Laboratory Estimation of Rock Joint Stiffness and Frictional Parameters
Laboratory Estimation of Rock Joint Stiffness and Frictional Parameters
Laboratory Estimation of Rock Joint Stiffness and Frictional Parameters
net/publication/292338937
CITATIONS READS
29 1,028
6 authors, including:
Some of the authors of this publication are also working on these related projects:
Development of a new rock mass strength criterion for rock masses having approximately orthogonal fracture systems (supported by CDC-NIOSH) View project
All content following this page was uploaded by Srisharan Shreedharan on 19 October 2019.
ORIGINAL PAPER
Abstract Numerical modeling of complex rock friction angles have also been estimated as by-
engineering problems involves the use of various products in the small scale direct shear tests.
input parameters which control usefulness of the
output results. Hence, it is of utmost importance to Keywords Joint normal stiffness Joint shear
select the right range of input physical and mechanical stiffness Normal stress Joint basic friction angle
parameters based on laboratory or field estimation, and
engineering judgment. Joint normal and shear stiff-
nesses are two popular input parameters to describe 1 Introduction
discontinuities in rock, which do not have specific
guidelines for their estimation in literature. This study An understanding of discontinuities in rock, specifi-
attempts to provide simple methods to estimate joint cally rock joints, is crucial to the sound construction of
normal and shear stiffnesses in the laboratory using the structures in and on rock such as foundations, dams,
uniaxial compression and small-scale direct shear tunnels, underground chambers for hydro power and
tests. Samples have been prepared using rocks waste isolation and slopes. Hence, it is important that
procured from different depths, geographical locations rock joints and fractures be adequately characterized
and formations. The study uses a mixture of relatively and their properties quantified before making engi-
smooth natural joints and saw-cut joints in the various neering decisions for construction in rocks. Rock
rock samples tested. The results indicate accept- masses have traditionally been described by rock mass
able levels of uncertainty in the calculation of the classification systems such as the Rock Quality
stiffness parameters and provide a database of good Designation (RQD) proposed by Deere et al. (1967),
first estimates and empirical relations which can be the Rock Mass Rating (RMR) by Bieniawski (1974),
used for calculating values for joint stiffnesses when Q-system (Barton et al. 1974) and the Geological
laboratory estimation is not possible. Joint basic Strength Index (GSI) by Hoek et al. (1995), with
multiple modifications to suit specific locations (Stille
et al. 1982; Hoek et al. 2002; Romana 1985).
With the advent of powerful computers and
P. H. S. W. Kulatilake (&) S. Shreedharan numerical modeling capabilities, rock masses could
T. Sherizadeh B. Shu Y. Xing P. He be modeled more quantitatively by dissecting the rock
Rock Mass Modeling and Computational Rock Mechanics mass properties into the intact rock and joint physical
Laboratories, University of Arizona, Tucson, AZ 85721,
USA and mechanical properties. While some properties
e-mail: [email protected] such as the rock density, Young’s modulus, Poisson’s
123
Geotech Geol Eng
ratio, uniaxial strength, cohesion, angle of internal 2 Review of Literature on Joint Normal and Shear
friction and tensile strength of intact rock are easily Behavior
estimated in the laboratory for the purpose of assign-
ing to the numerical models, other properties may not Goodman (1974, 1976) suggested a hyperbolic rela-
be as easy to estimate. Often, numerical models make tion to describe joint closure under normal compres-
use of parameters which are obscure in definition or sive stress (Fig. 1). Bandis et al. (1983) presented a
difficult to estimate in the laboratory, since their usage modification of Goodman’s hyperbolic model, which
may describe the rock mass more efficiently during was shown to provide a better fit to experimental data
stress computation. Joint normal and shear stiffnesses across the whole range of stress and closure values. It
are two such parameters which are not easily estimated is necessary to estimate two parameters: (a) initial
in the laboratory and no formal guidelines are normal stiffness and (b) maximum joint closure to fit
prescribed for their estimation. In this paper, we these hyperbolic models. Shehata (1971) used a semi-
propose methods to estimate the joint normal and logarithmic empirical model to describe the same
shear stiffnesses in the laboratory, and estimate basic relation. Alternative models, based on Hertzian con-
friction angles, for smooth saw-cut joints or relatively tact theory, have also been suggested to describe the
smooth natural joints with relative ease and provide a non-linear stress deformation behavior (Sun et al.
framework for their subsequent usage in numerical 1985; Swan 1983; Matsuki et al. 2001). Malama and
models. The methodology may also be extended to Kulatilake (2003) proposed two functions: one
rough natural joints without significant modification.
