A PLASTIC-DAMAGE MODEL FOR CONCRETE

J. LUBLINER
Department of Civil Engineering. University of California. Berkeley. California. U.S.A.

and

1. OLIVER. S. OLLER and E. O%ATE

Universitat PoliGcnica de Catalunya. E.T.S. Inpenieros de Caminos. Canales y Puertos.
Barcelona. Spain

AWract-fn this paper a constitutive model based on an internal variable-fo~uiation of plasticity

theory for the non-linear analysis of concrete is prcscnted. fhc mod4 uses a new yield criterion
which matches expcrimcntal data quite well and it accounts for both elastic and plastic stiffness
degradations etfects. On.wt and amount of crackillg can bc studitd by a simple postprocessing of
the ~njte-element plasticity solution. The xcuracy of the model is checked with some examplss of
application.

I. INTROIXJ(I’ION

The &t&al theory of pl:tstioity. like Tony twthcnluticnl roprcscntation of the mcchnnicitl
bchnvior of solids. may hc vicwcd in two ways : iIS il tr;illSliltiOIl 01’ phySiCill reality. ilfld iIS
il m&l that ilppronirn;ltcs the hchnvior under certain circumstances. The lirst view is often
held with royard to ductile crystalline solids, cspcciillly metals. :llthough the attempts to
rclatc mathematical plasticity theory to dislocations hvo not been markedly successful.
With reg;trd to concrctc and rock, howcvcr, it hits gcnc~\lly been ~lckn~~w~cd~c~~
that such
pr~~mincnt fc:tturcs of pl:isticity theory ;is ii well dcfincd yield criterion and strictly ClilStiC
uni~~~Iin~ iwe ~pproxifniltions ilt bCst. ~~v~rt~~~~~ss, many problems involving thcsc
mlttcrinls hitvc bren quite succcssfufly trc;ltcd by mcuns of plasticity theory, and these
results are not inv~llid~ltc~i by frtct that other problems have not hccn so successfully treated.
nor by the fact that other models may h;lvc been cquuily cfl&tivc.
The broitdcst arelt of success of pklsticity theory with conorctc is the treatment of
reinforced concrete (see Chcn, 1982 for 3 survey of the results) und other situations in which
the material acts primarily in compression. In problems in which tension, with the attendant
crack development, plays ;L significant rolc- such i\s shear Failure in reinforced-concrete
structures-the usual proctdurc nowadays is to apply plasticity theory in the compression
zone. tend trtxt the zones in which iit Icnst one principal stress is tensile by one of several
versions of fracture mechanics. such as: linc;lr elastic fracture mechanics. Bcinnt and
Cedolin (1980) : smeared-crack mod&. Rushid (1968). Suidan and Schnobrich (1973). de
Borst and Nrtuta (1984); the fictitious-crack modct, fiiltcrborg ct uf. (1976); and crack-
band theory, B6tmt rend Oh (1983).
In spite of the success of this :tpproitch in solving numcrotjs problems, it presents some
inconvenient fcitturcs that limit its usefulness. such ;ts the need for defining uncoupled
behavior along each principal stress (or strain) direction. the USC of a quite arbitrary shear
retention factor to ensure some shcrtr rcsistancc illong the crack. the lack of equilibrium at
the cracking point when mocc than one crack is formed (Ofiatc ct ui.. 1986), the difficulties
of defining stress p:lths following the opening and closing of cracks under cyclic loading
conditions. and the diflicultics of dealing with the combined cffcct of crrlcking and plilsticity
at the damaged points (dc Borst. 1987).
Some of these limitations could bo avoided if a single constitutivc model could be used
that governs the non-linear behnvior of concrete. including failure. in both tension and
compression, with appropriate allowance mrtde for the different values of the parameters
describing the two modes. It is the purpose of this paper to formulate such a model in the

30
 A plastic-damage model for concrete 301

variable K. similar to the former in that it never decreases, and increases if and only if plastic
deformation takes place. However. the plastic-damage variable cannot increase beyond a
limitingvalue, and the attainment of this value at a point of the solid represents rota1
damage, which can be interpreted as the formation of a macroscopic crack. The variable K
can be non-dimensionalized so that its maximum value is unity.
As was also said. the model is intended to apply to frictional materials, in which total
damage is assumed to correspond to the ranishing afthe cohesion. The yield criteria most
often usedfor such materials are the Mohr-Coulomb and Drucker-Prager criteria, both
of which may be written in the form

F(a) = c, (2)

where c is the cohesion. and F(a) is a function that is homo_eeneous in the first degree in
the stress components. However. both these yield criteria have notoriously poor correlation
with experimental data for concrete and geomaterials. Numerous improvements to these
yield criteria have been proposed in recent years : Chen and Chen (1975). Ottosen (1977).
Chen (19SZ). Podgbrski (1985). Fnrdisand Chen (1986). However. they do not take the form
ofeqn (2) with F(a) a homogeneous function of the stress components. and consequently do
not permit an unambiguous definition of the cohesion. A first attempt of the authors to
obtain suitable yield criteria for concrete in the form of eqn (2) was to modify the Mohr-
Coulomb criterion to tit enpcrimsntal data. Numerical results obtained in the finite-clement
analysis of various concrctc structures using tbc framework of standard plasticity theory
wcrc cncnuraging and motivated the present rcscarch. A new yield criterion of the form
(2). whose iit with ~~p~ri~n~nt~ll data is about as good as that of any other proposed criterion
(oxccpt possibly in the dom;~in of high hydrostatic prcssurcs) will hc prcscntcti in Section
1.4.
Tho cohesion < will lx scaled so that its initial value is,fi.,,. the initial yield strength in
uniaxial compression. which may bc iclcntificd with the “discontinuity” stress, ix. the stress
at which the volume strain attains its cxtrcmum. Conscqucntly (* =jC,, when K = 0 and
(+= 0 when K = I. Howcvcr, unlike the usual plasticity models with isotropic hardcniny, c
is not nocsssarily taken simply as :I function of K. Rather, the value of (* at ;I given K may
CIC~CIIJ on the process-that is. the cohesion c is itself assumed to be an internal varii\blc,
governed by a rate equation in which i is proportional to I& the proportionality factor
being a function of the state variables.
If degradation of the elastic stiffness is not taken into account, and if the elastic stilrncss
tensor is dcnotcd by D, then the governing equations of the model consist of: (a) the yield
cri&rion ( 3) ; (b) the ehtic-pltrsric struirr cl~lcont~i~.~iriort

e= 0 -‘a+e”; (3)

ip = ig. (4)

where x the plastic loading factor and g = c?G/c!ais the plastic flow vector normal to the
plastic potential surface G = const; (d) the rule cqtrcrrion/br K, which will be assumed to
be of the form

k = h’(a, c. zc)t?’; (5)

and (e) the rate eq~~~i~~~

for c, of the form

r: = k(a. c. K)l;.. 16)

