International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Shear strength of rock joints influenced by compacted infill

Buddhima Indraratna a,n, Wuditha Premadasa a, Edwin T. Brown b,
Antonio Gens c, Ana Heitor a
a
Centre for Geomechanics and Railway Engineering, Faculty of Engineering, University of Wollongong, Wollongong, NSW 2522, Australia
b
Golder Associates Pty Ltd., Milton, NSW, Australia
c
Department of Geotechnical Engineering and Geosciences, Universitat Politècnica de Catalunya, Barcelona, Spain

art ic l e i nf o a b s t r a c t

Article history: Discontinuities such as fault planes, joints and bedding planes in a rock mass may be filled with different
Received 3 October 2013 types of fine-grained material that are either transported or accumulated as gouge due to weathering or
Received in revised form joint shearing. Previous laboratory studies have mainly examined the role of saturated infill that exhibits
6 February 2014
the minimum shear strength. However, in practice, the infill materials are often partially saturated
Accepted 28 April 2014
generating matric suction within the joint that can contribute to increased shear strength. To the
authors’ knowledge this is the first study to examine the influence of compacted (unsaturated) infill on
Keywords: the joint shear strength. A series of laboratory triaxial tests on idealised model joints and imprinted
Compacted infill natural joint profiles was carried out, with constant water contents of the infill being maintained. From
Rock joint
the laboratory results, it is observed that the peak shear strength of infilled joints increased with the
Rock mechanics
decrease of degree of saturation from 85% to 35% for both idealised joints and replicated natural joints.
Shear strength
Unsaturation Based on the laboratory observations an empirical model for describing the infilled joint shear strength
was developed.
Crown Copyright & 2014 Published by Elsevier Ltd. All rights reserved.

1. Introduction material and because of their high transmissivity, joints are often
conduits for fluid flow [2]. Laboratory testing of infill materials
Rock masses present in nature are generally characterised from a rock mass failure site at Kangaroo Valley, New South Wales,
by discontinuities such as joints, fractures and other planes Australia, confirmed that the soil can reach more than 95% of
of weakness. Discontinuities that are infilled with fine-grained saturation after a period of heavy rainfall. During dry seasons, the
material which is either transported or appears as a result of infill saturation will gradually decrease, contributing to an increase
weathering or joint shearing, will adversely affect the behaviour of in the overall shear strength of the jointed rock mass. While
the rock mass. These fine infill materials may drastically reduce studies have been carried out to investigate the behaviour of
the shear strength of the rock joints compared to an unfilled or infilled rock joints [3–9], the majority have considered either a
clean joint, because they may prevent the walls of the rock joint fully saturated infill condition or a specified saturation level..
from coming into contact during shear. Recently, Alonso et al. [10], and Zandarin et al. [11] conducted
The degree of saturation of the infill is a governing parameter of a study on partially saturated bare rock joints, but the role of infill
the shear strength of a filled joint, and it can vary noticeably, within the joint was not considered. From a practical perspective,
depending on the groundwater and climate patterns. Barton [1] most infill materials will likely be compressed over time and
carried out an extensive study of filled discontinuities in rock in remain typically in an unsaturated state, unless the joints are
which the in-situ water content of the infill was found to be a submerged by groundwater which may happen in the event of
principal parameter controlling the shear strength of a filled joint. groundwater inflows occurring through specific discontinuities. In
Furthermore, for adverse climatic conditions, i.e. heavy precipita- this instance, grouting of the joints may be considered as a method
tion and long periods of rainfall, Barton [1] has reported that the to prevent the infill materials from reaching full saturation, thus
joints may act as conduits of water, leaving the fine infill material reducing the probability of catastrophic rock slides occurring.
basically in near saturated conditions. Most rock masses contain The shear strength of a filled joint is often assumed to be that of
complex, interconnected networks of joints filled with gouge the infill material alone. While this assumption may be acceptable
if the infill thickness is higher than a certain critical value,
for smaller values of infill thickness in relation to the joint rough-
n
Corresponding author. Tel.: þ 61 2 4221 3046; fax: þ61 2 4221 3238. ness or asperities it neglects the possibility of rock-to-rock
E-mail address: [email protected] (B. Indraratna). contact taking place. In these conditions, the rock-to-rock contact

http://dx.doi.org/10.1016/j.ijrmms.2014.04.019
1365-1609/Crown Copyright & 2014 Published by Elsevier Ltd. All rights reserved.
 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307 297

