Belle, B, Webb D. (2021). Goaf gas drainage and its impact on coal oxidation
behaviour: A conceptual model, International Journal of Coal Geology, Volume 248, 103878.
376
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
G. Bogin
Mechanical Engineering Department, Colorado School of Mines, Golden, CO, USA
1 INTRODUCTION
Despite alternative energy sources, worldwide coal production is still increasing each year
(Dudley and others 2018). Longwall mining is the most utilized coal mining method, due to its
high productivity and safer operating conditions (Peng 2019). However, in usual mining oper
ations and conditions, coal mining still faces serious challenges (Qiao et al. 2022). Despite the
advancements in technology and safety management, longwall face explosions from accumu
lated methane gas are known to be the most common cause of methane explosions (Juganda
2020a). Existing industry practices depend on point-type methane sensors in critical regions to
prevent explosive gas accumulations (Juganda et al. 2020). However, point sensors are not
reliable at spotting and warning about explosion hazards, especially in crucial areas, such as
near the cutting drum, tailgate, and headgate areas of the longwall face (Davis, Engel, and
van Wingerden 2015). One catastrophic example of a methane explosion accident was the
2010 Upper Big Branch Mine in West Virginia, U.S. (Davis et al. 2015). Although atmos
pheric monitoring systems can report real-time methane concentrations, they fall behind due
to their limited number of sensors and locations (D C Demirkan et al. 2022), which lack full
coverage of the whole longwall face. Computational fluid dynamics (CFD) was employed to
simulate ventilation conditions in longwall faces to reproduce airflow aerodynamics and the
formation of hazardous gas mixtures, which are not detectable with conventional monitoring
and ventilation inspection practices (Juganda et al. 2020). Although CFD modeling can accur
ately predict explosive gas zones, high computational power and time requirements diminish
its use for real-time ventilation monitoring purposes (D C Demirkan et al. 2022).
DOI: 10.1201/9781003429241-39
377
In 2008 Karacan et al. proposed principle component analysis and an artificial neural network-
based approach to predict methane emission rate throughout 63 longwall mines. The study
showed that the volume of daily methane emission from each mine could accurately be predicted
(Karacan 2008). In the following years Dougherty and Karacan utilized the prediction model in
Karacan’s previous work (Karacan 2008) and developed software that can predict ventilation
emissions with elastic properties (Dougherty and Özgen Karacan 2011). In 2018 Duda and Krze
mień proposed a framework for forecasting methane emissions from seams to goafs; and predicted
the average volume of methane per minute for each year of mine life (Duda and Krzemień 2018).
In 2021 Sidorenko et al. provided the necessary parameters to predict methane emissions from
seams to goafs (Sidorenko, Dmitriev, and Sirenko 2021). Although these studies discussed predic
tion, these predictions were neither real-time nor spatial outputs.
Previous studies demonstrated the successful implementation of Artificial Intelligence (AI) in vari
ous fields with real-time predictions. For example, Nyanteh et al. implemented an AI for real-time
fault detection (Nyanteh et al. 2013). Dong et al. used the AI model for real-time monitoring and
prediction of slope failures (Dong et al. 2021). Rodríguez-Rangel et al. incorporated big data analy
tics for autonomous vehicles’ speed estimation (Rodríguez-Rangel et al. 2022). Wahyono et al. pro
posed combining AI with data mining for real-time forest fire detection (Wahyono et al. 2022).
Real-time prediction of methane in a longwall face requires predicting explosive gas zone forma
tion in time, and its location in 3D (x, y, and z coordinates). Predictions in time are mainly con
ducted using time series classification/analysis (Bagnall et al. 2017). Spatiotemporal AI models and
time series classification are relatively new to the field. Moreover, the vast majority of successful
models take into account only one or two spatial dimensions, such as x and/or y coordinates, and
the data of interest (Dempster, Petitjean, and Webb 2020; Deng et al. 2013; He et al. 2016; Ismail
Fawaz et al. 2020; Li and Principe 2017; Lines, Taylor, and Bagnall 2018; Lucas et al. 2019; Schä
fer and Leser 2017; Shifaz et al. 2020). For example, climate change studies that track carbon emis
sions use latitude, longitude, and carbon content (Alléon et al. 2020), whereas water quality
studies use one-dimensional distance of the intersections and water content (Li 2011)
Current explosive gas accumulation monitoring practices in longwall coal mines rely on two
methods, namely, point sensors and CFD modeling. Point sensors take real-time measurements
along the face and provide methane concentration values at the installed locations. CFD modeling
provides methane prediction for the whole longwall face. However, due to computational cost and
load, predictions take days or weeks, depending on the resolution of the study area. This study is
a continuation of our previous works, i) benchmarking and analyzing the suitability of the dataset
and off-the-shelf algorithms (Demirkan 2022), and ii) adopting a modified long-short term
memory (LSTM) algorithm and methodology as a real-time explosion hazard warning system
(Doga Cagdas Demirkan et al. 2022). This study explores the use of a 3D spatiotemporal predic
tion model for real-time prediction of methane in the coal face with a small data set.
2 DATA PREPARATION
Version 18.2 of the Ansys Fluent software was used to model and simulate a longwall face. The
modeled longwall face had a length of 300 meters, a mining height of 3 meters, and a depth of 6
meters. Also represented were two basic pieces of equipment: i) support equipment (shields) and
(ii) cutting equipment (shearers). Each shield was 2 meters long and was affixed to the model. One
10 m long shearer was set in the headgate (HG) part of the long mine and between 265 to 275
meters. The region that was modeled was then covered with about one thousand hexagonal and
octagonal meshes. The simulation exemplified a transient CFD model of methane (CH4) gas emis
sion from the coal face based on a bleeder ventilation system with a tailgate (TG) back return
setup. The transient model was run for 20 seconds with 1-second recording intervals. Raw data
from the Fluent software were preprocessed, which included data conversion into CSV files,
removal of empty fields, the addition of cutting directions, and correction of all fields’ two signifi
cant numbers. The LSTM network, which had been modified in our previous work (Doga Cagdas
Demirkan et al. 2022), was trained and tested with new CFD outputs. The inputs to the modified
LSTM model were; x, y, and z coordinates, airflow velocity, methane concentration, and the
378
volume of each cell for 20 seconds, which were recorded at 1-second intervals. In Figure 1, long
wall mining model along with the shearer location and data collection locations are given.
3 RESULTS
This research utilized a high-end gaming PC with the following specifications: CPU: 10th Gen
Intel Core i7-10875H (8-Core, 16MB Cache, up to 5.1GHz with Turbo Boost 2.0), GPU:
NVIDIA(R) GeForce RTX(TM)2070 8GB GDDR6, Memory:32GB DDR4 2666MHz, Stor
age: 2TB PCIe M.2 SSD. The training of the algorithm took around six hours, and the time
required to make a prediction was twenty seconds.
Depending on the dataset, the usual practice of separating data into train, validation, and
test sets can range from 50–50 percent to 80–20 percent. When there are fewer data points (n
< 10,000), 80%–20% splitting is utilized. With higher numbers of recordings, however (n >
1,000,000), the significance of split ratios diminishes and ratios can be 50%-50% . Overall,
good data representation in the train and test sets is the most critical aspect of partitioning
a dataset (Dubbs 2021; Gholami, Torkaman, and Dalir 2019; Gholamy, Kreinovich, and
Kosheleva 2018; Joseph 2022; Owusu, Bansah, and Mensah 2019). In this study, there were
one thousand recordings; hence, training and testing were split as 80% to 20% for each
instance using the stratified K-folds cross-validation method. This resulted in a balanced data
division that maintained the proportion of samples for each methane content. Eighty percent
of the data was used to train the model.
Validation data were separated from training data, which validated the AI’s performance.
Training and validation accuracy helps users evaluate their mode. Figure 2 illustrates
a commonly used metric for assessing algorithm performance, validation, and training accur
acy versus an epoch.
In the training and validation accuracy graph, the curves’ slopes approached horizontal
after the 15th epoch, which indicates that data did not make a significant learning process
over the algorithm. At approximately the 19th epoch, the learning curve became almost hori
zontal. This indicates that no further training was necessary after the 20th epoch, as the accur
acies did not change considerably and ranged from 83.8% to 85.8%. Lastly, the validation
curve above the training curve indicates that the model was fed a good representation of data,
was ready for testing and was able to provide reliable predictions.
After training and validation, test data were used to evaluate the AI model’s performance in
predicting methane. The model’s performance was analyzed using the actual and predicted
methane content of the testing data as a non-explosive and explosive zone. The results are
given in Table 1.
379
Figure 2. Validation and training accuracy of the algorithm.
As was the case with the Upper Big Branch accident, a longwall mine methane explosion can
be fatal. Current explosive gas zone management practices are carried out either with point
sensors or CFD modeling. Whereas explosive gas zone monitoring relies on point sensors, the
critical regions of the longwall face cannot be tracked in real time. Although CFD modeling
can overcome the sensors’ coverage, the required prediction time could be days to weeks,
depending on the resolution of the simulation. Moreover, training the AI with whole mine
data needs intensive training for accurate results.
Leveraging the power of AI might be crucial for monitoring explosive methane concentra
tion. The primary objective of this study was to combine the advantages of current methane
monitoring practices and reduce their disadvantages. For this purpose, this study utilized
modified LSTM architecture for real-time methane prediction with a small data set and with
the proposed methodology in our previous studies.
Although each mine ventilation condition is unique, previously conducted studies
(KRICKOVIC S and FINDLAY C 1971; Marts 2015; Marts et al. 2014; Schatzel et al. 2012;
Schatzel, Krog, and Dougherty 2011) prove that simulated data (CFD modeling) represents
the actual ventilation that includes but is not limited to “(i) continuous leakage of fresh air
from the face to the gob, and the higher accumulation of methane as the supplied air travels from
the headgate to tailgate side of the face; (ii) higher leakage around the headgate and tailgate
corners of the face due to the high porosity and permeability around the edge of the gob; (iii)
Methane accumulation seems to follow linear regression based on ventilation surveys done in sev
eral longwall operations.” (Juganda 2020b).Therefore this study utilized the CFD data as val
idation while calculating accuracies. As a result, the overall accuracy was 82.30% with a small
data set, n=1000, when compared to CFD results. This concludes that the modified LSTM
380
algorithm could combine compelling aspects of CFD modeling (full coverage and representa
tion) and point sensor measurements (real-time). AI algorithms could achieve 3D coverage of
CFD modeling and real-time point sensor data measurements.,
However, the AI’s prediction capabilities depend on the simulated CFD model results.
Therefore, predictions can only be as accurate as similar longwall face models. Training the
AI model by using different longwall face models can increase the AI model’s capacity. If the
AI model can be trained with more data, it might be used for methane monitoring purposes.
Although the proposed methodology successfully predicted methane concentration throughout
the longwall face, the results only contained numbers representing location, time, and methane
content. Results consisted of lines of numbers that could not be interpreted or used by engineers
and/or miners to determine if the explosive gas accumulation was hazardous.
As future research associated with this study, the authors will train the algorithm by using
other mines’ methane emission models. This is expected to increase the algorithm’s prediction
capabilities and enable its usage and implementation in more longwalls in the world.
ACKNOWLEDGMENTS
This study was supported by the National Institute for Occupational Safety and Health
(NIOSH), grant number 0000HCCR-2019-36403.
This study utilized resources from Colorado School of Mines’ CyberInfrastructure and
Advanced Research Computing (CIARC) team and the University of Colorado Boulder
Research Computing Group, which is supported by the National Science Foundation (awards
ACI-1532235 and ACI-1532236), the University of Colorado, Boulder, and Colorado State
University RMACC Summit.
The authors declare no conflicts of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or
in the decision to publish the results.
REFERENCES
Alléon, Antoine, Grégoire Jauvion, Boris Quennehen, and David Lissmyr. 2020. “PlumeNet: Large-Scale
Air Quality Forecasting Using A Convolutional LSTM Network.”
Bagnall, Anthony, Jason Lines, Aaron Bostrom, James Large, and Eamonn Keogh. 2017. “The Great
Time Series Classification Bake off: A Review and Experimental Evaluation of Recent Algorithmic
Advances.” Data Mining and Knowledge Discovery 31(3):606–60.
Davis, Scott G., Derek Engel, and Kees van Wingerden. 2015. “Complex Explosion Development in
Mines: Case Study—2010 Upper Big Branch Mine Explosion.” Process Safety Progress 34(3):286–303.
Demirkan, D C, S. Duzgun, A. Juganda, J. Brune, and G. Bogin. 2022. “Evaluation of Time Series Arti
ficial Intelligence Models for Real-Time/near-Real-Time Methane Prediction in Coal Mines.” CIM
Journal 0(0):1–10.
Demirkan, Doğa Çağdaş. 2022. “NEAR-REAL-TIME/REAL-TIME METHANE PREDICTION
AND VISUALIZATION IN UNDERGROUND LONGWALL COAL MINING by Doga Cagdas
Demirkan.”
Demirkan, Doga Cagdas, H. Sebnem Duzgun, Aditya Juganda, Jurgen Brune, and Gregory Bogin. 2022.
“Real-Time Methane Prediction in Underground Longwall Coal Mining Using AI.” Energies 15
(17):1–12.
Dempster, Angus, François Petitjean, and Geoffrey I. Webb. 2020. “ROCKET: Exceptionally Fast and
Accurate Time Series Classification Using Random Convolutional Kernels.” Data Mining and Know
ledge Discovery 34(5):1454–95.
Deng, Houtao, George Runger, Eugene Tuv, and Martyanov Vladimir. 2013. “A Time Series Forest for
Classification and Feature Extraction.” Information Sciences 239:142–53.
Dong, Mei, Hongyu Wu, Hui Hu, Rafig Azzam, Liang Zhang, Zengrong Zheng, and Xiaonan Gong.
2021. “Deformation Prediction of Unstable Slopes Based on Real-Time Monitoring and Deepar
Model.” Sensors (Switzerland) 21(1):1–18.
Dougherty, Heather N. and C. Özgen Karacan. 2011. “A New Methane Control and Prediction Software
Suite for Longwall Mines.” Computers and Geosciences 37(9):1490–1500.
Dubbs, Alexander. 2021. “Test Set Sizing Via Random Matrix Theory.” (December).
381
Duda, Adam and Alicja Krzemień. 2018. “Forecast of Methane Emission from Closed Underground
Coal Mines Exploited by Longwall Mining – A Case Study of Anna Coal Mine.” Journal of Sustain
able Mining 17(4):184–94.
Dudley, Bob and others. 2018. “BP Statistical Review of World Energy.” BP Statistical Review, London,
UK, Accessed Aug 6(2018):116.
Gholami, V., J. Torkaman, and P. Dalir. 2019. “Simulation of Precipitation Time Series Using
Tree-Rings, Earlywood Vessel Features, and Artificial Neural Network.” Theoretical and Applied Cli
matology 137(3–4):1939–48.
Gholamy, Afshin, Vladik Kreinovich, and Olga Kosheleva. 2018. “Why 70/30 or 80/20 Relation Between
Training and Testing Sets : A Pedagogical Explanation.” Departmental Technical Reports (CS) 1–6.
He, Kaiming, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. “Deep Residual Learning for Image
Recognition.” Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern
Recognition 2016-Decem:770–78.
Ismail Fawaz, Hassan, Benjamin Lucas, Germain Forestier, Charlotte Pelletier, Daniel F. Schmidt,
Jonathan Weber, Geoffrey I. Webb, Lhassane Idoumghar, Pierre Alain Muller, and
François Petitjean. 2020. “InceptionTime: Finding AlexNet for Time Series Classification.” Data
Mining and Knowledge Discovery 34(6):1936–62.
Joseph, V. Roshan. 2022. “Optimal Ratio for Data Splitting.” Statistical Analysis and Data Mining
(February):531–38.
Juganda, Aditya. 2020a. “Evaluation of Point-Based Methane Monitoring and Proximity Detection for
Methane Explosive Zones in Longwall Faces of Underground Coal Mines.” Colorado School of Mines.
Juganda, Aditya. 2020b. “Evaluation of Point-Based Methane Monitoring and Proximity Detection for
Methane Explosive Zones in Longwall Faces of Underground Coal Mines.” 255.
Juganda, Aditya, Claire Strebinger, Jürgen F. Brune, and Gregory E. Bogin. 2020. “Discrete Modeling
of a Longwall Coal Mine Gob for CFD Simulation.” International Journal of Mining Science and
Technology 30(4):463–69.
Karacan, C. Özgen. 2008. “Modeling and Prediction of Ventilation Methane Emissions of U.S. Longwall
Mines Using Supervised Artificial Neural Networks.” International Journal of Coal Geology 73(3–
4):371–87.
Krickovic S and Findlay C. 1971. “Methane Emission Rate Studies in a Central Pennsylvania Mine.”
U S Bur Mines, Rep Invest 7591.
Li, Kan and Jose C. Principe. 2017. “Transfer Learning in Adaptive Filters: The Nearest Instance
Centroid-Estimation Kernel Least-Mean-Square Algorithm.” IEEE Transactions on Signal Processing
65(24):6520–35.
Li, Lei. 2011. “Fast Algorithms for Mining Co-Evolving Time Series.” (September):217.
Lines, Jason, Sarah Taylor, and Anthony Bagnall. 2018. “Time Series Classification with HIVE-COTE:
The Hierarchical Vote Collective of Transformation-Based Ensembles.” ACM Transactions on Know
ledge Discovery from Data 12(5).
Lucas, Benjamin, Ahmed Shifaz, Charlotte Pelletier, Lachlan O’Neill, Nayyar Zaidi, Bart Goethals,
François Petitjean, and Geoffrey I. Webb. 2019. “Proximity Forest: An Effective and Scalable
Distance-Based Classifier for Time Series.” Data Mining and Knowledge Discovery 33(3):607–35.
Marts, J. A. 2015. “Nitrogen Injection in Progressively Sealed Longwall Gobs and the FORMATION
OF A COMPLETE AND DYNAMIC SEAL.”
Marts, J. A., R. C. Gilmore, J. F. Brune, G. E. Bogin, J. W. Grubb, and S. Saki. 2014. “Dynamic Gob
Response and Reservoir Properties for Active Longwall Coal Mines.” Mining Engineering 66(12):41–48.
Nyanteh, Yaw, Chris Edrington, Sanjeev Srivastava, and David Cartes. 2013. “Application of Artificial
Intelligence to Real-Time Fault Detection in Permanent-Magnet Synchronous Machines.” IEEE
Transactions on Industry Applications 49(3):1205–14.
Owusu, Obed, Kenneth Joseph Bansah, and Albert Kobina Mensah. 2019. “‘Small in Size, but Big in
Impact’: Socio-Environmental Reforms for Sustainable Artisanal and Small-Scale Mining.” Journal of
Sustainable Mining 18(1):38–44.
Peng, Syd S. 2019. Longwall Mining. CRC Press.
Qiao, Ming, Ting Ren, Jon Roberts, Xiaohan Yang, Zhongbei Li, and Jianming Wu. 2022. “New Insight
into Proactive Goaf Inertisation for Spontaneous Combustion Management and Control.” Process
Safety and Environmental Protection 161:739–57.
Rodríguez-Rangel, Héctor, Luis Alberto Morales-Rosales, Rafael Imperial-Rojo, Mario
Alberto Roman-Garay, Gloria Ekaterine Peralta-Peñuñuri, and Mariana Lobato-Báez. 2022. “Ana
lysis of Statistical and Artificial Intelligence Algorithms for Real-Time Speed Estimation Based on
Vehicle Detection with YOLO.” Applied Sciences (Switzerland) 12(6):1–20.
382
Schäfer, Patrick and Ulf Leser. 2017. “Fast and Accurate Time Series Classification with WEASEL.” Pp.
637–646 in Proceedings of the 2017 ACM on Conference on Information and Knowledge Management,
CIKM ’17. New York, NY, USA: Association for Computing Machinery.
Schatzel, S. J., R. B. Krog, and H. Dougherty. 2011. “A Field Study of Us Longwall Coal Mine Ventila
tion and Bleeder Performance.” SME Annual Meeting and Exhibit and CMA 113th National Western
Mining Conference 2011 65–70.
Schatzel, Steven J., C. Özgen Karacan, Heather Dougherty, and Gerrit V. R. Goodman. 2012. “An Ana
lysis of Reservoir Conditions and Responses in Longwall Panel Overburden during Mining and Its
Effect on Gob Gas Well Performance.” Engineering Geology 127:65–74.
Shifaz, Ahmed, Charlotte Pelletier, François Petitjean, and Geoffrey I. Webb. 2020. TS-CHIEF:
A Scalable and Accurate Forest Algorithm for Time Series Classification. Vol. 34. Springer US.
Sidorenko, Andrey Alexandrovich, Pavel Nikolaevich Dmitriev, and Yury Georgiyevich Sirenko. 2021.
“Predicting Methane Emissions from Multiple Gas-Bearing Coal Seams to Longwall Goafs at Russian
Mines.” ARPN Journal of Engineering and Applied Sciences 16(8):851–57.
Wahyono, Agus Harjoko, Andi Dharmawan, Faisal Dharma Adhinata, Gamma Kosala, and Kang
Hyun Jo. 2022. “Real-Time Forest Fire Detection Framework Based on Artificial Intelligence Using
Color Probability Model and Motion Feature Analysis.” Fire 5(1).
383
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Lithium-ion (Li-ion) batteries are finding more use as power sources in the
mining industry. However, they are known to pose significant fire and explosion hazards.
When a Li-ion battery is exposed to excessive operating conditions, its internal temperature
may exceed a normal operating range, allowing the active component materials to decompose
or react with each other, eventually leading to thermal runaway. A Li-ion battery contains
certain oxidizing agents making suppression of a battery fire very challenging. A series of Li-
ion battery fire suppression tests were conducted by researchers at the National Institute for
Occupational Safety and Health (NIOSH) to evaluate the effectiveness of different fire sup
pression test systems including dry chemical, water spray/mist, and Class D extinguisher
powder. The batteries tested are commercial nickel manganese cobalt (NMC) and lithium iron
phosphate (LFP) battery packs. The results indicated that dry chemical and Class D powder
could extinguish the fire temporarily, but a reignition occurred. Water mist was able to extin
guish the battery fire completely with continuous cooling of the battery to prevent the reigni
tion. The suppression results for both NMC and LFP chemistries were also compared. These
test results can be used to develop appropriate firefighting strategies for safe and effective sup
pression of battery fires in a mine.
1 INTRODUCTION
Li-ion batteries are known to pose significant fire and explosion hazards as the use of these
batteries has become more widespread, ranging from consumer electronics, battery electric
vehicles, to energy storage systems. The mining industry has also seen increased utilization of
Li-ion batteries for powering equipment in both surface and underground mining operations.
The safety issues related to Li-ion batteries pose a great challenge to their wider application of
Li-ion batteries in large mining equipment which may consist of thousands of battery cells.
Several battery fire incidents have occurred in underground mines in the early stages of bat
tery electric vehicle deployment (Gillet, 2021). A battery fire is usually caused by thermal run
away that occurs when heat generated from exothermic reactions inside a battery outpaces
heat dissipated from the battery. If unmitigated, thermal runaway will proceed to cell rupture
and the venting of hot, toxic, and highly flammable gases, leading to a possible fire or explo
sion. These potential fire hazards need to be assessed and techniques developed to effectively
suppress Li-ion battery fires.
Many laboratory-scale experiments have been conducted on the battery fire suppression
techniques (Xu et al., 2021; Zhang et al., 2021; Cui et al., 2022; Said et al., 2022). Li et al.
(2015) studied the fire-extinguishing efficiency of ABC powder, carbon dioxide, aqueous film-
form foam (AFFF), and water mist on the 18650-type LiCoO2 Li-ion battery pack. The ABC
powder, carbon dioxide, and 3% aqueous film-form foam were able to extinguish the open
flames of the Li-ion battery pack, but the reignition could not be avoided. The water mist
DOI: 10.1201/9781003429241-40
384
system could not suppress the battery fire in the experiments because of low water flow rate.
The laboratory experimental results of Xu et al. (2020) indicate that water mist/spray has
a better suppression effect than CO2 and HFC-227ea. The Federal Aviation Agency (FAA)
conducted experiments to screen fire-extinguishing agents for battery fire involving consumer
electronic products powered by Li-ion batteries (Maloney, 2014). The experimental results
demonstrated that water-based extinguishing agents were the most effective, while nonaqu
eous extinguishing agents were the least effective. Russo et al. (2018) conducted experiments
to evaluate the effectiveness of different fire extinguishing agents for single Li-ion cells and
a battery pack. Out of the five fire-extinguishing agents including water, foam, dry powder,
carbon dioxide, and water mist, water and foam were the most effective by rapidly reducing
the temperature of the cell and extinguishing the fire.
There are few studies on the suppression of large Li-ion battery pack fires. Egelhaaf et al.
(2014) carried out firefighting tests of Li-ion traction batteries, and the test results showed
that the amount of water required to extinguish such a fire is a lot larger amount than that
used for firefighting of conventionally driven vehicles. In National Transportation Safety
Board (NTSB) investigations of battery fires in electric vehicles, the total amount of water
used to suppress high-voltage Li-ion battery fires ranged from 300 to 20,000 gallons (NTSB,
2020). These studies indicated that water can be an effective suppression agent for fighting
battery fires if a sufficient amount of water is available. For battery electric vehicles used in
underground mines, water may not be readily available, or the amount of water may be
limited. Therefore, research is needed to investigate the characteristics of different Li-ion bat
tery fires in mines and effective fire suppression techniques. In this study, fire suppression
experiments were conducted for both NMC and LFP battery packs using water mist, dry
chemical, and Class D extinguishing powder, and the suppression results for these two battery
chemistries were compared. Experiments were also conducted to study the effect of battery
size on the suppression of NMC battery pack fires. The experimental results from this study
may help develop appropriate firefighting strategies for safe and effective suppression of Li-
ion battery fires in underground mines.
2 EXPERIMENTAL
Li-ion battery fire suppression experiments were conducted in a container with dimensions of
12-m long, 2.4-m wide, and 2.85-m high (40 ft × 8 ft × 9.5 ft) and using a data acquisition and
test observation trailer located nearby the container. A fan is installed at one end of the con
tainer and the opposing end is open as shown in Figure 1. The ventilation velocity was set at
0.2 m/s (40 fpm) for all tests. The container is equipped with three fire suppression systems:
dry chemical, water spray/mist, and Class D powder. The fire suppressant discharge nozzle
direction is located above the battery. The suppression system is manually activated from out
side the container. Thermocouple trees and gas sampling tubes were installed at the exit of the
container to measure exit gas temperatures and collect gas samples. Two cameras were
installed inside the container to monitor testing.
The batteries used for the suppression tests were a 12V 30Ah NMC battery pack consisting
of 36 cylindrical cells and a 12V 35Ah LFP battery pack consisting of 24 cylindrical cells.
Before each test, the battery pack was charged to the 100% state. During the tests, the battery
pack was placed onto two heater strips for heating up with each of two 750-Watt heaters. In
addition, 12 to 18 K-type thermocouples were attached to the surfaces of the battery cells of
the battery pack to measure the battery temperature as shown in Figure 2. During the tests,
when the battery reached thermal runaway, the heaters were turned off, and then, after
a stable flame was established, the fire suppression system was activated. The suppression
nozzle was placed 1.5 m above the battery pack. For the dry chemical and Class D powder
suppression systems, the release time was between 30 and 50 seconds. For the water mist
system, the flow rate was 3.6 gallon per minute and water was on for 2 to 5 minutes. Battery
surface temperatures were measured by the data acquisition system. Air exiting the container
was sampled using an infrared gas analyzer to determine concentrations of carbon monoxide
385
Figure 1. Battery fire suppression test setup.
Figure 2. Battery packs and attached thermocouples. NMC (left), LFP (right).
(CO) and carbon dioxide (CO2). Cameras were used to observe and record the fire and sup
pression behaviors. Observations were made on a monitor in the trailer to determine the times
for the first appearance of smoke and flame and the final extinguishment of the fire if there
was one.
386
Compared to the NMC battery pack, significantly heavier smoke was generated before the
first flame. The first cell explosion also occurred about 1.5 minutes after the first flame. How
ever, very few cells exploded, and the battery pack was not disintegrated. The total burning
time was over 19 minutes. The battery did not burn as violently as the NMC battery. Instead,
the LFP battery only burned violently for 1-2 minutes, and then maintained a local small
flame for 4-5 minutes, followed by another violent but short burn Figure 3 (a) shows the vio
lent burn with cell explosion for the NMC battery pack, and Figure 3 (b) shows the flaming of
the LFP battery.
The apparently different combustion behaviors of NMC and LFP batteries can be further
examined in Figure 4. Figure 4 (a) shows the exit gas temperature increase above the ambient
temperature, while Figure 4 (b) shows the measured CO2 concentrations at the exit for the
two batteries. The NMC battery had a slightly higher maximum gas temperature increase, but
a lower maximum CO2 concentration than the LFP battery. As observed, the NMC battery
burned violently lasting for only 4.5 minutes. This can be seen clearly in the gas temperature
increase and CO2 concentration data. For the LFP battery, the gas temperature increase and
CO2 concentration data show clearly that there were a few peaks after the first peak indicating
the existence of small flames and burning for more than 19 minutes. This is consistent with the
observation on the camera monitor.
Figure 3. 12V battery pack free burn (a) NMC (left) and (b) LFP (right).
Figure 4. (a) Comparison of exit gas temperature increase (left) and (b) CO2 concentrations (right).
3.2 NMC and LFP battery pack fire suppression using dry chemical and Class D powder
Dry chemical is a commonly used fire suppressant for mine equipment fires. In this study,
a dry chemical suppression system was used to suppress fires from both 12V NMC and LFP
battery packs under the same experimental conditions. For the NMC battery, the dry chem
ical suppression system was activated after the first cell explosion occurred, and the
387
suppressant discharge lasted for 10 seconds. The flame was extinguished by the suppressant.
Figure 5 (a) shows the burning of the NMC battery before the discharge of the suppressant,
and Figure 5 (b) shows the NMC battery pack after the flame was extinguished. The battery
pack was dislocated a little and broke into two sections by the cell explosion. After 2.5 min
utes, the reignition occurred and the NMC battery exploded again as shown in Figure 5 (c).
Eventually, the burning slowed down as shown in Figure 5 (d). The reignition lasted for about
8 minutes. Class D extinguishing powder was also tested for suppression of the battery pack
fires. Similar to the dry chemical, the Class D powder was able to quench the flame quickly
for the NMC battery pack; however, the flame came back in a few seconds after the discharge
of the suppressant. The NMC battery continued to burn for about 5 minutes to self-
extinguishment.
Figure 6 shows the LFP battery fire and suppression process with the dry chemical suppres
sion system. Figure 6 (a) shows the flame before the discharge of the dry chemical. Figure 6
(b) shows the battery after the flame was extinguished by the suppressant with everything
covered by the white chemical. After roughly 2 minutes, reignition occurred as shown in
Figure 6 (c) because the dry chemical was not able to cool the LFP battery to stop the exother
mic internal reactions. The LFP battery continued to burn all the way to self-extinguishment
lasting for more than 10 minutes as shown in Figure 6 (d). During the reignition period, there
was a cell explosion, and the flame went out for about half a minute and came back again.
For the LFP battery pack, the Class D powder was not able to extinguish the flame, and the
battery continued to burn for 10 minutes.
Figure 5. NMC battery fire suppression using dry chemical: (a) flame before suppression; (b) after sup
pressant discharge; (c) explosion following reignition; (d) flame near burnout.
3.3 NMC and LFP battery pack fire suppression using water mist
A water mist system was applied to suppress the battery fires from both the NMC and LFP
battery packs. For the NMC battery fire, water mist was turned on one minute after the first
flame, and the fire was extinguished in 65 seconds. There was one cell explosion during the
suppression. Figure 7 (a) shows the burning of the NMC battery before water suppression
388
Figure 6. LFP battery fire suppression using dry chemical: (a) flame before suppression; (b) after sup
pressant discharge; (c) flame after reignition; (d) flame near burnout.
was turned on, while Figure 7 (b) shows the NMC battery after the fire was extinguished.
Water was on for 2 minutes and there was no reignition. For the LFP battery fire, the water
mist extinguished the flame in 43 seconds, but the flame occurred again in half a minute. As
water was still on, the fire was extinguished in 14 seconds. The total extinguishing time was 86
seconds, and there was no reignition. Figure 8 (a) shows the burning of the LFP battery
before the release of water mist, and Figure 8 (b) shows the LFP battery pack after the fire
was extinguished. The results indicate that water mist was effective in extinguishing both the
NMC and LFP battery pack fires as water could cool the battery continually to prevent the
reignition of the battery.
Figure 7. NMC battery fire suppression with water mist; (a) flame before suppression (left) and (b)
flame out after water mist (right).
389
Figure 8. LFP battery fire suppression with water mist; (a) flame before suppression (left) and (b) flame
out after water mist (right).
Figure 9. Comparison of exit gas temperature increases between different NMC battery pack sizes
using (a) dry chemical (left) and (b) Class D powder (right).
390
Figure 10. Scattered NMC cells after flame was out with (a) dry chemical system (left) and (b) Class
D powder system (right).
Figure 9 (a) and Figure (b) show exit gas temperature increases for both the 12V and 24V
NMC battery packs during the fire suppression process using the dry chemical and Class
D powder system, respectively. It can be seen that the dry chemical was able to extinguish the
flame immediately after the discharge of the suppressant. But reignition occurred after 15 min
utes as the dry chemical was not able to cool the battery. Dry chemical smothered the battery
completely and stopped the radiative heat transfer from the flame to the battery as the flame
was extinguished. The chemical reactions inside the battery were still ongoing, but there was
no external heat transfer any longer. This may be the major reason that it took 15 minutes for
the reignition to occur, and the battery burned very violently with many cell explosions. Using
the dry chemical suppressant, 24V NMC battery pack was completely disintegrated, and cells
were scattered as shown in Figure 10 (a). The maximum exit gas temperature increase was
about 38° C for the 24V NMC battery pack fire while using the dry chemical suppressant,
Figure 9 (a). On the other hand, the Class D powder was not able to extinguish the flame
completely, but just reduced it to a small flame. The small flame continually radiated heat to
the adjacent battery cells. The chemical reactions inside those cells were accelerated by the
24V NMC battery pack flame heat, and it took 70 seconds for those cells to explode produ
cing a peak exit gas temperature increase of 20° C, Figure 9 (b). About 5 minutes later, there
was another round of cell explosions with another peak exit gas temperature increase of 21°
C. As shown in Figure 10, more of the 24V NMC battery pack cells exploded with the dry
chemical system than with the Class D powder system.
4 CONCLUSIONS
Experiments were conducted to study the fire behaviors of 12V NMC and LFP Li-ion battery
packs. Under the same conditions, the NMC battery pack burned more violently with more
cell explosions than the LFP pack, while the LFP pack burned for a longer time and generated
more CO2. Fire suppression experiments were then conducted for both battery chemistries
using dry chemical, Class D powder, and water mist systems. The experimental results demon
strate that water was more effective in extinguishing the battery pack fires than dry chemical
and Class D powder because water was able to cool the battery continually. With dry chemical
and Class D powder, the flame was extinguished temporarily, but reignition occurred after 2
to 2.5 minutes with the dry chemical, and a few seconds with the Class D powder.
The effect of battery size on the suppression of NMC battery pack fires was investigated
through suppression tests using a larger 24V battery pack. The experimental results show that
even though water was still able to extinguish the larger 24V battery pack fire, the extinguishing
time increased significantly as compared to the smaller 12V battery pack, from 65 to 172 seconds
indicating a large amount of water is needed to extinguish a large format Li-ion battery fire. With
the larger 24V battery pack and the dry chemical, the flame was also extinguished temporarily,
but it took a much longer time for the reignition to occur as compared to the smaller 12V battery
pack, from 2.5 to 15 minutes, indicating the larger the battery size, the longer time for the
391
reignition to occur. With the Class D powder system, the flame was not completely extinguished,
and the battery continued to burn, indicating that the Class D powder is least effective for sup
pressing the Li-ion battery fires. The experimental results also demonstrate that the longer time
before the reignition results in a more violent explosion, as more cells become involved.
5 DISCLAIMERS
The findings and conclusions in this report are those of the authors and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health,
Centers for Disease Control and Prevention. Mention of any company or product does not
constitute endorsement by NIOSH.
REFERENCES
Cui, Y., Liu, J., Han, X., Sun, S., Cong, B. (2022). Full-scale experimental study on suppressing
lithium-ion battery pack fires from electric vehicles, Fire Safety Journal, 129, 103562.
Egelhaaf, M., Wolpert, D., Lange, T. Fire Fighting of battery electric vehicle fires. Proceedings of the
Third International Conference on Fire in Vehicles, October 1-2, 2014, Berlin, Germany. pp. 221–230.
Gillett, S., 2021. Battery Electric Vehicle Emergency Response Incident Review and Best Practices.
Workplace Safety North Virtual Symposium: Battery Electric Vehicle Safety in Mines, January 20,
2021.
Li, Y., Yu, D., Zhang, S., Hu, Q., Liu, X., Wang, J. (2015) On the fire extinguishing tests of typical lith
ium ion battery. Journal of Safety Environment, 15:120–125.
Maloney, T. (2014). Extinguishment of lithium-ion and lithium-metal battery fires, US Department of
Transportation; Federal Aviation Administration: Washington, DC, pp.46–51.
NTSB, 2020. Safety risks to emergency responders from lithium-ion battery fires in electric vehicles,
Safety Report NTSB/SR–20/01.
Russo, P., Di Bari, C., Mazzaro, M., De Rosa, A., Morrielo, I. (2018). Effective fire extinguishing sys
tems for lithium-ion battery. Chemical Engineering Transactions, 67, 727–732.
Said, A.O., Garber, A., Peng, Y., Stoliarov, S.I. (2022) Experimental investigation of suppression of
18650 lithium ion cell array fires with water mist, Fire Technology, 58, 523–551.
Xu, J., Guo, P., Duan, Q., Yu, X., Zhang, L., Liu, Y., Wang, Q. (2020). Experimental study of the effect
iveness of three kinds of extinguishing agents on suppressing lithium-ion battery fires. Applied Ther
mal Engineering, 171, 115076.
Xu, J., Duan, Q., Zhang, L., Liu, Y., Zhao, C., Wang, Q. (2021). Experimental study of the cooling
effect of water mist on 18650 lithium-ion battery at different initial temperatures, Process Safety and
Environmental Protection, 157, 156–166.
Zhang, L., Duan, Q., Liu, Y., Xu, J., Sun, J., Xiao, H., Wang, Q. (2021). Experimental investigation of
water spray on suppressing lithium-ion battery fires, Fire Safety Journal, 120, 103117.
392
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
Robert Ilango Pushparaj, Guang Xu*, Amir Iqbal & O.B. Salami
Department of Mining and Explosives Engineering, Missouri University of Science and Technology, Rolla,
MO, USA
ABSTRACT: Any fire could quickly destroy human lives including casualties. Fire risk
exposure by diesel-powered vehicles in confined area is critical. Also, turning to battery
powered vehicles, fire risk is worse. Li-ion battery fire produces intense heat, smoke, and com
plex toxic gasses. Among them, toxic gases constitute a significant threat to human health.
Therefore, it is necessary to study the detailed identification of gas emissions from the battery
fire. Previous studies reported the emission of toxic gases after firing the different types of bat
teries, still unknown gases are not studied or quantified. Remarkably, real-time gas analysis is
not explored very well which motivates us to conduct this work. In this primary study, we
conducted a large-scale diesel fire test at the predesigned fire station. The fire was recorded
using the IR radiation from the thermal camera. The total fire lasted about 20 min with
a maximum fire temperature of 601ºC. Thermocouple 1 (T1) measured the maximum
temperature of 39ºC which is comparatively lower than Thermocouple 2 (T2) due to the con
siderable distance between them about 2 ft. This primary fire characteristics investigation will
provide a new way to conduct large scale battery fires and a detailed understanding of battery
fire and associated health risks.
1 INTRODUCTION
Rechargeable Li-ion batteries (LIB) play a vital role in the energy storage system due to their
high gravimetric and volumetric energy density (Scrosati and Garche, 2010). Besides, electrical
vehicles are emerging in the mining industry to substitute the current extensively used mobile
equipment operated primarily by diesel internal combustion engines owing to its low energy
consumption and very less heat emission (Stampatori et al., 2020). Furthermore, the mainten
ance and ventilation infrastructure cost are lower than diesel powered vehicles(Skidmore,
2021). As an example, Newmont is a famous gold mine industry have agreement with caterpil
lar to achieve zero emission by 2027 through investing primarily $100 billion on battery elec
trification of mining vehicles (Release, 2021).
Nevertheless, using Li-ion battery is not safe due the thermal insecurity that leads to Li-ion
battery fire and explosion (Sun et al., 2020). Generally, the battery fire releases hazardous
gases and heat that consequently impact the potential health risks. For example, ‘B747ʹ cargo
flights crashed and reported two pilots death in 2011 due to the LIB fire (Brett, n.d.). Also,
the Occupational Safety and Health Administration report says twenty-five thousands of Li-
ion battery fire incidents happened over 5 years (Yuan et al., 2020). Therefore, understanding
the risks of BEV fire enable us to prepare for preventing BEV incidents that enables safe mas
sive use of LIB.
State of charge (SOC) may impact the battery fire characteristics such as heat, flame, and
gas emissions. Recent studies focused thermal and fire behaviors of 38 Ah and 243 Ah LIBs
with different SOC and analyzed the relationship of thermal runaway and gas generations.
DOI: 10.1201/9781003429241-41
393
For example, M. Chen et al, conducted experimental Li-ion fire study with primary batteries
first and later with modern battery like 18650 battery type and reported that the combustion
efficiency and yield of CO2 varied from single to bundle primary battery fire. Further, solid
electrolyte interface layer breakdown and the separator drop-off are the main key reasons for
the Li-ion battery fire (Chen et al., 2016)(Chen et al., 2015). Afterwards, V. Somandepalli
et al, examined quantification of combustion hazards, but they only suggested the gas com
position highest over pressure and flammability that helps to design vehicles (Somandepalli
et al., 2014). Also, P. Huang et al, studied combustion behavior using large scale lithium titan
ium battery, revealed risk increases according to SOC (Huang et al., 2015). Very recently,
F. Larson et al, investigated the fluoride gas emission with different types of batteries and this
investigation is limited to small or very few batteries gas emission observations (Larsson et al.,
2017). Though several reports identified the toxic gases emissions and quantification, still
other hazardous gasses not yet clearly studied or quantified. Moreover, previous studies are
tested with very few batteries (less than 100 g) and estimated the thermal runway and predic
tion of fire suppression that is not enough for the understanding of Li-ion battery fire safety,
particularly underground mine battery vehicles.
In the past, diesel pool fire has been studied and their impact has been examined. For
example, L.Yuan et al (Yuan et al., 2018) studied diesel fire and reported the sensor response
time to early detect the fire risk. Later, R. Hansen (Hansen, 2019) researched a number of
design fire situations which were created from risk evaluations. Also, a deterministic approach
was executed and where the findings from previous full-scale fire experiments in underground
mines. Very recently, G.V Kuznetsov et al (Kuznetsov et al., 2022) classified fire and analyzed
fire characteristics, gas analysis and finally recommended fire prevention systems.
The objective of this paper is to determine the diesel pool fire characteristics before setting
up a large battery fire station to identify and quantify toxic gases emitted in LIB fires and
evaluate their risk in underground mines. This diesel pool fire characteristic includes fire sta
tion setup, positioning the fire pool, measuring the surface temperature, and flame character
istics. This preliminary work will help to improve the battery fire station and control external
factors such as air direction and velocity before battery fire risk analysis.
2 EXPERIMENTAL SET UP
Figure 1(a) describes the simple schematic representation of a full-scale diesel fire test station.
A few significant devices that have been used to test such as a thermal camera, thermocouple,
and exhausting fan. In brief, a pan container (12-inch diameter) and a 0.5-gal diesel pool were
used in this experiment. The thermal camera was placed 10 ft away from the pool fire. We
used two thermocouples (T1, T2), and they measured the instantaneous temperature at two
Figure 1. (a) Schematic representation of diesel fire test (b)Photograph of actual fire station.
394
different locations with a distance of 2 ft. The approximate length of thermocouple from the
center of the pool fire is 5 ft. The fire was ignited using an external source, and the fire was
slowly started and stabilized over 5 min, and total fire lasted about 20 min. At the end of the
duct the fan was connected, and it was able to pull some amount of smoke out of the gas sta
tion indicating that this fan could pull enough smoke. This smoke will be transferred the gas
cell for the further toxic and toxic gas evaluation using the FITR shown in the Figure 1(b).
The same procedure will be used to conduct the large-scale battery fire test and gas evaluation
which will be discussed in our future research.
Figure 2 illustrates the combustion behavior of the diesel fire corresponding to the various time
stamps. Noted that, the diesel fire was not stable due the uncontrolled air direction and velocity.
According to the burning characterization, the fire was classified in to three types. (i) ignition,
(ii) stable combustion and (iii) abatement (Liu et al., 2021). The fire was visible after 2 minutes
of ignition. The fire intensity reached to its maximum value at the reported time. Therefore,
t = 2.0 min was chosen as the representative time for stage I of combustion. The images depict
ing the 3.0 min somewhat higher temperature. It can be seen that stage I has low flame intensity
with a temperature of 199 ºC after 2 min which further burned and reached to stage II represent
ing stable combustion. The stage II was lasted about 10 min with a high intensity of fire and
read the maximum temperature of 601ºC. Finally, the fire reached to the stage III which called
as abatement where the fire last 2 min and temperature was recorded about 150ºC indicating
the burning was slow down. In comparison with each stage, the stage II has major vertical fire
than others. However, the fire was going all the way to all the direction due to the air flow direc
tion. This experiment was set up in a place where too many factors were associated.
From this research finding, we can understand the fire temperature and direction. This
research will be extended at the place where you can control the external factors to determine
the full fire characteristics. The thermal camera was used to record the video and then classify
the burning behaviors of the fire. The thermal camera uses IR radiation to measure the fire
395
temperature. IR camera visualizes the surface’s temperature gradient by converting changes in
electrical resistance made by infrared radiation into a thermogram (Spieß et al., 2021). In
general, thermal camera will work based on the Stefan-Boltzmann law. This law states that
the total energy radiated per unit surface area of a black body across all the wavelengths per
unit time is s directly proportional to the fourth power of its absolute temperature. Previously,
similar studies were conducted to analyze the 2.9Ah powered Li-ion battery burning behaviors
(Ping et al., 2015). Therefore, this case study will help to further research on large scale battery
fire classification and gas quantification.
Surface temperature is critical to trigger the battery thermal runaway. Hence, it should be
measured using an appropriate tool. Thermo couple is powerful tool that has been used to
detect the battery surface temperature during the thermal runaway process (Chen et al., 2019).
Primarily, we calculated the diesel fire the axial temperature using thermocouple. Figure 3 rep
resents the temperature vs time profile of the diesel fire. T1 and T2 representing the two differ
ent locations. Position T1 shows clearly that the maximum temperature reached to 39ºC while
T2 shows only 23ºC. The temperature difference is due to the location and no uniform air
inlet and external turbulent. Due to external factors such as air direction, air velocity, climate
conditions, the temperature may not be reproduced at open conditions. In our future study,
we will use this temperature logger to determine the maximum surface temperature of the bat
tery instead of diesel fire. Also, increasing the thermocouples at different locations around the
fire may ensure temperature detection accuracy. In the past, many researchers have been used
1-16 thermocouple to determine the axial temperature from various fire sets up (Xu and Lei,
2022). Hence, using many thermocouples, we can predict the average temperature ranges
around the battery fire. We extend the use of thermocouple on the battery surface to know the
surface temperature to trigger the thermal runaway. This case study will provide the risk
evaluation during thermal management design.
4 FUTURE WORKS
To execute the battery fire research, we will use the diesel fire test station with high-accuracy
analytical instruments, as shown in Figure 4. The dimension of the fire station is 20 x 10 x 10
(L x H x W in ft). The fire station is covered with a metal roof to avoid fire, or any explosion.
Primarily, the fire station is closed on three sides. A gap of 1 ft from the ground was left to
allow air venting into the chamber. The smoke duct dimension is 8.0 cm in diameter and
1.0 m in length in vertically. It will be connected horizontally with a 3.0 cm duct, and the total
396
length will be 5 ft from the center. The duct end will be connected to the exhaust fan (AC
Infinity, model AI-CLA4) to pull the fumes.
To collect and transfer the gas, we will use a vacuum pump (Dia-Vac, model R222-At-
AA1-C), which will be connected to the outlet of the FTIR gas cell (Antaris IGS gas
analyzer) to create the negative pressure. The gas cell size is 2 M in length which will
maintain temperature and pressure to resist gas from condensation. Finally, K-type
(model TP-01A) thermocouples will be used to measure the temperature and SD data
logger will be used to record the temperature data. We will use thermal camera (FLIR
E86, wavelength 650 mn, temperature range º0C-650ºC). Commercial large-sized Li-ion
battery pack or module (Lithium-Ion phosphate, 500-600 kWh) will be used first to iden
tify the battery fire risk. We will also use different gas collection methods, such as gas
washing bottles (Iplusmile,60 ml glass) and gas cell (thermo scientific) collection to check
the offline gas measurement. Ion chromatography (thermo scientific) will be used to ana
lyze the offline gas measurement and then the data will be compared with the real-time
measurement.
5 CONCLUSIONS
In this study, a large-scale diesel fire (0.5gal capacity) was tested and measured the maximum
surrounding temperatures at different locations. Also, the maximum temperature achieved by
full-scale diesel fire temperature using IR radiation. The maximum fire temperature was
601ºC during the burning. This study suggests that air direction and velocity to be controlled
to reproduce the data. Overall, this fire can be distinguished with respect to time which opens
a way to understand more about fire characteristics. The primary analysis of the combustion
fire and changes in temperature reveals that the thermocouple and thermal camera will be
handy in determining the fire-related properties. Also, this study, provides the basic informa
tion of fire behaviors air direction, and gas collection. This research will be conducted using
a real battery used in mining vehicles to identify toxic gases, suppression techniques and ther
mal management systems.
397
REFERENCES
Brett, D., n.d. airportwatch:Another cargo aircraft carrying batteries crashes – off South Korea.
Chen, M., Dongxu, O., Liu, J., Wang, J., 2019. Investigation on thermal and fire propagation behaviors
of multiple lithium-ion batteries within the package. Appl. Therm. Eng. 157, 113750.
Chen, M., He, Y., De Zhou, C., Richard, Y., Wang, J., 2016. Experimental Study on the Combustion
Characteristics of Primary Lithium Batteries Fire. Fire Technol. 52, 365–385. https://doi.org/10.1007/
s10694-014-0450-1
Chen, M., Zhou, D., Chen, X., Zhang, W., Liu, J., Yuen, R., Wang, J., 2015. Investigation on the ther
mal hazards of 18650 lithium ion batteries by fire calorimeter. J. Therm. Anal. Calorim. 122, 755–763.
https://doi.org/10.1007/s10973-015-4751-5
Hansen, R., 2019. Design of fire scenarios for Australian underground hard rock mines–Applying data
from full-scale fire experiments. J. Sustain. Min. 18, 163–173.
Huang, P., Wang, Q., Li, K., Ping, P., Sun, J., 2015. The combustion behavior of large scale lithium
titanate battery. Sci. Rep. 5, 1–12. https://doi.org/10.1038/srep07788
Kuznetsov, G. V, Volkov, R.S., Sviridenko, A.S., Strizhak, P.A., 2022. Fast detection of compartment
fires under different heating conditions of materials. Process Saf. Environ. Prot. 168, 257–274.
Larsson, F., Andersson, P., Blomqvist, P., Mellander, B.E., 2017. Toxic fluoride gas emissions from
lithium-ion battery fires. Sci. Rep. 7, 1–13. https://doi.org/10.1038/s41598-017-09784-z
Liu, P., Li, Y., Mao, B., Chen, M., Huang, Z., Wang, Q., 2021. Experimental study on thermal runaway
and fire behaviors of large format lithium iron phosphate battery. Appl. Therm. Eng. 192, 116949.
https://doi.org/10.1016/j.applthermaleng.2021.116949
Ping, P., Wang, Q., Huang, P., Li, K., Sun, J., Kong, D., Chen, C., 2015. Study of the fire behavior of
high-energy lithium-ion batteries with full-scale burning test. J. Power Sources 285, 80–89.
Release, N., 2021. Newmont and Caterpillar Announce Revolutionary Strategic Alliance to Achieve
Zero Emissions Mining1–4.
Scrosati, B., Garche, J., 2010. Lithium batteries: Status, prospects and future. J. Power Sources 195,
2419–2430. https://doi.org/10.1016/j.jpowsour.2009.11.048
Skidmore, Z., 2021. Battery electric vehicles – facilitating low-carbon mines.
Somandepalli, V., Marr, K., Horn, Q., 2014. Quantification of combustion hazards of thermal runaway
failures in lithium-ion batteries. SAE Int. J. Altern. Powertrains 3, 98–104. https://doi.org/10.4271/
2014-01-1857
Spieß, B., Metzsch-Zilligen, E., Pfaendner, R., 2021. Mechanistic evaluation of flame retardants during
UL94 standard testing via IR-camera. Polym. Test. 103, 107320.
Stampatori, D., Raimondi, P.P., Noussan, M., 2020. Li-ion batteries: A review of a key technology for
transport decarbonization. Energies 13. https://doi.org/10.3390/en13102638
Sun, P., Bisschop, R., Niu, H., Huang, X., 2020. A Review of Battery Fires in Electric Vehicles, Fire
Technology. Springer US. https://doi.org/10.1007/s10694-019-00944-3
Xu, T., Lei, P., 2022. Experimental study on flame height and heat release rate estimation of
diesel-wetted wood powder fire. Case Stud. Therm. Eng. 33, 101906.
Yuan, L., Dubaniewicz, T., Zlochower, I., Thomas, R., Rayyan, N., 2020. Experimental study on ther
mal runaway and vented gases of lithium-ion cells. Process Saf. Environ. Prot. 144, 186–192. https://
doi.org/10.1016/j.psep.2020.07.028
Yuan, L., Thomas, R.A., Rowland, J.H., Zhou, L., 2018. Early fire detection for underground diesel fuel
storage areas. Process Saf. Environ. Prot. 119, 69–74.
398
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
A.R. Kumar
Department of Energy and Mineral Engineering, The Pennsylvania State University, University Park, PA, USA
Keywords: Full-scale fire, underground mines, temperature attenuation, heat release rate
1 INTRODUCTION
Fires in confined underground spaces are very deadly due to their impending death (Fernán
dez-Alaiz et al., 2020). The complexity of underground mines owing to their layout, ventila
tion limitations, and the presence of hazardous and flammable materials in such environments
make underground fire incidents particularly perilous (De Rosa, 2004, Fernández-Alaiz et al.,
2020). Studies have shown that mine fires produce significant levels of toxic gases and smoke
which can be transported through the ventilation network to various parts of the underground
workings. These noxious gases released during fires further reduce visibility and escape
chances of the underground personnel, and research has shown that most fatalities due to
underground fires can be attributed to the inhalation of poisonous gases which is mainly
carbon monoxide (Bahrami et al., 2021, Zhou, 2009, Zhou et al., 2018).
Contemporary studies have investigated temperature decay and maximum gas temperature
beneath the ceiling for different ceiling extraction strategies (Tang et al., 2018a, Tang et al.,
DOI: 10.1201/9781003429241-42
399
2018b, Tao et al., 2022). In one of the studies, Tang et al. (Tang et al., 2018a), examined the
combined effect of longitudinal wind and single smoke ceiling extraction on the temperature dis
tribution in a model-scale horizontal channel with a similarity scale of 1:66 based on Froude
correlation. They proposed an empirical model for the set experimental conditions which could
be applied to similar practical situations. The research showed that the longitudinal temperature
decays faster in upstream positions than in downstream positions. In another study, Tang et al.
(Tang et al., 2018b) argued that most of the previously suggested correlations were developed
using a square or circular fire source. They opined that it is important to extend the geometry of
the fire source to account for other practical tunnel fire scenarios such as rectangular fire
source. Hence, a new empirical correlation for temperature evolution with combined smoke ceil
ing extraction induced by a rectangular fire source was established. Tao’s team (Tao et al.,
2022) on the other hand, used a model scale tunnel with a Froude similarity criterion of 1:10 to
explore the temperature distribution in tunnels with the combined effect of a single-side central
ized smoke extraction system. Their study showed that the maximum temperature initially
increases and then decreases when the smoke extraction rate is increased with fires having the
same HRR. A new temperature decay model for the single-side centralized smoke extraction
system was established based on the experimental results.
Most of our knowledge on fire and toxic gases spread in underground mines environment is
based on previous studies from straight tunnel fires or scaled model tests. This makes it impossible
to preserve all the dimensionless terms derived from the scaling theory (Li and Ingason, 2012).
Also, most existing studies do not account for the multiple bifurcations and environmental condi
tions common in underground mines. Studies showed that the thermal properties of the material
used for building the model scale tunnel may significantly influence temperature and heat transfer
(Li and Ingason, 2012). Hence, there is a need to conduct full-scale fire tests for a deeper under
standing of fire behavior and temperature decay in real underground mining situations.
This study investigates the interactive influence of ventilation conditions induced by ceiling
smoke extraction rate and fire size on temperature attenuation, and maximum excess gas tem
perature in an experimental underground mine. The schematic of the experimental rig and
data communication system used in this study is illustrated in Figure 1. The scale was first set
up and tarred to zero. The fire pan was then placed on a 30.0 cm by 30.0 cm frame installed
on the scale to alleviate the exposure of the sensitive scale electronics to high temperatures.
Diesel fuel was then poured into the pan. The monitors and sensors were turned on and syn
chronized to log the data from a remote computer before the fire pool is ignited. Full-scale
fire tests were then carried out in the main drift of the mine to evaluate the effect of varying
ventilation conditions on temperature and toxic gas distribution. Based on the results, an
empirical correlation for temperature decay in the mine drift was developed. A new model for
a single-side non-centralized smoke extraction system in an underground space was proposed.
The obtained data from this study will also be very useful to validate existing mine fire compu
tational fluid dynamics (CFD) models and model scale test results based on Froude similarity.
Figure 1. Schematic of the experimental rig and data communication system used for this study.
400
2 FULL-SCALE FIRE TEST
Figure 2. Plan view of the study location and the exhaust fan.
The diameters of the fire pool containers were 0.3 m, 0.5 m, and 0.7 m as shown in
Figure 4 (a). Three ventilation conditions were studied. They were varied by adjusting the ceiling
smoke extraction rate between 0.0-100.0 %. The fan was turned off to eliminate mechanical venti
lation; this case was denoted as the ‘0.0 %’ ventilation condition. The mine was under natural
ventilation in this condition. The second case is the ‘50.0 %’ ventilation rate which was realized by
401
turning the fan on but blocking half of the fan duct cross-section area using a circular wooden
board as shown in Figure 4 (b). Whereas for the third case, the fan was turned on and the duct
was kept fully open to realize the ‘100.0 %’ ventilation condition (as depicted in Figure 4 (c)).
A constant diesel volume, 0.5 gals, was used for each test. After each test, the fan was run at
100.0 % ventilation condition for 10.0 mins, to remove the smoke and allow the underground
temperature to reach ambient conditions. The fire pool was placed at 13.0 m from the in-by
portal along the main drift (indicated by a fire icon in Figure 2). A ventilation door is located at
about 12.0 m from the fire pool downstream which separates the testing region from the blind
heading of the main drift in the underground and thus, the fire could be considered axisymmetric
within the test region in the underground mine. The smoke temperature at the mine ceiling was
determined using K-type thermocouples with an accuracy of 0.025°C and a diameter of 0.6 mm
connected to a data logger. At the roof of the mine, thermocouples were installed at all seven
locations (S1-S7) as depicted in Figure 5.
3 TEMPERATURE DISTRIBUTION
402
while transverse attenuation measures the temperature decay between the walls of the tunnel.
However, the temperature variation along the tunnel or mine airways is more important as
occupants tend to evacuate along the tunnel instead of moving zigzag between its walls. In
this study, only the longitudinal temperature attenuation and maximum excess smoke tem
perature beneath the ceiling were considered. The theoretical background of temperature
decay beneath the ceiling is presented in section 3.1.
where ΔT0 is the average temperature rise near the ceiling over the fire source, ΔT is the
average instantaneous temperature rise at the specified location from the fire axis, lb represents
the beam depth, H is the separation between the fire surface and the ceiling, and St is the Stan
ton number which was estimated to be 0.03 from the correlation of experimental data
obtained from the study. The temperature attenuation was found to conform with exponential
decay laws and subsequent investigations have corroborated this finding. A later study by Hu
et al. (Hu et al., 2013, Hu et al., 2007) which involved a series of full-scale tests in large tunnels
and vehicular tunnels showed that temperature beneath the ceiling in a longitudinally venti
lated tunnel can be expressed as an exponential decay given by:
403
where ∆Tx is the maximum increase in temperature x m away from a reference point, ∆Tt is the
maximum increase in temperature at the reference point, x is the distance from the coordinate
origin, xr denotes the distance of separation between the fire and the coordinate origin, and β is
the temperature attenuation factor. More recently, Zhao et al. (Zhao et al., 2019) investigated
the ceiling temperature distribution in a 1:20 reduced-scaled tunnel model tunnel and estimated
the attenuation factor to be 0.536 for a tunnel fire with a two-point extraction ventilation
system. Thus, using their attenuation factor, a modified Hu’s equation was obtained:
In another study which involved a total of twelve tests in a scaled-model tunnel (1:23), Inga
son and Li (Ingason and Li, 2010), demonstrated that the distribution of excess gas tempera
ture beneath the ceiling without mechanical ceiling smoke extraction can be expressed as
a sum of two attenuations given by:
Therefore, the generalized temperature decay model beneath the tunnel/drift ceiling can be
modified using the exponential model as follow:
Table 3. Point traverse airflow upstream of the fire for different experimental cases.
