You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/269984574

Empirical methods in mining geomechanics – Reflections on current state-


of-the-art

Conference Paper · June 2014


DOI: 10.13140/2.1.1217.9520

CITATIONS READS

3 4,075

1 author:

Fidelis Tawiah Suorineni


Thinkrox Inc
68 PUBLICATIONS   804 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

underground mininig View project

Mining Systems View project

All content following this page was uploaded by Fidelis Tawiah Suorineni on 24 December 2014.

The user has requested enhancement of the downloaded file.


Empirical methods in mining geomechanics – Reflections on current state-of-the-art

F.T. Suorineni
UNSW Australia, Sydney, Australia

ABSTRACT: Geomechanics rely greatly on empirical methods for excavation design


as a means of overcoming the complex nature of the rockmass and the associated difficulties in defining its
constitutive behaviour. These empirical methods include the Rockmass Rating (RMR) systems (Bieniawski,
1973), the Tunnelling Quality Index (TQI) or Q-system Barton et al. (1974), Mining Rockmass Rating
(MRMR) system (Laubscher & Taylor 1976), The Geological Strength Index (GSI) (Hoek 1994), Hard rock
pillar desigm chart (Lunder and Pakalnis, 1997), The Equivalent Linear Overbreak Sloughage (ELOS) stabil-
ity graph (Clark & Pakalnis 1997), the cavability prediction chart (Laubscher, 1994) and the Hoek and Brown
failure criterion (Hoek & Brown, 1980). This paper gives credit to the developers of these methods who
through their diligence and observation brought some sanity to somehow various chaotic periods in the history
of mining geomechanics. Current state-of-the-art shows some of these methods, while valid at the times of
their developments, have reached their limits in our time and new thinking is required. There is also evidence
that these methods are misused. The misuse sometimes leads to unjustified criticisms of the methods and
originate from lack of understanding of the various databases and the underlying assumptions that governed
the development of the various methods. The paper draws attention to the significant roles played by empirical
methods in geomechanics and emphasizes the need for further development and continuous improvement of
these methods to manage the complexity of the rock mass. To achieve this requires consistency in the expan-
sion of the existing databases, calibration of any new factors and the collection of more quality data. The pa-
per cautions users of these methods to pay due attention to the limits of the databases and their underlying as-
sumptions for reliable designs..

