Mechanics of Engineering Materials

Edited by C. S. Desai and R. H. Gallagher

© 1984 John Wiley & Sons Ltd

Chapter 3

Microplane Model for

Strain-controlled
Inelastic Behaviour
z. P. Bazan!

3.1 INTRODUCTION
Various heterogeneous brittle aggregate materials such as concretes, rocks, or sea
ice, are inelastic but cannot be described as plastic, except at extremely high
hydrostatic pressures. A characteristic property of such materials is that they
exhibit strain-softening, i.e., a decline of stress at increasing strain, which results
from progressive development of fracture. Since these materials can undergo
strain-softening within a relatively large zone, a non-linear triaxial constitutive
relation is needed for its description. There are, however, some important
differences from the classical modelling of inelastic behaviour, i.e., from the theory
of plasticity.
First, rather than determining the inelastic phenomena in terms of stresses, as
in plasticity, one must determine them in terms of strains. This is because in terms
of stresses the description is not unique, as two strains correspond to the same
stress, in the case of strain-softening, while still only one stress corresponds to a
given strain. Second, the normal inelastic strains are, in contrast to plasticity,
important, in fact dominant. They describe the cumulative effect of rnicrocrack-
ing. Third, the inelastic phenomena are highly oriented and happen almost
independently on planes of various orientation within the material as a function
of normal strains across the planes.
In the present work, it is proposed to describe this behaviour independently on
planes of various orientations in the material, called micro planes, and then in a
certain way superimpose the inelastic effects from all the planes. This type of
approach has a long history. First proposed in 1938 by Taylor [1], the idea was
exploited by Batdorf and Budianski in their slip theory of plasticity [2]. A
number of subsequent investigators adopted this approach for plasticity of
polycrystalline metals [2 to 6]. Zienkiewicz and Pande [7] and Pande et al. [8-9]
developed an approach of this type in their multilaminate models for rocks and
soils.
46 Mechanics of Engineering Materials Microplane Model for Strain-controlled Inelastic Behaviour 47
In the aforementioned models, the stress on each plane within the material is pieces are nearest. The deformation of the thin layer of matrix between two
assumed to correspond to the same macroscopic stress and the inelastic stresses aggregate pieces, which is the chief source of inelastic behaviour, seems to be
are superimposed. As mentioned, however, for certain materials the inelastic determined mainly by the relative displacements of the centroids ofthe aggregate
behaviour is predominantly strain-controlled, and it is then more appropriate to pieces, which roughly correspond to the macroscopic strain. The microplanes
assume that strains, not stresses, correspond on the planes of all orientations to may be imagined to represent the thin layers of matrix and the bond interfaces
the same macroscopic strain. In this case it is necessary to superimpose in some between the adjacent aggregate pieces (Figure 3.1 (a)), since microcracking is
way the inelastic stresses (relaxations) from the planes of all orientations. This chiefly concentrated there.
approach was adopted for concrete and geomaterials in ref. [11] and was According to Hypothesis I (equation (3.1 », the virtual work of stresses per unit
summarized in ref. [12]. In these works, the inelastic shear stresses on planes of all volume may be written as 8W = Ci/jaij = f.~j8a~j + eil8aiJ, in which erj and eil
orientations within the material were neglected. However, although their role is represent the strains associated with the additional elastic stress and the stress
no doubt secondary, in case of concrete and geomaterials at high hydrostatic resulting from the micro planes. At the same time, 8W = eijbat + EijbaiJ. Since
pressures, they certainly have some effect. The purpose of this work is to both expressions must hold for any ba~j and any bail, we must have E~j = EiJ = Eij'
generalize ref. [11] to include the effect of inelastic shear stresses. According to Hypothesis II, the components of the strain vector en on any
micro plane are
3.2 BASIC HYPOTHESES (3.2)
The macroscopic stress tensor will be denoted as aij, and the macroscopic strain in which ni are the cosines of the unit normal to the microplane. The normal
tensor as Eij' With regard to the interaction between the macro- and micro-levels, microstrain, i.e., the normal component of strain vector en, may be denoted as eN,
the following three hypotheses may be introduced. and the components of the vector of the shear component &T may be denoted as
Hypothesis I. The tensor of macroscopic stress, aij' is a sum of a purely elastic
macrostress alj that is unaffected by inelastic processes on planes of various
(0') (b)
orientation, and an inelastic macrostress 1:ij which reflects the stress relaxations
from micro planes of various orientations, i.e.,
(3.1)
(latin lower case subscripts refer to cartesian coordinates X;, i = 1,2,3).
Hypothesis II. The normal microstrain EN and the shear microstrain ET on
each micro plane of any orientation is the resolved component ofthe macroscop!c.
strain tensor Eij'
Hypothesis III. There exist an independent stress-strain relation for each
microplane of any orientation.
Hypothesis II is opposite to that made in the slip theory of plasticity, in which
the stresses rather than strains on the planes of all orientations are assumed to be
the resolved components of the macroscopic stress. One can offer three reasons
for this. First, if the material state were characterized by stress rather than strain,
the description would not be unique since, in the case of strain-softening, there are
two strains corresponding to a given stress. Second, the relationship between the
micro- and macro-levels would not be stable in the case of strain softening, which
y=x
has been confirmed numerically. Third, the use of resolved strains, rather than 2