S/N P1 P2 P3 P4 P5 P6 P7 P8 P9
Exp 1 0.47 0.33 0.27 0.32 0.12 0.00 0.33 0.16 0.04
Exp 2 0.16 0.16 0.21 0.00 0.00 0.07 0.07 0.11 0.06
Exp 3 0.00 0.00 0.00 0.00 0.00 0.00 0.07 0.00 0.04
Exp 4 0.00 0.00 0.00 0.00 0.00 0.00 0.07 0.00 0.04
Exp 5 0.16 0.16 0.21 0.00 0.00 0.07 0.07 0.11 0.06
Exp 6 0.16 0.16 0.21 0.00 0.00 0.07 0.07 0.11 0.06
Exp 7 0.47 0.33 0.27 0.32 0.12 0.00 0.33 0.16 0.04
Exp 8 0.47 0.33 0.27 0.32 0.12 0.00 0.33 0.16 0.04
Exp 9 0.00 0.00 0.00 0.00 0.00 0.00 0.07 0.00 0.04
where m'' is the burning rate or mass loss rate per unit area per unit time (kgm-2s-1), ∆Hc,eff is
the effective heat of combustion (kJ.kg-1), kβ is the empirical constant (m-1), and D is the
405
diameter of the burning area (m). The equation could be simplified by eliminating the term in
the parentheses and the HRR could be obtained from the product of mass loss rate and heat
of combustion as given by the equation (7) (Yuan et al., 2016, Zhou et al., 2018).
For this study, the HRR was determined from the mass loss rate of the diesel using the sim
plified equation (7). The mass loss rate and the corresponding HRR results in the order of
decreasing maximum temperature are presented in Table 4. Table 4 shows that the fire size
has a significant influence on the burning rate. Longitudinal ventilation has a negligible effect
on the burning rate due to the very low ventilation velocity observed in the mine drift during
the fire tests. For instance, considering test experimental cases 5 and 6, a decrease in fire size
to about half may cause the burning time to increase with a factor of up to 4.5 if the longitu
dinal ventilation is kept constant.
Table 4. Heat release rate values obtained for the experimental cases.
S/N Test Case Pool diameter (m) Ventilation Opening (%) Mass loss rate (g/s) Max HRR (KW)
For the calculation model of temperature attenuation upstream and downstream of fire, an
improved technique (Tao and Zeng, 2022) that incorporates a two-dimensional spreading
phase for temperature distribution was introduced instead of the conventional one-
dimensional temperature decay model found in many works of literature. First, normalized
the smoke temperature Tx was normalized, measured by the thermocouples from a distance
x from the fire to a dimensionless temperature ∆TND,x by a reference value Tr and the ambient
air temperature Ta as follows:
The reference temperatures Tr for the upstream and downstream positions were taken at
station S3 and station S5 (as shown in Figure 9 and Figure 10), 2.0 m upstream and
406
Figure 7. Temperature curves measured at different distances for Test 2 (Centre point experiment).
Figure 8. The maximum rise in smoke temperature at mine roof for the different ventilation conditions.
downstream of the fire source respectively. The dimensionless normalized temperatures are
plotted against the distance of separation of the thermocouples measuring stations to the fire
source as seen in Figure 9 to Figure 11. The experimental data from each test was fitted to an
exponential function using the following equations:
The dimensionless normalized temperatures can be properly fitted using the exponential
function (Zhao et al., 2019). The average value of the attenuation factors was determined and
used for developing the empirical model in this study. Finally, a two-dimensional spreading
phase empirical model was developed that could predict the temperature distribution for an
axisymmetric fire in a mine drift with multiple bifurcations. From the results in Table 5, the
semi-empirical calculation model is summarized as follows:
The exponential fitting of dimensionless smoke temperature distribution along the under
ground drift is presented in Table 5. The average values of a, b, and R2 are calculated. The R2
value for upstream and downstream are well-fitted which indicates a strong correlation of
temperature with the distance of separation from the fire source.
A comparison of the proposed temperature attenuation model with existing models was
done to highlight the difference between the temperature decay models. Ingason and Li
407
Figure 9. Correlation of temperature evolution in the mine drift for 425.0 kW (0.7D) fire under different
ventilation conditions.
Figure 10. Correlation of temperature Evolution in the mine drift for 212.5 kW (0.5D) fire under differ
ent ventilation conditions.
Figure 11. Correlation of temperature Evolution in the mine drift for 85.0 kW (0.3D) fire under differ
ent ventilation conditions.
408
(Ingason and Li, 2010) developed a model using a reduced-scale tunnel with no mechanical
smoke extraction. Zhao et al. (Zhao et al., 2019) developed a temperature attenuation model
using a reduced-scale model tunnel too but with a two-point mechanical smoke extraction
point. However, the attenuation model developed in this study was based on a single-point
smoke extraction system. It can be deduced from
Figure 12. Comparison of the temperature attenuation model developed and existing typical tempera
ture attenuation models.
5 CONCLUSIONS
Full-scale fire tests were conducted in underground mine drift. The analysis of temperature
attenuation and maximum excess gas temperature beneath a mine drift is presented in this
study using experimental and theoretical approaches. The temperature evolution upstream
and downstream of the mine drift were fitted using exponential correlation and a new empir
ical model of temperature attenuation given by Equation (10) was proposed. The attenuation
model upstream of the fire could predict the temperature rise as we approach the fire source
from a reference point upstream of the fire. This correlation is valid for HRR value between
85-425 KW and maximum excess gas temperature of up to 160 °C. This type of prognostica
tive attenuation model will further improve firefighting and suppression strategy as responders
move towards the fire zone during an emergency. Finally, the findings from this study could
serve as fire-field data in building and optimizing emergency evacuation plans, in addition to
validating existing CFD models for underground mines and other confined space
environments.
REFERENCES
Bahrami, D., Zhou, L. & Yuan, L. (2021) Field Verification of an Improved Mine Fire Location Model.
Mining, Metallurgy & Exploration, 38, 559–566.
De Rosa, M. I. (2004) Analysis of mine fires for all US metal/nonmetal mining categories: 1990–2001.
Delichatsios, M. A. (1981) The flow of fire gases under a beamed ceiling. Combustion and Flame, 43,
1–10.
Fernández-Alaiz, F., Castañón, A. M., Gómez-Fernández, F. & Bascompta, M. (2020) Mine fire behav
ior under different ventilation conditions: Real-scale tests and CFD modeling. Applied sciences, 10,
3380.
409
Hu, L., Chen, L., Wu, L., Li, Y., Zhang, J. & Meng, N. (2013) An experimental investigation and correl
ation on buoyant gas temperature below ceiling in a slopping tunnel fire. Applied Thermal Engineering,
51, 246–254.
Hu, L., Huo, R., Wang, H., Li, Y. & Yang, R. (2007) Experimental studies on fire-induced buoyant
smoke temperature distribution along tunnel ceiling. Building and Environment, 42, 3905–3915.
Huang, Y., Li, Y., Dong, B., Li, J. & Liang, Q. (2018) Numerical investigation on the maximum ceiling
temperature and longitudinal decay in a sealing tunnel fire. Tunnelling and Underground Space Tech
nology, 72, 120–130.
Ingason, H. & Li, Y. Z. (2010) Model scale tunnel fire tests with longitudinal ventilation. Fire safety jour
nal, 45, 371–384.
Iqbal, N., Salley, M. H. & Weerakkody, S. (2004) Fire dynamics tools (FDTs): quantitative fire hazard
analysis methods for the US Nuclear regulatory commission fire protection inspection program,
United States Nuclear Regulatory Commission, Office of Nuclear Regulatory . . ..
Li, Y. Z. & Ingason, H. (2012) The maximum ceiling gas temperature in a large tunnel fire. Fire safety
journal, 48, 38–48.
Marková, I., Lauko, J., Makovická Osvaldová, L., Mózer, V., Svetlík, J., Monoši, M. & Orinčák, M.
(2020) Fire size of gasoline pool fires. International journal of environmental research and public health,
17, 411.
Tang, F., Cao, Z., He, Z., Ling, X. & Wang, Q. (2018a) Thermal plume temperature profile of
buoyancy-driven ceiling jet in a channel fire using ceiling smoke extraction. Tunnelling and Under
ground Space Technology, 78, 215–221.
Tang, F., Cao, Z., Palacios, A. & Wang, Q. (2018b) A study on the maximum temperature of ceiling jet
induced by rectangular-source fires in a tunnel using ceiling smoke extraction. International Journal of
Thermal Sciences, 127, 329–334.
Tang, F., Chen, L., Chen, Y. & Pang, H. (2020) Experimental study on the effect of ceiling mechanical
smoke extraction system on transverse temperature decay induced by ceiling jet in the tunnel. Inter
national Journal of Thermal Sciences, 152, 106294.
Tao, L. & Zeng, Y. (2022) Effect of single-side centralized exhaust on smoke control and temperature
distribution in longitudinal ventilation tunnel fires. Tunnelling and Underground Space Technology,
119, 104241.
Tao, L., Zeng, Y., Li, J., Yang, G., Fang, Y. & Li, B. (2022) Study on the maximum temperature and
temperature decay in single-side centralized smoke exhaust tunnel fires. International Journal of Ther
mal Sciences, 172, 107277.
Yuan, L., Zhou, L. & Smith, A. C. (2016) Modeling carbon monoxide spread in underground mine fires.
Applied Thermal Engineering, 100, 1319–1326.
Zhang, X. (2012) Underground mine fire simulation using multiscale modeling approach, Missouri Univer
sity of Science and Technology.
Zhao, P., Yuan, Z., Yuan, Y., Yu, N. & Yu, T. (2019) A study on ceiling temperature distribution and
critical exhaust volumetric flow rate in a long-distance subway tunnel fire with a two-point extraction
ventilation system. Energies, 12, 1411.
Zhao, S., Liu, F., Wang, F. & Weng, M. (2018) Experimental studies on fire-induced temperature distri
bution below ceiling in a longitudinal ventilated metro tunnel. Tunnelling and Underground Space
Technology, 72, 281–293.
Zhou, L. (2009) Improvement of the mine fire simulation program MFIRE.
Zhou, L., Yuan, L., Bahrami, D., Thomas, R. A. & Rowland, J. H. (2018) Numerical and experimental
investigation of carbon monoxide spread in underground mine fires. Journal of Fire Sciences, 36,
406–418.
410
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
Charles Kocsis, Felipe Calizaya, Jeffrey Johnson, Tulio Dias & Natanna Nunes
Department of Mining Engineering, University of Utah
John Poulos
John Poulos, Geosyntec Consultants Inc., Lakewood, Colorado
1 INTRODUCTION
DOI: 10.1201/9781003429241-43
411
mine uses a blower-type ventilation system equipped with two surface fans. The longwall
panel is ventilated by a bleederless ventilation system where the mined-out areas are isolated
by means of high-pressure seals and the gob is inertized with nitrogen gas that is injected from
two to five cross-cuts inby to the face from the headgate side (Calizaya et.al. 2016).
The aim of this project is to construct two pressure chambers at the West Elk mine: one in
a mined-out (passive) area, and another near an active longwall mine gob. Each chamber would
be equipped with a nitrogen injection system, flow control valves, and pressure sensors. The
chambers would be tested under different ventilation conditions. This study presents a summary
of the results achieved, and the steps taken to minimize the risk of starting a fire in a mine gob.
412
causing emissions of the air-gas mixture from the gob. Similarly, when the barometric pressure
increases the gas volume will contract to allow the ventilation air to migrate into the gob
(Francart & Beiter 1997, & Schatzel et al. 2012). Pressure changes of this kind can be con
trolled by operating an active pressure-balancing system.
Following a mine visit, the University of Utah and the mine personnel decided to develop an
action plan to start the construction of the 1st chamber at the West Elk mine. The chamber would
be established by constructing a Kennedy stopping at about 3 m (10 ft) in front of an isolation
seal in a mined-out area. The standard stopping design used in the mine would be modified to
include two personnel doors and curtains to ventilate the chamber and to allow the mine person
nel to inspect the seal. The chamber would be equipped with a set of manometers, pressure trans
ducers, gas sampling ports, and a nitrogen gas injection system, connected through control valves
to the mine’s nitrogen pipeline.
413
Figure 1. Plan view of the modified pressure chamber.
Figure 2. West Elk mine pressure chamber construction details: (a) Kennedy stopping and digital man
ometers, and (b) Nitrogen injection pipeline and flow control valves.
414
Figure 3. Second pressure chamber: (a) Kennedy stopping and nitrogen injection system and (b) Hand-
held pressure sensors tapped to mine’s AMS.
In a mine ventilated by a blower fan, such as at the West Elk mine, a negative pressure differ
ence across the isolation seal (P3 – P2 < 0) means that the ventilating air (oxygen) is flowing to
the gob through or around the seal. This may be sufficient to start the Sponcom process. The
risk can be reduced by injecting pressurized nitrogen into the chamber.
Several pressure balancing tests were conducted on each pressure chamber. Once the cham
ber was constructed and equipped with a nitrogen injection system and pressure sensors, pres
surized nitrogen was injected into the chamber and the pressure differentials were monitored.
Figure 4 shows a schematic of the nitrogen injection system. The system consists of a 5 cm
HDPE pipe extending between the main nitrogen source and the pressure chamber. Valves V1
and V2, regulators R1 and R2, and pressure gages G1 through G3 were all used in the 1st
chamber. However, for practical reasons, regulator R2 and gage G3 were removed in the 2nd
chamber. During each test, nitrogen flow rate and pressure differentials across the seal and
stopping were monitored continuously. While the seals held high-pressure differentials effect
ively, the stoppings did not, particularly in the 1st chamber. To overcome the problem, the
Kennedy stopping was reinforced and sealant material was applied between the panels from
the high-pressure side. This helped but was not sufficient to hold a pressure difference of 1,000
Pa. Lessons learned from the 1st chamber were applied in the construction, testing, and moni
toring of the 2nd chamber. The results of each test are presented below.
415
variations on the pressure chamber, then, the chamber was filled with pressurized nitrogen and
evaluated for leakage through the Kennedy stopping. When a fixed flow rate of nitrogen was
injected into the chamber, the chamber pressure increased to a higher level. When the flow was
shut off, the chamber pressure dropped to its initial level rapidly. This was mainly due to gas
leakage through the Kennedy stopping. The results of the two experiments are presented below.
Figure 5. Barometric pressure and differential pressure variation across the seal and stopping.
416
Figure 6. Pressure differentials series after nitrogen gas injection (December 2-3, 2021).
11:25 AM. For regulator R1 set to minimum flow, Valves V1 and V2 were open. Pressure
gages: G1 = 400 kPa (58 psi), G2 = 0, DP2-1 = 42 Pa (0.17 in w.g)
11:30 AM. Regulator position was increased by ½ turn. Pressure gages recorded: G1 = 385
kPa (56 psi), and G2 = 20.7 kPa (3 psi), and DP2-1 = 243 Pa (0.97 in-wg)
11:37 AM. Regulator position increased by another ½ turn. Gage pressures: G1 = 365 kPa
(53 psi), and G2 = 20.7 kPa (3 psi), and DP2-1 = 492 Pa (1.97 in-wg)
11:41 AM. Regulator position increased by another ½ turn. Gage pressures: G1 = 379 kPa
(55 psi), and G2 = 34 kPa (5 psi), and DP2-1 = 630 Pa (2.52 in-wg)
11:42 AM. Regulator position increased by ½ turn. Gage pressures: G1 = 379 kPa (49 psi),
and G2 = 28 kPa (4 psi), and, DP2-1 = 808 Pa (3.23 in-wg)
11:48 AM. Significant leakage of nitrogen through door gaskets was detected. The nitrogen
control valves were shut off. It took 20 sec. to depressurize the chamber.
Figure 7 shows barometric pressure variations (actual, not standardized), and pressure differ
entials (chamber-entry and gob-chamber) recorded during this test. Notice that when the gas
flow rate increased, the pressure differentials across the seal and stopping increased rapidly
and remained constant thereafter, indicating that both walls can hold pressurized nitrogen
with reduced leakage. Table 1 summarized the differential pressures across the chamber walls.
417
Figure 7. Pressure differentials across the chamber stopping and seal – Test 1.
Table 1. Pressure differentials across the chamber stopping and seal – Test 1.
418
Figure 8 shows barometric pressures and pressure differentials across the stopping and seal
during the test. Notice a slight decline when the differential pressure across the stopping
reached 6.41 in w.g. Table 2 shows the differential pressures across the chamber walls.
Figure 8. Pressure differentials across the chamber stopping and seal – Test 2.
Table 2. Pressure differentials across the chamber stopping and seal – Test 2.
The above results show that the balance chamber will hold pressurized nitrogen as long as
a steady flow of nitrogen gas is supplied to the chamber. Because of leakage though the per
sonnel doors and fractured coal around the periphery of the stopping, the chamber was not
able to hold pressurized gas for a long time. In test 1, when the flow control valves were shut
off, it took only 20 seconds to depressurize the chamber. When the leakage points were fixed
for the second test, the chamber took approximately 45 seconds to depressurize. To improve
the quality of the stopping and to reduce leakage, it is recommended to reverse the position of
the access doors so that the door gaskets are compressed when pressurizing the chamber.
Two pressure chambers were constructed at the MCC’s West Elk Mine. Each chamber was
established by installing a Kennedy stopping at about 3 m (10 ft) in front of an existing seal
equipped with a nitrogen injection system and a set of monitors linked to the mine’s AMS.
419
Several pressure balancing tests were conducted on each chamber. Each test started by
injecting pressurized nitrogen into the chamber and monitoring for changes in barometric
pressure, along with the pressure differentials across the chamber’s seal and stopping. Signifi
cant leakage of nitrogen through the Kennedy stopping was detected.
Changes in barometric pressure in an underground coal mine can cause “in-gassing”, or
“out-gassing” conditions at the permanent seals. At the West Elk Mine, the barometric pres
sure varies daily as the air temperature on the surface increases or decreases. In this mine, on
average, this daily change is in the order of 2.5 in w.g.
When a fixed rate of nitrogen was injected into the 1st chamber, the pressure differentials
across the seal increased from -0.4 in. w.g. to +3.48 in. w.g. No major gage pressure changes
across the stopping were observed. The results showed that while the permanent seal is doing
the intended work, the Kennedy stopping still needs additional work to reduce leakage.
Once established and equipped, the 2nd chamber was tested for different nitrogen flow
rates. The results showed that the balance chamber will hold pressurized nitrogen as long as
a steady flow of nitrogen is supplied. Under the existing conditions, when the nitrogen flow to
the chamber is stopped, the chamber will depressurize in less than one minute.
ACKNOWLEDGMENTS
The authors would like to acknowledge the Alpha Foundation for providing financial support
for this project. The authors would also like to express their gratitude to Mr. John Poulos,
Mining Consultant at Geosyntec Inc., Jessica Wilczek, Manager of Engineering and Environ
mental Affairs at MCC, Elizabeth Lindgren, Ventilation Engineer at MCC, Rick Chermak,
Project Construction Manager at MCC, and Garren Atchley, Mine Engineer at MCC for
their support in developing this project.
DISCLOSURE
This study was sponsored by the Alpha Foundation for the Improvement of Mine Safety and
Health, Inc. The views, opinions, and recommendations expressed herein are solely those of
the authors and do not imply any endorsement by the ALPHA FOUNDATION, its Direct
ors, and its staff.
REFERENCES
Bessinger, S.L., Abrahamse, J.M., Bahe, K.A., McCluskey, J.E. & Palm, A.T. 2005. Nitrogen inertization
at San Juan Coal Company’s longwall operation. SME Annual Meeting, Salt Lake City, UT.
Bhowmick, B.C., et al. 1992. Dynamic Balancing of Pressure for Control of Fire in Underground Coal
Mines. Journal of Mining Research. 1:1. New Delhi.
Calizaya F., Nelson M. G., Bateman, C., and Jha A. 2016. Pressure Balancing Techniques to Control
Spontaneous combustion. SME Annual Meeting, Preprint 16–050. Phoenix, AZ.
Chalmers, D.R. 2008. Sealing design. Proceedings of the 12th North American/US Mine Ventilation
Symposium, Reno, NV: University of Nevada. pp. 219–223.
Francart, W.J. & Beiter, B.A. 1997. Barometric Pressure Influence in Mine Fire Sealing. Proceedings of
the 6th International Mine Ventilation Congress. Littleton, CO:S ME. pp. 341–342.
Grubb, J.W., 2008. Preventative Measures of Spontaneous Combustion in Underground Coal Mines.
Ph.D. dissertation, Colorado School of Mines, Golden, CO.
Moreby, R. 2009. Ventilation. Australasian Coal Mining Practice. Carlton, Victoria: Australasian Insti
tute of Mining and Metallurgy.
Ray, S.K. & Singh, R.P. 2007. Recent Developments and Practices to Control Fire in Underground Coal
Mines. Fire Technology (Central Mining Research Institute, Dhanbad, India). 43: 285–300.
Schatzel, S.J., Krog, R.B. & Dougherty H. 2012.Methane Emissions and Airflow patterns on a Longwall
Face. SME Annual Meeting, Preprint 12–016. Seattle, WA.
Smith, A.C. & Lazzara, C.P. 1987. Spontaneous Combustion Studies of U.S. Coals, USBM RI 9079.
420
Mine gases
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
D.T. Hristopulos
Technical University of Crete, Chania, Crete, Greece
ABSTRACT: Prediction of methane gas concentrations in underground coal mines has been
a topic of interest for academia and the mining industry for decades as methane gas is con
sidered a significant hazard to mining safety and productivity. This study utilizes multivariate
time series forecasting approaches to predict methane emissions in an active mine. The multi
variate method incorporates the past values of methane gas concentrations as a predictor of
its future values but also integrates the correlations between gas concentration and barometric
pressure as well as barometric pressure and coal production rate in predictive models of
methane gas emissions. The data used in this study were collected from the atmospheric moni
toring systems of an underground coal mine operation in the eastern USA.
The data were stored, pre-processed, and statistically analyzed to develop multivariate fore
casting models based on the Vector Autoregression (VAR) and Autoregressive Integrated
Moving Average with Exogenous Input (ARIMAX) methods. The optimal model for each
method was selected based on the Akaike Information Criterion. The forecasting performance
and accuracy of each model were assessed using cross-validation to establish the best overall
model. It was concluded that both VAR and ARIMAX could, in most cases, adequately fore
cast methane gas concentrations in underground coal mines.
1 INTRODUCTION
The number of fatalities due to methane gas explosions in underground coal mines has signifi
cantly decreased in recent decades, mainly due to the extraordinary effort made by different
organizations such as the mining industry, governmental agencies, and academia to develop
new methods and techniques to monitor and more efficiently control methane gas emissions
and concentrations in underground coal mine operations (Taylor and Karacan, 2010; Diaz
et al., 2022). However, methane gas explosions still represent the most feared threat in the coal
mining industry worldwide. For example, in 2009, more than 750 coal miners lost their lives in
over 150 methane gas explosions in China (Cheng, 2018). Furthermore, in the U.S. coal mining
industry, methane gas explosions are also considered the most dangerous hazard; since 1900,
more than 11,600 underground coal mine workers have died in more than 500 mining disasters
(NIOSH, 2022). Other examples include the catastrophe at the Soma Coal Mine (Turkey) in
2014, which killed 300 underground coal miners and has been documented as the worst disaster
in the recent history of the mining industry (Düzgün and Leveson, 2018).
Methane gas disasters have been the primary motivation for exploring different methodolo
gies that could accurately predict methane gas concentrations in underground coal mine oper
ations. As a result, different approaches have been implemented to tackle this problem. For
example, these include empirical, numerical, and statistical techniques. Indeed, a statistical
approach based on time series analysis has recently been employed to develop different
DOI: 10.1201/9781003429241-44
423
forecast models capable for methane gas concentrations in underground coal mines using uni
variate and multivariate time series data (Diaz et al., 2021a, b; 2022).
Past research has demonstrated that methane gas concentrations are autocorrelated and
that the past values of methane gas time series can be used to predict its future values allowing
the implementation of univariate time series analyses (Dixon, 1992; Dixon and Longson,
1993). On the other hand, other studies have established that methane gas concentrations cor
relate with barometric pressure (BP) and coal production rate, permitting the use of multivari
ate time series analysis (Wasilewski, 2014; Hemp, 1994; Xu et al., 2014). For example,
methane gas and coal production rate are positively correlated, while methane gas and baro
metric pressure are negatively correlated (Diaz et al., 2021a, b; 2022).
The research presented in this paper explains the main steps for developing different multivari
ate forecasting methods to predict methane emissions at an active mine. The multivariate method
incorporates the past values of methane gas concentrations as a predictor of its future values but
also integrates barometric pressure as well as barometric pressure and coal production rate as
predictors of methane gas concentration. The mine data (methane gas and coal production rate)
were collected by the atmospheric monitoring system (AMS) of an active, underground coal
mine in the eastern US. The weather data (barometric pressure) were retrieved from the closest
public weather station to the case study. Furthermore, various statistical techniques were imple
mented to assess the association (cross-correlation) between the methane gas concentration and
the exogenous variables. Such associations were employed to develop two multivariate forecast
ing approaches. Finally, the optimal model for each approach was selected using the Akaike
Information Criterion (AIC). The results obtained from the two forecasting approaches were
compared using cross-validation metrics to determine the best model overall.
424
based on someone’s intuition, knowledge, and experience. Quantitative forecasting methods,
on the other hand, are based on mathematical modeling and are the focus of the study pre
sented in this paper (Shumway and Stoffer, 2017). Univariate and multivariate forecasting
methods are the most common quantitative forecasting models employed.
Univariate techniques estimate the future values of a process utilizing past values of the pro
cess (time series). In contrast, multivariate forecasting methods incorporate more than one
correlated time series and utilize past values from all the variables that are mutually analyzed.
In addition, multivariate methods offer a more comprehensive understanding of the studied
process and, in most cases, higher accuracy than univariate forecasting methods (Chatfield,
2004; Shumway and Stoffer, 2017). Therefore, multivariate forecasting approaches with one
and two auxiliary variables are employed herein to forecast methane gas concentrations.
425
methane gas and barometric pressure in order to predict methane gas concentrations. Further
more, The ARIMAX(p,d,q) model is employed herein using d = q = 0 in order to compare its
performance and accuracy with the VAR(p) model. The autoregressive order p* of the optimal
ARIMAX model is selected based on the lowest AIC.
This section introduces the results of methane gas concentration forecasting using the ARIMAX
(p, d, q) and the VAR(p) models. Four datasets (from 1 to 4) with daily averaged time series were
employed to run both models. The time series in datasets 1 and 3 correspond to three months
(from October 2019 to December 2019), while those in datasets 2 and 4 correspond to six months
(from October 2019 to March 2020). Datasets 1 and 2 consist of two time series: (i) the methane
concentration time series (dependent variable) and (ii) the barometric pressure time series (inde
pendent variable). Datasets 3 and 4, on the other hand, include three time series: (i) the methane
concentration time series, (ii) the barometric pressure time series, and (iii) the coal production
rate time series. The last two time series are considered auxiliary predictors.
Figures 1 and 2 consist of four plots (a to d). Graphs (a) and (b) are the forecasts obtained
using the multivariate (VAR and ARIMAX) one-step-ahead approaches with one auxiliary
(barometric pressure) variable, while plots (c) and (d) are the forecasts developed using two
auxiliary (barometric pressure and coal production) variables, respectively. Additionally, the
blue line represents the training data (methane gas time series), the upper and lower black
dashed lines signify the 95% prediction confidence interval, and the red line represents the
forecast. The validation and forecast periods contain five percent (5%) of the total points
(data records) used in the training time series.
Figure 1 presents the forecasting results using the ARIMAX(p,d,q) model. Figure 1a shows the
forecast for dataset 1; the optimal autoregressive order is p = 4 based on the lowest AIC. Visual
inspection demonstrates that the forecast (red line) is close to the actual values in the validation
period (blue line). Furthermore, the correlation coefficient (henceforth, R) is calculated as R = 0.78,
implying a strong correlation between the validation data and the forecast. Similarly, Figure 1b dis
plays Forecast 2; the best possible autoregressive order is p = 9 according to the AIC and R = 0.67,
indicating similar variations between the validation data and the forecast. Likewise, Figures 1c and
1d show forecasts for the gas concentration time series over the same periods presented in Figures 1a
and 1b, respectively, using two auxiliary (independent) variables. Figure 1c shows that the optimal
autoregressive order is p = 1 based on the lowest AIC. In this case, the forecast (red line) is closer to
the actual values in the validation period (blue line), with a higher correlation coefficient R = 0.87
than the one obtained in Figure 1a (see Table 1). Similarly, in Figure 1c, the best possible autoregres
sive order is p = 2 and R = 0.76, also higher than the one in Figure 1b. Finally, in all cases, the
actual values and the forecasts are captured by the 95% prediction confidence interval.
Figure 2 introduces the forecasting results using the VAR(p) model. Figures 1a and 1b show the
forecast for datasets 1 and 2, respectively. The best model in Figure 1a has p = 2. As a result, the
correlation coefficient between the forecasts and the validation data is R = 0.63, which is lower than
the value obtained with the ARIMAX model with one exogenous variable (Figure 1a). Similarly, in
Figure 1b, the best model is obtained for p = 2. The correlation coefficient is R = 0.64, slightly lower
than the one shown in Figure 1b. Furthermore, Figures 1c and 1d present the forecast for datasets 3
426
and 4 correspondingly. In both cases, the best model is obtained for p = 2 according to the lowest
AIC. The correlation coefficients are R = 0.66 and R = 0.65, respectively, which are lower than
those shown in Figures 1c and 1d, as presented in Table 2.
Tables 1 and 2 demonstrate that the multivariate ARIMAX(p,d,q) and VAR(p) forecasting
methods with one and two auxiliary variables can predict methane gas concentrations. In
most cases, the methane gas concentration forecasted by both models matches the direction
(increase/decrease) of the validation data. Furthermore, the observed methane gas levels are
continuously captured by the 95% prediction intervals, as illustrated in Figures 1 and 2. In
addition, the CV correlation between the forecasts and the validation data is strong and posi
tive, and the values of other CV measures are similar (see Tables 1 and 2).
The results obtained herein show that the forecasting model is improved (a higher correl
ation between the forecast and the validation data is achieved) by using two auxiliary (baro
metric pressure and coal production rate) variables than one (barometric) pressure auxiliary
variable. Furthermore, in all cases presented in Figure 2 (VAR model), the results were infer
ior to those shown in Figure 1 (ARIMAX model).
Figure 1. ARIMAX(p,d,q) one-step-ahead CH4 concentration forecasts using one and two exogenous vari
ables: (a) forecast for dataset 1; (b) forecast for dataset 2; (c) forecast for dataset 3; (d) forecast for dataset 4.
4 CONCLUSIONS
Implementing barometric pressure time series and coal production time series as auxiliary pre
dictors provided better methane gas concentration forecasts than those obtained with only
one auxiliary variable (i.e., barometric pressure).
The ARIMAX(p,d,q) one-step-ahead forecast model offered better cross-validation per
formance (higher correlation between the forecast and the validation data) than the ones
427
obtained using the VAR (p) one-step-ahead forecast. On the other hand, both approaches
gave similar results for the other cross-validation measures (i.e., RMSE, MAE, and ME). The
performance of both forecasting approaches should be further assessed with different datasets
and operating conditions.
The ARIMAX and VAR forecast methods proposed in this research provide a potential
solution to the critical problem of forecasting methane gas concentrations in underground
coal mines. Accurate forecasts of methane concentration can lead to a warning system that
will improve the safety conditions of underground coal miners worldwide.
Table 1. Summary of methane gas forecasting results for the ARIMAX(p,d,q) one-step-ahead model
(using one exogenous (barometric pressure) and two exogenous (barometric pressure and coal production
rate) time series as independent variables). p*: Optimal autoregressive order of ARIMAX model based
on the lowest AIC.
Data Length (days)
Number of Auxiliary Training Validation
Variables Data Data p* R RMSE ME MAE
Figure 2. VAR(p) one-step-ahead CH4 concentration forecasts using one and two exogenous variables:
(a) forecast for dataset 1; (b) forecast for dataset 2; (c) forecast for dataset 3; (d) forecast for dataset 4.
428
Table 2. Summary of methane gas forecasting results for the VAR(p) one-step-ahead model (using
a multivariate vector of methane gas concentration, barometric pressure, and coal production rate).
p*: Optimal autoregressive order of VAR model based on the lowest AIC.
Data Length (days)
Number of Exogenous Training Validation
Variables Data Data p* R RMSE ME MAE
ACKNOWLEDGMENT
This study was sponsored by the Alpha Foundation for the Improvement of Mine Safety and
Health, Inc. (ALPHA FOUNDATION). The views, opinions, and recommendations
expressed herein are solely those of the authors and do not imply any endorsement by the
ALPHA FOUNDATION, its Directors, and staff.
REFERENCES
Andrews, B.H. Dean, M.D. Swain, R. Cole, C. 2013. Building ARIMA and ARIMAX models for pre
dicting long-term disability benefit application rates in the public/private sectors. Society of Actuaries.
University of Southern Maine, Maine, USA.
Box, G.E. Jenkins, G.M. Reinsel G.C. Ljung, G.M. 2015. Time Series Analysis: Forecasting and Control,
5th edn. John Wiley & Sons, Hoboken, NJ, USA.
Brockwell, P. J. and Davis, R. A. 2016 Introduction to Time Series and Forecasting. 3rd edition,
Springer. https://doi.org/10.2307/2965440.
Cerqueira, V. Torgo, L. Mozetič, I. 2020. Evaluating time series forecasting models: an empirical study
on performance estimation methods. Machine Learning. https://doi.org/10.1007/s10994-020-05910-7.
Chatfield, C. 2004. The Analysis of Time Series - An Introduction. 6th Edition. Chapman & Hall/CRC,
New York, USA.
Cheng, J. 2018. Explosions in Underground Coal Mines - Risk Assessment and Control. Springer.
https://doi.org/10.1007/978-3-319-74893-1.
Diaz, J. Agioutantis, Z. Hristopulos, D.T. Schafrik, S. 2021a. Managing and utilizing big data in atmos
pheric monitoring systems for underground coal mines. Materials Proceedings 5(1):78. https://doi.org/
10.3390/materproc2021005078.
Diaz, J. Agioutantis, Z. Hristopulos, D.T. Schafrik, S. 2021b. Towards atmospheric monitoring data
analysis in underground coal mines. In: Tukkaraja P (ed) Proceedings of the 18th North American
Mine Ventilation Symposium (NAMVS 2021), June 21–23. CRC Press, Boca Raton, FL, pp 498–506,
https://doi.org/10.1201/9781003188476-51.
Diaz, J. Agioutantis, Z. Hristopulos, D.T. Schafrik, S. Luxbacher, K. 2022. Time Series Modeling of
Methane Gas in Underground Mines. Mining, Metallurgy & Exploration. Springer, https://doi.org/
10.1007/s42461-022-00654-5.
Dixon, D.W. 1992. A statistical analysis of monitored data for methane prediction. PhD thesis, Univer
sity of Nottingham, URL http://eprints.nottingham.ac.uk/12977/1/334922.pdf
Dixon, D.W. Longson, I. 1993. A statistical method for methane prediction and improved environmental
control. In: Proceedings of the 6th US Mine Ventilation Symposium, Salt Lake City, UT, USA.
Düzgün, H.S. Leveson, N. 2018. Analysis of Soma mine disaster using causal analysis based on systems
theory (CAST). Saf. Sci. Volume 110, Part A, December 2018, 37–57. https://doi.org/10.1016/j.
ssci.2018.07.028.
Hemp, R. 1994. The effect of changes in barometric pressure on mines in the highveld of South Africa.
Journal of the Southern African Institute of Mining and Metallurgy 94(6):133–146.
Hristopulos D.T. 2020. Random Fields for Spatial Data Modeling: A Primer for Scientists and Engin
eers. Springer, Netherlands, https://doi.org/10.1007/978-94-024-1918-4
Kirchgässner G, Wolters J. 2007. Introduction to Modern Time Series Analysis. Springer, Berlin,
Germany.
429
Lütkepohl, H. 2005. New Introduction to Multiple Time Series Analysis. Edited by Springer Science &
Business Media. Berlin, Germany.
Shumway, R.H. Stoffer, D.S. 2017. Time series analysis and its applications, 4th edn. Springer.
Switzerland.
Taylor, C. Karacan, C.Ö. 2010. Historical development of technologies for controlling methane in under
ground coal mines. In: Extracting the Science: A century of mining research. Brune JF (ed), Littleton,
CO, Pittsburgh, USA.
NIOSH. 2022. NIOSH data and statistics. https://wwwn.cdc.gov/NIOSH-Mining/MMWC/MineDisa
sters/Count. The National Institute for Occupational Safety and Health Accessed August 02, 2022.
Tsay, R.S. 2015. Multivariate Time Series Analysis with R and Financial Applications. First Ed. New
Jersey, USA: John Wiley & Sons, Inc.
Wasilewski, S. 2014. Influence of Barometric Pressure Changes on Ventilation Conditions in Deep
Mines. Archives of Mining Sciences, 59(3), pp. 621–639. https://doi.org/10.2478/amsc-2014-0044.
Xu, L. Lin, X. Amen, J. Welding, K. McDermitt, D. 2014. Impact of changes in barometric pressure on
landfill methane emission. Global Biogeochemical Cycles, 28(7), pp. 679–695. https://doi.org/10.1002/
2013GB004571.
430
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
P.C. Thakur
West Virginia University, Morgantown, WV, USA
ABSTRACT: Longwall mining is the safest and most economical way to mine coal in under
ground mines. At present 37 longwall panels in 11 states are mining more than 50% of under
ground mined coal. Proper degasification of the panels prior to mining and post mining
promotes mine health and safety and improves productivity. Produced gas can be sold for
added profit. For proper degasification, all coal seams are divided as shallow, medium-depth
and deep coal seams. The degasification scheme for a typical mine in each category is dis
cussed in detail. Special attention is given to floor gas capture that has led to some recent coal
mine disasters.
1 INTRODUCTION
1.1 Background
Majority of global coal production comes from underground mining which is done by two
methods: (1) room and pillar and (2) longwall mining. As shallow coal seams are mined out
and mining reaches greater depths, longwall mining becomes the preferred method of mining.
In United States, more than 50% of all underground mined coal is mined by the longwall
method. Table 1 shows the distribution of 37 longwall panels in U.S. coal mines.
DOI: 10.1201/9781003429241-45
431
The trend for panel sizes and mining equipment in the coal industry is to continue to go
upward pushing production capacities and productivity to higher levels. Another reason for
increasing the longwall panel size is that development sections cannot keep up with the rate of
longwall advance. Increasing the width of the longwall panel slows down the rate of advance.
Today it is quite realistic to plan longwall panels in mildly gassy coal seams that are 1500 ft
wide and 15,000 ft long containing nearly 5 to 7 million tons of raw coal. Such large longwall
panels offer many benefits as follows:
• Improved safety and reduced injury rate
• Improved recovery of in situ coal reserve
• Improved productivity and reduced cost/ton
On the other hand, large longwall panels cause some concerns, such as follows:
• Ventilation and methane control
• Respirable dust control
• Increased likelihood of geological anomalies, e.g., faults, washouts, etc.
These problems become more serious when the gas content of coal seams increases with the
increase in mining depths. Costs of mine ventilation and coal seam degasification go up, and
it becomes necessary to optimize the two processes to keep the mine environment safe and the
combined cost at the minimum. Drilling process for degasification can also locate any geo
logical anomalies in the panel prior to mining and helps in respirable dust control.
The optimized plan for mine degasification and ventilation with added benefit of coal seam
exploration can be summarized as follows:
1. Recover as much coal bed methane (CBM) as possible prior to mining and postmining to
minimize ventilation requirements.
2. Optimize longwall panel sizes and the size of the district depending on the gas content of
the coal seams to keep the combined cost of degasification and ventilation down.
3. Plan mine ventilation to maximize ventilation efficiency measured as the ratio of air horse
power delivered to face to the total air horsepower created at the fans.
4. Seal all mined out areas with approved design of stoppings and recover as much CBM as
possible without making the atmosphere in the sealed area explosive.
5. Measure coal seam thickness while drilling and locate geological anomalies if any.
6. Gather all produced gases and market them after processing to defray the cost of degasifi
cation and ventilation.
In the development phase of mining, the only source of gas is the coal seam being mined.
Ideally, gas content should be reduced to 100 ft3/t for safe and efficient mining. This is
achieved by pre-mining methane drainage as discussed later in the paper.
Most of gas in longwall mining comes from the mined-out area, called ‘gob’. Longwall
mining subsides the overlying coal seams and makes the underlying coal seams heave. Various
estimates of the area called ‘gas emission space’ is shown in Figure 1.
The vertical dimension of the gas emission space can range up to 1000 feet above the mined
coal seam and about 250 feet in the floor.
Factors which govern the rate of gas emission in the longwall gobs can be summarized as
follows:
1. The density and proximity of gas-bearing zones in the gas emission space.
2. The gas content of coal seams and other gas-bearing horizons.
3. The extent to which overlying and underlying coal seams have been mined.
4. The rate of coal face advance. This is the most important factor.
5. The presence of geological disturbances and leftover coal pillars.
6. The subsidence characteristics of the area which determines the permeability of the gas
emission space.
432
Figure 1. Vertical limits of the gas emission space.
The depth of a coal seam is only a rough indication of its gas content. Direct measurement
of gas contents is highly recommended.
433
coal seam). This parameter is a combination of reservoir pressure, permeability, porosity,
gas content, and diffusivity of the coal seam. Table 3 shows specific gas production for some
U.S. coal seams.
These coal seams are shallow in depth and rarely have thick coal seams overlying or under
lying the coal seam being mined.
As shown in Table 2, these coal seams have a depth of 600 – 1500 ft and a gas content of
100 – 300 ft3/ton. They need to be degassed pre-mining and post-mining. Such mines are
located in the Northern Appalachian Basin of Pennsylvania and West Virginia.
434
In-mine horizontal boreholes are about 1000 – 1500 feet long. Two boreholes are drilled from
each location: one to degas the development headings; the other across the longwall panel.
A well-designed drilling program can remove 50 – 60% of the in-situ gas making mining much
safer and highly productive.
These coal seams are 1500 feet to 3000 feet deep with very high gas contents as shown in
Table 2. Extensive efforts are needed for both pre-mining and post-mining degasification.
These mines are located in Central and Southern Appalachian Basins of USA.
435
below 200 ft3/ton. The longwall panel is drilled at every 100- or 200-feet interval across the
panel. Gas pipelines are laid to take all gases to surface for utilization. A typical 1000 ft
x 10,000 ft panel may need 10-12 frac-wells. The fracking process is fully descried elsewhere
(Thakur 2016).
436
Gas emissions rate in gobs is highly dependent on the specific gob emission and the rate of
mining in ft/day. A typical longwall panel 750 ft wide x 10,000 feet may need 30-35 gob wells
of 12” diameter. Blowers on these gob wells have a capacity of 2-4 MMCFD with a suction of
120 inches of water gauge. The width of the longwall panel is limited by the number of gob
wells needed. The maximum width is reached when the spacing between the gob wells equals
half the width of the panel (Thakur 2008).
The portion of the borehole below the mineable coal seam is filled with coarse sand and
small gravel, and a solid cement plug is put across the working coal seam. When longwall face
is mined through the borehole and moved 100-150 ft outby, the floor heaves. The partially
filled gob well provides a channel for gas to come out in the gob at a point that is 100-150 ft
inby of the longwall face. Such gob wells can be placed at close intervals (150-500 ft apart) on
the longwall panels that are prone to floor gas emissions. Gas production is assisted by
vacuum pumps creating 1-5 psi negative pressure.
437
7 SUMMARY AND CONCLUSIONS
Longwall mining is the safest and the most efficient method of mining underground coal. In
USA, more than 50% of all underground coal production is done by longwall mining. With
increasing depth, it is becoming necessary to degas the coal seams both before mining and
post mining. Coal seams are divided into three categories depending on their gas contents in
ft3/ton.
Pre-mining and post mining degasification of a typical coal seam in each category is
discussed. Modification of gob well construction to capture excessive floor emissions is
also discussed. Some deep coal mines can produce 70 – 80 MMCFD from gob wells
alone. Combining this production with gas production from frac-wells drilled in advance
of mining can create a gas flow of 200 – 250 MMCFD. At the current price of natural
gas at $6-9/MMCF, it can create a sizeable additional income to the coal mines defraying
all cost of degasification. Increased productivity and reduced cost of mining are added
benefits.
REFERENCES
Thakur, P.C. 1981. Methane Control for Longwall Gobs, Longwall-Shortwall Mining, State-of-the-Art,
Chapter 10, SME, AIME, pp 81–86.
Thakur, P.C. 2008. Optimum widths of longwall panels in highly gassy mines – Part II, 12th U.S. Mine
Ventilation Symposium, University of Nevada, Reno, Nevada.
Thakur, P.C. 2016. Advanced Reservoir and Production Engineering, Elsevier, pp. 210. Chapters 2-5; pp
17–73.
Thakur, P.C. 2016. Advanced Reservoir and Production Engineering, Elsevier, Chapter 8, pp 105–133.
Thakur, P.C. 2018. Advanced Mine Ventilation. Elsevier, pp 512. Chapter 14: pp 227–246.
438
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
Study of in-situ coal seam gas content for Australian coal and gas
outburst management: A field data analysis and laboratory
experiment
ABSTRACT: The gas content and composition are widely utilized in underground coal
mining to evaluate the potential for coal and gas outbursts. This study investigates the in-situ
coal seam gas content and gas sorption capacity in three coal seam sections (A1, A2, and A3)
in Sydney basin to understand their implications for gas management and outburst risk in
Australian coal mines. The fast direct desorption method and associated devices were
described in detail and employed to measure the in-situ coal seam gas components (Q1, Q2,
and Q3) and gas composition. This method offers a reliable and accurate approach for assess
ing coal seam gas content and composition, with an error factor of ± 15%. The results
revealed gas contents in the A1 and A3 sections exceeding Level 2 Threshold Limit Values
(TLV), necessitating gas drainage measures. High-pressure isothermal gas sorption experi
ments were conducted on coal samples from the A1, A2, and A3 seam sections to investigate
their gas sorption capacity. The Langmuir model demonstrated excellent performance in
describing CO2 sorption behavior, with fit coefficients R² greater than 0.99. The study
observed a strong positive correlation (Pearson’s coefficient >0.855) between in-situ gas con
tent and Langmuir volume, suggesting that higher sorption capacities correspond to higher in-
situ gas content in coal seams. The findings of this study highlight the importance of under
standing the relationship between in-situ gas content and gas sorption capacity in managing
gas emissions and mitigating outburst risks in underground coal mines. The insights derived
from the study can contribute to the development of effective gas management strategies and
enhance the safety and efficiency of coal mining operations in Australia.
1 INTRODUCTION
Coal mine gas emissions and outbursts have been long-standing challenges in the coal mining
industry, posing considerable threats to the safety of miners and the efficiency of mining oper
ations (Thakur 2016; Lei et al. 2023). The gas content within coal seams is a vital factor for
comprehending and addressing these concerns (Liu et al. 2020). Accurate measurement and
analysis of this parameter help to estimate coal seam gas reserves, assess the risk of coal and
gas outbursts, and develop appropriate gas management strategies (Black 2018).
Gas content measurement techniques can be classified into two primary categories: indirect
and direct (Diamond and Schatzel 1998). Indirect approaches depend on gas adsorption prop
erties under particular pressure and temperature conditions or empirical data linking gas con
tent with other factors such as gas pressure, emission rate, coal rank, and overburden depth
(Hou et al. 2020). Nevertheless, indirect methods present various drawbacks, including diffi
culties in acquiring precise in-situ seam gas pressure measurements, the presumption of com
plete gas saturation in coal seams, and the complexity in determining the composition of
DOI: 10.1201/9781003429241-46
439
multi-component seam gases (Xue and Yuan 2017). In contrast, direct methods assess the real
amount of gas liberated from coal specimens and can be categorized into slow and fast desorp
tion techniques. The slow desorption method requires weeks or months for a coal sample to
reach a low desorption rate or to stop desorbing, while the fast direct desorption method
expedites the process by crushing the sample upon arrival at the laboratory, obtaining gas
content results within days (Xue, Xie, and Wang 2013).
Since the introduction of the direct gas content measurement method by Bertard et al. (Ber
tard, Bruyet, and Gunther 1970), various systems have been developed for direct measurement
of coal gas content. And these methods and their subsequent modifications have been effective
in measuring coal seam gas content in cored coal samples. Bertard and his colleagues used an
arrangement of three distinct containers (allocated for initial measurement, transportation,
and crushing), a U-tube manometer, and a crusher to assess the volume of gas desorbed from
coal samples collected from operational mining sites (Bertard, Bruyet, and Gunther 1970).
Kissell et al. tested the applicability of the “direct method” for determining the methane con
tent of virgin coalbeds in the Vesta No. 5 coal (Kissell, McCulloch, and Elder 1973). They
conducted experiments using exploration cores to assess the method’s effectiveness in estimat
ing gas content and predicting the total gas emissions from a potential mine situated at the
exploration site. The results indicated that the direct method, when applied to exploration
cores, successfully provided accurate estimates of coalbed gas content and reasonably pre
dicted the overall gas emissions from the prospective mine. Xue and Yuan developed an
adapted direct method to measure gas content in coal seams by employing coal cuttings from
underground boreholes, specifically in cases of soft, friable, or geologically intricate coal
seams where acquiring cores may pose significant challenges (Xue and Yuan 2017; Xue et al.
2014). The modified method showed satisfactory results when applied in a soft seam at Huai
nan, China, demonstrating its potential as a viable alternative to the standard procedure for
determining gas content when cores are not available. These findings suggest that the direct
method can be a valuable tool in gas content analysis and mine planning, contributing to
improved safety and efficiency in coal mining operations.
The Sydney Basin in Australia is known for its abundant coal resources and complex geo
logical conditions, which make the assessment of gas content and sorption capacity particu
larly crucial for local coal mining operations (Faiz et al. 2007). Previous studies have
primarily focused on the measurement of gas content and composition using various direct
and indirect methods. However, there is a lack of comprehensive research on the relationship
between in-situ gas content and gas sorption capacity in the context of coal seams from the
Sydney Basin.
In this study, we aim to address this gap by investigating the in-situ coal seam gas content
and gas sorption capacity in three coal seam sections (A1, A2, and A3) in the Sydney Basin.
The fast direct desorption method and associated devices were employed to measure the in-
situ coal seam gas components (Q1, Q2, and Q3) and gas composition. High-pressure isother
mal gas sorption experiments were conducted on coal samples from these seam sections to
investigate their gas sorption capacity. The relationship between in-situ gas content and gas
sorption capacity is analyzed to understand its implications for gas management and outburst
risk in Australian coal mines.
2 METHODOLOGY
440
gas to desorb from the intact core before the final crushing stage. Coal gas content is exam
ined using the quick direct desorption method at the Coal Gas Laboratory A, and the gas
composition is assessed employing a 4-channel Micro Gas Chromatograph.
where q = the accumulated volume of desorbed gas since coal is placed in the desorption can
ister and measurement commenced, m3/t; t = the time since the start of gas desorption in the
hole, min; k = the slope of a regression line passed through canister desorption data.
Note: the desorbed volumes are expressed at STP conditions (20 ℃, 101.325kPa).
Figure 1. Lost gas content (Q1) calculation illustration (a) and its measuring apparatus (b) (Australia
2016).
441
increments determined over the measurement period. If there are n steps to complete the
measurement, the Q2 value can be expressed as:
where m = the mass of sample, g. ΔVi = the incremental volume of gas desorbed, which is the
difference between the current corrected gas volume at instant i and the previous corrected
volume in the system at instant i - 1 (provided the cylinder has not been refilled). i = 1 of the
measurement corresponds to the beginning of measurement and associated V is the volume of
void in the system at this instant.
442
Figure 3. Residual gas content measurement apparatus for estimation of Q3.
443
Figure 4. Sample preparation.
c) Degas the sorption canister for 2 hours using a vacuum pump till relative pressure reaches
-100kPa, and let it stand for 24 hours; inject helium to measure dead space volume.
d) Charge CO2 into the canister with samples, and test adsorption and desorption isotherms
at pressure gradient of 0.5 MPa, maximum pressure of 5.0 MPa, and equilibrium time of
24 hours.
e) Maintain thermostatic water bath and room temperature at 30°C during sorption test.
The Langmuir equation is a mathematical model that is commonly used to describe the
adsorption of gases onto a porous surface.
where V = the adsorption volume under equilibrium pressure P, cc/g; VL = the Langmuir
volume, which represents the limit gas adsorption capacity when the gas pressure is infinite,
cc/g; PL = the Langm uir pressure, which represents the pressure at which the adsorption
volume V reaches half of the VL, MPa.
3.1.1 Coal seam gas content and outburst threshold limits (TLV)
The Australian outburst TLV varied linearly based on gas composition, increasing from
a minimum in CO2 rich conditions (5 m3/t) to a maximum in CH4 rich conditions (9 m3/t), as
444
shown in Figure 6, with the presence of CO2 seam gas considered a significantly higher out
burst risk than CH4 seam gas. The Level 1 TLV for ‘normal’ mining was 9.0 m3/t in pure CH4
conditions and 5.0 m3/t in pure CO2 conditions. If gas content was not reduced below the
Level 1 TLV, mining allowed only under outburst mining procedures. The Level 2 TLV for
outburst mining was 12.0 m3/t in pure CH4 conditions and 10.0 m3/t in pure CO2 conditions.
If gas content did not fall below the Level 2 TLV, mining was only permissible using remote-
operated equipment, with all personnel staying clear of the outburst risk zone (Black 2019;
Black and Aziz 2010).
Based on the measuring results from coal mine A, the gas content of the upper A1, mid-
section A2, and lower section A3 of the same seam has been analyzed. Figure 6 demonstrates
that the CO2 composition in all three seam sections exceeds 95%. The virgin gas contents in
both A1 and A3 sections surpass the Level 2 Threshold Limit Value (TLV), while the gas con
tent in the A2 seam section falls between Level 1 and Level 2 TLV. Hence, is crucial to imple
ment gas drainage measures to reduce the gas content below the respective TLVs in each seam
section.
Figure 6. In-situ coal seam gas content of Mine A and outburst threshold limits (TLV).
445
Figure 7. Q1, Q2, Q3 and QT gas content of different seam sections of coal seam A.
446
performance in describing the sorption behaviour of CO2, with fit coefficients R2 greater than
0.99, emphasizing the model’s suitability for representing the CO2 sorption process in these
coal samples.
Analysis of the experimental and characterization results reveals variations in adsorption
and desorption capacities for CO2 among the tested coal samples from the different seam sec
tions. Notably, the desorption Langmuir volume of CO2 (de-VLCO2) is smaller than ad-VL
CO2, varying between 18.11 m3/t and 26.63 m3/t, with a considerable proportion of residual
gas content Vr (5.38-6.49 m3/t). Sample A3 has the highest Langmuir volume and a moderate
Langmuir pressure for adsorption, indicating a strong adsorption capacity, which could be
attributed to the presence of developed micropores in the coal reservoir. The highest residual
gas content and high hysteresis degree of A3 could be related to the presence of poorly con
nected pores, which may contribute to the trapping of CO2 gas.
Table 1. Model fitted parameters for CO2 ad-/desorption isotherms and its hysteresis index.
Adsorption parameters Desorption parameters
447
3.3 The relationship between in-situ gas content and gas sorption capacity
The in-situ gas content and gas sorption capacity of a coal seam are interconnected factors
that significantly impact the determination of coal seam gas reserves and the propensity for
outbursts in underground mines. In this section, we examined the relationship between in-situ
gas content and gas sorption capacity, utilizing the data obtained from the A1, A2, and A3
seam sections. So as to comprehend how variations in sorption capacity affect the in-situ gas
content and to identify factors that could impact gas management and outburst risk.
Figure 10 presents a scatter plot comparing in-situ gas content (QT) with the Langmuir
volume for the coal samples from the three seam sections. A strong positive correlation (Pear
son’s coefficient >0.855) is observed between gas content and Langmuir volume. This suggests
that higher sorption capacities correspond to higher gas content within coal seams.
Figure 10. The relationship between in-situ gas content and gas sorption capacity.
The previous data show that the A3 seam section has the highest in-situ gas content (QT) and
the highest adsorption Langmuir volume. Conversely, the A2 seam section exhibits the lowest
in-situ gas content and the lowest adsorption and desorption Langmuir volume. These results
further reinforce the positive correlation between in-situ gas content and gas sorption capacity.
The isotherm-based residual gas content (Vr) also plays a critical role in understanding the rela
tionship between gas content and sorption capacity. As mentioned earlier, the A1 and A2 seam
sections have higher residual gas content (Q3) than the A3 seam section, indicating that these sec
tions are harder to drain and exhibit lower gas drainage efficiency. This could be attributed to the
coal’s pore structure, which may influence the connectivity of pores and the trapping of gas within
the coal matrix.
In summary, he connection between gas content and gas sorption capacity is evident.
Higher sorption capacities correspond to higher in-situ gas content in the coal seams, and
variations in residual gas content can impact gas management strategies and outburst risk.
Understanding these relationships is crucial for developing effective gas drainage techniques
and ensuring safe mining operations in coal mines with high gas content.
4 CONCLUSIONS
In this study, we examined the in-situ coal seam gas components (Q1, Q2 and Q3) and corres
ponding measurement methods and associated devices, analyzed their limitations and controls.
448
Utilizing the methods, the coal seam gas content and gas composition of the A1, A2, and A3
seam sections are determined. Furthermore, the gas sorption capacity for samples taken from
the A1, A2, and A3 seams are investigated through high-pressure isothermal gas sorption
experiments. Based on the results and discussions presented, the following four key conclusions
can be drawn:
The direct desorption method, specifically the fast direct desorption method, proves to be
an effective and reliable approach to measure the in-situ coal seam gas content and gas com
position for the coal seams investigated in this study. By implementing various procedural,
measurement, and calculation controls, this method ensures accurate results with a low error.
The method can be a well-suited for assessing coal seam gas content and gas composition in
Australian coal mining industries.
The in-situ gas content and gas composition of the coal seams exhibit significant variations
among the A1, A2, and A3 seam sections. The A3 seam section shows the highest in-situ gas
content, while the A2 seam section has the lowest. The CO2 composition of all three seam
sections is higher than 95%, highlighting the importance of understanding CO2 sorption
behavior in these coal seams.
The relationship between in-situ gas content and gas sorption capacity is evident, with higher
sorption capacities corresponding to higher in-situ gas content. The A3 seam section demon
strates the strongest adsorption capacity, which could be attributed to the presence of developed
micropores in the coal reservoir. The higher residual gas content in the A1 and A2 seam sections
indicates that these sections are harder to drain and exhibit lower gas drainage efficiency, pos
sibly due to the coal’s pore structure and the trapping of gas within the coal matrix.
The analysis of gas content components (Q1, Q2, and Q3) reveals that the A3 seam section
has a higher outburst propensity than the A1 and A2 seam sections, due to its higher Q1 and
Q2 gas contents. This finding underscores the need for effective gas management strategies
and outburst prevention measures in coal mines with high-gas content and varying gas sorp
tion capacities.
These conclusions offer valuable insights into the relationship between in-situ gas content
and gas sorption capacity in coal seams, the importance of accurate measurement methods,
and the factors influencing gas management and outburst risk in coal mines. Future research
could explore the impact of different coal properties, such as pore structure, sorption hyster
esis and coal rank, on the relationship between in-situ gas content and gas sorption capacity
to optimize gas drainage techniques and ensure safe mining operations.
REFERENCES
Australia, Standards. 2016. “Determination of gas content of coal and carbonaceous material - Direct
desorption method.” In. Sydney: SAI Global Limited.
Bertard, C., B. Bruyet, and J. Gunther. 1970. ‘Determination of desorbable gas concentration of coal
(direct method)’, International Journal of Rock Mechanics and Mining Sciences & Geomechanics
Abstracts, 7: 43–65.
Black, Dennis. 2018. “Control and management of outburst in Australian underground coal mines.” In
18th Coal Operators’ Conference, edited by Naj Aziz and Bob Kininmonth, 246–63. Mining Engineer
ing, University of Wollongong.
Black, Dennis, and Naj Aziz. 2010. “Outburst threshold limits-current research outcomes.” In 2010 Coal
Operators’ Conference, edited by Naj Aziz and Bob Kininmonth. Mining Engineering, University of
Wollongong.