30’ J. LIAILIXR et al.
(al ( bl

Fig. I. Unianial curves (a-f) : (a) tension: (b) compresson

2.2. DeJhirion of K
Uniasial stress states. Let us suppose that we have available experimentally derived
stress-strain diagrams in uniaxial tension and compression, and that these may be converted
into c--$ curves. as in Fig. I. Let us further assume that the areas under these curves arc
finite and equal to .c/,and y,, cespectivcly. For the tension test. let us define

(7)

With K as the indcpcndcnt variable. curve (a) of Fig. I may further bc convert4 into ;I
function CT=J;(K). such that/;(O) =I;,, andj;( I) = 0. Similarly. for the compression test WC
may dclinc

and convert curve (b) of Fig. I into CJ=J~(K) such thatjL(0) =/;” and/,( I) = 0.
An analytically convcnicnt function J’(K) that may scrvr as cithcr j:(x) or j;(k) and
which is consistent with the fact that cxpcrimcntally observed stress-strain curves tend to
attain the zero-stress level asymptotically (rather than at a finite “ultimate strain”) may be
dcrivcd from the CT--~!relation given by

u =fo[( I +u) exp ( -htzP) -uexp (- %P)],

whrrc (I and h are dimensionless constants that may bc obtained if y = (TdP’ and

(da/dr:P)l,,P_,, are given :

Note that N > I implies initial hardening, while a < I implies softening immediately after
yielding.
By integration we obtain
 A plastic-damage model for concrete 303

Fig. t. Unia.xial curves (o-w) : (a) tension : (b) compression.

h’ =
I
.-

g
8
s
”
od.F= I- &[?(I+a)exp(-hsP)-nexp(-%&P)].

and thercforc

where C/)(K) = 1 fu(:!+N)K. If N > I, then f(x) attains a maximum value of

/in = /;,( I + (0 l/4,1, or

ctfgt ~~1~1~~.It has been made ~~buj~d~lntly clear OVC~ the past decade
Th* .si~F~~i~*~i~J~~
that the strain-softening branch of thr stress-strain curves of concrete and rock cannot
represent a local physical property of the material. The arguments have been advanced
both on physical grounds and on the basis of the mesh-sensitivity of numerical solutions
obtained by means of the finite-element method. The mesh-sensitivity can be largely elim-
inated if one defines g, = G,/1 and .gc = GJI, where 1 is a characteristic length related to the
mesh size. and G, and G, are quantities with the dimensions of enorgy/aroa that iire umcnzerf
to be material properties.
In problems involving tensile cracking. G, may be identified with the specific fracture
energy Cc, defined as the energy required to form a unit area of crack. It has generally
been assumed that Gr is I true material property, and methods have been developed for
determining it (Rots CI (II., 19YS). For the characteristic length I, various approaches have
been proposed : hLilnt and Oh (1983), Crisfeld (1986). Ccrvcra el ul. (1987), Oliver (1988).
Not so much attention has been paid to the corresponding compressive problem.
Compressive failure may occur through several mechanisms-crushing, shearing. and trans-
verse cracking-and consequently G,. if it is indeed a material property, cannot readily be
idcnti~ed with any particular physical energy. Moreover. it must be kept in mind that it is
only the dcsccnding portion of the stress-strain curve that is mesh-sensitive. Consequently,
a consistent definition ofgc must take the form gcQ+gE,, where g& is the area under the 6-
6” curve up to the peak stress, and gel is the remainder. Now, gco is mesh-independent and
is therefore a material property. For gv, we postulate. for convenience, that gE, =I G,,/f,.
where 1 is the forementioned mesh-dependent characteristic length for tensile cracking, and
x3-t J. LUI)LS~~PI al.

G,, is an assumed material property chosen in such a way that the numericai analysis of a
standard compression test gives results that coincide with experimental data,

Muffiu.riuf stress spates. In attempting to extend the preceding definitions to multiaxial

stress states. we consider first pure (but not necessarily uniaxial) tension nnd compression
states; that is. if the principal stresses pi!. 0:. o3 are ordered such that crl k 6: 2 a,, then
either g3 k 0 (pure tension) or ETA< 0 (pure compression).
More specificsliy. consider bia?riaI compression, i.e. B, = 0. It is tempting to use the
total energy dissipated, f (o, dsep + gi dc’; ). suitably normalized. as a dama_ee variable.
However. experiments by Kupfer et ul. (1969) and Tasuji et ai. (1975) indicate that even
~(a,dz~ alone is greater in the biaxial than in the uniaxial case. if& is identi~ed with the
offset strain (total strain less extr~pol~~ted elastic strain, allowing for stitfness de~r~d~tion~.
Moreover, on the CF~--E~
curves, the pcrlk stress occurs at a value of the t-5 that is npproxi-
mately independent of the stress-component ratio ot/a2. Consequently the ares under the
a& curve up to the peak is approximately proportional to the ratio jb.Jrnl.. to the uniaxial
vniuejl., (conventionally denotedJ:f. In order that the value of K correspondins to the peak
stress may be the same in the biaxial as in the uniaxial compression case. the definition (8)
cannot be used in the biaxial c~sc either. An altcrnntivc definition of K, which reduces (8)
in the uni:rGnl WSC, is cmhodied in the rate equation

Anaiogc~usty, in pure multiaxial tcnsictn wc can ;issunfc that

I~ittalty, we cctnsirtcr 3 state ofstressth;tl is neither pure tension nor pure compression,
i.e. (rl > 0 :tnrl 6 1 < f) (I’m cuamplc, simple shear), We ncccfan equation for K whose right-
hand side tends to that ofccln (9) or (IO) as ~;r, -+ 0 or TT, -+ 0. rcspcctivcly. Such an cquntion
mny bc JccIuccd from the more general equation. presumably valid at all stress states,

wht~c r(a) is ;I weight factor dc~cnd~n~ cor~tinuously on G such that 0 6 r(a) < 1, with
r(a) = I if 6, 23 0 for all i, i = 1, 7. 3, and f(b) = 0 if (i, < 0 for all i. A particular form of
r(a) is

where (x) = !(].~I +.Y). The special cuscs of pure tension and compression follow obviously
from cqn (13). For biaxial tension-compression (a, > 0, oL = 0. ~7~< 0), cqn (I?) yields
the following rate equation for K :