1 INTRODUCTION while in numerical modelling the constitutive physi-


cal descriptions must be externally linked to the
1.1 Empirical methods in mining geomechanics rockmass characteristics in question. To ensure that
Predictive tools in geo-engineering can be grouped the formulated rockmass constitutive equations be-
into empirical methods and numerical modelling. hind a numerical modelling code that respect physics
Empirical methods are based on field observations closely represent the behavior of the rockmass in
of the rockmass behaviour and field experience. service numerical models often have to be calibrated
They are a methodical documentation of expert against field observations for their predictions to be
knowledge. Franklin (1993) defines empirical de- close to reality. For this reason Franklin (1993)
sign as quantified judgment based on experience. states that empirical methods often proof ‘closer’ to
Numerical modelling codes are based on the phys- the ‘truth’ than the apparently more precise predic-
ics of the problem and are attempts to predict the tions of theoretical analysis, adding that based on re-
rockmass behaviour from its constitutive physical al data empirical methods provide a standard against
description. Mathews et al. (1981) stated that empir- which theoretical predictions are measured and can
ical methods based on back analysis are powerful be judged.
predictive tools, particularly if combined with nu- The complexity of the rock mass cannot be fully
merical modelling and analysis techniques. The au- accounted for in our constitutive models that under-
thor believes in the power of empirical methods. lie numerical models. While these numerical models
Both empirical methods and numerical modelling can be tricked to give us the answers we want, the
rely heavily on the adequate characterization of the rock mass is so arrogant it will behave in the manner
rockmass. In empirical methods the rockmass char- it wants. Müller (1988) noted that the dominating
acterization is implicit in its observed behaviour, geologic conditions at site do not care what kind of
theoretical ideas we may have and what our econom- 2 REFLECTIONS ON CONTEMPORATY
ic situations may be. He continued to state that the EMPIRICAL METHODS
geologic conditions will act in the way which is pre-
2.1 The rockmass rating (RMR) system
determined by their nature on one hand as well as by
the manner in which we treat it during excavation RMR (Bieniawski 1973) became the first widely ac-
and support. cepted and used rock mass classification system fol-
lowing its development in 1973. Its popularity
stemmed from the fact that it could be used for ex-
1.2 Contemporary empirical methods cavation design in rock with significant capacity to
The following empirical methods are discussed for predict excavation stand-up time. RMR was also
their popularity and widespread use in geomechan- adopted by Hoek and Brown (Hoek & Brown 1980)
ics: for the determination of the Hoek and Brown failure
− The Rock Mass Rating (RMR) system criterion parameters. RMR was later replaced by the
(Bieniawski 1973) Geological Strength Index (GSI) in 1994 (Hoek
− The Tunnelling Quality Index (Q) system (Barton 1994) for that role because it became ineffective in
et al. 1974) performing that function for values of RMR less
− The Mining Rockmass Rating s(MRMR) system than about 25.
(Laubscher & Taylor 1976) Various critiques of RMR have been published,
− The Laubscher rules in block cave mining out of which evolved the Modified or Mining Rock
(Laubscher 1994) Mass Rating (MRMR) system (Laubscher & Taylor
− The Geological Strength Index (GSI) (Hoek 1976). MRMR later became the source of cavability
1994) prediction in block cave mining, and remains so to
− The open stope stability graph (Mathews et al. today. These later developments of the RMR are dis-
1981; Clark & Pakalnis 1997) cussed in later sections.
− The Equivalent Linear Overbreak Slough (ELOS) Milne et al. (1998) provide a chronological devel-
stability graph (Clark & Pakalnis 1997) opment of RMR between 1973 and 1989. Within
− The hard rock pillar design graph (Lunder & Pa- this period the RMR factors have been modified as
kalnis 1997), and more experience in its application was gained.
− The Hoek and Brown failure criterion (Hoek & Suorineni (2013) critically re-examined the origi-
Brown 1980). nal RMR database for its validity, robustness. Focus
was on its application independent of the geological
It cannot be overemphasized that the great pio- environments or rock types and depth. The validity
neers of rock mechanics, geomechanics and engi- of the method with regards to its stand-up time pre-
neering geology (including Karl Terzaghi, Ralph diction was also examined.
Peck, Leopold Muller, Evert Hoek, John Franklin, Figure 1 provides the composition of the RMR da-
Denis Laubscher, Nick Barton, Z.T. Bieniawski and tabase in terms of rock origin as published in
Rimas Pakalnis) recognized the importance of geol- Bieniawski (1989).
ogy, observed and learnt from the arrogant rockmass
behaviour and guided by their intuition managed to
discipline it. Empirical methods are the outcome of
patient observations, intuition and keen interest in
geology. Igneous
This paper pays glaring tribute to these great men (20%)
by reflecting on their achievements. It then recogniz- Sedimentary
es that while some of these contributions worked (63%) Metamorphic
well to solve the problems of their time they have
now reached their limits and new thinking is urgent- (17%)
ly required. The paper then draws on the lessons
from medicine and science to suggest that for signif-
icant breakthroughs in rock mechanics or geome-
chanics, genuine, multidisciplinary approach sup-
ported by generous funding and rigorous supervision
is required. Figure 1. Composition of the 1989 RMR database with refer-
ence to rock origin.