Black, Dennis, Naj Aziz, Matt Jurak, and Raul Florentin. 2009. ‘Gas content estimation using initial
desorption rate’.
Black, Dennis J. 2019. ‘Review of coal and gas outburst in Australian underground coal mines’,
Int J Min Sci Technol, 29: 815–24.
Diamond, William P., and Steven J. Schatzel. 1998. ‘Measuring the gas content of coal: A review’, Inter
national Journal of Coal Geology, 35: 311–31.
Faiz, M., A. Saghafi, N. Sherwood, and I. Wang. 2007. ‘The influence of petrological properties and
burial history on coal seam methane reservoir characterisation, Sydney Basin, Australia’, International
Journal of Coal Geology, 70: 193–208.
449
Hou, Xiaowei, Shimin Liu, Yanming Zhu, and Yun Yang. 2020. ‘Evaluation of gas contents for a
multi-seam deep coalbed methane reservoir and their geological controls: In situ direct method versus
indirect method’, Fuel, 265: 116917.
Kissell, Fred N., Charles M. McCulloch, and Curtis H. Elder. 1973. “The direct method of determining
methane content of coalbeds for ventilation design.” In.: US Department of Interior, Bureau of
Mines.
Lei, Yang, Yuanping Cheng, Liang Wang, Ting Ren, and Qingyi Tu 2023. ‘Mechanisms of Coal and Gas
Outburst Experiments: Implications for the Energy Principle of Natural Outbursts’, Rock Mechanics
and Rock Engineering, 56: 363–77.
Li, Zhongbei, Ting Ren, Xiangchun Li, Ming Qiao, Xiaohan Yang, Lihai Tan, and Baisheng Nie. 2023.
‘Multi-scale pore fractal characteristics of differently ranked coal and its impact on gas adsorption’,
International Journal of Mining Science and Technology.
Liu, Yanwei, Yang Du, Zhiqiang Li, Fajun Zhao, Weiqin Zuo, Jianping Wei, and Hani Mitri. 2020.
‘A rapid and accurate direct measurement method of underground coal seam gas content based on
dynamic diffusion theory’, International Journal of Mining Science and Technology, 30: 799–810.
Moore, T. A. 2012. ‘Coalbed methane: A review’, International Journal of Coal Geology, 101: 36–81.
Thakur, Pramod. 2016. Advanced reservoir and production engineering for coal bed methane (Gulf Profes
sional Publishing).
Xue, Sheng, Jun Xie, and Yu Cang Wang. 2013. ‘A Direct Measurement System to Estimate Gas Con
tent in Coal’, Applied Mechanics and Materials, 313-314: 604–07.
Xue, Sheng, and Liang Yuan. 2017. ‘The use of coal cuttings from underground boreholes to determine
gas content of coal with direct desorption method’, International Journal of Coal Geology, 174: 1–7.
Xue, Sheng, Liang Yuan, Jun Xie, and Yucang Wang. 2014. ‘Advances in gas content based on outburst
control technology in Huainan, China’, International Journal of Mining Science and Technology, 24:
385–89.
Zhang, Lei 2013. ‘Study of coal sorption characteristics and gas drainage in hard-to-drain coal seams’,
University of Wollongong.
450
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
1 INTRODUCTION
As the near surface mineral resources deplete and mineral consumption increases, the under
ground will become more extensive and deeper. Underground mining can pose several challenges
such as support, drainage, transportation, subsidence, illumination, and ventilation. The mining
method largely determines these challenges. A well-thought-out plan is necessary to manage these
challenges in an underground mining operation. One of the main aspects of underground mining
that is often overlooked is ventilation. Ventilation is required to control the level of air contamin
ants, such as dust, fumes, and gases, and to provide a comfortable working environment for
miners. In addition to maintaining air quality, ventilation can also be used to control the tempera
ture and humidity in the mine, which can affect worker comfort and safety. The ventilation
system can also play a critical role in emergency situations, such as fires or explosions, by helping
to control smoke and toxic gases and facilitating evacuation. Mine health and safety depend on
ventilation, and it requires significant capital and operating cost. Given the importance of ventila
tion in underground mining, it is essential that mining companies invest in high-quality ventilation
systems and ensure that they are properly maintained and monitored.
In mining, when the geological conditions are favorable, methods that involve large openings
are often preferred. This can provide several advantages over other techniques, including
improved access, increased safety, higher production rates and lower costs. Some of the challenges
associated with large openings could be complex airflow patterns, pressure differentials, heat and
energy loss, pockets of stagnant air and gas accumulation in enclosed areas. These potential issues
need to be studied and understood for better design, planning and health and safety of the miners.
Blasters and ventilation engineers must ensure that adequate ventilation is provided to
dilute pollutants below threshold limit values before miners can enter the working face. Trad
itionally, most underground mines estimate post-blast dilution time based on observations
and experiences, often resulting in an erroneous selection of post-blast dilution time. Few
mines use a theoretical approach; however, as they are unable to account for changes in blast
ing conditions (Tiile, 2019). Most underground mines perform blasting operations at the end
DOI: 10.1201/9781003429241-47
451
of the working shift to allow sufficient time to clear the blast fumes. Due to prolonged waiting
time, a significant production delay can occur in mines when multiple blasts are required in
a day (Gillies et al., 2004).
The waiting time or re-entry time depends on a number of factors such as the air velocity at
the inlet, vent duct diameter, and the amount of gases or fumes produced during blasting.
While there are different algorithms and models available to calculate the post-blast dilution
time in a blind heading, they lack accuracy for some conditions and geometries (Torno and
Torano, 2020). An exponential function was obtained for calculation of gas clearance after
blasting with a correct statistical adjustment by means of experimental values. Area of the
face advance, mass of explosive used in the blast, time for the air to clear, distance from the
end of the ventilation duct to the face and quantity of air (Torno and Toraño, 2020).
Airflow behavior is a complex phenomenon and depends on the geometrical configuration,
pollutant source and ventilation parameters. Computational Fluid Dynamics (CFD) can help
predict the airflow behavior, visualize the airflow patterns, and measure the performance of
the ventilating system. Gillies et al. (2004) developed assessment tools to minimize safe after
blast re-entry time in an underground metal mine in Australia. Lu (2011) studied blast fume
diffusion in eight different conditions of different drift lengths and outlet positions. The results
showed that the fume discharge time had a linear relationship with the drift length and pipe
outlet position. Stewart (2014) performed experimental investigation to evaluate the re-entry
prediction methods. Akash et al. (2023, 2022) investigated the effect of trapped fumes on the
re-entry time using CFD and developed a machine learning model to predict the dilution time.
Many researchers have successfully used the CFD techniques to model the ventilation
system and evaluate the performance of fume and dust removal (Qian et al., 2020). Purusho
tham et.al (2021) investigated the airflow and radon dispersion for block cave mining method
using CFD studies. Numerical simulation of blast fume dispersion in large openings can help
understand the distribution of fumes in the underground working areas, role of auxiliary ven
tilation on the fume dispersion, and air flow patterns.
Large-opening mine ventilation is unique for the following reasons: (1) it is challenging to
keep airflow velocities high enough to effectively remove or dilute airborne contaminants, (2)
large air volumes can be moved through the mines with little static pressure drop, and (3)
stoppings to direct ventilation airflows are costly to construct and maintain (Grau III et al.,
2006). The knowledge of the airflow patterns and pollutant dispersion in the underground
working environment is useful in designing an effective ventilation system (Krog et al., 2004).
Using CFD techniques, this study examines air flow patterns and blast fume dispersion in
large openings:
i) Sanford Underground Research Facility (SURF) cavern,
ii) large room and pillar working panel, and
iii) longhole open stope.
2 METHODOLOGY
A brief overview of the methodology is presented in this section. Three geometry models are
considered for the airflow pattern and fume dispersion study.
Blast fume is a mixture of toxic gases like carbon monoxide (CO), carbon dioxide, nitrogen
oxides, etc. Out of these gases, CO is of major concern due to the amount released in a blast and
the permissible exposure limit (PEL). The fluid domain in CFD (air) is assumed to be a mixture
of Nitrogen, Oxygen and CO. The amount of CO released from the blast is calculated using the
NIOSH study (Rowland and Mainiero, 2000) to determine the effect of ANFO fuel oil content on
CO production i.e., 1 Kilogram of ANFO produces 16 liters of CO. Since a forcing ventilation
system is a highly efficient and commonly used auxiliary ventilation system, a forcing ventilation
system was considered to study the blast fume dispersion problem in large openings.
First, the airflow patterns and blast fume dispersion from an auxiliary ventilated drift in
a regular room and pillar panel is investigated using CFD. Next, a large room and pillar panel
is investigated and compared with the regular room and pillar airflow characteristics.
452
2.1 Room and pillar
Room and pillar mining is one of the earliest and most common mining methods. Figure 1
shows the dimensions and extraction sequence of a typical room and pillar panel. In this
study, the panel is assumed to be at the extraction stage 11.
2.2 SURF
SURF is located at the site of the former Homestake Gold Mine in Lead, South Dakota. This
facility has since been repurposed and extensively modified to accommodate numerous under
ground science experiments. The site is owned and operated by the South Dakota Science and
Technology Authority (SDSTA). A global neutrino physics community is developing a leading-
edge, dual-site experiment for neutrino science and proton decay studies, known as the Deep
Underground Neutrino Experiment (DUNE). The facility required for this experiment, the Long-
Baseline Neutrino Facility (LBNF), is constructed at the Sanford Underground Research Facility
(SURF) in Lead, South Dakota. The construction project will excavate three caverns a mile
underground as shown in Figure 3. Two of the caverns will house the detector modules
453
Figure 2. Layout of the large room and pillar geometry modeled in the CFD model.
Figure 3. Caverns constructed at the SURF for the DUNE experiment (Fermilab, 2022).
The final stage of cavern development considered for this study is shown in Figure 4. At
this stage, the opening volume is at the largest.
454
Figure 4. SURF Geometry with auxiliary ventilation.
primarily used for extraction of large ore bodies with a steep dip, regular shape, and defined ore
boundary.
Development of sublevel stoping includes a haulage drift, raises to access sublevels, under
cuts below the stope, draw points, and a slot raise at the end of the stope (Figure 5). The ore
455
body is divided into parallel sublevels of typically 20-40 metres vertically apart. Once this is
completed, a slot raise is developed and subsequently increased by blasting and drilling
accesses that are developed into the ore, filled with explosives and blasted sequentially.
Boundary conditions used in the simulation of different mining methods are presented in
Table 1.
The blast fumes are released from the central room as shown in Figure 7. The other two
headings do not have any active blasting going on and hence there is no auxiliary ventilation
in those headings. The blast fume dispersion in the panel is shown at different times in the
Figure 7. The results show that the blast fumes accumulate in the headings without auxiliary
ventilation. The asymmetry in the fume dispersion is due to the auxiliary ventilation and the
negative pressure at the auxiliary duct inlet.
456
Figure 7. CO dispersion in the room and pillar panel at time 10 s, 90 s, 300 s, 720 s, 900 s, and 1200 s.
457
The velocity vectors in Figure 6 shows that auxiliary ventilation duct is not recirculating the
fumes. However, the CO dispersion plots in Figure 7 show that the duct inlet is positioned in
the region where the blast fumes could diffuse and reach higher concentrations. This demon
strates the need for identifying the appropriate location for the auxiliary ventilation system.
Figure 9. CO dispersion in the room and pillar panel at time 10 s, 60 s, 180 s, 300 s, 480 s, 900 s, 1200 s,
and 1500 s.
458
3.3 SURF
A steady state simulation of airflow and gas dispersion is performed for the SURF cavern
geometry to identify the regions of dead spaces. To visualize regions of gas accumulation in
steady state condition, high gas release rate at the working face is assumed as shown in Table 2.
The duct inlet is at high velocity (10 m/s) and the velocity inside the cavern is less than 1 m/s.
The color coding for the streamline of velocity shown in Figure 10 is customized to clearly visual
ize low velocities i.e. velocity magnitude more than 1 m/s are shown in red. Similarly for the
volume rendering of gas concentration in Figure 10, gas concentrations more than 50 ppm of CO
are shown in red. In Figure 11, the dead spaces immediately below the inlet can be clearly seen.
Figure 10. Streamlines of airflow inside the cavern. Figure 11. CO concentration in the cavern.
Figure 12. Velocity distribution in the drifts and crosscuts of the longhole stope (left); CO distribution
in the longhole stope after 20 minutes of ventilation (right).
459
3.4 Longhole open stoping
The velocity distribution in the drifts, crosscuts, and stope are shown in Figure 12. The
broken rock in the stope is modeled as a porous media hence the low velocities in the porous
media. Auxiliary ventilation in the upper and lower levels help in the dilution of the fumes in
the stope. Figure 12 shows the high CO concentration within the stope due to the fumes
trapped in the broken rock.
4 CONCLUSIONS
This study shows that CFD is a useful tool in analyzing the blast fume dispersion and airflow
patterns under various airway configurations used in different mining methods. While this
study focuses on underground mining methods, understanding of airflow pattern and pollu
tant distribution in large underground space will be beneficial to the study of the flow and
explosion of methane, mine fires, and diesel particulate matter.
The results show that the change in airway dimensions (large openings) changes the airflow
behavior and gas dispersion. The large openings are characterized by the lower velocities in the
airway and slower removal of gases in low velocity areas. The study shows that air velocity plays
a major role in the dilution of the fumes. Strategic positioning of the auxiliary duct and low-
pressure propeller fans can help in ventilating large underground openings. Because of the porous
media (broken ore), stopes take longer for dilution. The methodology and the study results will be
helpful in investigating the gas dispersion and airflow patterns of other underground bulk-mining
methods such as caving and stoping methods. The extra-large dimensions make the cavern one of
a kind and the study results will be pioneer for such large excavations in the future.
REFERENCES
Adhikari, A., Jayaraman Sridharan, S., Tukkaraja, P., Sasmito, A. & Vytla, S. (2022) The Effect of
Trapped Fumes on Clearance Time in Underground Development Blasting. Mining, Metallurgy &
Exploration, 39, 1811–1820.
Adhikari, A., Tukkaraja, P., Sridharan, S. J. & Verburg, A. (2023) Machine learning model for predicting
blast fume dilution time in underground working areas. CIM Journal, 14, 64–73.
Copco, A. (2000) Sublevel Stoping Operation Atlas Copco Group.
Fermilab (2022) DUNE at LBNF. Department Of Energy.
Gillies, A. D. S., Wu, H. & Shires, D. (2004) Development of an assessment tool to minimize safe after
blast re-entry time to improve the mining cycle. Proceedings of the Tenth US/North American Mine
Ventilation Sm. Alaska, USA, AA Balkeme publishers.
Grau Iii, R., Krog, R. & Robertson, S. (2006) Maximizing the ventilation of large-opening mines. Pro
ceedings of the 11th US/North American Mine Ventilation Symp. University Park, PA.
Kohler, J. (2022) Room and Pillar Method. PennState.
Krog, R., Grau, R., Mucho, T. & Robertson, S. (2004) Ventilation planning layouts for large opening mines.
Lü, W.-S., He, L. & Yang, P. (2011) Simulation study of blast fume diffusion characters of driving face.
Journal of Coal Science and Engineering (China), 17, 47–51.
Qian, J., Wang, J., Liu, H. & Xu, H. (2020) CFD Modeling of Ventilation and Dust Flow Behavior in
Polishing and the Design of an Innovative Wet Dust Removal System. International Journal of Envir
onmental Research and Public Health, 17, 6006.
Rowland, J. H. & Mainiero, R. (2000) Factors Affecting ANFO Fumes Production. Proceedings of the
26th Annual Conference on Explosives and Blasting Technique, Vol. 1.
Stewart, C. (2014) Practical prediction of blast fume clearance and workplace re-entry times in develop
ment headings.
Tiile, R. N. (2019) Investigating blast fume propagation, concentration and clearance in underground
mines using computational fluid dynamics (CFD). Mining Engineering. Missouri University of Science
and Technology.
Torno, S. & Torano, J. (2020) On the prediction of toxic fumes from underground blasting operations
and dilution ventilation. Conventional and numerical models. Tunnelling and Underground Space
Technology, 96.
460
Torno, S. & Toraño, J. (2020) On the prediction of toxic fumes from underground blasting operations
and dilution ventilation. Conventional and numerical models. Tunnelling and Underground Space
Technology, 96, 103194.
Tukkaraja, P., Bhargava, R. & Jayaraman Sridharan, S. (2021) Radon in Underground Mines. Mining
Technology.
461
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J.E. Fox
Ph.D., Senior Consultant, Mine Ventilation Services, a business unit of SRK Consulting, USA
ABSTRACT: One of the largest sources of methane (CH4) emissions is ventilation air
methane (VAM) from underground coal mines. Prabhu Energy Labs is developing the “Oxi
perator,” intended to be a significant step in achieving low-cost, scalable VAM destruction via
oxidation. The Oxiperator is a patented, all-metal, porous heat exchanger for oxidizing CH4.
For VAM applications, Oxiperator units will eliminate CH4 at concentrations as weak as
0.2%, all the way up to 100% (by volume in air). For CH4 concentrations 1.5% or higher, the
Oxiperator can be coupled to a gas turbine to generate power. A 40,000 CFM Oxiperator
operated out of a 40-foot shipping container can be quickly delivered and installed. Compared
to a Regenerative Thermal Oxidizer (RTO), the Oxiperator will have a much smaller foot
print, cost less, and perform better. Oxiperators VAM air always flows continuously in the
same direction, avoiding the complicated pipes and poppet valves needed for RTO flow rever
sal every few minutes. This paper discusses Oxiperator design, the benefits of high temperature
3D additive manufacturing, lab-scale/field-scale experimental results and future plans.
1 INTRODUCTION
Underground coal mine VAM accounts for roughly 40 million tons of CH4 a year globally .
Using a modest GWP (global warming potential) of 25, this is equivalent to over a gigaton of
CO2e (carbon dioxide equivalent). VAM concentration in air varies considerably from mine to
mine and hour to hour. For many mines, VAM is emitted in the range of 0.3% to 1.0% in air
(30,000 - 100,000 ppm by volume) . It is imperative that as much VAM as possible be eliminated
to meet global greenhouse gas targets. Several studies have documented the magnitude of the
issue in the US and elsewhere. These include the United States Environmental Protection Agency
(US EPA, 2021), and the International Journal of Coal Geology (C.Ö. Karacan et.al, 2011).
The only technology currently available for VAM destruction is oxidization using
a Regenerative Thermal Oxidizer (RTO). RTOs have seen limited adoption despite growing
international demand for coal VAM reduction as governments and stakeholders recognize the
damaging effect of CH4 on regional air quality and global climate change. Several papers have
been published on oxidation. The International Journal of CH4 Geology addresses RTO tech
nologies (C.Ö. Karacan et al, 2011).
RTOs are large vessels filled with ceramic shells. The ceramic shells are maintained at about
1500° to 1600° F. Every few minutes the VAM flow directions must be reversed to keep the
temperatures hot enough for oxidation to occur. The flow reversal is achieved by large poppet
valves that must operate at high temperatures. RTOs take years to install, and the flow rever
sal is challenging. They are difficult to move to other locations. By contrast, the Oxiperator
will be delivered and operated in a cargo container, ready of rapid deployment, or relocated
after new bleeder wells are brought into service.
DOI: 10.1201/9781003429241-48
462
Prabhu Energy Labs is developing the Oxiperator, a new technology for oxidizing VAM.
Oxiperators have the potential to oxidize VAM in the range of 0.3% CH4 to 1.0% CH4 and
higher, at a lower cost and smaller footprint with relatively minimal operating costs.
2 THE OXIPERATOR
The Oxiperator is the first all-metal, high-temperature porous heat exchanger for oxidation of
low-concentration (weak) CH4 emissions mixed in air, protected by two US patents. Unlike
other oxidizers, the Oxiperator uses no catalyst or ceramics. The Oxiperator is made up of
two all-metal sections: a conventional heat transfer section and a porous section. The porous
section of the heat exchanger is the key to continuous oxidation. Due to its all-metal construc
tion and modular design, the Oxiperator has the potential to take full advantage of the bene
fits of current 3D printing technology using extreme temperatures.
Figure 1 is a schematic of the Oxiperator processing a mixture of 0.3% CH4 in air, which is
a widely measured VAM concentration from many coal mine shafts. The intake air dry bulb
temperature is assumed, for this illustration, to be 60° F.
Incoming air is heated by passing through processed hot gas mixture in the traditionally
designed section of the first heat exchanger. At this point the temperature has been raised to
1500° F, at which point the CH4 is already oxidizing rapidly. As the CH4 laden intake air oxi
dizes, the first heat exchanger temperature is raised commensurately. The gas mixture then
enters the porous section where fresh, intake gas mixture mixes through pores with return hot
gas, promoting remaining gas mixture oxidation. All processed gas mixture passes thusly
through the porous section, reaching 1650° F, thereby oxidizing any remaining CH4.
The hot gas mixture returns through the non-porous side and exits the Oxiperator at 210°
F. When 0.3% CH4 in air is oxidized, it raises the temperature by 150° F. The gas mixture
exhausts the system at 210° F, which is 150° F hotter than the intake gas mixture. At higher
CH4 concentrations, the Oxiperator can generate power without using previously researched
catalysts. At 1.5% CH4 the Oxiperator could potentially generate power without a catalyst.
An experiment with a catalytic combustor was reported in Energy (S. Su et. al. 2015). Other
operators have used RTOs to produce steam and run a steam turbine for downstream energy
recuperation.
463
3 TECHNOLOGY DEVELOPMENT AND SCALE-UP
A 25 cfm Oxiperator was successfully tested in California. Plans are underway for rapid scale-
up in size. Prabhu Energy Labs is currently manufacturing a 3D printed 100 cfm Oxiperator
that will be used to oxidize CH4 slip from a lean-burn gas engine. CH4 slip from lean burn
engines refers to the fugitive fuel gas that escapes from the exhaust of lean burn engines. It is
a major concern for the oil and gas industry where tens of thousands of lean burn engines are
employed in gathering and boosting stations, all of which have CH4 slip that is about 0.3% con
centration. The work on CH4 slip will provide valuable information for VAM Oxiperators.
Upon successful testing, a 1,000 cfm Oxiperator will be manufactured for a larger lean burn
engine, then a 3,000 cfm Oxiperator for a one-megawatt lean burn engine.
The work on CH4 slip will provide valuable information for VAM Oxiperators. A modified
version of the 3,000 cfm Oxiperator will then be tested on VAM from a coal mine in the US.
The next Oxiperator will be 10,000 cfm, followed by a 40,000 cfm Oxiperator for demonstration
at a coal mine. At the commercial scale a 40,000 cfm Oxiperator system, capable of oxidizing
2,000 tons of CH4 a year, will be installed inside a 40 foot shipping container including the fan,
filter and controls. The Oxiperator system would be delivered to the mine site and commissioned
in several days, operating from the container installation using instrumentation and controls for
safety concious remote operation. Several 40,000 cfm Oxiperators could be placed around the
mine exhaust, which may be exhausting more than 300,000 cfm of airflow. As mining produc
tion changes, Oxiperators can be moved to other mine site objectives, or swapped out if neces
sary. Figure 2 is a conceptual description of an Oxiperator working at a coal mine.
4 3D PRINTING
464
Heat transfer rates using microchannels are much higher than traditional heat exchangers.
Ultra-thin (0.3mm thick) walls are 1/10th the thickness (and mass) of conventionally manu
factured units but have been shown to withstand pressures up to 80 bar (1200 psi). 3D printing
will reduce high temperature materials by more than 90% compared to conventional manufac
turing. The original Oxiperator prototype had wall thicknesses of 3 mm, with a total mass of
60 kg. Preliminary designs for an Oxiperator made with 3D printing have wall thickness of
0.3 mm, for a total mass of 4 kg. Figure 3 shows the potential order of magnitude difference
between traditional manufacture and 3D printing.
Oxiperators can be used for many other applications. They can be installed at abandoned
mines to tackle residual gas emissions. For gas concentrations 1.5% or higher, Oxiperators
can be coupled to a gas turbine to generate electricity. If there is drainage gas or other CH4
available, VAM could be sweetened to 1.5% strength, still well below explosive range, and
used to generate power.
Applications are not limited to coal mines. For example, one of the challenging issues for
natural gas distribution networks is CH4 slip from thousands of lean burn gas engines that are
used to drive compressors. The CH4 slipping into the exhaust is commonly about 0.3%. The
project is already underway and will reduce the technology risk before the scale up in size and
before the coal mine demonstrations.
465
Methane is explosive in concentrations between 5% and 15%. Coal mines must adhere to
stringent safety standards set by MSHA, the US Mine Safety and Health Administration. It is
essential to obtain prior MSHA approval of all ventilation specific safety systems before an
Oxiperator is installed at a coal mine. It is planned to keep the Oxiperator at a safe distance
from the mine exhaust, with a sufficient “safety” gap between the mine exhaust and the Oxi
perator inlet. This distance will allow some of the VAM to escape, but it is considered neces
sary to maximize safety when workers are underground. Several monitors will also sense CH4
concentrations and the Oxiperator will be rapidly isolated as concentrations approach to the
explosive range. Methane sensors would be installed at the bottom of the shaft, if possible,
providing early warning of a gas concentration rise. This early warning will allow precious
moments for protective safety action at the Oxiperator.
According to the United States Environmental Protection Agency (EPA, 2021), 1 tonne
CH4 in equivalent to 25 tonnes of CO2. Also, the EPA has recommended that coal mining
operators should be charged an empirical CH4 emission (metric tons of CH4 released per ton
of coal mined) rate multiplied by tons of coal mined per year. The suggested cost for each
metric ton of CO2 released into the open atmosphere is currently $51. The EPA has intro
duced legislation that will nearly quadruple that cost to $190 per tonne of CO2 equivalent by
the end of 2024! If one 40,000 cfm VAM Oxiperator destroys 2,000 tonnes of CH4 a year,
which would potentially save the coal mine operator US$9,500,000 annually!
REFERENCES
C. Ö. Karacan, F. A. Ruiz b, M. Cotè, S. Phipps: “Coal mine methane: A review of capture and utiliza
tion practices with benefits to mining safety and to greenhouse gas reduction”. International Journal
of Coal Geology, 2011.
S. Su et al.: “A 25 kWe low concentration methane catalytic combustion gas turbine prototype unit”.
Energy Magazine, 2015.
US EPA Report: “Ventilation Air Methane (VAM) Projects in the United States: Barriers and Potential
Opportunities”, 2021.
466
Mine heat
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
1 INTRODUCTION
2 GEOTHERMAL GRADIENT
The geothermal gradient is derived from virgin rock temperature (VRT) measurements. Such
measurements can be performed in short boreholes underground or in long exploration boreholes.
VRT measurements at the Malmberget mine showed that short boreholes were less suitable for
DOI: 10.1201/9781003429241-49
469
Figure 1. Overview of approximate site locations. Modified from OpenStreetMap (CC BY-SA 2.0).
obtaining a geothermal gradient, so only the long borehole method was continued at all other
sites. The quasi-continuous depth-temperature logs were obtained using a commercial tempera
ture logging device in a manner identical to that described in Klose, Jones and Martikainen,
2022.
2.2 Results
In general, data was processed as described by Klose, Jones and Martikainen (2022). The first
40m of borehole data was removed to eliminate surface temperature effects. The geothermal
470
gradient was then determined as a line of best fit through the remaining downhole and uphole
data. True depth corrections were applied to all measurements using data on the encountered
water levels and borehole collar elevations. Collar elevations of the Malmberget and Kiruna
boreholes refer to each mine’s reference coordinate system. This was corrected during post-
processing to instead use sea level as a reference elevation, ensuring comparability between
sites (Figure 2, expanded in Figure 3). Borehole data for surface boreholes in Svappavaara
and Per Geijer referred to the true elevation above sea level.
Figure 2. Compiled depth-temperature profiles at 40 m below collar and referenced to sea level.
Data was collected from a total of 13 boreholes on levels 600, 970 and 1250 in Malmberget.
A good spread of data was achieved despite some targeted areas no longer being accessible and
blockages in boreholes preventing full-length measurements. Three boreholes were measured from
level 600 to complement measurements on lower levels and to help establish a gradient trend.
A total of seven boreholes were successfully surveyed at Kiruna. Five of these boreholes are
located on the hanging wall side with boreholes directed towards the orebody. The remaining
two boreholes are on the foot wall and are dipping away from the orebody. Measurements
spanning up to 1000 m in borehole length were possible at nearly all selected sites and differ
ent gradients were observed for the hanging wall and foot wall, which indicates that geology,
and thus distinct thermal properties, can play a greater role than initially thought.
At Svappavaara, a total of six boreholes were surveyed, with the Leveäniemi and Gruvber
get deposits each targeted by three boreholes. Four boreholes were located outside of the
mine’s industrial area, while the two remaining boreholes were located close to the active pit
but targeting Leveäniemi and Gruvberget, respectively. Given that the exploration holes are
depth-limited according to the depth of the planned mine extension, the measurements were
limited as well and only the final sections of the boreholes were used to determine
a geothermal gradient. The start of these final sections was determined by visually identifying
the depth where a change in gradient occurred.
The same approach was used for the three boreholes surveyed in the Per Geijer exploration
area. All three boreholes are located on the hanging wall side. Poor accessibility is the main
reason for the limited number of boreholes surveyed.
471
Figure 3. Site-specific depth-temperature profiles at 40m below collar and referenced to sea level.
2.3 Discussion
Boreholes closer to the pit had significantly lower water levels than those further away. Just as
it is important in an underground mine to measure thermal gradient away from the influence
of the underground mine infrastructure, in a surface mine, it is important to intersect depths
that are below the currently active pit. Specifically, it is important to measure an area
unaffected by the cone of depression resulting from pit dewatering.
Consistencies in temperatures between different sites are noted when viewed with sea level
as the common reference elevation (Figure 2). The depth-temperature profiles of the Svappa
vaara measurements appear consistent with that of from level 600 at Malmberget. The Per
Geijer measurement profiles seem consistent with hanging wall measurements at Kiruna.
These are four different mineralizations found in the four similar, yet variable, geologies
located up to 75 km apart. Although the deposits are at different stages of mine development,
use different mining methods, and have varying depths, the overall geothermal gradients are
the same for all of them.
This indicates that on a macro-scale, geothermal gradient is quite independent of human
influence, except for the role of groundwater, and that the gradient between the different sites
is less variable than originally thought. Mines spaced so closely together can use the same
472
geothermal gradient, generally. This will be true unless there is a clear boundary condition
that changes the base-state in which the mine is developed. Examples of this might be presence
of a localized highly radioactive geology, known geothermal activity limited by a geological
boundary, or other such factors.
However, on a micro-scale, used for mine ventilation planning and simulation, the role of
geology becomes more important. On this scale, variable geothermal gradients within one
mine can exist and are influenced by geological properties, which are discussed in greater
detail in the next section.
3 THERMAL PARAMETERS
Apart from the analyses of thermal conductivities by Parasnis (1975, 1982), little information
is available on the thermal properties of the host rocks in LKAB mines. For a higher degree
of confidence and more accurate consideration of site-specific conditions, thermal properties
need to be determined. The parameters are related to one another through density as follows:
Where α is thermal diffusivity (m2 s-1), λ is thermal conductivity (W m-1 K-1), cp is specific
heat (J kg-1 K-1), and ρ is density (kg m-3).
Conductivity is typically measured using the divided bar (DB) method. Specific heat can be
measured using calorimetric methods. With the density value obtained through Archimedes’
principle, thermal diffusivity of the sample can be determined using the Equation 1. Conduct
ivity can also be measured using Thermal Conductivity Scanning (TCS). This method requires
minimal and non-destructive sample preparation for core sections. Thermal Diffusivity Scan
ning (TDS) can be obtained concurrently, which is why the method will be referred to as
“TCS/TDS”. Specific heat can be calculated once the density value is known by rearranging
Equation 1.
473
(nHW = 3; nFW = 12). The second laboratory used the DB method to analyze 20 Svappavaara
samples (nHW = 10; nFW = 10) and 15 Per Geijer samples (nHW = 5; nFW = 10). The results are
summarized as boxplots in Figure 4, with Malmberget, Svappavaara, Kiruna and Per Geijer
respectively indicated with “M”, “S”, “K” and “P”.
Figure 4. Comparison of laboratory results across all sites and methods, showing distribution of sample
a) thermal conductivity, b) thermal diffusivity, c) specific heat and d) density.
The reported standard procedure for the TCS/TDS method was for each sample to be
scanned at least two times, with the sample being rotated or shifted the second time. Mean
values were calculated for each sample. When measurements were potentially affected by
inhomogeneities such as inclusions or banding, samples were reported being split into separate
scanning sections or the scans being repeated on a different side of the specimen. Grain size
and cavities was another reported issue. Scanning biases were only reported for core and are
presumed lower for massive, sawed samples due to the more ideal surface and depth.
For the divided bar method, disks were cut from homogeneous sections of the core. At
times, the cut samples were found to not meet the method specifications. Lack of homogen
eity, smoothness, thinness, and breakage reported as the most common issues. In these cases,
new samples were cut that met the required testing criteria. It is important to note that all
disks were cut to have the same thickness, regardless of the diameter of the core.
474
The core diameter (BQ or NQ) had no noticeable influence on thermal parameters as deter
mined by the divided bar method as shown in Figure 6.
Figure 5. Comparison of results for the same Malmberget samples analyzed using the TCS/TDS and
DB methods, showing distribution of a) thermal conductivity, b) thermal diffusivity, c) specific heat and
d) density.
Figure 6. Comparison of results for different core diameters for the DB method, showing distribution
of a) thermal conductivity, b) thermal diffusivity, c) specific heat and d) density.
A direct comparison of results for different drill core diameters analyzed for identical litho
logical groups indicated that measured values of thermal conductivity and specific heat are
within ±5%. The value spread can be explained by slight variations in mineralogical compos
ition, as the samples originate from different sections and boreholes. A similar assessment of
size effects for the TCS/TDS method was not possible with such a limited sample set.
475
identified ones. Sandberg also provided a possible range of volume fractions for each mineral
identified in the thin sections. The lower and upper boundaries of each range were separated
by 10 percentage points in most cases. Only the more common minerals listed by Sandberg
(2018) were used for the purpose of calculating the thermal properties because the volume
fraction was greater than 1% for only a few of the more uncommon minerals. The properties
of the high purity mineral samples were based on data compiled by Waples and Waples
(2004), Čermák and Rybach (1982) and Boulanger-Martel, Bussière and Poirier (2018), citing
Côte and Konrad (2005) as seen in Table 1.
Amphibole
Actinolite 3200 705 3.483 1.54
Hornblende 3080 710 2.85 1.30
Amphibole
Actinolite 3200 705 3.483 1.54
Hornblende 3080 710 2.85 1.30
Anhydrite 2900 585 4.76 2.81
Biotite 3000 770 1.17 0.51
Calcite/“Carbonate” 2745 815 3.57 1.60 Calcite assumed
Chlorite 2800 600 5.14 3.06
Clinopyroxene Augite 3380 670 3.82 1.69 Augite assumed
Epidote 3400 780 2.82 1.06
Feldspar Mix of feldspar minerals assumed
Albite 2600 730 2.31 1.22
Orthoclase 2600 628 2.31 1.41
Hornblende 2580 700 2.49 1.38
Hematite 5150 620 11.3 3.54
Ilmenite 4750 849 2.38§ 0.59
Magnetite 5175 586 5.10 1.68
Pyrite 5010 510 19.2 7.51
Quartz 2648 740 7.69 3.92
Titanite 3500 810 2.33 0.82
* Waples (2004), †Cermak and Rybach (1982), §Boulanger-Martel, Bussière and Poirier (2018), citing
Côte and Konrad (2005).
Using the Sandberg (2018) data and the compiled thermal parameter values for different
minerals, average volume fractions were then calculated for each mineral with the reported
lower and upper boundaries as input. The individual volume fractions were then cumulated.
This cumulative value was used to correct each individual volume fraction so that the new
cumulation would reach exactly 100%.
Thermal conductivity was calculated using Equation 4, after information compiled by
Robertson (1988). All other properties, including density, were calculated using Equation 2
after substituting thermal conductivity with the desired property.
476
Where λ is thermal conductivity (W m-1 K-1), n is volume fraction (1), p is parallel and s is
serial mineral grain arrangement.
Two comparisons can then be performed. The first is a direct comparison between calcu
lated (“Calc.”) and measured (“Meas.”) values for four thin sections and their corresponding
core samples, seen in Figure 7. The second, indirect comparison in Figure 8 focuses on the
properties of the foot and hanging wall and includes a greater number of samples, leading to
an averaging effect.
Figure 7. Direct comparison of calculated and measured properties for four Kiruna thin sections and
their core samples, showing distribution of a) thermal conductivity, b) thermal diffusivity, c) specific heat
and d) density.
Figure 8. Comparison of TCS-measured and calculated foot wall and hanging wall properties for
Kiruna samples, showing distribution of a) thermal conductivity, b) thermal diffusivity, c) specific heat
and d) density.
3.5 Discussion
Two different methods of analysis were used to obtain thermal parameters for the selected
samples. The TCS/TDS method is applied lengthwise to a core sample while the divided bar
477
method uses a disk cut from a homogenous core section. Especially the latter method can be
considered firmly established, while core is a readily available source of information. Differ
ences in laboratory results could be explained by a size effect, with TCS/TDS results being
seen as more representative of a rock mass due to the inclusion of defects. However, this is
a cautious assessment. In general, laboratory analyses and calculated values do not reflect the
conditions of the rock mass in-situ. However, in-situ studies are potentially more costly and
time-consuming than laboratory analyses. For this reason, further study of the impact of core
diameter and sample geometry on the results is suggested. This would improve overall confi
dence in the results for simulation applications. For the divided bar method at least, core
diameter does not appear to impact results.
Average thermal conductivities highlight similarities and differences between the geologies
of each site. Conductivity is the only property that is measured directly by both the TCS/TDS
and DB methods. The results for this property show that similarities appear to exist between
samples from Malmberget, Svappavaara and the Kiruna foot wall. The Kiruna hanging wall
and Per Geijer foot wall samples are also somewhat similar, although the low number of sam
ples for the former influences this assessment. The Per Geijer hanging wall is distinct from the
other sample sets.
Meanwhile, the large range of densities for Malmberget points towards distinct properties
for a number of the sampled lithologies. This is also seen in the extended range seen for the
Svappavaara foot wall, which in terms of sample size and sampled lithologies, is similar to the
hanging wall. The otherwise consistent average densities and ranges suggests more homoge
neous geologies contained within the sample sets.
Calculations of thermal parameters seem to be influenced by several factors. The first factor
is the identification of minerals and volume fractions as well as the choice of mineral proper
ties to represent generic groupings such as amphibole and feldspar, which contribute to large
volume fractions. Figure 7 appears to show inconsistencies between the measured and calcu
lated parameters, but having only four data points for comparison, sample size is likely to
play a significant role in this case. This changes when the sample populations are increased,
leading to an averaging effect visible when comparing Figure 7 with Figure 8. The results
show that the calculation of parameters from modal values, if available, could be a valid
option. It is an example of how data that is routinely collected in a mineral processing setting
can be used to generate input parameters for mine climate models.
The calculation of thermal parameters assumes a homogeneous rock mass that has no pores
or pore fluids. The DB method, which analyses a uniform thin section may provide results
that are more similar than the TCS/TDS method, which is conducted on a full piece of rock
core. This is supported by the comparisons resulting from this work and is evidenced in Fig
ures 5 and 8. While the results seem to support this conclusion, the small sample size means
that further investigation regarding the comparison between the TCS/TDS method, the DB
method, and thermal parameter calculation is required for validation.
The VentsimTM DESIGN (Howden Group Ltd 2022) software package is used for climate
modeling of underground mines. Global settings for geothermal gradient and thermal param
eters can be overridden by applying custom rock types and localized changes in the geother
mal gradient to individual airways using VentsimTM DESIGN presets. The filtering of airways
greatly assists this process.
Two gradients were applied in Malmberget. The first of these was applied for elevations
above level 1000 and consists of the only valid measurement for this elevation range, which is
also offset from the other measurements. Due to their consistency (Figure 4), these were used
to generate a second gradient that was applied to elevations below 1000. This was seen as
a conservative measure, since part of the downcast air in the model would interact with a rock
mass with a higher VRT (Klose, Jones and Martikainen 2022).
478
Different gradients were found for the Kiruna hanging wall and foot wall, which simplifies
the application of a climate simulation. A global gradient was applied to Svappavaara and Per
Geijer as no underground infrastructure is currently present and surface borehole data was col
lected to only a limited depth. For Svappavaara, the mean was calculated from four gradients
determined from the bottom-most stable sections, while the other two were excluded due to
insufficient length. As only two of three Per Geijer measurements are without major surface
influence and show different gradients, the Kiruna hanging wall gradient was used instead. This
was deemed appropriate given the proximity of the deposit to Kiruna, the overall gradient
macrotrend as shown in Figure 3, and similar thermal conductivities as shown in Figure 4.
A rock surface temperature was calculated as a model input as a final step in gradient
implementation. This is used as an input to calculate VRT. The rock surface temperature is
hypothetical and can be interpreted as the intersection of the geothermal gradient line with the
mine grid surface datum. The VRTs calculated from these inputs should coincide with actual
measurements. This is applicable to Malmberget and Kiruna. While the hanging wall gradient
and surface rock temperature of the Kiruna hanging wall was applied for Per Geijer, the aver
age surface rock temperature referring to sea level was implemented for Svappavaara.
After the implementation of the gradients for each site, thermal parameters were applied.
For Malmberget, general thermal properties were calculated based on logged meters available
in a core logging database using Equation 5 (Klose 2022).
Where P is the rock property to be calculated, xi is the fraction (1) of lithology i out of the
n main lithologies, pi is the property of lithology i, and p̄ is the arithmetic average of the prop
erty for all lithologies. The units of P, pi and p̄ are identical and depend on the property to be
calculated.
The same approach and database were used to determine more individual thermal proper
ties for two target areas of the mine expansion, Area 1 and Area 2, by filtering the database
accordingly. These properties were applied to airways located in the specific area only, while
all other airways were assigned the general values. For Malmberget, the results of the reana
lyzed samples were included.
In the case of the Kiruna mine, nearly all infrastructure is located in the foot wall. The clear
geological distinction from the hanging wall greatly simplified the assignment of thermal prop
erties to airways, as the hanging wall could essentially be excluded from further consideration.
For the hypothetical Svappavaara and Per Geijer operations, average foot wall and hanging
wall properties appeared adequate at this early stage of evaluation.
The properties implemented in each climate simulation model are summarized in Table 2
below.
479
4.1 Case study: Malmberget mine
The impact of implementing new parameters was evaluated through a case study, for which
the most recently updated VentsimTM model for Malmberget mine was selected. The primary
focus of this section is to highlight the approach taken towards this case study as modelling
work and calibration is ongoing.
The Malmberget ventilation system features several ventilation shafts and a total airflow
capacity of 2300 m3 s-1. Approximately 18% of this capacity is currently directed from two
shafts (Shaft 1 and Shaft 2) towards Area 2. Area 1 is supplied by its own shaft, which has
a capacity of 540 m3 s-1.
Before implementing the parameters, the model settings were evaluated. Particular attention
was paid to the mine grid elevation, which acts as reference elevation to parameters relevant
to the implementation of the geothermal gradient as well as the input surface conditions
(Howden Group Ltd n.d.). In the original model, this elevation was set to the approximate
elevation of the nearest active weather station operated by the Institute of Meteorology and
Hydrology of Sweden (Sveriges meteorologiska och hydrologiska institut, SMHI). To allow
appropriate comparisons to be made, the mine grid and surface datum rock temperatures
were adjusted to the Shaft 1 collar elevation. The virtually identical VRT calculated for differ
ent airways indicated the success of the mine grid adjustment.
A dataset from summer 2022 was used in this case study. Near-shaft dry bulb temperature
DBT and wet bulb temperatures WBT (°C) and barometric pressures BP (kPa) were collected
using a Kestrel 5500 weather meter in logging mode. Data was logged every 20 minutes (Shaft 1)
and 30 minutes (Shaft 2). An identical instrument was used to perform underground measure
ments within the same timeframe at locations shown in Figure 9. The measurements of the dataset
focused on airflow directly from the local shaft connections. Considering the depth and compli
cated ventilation route, it is likely that DBT (°C) and WBT (°C) are over- and underestimated,
respectively, compared to the airway values. No mining or development activity was observed in
the area during the measurements and all measurements were performed the same way.
As the times of surface and underground measurements did not align directly, surface condition
inputs were calculated using weighted averages by assigning underground measurements to the
closest surface measurement time. Weighted averages were calculated for DBT (°C), WBT (°C)
and BP (kPa), which were then used as model inputs. This was deemed acceptable because of the
Figure 9. Partial overview of Area 2 in VentsimTM, showing locations of underground spot measurement.
480
steady surface conditions during the measurement period, the depth of the underground measure
ment locations and the thermal flywheel effect contributing to the stability of the underground
climate. The calculated values for surface conditions at Shaft 1 were entered into the model with
the surface temperature adjustment setting activated. The success of this approach indicated by
simulated inlet temperatures at Shaft 2 being in general agreement with actual measurements at
that location.
Heat simulation results for dry bulb and wet bulb temperatures (°C) were compared following
the adjustment of input parameters, which is summarized in Table 3. The comparisons covered the
adjustment of rock thermal parameters and gradient individually and in combination. Only airways
with airflow on levels within the depth range of performed geothermal gradient measurements were
part of the assessment. Simulation results were cross-checked with underground spot measurements
taken between levels 1000 and 1250 in Area 2. The expectation was an improved agreement
between actual measurements and simulated values at these locations. Measured and simulated
values would be identical if the absolute difference between them were 0°C, whereas a positive dif
ference would indicate that the simulated temperatures are higher than those measured.
Table 3. VentsimTM parameters used in the original and adjusted models, with reference to values used
during the parameter implementation.
Adjusted
Global Parameter Unit Original Adjusted (parameter implementation)
The simulation results (Figure 10) show that the rock thermal parameters themselves affect nei
ther DBT (°C) nor WBT (°C) beyond 0.1°C. Likely explanations for this behavior are similarities
in certain values such as thermal conductivity, as well as the applied airway age reducing the heat
flow into the opening due to cooling of the surrounding rock. This was not investigated further.
On the other hand, changing the geothermal gradient generally decreases both DBT (°C) and
WBT (°C). Where the simulation approaches the measured DBT (°C) values, such as at Site 3 and
Site 5, WBT (°C) generally increases. This could be explained by increased evaporation rates.
In general, the temperature changes can be explained by changes in simulated VRT. To
illustrate this, the surface rock temperature and presumedly linear gradient for the adjusted
model result in a theoretical VRT of 24.9°C. On the other hand, extending the other gradients
down to the current main level of 1250 leads to the respective theoretical VRT values of 15.4°C
using the gradient “Above 1000” and 14.1°C for the other. Changing the gradient therefore has
consequences for the heat transfer between the air and the rock mass from the moment air
enters the mine. This indicates that closer scrutiny of near-surface rock mass temperatures is
necessary, possibly requiring a seasonal adjustment.
Finally, some improvement of temperature values can be observed for Sites 1, 2 and 4
when both thermal parameters and gradients are implemented. This emphasizes not only the
importance of the gradient as a key influencing value, but also that the implementing of
mine-specific parameters and gradients should not be considered separately.
481
Anomalies such as DBT (°C) being lowered significantly such as at Site 2 and Site 4 when
applying the gradient only, or the greatly increased WBT (°C) at Site 2 compared to the actual
measurements, require further investigation and suggest the need for recalibrations following
the implementation of new parameters.
Figure 10. Difference in simulated a) DBT (°C) and b) WBT (°C) relative to measured values.
Generalized thermal gradient values taken from other nearby operations are suitable for use
during the first stages of ventilation planning for a new mine with no pre-existing data, as
long as extenuating circumstances are not present. For existing operations with experience,
data, and a need for more precise local modeling for expansion purposes, geothermal gradient
measurements should be conducted in boreholes intersecting rock where future infrastructure
is expected to be located. This ensures the highest relevancy of geothermal gradients and sim
plifies application in climate models. Thermal gradients determined from surface borehole
data can be used to infer a temperature at a mine-relevant depth, although the extrapolation
of the gradient is affected by uncertainty in measurements. As such, measurements should go
as deep as possible and data collection near active open pits should be avoided, unless meas
urements beyond the extent of the pit and the cone of depression can be guaranteed.
The collection of thermal parameters is a complex endeavor, as illustrated by the case studies
of the two existing and two potential underground mines. Knowledge of the present and
482
potential future mining situations can help in obtaining relevant parameters. Use of core sam
ples to obtain thermal parameters for individual rock types allows the use of large sample col
lections and later manual adjusting of possible rock mass properties. This is advantageous for
large mines with complex geologies.
The choice of laboratory method to obtain simulation parameters matters less than obtaining
representative samples of the main rock types of the rock mass in which infrastructure will be
placed. Most borehole meters originate from exploration boreholes that are drilled from the sur
face or from underground exploration tunnels towards the orebody. Borehole location and direc
tion need be considered during sample selection, as the approach of identifying the most common
lithologies through drilled meters alone can introduce a bias towards hanging wall lithologies.
Using a mine climate simulation software allows the assignment of multiple gradients and
rock types in the same mine. As the software is a tool that can be used to predict underground
conditions, it is important to know the software capabilities and limitations. These would
guide the application of the available data and can guide parameter collection efforts as well.
Implementing thermal parameters is potentially more complex than geothermal gradients
and largely depends on the level of detail required in the model. However, the Malmberget
case study shows that neither thermal parameters nor gradient should be viewed in isolation.
Ultimately, replacing placeholder thermal parameters and gradients with site-specific ones
can improve reliability and accuracy of climate simulation results for strategic planning pur
poses. This is of particular value for potential greenfield sites such as Svappavaara and Per
Geijer. Existing sites such as Malmberget and Kiruna benefit first and foremost from the
transparency and understanding that comes with the updating process and require subsequent
model calibration checks to use effectively.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the assistance from LKAB personnel, particularly from the
R&D and Exploration departments, in realizing this work. The authors would also like to extend
their thanks to the anonymous reviewers for their valuable suggestions and improvements.
REFERENCES
Boulanger-Martel, V., Poirier, A., Côté, J. & Bussière, B. 2018. Thermal conductivity of Meadowbank’s
mine waste rocks and tailings. In GeoEdmonton 2018 – 71st Canadian Geotechnical Conference, Edmon
ton, 23-26 September 2018. 8pp.
Čermák, V. & Rybach, L. 1982. Thermal properties: Thermal conductivity and specific heat of minerals and
rocks. In G. Angenheister (Ed.), Landolt-Börnstein Zahlenwerte und Funktionen aus Naturwissenschaften
und Technik, Neue Serie, Physikalische Eigenschaften der Gesteine. Berlin, Heidelberg, New York:Springer.
Howden Group Ltd 2022. VentsimTM DESIGN (Version 5.4) [Computer software]. Retrieved from
https://ventsim.com/.
Howden Group Ltd n.d. VentsimTM DESIGN User Guide Version 5.4. Retrieved from https://ventsim.
com/files/VentsimManual.pdf.
Klose, F.K.R. 2022. Heating and cooling in LKAB underground mines. MS thesis. University of
Leoben, Austria.
Klose, F.K.R., Jones, T.H. & Martikainen, A.L. 2022. Geothermal gradient determination for ventila
tion and air conditioning modelling at Malmberget mine. In Proceedings of the Australian Mine Venti
lation Conference, Gold Coast, 10-12 October 2022: 66–80.
Parasnis, D. S. 1975. Temperature phenomena and heat flow estimates in two precambrian ore-bearing
areas in north Sweden. Geophysical Journal International 43(2): 531–554.
Parasnis, D. S. 1982. Geothermal flow and phenomena in two Swedish localities north of the arctic circle.
Geophysical Journal International 71(3): 545–554.
Robertson, E.C. 1988. Thermal Properties of Rocks. Open-File Report 88–441.
Sandberg, H. (2018): A petrological investigation of the host rocks for the Kuj-Kiirunavaara ore. BS
thesis. University of Uppsala, Sweden.
Waples, D.W. & Waples, J.S. 2004. A review and evaluation of specific heat capacities of rocks, minerals,
and subsurface fluids. Part 1: Minerals and nonporous rocks. Natural Resources Research 13(2): 97–122.
483
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
J.A. Buaba
South Dakota School of Mines and Technology, Mining Engineering and Management, Rapid City,
South Dakota, USA
A.J. Brickey
South Dakota School of Mines and Technology, Mining Engineering and Management, Rapid City,
South Dakota, USA
1 INTRODUCTION
Heat build-up in underground mines, i.e., mines below the surface, is a safety concern for mine
operators because it may impact a miner’s health. The human body’s ability to cool down is con
trolled by air temperature and humidity levels in an environment, i.e., thermoregulation; high air
temperature and humidity levels increase the potential for heat-related illnesses such as heat faint
ing, heat exhaustion, heat rash, and heat stroke (Ryan & Euler, 2017). Further studies show
a negative correlation between heat stress (total heat load placed on the body) and cognitive per
formance which may cause workplace safety incidents and affect worker productivity (Yeoman
et al., 2022). Heat is an innate part of all mining activities in both surface and underground mines.
Mining activities such as drilling, blasting, loading, and hauling use mechanized equipment, or by
the nature of the activity, generate heat which is transferred to the mine atmosphere. In surface
mines, the working areas are exposed to the surface atmosphere, and the heat generated is dis
persed into the environment whereas in underground mines, a ventilation system is required to
control and manage the mine environment in order to create a safe working condition for the
miners. The Mine Safety and Health Administration (MSHA) defines a “hot” work site as
a combination of high air temperatures, humidity, radiation, and relatively low airspeed that
exceeds a wet bulb globe temperature of 26°C (79°F) (U.S. Department of Labor, 2012). This
temperature limit captures the convective, evaporative, and radiative heat exchange around the
work site. Apart from the heat generated by mechanized equipment, heat from the surrounding
DOI: 10.1201/9781003429241-50
484
rocks (strata heat) and autocompression, i.e., the rise in the temperature of air due to the increase
in pressure at lower elevations, increase the ambient air temperature in underground mines. Heat
is transferred between and/or within mediums through conduction, convection, and radiation in
the presence of a thermal gradient (McPherson, 1993a). In underground mines, the airflow is sup
plied using large primary fans that could be on the surface or underground, and the air is circu
lated through the mine drifts, i.e., the airways, to remove heat via convection. Booster and
auxiliary fans in conjunction with ventilation controls such as ventilation doors, stoppings, and
regulators are also used to distribute the airflow to the working areas. As the air moves from the
surface to the lower levels, some of its potential energy is converted to heat which increases the air
temperature (autocompression). The air temperature can also increase due to the heat emitted by
the strata and mechanized equipment via thermal convection and radiation as it circulates through
the mine drifts. The wet bulb temperature of the air, i.e., the temperature at which air can be
cooled by evaporation, increases at the rate of about 4°C (7.2°F) per 1000 m vertical descent
(Ramsden, et al. 2007). The maximum working temperature for most mining operations ranges
from 27.5°C (81.5°F) to 28.5°C (83.3°F) wet bulb temperature (Ramsden, et al. 2007). The wet
bulb temperature limit used in this study is 28 °C. This is based on the regulations of the host
country for the case study mine. Heat transfer from the strata to the air and vice versa varies by
the virgin rock temperature present in the mine. The heat generated from the equipment depends
on the rated engine capacity, fuel consumption, and the percentage of time it is utilized (van der
Walt, et al. 1983). Heat in underground mines may be reduced by (i) engineering controls, such as
mine planning, ventilation, and air conditioning, (ii) administrative controls and work practices
such as developing a work-rest routine, rotating personnel on hot jobs, no idling of equipment,
and (iii) providing personal protective clothing and equipment such as reflective clothing to stop
the body from absorbing radiant heat (U.S. Department of Labor, 2012). As mines become
deeper, the problems of heat become more critical, requiring an air conditioning system to control
the air temperature and humidity levels in the mine environment.
In this paper, we evaluate engineering controls, i.e., mine planning, ventilation, air condi
tioning, and optimization to understand heat emissions and the impact on production in
underground mines. Temperature levels associated with heat generated from equipment and
other sources are modeled and incorporated into a medium-term production scheduling opti
mization model developed by Ogunmodede et al. (2022). The scheduling model maximizes the
net present value while adhering to precedence, resource, and heat constraints. The resulting
schedule considers temperatures by production level based on ventilation capacity, production
rates, and equipment usage. We evaluate the usefulness of this model by analyzing various
production and ventilation scenarios to determine if heat can be managed via production
scheduling or if modifications to the ventilation system is needed.
1.1 Background
Underground mine planning is an iterative process to determine the economic viability of
a mineral deposit. The planning process consists of resource modeling, mine design, production
scheduling, ventilation planning, equipment selection, infrastructure design, reconciliation, and
financial evaluation which continues throughout the life of the mine operation (Bullock, 2011).
The information needed to develop the mine plans is typically based on historical operational
data and/or engineering estimates. As new information becomes available, the mine plans are
updated accordingly. One outcome of mine planning is a production schedule that considers the
time value of money to verify the economic viability of the mineral deposit. The production
schedule determines the start time and extraction sequence of the mineral deposit. The schedule
is subject to precedence constraints and resource constraints such as production (mining equip
ment) and processing (crushing, milling) capacities. This process can be used to generate strategic
(long-term), tactical (medium-term), and operational (short-term) production schedules (Hall,
2017). Long-term production schedules define the overall business objective for the mining oper
ation, e.g., capital projects and life-of-mine production based on corporate goals. Medium-term
production schedules guide the operation to achieve the long-term goals up to a five-year time
horizon at monthly or quarterly fidelity. Short-term production schedules define the day-to-day
485
operation of the mine and focus on achieving the targets set in the medium-term schedule. The
time horizon typically ranges from weekly up to two years at hourly, by shift, or daily fidelity.
This integrated approach to mine planning requires the use of complex tools such as operations
research techniques to make informed decisions to maximize the value of the mineral deposit.
Operations Research is a scientific approach using mathematical models to represent
a problem and analyzing the solution to make better decisions (Rardin, 2017). The process
begins with formulating or modeling the problem as close to reality as possible. The decision
variables to be determined are defined together with the parameters, objective function, and
constraints. The parameters are a set of known information about the problem, the objective
function maximizes or minimizes a function of decision variables, and the constraints set
a limit on the solution. Various algorithms are used to solve the model and the results are
analyzed to make meaningful conclusions that can be applied. The common types of models
used in operations research and applied in underground mining include linear programming
(LP), integer programming (IP), mixed integer programming (MIP), networks, non-linear pro
gramming (NLP), constraint programming, and stochastic programming.
2 LITERATURE REVIEW
Operations research techniques were first applied in underground mine production scheduling to
determine a feasible schedule to meet the annual contract agreement for a copper mine (Williams
et al. 1973). The authors formulated a linear programming model to produce a “good” schedule;
however, it was difficult to define the qualities of a “good” schedule. They proceeded to develop
a tool that allowed a mine planner to evaluate multiple scenarios and select the “best” option.
Though optimality was not achieved, the resulting schedule showed significant improvement
over manual scheduling. Over the past few decades, improved computational capability, new
solution algorithms, and mathematical modeling techniques have allowed researchers to develop
integer programs that incorporates binary variables and allow for “yes-no” decision-making.
Trout (1995) formulated a mixed integer programming (MIP) multiperiod production scheduling
model for a sublevel stoping copper operation located in Mt Isa, Australia. The objective of the
model was to maximize net present value (NPV) by scheduling stopes for production over a two-
year horizon. The solution could not solve to optimality due to insufficient memory; however,
the model improved the NPV by 23%. Trout’s model was improved by Nehring and Topal
(2007) who formulated an additional constraint function to limit multiple fill mass exposures
without violating other constraints. This made the model more applicable to the sublevel stoping
method. The model was tested on a conceptual mine stoping operation. Their model was then
compared to a manual schedule which showed a 0.66% increase in NPV. Carlyle and Eaves
(2001) used an integer programming model to plan a production schedule for a sub-level stoping
operation at Stillwater Mining Company for a 10-quarter planning period, to maximize revenue
from mining platinum and palladium. The flexibility in their formulation allowed engineers to
investigate multiple production scenarios quickly. McIsaac (2005) used an MIP model to gener
ate a long-term mine plan for a polymetallic mine to maximize NPV over four years at quarterly
fidelity. The model solved for 30 minutes; however, no mention is made if the solution obtained
was optimal. Little et al., (2008) revised Nehring’s and Topal (2007) model by reducing the
number of binary decision variables, which significantly improved solution time by 92% without
altering the solution. Smith et al. (2003) formulated an MIP model to produce a life-of-mine
schedule at Mt Isa. Their solution showed continuity of production and improvement in NPV
during the mine’s life when compared to the traditional method of planning using spreadsheets
and other commercial mine schedulers. Kuchta et al. (2003) implemented an MIP model to
schedule activities at LKAB’s Kiruna Mine in Sweden. The model minimizes deviation from
monthly planned production quantities while adhering to operational constraints. Other opti
mization models implemented at LKAB used aggregation and decomposition heuristics to
improve model tractability (Kuchta et al. 2004; Martinez & Newman, 2011; Newman & Kuchta,
2007). O’Sullivan & Newman (2014) scheduled the underground Lisheen Mine, located in Ire
land, with an MIP model. The objective maximizes discounted metal (zinc and lead) with
486
complex precedence constraints. Aggregation techniques were used to reduce variables and devel
oped methods for assigning precedence to the various activities; however, aggregation was limited
because of the complicated precedence structures. Lisheen Mine utilized the model to make end-
of-mine-life production scheduling decisions.