Engineering rock mechanics has usually described
rock joint friction without much consideration to the
complexities involved in the frictional response of
rock joints/faults, since their importance is paramount
when discussing earthquake mechanics and less so for
engineering applications. A classical research on this
topic is by Byerlee (1978), which introduces Byerlee’s
Law and assumes constant friction angles of 40o and
31° for the joint/fault gouge depending on whether the
normal stress is less than or more than 200 MPa.
While this serves as a good first estimate for the
purpose of numerical modeling, other researchers
argue that the frictional response may be a function of
the rate of sliding (Dieterich 1992; Ruina 1983). A
comprehensive review of the rate-and-state friction
laws and their application to earthquake physics is
discussed by Marone (1998). Even though they are
discussed in conjunction with earthquake processes,
some researchers (Helmstetter et al. 2004; Veveakis
et al. 2007) suggest that landslides may follow similar
mechanisms and it may be necessary to analyze them
using rate-and-state friction models. Hence, readers
are urged to exercise caution when applying labora-
tory estimated friction angles for analyses of land-
slides, since they may be treated as static events
obeying classical rock joint frictional formulations or
dynamic events following rate-and-state formulations.
Fig. 1 Normal stress versus rock and joint deformation for
It may be necessary to carry out site-specific tests to
intact rock specimen and specimen with single fracture for
discern the applicability of various friction laws to a granodiorite (Goodman 1976). Here, the parameter Vm is the
specific landslide event. same as the parameter Dj introduced in Eq. 2
123
Geotech Geol Eng
123
Geotech Geol Eng
standard D7012 (2014). A single saw-cut was intro- line in Fig. 3). Then, the joint deformation was
duced halfway through the length of the prepared estimated for each normal stress value by subtracting
cylindrical specimen. The ends of both smaller intact rock deformation from the experimental rock
cylinders were ground and smoothened so as to mass deformation curve (Fig. 4). A best fit regression
eliminate any frictional or roughness effects. A function for this curve has been found to be the
masking tape was wrapped around the saw-cut joint exponential function as shown in Eq. (3), where A and
to further ensure that no shearing or slipping occurred B are empirical constants and r and Dj have the usual
along the horizontal joint. This sample, now ready for meanings as previously defined elsewhere.
testing, was uniaxially compressed to failure at a
r ¼ AeBDj ð3Þ
constant load rate. To estimate the stiffness properties
of interfaces between two rock types, the samples were Applying the natural logarithm function on both
created mismatched, i.e., the top half of the sample sides of Eq. (3) and subsequently differentiating with
belonging to one rock type and the bottom half to respect to r yields
another. Reasonable effort was made to ensure that the
dr=dDj ¼ Br ð4Þ
samples were free from preexisting fractures, joints or
voids. Figure 2 shows some of the prepared samples The left hand side of Eq. (4) is, by definition, Kn.
for the uniaxial compression test for samples with a Hence,
single horizontal joint. During the compression, a
curve of normal stress versus normal deformation was Kn ¼ Br: ð5Þ
plotted. Figure 3 shows the total normal deformation
of the sample expressed as a function of the applied
normal stress. As can be observed from it, the curve 3.2 Estimation of Joint Shear Stiffness
contains an initial non-linear part due to the closure of
the saw-cut joint and a linear part from the intact rock. The small scale direct shear (SSDS) test is particularly
A straight line parallel to the linear part and passing useful for estimating the frictional properties of rock
through the origin was drawn and designated as the joints and faults in the laboratory. For our purposes,
theoretical intact deformation for that sample (orange we follow ASTM standard D5607 (2008) to prepare
Fig. 2 Samples prepared for uniaxial compression test to estimate joint normal stiffness
123
Geotech Geol Eng
Fig. 4 Normal stress as a function of joint deformation and the fitted exponential curve for sample LSS (sample number 6 in Table 1)
the direct shear samples and to perform the tests. A increasing horizontal shear force pushes the bottom
labelled illustration of a shear box used for direct shear frame, thus shearing the natural or saw-cut joint
testing is given in Fig. 5 and the actual shear box used surface. A dial gauge connected to the shear box
for the testing, a Humboldt 25SIXTYSHEAR, is records the shear displacement of the bottom frame.