In particular, in n simple shearing test (al = 1: = --c~, rr2 = 0). the rate equiltion for ti is
 A phstic-d~m~gc model for concrete 305

It follows from the discussion in Section 2.1 that the evolution of the cohesion c must
be such that c-c 0 as h’+ 1 in any process. In particular. in view of the scaling of c, the
rate equation for c, eqn (6). must have the solution c =fc(rc) in a compression test and
c = (flQ,‘ft,,)J;(h’)in a tension test.
If. as some investigators. for example Chen (1982). hold, the compressive and tensile
stress-strain curves are similar. withf;(k‘)/J;o =f;(k-)&,. then the rate equation (6) may be
replaced by the functional relation c =Ji(h-). Otherwise, a form must be assumed for
k(a, r~ c) in eqn (6) which leads to the solutions discussed above. A possible form is

[
44
k(t7. K. c) = c Jrj--p) + !+V4]. (13)

where r(a) is a weight factor similar to the one discussed previously, and may in fact be
likewise assumed to be given by eqn (I’).
To show that c -, 0 as K + f in any process. we note, first, that r will, in general. vary
in the course of the process. With the hofp of eqns (12) and (I 3). (6) may be written as

anti, through intcgratioti by parts, rcwrittcn as

tl In L*= d(r InI; + (I -r) d InjJ + In (~~~~)dr.

f~inally, it can be intcgratcd to yield

Since there is no reason to rxpeet the integral f:I: In (j$J dr to become infinite in any
physicalfy realistic process, and sincef,( f ) = ji( I) = 0, it follows that c = 0 when x = 1.

In biaxial tests on concrete and gromaterials it is usually found (Kupfer et (II., 1969)
that the various critical surfaces in stress space (proportional limit, “discontinuity”,
“failure”) are similar. The same result is nob found in triaxiai compression tests, at least
at sufficiently large hydrostatic pressures; under these conditions it is found that the
hardening goes on indefinitely. In order words, while the yield surface (however defined) is
closed. the failure surface is open in the direction of hydrostatic compression. The so-celled
crrp model (DiMaggio and Sandlcr, 1971) has been used to describe this discontinuity.
The present model is directed toward failure analysis, and therefore no attempt will be
made to formulate the yield criterion in the region in which failure does not occur (should
this &come necessary, the cap model may be resorted to). Equation (2). consequently, is
assumed valid only in that part of stress space in which radial loading leads to failure. With
c = f‘,,. this equation describes the corresponding part of the initial yield surface, while the
failure surface is attained when t reaches its maximum along a given loading path. It is an
essential feature of this model that the same function F(a). homogeneous in the first degree
in the stress components. describes both.
As was discussed in Section 2.f, the many analytical forms that have been proposed
for the failure surface of concrete are not of this nature, except for the Mohr-Coulomb and
Drucker-Prager criteria. As a rule. they are quadratic in the octahedral shear stress (or,
306 J LUBLIYER rt UI
--
equivalently in , J:. where J2 is the second invariant of the stress deviator) and linear in
the mean normal stress (or in I,. the first invariant of stress); the third invariant enters
through the polar angle 0 in the deviatoric (z) plane (Chen and Chen. 1975 ; Ottosen. 1977 ;
Chen. 1982; Podgorski. 1985: Fardis and Chen. 19%). With these forms. the meridians in
the G,c~~G,space are curved. so that the failure surface tends to a circular cylinder as I, -+
- x. If. however, the high-pressure region is excluded. then virtually all available failure
data can be fitted quite well into eqn (2). in which F(a) has the form

where z, /I and ;’ are dimensionless constants. This form uill be adopted in the present work
for the yield surface. Note that when ci,,,,, = 0. i.e. in biaxial compression. this is just the
Drucker-Prayer criterion. the only parameter then being r. which can be obtained by
comparing the initial equibinxial and uniasial compressive yield stresses&, and/;,,:

yicldinp

(./L /;,,I -- 1
x=Z( 1;,,./L,,,- 1 (15)

Expcrimcntal values of/~,,& lit bctwccn I. IO :trirl I. IO, yicltling z bctwccn 0.0s and 0. I?.
Once r is known. /I can bc dctcrrnir~ctl from /i,,//,,. whcrcj;,, is the initial unia.ltial tcrisilc
yield stress :

Lo I +x+/l
./;,, = I- z ’

or

- (I
P = ( I - ~)(/L,,./;I,) + z).

For example, if r = 0. I2 and_/;,, = I().(/;,,, then /I = 7.6X.

The paramrtcr 7 appears only in triaxial compression. that is. in stress states with
a,,,~1< 0. Let TM and CM dcsignatc. rcspcctivcly, the “tcnsilc meridian” (a, > (T: = a,)
and the “compressive meridian” (a, = ci: > CT,)on the yield surface. On the former,

a,,,,,= :(I, + 3J3-U

and on the latter

With a,,,, < 0. the equations of the rcspcctivc rncridians arc thcrcforc

(7;.+3)/3J:+(;l+3r)l, = (1 --XV: (TM)

and
 A plastiwhnag~ modrl for concrete 307

(~+3)J5&+3x)1, = (1 -r)_&: (CM)

here fc is the critical stress in uniaxial compression, whether yield stress (for the yield
surfa~) or ultimate stress (for the failure surface). Let us define

the present model then yields

that is. a constant. The meridians thus described are therefore straight. In spite of the claim
of Ottosen (1977) that p “increases from 0.5. . . but remains less than unity”, most available
experimental failure data are fitted just as well with straight as with curved meridians-
that is. with constant values of p. Typical values range from about 0.64 (Schickert and
W~nk~er, 1977) and 0.66 {Richart EI ~rl., 1982). to about 023 (Mills and Zimmerman, 1970).
From eqn (f6) we obtain

.{”_I 3(1 -P)

..___..--
2/)-f *

A ViIlUCof p = { Icads to y = 3.
The form taken by tho proposed yitld surface on ditrcrcnt planes of the stress space is
shown in Fig. 3.

