Вы находитесь на странице: 1из 15

International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

A theoretical model for rock joints subjected to constant normal


stiffness direct shear
Julian P. Seidela,b, Chris M. Haberfielda,c,*
a
Foundation QA Pty Ltd, Croydon, Victoria 3136, Australia
b
Department of Civil Engineering, Monash University, Victoria 3800, Australia
c
Golder Associates Pty Ltd, Hawthorn West, Victoria 3122, Australia

at ly
in on
n
Abstract

io
em se
The authors have conducted an investigation into the behaviour of rock joints subjected to direct shear. Both concrete/rock and
rock/rock joints were investigated. The behaviour of rock/rock joints is important for the assessment of stability issues involving

ss u
rock masses (e.g. rock slope stability). Concrete/rock joints are vital to the assessment of performance of concrete piles socketed into
rock, rock anchors and concrete dam foundations.
di ch
This investigation included an extensive series of direct shear tests under a range of stress boundary conditions. The rock used for
the tests was a soft artificial siltstone, called Johnstone. The results from the tests on concrete/Johnstone joints have been presented
or ar
in Seidel and Haberfield (Geotech. Testing J. (2002), accepted for publication) and on Johnstone/Johnstone joints in Fleuter
(MEngSc Dissertation, Department of Civil Engineering, Monash University, Australia, 1997) and Pearce (Ph.D. dissertation,
e

Department of Civil Engineering, Monash University, Australia, 2001, in preparation). This paper describes the theoretical models
ng s

developed to simulate the observed behaviour, including asperity sliding, asperity shearing, post-peak behaviour, asperity
yi re

deformation and distribution of stresses on the interface. These models have been combined into a micro-mechanical simulation of
joint shear. Comparisons between program predictions and measured performance are presented and discussed. r 2002 Elsevier
op al

Science Ltd. All rights reserved.


r c on
fo rs

1. Introduction tested [3], this paper will concentrate on joints contain-


ot pe

ing two-dimensional roughness profiles—that is, rough-


As part of an investigation into the performance of ness in the shearing direction only.
piles socketed into weak rock, the authors have under- In essence, the experimental program comprised two
N r
Fo

taken a laboratory-based experimental investigation parts—testing of simple and regular triangular idealiza-
based on direct shear testing. As discussed in Seidel tions of a rough interface, and testing of interfaces with
and Haberfield [1], these tests were all performed under complex roughness, which are referred to as fractal
constant normal stiffness conditions. The tests con- profiles. Profiles of both types are shown in Fig. 1. The
ducted for this investigation were performed on genesis of the fractal profiles [5] is not central to the
concrete/rock samples with a range of geometries and purpose of this paper. The triangular profiles were tested
stress boundary conditions. Fleuter [2] and Pearce [3] in order to provide experimental data for the develop-
completed a similar set of tests on rock/rock joints. The ment of the theoretical micro-mechanical models. It was
rock used in the tests reported herein is a reconstituted hypothesized that profiles with complex geometry could
siltstone known as Johnstone. Johnstone has been be idealized as a series of simple triangular asperities. It
demonstrated to have strength and deformation proper- followed that if appropriate consideration was made for
ties close to the natural siltstone from which it is derived interaction between these individual asperities, the shear
but without the inherent variability [4]. Although joints behaviour of the more complex fractal profiles could be
containing three-dimensional roughness have been predicted from the models developed from triangular
asperities. The fractal profiles were tested to validate this
hypothesis.
*Corresponding author. Golder Associates Pty Ltd, PO Box 6079,
Hawthorn West, Victoria 3122, Australia. Tel.: +61-3-8862-3500;
The tests on simple regular triangular profiles
fax: +61-3-8862-3501. indicated that shear behaviour involved two basic and
E-mail address: chaberfield@golder.com.au (C.M. Haberfield). apparently independent mechanisms; initial sliding

1365-1609/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 5 - 1 6 0 9 ( 0 2 ) 0 0 0 5 6 - 4
540 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

1200
5 x 3.75 mm high regular triangles

10 x 7.5 mm high regular triangles 900

12.5 x 6 mm high regular triangles

τ(kPa)
600
15 x 7.5 mm high regular triangles

17.5 x 9.5 mm high regular triangles


300
Triangular profile
22.5 x 9.5 mm high regular triangles Fractal profile

0
27.5 x 11.5 mm high regular triangles 0 10 20 30 40 50

at ly
(a) X (mm)
Fractal Class A

in on
n
1200

io
Fractal Class B
Triangular profile

em se
Fractal profile
900

ss u
Fractal Class C

Fractal Class D
di ch τ(kPa)

600
or ar
Fractal Class E
e
ng s

0 50 100 150 200 250 300 350 400 450 500 550 300
yi re

Length (mm)
Fig. 1. Roughness profiles used in direct shear testing.
op al

0
0 10 20 30 40 50
r c on

along the surface of the asperities and then simultaneous (b) X (mm)
shearing through all of the intact asperities. These two
fo rs

Fig. 2. Typical shear stress–displacement responses for triangular and


mechanisms have been reported previously by several fractal profiles. (a) J/C joints (regular 101 asperity profile with 42.5 mm
ot pe

researchers including Patton [6], Ladanyi and Arch- chord length and Class B fractal profile with 16 mm chord length tested
ambault [7] and Lam and Johnston [8]. Post-peak with initial normal stress, sn0 ¼ 300 kPa and normal stiffness,
behaviour is also important to the overall performance K ¼ 300 kPa/mm). (b) J/J joints (regular 101 asperity profile with
N r

48 mm chord length tested with initial normal stress, sn0 ¼ 400 kPa,
Fo

of the joint. Typical shear stress–displacement responses normal stiffness, K ¼ 267 kPa/mm and fractal profile with a mean
for Johnstone/Concrete (J/C) and Johnstone/Johnstone asperity angle of 51, chord length of 16 mm, length tested with initial
(J/J) joints containing 101 regular triangular roughness normal stress, sn0 ¼ 400 kPa and normal stiffness, K ¼ 450 kPa/mm).
profile are shown in Fig. 2.
Although the shear behaviour of joints with more
complex geometry is also controlled by these same two ment responses for Concrete/Johnstone and J/J joints
basic mechanisms, the mechanisms cannot be readily containing fractal roughness profiles. The sliding and
isolated in the joint response. Shear displacement is shear mechanisms cannot be separately identified.
initially affected by sliding on the steepest asperities. The This paper deals only with the shear displacement
consequence of this sliding is joint dilation, which causes behaviour of clean rock joints and in particular, the
shallow asperities to be lifted out of contact, and stresses micro-mechanical models developed by the authors to
to be localized on the steep asperities in contact. At a simulate the behaviour observed in constant normal
critical displacement, the shear stresses on the steepest stiffness direct shear tests. The influence of joint infill
asperity exceed the asperity strength. The asperity fails, and joint degradation have not been considered and the
and its load is shed to other asperities. Dilation is then reader is referred to Indraratna and Haque [9] (for
controlled by the next steepest asperity until it too fails example) for a review of other joint shear models and
in shear. Thus, sliding, progressive asperity shear and the effects of infill and degradation. The paper
post-peak sliding occur simultaneously in complex concludes with a comparison of measured and predicted
profiles. Fig. 2 also shows typical shear–stress displace- joint shear responses.
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 541