In recent years, several researchers have made attempts to standardize the underground mine pro
duction scheduling problem as a Resource Constrained Project Scheduling Problem (RCPSP)
(Brickey, 2015; Fava et al. 2012; Maybee, 2011). The RCPSP optimization model consists of an
objective function that maximizes NPV or minimizes cost, precedence constraints, and resource con
straints. The decision variables can be binary, i.e., either 0 or 1, which determine when to start
a given activity. A novel algorithm for solving the linear programming (LP) relaxation of the
RCPSP problem has allowed researchers to model and solve more complex mine production sched
uling problems (Bienstock & Zuckerberg, 2010). Terblanche and Bley (2015) formulated an under
ground mine scheduling optimization problem as an MIP using the generic RCPSP structure. Their
formulation applies to selective mining where ore grade is highly variable. Extracting only the high-
grade areas in the mine reduces cost and improves mine profitability. Brickey (2015) formulated
a generalized underground mine production scheduling model using the RCPSP structure to maxi
mize discounted gold ounces mined and determines the optimal sequence of activities in relation to
development, extraction, and backfilling of an underground mine with ventilation as a consumable
resource. The model formulation is flexible in that constraints can be customized and used in other
underground mines with similarly structured data. The model was solved using the OMP Solver
(Rivera et al. 2016) which utilizes the Bienstock-Zuckerberg algorithm (Bienstock & Zuckerberg,
2010) and a TopoSort heuristic (Chicoisne et al. 2012) to find the global optimal solution. Lopes
(2017) formulated a binary integer programming mine production scheduling model for a large-scale
underground hard-rock mine at daily fidelity to maximize net present value (NPV) using the model
structure developed by Brickey (2015). The author determined solutions for the model using the
OMP Solver and the sliding window heuristics to expedite solutions for scheduling problems. Huang
et al. (2019) developed a mixed integer linear programming (MILP) production scheduling model
for an underground cut-and-fill mine. The objective function maximizes NPV subject to mining and
processing constraints. The model was applied to two case study mines and the results showed a 9%
to 17% increase in NPV, respectively, when compared with the manual schedule. Brickey et al.
(2020) applied the RCPSP production scheduling model to determine a medium-term production
schedule for the Barrick Turquoise Ridge mine. Ogunmodede et al. (2022) formulated an MIP using
the RCPSP structure, to maximize NPV subject to precedence and capacity constraints with consid
eration of heat generated by equipment, the strata, and autocompression. The authors generated
more realistic schedules and indicated when refrigeration was activated to mitigate heat emissions.
We extend the work of Ogunmodede et al. (2022) by analyzing various production and ventilation
scenarios to make better investment decisions and describe their usefulness to mine planning engin
eers. More detail on the application of operations research in underground mine planning can be
found in Newman et al., (2010), Acuña and Lowndes (2014), and Chowdu et al. (2020).
3 METHODOLOGY
We apply the mathematical model developed by Ogunmodede et al. (2022) to a case study mine,
herein referred to as Mine X, to evaluate the impact of heat on the production schedule. Ogun
modede’s model, i.e., model (R+), quantifies heat from three main sources in underground mines
using thermodynamic principles. These heat sources— (i) diesel equipment, (ii) autocompression,
and (iii) strata-rock— are modeled given the following assumptions: (i) the medium-term produc
tion schedule with heat constraints is solved at steady state, meaning that heat from the previous
time period is not carried over to the current time period; (ii) the air velocity is constant, assum
ing there are auxiliary fans on each level, (iii) the mass flow rate, ṁ , is constant on each mining
level; (iv) the relative humidity is constant throughout the mine; and, (v) sensible heat from
equipment is absorbed into the air via convection on any given mine level.
The heat from diesel equipment is quantified using GT-SUITE software (Gamma Technolo
gies, 2021), a one-dimensional modeling software that utilizes unsteady nonlinear Navier-Stokes
487
equations based on engine specifications and ambient conditions on a mine level. The result from
GT-SUITE is the heat emitted from a given piece of equipment in a specific cycle stage, i.e., trav
eling up or down a ramp, loading or idling, and/or traveling on the surface. The total heat from
the equipment is determined by the proportion of time the equipment operates in a specific cycle
stage. Heat from autocompression is determined using Equation 1 (McPherson, 1993b).
where:
Tfinal = final air temperature on a mine level, °C
Tinitial = initial air temperature from surface, °C
g = acceleration due to gravity, m/s2
z1 = surface elevation of the mine, m
z2 = mine level elevation, m
cp = specific heat capacity of air, J/kg °C
Strata heat from newly excavated airways is quantified using Equation 2 (Mutmansky et al.
2012).
where:
l = length of airway, m
p = perimeter, m
k = thermal rock property, W/m °C
r = hydraulic radius, m
Tv = virgin rock temperature on a mine level, °C
Tair = air temperature on a mine level, °C
ω = Goch-Patterson coefficient on a mine level, unitless
All the heat sources are quantified a priori and included as parameters in the production sched
uling model. The medium-term production scheduling model used in this study is formulated as
an RCPSP optimization model. We present the generalized mathematical formulation of an
RCPSP in Appendix A. We ask our audience to refer to the detailed formulation in Ogunmodede
et al. (2022). The RCPSP model is solved using the Bienstock-Zuckerberg algorithm (BZ) (Bien
stock & Zuckerberg, 2010) and the TopoSort heuristic (Chicoisne et al., 2012). The BZ algorithm
provides continuous solutions to the RCPSP model by relaxing it as a linear programming
model. These continuous solutions are then converted into integer-feasible solutions using the
TopoSort heuristic. The open mine planner (OMP) solver (Rivera, Goycoolea, Moreno, & Espi
noza, 2021) is used to determine solutions to the RCPSP model by executing the BZ and Topo
Sort algorithms. The OMP solver process begins by exporting the mine schedule from Deswik
(Deswik Mining Consultants Pty Ltd, 2022) as a CSV file to a python script to extract the time
periods, precedence, and resource constraints into OMP files. The OMP solver outputs the time
periods for each activity, the quality of the solution, and the constraints which were violated.
This information is imported back into Deswik® for analysis.
4 CASE STUDY
488
a decline and several raises. Total airflow is approximately 210 m3/s (445,000 cfm). We obtain
the medium-term and ventilation plan for one of the active mines as a Deswik CAD and
scheduler, and Ventsim™ file format, respectively. Activities in the medium-term plan from
the mine are scheduled at a daily fidelity over a period of 5 years. From this, we obtain infor
mation on the types of activities, their duration, resource assignment, and precedence struc
ture. Table 1 shows the resource and equipment capacities used in the model. The operation
would like to optimize their ventilation resource to create a safe working environment for
miners, most especially heat, while optimizing production.
Using the data from the case study mine, we solve and analyze two production scheduling
models. The first model, defined here as base model, is constrained by the precedence structure
and the resource capacities outlined in Table 1. The heat generated from the activities is uncon
strained in this model. The second model, defined as heat-constrained model, includes the con
straints defined in the base model with additional constraints to limit the heat generated from
activities per level. The heat constraints ensure that the wet bulb temperature on a mine level is
below the allowable limit. We use these models to analyze the following scenarios in this study:
1. Normal operating conditions
2. Increase in equipment heat load
3. Decrease in ventilation
Under normal operating conditions, the initial temperature, relative humidity, and airflow
per level are 18.9 °C (66°F), 76.5%, and 52.5 m3/s respectively. We assume that four levels are
active daily, and the airflow is evenly distributed to the levels. We vary the equipment heat
load and ventilation in the other two scenarios to observe how changes in these parameters
affect the production schedule. The following sections present the results for each scenario.
5.1 Base model versus heat-constrained model under normal operating conditions
The result for this scenario using the base model and heat-constrained model is shown in
Figure 1. The graph shows the annual ore tonnes produced and development meters completed.
The base model ramps up production in the early years of the mine life. Production peaks in
2023 and gradually declines towards the end of the mine life. There is a decline in development
meters in the same year but gradually peaks in 2025. The objective function of the base model
maximizes NPV hence, it schedules development activities just in time for ore production. The
schedule from the heat-constrained model provides ore tonnes at a slower rate and gradually
peaks later in the mine life in year 2025. For development meters, the heat-constrained model
completes development activities faster than the base model. This is because of the lower heat
load from development activities compared to stope extraction and backfill activities. The equip
ment used for development and production drilling are mostly electric powered. Given the lower
heat load from development and production drilling activities, the heat-constrained model
489
schedules much of these activities at the lower levels where temperatures are high due to auto
compression and strata heat, while adhering to precedence and resource constraints. The heat-
constrained model optimizes NPV by scheduling stope extraction activities on the upper levels
until stopes are available to mine on the lower levels. The heat constraints included in the model
ensures that the total heat load from activities do not exceed the allowable limit for each level in
the mine. Figure 2 shows the wet bulb temperature on the active mine levels for a 6-month
period in 2025 where we first observe a temperature violation. The solution obtained for the
base model violates the maximum allowable temperature, i.e., 28 °C (82.4°F) wet bulb, while the
heat-constrained model adheres to the heat limits.
Figure 1. Annual ore production and development meters using the base model and heat-constrained
model.
Figure 2. Wet bulb temperature on an active mine level under normal operating conditions using the base
model.
5.2 Base model versus heat-constrained model with an increase in equipment heat load
The equipment heat load in this scenario is raised at 10% increments under normal operating
conditions, i.e., 18.9 °C (66°F) initial temperature, 76.5% relative humidity, and 52.5 m3/s
490
airflow per level. This modification to the parameters reflects an underestimation of the equip
ment heat load used in the previous scenario (normal operating conditions). The production
schedules generated in this scenario are similar to the previous scenario hence we focus our
attention on the heat generated from these activities. We observe an increase in the number of
active mine levels that violate the allowable temperature limit as the equipment heat load
increases using the base model. Figure 3 shows the maximum and average wet bulb temperature
for the 50% increase in equipment heat load using the base model for a 6-month period in 2023.
We observe that Levels 400 and 425 violate the allowable temperature limit multiple times.
Data from the production schedule show that several stope extraction, backfill, and develop
ment activities are happening on these levels. Information from the temperature helps the mine
planner to identify mining areas that are likely to violate the allowable temperature limit and
adopt strategies to mitigate high temperatures on the level. Figure 4 shows that the wet bulb
temperature for the same period using the heat-constrained model is below the allowable limit.
Figure 3. Wet bulb temperature on active mine level for a 50% increase in equipment heat load using
the base model.
Figure 4. Wet bulb temperature on active mine level for a 50% increase in equipment heat load using
the heat-constrained model.
491
5.3 Base model versus heat-constrained model with a decrease in ventilation scenario
In this scenario, we maintain all the parameters used under normal operating conditions, i.e.,
the equipment heat load, initial temperature, and relative humidity, but vary the airflow to the
mine levels. This scenario simulates situations where damage to the ventilation fan may dis
rupt production. Similar to the scenario discussed in Section 5.2, we observe an increase in the
number of mine levels that violate the allowable temperature as the airflow is reduced at 10%
increments. Figure 5 shows the wet bulb temperature for a 50% decrease in ventilation. We
observe that the wet bulb temperature per level increases as high as 45°C (113°F) and multiple
levels violate the allowable temperature limit in the same period. This implies that the ensuing
production schedule is unachievable. This information helps the mine planner to make
informed decisions such as reducing the number of active levels, and focusing on high priority
areas, i.e., areas that contribute significantly to NPV.
Figure 5. Wet bulb temperature on active mine level for 50% decrease ventilation using the base model.
In this study, we model heat associated with auto compression, geothermal gradient, and
mining equipment, and incorporate it into a medium-term production schedule developed by
Ogunmodede et al. (2022). We define two models: the base model, where the heat generated
from activities are unconstrained, and the heat-constrained model. Using data from a case
study mine, we demonstrate the strength of this model by simulating scenarios including an
increase in equipment heat load and a decrease in ventilation that could disrupt production.
The result show that the heat-constrained model generates operationally feasible schedules that
manage heat within a mine. The heat-constrained model schedules activities such that the heat
generated is below the allowable wet bulb temperature limit.
As underground mine extraction depths increase, the natural sources of heat, i.e., autocom
pression and strata can increase to uncontrollable levels. The controllable factor is to manage
the heat from mining activities. The heat-constrained model, evaluated in this study, can serve
as a tool for mine planners to evaluate both the production schedule and the environmental
conditions associated with the schedule to ensure safety while efficiently managing produc
tion. Also, the model can help mine planners perform further analysis on the impact various
disruptions could have on the schedule. Furthermore, this model can be used for financial
analysis in the feasibility stage of a mining project to make better investment decisions. Future
research would incorporate other contaminants such as diesel particulate matter and respir
able dust to evaluate their impact on the production schedule and ventilation system.
492
ACKNOWLEDGMENTS
We thank the National Institute for Occupational Safety and Health (NIOSH) for funding
this research under grant 0000HCCR-2019-36404. We also thank Aaron Swift and John Aya
buri for their assistance in modeling the heat values for diesel equipment.
REFERENCES
Acuña, E. I., & Lowndes, I. S. (2014). A review of primary mine ventilation system optimization. Inter
faces, 44(2), 163–175. https://doi.org/10.1287/inte.2014.0736
Bienstock, D., & Zuckerberg, M. (2010). Solving LP relaxations of large-scale precedence constrained
problems. In F. Eisenbrand & F. B. Shepherd (Eds.), Integer Programming and Combinatorial Opti
mization 2010 (pp. 1–14). https://doi.org/10.1007/978-3-642-13036-6_1
Brickey, A. J. (2015). Underground production scheduling optimization with ventilation constraints.
(Doctoral Thesis). Colorado School of Mines. Arthur Lakes Library.
Brickey, A. J., Chowdu, A., Newman, A., Goycoolea, M., & Godard, R. (2020). Barrick’s Turquoise
Ridge gold mine optimizes underground production scheduling operations. INFORMS Journal of
Applied Analytics. https://doi.org/10.1287/inte.2020.1027
Bullock, R. L. (2011). Introduction to underground mine planning. In P. Darling (Ed.), SME Mining
Engineering Handbook (3rd ed., Vol. 1, pp. 1135–1141). Littleton: Society for Mining, Metallurgy &
Exploration.
Carlyle, W. M., & Eaves, B. C. (2001). Underground planning at Stillwater Mining Company. Interfaces,
31(4), 50–60. https://doi.org/10.1287/inte.31.4.50.9669
Chicoisne, R., Espinoza, D., Goycoolea, M., Moreno, E., & Rubio, E. (2012). A new algorithm for the
open-pit mine production scheduling problem. Operations Research, 60(3), 517–528. https://doi.org/
10.1287/opre.1120.1050
Chowdu, A. A., Nesbitt, P., Brickey, A. J., & Newman, A. (2020). Operations research in underground
mine planning : A review. INFORMS Journal on Applied Analytics. Forthcoming.
Deswik Mining Consultants Pty Ltd. (2022). Deswik - Delivering mining efficiency. Retrieved from
https://www.deswik.com/
Fava, L., Maybee, B., & Millar, D. (2012). Decision support for an underground gold mining operation -
A case study using the Schedule Optimisation Tool. Proceedings Project Evaluation 2012, (May),
41–48. Melbourne: Australasian Institute of Mining and Metallurgy.
Hall, B. (2017). Long-term mine planning and the mine-planning process. Retrieved October 18, 2022,
from https://www.amcconsultants.com/experience/long-term-mine-planning-and-the-mine-planning-
process
Huang, S., Li, G., Ben-Awuah, E., Afum, B. O., & Hu, N. (2019). A robust mixed integer linear pro
gramming framework for underground cut-and-fill mining production scheduling. International Jour
nal of Mining, Reclamation and Environment, 1–18. https://doi.org/10.1080/17480930.2019.1576576
Kuchta, M., Newman, A., & Topal, E. (2003). Production scheduling at LKAB’s Kiruna Mine using
mixed-integer programming. Mining Engineering, 55(4), 35–40.
Kuchta, Mark, Newman, A., & Topal, E. (2004). Implementing a production schedule at LKAB’s
Kiruna Mine. Interfaces, 34(2), 124–134. https://doi.org/10.1287/inte.1030.0059
Little, J., Nehring, M., & Topal, E. (2008). A new mixed-integer programming model for mine produc
tion scheduling optimisation in sublevel stope mining. In P. Knights (Ed.), Australian Mining Technol
ogy Conference 2008. Twin Waters.
Lopes, T. V. F. (2017). Underground mine production scheduling using OMP solver: analysis of modifica
tions to an integer programming model and implementation of a sliding time window heuristic. South
Dakota School of Mines & Technology.
Martinez, M. A., & Newman, A. M. (2011). A solution approach for optimizing long- and short-term
production scheduling at LKAB’s Kiruna mine. European Journal of Operational Research, 211(1),
184–197. https://doi.org/10.1016/j.ejor.2010.12.008
Maybee, B. (2011). A risk-based evaluation methodology for underground mine planning. ProQuest Disser
tations Publishing.
Mcisaac, G. (2005). Long-term planning of an underground mine using mixed-integer linear
programming. CIM Bulletin, 98(1089), 89.
McPherson, M. J. (1993a). Background to subsurface ventilation and environmental engineering. In Sub
surface Ventilation and Environmental Engineering (pp. 1–11). https://doi.org/10.1007/978-94-011-1550-
6_1
493
McPherson, M. J. (1993b). Subsurface ventilation and environmental engineering. In Subsurface Ventila
tion and Environmental Engineering. https://doi.org/10.1007/978-94-011-1550-6
Mine X Corporation. (2021). 2021 Annual report (Vol. 2021).
Mutmansky, J. M., Hartman, H. L., Ramani, R. V., & Wang, Y. J. (2012). Mine ventilation and air condi
tioning. Germany: Wiley.
Nehring, M., & Topal, E. (2007). Production schedule optimisation in underground hard rock mining
using mixed integer programming. In R. Flanagan (Ed.), Project Evaluation Conference 2007 (pp.
169–175). Melbourne: The Australasian Institute of Mining and Metallurgy.
Newman, A. M., & Kuchta, M. (2007). Using aggregation to optimize long-term production planning at
an underground mine. European Journal of Operational Research, 176(2), 1205–1218. https://doi.org/
10.1016/j.ejor.2005.09.008
Newman, A. M., Rubio, E., Caro, R., Weintraub, A., & Eurek, K. (2010). A review of operations
research in mine planning. Interfaces, 40(3), 222–245. https://doi.org/10.1287/inte.1090.0492
O’Sullivan, D., & Newman, A. (2014). Extraction and backfill scheduling in a complex underground
mine. Interfaces, 44(2), 204–221. https://doi.org/10.1287/inte.2013.0730
Ogunmodede, O., Lamas, P., Brickey, A., Bogin, G., & Newman, A. (2022). Underground production
scheduling with ventilation and refrigeration considerations. Optimization and Engineering, 7(3),
219–242. https://doi.org/10.1007/s11081-021-09682-4
Radin, R. L. (2017). Optimization in operations research (2nd ed.). Hoboken, NJ: Pearson Higher Educa
tion Inc.
Ramsden, R., Branch, A.R. and Wilson, R. (2007). Factors influencing the choice of cooling and refriger
ation systems for mines. Journal of the Mine Ventilation Society of South Africa, 60(3), 92–98.
Rivera, O., Goycoolea, M., Moreno, E., Alexandra, N., & Brickey, A. (2016). The OMP user’s guide.
Unpublished material.
Rivera, O., Goycoolea, M., Moreno, E., & Espinoza, D. (2021). The Open Mine Planner Guide.
Ryan, A., & Euler, D. S. (2017). Heat stress management in underground mines. International Journal of
Mining Science and Technology, 27(4), 651–655. https://doi.org/10.1016/j.ijmst.2017.05.020
Smith, M., Sheppard, I., Karunatillake, G., & Camisani-Calzolari, F. (2003). Using MIP for strategic
life-of-mine planning of the lead/zinc stream at Mount Isa Mines. 31st International APCOM Sympo
sium Proceedings, Capetown, South Africa, (Figure 3), 465–474. Retrieved from http://www.saimm.co.
za/Conferences/Apcom2003/465-Smith.pdf
Technologies, G. (2021). GT-SUITE. Retrieved from https://www.gtisoft.com/gt-suite/
Terblanche, S., & Bley, A. (2015). An improved formulation of the underground mine scheduling opti
misation problem when considering selective mining. ORiON, 31(1), 1. https://doi.org/10.5784/31-
1-422
Trout, L. P. (1995). Underground mine production scheduling using mixed integer programming. The
25th International APCOM Symposium, (July), 395–400. Retrieved from https://ausimm.com/product/
underground-mine-production-scheduling-using-mixed-integer-programming/
U.S. Department of Labor. (2012). Heat stress in mining. In National Mine Health and Safety Academy
Safety Manual Series (SM 6). Retrieved from http://www.ergonomics.org.au/downloads/EA_Jour
nals/Heat_Stress_in_Mining_-_leveritt.pdf%5Cnpapers2://publication/uuid/F2AECEBC-BA0E-4924-
94E6-3A9E247C01A8
van der Walt, J., de Kock, E. M., & Smith, L. K. (1983). Analysis of ventilation and cooling requirements
for mines. Journal of The South African Institute of Mining and Metallurgy, 83(2), 25–34. https://doi.
org/10.10520/AJA0038223X_1371
Williams, J. K., Smith, L., & Wells, P. M. (1973). Planning of underground copper mining. 10th Interna
tational Applied Computer Operations Research in Mineral Industries, 251–254.
Yeoman, K., Weakley, A., DuBose, W., Honn, K., McMurry, T., Eiter, B., Poplin, G. (2022). Effects of
heat strain on cognitive function among a sample of miners. Applied Ergonomics, 102, 103743. https://
doi.org/10.1016/j.apergo.2022.103743
494
APPENDIX A – GENERALIZED RESOURCE CONSTRAINED PROJECT
SCHEDULING PROBLEM (RCPSP)
Sets:
A set of activities
R set of resources
P set of activity precedence relationship
T set of time periods
Parameters:
d^a duration of activity a 2 A
cat objective value to be derived if activity a is initiated in time period t
δt discount factor for objective function in time period t
d^a;a delay between activity a0 and activity a
^rrt quantity of resource r available in time period t
qar quantity or resource r consumed by executing activity a
Decision variables:
Objective function:
Constraints:
The objective function (3) maximizes the discounted total value of parameter c from all
activities over the entire time horizon. This could represent the NPV from production oper
ations or quantity of mineral commodity produced. The objective function could as well be
altered to minimize parameters such as production costs or deviations from production target.
Constraint (4) ensure that no activity is executed more than once. For instance, a given stope
cannot be mined more than once. Constraint (5) impose precedence relationships on activities,
while constraint (6) enforces resource consumption limits. Constraint (7) ensure that the deci
sion variables are binary.
495
Mine ventilation and automation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
M.L. Harris, S.J. Schatzel, K.M. Ajayi, M. Van Dyke, P. Zhang, V. Gangrade &
J.D. Addis
National Institute for Occupational Safety and Health, Pittsburgh, PA, USA
ABSTRACT: This paper summarizes the changes in permeability for two boreholes located
above an abutment pillar in a longwall coal mine to characterize potential interaction between
shale gas wells and the coal mine operations under deep cover. To determine the safety of the
mine environment in case of a potential well breach, fracture network characteristics are
needed to conduct a comprehensive hazard assessment. Permeability was measured using
a falling-head slug test and calculated in accordance with the Hvorslev model during the
mine-by of a longwall panel on one side of the pillar. The two boreholes had screened lengths
at different depths to evaluate stratigraphic zones of interest above the active mining seam.
The permeability at each borehole increased from pre-mining to post-mining and was highest
while the face was close to and during mine-by of the test site.
1 INTRODUCTION
Within the past 15 years, unconventional shale gas wells have been drilled through current and
future coal reserves in Pennsylvania, West Virginia, and Ohio. Impacts on the mechanical
integrity of these wells becomes a concern when mining occurs in and around these wells. The
shale gas wells penetrate the coal seams and the coal seams are subsequently mined. The coal
and shale gas industries need to know what the impacts of longwall mining are on the shale gas
wells, what the potential deformation may be, and what stresses are imposed on the gas wells.
In 2012, the Pennsylvania Department of Environmental Protection (PADEP) recognized
the 1957 Pennsylvania Gas Well Pillar Regulation (Commonwealth of Pennsylvania, 1957)
was created without current data from modern day longwall mining and called for research to
revise the outdated regulation. Also, the 1957 Pennsylvania Gas Well Pillar Regulation has
been widely used by the Mine Safety and Health Administration (MSHA) along with CFR
75.1700 to manage gas well pillar stability issues. Therefore, given the posed questions and the
need for further guidance given modern mining technologies and practices, the National Insti
tute for Occupational Safety and Health (NIOSH) initiated research to address these issues.
One of the first steps in quantifying the interaction between the two processes is to determine the
change in ground permeability at depths where longwall-induced deformations are of highest con
cern. If longwall-induced subsurface strata deformations and stresses should affect shale gas well
casings and if a breach occurs, the changes in the ground permeability could potentially increase
the rate of shale gas infiltration into an underground coal mine. Permeability field measurements
have been conducted under various differing conditions. Boreholes placed in the barrier pillar and
in the longwall gob areas under differing depths of cover have been monitored during completion
of the longwall coal seam. Previously, three monitoring boreholes were monitored to experimen
tally measure borehole permeability and changing water head under shallower cover (deepest at
DOI: 10.1201/9781003429241-51
499
127 m or 417 ft). Watkins et al. (2021) assessed permeability changes around boreholes above an
abutment pillar under shallower cover. This paper will discuss a similar scenario except under
deeper cover conditions (up to 238 m or 1,047 ft) and during the first longwall panel mine-by.
2 TEST METHOD
The study included drilling two boreholes to various depths to monitor the Sewickley and Union
town horizons and performing field experiments to gather permeability data. Although Marcellus
shale gas reservoirs are at much greater depths than the Pittsburgh coal bed longwall mines, the
risk of a hypothetical casing failure is thought to be greatest where the maximum zone horizontal
movement occurs in response to mining. This zone is typically shallower than the mined coalbed
and within a few hundred meters of the unit. Additionally, the areas above abutment pillars are
particularly important as shale gas wells are required to be drilled through them.
Figure 1. Generalized stratographic profile for the VEP boreholes (not to scale).
The primary factors creating the gas transport network associated with longwall gobs are ground
movements related to subsidence. Previous ground control studies (Su et al., 2019) indicate the pri
mary movement component that potentially deforms well casings in longwall pillars is subsidence or
horizontal movement. FLAC3D™ modeling estimations show an insignificant amount of vertical
movement for borehole locations above abutment pillars. However, higher magnitudes of displace
ment perpendicular to the panel are typical and are shown for this area in Figure 2. Although strati
graphic zones of maximum horizontal movement in the region can be variable, previous modeling
shows that the Uniontown and Sewickley coal seams are among the likely locations (Su et al., 2019).
500
Most of the horizontal movement was estimated to be perpendicular to the longwall panel due to
the ground subsidence pulling the strata above the pillar toward the caved gob zone, which is per
pendicular to the panel direction. Figure 2 presents data for a nearby location and is representative
for this monitoring site. Here, a horizontal displacement of about 0.7 cm (0.3 in) with a predicted
casing diameter reduction of 0.15 cm (0.16 in) is predicted. A depth of approximately 155 m (509 ft)
is when horizontal displacement increases to about 2.0 cm (0.8 in) and the predicted casing diameter
reduction increases to 0.70 cm (0.3 in). These zones of maximum movement are also important loca
tions for fluid transport as these are typically high permeability zones for the transport of gas from
a hypothetical breach.
Figure 2. Shale gas well horizontal displacement by depth perpendicular to the panel from field meas
urements and FLAC3D™ modeling.
501
Figure 3. Top view of a three-entry longwall layout. Test boreholes and mining direction are shown.
completed. These test zones were chosen to correspond to probable locations of high ground
movement where increases in permeability were considered most likely. An INW PT2X (Sea
metrics™) piezometer tracked the long-term equilibrium water height and water slug height
(Figure 4). For the falling-head slug tests, a water slug height of up to 3 m (10 ft) was added
to the boreholes. The previous study (Watkins et al., 2021) recorded the fall of water head at
5, 30, or 60 s intervals, depending on the expected drainage rate of the water. However, during
the current study the rate of loss was at a much slower rate and, therefore, the data was
recorded every minute, and the falling rate was later determined from the downloaded data.
Figure 4. The sideview of the VEP-S and VEP-U monitoring boreholes (not to scale).
When the longwall face was 364 m (1193 ft) away from the monitoring site, the slug tests
were conducted on each of the boreholes twice a week. The frequency increased to three times
per week when the longwall face was within 100 m (328 ft) of the boreholes. When the active
longwall face passed the borehole locations and was further away, the sampling rate for slug
tests was approximately once per week. Once the longwall panel was completely mined, long-
term continuous water pressure readings were recorded every eight hours. The piezometer
readings, water slug height (Hw), and initial slug height (Ho) were converted to Hw/Ho values.
For the calculation of permeability from falling-head slug tests, Hw/Ho values were plotted on
a semi-log graph to determine the T37 time, which is the time, in minutes, in which the water slug
drained to 37% of the initial water slug height (Dawson & Istok, 1991; Watkins et al., 2021). For
a 10-ft water slug, the T37 time would be calculated as the time it took for the slug to drain to 3.7 ft.
If the T37 value did not fall within the given data range, two points were chosen along the most
linear section of the semi-log plot of the slug test, and the slope of those two points estimated the
T37 time. Watkins et al. (2021) details the application of T37 in the Hvorslev method to determine
the hydraulic conductivity. The hydraulic conductivity is dependent upon the well casing radius, the
well screen radius, and the length of the well screen. For both VEP boreholes, the well casing radius
502
is 0.052 m (2 in), and the well screen radius is 0.104 m (4 in). The well screen length is 4.6 m (15 ft)
for VEP-U and is 3.0 m (10 ft) for VEP-S.
As with the previous study, multiple assumptions about the slug test and ground aquifer are
applied when conducting the Hvorslev method to calculate permeability. This method assumed that
the water slug is added instantaneously, the groundwater flow is described by Darcy’s Law, and the
volume of water that flows into the aquifer is equivalent to the change in water volume within the
well casing (Watkins et al., 2021). The aquifer is assumed to be incompressible, lithologically homo
geneous, isotropic, and vertically confined by aquicludes. The injection well is assumed to have
a negligible radius in relation to the size of aquifer, the well has a screen with a negligible head loss,
and the water slug flow travels horizontally away from the well in all directions.
3.1 VEP-U
Figure 5 displays the changes in water head from experimental data in VEP-U. The data was col
lected when the longwall panel was at varying distances from the monitoring boreholes. In this
figure, the negative distance (– 54 m or -177 ft) in the legend refers to an approaching longwall
face and the positive distance (+61 m or +200 ft) refers to the longwall face moving away. Zero
distance is where the longwall face passed the center of the test abutment pillar and is the basis for
position measurements. As seen in Figure 5, the permeability was 3.19 x 10-15 m2 (3.2 mD) when
the longwall was 54 m (177 ft) from mining by the monitoring holes. The permeability de-creased
slightly to 2.51 x 10-15 m2 (2.51 mD) during the mine-by before rising to 5.33 x 10-15 m2 (5.33 mD)
when approximately 61 m (200 ft) beyond the monitoring boreholes.
Figure 5. VEP-U water height curves for slug tests at various longwall positions. The negative value
refers to the approaching longwall face and the positive value refers to the face moving away from the
test location.
Figure 6 shows the permeability, m2, and pressure measurements, kPa, at the Uniontown hori
zon from VEP-U. The numbers above the bars are the distances (m) of the longwall panel from
the monitoring boreholes. The black negative numbers above the bars indicate the longwall panel
approaching the monitoring location and the red positive numbers above the bars indicate the dis
tance as the longwall panel passes the monitoring location. Prior to mining, the permeability in
VEP-U was 0.19 x 10-15 m2 (0.19 mD). The borehole water level steadily rose prior to 08/23/2021.
This is seen in the rise in associated pressure measurements (orange plots) on Figure 6. The longwall
panel was mined by the abutment pillar on 08/25/2021 when the permeability was 2.51 x 10-15 m2
503
(2.54 mD). The water levels and pressures were also the highest at this time with pressures just over
910 kPa (132 psi). When the longwall panel passes the monitoring point, the permeability increases
and reaches a maximum of 5.33 x 10-15 m2 (5.4 mD) and the pressure in the hole drops to 585 kPa
(85 psi). After this maximum, the permeability drops to a low of 8.78 x 10-16 m2 (0.89 mD) before
slightly rising. The downhole pressure is relatively steady during the first half of September and
hovers around 500 kPa (73 psi) before dropping at the end of September.
Figure 6. Permeability values and pressure measurements from VEP-U at the Uniontown horizon.
Black numbers on the graph indicate the distance (m) of the longwall face advancing toward the monitor
ing point, and red numbers indicate the distance (m) of the longwall.
3.2 VEP-S
Figure 7 displays the water changes before, during, and after the longwall panel mine-by of VEP-S.
When the longwall face was approaching the monitoring site (54 m or 177 ft away), the permeabil
ity was 1.41 x 10-15 m2 (1.43 mD). The permeability then increased to 2.70 x 10-15 m2 (2.74 mD)
during the longwall panel mine-by before increasing again to 1.27 x 10-14 m2 (12.8 mD).
Figure 7. VEP-S water height curves for slug tests at various longwall positions. The negative value refers
to the approaching longwall face and the positive value refers to the face moving away from the test location.
504
Figure 8 shows the pressure measurements and the corresponding permeability values. The
highest permeability, 1.755 x 10-14 m2 (17.8 mD), occurred two days after the mine-by when
the panel was 19 m (62 ft) past the monitoring point (Figure 8). The associated water pressure
at this point was 885 kPa (128 psi). After the longwall panel mine-by, the pressure decreased
and held steady at around 500 kPa (73 psi), while the permeability values varied in the range
of 0.97 x 10-15 m2 – 1.93 x 10-15 m2 (0.98 mD – 1.96 mD). Towards the end of September, the
pressure started decreasing again as the permeabilities slightly increased.
Figure 8. Permeability values and pressure measurements from VEP-S at the Sewickley horizon. Black
numbers on the graph indicate the distance (m) of the longwall face advancing toward the monitoring
point and red numbers indicate the distance (m) of the longwall.
3.3 Comparison
Table 1 lists the minimum and maximum measured permeability values after the first longwall
panel mine-by as well as the predicted horizontal displacements. As with the previous study,
the deeper Sewickley horizon saw higher permeability values since it is closer to the mining
horizon and would be most impacted by the compression of the fractured rock. The maximum
permeabilities under deeper cover are smaller than for shallower cover by an order of magni
tude of two for both horizons. As seen in Table 1, the maximum permeability for the Sewick
ley horizon under deeper cover was 1.75 x 10-14 m2 (17.8 mD) and 1.34 x 10-12 m2 (1358 mD)
under shallower cover. For the Uniontown horizon, the maximum permeability under deeper
cover was 5.33 x 10-15 m2 (5.4 mD) and 2.42 x 10-13 m2 (245 mD) under shallower cover.
Table 1. Minimum and maximum permeability values for VEP-S and VEP-U resulting from this study
and for corresponding FEB-1 and FEB-2 (Watkins et al., 2021).
Longwall Panel 1
Measured Permeabilities Predictions
Horizontal
Borehole Horizon Overburden Minimum Maximum Displacement
ID of Interest Depth m (ft) m2 (mD) m2 (mD) cm (in)
VEP-S Sewickley 328 (1,076) 4.64E-16 (0.47) 1.75E-14 (17.8) 0.6 (0.2)
VEP-U Uniontown 291 (955) 1.84E-16 (0.19) 5.33E-15 (5.4) 0.3 (0.1)
FEB-1 Sewickley 127 (417) 4.06E-13 (411) 1.34E-12 (1358) 17 (6.7)
FEB-2 Uniontown 76 (249) 2.16E-14 (22) 2.42E-13 (245) 4 (1.6)
505
The previously stated trend also correlates with the predicted horizontal displacements. The
predicted displacements for both shallower and deeper covers are larger for the deeper Sewick
ley horizons than for the Uniontown. The horizontal displacement prediction was 0.6 cm (0.2
in) for the deeper cover and 17 cm (6.7 in) for the shallower cover and, for the Uniontown
seam, it was 0.3 cm (0.1 in) and 4 cm (1.6 in) for the deeper and shallower covers, respectively.
The predicted horizontal displacements are also 1-2 orders of magnitude in difference with the
shallower cover having higher values than that of the deeper cover.
4 CONCLUSIONS
DISCLAIMER
The findings and conclusions in this report are those of the author(s) and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health,
Centers for Disease Control and Prevention. Mention of any company or product does not
constitute endorsement by NIOSH.
REFERENCES
CFR. Code of Federal Regulations. Washington, DC: U.S. Government Printing Office, Office of the
Federal Register.
Commonwealth of Pennsylvania, Department of Mines and Mineral Industries, Oil and Gas Division.
1957. Joint Coal and Gas Committee, Gas Well Pillar Study. Harrisburg, Pennsylvania, 28 pp.
Dawson, K. J., & Istok, J. D. (1991). Aquifer testing: Design and analysis of pumping and slug tests (pp.
297–305). Lewis Publishers, Inc.
Mohan, G. M., Sheorey, P. R., & Kushwaha, A. 2001. Numerical estimation of pillar strength in coal
mines. International Journal of Rock Mechanics & Mining Sciences, 38, 1185–1192.
Palchik, V. 2005. Localization of mining-induced horizontal fractures along rock layer interfaces in over
burden: Field measurements and prediction. Environmental Geology, 48, 68–80.
Rusnak, J., & Mark, C., 2000. Using the point load test to determine the uniaxial compressive strength
of coal measure rock. In Proceedings of the 19th international conference on ground control in mining
(pp. 362–371). West Virginia University.
Su, D. W. H., Zhang, P., Schatzel, S. J., Gangrade, V., Watkins, E., et al. 2019. Longwall-induced sub
surface deformations and permeability changes shale gas well casing integrity implication. In Proceed
ings of the 38th international conference on ground control in mining (pp. 49–59). Morgantown, WV:
West Virginia University.
Watkins, E., Karacan, C. Ö., Gangrade, V., & Schatzel, S. 2021. Assessing Gas Leakage Potential into
Coal Mines from Shale Gas Well Failures: Inference from Field Determination of Strata Permeability
Responses to Longwall-Induced Deformations. Natural Resources Research, 30(3), 2347–2360.
506
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
1 INTRODUCTION
The integrity of an unconventional shale gas well positioned in the abutment pillar of a longwall
mine could be compromised due to mining-induced deformation. Consequently, gas from the
compromised casing(s) is released in the overburden and could flow towards the mine. This
increases the risk of an explosion or accumulation of explosive gas beyond the mandated limits
(Mine Safety and Health Administration, 2018). This is a novel problem as the coexistence of
coal and shale gas is an emerging field in current and future coal reserves in Pennsylvania, West
Virginia, Ohio, Virginia, and Tennessee. Currently, it is required that shale gas wells near mining
should be plugged or adequately protected by a designed pillar following the design criteria devel
oped from the 1957 Pennsylvania Gas Well Pillar Study (Commonwealth of Pennsylvania, 1957),
which predates longwall mining. This current study requires an update to consider new mining
methods such as longwall mining and the impact of deep-cover mining. Hence, researchers at the
National Institute for Occupational Safety and Health (NIOSH) have been conducting research
to provide engineering guidelines for shale gas wells influenced by longwall mining under shallow,
medium, and deep cover. Findings from these studies have predicted and measured horizontal
displacement within the abutment pillar where the unconventional gas wells are positioned (Su
et al., 2019, Su et al., 2018, Zhang et al., 2019a, Zhang et al., 2019b). For a medium and deep-
cover site in southwestern Pennsylvania, Su et al. (2019) predicted that the horizontal displace
ment under deep cover (at 361-m depth) is one order of magnitude smaller than the medium-
cover site. Even though this observation might be site specific, the findings show that the impact
of mining-induced deformation could vary with depth.
DOI: 10.1201/9781003429241-52
507
The Pennsylvania longwall mine data shows that mining depth could be as shallow as
30.48 m (100 ft) in stream locations and as deep as 366 m (1,200 ft) (Pennsylvania Department
of Environmental Protection, 2022). Similarly, the depth of the target Marcellus Shale could
vary from 1,500 to 2,700 m (Zhang et al., 2019b). Therefore, these unconventional shale gas
wells have been drilled through coal seams, and the impact of mining-induced deformation on
the gas well casing could vary due to the mining depth. Even though the current research by
NIOSH is aimed at providing engineering principles to ensure a safe co-existence of both oper
ations (coal mining and shale gas extraction), it is also important to consider the impact of
a hypothetical shale gas well casing breach. The possibilities that the gas could flow towards the
mine is dependent on the permeability of the surrounding strata from the breach location to the
mine. Schatzel et al. (2012) observed that mining-induced permeability could change up to 7
months after mining with a hundred or thousands millidarcy increase. Hence, this research
group has conducted field studies to measure mining-induced permeability at study sites in
southwestern Pennsylvania under shallow and deep cover (Watkins et al., 2021). Due to the
limitations of field studies in the number of strata that could be monitored, a discrete fracture
network (DFN) model was developed using the core log data of the study sites (Ajayi et al.,
2022, Khademian Z. et al., 2022). DFN models are preferred because the fractures are modelled
explicitly similar to field conditions unlike the continuum approach. It assumes that the fracture
permeabilities are notably greater than the permeability of the rock matrix such that it can be
ignored . The model is stochastic because it is impractical to exactly represent the fracture net
work in the overburden. It is used to predict the permeability of all the strata in the overburden,
and the results are validated with the field measurements. Previous numerical models for
a shallow-cover site (Ajayi et al., 2021) and a deep-cover site (Khademian Z. et al., 2022) have
focused on the permeability over the abutment pillar. However, this study presents the perme
ability for the overburden from the center of the abutment pillar to the gob (Figure 1). The
difference in the induced permeability is used to analyze the impact of a hypothetical casing
breach on the mine under different covers. The research approach is presented in Section 2.0
and the findings are presented in Section 3.0 with conclusions in Section 4.0.
2 RESEARCH APPROACH
508
shale, limestone, shale, coal, and shaley limestone. The mining horizon for the shallow-cover
site is 146.9-m (482-ft) and four monitoring boreholes were drilled over the abutment pillar.
Three of the boreholes monitored mining-induced permeability changes during the mining of
both panels, and the fourth borehole monitored the ground movement due to longwall
mining. The boreholes monitored permeability changes at specific horizons: the first borehole
(tagged FEB 1) monitored the Sewickley coal horizon at 124.97-m (410-ft) depth, the second
borehole (tagged FEB 2) monitored the Uniontown coal horizon monitored at 79.29-m (260-
ft) depth, and the third borehole (tagged FEB 3) monitored 41.15-m (135-ft) depth, which tar
gets the projected top of the fractured zone. The choice of these monitoring locations is based
on an earlier geomechanical study for the shallow-cover site, which identified that the Sewick
ley and Uniontown horizons are the primary zones of horizontal displacement of 17 cm and
4 cm, respectively. The deep cover site is a longwall mine with an overburden depth of
341 m (1,119 ft). Tables 1 and 2 present a summary of the field permeability measurements
conducted using a falling-head test for the shallow and deep-cover sites, respectively.
Table 1. Field permeability measurements over the abutment pillar for shallow cover (Watkins E, 2020).
Borehole ID First Panel Range m2 (mD) Second Panel Range m2 (mD)
13 12 12 12
FEB 1 4:07 � 10 1:35 � 10 (411–1,360) 1:09 � 10 5:03 � 10 (1,100–5080)
14 13 14 13
FEB 2 2:17 � 10 2:43 � 10 (21.9–245) 1:14 � 10 3:82 � 10 (11.5–386)
12 11 11 10
FEB 3 2:70 � 10 3:26 � 10 (2,730–32,900) 3:81 � 10 1:31 � 10 (38,500–132,000)
Table 2. Permeability measurements over the gob for deep cover (Khademian Z. et al., 2022).
Borehole ID Location First Panel Range m2 (mD)
Pre-mining- 295-m depth and 98 m away from the tailgate gateroad 2:42 � 10 13 2:81 � 10 13
Borehole A (245-285)
Post Mining- 265-m depth at the Uniontown coal horizon and at the 6:61 � 10 14 1:21 � 10 13
509
Figure 2. Summary of the research approach.
2.3 Development of multizone DFN model and numerical model for permeability
As described in Section 1.0 (Introduction), it is impractical to exactly represent the fractures
in the overburden. Therefore, a stochastic DFN model is developed in Fracture Flow Code
(FFC) to predict mining-induced permeability. In a previous study conducted with FFC, the
overburden was classified into three zones: Tailgate Abutment pillar area, Gob area, and
Headgate chain pillar area. However, this approach has its limitations because a mining-
induced aperture could change within a short interval. Therefore, FFC is updated such that it
discretizes the overburden. The region over the abutment pillar is classified into 5-m zones
toward the edge of the panel, and the section over the gob is classified as 10-m zones. It is
assumed that these zones completely map the exact values of the aperture value from the geo
mechanical model in 3DEC. Within each zone, the center location of the vertical fractures is
generated using uniform distribution, the length of the fractures is modeled with lognormal
distribution, the fracture density is imported from the 3DEC model, the aperture is modeled
with lognormal distribution using the peak value of each zone from 3DEC, and the fracture
orientation is modeled with Von Mises Fischer’s distribution. The model generates 100 DFN
realizations to replicate the potential variation in fracture geometries, and the average perme
ability is presented in this study. Figure 3 shows one of the DFN realizations of the multizone
model generated in FFC for the shallow-cover site.
For each zone in Figure 3, the vertical and horizontal permeabilities are determined using the
method presented in a previous study (Zhang et al., 1996). Cubic law is used to model flow through
the fractures and Darcy’s equation is used to predict the permeabilities (Zhang et al., 1996):
In Equation 1, q indicates the direction of flow and velocity (m/s), and it is calculated by divid
ing the sum of boundary flow from cubic law (Q, m3/s,) by the boundary area (A, m2),
rHx and rHy are the pressure head gradients in the x and y directions, respectively, and K is
the permeability coefficient or conductivity (m/s). The principal (qyy and qxx) and cross flow (qxy
and qyx) in the y and x directions, respectively, are obtained by applying boundary pressure inde
pendently in both directions. The absolute permeability tensor (k in m2) is obtained from:
ρ is the density of fluid kg/m3, g is the acceleration due to gravity (m/s2), μ is the dynamic
viscosity of the fluid (Ns/m2). The principal absolute permeabilities in the x (kxx) and y (kyy)
directions are the horizontal and vertical permeabilities.
510
Figure 3. Stochastic DFN model generated for FFC.
This section presents the permeability results from FFC for shallow and deep cover from the
center of the abutment pillar to 350 m (1,148 ft). Even though the width of the panel is
457.2 m (1,500 ft), the choice of 350 m (1,148 ft) is to focus on the area of interest closer to the
gas well in the abutment pillar and save on the computational time required for the full
model. Figures 4 and 5 show the average permeability from 100 DFN realizations for each of
the zones in the overburden for shallow and deep cover, respectively. These permeability maps
are developed using the same range for ease of comparison.
It is observed that from the permeability map in Figure 4 that the values over the abutment
pillar (within 0–15 m) are about an order of magnitude lower than the edge of the panel. For
most of the strata, the permeability at the edge of the panel is highest and gradually decreases
towards the gob. Under this shallow-cover condition, the impact of mining-induced
511
deformation is significant within 100 m from the center of the panel and gradually decreases
towards the gob except for the strata within 40 m from the mine roof and about 10 m from
the surface. The strata directly closest to the mine roof (within 40 m) seems to have an
extended higher permeability region into the gob. The lowest permeabilities are toward the
center of the gob for strata between 10–100 m. Even though these observations are unique for
this site, the results show that mining-induced deformation could significantly impact the over
burden under shallow cover. By comparing Figures 4 and 5, the mining-induced permeability
for most of the zones under deep cover in Figure 5 is significantly lower than in shallow cover
(Figure 4).
The highest permeability under the deep cover is near the edge of panel towards the surface
and could be due to the impact of subsidence. Like the shallow-cover site, the strata within
30–40 m from the mine roof have higher permeability compared with the strata above. Apart
from these locations, there are multiple extremely low permeability regions/zones in the over
burden close to the range of in-situ or undisturbed rock mass. This is a clear distinction
between the mining-induced permeability for deep and shallow cover. Based on the difference
in induced permeability, it is predicted that the impact of a shale gas casing breach under shal
low cover could create more concerns for the mine compared to deep cover. For the shallow
and deep cover, Figure 6 shows the exact plot of the permeabilities at about 100 m from the
center of the pillar for the stratum directly at the mine roof. The results show the permeability
at the edge of the panel (between 15–20 m) is about one to two orders of magnitude greater
than the permeability over the abutment pillar (0–15 m). The underlying reasons for perme
ability differences are the abutment loading and caving process. At the center of the pillar, the
loads from retreated panels compact the fractures and reduce their aperture. Approaching the
panel edge, bedding plane separation and caved blocks significantly increase the fracture aper
tures and thus their permeability.
512
Figure 6. Comparison of permeability for shallow and deep cover.
4 CONCLUSIONS
This study presents the comparison of mining-induced permeability for a deep and shallow
site in southwestern Pennsylvania. A discrete fracture network model is developed in FFC
with aperture values imported from the geomechanical model in 3DEC. This stochastic DFN
model generates 100 realizations to account for potential variation in fracture geometry and
the average for each zone is compared. The findings from this study shows that:
i. For both shallow and deep cover, there are notable permeability changes for the strata
within 30–40 m of the mine roof, and a shale gas casing breach within this horizon could
provide pathways for gas flow toward the mine,
ii. Under shallow cover, the high permeability zone for each stratum at the edge of the panel
could extend into the gob to about 100 m and,
iii. Under deep cover, there are multiple locations over the gob with very low permeability
values such as in-situ or undisturbed rock mass.
These findings provide insights into the potential impact of a casing breach under different
mining depth and could help toward development of best practices for the safe operations of
both coal mining and shale gas extraction.
DISCLAIMER
The findings and conclusions in this paper are those of the authors and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health,
Centers for Disease Control and Prevention. Mention of any company or product does not
constitute endorsement by NIOSH.
REFERENCES
Ajayi, K., Khademian, Z., Schatzel, S., Watkins, E. & Gangrade, V. (2021) Numerical modeling of
longwall-induced permeability under shallow cover. Mine Ventilation. CRC Press.
Ajayi, K., Khademian, Z., Schatzel, S., WATKINS, E. & Gangrade, V. (2022) A Discrete Fracture Net
work Model for Prediction of Longwall-Induced Permeability. Mining, Metallurgy & Exploration, 1–8.
Commonwealth of pennsylvania (1957) Joint Coal and Gas Committee, Gas Well Pillar Study. Harris
burg, Pennsylvania:, Department of Mines and Mineral Industries Oil and Gas Division
513
Khademian Z., Ajayi, K. M., SU, D. W. H., Schatzel, S. J., Kim, B. H. & Esterhuizen, G. (2022) Rock
mass permeability induced by longwall mining under deep cover: potential gas inflow from a sheared
gas well. SME Annual Conference & Expo. Utah.
Mine safety and health administration (2018) CFR § 75.323, Title 30, Electronic Code of Federal Regula
tions. Mineral Resources, Department of Labor.
Pennsylvania department of environmental protection (2022) Longwall Mine Data. https://www.dep.
state.pa.us/dep/deputate/minres/LongwallForestlandStudy/Table1.pdf.
Su, D. W., Zhang, P., Van Dyke, M. & Minoski, T. (2018) Effect of Cover Depth on Longwall-Induced
Subsurface Deformations and Shale Gas Well Casing Stability. 52nd US Rock Mechanics/Geomecha
nics Symposium. American Rock Mechanics Association.
Su, D. W., Zhang, P., Van Dyke, M. & Minoski, T. (2019) Effect of longwall-induced subsurface deform
ations on shale gas well casing stability under deep covers. International Journal of Mining Science and
Technology, 29, 3–8.
Watkins E, G. V., Schatzel S, Hollerich C, Addis J (2020) Permeability Determination for Potential Inter
action between Shale Gas Wells and the Coal Mine Environment due to Longwall-induced Deform
ations. 2020 SME Annual Meeting and Exhibit. Phoenix, AZ.
Watkins, E., Karacan, C. Ö., Gangrade, V. & Schatzel, S. (2021) Assessing Gas Leakage Potential into
Coal Mines from Shale Gas Well Failures: Inference from Field Determination of Strata Permeability
Responses to Longwall-Induced Deformations. Natural Resources Research, 1–14.
Zhang, P., Dougherty, H., su, D. & Trackemas, J. (2019a) Influence of Longwall Mining on the Stability of
Gas Wells in Chain Pillars. Proceedings of the 38th International Conference on Ground Control in Mining.
Zhang, P., su, D. & Lu, J. (2019b) Influence of Longwall Mining on the Stability of Shale Gas Wells in
Barrier Pillars. American Rock Mechanics Association. New York.
Zhang, X., Sanderson, D., Harkness, R. & LAST, N. (1996) Evaluation of the 2-D permeability tensor
for fractured rock masses. International journal of rock mechanics and mining sciences & geomechanics
abstracts. Elsevier.
514
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
C. Stewart
Minware, Brisbane, Australia
ABSTRACT: Jet fans are extensively used in civil road tunnels, large enclosures (such as under
ground car parks) and other applications requiring the movement of air in large low-pressure
voids but are rarely systematically used in underground mining. Despite a relatively low cost and
compact size compared to typical mine axial fans, jet fans have found little practical use in most
mines, apart from occasional applications in coal and room and pillar mines. Jet fans have limited
pressure capabilities and are inefficient in moving airflow through higher-resistance airways.
This research explores the technical merits of jet fans for new applications in mining envir
onments and highlights areas where the fans may be highly beneficial compared to conven
tional fans or ventilation controls such as louvres or doors. Jet fans can excel in applications
that benefit from producing targeted ventilation flows while allowing unrestricted vehicle
movement that would otherwise be hindered by doors or ventilation controls.
An overview of the theory and application of jet fans in mining environments is provided
and examples and theoretical effectiveness of potential uses in ramp flow control and block
cave extraction ventilation is researched using modelling software. Further detailed design
and practical test work for specific applications is intended for future research.
1 INTRODUCTION
1.1 Background
Jet fans, sometimes known as impulse or induction fans are derived from a specialised vari
ation of axial fans and came to prominence in the early 1960s in US road tunnels (ErLeong &
Mashrae 2022). The technology was later enhanced by Japanese fan manufacturers in the
1970s and then in South Korea in 1986.
While jet fans can be based on either axial or centrifugal design, the most common type is axial
due to compact sizing and the useful horizontal in-line thrust vector produced for tunnelling.
Axial jet fans consist of a cylindrical fan casing designed to generate a high-velocity jet of air
from the discharge of the fan. The accelerated airflow through the fan generates an impulse or
‘thrust’ which is imparted and spread to the surrounding airflow via the jet stream. The jet can be
directed into large chambers to create air movement (Drenda et al., 2018) or for improved fan
ventilation (Konduri et al., 1998), however, the primary use of jet fans is for tunnel ventilation by
creating a net positive pressure and increasing (or decreasing) overall air movement to greater
amounts than the discharge air from the fan alone (ErLeong & Mashrae 2022).
The overall static pressure induced for large tunnels is relatively low, rarely exceeding a few
tens of pascals (10 – 30 Pa) however in low resistance applications this may be sufficient to
generate significant changes in airflow, particularly when multiple fans are used in parallel or
spaced along the tunnel in series. Jet fan thrust produced is not significantly influenced by sur
rounding tunnel air velocity, working well in contraflow, counterflow or even stagnant tunnel
conditions.
DOI: 10.1201/9781003429241-53
515
1.2 Jet fan design
Jet fans are a relatively simple and robust design consisting of an inline impellor, motor, motor
shroud and straightener arrangements within a cylindrical casing, and are often supplied with
inlet and exhaust silencers to reduce noise. Unlike higher-pressure axial fans, the blade size to
hub ratio tends to be high to optimise low pressure movement of the high air volumes and the
casing design from inlet to exhaust is usually straight with no expanding discharge diffuser.
Fan capabilities are normally specified by thrust (N) produced in a static air environment,
and specifications may also include volume flow, discharge velocity, fan power, and relevant
dimensions data such as diameter and width. Fans can also be designed in variable speed and
reversible configurations to provide optimum control of thrust and direction. Fully reversible
or bidirectional fans can produce similar thrust in both directions. Partially bidirectional fans
can produce highly efficient thrust in one direction but lower efficiency in reverse, while uni
directional is only designed to produce thrust in one direction. Theoretical fan efficiency is
based on the motor shaft power versus the velocity pressure and air volume flow produced
but specifications often rated efficiency on the thrust (N) to power (kW) ratio, with high
ratios indicating better efficiency.
The most important data for performance calculations is the theoretical thrust produced,
which is a product of the kinetic energy imparted to the air mass through the fan into the dis
charging jet stream. The momentum of the discharged air mass is diffused and transferred
into the surrounding air mass creating a net positive pressure. Thrust in Newtons can be calcu
lated as follows.
Where ρ = air density (kg/m3), Q= airflow volume (m3/s) and A = Fan outlet area (m2)
The theoretical thrust however is eroded based on installed conditions, and the actual or
installed thrust depends on the prevailing air velocity into the fan, the placement of the fan in
relation to the centreline of the drift, and the angle of the airflow relative to the surrounding
wall. The installed thrust can be estimated by the following equation.
Where tunnel velocity correction, wall proximity correction factor and jet inclination correc
tion factor, ranging from 1.0 (horizontal) to 1.1 (5° - 7° tilt to centre).
To convert thrust into a pressure rise across a tunnel, the following equation can be used.
516
where A = Area of Tunnel (m2)
where DF = Jet Fan Diameter (m), DT = tunnel diameter (m), Z = distance of jet fan axis to
tunnel wall or ceiling (m)
1.2.1 An example
A fan manufacturer offers an 8.5kW jet fan at a specified 278N of thrust, an outlet velocity of
30.5 m/s, a diameter of 560mm, and is installed in a 5m x 5m tunnel with airflow of 2.0m/s. The
jet fan is installed horizontally in line with the tunnel axis, with the fan centre axis 1.0m below
the ceiling. The applied fan thrust and pressure rise due to the fan can be calculated as follows.
Separation Factor =(2 × 1.0 – 0.56/(5.0 - 0.56) = 032
k1 = 0.93, k2 = 0.94 (from Figure 2), k3 = 1.0 (horizontal alignment)
Installed Thrust (Ti) = 278 × 0.93 × 0.94 × 1.0 = 243N
Pressure Rise Δ P=243/25 =9.7 Pa
The example above shows the induced pressure of a jet fan can be relatively small, and of
little use for conventional air movement through the shafts and tunnels of a large mine. In
comparison, the applied pressure of a typical mine axial fan through a pressure seal such as
a door or bulkhead can be several magnitudes greater than the 9.7 Pa shown in the above
example.
The advantage of a jet fan however is that it can help distribute or correct airflow in low-
pressure, open environments, without the need for a sealing door, brattice or bulkhead that
would otherwise restrict traffic flow or add to ventilation infrastructure cost.
Mines have a safety and legal responsibility to ensure adequate airflow is delivered to all
working areas of the mine to meet or exceed minimum velocities, satisfy diesel equipment
requirements, and achieve dilution standards. In a large dynamic mine with many work areas,
517
this requirement can be challenging and may require complex and expensive ventilation circuit
arrangements, frequent adjustment of controls, or overventilation of areas to ensure minimum
standards are always met.
In most cases, the level of control required comes at the cost of substantial resources and
mining infrastructure, expensive and restrictive (to traffic) controls such as doors and louvres,
energy for ventilation fans, and of course manpower to manage and ensure everything works
as required.
In some applications however, jet fans may alleviate much of this burden by removing the
need for some controls and allowing optimisation of airflow without wastage.
In road tunnels for example, as illustrated in Figure 3, the airspeed is often regulated
depending on the volume of traffic, gas concentrations and the time of day. A control system
using air velocity sensors will activate jet fan operations, adjusting the fan speeds and some
times the direction to ensure the targeted air velocity is achieved. Vehicle movement piston
effects are often enough to drive much of the flow, but during traffic slowdowns or jam condi
tions, the jet fans will ensure that targeted flow is achieved.
Figure 3. Jet fan components & considerations for tunnel air velocity control (not to scale).
The figure above shows an example of methods of controlling air velocity depending on
traffic volume and speed at the time. The jet fan operation is controlled via a linked PID (pro
portional, integral, derivative) control system that operates jet fan numbers, speed, and direc
tion. During times of rapid, high-volume vehicle movement, piston effects may be sufficient to
move the majority of required air volume through the tunnel, and minimal jet fan use may be
required. During low-volume vehicle times or when traffic slowdowns occur, more jet fans are
run (and at higher speeds) to achieve safe air volumes. Reversible jet fans can even be used to
retard ventilation flow, so excessive ventilation is not consumed at the expense of adjoining
tunnels, or to prevent tunnel fumes from exiting the portal.
