You are on page 1of 7

The modified stability graph method; more than 30 years later

Y. Potvin
Australian Centre for Geomechanics, The University of Western Australia, Perth, Australia

ABSTRACT: The modified stability graph method is the prime tool to design open stopes in underground
hard rock mines. This Keynote provides some historical narrative on the early development of the method.
After more than 30 years of the method being applied by a large number of practitioners, some observations
based on the authors experience are given on common challenges, mistakes and misuse of the method.
1 THE ORIGINAL STABILITY GRAPH
METHOD
In 1981, the Vancouver office of Golder and Associates were given a contract from the Canadian Government agency Canmet to develop an approach to
design open stopes. As a result of this work,
Mathews et al. (1981) proposed the stability graph
method, based on an extension of the Barton et al.
(1974) NGI rock mass classification system. The
reader is referred to Potvin (1988), Hoek et al.
(1995), and the abundant literature on this subject
for a detailed description of the method.

Figure 1. The original stability graph after Mathews et al.


(1981)

The original stability graph (Figure 1) relates individual stope surface dimensions expressed as hy-

draulic radius (or shape factor) in the X-axis, to a


rock stability index called the stability number N,
plotted on a log scale in the Y-axis. Based on
26 case studies from 3 mines complemented with
29 case studies from literature, three zones were defined on the graph; a stable zone (top left), a potentially caving zone (bottom right), and a potentially
unstable zone (in the middle).
The method was documented in a Canmet report
produced by Golder and Associates (Mathews et al.
1981), but was not found in the public domain at the
time. It is in 1983, during a site visit to Norandas
Heath Steel Mine, one of the three mines contributing the 26 case studies, that I found in one of the
mines filing cabinet, the Canmet report describing
the stability graph method. The method appeared to
have a lot of potential, being easy to use and logically assessing the effect of each of the main parameters influencing stope stability. However, the method
was unknown and unproven. The small database
used for its development did not allow for a clear
definition of the stability zones. In practice, this
made the graph difficult to use for designing stopes.
For example, consider the case of a stope hanging
wall where the stability number N is 10. A range of
options available to the designer are illustrated in
Figure 2. It could involve designing conservatively
for the limit of the stable zone (point 1) or to have
an aggressive design going in the potentially unstable zone (point 2). If the stope height has been set to
be 20m due to the sub-level interval and the designer
aims at establishing a feasible strike length based on
the design point 1, which has a hydraulic radius of 3,
it would result in a stope length of 8.5m. If the designer decides that an aggressive design is more appropriate, point 2 suggests a hydraulic radius of 9,
and this would correspond to a stope length of

180m. This example, where the design range varies


from a conservative 8.5 m to an aggressive 180 m
long stope, demonstrates that the guidelines provided by this graph are simply too broad for usage in
practical stope design applications.
For a 20 m high stope
HR = 3 =>

L = 8.5 m

HR = 9 =>

L = 180 m

Point 1

tion). Therefore, caving in the modified stability


graph is not associated with block or sublevel caving
behaviour, which generally involves continuous unravelling of the caving surface. The scarcity of
points with hydraulic radius greater than 12 in Figure 3 is also noted. This will result in some uncertainties for stability predictions in this part of the
graph. In fact, work from Mawdesley (2002) suggests that the transition zone becomes flatter with
larger hydraulic radii.

Point 2

Figure 2. Example illustrating that the use of the original stability graph for design provides guidelines that are too broad to be
practical

2 THE MODIFIED STABILITY GRAPH (POTVIN


1988; NICKSON 1992)
As part of my PhD study beginning in 1985, I undertook, with the assistance of Marty Hudyma, to populate the stability graph with more case studies, which
lead to the modified stability graph method. It is
noted that the modifications proposed to the method
within my doctoral research were overviewed by the
lead original developer of the method Mr Ken
Mathews and externally examined by another author
of the original method, Dr Evert Hoek.
A further 176 case studies from 32 mines were
added to the original database. Some minor modification for Factors A, B and C were proposed, but the
major contribution from this work came from refining the stability zones on the graph (Figure 3, after
Potvin 1988).
It was the first refinement of the original stability
graph method and the transition zone between stable
and caving had been reduced significantly, to a
point where it became a very useful stope design
tool. However, the following clarifications are useful. The choice of the word caving in this graph was
somewhat unfortunate. In the Canadian open stope
mining context of the mid-eighties, the terminology
caving of a stope surface was commonly used and
meant that the stope would suffer very significant
stability and dilution problems (say over 30% dilu-