influence becomes increasingly prominent. In contrast, the effect throughout shearing. Indraratna et al. [8] has shown that this thin
of infill saturation can be distinctly observed for a thicker infill layer of waterproofing sealant has negligible effect on the shearing
where the strength is governed by the infill alone. This effect resistance of the joint surface. Additional specimens replicating
decreases as the infill becomes thinner, because in such conditions the roughness profiles of natural joints were also prepared. The
the shear strength of the joint is basically governed by the shear details of the preparation method are given in the subsequent
behaviour of the surfaces of the rock. sections.
In this study, a series of constant water content undrained A silty clay (25% fine sand and 75% kaolinite) with a liquid limit
triaxial tests on idealised models of rock joints and replicated of 39 and a plastic limit of 20 was used as the infill material.
natural joints has been conducted to investigate the effect of infill The shear strength behaviour of the infill material under saturated
saturation on the shear strength of filled joints. Although the shear conditions was studied using a direct shear box apparatus
strength of soil at constant water conditions has been studied in (AS 1289.6.2.2 [15]), and a friction angle of ϕ0 ¼ 211 and a cohesion
the past [12,13], no literature is available on infilled joints tested intercept (c0 ) of 13.4 kPa were obtained. The infill material was
under unsaturated infill conditions. This study proposes a novel mixed in the laboratory to known moisture contents and then
approach for laboratory testing of infilled joints under different spread over the surfaces of the joints with a spatula. The joints
initial degree of saturation of compacted infill. were then statically compacted to a given infill thickness to
asperity height (t/a) ratio with an infill dry density of 1250 kg/
m3. Note that despite having infill specimens prepared with
varying (t/a) ratios, the initial dry unit weight remained the same,
2. Laboratory investigation
and thus different water content resulted in initial degree of
saturations varying from 35% to 85%, corresponding to a suction
2.1. Specimen preparation
varying from 860 kPa to 165 kPa (see soil water characteristic
curve later). An example of the final joint profile obtained once
Calibration of any shear strength model requires a series of
the infill was spread to a certain thickness is shown in Fig. 1c.
identical joint specimens, and this surely restricts the use of real
Although the behaviour of the joints with infill is studied for a
joint surfaces. Therefore, for reasons of simplicity and reproduci-
wide range of the infill initial degree of saturation, fully saturated
bility, idealised model rock joints with regular saw toothed
conditions were not considered. This is owing to the difficulty in
surfaces were cast with a mean dip angle of 601 [8]. The jointed
preparing viable specimens using this procedure, that is, because
specimens were 54 mm in diameter with an asperity height of
the infill became slurry-like when it approached saturation. After
2 mm and an initial asperity angle of 181. Fig. 1 shows the mould
assembly the infilled joint specimens were wrapped in a thin
used to prepare the specimens and a specimen obtained after
neoprene (impervious) membrane. During testing, the sealant
casting. Indraratna [14] proposed the use of gypsum cement
applied on the surfaces of the joint and the impervious membrane
(CaSO4  H2O hemihydrates, 98%) to model soft sedimentary rocks
ensured that the clay infilled joints maintained constant water
and to prepare idealised joints. This material is readily available
content conditions. Furthermore, as the permeability of the model
and relatively inexpensive and can be moulded into any shape
rock is much lower than that of the infill (kinfill/krock 41000), it
when mixed with water. The unconfined compressive strength is
could be assumed that undrained conditions would still prevail.
independent of time once the chemical hydration is complete. The
properties of the material depend on the gypsum cement to water
ratio used to mix it. A consistent unconfined compressive strength 2.2. Testing procedure
(sc) of 65–70 MPa was obtained for a plaster to water ratio of 7:2
by weight after two weeks of curing. After being removed from the The high-pressure two-phase triaxial apparatus (Fig. 2) devel-
mould, the plaster specimens were cured for two weeks in an oven oped at the University of Wollongong [16] was customised for this
at a controlled temperature of 40–45 1C. The plaster specimens study. The cylindrical chamber can accommodate samples of
were immersed in water for at least 72 h, and subsequently an 54 mm and 60 mm in diameter with a height-to-diameter ratio
organic waterproof sealant was applied over the surfaces of the up to 2.0. Silicon oil was used as the confining fluid as it does not
joints to ensure that the infill water content remained constant react with the steel cell or with the latex membrane. A variable

Fig. 1. (a) The mould and (b) simulated joint surface profile of the idealised saw toothed joint (c) with the infill material.
298 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307

Fig. 2. High pressure triaxial apparatus for rock joint shearing.

constant strain mechanical driving system was installed to apply a of 35%, 50%, 60%, 70% and 85% corresponding to water contents of
constant strain rate. Vertical displacement was measured using a 15%, 20%, 25%, 30%, and 35%, respectively. The stress–strain and
Linear Variable Differential Transformer (LVDT) with an accuracy dilation responses were observed and the peak shear strength
of 0.001 mm while a system with laser beams with accuracy of was considered as the maximum shear strength obtained on the
0.001 mm was used to measure the horizontal movement of the stress–strain curve. Dilation of the joint was calculated normal to
joint. All the measuring devices were connected to a data-logger the joint surface using independent measurements of axial and
which was then connected to a computer for continuous data horizontal displacements. At the end of each test, the moisture
logging. content of the infill was measured and compared with the initial
Although the testing was carried out adopting undrained value. The difference between the two measurements was typi-
conditions for the water phase, the air phase remained drained cally less than 0.1, which indicates that constant water conditions
throughout the test. The specimen of infill was tested at its initial were attained.
water content. While the change in suction could not be mon- To adequately describe the hydraulic properties of a material
itored during the compression stage, its value at the beginning of under unsaturated conditions, it is necessary to establish a
the shearing stage could be captured following the procedure relationship between suction and the amount of water present,
outlined in [17]. A strain rate of 0.01 mm per minute was adopted which can be expressed in terms of either water content also
for the test series. This shearing rate is consistent with those known as gravimetric water content (mass), volumetric water
adopted for constant water conditions in the previous studies content, or degree of saturation (volume). This relationship is often
[12,19]. referred as the soil–water characteristic curve or SWCC. The soil–
The infilled jointed specimen with known moisture content (or water characteristic curve for the infill material prepared at the
initial degree of saturation) was assembled inside the cell and the same dry unit weight was developed using test data obtained
confining pressure was applied. Three different confining pres- from the pressure plate apparatus and chilled mirror hygrometer,
sures of 300 kPa, 500 kPa and 900 kPa were used in this study to according to the standard procedure described in ASTM [19]. Two
investigate the effect of normal stress (or the confining pressure) distinct data sets obtained from the pressure plate and chilled
on shear strength. As shown in Fig. 1c, the mean joint plane is mirror hygrometer are shown in Fig. 3 for a wide range of matric
inclined at 601 to the horizontal plane, and the normal stress suction that defines the SWCC. Pressure plate apparatus was used
acting on the joint plane varies during shearing even though the to obtain data under matric suctions up to 1000 kPa and the
confining pressure is kept constant. A series of tests was carried chilled mirror hygrometer was used for the range above 1000 kPa.
out for (t/a) ratios of 0.5, 1.0 and 3.0 and a starting infill saturation The pressure plate extractor consists of a saturated high air entry
 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307 299