Figure 1 shows that the RMR (1989) database


consists of 63% sedimentary rocks, 17%% meta-
morphic and 20% igneous rocks. Out of the 63%
sedimentary rocks about 45% are shale (Figure 2).
Figure 2. Composition of the RMR sedimentary rock origin by
rock type. Figure 4. RMR span-stand-up time concept (From Barton &
Bieniawski 2008).
Figure 3 is a plot of depth (m) against frequency
data points. The figure shows that about 90% of the While stand-up time may also also apply to some
RMR data came from depths less than 500 m below igneous rocks such as olivine rich rocks, it is diffi-
surface. cult to argue the concept of stand-up time for exca-
vations in these rocks. As pointed out by Müller
(1988) while timing is important in rock engineering
or geomechanics it can hardly be computed or even
assessed without deep geological knowledge and in-
tuition. This fact is further buttressed by the fact that
“one cannot wait until the rock itself announces its
stand-up time by roof falls and slabbing of the side
walls” (Müller 1988). One must make decisions be-
fore the stand-up time is reached.
Isaacson (2007) states that Albert Einstein’s intel-
ligence and breakthroughs hinged on his ability to
mix intuition with a feel for the patterns of nature to
be found in experimental data. He adds “The scien-
tist has to worm these general principles out of na-
ture by discerning, when looking at complexes of
empirical facts, certain general features.” These ob-
servations reflect on the complexity of the stand-up
time concept and what is required to make it mean-
ingful.
Thus, while RMR brought discipline and logic in-
Figure 3. RMR 1989 database composition relative to depth. to rockmass characterization at a time of chaos, and
contributed significantly to understanding this com-
From Figures 1 to 3 it is obvious that RMR con- plex medium, we must admit its limitations in con-
sist of soft rocks dominated by shales from depths temporary applications.
less than 500 m below surface. The significance of
this revelation on the composition of the RMR data- 2.2 Q-system
base is discussed under the Laubscher block caving
rules (Laubscher 1994, 2001) later in this paper. The The Q-system (Barton et al. 1974) (Equation 1) is
implication of the database as it relates stand-up time well detailed in several publications, and the discus-
is discussed next. sion here is limited to what many never venture to
Sedimentary rocks, and in particular, shales are look – the foot notes and assumptions.
susceptible to the vagaries of the environment on
exposure depending on their composition. Domina-  RQD  J r  J w 
tion of the RMR database by sedimentary rocks Q =     (1)
gives logic to the stand-up time concept (Figure 4).  n  J a
J  SRF 
where RQD = Rock Quality designation, Jn= joint up of discrete blocks defined by continuous joints.
set number, Jr = joint roughness number, Ja = joint This interpretation implies that the three rock masses
alteration number, Jw = joint water reduction factor shown in Figure 5 will have the same Q value and
and SRF = Stress reduction factor. therefore the same self-supporting capacities. Obvi-
The Q-system is a major contribution to rock me- ously, this is not the case intuitively and in our expe-
chanics/geomechanics and Barton et al. (1974) de- rience.
serve commendation. Today, there is a general feel-
ing that its applications have been too much
extended beyond its database limits. QTBM is one
such unproven versions of the extended application.
Palmström and Broch (2006) provide a critical re-
view of the Q-applications and chronicled its various
developments between 1974 and 2002. They advised
that potential users of the Q-system and for that mat-
ter any other empirical method, should carefully
study the limitations of the system before using
them.
The quotient (RQD/Jn) in Q is defined as the Figure 5. Hypothetical rockmasses: A contains continuous joint
block size, (Jr/Ja) the interblock shear strength and sets; B contains two discontinuous joint sets; C continuous and
(Jw/SRF) the active stress. What seems to be an im- one discontinuous joint sets.
plicit problem in Q is the definition of the term
block size. The implication of interpreting RQD/Jn For the reasons given above, Suorineni et al.
as block size is that any rockmass will have at least (2008) proposed a modified Q called Q-star (Q*)
three continuous joint sets to define finite blocks. (Equation 2) that accounts for inpersistent joint sets
Hence, for rockmasses with less than three continu- in the conventional Q-system. Care was taken to re-
ous joint sets Q will underestimate their self-support tain the original Q-system procedure.
capacity. This has implications in using the stability
graph for predicting cavability in block cave mining.  RQDo  J r* 
Q = 
*
J d   (2)
This is discussed later in a different section.  J n  J a 
Some users do not pay attention to the detailed
guidelines in the use of the Q-system, in particular Where RQDo is RQD measured in the direction of
the footnotes. The Q-system footnotes are critical in excavation, Jn is joint set number based on total
its proper application. Oversights of the footnotes number of joint sets, Jd and Jr star account for non-
have resulted in unwarranted criticisms of the meth- persistent joint sets and are determined from Tables
od by some users. Here are some significant points 1 and 2 respectively.
in Barton et al. (1974):
− Joint orientation relative to tunnel axis did not Table 1. Values for Jd (Suorineni et al. 2008)
Number of persistent joint sets Correction factor, Jd
appear to be a significant factor because the data- 0 4
base is from civil tunnels where tunnels are often 1 3.5
placed in the best orientation. 2 1.5
− Different personal, national and continental engi- ≥3 1
neering practices lead inevitably to variations in
methods of support, even for the same quality of Table 2. Values for Jr-star (Suorineni et al. 2008)
rock. The author adds that in mining, regulations Joint persistence (m) Description Jr-star
will also dictate support levels and practice. <1 Very low 5.0
− Support recommendations in the original Q sys- 1-2 Low 4.5
tem support chart are for permanent support. 2-3 Moderate 4.0
3-4 High 3.5
− Support recommendations assume good blasting ≥4 Very high 3.0
or excavation practice. For better drilling and
blasting or poorer drilling and blasting (deviation
from average) the Q system recommended sup- By virtue of the fact that most of the initial data in
port may tend to be conservative or inadequate Q came from civil engineering tunnels, it is not sur-
respectively. To account for poor blasting practice prising that relative orientation was not considered a
adjust Jn and RQD accordingly. significant factor for inclusion in the Q-system since
− The joint set with minimum Jr/Ja should always most civil tunnels are aligned in the best orientation
be used in computing Q. for stability. In mining, engineers have no luxury to
choose the drift alignment in the most stable orienta-
The block size and interblock shear strength defi- tion. Often, engineers have to deal with the effect of
nitions seem to imply that every rock mass is made
relative orientation of excavations with geologic ficiencies of RMR were provided in Section 2.1.
structures in stability analysis. Thus, the Laubscher block caving rules are based on
Another implicit problem with the Q-system is weak sedimentary rocks largely made of shale from
the difficulty in applying it to weak rockmasses. In shallow depths. Additionally, the rules were devel-
weak rock masses the Q-system parameters are diffi- oped in the South African geological environment.
cult to determine and excessive deformation may be Figure 6 and Figure 7 compare stresses in undercuts
the mode of failure rather than structural or brittle in stress environments where ko is 1/3 and 3 as may
failure. This weakness of the Q-system appears not be in S. Africa and Australia or Canada respectively.
to be obvious to many users of the method. Løset While in Figure 6 caving will easily propagate in
(1999) states as follows: poor quality rockmass in Figure 7 conditions do not
“The Q-system was primarily suited to hard joint- favour cave initiation in the same rockmass. Cave
ed rocks …….. but for the classes of poorest rock initiation will later likely stall with cave back shape
quality the system has not provided detailed descrip- changes in Figure 6 while in Figure 7 cave growth
tion of the support constructions. This means that for may later become rapid after initial preconditioning
weak rock masses the dimensioning of the support to initiate growth.
must usually be verified by numerical modelling or
some other means of calculation.” 6
5 σc
Løset (1999) identify 6 types of weak rock masses σz=3σx
as follows: 4
− Heavily jointed rocks in strong (Q<0.01) 3
− Zones with altered or strong rock (Squeezing or σ/σz 2 σc