2.5. f+-h k-lJiC* fJ?lli flJtl~/Clr’r? f Sf#if CW

It is known that granular materials such as concrete can exhibit a significant

woii

vuiutnc change when subjected to severe inelastic states. This change in volume, caused by
plastic distorti~)n, can bc rcprc~~~l~cedwell by using PII adequate plastic potentiaf function
G in the dc~jnjti~~nof the tlow rule as given by cqn (4). For the examples analyzed with the
present modct we have chosen for C the classicat Mohr-Coulomb yield Function with the
an& ofd~~~~t~lncy# substituting &>rthr angle of internal friction 4 (Oiiate et at., 1988):

f, sin [Isin *
C(U. II/) = j sin l/J+ JJ? co!i U- -I__-- , (17)
(
J3 1

In the absence of stitrness degradation, the tangent elastic-plastic stiffness operator

has the usual form ofclassicat ~~~~stjc~ty
theory. given in the six-dimensional vector notation
for symmetric second-rank tensors by

whcrc f = dF/&. g I= c’G,L?aand H is the plastic modufus, given in the case of the prcscnt
model by X-h’g. Regardloss of whcthcr an associated flow rule (g = f) is used or not, eqn
(18) rcquircs the values of the normal vector f to the yield surface. However, the yield
surface prcscntcd in Section 2,4 is not smooth, and a dt&sion must be made as to the value
off at the singular points.
The singular points of the yield surfaccarc the following : (a) those where the maximum
normal stress changes direction. comprising the compression meridians; (b) those where
the maximum normal stress changes sign, belonging to the intersection of the yield surface
with the three quarter-planes that bound the triaxid compression octant. namely fi) bi = 0,
a2<0,0,<0,(ii)a, <&a, = 0,~~ < 0. (iii) 0: c 0, (r2 < 0. f13 = 0.
308 J. LCBLINER et al.

ab b”

f
 A plasticdamage model for concrac 309

The compression meridian, say gt = a2 > CT~.inctudes cases of axial symmetry. in

which, by virtue of said symmetry, one must havef, =f2. It is not unreasonable to extend
this rule to ali cases corresponding to compression meridians-that is. to place f exactfy
midway between the bounding normals.
The other set of singular points is of great importance because it includes biaxial (and,
as a special case. uniaxial) compression. Suppose that e, > u2 >, rr3. Differentiating the
right-hand side of (I 4) with respect to G I yields

fi =~(~+~+,).
where 6 = /I if O# > 0.3 = y ifdt < 0. and 6 is ~~dete~i~ate in the range between fr and 7
ifa, = 0. In this case there is no symmetry argument to dictate a choice of 8. and recourse
must be made to empirical information.
Andanaes et al. (1977) report that eoncrete under biaxial stress apparently does nat
exhibit normality of the plastic strain-rate vector to the yield surface, although the results
are not altogether clear because the authors do not spetI out how they define elastic strain-
that is, whether it is defined by the initial modulus or fas it should be) by the current
(degradodf secant modufus. Jf it is the former. then a correction would bring the results
closer to normality. Even without this correction, however, normality is fairly closely
approacticd ilt uniaxinl compression in lllc limit from the biasid conrpr~~.ysiimcpaclrant.
This indicates that. if an associated flow rule were used with the present yield surface. then
at the singular points ~~?rr~sp~?n~iin~to fr,;,, = 0, S would have the value y. This resuft will
he xlopt4 as 3 working rule for the dctinition of f at the singuhtr points bcfonging to
category (h).
Uniaxial compression is rcprcscntcd by :I point that is the intcrscction of the just-
discusscd singularity locus with a cornprcssion mcriclian. If ~7, = o2 = 0 and 0) < 0, the
rule WC have itdoptcd yields

j; =ff = (;+.+,).
;.A-;
L?=$-&l++= -1.

With an associated Row rule, this result gives the following value of the ratio of the
transverse to the axial rofrrl strain rates :

_
‘I
E
..- = _
ET
_I

ci 23

where E is the elastic Young’s modutus, v is the elastic Poisson’s ratio and E, is the current
tangent modulus.

3. I . Gtwtw~l
rmt.si&ra rims
In the presence of stiffness degradation, i.e. when the elastic stiffness D varies in the
course of
deformation. the preceding results must be modified, in particular with regard to
the formulation of an associated flow r&e,
it wifi be assumed that stiffness degradation can be described through the deptxdence
of the stiffness D on two sets of scalar-valued internal variables. to be called the ehsric and
phric degradation variables, and to be denoted d,, n2. . and 3,. &. . . I , respectively.
The former are similar to the ciamqe cariddes such as those considered by Simo and Ju
310 J. LCBLISER Ed ai

(1957) and others. in that their variation is associated with the total deformation. but
without the necessity of a Lfamagr rriteriun. Their evolution wilf be assumed to be governed
by rate equations of the form

d, = #,(k]ri>, (19)

where k, is a vector in stress space denoting the direction of what may be called cfegrcdution
louciing associated with the variable ci,. and 4, is a positive scalar factor; the 4, and k, are
functions of u and the 4. and the #Jma_~incorporate a damage threshold in the sense that
they may vanish inside some surface in the space of the state variables.
The plastic degradation variables 4, are those associated with plastic deformation. and
constitute a special case of the damage variables introduced by Chaboche (l977), in that
they are governed by the rate equations

15,= i/f, (20)

where i is the plastic loading factor in the tlow rule (4). With the further choice ~1,= i,rg,
these rate equations become

whore the I, arc likcwisc vectors in stress space.

A distinction bctwccn the degradation vitriables used hcrc and the damage varinblcs
that i:ppc;tr in continuum tl:tmitgc mechanics is that we do not invoke the concept of
e,,@litx~ .~rcsx introduced by Kachanov ( t958) itnd devclopod by Rnbotnov ( I WI) ;ttId
others, such as Martin and Lcckie (1972) and fhtyhurst antf Leckic (1973). According to
this concept, the equations governing the behavior of the thtm;t~cti material arc obtnincd
from those of the l~rl~i~~~~l~~~c~l
material by rcphrcin_c the true stress d by the ctfcctivc stress
a,‘( I -cl). whcrc tl is the rfamagc vari:thlc ; thcrcfore the stilfncss I) is rcpiaccd by ( I -t/)1).
tlowcvcr. while the notion of cffcctivc stress may be ~lppr(3pri~~te in metal plasticity and
creep in view of its microphysical basis, it is kss so in the case ofconcrctc and gcomatcriafs.
It will conscqucntly bc assumed. at least to begin with. that D dcpcnds on the do~radation
variables in a general way. so that

u = D(tl1,. . : s,,. . .)(e-&P)

It follows by difhxentiation with respect to time that

ti = ril, -‘a-i-D&-S).

