Mechanical Modeling of Material

Damage
A systematic theory to describe the anisotropic damage states of materials and a
S. Murakami consistent definition of effective stress tensors are developed -within the framework
Professor, of continuum damage mechanics. By introducing a fictitious undamaged configura-
Department of Mechanical Engineering, tion, mechanically equivalent to the real damaged configuration, the classical creep
Nagoya University, damage theory is extended to the general three-dimensional states of material
Furo-cho, Chikusa-ku, damage; it is shown that the damage state can be described in terms of asymmetric
Nagoya 464-01, Japan
second rank tensor. The physical implications, mathematical restrictions, and the
limitations of this damage tensor, as well as the effects offinite deformation on the
damage state, are discussed in some detail. The notion of the fictitious undamaged
configuration is then applied also to the definition of effective stresses. Finally, the
extension of the effective stresses incorporating the effects of crack closure is
discussed. The resulting effective stress tensor is employed to analyze the stress-path
dependence of the elastic behavior of a cracked elastic-brittle material.

1 Introduction
Proper understanding and the mechanical description of the M., 1971; Onat and Leckie, 1988). The damage variables to
damage process of materials brought about by the internal model such material damage should have not only clear
defects are of vital importance in discussing the mechanical ef- physical implication, but also have appropriate mathematical
fects of the material deterioration on the macroscopic and mechanical properties. As regards this point, previous
behavior of materials, as well as in elucidating the process modeling of material damage needs further refinement.
leading from these defects to the final fracture. A systematic The present paper is concerned with a systematic modeling
approach to these problems of distributed defects can be pro- of material damage and the extension of the aforementioned
vided by continuum damage mechanics (Chaboche, 1981; Kachanov-Rabotnov notion to the general three-dimensional
Hult, 1979; Kachanov, 1986; Krajcinovic, 1984; Lemaitre and anisotropic damage. By postulating that the material damage
Chaboche, 1978, 1985; Murakami, 1983). The fundamental can be characterized mainly by the decrease of load-carrying
notion of this theory, attributable originally to Kachanov effective area caused by the development of microscopic
(1958) and modified somewhat by Rabotnov (1969), is to cracks and cavities, the state of material damage is described
represent the damage state of materials characterized by by introducing a fictitious deformation gradient from the cur-
distributed cavities in terms of appropriate mechanical rent damaged configuration B, of the material element to the
variables (internal state variables), and then to establish corresponding fictitious undamaged configuration Bj with the
mechanical equations to describe their evolution and the decreased net area due to damage; two kinds of damage ten-
mechanical behavior of damaged materials. sors D and D are defined to represent the damage state with
The nucleation and the growth of voids observed in the pro- respect to the current damaged configuration B, and to the
cess of material damage depend significantly on the direction elastically unloaded damaged configuration B,. Then, the
of the applied stress or strain and, hence, are essentially relations between the damage tensors D and D as well as their
anisotropic. The generalization of Kachanov-Rabotnov's no- mathematical restrictions are elucidated; they are shown to be
tion (Kachanov, 1958; Rabotnov, 1969) to multiaxial states of second rank symmetric tensors. The mechanical interpretation
damage, therefore, necessitates the proper description of the of these damage tensors and the definition of the related effec-
oriented features of material damage. Thus considerable ef- tive stress also are discussed. Finally, the proposed theory is
forts have been done to find rational modeling of anisotropic extended to incorporate the effects of the microcrack closure
damage phenomena (Betten, 1983; Chaboche, 1984; Cor- and is applied to analyze the elastic behavior of a cracked
debois and Sidoroff, 1982; Hayhurst and Stonlkers, 1976; elastic-brittle material under combined hydrostatic pressure
Leckie and Onat, 1981; Kachanov, M., 1980; Krajcinovic, and shear.
1983; Murakami and Ohno, 1981; Vakulenko and Kachanov,
2 Mechanical Description of Damage States
Contributed by the Applied Mechanics Division for publication in the JOUR-
NAL OF APPLIED MECHANICS. 2.1 Fictitious Undamaged Configuration and Fictitious
Discussion on this paper should be addressed to the Editorial Department, Deformation Gradient. Let us first take a tensile bar of Fig.
ASME, United Engineering Center, 345 East 47th Street, New York, N.Y.
10017, and will be accepted until two months after final publication of the paper
1. We denote its cross-sectional area in the undamaged initial
itself in the JOURNAL OF APPLIED MECHANICS. Manuscript received by ASME state and that in the current damaged state under stress a by
Applied Mechanics Division, May 27, 1987; final revision, October 20, 1987. A0 and A, respectively. According to the classical damage

280/Vol. 55, JUNE 1988 Transactions of the ASME

Copyright © 1988 by ASME
Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 t=(7A*

(a) (b) (c)

Fig. 1 Damage of a bar under uniaxial tension; (a) initial undamaged

state; (b) Current damaged state; (c) Fictitious undamaged state
<«> (•)

