You are on page 1of 688





Proceedings of the Third International

Symposium on Block and Sublevel Caving
5-6 June 2014, Santiago, Chile

Ral Castro
Universidad de Chile
Copyright 2014. Universidad de Chile. All rights reserved. No part of this publication may be reproduced,
stored or transmitted in any form without the prior permission of The Universidad de Chile.


The information contained in this publication is for general education and informative purposes only.
Except to the extent required by law, the Universidad de Chile makes no representations or warranties
express or implied as to the accuracy, reliability or completeness of the information stored therein. To
the extent permitted by law, the Universidad de Chile excludes all liability for loss or damage of any kind
at all (including indirect or consequential loss or damage) arising from the information in this publication
or use of such information. You acknowledge that the information provided in this publication is to assist
you with undertaking your own enquires and analyses and that you should seek independent professional
advice before acting in reliance on the information contained therein. While all care has been taken in
presenting this information herein, no liability is accepted for errors or omissions. The views expressed
in this publication are those of the authors and may not necessarily reflect those of the Universidad de

The papers contained in this publication are for general information only, and readers are cautioned to
take advice on cave mine projects.

Photographs courtesy of Codelco.

ISBN 978-956-19-0857-4

Av. Libertador Bernardo OHiggins 1058, Santiago de Chile | Telfono: (56 2) 29782000
Universidad de Chile

The Universidad de Chile was founded in 1842 being the oldest higher education institution in Chile.
Generating, developing, integrating and communicating knowledge in all the areas of knowledge and
culture are the mission and basis of the activities of the University.

The Universidad de Chile (UCH) has also a 160-year tradition of educating mining engineers. The first
mining engineering program was created under the leadership of Andrs Bello in 1853, during the
presidency of Manuel Montt. Since then, several hundred mining engineers have been trained at the
UCH contributing greatly to the development of the Chilean mining industry. Mining engineers from the
UCH have lead important technological changes, institution and to open new horizons in the mining and
metallurgy industry. Examples of the involvement of UCH graduates are numerous including technological
developments in block caving, the generation of El Tenientes convertor and the development of heap
leaching technologies.

Today the mining training activities at the UCH are multiple and largest than the first Bellos dream.
The Mining Engineering Department is in charge of delivering undergraduates, postgraduates (master
and doctorate) and continuous mining education programs. Fundamental and applied research in mining
is achieved through the Advanced Mining Technology Center (AMTC) a multidisciplinary center aimed
to develop technology-based applied solutions for the industry. In terms of underground mining, block
caving research is conducted at the Block Caving Laboratory, where the next generation of underground
mining specialists is being trained.

The communication of knowledge is one of the missions of the University of Chile. Therefore, seminars and
publications in mining are the platforms through which we present and discuss the latest advancements
in mining related technology and knowledge. The Universidad de Chile is honored to be the organizer of
Caving 2014 and to host it here in Santiago.

The editor thanks the following people who contributed their time and expertise as reviewers of
manuscripts for the Third International Symposium on Block and Sublevel Caving held in Santiago, Chile.
Dr. Eleonora Widzyk-Capehart, Universidad de Chile, Chile
Prof. Juan Pablo Vargas, Universidad de Santiago de Chile, Chile
Prof. Javier Vallejos, Universidad de Chile, Chile
Prof. Nelson Morales, Universidad de Chile, Chile
Prof. Xavier Emery, Universidad de Chile, Chile
Prof. Yves Potvin, Australian Centre for Geomechanics, Australia
Prof. Hans Gpfert, Universidad de Chile, Chile
Dr. Matthew Pierce, Itasca, United States
Prof. Italo Onederra, University of Queensland, Australia
Prof. Leandro Alejano, Universidad de Vigo, Spain
Dr. Enrique Rubio, REDCO, Chile

Local Organizing Committee Universidad de Chile

Carolina Bahamondez
Sebastin Valerio
Mara Elena Valencia
Vernica Moller
Paula Alfaro
Marcela Muoz
Bernardita Ponce
Javier Gutirrez

International Organizing Committee
Andrzej Zablocki Atlas Copco
Dr. Matthew Pierce Itasca Consulting Group
Prof. Gideon Chitombo University of Queensland
Gustavo Reyes Hatch
Danie Burger Sandvik
Jarek Jakubec SRK
Victor Encina JRI
Alfonso Ovalle AMEC
Mauricio Larran Codelco

To be profitable, the extraction of large amounts of valuable minerals from the ground requires the use
of efficient mine technologies. Equally important is the sustainability of the operations and high safety
standards. Underground mining methods produce less impact on the environment than open pit practices.
Caving methods are also the natural replacement for open pit operations as the ore reserves near the surface
become depleted. Mine caving offers the lowest cost and highest production, provided that this method is
correctly selected and implemented for the orebodys geotechnical and geological conditions.

Australia, Canada, Chile, Indonesia, Mongolia, China, Sweden, South Africa and the USA all have cave mines.
Currently, worldwide mine caving research is being pursued within mining companies, universities and research
centers. Current research is analyzing some of the technical challenges that the block caving industry faces,

Large amount of development required in a short period of time.

Scarcity of highly qualified people.

Need for high productivity material handling systems.

Understanding and tracking of the cave and the material flow.

Mud rush and rockburst prediction and control, especially when the mud has a high grade content.

Mine costs and dilution control.

High stress conditions.

Ventilation and high temperature conditions.

Stability of the mine infrastructure.

Many operations are considering, or have decided, to use block caving as their preferred mining method.
Currently, about 400,000 ton per day are extracted by caving methods. It is estimated that this figure will
increase to a rate of 1 M ton per day by 2018. Production rates will also increase. This will present new and
exciting challenges and opportunities for the mining industry and for the R&D community.

I would like to present to you the proceedings of the Third International Symposium on Block and Sublevel
Caving, which will be held in Santiago on the 5th and 6th of June 2014 in Chile. In this Proceedings, you would
find two key notes and sixty eight technical articles written by people from all over the globe. Technical topics
include innovation, mine planning, mine geomechanics gravity flow, seismicity, production and development
planning, ventilation, blasting and case studies.

I would also like to acknowledge the people that believed in the dream of making Chile not only a center of
copper production but also a center of knowledge production: Fidel Baez, Sergio Fuentes, Ernesto Arancibia,
Gideon Chitombo, Octavio Araneda, Mauricio Larran, Marko Didyk and to the many other professionals and
friends that have contributed to the dream.

We hope that this book, the presentations and the workshops would contribute to define the state of the art
of caving and to help us think about our future, the future of mining.

Prof. Ral Castro

Co-chairman Caving 2014
Universidad de Chile

The Universidad de Chile proudly thanks and acknowledges the Principal and Major Sponsors of the
Third International Symposium on Block and Sublevel Caving


Caving 2014, Santiago, Chile



Future Challenges and Why Cave Mining Must Change

German Flores, Newcrest Mining Limited, Australia 23

Its Not Mine Safety But Mind Safety - A Henderson Approach

GK Carlson Climax Molybdenum Company, USA 53


Fracturing in the footwall at the Kiirunavaara mine, Sweden

M Nilsson Lule University of technology, Sweden
D Saiang SRK Consulting (Sweden) AB, Sweden
E Nordlund Lule University of technology, Sweden 63

Draw control strategy at the New Gold New Afton Mine

A Chaudhary New Gold, Canada
K Keskimaki New Gold, Canada
S Masse New Gold, Canada 72

Caving experiences in Esmeralda Sector, El Teniente Mine

M Orellana Codelco, Chile
C Cifuentes Codelco, Chile
J Daz Codelco, Chile 78

Undercut advance direction management at the North 3rd Panel, Rio Blanco Mine,
Divisin Andina Codelco Chile
L Quiones Codelco, Chile
C Lagos Codelco, Chile
F Ortiz Codelco, Chile
E Faras Codelco, Chile
L Toro Codelco, Chile
D Villegas Codelco, Chile 91

New growth strategy in Esmeralda Mine

N Jamett Codelco, Chile
RQ Alegra Codelco, Chile 98


Assessment of broken ore density variations in a block cave draw column as a

function of fragment size distributions and fines migration
L Dorador University of British Columbia, Canada
E Eberhardt University of British Columbia, Canada
D Elmo University of British Columbia, Canada
B Norman University of British Columbia, Canada
A Aguayo Codelco, Chile 109

Assessing the state of the rock mass in operating block caving mines: A review
D Cumming-Potvin, University of Western Australia, Australia
J Wesseloo, University of Western Australia, Australia 118

Caving 2014, Santiago, Chile

Influence of secondary fragmentation and column height on block size

distribution and fines migration reaching drawpoints
L Dorador University of British Columbia, Canada
E Eberhardt University of British Columbia, Canada
D Elmo University of British Columbia, Canada
B Norman B. University of British Columbia, Canada
A Aguayo Codelco, Chile 128

Analysis of hangup frequency in Bloque 1-2, Esmeralda Sur Mine

E Viera Codelco, Chile
E Diez Codelco, Chile 138

A 3DEC-FLAC3D Model to predict primary fragmentation distribution in Cave Mines

T V Garza-Cruz Itasca Consulting Group, Inc., USA
M Fuenzalida Itasca Consulting Group, Inc., USA
M Pierce Itasca Consulting Group, Inc., USA
P Andrieux Itasca Consulting Group, Inc., USA 146

ALCODER, challeges of paradigms in caving methods

Gl Krstulovic Geomecnica Ltda., Chile
GA Bagioli Tetra Tech Metlica, Chile 159

Characterization and synthetic simulations to determine rock mass behaviour

at the El Teniente Mine, Chile. Part I
A Brzovic Codelco, Chile
P Schachter Codelco, Chile
C de los Santos Codelco, Chile
JA Vallejos, University of Chile, Chile
D Mas Ivars Itasca Consultans AB, Sweden 171

Characterization and synthetic simulations to determine rock mass behaviour

at the El Teniente mine, Chile. Part II
JA Vallejos University of Chile, Chile
K Suzuki University of Chile, Chile
A Brzovic Codelco Chile, Chile
D Mas Ivars Itasca Consultans AB, Sweden 179


Fragmentation estimates using BCF software Experiences and pitfalls

J Jakubec, SRK Consulting Ltd., Canada 191

An alternative approach to verifying predicted fragmentation in weak rock

RN Greenwood SRK Consulting Inc., Canada
BN Viljoen SRK Consulting (Canada) Inc., Canada 201


Block Caving using Macro Blocks

S Fuentes Codelco, Chile
F Villegas Codelco, Chile 211

La Encantada: An inclined cave design

J Valencia NCL Ingeniera y Construccin, Chile
P Paredes NCL Ingeniera y Construccin, Chile
F Macas First Majestic Silver Corporation, Mexico 217

Caving 2014, Santiago, Chile


Considerations for designing a geomechanics monitoring plan for each engineering stage
AE Espinosa Codelco, Chile
P Jorquiera, Codelco, Chile
J Gltzl, Gltzl GmbH, Germany 227

Integrated support quality system at El Teniente Mine

MS Celis, Codelco, Chile
RA Parraguez, Codelco, Chile
E Rojas, Codelco Chile, Chile 234

Management indicators for the cave geometry control, El Teniente mine

J Cornejo Codelco, Chile
C Pardo Codelco, Chile 243

Geomechanical issues and concepts associated with scoping study and prefeasibility
stage of a Block/Panel Caving Project
J Daz DERK Ltda., Chile
P Lled DERK Ltda., Chile
F Villegas Codelco, Chile 250


Ciresata geotechnical evaluation and caving study, Romania

N Burgio Stratavision Pty Ltd, Australia 263

Identification of different geomechanics zones in panel caving- application to Reservas

Norte El Teniente
P Landeros Codelco, Chile
J Cornejo Codelco, Chile
J Alegra Codelco, Chile
E Rojas Codelco, Chile 271

Geostatistical evaluation of fracture frequency and crushing

SA Sguret MINES ParisTech, France
C Guajardo Codelco, Chile
R Freire Rivera Codelco, Chile 280

Geomechanical ground control in block/panel caving

J Daz DERK Ltda., Chile
Y Seplveda DERK Ltda., Chile
P Lled DERK Ltda., Chile 289


Use of experiments to quantify the flow-ability of caved rock for block caving
RE Gmez, University of Chile, Chile
R Castro, University of Chile
D Olivares, University of Chile, Chile 299

An analysis of the lateral dilution entry mechanisms in Panel Caving

PS Paredes University of Chile, Chile
MF Pineda University of Chile, Chile 307

Caving 2014, Santiago, Chile

Application of a methodology for secondary fragmentation prediction in cave mines

MA Fuenzalida Itasca Consulting Group, Inc., USA
T Garza-Cruz Itasca Consulting Group, Inc., USA
M Pierce Itasca Consulting Group, Inc., USA
P Andrieux Itasca Consulting Group, Inc., USA 318

Case Study: Improving SLC recovery by measuring ore flow with electronic markers
S Steffen Elexon Mining, Australia
P Kuiper Elexon Mining, Australia 328

Stochastic models for gravity flow: numerical considerations

WH Gibson SRK Consulting (Australasia) Pty Ltd, Australia 337

First steps in monitoring gravity flow at El Teniente Mine: installagion stage in

Block-2, Esmeralda Mine
E Viera Codelco, Chile
M Montecino Codelco, Chile
M Melndez Codelco, Chile 348

Experimental study of fines migration for caving mines

F Armijo BCTEC Engineering and Technology, Chile
S Irribarra Universidad de Chile, Chile
R Castro Universidad de Chile, Chile 356

Towards an understanding of mud rush behaviour in block-panel caving mines

ME Valencia University of Chile, Chile
K Basaure University of Chile, Chile
R Castro University of Chile, Chile
J Vallejos University of Chile, Chile 363

Statistical analyses of mud entry at Diablo Regimiento sector-El Tenientes Mine

IM Navia Universidad de Chile, Chile
RL Castro Universidad de Chile, Chile
MA Valencia, Universidad de Chile, Chile 372


Hybrid composite, a way to enhance the mechanical properties of breakable ground support
V Barrera Mining and Metallurgy Innovation Institute IM2 Codelco, Chile
P Lara Mining and Metallurgy Innovation Institute IM2 Codelco, Chile
G Pinilla Codelco, Chile
F Bez Codelco, Chile 381

Pilot tests as a tool for the design of autonomous mining systems

J Riquelme University of Chile, Chile
R Castro University of Chile, Chile
S Valerio University of Chile, Chile
J Baraqui Codelco Chile, Chile 386

Implementation of LiDAR technology to evaluate deformation field induced by panel caving

exploitation, Codelco Chile El Teniente Division
AE Espinosa Codelco, Chile
P Landeros Codelco, Chile 394

Caving 2014, Santiago, Chile

Semi-autonomous Mining Model

M Fishwick Codelco, Chile
M Telias IM2-Codelco, Chile 403

Future automated mine operation: Synergistic collaboration between humans

and automated systems
J Ruiz-del-Solar University of Chile, Electrical Engineering Dept-AMTC, Chile
E Widzyk-Capehart University of Chile-AMTC, Chile
P Vallejos University of Chile-AMTC, Chile
R Asenjo University of Chile-AMTC, Chile 415


Mine sequence optimization for Block Caving using concept of best and worst case

Fast-track Detailed Engineering for Panel Caving

JC Vienne, AMEC Internacional, Chile 437

Optimizing Hill of Value for Block Caving

A Ovalle, AMEC International, Chile
M Vera, AMEC International, Chile 442

Footprint and economic envelope calculation for Block/Panel Caving Mines under
geological uncertainty
E Vargas University of Chile, Chile
N Morales University of Chile, Chile
X Emery University of Chile, Chile 449

Determination of the best height of draw in block cave sequence optimization

F Khodayari University of Alberta, Canada
Y Pourrahimian University of Alberta, Canada 457

Block Caving strategic mine planning using Risk-Return Portfolio Optimization

E Rubio REDCO Mining Consultants, Chile 466


Numerical modelling of Pilar Norte Mine development using Abaqus

R Cabezas MVA Geoconsulta, Chile
F Garca MVA Geoconsulta, Chile
M Van Sint Jan MVA Geoconsulta, Chile
R Zepeda CODELCO, Chile 479

Geomechanical evaluation of large excavations at the New Level Mine - El Teniente

E Hormazabal SRK Consulting, Chile
J Pereira Codelco,Chile
G Barindelli, Codelco, Chile
R Alvarez SRK Consulting, Chile 486

Design of 3-D models in mining

E Crdova Codelco, Chile
P Gonzlez, Codelco, Chile
C Pardo Codelco, Chile 501

Caving 2014, Santiago, Chile


Study of the impact of rock mass preconditioning on a Block Caving Mine Operation
C Castro IM2-Codelco, Chile
F Bez Codelco, Chile
E Arancibia Codelco, Chile
V Barrera, Im2-Codelco, Chile 515

Pre-conditioning with hydraulic fracturing when and how much?

C Valderrama Pontificia Universidad Catlica de Chile-IM2 Codelco, Chile
F Bez Codelco, Chile
E Arancibia Codelco, Chile
V Barrera IM2-Codelco, Chile 525

Caving propagation and dilution control through the preconditioning technology

V Barrera Codelco, Chile
C Valderrama Codelco, Chile
P Lara IM2 Codelco, Chile
E Arancibia Codelco, Chile
F Bez Codelco, Chile
E Molina Codelco, Chile 532

Numerical analysis of preconditioning using blasting and its relationship with the geomechanical
properties of the rock mass and its interaction with Hydraulic fracturing
F Bez Codelco, Chile
E Arancibia Codelco, Chile
I Pieyro IM2 S.A., Chile
J Len IM2 S.A., Chile 538

Intensity rock mass preconditioning and fragmentation performance at the

El Teniente Mine, Chile
A Brzovic Codelco, Chile
JP Hurtado Universidad de Santiago de Chile, Chile
N Marn Codelco, Chile 547


Improved microseismic event hypocentre location in Block Caving Mines using

local earthquake tomography
J Philippe Mercier, Golder Associates, Canada
W de Beer, Golder Associates, Canada
J Pascal Mercier, Advanced GeoScience Imaging Solutions, Canada 559

Seismic risk management for underground miningprojects - Codelco Chile Divisin El Teniente
AE Espinosa CODELCO Chile Divisin El Teniente, Chile
RA Fuentes CODELCO Chile Divisin El Teniente, Chile
EG Moscoso ERDBEBEN Ltda, Chile 567

Seismic hazard analysis at the El Teniente Mine ising a clustering approach

J Cornejo Codelco, Chile
J Vallejos University of Chile, Chile
X Emery University of Chile, Chile
E Rojas Codelco, Chile 575

Caving 2014, Santiago, Chile

Modeling induced seismicity in 4D

E Cordova Codelco, Chile
M Nelson University of Utah, USA 586


Application of InSAR technologies to measure the subsidence at El Tenientes Mine

AE Espinosa Codelco, Chile
O Mora Altamira Information, Espaa
F Snchez Altamira Information, Espaa 603

Chuquicamata Underground Project subsidence analysis

A Aguayo Codelco, Chile
D Villegas Codelco, Chile 611


Methodology for up-hole drilling accuracy measurements at Kiruna SLC mine

M Wimmer LKAB, Sweden
AA Nordqvist LKAB, Sweden
D Billger Inertial Sensing One AB, Sweden 625

Analysis of geometric design in ventilation raises for Block Cave production level drifts
JP Hurtado, Universidad de Santiago de Chile, Chile
YH San Martn, Universidad de Santiago de Chile, Chile 638

Simulating the logistic of an underground mine

M Moretti Paragon Decision Science, Brazil
L Franzese Paragon Decision Science, Brazil
M Capistran Paragon Decision Science, Brazil
J Cordeiro Alkmim/AngloGold Ashanti, Brazil
B Penna Alkmim/AngloGold Ashanti, Brazil
G Mendes Alkmim/AngloGold Ashanti, Brazil 647

Engineering approach for the design and analysis of drawbell blasting in block
and panel caving
Altamirano BCTEC Ingeniera y Tecnologa SpA, Chile
R Castro Universidad de Chile, Chile
I Onederra University of Queensland, Australia 656

Analysis of induced damage due to undercut blasting

D Morales Hatch, Chile
R Olivares Codelco, Chile 665

How high a draw column in Block Caving?

C Cerrutti AMEC International, Chile
A Ovalle AMEC International, Chile
Y Vergara Universidad de Chile, Chile 674

Caving 2014, Santiago, Chile

Keynote Speakers

Keynote Speakers

Future Challenges and Why Cave Mining Must


German Flores, Newcrest Mining Limited, Australia

The evolution of the cave mining industry has been driven by the requirement to adapt to change. In the late
70s, the driver for change was when lower grade and hard ore rock was encountered, after easier caving in
near surface weaker rocks. Arguably, this major change was first introduced at Codelcos El Teniente mine
in order to mine the hard rock efficiently, safely and economically. This step change was the introduction
of mechanised panel caving based on load, haul and dump (LHDs) machines subsequently resulting in
the development of different cave mining layouts. Given the hard rock and the size of the drives needed to
accommodate the LHDs at that time, jumbos and new rock support systems were also introduced. Since
then, incremental changes have been introduced into the cave mining industry primarily to increase safety,
mining efficiencies and reduce mining costs. These have included increasing LHD capacity to handle up
to 2 m3 rocks and increasing productivity, electric LHDs to improve underground environment, and rapid
development technology in order to increase development rates and access orebodies quicker. During this
same period, semi- autonomous technology has been introduced for the purposes of increasing productivity,
safety and further reducing mining costs. Preconditioning techniques were introduced with the view to
change the characteristics of the rockmass in order to enhance the caving process, especially the cavability
and fragmentation.

The cave mining industry is now moving rapidly into a new and less certain environment where arguably,
another revolutionary change is required in order to continue sustaining the industry. The potential
challenges include technical, economical, licence to operate and human capital issues. As it was the case
in the late 70s when hard ore rock was first encountered, the industry must now change in order to sustain
itself technically and economically.

This paper, which supplements a keynote address by the author, argues that in some geotechnical
environments, future cave mining may not be effectively applied with todays practice and technology that
has evolved in the last 30 years. It is also argued that the development of future cave mining systems
can be accelerated covering a much wider range of mining conditions, requirements and even mining
philosophies. Revolutionary changes are required in order for the industry to sustain its future. This means
that the cave mining industry must change.

1 Introduction
Cave mining methods have become viable and preferred mass mining options where the objectives are
low cost and high production rates. However, the cave mining industry is now entering a potentially less
certain period where current cave mining methods may not be suitable to achieve the low cost and high
productivity objectives. This environment includes greater depths, lower average grade deposits, demand
for increased productivity, escalating mining cost (capital and operating), harder and heterogeneous rock
masses, higher stress and, in some cases, higher temperature environments. In addition, there is increasing
shortage of technical skills, becoming more difficult to access capital and communities are after higher
environmental standards. Radical changes to current practices are thus needed.

In the 70s, Codelco was successfully applying block caving methods designed for weak rock mass in
low stress environment and with relatively high grades (Ovalle 1981; Baeza et al. 1987; Kvapil et al.

Caving 2014, Santiago, Chile

1989). During this period, Codelco then encountered hard ore rock with much lower average grades.
The consequences of these two issues were: very slow cave propagation, coarse fragmentation and low
productivity leading to higher mining costs.

In order to continue achieving similar levels of productivity and profitability, step changes were required to
the block caving practices at that time. Chacn et al. (2004) discussed these step changes, which included
the introduction of the mechanised panel caving method using LHDs, suitable layouts, new support systems,
alternative undercutting sequence and underground primary crushers.

Historically, block caving had been the preferred method because of the weak rock masses being caved
at the time. The area required to achieve continuous caving in such rock masses was small (e.g. 90 m x
60 m = 5,400 m2) making block caving suitable for a wide range of conditions (Figure 1). Because hard
ore rock required much bigger footprint to achieve caving (e.g. 15,000 m2), the concept of panel caving
was introduced (Chacn et al. 2004). Larger 3.5 t capacity LHDs were introduced for the first time in
underground cave mining as a means to handle the coarse fragmentation (Haley 1982). In order to increase
the productivity from LHDs, a different horizontal extraction level layout was developed. Following a
number of trials, the El Teniente layout was created (Figure 2). The introduction of larger LHDs required
the development of bigger drives of up to 3.6 m x 3.6 m. The development of these size drives resulted in
the introduction of development jumbos and alternative support systems including grouted rebars, mesh
and shotcrete as shown in (Figure 3).

In addition to the use of LHDs, part of the strategy to manage the big rocks was the introduction of an
underground gyratory crusher to improve the efficiencies of subsequent material handling.

The above changes formed the basis of current mechanised caving. Since then, there have been a number
of incremental changes that have been introduced to further increase mining efficiencies, safety and reduce
mining costs. Such changes, which include rapid development, undercutting strategies and geometries and
material handling systems, are discussed in this paper. However, in themselves, they are not expected to
effectively deal with the future challenges listed earlier.

Additionally, the relatively near surface orebodies where current mechanised caving techniques were
developed are now being exhausted and new orebodies are increasingly been found at much greater depth
than current. Such orebodies are bringing with them new challenges.

This paper discusses the future challenges and why cave mining, in particular, must change in order to
exploit the future orebodies efficiently and economically. The need for the industry to change is reinforced
in recent published work by Ernst & Young (2014) and Deloitte (2013). They discuss business risk facing
mining and metals during 2013-2014 and the top ten issues that the mining companies will face in the
coming year, respectively. The risks discussed by these authors are consistent with those presented in this

Keynote Speakers

Figure 1 Block caving method used in weak ore rock at El Teniente mine (Sisselman 1978)

Figure 2 Panel caving method used in hard ore rock at El Teniente mine (Hartman 1992)

Caving 2014, Santiago, Chile

Figure 3 Ground support for hard ore rock (Ovalle & Albornoz 1981)

2 Incremental changes in the last 30 years

Following engineering studies, the key processes in cave mining include design, cave establishment and
production. Numerous and significant incremental changes in the last 30 years have mainly been focussed
in these areas. In this context, incremental changes refer to improvements but still within the current
practice umbrella or changes that only affect a component of the entire cave mining process. In spite of
their significance, they do not necessarily result in a complete transformation of the caving practices such
as when cave mining moved from weak ore rock to hard ore rock.

2.1 Design

Once the cavability and fragmentation have been assessed, the key design features in cave mining practices
are mining strategy, block height, extraction level layout and undercutting (strategy and geometry).

The effective design of the caving operation is pivotal to the success of any mining business. It is crucial that
proper orebody knowledge (geological, geotechnical, hydrogeological, metallurgical and environmental)
including associated uncertainties is collected early in the design process. This should be followed by
a proper and rigorous analysis in order to establish most appropriate design parameters to suit a given
orebody. This is instead of simply adopting parameters from existing operations which unfortunately
remains common practice.

Of the above design activities the key incremental changes have been in the following areas:

Keynote Speakers

2.1.1 Mining strategy

The move from weak to hard ore rock masses resulted in the change from block to panel caving strategy
as shown in Figure 4. This was due to the size of the area required to initiate the caving in harder ore rock.
As indicated earlier, the area required to achieve continuous caving in weak rock masses was small and
general less than 10,000 m2 (Flores & Karzulovic 2004). For harder rock masses, the area required to
initiate continuous caving has been reported to be as high as 25,000 m2 (Cataln et al. 2010). However,
more recently and given a number of technical and operational problems associated with large panels such
as discussed by Araneda & Sougarret (2008), there are now designs and operations utilising block caving
strategies but at a much larger scale. These are referred to as macroblocks (Aguayo et al. 2012; Madrid &
Constanzo 2013; Villegas & Fuentes 2014). The advantages of this move back to block strategy includes
better management of cave establishment, production and panel cave front stability and, in some cases,
better management of potential operational hazards (e.g. collapses). An added advantage of the macro block
concept, in cases where the orebody footprint is very large, is the ability to develop a mining strategy or
sequence to reduce the payback period of the project thereby maximising the return of the entire deposit.
Some refer to this strategy as value engineering. Figures 5 and 6 are illustrations of this concept using the
Cadia East deposit (Manca & Flores 2013).

Block caving for weak ore rock Panel caving for hard ore rock

Figure 4 Cave mining step change from block caving to panel caving method (Chacn et al. 2004)

2.1.2 Block heights

Block height is defined, in this context, as the height of the block to be caved from the extraction level to the
surface, the base of a pre-existing open pit, a level or a mined-out area above (after Brown 2003 & 2007).
Block heights have today ranged from 150 m to approximately 500 m (Flores et al. 2004). However, more
recently block heights of up to 1,000 m or slightly greater have been designed such as shown in Figure 7
(Manca & Flores 2013). The main driver behind this increase in block height has been the requirement to
exploit low grade orebodies profitably (i.e. economic consideration).

Caving 2014, Santiago, Chile

Figure 5 Value engineering vertical section (Manca & Flores 2013)

Figure 6 Value engineering plan view (after Manca & Flores 2013)

Keynote Speakers

Figure 7 Block height (Manca & Flores 2013)

2.1.3 Extraction layout

In a paper by Chacn et al. (2004) on Thirty Years Evolution of block caving in Chile, the different
extraction layouts established since caving of hard rock was encountered are discussed. These were
primarily designed to improve the efficiency of LHDs in hard rock mining with coarse fragmentation. The
layout included the herringbone design specifically adopted for Salvador mine (Figure 8) and later modified
for Andina mine (Figure 9). At the same time, Henderson operations in the USA introduced the herringbone
layout (Figure 10). Based on detailed analysis of these geometries, the El Teniente layout was introduced
for the first time in the El Teniente-4 South production sector. The advantage of this new layout was an
increase in the 5 t LHD productivity from approximately 100 to 150 tph (Chacn et al. 2004).

In the last 30 years, the most commonly used layouts are the herringbone and the El Teniente as shown
in Figures 11 and 12 (Leach et al. 2000; Botha et al. 2008). An advantage of the herringbone layout is the
LHD manoeuvrability when electric tethered machines are used. In the case of the El Teniente layout, the
advantages are the easiness of construction, the effectiveness of LHD digging (attacking the muckpile
head-on) resulting in a better ore flow into the drawpoint and the stability of the extraction level pillars. The
extraction level geometry (grid) has changed from the original 24 m x 12.5 m grid (Chacn et al. 2004) to
34 m x 20 m (Castro et al. 2012), with the most common being 30 m x 15 m (Chitombo 2010). Laubscher
(1994), however stresses that the grid size should be a function of fragmentation and the requirement to
achieve flow interaction between adjacent drawpoints.

Caving 2014, Santiago, Chile

Figure 8 Herringbone layout adapted at Salvador mine (Chacn et al. 2004)

Figure 9 Herringbone layout adapted at Andina mine (Chacn et al. 2004)

Keynote Speakers

Figure 10 Herringbone layout adapted at Henderson mine (Chacn et al. 2004)

Figure 11 Typical Herringbone layout (after Brown 2007)

Figure 12 Typical El Teniente layout (after Brown 2007)

Caving 2014, Santiago, Chile

2.1.4 Undercutting strategy

One of the most critical processes in cave mine design is undercutting. This is in terms of effectively
initiating the caving of a block or panel, ensuring earlier start of production and depending on the
undercutting sequence used, managing of stresses. Brown (2003 & 2007) emphasises that the undercutting
strategy adopted can have a significant influence on cave propagation and on the stresses induced in, and
the performance of, the extraction level installations. The three mostly used undercutting strategies are
post, pre and advanced undercutting as shown in Figure 13 (Rojas et al. 2000; Barraza & Crorkan 2000;
Barber et al. 2000). Historically, the post-undercut was used and later the most commonly used became the
advanced. Currently, there is an increasing interest in applying post-undercutting strategy (Manca & Flores
2013). The reason for this shift is mainly to reduce the interaction between the activities associated with
cave preparation (i.e. undercut and extraction level development) and those associated with production. The
main benefit of this is the reduction of the overall cave establishment time.

Figure 13 Undercutting strategies post, pre and advanced undercutting (after Brown 2007)

2.1.5 Undercutting Geometry

With respect to geometry, the high undercut (sublevel caving ring geometry), narrow and flat and, narrow
and inclined options have been used as shown in Figures 14, 15 and 16 (Jofr et al. 2000; Barraza &
Crorkan 2000; Flores et al. 2004; Silveira 2004). The advantages and disadvantages of these options have
been debated widely in the industry and have included the following:

Keynote Speakers

1. High undercuts (up to 20 m) have historically been the most common geometry used for
undercutting as shown in Figure 14 (Ovalle & Albornoz 1981; Manca & Flores 2013). They are
relatively flexible in that they can be drilled with a range of drilling equipment and hole diameters.
In addition, high undercut can be used to produce tonnage with finer fragmentation for the mill at
the beginning of the caving process. However, a common problem associated with high undercuts
is blast hole deviation and hole loss resulting in poor ring breakage necessitating redrilling of the
undercut rings.

2. As the term implies, a flat undercut is formed by using flat lying drill holes rather than fans or
steeply inclined holes (Figure 15). As a result, the undercut is narrow with a height equal or
slightly greater than that of the drill drives (e.g. 4 m). The advantages of narrow flat undercut as
reported by Butcher (2000a) include that they produce higher advance rates because less drilling
and charging is required and reduce the magnitudes of the induced stresses which may otherwise
cause problems. However, the major disadvantage of narrow flat undercut is the potential of the
formation of pillars (remnants) due to either blast hole loss and deviation or confined blasting
conditions arising from inadequate cleaning of the previously blasted undercut rings. In addition,
coarse cave fragmentation is generally encountered earlier in the natural caving process (Leiva &
Duran 2003).

3. In order to offset one of the disadvantages of the narrow flat geometry (i.e. assuring complete
breakage), the narrow inclined undercut was introduced (Figure 16). This allowed easier flow
of the blasted material in the inclined section of this geometry. In spite of this, the problem of
encountering coarse cave fragmentation during earlier caving still remains. (Calder et al. 2000).

Because of the disadvantages associated with flat undercut discussed earlier and in particular poor undercut
breakage and formation of remnant pillars with the potential of causing collapses in the extraction level
below, some operations are now implementing and/or reconsidering high undercuts but utilising better
drilling and blasting practices and technologies (Manca & Flores 2013; Manca & Dunstan 2013).

Figure 14 High undercut geometry (after Manca & Flores 2013)

Caving 2014, Santiago, Chile

Figure 15 Narrow and flat undercut geometry (after Brown 2007)

Figure 16 Narrow and inclined undercut geometry (after Brown 2007)

2.2 Cave establishment

Cave establishment includes activities associated with mine set-up with those associated with cave set-up as
described in Manca & Flores (2013) and shown in Figures 17 and 18. More specifically, the associated key
mining activities in cave establishment include access development as well as associated material handling
systems; mining services (ventilation, power, water); extraction, undercut and haulage level development;
civil works (permanent ground support and concrete roadways); drawbell opening and undercutting rate.
Where preconditioning is applied, this activity becomes integral part of the cave establishment.

Keynote Speakers

Figure 17 Mine set-up strategies (after Manca & Flores 2013)

Figure 18 Cave establishment (after Manca & Flores 2013)

Of the above cave establishment activities the key incremental changes have been in the following areas:

Caving 2014, Santiago, Chile

2.2.1 Access Development

In the last 10 years, the focus on access development has been on single heading rapid development but
still using conventional drilling and blasting methods. The advance rates using such methods have been of
the order of 160 m/month (Willcox 2008). As part of this focus, rapid development technology has been
introduced leading to advance rate of up to 265 m/month with a record of 311 m/month in a single heading
(access decline) for a 5.5 m x 6.0 m decline access mined at a gradient of 1:7 (Willcox 2008; Zablocki &
Nord 2012) as shown in Figures 19 and 20. The use of mechanical excavators (TBM, road header) for rapid
access development continues to be an active area of R&D.

Figure 19 Rapid development (Flores & Logan 2008)

Figure 20 Rapid development (Willcox 2008)

Keynote Speakers

2.2.2 Undercut, extraction and haulage level development

The rapid development technology has been extended to footprint development, which includes multiple
heading undercut, extraction and haulage level development. An average development rate of 540 m/month
for a single long round jumbo, has been achieved for such headings (Manca & Flores 2013).

2.2.3 Civil Works

In cave mining, civil works refers mainly to permanent ground support and concreting of roadways. With
respect to permanent ground support associated with drawpoint support, the changes have mainly been
associated with the number and type of steel sets used in conjunction with cable support and concreting.
The number of steel sets have been reduced from as many as 7 down to two for a single drawpoints,
however there have been cases where no steel sets have been used as shown in Figure 21 (Bartlett 1992;
Rojas et al. 1992; Golden & Fronapfel 2008; Dunstan & Popa 2012). Recently, Andina and El Teniente
operations have trialled the use of pre-fabricated support systems for drawpoint to reduce the installation
time by 50% as shown in Figure 22 (Fuenzalida & Baraqui 2012).

Figure 21 Drawpoint support with and without steel sets

Figure 22 Drawpoint using prefabricated support (Fuenzalida & Baraqui 2012)

Caving 2014, Santiago, Chile

Shotcrete was introduced as part of the support system in the late 70s (Wilson 2000). In most cases it
was used in conjunction with mesh (Dolipas 2000). The incremental change associated with shotcreting
included the thickness 50 mm to 100 mm and the introduction of steel fibre reinforcement. At the same
time, there were significant improvement in the delivery and spraying systems.

In the 80s, the practice of concreting extraction level roadways (extraction drives and drawpoint drives) was
introduced. This was purely to achieve high speed tramming and therefore high LHD productivity as well
as operator comfort (Butcher 2000b; Duffield 2000). The incremental change in concreting of roadways
was the number of layers and the strength of the concrete used. Current practice includes a bottom layer of
25 to 30 MPa and an upper layer of 70 to 85 MPa as shown in Figure 23.

Figure 23 Roadways design (Duffield 2000)

2.2.4 Drawbell opening

The opening of the drawbells was traditionally a very lengthy process requiring up to 3 to 4 blasting stages
(Music & San Martin 2012). However, there have been cases where the process has been significantly

With the introduction of accurate small diameter drill rigs, the onset of electronic detonators, emulsion
products and large diameter blind hole drilling, drawbell blasting has now been reduced to a single step
blasting as shown in Figure 24 (Silveira et al. 2005; Dunstan & Popa 2012). As a result of these changes,
drawbell heights have been increased from around 10 m to up to 18 m and opening rates from around 3 to
up to 12 drawbells/month are being achieved (Silveira 2004; Casten et al. 2008; Manca & Flores 2013).

2.2.5 Undercutting Rate

The incremental change associated with undercutting has been the rate of undercutting expressed in m2/
month due to the improvements in drilling and blasting technologies. The most recent industry benchmark
for block and panel caves indicated that the undercutting rate is in the range of 2,000 to 4,000 m2/month
for low undercuts (Chitombo 2010). Some of the current operations have achieved undercutting rates in the
range of 4,000 to 6,000 m2/month. This has enabled rapid cave establishment and reduction of the ramp-up
time (Silveira 2004; Manca & Flores 2013).

Keynote Speakers

Figure 24 Drawbell fired in a single blasting

2.2.6 Preconditioning

Preconditioning was an incremental change introduced to better manage the caving process. The intent
was to alter the characteristics of hard rockmass such that it behaves similar to a weak rockmass (van As &
Jeffrey 2000; Chacn et al. 2004; van As et al. 2000). In cave mining, the processes of interest are cavability
and fragmentation (Sougarret et al. 2004, Cataln et al. 2010; 2012). In addition to the enhancement of the
caving processes, preconditioning is also being used to manage seismicity (Araneda & Sougarret 2008).
The techniques currently used are hydraulic fracturing, confined blasting and combination of the two as
shown in Figure 25. Cataln et al. 2012 refers to the latter as intensive preconditioning.

Figure 25 Intensive preconditioning (after Manca & Flores 2013)

Caving 2014, Santiago, Chile

Preconditioning for caving applications remains a very active area of research. From technical and
operational perspectives, the objective is to engineer the volume of hard rock to be caved in order to achieve
faster cave propagation rates and therefore higher draw rates, finer fragmentation resulting in less hang-
ups and secondary breakage activities, reduce area required to initiate caving and reduce the magnitude of
the seismicity due to caving. From a business perspective, these benefits will translate into shorter ramp
up times, more continuous production process, smaller underground primary crushers and therefore lower
mining costs. From a safety point of view, preconditioning should enable better management of the stresses
resulting in safer working conditions.

2.3. Production

With respect to production, the most significant incremental changes have been associated with the

2.3.1 LHD capacity and type

The LHD capacity has progressively been increased from 3.5 t in the 80s to current 21 t as shown in Figures
26. This was for the purposes of handling coarse fragmentation and achieving higher productivity (Stevens
& Acua 1982).

3.5 t LHD ( 17 t LHD (Cadia Valley Operations February 2013)

Figure 26 LHD capacities from 3.5 tonnes to 21 tonnes

The two LHD types, currently used are diesel and electrics as shown in Figure 27. The diesel LHDs are the
most commonly used and the electric, which was introduced in the last 10 to 15 years, are more suited to the
herringbone layout and are becoming more popular given the mining environmental restrictions (i.e. diesel
particulate). Electric LHDs are currently all tethered but there is active research to develop untethered

Keynote Speakers

Figure 27 Diesel 21 t and electric 14 t Sandvik LHDs (

2.3.2 Automation

The idea of automated LHDs (semiautonomous) was introduced in the early 2000s in order to reduce the
exposure of underground workers to severe and unsafe mining environment or conditions including hotter,
wet muck or mudrush conditions, dusty, noisy and seismic prone areas (Gustafson et al. 2013; Schunnesson
et al. 2009; Metsnen 2004; Schweikart & Soikkeli 2004; Varas 2004). In addition, automated LHDs were
introduced as a potential means of achieving consistent productivity however this remains an active area
of R&D. The application of fully automated LHDs is however yet to be achieved and remains an area of
active research by different suppliers.

With respect to underground automated trucks, there is only one known and documented case (Burger &
Cook 2008; Cook et al. 2008). Figure 28 shows the semiautonomous LHD and truck used in cave mining.

Figure 28 Semiautonomous underground LHD and truck (Burger 2006; Cook et al. 2008)

2.3.3 Drawpoint secondary breakage

A number of incremental changes have been made in this area and have ranged from the use of a single
boom jumbo in combination with either explosives or penetrating cone fracture (PCF), specialised high
reach drill rig, water cannon to mobile rock breakers (see Figure 29). Additional research is been carried
out on the use other more exotic techniques such as plasma rock breakage and pulse water jet. The goal in
this area is to develop systems, which can be deployed rapidly with minimal evacuation, ventilation and
therefore much less production interruptions (Singh 1998; Moss et al. 2004).

Caving 2014, Santiago, Chile

Jumbo with a single boom Commando

Mobile rock breaker Water cannon

Figure 29 Drawpoint secondary breakage systems used in caving operations

2.3.4 Crushing and tipping arrangements

The conventional primary crushing systems in hard rock cave mining are gyro and jaw (Calder et al. 2000;
Casten et al. 2000; Botha et al. 2008). In the last 10 years, jaw-gyro crushers (Duffield 2000; Manca &
Dunstan 2013) and mineral sizers have been introduced as shown in Figure 30 (Arancibia et al. 2012;
Fuenzalida et al. 2012). With respect to mineral sizers, the biggest change has been the ability to crush rock
above 200 MPa. In addition, developments are being made to achieve throughputs higher than 2,500 tph
using jaw-gyro crushers.

Gyratory crusher Jaw crusher (Flores et al. 2007)

Keynote Speakers

Jaw- gyratory crusher Sizer

(Cadia Valley Operations April 2013) (LEstrange 2009)

Figure 30 Crushing systems

The tipping arrangements have evolved from single to multiple as shown in Figure 31 and from tipping
into a hopper-feeder-crusher to tipping directly into the crusher (Krek et al. 2008; Manca & Dunstan 2013).

Figure 31 Tipping arrangements at Cadia East mine with 4-tipping points (Manca & Flores 2013; Cadia
Valley Operations September 2013)

2.3.5 Draw Rate

Draw rate is the rate at which caved ore is drawn from individual drawpoints or a group of adjacent
drawpoints and it is expressed in millimetres per unit time or tonnes per area per time period (mm/day or
t/m2-day). The incremental change has been an increase of draw rate from 25 mm/day to 115 mm/day at
cave initiation. The draw rates established for Cadia East during the cave initiation (up to 30% of the block
height) vary from 115 mm/day to 280 mm/day with an average of 190 mm/day. Higher than 30% to the top
of the block, the draw rates vary from 280 mm/day to 400 mm/day with an average of 320 mm/day. This
increase in draw rates is being attributed by some to impact of preconditioning (Manca & Flores 2013).

2.3.6 Main Material Handling System

The main material handling systems to the surface stockpile or to the mill used in cave mining are trucks,
trains, shafts and more recently conveyors as shown in Figure 32 (Tyler et al. 2004; Botha et al. 2008;
Brannon et al. 2008; Ferguson et al. 2008; Pinochet, et al. 2012; Sinuhaji et al. 2012). The associated

Caving 2014, Santiago, Chile

incremental changes with respect to the main material handling systems have been mainly in the areas of
capacity, depth and speed (Taljaard & Stephenson 2000; Brannon et al. 2012).

Train system (Flores et al. 2007) Truck system (Willcox 2008)

Conveyor belt system Shaft system

(Cadia Valley Operations April 2013) (Moss 2004)

Figure 32 Material handling systems used in cave mining

2.4. The impact of the incremental changes

The incremental changes that have occurred in the last 30 years and discussed above, have arguably been
significant in terms of increasing safety, mining efficiencies and productivity as well as reducing costs during
cave mining of hard rock. Collectively, these changes have also enabled the industry to effectively mine
in conditions that would otherwise have been uneconomic using conventional methods and practices. In
addition, the moderate depths orebodies where current mechanised caving techniques have been developed
are now being exhausted and new orebodies are increasingly been found at depth much greater than current.
Such orebodies are bringing with them new challenges.

The incremental changes designed to get more efficiently out of the 1970-80s step change have nevertheless
not resulted in a generational transformation in cave mining. The future challenges constitute a radical shift
from current experiences and therefore necessitate radical changes in order to effectively mitigate the risks
that the cave mining business may face in the future. These challenges can be categorised under the broad
topics of technical, economical, licence to operate and human capital. Now is the time to make fundamental
and dramatic changes.

Keynote Speakers

3 Future Challenges
The future high risk environments or challenges identified in this paper and with the potential to significantly
impact on the efficiency, economic and sustainability of cave mining include the following:

3.1. Challenge 1: technical

The technical challenges that need to be overcome include those associated with greater depths, lower
average grade deposits and meeting the demand for increased productivity.

3.1.1 Greater depths

The issues impacted by depths and that need to be overcome are acquisition of reliable deposit knowledge,
access to the orebody, harder rock, higher rock stresses, extreme work environment (e.g. ventilation,
temperature and humidity), higher demand for power, longer distances for material transport to surface and
effective working hours due to the transport the personnel from and to surface. Engineering solutions are
required in order to be able to mine under these new conditions which are outside current practices.

3.1.2 Lower average grade deposits

It is widely recognised that the future will predominantly be associated with the mining of lower grade
deposits. Cave mining operating costs are often not reported in the public domain, however these costs have
arguably been escalating in recent times using current practices and estimated to be in the range to USD 7/t
to USD 12/t. In order to economically mine future lower grade deposits, will necessitate the development
and application of technologies, such as discuss later, in order to ensure that operationally, cave mining
remains low costs (e.g. <USD 5.0/t). Such low cost will enable mining of low grade deposits which would
otherwise be considered uneconomic using today operating practices.

3.1.3 Demand for increased productivity

Cave mining will continue experiencing the demand for increased productivity (Ernst & Young, 2013). This
will be exacerbated by the mining of lower grade deposits. In order to be economical under these conditions
will require the industry to at least double its current best productivity from a single footprint as discussed
in paper by Wellman et al. (2012). Such productivity levels can only be achieved through application of new
technologies for continuous and automated production systems.

Broadly speaking, the technical challenges associated with depth, low grades and demand for increased
productivity; suggest that cave mining cannot be business as usual. Fundamental changes are required in
order to effectively address these challenges. As such, it is proposed that:

1 Methods are developed to enable safe and rapid access to the deep orebodies as well as rapid
footprint establishment. With respect to single heading access development, the current mining
industrys best using drilling and blasting techniques assisted by long run drilling jumbos has been
reported to average 265 m/month (Willcox 2008). During the same time, tunnels constructed using
civil engineering mechanical excavators have reported averages of up to 670 m/month (Cigla et
al. 2001). Whilst the conditions and requirement associated with civil tunnels are significantly
different from the mining tunnelling requirements, considerable effort needs to be made to develop
mechanical excavation technology for mining to approach similar development rates as civil
tunnels. It should be possible to also adapt the same technologies for rapid footprint establishment

Caving 2014, Santiago, Chile

2 The current industrys practices to enhance cavability and fragmentation include the use of
hydrofracturing and confined blasting techniques and more recently the combination of the two
(Cataln 2012). The true effects of these techniques continue to be a subject of research in mining.
More effective preconditioning methods should be developed in order to achieve unconstrained
caving rates, better fragmentation and therefore better cave drawpoint flow with much higher
degree of confidence. These methods could include ultra preconditioning and unconfined blasting.

3 The current production systems are batch in nature because of the use of LHDs as both a digging
and tramming machine. In addition, with current layouts, the LHD spends more time tramming
than digging. This results in a batch process with low productivity (e.g. 120 to 150 tph). To achieve
the demand for higher productivity, continuous rather than batch systems need to be developed.
This could include the use of compact, flexible and mobile machines integrating the loading
and crushing processes linked to conveyor systems. Alternatively, systems such as the rock flow
continuous production system could be adopted (Steinberg et al. 2012).

4 The current layouts such as El Teniente and Herringbone were developed to improve LHD
productivity (Ovalle 1981, Chacn 2004). Alternative layouts to enable effective implementation
of continuous mining systems need to be developed.

The above changes combined are shown in Figure 33 and they should help reduce escalating mining costs
thereby improving the cave mining business of the future in spite of the new technical challenges discussed.

Figure 33 Conceptual future cave mining continuous mining system 3.2. Challenge 2: economical

Keynote Speakers

Both, capital and operating cave mining costs have been escalating in the last few years and this trend
is expected to continue in the future. The result has been a reduction in the margin due to the lower
commodities prices and the increasing mining costs. This challenge needs to be given priority if large
scale and low grade caving operations are to be viable. Ernst & Young 2013 discuss strategies to manage
cost escalation that could equally be considered for the cave mining business. The suggested strategies
include reviewing the soundness and probabilistic risk of different cave mining projects, better definition
of orebody characteristics, greater analysis of the return of investment from individual projects, building
mines in phases and in smaller volumes when commodity prices are low and then scaling up as demand
fundamentals shift, and shifting from asset expansion to operational excellence thereby maximizing cash
flow from existing assets.

3.3. Challenge 3: social license to operate

Public focus on mining activities and the manner in which community and environmental concerns are
addressed continues to increase. This increases the importance of including social licence to operate
considerations in the development and assessment of mining project proposals. A successful project should
include community and environmental considerations in an enterprise risk management framework with
clear and proactive risk mitigation strategies. A balanced assessment of a project, which identifies both the
potential impacts of a project development and the project benefits, combined with early and continuing
engagement with stakeholders at all levels, will have a significant impact on project outcomes.

3.4. Challenge 4: skills shortage

In the last 10 years there has been an increasing number of experienced cave mining technical and operators
either at or nearing retiring age. At the same time, there has been an increasing number of young people
pursuing careers other than in mining (e.g. information technology, health, business, social services). These
two trends have resulted in mining skills shortage in almost all major mining countries. This is an issue that
the industry needs to address not only in term of filling the current skills shortage but also to prepare the
new generation of mining talents equipped to adequately address the challenges discussed earlier. Some
of the strategies discussed in the literature include better training or reskilling of existing talents, retaining
modern talents, flexibility and mobility of the workforce, adopting new technology and closer links with
educational and training institutions.

4 Conclusions
Cave mining will continue being an attractive method for quite some time mainly due to the fact that the
method can be low cost and high productivity. However, a number of the future challenges, which are
significantly outside past and or present experience, dictate that significant changes to current practices are
required. It is argued in this paper that the number of incremental changes introduced to the industry in the
last 30 years may not necessarily be adequate in themselves to effectively address the future challenges.
Some of the envisaged step changes for future challenges are discussed in the paper.

Whilst some major mining companies may have sufficient resources to address a number of the challenges
in isolation, a global and collaborative approach is required in order to accelerate and more effectively
develop new technologies, practices and skills required to successfully manage the challenges thereby
sustaining the cave mining industry globally.

Caving 2014, Santiago, Chile

The author would especially like to acknowledge Andrew Logan of Newcrest, Professor Gideon Chitombo
of Bryan Research Centre, The University of Queensland, Australia and, Jock Macneish and Robert Black
of Strategic Images who have contributed to this paper.

Aguayo, A, Uribe, G & Pedrero, J 2012, Multi-lifts Production scheduling at the Chuquicamata
Underground project, Proceedings MassMin 2012, Sudbury, Paper 6822, Canadian Institute
of Mining, Metallurgy and Petroleum: Ontario.

Arancibia, E, Puga, G, Bez, F & Raa, F 2012, Impact of emerging technologies in the panel caving
mining, Proceedings MassMin 2012, Sudbury, Paper 6900, Canadian Institute of Mining,
Metallurgy and Petroleum: Ontario.

Araneda, O & Sougarret, A 2008, Lessons learned in cave mining at the El Teniente mine over the period
1997-2007, Proceedings MassMin 2008, Lulea, (Eds: H Schunnesson and E Nordlund), pp.
43-52, Lule University of Technology: Sweden.

Baeza, L, Rosenthal, J & Kvapil, R 1987, Geomechanics at El Teniente mine, Proceedings 28th US
Symposium on Rock Mechanics, Tucson, AA Balkema, Rotterdam, Boston, 1987.

Barber, J, Thomas, L & Casten, T 2000, Freeport Indonesias Deep Ore Zone mine, Proceedings
MassMin 2000, Brisbane, (Ed: G Chitombo), pp. 289-294, Australasian Institute of Mining
and Metallurgy: Melbourne.

Barraza, M & Crorkan, P 2000, Esmeralda mine exploitation project, Proceedings MassMin 2000,
Brisbane, (Ed: G Chitombo), pp. 267- 277, Australasian Institute of Mining and Metallurgy:

Bartlett, P 1992, Support in a mechanised cave at Premier mine, Proceedings Massmin 1992, Johannesburg,
(Ed: H Glea), pp. 99-109, SAIMM: South Africa.

Botha, J, Watson, S, Arkadius, T & Samosir, E 2008, Simulation applications at PT Freeport Indonesias
DOZ / ESZ block cave mine, Proceedings MassMin 2008, Lulea, (Eds: H Schunnesson and E
Nordlund), pp. 237-246, Lule University of Technology: Sweden.

Brannon, C, Casten, T, Hewitt, S & Kurniawan, C 2008, Design and development update of the Grasberg
block cave mine, Proceedings MassMin 2008, Lulea, (Eds: H Schunnesson and E Nordlund),
pp. 433-442, Lule University of Technology: Sweden.

Brannon, C, Patton, M, Toba, R & Williams, G 2012, Grasberg block cave: logistical support system
design, Proceedings MassMin 2012, Sudbury, 6777, Canadian Institute of Mining, Metallurgy
and Petroleum: Ontario.

Brown, ET 2003, Block Caving Geomechanics, 1st edition, JKMRC Monograph Series in Mining
and Mineral Processing, Julius Kruttschnitt Mineral Research Centre, The University of
Queensland: Brisbane, 515p.

Brown, ET 2007, Block Caving Geomechanics, 2nd edition, Julius Kruttschnitt Mineral Research Centre:
Brisbane, 696p.

Keynote Speakers

Brunton, I, Sharrock, G & Lett, J 2012, Full Scale Near Field Flow Behaviour at the Ridgeway Deeps
Block Cave Mine, Proceedings of MassMin 2012, Sudbury, Paper 6826, Canadian Institute
of Mining, Metallurgy and Petroleum: Ontario.

Burger, D and Cook, B 2008, Equipment automation for massive mining methods, Proceedings MassMin
2008, Lulea, (Eds: H Schunnesson and E Nordlund), pp. 493-498, Lule University of
Technology: Sweden.

Butcher, R 2000a, Block cave undercutting - aims, strategies, methods and management, Proceedings
MassMin 2000, Brisbane, (Ed: G Chitombo), pp. 405-411, Australasian Institute of Mining
and Metallurgy: Melbourne.

Butcher, R 2000b, The role of mass concrete in soft rock block cave mines, Proceedings MassMin 2000,
Brisbane, (Ed: G Chitombo), pp. 422-428, Australasian Institute of Mining and Metallurgy:

Cadia Valley Operations, 2013. Across the valley magazine. Newcrest Mining Limited, Internal magazine,
Edition February 2013.

Cadia Valley Operations, 2013, Across the valley magazine, Newcrest Mining Limited, Internal magazine,
Edition April 2013.

Cadia Valley Operations, 2013, Across the valley magazine, Newcrest Mining Limited Internal magazine,
Edition September 2013.

Calder, K, Townsend, P & Russell, F 2000, PC-BC: a block cave design and draw control system,Proceedings
MassMin 2000, Brisbane, (Ed: G Chitombo), pp. 469-484, Australasian Institute of Mining
and Metallurgy: Melbourne.

Casten, T, Golden, R, Mulyadi, A & Barber, J 2000, Excavation design and ground support of the gyratory
crusher installation at the DOZ mine, PT Freeport Indonesia, Proceedings MassMin 2000,
Brisbane, (Ed: G Chitombo), pp. 295-299, Australasian Institute of Mining and Metallurgy:

Casten, T, Rachmad, L, Arkadius, T, Osborne, K & Johnson, M 2008, P.T. Freeport Indonesias Deep Ore
Zone mine - expanding to 80,000 tonnes per day, Proceedings MassMin 2008, Lulea, (Eds: H
Schunnesson and E Nordlund), pp. 265-274, Lule University of Technology: Sweden.

Castro, R, Vargas, R & Huerta F 2012, Determination of drawpoint spacing in panel caving: a case study at
the El Teniente Mine, The Journal of the Southern African Institute of Mining and Metallurgy,
vol. 112.

Catalan, A, Sinaga, F & Qudraturrahman, I 2010, The role of geotechnical engineering during the
prefeasibility studies and early works of Cadia East panel caving project, New South Wales,
Australia, Proceedings of Caving 2010 Conference, Perth, Australia.

Catalan, A, Dunstan, G, Morgan, M, Green, S, Jorquera, S, Thornhill, T, Onederra, I & Chitombo, G 2012,
How can an intensive preconditioning concept be implemented at mass mining method?
Application to Cadia East panel caving project, Proceedings 46th Congress, US Rock
Mechanics/Symposium, Paper No ARMA 12-681, July 24-27, Chicago, IL, USA.

Chacn, J, Gepfert, H, Ovalle, A 2004, Thirty years evolution of block caving in Chile, Proceedings
MassMin 2004, Santiago, (Eds: A Karzulovic and M Alfaro), pp. 387-392, Chilean Engineering
Institute: Santiago.

Caving 2014, Santiago, Chile

Chacn, E, Barrera, V, Jeffrey, R & van As, A 2004, Hydraulic fracturing used to precondition ore and
reduce fragment size for block caving, Proceedings MassMin 2004, Santiago, (Eds: A
Karzulovic and M Alfaro), pp. 529-534, Chilean Engineering Institute: Santiago.

Chitombo, GP 2010, Cave mining - 16 years after Laubschers 1994 paper Cave mining state of the art,
Caving Conference 2010.

Cigla, M, Yagiz, S, Ozdemer, L 2001, Application of tunnel boring machines in underground mine
development, 17th International Mining Congress, Ankara, Turkey.

Cook, B, Burger, D, Alberts, L & Grobler, R 2008, Automated loading and hauling experiences at De Beers
Finsch mine, Proceedings Tenth Underground Operators Conference 2008, Launceston, 231-
238, Australasian Institute of Mining and Metallurgy: Melbourne.

Deloitte, 2013, Tracking the trends 2014, The top 10 issues mining companies will face in the coming
year, Accessible at
Industries/Energyandresources/Mining/Deloitte Tracking_the_trends_2014_final_Dec2013.

Dolipas, R 2000, Rock mechanics as applied in Philex block cave operations, Proceedings MassMin 2000,
Brisbane, (Ed: G Chitombo), pp. 395-404, Australasian Institute of Mining and Metallurgy:

Duffield, S 2000, Design of the second block cave at Northparkes E26 Mine, Proceedings MassMin 2000,
Brisbane, (Ed: G Chitombo), pp. 335-346, Australasian Institute of Mining and Metallurgy:

Dunstan, G & Popa, L 2012, Innovative cave establishment practices at Ridgeway Deeps, Proceedings
MassMin 2012, Sudbury, Paper 6792, Canadian Institute of Mining, Metallurgy and Petroleum:

Ernst & Young, 2013, Business risks facing mining and metals 20132014. The business risk report.
Mining and metals 20132014, Accessible at
Business_risks_facing_mining_and_metals_20132014_ ER0069/ $FILE/Business_risks_

Ferguson, W, Keskimaki, K, Mahon, J & Manuel, S 2008, Henderson 2000 conveyor update, Proceedings
MassMin 2008, Lulea, (Eds: H Schunnesson and E Nordlund), pp. 575-584, Lule University
of Technology: Sweden.

Flores, G, Karzulovic, A & Brown, ET 2004, Current practices and trends in cave mining, Proceedings
MassMin 2004, Santiago, (Eds: A Karzulovic and M Alfaro), pp. 83-90, Chilean Engineering
Institute: Santiago.

Flores, G, Logan, A & Cuthbert, B 2007, Codelco November 2006 Visit Lessons Report, Newcrest
Technical Visit report to Codelco operations, April 2007.

Flores, G & Logan, A 2008, Caving technology development and its application at Cadia East project,
Proceedings 3rd International Conference on Innovation in Mine Operations, Minin 2008
(Eds: J Arias, R Castro Tadeusz Golosinski), pp. 257-269, Santiago.

Fuenzalida, P, Baraqui, J & Castro, R 2012, Mining with low profile cruhsers: The experience at Codelco
Chile, Proceedings 5Th International Conference on Innovation in Mine Operations, Minin
2012 (Eds: R Kuyvenhove, J Morales and C Vega), pp. 90- 91, Santiago.

Keynote Speakers

Fuenzalida, P & Baraqui, J 2012, Precast structure for draw point in block/panel caving, Proceedings 5th
International Conference on Innovation in Mine Operations, Minin 2012 (Eds: R Kuyvenhove,
J Morales and C Vega), pp. 94-95, Santiago.

Golden Jr, R & Fronapfel, L 2008, Evolution of ground support practices on Hendersons lower levels,
Proceedings MassMin 2008, Lulea, (Eds: H Schunnesson and E Nordlund), 717-728, Lule
University of Technology: Sweden.

Gustafson, A, Schunnesson, H, Galar, D & Kumar, U 2013, The influence of the operating environment on
manual and automatic load- haul-dump machines: a fault tree analysis, International Journal
of Mining, Reclamation and Environment, vol. 27 Iss 2, 2013, pp. 75-87.

Haley, W 1982, Adaptation of surface mining machines to underground mining, Design and Operation
of Caving and Sublevel Stoping Mines, (Ed: D R Stewart), 1198-1219, Society of Mining
Engineers, AIME: New York.

Jofre, J, Yez, P & Ferguson, G 1992, Evolution in panel caving undercutting and drawbell excavation,
El Teniente Mine, Proceedings Massmin 1992.

Krek, R, Leonforte, A, Pratt, A & Dunstan, G 2008, Underground infrastructure requirements for
underground cave mining operations, Proceedings Tenth Underground Operators Conference
2008, Launceston, pp. 205-217, Australasian Institute of Mining and Metallurgy: Melbourne.

Kvapil, R, Baeza, L, Rosenthal, J & Flores, G 1989, Block caving at El Teniente Mine, Chile, Trans Instn
Min Metall, Sect A: Min Industry, 98: A43-56.

Laubscher, DH 1994, Cave mining the state of the art. J S Afr Inst Min Metall, 94(10): pp. 279-293.

Leach, A, Naidoo, K & Bartlett, P 2000, Considerations for design of production level drawpoint layouts
for a deep block cave, Proceedings MassMin 2000, Brisbane, (Ed: G Chitombo), pp. 356-366,
Australasian Institute of Mining and Metallurgy: Melbourne.

Leiva, E and Duran, L, 2003. Pre-caving, drilling and blasting in the Esmeralda sector of the El Teniente
mine. Fragblast 2003, Vol 7, No. 2, pp. 87-104.

LEstrange, H 2009, Cadia East Technology, Chile Visit Report Sizer Investigation, Newcrest Technical
Visit report to Codelco operations, October 2009.

Madrid, A & Constanzo, H 2013, Operational definition of three undercutting fronts for the New Mine
Level project, Proceedings of 3rd International Seminar on Mine Planning, Santiago (Eds: J
Beniscelli, C Bottinelli, J Crdenas, H Constanzo, H Gpfert and E Henrquez), pp. 171-179,

Manca, L & Dunstan, G 2013, Cadia East a case study in applied innovative design, Proceedings of 3rd
International Seminar on Mine Planning, Santiago (Eds: J Beniscelli, C Bottinelli, J Crdenas,
H Constanzo, H Gpfert and E Henrquez), 181-190, Chile.

Manca, L & Flores, G 2013, Modern Planning Practices for Cave Mining, Proceedings of 3rd International
Seminar on Mine Planning, Santiago (Eds: J Beniscelli, C Bottinelli, J Crdenas, H Constanzo,
H Gpfert and E Henrquez), pp. 191-204, Chile.

Metsnen, A 2004, Supplier as solution provider for the mining industry, Sandvik Mining and Construction
vision of the future in mining, Proceedings MassMin 2004, Santiago, (Eds: A Karzulovic and
M Alfaro), pp. 659-661, Chilean Engineering Institute: Santiago.

Caving 2014, Santiago, Chile

Moss, A, Russell, F & Jones, C 2004, Caving and fragmentation at Palabora: prediction to production,
Proceedings MassMin 2004, Santiago, (Eds: A Karzulovic and M Alfaro), 585-590, Chilean
Engineering Institute: Santiago.

Music, A & San Martin J 2012, Great volume draw bells blast at El Teniente, Proceedings MassMin 2012,
Sudbury, Paper 6779, Canadian Institute of Mining, Metallurgy and Petroleum: Ontario.

Ovalle, AW 1981, Analysis and considerations for mining the El Teniente ore body, Design and Operation
of Caving and Sublevel Stoping Mines, (Ed: D R Stewart), pp. 195-208, Society of Mining
Engineers, AIME: New York.

Ovalle, AW & Albornoz, HR 1981, Block caving with LHD equipment at El Teniente, Design and
Operation of Caving and Sublevel Stoping Mines, (Ed: D R Stewart), pp. 355-361, Society of
Mining Engineers, AIME: New York.

Pinochet, A, Constanzo, H & Larran, M 2012, Value generation at El Teniente mine by using the main
transport system flexibility, Proceedings MassMin 2012, Sudbury, Paper 6979, Canadian
Institute of Mining, Metallurgy and Petroleum: Ontario.

Rojas, E, Cuevas, J & Barrera, V 1992, Analysis of the wear in drawpoint at El Teniente mine, Proceedings
Massmin 1992, Johannesburg, (Ed: H Glea), pp. 303-310, SAIMM: South Africa.

Rojas, E, Molina, R, Bonani, A & Constanzo, H 2000, The pre-undercut caving method at the El Teniente
Mine, Codelco Chile, Proceedings MassMin 2000, Brisbane, (Ed: G Chitombo), pp. 261-
266, Australasian Institute of Mining and Metallurgy: Melbourne.

Silveira, AC 2004, Undercutting at E26 lift 2 Northparkes, Proceedings MassMin 2004, Santiago, (Eds: A
Karzulovic and M Alfaro), pp. 410-414, Chilean Engineering Institute: Santiago.

Silveira, C, Lovitt, M & Hewitt, T 2005, Off to a Good Start with Lift #2: Drawbell Extraction
Northparkes, Proceedings Ninth Underground Operators Conference 2005, Perth, pp. 75-80,
Australasian Institute of Mining and Metallurgy: Melbourne.

Schunnesson, H, Gustafson, A & Kumar, U 2009, Performance of automated LHD machines: A Review,
Proceedings of International Symposium on Mine Planning and Equipment Selection, Banff,
Canada, pp. 773-782.

Keynote Speakers

Its Not Mine Safety But Mind Safety - A Henderson


GK Carlson Climax Molybdenum Company, USA

As all industries have come to realize the implications and value of a safe workplace, and while it continues
to be a topic that dominates conversation, we continue to have injuries. Henderson is no exception. Safe
Production is our number one priority and Henderson has improved its safety performance; but we still
have not achieved the ultimate goal of no accidents and incidents. With all the rules, programs, tools,
experience, and studies at our disposal today, the question of why they have not been eliminated comes
down to one thing: not using the greatest tool at our disposal, OUR BRAINS!

Henderson has been working for years to get all employees, top to bottom, to understand that Safe
Production is truly the number one priority every minute of every day and tries to create an atmosphere
that encourages people to make good decisions in all they do at work. To succeed at this, all employees
must feel comfortable with the concept that safety has priority over production. It can also be summed up
by a single statement from our General Manager I want you all to be intensely selfish when it comes to
making safe choices. After all, you and your family depend on you to exercise your intelligence to that end.
You are the only one that has complete control. We all know the weakest link in a chain will be the failure
point. As in all workforces, from the top down, anyone not intensely selfish about safety becomes the weak
link, and the chain will break. This paper will present some of the structure and belief systems Henderson
is employing toward the ultimate goal of no accidents and incidents. To achieve this everyone must use the
best tool available, their brains!

1 Introduction
The Henderson Mine is a post-undercut, panel caving operation located near Empire, Colorado, USA. The
mine produces high quality primary molybdenum from a complex of granitic intrusives. Mine safety is not
a separate topic from general safe work practices. Maintaining a safe workplace is required of all industries
and businesses. Pigeonholing ourselves into a specific industry rather focusing on the root causes and
mitigation of injuries does not assist in elimination of accidents and incidents. Mining does have certain
hazards that are somewhat unique, but as an industry we should be focusing on safe work practices in
general. An open hole is an open hole, whether it is an ore pass or a catwalk with a section of grating
removed. Electricity can kill if it is an energized cable in a mine or an improperly wired outlet in a house.
Confined spaces are found in mines as well as a citys sewer systems. Tripping hazards can be anywhere.
Henderson, like all other mines and mills, works diligently to mitigate and minimize all hazards that may
affect our workers. However, without the full and complete buy-in of employees into a true safety culture,
injuries and incidents will continue.

Henderson has rules, lots and lots of rules. Coupled with Hendersons requirements are the government
authorities rules and regulations concerning areas not necessarily directly involved with safety. How can
anyone possibly keep all this information at the forefront of their minds and perform daily work? So why
do we have so many rules? Obviously, these rules have been developed in order to give workers a reference
and guidance on safe work practices as all were likely created due to some historic injury or fatality. The
rules are intended to help prevent repetition of past incidents by providing training so people can learn, and
hopefully, act accordingly.

Caving 2014, Santiago, Chile

Thus, how do we get workers to follow all these rules? Many programs, incentives, studies, and systems
have been developed to accomplish this goal. The question is: do they work? The answer is: sort of. No
matter how much businesses try, individuals act according to the way they think and their perception of the
world around them. That perception and method of processing inputs has been developing in a person from
the day they were born. It is influenced by the people, experiences, and beliefs they have been interacting
with throughout their lives. Changing a lifetime of reasoning is a daunting task. However, in most cases,
some facet of a persons life learnings must be overcome in order to develop a worker who consistently
makes safe choices. It is vital that we diligently pursue the re-education and training of people to become
safe and productive employees. Although one would think just going home to our loved ones uninjured
would be incentive enough, it appears to need reinforcement. Furthermore, positive feedback or recognition
for good decisions can be a more powerful incentive to reinforce a behavioural change. Henderson has
found that accountability for ones actions is a key element to changing behaviour. If people are not held
accountable for making poor decisions, there is little external incentive for them to change their behaviours.
This must be consistent for all or the system will fail. In other words, we need to do all we can to get people
to use the best tool available to them, their brains.

Henderson continues to re-examine its safety culture, programs, systems, and behavioural evaluations.
Henderson utilizes an OHSAS 18001 certified safety system to manage all training documents, all hold Safe
Operating Procedures (SOPs), develop and classify all compliance documentation and assist Henderson
in auditing system compliance. Sub-categories of this system include; task risk assessment, consequence
thinking, and engineered or administrative controls to reduce reliance on behavioural decision making for
any job. All of these are elements that must be used as training, enforcement and reward opportunities.
Henderson believes it has the appropriate procedures and controls in-place for all our employees to complete
all tasks safely, as long as they follow the rules. More importantly, however, Henderson is attempting
to redirect and give positive influence to the way a person thinks, reasons and reacts. Until people believe
passionately in their own wellbeing as the number one priority, fully understand the combined knowledge
that has contributed to all the rules that are in place, and use that knowledge to think before acting, we will
continue to have accidents. We must strive for all employees to exercise their minds and use their brains,
the best tool out there.

2 Beyond the System

Hendersons OHSAS 18001 certified safety system is the basis for its safety program. This system is used to
ensure Henderson does what it says it is doing as well as track all relevant documentation. The scope of this
paper though is not, however, to discuss the OHSAS system. This paper is intended to go beyond the rules
and focus on other attributes that Henderson believes are the core of a successful safety culture.

Lets first focus is on some of the very basics. A clean, well-organized workplace: To this end, Henderson
has embraced a higher level of housekeeping compliance and workplace inspections, making it incumbent
upon frontline supervisors and employees to adhere to this principle. Personal accountability is paramount
in maintaining this high standard. Throughout the operation, there is no tolerance for saying its not my
area, so not my problem. All personnel are to report any issues observed and if they can fix it, then fix
it! This includes simple things like picking up any stray trash; making sure all equipment is kept clean,
cleaned for inspection, and ready for the next user; keeping all storage area orderly and supplies secured;
and barricading any areas that may have safety issues (and promptly reporting such issues). In general,
Henderson encourages its employees to take pride in the operation.

In order to achieve such high standards, the workforce must act as a cohesive unit. Individuals cannot be
working in multiple directions based on their own attitudes and pre-conditions. Henderson relies on its
frontline supervisors to be fully engaged with the operation and their own crew to make sure the company
goal of Safe Production is at the forefront of all we do, and is not compartmentalized. This requires constant
communications and further makes the frontline supervisors have to think and understand their roles and

Keynote Speakers

responsibilities, as well as how they interact with those downstream. Henderson further conducts mock
audits in which the Senior Supervisor of each department routinely audits their work areas, trying to think
as a Federal inspector might. This allows supervisors to find any issues and mitigate them before the issues
become enforcement citations. All supervisors carry cameras for documenting observations (both good and
bad) to share with others. Equipment pre-operational checks are examined to prove they have been properly
done and all supervisors and employees are held responsible for their work areas.

The benefits of this effort have been impressive. Employees are taking ownership and the level of issued
citations from Federal inspections, despite increased scrutiny, is falling (Figure 1). Equipment availabilities
have risen as employees take greater care of their machines and subsequently so has productivity. This did
not happen overnight. It has taken many years of perseverance and enforcement by all levels of supervision
to re-train themselves and the workforce to reach these levels.

Figure 1. Henderson Quarterly Issued Citations (Mine and Mill)

3 Accountability
It is probably well accepted in our industry that we have lots of rules. So many so that it is beyond the
human capacity to remember all of them. However, most rules are a result of some poor individual being
injured or killed before the rule became a rule. Certainly, these rules are good, but if they are not followed
or fully understood, it can lead to accidents. Accountability and impartial enforcement is mandatory if an
effective safety culture is to take root. The excuse I didnt know is not acceptable at Henderson. If a
worker does not know the procedures, they should not be doing the task. If they are unsure, they should get
confirmation and/or training before proceeding.

Henderson takes the approach that individuals must be fully engaged in their own personal safety to
appreciate the gravity and dimensions of the decisions that govern their work ethic. Concurrently, how
that individual responds to their supervisors directions and how the supervisor reflects the safety attitude
will dictate how a choice is made by each individual. Henderson has placed particular emphasis on the
supervisors roles and responsibilities to ensure that they are fully engaged and that communications to the
workforce are in line with Safe Production. This continues up the chain of command to the very top.

