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

net/publication/252527811

Ground Support in Mining and Underground Construction

Article

CITATIONS READS

7 941

4 authors, including:

Ernesto Villaescusa Yves Potvin

WA School of Mines - Curtin University Australia University of Western Australia
141 PUBLICATIONS 752 CITATIONS 70 PUBLICATIONS 761 CITATIONS

SEE PROFILE SEE PROFILE

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

Stope reconciliation View project

Early strength of shotcrete View project

All content following this page was uploaded by Ernesto Villaescusa on 03 July 2014.

The user has requested enhancement of the downloaded file.

 Ground Support in Mining and
Underground Construction
Proceedings of the Fifth International Symposium on Ground Support
28-30 September 2004, Perth, Western Australia
Edited by
Ernesto Villaescusa
Yves Potvin

Rock reinforcement design for overstressed rock using three dimensional numerical modeling

T. Wiles, E. Villaescusa & C. R. Windsor

p. 483-489
Rock reinforcement design for overstressed rock using three dimensional
numerical modeling
T. Wiles
Mine Modelling Pty Ltd, P.O. Box 637, Mt Eliza, Victoria 3930, Australia
E. Villaescusa & C. R. Windsor
Western Australian School of Mines, Curtin University of Technology, PMB 22, Kalgoorlie 6430, Western
Australia

ABSTRACT: A procedure is presented for the design of reinforcement for highly stressed rock based on 3D
numerical stress analysis using the MAP3D code. Modelling requires extensive characterisation of the rock
mass strength and deformability and appropriate characterisation of the stress field. The numerical model is
calibrated using a Rock Mass Damage Criterion and a Rock Mass Failure Criterion that are calibrated to ob-
servations of in situ cracking. These criteria define an outer damaged or cracked zone and an inner, failed or
broken zone. Examples are used to show how the extent and dimensions of these zones can be determined by
post-processing the modelling results and how the boundary of these zones can be used to dimension the pri-
mary reinforcement scheme.

1 INTRODUCTION derground excavations. In modern engineering mine

design, the rock mass strength is usually estimated
The effectiveness of a particular reinforcement prior to excavation using borehole data (i.e. as part
scheme in terms of its density and length can be as- of the orebody delineation process, when full core is
sessed using empirical strategies, theoretical meth- available) to determine the variability of the intact
ods and geotechnical instrumentation (Windsor and rock properties and the rock mass classification pa-
Thompson, 1993). In addition, detailed stability rameters throughout the orebody. The intact rock pa-
analysis may now also include the calibration of a rameters (uniaxial compressive strength and the
numerical model program to simulate the excavation elastic constants E and ) can be determined from a
sequences in order to assess the role of stress change number of representative holes taken from the full
on the rock mass environment. Three dimensional set of exploration holes (Figure 1) thus allowing the
numerical modelling results can then be used to de- characterization of the entire orebody.
termine the overall stability around underground ex-
cavations, where zones of damage, or failure can be
estimated and then used to determine the required
length and capacity of a reinforcement scheme. The
required input parameters consist of the in-situ stress
profile with depth, the strength and deformational
properties of the rock mass as well as the excavation
steps and their sequence.