2. Triangular asperity models 2.2. Asperity shearing

2.1. Asperity sliding Initially, before the commencement of sliding, the two
halves of the joint are assumed to be in intimate contact
The direct shear tests [1–3] confirmed the sliding with both faces of each asperity in full contact. After the
models of Patton [6] and Ladanyi and Archambault [7] initiation of interface slip, the contact area between the
for unbonded purely frictional surfaces. The average two halves of the joint is restricted to one asperity face,
shear stress for sliding, t, is given by and progressively reduces as shear displacement pro-
gresses. This is demonstrated in Fig. 3 for an interface
t ¼ s tanðfb þ yÞ; ð1Þ comprising regular triangular asperities. Local normal
stresses increase both as a consequence of the reduced
where s is the average normal stress applied to the joint, contact area and as a result of the increasing normal
fb is the base friction angle of the joint and y is the stress due to the constant normal stiffness (CNS)
asperity inclination. It should be noted that Eq. (1) was condition. A critical normal stress is reached at which
derived for rigid asperities, which may be a reasonable the asperity can no longer sustain the loading and

at ly
representation of hard rock. However, for the soft rock individual asperity failure results.

in on
(uniaxial compressive strength of about 3.5 MPa) used The asperity shearing mechanism was observed to

n
in this investigation, the compressibility of the rock is differ between J/C and J/J joints. For J/C joints, the

io
significant and cannot be neglected. Seidel and Haber- much stronger half of the joint constrained failure over

em se
field [10] adopted the energy approach used by Ladanyi the full contact length of each asperity. However, for J/J
and Archambault [7] to show that Eq. (1) also holds for joints the material on both sides of the interface is

ss u
elastic asperities. This result in itself is not startling, but similar, allowing failure to occur at localized regions of
di ch
as demonstrated later, it is important when considering high stress that occur at the leading and trailing points
the shear behaviour of irregular profiles consisting of of contact of each asperity. Failure gradually progressed
or ar
multiple asperities of varying angle. from these localized regions until complete failure of
The results from the tests on regular triangular each asperity (and therefore of the whole interface)
e

asperity profiles indicated an average sliding or base occurred. This resulted in a significant reduction in the
ng s
yi re

friction angle of 24.51 [1,2]. This angle compares measured strength. The finite difference program FLAC
favourably with the residual sliding friction angle for was used to investigate the failure of both J/J and J/C
op al

horizontal concrete/Johnstone interfaces of 231 deter- interfaces. The results of this analysis indicated that the
mined by Lam [11] and the residual angle of 241 ultimate failure mode in J/J joints was similar to that of
r c on

reported by Kodikara [12] for flat J/J interfaces. The J/C joints, but failure occurred at a lower stress. A stress
tests did not indicate any appreciable difference between reduction factor of 1.38 was found to be appropriate for
fo rs

peak and residual angles of friction. J/J joints [3].


ot pe

The purely frictional behaviour is contrary to the Observation of video records of the direct shear tests
behaviour reported by Lam and Johnston [8] and of both the J/C and J/J joints clearly shows asperity
Kodikara and Johnston [13] for tests on similar J/C failure to be rotational. This is in direct contrast to other
N r
Fo

joints. Both of these investigations reported significant models, such as Patton [6] and Johnston and Lam [14],
levels of cohesion during sliding. The reason for the which were based on a planar failure surface. Similar
cohesion observed by these researchers is not well observations of rotational failures are attributed to
understood. Nelson [15].

σn = ( 2λ ). ( σ + Kψ)
no
opposing λ–x
interface

ψ
rock
x


Fig. 3. Reduction of asperity contact area with progressive shear displacement.
542 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

The form of the failure planes observed in the direct w


shear testing, along with geometric considerations, δ q
suggested the use of slope-stability methods to model
Y
the shear failure of the rock asperities. However, the
θo
implementation of a slope-stability analysis within an r = ro exp [ tanϕ .(θ–θo )] θ µ
µ
overall computer program designed to predict the
µ– ρ –ρ
behaviour of complex joint geometries is very cumber- so
some. As a result, a closed-form solution for the failure Major
principal stress
of a weightless slope with slope angle g in a c2f soil
subjected to an inclined load [16] was adopted. Given
µ Failure Surface
the negligible influence of self-weight on the strength of
the asperity, the assumption of weightlessness is not
ψ
significant. Fig. 4 shows the principal features of X
Sokolovsky’s solution. Sokolovsky’s original axis con-
Fig. 4. Solution for weightless c  f soil (after Sokolovsky [16]).
vention, which comprises an X-axis vertically down and

at ly
a Y-axis horizontally away from the slope is retained.

in on
The analytical solution comprises two failure planes

n
connected by a log-spiral as shown in Fig. 4. In the case solution is valid. However, if the ratio is smaller, the

io
under consideration, the load inclination is assumed to ‘‘free’’ failure mechanism will be affected, and a

em se
be equal to the sliding friction angle of the joint. The ‘‘constrained’’ failure mode, at a higher failure stress
solution is obtained indirectly by the simultaneous will result. It can be seen from Fig. 5, that s=wos0 =w in

ss u
solution of the following three equations: the initial mated position, but then equals and exceeds
q cos d ¼ sð1 þ sin f cos 2rÞ  H;
di ch ð2Þ s0 =w as shear displacement proceeds. Values of s0 =w are
listed in Table 1 for different slope angles. As no
or ar
q sin d ¼ s sin f cos 2r ð3Þ analytical solution exists for such constrained modes of
failure, a numerical approach was adopted to extend the
e

and analytical solution for unconstrained failure to cases


ng s
yi re

H where a limited slope length to load width ratio exists.