Caving 2014, Santiago, Chile

In this way, all are held accountable and responsible for the actions of those below them. Continued
discussions between supervisors and subordinates have slowly begun to move the pendulum towards the
safety culture Henderson wants. It is now considered alright for an employee to refuse to do a job they feel
is unsafe, but it is not acceptable to just say something is unsafe. The employee must identify the issues
that they believe make it unsafe, whether it is insufficient training, specific controls are not in place, the
equipment to be used is unsatisfactory for performing the task, or whatever the issue is. Supervisors work
with employees to the degree necessary to ensure that the task has been properly vetted and risk has been
minimized to an acceptable level. If this cannot be agreed upon, the task will be reviewed by an appropriate
team until all agree the risk has reached acceptable levels. This rarely occurs, however, since the OHSAS
18001 system dictates that crews have reviewed their tasks, assessed risks, apply proper controls, and have
a safe procedure in place before any work begins. These are the SOPs that have been developed and supply
the rules to be followed in order to complete a task safely. Again, this is using brain power upfront as an
effort to stem the possibility of unintended consequences in the future.

Accountability does not always have to be negative. When we observe people making good choices (like
stepping back and thinking when things are not proceeding correctly) and publicly acknowledging it as a
desired behaviour, this can potentially be more powerful and lasting than punishment for an undesirable

4 The Frontline Offense

Safe Production is not a defensive strategy and should be proactive rather than reactive. It is all offense.
We must make the first moves, be relentless in our efforts, and continually communicate the expectations
and goals to be achieved.

Henderson is very successful in safely bringing large projects to completion, yet has tended to have less-
favorable experiences on normal day-to-day tasks. This may be due to the recognition that large projects
have large potential for accidents. We focus our minds and efforts during these large projects in a higher
fashion to ensure that the project is done safely. The human brain has a tendency toward running in autopilot.
Thus, an individuals attention tends not to be focused to the degree necessary to produce the same results
as large projects. Obviously, the people who do the work are those with the best knowledge of how to do it
safely and efficiently, but for some reason may make poor choices.

Hendersons approach is to place a high degree of responsibility on the frontline supervisors to have a more
thorough knowledge of their personnel to properly engage them, know what motivates or demotivates
them, what life issues might affect their work focus, and direct them so that their choices are in line with
the appropriate goals. It is incumbent on the frontline supervisors to have the greatest awareness of those
under them to ensure that they are the right people for the job, are focused and ready to perform the work,
and are not being distracted by other factors that could put them or others at risk. Henderson requires the
frontline supervisors to spend 80% or more of their time visiting their work areas and communicating with
their personnel, keeping a pulse on anything that might detract from the work focus. To assist in this effort,
Henderson sets limits on crew sizes to allow the frontline supervisor to have the extra time to truly engage
their subordinates. This engagement fosters communications, thinking (using the mind) and examining and
analysing any pertinent issues to create the intended result of working in a safe and productive manner.
Similarly, this allows more one-on-one interaction between the supervisor and subordinate, giving each a
better understanding of expectations and distractions that might be barriers to reaching the desired results.

5 Its Not Just Us

Another aspect that Henderson has changed in recent years is how we work with and manage our
contractors on-site. Contractors are considered an addition to the workforce and as such must abide by all
the values and rules of Henderson. There are no exceptions and contractor work hours and incidents are

Keynote Speakers

compiled with the sites reported safety numbers. Not only do contractors have to submit a site safety plan
prior to commencing any work on the property, but Henderson has placed additional requirements and
accountability on project managers to ensure full compliance with their plan and Hendersons regulations.
Since contractor safety performance impacts the Henderson safety statistics, every employee is empowered
to question a contractors perceived unsafe acts to ensure we are all moving forward in a safe manner. It
is not just the project managers responsibility; everyone in the area is responsible to ensure all work is
being performed safely. Accountability, as mentioned before, is key for both Henderson and contractor
personnel, and those people who fail to meet the standards have been removed. Again, this has taken time
and diligence, and the improvement is demonstrated by the incident graph (Figure 2) showing a progressive
decline in recordable incidents.

One normally thinks of contractors as those doing heavy work, but at Henderson it applies to all contractors,
including, for example, the janitorial staff to freight haulers. All are subject to the same rules and standards.
This also requires additional diligence as many smaller companies may not have taken measures to ensure
safe work practices to meet Hendersons expectations. The educational and communication demands to get
everyone utilizing their brains and acting accordingly is enormous, but must be done if we expect to have
all workers making safe choices on a routine basis.

Figure 2 Henderson Total Recordable Incident Rate(TRIR)

6 Incentives
Over the years many ideas have been tried to incentivize the safety culture. Historic approaches usually
included extrinsic motivators such as awards, bonuses and gifts. These motivators only seemed to have
worked for some. The question we need to ask is: What better incentive is there than to not be hurt, disabled
or fatally injured while at work? How do you teach someone that the end of a finger, or a toe, or a leg, or
an arm is worth thinking about every minute of every day? Henderson has come to recognize that intrinsic
motivation is the key. We are trying to help people understand that they need to be focused on the task at
hand for their own wellbeing and their loved ones rather than to earn some trinket or to prevent disciplinary

Most would agree that when a task becomes routine, the person(s) doing the work will become complacent
in performance since they have done it many times without injury. This is one of our greatest challenges.
When the mind goes into autopilot, the consequences of actions and choices become less at the forefront of
the work, until something goes wrong.

Caving 2014, Santiago, Chile

What is the meaning of Safe Production? Is it just words or is there a true commitment? Henderson
believes that Safe Production is just as it says. Hendersons bonus structure is based on the entire operations
performance and corporate goals. Initially, this was structured as a breakdown of 35% of the total bonus
(paid quarterly) based on the operations safety performance. This placed the burden on all employees
to work safely or all could potentially lose a significant percentage of the bonus. This may seem unfair
to some, punishing the innocent, however it is also meant to act as an incentive to have employees pay
attention to each others work practices, speak up and correct inappropriate behaviours and actions. How
would you feel if you chose to ignore an action you observed, knowing it was unsafe, only to find out later
that the person was injured or killed? Getting the workforce to accept critical coaching from peers and/
or subordinate as well as overcoming the intimidation one may feel in approaching a peer or a contractor,
is challenging. This requires getting all individuals to accept coaching not as a criticism but rather as an
opportunity to think and analyse what they are doing and make sure it is the safest and best approach. This
has taken a lot of time and effort to change the way the workforce behaves and is by no means complete.
However, there is noticeable positive change taking place and Henderson is committed to pursuing this act
as we believe the new culture is taking root.

Recently, the bonus structure was modified to allow for 50% of the bonus pay be based on the operations
safety performance, with the caveat that any employee injured in an accident and found culpable in a
safety violation forfeits their safety portion of the bonus. In other words, Henderson is holding people
accountable for poor decisions. Some might say this will make employees not report injuries and, in some
relatively minor incidents, this might be true. However, a major accident with injuries will get reported.
The subsequent investigation will also reveal the decision processes leading up to the occurrence and
this information is then shared throughout the company. Additionally, contractor hours and incidents are
included in the calculations for bonus pay. Although they are not participants in the program, their safety
record can affect the bonus calculations.

Supervisors are also rated on their crews performance which can have implications towards advancement
and merit pay, and on up the chain.

7 Consequence Thinking
Now that we have rules, controls, accountability, enforcement and incentives, why do we continue to have
accidents and incidents? As the old saying goes, you can lead a horse to water but you cannot make him

Each person has a different perspective and attitude about the world around them. The way they process
information, weigh alternatives and make decisions are all different. Likewise, the way each individual
communicates desires or concerns is equally unique.

It is impractical and impossible to have a supervisor with every employee at every moment. We must
rely on the employees to understand what is expected of them and trust them to act accordingly. Enter the
demon; the human capacity to not use the best tool in our arsenal, our brains.

Consequence thinking involves using ones brain to analyse what might happen and weigh the risks if a
particular choice is made before proceeding with the action/decision. Henderson calls it the 15 second
rule what could happen in the next 15 seconds of work. This sounds simple but as a matter of practice it
goes against human nature to engage the brain that much. Since the human brain prefers to run in autopilot,
for each individual to think before each action with hundreds of actions taking place over the course of a
workday is challenging. However, this is what must be done in order to truly have Safe Production.

No one wants to be injured and consciously decides to proceed with a task knowing they will be hurt.
Unfortunately, people faced with a choice can gamble that they will not be hurt (it wont happen to me).
This may be due to several factors: it is a routine task done many time before without injury (complacency),

Keynote Speakers

they ignore or disregard some aspect of the procedure thus the job will be done more quickly (shortcuts),
it is more of a bother to get the right tool than use what is at hand (improvising/laziness), they believe they
can handle the load without assistance (bulletproof), and on and on. In each instance, there is probably
some small voice in the brain telling them not to proceed, but it is often ignored, sometimes with serious
consequences. In the injuries Henderson has suffered over the past few years, the follow up investigation
demonstrates a common element of human behaviour that has caused a poor choice to be made. How do we
get people to engage their brains and really listen to that little voice?

8 Managing Production Pressure, Frustration, Upset Conditions

In keeping Safe Production as the priority, Henderson tries to convey to employees that production pressure
is really an individual choice. Although there may be emphasis communicated to get a task done, it is
Hendersons policy that it is not to be construed as an order to shortcut any process and put oneself at risk
in order to complete the task quickly. What it is meant to be conveyed is that all efforts should be taken to
ensure the task is completed correctly, in the shortest possible timeframe. Production pressure at this point
becomes an individuals personal decision. Knowing that the task is to be done correctly, employees need to
exercise their minds, use their brains, to determine the most effective and safe manner in which to proceed
to minimize task completion time. It is also required that the supervisor be engaged with, and understand,
their people to make sure all work is being done appropriately.

Likewise, if things are not going well as tasks are being performed, it is better to take the time to stop, re-
examine the issues, get additional help if needed (more brains), develop a plan moving forward, and then
begin again. Taking the time to incorporate all these steps should, if the brain is used, result in the issue being
resolved and the task being completed safely and correctly in a practical timeframe. It is when frustration
is allowed to overshadow consequence thinking, ignoring that little voice and forcing a resolution, that
accidents are likely to occur.

9 Core Values
What Henderson wants is for safety to become a core value of each persons life. If a person fully embraces
safety as part of his or her true self, then this attitude is carried beyond the workplace to home life as well.
An employee seriously injured at home, suffering lost work time, impacts the company just as much as one
injured at work. The result is the same, lost time and productivity. A true belief in safety is not checked in
and out at the gate to work. What does it take to make good behaviours become good habits? Constant and
concise training, reminders, and continual re-enforcement of good behaviours are a key element. Training
is not just how to operate or work on a piece of equipment, it is also the safety aspects of that operation. We
must begin by taking the time to fully train our workforce in the safety aspects of their jobs. We must have
constant reminders of that training and on a day-to-day basis follow up to make sure these practices are being
utilized. This cannot be done by only a few individuals, but must be a part of the day-to-day work by all.

Henderson has been cultivating a culture where all workers can and should question another when they see
something that does not appear to be safe. Likewise, it is obligatory for the employee being questioned to
not be intimidating or dismissive, but instead to accept and examine or explain their own practices. Call it
forced re-focusing, a reminder of consequence thinking, using the brain.

Henderson also helps people focus on safety by having any group meeting begin with a safety share. This may
be a management meeting, crew tailgates, or any other meeting. People are expected to come to meetings
with something to share on safety. Management also holds quarterly supervisor meetings in order to bring
all frontline supervisors and senior supervisors together with the superintendents to discuss safety values and
issues in addition to production matters so all can share experiences and learn from each other. This helps
drive, from the top down, the core value and expectation that Safe Production is the first priority. It helps

Caving 2014, Santiago, Chile

break down the barriers to a true safety culture that can then be communicated further down to the workforce.
It also reinforces the attitude that supervisors will support good decisions with regards to safety, even if it
disrupts or hampers production. Good decisions are recognized and shared throughout the operation.

10 Conclusions
As a business, we lead people toward the results we desire and must utilize our best efforts to influence and
positively redirect a persons thoughts, attitudes and actions to consistently make safe choices. Henderson
is continually trying to engage each employee, from top to bottom, to encourage the belief that Safe
Production is not only in the companys best interest, but in each individuals best interest. We provide
guidance and set examples for others to follow. Holding people accountable for their poor decisions, as well
as holding supervisors accountable for the actions of their employees, is mandatory if an effective safety
culture is to take root. If a person does not hold safety as a core value in their life, then true safety is not
an intrinsic value and the person is gambling with their body and/or life and potentially the lives of their
fellow employees. Recognizing and communicating the safe choices that have been made is a positive and
potentially greater motivator that all must share.

Safety is a mind game. We must continually focus on each individual, day by day, to reinforce the belief that
safety is a core value they need to adopt. We cannot let up; to do so invites the inevitable accident. This is
not easy and will never be complete because the brain wants to run on autopilot. But in order to have Safe
Production it must be continually active and focused. Safety is uncompromising, repetitious and demands
lots of continuous effort, and the very best safety tool provided to everyone is their brain. We must be
relentless in the pursuit that all need to use it!

Caving 2014, Santiago, Chile

Case Studies

Case Studies

Fracturing in the footwall at the Kiirunavaara mine,

M Nilsson Lule University of technology, Sweden
D Saiang SRK Consulting (Sweden) AB, Sweden
E Nordlund Lule University of technology, Sweden

The Kiirunavaara mine is a large scale sub level caving (SLC) mine located near the city of Kiruna in
northern Sweden. It is owned and operated by LKAB (Luossavaara-Kiirunavaara AB). The mine produces
approximately 28 million tonnes of iron ore annually. Over the last 30 years the mine has experienced a
slow but progressive fracturing and movement in the footwall rock mass induced by the SLC operations.
The footwall contact which assumes a slope-like geometry is partially supported by the caved material
from the hangingwall. However, since the late 1980s damage has been observed on the footwall crest
as well as within the footwall. Progressive rock mass movement in the footwall is indicated by surface
subsidence and visual observations underground. The extent of the damage has traditionally been estimated
using empirical relations. Most of the current long term underground infrastructure within the footwall
is located at a considerable distance from the ore contact. However, for new developments on deeper
levels it is imperative to predict the future extent of the damage volume. Approximating the position of
the damage boundary in the footwall at the current state of mining would assist in predicting the extent
and characteristics of the damage volume as the mine deepens. LKAB and LTU (Lulea University of
Technology) have therefore initiated a joint research project to study the long term stability of the footwall
at the Kiirunavaara mine. This paper constitutes part of the work in this research.

The paper describes a damage mapping campaign and subsequent analysis of the Kiirunavaara mine
footwall to approximate the outer boundary of the damage. The footwall was systematically mapped on
6 levels between 320 and 800 m. The mapping results were then used to interpolate damage lines on the
respective levels. The damage lines were used to construct a continuous damage surface between the studied
levels. Existing records of damage mapping, monitoring and predictions were reviewed and compared to
the results from the current campaign. The new results show that, the outer damage surface appears to
remain stationary on the upper levels while new damage was observed on the deeper levels. At levels above
740 m the damage is judged to be mainly controlled by movements along natural discontinuities. At levels
below 740 m the majority of the damage seems to be stress induced.

1 Introduction
The Kiirunavaara mine, located near the city of Kiruna in northern Sweden, is a large scale sublevel caving
operation producing 28 Mt (million metric tons) of iron ore per year. Originally an open pit operation the
mine later transited to underground operations in the late 1950-s. Today the main orebody is mined using
SLC and the mine is currently transitioning from a main haulage at level 1045 to the new one at level 1365
situated at a depth of roughly 1100 m (actual depth from ground surface).

The main host rock in the footwall is the Precambrian aged low quartz syenite porphyry. The porphyry is
subsequently replaced by other rock types farther into the footwall. The footwall porphyry is subdivided into
5 categories; denoted SP1-SP5 according to strength properties. The RMR for the footwall was compiled
and referenced by Sandstrm (2003) to range between 49 and 68.

Caving 2014, Santiago, Chile

In principle the mine is oriented north-south, with the footwall on the west side and the ore body dipping
around 60 towards the east (see Figure 1). Orientation and naming of objects and infrastructure are
referenced to a local 3D coordinate system with vertical z-axis originating at the pre-mining peak of the
Kiirunavaara mountain. The local y-axis is roughly oriented north to south and follows the general strike of
the orebody, the x-axis is oriented roughly west to east, z-coordinates increases with depth, y-coordinates
increases southwards and x-coordinates increases eastwards into and beyond the hangingwall.

Figure 1 Coordinate orientation of the Kiirunavaara mine

The surface and underground mining infrastructures at Kiirunavaara are located on and within the footwall.
The general layout of the underground infrastructure is aligned parallel to the strike of the orebody. Most of
the permanent infrastructures such as crushers, skip shafts and workshops are located at a relatively large
distance from the footwall-ore contact. The infrastructure close to the contact consists mainly of roads, ore
passes and footwall drifts.

Over the last 30 years the mine has experienced a slow but progressive fracturing and deformation in the
footwall rock mass. This movement of the rock mass is directly related to the sequential sub-level caving
(SLC) operations. As the orebody is removed through SLC the footwall contact becomes de-stressed and
assumes a slope-like geometry, see Figure 2. The footwall slope is partially supported by the caved rock
masses from the hangingwall (Villegas & Nordlund 2008; Stckel et al. 2013).

Damage has been observed since the late 1980s both on the footwall crest and at the underground
infrastructure. Surface cracks have been systematically mapped and tracked in varying regularity since
1992. The mapping was first focused within the extent of the open pit area but has later shifted to the
northernmost parts of the crest where cracks have been observed outside the open pit area (Lupo 1996).

Case Studies

Figure 2 Concept of the footwall slope

Even though most of the production critical infrastructure (skip shafts, crushers, etc.) is located at a
considerable distance from the ore contact a large scale movement/failure in the footwall could drastically
impede the mining operations due to damage on roads, ore passes and footwall drifts. It has thus been
a focus of LKAB since the early 1990s to accurately forecast the global stability of the footwall with
increasing mining depth. Several short studies have already been published with this aim. However,
these studies have produced inconsistent and inconclusive results. LKAB and LTU (Lulea University
of Technology) have therefore initiated a joint research project to study the long term stability of the
footwall at the Kiirunavaara mine.

This paper constitutes the initial phase of this research project and aims to determine the present extent of
the footwall fracturing (damage line) and, if possible, confirm the assumed failure modes.

2 Kiirunavaara footwall fracture studies

2.1 Previous studies

In the 1990s several studies were published on the large scale footwall stability at the Kiirunavaara mine.
These studies were aimed at identifying the footwall failure mechanisms. Dahner-Lindkvist (1992) analysed
the observed damage by using the slope stability charts developed by Hoek & Bray (1981). Singh et al.
(1993) evaluated the progressive failure of the hanging and footwall at Kiirunavaara mine and Rajpura
Dariba SLC mine in India. They interpreted tensile failure as the main mechanism driving the footwall
instability and henceforth predicted formation of tensile cracks at the ground surface.

Lupo (1996) assumed that the failure surface in the footwall was planar, either comprising of a pre-existing
structure or a combination of geological structures and failure surfaces through intact rock in combination
with a surface tension crack. With this assumption the surface deformations and the correlated underground
damage were predicted to follow the inclination of the ore/host rock contact and not to propagate significantly
westward as the mining depth increased.

Caving 2014, Santiago, Chile

Sjberg (1999) performed numerical and limit equilibrium analyses of the Kiirunavaara footwall and
identified circular shear failure as the primary failure mechanism.

The subject of footwall stability was revisited by Henry & Dahner-Lindqvist (2000) in which the subject
of predominant failure mechanisms was briefly discussed but not examined in depth. Later, Villegas &
Nordlund (2008) modelled the progressive failure in Kiirunavaara using the code PFC2D. Both the rock
mass and the caved material were modelled, but major structures were not considered and the size of the
particles constituting the caved material had a radius of 1 m. The model showed that the caved material
supports the footwall even during ore draw. The effect of the traction from the caved material on the
footwall induced only local failures on the footwall face close to the undercut level. These local failures did
not progress significantly into the wall. This contradicts some indications by Lupo (1996) that the caved
masses added to the shear forces in the footwall during ore draw.

3 Current monitoring

3.1 Surface monitoring

Quantitative deformation monitoring is performed almost exclusively on the ground surface. Both the
footwall and hangingwall are monitored by GPS along predetermined monitoring lines (Stckel et al. 2013.
The footwall monitoring lines includes a total of 84 measurement hubs of which the majority are measured
once a year. Some specific points of special interest where larger deformations are expected are measured
quarterly. InSAR technology has been used since 2009 but is primarily evaluated only for the hangingwall
ground surface (Stckel et al. 2013). In addition, aerial photographs by helicopter over-flights have been
captured yearly since 2008. So far the footwall crest suffers only minor and continuous deformations.

3.2 Underground monitoring

Deformations are more apparent in the underground infrastructure than they are on the surface. Large
scale footwall fracturing was first observed underground in the 1980s. To monitor this deformation Time
Domain Reflectometry (TDR) cables were utilized. However, due to the early installation many of the
measuring points are now inside the failure volume and can no longer be accessed. The high cost of
installing new TDR cables led to LKAB not replacing this system as measuring points were lost. Instead
a macro-seismic system from CANMET was brought in around the year 2000 to track volumes suffering
rock mass deformation. This system was subsequently replaced by a local micro-seismic system from IMS
in 2003 constituting around 10 geophones. The 2003 system was later expanded to a mine-wide system in
2008 which included around 220 geophones in late 2013 (Stckel et al. 2013).

The majority of the current IMS geophones are installed in the footwall close to the active mining areas,
while only a few are located at the higher levels. Due to the resulting low azimuthal coverage of the upper
levels the system has of yet not been evaluated for monitoring the footwall failure in any larger extent using
the current analysis methods.

The seismic system is currently the only method that is quantitatively monitoring the footwall underground.
Qualitative measurements are performed by routine damage mapping concentrated near the production
areas. The upper decommissioned areas are mapped only in relation to specific projects, meaning that
several years may pass between the mapping campaigns. Internal LKAB memos document several of these
campaigns, the two most recent were performed in 2004 and 2012 respectively but they only covered the
mine section Y22 to Y24.

Case Studies

3.3 Quantitative damage mapping

In the campaign reported in this paper damage mapping was performed between the coordinates Y15 and
Y45 (roughly 3 km along the strike of the orebody), see Figure 3. The aim was to update the underground
observations on decommissioned levels and to present them in the context of large scale footwall fracturing.

To achieve the best overview a number of evenly spaced levels were studied. Old haulage levels were found
to be both quite easily accessible on most Y-coordinates and evenly spaced in depth. The levels 320, 420,
509, 540 and 775 m were therefore initially considered for mapping. In addition, level 740 m was included
as the 2004 and 2012 campaigns (between Y22 and Y24) indicated a break in trend of the location of
damage on the overlying levels.

In order to single out specific regions to investigate previous damage mapping protocols for the respective
levels were used. A line was drawn on each level following the outlined contour of the outermost previously
mapped damage. Areas where the contour lines intersected existing drifts running semi-perpendicular to the
ore strike were marked as candidates for mapping. Areas hosting known large scale discontinuities where
movement had previously been recorded were also included.

Figure 3 Sketch of the footwall damage mapping area (adapted with courtesy of LKAB)

In some areas, such as parts of level 740 m, damage was already mapped in the outermost (farthest from the
ore contact) parts of the infrastructure. In these areas, the position of the outer damage surface cannot be
explicitly determined utilizing damage mapping alone but needs to be interpolated between mapped points.

Caving 2014, Santiago, Chile

4 Results
By interpolating damage contour lines between observed points on the respective levels a damage surface
geometry was approximated (Figure 4). The damage contour lines were then updated based on new
observations when damage was confirmed outside the lines derived from the previous campaigns (i.e.,
farther into the footwall).

Figure 4 Plan map showing contour lines for mapped damage on the respective levels

The observed fallouts associated with footwall damage were predominantly structurally controlled above
level 740 m. At level 740 m the majority of the observed damages were stress induced. At this level the
initiation and propagation of new local fractures also became more apparent.

Figure 4 outlines a failure surface that roughly follows the general dip and strike of the orebody. For further
analysis all parts of the contour lines were given the same credibility regardless of if the points were derived
from direct observation or from interpolation. On the basis of this assumption a damage surface can be
interpolated also between levels.

An averaged damage plane was generated by tracing the projected edges of the damage contour lines at
Y12 and Y49 between levels 320 m and 740 m using straight lines. The average damage plane obtained
was a flat surface dipping 56 and striking parallel to the orebody. The plane was used as a reference for
setting the colour scheme of a refined damage surface generated from the full contour lines. The refined
damage surface is hereafter referred to as simply the damage surface. Dark colour means that the damage
surface lies east from the average plane, while light colour means that it lies west from the average
plane, black line indicates the reference average plane (Figure 5). The reason behind the different
colouring was to simplify visualisation of the complex surface geometry.

Case Studies

Figure 5 Profile view of the damage surface with colour reference plane (left) and with levels 320 and 775 m

From the colour scheme in Figure 5 it is clear that the shape of the damage surface is complex. The earlier
postulated circular shear surfaces are not discernible nor can any other clear mechanism be identified from
the delineated surface.

Domains are established to prepare for future numerical models of the footwall. The volume is separated into
domains with similar damage surface geometry. This means that a 2D numerical model can be calibrated
with respect to each domain using the trace of the damage surface and an assumption of plane strain. In
Figure 6 the damage surface has been divided into 7 domains. An arbitrary vertical profile within a domain
should capture the representative behaviour of the damage surface within that domain.

Figure 6 Footwall domains with respect to the damage surface geometry, front view from the east (upper
image) and top view (lower image) along the average plane

Caving 2014, Santiago, Chile

The footwall was divided into domains based on the orientation of the major joint sets by Rdberg (1991).
In comparison with the joint domain for level 795 m presented by Rdberg (1991) a number of similarities
appear to the domains proposed in this paper, see Figure 6 and Figure 7.

Figure 7 Joint set domains on level 795 m by Rdberg (1991), top view

The results from both Rdberg (1991) and the current investigation indicate domain boundaries at roughly
Y33, Y39 and Y44. The agreement between the two studies is an indication of a change in the rock mass
properties at these sections. The northern part of the mine i.e. between Y15 and Y33 was not divided into
domains by Rdberg (1991).

5 Conclusions
Using damage observations connected by contour lines on multiple levels a 3D representation of a large
scale damage zone can be made. A study of this 3D representation allowed a number of conclusions related
to the damage behaviour of the footwall to be made:

A continuous but complex damage surface can be approximated from mapped underground damage.
The movements in the footwall causing this damage does not directly transfer to the ground surface
as surface deformation measurements by GPS indicate small and continuous deformation only.

Comparing with previous mapping it is clear that the progression of the damage surface into the
footwall does not follow the mining depth linearly. The rate of progression seems to be lower on
shallow levels than on levels closer to the excavation level (i.e. deeper levels).

At levels above 740 m the observed fallouts appeared to be predominantly structurally controlled.
Below 740 m the damage seemed to be mainly stress induced.

No single failure mode could be discerned from the damage surface. The derived geometrical
shape was complex which would indicate that two or more mechanisms are acting in combination.

The surface approximation shows a plausible position of an estimated damage surface. That is, it indicates
the boundary between mobilised damaged de-stressed rock mass and non-mobilised load bearing rock
mass. The fact that the delineation results in a relatively smooth surface suggests that this boundary
represents an explicit zone of fracturing. The zone of fracturing is indicated to be continuous and sub-
parallel to the ore strike and dip. Perpendicular to the ore strike the zone seems to be limited in width with
the activity concentrated in a narrow region. It might thus be detectable through deformation monitoring.
A monitoring/measuring system should be designed to confirm and monitor the position of the damage
boundary in-situ by.

Case Studies

The authors acknowledge the funding and data access granted for this study by LKAB. Thanks are also
due to Centre of Mining and Metallurgy (CAMM) at LTU. The authors also acknowledge the on-site
contributions during mapping by LKAB staff, especially Karola Mkitaavola, Hkan Krekula and ke
hrn. Finally the authors wish to thank Jonny Sjberg (Itasca/LTU) for valuable comments during the study

Dahner-Lindkvist, C 1992, Liggvggstabiliteten i Kiirunavaara (in Swedish). Bergmekanikdagen, pp. 37-
52), Stockholm: BeFo.

Henry, E, & Dahner-Lindkvist, C 2000, Footwall stability at the LKABs Kiruna sublevel caving operation,
Sweden, Massmin 2000, pp. 527-532, Brisbane, Queensland, Australia: The Australasian
institute of mining and metallurgy.

Hoek, E, & Bray, JW 1981, Rock slope engineering. London: Institution of Mining and Metallurgy, 358 p.

Lupo, JF 1996, Evaluation of deformations resulting from mass mining of an inclined orebody, Colorado
School of Mines: Doctoral Thesis.

Rdberg, G 1991, Strukturkarteringar i Kiirunavaaragruvans liggvgg, Niv 795m avv (in Swedish).
Technical Report: Tekniska Hgskolan i Lule.

Sandstrm, D 2003, Analysis of the virgin state of stress at the Kiirunavaara mine, Lule University of
Technology: Licentiate Thesis 2003:02.

Singh, UK, Stephansson, OJ, & Herdocia, A 1993, Simulation of progressive failure in hangingwall and
footwall for mining with sub-level caving, Transactions- Institution of Mining and Metallurgy,
vol A102; pp. A188-A194.

Sjberg, J 1999, Analysis of large scale rock slopes. Lule University of Technology: Doctoral Thesis

Stckel, BM, Mkitaavola, K & Sjberg, J 2013, Hanging-wall and footwall slope stability issues in
sublevel caving, Slope Stability 2013, pp. 1045-1060, Brisbane, Australia: Australian Centre
for Geomechanics.

Villegas, T & Nordlund, E 2008, Numerical simulation of the hangingwall subsidence using PFC2D,
Massmin 2008: Proceedings of the 5th International Conference and Exhibition on Mass
Mining, pp. 907-916, Lule: Lule University of Technology.

Caving 2014, Santiago, Chile

Draw control strategy at the New Gold New Afton

A Chaudhary New Gold, Canada
K Keskimaki New Gold, Canada
S Masse New Gold, Canada

The New Gold New Afton mine is a 4.5 Million tonne per year operating block cave mine located 8 km
outside of Kamloops, British Columbia. The current production level at New Afton is the 5070m level and
has a mineable reserves base of 48.8 Mtonnes at 0.56 g/t Au and 0.84% Cu, with a lower production level
in the planning stage. At its size, New Afton is one of the smaller producing block caving operations in the
world, with a long and narrow footprint atypical of most existing caving operations. In 2012, to further
optimize the projects economics, the decision was made to separate the New Afton cave into three distinct
areas; west, east and central. The first drawbell in the east cave was blasted in June 2013 forcing the
transition from a panel cave mentality in the west cave to a self-contained block cave. This transition was
necessary to manage the cave back profile, minimize the potential for rilling of broken cave material and to
establish an even draw profile to reduce or delay the effects of dilution entry. Key tools used at New Afton
to manage this change in philosophy and to monitor results include: extensive draw point grade sampling
and trend analysis, cave monitoring systems such as micro-seismic system and time-domain reflectometry
(TDR) cables, and weekly height of draw (HOD) sections developed from daily drawpoint production
reporting. A key challenge encountered while transitioning to a block cave draw strategy was the balancing
of grade and production tonnes to ensure consistent mill feed material, as well as maintaining draw focus
to minimize stress generation on the production level. Verification of cave performance compared to PCBC
modeling, tracking drawpoint grade changes vs. column height, and evaluating for dilution entry is still
in its infancy at New Afton. These areas continue to be studied to verify cave performance and for model
calibration. New Aftons draw strategy and adjustments so far have proven to be successful. Learnings from
the west cave are currently being applied to the east cave in order to ensure healthy cave growth, as well as
maximizing resource recovery and project value.

1 Introduction
New Golds New Afton mine is a 4.5 Million tonne per year operating block cave mine located 8 km outside
of Kamloops, British Columbia. The 2013 year end reserve base is estimated at 48.8 Mtonnes with average
grades of 0.56 g/t Au and 0.84% Cu. New Afton is a traditional rubber tired caving operation with 4 CAT
R1600 scoops operating on the extraction level. Material from the cave is dumped down ore passes to the
haulage level where 7.6 m3 scoops and 45 tonne trucks transfer it to the underground gyratory crusher.
Material is then conveyed by a 4.2 km long conveying system to the surface stockpile.

In 2012 the New Afton cave was separated into three distinct areas, the west, east, and central caves, each
having 130, 203, and 56 planned drawpoints respectively. The separation of the three caves required the
adaptation of draw strategy from a panel-caving scenario in the west cave where the oldest drawpoints are
pulled the hardest, to a block caving strategy where the cave is drawn down evenly. In order to achieve this,
a complete switch in draw strategy was required to catch up the east portion of the cave to promote cave
propagation and locate the height of draw (HOD) peak closer to the center of the footprint.

Case Studies

2 Cave development
The west cave at New Afton was constructed from west to east (Figure 1), and as drawbells were blasted
the tonnes produced from each drawpoint steadily increased over time to reach suitable production targets.
This method of cave advancement naturally produces a skewed cave profile and an inclined advancing
cave face. This can be beneficial and is preferable for larger caves as older drawpoints are depleted sooner.
Therefore, as the cave advances there is not an oversupply of drawpoints that cannot be maintained. It was
realized at New Afton that a continued uneven draw profile for smaller caving footprints can promote rilling
of material down the steep cave face, and early dilution entry by drawing overlying waste zones sooner than
other parts of the cave. Uneven draw could potentially affect expected cave performance and recovery of
the ore deposit. A proactive approach was taken to continuously monitor and manage the cave back profile
as well as having outside consultants analyze draw control strategies to ensure optimum cave growth.

Figure 1 New Afton cave footprint (as of April 1, 2014)

3 Draw control strategy

The decision to adjust draw strategy to a block cave scenario was made to balance the cave back shape and
promote even vertical draw down of columns. The decision was made after the cave had broken to surface
and subsidence was observed on surface to confirm cave breach. At this point the entire west cave footprint
was constructed and actively caving. To increase cave growth in the east, even out the steep cave face, and
minimize potential dilution entry, the decision was made to balance the cave back by focusing draw on the
east side. At typical draw rates ranging from 15 to 20 buckets/day (126 to 168 tonnes/day) per drawpoint,
several months of draw would be required to balance the cave shape.

Key rules followed during draw strategy adjustment included: 1) ensuring even grade and tonnage
distribution between shifts 2) ensuring the differential in daily draw between adjacent drawpoints does not
exceed 50%, and 3) maintaining draw emphasis on higher than normal convergence drawpoints

Even grade distribution was managed by utilizing data from the extensive draw point sampling program in
place at New Afton. Grab samples are taken from drawpoints at varying frequencies to develop a database

Caving 2014, Santiago, Chile

of assay results. These grades are tonnage weighted between samples to provide actual drawpoint grade
performance each month. Draw rates were adjusted to balance grade and provide consistent mill feed
material. This was a key request from the mill as a consistent feed material is beneficial to maintain steady
state operations and improved recovery rates.

A key focus at New Afton is ensuring neighboring drawpoints do not have large daily draw rate differentials.
High draw differentials between adjacent drawpoints can result in local problems of uneven mixing and has
the potential to promote packing and convergence in weaker ground. Therefore, maintaining a strict rule for
a maximum daily draw differential of 50% between adjacent drawpoints helps alleviate such problems and
ensures the progression of a consistent cave shape.

A group of 6 drawpoints at New Afton have experienced greater than normal convergence rates at 1 to
1.5 mm per day of movement. It was seen that vertical convergence in drawpoints can be controlled
with increasing the rate of draw by a magnitude of 50%-100% in that particular drawpoint for a week.
However, the presence of horizontal convergence and very minor vertical convergence was seen in these 6
drawpoints. Experimentation with maintaining steady draw rates, increasing draw in the affected area for
varying periods of time, and continuous draw on both day and night shifts was conducted for this scenario.
It was noted that continuous draw proved to be the most effective in controlling horizontal convergence.

4 Cave profile tracking

At New Afton, visual interpretations of the cave back are completed on a weekly basis by utilizing cave
growth data from TDR cables and seismic activity detected by micro seismic sensors installed around the
cave. Data from these instruments gives indications of where the seismogenic and fracture zones of the cave
are located. The plotting of instrumentation data along with HOD profiles allows for visual interpretation of
the cave back and cave growth rates can be determined as shown in Figure 2.

Figure 2 Cave profile section view

Case Studies

Through analysis of this data, it was found the New Afton cave follows a 4:1 growth rate. This means
the cave back grows at roughly 4x the HOD height. Locally this cave growth rate may vary, however on
a global scale this relationship gives a good indication of expected cave growth, approximate timing to
breach surface, and timing for interaction with any infrastructure located above the growing cave.

A 4:1 cave growth factor is equivalent to 25% swelling of the cave material as it is broken and is indicative
of fractured and fragmented cave material. New Afton has a rockmass rating (RMR76) of 35 -50 with some
areas significantly weaker due to clay infill. The main rock types encountered include BXF, Mozonite,
and Diorite. The fine fragmentation has been observed through the operation of the New Afton cave as
drawpoint availability has been excellent. It is very seldom a drawpoint is unavailable for production due to
oversized material requiring remediation work lasting longer than a single shift. This high availability has
made it possible for New Afton to reach higher than planned production rates as the ability to move material
from the cave has not been the limiting factor.

In addition, the Cave Management System (CMS) application in PCBC allows for the utilization of
daily draw data to develop cave HOD maps. Sections cut along each strike drive are analyzed for HOD
progression, cave back estimations, as well as calculations for potential air gap generation. These metrics
are analyzed weekly and draw is adjusted in areas of the cave that has been under or overdrawn. Weekly
draw compliance to plan is also measured and adjustments are made based on performance. Height of draw
contours and percent of total column drawn contours are valuable tools to assess overall cave shape and to
plan long-term draw strategies. Figure 3 shows the progression of the New Afton HOD profile over time.

Figure 3 HOD progression tracking

5 Reconciliation process
The reconciliation process is still fairly young at New Afton after only one year of full production. The
cave has produced as planned in its first year which can be expected in younger caving operations as there
is little opportunity to see dilution entry. Copper grades have tracked well against expectations while gold
grades have outperformed expectations in 2013. The reconciliation process going forward will become
more crucial to track cave performance as the potential for mixing and dilution entry may affect cave
performance. PCBC is used to reconcile mine assay grades to the predicted block model grades and mill feed
grades shown in Figure 4. To accomplish this reconciliation process, accurate records of tonnes produced
per drawpoint and assay data are required to have a good understanding of actual performance over time.

Caving 2014, Santiago, Chile

Figure 4 Cave grade performance over time

Overall cave grade performance can be analyzed against plan and mill performance. A close relationship
between the three measures is ideal to confirm cave performance to plan. If these measures begin to diverge
than further investigation will be required. Issues such as unplanned mixing, block model uncertainty and
grade sampling procedures are some of the issues that can lead to diverging trends.

Another key graphic developed to analyze drawpoint performance is a grade differential to plan vs. column
height graph shown in Figure 5. These graphics are developed to track drawpoint performance over time
and observe trends as drawpoints mature and climb to higher heights of draw.

Figure 5 Drawpoint grade performance vs. HOD

Case Studies

6 Lessons learned
Through the operation of the west cave at New Afton, many lessons have been learned that will be applied
to the operation of the east cave. The achievement of the minimum number of drawpoints required to reach
maximum production rates from the cave is a key learning to be transferred to the east cave. 84 drawpoints
(42 drawbells) has been determined as the critical values to produce 9,000 to 11,000 tonnes per day from
the cave. At an average expected drawbell construction rate of 3 bells per month, earlier blasted drawpoints
will have a ramp-up period of 14 months to achieve peak production rates of 15 to 20 buckets per day (or
128 to 170 tonnes per day). This draw rate is equivalent to 450-500 mm of column draw down per day. As
the cave is developed from west to east, earlier blasted drawbells will have a higher HOD than the center
of the cave and the newly blasted bells in the east. The draw rates will be reversed to catch-up the HOD in
the east once the newer blasted bells reach the ramp-up period. This strategy will continue until a balanced
cave angle is reached.

7 Conclusion
The New Afton block cave has been very successful through its first full year of production. The cave has
performed as expected based on metal production and as operational efficiencies continue to increase,
greater than nameplate daily throughputs are being achieved. The New Afton west cave was transitioned
from a panel cave to a block cave draw strategy in order to balance the cave back profile. This was a key
measure to maintain long-term cave performance and to work on ensuring cave growth across the footprint.
The operation of the west cave has provided valuable experience on how the New Afton rock mass responds
to caving and has proved to be very suitable for block caving. East cave development continues as planned
and a mill expansion to 14,000 tonnes per day is underway to better match the mine production capabilities.

Diering, T 2004, Combining long term scheduling and daily draw control for block cave mines, in
Proceedings MassMin 2004, A Karzulovic and M Alvaro eds, 2225 August, Santiago, Chile,
Instituto de Ingenieros de Chile, Santiago, pp. 486490.

Laubscher, DH 1994, Cave mining the state of the art, Journal of the South African Institute of Mining
and Metallurgy, vol. 94(10), pp. 279293.

Caving 2014, Santiago, Chile

Caving experiences in Esmeralda Sector, El Teniente

M Orellana Codelco, Chile
C Cifuentes Codelco, Chile
J Daz Codelco, Chile

Collapse processes occurred in the Esmeralda sector, particularly those located ahead of the undercutting
front during years 2009-2010, did not allowed the mining advance. A new exploitation strategy was created,
and two new sector Block 1 and Block 2 were developed as virgin caving. Mining method was defined as
conventional Panel caving with hydro-fracturing preconditioning.

To deal with the collapse and rockburst risk due to the stress redistribution during the connection stage, a
new operational strategy was designed. New rates for drawbell opening, a new extraction policy and the
undercut front advanced taking into account the geological features were established.

Currently, Block 1 is in permanent caving regime and Block 2 is in the connection process. From these
two experiences, it is possible to highlight some results related to the mining management under different
geological settings. The seismic response and the duration of the connection processes have been modified
by the different mining strategies. For instance, a distinct result is the different seismic response in both
blocks due to the differences in geological setting and stress field.

1 Introduction
The Esmeralda Mine currently extracts a total of 25,000 tpd from the exploitation of three main areas:
Block 1, Block 2 and Panel 1. Blocks 1 and 2 were started as part of a new exploitation strategy designed for
Esmeralda after the most recent collapses in 2010, and Panel 1 is being worked to recover the reserves from
the central collapse area. Blocks 1 and 2 are independent sectors designed to start new caving processes
away from the old Esmeralda cavity (Figure 1). The exploitation sequence starts with Block 1, which covers
roughly 43,000 m2, and then continues with Block 2, with an area of 41,000 m2.

Case Studies

Figure 1 Historic collapses in Esmeralda Mine. Block 1 and 2 locations

Conventional panel caving is the exploitation method used in both areas, with preconditioning of the first
100 m of rock above the Undercut Level. The method first completely develops production and undercut
levels, followed by the firing of drawbells and, finally, advancing by blasting at the undercut level (Figure 2).

Figure 2 Conventional Panel Caving sequence

Exploitation of Block 1 began in June of 2011 with the first drawbell blasting. Caving started approximately
one year later after connection was made with the upper level of Teniente 4. Exploitation of Block 2 started
in July of 2012, and its connection process is currently being completed. The two blocks, being of different
geological and structural conditions and with different stress fields, produce different seismic responses at
each stage of the connection process. These responses will be explained in this paper.

Caving 2014, Santiago, Chile

2 Geology and Geotechnical Conditions

The column height in the sector where Blocks 1 and 2 are located varies from 650 m at the west end to
1,000 m at the east end. This column height has first 160 meters of in situ column, and the rest is broken
material up to the surface.

The lithology of Block 1 is primarily competent rock mass made up principally of the El Teniente Mafico
Complex (CMET) and diorite porphyry. The main structures of this Block 1 face NE and NNW; these
structures are Fault P and Faults H and J, respectively. The quality of this rock mass is Regular-Good on
the IRMR scale.

Block 2, which consists mainly of a Brechas unit and CMET, has smaller structures. The Lamprofido dike
crosses it and it faces NE. The geotechnical quality of the CMET portion of the rock is Regular, and the
Brechas Complex portion is classified as Good.

Table 1 shows the average value of the sectors stress fields the major stress in pre-mining condition and
rock mass characterization.

Table 1 Geology and geotechnical conditions, Block 1 and Block 2 (Quiroz et al. 2010)

Parameter Block 1 Block 2

CMET-Teniente Mafic Complex CMET-Teniente Mafic Complex (55%)

Lithology (60%) Breccia complex (40%)
Diorite Porhyry (40%) Tonalite (5%)

UCS [Mpa] 130 145

Major geological
P Fault, J Fault, H Fault Lamprofid Dike
Structure frecuency 0.37 in CMET
(ff/m) 0.26-0.31 in Breccia
Geotechnical quality 3 CMET (Regular)
1-3 (Good-Regular)
(IRMR) 2 Breccia (Buena)
Column height 650-750 800-1000
S1/S2/S3 40/36/21 43/34/20
Dip/dip direction S1 353/30 202/20

3 Block cave parameters

3.1 Mine design

The two blocks are very similar in terms of design. The draw layout is 15x20 m throughout Block 1, while
the north half of Block 2 has a 15x20 m and that of the south half is 15x24 m. The reason for this difference
in design is to increase the safety factor and strength of the pillars. Both blocks have a similar footprint:
280x200 m in Block 1 and 240x200 m in Block 2.

The undercut level is 14 m above the extraction level. The exploitation method used with the blocks,
conventional panel caving, uses high 16 m undercutting and a 2 m burden. The drawbells are opened in two
stages, and the undercutting blasts are done three rings at a time.

Case Studies

The preconditioning design uses ascending hydraulic fracturing on each blocks entire footprint to mitigate
seismic risk and to promote caving propagation; the perforations are made in a 40x35 m mesh. Downward
preconditioning is also being done on an experimental basis, with 70 m holes drilled from the production
level downwards in order to diminish the seismic response during the caving process.

3.2 Mining sequence and extraction strategies

In Block Caving in primary rock, the initial exploitation phase is one of the most important in the project
since it influences the macro sequence of the exploitation of the area, caving propagation, and the time
needed to connect to upper levels. Exploitation of both blocks began at the NE end and progresses toward
the NW so that the undercut front advances perpendicular to the structures and major faults (Fault P in
Block 1 and the Lamprofido Dike in Block 2).

As mining progresses, the strategy is to open 4x5 drawbells over an area of approximately 12,000 m2 that
is open and available for extraction to begin of caving propagation.

When Block 1 exploitation began, the average monthly advance rate was 1,800 m2/month, as shown in
Figure 3. In June 2012, this rate dropped to the current average advance rate of 1,000 m2/month in order
to minimize the seismic risk generated by the increase in seismic frequency and the high-magnitude, high-
energy events that occurred at the beginning of the caving process. In Block 2, in contrast, areas were first
opened at advance rates of roughly 700 m2/month since it was expected that greater stress fields would be
produced in the area due to the higher column and the presence of lithological contacts. Since little seismic
activity was recorded in the area, this advance rate was increased to 1,200 m2/month in April 2013.

Different extraction strategies were established for the two blocks based on the caving stages: prior to
reaching the critical area to start caving, and once the critical area has been reached. The drawbells are
incorporated into production at a extraction rate of 0.1 or 0.2 tons per m2 per day. Once the critical area
to start caving has been reached, the rate increases based on the height at which extraction is taking place.
Daily extraction from both blocks increases as the amount of productive area available increases. Block 1
currently has a daily extraction rate of 16,000 tons, with an open area of some 40,000 m2 available, while
Block 2 extracts 3,000 tpd with an open area of 10,000 m2 available.

3.3 Cave initiation

There are various theories as to how the different configurations of stress magnitude and direction cause
caving (caving mechanisms). Coates (1981) suggests that there are two different mechanisms that, acting
independently or together, cause caving to start: horizontal stress traction in the center of the undercut,
and high subvertical compression stress at the corners of the undercut. Caving begins when these stresses
exceed the rocks resistance. Heslop and Laubscher (1981), on the other hand, propose that there are two
fault mechanisms present in caving propagation. The first is known as stress caving, which involves the
combination of flat dipping discontinuities that cave due to shearing in high compression stress fields. In
the second mechanism, called subsidence caving, a solid rock mass quickly caves in due to shear stress
near the vertical edges of the block. This occurs because the normal stress acting on the edges of the block
is lower than the slide resistance created along the length of these edges and is not high enough to support
the block.

These phenomena are seen indirectly in seismic activity, as seismic events involve the breakage of rock
masses due to concentration of stress. For example, when mining on these blocks began, intense seismic
activity was seen near the production level and around the undercut front as stress accumulated at the base
of the blocks when a small, unstable, dome-shaped cavity was formed.

Caving 2014, Santiago, Chile

Case Studies

Figure 3 Cumulative area, monthly area and extraction per day, Block 1 and Block 2

Then, when high-level caving propagation began, the connection process was accompanied by an increase
in seismic activity in the pillar, whose thickness decreased as the cavity grew. Stress was redistributed over
the bases of the cavity as a result of the connection, making it impossible for the stress to continue to be
transmitted through the rock mass.

3.4 Seismicity induced by the caving process

The stages of a caving process may be identified from the frequency of seismic activity, the location of the
seismic events, and the characteristics of relevant events. The following topics detail the behavior of each
of these parameters for the different stages of caving.

3.4.1 Seismic frecuency

There was relatively little seismic activity when mining began at Block 1, averaging 10 events per day from
August 2011 to February 2012, as shown in Figure 4. The activity was located chiefly near the production
level and around the undercut front as a result of abutment stress. As little seismic activity was recorded
at high altitude, an extraction test was done once an area of 4,700 m2 had been opened up. Extraction rate
was first increased to 0.4 m2/day and was later halted to avoid the risk of air blast since the slight seismic
activity recorded, even with increased extraction, indicated that there was not yet enough open area to cause

Case Studies

A seismic activation process began in April 2012, with frequency increasing to an average of 20 events
per day and peaking at 30 to 40 seismic events per day. This activation was associated with an increase in
the caved area and, therefore, to an increase in the destabilized area. During this pre-caving stage, which
occurred from April to June 2012, the seismic activity was located at the cave front and, particularly, clusters
began to appear in the western sector of Block 1, specifically between faults J and H, which manifested in
high-altitude seismic events.

The first overall peak of seismic frequency of 200 events per day was observed at the end of July 2012, with
an extraction area of 13,500 m2 and a caved area of 16,000 m2. It was at this time that the first evidence
of breakthrough with the upper level of Teniente 5 was observed, 90 m above the production level at

Seismic activity remained high after the blocks first breakthrough with Teniente 5, with some 100 events
occurring per day. The second overall peak of 350 events/day was reached in October 2012 when the
connection was made between Block 1 and the Teniente 4 level located 160 m above the Esmeralda
production level. At that time, the seismic activity was occurring chiefly between the Teniente 5 and
Teniente 4 levels. Block 1 was declared connected with a total of almost 24,000 m2 caved and an estimated
connection time of 15 months.

After the connection with Teniente 4 was made, seismic activity began to decline considerably, down to an
average of 40 events/day during December 2012.

Figure 4 Total seismic frequency, seismic frequency above UCL and seismic frequency below UCL in Block 1
and Block 2

Caving 2014, Santiago, Chile

Mining at Block 2 began in July 2012 with the same seismic frequency as at the start of Block 1, an average
of 10 events per day. However, in October of 2011, before mining started in this sector, seismic events
reached a peak of 50 events/day as a result of the connection made in Block 1. This indicated that the two
blocks are not completely independent from each other despite the fact that they are some 200 m apart.

The first sign of high altitude caving propagation was a peak of 40 seismic events per day recorded at the
end of July 2013. The first signs of connection with the Teniente 5 level were observed on this date, with
approximately 10,000 m2 caved.

Finally, the connection with the Teniente 4 level occurred in October 2013 with 14,000 m2 caved and an
estimated connection time of 15 months.

As shown in Figure 5, Blocks 1 and 2 underwent similar processes in terms of frequency of events, with
both connection processes experiencing increased seismic activity throughout the breakup of the crown
pillar. Afterwards, seismic activity decreased back to a state of equilibrium. However, seismic activity in
Block 1 was much greater than in Block 2, probably because the high level of stress due to a greater column
height and increased fracturing frequency gave rise to conditions favorable for caving.

It took approximately four months to break up the crown pillar in each of the two blocks, starting with the
time the first peak in seismic activity was recorded until the blocks seismic activity returned to equilibrium.

3.4.2 Rock bust and event magnitude greater than 1.

Regarding the rock bursts that occurred while connecting the two blocks, it is important to note that there
were two rock bursts at Block 1: the first while expanding a cavity, and the second while connecting a cavity
to the upper level of South Teniente 4. One rock burst occurred at Block 2 while connecting that block to
the upper level.

Table 2 presents a summary of the linear meters damaged by the rock bursts. The damage caused by rock
bursts ranges from minor spalling of shotcrete to projection of rock mass. Block 1 suffered damage in the
ventilation level and hauling level, while the Block 2 burst caused damage at the production level, but not
at the lower levels.

Table 2 Lineal Damage in meters by Rock Burst, Block 1 and Block 2

Block 1 15-05-12 Block 1 29-09-12 Block 2 22-10-13

Heavy Moderate Minor Heavy Moderate Minor Heavy Moderate Minor
Production 45 m 10 m 70,5 m 22 m 103 m 125 m
Ventilation 103 m
Begin caving
in Esmeralda
7m 284 m 12 m 40 m
Mine, Division
El Teniente.
Total 439 m Total 133 m Total 250 m

Case Studies

The seismic events that damaged the tunnels support and rock mass were triggered by mining blasts. Table
3 shows the date and time of the blasts and the seismic events as well as the time lapse between the blasts
and the seismic events connected with the rock bursts.

Table 3 Date and time after blasting to triggers the rock burst, Block 1 and Block 2

Block Date Magnitude Distance to blasting Time since blasting

1 15/05/2012 1,7 29 0:00
1 29/09/2012 2,1 71 3:38
2 22/10/2013 1,9 29 0:00

Figure 5 show the blasting locations and the seismic events of each rock burst.

Figure 5 Event and blast location that triggers the rock burst

Events with magnitudes greater than 1 occurred in both blocks, 35 in Block 1 and 26 in Block 2, triggered by
undercut and drawbell blasts and by caving propagation, 80% in Block and 46% in Block 2 was registered
in the first 24 hours since the blast. Figure 6 shows plan and cross-section views of the seismic events equal
to or greater than 1 in Blocks 1 and 2.

Caving 2014, Santiago, Chile

Figure 6 Event magnitude great tan 1 Plant View and cross cut

3.4.3 Breakthrough and cave back

The cave back is estimated using information of seismic activity, inspections drillholes, subsidence
inspections, and the lithology and granulometry at extraction points.

The first evidence of expansion in Block 1 caving was recorded at the end of May 2012. This expansion
was associated with an increased frequency of seismic events and was furthered by the fact that the H and
J structures are situated semi-parallel to the undercut front of Block 1.

The seismic activity recorded at Block 1 after the first evidence of growth in the cavity manifests increased
collapse, spreading in altitude, and an increase in the collapsed area at the altitude of Teniente 5. Figure 7
shows the estimated cave back at the end of May 2012.

Case Studies

Figure 7 Estimated cave back of Block 1 at May 2012.

After this expansion of the Block 1 cavity, seismic activity increased from 10 to 20 events/day. A first peak
of 200 events/day was reached, followed by a new peak of 350 events/day, evidence of the increased height
of the Block 1 cavity, which was connected in November 2012 (Figure 8).

Figure 8 Estimated Cave Back of Block 1 at November 2012

Caving 2014, Santiago, Chile

Exploitation of Block 2 began in July 2012, with the opening of the first drawbell. The block did not
generally provide evidence of high-altitude caving through seismicity until the records began to increase in
July 2013, when a variety of different seismic peaks occurred; this continued until November 2013. After
the connection to the upper level was made, seismic activity returned to equilibrium (Figure 9).

Figure 9 Estimated Cave Back of Block 1 at November 2013

It took a total of 15 months for both blocks to be connected to the upper level, and the differences between
the areas that caved in during this process are relevant. At Block 1, an area of 24,000 m2 was caved to make
the connection with the upper level, while 14,000 m2 was caved at Block 2. These differences in rate of
caving propagation have to do with the blocks stress conditions and the quality of their rock mass; Block
2 had more soft fractures and greater in situ stresses, resulting in caving propagation over a smaller area.
Figure 10 show the differences between Blocks 1 and 2 regarding the cave back heights and the tonnage
extracted, as well as the open area needed to reach cave back heights of approximately 160 m. They also
reflect the differences between mayor and minor principal stress of 19 MPa to Block 1 and 31 MPa to
Block 2.

Figure 10 Caving area and draw tones vs Cave Height, Block 1 and Block 2

Case Studies

4 Conclusions
Seismic activity, in terms of frequency and relevancy of events, reflects the different features of the initial
processes, breakthroughs, and the regimes for caving in virgin areas. There was seismic activity before the
connection was made with the Teniente 4 and Teniente 5 levels, reflected in frequency peaks of roughly
200 events/day in Block 1 and 40 events/day in Block 2. Rock bursts also occurred prior to connection
due to the concentration of stresses generated by the connection process in the area of the undercut front
and underneath the production level. Therefore, it is important that, prior to and during the initiation of the
caving connection, mining activity be decreased in terms of m2 blasted per month. An example of this is
what occurred in Block 1, where the undercutting rate had to be reduced from 1,800 m2 to 1,200 m2 per
Different area were needed for the two blocks to be connected due to the geological and structural
characteristics of each sector. In Block 1, the H and J structures were instrumental in beginning the caving
propagation process since the release of these structures produced a free face for material to collapse, thus
generating a sort of toppling failure toward the east sector. These structures were activated and behaved
as they did due to the effect of the fronts advance with respect to these failures (sub-parallel) combined
with the direction of the main stresses that caused the release. The P fault, in contrast, did not influence
the caving propagation process since the undercut front advanced perpendicular to the fault and, together
with the direction of the largest main stress, kept this structure confined, which caused it to respond with
significant seismic events.
The caving process in Block 2 was more benign than that of Block 1 in terms of frequency and relevance
of seismic events, probably because the intense stresses generated by the higher column and more frequent
fractures created favorable caving conditions.
Both caving processes took 15 months from the first drawbell blast until connection was made with the
upper level, compared to the previous experience at North Esmeralda, which took 46 months. This time
difference in connection may be explained by the fact that hydraulic fracturing was used with the blocks.
In both Block 1 and Block 2, four months of seismic activity was recorded that reflected the increasing
height of the cavity as the crown pillar was broken, with peaks of 200 to 350 events per day. Seismic
activity later returned to a frequency of less than 30 events per day.
The three rock bursts in the blocks were triggered by the caving or undercutting blasts, and two of the three
rock bursts that occurred during the blocks caving connection process were recorded during the seismic
event peaks and, therefore, while connection was being made with the upper cavity. While blasting was
halted at Block 1 during this process, this was only done during two weeks and after the rock burst that had
occurred in Block 1. During seismic peaks caused by caving propagation, the undercutting rate must be
decreased to keep from triggering major seismic events.

Quiroz, R, Vega H, Cuello D, Cifuentes C, Quezada O, Milln J & Barraza M 2010, Esmeralda Sur
definiciones de crecimiento, Informe Interno DPL-I-2010.
Coates, DF 1981, Rock Mechanics Principles. Monograph 874, pp. 5-1 to 5-37, Energy, Mines and
Resources, Canada.
Cuello, D, Cavieres P, Cifuentes C 2011, Informe Lineamientos Geomecnicos para Planificacin Minera
Bloque 1 Mdulo A, Proyecto Esmeralda Sur, Informe Interno SGM-I-006/2011.
Diaz, J, Cifuentes C, Orellana M 2013, Back anlisis conexin de bloque 1, mina esmeralda sur, Informe
Interno SGM-NI-10-2013.
Diaz, J, Cifuentes C, Orellana M 2013, TTAB, Esmeralda Bloque 1, Presentacin Interna.

Caving 2014, Santiago, Chile

Heslop, TG & Laubscher, D 1981, Draw Control in Caving Operations on Southern African Chrysotile
Asbestos Mines, Design and operation of caving and sublevel stoping mines, NewYork,
(Ed(s): D. Stewart), 755-774, Society of Mining Engineers -AIME.
Milln, J 2010, Antecedentes geolgico-geotcnico entre XC-Acceso 3 y XC-Acceso 4, Mina Esmeralda,
Milln, J & Gonzalez F 2011, Antecedentes geolgicos y geotcnicos del rea a incorporar el ao 2012
(P0), SGL-I-072-2011.

Case Studies

Undercut advance direction management at the North

3rd Panel, Rio Blanco Mine, Divisin Andina Codelco
L Quiones Codelco, Chile
C Lagos Codelco, Chile
F Ortiz Codelco, Chile
E Faras Codelco, Chile
L Toro Codelco, Chile
D Villegas Codelco, Chile

The Rio Blanco Mine, property of Codelco Chile and one of the largest copper deposits in the world began
the mining operation of the 3rd Panel in 1997 by Panel Caving with conventional undercutting and LHD
extraction. This panel has an overburden of 500 meters and a low-medium stress regime. The main topic
treated in this document is the Undercut Advance Direction Management at the North 3rd Panel, where the
Design and Ground Control of front cave are the key factors for a successful operation.

1 Introduction
The Ro Blanco Underground Mine is located 50 km North of Santiago. The extraction process began in
1970 with the 1st Panel, which was mined until 1982 using Block Caving with grizzly treatment. Between
1982 and 1997 the 2nd Panel was extracted using the same mining method. Both panels were situated in
secondary rock.

The mine operation of the 3rd Panel started in 1997 by Panel Caving with conventional undercutting and
LHD extraction. It is placed in primary rock, with an overburden of 500 meters and a low-medium stress
regime. Subsequent to a southwest caving advance (1997 to 2004), the mining process stopped for eight
years in the area, continuing at the North zone. There are plans to continue with the exploitation in this
sector until 2017.

The main topic treated in this document is the front cave orientation control at the Undercut Level and its
usage as an instrument to reduce geotechnical problems, emphasizing Design and Ground Control of front
cave as key factors.

2 Risk reduction in Panel Caving

With the acquired experience and the situations that occurred during the years of exploitation of the 3
Panel, there are facts that must be taken into account when it comes to reducing risk in the exploitation of
a Panel Caving. These are:

Caving 2014, Santiago, Chile

2.1 Front Cave orientation considering the most important geological structures.

In order to orientate the Front Cave in relationship with the geological structures, two aspects must be

1. The Front Cave must not be orientated parallel to the main structures. In the case of the III Panel,
an angle of N40E has been defined, as shown in Figure N 1.

Figure 1 Front Cave orientation in relation to main structures, Production and Undercut levels

2. Faults or discontinuities must not be left immediately after the Front Cave because this could
cause collapsing of the brow. In this case, two options could be considered; either the caving has
to be stopped at a reasonable distance before the discontinuity to avoid activation, or it has to be
continued beyond the discontinuity in order to have the problematic area blasted, thus preventing
the activation of the structure (Figure 2).

Figure 2 Structural condition in the blast N11, GH-81, March 2014

2.2 Principal stresses orientations

The main stress (s1) is orientated in an E-W direction, which is related to the present tectonic regime. Front
Caving parallel to s1 has not been done and thus its maximum augmentation has not been observed.

On the other hand, if blasting is done with long spacing between fronts, a concentration of stress in the
edges may be observed.

To avoid these concentrations, which could endanger the stability of the front, the spacing must be no more
than 5 ring holes (Figure 3).

Case Studies

Figure 3 Spacing between Ring Holes

2.3 Rate of the undercut advance

The undercut advance rate depends on the monthly area to be blasted. This rate varies between 1,000 m2
a 1,500 m2, allowing an annual undercut advance of 10,000 to 15,000 m2, always maintaining the defined
angle (Figure 4).

Figure 4 Layout of undercut from January to December (2014)

2.4 Rock mass Preconditioning of primary rock

The use of Hydraulic Fracturing and Confined Blasting (DDE) methods, allows decreasing of the primary
fragmentation size and increasing of the caving propagation speed.

Caving 2014, Santiago, Chile

2.4.1 Safety zone built behind the front cave

In 2004, there was a collapse in the southern 3rd Panel. After analyzing the causes, a security band was
defined at a certain distance from de Front Cave. The length of this band was determined by considering
the abutment stress effect, which in this case has been estimated in 50 m. The band needs to be completely
built and fortified before the beginning of the caving process and it must be maintained during the advance
of the Front Caving. This procedure must be controlled monthly as shown in Figure 5.

Figure 5 Monthly safety zones in the Production Level

2.5 Support design for Production and Undercut Level

A support system design for the Panel Caving method is defined for the most critical condition, which are
the Front Caving advance and the abutment stress effect on the drift in the Production Level.

For this purpose, a standard fortification has been defined for the caving and Production Levels, as presented
in Figure N6. This fortification is determined based on the geomechanical evaluation of the behavior of the
supporting structure during the caving process.

Although the design considerations are important for successful caving in a Front Caving advance with
controlled risk, ground control is of vital importance. For this parameter, the following aspects need to be

2.6 Brow damage at the Undercut Level

The damage produced in the brow by the detonation is evaluated after every blast with the objective
of eliminating the risk of falling blocks, which could endanger operators. This evaluation is informed
accompanied by roof control and damaged supporting structure replacement recommendations.

Case Studies

Figure 6 Support design for Production Level and Undercut Level

2.7 Detentions of the extraction in the front cave

Once an area has started its production, extraction must go on according to the production program. If
the program is not followed, pressure points might be originated, which would result in collapse in the
Production Level. To control this risk, the Production Level is continually inspected, a monthly control of
extraction and extensometer monitoring are used as an alert system (Figure N7).

Figure 7 Accumulated extraction rate, January 2014. The black line indicates the advance of the front cave

2.8 Extensometer monitoring.

Presently, the Production Level has a monitoring system based on extensometers, each one with measuring
points at a distance of 1, 3, 5 and 7 m. The extensometers are installed at production drift. A total amount
of 10 extensometers are distributed in approximately 10,000 m2. These are connected to a data logger for
data transmission (Figure N8).

Caving 2014, Santiago, Chile

Figure 8 Displacement of extensometer by Undercut advance

2.9 Seismic activity in the front cave

Since 2005, induced seismicity is monitored by an ISS 30 channel seismic system. This method has been
used to support operational decisions in front caving. The system is modified annually, as the front advances
with the installation of new sensors at a maximum height of 40 m. Results indicate that seismicity is mainly
associated to structure activation and caving propagation at the undercut, as shown in Figure N9.

Figure 9 Seismicity in structures at the Undercut Level

Case Studies

The magnitude of the seismic moment measured up to now shows values in the range of -0.2 and 0
(Figure 9).

2.10 Crater advance

The crater advance is monitored every year by aerial photography. This system allows calibration of the
subsidence model used to estimate the crater advance in the future years.

3 Conclusion
Undercut management, with reduction of the associated risk, is a complex task that requires the consideration
of diverse variables and disciplines, like geomechanics, geology and geophysics. These aspects contemplate
from design to ground control and instrumentation monitoring.

In Divisin Andina, a methodology that considers all the previous aspects related to a successful caving
operation has been implemented, assuring the viability of production and reducing associated risks.

The authors would like to thank Codelco Chile Divisin Andina for allowing the publication of this paper.

Brady, BHG & Brown, ET 2004, Rock Mechanics for Underground Mining, 3nd edition, Chapman &Hall;

Brown, ET 2003, Block Caving Geomechanics. The International Caving Study Stage I 1997-2000. JKMRC
Monograph Series in Mining and Mineral Processing 3.

Hoek, E & Brown, ET 1980, Underground Excavations in Rock. Institution of Mining and Metallurgy,

Hoek, E 2006, Rock Engineering, Course notes. Available from <


Institute Mine Seismology (IMS) JMTS 2012, Anlisis de informacin ssmica, Australia.

Institute Mine Seismology (IMS) JDI v 5.0 2012, Visualizador tridimensional de informacin ssmica,

Lagos C, Toro A, Diario-SGEOM_MS_27Nov2012, Informe geomecnico diario, Nivel 16 Hundimiento.

Merino A, Quiones L 2009, Plan de Instrumentacin Geotcnica III Panel LHD. Aos 2009-2016, Nota
Interna GRMD-SGEOT-063-09.

Ortiz, F, Gallardo, G 2012, Caracterizacin Geotcnica reas 18 y 19, Sector Norte III Panel.

Soto, C, Merino, A, Quiones, L, Ortiz F, 2009, Revisin Geomecnica Programa Quinquenal de Obras
2009 2014, Mina Subterrnea - III Panel Mina Ro Blanco, Nota Interna GRMD-SGEOT-
120 08.

Villegas, D 2013, Cartilla No. 1, Primera Semana de Febrero 2013.

Caving 2014, Santiago, Chile

New growth strategy in Esmeralda Mine

N Jamett Codelco, Chile
RQ Alegra Codelco, Chile

Esmeralda mine is the third major panel caving fully developed in primary ore at Mina El Teniente
(CODELCO). Conceptual engineering was performed during the years 1992 and 1993, mainly based on
the experience gained from the exploitation of former sectors along with geomechanical and mine design
knowledge of that time. Esmeralda is one of the main productive sectors of the Mina El Teniente, with
reserves of 205 Mt @ 0.92% Cu located on a footprint of 629,000 m2. Design for production goal of 45,000
tpd and applying mining method Advanced Panel Caving, CODELCO began its undercut process in 1996
and then drawbells blasting a year later in September 1997.

At the beginning of 1999 with 12,000 m2 of open area, Esmeralda experienced the first connection
process at higher levels, which added to the fulfillment of production targets for the year 2000 and which
established a high quality in its operation and big expectations for the future. However, since 2001 an
instability phenomena of collapse type began to manifest itself, which continued until late 2004 causing
loss of galleries and infrastructure, totaling 26,600 m2 of collapsed area behind the undercut front.

Given this situation in terms of active area and its own production capacity, it was impossible to achieve
the production goal and another, new growth strategy of undercut process was put in place in August 2008.
However, in December of that year, the instability phenomena began that continued until 2010, causing a
full stop of the process and a loss of galleries and infrastructure equalling 30,605 m2 of loss area but this
time ahead of the undercut front.

Given this critical scenario, in 2010, a new task force generated a robust proposal allowing to resume and
continue the growth of Esmeralda mine towards the production goal. As a result, a sequence of exploitation
with smaller fronts (blocks) and a change in the mine design, mining method and growth macro-sequence
commenced, that is, Conventional Panel Caving plus preconditioning (hydraulic fracturing) and an
orientation of the undercut front as a function of relevant structures and lithology, including updates of the
mine plan.

In July 2011, growth activities commence based on this new strategy, giving good results and also continuity
in the growing process, achieving the milestones and targets established in the mining plan. Currently, two
blocks are being excavated simultaneously reaching 30,000 m2 of open area.

1 Introduction
The collapse processes lasted from 2004 to 2010, particularly those generated ahead of the undercut front
between 2009 and 2010, which compromised the mine plan due to the impossibility of mining in the
affected sectors. This damage caused a full stop of the blasting process of new drawbells and stopped all
the activities related to growth and sustainability of Esmeralda mine (open area and available reserves).
In response to this critical situation, a project team composed of experts of the Division was established
with the aim to define a plan of action to return to the growth trajectory initially generated. As a result, a
new operating strategy that involved radical changes to the original project in terms of the minig method,
mine design, macro-sequence and operational practices was generated. This is the Esmeralda Sur Project,
which generally consists of a series of modules (blocks) that are operated independently by applying Panel

Case Studies

Caving Conventional plus hydraulic fracturing (FH) method to all rock mass being mined. The first block to
be incorporate corresponds to Block 1, which has an area of 43,250 m2 and reserves of 29 Mt @ 1.04% Cut.

2 Objectives
The objective of this paper is to describe the most relevant aspects of design criteria, scope, implementation
and operation of the Project Esmeralda Sur to excavate Block 1.

3 Background

3.1 Geology

Esmeralda Mine is located in a sector composed mainly of andesite rock and a series of intrusive and
geological structures with preferential trend NE and subordinately NS to NNW (Figure 1). Block 1 is
mainly composed by rocks of three lithological units: Mafic Complex El Teniente (CMET) and Diorite
Porphyry, and small bodies of igneous and hydrothermal breccias. The CMET corresponds to a volcano -
plutonic complex consisting of basalts, andesites, gabbros and joints. The Diorite Hw or Diorite Porphyry
corresponds to a felsic intrusive intermediate composition, which has been associated with the development
of magmatic breccias (igneous) and hydrothermal. Related to the location of felsic intrusive Hw, there
are different bodies of igneous and hydrothermal breccias, some of which are associated with high grade
copper mineralization. Two main structural systems are defined; the first corresponds to P Fault System and
Lamprophyre Dike with the main branch of this system corresponds to the P Fault and has a persistence
estimate of 900 m in the horizontal and fillings of low cohesion. The second system is NS-NNW system
(J Fault and Latita Dike). The failures of this system have cloaks that tend to be subvertical and filled
of anhydrite, molybdenite, carbonates, bornite and plaster. In this domain, the presence of the J Fault is
dominant, which has an estimated extent above 300 m horizontally and 200 m vertically.

3.2 Stress State

Regarding the stress state, the information is integrated from in-situ stress measurements and the use of a
three-dimensional numerical model to interpolate and extrapolate the stress tensor in those areas of interest,
where no information is taken in-situ (Table 1).

4 Strategy & Macro-sequence

After a series of analyzes and studies the option to resume growth under the concept of operating modules
(blocks ) in order to decouple from a large single front to smaller fronts (Figure 2) was determined. The
mine method established was the Conventional Undercut Design with Hydraulic Fracturing (FH) applied
to the entire rock mass to be incorporated. In addition, the orientation of growth in these blocks had to take
into account the main geological structures to avoid or reduce the parallelism previously experienced on
the single front. Finally, it was decided that all future mining will be under shadow, i.e. under Teniente 4
Sur, the old sector located above Esmeralda.

In the case of Block 1, the direction front would be South-West in order to face directly the P fault system
through a growth perpendicular to it. This strategy also sought to better address the diorite complex, together
with the J and H fault, and thus ensure the connection of caving in the most complex area toward higher
stability more favorable spread. In addition, for the first block, negative FH was made from production level
on failures and relevant lithologic contacts (P fault, diorite Hw).

Caving 2014, Santiago, Chile

Figure 1 Esmeralda Mine Geology and Location of Esmeralda Sur Block

Table 1 Tensional pre-mining field for the first blocks in Esmeralda

Block Principal Stresses Magnitude Trend Plunge

S1 40 353 30 1 30
Block 1 (Hw) S2 36 257 25 5 25
S3 21 123 40 81 20
S1 43 202 20 9 23
Block 3 (Central) S2 34 112 40 5 10
S3 20 353 8 80 20
S1 50 191 13 10 12
Block 2 (Fw) S2 31 95 16 16 17
S3 19 327 23 65 9

5 Mine Plan
The mine plan for Esmeralda Sur was defined for each block independently, integrating all information
obtained from back analysis of Teniente 4 Sur as well as geological and geomechanical aspects, mining,
mine design and planning. Each block was planned to use south and north access under both shadow
and two lines of orepass systems (crosscut) for material handling (Figure 3), with a nominal capacity of
production of 3,500 tpd, each drift at production level and 15,000 tpd for each crosscut at haulage level.
Block 1 was planned to begin production in July 2011 and, the following year, the exploitation of Block 2
was planned to commence reaching, in total 25,000 tpd, in August 2014 (Figure 4).

Case Studies

Figure 2 Macro-sequence blocks Esmeralda Sur

Figure 3 Blocks Design Esmeralda Sur

Caving 2014, Santiago, Chile

Figure 4 Five Year Plan Esmeralda Sur 2011 -2015

6 Unit Operation

6.1 Activities Sequence

The developed sequence of activities relevant to the unit operations that were part of the mining operation
and a Conventional Panel Caving preparation, considered developments of the undercut front to a variable
distance between 100 to 150 m and a security zone or transition zone of 70 m from the front, where all the
buildings and fortifications had to be made. For Block 2, the area was extended to 100 m. In parallel, while
advancing the mining preparation, drilling drawbells in the production level and radial fan at undercut
level began. This was followed by the first blast of drawbell (two stages), leaving a 3 m slab between the
roof. Subsequently, the undercut level and material removal were completed until the void was generated
necessary for the fan blasting at the undercut level towards the completion of the drawbell (approximately
10 fans).

6.2 Drilling and Blasting Design

The design used at the Teniente 4 Sur mine with some improvements to the length of the holes was used
for this new production levels. The drawbell design (Figure 5) considered two stages corresponding to
the raise and its north and south sides as the first blast phase. The lateral parts (W-E) were planned for the
second phase. The design considered 58 holes of 15 m high leaving a pillar of 3 m between the roof and
the undercut level. Total drilling length was 731 m, area was 588 m2 and total volume removed was 1.824
m3. At the undercut level, which corresponded to the radial fan drilling, 20 holes with 270 m total drilling
length, burden and spacing of 2 m and undercut height 16 m were completed. This design also had negative
holes for connecting drawbell to the undercut level and eliminating the pillar of 3 m (Figure 6).