Now, from the chain rule, and cqns (19) anti (7-l),

We define the operator C, by

(27,

note that C, is ;I nonlinerrr operator in that it depends on the direction of d.

We also define the linear operator
 A plastic-damage model for concrete 311

CP= D-~2D,‘dd,D-‘@I;. (231

From the definitions (27) and (23) it follows that

d = C,&cgp. (24)

Let the yield criterion be given. as before, by eqn (2). Note that we are omitting any
dependence of the yield criterion on the degradation variables. in contrast to the continuum
damage mechanics model based on the aforementioned notion of effective stress. which
necessarily brings the damage variable into the yield criterion. As we said above. this model
has a physical basis for metals, but in the case of concrete and geomaterials the theory of
plasticity is at best an approximate mathematical model of behavior, and its usefulness
would be impaired by complications.
The consistency condition

where f = ?F/c@. combined with eqns (5). (6). (9). and (24). yields

where

If = kh’g (25)

is the plastic modulus. whencvcr A > 0. Stability under strain control implies that

Ii + f ‘t,g > 0,

and this condition will be assumed to bc met. Conscqucntly

and thcrcfore the elastic-plastic tangent st#ness operator is defined by

-
&-Cd
e- c,-c”$e_ (27)
p Hff’C,g

in place of cqn (I 8). The tangent stiffness, as a piecewise linear operator, is symmetric if C,
is symmetric crrrtlif C,g is proportional to Cff. The symmetry of C, will be examined in the
next section, while in the following one it will be shown that the aforementioned pro-
portionality is an appropriate form of the associated flow rule in the prcscncc of stiffness
degrad~ttion.

3 2. Eimtic ckymiii~ tion

The simplest hypothesis of elastic degradation is based on a single variable d (Kach-
anov, 1958). such that

D = (I -d)D,, (28)

where D,, is the initial stiffness. For an isotropic solid, this hypothesis implies a constant
Poisson’s ratio. which several investigators, e.g. Kupfer et al. (1969). have reported for
concrete up to a stress level that approximately coincides with the onset of major cracking
213 1. LUfw.*ZR es al.

(typically some 75430% of the ultimate stress) and that therefore may be reasonably taken
as representing the yield stress. Equation (28) leads to the operator C, given by

Thus Ck is symmetric if and only if k is proportional (or equal. with no loss of generality,
since any proportionality factor can be absorbed in 4) to (I. In other words, elastic stiffness
denudation is associated with an increase in the total defo~ation work. Note that, in the
elastic rang, trTti = ( I -4 Et’,. where 2 IV, = eTDs, is the square of the unrinmaged f~erg~
norm of the strain (Simo and Ju, 1967).
However, the majority of investigators of the multiaxial behavior of concrete (see
Cedolin tr rrl., 1977 for a survey; also Andanaes et al.. 1977) have found degradation
behavior that is not described by eqn (78). Instead, they have found that the bulk modulus
depends primarily on the volume strain, and the shear modulus on the octahedral shear
strain, (n the elastic range. this dependence is equivalent to one in which the bulk and shear
moduli are determined respectively by the volume work and dist~rtion~~l work, and can be
described by assuming. in ptace of eqn (28) that

K = f 1 -tf,)&,, G = I1 -d,)G,,, (SO)

wflcre I\’ (G) ;~ntl K,, (G,,) dcnotc rcspsctivoly the current and initial values of thu secant
hlk (StlCilr) lllOdlllllS.
7’hc 11 x 6 stillness nurtrix of an isotropic solid can bc written in the form

D = Kl 1I’+ GUdcv,

whcrc t = (I, I. l,~l,O.O], dcv = I - ill ’ (f being the 6 x 6 identity matrix), and

2 0 0 0 0 0
0. 7 0 0 0 0
0 0 2 0 0 0
u= 0 0 0 I 0 0
0 0 0 0 1 0
00000I

Thus

where

and therefore, since U dcv U -‘dev = dw,

 A plastic-damage model for concrete 313

where aa = jlTa is the mean normal stress, and s = dew is the stress deviator. Consequently
C, is symmetric in all processes if and only if k, = a,1 and k2 = s---that is, if the bulk and
shear stiffnesses are associated, respectively, with an increase in the volume and distortional
deformation work.
Let us consider. for example, an exponential dependence of the moduli on the cor-
responding strain, i.e. a dependence given by

where a, and a2 are constants, and E, = lTs and yWc= 2,,&- are the volume strain and
the octahedral shear strain. respectively. It can be shown that these functions can be
obtained, in monotonic radial loading, from the following forms of (6, and 42 :

3.3. T/w t~ssociati~i

f jlr, w rdc
It is well known in classical plasticity theory that what is generally known as the
associated ftow ruic-that is. the normality of the plastic strain-rate tensor to the yicfd
surfrtcc in stress space-is cquivafcnt to the axiom of mchtrum pfastic dksipatiu~ (MPD)
wbcn the clsstic still&s is constant, or, more gcncrally, when the frtr cncrgy (at a given
tcln~r~ltllrc) may bc dccomposcd as

whcrc the a, ilre the phtic internal variables i.e. variables whose rates vanish whenever
ip = 0. Such conseyucnces of the associated flow rufc as uniqueness of solutions to boun-
dary-v:llue problems and the theorems of limit analysis ilow largely from the MPD axiom,
which hits also been helpful in clarifying the form taken by the associated flow rule in large-
dcformntion plasticity (Lubliner, f984, 1986).
In its most general form, the MPD axiom may bc expressed as follows. Let the free
energy be written as J/(e’,a,, . . , , j?,, . . .). where e’ = e--ep is the elastic strain, the a, are
the plastic internal variables as above and the 11,are any other internal variables. The total
dissipation is

D = a’i-4.