stresses, appears to reflect the microstructure of a brittle aggregate material more

realistically. In contrast to polycrystalline metals, brittle aggregate materials x=x,
consist of hard inclusions embedded in a relatively soft matrix. The microstresses
Figore 3.1 (a) Example of idealized microstructure, (b)-(c) explanation of notations, (d) stress-strain
are far from uniform, having sharp extremes at the locations where the aggregate relation on a microplane (<T" = 'N,E" = EN)
48 Mechanics oj Engineering Materials Microplane Model Jor Strain-controlled Inelastic Behaviour 49
With regard to the elastic parts of the shear components of stresses and strains
BT;'
Figure 3.1 (c» £T=S-£N, and its magnitude is BT=[ls"12 -(BN)2]1/2 or
on the micro plane, we wiII now introduce an additional assumption, namely that I2
GT= ISTI = (BT/TY • Thus, we obtain the following expressions for the vector
the vectors of these shear components are parallel. This precludes anisotropic and the magnitude of the shear strain component on the microplane:
behaviour within each microplane, although overall anisotropy remains possible
by considering different properties on various microplanes. According to BT, = (nk8ij - njnJnJB jk (3.9)
Hypothesis III we may now write BT = [njEjjllk(Bjk - IIjn m Bkm )] 1/2 (3.10)
iN = CE N - iN According to equation (3.6), the normal and tangential components of the
(3.3)
iT, = BET, - iT, stress relaxation rate on the micro plane can be expressed as
in which C and B are the elastic constants for the normal and shear response on
the micro plane, and TN and TT, are the inelastic stress relaxations in the normal (3.11)
and tangential directions on the microplane. Superimposed dots denote
time rates. For the magnitudes of the shear components, equation (3.3) implies The derivatives of the inelastic potential in these equations may be calculated
i = BiT - t;.. as
Further we need to specify the inelastic stress relaxations. For this purpose, we
aJp aJp aBT alp a 1/2 oJp BT
assume the existence of inelastic potentialsJp and loading surfaces gp({3 = 1, ... ,11) - = - - - = - - ( B T BT) =---'. (3.12)
OBT, OBT OET, OBT OET, J J OET BT
for each microplane. They must be defined in terms of strains rather than stresses,
i.e., We see that the vector normal to the potential surface i p is parallel to the vector
!P(BN,f.T) =0 ({3=I, ... ,n) (3.4) of the tangential component of strain on the microplane, i.e.,

gp(E N, BT) = 0 (3.5) (3.13)

The rates of inelastic stress relaxations may be assumed to be given by the The derivatives of the inelastic potentials and loading functions appearing in
equations (3.6)-(3.7) may be calculated as
normality rule
."
Tjj=
" ar
\" JP·
L.... -f./.p (3.6)
oJp
~
oJp
= Pij~ + %.,,-,
,oJp ogp ogp ,ogp
-~- = Pij-:)- + qjj-o- (3.14)
p~ I aBij ~ij u~ u~ (~ (~ v~

in which we introduce the notation

(3.7)