cosd A parametric study was conducted using the multiple-

op al

1 þ sin f cos 2r wedge slope-stability program EMU [18]. EMU is based


 exp½ðp  2g þ 2rÞ tan f  1 ; on an upper-bound energy method approach, and has
1  sin f
r c on

been found to agree closely with available analytical


ð4Þ
solutions [19]. Analyses were performed varying the
fo rs

where s ¼ ðs1 þ s3 Þ=2 and H ¼ c cot f: slope angle, the ratio of slope length to load width, and
ot pe

For the case of regular asperities considered here, d ¼ the Mohr–Coulomb strength parameters. The analyses
fb ; g ¼ 2a (a=asperity angle), and c and f are the for slope to load width ratios in excess of s0 =w; were
intact strength parameters of the rock. The solution found to be in good agreement with the analytical
N r
Fo

for q is obtained numerically by application of the solution. On the basis of this parametric study, a
Newton–Raphson method [17], for which there is rapid constrained failure correction factor, k; was defined to
convergence. relate failure stress, qc to Sokolovsky’s [15] failure stress
It should be noted that the Sokolovsky solution is for for a long slope, qf (k ¼ qc =qf ):
a slope of infinite length and assumes that the failure  s0 
envelope intersects the slope above the toe. Any sos0 ; k ¼ 0:85 exp 0:16
s
constraint imposed by a limited slope length would
increase the predicted failure stress. The solution, and
however, does allow the theoretical ratio of the distance, sXs0 ; ks ¼ 1: ð6Þ
s0 (from the crest of the slope to the point of intersection
of the failure plane with the slope) to the width of the As indicated in Fig. 5, the minimum value of s=w for
loaded area, w, to be determined as an asperity is 1.0, which occurs in its undisplaced, fully

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s0 2 cos m tanðp=2  m  jrjÞ 1 þ tan2 ðp=2  m þ jrjÞ expftan fðp=2  g  jrjÞg
¼ : ð5Þ
w tanðp=2  m  jrjÞ  tanðp=2  m þ jrjÞ

If the actual slope length to loaded area width ratio, mated position. The maximum value of s0 =s occurs
s=w (as shown in Fig. 5) exceeds s0 =w; the closed-form when s ¼ w: From Table 1, for a 12.51 asperity, the
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 543

opposing interface
opposing σn
w s interface

so
τ
ψ Rock
Mated position (constrained failure-worst case)
x θ

opposing interface 2λ
w
s Fig. 6. Simplified rigid asperity model for slope-stability analysis
predictions.
rock so

Critical shear displacement for transition from


constrained to unconstrained failure. argument, asperities will fail when the local stress is
equal to the ‘‘constrained’’ failure stress, qc ; computed

at ly
by the corrected Sokolovsky solution.

in on
At the failure shear displacement, xf ; the joint

n
opposing interface
w dilation, c; for the rigid asperity is given by c ¼

io
s xf tan y and the global normal stress, sn ¼ sn0 þ

em se
rock so Kxf tan y: From the frictional asperity sliding model,
the global failure shear stress, tf ; is given by tf ¼

ss u
General position-unconstrained failure. sn tan ðfb þ yÞ:
di ch
Fig. 5. Determination of the ratio s=w in mated and displaced At failure, the local stress perpendicular to the face of
positions. the asperity, snl ; can be determined as
or ar
snl ¼ ðsn0 þ Kxf tan yÞ
e

Table 1  ðcos y þ sin y tanðfb þ yÞÞ2l=ðl  xf Þ: ð7Þ


ng s

Comparison of EMU and Sokolovsky solutions


yi re

Defining a shape factor, zy ¼ cos y þ sin y tanðfb þ


Asperity s0 Predicted failure stress, qf EMU yÞ; and equating the local stress and the normal
w
op al

angle a1 (MPa) Sokolovsky component of the corrected Sokolovsky failure stress,


scr ; allows the failure shear displacement to be obtained
r c on

EMU Sokolovsky
as
0.0 1.86 16.35 15.89 1.029
fo rs

5.0 1.73 12.93 12.62 1.025 lðscr  2zy sn0 Þ


10.0 1.62 10.27 10.03 1.024 xf ¼ : ð8Þ
scr þ 2lzy k tan y
ot pe

12.5 1.58 9.14 8.94 1.023


15.0 1.54 8.10 7.96 1.018 The global failure shear stress, tf ; can hence be
17.5 1.51 7.16 7.09 1.010 computed as
N r

22.5 1.45a 5.54 5.61 0.986


Fo

27.5 1.41a 4.37 4.42 0.986 tf ¼ tanðfb þ yÞðsn0 þ Kxf tan yÞ: ð9Þ
a
Extrapolated values, as s0 =w undefined for a > 22:51: Fig. 7 compares the experimentally determined global
failure shear stresses for the regular triangular asperity
tests with predictions from the rigid asperity model. The
maximum value of s0 =s is therefore 1.58, which implies a stress reduction factor of 1.38 has been applied to the J/J
maximum correction factor, k ¼ 1:09 for this asperity tests. In general, predicted peak shear stress values are
angle. That is, corrections necessary to the Sokolovsky within 5–10% of the experimental values, which is
solution are relatively modest. considered to be within the expected scatter of the
In order to validate that the Sokolovsky solution experimental results.
(with correction as necessary) was appropriate to model Deformations have not been taken into account in
asperity failure, this approach was applied to prediction this simple model. Dilation in the shear test will be
of the global shear stress at failure, tf ; for the triangular reduced by asperity elastic compression. This results in a
asperity tests. For the purposes of this validation, a slower development of normal and shear stresses than
simplified rigid asperity model was adopted. The assumed in the ideal rigid assumption. Experimental
interfaces are assumed to comprise regular symmetrical evidence confirmed that contact lengths at failure were
asperities of angle, y; and half-length, l (Fig. 6). The shorter than those predicted by the simple model.
initial global normal stress is designated, sn0 ; and the This comparison supports the contention that the
global stiffness, K. On the basis of the preceding failure mechanism of regular triangular asperities in
544 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

2000 mechanism which ensures that the failure surface must


5 deg (J/C)
10 deg (J/C)
emanate from the rear of the applied load.