2 ROCK MASS STRENGTH &

DEFORMABILITY

The rock mass compressive strength is a measure of

the peak load carrying capacity of a rock mass. It is
defined as a proportion of the intact rock strength
due to the presence of geological discontinuities
(Hoek and Brown, 1980). The rock mass strength is
defined here as the limiting load required for stress Figure 1. Typical distribution of exploration holes in
driven failures to initiate and propagate around un- an orebody.
 3 IN SITU STRESS
Usually, the uniaxial compressive strength and its
variation for each rock type present can be defined Reliable evaluation of in situ stress is an important
such as in Table 1. Alternatively it may be pre- phase in the analysis and design of underground ex-
sented using modelled contours across a particular cavations, particularly when evaluating excavation
unit such as the strength distribution in the hanging- stability with the aim of preventing stope/pillar wall
wall boundary as shown in Figure 2. failures. Consequently, over the last seventy years
considerable effort has been invested by numerous
research organizations in finding suitable methods to
Table 1. Average Uniaxial Compressive Strength per quantify Earth s crustal stresses. In cases where ex-
rock type. cavation access is available, the overcoring method
Rock Sample UCS STDev using the CSIRO HI cell has proven to be an accu-
Type Number rate and reliable method of measuring the complete
(MPa) (MPa) 3D stress tensor. In addition, in the last few years the
Hangingwall Western Australian School of Mines (WASM) has
9 114 25.1
rock
developed a technique to determine the complete 3D
Orebody 11 107 42.2
Footwall stress tensor using the Acoustic Emission method
7 139 40.1 (Villaescusa et al, 2003). The practical advantages of
rock A
Footwall the WASM AE technique revolve around the fact
15 107 23.0
Rock B that the state of stress may be quantified for any
point where oriented exploration core can be ob-
tained. This negates the previous restriction for the
- 140 UCS
(MPa) existence of an excavation from which to conduct
2200 RL
- 130
the measurements.
- 120 The results shown in Figure 3 compare the stress
- 110 tensors obtained by the CSIRO HI cell and the
- 100
WASM AE method for the orebody shown in Fig-
ures 1 and 2. The comparison involves a two point
- 90 2000 RL
WASM AE stress profile defined over a 100m inter-
- 80
val and a single CSIRO HI overcoring result at a
- 70 shallower depth separated by 150m. The principal
- 60
stress magnitude-depth relationships and the princi-
8400 E

8600 E

9000 E
8800 E

- 50 1800 RL
pal stress orientations for the two point WASM AE
profile and the single CSIRO HI result are given to-
gether in Figure 3.
Figure 2. Modelled UCS variability across an ore-
The data shows that there is excellent agreement
body hangingwall boundary.
between the extrapolation of WASM AE magnitudes
of the principal stresses compared with those ob-
The actual value of rock mass strength and deform-
tained by the CSIRO HI cell overcoring and between
ability depends upon the geometrical nature and
the principal stress orientations indicated by over-
strength of the geological discontinuities, which can
coring at 363m depth compared to that obtained by
be estimated by using empirical methods that rely on
WASM AE at 493m depth. Comparison of the two
rock mass classifications. Table 2 presents some
WASM AE measurements at 493m and 595m indi-
typical average results for the same rock units de-
cate a small rotation that effectively flips the major
scribed before in Table1.
and intermediate principal directions to diametrically
opposite positions on the projection. The relative
Table 2. Average rock mass properties per rock type.
variation for each principal stress magnitude is
Rock mass Em cm m shown in parenthesis under each projection of
Type (GPa) (MPa) ( ) WASM AE principal orientations.
Hangigwall
32 46 40
Rock
Orebody 31 44 44
Footwall
32 57 40
Rock A
Footwall
31 44 44
Rock B
Where: E m is the deformation modulus, cm is the
uniaxial compressive strengths, and m is the fric-
tion angle of the rock mass.
 Principal Stresses (MPa)
0 20 40 60
-250
HI 1
HI 2
HI 3

-300
AE 1
AE 2
AE 3 HI at 363m and AE at 493m
-350 Principal stresses and planes

-400
Depth (m)

-450

AE principal orientations at 493m

-500 1(1.2%), 2 (1.7%), 3(2.9%)

-550

-600

AE principal orientations at 595m

-650 1(3.2%), 2 (1.4%), 3(6.6%)

Figure 3. Comparison of a three point AE stress measurement profile with a single point CSIRO HI Cell
stress measurement.