Case Studies

Figure 5 Drilling and Blasting Design Standard Drawbell Blocks Esmeralda

Figure 6 Drilling and blasting design: Standard Fan Undercut Blocks Esmeralda

6.3 Blasting

In March 2011, drilling on the north-east side of Block 1 began, given the launch schedule blasting in July
of the same year, which was successfully achieved on the 27th of July and making it the first blasting in
Esmeralda Sur. Subsequently, blasting along the entire area of this block was achieved (undercut rates
1500 m2/month) together with the fulfillment of production goals. In June 2012, the first evidence of
caving connection was already achieved at higher levels, generating damages at the upper galleries located
at Teniente 5 level under an active area of 16,000 m2 (5x5 drawbell). Together with this process, there

Caving 2014, Santiago, Chile

was a rise in seismic events, typically on every breakthrough process, reaching peaks of 300 events / day.
Blasting activities and drawing continued normally until three months later (October 2012) with 19,000 m2
of incorporated area. The second process of connecting to the Teniente 4 level that caused the breakthrough
process on Block 1 was then declared as officially achieved.

6.4 Drawing

The drawing strategy was influenced by the evolution of the connection process. The extraction rate applied
to the incorporated drawbells in production from the beginning until critical area was reached was 0.2
ton/m2-day. After the first connection process was reached, the strategy changed in order to accelerate
the spread of the caving. Two distinct areas were defined: the first was on the periphery of the extraction
area that included a line of drawbells and the other was a central zone composed of drawbells inside the
active area, where higher extraction rate was applied. The release of the first draw point was performed in
December 2012 in which 2,700 m2 were released. In terms of production and new drawbells incorporated,
continuity and growth of both processes was achieved (Figure 7).

Figure 7 Monthly Production and New Area Block 1

7 Conclusions
Based on the experience during the various stages of implementation, drawing and continued growth of
active area and subsequent breakthrough process in Block 1, the new strategy and macro-sequence showed
great promise towards achieving production goals.

This conditions created high expectations towards, firstly, achieving the goals established in the original plan
of Esmeralda and, secondly, towards suggesting that the application of Conventional Panel Caving was a
viable alternative to be applied to future projects at Teniente or another Mine. These high expectations were
based on improvements, such as, hydraulic fracture of the rock mass, small undercut front and orientation
of growth taking into account the main geological structures.

Case Studies

Nevertheless, the operational discipline, quality and better control of the activities should always be used
as an underlying robust standard towards successfully achieving the goals and milestones defined by the
mine plan.

Barraza, M, Quiroz, R, Vega, H 2010, Esmeralda Sur Definiciones de Crecimiento, DPL-I-009-2010.

Cifuentes, C, Daz, J & Orellana, M 2013, Back Anlisis Conexin de Bloque 1, Mina Esmeralda Sur.

Cuello, D, Gallardo, M, Daz, S & Cavieres, P 2010, Ingeniera Geomecnica Proyecto Esmeralda Sur,

Rojas, E, Quiroz, R, Leiva, E, Gaete, S 2005, Diagnstico Mina Esmeralda, SGM-I-024/2005.

Caving 2014, Santiago, Chile

Caving Mechanics

Caving Mechanics

Assessment of broken ore density variations in a block

cave draw column as a function of fragment size
distributions and fines migration

L Dorador University of British Columbia, Canada

E Eberhardt University of British Columbia, Canada
D Elmo University of British Columbia, Canada
B Norman University of British Columbia, Canada
A Aguayo Codelco, Chile

The broken ore density (BOD), commonly related to the swell or bulking factor, is an important parameter
for block caving design. It is well known that the ore column density decreases (and swell factor increases)
at the drawpoint due to the development of a loosening zone generated by ore extraction. However, the
broken ore in the draw column also potentially experiences stress and density heterogeneities throughout,
depending on the block properties (e.g., shape, aspect ratio and size distribution). Other important factors
include the air gap thickness, draw rate and draw sequence. In addition, the blocks undergo abrasion and
breakage (i.e., secondary fragmentation) which increases with draw column height. This generates rounder
block shapes and smaller particles, enabling different block shape configurations and finer broken ore
size distributions. These smaller particles migrate downwards into the draw column increasing the BOD.
Important advances with respect to the calculation of minimum and maximum packing of coarse granular
soils and rockfill makes it possible to estimate the ranges of draw column densities using as input the block
size distribution. This information is used in this work to obtain the ranges of broken ore density and its
spatial and temporal distribution in an ideal draw column. From this analysis, several broken ore density
distributions are proposed for an ideal draw column and initial block arrangement based on data taken
from the literature, and conceptual models regarding fines migration and broken ore size distributions.

1 Introduction
The swell of a caved rock mass plays a significant role in the planning and design of block and panel cave
mines, especially in terms of cave propagation, subsidence extent and ore recovery (Van As & Van Hout
2008). Key parameters such as draw height and draw rate are influenced by the ratio of caved (Vcaved)
and in-situ volume (Vin situ). Hence, the bulking factor is defined as B = Vcaved / Vin situ -1 and the swell
factor is (1+B) x 100 (Brown 2007). With respect to the block/panel caving method, there is a lack of data
regarding the swell factor. Laubscher (1994) recommends swell factors of 108%, 112%, and 116% for
coarse, medium and fine fragmentation, respectively. In comparison, Gonzalez & Duplancic (2012) have
suggested values of 130-140% based on experiences at the Teniente mine, Chile, and Alcalde et al. (2008)
obtained values of 115% and 120% at the Andina mine, Chile. Regarding a maximum swell factor, Lorig &
Pierce (2000) and Hancock et al. (2012) have suggested porosity values of 0.4 and 0.5, respectively, based
on numerical studies, which correspond with swell factors of 166% and 200%.

The swell of a caved rock mass depends largely on the rock mass characteristics and properties (e.g.
strength). Important factors include the number of joints, their orientation, spacing and persistence, which
control the in situ fragmentation. Other key factors include the in situ stress conditions and whether an
air-gap is present. These, respectively, influence the primary fragmentation and fall height of the blocks

Caving 2014, Santiago, Chile

from the cave back, and subsequently, the initial block size and shape distributions. The resulting initial
configuration of blocks then undergoes changes (secondary fragmentation) as the material moves down
through the draw column, with further fragmentation and swell depending on the changing column height
(Ross & Van As 2012) and changing gravity flow mechanisms with changing column height (Castro et al.
2014). Ross & Van As (2012) introduced the fact that the broken ore undergoes a change in the bulking
factor along the ore column, as also suggested by Sharrok et al. (2012). Herein, the term broken ore density
(BOD), which is the ratio between the bulk weight and caved volume, will be used together with swell or
bulking factor.

2 BOD distribution along a draw column

The broken ore in a draw column can be expected to undergo changes that lead to a heterogeneous density
distribution. Single blocks released from the cave back can experience different block arrangements
depending on the initial block size and shape distributions, as well as air gap height. As will be discussed later
in this paper, the initial broken ore density will subsequently undergo changes in response to compression
and dilatancy.

2.1 Initial block arrangement

Single blocks released from the cave back can align to form numerous block arrangements. The air gap
height is a relevant parameter in this regard. In the case of a negligible air gap, the blocks released from the
cave back will have less chance to rotate and thus will retain their contact with adjacent blocks. This would
lead to a tighter packing and smaller initial swell factor. In contrast, the presence of a sizeable air gap would
facilitate a more disordered block arrangement, increasing the initial swell factor. Block shape and aspect
ratio are also important in the initial block arrangement; for example a cubic block shape would allow
a tighter packing compared to blocks with high aspect ratios. Research on the influence of block shape
distribution, using characterization techniques presented by Kalenchuk et al. (2006), are currently ongoing.

2.2 Broken ore compression - increasing BOD with depth

After their initial arrangement, individual blocks will start to move down through the draw column. As the
height of the draw column increases above a block in response to continued draw and caving, the broken ore
will be subjected to increasing vertical load. The resulting stresses experienced will depend on its location
within the draw column as shown in Figure 1. Experimentally, the different stress regimes present in an
ore column can be represented by a triaxial compression test incorporating shear/compression and one-
dimensional compression zones.

Using triaxial compression testing, Valenzuela et al. (2011) investigated density increases in waste rock
under a range of high confining pressures (Figure 2). This data is from waste rock with an average specific
gravity (Gs) of 2.7 and maximum block size of 0.2 m, which is 20 times less than a typical oversize of 4 m.
However, the size distribution in terms of gradation and block shape distributions are similar, both of which
are critical parameters controlling BOD. Thus, it is feasible to employ this chart to estimate increases in
BOD. Based on this chart, a density increase from 1.85 t/m3 to 2.12 t/m3 is seen for an overload of 200 m
height (effective mean stress of 2.25 MPa). Note that this chart is suitable to assess the BOD from the upper
part of a muckpile moving down through the draw column, excluding the dilatancy zone and any potential
arch effect close to the drawbells.

Caving Mechanics

Figure 1 Different stress regimes acting on broken ore in a draw column

Figure 2 Influence of confining stress on broken ore density (Modified from Valenzuela et al. 2011)

Caving 2014, Santiago, Chile

2-3 Broken ore dilatancy - decreasing BOD near the drawpoint

The broken ore close to the drawpoint tends to dilate due to ore draw (Melo et al. 2008), likely as arching
develops and ore passes into the drawbell. Without considering secondary fragmentation and fines migration
during ore draw, the broken ore has a maximum dilation, which can be associated with a minimum packing
of the broken ore. This minimum packing can be associated with minimum density tests performed for
cohesionless soils (ASTM D 4254 00), together with geotechnical characterization of gravels and rockfill
reported in the literature (De la Hoz 2007, Dorador 2013). These demonstrate that the minimum packing
depends directly on the block size distribution (BSD) and block shape (assuming constant specific gravity).

3 Secondary fragmentation and fines migration

3.1 Impact of secondary fragmentation on ranges of BOD

Secondary fragmentation is commonly associated with comminution, point load fracturing (splitting),
corner rounding and crushing of blocks due to shear and compressive stresses imposed during vertical
movement of the broken ore. This fragmentation process is important for the BOD because the average
size distribution decreases, impacting the ranges of broken ore densities. As shown in Figure 1, a broken
ore zone moving down through an ore column can undergo a combination of two modes of stress. The first
can be associated with compression within the central axis of the column, where the broken ore would
predominantly experience a significant amount of splitting. The second involves shear/compression outside
the central axis of the draw column, where the broken ore would generate more fines due to shearing and
rounding along the block edges. Hence, the secondary fragmentation of broken ore will result in a dual-
mode weighted gradation curve from splitting and fines generation derived from two modes of induced

3.2 Fines migration impacting ranges of BOD

Fines migration is another key process influencing the BOD distribution within a draw column. Fines travel
down through the column and fill the voids in between larger blocks, increasing in concentration towards
the bottom of the column. In cases where there is a significant amount of fines close to the drawpoint, large
blocks may be found floating in a fine matrix of sand and gravel. In order to study the evolution of the block
size distribution at a drawpoint in terms of adding fines into the broken ore, four fluvial material gradation
curves (G1, G2, G3 and G4) are presented in Figure 3 and Figure 4. In qualitative terms, the G-1 curve
would correlate with the block size distribution at a drawpoint when caving starts and G-2 to G-4 would
represent the transition to a finer gradation due to fines migration.

Figure 5 shows the corresponding ranges of minimum and maximum densities for each of these gradations.
This shows that the density increases from the uniform gradation (G-1) through G-2 and G-3. However,
this then slightly decreases for G-4 producing a peak density with gradation G-3, consistent with the Fuller
& Thompson (1907) curve of maximum density for aggregates. This analysis will be employed in a later
section of this paper to study the broken ore density distribution.

It should be noted that there is no standard procedure to evaluate fines migration through a draw column.
Dorador et al. (2014) have investigated using segregation and internal erosion relationships derived for earth
dams to approximate fines migration through a draw column. However, this requires further calibration
using data from active block cave operations. Finally, as shown in Figure 6, the block size distribution along
the ore column varies depending on the secondary fragmentation and fines migration, which generates a
broken ore density distribution along the draw column. As follows, the ranges of BOD in a ore column
can be discussed using the void index (e) parameter, typically used in geotechnical characterization,
equivalent to the bulk factor B.

Caving Mechanics

Figure 3 Gradation curves for start of caving (G1) Figure 4 Gradation samples
and transition due to increasing fines migration (G2 to
G4) (Modified from Dorador 2010)

Figure 5 Minimum and maximum densities for Figure 6 Example of influence of fines migration
gradation curves (Modified from Dorador 2010) and secondary fragmentation on the final BSD at
the drawpoint

4 Ranges of BOD in a draw column

Several authors have observed that the ranges of minimum and maximum densities for gravels and sands,
or in terms of void ratio emax and emin, depends primarily on the particle size and shape distributions. The
former factor has been reported by Biarez (1994) and Dorador (2013) and correlates well with the maximum
packing (or emin) using the uniformity index Cu, defined as the ratio between D60/D10 (Figure 7). The density
ranges in sands and gravels have been found to be linear when plotting emax vs emin (Cubrinovski & Ishihara
2002, De la Hoz 2007). In order to extend these correlations, data from Kezdi (1979), Gesche (2002),
De la Hoz (2007) and Dorador (2010) have been reviewed. This data consists of laboratory testing of
minimum and maximum void ratios on subangular gravels (Figure 8). The minimum packing is typically
determined by pouring the material into a cylindrical mould as suggested by ASTM standard D 4254 00.
The maximum packing is carried out in the same mould by means of a compaction process using a vibratory
table (ASTM standard D 4253 00). Hence, these two correlations (Figure 7 and Figure 8) can be used
for estimating the minimum and maximum densities of the broken ore when it undergoes an initial loose
packing (see section 5.1) comparable to the packing of rockfill and gravel materials.

Caving 2014, Santiago, Chile

Figure 7 Correlation between BSD gradation (Cu) Figure 8 Minimum and maximum packing in
and maximum packing (emin). Modified from gravel
Dorador (2013)

5 Estimation of BOD distribution into a draw column

Two scenarios have been proposed to analyse the BOD distribution in a draw column. The first refers to a
loose initial packing of broken ore material, involving blocks that were free to rotate and fall through an
air gap onto the muckpile surface. Due to the initial loose state of the material, correlations from Figure 2,
figure 7 and Figure 8 can be applied. Conversely, the second case is where an air gap is not present and the
broken ore released from the cave back remains in a dense packing. Because the blocks retain their contacts
with adjacent blocks in a tight assemblage, the comparison between broken ore and rockfill or gravels is not
applicable, and the correlations previously mentioned do not apply.

5.1 Loose packing of broken ore

Block Size Distribution (BSD) curves applicable for drawpoints derived by Dorador et al. (2014) have been
applied here to study the loose packing of the broken ore in a draw column (Figure 9). The initial BSD
corresponds to an ideal size distribution at the top of the muckpile (after primary fragmentation and rock
fall impact). This has a Cu = 27, which is higher than several Cu values of primary fragmentation reported in
the literature for operating cave mines (e.g., Salvador and Palabora mines; see International Caving Study
II). Hence, this BSD will be used in a qualitative estimation of the broken ore density distribution. The
scope of this work will be limited to the use of a block size distribution (Cu = 87) for a drawpoint under a
220 m high draw column (Figure 9).

The procedure to obtain the range of BOD in a draw column for both initial and final BSD is based on
the method explained in section 4. The average value curve in Figure 7 is used to obtain the maximum
packing, which is emin = 0.26 or BOD = 2.14 t/m3 for the initial BSD (considering a specific gravity of 2.7),
and emin = 0.22 or BOD = 2.21 t/m3 for the final BSD. To obtain the minimum packing, the correlation
provided in Figure 8 is employed. Thus, the initial and final BSD produce an emax = 0.59 (BOD = 1.70 t/
m3) and emax = 0.53 (BOD = 1.76 t/m3), respectively.

Hence, the BOD distribution for the two different initial densities is analysed in Figure 10 (alternatives A
and B). A corresponds to an initial BOD of 1.80 t/m3, which increases with column depth following the
dashed curves depicted in Figure 2, until it reaches its maximum value of 2.0 t/m3. At this point (50 m above
the draw point), dilatancy is assumed to begin. Inside the dilatancy zone, the broken ore is hypothesized to
decrease in density until it reaches its minimum packing (1.76 t/m3). Conversely, alternative B starts with a
BOD of 1.92 t/m3 increasing in a similar manner up to a maximum of 2.12 t/m3 and then decreasing with
increasing dilatancy until the minimum packing of 1.76 t/m3 is arrived at.

Caving Mechanics

Figure 9 Block size distribution at drawpoint for a Figure 10 BOD distribution along a draw column of
column height of 200 m. After Dorador et al. (2014) 200 m height, with loose packing

5.2 Dense packing of broken ore

Two cases have been considered to study the evolution of the BOD distribution for an initial dense packing
of broken ore (Figure 11), for example where a negligible air gap may be present. The first is a well-
arranged packing of blocks without any swell factor (Sf = 100%). In this case, the density does not undergo
significant changes until the dilatancy zone where the broken ore exhibits a significant decrease in density
splitting into minimum and maximum density scenarios at the drawpoints (Figure 12). The second is the
case of a small swell factor of Sf = 113%. It is assumed that a slight reduction of the BOD with depth occurs
until the start of the dilatancy zone, where a significant decrease in density then occurs.

Figure 11 Two examples of different initial

Figure 12 BOD distribution along a draw column of 200
block arrangements with dense packing
m height, with dense packing

Hence, based on these qualitative BOD distributions, it is possible to observe the importance and impact
of the initial block arrangement on broken ore density. It is believed that the rock mass characteristics
and subsequent secondary fragmentation, as well as the air gap thickness play a key role in this initial
arrangement of the blocks. Continued investigations are being carried out on these topics.

Caving 2014, Santiago, Chile

6 Conclusions
This preliminary study addresses the density distribution of the broken ore along a draw column. The Broken
Ore Density (BOD) undergoes changes with increasing depth, relative to the top of the draw column, and is
strongly controlled by the initial block arrangement. This in turn is conditioned by the orientation, spacing
and persistence of the discontinuities within the rock mass (i.e., in situ fragmentation), as well as the air
gap thickness, presence of veining, rock mass strength, and other factors that influence the primary and
secondary fragmentation. Preliminary research has shown that the discontinuity network and corresponding
block size and shape distributions are dominant factors, for example, cubic shaped blocks tend to form
a more ordered packing with respect to other block shapes. As noted, air gap is also a significant factor
because negligible air gap will result in a tighter initial arrangement of blocks; with increasing air gap, the
blocks released from the cave back will have enough space to rotate causing a more disordered packing,
generating a higher bulking factor and smaller BOD. Numerical models as well as operational mine data can
help to improve this understanding of the influence of the initial block arrangement on BOD distribution,
and is the subject of ongoing research.

The analysis presented includes the influence of secondary fragmentation and fines migration. Both of
these factors reduce the average size of the block size distribution (BSD). Using correlations and charts
from empirical relationships derived for coarse granular materials, the broken ore density was shown to
increase until a critical BSD. For finer BSD, this produced a more gradual decrease in density. However, it
is believed that the impact of secondary fragmentation and fines migration on BOD is not as significant as
the initial block arrangement.

The compression and dilatancy of the broken ore during caving and its movement through the draw column
was also studied. In the case of an initial loose packing of broken ore, the BOD increases with depth
(i.e., confining stress) down through the draw column until a point where the BOD exhibits a decrease
when it enters into the dilatancy zone. Conversely, the initial dense packing in the broken ore undergoes
a different BOD distribution. The BOD is assumed to experience minor changes along the draw column
before entering into the dilatancy zone, but then exhibits a greater reduction in density at the drawpoints.
These results reflect findings from the first phase of a detailed investigation for Codelco Chuquicamata
(PMCHS), with further research planned to improve understanding of BSD tendencies and evolution down
through the draw column.

ASTM D 4254 00. Standard Test Methods for Minimum Index Density and Unit Weight of Soils and
Calculation of Relative Density.
ASTM D 4253 00. Standard Test Methods for Maximum Index Density and Unit Weight of Soils Using
a Vibratory Table.
Alcalde, F, Bustamante, M & Aguayo, A 2008, Estimation of remaining broken material at division
Andina, In 5th International Conference and Exhibition on Mass Mining, Lule, pp. 179-189
Biarez, J & Hicher PY 1994, Elementary Mechanics of Soil Behaviour: Saturated Remoulded Soils, A.
A. Balkema, Rotterdam.
Cubrinovski, M & Ishihara K 2002, Maximum and minimum void ratio characteristics of sands, Soils and
Foundations, vol. 42, pp. 65-78.
Castro, RL, Fuenzalida, MA & Lund, F 2014, Experimental study of gravity flow under confined
conditions, Int J Rock Mech Min Sc, vol 67, pp. 164-169.
De La Hoz, K 2007, Estimacin de los parmetros de resistencia al corte en suelos granulares gruesos,
Tesis de Magister en Ciencias de la Ingeniera, University of Chile, Santiago (in Spanish).

Caving Mechanics

Dorador, L 2010, Anlisis experimental de las metodologas de curvas homotticas y corte en la evaluacin
de propiedades geotcnicas de suelos gruesos, Tesis de Magister en ciencias de la ingeniera.
University of Chile, Santiago (In Spanish).
Dorador, L & Besio, G 2013, Some considerations about geotechnical characterization on soils with
oversize. In Fifth International Young Geotechnical Engineering Conference - 5iYGEC13,
Paris, pp. 407-410.
Dorador, L, Eberhardt, E, Elmo, D, Aguayo, A, 2014, Influence of secondary fragmentation and column
height on block size distribution and fines migration reaching drawpoints, In Proceedings of
the 3rd International Symposium on Block and Sublevel Caving, Santiago.
Fuller, W & Thompson SE 1907, The laws of proportioning concrete. Trans. of the American Society of
Civil Engineers; vol. 59, pp. 67-143.
Gesche, R 2002, Metodologa de evaluacin de parmetros de resistencia al corte de suelos granulares
gruesos, Thesis of Civil Engineering, University of Chile, Santiago (In Spanish).
Gonzalez-Carbonell, P, Duplancic, P & Thin, I 2012, A generic overview of the interaction of a block cave
draw strategy and cave monitoring. In 6th International Conference and Exhibition on Mass
Mining, Sudbury.
Hancock, W, Weatherley, D & Chitombo, G 2012, Modeling the gravity flow of rock using the discrete
element method. In 6th International Conference and Exhibition on Mass Mining, Sudbury.
Kalenchuk, KS, Diederichs, MS & McKinnon, S 2006, Characterizing block geometry in jointed
rockmasses, International Journal of Rock Mechanics and Mining Sciences, vol. 43, pp.
Kezdi, A 1979, Soil physics selected topics, Elsevier Scientific Publishing Co., Amsterdam.
Laubscher, D 1994, Cave mining the state of the art. Journal of South African Inst. of Mining and
Metallurgy, vol. 94, pp. 279-293.
Lorig, L & Pierce M 2000, Methodology and guidelines for numerical modelling of undercut and extraction
level behaviour in caving mines, Itasca Consulting Group Inc, Report to International Caving
Marsal, R 1973, Mechanical properties of rockfill, in Embankment-dam engineering: Casagrande Volume,
Wiley, New York.
Melo, F, Vivanco, F, Fuentes, C & Apablaza V 2008, Kinematic model for quasi static granular displacements
in block caving: dilatancy effect on drawbody shapes, Int J. Rock Mech. Min. Sci., vol. 45,
Ross IT & Van As, A 2012, Major hazards associated with block caving, In 6th Int. Conf. and Exhibition
on Mass Mining, Sudbury.
Sainsbury, B, Pierce, ME & Mas Ivars, D 2008, Analysis of caving behaviour using a synthetic rock mass
- ubiquitous joint rock mass modelling technique, in Proceedings, SHIRMS, Perth, vol. 1, pp.
Sharrock, GB, Beck, D, Capes, GW & Brunton, I 2012, Applying coupled Newtonian Cellular Automata
- Discontinuum Finite Element models to simulate propagation of Ridgeway Deeps Block
Cave, in 6th International Conference and Exhibition on Mass Mining, Sudbury.
Valenzuela, L, Bard, E & Campaa, J 2011, Seismic considerations in the design of high waste rock
dumps, in 5th International Conference onEarthquakeand GeologicalEngineering ICEGE,
Van As, A, Van Hout, GJ 2008, Implications of widely spaced drawpoints, in 5th International Conference
and Exhibition on Mass Mining, MassMin, Lule, Sweden, pp. 147-154.

Caving 2014, Santiago, Chile

Assessing the state of the rock mass in operating block

caving mines: A review
D Cumming-Potvin, University of Western Australia, Australia
J Wesseloo, University of Western Australia, Australia

The block caving mining process limits access to the orebody, in turn limiting opportunities to gauge the
state of the rock mass once the rock mass degradation and caving process has initiated. Poor knowledge of
the evolving state of the rock mass inside the column and of the propagation of the cave can lead to some
of the key geotechnical risks in block caving, namely uncontrolled, dynamic large scale caving events,
caveback hang-ups and undesirable cave propagation outside of the orebody.

Due to this lack of access, geotechnical monitoring is often conducted to gain an understanding of the
rock mass. Open hole and TDR monitoring give point measurements of the cave back which are usually
reliable (Chen 2000). Extensometers give point measurements of displacement, however the results are
not always entirely reliable (Brown 2003). These monitoring methods, while still important for gaining
physical measurements of the cave propagation, are only point measurements. In order to monitor the entire
cave volume, microseismic monitoring is commonly used (Lett and Capes 2012).

While some attempts have been made to quantify the rock mass degradation process in the cave column or
to identify the location and profile of the cave back, there is no accepted and proven method for doing so
based on monitoring data. Therefore, there is no accepted scheme for assessing the state of the rock mass
in a block caving setting. This affects the knowledge that mines have of their cave as it propagates, but can
also affect calibration of numerical models. If there is no reliable data on the state of the rock mass, then
any calibration will subsequently be unreliable.

This paper reviews the existing literature on assessing the damage state of the rock mass in block caves,
as well as other mining environments. The strengths and shortcomings of the different analysis methods
are discussed. Based on this review it is proposed that proper systematic verification of the available
suggested methods is lacking, as well as a systematic process of integrating the different methods into a
single coherent system for the spatial and temporal evaluation of the damage state of the caving rock mass.

1 Introduction
Block caving is a mining method which has been increasingly implemented in recent times due to its cost
efficiency and high production rates. The lack of access to the orebody, which is characteristic of the block
caving method, results in a poor knowledge of the state of the rock mass inside the ore column. This leads
to some of the key geotechnical risks in block caving, namely, uncontrolled, dynamic large scale caving
events, cave back hang-ups, poor fragmentation and undesirable cave propagation outside of the orebody
(Hebblewhite 2007; Westman, Luxbacher & Schafrik 2012).

In order to mitigate these risks, geotechnical monitoring is carried out in order to gain information on the
initiation and development of caving. Brown (2003) defines four general types of monitoring methods for
measuring the propagation of a cave in block caving mines: manual methods (such as depth measurement
from open holes), Time Domain Reflectometry (TDR), microseismic methods and Cavity Monitoring
Systems (CMS).

Caving Mechanics

Whilst some analysis methods for assessing cave development have been applied to caving, no review of
the different methods used has been published. This paper will review the literature available on defining
the location of the cave profile and the damage ahead of it. We will be a focus on the use of microseismic
data and analysis techniques, as microseismic techniques provide time continuous three-dimensional data
on the cave.

A number of analysis methods have been used to quantify rock mass damage in open stoping mines
(Falmagne 2001; Coulson & Bawden 2008), in laboratory tested samples (Eberhardt et al. 1998; Chang &
Lee 2004) and in quantifying the effects of pre-conditioning (Reyes-Montes, Young & Van As 2012). These
methods could be applied to caving mines for the purpose of identifying the damage state of the rock mass
and cave profile, however the scope of this paper will be limited to analysis conducted on block caving

2 Microseismic analysis methods

Duplancic (2001) divided the caving profile into five zones; the pseudo-continuous domain, seismogenic
zone, zone of loosening, air gap and caved zone (Figure 1). This model for the caving front is widely
accepted and has been adopted by the industry as the framework within which monitoring results are
interpreted (Brown 2003).

Figure 1 Zones of ground behaviour in a block caving mine (Duplancic 2001)

A number of analysis methods have been used in caving mines to obtain information on the caving process
from microseismic data. These are performed for two main objectives, with different analysis techniques
being applied to each of them and can be summarised as follows:

Finding cave/seismogenic zone geometry

o Event location

o Apparent stress/energy index

o Passive seismic tomography

Caving 2014, Santiago, Chile

Identifying rock mass damage in the cave column

o Passive seismic tomography

o Shear wave splitting

3 Finding cave/seismogenic zone geometry

3.1 Event location

A number of authors have used the vertical location of seismicity in caving mines to estimate the location
of the seismogenic zone (Duplancic 2001; Trifu et al. 2002; Glazer 2007; Hudyma et al. 2007a; Hudyma,
Potvin & Allison 2007b; Hudyma & Potvin 2008; Dixon et al. 2010; Abolfazlzadeh 2013). The cave is often
divided into different cross-sections to determine the seismogenic zone. The criterion for determining the
limits of the seismogenic zone is usually a contour along the cross-section above or below which a certain
proportion of the events lie (as seen in Figure 2).

Some of these studies have been compared to open hole dipping results for validation. It should be noted
that this approach implicitly assumes a zone of loosening with a constant thickness over space and time. As
the zone of loosening does not have a constant thickness through space and time (Hudyma et al. 2007b), the
two methods cannot be used to validate each other.

Abolfazlzadeh (2013) created a detailed case study of tracking the seismogenic zone at Telfer mine. He
suggested using a two-dimensional grid (in plan view) that each columns event vertical location distribution
is used to define the limits, with 10% and 90% vertical locations suggested as the effective limits of the
seismogenic zone. This approach implicitly assumes that the cave back migrates only vertically. This is
generally not the case. Abolfazlzadeh acknowledged that ultimately, there is no ideal methodology to
define the seismogenic zone. The results of the method are dependent on the subjectivitely chosen to cut-
off values and direction and spacing of the cross-sections used. Although this method generally corresponds
with the framework presented by Duplancic (2001), the clear delineation of different zone remain uncertain.

Figure 2 Seismogenic zone for Northparkes lift 2; May-August 2003 (Hudyma el al. 2007b)

Caving Mechanics

3.2 Apparent stress/energy index

Apparent stress is a seismic source parameter often used as an indicator of the stress level in the rock mass
where the seismic event originated. Energy Index (EI) is conceptually similar but, in contrast to apparent
stress, is a model dependent and scale independent measure of the stress levels (Van Aswegen & Butler

Hudyma et al. (2007b) used the apparent stress of microseismic events to identify the seismogenic zone at
Northparkes E26 lift 2. The hypothesis was that high apparent stress events (in this case apparent stress >
10 kPa) were more likely to appear on the edge of the seismogenic zone, which is expected to be an area
of high stress. The high apparent stress events occurred in a tight spatial distribution in the upper bound
of the seismogenic zone. Abolfazlzadeh (2013) came to similar conclusions. Although the premise of the
approach is reasonable neither of these studies included any verification.

Chen (1998) used EI to indicate the location of a de-stressed zone immediately above the cave back and a
highly stressed zone above this de-stressed zone (i.e. a seismogenic zone). While this is broadly consistent
with the caving zones of Duplancic (2001), the Chen noted that EI could not be used to infer fracturing. He
also noted that a delineatiation of the cave profile using the event locations, could not be achieved.

3.3 Passive seismic tomography

Passive seismic tomography estimates the velocity structure of an area of interest using microseismic events.

Although it shows great potential, passive seismic tomography is not often used to define the cave back
or seismogenic zone. Its use seems to be limited to two separate studies performed at Ridgeway mine
(Pfitzner et al. 2010; Westman et al. 2012). Some confidence in the potential of the method is provided by
a theoretical study performed by Lynch and Ltter (2007).

Lynch and Ltter (2007) used synthetic data to test the passive seismic tomography technique in finding
the velocity structure and geometry of theoretical block caving mines. By using a given velocity structure
and randomly placed events and sensors, they converged to velocity results by minimising the residuals of
the travel times. They tested three different geometries; a simple homogeneous model, a single cave and a
model with two caves of different heights. They were able to find the geometric parameters of the models
within 5 - 7% of the true synthetic values. The study shows that it is technically feasible to use passive
seismic tomography to find the geometry of the cave back. The geometries tested were, however, limited to
simple parabolic shapes which do not accurately reflect the shapes of cave backs in reality. The synthetic
seismic events are also a simplification of reality. Whilst the study included the bending of ray travel paths
around the cave, it did not take into account the reflection and refraction often seen in true seismic events
(Daehnke 1997).

One of the studies that used double difference passive seismic tomography in order to identify the
seismogenic zone at Ridgeway was performed by Pfitzner et al. (2010). In this study the velocity increase
was used as an indicator of higher stress, which was interpreted as the location of the seismogenic zone
(Figure 4). This zone was not compared to seismogenic zone limits defined by the event locations, such as
those found in Hudyma et al. (2007b).

The second study at Ridgeway using double difference passive tomography was performed by Westman, et
al. (2012). They used passive seismic tomography over an 18 month period to investigate the changing rock
mass and stress conditions. The location of the seismogenic zone was also identified as part of the study.

Caving 2014, Santiago, Chile

Figure 4 Seismogenic zone inferred by velocity increase (Pfitzner et al. 2010)

Figure 5 shows examples of the results from the study. The event locations were found to mostly lie within
the higher velocity zone above the cave. The study was limited to qualitative assessments and no attempt
was made to correlate the seismogenic zone with the event locations or to quantify the boundary of the
seismogenic zone.

Figure 5 Contour plot of velocity for March 2010 with both the block and sub-level caves. The low velocity
isosurface represents 5400m/s (Westman et al. 2012)

4 Identifying rock mass damage in the cave column

4.1 Passive seismic tomography

Passive seismic tomography, despite showing strong potential, has not been extensively used in block
caving mines to determine rock mass damage. Glazer and Lurka (2007), Pfitzner et al. (2010) and Mercier
et al. (2012) have used the technique to infer both rock mass damage and stress state from seismic velocity
inferred through tomography.

Caving Mechanics

Glazer and Lurka (2007) used passive seismic tomography to quantify velocity change at Palabora block
cave mine which they related to stress change. The results indicated an area of higher velocity towards
the east of the mine which was interpreted as either being an area of higher stress, or being an area of
compacted coarser cave material. The uncertainty of whether it indicates an area of solid rock under high
stress or compacted cave material highlights the subjectivity of the interpretation involved with the passive
seismic tomography method, as described by (Glazer & Lurka 2007).

Glazer and Lurka (2007) performed a verification exercise by comparing the higher velocity areas to a
crosscut with large convergence and areas where the seismicity had a higher energy index. Whilst this
verification used independent methods for comparison, it was limited to a broad qualitative comparison.

Pfitzner et al. (2010) used the drop in velocity to infer zones of damage at Ridgeway mine. The inferred
limit of damage (Figure 6) roughly concurs with Duplancics (2001) caving model, however there was no
reconciliation of the area with independent measurement techniques. Conventional cross-hole tomographic
surveys were carried out with the aim of quantifying the change in rock mass modulus, however the survey
results were inconclusive (Morgan 2009, in Pfitzner et al. 2010). Whilst the parameter used to assess rock
mass damage (velocity drop) was quantitative, no cut-off values or guidelines for the quantification of
damage were presented. The interpretation of the meaning of the different velocity values seems to be
uncertain and the application of the method seems to be somewhat subjective without a means to validate
the results.

Figure 6 Rock mass damage inferred by velocity decrease (Pfitzner et al. 2010)

Mercier et al. (2012) used double difference passive seismic tomography to investigate the changes in
velocity at the Northparkes E48 block caving operation and the relationship of the velocity changes to
stress changes in the rock mass. They built p-wave and s-wave velocity models for different time periods.
An example of s-wave velocity tomography results for four different periods is shown in Figure 7. The
four models correspond to the progressive undercutting (January-July 2010) and self-propagating (October
2010) periods in the cave progression.

Caving 2014, Santiago, Chile

Figure 7 S-wave velocity models for Northparkes E48, 2010 (modified after Mercier et al. 2012). The
approximate position of the undercut is given by the red line. The isosurface respresents a velocity of 3000 m/s.

The resulting velocity model shows the progression of the higher velocity zone across the orebody. They
associated this high velocity zone with stress increase from the undercutting activity. For example, the
overall decrease in velocity seen between July and October 2010 was associated with completion of the
undercut along with the upwards migration of the high stress zone with the propagation of the cave column.

The only validation conducted for the study was to note that The results are internally consistent and in
accordance with accepted views of the caving sequence. The analysis conducted was largely qualitative and
based on interpretation of broad trends in the velocity model. While this may improve overall knowledge
of the cave behaviour, it does not aid in the identification and quantification of rock mass damage through
space and time.

4.2 Shear wave splitting

Shear wave splitting is a seismic analysis technique for microseismic events which travel through isotropic
media. As the shear wave enters the medium, it is split into two orthogonally polarized waves. The wave
arrival times are separated by a delay which is proportional to the degree of anisotropy and the travel path
length (Wuestefeld et al. 2011).

Wuestefeld et al. (2011) used shear wave splitting to identify fracture evolution at Northparkes E26 block
cave. The shear wave anisotropy was calculated for over 13 000 events. They found variations in anisotropy
over time, which was attributed to the generation of new fractures, however the analysis was never conducted
for different areas of the mine (only the whole cave column), and spatial changes in fracturing were not
considered. There was no validation done for the analysis, however it is difficult to independently validate
measures of fracturing without direct visual observation.

4.3 Discussion

As previously mentioned, Brown (2003) described four different monitoring methods for block caving
mines: manual methods, TDR, microseismic methods and CMS. Conducting a CMS in a block cave is usually
impractical, due to the limited access inherent to the mining method. Manual methods and TDR monitoring
give useful information on the location of the cave back, however they only give point measurements
in space which have to be taken incrementally (i.e. they are not continuous monitoring methods). These
methods only give an indication of the location of the cave back, without any information on the state
of the rock mass. Microseismic methods have the advantage of being a continuous three-dimensional

Caving Mechanics

representation of the cave development in real time, however, they do not give direct information on the
cave profile and rock mass damage, which necessitates further interpretation of the microseismic data. In
order to get a complete picture of cave development with monitoring, microseismic methods need to be
properly validated so that confidence can be placed in the results.

In block caving, monitoring results and subsequent analyses are most commonly interpreted based on
the framework provided by Duplancic (2001). Complicated analysis techniques such as passive seismic
tomography have emerged to interpret stress change from velocity in order to find the seismogenic zone.
The interpretation of these, however, is based on the framework of a model derived from event location
and so the results may not be an improvement upon simply using the location of events to determine the
seismogenic zone.

There is a discord in the interpretation of velocity change in the passive seismic tomography technique.
Some authors (Pfitzner et al. 2010) have interpreted velocity change as rock mass damage, whereas others
(Glazer & Lurka 2007; Westman, Luxbacher & Schafrik 2012; Mercier et al. 2012) have used it purely as an
indicator of stress. It is reasonable to assume the reduction in velocity to be a result of both reduction in stress
and an increase in damage. The two phenomena happen concurrently. It seems that further understanding of
the caving process is required before meaningful separation of these two effects can be achieved.

Whilst some show promising results, none of the analysis techniques discussed have shown an ability
to adequately describe the caving profile and damage zone across space and time. A system combining
these techniques could give improved performance over any individual technique. In order to quantify any
improvement, a verification method which can give an independent measure of rock mass damage and
the three-dimensional location of the cave back is needed. None of the verification methods in the studies
presented can produce this sort of independent measure.

There is a systematic lack of quality validation through all studies and so there are no obvious criteria by
which to judge which of the techniques is most successful in describing the caving profile. An independent
method of verification which can identify the location of the cave profile (including damage above the cave
back) across space and time is necessary to evaluate caving analysis techniques. This could potentially be
achieved through the use of a physical model, where direct visual observation of the scaled representation
of the cave can be tied with acoustic emission monitoring.

5 Conclusion
Table 1 summarises the pros and cons of the analysis techniques which have been applied to block caving

Following a review of the different analysis techniques which have been used to quantify cave development,
it is unclear which of these methods gives the best results. Each has strengths and weaknesses, and we
suggest that better result could be achieved by combining these techniques into a single (calibrated) analysis

Physical modelling may provide a means of furthering our understanding of the caving process and provide
a strong empirical basis for verifying and improving current techniques for assessing the rock mass state
throughout the caving process.

Caving 2014, Santiago, Chile

Table 1 Summary of pros and cons of analysis techniques

Pros Cons
Finding Cave/seismogenic zone geometry
- Potential for high quality - Interpretation of velocity limits for
Passive seismic
definition of all zones in the cave seismogenic zone are subjective
profile - Not routinely used
- Unless further temporal information is used,
Apparent stress
- Conceptually simple can only give information on current seismically
/energy index
active areas
- Unless further temporal information is used,
can only give information on current seismically
active areas
Event location - Conceptually very simple - Definition of sections and proportion of
events to use is subjective
-Grid methodology assumes cave propagates
Identifying rock mass damage in the cave column
- Geology or virgin state must be known
- Ability to get quantitative
precisely, else only relative changes can be
information on velocity which
Passive seismic observed
can be related to fracturing
tomography - Interpretation of damage from velocity is
- Can get information on
somewhat subjective
aseismic zones of rock mass
- Not routinely used
- Has not yet been used to define spatial
Shear wave - Possibility to find orientation of
changes in fracturing in block caves
splitting fracturing
- Not routinely used

Abolfazlzadeh, Y 2013, Application of Seismic Monitoring in Caving Mines - Case Study of Telfer Gold
Mine, Thesis, Laurentian University.

Brown, ET 2003, Block Caving Geomechanics, Julius Kruttschmitt Mineral Research Centre, The
University of Queensland, Indooroopilly, QLD.

Chang, SH & Lee, CI 2004, Estimation of cracking and damage mechanisms in rock under triaxial
compression by moment tensor analysis of acoustic emission, International Journal of Rock
Mechanics and Mining Sciences, vol. 41, pp. 1069-1086.

Chen, D 1998, Application of a microseismic system in monitoring E26 Block Cave at Northparkes Mines,
International Conference on Geomechanics and Ground Control in Mining and Underground

Coulson, A & Bawden, W 2008, Observation of the Spatial and Temporal Changes of Microseismic Source
Parameters and Locations, Used to Identify the State of the Rock Mass in relation to the Peak
and Post-Peak Strength Conditions., 42nd US Rock Mechanics Symposium.

Daehnke, A 1997, Stress Wave and Fracture Propagation in Rock, PhD Thesis, Technischen Universitiit

Caving Mechanics

Dixon, RA, Singh, U & McArthur, C 2010, Interaction between a propagating cave and an active pit at
Telfer Mine - Part II: monitoring interaction, 2nd International Symposium on Block and
Sublevel Caving.
Duplancic, P 2001, Characterisation of caving mechanisms through analysis of stress and seismicity, PhD
Thesis, University of Western Australia.
Eberhardt, E, Stead, D, Stimpson, B & Read, RS 1998, Identifying crack initiation and propagation
thresholds in brittle rock, Canadian Geotechnical Journal, vol. 35, pp. 222-233.
Falmagne, V 2001, Quantification of rock mass degradation using micro-seismic monitoring and applications
for mine design, PhD Thesis, Queens University.
Glazer, S 2007, Applications of mine seismology methods in block cave mining, 1st International
Symposium on Block and Sublevel Caving.
Glazer, S & Lurka, A 2007, Application of passive seismic tomography to cave mining operations based
on experience at Palabora Mining Company, South Africa, 1st International Symposium on
Block and Sublevel Caving.
Hebblewhite, BK 2007, Management of geotechnical risks in mining projects, School of Mining
Engineering, The University of New South Wales, Sydney, NSW.
Hudyma, M & Potvin, Y 2008, Characterizing caving induced seismicity at Ridgeway gold mine,
MassMin 2008.
Hudyma, M, Potvin, Y & Allison, D 2007a, Seismic monitoring of the Northparkes lift 2 block cavePart
2 production caving, 1st International Symposium on Block and Sublevel Caving.
Hudyma, M, Potvin, Y & Allison, D 2007b, Seismic monitoring of the Northparkes lift 2 block cavePart
I undercutting, 1st International Symposium on Block and Sublevel Caving.
Lynch, R & Ltter, E 2007, Estimation of cave geometry using a contrained velocity model inversion with
passive seismic data, 1st International Symposium on Block and Sublevel Caving.
Mercier, J, Mercier, J, De Beer, W & Morris, S 2012, Beyond Coloured Balls: Passive Source Tomography
of Microseismic Data for Block Caving, MassMin 2012.
Pfitzner, M, Westman, E, Morgan, M, Finn, D & Beck, D 2010, Estimation of rock mass changes induced
by hydraulic fracturing and cave mining by double difference passive tomography, 2nd
International Symposium on Block and Sublevel Caving.
Reyes-Montes, J, Young, R & Van As, A 2012, Quantification of preconditioning efficiency in cave
mining, MassMin 2012.
Trifu, C, Shumila, V & Burgio, N 2002, Characterization of the caving front at Ridgeway mine, New South
Wales, based on geomechanical data and detailed microseismic analysis, 1st International
Seminar on Deep and High Stress Mining.
Van Aswegen, G & Butler, A 1993, Applications of quantitative seismology in South African gold mines,
3rd International Symposium on Rockburst and Seismicity in Mines.
Westman, E, Luxbacher, K & Schafrik, S 2012, Passive seismic tomography for three-dimensional time-
lapse imaging of mining-induced rock mass changes, The Leading Edge, vol. 31, no. 3, pp.
Wuestefeld, A, Kendall, J, Verdon, J & Van As, A 2011, In situ monitoring of rock fracturing using shear
wave splitting analysis: an example from a mining setting, Geophysical Journal International,
vol. 187, pp. 848-860.

Caving 2014, Santiago, Chile

Influence of secondary fragmentation and column

height on block size distribution and fines migration
reaching drawpoints
L Dorador University of British Columbia, Canada
E Eberhardt University of British Columbia, Canada
D Elmo University of British Columbia, Canada
B Norman B. University of British Columbia, Canada
A Aguayo Codelco, Chile

In block and panel caving projects, the secondary fragmentation and its effect on the final block size
distribution (BSD) reaching the drawpoints are key considerations in the design and success of a caving
operation. Although there are existing empirical methods to predict these (e.g., Laubschers size distribution
chart, Esterhuizens BCF, etc.), these incorporate several rules of thumb that can be improved upon
through a more mechanistic understanding of the complex processes involved. This paper first explores
the techniques commonly used in practice to assess secondary fragmentation as well as the key influencing
mechanisms: comminution, fines migration and BSD into a drawbell. Comminution originates from point
load breakage, shearing, crushing, and abrasion between rock blocks as they migrate downward into a
drawbell, increasing the finer broken ore size distribution with depth. A simple methodology is proposed to
estimate an approximate range of fines migration for different draw column heights, based on the technical
literature published on internal erosion and fines segregation in earth dams. In addition, the shape of the
BSD curve into a drawbell as a function of column height and undercut depth will be examined. The latter
will account for the influence of the in situ stresses on the primary fragmentation and initial BSD below
the cave back as the cave propagates and the column height grows. Experimental data from the literature
examining particle breakage under compression/shear will be considered in order to characterize the BSD
curve as a function of column height and depth.

1 Introduction
Rock fragmentation is one of the most important factors in the performance of a block caving operation (Van
As & Van Hout 2008; Moss 2012). In addition, it is well accepted that caving fragmentation incorporates
three components: the in-situ fragmentation, representing the natural discrete fracture network distributed
throughout the rock mass; the primary fragmentation, arising from stress-induced fractures propagating
in the cave back; and the secondary fragmentation, resulting from block impact, comminution and other
fragmentation processes occurring within the draw column (Laubsher 1994; Eadie 2003). In the context
of secondary fragmentation, this involves a variety of mechanisms not all of which are well understood
(Brown 2007). To date, empirical design charts produced by Laubscher (1994) as well as several numerical
approaches described below are generally used to assess secondary fragmentation.

The Block Caving Fragmentation (BCF) model devised by Esterhuizen et al. (1996) employs empirical
relationships to assess the primary and secondary fragmentation as well as hang-up potential. Although this
approach is able to quantify secondary fragmentation and has been calibrated using mine data, its reliability
has been questioned (Butcher 2007). Experiences at Palabora found the BCF over predicted the percentage
of oversized blocks (> 2 m3) and under predicted the number of hang-ups (Ngidi & Pretorius 2011). Pierce

Caving Mechanics

(2009) proposed an alternative methodology built on a particle flow model (REBOP) and laboratory testing
with an annular shear cell.

In addition, a hybrid approach was developed during MMT1, which includes empirical rules based
on comminution theory (Kojovic 2010). Weatherley & Pierce (2011) compared the performance of
these methods, but the predictions did not fully match the data collected at Ridgeway Deeps (RWD),
underpredicting the fines production. Instead, they concluded that Pierces method (REBOP) performed
better. Finally, Rogers et al. (2010) has proposed another methodology based on a stochastic Discrete
Fracture Network (DFN) approach, which captures the breakage of blocks as they move through the cave,
but lacks computing the fines production.

Although there is significant interest in developing the basis and fundamentals of secondary fragmentation,
some topics have received less attention, such as the influence of: fall height, block rotation and rockfall
impact on the muckpile surface where an air gap is present; vein, rock strength and non-persistent joint
distributions in the blocks; the initial arrangement of caved blocks and subsequent block interactions; the
broken ore density and its distribution within a draw column; and the role of fines in cushioning block
interactions. Several of these are influenced by BSD and its evolution down through the draw column
height over time.

2 Secondary fragmentation assessment by means of large compression tests

Secondary fragmentation is commonly attributed to a combination of block splitting and rounding, with
block movement being controlled through a combination of shear and compressive stresses occurring in the
draw column zones (Pierce 2009). Replicating these conditions through laboratory testing provides a useful
means to develop empirical rules of thumb or numerical model calibration. Accordingly, published results
involving large triaxial compression tests CID (Consolidated Isotropically Drained) are a valuable source of
data to evaluate secondary fragmentation of broken ore within a draw column. These are discussed below.

2.1 Large compression tests to evaluate secondary fragmentation

As shown in Figure 1, secondary fragmentation can be linked to two modes of stresses acting within a draw
column. In the center, the broken ore undergoes anisotropic compression. This is similar to the load path
conditions applied in an oedometer test. Adjacent to this, towards the outer periphery of the column, the
broken ore experiences shear stresses. This is similar to the load path conditions applied in direct simple
shear tests.

Of interest are laboratory results involving a unique, large triaxial device capable of testing samples with
1 m diameter, previously applied to rockfill characterization studies (Marsal 1973; Verdugo et al. 2007).
These serve as a proxy for the load path experienced in the draw column (Figure 2), which includes the
development of both compression and shear zones (Figure 3). Maximum particle sizes of the rockfill and
waste rock tested in this facility have reached 15-20 cm. These represent a valuable data source that can
be extended to secondary fragmentation studies, given similarities in the intrinsic rock properties (block
strength, angularity, aspect ratio), as well as block size distribution, initial material density, and confining
pressure (i.e., column height). Two main contributions of these large scale tests is the shearing strength
chart for rockfill by Leps (1970) and geotechnical characteristics of large waste rock dumps (Valenzuela et
al. 2007). Large diameter triaxial tests offer a useful alternative to estimate the comminution of broken ore
in a draw column under shear and compression stresses based on samples with a maximum particle size
of 15 cm and load paths simulating a draw column of 200 m overload height. Results using this device for
specific testing of secondary fragmentation are reported in a later section.

Caving 2014, Santiago, Chile

2.2 Size scaling from large compression samples to broken ore sizes

Blocks in a draw column are much larger than the maximum particle size tested in the large triaxial
compression tests discussed above; thus a scaling relationship is needed to apply test results to block size
distribution relationships. The most popular technique to scale geotechnical properties from small granular
samples to large broken rock is based on the parallel gradation method or parallel size distribution method
first proposed by Lowe (1964). This technique involves shifting the size distribution curve (on a semi-log
plot) by a factor S to scale the smaller sized triaxial compression tests to the larger scale in situ material
(Figure 4). De la Hoz (2007) demonstrates that this technique is suitable in sands and gravels, but for larger
sample sizes, the strength and stiffness tends to decrease (Frossard 2013).

It is also well reported that the strength of individual blocks decreases as block size increases (Hoek &
Brown 1980; Santamarina & Cho 2004). On the other hand, the block coordination number (i.e., number
of contact points) resulting from the particle packing also influences block fragmentation. Particles with
more contact points are generally subjected to a lower probability of secondary fragmentation due to the
loads being more distributed (McDowell et al. 1996). Thus, a large block adjacent to a number of smaller
blocks is less susceptible to fragmentation due to its higher coordination number but more susceptible to
containing strength reducing defects (e.g., veining, non-persistent joints, etc.). Some authors agree that the
coordination number is more significant than strength reduction due to block size; however there is not
enough experimental data to confirm this assumption.

Moreover, another effect related to the coordination number is the contact nature among adjacent blocks.
Large block are susceptible to breakage depending on how its flaws are aligned relative to the contacts
acting on it (e.g., corner-side or side-side). Research applying empirical and numerical techniques is
currently underway to quantify the influence of the particle arrangement around large blocks.

Figure 1 Stresses within a broken ore zone Figure 2 Representation of draw column
assuming narrow flow width (Laubscher 1994) and stress modes to triaxial compression loading
interactive flow (Susaeta 2004) conditions

Caving Mechanics

Figure 3 Stresses developed in a triaxial Figure 4 Gradation parallel method

compression test

3 Methodology to assess fines migration in a draw column

3.1 Background

Fines migration (Figure 5) has been investigated by several authors conducting empirical studies (Hashim
& Sharrock 2012; Chen et al. 2009; Castro 2006) and numerical modelling (Leonardi et al. 2008; Pierce
2009). This segregation process has been identified as a key element in draw control and cushioning of large
blocks (Laubscher 1994), and mudrush risk due to the presense of water (Jacubek et al. 2012). Although
significant advances have been reported on this theme, there is a lack of a methodology to quantify the
fines migration for different draw column heights in caving operations. In contrast, numerous studies exist
investigating fines segregation in granular materials related to internal erosion, piping, suffusion and filter
design in earth dams. These are reported in standard design manuals such as the Earth & Rock-Fill Dams
General Design & Construction Considerations (2004) and Design and Construction of Levees (2000).

Major advances in Dam Engineering by Kezdi (1979), Sherard (1979) and Kenney & Lau (1985) make
possible the assessment of segregation potential of fines (< 4.75 mm size) from larger particles (> 4.75 mm
up to 1,000 mm). These methods focus on fines segregation due to seepage through an earth fill dam, which
is not fully comparable to the fines migration in a block cave draw column. However, the fines segregation
from a broken ore zone is a dynamic process involving the continuous downward progression of blocks,
including internal movements among blocks, facilitating the migration of fines from the broken ore. Hence,
the fines migration in caving is somewhat comparable to the internal erosion in dams.

Applying Kezdis (1979) method to an ore column, the initial gradation can be divided into a coarse and
fine gradation as shown in Figure 6. The key hypothesis of this method is that the segregation of the fines
gradation will occur if the ratio D15 /d85 is higher than 4, where D15 is the particle diameter for the 15%
of mass passing of the coarse gradation and d85 is the particle diameter for the 85% of mass passing of the
fine gradation. Here it is necessary to establish the initial gradation, which then allows the calculation of the
fine and coarse gradations relative to a specific block size (black dashed line in Figure 6). The segregation
potential can then be checked by applying this procedure to several block sizes. A complete explanation of
this method can be found in Kezdi (1979), Chapuis (1992) and Li & Fannin (2008).

3.2 Fines migration and broken ore size distribution

It is well accepted that fines move more rapidly than coarser particles through the draw column (Laubscher
2000). This can be used to develop a conceptual fines migration sequence occurring down through an ore

Caving 2014, Santiago, Chile

column, guided by Kezdis method together with results from large diameter triaxial tests (Figure 79).
The broken ore zones in the column are defined with roman numerals adjacent to the column, with each
representing either the addition of broken ore to the top of the muck pile (through caving from the cave back)
or subtraction representing removal of the zone through extraction at the draw point. Other nomenclature
in these figures includes the numbers 0, 20, 40until 200, which are related to the overload height of
broken ore. Included with the overload notation are the letters C and F, which refer to the coarse and
fine gradations presented in Figure 6. Thus, a C40 is a coarse gradation under 40 m overload height and
F120 is a fine gradation under 120 m overload height. Furthermore, the sequence takes into account 11
stages, which in turn includes 3 sub-stages (letters a, b, c). Letter a corresponds to the initial secondary
fragmentation of the ore after a vertical movement of 20 m; b is related to the corresponding migration
of fines (blackened gradations); and c corresponds to the broken ores response to the overload pressures.

The sequence starts at stage 0 in the undercut level (Figure 7). The blasted rock is assumed to be mined,
so that the cave initiates and broken ore falls into the undercut. Next, the procedure assumes that two new
portions of broken ore are released from the cave back (Stage 1a), comprising a coarse and fine gradation
(C0+F0). At this early stage, no fines migration is assumed for the b sub-stage because the draw column
is not developed enough to permit significant internal movement of the broken ore. Thus stage 1b remains
the same as stage 1a (this is the same for stages 2a and 2b).

At stage 1c, an overload of 20 m is applied and the stresses are disproportionately concentrated on the
coarse gradation. In response, C0 changes to C20 and some fines are generated (equal to F20 minus F0).
Thus, the fines production is increased with every sub-stage c (i.e., F40 F0, F60 F0, through F200
F0) due to overloading through the progression of the draw column (roman numerals i through xx). A
similar sequence repeats for stage 2. In stage 3, the ore column is mature enough to consider the migration
of fines. Thus, for stages 3 and 4, it is assumed that half of the fines will migrate downward to the broken
ore zone below and, for stages 5 to 7, all of the fines are assumed to migrate into the underlying broken
ore zone. For stages 8 to 10, it is assumed that the fines will migrate two broken ore zones down (or 40 m
downward). Finally, this sequence is described until stage 10, which corresponds to a draw column height
of 220 m.

Figure 5 Segregation process along a Figure 6 Kezdis method to divide an initial gradation into
draw column a fine and coarse gradation. Thus, the initial gradation is the
mixing product of both fine and coarse gradations. Fine and
coarse particles are defined as particles smaller and larger
than 20 mm, respectively

Caving Mechanics

Figure 7 Fines migration sequence (Stages 0 - 2)

Figure 8 Fines migration sequence (Stages 3 - 4)

Figure 9 Fines migration sequence (Stages 5 10)

4 Evolution of BSD at different ore column heights

In order to study the evolution of the BSD within an ore column, a large triaxial compression test on
saturated waste rock material was carried out. The material tested corresponds to a granodiorite with a UCS
of 140-150 MPa and specific gravity (Gs) of 2.77. The triaxial test was performed applying a confining
pressure of 2.5 MPa, resulting in a deviatoric stress at failure of Dsf = (s1- s3) of 9.5 MPa and vertical
deformation at failure of 18%. The specimen dimension was 100 cm diameter and 180 cm height, with a
maximum particle size of 15 cm and specimen density of 19 KN/m3. Thus, this triaxial test approximates a
broken ore zone with an overload of 200 m.

Using the parallel gradation method, the initial gradation and gradation after testing were scaled to a
maximum block size of 4 m, imitating block sizes within a draw column (Figure 10). Thus, the initial
scaled gradation represents the primary fragmentation curve. Finally, it is possible to interpolate several

Caving 2014, Santiago, Chile

gradation curves (nine in total) representing different confining pressures (or overload heights) between the
initial gradation (no stress) and gradation after testing (s3 = 2.5 MPa), as shown in Figure 10. Note that
these interpolated curves can be used to characterize the coarse gradations (C0, C20C200) as well as the
fine gradations (F0, F20F200) in the fines migration sequence (Figures 7-9).

After scaling the gradations to the in situ block sizes, the fines migration sequence is applied. Figure 11
presents the evolution of the scaled gradation (or BSD) for three different column heights. Based on these
results, it is possible to observe how the initial BSD curve transitions at stage 10 to a more linear trend than
an S shaped curve.

The magnitude of the fines migration on the final BSD at the drawpoint depicts a reduction in the average
size of the material. For example, in Figure 11, the 50% of mass passing due to loading only (dashed curve)
drops from 0.6 m to 0.3 m (two times less). However, this reduction considering both loading and fines
migration (stage10 curve) drops from 0.6 m to 0.04 m (15 times less). Thus, the combination of large triaxial
tests and the fines migration analysis is able to capture the evolution of the BSD from its initial gradation
(primary fragmentation) to that mined at the drawpoints. Moreover, the impact of the fines migration on the
BSD can be determined by comparing the initial scale gradation after testing curve (black dashed curve in
Figure 10) and the stage 10 curve (grey color). The former is representative of the fragmented ore close
to the drawpoint with a muckpile overburden of 200 m but no fines migration, and the latter represents the
same overburden but including the fines migration process.

Figure 10 Initial gradation (BSD) and gradation Figure 11 Evolution of Initial gradation for three
after testing, including scaled curves (parallel different stages. Stages 3, 6 and 10 represent a
gradations) column height of 80, 140 and 220m, respectively.

5 Discussion
The shape of the block size distribution has been addressed in terms of applying relationships drawn from
large compression tests (triaxial CID test) and a fines migration sequence analysis based on techniques
developed to study internal erosion in earth dams. Some assumptions included in this work are discussed
as follows:

Triaxial tests used as a proxy for simulating the shear and compression zones within a broken
ore column: As explained in section 2.1, a triaxial test combines elements of both simple shear
and oedometer tests but this hypothesis needs to be corroborated with a conventional laboratory
testing program involving simple shear, oedometer and triaxial CID testing, with special focus
on particle breakage. For example, simple shear tests could involve significant splitting but much

Caving Mechanics

less fines generation compared to an oedometer test, which in turn could experience much more
fines generation due to stress concentrations among particles but less splitting than simple shear.

Fines migration sequence: This sequence applies segregation and piping criteria used for earth
dams. However, due to the influence of water flow on fines migration, this methodology could
overestimate the fines arriving at the drawpoints. In addition, this sequence assumes different
segregation rates, which still require validation from field data and percolation rate studies (e.g.,
Bridgwater et al. 1978; Cardew 1981; Pierce 2009).

Block size scaling effects on broken ore fragmentation: This issue has not been addressed in this
study and requires more experimental and numerical studies to better understand the influence of
veins and smaller discontinuities present in larger blocks on their overall fragmentation within
a draw column.

6 Conclusions
A conceptual model has been developed accounting for the influence of secondary fragmentation and fines
migration on the block size distribution of broken ore encountered at a drawpoint. Firstly, the broken ore
fragmentation by shear and compression has been investigated using a large diameter compression test on
waste rock (Triaxial CID). Secondly, the parallel gradation method is used to scale the standard particle
size from the triaxial test sample to that representing the in situ block sizes. Thirdly, a simple method is
proposed to estimate an approximate range of fines migration for different ore column heights. This suggests
a more linear BSD than an S or exponential shape. Finally, more efforts are required to understand the
fundamentals of secondary fragmentation and the prediction of the drawpoint BSD with a higher accuracy.

Bridgwater, J, Cooke, MH & Scott, AM 1978, Inter-particle percolation: Equipment development and
mean percolation velocities, Transactions of the Institution of Chemical Engineers, vol. 56,
pp. 157-167.

Brown, ET 2007, Block Caving Geomechanics, Indooroopilly: Julius Kruttschnitt Mineral Research Centre.
ISBN 978-0-98003622-0-6, Queensland.

Butcher, RJ & Thin, IGT 2007, The inputs and choices for predicting fragmentation in block cave projects,
in Proceedings First International Symposium on Block and Sub-level Caving, Southern
African Institute of Mining and Metallurgy, Johannesburg, pp. 3549.

Cardew, PT 1981, Percolation and mixing in failure zones, Powder Technology, vol.28, no. 1, pp.119-128.

Castro, R 2006, Study of the mechanisms of granular flow for block caving, PhD Thesis, University of

Chapuis, RP 1992, Similarity of internal stability criteria for granular soils, Canadian Geotechnical
Journal, vol. 29, no. 4, pp. 711713.

Cheng, YM,Liu, ZN,Song, WD &Au, SK 2009, Laboratory Test and Particle Flow Simulation of Silos
Problem with Nonhomogeneous Materials, Journal of Geotechnical and Geoenvironmental
Engineering, vol. 135, no. 11, pp. 1754-1761.

Eadie, B 2003, A Framework for modeling fragmentation in block caving, PhD Thesis. The University of
Queensland. Australia.

Caving 2014, Santiago, Chile

Esterhuizen, GS, Laubscher, DH, Bartlett, PJ & Kear, RM 1996, An Expert System Approach to Predicting
Fragmentation in Block Caving, Proceeding Massmin Mining Methods, SAIMM.

Frossard, E, Ovalle, C, Dano C, Hicher, PY, Maiolino, S & Hu, W 2013, Size effects due to grain crushing in
rockfills shear strength, Proceedings of the 18th International Conference on Soil Mechanics
and Geotechnical Engineering, Paris.

Hashim, MHM, & Sharrock, GB 2012, Dimensionless percolation rate of particles in block caving mines,
In: MassMIN 2012 Conference Proceedings.MassMin 2012: 6th International Conference and
Exhibition on Mass Mining,Sudbury, Ontario, Canada.

Hoek, E, & Brown, ET 1980, Underground excavations in rock, Instn Min. Metall, London.

Kenney, T, & Lau, D, 1985, Internal stability of granular lters, Canadian Geotechnical Journal, vol. 22
pp. 215225.

Kezdi, A 1979 Soil physics selected topics. Elsevier Scientific Publishing Co., Amsterdam.

Kojovic, T 2010, Application of the Hybrid Model to RWD, Subproject Report submitted to MMT2
Secondary Fragmentation Project.

Laubscher, D 1994, Cave mining the state of the art, The Journal of The South African Institute of
Mining and Metallurgy, pp. 279-293.

Leonardi CR, Owen DRJ, Feng, YT & Ferguson WJ 2008, Computational modelling fines migration in
block caving operations, Proceedings of the 5th international conference and exhibition on
mass mining, Lulea, Sweden.

Li, M & Fannin, RJ 2008, Comparison of two criteria for internal stability of granular soil, Can. Geotech.
Journal, vol. 45, pp. 1303-1309.

Lowe, J 1964, Shear Strength of Coarse Embankment Dam Materials, Proc. 8th International Congress
on Large Dams, vol.3, pp. 745-761.

Marsal, RJ 1973, Mechanical Properties of Rock Fill Embankment- Dam Engineering, (ed.) Hirschfelt
and Poulos, John Wiley. New York.

Mcdowell, GR, Bolton, MD, & Robertson, D 1996, The fractal crushing of granular materials, Journal of
the Mechanics and Physics of Solids, vol. 44, no. 12, 2079-2102.

Ngidi, SN & Pretorius, DD 2011, Impact of poor fragmentation on cave management, In 6th Southern
African Base Metals Conference. The Southern African Institute of Mining and Metallurgy,
pp. 111-122.

Pierce, M 2009 A Model for Gravity Flow of fragmented rock in Block Caving Mines, PhD Thesis, The
University of Queensland.

Rogers S, Elmo D, Webb, G, & Catalan, A 2010, A discrete fracture network based approach to defining
in situ, primary and secondary fragmentation distributions for the Cadia East panel cave, In
Caving 2010, Proceedings of the 2nd International Symposium on Block and Sublevel Caving,
Perth, (Edited by Y. Potvin), Australian Centre for Geomechanics.

Santamarina, JC & Cho, GC 2004, Soil Behaviour: The role of particle shape, Proc. Skempton Conference,

Caving Mechanics

Sherard, JL 1979, Sinkholes in dams of coarse, broadly graded soils, Transactions, 13th International
Congress on Large Dams, New Delhi, India, vol. 2, pp. 25-35.

Susaeta, A 2004, Theory of gravity flow (part 1), In Proceedings of the 5th international conference and
exhibition on mass mining, Santiago, Chile, pp. 167-178.

U.S. Army Engineer Manual 2004, Earth & Rock-Fill Dams General Design & Construction Considerations,
EM 1110-2-2300, available from: <
Publications/EngineerManuals/EM_1110-2-2300.pdf>. [1 April 2014].

U.S. Army Engineer Manual 2000, Design and Construction of Levees, EM 1110-2-1913, available
from: <
EM_1110-2-1913.pdf>. [1 April 2014].

Valenzuela, L, Bard, E, Campana, J & Anabalon, ME 2008, High waste dumps - challenges and
developments, In: Rock Dumps 2008, Fourie, A. (Ed.), Australian Centre for Geomechanics,
Perth, pp. 65-78.

Verdugo, R, Peters, G, & Bejarano, I 2007, Evaluacin de parmetros geomecnicos de suelos gruesos,
VI Chilean Geotechnical conference, Valparaso.

Weatherley, D, & Pierce, M 2011, Progress Report - Fundamentals of Caving Fragmentation, Report to
the Mass Mining Technology 2 Project, February.

Caving 2014, Santiago, Chile

Analysis of hangup frequency in Bloque 1-2,

Esmeralda Sur Mine
E Viera Codelco, Chile
E Diez Codelco, Chile

Monitoring and analysis of secondary breakage is relevant in productivity of Block- Panel Caving mines,
especially in mines where production area with low percentage of draw column (<30% of primary column)
is majority in comparison to an area without geomechanical constraints. A smaller capacity of secondary
breakage can generate deviations in draw strategy, which may compromise production plans in the short
and the long term.

Bloque-1 of Esmeralda mine (located in the Sub -5 level El Teniente mine ) reflects the condition of
extraction described, because 35% of production area has draw rate constraints (percentage of primary
draw column <30%) and 42% of the open area has no draw rate constraints. In this particular sector, a
low capacity of secondary breakage (S.B) can generate wrong practices of draw rate strategy that may
impair considered planning strategies (dilution control, angle control and regularization of draw heights),
due to the reduced availability of area in state of caving propagation process (which has higher hanging
frequencies) and increase of extraction in area without draw rate constraints (lower hanging frequency).

The main goal to this work is to perform a quantitative analysis of the hanging frequency to Bloque-1,
correlating this variable with the extraction percentage of primary column and granulometry curves.

For Bloque -1 a record of hangings and activities related with secondary breakage of draw points for the
year 2013 has been made, which we have calculated the rate of extraction between hanging (REH) of
production drifts throughout the detailed period. This index has been related to granulometry curves and
the average percentage of primary column in each production drift.

It is aimed to establish a pattern that allows including the frequency of hanging by draw point in short-term
plans, considering the predominant granulometry curves and the percentage of average primary column
for future projects to be undertaken in Esmeralda Mine.

1. Introduction
A greater understanding of rock fragmentation is vital for Block/Panel Caving mines design and planning
stages, as well as the variables that depend on the rock fragmentation. They go from design of layout
(spacing of drawpoints), to the daily secondary breakage requirements. Because of its importance, it is also
necessary to perform back analysis studies that allow to estimate fragmentation predictive models, and
management tools, in order to be able to monitor the predominant granulometry behavior.

The following study performs a back analysis with the data recorded in Esmeralda Sur, which have the
observed granulometry reports in ranks, and the secondary breakage (SB) made in 2013.

Currently, Esmeralda Sur is formed by Bloques 1 and 2, which, until March 2014, showed an incorporated
area of 30,589 m2 and 13,163 m2 of production, respectively.

Esmeraldas Mine Sector Sur has a volume of mafic rocks, known as El Teniente Mafic Complex (Figure
1), intrusioned by subvertical matters of felsic rocks. Felsic rocks present in Bloque 1 are located between

Caving Mechanics

production drifts 23 and 33, and trenches 30 to 35 to the mine, where dioritic porphyry can be found in
contact to hydrothermal breccias of dioritic porphyry in the NE.

On the other hand, Bloque 2 had one intruded of tonalite that occupied half of the surface, cutting it
diagonally from NW to SE, which suffered later intrusions from matters of anhydrite hydrothermal breccias,
biotite hydrothermal breccias, and finally from a microdiorite porphyry:

Figure 1 Esmeralda Sur Lithology

2 Objectives
The following study aims to analyze the Esmeralda Sur hanging drawpoints frecuency, by considering the
secondary breakages made, and the recorded granulometry during 2012 and 2013.

3 Methodology

3.1 Information record

The hang drawpoints condition information is recorded in 2 databases:

Control Production Mine database: the mentioned database is hold by analysts routes that record
relevant information about the drawpoints condition, such as humidity, hangs and observed
granulometry conditions. Ideally, the most relevant productive areas must be checked daily.
Esmeralda Mine Operational database: the draw area records every secondary breakage performed
in the drawpoints. This database provides the secondary breakage date (day and shift), amount of
explosives used, drift and trenches in which the breakage labors are made.

When comparing both databases, which shows a greater reliability analyzing the hanging drawpoints
frequency is the second one, since it covers a greater temporality (the first one fairly shows 1 daily shift
with information, the second one shows 3 daily shifts with records).

Caving 2014, Santiago, Chile

In order to quantify the effect generated in production by the secondary breakages, made at drift level, the
Secondary Breakage Index is defined as it follows:

This index must be defined on a mine unit (from mine scale size to drawpoints level) and at a certain time

3.2 Granulometry information record.

This information is compiled by a team of eight production control analysts, at least once a day in major
areas. During their routes, they pick up information related to drawpoint condition, granulometry, humidity,
lithology and dilution.

The information used for this work was the drawpoints granulometry, according to the following overlaps:
fragments lower than 5 cm (A), between 5 and 25 cm (B), from 25 to 50 cm (C), from 50 to 100cm (D),
and rocks greater than 100 (E). When classifying rocks into one of the overlaps, the three right-angled
sizes arithmetic mean to the volume it should hold it is obtained. The granulometry information is a visual
appreciation to the percentages of 5 granulometry overlaps on the slope surface made in the drawpoints and
it is calibrated by point measures made on ground with a tape measure. The information obtained on ground
and stored into a database has continuity from the beginning of Mina Esmeralda. Note that, additionally
the drawpoints can be hang, which makes the data collect not to be safe for the field personnel, thus not
collecting granulometric information. This lack of information creates a bias within the database, which
affects the thicker granulometry representativeness. For this condition with no granulometry surveys, a
100% of the granulometry greater than 1m is assigned.

The analyzed data for this article were the thicker granulometry information (>50cm), obtained from
January 2012 until March 2014, which covers almost the entire study area productive life so far.

Finally, the scheme of stages to consider for the hang drawpoints and granulometry analysis is detailed

All of the secondary breakage information recorded between January-November 2013 will be considered.
During 2012, the database information was not 100% completed, therefore it is not considered within the

Note that the secondary breakage analysis only considers the drawbells in which the caving heading has
completely passed (production drawbells). The previous criterion leaves out of the analysis the drawbells
with swelling condition, which do not show hanging conditions.

As for a greater understanding concern, the analysis will be developed at production drift level. This
consideration allows adding the secondary breakage variable within the short term plans in an easy way.

4 Results
When considering the secondary breakages labors made for drift and the draw during 2013, the REH
index is calculated for every month, which is compared to the average primary column draw, obtaining the

Caving Mechanics

Figure 2 Information process for analysis scheme

Figure 3 REH as a function of extraction

As the primary column draw rate increases, the REH index increases proportionally, thus showing a greater
dispersion of amounts above the 30% of the average primary column draw. The previous behavior can be
explained by the following points:
Increasing geomechanical draw rates: the continuous geomechanical increasing draw rates occurs as
certain limits, previously defined, are exceeded. Therefore, the greater extraction of primary column,
the greater productive capacity per area (no operational restrictions).

Caving 2014, Santiago, Chile

Variations in the material handling system: as for the Esmeralda Sur case, this variation occurs when
ore passes are incorporated, which becomes into a more efficient material handling system, since it
reduces the ore hauling mean distance.
Reduction of hanging drawpoints and/or rock sizes: when reducing the hangings and/or rock sizes,
it reduces the operational interferences, which increases the material handling system productivity
(more operational continuity).