%
Observation of video camera records of the shear tests

0
12.5 deg (J/C)

+1
15 deg (J/C)
1600 on profiles comprising only regular triangular asperities
Predicted Shear Stress (kPa)

17 deg (J/C)
22.5 deg (J/C)
0% confirms the development of a curved failure surface
27.5 deg (J/C) -1 emanating from the apex of the loading concrete
5 deg (J/J)
1200 10 deg (J/J) asperity, and intersecting the trailing face of the rock
20 deg (J/J)
asperity, as shown in Fig. 8. The geometry of the
observed failure surfaces, and in particular the location
800 of the intersection point on the trailing rock face, are
predicted reasonably well by the closed-form solution of
Sokolovsky.
400 However, during the laboratory tests, it was observed
that further shear displacement did not occur on the
curved failure surface, but rather on a chord linking the

at ly
0
0 400 800 1200 1600 2000 intersection points of the initial failure surface with the

in on
leading and trailing asperity faces (see Fig. 8). The

n
Measured Peak Shear Stress (kPa)
inclination of this chord is defined in Sokolovsky’s

io
Fig. 7. Predicted and measured global failure shear stresses for J/C method by the angle of the slope, and the relative

em se
and J/J joints containing regular triangular asperities.
lengths of slope distance, s0 ; to load width, w; as given in
Eq. (5). This chord defines the lower planar surface for

ss u
shear movement. As noted, this model has been
weak rocks such as Johnstone is analogous to a slope- di ch postulated on the basis of experimental observation
stability failure, and can be modelled using the alone, and has not been derived from theoretical
or ar
Sokolovsky analytical solution. It is recognized that considerations.
this failure mechanism will be affected in natural rock by
e

unfavourably oriented planes of weakness. This ap-


ng s
yi re

proach has been found to be applicable to other weak 2.4. Post-peak friction angle
rocks, but does not appear to apply to harder rocks such
op al

as basalt and granite [3]. It is apparent from the previous discussions that the
It should also be noted from Fig. 4 and Eqs. (2)–(4) planar shear surface along which the rock wedge moves
r c on

that the Sokolovsky solution is a function only of the is entirely within failed material at residual strength
material strength parameters, the load orientation and (f ¼ 24:51; c ¼ 0 kPa). The failed rock between this
fo rs

the asperity angle. As the solution involves no dimen- shear surface and the opposing asperity is effectively a
ot pe

sional term, the solution is scale-independent. This granular material, held in place by the high local
suggests that this approach can be applied equally to the confining stresses. Fig. 9 shows the sheared asperity
evaluation of joints with small scale roughness, e.g. pile after some post-peak displacement has already occurred,
N r
Fo

sockets or rock blocks in an underground excavation, or and it is evident that part of the zone of sheared material
to large scale roughness, such as at the interface of a extends beyond the trailing face of the failed rock
concrete gravity dam with the founding rock or for a asperity, where it is unconfined, and spalls as debris
major rock slide. down the trailing face.
Careful examination of the video records reveals that
the zone of granular material exhibits a rolling motion
2.3. Post-peak behaviour at the contacts with both the opposing intact asperity
and the rock shear plane. This is accompanied by
The Sokolovsky closed-form solution described in the relative motion both of the zone of granular material
previous section is also convenient in that it defines the with respect to the intact rock, and of the opposing
geometry of the failure surface (Fig. 4). As pointed out intact asperity with respect to the granular zone.
in the previous section, the failure surface geometry, and As shown in Fig. 10, a relative post-failure movement
the failure strength are functions of asperity angle, of dx between the opposing and rock asperities is
independent of any linear dimension. It follows that all composed of movements of both the opposing asperity
the material bounded by the failure surface, the slope, and the failed rock, and between the failed and intact
and the loaded portion of the crest will simultaneously rock. For the sake of analysis, the movement on the
reach failure, and reach residual strength. However, the post-peak shear plane (angle b) is denoted as e dx.
rigidity of the opposing asperity which imposes the crest The movement between the opposing asperity and the
load superimposes a kinematic restraint on the failure compressed granular material (angle y) is then
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 545

Fig. 8. Schematic representation of post-peak shear failure.

at ly
in on
n
io
em se
ss u
di ch
or ar

Fig. 9. Schematic representation of post-peak shear displacement.


e
ng s
yi re
op al
r c on
fo rs
ot pe
N r
Fo

Fig. 10. Relative sliding movements of failure wedge.

ð1  eÞ dx. The ratio, e; is dependent on the relative function of the friction angle, and the inclination of the
angles of the two planes. planes. The general equation which applies is found to
In order to determine the value of e; two idealized be
limiting cases are shown in Fig. 11. For the two parallel
planes separated by rollers, a movement dx of the upper e ¼ tanðfb þ yÞ=½tanðfb þ yÞ tanðfb þ bÞ: ð10Þ
plane relative to the fixed plane will cause a movement
dx=2 of the rollers. For this limiting case, therefore, Using an energy approach similar to Ladanyi and
e ¼ 0:5: For the two planes inclined at a relative angle of Archambault [7], Seidel and Haberfield [9] were able to
ðp=2  fÞ; dilation of the upper plane is not possible, derive the following expressions for the required
and a movement dx of the upper plane relative to the components of shear force.
fixed plane, therefore causes a movement dx of the For the component required to dilate (or contract)
rollers. In this case, e ¼ 1:0: against the normal force:
From the limiting cases, it can be seen that the relative
movement between the rollers and each plane is a t1 ¼ sn ðe tan b þ ð1  eÞ tan yÞ: ð11Þ
546 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

Table 2
Post-peak friction angles for regular triangular asperities

Asperity Failure plane angle b1 Post-peak friction angle (1)


angle a1
Predicted Observed Predicted Observed

5.0 1.3 4.2 25.9 27.5


10.0 2.4 6.5 26.8 27.5
12.5 2.9 2.6 27.0 27.0
15.0 3.3 3.4 27.2 27.2
17.5 3.7 6.3 27.5 27.5
22.5 4.2 5.9 27.8 27.5
27.5 4.6 5.7 27.8 o30a
a
Residual had not been reached when the test was terminated.

at ly
Simple application of Patton’s Eq. (1) in the post-

in on
peak phase during which the joint contracts (implying a

n
negative value of y) would result in a predicted post-

io
Fig. 11. Limiting cases for sliding between two planes.
peak friction angle less than the residual friction angle of

em se
24.51. In fact, all the observed post-peak friction angles
It should be noted that where b is a negative angle, exceed the residual friction angle, and are relatively

ss u
this implies some release of energy. constant for all asperity angles (27.01 to 27.51). The
di ch
The additional friction due to contraction on the values predicted by Eq. (15) are also in excess of the
shear plane (or dilation if the shear plane is a positive residual friction angle and on average only 0.41 less than
or ar
angle), is given by observed (range from 25.91 to 27.81).
e

t2 ¼ t tan fjtan bj: ð12Þ


ng s
yi re

As the concrete asperity of the opposing interface 2.5. Multiple asperity model
physically follows the post-peak shear plane, no
op al

component due to additional work from dilation at As argued in the introduction to this paper, the fractal
the angle of the concrete asperity is needed. model of roughness [5] allows one to simulate complex
r c on

