Vous êtes sur la page 1sur 7

Rock mass properties for slope and tunnel design

Propiedades de macizos rocosos para el proyecto de taludes y tneles


Tarcsio B. Celestino
EESC, University of So Paulo, So Carlos, Brazil
Themag Engenharia, So Paulo, Brazil

Abstract
Rock mass properties depend on scale, on damage during excavation, and on the stress path along the construction
sequence. The latter is more important for high slopes and for deep tunnels. Empirical evidence of this fact is
shown here with respect to high slopes. Comments are made about limitations of current correlations between rock
mass classification and engineering parameters. A method is also presented for quantitatively obtaining the
decrease in strength due to damage caused along the stress path for open or underground excavations.

Resumen
Las propiedades de los macizos rocosos dependen de las dimensiones, de los daos durante la excavacin y de la
trayectoria de las tensiones a lo largo de la construccin. Este ltimo es ms importante para taludes de gran
altura y para tneles profundos. Se presentan aqu evidencias empricas de la validez de ese enunciado para
taludes de gran altura. Son tambin presentados comentarios sobre las limitaciones de las correlaciones usuales
entre clasificacin del macizo rocoso e sus propiedades de ingeniera. Se expone un mtodo para obtener de forma
cuantitativa la disminucin de resistencia debida al dao causado por la trayectoria de tensione para excavaciones
a cielo abierto o subterrneas.

1 INTRODUCTION

and tunnels is mechanically inconsistent. In other


words, the parameters are not material properties.
In order to improve the knowledge in this area, it
is important to understand more about the scale
dependent failure phenomena, and the strength of
non-persistent joints. Progress can be made if the
cumulative damage process of the intact material
is taken into account during the whole excavation
process, instead of simply adopting parameters
for the final excavation stage. Hoek et al. (2002)
have already introduced a damage parameter into
the rock mass failure criterion, but how to obtain
that parameter is not a straightforward procedure.
The use of cohesion softening models in
conjunction with stepwise excavation analysis is
shown to simulate the progressive damage of the
intact material, substantially decreasing the factors
of safety with respect conventional analyses.

Obtaining appropriate rock mass engineering


parameters for the design of major excavations
has always been a challenge in rock mechanics
pratice. Recently, some failures of high slopes has
raised even more the attention to that subject.
Comparison of an empirical guideline for
engineered slopes from 20 years ago with data
from recent failure of high slopes seems to show
that the previous guideline was too optimistic. In
other words, high slopes seem to involve different
parameters or failure mechanisms with respect to
moderately high slopes. Conceptually, the same
limitations apply to deep tunnels.
The process of obtaining engineering
parameters from rock mass classification systems
is critically reviewed. It is shown that the
difference of parameters for the design of slopes
115

The procedure may bring an improvement to


current analysis methods. Back analysis of recent
failures however is needed to validate the
procedure, before it is used for design.

conceptual analyses of the failure phenomenon


must go on in order to improve the process of
obtaining
consistent,
material
dependent
parameters. This will be discussed in item 4.

2 EVALUATION
OF
ROCK
MASS
PARAMETERS FROM CLASSIFICATION
SYSTEMS
2.1 General procedure
Geomechanical classification systems were first
introduced relating rock mass attributes to same
characteristics (e.g. support type and quantity) of
engineering work. A good example of that is
Terzaghis (1946) classification. Bartons Q
system (Barton et al., 1974) and Bieniawskis
RMR (Bieniawski, 1974) originally followed the
same trend for tunnels.
However, the need to extend the results of rock
mass classification to more general types of
engineering works led to a new procedure:
classification systems have been used to obtain
mechanical parameters of the rock mass (strength
and deformability) which in turn are used to feed
mathematical models to analyze and design rock
slopes and tunnels alike. The procedure, however,
has presented some inconsistencies. The most
important is that strength parameters for tunnels
and slopes are different.
Hoek et al. (2002) suggest different expressions
for strength parameters for analyzing tunnels and
slopes. The expressions are obtained by fitting a
straight line (Mohr-Coulomb envelope) secant to
the curved Hoek-Brown envelope as shown in
Figure 1. One could argue that different works
lead to different levels of stress, which could
consistently explain the need for different
parameters.
A possible solution would be the adoption of
the curved Hoek-Brown envelope. However,
analyses considering cohesive strength determined
by fitting a tangent to curvilinear envelope are
considered to be optimistic by Hoek et al. (2002).
This suggests that classification systems may be
more reliable when used in the original procedure,
as an empirical tool relating rock mass quality to
characteristics of the engineering work. Large data
bases support such relationships. The second
procedure (classification system rock mass
parameters mathematical model design)
should be used with engineering judgment.
Constant updates of parameters (e.g. Hoek-Brown
criterion) are only one of reasons for that.
While the use of classification systems to obtain
rock mass parameters is a very popular procedure,