In the following section, the reduction of hanging drawpoint and/or rock sizes will be explained in detail,
by analyzing if there is a grater granulometry record reduction (type E), as the extraction of percentage of
primary column increases. The Figure 4 shows the granulometry evolution according to the primary draw
percentage, by considering a data record from 2012 and 2013 in Esmeralda Sur:

Figure 4 Fragmentation observed at drawpoints as a function of extracted column

According to the Graphic 2 information, it is evident that the coarse fragments remains between 50% and
60% from the beginning of the draw up to the 80% of the primary rock column, getting to its maximum in
this point, and decreasing to the 30% in drawpoints with higher extraction percentage of primary column
(120%). Regarding the two granulometry overlaps behavior, it can be ensured that it is despair, as the D
granulometry remains relatively stable, on the other hand, coarse fragments varies according to the draw.
The aimed granulometric curve for this mining method foresees a high raise in fine graining, after it exceeds
the columns 100%, the primary rock is replaced for fragmented material. Such decrease is mitigated at
Bloque-1, where the points having 120% of draw remain an average of 20% of size E rocks with the
capability of blocking the ore flow.

Starting at 80% of primary column extraction, a decrease in performed secondary breakage should be seen,
by which it should raise the REH index defined above. This result is important from the short term planning
point of view, since it is possible to take this value as a reference to increase the productivity per drift,
within the short term.

Linking the Figures 1 and 2, we can see that the REH rate increases proportionally with the extraction,
however reports of fragmentation type E no evidence of variation up to 80% extraction of the primary
column, whereby the increasing of geomechanic draw rates and improvements in the mineral handling
system are more relevant in that stretch.

Caving Mechanics

When reducing the plotted values (in Figure 2), calculating the REH index within the entire period (January-
November 2013 period and not monthly as performed in the first case), the obtained results are as shown
in Figure 5.

Figure 5 Average extraction of primary column vs REH

Figure 5 shows the same behavior seen when plotting all of the monthly data. However, the difference
is that in this graph some regression function can be found. The logarithmic regression (explained in the
graph) shows an acceptable correlation, and then it is possible to use the regression equation within the
monthly planning exercises.

In order to exemplify this analysis usage in the short term planning, the defined parameters for production
method during March at Bloque-1 are considered as listed in Table 1.

Table 1 Parameters Bloque-1 March 2014.

Production Drift Productivity (tpd) Amount of drawpoints Primary column average

extraction (%)
21 301 3 14
23 810 7 18
25 1.244 10 19
27 2.153 13 46
29 2.579 15 58
31 2.778 16 73
33 2.817 16 76
35 2.879 16 57
37 1.379 8 38
53 348 5 6
55 544 8 13
57 1.095 10 15
59 1.280 11 11

Caving 2014, Santiago, Chile

With the average percentages we can obtain the REH index for each one of the production drifts detailed
above (in the regression formula). Considering that we have average planned productivity, it is possible to
estimate the entries amount in production drifts, to perform daily secondary breakage labors (according to
REH formula, Table 2):

Table 2 Amount of overlaps per secondary breakage labors

Production Drift REH Production drift entries to S.B labors per Amount of points in S.B
day per day (*)
21 826 0,36 1
23 959 0,84 3
25 1.244 1,26 4
27 2.153 1,55 5
29 2.579 1,73 5
31 2.778 1,74 5
33 2.817 1,74 5
35 2.879 1,94 6
37 1.379 1,06 3
53 412 0,84 3
55 789 0,69 2
57 867 1,26 4
59 725 1,77 5
(*) in base to 3 points in S.B per production drift entry.

For the demanded rate during the March program and for the average primary column draw level per drift,
it is necessary to perform secondary breakage labors in 50 distributed points, as shown in table 2. When
considering the full month and the time this labors per drift take, it is possible to estimate the monthly
required time for secondary breakage labors, which can be included in the monthly plan.

Note that to satisfy these high productivity rates, is necessary the detailed quantity of secondary breakage per
production drift. Significant deviations in the distribution of secondary breakage impact in draw strategies
in the short term planning (for example humidity control, dilution control, extraction angle control, etc.).

This is exemplified in production drift C-25 and C-27 of Bloque-1, which has completely different conditions
of granulometry (Figure 6). In production drift C-25, the hangings frequency is higher as compared to
C-27, whereby the requirement of secondary breakage in C-25 is increased. However more resources are
allocated to reduce hanging drawpoints on production drift C-27, which helped to create a difference
between the two extraction areas (due to increased availability of area in production drift C-27), causing a
difference in extraction height in both areas:

3 Discussion and conclusions

A correlation is seen between the primary draw percentage and the REH index, which is directly proportional
to the primary column draw percentage.

Caving Mechanics

Figure 6 Extraction height between production drift C-25 and C-27

It is possible to state that starting at 80% of primary column extraction, a change was seen in the
granulometric distribution reports as the recorded ranks, decreasing in a progressive way the fragments
percentage reports above 1 m, starting at average primary column draw level. This topic development will
be useful for planning and control of the following mine Esmeralda Sur productive block.

It is possible to estimate the secondary breakage level to be made considering the primary column draw
percentage and the planned productivity in the short term programs (at productive drifts level). Differences
in the secondary breakage requirements make deviations in the short term program, which could affect
draw strategies (draw angle increase, dilution control, humidity control, etc.).

Contributions from Gabriel Tapia (Universidad de Chile) and Boris Leal (Universidad de Santiago de
Chile) are gratefully acknowledged.

Montecino, N, Castro, R 2013,Modelo de mezcla de fragmentacin secundaria en minera de block/panel
caving, Laboratorio de Block Caving, Santiago, Chile.

Moss, A, Russell, F & Jones, C 2004, Caving and fragmentation at Palabora: prediction to production,
Proceedings MassMin 2004, Santiago, Chile.

Caving 2014, Santiago, Chile

A 3DEC-FLAC3D models to predict primary

fragmentation distribution in Cave Mines
T V Garza-Cruz Itasca Consulting Group, Inc., USA
M Fuenzalida Itasca Consulting Group, Inc., USA
M Pierce Itasca Consulting Group, Inc., USA
P Andrieux Itasca Consulting Group, Inc., USA

Prediction of primary fragmentation is of great importance in cave mining. Numerical models provide an
opportunity to link caving induced stresses at failure with associated fragmentation of rock masses. In this
paper, the results of a set of 3DEC simulations, based on a methodology developed by Itasca to model the
fragmentation of massive veined rock masses, are used to relate induced stresses at failure to percentage
of large fragments produced. A new approach is described in which such fragmentation relationship is
applied to predict the primary fragmentation distribution in a FLAC3D cave model based on computed
stresses at failure.

1 Introduction
The spatial distribution of primary fragmentation as a result of mining operations becomes critical in the
context of cave mining. The ability to predict an approximate size distribution and likelihood of generating
large fragments as a function of rock mass strength and caving induced-stresses can provide significant
insight into the mine design and risk assessment.

Discontinuum approaches, such as DEM, can realistically simulate mechanisms such as spalling and
bulking, but are too computationally expensive to be implemented in a large-scale caving model. For this
reason continuum approaches are required to ensure reasonable computational times.

A methodology is presented here that combines the characterization of a heavily veined massive rock
mass (by construction and testing of a synthetic rock mass), with results of a continuum cave-scale model
(informed by the SRM-derived rock mass strength) to predict the distribution of fragment sizes as a function
of rock mass strength and cave back stress as illustrated in Figure 1 .

Figure 1 Diagram illustrating the proposed methodology for primary fragmentation prediction

Caving Mechanics

2 Synthetic rock mass (SRM) modeling

A synthetic rock mass was developed following the methodology described by Garza-Cruz & Pierce (2014)
using 3DEC to model a heavily veined massive rock mass. The rock mass was represented by a collection
of interlocked tetrahedral blocks bonded at their contacts. The block contacts represent a network of low
persistence veins as well as unveined intact rock. The advantage of 3DEC (Itasca, 2013) is that it allows for
the construction of samples with zero initial porosity and highly interlocked irregular shapes that provide
resistance to moment after contact breakage. This allows one to mimic the high uniaxial compressive
strength to tensile strength ratios and friction angles typically exhibited by hard rock. Virtual testing of
SRM samples populated with real field data allows us to characterize the mechanical behavior of a rock
mass and, when subject to caving stress paths, provides information about the associated fragment size
distribution, both of which are emergent results.

2.1 Sample construction

Two different samples (a 8x8x8-m and 18x18x18-m) were constructed by assembling a collection of highly
interlocked tetrahedral blocks with approximate edge length of 0.5 m using KUBRIX-Geo and importing
them into 3DEC to be populated with pertinent material properties (Garza-Cruz and Pierce 2014).

Block contacts were populated with three different tensile strength distributions (accounting for intact rock
and veins) to examine their impact on rock mass strength. The three distributions correspond to three
different geological domains and include a certain percentage of zero strength veins that constitute 12%,
13% and 22% of all strength measurements for the strong, medium and weak domains, respectively
(Figure 2). In order to populate a sample, each block contact was assigned a tensile strength value randomly
selected from the cumulative distribution of rock tensile strength of interest (Figure 2), and its local cohesion
was set to be 2.5 times such tensile strength. The used cohesion-to-tensile-strength ratio was based on a
sensitivity study in which such ratio produced macro UCS/tensile strength ratio in the order of 10-20, which
is consistent with typical observations. In all models, the blocks were defined as elastic and zoned with an
approximate edge length of 0.5 m. The blocks and block-contacts micro-mechanical properties used are
summarized in Table 1.

Figure 2 Synthetic rock mass sample generation procedure. 3DEC 8x8x8-m sample after trimming 1 m off
all sides (left). Tensile-strength cumulative distribution for the strong, medium and weak categories of
rock used in the 3DEC models (center). Vertical cross-section of the sample showing the block contact tensile-
strength distribution based on the medium strength category (right). Block contacts in black have zero
tensile strength and cohesion value, and account for 13% of contacts.

Caving 2014, Santiago, Chile

Table 1 Elastic block and contact mechanical properties used in the model
Block Properties
Youngs Modulus 51 GPa
Poissons Ratio 0.25
Density 2650 kg/m3
Contact Properties
Normal stiffness 163.2 GN/m
Shear stiffness 81.6 GN/m
Peak friction angle 21
Residual friction angle 42
Dilation angle 10
Peak tensile strength Variable (see Figure 3)
Residual tensile strength 0
Peak cohesive strength 2.5*tensile strength
Residual cohesive strength 0

An additional 8-m cubic sample with more persistent veins was constructed following the same procedure
outlined and superposing a discrete fracture network (DFN) in the sample by cutting the blocks intersected
by the DFN. The contacts formed by the introduction of the DFN were randomly populated with a strength
distribution from samples that failed on a structure for the weak, medium and strong subdomains
(Figure 3). Such sample was used to address the impact more persistent joints have in primary fragmentation.

Figure 3 DFN with fractures colored by area used to cut the SRM sample (left). Tensile strength distribution
of vein strength for the strong, medium and weak subdomains (center). Vertical cross-section of the
SRM sample with DFN showing the block contact tensile-strength distribution based on the medium
strength category and DFN populated with the medium vein strength distribution (right)

2.2 Emergent rock mass strength

Failure envelopes were estimated for the weak, medium and strong subdomains by subjecting samples to
a suite of virtual numerical triaxial tests under different confinement levels, as well as to direct tension
and uniaxial compression. The uniaxial compression test was performed by compressing the sample at a
constant strain rate until it failed. The model represents the compression of a rock mass sample between
frictionless roller boundary conditions which is relevant to spalling conditions. In an analogous way, a
direct tensile test was performed by pulling each sample apart at a constant rate in a quasi-static fashion
until it failed. The results of the UCS and direct tensile tests as well as the approximate failure envelopes for

Caving Mechanics

the different synthetic veined rock mass samples analyzed are shown in Table 2 and Figure 4 respectively.
Corresponding Hoek-Brown failure envelopes were estimated, resulting in an average mi = 13. This
information was used as input to model the different rock masses in a cave-scale FLAC3D model.

Table 2 Summary of UCS and Tensile Strength Results for the Different Veined Rock Mass Samples Tested

SRM Sample SRM Sample SRM Sample

Strength Distribution
Strong Medium Weak
Tensile Strength [MPa] 1.45 1.15 0.45
UCS [MPa] 20.5 16 8.7

Figure 4 Approximate failure envelopes derived from strength testing of veined rock mass samples with
different vein strength distributions

2.3 Primary fragmentation under caving-induced stress path

A series of SRM samples were tested under caving-induced stress paths to predict primary fragmentation.
The 3DEC model allows the blocks forming the sample to break at their subcontacts as a result of stress
concentrations, mimicking the initiation of cracks that can coalesce and/or propagate to fracture the rock
mass. This results in an emergent fragment size distribution.

The testing environment for fragmentation studies consisted of either an 8-m or 18-m veined rock mass
cube embedded in an elastic boundary as shown in Figure 5 (Garza-Cruz & Pierce 2014). The top and
sides of the synthetic rock mass cube were bonded to the elastic boundary. The elastic boundary facilitated
the application of stress boundary conditions to represent the induced stresses in the cave back and would
provide a way for stresses to shed upwards during failure (as would be expected in situ). Boundary
conditions were applied to the model as shown in Figure 5. The model was cycled to equilibrium, and
the vertical stress on the sample bottom-face relaxed in ten increments, while allowing the model to reach
equilibrium each time, simulating the upward advance of the cave back from below.

Caving 2014, Santiago, Chile

Figure 5 8-m edge-length cubic sample embedded in an elastic boundary for fragmentation testing. a) and c)
show the boundary conditions applied to the model. b) Elastic boundary made transparent to aid visualization

As the vertical stresses at the stope back decrease, tensile fractures develop sub-parallel to the face. A wide
range of SRM tests were conducted to examine the impact of cave back stress, block size, sample size
and vein persistence on predicted fragmentation. Figure 6 shows selected results for the case of 8-m cubic
samples of the weak, medium and strong domain (the different colors represent fragments, defined
as any collection of blocks that were still bonded together at one or more subcontacts). Figure 7 and Figure
8 show example results for the case of 18x18x18m strong sample and 8x8x8m medium sample with a
superposed DFN, respectively. The models revealed that for a given stress state, the volume of stope back
that undergoes spalling decreases as the rock mass strength increases.

In order to characterize the spalling behavior in a quantitative way, the cumulative percentage by volume
as a function of approximate fragment volume for all the strong, medium and weak samples were
calculated (results corresponding to the three cases presented in Figure 6 are shown in Figure 9 and Figure
10). The fragmentation test indicated that as the strength of the rock mass increases, so does the size of
fragments that can be expected to result from the rock mass failure and redistribution of stresses, which in
turn results in a larger size distribution. It is important to mention that the size of the smallest fragment that
can be created is limited by the size of the elemental tetrahedral forming the modeled rock mass.

Figure 9 also summarizes the 25th, 50th and 75th percentile fragment sizes by volume for the different
strength samples subject to a maximum cave back stress of 70 MPa. Figure 10 lists the percentage of the
modeled synthetic rock mass that underwent fragmentation, as well as the percentage of fragments with
volume in excess of 1.3 m. It is interesting to note that weaker rock would fragment finer, and in this
specific example would produce no large fragments, while stronger rock masses would tend to produce
more large fragments than medium strength rock masses. Therefore, the probability of generating large
fragments increases as the quality of the rock mass increases.

Caving Mechanics

Vertical plane through center of the Fragmentation at stope back

Weak SRM

Medium SRM

Strong SRM

Figure 6 Fragmentation of a weak (top), medium (center) and strong (bottom) SRM samples after the
vertical stress was fully relaxed from below. SRM sample dimensions: 8x8x8m. Views along a vertical section
(left) and looking up from below (right). Stress state tested: v = 23.4 MPa, H = 3*v = 70.2 MPa, h = 1.6*v
= 37.44 MPa

Caving 2014, Santiago, Chile

Vertical plane through center of the Fragmentation at stope back

Strong SRM

Figure 7 Fragmentation of a strong SRM sample after the vertical stress was fully relaxed from below. SRM
sample dimensions: 18x18x18m. Views along a vertical section (left) and looking up from below (right). Stress
state tested: v = 15.6 MPa, H = 3*v = 46.8 MPa, h = 1.6*v = 25 MPa.

Vertical plane through center of the Fragmentation at stope back

SRM with DFN

Figure 8 Fragmentation of a medium SRM samples with superposed DFN (see Figure 3) after the vertical
stress was fully relaxed from below. SRM sample dimensions: 8x8x8m. Views along a vertical section (left) and
looking up from below (right). Stress state tested: v = 15.6 MPa, H = 3*v = 46.8 MPa, h = 1.6*v = 25

Caving Mechanics

Edge Length Strong Medium Weak

25th percentile
0.04 0.03 0.02

50th percentile
0.12 0.07 0.03

75th percentile
0.27 0.17 0.07

10.2% 13.7% 23.4%

5.37% 0.0% 0.0%

Figure 9 SRM-derived primary fragmentation distribution for cave back stress of 70 MPa.

Fragment Diameter Distribution Assuming Disk Shape

(Cave Back Stress = 70 MPa)

Figure 10 SRM-derived primary fragmentation distribution for cave back stress of 70 MPa, assuming disk
shape for fragments

Caving 2014, Santiago, Chile

A wide range of SRM tests were conducted to examine the impact of cave back stress, block size, sample
size and vein persistence on predicted fragmentation. The results of all tests are summarized in Figure 11,
which relates the fragmentation to the strength of the domain and the induced stress at failure in the cave
back. In general, it can be seen that the weak and moderate domains generally produced fragments <1.3 m3
under both low and high stress, whereas the strong domain has the potential to produce large slabs (several
meters in diameter) when failing under low stress.

The SRM-derived fragmentation prediction chart was combined with the results of cave-scale modeling to
estimate how primary fragmentation might vary through the column as a result of differences in cave back
stresses alone for a given rock mass strength.

Figure 11 Primary fragmentation prediction chart the strong, medium and weak domains generated
from SRM sample testing

3 Numerical approach to analysis of caving

Five key geomechanical zones are associated with block and panel caving, as shown in the conceptual
model sketched in Figure 12. This builds on the conceptual model developed by Duplancic and Brady
(1999). The following are defining characteristics of each of the five zones.

Elastic zone Induced stresses may be high here, but are insufficient to induce measurable

Seismogenic zone Where microseismicity occurs within the jointed rock via joint slip and fracture
extension. This is commonly defined via an empirical damage threshold criterion that is a function of
the deviatoric stress and intact UCS [0.3 < (1 3) / (UCSintact) < 0.5 ].

Yielded zone Where the rock mass has disintegrated and lost all of its cohesive and/or tensile
strength, but has not moved a significant distance yet. The outer limit of this zone generally coincides

Caving Mechanics

with the fracture limit, where visible fractures are evident in intersected openings or on ground surface,
significant offset occurs in open boreholes and TDR cables break.

Air gap An air gap can exist if the overlying rock mass retains some level of cohesive and/or tensile
strength. As an air gap expands in size, the overlying rock mass may weaken further, causing advance
of the yielded zone and collapse into the air gap.

Mobilized zone This is where the disintegrated rock mass has moved a significant distance and is
starting to dilate and bulk as a result. The criterion depends on the scale of the cave and the modulus of
the rock mass; a total or vertical displacement of 1-3 m is generally employed.

The caving and stress-redistribution process inherently involves large deformations, shear along pre-
existing joints and bedding surfaces, fracturing of intact rock blocks and fragmentation of the rock mass
above the undercut level. Ideally, one would model this process using a discontinuum (e.g., 3DEC or
PFC3D) approach in which pre-existing fractures are explicitly represented in the model. However, the
computational size and time requirements to solve mine-scale problems currently still make it impossible
to attack a problem completely with the discontinuum approach.

Instead, an algorithm to simulate caving within the concept of a continuum-based model has been developed
over the past 15 years during the industry-funded International Caving Study (ICS I & II) and Mass Mining
Technology (MMT I & II) projects. The constitutive rock mass response required to represent caving (i.e.,
the rock mass yield, dilation and bulking) was developed using strain-softening material models, with
strain-dependent properties that are adjusted to reflect the dilation and bulking that accompany caving. The
caving algorithm as implemented in FLAC3D (Itasca, 2012) attempts to predict the limits of these zones as
a function of production from the cave. In addition to these cave limits, the results of cave-scale modeling
are used to derive estimates of the following:

Abutment and cave stresses;
Bulking factors, caving rate and breakthrough timing; and
Successful comparisons between predicted (via FLAC3D) and actual cave behavior have been achieved at
a number of operations worldwide (e.g., Northparkes E26, Ridgeway Deeps, Palabora, Grace Mine).

Figure 12 Conceptual model of caving

Caving 2014, Santiago, Chile

Within the caving model, a rigorous mass-balance routine is implemented to ensure that the tonnes-
based production schedule is represented accurately within the numerical model. Although the routine
is computationally intensive, and can lead to relatively long model run-times (two to three weeks), the
numerical approach is required to accurately capture the mechanisms of damage, yield, dilation and bulking
necessary to correctly reproduce the evolving cave shape and propagation rates in response to a specific
production schedule.

3.1 Numerical modelling of caving and primary fragmentation predictions based on SRM

A series of heterogeneous models were constructed in FLAC3D using the strength of domains derived from
SRM testing (UCS = 8.7, 16 and 20.5 MPa for weak, medium and strong domains, respectively and mi=13)
along with brittleness derived from back-analysis of stope overbreak (not presented here). The modelled
undercut layout consists of 350 m N-S and 140 m E-W, with a hydraulic radius= 50 m. The in-situ stress
state used is shown in Figure 13 (H applied at 058).

Figure 13 In-situ stress regime used in the FLAC3D cave model

The production process was simulated assuming weak, medium and strong rock everywhere. The
model results suggested that it would be possible to sustain caving all the way to ground surface. Figure 14
shows the stress redistribution around the mobilized and yielded zones as well as the predicted shape of the
inclined cave after two years of full production (strong case).

Figure 14 Predicted shape of the inclined cave two years after start of full production. Contours show the
major principal stress redistributing around the mobilized and yielded zones

Caving Mechanics

As the cave is propagated, the model records the induced stresses at failure. Such stresses at failure for the
case of strong domain after caving to ground surface took place are shown on the left side of Figure 15.
The SRM-derived fragmentation prediction chart (Figure 11) was combined with the results of cave-scale
modeling to relate induced stresses at failure to percentage of large fragments produced and estimate how
primary fragmentation might vary through the column as a result of differences in cave back stresses alone
(Figure 15).

Figure 15 Caving-induced stresses at failure assuming strong rock everywhere (left). Associated primary
fragmentation as a percentage of fragments larger than 1.3 m3 for given stresses at failure (right)

At the beginning of cave life, brittle spalling is promoted by the large hydraulic radius, which limits
arching and the associated buildup of confinement in the cave back, leading to a higher percentage of large
fragments as derived from the SRM fragmentation analysis (~50% of fragments could be >1.3m3). As
caving progresses, finer fragmentation should be expected at the column mid-height, where the cave back
stresses are highest (~30% of fragments could be >1.3m3). As the cave reaches ground surface, stresses are
low and fragmentation relies more on gravitational pull, leading to a higher percentage of large fragments
(~60% of fragments could be >1.3m3).

4 Conclusions
A methodology has been developed using 3DEC and FLAC3D to predict primary fragmentation as a
function of rock mass strength and stresses induced at failure in a cave mine. SRM samples of heavily
veined massive rock masses can be constructed using 3DEC by assembling a collection of tetrahedral
blocks bonded at their contacts; while contact strength heterogeneity is introduced based on field data.
Emergent SRM strength is used to inform a FLAC3D caving model. SRM samples were also tested under
cave-like stress paths to predict primary fragmentation. The SRM-derived fragmentation prediction chart
was combined with the results of cave-scale modeling in FLAC3D to relate induced stresses at failure
to percentage of large fragments produced and estimate how primary fragmentation might spatially vary
through the column. The results suggest that primary fragmentation could be quite coarse where medium
and strong domains cave under low stresses, which is expected at the bottom and top of the ore column.

This methodology can be applied to models where the spatial variation of geologic domains is known,
to predict primary fragmentation as a function of changes in rock mass strength and induced stresses.

Caving 2014, Santiago, Chile

Furthermore, the estimated primary fragmentation size distribution can be used as an input to predict cave
drawdown and secondary fragmentation using REBOP (Fuenzalida et al, 2014).

Further studies are warranted to refine the strength characterization of rock masses using SRM as well as
their associated fragmentation under caving stress paths.

Duplancic, P, & Brady, BH 1999, Characterization of caving mechanisms by analysis of seismicity and rock
stress, Proceedings 9th International Congress on Rock Mechanics (Paris), vol. 2, pp. 1049-
1053. Balkema, Rotterdam.

Fuenzalida, M, Garza-Cruz, TV, Pierce, M & Andrieux, P 2014, Application of a methodology for
secondary fragmentation prediction in cave mines, Proceedings 3rd International Symposium
on Block and Sublevel Caving, Santiago, Chile.

Garza-Cruz, TV, & Pierce, M 2014, A 3DEC Model for Heavily Veined Massive Rock Masses, Proceedings
48th US Rock Mechanics / Geomechanics Symposium. Minneapolis, USA.

Itasca Consulting Group, Inc. 2013, 3DEC Three-Dimensional Distinct Element Code, Ver. 5.0 Users
Manual, Minneapolis: Itasca.

Itasca Consulting Group, Inc. 2012, FLAC3D Fast Lagrangian Analysis of Continua in 3 Dimensions,
Ver. 5.0 Users Manual, Minneapolis: Itasca.

Caving Mechanics

ALCODER, challeges of paradigms in caving methods

Gl Krstulovic Geomecnica Ltda., Chile
GA Bagioli Tetra Tech Metlica, Chile

More than 90% of the Chilean metal production from underground mining sites is extracted through collapse
mining methods. The experience gained in these mining sites has originated a series of assumptions, which
largely respond to popular belief and have no sufficient analytical support to be considered as true to the
behavior of in-situ rock.

The reformulation of old concepts in classic rock mechanics has allowed establishing new criteria in
order to explain the behavior of rocks excavated thus, which include the deterioration criterion as
an alternative to the rupture criterion; this allows reviewing the most frequent paradigms in mining
operations by caving.

In our case, the Janb-Kulhawi-Krstulovic concepts, which are alternative to the traditional Mohr-
Coulomb-Hoek, allow anticipating the orientation that the collapsing rock will adopt, including the
resulting seismicity, among other things. It may be assumed from the foregoing that this review concludes
on geometric configurations that favor collapse, including the geometry of the anomaly that is currently
known as pre-conditioning of rock, in order to also favor collapse.

The analytic formulation of this deterioration criterion has been incorporated in the ALCODER computer
simulator. The necessary input data for these processes require identifying the behavior of the deformation
module of the rock, and the variation with their surrounding tectonic confinement. Complementing the
foregoing, the maximum deformation energy (DE) tolerated by such rock based on lab tests, constitutes a
comparative pattern for establishing seismicity and potential rockburst in-situ.

1 Brief introduction to ALCODER

More than 90% of the Chilean metal production from underground mining sites is extracted by collapse
mining methods. ALCODER is a Fortran simulator by finite state-of-the-art elements, expressly designed
to adequately respond to mining problems in relation with collapsing rock.

The ALCODER algorithm is based on original computer programs from Utku (1968) and Kulhawy (1972).
Both algorithm well validated as per references:

ELAS-A., Senol Utku et al, 1968.

Kulhawy, Fred H., 1972.
These original concepts are modified according to the references indicated below:
Krstulovic G.,2004
Bagioli G., Krstulovic G., 2008.
The recent uses of ALCODER in Caving and SLS Mass Blast are the following:

Caving 2014, Santiago, Chile

CODELCO Salvador, Geomecnica Ltda. / Metlica Ing., 2000, Panel Caving. Basic Eng. for
Inca Oeste.

Vasante Metals Inc. Geomecnica Ltda. / Metlica Ing., 2004, SLS Mass Blast (Brasil).
Free Port Mc Moran Santos, Alcaparrosa and Candelaria Ore Deposits, Geomecnica Ltda.,
2000-2009, Internal project reports and design results for Mass Blast in SLS stoping.

CAP Romeral, Geomecnica Ltda. / Metlica Ing., 2010, SLC Project Under Current Open Pit.
Yamana Gold Inc., Geomecnica Ltda. / Metlica Ing., 2010, Mining Projects for QDDL
(Argentina), Jernimo (Chile) Ore Deposits for Mass Blast in SLS Stoping.

IM2, Geomecnica Ltda., 2010-2011, Internal Consulting Reports (In House Consultant) for
Rock Pre Conditioning in Caving.

CODELCO Chuquicamata, Geomecnica Ltda., 2011-2012, Caving Propagation Estimation


CODELCO Andina, Geomecnica Ltda. / SKM Ing, 2012-2013, Support Analysis for Haulage
III Excavation Alternative.

CODELCO El Teniente, Geomecnica Ltda. 2013-2014. ALCODER Validation of 7 RockBurst

at Pilar Norte Orebody

More information on ALCODER results for full scale trial can be obtained from Freeport McMoran for
MassBlast in SLS stoping , and from CODELCO El Teniente for RockBurst.

2 The deterioration in rock criterion that rules the ALCODER algorithm

Since its beginnings, Rock Mechanics for Geomechanics in Mining has inherited, from its civil pair, and
according to classic mechanics, the Mohr-Coulomb-Navier Rock Rupture Criterion concept. This concept
was modified some years ago by an empirical equivalent: the Hoek Rupture Criterion. In both cases, the
Criterion aims to explain the conditions that rule the rupture phenomenon in rock. As a result, the software
that are commercialized for rock stability evaluations during mining excavations invariably contain these
rupture criteria to explain the behavior of such rock. The Output of these software invariably provides:

Safety conditions of the remaining rock after mining excavations through a Safety Factor.
Deformation conditions of this same remaining rock.
Both outputs require interpretation and validation, which are not always sufficiently achieved for the
purposes of the study under evaluation.

Alternatively, in mining excavations (unlike civil excavations), one must coexist with deteriorated rock that
are still capable of supporting the mining building. Consequently, the expected deterioration degree in
this mining excavation becomes a relevant factor for the design in Caving.

Here we formulate that deterioration in rock is directly associated to the deformation that these materials
could suffer in a confinement / deconfinement process resulting from mining excavations. Rock taken from
virgin (raw) conditions to deconfined conditions (near an excavated wall) suffer micro fractures which
compensate (in volume) the deformation of such walls towards the excavated space. The occurrence of
these micro fractures implies a reduction of the competence capacities of such rock, i.e., a reduction that
responds to a deterioration of the competence indexes that define the quality of this rock.

Caving Mechanics

The competence index that is well detected by this deterioration, is the Deformation Module (E). It is
empirically proven that rock with lower (E) have lower competence qualities, i.e., there are widely accepted
empirical formulas that relate (E) / RMR, (E) / GSI, (E) / RQD, where: RMR, GSI, RQD are quality indexes
collected through geotechnics in-situ.

In other words, (E) is a good indicator of in-situ rock quality i.e., (E) can suffer modifications according
to the confinement / deconfinement process resulting from mining excavations in-situ according to the
tectonics of the area. The law that rules this variation of (E) in rock was initially formulated by (Kulhawi

The ALCODER incorporates this Law of Variations of (E) under confinement, thus concluding the resulting
new (E) in the remaining rock at the excavation. The variations of (E) are transferred to variations of RMR,
GSI, RQD, as appropriate, according to the aforementioned empirical formulas.

Therefore, the ALCODER Output is user-friendly in RMR, GSI, or RQD indexes, which allow configuring
the deterioration experienced by rock undergoing an excavation process.

3 Collective imagination myths regarding the caving process

Mining by caving has the following unchallenged paradigms:

1. After causing collapse over the Undercut, Caving progresses according to dome geometry.
2. The advance perimeter or caving face in Panel Caving must be concave towards the collapse.
3. For Caving simulator effects, Modules (E) in rock can be assumed as invariable.
4. Pre-Conditioning (PA) of in-situ rock always helps to accelerate collapse.
5. The caving policy from extraction points can be made independently from the collapse

6. The resulting granulometry in extraction points is independent from the collapse process.
7. Computer simulators cannot detect faults that have not been pre-established.
8. Rock Burst can only be anticipated with seismic records.
In the following Sections, we present results of the ALCODER simulator with the deterioration criterion in
rock, which challenge the accuracy of these myths.

3.1 Myth 1

After causing collapse over the Undercut, Caving progresses according to the dome geometry. This as
schematically described in Figure 1.

INCORRECT. The geometric configuration of collapse in progress depends on the quality of in-situ rock
and the tectonics of the area. Under normal conditions, collapse tends to advance towards a geological
discontinuity or slopes in mountain topography.

Figure 2 shows the deviation of the advance in collapse. Example of an ALCODER simulator Output after
7 iterations of collapse in Caving under an Open Pit topography and without an express extraction policy.
In other words, retraction according to spontaneous collapse. The black line in figure 2, is the main fault.
Collapse Criterion: Spontaneous by Hydraulic Radius (HR) in rock with MRMR index lower than 40.

Caving 2014, Santiago, Chile

Figure 1 Collective Imagination Regarding Caving Process

Figure 2 Deviation of the Advance in Collapse

3.2 Myth 2

The caving face perimeter in Panel Caving must be concave towards the collapse.

INCORRECT. The geometric configuration of the collapse cavity in progress / in-situ rock outlined in plant
projection and in vertical projection makes it look like a vault which auto-supports itself with the concavity
of its walls. To facilitate collapse, both vertical and horizontal projections must be as straight as possible,
so that the abutment effort is minimized with the straight faces.

Figure 3 shows an isometric view of ALCODER simulator for Caving.

Figure 4 shows ALCODER result of the variation in the abutment Stress according to the geometry of the
face in collapse. In other words, the closer the face gets to the vertical, the abutment Stress will diminish.

Caving Mechanics

Figure 3 Isometric View of ALCODER Simulator Devoted to Results on Figure 4.

Figure 4 ALCODER Result of the Variation in the Abutment Stress

3.3 Myth 3

For Caving simulation, Modules (E) in rock can be assumed as invariable.

INCORRECT. Module (E) varies according to the confinement / deconfinement of the location. Figure 5
shows the variation of (E) according to lab tests for the case of Porphyry in Chuquicamata.

The Law that rules the variation of (E) according to (Kulhawi 1972) is function of the experimental constants
(K) and (n), where Pa is the atmospheric pressure and Sigma 3, the lower confinement in-situ.

Caving 2014, Santiago, Chile

Figure 5 Variation of (E) According to Lab Test on Porphyry Rock

= 3
E = Young`s Elastic Modulus
Pa = Atmospheric pressure expressed in the same units as E
3= Minimum principal stress
n = Modulus Exponent , K = Modulus Number
K, n= are pure numbers

Figure 6 shows, according to (Barragan & Krstulovic 2013), a recent compilation with the empirical relation
(K) / (n). It is estimated that in rock with RMR under 60, the mistake of not considering the variation of (E)
can seriously affect the Simulator results.

Figure 6 Empirical Relation (K) / (n)

Caving Mechanics

3.4 Myth 4

Pre-Conditioning (PA) of in-situ rock always helps to accelerate collapse.

INCORRECT. The (PA) aims to cause a quality anomaly in the in-situ rock. In such condition, in order to
achieve the acceleration of collapse, the (PA) must satisfy a requirement of adequate geometric shape and
intensity in the reduction of the quality of in-situ rock.

Figure 7 shows an ALCODER example of a (PA) geometry 20% - 40% deterioration in (E), which
successfully accelerates (compared with Figure 2) Caving under mountain topography. The black line is
the main fault.

Figure 7 Example of (Pa) Geometry Deterioration in (E)

Figure 8 shows an ALCODER example of an insufficient application of (PA) with deterioration in 20%
in (E) to accelerate the collapse process in a tectonic environment. This Figure 8 includes (%) of the
collapsed material according to RMR index. In this case, the quadrant with 20% deterioration in (E) does
not accelerate rock collapse in the object sector of (PA).

Figure 8. ALCODER Example of Insufficient (PA)

Caving 2014, Santiago, Chile

3.5 Myth 5

The mining production policy from extraction points can be made independently from the caving process.

INCORRECT. Ideally, mining from the extraction points should be made with previous knowledge of the
way in which Caving collapses spontaneously.

If a mining policy that coincides with the spontaneous collapse of Caving is not maintained, this could
cause extreme conditions at the production level:

If mining happens faster than the collapsed material in Caving, this could cause a space between
the ground in-situ and the already collapsed material, originating an abutment condition, or else,
the detachment of wedges as in-situ rock.

If mining happens slower than the collapsed material in Caving, the non extracted column
is compacted and serves as temporal support for Caving and causes a collapse option at the
production level.

Figures 9 and 10 are records of collapse due to inconsistency between mining / Caving according
to the ALCODER Output. The dark bodies located in the collapsed material, are wedges incorporated
spontaneously in the Caving. In other words, the ALCODER allows forecasting the spontaneous collapse
process, and thus allows adjusting the mining policy in-situ.

Figure 9 Collapse at the Production Level Figure 10 Wedges Incorporated in the Caving

3.6 Myth 6
The resulting granulometry at the extraction points is independent from the collapse process.

INCORRECT. During the collapse process, the confinement / deconfinement conditions in the in-situ rock
cause spontaneous collapse of various granulometries. Thus, it is incorrect to assume that initially there is a
coarse granulometry that subsequently is reduced along its transit towards the extraction point.

The spontaneous granulometry produced by Caving depends on the rock quality, tectonic stresses and the
geometry of the spontaneous collapse, which could include extreme cases:

Caving Mechanics

Abundant spontaneous coarse granulometry at the beginning that subsequently limits itself to
spontaneous collapse of fine granulometry.

Alternatively, only fine granulometry at the beginning, and/or subsequently, large boulders due
to collapse in higher elevations.

Figure 11 describes an ALCODER Output with granulometries differentiated by colors as caving outcrops
under an open pit slope. ). Note that fragmentation is result of different RMR/ E values. Validation by
correlation RQD/E/RMR /Blok Size is in progress.

Figure 11 ALCODER Output with Differentiated Granulometries

Figure 12 describes the configuration of the collapsed material by Caving stages according to ALCODER
Output in 14 extraction columns over the undercut.

Figure 12 Extraction Columns Over the Undercut After 6 Caving Stages

3.7 Myth 7

Simulators cannot detect faults that have not been pre-established in the simulation model.

INCORRECT.- ALCODER detects the spontaneous displacement in rock bodies as Caving progresses, and
sheds light on subsidence Faults/Cracks or Faults/Cracks in the mining infrastructure near the collapse.

Caving 2014, Santiago, Chile

Figure 13 schematically shows an SLC project under an exhausted Open Pit. In Figure 14, the ALCODER
simulator identifies cracks in the SLC mining sequence. These cracks are from subsidence in overturn, and
cracks in ramp infrastructure of the SLC as mining operations progress.

Figura 13 SLC Project Under Exhausted Open Pit

Figure 14 Cracks in the SLC Sequence After ALCODER Out Put

3.8 Myth 8

Rock Burst can only be anticipated with seismic records.

INCORRECT. There are no bibliographic records regarding Rock Burst forecasts in mining sites. Although
more than 40 years have passed since micro seismic hearing systems were implemented in mining
operations, the Rock Burst issue has still not been solved.

According to Classic Mechanics, Rock Burst occurs in rock when the Deformation Energy (DE) per unit
volume exceeds the tension resistance in the rock. To determine this (DE) per unit volume, rock is assumed
as a block with differential dimensions, which undergoes the action of the main normal stresses (S). (ED)
is the work executed by these stresses when deforming the cube one magnitude (dl) (Obert L. & Duvall W.

Caving Mechanics

In other words, Magnitude (ED) is equal to the work necessary to deform the rock cube in (dl). Rock Burst
occurs when the work exceeds the tension resistance of the rock (Example: Resistance at Atmospheric
Confinement). The (ED) can be estimated according to formulas (1) and (2).



E =Young`s Elastic Modulus

= Poisson`s Ratio
= 1, 2, 3 Principal Confining Stresses

Figure 15 shows an ALCODER simulator with Output in collapse after 10 iterations. For each of the
iterations, the ALCODER identifies the (ED) that exceeds the maximum value accepted by this rock
according to (ED) verifications in lab. In this case, points 1 to 5 present numbers for (ED) Rock Burst.

Figure 15 ALCODER Out-Put with Maximum (ED)

In other words, the ALCODER can anticipate the opportunity and the place where Rock Burst would occur
during the caving process. Magnitude of Rock Burst anticipated by ALCODER can be estimated from
Krstulovic (1977).

4 Conclusions
The Rock Deterioration Criterion based on the (E) Module deformation index suggested here has analytical
grounds in classic mechanics, and empirical verifications in Figure 6. Taken to applications of the mining
business through the ALCODER algorithm, this criterion is adequate for addressing a series of typical
caving issues, i.e., problems that range from abutment stress, collapse in production levels, to Rock Burst.

Caving 2014, Santiago, Chile

The authors thank Tetra Tech Metlica and Geomecnica Ltda. for the considerations granted to support
this document.

Bagioli, G, Krstulovic G 2008, An ALCODER for Computer Monitoring of Slopes Stability During WTI
Program in Open Pit Mining, ISRM Congress, Lima, Per.

Barragan, JL, Krstulovic G 2013, Lab. data compilation from different authors.

Duvall, W & Obert, L 1967, Rock Mechanics and the Design of Structures in Rock, John Wiley & Sons,

Fred, H. Kulhawy, 1972, Finite Element Modeling Techniques for Underground Opening in Rock. Contract
N H0210023. Advanced Research Projects Agency, Washington USA.

Krstulovic, G 1977, Mtodos y Tcnicas Micro Ssmicas en la Evaluacin de Estabilidad Dinmica de

Macizos Rocosos. RI-77-1 Centro de Investigaciones Minero y Matalurgica CIMM- Chile

Krstulovic, G 2004, ALCODER A New Method for Evaluating Stability of Rock Excavations, Mass Mine

Senol Utku et al. 1968, General Purpose Computer Program for the Equilibrium Problems of Linear
Structures. TR 32-1240. Jet Propulsion Lab. CALTEC Pasadena, California.

Caving Mechanics

Characterization and synthetic simulations to

determine rock mass behaviour at the El Teniente
Mine, Chile. Part I
A Brzovic Codelco, Chile
P Schachter Codelco, Chile
C de los Santos Codelco, Chile
JA Vallejos, University of Chile, Chile
D Mas Ivars Itasca Consultans AB, Sweden

A comprehensive geotechnical characterization has been undertaken at the El Teniente mine to describe
and determine the rock mass behaviour strength properties of the primary copper ore. This type of rock can
be considered as a heavily veined massive and unfractured rock mass. This compressive work is focused
sequentially on; 1) an intensive structural data collection campaign from several oriented core placed
within main geotechnical units, 2) discrete fracture network modelling, 3) laboratory testing of the intact
rock and veins materials combined with scaling procedures, and 4) application of the Synthetic Rock Mass
(SRM) approach to study the strength and deformation behaviour of the main geotechnical units.

The Synthetic Rock Mass (SRM) modelling approach, based on particle mechanics, has been developed
to simulate the mechanical behavior of jointed rock mass. This technique uses the bonded particle model
for rock to represent intact material and the smooth-joint contact model (SJM) to represent the in situ
joint network. The macroscopic behavior of an SRM sample depends on both the creation of new fractures
through intact material and slip/opening of pre-existing joints. SRM samples containing thousands of non-
persistent joints can be submitted to standard laboratory tests (UCS, triaxial loading, and direct tension
tests) or tested under a non-trivial stress path representative of the stresses induced during the engineering
activity under study.

This paper describes the first part of the study, with focus on structural data collection campaign (points 1)
and laboratory testing (point 3).

1 Introduction
The primary copper ore at the El Teniente mine is described as very competent and massive, due to it exhibits
a brittle behavior, often violent failure under high stress conditions (Rojas et al 2001). This description is
coherent with the geological description of the rock mass, which does not have discontinuities match as the
definition provided by International Society of Rock Mechanics (ISRM, 1981). Only faults can be classified
as discontinuities, but they are widely spaced. The primary copper ore has a high frequency of veins, where
the cooper mineralization is hosted, these vein network structures are known as stockwork (Figure 1). Soft
veins containing weak minerals as infill (chalcopyrite and anhydrite) control the disassembling of the rock
mass during caving (Brzovic & Villaescusa 2007; Brzovic 2011).

Nowadays, there are two traditional methods used to estimate the strength of the rock mass: 1) determination
of the strength envelope of the rock mass using scaling parameters from laboratory tests, for example Hoek-
Browns failure criterion and 2) using numerical modeling based on back analysis of previous experiences
of failure observed and measured in mining or civil excavations.

Caving 2014, Santiago, Chile

Figure 1 Panel caving method currently used at the El Teniente Mine (a), Intense vein network stockwork at
a development ahead of the cave front (b), and Weak Veins as faces of caved rock blocks (c) (modified from
Brzovic & Villaescusa 1997)

The primary copper ore of El Teniente mine practically has no fractures or joints; therefore it is difficult
to determine a rating able to scale laboratory data, such as RMR (Laubscher 1977) or GSI (Hoek 1994).
Therefore, the way number 1 cannot be used properly unless the RMRs or GSIs input parameters are
manipulated or adjusted in order to obtain reasonable results. The second option requires, in general, a good
characterization of a previous failure event in the rock mass, not always available.

In recent years, new techniques of numerical modeling associated with the concept called Synthetic Rock
Mass have been developed (PFC3D, ELFEN, Abaqus) aiming to capture the real behaviour of the rock
mass. Those methodologies are the third way to estimate the strength of the rock mass, but they are still in

This paper is composed by two parts. It is aimed to implement the concept of SRM developed by
Itasca (Pierce et al. 2007; Mas Ivars et al. 2011). The methodology is divided in making a geotechnical
characterization (mapping and laboratory tests), developing scaling laws and applying the SRM approach.
This part describes the results of comprehensive effort to characterize the El Teniente rock masses, which
include; core logging, field work, structural data analysis, Discrete Fracture Network (DFN) modeling, and
laboratory testing. In a following paper (Vallejos et al. 2014), the strength and deformation behaviour of
four rock mass domains from the El Teniente mine are studied.

2 Intensive structural data collection

An intensive structural data collection campaign from several oriented core placed within main geotechnical
units were undertaken as part of this study. In order to avoid orientation bias during data collection at each
location or geotechnical unit, three oriented cores were drilled in three almost orthogonal orientations
to each other. Those groups of cores, 9 in total, are called triada and represent more than 2000 meters
of structural mapping (between 240 and 300 meters each triada). Geological and full structural core
logging (similar to scanline mapping) was undertaken to determine the intensity of weak veins at each

Caving Mechanics

location. Quantitative mineralogical assemblage, orientation and geometric features of veins were the main
characteristics obtained during core logging.

Weak veins and fault lineal intensity then were determined for each geotechnical unit, the mean value
measured of the lineal frequency (P10 according to Dershowitz & Einstein 1988) at each triada location,
range from 1m-1 to 12m-1 (excluding the low grade central brecha braden). On the other hand, fault lineal
intensity measured on the same cores, range from P10 0.05 m-1 to P10 0.4 m-1. Those intensity values
agree with the geological description of primary rock mass by Brzovic & Villescusa (2007).

Additional core logging (more than 10,000m from un-oriented cores) and historical structural drive mapping
were used to generate a new geotechnical zoning of primary rock mass, which agree with the alteration
and genetic geological model of the El Teniente porphyry copper. The new geotechnical zone are shown in
Figure 2, detailed sequence of zoning building was presented by Brzovic & Schachter (2013).

Figure 2 Plan view of the geotechnical model at the El Teniente mine (level 2121 and 2210) based on weak
veins intensity measured in oriented cores

3 Discrete fracture network modelling

Structural data analysis were also undertaken to build Discrete Fracture Network (DFN) as the best way to
represent the rock structure of the stockwork veins nature (Figure 3). The methodology followed to build
a DFN from each geotechnical unit is fully described in Brzovic & Herrera (2011). Discontinuity size, for
small scale geological structures were obtained from scanline data collected in mine drives, and for faults,
from general plan view of fault interpretations. Based on the structural data analysis, it was possible to
obtain the weak veins volumetric intensity P32 (Dershowitz & Einstein, 1988) of each geotechnical unit.
DFN in Figure 3, were built using commercial FracMan software, which allowed to readily determining in
Situ Fragmentation of primary rock mass. The P32 of weak veins determined at the El Teniente mine range
from 2m2/m3 to 15m2/m3, which represent a small percentage of the stockwork veins of the primary rock
mass. On the other hand, fault intensity P32 were determined from 0.15m2/m3 to 0.40m2/m3.

Caving 2014, Santiago, Chile

Figure 3 DFN model as the way to better represent the nature of the stockwork veins at the El Teniente mine
(Adapted from Brzovic & Herrera 2011)

4 Laboratory testing results

Rock and veins samples of primary ore were tested in several laboratories to determine strength properties
of intact rock and geological structures. Laboratory testing included; UCS and triaxial test of the standard
rock sample size, UCS of large rock specimen to develop scaling law relationship for main rock types,
direct tensile and shear test of all vein types. That information was complemented with data analysis of the
historical lab information from the mine site.

4.1 Intact rock properties

In general, at the El Teniente mine, there are two main factors that control the strength of a laboratory
specimen that represent the intact rock material of primary rock mass: the proper intact rock material and
the veins features contained on the small rock sample. This aspect describe a fundamental characteristic of
the primary ore, stockwork veins intensity are so high that even a small core sample contain several veins
within. Based on that fact, Marambio et al. (2000) suggested a classification of rock sample failure (or
failure mode) during testing, which is summarized in Figure 4. The same figure also presents the standard
values of UCS and triaxial testing from main geological units of the El Teniente mine. In addition, Figure
5 presents historical values of Young Modulus.

Caving Mechanics

Figure 4 Historical values of Uniaxial and Triaxial tests undertaken for main rock types at the El Teniente
mine, which were classified according the failure mode suggested by Marambio et al (2000)

Figure 5 Historical values of Young Modulus and UCS to different rock sample sizes undertaken to main rock
types at the El Teniente mine

Caving 2014, Santiago, Chile

Figure 5 also presents UCS values for different rock sample sizes undertaken to find out scaling law
relationship. In this figure only failure mode A, B and C were included. As would be expected, large
variability of intact rock strength properties are found out due the vein influence during rock sample

4.2 Vein strength properties

Standard direct shear tests non-conventional direct tensile tests were undertaken for several vein types
in order to obtain strength properties of the El Teniente geological structures. Those tests included full
geological and geometrical description of each veins typed tested. More than 40 samples were undertaken
for direct shears test, most of them at the SP Technical Research Institute of Sweden, and more than 50
samples were undertaken for direct tensile test at both the SP Technical Research Institute of Sweden and
the IDIEM Laboratory of University of Chile. Historical information from the mine site was also included
during data analysis. Description of the methodologies used can be seen in: De los Santos and Brzovic
(2013), Baraona (2012), and De los Santos (2011). Figure 6 present some shear and tensile strength values
of the El Teniente veins correlated to the main mineralogical assemblage as infill.

Figure 6 Shear and tensile strength values of veins at the El Teniente mine (adapted from: De los Santos &
Brzovic 2013; Baraona 2012; De los Santos 2011)

Caving Mechanics

5 Conclusions
From the comprehensive geotechnical characterization undertaken at the El Teniente mine in order to
describe and determine the rock mass strength properties of the primary copper ore, it was found that:

A new geotechnical zonation was achieved based on the concept of weak veins occurrence. Average
lineal vein intensity (P10) of main geotechnical unit of primary rock mass range from 1m-1 to 12m-
1. Average fault lineal intensity also range from P10 0.05m-1 to P10 0.4m-1. These new concept/
zonation are in more agreement with rock mass behavior at the mine site than previous, for instance
Fragmentation and Seismicity.

Structural data analysis allowed building Discrete Fracture Network model that honored core logging
of main geotechnical units at the mine site. DFN output provide veins volumetric intensity of the
primary rock mass (P32), which range from 2m2/m3 to 15m2/m3. Fault volumetric intensity P32 range
0.15m2/m3 to 0.40m2/m3.

Intensive laboratory testing and historical mine site data were used to determine strength properties of
both intact rock and geological structures of the El Teniente mine. Data analysis also included scaling
law relationship. All those basic information gathered were used to apply the Synthetic Rock Mass
approach to study the strength and deformation behaviour of primary rock mass.

The authors acknowledge to The El Teniente Division of Codelco-Chile for their permission to publish
the data and for supporting this work. This study was commanded by API T10E202 of Codelco-Chile.
FONDECYT Initiation Grant #11110187 also financed this study.

Baraona, K 2014, Comportamiento de Vetillas sometidas a Ensayos de Traccin Directa, mina El Teniente,
Internal report of the Superintendence Geology, CODELCO-Chile El Teniente Division, API
T10E202. [in Spanish].

Brzovic, A & Villaescusa, E 2007, Rock mass characterization and assessment of block-forming
geological discontinuities during caving of primary copper ore at the El Teniente mine, Chile,
International Journal of Rock Mechanics and Mining Sciences, vol. 44, pp. 565-583.

Brzovic, A 2009, Rock mass Strength and Seismicity during Caving Propagation at the El Teniente Mine,
Chile , In Proceedings of 7th International Symposium on Rockburst and Seismicity in Mines
(RaSiM07). Tang, C.A. editor. Dalian University. (2) 838-52.

Brzovic, A & Herrera, S 2011, Assessing Geological Vein Size and Intensity using Discrete Fracture
Network Modeling at the El Teniente Mine, Chile, InProceedings of the 45th US Rock
Mechanics / Geomechanics, ARMA Symposium, San Francisco, EEUU. 11-252.

Brzovic, A & Schachter, P 2013, Rock Mass Geotechnical Characterization based on the Weak Stockwork
Veins at the El Teniente Mine, Chile, In Proceedings of 3th International Seminary of Geology
for the Mining Industry, GEOMIN. Santiago, Chile.

De los Santos, C 2011, Efecto de la Mineraloga, Alteracin, y Geometra en la Resistencia Mecnica de

las Vetillas, Mina El Teniente, Regin del Libertador Bernardo OHiggins, Chile. Memoria
Para Optar Al Ttulo De Gelogo. Universidad de Concepcion. [in Spanish]

Caving 2014, Santiago, Chile

De Los Santos, C & Brzovic, A 2013, Geotechnical Properties on Cemented and Healed Stockwork Veins
at the El Teniente mine, Chile, In Proceedings of 3th International Seminary of Geology for
the Mining Industry, GEOMIN. Santiago, Chile.

Dershowitz, W & Einstein, H 1988, Characterizing rock joint geometry with joint system models, Rock
Mechanics and Rock Engineering, vol. 21, pp. 21-51.

Hoek, E 1994, Strength of rock and rock masses, ISRM News Journal 2, pp. 416.

Hoek, E & Brown, E 1988, The HoekBrown failure criterion a 1988 update, in: Proceedings of the 15th
Canadian Rock Mechanics Symposium, pp. 3138.

ISRM 1981, Suggested methods for the quantitative description of discontinuities in rock masses in Rock
characterization, testing and monitoring, ISRM Suggested methods, (edited by ET Brown),
Pergamon Press, pp. 3-52.

Laubscher, D 1977, Geomechanics classification of jointed rock masses mining applications, Trans.
Inst. Min. Metall., 86, A1-A8.

Mas Ivars, D, Pierce, M, Darcel, C, Reyes-Montes, J, Potyondy, D, Young, P & Cundall, P 2011, The
Synthetic Rock Mass approach for jointed rock mass modeling, International Journal of Rock
Mechanics and Mining Sciences, vol. 48, pp. 219244.

Marambio, F, Pereira, J & Russo, A 1999, Comportamiento Estudio Propiedades Geotcnicas Proyecto
Pipa Norte, Internal report SGL-280/1999 of the Superintendence Geology, CODELCO-
Chile El Teniente Division [in Spanish].

Pierce, M, Mas Ivars, D, Cundall, P & Potyondy, D 2007, A synthetic rock mass model for jointed rock,
In Proceedings of the 1st Canada-US Rock Mechanics Symposium, Vancouver, Canada, vol.
1, pp. 341-349.

Rojas, E, Cavieres, P, Dunlop, R, & Gaete, S 2000, Control of Induced Seismicity at the El Teniente Mine,
Codelco Chile, In Proceeding Massmin, Chitombo, G, editor, Brisbane, Australia, AusIMM,

Vallejos, J, Brzovic, A, Lopez, C, Bouzeran, L & Mas Ivars, D 2013, Application of the Synthetic Rock Mass
approach to characterize rock mass behavior at the El Teniente Mine, Chile, Continuum and
Distinct Element Numerical Modeling in Geomechanics: Proceedings of the 3rd International
FLAC / DEM Symposium, Hangzhou, China, paper: 07-02.

Vallejos J, Suzuki, K, Brzovic, A & Mas Ivars, D 2014, Characterization and Synthetic Simulations to
Determine Rock Mass Behaviour at the El Teniente Mine, Chile. Part II, In: Proceedings of
the 3rd International Symposium on Block and Sublevel Caving, Santiago, Chile.

Caving Mechanics

Characterization and synthetic simulations to

determine rock mass behaviour at the El Teniente
mine, Chile. Part II
JA Vallejos University of Chile, Chile
K Suzuki University of Chile, Chile
A Brzovic Codelco Chile, Chile
D Mas Ivars Itasca Consultans AB, Sweden

A comprehensive geotechnical characterization has been undertaken at the El Teniente mine to describe
and determine the rock mass behaviour strength properties of the primary copper ore. This type of rock can
be considered as a heavily veined massive and unfractured rock mass. This comprehensive work is focused
sequentially on; 1) an intensive structural data collection campaign from several oriented core placed
within main geotechnical units, 2) discrete fracture network modelling, 3) laboratory testing of the intact
rock and veins materials combined with scaling procedures, and 4) application of the Synthetic Rock Mass
(SRM) approach to study the strength and deformation behaviour of the main geotechnical units.

This paper describes the second part of the study, with focus on point 4.

The Synthetic Rock Mass (SRM) modelling approach, based on particle mechanics, has been developed to
simulate the mechanical behaviour of jointed rock mass. This technique uses the bonded particle model for
rock to represent intact material and the smooth-joint contact model to represent the in-situ joint network.
The macroscopic behaviour of an SRM sample depends on both the creation of new fractures through intact
material and slip/opening of pre-existing joints. SRM samples containing thousands of non-persistent joints
can be submitted to standard laboratory tests (UCS, triaxial loading, and direct tension tests) or tested
under a non-trivial stress path representative of the stresses induced during the engineering activity under

The micro-parameters of the bonds and the smooth-joint contacts between the particles have been calibrated
against the mechanical properties and scaling laws for intact rock and veins, so that representative virtual
SRM samples of the four different geotechnical units could be generated and tested.

Results from the SRM simulations include pre-peak properties (modulus, damage threshold, peak strength,
etc.) and post-peak properties (brittleness, dilation angle, residual strength, fragmentation, etc.). Of
particular interest is the ability to obtain predictions of rock mass scale effects, anisotropy, and brittleness,
properties that cannot be obtained using empirical methods of property estimation.

1 Introduction
Presently, there are mainly two traditional methods used to estimate the strength of the rock mass: 1)
determination of the strength envelope of the rock mass using scaling parameters from laboratory tests,
for example Hoek-Browns failure criterion, and 2) using numerical modelling based on back analysis of
previous experiences of failure observed and measured in mining or civil excavations.

Caving 2014, Santiago, Chile

The primary copper ore of El Teniente mine has no fractures; therefore, it is difficult to determine a rating
that would enable to scale laboratory data, such as, RMR (Laubscher 1977) or GSI (Hoek 1994). This is the
reason why the first option is not valid unless the entry data is manipulated or adjusted in order to obtain
reasonable results. The second option requires, in general, a good characterization of a previous failure
event in the rock mass, which is not always available.

In recent years, new techniques of numerical modelling have been developed (PFC3D, ELFEN, Abaqus)
aiming to capture the real behaviour of the rock mass. Those methodologies are the third way to estimate
the strength of the rock mass, but they are still in development.

This paper is composed of two parts. The first one is included in Brzovic et al. (2014). The second part
includes results from numerical modelling in the primary copper ore in El Teniente mine, particularly
it has been implemented the concept of SRM developed by Itasca (Pierce et al. 2007; Mas Ivars et al.
2011). The methodology is divided into making a geotechnical characterization (mapping and laboratory
tests), developing scaling laws and applying the SRM approach. This paper aims to study the strength and
deformation behaviour of four rock mass domains from the El Teniente mine (Dacite, Diorite and CMET)
and compares these results with the estimations based on classification systems and other numerical models.
This study endeavours increased knowledge based on a previous work with this technique in El Teniente
veined rock mass (Vallejos et al. 2013).

2 Synthetic rock mass components

The SRM method is based on the generation and testing of three-dimensional synthetic rock mass samples
in order to simulate the mechanical behaviour of jointed or veined rock masses. SRM is implemented in
PFC3D 4.0 software (Itasca 2008) and uses the interface SRMLab 1.7 (Itasca 2012). Figure 1 summarizes
the main components of the model, which represents the intact rock as an assembly of bonded particle
(Figure 1a), using the Enhanced Bonded Particle Model (BPM), and an embedded Discrete Fracture
Network (DFN) to represent joints (Figure 1b). Each joint is represented explicitly using the smooth-joint
contact model (SJCM).

(a) (b) (c)

Figure 1 Sample constructed with PFC3D particles (a), DFN superimposing onto the previous sample (b) and
Synthetic Rock mass sample (c) (Board & Pierce 2009)

Two models compose the Enhanced Bonded Particle Model (BPM), which represents intact rock; the particle
contact and the parallel bond model. A more detailed explanation of the standard and enhanced BPM for
rock can be found in Potyondy & Cundall (2004) and Potyondy (2011). The smooth-joint contact model
(SJCM) represents joints in the SRM samples simulating the behaviour of a smooth interface, regardless of
the local particle contact orientations along the interface. This model makes possible the creation of large

Caving Mechanics

volumes of synthetic rock containing thousands of non-persistent joints. A more detailed explanation of
SJCM can be found in Mas Ivars et al. (2008b).

The most logical choice to represent explicitly a vein network is by using a Discrete Fracture Network (DFN)
(Dershowitz & Einstein 1988). Each DFN represents different rock mass domains, and are characterized
by fracture intensity and orientation of each set of veins. The data and the procedure used to develop each
DFN model is described in Brzovic et al. (2014).

3 Calibration of the SRM model

In order to calibrate a SRM model and to create a large-scale model, laboratory tests and fields observations
are needed. A summary of data for intact rock and veins properties and scaling procedures are described in
Brzovic et al. (2014).

3.1 Intact rock

Laboratory data for UCS tests is adjusted using the relation proposed by Yoshinaka et al. (2008), defining a
scaling power law for each lithology:



c: is the uniaxial compressive strength of a cylindrical specimen with a diameter ,

k: is a material constant.

Even though exists a calibration procedure, the basic way to define a set of micro-parameters is by a trial
and error approach (Itasca, 2008). The size of calibration is defined by the intact block within each DFN.
The size of particle is selected equal to four particles along the average intact block size, and the aspect
ratio of the calibration sample is 2.1:1. The assumptions taken for the intact rock calibration are detailed
in Vallejos et al. (2013). To sum up, there are assumptions to reproduce better the brittle behaviour of El
Teniente rock masses and other ones suggested on previous studies for reproducing hard rock behaviour
(Potyondy & Cundall 2004; Potyondy 2011).

The model cannot reproduce Poissons ratios larger than approximately 0.10 if a reasonably brittle response
is desired. Due to the rock mass response being more influenced by the veins behaviour, Poissons ratio is
not considered in the calibration. The rest of macro-parameters were matched with less than 1% of error.

3.2 Veins

Brzovic & Villaescusa (2007) suggest that veins with thicknesses greater than 2 mm and with less than
1/3 of hard minerals play a relevant role controlling fragmentation and in the seismicity during caving
propagation. The present study includes only soft veins with thicknesses greater than or equal to 1 mm,
assuming that veins with thickness between 1 and 2 mm affect the rock mass behaviour. Macro-parameters
have to be scaled to represent the average in-situ conditions of each rock mass unit. In this case the average
length of veins is 1 m, therefore all macro-parameters are scaled to this length. It is considered that friction
angle is not influenced by scale effect. The procedure to calibrate the model considers estimating micro-
parameters based on results of previous simulations. The calibration procedure is detailed in Vallejos et al.
(2013). The assumptions taken for veins calibration consider peak and residual friction angles of veins to be
40 and dilation angle to be 0. It is assumed that most of the rock mass dilation comes from block rotation
and the relative large size of the particles.

Caving 2014, Santiago, Chile

4 Characterization of rock mass units and behaviour

4.1 Geometry and test configurations

The tested geometry is a cylinder of width equal to ten times the intact block size with an aspect ratio height
to width of 2.1:1. Each DFN is a statistical representation of the network of veins forming the stockwork in
a 30 m x 30 m x 30 m volume, but the SRM specimen widths are set to only 10 DFN average spacing for
each lithology and associated DFN in order to minimize simulations time. Table 1 summarize the geometry
used in the modelling. It has to be taken into account that the SRM sample sizes used in this study are not
large enough to reach the REV (Esmaieli et al. 2010). Considering all lithological units, P32 in function of a
cubic sample width converges to a mean value in samples with over 10 m width.

Table 1 DFNs and SRMs dimensions for the block geometrical analysis

DFN Average SRM Width SRM Particle SRM Number

spacing (m) (m) diameter (m) of particles
Dacite 0.70 7.0 0.174 169,738
Diorite 0.16 1.6 0.040 166,845
CMET HW 0.22 2.1 0.052 171,704
CMET FW 0.13 1.3 0.033 159,375

4.2 Characterization of rock mass behaviour

For each test, the pre-peak rock mass parameters and stress-strain behaviour are registered. Direct tension
tests and triaxial tests are performed to characterize the SRM response of the rock masses. The testing
directions include two orthogonal horizontal directions (direction 1 refers to the E-W direction, direction 2
refers to the N-S direction) and the vertical direction (direction 3) for each lithology.

Figure 2 presents the stress-strain behaviour for triaxial tests in the three testing directions for the four
lithologies. It is observed that Dacite presents a degree of anisotropy. As expected, the peak strength
increases with confinement. However, the post-peak behaviour tends to be more brittle as confinement
increases in Dacite, while in the others lithologies post-peak behaviour is brittle only in low confinements.

4.3 Characterization of rock mass parameters and comparison with classification systems

Table 2 summarize the main geotechnical parameters of each lithological unit. These data is used to compare
SRM results with estimations based on the classification systems.

Table 2 Geotechnical parameters for each lithology (Brzovic 2001)

Lithology Ei (GPa) UCS (MPa) mi GSI D RMR

Dacite 43 167 10.6 75 - 90 0 72 - 77
Diorite 45 140 9.2 70 - 90 0 68 - 72
CMET FW 55 97 12.1 70 - 85 0 66 - 72
CMET HW 55 121 12.1 70 - 90 0 66 - 74

Caving Mechanics

Figure 2 Axial stressstrain and volumetricaxial strain curves for triaxial tests in three directions

SRM envelopes are compared with the resulting envelopes obtained from a previous investigation in El
Teniente, where a numerical model based in back analysis of documented collapses at Esmeralda was
developed (Pardo et al. 2012). Also, the peak strength envelope of the rock mass is compared with the one
proposed by Hoek et al. (2002):




GSI: Geological Strength Index, D: factor of disturbance (blast damage and stress relaxation), c uniaxial
compressive strength of the intact rock material and, mi material constant.

Figure 3 presents a comparison of peak strength envelopes estimated with different methodologies.

The strength envelopes estimated with the SRM technique are non-linear and they are not consistent with
the envelopes estimated with the Hoek-Brown criterion based on the parameters of Table 2. None of the
envelopes are close to the envelope estimated with the minimum GSI. However, SRM envelopes are similar
to the envelopes estimated for the mine scale elastic-plastic numerical modelling study (Pardo et al. 2012).
Table 3 shows a summary including Hoek-Brown and Mohr-Coulomb parameters adjusted in RocData. The
range of values includes results in the three tested directions. These values indicate that the GSI resulting
from SRM modelling is between 41 and 60; therefore all rock masses have a fair quality. Geological
information of El Teniente mine indicates that El Teniente rock masses have a good to very good quality.

Caving 2014, Santiago, Chile

Figure 3 Comparison of peak strength envelopes for each lithological unit

Table 3 Geotechnical parameters adjusted from SRM modelling for each lithological unit

Lithology GSI c (MPa) ()

Dacite 19.4 22.9 55.5 57.7 3.7 4.1 50.0 53.2
Diorite 11.3 13.1 45.3 47.3 2.6 2.7 42.6 44.7
CMET FW 12.8 15.7 51.8 54.5 2.6 2.9 42.6 45.0
CMET HW 11.0 13.7 48.6 52.2 2.7 2.8 43.2 44.5

Uniaxial compressive strength and Yongs modulus of the rocks masses are compared with empirical
formulas based on classification systems:

1. Uniaxial compressive strength of the rock mass (Table 4). Empirical formulas proposed by Hoek
et al. (2002) and Hoek & Brown (1988) are used to compare modelling results. These formulas
are based on s and a defined in equation (4), and RMR, the rock mass rating of Bieniawski (1974).
2. Youngs Modulus of the rock mass (Table 5). Empirical formulas proposed by Serafim & Pereira
(1983) and Hoek & Diederichs (2006) are used to compare modelling results. These formulas are
estimated using the intact rock modulus (Ei). The Youngs modulus estimated with SRM approach
has a slight dependence on the minor principal stress and no significant evidence of anisotropy.

Caving Mechanics

Table 4 Comparison of uniaxial compressive strength of the rock mass using different methods

Table 5 Comparison of Youngs Modulus of the rock mass using different methods

4.4 Effect of preconditioning by hydraulic fracturing

To study the effect of the hydraulic fracturing on the rock mass behaviour, sub-horizontal fracture planes
were included explicitly in the SRM sample. Figure 4 presents the stress-strain curves resulting from triaxial
tests in directions 1 (E-W), 2 (N-S) and 3 (vertical direction). Comparing these results with stress-strain
curves in Figure 2, it is clear the impact on elastic parameters of the sample tested in the vertical direction.
These results can be complemented with other studies that have shown the relevance of preconditioning by
hydraulic fracturing in cave propagation and primary fragmentation (Snchez Juncal et al. 2014).

Figure 4 Stress-strain curves resulting from simulations with fractures due to preconditioning

5 Conclusions
The SRM approach has been used to characterize the behaviour of four lithological units from El Teniente
mine. The results are promising and show an improvement compared to those reported in the previous
papers (Vallejos et al. 2013; Mas Ivars et al. 2013).

The main advantage of the SRM approach is that it allows estimating the behaviour of a synthetic sample
as a result of a geotechnical and geological characterization, and not as a result of a back analysis or

Caving 2014, Santiago, Chile

empirical formula. Results are consistent with the fundamental principles of rock mechanics. Nevertheless,
some limitations have to be overcome, such as the fact that it is not possible to fit the Poisson ratio keeping
a brittle post-peak behaviour. Furthermore, the particle assembly is calibrated to the strength and elastic
behaviour of the average rock blocks in the rock mass; therefore the rock block scale effects is not captured

SRM standard tests show acceptable results scaling veins macro-parameters to the average length within
the DFN. Due to all veins having the same macro-parameters, the only differences between models for each
lithology are intact rock micro-parameters and the influence of the vein network geometry (DFN). These
assumptions results in envelopes comparable with another numerical modelling estimation.

The effect of the hydraulic fracturing on the rock mass behaviour, resulting from SRM modelling, show a
high potential for the SRM approach to evaluate the effect of including new fractures due to pre-conditioning
in the field.

It is recommended to further investigate the effect of increasing the SRM sample size and also make
additional effort in complementing laboratory and field data to support the numerical modelling results.

The authors acknowledge The El Teniente Division of Codelco-Chile for their permission to publish the
data and for supporting this work. This study was commanded by API T10E202 of Codelco-Chile (contracts
4501127645 and 4501142662) and by FONDECYT Initiation Grant #11110187. Caroline Darcel, Romain
Le Goc and Lauriane Bouzeran from Itasca Consultants SAS are also acknowledged for their contribution
to this work.

Board, M & Pierce, M 2009, A Review of Recent Experience in Modeling of Caving, International
Workshop on Numerical Modeling for Underground Mine Excavation Design: Proceedings of
the 43rd US Rock Mechanics Symposium, Asheville, United States.

Brzovic, A, Schachter, P, de los Santos, C, Vallejos, J & Mas Ivars, D 2014, Characterization and Synthetic
Simulations to Determine Rock Mass Behaviour at the El Teniente Mine, Chile. Part I,
Proceedings of the 3rd International Symposium on Block and Sublevel Caving, Santiago,

Brzovic, A & Villaescusa, E 2007, Rock mass characterization and assessment of block-forming
geological discontinuities during caving of primary copper ore at the El Teniente mine, Chile,
International Journal of Rock Mechanics and Mining Sciences, vol. 44, pp. 565-583.

Brzovic, A 2001, Fundamentos geolgicos para un sistema de clasicacin geotcnico del macizo rocoso
primario, mina El Teniente, Internal report SGL-187/2001 of the Superintendence Geology,
CODELCO-Chile El Teniente Division [in Spanish].

Brzovic, A 2009, Rock mass Strength and Seismicity during Caving Propagation at the El Teniente Mine,
Chile , In: Proceedings of 7th International Symposium on Rockburst and Seismicity in Mines
(RaSiM07). Tang, C.A. editor. Dalian University. (2) 838-52.

Dershowitz, W & Einstein, H 1988, Characterizing rock joint geometry with joint system models, Rock
Mechanics and Rock Engineering, vol. 21, pp. 21-51.

Caving Mechanics

Hoek, E & Diederichs, M 2006, Empirical estimation of rock mass modulus, International Journal of
Rock Mechanics and Mining Sciences, vol. 43, pp. 203-215.

Hoek, E, Carranza-Torres, C & Corkum, B 2002, Hoek-Brown failure criterion-2002 edition, Proceedings
of the fifth North American Rock Mechanics Symposium, Toronto, Canada, vol. 1, pp. 267

Hoek, E 1994, Strength of rock and rock masses, ISRM News Journal 2, pp. 416.

Hoek, E & Brown, E 1988, The HoekBrown failure criterion a 1988 update, Proceedings of the 15th
Canadian Rock Mechanics Symposium, pp. 3138.

ISRM 1981, Suggested methods for the quantitative description of discontinuities in rock masses in Rock
characterization, testing and monitoring, ISRM Suggested methods, (edited by ET Brown),
Pergamon Press, pp. 3-52.

Itasca Consulting Group, Inc. 2012, SRMLab version 1.7, Minneapolis, United States.

Itasca Consulting Group, Inc. 2008, PFC3D Particle flow code in 3 dimensions, Version 4.0, Minneapolis,
United States.

Laubscher, D 1977, Geomechanics classification of jointed rock masses mining applications, Trans.
Inst. Min. Metall., 86, A1-A8.

Machuca, L & Villaescusa, E 2011, Summary of intact rock property values for Codelco ChileDivisin El
Teniente, Western Australian School of Mines - Geomechanics laboratory report to Divisin
El Teniente, Codelco Chile, API T10E202.

Mas Ivars, D, Bouzeran, L, Le Goc, R & Darcel, C 2013, Final report on Synthetic Rock Mass (SRM)
Fragmentation Analysis El Teniente March, 2013, Itasca Consulting Group report to
Divisin El Teniente, Codelco Chile, API T10E202.

Mas Ivars, D, Pierce, M, Darcel, C, Reyes-Montes, J, Potyondy, D, Young, P & Cundall, P 2011, The
Synthetic Rock Mass approach for jointed rock mass modeling, International Journal of Rock
Mechanics and Mining Sciences, vol. 48, pp. 219244.

Mas Ivars, D, Pierce, M, DeGagne, D & Darcel, C 2008a, Anisotropy and scale dependency in jointed rock-
mass strengthA synthetic rock mass study, Continuum and Distinct Element Numerical
Modeling in Geomechanics: Proceedings of the 1st International FLAC / DEM Symposium,
Minneapolis, United States, paper 06-01, pp. 231- 239.

Mas Ivars, D, Potyondy, D, Pierce, M & Cundall, P 2008b, The smooth-joint contact model, Proceedings
of the Eighth World Congress on Computational Mechanics and Fifth European Congress on
Computational Methods in Applied Sciences and Engineering, Venice, Italy, paper a2735.

Pardo, C, Villaescusa, E, Beck, D & Brzovic, A 2012, Back Analysis of intensive rock mass damage at the
El Teniente Mine, CRC-Mining Conference, Brisbane, Queensland University.

Pierce, M, Mas Ivars, D, Cundall, P & Potyondy, D 2007, A synthetic rock mass model for jointed rock,
Proceedings of the 1st Canada-US Rock Mechanics Symposium, Vancouver, Canada, vol. 1,
pp. 341-349.

Potyondy, D 2012, The bonded-particle model as a tool for rock mechanics research and application: Current
trends and future directions, The Present and Future of Rock Engineering, Proceedings, of the
7th Asian Rock Mechanics Symposium, Seoul, Korea, pp. 73-105.