(3.15)
in which lip are material softening parameters depen?ing on .t?e current stat~ of
the material and possibly also its history, and H IS Heavlslde step funchon.
From equation (3.8) we can further calculate
Similarly to Drucker's stability postulate in plasticity, equations (3.6)-(3.7) can be
easily derived from a more plausible hypothesis (postulate) for a strain cycle,
called II'yushin's postulate, as previously used for macroscopic inelastic theories (3.16)
based on loading surfaces in the strain space [13 to 20]. The Heaviside function in
while from equation (3.10) we obtain
equation (3.7) distinguishes between loading and unloading.
By projecting the microplane strain vector (equation (3.2» on to the direction n , 1 a
of the normal, we obtain the magnitude of the normal strain component on the % = 2-.,,-(n p i:qp l1,i:q, - I1 pllqBpql1,l1 s C,,)
BT uBij
microplane and its vector: which reduces to
(3.8)
(3.17)
The magmtude . af ' vector on t h
the stram '
e mlcrop . 1"1
Iane IS £ = ("
EjEj")1/2 =
(njEj;llkEjk) I 12. The vector of the tangential (shear) strain component is (see
The last tensor is non-symmetric. Later we will need its symmetric part, which
50 Mechanics of Engineering Materials Microplane Model for Strain-controlled Inelastic Behaviour 51
reads Furthermore, using equations (3.14) and (3.7), we may express

(3.18) f alp. f
L... -,;- Jl.,/)Eij = L... hp [afp - afp]
Pij,,- + (qij + %)-,;-
p= 1 UEij p= 1 UEN UeT

in which we introduce the notation

(3.26)
(3.19)
We need to establish now the equilibrium relation between the microstresses in which qij is the antisymmetric part of qij, i.e., qij = qij - qij. Noting that the
on the microplanes of all orientations and the macroscopic stress tensor. We may antisymmetric parts give zero products with symmetric tensors, i.e., qiji5eij = 0,
use for this purpose the principle of virtual work, which requires that the virtual qkmDEkm = 0, we find that
work of macroscopic stress rates on any macroscopic strain variations within a f
L...
afp. R .
-ae Jl.p = ijkmEkm (3.27)
small unit sphere of unit radius be equal to the virtual work done on all the p= 1 ij
microplanes tangential to the unit sphere. This condition may be written as in which
follows

Jsr
(3.28)
. 4n
DW = 3 iiji5eij = 2 (i N i5e N + i T i5eT )f(o)dS

= 21 (C6NDEN + BBT,DeT; - iijDEij)f(o)dS (3.20)

Substituting equations (3.27) and (3.25) into the variational virtual work
relation in equation (3.21), and noting that this relation must hold for any
variation De ij, we find that
where S is the surface of a unit hemisphere, and f(o) describes the frequency of (3.29)
microplanes as a function of orientation o.
Substituting here fro.n equations (3.8), (3.9), and (3.6), we may obtain the or
relation (3.30)
in which
(3.21)

in which
DUkm = ;n Is (aijkm + bijkmB - Rijkm)f(o)dS,
C

bijkm = (nmD'k - n,nkn",)(njD'i - n,ninj) mjkm = }n Is (aijkm + bijkmB)f(o)dS

C (3.31)
= njn",Dik - njnknmni - n",ninjn k + ninjntn",(n,n,) (3.22)
= Diknjn", - aijk", iii = ;n Is Rijkmf(o) dSekm (3.32)
This fourth order tensor is symmetric when ij is interchanged with km but non-
symmetric when i is interchanged withj or k is interchanged with small m. The Here DUk'" is the tensor of tangential moduli corresponding to the microplanes,
tensor may be written as a sum of a symmetric part and an antisymmetric part, Dj}k", is the elastic part of this tensor, and 67j is a tensor of the rate of inelastic stress
relaxation. The integrals in equations (3.31) and (3.32) extend over the surface S of
(3.23) a unit hemisphere.
in which the symmetric part is Consider now the special case of isotropic materials, for which f(o) = 1. For
this case the elastic stiffness matrix Djjk", must be equivalent to an isotropic
bijk'" = i(Diknjn", + Djknin", + Di",njnk + Dj",nink) - aijk", (3.24) material stiffness matrix characterized by some shear modulus am and Poisson
For the antisymmetric part it is true that bijkmDeiikm = 0 for any Deij . Therefore, ratio vm • Their values may be easily calculated. To this end, consider a uniaxial
strain rate 633 = 1 while all other components of 6ij = O. We may now substitute
(3.25) n 1 = sin rjJ cos e, n2 = sin rjJ sin e, n3 = cos rjJ and equations (3.29) and (3.31) for
52 Mechanics of Engineering Materials Microplalle Model for Strain-controlled Inelastic Behaviour 53
iii = 0 then give This relation permits us to choose the elastic constants of the material as a