Figure 1 Relationships between major and minor


principal stresses for Hoek-Brown and equivalent
Mohr-Coulomb criteria. (Hoek et al., 2002)
2.2 Rock mass damage
Hoek et al. (2002) introduced a damage
parameter D in the failure criterion to take into
consideration the degree of disturbance to which
the rock mass has been subjected due to blasting
and stress relaxation. Those authors present a few
suggestions of values to be adopted for the
parameter D, most of them related to the blasting
quality. For one case of tunnel, substantially
different values for D are suggested when an
invert is or is not adopted. This is a clear
statement that stress concentrations cause damage
to the rock mass which has to be taken into
consideration when adopting strength parameters.
In the case of large open pit mine slopes, a high
value is suggested (D = 0.7) even when
excavation is carried out without blasting, by
ripping and dozing. The reason for damage are
stress concentrations. Nevertheless, Hoek et al.
(2002) suggest only isolated values for a few
different cases, and a rational method to obtain D
has not been presented.
116

unstable slopes proposed by Hoek and Bray


(1977).
It can be argued that the data of higher slopes
leads to a more conservative boundary of stable
slopes than the one proposed by Hoek and Bray
(1977). In fact, taking the trend of the boundary
line proposed by Hoek and Bray (1977) the height
of stable slopes seems to be about 100m in
excess of the height of slopes which actually
failed. This fact supports the argument that the
stability of high slopes involves concepts different
from those well known to govern the stability of
lower slopes.
Many researchers have been concerned with
this fact, and Sjberg (1999) wrote:
Unfortunately, failure mechanisms of high
slopes, especially in hard rocks, are generally
poorly understood and/or known.

The amount of stress concentrations can be


evaluated for intermediate steps of the excavation. The
amount of damage caused to the intact material by
stress concentrations can be used to evaluate the
amount of disturbance of the rock mass, instead of
attributing a value for D based on experience. A real
example of this procedure will be presented in item 5
for a high excavated slope.
It is also important to note that the amount of
disturbance (or the strength decrease) is scale
dependent, even above the scale of thousands of cubic
meters.Further details will be given in item 4.
Back analysis of some large scale failure case
histories reported by Sjberg involving volumes of the
order of 108m shows that the strength decrease (or the
amount of disturbance") is so high that the value for
the parameter D would be even higher than 1, which
was the upper limit proposed by Hoek et al. (2002) for
very disturbed rock masses.

Hoek & Bray (1977)

3 IMPORTANCE OF ROCK MASS DAMAGE


ON SLOPE HEIGHT - EMPIRICAL
ASSESSMENT

Failure (Sjberg, 1999)

700
600
500

H (m)

The collection of data related to successful and


usuccessful case histories is a common practice in
engineering, with the purpose of establishing the
boundaries of the state of practice. For rock slope
engineering, the boundary is usually established in
graphs relating slope height to slope angle. Hoek
and Bray (1977) presented information of several
cases of stable and unstable slopes of open pit
mines, quarries, dam foundation excavation and
highway cuts in hard rock. In addition to the data
collected, they also presented an envelope of the
highest and steepest slopes excavated to that date,
meaning a useful and practical guide to the
highest and steepest slopes which can be
contemplated for normal open pit mine planning,
in the authors words.
The data comprised 99 case histories, with
heights ranging from 60 to 670m, 88% of which
were lower than 200m. It is also clear that the
cases of unstable slopes collected involved
mechanisms with failure of intact rock. There are
at least four cases of unstable slopes with heights
in the range of 200m, and in one case the height
exceeds 250m.
More than 20 years later, Sjberg (1999)
collected similar data and the increase of slope
heights is remarkable. Seven case histories of
failed open pit mine slopes were collected, with
heights ranging from 200 to 658m. Figure 2 shows
the data of unstable slopes collected by Sjberg
(1999), as well as the boundary of stable and

