You are on page 1of 25

lnt. J. Rock Mech. MIn. Sct.& GeomeclL Ab$tr. Vol. 10. pp. 3t 1-335.Pm'pmoaPress 1973.

Printedin Great Britalu

CONTINUUM CHARACTERIZATION OF JOINTED


ROCK MASSES
PART l--THE CONSTITUTIVE EQUATIONS

B. SINGH
University of Minnesota, Minneapolis, Minnesota

( R e c e i u e d I O c t o b e r 1972)

Abstract--Hill has proved that the average strain energy density in any region of an elastic
and inhomogeneous material can be calculated from the average values of the stresses and
strains within that region. This concept has been used to derive the general constitutive
equations of a rock mass containing an orthogonal set of discontinuousjoints intersecting an
anisotropic rock material. The constants required for the continuum characterization of the
jointed mass are the 'joint stress concentration factors" Br,~ and Br~ (see Fig. ! of this paper).
These are defined as the ratio of stresses along the joint to the overall stresses in the rock.
Exact expressions for the stress concentration factors have been obtained for the case of a
rigid rock (i.e. the intact material between the joints) containing staggeredcompliant joints, and
a correction is propo~d for the case where the rock is (elastically) deformable. These results
suggest that interlocking between blocks of rock may become significant even for a slight off~t
along the joints, such as may occur during shear deformation of a rock mass containing
an initially continuous orthogonal joint set.
Stress concentratk~n factors computed independently from the results of a finite element
program for a jointed mass compare exccllently with the above-mentioned theoretical results.
It is further shown that tensile stresscs are developed inside a rock with staggeredjoints, and
may be as high as twice the overall shear stresses or the overall compressive stresses. Typical
stress distributions within a block are given in Figs 6(a) and 6(b) of this paper.
It is concluded that a rock ma.~ is rendered anisotropic by any joint set having a preferred
orientation. Expressions from which the elastic moduli of the equivalcnt continuum anisotropic
rock ma~ may be obtained are presented in Table I.

INTRODUCTION
LARGE-SCALE discontinuities such as j o i n t s and bedding planes are characteristic structural
features o f rock masses. M o r e o r less regularly orientated and spaced, the discontinuities
can exert a d o m i n a n t influence on the overall mechanical response o f a mass. It w i l l be
decidedly a n i s o t r o p i c in its d e f o r m a t i o n behavior, and failure (when it occurs) w i l l be
strongly influenced by the o r i e n t a t i o n and properties o f the discontinuities. It is conse-
quently not surprising that attempts to analyse the behavior o f rock masses on the basis of
classical isotropic c o n t i n u u m theory met with little success.
The advent o f the finite element method of analysis gave considerable impetus to theor-
etical study of rock mass behavior. With this method, discontinuities, inhomogeneities and
a n i s o t r o p y can be taken into a c c o u n t with relative ease. Even so, for m a n y problems,
finite element amdysis suffers from two distinct disadvantages:

I. T o take a c c o u n t o f all the significant structural features, such as j o i n t s and bedding


planes, in the detail desired, frequently requires storage capacity in excess of that
available in the present generation o f computers.
311
31_* B. SINGH

2. Even if the capacity were unlimited, the geological field information is not available
in sufficient detail to permit accurate computer modelling of the structure. Joint
orientations and spacings, joint filling materials, etc. will vary from place to place.
It is also impracticable in many engineering situations, to consider in detail every structural
feature even if these were known and could be modelled on the computer. In tunnel driving,
for example, the shape of the tunnel, and type and size of support, must be decided prior
to the start of excavation. These decisions must be based on average rock conditions, and
the design must be capable of accommodating moderate variations from these conditions.
Similar considerations apply in the planning of field experiments ~e.g. plate-bearing
tests) to ascertain the mechanical properties of the rock mass. The effects of variation in
(perhaps unknown) structural conditions should be theoretically examined in advance in
order to ensure that the experiment itself is 'well-conditioned" (i.e. correctly designed so as
to reveal, with maximum accuracy, the desired information). The optimum location of
measuring instruments, for example, varies appreciably depending on the model assumed in
advance of testing, to represent the mass.
There would appear, then, to be considerable practical merit in developing procedures
whereby the average behavior of a jointed or bedded rock mass can be studied without
excessive demands on computer capacity.
This paper describes the development of an anisotropic continuum model in which the
average influence of joints, bedding pkmes and similar planar features can be taken into
account. It will be shown in Part !! of the paper that this continuum model is quite accurate,
when compared with the finite element model, in regions of lower stress gradient. The
accuracy falls off considerably in regions of high stress gradient. This disadvantage is
readily overcome, however, by using the anisotropic continuum model to represent the
major parts of thc mass (in which stress gradients are moderately low), and reserving finite
element analysis to the regions where detailed modelling of discontinuities is n~'essary, as
in regions of high stress gradient. In this way it is possible to make much more effective use
of the capabilities of high-speed digital computers.
Other investigators have used anisotropic continuum theory to describe rock masses
although progress has been limited. SM^R'r [I], for example, has considered the case of a
mass containing three orthogonal equally spaced joint system, each joint containing gouge
which he assumed to be as an elastic, isotropic material. SAL^MON ['2_] derived a general
solution for stratified rock in which the elastic constants of the strata arc arbitrarily distri-
buted. This solution can be used to represent a rock mass containing horizontal, filled
joints. Continuum models for more general jointed rock mass systems have not been
reported, presumably because of the increased mathematical complexities of such problems.
The progress reported in this paper has been made possible by assuming a different model
for the joints. In this model, the joints are considered as a surface of discontinuity in the
rock mass along which displacements are uniquely related to the corresponding stresses.
In the continuum characterization, the joints and the intact rock are considered the two
component phases of an elastic system. The elastic theory of composite materials is thus a
useful base upon which to develop the characterization.

THEORY OF ELASflCITY OF COMPOSITE MATERIALS


A jointed rock mass may be regarded as a composite material involving two main com-
ponents: (I) relatively intact blocks, and (2)joints or planes of discontinuity. Development
CONTINUUM CHARACTERIZATION OF JOINTED ROCK MASSES. PART I 313

of the rock-mass continuum model is based on the theory, of composite materials. It seems
appropriate, therefore, to briefly review the more pertinent recent contributions.
The basic principles of the elasticity of composite materials have been propounded by
HILL [3]. The theory involves several assumptions concerning the interaction between the
various phases of the material. Thus:
(i) All phases of the material are firmly bonded to each other so that there is no slip
between adjacent phases along a boundary during the deformation.
(ii) The material is macroscopically homogeneous. This requires all characteristic
dimensions of the inhomogeneities to be small in comparison to the dimensions
of the (analytical) elements Isuch as used in finite element analysis) of the material.
{iii) All inhomogeneities {inclusions) are aligned.
HIt.L [3] showed that the a~erage density of the strain energy in any region can be calcu-
lated from the averages of the stresses and strains within the body, when the boundary
traction is 'macroscopically uniform', in the sense that deviations from the mean are small
compared with the mean value. Thus the 'overall" or 'effective' stress strain relations are
derived from the following energy equations:
if"-: t/2[,~lT[~]--i/2[,~]T[M][,~I {t)
where [M] is the compliance matrix of tile composite material, and the bar denotes the
average value of the function, lit may be noted that similar relations do not hold invariably
for the average density of the energy in an individual phase, in terms of the average stress
and strain in that phase.) Equation (I) obviates the necessity of knowing the local distri-
bution of stresses ~ithin e;tch phase. Proceeding from this basis the following chtssical law
of el:lsticity of a two-phase material has been propounded:
[MI ~ [.tt,1 .~ e, {[MjI - [.~t,l} [Oil (2)
where the suffixes r :mdj denote the two different pha~es, cj is the volumetric concentration
of thej-th phase.
The matrix [Bj] relates the average of the stresses in thej-th phase to the overall average
stress [,-~].

[,~jl = [Bjl[,~]. 13)


Composite elasticity problems may then be solved completely simply by determining the
term [Bj] which I-llt.t. [4] has termed the "'stress concentration factor". This m:ty be accom-
plished by linite element analysis of a representative volume of the material.