swelling may take place)


σx H

− Weak rocks with joints (Young sedimentary rocks 1 σt B


(Sandstone) or weakly metamorphic rocks such as 0
shale, slate or phyllites) -1
− Weak rocks with excavation induced fractures 1 2 34 5
(e.g. Young homogeneous sedimentary rocks B/H
(chalk, sandstone) or low metamorphic rocks Figure 6. Stresses in undercut where ko=1/3 (Redrawn from
(Shale, phyllites) Coates (1981).
− Weak rocks without joints or induced excavation
fractures (Young homogeneous sedimentary rocks It is concluded that while the Laubscher cave de-
such as chalk, sandstone and mudstone), and sign rules worked well at the time of their develop-
− Weathered rocks. ment when mining was at shallow depths and in
weak rockmasses, these rules have reached their lim-
its as current mining is taken place at depths far
2.3 Mining rockmass rating (MRMR) system greater than 500 m and in more competent rock-
masses.
MRMR is obtained by applying adjustment to RMR
to account for mining induced factors. Adjustments 6
are made to account for weathering, blast damage, 5 σz=σx/3
orientation and induced stresses. 4 σc
3 σc
By virtue of RMR database limitations discussed σ/σx 2 σx
in Section 1.2, the MRMR system is subjected to the σt H

1 B
same limitations. 0
There are however additional benefits of the -1
MRMR system over RMR apart from the adjustment 1 2 34 5
for mining purposes. MRMR uses modified ratings B/H
of RQD and uniaxial compressive strengths to make Figure 7. Stresses in undercut where ko=3 (Redrawn from
rockmasses more sensitive to changes in these pa- Coates 1981).
rameters.