pirical failure criterion established for the different

domains around the underground excavations within
4 NUMERICAL MODELLING an orebody. It must be emphasized that any predic-
tive models must be calibrated (validated) against
field data and observations using either visual meth-
In most cases of underground mining, the induced ods and/or geotechnical instruments.
stresses may be determined using linear elastic nu- Although linear elastic modelling can be used to
merical modelling. The required inputs are the in- estimate the level of damage and the extent of the
situ stress field with depth, the estimated deforma- failure zones, non-linear models are required to
tional properties of the rock mass and the chosen ex- simulate any resultant stress re-distribution from
traction sequence. In this study the elastic version of such failures. Progressive orebody extraction may
the computer program Map3D was used to deter- induce several phases of post-peak behaviour in a
mine the stress distribution around the underground rock mass and similarly, small changes to the stress
excavations. It must be understood that the results field induced by distant extractions may induce sig-
are used in conjunction with structural information nificant rock mass damage around a particular exca-
(for example large fault behaviour) in order to inter- vation wall.
pret any excavation option analyzed as well as their
respective reinforcement strategies. Typical output
from numerical modelling includes stresses and dis-
placements. These can then be compared with em-
5 FAILURE CRITERION tegrate if it is not held together by a ground support
scheme.
Experience through correlation of underground ob-
servations and geotechnical instrumentation with
numerical modelling results suggest that a rock mass
is damaged when a range of induced stress levels
exceeds a certain site dependent threshold as shown
in Figure 4. Below the damage threshold the re-
sponse is elastic and usually very little damage can
be observed. However, with increased overstressing
increased damage is experienced. The actual damage
level reached depends upon the amount of over-
stressing and beyond the initial damage threshold a
zone of potential overbreak (POB) is reached. In-
creased stress beyond this level may cause stress
driven failures and eventually the rock mass may 140
become unsupportable. 120

3 (MPa)
100
Unsupportable
80 HW
driven failure FW
60

1-
POB 40
1 1
20

0
-5 0 5 10 15 20 25 30

Damage driven Unsupportable

3 (MPa)
POB
threshold failure

Figure 5. CMS profile of failure and damage crite-

3 - Confinement 1 rion from back analysis using Map3D.

Figure 4. Different levels of stress driven damage 1 1 =B+q 3

and failure. Stress driven
failure
Consequently, a rock mass is neither strictly q
failed nor unfailed, but rather for similar confine- 1 k mass
roc 1 =A+ 3
ments, there is a range of stress levels where increas- ag ed
ing excavation damage is experienced. A Rock Mass Dam
B
Damage Criterion can be defined as follows:
A
1= A+p 3 (1) Undamaged rock mass

where A and p are site dependent constants. Back

analysis of numerical modelling over a number of 3

years (Wiles 1998, 2004) suggests that p normally

takes on a value near unity. Figure 6. Rock mass damage zones
Figure 5 shows an empirical damage criterion es-
tablished from back analysis of cavity monitoring Another criterion that can be readily identified
system (CMS) surveys for a primary stope located at (upper line in Figure 6) is commonly called the Rock
a deep underground operation in Western Australia. Mass Strength Criterion and is defined as follows:
The criterion expressed by Equation 1 is also con-
ceptually represented by the lower line in Figure 6 1= B+q 3 (2)
and can be interpreted to represent the stress level where B is the Uniaxial Compressive Strength ( cm)
where seismicity is observed to occur. In addition, of the rock mass and q is related to the rock mass
borehole cameras can be used to directly observe the friction angle ( m) by tan (45+ m/2).
amount of damage in the form of increased fracture
frequency. Geotechnical instrumentation shows that cm 3 tan 2 (45 m 2)
when this stress level is exceeded a loss of rock Strength Factor A =
1
mass cohesion is experienced. However, a consider-
able degree of residual frictional strength (i.e. inter- This criterion represents a stress level at which
locking) is still available. Nevertheless, the rock failures can be considered to be stress driven. When
mass is visibly cracked and may unravel and disin-
the stresses reach this level the interlocking is over-
come and the rock mass undergoes considerable 300 Blk 27 - 6800

non-linear deformation. This deformation is driven 250

Blk 71 - 6800
Blk 72 - 6800
by large forces that may not be held back by ground Blk 73 - 6800
200
support schemes. In fact, ground support must be