Determination of the stress conc'c, tration factor [B~]


In the case of a rock mass, [Bj] is determined by analyzing a large ('prismatic) body of the
rock mass containing several representative joints. The body is subjected to uniform
boundary displacements or tractions. The stress distributions predicted by the linite element
analysis may now be used to compute the ratios between the average stresses along a joint
and the o~erall stresses on a plane parallel to that joint. These ratios are the stress concen-
tration lactors [Bj], as delined by equation (3).
Finally. the compliance matrix [M] of the rock mass may be found from equation (2)
by substitution for the compliance matrices of the rock [M,] and the joint material [Mj].
Unfortunately. there is very little practical information to indicate appropriate values for
314 B. SINGH

the elastic constants of the joint material. Experiments to determine the p r ~ s e relation-
ship between [Mj] and stiffness of the joint are very much needed in order to allow realistic
application of Hill's analysis.
An alternative approach has been adopted by HASmN and SHnUKMAN [5], who deter-
mined upper and lower bounds for the moduli of deformation of a heterogeneous material.
According to the principle of minimum complementary energy, the actual strain energy
of an elastic body will be less than the strain energy corresponding to any fictitious equili-
brium state of stress. Using this principle, we may obtain an upper bound estimate of the
compliance of a composite material, by assuming the stress concentration factor to be unity.
This is known as the 'Reuss" estimate. Similarly, a lower bound estimate of the compliance
may be obtained by assuming a uniform distribution of the strains within all phases. A
simple calculation will show that the actual stiffness of a heterogeneous material is less than
or equal to the volumetric average of the stiffnesses of its phases. This lower bound estimate
is known as the "Voigt' estimate.
HASmN and S~rrmKMAN [5] proved that the upper bound estimate of the stress concen-
tration factor may be significantly less than unity for the case of isotropi¢ materialcontaining
randomly orientated inhomogeneities. Their bounds are quite close for all practical purposes.
Such an approach may be useful in studying the elasticity of rock masses, which could be
assumed to consist of essentially isotropic material with randomly orientated and infilled
joints.
The problem of heterogeneities which may be represented as ellipsoidai inclusions has
received considerable attention. HILL [4] employed the 'self-consistent method' to derive
expressions for the stress concentration factors. The self-consistent theory examines the
interaction between inclusions by considering the inclusions to exist isolated (without
interaction) in a medium with an elastic modulus which is the same as the effective modulus
of the heterogeneous material. This hypothesis allows direct use of the classical results for
inclusions (of compliance [MjD embedded in an infinite elastic, homogeneous medium
whose continuum characteristics [M] have been derived from the microscopic character-
i.stics of a heterogeneous material. The expression for the stress concentration factor takes
the form
[Bj]-' = [I] -t- {[I] -- [s]} { [ M ] - ' [Mj] -- [I]} (4)
where [!] is a unit matrix whose diagonal coefficients are unity. The matrix Is] is dimension-
less. Its coefficients depend upon the shape of the inclusions, and Poisson's ratio and the
nature of anisotropy of the media. The coefficients of matrix Is] have been derived by
Est,~LBY [6]. His results suggest that the diagonal coefficients of [s] will be between zero and
unity. It is interesting to note that equations (2) and (4) may be combined to yield either the
Reuss or the Voigt result by setting [s] equal to unity or zero respectively. Thus the whole
problem reduces to the proper selection of coefficients for the matrix [s]. In the special case
of open cracks of ellipsoidal shape, the solution may be further simldified by setting the
stiffness of the inclusion [ M j ] - = equal to zero. In this case, equations (2) and (4) combine to
give
[M] = [M,] A- c j [ M ] { [ l ] -- Is]} - t (5)
where cj is the crack porosity. Equation (5) suggests that the modulus of deformation may be
correlated to the crack porosity of the fractured rock, provided the cracks do not intersect
each other, it therefore appears that the above-mentioned analysis may help more in studying
CO,WI'INUUM CHARACTERIZATION OF JOINTED ROCK MASSES. PART I' 315

the deformation characteristics of the rock material itself(i.e, between the joints) rather than
the rock mass.

Choice ofl elastic moduli


It is important to know which values of the elastic modulus to use for the rock material
when laboratory tests on rock cores yield widely scattered results. BeRA~ and MOLYNEUX
[7,8] studied the effective bulk modulus of deformation o f a stochastic elastic material
in which the elastic constants fluctuate randomly everywhere. Subsequently, KRoner [9]
obtained the effective shear modulus of the perfectly disordered composite material. The
results of these investigations show that the effective elastic modulus may be significantly
less than the mean value of the population of the (experimentally determined) elastic moduli,
depending upon the variance of the latter. The conclusion of Su, WaNG and STEFANKO[10],
that the mean stresses in a stochastic body are not significantly different from the stresses
in a classical elastic body if the modulus of deformation of the latter is equal to the mean of
the variable moduli of the stochastic body, is therefore not general. However, in the special
case of a homogeneous stochastic body, their conclusion may be approximately correct,
because the stress distribution in an elastic body is rather insensitive to the elastic constants.
Further it is felt that a stochastic rock mass may be simulated in the finite element model by
randomly varying the elastic modulus of the rock, the orientation and stiffness of the joints
as these authors suggest, provided that care is taken to account for the size of the finite
element. Such a model may then be studied to evaluate the effective modulus of deformation
of the stochastic rock mass, as discussed earlier.

CONSTITUTIVE RELATIONS
The problem ofdctermining the elastic behavior of a rock mass differs somewhat from the
corresponding problem for a composite material, in that in the former the actual behavior of
the joint phase is rather unknown, in conventional finite element analysis [I I] the joint is
considered to be one-dimensional linite element of zero thicknesses which has shear stiffness
Kr and normal stillness Kn. These joint stiffnesses are not known a priori. As such it does
not seem possible to predict the mass modulus simply from the results of laboratory tests on
rock cores. However, if the displacement field is observed during a large-scale test, the
interpretation of the results is made easy by presuming a joint model. As will be shown later,
only two constants need to be evaluated in order to determine the complete stress-strain
relations of the rock mass, provided the elastic moduli of the rock are known.
The philosophy of approach is to use the same joint model in the design as in the inter-
pretation of field results. Consequently any error made in inferring the test result is compen-
sated by the opposite nature of the error in assessing the performance of the structure. It
would be justifiable therefore to use the simple one-dimensional joint model for continuum
characterization of the rock mass. It may be recalled here that the effective moduli of the
rock mass may be estimated from the results of a finite element analysis of part of the rock
mass. The object of the following discussion is also to derive exact constitutive relations,
for a better understanding of the behavior of the rock mass.
Consider the problem of a single system of joints which may be discontinuous but are
parallel, each having the same normal and shear stiffnesses, KMt and Krt. The rock is
homogeneous but may be anisotropic. The joints may be randomly spaced and have random
316 B. SINGH

stiffnesses. We need to know only the average fracture frequency n=. and the average compli-
ance of the joints. The detailed analysis is included in Appendix A. The compliance matrix
[M] of the rock mass may be expressed by an equation similar to equation (2):
[M] = [.14.] + .,[Kjtl-t[e~t] (6)
where
[M,] = compliance of the rock [given by equations (A.24) or (A.25) in Appendix
A]

[K,,I-' =
I ° °i]
o

0
r;,,

0 KT= f

oo
[aj,] = 0 e~,

0 0 Br,

and B~,, BT, are the stress concentration factors for normal stress and shear stress respec-
tively along the joint.
The effect on the rock mass of compliance of a second system of joints orthogonai to the
first may b¢ taken into account, in this case equation (6) is expanded as shown in equation
(7), after defining the rock mass with the first joint set to be a rock* of compliance M
(Fig. 1). Then
[M]..,=.=.., = [M] + . , [ K ~ , I - ' [ a , = ] (7)
where the subscript 2 refers to the second joint system.

Rock block O v s r O I I $ h e o r =.tress 1"

Ovsroil
norm=it
stress ¢7

2 C r o s s - i s * h i set I " J o i n t set Z:


Noemol s t i f f n e s s KNI Noemol s t i f f n e s s KN2
Sheor s t i f f n e s s KTI Shel=¢ s t i f f n e s s KTZ

= I
A l e s of G n i t o t r o p y

Fro. I. Rockrmumcon,-;ninl s ~ joints.