0.5 to 3.0). Since this was a relatively thin seam, it was not feasible
to install a pore water pressure or a suction probe inside the joint
to measure the negative pore pressures, as there was a distinct
possibility of it being damaged during shearing. It was therefore
not feasible to monitor the change in the matric suction while the
joint was being sheared. The direct shear tests conducted on
statically compacted kaolin reported by Tarantino and Tombolato
[21] show that for applied vertical stresses of 300 kPa and 600 kPa
the changes in suction monitored during direct shearing were
generally small. Limited variations of suction during the shear
stage under constant water content conditions have also been
reported by [12] and [22] for tests of the same material. Therefore,
the value of suction at the start of testing incorporated in the
current shear strength analysis is reasonable. In addition, during
the shearing process, the broken asperities commonly contam-
inate the infill material due to asperity overriding in thin infill
Fig. 3. Soil–water characteristic curve for the infill material. seams. In this instance, the final value of suction computed using
methods such as the filter paper technique may not be realistic.
Therefore, in subsequent sections of this paper the shear strength
value (HAEV) ceramic disk contained in an air pressure chamber. behaviour of the infilled joints is modelled considering the
In this study, two ceramic disks with HAEVs of 500 kPa and initial suction of the infill computed at the start of test as a refer-
1500 kPa were used. The HAEV ceramic disk was saturated and ence value.
was in contact with water in a compartment below the disk that
was connected to an exterior burette. The infill material was then
prepared in a retaining cell to a dry density of 1250 kg/m3 and was 3. Results and discussion
saturated by inundation in a tray. After saturation the sample
weight was measured and it was placed inside the pressure Forty eight undrained triaxial tests were carried out on idea-
chamber. Then the desired suction was applied to the system by lised saw-tooth joints for confining pressures of 300 kPa, 500 kPa
the pressure differential of air and water phases (i.e. ua  uw; air and 900 kPa, and at initial degrees of saturation of 35%, 50%, 60%,
pressure was increased while the water pressure was kept at 70%, 85% with the t/a ratios of 0.5, 1.0 and 3.0. Test results for 63
atmospheric pressure) that was maintained until the specimens triaxial tests on replicated natural joint surfaces were also carried
reached equilibrium. The equilibrium state can be identified by out and will be discussed later in this paper. An example of the
observing the air–water interface of the burette attached to the test results obtained from the undrained triaxial testing of infilled,
water compartment of the pressure chamber. Once the equili- idealised rock joints is presented and discussed here. Fig. 4
brium state was established (typically after seven days or when illustrates the selected plots of deviator stress and dilation against
the air–water interface changed less than 1 ml over a period of axial strain for varying values of infill thickness and initial degrees
48 h), the pressure chamber was opened and the specimens and of saturation for the specimens tested under a confining pressure
their retaining rings weighed to determine the corresponding of 900 kPa for t/a ratios varying from 0.5 to 3.0. For the purpose of
water content. The specimens were then placed back on the brevity, only the typical results for the degree of saturation of 50%
porous plate and the next increment of suction was applied by and 85% are shown in Fig. 4(a) and (b). The associated dilation was
increasing the air pressure. The process was continued until the measured as a function of the normal displacement of the joint
required matric suction increment was reached. with respect to the shear plane, considering both the horizontal
For the matric suction values exceeding the air entry values of the and vertical components of the joint displacement vector, as
pressure plate ceramic disk, alternative techniques such as the chilled presented in Fig. 4(c) and (d). The shear stress of an infilled joint
mirror hygrometer can be used. Note that while the chilled mirror depends mainly on the thickness of the infill. When the infill is
hygrometer apparatus measures total suction, the osmotic component comparatively thin (i.e. t/a ¼0.5, 1.0), the stress–strain plot exhibits
should be negligible as there are no salts present in the infill material. two distinct stages of change in curvature (Indraratna et al. [8]).
The chilled mirror hygrometer consists of sealed chamber with a fan, a The first stage corresponds to the yielding of the soil infill. Beyond
mirror, a photoelectric cell, and an infrared thermometer. In this this stage asperity interference starts to prevail causing a signifi-
case, the infill material is placed in a plastic container with a diameter cant increase in dilation. This is particularly evident in the t/a ¼0.5
of 40 mm, which is then placed on a tray and inserted in to a temper- data shown in Fig. 4 for an axial strain of approximately 0.5%.
ature controlled chamber at which point, the infill material reaches The second stage corresponding to the peak deviator stress is
equilibrium with the chamber environment in a matter of minutes. largely influenced by rock–rock contact (albeit some effect of the
Inside the chamber the temperature is reduced until it the material infill is still present). This double-stage phenomenon becomes less
reaches its dew-point. At this time, the chilled mirror hygrometer pronounced as the infill increases in thickness. When t/a reaches
computes the total suction based on the vapour pressures and a value of 3.0, there is no noticeable double-stage response
temperature inside the chamber using Kelvin's equation. Prior to the in deviator stress, which implies that the shear behaviour is
suction measurements the device was calibrated using a solution of largely controlled by the soil infill alone. This is also confirmed
0.5 M KCL provided by the manufacturer (Decagon Devices). by the dilation response since fully compressional behaviour was
The data obtained by the two methods (i.e. pressure plate and observed with thicker infill. When the normal stress or confining
chilled mirror hygrometer) was interpolated using the well-known pressure increased from 300 kPa to 900 kPa, the deviator stress
van Genuchten [20] relationship fitted to the experimental data increased as expected showing a higher peak shear stress.
using the least square method. The interpolated best fit para- When the degree of saturation of the infill increased from 50% to
meters are shown in Fig. 3 along with the experimental data. 85%, the deviator stress was reduced and the axial strain required to
The silty clay seam inside the rock joint that was tested varied reach peak deviator stress increased slightly (Figs. 4 and 5). In addition,
from 1 mm to 6 mm in thickness (corresponding to t/a ratios of the volumetric behaviour observed is mainly compressive when the
300 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307

Fig. 4. Shear behaviour of idealised infilled joints with different initial degrees of infill saturation under confining pressure of 900 kPa.