The final component of sliding friction, t3 ; remains profiles as a series of triangular asperities of constant
unchanged, i.e. length. Both the included angle of the asperities and the
fo rs

orientation may vary, as dictated by the particular


t3 ¼ sn tan f ð13Þ
ot pe

profile being simulated [1]. The behaviour of the profile


leading to can be determined by accounting for interaction effects
between asperities and summing the responses from
N r

sn ftan fb þ e tan b þ ð1  eÞ tan yg


Fo

t¼ ð14Þ individual asperities. This approach has been used to


½1 þ tan fb tan b
extend the single asperity model to irregular profiles
and by substitution from Eq. (10): containing multiple asperities.

sn ftanðfb þ yÞðtan fb þ tan bÞ þ tanðfb þ bÞðtan fb þ tan yÞg


t¼ : ð15Þ
½1 þ tan fb tan b½tanðfb þ yÞ þ tanðfb þ bÞ

The Sokolovsky failure plane angles and the predicted In a theoretical profile comprising rigid asperities of
post-peak friction angles (computed from Eq. (15)) for varying angle, shear displacement is accompanied by
regular triangular concrete/Johnstone joints are com- dilation on the steepest asperity face. All other asperities
pared with observed values in Table 2. The predicted are lifted out of contact. The total sample load is
failure plane angles, b; progressively increase from 1.31 concentrated on a single asperity (the steepest), resulting
to 4.61 for asperity angles from 51 to 27.51. By in very high local stresses. By contrast, in the same
comparison, the observed failure plane angles vary in an profile comprising compressible asperities, the deform-
irregular manner from 2.61 to 6.31. On average, the ability of the sample ensures that load is shared between
predicted failure plane angles are 1.81 less than those asperities of different angles. If the high local stress
observed. computed for the steepest asperity in the rigid profile
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 547

were imposed on the equivalent compressible asperity, A further consequence of the deformations in an
the accompanying deformations would result in recon- elastic profile is that the distribution of stresses between
tact with other asperities of lower inclination. As a adjoining asperities is non-uniform. The distribution of
consequence, sliding occurs not only on the currently loads between individual asperities is a complex function
steepest asperity slope, but also on asperities of lower of asperity angles, the relative position of asperities, and
slope angles [20]. the ratio of contact area to sample depth. Under the
The energy approach of Ladanyi and Archambault [7] assumption of elasticity, the deformation field and the
can be used to establish the sliding resistance of an distribution of loads are uniquely related. It follows
individual asperity of angle y; within an interface therefore, that if the deformation field can be estab-
dilating at angle y0 ; which is greater than y: The imposed lished, the distribution of loads to individual asperities
asperity normal stress is sn (Fig. 12). This problem can be computed using matrix methods and enforcing
differs from the solution for a rigid asperity only in as global compatibility.
much as the dilation angle is y0 ; not the asperity angle y:
By assuming that the deformation of the asperities is 2.6. Asperity deformations
elastic, it can be shown that the additional increment of

at ly
work, dW ; required to dilate against the imposed The deformation of individual asperities is derived

in on
normal stress over a shear displacement, dx; and the from two components; the deformation due to stresses

n
increment of strain energy, dU; released by the faster applied directly on an asperity’s surface and the

io
rate of dilation are equal and opposite, i.e. concomitant deformations resulting from loading of

em se
dW ¼ dU ¼ dxðtan y0  tan yÞsn : ð16Þ other surrounding asperities. The zone of influence of
deformations around an individual asperity is a function

ss u
As there is no change in the energy requirement for of the interface geometry. Again, deformations are
di ch
shear displacement, the shear force required to slide on
the elastic asperity is shown to be a function of the
assumed to be elastic.
The CNS laboratory shear test samples are profiled in
or ar
asperity geometry, independent of the interface dilation such a way that the asperities are prismatic elements
angle. The global shear force required for sliding of an which extend the full width of the sample. When these
e

elastic joint profile comprising n asperities of variable asperities are loaded, the load will be applied in a strip
ng s

angles, yj ; will be the sum of the individual asperity


yi re

across the full face of the asperity. The load strip on


shear forces. The global shear stress is then given by each contacting asperity face will produce an elastic
1X
op al

n deformation profile in a one-dimensional zone of


t¼ aj snj tanðfb þ yj Þ; ð17Þ influence (along the length of the sample) about that
A j¼1
r c on

asperity. The deflection of every asperity in the profile


where A is the total joint plan area, and aj and anj are will be affected by the load on each asperity (including
fo rs

the contact areas of, and normal stresses acting on the itself).
ot pe

individual asperities. As it is a general equation, no In order to establish the distribution of load on


distribution of asperity angles is implied. The equation individual asperities in the test sample, it is necessary to
holds for elastic asperities, because the potential energy determine the compatible and unique set of asperity
N r
Fo

components in Ladanyi and Archambault’s rigid asper- elastic deformations and asperity stresses that match the
ity analysis which reduce due to elasticity effects are applied global stress. This distribution will be a function
exactly compensated for by an increase in stored strain of interface and individual asperity geometry, the
energy in the asperity [9]. current shear displacement and contact lengths.
Unfortunately, the influence of a finite loaded area on
an elastic medium of finite depth is beyond simple
analysis and, to the knowledge of the authors, general-
n
dψ ized elastic solutions are not available. Solutions could
be obtained from numerical techniques such as finite
element or boundary element solutions. However, the
 implementation of these solutions into a micro-mechan-
ical solution approach would be cumbersome.
ψ A more practical solution is to adopt the analogy of a
footing at the surface of a finite elastic medium
underlain by a rigid base. The approximate and
′ numerically convenient method of Steinbrenner [21]

was adopted. Steinbrenner’s method is a widely known
x technique used by geotechnical engineers for the
Fig. 12. Schematic representation of general asperity geometry. determination of the settlement of surface footings
548 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

situated on an elastic layer of finite depth. The method is of asperities in the profile. It is believed that violation of
based on the assumption that the vertical deflection, the assumption of a horizontal bounding surface will
DrD ; below the corner of a uniformly loaded rectangular not significantly affect the computation of relative
area on the surface of an elastic medium of finite depth, deflection profiles. Second, the computation only relates
D; is equal to the difference between the vertical to elastic deformations normal to the interface, and
displacement at the surface, Dr; and the vertical takes no account of shear deformations. Taking
displacement Dr0 at a depth D below the same loaded simultaneous account of normal and shear stresses,
area. although ideal, would add a significant level of
The settlement of points both within a uniformly computational complexity, and analysis time. It is
loaded area, and outside the loaded area can be rationalized that the distribution of shear stresses in
determined by algebraic summation. Thus, deforma- the sample will reflect the distribution of normal
tions in the zone of influence around each asperity can stresses, and that as a result, the net effect on
be estimated by this method. The method is particularly computation of relative deformations (and stresses)
attractive as a wide range of load width to depth ratios should be small.
can be accommodated.