* In the caseof the nonortholp)naJ joints, the con~pliam:¢ matrix M must b¢ transformed to the new coordin-
ate axis which is perallel to joint set 2.
CONTINUUM CHARACTERIZATION OF JOINTED ROCK M-~SSES. PARI" I 317

Further, in the case o f a stochastic joint system with randomly v a ~ i n g joint stiffnesses
and spacing, equations (6) and (7) may be generalized by replacing the joint parameters n.
K~-t and K r - t by their average values. Futhermore it has been pro~ed that the stress
concentration factors are unity if the joints are continuous, since the stresses along a joint
must be equal to the overall stresses on a plane parallel to that joint in order to maintain
equilibrium. It is concluded from equation (7) that the off-diagonal terms o f the compliance
matrices of the rock mass and the rock are the same. This is a consequence o f a d o p t i n g a
one-dimensional joint model. An interesting feature o f the anisotropic continuum model is
that the principal strain directions may not coincide v,ith the principal stress directions. Table
I summarizes the results in the form o f the input data required for the finite element program.

TABLE I. ELASTIC MODULI OF THE ANISOTROPIC CONTINUUM MODt-L OF A JOINTED R(I<'K MASS (Fig. I)

Moduli of Et = R F t . E, E~ - R F , . E, PI vz
rock mass

Joint I I I
due to due to
sets kF, tCF, Gtz
only ~t only ,,~

Single joint set normal to I L. E, v, v,. RF~


I I
axis 2 S2KN2 -~, + SaKr-'-"'~

Orthogonal joint sets normal F.., I ! I


I F - - I v,.RFt ,',.RF~
to axes I and 2 St K.,l if,, + s-;-,~ , , + s , x ,---]

Orthogonal joint sets with I ;. Brt + - -I


staggered cross joints normal I ~- - - I !- v,.RFt ,',.RFa -

St K.,tt S2K.,~z (;, " S t K r l S2Kr2


to axis I (obtain B,,. and B r t
fr(lm equations ( I I ) and (I 2))

Note: Elastic m~duli are delined by the following constitutive equations:


eI (aI/RF t -- ol, , -- ajv,)/f,
"2 (-o,~. + . a l l C F . - oj..)/E,
• , - (-.'.~, - o2", t - o ~ ) I E ,
~'z. :: f l 2 / G t 2
where subscript
r denotes properties of intact rock material
1,2 denotL.'saxes of anisotropy which are orthogonal to joint sets I and 2
3 denotes third axis orthogonal to axes 1 and 2

H o o k e ' s law requires the stress-strain matrix to be symmetric, so that


I V2 ~'r
E_ = E-S = ES" (8)
This relationship is also reported by CALCOrE [121 for the similar problem o f laminated
composite structures. Equation (8) indicates that the reduction factor should be applied to
Poisson's ratio as well as to the elastic modulus, it may be remarked here that, in the finite
element program currently used (adopted from HEUZE [I I ]) for study o f these joint problems,
the shear modulus of an anisotropic material is c o m p u t e d from the relationship
! ! I 2.,
a,--S = + E + e--T (9)
318 B. SINGH

Strictly, this equation holds for the shear modulus only when the coordinate axes are
inclined at an angle of 45 ° to the axes of anisotropy. This kind of anisotropy [i.e. equation
(9)] was proposed by St Venant in 1863 (LEKH~rrsKll [13]) and is currently used because of
its great simplicity. It allows one to study an orthotropic anisotropic body with only three
independent elastic constants.
If the shear modulus G t2 of the rock mass is defined as the ratio of half the deviator stress
to the deviator strain it is easily shown that equation (10)
1 2(1 + v,) 1
- - - + ~ (lO)
G,~ E, St K..,.
may be derived by considering a rock mass which contains joint set 2, and which is subjected
to a unit-compressive stress normal to the joints a unit tensile stress parallel to the joints.
The rock mass will then behave as a St Venant body for the special case where the shear
stiffness Kr2 is equal to the normal stiffness K.vz of the joints (Table 1).
A practical solution may be found to the problem of determining the deformability of
rock masses containing joints filled with gouge of known characteristics. The shear stiffness
Kr of the joint is simply equal to the ratio of shear modulus of the gouge to the thickness of
the joint (7). Further, since the joint thickness is very small compared to its length and
width, the gouge will tend to compress under *zero lateral strain' condition. Consequently
the normal compliance of the joint may be equated to the product of the compressibility
of the gouge and the thickness of the joint. The compressibility of the gouge may be
determined from laboratory consolidometer tests. It may be noted here that the constitutive
equations derived from S^LAMON[2] and SMAzr [I ] simplify to the expressions listed in the
first and second rows of Table 1 respectively, provided it is assumed simply:
(i) That the product of the volume nTand the elastic moduli of thegouge are negligible
compared to the elastic moduli of the rock material.
(ii) That the volume of the joints is small compared to the volume of rock in the rock
mass.

F_~TIMA'rlON OF S'i'RE..~ CONCENTRATION FACTORS FOR STAGGERED JOINTS


The stress concentration factor has been defined as the ratio of the average stresses along
the joint to the corresponding overall stresses on a plane parallel to that joint within the
rock mass. In the special case of continuous joints, equilibrium requires that the average
stresses along thejoint be equal to the overall stresses. However, this is not true for staggered
joints. The magnitude of the stress concentration factors would depend upon the extent
of interlocking between the blocks of the rock.
The stress concentration factors have been derived in Appendix B assuming that the rock
is rigid (relative to the joints), and may be expressed as follows:

=
[ *-*( ,)],
+ l - (11)

[ (, ,)]'
where subscripts I and 2 refer to the cross joints and the continuous joints respectively
(Fig. I). Br, and Bsl are stress concentration factors for shear stress and normal stress
CONTINUUM CHARACTERIZATION OF JOIN"TED ROCK MASSES. PART [ 319

: ~.~f4T~NiJC'r',.3 , 3" JC.~NT OFI:SET '.~

• "
rLI~'~ 3TI":;dE33 .] ,:1
..... ii'
[1 ;;
. . . . . ]
Y s~ .I
Z

E__x (=)
s~lz ,I

F 5Z

×
Ib)

Fw(i.2. A part of the rock mass (containing two orthogonal staggered joint sets) analysed by the finite element
method to compute the stress concentration factors B,,. Br~.

respectively. The joint offset is denoted by S. An interesting result may be noted that the
stress concentration factors are the same for shear and normal stresses when shear stiffness
is equal to the normal stiffness of the joint as envisaged in St Venant's type anisotropie
continuum model of a rock mass. it may be noted that the stress concentration factors
decrease with increasing stiffness of the continuous joints and increasing joint offset. Thus,
there may be significant interlocking between joints even with small joint offset when the
normal stiffness of the continuous joints is high in comparison to that of the cross joints.
Such conditions are likely to exist when continuous joints have been closed and shear
deform:ttions have resulted in offset among originally continuous cross joints. The above
expressions for the stress concentration factors may be substituted in equations (6) and (7)
to obtain the compliance matrix of a rock mass containing staggered joints.
The stress concentration factors have also been computed from finite element analysis.
Figure 2(a) shows part of a rock mass analysed in plane strain. The following boundary
conditions have been used to determine Br~ :
(i) Normal stresses along all sides are zero.
(ii) Horizontal deflections along AB are zero but uniform and of a given magnitude
along CD.
(iii) Vertical deflections along sides BC and AD are zero.
(iv) Points ABCD are hinged.
To determine B,~:

6) Shear stresses are zero along all sides.


(ii) Vertical deflections are zero along sides AB and CD.
(iii) Horizontal dellection is zero along side AD but uniform and of a given magnitude
along BC.
The shear stiffness of the continuous joint was kept equal to the normal stiffness, corres-
ponding to a modulus reduction factor of 1/20. The joint offset was varied as 1/8, i/4 and 3/8
320 B. SINGH

[ ,
F,nLte element
""~%A~A__ I / 5 onotyst$

c~

o is.
~J
c]
06 --
S, =4
S.

o=°°
T~eoretqcQl
.j
onalysls
v ~,.1) S t r e s s concentration
facfor for shear
~v Stress [3. t

~o
I -- = [ 'l' '' ,'4
0 0 25 C 50 0 ~5 ~
_s E,_S_ ]
S: S,
I0
F,nlte element
7ncllysls

m o8
Z,

S.
§ 06
,.
¢ 04 o%:4
v ! S.
g
Theoret,col
onalysls

~"'I'-- ,oo
0 0 ,?.5 0 ~0 Or5
S [';_ S
s~ T, 1
(b) Stress c0ncentrott0n foctor for normtll s t r e s s 8N;

F,(;. 3. Comp~isun of ~int stress conomtration factors predicted thcorctE:aUy assumim~r~d rock with
those computed by the finite element analysis.