2.1 Performance and efficiency comparison of high-pressure auxiliary fans vs jet fans
Auxiliary fans are occasionally used in mines for jet fan applications and are free-hung with
out ducts or walls to boost flow along stagnant drives or push air into dead-end headings or
large chambers. However, the efficiency and effectiveness of using conventional auxiliary fans
in jet fan applications should be considered. The fan total pressure efficiency of a high-
pressure auxiliary fan operating under no static pressure may be only 40% or lower.
A comparison of an auxiliary and a jet fan is provided below.
A 1259mm diameter and 55kW motor power (Howden AXN-1259) using Equation 1 sug
gests a theoretical thrust of up to 1185N would be possible. A similar specialised jet fan design
of the same power and size (Howden APA1250/578) is catalogued (Howden Ventilatoren
GmbH, 2022) to produce almost double the thrust at 2025N. A smaller, lower power jet fan
(e.g., Howden APA 1000/403 1000mm fan with 30kW motor) will produce the same thrust as
the higher powered larger 55kW mine auxiliary fan.
518
In summary, it is rarely an economical solution to employ high-pressure auxiliary fans in jet
fan applications unless specialised discharge reduction cones are used to improve fan jet per
formance (Wolski, 1995) or the fan is employed in very short-term applications for convenience.
Figure 4. Theoretical pressure balance equation for parallel drift flow control.
Kirchhoff’s law applied to ventilation and used in the Hardy Cross algorithm indicates that
the sum of the pressure losses (or gains) in any loop circuit must equal zero. The figure above
can be described by the following equation.
Where P1ab and P2ab is the frictional loss (Pa) of the perimeter drift between crosscuts
Pja and Pja is the combined crosscut frictional pressure loss and jet fan addition (Pa).
Pva and Pvb is the piston effect (Pa) of the vehicle movements in the drifts.
Assuming similar system pressures at P1ab and P2ab at both ends, Atkinson’s formula can be
used to estimate airflow influence from a jet fan. Assuming the piston effect of the vehicle is
negligible, a +/- 15 Pa jet fan pressure influence is available (the capability of a modest-sized
11kW jet fan). The airflow difference in a 500m long 20m2 drift can be calculated as shown
below.
519
In summary, the application of jet fans in equal pressure crosscuts with a thrust influence of
+/-15 Pa can change the airflow in each by over +/- 30 m3/s. The movement of vehicles may
increase or decrease this amount due to piston effects, depending on direction, however, the
average overall increase remains the same. This offers a huge range of ventilation control poten
tially suitable for most large diesel equipment. The above calculations of course can be modelled
more quickly by ventilation simulation software to help design and prove larger systems.
2.4 Conceptual application of jet fans in large block cave haulage and extraction
Block cave mines are usually designed with multiple parallel extraction and haulage drifts
accessed by perimeter or ‘rim’ drives and ventilated with intake air entering one or both sides
520
Figure 5. Variations to airflow (left) with no jet fan control at the model ramp location shown (Right).
Figure 6. Airflow controlled (left) with jet Fan variable pressure (right) to achieve 25 m3/s.
and exhausted via regulated raises in the mid-section or opposite side to a separate ventilation
level. Multiple parallel pathways create unique ventilation difficulties for control, with even
small variances in rim drive pressure, airways resistance or vehicle activities causing uneven
flow distribution. For effective jet fan control, the resistance or pressure differences must be
small, which restricts a ventilation design layout to one that has similar resistance along all
available pathways.
The concept is shown in Figure 7 where the left design has similar distances from primary
intake to exhaust regardless of the parallel path taken, while the right design has increasingly
long distances between intake and exhaust, the further away the parallel drift is from the pri
mary circuit. The variable resistance caused by different distances and therefore the large vari
ation in pressure losses means the right-hand design is likely unsuitable for low-pressure jet
fan control.
For efficiency and productivity, both sides of the cave footprint are often worked from the
same extraction drive, so a central exhaust raise and regulator must be used and adjusted to
suit activities at the time. Ideally, for two-sided access, the central exhaust would draw in
equal amounts of air from both sides however this is rarely the case because of unequal pres
sure bias between each extraction drift side. The concept is shown in Figure 8 where an intake
air drift distributes fresh air via raises and regulators onto the extraction footprint perimeter
rim drive to equalise pressure. A central exhaust drift below the extraction level draws exhaust
air through connected raises and regulators.
Methods can be implemented to control the airflow variability, including:
521
Figure 7. Schematic of parallel drift showing equal path resistance design vs unequal path resistance.
• An alternative single-sided entry and production with an exhaust raise at the end.
• Doors centrally or on both sides to control flow.
• Regulating intake flow and pressure along the perimeter rim drives.
• Over-ventilating the area to ensure minimum flows are achieved.
None of these methods however are particularly efficient or cost-effective and may impede
traffic or result in poor control of airflow and dust.
Figure 8. Conceptual block cave extraction level layout with example jet fan locations.
522
airflow into each extraction drift. A model simulation of a Howden APR 800/330 (800 mm
diameter) reversible 22 kW jet fan with a static thrust of 710 N was used to provide an esti
mated maximum 30 Pa pressure influence in a 20 m2 drift.
Network modelling of jet fan systems cannot simply use ‘fixed’ flows as this gives an unreal
istic expectation of available jet fan pressures. Instead, an automated flow-regulated jet fan
effect was modelled by using a conventional fan curve (shown in Figure 9) to apply positive or
negative jet fan pressure bias of up to 30 Pa to target a set flow point of 20 m3/s for a loader
operation. Slightly higher bias pressures were allowed outside the target range to improve
simulation convergence and reflect the change in jet fan effects in counterflow conditions.
Figure 9. Modified Drift Pressure ‘Fan’ Curve for Jet Fan Operation.
Three types of extraction drive activities were modelled within the same footprint.
• dual production requiring full airflow on both sides (20 m3/s per side, 40 m3/s exhaust)
• single-sided production requiring full airflow on one side only (20 m3/s exhaust).
• temporarily no activity with minimised flow (assumed 2m3/s exhaust leakage)
An attempt to model equal flows in extraction drifts without jet fans is shown in Figure 10.
The solid line represents a target airflow, and the columns show modelled airflow achieved.
Figure 10. Airflow modelling in extraction drifts #1 to #26 without jet fans.
523
Drifts 1-4 and 24-26 are open but not operational and require only minimal airflow, while
the remaining central drifts are in production and require 20m3/s each. Only one side of the
footprint results are shown in the figure. Airflow is well outside the target range in many
extraction drifts, with some having insufficient airflow, and others wasting too much airflow.
Further design work could be done to try to improve shortfalls by balancing regulators, how
ever this was generally found to increase air wastage in unused drifts or drifts that already
have adequate ventilation.
When jet fans are applied along each drift as shown in Figure 11, an even flow distribution
at the targeted airflow is achieved in all drifts. The required jet fan pressures (shown in the
shaded curve) vary from -20 Pa (reversed) to 10 Pa (forward), with an average jet fan influence
of 8 Pa or around 25% of maximum jet fan power. The requirement for fine adjustments of
most regulators is removed with only broad intake system adjustments required to ensure
both perimeter rim drifts have sufficient airflow, and the jet fan pressures remain operating
within their working range.
Of particular benefit is the ability of jet fans to use counter-flow to reduce airflow in tem
porarily inactive crosscuts provided some airflow remains for cooling the fans. While outside
the scope of this paper, a similar arrangement could also be applied to haulage level ventila
tion controls where multiple parallel pathways to different haulage chutes or areas could be
influenced without the need for ventilation controls and doors which impede traffic.
Figure 11. Airflow modelling in extraction drifts #1 to #26 with jet fans.
524
Blast Damage: Jet fan design is relatively simple and robust, and the fans are free hanging
and therefore not subject to stagnating blast pressures like regulators or doors. To increase
proximity distances from blasting and production on a two-side extraction footprint, instead of
using a mid-extraction drive located bi-directional fan, a single-direction jet fan could be located
at both ends of the extraction drive on the rim drive directing jet airflow into each side.
Heat and Recirculation: Generated heat from jet fans is expected to be relatively small com
pared to mobile equipment present, with an expected wet bulb increase (for example) of 0.3°C
for a 30kW jet fan at full power inducing 1m/s velocity of air in a 25m2 drift. In most cases, jet
fans will run substantially less than full power. For lower velocity ‘retarding’ operation, recir
culation and heat buildup are a concern, however, even at 0.25m/s velocity for the above
example, the temperature increase is only 1.2°C (unless operated in a recirculating restriction
mode to reduce airflow).
Cost: The electrical cost for a single 30kW jet fan operating at 25% duty at a mine electrical cost
of 15c/kWh will be $9800 per year, a modest cost to pay for the control provided, and the likely
significant saving compared to the additional airflow that would otherwise be required to ensure
minimum flow standards are met. As within auxiliary fans, occasional maintenance of blades and
motors through wear and tear will be required, however, the relatively low capital cost of jet fans
may relegate them to a disposable item after several years of use in dusty environments.
Communications: Digital communication networks and automation and a standard features
in most modern mines, and are robustly constructed to ensure any communication, automa
tion or sensors can be installed in most areas. Wi-Fi communication can be an option in par
ticularly hazardous areas where communication cables may be damaged.
3 CONCLUSION
The capabilities of jet fans for use in underground mining have likely been underappreciated,
and the application of jet fans may have significant ventilation control, cost, and safety bene
fits. The use of jet fans and ventilation control systems for road tunnels is a well-proven tech
nology but the methods have not generally been transferred to underground mines, perhaps
because of preconceived notions of the limited effectiveness of jet fans in high-pressure, high-
resistance environments.
The examples presented in this paper demonstrate the potential of jet fans in unique appli
cations which would otherwise need to be solved with over-designed ventilation airflows to
counter wastage, or with numerous, expensive, and complex ventilation controls and infra
structure. Other issues such as improved control of gases, dust and radon could also be ana
lysed. While the flexibility, effectiveness, and robustness of jet fans in these applications have
only been demonstrated as a concept with computer modelling, with further test work, control
system development, and cost analysis, these examples and many more applications may bene
fit. Jet fans have the potential to remove significant complexity and cost while improving effi
ciency, productivity, and flexibility in many underground mine applications.
REFERENCES
Drenda, J., Domagała, L., Musioł, D., Pach, G. & Różański, Z. (2018) An analysis of selected jet fans
used in chambers of KGHM mines with respect to the air stream range. Tunnelling and Underground
Space Technology, 82, 303–314.
Erleong, C. W., Mashrae Mies (2022) Enhancing the design of ventilation systems. The Singapore Engin
eer. The Singapore Engineer.
Fläktwoods (2020) Jetfoil and Large JM Aerofoil Fans for Tunnel Ventilation. FläktWoods.
Howden Ventilatoren Gmbh (2022) Air for Underground Traffic. Howden.
Konduri, I., Mcpherson, M. & Topaz, E. (1998) Model and field investigations of jet fans for face
ventilation.
Wolski, J. K. (1995) Performance analysis of the jet fan used as a booster fan. Society for Mining, Metal
lurgy, and Exploration, Inc., Littleton, CO (United . . ..
525
Occupational health and safety in mine ventilation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
S.J. Schatzel, K.M. Ajayi, Z. Khademian, R. Kimutis, M.L. Harris, M. Van Dyke &
J.D. Addis
CDC/NIOSH/PMRD/MSSB
ABSTRACT: Unconventional gas wells continue to be drilled through current and future
coal reserves in Pennsylvania, West Virginia, and Ohio. The ability for both mining and gas-
producing activities to coexist safely is a continuing question for federal and state regulatory
entities as well as for industry representatives. A hypothetical, gas well casing failure result
ing from mining-induced ground movements could produce unsafe conditions and an explo
sion hazard in nearby operating mines. The National Institute for Occupational Safety and
Health (NIOSH) is conducting research to characterize a hypothetical breach from an
unconventional gas well near an operating longwall coal mine and any resulting mine safety
consequences for a range of mining conditions. Regional conditions include overburden
depths of under 152 m (500 ft), between 152 m (500 ft) and 274 m (900 ft), and overburden
depth over 274 m (900 ft). Geologic conditions also include typical southwestern Pennsylva
nia topography and stream valley environments. Differing ventilation schemes commonly
used in Pittsburgh coal bed mines have been considered in assessing the distribution of gas
from a hypothetical breach in mine workings. Multiple technical approaches are utilized to
address the ventilation research questions. These methods include experimental techniques,
analytical methods and interpretations, and numerical and physical modeling tasks. The
NIOSH team is providing scientific input to regulatory and industry partners in the develop
ment of new recommendations for shale gas wells influenced by longwall mining.
A summary of the current findings by the research team, across all tasks, is provided.
Although this summary represents years of research by the NIOSH team, refinements to the
technical interpretations may continue to be produced over the life of the project as more
research data become available.
1 INTRODUCTION
Longwall mining induces surface and subsurface subsidence producing stress changes in the
overburden. Gas wells near longwall mining are protected by abutment pillars or barrier pil
lars. Interested parties have raised some concerns about the possibility that mining condi
tions could produce excessive stresses that result in deformation and damage to gas well
casings and suggested that regulatory guidelines need to be updated (Commonwealth of
Pennsylvania, 1957, 2017; John T. Boyd Co., 2016; Mark & Rumbaugh, 2020). NIOSH is
providing scientific findings from mining research to improve worker and safety and health
(Su et al., 2020; Schatzel & Su, 2020; Zhang & Su 2021).
DOI: 10.1201/9781003429241-54
529
2 EXPERIMENTAL METHODS
This research endeavor is a combined ground control and ventilation research effort. To be
successful, the two overall research components broadly interact as the movement of ground
produces fracturing which is a critical component of the hypothetical inflow hazard. Mining-
induced fracturing related to the full extraction mining of the coal seam produces transport
pathways for the movement of gas away from the hypothetical breach towards the operating
mine resulting in a significant safety risk. This discussion is focused on the ventilation portion
of the research with our goal of providing scientific input for the industry and operations asso
ciated with the longwall mining of coal near unconventional gas wells.
To conduct research on ventilation objectives, a series of technical approaches were devel
oped Figure 1 shows the technical approaches that continue to progress in this research. They
can be broadly categorized as methods to estimate the amount of hypothetical inflow, tech
niques to produce portrayals of movement through overburden and mine ventilation systems,
and a new methodology to distinguish gas sources. A brief summary of each of the technical
approaches is included in the subsections that follow.
The review of these approaches follows the path and chronological order of gas migrating
away from a damaged well in the event of a hypothetical casing breach. Gas is transported
away from the well through the fractured overburden for different sites, overburden depths
and stratigraphic units. The fractures in the overburden are characterized in terms of the frac
ture aperture and permeability in the strata and then an estimated inflow quantity to the
nearby mine is given. Gas movement from the hypothetical breach is simulated moving
through the ventilation system and concentrations of added gas are quantified in key areas of
the mine. Findings from NIOSH’s proposed methodology for distinguishing gas sources as
contributors to the gas emissions underground are reviewed.
530
stratigraphic unit. Falling head slug tests are performed with the addition of incremental
amounts of water and head loss rates are measured with a downhole piezometer.
2.3 Discrete Fracture Network (DFN) Fracture Flow Code (FFC) simulations
The DFN FFC analysis simulates fluid transport through a stochastic representation of fracture
systems. This technology is used to map fracture apertures in longwall gobs following a log
normal distribution. This study develops a stochastic DFN model for the site using fracture
parameters obtained from the site core log data and geomechanical analysis. Flow from
a breached gas well casing is modeled through the strata’s fracture network using cubic law. Frac
ture position data is produced from lithologic input based on information from field sites and
from DFN 3DEC results to generate 100 two-dimensional realizations (Ajayi & Schatzel, 2020).
531
2.6 Longwall Instrumented Aerodynamic Model (LIAM)
A 1:30 scale physical model of a longwall panel that replicates a cooperating Pittsburgh coal
bed mine was built. The ventilation and reservoir properties have a high degree of similitude
with full-scale mines operating in the study area. To ensure good geometric scaling, the LIAM
is built with a very high attention to detail, including shields, cribs, 3D printed shearer, regu
lators, standing supports, gob vent boreholes, and curtains. To simulate the reservoir proper
ties of the gob for the Pittsburgh seam, five types of materials having of different
permeabilities are used. The model was adapted to incorporate known quantities of dilute
sulfur hexafluoride tracer gas as a proxy for hypothetical breached gas mixed in mine air.
Flow paths can be defined, and the increase in methane (CH4) concentrations can be quanti
fied in the LIAM using this technology (Gangrade et al. 2019, 2022).
In NIOSH’s efforts to protect miners throughout the range of scenarios that may present
themselves during longwall mining near unconventional gas wells, several key, impactful vari
ables were identified as most influential. The described technical approaches were utilized to
provide scientific input for safety enhancements in the event of a hypothetical well breach.
These variables and the current technical findings are discussed below.
532
Figure 2. Panel 1 mining under deep cover, Uniontown horizon. Mine-by shown by vertical, dotted red
line. Maximum permeability during monitoring reached about 5.4 mD, purple line shortly after the face
passed the monitoring site. The undisturbed rock permeabilities are shown by the initial readings. Max
imum permeabilities are two orders lower at the deep cover site compared to the shallow cover site.
Figure 3. Panel 1 mining under shallow cover, Uniontown horizon. Mine-by shown by vertical, dotted
red line. Maximum permeability of about 246 mD, purple line. Permeabilities increased as the face
approached but reached a maximum almost two months after the face passed, possibly effected by the
stream valley location of the monitoring site.
behavior than sites near stream valleys. For more typical topography away from stream val
leys, the first panels mine-by showed permeability dropping to about 10% to 20% of the max
imum value achieved for about 610 m (2000 ft) after mine-by (Figure 4). For sites near stream
valleys, first panel mine-bys show the permeability remains in the range of 90% of the max
imum value achieved for about 610 m (2000 ft) after mine-by (Figure 5). Note there is also
a significant difference in overburden depth between the two sites which may contribute to
these findings.
533
Figure 4. Panel 1, typical regional topography, Sewickley horizon. Mine-by of longwall face shown by ver
tical, dotted, red line. Green, horizontal line shows 20% of maximum permeability value, which is close to the
post mining values during this time interval. Maximum permeability was achieved shortly after face passed.
3.3 Importance of the location of maximum horizontal movement of strata in gas transport
There is a high degree of importance regarding the location of maximum horizontal move
ment of strata in gas transport and the amount of predicted inflow quantities associated with
a hypothetical well breach. The zone of maximum horizontal movement or unconventional
subsidence was assumed to be the most likely position for the hypothetical well breach. From
our ground control research partners, it is understood that zones of maximum unconventional
subsidence are positioned where large contrasts in bending strength occur in adjacent strata.
Multiple stratigraphic zones of these high contrasts in bending strength were identified at the
study sites by monitoring ground movement, and through ground movement simulations
using numerical modeling software FLAC 3D. This software can be used to solve complex
geotechnical problems for three-dimensional analyses of rock.
These factors contributing to overburden deformation are incorporated in the overburden
fracture characterizations completed by DFN modeling. The two simulation methods por
tray the changes to vertical fracture in response to ground movement (3DEC) and to fluid
transport through fractured rock, or FFC. The combination of these DFN simulations pro
duce a simulation of inflow to an active mine in the event of a hypothetical casing breach.
The summation of these results to date has shown that, for flow simulations under inter
mediate cover (between 152 m, 500 ft and 274 m, 900 feet) and deep cover (>274 m, 900 ft)
where the zone of maximum movement was above the mining-induced fracture zone (more
than about 300 ft above the Pittsburgh seam), no hypothetical breached gas reached the
mine (Table 1).
When the horizon of maximum horizontal displacement occurs above the zone of mining-
induced fracturing (typically more than 91 m, 300 ft above the Pittsburgh seam in the region),
the likelihood of hypothetical breached gas reaching the mine is low. Research has shown the
dominant means of gas transport is through the induced fracture network formed by longwall
mining. When the induced fracture network is not in contact with the location of maximum
horizontal movement, the opportunity for appreciable gas migration from a hypothetical well
breach is limited. For simulations under intermediate cover (between 152 m, 500 ft) and deep
cover (>274 m, 900 feet) where the zone of maximum movement was above the mining-
induced fracture zone, no gas from a hypothetical well breach reached the mine (Table 2).
534
Table 1. A short summary of inflow simulations for intermediate depth (between 152 m, 500 ft and
274 m, 900 ft) at 20,700 kPa (3000 psi).
Sites Hypothetical breach location 1 borehole 10 boreholes
Tunnel Ridge 51.2 m (168 ft) below surface 0 m3/s (0 cfm) 0 m3/s (0 cfm)
NV35 120 m, 395 ft below surface <5-4 m3/s, <1 cfm Not determined
Figure 5. Sewickley horizon monitoring borehole at the, shallow, stream valley monitoring site.
Mine-by of longwall face shown by solid, red line. Dashed purple line shows 90% of maximum perme
ability value, which is close to the post mining permeability values.
An additional consideration for these unconventional well pads is the number of wells
drilled from a site. For the locations adjacent to mining, the well pads can contain different
numbers of wells. As part of this study, hypothetical well breach inflows were assessed for dif
ferent numbers of wells on a pad, both at flowing and shut-in pressures. Increasing from 1 to
10 boreholes typically increases inflow prediction quantity about 2 to 3 times the single well
rate at 20,700 kPa (3000 psi).
535
Table 2. Summary of hypothetical inflow findings for differing depths.
20,700 kPa 20,700 kPa
(3000 psi) (3000 psi)
Overburden depth 1 casing 10 casing Comments
Shallow cover sites (under 152 m, 500 ft, 0.347 m3/s, 0.672 m3/s, Maximum movement
depth of overburden cover at well site) (736 cfm)1 (1420 cfm)1 within fractured zone2
Intermediate cover sites (between 152m, 500 ft 0 m3/s, (0 0-0.00656 m3/s, Maximum movement
and 274 m, 900 ft, depth of overburden cover cfm) inflow (0–13.9 cfm) above fractured zone
at well site) inflow
Deep cover sites (over 274 m, 900 ft depth of 0–0.035 m3/ 0–0.0854 m3/s, Maximum movement
overburden cover at well site) s, (74 cfm) (181 cfm) above and within frac
inflow inflow tured zone2
1
Inflow data for shallow cover site in stream valley at 2410 kPa (350 psi)
2
Inflow data with hypothetical breach above fractured zone expected to produce different behavior
Under deep cover, quantities of inflow from a hypothetical breach are anticipated to be low
under shut-in pressures—20,700 kPa (3000 psi). No inflow quantities were predicted from
a hypothetical breach under deep cover above the mining-induced fractured zone under shut-
in pressures.
3.5 Behavior of gas transport in underground mine ventilation systems and possible input for
monitoring or mitigation
Different inflow scenarios were simulated using the model gas migration methods shown in
Figure 1. Two fundamentally different ventilation schemes were used in these simulations and
are described in Dougherty et al., 2022. If estimated inflow quantities from a hypothetical well
breach do not exceed about 0.19 m3/s (400 cfm), mitigation by the mine’s standard ventilation
system may be possible. Ventilation systems using conventional Pittsburgh coal mine strat
egies showed good potential for controlling this level of additional inflow.
In the Pittsburgh ventilation systems studied, gas from a hypothetical well breach of
0.19 m3/s (400 cfm) was primarily concentrated at the bleeder evaluation points (BEPs) and
bleeders but below statutory limits, with some limited amounts of the hypothetical breached
gas in the gob and no gas on the active longwall face (Table 3). The lower simulated inflow
quantity of 0.16 m3/s (340 cfm) showed identical distribution patterns regarding of where the
hypothetical breached gas accumulated, was absent.
A second, higher level of hypothetical breach inflow was considered. These simulations
showed 0.99 m3/s (2100 cfm) of gas entering the mine through roof fractures. Ventilation air
flow simulations for a conventional Pittsburgh coal mine and a hypothetical breach of
0.99 m3/s (2100 cfm) showed methane concentrations at elevated levels, above or near statu
tory limits in the BEPs and the bleeders. Hypothetical inflows of this magnitude or greater
will likely require mitigation to re-establish safe conditions underground.
536
Table 3. The Table shows results from the LIAM physical model for ventilation testing using tracer gas
mixture as a proxy for introduced gas from a hypothetical well breach through roof fractures. Results are
in close agreement with other simulation methods for transport in the ventilation network. Multiple tail
gate, bleeder, and gob vent borehole/gab gas venthole locations are standardized test locations in entries
or the gob.
0.19 m3/s, 400 cfm hypo
Monitoring location 0.16 m3/s, 340 cfm hypothetical inflow thetical inflow
Increased CH4 concentration due to hypothetical well breach, % CH4
how the gas moves through the mine and the magnitude of concentration change due to the
added gas from the hypothetical breach.
3.7 Methodology for distinguishing gas sources near well sites or within mines
A GC-based methodology was developed to discriminate coalbed gas, unconventional
(shale) gas, and underground storage field gas. The anticipated need to retrieve and ana
lyze samples with a relatively short turnaround time makes GC methods attractive for
this application. Gas analysis for stable isotopes of carbon and hydrogen have been used
to discriminate gas sources (Coleman, 1991). These analyses can have limitations for rou
tine, frequent sampling and analysis in this application where short turnaround times for
results and lower analytical costs for multiple samples are desirable. Bivariate plots using
the hydrocarbon index and carbon dioxide (CO2) concentration show coalbed gas plot
ting above 2% CO2 with shale gas and storage field gas below this value and below 80
on the hydrocarbon index (Figure 6).
Plots of the hydrocarbon index and CH4/CO2 ratio show coalbed gas below 80 for the CH4
/CO2 ratio with shale gas and storage field gas over 110 for the CH4/CO2 ratio and below 80
on the hydrocarbon index. A strategy for implementing this technique uses small diameter
boreholes to collect samples that can be analyzed by GC to identify incursions of non-coal gas
(Schatzel et al., 2022).
The limitations of this methodology have not been completely established. In southwestern
PA, differences in both organic matter type and rank for coal and non-coal sources create
a good theoretical basis for discrimination of the produced gases. The accumulation of more
data is in progress and will better define performance parameters for the methodology.
537
Figure 6. Graphical methods of discriminating gas sources using gas chromatography and graphical,
interpretive methods. Plots of the hydrocarbon index and CO2 concentrations (a) and CH4/CO2 ratio (b)
with potential gas sources occupying different portions of the plot.
Differences in topography at the monitoring sites were also very influential for mining-
induced fracture permeability. Where topography near the monitoring site included a stream
valley, the site retained much of the increase in permeability after the face passed, up to 90%
of the maximum value. In more typical western PA topography, away from stream channels,
induced permeability dropped to 10 to 20% of the maximum value in the time frame of moni
toring presented.
The position of maximum horizontal movement has a great effect on the amount of hypo
thetical breached gas that can reach the mine workings. Enhanced inflows were estimated at
the shallow cover, stream valley site, within the fractured zone. The intermediate and deep
cover sites showed essentially no inflows when the hypothetical breach location is above the
fracture zone.
538
Gas transport in the ventilation system for a hypothetical 0.19 m3/s (400 cfm) gas breach
showed a good possibility for managing gas within statutory limits. If a hypothetical breach
of 0.99 m3/s (2100 cfm) were to occur, this quantity would be excessive and require mitigation
to comply with statutory limits for the mine.
A methodology for distinguishing gas sources near well sites or within mines was reviewed.
A GC-based method was developed to distinguish coalbed methane, shale gas, and storage
field gas sources. The proposed method successfully distinguished the gas sources in the data
set. The goal of the ventilation research is to allow for the intersection of these two industries
to co-exist with science-based recommendations forming a framework for worker safety.
REFERENCES
Ajayi, K. M. & Schatzel S. J., 2020. Transport model for shale gas well leakage through the surrounding
fractured zones of a longwall mine. International Journal of Mining Science and Technology, DOI:
10.1016/j.ijmst.2020.05.0, published online May 29, 2020.
Coleman, D. D., 1991. The Use of Geochemical Fingerprinting to Identify Migrated Gas. 67th Annual Tech
nical Conference and Exhibition of the Society of Petroleum Engineers, Washington, DC, October 4–7,
1992.
Commonwealth of Pennsylvania, Department of Mines and Mineral Industries, Oil and Gas Division.
1957. Joint Coal and Gas Committee, Gas Well Pillar Study. Harrisburg, Pennsylvania, 28 pp.
Commonwealth of Pennsylvania, Department of Environmental Protection Oil and Gas Management,
2017. Guidelines for Chain Pillar Development and Longwall Mining Adjacent to Unconventional
Wells, Technical Guidance Document 800-0810-004, Harrisburg, Pennsylvania, 27 pp.
Dawson KJ, Istok JD, 1991. Aquifer testing: design and analysis of pumping and slug tests. Chelsea, MI:
Lewis Publishers, Inc.
Dougherty, H., Watkins, E., Kimutis, R., 2022. A network model analysis of an unconventional gas well
breach above an underground coal mine. MineXchange: 2022 SME Annual Conference and Expo, Feb
ruary 27-March 2, 2022, Salt Lake City, Utah, preprint 22–008. Englewood, CO: Society for Mining,
Metallurgy & Exploration, 2022 Feb, 33–36.
Gangrade., V., Schatzel, S. J., Harteis, S. P., Addis, J. D. 2019. Investigating the Impact of Caving on
Longwall Mine Ventilation Using Scaled Physical Modeling. Springer, Mining, Metallurgy & Explor
ation, 36(4),729–740.
Gangrade, V., Kimutis, R. A., Watkins, E. A., Schatzel, S. J., Addis, J. D., Hollerich, C. A. 2022. Simu
lating the Impact of a Shale Gas Well Breach on Longwall Mine Ventilation Utilizing a Scaled Phys
ical Model. Society of Mining, Metallurgy and Exploration Mine Expo 2022, Salt Lake City, UT,
February 27 to March 2, 2022.
Khademian, Z., Ajayi, K. M., Su, D. W. H., Schatzel, S. J., Kim, B. H., Esterhuizen, G. S. 2022. Rock
mass permeability induced by longwall mining under deep cover: potential gas in-flow from a sheared
gas well. Society of Mining, Metallurgy and Exploration Mine Expo 2022, Salt Lake City, UT, Febru
ary 27 to March 2, 2022.
John T. Boyd Co., 2016. The Pennsylvania Department of Environmental Protection’s Review of the Gas
Well Pillar Study Update. PO 4300311202 and 4300400813. September 6, 22 pp.
Mark, C., & Rumbaugh, G. M., 2020. Assessing risks from mining-induced ground movements near gas
wells. International Journal of Mining Science and Technology, 30. 11–16.
Ren, T., Balusu, R., Claassen, C., 2011. Computational Fluid Dynamics Modelling of Gas Flow Dynam
ics in Large Longwall Goaf Areas. In: 35th APCOM Symposium, Application of Computers and Oper
ations Research in the Minerals Industry, 603–613.
Schatzel, S. J. & Su, D. H. W., 2020. The origin of gas in coal and shale. In Pramod Thakur, Steven
J. Schatzel, Kashy Aminian, Gary Rodvelt, Morgan H. Mosser, Joseph S. D’Amico (eds) Coal Bed
Methane: Theory and Application, Amsterdam, The Netherlands, Elsevier, 25–34.
Schatzel, S. J., Ajayi, K., Watkins, E. A., Gangrade, V., 2022. Gas source discrimination methods to
identify the occurrence of a hypothetical, unconventional gas well breach into a nearby longwall mine.
Society of Mining, Metallurgy and Exploration Mine Expo 2022, Salt Lake City, UT, February 27 to
March 2, 2022.
Su, W. H., Zhang, P., Dougherty, H., Van Dyke, M. A., 2020. Shale gas production under the influence
of underground longwall coal production. In: Proceedings of the IV International Conference on Fossil
& Renewable Energy, Houston, TX, Feb. 17–19.
539
Watkins, E. A., Karacan, C. O., Gangrade, V., Schatzel, S. J., 2021. Assessing Gas Leakage Potential
into Coal Mines from Shale Gas Well Failures: Inference from Field Determination of Strata Perme
ability Responses to Longwall-induced Deformations. Natural Resources Research. 10.1007/s11053-
021-09859-9
Watkins, E. A., Gangrade, V., 2022. Optimization of Auxiliary Fan Placement for Large-opening Under
ground Stone Mines. Society of Mining, Metallurgy and Exploration Mine Expo 2022, Salt Lake City,
UT, February 27 to March 2, 2022.
Zhang, P., & Su, W. H., 2021. FLAC3D Modeling Evaluation of the Comprehensive NV-35 Gas Well
Pillar Study. In: Proceedings of the 55th US Rock Mechanics/Geomechanics Symposium, June 18–25,
ARMA-21-A–1193
540
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: During an underground mine fire, the presence of smoke and toxic gasses,
low visibility, and changes in the ventilation system will make it extremely difficult to identify
evacuation measures and the optimum path to safety. This paper presents an algorithm for
solving the maximum cost flow network problem (MCFNP) for fire evacuations considering
the distribution of toxic gasses inside the mine in real-time. The proposed algorithm will iden
tify the optimum evacuation paths requiring minimum decision-making time. The algorithm
holds data in nodes and arcs accumulating values at each iteration and updating the network
depending on the mine conditions. To achieve this, a fire simulation was performed in
VentSimTM software to extract the air velocity, gas concentrations, visibility, and changes in
the ventilation system data throughout the incident zone. The presented algorithm finds the
evacuation paths improving time-response and comparing the decision-making of miners in
a successful evacuation. The prediction of the concentration of toxic gasses, recommends
a safety path and avoids exposure to the danger zone despite the presence of the shortest-path
and road capacity toward the surface.
1 INTRODUCTION
Underground mine workers may encounter situations such as explosions, fires, floods, roof
falls, and other emergencies that put their lives at risk (Ronald S. Conti 2005; Mahdevari,
Shahriar & Esfahanipour 2014). One of the most dangerous dynamic situations in under
ground coal mines is fire. It is the second most frequent cause of death throughout history
with more than 734 fatalities from 1900-2006. Despite all the technological advancements, this
number is continually increasing due to fire accidents underground (Centers for Disease Con
trol and Prevention 2009; An et al. 2021). Fire incidents also generate large economic losses in
the mining industry (Saleh & Cummings 2011). Some critical impacts of fire in the under
ground environment are damage to the mine structure, potential explosion, generation of
smoke and toxic gases, visibility, and production loss (Hansen 2010). The most common cause
of death is monoxide carbon (CO) poisoning, (U.S Department of Labor 2007; Yuan, Zhou &
Smith 2016). Table 1 shows the quantity of CO, the allowable exposure time by the law, and
the symptoms produced by the gas(Centers for Disease Control (CDC) 1988; U.S Department
of Labor 2007).
Effective escape from the contaminated environment will significantly reduce the risk of
fatalities and injuries during a fire event (Adjiski et al. 2015). Mine evacuation refers to
a series of protocols and procedures that allows a mine to respond, establish control, and
escape as soon as possible from the danger (Kingery 1960). During a fire event, the mine ven
tilation system plays an important role in delivering fresh air to the miners. However, the
mine ventilation system also transports contaminants to different parts of the mine causing
contaminations in workplaces (Kingery 1960; Conti 1993). The mine ventilation network and
DOI: 10.1201/9781003429241-55
541
Table 1. Health effects from exposure to CO levels in Underground mines(U.S Department of Labor
2007).
CO levels Time exposure
Exposure level [ppm] [h] Symptoms
spatial distribution of the mine depend on the method of coal extraction, room and pillar is
a traditional method for extracting coal (Kingery 1960).
This method consists of mining the rock forming “rooms” and leaving behind “pillars” as
a support of the overburdened rock weight. These procedures form “grids” as a tunnel net
work pattern inside the mine (Khayreddine Rahab 2011). The ventilation network analysis is
concerned with the interactive behavior of airflows within connected branches of an integrated
tunnel network. Knowing how those branches are interconnected then we can predict, quanti
tatively, the distribution of airflow for given locations and duties of fans(Mcpherson 1984).
During a fire event, the carried away smoke and toxic gasses through the tunnels force the
workers to evacuate the mine or find a safe place to refuge during the incident(Onifade 2021).
Miner’s decision-making capabilities determine the success of survival and control of the evacu
ation flow (Ray & Singh 2007). However, identifying the highest concentration of the toxic gasses
is of vital importance to optimize the escape path reducing miner exposure to contaminations.
Ford-Fulkerson is a maximum flow cost optimization algorithm to find the “cheapest” possible
way of sending a certain amount of flow through a network [17]– [19]. Maximum Cost Flow
Problems are composed of a series of nodes (points) and arcs (arrows), with a beginning and an
end corresponding to the uses of graph theory (Harary & Norman 2013; Lavrov 2020).
2 LITERATURE REVIEW
542
3 METHODOLOGY
The simulation of the multivariable model has been executed using VentFIRETM to execute fire
simulation. We used the software to collect the amount of carbon monoxide (CO) produced by
the fire. The data extracted was used as input in a code developed in Python (Yuan et al. 2016).
For this paper, only the concentration of carbon monoxide produced by the fire inside the mine
was considered to identify the critical points where miners should not take during the evacuation,
and how this prediction can contribute to the decision-making. To execute the simulation, the fol
lowing assumptions were carried out to develop a realistic fire environment for accurate results:
A. Fire event: Mobile diesel equipment fire accident; mine truck MT2010 (Epiroc 2022).
B. The burning substance is diesel combined with coal; it is assumed that the mine truck was
loading around 50% of its capacity (10000 kg), and the tank was full of diesel (380 liters).
C. The fire occurred due to a vehicle rollover; the diesel spilled 50% of the tank on the surface
(An et al. 2021). The fire cannot be extinguished via an onboard suppression system or
portable extinguishers.
D. Ten miners are 1 km from the surface. The speed at which they walk is 1.1 m/s (Howden
2020).
E. Wireless Sensor Network Type: Carbon monoxide sensors. The sensors were distributed
every 100 m starting from the workplace (Conti 1993). In total, 15 sensors were put in the
VentFIRETM simulation. Refer to Figure 3.
F. The duration of the fire is 1 hour with a growth period of 30 min, the delay fire start is 5
minutes, and the growth and delay periods are 30 minutes each.
The ventilation system corresponds to an underground coal mine shown in Figure 2. The
direction of the gasses flowing south to north is for the intake air (dark blue). Thus, fresh air
is delivered through the central airways, and the contaminated air is exhausted along the side
of the coal panel (red). The conveyor is in a center airway with a brattice curtain regulating
the airflow, where the fire accident occurs (brown).
Figure 1. Underground coal mine ventsimTM model: Airway type of the mine ventilation.
Part of the mine is extracted from the whole model to have a clearer visualization of the
simulation (Figure 2). The data collected is only from the zone that covers the workplace, the
fire accident, and the exit to the surface, where the 15 sensors are located. However, the entire
ventilation system is working as a whole mine model to conserve logical results.
The ventilation system is crucial to understand how the gasses are being transported
through the fire model and predicting the hazards once the miners are notified about the fire
event. The model and the evacuation are dependent parameters. However, the velocity of the
airflow is relevant to see how fast the gas is being transported and it’s considered for the ana
lysis of the distribution of the gas.
543
Figure 2. The ventilation model was extracted to perform the fire simulation.
In this section, the multiple emergency flow route (MEFR) model is carried out. The first
flow path is the distribution of the parameters created by the fire (source 1) and how the venti
lation delivered them to the miners’ location (sink 1). The second flow path is the evacuation
of the mine from the workplace (source 2) to the surface (sink 2). The objective of MEFR,
using a Ford-Fulkerson algorithm, is to minimize the time evacuation and the risk generated
by the CO. Knowing which the maximum-cost flow is delivering through the mine, we deter
mined where is the highest concentrations of CO. To develop the mathematical model,
a generic maximum-flow algorithm is modified, to present the solution to the evacuation
problem. Given a directed graph with quantities, G = ((N, A)) be a directed graph with node
set N and edges set A, the graphic G = ((N, A)) representing the maximum capacities of the
tunnels. The capacity c(i,j) is a positive real value for each arc (i,j) ϵ A, this capacity repre
sents the quantity of the hazardous parameters. delivered into the mine or the distribution of
the fresh air. The new graph G = (N, Af) for a pre-flow fp content the new edge with the
residual values Af. The algorithm is presented in (Ford 1955; Ahuja 1987; Goldberg & Tarjan
1988; Kyi & Naing 2018; Lavrov 2020).
Figure 3. A General max-flow cost algorithm for evacuation problems (Goldberg & Tarjan 1988).
The maximum permissible limit for CO concentration in coal mines is 75 ppm for 15 min
utes without proper approved respiratory protection (U.S Department of Labor 2007). In this
study, the worst-case scenario where the miners do not have proper respiratory protection is
544
considered. Figure 4 shows the CO concentration map of the mine after the end of the fire
simulation, while Table 2 contains the data collected by the CO concentration sensors during
the simulation. The least affected part of the mine is the ‘intake airway’ part that lies north of
the fire (see sensors no. 1 and 3), as the direction of the intake air forces the fire fumes towards
the interior of the mine. At the same time, the CO concentration at the ‘return airway’ part of
the mine that lies north of the working area exceeds the permissible limit (of 75 ppm) only at
the last 10 minutes of the simulation (see sensors no. 2, 4, 8, 9, and 11). This means that the
CO concentration would allow the miners to safely evacuate moving along the return air path
towards the main exit up to 1 hr. after the fire occurs without protection.
Naturally, the area where the CO accumulates mostly and the fastest is at the site of the fire
and south of the fire: at the site of the fire the CO concentration increases from 0 to 216 ppm
within the first 10 minutes from the beginning of the fire (see sensor no. 5), while the part
exactly south of the fire exceed the safety limit within the next 5 minutes (15 minutes from the
fire – see sensors no. 6 and 7). As time passes, the CO fumes move further inside the mine and
the CO accumulation at the west part of the active coal panel (refuge chamber location) has
exceeded the safety limit between 25 and 30 minutes from the starting of the fire (see sensors
no. 10, 12, 13, and 15). This means that the workers would have approximately 20 minutes
available to reach the vicinity of the refuge chamber and approximately 35 minutes to safely
enter the chamber (without protective equipment).
Table 2. Carbon monoxide concentration results in ppm as read by the 15 sensors during 1 hr of
simulation.
Time [s] 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
300 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
600 0 0 0 0 216 57 19 0 0 0 0 0 0 0 0
900 0 0 0 0 871 287 170 0 0 4 0 0 42 0 0
1200 0 0 0 0 1960 680 476 0 0 67 0 4 228 0 43
1500 0 0 0 0 3482 1246 948 0 0 237 0 65 570 0 217
1800 0 1 0 1 5435 1955 1564 0 0 545 0 235 1071 0 541
2100 0 2 0 2 7830 2789 2307 0 0 983 1 537 1711 2 1024
2400 0 5 0 6 7845 3009 2763 4 7 1543 17 973 2483 24 1645
2700 0 22 0 26 7852 3010 2788 33 45 2089 77 1540 2723 97 2383
3000 0 74 0 85 7881 3010 2790 117 144 2238 213 2076 2737 251 2637
3300 0 181 0 202 7875 3011 2791 282 329 2266 444 2258 2740 506 2684
3600 0 357 0 392 7884 3012 2793 547 617 2273 786 2305 2741 873 2699
545
Table 3 shows the CO concentrations from the fire simulation and are compared with the
permissible exposure limit time according to MSHA(U.S Department of Labor 2007)
The data collected from VentSimTM is the flow used for each arch of the model as the cap
acity to deliver the gas. The miners’ decision-making depends on the points with the highest
concentrations of CO. As the simulation was run, the fire incident was reported after 10 min
utes (i.e., the capacity and flow data extracted from the model were taken from minute 10
onwards). After obtaining the data, we used the code to find the safest path considering the
highest concentrations of CO in the mine. In addition, based on the distance between each
airway that the miners must travel, the algorithm selects the longest routes and discards them
until the shortest one is obtained, optimizing the desired escape route.
Figure 5 shows the capacity and flow that is delivered in each tunnel A = arc (i,jn).
The sequence of graphs also shows the update of the data according to the algorithm
proposed. In addition, the intake shaft has a direction from west to east and from north
to south. To make the graph simpler and the values more visible, every two real mine
junctions V = nodes (i,jm) represents one in the proposed scheme, and the values calcu
lated between each arc will be interpolated so that the results do not change (Goldberg
& Tarjan 1988; Lavrov 2020).
Figure 5. Distribution of CO based on the Ford-Fulkerson algorithm after 20 minutes of the fire event.
4 RESULTS
The number of iterations generated by the algorithm depends on the possible number of
routes that can reach a point or sector. However, the algorithm selects all the possible paths
and accumulates the flow, and calculates the maximum flow from the destination sink.
Table 4 shows the results of the first flow path due to CO concentrations in the mine section
10 minutes after the fire alarm. It shows the dangerous points through which miners cannot
546
cross. Each color (green, blue, and yellow arrows) represents the arbitrary routes selected by
the algorithm in Figure 5 and represents the column 1,4, and 7 in Table 4, respectively. The
source 1 is the fire and the sink 1 is the workplace.
Table 4. Distribution of the CO concentrations results using the algorithm Ford- Fulkerson.
The first iteration of the flow path 1.
CO Flow CO Flow CO Flow rc Max-
1st Route [ppm] 1 2nd Route [ppm] 2 3rd Route [ppm] 3 (i,j) flow
Figure 6 shows the safest and shortest Evacuation Route (SSER) based on the information
analyzed previously. The values in green are the results of the flow after the code has per
formed in Python. The results are scaled by 1000 to understand the maximum values.
Table 5 shows the results of the safest path as a function of the distance from each branch
and the walking speed at which the miners evacuate. The time taken to exit depends on the
constant speed at which miners walk and the distance from each arch. The code selects the
minimum value of time it takes for the miners to travel from one node to the other until they
reach the exit.
547
Table 5. SSER using Ford-Fulkerson algorithm form the workplace to surface.
Accumulative time [s]
Paths Flow Time [s] Bifurcation 1 Bifurcation 2
Workplace ->1 12.2 44.6 45 45
1→2 14.7 56.4 101 101
1→3 30.39 45.5 90 90
2→4 14.7 45.5 146 146
3→4 30.39 46.3 136 136
3→5 0 45.5 136 136
4→6 9.39 28.9 165 175
4→7 35.7 57.8 194 204
5→6 0 45.5 181 181
6→9 9.38 89.1 254 270
7→9 35.7 47.2 241 241
7→8 0 46.8 241 241
8 → 10 0 30.8 272 272
9 → 10 0 46.7 288 301
9 → 11 22.5 106.3 348 361
10 → 12 0 107.6 379 396
11 → 12 7.38 45.6 393 406
11 → 13 14.9 45.9 394 407
12 → 14 7.38 46.0 425 439
13 → 14 14.9 45.9 440 440
13 → 15 0 45.3 439 452
14 → 16 7.38 29.3 455 469
15 → 16 0 45.8 485 485
16 → surface 7.38 133.4 588 618
SSER Workplace→1→3→4→6→9→11→12→14→16→Surface
5 CONCLUSIONS
This study used Ford-Fulkerson algorithm to identify the safe path to safety considering the
concentrations of CO during a fire event in an underground coal mine. We used VentSimTM
software to capture various ventilation data, including airflow and airflow direction, visibility,
oxygen level, and CO concentrations. The algorithm helps miners to make a more assertive
decision and mitigate the risks involved in the prolonged inhalation of carbon monoxide.
Improving the code can positively impact the mining industry by decreasing fatality rates and
the costs of lost miners. Similar studies can be conducted using the proposed algorithm to
identify the path to safety in various scenarios, such as roof fall, explosion, and other mine
disasters. This will allow the mine operators to utilize a data-driven approach for mine evacu
ation planning. The incorporation of the code in an interface that miners can control will help
rescue team to establish better communication with the miners involved in the fire emergency.
It should be noted that several parameters play an important role to choose the optimum
evacuation route.
ACKNOWLEDGMENTS
The project is funded by the National Institute for Occupational Safety and Health
(U60OH012351).
REFERENCES
Adjiski, V., Mirakovski, D., Despodov, Z. & Mijalkovski, S., 2015, ‘Simulation and optimization of
evacuation routes in case of fire in underground mines’, Journal of Sustainable Mining, 14(3), 133–143.
548
An, W., Li, Q., Liang, K., Yin, X. & Wang, Z., 2021, ‘Study on Combustion Characteristics and Evacu
ation during Intake Airway Fire in Coal Face under Different Ventilation Conditions’, Combustion
Science and Technology, 193(8), 1378–1399.
Brake, R., 2013, Fire modelling in underground mines using Ventsim Visual VentFIRE software, THE
AUSTRALIAN MINE VENTILATION CONFERENCE, Adelaide.
Centers for Disease Control and Prevention, 2009, ‘Underground Coal Mining Disasters and Fatalities—
United States, 1900–2006ʹ, MMWR, 57, 1–7.
Centers for Disease Control (CDC), 1988, ‘NIOSH recommendations for occupational safety and health
standards 1988.’, MMWR supplements, 37(7), 1–29.
Chavez, J.D. and H.L.H., 1993, ‘Duality Theorems for continuos analog of Ford-Fukerson Flows in
Networks’, Advances in applied mathematics, 14, 369–388.
Conti, R.S.L.C.D., 1993, Fire sensors and the detection of underground mine fires, Longwall USA confer
ence, Pittsburgh, PA.
Epiroc, 2022, Minetruck MT2010 Underground articulated truck with 20-tonne load capacity.
Ford, L.R. and F.D.R., 1955, ‘Maximal Flow through a network’, 1st edn.
Goldberg, A. V & Tarjan, R.E., 1988, ‘A New Approach to the Maximum-Flow Problem’, Journal of the
Association for Computing Machinery, 35(4), 921–940.
Hansen, R., 2010, Smoke spread calculations for fires in underground mines.
Harary, F. & Norman, R.Z., 2013, Graph Theory as a Mathematical Model in Social Science *.
Howden, B., 2020, VentSim DESIGNTM User Guide VentSim DesignTM Software, vol. Version 5.4.
Khayreddine Rahab, 2011, SURFACE AND UNDERGROUNDMINING TECHNOLOGY, Wrocław.
Kingery, D.S., 1960, Introduction to mine ventilating principles and practices., Bulleting 589, vol. 1,
Bureau of mines, Washington.
Kyi, M.T. & Naing, L.L., 2018, ‘Application of Ford-Fulkerson Algorithm to Maximum Flow in Water Dis
tribution Pipeline Network’, International Journal of Scientific and Research Publications (IJSRP), 8(12).
Lavrov, M., 2020, When augmenting paths fail Proving the residual graph theorem Max-flow algorithms.
Mahdevari, S., Shahriar, K. & Esfahanipour, A., 2014, ‘Human health and safety risks management in
underground coal mines using fuzzy TOPSIS’, Science of the Total Environment, 488–489(1), 85–99.
Mcpherson, M.J., 1984, ‘Mine ventilation planning in the 1980s’, International Journal of Mining Engin
eering, 2(3), 185–227.
Mine Safety and Health Administration, 2007, ‘Accident, illness and injury, and employment self-
extracting files (part 50 data): 2006ʹ, US Department of Labor, Mine Safety and Health Administration.
Neumann, M.M., 1984, A Ford-Fulkerson type theorem concerning vector-valued flows in infinite networks,
vol. 34.
Onifade, M., 2021, Towards an emergency preparedness for self-rescue from underground coal mines,
Process Safety and Environmental Protection, 149, 946–957.
Ray, S.K. & Singh, R.P., 2007, ‘Recent developments and practices to control fire in underground coal
mines’, Fire Technology, 43(4), 285–300.
Ahuja, R.K.& O.J.B., 1987, ‘A fast and simple algorithm for the maximum flow problem’, Massachusetts
Institute of technology, 87, 10–48.
Ronald S. Conti, L.L.C.W.J.W. and C.P.L., 2005, ‘Fire Response Preparedness for Underground
Mines’, CDC Workpalce Safety and Health, DEPARTMENT OF HEALTH AND HUMAN SER
VICES, Pittsburgh, PA.
Saleh, J.H. & Cummings, A.M., 2011, Safety in the mining industry and the unfinished legacy of mining acci
dents: Safety levers and defense-in-depth for addressing mining hazards, Safety Science, 49(6), 764–777.
U.S Department of Labor, 2007, Carbon Monoxide and Mining Health Hazard Information Card HH-28.
Wang, K., Jiang, S., Ma, X., Wu, Z., Shao, H., Zhang, W. & Cui, C., 2016, ‘Numerical simulation and
application study on a remote emergency rescue system during a belt fire in coal mines’, Natural Haz
ards, 84(2), 1463–1485.
Yang, B., Yao, H. & Wang, F., 2022, A Review of Ventilation and Environmental Control of Underground
Spaces, Energies, 15(2).
Yuan, L., Zhou, L. & Smith, A.C., 2016, ‘Modeling carbon monoxide spread in underground mine fires’,
Applied Thermal Engineering, 100, 1319–1326.
549
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Federal regulations require mines to place refuge alternatives (RAs) in under
ground coal mines in the United States. Because miners trapped in an RA could experience
heat exhaustion or heat stroke, RA regulations mandate an apparent temperature (AT) limit
of 35°C (95°F) in RAs. The National Institute for Occupational Safety and Health (NIOSH)
conducted in-mine AT tests to evaluate three RA heat mitigation devices: a borehole air
supply (BAS), a battery-powered air conditioner (BPAC), and a cryogenic air supply (CAS).
The BAS limited the AT in a 60-person built-in-place (BIP) RA to 25°C (77°F) when the out
side air temperature was 10°C–23.9°C (50°F–75°F) with relative humidity from 50%–95%.
During tests on a portable tent-type RA with a 29.4°C (85°F) mine air temperature, the
BPAC and CAS kept the AT below 35°C (95°F). AT tests showed the CAS allowed for 12
miners in a 29.4°C (85°F) 30-person BIP RA. RA manufacturers and mines can use this infor
mation to ensure RAs can meet the AT limit.
1 INTRODUCTION
Heat buildup inside a refuge alternative (RA) is a serious concern. Heat sources for an RA
include the RA occupants and the RA’s carbon dioxide (CO2) scrubber system. RAs have
a limited ability to dissipate heat due to the low thermal conductivity of mine strata materials.
If conditions become severe enough, RA occupants could experience heat exhaustion or heat
stroke (Bernard, 2011, OSHA, 2003). To guard against severe heat stress, federal RA regula
tions for underground coal mines specify a maximum apparent temperature (AT) of 35°C (95°
F). The AT is a heat index that can be used to determine how severe a thermal environment
appears (Steadman, 1979).
To ensure that the AT limit is not reached inside an RA, heat mitigation devices or occu
pancy derating could be necessary for warm or hot mines. The National Institute for Occupa
tional Safety and Health (NIOSH) examined the cooling capabilities of an air-conditioned
borehole air supply (BAS), a battery-powered air conditioner (BPAC), and a cryogenic air
supply (CAS) through RA AT testing in its Experimental Mine in Bruceton, PA. This paper
discusses the results of heat mitigation device AT testing.
DOI: 10.1201/9781003429241-56
550
Figure 1. Pictures of a BAS for RAs.
BAS can supply fresh air for 96 miners. To use the BAS, a flexible duct is connected from its
outlet to the top of a borehole.
551
In the event of a mine emergency requiring breathable air from the CAS, the following pro
cedure would be followed. First, power to the CAS would be turned off. Next a vacuum-
jacketed hose would be connected from the CAS to a heat exchanger positioned within an
RA. A pressure buildup valve on the CAS would be activated to allow vapor pressure from
the dewar to force liquid air to flow to the heat exchanger. As the liquid air passes through the
heat exchanger, it would absorb heat from the surroundings and become gaseous air at
roughly 5°C–15°C (41°F–59°F), thus providing cooling to the RA.
NIOSH researchers developed an AT test method to evaluate the thermal environment of RAs
(Yantek, 2014, 2019). This test method uses devices called simulated miners (SMs) to provide
the representative metabolic heat from actual miners and heated water tanks to provide the rep
resentative heat input from an RA’s CO2 scrubber system. The SMs and heated water tanks use
electrical resistance heaters as heat sources. The SMs also served to provide moisture input to
the RAs via vapor or simulated sweat using water from the heated water tanks. The heat input
for the SMs was set at 117 W per person, and the heat input of the heated water tanks was set
to 27.5 W per SM (Yantek, 2014, 2019). During the AT tests, the total heat input was controlled
using programmable variable autotransformers (PVAs). The heat input from the heated water
tanks was controlled via manual variable autotransformers (MVAs).
Each of the heat mitigation devices were tested using either a built-in-place (BIP) RA, port
able tent-type RA, or both. The BAS was tested on a 60-person BIP RA constructed inside
the NIOSH Experimental Mine in Bruceton, PA. The BPAC was tested on a 20-person port
able tent-type RA that was installed in a crosscut in the Experimental Mine. The CAS was
tested on the 20-person portable tent-type RA and a 30-person BIP RA.
552
volumetric flow rate of 1317 m3/hr (775 SCFM), which is slightly higher than the required
21.2 m3/hr (12.5 SCFM) of fresh air per person for 60 people. The BAS air conditioner was
set to deliver air at a dew point of 12.8°C (55°F).
The air temperatures and relative humidity (RH) levels inside the RA were measured using
Vaisala HMP155 (Vaisala, 2022b) and HMP110 Humidity and Temperature Probes (Vaisala,
2022a). For ease of reference, the RA was split into three equal-sized hypothetical sections:
Section 1, Section 2, and Section 3. One Vaisala HMP155 was placed in the center of Section 2
at mid-height, one Vaisala HMP110 was placed at the center of Section 1 at mid-height, and
one Vaisala HMP110 was placed at the center of Section 3 at mid-height. Averaging resistance
temperature detectors (RTDs) with a length of 1.2 m (4 ft) were used to monitor the air tem
perature at additional locations.
The nominal total heat input was set to 8670 W to provide the representative metabolic
heat input of 60 occupants. The SMs were evenly distributed using 6 SMs across the width of
the BIP RA with 10 SMs along the length of the BIP RA. Because borehole air supplies pro
vide enough air to dilute miner-generated CO2, heated water tanks were not needed. The
water tanks were used only to provide moisture via the SMs.
3.2 Setup for BPAC and CAS testing in 20-person portable tent-type RA
The 20-person portable tent-type RA was deployed in an entry in the Experimental Mine. Two
insulated walls were built to create an isolated test area. One wall was about 3.0 m (10 ft) from
the metal box of the RA, and the other wall was about 3.0 (10 ft) from the tent end of the RA.
The insulated-wall frames were built using 2x4 wooden studs spaced 40.6 cm (16 in) from center
to center with a layer of 50.8-cm-thick (2-in-thick) polystyrene foam attached to each side of the
wall frame. The cavities of each wall were filled with fiberglass insulation batts.
The air temperatures and RH levels inside the RA were measured using Vaisala HMP155
(Vaisala, 2022b) and HMP110 Humidity and Temperature Probes (Vaisala, 2022a). For ease
of reference, the RA was split into three equal-sized hypothetical sections: Section 1, Section 2,
and Section 3. One Vaisala HMP155 was placed in the center of Section 2 at mid-height, one
Vaisala HMP110 was placed at the center of Section 1 at mid-height, and one Vaisala
HMP110 was placed at the center of Section 3 at mid-height.
The mine air temperatures around the RA were measured using Lascar Electronics EL-
WiFi-T+ High Accuracy WiFi Temperature Data Loggers (Lascar Electronics, 2022). Four
data loggers were positioned over Sections 1, 2, and 3 of the tent and over the RA’s metal box
midway between the RA and the mine roof. Four data loggers were positioned around the
sides of the RA, and one data logger was positioned at either end of the RA. These six data
loggers were positioned at mid-height and 30.5 cm (12 in) from the RA.
The heat input was set to the nominal value corresponding to 30 occupants, 4335 W. This
value was selected because each device was designed to provide cooling for an RA with 30 occu
pants. The SMs provided a nominal 3510 W, and the heated water tanks provided a nominal
825 W. Be-cause the space and volume inside the 20-person RA is less than would be required
per RA regula-tions, these conditions are more severe than those of an actual 30-person RA.
To represent a hot mine, the mine air temperature around the RA for these tests was ele
vated to 29.4°C (85°F) using portable thermostatically controlled space heaters that were posi
tioned around the RA. The heaters were on for the entire duration of the tests. The test was
begun as soon as the desired mine air temperature was reached.
553
The air temperatures and RH values were measured using Vaisala HMP155 and HMP110
Humidity and Temperature Probes located at the centers of three equal-sized hypothetical sec
tions: Section 1, Section 2, and Section 3. One Vaisala HMP155 was placed in the center of
Section 2 at mid-height, one Vaisala HMP110 was placed at the center of Section 1 at mid-
height, and one Vaisala HMP110 was placed at the center of Section 3 at mid-height. The
mine strata surface temperatures and temperatures at depth were measured using patch-style
Class A RTDs and PVC rods with RTDs affixed.