shown in Fig. 6. Samples used have either natural During the experiment, a plot of shear stress versus
joints or saw-cut joints. Each sample contains a top shear displacement was generated as shown in Fig. 7.
half and a bottom half, each embedded in a hydro- This process was repeated on each sample for four
stone mold so as to be held securely within the frames different normal stress values. Due to the linear nature
in the shear box. During the experiment, a normal of the shear stress versus shear displacement curve in
force is applied on the top frame while a constantly most of the pre-peak range, the shear stiffness was
123
Geotech Geol Eng
123
Geotech Geol Eng
0.8
JKS = 0.665σn
0.6
R² = 0.977
0.4
0.2
0
0 0.5 1 1.5 2
Coulomb strength criterion given in Eq. (7). Tests on limestones and mudstones (Devonian Rodeo Creek
natural joints can also be used to estimate peak friction and Devonian Popovich in Tables 1 and 2) associated
angle, residual friction angle and joint cohesion with a gold mine in USA, and dacite, conglomerates,
(usually assumed to be zero since joint cohesion mudstones and limestones from another gold mine in
values are significantly lower than intact rock cohe- USA have been analyzed in the study. All groups of
sion values). samples come from different depths, geographical
locations and periods of formation.
s ¼ rtan/: ð7Þ
4.1 Results Pertinent to Joint Normal Stiffness
4 Discussion of Results Table 1 shows the results and statistical analysis for
the Joint Normal Stiffness tests performed on different
The results are obtained from samples consisting of rock specimens as per the methodology detailed in
four broad groups based on the geographical location Sect. 3.1. Tests have been performed on coal measure
from where they were collected. Coal measure rock rocks (sample numbers 1–14) and rocks associated
samples (sandstones and shales) obtained from the with gold (samples 15–36). Experiments where two
United States of America, coal samples from China, types of samples have been tested (different top and
123
Geotech Geol Eng
bottom pieces) to determine interface properties are discrimination has been made for calculating the
indicated as LSS/SST, SST/SS etc. This means that the mean, standard deviation and coefficient of variation
top piece was the first sample (LSS, SST etc.) and the on the basis of joint type. All samples were tested at
bottom piece was the second sample (SST, SS etc.). four different normal stress values. Column 9 in
All tests associated with each rock type have been Table 2 gives the value of the constant F introduced in
grouped together in obtaining values for statistical Eq. (6) calculated from a best fit curve of Ks versus r
parameters. It can be seen from the low coefficient of passing through the origin. It may be noted that
variation under 20 % that this method has worked well different groups of samples have been tested under
for all the rock types except sandstone. Even for sets of slightly different effective normal stresses at
sandstone a medium level coefficient of variation the authors’ discretion. Similar to results obtained for
value has resulted. Overall, the results show low the joint normal stiffness tests, the shear stiffness
coefficients of variation, indicating reliability in the coefficient F takes values inside a narrow range of
test results. The constant B introduced in Eq. (5) 0.65–2.15 with dacite being the exception and having
(columns 4 and 5 in Table 1) is seen to consistently 2.68 as the mean F value. Since both normal and shear
take values in the narrow range of 10–20 despite the stiffness can be expressed as linear functions of the
samples having come from a variety of regions and applied normal stress, a one-to-one relation can be
origins. established between the normal and shear stiffness. It
can be done by simply dividing Eq. (5) by Eq. (6) (B/F
4.2 Results Pertinent to Joint Shear Stiffness ratio). This can be used to empirically estimate either
parameter when the other is available. Table 3 shows
The results and statistical analyses for the joint shear the relation between the normal and shear stiffness for
stiffness tests are tabulated in Table 2. Samples 1–17 the different rocks tested. Due to the grouping
are from coal and coal measure strata and the procedure followed for the sandstone and shales for
remaining samples are from gold mines. The samples the estimation of joint normal stiffness, the parameter
tested have, for the most part, saw-cut joints, i.e., associated with shear stiffness, F, has also been
perfectly smooth joints created artificially by sawing averaged in a similar way in Table 3. It can be
an intact specimen in half. Some samples were observed from Table 3 that the normal stiffness is
available with smooth natural joints and have also consistently larger than the shear stiffness by 5–25
been used for the study. Since the smoothness of the times. This provides a reasonable and usable guideline
natural joints was comparable to the saw-cut joints, no for estimating one stiffness parameter given the other.