2.5 The geological strength index (GSI)


2.4 The Laubscher block caving rules
The Geological Strength Index (GSI) was developed
The Laubscher block caving rules are worthy of to replace RMR in the determination of the Hoek-
credit. To date, they continue to provide guidance for Brown failure criterion rockmass constants m, s and
block cave mine design, safety and productivity. a. This followed several years of use and experience
The Laubscher block caving rules are derived with the RMR in the determination of the Hoek-
from MRMR that has its source from RMR. The de- Brown failure criterion constants. Hoek (1994) ob-
served that while the use of RMR in determining the authors who did not understand the assumptions and
Hoek-Brown failure criterion constants was accepta- the stability graph database. Criticisms that the graph
ble for rockmasses with RMR values of more than does not account for tension (Diederichs & Kaiser
about 25 it did not work well for rockmasses with 1999; Mitri et al. 2010) are good examples.
RMR less than 25. Hence a new rockmass classify- Mathews et al. (1981) advise that for tension the
cation system specifically for the determination of stress factor should be set to 1 since for these condi-
rockmass properties was required. GSI (Hoek 1994) tions bridging type failures will not occur and only
was introduced as that rockmass classification sys- gravity driven structural failures are likely for which
tem. the gravity factor C takes over. Thus tensile condi-
The GSI was developed following many years of tions were not an oversight in the development of
discussions with engineering geologists leading to the stability graph.
precise wording in each box of the block size col- The stability graph is also misused as pointed out
umn and to the relative weights assigned to each in Suorineni (2011). The stability graph was devel-
combination of structural and surface conditions oped for large or wide orebodies. Application of the
(Hoek 1999). GSI is based on an assessment of the method to narrow vein orebodies often shows the
lithology, structure and condition of discontinuity method does not work in these orebodies. The meth-
surfaces in the rockmass and it is estimated from od is also supposed to be applied to steeply dipping
visual examination of the rockmass exposed in sur- orebodies in good quality rockmasses and should not
face excavations. It combines two fundamental pa- be applied to orebodies in poor quality and or low
rameters of geological processes that cannot be cap- dipping orebodies.
tured in any equation, the blockiness of the mass and The developers of the stability graph stated at the
the conditions of discontinuities that govern rock- time that while the factors for calculating the stabil-
mass behaviour. Marinos et al (2007) wrote: “….. at- ity number N (Equation 3) were sufficient to concep-
tempts to quantify the GSI classification to the per- tually prove the method more data was needed to
ception that engineers are happier with numbers are validate their reliability.
interesting, they have to be applied with caution in
order not to lose the geologic logic in the system.
Unfortunately, Hoek et al. (2013) wrote “…….  RQD  J r 
N =    ABC (3)
Some uniformity and quantification of the chart (the  J n  J a 
GSI chart) is necessary.” The author considers this
reversal of logic after the previous statement as un- The conceptual development of the method used 26
fortunate. case histories from three mines. Following this de-
It is also recognized that at depth joints are tight velopment Potvin (1988) expanded the database to
and jointed rockmasses may behave elastically. GSI, 175 case histories from 34 mines and calibrated the
like RMR does not account for confinement at stability graph factors A, B and C with a redefinition
depth. Confinement at depth will tend to strengthen of the transition zones. Potvin’s work undoubtedly
rockmasses. It is therefore concluded that at depth enhanced confidence in the method as an open stope
GSI in its current form will tend to underestimate the design tool. Consequently, it is expected that in re-
strength of rockmasses. cent applications of the method and databases the
calibrated stability graph factors should be used.
The extended Mathews stability graph method for
2.6 The open stope design stability graph open stope design (Mawdesley 2002; Trueman and
The open stope design chart by Mathews et al. Mawdesley 2003) consists of an impressive 483 case
(1981) is extensively discussed in the literature fol- histories. Unfortunately, this database used the orig-
lowing its development in 1981. These discussions inal uncalibrated stability graph factors. More dis-
focus on the limitations and suggested improvements turbingly, the database consists of not just data from
to the method in the nearly three decades of its exist- open stopes but data from other mining methods as
ence. Major contributions to the method include the shown in Figure 8. These observations clearly show
work of Potvin (1988), Nickson (1992), Stewart and that guidelines for the development and use of em-
Forsyth (1995), Hadjigeorgiou et al. (1995), Milne et pirical methods are urgently needed and hence the
al (1998), Suorineni (1998, 2010), Clark and Pa- significance of this conference.
kalnis (1997) and Trueman and Mawdesley (2003).
Further details of the evolution of the stability graph
can be found in Suorineni (2010).
Admittedly, some criticisms of the open stope
stability graph are due to some users not understand-
ing the assumptions of behind the development of
the method and the foundation databases. Suorineni
(2011) focused on criticisms of the stability graph by
1000 2.8 The hard rock pillar design chart
Stable
Stability Number (N)

The hard rock pillar design chart (Figure 10) was


10 developed by Lunder and Pakalnis (1997) following
Failure-Major Failure
the work Hudyma (1988). Figure 10 is a simple pil-
lar design chart commonly used in hard rock under-
0.1 ground mines around the world.
Cave
0.7

Average Pillar Stress (Ps)/UCS


FS=1
0.001 0.6
0 5 10 15 20 25 30 35 40 45 50 55 60
0.5
Hydraulic Radius (m) FS=1.4
0.4
Open stope Cut-and-Fill Longwall
Narrow-vein Sublevel Caving Block Caving 0.3
Figure 8. The Extended Mathews Stability Graph database plot-
0.2
ted by mining method.
0.1