In order to evaluate the CAS in hot-mine conditions, the interior of the BIP RA was pre
heated to a strata surface temperature of 29.4°C (85°F) using a 30-kW heater that was posi
tioned near the center of the BIP RA and six MVA-controlled baseboard heaters that were
distributed within the BIP RA. The 30-kW heater and the baseboard heaters were operated
together to rapidly increase the BIP RA air and strata surface temperatures. After the initial
preheat period, the 30-kW heater was turned off and the MVAs were iteratively adjusted so
that the mine strata temperature stabilized at about 29.4°C (85°F) for 8 hours prior to begin
ning the test.
The number of occupants that could be housed in the BIP RA with the CAS operating was
determined through a set of iterative tests. To provide cooling, liquid air was delivered to the
BIP RA from the CAS to a heat exchanger located inside the BIP RA. Because it was assumed
that CO2 scrubbers would be necessary for a CAS, the total heat input for each test was the
number of SMs used times 144.5 W, which includes 117 W to represent metabolic heat input
and 27.5 W to represent CO2 scrubber heat input. The liquid airflow rate from the CAS was
kept roughly constant. Small adjustments to the flow rate were made in an attempt to keep
the AT below 35°C (95°F) and to ensure that the liquid air would last for 96 hours.
3.4 AT calculation
To develop an equation to calculate the AT, or heat index, using the dry-bulb air temperature
and relative humidity, Rothfusz carried out a multiregression analysis on the tabular data
from Steadman (Rothfusz, 1990). The National Weather Service refined this approach to cal
culate the AT. For dry-bulb air temperatures above 26.7°C (80°F) and an RH above 13%, the
AT was calculated using (National Weather Service, 2023):
where T is the dry-bulb air temperature in °F amd RH is the percent relative humidity. If
Equation 1 resulted in an AT below 26.7°C (80°F), the following simplified formula was used
to calculate AT (National Weather Service, 2023)
4 RESULTS
554
(48.6°F–77.5°F) with an average of 17.7°C (63.8°F), and the outside RH ranged from 36.8%–
95.2% with an average of 72.0%.
Figure 4. (a) Mine air average temperature and (b) RA interior temperature and %RH for BPAC AT
test.
555
Figure 5. (a) Average mine air temperature and (b) average RA interior temperature, %RH, and AT
for CAS AT test.
Table 1. Summary of CAS AT test results for a 29.4°C (85°F) 30-person BIP RA.
Parameter Test 1 Test 2 Test 3 Test 4
Number of SMs 24 16 12 12
Test Duration 96 hrs 96 hrs 96 hrs 8 hrs
CAS used Yes Yes Yes No
Time when CAS empty 91 hrs 96 hrs 91 hrs –
AT when CAS empty 37.4°C (99.4°F) – 30.3°C (86.6° F) –
48.1°C 35.2°C 37.9°C 40.7°C
AT at end of test (118.6°F) (95.3°F) (100.3°F) (105.2°F)
The CAS kept the AT below 35°C (95°F) with the heat input corresponding to 12 miners, but
at 91 hours, the CAS ran out of liquid air. At this time, the AT was 30.3°C (86.6°F). For the
test without the CAS with the heat input of 12 miners, the AT reached 40.7°C (105.2°F) after
just 8 hours.
5 DISCUSSSION
The BAS was able to keep the AT well below the AT limit. In addition, the BAS had no perform
ance issues during testing. However, the outside atmospheric conditions were not as severe as
desired, as the dry-bulb air temperature was a maximum of only 25.8°C (77.5°F). Ideally, the
BAS would be evaluated with an outside air temperature matching that of southern mines, such
as the ones in Alabama, which are hotter than other United States underground coal mines. In
Brookwood, AL, July is the hottest month with an average high temperature of 33.1°C (91.6°F)
and an average RH of 73% (Weather U.S., 2022). Furthermore, to ensure adequate cooling in
Alabama-mine-like conditions, the BIP RA should be preheated to 29.4°C (85°F) as was done
with the evaluations of the BPAC and the CAS.
The BPAC was able to keep the AT below the limit, but higher than expected power con
sumption caused issues with the unit’s power system. Because the device uses a compressor
that adjusts to the cooling demand, and the cooling demand was greater than expected, the
power consumption exceeded the capacity of the batteries and of the circuit breaker. In order
to ensure the device could last for an entire 96 hours, the battery capacity would have to be
increased. Under the test conditions here, the unit operated for about 60 hours. Therefore, the
battery capacity would need to be increased by about 60% to ensure the BPAC could operate
for 96 hours. The BPAC would need to satisfy federal requirements for permissibility in under
ground coal mines. As tested, the BPAC was not evaluated subject to these requirements.
556
The CAS was able to provide enough cooling to keep the 20-person portable RA below the
AT limit, but it had issues with ice buildup on the heat exchanger, and it ran out of liquid air
prior to the end of the 96-hour test. The heat exchanger design must be capable of having ice
buildup removed easily without risking damage. For the 20-person portable RA test, the CAS
2000-liter dewar had about 1850 liters of liquid air inside at the beginning of the test. During the
test, the CAS was set to deliver liquid air to the heat exchanger inside the tent at a nominal rate
of 220 liters/min. However, the CAS had limited flow rate control capability, so the actual flow
rate exceeded the target amount. Better flow control capability would be needed for a final
product.
The CAS tests on the 30-person 29.4°C (85°F) BIP RA showed promising results. With the
heat input of 24 occupants, the unit was able to keep the AT below 37.8°C (100°F) for 91 hours.
A small increase in liquid airflow rate and a larger volume of liquid air would have allowed the
unit to provide enough cooling to bring the AT below 35°C (95°F) and to last the full 96 hours
with 24 occupants. The tests with the heat input of 16 occupants showed the CAS would have
kept the AT below 35°C (95°F) with a slight increase in liquid airflow rate. The tests highlighted
the importance of having good flow control.
One of the challenges with the CAS is that this is a novel technology for mining applica
tions. Because gas suppliers do not currently offer liquid air, mines would have to mix their
own liquid air, or a supplier would need to be convinced to provide this commodity. Another
challenge is the maintenance of a CAS in a mining environment. This would require the devel
opment of a maintenance schedule and instructions. It is likely that a vendor would need to
provide maintenance services.
6 CONCLUSIONS
All three devices were able to reduce heat buildup within the RAs. The BAS would need to be
tested in more severe thermal conditions like those in southern mines to ensure the device would
be capable of keeping the apparent temperature (AT) below 35°C (95°F). Both the BPAC and
CAS were able to keep the AT under 35°C (95°F), but both devices had issues. The BPAC needs
improved battery capacity. In addition, the BPAC would need to pass permissibility testing. The
CAS would benefit from a heat exchanger design that is less prone to ice buildup and has
improved flow control.
To ensure hot mines can be kept under the AT limit without occupancy derating, cooling
devices would need to be used with RAs. The devices tested here could all serve this need.
The BAS is closest to being a usable product. The BPAC and CAS need some additional
improvements before they are ready to be used to provide cooling to RAs in productions
mines.
7 LIMITATIONS
The heat mitigation devices were evaluated only in the NIOSH Experimental Mine. If an
actual mine has less conductive strata, the resulting ATs would be higher. The BAS evaluation
was not conducted with hot outside ambient conditions or a preheated mine, both of which
would be more severe than the tested conditions. All of the devices described in this publica
tion would require additional development prior to implementing them in commercial mines.
8 DISCLAIMER
The findings and conclusions in this report are those of the author(s) and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health,
lefts for Disease Control and Prevention. Mention of any company or product does not con
stitute endorsement by NIOSH.
557
REFERENCES
Bernard, T. E. 2011. Physiological analysis of human generated heat in a refuge alternative. NIOSH
Contract Report 254-2011-M-40932. Tampa, FL, University of South Florida.
CFR. 2018. Code of Federal Regulations, 30 CFR 7.506, Breathable air components. Washington, DC,
U.S. Government Publishing Office. https://www.govinfo.gov/content/pkg/CFR-2018-title30-vol1/pdf/
CFR-2018-title30-vol1-sec7-506.pdf
CFR. 2019. Code of Federal Regulations, 30 CFR 75.1506, Refuge alternatives. Washington, DC, U.S.
Government Publishing Office. https://www.govinfo.gov/content/pkg/CFR-2019-title30-vol1/pdf/
CFR-2019-title30-vol1-sec75-1506.pdf
CFR. 2021. Code of Federal Regulations, 30 CFR 7.508, Harmful gas removal components. https://
www.govinfo.gov/content/pkg/CFR-2021-title30-vol1/pdf/CFR-2021-title30-vol1-sec7-508.pdf.
Cryo Life Support Systems. 2016. Refuge alternative heat mitigation system utilizing advanced liquid air
technologies. Broad Agency Announcement (BAA) Contract 200-2016-91194. Pittsburgh, PA, USA,
the National Institute for Occupational Safety & Health. https://www.cdc.gov/niosh/mining/research
program/contracts/contract_200-2016-91194.html.
DRS. 2016. Development of Heat Mitigation Systems for Refuge Alternatives (ras) used in Underground
Coal Mines. Broad Agency Announcement (BAA) Contract 200-2016-91901. Pittsburgh, PA, USA, the
National Institute for Occupational Safety & Health. https://www.cdc.gov/niosh/mining/researchpro
gram/contracts/contract_200-2016-91901.html.
Lascar Electronics. 2022. Lascar Electronics EL-WiFi-T+ High Accuracy WiFi Temperature Data
Loggers. Accessed on 1/13/2023. https://www.lascarelectronics.com/easylog-el-wifi-tplus.
National Weather Service. 2023. The Heat Index Equation. Weather Prediction Center. Accessed on 1/
13/2023. https://www.wpc.ncep.noaa.gov/html/heatindex_equation.shtml.
OSHA. 2003. 1910.146 - permit-required confined spaces. Occupational Safety and Health Administra
tion (OSHA). https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.146.
Rothfusz, L. P. 1990. The heat index equation (or, more than you ever wanted to know about heat
index). Fort Worth, Texas: National Oceanic and Atmospheric Administration, National Weather Ser
vice, Office of Meteorology, SR 90–23. https://www.weather.gov/media/ffc/ta_htindx.PDF.
Steadman, R. G. 1979. The assessment of sultriness. Part I: A temperature-humidity index based on human
physiology and clothing science. Journal of Applied Meteorology and Climatology, 18, 861–873. https://
journals.ametsoc.org/view/journals/apme/18/7/1520-0450_1979_018_0861_taospi_2_0_co_2.xml.
Vaisala. 2023a. Humidity and Temperature Probe HMP110. Accessed on 1/13/2023. https://www.vaisala.
com/en/products/instruments-sensors-and-other-measurement-devices/instruments-industrial-measure
ments/hmp110.
Vaisala. 2022b. Humidity and Temperature Probe HMP155. Accessed on 1/13/2023. https://www.vaisala.
com/en/products/weather-environmental-sensors/humicap-humidity-temperature-probe-hmp155.
Weather U.S. 2022. July weather forecast Brookwood, AL. Accessed on 1/13/2023. https://www.weather-
us.com/en/alabama-usa/brookwood-weather-july#temperature.
Yan, L., Fernando, R., Yantek, D., Carr, J., Reyes, M., Degennaro, C., Yonkey, J. & Srednicki, J. 2021.
Storage time and venting characteristics for cryogenic air supplies on cryocooler shutdown. The 21st Inter
national Cryocooler Conference. Boulder, CO. https://cryocooler.org/resources/Documents/C21/085.pdf.
Yan, L., Yantek, D., Lutz, T., Yonkey, J. & Srednicki, J. 2017. Heat mitigation for underground coal
mine refuge alternatives using cryogenic air or borehole air supplies. ASME International Mechanical
Engineering Congress and Exposition. American Society of Mechanical Engineers.
Yan, L., Yantek, D., Reyes, M., Whisner, B., Bickson, J., Srednicki, J., Damiano, N. & Bauer, E. 2020.
Cryogenic air supply for cooling built-in-place refuge alternatives in hot mine. Mining, Metallurgy &
Exploration, 37, 861–871.
Yantek, D. 2014. RI 9695 - Investigation of temperature rise in mobile refuge alternatives. Publication
No. 2014–117, US Department of Health and Human Services, Centers for Disease Control and Pre
vention. National Institute for Occupational Safety and Health, Office of Mine Safety and Health
Research. https://www.cdc.gov/niosh/mining/UserFiles/works/pdfs/2014-117.pdf.
Yantek, D.S., Yan, L., Damiano,N.W.,Reyes, M.A., and Srednicki, J.R.. 2019. A test method for evalu
ating the thermal environment of underground coal mine refuge alternatives. International Journal of
Mining Science and Technology 29(3): 343–355. https://www.ncbi.nlm.nih.gov/pmc/articles/
PMC9278555/.
558
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
1 INTRODUCTION
On December 31, 2008, the Mine Safety and Health Administration (MSHA) published its
final rule “Refuge Alternatives for Underground Coal Mines” (Federal Register, 2008), which
became effective on March 2, 2009. This rule mandates underground coal mines to install
refuge alternatives (RAs) so that all miners, contractors, and visitors have access to an RA in
case of an inescapable mine emergency. The RA regulations require RAs to provide
a breathable air environment, food and water, and supplies needed to sustain trapped miners
for up to 96 hours. The regulations also provide requirements for the breathable air environ
ment, available space and volume, and thermal conditions.
Due to human metabolic heat, carbon dioxide (CO2) scrubber system heat, and a limited
ability to dissipate heat, the thermal environment of an occupied RA can become severe
enough to cause heat exhaustion or heat stroke, which can be deadly (OSHA, 2003, Bernard,
2011). RA regulations for U.S. underground coal mines specify a maximum apparent tem
perature (AT) of 35°C (95°F) (Steadman, 1979). For indoor environments without solar load
ing, the AT is calculated using dry-bulb temperature and percent relative humidity (%RH)
(Rothfusz, 1990). The regulations require RA manufacturers to conduct testing to determine
an RA’s maximum occupancy without exceeding the AT limit.
Many factors affect the resulting RA thermal conditions including the number of RA occu
pants, the mine air and strata temperature, and the mine strata thermal properties (Yan et al.,
2020, Yantek et al., 2017). As the number of occupants increases, the heat input also increases,
and the resulting RA thermal conditions become more severe. Higher initial mine air and
strata temperatures also increase the severity of an occupied RA’s thermal conditions. Mine
strata materials have low thermal conductivity, which limits conductive heat transfer from an
RA to the mine.
DOI: 10.1201/9781003429241-57
559
To ensure that the AT limit is not reached, heat mitigation devices or occupancy derating
could be necessary for warm or hot mines. The National Institute for Occupational Safety and
Health (NIOSH) used test-result-validated RA thermal simulation models with actual mine
temperatures, %RH values, and strata compositions to examine occupancy derating for two
portable RAs: a 23-person portable tent-type RA and a 6-person portable steel RA. The
stated occupancies are based on the space and volume requirements of the RA regulations for
underground coal mine RAs (CFR, 2019). This paper discusses the results of the occupancy
derating simulations.
The concept of occupancy derating of an RA is to limit its occupancy to less than full capacity
in order to ensure that the conditions within the RA remain below the 35°C (95°F) AT limit.
The resulting thermal conditions within an RA are a function of the number of occupants, the
RA design, the mine strata thermal properties, and the mine air and strata temperatures. For
a given RA, it would be impractical to conduct testing to represent the conditions in all mines.
In order to examine the need for occupancy derating, NIOSH used validated thermal simula
tion models (refer to Figure 1) of a 23-person portable tent-type RA and a 6-person portable steel
RA that were tested in the NIOSH Experimental Mine (EM) (Yan et al., 2016a, Bissert et al.,
2016, Yan and Yantek, 2018, Yan et al., 2020, Klein et al., 2017, Klein and Hepokoski, 2017)
(Yan et al., 2016b) (Klein, 2017a) (Klein, 2017b). In these models, the mine strata were modeled
as twenty-four 7.62-cm-thick (3-in-thick) layers up to a modeling depth of 1.83 m (6 ft). The
models included transient heat conduction into the mine roof, rib, and floor. The mine strata ther
mal properties for each mine were used for the layers in the models. Human thermal models
(HTMs) were used to input the heat of actual occupants (Bernard et al., 2018, Klein and Hepo
koski, 2017). The HTMs model the complex nature of heat transfer from the human body and
account for all manners of body heat transfer including vasodilation, perfusion, and evaporation
of sweat. The CO2 scrubber system heat input was set to 27.5 W per person (Shumaker, 2013).
For each RA, the mine dry-bulb air temperature, %RH, and mine strata temperatures were used
as initial conditions. A 96-hour transient thermal analysis was conducted with the maximum RA
occupancy to determine the final RA air temperature and %RH. The AT limit was calculated
from the predicted RA air temperature and %RH (Rothfusz, 1990). If the AT limit was reached,
the number of HTMs was reduced and the analysis was repeated. The maximum number of occu
pants was then found by iteratively increasing or decreasing the number of HTMs.
Figure 1. Thermal simulation models used for occupancy derating analysis (a) 23-person portable
tent-type RA and (b) 6-person portable steel RA.
Initially, NIOSH examined occupancy derating for both RAs based on the EM. This ana
lysis was conducted using the thermal simulation models of the RAs with the mine strata ther
mal properties and thermal conditions of the EM. The EM floor strata was modeled as
560
a 15.2-cm-thick (6-inch-thick) layer of concrete over 168 cm (66 in) of siltstone, the EM rib
strata was modeled as a 2.5-cm-thick (1-inch-thick) layer of shotcrete over bituminous coal,
and the EM roof strata was modeled as 31 cm (12 in) of slate, 61 cm (24 in) of bituminous
coal, and 91 cm (36 in) of shale. To simplify the model, the 2.5-cm-thick (1-in-thick) layer of
shotcrete on the EM roof was ignored in the analysis. Thermal simulations were conducted
across a range of initial mine air, mine strata, and RA temperatures from 12.8°C–23.9°C (55°
F–75°F) in 2.8°C (5°F) increments. The initial %RH for these simulations was 85% based on
measured data from the RA tests in the EM.
Next, NIOSH examined occupancy derating for the portable tent-type and metal RAs for
five U.S. mines located in the Southwestern, Midwestern, Southern, Western, and Eastern
regions of the U.S. NIOSH collaborated with these mines to determine the mine temperatures,
%RH values, and mine strata properties to be used in the simulations (Bissert et al., 2016,
Yantek et al., 2017). The mine air temperature, %RH, and strata temperatures near existing
RAs were measured by mine personnel using air-temperature-measurement resistance tem
perature detector (RTD) probes, a %RH instrument, RTD-instrumented PVC rods (refer to
Figure 2), and RTD thermometers provided by NIOSH. The mine strata temperatures were
measured at the surface and at depths of 0.61 m (24 in) and 1.22 m (48 in). Mine personnel
documented the temperatures at least once per month for an entire year. The worst-case com
bination of mine air temperature, %RH, and mine strata temperatures were used as initial
conditions in the thermal simulations (see Table 1). Through conversations with mine person
nel and NIOSH ground control researchers, the mine strata composition and layer thicknesses
were assumed based on the geographic location of each mine (see Table 2). The strata proper
ties used in the thermal simulations are shown in Table 3.
Figure 2. (a) RTD-instrumented PVC rod for measuring strata temperatures and (b) picture showing
RTD-instrumented rod installed into the mine rib.
3 RESULTS
561
Table 1. Mine dimensions and worst-case combinations of %RH, mine air temperature, and mine
strata temperatures for five mines across the U.S.
Strata Temperatures
Mine
Mine Air Air % 0.61 m (24 1.22 m (48
Mine Width Height Temp. RH Surface in) Depth in) Depth
Table 2. Mine strata composition of five mines used in occupancy derating study. Note: If only one mater
ial is listed, the entire 1.8-m-thick (6-ft-thick) strata layer is assumed to consist of the listed material.
Strata Composition
Mine Floor Ribs Roof
Table 3. Mine strata thermal properties used in the occupancy derating studies.
Thermal Conductivity Specific Heat
Material Density (kg/m3) (W/m-K) (J/kg-K)
Note: For measured values, core samples collected by NIOSH were tested by Decagon Devices, Inc., 2365 NE
Hopkins Court, Pullman, WA 99163.
The analysis of the 6-person portable steel RA with the EM strata properties also showed
that occupancy derating would be necessary for an initial temperature of 21.1°C (70°F) or
greater (see Table 4). At an initial temperature of 21.1°C (70°F), the occupancy would have to
562
be decreased from 6 to 4, a 33% reduction. For an initial temperature of 23.9°C (75°F), the
occupancy would have to be reduced from 6 to 2, a 67% reduction.
Table 4. Derating results for 23-person portable tent-type RA from simulations of the EM. Red indi
cates the AT exceeded 35°C (95°F). Note: The initial %RH for these simulations was 85 %RH.
Number of Final RA Air
Initial Temperature Occupants Temperature Final RA %RH Final AT
Table 5. Derating results for 6-person portable steel RA from simulations of the EM. Red indicates the
AT exceeded 35°C (95°F). Note: The initial %RH for these simulations was 85 %RH.
Number of Final RA Air
Initial Temperature Occupants Temperature Final RA % RH Final AT
563
Table 6. Derating results for 23-person portable tent-type RA from simulations of five mines across the
U.S. Red indicates the AT exceeded 35°C (95°F).
Number
Initial AirTem Initial of Occu Final AirTem Final
Mine perature % RH pants perature % RH Final AT
Table 7. Derating results for 6-person portable metal RA from simulations of five mines across the U.S.
Red indicates the AT exceeded 35°C (95°F).
Number
Initial AirTem Initial % of Occu Final RA Final
Mine perature RH pants AirTemperature % RH Final AT
4 DISCUSSION
The derating simulation results show that the occupancy derating percentages depend on
the initial mine temperature and mine strata composition based on geographical location.
For the portable tent-type RA, the predicted AT was just under the AT limit when the
mine initial temperature was 18.3°C (65°F) for the EM and for the Southwestern mine. If
the mine strata thermal conductivity was slightly lower, the AT limit would have been
reached because the final RA air temperature would have been higher. At an initial mine
temperature of 21.1°C (70°F), the portable tent-type RA needed to be derated by 30% and
the portable metal RA needed to be derated by 33% for the EM. In addition, for the Mid
western mine with a mine initial temperature of just under 21.1°C (70°F), the portable
tent-type RA needed to be derated by 10% and the portable metal RA was almost at the
AT limit. For the EM, at an initial mine temperature of 23.9°C (75°F), the portable tent-
type RA and portable metal RA would have to be derated by 60% and 67%, respectively.
For the EM at an initial mine temperature of 26.7°C (80°F), using the dry-bulb tempera
ture rise per occupant and relative humidity for the 23.9°C (75°F) initial mine temperature,
the occupancy of the portable tent-type RA would have to be derated by 90% and the port
able steel RA would not be able to be used without cooling. For the Southern mine, which
had an initial mine temperature of about 26.7°C (80°F), the derating analysis showed that
the portable tent-type RA would have to be derated by 96% and the portable metal RA
would require cooling.
564
Based on the occupancy derating analysis results for the EM and the actual mines, a table
of expected percent occupancy derating needed for compliance was developed (refer to
Table 7). In the development of this table, we considered the results from more than 30 in-
mine tests of the portable 23-person tent-type and 6-person metal RAs. In addition, we con
sidered simulation results that examined the resulting thermal conditions for mine strata
compositions having “low” to “high” thermal conductivity and a range of mine entry sizes
(Bissert et al., 2016). Furthermore, we considered simulation results that examined how ini
tial mine air temperatures, initial mine strata surface temperature, and initial mine strata
temperature at depth affects the resulting thermal conditions (Yantek et al., 2017). To
account for variations in mine strata composition, entry size, and RA thermal properties,
the lowest percent occupancy reductions from the analysis were decreased slightly, and the
highest percent occupancy reductions from the analysis were increased slightly. This table
should not be interpreted as a rigid set of percent occupancy reductions as it is based on
thermal simulations and has not been validated by testing. Furthermore, the table is based
on results from only one portable tent-type RA and one portable metal RA and a few mines
with assumed mine strata composition based solely on geographical location. However,
Table 7 can be used as a guide to provide an expected possible range of percent occupancy
reductions. It should be noted that above 26.7°C (80°F), it is likely that cooling or dehumidi
fication would be required because the buildup of moisture inside an RA would likely result
in a relative humidity of greater than 90%. At this relative humidity, the 35°C (95°F) AT
limit would be reached at an RA air temperature of 28.3C (83°F). For initial mine temperat
ures of 29.4°C (85°F), the AT limit would be reached for a relative humidity of 75%. Based
on all NIOSH testing and thermal simulations, the relative humidity would exceed 90%.
Therefore, cooling or dehumidification would be necessary for mine air temperatures above
29.4°C (85°F).
Table 8. Approximate percent occupancy derating needed for a 23-person portable tent-type
RA and a 6-person portable steel RA.
Range of Initial Mine Temperatures % Occupancy Derating Needed for Compliance
5 CONCLUSIONS
Although it would be impractical to conduct testing for all possible mine temperatures and
mine strata compositions, our analysis results can be used as a rough guide for planning pur
poses. In order to ensure compliance with the 35°C (95°F) AT limit, testing or simulations
using validated thermal simulation models should be used to estimate the final AT, which pri
marily depends on initial mine temperature and mine strata composition.
In the absence of cooling or dehumidification, our analysis indicates that occupancy derat
ing of RAs may be required for some mines. For mines below 18.3°C (65°F) with very low
mine strata thermal conductivity, the occupancy may need to be slightly reduced. For mines
between 18.3°C–21.1°C (65°F–70°F), occupancy derating would be necessary for mines with
strata having low thermal conductivity. From 21.1°C–26.7°C (70°F–80°F), the results indicate
that occupancy derating would be necessary. Above (80°F), occupancy derating would
approach 100%, and cooling or dehumidification would be needed to ensure that the thermal
conditions inside an RA stay below the 35°C (95°F) AT limit.
565
6 LIMITATIONS
Because the analysis presented here was performed on only two RAs with conditions from only
five mines across the U.S. and one nonproduction mine, the results should not be assumed to
directly apply to mine-specific RA installations. Additional testing or analysis would be neces
sary to determine occupancy derating percentages for a specific RA installation.
REFERENCES
Bernard, T., Yantek, D. & Thimons, E. 2018. Estimation of metabolic heat input for refuge alternative
thermal testing and simulation. Mining Engineering, 70, 50.
Bernard, T. E. 2011. Physiological analysis of human generated heat in a refuge alternative. NIOSH Con
tract Report 254-2011-M-40932. Tampa, FL, University of South Florida.
Bissert, P., Yantek, D., Klein, M. & Yan, L. 2016. Analysis of heat loss mechanisms for mobile tent-type
refuge alternatives. Transactions of Society for Mining, Metallurgy, and Exploration, Inc, 340, 70.
CFR. 2019. Code of Federal Regulations, 30 CFR 75.1506, Refuge alternatives. Washington, DC, U.S.
Government Publishing Office. https://www.govinfo.gov/content/pkg/CFR-2019-title30-vol1/pdf/
CFR-2019-title30-vol1-sec75-1506.pdf accessed on 2/24/2023.
Federal Register. 2008. Refuge Alternatives for Underground Coal Mines; Final Rule, Department of
Labor, Mine Safety and Health Administration, 30 CFR Parts 7 and 75. Washington, DC, US Gov
ernment Printing Office.
Klein, M. 2017a. Mine shelter thermal analysis - 23-person inflatable RA TAITherm model validation and
analysis. Report 3018–001, Revision 4.1. ThermoAnalytics, Inc., 23440 Airpark Blvd., Calumet, MI.
Klein, M. 2017b. Mine shelter thermal analysis - steel RA TAITherm model validation and analysis.
Report 3090–001, Revi-sion 4.1. ThermoAnalytics, Inc., 23440 Airpark Blvd., Calumet, MI.
Klein, M. & Hepokoski, M. 2017. Human thermoregulation model for analyzing the performance of
mine refuge alternatives. Mining Engineering, 69.
Klein, M., Yantek, D., Hepokoski, M. & Yan, L. 2017. Prediction of human core temperature rise and
moisture loss in refuge alternatives for underground coal mines. Transactions of Society for Mining,
Metallurgy, and Exploration, Inc, 342, 29.
OSHA. 2003. Permit-required confined spaces. Department of Labor, Occupational Safety and Health
Administration, https://www.osha.gov/laws-regs/regulations/standardnumber/1910/1910.146 accessed
on 2/24/2023.
Rothfusz, L. P. 1990. The heat index equation (or, more than you ever wanted to know about heat
index). Fort Worth, Texas: National Oceanic and Atmospheric Administration, National Weather Ser
vice, Office of Meteorology, SR 90–23. https://www.weather.gov/media/ffc/ta_htindx.PDF accessed on
2/24/2023
Shumaker, W. A. 2013. Information relayed from Wesley Shumaker, Mechanical Engineer, MSHA
Approval and Certification Center, to NIOSH Pittsburgh Mining Research Division regarding the
heat used to represent a refuge alternative CO2 scrubber system during heat and humidity testing. U.S.
Department of Labor, Mine Safety and Health Administration, Approval and Certification Center.
Triadelphia, WV.
Steadman, R. G. 1979. The assessment of sultriness. Part I: A temperature-humidity index based on
human physiology and clothing science. Journal of Applied Meteorology and Climatology, 18, 861–873.
Yan, L. & Yantek, D. 2018. Portable refuge alternatives temperature and humidity tests. Mining engin
eering, 70, 43.
Yan, L., Yantek, D., Klein, M. & Bissert, P. 2016a. Interior thermal environment of a 6-person metal-
type refuge alternative (RA). ASME International Mechanical Engineering Congress and Exposition.
Phoenix, Arizona, USA, American Society of Mechanical Engineers.
Yan, L., Yantek, D., Klein, M., Bissert, P. & Matetic, R. 2016b. Validation of temperature and humidity
thermal model of 23-person tent-type refuge alternative. Mining engineering, 68, 97.
Yan, L., Yantek, D. & Reyes, M. 2020. Underground mine air and strata temperature change due to the
use of refuge alternatives. Mining, Metallurgy & Exploration, 37, 773–781.
Yantek, D., Yan, L., Bissert, P. & Klein, M. 2017. Effects of mine strata thermal behavior and mine ini
tial temperatures on mobile refuge alternative temperature. Mining engineering, 69, 41.
566
Renewable / alternative energy in mine ventilation
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
M. Mohit
McGill University, Montreal, Quebec, Canada
S. Akhtar
Polytechnique Montreal, Montreal, Quebec, Canada
ABSTRACT: Spray freezing technology has been shown its exceptional efficiency, safety, and
sustainability for underground mine heating. Designing spray freezing systems requires
a mathematical model with the rigorous formulation and fast computation methods, capable of
predicting performance indicators. Existing models for spray freezing often take the droplet
motion and velocity distribution as a priori, thus making it less feasible in practice. In the present
work, a novel reduced-order model is developed to calculate the distributions of droplet velocity
and residence time for various spray configurations (namely, flat fan, hollow cone, and full cone)
and droplet diameters. The velocity distributions are then incorporated into a robust heat transfer
model for mine heating to improve the prediction of the droplets freezing time and overall heat
transfer rate (HTR). Consequently, the residence time distribution is used along with droplets
diameter distribution to calculate the ice packing factor (IPF) and cooling capacity of the system.
1 INTRODUCTION
In recent years, studying different clean technologies for providing the heating demand of
underground mines has attracted researchers’ attention due to the importance of employing
eco-friendly systems in mining industries. Spray freezing technology as a renewable and effi
cient method for air heating in under-ground mines of sub-arctic regions is one of these tech
nologies. In this method, the sub-arctic air stream is heated by using the water latent heat of
solidification. An example of spray freezing application for underground mine heating has
been introduced by Stachulak (1991). Through spraying water into the cold air stream, water
droplets freeze, and the air temperature increases. The schematic of the spray freezing setup is
illustrated in Figure 1 (a). To efficiently evaluate the performance of spray freezing setups, it
is necessary to develop mathematical models with high accuracy and low computational cost.
The two important aspects of spray freezing modeling are the droplet freezing and droplet
dynamics. Droplet freezing process has been the topic of study by many scholars due to its
importance from both the theoretical and practical point of view. The solid-liquid phase
change phenomenon of water droplets is categorized under the subject of Stefan problem by
many scholars. Various numerical, analytical, asymptotic, and hybrid methods have been
developed in the literature to solve the Stefan problem in different conditions and coordinates
(Rubinšteĭn, 2000; Meirmanov, 2011). One powerful method to investigate Stefan problems
subject to various boundary conditions is the perturbation series analysis which has been used
in many works (Caldwell and Kwan, 2003; Akhtar, Xu and Sasmito, 2021b). It should be
DOI: 10.1201/9781003429241-58
569
noted that most results on droplet solidification modeling do not take nucleation and den
dritic growth stages into account. This shortcoming has been addressed by Akhtar, Xu and
Sasmito (2021a) by developing a reduced-order semi-analytical model for droplet solidifica
tion that can capture all freezing stages. On the other hand, several researchers have studied
droplets motion in different conditions. Analyzing droplets motion and dynamics is essential
to calculate spray characteristics in various fields like spray cooling, spray freezing, and pesti
cide spraying. The equation of motion for droplets, bubbles, and solid particles has been dis
cussed by Michaelides (1997). This equation of motion can be numerically solved for droplets
to find their trajectory and velocity (Deshpande, Gao and Trujillo, 2011; Zhang, Zhang and
Bai, 2022). However, droplets dynamic has not been considered in the literature of spray-
freezing modeling to the best of authors’ knowledge. Furthermore, to fully capture the drop
lets dynamic, it is necessary to take the spray shape into account. This issue has not been
addressed in the literature as well.
Akhtar, Xu and Sasmito (2021) studied the thermal characteristics of the spray freezing setup
by developing a reduced-order theoretical-statistical model. However, the effects of droplets
motion are not considered in that study. In the present work, a reduced-order model for droplets
motion has been combined with the droplet freezing model introduced by Akhtar, Xu and Sas
mito (2021a) to accurately capture system-scale characteristics of spray freezing setup. Different
spray shapes have been studied and the droplets velocity distribution and residence times have
been calculated by numerical approaches. After that, the results have been coupled with the drop
let freezing model to calculate heat transfer coefficient (HTC), HTR, IPF, and cooling capacity.
2 MATHEMATICAL MODELING
The mathematical modeling of the spraying freezing system can be divided into three parts,
namely, a) Droplet freezing modeling; b) Droplet dynamics modeling; c) Spray modeling. The
first two parts are treated as particle-scale processes, while the third one can be seen as
a macro-scale process. The following assumptions are made in the present work:
i) Droplets are spherical with no deformation and there is no change in the droplet volume
during solidification;
ii) Droplets break-up and coalescence are neglected;
iii) The thermo-physical properties are assumed to be phase-dependent in the droplet freezing
model (Sections 2.1.1 and 2.1.2);
iv) When a droplet collides with the system boundaries, it does not contribute to the heat
transfer anymore.
In the following subsections, each modeling part will be discussed.
where:
570
The dimensionless variables are defined as:
where ri and rI,i are the droplet and interface radiuses, respectively, t is time, and αs is the ice
thermal diffusivity. Ste and Bi are the Stefan and Biot numbers, respectively, as defined below:
Equations (1)-(3) will be used to calculate the interface position of each droplet in the spray
freezing process. The schematic of the solid-liquid interface motion is shown in Figure 3 (c).
Figure 1. (a) Spray freezing setup (b) Droplet freezing curve (c) Solid-liquid interface motion for
a droplet.
where xd and ud are the droplet position and velocity vector, respectively. The drag force and
gravity are the two significant forces acting on a droplet and other terms in the equation of
571
motion can be neglected (Zhang, Zhang and Bai, 2022). Writing the second Newton’s law for
a single droplet gives:
where:
In above equations, md, ρd, rd, and Dd are the droplet mass, density, radius, and diameter,
respectively, ug, ρg, μg are the gas velocity vector, density, and dynamic viscosity, respectively,
g is the gravitational acceleration vector, and Red is the Reynolds number according to the drop
let-gas relative velocity. CD is the drag coefficient which can be calculated for different values of
Red as below (Zhang, Zhang and Bai, 2022):
From (5) to (8), the droplets position and velocity can be calculated over time using numer
ical approaches. Furthermore, the residence time of each droplet, i.e., the total time that each
droplet remains within the system domain, can be calculated by comparing its position with
the system boundaries.
where indices eq, c, r, s, and e are referring to terms “equivalent”, “convection”, “radiation”,
“sublimation”, and “evaporation”, respectively. Coefficient hr is not affected by the droplets
velocity because it is not a function of Reynolds number (Red), but hc, hs, and he are func
tions of Red. In previous works (Akhtar, Xu and Sasmito, 2021; Akhtar, Xu and Sasmito,
2021a), since droplets velocity is neglected, Red is calculated based on the air velocity:
This value will be used in the calculation of hc, hs, and he instead of Reold
d . It should be
noted that ud is obtained from the droplets dynamic model given in Section 2.1.2 and is
572
a function of droplet diameter, droplet initial direction, and spray shape. In the following
part, a novel approach for considering the spray shape and droplets initial velocity direction
in the droplet’s velocity calculation will be discussed.
where m_ s is the mass flow rate of the ice formed during the operation and m
_ t is the total mass
flow rate of sprayed water. To calculate IPF, the frozen portion of each droplet should be
calculated and then integrated for all droplets:
in which f(Di) is the probability of a droplet with diameter Di to exist, N_ is the particle injec
tion rate of the spray, and ms,i is the mass of the frozen portion of a droplet with diameter Di.
N_ can be calculated as:
where mi is the mass of a droplet with diameter Di. For a droplet with diameter Di,
variable ms,i in (14) is calculated by finding the interface position based on the droplet resi
dence time (τr,i) and freezing time (τf,i) using equations (1)-(3). It should be noted that the
573
volume change occurring during the water freezing process is neglected in the droplet freezing
model developed by Akhtar, Xu and Sasmito (2021a). This will result in the mass conservation
violation due to the phase-dependency of density which yields a considerable error in IPF cal
culation. Hence, a correction is applied to IPF calculation by using the mass conservation
equation. The total mass of water is conserved during the process, hence:
where ml.i is the volume of unfrozen water of the droplet with diameter Di and Δm is the mass
difference due to neglecting the droplet diameter movement. Δm can be added to the numer
ator of the IPF correlation (14) to compensate for the effect of droplet volume increase during
the freezing. Therefore:
in which ei is the total heat emission of a droplet with diameter Di during solidification and
can be calculated as:
Figure 2. Probability distributions (a) Flat spray (Liu, 1999) (b) Cone sprays (Liu, 1999) (c) Experimen
tal data from Santangelo (2010).
where hst,i is the HTC of each freezing stage, Ti is the temperature profile over time, T∞ is the
air temperature, and Ai is the droplet surface area.
3 MODEL VALIDATION
To validate the model, the experimental data from the frood-stobie mine reported by Stachulak
(1991) has been used. The droplet distribution in this part is adopted from the experimental
results proposed by Santangelo (2010) which suitably mimics the situation in frood-stobie mine.
This distribution is depicted in Figure 2 (c). As discussed by Stachulak (1991), the total amount
of ice produced during the winter operation of the frood-stobie mine is approximately mice,exp
=145.000 tons. Using the developed model in the present work with the distribution shown in
574
Figure 2 (c), the IPF will be equal to 92%. Considering 1,200 operating hours of two stopes
during the winter and an average water flow rate of 18.9 kg/s as reported by Stachulak (1991)
and Trapani and Chen (2017), the total amount of ice formed during winter can be calculated
using the model output as below:
Figure 3. Modeling droplet initial directions (a) Flat spray (b) Hollow cone spray (c) Full cone spray.
The developed model has been run using the specifications given by Stachulak (1991) for the
frood-stobie mine. The spray shapes discussed in Section 2.2 along with their corresponding
diameter distributions have been considered in running the code one by one. For each spray,
different parameters have been calculated and averaged using the given distribution. The
results are summarized in Table 1. As can be seen from this table, the values of IPF and HTR
are greater for hollow cone and full cone sprays compared to flat spray. This difference is due
to the different diameter distributions or SMDs of the sprays as reflected in Figure 2. While
larger droplets require more time for freezing, their residence time is smaller compared to
finer droplets due to the larger gravity force imposed on them. Hence, the ratio of the frozen
volume over the droplet volume is smaller for droplets of larger diameter which in turn results
in a smaller IPF for sprays with larger SMDs.
Comparing the results for hollow cone and full cone spray shapes shows that although the
spray shapes and droplet trajectories are different, but the average values listed in Table 1 are
quite close. The reason is that the diameter distribution is the same for both cases as they are
produced by a specific type of nozzle.
575
Table 1. Average values for different spray shapes.
Spray shape
Parameter and unit Symbol Flat spray Hollow cone Full cone
The ice produced during the cold days of winter can be stored and utilized in summer for
cooling purposes. The cooling capacity (Cc) of a spray-freezing setup is calculated by multiply
ing the rate of ice production by the water latent heat of fusion and the total operating time:
where top is the total operating time. As can be inferred from (21), a larger IPF will result in
a greater cooling capacity, hence using cone-sprays will provide more cooling capacity during
the summer operations.
5 CONCLUSION
In the literature of spray freezing modeling, there is a research gap in bridging between the
models of small-scale droplet freezing phenomena and system-scale spray characteristics. The
present work aimed to address this shortcoming by developing a reduced-order model for cal
culating droplets velocity and residence times and incorporating this model into the existing
droplet-scale models of freezing phenomena and experimental correlations of spray character
istics. For modeling the droplets dynamic, different spray shapes, i.e., flat, hollow cone, and
full cone sprays have been modeled in a computationally efficient manner. Droplets diameter
distribution is then included in the model for each spray shape to determine the system-scale
characteristics of the setup like the HTR and IPF. The total amount of ice formed during the
winter operations has been used as a benchmark to validate the model. The model results indi
cate that sprays with finer droplets (like cone sprays) will provide greater HTR and IPF. The
developed model can be extended to capture other contributing phenomena like droplet-
droplet and droplet-wall collisions.
ACKNOWLEDGEMENT
The authors wish to thank the Fonds de recherche du Québec - Nature et technologies (FRQ-
NT) (PR-300597) and the Natural Sciences and Engineering Research Council of Canada
(NSERC) (RGPIN-2021-02901). The first author would also like to show his gratitude to the
McGill Engineering Doctoral Award (MEDA) for supporting his doctoral research.
REFERENCES
Akhtar, Saad, Xu, M. and Sasmito, A. P. 2021a. Development and validation of a semi-analytical frame-
work for droplet freezing with heterogeneous nucleation and non-linear interface kinetics, Inter-
national Journal of Heat and Mass Transfer. Elsevier Ltd, 166, p. 120734.
576
Akhtar, Saad, Xu, M. and Sasmito, A. P. 2021b. Development and validation of an asymptotic solution
for a two-phase Stefan problem in a droplet subjected to convective boundary condition, International
Journal of Thermal Sciences. Elsevier Masson SAS, 164(December 2020), p. 106923.
Akhtar, S., Xu, M. and Sasmito, A. P. 2021. Spray freezing for mine heating a statistical perspective,
Mine Ventilation, (2008), pp. 357–365.
Caldwell, J. and Kwan, Y. Y. 2003. On the perturbation method for the Stefan problem with time-dependent
boundary conditions, International Journal of Heat and Mass Transfer. Elsevier, 46(8), pp. 1497–1501.
Deshpande, S., Gao, J. and Trujillo, M. F. 2011. Characteristics of hollow cone sprays in crossflow,
Atomization and Sprays. Begel House Inc., 21(4).
Liu, H. 1999. Empirical and Analytical Correlations of Droplet Properties, in Liu, H. B. T.-S. and E. of
D. (ed.). Norwich, NY: William Andrew Publishing, pp. 238–314.
Meirmanov, A. M. 2011. The stefan problem. Walter de Gruyter.
Michaelides, E. E. 1997. Review—the transient equation of iviotion for particles, bubbles, and droplets,
Journal of Fluids Engineering, Transactions of the ASME, 119(2), pp. 233–247.
Rubinšteĭn, L. I. 2000. The stefan problem. American Mathematical Soc.
Santangelo, P. E. 2010. Characterization of high-pressure water-mist sprays: Experimental analysis of drop
let size and dispersion, Experimental Thermal and Fluid Science. Elsevier Inc., 34(8), pp. 1353–1366.
Stachulak, J.S., 1991. Ventilation strategy and unique air conditioning at Inco Limited. CIM Bul-letin
(Canadian Institute of Mining and Metallurgy); (Canada), 84(950).
Trapani, K. and Chen, Z. 2017. Computational fluid dynamic modelling of the Frood-Stobie ice stope
thermal storage for mine ventilation heating, Proceedings of the Eighth International Conference on
Deep and High Stress Mining, pp. 289–298.
Zhang, Haibin, Zhang, Hao and Bai, B. 2022. Numerical Study on Flow Structures of a Hollow Cone
Spray in Crossflow, Heat Transfer Engineering. Taylor & Francis, 43(8–10), pp. 737–753.
577
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Lithium-Ion Batteries (LIB) have dominated the energy market for several
decades due to their high energy density and long-life cycle. However, several fire accidents in
electric vehicles have raised questions about their safety concerns. Researchers have identified
thermal runaways to be the major reason for the fire susceptibility of LIBs. One of the major
risks in accidents involving LIB fires is toxic gaseous emissions. Hence, it is necessary to
understand those toxic emissions to properly counter them. In this study, small-scale battery
thermal abuse tests will be performed, and the toxic gases will be analyzed using different
tools such as Fourier Transform Infrared (FTIR) spectroscopy ultimately. 5 different chemis
tries of roughly the same nominal capacity i.e., 3000 mAh, will be tested and a comparison of
the gaseous emissions will be presented in future studies, which has not been done before.
However, so far, we have performed preliminary experiments to study the surface temperature
of cells. The results of this study showed that the thermal runaway of the cell is triggered when
the surface temperature of the cell reaches 190°C, and it keeps on increasing for up to 208°C,
even without the use of the heating source, the duration of the experiment varies from 30min
to 45min. Based on the study results, suitable experiment scenarios are designed the study the
gaseous analysis of the LIB on the cell level.
1 INTRODUCTION
Lithium-ion batteries (LIBs) are the most commonly used batteries in battery electric vehicles
(BEV) due to their high energy density and long lifetime (Larsson et al., 2016). However, sev
eral BEV accidents involving battery fires have raised concerning alarms about the fire risks
of LIBs (Sun et al., 2020)(CMTeam, 2022). Since 2006, Federal Aviation Administration
(FAA) has recorded about 433 fire accidents involving LIBs in aviation (FAA, 2022). In 2012,
A Toyota Prius (hybrid) was recorded to have caught fire after being flooded by hurricane
Sandy (Labovick, 2021)(GARTHWAITE, 2012). In 2013, two Tesla Model Ss were reported
to have caught fire after being involved in two separate collisions(Musk, 2013)(George, 2013).
Further down the timeline, different car makers have experienced BEVs catching fire accidents
such as Volkswagen in 2017 (Traugott, 2021), BMW i8 (Bruce, 2019), and two separate events
involving Porsche Panamera in 2019 and a Taycan in 2020 happened in Portugal and USA
respectively (Labovick, 2021). Several other events have happened when BEV caught fire
while charging or parking idle (Wayland, Micheal. kolodny, 2021)(Yoney, 2021).
Existing research has identified thermal runaways to be the probable reason for most of the
fire accidents in BEVs (Sun et al., 2020)(Liu et al., 2021)(Liu et al., 2016). Thermal runaway
happens when the battery pack ignites spontaneously due to overheating, according to the lit
erature. It is caused by either electrical abuse, mechanical abuse/collision, or thermal abuse (Sun
et al., 2020)(Liu et al., 2016)(Zhong et al., 2019)(Zhao et al., 2021). Once thermal runaway is
DOI: 10.1201/9781003429241-59
578
triggered, chemical reactions start to take place inside the cell/pack, causing the emission of
toxic fumes. Due to the presence of fuel and oxidizer inside the cell, these chemical reactions will
most likely continue spontaneously, giving rise to heat and toxic gases during the process
(Doughty and Roth, 2012). Several studies have analyzed the thermal characteristics of LIB
fires, for instance, Yangyang Fu et.al (Fu et al., 2015) concluded from their experimental study
that the stored electrical energy has a significant impact on the burning behaviors of LIBs. In
the contrast, Larsson et.al (Larsson et al., 2016) stated that the stored electrical energy has nom
inal dependence on the thermal properties of the battery.
One of the most under-explored areas of BEV fires is the amount and types of gases emitted
during fires. These toxic gases pose an immense threat to human health (Peng et al., 2020), as
fatalities and injuries are mostly caused by the inhalation of toxic gases in fire accidents (Stec,
2017). Several researchers have performed experiments on the fire behavior and propagation
of the fire in LIB cells (Fu et al., 2015)(He et al., 2020). However, very little information is
present on the types and amount of toxic gases as suggested by a review performed by Sun
et al. (Sun et al., 2020). The presence of toxic gases such as CO, CO2, HCl, HF, SO2, and
POF3 has been confirmed by different studies(Larsson et al., 2017) (Ribière et al., 2012), How
ever, the amount of the gases has still not been well understood. Larsson et al (Larsson et al.,
2017) indicated the presence of HF, POF3, and PF5, However, the cells used by Larsson et.al
(Larsson et al., 2017) were of different chemistries and the number of cells in each battery
sample varied. In a separate study (Lecocq et al., 2016), the fire behavior of Li-ion batteries
with LiPF6 chemistry was analyzed and the presence of toxic gases such as HF, and SO2 was
confirmed, however, they were also unable to quantify the amount of the gases and also the
experiment was done on only one chemistry of LIB. This research is also endorsed by
a different study in which LIB cells were dissembled and the heat value and gases analysis was
performed, confirming the presence of PF5, CO2, and HF (Kriston et al., 2019). However, the
quantification and complete analysis of the gases is unavailable. Besides, reignition is another
major problem related to LIB fires and several researchers have studied the effect of different
extinguishers on LIB fires, and so far, a specifically effective extinguisher against LIB fires is
non-existent (Maloney, 2014)(Luo et al., 2018)(Su et al., 2014).
The main objective of this study is to evaluate the toxic fumes emitted during fire incidents
involving Li-ion batteries. However, so far, only preliminary tests are performed, and the results
are used as a guideline for the future study of the project. Similar studies have been performed
by other researchers; however, the difference lies in the number of chemistries of LIB used, the
types of gases analyzed, and the SOC relative to the LIB chemistries tested. The objective of the
research will be achieved by first identifying and quantifying the toxic fumes by performing fire
abuse tests. Applying fire directly to the battery cells, the cells will be left to burn out completely
at different states of charge (SOC). The smoke of the batteries will be analyzed for the type and
amount of gases, with the help of a multi-gas detector and also spectroscopy using FTIR. This
analysis will be useful in identifying and quantifying the gases emitted during battery fires,
which can further be used in designing specialized protective equipment and fire extinguisher
designs, to elevate the health and safety standards of battery-powered vehicles.
2 METHODOLOGY
579
Figure 1. Lithium-ion battery incinerator.
580
shown in Figure 2 and is gradually heated up at a rate of 4°C per min until the cell starts
smoking off. A high-resolution thermal camera (FILR E4) Figure 7a, is used to capture the
burning process of the battery based on the flame profile and temperature data, through the
transparent window of the chamber.
Figure 2. a: the Nickel Chromium resistance wire, b: application of heat using Ni-Chrome resistance wire.
For the analysis of the smoke, we made used Fourier Transform InfraRed (FTIR) spectros
copy. Thermo Scientific Antaris IGS analyzer Figure 5, which is also reported to have been
used by F Larsson (Larsson et al., 2017). The sample is burned in the chamber, and the smoke
is pulled out by the inline duct fan at an average speed of 40 L/sec, however before the smoke
is released into the atmosphere, the sample is sucked into the FTIR gas chamber using an
external vacuum pump at 0.05 L/sec. The gas cell conditions are set at 180°C and 650 mmHG,
and these were the conditions at which the FTIR Fire Science Calibration Library will be used
for gaseous analysis. The experiment lasted for 45 min and the results are based on real-time
analysis of the smoke.
581
Figure 4. Location of thermocouples on the cell.
Fire will be directly applied to a battery cell in the chamber, once it starts to smoke, it will
be collected and analyzed in by the multi-gas detectors and the sample of the gas will also be
collected for FTIR analysis for the POF5, PF3, and HF. The experimental setup was bor
rowed and modified from a study performed by F. Larsson (Larsson et al., 2017) in 2017.
Once the gas sample is analyzed, and based on those analyses, PPEs designs, and Fire suppres
sion techniques will be recommended.
Figure 6 shows the illustration of the experimental setup used for the analysis of Toxic emis
sions of LIB cells. Starting from Step 1 includes lithium-ion battery cells indicated by A1,
which are first charged to the desired state of charge with a MIBOXER Cell charger indicated
by A2 after that in step 2, J thermocouples (B1) are attached to the cell surface which is con
nected to Anabi AT44532 temperature data logger (B2). In the next step, the 18 Gauge
Kenthal A-1 Annealed round Nickle Cobalt resistance wire (C1) is wound into a coil and
powered with a DC power supply (C2), and the system is placed in the incinerator. The cell is
then heated till it starts smoking and the smoke is directed toward the FTIR (D1) in step 4,
and the gases are analyzed with OMNIC software using a personal computer (PC). The data
from the temperature data logger is also stored in the PC using AT4500 software, as shown in
Figure 6, Step 4 and step 5 are not a part of this report.
Compared to the similar setup used by other researchers, the advantage is that this setup is
easily and commercially available in the market, this setup is much easier and economical to
build and use for battery experiments. Other researchers have used a similar setup, however,
they have some equipment such as SBI apparatus, which is much more lab specific and cost too
582
Figure 6. Experimental setup used for toxic gas analysis of the LIB emissions.
much. This setup is a modified version of that used by Larsson et.al (Larsson et al., 2017) How
ever, the difference is that the designed experiment can collect the temperature data using both
thermocouples and a thermal imaging camera Figure 7(a), which will be further processed for
fire profiling analysis. This will help us to better understand the fire profile and devise effective
suppression solutions and help in the development of emergency PPEs for LIB fires.
Another aspect of the novelty of this research is that so far, no researcher has identified and
analyzed all the gases at once while studying five different LIB chemistries with the same
power and charge capacities. For instance, F Larsson et.al (Larsson et al., 2017) studied three
chemistries but with variable capacities, P. Ribière et.al (Ribière et al., 2012) and Yangyang
Fu et.al (Fu et al., 2015) studied only one chemistry. Hence studying different chemistries at
the same capacity level will give us an insight into how different chemistries behave under the
same circumstances, and it will be a realistic comparison.
3 PRELIMINARY RESULTS
Initially, trial experiments were performed to find a suitable method and apparatus for the
experiments. The experiments were repeated 6 times, and most of the results are the same as
described below. Different types of abuse methods were applied including propane burner,
heating strip, and Ni-Chrome resistance wire. Tests indicated problems with controlling the
heat applied to the sample using the propane burner. However, the heating strip and the Ni-
Chrome resistance wire techniques performed better. Ni-Chrome resistance wire with a DC
voltage controller, as it is economical compared to the heating strip, is recommended, and will
be used further in this research.
The surface temperature of the cell was measured with the help of thermocouples, TC1 is
attached to the top (valve) end while TC2 is connected to the bottom end Figure 4, data were
recorded continuously every 10 sec. In these preliminary experiments, it was observed that the
cell starts smoking when the surface temperature of the cell exceeds 190°C. The electrolyte is
ejected from the cell. A mild hissing sound is evident that intensifies over the remainder of the
experiment until all the gases are emitted from the cell. The burning process was also captured
with a thermal camera to analyze the flame profile and surface temperature of the cell
Figure 7.
583
Figure 7. A: FLIR E4 Thermal Imaging Camera, b: Thermal image of the burning process of the
cell.
The power voltage of the DC power supply was increased after each 10 min. which can be
seen in Figure 8. The power of the supply was kept constant at 20W for the first 20 min. It can
be seen on the surface of the temperature profile of the cell. Here, the rate of temperature
increase doubles in the third section of the experiment. When the temperature of the cell
reached 220°C, it started smoking vigorously, and the power supply was turned off. The cell
kept smoking for 10 more mins and the temperature of the cell decreased thereafter. Figure 8
shows the data captured by only two thermocouples because only two of them were attached
to the cell.
Figure 8. Data Logger results of the surface temperature of the cell trial 1; TC1 (red), TC2 (yellow).
In another trial was repeated with the same procedure and different results were achieved
this time for the surface temperature of the battery. The experiment lasted for more than
one hour and the liquid was still confined within the cell. The maximum surface temperature
was seen to be up to 280 °C. At the time, t=34.0 min, the power source was turned off. The
surface temperature was seen to rise to 190 °C. It further rose to 208 °C without any external
source. It can, therefore, be inferred that thermal runaway occurs when the temperature of the
cell reaches 190 ° Figure 9.
584
Figure 9. Data logger results of the surface temperature of the cell.
4 CONCLUSIONS
The preliminary tests run on the small-scale LIB yield some scientific results. They are
described below:
• It takes 45 mins for the cell to burn out with the 50W DC power supply.
• The cell starts to smoke when the temperature reaches 190 °C. This also could be assumed
to be the temperature at which the thermal runaway begins.
5 FUTURE STUDIES
For future recommendations, this study provides valuable input on cell behavior under ther
mal abuse. The authors will continue to work on capturing the smoke coming out of the cell
and analyze the nature of the smoke for the type and amount of different toxic gases with the
help of FTIR, further in the project.
NOTE
This is a work in progress, and the study is still not complete yet.
REFERENCES
Bruce, C., 2019. Smoking BMW i8 Dumped In Water By Firefighters [WWW Document]. URL https://
www.motor1.com/news/315476/bmw-i8-fire-reponse-netherlands/
CMTeam, 2022. How Many Electric Cars Catch Fire Every Year: Electric Car Fire Statistics (2022 Find
ings) [WWW Document]. URL https://www.carsmetric.com/electric-car-fire-statistics/#:~:text=Key
Electric Car Fire Statistics 1 Electric-powered vehicles,vehicles and 3.4%25 for hybrid vehicles. More
items.
Doughty, D.H., Roth, E.P., 2012. A general discussion of Li ion battery safety. Electrochem. Soc. Inter
face 21, 37.
FAA, 2022. Lithium Battery Incidents [WWW Document]. URL https://www.faa.gov/hazmat/resources/
lithium_batteries/incidents
Fu, Y., Lu, S., Li, K., Liu, C., Cheng, X., Zhang, H., 2015. An experimental study on burning behaviors
of 18650 lithium ion batteries using a cone calorimeter. J. Power Sources 273, 216–222.
Garthwaite, J., 2012. Mystery at Port Newark: Why Did 17 Plug-In Cars Burn? NewYork Times.
585
George, P., 2013. Another Tesla Model S Caught Fire After A Crash In Mexico [WWW Document].
URL https://jalopnik.com/another-tesla-model-s-caught-fire-after-a-crash-in-mexi-1453376349.
He, X., Restuccia, F., Zhang, Y., Hu, Z., Huang, X., Fang, J., Rein, G., 2020. Experimental study of
self-heating ignition of lithium-ion batteries during storage: effect of the number of cells. Fire Technol.
56, 2649–2669.
Kriston, A., Adanouj, I., Ruiz, V., Pfrang, A., 2019. Quantification and simulation of thermal decompos
ition reactions of Li-ion battery materials by simultaneous thermal analysis coupled with gas analysis.
J. Power Sources 435, 226774.
Labovick, 2021. ELECTRIC VEHICLE FIRE INCIDENTS AND STATISTICS [WWW Document].
URL https://www.labovick.com/blog/electric-vehicle-fire-incidents-and-stats/
Larsson, F., Andersson, P., Blomqvist, P., Mellander, B.-E., 2017. Toxic fluoride gas emissions from
lithium-ion battery fires. Sci. Rep. 7, 1–13.
Larsson, F., Andersson, P., Mellander, B.-E., 2016. Lithium-ion battery aspects on fires in electrified
vehicles on the basis of experimental abuse tests. Batteries 2, 9.
Larsson, F., Mellander, B.-E., 2014. Abuse by external heating, overcharge and short circuiting of com
mercial lithium-ion battery cells. J. Electrochem. Soc. 161, A1611.
Lecocq, A., Eshetu, G.G., Grugeon, S., Martin, N., Laruelle, S., Marlair, G., 2016. Scenario-based pre
diction of Li-ion batteries fire-induced toxicity. J. Power Sources 316, 197–206.
Liu, X., Wu, Z., Stoliarov, S.I., Denlinger, M., Masias, A., Snyder, K., 2016. Heat release during
thermally-induced failure of a lithium ion battery: Impact of cathode composition. Fire Saf. J. 85,
10–22.
Liu, Y., Sun, P., Niu, H., Huang, X., Rein, G., 2021. Propensity to self-heating ignition of open-circuit
pouch lithium-ion battery pile on a hot boundary. Fire Saf. J. 120, 103081. https://doi.org/10.1016/J.
FIRESAF.2020.103081
Luo, W., Zhu, S., Gong, J., Zhou, Z., 2018. Research and development of fire extinguishing technology
for power lithium batteries. Procedia Eng. 211, 531–537.