degree of saturation of the infill is relatively high whereas specimens

with a lower degree of saturation exhibit predominantly dilative
behaviour. This is owing to the fact that when matric suction increases,
the infill becomes stiffer (compressibility reduced) as ‘bonding’ of infill
material particles is enhanced. Similar experimental observations have
been made in constant water content tests of residual soil [12]. The
predominately dilative behaviour observed after the initial compres-
sion is due to both asperity interference (i.e. local sliding and over-
riding) and some asperity degradation.
To better understand the shear strength behaviour of the infill
material alone, a series of undrained triaxial tests was conducted
on specimens compacted with different initial water contents
yielding the same dry densities and initial degrees of saturation
adopted for the joint testing. As expected, the peak shear stress of
the silty-clay infill material was observed to decrease with an
increase in the initial degree of saturation (S0). In Fig. 5 the
behaviour of infill alone is contrasted to that of the infilled joints. Fig. 5. Comparison of shear behaviour of infilled rock joints and infill alone at a
confining pressure of 900 kPa.
It is observed that the ductile infill specimens continue to shear at
high axial strains to attain the peak deviator stresses (4 15%), in
comparison with the relatively brittle infilled joints (t/a ¼0.5 and In addition, the specimens of infill alone showed significant
1.0) that attain their peak deviator stresses at axial strains of less bulging without producing a distinct slip surface or shear band.
than 5%. For the infilled joint specimens, large post-peak strains
could not be accommodated in the high pressure triaxial equip-
ment in order to prevent damage to the thin latex membrane. As 4. Theoretical background
shown in Fig. 5 for t/a¼ 3.0, the infilled joints behave in a ductile
manner similar to that of the infill alone. However, the influence of The thickness of the infill is the most important parameter
the rock interfaces is still present producing higher peak stresses controlling the shear strength of an infilled joint. Several appro-
than for the infill alone. The large difference in size between the aches have been taken to investigate the effect of the infill
cylindrical triaxial specimens of infill material and the relatively thickness. In most of these studies the infill thickness (t) is usually
thin seam of infill sheared between the joint walls within narrow normalised by the height of the joint asperity (a), giving the t/a
and constrained boundaries could contribute to this difference. ratio to account for the effect of joint roughness. These models are
 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307 301

5. Shear strength model with unsaturated infill

Even though the shear strength of an infilled joint is primarily

governed by the infill thickness, the infill properties such as the
degree of saturation can also be expected to have an influence on
its strength. It is possible that the dilation of the joint and the
critical t/a ratio will also be influenced by the extent to which the
infill has been saturated. To incorporate the effect of infill satura-
tion, the peak shear strength model proposed earlier in [7] is
modified in this study. For a general soil-infilled joint, the peak
unsaturated shear strength (τp,unsat) can be expressed as:
τp;unsat ¼ A0 þ B0 ð4Þ
0 0
where A is the strength contribution of the joint surfaces and B is
the strength contribution of the infill material.
Function A0 will have an optimum value equivalent to the shear
strength of a clean joint when there is no infill in the joint, and a
minimum value of zero where there is no contact with the rock
wall. Function B0 increases from zero to its maximum value (i.e. the
shear strength of the infill alone) with infill thickness increasing
up to the critical t/a ratio (Fig. 7). The change in degree of
saturation of the infill only affects the shear strength of the infill
Fig. 6. Conceptual normalised shear strength model for infilled rock joints (modified material, and therefore will not impact on function A0 which will
from [7]).
remain exactly the same as in the conceptual model (Eq. (1)) of
Indraratna et al. [7]. In contrast, function B0 needs to be modified
based on a concept of a reduction in the shear strength with an as it models the variation in the shear strength of the infill
increasing t/a ratio [3,6–9,16,23,24]. Indraratna et al. [7] proposed material which will vary with the degree of infill saturation. The
a conceptual model to describe the peak shear strength of a soil- shear strength of a partially saturated soil includes a strength
infilled joint (Fig. 6). According to this model, the peak shear contribution from the matric suction in addition to the strength of
strength of an infilled joint is the sum of two algebraic functions A a soil in a fully saturated state. The shear strengths of partially
and B which are assumed to provide a normalised shear strength saturated soils have been studied in the past [18,26–28], and these
for ratios of infill thickness to asperity height (t/a) that are less studies have proposed a range of models. A comprehensive review
than the critical value of the ratio of infill thickness to asperity of unsaturated soil models has been presented in Ref. [29]. The
height (t/a)cr. The relevant shear strength parameters are given by: model proposed in [27] predicts the shear strength using the soil–
water characteristic curve. The revised model capturing the degree
A ¼ tan ðϕb þ iÞð1 kÞα ð1Þ
of saturation can be expressed by:
  
 β S S
2 τp;unsat ¼ c0 þ ðsn  ua Þ tan ϕ0 þ ðua  uw Þ ð tan ϕ0 Þ r res ð5Þ
B ¼ tan ϕfill ð2Þ 100  Sres
1 þ1=k
where τp,unsat is the shear strength of an unsaturated soil, c0 is
 β the effective cohesion of saturated soil, (sn  ua) is the net normal
τp 2
¼ A þ B ¼ tan ðϕb þ iÞð1  kÞα þ tan ϕfill ð3Þ stress, ϕ0 is the effective angle of shearing resistance for a
sn 1 þ 1=k
saturated soil, (ua  uw) is matric suction of the soil, Sr is the
where ϕb is the basic friction angle of the joint surfaces, i is the degree of saturation of the soil, and Sres is the residual degree of
initial asperity angle of the undulations, ϕfill is the friction angle saturation of the soil.
of the infill, k ¼ ðt=aÞ=ðt=aÞcr , (t/a)cr is the critical t/a ratio of the
joint, sn is the normal stress and α and β are empirical constants
defining geometric loci of the functions A and B.
For rough joints without infill, the normalised shear strength is
equivalent to tan ðϕb þ iÞ, as proposed by Patton [25] for clean
joints. Function A models the decrease in the influence of rock-
to-rock contact with an increasing (t/a) ratio, while Function B
gradually increases the frictional component of the infill strength
until the critical thickness to asperity height ratio (t/a)cr is reached
[7]. Depending on the critical t/a ratio, two main regions can be
identified in the model: the ‘zone of interference’ (t/a o(t/a)cr) and
the ‘zone of non-interference’; t/a 4(t/a)cr (Fig. 6). In the non-
interference zone the shear strength is governed by infill only
as no rock-to-rock contact occurs during shearing. Therefore, the
normalised shear strength is given by the peak friction angle of the
infill, tan ϕfill , in that zone.
In this model, the infill was not considered to be partially
saturated in which case the matric suction plays a significant role.
Moreover, the cohesion of the infill was ignored in [7]. The model
proposed in the present study considers both the cohesion and the
effect of matric suction for unsaturated infill material. Fig. 7. Extended peak shear strength model including the effect of infill saturation.
302 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307