Figure 3. The modified stability graph with the 176 new case
studies populating the new graph and refining the stability
zones. The round shape points represent stable cases, the triangular are cases with extensive stability problems and the square
points represent cases with some stability problems (Potvin
1988)

As part of the modified approach to the method, I


also proposed a zone where the stope surface can be
stabilised with cable bolts. Based on 66 case studies
of cable bolted stope backs and hanging walls, the
stable with support zone was delineated, in between the dashed line and the transition zone in Figure 4. A cable bolting density graph was also produced (Figure 5) from this database, relating the
block size of the rock mass relative to the stope surface against to the density of the bolting pattern. The
simple principle applied is that the smaller the rock
blocks in larger surfaces will require more cable
bolts compared to larger blocks in smaller surfaces.
Nickson (1992) further refined the cable bolt stability zone by adding 46 supported case studies (Figure 6). This version of the graph which combines the
transition zone that I proposed in 1988 and Nicksons cable bolt support zone has been re-published
in widely used references such as Support of Underground Excavations in Hard Rock (Hoek et al. 1995)
and
Cablebolting in Underground Mines
(Hutchinson & Diederichs 1996). To my know-

ledge, this graph has become a de-facto industry


standard.
There have been many other proposed modifications to the method and to the graph itself since the
modified Stability Graph was first published in
1988. The review of all these proposals is beyond
the scope of this keynote but it is my observation
that very few of these modifications have been widely accepted or used by the industry. Most of them
are either unproven or site specific and they have
added more confusion than value.
Nevertheless, a few modifications such as the radius factor (Milne et al. 1996) and the Equivalent
Linear Overbreak Sloughing; ELOS (Clark and Pakalnis, 1997, Capes 2009) are very useful and extend
the method capabilities and have been adopted by a
number of mine sites.
Figure 6. The stability graph according to Nickson 1992, showing a refined zone stable with support

3 APPLICATION OF THE MODIFIED


STABILITY GRAPH METHOD

Figure 4. The Modified Stability Graph showing the stable


with cable bolt support zone between the transition and the
dashed line

Figure 5. The cable bolt density graph relating the relative block
size of the rock mass to the density of the cable bolt pattern

The modified stability graph method can be applied


at different stages of a mining project with different
objectives. It remains a design method that provides
a ball-park estimate and as such it has limited accuracy, but it is a very powerful and robust technique to apply at the feasibility study stage. Mine
designers at the feasibility stage aim at developing
systematic stope layouts that will be stable. In green
field feasibility studies in particular, where there is
no opportunity to calibrate numerical models, the
empirical modified stability graph method has a
clear advantage of being intrinsically calibrated
through the thousands of case studies in which it has
been applied during the last 30 years. It allows mine
designers to assess with some confidence many of
the critical mine design parameters such as sub-level
interval, stope dimension and the need for cable bolt
support stabilisation.
The method is also often being applied again
whilst the mine is operating, at the individual stope
planning stage, when more data and more
knowledge on the rock mass behaviour exist. The
mine planners at that stage can benefit from the
knowledge gained from mined out stope performance.
The empirical method is best used in conjunction
with numerical modelling at the stope planning
stage, because by then, there will have been some
opportunities to calibrate a numerical model. When
the mine is in operation, most of the infrastructure is
already in place which limits the flexibility to implement changes in the stope design. However, the
objective is then to verify the original design and