400
300

stable

200
100

unstable

0
0

10

20

30

40

50

60

70

80

90

Slope angle (degrees)

Figure 2 Relationship between slope height and


slope angle: unstable slopes (from Sjberg, 1999)
and boundary between stable and unstable slopes
(from Hoek & Bray, 1977).
Still according to the same author, rock mass
failure criteria which are available, have not been
verified for large scale slopes. On the other hand,
if the strength parameters are chosen for the range
of confining pressures compatible with those
actually installed in the rock mass, the results of
safety evaluation should be appropriate for any
height of slopes. However, this procedure has not
led to safe design in some instances. Therefore, it
is wise to consider that other factors may be
responsible for the apparently unpredictable
behavior of high slopes. In particular, the higher
the slope, the higher the damage caused to the
rock mass during excavation. Indeed, natural high
slopes (much less affected by damage) compiled
117

by Sjstrom (1999) tend to follow more closely


Hoek and Brays (1977) envelope.
A quantitative procedure for evaluating damage
for excavated slopes will be presented in item 5.
4 CONCEPTUAL ASPECTS
MASS STRENGTH

OF

proportional to the specimen diameter, the ratio of


strength decrease was 0.45, as compared to 0.5 of
the theoretical model.
Many arguments have been raised about the
strength decrease with dimensions, and how far is
this trend actually observed. Bieniawski (1984)
for instance discusses the concept of critical size,
above which no strength decrease is observed. For
specimens with dimensions of the order of
magnitude of a few meters or bigger no further
strength decrease is supposed to be observed.
Hoek and Brown (1980) present strength
parameters for intact laboratory specimens and
for rock masses, but no influence of the
dimensions of the rock mass is presented. The
same concept holds for the more recent versions
of their criterion.
However, back analysis of failures reported by
Sjberg (1999) involving volumes ranging from
104 to 108m shows significant strength decrease,
as presented in Figure 4. The decrease should not
be attributed only to scale effect, as the amount of
damage to the intact material due to stress
concentrations (as discussed in item 5) could be
different. However the very low normalized
strength for very large volumes is a very strong
argument that scale effects are still important at
large dimensions.

ROCK

c / c ref

The subject of rock mass strength is too


complex to be consistently treated with a
theoretical model. Significant progress, however,
has occurred, based on the interpretation of model
tests, some of which will be described.
The behavior of persistent joints is rather well
known. They mainly affect failures of small-scale
slopes or localized block fall into underground
excavation. Several models for persistent rock
joint strength are available (e.g. Barton and
Choubey, 1977) and obtaining their parameters
and applying them are straightforward operations.
Significant uncertainly arises when nonpersistent rock joints exist. This is the case of the
failure of high slopes, or global failure of
underground works. The coupled role of rock
bridges and dilation due to joint asperities is
complex, and consistent consideration of all the
phenomena involved is not practical.
Several aspects related to the behavior of nonpersistent joints and their strength have been
analyzed by means of physical model tests
interpreted with theoretical models, like fracture
mechanics. Some of these models will be
presented.
4.1Scale effect
It has been recognized that rock strength is
highly dependent on scale. Results of compression
test on models with flat joint dimensions
following different rules with respect to the
specimen dimension were presented by Bortolucci
(1993) and Bortolucci and Celestino (1996).
Figure 3 shows the results of relative uniaxial
strength (strength of specimen with joints divided
by the strength of intact material 50-mm diameter
specimens) for different specimen diameters. For
joints with constant dimension, not varying with
the specimen diameter (series C and CD) the ratio
of strength decrease with the logarithm of the
diameter was 0.09 and 0.06, as compared to 0.0
according to a model based on fracture mechanics.
Even though the dimension of the artificial joints
was constant, it is wise to consider that the
dimension of the natural intact material
microcracks were not constant, and this explains
the non-zero value. For joints with dimension

D (mm)

Figure 3 Variation of relative strength with


specimen diameter (Bortolucci, 1993)
4.2 Shear strength of non-persistent joints
The role of flat non-persistent joints in rock
masses has also been evaluated in shear, by means
of model tests subjected to biaxial compression.
Figure 5 presents results obtained by Jamil (1992)
in the form of strength envelopes of non-persistent
joints with different orientation with respect to
the major principal stress, joint length Lj, rock
118

bridge length Lr and spacing d. Strength


envelopes of the intact material and persistent
joints are also presented.