of spacing along cross joints. The ratio of spacing along cross joints and continuous joints
was taken as 2 and 4. The stress concentration Pdctors wcrc c',deu'lated from the stress
distribution between central joints so that the influence of idealized end conditions is
minimized. The results have been compared with those from equations (11) and (12) in
Figs 3(a) and 3(b). They are seen to be in excellent agreement.
The equations (11) and (12) have been modified to account approximately for the ©iasticity
of the rock when the joint offset is half of the spacing between cross joints. The detailed
derivation is given in Appendix B. The expressions for the stress concentration factors arc
found to be as follows:
K~2 S t __ [ 1 + -K..S.
- S,']-' (13)
B;~ = 1 + 4Krt $2 40E, ~'iz'J
Kr2 St[ 4Kr,S____~zKr2S, Siz_] -'
B.~tt = 1 + 4K~-"~tS-: i + 3E, + 10~, S=2J (14)
where E, is thedastic modulus of an isotropic rock block. These relations arc approximately
CONTINUUM CHARACTERIZATION OF JOINTED ROCK MASSES. PART I 321

valid for plane strain conditions. It may be seen that the influence of the elasticity of rock
is not significant when the term K.~,S,./E, or the modulus reduction factor across con-
tinuousjoints is less than 1,10.
Stress concentration factors were also computed by the finite element analyses. Only a
small part of the rock mass needs to be considered because of symmetry, as shown in Fig.
2(b). (Joints AF and CD are half-joints because lines AE or BD pass through the middle of
the cross joints and represent planes of symmetry. Thus a half-joint will have half the
compliance of an actual joint.) The boundary conditions were taken to be the same as for
the staggered joint problem. The computations were carried out considering joint stiffnesses
corresponding to a modulus reduction factor of 1/20 and 1/4 in the direction normal to
the continuous joints. The shear stiffncsses of both joints were set equal to the normal stiff-
nesses. Figures 4(a) and 4(b) show the results of the above analyses and, for comparison,
those from equations (I I) and (12), for a modulus reduction factor of 1/20. Excellent agree-
ment was again noted. Further the stress concentration factors predicted by equations (13)
and (14) are fairly close to those computed by finite element analysis for the case of a
modulus reduction factor of I/4, as shown in Figs 5(a) and 5(b).
,0 . ~ ~ . __ l,,oo

0a O~j x

o s L

o Theoret,cal
- 04
c Fin,re element
ue 0 n a 1y ' ; , ' ~ X
c
o
u 02--

0 ~ 2 3 4
S,/$2
(a) Stress concentraf,on factor for shear stress BTI

; oe S.I

o ~ I/2
= 06
o

c 04 Finite e l e m e n t

,,, 02--

0 i 2 3 4

S~/Sz
( b ) S t r e s , ~ c o n c e n t r a t i o n factor for normol '~fress B,~r
F t t ; . 4. Comparison of joint stre~s concentration factors predicted theoretically assuming rigid rock with
those computed by the finite element analysis.
LW.M.S. I 0/4.---~
322 B. SINGH

'°1 : 00
x O5

0"6 X KN2 2
Theoretical - -
Finite element
ana|ysil X

O I Z 3 4
S=/S:
(a) Stress concentration factor for shear stress BT!

-~o s,'g
06
Theoretical
~- Finite element
g
u 04 4 onalylds X

O I 2 3 4
St/S;,
(b) S tress concentretion factor for normal stress BNI

FIG. 5. Comparison of joint stress concentration factors predicted theoretically, considering elm ~.qasticity
o f rock with those computed by linit¢ clement analysis.

The analysis (Appendix B) indicates that continuous joints are subjected to tensile stresses
when the rock mass is deformed by shear stresses. The analysis is, therefore, vM/d only when
the normal stresses in situ are sufficiently high that the joint does not experience any exten-
sion. For the case where the initial normal stress along the continuous joints is zero, the
normal stiffness Ks= should be taken as half of the actual normal stiffness in equations (1 !)
and (13) following the arguments used to derive these equations (Appendix B).

LOCAL S T R E . ~ DISTRIBUTION IN A ROCK MASS CONTAINING STAGGERED JOINTS


The elastic analysis based on the continuum model gives an overall stress field. As
already noted, the average of the stresses within any region of the rock mass tends to be
equal to the overall stresses within that region (H,u. [3D provided it is not subjected to
significant stress gradients. The stress distribution obtained from the joint model will thus
tend to conform to the distribution obtained from the continuum model of the rock mass.
However, local stresses determined by the joint model may differ sipificantly from the
overall stress at any point even thoush there are no overall sties= gradients. Obviously this
C O N T I N U U M C H A R A C T E R I Z A T I O N O F JOINTED ROCK MASSES. PART I 323

difference will d e p e n d u p o n the extent to which the stress concentration factors B.~t and
Bri deviate from unity. Thus in the case of continuous joints, the stress fields computed
from both models should be virtually the same (in the region of small stress gradients)
because the stress concentration factors are equal to unity. On the contrary, the local stress
distribution within a rock mass containing staggered joints will have to be determined separ-
ately from the known overall stresses or strains, following the procedure used to compute
the stress concentration factors.
The local stress distribution inside a block of rock is plotted in Figs 6(a) and 6(b).
Tensile stresses are indicated as positive values. The stresses were computed by finite

r ,,-> ]c i
.i~lTT,~,.,,,,o, ,, co~p
i---! - :ill. ,
I~IL----- ~ .... JI I r o c l , ,',,oss
I lCor.o. ,..°.,o.!1i
' | I "

iJ , a I

J o i n t 2 ~,
C

Crc" iO [~,
iumt

i,)5@

(a)

I T 1
ibr ...... "o --~i
• [,Lines of symrnetryll
;L. . . . i - - - -- -il I t~ock rnac,~
A= B II

~- .......... ~_7 ...... i


*",~ J o i n t 2 ~
=40 . . . . . . . . . . . . . . . . . . . . :') C

CI'OSS i '"')
io~nt

(b)

FIG. 6(a). M a x i m u m principal stres.~-s (psi) in a quarter block o f the rock mass subjected to an overall ~hear
stress o f 519 psi (positive values indicate tensile stresses).
FiG. ~ b ) . M a x i m u m principal stresses (psi) in a q u a r t e r block o f the rock mass subjected to an overall
h o r i z o n t a l strc~,s o f - 1 8 5 0 psi.
3".4 B. SINGH

element analyses of a part of the rock mass shown in Fig. 2(b). The same boundary condi-
tions were used as in analyses of the stress concentration factors discussed earlier. The
stiffness of cross joint set 1 was taken to be 1/100 of the stiffness of continuous joint set 2.
This corresponds to a modulus reduction factor of 1/4 in the direction perpendicular to the
joint set 2.
Figure 6(a) shows contours of maximum principal stress for an overall shearing strain of
0.5 per cent. The overall shear stress was found to be 519 psi. It is interesting to note that
significant tensile stresses are developed in this case. The maximum principal stress planes
were inclined at about --45 ° to the horizontal.
Figure 6(b) shows contours of the maximum principal stress for an overall horizontal
compressive strain of 0-5 per cent and no overall vertical strain. The average horizontal
stress was computed to be -- 1850 psi. It was surprising to note that tensile stresses as high
as 1200 psi could be developed.
Figure 7 shows the relationship between joint parameters and the maximum tensile
stress developed on planes parallel to the cross joints when the rock mass is distorted by a
pure overall shearing strain. The stiffness of the continuous joints was held constant at a
value corresponding to a modulus reduction factor of 1/20.
As can he seen from Fig. 7(a), the maximum tensile stress is rather insensitive to the joint
offset S in the range of I/8-1/2 of the jointed spacing S~. The tensile stress is also not sensitive
to the ratio of stiffnesses of the cross joints and of the continuous joints [KMz/Kr~]as
shown in Fig. 7(b).
The maximum tensile stresses (,r.),,,, may also be obtained, approximately, by applying
the theory of bending of beams [equation (B.15) in Appendix B] as follows:

(")" = ~(i - Br,) S, 05)

2.0-- J
K,,zl KTI SIIS z

o~
Z 10 .~
. I --- 1_Th.o.e,i©a,
--o-, ---~

~ ' Finite element


I/5 ~ ~ analysis
i I i f I, I l J
o o.tzs o zs o.~,?s' o.~o o so ;~s "r~, ,00
S/51 KN~ KTI
(a) Influence of the ratio of iolnt offset S to (b) Influence of the ratio normal
the spolcttt9 of creel; joints ~ ( $ t ~ 4 , St i f ~ 4)~ooatinue~l joint s
Sz= spacing of continuouS joints ]. (K~)~o the shear stiffness of
the Cress joints(KT,) for various
ratios of sWscings of crossand
continuous joints (S,/S~) IS/S,,
0"50 for oil cases).