look for opportunities to optimise it. These investigations may lead to slight changes in stope dimensions, or to the development of control measures to
reduce the risk of stope instability, including cable
bolt support, changes in stope blasting methods and
extraction strategies to better control dilution.
4 GATHERING INPUT DATA
As with any design method, the reliability of the
output can only be as good as the quality of the input
data. Although the method owes its popularity to its
simplicity, there are many ways to make mistakes in
assessing the data required to apply the modified
stability graph. In fact, the characterisation of the
rock mass at mine sites is often left with the most
junior geotechnical engineers or with geologists having limited or no geotechnical training. This in itself
can lead to major sources of inaccuracy in the design.
When discussing rock mass data gathering, it is
important to make a distinction between rock mass
characterisation and rock mass classification. Potvin
et al. (2012) make this distinction as follows: Rock
mass characterisation should be generic in nature,
capturing the basic input parameters that can be used
in classification systems and empirical design methods. Rock mass characterisation is the background
field work required to perform rock mass classification. It should concentrate on measurements and information about the intact rock strength, the intensity of natural fractures in the rock mass and the
conditions of these fractures. It should, however, not
be limited to obtaining parameters used in the classification systems. The characterisation should also
aim to provide a context or framework for further
design decisions and the use of rock mass classification systems.
Rock mass classification is the subsequent step to
the characterisation. Parameters that vary according
to the design, such as the relative orientation of geological structures compared to the opening or the
pillar, the induced stresses, the ground water, etc.,
should be calculated as part of the rock mass classification and design process, rather than during the
rock mass characterisation process.
If one considers a volume of rock mass, this volume will have a unique rock mass characterisation,
but at the same time, it can have multiple rock mass
classification values depending on what is being designed, the orientation of the designed structures
(slope, stope, drive, pillar, etc.) and the scale of these features. If the design or scale changes, the classification value can change despite the fact that the
rock mass considered is exactly the same.
The rock mass characterisation data can be assessed from core and from underground mapping
when access is available. Underground mapping will

generally produce better quality data, as the exposure to the rock mass is significantly improved compared to core. However, assuming that oriented core
is available, all the parameters to perform a modified stability graph analysis can be assessed from
core, except for the large scale roughness (planar
versus undulating). The impact of large scale roughness on the Q value is double when comparing a
planar versus an undulating joint. This is generally
an acceptable range of Q values to carry throughout
the analysis, at the feasibility study stage.
5 COMMON MISTAKES IN APPLYING THE
METHOD
As mentioned before, the most common and important source of inaccuracy in applying the modified stability graph method is likely made while using the rock mass classification system. More
specifically, the joint roughness (Jr) and joint alteration (Ja) parameters must be assessed on the joint set
that potentially will most critically affect the stability of the stope surface considered for design. This
critical joint is the same joint that will be selected
for Factor B assessment. In the case of stope back, it
will generally be a relatively flat joint and for stope
wall it will be a sub vertical and sub parallel joint set
to the designed wall. Therefore, for the same location, the Q (wall) can be drastically different than
the Q (back). In fact, unless the rock mass is isotropic and all joint sets have the same Jr and Ja, a
different Q must be used for stope back and walls.
In practice, it is rather common that the rock
mass characterisation step is skipped and classification is applied directly to areas of interest of the
mine. Q values are assigned to rock mass on a domain basis, without considerations to the stopes or
other mine structure being designed. Often, the Jr
and Ja values are assessed on the most prominent
joint set or alternatively, to be conservative, it is assigned to the joint set with the lower Jr and Ja values. This could lead to very significant mistakes.
Amongst other sources of mistakes, the graph that
I proposed to assess Factor B (Figure 7, Potvin 1988)
can also lead to confusion in certain hanging wall
situations. This is because the graph only covers differences in strike from 0 to 90 degrees. It is important to visualise in three dimensions whether a
critical joint can cause stability problems in a hanging wall as a reality check rather than blindly subtract strike and dip numbers and read a Factor B on
the chart. A critical joint with strong influence on
stability should have a Factor B approaching 0.2 and
a joint with little influence on stability (toppling situation, or joints sub-perpendicular to the stope surface) should have a Factor B approaching 1.0.