Caving 2014, Santiago, Chile Caving Mechanics

Potyondy, D 2011, Parallel-bond refinements to match macroproperties of hard rock, Continuum and
Distinct Element Numerical Modeling in Geomechanics: Proceedings of the 2nd International
FLAC / DEM Symposium, Melbourne, Australia, Paper 08-04, pp. 459-465.

Potyondy, D & Cundall, P 2004, A bonded-particle model for rock, International Journal of Rock
Mechanics and Mining Science, vol. 41, pp. 1329-1364.

Rojas, E, Cavieres, P, Dunlop, R, & Gaete, S, 2000, Control of Induced Seismicity at the El Teniente Mine,
Codelco Chile, Proceeding Massmin, Chitombo, G, editor, Brisbane, Australia, AusIMM,

Snchez Juncal, A, Mas Ivars, D, Brzovic, A & Vallejos, J 2014, Simulating the effect of preconditioning
in primary fragmentation, to be published in Proceedings: Eurorock 2014, Vigo, Spain.

Serafim, J & Pereira, J 1983, Considerations on the geomechanical classification of Bieniawski,

Proceedings of the International Symposium on Engineering Geology and Underground
Construction, Lisbon, Portugal, vol. 1, pp. 33-44.

Vallejos, J, Brzovic, A, Lopez, C, Bouzeran, L & Mas Ivars, D 2013, Application of the Synthetic Rock Mass
approach to characterize rock mass behavior at the El Teniente Mine, Chile, Continuum and
Distinct Element Numerical Modeling in Geomechanics: Proceedings of the 3rd International
FLAC / DEM Symposium, Hangzhou, China, paper: 07-02.

Yoshinaka, R, Osada, M, Park, H, Sasaki, T & Sasaki, K 2008, Practical determination of mechanical
design parameters of intact rock considering scale effect, Engineering Geology, vol. 96, pp.

Caving 2014, Santiago, Chile



Fragmentation estimates using BCF software

Experiences and pitfalls
J Jakubec, SRK Consulting Ltd., Canada

Fragmentation estimates are one of the key inputs for cave mine design. Block cave fragmentation (BCF)
software has been the industry standard for decades. However, experience from several mines has shown
poor correlation between the initial fragmentation estimates and reality. Realistic input parameters in the
BCF software are key to realistic fragmentation estimates. Often, such parameters are based on drill core
data, but correct assessment of rock mass parameters based on such data could be a challenging task.
The author of this paper discusses some of the reasons for poor reconciliation and shares his experience
and methodology with using BCF software. Experience has shown that BCF software may overestimate
fragmentation because of conservatism during the feasibility stage, drill core bias, ignoring fines and
weathering, and inadequate accounting of rock block defects. The quality of fragmentation predictions
using BCF software can be improved significantly through careful evaluation of these factors.

1 Introduction
The BCF software was developed and introduced to the mining industry in the 1990s (Esterhuizen 1994;
Esterhuizen et al. 1996). There have been several changes to the software code since then and currently the
most up to date version is BCFV305.

Although there are other techniques to assess block caving fragmentation, BCF software remains a proven
and practical method that enables the rapid evaluation of different scenarios.

However, the current general experience in the mining industry is that BCF software predicts coarser
fragmentation than the actual fragmentation present. The author of this paper uses his experience from
a number of operating cave mines to analyze and discuss potential reasons for such a discrepancy and to
suggest solutions.

2 Fragmentation in block caves

During the rock mass caving process, the rock blocks are formed by four mechanisms:

Gravity liberation of existing blocks bounded by open or weakly healed joints.

Stress fracturing via intact rock or via rock block defects.

Dynamic impact breakage due to rockfalls.

Breakage during the communition processes in the cave.

When describing the rock mass fragmentation in caving mines, three types of fragmentations are recognized:

In situ fragmentation.

Primary fragmentation.

Secondary fragmentation.

Caving 2014, Santiago, Chile

2.1 In situ fragmentation

In situ fragmentation is the block size distribution of naturally formed blocks that are bounded by open or
weakly healed joints. This block size distribution would be formed by gravity as if the rock mass simply
fell apart. No new stress fractures would be induced or considered.

2.2 Primary fragmentation

Beside gravity, the rock mass caving processes typically involve induced stresses acting in the cave
envelope. Such stresses cause rock fracturing through intact rock or along the pre-existing defects, further
reducing the in situ fragmentation. Laubscher (2000) defines primary fragmentation as the fragmentation of
the rock as it parts from the surrounding rock mass.

2.3 Secondary fragmentation

Secondary fragmentation is the further break down of the rock as it moves down the draw column (Laubscher
2000). The primary blocks are subjected to communition processes within the cave, further reducing the
primary blocks and generating fines. In a cave where an air gap is present, the primary blocks could also be
reduced by dynamic breakage during rockfall and drilling.

3 Rock mass characterization

Two rock mass parameters that are critical to define for realistic fragmentation estimates are rock mass
defects and rock strength.

3.1 Rock mass defects

Geologic processes prior to mining such as brittle deformation of and/or sedimentation can introduce defects
that have variable geometry, continuity, shear strength, and cohesion. Such defects could significantly
reduce the rock block or rock mass strength, especially in an unconfined situation (Jakubec 2013). Rock
mass defect characterization should include a range of different scale structures, from large-scale structures
through joints to rock block defects.

3.1.1 Large-scale structures

Large-scale structures in the context of this paper include all types of large rock mass structures such as
faults, shear zones, or closely spaced joint clusters.

Although large-scale structures do not typically influence rock fragmentation processes, they are sources
of fines (typically, fragments are smaller than 0.001 m3) and hence should be defined for fragmentation
analysis. Typical rock mass in caving operations includes 315% of in situ fines contained within the large-
scale structures (Figure 1).

3.1.2 Open and cemented joints

It is typical industry practice, when describing a drill core or during the mapping of tunnel walls, to describe
defects that do not have bonding cement as open joints (Figure 1). In the context of fragmentation analysis,
the open joints should have sufficient continuity that they form an in situ block in other words, they
should be block-bounding joints. Often, the joints are cemented with mineral infill of variable strength
ranging from very weak to strong with strength similar or occasionally exceeding the intact rock strength


(Figure 2). Fragmentation analysis using BCF software for rock masses that do not have open or weakly
cemented block-bounding joints is not reliable. In such cases, a more sophisticated analysis such as a
synthetic rock mass (SRM) approach should be used (Jakubec et al. 2012).

Figure 1 Example of fines generating fault (left) and open joints (right) in the drill core

3.1.3 Rock block defects

A special category of defects are small discontinuous fractures and veins or micro-fractures (Figure 2). Such
defects have limited continuity and are contained within the in situ blocks (Laubscher & Jakubec 2001).

Figure 2 Example of cemented joint (left) and rock micro-fractures (right)

The description of such defects and the challenges in their characterization have been discussed in several
papers (Jakubec 2013). To avoid double dipping and incorrect material strength characterization, the
micro-defects contained within the hand specimen and affecting laboratory unconfined compressive
strength (UCS) tests should not be taken into the subsequent IRS strength reduction process. The defects
that do not affect UCS tests must be included in rock block strength reduction.

3.2 Rock strength

Both primary and secondary fragmentation are influenced also by rock strength, stresses acting on the
rock in the cave back, and point loading in the cave column. The following categories of rock strength are

Intact rock strength.

Rock block strength.
Rock deformation.

Caving 2014, Santiago, Chile

The MRMR classification system developed by Laubscher & Jakubec (2001) recognizes and addresses the
need to include all important defects in the rock mass classification and rock mass strength.

3.2.1 Intact rock strength (IRS)

Intact rock strength in terms of BCF analysis is defined as the unconfined compressive strength of the rock
specimen that can be directly tested, including all internal micro-defects contained in the specimen.

3.2.2 Rock block strength (RBS)

Rock blocks bounded by open or cemented joints will have a lower strength than the intact rock strength
if their dimensions exceed approximately 50 mm. Discontinuous joints, fractures and veins that terminate
within rock blocks and do not take part in the formation of blocks will further reduce the rock block
strength. The intact block strength in the BCF software is equivalent to rock block strength in terms of the
Laubscher-Jakubec MRMR classification. The concept of strength adjustment to rock block and rock mass,
and strength reduction of rock block strength is illustrated in Figure 3.

Figure 3 Example of RBS concept (left) and RBS reduction from the Chuquicamata mine in Chile (after
Jakubec et al. 2012)

3.2.3 Rock failure criterion

Rock mass failure characteristics are also included in the BCF software approach via Hoek & Brown mb-
value for rock mass. The Hoek & Brown criterion is used to determine the triaxial strength of the rock and
the value may be estimated using published tables for typical materials or determined by laboratory tests.

4 Block caving fragmentation

The BCF software is a commercially available computer program authored by Dr. G. Esterhuizen and has
been used by the cave mining industry for the past two decades.

The program is based on analytical and empirical rules describing the fragmentation processes and
factors that play a role in block cave fragmentation. (Esterhuizen 2005)

The program consists of three main modules:

Primary fragmentation module.

Secondary fragmentation module.

Hang-ups module.


The latest available version is BCFV3.05 from 2005.

It is not an objective of this paper to describe in detail the BCF engine or algorithms and the author of this
paper recommends that the reader refers to the BCF Technical Reference and User Guide (Esterhuizen

4.1 BCF software input data

A comprehensive data set is required for all three BCF software modules and the input data are organized
in three main groups:

Geology (or rock mass) input.

Cave input.

Draw input.

The first two inputs are used to calculate primary fragmentation and the last input is used to calculate
secondary fragmentation and conduct the hang up analysis.

4.1.1 Geology input

The primary fragmentation module requires input from two areas: geology and cave information. The
geology input consists of:

Rock mass information

Rock type simple rock type name or abbreviation.

Rock mass rating MRMR (Laubscher 1990).

Hoek & Brown m value for rock mass (mb) from published tables or calculated from laboratory

Intact rock strength (UCS in MPa).

Fracture/veinlet frequency/m, ff/m, (rock block defects).

Fracture/veinlet conditions (Laubscher 1990 joint conditions equivalent).

Intact block strength (same as rock block strength) this value is calculated or can be manually

In situ block bounding joint information

Joint set number three sets define the block, an additional set will shape it; there is typically no
material benefit to inputting more than five sets.

Joint dip average and range.

Joint dip direction average and range.

Joint spacing average and range (minimum and maximum).

Joint condition average and scatter (Laubscher 1990 joint conditions, 1-40).

Joint orientation is typically obtained from stereonet analyses of joint data.

Caving 2014, Santiago, Chile

4.1.2 Cave input

The cave input table is the second input group of information that has to be completed in order to calculate
primary fragmentation. The cave input data include:

Cave orientation dip and dip direction of the caving face.

Stresses dip and strike direction, and normal stress of the induced stresses in the cave back
(values could be estimated using the BCF software manual guidelines or obtained from numerical
modelling analyses).

Stress spalling and fracturing stress fractures may form in the cave face if the stresses are high
enough to cause compressive failure of the rock. Spalling provides a second option to model the
effects of stress fracturing. The amount of spalling is entered as a fixed percentage of the volume
of rock.

4.1.3 Draw input

Draw input data are required to model secondary fragmentation and hang up analysis. The data include the

Primary fragmentation file.

Draw data:

o Draw height.

o Maximum caving height.

o Draw width.

o Swell factor.

o Rock density.

o Additional fines.

o Rate of draw.

Draw bell size.

5 Experiences of using BCF software

As mentioned before, the general experience in the mining industry is that the BCF software predicts
coarser fragmentation than the actual fragmentation present. This view is supported by observations from
several caving operations that the author has been involved with.

Experience shows that there are several potential reasons for this discrepancy:

Focus on an oversize portion (+2 m3) of the block distribution curve during the feasibility study,
hence a coarser, more conservative curve is selected as the base case.

The quality of the input data, which are often based on drill core with orientation bias and under
sampled joints populations.


Ignoring fines sourced from faults, shear zones, and joint clusters.

Not considering the weathering effect on fragments in the cave zone.

Ignoring portion of the joint populations when selecting major joint sets and not reconciling with
the ff/m value measured in the core.

Ignoring rock block defects.

5.1 Focus on course blocks

The range of input data results in a range of fragmentation curves for the same rock mass. It is often the
tendency to select the more conservative analysis to acknowledge uncertainty during the feasibility study
stage because coarse blocks during the cave mining operation can cause operational difficulties that result
in costly delays, due to hang ups and secondary blasting. Although this is a valid concern, inevitably coarse
fragmentation distribution curves are selected as the base case scenario and this can potentially result in a
discrepancy with the actual experience during production. Typically, finer fragmentation realization is not
considered as important. However, it is important to consider both sides of the possible range: a coarse
curve for material handling and production rates and a fine curve mainly for drawpoint spacing.

5.2 Orientation bias of the drill core

Preferred drill hole orientation can under sample joints populations resulting in the underestimation of the
ff/m value and/or the joint set number, see Figure 4. It is important to orient drill holes in different directions
to capture a complete family of joints. It is also important to apply a Terghazi correction in the stereoned
analysis to minimize orientation bias. Acoustic or optical televier data could also improve understanding of
the joints populations.

Figure 4 Example of drill hole orientation bias

5.3 In situ fines

In situ fines are sourced from large-scale structures such as faults, shear zones, and joint clusters. Generally,
fragments smaller than 10 cm3 are to be considered as fines, which means that rock with a ff/m value of
10 can produce significant amount of fines. It is relatively easy to estimate amount of in situ fines if core
logging did not ignore large-scale structures. Experience shows that typical caveable rock masses have
315% in situ fines.

Caving 2014, Santiago, Chile

5.4 Weathering of the caved rock

Weathering susceptibility and oxidization of the rock has to be considered. Broken rock in the cave zone
usually resides in the cave for several years with constant attrition and often water migration. Some
rocks, such as kimberlites and serpentinite, are susceptible to weathering, while oxidization is typical for
sulphidic ore. Both processes could generate considerable amount of fines that is not accounted for in the
fragmentation analysis.

5.5 Under sampling joints populations

The under sampling of joints populations is considered as one of the leading causes of discrepancy between
analysis and reality. Joint sets are often selected graphically on stereonets using fences around the joint
pole concentrations (Figure 5a). Plotting poles revealed waste number of joints outside fences.

Figure 5 Selected joint sets based on stereonet without poles (a) and with poles (b)

The problem is that a large amount of more randomly distributed and oriented joints is not accounted for.
If a fracture frequency check as discussed above is not undertaken, the fragmentation curves could be very
coarse and far from reality. Depending on the structural character of the rocks there are two main ways to
rectify this. One way is to decrease the joint spacing for one or more joint families to match the expected
ff/m value. The other way is to create a new joint set and assign the spacing to match the ff/m value. The
BCF software tool for verifying the ff/m value is powerful and should be used to make sure that all joints
are accounted for.

5.6 Ignoring rock block defects

Another common reason for underestimating fragmentation distribution curves is ignoring rock block
defects. Currently, the only classification system that includes rock block defects is Laubscher & Jakubec
MRMR (2001). If logging is done for example using Beniawski RMR or Hoeks GSI, rock block defects
are ignored and rock block strength will be overestimated. A comparison of primary fragmentation results
with and without defects is illustrated in Figure 6.


Figure 6 Comparison of fragmentation curves with and without defects

6 Conclusions
The BCF software remains one of the leading practical tools to assess block caving fragmentation. Despite
its limitations, it enables the rapid assessment of different scenarios and investigation of sensitivity to
individual parameters. Because of the complexity of fragmentation analyses, it is very difficult to guess the
outcome by changing various parameters. For example, for a strong rock mass in a low stress environment
the stress level variation will probably not produce realizations that are materially different. On the other
hand, for rock masses where the stress and rock block strength values are close, a small change in stress
level could produce significantly different outcome. It is difficult to comment on the accuracy of the
fragmentation analysis. The accuracy is different for each case and dependent on input data quality, stress
strength relationship, etc. In the case of the Chuquicamata underground study, two approaches, BCF and
SRM combined with numerical analysis were used and yielded similar results (Figure 6).

One of the biggest challenges for realistic fragmentation estimates is to correctly define block bounding
joints. In cases where such joints are cemented with strong infill, a more sophisticated approach such
as SRM should be used. For examples, the SRM approach in fragmentation was discussed in papers by
Jakubec et al (2012) and by Jakubec (2013).

As in any analysis, it is important to have realistic and complete data for input in order to produce realistic
fragmentation estimates. This paper illustrates some of the reasons for the discrepancy experienced by the
mining industry between the estimated and actual fragmentation. However, caution has to be exercised
always to manage expectations. None of the techniques can and most like will not predict fragmentation
with the often expected high degree of accuracy. The geological nature of the orebodies is too complex and
it is not realistic to capture such complexity in the analysis without calibration to the real data. However, by
considering all the available information we can produce estimates that are roughly right and not precisely

The author would like to thank Dr. Gabriel Esterhuizen for his valuable comments, and Ms. Van Ngo and
Ms.Sophia Karadov for their assistance in preparing the document for publication.

Caving 2014, Santiago, Chile

Esterhuizen GS 1994, A model for predicting block cave fragmentation, in Application of Numerical
Modelling in Geotechnical Engineering, South African National Group of the ISRM, Pretoria,
South Africa, September 1994, pp 147-151.

Esterhuizen, GS, Laubscher, DH, Bartlett, PJ & Kear, RM 1996, An expert system approach to predict
fragmentation in block caving, Massive Mining Methods, South African Institute of Mining
and Metallurgy, Colloquium, pp. 2-11.

Esteruhisen, GS 2005, A program to predict block cave fragmentation - Technical reference and users

Hoek, E and Brown, ET 1997, Practical estimates or rock mass strength, Int. J. Rock Mech. Min.g Sci. &
Geomech. Abstr., vol 34, N8, pp. 1165-1186.

Jakubec, J, Board, M, Campbell, R, Pierce, M, Zaro, D 2012, Rock mass strength estimateChuquicamata
case study, in Proceedings MassMin 2012, June 10-14, Canadian Institute of Mining,
Metallurgy and Petroleum (CIM), Sudbury, Canada, CD-rom only.

Jakubec, J 2013, Role of Defects in Rock Mass Classification, Ground Support 2013 Conference, ACG,
Perth, Australia.

Laubsher, DH, A geomechanics classification system for the rating of rock mass in mine design, J.S. Afr.
Inst. Min. Metall., vol. 90, no. 10. Oct. 1900. pp, 257-273.

Laubscher, DH 2000, A practical manual on block caving, International Caving Study (1997-2000),
University of Queensland, Brisbane, Australia.

Laubscher, DH and Jakubec, J 2001, The MRMR Rock Mass Classification for Jointed Rock Masses, in
Underground Mining Methods: Engineering Fundamentals and International Case Studies,
eds. W.A. Hustrulid and R.L. Bullock, Society of Mining Metallurgy and Exploration, SMME,
pp. 475481.


An alternative approach to verifying predicted

fragmentation in weak rock
RN Greenwood SRK Consulting Inc., Canada
BN Viljoen SRK Consulting (Canada) Inc., Canada

Fragmentation of ore and waste rock in block cave operations influence several aspects of mine design
including draw-point spacing, dilution entry, secondary blasting, and material handling systems. Current
accepted practice to estimate fragmentation and expected block size distribution includes empirical and
numerical analyses, such as Block Cave Fragmentation software (BCF), discrete fracture network, and
particle flow code. BCF is the most widely used and generally accepted numerical method to determine
potential fragmentation within hard rock environments with high stress levels. However, the considered
greenfields caving project, which is in a highly variable and relatively weak rock mass, has required
an alternative approach to be compared with fragmentation estimates from the BCF program. A 5 x 5
observational matrix combines observations from drill core photographs of brokenness/breakability with
weakening alteration/rock strength. The result is a fragmentation point estimate o for the percentage of rock
not passing 0.3 and 1.0 m3 (mine-specific requirements for the materials handling system). The evaluation
results show the spatial distribution of the fragmentation estimates across the project area as compared
to the fragmentation estimates based on the more conventional geotechnical inputs in the BCF analyses.

1 Introduction
Prediction of rock mass fragmentation is used, among other things, to select mobile equipment, design
material handling systems and to estimate draw-point spacing, attrition, and draw control. It is also used to
budget for secondary rock breaking. Primary fragmentation is related to the rock fabric, condition of in situ
rock blocks, and induced stresses, while secondary fragmentation considers comminution of the primary
blocks as they are drawn down through the cave and finally report to the production level draw-point. A
range of industry accepted methods are available to predict primary and secondary fragmentation. These
typically include software packages developed from analytical and empirical rock engineering principles
and refined through benchmarking studies and visual assessments.

As with many early stage projects, data sources are typically limited to drill core logging, downhole
geophysics and laboratory testing of core samples. Mapping in early stage or exploration drives is not
always available.

This paper reviews an alternative fragmentation assessment conducted for a greenfields caving project
in a highly variable and relatively weak rock mass, where the BCF fragmentation prediction was coarser
than expected based on the observed drill core condition. Predicted oversize ore negatively impacted on
the material handling system design and an alternative assessment was developed to predict the expected
the size distribution. This approach is considered suitable as an alternate estimation methodology in data-
limited projects where core photographs are the primary data source used.

2 Numerical fragmentation prediction

In most established methods of fragmentation prediction, generalized rock mass properties are used as
inputs to predict the primary and secondary fragmentation. Data from core logging or mapping is processed

Caving 2014, Santiago, Chile

to identify the type and orientation of discontinuities present (e.g., joints, cemented features, veins, and
micro-defects), joint properties (e.g., frequency or spacing, roughness, infill, and persistence) and the intact
rock strength. Data is consolidated to produce a set of inputs that is reasonably representative of the entire
caving area (at the coarsest level) or individual geotechnical domains.

Drill core logging data is typically the most common source of information and can be collected to a
high level of detail including many parameters used as input to various classification schemes, such as in
Laubscher RMR (1990), Jakubec IRMR (2000), Call Nicholas Core2Frag (2004), and Bieniawksi RMR
(1989). Early stage projects typically do not have this luxury, and despite the large volume of data points
potentially available, core logging data is limited in value providing small, three-dimensional samples of
the rock along many one-dimensional lines through the rock mass. This data is useful to establish dominant
joint orientations and spacing; however, properties such as joint persistence, primary block size, and
block aspect ratios can only be inferred but not directly measured. The presence of micro-defects is often
misinterpreted especially if drill core does not separate during the drilling and handling process. Specific
tests have been developed to estimate the intensity and strength of micro-defects in core.

Stresses within and around the cave can also influence the caving process and fragmentation. During the
early stages of a project the virgin stress state is often uncertain and assumptions are based on global stress
databases (e.g., World Stress Map, Heidbach et al, 2008) or estimated based on the regional tectonic and
structural setting. Induced stresses are then derived for the perceived current stress state and the planned
cave geometry.

2.1 Block cave fragmentation

BCF software (Esterhuizen, 2005) is a widely used fragmentation prediction tool with generally accepted
results. Primary fragmentation is derived from joint spacing, orientation, and joint conditions. Intact
rock strength, primary block dimensions, micro defects, and induced stresses determine the secondary
fragmentation. BCF was developed and calibrated for caving mines in hard rock environments where joints,
stress, and rock strength are the main contributors to fragmentation. The algorithms were not calibrated
for weaker rock caves where altered weak rock tends to break up with minor disturbance. The software
is most useful for analysis in cases where poor fragmentation is expected. This is, when rocks are highly
jointed, fragmentation is unlikely to be an issue, and BCF will simply confirm that fine fragmentation can
be expected (Esterhuizen, 2013).

BCF requires explicit inputs with limited allowance to account for variability. Inputs to the fragmentation
prediction are normally a generalization of the rock mass and geotechnical properties with a set deviation.
Mine wide data is applied and localized variability is often overlooked. The unique properties associated
with certain rock types or geotechnical conditions are not considered in the evaluation.

An additional review of the BCF fragmentation results was requested to re-assess fragmentation predictions
and over size.

3 Alternative fragmentation assessment

At the project site, a large number of diamond drill holes were completed during the exploration phase
of the project to define the extent of the deposit and develop a resource model. The recovered core was
logged geologically and geotechnically by on-site staff according to the various owners procedures. This
procedure has resulted in data of variable quality, with geotechnical parameters related to multiple rock mass
classification schemes. High quality core box photographs, and more recently in split-tube photographs,
have also been collected for the majority of the project drilling.


Figures 2.1 and 2.2 provide a comparison of core degradation through separation along pre-existing
cemented features or discontinuities. The samples also show formation of new mechanical breaks. The core
in Figure 2.1 was considered representative of the in situ rock, while the core in Figure 2.2 was considered
more representative of fragmented ore in an established cave after being subjected only to standard triple
tube drilling and handling practices.

Figure 2.1: Photo of core in split tubes during geotechnical logging

Figure 2.2: Photo of core in the core box after geotechnical logging

3.1 Core photograph re-logging

The difference in core condition highlighted in Figures 2.1 and 2.2 presented an opportunity to re-evaluate
the expected fragmentation and size distribution. A concept was developed in which the core photos were
individually reviewed and used to predict the perceived condition of the ore at the drawpoint based on the
developed parameter descriptions. Re-logging was based on the assumption that all core was handled in
a similar manner and potential inconsistencies had no insignificant influence on the interpretation. The
following conditions were applied for the re-logging:

Brokenness/breakability is the degree to which the rock was broken, irrespective of whether the
discontinuities were open joints, drilling-induced fractures, or mechanical breaks. Brokenness/
breakability was defined by the intensity and spacing of discontinuities and the length of intact core.

Alteration/hardness is the degree of core weathering or deterioration and the perceived hardness. The
rating was quantified based on the visual weathering or weakening of the rock, the amount of fines, and
the rock hardness, based on experience with the core and similar material.

Fragmentation is the extent to which caved rock was expected to fragment, based on the core
brokenness, degree of alteration, and fines content.

Caving 2014, Santiago, Chile

The photo re-logging was not based on estimating measurable numbers; rather than attempting to derive an
accurate fragmentation analysis, the process was used to qualify the fragmentation distribution throughout
the deposit. The size distribution was considered in terms of the percentage of core that would not pass
specific dimensions of 300 or 1,000 mm. These dimensions were guided by the infrastructure design of
grizzlies, conveyors, and crushers.

The core from 10 selected holes was evaluated and the brokenness/breakability, alteration/hardness, and
fragmentation were rated individually. The results with core box photographs were used as case examples
to establish the relationship between the input fields (brokenness/breakability and alteration/hardness) and
fragmentation in a 5 5 matrix (Figure 2.3). This illustrates the expected core condition for the various
ratings. The matrix was used as a template to guide further re-logging and to maintain standardized ratings
using the brokenness/breakability and alteration/hardness ratings as the input fields to predict the expected
fragmentation and size distribution.

Figure 2.3 Core photographs representing brokenness/breakability, alteration/hardness, and fragmentation


3.2 Fragmentation assessment

The re-logging procedure was then applied to assess fragmentation in and around the deposit. The selection
of drill holes to be reviewed was guided by the position of the cave footprints and the caving influence zone.
The core photographs from 75 drill holes (8,727 core box photographs) were re-logged. Each core box was
considered as a logging interval (a single rating represents the entire box) and the interval was spatially
referenced by depth and drill hole. This allowed any further assessment to be spatially constrained to a
horizon of interest (e.g., a production level) or to a particular domain or lithology within the cave footprint.
The distribution of ratings (Table 3.1 and Figure 3.1) shows the number of boxes for each rating and the
percentage of all the logged boxes.


Table 3.1 Summary of photo re-logging results

Rating Brokenness/Breakability Alteration/Hardness Fragmentation

Number of Number of Number of
Percentage Percentage Percentage
Boxes Boxes Boxes
1 106 1% 784 9% 200 2%
2 365 4% 4469 51% 367 4%
3 2155 25% 2526 29% 1589 18%
4 4640 53% 842 10% 4986 58%
5 1461 17% 106 1% 1585 18%

The re-logging provided an overview of the core condition and individual rating distribution, but no
measurable output in terms of size distribution or fines content. The experience gained during the re-
logging made it possible to relate the assigned ratings to a modified size distribution and a percentage of
contained fines.

3.3 Size distribution

The size distribution analysis produced for this project was not the typical S-curve but rather indicated
the percentage of ore expected to exceed specific dimensions. The assigned percentages (Table 3.2) are
subjective estimates based on the percentage of core expected to exceed the 300 and 1,000 mm limits.
The evaluation was based on the actual core length. Block aspect ratio was not considered as it cannot be
reliably estimated from drill core.

Table 3.2 Core size distribution for the range of brokenness ratings
Brokenness/ Percentage Not Percentage Not
Breakability Rating Passing 300 mm Passing 1000 mm
1 90% 30%
2 70% 10%
3 40% 0%
4 0% 0%
5 0% 0%

The modified size distribution for the deposit was estimated based on the distribution of fragmentation
(Table 3.1) and the estimated percentage of core that would not pass 300 mm and 1,000 mm (Table 3.2).
A weighted average method was used to calculate the total percentage core that was expected not to pass
the 300 or 1,000mm limits. The values were calculated for each fragmentation rating and expressed as a
percentage of the total core volume (Figure 3.2). The lines show the cumulative percentage of core that
would not pass 300 and 1,000 mm, respectively.

Caving 2014, Santiago, Chile

Figure 3.2 Percentage of rock not passing 300 and 1,000 mm

The results from the core photo review were then used to indicate the expected fraction of core not passing
the two specified dimensions of 300 mm and 1,000 mm. These results were related back to size distribution
curves by assuming an aspect ratio of one, and applying the calculated percentage passing to the respective
block volume. Figure 3.3 presents the BCF distribution curves resulting from varying the input parameters
within the identified limits of the rock mass (i.e., fine, average, and coarse) and the two points derived from
the core photo review. The points estimated from the core photo review predict much finer fragmentation
when compared to the BCF size distribution curve.

3.3 Fines estimation

For the purpose of this study, fines were defined as any material consisting of small pieces in which the
volume of the individual fragments did not exceed 1 cm3. The percentage of fines content was estimated
based on the alteration/hardness ratings, where more intense weakening alteration typically results in an
increase in fines content. The alteration/hardness ratings do not specifically consider the fines content, but
rather on the overall condition and appearance of the core. This resulted in varying percentage fines for core
with the same alteration/hardness rating. A sensitivity analysis approach was adopted to test the potential
range of contained fines, and the upper and lower limits (Table 3.3) were derived from visual assessment
of the core photographs. These represent the likely maximum and minimum percentage fines material for
each alteration/hardness rating.

Table 3.3 Estimated fine material associated with alteration/hardness rating

Alteration/Hardness Lower Fine Material Upper Fine Material
Rating Limit Limit
1 0% 0%
2 0% 5%
3 5% 10%
4 10% 50%
5 50% 90%


Figure 3.3 Comparison between BCF and core photo review predicted size distribution

The calculation of the expected fines was similar to the calculation of core size distribution. The upper
and lower fine material limits were combined with the alteration/hardness ratings distribution to calculate
the fine material volume. The expected amount of fines for each of the alteration/hardness ratings is a
percentage of the total core volume (Figure 3.3).

Figure 3.3 Fine material distribution based on alteration/hardness ratings

Caving 2014, Santiago, Chile

3.4 Further evaluation

Incorporation of the fragmentation ratings into the previously established geological model prevented the
evaluation from being an isolated product. Each drilling interval contained data on the rock and alteration
type, alteration intensity, fragmentation ratings, and spatial locations. Review of this information highlighted
the relationship between the different rock properties and the expected fragmentation. Block modelling the
fragmentation data creates a three-dimensional representation of the expected size distribution and fines
content throughout the cave footprint, which can better manage potential bias resulting from drill hole
spacing, orientation, and depth. Improved fragmentation distributions can be predicted for the entire mine,
individual horizons, or specific locations within cave areas.

4 Conclusions
A visual evaluation of core photographs was used to re-log drill core to determine both the brokenness/
breakability and the alteration/hardness of the rock. The observations were then applied to a rating system
to determine the expected fragmentation. The percentage of material exceeding the specified size limits was
based on the rating of the individual core boxes and the generalized physical characteristics of the specific
brokenness/breakability classification. The method estimates fragmentation from the condition of actual
core rather than an empirical estimate based on generalized properties of the rock mass.

The re-evaluation outcome provided size distribution point estimates for comparison with the BCF analysis.
The detailed review of the core after drilling and handling indicated the fragmentation predictions made by
the BCF software may be biased towards coarse fragmentation. Upon comparison of the two analyses
(photo re-logging and BCF), the fragmentation results determined from the photo re-logging are expected
to be a more reliable predictor of secondary fragmentation in a well-established cave within a similar weak
rock mass.

The authors would like to thank Messrs. Chris Page and Jarek Jakubec for their input during the
development and review of the alternative approach to fragmentation estimation. As well, thank you to
Gabriel Esterhuizen for his availability to discuss the BCF program.

Bieniawski, ZT 1989, Engineering rock mass classifications, New York: Wiley.
Esterhuizen, GS, BCF Version 3.04 A Program to Predict Block Cave Fragmentation - Technical Reference
and Users Guide, 2005.
Esterhuizen, GS, Personal Communication, April, 2013.
Heidbach, O, Tingay, M, Barth, A, Reinecker, J, Kurfe, D, and Mller, B, 2008, The World Stress Map
database release 2008 doi:10.1594/GFZ.WSM. Rel2008.
Laubscher, DH 1990, A geomechanics classification system for the rating of rockmass in mine design,
Journal of the South African Institute of Mining and Metallurgy, 90, pp. 257-273.
Laubscher, DH & Jakubec, J 2000, The IRMR/MRMR Rock Mass Classification System for Jointed Rock
Masses, SME 2000.
Nicholas, DE, & Srikant, A 2004, Assessment of primary fragmentation from drill core data, In
Proceedings of MassMin 2004, A. Karzulovic & M.A. Alfaro (Eds.), Santiago, Chile: Instituto
de Ingenieros de Chile, pp. 55-58.

Caving 2014, Santiago, Chile

Future Projects

Future Projects

Block Caving using Macro Blocks

S Fuentes Codelco, Chile
F Villegas Codelco, Chile

This paper presents the Block Caving Macro Blocks concepts and genesis. Chuquicamata Underground
Project has considered using this configuration as an underground exploitation method in order to improve
the management of mining operations and control of geomechanical problems.

1 Introduction
After more than 100 years of operations, the Chuquicamata open pit will reach the end of its economic
life at the end of this decade. However, the geological exploration program that Codelco undertook some
years ago showed a large amount of remaining resources beneath the final shell pit. Due to the great depth,
experience suggested that the only feasible exploitation method could be an underground operation.

The Chuquicamata Division commenced studies in early 2000 to assess the technical feasibility and
economic potential of a massive underground mining operation, which could maintain the historical
production level. As a result, the Chuquicamata Underground Mine Project has been designed to recover
approximately 1,760 million tonnes of ore, with an average ore grade of 0.71% Cu, 512 ppm of Mo and 492
ppm of As, over a 39-year time horizon, preceded by a period of 8 years of construction and commissioning.
Today, the feasibility study is completed. The construction of the permanent infrastructure (main access
tunnels, intake tunnels, exhaust and shafts) commenced in 2012. The projects master schedule considers
intensive drifting and construction until 2018, followed by a 7-year ramp-up period to achieve the 140,000
tonnes/day design capacity.

1.1 Project location

The Chuquicamata Mine is located 1,240 kilometers north of Santiago, the Chilean capital, at 2,900 meters
above sea level. The site is very close to the city of Calama; it can be reached by highway and the nearest
airport, Aeropuerto El Loa, is only 20 kilometers (Figure 1).

The mine is located in the heart of one of the most important copper-producing districts in the world. It
started its current operations in 1910, although the high quality ore deposit has been well-known since pre-
Hispanic times.

2 Block caving macro blocks origin

The concept of Macro Blocks was analyzed and developed as an extension of the classical Block Caving
method, incorporating the latest of the Codelcos experiences in Block and Panel Caving exploitation,
especially, in the mining operations management, geomechanics and ore body geometry related topics.

2.1 Necessity of new production area

The Chuquicamata Underground Mine planning considers a high production rate, 140,000 tonnes/day. To
achieve this target, it is necessary to prepare a very large area of 102,000 m2 (400 draw-points) for the first

Caving 2014, Santiago, Chile

productive year and an average of 70,000 m2/year (270 draw-points/year) for the rest of the production
horizon (Figure 2).

Considering these huge preparation requirements and the experience at El Teniente Mine on how to deal with
the interferences produced by simultaneous preparation and exploitation activities (Araneda & Sougarret
2007), the project team analyzed and defined the implementation of an exploitation configuration method,
which facilitates the management of interferences during the mining cycle. This method divides or separates
the area under development from the area being undercut/blasted and from the area under exploitation. Each
area is independent and is called a Macro Block (Figure 3). Additionally, this productive configuration is
perfectly suitable for the geometry of the Chuquicamata ore deposit (long and thin).

Figure 1 Project location Chuquicamata Underground

Figure 2 New production and drawpoints number

Future Projects

Figure 3 Macro block method (Fuentes 2009)

Figure 4 shows the sequencing of the mining activities, as considered by the Chuquicamata Underground
Mine Projects, using the Macro Blocks (MB) configuration. The central MB is in its productive stage, while
the surrounding Macro Blocks have initiated the undercutting process.

It appears to be an easy definition, but actually the discussion with internal and external experts was hard
and took a very long time to reach an agreement. Today, Codelco is starting to mine many other areas
with this modular configuration approach, following the same principles considered for the Macro Blocks

Figure 4 Macro block method in Chuquicamata underground

2.2 Management of geomechanical problems

The main geomechanical problem in Block and Panel Caving is the production level collapse. The most
used strategy to resolve this type of issue has been to abandon the collapsed area and re-initiate the caving

Caving 2014, Santiago, Chile

(Pardo & Villascusa 2012), which means to leave a separation pillar between the affected area and fresh
production area (Figure 5).

This concept has been incorporated into the MB design. A 30 m pillar has been left between each MB in
order to prevent any geomechanical problem and, also, to have the option to isolate this sector to continue
with the exploitation. Pillar dimensioning criteria consider the estimation of the local abutment stress,
which means that the pillar is big enough to avoid the effects of the abutment stress generated by the
previous Macro Block on the production drifts located in the following MB in the mining sequence.

Figure 5 Boundaries of production drifts collapsed 2001 2010

3 Macro Block, an exploitation unit

In order to maximize the productivity of each MB, each MB was defined as an independent exploitation
unit with its own ore passes, crushing and accessing system. In other words, each MB is considered as an
independent mine that produces mineral and delivers it to the main transportation system.

In addition, the Chuquicamata Underground Mine Project has considered the pre-conditioning of the rock
column in each MB to improve the caving propagation and extraction, maximizing the production capacity
of each MB system.

Figure 6 shows an MB unit. Each MB should have enough area to initiate a new caving, and also, due to
the possible dilution of the neighboring blocks, each new MB has considered a special draw control policy,
based on a strict planning and production operation, to avoid or control the possible lateral dilution.

4 Conclusions
The outcomes of the analysis of using Macro Block in Block Caving operations can be summarized as

Block Caving using Macro Block configuration as mining method gives flexibility in production
planning, development of mining infrastructure and other operations, improving the likelihood of

Future Projects

Figure 6 Macro Block exploitation unit, Chuquicamata Underground Project

Macro Blocks are suitable for the geometry of the Chuquicamatas ore body (long and thin).

The modular design allows the incorporation of technological changes more easily.

Major collapses in the production areas could be easily isolated.

The activities of development, construction, and undercutting are separated from the production
processes with the possibility of achieving an improved productivity compared to all other caving

The authors would like to thank Codelco Chile for sponsoring the presentation of this paper and all those
who helped us in some way to properly write this paper.

Araneda, O & Sougarret, A 2007, Lessons learned in cave mining: 1997 - 2007, International Symposium
on Block and Sub-Level Caving Cave Ming Keynote address. The Southern African Institute
of Mining and Metallurgy, South Africa, pp. 57-71.

Chitombo, GT 2010, Cave mining - 16 years after Laubschers 1994 paper Cave mining state of the art,
Sustainable Minerals Institute The University of Queensland, Australia, Perth, pp. 45-61.

Fuentes, S 2009, Key Decisions Document. Pre-feasibility study, Vice-presidency of project Codelco-
Chile. (Internal Report, Codelco Chile).

Jofre, J. Yanez, P & Ferguson, G 2000, Evolution in panel caving underground and drawbell excavation El
Teniente Mine, MassMin 2000 Brisbane, Australia, pp. 249 260.

Caving 2014, Santiago, Chile

Laubscher, DH 1994, Cave mining - The state of art, The Journal of the South African Institute of Mining
and Metallurgy.

Moss, A, Diachenko, S & Townsend, P 2006, Interaction between the block cave and the pit slopes at
Palabora Mine In Stability of Rock Slopes in Open Pit Mining and Civil Engineering Situations,
Johannesburg, SAIMM, Symposium Series S44, pp. 399410.

Pardo, C & Villascusa, H 2012, Methodology for back analysis of intensive rock mass damage at the
Teniente Mine, 6th International Conference & Exhibition on Mass Mining, Sudbury, Canada.

Future Projects

La Encantada: An inclined cave design

J Valencia NCL Ingeniera y Construccin, Chile

P Paredes NCL Ingeniera y Construccin, Chile
F Macas First Majestic Silver Corporation, Mexico

Massive caving methods offer low cost and high productivity alternatives when they are able to fit the
orebodys geometry and geomechanical characteristics. Nevertheless, when the orebody does not fulfil the
typical caveable characteristics and stopping methods are not applicable, difficulties appear in finding a
caving method that suits the orebodys geometry and competence. An alternative to this problem is the use
of inclined caving methods.

This paper presents the methodology and main results of the preliminary study for the exploitation of the
Breccias sector in First Majestics La Encantada mine. Some parts of this sector have been previously
mined with Cut & Fill methods, leaving several excavations in the orebody. Difficulties related to the
low competence of the rockmass and lower grades have led to the need of exploring alternatives to the
traditional Cut & Fill mining methods. Attending the orebodys geometry in the sector, the use of low cost
and non selective mining methods, such as caving methods, has been considered.

The use of sublevel caving implicates the construction of several excavations in the orebody, attending
to the orebodys conditions related to its low competence and representing the higher costs within the
caving methods. Two horizontal Block Caving layouts were proposed: (1) a 3.5 yd3 LHD operated offset
herringbone layout and (2) a scrapper operated regular layout. Attending to recovery considerations,
both layouts consider a small drawpoint spacing, which results in small pillars that would cause stability
issues and high excavation density. Thus, an inclined caving layout is proposed, solving recovery, costs,
productivity and stability problems.

1 Introduction
La Encantada Silver Mine, from First Majestic Silver Corporation (FMS), consists of silver/lead/zinc
oxidized mineral deposits located in the State of Coahuila, Mxico, 708 kms northeast of Torreon (Figure
1). The mine comprises numerous mineral concentrations within the underground development area,
including some exhausted deposits and additional geologic potential in other areas.

The Breccias Sector consists of two breccias (Milagros and San Javier) and a partly mineralized magmatic
intrusive. This sector has been mined by FMS and previous owners using selective mining methods and
contains, therefore, several remaining tunnels and mine openings (Figure 2). Difficulties related to the low
rock mass competence have led FMS to seek an alternative for the current cut & fill methods applied in
the mine. Attending the massive shape of the orebody in the sector and lower grades, FMS has considered
the use of non-selective massive mining methods, such as, caving methods. The following paragraphs
describe the caving alternatives and the final inclined cave design proposed by NCL for La Encantada
mines Breccias Sector.

Caving 2014, Santiago, Chile

Figure 1 La Encantada mine site location

2 Background and considerations

The Breccias sector has the following characteristics (Table 1):

Massive shaped and low silver graded orebody.

Low rockmass competence in the Breccias units (MRMR ranging from 21 to 34).
Regular to fair rockmass competence in the foot and hanging walls.
Orebody limited by topography, no waste overload.

Table 1 Geotechnical parameters for the geological units

RMR Bieniawski MRMR Laubscher

Milagros Breccia 39 21
San Javier Breccia 49 23
Milagros Intrusive 60 34
Limestone 61 35

Figure 2 a) Isometric view of remaining mine openings and orebodies (in red San Javier Breccia, in blue
Milagros Intrusive and in green Milagros Breccia); b) schematic plan view of the geological units

Future Projects

These characteristics allow the use of caving methods given that:

Self-supported methods require high rockmass competence in the orebody, in order to obtain fair

Cut & Fill methods are oriented to high graded orebodies, due to their high costs and low

Attending the above mentioned, sublevel and block caving designs were proposed for the sector in order to
define which method better complied the following:

Represent a low cost and high productivity alternative.

Minimize the volume of mine openings in the low rockmass competence.

3 Method selection

3.1 Sublevel caving alternative

The Sublevel caving method (SLC) consists of drilling and blasting the orebody in several superimposed
levels and of caving. As a consequence, the waste overload from the roof and hanging wall occurs. This
method has, therefore, an intensive excavation density in the orebody and, in the particular case of La
Encantada mine, would implicate safety hazards for both people and equipment. On the other hand, the
cost of a SLC alternative is higher than a Block Caving (BC) operation. Thus, due to safety hazards, related
to the excavation density in the orebody, the fact that the operational cost of a SLC is the highest among
caving methods and higher dilution potential; the SLC alternative was discarded for the Breccias sector.

3.2 Block caving alternatives

Block caving and its variants consist of generating a drawpoint base at the bottom of the orebody and
undercutting its base in order to allow natural caving of the rockmass. Considering the previously exposed
characteristics of the deposit, this method was selected due to the following reasons:

The general shape, grade distribution and structural characteristics of the deposit fit the main
characteristics of the caving (undercutting) variants.

Minimizes the volume of excavations inside the orebody, reducing them to only the extraction
and undercut levels.

Represents the lowest cost and highest productivity option.

Considering the rockmass classification (Table 1), and Laubschers abacus for drawpoint spacing (Figure
3), the suggested spacing for a horizontal BC layout should be between 7 and 13 m, considering a 3 m
drawpoint width. Taking into account the fact that the MRMR of the Breccias orebodies (San Javier and
Milagros) are in the lower limit of the rock mass class (4), a horizontal 10 m x 10 m layout was proposed,
considering two tramming options:

1. Using 3.5 yd3 LHD, which is the largest loader that fits the geometry

2. Using scrappers

Caving 2014, Santiago, Chile

Figure 3 Laubschers abacus for drawpoint spacing (Laubscher 1994)

Figure 4 Horizontal layout using LHD

Figure 5 Horizontal layout using scrappers

Future Projects

Figure 4 and 5 show the horizontal layout design for LHD and scrappers, respectively. It is possible to
appreciate that high excavation density is needed to satisfy a 10 m x 10 m spacing draw pattern. Previously
mentioned stability problems in the excavations of the Breccias sector make the horizontal BC layouts
a high risk alternative in terms of the stability, due to the high excavation density that they implicate.
Therefore, the horizontal layout BC alternatives were also discarded for the Breccias Sector.

Once the typical caving variants had been discarded, the evaluation of an inclined cave layout was considered,
due to the higher robustness that it represents, by reducing the excavation density at the same elevation,
using several levels to build drawpoints. Despite the fact this is not a common arrangement, inclined cave
alternatives have been proposed and implemented in several mining countries, such as, Australia, Canada,
South Africa and Zimbabwe (Jakubec 1992; Carew 1992; Hangweg et al. 2004, Laubscher & Jakubec
2000; Jakubec & Laubscher 2012) resulting in successful experiences in some cases.

Figure 6 shows the inclined cave layout proposed for La Encantada mine. This layout was selected due
to the lower excavation density per level and, therefore, the higher stability that it represents. It is worth
noting that, in contrast to the conventional BC, this design does not consider an undercut level. This relies
on the fact that the poor rockmass quality would enable the orebody to cave with a long drawbell blasting

Figure 6 Inclined cave layout

4 Mine Design
A single lift inclined cave design was proposed for the deposit using the internal tool Block Cave for
footprint elevation definition, based on Laubschers vertical mixing algorithm (Laubscher 1994). Figure 6
shows the general mine design, which considers four drawpoints lifts separated 10 meters in vertical and
main drifts at every drawpoint elevation that connect these drawpoints with production drifts. The higher
level is located 300 m below surface. Drawbells of 17 m height connect the different production levels.
Production drifts have a total length of 26 m, 14 m of which are drilled to generate the drawbell. Ore is
dumped in an ore-pass, transferring it to existing haulage levels. Tramming is performed by 3.5 yd3 LHDs
and haulage is divided between a conventional trucking system (800 tpd capacity) and a railway-shaft
system (1,200 tpd capacity). Ventilation shafts located at the north east side inject fresh air from surface at
the end of every main drift.

Caving 2014, Santiago, Chile

Figure 6 Breccias sector general mine design

Figure 7 Breccias sector general mine design isometric views

5 Mining Sequence
An initial area of 4,800 m2 (within a total of 11,200 m2) allows a 17 m Hydraulic Radius (HR) footprint,
which is enough to initiate caving at the mine, considering the caveability assumed for the Breccias sector.
On the other hand, a 2 drawbells per month incorporation rate is proposed. This would allow the Breccias
sector to achieve full production at 2,000 tpd during the second year of mine life. This 2,000 tpd production
rate is limited by the processing plant capacity, further production rate increase could be studied for the
same design if a plant capacity expansion is evaluated. Finally, the mining sequence proposed goes from the
top north-east to the bottom south-west drawpoints, obeying the geometrical restrictions in order to avoid
undercutting a non-opened drawpoint.

Future Projects

6 Industrial scale test

In order to identify possible risks and further issues without significantly compromising capital and mine
production, an industrial scale test is proposed for the inclined cave design. This test consists of building
2 drawpoints in the upper level and operate them for a given period of time in order to capture as much
experience as possible to be used in the engineering project for the rest of the area. This test is not intended
to produce any caving at all, but it will generate valuable information related with the construction of the
drawpoints, support requirements and drilling and blasting procedures for undercutting, providing FMS
with some experience in the subject.

Figure 8 Breccias sector industrial scale test proposal plan view

7 Conclusions
Inclined caving methods represent a viable solution to low graded deposits with weak rock massesl as they
are a flexible alternative when traditional caving methods implementation presents technical difficulties.
The use of an inclined scheme allows defining an adequate drawpoints pattern and at the same time account
for a reduced excavations density in a single level, improving stability conditions. In particular, Breccias
sectors technical challenge can be overcome by a simple solution using inclined caving methods. The
implementation of an industrial scale test would be of great utility to improve both technical and economical
information in order to develop the project. Finally, the use of an inclined caving layout represents a valid
alternative for the application of caving methods in medium scale mining.

Carew, TJ 1992, Footwall drawpoint caving at Cassiar Mine: In proceedings MASSMIN 92, Johannesburg,
South Africa, pp. 295-301.

Hannweg, L et al. 2004, Koffiefontein mine front cave Case History: In proceedings MASSMIN 2004,
Santiago. Chile, pp 393-396.

Jakubec, J 1992, Support at Cassiar underground mine: In proceedings MASSMIN 92, Johannesburg,
South Africa, pp. 111-123.

Janelid, I 1978, Method for mining of rock or ore according to the block caving principle in massive
formations. U.S. Patent 4,072,352.

Caving 2014, Santiago, Chile

Laubscher, DH 1994, Cave mining - the state of the art, The Journal of the South African Institute of
Mining and Metallurgy, vol 94, N 10, pp. 279-293.

Laubscher, DH 2012, Incline Cave Mining A Viable Alternative to Horizontal Layout: In proceedings
MASSMIN 2012, Sudbury, Canada.

Laubscher, DH & Jakubec, J 2000, Block Caving Manual Incline Cave, ICS 2000 internal document.

NCL SpA 2013, Proyecto de explotacin sector brechas, Mina La Encantada. Estudio preliminar, Study
report for First Majestic Silver Corporation (In Spanish).

Caving 2014, Santiago, Chile

Geomechanic Design

Geomechanic Design

Considerations for designing a geomechanics monitoring

plan for each engineering stage
AE Espinosa Codelco, Chile
P Jorquiera, Codelco, Chile
J Gltzl, Gltzl GmbH, Germany

Frequently, plans for monitoring mining geomechanics are designed with the aim of measuring deformations
or displacements that allow early identification of the onset of potential instability. This approach is suitable
for the control of risks pertaining to the field of geomechanics. For this purpose a methodological support
is necessary to link with a systemic functional approach, the process from the definition of purpose to the
performance evaluation and compliance targets.

From the experience of the last ten years in the development of implementation plans and monitoring
geomechanics in the El Teniente Division of CODELCO - Chile, this paper proposes a methodology that
guides the development of a plan for implementation and monitoring geomechanics. The methodology
applies particularly for each project with a clear focus on the applicability of the records obtained in stages
of conceptual design, functional implementation, procurement start to evaluation of results and fulfillment
of objectives. All this will be done to finally close the loop with a stage design which fits oriented to the
utility for operation over the life of the mine.

The methodology proposed here uses, as a structure for defining purposes and objectives, the different
stages of the mining project engineering, ranging from all engineering stages and then binding steps with
the start of production operations. Finally the result is a map that identifies the processes required to
develop an implementation plan and monitoring geomechanics to be considered as a necessary requirement
and functional utility for the safe performance of mining operations activities.

1 Introduction
Normally, in relation to the geomechanical monitoring in underground mines, the plans, in such technical
application, are indicated as tools to control losses caused by geomechanical instabilities which occur along
the mining working. For the appropriate design, the objective has been mainly centered in the acquisition of
proper data. The main focus should be on the use and the evaluation instead of just collecting data.

One paradoxical aspect is that the design is directed form a model of behavior previously estimated and
the expectation resides in the confirmation of the assumptions of the original design. If this occurs, it
necessarily would be an invitation to modify the model in which such design was elaborated. However,
understanding that monitoring (as a control) is only made once the processes are initiated, this could mean
to implement modifications.

In that way, the design/setup of a geotechnical instrumentation and monitoring plan is a constant process,
in which the main requirement is to define the main purpose for each engineering stage, considering
the geotechnical, geomechanical mining designs and most of all the tolerance records, according to the
associated geomechanical risks.

Caving 2014, Santiago, Chile

The development of this working action plan respectively project has been motivated by the final results
obtained by evaluation of monitoring of geomechanical plans, implemented in the last ten years in Divisin

In that period it was possible to identify that the focus on the design is very helpful, getting very detailed
records which can be used and applied in the mining process. The previous consideration has driven to
check the purpose of the implementation of these monitoring and geomechanical systems as well as its
contribution to the mining development.

This project proposes the considerations to elaborate a methodology that can guide properly toward the
development of a geotechnical instrumentation, monitoring and in the end take geomechanical actions.
This can be done for each individual case, with a clear guidance to the use and application of the different
registrated files obtained. For the control of geomechanical risks or losses (also developed among the stages
of conceptual and detailing engineering) properly evaluated after the different mining jobs.

The geomechanical instrumentation in underground mines and particularly the available experience in El
Teniente de Codelco Chile, has played a major role as a supporting tool for the understanding of caving
processes in general, with regards to the comprehension of mechanical behavior of the surrounding massif
and exploitation.

However, applications on decisions in an emergency were more an instance inspired by previous experiences
than using actual measurements. In principal this situation arose from historical cases and had an unknown
consequence. The results simply need to be linked to the records with the final conclusions and also with
unknown conditions along the different stages of the project and more important at the start of production.

For this diagnosis it appears as a relevant fact, to become properly acquainted with a structure, allowing
the definition of objectives which direct to the fulfillment of the expectations. According to the stage of
engineering that will be developed during a determinate instance, making geomechanical monitoring
coherent with mining design. Also taking into account the geomechanical vulnerability implied to the
exploitation and the plans for the mining development.

A very good example for the application of a system for geomechanical monitoring is the seismic system
available in El Teniente (ISS Mina). Actually here Analysis of the record allows taking concrete decisions
with regards to blasting and the isolation of sectors (for instance, personnel entering and leaving). Besides
the contribution of data for making progress evaluating seismic menacing or dangers, among other relevant
effects it is important for long term planning for the mining development.

At present we are working on a summary of a methodological proposal, to approach functional conceptual

aspects as well as an instrumentation and geomechanical monitoring plan, with the general purpose of
including it in different stages of engineering, according to the project expectations and the contribution to
control geomechanical risks.

At least four relevant stages are considered in the project:

IP&GM expectations
Method to focus on points of each engineering stage
Designs for engineering and ground implementations
Evaluation of the results and fulfillment of the expectations
From this perspective design and evaluation stages will occur, but previously expectations need to be
defined that allow a decision carry out a IP&GM. All this implies that geomechanical monitoring is a

Geomechanic Design

means but not the purpose. In the end it depends on the project owner, his personal attitude to face the risk
and the available alternatives and measurements in the particular case.

1.1 Background consideration

Checking the background that contain the different plans of geomechanical instrumentation and monitoring,
developed by the mine of EL TENIENTE de CODELCO CHILE in the last twelve years, it shows an
emphasis orientated to get measurements mostly with regard to the application. This is evident in the
limited documents which lead to the post evaluation and due to this actions were imposed by the results of
the application of geomechanical monitoring plans.

2 Methodological development
Following this exposition the methodological focus including each stage will be described, in order to
elaborate an Instrumentation Plan and Geomechanical Monitoring (IP&GM), that illustrates its contribution
to value the mining process and that inserts itself in each one of the four geomechanical engineering stages.

2.1 Expectations of IP&GM for each engineering stage

The methodology considers one first step where the expectations for each engineering stage will be indicated.
These expectations are reduced to the following definitions: Purposes, Objectives, Goals and Products.

Table 1 Expectations on the IP&GM

Profile Pre-feasibility Feasibility Details

Define monitoring
Identify requirements
Estimate costs of IP Insert into the
Purpose geomechanical according to the
&GM mining plan
potential risks geomechanical
Determine type,
Develop work plan Size requirements Develop location
amount and use of
objective (time and costs) for and evaluate plans and monitoring
instruments (CAPEX
the next steps technologies plans
and OPEX)
Consider available
technologies and Design with technologies commercially
GOAL shorter range of available and accessible in terms defined by
and background
innovation engineering

Report of Report to size Planes with all

Design and implement
descriptive requirements in locations and
Product a plan for monitoring
scenarios and time and costs as monitoring
potential risks technologies frequency

Caving 2014, Santiago, Chile

2.2 Methodological design for each engineering stage

The fundamental objective of this work is to materialize tasks, activities and/or concrete products. The
aspects that are going to be developed in each engineering stage and the goal to implement an instrumentation
and geotechnical monitoring plan have a valuable contribution to the mining process.

2.2.1 Profile of engineering stage

This stage defines the business potential, it describes the principal risks in qualitative terms and uses bench
marking records for the IP&GM. This is necessary to get an overview of the level of costs and the magnitude
of works, using similar experiences and present identified risks of mayor relevance.

Table 2 Key contents for IP&GM in the Profile engineering stage

Process IP&GM Engineering stage

1.- review of
available: It consists
of a compilation of
instrumentation records
made around the area of
interest or comparable
geomechanical conditions
(geotechnical, mining,
environment stress). This
Delivery details

background is useful to
support new requirements
if necessary.
2.- Budget and business
Must be considered
costs associated with the
construction of conceptual
geomechanical model and
plan that the proposed
implementation is a
product that is made after
the availability of the
Offer process: Background required

Start a plan for developing a system to monitor the impact of 1.- Geographical location and
geomechanical instabilities, adding value to define the true timing of the operation.
dimension of the requirements that meet the stated objective. 2.- Security Policy, Standard
fatalities control and risk

Geomechanic Design

2.2.2 Pre-feasibility engineering stage

In this stage different exploitation alternatives, mining designs and stages of exploitation are evaluated.
The work ends with the selection of the best alternative that will be focused in the following stage. The
instrumentation and geomechanical monitoring plan has been defined conceptually according to the
technology and the type of measurement normally used for that kind of activity.

Extensometers and periodical measurements in a manual way.

Local deformation measurements according to non-systematic requirements.
Seismic activity records, remote and continuous automatic monitoring

2.2.3 Feasibility engineering stage

For each useful stage a definition about the method and the dimension of the exploitable sector is available.
It corresponds to development of parameters that allows elaborating the mining plan. For this stage, where
the temporality is defined and where the constructions are realized, results are very important to incorporate
the installation works that develop mining activity.

Table 3 Key contents for IP&GM in prefeasibility engineering stage

Process IP&GM to prefeasibility engineering

1. - Report with
instrumentation plan that
incorporates definitions: the
kind of device and the amount
estimates (in the range of 25%)
2.- Report with the qualitative
Delivery details

assessment of the expected

risk control: is defined the
scope and range of functionality
for design, should be clearly
established for what is the
3.- Report with the evaluation
of the expectation value
contribution of IP&GM in
mining development indicators
are definite to assess the
effectiveness of the design
Offer process: Background required

Identify relevant aspects of the expected geomechanical 1.- Mining method.

behavior and incorporate them into a rational design of 2.- Sizing of mining

Caving 2014, Santiago, Chile

2.2.4 Detailed engineering stage

In this stage plans for installation are elaborated as well as technical bases for allowances (permissions).
Practical ground plans are designed and the strategy to use the available resources in construction and
different developments (availability for perforations, constructions, electric installations).

3 Performance Indicators
As a manner to identify the necessity of modifications and corrections of the parameters that define
instrumentation, monitoring and record analysis it is proposed to use indicators that inform about efficacy
and the use of the system. Like any evaluator to give the final value to the final product of the system. The
premises for the works on the comportments of the geomechanical monitoring systems are:

Matrix results: the options of the final results are reduced to four stages that depend on whether the
alert was right or not.

Modifications of the instrumental monitoring parameters amend the original costs of the system.
Quantifications of these modifications are an available economic indicator of the system.

Table 4 Possibilities on the final results of the geomechanical system

Consequences No consequences

Alert Success Minor fail (cost)

No Alert Major Fail (Safety) Monitoring

Modifications of the system are realized fundamentally on the base of acceptable criteria of the negative/
positive decisions registered. In this sense the cost indicator depends on the system work. Regarding the
operative implementation, each case will be specially analyzed in order to take decisions on the modification
of the available system.

4 Conclusions
The described methodology allows a structured focus headed to identify the exact values contributed
by a system for a geomechanical monitoring mining project.

The inclusion of appropriate indicators, gives great reliability of the present initiative.
The advantages of identifying the role of a geomechanical monitoring system for mining processes
allows to plan deliverable products, value engineering at each stage, satisfying expectations from
technical and also economic perspectives.

Geomechanic Design

Table 5 Key contents of IP&MG in feasible engineering stage.

Process PI&GM to engineering feasibility

Report (conceptual Plan)
description the kind
of the geomechanical
instrumentation for
monitoring excavation and
for control of the process of
mining on the rock mass.
Report (Instrumentation)

Product details
technical specifications
and cost estimate for the
purchase and installation of
the required instruments.

Report (Monitoring)
definitions of the frequency
of measurements, data
analysis, threshold values
and actions in case of
deviations in the expected
Offer process: Background required

Design geomechanical monitoring system for the control of major 1.- Geomechanical Conceptual
excavations (caves) and for control over the response of the rock model
mass against the advance of mining 2.- Plan developments in mining

Espinosa, Cornejo, Fuentes 2012, Geomecnica proyecto Dacita Enlace ingenieras bsica y detalles,

Morrison, RGK 1976, A philosophy of ground control, Department of mining and metallurgical
engineering McGill University, Montreal.

Caving 2014, Santiago, Chile

Integrated support quality system at El Teniente Mine

MS Celis, Codelco, Chile
RA Parraguez, Codelco, Chile
E Rojas, Codelco Chile, Chile

The improvement of the design and installation of underground support requires an integrated quality
control system. It rises from the relevance of the support quality in its response to loadings associated
to mining (especially rock burst and collapses), as support quality can achieve a better damage control,
protecting personnel and mining infrastructure.

The integrated quality support system has been implemented during year 2013 and it is based on the legal
framework. It considers 3 main aspects: design - monitoring - post evaluation.

Ground control engineers check the quality of the installed support and compare it to accepted standards.
Considering the limited resources, monitoring is focused in some critical areas defined according to seismic
activity, mining and stress field. Technical reports have been prepared including the main aspects surveyed
in the field for each support system. Post-evaluation reports are generated including lessons that should
feedback designs, establishing a support improvement cycle. The resulting information is monthly sent to
the areas of interest (Critical Risks, Mine Operations, construction companies). Depending on the critical
level of findings causality analysis or corrective actions were undertaken.

1 Introduction
In Tenientes mine the Geomechanics Superintendence has the following mission:

To contribute to maximization of economics value in long term of El Teniente Division and Corporation,
support mining explotation with Geomechanical application, with emphasis in rock burst risk control.

To control rock burst risk, 3 points are involved:

Source: mining control or rock mass pre conditioning can reduce seismic event magnitude.
Damage control: installation of support with better response to dynamics loadings can control the
damages level in a better way.

Personnel exposition: definition of exclusion criteria, abutment stress zone and use of remote-
controlled equipments looking for reducing exposition of personal and equipment in higher risk

In accordance with that, we deduce the importance of an Integrated Support Quality System, considering the
relevance of the support quality in its response to loadings associated to mining and because its necessary
to improve our management of findings and interactions with others Geomechanics areas.

Geomechanic Design

2 Component of System
The integrated Support Quality Index (ISQS) includes the main components as indicated in Figure 1:

DESIGN considers as inputs the mining design variables: gallery sizes, lifespan, use, dynamic loadings,
corrosion, lithological and structural conditions, stress field among others. This data is used to produce
drawings, calculation logs, technical specifications, quality standards. Those products are the inputs for
monitoring support behavior.

MONITORING: in addition to the survey of the installed support elements, it includes also the testing
of other innovative support systems that could be used in some specific underground conditions and daily
solutions to operational requirements for underground singularities.

POST EVALUATION. It allows to put together information notes, loading characteristics (rock bursts,
collapses, abutments stress levels, falling wedges) and the expected support behavior. Post evaluation notes
includes the lessons that should feedback designs, establishing a support improvement cycle.

Figure 1 Components of the Integrated Support Quality Index (ISQS)

The 3 components, design, monitoring and post evaluation, define a working cycle. It includes continuous
improvement of designs. But besides, it produces an interaction with the Critical Risk area responsible for
the management of areas considered critical due to their impact in the personnels safety and for the mining
business. This allow implementing corrective actions in a faster and effective way.

3 Monitoring

3.1 Definition of attention focus

Ground control engineers check the quality of the installed support and compare it to accepted standards.
Considering the limited resources, monitoring is focused in critical areas that have been defined as seismic

Caving 2014, Santiago, Chile

activity, mining and stress field. Notes are prepared including the main aspects surveyed in the field for
each support system.

Figure 2 Parameters to define attention focus

Abutment Zone: Zone in the vicinity of caving front, where rock mass show evidences of the concentration,
variation and rotation of stresses. The width of this zone is defined for each sector depending on the applied
caving method and geomechanical and geotechnical conditions.

Table 1 Analysis criteria for Abutment Zone

Variable Evaluation


to include
ZT ahead of
extraction limit

Energy Index: ratio between event radiated energy v/s expected radiated energy. EI is calculated for each
event (Mendecki, 1997). Besides for seismic activity the criteria involve cluster seismic magnitude (Table
2 and 3).

Geomechanic Design

Table 2. Analysis for Energy Index

Variable Evaluation

Criteria: to
include zones with

Table 3 Analysis criteria for Seismic Activity

Variable Evaluation

Criteria: to include
event cluster
magnitude 0,5 in
last year

Fractures Pressures: characterization with geostatistical models of the spatial distribution of propagation
fracture pressures induced by hydraulic fractures. In this case the critical are defined as these that reaches
more than 30 MPa of pressure to break a fracture (Table 4).

Caving 2014, Santiago, Chile

Table 4 Analysis criteria for Fractures Pressures

Variable Evaluation


to include fractures
pressures greater
than 30 MPa

Peak Particle Velocity: maximum vibration velocity estimated from seismic sensor records. In this case a
critical zone is defined as these where the ppv reaches xx mm/s as shown in Table 5.

Table 5 Analysis criteria for Peak Particles Velocity

Variable Evaluation

Criteria: to
include the
greatest values
of PPV

Geomechanic Design

The above indexes are considered to define a critical zone when all criteria are meet. That is when conditions
described from Table 1 to 5 are reached.

Figure 3 Identification of Attention Focus

3.2 Survey Information

Technical reports are prepared including the main aspects surveyed in the field for each support system
(survey date, name of geomechanical engineer, identification of evaluation site, contractor company
responsible of the site and evaluated items).

4 Management of Critical Findings

Geomechanics group send monthly all the surveyed information. Depending on the critical level of findings,
Critical Risk area request causality analysis or corrective actions that have to be sent to Mine Operations
or Construction Companies.

Reports include critical findings with their impact in the productive process, verification of correctives
actions obtained from these findings and inform the unfulfilment of support system quality of surveyed
galleries (summary of technical reports).

Caving 2014, Santiago, Chile

This is the evaluation scale:

: Agree with the design : Possibility of improvement : Major flaw

Table 6. Example of Technical report. Bolt plate nut and mesh system
Technical Report Evaluation Point % compliance
Bolt length agree with design 100 98-99 0-97
Spacing between bolts agree with
100 86-99 0-85
All bolts installed with plate and nut 100 98-99 0-97
Plate in correct position (relative to
100 <100
Good connection between plate, nut,
100 <100
bolt and mesh
30 cm maximum height from floor to
100 <100
first bolt
Saw grouting outside drilling 100 90-99 0-89

Mesh without corrosion 100 <100

Overlap mesh with one bolt line 100 <100

Mesh from floor to floor 100 90-99 0-89

Intact mesh without rock slabs 100 <100

Figure 4 Actual reduction of major flaw

Geomechanic Design

Figure 5 2013 Evaluation sites in Production Level (Esmeralda, Reno Dacita and Diablo Regimiento sectors)

5 Conclusions
As a conclusion, we have identified some contributions of this Integrated Support Quality System:

It improves the management of findings in areas considered critical due to their impact in the
personnel safety for the mining business, executing corrective actions in a more rapid and effective

It gives simple solution to some operational problems and includes learned lessons of post
evaluation of the support designs.

It allows the same standard for evaluating the installation quality of underground support in all the
Mine sectors.

A database can be created to feed post evaluations.

It is an indicator for management regarding the installed support quality. It is a warning in relation
to practices that impact negatively the quality of installation and motivates the searching of
solutions for them.

Caving 2014, Santiago, Chile

For this year, we expect to extend design and post evaluation aspects, to implement this System in the new
projects of the Division and to structure a computational platform with all the system information.

Cornejo, J, Muoz, A & Rojas, E 2012, Estimacin de presiones de propagacin de hidrofractura en
volmenes pre acondicionados PQ2013, Mina El Teniente, Internal report SGM-I-056/2012.

Dunlop, R, Gaete, S & Rojas, E 1999, Sismicidad inducida y estallidos de roca en Mina El Teniente,
Internal report PL-I-099/99.

Juran, M, Gryna FM 1995, Anlisis y Planeacin de la Calidad, ed. McGraw-Hill.

Mendecki, A 1997, Seismic Monitoring in Mines. Chapman & Hall.

Geomechanic Design

Management indicators for the cave geometry control,

El Teniente mine
J Cornejo Codelco, Chile
C Pardo Codelco, Chile

The fulfillment of production targets for massive cave mining is achieved by a successful application
of cave back management. In this context, the planning tools used to identify the quality of extraction
process are essential. For panel caving operations, the cave back could be managed through the extraction
process and the incorporation of drawbells into production. He main geomechanical hazards, such as,
rock bursts, collapses and early dilution could be related to unfavourable cave geometry. Therefore, the
constructions and follow up of cave back management indicators are essential for an effective planning of
caving operations. In this article, the authors present the results from the application of the angle of draw
as an indicator of the performance of the cave back at Reservas Norte sector of El Teniente. The results
indicate that there is need to improve the draw control over the sector as the indicator is underperforming.
This would help to identify and to reduce the probability of occurrence of major geomechanical hazards.

1 Introduction
Many geotechnical risks have been associated with unfavorable conditions of cave back geometry. Some
rock bursts and collapses experienced at different operations have been facilitated by deviations from
planned draw process. Normally, the geotechnical guidelines set out the requirements that must be met;
however, it is crucial to look for some geometry indicators for evaluating the conceptual models used as
caving rules.

In the present study, management indicators were developed as key process indicators (KPIs), reflecting
the general state of the cavity regarding the geotechnical guidelines. Based on the key process indicators, it
is possible to anticipate geometries issues that increase the plan vulnerabilities reducing the occurrence of
major detentions and risk of personnel exposure.

The cave back geometry would be strongly associated the stress variations around the cave front during
complete extraction steps in any block caving mine. It is crucial to keep a geometry control at any panel
caving, because any deviation could lead to unfavorable conditions.

Figure 1 shows (Flores et al. 2004) representative phases for connecting process by block caving. During
the firsts stages the flat undercut induce an active zone located upper this cavity. This affected volume
of rock mass corresponds to a combination of seismogenic zone and loosening area which induce stress
changes towards edges of the cavity mean cavity progress (Duplancic & Brady 1999).

Although one of the main geotechnical aspects of caving process is the crown pillar breaktrough and
connecting to upper cavities, there are other parts of the process that would be almost as strategic as the
previous ones. The exploitation method considers incorporating area after connections, increasing the
extension of the undercut, advancing with the cave front and increasing the rock mass affected by mining
advance. Therefore, for wider undercutting fronts, such as, at Reservas Norte case, the progressive process
of the breaking the ore column properly becomes a very important issue (Landeros et al. 2012).

Caving 2014, Santiago, Chile

(a) Initial undercutting, the main mechanisms

for propagation is the unconfinement of rock

(b) Cave propagation induces a curve shape

increasing the effect of stress caving.

(c) Additional upward caving propagation

increases the curvature of the cave back,
and makes the stress caving mechanism

Figure 1 Evolution of the cave back and caving mechanisms through time due to the upward propagation of
caving (Flores 2005)

Experience at El Teniente Mine allows conclude that unfavorable caveback geometries increase potential
geotechnical hazards, such as rockbursts, collapses, hang-ups and airblasts and dilution entry. In this
context, there is evidence of seismic activity which has induced rock-bursts and also collapses experienced
in several levels, affecting recovery of reserves and staff safety working in different operations (Landeros
et al. 2012).

In order to identify unfavourable geometry condition of cave back, a number of direct and indirect
methodologies for measuring and/or estimating the cave back shape have been implemented. One of the
most widely used indirect relations is the ratio of extraction surface (angle) with the cave back. This concept
is a simplification defined as the average angle between the effective extraction height and distance to a
point reference to a specific direction of advance (Araneda & Gaete 2004); in Figure 2 a scheme of this
concept is shown.

From the above concept, it is assumed that the extraction rate is linked to the drawbell incorporation rate
by the extraction surface (simplified as an angle). This geometrical relationship between the extraction rate
(Pi+1, Pi, Pi-1) and caving rate is defined by the following equation:

Geomechanic Design

Figure 2 Simplified caving model

= tan(x) (1)

ve : Extraction rate [t*m2/month].

vn : Incorporation rate [m2/month].

x : extraction angle for a drawpoint

Thus, there is a lineal relationship between the drawbell incorporation rate and the extraction rate. Therefore,
in order to get to a constant draw angle, any increase of the extraction rate will have to be accompanied by
an increase on the drawbell incorporation rate.

One of the research consideration is that an accelerated incorporation rate relative to the effective extraction
rate would mean a low extraction angle and a potential unfavorable stress condition around the cavity.
On the other hand, a higher extraction rate with respect to the incorporation rate could induce potential
or dilution (mud or dilution) resulting in loss of reserves. Furthermore, the activation of geological faults
could be facilitated by a higher angle of extraction.

4 Cavity control indicators for Reservas Norte Sector

An indicator could be defined as a number that describes the performance of a specific activity. Key
performance indicators (KPIs) are defined as indicators of strategic significance, which are perceived
as critical under current business circumstances (Tomkins 1988). In this study, a methodology based on
the current practice of calculating profiles extraction was used. This methodology consists of calculating
the height draw at specific reference lines along the mine. In this study, reference planes were used to
calculate the indicators and the state of each point is measured against these planes of reference, these
points are grouped into zones according to the preferential growth of the cavity. In Figure 3, an outline of
the methodology and the description of how to calculate the indicator are shown.

Caving 2014, Santiago, Chile

Figure 3 Plan view of zones defined to calculate the angle of draw (zones 1 to 6)

The criterion to define the categories took into account the influence of the distance from the height of
draw of a given drawpoint to the edge of the cavity, i.e., it was considered that the closer a point on the
edge of the cavity, the greater its influence on stability condition in the surrounding infrastructure, while its
influence decreases as the distance increases. In Figure 4 the categories associated with different conditions
of points with respect to the reference planes, where the critical conditions are given by the category red
and magenta. In this four categories are defined:

1. Low angle of draw, that is when the calculated angle is smaller than (magenta).
2. Good angle of draw, when the draw angle is within - range (green).
3. Intermediate angle of draw (yellow), when the angle od draw is larger than and smaller that a
critical angle (CA).