i33 = }n [C f: f: cos 4
cfJ sin cfJ dcfJ d(H B f: f>OS2 3
cfJ sin cfJ dcfJ de ]e33
whole, as well as of the microplane system, and also choose the Poisson ratio
for the additional elastic stress.
In general, the total tangential elastic moduli are
=! (3C + 2B) (2.33)
(3.38)
ill
3
= -(C - B)
fn fn sin 3 cfJ cos 2 2
cfJ cos OdOdcfJ (;33 = t(C - B) (2.34) For isotropic materials, Drikm represent the elastic moduli tensor of an
2n 0 0 isotropic material;
According to Hooke's law, i II/i33
= ym/(l - ym) for uniaxial strain, and so
ym = i I d(i II + i 33)' Substituting from equations (3.33) and (3.34), we thus get (3.39)
C-B
vm = _ __ (2.35) in which bij = Kronecker delta = 1 if i = j, and 0 if if- j.
4C+B
Furthermore, according to Hooke's law we have, for uniaxial strain, 3.3 CASE OF ZERO SHEAR RELAXATIONS ON MICROPLANES
vm = i I t/(i II + i33 ),
from which we may solve
For tensile strain-softening of concrete, it seems that one may neglect the shear
m
Gm = 1 (I - 2v )(3C + 2B) (3.36) stress relaxations and consider only the normal stress relaxations on the
m
10 (I - v ) microplanes, which correspond to the formation of microcracks in the direction
From equations (3.35) and (3.36) we may solve the constants C and B from of the micro planes. In this case B = !T; = !T; = O. One may consider here for
desired values of Gm and ym. Equation (3.35) yields the following values of each microplane only one loading surface (P = 1), fl = gl = CN = const. In this
case we get
Poisson ratio:
B/C=O vm =0.25 (3.40)
14/59 0.18 The relationship between the normal stress and the normal strain on the
1 o microplane may be conveniently described by the formula
00 -1 (n "" 2) (3.41)
It is interesting to observe that Poisson ratios greater than 0.25 cannot be.
obtained. The range appears suitable for geomaterials.
In some situations, however, an adjustment of the Poisson ratio provided by 300
the system of micro planes may be needed. For example, one might desire for
some material an overall Poisson ratio v> 0.25, or one might simply need for ;;;
c.
the best fit of test data a different Poisson ratio for the rnicroplane system than :: 200
for the material as a whole. Such an adjustment of Poisson ratio is made possible ~
by equation (3.1) (Hypotehsis I). Let the additional elastic stresses uri be given cii Cb/Co • 1.000000008
Ce/C o • 1.000000009
by an isotropic stress-strain relation to Eij, characterized by shear modulus Ga
and Poisson ratio va. Then, for uniaxial strain E33 = 1 (Ell = e22 = 0) we have
a33 = 2G(1 - v)/(l - 2v). Summing the stress from the microplane system and
'. Cdl Co • 1.000000002
C.I Co • 1.000000008
Cfl Co • 1.000000008
CQI Co • 1.000000005

the additional elastic stress, we also have a33 = 2Ga(1 - v")/(l - 2v") +
2Gm (1 - vm)/(I - 2v m). From these equations we may solve 0.0002 0.0004 0.0006 00008
Strain
Ga =
a m
1 - 2v (G(l - v) _ G (1- V
m