at ly
The deflection profiles predicted by the Steinbrenner
2.7. Stress distribution determination

in on
method for a strip load are compared with the results of

n
a plane strain linear elastic finite element analysis using

io
As discussed in the previous section, the interface
the Strand6 package [22] in Fig. 13 for a the load width

em se
stress and deformation distributions are uniquely
to elastic depth ratio shown. The FE analyses were
related. By applying a unit stress to each asperity in
performed for rigid loaded areas, whereas the Stein-

ss u
turn, a vector of compliance values can be established to
brenner method assumes a uniform load (i.e. flexible
di ch relate displacements at every other asperity to the
area). The average deflections over the loaded area are
applied unit stress. A global compliance matrix can
used for comparison with the FE analyses. It can be seen
or ar
then be compiled for the entire profile. The stresses,
that the comparative deflection profiles, although not
however, are not known but must be determined. This
exactly in agreement, show reasonably close correlation
e

requires a reformulation of the problem by inversion of


for a variety of depth to load width ratios.
ng s

the compliance matrix to give a stiffness matrix-relating


yi re

Two limitations in the use of this approach are noted.


load on any asperity to a unit displacement of any other
First, the Steinbrenner method assumes an elastic layer
asperity. The relative displacements of the asperities are
op al

with horizontal upper and lower bounding surfaces.


fixed by geometrical considerations, and the absolute
Clearly, the asperity surface in this case is not
r c on

displacements can be determined using the equilibrium


horizontal. The authors adopted the average sample
equation established by the CNS condition which relates
height as the layer thickness. It is stressed, however, that
fo rs

global normal stress to dilation.


the Steinbrenner analysis is only undertaken in order to
A generalised interface profile is shown in Fig. 14 of
ot pe

establish the relative rather than absolute deformations


length, LT ; comprising n chords of equal horizontal
projected length, s: The depth of the elastic medium, D is
N r

adopted as the average depth of the rock sample, and


Fo

120 the load width, Lbs; to simulate the plane strain


conditions in a CNS direct shear test.
Percentages of loading strip deflection

Finite Element (Strand6)


Consider chord i loaded with stress, si : The stress
100
Steinbrenner Method applied to this chord alone will result in a displacement
profile such as that indicated in Fig. 15. The displace-
80 ment, rji ; of chord j due to the stress at chord i can be
evaluated from
si D
60 rji ¼ Iji ; ð18Þ
E

where the displacement influence factor, Iji is non-zero


40
within the zone of influence of chord i; and zero
elsewhere. By the form of this relationship, it can be seen
20 that this influence factor is essentially a multiplier of the
0 2 4 6 8 10 12 one-dimensional compression displacement. From con-
Distance from strip centreline (inhalf-width multiples) siderations of symmetry, it can also be shown that the
Fig. 13. Comparison of deflection profiles as determined from influence factor for the displacement of chord i due to
Steinbrenner and finite element analysis. loading of chord j is identical, i.e. Iji ¼ Iij :
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 549

2000

2.00

Shear and Normal stress (kPa)


1500

Dilation (mm)
1.00

1000

0.00
500
Normal stress (or dilation)
Shear stress
Predicted normal stress / dilation
Predicted shear stress
0

at ly
0 10 20 30 40 50
(a) Shear Displacement (mm)

in on
n
io
1600 3
Fig. 14. Generalized CNS test profile parameters used in Steinbrenner Shear stress

em se
Predicted shear stress
method.

Shear and Normal stress (kPa)


Normal stress (or dilation)
Predicted normal stress/dilation

ss u
1200 2
di ch
Assuming the zone of influence extends for k chords1

Dilation (mm)
about the loaded chord i; the displacement influence
or ar
factors can be calculated at the mid-point of chords 800 1
i  k through to i þ k: Thus a stress at chord i will cause
e

displacements from chords i  k through to i þ k:


ng s
yi re

Conversely, the total displacement of chord i; ri ; will 400 0


result from the influence of stresses on chords i  k
through to i þ k:
op al

X 0 -1
r c on

iþk
Iij sj D 0 10 20 30 40 50
ri ¼ : ð19Þ
ik
E (b) Shear Displacement (mm)
fo rs

Fig. 15. Comparison of measured and predicted responses for regular


If a uniform stress s ¼ si were applied to all chords,
ot pe

triangular joints. (a) J/C joint with 51 asperities and K ¼ 600 kPa/mm,
then the resultant one-dimensional compression of all sn0 ¼ 600 kPa. (b) J/J joint with 51 asperities and K ¼ 400 kPa/mm,
chords, r; would be given by sn0 ¼ 400 kPa.
N r

si D
Fo

r ¼ ri ¼ : ð20Þ
E
Since under these conditions, r ¼ ri ¼ rj and s ¼ From Eq. (18), it can be seen that the influence factors
si ¼ sj ; it can be shown that are multiplied by D/E to obtain the compliance factors,
Cji : The compliance factors, Cji ; are assembled in the
X
iþk X
iþk
Iij ¼ Iji ¼ 1: ð21Þ global compliance matrix, ½C; to represent the relative
ik ik influence of stress on any chord on the displacement of
any other chord, i.e. frg ¼ ½Cfsg:
That is, the sum of the influence factors pertaining to The stiffness matrix, ½S; is determined by matrix
any loaded chord must be equal to unity. It is noted that inversion in order to establish the stresses as a function
this is an artificial enforcement of Steinbrenner’s method of chord displacements, i.e.
on a finite sample. Any error in this summation due to
the effects of discretization of the profile or other fsg ¼ ½Sfrg where ½S ¼ ½C1 : ð22Þ
inaccuracies are corrected by proportional adjustment The individual chord displacements, ri ; can be
of the individual influence factors. established from ri ¼ x tan yi  c; where x is the shear
1 displacement, yi ; is the angle of chord i; and c is the
The value of k is chosen such that the influence factors of asperity
i+k are extremely small. For the CNS samples, no limit was defined, total dilation of the interface, which must be deter-
and the influence of each asperity on every other asperity was taken mined. The dilation can be established from considera-
into account. tion of equilibrium normal to the interface. The global
550 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

normal stress boundary conditions including the initial Table 3


normal stress, sn0 and the global stiffness, K give rise to Engineering properties of Johnstone
the following equation: Saturated water content 15.5%
Secant Young’s modulus (50% peak load) 360 MPa
X
n
si wi ¼ LT ðsn0 þ KcÞ; ð23Þ Poisson’s ratio 0.25
Uniaxial compressive strength 3.5 MPa
i¼1
Drained cohesion 750 kPa
where wi is the load width on chord i; si is the projected Peak drained friction angle 361
Residual friction angle 24.51
length of the chord, and wi =si  x:

3. Comparison with laboratory test results 1600


4.00
The analytical models described above have been

Shear and Normal stress (kPa)


incorporated into a computer program called Rocket

at ly
1200 3.00
[23] which allows the complex interactions to be solved

Dilation (mm)
in on
quickly. Comparisons between the computer simula-

n
tions and the laboratory tests are detailed hereafter.

io
2.00
These simulations do not purport to be ‘‘Class A’’ 800

em se
predictions [24]—they are made in hindsight, with the
tests on regular triangular asperities forming the 1.00

ss u
basis for the models which have been adopted.
di ch
However, since the various components of the models,
which cover asperity sliding, shearing and elastic
400
Normal stress (or dilation)
Shear stress
0.00
or ar
Predicted normal stress / dilation
distribution of stresses are based only on theoretical Predicted shear stress
concepts, it does not matter that the predictions are
e

0
0 10 20 30 40 50
made a posteriori.
ng s

(a) Shear Displacement (mm)


yi re

In order to carry out simulations of the direct


shear tests, the following basic input parameters are 1500 4
required:
op al

Shear and Normal stress (kPa)


r c on

* Young’s modulus and Poisson’s ratio of the rock (for 1200 3


concrete/Johnstone joints the assumption is made

Dilation (mm)
fo rs

that the concrete is rigid relative to the rock);


900 2
peak drained cohesion and friction angle of the rock;
ot pe

* residual friction angle of the rock;


* geometry of the roughness profiles, i.e. chord length 600 1
N r

and inclination of each chord; and


Fo

* test boundary conditions; i.e. initial normal stress and


normal stiffness. 300 Shear stress 0
Predicted shear stress
Normal stress (ordilation)
Predicted normal stress/dilation
The following predictions are based on the average 0 -1
rock properties listed in Table 3, and the actual 0 10 20 30 40 50
roughness profiles and boundary conditions used in (b) Shear Displacement (mm)
the tests. Typical comparisons between actual and Fig. 16. Comparison of measured and predicted responses for regular
predicted responses of selected CNS tests on regular triangular joints. (a) J/C joint with 151 asperities and K ¼ 300 kPa/
triangular asperity profiles are presented in Figs. 15 and mm, sn0 ¼ 300 kPa. (b) J/J joint with 101 asperities and K ¼ 300 kPa/
mm, sn0 ¼ 400 kPa.
16. The shear stress, normal stress and dilation vs. shear
displacement predictions are compared in a single graph
for each test in these figures. In all cases, the left-hand
ordinate is in stress units; shear stress and normal stress evaluated from a single curve (using left and right
plots can be referred to this axis. In addition, the right- ordinates, respectively).
hand ordinate is in displacement units, and the interface Although predictions have been carried out for all
dilation can be determined from this axis. Because of the tests (which include regular triangular asperities varying
relationship between normal stress and dilation in a from 51 to 27.51 and a range of stress and stiffness
CNS test, both normal stress and dilation can be conditions), only comparisons for four tests on regular
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 551

1200 1.00 1200 3.00

Normal stress (or dilation) Normal stress (or dilation)

Shear and Normal stress (kPa)


Shear stress Shear stress
Shear and Normal stress (kPa)

Predicted normal stress / dilation Predicted normal stress/ dilation


900 Predicted shear stress 0.67 900 Predicted shear stress 2.00

Dilation(mm)

Dilation (mm)
600 0.33 600 1.00

300 0.00 300 0.00

-0.33 0 -1.00
(a) 0

at ly
0 10 20 30 40 0 5 10 15 20
(a) Shear Displacement (mm)

in on
Shear Displacement (mm)

n
io
1600 3 1600 3

em se
Shear and Normal stress (kPa)

Shear and Normal stress (kPa)


Shear stress Shear stress

ss u
Predicted shear stress Predicted shear stress
1200 Normal stress (ordilation) 2 1200 Normal stress (ordilation) 2
di ch
Predicted normal stress/dilation Predicted normal stress/dilation
Dilation (mm)

Dilation (mm)
or ar
800 1 800 1
e
ng s
yi re

400 0 400 0
op al
r c on

0 -1 0 -1
0 5 10 15 20 25 0 5 10 15 20 25
(b) Shear Displacement (mm) (b) Shear Displacement (mm)
fo rs

Fig. 17. Comparison of measured and predicted responses for fractal Fig. 18. Comparison of measured and predicted responses for fractal
ot pe

joints. (a) J/C joint with Class A fractal profile and K ¼ 300 kPa/mm, joints. (a) J/C joint with Class B fractal profile and K ¼ 300 kPa/mm,
sn0 ¼ 900 kPa. (b) J/J joint with fractal profile with mean asperity sn0 ¼ 300 kPa. (b) J/J joint with fractal profile with mean asperity
angle of 51 and K ¼ 400 kPa/mm, sn0 ¼ 400 kPa. angle of 101 and K ¼ 400 kPa/mm, sn0 ¼ 400 kPa.
N r
Fo

triangular asperity profiles are included here. Similarly 4. Conclusions


close comparisons were obtained for all of the other
tests on regular asperity profiles. These figures clearly This paper describes the theoretical models developed
demonstrate that the sliding and shearing models to predict the shear behaviour of soft rock joints. The
adopted capture the essential behaviour of regular physical processes modelled include sliding between
triangular asperity profiles. rough surfaces of essentially elastic materials, asperity
Figs. 17–19 present corresponding observed and shearing, post-peak behaviour and load sharing between
predicted behaviour for fractal profiles [5] for various asperities in a complex interface.
constant normal stiffness and initial normal stress It has been shown that the analytical techniques are
conditions. All predictions are reasonably close. Close able to simulate the shear behaviour of interfaces
predictions were also obtained for the remaining fractal comprising simple triangular asperities. Furthermore,
tests. by taking account of elastic interactions between
Fig. 20 compares the measured and predicted asperities of varying angles, the shear behaviour of
peak shear stress for all tests on fractal profiles profiles with complex and random roughness can also be
involving Johnstone. Predictions lie within expected simulated with reasonable success.
experimental scatter and within 710% of the observed The simplicity of the models which underlie this
values. micro-mechanical simulation are duly acknowledged,
552 J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553