Figure 7. Determination of the Orientation Factor, after Potvin


(1988)

Another mistake that I have observed occasionally is in the application of Factor C, for the case of
sliding (Figure 8) as proposed by Hadjigeorgiou et
al. (1995). This factor was developed to address the
very specific situation where a joint oriented parallel, but with a shallower dip than the footwall of a
stope created a potential for sliding instability. This
graph should not be applied to vertical walls. In the
original database used for developing the modified
stability graph (Figure 3, 4 and 5), all vertical walls
had a Factor C of 8, in accordance to the gravity
Factor C graph (Figure 9), no matter whether there
was potential for sliding or not. Applying Figure 7
for vertical walls and normal footwall conditions
over-penalises the calculation of the stability number, as Factor B already accounts for unfavourable
critical structures.

Figure 9. Graph developed to assess the gravity Factor C for


cases other than footwall experiencing sliding issues

Another relatively common practice observed in


the industry that deserves a word of caution is the
use of designed spreadsheets where the raw input
data are entered into cells and points are automatically generated on a modified stability graph by the
spreadsheet. Fundamentally, there is nothing wrong
with this approach and it can be quite useful for sensitivity analyses. The potential problem is that it bypasses the thinking process and one of the great benefits of applying the method manually, which is to
facilitate the scrutiny of each factor potentially affecting the stability of the stope and provide an understanding of the expected behaviour of the stope.
As such, the reality check can be totally overlooked with the spreadsheet approach.
6 SOME CHALLENGES IN APPLYING THE
METHOD

Figure 8. Graph specifically developed to assess Factor C for


footwall having sliding issues

Applying the modified stability graphs in some specific situations where the design lies outside the generic open stope mining situations can be challenging. For example, the concept of hydraulic radius
applies to straight surface, as opposed to convex or
concave, which admittedly is somewhat unusual but
possible for open stopes. It can also be challenging
to assign a hydraulic radius to complex polygon
shapes (instead of traditional rectangular and square
shapes) or stope surface involving a brow. The radius factor (Milne et al. 1996) can be a great assistance to overcome some of these problems. Otherwise, good engineering judgement must be used to
interpret the analysis of a complex surface simplified into a rectangle. For example, fitting an approximately equivalent rectangle surface over a polygon
can be acceptable, if the fit is reasonable, given the
intrinsic inaccuracy (ball park estimation of stope
dimension) of the empirical method.

Assessing the induced stress for determining Factor A is also not always straight forward and sometimes requires some degree of interpretation and
judgement. For example, Figure 10 shows a Map 3D
modelling output of the induced stress around an
open stope layout. Looking at the stope bottom
(point C), the induced stress acting on this stope
back is around 40 MPa (green colour). However, the
stope wall (point B) is not so evident. There is clearly high stress from the top corner (red colour
90 MPa) and lower stress towards the bottom (yellow 65 MPa). In such a case, one would be advised
to use the level of stress acting towards the middle
of the designed stope surface (yellow/maroon interface 70 MPa), away from the edge effect of the
stope surface.

Figure 10. Numerical modelling output showing the induced


stress around a stoping layout

Using the modified stability graph for very large


stopes can also be problematic since the modified
stability graph (Figures 3-5) cannot be used with
confidence in cases where the hydraulic radius exceeds a value around 12, as mentioned in section 2.
For cases where hydraulic radii exceed 12 up to
about 20, the readers can double check their design
using the stability graph Figure 11, after Mawdesly
2002). This is a log-log graph and the transition zone
is shown as a straight line.