123
Geotech Geol Eng
Table 1 Results and statistical analysis for joint normal stiffness estimation
Sample number Rock type Sample name Kn/r (B) Mean Coefficient
of variation
It may be noted here that previous studies (Wang et al. hypothetical situation where two identical rock sam-
2012) have shown that rock masses are usually not ples, each with a single joint with different normal and
significantly sensitive to changes in the stiffness shear stiffnesses, are subjected to the same stress state.
values unless the changes are of the order of at least The samples would exhibit similar deformation char-
10±2 times the original value. For example, consider a acteristics if the difference between their
123
Geotech Geol Eng
Table 2 Results and statistical analysis for joint shear stiffness estimation
Sample Rock type Sample Joint Ks (GPa/m) at different normal stresses Ks/r Mean Coefficient
number name type (r expressed in MPa) (F) of
variation
r = 1.72 r = 0.83 r = 0.42 r = 0.21
1 Lam. Silty LSS1 N 1.07 0.60 0.21 0.10 0.58 0.68 0.20
2 Shale LSS2 N 1.11 0.65 0.23 0.14 0.67
3 LSS3 N 0.95 0.47 0.38 0.05 0.57
4 LSS4 N 1.44 0.81 0.47 0.14 0.90
5 Sand-stone SST1 SC 2.91 1.37 1.1 0.36 1.92 1.32 0.26
6 SST2 SC 1.97 0.98 0.46 0.19 1.08
7 SST3 SC 2.23 1.16 0.68 0.06 1.15
8 SST4 N 1.52 0.71 0.46 0.36 1.14
9 Sandy Shale SS1 SC 1.79 1.10 0.52 0.29 1.25 1.23 0.29
10 SS2 SC 1.84 1.07 0.85 0.43 1.61
11 SS3 SC 1.15 0.52 0.40 0.08 0.66
12 SS4 SC 1.68 1.26 0.64 0.35 1.42
13 Shale S1 N 1.62 0.83 0.53 0.11 0.93 0.92 0.11
14 S2 SC 1.65 1.25 0.38 0.17 1.04
15 S3 SC 1.10 0.59 0.43 0.17 0.80
r = 1.00 r = 0.75 r = 0.50 r = 0.25
18 Dev. Rodeo DRC1 N 1.02 0.66 0.40 0.18 0.77 1.08 0.28
19 Creek DRC2 N 1.87 0.99 0.28 0.11 1.26
20 DRC3 N 0.89 0.54 0.40 0.07 0.66
21 DRC4 N 2.13 1.77 0.69 0.47 1.71
22 DRC5 N 1.87 0.98 0.50 0.19 1.30
23 DRC6 N 1.40 0.68 0.18 0.11 0.93
24 DRC7 SC 1.28 0.86 0.25 0.09 0.93
25 DRC8 SC 1.52 0.48 0.27 0.06 0.94
26 DRC9 SC 1.31 0.86 0.36 0.08 0.96
27 DRC10 SC 1.88 1.22 0.25 0.18 1.34
28 DRC11 SC 2.16 1.46 0.64 0.11 1.60
29 DRC12 SC 1.08 0.45 0.20 0.16 0.71
30 DRC13 SC – 1.01 0.42 0.11 1.15
31 DRC14 SC – 0.73 0.47 0.23 0.94
32 Dev. Pop. DP1 SC 1.38 1.28 0.53 0.16 1.15 1.05 0.30
33 DP2 SC 1.79 1.05 0.43 0.27 1.27
34 DP4 SC 1.34 0.67 0.60 0.36 0.97
35 DP5 SC 1.35 1.08 0.41 0.14 1.05
36 DP6 SC 0.88 0.54 0.35 0.13 0.65
37 DP8 SC 2.15 1.47 0.61 0.19 1.60
38 DP9 SC 1.70 1.44 0.59 0.34 1.38
39 DP10 SC 1.08 0.80 0.51 0.09 0.83
123
Geotech Geol Eng
Table 2 continued
r = 1.38 r = 0.83 r = 0.42 r = 0.21
corresponding stiffnesses is less than approximately joints are perfectly smooth, no cohesion has been
100 Pa/m. Hence, the inherent uncertainty present in assumed for them. Similarly, due to the extremely low
the calculation of joint normal and shear stiffnesses in values of cohesion for natural joints-of the order of
the experimental results presented in this manuscript ‘kPa’ as compared to ‘MPa’ for intact rock, cohesion
may be deemed acceptable. has not been presented in the table. The range of the
basic friction angles for the thirteen rock types
4.3 Joint Friction Angle reported is between 17° and 31°, with friction
coefficients varying between 0.313 and 0.597. Most
Table 4 shows the calculated basic friction angles and rocks have acceptable statistical uncertainties under
the associated statistical summary. Since saw-cut 25 % in the estimated friction angles. Only the
123
Geotech Geol Eng
Table 4 Friction angle estimation from small scale direct shear experiments
Rock/formation type Number of samples Mean basic/residual Coefficient of Coefficient of