Maloney, T., 2014. Extinguishment of lithium-ion and lithium-metal battery fires. US Fed. Aviat. Adm.
Washington, DC, USA 46–51.
Musk, E., 2013. Model S Fire [WWW Document]. URL https://www.tesla.com/blog/model-s-fire
Peng, Y., Yang, L., Ju, X., Liao, B., Ye, K., Li, L., Cao, B., Ni, Y., 2020. A comprehensive investigation
on the thermal and toxic hazards of large format lithium-ion batteries with LiFePO4 cathode.
J. Hazard. Mater. 381, 120916.
Ribière, P., Grugeon, S., Morcrette, M., Boyanov, S., Laruelle, S., Marlair, G., 2012. Investigation on
the fire-induced hazards of Li-ion battery cells by fire calorimetry. Energy Environ. Sci. 5, 5271–5280.
Stec, A.A., 2017. Fire toxicity–The elephant in the room? Fire Saf. J. 91, 79–90.
Su, C.-H., Chen, C.-C., Liaw, H.-J., Wang, S.-C., 2014. The assessment of fire suppression capability for
the ammonium dihydrogen phosphate dry powder of commercial fire extinguishers. Procedia Eng. 84,
485–490.
Sun, P., Bisschop, R., Niu, H., Huang, X., 2020. A review of battery fires in electric vehicles. Fire Tech
nol. 56, 1361–1410.
Traugott, J., 2021. Two-Day-Old VW Golf Hybrid Explodes While Driving [WWW Document]. URL
https://carbuzz.com/news/two-day-old-vw-golf-hybrid-explodes-while-driving.
Wayland, Micheal. Kolodny, L., 2021. Fires, probes, recalls: The shift to electric vehicles is costing auto
makers billions [WWW Document]. URL https://www.cnbc.com/2021/08/19/fires-probes-recalls-auto
makers-spend-billions-in-shift-to-evs.html
Williams, F.W., Back, G.G., 2014. Lithium battery fire tests and mitigation. NAVAL RESEARCH LAB
WASHINGTON DC CHEMISTRY DIV.
Yoney, D., 2021. Hyundai Kona Electric Fires In Norway And Korea Cause Concern [WWW Docu
ment]. URL https://insideevs.com/news/515983/kona-electric-fire-norway-korea/.
Zhao, J., Xue, F., Fu, Y., Cheng, Y., Yang, H., Lu, S., 2021. A comparative study on the thermal runaway
inhibition of 18650 lithium-ion batteries by different fire extinguishing agents. Iscience 24, 102854.
Zhong, G., Mao, B., Wang, C., Jiang, L., Xu, K., Sun, J., Wang, Q., 2019. Thermal runaway and fire
behavior investigation of lithium ion batteries using modified cone calorimeter. J. Therm. Anal.
Calorim. 135, 2879–2889.
586
Ventilation monitoring and measurement
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
D.J. Brake
Mine Ventilation Australia
ABSTRACT: Gauge and Tube (also known as Trailing Hose or P-Q) Surveys have been
widely used in underground mine ventilation for frictional pressure loss surveys and ventilation
model calibrations for a very long time. They are one of the most basic types of underground
ventilation measurement. However, there are many factors that have been contributing to and
accelerating the decline of this type of survey and underground measurements more generally. In
most cases, barometric pressure surveys are now the preferred method for validating ventilation
circuits and models and can be completed more quickly, more cheaply, with fewer personnel and
with less disruption to mine operations. In addition, the growing use of real-time instrumentation
of the ventilation system combined with Big Data analysis techniques is reducing the amount of
manual measurements and other data collection required. Artificial Intelligence is also likely to
assist in the future with providing statistically reliable advice regarding the location, type and fre
quency of underground measurements, as well as answers to many other ventilation questions
including validating ventilation models and providing site-specific advice on management of
upset conditions such as fan or ventilation control failures or outages, fires or explosions. This
paper discusses the factors behind these trends and identifies some of the advantages and disad
vantages of both gauge and tube and barometric pressure surveys, as well as the broader poten
tial use of Big Data and Artificial Intelligence to assist with decisions regarding underground
measurements generally and assisting with other ventilation-related advice.
1 INTRODUCTION
Gauge and tube ventilation surveys have been the mainstream ventilation survey technique
for measuring underground flow directions and volumes for a very long time. However, this
type of survey started falling from favour in Australian hardrock mines from the early 1980s
and was almost entirely abandoned by the early 1990s. Gauge and tube surveys persevered
somewhat longer in coal mines but have also now been abandoned for many years.
The coup de grâce for the decline of this type of survey occurred during the transitional
period when underground mines went from 5-day operations to 7-day operations. This major
structural change in the industry was usually associated with moving from 8-hour to 12-hour
shifts and in many cases in Australia, also associated with moving from residential to fly-in,
fly-out (FIFO) operations.
The nature of gauge and tube surveys means they need to be undertaken when there is the
minimum possible activity in the mine resulting in the minimum disruption to the ventilation cir
cuits. This was manageable when gauge and tube surveys could be undertaken when the mine
was mostly idle over a weekend and ventilation staff lived in a nearby town and could complete
the survey on “overtime”. When mines moved to 7-day operations, gauge and tube surveys
became increasingly problematic.
DOI: 10.1201/9781003429241-60
589
However, the reasons for their declining use were not only related to the loss of available
non-production time underground to conduct the survey but also to the availability of suitable
personnel and their time and the rise of alternative measurement technologies and processes.
In addition, the rapid advances in the use of real-time instrumentation, Big Data analysis
techniques and Artificial Intelligence are likely to become a significant factor in the science
and practice of underground measurements and ventilation model validation, fault-finding
and decision-making in the medium term.
This paper summarises the author’s opinions about Guage and Tube Surveys in particular
and extends these reflections to the case of underground measurements more generally.
2 VENTILATION SURVEYS
The theory and techniques for gauge and tube (G&T) surveys are well described by McPher
son (2007). However, there are a number of drawbacks of using gauge and tube surveys in
modern mines, despite their historical popularity.
• In a gauge and tube survey, the resistance of every section of airway (calculated from every
flow and pressure measurement of the gauge and tube as it is advanced) has to be added to
the resistance of the previous gauge and tube measurement. Any errors that creep in even
tually get carried through the system (at least unless a second closed loop can be formed).
• Airway resistances often change over time for a variety of reasons, so that just surveying
the incremental new parts of the ventilation circuit since the last survey can and will
590
eventually result in problems in the model as older airways are altered and these changes
are not necessarily recognised.
• There is significant manpower and time required and hence significant labour cost.
• There is the opportunity cost to these time-consuming surveys. While the ventilation team
is undertaking a gauge and tube survey, they are not doing other, often critical, ventilation
work and this lost time cannot be made up
• There is the disruption to the operations with restrictions on equipment travel and/or venti
lation control operation during the period of the survey to obtain accurate results
• There are potential safety issues for the survey team if surveying in high activity areas, e.g.
on main ramps or travelways or in exhaust airways during production periods.
• It is not possible or very difficult to undertake a gauge and tube survey in many vertical
raises or other airways in which travel through is not possible.
Expanding the remit of this discussion from gauge and tube surveys to the role of underground
ventilation measurements more generally, it could easily be argued that most of the underground
measurements taken by the ventilation professional are never used again and contribute little to
the effective running of the mine, or the validation of the ventilation model in the mine. In some
cases, the collection of underground measurements actually reduces the ability of the ventilation
professional to perform other more value-adding activities and therefore decreases the overall
effectiveness of the ventilation system and negatively impacts on the ventilation outcomes on site.
4 CHANGING TECHNOLOGIES
There are four trends over the past few decades impacting on the methodology for conducting
ventilation measurements surveys underground.
Firstly, the development of very comprehensive and accurate ventilation modelling pro
grams. Compared to the pre-computer era or even the period of early ventilation modelling
software, modern mine ventilation modelling programs can comprehensively include all the
factors that drive flow and pressure in underground ventilation networks including variable
density, compressible airflow, shock losses and full thermodynamic modelling as required. To
some extent they can also not only model steady-state conditions but even transient condi
tions. Any errors in current ventilation models are due to errors in the physical inputs of the
airways and leakage resistances etc., rather than due to assumptions or approximations in the
software itself, as was previously the case.
Secondly, the ready availability of inexpensive high accuracy barometers. In many respects,
the traditional method of undertaking ventilation surveys in a mine is analogous to the trad
itional method of undertaking a topographical survey of surface land: in historical land survey
ing practices, surveyors would use levels or plane tables and theodolites to comprehensively
measure the three dimensions of many surface points and create contour maps from these
measurements. However, with the advent of, firstly, accurate aerial mapping and more recently
differential GPS, these older methods of land surveying have been mostly superseded. It is very
possible today to use aerial photographs to develop high resolution contour maps with only
the occasional spot heights (or spot elevations) surveyed by humans, if at all. In a similar way,
high accuracy barometers have made similar advances possible with ventilation surveys as will
be discussed in the next section.
Thirdly, the increasing trend to instrumenting the ventilation system in a mine, often part of
a more comprehensive Ventilation-on-Demand system. The use of fibre-optic communication
backbones has also made real-time Big Data type of analyses now possible. Instrumentation,
VOD and Big Data will continue to decrease the number of manual measurements needing to be
taken, assuming suitable software becomes available to process this data into useful information.
Fourthly, the development of so-called Artificial Intelligence is beginning to have a major
impact on assisting with solving many complex problems in society and industry ranging from
medical diagnosis to setting up spreadsheet formulas. AI is likely to have an equally profound
impact on the profession of mine ventilation over the decade to come.
591
5 BAROMETRIC PRESSURE SURVEYS
The theory and techniques for barometric (or altimeter) pressure (BP) surveys are well
described by McPherson (2007). BP surveys are not new and date back to around the 1920s
(McElroy & Kingery, 1957; Harris et al, 1973). However, it is the ready available and low cost
of high accuracy barometers (Derrington, 2015) that has made the technique a far more
attractive option than previously.
A barometric pressure survey can be used in at least two ways:
• For airways which have a significant frictional pressure loss along their length, a BP survey
can be used to directly measure frictional pressure loss, resistance and friction factor for
that airway
• For validation of a ventilation model, effectively checking the integrity of the model against
measured “spot heights” (in the pressure sense).
In the latter case, a number of nodes are selected throughout the mine and the barometric
pressure (and wet and dry bulb temperatures) are accurately measured at those locations. If
the airflows in the mine are known (and these can be measured easily and accurately), and the
elevation of the BP positions is accurately known (to within about 0.5 m, which can now usu
ally just be done simply and easily using the survey floor strings in the as-built survey data
base), then the pressure loss between surface and those node points can be accurately
measured and compared to the predicted values from the ventilation model. In addition to
correlating the model by comparing predicted and actual BP values, it is good practice to also
measure the differential pressure (by manometer) across key ventilation controls, which can
also then be checked against the modelled values of differential pressure (DP) at these loca
tions. Ventilation controls are also inspected to estimate leakage (m3/s) so that this can be
accurately modelled. Measuring DP and flow across ventilation controls (effectively a gauge
and tube survey of a single “airway”) remains very useful in terms of measuring the leakage
resistances of different types and constructions of ventilation controls (e.g. doors, air cross
ings). Any discrepancies between measured and modelled values of BP, DP and leakage can
be investigated. Once any differences are within a reasonable tolerance, the model can be con
sidered to be validated and fit for purpose.
The main advantages of barometric pressure surveys are:
• They are quick and inexpensive (compared to gauge and tube)
• There are no cumulative errors as each measurement stands alone as a fixed reference point
for validation in the model
• Any changes in the resistance of old airways or significant new sources of leakage are
detected because each BP survey is, in effect, a full new mine survey.
• Very importantly, if the as built ventilation model has failed to include some new piece of
development somewhere, or some important development has not been removed from the
model (i.e. these changes have inadvertently not been included in the ventilation model)
then the BP survey will detect this, but an incremental gauge and tube survey of only the
new development in a mine will not.
• A BP survey can be conducted by one person, assuming the base (or control) station tech
nique is used, with the surface barometer at a suitable location in data-recording mode
• A BP survey can be conducted with the mine operating as normal with generally little loss
of accuracy.
• A BP survey results in negligible disruption to mine operations.
Barometric pressure surveys are particularly useful for very large operations or those with
many airways that are difficult or impossible to traverse. For example, the three integrated
Mount Isa underground mines circulated 6000 m3/s of air through over 600 km of ventilated
airways over a strike length of 5 km, from surface to 1900 metres below surface and with
levels every 40 m (every 20 m in the upper areas).
592
6 THE IMPACT OF VENTILATION SYSTEM INSTRUMENTATION,
VENTILATION-ON-DEMAND AND BIG DATA
The increasing range of responsibilities of the mine ventilation engineer has already put great
pressure on his or her ability to collect manual measurements from underground. Even where
these are collected, sometimes for statutory purposes, they are often not used for any useful
subsequent analysis. Most new mines are now being set up with a fibre optic communications
backbone and larger mines are implementing an increasing degree of real-time instrumenta
tion and monitoring, and in some cases, adopting a form of remote and semi-autonomous
control of the ventilation system (VOD). The ability to collect and usefully manage large sets
of data (Big Data) will be an essential precursor to the development and refinement of Artifi
cial Intelligence systems in mine ventilation engineering.
Artificial Intelligence (AI) has been defined (Copeland 2023) as “the ability of a digital computer
or computer-controlled robot to perform tasks commonly associated with intelligent beings. . .
developing systems endowed with the intellectual processes characteristic of humans, such as the
ability to reason, discover meaning, generalize, or learn from past experience”. Examples would
include some robotics, self-driving cars, chatbots, virtual travel agents and many others.
So how could AI assist with ventilation decision-making and model validation in the future?
With respect to ventilation model validation, an interesting paper by Griffith and Stewart
(2021) effectively described using a limited number of gauge and tube survey data to improve the
calibration of a mine ventilation computer model by setting up a sensitivity matrix of actual
versus measured values to modify the resistance of the most probable airways that need resist
ance adjustment. It is likely that this approach could be modified to do one of the following:
• Use the same limited (or nil) P-Q data, along with accurate barometric pressure data and
differential pressure measurements, and with the option for the user to set or relax friction
factors and airway sizes and other parameters, to develop the most reliable true estimate of
the ventilation model of the mine, or
• Assess the model and provide a list of ventilation controls or fan locations and the type of
measurements needing to be undertaken at each location (e.g. pressure difference across
a ventilation control, pressure and flow through a fan, barometric pressure difference to
surface), to provide a statistically-valid confidence level for the model. For example, the AI
system could determine that if certain specific measurements are taken, it would then be in
a position to use these measurements in the model to determine how much uncertainty is
left in the reliability of the overall model, or perhaps even provide advice about the uncer
tainty regarding a particular question from the ventilation engineer such as this: “Given the
existing ventilation model and the results of the last flow and pressure survey and the accur
acy of these measurements, what additional measurements would need to taken to be 95%
confident that a predicted fan pressure and flow at (for a new fan duty at some location)
would be within 3% on both flow and pressure”.
• In fact, all five questions posed in the dot points in section 2 of this paper could be candi
dates to put to an AI system in the future.
Artificial Intelligence will certainly impact on the role of measurements and data collection
and analysis more generally for the mine ventilation professional in the future. However, this
author’s view (Brake, 2022) is that the impacts will mostly be positive. AI could provide statis
tically sound advice answering many of the questions that come up frequently (or less fre
quently) for the ventilation engineer allowing for faster, more reliable and more timely
decisions. In addition, AI could remove the need to take many of the often-tedious and time-
consuming measurements that perhaps add no real value to the mine ventilation monitoring
593
program whilst simultaneously improving the reliability of the analysis, freeing up the ventila
tion professional for more important parts of the job.
8 CONCLUSION
Given the increasing demands on the ventilation engineer’s time, the shrinking size of the ven
tilation department on most mine sites, and the 24/7 nature of mine operations, barometric
pressure surveys, when combined with volume surveys, are increasingly the most practical
method of validating ventilation models. This very useful survey technique should be part of
the repertoire of every practicing mine ventilation engineer and its specific requirements (e.g.
timing) should be included in the mine’s Ventilation Management (Control) Plan.
With regard to ventilation measurements more generally, it is likely that Big Data analysis
and Artificial Intelligence will play an increasingly useful role in many technical related mat
ters, including those in mine ventilation, over the next decade. AI is likely to develop the point
where it can provide independent and statistically sound advice regarding the types and loca
tions of measurements to be taken underground to achieve a statistically reliable estimate of
the overall validity of a ventilation model or to rapidly answer specific questions regarding
fault-finding or diagnosis of the ventilation system including under upset conditions such as
power failures, fan failures, fires or explosions.
REFERENCES
Brake D J, 2015. Quality Assurance Standards for Mine Ventilation Models and Ventilation Planning,
Proc 3rd Aust Mine Vent conf (Belle B (ed)). AusIMM. pp. 221–228.
Brake D J, 2022. Developing and maintaining mine site ventilation capability and new trends in mine
ventilation career paths. The 6th Australian Mine Ventilation Conference (Belle, B (ed). AusIMM.
Gold Coast, 10-12 Oct 2022. pp 181–204.
Copeland, R J, 2023. Artificial Intelligence. Artificial intelligence | Definition, Examples, Types, Applica
tions, Companies, & Facts | Britannica. Encyclopaedia Britanica.
Derrington, A.S. 2015. Development of a Low-cost Instrument for Barometric Pressure Surveys. The
Australian Mine Ventilation Conference. AusIMM. Chalmers, D (ed). Sydney, NSW. AusIMM. 31
Aug-2 Sep 2015.
Griffith M D & Stewart C M, 2021. Ventilation model calibration with limited survey data. Proc 18th
NAMVS, Tukkaraja, P (ed). 12–17 Jun 2021. Online. Taylor & Francis. pp 509–517.
Harris, E.J., Dalzell, R.W, Kline, R.J & Miller, E.J. 1973. A Method for Calculating Mine Ventilation
Pressure Losses Using Computers and Desk-Top Calculators, A Supplement to Information Circular
7809. Information Circular 8594. US Bureau of Mines.
McElroy, G.E & Kingery, D.S. 1957. Making ventilation-pressure surveys with altimeters, Information
Circular 7809. US Bureau of Mines.
McPherson, M.J. 2007. Subsurface Ventilation Engineering. http://www.mvsengineering.com/ 835 pp.
Rowland, J.R. 2011. Ventilation Surveys and Model Building–A Consultant’s Opinion. Australian Mine
Ventilation Conference/Sydney, NSW, 5-6 September 2011.
594
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: Accurately measuring the airflow quantity delivered by tunnel ventilation fans
during site acceptance testing can be difficult due to the fan configuration and constrained
sizes of adjoining plenums. Most axial flow tunnel ventilation fans are reversible with the
drive motor attached directly to the fan impeller. Obstructions created by motor supports pre
clude performing pitot tube traverses in the fan section and the presence of transition duct
work and sound attenuators on each side of the fan dictate that airflow measurements be
taken at the end of the fan equipment train. Point air velocity measurements taken with an
anemometer at the fan room and plenum interface are averaged to calculate the fan airflow
quantity. Limited floor space in ventilation buildings may result in narrow plenums adjacent
to fan rooms that produce non-uniform flow with high velocities concentrated on one side of
the fan ductwork, making the accuracy of anemometer point traverses problematic. This
paper presents the results of a study of alternative grid spacing of point anemometer traverses
to yield more accurate field measurements using CFD modeling of a sample fan room and
plenum arrangement.
1 INTRODUCTION
The NFPA (National Fire Protection Association) 130 Standard for Fixed Guideway Transit
and Passenger Rail Systems (2020) requires passenger rail and transit (subway) tunnels and
stations to be equipped with emergency ventilation systems that provide tenable egress condi
tions for passengers evacuating from a train fire incident. To meet the standard, a series of
shafts are typically spaced along the length of the tunnels to facilitate a “push-pull” ventilation
system utilizing reversible axial flow fans. For a fire emergency in the tunnel, passengers
would walk towards the shaft supplying air, while smoke is exhausted from a different shaft in
the opposite direction. Passenger rail stations encompassing multiple tracks with much larger
enclosed space than tunnels may be equipped with an exhaust emergency ventilation system to
maintain smoke layers above the walking height of passengers. Station platform ventilation
fans are also usually reversible to allow them to work in tandem with adjoining tunnel ventila
tion shafts if required.
As noted by Ray and Gamble (2010), since ventilation systems serve a critical life safety
function for passenger rail and transit tunnels used by the public, factory acceptance testing of
tunnel ventilation fans is required for most projects to confirm that the fans deliver the design
duty airflow. Factory testing is conducted in accordance with procedures described in AMCA
210-07, “Laboratory Methods of Testing Fans for Certified Aerodynamic Performance
Rating,” with pitot tube traverses performed across a measurement plane located a minimum
of 8.5 test duct diameters away from the fan with essentially a uniform air velocity profile
across the duct.
In addition to factory acceptance tests, site acceptance testing is also performed to verify
that tunnel ventilation fans deliver the design duty airflow quantity in the field. As shown in
DOI: 10.1201/9781003429241-61
595
Figure 1, an elevation view of a typical reversible axial fan equipment train, round-to-
rectangular transition ducts are attached to each side of the fan with sound attenuators placed
downstream of the transitions. Measuring velocity and total pressures with a pitot tube in the
fan section is not feasible since the drive motors are coupled directly to the fan impeller and
motor support structures create disturbances to airflow and obstruct insertion of the tube.
The increase in cross-sectional area through the transition ducts creates angular airflow that
also precludes using a pitot tube since the bent tip section of the pitot tube needs to be aligned
with the flow. As a result, measurement of fan total pressure in the field is not feasible. Since
parallel full height baffles in the sound attenuators create channeled airflow with no access for
measurement instrumentation, suitable locations for measuring fan airflow quantity are gener
ally limited to the ends of the fan equipment train.
Air velocity traverses taken across the end of the fan equipment train in an adjoining air plenum
can be performed by direct measurement of velocity using vane anemometers or by measuring
velocity pressure with a pitot tube and calculating the velocity. Due to the large size of the duct
work, pitot tube traverses are cumbersome to perform, requiring custom tubes long enough to
reach the center of the ducts and sufficiently stiff to avoid deflection with the fan operating.
Digital anemometers with remote readouts and attached to extension rods are easier to use and
can allow testing personnel to remain to the sides of the duct for all measurements.
Field testing specifications typically require anemometer traverses to be performed twice
across the measurement plane and averaged. Conducting continuous traverses at the plenum
interface without personnel standing in front of a portion of the duct would be difficult, and
repeatability of successive traverses would be problematic since testing contractors do not
have experience with this method. Instead, point air velocity measurements taken at the end
of the fan equipment train are averaged and multiplied by the duct cross-sectional area to esti
mate the fan airflow quantity. As noted by Jorgensen (1999), the most frequently used method
to locate traverse stations is to divide the duct into equal areas and measure at the centroid of
the subdivided areas. In determining the number of point measurements comprising the tra
verse grid, practical considerations need to be weighed against selecting a sufficient number of
locations to establish the airflow distribution exiting (or entering) the fan ductwork. A test
grid with too few velocity measurements will not accurately reflect the airflow quantity
delivered by fan, while a grid with a large number of measurement points may take an exces
sive amount of time to complete. For a facility with multiple fans operating in parallel, the
time required to complete two traverses for each fan in both exhaust and supply operation can
be substantial. McElroy (1935) recommended not less than 16 traverse points for square cross
596
sections and not less than 24 for rectangular cross sections where one dimension is twice the
other for “precise traversing methods” in major underground air currents or tests of primary
fan performance. He further noted that a minimum spacing of 0.304 m (1 ft) between grid
points to be satisfactory in mine airways, and that if closer spacing appears to be required, “it
is very probable that the conditions of turbulence would prevent a high degree of accuracy by
any method of measurement”.
Fan airflow quantity and power measurements taken during site acceptance testing recently
completed for a passenger rail station are summarized in Table 1. Point anemometer traverses
were performed at the end of the fan equipment trains for both the exhaust and supply airflow
directions for eight reversible axial fans—four, 2.59-m (102-in.)-dia., 184 kW (250 hp) fans
and four, 2.16-m (85-in.)-dia., 110 kW (150 hp) fans. Velocity measurements were taken with
a digital vane anemometer equipped with a remote readout at a total of 36 points at the end
of the 3.66-m (12-ft) by 3.66-m (12 ft) fan ductwork for the 2.59-m (102-in.)-dia. fans and at
the end of the 2.44-m (8-ft)-wide by 2.74-m (9-ft)-high ductwork for the 2.16-m (85-in.)-dia.
fans. The measurement points were located at the center of 36, 0.61-m (24-in.) by 0.61-m (24-
in.) subdivided square areas in the larger ducts. This spacing placed measurement points
0.31 m (12 in.) from the duct walls and 0.61 m (24 in.) apart in between. For the smaller ducts,
measurement points were located at the center of six-wide by six-high, 0.41-m (16-in.) by 0.46-
m (18-in.) rectangles. This grid spacing placed measurement points 0.31 m (8 in.) from the
duct side walls and 0.41 cm (16 in.) apart across the duct, and 0.23 m (9 in.) from the top and
597
bottom duct walls and 0.46 cm (18 in.) vertically apart. The grid measurement locations were
marked on the screens installed across the ends of the ductwork with electrician’s tape. The
anemometer was secured to an adjustable painter’s extension pole to allow testing personnel
to remain to the side of the ductwork while taking measurements, as shown in the photograph
provided in Figure 2. A ladder was needed to help reach some measurement stations in the
top center of the larger duct. At each measurement station, the anemometer was set up to
read continuously over a 10-second interval and the high and low velocities were averaged.
A three-phase power quality analyzer wired directly to the fan motor terminals was used to
directly measure current, voltage and power factor and calculate power.
In addition to listing the field test airflow quantities and power for each fan in both direc
tions of flow, Table 1 also displays the fan airflow quantity derived from plotting field-
measured power (corrected to the project air density) on the factory test airflow quantity vs.
bhp curves. This is considered the most accurate method of determining fan airflow in the
field for tunnel ventilation projects. The field-measured airflow quantities ranged from
approximately 112% to 132% of the airflow quantities determined from the bhp factory curves
for the 2.59-m (102-in.)-dia. fans in exhaust and 114% to 131% in supply. The field-measured
airflow quantities for the 2.16-m (85-in.)-dia. fans varied from approximately 99% to 109% of
the airflow quantities determined from the bhp factory curves in exhaust and 99% to 115% in
supply. It should be noted that the repeatability of the two traverses for each fan test was
excellent, with the differences in average air velocity between successive traverses falling in
a range of 0.1% to 3.1%. This indicates that the variations seen between the field-measured
airflow quantities and the airflow quantities determined from the factory airflow quantity vs.
bhp curves was not due to inconsistent handling of the anemometer and extension pole setup
or failing to measure the same grid point locations in successive traverses. It is also assumed
that the installed fan arrangement does not degrade the factory performance curves since the
field airflow measurements exceed those predicted from the factory curves and AMCA 201-02
(R2011) addresses “system effects” causing non-uniform flow conditions at the fan inlet or
outlet by increasing the fan system resistance, not de-rating the fan.
An additional velocity traverse was subsequently performed for SVF-4 operating in
exhaust with an increase from 36 to 64 measurement points to evaluate whether the accuracy
Table 1. Site acceptance testing airflow measurements for station ventilation fans.
EXHAUST TVF-1 TVF-2 TVF-3 TVF-4
598
of a 36-point traverse was significantly impacted by the spacing of velocity measurement
points. For the revised test, measurement points were placed in the center of 64, 0.41-m by
0.41-m (16-in. by 16-in.) subdivided duct sections. These points were within 0.20 m (8 in.) of
the duct walls and 0.41 m (16 in.) apart, compared to 0.31 m (12 in.) from the duct walls and
0.61 m (24 in.) apart for the original test traverses. Despite a 78% increase in velocity meas
urement locations, the airflow quantity measured in exhaust was 151.8 m3/s 323,000 cfm,
only 1% less than the original measurement and still 131% of the airflow quantity predicted
from power measurements.
Velocity contours across the end of the fan equipment train developed from the field test
traverse measurements are provided in Figure 3 for TVF-4 operating in supply. Field airflow
quantity measurements for TVF-4 varied the most from those predicted from power measure
ments (132.1% exhaust/131.2% supply) while field airflow quantity measurements for TVF-5
were closest to those predicted from power measurements (98.7% exhaust/99.3% supply). The
simplest explanation for these differences is that measurement points were spaced closer
together and closer to the duct walls for the smaller ductwork connected to TVF-5 versus the
larger ductwork connected to TVF-4. However, the velocity contours displayed in Figure 4
for TVF-4 supply operation indicate that the airflow is more concentrated in the upper center
to lower right side of the duct with velocities exceeding 13.2 m/s (2,600 fpm) in the upper
center. The overall average measured duct velocity was 8.8 m/s (1,736 fpm). The uneven flow
distribution may make it more difficult to accurately measure the average velocity for SVF-4
in supply.
Based on the results of the 64-point traverse for TVF-4 operating in exhaust, field airflow
measurements were not repeated for TVF-4 or any other fans with additional grid points.
Instead, it was decided to study alternative velocity traverse grid spacings using computational
fluid dynamics (CFD) analysis to evaluate the number of points that would provide the closest
measurement to the actual airflow rate of a highly skewed flow distribution. The primary goal
599
of this analysis was to use CFD as a tool to easily compute airflow quantity from skewed vel
ocity contours using different measurement patterns, not to simulate the internal airflow
through the fan equipment train with a high degree of accuracy. A three-dimensional model of
the TVF-4 fan room and adjoining plenums originally developed during construction using
AutoCAD was updated to include more detail of the fan equipment train. As illustrated in
Figure 4, representations of the fan dampers, the full height baffles installed in the sound attenu
ators located on each side of the fan and the fan drive motor and foot-mounted motor pedestal
were added to the fan equipment train portion of the CFD model. The only missing features in
the geometry compared to the physical installation were the fan impeller and wire mesh screens.
Transient simulations of TVF-4 operating in supply were performed using ANSYS Fluent soft
ware. The cases utilized the realizable k-ϵ turbulence model and the grid size was approximately
10 million polyhedral cells. The TVF-4 supply airflow quantity was set to approximately
117.5 m3/sec (250,000 cfm) to match the quantity determined from the factory performance
curves using the field-measured power. The upstream inlet and downstream outlet were modeled
as pressure boundaries, and to achieve the desired non-uniform flow at the plane of measure
ment the Fluent “fan” internal boundary condition utilized was set to swirl the flow to produce
a similar velocity pattern to that depicted in Figure 3. An acceptably skewed pattern was devel
oped after iterating the fan boundary settings. To determine if the solution was grid independ
ent, two sensitivity cases were run, and the results were nearly identical.
Air velocities were measured for four different traverse grids across the end of the ductwork
adjacent to the exhaust plenum in the CFD model: (1) 36 points, (2) 49 points, (3) 81 points
and (4) 144 points. The 36-point grid was identical to that used in the field testing, while the
81-point and 144-point grids placed the air velocity measurement points at the center of 9 by 9
and 12 by 12 equal subdivided areas of the overall 3.66-m (12-ft) by 3.66-m (12-ft) duct,
respectively. Traverse points for the 81-point grid were within 0.20 m (8 in.) of the duct walls
and 0.41 m (16 in.) apart, compared to 0.15 m (6 in.) from the duct walls and 0.31 m (12 in.)
apart for the 144-point grid. The 49-point traverse was based on the log-Tchebycheff method
that utilizes a logarithmic distribution adjacent to the duct walls and a polynomial distribution
of velocity for the interior points. Spacing of the 49-point traverse grid points is depicted in
Figure 5. As noted by Jorgensen (1999), all points are also equally weighted in this method.
The velocity traverse points were represented as 6.99-cm (2.75-in.)-dia. circles in the CFD
model that mimic the size of the digital anemometer. Area-weighted averages of velocity for
each circle were averaged across the duct cross section to determine the airflow quantity meas
ured by each traverse grid configuration.
Figure 4. Three-dimensional CFD model of TVF-4 fan equipment train and plenums.
600
Figure 5. Log-Tchebycheff traverse grid for 3.66-m (12-ft) by 3.66-m (12 ft) cross section.
Velocity contours at the measurement plane at the end of the fan equipment train adjacent to
the exhaust plenum for the CFD simulation of TVF-4 operating in supply with the 36-point
traverse grid are shown below in Figure 6. As noted previously, the Fluent “fan” internal
boundary condition utilized in the CFD model was set to swirl the flow to attempt to skew the
higher air velocities in the 11.2 m/s (2,200 fpm) to 14.2 m/s (2,800 fpm) range to the upper
center and lower right portions of the cross section as was measured in the field. Table 2 pro
vides a summary comparison of the actual airflow quantity set in the CFD model (as close as
possible to the quantity derived from the power measurement) and the airflow quantity com
puted from the traverse of the CFD simulation cross section at the end of the fan equipment
train for each of the four alternative grid spacings. An examination of the airflow data shown
in Table 2 confirms that the CFD simulations predict that the supply airflow quantity
delivered by TVF-4 would still be overestimated by using any of the four alternate grid spa
cings. The simulation of the “base case” of 36 points was somewhat closer to the assumed air
flow quantity vs. the 36 points measured in the field (124.1% of assumed flow vs. 131.2%).
Using the centroid of subdivided equal areas, increasing the number of measurement points
from 36 to 81 would reduce the “overmeasurement” of the TVF-4 supply airflow from 124.1%
to 117.2% of the assumed airflow. Increasing to 144 measurement points would further reduce
the “overmeasurement” of airflow to 114.9% of the assumed airflow. Using the log-
Techebycheff grid spacing of 49 points results in a predicted TVF-4 supply airflow quantity of
120.3% of the assumed airflow. It would appear that using the log-Techebycheff method to
space measurement locations does not provide advantages over using centroids of subdivided
equal areas since the simulation data shows that its overestimated airflow of 120.3% of the
assumed airflow for 49 measurement points falls in between the 36 and 81-point equal area
centroid spacing overmeasurements. A comparison of the data contained in Tables 1 and 2
indicates that the field measurements for the smaller 2.16-m (85-in.)-dia. fans (TVFs-5, 6, 7
and 8) are all closer to the assumed airflows (98.7% to 114.8%) than the simulated measure
ments for similar or even closer traverse point spacing for the 2.59-m (102-in.)-dia. fans.
601
Figure 6. Air velocity contours at inlet end of ductwork for CFD simulation of TVF-4 operating in
supply.
Table 2. TVF-4 actual CFD model supply airflow quantity vs. traverse CFD airflow quantity.
Actual CFD Q Traverse CFD Q
Traverse Grid m3/sec (cfm) m3/sec (cfm) Traverse CFD Q/ actual CFD Q
6 CONCLUSIONS
Obtaining accurate measurements of the airflow quantity delivered by tunnel ventilation fans
during site acceptance testing is challenging for a number of reasons, including skewed airflow
distribution through the fan caused by narrow adjoining plenums and the difficulty locating
measurement cross sections sufficiently distant from obstructions and transitions in the duct
work. Recent field test data presented for a station with eight reversible tunnel ventilation
fans showed airflow quantities measured with anemometer traverses overestimating the fan
airflow quantity determined from field power measurements applied to factory test curves by
12 to 32 percent for 250,000 cfm fans and as much as 15 percent for 125,000 cfm fans. CFD
modeling of the fan with the highest overmeasurement in the field showed a reduction in over
measurement from 24% to 15% by increasing the number of measurement points from 36 to
602
144 across 3.66-m (12 ft) by 3.66-m (12 ft) fan ductwork. Two complete traverses over the
cross section being measured are required for field testing of tunnel ventilation systems, how
ever, and performing two, 144-point traverses for each direction of flow for a large number of
fans would be time prohibitive. In the case of the 117.5 m3/sec (250,000 cfm) fans, increasing
the number of velocity measurement points from 36 to 81 points would be more practical and
result in more accurate results, although still 17 percent high based on the results of the CFD
analysis described above.
Since nearly all of the fan airflow quantities measured in the field for the sample station
exceeded the derived values using power measurements, it is clear that the lower velocities adja
cent to the duct walls are underrepresented in the traverses. Based on the CFD analysis, this
would also apply to the traverse with the largest number of measurement points (144), with
points 0.15 m (6 in.) from the duct walls. Because the CFD simulations also predict overmeasure
ment of average air velocities for the unequal airflow distribution entering the fan equipment
train, it is unlikely that anemometer error is contributing to the overmeasurement of the fan
airflow.
REFERENCES
AMCA. 2011. Publication 201-02 (R2011), Fans and Systems. Arlington Heights, IL: Air Movement and
Control Association International, Inc.
ANSI/AMCA. 2008. Standard 2010-07 (ANSI/ASHRAE Standard 51-07), Laboratory Methods of Test
ing Fans for Certified Aerodynamic Performance Rating. Arlington Heights, IL: Air Movement and
Control Association International, Inc.
Jorgensen, R. (ed). 1999. Fluid flow. In Fan Engineering, Chapter 2: 2–90. Kingsport, TN: Quebecor
World Book Services.
Jorgensen, R. (ed). 1999. Fluid flow. In Fan Engineering, Chapter 2: 2–94. Kingsport, TN: Quebecor
World Book Services.
McElroy, G.E. 1935. Engineering Factors in the Ventilation of Metal Mines: 12–13. Washington, D.C.;
U.S. Department of the Interior, Bureau of Mines.
National Fire Protection Association. 202. NFPA 130: Standard for Fixed Guideway Transit and Passen
ger Rail Systems: 130-22-130-24. Quincy, MA.
Ray, R. E. & Gamble, G.A. 2010. The benefits of conducting factory performance tests for main mine
fans. In S. Hardcastle & D.L. McKinnon (eds.), Proceedings of the 13th U.S./North American Mine
Ventilation Symposium: 529–539. Sudbury, Ontario: MIRARCO – Mining Innovation.
603
Ventilation network analysis and optimization
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
ABSTRACT: A stable and well-maintained mine ventilation system is the key to ensuring
a safe and healthy working environment for miners. A sudden, unplanned, and significant
change in airflow termed as abnormal airflow is frequently observed in mine ventilation.
Some abnormal airflows can return to normal without manual intervention; however, some
abnormal airflows may cause catastrophic accidents if left unattended. In addition, abnormal
airflow may be a consequence of an accident such as a blocked airflow route due to a roof
fall. Promptly diagnosing and locating the cause of abnormal airflow can help prevent acci
dents. Researchers at the National Institute for Occupational Safety and Health (NIOSH)
have developed a method to diagnose the cause of abnormal airflow for underground mine
ventilation systems. The purpose of this paper is to verify the developed method using experi
mental tests conducted at NIOSH’s experimental mine. The airflows were monitored by
a real-time atmospheric monitoring system installed in the experimental mine during the tests.
As demonstrated in this case study, the developed abnormal airflow diagnosing method,
based on the resistance sensitivity and matching method, has been proven to be reliable.
1 INTRODUCTION
Ventilation is the lifeline of underground mines to provide the underground workings with fresh
air in sufficient quantity and quality, dilute and remove dust and noxious gases, and regulate tem
perature and other conditions. A stable and well-maintained mine ventilation system is the key to
ensuring a safe and healthy working environment for miners. The efficiency of mine ventilation
highly relies on airflow distributions and airflow quantities in the mine ventilation system. The
airflow distributions are closely monitored and controlled during the life of an active mine.
Changes in airflow quantity or direction are not rare in the routine management of ventilation.
The airflow changes can be categorized as: 1) planned changes for the purpose of maintaining a
safe and healthy working environment; 2) periodic changes associated with atmospheric condi
tions such as barometric pressure, temperature, humidity, and so on; 3) expected temporary
changes from the interference of moving equipment; 4) sudden, unplanned, and significant
changes due to some hidden hazardous conditions or malfunction of ventilation fans. The fourth
category of airflow changes can be termed abnormal airflow if the changes in quantity exceed
a certain level. The abnormal airflow can be caused by some kinds of failures in ventilation. For
instance, unattended opening/closing of a ventilation door, airpath blockage due to roof collapse,
fan stoppage, or any other unforeseen hazardous condition. The abnormal airflows may cause
catastrophic accidents if left unattended. For example, insufficient air supplies resulting from
abnormal airflow may induce elevated levels of hazardous gas accumulation. Therefore, promptly
DOI: 10.1201/9781003429241-62
607
diagnosing and locating the cause of abnormal airflow can help prevent accidents. Researchers at
the National Institute for Occupational Safety and Health (NIOSH) have developed a method,
named as abnormal airflow diagnosis method, to assist the diagnosis of the cause of abnormal
airflows for underground mine ventilation systems (Bahrami & Zhou, 2022). Bahrami & Zhou
(2022) also verified the abnormal airflow diagnosis method using the cases generated with ventila
tion network modeling. The method was found to be reasonably accurate in locating an unknown
source airway caused by a flow resistance change in a ventilation network. However, the verifica
tion was based on model-based numerical data which tends to provide “perfect” cases without
involving any interferences from the real world. There is a need to test the proposed method using
real-world scenarios. For this reason, we have conducted a ventilation test at NIOSH’s experimen
tal mine by changing airflow distribution while monitoring of the Atmospheric Monitoring
System (AMS) and testing the abnormal airflow diagnosis method using the experimental data. In
this paper, while we refer to our earlier work (Bahrami & Zhou, 2022), the focus is to experimen
tally verify and demonstrate how to diagnose an abnormal airflow using the proposed method
with the monitored AMS data.
2 METHODOLOGY
The abnormal airflow diagnosis method referred to in this paper is based on the availability of
the sensitivity matrix for a given mine ventilation network and the matching method which
compares the monitored airflow changes with their sensitivities from the sensitivity matrix.
For a better understanding of the diagnosis procedure, a brief introduction to the sensitivity
matrix and the matching method is necessary. These have been described comprehensively by
the authors previously (Bahrami & Zhou, 2022; Zhou & Bahrami, 2022).
Where Sij is the sensitivity of airway i to the change of resistance, ΔR, in airway j; Rj is the
resistance in airway j, and Qi is the volumetric airflow rate in the airway i. Given a ventilation
network with N airways, each airway can have N sensitivities to the rest of the airways and
itself. Therefore, an N � N sensitivity matrix S can be formed for the ventilation network
(Zhou & Bahrami, 2022):
608
are detected in a ventilation system, the airflow changes can be compared to their sensitivity
columns in the sensitivity matrix for each airway to find out which airway can cause a similar
airflow change pattern and has the closest airflow change quantities. Bahrami & Zhou (2022)
employed Root Mean Square Error (RMSE) to measure how well the changed airflows match
the sensitivity of each airway. For each airway in the network, a corresponding scale factor is
determined by minimizing the corresponding RMSE. It is assumed the airway with the min
imum RMSE is the airway that has caused the abnormal airflow due to its resistance change.
For detailed information about the matching method, refer to Bahrami & Zhou (2022).
The Safety Research Coal Mine (SRCM) is one of the two experimental mines located at the
Bruceton Campus of NIOSH in Pittsburgh, PA. Despite the smaller size of entry (6.5 ft high
by 14 ft wide) compared to the operating mines nowadays, this room-and-pillar mine has
a complete ventilation network and has been frequently used for research regarding mine ven
tilation, mine fires, and other studies (Zhou et al., 2017, 2018, 2019, 2020). The main fan,
installed at the surface above the return shaft, exhausts air from the mine. Stoppings, doors,
regulators, and brattices are used in the mine to direct the airflow to the desired routes. The
quantity of main airflow getting into the mine is controlled by the main fan and a door at the
main return entry. The layout of the SRCM is shown in Figure 1.
The SRCM is equipped with a mine-wide AMS to continuously monitor the atmospheric
condition of the mine. Nine air-velocity sensors are arranged throughout the mine to monitor
the real-time air velocity (as shown in Figure 1) for the ventilation tests conducted in this
study. The sensors are all mounted close to the roof in the middle of the entry. The airflows
are continuously monitored at these locations with a server on the surface to store air-velocity
readings every minute.
609
Within the SRCM, a door in the entry connecting B Butt and A Butt remains closed for
normal ventilation (as shown in Figure 1). To create an abnormal airflow scenario, we intention
ally open this door and maintain the open status for approximately 17 hours for our testing. The
real-time air velocities at the nine AMS monitoring locations were recorded using air-velocity
sensors every minute. Figure 2 shows the monitored air velocities from Sensors S2, S3, S4, and
S5 during the time the door was open There was a large air-velocity drop at Sensor S2 from
around 500 ft/min to 250 ft/min when the door was opened. The air velocity returned to 500 ft/
min after the door was closed. Upon opening the door, a slight air-velocity increase can be seen
at Sensors S3 and S4. Sensors S1, S6, S7, S8, and S9 do not see significant air-velocity changes
upon opening and closing the door. Therefore, they are not plotted in the figure.
Figure 2. Monitored air velocities at Sensor S2, S3, S4, and S5.
As it was mentioned above, the abnormal airflow diagnosis method proposed in this paper con
sists of two major parts: one is the resistance sensitivity matrix for a ventilation network, and
the other is the matching method using RMSE. The generation of the resistance sensitivity
matrix of a ventilation network is based on one-time ventilation simulation results. Therefore,
the collaboration of the network model and the mine is critical to ensure a good application of
the abnormal airflow diagnostic method. Continuous efforts have been made to establish and
maintain a well-calibrated ventilation network model for the SRCM (Iannacchione et al., 2015;
Zhou et al., 2022). The schematic outline drawing of the SRCM ventilation network including
166 airways is displayed in Figure 3. The source airway of the abnormal airflow, where the
door is maintained open for the test, is labeled as Airway #40 in the network drawing.
Table 1 displays the comparison of the measured airflows from all nine air-velocity sensors
and simulated airflows using MFIRE, which is an open-sourced mine fire simulation program
having ventilation network simulation capability. The calculation of resistance sensitivity is
610
Figure 3. Schematic drawing of the SRCM ventilation network with airway numbers.
implemented in MFIRE as well (Zhou & Bahrami, 2022). Due to the fluctuation in the air-
velocity sensor readings, a week-long average air velocity at each sensor was used as the value
of the measured air velocity. With the known cross-sectional area for each sensor location, the
airflow rate can be obtained by multiplying the average sensor reading by the cross-sectional
area at each sensor location. However, the calculated airflow rate is based on the point reading
of the air-velocity sensor instead of the average airflow rate through the whole cross-sectional
area as the air-velocity sensor measures the air velocity where the sensor is mounted. The air-
velocity sensors used in the SRCM are typically installed at the upper section of the entry,
close to the roof or ribs, to avoid interference from the moving equipment and miners.
A correction factor for each sensor location was introduced and calculated to convert the
point airflow rate to the average airflow rate following the same procedure illustrated by Zhou
et al. (2017) and by Zhou & Bahrami (2022). The adjusted airflow rates using these correction
factors are considered the final airflow rates. The comparisons were made between the adjusted
airflow rates and the simulated airflow rates using MFIRE. The average difference for all the
sensor locations was 4.63% with the largest difference occurring at Sensor 4.
611
5 RESULTS AND DISCUSSION
By applying the derivative method implemented in MFIRE (Zhou & Bahrami, 2022), a 166-
by-166 resistance sensitivity matrix was obtained for the SRCM ventilation network. In the
sensitivity matrix, if we say the rows represent the source airways where resistance changes are
made and the columns represent the target airways whose airflow changed according to the
resistance change at each airway, an element Sij in the ith row and jth column is the rate of
airflow change in airway j responding to the resistance change in row i.
As shown in Figure 2, airflow changes occurred at Sensors S2, S3, S4, and S5 while opening
the door at Airway #40.
Table 2 summarizes the monitored airflow changes at all nine sensors and the correspond
ing airway number (#) of each sensor in the SRCM ventilation network.
In this case, the airflow change at Sensor S2 located in Airway #155 is obviously abnormal
due to the significant airflow reduction of 13,167 cfm. To find out the possible source airway
(s), whose resistance change has caused similar airflow changes in all sensor airways including
Airway #155, all the monitored airflow changes, including zero changes, at all nine sensor air
ways are set as target airflow changes. Then the columns, representing the corresponding
sensor airways, in the resistance sensitivity matrix are extracted out to be used to match the
input airflow changes. Before applying the abnormal airflow diagnosis method, each row is
screened for the sorted sensitivities to eliminate all-zero rows. An all-zero row means that the
resistance change in the airway represented by this row has no influence on all monitored air
ways, and the calculated RMSE of the all-zero row airway will be zero as well. Figure 4 is
showing the RMSE for each non-zero sensitivity row. It can be seen that Airway #40 has the
least RMSE which indicates Airway #40 is the best matching airway to cause the airflow
changes in the monitored airways, as well as the abnormal airflow change in Airway #155. In
addition to Airway #40, Airways #145, #155, and a few others can also be the source airways
causing the airflow changes in the monitored airways.
612
6 CONCLUSIONS
DISCLAIMER
The findings and conclusions in this report are those of the authors and do not necessarily
represent the official position of the National Institute for Occupational Safety and Health,
Centers for Disease Control and Prevention. Mention of any company or product does not
constitute endorsement by NIOSH.
ACKNOWLEDGEMENT
The authors wish to thank John Soles of the Pittsburgh Mining Research Division for con
ducting the full-scale AMS test.
REFERENCES
Bahrami, D. Zhou, L. 2022. A novel methodology to locate an abnormal airflow in underground mine
ventilation networks. Proceedings of the 2022 SME Annual Conference & Expo, Preprint 22-018, Salt
Lake City, February 27-March 2, 2022.
Dziurzynski, W. Krach, A. Palka, T. 2017. Airflow sensitivity assessment based on underground mine
ventiltion systems modeling. Energies, 10(10), 1451.
Griffith, M. & Stewart, C. 2019. Sensitivity analysis of ventilation models – where not to trust your simu
lation. Proceedings of the 17th North American Mine Ventilation Symposium. Montreal, Canada:
The Canadian Institute of Mining, Metallurgy and Petroleum, pp. 654–660
Iannacchione A. Joseb A. Hornc, T. Iannacchione, G. Iannacchione, S. 2015. Modeling the ventilation
network at NIOSH’s Safety Research Coal Mine. Proceedings of the 15th North American Mine Ven
tilation Symposium, June 20-24, 2015, Blacksburg, VA.
Jia, J. Jia, P. and Li, Z. 2020. Theoretical study on stability of mine ventilation network based on sensi
tivity analysis. Energy Science & Engineering, 00, 1–8. doi:10.1002/ese3.699.
Li, G. Kocsis, C. Hardcastle, S. 2011. Sensitivity anlysis on parameter changes in underground mine ven
tilation systems. Journal of Coal Science & Engineering, 17(3), 251–255.
Zhou, L. Luo, Y. Wu, F. 2007. Identificataion of abnormal airflow quantity in underground coal mines.
International Journal of Mineral Resources Engineering, 12(4), 257–265.
Zhou, L. Yuan, L. Thomas, R.A. Iannacchione, A. 2017. Determination of velocity correction factors
for real-time air velocity monitoring in underground mines. International Journal of Coal Science and
Technology, (2017) 4: 322–332.
613
Zhou, L. Yuan, L. Bahrami, D. Thomas, R.A. Rowland, J.H. 2018. Numerical and experimental investi
gation of carbon monoxide spread in underground mine fires. Journal of Fire Sciences, 36(5): 406–418.
Zhou, L. Yuan, L. Bahrami, D. Thomas, R.A. Cole, G. 2019. Study on integration of real-time atmos
pheric monitoring system monitoring data and MFIRE simulation. Proceedings of the 16th North
American Mine Ventilation Symposium, Montreal, Canada, April 28-May 1, pp. 794–803.
Zhou, L. Yuan, L. Thomas, R.A. Bahrami, D. Rowland, J.H. 2020. Determination of a mine fire inten
sity using atmospheric monitoring system in a ventilation network. Proceedings of the 2020 SME
Annual Conference & Expo, Preprint 20-005, Phoenix, AZ, February 23–26.
Zhou, L. Thomas, R.A. Yuan, L. 2022. Experimental study of improving a mine ventilation network
model using continuously monitored airflow. Mining, Metallurgy & Exploration. 39, 887–895. https://
doi.org/10.1007/s42461-022-00574-4.
Zhou, L. & Bahrami, D. 2022. A derivative method to calculate resistance sensitivity for mine ventilation
networks. Mining, Metallurgy & Exploration. 39, 1833–1899 https://doi.org/10.1007/s42461-022-00630-z.
614
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
I. Peña
Nuclear Waste Partnership LLC, Carlsbad, NM, USA
ABSTRACT: The Waste Isolation Pilot Plant (WIPP) facility, operated by the U.S. Depart
ment of Energy (DOE), is the only transuranic waste repository in the United States. The
facility is designed for the permanent disposal of transuranic radioactive waste generated
through nuclear weapons production and research. On February 14th, 2014 a continuous air
monitor (CAM) alarmed indicating a radioactive contamination event had occurred in the
active emplacement panel underground. The ventilation system automatically switched to
a filtration mode where all exhaust air is sent through a bank of filters including High Effi
ciency Particulate Air (HEPA) filters. The original filtered ventilation system had limited cap
acity. Interim projects have included adding additional fans and filtration units to the surface
and an underground fan to enhance airflow to mining areas. The DOE has decided to main
tain the WIPP facility for TRU waste disposal and to this end has authorized several signifi
cant capital projects. These projects, which are currently under construction, includes a dust
and water extraction system on surface combined with one of the world’s largest HEPA filtra
tion systems. The new Safety Significant Confinement Ventilation System (SSCVS) will be
capable of filtering up to 540,000 cfm. The dust extraction system will contain six 100,000 cfm
capable filtration units to remove salt dust and water before the air reaches the HEPA filters.
The second project is a new 28 ft diameter intake shaft 2,150 ft in depth capable of passing
500,000 cfm. New surface intake fans are designed for this shaft. An intake shaft at the site is
being reconfigured as an exhaust for the new intake fans. Both capital projects are significant
with expenditures expected to exceed $400 million. This paper describes the analyses per
formed to size the new fans, filtration units and shaft to support work activities and to main
tain proper flow alignment in the underground. The status of the capital projects will also be
described.
1 INTRODUCTION
In 1979, the United States Congress authorized the Waste Isolation Pilot Plant (WIPP) as
a research and development facility to demonstrate the safe disposal of transuranic radioactive
(TRU) waste from defense activities. TRU waste results chiefly from the production of nuclear
weapons from plutonium and enriched uranium. The term transuranic indicates that the waste
contains radionuclides with atomic numbers greater than 92, that is, greater than that of uranium.
TRU waste consists of a wide variety of contaminated materials from laboratory and production
operations, including discarded protective clothing, laboratory test equipment and reagents,
machine components, and solidified sludge. This waste has accumulated over the past 75 years
because of weapons development and production at U.S. defense facilities. The goal of WIPP is
to dispose legacy TRU waste from 22 generator sites nationwide.
The facility is managed by the Department of Energy (DOE) with Nuclear Waste Partnership,
LLC being the operating contractor. WIPP is the first geological repository in the nation for
which an application for permanent geological isolation has been approved. WIPP is located
approximately 35 miles southwest of Carlsbad, New Mexico (see Figure 1). The repository
DOI: 10.1201/9781003429241-63
615
horizon is located 655 m (2150 ft) below surface in the Salado salt horizon. The salt horizon was
formed about 250 million years ago and is 610 m (2,000 ft) thick. This salt horizon was selected
based on its stable geology, lack of water, high thermal conductivity and because of its “creep”
characteristics which will eventually encapsulate the waste in the emplacement rooms. The first
shipment of TRU waste was received in 1999.
WIPP currently consists of four shafts and eight panels for the disposal of TRU waste. The
original ventilation system at the WIPP facility utilizes three shafts for intake air, the salt-
handling shaft (SHS), waste handling shaft (WHS) and air-intake shaft (AIS). A single
exhaust shaft (ES) is utilized for the return airway at the facility. During normal operations,
fresh air is supplied primarily by the AIS with secondary intake from the SHS. Additionally,
the SHS is used for personnel and material access as well as removal of the mined salt.
Nuclear waste is lowered through the waste handling shaft which is equipped with an enclosed
headframe. The 6.1 m (20 ft) diameter waste handling shaft provides access for personnel and
equipment to the repository with limited intake downcasting the shaft. The air from the
waste-handling shaft is directly routed to the exhaust shaft after ventilating the waste-
handling shaft station. Figure 2 gives a cross-section of the WIPP facility and the original
infrastructure.
Figure 3 illustrates the original surface fan configuration at WIPP. Primary ventilation
through the facility was achieved by running of one or two of the 450-kW (600 hp) centrifugal
main fans (labeled 700A, B and C). During concurrent mining and waste-handling operations,
two of the three fans operate in parallel to provide 230 m³/s (490,000 cfm). The surface fan
arrangement also included a filtration mode. In this mode the airflow decreases to 28 m³/s
(60,000 cfm). The shift from full airflow to filtration mode is accomplished by turning off the
main fans and starting one of the three 175 kW (235 hp) 860 centrifugal standby filtration
fans. A series of isolation dampers diverts the air through the filtration system where the air is
routed through a series of high-efficiency particulate air (HEPA) filters.
The underground ventilation system is divided into four circuits. These are the North area,
where the primary equipment shop is located along with a salt disposal investigation (SDI)
experimental area, the construction circuit, TRU waste disposal circuit and the waste shaft
station circuit. Regulators strategically located in each circuit ensure airflow courses from an
area where there is no TRU waste towards the disposal circuit. Figure 4 illustrates the four
underground circuits at the beginning of 2014.
616
Figure 2. Geology at WIPP with original shaft and underground design.
3 CONTAMINATION EVENT
On February 14, 2014 a radiation detection sensor, called a continuous air monitor (CAM),
alarmed in Panel 7. The CAM measures airborne radioactivity within the exhaust of the panel
where active waste emplacement operations and final storage occurs. The CAM alarmed in the
evening when no personnel were underground at the time of the incident and triggered an
617
Figure 4. Original fan system at WIPP.
automatic ventilation system switch from one primary unfiltered fan (700C) to filtration mode
(860C) where all air exhausting the underground passes through high efficiency filter banks. The
cause of the release was due to a single drum that contained incompatible materials resulting in
a spontaneous combustion event in Room 7 of Panel 7. Figure 5 shows the damage to the TRU
container. The release contaminated certain areas in the underground, namely downstream of
where the release occurred. After the release, DOE and NWP have implemented numerous tem
porary fan systems to increase the available airflow to the underground. The interim ventilation
system (IVS) consists of two additional surface HEPA filtration units tied to dedicated exhaust
fans. In 2016 the IVS was placed into service, increasing filtered exhaust by an additional 54,000
cfm raising the total filtered exhaust to 114,000 cfm. The Supplemental Ventilation System
(SVS) fan was installed in 2017. This fan is located at the base of the AIS and pushes air into
the construction circuit. This increased the ventilation system capacity for the construction and
the north area of the repository by 46,000 cfm for a total intake volume of 180,000 cfm. With
SVS operating, airflow is routed in series from construction to the disposal circuit. The north
circuit is ventilated by using the SHS as an exhaust. The IVS and SVS were installed as tempor
ary measures until a permanent ventilation system could be engineered and installed.
618
4 FUTURE MINING AND VENTILATION PLANS AT WIPP
The original operational life of WIPP, at the time of the radiologic release, was to finish emplacing
waste in Panels 7 and 8 and in the main north/south airways between Panels 1 and 4. At that time
this would have ended the operational phase for WIPP. However, after the radiologic event, DOE
re-evaluated the amount of TRU waste currently emplaced at WIPP and concluded that the facil
ity needed to continue operation under the current waste operating permit. This resulted in
a major study to determine where new potential panels could be mined, a new surface fan config
uration and an option for a new shaft to access the facility. DOE developed two primary criteria
as a design basis, the first was that for lifecycle planning the facility was to operate to the year
2050. The second criterion was that all exhaust air from the disposal circuit will be filtered with
HEPA systems.
A series of tradeoff studies were performed to determine where additional panels could readily
be developed. The Land Withdraw Act of 1992 set aside 16 sections of land for WIPP. The cur
rent WIPP underground is in the approximate center of this area. The selected alternative was to
mine additional panels to the west of the existing repository. The primary reason for this selection
was ease of access to the west for mining, faster development to the next panel for waste handling
after Panel 8 is filled, straight forward ventilation system design, and sufficient area to the west for
panel development to achieve the design goals. The current design is to include two additional
panels (11 and 12) to the west with the ability to add additional panels.
Once the mining direction was determined, the engineering team began assessing geotechnical
considerations for the new design. The cross sectional area of new airways was slightly adjusted as
well as the distance between cross-cuts and distance between parallel airways. The panel design
was also adjusted with a greater length between the first room in the panel and the main access
west drives.
The original mains at WIPP are oriented north/south and consist of four entries. South of the
SHS, the second main from the west is a common intake to a regulator that controls air to the
disposal circuit. Air not entering this regulator continues in this drive to ventilate the construc
tion circuit where it returns in the furthest west main entry. The third entry from the west is the
disposal intake and the fourth is the disposal exhaust. In the new design a second disposal
intake is included. The reason for this is for ground control considerations. With a single dis
posal intake, any ground control function interrupted waste handling operations. With two
intakes this challenge is virtually eliminated.