m
») (3.37) Figure 3.2 Distribution of integration points for 2 x 25-point formula defined in Table 3.1, and
1 - va 1 - 2v 1 - 2v response curves for uniaxial stress applied at directions a, b, .. ,f.
54 Mechanics of Engineering Materials Microplane Model for Strain-controlled Inelastic Behaviour 55

in which C, k, and n are material constants. For large eN the value OfrN becomes Table 3.1 Direction cosines and weights for 2 x 25 points with error of 10th order (after Bazant and
Oh [11],
essentially zero (n = 2 is a suitable exponent). Comparing this with
equation (3.40), we have IX x~ xi X; w"

hi = C - F'(eN) = C - C(1 -kneN)exp( - keN) (3.42) I 1 0 0 0.01269841058

2 0 1 0 0.01269841058
The present special case has been considered in refs [11 and 12], and good 3 0 0 1 0.01269841058
4 0.7071067812 0.707106 7812 0 0.02257495612
fits of various tensile strain-softening test data have been demonstrated. Since, - 0.707106 7812 0 0.022574956 12
5 0.7071067812
in absence of shear relaxations, the microplane model yields Poisson ratio 0.25, 6 0.7071067812 0 0.7071067812 0.02257495612
the additional elastic deformation was used to correct this value to 0.18, typical 7 0.7071067812 0 - 0.707 106 7812 0.022574956 12
8 0 0.707106 7812 0.707106 7812 0.02257495612
of concrete. There was a slight difference from the present formulation in that - 0.707 106 781 2 0.022574956 12
9 0 0.707106 7812
the superimposed additional elastic value was strain rather than stress. The 10 0.301511 3354 0.301 511 3354 0.904 534039 8 0.02017333557
present formulation is, however, more efficient. One of the comparisons with 11 0.3015113354 0.3015113354 - 0.904 534039 8 0.02017333557
12 0.301511 3353 - 0.3015113354 0.904 534039 8 0.02017333557
the test data of Evans and Marathe [21], made in ref. [11], is reproduced in - 0.904 534039 8 0.02017333557
13 0.3015113354 - 0.301 511 3354
Figure 3.2, and the associated stress-strain curve considered for the microplanes 14 0.301511 3354 0.904 5340398 0.3015113354 0.020 173 335 57
is shown in Figure 3.1(d). 15 0.301511 3354 0.904 534 039 8 - 0.3015113354 0.020173 335 57
16 0.3015113354 - 0.904 534 039 8 0.3015113354 0.020 173 335 57
17 0.301511 3354 - 0.904 534 039 8 - 0.3015113354 0.020 173 335 57
18 0.904 534 0398 0.3015113354 0.301511 3354 0.02017333557
3.4 NUMERICAL INTEGRATION ON - 0.3015113354 0.02017333557
19 0.904 534 039 8 0.3015113354
THE SURFACE OF A SPHERE 20 0.904 534 039 8 -0.3015113354 0.301 511 3354 0.02017333557
21 0.904 534039 8 -0.3015113354 - 0.3015113354 0.02017333557
In general situations, the integral in equation (3.31) over the surface of a unit 22 0.577 350 269 2 0.577 350 269 2 0.5773502692 0.02109375117
hemisphere has to be evaluated numerically, approximating it by a finite sum: 23 0.577 350 269 2 0.577 350 269 2 - 0.5773502692 0.02109375117
24 0.5773502692 - 0.577 350 269 2 0.5773502692 0.02109375117
n 0.02109375117
25 0.577 350 269 2 - 0.577 350 269 2 - 0.577 3502692
DVkm = I 6w.[(a jjkm C + bjjkmB - Rjjkm)f(n)]., I w. = t (3.43)
a=l a (J = 25.239401°.
in which (X refers to the values evaluated at certain numerical integration points
on the spherical surface (i.e., certain characteristic directions), and w. are the orientations of the uniaxial stress with regard to the (X-directions. Ideally, the