4. Large angle of draw, this is when angle of draw is larger than the CA (red).
From this classification, sectors under critical conditions were 1) and 4). In these cases, actions are

1) In the case of areas reaching 1), they are considered as priority extraction areas. In this case, the
continuity of the operation suggests to increase the extraction and/or decelerates the incorporation
of new drawbells.

2) In the case of areas reaching 4), these were termed Areas of over extraction. In this case, there
is a need to accelerate the incorporation of new drawbells and not to over-extract drawpoints from
what has been planned.

As example, the methodology was applied in Reservas Norte sector at the El Teniente mine (Figure 5),
where the following features were considered:

Geomechanic Design

Figure 4 Angle of cave back and categories used in this study

The indicator was calculated only for active fronts.

Based on the empirical relation that indicates connection to upper cavities when 33% of column
height of primary rock has been extracted, a distance of 60m for evaluation has been calculated.
This evaluation distance is fixed between the incorporation front (first draw bell incorporated) and
the last point used to calculate the angle of the cave back.

The sector has been divided in zones to improve the interpretation(Table 1 and Figure 5):

a. Panel caving variants.

b. Lithological and structural condition.
c. Column height of primary rock.
It should also be considered that in Reservas Norte there are different mining methods taking place as
indicated in Table 1.

The results of the analyses are indicated in Figure 6 which shows that Zone 1 has the largest numbers of
drawpoints with overdrawn. Zone two and four have the largest number of drawpoints with under drawn.

Finally, through the use of the proposed tools, it is expected to maintain control of geometry cavity,
which allows improving consistently the results of the indicators. In particular, expected to the estimated
compliance from the five-year plan for 2014 through management in the short and medium term, it is
possible to reach levels that allow non-stop operation.

Caving 2014, Santiago, Chile

Table 1 Zoning used mina Reservas norte


- Advanced panel
Variation of panel - Advanced - Conventional - Conventional panel - Conventional panel
caving panel caving panel caving caving caving
from XC5N hacia el
-Faults N1 AND2,
-Faults G y F. AND3 y AND 4.
-Faults G and
Lithological -Falla de Agua y
East-West -Dacita - CMET
and structural And 1.
condition -Breccia -Brecha Anhidrita
-CMET Anhidrita - Prfido Diortic
Column height of 160 to 180 160 to 180
160 to 180 meters. 180 to 360 meters.
primary rock meters. meters.


Figure 5 Plan view of zoning used to cavity control at Reservas Norte Sector

Geomechanic Design

Figure 6 Results of the application of performance indicators, November 2013

5 Conclusions
Surface extraction (draw angle) is one of the key indicators of effective cave back control at El Teniente mine.
The results so far have identified some specific spots, where the extraction does not follow expectations on
the draw angle. Based on that finding, this tool has allowed improving the operational control over the draw
strategy, reducing the possibility of early mud entry and instability problems at the mine. In the future, to
achieve continuous improvement goals, an attempt will be made to use this criterion selectively.

The authors would like to thank Codelco Chile, El Teniente, to authorize the publication of this document
and, in particular, to all the people who make the Superintendency of Geomechanics GRMD.

Araneda, O & Gaete, S 2004a, Continuous modelling for caving exploitation, MassMin 2004, A Karzulovic
& M Alfaro eds., Chilean Engineering Institute, Santiago, Chile.

Duplancic, P and Brady, BH 1999, Characterisation of caving mechanisms by analysis of seismicity and
rock stress, Proc. 9th Congr., Int. Soc. Rock Mech., Paris (eds G. Vouille and P. Berest), vol.
2, pp. 104953, A. A. Balkema: Rotterdam.

Flores, G, Karzulovic, A and Brown, ET 2004, Current practices and trends in cave mining, MassMin
2004, A Karzulovic and M Alfaro (eds), Chilean Engineering Institute, Santiago, Chile,.

Landeros, P, Cuello, D & Rojas, E 2012, Caveback management at Reservas Norte Mine, Codelco Chile, El
Teniente Division, 6th Conference and Exhibition on Mass Mining, Massmin 2012, Canada.

Tomkins, J 1988, The warehouse management handbook, Mc Graw Hills eds, pp 513-559.

Caving 2014, Santiago, Chile

Geomechanical issues and concepts associated with

scoping study and prefeasibility stage of a Block/Panel
Caving Project
J Daz DERK Ltda., Chile
P Lled DERK Ltda., Chile
F Villegas Codelco, Chile

Considering the amount of underground mining projects in development that have studied the application
of the caving methods together with other ore bodies that will also evaluate the feasibility of a future
underground caving, this document summarizes the main geomechanical issues associated with the stages
of a scoping study and a prefeasibility study for the Block/Panel Caving operation that must be considered.
In addition, this technical paper will present the usual practices adopted and other considerations during
the development of this type of studies. Equally, this material could be used as a general reference for
engineering professionals in technical offices and/or mining companies interested in caving issues as well
as in education and professional training.

1 Introduction
The development of the first engineering stages in a Block or Panel Caving underground project, as any
other project, require the definition of various key geomechanical aspects, which scope and depth must be
in accordance with the stage of the development of the project. The terms and targets of each engineering
stage play a relevant role in the adequate supply of resources in Geomechanics during the project design
and planning, avoiding excessively advanced or unnecessary definitions, which will probably be reviewed
and modified again in the following stages of the project.

This work is focused on the Scoping and Prefeasibility stages of Block or Panel Caving underground
mining projects because, according to the authors, occasionally, the scope and depth of the geomechanical
issues tend to be confused between the above mentioned engineering stages.

2 Stages of a mining project

In general terms, a mining project has five major stages:

1. Planning Stage: It considers the period since the conception of the project, at the level of
idea or profile, up to the demonstration of its economic feasibility at the level of scoping

The Scoping Engineering identifies the projects business potential, key factors, fatal risks,
investment order of magnitude and operating costs together with the relevant technical issues.

The Prefeasibility (Conceptual) Engineering studies the possible alternatives of the project
to establish the most favourable case. The technical and economic feasibility of the mining
configurations, technologies, capacities, and others are determined. In this stage, the investment
sums (CAPEX) and operating costs (OPEX) are established.

Geomechanic Design

2. Design Stage: This is the stage where the project is DESIGNED, including the activities that
belong to the Basic and Detailed Engineering stages or contexts. Together with the former stage,
this one corresponds to the project category.

The alternative selected in the Prefeasibility stage is developed in this stage to demonstrate its
technical and economic viability.

3. Development Stage: This stage corresponds to the construction or development of the project,
according to the design specified in the former stage. However, there can be design modifications
during this stage without changing or altering the core concepts.

4. Operational Stage: This stage corresponds to the project operation during its life, including
the pre-production or ramp-up stage.

5. Closure Stage: This stage corresponds to the project closure after the end of its life.

To frame these general stages with the projects engineering phase, Figure 1 presents a scheme with these

Figure 1 Relationship between the general stages of a mining project with the engineering stages and the
category or stat

As explained previously, this work is framed within the planning stage and addresses the geomechanical
issues to be defined in the Scoping and Prefeasibility Engineering stages of a Block and/or Panel Caving
underground project.

A geomechanical problem can have a significant impact on an underground project; thus, it is necessary
to reduce or minimize the risk of making errors in the projects early stages (Scoping and Prefeasibility

Caving 2014, Santiago, Chile

3 Geomechanical issues of a Block/Panel Caving mining project

The major geomechanical issues in a Block/Panel Caving project to be addressed are the following:

Geological, Structural, Hydrogeological and Geotechnical characterization.

Intact rock, geological structures and rock mass properties.
Stress field characterization.
Caveability assessment.
Fragmentation (Primary and Secondary).
Induced Seismicity.
Geotechnical assessment for design.
Geotechnical assessment for planning.
Subsidence analysis.
Geotechnical hazards.
Although these issues can vary depending on the characteristics of each project and type of deposit, its
possible to cluster these issues in the following categories:

1. Geometry: Geometry establishes the factors that define the shapes, sizes and distributions in
space. It can be divided in two sub-groups: Natural and Unnatural. The natural geometry considers
the geologic, structural, hydrogeological and geotechnical models in addition to the fracturing
level of the environment (primary fragmentation) and the surface topography (geomorphology).
The Unnatural geometry involves the mining method scenario or context, greatly measured by
the geomechanical issues for design and planning.

2. Geomechanical Context: The geomechanical context is specific for each deposit and project;
it can be divided in two large groups: Pre-mining properties and Induced Loads. Pre-mining
properties consider the material characterization and contact zones in the geological, structural,
hydrogeological and geotechnical models, in particular, the parameters to evaluate the strength
and deformability of the rock mass and its components (intact rock and geological structures).
This group also characterizes the in situ or pre-mining stress state. On the other hand, Induced
Loads correspond to the effects that will be produced by caving on the environment, with
the review of aspects associated to caveability, cave propagation, abutment stress, secondary
fragmentation and induced seismicity, among other issues.

3. Interaction: Interaction considers the impacts that the project can create on the environment,
with its subsidence phenomena and the occurrence of geotechnical events.

Figure 2 shows a diagram that summarizes the issues involved in each one of the major groups or categories
that the Geomechanical Discipline must manage in a Block/Panel Caving project.

4 Geomechanical concepts and parameters in the scoping and prefeasibility

engineering stages
In practice, during the development of engineering projects, it has been possible to observe that there
are requirements and contributions made to Geomechanics that mostly dont have enough and reliable
information to respond according to the projects engineering stages. In accordance to this, it is always
highly necessary to define which subjects and parameters must be addressed, both in terms of their scope

Geomechanic Design

and depth from the geomechanical caving methods viewpoint in the Scoping and Prefeasibility (Conceptual)
Engineering stages.

Figure 2 Geomechanical Issues and Parameters considered in a Block / Panel Caving Mining Project

Table 1 provides a summary of the issues or subjects of interest and the geomechanical parameters the
authors suggest to be addressed during the Scoping and Prefeasibility Engineering stages in a Block / Panel
Caving Mining Project.

Table 1 Geomechanical Issues and Parameters in the Scoping and Prefeasibility Engineering stages

Issues of interest / Parameter Scoping Engineering Prefeasibility Engineering

Characterization Geological Plan views and sections with Geological model (3D) in a
Lithology, Alteration and computer platform that includes at
Mineralization, preferably least the lithologies, alterations and
systematically spaced. Plan mineralization attributes.
views must consider nearness to
The first geomechanical model (or
undercutting and production levels.
geotechnical model) corresponds
Sections must be oriented in the
to the geological model; thats why
geographic NS and EW and/or
its important to have it as early as
mina (local); spaced between 100
and 200 m.
Structural Plan views and sections with major Structural model (3D), in a
and/or principal structures traces computer platform with the major
and projections. structures identified, including
structural domains with the
Plan views near the undercutting
interpretation of the information
and production levels. The plan
about the minor structures
views and sections must spatially
surveyed from drillholes and/
coincide with the geological
or outcropping or underground
exploration works.

Caving 2014, Santiago, Chile

Table 1 Geomechanical Issues and Parameters in the Scoping and Prefeasibility Engineering stages
Issues of interest / Parameter Scoping Engineering Prefeasibility Engineering
Characterization Hydrogeological Recording of water levels in probe Conceptual Hydrogeological Model,
holes and/or drillholes. Water table and identification of permeable and
recharge resulting from precipitation impermeable units.
(water and/or snow). Knowledge of
surface run-offs or presence of lake
Geotechnical Geotechnical characterization, at least Geotechnical model (3D) able
one rock mass quality method (i.e., RMR to display at least two rock mass
Laubscher) in the geotechnical database. quality methods (for example RMR
Plan views near the undercutting/ Laubscher, GSI f(RMR Bieniawski)
production levels and sections with the or Q).
geotechnical quality distribution.
Development of intact rock samples
In case there is not geotechnical lab test campaigns for the principal
characterization, there at least must be units; uniaxial tests must be included
the spatial distributions and/or RQD determining the elastic constants (15
values and/or Fracture Frequency, FF. tests per unit), triaxial tests (5 tests for
each confinement level, preferably 5
Uniaxial Compressive Strength (UCS)
levels), indirect tensile tests (15 tests
estimations and Point Load Tests (PLT)
per unit); in addition to the index
in major units.
properties (unit weight, porosity, P and
S wave propagation velocity).
Stress State Preliminary estimation of in situ stresses Carry out preliminary stress
according to topography and lithostatic measurements, preferably through
column, benchmarking of the region hydraulic fracturing to obtain the deep
and/or mining district, use of the world stress distribution and using some
stress map, etc. mine-scale numerical models (3D).
It is suggested to review the technical It must be considered that these
literature (Hoek & Brown (1980); numerical models could reach error
Amadei & Stephansson (1997); Daz & levels between 25% and 50%.
Lled (2005); Lled & Daz (2006)).
Primary Fracture frequency FF and RQD analysis Primary fragmentation simulation
Fragmentation Primary fragmentation estimation for the geotechnical units of interest
from benchmarking and preliminary for each structural domain under the
simulations for different rock qualities, expected stress states and applying
structural domains and stress states. tools such as BCF, Size, etc. (Daz,
Lled, Aguilar & Seplveda (2013)).
Surface Topography Updated surface topography. Updated surface topography.
Mine Design Mining Method Analyzing the possible mining Analyzing the possible mining
alternative(s), development of decision alternative(s), development of
matrices for the undercutting variants decision matrices for the undercutting
comparatively evaluating geomechanical variants comparatively evaluating
concepts. geomechanical concepts with the
highest detail and information
It must be possible to detect possible
fatal flaws as major focus of attention.
Levels elevation Knowing preliminary elevations of It is necessary to know the elevations
the main levels (undercutting and of the undercutting, production,
production) of the mining alternative(s). ventilation and haulage levels of the
mining alternative(s).

Geomechanic Design

Table 1 Geomechanical Issues and Parameters in the Scoping and Prefeasibility Engineering stages
Issues of interest / Parameter Scoping Engineering Prefeasibility Engineering
Mine Design Levels Extraction grid Carrying out a comparative The alternative(s) for selected
Layout analysis of the different drawpoint spacing must be analysed,
types of extraction grids, hopefully with 3D numerical models.
benchmarking support and This model must include excavations
expert judgement. and pillars from the Production
level, ore pass system, draw bells
and undercutting in the undercutting
Crown Pillar The nominal thickness of the The thickness of the crown pillar
crown pillar can be defined the proposed can be analysed through
support of benchmarking and available 3D numerical models.
expert judgement. In addition,
it must be in agreement with
the undercutting variant and
geotechnical scenario.
Draw point The draw point design The thickness of the crown pillar
must consider the expected proposed can be analysed through
geotechnical scenario. available 3D numerical models and
must comply with the geomechanical
design criteria.
Undercut type and height Carrying out a comparative Defining the undercut type and
analysis of undercut height height alternative, it must consider
(low, medium or high), the expected geotechnical scenario,
benchmarking and expert the use of caving assistance methods
judgement are used. (preconditioning), geomechanical
design criteria and drilling and
blasting criteria.
Levels Undercutting Generation of major drawings with
Ground Production support recommendations, support
Support systems and elements arrangement in
Ventilation Definition of general ground
addition to technical specifications.
Reduction support criteria, usually from
geomechanical classifications, It is possible to make a support
Transportation experience obtained at the zoning from the additional
Temporary site and/or other companies, information available.
Infrastructure in addition to the specialists
Verification of the support
expert judgement.
recommendations through 2D
numerical modelling for different
geotechnical scenarios.

Caving 2014, Santiago, Chile

Issues of interest / Parameter Scoping Engineering Prefeasibility Engineering

Mine Planning Block Height The in situ and extractable A deeper analysis is made to the
block height proposed must assessment of block height impact
be analysed from empirical in a similar manner as the Scoping
relations, benchmarking and stage, including the draw point
expert judgement to make support design and draw bell design
caving propagation analyses, to in the analysis.
estimate the in situ and induced
It is possible to develop 2D and 3D
stress state and the subsidence.
numerical models depending on the
Parametric 2D numerical
complexity of the project (Daz &
models can be developed.
Lled 2008).
(Karzulovic et al. 2004)).
Advance Sequence The advance sequence must The definite advance sequence
consider the principal stress must consider the orientation of
orientation and the orientation major stresses, the orientation of
of major geological structures. major geological structures, the
undercutting front geometry, the
undercutting and production levels
layout and the distance between
Possible it will be necessary to
make 3D modelling in the definite

Geomechanic Design

Table 1 Geomechanical Issues and Parameters in the Scoping and Prefeasibility Engineering stages

Issues of interest / Parameter Scoping Engineering Prefeasibility Engineering

Mine Planning Caving Rate The definition of the average It is necessary to define the
caving rates must be made through operationally reasonable
benchmarking and/or expert caving rate ranges considering
judgement. differentiated rates for design,
planning and geomechanical
conditions. This tries to mitigate
the eventual occurrence of
geotechnical hazards.
It is recommended to use expert
judgement and the experience
obtained in sites with similar scale
and geomechanical context.
Draw Rates The definition of the average It is necessary to define the
draw rates must be made through operationally reasonable draw rate
benchmarking and/or expert ranges considering differentiated
judgement. rates for different design, planning
and geomechanical conditions.
It is recommended to use expert
judgement and the experience
obtained in sites with similar scale
and geomechanical context.
Secondary Fragmentation Normally based on benchmarking and A secondary fragmentation
project requirements for the expected analysis is required to establish the
productivity of the draw point and expected productivity in the draw
production rate. point and production rate.
Sensitivity analysis is made for
the different geomechanical units,
structural domains and stress state.
Caveability Caveability Use of empirical methods It is recommended to use empirical
Assessment (Laubschers caveability chart (1990). methods (Laubschers caveability
chart (1990) or Laubscher &
Jakubec (2001) including the
data of sites using caving with a
geotechnical scenario similar to the
projects one.
Caving Propagation Use of empirical relations and In addition to empirical relations,
benchmarking (Karzulovic et al. make use of numerical modelling
2004). if an unfavourable condition is

Caving 2014, Santiago, Chile

Table 1 Geomechanical Issues and Parameters in the Scoping and Prefeasibility Engineering stages
Issues of interest / Parameter Scoping Engineering Prefeasibility Engineering
Geotechnical Collapses Collapse occurrence potential analysis Collapse risk analysis.
Hazards to identify if it corresponds to a fatal Identification of potential
flaw or a serious condition for the collapse occurrence zones.
project. Recommendations on design and
planning measures for mitigation
Rockbursts Rockburst occurrence potential Rockburst risk analysis; seismic
analysis based on the rocks risk analysis.
mechanical properties and stress state
Recommendations on design and
in order to identify if it corresponds to
planning measures for mitigation
a fatal flaw or a serious condition for
the project.
Hang-ups (Air Blast) Hang-ups (air blast) occurrence Hang-ups (air blast) risk analysis.
potential analysis in order to identify Identification of mitigation
if it corresponds to a fatal flaw or a measures.
serious condition for the project.
Recommendations on design and
planning measures for mitigation
Water/Mud Rushes Water/mud rushes occurrence potential Water/mud rushes risk analysis.
analysis based on hydrogeological
Recommendations on design and
information in order to identify if it
planning measures for mitigation
corresponds to a fatal flaw or a serious
condition for the project.
Geotechnical Subsidence Subsidence Preliminary estimation of the final Preliminary estimation of the final
events Crater extent of the subsidence crater on the extent of the subsidence crater
surface, use of empirical methods and/ on the surface, use of empirical
or benchmarking. General subsidence methods and 3D numerical
angles are defined. modelling. Angles of subsidence
are defined by period and by crater
wall, considering the geotechnical
Zone of Preliminary estimation of the zone of Estimation of the zone of influence
Influence influence around the final subsidence around the subsidence crater
crater, use of empirical methods and by period, use of 3D numerical
benchmarking. modelling.
Major Infrastructure Orientation and location of In addition to deepen the work
excavations with respect to the made in the Scoping Engineering,
principal structures, stress state (in the geometries, sizes and level of
situ and induced), geotechnical quality stability of the excavations must be
of the rock mass and subsidence analysed.
Preparing the construction
sequence of the excavations and
defining systematic ground support

Geomechanic Design

5 Conclusions
The input information in each one of the engineering stages must be considered as strategic information.
Consequently, the quality and reliability of this information define the quality and the reliability of the
analyses made to support the projects geomechanical recommendations.

It is relevant for the client to have a team of at least one geotechnical geologist and a geomechanical engineer
with experience in underground caving. This will substantially improve the quality of the geomechanical
outputs associated with the scoping and prefeasibility engineering. This becomes increasingly relevant
when there is a transition from open pit mining to underground mining.

Hoek, E & Brown ET 1980, Underground Excavations in Rock, Institution of Mining and Metallurgy,
ISBN 0 419 160302, E & FN SPON.

Amadei, B & Stephansson, O 1997, Rock Stress and its Measurement, ISBN: 0 412 44700 2, Chapman &
Hall, Londres - Inglaterra.

Daz, J & Lled, P 2005, Estado Tensional In Situ y/o Preminera en Mina Chuquicamata - Divisin Codelco
Norte, Informe Tcnico IT-DCN-E01-01-05, Divisin Codelco Norte de Codelco Chile por
Derk Ingeniera y Geologa Ltda.

Lled, P & Daz, J 2006, Mediciones de Esfuerzo In Situ Mediante Tcnicas de Hidrofracturamiento, in
Proceedings of Mining 2006 II International Conference on Mining Innovation, May 23 to 26,
Santiago Chile.

Lled, P & Daz, J 2008, Modelos Numricos 2D para Diseo de Soporte, Ingeniera Conceptual
Chuquicamata Subterrneo Codelco Chile VCP, Informe Tcnico IT-VCP_CHS-E09-01-08
emitido por Derk Ingeniera Limitada al Proyecto Chuquicamata Subterrneo.

Laubscher, DH 1990, A Geomechanics Classification System for the Rating of Rock Mass in Mine Design,
South African Journal of Mining and Metallurgy, vol. 90, no. 10, Oct. 1990, pp.257-273.

Daz, J, Lled, P, Aguilar, J & Seplveda, J 2013, Geotechnical Pre-Feasibility Study Carrrapateena
Project, Technical Report IT-NCL-E01-02-2013, Derk Ingeniera y Geologa Ltda. To NCL.

Karzulovic, A, Flores, G & Brown, T 2004, Current Practices and Trends in cave mining, Massmin 2004,
Codelco Norte Division, Codelco Chile.

Laubscher, DH & Jakubec, J 2001, The MRMR Rock Mass Classification for Jointed Rock Mass, eds.
Hustrulid, W. A. & Bulock, R. L. Underground Mining Methods. Engineering Fundamentals
and International Case Studies, 718 p. SME: Littleton, Colorado.

Codelco 2014, Available at


Caving 2014, Santiago, Chile

Geotechnical Characterization

Geotechnical Characterization

Ciresata geotechnical evaluation and caving study,


N Burgio Stratavision Pty Ltd, Australia

The Ciresata deposit is located in the Golden Quadrilateral Mining District of the South Apuseni Mountains
in west-central Romania. Carpathian Gold Inc. commissioned a prefeasibility study to determine a suitable
bulk scale underground mining method to achieve a production capacity of 30,000 t/day. Geotechnical
investigations focused on the sublevel and block cave mining options.

Material property testing of diamond drill core was undertaken to provide essential information for
caveability, fragmentation and stress modelling analysis. Anomalous rock strength information required
a second testing laboratory to be utilised for data verification. Initial predictions of fragmentation were
highly sensitive to rock strength and local stress field information which impacted undercutting strategies
for the block cave option. This paper outlines how engineering judgements were applied to cover data
uncertainty in the early stages of the prefeasibility study and how the ongoing geotechnical evaluation was
managed as new information became available.

1 Introduction
The Ciresata deposit is located in the Golden Quadrilateral Mining District of the South Apuseni Mountains
in west-central Romania (Figure 1). Historic gold production from the Golden Quadrilateral exceeds 55 MM
Oz. Ciresata is one of the three main deposits occurring within the Rovina Exploration license, managed by
Carpathian Gold Inc. The Rovina and Colnic deposits are planned as open pit operations whilst Ciresata is
undergoing studies for a bulk underground mining technique. The license is located approximately 20 kms
southwest of Rosia Montana. A prefeasibility study concluded that Ciresata was amenable to the sub-level
caving or block caving based on a production capacity of 30,000 t/day.

Figure 1 Ciresata Location Map

Caving 2014, Santiago, Chile

2 Ciresata geology & geotechnical domains

Gold-copper mineralization is hosted by subvolcanic intrusions of Neogene age and adjacent hornfelsed
Cretaceous sediments (Nebauer et al. 2005). The porphyries comprise of small, vertically attenuated, coarse-
grained hornblende-plagioclase intrusives. Mineralisation occurs as pyrite-chalcopyrite disseminations
associated with sheeted and stockwork quartz veining. Alteration types range from magnetite progressing
outwards to potassic, phyllic, and propylitic assemblages. Mineralisation commences between 50 m and
150 m below the present day surface due to a hornblende porphyry which caps mineralisation near the
surface. The host sediments dip at moderate angles to the northeast.

Geotechnical domains were defined based on lithology, structure, alteration and rock mass properties
(Figure 2). Several intrusive phases were grouped into a single domain referred to as the Intermineral
Porphyry. Weaker geotechnical conditions occur closer to surface due to increases in argillic alteration and
elevated fracturing. The rock mass was characterised based on Laubschers RMR system of classification
(Laubscher 1990) and summarised in Table 1.

Preliminary fault interpretations were prepared to identify structural features that could assist caveability
and cave propagation behaviour. Sub-vertical faults strike along a north to north-westerly direction. Haloes
of carbonate wall rock alteration are often observed next to faults zones. Milled breccias, representing
thrust faults, dip sub-parallel to bedding at moderate angles to the north and northeast. The thrust faults
appear to be offset by earlier vertical faults.

The fracture frequency below the Argillic Cap is generally low and typically ranges from 1.5 to 2 fractures
per meter. There appears to be no preferential development of jointing along the bedding planes, despite
sediments occasionally displaying laminated textures.

Figure 2 Ciresata Geotechnical Domains

Geotechnical Characterization

Table 1 Average Rock Mass Ratings for Geotechnical Domains

Domains Description RQD RMRL90

Surface fracture Zone Oxidation (1-5m) and elevated fracturing to 50m depth 38 43
Argillic Cap Argillic-carbonate alteration extending to 150m depth 75 55
Sediments Sediments dipping moderately towards the northeast 83 60
Intermineral Porphyry Sub-vertical intrusive pipes and dykes 85 60
Western Porphyry Northwest striking intrusive located west of the ore zone 73 57
Eastern Capping Porphyry Intrusive sill which caps mineralisation 82 58
Lower Bedded Zone Sedimentary zone with occasional laminated bedding 71 55
Lower Fracture Zone Zone of increased fracturing due to faulting 67 51

3 Caveability
A conceptual block cave layout was positioned 650 m below surface.The hydraulic radius (HR) of the
production area (HR=92) is well in excess of that required to initiate caving (HR=42). By contrast, a
minimum span of 120m (HR~28-30) was established for the primary SLC level within the narrower
carapice of the mineralised zone approximately 120 m below surface (Figure 3). The rock mass above
the SLC level has an MRMR range from 42 to 60, hence, the minimum span offers a reasonable balance
between proximity to surface whilst enabling a progressive stepout for the expansion of subsequent SLC
production levels at depth.

Figure 3 Ciresata SLC and Block Cave Caveability Chart

Caving 2014, Santiago, Chile

4 Stress field estimation

The likely stress field conditions were estimated based on a review of local and regional tectonic
information. Several geologically recent extensional basins and over-thrusting combined with current
earthquakes, associated with subducting remnant tectonic plates, has generated a tectonically complex area.
The generalised map of European stress trends implies there is little stress data near the Ciresata deposit
and it is not certain if the local stresses are European (NW-SE) or related to the Anatolian Province to the
southeast. In addition the European stress trends are based on deep earthquake and hydraulic fracturing data
rather than shallow in-situ rock stress measurements in mines or civil constructions (Lee 2012).

Deposit scale structures (i.e. dykes, contacts, faults shears, bedding etc) are thought to influence local stress
orientations and principle stress ratios, rather than regional tectonic considerations. As the lithologies at
Ciresata are geologically recent it can be argued that the styles, and offsets of the main faults, are likely
to be good indicators of the current regional stress field. The surrounding over-thrusting implies that high
horizontal stress conditions cannot be sustained within the ringing Carpathians.

The interpreted steeply dipping NNW-SSE faults and have a similar orientation to the regionally developed
structures and may imply N-S orientation for the local maximum horizontal stress. There have been no
observations of drilling difficulties such as squeezing or core discing. Two acoustic televiewer surveys were
completed, however these were shallow and borehole breakouts or drilling induced fracturing would not
be expected at these depths. Using current interpretations, it was concluded that the most likely orientation
for the maximum horizontal stress at Ciresata is NNW-SSE i.e the bisector of the acute angle between the
interpreted fault structures. Estimated rock stress relationships were developed for the most likely scenario
as well as more and less deviatoric options (Table 2).

Table 2 Estimated Rock Stresses for Ciresata

Geotechnical Characterization

5 Rock strength testing

Field observations and point load testing indicated hard rock strengths, yet initial UCS tests were
conspicously low at almost half the expected value. A decision was made to apply the higher rock strength
parameters inferred from field observations, for preliminary geotechnical analysis. Further UCS testing
from an alternate laboratory (Laboratory B) generated much higher results which more closely matched
field observations and verified engineering assumptions (Table 3).

Table 3 Initial UCS Test Results

Rock Type Laboratory A (MPa) Laboratory B (MPa)

Min Average Max Min Average Max
Porphyry 30 56 118 28 116 180

Sediments 22 42 69 40 76 120

An independent suite of samples were sourced and supplied courtesy of Newcrests Cadia Valley operation
in Australia to check the performance of both laboratories. Sample duplicates were prepared including a
standard comprising of dental plaster. The results confirmed earlier observations that Laboratory A was
almost consistently generating lower than expected results. There remains some uncertainty if the lower
results were due to sample preparation, equipment, or some other systematic error.

Table 4 Comparative UCS Testing using Cadia Valley Drill Core

Laboratory A Laboratory B
Cadia Valley Sample Descriptions (MPa) (MPa)
Volcanoclastic rock 116 171

Volcanoclastic rock with epidote 90 68

Pyroxene phyric volcanoclastic rock 64 102

Pyroxene phyric volcanoclastic rock 61 184

Dental plaster mould (15MPa standard) 8 12

6 Fragmentation
Fragmentation predictions are highly susceptible to the degree of natural fracturing and also the ratio between
induced stress and rock strength. The Ciresata rock mass has a widely spaced jointing system, hence, the
degree of stress induced fracturing becomes an important factor in reducing fragmentation provided the
stress/strength ratios are favourable. The analysis was undertaken using the Block Cave Fragmentation
program (BCF v3.05) developed by Esterhuizen (2005). Fragmentation profiles were generated for low and
high strength scenarios for the sedimentary and intrusive host rocks (Figures 4 and 5).

Caving 2014, Santiago, Chile

Figure 4 Fragmentation Domains

Figure 5 Primary Fragmentation for Sediments and Intrusives

Geotechnical Characterization

There is a substantial difference in primary fragmentation for each scenario based on the rock strength
inputs. The intrusives are expected to generate very coarse fragmentation albeit over a relatively small
production area. Secondary fragmentation analysis suggested that 50 m to 75 m of draw would be required
before more manageable levels of oversize were encountered from the sediments for the block cave option.
The design of a high undercut would mitigate the initial coarse fragmentation to more manageable levels
(Figure 6).

Figure 6 Undercut Design and Secondary Fragmentation

7 Conclusions
Comparative laboratory testing became necessary to address and confirm discrepancies in UCS results.
Sample standards and duplicates offered an opportunity to test the quality and reliability of laboratory results
and supports the engineering judgments applied to the base case scenarios for Ciresata. The decision to use
an alternate material testing laboratory was significant as lower rock strengths would have underestimated
fragmentation and overestimated the rock mass deformation response.

The Ciresata rock mass and geometry is amenable to block cave and sub-level cave mining methods.
Minimum dimensions were established for the primary SLC level to initiate caving. The block cave
footprint far exceeds the hydraulic radius required for caveability, however, cave propagation could be
slow due to the competent rock mass and limited faulting. A high undercut design was incorporated into
the block cave option to reduce secondary breakage requirements and improve early productivity. An
improved understanding of the rock mass will be achieved once underground exposures become available
and site stress measurements are undertaken. Opportunities also exist to consider preconditioning to reduce
fragmentation and assist cave propagation.

I wish to acknowledge Mr Randy Ruff of Carpathian Gold Inc. for permitting the publication of this paper.

Caving 2014, Santiago, Chile

Esterhuizen, G 2005, BCFV3.05 A program to predict block cave fragmentation, Technical Reference and
Users Guide.

Laubscher, DH 1990, A geomechanics classification system for the rating of rock mass in mine design,
Transactions A. Afr. Inst. Min. Metall, vol.90, N10.

Lee, M 2012, Ciresata Rock Stress Estimate, Internal Memo, Carpathian Gold Inc.

Neubauer, F, Lips, A, Kauzmanov, K, Lexa, J & Ivascanu, P 2005, Subduction, slab detachment and
mineralization: The Neogene in the Apuseni Mountains and Carpathians, Ore Geology
Reviews, Elsevier, pp. 13-44.

Geotechnical Characterization

Identification of different geomechanics zones in panel

caving- application to Reservas Norte El Teniente
P Landeros Codelco, Chile
J Cornejo Codelco, Chile
J Alegra Codelco, Chile
E Rojas Codelco, Chile

The identification of different geomechanics behavior has been a constant focus of interest for the rock
mechanics engineering applied to planning and projects at El Teniente Mine, considering that most
commonly used rock mass classification systems (i.e.: RMR, RQD and Q), represent similar values where
most important exploitation sectors are taken place up to now. From that point of view, geologists have
reached a good advance for establishing differences in terms of geotechnical aspects, mainly associated to
vein infilling and in situ fragmentation.

Then, for the planning process, it is very important to be able to have a geomechanics model which
interprets geotechnical characterization against mining exploitation on different scales, such as local scale
(excavations) or global behavior (cavities).

In this study, a methodology for the analysis and evaluation of geomechanics behavior is presented,
considering a case study with panel caving exploitation, emphasizing aspects like induced stresses,
seismicity induced by mining, hydraulic fracturing and damages ahead the undercutting front.

Finally, discussion of results is focused on its relationships with geotechnical characterization and
geomechanics guidelines for mine planning.

1 Introduction
El Teniente Mine includes different production sectors (see Figure 1), all of them located around a chimney
of sub-volcanic breccias with an inverted cone shape, known as Braden Pipe. Reservas Norte also
known as Sub6 Sector at the beginning is located on the north-eastern side, and its exploitation started
in 1989 using conventional panel caving. Several rockbursts occurred during the 1990s and different
exploitation sequences were tested to ensure continuous operation. Up to date, Reservas Norte is mined
by advance panel caving, considering a production plan close to 40,000 tons/day. Up to date, the cave at
Reservas Norte mine is in a steady-state condition, and its geometry consists of an active 700 meters wide
undercutting front.

One of the main constrain to the explotaition at El Teniente is related to the geomechanical hazards. This
hazards also mean planning constrains (rate of draw, .etc) that depends on the sector to be analised. In
this article we present the methodology used at El Teniente to define the geomechanical areas for planning
purposes. These are based on an understanding of the geotechnical environment, understanding of the
induced stresses on mine infrastructure, seismic potential and blasting induced seismicity. In the following
sections, each criteria is described finalizing with the definition of zones for Reservas Nortes sector.

Caving 2014, Santiago, Chile

Figure 1 Productive sectors and projects, El Teniente Mine (Business Plan 2013)

2 Geological and geotechnical data

Predominant lithology corresponds to Andesite intruded by three major ore bodies: Dacite Porphyry and
Anhydrite Breccia on the western side and Diorite Porphyry, on the central and eastern sides. In terms of
intact rock properties, all of them are very stiff with average Youngs Modulus approximately 60 GPa.
On the other hand, in terms of rock mass quality indexes, most of these ore bodies are very similar and
competent with GSI in the range of 75 to 90 and IRMR in the range of 55 to 62. Most important geological
faults are classified as master faults (faults G, C, N1 and N2) and major faults (faults F, AND-1, AND-
2, AND-3, AND-4 and AND-5), see Figure 2.

3 Induced stress state

2.1 Numerical modelling

The geometry of the broken material cavity controls major differences in terms of induced stress conditions.
For the evaluation of this impact into the mining plan, comparative analysis are developed based on a
tri-dimensional numerical analysis with a linear elastic boundary element software, calibrated with field
information such as stress measurements, damage ahead the undercutting front and propagation pressures
of hydraulic fracturing (Cuello et al 2010). Reservas Nortes numerical model is part of a bigger mine scale
model, considering cavities of different productive sectors and lithological aspects such as Braden Pipe.
This model is calibrated to get a good approach of pre mining stress states in different zones of the mine.
Then, to get a local stress condition at Reservas Norte mine, improved geometries are incorporated such as
specific caveback and undercutting surfaces, as shown in Figure 3.

Geotechnical Characterization

Figure 2 Plan view of major geological aspects, Reservas Norte mine, undercutting level (modified from
Gonzalez 2013 and Gallardo 2013)

Figure 3 Mine scale numerical model (left) and local conditions for Reservas Norte sector (right), map3d code

Input parameters are defined, mainly considering three predominant materials: Braden Pipe, Andesites and
broken material. Numerical modelling strategy considers the utilisation of field information to calibrate and
validate the model. Calibration process is described by Cuello et al 2010, and it considers historical record
of damage ahead the undercutting front and propagation pressures obtained during the hydraulic fracturing
process. Differences between model estimation and field information are shown in Figure 4.

2.2 Propagation pressures of Hydraulic Fracturing (HF)

Average magnitude of the propagation pressure obtained from the pre-conditioning process by hydraulic
fracturing allows obtaining a very good approximation of confinement magnitudes inside the rock mass.
This valuable information is also used for calibrating numerical modelling.

Geostatistical analysis have been developed using this information and a block modelling by kriging
techniques (Cornejo and Landeros 2013). Results for the zone of interest are shown in Figure 5.

Caving 2014, Santiago, Chile

Figure 4 Severe damage criterion applied to undercutting layout, Reservas Norte (modified from
Cuello et al 2010)

Figure 5 Propagation pressures of HF (left) and kriging results (right), from Cornejo and Landeros 2013

4 Analysis of seismicity induced by mining

3.1 Evaluation of seismic hazard

For the evaluation of the seismic hazard, the selection of the volume of analysis covered 375*10^6 m3,
including both Reservas Norte and Pilar Norte sectors. All the events recorded by the seismic monitoring
system from December 2011 to April 2013 were considered.

Geotechnical Characterization

Initial filtering is related to spatial location of each event and the general uncertainty of its estimation.
Then, a distance is estimated with the 95% of the data lower error. By the other hand, modal magnitude
is used for estimating minimal sensibility of the seismic system for the zone in analysis, in this case, this
local magnitude corresponds to Mw=-0.7. The objective of the application of a spatial-time filtering is
to eliminate all the events that do not interact with principal clusters, because their location is further
according to the data base characteristics or because their occurrence in time is not representative of the real
latency of the zone of analysis. Then, only events with an effective interaction are included on the analysis,
as shown in Figure 6.

Figure 6 Seismic events considered for the analysis (left) and the spatial-time filtering applied (right)

All the selected events are clustered in a later stage of analysis, using agglomerative hierarchical techniques,
maximizing distance between clusters. At first, a dendogram is built to calculate the distance matrix between
events and then, an optimal number of groups are determined with at least 250 events each. The whole
process is described in Figure 7.

Figure 7 Identified groups according to agglomerative hierarchical techniques

Caving 2014, Santiago, Chile

Rockburst database was used as base information to categorize the seismic hazard, where 95% of rockbursts
are related to local magnitudes higher than Mw=1.5 and energy release higher than 10^6 J. There is also a
high correlation between magnitudes, energy releases and location of damage. Then, a dispersion graph was
used to examine tendencies, estimating 50 meters of linear damage.

Finally, different hazard levels were defined, considering the occurrence of those conditions. Higher hazard
level are related to all defined conditions; by the other hand, lower hazard levels are related to clusters
where there is no occurrence of magnitudes and energy releases higher than previously estimated, as shown
in Figure 8.

Figure 8 Zones identified with different seismic hazard

5 Seismic events after blasting

This stage of the analysis considered all seismic events with magnitude higher than Mw>0 recorded in on
each single blasting. Several assumptions were defined as it follows:

The radius of influence of each blasting corresponds to 100 meters spherically measured.
Time of influence of each blasting corresponds to 24 hours after.
Spatial coordinates are referred to the centre for each blasting.
The analysis was developed per year, and considering geotechnical zones as shown in Figure 9. The
geotechbical domains are based on different kind of vein infilling, establishing limits between rock mass
characteristics. In this Figure also thje magnitude and location of events are plotted. This shows that the

Geotechnical Characterization

blasting related seismic events are located to the West of the sector. As the front moves to the west so did
the events.

In general terms, it is possible to identify that seismic events related to blasting events with higher magnitudes
have a very similar spatial distribution during 2008 and 2009, when the undercutting front were positioned
in geotechnical zones with similar characteristics. Once the undercutting front was moving ahead to the
west (years 2010 and 2011), the frequency of these kinds of events increased. As mining conditions could
be considered constant similarly, the increase could be related to a different rock mass.

Figure 9 Seismicity recorded after blastings, including undercutting front at the end of each year (modified
from Riquelme 2012 and Benado 2008)

6 Delimitation of different geomechanics zones

The process of identifying areas with different geomechanical behavior includes all the parameters described
in the preceding paragraphs, adding some operational aspects such as orientation of undercutting front, the
extension of the transition zone and orientation of the drifts.

Results are focus on the definition of polygons, used for the daily operational activities and the short term
analysis developed by geomechanics engineers, supporting the operational process. A plan view with the
zones is shown in Figure 10, where x polygons were defined for geotechnical control.

Caving 2014, Santiago, Chile

Figure 10 Plan view with the current operational polygons for geomechanics ground control (modified from
Gallardo 2013)

7 Conclusions
The analysis and evaluation of geomechanics aspects is a relevant focus for mine design and planning. The
identification of different geomechanics behaviour is based on a complementary analysis between induced
stresses, seismicity induced by mining, hydraulic fracturing, damages ahead the undercutting front and
some operational aspects.

This process is iterative permits to define polygons which represent the base for daily evaluation of seismic
activity and for further analysis. These polygons are updated periodically by geomechanics staff of El

The authors wish to thank Codelco Chile, El Teniente Division for allowing the publication of this paper
and the Geomechanics Staff that supplied data and information.

Benado D 2008, Geotechnical zones based on vein infilling stockwork, Internal Report, Codelco Chile
El Teniente.

Cornejo, J, Landeros, P 2013, Estimation of propagation pressures associated to hydraulic fracturing

process, using geostatistics techniques, El Teniente Mine, XVIII Symposium of mining
engineering, Simin 2013, Universidad de Santiago, Chile.

Geotechnical Characterization

Cornejo, J 2013, Identification of hazard zones, using seismic events clustering, Msc thesis, mining
engineering department, Universidad de Chile.

Cuello, D, Landeros, P & Cavieres, P 2010, The use of a 3D elastic model to identify rock mass damaged
areas in the undercut level at Reservas Norte sector, Proceedings of 5th international
conference on deep and high stress mining, Santiago, Chile.

Gallardo, M 2013, Polygons for seismic control and frequency events criterion, Internal Report, Codelco
Chile El Teniente.

Riquelme, O 2012, Analysis and evaluation of the geomechanics behavior of high drawbell, Mining
Engineer Thesis, Mining Engineering Department, Universidad de Santiago de Chile.

Caving 2014, Santiago, Chile

Geostatistical evaluation of fracture frequency and

SA Sguret MINES ParisTech, France
C Guajardo Codelco, Chile
R Freire Rivera Codelco, Chile

This work details how to estimate the Fracture Frequency (FF) ratio of a number of fractures divided by a
sample length. The difficulty is that, often, a part of the sample cannot be analysed by the geologist because
it is crushed, a characteristic of the rock strength that must also be considered for the Rock Mass Rating.
After analysing the usual practices, the paper describes the (geo)-statistical link between fracturing and
crushing and the resulting method to obtain an unbiased estimate of FF at a block or point support scale.
Some concepts are introduced: True FF, Crushed FF, crushing probability and crushing proportion.
The study is based on a real data set containing more than 13,000 samples. An appendix gives a very
general formal demonstration on how to obtain unbiased ratio estimation.

1 Introduction
One of the most important attributes used in the Rock Mass Rating (RMR) is the Fracture Frequency
(FF); a ratio of a number of fractures counted by the geologist divided by the sample length. However,
the calculation is not that simple because it often happens that a significant part of the sample is crushed,
making the fractures counting impossible, and FF becomes the ratio of two quantities that both change
from one location to another one in the deposit, making the evaluation difficult, whether at sample or block
scales - in other words, this ratio is not additive (Carrasco et al. 2008). To get around this difficulty, the
usual practice consists of using an additive formula that combines fractures number and crush length. The
aim of this paper is:

Analyzing the geostatistical link between fracturing and crushing.

Proposing an unbiased way to estimate FF.
Introducing the concept of crushing probability.

2 Formalization
Figure 1 shows a sample with fractures and defines the vocabulary.

In the following, all the samples are supposed to have the same length (1.5 m). For simplification, one will
consider just one location x (center of gravity of the sample) for LNC, LC and Nfract. The quantities LNC,
LC and Nfract, counted by 1.5 m length, are additive and can be estimated by the basic geostatistical method
called kriging (Matheron 1963). Nfract plays the role of a fractures accumulation, the equivalent of
the metal accumulation in conventional mining, i.e. the product of the grade by the thickness of the vein.

Geotechnical Characterization

Figure 1 Scheme presenting the useful variables, Crush Length and Fractures Number

The quantity:

N fract ( x )
FFtrue ( x ) = (1)
L NC ( x )

is the key frequency as it represents the true fractures frequency in the non-crushed part of the material.
N ( x) and L NC ( x) change and the average
However, it is not additive: when x moves in the space, fract
frequency between two measurements located at x1 and x2 is:

N fract ( x1 ) + N fract ( x2 )
FFtrue ( x1 U x2 ) =
L NC ( x1 ) + L NC ( x2 )

FFtrue ( x1 ) and FFtrue ( x2 ) only if L NC ( x1 ) = L NC ( x2 ) .

This latter ratio is equal to the average of
FFtrue ( x0 ) for any x0, using surrounding measurements FFtrue ( xi ) , is
Therefore, a direct kriging of
not possible.

This is the reason why practices use the formula:

N fract ( x ) + aLC ( x )
FFcorrected ( x ) = (2)

In Equation (2), the coefficient a represents an arbitrary quantity supposed to give more or less importance
to crushing in comparison with fracturing (a=40 in our case). By this way, the geotechnician incorporates
the information given by crushing. Equation (2) has also the advantage of combining additive quantities
that can be estimated separately and then combined:

N *fract ( x ) + a.L*C ( x )

FFcorrected ( x ) = (3)

Caving 2014, Santiago, Chile

In Equation (3), the exponent * denotes various estimates.

To understand what the coefficient a represents, let us develop Equation (2):

L NC ( x )FFtrue ( x ) + LC ( x )a L NC ( x )FFtrue ( x ) + LC ( x )FFcrushed ( x )

FFcorrected ( x ) = = (2)
L NC ( x ) + LC ( x ) L NC ( x ) + LC ( x )

Presented in this way, Equation (2) appears as an additive formula combining two frequencies: a being
FFcrushed ). This latter quantity must be at least greater than
the one associated with crushing (now written
FFtrue and we will detail this point in the following.
any observable

First, let us analyse the link between fracturing and crushing.

2 Observation of a natural phenomenon

We start by the examination of two samples:

Figure 2 Two samples: Few crushing and fractures (a) and important crushing, numerous fractures (b)

Figure 2a presents a drill core where the crush length is only 11 cm with just one fracture in the non-crushed
part; Figure 2b presents the contrary: crush length is important (74 cm over 1.5 m) and 16 fractures in the
remaining part. Is it a particular example or is there a statistical link between Nfract and Lc? We have
analysed 13,000 samples (1.5 m length) coming from an underground mine in a 1000 x 2300 x 1000 m3
box along x, y, z. (Figure 3).

Geotechnical Characterization

Figure 3 Planes presenting projections of the data

The scatter diagram between Nfract and LC (Figure 4a) leads to mixed conclusions:

The correlation coefficient is important (0.75).

70% of the population lies inside the confidence interval defined by the conditional expectation
curve, the remaining part does not present significant correlation.

4 True frequency estimation

Figures 4b, 4c and 4d present, respectively, the direct Nfract variogram (Matheron 1962, or a possible
alternative calculation given by Emery 2007), Lc variogram and their cross variogram. All these variograms
can be modelled by a unique model, up to a multiplicative factor, in other words, Nfract and Lc are in
intrinsic correlation (Wackernagel 1995).

Two important consequences result from this experimental property:

It is not useful to use cokriging (Wackernagel 1995) for estimating Nfract or Lc.
The ratio of both estimates obtained by kriging is non biased (see Appendix).
This latter property leads immediately to the method for estimating the non additive quantity FFtrue at a
block scale V located at coordinates x:

N Kfract (Vx )
FFtrue (Vx ) = (4)
LKNC (Vx )

Caving 2014, Santiago, Chile

Figure 4 Scatter diagram between crush length (Lc, horizontal axis) (a) and Fractures number (Nfract).
Line represents the linear regression of Nfract against LC, and the conditional expectation curve. Red
dotted lines represent the standard deviation around the conditional curve. Resp. Nfract, Lc, and Nfract
cross Lc variograms. Points are experimental, continuous curves the intrinsic model (all the variograms are
proportional) (b-c-d)

In Equation (4), exponent K denotes the estimate of the variable by kriging, using a set of around 50
surrounding samples that change when the location x changes (moving neighbourhood, Chils & Delfiner
1999). The samples used for numerator and denominator must be the same to preserve the non bias of the

Figure 5a presents a map of
FF (Vx ) , when Vx is sized 10 x 10 x 9 m3. Geotechnicians prefer the

reverse of the frequency because it represents the average size of non-fractured core. When this quantity
is small, the strength of the rock is bad and a low RMR is associated with the block. Another consequence
of intrinsic correlation between both terms of the ratio is that estimating the ratio or its reverse is the same
problem. Generally, this is not the case. For example, the reverse of an additive grade is not additive.

5 Crushing percentage or probability

Equation (4) is a ratio of two estimations that can be used separately. When we divide the denominator by
the sample length, we can obtain an unbiased and optimal estimate of the crushing proportion:

Geotechnical Characterization


Figure 5 Map of inverse True Fracture Frequency using block kriging (a). Map of inverse Usual Fracture
Frequency that incorporates crushing estimate and arbitrary frequency for crushing equal to 40 (b). Same as
(b) but with crushing frequency inferred from statistics and set to 80 (c). Crushing proportions at block scale
estimated by kriging (d)

Figure 5d shows a cross section of the result with important crushing proportions at the West of the domain,
that correspond to a well known damage zone due to a major fault.

6 Usual formula improvement

The intrinsic correlation between crushing and fracturing leads to the optimal and unbiased estimate of
formula (2) at block scale, for example:


Figure 5b shows a cross section of , a combination of Figure 5a and Figure 5d, with the result
that the West damaged zone is reinforced by accounting for crushing proportions.

Caving 2014, Santiago, Chile

7 Crushing frequency inference

Equation (2) shows that the coefficient a used in Equations (2) and (6) plays the role of a fracture frequency
associated with crushing and named FFcrushed. In our case, for some reasons unknown when writing this
paper, this quantity was set to 40 and the question is: could this parameter be obtained experimentally?
Let us consider the scatter diagram between Lc and FFtrue calculated using the 13,000 samples at our
disposal (Figure 6).

Figure 6 Scatter diagram between crush length (Lc, horizontal axis) and FFtrue as defined by (1). Solid line
represents the conditional expectation curve; dotted segment represents a conservative extrapolation

When Lc increases, FFtrue increases, this is a consequence of the correlation between crushing and
fracturing (the number of fractures are in average more numerous when crushing length is important). The
increasing rate is not linear but hyperbolic because we divide Nfract by a quantity that tends to zero when
Lc increase.

If we suppose that:

The crushing phenomenon appears when FFTrue is high, FFcrushed > FFTrue.

On average, FFcrushed is independent from LC, then FFcrushed can be characterised by its average (reference to
the conditional expectation curve) and must be at least equal to the limit of FFTrue when LC tends to 1.5 m.
Figure 6 shows that FFTrue = 40 for LC around 1 m. There is still a part of the sample that is not crushed, in
contrary to the previous hypothesis and FFcrushed must be at least greater than the maximum of E[FFtrue | LC]
we can calculate, here 50 at Lc = 1.14 m. If we make a crude linear extrapolation of the curve we obtain,
for Lc = 1.5 m:

FFcrushed > FFTrue = 85 (7)

As every extrapolation, this result is extremely sensitive to the hypothesis on the non linear regression
modeling. The mapping of the Fracture Frequency obtained when we replace 45 by 85 in Equation (2) is
presented in Figure 5c. Compared to the map using the traditional formula (Figure 5b), the West damage
zone is reinforced because the influence of crushing is multiplied by more than two.

8 Conclusions
Analysis of usual practices and properties of the two variables involved in the Fracture Frequency: the
Crush Length and the Fracture Number, does not require inclusion of both quantities in a single arbitrary

Geotechnical Characterization

formula. Analysis of a data set showed that both variables are statistically highly correlated as well as
spatially and they share the same variogram. This circumstance makes possible to estimate directly the real
interesting quantity that is the ratio of fractures number divided by the sample length to shortcut the lack of
additivity of this ratio. The resulting estimate is unbiased, a basic requirement when evaluating a quantity.

On the other hand, the crushing phenomena must be estimated separately, giving a crushing proportion (at
block scale) or a crushing probability (at point support scale) that must be incorporated in RMR in the same
way as FF and other geotechnical attributes.

All these possibilities depend directly on the mutual behaviour of Fractures Number and Crush Length and
any study on the subject should start by the geostatistical analysis of these two variables. A more detailed
analysis of their link, and another case study that will be published in the next future, showed that the
present observed correlation is not due to hazard: fracturing sometime contributes to crushing, sometime
not, depending on the mutual organization of the fractures. Finally, with such studies, we evaluate the
mechanical properties of the rock.

The authors would like to acknowledge Sergio Fuentes Sepulveda, Vice President of the Projects Division
of CODELCO, Chile, and his company, for their strong support in the implementation of good geostatistical
practices along the copper business value chain, as well as anonymous reviewers who greatly contributed
to improving the quality of the manuscript.

Appendix: Unbiased ratio estimation

Consider Z1(x) and Z2(x), two unknown values to be estimated using a set of 2n measurements {Z1(xi),
Z2(xi), i:1,n}. Let * denote any estimate and wi any scalars. If:


then the ratio is unbiased on average if we assume its order one stationarity at the neighbourhood scale.



If * is Kriging (whether Ordinary or Simple, Rivoirard 1984), with the same variogram for Z1 and Z2 and
same sample locations for both variables (isotopy), then the ratio is unbiased.


As the kriging weights i are identical for both terms of the ratio, we have


Caving 2014, Santiago, Chile



Equation (8) is verified, the ratio is unbiased.

The authors did not find any reference concerning Fracture Frequency estimation in Geostatistics, which
essentially focus on fracture network characterization and simulation (Chiles 1999). They notice some
papers mentioning the use of Artificial Neural Network (Fitgerald &Al 1999) and crushing phenomena is
never studied as a Regionalized Variable.

Carrasco, P, Chils, JP & Sguret, SA 2008, Additivity, Metallurgical Recovery, and Grade, in Proceedings
of Geostatistics-the Eighth International Geostatistics Congress, ed. Emery, X., vol. 1, pp. 237-

Chils, JP & Delfiner, P 1999, Geostatistics. Modeling Spatial Uncertainty, Wiley, 703 p.

Chils, JP & de Marsily, G 1999, Stochastic models of fracture systems and their use in flow and transport
modeling, in Flow and Contaminant Transport in Fractured Rock, eds. Academic Press, San
Diego, Ca, Chapter 4, pp. 169-236.

Emery, X 2007, Reducing fluctuations in the sample variogram, Stochastic Environmental Research and
Risk Assessment, vol. 21(4), pp.391-403.

Fitzgerald, E. M., Bean, C. J., Reilly, R., Fracture-frequency prediction from borehole wireline logs using
artificial neural networks, Geophysical Prospecting Journal, vol. 47, N 6, pp. 1031-1044.

Matheron, G 1962, Trait de Gostatistique Applique, Tome I, Mmoire du Bureau de Recherche

Gologique et Minires, No 14, Editions Technip, Paris, France.

Matheron, G 1963, Principles of Geostatistics, Economic Geology, vol. 58, pp. 1246-1266.

Rivoirard, J 1984, Le comportement des poids de krigeage, Doctoral Thesis, E.N.S. des Mines de Paris.

Wackernagel, H 1995, Multivariate Geostatistics, Springer, Berlin.

Geotechnical Characterization

Geomechanical ground control in block/panel caving

J Daz DERK Ltda., Chile

Y Seplveda DERK Ltda., Chile
P Lled DERK Ltda., Chile

Numerous underground mining projects that use caving methods will be in operation in the next 10 years.
For example, we have the case of Chuquicamata Underground Mine (2018) and El Teniente New Mine
Level (2017) in Chile, Oyu Tolgoi in Mongolia (2015), and Grasberg Block Cave (2017) in Indonesia,
among other smaller projects. Basis for their success is both the constructive capacity of their mining
and major infrastructure works in the initial-phase and in the projects operation. On the other hand,
currently, there are many mining sites operating that apply caving methods, which have faced different
geotechnical contexts and various operational challenges, where the ground geomechanical control has
been fundamental in the mine development and a complement to operational decisions. This geomechanical
control directly receives the acquired experience, both bad and good, of what we are doing as a mining

A field geomechanical engineer has the main goal of controlling the potential deviations from the various
operational activities that directly influence the geotechnical design and planning parameters defined in
the previous project study stages and, in turn, facing the short-term geomechanical problems that typically
arise from the daily field activities.

From the geomechanical viewpoint, three general lines associated with the field control work can be
established; namely, (1) Control line associated with Mine Preparation (developments construction); (2)
Control line associated with Mine Production activities (short-term planning and mining), both in a short-
term horizon, and finally, (3) Control line associated with the Principal and Permanent Infrastructure in a
short and medium term horizon.

This work consolidates and summarizes the major activities to be executed by a field geomechanical
professional during a mine shift, the types of outputs or deliverables of his technical work to be normally
used during the caving of a sector and the resources that must be available for its adequate field geotechnical
or geomechanical execution.

Finally, this work includes a general description of the most adequate technical profile for the professional
that will be in charge of the field geotechnical or geomechanical control in a short and medium term

1 Introduction
As shown in Figure 1, Geomechanics identifies three major focuses or clients: Planning, Design and
Mine Operations, the latter one being the context of interest for this work.

Caving 2014, Santiago, Chile

Figure 1 General diagram with main focus on the Geomechanics discipline in caving mining

In general, the successful preparation and exploitation of a mining project according to quality standards
and in geomechanical terms is based on the adequate and permanent control, recording and analysis made
on the mines short term activities.

The objective is to know and control the rock mass and cave behaviour since their beginning and therefore,
proactively detecting, evaluating and correcting any sign of deviation that could cause any type of potential
geotechnical hazard that affects the normal operation, safety and continuity of the methods productive
process and its associated major infrastructure.

Hence, the routine visual inspection is vital in the work of a field geomechanical engineer, which must be
carried out on a daily basis, supported by instruments and geological geotechnical characterization for an
adequate recording of the information.

This visual inspection is developed in all the mine levels and on the surface with the frequency and
periodicity associated to the importance and timing of the facilities. That is why the inspections at the
strategic levels and at the highly geotechnically exposed levels (i.e., ventilation levels, size reduction,
main accesses, permanent transportation and infrastructure, among others) are usually scheduled in more
extended periods, according to the geomechanical context of each site and that usually can be: once a week,
once a month, once every semester and even once a year.

2 Operational geomechanical control lines

Based on the schematic in Figure 1, we can establish three general lines related to field control, in particular:

1. Control line associated to Mine Preparation (developments construction).

2. Control line associated to Mine Production activities (Short term planning and draw rate), both
in a short term horizon, and finally
3. Control line associated to Principal and Permanent Infrastructure in a short and medium term

Geotechnical Characterization

The first geomechanical control line considers activities related to the field of construction quality of a
design, such as:

Marking, drilling, blasting and number of draw bell commissioning stages to control damage and
achieve designs (i.e., rock brow width at draw point, draw point support damage, etc.)

Marking, drilling, blasting and amount of caving runs to control damage and achieve designs
(i.e., prevent remnant pillars and support damage at the undercutting and production level,
prevent geometrical singularities in the undercutting level, etc.).

Review of the topographical control of the advance in the production and undercutting level
infrastructure (mining pattern), specifically, pillar overbreak.

Systematic and definite advance support installed: pre-mining zone, and transition zone
established in an as-built ground support drawing.

The second geomechanical control line considers activities related to the field of planning, such as:

Advance sequence management, mainly to allow the evaluation of the undercutting front
geometry and orientation.

Strategy to mine the mineralized and in situ column to assess the caveback geometry and caving
propagation .

Subsidence propagation and behaviour; crater and influence zones.

The third control line specifically refers to the periodical control of the status of the service and major
infrastructure work that is operational and/or in a preparation stage in a short and medium term horizon.

A well documented and timely field geomechanical control will favour continuous improvement in the
medium long term mining, especially giving a feedback to the design and mine planning areas (Figure 2).

Figure 2 Feedback from operational planning to the short, medium and long term planning and mine design

3 Deliverables and activities

The field geomechanical control considers the observation and recording of various pieces of information.
If the information is timely analysed and evaluated, it allows improving mining and foresee important
geotechnical hazards. In the following, we describe the main activities and deliverables associated to the
operational geomechanical work:

Caving 2014, Santiago, Chile

1. Geomechanical Behaviour As-Built Drawing: This drawing corresponds to the recording of the
excavation behaviour in terms of associated damage and presence of abnormal conditions, such
as draw point hang-ups, brow damage, notorious overbreak, loaded pillars, etc. Damage can be
observed both in the installed ground support and in the rock mass. The objective is to qualify
and evaluate the type of damage or condition and their evolution through time to determine
their potential causes and to adopt remedial, control or restriction measures in the operational
activities. It is important to verify if there is any relation between the documented behaviour,
the mining in the sector and the presence of any geotechnical hazard that is typical of the caving
method. Monthly delivery.

2. Geomechanical Hazards As-Built Drawing: This drawing maintains the records of all the
geomechanical events or hazards occurred during the mines life. This allows the traceability
of the information to make evaluations at different periods (retrospective, present and future)
that can contribute to the identification of any characteristic phenomenon, their conditions and
therefore their potential solutions. These drawings must include the potential geomechanical
hazards (i.e., collapses, airblasts, rockbursts, etc.) identified by the Project and that need to be
validated period by period with updated geological geotechnical information. Monthly delivery.

3. Ground Support As-Built Drawing: This drawing corresponds to a detailed record of all the
ground support installed both in the advance and definite excavations and pillars. This drawing
specifies the type of support system and element, amount, distribution and their relevant
characteristics. Based on these records, an evaluation is made to know if what has been installed
actually corresponds to what is established in the design guidelines or not. If there is any deviation,
it must be reported and corrected. As the former one, this record or deliverable can also be used
for later studies and retrospective geomechanical hazard analyses. Monthly delivery.

4. Draw bell Commissioning Control: This control consists of recording the draw bell development
status, its characteristics (commissioning in 1, 2 or 3 stages), location and amount. This
information allows controlling the commissioned draw points, the achievement of the defined
design (i.e., brow width) and the draw bell commissioning rate in agreement with the established
mining plans. Weekly or as required delivery.

5. Undercutting Control: This control consists of recording the undercutting front advance, the
characteristics and conditions of the commissioned area. This implies controlling the amount of
blasted rounds, amount of associated explosives, round condition (charged and cut blastholes),
type of undercutting (low or high), undercutting design, and blastholes firing diagram. In
addition, this information allows controlling the undercutting front geometrical characteristics,
presence of remnant pillars and association of potential damage due to the execution, and the
area commissioning rate, characteristics and amount, in agreement with the established mining
plans. Daily delivery.

6. Review of Development Advances: This implies having available a development progress

drawing with the condition of the daily advance fronts condition to detect any instability
condition or failure mechanism that recommends applying preventative or control measures.
In addition, this review includes the topographical activity records in order to characterize the
excavations developed, which allows detecting potential overbreaks, underbreaks and/or their
deviations. This could influence the infrastructure construction quality in the levels, namely, the
condition of the draw rate module (excavations and pillars) in the production level. Monthly

Geotechnical Characterization

7. Draw Control: This control consists of recording the draw at the draw points to create draw iso-
curves, to verify the low and high draw, executed rates in the front perimeter and in the caving
regime zones. The draw rate is a parameter that is related to seismic activity and the geotechnical
hazard potential in caving. Weekly delivery and monthly summary.

8. Fragmentation Control: This control consists of recording the particle size by means of digital
photography at the draw points in a systematic manner, correlating it with the draw point life
stage, associated column height and draw rate. Occasionally, its necessary to create control
sheets for each draw point and monthly control. Weekly delivery and in accordance to the draw

9. Subsidence Advance Control: This control consists of recording the different types of damage
associated to subsidence zones. Ideally, this recording must be made in old workings located in
the sub-surface intermediate zone and on the ground surface. This information allows controlling
the crater geometry, its behaviour, influence zone and correlating its growth with the existing
mining activity. If possible, this work must be supported by semi-annual or annual ortho-
referenced and/or satellite aerial photography. Semestral and annual delivery.

10. Seismicity Review: This review implies updating and reviewing the daily, weekly and monthly
seismic records statistics and trends. This type of instrumentation and information allows
analysing the behaviour in zones that cant be observed by the human eye, namely, the
behaviour of the caveback zone in the areas and their relationship with draw rates and caving
propagation. Daily, weekly and monthly delivery.

11. Blasting Damage Characterization and Evaluation: In specific situations, mainly when there is
no dedicated service to control the blasting-induced vibrations (development and production), it
is necessary to have the field Geomechanical Engineer carrying out some measurements and near
and/or far field vibration analysis, as required, to evaluate the effects of the development and
production blasting in the rock mass and surrounding excavations, in addition to the construction
and/or calibration of the damage models available for the geotechnical units. This commonly
happens with the draw bell commissioning blasting (phases) or caving (undercutting) blasting
that creates potential damages in the production and undercutting level pillars and excavations.
Delivery frequency according to studies.

12. Review of Geotechnical Instrumentation: This implies being updated and reviewing the statistics
and trends of the installed instruments monitoring on a daily, weekly and monthly basis
(production level pillars, undercutting, pre-mining zone, transition, de-stressing), based on the
monitored parameters: deformation, displacement, and stress state, among others.

Figure 3 shows the documents scheme for the deliverables developed by the operational geomechanical
control and their delivery frequency. This example underlines the importance of including the operational
geomechanical works in the mine integral management system through each one of the updated and
available working procedures (DERK 2013).

Caving 2014, Santiago, Chile

Figure 3 Document management examples for the operational geomechanical control in caving (DERK 2009;
Daz 2010; DERK 2011)

3 Resources required
This section provides a summary of equipment, instruments, tools and accessories that must be included in
the basic resources that must be available for the field geomechanical control personnel:

Personal computer with Windows platform, with Office, Grapher and Autocad software

Software for fragmentation estimation from digital photography processing.

Tablet for damage digital mapping and field geotechnical information.
State-of-the-art personal protection equipment and/or elements.
Compass and 50 meter measuring tape.
Digital distance meter.
Photo camera and high-resolution digital film camera.
Long range flashlights.
Tripod halogen floodlights.
Geology hammer.
Triaxial geophones and/or accelerographs and cables.
Seismograph (for example, Minimate Plus Equipment).
Borehole Camera equipment.

Geotechnical Characterization

Adhesive agent as required by humidity and pollution (for example, Poxipol).

Laser Scanner equipment (for example, Mine Scan Geosite).
Logbook in three copies.
4-wheel drive diesel light truck equipped with radio broadcasting.
1 field technical assistant.
Drillhole lifting equipment (drillhole deviation).

4 Technical profile
There are several technical aspects that must be known by the professional that will work on the field
geomechanical control or in the operational Geomechanics. In particular, he/she must know and manage
concepts related to the following:

Operational processes in a mine that uses caving methods.

Mine Caving planning and design.
Geology and geotechnical aspects associated to caving.
Induced seismicity.
Civil works construction.
Management of support computer tools for geomechanical analysis and evaluation; i.e.,
Rocscience, Autocad, Grapher, etc.

Geotechnical instrumentation; i.e., convergence stations, extensometers, stress measurement,

TDRs, among other typical instruments applied.

Vibration control and modelling of blasting-induced damage.

The professional shall preferably be a Mining Civil Engineer, with minimum general experience in
underground mining from 2 to 5 years, and with specific expertise in Geomechanics for caving mining of
at least 1 to 3 years.

5 Conclusions
The short-term geomechanical knowledge management and documentation in a mine site is the responsibility
of the operational Geomechanics and it is of vital importance for its medium to long-term sustainability.

An efficient and constant communication among the mine operations areas, planning, geology and
Geomechanics is fundamental to successfully mine a productive sector and to control its losses.

The on-going training of the mine personnel on operational Geomechanics increasingly reduces the
deviations that can take to geotechnical hazards that affect safety and the continuity of the productive

The operational geomechanical management must be included in the mines integral management systems
through its updated and available working procedures.

Caving 2014, Santiago, Chile

DERK Ltda. gives thanks to its geomechanical team who worked with a high quality in El Salvador
Division contracts (Codelco Chile) since 2009 to 2013.

DERK 2013, Sistema de Gestin Integral DERK Ingeniera y Geologa Ltda.

DERK 2011, Servicio de asesora integral en geologa, geotecnia, geomecnica y planificacin minera
Divisin Salvador, Gerencia de Operaciones Minas-Plantas, Superintendencia de Geologa y
Planificacin Minera-Metalrgica, Divisin Salvador Codelco Chile.

Daz, J 2010, Productos tcnicos servicio geomecnica operacional mina subterrnea (SGOMS), DERK
Ltda. PR-DSAL_SAEG_MS-P02-01-2011, Servicio de Asesora Integral en Geologa,
Geotecnia, Geomecnica y Planificacin Minera Divisin Salvador de CODELCO CHILE.

DERK 2009, Servicio de asesora integral en geotcnia y geomecnica Divisin Salvador, Gerencia
de Operaciones Minas-Plantas, Superintendencia de Geologa y Planificacin Minera-
Metalrgica, Divisin Salvador Codelco Chile.

Caving 2014, Santiago, Chile

Gravity Flow

Gravity Flow

Use of experiments to quantify the flow-ability of caved

rock for block caving
RE Gmez, University of Chile, Chile
R Castro, University of Chile
D Olivares, University of Chile, Chile

Block/panel caving mining is a massive underground method, in which an ore column of broken rock is
generate above the production level, as the cave propagates upwards through the ore body. ,As reserves
deplete from near surface, the next generation of block caves will be carried out in deeper condition than
currently known, with large column heights and therefore higher vertical stresses. There are unknowns
related to the flow characteristics that deeper caves would face. The aim of this study is to quantify the impact
of large vertical pressure on the flow-ability of fragmented rock. For this reason, experiments representing
the stress and geometry conditions of deep caves were conducted under a range of vertical pressures,
materials and humidity conditions. The results indicate that the flow-ability of caved rock depends on the
vertical stresses, fines content and humidity conditions.

Keyword: caving mining, flow-ability, hang up, vertical stresses, fines and humidity content.

1 Introduction
In block/panel caving, ore production is affected by interferences associated with the caving process,
especially, those related to the gravity flow, such as, hang-ups and oversize rocks. The mine design capability
to provide a given production rate is affected, among other factors, by the ore flow.

Flow-ability is defined as the flow condition or ability of a granular material to flow under a given set
of material properties, infrastructure geometry and stress conditions. The flow-ability can be classified
into free flow, intermittent flow, assisted flow and no-flow (Castro 2014). Kvapil (2008) indicates that
flow-ability depends on many parameters including particle size, extraction rate, particles shape, surface
roughness between particles, friction between particles, moisture content, compressibility, compaction,
particle resistance, and magnitude, distribution and direction of external loads and forces. However, despite
being listed, the flow-ability under all those sets of parameters has not been quantified.

Flow-ability could be characterised both qualitatively and quantitatively. In terms of qualitative

characterization, the flow could be qualified as free flow, intermittent flow and no flow, depending on the
ratio between particle size and opening (Laubscher 2006). Studies on gravel have shown also that the flow-
ability of granular material is influenced also by the vertical load (Fuenzalida 2012). Castro et al. (2014)
have proposed a flow-ability chart for coarse and dry rock, which is presented in terms of vertical stress
and drawpoint width/d50.

Caving 2014, Santiago, Chile

Figure 1 Influence of vertical stress in flow-ability

In quantitative terms, flow-ability can be characterised in terms of the number of hang-ups every 1,000
tons or broken rock drawn. Hang ups are one of the flow interferences that affect productivity. Moreover,
hang ups can be used to measure the flow-ability of material because when a hang up occurs it means the
flow of material has been interrupted and the broken rock cannot be extracted (Troncoso 2006). Two kinds
of hang-ups can be formed in coarse material: cohesive and mechanical (Kvapil, 2008. Beus et al., 2001;
Hadjigeorgiou & Lessard 2007). The formation of arches on a rough wall is generated by the rotation of
the principal stresses on the wall and by induced wall pressures (Handy 1985). The dimensions of the arch
depends on the friction angle of the material, depth or vertical stress, inclination of the walls in a drawpoint,
draw rate, shape and strength of the particles, and humidity (Kvapil 2008). At the mine it has been observed
that as more material is extracted from a drawpoint the frequency of the hang ups decreases (Maass 2013).
This phenomenon is probably related to the decrease of the particle size during the extraction of an ore
column (Montecino 2001).

There are many unknowns related to the flow of materials especially under confined conditions. For example,
what is the role of the fines, water and stresses on the flow-ability of the broken rock. In this article, we
present the experiments conducted to evaluate the flow-ability of caved rock under high vertical load for
different fines and humidity conditions. Extraction is carried out by a scaled LHD system to represent
current caving characteristics.

2 Laboratory scaled model and material characterization

2.1 Experimental set up and materials

The experiments were conducted in a set up to study confined flow. This consist of a steel cylinder, which is
filled with broken rock (70-80 kg of crushed ore) under a hydraulic press machine with a capacity of 1,800
kN. The steel cylinder has diameter of 340 mm, as shown in Figure 2. A cylindrical shape was chosen to
avoid the concentration of stresses at singularities. The height of the designed cylinder is 700 mm in order
to hold the desired volume of gravel and to suit the emulated Andina mine drawbell with a scale of 1:75
(Figure 2b), with a rectangular opening of 53 x 96 mm2. Since the drawbell is located in the center of the
cylinder, ow zones will not intersect the walls of the model. A steel extraction system was built to replicate
the extraction the same as LHD does from an extraction point (Figure 2a).

Gravity Flow

Figure 2 Experimental equipment: (a) cylindrical model in a press machine which changes the vertical load,
V, and (b) extraction system, located in the bottom, center of the model

The material used in the experimental tests was crushed sulphide ore with a high aspect ratio to represent the
geometry of caved rock (sphericity of 0.58 and a roundness of 0.25). Two different particle size distributions
of this material were prepared and tested: one with a passing size d80 of 11.8 mm and the other one with a
d80 of 15.6 mm. Both samples have the same uniformity coefficient (Cu=d60/d10) of 2. Those particle size
distributions were scaled from the size distribution of the primary fragmentation curve of the undergrounds
Chuquicamata project (Figure 3). Table 1 summarizes the characteristics of the samples.

Figure 3 Particle size distribution of samples used in the experiments

Caving 2014, Santiago, Chile

Table 1 Samples characteristics

*Drawpoint width for non-square geometry can be represented by hydraulic diameter (Jennings & Parslow

2.2 Experiments

A total of 18 experiments have been carried out to date described in Table 2. Ten experiments were
performed without fines or humidity in order to define a base case considering different vertical loads.
Then humidity and fines were added to the samples to measure their impact on the flow-ability. Fine
material used in this study had a uniform distribution with d100 equals 1 mm. Humidity used in this study
was 1.5 liters per 10 kg of fine material that is 15% of water.