Since d = c?#/&?. the dissipation may be decomposed as

D = D&i-D,,

whcrc

and
The latter is the plastic dissipation. a function of &-E~, I!. . . . . fi,. . . . ; Cp. i,, . . . . The

MPD axiom assumes that

Dp(~-~P.rI..... p ,.... :i”.i, 1... 12 Dp(&*--&p.5L

I...., p I*.., ;cp,ci,,...) (31)

if E* is a state of strain that can be attained from E by means of an elastic process, i.e. one
with no change in sp. the z, or the a,. In simple cases inequality (31) reduces to the well-
known form

With D, a differentiable function of&-SEP. inequality (31) may also be expressed in the local
form

For an el~~stic-pl~lsticsolid exhibiting linear elasticity with de~r~~d~itionas dcscribcd in

Section 3.1, the free cncrgy may bc written as

* = j(z -EP)'-I)():--tf) -t-l&(2,, . . .),

whcrc D dcpcnds on the d, and 6,. as hofore. Note that fftc 6, arc incfudcd among the z,.
The plastic dissipation is, ~~cc~~r~iin~iy,

and the local form of the MPD axiom, incylI~~lity (32), becomes

L
i’ 11-~f?0/3s,(e--e’y,’
I 1
cp2 0,

or, in view ofeqn (23).

i’l’C,iP2 0.

The flow rule (4) implies that

i”c& 3 0 (33)

whenever ; > 0, that is, whenever

i’Pf c >
r, 0 - f34)

in view ofeqn (26). The simultaneous satisfaction of inequalities (33) and (34) requires that
C,g be proportional to or, with no loss of generality, equal to CIf :

c,g = C,f. (35)

In a solid with elastic and plastic stiffness degradation as defined hcrc. than, the flow rule
 A plastic-damage model for coocrete 315

(4) may be regarded as associated if g obeys eqn (35). In the absence of degradation,
c, = c, = D and therefore the classical form g = f is recovered.
If eqn (35) holds, the tangent elastic-plastic stiffness is defined by

( rTc,r >
ci = C$ = C$--CTf (36)
H+fTCTf’
c

It thus differs from the classical form only in that the constant elastic stiffness D is replaced
by C,.

3.4. Plastic degradation-an example

A simple model of plastic degradation that may be appropriate for concrete is based
on the assumption that plastic degradation occurs only in the softening range, and that the
stiffness is then proportional to the cohesion-that is, there is only one plastic degradation
variable 5 = 6,. with the current stiffness given by

D = (I -J)D&i,. . . .).

and with S governed by the rate equation

the initial value 5 being zero. so that, in the softoning range (c’ < 0).

The vector I = I, appearing in cqn (31) is accordingly given by

the quantities h and k being those of eqns (5) and (6). Thus

c, = D+ ki!&‘.
c

Equation (35) may now be solved for g:

1
( - k)h’D - ‘C:f
g = D-1 cyr- c+( -k)h”ri” (37)

The plastic modulus I-f as dcfincd by cqn (25) is given by

ckhTD - ‘C,‘f
H=
c+(-k)h*D-‘a’

4. NUMERICAL EXPERIMENTAlION

The constitutive model presented has been implemented in a standard finite element
program for non-linear analysis of structures and applied to evaluate the numerical response
of several specimens for which experimental results are known. Before entering in the
 3t6 f LUBLIHER
et rtl

Fig. 4. Direction of cracking.

discussion of the examples analyzed. some general considerations on the definition and
representation of cracking have to be mentioned.
Cracking is the most important external mrmifestation of damage in a concrete struc-
ture. In order to obtain a graphic representation of this kind of damage. some parameters
are CV:&KIM itt each intcgrrttion point (I prts~iori, once convergence of the non-linear
solution has ban rcsched. This can be interpreted 3s a postprocessing of the results in
which conditions for onset of cracking, crr~ck dircrtic~?tf. pfusric strcrins {as a mcwwc st the
opening of the crncks). t~rc*r,q_~rlissiptrrion and shear-rr~lcrrlionJil~tnr arc computed by means
of the following proccdurc :
(a) Crackinp initiates at ;I point when the plastic-dsm:lge vilriitbtc K is grattcr than
zero, and the maximum principal plastic strain is positive. The direction c3f cracking is
assumed to be orthogonaI to that of the maximum principrtf plastic strain at the &ma&
point. Other criteri:t for defining onset and dirccticms ofcracking as the hwiizttiotr c.cmfitimt
b:rscd on the ircoustic tensor (Willam and Sobh, 1987). or nrtr.rirrucnc cwcr;): rchy~.rc ( Ihk~nt,
1986). arc also possible.
fb) The vector of plastic strains along the directions of the crack, c”. can he otminrt!
in terms of the plastic strrtins e” cxpresscd in grobtli Cartelsian axes xs :

(39)

where U is the angle which the direction of the maximum principal strain forms with the
global s-axis (see Fig. 4). The vector ecr is used as a measure of the opening and sliding of
the crack.
(c) The energy dissipitted in the structure due to cracking in a load increment is
obtained as

where V is the volume of the structure.

(d) The shear retention factor (Rots el uf., 1985). at a crack is obtained as
/‘I= x,~.~./z:-,.,where r,.,. is the shear stress par&t to the direction of the crack and r:.,‘ is
the value obtained from a linear-elastic analysis as rr.v. = G<;,,,.., whcrc GO is the elastic
shear modulus.

The first example presented is the application of the model to the well known biaxial
compressor test of Kupfer et al. (I 9699).The exampte consists of the study of the behavior
 A plastic-damage model for concrete 317

Eo:395K?u.00 kg cm“
r,:O 24
t.0 :: -229 00 kg cm“

+ ,,,.I : -328 00 kg Cm”

ttp: 22.90 kg cm“
a:0 12- so ~1.16

y--30
G, -0 16kq cm-’
C. :16GOkq cm ’
Asswatsd tlow hid and potential twut~on. drowsed

of 9 concrrtc spcrcirncn of 20 x 20 x 5 cm (se0 Fig. 5) subjoctsrl to dilfcrcnt loading con-

ditions - pure compression, douhic symmetric compression and double non-symmetric
~ornpf~ssi~~n.
The geometry of the test. material properties data and the finite elumont mesh of four
standard 4-nodcd rectangular clcmcnts used for the analysis arc shown in Fig. 5. In all casts
an initial stress approach, which circumvents the problem of non-symmetry of the stitfness
matrix due to non-associated plasticity, has been used. Also, displacements have been
controlled using a standard spherical path technique (Crisficld, 1981).

4.1 . 1. Ptttt! ‘.~tttpt%%itttt : flzV/a, , = - t/o tttd d J J = 0. In Fig. 6(a) numerical results
obtained with our model for associated and non-associated plasticity have been plotted.
Excellent agreement of numerical results for the u 12-t:22 curve with experimental test, also
plotted in the Fig. 6, is obtained for both cases. Good agreement between experimental and
numericat results is also obtained for the cr22-~, , curve. However, associated plasticity seems
to match experimental results better after the stress peak in this case.
In Fig. 6(b), numerical results obtained by diKerent researchers for the s&me problem
are shown for comparison. Finally. in Fig. 7, the crack directions obtained at total failure
in the Gauss points are shown.