Fla. 7. F,/rect o f spi:ini~ ~ and uilTnml of joints on the m ( t a m k ) stress (or,),,= due to an
ovendl shorn" stress ~.,'.
CONTINUUM CHARACTERIZATION OF JOIN-rED ROCK MASSES. PART I 325

I00 KT?./KNI

-O.i

:<
b -0.5
u

I15
0125 0-25 0575 0.50
S/So
{o) I n f l u e n c e of the ratio of joint o f f s e t S to the spacing of cross
j o i n t s S I ( S I / S : , = ? , S2is spacing of c o n t i n u o u s j o i n t s ) .

b~
y
f
4

1 1 I I
0 25 50 15 t00
Kr:/Km
[b) I n f l u e n c e of the r a t i o shear s t i f f n e s s of c o n t i n u o u s j o i n t s
(K r2) to n o r m a l s t i f f n e s s of the cross j o i n t s ( K m t ) f o r v a r i o u s
r a t i o s of spacinqs of cross and c o n t i n u o u s j o i n t s (SI/S~.)
[ S / S i l O 5forallcases].

F , ;. 8. Effect of'~pacing, offset and st ill'ness o f joints o n the m a x i m u m tensile stress (,,.).,.= d u c to an overall
horizontal stress ,,~,'.

where ~-x:° is the overall shear stress. The tensile stresses predicted by equation (15) are
also plotted in Fig. 7(b). They are seen to be quite close to those found from finite element
analyses. The beam theory seems to be applicable for a range of S,/$2 between 1 and 4. If
there is no in situ normal stress across continuous joints, the tensile stresses produced by
shearing the rock mass will result in opening of the continuous joint. In this case, the theory
predicts tensile stresses of half the magnitude given by equation (15). These stresses are
still sumciently high to cause fracture of the rock block.
Significant tensile stresses a= are also developed on planes parallel to the continuous
joints when the rock mass is compressed in the direction of the continuous joints (by an
overall stress a~°). The maximum stress ratio (,,:),,,,/ax ° is plotted in Figs 8(a) and 8(b) for
various values of joint offsets and different ratios of still'nesses Kr:/K.v, (i.e. ratio of shear
stiffness of the continuous joint to the normal stiffness of the cross joint).
The maximum tensile stresses appear to be rather insensitive to the joint offset S and the
ratio of stiffness Kr:/l(,, as in the case of maximum tensile stresses caused by shearing.
it may be mentioned here that the joints are not subjected to any tensile stresses due to
compression of the rock mass, unlike the case of shearing discussed above.
Tensile stresses, and consequent opening of joints, in a block of a rock mass containing
staggered joints may be of signilicance in estimating the peak strength of rock masses.
326 B. SINGH

CONCLUSIONS
I. The constitutive equations have been derived to characterize an anisotropic rock
containing two orthogonal sets of staggered joints.
2. Continuum characterization of a rock mass requires the determination of new
parameters called 'joint stress concentration factors'. Approximate expressions
for the same have been de~'eloped and verified by the finite element analyses.
3. The off-diagonal terms of the compliance matrix of a rock mass are the same as
those of its rock component provided the joints are orthogonal.
4. Tensile stresses are developed inside the block of a rock embraced by staggered
joints, and may be as high as twice the overall shear stress or the overall compressive
stress.
Acknowledgments--The author gratefully acknowkdltes the assistance of Dr C. NEt.sos in evaluating the
stress concentration factors. The author also acknowledges the assistance of Professor C. FAnu4uItsr who
kindly edited the paper.
This research was supported in part by the Advanced Research Projects Agency of the Department of
Defen~ and was monitored by the United States Bureau of Min~, Twin Cities R e s t ' o h Center, under
Contract Number 140101610, Principal lnvestiptor--professor C. FAmuup.sr and in part by a IWant from
the University of Minnesota Computer Center.

REFERENCES
I. SUAaTP. Strength of weathered rock. Inc. J. Rock Mech. ?,fin. Sci. 7, 371-383 (1970).
2. S^LAMONM. D. G. Elastic moduli of a stratified rock ma~. Int. J. Ra~ck Mech. Min. Sci. 5. 519--527
(1968).
3. HILL R. Elastic properties of reinforced solids: Some theoretical principles. J. Mech. Phys. Solids II,
357-372 0963).
4. HtLL R. A self-gonsistgnt mechanics of composit© materials. J. Mech. Phys. So//d~ 13, 213-222 (1965).
5. IIAsmN A. and S~rrRIgMANS. A variational ~ h to the theory of the elastic behaviour of multi-
phase materials. J. ?,fech. Phys. Solids i I, 127-140 (1963).
6. E~IELBYJ. D. Elastic inclusions and inhomolpmcitics, Progress in Solid MectKmics ([. N. S~ddon and
R. Hill, Eds), Vol. I!, Chap. 3, Intersck~-e 0966).
7. Be:RANM. J. and MoLYNeUXJ. U.,~ of ¢"lassical variational principtcs to determine bounds for the
effective bulk modulus in heteroscncous media. Q. appl. Math. 24, IO7-118 (1966).
8. MOLYN£UXJ. and IkR,qN M. Stat~tical pfoperti~ of the streB and Stl~n fickls in a medium with small
random variatitms in elastic co¢ffgk.nts. J. Math. Mech. 14(3) 337-351 (1965).
9. KRONERE. Elastic moduli of perfectly disordered compmite maturiah. J. Mech. P~ys. So//ds 15, 319--329
(t967).
10. Su Y. L., WANG Y. J. and SrEEANgO R. Finite dement analysis of undeq~ound str~_~ utilizing
stochastically simulated material properties, Proceed/rigs of tie Eletwnth Syml~au~n on Rock MeekwMt's,
University of California, pp. 253-266 (1970).
! !. HEuzt' F. E. Tke Des/gn of Raom m / P i l l a r Structures is, Competem Jointed Rock, Ph.D. Theft, Uni-
versity of California, Appendix B, pp. 249-259 (19"/0).
12. CAt.COT~L. R. The Analysis o/Laminated Composite Structures, Article 2.3, Van Nostrand:Reit~old
(1969).
13. I . ~ t m r r ~ n S. G. Theory o/Elasticity of Anisotropic Elastic Body, (P. Fearn, Tramd.), pp. 31 & 54,
Holden-Day (1967).
14. TIMOStlENgOS. P. and ~ J. N. ?Twory o f Elasticity, 3rd edn, Chap. 3, McGraw-Hill (1970).

APPENDIX A
Derieotion o f dee Gewe¢~ Co~titutit~ Equattoas

The following analym ht haecd g l ~ m i a ~ oa the ~ work of Htt.t. [3,41 who thowod that the ttraiu
~ s y of a ~ body t a m to be equal t o . h e the product o.f.the a,~e. m l avenge
strains, provided the distribution of the surface tracttmt tends to be uaiform m the statiSttullsterne.Thls
approach has also been fellowed by SAt.~MON 12l and SMAaT [I].
C O N T I N U U M C H A R A C T E R I Z A T I O N OF JOINTED ROCK MASSES. PART [ 327

Vb-Vo
Ub- Uo
-b~, ~
Area dA-
i
u
i
i
i
w
~'
i
J

7
I

J i
i
i
p
i i, T
i :
z -i - ~ /

!z o
a
Im
.li
t ~ X ' 1"-- l"-'t
RG. 9(a). Deflected shape of unit cube due to overall stresses [,7°1.
FIG. 9(b). Stresses along joints due to overall stresses [~*].

Representatire voh~me, it has been assumed that the stress distribution along the joint is nearly uniform.
Hence, the size of the loaded area should be sufficiently large that stress gradients along the joints are not
significant. The contribution o f t b e strain energy due to the joint is then given almost uniquely by the average
magnitude of the stresses along the joint. The assumption is usually valid for the large prototype structures
considered in rock mechanics except in the special case of one joint set when rock layers are subjected to
significant bending moments. But one joint set is an exception rather than a rule. Further. tl~e joints should
pervade the entire rock mass under consideration. Thus, faults, shear zones and bedding planes (between
different rock formationsl are exchtded from the delinition 'joint" for the purpose of this continuum character-
ization. Furthermore, a representative volume only of the rock mass has been considered here; it contains a
completely representative set of joints. However. it is not necessary that all the joints in the reprc~ntative
volume be spaced uniformly and be of the ~ame still'm,.'ss.
Arcra,¢e strc'ss and arera,~,e strain. The volume of joints in a rock mass is usually negligible compared to the
volume of the rock. Moreover, the average of the stresses along both sides of a joint must be equal in order
to maintain equilibrium, tlence the overall average stress

[ ~ ' l = [a,l (A.I)

where the bar denotes the average value of the function.