7 CALIBRATING THE STABILITY GRAPH


METHOD
It is very common in the industry to calibrate the
modified stability graph for a specific mine. The
mine site calibration generally involves the back
analysis of several case studies, plotting them on the
modified stability graph, and shifting the transition
zone up, down, right or left, aiming at finding a better fit for the local data.
The need for site calibration, in some ways, contradicts the following statement I made in section 3 of
this keynote;
the empirical modified stability graph method has a clear advantage of being intrinsically calibrated through the thousands of case studies in
which it has been applied during the last 30 years.
It is my strong view that the need for calibration
is not because the generic modified stability graph
needs to be adapted for local site conditions, as the
database behind the graph is extensive and covers a
very wide variety of ground conditions, likely encompassing most local conditions. It is rather the users that need to calibrate themselves to the generic
graph. Anyone that has performed rock mass characterisation, rock mass classification, structural mapping, and numerical modelling will know that there
is a significant amount of interpretation involved in
obtaining representative data from these techniques.
Furthermore, as discussed in sections 4 and 5 of this
paper, there are many common mistakes made when
gathering data and applying the modified stability
graph method. If the interpretation during gathering
and/or mistakes in applying the method is systematically done by a user, the result is likely to be that all
back analysis points on the graph will systematically
plot either over- or under-conservative compared to
the generic transition zone. The calibration of the
users will then cause the transition curve to move
toward the top left (more conservative) or bottom
right (less conservative).
8 SUMMARY

Figure 11. Log-log version of the stability graph (after


Mawdesley 2002) featuring many cases with hydraulic radii
(shape factor) between 12 and 20

The modified stability graph, like a number of empirical methods in geomechanics, has met with wide
mining industry acceptance and extensive worldwide utilisation at feasibility study and stope planning stages. The modified stability graph provides a
ball-park estimate of stope dimension and as such
it has limited accuracy but it is a very powerful and
robust technique to plan stope layout.
Many authors have proposed subsequent modifications to the method but very few have been widely
adopted in practice. Amongst those proposals, the
ELOS and the radius factor concepts are perhaps the
most useful extension of the method.

Although the modified stability graph method is


simple to apply, there are several potential sources
of inaccuracies and mistakes associated with the application of the method. In particular, the classification of the rock mass is often conducted without
identifying the critical joint leading to error in estimating Jr and Ja. The estimation of Factors A, B and
C also have challenges and are the source of common mistakes from inexperienced users.
The need for calibrating the method for local
mine site conditions is often misconstrued with the
need to calibrate how the users classify rock masses
and apply factors A, B and C in the method. In general, if a significant adjustment of the transition zone
is required, it is likely due to a systematic error in
the way the method is applied, including rock mass
classification assessment.
REFERENCES
Barton, N., Lien, R. & Lunde, J. 1974. Engineering
classification of rock masses for the design of tunnel
support. Rock Mechanics 6(4): 189-236.
Capes, G.W. 2009. Open stope hangingwall design based on
general and detailed data collection in rock masses with
unfavourable hangingwall conditions, Ph.D. thesis,
University of Saskatchewan, 248 p.
Clark, L.M., & Pakalnis, R.C. 1997. An empirical design
approach for estimating unplanned dilution from open stope
hangingwalls and footwalls. 99th CIM-AGM, Vancouver,
CD-Rom.
Hoek, E., Kaiser, P.K. & Bawden, W.F. 1995. Support of
Underground Excavations in Hard Rock. Rotterdam:
A.A. Balkema, 215 p.
Hutchinson, D.J. & Diederichs, M. 1996. Cable bolting in
Underground Mines. BiTech Publishers, Richmond, 406 p.
Mathews K.E., Hoek, E., Wyllie, D.C. & Stewart, S.B.V. 1981.
Prediction of stable excavation spans for mining at depths
below 1000 m in hard rock mines. Canmet Report DSS
Serial No. OSQ80-00081.
Mawdesley, C.A. 2002. Predicting rock mass cavability in block
caving mines. Ph.D. thesis, University of Queensland, 410 p.
Milne, D., Pakalnis, R.C. & Felderer, M. 1996. Surface
geometry assessment for open stope design. In Proc. North
American Rock Mechanics Symposium, Montreal, Balkema,
pp. 315-322.
Nickson, S.D. 1992. Cable support guidelines for underground
hard rock mine operations. M.A.Sc. thesis, The University
of British Columbia., 223 p.
Potvin, Y. 1988. Empirical open stope design in Canada. Ph.D.
thesis. The University of British Columbia, 350 p.
Potvin, Y., Dight, P. & Wesseloo, J. 2012. Some pitfalls and
misuse of rock mass classification systems for mine design,
Journal of the Southern African Institute of Mining and
Metallurgy 112(8). August 2012.

You might also like