The strength gain due to dilation is substantial.


Dilation induces tension at the rock bridges,
which results in increased normal compressive
stress on the joint. The specimens were optically
instrumented, and strains could be measured in
many locations.
Failure modes of rough non persistent joints are
complex, and far from being fully understood as
yet. Localized shear and tension occur, and failure
can be step-wise or along joint planes, depending
on joint arrangement, roughness and external load.
Progress in this area is still needed for better
understanding of rock mass strength.
N o r m a liz e d P e a k S t r e n g t h v s S t r a in

Figure 4 Normalized rock mass strength vs.


volume of construction element (Sjberg, 1999)

0 .7

0 .6

0 .5

(12)/c

Flat joints do not frequently occur in nature, and


tests results on them are not realistic. When
sheared, wing failure occurs; for rough joints,
coplanar fracture propagation takes place,
according to results obtained by Holzhausen
(1978). Model tests of rough joints under shear are
complex to construct, but much more realistic.
Results of stress-strain curves of biaxial
compression tests obtained by Gaitn et al. (2004)
for three different values of joint roughness
coefficient JRC (Barton and Choubey 1977) are
presented in Figure 6. Peak strengths for the same
confining stress, for different roughness
coefficients are presented in Figure 7. Values for
persistent joints are also presented.

JR C = 8
JR C = 4
JR C = 0

0 .4

0 .3

0 .2

0 .1

0
0 .0 0 0

0 .0 0 2

0 .0 0 4

0 .0 0 6

0 .0 0 8

0 .0 1 0

0 .0 1 2

S tra in

Figure 6 Normalized peak strength vs. strain


curves for different values of JRC (Gaitn et al.,
2004).

(1-2)/c

0,6

0,4

persistent joint
non-persistent

0,2

0
0

JRC

10

Figure 7 Normalized peak stress vs. JRC (Gaitn,


et al., 2004).
5 QUANTITATIVE
EVALUATION
DAMAGE FOR HIGH SLOPES

OF

Hoek et al. (2002) propose a damage parameter


D to account for the decrease of the rock mass
strength due to blasting and stress relief. Those
authors, however did not propose any quantitative
process for the evaluation of the parameter D.

Figure 5 Strength envelope of non-persistent


joints (Jamil, 1992)
119

Zea (2004) showed, by means of stress analyses


simulating all the steps of excavation of a high
open pit slope in hard rock, that the stress
concentrations at the of the slope cause damage
responsible for significant decrease of strength.
Figure 8 shows the stress paths at different
locations of the rock mass, corresponding to toes
of temporary configurations. The strength damage
was evaluated by implementing the model of
cohesion softening and friction hardening
proposed by Martin and Chandler (1994). The
distribution of cohesion in the damaged rock mass
is shown in Figure 9. Peaks of strength decrease
are located at elevations where slope toes were
simulated. It can be seen that strength decrease is
more significant for higher slopes. Down to 150m
depth it is negligible.

Figure 9 Cohesion loss due to stress


concentrations during open pit excavation (Zea,
2004)

16
P1
P2
P3
P4
P5

14

1 [ MPa]

12
10

60 m
120 m
180 m
240 m
300 m

The decrease of cohesion is anisotropic,


following the direction of the microcracks
developed. This happens because the orientation
of the principal stresses at a point rotate during the
excavation process. The slope stability was
evaluated using program NONCIR (Celestino and
Duncan, 1981) which estimates the factor of
safety for the critical noncircular slip surface,
taking anisotropy into consideration. The factor of
safety decreased from 2.12 (no damage
considered) to 1.32 (damage considered).
For tunnels, similar analyses can be performed,
considering the damage caused by stress
concentrations due to different locations of the
excavation face. The shape of the cross section
plays an important role on the amount of rock
damage.