friction angle (/, °) variation friction (tan /)
limestone-dacite interface test sample appears to have when building numerical models which use stiffnesses
a high coefficient of variation, exceeding 25 % to describe joints. This is because it allows for using
(35 %). Hence, the tests provide a useful addition to dynamically changing values of stiffnesses rather than
literature in the form of residual friction angles for a single static value which becomes an underestima-
joints in coal measure rocks including shales, sand- tion at high normal stress values. Also, it allows one to
stones, coal, and some medium hard rocks including select physically suitable values rather than arbitrary
limestone, mudstone, tuff and dacite. values when building models where laboratory testing
is not possible. Finally, the stiffness and friction angle
values serve as a useful addition to the existing
5 Conclusions literature on mechanical property values for joints and
fractures in different rocks.
This study presents simple methods to estimate rock
joint normal and shear stiffnesses in the laboratory. Acknowledgments The support provided by the various mining
Joint normal stiffness has been estimated using a companies through providing geological data, rock core and/or
block samples, and allowing access to the mine to perform field
modified uniaxial compression test which requires no
investigations is very much appreciated. The work was funded by
cyclic loading. Joint shear stiffness has been estimated the NIOSH of the Centers for Disease Control and Prevention
through multi-stage small scale direct shear tests. The (Contract No. 200-2011-39886). We also thank two anonymous
tests have been performed on rocks obtained from reviewers for their critique which helped improve the paper.
around the world, varying in origin, depth and
associated mineral. Despite this, the variability in
results falls within a narrow range due to which it can References
be used as a foundation for estimating joint stiffness
parameters when none are available. It has been Bandis SC, Lumsden AC, Barton NR (1983) Fundamentals of
observed that the joint normal and shear stiffnesses are rock joint deformation. Int J Rock Mech Min Sci
20(6):249–268
linear functions of the normal stress applied on the Barton N, Lien R, Lunde J (1974) Engineering classification of
rock sample. The normal stiffness, with units GPa/m, rock masses for the design of tunnel support. Rock Mech
has been found to typically be approximately 10–20 6:189–236
times the applied normal stress expressed in MPa. Bieniawski ZT (1974) Geomechanics classification of rock
masses and its application in tunneling. In: Proc. 3rd Int.
Similarly, the shear stiffness, expressed in GPa/m, is Cong. Rock Mech Vol. 2, pp 27–32
between 0.65 and 2.15 times normal stress expressed Byerlee J (1978) Friction of rocks. Pure Appl Geophys
in MPa. This observation serves as a useful guideline 116(4–5):615–626
123
Geotech Geol Eng
Deere DU, Hendron AJ, Patton FD, Cording EJ (1967) Design of Romana M (1985) New adjustment ratings for application of
surface and near-surface construction in rock. In: The 8th Bieniawski classification to slopes. In: Proceedings of the
US Symposium on Rock Mechanics (USRMS). American International Symposium on the Role of Rock Mechanics
Rock Mechanics Association in Excavations for Mining and Civil Works. International
Dieterich JH (1992) Earthquake nucleation on faults with rate- Society of Rock Mechanics, Zacatecas, pp 49–53
and state-dependent strength. Tectonophysics Ruina A (1983) Slip instability and state variable friction laws.