1 (MPa)
Blk 22 - 7000
able to move with this deformation if the failed ma- 150 Blk 34 - 7000

terial is to be contained. 100

Blk 33 - 7200
Blk 42 - 7200
Data from a number of mines exhibiting brittle 50 Blk 72 - 7200
rock response suggests that A and B have similar Blk 74 - 7200
0
magnitudes, and the two criterion may meet at 3=0. 0 5 10 15 20 25 30
Best Fit
Observed
It is anticipated that this may not be true for more Cracking
3 (MPa)
compliant rock types. In addition, the rock located
within the zone defined between the two criteria can
be considered to be damaged. As overstressing in- Figure 8. Back analysis of pillar failures.
creases from the lower criterion to the upper one, the
rock mass becomes progressively more sensitive, in Also shown in Figure 8 are results from a back
that it is easier to trigger an unravelling failure, for analysis of locations where cracking was observed in
example by blasting nearby. boreholes. This provides values to be used in the
While the rock mass strength failure criterion dis- failure criterion described above (Equations 1 and
cussed above can be estimated by using empirical 2):
methods that rely on rock mass classification, corre- 1- 3 = 120 Rock Mass Damage Criterion
lation of underground observations and geotechnical
instrumentation with back-analyses is used to verify 1= 124 + 4.1 3 Rock Mass Failure Criterion
whether the estimate is correct and refine the actual
values.
For example in Figure 7, the two pillars towards
the back were observed to fail right through to the 6 ROCK REINFORCEMENT DESIGN
core, while the pillar in the foreground experienced
side wall spalling only. Elastic modelling can be Whether the strength parameters defined in Equa-
used to determine the stress levels respectively in the tions 1 and 2 are determined from back analysis, or
core and side walls. These stresses can then used to estimated using empirical methods that rely on rock
verify the failure criterion. By repeating this type of mass classifications, the design of the rock rein-
back-analysis for many observations in situ, the site forcement in overstressed rock can be achieved us-
specific rock mass compressive strength represent- ing three dimensional numerical modeling . This can
ing stress driven failure can be determined as shown be achieved by assuming that the ground response
in Figure 8. can be described by two categories: broken ground
and cracked ground as shown in Figure 9.

Anchor
length
Bolt
Cracked length
Broken ground
ground

Span

Figure 9. Zones used for rock reinforcement design.

The broken ground is ground that has undergone

stress driven failure and represents the dead weight
that our support needs to suspend. This will be de-
termined using the rock mass stress failure criterion.
Figure 7. Modelled induced stresses in pillars using Consider a highly stressed location shown in Figure
Map3D. 10. To determine the depth of broken ground, the
values of strength divided by stress have been con-
toured, or (124 + 4.1 3)/ 1. The results show that At other locations stress levels may be insuffi-
the broken ground depth extends 2 metres into the cient to induce stress driven failures.Therefore, at
back. such locations this method would predict zero depth
of broken ground and alternative failure mechanisms
and alternative analysis methods depending on the
structural geology and geometry of the openings
must be considered. In medium to low stress condi-
tions there are basically three cases of rock mass to
consider: massive, stratified, and jointed rock. In
massive rock a simple two dimensional, elastic
analysis of the opening and the stress field may pre-
Broken
dict mild spalling as opposed to deep fracturing. In
Ground 2m stratified rock, beam or plate theory may predict
shearing and dilation to occur depending on the ori-
Span 8m entation of the stratigraphy and the orientation and
shape of the opening. This may lead to bending and
buckling with step path failures through the layers or
Figure 10. Contours of (124 + 4.1 3)/ 1 cantilever action and guttering of the layers, which is
common in coal mining collapse mechanisms. In
The cracked ground defines where the reinforce- jointed rock, block theory may be used to the stabil-
ment anchoring begins. This will be determined us- ity of blocks of rock that may translate or rotate to-
ing the rock mass damage threshold criterion defined wards the opening. This mechanism may initiate
earlier. Consequently, the depth to the damage with the loss of individual blocks but may propagate
threshold is determined using the contoured values to a progressive collapse of the block assembly
of ( 1 - 3) as shown in Figure 11. around the opening. In each case it will be neces-
sary to predict the dimensions and extent of the fail-
ure zone and provide a reinforcement and or support
scheme suitable for both global and local stability.
The analysis methods and procedures for reinforce-
ment design in these circumstances have been given
by Hoek and Brown (1981) and Brady and Brown
(1985).
Damage Once an initial design has been formulated this
Threshold 5m
modeling method may also be used to evaluate how
other excavation stabilization techniques affect the
rock reinforcement requirements. This would in-
Span 8m
clude modelling of alternative sequences, reduced
spans and using backfill.
Figure 11. Contours of ( 1- 3)