The mining of salt at WIPP is primarily by full face continuous mining machines. No dust
suppression is applied during mining. Because of this, exhaust airflow from the construction
circuit can have significant airborne salt dust. In the original ventilation system design, this
dust reported to the surface and was exhausted directly to atmosphere. The new design is to
have the air pass through HEPA filters prior to discharge. In addition to dust, the exhaust
airflow can have significant moisture. This moisture is introduced during certain times of
the year. The facility is in the high desert of New Mexico and monsoonal moisture, typically
coming from the south in the summer, is very common. As the air upcasts the exhaust shaft,
this moisture condenses resulting in significant water discharging through the fans.
Water and salt dust could cause plugging of the pre-filters in the HEPA filtration system. This
would result in constant filter change outs. To minimize the impact of salt and moisture, a surface
Salt Reduction Building (SRB) is being constructed. The purpose of this system is to remove mois
ture through a demister and salt dust through filtration before it reaches the HEPA filter system.
The first ventilation analyses assumed a new surface salt and water reduction system upstream
of a HEPA filter system. These structures will replace the system shown on Figure 3. The cur
rent design included Panels 11 and 12 as shown on Figure 6. Modelling was performed incorp
orating natural ventilation pressure (NVP) for both cold (winter) and hot (summer)
conditions. Air expansion/contraction was accounted for in each intake and exhaust shaft.
Airflow requirements were based on diluting diesel exhaust, namely from the small mine
619
Figure 6. Ventilation system design for future WIPP panel development.
trucks, and the equipment used to transport the TRU waste from the Waste Shaft to the
active emplacement panel room. In addition, there are ventilation requirements for each waste
handling room of 20 m3/s (42,000 cfm). It is assumed three rooms are actively ventilated in
a normal operation. The construction area requires 35.4 m3/s (75,000 cfm) in the active mining
zones. Ventilation for the main shop and SDI is assumed at 28.3 m3/s (60,000 cfm) and the
waste shaft station at 11.8 m3/s (25,000 cfm). Bulkhead resistance values were based on his
toric values at WIPP. Airway resistances were based on measured friction factors and pres
sure/quantity measurements. Regulators were incorporated with fixed quantity branches and
adjusted until the airflow design values given above were achieved.
The base model for WIPP determined that the maximum exhaust airflow required is
245 m3/s (520,000 cfm) at a static pressure of 2.5 kPa (10.0 in. w.g.) (as measured at the collar
of the exhaust shaft. All system losses downstream of the collar were calculated by HVAC
engineers assigned to specify the primary fans).
Once the total flow and required delivered pressures were determined, design engineers began to
develop concepts for a salt reduction system and parallel HEPA filtration system. The salt reduc
tion system consists of six independent moisture and salt reduction units. Each unit is comprised
of a CFT GmbH demister upstream of a CFT GmbH Model HTKK 1/1500-2S dry scrubber.
A Spendrup 152-091-1800-C-1 600 hp fan is connected to each unit. The fan motors are on vari
able speed drives for airflow control. Each unit is capable of handling a throughput of 120,000
cfm. The fans are fitted with an outlet silencer. Figure 7 shows a SRU during factory acceptance
testing. The salt reduction units are in a separate building called the Salt Reduction Building
(SRB). The SRB has one primary intake duct tied to a manifold to each SRU. A common exhaust
plenum takes the air back to the intake to the Safety Significant Confinement Ventilation System
(SSCVS) in the New Filter Building (NFB). Photos of an installed SRU is shown on Figure 8.
620
Figure 7. Demister, filter and fan system for a single SRU (at the time of factory acceptance testing).
The SSCVS was designed based on each HEPA filter unit being capable of 12.7 m3/s (27,000
cfm). To achieve a design flow of 255 m3/s (540,000 cfm) this resulted in a minimum of 20 filter
units being available. The system was designed with 22 units. A filter unit is comprised of four
in line filters. The first filter is a MOD filter followed by a HIGH filter then two HEPA filters.
With this train contamination can be removed from the airstream. The basic design of the SRB
and SSCVS system is shown on Figure 9. This figure also shows a typical HEPA filter unit.
Mechanical engineers calculated the pressure losses through the HEPA filters, ducts, dampers
and elbows and computed a maximum possible pressure for the SSCVS fans. A total of six fans
were selected for the design, each fan capable of 70.8 m3/s (150,000 cfm) at a design pressure of
7.0 kPa (28.1 in. w.g.) (at an air density of 0.96 kg/m3 [0.06 lbm/ft3]). At maximum flow four fans
would be in operation each at approximately 63.7 m3/s (135,000 cfm). A Clarage 86.625 SW5731
CHS Fan was selected for the project. A 746 kW (1000 hp) motor on a variable speed drive is
included in the fan design. Because the fans are internal to the SRB, the fan motors will be water
cooled.
Because the resistance in each HEPA filter unit will vary, depending on the amount of load
ing, the airflow through each unit will be controlled by a flow damper connected to an airflow
sensor on the intake to the HEPA filter unit. The damper will not modulate continuously,
rather be set and compared by personnel operating the system. Differential pressures will also
be monitored across each filter bank within a HEPA unit. Increases in pressure for a constant
flow will indicate when filters should be changed out.
621
Figure 9. Isometric view of SRB and NFB system.
The SRB system is designed to have the internal fans simply overcome the losses of the SRU
and duct work. Flow through the system will be balanced with the SSCVS total flow. In the
event of a radiological release, the SRB will be isolated from the system with high efficiency
dampers and all airflow from the underground will be directed through the SRB bypass duct.
The surface SRB and SSCVS are currently under construction. Completion is expected by
late 2024. The existing fan system connects the north side of the exhaust shaft. The SSCVS con
necting duct will be to the east. Figure 3 shows the direction of the SSCVS in relation to the
existing fan system. Prior to connecting the SRB and SSCVS to the Exhaust Shaft, a series of
tests will be performed on the system. The 4.3 m (14 ft) diameter horizontal duct from the
Exhaust Shaft location will be fitted with a regulator. The regulator will be partially closed to
simulate the resistance required to support the repository. Testing will include various fan oper
ations, including full flow, filter unit configurations combined with fan operation, duct integrity,
damper integrity, filter flow adjustment other operational considerations. It is expected that by
conducting a comprehensive testing program the operation once the system is connected to the
Exhaust Shaft will be simplified.
Historically at WIPP extracting salt is a challenge as the SHS is only 3.04 m (10 ft) in diameter
and consists of a single 7.25 tonne (8 ton) skip. The shaft and its internal components were
constructed in the early 1980s. To expect this infrastructure to last the life-cycle time to 2050
would be a challenge. After the SSCVS study was commenced, DOE embarked on a second
study to evaluate the addition of a new shaft. A series of alternative studies were performed to
evaluate the location of a new shaft, the size and the long term use of the shaft.
The results of these studies concluded that a new 8.54 m (28 ft) diameter shaft would be
constructed due west of the AIS and SHS. The location is shown on Figure 6. This location
was selected since the new mining panels were being constructed to the west and the location
was readily accessible on surface. The shaft diameter was selected to accommodate a possible
new hoisting system in the future. This new hoist would replace the SHS for primary salt
removal and provide personnel and equipment access to the repository. A large conveyance
for personnel and equipment with two salt skips are envisioned for the shaft.
The shaft will be an intake to the repository providing air to the construction and disposal
circuits. A surface intake fan is to be installed on the shaft. The design has two parallel fans
installed on the surface connected to a plenum (see Figure 10). Only one fan will operate at
622
a time with the second being a fully redundant fan. The fans are designed to intake over
235 m3/s (500,000 cfm). With a forcing fan on the collar of the shaft, the airflow from the
construction panel will course to the AIS where it will exhaust on surface. This flow path sig
nificantly reduces the salt dust loading passing to the Exhaust Shaft.
To facilitate airflow exhausting the AIS a new horizontal duct with stack was constructed
on the collar of the AIS. This design will allow the dust to settle away from adjacent buildings
and parking areas. Figure 11 shows the exhaust duct and stack on the AIS. The shaft project
includes underground excavations around the shaft station and two airways connecting to the
primary north/south drifts. The shaft project completion date is expected to be midyear 2024.
Figure 10. Intake fan with plenum to new Figure 11. Exhaust stack at the AIS.
shaft.
8 SUMMARY
This paper describes two significant ventilation upgrades at the WIPP facility. The SSCVS con
sists of a 255 m3/s (540,000 cfm) exhaust fan system. The entire exhaust system will be capable
of being filtered for any radiological release. This filter system includes HEPA filters. The result
ing design is the largest HEPA filtration system in the world. A total of six 746 kW (1000 hp)
fans are included in the design with a maximum of four fans operating in parallel. A water and
salt dust removal system are being installed upstream of the HEPA filtration units to reduce
exhaust contaminants from reaching the filters. This system incorporates six demister and dry
dust filters each tied to a vane axial fan. It is designed to filter the entire maximum flow through
the system. In the event of a radiological release underground, the water and salt dust filtration
system will be turned off and the air diverted directly to the SSCVS.
In addition to the primary exhaust system upgrade, WIPP is installing a new 8.54 m (28 ft)
diameter intake shaft with surface intake fans. This system will provide upwards of 236 m3/s
(500,000 cfm) of intake air to the underground. A new exhaust path will be established to take
salt laden air from the construction system and route it to the air intake shaft which will be an
exhaust. On the surface of this shaft is an exhaust duct and stack to minimize salt dust over adja
cent buildings and the parking lot. The system can be adjusted to operate without the intake fans
on the new shaft since the SSCVS can handle the maximum expected airflow at the facility.
With this system it is expected that WIPP can maintain operations for construction of Panels
11 and 12 and beyond should DOE elect to do so. In addition, the system will be robust and
can be managed to adapt to any normal or off-normal operating condition. A future project to
install a new hoisting system in the new shaft will further enhance WIPPs operational flexibility.
REFERENCES
623
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
F.S. Bergh
Howden Africa
ABSTRACT: The COVID-19 pandemic has certainly shocked the world and made everyone
to rethink, rework and reimagine the workplace of tomorrow. The 21st-century engineer will
have to work smarter to manage the changing environment and employ technological tools to
aid them in the quest for zero harm. This paper outlines the concept of remotely and autono
mously supplying ventilation to underground workings, commonly referred to as Ventilation-On
-Demand (VOD). For ease of reference, the presentation refers to VOD. However, the proposed
system should rather be viewed as a Ventilation Optimization Solution. Through remote moni
toring of working conditions, these systems can be used extensively to improve workplace condi
tions, whilst offer financial benefits. This is made possible by employing sensors for real-time
monitoring integrated with hardware to address sub-standards in real time.
1 INTRODUCTION
Globally, the mining industry is grappling with a number of issues which include shortage of
electricity supply and associated rising costs. A casing point is a South African metalliferous
mining industry which is heavily reliant on electricity and highly sensitive to energy cost. This
begs the question: how will mines be able to supply qualitative and quantitative air in the
underground working in a cost-effective manner?
Historically, South African mines would afford to adopt the ‘brute-force’ ventilation strat
egy which means ventilating all the underground workings at full capacity 24/7 a day regard
less of activity or the underground environmental conditions. The year 2007 can be marked as
a defining point of paradigm shift as South Africa started experiencing unprecedented electri
city tariff increase and load-shedding. This is one of the major reasons why the industry is
looking at reducing the electricity cost by focusing research on energy efficient projects such
as Ventilation On Demand (VOD). The guiding principle of energy cost saving using VOD is
based on the cubic relationship between power and air quantity (P α Q3). This means any
reduction in quantity will result in significant energy savings. This research will also go a long
way in addressing the economic sustainability and indirectly reducing carbon footprint by
using energy derived from fossil fuel efficiently.
The scope of this paper is on reducing the cost of ventilation using VOD system in under
ground metalliferous mines and also improving the health and safety.
This is a literature review paper, and the layout is made up of research background, litera
ture review, health and safety, conclusion and recommendations. The literature review pro
vides views/conclusions from different literature illustrating how real time monitoring, in
conjunction with the remote control of ventilation devices can support mines on their journey
to zero harm. In addition, the literature review provides the feasibility study which provide
the cost of VOD and the payback period.
DOI: 10.1201/9781003429241-64
624
2 RESEARCH BACKGROUND AND CONTEXT
The dynamic underground environment, and the dependence on human intervention, are major
challenges in supplying adequate fresh air to areas where employees are required to travel and/
or work. This can result in a deterioration of workplace conditions and is a constant threat to
the health and safety of employees. Covid-19 forced skilled and experienced personnel to work
from home, which increased the risk of mismanagement of the ventilation systems.
Inadequate supply of fresh air has detrimental effects on the health and safety of the
employees. Mine ventilation professionals must empower themselves with the available tools
to ensure workplaces remain safe and risk-free to the health of employees.
Mines need to become more innovative and utilise the available infrastructure to distribute
the air in a more optimised and autonomous fashion by adopting optimisation systems. This
approach will minimise the dependency on human intervention and make it possible for
skilled and experienced personnel to work from home.
3 LITERATURE REVIEW
Historically, South Africa (SA) had very competitive electricity tariffs. The abundance and
rates did not warrant an energy efficient approach towards mine ventilation. This notion is
supported by Johnson (2012) in his study which concluded that South African mines are pro
duction driven where energy efficiency is not part of the mind set and culture.
Resultantly, the forbearers of mine ventilation in SA could afford to design ventilation sys
tems based on the worst-case scenario as highlighted by Kocsis, et al (2003). The approach
was to “set and forget” as the brute-force ventilation system would always ensure more than
the required air supply to the underground workings (Mochubele, 2014).
As early as 2003, Kocsis et al recognised the potential of VOD systems for energy savings,
with Johnson in 2012 and Mochubele (2013) all reaffirming the same notion. The health and
safety improvements of the systems are neglected in the literature, with only claims to ‘main
tain’ health and safety being made. The literature review focused on VOD implementation
and the associated health and safety improvements and energy savings.
625
Figure 1. VOD levels of control (Howden, 2018).
Energy savings can be achieved through the system when used with variable speed drives
(VSDs) on the fan motors instead of guide vanes which reduce quantity by reducing the fan
rotational speed. For health and safety improvements in the mine, it is worthy to invest in
level 3 of controls which require extensive monitoring of the environmental conditions and
level 4 which track the locations of personnel and machines. Level 5 refers to completely
autonomous dynamic ventilation-on-demand systems which incorporates time-based, quality-
based and event-based ventilation optimization.
626
There is a need for re-entry examinations in underground mines after blasting, to ensure the
environmental conditions are within acceptable limits before re-entry into working areas.
These are usually done manually with hand-held devices, increasing the risk of exposure to the
observer, to not just noxious fumes, but seismicity risks as well. Through VOD, examinations
can be done remotely by checking the real-time gas readings in the system. This reduces the
risk of exposure and hence improve the health and safety position of the mine.
A hypothetical mine has a re-entry standard of 45 minutes, with early-entry examinations
conducted manually by the re-entry team. The re-entry team involves the shift supervisor
(miner); the crew leader; and two safety representatives. Hand-held gas detection instruments
are used for this and the following hazards are identified for this process:
i) Short term overexposure to CO
ii) Short term overexposure to NO2
iii) Flammable gas accumulation and fire risk
The time-weighted average occupational exposure limits (TWA.OEL) represent the minimum
acceptable exposure over a longer period i.e. 8 hours. The Short Term Exposure Limit (STEL)
refer to acceptable exposure over a short period, usually 15 minutes, as long as the time-
weighted average is not exceeded. Please see the TWA and STEL values in Table 1 below.
The acute effects of CO exposure include but are not limited to mental confusion, vomiting,
loss of muscular coordination, loss of consciousness and death.
High exposures, above 70ppm, leads to victims rapidly becoming mentally confused and
can lose muscle control and consciousness without experiencing any mild symptoms
CO takes longer than all the gases to dissipate to the OEL.TWA levels. This occurs after more
than 45 minutes from blasting according to the figure below. With a re-entry time of 45 minutes
in place at the mine, the probability of acute exposures of the re-entry team is quite high.
627
Potential accumulations in muck piles and lingering accumulations caused by ineffective and
damaged ventilation infrastructure poses a risk to the re-entry team. The likelihood for this
event to occur is probable (35% - 65%) Figure 2 illustrates the dissipation trend of blast fumes.
628
The implementation of a Ventilation Optimisation Solution can:
• Schedule events for reduced flows during shift-change and breaks;
• Reduce ventilation in zones where no production is occurring
• Reduce ventilation rates in zones where ventilation exceeds the requirements
• Monitor and control air quality within regulatory limits to maintain an adequate level of
gases, airborne pollutants and temperature
• Post-blast ventilation management to minimise blast clearing times.
Through these controls, the mine can achieve cost savings and improve health and safety
through increased monitoring and control over the airflow to respond to conditions in real-
time.
A case study at Newmont’s Eleonore Gold Mine in Canada, shows the mine has achieved
• a 43% reduction in mine heating costs
• a 56% reduction in underground ventilation electricity costs; and
• a decline in surface ventilation electricity costs of 76% through VOD implementation
(Gleeson, 2019).
629
3.4.2 Ventilation optimisation of underground ventilation network and booster fans
The proposed design, with air crossings placed strategically within the underground work
ings, eliminated the need for underground booster fans. Ventilation control devices comprise
of air crossings and automated regulators in the return airway of each section. The running
cost of the booster fans add to annual savings for the mine. Figure 3 below depicts an auto
mated louvered regulator used to automatically regulate airflow flowing into the ventilation
districts.
630
Figure 4. Financial analysis USD.
discarded. VOD systems have short payback periods of about 2-3 years as they pay for them
selves through the energy cost savings achievable through this dynamic control of airflow.
These systems have proven to be critical not just for energy savings but for improving the
health and safety in underground mines as well.
VOD systems allow mines to view environmental condition data in real-time, whilst logging
changes in the ventilation system and notifying officials of sub-standards for localized evacu
ations. The system also allows the user to implement controls in real-time that will safeguard
worker health.
VOD systems help limit the time of exposure to airborne pollutants by preventing a build-
up of contaminants in production zones through re-directing noxious gases directly to return
airways. This will aid the operations existing controls in creating an environment that will go
a long way to achieving “ZERO HARM”.
5 CONCLUSION
631
6 RECOMMENDATIONS
Based on the conclusion and information contained in this technical paper, it is evident that
VOD or Ventilation Optimization solutions are advantageous to improving health and safety.
The removal of employees from dangerous work places or conditions, such as re-entry exam
inations, shows to be the most effective way of improving health and safety. The current
system only removes the re-entry team from the hazardous area during re-entry examinations.
The team will still go and work in the very same areas for an entire shift where environmental
conditions can be readily reduced to a sub-standard. Mines should explore modernisation
techniques such as remote or automated mining solutions. VOD systems are an enabler to
such solutions according to Mochubele (2014).
REFERENCES
Brake, R., 2015. A Review of Good Practice Standards and re-Entry Procedures after Blasting and Gas
Detection Generally in Underground Hardrock Mines. Virginia, 15th North American Mine Ventila
tion Symposium.
Dominguez, C. R., Martinez, I. V., Pena, P. M. & Ochoa, R. A., 2019. Analysis and evaluation of risks
in underground mining using the decision matrix risk-assessment (DMRA) technique, in Guanajuato,
Mexico. Journal of Sustainable Mining, 18(1), pp. 52–59.
Gleeson, D., 2019. International Mining. [Online] Available at: https://im-mining.com/2019/11/07/how
dens-eleonore-ventilation-on-demand-solution-wins-award/ [Accessed 25 August 2020].
Howden, 2021. VentSim Control: Eleonore. [Online] Available at: https://www.howden.com/en-us/case
studies/ventsim-control-eleonore#pdf [Accessed 16 November 2021].
Howden, V. S., 2018. Levels of control within a ventilation optimization system. Montreal: Howden
Ventsim Solutions.
Johnson D, F. C., 2012. An overview of energy efficiency in South African Hard Rock Mines, ??:
Research gate.
Kocsis, C. K., Hall, R. & Hardcastle, S. G., 2003. The integration of mine simulation and ventilation
simulation to develop a ‘Life Cycle’mine ventilation system. South African Institute of Mining and
Metallurgy, pp. 223–230.
MHSA, 1996. Mine Health and Safety Act 29 of 1996. [Online] Available at: Mine Health and Safety Act
[Accessed 21 August 2020].
Mining Weekly, 2020. Getting the best out of mining starts with policy certainty, says Gold Fields’ Hol
land. [Online] Available at: https://www.miningweekly.com/article/getting-best-out-of-mining-starts-
with-policy-certainty-holland-2020-08-20 [Accessed 21 August 2020].
Mochubele, E. M., 2014. Effects of Increasing Rejection Temperatures on Electricity Demand for Venti
lation and Cooling in Automated Metalliferous Underground Mines, Johannesburg: University of the
Witwatersrand.
Mohapi, G. & Zarske, R., 2018. Health and Safety in South African Mines: A Best Practice Report,
Frankfurt: Southern African-German Chamber of Commerce and Industry.
New Jersey Department of Health, 2000. Hazardous Substance Fact Sheet: Nitrogen Dioxide. New
Jersey, New Jersey Department of Health and Senior Services.
Oxford Reference, 2021. Quantitative Research. [Online] Available at: https://www.oxfordreference.com/
view/10.1093/oi/authority.20110803100357649 [Accessed 16 November 2021].
US Consumer Product Safety Commission, 2021. Carbo-monoxide Questions and Answers. [Online]
Available at: https://www.cpsc.gov/Safety-Education/Safety-Education-Centers/Carbon-Monoxide-
Information-Center/Carbon-Monoxide-Questions-and-Answers [Accessed 16 November 2021].
Wallace, K., Prosser, B. & Stinnette, J. D., 2014. The practice of mine ventilation engineering, California:
Mine Ventilation Services, Inc. Fresno.
Yahnke, K., 2021. Risk Assement Matrix. [Online] Available at: https://www.i-sight.com/resources/risk-
assessment-matrix/ [Accessed 16 November 2021].
632
Ventilation planning and design
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
M. Pinheiro-Harvey
Agnico Eagle, Canada
ABSTRACT: The 4 Shaft Project represents a step change for ventilation capacity at
Agnico Eagle’s Macassa Mine in Kirkland Lake, Ontario. This paper outlines the process
design, equipment selection, and experience in procuring new primary ventilation equipment
for this project. Equipment includes new 64 MMBtu/hr natural gas direct fired mine air heat
ers with integrated silencers; twin 600 hp vane axial fresh air fans; and twin 3,000 hp centrifu
gal primary exhaust fans with demisters and high-performance silencing measures. This paper
presents the description of obstacles tackled during the engineering design of the primary
mine ventilation equipment, which included space limitations on surface and noise concerns
due to proximity to local communities.
1 INTRODUCTION
In Northern Ontario, Canada, the town of Kirkland Lake has seen seven gold mining oper
ations in the 20th century. After a successful exploration campaign in the early 2000’s, Kirk
land Lake Gold re-established gold mining operations at the Macassa mine, which included
using the existing 3 Shaft as a service, skipping, and access shaft, and the 2 Shaft as secondary
egress. The ventilation circuit was driven by underground booster fans operating in parallel
within several of the historic tracked mining levels. The most recent configuration had fresh
air downcast in 3 Shaft and exhaust in 1 Shaft and 2 Shaft. After successful exploration at
depth, a new access shaft, called 4 Shaft, was approved, and began construction in 2018. To
accommodate the increased fresh air capacity, a pair of new exhaust raisebores to surface
were also approved for construction in parallel with the shaft.
Now merged with Agnico Eagle, the company’s 4 Shaft Project is now in the final stages of
construction in 2023. This shaft represents an opportunity to renew and improve mine access
in proximity to the deep SMC and Lower North orebodies including ore handling systems;
mine services and utilities; and ventilation. Additional fresh air will support mining activities
at depth which promises to improve safety and productivity.
To support production before the four-year shaft construction period was complete, the
SMC230 project was developed at Macassa in 2020 which included three new underground
fan stations and refit of three others which increased fresh airflow to operations at depth –
achieving the namesake of 230 kcfm airflow to the SMC orebody.
New ventilation infrastructure to leverage the new 4 Shaft and raisebore airways was div
ided among two projects:
1. 4 Shaft Project, which would complete the 4 Shaft Heaters and 160L Booster Fans.
2. Raisebore Exhaust Fans Project, which would construct the new fans for the raisebores.
DOI: 10.1201/9781003429241-65
635
2 PROCESS DESIGN
The life-of-mine (LOM) ventilation schematic for Macassa can be seen in Figure 1. The venti
lation circuit is negatively induced via the twin raisebore exhaust fans with the fresh air
coming from 3 Shaft and 4 Shaft. Multiple booster fans are installed mine-wide to direct the
airflow through various mining areas.
The design for the 4 Shaft Heaters, 160L Booster Fans, and Raisebore Exhaust Fans com
menced with a studies phase to develop the basis of design including tradeoffs to confirm equip
ment selection and configuration. New ventilation equipment was sized to maximize the capacity
of the new shaft and exhaust raisebores – to provide the highest degree of future flexibility.
Key process conditions forming the basis of the 4 Shaft Heater, 160L fans and the Raise
bore Exhaust Fans are summarized in Tables 1 and 2.
A schematic of the process design for the 4 Shaft Heater and 160L fans are shown in
Figure 2. The layout of various components was designed with the following key design aspects:
• 4 Shaft Heaters would include an acoustically insulated steel plenum above the Vent Shaft
with removable roof panels to allow future access for fan replacement.
• Relatively low-pressure application – with the 160L fans only having to overcome pressure
losses up to 4 Shaft Station on 160L.
• Fan flow slightly higher than 4 Shaft downcast demand to upcast 30 kcfm to 4 Shaft collar
which provides some make-up air for the headframe.
Prior to selection of equipment, a noise study was carried out to define the maximum allow
able sound power level to be specified for the heater house intake that will meet community
noise limits at the nearest receptors and determine whether the underground fan discharge
levels can meet occupational noise exposure limits at 160L and at the Shaft collar work area.
Two fan operation conditions were evaluated:
• Both fans operating under typical conditions
• Maintenance scenario where one fan runs while the second is being maintained.
636
Figure 1. Macassa life of mine ventilation schematic in long section view process design.
The Raisebore Exhaust Fans are planned to be installed on surface, which is the preferred
location, however there are concerns regarding the noise impact on the adjacent community
due to limited mitigation opportunities with a surface installation. An underground installa
tion to alleviate these concerns was considered, but ultimately the decision was made to go
forward with a surface installation due to technical and economic considerations.
This decision required modifications to the design of the fan system to include an independ
ently supported acoustic enclosure surrounding the fan casing and all ductwork before the
existing silencer, and the addition of a second silencer of similar performance in series. Both
the considered underground and chosen surface layouts are shown in Figure 3.
The VFD cutouts for the 160L fans are located downstream of the fans (as shown in
Figure 4 Figure 4). To protect the equipment inside the cutout from high circulating/turbulent
air movement induced by the passing airstream, placement of deflection shields was suggested
based on CFD analysis. A sketch of predicted velocities and net forces are shown in Figure 5.
Figure 5 Finally, a standard operating procedure (SOP) was developed for accessing 160L to
promote safety given airflow velocities around the 160L shaft station.
Demisters were selected for the Raisebore Exhaust Fan application to avoid contamination of
local soils with dirty water droplets. The outlet side of the fan was selected for the demister loca
tion, which reduced ductwork and footprint while also offering less velocity transition. In other
words, the exhaust air only requires one expansion transition to reach a low velocity required for
the demister, outlet airflow silencer, and stack. Turbulence from the fan discharge moving through
the evasé was a concern and CFD modeling was conducted to demonstrate to project teams and
demister vendor that the velocity profile was suitable for equipment integrity and performance.
637
Figure 2. Schematic showing 4 shaft heaters and 160L Booster fans in elevation view.
Figure 3. Section view of exhaust fan concept in underground chamber (a) and Selected configuration
showing elevation view of one fan on surface (b).
Equipment selection trade-off studies, which included workshops with Operations and vendor
consultations, were conducted by the project team to inform specification packages for competi
tive bidding. After clarifications, the technical bid evaluation process identified compliant
638
Figure 4. Isolation view of the underground 160L Booster Fans with VFD location on right hand side.
Figure 5. CFD study results indicating high circulating flows in left view and shield walls with improved
conditions on the right.
offerings and a single vendor, Howden, a Chart Industries Company, was selected to provide all
the packages.
Figures 6 and 7 show the general arrangement of the 4 Shaft Heaters and 160L fans. To main
tain access to the ventilation shaft for material movement, namely for 160L fan components, the
heater house is equipped with removable roof panels. Each burner will be connected to its individ
ual valve train mounted on a steel rack inside a control room adjacent to the heater house. To
remove any sort of lifting hazards, the four burner valvetrains were divided into pairs and located
along opposite walls in the control room. The control room will house a control panel with
a burner management system (BMS) as required by the Technical Safety Standards Authority
(TSSA), and an integrated PLC system which will allow the Operators to monitor and control
the system either directly from a local operator station (panel mount touchscreen) or indirectly
from any HMI location elsewhere on the site.
Split baffle-type silencers were selected for the 160L fans so that a low-profile lifting trolley
can be installed within the drift excavation which will be used for initial installation and main
tenance needs.
639
Figure 6. General arrangement for 4 Shaft Heater in Long Section across the Burner Room.
Figure 7. Elevation view arrangement showing one of the two of 160 Fan Systems.
• To fit the four 15-foot-long grid burners, a two-story heater building was used, making all
burners fully accessible without working at heights.
• Due to strict community noise limits, noise from combustion blowers and natural gas com
bustion in the burners was a concern. A field study at an existing heater was conducted so
that a realistic sound power level for these components could be incorporated into the
noise control monitoring.
• Like the 3 Shaft Heaters, the burners and burner control systems utilized similar compo
nents including flame rods for combustion monitoring, Maxon APX forced combustion air
grid burners, and Honeywell 7800 series flame safety relay.
• The plenum connecting the Vent Shaft and the heater was constructed as modular acoustic
ally insulated panels to reduce site construction time.
640
• Motor brakes were included to allow any free-wheeling fan to be brought safely to a stop.
Disc-type brakes were used with rotors mounted to shaft extensions on the motor non-
drive end.
• Fan preventative maintenance was prioritized, and a Bently Nevada 1900/65 monitor was
selected to monitor a horizontal vibration probe on each fan bearing. With two vibration
probes per fan, a single four-channel monitor can monitor both fans.
• Temperatures of both motor bearings and windings are also monitored to improve main
tenance planning.
• Variable speed drives were selected for fan modulation and are expected to have a long ser
vice life in the relatively clean conditions on 160L.
• With the relatively unobstructed and short airway these fans are responsible for, stall sen
sors were not required.
Features included on the Raisebore Exhaust Fan are as follows:
• Full condition monitoring instrumentation for 3,000 hp fan motors and fan bearing/shafts.
• Modulation with variable speed drive and option to also modulate with parallel-blade inlet
dampers. The inlet dampers also serve to isolate a fan system during maintenance or repair.
• Fitted wash bars were allocated for the stainless steel, blade-type demisters.
• Extensive noise engineering was applied to optimize noise control measures for the fans –
which includes a combination of baffle-type outlet silencers, acoustic lagging on exposed
ducting surfaces, and acoustic enclosure around the fan body.
641
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
C. Gobbs
Lundin Mining Corporation, Eagle Mine, USA
ABSTRACT: Lundin Mining operates the Eagle Mine in the Upper Peninsula of Michigan
USA. The mine is preparing to construct the new Upper Keel zone – another high-grade
nickel orebody in proximity to the existing mine. This paper provides an overview on several
aspects of the ventilation design and strategy including the options for development and con
struction, life-of-mine operations, and the innovative ability for on-shift blasting during the
construction phase while maintaining operations in the existing Eagle and Eagle East zones.
1 INTRODUCTION
Eagle Mine is owned by Lundin Mining Corporation, a global diversified base metal company
headquartered in Toronto. Eagle Mine started operations in 2014 with an expected life of
mine into 2027. The hard rock underground mine is located in the Upper Peninsula of Mich
igan, USA. Nickel/copper ore is mined using both stoping and cut-and-fill mining methods
for the Eagle and Eagle East ore bodies. A satellite ore body known as the Upper Keel deposit
is located approximately 800m east of the existing portal and approximately 430m below
surface.
2 MINE-WIDE VENTILATION
The primary ventilation circuit for the mine uses two fresh air intakes: one through the mine
portal that is equipped with fresh air fans and direct-fired heaters and a Fresh Air Raise
(FAR) that is equipped with direct-fired heaters. Fresh air from the main decline and FAR is
distributed to Eagle by two 700hp exhaust fans located on the surface, at the collar of
a Return Air Raise (RAR). The Eagle East ventilation uses a branch circuit that is powered
by two 400hp fresh air fans to “push” fresh air from the main decline and FAR to the mining
zone. All exhaust air from both zones is collected on the 265L return air drive (RAD) via
exhaust raises and exhausted out the RAR. Additional details can be found in (Witow et al.,
2019).
The Upper Keel deposit is found adjacent to the main decline, at a shallower depth than the
other existing zones. During initial construction of this main decline, muck bays were con
structed to allow the staging of development muck. Muck Bays 2 and 4 will be re-purposed to
establish development drives to access the top and bottom portions of Upper Keel, respect
ively. The proposed ventilation plan will utilize fresh air from the upper portion of the main
decline drawn off at Muck Bay 2, circulated through this zone, down a series of FARs, and
returned to the Upper Keel ramp to the main decline at Muck Bay 4. A schematic of the over
all proposed ventilation network is shown in Figure 1.
DOI: 10.1201/9781003429241-66
642
Figure 1. Eagle mine overall ventilation schematic outlining the ventilation of Eagle, Eagle East and
Upper Keel zones.
A ventilation schematic showing the ventilation of the Upper Keel is shown in Figure 2. Air
for the Upper Keel will be drawn from the Main Decline at Muck Bay 2 by the new Upper
Keel Booster Fans. Fresh air will be delivered to the Upper Keel through a set of internal
fresh air ventilation raises. All the air will exit into the Upper Keel haulage ramp and exhaust
air will travel up the decline to rejoin the remaining air from the portal that has not been
directed to the Upper Keel.
643
The ramp section between Muck Bay 2 and 4 will have a reduced flow rate due to the
Upper Keel ventilation supply “bypassing” this area. Appropriate measures will ensure portal
air flow is greater than the Upper Keel demand resulting in s sufficient air flow to support
trucking in this ramp.
The proposed Upper Keel ventilation design draws and returns air to the same branch of
the primary ventilation circuit, creating an independent circuit resulting in:
1. No impact on fan flows or pressures within the Upper Keel.
2. Exhaust air from Muck Bay 4 will recirculate back up to Muck Bay 2 and through the
Upper Keel zone.
3. No increase or decrease in air flows or pressures in the Eagle primary circuit or fan stations.
To control the ventilation within the Upper Keel zone and to maintain a simple control
schematic, all the fresh air from the raises will be directed into the Upper Keel ramp on the
lowest mining level by constructing timber bulkheads which will block airflow from each of
the internal levels with the exception of the lowest one. Air from the lowest level will enter the
Upper Keel ramp which will upcast toward the main decline. Auxiliary fans will be placed in
the ramp to ventilate each active level.
As seen in the schematics above, exhaust air from Upper Keel rejoins the fresh air circuit for
Eagle and Eagle East. On-shift blasting could allow development crews to maximize develop
ment rates in Upper Keel, where crews can blast at the face as soon as drilling and loading are
done and not wait until the end of the shift blasting window. Therefore, to enable on-shift
blasting in the Upper Keel, consideration must be made to isolate blasting gases to avoid
fumes re-entering the main decline and downstream workplaces. To achieve this, a single 54”
steel exhaust duct is proposed to pick up the exhaust air from the Upper Keel decline connec
tion in Muck Bay 4 and direct it to the 265 RAD and through the airlock doors. Planning for
blast gas clearing was informed by previous testing (Carriere et al. 2017).
This configuration is chosen based on the historical development of both Eagle and Eagle
East declines that were completed with a similar single 54” steel duct. Thus, the headroom in
the ramp is sufficient so that the duct can be installed without impacting haulage activities.
The configuration of the fans at the 265 RAD airlock doors (Figure 3) was used during the
development of the Eagle East decline and should be familiar to site operations.
644
A ventilation schematic showing the ventilation flows during blast clearing and the general
airflow is shown in Figure 4. Following an on-shift blast, the Upper Keel Blast Fans located
at the 265L RAD will be turned on to draw all the air from the Upper Keel straight to
exhaust. To ensure that no blasting gases enter the main decline, the fresh air booster fans at
the top of the raise in Muck Bay 2 will be turned down and a door installed in Muck Bay 4
will be closed. This will completely isolate the Upper Keel and ensure that any fresh air drawn
through the portal and down the Eagle decline will be free of blasting gases, allowing produc
tion activities to continue in either Eagle or Eagle East without delay.
The 80 kcfm airflow volume during blast mode is based on the practical amount of air that
can be induced at the 265L RAD by two 54”200 hp fans without the use of inline booster
fans. Having no in-line booster fans in the exhaust duct will keep all parts of this system under
negative pressure, ensuring that any leakage at duct joints or from mobile equipment damage
to a duct would not result in the escape of blast smoke.
Instrumentation for remote monitoring and control will be used for the Upper Keel Booster
Fans, Blast Door, and Upper Keel Blast Fans to ensure the safe and stable operation of the
blast mode system. In the event of any upset conditions, alarms will be programmed in the
mine’s control system, and appropriate system interlocks, including the shutdown of Upper
Keel Booster Fans, which can be programmed for additional layers of redundancy.
Initial development from Muck Bay 4 into the Upper Keel decline is proposed to occur under
a traditional forcing auxiliary ventilation arrangement until the steel exhaust duct would not
be damaged by concussion and fly rock. These initial development blasts will require end-of-
shift blasting allowing blast smoke to clear through the mine’s primary ventilation circuit,
without impacting Eagle and Eagle East operations.
The blast mode exhaust duct can be installed after development has commenced; at which
time the remaining development can utilize on-shift blasting. A fresh air fan will be required
645
to direct air to the working face and flush blast gases back up the ramp, where the exhaust
duct will collect all exhaust and direct it through the 265L RAD Airlock to the RAR. The
fresh air fan should be installed in an overlapping configuration with the intake of the exhaust
duct to prevent recirculation of blast smoke into the intake duct (Figure 5). The ducting over
lap distance has not yet been calculated – and may be limited by available space.
After completion of the first internal FAR, the Upper Keel Booster Fans will be installed,
and flow-through ventilation will be instituted. The construction of the Blast Door should be
scheduled to complete prior to the start-up of the Upper Keel Booster Fans.
The Upper Keel Booster Fan will be the driver of the ventilation circuit that begins at the top
of the FARs in Muck Bay 2. An example sketch of a potential installation for the fans is
shown in Figures 6 and 7. The fresh air booster fans will be controlled by a variable frequency
drive (VFD). The VFD will be networked to allow remote control of the fans from the
surface. This will be critical to ensure fan turndown is achieved during blast clearing.
In the blast clearing mode, the Upper Keel ventilation flow is matched to the capacity of
the exhaust duct at 80 kcfm. Based on this balance, all of the blasting fumes should be cap
tured before entering the main ramp. As an additional measure of safety, a ventilation door
(blast door) is proposed for Muck Bay 4 which will prevent an imbalance of flows and prevent
any fugitive blasting gasses from reaching the main ramp. This door will normally be open
during production and will not affect truck haulage. In blast mode, the doors will close and all
air from the zone will be captured in the exhaust duct. The blast door will be required only
after the ramp and Upper Keel FAR are connected and a flow-through circuit is made.
An overhead steel sectional door, comparable to those currently used on 265 RAD, is pro
posed for the blast door. The door system is shown in Figure 7. A higher-pressure door has
been selected to withstand upset conditions in the case the Upper Keel Booster Fans speed is
646
Figure 6. Design Sketch for upper keel booster fan station - Plan View.
Figure 7. Design sketch for upper keel booster fan station – Section view.
647
not correctly reduced or if there is an exhaust duct fan failure. This pressure would be too
high for a rubber roll-up door.
7 CONCLUSIONS
Having the Upper Keel zone located adjacent to the primary mine access presents an oppor
tunity for a simple ventilation circuit both in construction and operations phases. Challenges
for minimizing blast clearing time can be overcome with a reduced-flow blasting mode in the
zone and an isolated exhaust duct connection to the return air raise. If Operations elect to
pursue on-shift blasting, the addition of a blast isolation door can provide a failsafe
configuration.
REFERENCES
Witow, D., Gobbs, C., Lam, J. & Harris, W. (2019) Lundin Eagle East Project – Mine Ventilation
Update. 17th North American Mine Ventilation Symposium. Canada.
Carriere, R., McGuire, C., McLaren, E., Witow, D. (2017) Studying Operational Improvements in Blast
Gas Clearing Using Ventilation Control. 16th North American Mine Ventilation Symposium. Colorado,
USA.
648
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
P. Tukkaraja
Department of Mining Engineering and Management, South Dakota School of Mines, Rapid City, SD, USA
P. Chang
WA School of Mines: Minerals, Energy and Chemical Engineering, Curtin University, Australia
ABSTRACT: This paper uses vortex flow modelling to find the optimal length of eddy air
flow in dead-end gallery using air velocity, pressure and diesel particulate matter (DPM) simu
lations and field investigations. Computational fluid dynamics (CFD) modelling conducted
for four different dead-end crosscut lengths (10 m, 15 m, 20 m, and 25 m), three different
crosscut angles (45°, 90° and 135°) and different air velocities in adjacent galleries revealed
that a distinct vortex flow develops in the dead-end crosscut. Results indicated that an eddy
airflow revolved in a curved form, while the air velocity and pressure decreased towards the
centre of the vortex and DPM concentration increased towards the centre of the vortex. The
eddy airflow influence distance in a dead-end crosscut depends on the crosscut angle and air
velocity in the adjacent gallery. If the air velocity in the adjacent gallery is one m/s and the
crosscut angle is 90°, eddy airflow ventilates up to 20 m from the entrance. Though the air
velocity in the adjacent gallery is 4 m/s, eddy airflow is not ventilating the crosscut after
30 m from the crosscut entrance. The eddy flow distance is lower in obtuse-angled crosscuts
than the acute-angled crosscuts.
1 INTRODUCTION
Dead-end workings are common in underground areas during development and production
mining (García-Díaz et al., 2019). Dead-end crosscuts adjacent to the main airways, like
decline, incline, tunnel etc., are commonly used in the subsurface to install substations,
pumps, stockpiles, etc. (Morla et al., 2021, Toraño Álvarez et al., 2002). Figure 1 shows a part
of a mine near portal. From the figure, it can be observed that the substation, stockpile, mono
pump, fuel bay, underground (u/g) toilet, sump, u/g store, etc. are in dead-end tunnel/drive/
cuddy/crosscuts/gallery. These areas are located adjacent to the main air gallery and most of
them are ventilated by eddy airflow. If the dead-end cuddy length is too long, eddy air flow
may not reach the endpoint of the gallery, causing accumulation of higher temperatures or
dust or DPM concentrations.
Diesel-powered vehicles are sometimes operated in unventilated areas like parking cuddies,
stockpiles, etc. (Morla et al., 2020a). As the lower airflow in dead-end workings, it takes longer
time to disperse exhaust fumes, including DPM. The DPM particles are made of carbon,
DOI: 10.1201/9781003429241-67
649
elemental (EC) and organic (OC), ash, sulphates and nitrates (Chang et al., 2019), particles are
small in size (nm) and their densities are low (0.3 gm/cm3 to 1.2 gm/cm3) (Bugarski et al., 2004),
they do not tend to settle quickly under their own weight (Morla et al., 2020a, Morla et al., 2019).
Prolonged exposure to DPM causes adverse health effects (Ristovski et al., 2012, AIOH, 2013,
Morla and Karekal, 2017, Morla et al., 2018, Chang and Xu, 2017).
As per Australian mine regulations (AIOH, 2013, MDG, 2008, WHSR, 2022) to minimize
the adverse health effects, the 8-hour time-weighted average exposure of mine personnel to
EC is limited to 0.1 mg/m3, total carbon (TC), 0.16 mg/m3 and diesel particulate (DP), 0.2 mg/
m3. As per the Western Australian mines regulations, to dilute diesel emissions, each location
where a diesel engine operates must be ventilated with a minimum airflow of at least 0.05 m3/
s/kW of the engine capacity (WHSR, 2022). Western Australian mine regulations also indicate
(WHSR, 2022), an average air velocity of at least 0.3 m/s is to be supplied at all working areas
of the mine where persons work.
The design of eddy airflow ventilated dead-end tunnels is important to maintain regu
latory ventilation requirements. Mapping airflow patterns in dead-end crosscuts will
better understand air distribution and develop the optimum design of eddy flow venti
lated tunnels. In this paper, studies were conducted using air velocity, pressure and
DPM field and modelling investigations to find the optimum length of eddy airflow in
a dead-end crosscut.
650
2.2 CFD modelling
CFD studies have been used in the mining industry to solve various critical ventilation issues
(Morla, 2013, Morla et al., 2013, Morla et al., 2015, Chang et al., 2020a, Chang et al., 2020b,
Chang et al., 2019a, Chang et al., 2019b). The commercially available CFD package ANSYS
Fluent was used for this DPM modelling (Tanguturi et al., 2013). The CFD simulations were
carried out in the sequence of steps outlined in the following paragraphs.
Figure 2 shows a CAD model representing the experimental gallery and the dead-end cross
cut. The sampling points are at 5 m from the dead-end face. Figure 2 shows the mesh gener
ated for the volume of the experimental gallery. The finer mesh cells were also used adjacent
to gallery walls, with seven layers of cells accommodated in the boundary layers. The compu
tational domain and mesh consist of almost half a million tetrahedron-shaped computational
cells. Convergence and mesh-independent studies were conducted in this modelling. The
residual RMS error value is 10-4, and the domain has imbalances of less than 0.01. The min
imum wall Y+ value for the model is 1.
The boundary conditions of the model were considered as having an intake air velocity of 3 m/s
at 300 K of dry bulb temperature and the initial DPM concentration in the dead-end crosscut was
considered as 820 µg/m3. The transient state modelling was used to model DPM concentration dis
persion in a dead-end gallery. For this modelling, the diameters of the DPM particles are considered
between 1e-9 m to 1e-7 m with a mean diameter of 1e-8 m. DPM particles were treated as inert mater
ials and the Rosin-Rammler diameter distribution was used. For physical models, the spherical drag
law was used as a drag parameter. As the air velocity is low (below 4 m/s), the standard k-ε turbulent
model was used. For stochastic tracking, a discrete random walk model with 10 tries and a 0.15-time
scale was used. The intake air and DPM are considered as two different phases. The Eulerian-
Lagrangian approach is used whereby the gas phase (air) was solved using the Eulerian approach,
and the particle-phase (DPM) was tracked using the Lagrangian approach.
The airflow in the tunnel was treated as a turbulent flow. The standard k-ε model, which is
one of the most commonly used turbulent models, is applied to simulate the airflow. The model
transport equation for k was derived from the exact equation, while the model transport equa
tion for ε was obtained using physical reasoning and bears little resemblance to its mathematic
ally exact counterpart. In the derivation of the k-ε model, the assumption is that the flow is fully
turbulent, and the effect of molecular viscosity is negligible. As the mine air is considered as
fully turbulent flow, the k-ε model is valid for mine air. The turbulent kinetic energy, k, and its
rate of dissipation, ε, are obtained from the following governing equations (ANSYS, 2013):
651
Figure 3. Meshed model of dead- end gallery.
where Gb is the generation of turbulent kinetic energy due to buoyancy, and Gk is the produc
tion of turbulent kinetic energy due to the mean velocity gradient.
The particle flow is modelled using the Euler-Lagrange approach, where particle properties
are studied along the particle flow path (Chang et al., 2019a). These models define particle
flow by considering the various forces that act on the particle (Thiruvengadam et al., 2016).
The forces commonly encountered are the drag force between the fluid and the particle, the
lift force, the virtual mass force, etc.
The fluid phase is treated as a continuum by solving the Navier-Strokes equations, while
the dispersed phase is solved by tracking a large number of particles, bubbles, or droplets dis
persed through the calculated flow field. The dispersed phase can exchange momentum, mass,
and energy with the fluid phase. DPM particles are tracked using the Lagrangian method in
the discrete phase, and the particle or droplet trajectories are computed individually at speci
fied intervals during the fluid phase calculation. The force balance equation relates the particle
inertia with the forces acting on the particle and can be written as
�! �
Where ~ F is an additional acceleration (force/unit particle mass), FD ~
u up is the drag
force per unit particle mass, and
Here ~ u is the fluid phase velocity, �!up is the particle velocity, µ is the dynamic viscosity of
the fluid, ρ is the fluid density, ρp is the density of the particle material, and dp is the particle
diameter. Re is the relative Reynolds number, which is defined as:
The additional forces induced on the particle due to the fluid surrounding the particle due
to growth in the boundary layer is called the virtual mass force and is given by
652
Where Cvm is the virtual mass factor with a default value of 0.5, the fluid and the particle are
coupled together mathematically in the form of slip velocity.
3.1 Results of field experiment and model validation for air velocity
Figure 4 shows the results of the airflow distribution at a cross-sectional area in the crosscut.
Lower air velocity is at the middle of the gallery and air velocity towards the gallery sides is
increasing. Table 1 compares the results of field experiments and modelled air velocities at vari
ous locations shown in Figure 4 for validation purposes. To measure air velocity in the mine,
a rotating vane anemometer was used. This anemometer can measure the minimum air velocity
of 0.15 m/s; therefore, the field measurements did not consider air velocity below 0.15 m/s. It
can be observed from Table 1 that the base case simulated results were in good agreement with
the measured data in most cases. Some discrepancies between the simulated and measured
results can be due to the unevenness in the gallery wall surfaces that was not considered while
modelling. Overall, the difference varies from – 10% to + 8.6%, which are acceptable. From the
modelling studies, it is also observed that center of the vortex is 5 m from the entrance.
Note: Difference % is the difference between simulation results and test results and is calculated as (Simulated
value– Experimental value)/ experimental value) × 100.
653
3.2 Vortex flow modelling of air velocity and pressure with different crosscut lengths and angles
After the initial dead-end crosscut modelling studies were conducted with a base case model,
detailed parametric investigations were carried out on the airflow patterns in unventilated
dead-end crosscuts of different geometries (crosscut lengths, orientation with respect to the
main airflow gallery and air velocities).
Figure 5 shows the CFD modelling results of air velocity distribution for dead-end crosscuts
10 m, 15 m, 20 m, and 25 m deep and two different orientations (45° and 135°) were con
sidered for the 20 m long dead-end crosscut. For this simulation, intake air velocity was
assumed to be 3 m/s. The vortex flow patterns are plotted on a plane 1.2 m above the ground.
The air velocity in the crosscut is relatively low and the velocities are lower than 0.1 m/s in
some areas, as shown in Figure 5. The lower air velocities are at the centre of the vortex and
air velocity increases towards the periphery of the vortex. The locations of the centre of the
vortex for the 10 m, 15 m, 20 m and 25 m long crosscuts at 90° to the main gallery were found
to be at 4.5 m, 8.5 m, 10 m and 10 m from the crosscut entrance. The centres of the vortices in
the 20 m long crosscut oriented at 45° and 135° were found to be at 10 m and 6.5 m, respect
ively, from the crosscut entrance.
Figure 6 shows the pressure distribution in a dead-end crosscut. For this modelling con
sidered dead-end crosscuts of 10 m, 15 m, 20 m (90°), 25 m, 45° (20 m) and 135° (20 m), and
an air velocity in the main gallery of 3 m/s. The results show that air pressure is reducing
towards the centre of the vortex. There is higher air pressure in the main air gallery; the pres
sure difference between the main air gallery and the dead-end crosscut (ΔP) is very low and is
below 0.4 Pa. A higher air pressure exists in the main air gallery.
654
Figure 6. Pressure streamlines pattern in dead-end crosscut.
difference varies from – 2.5% to + 2.3%, which are acceptable. From the modelling studies, it
is also observed that center of the vortex is 5 m from the entrance Figure 7(a).
Note: Difference % is the difference between simulation results and test results and is
calculated as (Simulated value– Experimental value)/ experimental value) × 100.
3.4 Vortex flow modelling of DPM concentration with different crosscut lengths and angles
Figure 8 shows the DPM distribution in different dead-end crosscuts. For this modelling,
dead-end crosscuts of 10 m, 15 m, 20 m (90°), 25 m, 45° (20 m) and 135° (20 m), an air velocity
of 3 m/s in the main gallery, and an initial DPM concentration in the dead-end crosscut of
655
820 µg/m3 were considered (Morla et al., 2020b). In all cases, transient flow modelling studies
were used for 180 sec for this modelling. A higher DPM concentration is observed at the
vortex centre and DPM concentration reduces towards the ends of the vortex. The DPM con
centration is higher in obtuse-angled crosscuts than the acute-angled crosscuts.
Figure 8. DPM concentration streamline patterns in dead-end crosscut after 180 sec.
Figure 9 shows the changes in DPM concentration concerning air velocity inside the dead-
end crosscut. DPM concentration increases with decreasing air velocity. A high DPM concen
tration is at the centre of the vortex. Figures 6 and 8 show changes in DPM concentration
concerning pressure differences between the main air gallery and dead-end gallery (ΔP). DPM
concentration is increases with decreasing air pressure. A high DPM concentration is at the
vortex centre, and the lower pressure drop is at the centre of the vortex.
Figure 9. Changes of DPM concentration with air velocity for 10 m, 15 m, 20 m, 25 m, 45° and 135°
dead-end crosscut.
656
3.5 Vortex flow modelling with different air velocities in adjacent gallery
Investigations have been conducted with different air velocities in dead-end crosscut’s adjacent gal
lery. For this modelling considers a 90° dead-end crosscut of 50 m long, and air velocities in the
adjoining gallery are 0.5 m/s, 1 m/s, 2 m/s, 3 m/s and 4 m/s. Figure 10 shows the results of simula
tions studies; if the air velocity in the adjacent gallery is 0.5 m/s, vortex-shaped airflow is spread up
to 15 m in the dead-end gallery. If the air velocity in the adjacent gallery is 1 m/s, 2 m/s, 3 m/s and
4 m/s, eddy airflow distance in the dead-end gallery is 20 m, 23 m, 27 m and 30 m, respectively.
Figure 10. Air velocity contours in 50 m long dead-end crosscut with different air velocities in adjacent
gallery.
4 CONCLUSIONS
Investigations concluded that eddy airflow was formed near the dead-end crosscut entrance and
revolved in a curved form, while the air velocity and pressure decreased towards the centre of the
vortex. DPM concentration increased towards the centre of the vortex. The eddy airflow distance
and velocity in a dead-end crosscut depend on the crosscut angle and air velocity in the adjacent
gallery. If the air velocity in the adjacent gallery is 1 m/s and the crosscut angle is 90°, the eddy
flow is up to 20 m from the entrance. The eddy airflow distance is lower in obtuse-angled crosscuts
than the acute-angled crosscuts. Though the air velocity in the adjacent gallery is 4 m/s, eddy flow
is not ventilating the crosscut after 30 m from the crosscut entrance. This research provided clear
information about ventilation distribution and DPM particle flow patterns in dead-end crosscuts.
REFERENCES
AIOH 2013. Diesel particulate matter Occupational Health Issues.: (The Australian Institute of Occupa
tional Hygienists, Inc).
Bugarski, A., Schnakenberg, G., Noll, J., Mischler, S., Patts, L., Hummer, J., Vanderslice, S., Crum, M.
& Anderson, R. The Effectiveness of Selected Technologies in Controlling Diesel Emissions in an
Underground Mine - Isolated Zone Study at Stillwater Mining Company’s Nye Mine. 2004. 1–86.
Chang, P. & Xu, G. 2017. A review of the health effects and exposure-responsible relationship of diesel
particulate matter for underground mines. International Journal of Mining Science and Technology, 27,
831–838.
Chang, P., Xu, G. & Huang, J. 2020a. Numerical study on DPM dispersion and distribution in an under
ground development face based on dynamic mesh. International Journal of Mining Science and Tech
nology, 30, 471–475.
Chang, P., Xu, G., Mullins, B., Abishek, S. & Sharifzadeh, M. 2020b. Numerical investigation of diesel
particulate matter dispersion in an underground development face during key mining activities.
Advanced Powder Technology, 31, 3882–3896.
657
Chang, P., Xu, G., Zhou, F., Mullins, B. & Abishek, S. 2019a. Comparison of underground mine DPM
simulation using discrete phase and continuous phase models. Process Safety and Environmental Pro
tection, 127, 45–55.
Chang, P., Xu, G., Zhou, F., Mullins, B., Abishek, S. & Chalmers, D. 2019b. Minimizing DPM pollution
in an underground mine by optimizing auxiliary ventilation systems using CFD. Tunnelling and Under
ground Space Technology, 87, 112–121.
MDG 2008. Guideline for the management of diesel engine pollutants in underground environments.
MDG-29.
Morla, R. 2013. Optimum Inertisation Techniques for Blasting Gallery Method. 3rd International Work
shop on Mine Hazard Prevention and Control,. Brisbane: CSIRO.
Morla, R., Balusu, R., Tanguturi, K. & Khanal, M. 2013. Prediction and control of spontaneous com
bustion in thick coal seams. Coal Operators conference. Wollongong: University of Wollongong.
Morla, R., Balusu, R., Tanguturi, K. & Ting, R. 2015. Inertisation options for BG method and optimisa
tion using CFD modelling. International Journal of Mining Science and Technology, 25, 401–405.
Morla, R., Godbole, A., Karekal, S., Bhattacharjee, R. M. & Balasubrahmanyam, N. 2018. Fundamen
tal understanding of diesel-operated man riding vehicle DPM dispersion–a case study. Journal of Sus
tainable Mining, 17, 105–110.
Morla, R. & Karekal, S. 2017. Diesel particulate matter investigations in underground coal mines.
Morla, R., Karekal, S. & Godbole, A. 2020a. CFD simulations of DPM flow patterns generated by
vehicles in underground mines for different air flow and exhaust pipe directions. International Journal
of Mining and Mineral Engineering, 11, 51–65.
Morla, R., Karekal, S. & Godbole, A. 2020b. Investigation of DPM dispersion in unventilated dead-ends
using transient flow modelling. International Journal of Mining and Mineral Engineering, 11, 121–133.
Morla, R., Karekal, S., Godbole, A., Bhattacharjee, R. M., Nasina, B. & Inumula, S. 2019. Effect of
ventilation air velocities on diesel particulate matter dispersion in underground coal mines. Inter
national journal of mining and Geo-Engineering, 53, 117–121.
Ristovski, Z. D., Miljevic, B., Surawski, N. C., Morawska, L., Fong, K. M., Goh, F. & Yang, I. A. 2012.
Respiratory health effects of diesel particulate matter. Respirology, 17, 201–212.
Thiruvengadam, M., Zheng, Y. & Tien, J. C. 2016. DPM simulation in an underground entry: Compari
son between particle and species models. International Journal of Mining Science and Technology, 26,
487–494.
WHSR 2022. Work Health and Safety (mies) Regulations. In: AUSTRALIA, W. (ed.).
658
Underground Ventilation – Tukkaraja (Ed)
© 2023 The Author(s), ISBN 978-1-032-55146-3
Author Index
659
Prabhu, E. 462 Sarver, E. 221, 230 Vanderslice, S. 104
Prabhu, M. 462 Sasmito, A.P. 569 Vanegas, A. 305
Prasojo, R. 25 Sastry, B.S. 192 Vaze, M. 54
Pushparaj, R.I. 393, 399, Schafrik, S. 423
578 Schatzel, S.J. 499, 507, 529 Wallace, K.G. 615
Schult, G. 353 Wang, X. 256
Qiao, M. 330, 439 Setiawan, I. 25 Watkins, E. 499, 529
Qureshi, A. 115 Shao, S. 541 Watson, K. 129
Shaw, J.K. 60, 163 Wijayanto, K. 135
Raj, K.V. 69 Sidrow, E. 264 Witow, D. 34, 183,
Rajapaksha, R. 305 Slouka, S. 264 635, 642
Rakhimov, D. 115 Soles, J. 384
Rawlins, C.A. 149 Stachulak, J. 129 Xu, G. 274, 285, 393, 399,
Rawson, T. 305 Stewart, C. 25, 515 578
Ray, R.E. 595 Xu, M. 569
Reed, R.J. 95 Tang, W. 384
Reed, W.R. 239 Taylor, S. 256 Yan, L. 550, 559
Ren, T. 330, 439 Thakur, P.C. 431 Yantek, D.S. 550, 559
Roberts, J. 330 Thomas, R.A. 384, 607 Young, D. 129
Roghanchi, P. 296, 305, Tom, K. 200 Yuan, L. 384
541 Torkmahalleh, M. 115
Rose, C. 135 Tsai, C. 264 Zaid, M.M. 274
Rubasinghege, G. 296 Tukkaraja, P. 451, 649 Zeinulla, A. 115
Rubeli, B. 129 Zhang, H. 83
Udofia, E. 484 Zhang, P. 499
Sabanov, S. 115 Uecker, L. 305 Zhao, Z. 313
Salami, O.B. 393, 399, 578 Zheng, Y. 239, 247
Salinas, V.P. 296 Van Diest, J. 353 Zhou, L. 607
Sandink, M. 256 Van Dyke, M. 499, 529 Zychowski, K. 296
660