The conceptual development of the shear strength model for ratios of 0.5 to 4.0. With close observation of the shear strength
partially saturated infilled joints is illustrated in Fig. 7. This model data with corresponding t/a ratios, it was noted that the critical t/a
defines two different zones distinguished by the thickness of the differs with the tested water contents. The variation of the critical
infill, the ‘interference zone’ and the ‘non-interference zone’. For t/a ratio with the initial degree of saturation of the infill is
convenience in modelling, a ratio ks is introduced so that the illustrated in Fig. 8 for wetting conditions. These critical values
boundaries of the zone can be clearly expressed, i.e. ks ¼(t/a)s/(t/ were obtained from the diminishing trends of peak shear stress
a)cr,s, where (t/a)cr,s is the critical t/a ratio of an infilled joint with a with increasing t/a ratio, for instance, as shown in Fig. 9. Idealised
degree of saturation of s, and (t/a)s is the t/a ratio of a given infilled joints tested with critical t/a values of 1.7 to 2.5 gave a slightly
joint with an infill saturation of s. non-linear relationship (3rd order polynomial) between the cri-
For ks o1 in the ‘interference zone’, the function A0 based on tical t/a ratio and the initial degree of saturation of the infill, for
rock joint surfaces is given by: the range of t/a values tested.
As (t/a)cr varies with the initial degree of saturation for the
A0 ¼ sn tan ðϕb þ iÞð1  ks Þα ¼ sn A ð6Þ
infilled joint, the parameter ks was introduced to better describe
where α is an empirical constant and A is the original parameter the interference and non-interference zones. Asperity interference
from Eq. (1). occurs when the infill thickness ratio is less than the critical
The unsaturated shear strength equation given by Eq. (5) is thickness ratio (i.e. ks o1). When ks 41, the shear behaviour is
used to develop the function B0 . The algebraic component, governed solely by the infill material. The decrease in peak shear
0
c0 þ ðua  uw Þ tan ϕ ½ðSr  Sres Þ=ð1  Sres Þ of Eq. (5) is a constant for strength with increasing t/a ratio is represented quite well by the
a given initial degree of saturation of the infill according to the soil ‘decay function’ introduced in Eqs. (6) and (9). As shown by Fig. 9,
water characteristic curve. This will be referred to here as the total the laboratory test data for the silty clay-infilled saw toothed joints
cohesion intercept (ct) given by: with different initial degrees of saturation verifies the mathema-
  tical model represented by Eqs. (10)–(12). The algebraic function A0
0 Sr  Sres
ct ¼ c0 þ ðua  uw Þ tan ϕ ð7Þ represents the decay of the maximum joint friction, while function
1  Sres
B0 models the increasing role of the infill angle of friction. The
Eq. (5) can then be re-written as: empirical parameters α and β for different initial degrees of
τp;unsat ¼ ct þ ðsn  ua Þ tan ϕ0 ð8Þ saturation determined using a regression analysis are tabulated
with the respective critical t/a ratios in Table 1.
0
The second function B , influenced by infill is given by The proposed peak shear strength model allows the shear
 β strength of a given combination of infill-joint profile at any t/a
0 2
B0 ¼ ct þ ðsn  ua Þ tan ϕ ð9Þ ratio and initial degree of saturation to be determined, as long as
1 þ ð1=ks Þ
the empirical constants α and β in Eqs. (6) and (9) can be
If the infill is saturated,B0 ¼ sn  B þ ct , where B is the original evaluated from the results of undrained laboratory tests. The
parameter in Eq. (2). For ks ¼0, there is no infill material in the required values of the basic friction angle of the joint surfaces
joint and the equation is simplified to the shear strength of a clean and the properties of the infill material can be determined by
joint, laboratory tests, if estimates cannot be made from the available
literature. The initial matric suction and residual degree of
τ ¼ sn tan ðϕb þ iÞ ð10Þ
saturation can be obtained using the soil–water characteristic
For the rock-infill interference zone (0 oks o1), curve of the infill material.
 β
2
τ ¼ ðsn  ua Þ tan ðϕb þ iÞð1  ks Þα þ ct þðsn  ua Þ tan ϕ0
1 þ ð1=ks Þ
ð11Þ
For the non-interference zone (ks 41), where the shear
strength is governed only by the infill material,
τ ¼ ct þðsn  ua Þ tan ϕ0 ð12Þ
The above equations can be simplified when the infill is fully
saturated (i.e. ua  uw ¼0). In this case Eqs. (11) and (12) revert
back to those proposed earlier by [7], except for the addition of the
cohesion term.

6. Verification of the shear strength model

The peak shear strength of a joint was taken as the point at

which the maximum deviator stress was reached. It was noted
that when the t/a ratio increased, the peak shear strength
decreased rapidly until a critical t/a was reached, beyond which
any further decline in strength was insignificant. This observation
leads us to hypothesise that beyond this critical ratio (t/a)cr the
shear strength is exclusively a function of the infill properties [8].
The results of 48 undrained triaxial tests conducted on silty
clay-infilled saw toothed joints were used to validate the shear
strength model presented earlier. Test data for five different initial
degrees of saturation were collected for joints having infills that
varied in thickness from 1 mm to 8 mm, corresponding to t/a Fig. 8. Variation of critical (t/a) ratio with initial degree of saturation.
 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307 303

Fig. 10. (a) Natural joint profile used for laboratory testing, (b) section A–A: (arrow
indicates the shearing direction), (c) section B–B.