8
6
4
2
0
0

60

120

180

240

300

Profundidade
de Escavao
[m]
Excavation
Depth (mm)
0

10

Estgios de Escavao

Excavation Steep

Comeo da escavao
2

3
Estagio de escavao
4
5

Talude Global

300 m
Bancada de 30m

8
9

4
2

Estagios de escavao

Pontos de monitoramento

14
12
10

Final da escavao

P1
P2
P3
P4
P5

3 [ MPa ]

6 CONCLUSIONS

10

40

Rock mass classification systems were


originally proposed as an empirical procedure
based on case history data bases correlating rock
mass attributes and underground excavations
characteristics. More recently, they have been
used to estimate parameters to feed numerical
models for general design. Some inconsistencies
exist in this procedure, and it is important to learn
about rock mass behavior from model tests and
simple theoretical models, as well as back analysis
of field-scale failure.
Compilations of failed and stable rock slopes
carried out 20 years apart seem to shown that high
slopes do not follow the trend of low slopes in
slope height vs. slope angle charts. This can be
explained by damage to the rock mass which is
more pronounced for high slopes. Rock mass

60 m
120 m
180 m
240 m
300 m

6
4
2
0
-2
0

60

120

180

240

300

Profundidade
de Escavao
[m]
Excavation
Depth (mm)
0

10

Estgios
de Escavao
Excavation
Steep

Figure 8 Stress paths at different locations of the


rock mass along excavation process (Zea, 2004)
120

strength decrease due to scale seems to affect even


very large volumes.
Significant decrease has been reported when
comparing failures involving volumes in the range
of 104 to 108m. Scale effects and damage may be
responsible for such decrease, and must be
understood to be rationally evaluated.
A quantitative procedure was presented to
quantify damage and the rock mass strength
decrease.

Symp and 17 Tunnelling Association Canada


Conference, NARMS-TAC, pp. 267-271.
Jamil, S.M. (1992) Strength of non-persistent
rock joints. Ph. D. Thesis, University of
Illinois at Urbana Champaign.
Martin, C.D. and Chandler, N.A. (1994) The
progressive fracture of Lac du Bannet
granite. International Rock Mechanics Min.
Sci & Geomechanics Abst., 31(4), 643-659.
Sjberg, J. (1999) Analysis of large scale rock
slopes. Lulea, Doctoral Thesis, Lulea
University of Technology.
Terzaghi, K (1946) Rock defects and loads on
tunnel sipports. In Rock Tunnelling with steel
support (eds. R.V. Proctor and T.L. white) 1,
17-99, Commercial Shearing and Stamping Co.
Zea, E.R. (2004) Failure mechanisms for high
open pit mine slopes. (in Portuguese). M.Sc.
Dissertation, So Carlos, University of So
Paulo, 124p.

REFERENCE
Barton, N., Lien R. and Lunde, J. (1974)
Engineering classification of rock masses for
the design of tunnel support. Rock Mech.,
6(4), 189-239.
Barton, N. and Choubey, V. (1977) The shear
strength of rock joints in theory and practice.
Rock Mech., 10 (1-2), 1-54.
Bieniawski,
Z.T.
(1974)
Geomechanics
classification of rock masses and its
application tunneling. Advanced in Rock
Mechanics 2, part A, pp. 27-32, National
Academy of Sciences, Washington.
Bieniawski, Z.T. (1984) Rock mechanics design
in mining and tunneling, Rotterdam A.A.
Balkema.
Bortolucci, A.A. (1993) Compression failure
model for brittle materials based on fracture
mechanics applied to scale effect (in
Portuguese).
Doctoral
Thesis,
EESC,
University of So Palo, 132p.
Bortolucci, A.A. and Celestino, T.B. (1996)
Probabilistic model for failure of brittle
materials under compression based on fracture
mechanics. In, M. Aubertin et al. (eds.), 2nd
North American Symposium Rock Mechanics,
2, 1715-1720.
Celestino & Duncan (1981) Simplified search for
nonciricular slip surfaces. Proc. 10th Int.
Cong. Soil Mech. Fdn. Eng., Stockholm (1)
391-394
Gaitan, V.H., Bortolucci, A.A., Celestino. T.B
(2004) Fracture propagation in rouge nonpersistent joints. Salzburg, EUROCK 2004,
v.1, 521-524.
Hoek, E. and Bray, J. (1977) Rock slope
engineering. London, The Institution of
Mining and Metallurgy.
Hoek, E. and Brown, E.T. (1980) Empirical
strength criterion for rock masses. J. Geotech.
Eng. Div., ASCE, 106 (GT9), 1013-1035.
Hoek, E., Carranza-Torres, C. and Corkum, B.
(2002) Hoek-Brown failure criterion - 2002
edition. 5th North American Rock Mech.
121

Vous aimerez peut-être aussi