0
2.7 The equivalent linear overbreak slough stability 0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2
graph Pillar Width to Height Ratio (W/H)

The stability graph is qualitative. All that it tells a Failed Unstable Stable
miner is that a stope of a given size is going to be Figure 10. Hard rock pillar design chart (Lunder and Pakalnis
probably stable, unstable or cave. However, miners 1997).
are interested in how much dilution a certain stope
size is going to produce and the relevant question is The author notes that the curves assigned factors
“will this dilution be acceptable?” Acceptable dilu- of safety of 1 and 1.4 in Figure 10 are simply lines
tion is mining method, mineral and or operation de- separating failed and unstable and unstable and sta-
pendent. ble pillars with no known degree of their conserva-
The Equivalent Linear Overbreak Slough (ELOS) tism. The The two curves in Figure 10 have been
stability graph (Figure 9) (Clark and Pakalnis 1997) misinterpreted by Kaiser et al (2011) and Martin
was developed to answer the key question miners and Maybee (2000) who argue that empirical pillar
keep asking: “Will my selected stope size give me strength equations such as those in Figure 10 tend to
acceptable dilution?” a horizontal asymptote and are wrong. They argue
that pillar strength equations should instead ap-
1000 proach a vertical asymptote after a pillar width to
height ratio of about 2.5. This argument is based on
Modified stability number N'

the results of numerical modelling which show fac-


100
0.5m tors of safety curves approaching a vertical asymp-
1m
2m tote at this pillar width to height ratio. Factors of
10 safety for pillars are different from pillar strength
and the two should not be compared.
Invariably, the arguments that pillars with width
1 to height ratios of about greater than 2.5 are inde-
structible, are based on comparison of the pillar fac-
tor of safety curves with pillar strength curves (ap-
0.1
0 5 10 15 20 25 ples and oranges).
Hydraulic radius HR (m) Suorineni (2013) argue that intuitively, pillar
Figure 9. The ELOS stability graph (Lunder and Pakalnis strength, pillar factor of safety and pillar stress
1997). should be related as in Figure 11.

The argument often advanced by the author is that


both the original and ELOS stability graphs are not
orebody size independent. An ELOS of 0.5 m in a 1
m wide orebody gives a dilution of 50% that will not
be acceptable in most mining operations while same
ELOS in a 10 m wide orebody will only result in 5%
dilution! This example demonstrates the need for the
development of a quantitative, dilution-based ore-
body width independent stability graph.
Today, nobody is making an effort to develop a
failure criterion for rocks that is based on fundamen-
tal rock behavior, geology and physics. A successful
and reliable rock failure criterion can only be devel-
oped through genuine multidisciplinary collaborative
research in geoscience and engineering.

Figure 11. Conceptual relationship between pillar stress, pillar


strength, Factor of safety and pillar width to height ratio.

In Figure 11 pillar stress decreases with increas-


ing pillar width to height ratio and to be constant af-
ter a width to height ratio of about 2 to 2.5 while pil-
lar strength increases to approach a horizontal
asymptote. Based on the trends of the pillar stress
and strength curves the pillar factor of safety should
increase with increasing pillar width to height ratio
to a vertical asymptote at about a pillar width to
height ratio between 2 and 2.5. Ironically, the pillar
strength-width to height ratio curve in Figure 11 is
identical to the Lunder and Pakalnis factor of safety
curves in Figure 10. Hence, the curves labeled factor
of safety curves of 1 and 1.4 are more likely pillar Figure 13. Material assumptions and predictions versus reali-
strength curves than factor of safety curves. ty/observed performance (Redrawn from Read and Martin
1996).