surface. The roughness amplitude or asperity height of the natural

joint was determined according to the ISRM [30] recommenda-
tions. Infill thickness (t) was measured at least at ten points along
the surface to obtain the mean infill thickness. Infill thickness to
asperity height ratio (t/a) was then determined using the mean
values of asperity amplitude and infill thickness. A series of
constant water content triaxial tests was conducted on infilled
joints with natural roughness similar to that of the idealised filled
joints at confining pressures of 300 kPa, 500 kPa and 900 kPa for
(t/a) ratios of from 0.26 to 2.05, and water content from 35% to
85%, giving a total of 63 tests. Since the shear strength and
volumetric data for all confining pressures showed similar trends,
only the results for a confining pressure of 900 kPa are given here
as an example (Fig. 11).
Compared to the idealised shear strength data, a key feature of
the replicated natural joint testing is that the laboratory data does
not show a distinct peak deviator stress even after 5% strain. This is
because the shearing of the triangular asperities in the idealised
joints corresponds to a noticeable drop in the shear stress, while
Fig. 9. Verification of shear strength model with experimental data for infill
saturations of (a) 50%, (b) 85%. the spatially distributed asperities of the natural joint profile do
not show this marked strength drop. At lower normal stresses, the
asperities can mobilize the peak strength by sliding over each
Table 1
Empirical constants and critical t/a ratios for different infill saturations for idealised other. When the steepest asperity is sheared off, the contacts will
saw-tooth joints. be moved to the next steepest asperity, and this process will
continue until a condition is reached where both sliding and
Degree of saturation (t/a)cr α Β shearing of asperities occur simultaneously [31]. However, if the
(water content, %)
infill thickness is higher than a critical value, the natural joints
35% (15%) 1.7 1.2 2.5 follow a similar trend to idealised joints, when the shear beha-
50% (20%) 1.9 1.4 2.2 viour is governed by the infill alone. For the natural joint
60% (25%) 2.1 1.6 1.9 with JRC ¼10–12, the (t/a)cr values varied from 1.0 to 1.4 (Fig. 8).
70% (30%) 2.3 1.8 1.7
It is observed that the critical t/a ratios for both idealised joints
85% (35%) 2.5 2.0 1.5
and natural joints with JRC ¼10–12 show a slightly non-linear
relationship (3rd order polynomial) with the initial degree of
7. Shear behaviour of replicated natural joints and saturation for the same infill material. At any given confining
model application pressure, the critical t/a ratios for idealised and natural joint
profiles are different. This is not surprising, because these ratios
The idealised joint profile used to develop the shear strength are governed by the joint characteristics, especially the heights
model has obvious limitations when compared to natural joint and distributions of the asperities. Idealised saw-tooth joints
behaviour. This section compares the relative differences briefly, consisted of an asperity height of 2 mm giving critical t/a ratios
and offers an extension of the idealised joint model that can be of from 1.7 to 2.5, corresponding to an infill thickness varying from
applied to natural joint profiles having a joint roughness coeffi- about 3.4 mm to 5.0 mm. In contrast, the replicated natural joints
cient (JRC) of 10–12. These joints were obtained from the rock of irregular profile had a mean asperity amplitude approaching
slide at Kangaroo Valley, New South Wales, and were replicated on 4 mm (i.e. almost double that of idealised joints for similar infill
to gypsum plaster using silicone rubber (Fig. 10). The samples were thickness), corresponding to critical t/a ratios in the narrower
54 mm in diameter (as were the idealised specimens) and a mean range of 1.0–1.4. It is also noteworthy that for the silty-clay infill,
asperity amplitude (a) of 3.91 mm was measured from the joint the critical t/a ratios (Tables 1 and 2) are mainly dependent on the
304 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307

Fig. 11. Shear behaviour of replicated natural infilled joints with different initial degrees of infill saturation under confining pressure of 900 kPa.

joint profile and the degree of infill saturation, and less dependent Table 2
on the applied confining pressure. Empirical constants and critical t/a ratios for different infill saturations for
replicated natural joints.
The peak shear strength of a filled joint having a natural profile
can be predicted using the model developed for idealised joints Degree of saturation (t/a)cr α β
with a subtle extension. For a general soil infilled joint with (water content, %)
partially saturated infill, the peak unsaturated shear strength (τp,
35% (15%) 1.0 1.3 3.0
unsat) can be expressed as:
50% (20%) 1.1 1.5 2.6
60% (25%) 1.2 1.7 2.3
τp;unsat ¼ A″ þ B″ ð13Þ 70% (30%) 1.3 2.0 1.8
85% (35%) 1.4 2.2 1.5
where, A″ is the strength contribution of the joint surfaces and B″
is the strength contribution of the infill material.
The function A0 in the idealised model that reflects Patton's [25] profile as:
equation for a clean idealised joint can be modified by introducing
  
the joint roughness characteristics proposed by Barton and Chou- JCS α
A″ ¼ sn tan ϕb þ JRC log ð1  ks Þ ð15Þ
bey [32]. For an irregular clean joint Barton and Choubey [32] sn
proposed a model to predict its peak shear strength using the Joint
roughness coefficient: where JRC is the Joint roughness coefficient and JCS is the joint
   wall compressive strength.
JCS Function B″ is only dependent on soil parameters and it is not
τp ¼ sn tan ϕb þ JRC log ð14Þ
sn dependent on the joint surface profile. Hence, the function B″
would be the same as function B0 of the original model.
where, τp is the peak shear strength, sn is the normal stress, JRC is
the joint roughness coefficient, JCS is the Joint wall compressive  β
0 2
strength and ϕb is the basic friction angle. This equation gives a B″ ¼ ct þ ½ðsn  ua Þ tan ϕ  ð16Þ
1 þ 1=ks
better application than to Patton's [25] equation as the term i in
Patton's equation does not account for the irregularity of the joint where
profile.  
0 Sr Sres
For ks o1 in the ‘interference zone’, the function A0 of the ct ¼ c0 þ ðua  uw Þ tan ϕ ð17Þ
idealised model is extended to A″ to capture the natural joint 1 Sres
 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307 305

Fig. 13. (a) Rock slope with a soil infilled discontinuity, (b) factor of safety with the
degree of saturation of the infill and initial matric suction.

sandy silt and/or sandy clay depending on the elevation above the
ground level below the slope. The strata close to the ground level
were significantly wetter than the upper slopes. More information
Fig. 12. Validation of shear strength model with experimental data for replicated on Kangaroo Valley jointed rock is given by Ref. [33]. In the current
natural joint profiles for infill saturations of (a) 50%, (b) 85%. analysis, an irregular rock joint of JRC ¼11 defining a wedge of
   weight W (simplified to 2D plane strain) with a slope angle of λ
JCS α
τp;unsat ¼ sn tan ϕb þ JRC log ð1  ks Þ þ ct and height H was considered, having a sediment-infilled joint at a
sn dip angle of θ (Fig. 13a). For a unit length normal to the plane of
 β
0 2 Fig. 13(a), the weight of the wedge W with unit weight γ can be
þ ðsn ua Þ tan ϕ ð18Þ
1 þ ð1=ks Þ simplified to:

For the non-interference zone (ks 4 1), where the shear W ¼ 0:5γ H 2 ðcot θ  cot λÞ ð20Þ
strength is governed only by the infill material, Considering a surcharge load F applied to the upper boundary
τp;unsat ¼ ct þðsn  ua Þ tan ϕ0 ð19Þ of the wedge, the factor of safety (FS) for sliding can be found by
limit equilibrium as:
Fig. 12 shows the validation of the model over an irregular
(natural) joint profile with JRC of 10–12 for different degrees of τH= sin θ
FS ¼ ð21Þ
infill saturation. It can be observed that the proposed model (Eqs. W sin θ þ F sin θ
(18) and (19)) gives a good agreement with the laboratory triaxial where, τ is the resisting shear stress along the joint described by
data obtained for natural joint replicas. The empirical parameters Eq. (18). The geometry of this particular rock wedge is approxi-
α and β for different degrees of saturation were determined by mated by H¼ 30.5 m, γ ¼ 27.5 kN/m3, λ ¼ 80o and θ ¼ 45o [33]. As
multiple regression analysis similar with the idealised joints and the length of the wedge is large compared to its width and height,
are presented in Table 2. the wedge stability can be conveniently analysed assuming 2D
(plane strain).
The clay-infilled joint model for a t/a ¼0.7 with the correspond-
8. Practical implications: Jointed rock slope stability ing model parameters given in Table 2 was used to calculate the FS
for different infill saturation of the wedge subjected to a surcharge
The use of the proposed model for natural joint profiles in a load F¼245 kN applied by the passage of a typical freight train.
practical situation is illustrated using a simplified slope analysis The value of FS increased significantly from 1.07 to 1.34 with the
problem as presented in Fig. 13(a), where a potentially unstable decrease of infill saturation from 85% to 35% (Fig. 13b). If the joint
rock wedge from Kangaroo Valley, New South Wales, Australia, is infill is fully saturated, the wedge is unstable as FS approaches
considered with simplified boundaries. The rock strata consisted unity, and the joint plane would require stabilisation, for example
mainly of sandstone and the joint infill was primarily composed of by rock bolting. In the case of relatively dry infill (Sr o40%), the
306 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307

wedge would have a FS 41.3. Not surprisingly, the FS of the wedge rock contact. Due to asperity interference, joint dilation can be
becomes asymptotic between the upper bound at which the infill observed in the interference zone. The ratio (t/a)cr varies with the
is totally dry and the lower bound where the infill is fully initial degree of saturation of the infill. It ranges from 1.7 to 2.5
saturated. This simplified example of the Kangaroo Valley rock as the initial degree of saturation increases from 35% to 85%.
wedge demonstrates the benefits of unsaturated infill existing For relatively thick infills (ks 41), the influence of the asperities is
within joints compared to the conservative assumption of fully suppressed and shearing takes place only through the infill.
saturated joints that is often adopted in the design of jointed rock Therefore, no dilation was observed in the laboratory for the
slopes. In actual practice, the sandy clay infill in Kangaroo Valley non-interference zone.
had degrees of saturation varying for 100% in the lower most Additional tests were performed to evaluate the similarity of
terrain to 40% in joints at higher elevations of the sandstone rock the shear strength data obtained with idealised saw-tooth joints to
stratum. replicated natural joints (JRC ¼10–12). The results indicated that
similar trends are obtained for both types of joints except that a
clear second stage is absent for the natural joints. This is attributed
9. Model limitations to the spatial variation of asperities over the natural profile. As for
idealised joints, the peak shear strength of natural joints decreases
The use of idealised saw toothed joint profiles in the experi- with an increase in the initial degree of saturation Also, for the
mental program was justified by the need to carry out repetitive, same variation of degree of saturation, the critical (t/a) ratio varies
reproducible tests with simple geometric profiles, in order to from 1.0 to 1.4 in a similar manner to the idealised joints. For both
understand and formulate a conceptual model. Even though idealised and natural joints, critical (t/a) ratios show a slightly
meaningful data could be obtained from these idealised profiles, non-linear relationship with the degree of saturation. The current
they do not accurately represent the shear behaviour of natural study proposed a conceptual model initially developed for idea-
joint surfaces that may be irregular and or wavy. Although some lised joints with varying degrees of saturation, that could be
steps have been taken by the authors to replicate and test natural conveniently modified to suit the natural joint profile of JRC ¼
joint profiles, there are still limitations due to the narrow range of 10–12. In addition, the stability study of potential rock wedge of
JRC (10–12) examined in this study. Therefore, further testing of Kangaroo Valley showed that the degree of saturation of the infill
different irregular joint profiles with different natural infills is governs the stability, highlighting important practical implications
required to validate the proposed models more comprehensively. of compacted infill rock joints. While this study provides insightful
Scale effects (the effects of changes in joint surface wave length results of the behaviour of infill saturation on the overall rock joint
and asperity height) were not studied. shear strength behaviour further studies with different infill
The effects of different infill material, parent rock type and the materials that may yield different water retention characteristics
extent of joint weathering were not considered in this study. are recommended.
The roughness of the joints were characterised using JRC
considering the 2D roughness only. An additional design para-
meter should be introduced for a comprehensive 3D shear analysis Acknowledgement
of rock joints.
The joints with silty clay infill alone cannot represent the entire
The Authors wish to acknowledge the contributions of Alan
range of joint types and infill materials represented in nature.
Grant (University of Wollongong) during the laboratory testing
Despite the advances the proposed model has demonstrated over
program. The second author would like to thank the Endeavour
previously existing models of infilled joints by capturing the role
Postgraduate Awards program for their financial support. The
of infill unsaturation, its application is still constrained by the
previous research in this area conducted by former PhD candi-
limitations summarised above.
dates at University of Wollongong (A. Haque, H.S. Welideniya,
M. Jayanathan and D.A.F. Oliveira) is gratefully acknowledged.
Support from the ARC Centre of Excellence for Geotechnical
10. Conclusions
Science and Engineering is also acknowledged.