Table 2 Summary of experimental conditions

Vertical load Humidity Size d80

Test Material Fines [%]
v [MPa] [%] [mm]
1 0 0 0 11.8
2 1.5 0 0 11.8
3 A1 3 0 0 11.8
4 6 0 0 11.8
5 10 0 0 11.8
6 0 0 0 15.6
7 1.5 0 0 15.6
8 A2 3 0 0 15.6
9 6 0 0 15.6
10 10 0 0 15.6
11 0 0 20 15.6
12 6 0 20 15.6
13 0 0 40 15.6
14 6 0 40 15.6
15 0 15 20 15.6
16 6 15 20 15.6
17 0 15 40 15.6
18 6 15 40 15.6

Gravity Flow

3 Results
The experiments were performed twice for coarse ore and once for fine material (the latter only for one
particle size distribution). During the tests, the flow-ability, the hang up frequency and the hang up height
were recorded.

3.1 Flow-ability

Flow-ability is classified into free flow, intermittent flow, assisted flow and no-flow (Figure 1). In terms of
flow-ability, the results (Table 3) indicate that for materials A1 and A2 flow-ability decreased from free flow
to no flow when vertical load increased from 0 to 10 MPa. When fines were added without humidity, the
flow condition was intermittent or assisted. When water was added, the flow was assisted flow and, when
there were a 40% of fines, the flow condition was no flow at all.

Table 3 Summary of experimental results

Vertical load v Interferences Standard dev.

Test Material Flow condition
[MPa] [g/hang up] [g/hang up]
1 0 Free Flow 1246 640
2 1.5 Assisted Flow 928 371
3 A1 3 Intermittent flow 1068 256
4 6 Assisted Flow 368 246
5 10 No Flow 0 -
6 0 Free Flow 1177 471
7 1.5 Intermittent flow 1036 471
8 A2 3 Intermittent flow 761 356
9 6 Assisted Flow 599 276
10 10 No Flow 0 -
11 0 Intermittent flow 1014 248
12 6 Assisted Flow 586 312
13 0 Assisted Flow 501 153
14 6 Assisted Flow 475 180
15 0 Assisted Flow 352 97
16 6 Assisted Flow 378 112
17 0 No flow 0 -
18 6 No flow 0 -

3.2 Hang up frequency

During the experiments, it was possible to detect flow interruptions or hang-ups, which were recorded in
terms of mass and height. The hang up frequency (Hg) is defined as the amount of material that can be
drawn before an interruption happens. The experimental results of Hg as a function of the vertical stress for
each laboratory test are presented in Figure 4.

Caving 2014, Santiago, Chile

Figure 4 Hang up result in laboratory test. A: coarse material test (duplicates included), B: 20% of fine
material, C: 40% of fine material and D: 20% of fine material with humidity

The experimental results in coarse material (A) show that increasing the vertical stress decreases the flow-
ability of material. For the media B, that is the one with 20% of dried fines, the hang-up frequency number
decreased. For materials C and D, the vertical load had no significant influence on the frequency of hang-
ups as they were not able to flow.

Field measured hang-ups are quantified by their hang-ups index (number of hang-ups per 1,000 ton of ore).
The measured hang-ups index in the experiments is similar to the observed index of primary sulphides in
mines. The index varies from 1.6 to 3.6 in mines and, as can be seen in Table 4, the scaled experimental
index varies from 0.75 to 3.95.

Table 4 Scaled hang ups index

Vertical load Hang ups index [# hang up/ 1000 ton]

[MPa] A1 A2
0 1.05 2.05
1.5 1.26 2.29
3 1.09 3.11
6 3.17 3.95
10 0.75 3.32

For the experiments with vertical load of 10 MPa the material got strongly compacted over the drawbell
generating a great hang-up above the drawbell and almost no hang-ups of lower height.

Gravity Flow

3.3 Height of hang-ups

A classification was made according to the height of the hang-up:

Low: in extraction point.

Medium: in drawbell (0-13.5m from the roof of the production level)
High: Above drawbell (over 13.5 m).
The geometry of the drawbell from which the dimensions were scaled is represented in Figure 5.

Figure 5 Drawbell geometry in real size

Average hang-up height of each experiment based on the vertical stress is shown in Figure 6. It can be seen
that as the vertical stress increases, the height of the hang-ups increases simultaneously. The dimensions of
each observed hang-up were scaled in order to quantify their height in the mine.

Figure 6 Hang-up heights in laboratory test. On coarse material A1: d80=15.6 [mm], A2: d80=11.8 [mm], B:
20% of fine material, C: 40% of fine material and D: 20% of fine material with humidity

In general, the height of the hang-ups increases with the vertical pressure for the coarse material (A1 and
A2). In the case of the addition of fine (materials B and C) the results indicate that vertical load has a small
impact in the increase of the height of the hang-ups. When fines and water were added (material D) there is
no effect of the vertical load on the height of the hang-ups.

Caving 2014, Santiago, Chile

4 Conclusions and discussion

Based on experimental tests, this paper shows that particle size as well as the moisture content and vertical
stress have a noticeable impact on the flow-ability of caved rock. The number and height of hang-ups
increases as the vertical load increases. Fines and humidity increase the number of hang ups. Also, high
hang-ups occurs when a vertical pressure was applied or when humidity and fines were acting together.

The scaled model was successful in understanding how confinement, particle size distribution, humidity and
fines presence affect the flow-ability of material. It is expected that as block caves get deeper the number
of hang ups would increase if fragmentation keeps constant. The results of the above experiments shows
that the hang-up number for mine design applications could be obtained from this kind of experiments.
This would require further research and analysis. It is expected that this kind of experiments would be will
become standard to the caving industry, especially for future and unknown conditions.

We want to thank Patricio vila and Alonso Vives for their help in these experiments, the staff of Block
Caving Laboratory for their support and encouragement and Asieh Hekmat for her advices. This project
was conducted under the partial funding of the basal project for Basal Excellences, which is an initiative of
the Chilean government.

Beus, M, Pariseau, W, Stewart, B & Iverson, S 2001, Design of Ore Passes, in Underground Mining
Method, pp. 627-634.

Castro, R, Fuenzalida, M & Lund, F 2014, Experimental study of gravity flow under confined condition,
International Journal of Rock Mechanics and Mining Sciences, vol 67, pp. 164-169.

Fuenzalida, MA 2012, Study of the confined gravity flow and its application to caving, Masters Thesis,
Santiago, Chile, Universidad de Chile. (in Spanish).

Jennings, BR & Parslow, K 1988, Particle size measurement: the equivalent spherical diameter, in
Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences,
419(1856), pp. 137-149.

Hadjigeorgiou, J & Lessard, JF 2007, Numerical investigations of ore pass hang-up phenomena,
International Journal of Rock Mechanics and Mining Sciences, vol. 44 N6, pp. 820-834.

Handy, RL 1985, The arch in soil arching, Journal of Geotechnical Engineering, vol. 111 N3, pp. 302-

Kvapil, DR 2008, Gravity flow in sublevel and panel caving A common sense approach.

Laubscher 2006, Cave Mining Handbook.

Maass, S 2013, Technological alternatives for hang ups removal, Masters Thesis, Santiago, Chile,
Universidad de Chile. (in Spanish).

Montecino, N 2011, Secondary fragmentation mix model in block/panel caving mining. Masters Thesis.
Santiago, Universidad de Chile. (in Spanish).

Troncoso, S 2006, Simulation of operational interferences impact on production planning, Thesis, Santiago,
Chile, Universidad de Chile. (in Spanish).

Gravity Flow

An analysis of the lateral dilution entry mechanisms in

Panel Caving
PS Paredes University of Chile, Chile
MF Pineda University of Chile, Chile

Dilution is an integral part of a cave mining operation and its behaviour affects different dimensions of a
caving project from the economic results to safety inside the mine.

One of the objectives of mine planning and draw control in panel caving is to maximize ore recovery by
minimizing the overall dilution content extracted. The existing gravity flow models that aim to predict
dilution entry and its behaviour do not consider the possibility of dilution particles to move laterally and
affect the content of drawpoints that are contiguous to a previously exhausted area.

This paper contributes to the understanding of this important variable in cave mining methods by presenting
an analysis of the lateral dilution entry mechanism hypothesis presented by Castro & Paredes (2014). The
objectives of this analysis were to determine the processes through which this mechanism enables dilution
particles to move laterally from a previously exhausted panel to a new panel in production and to set the
basis for further development of guidelines for mine design and planning of new panels contiguous to
previously exhausted panels. The analysis was performed with two approaches: a mechanical analysis
supported with mine data and experimental observations at a laboratory scale. The authors postulate that
there are two processes through which dilution can migrate lateral distances: lateral displacement over the
broken ore pile and lateral preferential flow from the dilution source. A mechanical analysis supported with
mine data shows that lateral displacement of dilution over the broken ore pile is feasible under cave mining
conditions. Consequently, experimental observations suggest that lateral preferential flow of dilution is
possible under certain conditions. Additionally, vertical fines migration through shear zones was observed.
Finally, for mine design and planning of a new panel contiguous to a previously exhausted panel, an
expression for the maximum caving face width as a function of several mining parameters is proposed.

Keywords: panel caving, dilution entry; gravity flow mechanics; draw control.

1 Introduction
Dilution control is crucial in managing a block/panel caving operation. This variable affects different
dimensions of a caving project from the economics to safety in the mine. Several authors have stated different
parameters affecting dilution entry (Table 1) and two main dilution behavior models have been developed
in order to provide tools for dilution entry prediction and control: Laubschers (1994) and Susaetas (2004).
Laubschers formulae for dilution entry prediction are based on the estimation of the height of interaction,
the swell factor of caved material and a measure of the uniformity of extraction. This last quantity is based
on the standard deviation of the tonnages extracted from the working drawpoints. He postulates that as the
draw uniformity increases, dilution entry will be delayed. Susaeta used mine data to expand Laubschers
model and postulated that dilution entry depends on the way draw is conducted, defining three behavioral
models: Isolated Flow, Isolated-Interactive Flow and Interactive Flow. Flow behavior can pass from one
model to another by increasing or decreasing uniformity of extraction. As the draw is more uniform, the
flow will be more interactive and dilution will be lower. Both approaches assume that the dilution source is

Caving 2014, Santiago, Chile

located above the area under analysis. Finally, they assume that the dilution particles will flow downward
to the drawpoints without being affected by the cave back propagation profile or rilling potential.

Table 1 Parameters affecting dilution (Castro & Paredes 2014)

Parameter Effect on dilution Author

Ore/waste To minimize dilution, ore/waste contact interface must be kept DeWolfe
contact surface at 45 to 50 moving away from the uncaved ore through draw (1981), Julin
inclination control. (1992)
Ore volume
The higher the ratio of ore volume to the surface area of the ore/ Laubscher
to ore/waste
waste interface, the lower the overall percentage of dilution. (2000)
interface area
Attitude of ore/
waste interface Irregular ore/waste interface areas will cause severe dilution entry.
Fragmentation Finely fragmented waste and coarse ore means early and extensive
range from ore dilution, whilst coarse waste and fine ore means a low overall
and waste extracted dilution percentage.
Good draw zone interaction and parallel flow will represent the
Height of the Laubscher
optimum conditions. Poor drawpoint interaction and draw zones
interaction zone (2000)
angled according to local variations will lead to high dilution.
Differences in
High-density ore and low-density waste lead to low dilution and Laubscher
density of ore and
vice versa. (2000)
Block or panel A block cave strategy will lead to more lateral dilution mixing than Laubscher
caving strategy panel caving. (2000)

Uniformity of High uniformity of the tonnage drawn from the neighbour Julin (1992),
draw drawpoints will lead to high interaction and late dilution entry. Susaeta (2004)

Despite the fact that the latest field studies using markers have shown how chaotic ore flow behavior is in
the near field (Brunton et. al. 2012; Castro & Armijo 2012), there is still significant potential to conduct far
field experiments to better understand dilution behaviour in the far-field.

Empirical observations of the historical dilution behavior at CODELCOs El Salvador and Andina mines
conducted by Castro & Paredes (2014) showed that 51% of the drawpoints with dilution (over a total of
1674 drawpoints with dilution analyzed) had a pulsed-shaped behaviour in which the dilution goes back to
zero after its first appearance. This non-continuous dilution entry behavior led to the definition of a dilution
entry criterion based on the cumulative dilution curves of the drawpoints. Thus, when the cumulative
dilution curve overcame a certain threshold (3% for Castro & Paredes analysis) significant dilution entry
was declared for the drawpoint at its corresponding extraction percentage. Using this criterion, Castro &
Paredes analyzed the dilution entry behavior at a drawpoint scale for 6 different sectors considering the
initial in-situ column heights, extraction sequence and uniformity of extraction using Susaetass Uniformity
Index. This analysis led to the development of a hypothesis for the mechanisms by which dilution entered
the drawpoints: (1) vertical entry by descending gravitationally from the source located above the sector,
(2) dilution entry after an air blast event due to sudden propagation of the cave back when new area is
incorporated, and (3) lateral dilution due to horizontal displacement of dilution particles from a source
located next to the panel under analysis.

Gravity Flow

The following sections correspond to an analysis of the lateral dilution entry mechanism hypothesized by
Castro & Paredes, incorporating experimental results derived by Pineda (2012), in order to contribute to the
understanding of dilution behavior in Panel Caving methods.

2 Lateral dilution entry mechanism hypothesis

According to Castro & Paredes (2014), when the dilution source is located next to the panel under analysis,
the cave back could preferentially propagate, in terms of rate, to the lateral interface of the in-situ and
caved waste rock. This preferential propagation of the cave back could enable dilution particles to enter
the draw columns and flow along the cave boundary (as observed in sand flow experiments by Kvapil
(2004)), travelling horizontal distances depending on the piles slope and producing early dilution entry
in drawpoints located far from the dilution source. This phenomena was observed at the Inca Central East
(ICE) and the Inca North (IN) sectors in El Salvador mine and at the Grizzly Cluster of Andina mines Panel
III sector.

Figure 1 Schematic sequential drawbell sections, showing lateral dilution mechanism (Castro & Paredes 2014)

Figure 2 illustrates ICEs drawpoints dilution entry behavior: as the sequence progresses going from the
dilution source (IC previously exhausted panel), allocated to the north of ICE, towards the south; dilution
particles enter the drawpoints earlier (DEP decreases as drawpoints are more distant from the dilution

According to Castro & Paredes, this behaviour suggests that dilution particles coming from the lateral
dilution source travelled large lateral distances rilling over the broken ore pile, following the process
described in the previous paragraphs.

The authors identified a second mechanism governing lateral dilution entry, that is, lateral percolation
preferential flow of particles from the dilution source into the drawpoints of a recently incorporated area.
Under certain conditions, when the new broken ore is in contact with waste, there is a potential for the
movement zone to develop towards the higher porosity zone, since it represents lower strength path for the
movement to develop. This, in the case of a new panel contiguous to a previously exhausted panel, could

Caving 2014, Santiago, Chile

promote lateral preferential flow of dilution particles into the new drawpoints. There is a laboratory scale
evidence of this phenomenon as observed in the experimental results derived from the use of a physical
model by Pineda (2012).

Figure 2 Plan view and Schematic cross-section of an ICE sectors drift showing decrease in DEP as the
sequence progresses from the lateral dilution source

3 Lateral displacement mechanical feasibility analysis

In order to probe the mechanical feasibility of this hypothesis, a limiting equilibrium analysis of the broken
material wedge, formed by the existence of an air gap, was performed. This consisted of analysing the
stability of the parameterized problem under the conditions in a panel cave operation (Figure 3).

: Failure surface angle.

: Granular material pile slope.
W: Vertical force due to the wedges weight.

z: Force applied by the overload.

, , c: Bulk density, friction angle and cohesion of the broken
h: Air gap height.
b: Length of the horizontal projection of the failure surface.
z: Overload height.
L: Failure surface length.

Figure 3 Limiting equilibrium analysis theoretical approach

The equilibrium analysis can be reduced to the determination of the safety factor (SF) defined as:

Gravity Flow

SF = FFE [1]

where FE and FD are the summation of stabilizing and destabilizing forces acting over the failure surface,
respectively. If the SF value is less than 1, the broken material wedge is unstable and the material will slide
over the failure surface to enter the granular material pile. Considering g as the acceleration of gravity, and
the Janssen (1985) approach for the force due to the overload, the destabilizing forces can be expressed as




For the broken material, the stabilizing force corresponds to the shear strength of the failure surface. Using
the Mohr-Coulomb criterion, the stabilizing forces can be expressed as follows:


Thus, an expression for the safety factor as a function of the problems parameters is:


From the evaluation of the safety factor using the geotechnical characterization of the broken material from
the Panel III sector, it is possible to observe that for overload heights over 8 m, the formation of a failure
plane with an incline of over 35 will cause instability in the wedge even for a 10 cm air gap height.

The previously described condition is sufficient to allow the detachment of a group of dilution particles.
Nevertheless, to allow a significant dilution entry on the granular material pile, the air gap height must be
enough to allow dilution particles to overcome the arching effect described by Hoek (2004). Hoek (2004)
states that the opening through which a group of particles can flow must be greater than 3 times the average
particle size in order to overcome the arching effect. Thus, considering Panel IIIs conditions, an air gap
height over 60 cm could allow a significant number of dilution particles to enter the granular material pile
(Paredes 2012).

Once dilution particles have been detached from the lateral dilution source and entered the granular material
pile, lateral displacement over the pile will be controlled by the piles slope () and the angle of repose of
these dilution particles (d). If is greater than d, dilution particles will be able to rill over the granular
material pile. In a Panel Caving operation, will take different values depending on the extraction rates,
and exceeding d in some cases. If we consider the long term case, where the granular material pile finds its
angle of repose (p), lateral migration of dilution particles will occur if this angle (p) is greater than d.
Considering Panel IIIs geotechnical characterization of dilution and ore, the angle of repose (32 and 53,
respectively) would allow the occurrence of lateral migration of dilution over the granular material pile.

4 Lateral preferential flow observations at lab scale

At the request of Agnico-Eagle, a physical model was built and tested by the Block Caving Laboratory
at the University of Chile in order to study the gravity flow mechanisms of a large stope (320 m height
and 45.000 m2 of production area) extracted using a block cave layout for blasted material. This free flow
material condition emulates a panel caving operation once the whole ore column has been caved.

Caving 2014, Santiago, Chile

4.1 Physical model experimental set-up and plan

For experimental purposes a 1:200 scale model was built. During the design stage of the experiments, all
the laws of kinematic similitude for granular materials (full geometrical similitude and extraction rate) were
taken into consideration as detailed by Castro et al. (2012).

Table 2 describes the experimental set-up and Figure 4 presents the experimental design.

Figure 4 Experimental design: Physical Model (right) and drawpoints and apex (left)

Table 2 Experimental set-up

4 dismountable plexiglass walls delineating the final geometry of the stope

Main Assembly Plexiglass assembly supported by iron vertical columns and rows
Dimensions of the model 1.6 m height x 1 m length x 0.25 m width at lab scale
Loading System Material and labelled markers loaded manually

11 drawpoints, each one includes shovel linked to a servomechanism device that

Extraction simulates the LHD extraction
Material System
Servomechanism allows varying the extraction rate
Coarse gravel with a d100 of 4.75 mm and a d50 of 2.2 mm allowing free flow
according to Hustrulid and Sun (2004)
Model Media Material dried to eliminate the humidity effect and capillarity between grains
For simulating dilution entry the model media is dyed using red dust

The experimental plan and its main objectives are outlined in Table 3.

Table 3 Experimental Plan

Experiment Draw strategy Objective

To determine isolated flow zone diameter geometry and to test the
1 Isolated draw
proper operation of the model.
This experiment simulates continuous dilution entry at the top of the
2 Uniform draw
stope. The aim is to quantify potential dilution entry mechanism

Gravity Flow

4.2 Experimental Results

4.4.1 Experiment 1: Isolated draw

In a first stage, an Isolated Draw Zone experiment was performed to calibrate the correct operation of the
extraction system for the physical model. For this experience, the extraction was conducted only in one
drawpoint. Once the Isolated Movement Zone (IMZ) reached the surface of the stope the extraction was
completed and the IMZ was measured. The main results of this experiment can be observed in Figure 5.

Figure 5 Progress of the Isolated Draw Zone

4.4.2 Experiment 2: Uniform draw with continuous dilution entry from the top

An experiment simulating dilution entry at the top of the stope was performed. Dilution was simulated
by adding a granular material on top of the stope dyed in red colour. This experiment was performed by
extracting from two extraction levels; the main extraction level located at the base of the model and a
secondary extraction level located in the stope footwall, 30 m (15 cm at model scale) above and 60 m (30
cm at model scale) away from the main extraction level; these distances are provided at prototype scale.
This horizontal distance was determined based on the IMZ experiment conducted previously in order to
avoid an early connection between flow zones.

The extraction for each drawpoint, in the main level, was continued until dilution was reported, at which
point draw from that individual drawpoint was stopped immediately. Afterwards, extraction was continued
for the rest of the drawpoints until all of them exhibited a significant content of dilution. Once all the
drawpoints of the main extraction level were closed, because of the dilution content, the drawpoints from
the secondary level were started into production following the same considerations stated above.

As a result from this experiment, Pineda (2012) observed how the flow zone developed faster at the lower
column height, generating an extraction profile; this profile is key in order to define the ore/waste interface
during the whole extraction process. Dilution moves downwards according to the flow velocity profile,
being faster in the proximity of the hangingwall as observed in sand experiments by Kvapil (2004). As
extraction progressed, it was noted that the fines contained in the red material (tinting red dust) migrated to
the drawpoints (Figure 6d).

Once the area relative to the extraction from the main level was exhausted (Figure 6g), the extraction from
the secondary level began. This generated a different extraction profile and promoted lateral movement
for the dilution source. Due to the lateral movement of the material, dilution particles move towards the

Caving 2014, Santiago, Chile

recently opened drawpoints diluting them in an early stage as hypothesized formerly (Figure 6h). Therefore,
in terms of mine design, a larger spacing would be recommended to avoid the flow zones to connect to the
diluting lower density zones.

Figure 6 Progress of the extraction under uniform draw

5 Avoiding lateral displacement of dilution particles over the ore pile

Considering the lateral dilution entry mechanisms described previously, in this section, an analysis of
the main variables in mine design and planning of a new panel is performed. The analysis proposes an
expression of the maximum caving face width that would prevent lateral displacement (rilling) of dilution
particles over the ore pile.

Block Caving methods need to maintain equilibrium between the material extracted and the material caved
to propagate caving during the early stages of a block or panel production. Therefore, it is not possible to
avoid the formation of short air gaps. Thus, one possible strategy to avoid lateral dilution migration over the
ore pile is to control the granular material piles slope during caving propagation.

Once the air gap is formed, if the local slope between the pile heights of two contiguous drawbells in
the sequence direction (the local pile slope) is greater than the angle of repose of dilution particles d,
dilution will be able to rill over the pile (Figure 7a). Using a simple approach, the time delay in days, defined
as the difference between the beginning of production for two contiguous drawbells, can be expressed as
a function of the new area incorporation rate (VD, in m2/day), the width of the caving face (W, in m), the
distance between the contiguous drawbells (d, in m) and the strike angle of the footprint (see Figure 7b):


Thus, for a given period during the caving propagation phase, the pile height difference between two
contiguous drawbells in the direction of the sequence, that have the same caving vertical propagation rate,
will be:

Gravity Flow


Where: VEX, in t/m2/day, is the extraction rate of the drawpoints; VCP, in m/day, is the caving vertical
propagation rate; S, in t/m3, is the ore solid density; and e is the void ratio inside the pile.

Then, the local pile slope () will be given by:


And dilution will be able to migrate laterally over the pile if:


From the parameters shown above, there are two main mining variables that can be controlled for design
and planning purposes: (1) the new area incorporation rate, and (2) the caving face width. The new area
incorporation rate is generally limited by several technical conditions, so the most flexible variable is the
caving face width. Thus, an expression for the caving face width that would ensure the prevention of early
dilution entry caused by rilling particles coming from a lateral dilution source is presented:


Figure 7 a) Schematic section view showing pile heights difference (h) and spacing (d) for two contiguous
drawbells; b) Schematic plan view of drawpoints showing caving width (W), sequence direction, contiguous
drawbell spacing (d) and footprint strike ()

The maximum caving face width for different vertical caving propagation rates Vcp is calculated,
considering the values in Table 4, as an academic example (Figure 8). From the figure, it is worth noting
that the maximum resulting caving face width varies significantly with small changes in Vcp.

Caving 2014, Santiago, Chile

Table 4 Variable values for maximum caving face width example

Variable Unit Value
e % 40
Vex t/m2/d 0.3
S t/m3 3
d 32
VD m2/d 30*

* equivalent to a rate of 2 drawbells per month in a 30 m x 15 m drawpattern

Figure 8 Example of maximum caving face with as a function of the vertical caving propagation rate
considering Table 4 values

6 Conclusions
Two hypotheses for the lateral dilution entry mechanism in panel caving were presented and described: (1)
displacement of dilution particles from a lateral dilution source over the broken ore pile (via rilling) and
(2) preferential flow of dilution particles from a lateral dilution source (via inclination of the IMZs towards
more free-flowing material). Additionally, the observation of fines vertical migration suggests that the
relation between ore and dilution particles size affects the vertical entry of dilution, as the tinting red dusts
particle size was significantly smaller than the model media used to represent ore.

Through a limiting equilibrium analysis, the first mechanism was proven to be feasible under panel caving
operation conditions. On the other hand, experimental evidence suggests that, under free flow conditions,
the second mechanism occurrence is possible.

Finally, an expression for the maximum caving face width that avoids dilution entry through the first
mechanism described is proposed for mine design and planning purposes.

Gravity Flow

7 Acknowledgments
The authors would like to acknowledge CODELCO-Chile, Agnico-Eagle and the Block Caving Laboratory
of the University of Chile for providing funding and data for the research. We would also like to thank Mr
Raul Hott for his helpful support during the writing of this article.

8 References
Brunton, I, Sharrock, G & Lett, J 2012, Full Scale Near Field Flow Behaviour at the Ridgeway Deeps
Block Cave Mine, Proceedings of MassMin 2012, Sudbury, Canada.

Castro, R, Pineda, M 2012, Draw control at Goldex mine. Internal report to Agnico-Eagle, Laboratorio de
Block Caving, Universidad de Chile.

Castro, R, Armijo, F 2012, Experimental design for the full scale flow test marker at El Teniente, Internal
Report to Codelco Chile, Laboratorio de Block Caving, University of Chile.

Castro, R, Pablo, P 2014, Empirical observations of dilution in panel caving, Accepted for publication.
Paper 12/110. Journal of South African Institution of Mining and Metallurgy.

DeWolfe, V 1981, Draw control in principle and practice at Henderson Mine, Desing and Operation of
Caving and Sublevel Stopping Mines, (Ed: D. R. Stewart), Society of Mining Engineers, USA.

Hoek, E 2004, Model to demonstrate how rockbolts work. [online] Hoeks corner < http://www.rocscience.

Hustrulid, W & Sun, C 2004, Some remarks on ore pass design guidelines, Proceedings of MassMin
2004, Santiago, Chile. (Ed(s): A.Karzulovic and M.Alfaro), Chilean Engineers Institute, pp.

Janssen, H 2004, Experiments regarding grain pressure in silos written in 1895, Proceedings of MassMin
2004, Instituto de Ingenieros de Chile, Chile, pp. 293-300.

Julin, D 1992, Block Caving, SME Mining Engineering Handbook, 2nd edition. Society for Mining,
Metallurgy and Exploration.

Kvapil, R 2004, Gravity flow in sublevel and panel caving - a common sense approach, Lule, Sweden:
Lule University of Technology.

Laubscher, D 1994, Cave Mining the state of the art, Journal of South African Institute of Mining and

Laubscher, D 2000, Block Cave Manual. Prepare for the International Caving Study 1997-2000. Julius
Kruttschnitt Mineral Research Centre, The University of Queensland. pp. 111-118.

Paredes, P 2012, Dilution Entry Mechanisms in Block and Panel Caving Operations, Master Thesis.
University of Chile, Chile, (in Spanish).

Pineda, M 2012, Study of the gravity flow mechanisms at Goldex by means of a physical model, Master
Thesis, University of Chile, Chile.

Susaeta, A 2004, Theory of gravity flow (Part 1), Proceedings MassMin 2004, Santiago, Instituto de
Ingenieros de Chile, Chile.

Susaeta, A 2004, Theory of gravity flow (Part 2), Proceedings MassMin 2004, Santiago. Instituto de
Ingenieros de Chile, Chile.

Caving 2014, Santiago, Chile

Application of a methodology for secondary

fragmentation prediction in cave mines
MA Fuenzalida Itasca Consulting Group, Inc., USA
T Garza-Cruz Itasca Consulting Group, Inc., USA
M Pierce Itasca Consulting Group, Inc., USA
P Andrieux Itasca Consulting Group, Inc., USA

This paper describes how primary fragmentation estimates can be combined with REBOP simulations to
predict the secondary fragmentation of caved rock. Primary fragmentation estimates are obtained from
separate numerical models that link caving induced stresses at failure (estimated with FLAC3D cave-scale
models) with associated fragmentation of rock masses (estimated through systematic Synthetic Rock Mass
testing with 3DEC). REBOP takes the predicted spatial distribution of primary fragmentation (along with
intact rock strength, drawpoint layout and draw schedule) as input and provides estimates of the evolving
secondary fragmentation in the column and at the drawpoints. The logic is based on laboratory studies that
link attrition (via block splitting and rounding) to the product of shear strain and normal stress (essentially
the work done on the fragments).

1 Introduction
Fragmentation of an orebody plays an important role in cave mines, dictating whether the operation will be
successful. According to Laubscher (1994, 2000), fragmentation will influence drawpoint size and spacing,
equipment selection, draw control procedures, production rates, dilution entry onto the draw column, hang-
ups and the need for secondary breakage.

As caved rock is extracted, the cave progresses upwards forming rock fragments in the cave back vicinity
due to the effect of gravity or induced stress acting on the inherent discontinuities in the rock mass, which is
a process known as primary fragmentation. Correspondingly, zones of moving material known as Isolated
Movement Zones (IMZ) develop in the ore column above the drawpoints. Within the IMZs, fragments
of rock are subjected to relatively low normal stresses (due to arching) but high shear strains (due to the
non-uniform nature of the velocity profile, which is characterized by a central plug flow region surrounded
by a shear annulus roughly 10-15 fragment diameters wide). In the surrounding stagnant zone outside the
IMZs, normal stresses are high but shear strains are very low (Lorig and Cundall 2000; Pierce et al. 2010).
Fragment attrition is believed to occur in both of these zones, a process known as secondary fragmentation
(Brown 2007).

Efforts have been made to provide means to predict secondary fragmentation based on rules established
through experience and engineering judgement (Esterhuizen 1998), and also through experimental work
(Castro et al. 2014). In the following sections, a methodology for secondary fragmentation estimation
is presented that is based on the results of experimental studies of attrition under shear combined with
REBOP flow simulation. The approach considers shearing-dominated attrition within the IMZs, but not
compaction-dominated attrition within the stagnant zones. Application to a case study is also presented.

Gravity Flow

2 Methodology to assess fragmentation in cave mines

A new methodology for drawpoint fragment size distribution prediction has been formulated (Figure 1).
The first step is to calculate the primary fragmentation through the use of a combination of numerical
models taking into account the discontinuities in the rock mass and the induced stresses that are likely to
develop in the ore column as the caving progresses. Using the resulting primary fragmentation estimates as
input, REBOP (Pierce 2010) is able to estimate the secondary fragmentation through the use of an attrition

Figure 1 Diagram illustrating the methodology to predict secondary fragmentation

2.1 Assessment of primary fragmentation

Synthetic rock mass (SRM) samples were constructed by assembling a collection of tetrahedral blocks and
populating their contacts with strength values randomly selected from a cumulative strength distribution
(Garza-Cruz & Pierce 2014). A variety of SRM samples were tested under a suite of triaxial tests as well as
uniaxial compression and tensile tests to characterize their strength. The model allows the blocks forming
the sample to break at their subcontacts as a result of stress concentrations, mimicking the initiation of
cracks that can coalesce and/or propagate to fracture the rock mass.

The rock mass strength derived from testing the SRM samples was used to inform a cave-scale FLAC3D
(Itasca 2013) model. As the cave is propagated, the model records the induced stresses at failure. The
stresses at failure then are combined with the SRM-derived associated fragmentation distribution to predict
spatial primary fragmentation through the column (Garza-Cruz et al. 2014).

Because REBOP considers primary fragmentation as a Gaussian distribution, a least squares technique was
used to best fit the different primary fragmentation curves for the different geological domains.

2.2 REBOP as a tool to estimate secondary fragmentation

REBOP was developed by Itasca for the industry-funded International Caving Study (ICSI and ICSII) and
Mass Mining Technology (MMT) projects as a tool for rapid simulation of material flow within block,
panel and sublevel caves. REBOP models flow by tracking the growth of IMZs associated with draw. An

Caving 2014, Santiago, Chile

IMZ in REBOP is comprised of a number of discrete disk-shaped layers stacked above the drawpoint. The
volume of an IMZ is tracked in REBOP by balancing the incremental mass drawn from the drawpoint
with the mass produced by a bulking of material at the periphery of the IMZ. The amount of bulking is
controlled at the local level by the average initial and maximum porosities of material at the IMZ periphery
as specified in the block model. The relative rate of upward and lateral growth (which controls IMZ shape)
is controlled by the fragmentation and friction angle of material at the IMZ periphery as specified in the
block model. Experiments (e.g., Castro 2006) and simulations (e.g., Lorig & Cundall 2000) reveal that
an IMZ tends to develop an approximately ellipsoidal or cylindrical shape with an evolving width that is
most strongly controlled by mean fragment size. As a result, IMZs tend to be narrow in finely fragmented
rock and wide in coarsely fragmented rock. This behavior, which can be simulated within REBOP, directly
affects the drawpoint spacing that is required to maximize mobilization of material throughout the volume
of interest (Pierce 2010). Material movements within the IMZs are tracked via a field of markers, the
positions of which are updated daily according to the local velocity profile within the IMZ. Markers are
assigned a tensile strength and fragment size randomly from the input distributions of their parent lithology
when they are first created. As markers move down the IMZs and mix, their diameters are averaged within
each disk-shaped layer to control the local IMZ expansion rate and internal velocity profile.

Secondary fragmentation in REBOP is handled by systematically reducing the fragment size associated
with markers based on their tensile strength and the stress and strain experienced as they transit through
the IMZ toward the drawpoint. More specifically, the degree of breakage experienced by a rock fragment
moving inside the IMZ is a function of the average stress inside the IMZ (estimated via bin theory) relative
to the strength of the fragments and the incremental shear strain. At present, REBOP makes use of the
shearing attrition model developed by Bridgwater et al. (2003):



= the mass fraction attrited from the mono-sized initial assemblies (percent);

= the normal stress applied to the assembly (MPa);

= the total shear strain applied to the assembly (dimensionless);

= the fracture or tensile strength of the constituent particles in the assembly (MPa); and

= empirical constants (dimensionless).

2.2.1 Strength Scale Effect

REBOPs secondary fragmentation methodology takes into account the rock block strength scale effect.
It has been established that the intact strength of rock decreases with increasing scale due to a higher
probability of the presence of discontinuities. Hoek & Brown (1980) developed an empirical scale effect
relation for intact strength on the basis of laboratory testing conducted by a number of different researchers
on homogenous hard rock samples (i.e., samples lacking significant microfracturing or alteration).


Gravity Flow


c = uniaxial compressive strength;

c50 = uniaxial compressive strength of a cylindrical specimen with a 2:1 height-to-width ratio and a
base fragment diameter of 50 mm;

de = arbitrary diameter in mm; and

k = constant, scale effect exponent.

While Hoek and Brown (1980) estimated k being equal to 0.18 for relatively homogenous hard rock, work
by Yoshinaka et al. (2008) demonstrated that the range in scale effect can be much wider when more flawed
intact rocks are considered, with k ranging from 0.1 to 0.9 (Figure 2). Adjustments to the strength scale
effect exponent, k, allow REBOP models to account for potential effects of defects that may exist in the
intact rock blocks, and therefore reduce rock block strength.

Figure 2 Scale effect relations for intact rock UCS proposed by Yoshinaka et al. (2008). The relation of Hoek &
Brown (1980) is shown for comparison (Pierce et al. 2009)

2.2.2 Modes of secondary fragmentation

REBOP accounts for the following three main effects that link attrition to the product of shear strain and
normal stress with respect to secondary fragmentation:

1. Block splitting occurs when the average normal stresses inside the IMZ are high relative to the
block strength. Splitting results from tension induced in a fragment via compression. If the induced
tensile stress exceeds the tensile strength of the grain, it will split.

2. Rounding of block corners and generation of fines refer to the removal of asperities from the
surface of the block as the fragments move down in the draw column.

3. Fines cushioning takes place when large blocks are surrounded by a large number of smaller
fragments, providing confinement and increasing strength and survivability as a result.

Caving 2014, Santiago, Chile

All three modes are responsible for the reduction of block size. The resulting characteristic evolution in
fragment size distribution is illustrated in Figure 3.

Figure 3 Main effects of secondary fragmentation results observed using REBOP: (1) corner rounding leading
to fines production, (2) block splitting leading to a breakdown in blocks of intermediate dimension and (3) fines
cushioning leading to survivability of large blocks

3 Application of the methodology in cave mines

3.1 Case study

The case of study consisted of an ore deposit with three main geological domains: weak, moderate and
strong representing different rock mass strengths.

The results of SRM primary fragmentation predictions were combined with the distribution of strength
domains and the predictions of cave back stresses at failure to derive a distribution of primary fragmentation
for input to REBOP. As illustrated in Figure 4, a relationship between induced stress and mean fragment
volume could be established assuming a linear behavior.

Based on the assumption that fragments are likely shaped as cubes, the following equations can be fitted
to the mean and standard deviation of the fragment diameters. Table 1 shows the fitted parameters for each
geological domain contained in the block model.


Gravity Flow

Figure 4 Mean fragment volume with respect to the induced stress at failure

Table 1 Parameters for the Mean and Standard Deviation Fragment Size

Type of Rock Parameter

a = -0.015, b = 1.9,
c = -0.014, d = 1.7
a = -0.011, b = 1.3,
c = -0.011, d = 1.2
a = -0.006 , b = 0.8,
c = -0.007, d = 0.8

Taking into account the induced stresses at failure from the FLAC3D simulations and using the relationships
in Equation 2 with the parameters established in Table 1, a block model can be populated with a primary
fragmentation curve fitted as Gaussian distributions. As a result, Figure 4 illustrates the populated block
model showing the mean fragment size for different geological rock domains.

3.2 Results

The secondary fragmentation logic in REBOP is sensitive to several inputs. Since IMZ shape is controlled
by the primary fragmentation and friction angle, a change in these inputs will have an impact on how the
size of fragments evolve as they go down through the column. Furthermore, the strength scale effect also
plays a major role in secondary fragmentation, as a weaker intact rock (low tensile strength) will produced
a finer fragmentation. This effect can be achieved by increasing the input exponent factor k. Hence, having
a coarser initial fragmentation, blocks will break more easily.

Caving 2014, Santiago, Chile

Figure 5 Cross-section of the block model mean fragment size for different geological rock domains, blue
being fine fragmentation and purple coarse fragmentation

Additionally, drawpoint spacing and draw schedule are also essential inputs for secondary fragmentation.
REBOP will predict overlap between IMZs and associated uniform drawdown if closely spaced drawpoints
and coarse fragmentation are considered. This will achieve a more uniform drawdown, lower shear strains
in the column and correspondingly lower secondary fragmentation. On the other hand, more breakage is
likely to occur with widely spaced drawpoints (isolated draw scenario).

It is anticipated that as the production progresses, the size of rock fragments will tend to decrease. Having
a differential draw will increase the effect of secondary fragmentation. On the contrary, having a uniform
extraction rate among drawpoints will result in more uniform drawdown, resulting in less secondary

Employing a user-defined drawpoint layout and draw schedule, and using the parameters detailed in Table
2 as inputs for REBOP, a drawpoint fragment size distribution can be estimated.

In order to capture the effect of having different lithologies with different fragment sizes in the draw
column, a first simulation for each zone was run with the mechanism of secondary fragmentation being
inactive (primary fragmentation curve). The effect of the secondary fragmentation with respect to the initial
fragmentation over the years of production is illustrated in Figure 6. In general, the size of rock fragments
tends to decrease over time due to the effects of block splitting, fines cushioning and rounding of block
corners and generation of fines as the blocks move down through the column as described in Section 2.2.
However, some of the changes in fragmentation can also be attributed to the spatial distribution of primary
fragmentation, which tends to be coarse at the top and bottom of the column and finer at the column mid-
height, and also varies laterally with lithology.

Gravity Flow

Table 2 Inputs parameters for REBOP model

Input Parameters
Initial/Maximum Porosity [%] Clay: 45/65, Rock: 0/45
Solid Density [tonnes/m3] 2.8
Angle of Repose [] 50
Friction angle[] Rock: 42, Clay: 20
Tensile Strength [MPa] Mean: 8.23 Stdv: 2.18
Fragment Size [m] Clay: 0.1, Rock: Mean [0.6-1]
Block Model Cell Size [m] 12
Base fragment Size for Scale effect relation [m] 0.05
Exponent in Scale Effect relation k = 0.18

Figure 6 Evolution of the global fragmentation over years of production

4 Conclusions
A new methodology for prediction of secondary fragmentation in cave mines has been presented. The
methodology relies on the use of flow simulation as well as experimental studies on shearing-induced
attrition. The primary fragmentation input to REBOP has been derived from separate numerical models
that link caving induced stresses at failure in the cave back vicinity with associated fragmentation of rock
masses. REBOP simulations of flow (employing a user-defined drawpoint layout and draw schedule)
estimate secondary fragmentation in the column and at the drawpoints as a function of shear strain, normal
stress and tensile strength.

Caving 2014, Santiago, Chile

It has been confirmed that the methodology is reliable; using the mechanisms found in the literature at the
same time serve the purpose of being practical and applicable to perform forward analyses in cave mines.

A direction to improve this methodology would be to include a secondary fragmentation mechanism via
impact loading, which can occur when rock fragments fall through an air gap. By including this approach
and performing further enhancements, this methodology could be used routinely as a practical tool to
estimate caved rock secondary fragmentation for the design of caving operations.

Bridgwater, J, Utsumi, R, Zhang, Z & Tuladhar, T 2003, Particle attrition due to shearing the effect of
stress, strain and particle shape, Chemical Engineering Science, vol. 58, pp. 4649-4665.

Brown, ET 2007, Block Caving Geomechanics, The International Caving Study 1997-2004, 2nd ed.,
JKRMC: Brisbane.

Castro, R, Fuenzalida, M & Lund, F 2014, Experimental study of gravity flow under confined conditions,
Int J Rock Mech and Min Sc, vol. 67, pp. 164-169.

Castro, R 2006, Study of the mechanisms of granular flow for block caving, PhD Thesis, University of
Queensland, Australia.

Esterhuizen, GS 1998, ICS Meeting Minutes, BCF Review, Brisbane, Australia.

Garza-Cruz, TV & Pierce, M 2014, A 3DEC Model for Heavily Veined Massive Rock Masses, Proceedings
48th US Rock Mechanics / Geomechanics Symposium, Minneapolis, USA.

Garza-Cruz, TV 2014, A 3DEC-FLAC3D Model to Predict Primary Fragmentation Distribution in Cave

Mines, Proceedings of the 3rd International Symposium on Block and Sublevel Caving, (ed.
R. Castro), Santiago, Chile.

Hoek, E & Brown, ET 1980, Underground Excavations in Rock, London, Instn. Min. Metall.

Itasca Consulting Group, Inc. 2013, 3DEC Three-Dimensional Distinct Element Code, Ver. 5.0.
Minneapolis: Itasca

Itasca Consulting Group, Inc. 2013, FLAC3D Fast Lagrangian Analysis of Continua in 3Dimensions, Ver.
5.0. Minneapolis: Itasca

Lausbcher, D 2000, Block Caving Manual. Prepared for International Caving Study. JKMRC and Itasca
Consulting Group, Inc.: Brisbane.

Laubscher, D 1994, Cave Mining-The State of the Art, J. South African Inst. Min. Metall., vol. 94, pp.

Lorig, LJ & Cundall PA 2000, A rapid gravity flow simulator, Final Report, International Caving Study,
E.T. Brown (Ed.), JKMRC and Itasca Consulting Group Inc., Brisbane, Australia.

Pierce, M 2010, A Model for Gravity Flow of Fragmented Rock in Block Caving Mines, Ph.D. Thesis,
University of Queensland.

Gravity Flow

Pierce, M, Weatherley, EK & Kojovic, T 2010, A hybrid methodology for secondary fragmentation
prediction in cave mines, in Proceedings of the 2nd International Symposium on Block and
Sublevel Caving, (ed. Y. Potvin), pp. 567-581.

Pierce M, Gaida, M & DeGagne, D 2009, Estimation of rock block strength, Proceedings of the 3rd
CANUS Rock Mechanics Symposium, ed. M. Diederichs and G. Graselli, Toronto, Canada.

Yoshinaka, R, Osada M, Park H, Sasaki T & Sasaki, K 2008, Practical determination of mechanical design
parameters of intact rock considering scale effect, Engineering Geology, vol.96, pp. 173-186.

Caving 2014, Santiago, Chile

Case Study: Improving SLC recovery by measuring ore

flow with electronic markers

S Steffen Elexon Mining, Australia

P Kuiper Elexon Mining, Australia

Many industries have embraced the concept of improving an operations performance by measuring a
process, analysing the results, implementing an improvement, and then measuring the effect of the
improvement. Sublevel caving lends itself well to such an approach, because the mining method involves a
process that is continually repeated. Changes can be implemented commencing from any blast ring, which
means each blast ring is an opportunity to improve the process, and that improvement can be extended to
all subsequent rings. Sublevel caving (SLC) is prone to dilution, which has a significant effect on a mines
profitability, so keeping dilution low or reducing dilution by process improvement is attractive. However,
most of the factors relating to dilution ultimately relate to ore flow and until recently, there were limited
means to monitor ore flow. Before Smart Markers, steel markers were used to measure ore flow; however,
their use was time consuming and interfered with production and they were therefore not systematically
used. Thus, SLC mining remained a black box mining method with limited control over production.

An SLC gold mine in Australia, managed by a dynamic and innovative team, has succeeded in closing the
process improvement loop: The mine used Smart Markers to measure ore flow and analysed the data. Based
on the data, it then reassessed its drill and blast processes and implemented changes. It then measured the
effect of the changes. The new data indicates a 4% improvement in primary recovery. An analysis of the
effect of this improvement indicates that, if the improvement in primary recovery is sustained, the mines
overall profit should increase by between 4% and 14%.

1 Introduction
With open stoping, mines assess the performance of their recovery by monitoring the mined-out cavity.
Under- and over-break are easily identifiable and can be targeted for improvement. However, with Sublevel
Caving, voids are automatically filled by gravity with broken rock. This mechanism does not usually
allow a visible assessment of ore recovery and flow, in particular where the desired ore and waste material
(dilution) is coming from, and which material has been left behind (ore loss). Understanding ore flow
is crucial to be able to systematically improve mining practices. For example, Smart Marker data has
shown that parts of targeted SLC rings are consistently not recovered and result in underperforming mineral
recovery. Knowing which parts of the ring are not recovered can enable mines to systematically target the
causes of the ore loss.

Since the 1960s, SLC research has concentrated on understanding and modelling material flow. The main
thrust has been to identify a generally valid ore flow mechanism that could then be used to optimise recovery
in sublevel cave mines. Investigations involved small-scale physical models, full scale flow monitoring
marker trials and computer modelling. To date, a number of theories have been proposed to explain the ore
flow mechanism in sublevel caving. These investigations have primarily been scientific in nature and the
results have not usually been integrated into mining production or used to improve mining procedures. This
has been partly due to the impracticality of using steel markers in production environments. The availability
of electronic Smart Markers has now made it possible to monitor flow and ore recovery with minimal
interference to production.

Gravity Flow

Over recent decades, the understanding of modern sublevel caving has advanced significantly mainly due
to academic research in conjunction with some larger mining operations. Newcrests Ridgeway SLC mine
is regularly cited as an example of what can be achieved by a well organised production team focused on
optimising mining. Their achievements included:

Mining production above 6 million tons per year which was well above the planned 4 million
ton per year name plate production (Popa et al. 2012),

Mined tonnes were 10% more than reserve tonnes (Manca 2012),
Au grade was down by 6% but Au quantity was 3% higher than reserve (Manca 2012), and
Cu grade was down by 2.5% but Cu quantity was 7% higher than reserve (Manca 2012).
Ridgeways SLC mining success can be attributed to a combination of disciplined mining and the
implementation of feedback loops to systematically improve mining procedures. The Ridgeway team used
steel markers to investigate dilution entry points and understand multilevel recovery (Power, 2004). Initially,
an improvement in primary recovery was not achieved (Power, 2004), although a 20% cost reduction was
achieved. Later work by Luca Popa showed that improved drill and blast techniques did increase primary
recovery (Popa et al, 2012).

2 The importance of controlling dilution

Uncontrolled ore flow is likely to happen if mining practices are not optimised for the local mine. The
consequence of uncontrolled ore flow from a recovery perspective is excessive grade dilution and ore
losses. (Uncontrolled ore flow can have other consequences, including safety issues, but that is not covered
here). Dilution and ore loss both have a detrimental effect on the economic performance of SLC mines;
however dilution has a greater economic impact than ore loss. Suppose a tonne of ore is lost. Another tonne
will be mined instead of it, and eventually, the mine close a little bit earlier. That means the profit on that
tonne is lost, and that loss is realised at the end of the mines life. However, if a tonne of waste material is
mined instead of ore, that tonne will still be hauled and processed. The costs involved in processing a tonne
of waste material are the same as processing high grade ore; however that tonne generates zero revenue.
That means mine profits are reduced by the revenue from the mineral in a tonne of ore, not the profit made
on a tonne of ore. That causes an amplified effect on profit. For example, with a mine like Ridgeway
SLC, an improvement of 0.1 g/t in mined grade through reduced dilution would have generated additional
revenue of USD 25,200,000 per year (6 Mt/y * 0.1g/t * 42 USD/g). Conversely, an increase in dilution of
0.1g/t grade would reduce revenue by US 25,200,000 per year. Higher than planned dilution rates quickly
erode mine profits.

Projects with marginal economic rates of return are particularly sensitive to changes in dilution rates: a
small change can mean the difference between a profitable mine and a loss-making one. In sublevel caving,
dilution and ore loss go usually hand in hand because a certain number of tonnes are extracted from each
drawpoint. If, for whatever reason, a certain number of tonnes of ore are left behind, material will flow from
somewhere else to make up the tonnes extracted. This materials grade is more likely to be diluted.

3 The importance of drill and blast to control ore flow

A mentioned above, exerting optimised control over ore flow is critical for achieving good mine performance.
Drill and blast procedures are the main instrument to influence ore flow and engineer ore recovery. The
purpose of drill and blast is to fragment the targeted rock and to mobilise and induce the flow of the blasted
ore to drawpoints. Appropriate drill and blast procedures are critical to the success of sublevel caving (Popa
2012). Production blasting in sublevel caving is confined blasting, meaning that the material lying flush

Caving 2014, Santiago, Chile

with the rock to be blasted is either previously blasted material, caved material or fill (Brunton, 2009).
Confined blasting is different to unconfined blasting as is generally used in open stoping. The effects of
suboptimal drill and blast in sublevel caving can include: Sub-optimal ore flow, Oversized rocks, hangups,
frozen rings; hole losses, brow damage and backbreak.

The consequences of this are safety issues (e.g. working near unstable brows), increased production costs
due to production interruption and reworks, and grade and production variability. Assessing the performance
of a confined blast is challenging due to the fact that the physical impact of drill and blast on the in-situ
rocks is not easily assessable. Confined blast performance is highly dependent on an adequate distribution
of explosive energy optimised for the local mine conditions, such as the mines design, rock properties,
stresses, mining sequence etc. The delivery of explosive energy depends on a multistep process. Blast
rings need to be designed with the optimal explosive energy distribution in mind. The blast rings need to
be drilled according to the plan. The blast holes need to be prepared and charged properly. The holes have
to be drilled accurately, as any deviation may change the geometry of the explosive energy distribution.
Holes found to be blocked during prepping may require cleaning or redrilling in order to charge them with
explosive. Boosters must be positioned appropriately and the detonator sequencing timing must be properly
implemented. Each step prior to the initiation of the blast is important to success. The key to improving
drill and blast results lies in measuring the performance of each step, including ore recovery and then
implementing changes to drill and blast design and practice to achieve improved results.

Optimised drill and blast can improve primary recovery, which has been achieved at Ridgeway SLC
(Popa, 2012). Primary recovery is the best opportunity to recover undiluted ore grade (Brunton 2009). The
potential consequence of non-primary recovery is that high grade ore may mix with low grade or waste
material while flowing over several sublevels to the extraction point. The result can be significant dilution
of mined ore grade.

4 Deviating flow
There are three main recovery target zones in an SLC ring: the core, shoulder and apex zone (Figure 1).

Figure 1 SLC ring recovery zones (Steffen & Clark 2013)

Gravity Flow

Marker data has shown that:

Most of the primary recovery tends to be extracted from the core zone, which is vertically
above the extraction drive. However, only a certain fraction of this material is recovered during
primary recovery.

Less material tends to be recovered from the shoulder zone, but as it is lies above the apex of
the level below it, it should be accessible as secondary recovery from the drive below.

Recovering the apex zone is important for overall recovery as it is a significant component of
the rings tonnage and it is also the key to recovering the shoulder zone material in the level
above it.

In cases where there is a 12.5 m spacing between drive centres, the core zone represents approximately 50%
of the rings tonnage, while the apex and shoulder zone each represent around 25% of the rings material.

Gavin Power (Power 2004), described the effect of deviating flow, which occurs when the apex is not
sufficiently fragmented and mobilised. This prevents ore flow from propagating through the apex and into
the shoulder zones targeted by secondary recovery. Instead, the ore flow deviates into the depleted drives of
the level above (Figure 2). This will introduce material of potentially diluted grade. The apex and shoulder
zone, which account for around 50% of the ring tonnage, are not being well recovered due the deviating
flow effect described. With tonnage-based draw control, a predetermined amount of ore is being recovered
from the production ring. If the amount of drawn material is larger than the material recovered from the
targeted ring or from the shoulder zone of the level above, it has to be coming from other places which
are of unknown and potentially diluted grade. If deviating flow occurs systematically, possibly to due to
inadequate drill and blast processes, a significant amount of high grade ore can be left behind in the apexes
and the shoulder zone.

Figure 2 Deviating flow illustration (Steffen & Clark 2013)

5 Case study
This case study shows an example of a mine that has actively used Smart Markers to understand their
recovery and made changes to mining procedures to improve their performance. This case study is of an
Australian gold mine employing sublevel caving production principles with zero grade waste backfill from

Caving 2014, Santiago, Chile

the top to reduce the likelihood of caving of the crown pillar. The mine is anonymous in this paper, because
there was insufficient time to have the mine management review this paper before the submission deadline.
The mine agreed to have their data used without their review, if their details were kept anonymous. The
mine has an annual production rate of around 1.5 Mt. The mine is in full production and to-date has met its
economic performance expectations. This can be attributed mainly to the efforts of the dynamic technical
team at the mine site. The mine has gone through several stages of optimising their mining procedures,
including drill and blast, using a Smart Marker system. This paper shows the actual ore flow patterns
recorded by the Smart Marker System in this mine. To date, Smart Markers have been installed on six
sublevels. On the first two levels, Smart Markers were installed in upholes. From the third level down,
Smart Markers were installed in down-holes from the level above the targeted sublevel. This approach was
chosen to simplify installation by lowering chains of Markers down the holes instead of having to push
them up individually. After mining had progressed through the production rings on levels two and three,
the Smart Marker data was analysed to evaluate areas of poor recovery and adverse flow effect. The data
collected indicated that deviating ore flow had occurred. Second quartile primary marker recovery had
dropped with every level and reached 30.3%. The primary recovery figure quoted here is somewhat low.
The actual figure is probably higher, because highly fragmented material close to the blast holes is more
likely to flow to the drawpoint; Markers can only be placed 65 cm from the blast ring. The Second quartile
marker recovery from the apex zones was 20.6%. Figures 3 and 4 show Smart Marker data signatures; these
show deviating flow and partial recovery of the apex and the shoulder zones on the level above.

Figure 3 Deviating Flow Pattern measured with Smart Markers (L) and Example of partial apex and shoulder
zone recovery (R)

An outcome of the marker data analysis was a recommendation to reassess the drill and blast processes,
which was subsequently done. The reassessment indicated potential areas for improvements in drill and
blast design and procedures. A number of changes were implemented and their results assessed. More Smart
Markers were installed to analyse the effect of the changes on primary recovery. The changes involved:

Increasing the distance between the toes of blast holes and any preconditioned ground (such as
drives and blasted ore at higher levels), in order to prevent explosive energy from venting into
preconditioned ground, to reduce hole blockage due to loose material and to decrease damage
to ring holes in neighbouring production drives.

reducing the powder factor.

changing the uncharged collar lengths.

Gravity Flow

improving drilling accuracy.

trialling different ring designs including an innovative asymmetrical ring design.
Reducing the number of blast holes would reduce production costs accordingly. The results from the first
round of process changes, which included a ring design with a lower number of holes, resulted in a notable
reduction of flow deviating through the depleted drive of the level above and improved recovery of the apex
zones. Second quartile primary marker recovery increased to 31.6% and apex recovery of 45%. The effect
on secondary recovery was unfortunately not observable because Smart Markers were not installed in the
rings on the sublevel above.

Figure 4 Improved Apex and primary recovery

Figure 5 Further improvements to apex and primary recovery

Caving 2014, Santiago, Chile

In a second set of trials, an innovative asymmetrical blast-ring design was trialled. The aim of this design
was to reduce the damage to shoulder blast holes in the neighbouring production drive and to increase
primary ore recovery. The results are shown in the graphic below. A further increase in the apex recovery
and overall primary recovery was observed. Second quartile primary recovery increased to 34.4% and apex
recovery of 50%. Again, secondary recovery was unfortunately not assessed because Smart Markers were
not installed directly above the relevant Smart Marker installation. These trials have shown some promising
initial results. The author believes that even more drill and blast improvements are achievable through
refining the current approach.

7 Financial effect of improving primary recovery

This section attempts to estimate the financial benefit of improving primary recovery, assuming the
improvement is implemented in all future rings. SLC mines have an overall rate of dilution. For the subject
mine, this is assumed to be 20%. Primary recovery extracts virgin ore, with zero dilution. That means the
dilution is always in the non-primary recovery. The subject mine draws approximately 1450 tonnes of ore
from each ring. The early data indicates that the implemented changes improved primary recovery by 4%.
This means that 4% x 1450 = 58 tonnes of ore came from the target ring instead of from somewhere else.
(As mentioned previously, the overall primary recovery figure is somewhat low; however a lower-than-
actual primary recovery figure understates, rather than overstates the benefit calculations).

Here are figures from the subject mines 2013 financial reports:

141846 ounces produced (4,411 kg at 31.1 grams per troy ounce)

AUD$1122 per ounce all in costs ($36 per gram)

AUD$1562 per ounce ($50 per gram)

Ore grade: 2.8 grams per tonne

High estimate: Let us assume that the improvement in primary recovery also results in an equivalent
improvement to secondary, tertiary, etc recovery. In that case, revenue will increase by 4%.

Using the above figures, revenue was 4,411x103 g x 50 dollars/g = $220 Million.

4% x revenue = $8.8 Million

The 2013 profit was 4,411x103 g x (50-36 dollars/g) = $62 Million

Therefore, the high estimate of improvement to profit, expressed as a percentage is $8.8 m / $62 m = 14%.

Low estimate: Let us assume that the changes only improve primary recovery, but do not improve secondary,
tertiary etc recovery at all. If that was the case, the 58 tonnes per ring of improvement in primary recovery
would be replacing 58 tonnes of diluted ore. To work out the improved ore grade, we need to estimate the
dilution in the non-primary recovery. If overall dilution (including primary recovery) is 20%, and primary
recovery is 30%, then all of the dilution is in the 70% non-primary recovery. Therefore, the non-primary
recovery has 20% / 70% = 29 % dilution.

If primary recovery improves from 30% to 34%, then primary recovery would have increased annually by

(0.34 0.3) x 4,411,000 g / 2.8 g/t = 64 kt

This tonnage would have had an improvement in grade of 29% (the difference between undiluted and
diluted ore). That works out to 0.29 x 64x103 x 2.8 grams = 52,000 g of additional gold.

Gravity Flow

This equates to 52000 x $50 = $2.6 million in additional profit.

From above, the mines overall profit was $62 Million. Therefore the low estimate of percentage improved
profit is 4%.

Overall estimate: The actual improvement will lie somewhere between the low and high estimate.
Secondary, tertiary etc recovery should improve, but we do not know by how much. For now, lets assume
halfway in between. See Table 1.

Table 1 Estimate of improved profit from implemented changes

Estimate Dollars As % of profit

High $8.8 Million 14%
Low $2.6 Million 4%
Probable $5.7 Million 9%

6 Conclusion
Now that there are suitable tools to measure ore flow, SLC mines are ideal to implement process improvement
by measuring performance, analysing the data, implementing an improvement and measuring the effect
of the improvement. This paper shows early results from a successful implementation of this approach.
Preliminary analysis shows that the improvements, see Table 1, are very promising, but the data was quite
fresh at time of writing. As more data is gathered, the picture will become clearer. Also, this is only the
beginning - further gains are very likely to be found by continuing this process of improvement.

The author would like to thank the following people for sharing their knowledge and for taking the time to
engage in many debates on how to improve cave mines (in no particular order): Stuart Long, Luca Popa,
Nigel Clark, Ian Brunton, Gideon Chitombo, Alan Guest and Otto Richter. This paper does not necessarily
represent their opinions.

Brunton, I 2009, The impact of blasting on sublevel caving material flow behaviour and recovery, PhD
thesis, W H Bryan Mining and Geology Research Centre, University of Queensland, St Lucia,

Jamieson, M 2012, Development of Sub Level Cave Draw Optimisation at Newcrest Mining, MassMIN
2012 Conference Proceedings.

Manca, L & Malone, E 2012, Cadia Valley Mines past, present and future, AUSIMM, Available from: [28
April 2014].

Power, GR 2014, Modelling granular flow in caving mines: large scale physical modelling and full scale
experiments, PhD Thesis, The University of Queensland, Brisbane.

Caving 2014, Santiago, Chile

Power, G 2004, Full scale SLC draw trials at Ridgeway Gold Mine, Proceedings of MassMin 2004,
(Karzulovic A. and Alfaro M. eds), 2225 August, Santiago, Chile: Instituto de Ingenieros de
Chile, pp. 225230.

Popa, L, Trout, P & Jones, C 2012, The evolution and optimization of sublevel cave drill and blast practice
at Ridgeway Gold Mine - production rings, in Proceedings Massmin 2012.

Vila, D 2012, Calibration of a mixing model for sublevel caving, PhD Thesis, the Faculty of Graduate
Studies, The University of British Columbia, Vancouver, Canada.

Steffen, S & Clark, N 2013, Sub-level caving: engineered to perform, The Mining Magazine, September

Gravity Flow

Stochastic models for gravity flow: numerical

WH Gibson SRK Consulting (Australasia) Pty Ltd, Australia

Material flow analysis is commonly required in mining methods where the rock has to move from its initial
position to the extraction point, commonly known as gravity flow. This happens in mining methods, such
as, sublevel caving, panel caving, block caving. During this process, the ore is diluted with waste from the
walls of the cave column or waste sitting on top of the ore. It is essential for the mining design to assess
the degree of dilution for different draw strategies to minimise the waste extraction and optimise the ore
recovery. Stochastic methods have proved to be good options for this type of assessment.

1 Introduction
Rules and probabilities are the bases of stochastic methods for describing phenomena, they are not founded
on a physical principle that forces the math to produce the right results (for example, the Finite Element
Method minimising potential energy to derive the equations used to solve the problem). Stochastic models
for gravity flow use just conservation of mass.

Despite the weak formulation, stochastic methods can be very powerful tools to solve material flow
problems, and some of the merits, limitations and pitfalls of these methods are explored in this paper.

2 Description of the program MFlow

Stochastic methods are modelling tools for estimating outcomes by allowing for random variations in one
or more inputs over time. When material is removed from a drawpoint, a void is created that is filled with
material from above the void. The exact source of that material is unknown; therefore, a random location
is assigned.

Figure 1 shows a plan view of a grid describing this concept. When part of the material is removed
from a cell below the centre of the grid, the void is filled from any of the nine cells above. Thisprocess
is random, making stochastic models ideal for this type of problem solving. In this particular case, a
probability is assigned to each cell indicating the chances that the void will be filled with material from the
cell immediately above (60%) or any of the surrounding cells (8% chance cells with adjacent sides and 2%
for cells with adjacent corners; percentages given as an example only).

The grid shown in Figure 1 does not match the axisymmetric nature of the problem; a better grid is shown
in Figure 2. This type of grid is better for analysing material flow because the grid can model naturally an
axisymmetric problem. The disadvantage of this type of grid is the shape of the cells, which complicates
the calculation.

Caving 2014, Santiago, Chile

Figure 1 Plan view of basic cube grid

Figure 2 Plan view of hexagonal prisms grid

Figure 3 shows a grid that combines the simple geometry of cubes and the circular location of hexagonal
prisms. MFlow uses this configuration of cells; p represents the probability that material will be transferred
from the cell above while (1-p)/6 is the probability that one of the cells around the central cell can transfer
material to the void below.

Figure 3 Plan view of cube shifted grid

Gravity Flow

3 Pascal cone and correlation between probabilities and ellipsoid width

The Pascal triangle can be used to understand how probabilities can be used to assess material flow in 2D.
Theconcept is extended to a 3D (Pascal Cone) to model the gravity flow in three dimensions.

3.1 Model in 2 Dimensions

The Pascal triangle can be used to assess the probability that a drunken man can reach home (depicted by
the shaded box in Figure 4), starting in a particular street and walking randomly (Harr 1987). At every
corner, there is a 50% chance that the man will turn either left or right. From Figure 5, it is possible to see
the probability of reaching home is 5/16.

In general, the probability of reaching corner r at street n can be calculated as follows:


Figure 4 2D random walk (after Harr 1987)

The probabilistic analysis indicates the chances of the man ending some blocks away from home. The
actual distance between the man and his home will depend on the size of the city blocks. The same problem
occurs in material flow as is explained in the following paragraphs.

Figure 5 presents the probabilities associated with this 2D random walk. If we imagine Figure 5 upside
down, we can see that the value indicated in any cell represents the probability of a cell being affected by
extraction of the cell with probability 1 (now at the bottom). This is a very simplistic 2D model, where each
cell has only two cells above with equal probability to fill the void generated in the cell below, in this case,
the chance that the cell highlighted in Figure 5 is affected by the extraction is 5/16. We can say, after this
analysis, that the cell affected is one (1) cell to the left of the cell of extraction but the actual distance of this
cell to the extraction point will be a function of the cell size.

A 2D model can be built with only three (3) cells that can transfer material to cells below. With this simple
model, it is possible to study the impact of the parameter p on the ellipsoid of extraction. Figure 6 presents
a vertical section showing probabilities that each cell will transfer material to the cell below.

Caving 2014, Santiago, Chile

Figure 5 Probabilities associated with 2D random walk

Figure 6 Probabilities within a simple 2D model

When this concept is applied in a 2D problem, it is possible to assess the chances that a particular cell will
be affected by the extraction. Figure 7 presents a cross-section with a single draw located at x=7 m. It
shows the probability that material located above the drawpoint is mobilised during extraction. The results
are presented for two values of p = 0.40 and 0.70. In both cases, the same amount of material is removed
from each drawpoint.

It is possible to observe that the parameter p controls the width and height of the ellipsoid of extraction.
A p-value closer to 1 produces a narrower ellipsoid of extraction than a smaller p-value. A correlation
between p and the actual width of the ellipsoid of extraction is discussed later.

The real challenge is in 3D, where the calculations of the probabilities are more complicated. If Figure 3 is
used as a reference, then there are seven (7) cells that can fill the void in the cell below and the probabilities
for each of the cells are not the same.

3.2 Model in 3 dimensions

The 3D problem can be viewed not as a Pascal triangle, but more as an inverted Pascal cone. It starts
with a cell with probability of 1 (certainly we will remove material from there) and the next layer above
represent the probability that one cell will fill the void created below. As we move up a layer and more
cells are involved in the calculation, the width of interaction between cells is controlled by the value of the
probability p indicated in Figure 3.

Gravity Flow

Figure 7 Probability of material being mobilised for parameter p=0.7 and p=0.4

A similar calculation can be carried out for the 3D case. With this calculation, it is possible to evaluate the
diameter of the ellipsoid of extraction for different values of p, as shown in Figure 8. Note that the width
is expressed in number of cells affected by the extraction, and not in actual distance measured in metres.

Figure 8 Width of ellipsoid of material mobilised as a function of parameter p

Caving 2014, Santiago, Chile

This effect can be visualised on a 3D model of two independent drawpoints in uniform material. Figure 9
shows the mobilised ellipsoids, clearly illustrating the effect of parameter p on the width of the ellipsoid
of extraction. This is an observed behaviour that, depending on physical properties of the broken rock, is
controlled by the width of the ellipsoid of extraction.

Figure 9 Extraction column width for different p parameters

3.3 Calibration of parameter p with observed material behaviour

It was shown that in these types of stochastic models, the cell size plays a role in the results; therefore, the
selection of the cell size has to be made considering other parameters in the model to be able to reproduce
the phenomena that have actually been observed.

The parameter p controls the width or diameter of the draw column in the model (not the actual length in
metres, but the number of cells that will be affected by the draw). If the width of the ellipsoid of extraction
is known, it is possible to select a cell size and a probability p to be used in the model.

The question that remains is how to relate the rock mass condition or other observed behaviour to parameter
p in order to build models that represent the actual material flow.

There have been some attempts to relate rock mass condition to ellipsoid width. Laubscher relates the
width of the ellipsoid with rock mass classification: the better the rock quality, the wider the ellipsoid
(Laubscher 1994). Kvapil recognised the same fact that an increase in the rock mass quality increases the
ellipsoid width (Kvapil 2008).

Sharrock (2008) made a review of several aspects of isolated draw and discusses the results of scaled
models. Unfortunately, the scaled models did not capture changes in ellipsoid width with material properties
or particle size. Susaeta (2004) presented a correlation between friction angle, particle size and ellipsoid

For mines in operation, it is possible to use information collected from the previous extraction to calibrate
p. Figure 10 shows the ounces extracted in a period of time compared with the values predicted using an

Gravity Flow

MFlow model. The parameter p was modified to minimise the difference between the curves and then
calibrated values were used in the future forecasting models.

Figure 10 Ore extracted used to calibrate parameter p

There are some guidelines to correlate observed material behaviour with ellipsoid width and this can be
used to define parameter p for a given cell size (using Figure 8). The process of assessing ellipsoid width is
based heavily on experience and observations, and less on a strong formulation including the characteristics
of the material. Nevertheless, there are some guidelines that can be used to build the stochastic model.

4 Stochastic modelling capabilities

Despite the simplicity of the formulation of stochastic models, they can address complex behaviour of
materials. Some of these are shown and discussed in the following paragraphs.

4.1 Variable flow of different materials (w Factor)

Several factors allow for some materials to travel faster than others in the draw column (Hashim 2009).
This type of phenomenon can be included in the model by introducing a weighting factor, w, that modifies
the probability of material moving from one cell to another.

A value of w=0 renders the material unaffected by draw and it does not flow. This can be used to define the
limits of the draw rings in SLC (Sub Level Caving) if it is assumed that the outside of the blasted ring will
not move. A value of w=1 allows the material to move freely.

Values between 0 and 1 can, therefore, be used to control the speed of material flow in the model. To show
the effect of the parameter w on the results, the model shown in Figure 11 was built. Above the extraction

Caving 2014, Santiago, Chile

point A there is fine ore under coarse waste, above the extraction point B the fine material is on top.

The results are shown in Figure 12. Due to fine waste material, there is a reduction in ore extraction in
drawpoint B due to a higher dilution of fine material travelling faster than the ore.

Figure 11 Section of simple model with different materials

Figure 12 Effect of fine material on extraction

4.2 Markers

The formulation of stochastic models is easy and that simplicity allows us to incorporate additional
calculation into the analysis without adding complexity to the overall analysis. In this case, markers are
added in the model to track the movement of the material.

Gravity Flow

Figure 13 shows the trajectory of markers located above drawpoint A. The sequence of extraction is A then
B then C. It is possible to see the change in trajectory of some of the markers and to compare the nature of
material that is extracted at the drawpoints.

Figure 13 Effect of fine material on extraction

4.3 Finger path (void diffusion)

It has been observed that an ellipsoid shaped extraction column is not always generated, and that material
can flow following a finger path - reaching surface earlier than expected (Brown 2003). This can be
modelled by modifying the weighting factor w and giving a higher probability of movement to material
already mobilised. This is shown in Figure 14 for a single drawpoint.

4.4 Modelling Surface Flow

Surface flow or unconfined flow is controlled by the repose angle; the material will flow until the repose
angle is reached. This introduces another constraint for the geometry, driven by material properties. Castro
(2009) mentioned that in order to model the surface flow, the height (h) and the side length of the cells (L)
should follow the relationship h/L=tanf, where f is the repose angle.

Figure 15 illustrates a failure on a bench and flow of the loose material.

Caving 2014, Santiago, Chile

Figure 14 Finger path for a single draw point

Figure 15 Surface Flow

5 Conclusions
Stochastic models have a much easier formulation than other methods, such as, the Finite Element Method.
However, the lack of formulation based on a physical principle makes them more difficult to set up unless
information about the rock mass to be modelled is available, thus enabling the modeller to calibrate the

The results are cell size dependent; therefore, cell size has to be selected along with other parameters
(probabilities) to ensure the model best represents the material behaviour observed.

Gravity Flow

It is suggested to calibrate a model using the ellipsoid of extraction width estimated from rock mass
characterisation or assessment of eccentricity of the ellipsoid, and a correlation presented between
probability p and number cells. Past draw performance at the mine can be used for calibration when
information about the type of material extracted, such as ore, waste and grades, is available.

Despite the limitations and the weak formulation, this paper shows that stochastic models can be used to
describe complex behaviour, such as, accelerated flow of fine material, finger path flow and dilution.


Brown, ET 2003, Block Caving Geomechanics, The International Caving Study Stage I, 1997-2000,
University of Quuensland, Julius Kruttschnitt Mineral Research Centre, Brisbane.

Castro, R, Gonzalez & Arancibia 2009, Development of a Gravity Flow Numerical Model for the Evaluation
of Drawpoint Spacing for Block/Panel Caving, J South African Inst Min and Metall. vol 119.

Harr, ME 2005, Reliability-based design in Geotechnical Engineering, Dover Publications Inc., Mineola,
New York.

Hashim, MHM & Sharrock, GB 2009, Numerical investigation of the effect of particle shape on
percolation, Proceedings 43rd U.S. Rock Mechanics Symposium & 4th U.S. - Canada Rock
Mechanics Symposium, American Rock Mechanics Association, 28 June-1 July, Asheville,
North Carolina, 8 pages.

Kvapil, R 2004, Gravity Flow in Sublevel and Panel Caving A Common Sense Approach, Lulea University
of Technology Press, Lulea, Sweden.

Laubscher, DH 1994, Cave mining the state of art, The Journal of The South African Institute of Mining
and Metallurgy, October 1994, pp. 278-293.

Sharrock, GB 2008, The Isolated Extraction Zone in Block Caving A Review, in Proceedings SHIRMS
2008, Perth, pp. 255-272.

The International Caving Study Stage I, 1997-2000, University of Quensland, Julius Kruttschnitt Mineral
Research Centre, Brisbane.

Caving 2014, Santiago, Chile

First steps in monitoring gravity flow at El Teniente

Mine: installation stage in Block-2, Esmeralda Mine
E Viera Codelco, Chile
M Montecino Codelco, Chile
M Melndez Codelco, Chile

The future scenarios under which panel caving is developing will be even more adverse, mainly because
of the complex geomechanical conditions and greater demands for production plans. The quantity and
quality of the required information for monitoring and controlling these variables fulfill a main role, thus
technology incorporation constitutes a key step in this topic.