2.9 The Hoek-Brown failure criterion


The Hoek-Brown failure criterion is discussed ex- 3 CONCLUSIONS AND RECOMMENDATIONS
tensively in the literature. Hoek and Marinos (2007) 3.1 Conclusions
provide a detailed summary of the evolution of the
Hoek-Brown failure criterion. The validity of the cri- This paper is a reflection on the state-of –the-art on
terion has been strongly questioned by Hoek himself empirical methods in geomechanics. The state-of-
when he stated as follows: “Our approach (refering the-art clearly shows our predecessors were innova-
to the Hoek-Brown failure criterion) was entirely tive in developing solutions to the questions of their
empirical and we worked from very limited data of time and deserve credit for their efforts. These solu-
rather poor quality. Our empirical criterion and our tions recognized the complexity of the rockmass
estimates of the input parameters were offered as a when compared to steel, soil or concrete.
temporary solution to an urgent problem.” Hoek There should be the recognition today that the so-
(1994). He went further to state: “The fact that the lutions developed by our predecessors at their time
criterion works, more by good fortune than because have reached their limits and we and subsequent
of its inherent scientific merits, is no excuse for the generations should face the challenge and develop
current lack of effort or even apparent desire to find solutions to the questions of our time.
a better way.” Our predecessors were bold to acknowledge the
The statements by Hoek should dispel any one limitations of their solutions. We have ignored such
from criticizing the Hoek-Brown failure criterion but honest views and criticized their solutions while of-
rather channel our energies towards finding the “bet- fering no alternatives of our own. This is the crux of
ter way” referred to by Hoek. Hence, it is not sur- the state-of-the-art of empirical methods.
prising when we produce predictive numerical mod-
elling results such as those shown in Figure 12. In 3.2 Recommendations
Figure 12 all the predictions following the assumed
material behaviours would have resulted in the It is strongly recommended that the use of empirical
placement of the support in the wrong areas in the methods be guided to ensure proper use and con-
tunnel and consequently creating fatal conditions for sistency of the databases. It is also recommended
any workers. that more effort be spent on the documentation of
field observations instead of spending more time on
computers to produce results that have no relevance Laubscher, D.H. 1994. Cave mining-the state of the art. Jour-
to actual experience and putting people to risks. nal of The Southern African Institute of Mining and Metal-
lurgy: 279-293.
Genuine multidisciplinary collaborative research Løset, F. 1999. Use of the Q-system in weak rock masses. NGI
involving geoscientists, engineers and technology Report No. 592048-1.
developers is required to solve the geomechanics Lunder, J. & Pakalnis, R. 1997. Determining the strength of
challenges of today. hard rock mine pillars. Bull Can Min Metall., 90(1013): 51-
55.
Marinos, P.G., Marinos, V. and Hoek, E. 2007. The geological
strength index (GSI): Characterization tool for assessing
REFERENCES engineering properties for rock masses. In Mark, Pakalnis &
Tuchman (editors), Proceedings International Workshop on
Barton, N. & Bienawski, Z.T. 2008. RMR anS Q-setting rec- Rock Mass Classification for Underground Mining, Infor-
ords. Tunnels and Tunnelling International: 26-29. mation Circular 9498. Pittsburgh: NIOSH.
Barton, N., Lien, R. & Lunde, J. 1974. Engineering classifica- Martin, C. D. & Maybee, W. G. 2000. The strength of hard-
tion of rock masses for the design of rock support. Rock rock pillars. International Journal of Rock Mechanics and
Mechanics, 6(4):189–236. Mining Sciences, 7(8): 1239-1246.
Bieniawski, Z.T. 1973. Engineering classification of jointed Mathews, K.E., Hoek, E., Wyllie, D.C. & Stewart, S.B.V.
rock masses. Trans. S. Afr. Inst. Civ. Engrs., 15: 335–344. 1981. Prediction of stable excavation spans at depths below
Bieniawski, Z.T. 1989. Engineering rock mass classifications. 1000m in hard rock mines. CANMET Report, DSS Serial
New York, Wiley. No. OSQ80-00081. 127 pp.
Clark, L.M. & Pakalnis, R.C. 1997. An empirical approach for Milne, D., Hadjigeorgiou, J. & Pakalnis, R. 1998. Rock mass
estimating unplanned dilution from open stope hangingwalls characterization for underground hard rock mines. Tunnel-
and footwalls. Proceedings of 99th Annual General Meet- ling and Underground Space Technology, 13(4): 393-391.