4. f .2. ~o~b~~ s~~~i~r~~~rjc

cu~fpr~ssiun: itrtl/ffI f = - I/- 1 fmi (T,~ = 0. numerical
results for the az2-ezt (or bl,-~,,) curve as shown in Fig. 8(a). Good agreement with
available expcrimcntal data (Kupfer et ul., 1969), is obtained for both associated and non-
associated plasticity. Howcvcr. some perturbations of numerical results towards the end of
the test can be observed. These are due to the well-known locking efic~ in 4-nodcd
rectangular elements working under incompressibiiity conditions, as is the cast in the
example analyzed when plastic strains develop. This spurious effect can be eliminated by
using reduced integration and other numerical techniques (Crook and Hinton, 1987).
Results obtained by different researchers for the same problem have been plotted in Fig.
8(b) for comparison.
318
 A plasticdamage model for concrete 319

I------ -----e--
7
* Orlglnat shape
I

II

Fig. 7. Biaxial compression test. Distribution of crauking for pure symmetric compression.

4. I .3. &~rh!c rrotr-sl’mntclri~ cotnflra.~i~tt : 42z-~ t t = - l/ - 0.52 r&a,? = 0. Numeri-

cal results obtained for the (T:-~ .:? and ~T~?-E,acurves ;trc shown in Fig. 9(a). Results agree
rcasonnhly well with those reported by Kupfer er al. (1969). with errors of 3% and IS%
in the peak stress and the corresponding strains, respectively, for the ~2_-r;22 curve for both
associated and non-associitted plasticity itn;tlysis. The USC of non-associated plasticity seems
to he more important to match experimental results for the ~z2-~:t, curve. itlthottgh the
peak response is not very precisely reproduced in this case. The dillicultics nt’itnalyzing this
problem itre evidenced in the reproduction of the numerical results obtained by various
researchers shown in Fig. 9(b).

4.2. f::uarrlplt~ 2 : urtidysi.~ ctj’cr rrolchvi hetmt rni.~ildJitr~rlrr~l trroiic

This exitmple is a reproduction of the experimental test performed by Arrea and

IttgrolYen (198 I). The geometry of the notched beam, mrttcriai data and loading conditions
used to induce a mixed fracture mode (modes I and II) are shown in Fig. IO. As it can be
seen in the figure. the steel beam, used to transmit the ioitds to the concrete beam, has also
been considered in titc analysis (assuming linear behaviors in order to take its rigidity into
account. The numcricttl analysis was performed using eight-noded two dintensionul linite
elements. and the mesh used is shown in Fig. IO(a). The crcrcli ~toulh slih~ cli.y&mr~wn~

(CMSD) at the notch tip (see Fig. IO(b)) was controlled using a spherical path technique
(Crisfieid, 198 I ).
Numerical results for the load -CMSD relation, showing the points ofonset ofcracking
(point A), instubiiity (point B) and ultimate state analyzed (point C) have been plotted in
Fig. I I. Good agrcemettt with the experimental results of Arrca and lngraffea (1981). also
plotted in Fig. I I, is ob;~ined.
In f:ig. 12 thecracking pattern itt the peak (point B of Fig. I I) and ultimrtte toad (point
C of Fig. I I itrc shown. it is interesting to note that cracking localizes in a narrow curved
band only after the peak load. for which all cracks are distributed almost vertically and
form an angle of approximately 60” to the horizontal axis (see Fig. 12(a)). Excellent
agreement between the localized cracking band obtained numerically and experimentally is
achieved, its can be seen in Fig. I Z(c).
The principal stress distributions at the onset of cracking (point A of Fig. 1I) and the
ultimate state (point C of Fig. I I) are shown in Fig. 13. It is worth noting the stress
relaxation in the zone where cracking localizes (see Fig. 13(b)). This localization can also
be clearly seen in Fig. 14. where the deformed shape of the beam (magnified 300 times) at
the end of the test is shown.
 l.
.
.
.
.
*

t . l

I
l

c
.

l
 l b~-ez,J ) Ovorhlnstol (19t37)

. (a,*-c*,J
Murray et al. (19791

Y
Fig. 9. Biaxial compression ITS:. Compared stress-strain results for double non-symmetric compression.
 J. LGBLMR et al.

10) ID)

203cm . 39 7cm 6 Icm 6 icm 397cm 20 3cm

- E~~erimentol test, Arrao and Xngroftsti (198’

o--9 Proposed modei

1” 100
*

6 I)0

60

20

0 I I
0’02 0’0. 0 06 0 06 0 10 0 ;. 0 to

CMSD x lo-’ em

Fig. t I. Nod~cci &am. Lud displxcmcnt cu~cs. comparison with cxpcrirncnrai rcsutts.
 A plastic-damage model for concrete

to)

Experimental bond

Fig. 12. Notched beam. Dixtributi~n and iocali?&ion of cr;t&ng: (;I) 11 the peak stress (B)--illi
cricks; (b) at the ultimate state (Q-crcks grcatcr than 3% of the maximum crack ; (c) at the
ultimate state (C)--comparison with expcrimcntal results of hrrcr and IngraiTca (lYIl)-crackr
grcatcr than 5% of the maximum cmck.

5. CONCLUSIONS

The authors belicvc that they have demonstrated that, when appropriately applied, the
cfassical theory of plasticity is a useful tool in the rate-independent inelastic analysis of a
material, such as concrete, which by no means can be regarded as elastic-plastic in the
usual sense. The si~ni~cance of the result is both theoretical and practical.
On the theoretical plane, we hope to have helped to show that plasticity theory, when
not interpreted too narrowly, is a very flexible model-one that can be used to describe a
wide variety of behavior, including dilatancy and other non-associative phenomena, stiffness
 J. LL.BLINER er al.

,. . . . .
,. . . . -
.,. . .. .. .. .
. . . . . . .‘~l#lnlnl*1*IHIO:~‘.~ . . .
. . .
,.u,r#r#r,*.*~.. . I,
_. . . : :t I IIIY,nInI”‘u::: : :_. : : : : :

. . .
. . .
. . .

. . . __._..
.-.-.. ..
_
.=::: : ::
. . ..
. . . . ..