Now consider a unit cube (i.e. with sides of unit length) of the rock mass. The cube contains a single set
of di~ontinuous joints which are parallel to one face of the cube [parallel to I m in Fig. 9(a)]. Figure 9(a)
shows arbitrary deflections imposed on the Ix)dy. Consider deflections along an element ab which inter,~ts
N joints. The net compression between a and b points is uniquely related to the strain (, in the rock and
compn,~sion t,j across the joints as follows:
N

V,- I/,=f,,d=+~rl. (A,2)


J : l

The average compre,s,sion may be found by integrating equation (A.2) over the area of the top of the unit
cube and is defined as overall strain ~°. Thus
N

,.o= f ( v 0 - V.)dA = ff..d:dA + ,a.3)


if there are no joints, vj is zero. So the second term of the right-hand side of equation (A.3) may be expressed
in term of average compression along joints 5 as
N

fZ~,dA = f ~aA, = A,O (a.4,


and hence equation (A.3) takes the form :
t, + A,ra -~ 17', -- 17, = ,.0 (A.5)
328 B. SINGH

where Aj is the total area of the joints within a unit cube. A similar relationship may also he proved for
overatl shearing strain.
~ -- Co = y.~" = ~,= :- Aid (A.6)
where ,~ is the averase shear displacement along joints. Further. the overall strain ~," will he equal to the
average strain ,r. in the rock since there would be no discontinuity in the horizontal strain field a/o,~ joints
within the unit cube.
The overall average of strains may be now expressed as follows:
[,°] ~ [~',] 4- .[t~,] (A.7)
where [fij] = [0, rS, d] and n is the average joint frequency which is equal to Aj i.e. total area of the joints
contained within a unit volume of the rock mass.
Stress concentration factor. Let the average str-"~es along a joint be ~s and ~a. These stresses generate
average displacement ~ and ~ according to the following constitutive relations:
Kra = ~, (A.8)
Kr,~ = ~ (A.9)
where K, and Kr are the normal and shear stiffnesses respectively of the joint. The strefaes along the joint
will be proportional to the overall average of the stresses, and are assumed to he related ~n~Hy let follows
[Fig. 9(b)]
• j = B,,~," (A.10)
and
~j : Br ~'.,'. (A.il)
The assumption has been made that the hydrecdatic overall stresses do not produce ~tear ~ ~ the
joint and, similarly, the overall shear stress~ do not produce normal s t ~ along the joint. T h ~ amump-
finn will be verified later on. Hut. [4] called the parameters 8 , and Br "'stn:~ ~ x n t m k m fit~toN", which
depend upon the configuration of inclusions (or joints) and their deformation ~ If the joints
shown in Fill. 9(b) are continuous, then the normal stress and shear stress ~ a j o l t will be exactly
equal to the over'aft normal stress and sbear stress rcspectively so as to satisfy the aplilibcium. CoMmlugntly
the stress cmgcntration factors are equal to unity in the case of o~,ntinuousjoints. T h ~ result is o f ~ b l e
importance in formulating the constitutive relations for a rock m a ~ having continuous joints.
Equations (A.8) and (A.9) together with equations (A.10) and (A.I I) may be combined in tbe following
matrix form:

:]
[~,] -~ [KjI-'tBjI[~'I (A.12)

where [K,]-' - [i ° I/K.

0 l/Kr

and [BjI - i - a,,

0
Constitutive equations. Let the compliance matrix of the rock be defined
[~,] ~ [M,][~,]. (A.13)

Substituting the vldtt¢ of Nraim [~,] from altmtioa (A.13) iato equation (A.7) and ~ , I t T ~ from equa-
tion (A.12), one thtaily obtains (remcmbe~ng from equation (A. l) that |6,] and [e"| ate ~ )
[.'] = {[/14r,] -I- n[Kj]-rIB, I} [0"] (A.14)
which yields the following compliance matrix for the rock mass (according to the definition ma~ compl/aoc=):
[M] = [M,] + n[K,]-'[~,]. (A.tS)
Since [Kj]- t and [Bj] are diagonal matriccs, the off..dis~ona]terms of the compliance matrices for the rock
CONTINUUM CHARACTERIZATION OF JOINTED ROCK MASSES. PART I 329

and the rock mass are the same. If the joints are continuous, the joint frequency n will be equal to the inverse
o f the joint spacing S, and [Bs] will have unit coefficients. Thus, the compliance matrix (for one set of
continuous joints) simplifies to the expression
I
[.Xtl = [.~t,1 - ~ [gsl-' (a.t6)

N o w consider another set o f orthogonal joints whose compliance is given by

"-K -I 0 ~ 1
[Ks: l - i = 0 (,A.17)

0 K;,'
Since the rock ma.~ containing the first set o f joints can be considered as a continuum, like the rock itself,
the modified compliance matrix [,~,lm:] o f the rock mass due to the presence of another set o f joints, may be
derived from equation (A.14) by replacing [ M , ] by [.tl|. Con~-qucntly it follows:
I
[Mmzl = [ M ] + ~ [Ks=]-'
or (A. 18)
I I
[M,z] = [,~4,] + ~, [K,,]-' + ~ IK, fl-'

where sub~:ripts I and 2 denote the first and second joint sets rcsr~-ctivcly.
Slocha.~licj(,inr .t5tlem. In reality, the stitTnc.-ss and spacing o f natural joints vary randomly. It is therefore
neccssary to consider the effect o f this on the continuum model.
The exact compliance matri,¢ can he derived for continuous but parallel joints. Consider the unit cube of
the rock mass which contains n joints of dilTcrcnt stilfncs.,~-'s [Ks] where j varit.~. I to n. Since the stresses D
and ~j along each joint will be exactly equal to the o~erall stru~s ~-,=" and ~,', acting on the top and bottom of
the unit euh~, the average displacements along each joint may bc obtained from the constitutive equations
of the joints
K r j 6 , :: r,=° (A.19)
Km ~+ = ~,°; i = I,n (A.20)
or simply in the matrb~ form from equation (A.12)
[as] = [Ks]-'[o'];j = I,n. (A.21)
Further, cquation (A.7) may be generalized easily with the help o f equation (A.3)

l,°l = [~,] + ~ [,is]. (A.22)

Substituting the value of [lis] from equation (A.21) and [¢,] in the above equation, the compliance matrix of
the rock mass may now he derived in similar fashion to that u ~ d for equation (A.I5)

[M] = [M,I + ~ [gs]-'

[M,] + n[Ks]- l (A.23)


wherc
0 0j
o X r '

and n ~ I/S = average joint frequency.


It is. therefore, proved that equations (A.16) and (A.18) can a l ~ predict the compliance o f a rock mass
containing a stt~chastic joint system, by using simply the mean values of joint spacing and compliance.
330 B. S I N G H

Compliance o f rock for ttw cases of plane stress or plane strain


It should be noted that the above analysis has not been restricted to isotropic material. In fact, it is not
necessary that the axes of anisotropy coincide with the joints. However proper complhtm~ matrix should he
used depending upon whether the rock is loaded in plane stress or plane strain. The compliance matrix of an
isotropic rock in plane stress is as follows for ready reference:

(M,] = ~ 1 1A.241

sym. 20 + v,)
and for plane strain

"1 --v, 0 "


i--v,

I m vez
I 0 (A.25)
[M,] = E,

2
sym. (1 -- v,)

Anisotropic joint model. The constitutive equations of a rock mass can also he obtained by considering the
joint to beof finite thickness 7". The volume of the joint is still assumed to he neillisible in comparison to the
volume of the rock mass. Let the characteristics of the joint material he defined as follows:
I
[Mj] = [At,] ~- ~ [K~J-*. (A.26)

According to HILL [4], the compliance of a heterot,cneous elastic material which contahts a set of parallel
cllipsoidal inclusions is given by equation (2). The volume Csof the inclusions (joints) per unit volume of the
matrix (rock mass) may he taken equal to TAj or Tn. On combining equation (2) with equation (A.26),
we obtain the following expression:
[M] : {M,] t- n[KA-'[BA (A.27)
which is precisely the same as equation (A. 15). Thus rock mass problems can also he studied with the help of
an anisotropic joint model whose duformation characteristics conform to equation (A.26).
Altcruatively for the spo=iai cas= of continuom joints, the elastic moduli of the rock mass can also be
computed from the constitutive ,~4uations for a stratilted rock mass (SALaMOn [2]). Conr~-ring n layers of
thicknt,.'ss 7" and compliance [Mj] as delincd by equation (A.26) and obs~rvinl~ the condition that nT
approaches z~ro, it may r~_adily be ~ that Salamon's comtitutive rehttions r c d t ~ to equation (A.27).