7 CONCLUSIONS
In this location the dimensions and extent of the
cracked or damaged zone (and within this zone the A procedure has been given for the design of rein-
dimensions and extent of the broken or fractured forcement for highly stressed rock based on 3D nu-
zone) have been determined. The results suggest that merical modelling using the Map3D code. The pro-
cable bolts are required across the 8m span and will cedure involves a sequence:
need to extend past the 5m deep damaged zone to be 1. Characterisation of the rock mass strength
anchored within intact rock. The density of cable and deformability.
bolts may be determined by considering the mass of 2. Characterisation of the stress field.
the damaged zone across the span of the opening. 3. Definition of the mine geometry and excava-
The type, stiffness and installation timing of the ca- tion sequence.
ble bolts chosen will depend on the expected veloc- 4. Modelling of the stress redistribution due to
ity of loading, both for the current circumstance and excavation.
for future mining induced stress changes. Investiga- 5. Back analysis to determine:
tion of the 2m deep broken or failed ground during a. A Rock Mass Damage Criterion
the back analysis stage will indicate if and what type b. A Rock Mass Failure Criterion
of rock bolts and mesh are required to retain the 6. Post-processing of stored analysis results to
broken ground between the cable bolt array spans. define the outer, damaged or cracked zone
and the inner, failed or broken zone.
 7. Primary reinforcement is dimensioned on the
geometry and mass of the damaged and
failed zones.
8. Secondary reinforcement and or support is
dimensioned on the geometry and likely be-
haviour of the failed zone local to the exca-
vation surface.
This procedure is suitable for hard rock mines
which have obtained sufficient data to properly
characterise the rock mass and the stress field. The
back analysis component is required in all cases in
order to calibrate numerical model predictions and
the damage and failure criteria to in situ observa-
tions of cracking. Closing the analysis with observa-
tions in this manner ensures a progression to appro-
priately dimensioned primary reinforcement.

REFERENCES:

Brady, B.H.G and E.T. Brown 1985. Rock Mechan-

ics for Underground Mining. London: George
Allen and Unwin, 527p.
Hoek E. and E.T. Brown 1980. Underground Exca-
vations in Rock. London: Istn Min. Metall.
Villaescusa, E., C.R. Windsor, J. Li, G. Baird and
M. Seto. 2003. Stress measurements from
cored rock. Minerals and Energy Research In-
stitute of Western Australia, Report No. 233.
Wiles, T.D. 2004. Map3D Course notes: Part 1:
Numerical Modelling Background, Part 2:
Rock Mechanics Model Interpretation, Part 3:
Energy Release Rate/Loading System Stiff-
ness/Non-Linear Modelling, Part 4: Integration
of Modelling with Seismic Monitoring.
Wiles, T.D. 1998. Correlation Between Local En-
ergy Release Density and Observed Bursting
Conditions at Creighton Mine, Unpublished
Mine Modelling Report to Creighton Mine.
Windsor, C.R. and A. Thompson, 1993. Rock rein-
forcement technology, testing, design and
evaluation. Comprehensive Rock Engineering.
J. Hudson (Ed). Vol4, Chp 16, pp 451-484,
Pergamon Press, Oxford.

View publication stats