This paper has discussed the influence of the initial degree of

saturation of the infill and the ratio of infill thickness to asperity References
height (t/a) on the shear strength of idealised saw toothed and
natural silty clay-infilled joints. The experimental results obtained [1] Barton N. A review of the shear strength of filled discontinuities in rock.
highlight how the t/a ratio contribute to a reduction of the shear Norwegian Geotechnical Institute Publication 1974.
strength from the maximum value associated with clean rough [2] Elsworth D. A model to evaluate the transient hydraulic response of three‐
dimensional sparsely fractured rock masses. Water Resourc Res 1986;22:
joints. The shear strength behaviour is governed by t/a up to a 1809–19.
critical thickness after which there is no significant change in [3] Ladanyi B, Archambault G. Shear strength and deformability of filled indented
strength. Further, for infill thickness exceeding this critical thick- joints. In: Proceedings international symposium on the geotechnics of struc-
turally complex formations. Capri. p. 317-326, 1977.
ness, the influence of asperities is suppressed and the shear [4] Lama RD. Influence of clay fillings on shear behaviour of joints. In: Proceedings
behaviour is mainly governed by the infill. third international association engineering geology. Madrid. p. 27-34, 1978.
The effect of initial degree of saturation of the infill material [5] Pereira JP. Mechanics of filled discontinuities. In: Proceedings international
conference on mechanics of jointed and faulted rock. p. 375-380, 1990.
also has an important effect on the shear strength. Indeed, it was [6] de Toledo PEC, de Freitas MH. Laboratory testing and parameters controlling
observed that as the degree of saturation increases, both the shear the shear-strength of filled rock joints. Géotechnique 1993;43:1–19.
strength and the joint dilation decrease. The effect of the confining [7] Indraratna B, Welideniya HS, Brown ET. A shear strength model for idealised
infilled joints under constant normal stiffness. Géotechnique 2005;55:215–26.
pressure is also shown by the test data obtained under three
[8] Indraratna B, Jayanathan M, Brown ET. Shear strength model for overconso-
different confining pressures, 300 kPa, 500 kPa, and 900 kPa. lidated clay-infilled idealised rock joints. Géotechnique 2008;58:55–65.
There is an apparent increase in shear strength when the confining [9] Indraratna B, Oliveira DAF, Brown ET. A shear-displacement criterion for soil-
pressure is increased. When the thickness of the infill is very low infilled rock discontinuities. Géotechnique 2010;60:623–33.
[10] Alonso E, Zandarín M, Olivella S. Joints in unsaturated rocks: thermo-hydro-
(ks o 1), the stress–strain behaviour is influenced by asperity mechanical formulation and constitutive behaviour. J Rock Mech Geotech Eng
interference and the peak shear strength is governed by rock-to- 2013;5:200–13.
 B. Indraratna et al. / International Journal of Rock Mechanics & Mining Sciences 70 (2014) 296–307 307

[11] Zandarin MT, Alonso E, Olivella S. A constitutive law for rock joints considering [23] Phien-wej N, Shrestha U, Rantucci G. Effect of infill thickness on shear
the effects of suction and roughness on strength parameters. Int J Rock Mech behaviour of rock joints. In: Proceedings of the international conference on
Min Sci 2013;60:333–44. rock joints, Loen, Norway. Rotterdam: Balkema: 289-294, 1990.
[12] Thu TM, Rahardjo H, Leong E-C. Shear strength and pore-water pressure [24] Papaliangas T, Hencher SR, Lumsden AC, Manolopoulou S. The effect of
characteristics during constant cater content triaxial tests. J Geotech Geoen- frictional fill thickness on the shear strength of rock discontinuities. Int J
viron Eng 2006;132:411–9. Rock Mech Min Sci Geomech Abstr 1993;30:81–91.
[13] Hamid TB, Miller GA. Shear strength of unsaturated soil interfaces. Can [25] Patton FD. Multiple modes of shear failure in rock and related materials. PhD
Geotech J 2009;46:595–606. thesis. University of Illinois; 1966.
[14] Indraratna B. Development and applications of a synthetic material to [26] Bishop AW, Alpan I, Blight GE, Donald IB. Factors controlling the shear
simulate soft sedimentary rocks. Géotechnique 1990;40:189–200. strength of partly saturated cohesive soils. In: Proceedings, ASCE research
[15] Standards Australia. Soil strength and consolidation tests – determination conference on shear strength of cohesive soils. University of Colarado.
of the shear strength of a soil – direct shear test using a shear box. AS p. 503-532, 1960.
12896221998. [27] Vanapalli SK, Fredlund DG, Pufahl DE. The relationship between the soil–water
[16] Indraratna B, Haque A, Aziz NI. Shear behaviour of idealised infilled joints characteristic curve and the unsaturated shear strength of a compacted glacial
under constant normal stiffness. Géotechnique 1999;49:331–55. till. Geotech Test J 1996;19:259–68.
[17] Oloo SY, Fredlund DG. A method for determination of phi(b) for statically [28] Sheng D, Fredlund DG, Gens A. A new modelling approach for unsaturated
compacted soils. Can Geotech J 1996;33:272–80. soils using independent stress variables. Can Geotech J 2008;45:511–34.
[18] Fredlund DG, Rahardjo H, Fredlund MD. Unsaturated soil mechanics in [29] Gens A. Soil–environment interaction in geotechnical engineering. Géotech-
engineering practice. 1st ed.. New York, NY: Wiley-Interscience; 2012. nique 2010;60:3–74.
[19] ASTM. Standard test methods for determination of the soil water chararcter- [30] ISRM (International Society for Rock Mechanics). Suggested methods for the
istic curve for desorption using a hanging column, pressure extractor, chilled quantitative description of discontinuites in rock masses. In: Brown ET, editor.
mirror hygrometer, and/or centrifuge. ASTM international. West Consho- Rock characterisation, testing and monitoring—ISRM suggested methods,
hocken, PA; 2008. Pergamon: Oxford; 1981. p. 319–368.
[20] van Genuchten MT. A closed-form equation for predicting the hydraulic [31] Kimura T, Esaki T. A new model for the shear strength of rock joints with
conductivity of unsaturated soils. Soil Sci Soc Am J 1980;44:892–8. irregular surfaces. In: . Proceedings international conference mechanics of
[21] Tarantino A, Tombolato S. Coupling of hydraulic and mechanical behaviour in jointed and faulted rock. Rotterdam: Balkema, p. 133-137, 1995.
unsaturated compacted clay. Géotechnique 2005;55:307–17. [32] Barton N, Choubey V. The shear strength of rock joints in theory and practice.
[22] Rahardjo H, Heng OB, Choon LE. Shear strength of a compacted residual soil Rock Mech Rock Eng 1977;10:1–54.
from consolidated drained and constant water content triaxial tests. Can [33] Indraratna B, Haque A. Shear behaviour of rock joints. Rotterdam: Balkema;
Geotech J. 2004;41:421–36. 2000.