ing, Vancouver: CIM. Milne, D., Pakalnis, R.C. & Felderer, M. 1996. Surface geome-
Clark, L.M. & Pakalnis, R.C. An empirical approach for esti- try assessment for open stope design. Proceedings of North
mating unplanned dilution from open stope hangingwalls America Rock Mechanics Symposium, Kingston: CIM
and footwalls. Proceedings of 99th Annual General Meet- Mitri, H.S., Hughes, R. & Zhang, Y. 2010. New rock stress fac-
ing, 27th April – 1st May 1997. Vancouver: CIM. tor for the stability graph method. International Journal of
Coates, D.F. 1981. Rock mechanics principles. Ottawa: Mines Rock Mechanics & Mining Sciences,
Branch Monograph 874, Information Canada. doi:10.1016/j.ijrmms.2010.09.015.
Diederichs, M.S. & Kaiser, P.K. 1999. Tensile strength and Nickson, S.D. 1992. Cable support guidelines for underground
abutment relaxation as failure control mechanisms in under- hard rock mine operations, M.A.Sc. thesis, University of
ground excavations. Int. J. Rock Mech. Min. Sci. 36: 69–96. British Columbia.
Franklin, J. A. 1993. Empirical design and rock mass character- Nickson, S.D. 1992. Cable support guidelines for underground
isation. In Hudson J.A. (ed), Comprehensive rock engineer- hard rock mine operations. M.Sc. thesis, Vancouver: Univ
ing, 2: 759–806. New York: Pergamon, British Columbia.
Hadjigeorgiou, J., Leclaire,J. & Potvin, Y. 1995. An update of Potvin, Y. 1988. Empirical open stope design in Canada, Ph.D.
the stability graph method of open stope design. Proceed- thesis, University of British Columbia, 1988.
ings of 97th Annual General Meeting. Halifax: CIM.. Read, R.S. & Martin, C.D. 1996. Technical summary of
Hoek E. 1994. Strength of rock and rock masses. News J ISRM AECL’s Mine-by Experiment. Phase1 excavation response.
2(2):4–16. Atomic Energy of Canada Limited Report, AECL-11311.
Hoek, E. & Brown, E.T. 1980. Underground excavation in Stewart, S.B.V. & Forsyth, W.W. 1995. The Mathews method
rock. London, IMM. for open stope design. CIM Bull. 88: 45–53.
Hoek, E. & Marinos, P. 2007. A brief history of the develop- Suorineni, F.T. 2010. The stability graph after three decades in
ment of the Hoek-Brown failure criterion. Soils and Rocks, use – Experiences and the way forward. International Jour-
2: 2-13. nal of Mining, Reclamation and Environment, 24(4): 307-
Hoek, E. 1999. Putting Numbers to Geology – an Engineer’s 3392.
Viewpoint. The Second Glossop Lecture, Quarterly Journal Suorineni, F.T. 2011. Factors influencing overbreak in the
of Engineering Geology, 32(1): 1-19. Barkers orebody, Kundana Gold mine: narrow vein case
Hoek, E., Carranza Torres, C.T. & Corkum, B. Hoek-Brown study. Paper by: P.C. Stewart, R. Trueman and I. Brunton,
failure criterion: 2002 edition. In Proc. of the 5th North 2011: Mining Technology, vol. 120(2), pp. 80-89. Letter to
American Rock Mech. Symp. 7-10 July 2002. Toron- the Editor, Transactions, Institution of Mining and Metal-
to:NARMS. lurgy (Section A): Mining Technology.
Isaacson, W. 2007. Einstein: His life and universe. New York, Suorineni, F.T. 2012. A critical review of the stability graph
Simon and Schuster. method for open stope design. MassMin 2012, 10-14 June
Kaiser, P.K., Kim, B., Bewick, R.P. & Valley, B. (2011). Rock 2012. Sudbury, CIM.
mass strength at depth and implications for pillar design. Suorineni, F.T. 2013a. Geomechanics challenges and its future
Mining Technology,120(3): 170 – 179. direction – Food for thought. Ghana Mining Journal, 14:
Laubscher D. H., 2001, Cave Mining – The State of the Art, 14-20.
Underground Mining Methods. In Hustrulid W.A., Bullock Suorineni, F.T. 2013b. The Geomechanics Challenges of Con-
R.L. (eds), Underground Mining Methods: Engineering temporary Deep Mining: Technology as the Pathway to In-
Fundamentals and International Case Studies: 455-464. creased Safety and Productivity. 16th Kenneth Finlay Lec-
Littleton, CO: Society for Mining, Metallurgy, and Explora- ture, 17th October. Sydney: UNSW Australia.
tion, Inc. Suorineni, F.T., Kaiser, P.K. & Henning, J.G. 2008. Safe rapid
Laubscher, D.H. & Taylor, H.W. The importance of geome- drifting – Support selection. Tunnelling and Underground
chanics of jointed rock masses in mining operations. Pro- Space Technology, 23: 682–699.
ceedings of the Symposium on Exploration for Rock Engi- Trueman, R. & C. Mawdesley, C. 2005. Predicting cave initia-
neering, 1st – 5th November 1976, Cape Town: SAIMM.. tion and propagation. CIM Bull. 96: 54–59.
Trueman, R. & Mawdesley, C. 2003. Predicting cave initiation
and propagation. CIM Bull., 96:54–59.

View publication stats

You might also like