1200 3.00 however, what is significant about the results is that the
Normal stress (or dilation)
shear behaviour of complex laboratory joint profiles in
Shear stress an artificial siltstone has been predicted on theoretical
Shear and Normal stress (kPa)

Predicted normal stress / dilation


900 Predicted shear stress 2.00
grounds alone.

Dilation (mm)
600 1.00
Acknowledgements

The authors gratefully acknowledge funding of this


work by the Australian Research Council, and the
300 0.00 financial support for Dr. Seidel’s Ph.D. candidature by
the Sir James McNeil Foundation.

0 -1.00
0 5 10 15 20

at ly
References
(a) Shear Displacement (mm)

in on
n
[1] Seidel JP, Haberfield CM. Laboratory testing of concrete–rock

io
1600 3
joints in constant normal stiffness direct shear. Geotech Test J,

em se
Shear stress
2002, accepted for publication.
Shear and Normal stress (kPa)

Predicted shear stress


Normal stress (ordilation) [2] Fleuter WF. Analytical and experimental investigation into the
Predicted normal stress/dilation

ss u
1200 2 shear performance of joints in soft sedimentary rocks. MEngSc
dissertation. Australia: Department of Civil Engineering, Monash
di ch University, 1997.
Dilation (mm)

[3] Pearce H. A micro-mechanical approach to the shear behaviour of


800 1
or ar
rock joints. Ph.D. dissertation. Australia: Department of Civil
Engineering, Monash University, 2001, in preparation.
e

[4] Johnston IW, Choi SK. A synthetic soft rock for laboratory
model studies. Geotechnique 1986;36(2):251–63.
ng s

400 0
[5] Seidel JP, Haberfield CM. Towards an understanding of joint
yi re

roughness. Int J Rock Mech Rock Eng 1995;28(2):69–92.


[6] Patton FD. Multiple modes of shear failure in rock. Proc 1st
op al

0 -1 Cong Int Soc Rock Mech Lisbon 1966;1:509–13.


0 5 10 15 20 25 [7] Ladanyi B, Archambault G. Simulation of shear behaviour of a
r c on

(b) Shear Displacement (mm) jointed rock mass. In: Somerton WH, editor. Rock mechanics;
theory and practice. Proc 11th Symp on Rock Mechanics,
Fig. 19. Comparison of measured and predicted responses for fractal
fo rs

American Institute of Mining and Metallurgy, New York 1970.


joints. (a) J/C joint with Class C fractal profile and K ¼ 300 kPa/mm, p. 105–25.
ot pe

sn0 ¼ 300 kPa. (b) J/J joint with fractal profile with mean asperity [8] Lam TSK, Johnston IW. Shear behaviour of regular triangular
angle of 151 and K ¼ 400 kPa/mm, sn0 ¼ 400 kPa. concrete/rock joints—evaluation. J Geotech Eng Div ASCE
1989;115:728–40.
N r

1000 [9] Indraratna B, Haque A. Shear behaviour of rock joints.


Fo

Class A (J/C) Rotterdam: Balkema, 2000.


%

Class B (J/C) [10] Seidel JP, Haberfield CM. The application of energy principles to
0
Predicted Shear Stress (kPa)

+1

Class C (J/C)
the determination of the sliding resistance of rock joints. Int J
800 Class D (J/C)
5 deg (J/J) Rock Mech Rock Eng 1995;28(4):211–26.
10 deg (J/J) 0% [11] Lam TSK. Shear behaviour of concrete–rock joints. Ph.D.
15 deg (J/J) -1
dissertation. Australia: Department of Civil Engineering, Monash
600 University, 1983.
[12] Kodikara JK. Shear behaviour of rock–concrete joints, side
resistance of piles in weak rock. Ph.D. dissertation. Australia:
400 Department of Civil Engineering, Monash University, 1989.
[13] Kodikara JK, Johnston IW. Shear behaviour of irregular
triangular rock–concrete joints. Int J Rock Mech Min Sci
Geomech Abstr 1994;31(4):313–22.
200 [14] Johnston IW, Lam TSK. Shear behaviour of regular triangular
concrete/rock joints—analysis. ASCE J Geotech Eng 1989;
115(5):711–27.
0 [15] Nelson J. M.Sc. thesis. MIT, 1977.
0 200 400 600 800 1000 [16] Sokolovsky VV. States of soil media, 2nd ed. [Jones DH,
Measured Peak Shear Stress (kPa) Schofield AN, Trans.]. London: Butterworth Scientific Publica-
tions, 1960.
Fig. 20. Predicted vs. measured peak shear stress for J/C and J/J [17] Kreyzig E. Advanced engineering mathematics, 3rd ed. New
joints. York: Wiley, 1972.
J.P. Seidel, C.M. Haberfield / International Journal of Rock Mechanics & Mining Sciences 39 (2002) 539–553 553

[18] Giam PSK, Donald IB. Soil stability analysis by an upper bound [21] Steinbrenner W. Tafeln zur Setzungsberechnung. Die Strasse
plasticity method. Civil Engineering Research Report No. 2/1989, 1934;1:121–4.
Monash University, ISSN 0155-6282, ISBN 0 86746 914 5, 1989. [22] G and D Computing. STRAND6, Version 6.0 Manual—Elastic
[19] Donald IB, Chen Z. Slope stability analysis by the upper bound and non-linear finite element software package, 1991.
approach: fundamentals and methods. Can Geotech J 1997; [23] Seidel JP. Rocket Ver 1.0 manual. Software package for
34(6):853–62. the simulation of drilled shaft load deflection response,
[20] Haberfield CM, Johnston IW. A mechanistically based model for 1994.
rough rock joints. Int J Rock Mech Min Sci Geomech Abstr [24] Lambe TW. Predictions in soil engineering. 13th Rankine lecture.
1994;31(4):279–92. Geotechnique 1973;23(2):149–202.

at ly
in on
n
io
em se
ss u
di ch
or ar
e
ng s
yi re
op al
r c on
fo rs
ot pe
N r
Fo

Вам также может понравиться