APPENDIX B
Strexg Coacentratio~f Factors for Rifid Rock with Stufgercd Joints
Figure ! shows a system of idealized joints, marked by double lines. The horizontal joint set is continuous
but the cross-joint set may be stasgen~. The spacini and stiffness are S,. KM., K r , for the extol joints and
St, K~z and Kzz for the horizontal joints. Ifa horizontal pfe~ure • is applied at tbe sm'face of the rock mass,
then the average normal stressesalong the cro.~ joints will be equal to Bs, ~ where B~: is the stress concentra.
tion f,,,-tor for normal stress. Similarly, if the rock mass is distorted by shear stresses 1"at the surface, the cross
joints will be subjected to an averatge shear stress Br,~' where B.r, is the stress concentration factor for shear
stress.
The following assumptions have been made in order to derive the exact expressions for the above mentioned
stress concentration factors:
(i) The stress concentration factors are the same for all ~.aross joints. Such a state of stress may be
achieved app<oximtcly by keeping a ~ distance between the loaded uarfac~.
(ii) The rock is a rigid body so that di~l~tiofts ~ by the flexibility of the rock are nqllilible in
comparison to the deflections alonl the j o i n . ~ the analysis will be valid when the stiffn¢~ of
the rock mass is small in comparison to that o f the rock.
(iii) The be s/m ttr~ses ane uJdqlkiently high to nmetralizv th¢ t c m ~ stress~ indu0ed by Mwarinlj of the
rock mass.
CONTINUUM CHARACTERIZATION OF JOINTED ROCK MASSES. PART I 331

I--s +- s,-s~
BOTTOM OF TOP BLOCK
---t- __.J
"2KN2~(1-SLY)- - 8
h,. ttttt, , , ,-,-,.o--g-
',

rST ,--~' r(2-Stt) S=


', '1 ~1, ', /
G •
£

-2KN2 8( I-S/Si. . .). ~'TOP OF BOTTOM BLOCK


(o) ENLARGED VIEW OF SECTION adeh (FIG.!)

.~1 I- ql-8, w

J.
-, S I--- ~ St-S ---'--- 5~/2

p q't"

(b) ENLARGED VIEW OF SECTION pqkn(FIG.I )

Fro. I0. Assumed displacement field in analysis of strc~s concentration factors for rigid ro~k.

Stress concentration factor f o r .vhcar .vtrc.~s


Figure IO(a) shows the rigid body displacements of a block abcdefgh marked in Fig. I. The deflections of
two sides of a joint arc plotted separately to iUustratc the net dellcction ax:ro~ thu joint. Let the vertical
dellcction at the corners of the block b¢ ,~. Since all the blocks will be dclh:cted in the .,~une way, the vertical
deflections at e.d will b¢ equal to those at h,a, i.e. S, but in thc opposite dircx:tion. Consequently the cross
joints will bc distorted by a shear displacement of 26 such that
B r l ~ = 25Kr=. (B.I)
For similar reasons, the vertical deflections at points g,b and f.c arc the same as t h o ~ of points a,h and
d,e respectively, again a. As such the net extension across the joint gh is
~ ~{I -- 2 S / S D or 26(I - S / S , )
and hence the tensile stress is
2;;(I -- S / S z ) K.v,
which is also equal in magnitude to the normal compressive stress along the joint ab. In case, the tensile
stress is more than the in sire vertical stress (primitive), horizontal joints will b¢ opened partly along gh and
cd.
Consider now the equilibrium of a part of the bkmk on the left-hand side o f the line joining g and b. Since
the average of shear stresses along gb must b¢ equal to ~-to maintain equilibrium on a plane passing through
the cross joints, the average shear stress on plane gb is found to be ~2 -- BrD. Equilibrium requires that
• B r i S z = ~2 -- Brt)S2 - 4KM, ~(1 -- S ] S t ) S . (B.2)
Substituting the value of 6 from equation (B.I), the stress concentration factor may be expressed as follows:
Br~ = [I + (I -- S/St) KN~S/S=Kr,I - t . (B.3)
It may be remarked here that these forces also satisfy the moment condition. Shear stresses have not been
shown along horizontal joint.
Since the cross joints are not subjected to any compression and the resultant normal force across the
continuous joints is zero. the assumption that no average normal stress along joints is developed due to pure
overall distortion is valid.
332 B. S I N G H

Stres~ coacewtrationfactor for ~ornml stre~


There will be no rotation of the block when only pressure is applied at the boundary of the rock mass.
It may be also noted that the lines pq and kn are the lines of syrnmm~ in Fig. I. Filure 10(b) shows the
rigid body displacements of the block. Let the horizontal displacements at I and m be 6t and/i, re=pectively
with respect to the centmlin¢ of the cross joint. Thus, one obtains from constitutive relations for the vertical
joints:
~t ÷ ~z = ~ B ~ t / K ~ . (B.4)
Compression of the cross joint will he followed by shear displacements ~iz and ~z along the horizontal
joints Ir and om respectively. Now the equilibrium of both the top blocks require
Krz~:$ = {=(2 -- B~t) - oB~: } Sz/2. (B.5)
and
Kraft(St -- S) = {o(2 -- BNt) -- oB.~t } Sz/2. (B.6)
Equations (B.4)-(B.6) may be solved to yield
B,~t = [l + (I -- $/St)KrzS/S,K~=] -t. (B.7)
it may be noted that the net shear force aloni; the horizontal joint is z=ro and the cro~t joint is not distorted.
Thus the assumption is justified that the overall normal s t ~ , ~ t do not produce any a v e r a ~ slqmr stre~
along the joints.

Influence of Rock Stiffness on Stress Cowcentration Factors for Staglrered Joints


it has been shown ¢arher that the stress concentration factors in~rcaz¢ am tbe stiJTn~s or the horizontal
joints decrease. Further the eft'oct of low rock stiffness is id=nt~al to the effect of low joint = z i l l l ' ~ on II1¢
ovel~ll ddollmlll~" y of ~ I"odt ~ II follows thai mdua:d stiffllem of Ihe rock will J ill ~ stress
concentration factors, provided other parameters remain unehanip:d. Cor~qucnfly realm ~ Ilmeratcd
in the rock will be of lower rnalp)ituth: than thou: pr,:dicted for the riilid rock. ~ exact ~ i t 1 ~ l ~ of stress
concentration factors may he obtained from finite element analysis ortbe rock mm~.
Since the ~ of the stiffn~t of the rock is o f . ~ m d a r y importance ~ to the ~ o f the
stiflrness of tbe JO~II, only II! approximate correction it ncedm;I to ~ th~ ~ f~r ~
concentration f l c t o ~ d~riw:d for rii$id rocks. Simplifying mmemptions ~ ~ Imem ~ to retain
the s i m p l i f y of the origk~l ~quatiom. Tbe rock has I ~ n a~mmcd to be ~ M llssti¢.
Filpure I I (a) shows the s t a l ~ joint sysl~n in which t he cross ~ cut t l ~ ~ ~ o f ~ block.
I.~ the a v ~ t l ~ horizontal str~,~s and shear ~tr~s in the rock he • and 1, re~ectively. Then the average
stn:s~ ~ the cross joinUt will be B~,~ and Brt~" according to the definition of ~ concentration
factors.