weights associated with the integration points. In finite element programmes for response curves for any orientation should be identical. However, large dis-
incremental loading, the numerical integration needs to be carried out a great . crepancies are found for a ten-point formula.
number of times. Therefore, a very efficient numerical integration formula is . Batant and Oh [23] derived numerical integration formulas with more than 10
required. For the slip theory of plasticity, a similar integration was performed points, which give consistent results even in the strain-softening range. The most
using a rectangular grid in the plane of spherical coordinates e and cPo This efficient formulas, with an almost uniform spacing of (X-directions, are obtained
approach is, however, computationally inefficient since the integration points are by certain subdivisions of the faces of an icosahedron or a dodecahedron [23].
crowded around the poles, and since, in the e- cP plane, the singularity arising Such formulas do not exhibit orthogonal symmetries. Other formulas which do
from the poles takes away the benefit from a use of a higher-order integration were also derived [23]. Taylor series expansions on a sphere were used and
formula. weights w. were solved from the condition that the greatest possible number of
Optimally, the integration points should be distributed over the spherical terms of the expansion of the error would cancel out. The angular directions of
surface as uniformly as possible. A perfectly uniform distribution is obtained certain integration points were further determined from the condition that the
when the micro planes normal to the (X-directions are the faces of a regular error term of the expansion be minimized. In this manner, formulas involving 16,
polyhedron. However, a regular polyhedron with the greatest number of sides is 21,25,33,37, and 61 points were established, with errors of 8th, lOth, and 12th
the icosahedron, for which N = 10 (2N is the number of faces), and this number order. Table 3.1 defines one of these numerical integration formulas, having 25
appears insufficient (a formula for this case was presented by Albrecht and points for a hemisphere; this formula exhibits orthogonal symmetry [23]. The
Collatz [22]). The need for greater accuracy is indicated when the response directions of the integration points are illustrated in Figure 3.2, and also shown
curves in a uniaxial tensile test with strain-softening are calculated for various are the stress-strain diagrams calculated for various directions of uniaxial tensile
56 Mechanics oj Engineering Materials
Microplane Model Jor Strain-controlled Inelastic Behaviour 57
4oor-----------------------------, m
(d) in whichk 1 = 2A(kT/h)t- exp( - Q/RT) and k2 = VJRT. Here Tis the absolute
temperature, Q is the activation energy of creep, R is the universal gas constant,
k is the Boltzmann constant, h is the Planck constant, V. is the activation
volume, and A,m are empirical constants. For the vectors of the tangential
';;; stress and strain components on the micro planes, equation (3.44) may be
..
0.