From the variables that rule the mining method, cavity control and gravitational flow mechanisms are of
vital importance. Technology development has allowed smart electronic devices (Smart Markers) usage,
which are installed on higher levels with respect to the production level (undercut level, haulage levels,
special drillholes, etc.) Once undercutting and the later caving process are started, the smart markers
are part of fragmented material which, once the mining process starts are subject to induced movements,
dependent on the draw strategies to be performed. The emergence of these markers in drawpoints and
their later transportation through LHD equipment, allow the installed readers at production level (or main
haulage level) to record the markers passing, under a certain detection radius.

The aim of the following work is to detail the Smart Markers installation stage, and to show that preliminary
results obtained at Esmeralda Mine. In this, near field flow tests (31 m maximum length drillholes between
ring blast hole) and far field test (vertical drillholes with a maximum length of 100 m), are being performed
on 3 undercut level drift (55, 57 and 59). The installation of 305 markers in the near field test is highlighted
(in 16 ring drillholes) from which 96 markers have been registered so far. In the far field tests, 92 markers
were installed in 3 vertical drillholes, from which 4 markers have been recorded.

1 Introduction
In the coming years, El Teniente will be facing even more challenging scenarios, characterized by complex
geomechanical views along with greater productive demands. Understanding the fragmented material flow
inside the ore column has been a very debated and investigated topic for several years. Different flow
mechanisms have been proposed, most of them based on scale size models. However, up to date trials to
monitor the material flow inside the cave at full scale have been few (Brunton 2012). Most of the design
and planning tools that have been developed throughout the years for the material flow predicting are based
on empirical relationships mainly. In order to improve this understanding, Smart Markers were installed at
Esmeralda Mines Bloque 2 aiming to understand and know, in a better way, the fragmented material flow
inside a through Block Caving method.

2 Objectives
The main goal of this work is to detail the installation process and markers registry performed at Bloque 2
Esmeralda Mine, by distinguishing between near and far field tests.

Gravity Flow

3 Methodology
For gravitational flow markers system deployment, stages for markers and readers installation are planned.
As for the markers installation, stages are defined for far and near field tests. Both of them detailed below.

3.1 Markers installation

3.1.1 Near field test

The near field are those measurement of flow under 30 from the undercutting level as shown in Figure 1.
The near field test were carried out between undercuts drifts C- 57 and C-59 which are spaced at 30 m apart,
with a maximum installation height of 31 m. The main goal of the tests is to study the interactions of flow at
the drawbell and through the minor and major apex. The markers are installed on radial drillholes ring (3
diameter) interspaced with blasthole rings used during the undercutting process.

Figure 1 Near field test design (isometric view and HW-FW section)

The installation in near field test considered the use of a spider with markers, which provides adherence
inside the drillhole. Once the first marker is installed, a 5 m long PVC bar is introduced, which allows to
keep the distance between each marker (avoiding vertical drifts from surrounding blasting). Once all of the
estimated markers are installed, each one of the drillholes is covered with wooden cones, which avoids the
whole markers complete column vertical descent.

Figure 2 Installation process in near field

Caving 2014, Santiago, Chile

The parameters used for markers installation in this test are listed in Table 1.

Table 1 Design parameters in short reach test

Near field design Number /Quantity

Number of stands of pipe 10
Number of drillholes per stand of pipe 9
Markers per stand of pipe 39
Drilled meters per stand of pipe (m) 217
Vertical spacing between markers (m) 5

3.1.2 Far field test

Far field flow measurements are carried out by installing markers above the 30 m, up to 70 100 m
(according to available drillhole length). The markers are installed in vertical drillholes with a maximum
length of 100 m, which are used in the pre-conditioning process. The goal is to obtain relevant information
(markers drifts, mainly), which allows vertical and lateral movement of the markers to be quantified for
medium/long term production planning. The used parameters for markers installation in this test are are
shown in Table 2.

Table 2 Design parameters in long reach test

Far field design Number /Quantity

Number of drillholes 6
Number of markers per drillhole 35
Drillholes length (m) 70-100
Spacing between markers (m) 2
Total number of markers 210

The design for markers installation in long drill holes is shown in Figure 3.

The steps for markers installations in vertical drillholes are the following:

1. Anchor installation: an anchor is installed in the drillhole bottom (6 diameter) with the rope linked to
the pulley. This process uses Wassara or Cubex drilling equipment to raise the anchor. One end is fixed
and the other one remains free. A slow advance must be considered, since any anchor tilting might
result in its loss.

2. Free end fixation to hoisting winch: once the anchor is installed, the fixed end must be adhered to a
hoisting winch strongly enough to rollback more than 100 m rope and rising the planned load (40 kg
approximately of rope).

3. On rope markers installation: markers are installed on rope with high resistance adhesive tape. They
are not installed with spiders, since it is possible they get stuck with the other ropes end.

Gravity Flow

Figure 3 Far field design

4. Rope lifting with markers: lifting is performed until the first installed marker gets to the drillhole
bottom or, until the ropes added load overcomes the winch hoisting capacity.

5. Free end fixation and security plug: the hanged load is secured and the vertical drillhole is blocked on
its base.

6. Grouted stage: at this stage a concrete bomb was used (Putzmeister equipment TK 40) given that the
grouted length is considerable. To do this a 2 shut-off valve was used at drillhole top, under which the
pipeline is connected to the equipment exit. A 1/2 diameter hose into the drillhole bottom allows to
know whenever it is completely grouted. This is shown in Figure 4.

Figure 4 Markers installation stages in vertical drillholes

Caving 2014, Santiago, Chile

3.1.3 Installation of markers from higher levels

Above Esmeraldas Bloque 2 is located the main haulage level; the railroad Teniente 5. The Superintendence
of Geology drilled a series of descending holes, aiming to perform auscultation of the rock mass, in order
to determine the caveback growing. Once these labors are performed, the drillholes are ready for markers
installation. The drillholes location and the markers installation can be seen in Figure 5.

Figure 5 Markers installation in XC-40 FFCC Teniente 5

The total installation in drillhole P-3 (see Figure 5) were10 markers. In this area markers will continue to
be installed as auscultation labors are carried out. Note that 10 markers we installed in the XC-40 box in a
row of 10 short drillholes of 60 cm each.

Considering the proposed design for the installation and the operational restrictions under which the process
was performed, there have been 397 markers as installation result (92 installed for far field test and 305 for
near field test):

Figure 6 Markers installation in near and far field tests

Gravity Flow

The near field installation covers a 3.192 m2 surface, which corresponds to 5 drawbells between production
drifts C-57 and C-59 (from drawbells Z-30 to Z-34).

The first vertical drillhole (P1T-C59) was installed in the area where Bloque 2 (drawbells Z-30 between
production drifts C-57 and C-59) mining started. The apparition of markers installed at heights of 30 m was
important for production control purposes, since it allows to identify through its extraction that the caving
was progressing as the extraction and area of the block increased.

3.2 Readers installation

In order to record most of the installed markers, it is necessary to place the readers, which capture a series of
information whenever the markers pass under the de antenna-reader set. The data that is collected includes
the date and time registration, ID number and marker type. Figure 7 shows the readers location in the
production level of Bloque 2.

Figure 7 Readers installation (1, 2 and 3) in Bloque 2 production level

These location of the readers near available in ore passes, ensures that all of the LHD equipment working
at the area had to pass through the antenna-reader set. Note that out of service readers were not detected
during the estimated time.

In March 2014, a reader was installed at Tenientes 8 main railroad haulage level, which is located at the
main haulage drift, on which Tenientes 8 main railway passes by. This reader will allow every marker
installed in project Bloque-2 (and also in the future ones) to be recorded, which turns into a better markers

4 Results

4.1 Markers registry

Up to March 2014 a total of 335 markers have been recovered. This equal to a 30% of recovery.

For the near field test case, 96 markers from a total of 305 have been recorded, which can be seen in Figure
8. Another four markers have been recovered from the far field. The recovery of the markers does not allow
to perform too many conclusions and this would require more information to be collected and analyzed.

Caving 2014, Santiago, Chile

Figure 8 Markers recovery in near field test markers installed under drawbells

In the near field test, 2 installation areas have been identified, in which the recovery percentage is different:

a. Markers recovery installed in drawbells: as for the markers installed under a drawbell mining direct
effect case, a greater markers recover can be seen, which gets to 58 markers (60% of the total markers
recovered in the near field test).

b. Markers recovery installed in minor apex: the markers recovery installed in the minor apex has been
lower in comparison to the markers installed in the drawbell (40% of the total recovered markers in
the near field test).

c. Markers recovery in far field test: as for the markers installed in vertical drillholes case, 4 markers have
been recorded up to date. These markers recovery and the heights evolution mined from the nearing
drawbells cluster can be observed below:

Figure 9 Markers recovery in far field test

In Figure 9, red color underlines the extraction height from the drawpoint, by which the markers were
recorded in the different indicated dates. Note that all of the entries match the area that shows greater
heights mined in the indicated date. In the near future further analysis will be carried out to describe the
flow from these results.

Gravity Flow

5 Discussion and conclusions

The smart markers test performed in Bloque 2 Esmeralda Mine validates the usage of this technology for
the future projects of the El Tenientes Division. This is based on the successful installation of 397 markers
installed, with no no major incidents and reaching a high level of reliability. Up-to-date, 100 markers has
been recovered at the mine.

Regarding the installation process, the main work in the future will be to improve the process of grouting
vertical drillholes above 70 m. This is key for this project as markers needs to be fixed to represent the flow
of the rock. One of the learnings of the process was the installation of a reader in the main haulage levels,
which allows increase possibilities record possibilities (and so getting a greater amount of information).

The markers record installed in vertical drillholes, that is above heights of 50 m, it is important, since
it helped to correlate to the information by the Operational Geomechanical area, which indicated the
connection of the caveback of Bloque 2 with Teniente 4-Sur, which is located in an upper level. From a
extraction point of view, this meant that the rate of draw was increased to 1 tpd/m2 on an area of 4.700 m2.

Brunton, I, Sharrock, G & Lett, J 2012, Full Scale Near Field Flow Behaviour at the Ridgeway Deeps
Block Cave Mine, in Proceedings of MassMin2012, Sudbury, Canada.

Caving 2014, Santiago, Chile

Experimental study of fines migration for caving mines

F Armijo BCTEC Engineering and Technology, Chile
S Irribarra Universidad de Chile, Chile
R Castro Universidad de Chile, Chile

Fine material has the potential to flow and migrate through the caved zone during its gravity flow. This
may have a large impact on the dilution content depending on the grades that fine rock could have. To date
there have been some theory and experiments which have tried to quantify the fines migration potential.
The objective of this paper is to present results of fine migration using experiments to quantify for a given
particle size and draw strategy to the fines migration. For this purpose, a pilot test was constructed to study
fines migration on block caving mine using a production level that is geometrical similar to a block cave.
In this case, fines and coarse particles were dried while the size of the coarse particles was thirty times the
size of the fine particles. The results show that no shear strain occurred when the draw was uniform from
drawpoints and then no migration occurs. On the other hand, shear strain occurred under isolated draw
and, therefore, fines migration was observed. The results shows that more research needs to be done in
terms of fine migration quantification.

1 Introduction
Gravity flow in block caving is one of the key mechanisms of ore drawing. Under this concept, gravity
is one of the most important parameters to allow the percolation of fines through coarse particles. If fines
fragments of caved rock can migrate more rapidly than coarse fragments, it may have significant impact
on ore recovery and dilution, particularly if the fines does not have mineral with economic interest (Pierce
2009). There is evidence to suggest that fines can accumulate in stagnant portions of the cave (Pierce 2009),
in this case, with the presence of water, highly unstable mud may be happened (Guest 2008, in Pierce 2009).
Then, fines migration is an important issue in mining business, therefore, understand and mitigate fines
migration is fundamental to generate value for the mining business.

According to (Laubscher 1994) the fines migration is related to the difference that exists between the rock
mass rating RMR of the rocks. Thus, an in-situ column with a large difference of RMR between the ore and
dilution has a higher height of interaction than an in-situ column with a small difference of RMR. A higher
height of interaction means also a smaller point of entry dilution (Figure 1).

Figure 1 Fines migration quantification through the HIZ (Laubscher 1994)

Gravity Flow

As noted by Pierce (2009) experimental studies of fines migration were performed by (Bridgwater et al.
1978, 9) who used a simple shear cell to deform a rectangular bed of spheres. The fines particles with the
diameter of dp were placed on the top of the bed of coarse particles with the diameter of db, then the rate
of percolation is given by:



- is shear strain;

- L (m) is defined as the mean percolation distance;

- k1 and k2 are two arbitrary constants equal to 20 and 8 respectively (Bridgwater et al. 1978, in Pierce 2009).

Note that it is necessary to apply a shear strain () to measure percolation (), shear strain and percolation
are directly proportional. Recently, the Equation 1 has been tested in PFC3D (Pierce 2009), (Hashim
& Sharrock 2009; Hashim 2011; Hashim & Sharrock 2012).This formulation has been included in the
software REBOP to calculate the percolation of fine particles (Pierce 2009).

From an experimental point of view, experiments large 3D physical model have been also included the
quantification of fines migration (Power 2004 and Castro 2007). In large 3D experiments, flow cannot be
observed, so researchers have considered the use of 2D experiments (Pineda 2012 & Orellana 2012). In this
article, it shows the results of an experimental program aim to quantify fines migration. The results of the
experiments are presented, so the audience could make their own interpretation of one of the paradigms of
block caving practice.

3 Methodology
3.1 Physical model design

In order to perform experiments some rules needs first to be complied, so the results of the model does
something to do with reality. There are three similarity types between a prototype (mine full-scale situation
under study and which include all the features of interest) and a model (simplified physical representation
of the prototype, in which the essential features are included), which are geometric, kinematic and dynamic

Two systems satisfy the geometric similarity when the distance between two homologous points
depends on a scale factor L.

Two systems satisfy the kinematic similarity when two events homologous occur at a time scale
factor T and

Two systems satisfy the dynamic similarity when the relation between the inertia and any external
force in two homologous points depends on the force scale factor F.

In this test, only geometry and kinematics similarity will be considered. Compaction and breakage could
only be observed if vertical load could be applied. In Table 1, the geometry scale factors are considered to
what have termed unconfined flow condition.

Caving 2014, Santiago, Chile

Table 1 Scale factor used in pilot model

From them and the geometry, experiments were conducted using a pilot model of 250 [cm] height, 70 [cm]
width and 23.3 [cm] length, which in a scale 1:200 represents 500 [m] of in-situ column and 140 [m] of
gallery. There are thirty-six drawpoints, which indicates three galleries; Figure 3 shows the pilot model. This
pilot model has been constructed considering geometry similitude conditions with a real mining operation.

(A) (B)
Figure 3 Front view of the physical model (A) and isometric view of draw system (B)

3.2 Physical model experiments

Two experiments were realized with different draw, uniform and isolated draw. Physical model is loaded
with coarse particle up to 240 [cm] and fine particles are loaded over coarse particle. The size distribution
of fine and coarse particles is showed in Figure 4.

Gravity Flow

Figure 4 Particle size distributions of fine and coarse materials

The average size of fine and coarse materials are 0.14 [mm] and 4.45 [mm] respectively, which correspond
to a ratio of 30. Each drawpoint in the model has a separate bin, so it is possible to determine the extracted
weight of each one. There is an electronic draw system which simulates the extraction at drawpoints. This
system is controlled by a computer and enables to move the particles in each drawpoint independently. Each
drawpoint has a sensor that identifies when material flows or produces a hang-up.

During the experiments, the point of dilution entry (PDE) was obtained, which is the drawn mass in a
drawpoint until the first fine particle is drawn, divided by the assigned mass for the point. For isolated draw
experiment the assigned mass corresponds to the total mass into the pilot model.

4 Results
4.1 Results of uniform draw

Figure 5 shows the flow pattern as a function of the draw stage which observed in experiments. In this case,
the dilution is represented by red color. The sequence is shown in terms of mass drawn. As illustrated in
Figure 5, fine migration is not happened in any percentage of drawn column. This is to the fact that there
is no shear strain when drawpoints are draw concurrently. At the approximate height of 20 centimeters the
irregular blue and green layers of material show no uniform movement so that we could infer that shear
strain is occurring (Figure 5 at 20% of mass drawn) . This is what Laubscher (1994) have termed height of
interaction which is a height at which fines migrates when 95% of the mass drawn have been drawn.

Caving 2014, Santiago, Chile

Figure 5 Images of uniform draw experiment.

4.2 Results of isolated draw

Because for this experiment fines migration not was observed, an experiment was set when only
one drawpoint was drawn. Figure 6 shows the flow pattern as a function of the stage of draw. In this
case, the dilution is represented by red color. The sequence is shown in terms of mass drawn. In this
case the flow zone was not uniform and deviated to the left side of the set up. Clearly also the isolated
draw created condition for shear strain. After drawing 60% of the total mass, fine migration occurred
due to the shear strain. Difference in movement speed in fine and coarse material were clearly identified.

Stagnant Zone


Figure 6 Images of isolated draw experiment

Gravity Flow

Table 2 shows the mean PDE for thirty-six draw points for uniform and isolated draw experiment
(considering dilution as fine red particles).

Table 2 Point of dilution of uniform and isolated draw experiment.

Uniform draw Isolated draw

PDE [% of material] 95% 60%
PDE [kg of material] 507 323

5 Conclusions
In this article we present some experiments aimed to quantify fines migration potential for gravity flow. For
uniform draw experiment there was no migration because there was no shear strain, which keeps straight
the most part of the time until they are close to the drawpoint so the migration occurs in the end. The
isolated draw experiments allow fines migration due to shear strain because the movement speed depends
on the height. It is expected that an isolated draw in a block caving mines precipitate the entry dilution when
it is located at the back; however, a uniform draw retards the entry dilution. These results should not be
taking as conclusive and many others experiments should be proposed and carried out to further quantify
fines migration. For example experiments should be conducted to find at which ratio of sizes (fines and
course) fines migration occurs. In addition, there are some complex phenomena that could affect migration
and are not modeled at the physical model specially the influence of water which could increase fines
migration for caving mines.

We would like to thank the Chuquicamata Underground Project for discussion of the above results and the
Block Caving Laboratory at University of Chile for the experiments being conducted.

Castro, R 2001, Escalamiento para modelo fsico de flujo gravitacional, Memoria para optar al ttulos de
Ingeniero Civil de Minas, Universidad de Chile, Santiago, Chile.

Castro, R 2007, Study of the mechanics of granular flow for block caving, PhD Thesis, University of
Queensland, Brisbane, Australia.

Laubscher, D 1994, Cave mining State of the art, Journal of the South African Institute of Mining and
Metallurgy, vol. 94 N10, pp. 279293.

Hashim, M & Sharrock, G 2012, Dimensionless percolation rate of particles in block caving mines,
MassMin 2012, 6th International Conference and Exhibition on Mass Mining, Canadian
Institute of Mining, Metallurgy and Petroleum.

Hashim, M 2011, Particle Percolation in block caving mines, PhD Thesis, University of New South Wales

Hashim, M & Sharrock, G 2009, Numerical Investigation of the Effect of Particle Shape on Percolation,43rd
US Rock Mechanics Symposium & 4th US-Canada Rock Mechanics Symposium, American
Rock Mechanics Association.

Caving 2014, Santiago, Chile

Orellana, L 2012, Estudio de variables de diseo del sistema de minera continua a partir de experimentacin
en laboratorio, Tesis de Magister en Minera, Universidad de Chile, Santiago.

Pierce, M 2009, A model for gravity flow of fragmented rock in block caving mines, PhD Thesis, University
of Queensland, Australia.

Pineda, M 2012, Study of the gravity flow mechanisms at Goldex by means of a physical model, Tesis de
Magister en Minera, Universidad de Chile, Santiago.

Power, G 2004, Modelling granular flow in caving mines: large scale physical modelling and full-scale
experiments, PhD Thesis, University of Queensland. Brisbane, Australia.

Gravity Flow

Towards an understanding of mud rush behaviour in

block-panel caving mines
ME Valencia University of Chile, Chile
K Basaure University of Chile, Chile
R Castro University of Chile, Chile
J Vallejos University of Chile, Chile

One of the most serious caving mine geotechnical risks are mud rushes. Mud rushes in block caving could
be defined as sudden inflows of saturated material from drawpoints. In the last years literature has been
written that tries to propose the causes for mud rushes. In general is accepted that mud rushes are due to the
presence of water, fines and extraction through drawpoints. This paper describes a framework to develop
a model to predict mud rush potential and to gain fundamental understanding from a geotechnical point of
view. In order to do that, the authors establish a limit equilibrium model and also carried out a geotechnical
characterization of mud obtained at El Tenientes mine at the lab. From this the main geotechnical indexes
are establish and a model to establish the shear strength of the mud as a function of density and water

1 Introduction
Block/panel caving operations can involve numerous hazards, one of those are mud rushes. Mud rushes
are sudden inflows of saturated fines from drawpoints or other underground openings (Butcher et al 2000).
The quick response of this phenomenon has terrible consequences for safety. Mud rushes are responsible of
numerous fatalities and severe damage to infrastructure.

Caving operations are inherently susceptible to mud rushes (Jakubec 2012). Due to the nature of caving
it has the potential of accumulating water from subsidence field as well as generating fines (comminution
process) during the extraction process. Persistence of both water and fine material could cause a mud rush.
El Teniente mine (CODELCO Chile) is not immune to the problem of mud rushes. According to Becerra
(2011), El Teniente history has plenty of examples. One of the last mud large rushes occurred in October of
2007. The incident inflicted an extensive restructure of the control and extraction in saturated drawpoints.
After this event, the operations policies have been set to limit the extraction rate and close areas when
drawpoints has presence of mud (Ferrada 2011). The strategy of restrict extraction has not only had a severe
impact on ore reserves but it is also unable to resolve the progressive appearance of mud in drawpoints.

There are four conditions necessary for a mud rush; water, mud forming material, a disturbance and a
discharge point. Operational experience shows these four are mandatory elements for the occurrence of a
mudrush (Butcher et al. 2000). According to published literature, there are several triggering mechanisms
of mud rushes which are classified based on the source of mud forming material and water. Table 1 resumes
the mud rushes classification proposed by Butcher et al. (2005).

Caving 2014, Santiago, Chile

Table 1 Mudrushes classification and proposed mechanism (Butcher et al. 2005)

Classification Mechanism
External Inflow of tailings. The material flows through a shaft, adit or open bench.
Failure of backfill in stopes. Material can flow through a barricade due to the poor
quality backfill.
Open pit slope failures. Mud flows due to the failure of a open cut slope.
Internal/Mix Formation of mud pockets in ore column with comminuted shale and rainwater. Rapid
muckpile compactation. Responsible of mud pocket discharge.

Today mine operations with mud rush risk deals with the problem of mitigation practices through:

Drawpoint categorization according to the percentage of fines and moist (Call and Nicholas et al.,

Draw control to ensure uniform draw (Butcher et al. 2000).

Limitation of ore reserves by height for specific drawpoints (Butcher et al. 2000).
Drainage to reduce the potential for mud rushes (Call and Nicholas et al. 1998).
Limitation of extraction rate and closure of areas with drawpoints containing mud (Ferrada 2011).
Mud rush score system (Holder 2013).
Geotechnical characterization of the mud was carried out by Call & Nicholas (1998) for IOZ mine (Freeport
McMoran Indonesia). On the other hand, Jakubec (2012) fulfil experiments about mudflow behaviour.
Both were conducted to know the material properties and size the flow potential. Likewise, they have
established failure mechanism for fine granular materials: mudflow and liquefaction. Nevertheless, they
have not suggested a model that explains a mechanism for mud rushes.

The main objective of this article is to provide a geotechnical model framework for mud rush potential
for a drawpoint. Then, results from the a geotechnical characterization of mud to El Tenientes mine are
described and tests to define the shear strength of the mud.

2 Framework for a geotechnical model of mud rush

Mud rushes are essentially a stability problem. Variations in water content and stress conditions can increase
the pore pressure and therefore the potential for sudden failure. In this case the granular material losses its
strength and behave as a fluid. Besides, they specify three kinds of mechanisms that make the granular
material fluid:

Static mechanisms: Related with the extraction of mud.

Dynamic mechanisms: Related with perturbations like blasting vibrations. These cause induced
liquefaction by seismic movement.

Water as a movement force: Related with the increase of water content. This can change the mud
properties, making the material fluid or drag along due the excess of pressure.

Gravity Flow

The mudrush issue in block caving could be considered similar to the case of mining with backfill,
particularly stopes with hydraulic and paste backfill. Hydraulic backfill are characterized by include fine
granular material and high water content from tails.

An analogy can be drawn between the geotechnical analysis of filled stopes and an isolated drawpoint. Both
are made in rock (drawbell and stope) filled with saturated fine material. Stope stability analyses allow
calculating the resistance of a bulkhead to seal off the extraction point. For block caving application stress
analysis is alike, except for the bulkhead. In block caving there is not a permanent bulkhead and only relies
on part of the broken ore at the drawpoint (see Figure 1).

As noted in Figure 1, the acting forces are from the weight of the material that fill the drawpoint (Fm + Fc)
and water (Fw). Filling material has two parts: broken ore and mud (Heights of this material are Hb and Hm
respectively). The horizontal component of this forces (FH) is trying to get the ore out of the drawpoint. The
shear resistance of this material (FB) is acting against the horizontal force.

Figure 1 Analytical model of a single extraction point

To be in equilibrium:

FB=Kh (Fm+Fc)+Fw (1)

A limit equilibrium analysis can be developed according methods to design barricades for backfilled stopes
under submerged conditions (Smith and Roettger, 1984). To understand how the mud pressure acts it is
necessary to acknowledge two components of total pressure: effective fill pressure and water pressure.
According to Smith and Mitchell (1982) the total bulkhead pressure for a fully saturated fill can be estimated
as equation (2). The first component represents the horizontal effective fill pressure on the pile of material
in drawpoint. On the other hand, the second component represents the water pressure on the drawpoint.


R is referred as a drainage ratio and is calculated according equation (3).


Caving 2014, Santiago, Chile


H = Fill height of mud in column (m)

m = Unit weight of the mud (kN/m3)

l = Pile of material length (m)

w = Pile of material width (m)

w = Unit weight of water (kN/m3)

P = Percolation rate in the drawpoint (cm/s)

P1 = Percolation rate in ore column (cm/s)

A = Total area of drawpoint (m2)

A1 = Ore column cross-sectional area (m2)

This model is useful for the first part of ore column, filled with mud. According to the incidents reported
to date, this height (H) should not exceed the height of the drawbell. The overload is considered. using the
Janssen (2004):



cr = Unit weight of caved rock (kN/m3)

= Internal friction angle ()

Rh = Hydraulic radius

= Friction coefficient ( tan ())

k = 1 sen2() / 1 + sen2 ()

z = Depth (m)

Then, the total stress on the pile of material in drawpoint it can be estimated as:


3 Case study

The use of the stress equation is illustrated with a sample application. The parameters used for the example
are showed in Table 2.

Gravity Flow

Table 2 Parameters for illustrative case

Parameter Unit Value

Depth (z) m 100
Unit weight of caved rock (cr ) kN/m3 19
Pile of material length (l) m 2
Pile of material width (w) m 4
Hydraulic radious (Rh) - 2.25
Internal friction angle () 45
Unit weight of the mud (m) kN/m3 27

Figure 2 shows the variation of the total stress on the pile of material as a function of drainage ratio through
the ore column. It is calculated for three heights of mud in a drawbell. It can be seen that stresses increases
when there is no drainage through the ore column. The stress of water only matters when it is accumulating
in the drawbell. When there are not drainage conditions (R = 0.1) more than a half of the total horizontal
stress acting at the drawpoint is due to the height of water .

Figure 2 Total stress on the pile of material as a function of drainage ratio (R) and mud height (H)

Once, stresses on drawpoint have been estimated, it is necessary to know the resistance of the pile of
material. This ore could be under different conditions of moisture and granulometry. In order to get the
correct parameters to estimate this resistance, a geotechnical characterization needs to be performed. This
was achieved through an understanding of the strength of the mud which is discussed in the next sections.

4 Geotechnical characterization
Samples of saturated fine material were collected from drawpoints classified as Critical in El Teniente
mine . The risk classification was implemented in the mine according to the fine material and water content

Caving 2014, Santiago, Chile

(Becerra, 2011).Following the standard procedure for representative sampling in El Teniente, three samples
were collected. This samples represent three different muds differentiated through the observed colour:
Grey (Sample 1), Yellow (Sample 2) and Mixture (Sample 3).

Two different test are implemented in this study:

1. Tests to define the index properties of material: Grain Size Distribution, Atterberg Limits, Specific
Gravity, Maximum and Minimum Void Ratio.

2. Tests to evaluate the relations between water content and compaction: Set of unconfined
compression and set of slump tests.

The outcomes of the first test are used define the conditions for the second tests. Full size distribution curves
were used to perform slump test and unconfined compression test. These tests are both carried out with
different saturation and relative density values.

Saturation, Void Ratio and Relative Density are defined using (6), (7) and (8) equations respectively.





Vw = Volume of water

Vv = Volume of voids

Vs = Volume of solids

Water content is a relation between mass of water and mass of solids. These measure are used to determine
the degree of saturation and mobility of the material. The Atterberg limits are a measure of critical water
content of fine soil. Those limits classify behavior of fine grain soil from solid to plastic (LP) and plastic
to liquid (LL). Density of the mud material is important in determining how much water and air can fill
the voids between particles, which effects the flow potential of the material. The void ratio is an important
material property in assessing strength permeability, and collapse potential. Tests are accomplished accord
ASTM standard and are described in Table 3.

Table 3 Performed tests for geotechnical characterization

Parameter Method
Water Content Oven Dry (ASTM)
Grain size distribution Sieve Analysis (ASTM)
Atterberg limits Method of Casagrande (ASTM)
Specific Gravity Picnometer, Submerged Mass (ASTM)
Void Ratio Maximum Minimum Density (ASTM)
Void Ratio Minimum Modified Proctor (ASTM)

Gravity Flow

5 Geotechnical Characterization results

Geotechnical indexes are used to understand the behaviour of saturated fine material. Table 4 synthetise the
main results. According to the results of laboratory test, the material is a well graded granular aggregate. It
contains gravel, sand, lime and clay in different proportions. Sample 2 seems to be the one with the most
portion of fines (22% limes and clays) and less gravel. While, sample 1 has the fewer portions of fines (11%
lime and clay) and most gravel (60%). Moisture contents range from 9.5 to 13.4 percent. Atterberg limits
produced a liquid limit range between 21.7 and 26.1 percent. The plastic limits are between 16.9 and 21
percent. These values indicate that the fines are classified as low plasticity silt and clay (ASTM D2487-00).

Table 4 Outcomes from geotechnical characterization

Sample 1 Sample 2 Sample 3

Water Content 9.50% 13.40% 12.50%
Grain size distribution D60=11.35 mm D60=3.907 mm D60=12.288 mm
D30=2.385 mm D30=2.43 mm D30=1.045 mm
D10=0.043 mm D10=0.006 mm D10=0.018 mm
Atterberg limits LL=21.7% LL=25.7% LL=26.1%
LP=16.9% LP=21% LP=19.1%
Specific Gravity (Gs) 2.76 2.68 2.72
Maximum Void Ratio (emax) 0.9 1.00 0.92
Minimum Void Ratio (emin) 0.26 0.28 0.22

Up to date tests to determine the strength for the mud have considered unconfined compression test. Figure
3 shows the unconfined strength for the three samples at different saturations and relative densities. Relative
densities are related to the maximum density achieved by the mud using a Proctor test. The results indicate
that resistance increases with a decrease on the relative density. At low relative density and high saturation
the mud shows no shear resistance. The water content for this case is over the plastic limit so the mud
behaves like a liquid.

Slump test indicate that in conditions of high relative density, there is no settling (no deformation). For a
relative density lower than 65%, the possibility of flow or plastic behaviors is depending on saturation. In
particular, when fluid behavior appears, water content is over or near the plastic limit. Sample 1, including
the most cases on fluid state, has the lowest limit.

The outcomes of the geotechnical characterization suggest that the resistance of the pile of material could
be overcome depending on the relative density. Relative density depends on the extraction of the mud at
drawpoints and it is not usually measured.

6 Conclusions
This paper outlines a geotechnical model framework for mud rush prediction. The new model represents an
instant before a mud rush. This model takes account three components: the weight of the mud in drawbell,
the pressure of water that is accumulated in the column and the overload of the broken material.

The simplified model is consistent with some mitigation practices like drainage. Drainage ratio is one of
the most important variable as well as relative density. When there are hardly drainage conditions (R = 0.1)
more than a half of the stress belongs to water pressure.

Caving 2014, Santiago, Chile

Figure 3 Unconfined compression test results (average between three samples)

Once the amount of load was estimated, a geotechnical characterization was performed regarding the
resistance of the pile of material. Saturated fine material was tested to know the behaviour of mud under
different conditions. Results indicate that plastic deformations could explain mud rushes for abrupt changes
of load or moist. Furthermore, relative density has an important role in the resistance of material.

Incorporate the extraction of material is the next step to understand the behaviour of mud under block
caving mining. This will be developing a numerical model and experiments at lab scale.

This paper has been prepared as an output of the Innova CORFO Project 12IDL2 - 15145. The authors wish
to thank Mauricio Melendez (CODELCO Chile) for provide the material of the experimental study, as well
as to CORFO that allowed the development of this research.

Becerra, C, 2011, Controlling drawpoints prone to pumping, El Teniente Mine, Proceedings of Second
international Seminar on Geology for the Mining Industry, Antofagasta.

Butcher, R, Joughin, W & Stacey, TR, 2000, A Booklet on methods of combating mudrushes in diamond
and base metal mines. Simrac.

Butcher, R, Stacey, T & Joughin, W, 2005, Mud rushes and methods of combating them, The Journal of
The South African Institute of Mining and Metallurgy, vol 105, no 11, pp. 807-824.

Call & Nicholas, 1998, IOZ Wet Muck Study, Freeport McMoRan Copper and Gold, C. & Hydrologic

Gravity Flow

Ferrada, M, 2011, Gravity flow under moisture conditions Control and management of drawpoint
mudflow, In 35th APCOM Symposium, Wollongong.

Jakubec, J, Clayton, R & Guest, A, 2012, Mudush Risk Evaluation, Proceedings of Massmin 2012,

Jansen, HA, 2004, Experiments regarding grain pressure in silos (Translated from german by W. Hustrulid
and Norbert Krauland), Proceedings of Massmin 2004, Santiago.

Holder, A, Rogers, AJ, Bartlett, PJ & Keyter, GJ, 2013, Review of mud rush mitigation on Kimberleys
old scraper drift block caves, The Journal of The South African Institute of Mining and
Metallurgy, vol 113, no 7, pp. 529-537.

Smith, JD & Mitchell, RJ, 1982, Design and control of large hydraulic backfill pours, CIM Bulletin, vol.
75, no 838, pp. 102-111.

Caving 2014, Santiago, Chile

Statistical analyses of mud entry at Diablo Regimiento

sector - El Tenientes Mine
IM Navia Universidad de Chile, Chile
RL Castro Universidad de Chile, Chile
MA Valencia, Universidad de Chile, Chile

Mudrushes have plagued block and panel caving operators with many fatalities and can have posed a
major hazard to safety in block and panel caving mining. Closing drawpoints with a high mudrush potential
can be introduced as an effective way of preventing mudrush hazards. Since draw point closure is due to
mudrush potential, not dilution, different amount of remnant saturated ore (RSO) would be remaining in
the block column. Tonnage and grade of RSO in this group were calculated based on the actual situation
of closed drawpoints. The second group contains drawpoints located in a zone with a high potential mud
entrance. In this group, the RSO that could potentially be removed once mud enters in drawpoints was
predicted based on the historical extraction data. The results indicated that RSO is itself an interesting
quantity in terms of tonnage and average grade. Respecting to occurrence of mud, the initial inflow of mud
was associated to drawn heights and draw uniformity that are similar to in situ height of the initial entrance
area. It is proposed that the subsequent entry of mud resulted not only in relation to the connection with
higher mined levels, but other mechanisms, such as the entry of water directly from surface.

1 Introduction
A flow of mud in block/panel caving, called mudflow, mudpush o mudrush, is defined as a sudden
and violent inflow of a mixture of water and fines to mine openings, with a high injury, death and damage
potential. A mudflow can damage equipment, cause operating losses and even, can cause fatalities (Butcher
et al. 2005).

The existence of mud at broken columns can cause two effects: mudflows, either violent mudflows or
less violent spills of mud; and the redefinition of reserves due to the cutting of the drawable heights. The
last with the purpose of not including mud drawing in mine planning due to safety actions and technical
capability (Barahona 2014, pers. comm., 03 February). This meant that there is ore that cannot be extracted
which has been termed remnant saturated ore (RSO). There is a need to quantify the economic potential of
RSO in drawpoints which have been closed to prevent the hazards of mud-water. Nowadays, three statuses
have been defined to face the mudrush hazard in El Tenientes sectors:

1. Mud-water status, or critical zone, has the most probability of mudrush occurrence and, thus have
been closed forever to prevent the entrance of mud. A wet muck classification matrix has been
developed, to define the mudrush risk considering fine material and moisture percentage (Becerra

2. Limited status, which is happened in the drawpoints surrounding the mud-water status. These
drawpoints have the hazard of lateral immigration of mud from the critical zone. Therefore,
the extraction rate from these drawpoints is limited. The result of extracting the limited status
drawpoints is mud immigration to operative drawpoints.

Gravity Flow

3. Barrier status, in which some drawpointss ore columns are used as a barrier to control the entry of
mud. The content of moisture and fine material in these drawpoints are not necessarily critical but
they are estimated as high risk points based on the flow direction of mud. Due to a null extraction
in this statue, the lateral advance of mud is paused or delayed (Vargas 2013, pers. comm., 30

In order to minimize the problem of dealing with mudrush in production areas, a number of researchers
have suggested to draw uniformly (Widijanto et al. 2012; Laubscher 2000). This strategy would allow
the extraction of mud in many drawpoints, as a result, would prevent mud concentration in just a few
drawpoints. An extensive dewatering program can also reduce water which leads to the wet muck runs
(Barber et al. 2000). Drainage strategy can be implemented in both surface water, which enters the cave
through rain falling onto the subsidence, or underground water. (Samosir et al. 2008; Barber et al. 2000).

In this article the economic potential of RSO in the closed drawpoints as well as those that could be closed
in the future due to the ingress of water-mud was calculated. Furthermore, the historical databases of
resources, reserves and extraction conditions at a mine of El Teniente known as Diablo Regimiento was
used to define the relationship between the appearance of mud and the drawn strategy based on the back
analysis statistical method. It should be noted that this database includes all resource and production history
of Diablo Regimiento from the initial date of extraction to November 2013.

2 Economic potential of RSO

The economic potential of the RSO at Diablo Regimiento was calculated based on the column model and
production data history of each drawpoint. Through this database it is possible to identify grade, tonnage
and density of each bench in every draw column. Due to mudrush hazards, some draw points have been
closed before reaching the economic drawable heights; therefore, two groups of drawpoints are introduced
to determine the accurate RSO as:

Drawpoints affected by mud-water: This group includes drawpoints with mud rush hazards
which are categorized in three statuses (Mud/Water, Limited and Barrier). To evaluate RSO,
both economical and marginal drawable heights are considered, under which the minimum and
maximum RSO defined respectively. Table 1 shows the results in each drawable height. As it is
indicated in table 1, in the case of marginal drawable height, two different cutoff grade was taken
into consider.

Drawpoints not affected by mud-water: This group includes drawpoints located in the zone under
upper mined levels which are already drawn (from East to West, they are Regimiento, Puente and
Fortuna). The drawpoints are considered with a high mud entrance potential. It should be noted that
the drawpoints considered in the previous section were excluded.

Table 1 shows the results of RSO calculation based on marginal drawable height for two various cut off
grades. As illustrated in table 1 the minimum RSO is 11.8 Mt ore material with the average grade of 0.63%.

3 General analysis of mud occurrence at Diablo Regimiento

According to critical matrix used in El Teniente copper mine (Becerra 2011), if the percentage of fine
material and moisture content in a drawpoint reaches the critical value the status of drawpoint will changes
to Mud/Water status. In this situation the drawpoint will be closed to prevent the hazard of mudrush. In this
paper the historical database of closed drawpoints is considered as a situation that mud occurrence.

Caving 2014, Santiago, Chile

Closure grades

Closure grade is defined as the final extracted grade of a drawpoint which is closed due to the mudrush
hazards. In figure 1 the closure grade of different drawpoints at Diablo Regimiento is illustrated. It can be
concluded from this figure that most drawpoints are closed with high copper grades.

Table 1 RSO at Diablo Regimiento

Drawpoints Total of Average grade
Considerations RSO (Mt) grade
considered RSO (Mt) (%CuT)
Affected by mud- Initial reserves at
4.4 0.72% 4.4 0.72%
water Diablo Regimiento
Affected by mud-
Considering marginal 14.9 0.62%
heights; cutoff grade 26.6 0.57%
Not affected by equal to 0.4%CuT 11.7 0.51%
Affected by mud-
Considering marginal 11.4 0.67%
heights; cutoff grade 18.8 0.63%
Not affected by equal to 0.5%CuT 7.4 0.57%

Figure 1 The frequency of closure grades at Diablo Regimiento

3.2 Drawn heights

The drawn heights were calculated using daily drawn databases as well as resources model per bench and
drawpoint. The historical database of Diablo Regimiento is used to analyses closed drawpoints in the case
of closure sequence and drawn heights.

Gravity Flow

3.2.1 Closure Sequence

A plan view of closed drawpoints at Diablo Regimiento is illustrated in figure 2. It is observed in figure 2(a)
that the initial mud entry drawpoint is at the center of the Diablo Regimiento sector, and it coincides with
the drawpoints where extraction began in order to generate the dome. Subsequently, the mud was always
appearing in neighboring points, and then appeared in the east sector. Based on the drawn height in figure
2(b) it is possible to compare the extraction height at different part of Diablo Regimiento sector.

Figure 2 (a) Closure sequence and (b) drawn height of closed drawpoints

3.2.2 Drawn heights at closed drawpoints

Figure 3 shows the frequency of drawn heights in closed drawpoints at Diablo Regimiento sector. The
high variability is illustrated for different drawpoints. Based on figure 3 and 2(b), it appears that the lower
elevations correspond to the drawpoints that begin connecting with mined and caved upper levels; that is
the center of the sector and the east side. Subsequently, the neighboring points to the aforementioned are
associated to a greater drawn height before the apparition of mud.

Figure 3 Drawn height frequency of closed drawpoints

It is observed that 30% of the drawpoints in Mud/Water status, were at a lower or equal to 170 m height
drawn, corresponding to the approximate distance between Diablo Regimiento and mined upper levels. In
addition, at that height is where the highest frequency of closure drawpoints occurred. Occurrence of mud
in these drawpoints may be through to the accumulation of water and mud in the overlying Regimiento
sector (Diablo Regimiento is below Regimiento). However, the entry of mud at a greater drawn height may

Caving 2014, Santiago, Chile

be attributable to the draw maintaining an irregular drawing profile combined with the accumulation of
mud-water on the surface. Mud entrance below 170 m may be due to vertical or lateral migration of mud
already within broken columns.

Analyzing the drawn height of closed drawpoints (figure 4) observed that as mining production progresses,
the range of possible heights closed drawpoints is increased. It can be concluded from figure 4 that the first
mud entrance is due to connection with upper sectors; but in other drawpoints that were extracted latter in
the sequence, it may have been other reasons for mud entrance (water from other sources).

Figure 4 Evolution of accumulated drawn height at Mud/Water status declaration

According to the analysis of the data the following could be concluded:

Rock caving commencement in virgin areas has a great influence on the potential of mud entrance
to the sector. For example, the dome shape of cave back in the center of a mine sector causes an
early interaction to the surface or to mined levels located above. This creates channels through
which fine material and water would entrance to production level.

In general an effective way of avoiding mud ingress is through uniform draw so as to bring the ore/
mud interface as horizontal as possible.

It is important to detect the sources of water and mud and to define a strategy for dewatering and
for a draw strategy to face high potential areas for mud ingress.

4 Determining the probability of mud entrance

As it is illustrated, the RSO could have an important role on the reserves evaluation; therefore the mid
and long term production planning would be changed based on RSO. Since RSO is the result of mud
occurrence in draw columns, it is essential to determine in advance the entrance of mud in drawpoints. As
a result, a model is proposed to predict the probability of mud occurrence employing a logistic regression.
This model can be used as a mine planning tool. The main data which are considered in this model are
temporal evolution of draw rate, fine material content, drawn height and season of the year. The last is due
to a correlation that could exist between water seasons and drawpoints closed due to mud. Some aspects
related to mudflows are described below.

Gravity Flow

A logistic regression model is proposed in this study to predict the Mud/Water status based on historical
data gather at Diablo Regimiento sector. Logistic regression is a technique for making predictions when
the dependent variable is a dichotomy, and the independent variables are continuous and/or discrete. In
order to predict the risk of mudrush hazard, a logistic regression used with dependant variable (p) as the
probability of persistence of mud. For each drawpoint the obtained value of p shows if there is mud in the
column (p=1) or not (p=0).

Formally, logistic regression model is defined as equation (1).


Solving for p, this gives equation (2).


Where i (i=0 ,, n) are the estimators and Xi (i=1 ,, n) the independent variables including:

X1: Draw rate

X2: Fine material content

X3: Drawn heigth

X4: Season

The variable season added to this model because in the probability of mud occurrence in spring is more than
other seasons of the year.

The estimators obtained for equation (1) are shown on equation (3).


Based on logistic regression model, the persistence of mud in various drawpoints in Diablo Regimiento
sector evaluated. Figure 5 shows the results. It is illustrated in figure 5 that this method enables to predict
mud occurrence in different part of the sectors. Moreover, the precision of model is 74%.

5 Conclusions
In this paper, statistical analyses of database at Diablo Regimiento sector was used to study the effect
of different parameters on mud occurrence in drawpoints. Based on this study, it is concluded that the
accumulated drawn height could be the most influence parameter in controlling mud entrance.

According to data analysis in this research, it is concluded that in the case of irregular profile of drawn
heights, uniformity and continuously strategy cannot solve the mudrush problem. In this situation, first
objective of short term production should be obtaining a uniform drawn height profile. After reaching this
objective, uniformity and continuously strategy seems to reduce the above mentioned problems.

Moreover, in this study the economic potential of RSO is evaluated. Even though closing drawpoints is
the best way to ensure safety in production level, the results of economic evaluation shows that RSO are
potential to provide at least one half year production of this sector.

Caving 2014, Santiago, Chile

Figure 5 Status of drawpoints on Diablo Regimiento sector; (a) real status of drawpoints and (b) predicted
status of drawpoints (W: wet drawpoints; D: dry drawpoints)

Finally, it is illustrated that logical regression method can be used to predict the mudrush hazards in different
part of sector. Further research needs to be conducted in order to evaluate the mud potential ingress for mine
planning purposes.

The authors acknowledge the assistance of Rodrigo Barrera, Max Barahona, Ricardo Vargas and Antonio
Pinochet from Codelcos El Teniente Mine, and Asieh Hekmat from BCLAB, for their helpful support.
This research has been funded through Corfo and by the Conicyt funds for the Advance Technology Center
(AMTC) of the University of Chile.

Barber, J, Thomas, L, Casten, T 2000, Freeport Indonesias Deep Ore Zone Mine, in Proceeding
of Massmin 2000, Brisbane, Queensland, 29-October-2-November 2000, ed. G.C., The
Australasian Institute of Mining and Metallurgy, pp 289-294.

Becerra, C 2011, Controlling Drawpoints Prone to Pumping - El Teniente Mine, in International Seminar
on Geology for the Mining Industry, 8-10 June 2011, Antofagasta, Chile.

Butcher, R, Joughin, W & Stacey, T 2000, Methods of Combating Mudrushes in Diamond and Base Metal
Mines, SRK Consulting, The Safety in Mines Research Advisory Committee (SIMRAC),

Butcher, R, Stacey, TR & Joughin, WC 2005, Mudrushes and methods of combating them, The Journal
of The South African Institute of Mining and Metallurgy, SAIMM, Volume 105, pp. 817-824.

Laubscher, D 2000, A Practical Manual on Block Caving, International Caving Study.

Samosir, E, Basuni, J, Widijanto, E & Syaifullah, T 2008, The Management of Wet Muck at PT Freeport
Indonesias Deep Ore Zone Mine, in Proceeding of Massmin 2008, Lule, Sweden, 9-11 June
2008, eds. H. S. & E. N., Lule University of Technology, Lule, pp. 323-332.

Widijanto, E, Sunyoto, WS, Wilson, AD, Yudanto, W & Soebari, L, Lessons Learned in Wet Muck
Management in Ertsberg East Skarn System of PT Freeport Indonesia, Proceedings of Mass
2012, Sudbury, Canada.

Caving 2014, Santiago, Chile



Hybrid composite, a way to enhance the mechanical

properties of breakable ground support

V Barrera Mining and Metallurgy Innovation Institute IM2 Codelco, Chile

P Lara Mining and Metallurgy Innovation Institute IM2 Codelco, Chile
G Pinilla Codelco, Chile
F Bez Codelco, Chile

When old mining workings levels are caved, the traditional ground support that has offered a safe place for
mining could become a problem for the continuity of the production process. This problem is related to the
flow of unbreakable traditional support elements (tramp iron) in the ore conveyance lines and comminution
circuits in underground mining. For this reason, the need to find breakable systems for ground reinforcement
is an open problem. This issue is partially addressed by market alternatives of fibre-reinforced polymers.
However, these products that warrant a good tensile strength show poor deformability due to their brittle
nature. IM2s Geological Mining Area and GT&I Codelco Chile are carrying out a research study to
develop breakable ground support elements based in the hybridization effect applied to fibre-reinforced
polymer to obtain systems with enhanced mechanical properties.

1 Introduction
Cave Mining, used to mine large ore bodies such as the tasks of Codelco, uses gravity and considers the
existence of different levels in their design. These levels, once mined, are successively abandoned and
the mining is deepened. Each of these levels requires the use of support systems in order to ensure a safe
working environment for both operators and machinery. The basic function of the support and containment
systems in the rock mass is to help self-supporting because every time an underground excavation is made,
the natural tendency is to occupy the empty volume and return to its undisturbed condition. This return to
balance is done by stress redistribution around the excavation, resulting in a gradual deformation of the
excavated cavity. However, when these processes exceed the mechanical strength of the rock surrounding
the excavation, this may cause the breakage and shedding of blocks and, in extreme cases, cause violent
rock bursts that occur when the existence of brittle rock is combined with a high stress concentration. It is
under these conditions that the support systems contribute to create a safe working environment during the
mining works.

The traditional support systems are made of ferrous elements due to their high plasticity and they become
unbreakable tramp iron when they are installed into new production caving levels. Table 1 provides a
description of the mechanical properties of the steel used in rock bolt making (Carvajal 2008).

Table 1 Minimum mechanical properties of steel used for rock bolts

Tensile Strength 440 MPa

Elongation 16.0%
Shear Strength 251 MPa
Density 7850 kg/m3

Caving 2014, Santiago, Chile

For this reason, the passage of these unbreakable materials in conveyance and crushing lines in mining
operations introduces disturbances in the continuity of the productive process. Sometimes, this scenario
will create unscheduled shutdowns, increasing the risk of not fulfilling the commitments stipulated in the
production planning.

A way to solve the problem of unbreakable underground support, chosen by IM2 Geomining - Codelco, is
to design new elements using materials that have the capability to withstand stresses inherent to the mining
works and that become breakable once they are absorbed into the caving. Modified fibreglass composite has
been chosen for this purpose because it meets both requirements. To increase its ability to deform (trying to
reach the properties of ferrous elements), the modified fibreglass composite has been provided with a core
of ductile material (minimum volumes of steel), using the hybridization effect (Marom et al. 1978) applied
to fibre-reinforced polymer. Tensile and shear tests in a universal testing machine have been made to obtain
measurable parameters under static loads.

2 Methodology

2.1 Fabrication

Two types of Hybrid Spun Bars (HSB-X1, with SAE 1045 steel core, and HSB-X2, with A630-420H steel
core) were manufactured by pultrusion. This production technique is a low-cost, high-volume manufacturing
process in which resin-impregnated fibres are pulled through a die to make the part. The process is similar
to metal extrusion, with the difference being that instead of material being pushed through the die initially,
it is pulled through the die in a second process. Pultrusion creates parts of constant cross-section and
continuous length (Mazumdar 2002). Figure 1 shows this forming technique.

Figure 1 Schematic pultrusion process

2.2 Test

A survey conducted in the Chilean supplier market, carried out by IM2-CODELCO, determined the strength
of Glass Fibre Reinforced Polymer (GFRP) bars (Barrera et al. 2013). This information is summarized in
Table 2.


Table 2 Properties for GFRP bars supplied in the Chilean market

Tensile Strength 506 MPa

Elongation 7.31%
Shear Strength 158 MPa
Density 709 kg/m3

To determine the properties of these elements, two types of static tests were considered: tensile and shear
tests. The breakability property will be research in further step of this investigation.

2.2.1 Tensile Test

The standard tensile test is the uniaxial tensile test. In this case, as the test sample was made of polymeric-
nature materials, the ASTM D7205 standard (ASTM 2006) was applied.

Figure 2 Tensile test and HSB specimen

2.2.2 Shear Test

The shear strength of the hybrid prototype bars was determined using a double shear test based on ASTM
D7617 (ASTM 2011). To execute this test, a piece specially designed for this purpose, which was assembled
in the universal testing machine, was used.

Figure 3 Parts and assembly of HSB specimen for double shear test

Caving 2014, Santiago, Chile

3 Results

3.1 Fabrication Results

Figure 4 shows the raw materials that were used and the result of the pultrusion process, which are the
hybrid spun bars, HSB.

Figure 4 Raw materials and HSB (right bottom corner picture)

3.2 Test Results

The tests performed was made to evaluate the mechanical strength characteristics of this elements. As it is
indicated, Figure 5 shows the fragile fracture mode in both, tensile and shear tests.

Figure 5 HSB specimens after tensile (left) and shear (right) tests

Table 3 summarizes the results of tensile and shear tests.

Table 3 Properties for HSB prototypes

HSB-X1 Tensile Strength 445 MPa; Elongation 10.8%; Shear Strength 191 MPa; Density 1,473 kg/m3
HSB-X2 Tensile Strength 392 MPa; Elongation 14.4%; Shear Strength 190 MPa; Density 1,592 kg/m3


In the light of these results, a comparison can be made with both ferrous elements as well as GFRP bars
provided in the market. This comparison is provided in Table 4, where the loss (gain) is shown in percentage
with respect to the base state for this research (rock bolt steel and GFRP bar).

Table 4 Comparison among prototypes and existing technology (rock bolt steel and GFRP bar)

Prototype With Respect to Tensile Strength Elongation Shear Strength Density

HSB-X1 -1.10% 32.5% 23.9% 81.2%
Rock bolt Steel
HSB-X2 10.9% 10.0% 24.3% 79.7%
HSB-X1 12.1% -47.7% -20.9% -108%
HSB-X2 22.5% -97.0% -20.3% -125%
-: Percentage gain with respect to; +: percentage lost with respect to

4 Conclusions
The tensile strength of the HSB-X1 prototype shows the best behaviour with respect to both base states:
rock bolt steel and GFRP bar. Furthermore, it also shows an increase of about 50% elongation and 21%
in the shear strength of the GFRP bar. The loss of about 32% in the shear strength with respect to the steel
rock bolt leaves a margin for further research to improve this property with new material cores. As for the
properties shown by the HSB-X2 prototype with respect to the GFRP bar, they are lower than the previous
prototype, except for elongation, where it shows a response close to the minimum accepted for steel. Both
prototypes show a lower density than rock bolt steel, however, this property is enhanced with respect to the
GFRP bar.

The authors acknowledge the sponsorship of Codelco in the context of the completion of Project API
M11DE12 Conceptualization and Experimentation of Breakable Ground Support Elements. In addition,
Patricio Lara is grateful for the valuable assistance of Cristian Welsch, who made it possible to fabricate the
breakable prototypes at PERNOMIN Ltda.

ASTM International 2006, ASTM D7205 Standard Test Method for Tensile Properties of Fibre Reinforced
Polymer Matrix Composite Bars, ASTM International, Pennsylvania.
ASTM International 2011, ASTM D7617 Standard Test Method for Transverse Shear Strength of Fibre
reinforced Polymer Matrix Composite Bars, ASTM International, Pennsylvania.
Barrera, V, Lara, P, Pinilla, G, Arancibia, E 2013, Breakable Ground Support a verification of mechanical
properties to diminish ferrous Solid Waste in Underground Mining, Proceedings of Copper
2013, ed. IIMCH, Santiago.
Carvajal, A 2008, Manual Sistema de Refuerzo de Rocas con Pernos Saferock, Gerdau Aza, Santiago.
Marom, G, Fischer, S, Tuler, FR, Wagner, HD 1978, Hybrid effects in composites, Journal of Material
Science, vol. 13, pp. 1419 1426.
Mazumdar, S 2002, Composite Manufacturing Materials: product, and process engineering, CRC Press,
Boca Raton.

Caving 2014, Santiago, Chile

Pilot tests as a tool for the design of autonomous mining

J Riquelme University of Chile, Chile
R Castro University of Chile, Chile
S Valerio University of Chile, Chile
J Baraqui Codelco Chile, Chile

For many years mining innovation has relied on full scale tests to verify the implementation of new
technologies. At the University of Chile, a methodology has been developed for using pilot tests for mining
engineering with the aim of speeding the process of design and implementation of novel technologies.
In this article, we present the potential of this approach for the conceptual design of an autonomous
and continuum mining system using scaled models. Autonomous means that the system could operate
without people taking decisions on when each dozer would draw. This is achieved using sensors (lasers
and cameras) and a control system for the dozers. This research also assists in the understanding of the
behaviour of interactions between the components of the Continuous Mining System. The results indicate
the extraction sequence and the use or not of controlling systems could increase the draw rate for this type
of material handling systems.

1 Introduction
Continuous Mining is a new material handling system designed to increase the rate of draw for block
caving mines (Encina et al. 2008). This technology has been tested at Codelcos El Salvador Division
between 2006 and 2007 and is currently under construction at Andinas Division. At the same time a
number of experiments using scaled and numerical modelling have been carried out at the University
of Chile (Alvarez 2010; Orellana 2011; Orellana 2012). The Continuous Mining system consist on the
use of novel-to-block cavings material handling equipment: (1) stationary feeders located in draw points
(dozers); (2) a continuous conveyor (panzer) which receives the material from dozers; and (3) a crusher in
order to reduce the size of material (Encina et al. 2008) (Figure 1-Left). The system which is currently being
tested at Andina Mine could operate 8 dozers per panzer as shown in Figure 1-Right.

Figure 1 (Left) General scheme of extraction level in a Continuous Mining module (Encina et al. 2008); (Right)
Plan view scheme of extraction level used in the current industrial test. The arrows indicate the direction of
movement of the material


One of the key factors of successful continuous mining systems is achieving higher extraction rates
compared to LHDs. Therefore, it is necessary to simulate the results of interaction between equipment
and materials. In this article the results of a pilot test program are presented to illustrate the interactions
between the components of the Continuous Mining System. The results also include the implementation of
Autonomous and Continuous Mining System for future applications in mining.

2 Pilot tests

2.1 Objectives

The aim of the pilot testing was to simulate the extraction of a production drift in a Continuous Mining
module (Figure 1). This research focused on the production rate and the potential interaction of the dozer
- panzer system. The steps performed in this research were as follows:

1. Conducting a scaling study and construction of a 1:50 scale pilot test with eight dozers and one
panzer as considered on the detailed engineering (JRI Ingeniera 2010). In this case both the
geometry and model media were scaled accordingly. The material with which the tests were
performed corresponds to approximately 1.2 tonnes of gravel, with a size range between 6.35 mm
and 40 mm with a mean diameter of 16mm (Figure 2).

2. Commissioning: detection and resolution of operating system problems.

3. Continuous Mining Experimentation: testing stage without sensing or control systems.
4. Autonomous Continuous Mining Experimentation: development of tests with sensing and control
systems for autonomous operation of the dozers.

5. Analysis of experimental results: comparison and analysis of results obtained after the completion
of the tests, generating conclusions and recommendations for future studies.

Figure 2 (Left) Materials sample used in experiments; (Right)Particle size distribution of the sample

3 Laboratory equipment
The laboratory equipment consisted of a model frame, extraction equipment, sensing system and control
system (Figure 3). The model framework consisted of a plexiglass framework of 1.58 m width x 0.57 m

Caving 2014, Santiago, Chile

depth x 0.82m height. In this model, eight drawbells were built also in plexiglass to observe the hang ups
and gravity flow during the tests.

Figure 3 Physical model used for experiments: (Left) Physical model with the electrical and actuation panel,
and a computer desk; (Right) Perspective of filled physical model

Figure 4 shows different views of the scaled dozer which is located under the drawbell with a width of 40
mm (Figure 4-Center). The extraction system operates by compressed air, pushes the upper part and mobile
part through a cylinder. In this scaled model it is possible to measure the pressure for each movement of
the dozer.

Figure 4 Dozer system in the scaled model: (Left) Frontal view of dozer; (Center) Side view of dozer;
(Right)Plan view of dozer gallery

The second equipment of the Continuous Mining System is a Panzer. Based on experiments, this part of the
system was represented by a belt conveyor with gaps as in the chain system (Figure 5). The belt operates by
two electrical engines located at the end of it.


Figure 5 Panzer construction in the scaled model: (Left) Perspective view of panzer gallery; (Right)Plan view
of panzer

The sensing system consists of four volume sensor prototypes designed and implemented for the studied
physical model. The system is based on the principle of structured-light photogrammetry, which is a
combination of image processing and structured light. A video camera, which obtains a static image of the
material on the panzer, is installed on the roof of the panzer gallery and is oriented at a specific inclination
angle to the horizontal axis of the gallery (Figure 6). The image processing algorithm outlines the edges or
the contours of the transported material based on the obtain image.

Figure 6 Sensing system in the scaled model: (Left) Side view of the sensing system; (Right)Perspective view
of panzer gallery with sensing system

Caving 2014, Santiago, Chile

The last component is the control draw system (hardware and software). The aim of the control system is
to maximize the amount of material as well as the uniform load in the panzer. To do that the system starts
with an assumption of the amount of material draw per dozer which is then compared to the measurements
at the panzer by the sensing system. The system then corrects the amount of material in the panzer and
actives if required other dozers in the panzer. The implementation of the control system is multi-platform:
it has been tested in Windows, Linux and OS-X systems. It has a control interface in which it is possible to
set extraction rates, dozers state (active or inactive), the speed of the panzer, and cycle times of the dozer.
It also shows a model representation of the load distribution in panzer.

4 Experiments and results

4.1 Experiments

Until now five experiments have been run at the pilot tests infrastructure. The experiments were run to set
up the pilot tests, to quantify the extraction of the dozer and panzer system, to test different draw sequences
and to test the autonomous system. The aim of each experiment is presented in Table 1. The main variables
are as follow:

Dozer dump length (Dl) (mm): is the panzer length which is used by the drawn rock from a dozer.
Dozer productivity (Dp) (g/cycle): is the amount of rock per cycle of the dozer.
Panzer utilization (Pu) (%): is the percentage of the total length of the panzer used by rocks. The
panzer utilization is calculated as Pu = Lg / Lt; In which Lg is the portion of panzer with visible
material (discounting panzer gaps) and Lt is theoretical length estimated considering the first
and last particle out of the panzer, independent of the vertical extension of material. Utilization
was estimated using ten consecutive cycles of the system (case without automation) and full test
duration (case with automation).

System production rate (Sp) (g/min; t/h): is the total broken rock extracted by the continuous
system (panzer).

Dozer sequence: experiments considered three operational dozers configurations (Figure 7).
TypeI:Two alternating dozers operating from the same panzer gallery side; Type II: Two extreme
and two central dozers operate; Type III: Only the four center dozers operate.

The main parameters used in tests are the following:

Panzer speed (Ps) (cm/s): fixed parameter scaled from the full scale speed of the panzer.
Dozer cycle time (Dt) (s/cycle): is the time takes to make a dozers dump.
Distance between operational dozers (Dd) (mm): is the distance between the centers of the
operational dozers galleries.

Maximum number of simultaneously running dozers (Nd): is the maximum number of permitted
operational dozers due to the full scale constraints.


Table 1 Experiments conducted in this research

Exp. Objectives
0 Check system function
1 Characterize the interaction ofdozerpanzer system
2 Determine main operational configurations and generate dozerssequences
Determine number of cycles required to reach a steadystate
Estimate panzer utilization

3 Determine productivity of the Continuous Mining for a given sequence (base case)
4 Determine productivity and panzerutilizationoftheAutonomous Continuous Mining system
with sensing and control components

Figure 7 Operational dozers configurations: (Left) Type I configuration; (Right) Type II configuration;
(Bottom)Type III configuration. Active dozers are the marked by grey colour

4.2 Results

The results of the dozer and panzer systems are shown in Table 2. Values were scaled using the 1:50 scale
factor. The results indicate that when a dozer is activated an average of 19 cm (9.6 m scaled) is used by the
drawn rock. This has a variability of 9 cm (4.8 m scaled) with each discharge from the dozers. Thus the
results indicate that when a dozer discharges the used length is variable as well as the amount of drawn rock.

A second series of experiments consisted in testing the continuous mining (experimental base case) and the
autonomous system for different draw strategies. The time to activate a dozer in the sequence was calculated
using the average dozers dump length, the distance between the dozers and the panzer speed (Barriga

Caving 2014, Santiago, Chile

2012). For both studied cases type III configuration is the one that produces more. This is possibly due to
the greater proximity of the active extraction points, resulting in the interaction of the respective movement
ellipsoids. Table 3 indicates the percentage increase of productivity and panzer utilization associated by
using an autonomous system, compared to the base case.

Table 2 Statistic results of dozers dump length and dozers productivity

Laboratory scale Scaled results

Dump length Productivity Dump length Productivity
[cm] [g/cycle] [m] [t/cycle]
Average 19.18 76.35 9.59 9.54
Standard deviation 9.76 74.38 4.88 9.30
Minimum 0.50 0.25 0.25 0.03
Maximum 50.50 423.72 25.25 52.97
Cycles number 400 400 400 400

Comparing the same draw sequence, the autonomous continuous mining productivity is, on average, 58%
higher than the uncontrolled system. Regarding panzer utilization, the average for automation case is 7%
higher than nonautomation case. Thus, it can be concluded that Autonomous Continuous Mining has also
a higher panzer utilization.

Table 3 Percentage increase of productivity and panzer utilization for studied cases

Configuration Type I Type II Type III Average

Productivity 50.2 60.6 62.0 57.6
Panzer utilization 1.0 21.4 -2.1 6.8

In general, productivities and utilization of the system increase through the implementation of the sensing
and controlling systems.

5 Conclusions
This research work confirmed the usefulness of pilot testing tool towards the understanding of behavior of
innovative mining systems. It was verified that, under experiments and particle size distribution conditions
presented, the dozer-panzer system is a high production system and that the draw per dozer is high and

The results of testing the Continuous Mining System subjected to experiments conditions showed that
configurations using central extraction points allow reaching higher extraction rates.

The results from the Autonomous Continuous Mining testing subjected to experiments conditions show
that the implementation of an autonomous control system allows the achievement of higher productivity,
extraction rates and panzer utilization.

Based on the results obtained, it is recommended to implement the autonomous system in the mine in order
to verify the increase of productivity of the system. It would be beneficial to perform new experiments
under a more demanding scenario. For example, incrementing the particle size distribution with the aim of
quantify the reduction in system productivity. It will be relevant to identify the potential problems, which


this technology could face in the future in case of not achieving the expected particle size distribution.
Finally, it is recommended to integrate uniformity indicators towards maximizing the production rates and
the extraction of reserves in long term production planning.

The authors would like to acknowledge the financial and technical support from Codelco Chile. The authors
would also like to acknowledge the contribution of Ernesto Arancibia from Codelco Chile and to IM2
(Institute for Innovation in Mining and Metallurgy) researchers during this project.

Alvarez, P 2010, Modelamiento fsico de la Minera Continua, Memoria de Ingeniera, Universidad de
Chile, Santiago, Chile. (in spanish)

Barriga, J 2012, Secuencia Accionamiento Dozer, Nota Tcnica N IA-004, IM2, Santiago, Chile. (in

Encina, V, Baez, F, Geister, F, & Steinberg, J 2008, Mechanized continuous drawing system: A technical
answer to increase production capacity for large block caving mines, Proceedings of Mass
Mining 2008 Conference, Lulea, Suecia, pp. 553-562.

JRI Ingeniera 2010, Informe de la Ingeniera Conceptual y Bsica de la Validacin Industrial Tecnolgica
de la Minera Continua, Santiago, Chile. (in spanish)

Orellana, L 2012, Estudio de variables de diseo del sistema de Minera Continua a partir de experimentacin
en laboratorio, Tesis de Magister, Universidad de Chile, Santiago, Chile. (in spanish)

Orellana, M 2011, Modelamiento Numrico de la Minera Continua, Tesis de Magister, Universidad de

Chile, Santiago, Chile. (in spanish)

Caving 2014, Santiago, Chile

Implementation of LiDAR technology to evaluate

deformation field induced by panel caving exploitation,
Codelco Chile El Teniente Division
AE Espinosa Codelco, Chile
P Landeros Codelco, Chile

Empirical experience and theoretical developments in rock mechanics have demonstrated that cavities
generated by extraction in cave mining induce changes in deformation fields, reaching different extensions
over time while their location varies according to the mine growth. Considering changes in deformation
field as a result of induced stress field on a rock material by caving, extensive recording of this information
is an important data that permits to evaluate the performance of mine design and allowing short term
engineers to have the chance to take appropriate actions, if deviations are identified.

LiDAR technology is based on the principal of calculating laser pulse time of flight (TOF). Therefore, if
the information, such as, laser pulse velocity, angular reference used to measure and the difference of time
between emitted and reflected ray is known, it is possible to determinate the relative distance of an obstacle
or object.

This work explains in detail the implementation of these concepts to geomechanical monitoring at Dacita
Project at El Teniente mine, showing results obtained as a the baseline measurement and during a
comparative analysis while considering mining activity and ground control information.

1 Introduction

1.1 El Teniente Mine overview

El Teniente Mine is a Codelco Chile underground copper mine. It is located in the Andes range in the
central zone of Chile, approximately 70 km SSE from the capital city, Santiago. El Teniente is the largest
known coppermolybdenum deposit in the world. It is hosted in a copper porphyry system. The main rock
types include Andesites, Diorites and Hydrothermal Breccias of the Miocene era. Since 1906, more than
1,100 million tons of ore have been mined. The mine is currently extracting approximately 140,000 tons/
day using mechanized caving methods. Panel and post-undercut caving methods, variations of the standard
block caving, were introduced in 1982 and 1994, respectively to exploit primary copper ore.

1.2 Dacita Project overview

El Teniente Mine includes different productive sectors, all of them located around a chimney of sub-
volcanic breccias with an inverted cone shape, known as Braden Pipe.

Dacita Project is located on the western side of Reservas Norte Mine and it corresponds geometrically to
an extension of that productive sector (Figure 1). Its exploitation started in November 2013 and it considers
a production plan close to 17,000 tons/day for the year 2019, using a conventional panel caving (post-cut
undercutting) and an integrated mining sequence of both mines.


Average primary columns heights are 180 meters, varying from 90 to 250 meters, finding lower values
beneath Sur Andes Pipa Mine and Pipa Norte Mine. Those values could be associated with low stress state
regime, however, the size of all the cavities around Dacita Project induces a higher level of pre-mining
stress very similar to Reservas Norte Mine.

Figure 1 Schematic view from the north of Dacita Project footprint. It is possible to identify its location
between cavities (Espinosa et al. 2012)

1.3 Geological and geotechnical data

According to Brzovic (2012), predominant lithology corresponds to Dacite Porphyry, which in terms of
intact rock properties is very stiff with average Youngs Modulus of approximately 60 GPa. In terms of rock
mass quality indexes, Dacite Porphyry is very competent with GSI in the range of 75 to 90. Most important
geological faults are classified as master faults (faults G, C, N1 and N2) and major faults (faults F, K, L).

2 Geomechanics monitoring plan concepts

Considering changes in deformation field as a result of induced stress field on a rock material by caving,
extensive recording of this information provides important data that permits to evaluate the performance of
mine design and allowing the short term engineers to take appropriate actions, if deviations are identified.

Therefore, some of the main objectives of the geomechanics monitoring plan are based on the following

To be able to identify changes, in terms of induced stresses and deformations, affecting

excavations in the productive levels.
To provide new field information for numerical modelling calibration.
To estimate zones affected by the mining advance.

Caving 2014, Santiago, Chile

Rockburst is one of the most relevant geomechanics risks identified in the development of the geomechanics
engineering of Dacita Project. To complement its evaluation, vulnerable zones are identified, using an
integrated methodology, considering geological aspects, size of the excavations and stress conditions,
selecting the places to be measured and then, comparing results with field information obtained by ground
control engineers. In specific cases, identification is complemented by numerical modelling analysis (Cuello
et al. 2010).

Major geomechanical hazards, such as, rockburst and collapses that affect large panel caving operations,
are highly dependent on a caveback geometry (Landeros et al. 2012). Based on the assumption that any
geometrical change is related to a change of the stress field, the geomechanics plan considered the concept
shown in Figure 2.

Figure 2 Interpretation process of monitoring results (Espinosa 2012)

According to all the aspects mentioned previously, one of the areas of geomechanics monitoring plan
was focused on the implementation of Light Detection and Ranging technology (LiDAR) as a tool for
measuring the geometries of excavations and their changes in time.

Development of LiDAR technology started in the 1970s in USA and Canada, used with satellites for
topography scanner with high cost and many limitations. With higher development of informatic technology,
it is currently used in many different fields. The device basically works emitting laser light pulses to
determinate the distance between surfaces and its position, generating clouds with millions of points. This
technology is highly accurate and precise and it is widely used for surveying measurements both open pit
mining and underground mining.

There are different types of LiDAR but it is not a matter of this study to describe all types. The one used
in this case corresponds to a laser-based ranging and imaging system, terrestrial and static (mounted on a
tripod), capable to capture data in medium and long distance inside the galleries.


Principal advantages of this technology, as compared with traditional methods, such as, convergence
stations or extensometers, are related to less operational interference and the capability to capture new
information in a wider extension and not only at one selected point.

3 Preliminary results and data processing

3.1 Preliminary results

First stage of measurements started in July 2012, in both the undercut and the production levels; it
represented the base line for future comparisons. During June 2013, new measurements were completed at
specific locations, based on an excavation vulnerability criterion; approximately 45% of the surface was
compared in each level. Figure 3 shows an example of captured data in a production drift.

Figure 3 Example of production level drifts LiDAR measurements at Dacita Project (Espinosa & Landeros

The decision to include LiDAR technology in the development of Dacita Project involved a new challenge.
There are at least four different documented algorithms for the calculations of the measurement results,
each one of those based on certain assumptions and limitations.

3.2 Existing distance measurement methods

The approaches described by Girardeau (2006) and Lague et al. (2013), used to measure the distance
between two point clouds in the context of geomorphologic applications, are shown in Figure 4 and are
described in the following paragraphs:

Digital elevation model (DEM) of difference:

DEM of difference is the most common method of point cloud comparison in earth
sciences when the large scale geometry of the scene is planar.

Caving 2014, Santiago, Chile

The two point clouds are gridded to generate DEMs either directly if the large scale
surface is near horizontal. The two DEMs are then differentiated on a pixel-by-pixel basis
which amounts at measuring a vertical distance.

This technique is very fast but it suffers from a major drawback: it cannot operate properly
on 3D environments or rough surfaces.

Figure 4 Existing 3D comparison methods between two point clouds PC1 and PC2 (modified from Lague et al.

Direct cloud-to-cloud comparison with closest point technique (C2C):

This method is the simplest and fastest direct 3D comparison method of point clouds as it
does not require gridding or meshing of the data, nor calculation of surface normal.

For each point of the second point cloud, a closest point can be defined in the first point
cloud. In its simplest version, the surface change is estimated as the distance between the
two points.

Improvements can be obtained by a local model of the reference surface either by a height
function or by a least square fit of the closest point neighbours.

The measured distance is sensitive to the clouds roughness, outliers and point spacing.
Cloud-to-mesh distance or cloud-to-model distance (C2M):
This approach is the most common technique in inspection software. Surface change is
calculated by the distance between a point cloud and a reference 3D mesh or theoretical

This approach works well on flat surfaces as a mesh corresponding to the average
reference point cloud position can be constructed. However, creating a surface mesh is
complex for point clouds with significant roughness at all scales or missing data due to


Interpolation over missing data introduces uncertainties that are difficult to quantify. Mesh
construction also smooth out some details that may be important to assess local roughness