Lacalizattm oi damage--/
Fig. 14. Nowhcd bum. Dcformcd shape al uhimatc state (d&uxmcnt magniliod 3W limes). and
localization of Ihc ctrrmzgc.
 A plastic-damage model for concrete 3’5

degradation. and others. Furthermore. it can be used to gain information about one
phenomenon-cracking-which is decidedly not one that is usually associated with
plasticity. In other words. plastic strain may be identified with any and all inelastic strain.
including cracking strain.
The practical significance of our results is that plasticity theory is a rather simple model
in comparison with models based on fracture mechanics or the more sophisticated versions
of continuum damage mechanics. In particular. a large volume of numerical codes for the
solution of problems in plasticity theory already exists. and is continually being enriched.
The excellent agreement with experiment obtained in the solution of a difficult problem
such as that of the notched beam shows that the potential of the present approach is great.
In future work. the authors hope to attack problems that include regions of triaxial
compression through the incorporation of a cap in the yield surface.

REFERENCES

Andanaes, E., Grestle. K. and Ko, H. (1977). Response of mortar and concrete to biaxial compression. J. Enyng
h/t& Dir. ASCE 103. 5 I S--CI6
Arrea. M. and Inpraffea. A. R. (1981). Mixed modecrack propagalion in mortar and concrete. Cornell University.
Department of Structural Engineering. Report X1-13. Ithaca. New York.
Batant. Z. (19X6). MLchanics of distributed cracking. dppl. .\I&. Rcr. 39. 675-705.
Ratnnt. Z. and Cedolin. L. (19X0). Fracture mechanics of reinforced concrete. J. .%g!cg hlcch. Dh,. rlSC& 106.
1287 1306

dc Borst. R. ;Ind N;IIII;I,P. (10X4). Smcarcd crack ;malysis of rcinforccd concrctc hrns and skths failing in shear.
I’rr~c. brr. <bn/: O,I (‘,~~~t~~~~I~‘r-~(ifk,d tsrtl I)c*si,qrr~~/‘~‘orrcmw Strtccrm*s Vol. I. pp. 26I 273. I’incridp
.~ltrir!,~.vrc
Press. SW;IIISC;I.
Buyuko~turk. 0. and Sharccf. S. ( IYX5). c’onstitutivc modeling of concrclc in linitc clcmcnt analysis. (‘on~{tctr.
s/rucf. 21. 5x1 610.
Ccdolin. L.. (‘rutrcn. R. ;III~ Dci Poli. S. (1077). Triuxial stress strain relationship for concrete. J. E$/ttq ,VPC~/I.
Div. ASC’E 103. EM3. 423 430.
Ccrvcra. M.. I linton, Ii. and I Iass;~n. 0. (19X7). Non-linsar analysis of rcinl’orccd plate and shell structures using
2tJ-node isoparamctric brick elcmcnts. <‘~nrprt/. S/nrcr. 25. X45 X70.
Chabochc. J. L. ( 1077). Sur I’utilis;ltion dcs variables tl’i’t;~l intrrnc pour Ia description du comportcmcnt
visa~plastiquc et de I;1 rupture par cntlolnrn;lScllle111. ONERA. Charillon (France). T. I’. No. 1Y77- I-IS.
Chcn. A. and Chcn. W. F. (1075). Constitutive relations for concrete. J. Grqrg .If&. /)ir. ,.l.SC’E 101, EM4%
465 4x1.
Chcn. W. I‘. (IYXZ). Phr.sricit~ in Reitr/imaf Concrc~c. McGraw-Hill. New York.
Crislisld. M. A. (198 I). A fast incremental iterative solution procedure that handles “snap through”. Conrpur.
Srrucr. 13.55-62.
Crislicld, M. A. (IYX6). Snap-through and snap-back response in concrete strusturcs and the dangers of undcr-
intcgrution. Inr. J. Nunr. Mdr. Entry/ 22. 75 I-756.
Crook, A. J. L. und Clinton, E. (lY87). Comparison of 10 quadrilateral finite elrments for plasticity problems.
Proc. IN. CMI/: Conqwcr~iunul Plusficiry (Barcelona) (Edited by 0. R. 1. Owen, E. f-linton and E. Ofiate) Part
I, pp. IXI 195. Pincridge Press. Swansea.
Di M&o. F. L. and Sandlcr. I. S. (1971). Material models for granular soils. J. &ng~g hlrch Dir. ASCE 97,
EM-t. 935 -950.
Dvorkin. E., Torrent. R. and Alvaredo. A. (19X7). A constitutive relation for concrctc. In Proc. /III. Con/:
C~mpururimd Plu.vricit.v (Barcelona) (Edited by 0. R. J. Owen, E. flinton and E. Driatc) Part 2. pp. 1415.-
1430. Pineridgc Press. Swansex.
Fardis. M. N. and Chcn, E. S. (lYX6). A cyclic multiaxial model for concrete. C’~>~rt~~r.M&r. I. 301 -305.
flan. D. J. and Chen. W. I’. (10X6). Strain space plasticity formulation for hardening -softcning material with
clastophstic coupling. //I/. J. .%tlicls Sfntcrurer 22, 935 950.
Hayhurst. D. R. and Lcckic. F. A. (1973). The clfect of creep constitutive and damage relationships upon the
rupture time of a solid circular torsion bar. 1. Mcch. Phys. SolilLf 21. 43 I .-t-$6.
Hillcrborg. A.. Modccr. M. and Pctersson. P. (1976). Analysis of crack formation and crack growth in concrctc
by means of fracture mcuhnnics and finite elcmcnts. Ccnrc*nr Concrctc Ref. 6. 773-782.
Kachanov. L. M. (IYSX). Time of the rupture process under crcvp conditions. /-_r. Ako~l. Nuuk S.S.S.R., Old.
T?kh. Nuuk 8. S-3 I.
Klisinski. M. and Mr& Z. (19X7). Description of inelastic deformation and degradation of concrctc. lnstitutc of
Fundamental Tcchnologicnl Rcscarch. lntcrnal report. Warsaawa. Pol;md.
Kupfcr. Ii.. flilsdorf. tl. K. and Ru.sch. fl. (1969). Behavior of concrctc under biaxial stresses. J. AC/ 66, 65&
666.
Lubliner.J. (19X4). A maximum-dissipation principle in generalized plasticity. AC/U .\frchunicu 52, 225-‘37.
Lubliner.J. (19Xh). Normality rules in large-deformation plasticity. ,v&. $/arls 5. 29-34.
Martin.J. B. and Lcckie. F. A. (1972). On the creep rupture of structures. 1. Ald~ Phy.r. Sdicb to. 2~3-235.