~200
generalized, in the inverted form, as follows

Vi
- Theory LT,
1 RT.
=k--:-'-smh- 1 (iT)
- (3.45)
2 eT kl
---- Evon., Morothe!l968)
This equation now replaces equation (3.3) of the present model, and
equations (3.6)-(3.11) become unnecessary. By following the same analysis as
before, one obtains the macroscopic stress-strain relation, replacing
°O~~--~--~O~,O=O~04~L-~--~~O,~OOO~8~ equation (3.29) as follows
Strain

Figure 3.3 Comparison with test data of Evans and Marathe [21], after Botant and Oh [11] '1jkrs -
-3f 1 . -1(iT)
2 bjkrsk--;-smh k- J(n)dS (3.46)
1t s 2 eT 1

stress with regard to the integration points (directions a, b, c, d, ...); the spread of Here '1jkrs represents the fourth order tensor of current viscosities, and J(n)
the response curves characterizes the range of numerical errors. For crude represents the distribution function for the frequency of clay platelets of various
calculations, the lowest required number of integration points is 16 [23]. orientations. For some clays, this distribution function has been measured
experimentally, using X-ray scattering techique. Applicability of equation (3.43)
3.5 APPLICATION TO ANISOTROPIC CREEP OF CLAY to test results is presently being studied at Northwestern University by J. K. Kim.
Complete description of clays further requires superposing equations for the
As another application, we may demonstrate an adaptation of the micro plane volume change. This may be best accomplished on the basis of the critical state
model to describe creep of an anisotropically consolidated clay and to correlate theory, for example, in a manner recently described by Pande et al. [8-9].
the stress-strain relation to known information about the distribution of the
frequency of platelets of various orientations within the clay. This problem has
been studied, for example, in ref. [24], using a micro mechanics model in which ' 3.6 CONCLUDING REMARKS
triangular cells of mutually sliding clay platelets are constrained to the same The micro plane model allows great versatility in constitutive modelling. The
macroscopic strains BU' same as here. It appeared, however, that this approach present form of the model, in which the strains on microplanes of all orientations
becomes quite complicated in the three-dimensional case, although it is not correspond to the same macroscopic strain, appears suitable for materials which
very difficult for two-dimensional analysis. In three dimensions, the present type exhibit progressive microcracking and tensile strain-softening. Constraining the
of micro plane model seems appropriate. microstructure to the same macroscopic strain is also important for numerical
In treating clay, the stress tensor aij must be interpreted as the effective stress reasons, not merely for the purpose of stability and uniqueness of representation.
tensor, i.e., aij = tij - DijP, in which P = pore-water pressure and tij = total stress The computational work required by a model of this kind is not as large as one
in the solid-water system. Let us consider only the case of deviatoric creep, for might think. The work required is greatly reduced by the recent development of
which the normal stiffness on the microplanes may be neglected, i.e. C = L~ = O. efficient numerical integration formulas for a spherical surface.
The microplanes of the present model may be interpreted, in the case of clay, One advantage of the model is that the stress-strain relations are primarily
as the planes of sliding in contact of adjacent clay platelets. As is well known, defined on the microplane level, on which one does not need to heed the tensorial
the sliding is governed by the rate-process theory, which yields the relation invariance requirements which are a source of great difficulty in constitutive
(3.44) modelling. Tensorial invariance is ensured subsequently, by combining the
responses from microplanes of all orientations within the material.
58 Mechanics of Engineering Materials Microplane Model for Strain-controlled Inelastic Behaviour 59
ACKNOWLEDGEMENTS 15. Z. P. Bazant and S. S. Kim, 'Plastic-fracturing theory for concrete', J. of the Engng.
Partial support during the writing of this work was obtained under Air Force Mech Div., Proc. Am. Soc. of Civil Engrs., lOS, June 1979, pp. 407-428, with Errata in
Vol. 106 (also as Preprint 3431, ASCE Annual Convention, Chicago, Oct. 1978).
Office of Scientific Research Grant No. 83-0009 to Northwestern University. The 16. W. D. Iwan and P. J. Yoder, 'Computational aspects of strain-space plasticity', J. of
preceding development of the mathematical model was partially supported Engineering Mechanics, Proc. ASCE, 109, 1,31-243 (1983).
under US National Science Foundation Grant No. CEE800-9050 to 17. P. J. Yoder and W. D. Iwan, 'On the formulation of strain-space plasticity with