211(1):115–134 J Geophys Res 88(10):359–10
Goodman RE (1976) Methods of geological engineering in Rutqvist J (1995) Determination of hydraulic normal stiffness of
discontinuous rocks. West, New York fractures in hard rock from well testing. Int J Rock Mech
Goodman RE (1974) The mechanical properties of joints. In: Min Sci 32(5):513–523
Proceedings of the Third Congress on ISRM, Denver, vol. Shehata WM (1971) PhD thesis, quoted in Sharp JC and Maini
1A, Washington, DC: National Academy of Sciences. YNT, in fundamental considerations on the hydraulic
pp 127–140 characteristics of joints in rock. Proceedings of the Sym-
Goodman RE, Taylor RL, Brekke TL (1968) A model for the posium on Percolation Through Fissured Rock, paper no.
mechanics of jointed rock. J Soil Mech Found Div Proc Am T1-F, Stuttgart, 1972
Soc Civ Eng 94(SM3):637–659 Standard ASTM D5607–08 (2008) Standard test method for
Helmstetter A, Sornette D, Grasso JR, Andersen JV, Gluzman S, performing laboratory direct shear strength tests of rock
Pisarenko V (2004) Slider block friction model for land- specimens under constant normal force. Annual Book of
slides: application to Vaiont and La Clapière landslides. ASTM Standards. American Society for Testing and
J Geophys Res: Solid Earth 109(B2):1978–2012 Materials, West Conshohocken
Hoek E, Marinos P (2007) A brief history of the development of Standard ASTM D7012 (2014) Standard test method for com-
the Hoek-Brown failure criterion. Soils Rocks 2:1–8 pressive strength and elastic moduli of intact rock core
Hoek E, Kaiser PK, Bawden WF (1995) Support of underground specimens under varying states of stress and temperatures.
excavations in hard rock. A. A. Balkema, Rotterdam ASTM International, West Conshohocken
Hoek E, Carranza-Torres C, Corkum B (2002) Hoek-Brown Stille H, Groth T, Fredriksson A (1982) FEM-analysis of rock
failure criterion-2002 edition. Proc NARMS-Tac mechanical problems with JOBFEM, vol 307(1). Stiftelsen
1:267–273 Bergteknisk Forskning–BeFo, Stockholm, p 82
Jiang XW, Wan L, Wang XS, Liang SH, Hu BX (2009) Esti- Sun Z, Gerrard C, Stephansson O (1985) Rock joint compliance
mation of fracture normal stiffness using a transmissivity- tests for compression and shear loads. Int J Rock Mech Min
depth correlation. Int J Rock Mech Min Sci 46(1):51–58 Sci 22(4):197–213
Jing L, Nordlund E, Stephansson O (1994) A 3-D constitutive Swan G (1983) Determination of stiffness and other joint
model for rock joints with anisotropic friction and stress properties from roughness measurements. Rock Mech
dependency in shear stiffness. Int J Rock Mech Min Sci Rock Eng 16:19–38
31(2):173–178 Veveakis E, Vardoulakis I, di Toro G (2007) Thermoporome-
Malama B, Kulatilake PHSW (2003) Models for normal fracture chanics of creeping landslides: the 1963 Vaiont slide,
deformation under compressive loading. Int J Rock Mech northern Italy. J Geophys Res: Earth Surf
Min Sci 40:893–901 112(F3):2003–2012
Marone C (1998) Laboratory-derived friction laws and their Wang JA, Park HD (2001) Comprehensive prediction of rock-
application to seismic faulting. Annu Rev Earth Planet Sci burst based on analysis of strain energy in rocks. Tunn
26(1):643–696 Undergr Space Tech 16(1):49–57
Matsuki K, Wang EQ, Sakaguchi K, Okumura K (2001) Time Wang X, Kulatilake PHSW, Song WD (2012) Stability inves-
dependent closure of a fracture with rough surfaces under tigations around a mine tunnel through three-dimensional
constant normal stress. Int J Rock Mech Min Sci discontinuum and continuum stress analyses. Tunn
38(5):607–619 Undergr Space Tech 32:98–112
123