Stress concentratioelfactor for shear stress


Let the corners of the block undergo a vertical displacement 8 due to the shear stress v as shown in Fig.
I I(b). Only vertic=J dispblcements have hecn plotted. Evidently, the total shell" disp41malm~t 28 skou~ be
related to the sheeu"str~s along the cross joint as follows:
26 -- Brl f / g r t . (15.8)
If the avera~ c o ~ along the horizontal joint AB is |, then the joint CO will be ¢~jlctcd to an
averq= =zc~on or ! ~ of an~ym~m, ry. C o e ~ t ~ v . the =.vp~e ~ . _ ~ ,d~...ehe
joint A B ~ tiN=~ tansi~ s t ~ aloz~ joint ( : ~ wm I~ ~ to KH2|. Now, ~ ~ I t e a[l~ibrhtrn
of the half-block ABCD, the following relationship is established:
l"BriSz = ~(2 -- B r l ) S z - - K~z[SL. (B.9)
In the ~ of a rigid rock 8 win he qLml to 6; othenvi~ the top and bottom sw'fac~ of the block will
deform to rcmmb~ an S . ~ curve. Thus the averap compmzsion alons I]~ joint is
$ = 6 -- 26" (B.10)

where 6" is th¢ n ~ a~'~r'qlg ddk~tion due to the curved shape of the deflected Surface of the b4ock.
The followinl a m m ~ i o m are next made to lind ¥:
(i) The emmml ~ ~loqz the joint is uJifonn.
(ii) The block behaves as a one-dimensional beam with hinges at the center as well as at both ends.
CONTINUUM CHARACTERIZATION O F J O I N T E D ROCK. MASSES. P A R T I 333

dr" ,if' It" dr" d dr' d' d

> J ~ 1 t
1H 1 ~ I
tlELo~' It., .
i.
<
1 1 l H i L.
<
2 r

! (o)

NET DEFLECTION OF BLOCK


+XN,~ . . . . ~-T

q" ? T f i t
ttt t t ~I
,i
~~'"/~
I< s, >1 (b)

¢ (2--BUt) i O6NI

. .(2_llm) T
SlI2
I l j , "K ' • X Jr.

TOTAL DEFLECTION (c)

FIG. I I. A~,sumcd displacement tield in analysis of stress concentration factors for an ela~,lic rock.

T h e beam AB is loaded with uniformly distributed load o f 2KM~g. T h e well-known equation for deflection is
as follows:
PV
I' = ~ [l J r -- 2L1:3 + .r*l (B.II)

v, here
y = vertical deflection a l o n g the beam
w = 2K.2~ = uniformly distributed load
i = S~3/12 = m o m e n t o f inertia o f the beam
I = S , / 2 = total length o f the simply supported beam
x = distance along the beam from o n e end.
T h u s the average net deflection ~' is determined from the above e q u a t i o n :
~ ' = wl'/12OEl

K,,..5;a S , "
(B.12)
- 80 E, S~'"
334 B. S I N G H

Equations (B.8)-(B.IO)and (B.12) are suffu:ientto determine the stressconcentrationfactoras in equation


(13).
For the special case of a rigid rock, we obtain, by setting E, = ~ in equation (13).
K , : S£
B~.tt = I + 4Krt S:" (B. I3)

This is the same expre~ion as equation (B.13) ( $ t / S = 2). It may he noted that the aomud stiffne~ of the
cross joims does not x e m to affect the stress concentration factor. Further, the influence of the rock stifl'ne~
upon Brt diminish~ sub~antially when the modulm of deformation of the rock mass is reduced to 10 per
cent of that of the rock across the horizontal joints.
One may still improve the accuracy in evaluating the net deflection ~ by considering the actual distribution
of normal stress along the joint, viz. 2K,= (3 - 2y). However, the resulting expression for 8rt is likely to be
rather complicated.
An approximate ¢~timate of the maximum tensile stress can be made by applying the beam a na!ogy to the
rock block ABCD. Considering the equilibrium of forces acting upon this block, the total uniformly distri-
buted load w is found to be 41-(I - Brt). The maximum tensile stress may now be evaluated from the
maximum bending moment as follows:
wl" 6 3~" $1
Maximum fibre str¢~ = 8 $2" - 4 (I -- BTt)~-~2. (B.141

A better approximation may be found by calculating the maximum bending moment rain8 a distributed load
equal to 2K~z (3 -- 2y) as suggested above but the simplicity of the above expression is then lost.

S t r e ~ concentration factor fi~r m~rmal stress


Figure I I(c) shows two quarter blocks of rock joined by half cross joints. The stresses will he symmetrical
about the vertical line pas.~ing through the middle of the rock blocks. Let the whole piece of the rock he
deflected horizontally by a uniform displacement ~ as shown in Fill. | I(c). Horizontal ~ of average
magnitude e are thereby generated. From the definition of the stress concentration far'tot', it foltows that the
averal~ stress along the joint will be egt,, and so the stressalong AB in the rock will he Q(2 - - Bnt). Singe the
shear strex~ along lines of symmetry AB and AD are zero. the aver'alp: shear stress along the horizontal
joint El(:: is
~" = 2o(! -- B~,)S~/S,. (B.15)
The boundary conditions along quarter-block AFICD of the rock are (A is origin):
(i) x , - O . u =. 0 a n d ~',= ~ 0
(ii) : = 0 , v = O a n d ~ ' , = = 0
(iii) x = - 5 , 1 2 , ~, = oB,~s and ~',. = 0
( i v ) : = S=/2, ~= =: 0 and ~,= = ~"

Let the average horizontal displacement along 13(: and CD be rite and ~ a respectively. Hence the vertical
joint CD is subjected to a net compression of
8 -- uco = ~B.~t/2K~,. (B. 16)
Further, the average shear displacctrgnt along the horizontal joint BC will ix:
-- 2~ac = T/Krz. (8.17)
Equations (B.15)-(B. 17) may now be solved to give
I
B~t : (B.18)
Kr= S,

where

The ~ t f may be determin~l from the d i ~ t fmkl in the ~ A..I~ ~ to above


mentioned boundary conditions."the sokatlonhas bc~n obtained J n l Airy s sttqtu11~n~'q~ a ~ : i l ~
by TIMOOIENKO ~ OCN3Og~ [141.
Consider the following Airy's streu function which satisfies the biharmonic equation:
÷ = c(xe/60 - - x4zZl8 4" z*/120 -F $1=X=:ZJl6 -- SiZz*/48) + d:z/2 + bxa/2. (B.19)
CONTINUUM CHARACTERIT'-~TION OF JOINTED R O C K MASSES. P A R T I 335

Stress m a y now be f o u n d after differentiating the above e q u a t i o n :

~ -- ~.:z - c ( - - x ' . 4 - StZx3;'8 - - St3zz/4 -:- : ' . 4 ) - d (B.20)

~.'~,
~z -- ~'x" - - c( = . r a . _"- 3 x ' -° " , -"* + S t Z z 2 ; 8 ) +-b (B.21)

--?'24~ -- c ( -- x J : -- St2.rz/4). (B.22~

it m a y be noted that the shear stresses sat isfy the b o u n d a r y conditions. T h e c o n s t a n t c is determined from the
condition i~;
0
j 'r,:dx = r S t / 2 ; z = S-,/2
-- St."2
a n d so
c = 6 4 T / S , S t 3. (B.23)

The displacement field u(.r,:) m a y be evaluated after integrating


/.~u !
.-- = % -- (=z -- v,o.) (B.24)
t,'.r ~ "

a n d substituting stress from e q u a t i o n s (11.20) anti (B.21) into the above e q u a t i o n :

t~ = - c/=', ~ - . r " ( I i ."*v , ) / ."*O i" :ax,: 4 - St zA'Z:(2 } v,)/8 t v ,.r a-:,'~
. ;, !. St2r~/24 ,, t_ x ( d - b y , ) . (B.25)

Similarly the displacement field t'(x.:) m a y be obtained. As c a n be readily vcrilied, they ~ t i s f y the necessary
~ t m d a r y conditions.
Equation (B.25) m a y now be used to c o m p u t e the average displacement tiac and lieD in o r d e r to derive the
stress concentration factor from equation (B. 181. T h u s

li,c = ~ t :/ O

u(.r,z) d x ; : .= S:/2.
-Sd2

cS,' Iv,-2 S:" 32 ]


= 128k.; - ~ 2St" ~' (.5Or, ÷ 2) S:2
St z cSt" (d - b,',] (B.26)

and
$d2
2 f u(.r,z)dz; x = - St~2.
0

cSt ~ [(I + 2v,) I 5" "~ 2s_, ~ 32(,_t -_t,v,q. (a.27)


64Ev L 1 0 3 lOSt 2 + 3St a cSt'* J

It follows from e q u a t i o n (B.23), (B.26) a n d (B.27):

2a,~ - a¢. = ~ + 3s, 2 t5 i sd 5s,'J ~- E + l-b--s?J" s , > s2 (B.28)

which, u p o n eliminating "r from e q u a t i o n s (B.18) and (B.28), leads to the final expression for the stress c o n -
centration factor as in equation (14).

You might also like