Northwestern University. Mary Hill is thanked for her outstanding secretarial multiple loading surfaces', J. of Applied Mechanics, Trans. ASME,48, 773-778 (1981).
assistance. 18. P. M. Naghdi, 'Some constitutive restrictions in plasticity', in 'constitutive Equations
in Viscoplasticity' in Computational and Engineering Aspects, AMD, Am. Soc. of Mecho
Engineers, 20, 79-93 (1976).
19. P. M. Naghdi and J. A. Trapp, 'The significance offormulating plasticity theory with
REFERENCES reference to loading surfaces in strain space', Intern. J. of Engng. Science, 13, 785-797
(1975).
1. G. I. Taylor, 'Plastic strain in metals', J. Inst. Metals, 63, 307-324 (1938). 20. J. Casey and P. M. Naghdi, 'On the characterization of strain-hardening plasticity', J.
2. S. B. Batdorf and B. Budiansky, 'A mathematical theory of plasticity based on the of Applied Mechanics, Trans. ASME, 48, 285-296 (1981).
concept of slip', NACA TNI871, April, 1949. 21. R. H. Evans and M. S. Marathe, Microcracking and stress-strain curves for concrete
3. T. H. Lin, Theory of Inelastic Structures, Chapter 4, John Wiley, New York, 1968. in tension', Materials and Structures (Paris), No.1, 61-64 (1968).
4. T. H. Lin, and M. Ito, 'Theoretical plastic stress-strain relationships of a polycrystal', 22. J. Albrecht and L. Collatz, 'Zur numerischen Auswertung mehrdimensionaler
Int. J. of Engng. Sci., 4, 543-561 (1966). Integrale', Zeitschrlft Fur Angewandte Mathematik und Mechanik, Band 38, Heft 1/2,
5. T. H. Lin and M. Ito, 'Theoretical plastic distortion of a polytcrystalline aggregate Jan./Feb., pp. 1-15.
under combined and reversed stress', J. Mech. Phys. Solids, 13, 103-115 (1965). 23. Z. P. Baiant and B. H. Oh, 'Efficient numerical integration on the surface of a sphere',
6. J. R. Rice, 'Continuum plasticity in relation to microscale deformation mechanisms', Report, Center for Concrete and Geomateriais, Northwestern University, Evanston,
in Metallurgical effects at High Strain Rates (Eds. R. W. Rohde et al.), Plenum 111., 1982.
Publishing Corp., New York, 1982. 24. Z. P. Baiant, K. Ozaydin and R. J. Krizek, 'Micromechanics model for creep of
7. O. C. Zienkiewicz and G. N. Pande, 'Time~ependent multi-laminate model of anisotropic clays', Journal of the Engineering mechanics Division. ASCE, 101,57-78
rocks-a numerical study of deformation and failure of rock masses', Int. Journal of (1975).
Numerical and Analytical Method In Geomechanics, 1,219-247 (1977). 25. C. R. Calladine, 'A microstructural view of the mechanical properties of saturated
8. G. N. Pande and K. G. Sharma, 'Multi-laminate model of clays-a numerical clay', Geotechnique, 21, 391-415 (1971).
evaluation of the influence of rotation of the principal stress axes', Report, 26. D. N. Wood, 'Exploration of principal stress space with kaolin in a triaxial apparatus',
Department of Civil Engineering, University College of Swansea, UK, 1982; see also Geotechnique, 25, 783-797 (1975).
Proceedings, Symposium on 'Implementation ofComputer Procedures and Stress-Strl,lin 27. J. L. Sanders, 'Plastic stress-strain relations based on linear loading function',
Laws in Geotechnical Engineering' (Eds. C. S. Desai and S. K. Saxena), pp. 575-590, Proceedings, 2nd US National Congress on Applied Mechanics, ASME, 455-460
held in Chicago, Aug. 1981, Acorn Press, Durham, NC, 1981. (1955).
9. G. N. Pande and W. Xiong, 'An improved multi-laminate model of jointed roclf 28. J. F. W. Bishop and R. HiII, 'A theory of plastic distortion of polycrystalline aggregate
masses', Proceedings, International Symposium on Numerical Model on Geomechanics under combined stress', Philosophical Magazine, 42, 327, 414-427 (1951) (see also
(Ed. by R. Dungar, G. N. Pande, and G. A. Studer), pp. 218-226, held in Zurich, Sept. continuation on pp. 1298-1307).
1982, Baikema, Rotterdam, 1982. 29. A. J. M. Spencer, 'A theory of the kinematics of ideal soils under plane strain
10. W. Wittke, 'New design concept for underground openings in rocks', Chapter 13 in conditions', J. Mech. Phys. Solids, 12, 337-351.
Finite Elements in Geomechanics (Ed. G. Gudehus), John Wiley, 1977 (see also
Erzmetall, 26, 2, 66-74 (1973».
11. Z. P. Baiant and B. H. Oh, 'Model of Weak Planes for progressive Fracture of
Concrete and Rock', Report No. 83-2/428m, Center for Concrete and Geomaterials,
Northwestern University, Evanston, Illinois, Feb. 1983.
12. Z. P. Baiant and B. H. Oh, 'Microplane model for fracture analysis of concrete
structures', Proc. Symposium on the Interaction of Non-nuclear Munitions with
Structures', held at US Air Force Academy, Colorado Springs, May 1983 (publ. by
McGregor & Werner, Washington DC).
13. J. W. Dougill, 'On stable progressively fracturing solids', ZAMP (Zeitschrlft for
Angewandte Mathematic und Physik), Vol. 27, Fasc. 4, 423-437 (1976).
14. J. W. DougiIl, 'Some remarks on path independence in the small in plasticity',
Quarterly of Appl. Math., 32, 233-243 (1975).