Fig. 2 Definition of three-dimensional states of damage; (a) Initial un-

damaged configuration; (b) Current damaged configuration; (c) Fic-
theory of Kachanov-Rabotnov (Kachanov, 1958; Rabotnov, titious undamaged configuration; (d) Elastically unloaded damaged
1969), the damage state of Fig. 1(b) can be represented by a configuration; (e) Elastically unloaded fictitious undamaged
configuration
damage variable (internal state variable) D (0<D<1)
specified as follows
D = 0 (initial undamaged state) (la) or micro voids, let us now extend the argument in the previous
D = 1 (final ruptured state) (lb) section. For this purpose, we will take an area element PQR of
an arbitrary orientation in the damaged material as shown in
The damage variable D of equation (1) is often interpreted Fig. 2(b), and call it the current (actual) damaged configura-
as the effective area (net area) reduction caused by the tion B,. Then, let us assume that the states of stress, strain,
distributed microscopic cracks and cavities due to material and damage are uniform in B,, and represent the line elements
damage (Hult, 1979; Kachanov, 1986; Lemaitre and PQ, PR and the area of PQR by the vectors dx, dy and vdA in
Chaboche, 1985); namely, it is assumed that the load-carrying the three-dimensional Euclidean vector space E. We further
net area of the bar of Fig. 1(b) is decreased from A to a certain take the initial undamaged configuration of the element by B0
smaller value A* because of the formation of distributed [Fig. 2(a)], and denote the corresponding vectors by dx0, dyQ
cavities, and the damage variable D is related to A * as follows and v0dA0, respectively. The deformation gradient from B0 to
1-D=A*/A (2) B, is represented by F.
Because of the three-dimensional distribution of
This reduction of load-carrying net area magnifies the effects microcavities, the load-carrying net area of PQR will be
of stress a induced by the external load t; in view of equation decreased. Then, as in the one-dimensional case of the
(2), the magnified stress is given as follows preceding section, we postulate that there exists a fictitious un-
a = t/A* = a/(l-D) (3) damaged configuration1 Bj [Fig. 2(c)] mechanically equivalent
to B,, and that the load-carrying net area of PQR is equivalent
and is usually called net stress or effective stress.
to the surface element P*Q*R* in Bf. The segments P*Q*,
The notion of Kachanov-Rabotnov's theory expressed by
P*R* and the area of P*Q*R* will be designated by the vec-
equations (l)-(3) has provided an important basis for the
tors dx*, dy* and v*dA* in the vector space E. Since the net
subsequent development of continuum damage mechanics,
area reduction due to damage is brought about not only in the
and considerable efforts have been made in order to extend the
plane PQR but also in the planes of other orientations, the
theory to the general multiaxial states of damage (Chaboche,
direction of the vectors vdA and v*dA* do not always coincide
1981; Hult, 1979; Kachanov, 1986; Krajcinovic, 1984;
with each other.
Lemaitre and Chaboche, 1985; Murakami, 1983). However,
the extension of the notion of equation (2) to model more According to the notion of equation (2), the damage
general states of damage has always been accompanied by variable D for general damage states can be defined properly,
significant difficulties of the proper description of anisotropic if the transformation between vdA and v*dA* of the two con-
damage states induced by the three-dimensional distribution figurations B, and Bf of Fig. 2 is specified in some way. This
of microscopic cracks and cavities. transformation between the area vectors of these two con-
figurations can be identified by introducing a fictitious defor-
In the one-dimensional case of Fig. 1, A* of equation (2) mation at point P from the current damaged configuration B,
implies the effective load-carrying area of the current dam- to the corresponding fictitious undamaged configuration Bf.
aged state (b). Thus we can assume a fictitious undamaged If we represent the fictitious deformation gradient from B, to
bar subject to the applied load t with the cross-sectional area Bj by G, the segments dx* and dy* in Bf are given as follows
A *, mechanically equivalent to the current damaged state (b)
as shown in Fig. 1 (c); we call it the fictitious undamaged state.
Then, the damage variable D, namely the damage state of Fig. dx* = Gdx, dy*=Gdy (4)
1(b), can be readily defined by equation (2), if the relation be-
tween the areas A and A * of the two mechanically equivalent In view of Nanson's theorem, the area vector v*dA* in Bf is
states is specified in some way. related to the vector vdA in B, as follows

2.2 Representation of Damage States With Respect to the

Current Damaged Configuration. By assuming that the prin-
Fictitious undamaged configuration defined in Fig. 2 is similar to the
cipal effect of the material damage consists of the net area physical configuration proposed by Stolz (1984) when he discussed the deforma-
decrease due to three-dimensional distribution of microcracks tion of materials with microstructure.

Journal of Applied Mechanics JUNE 1988, Vol. 55/281

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 v*dA* = (l/2)dx*xdy* K(G-l)T = (¥e)T[KG-1]T(¥e)T (12a)
= (l/2)(Gdx)x(Gdy) e T
v = (r ) v(¥ y e T
(126)
= K(G~l)T (vdA), K=detG (5) By representing the principal values and the corresponding
where ( )T denotes the transpose of the second rank tensor. principal directions of the tensors D and D by D, v and D, v,
The argument developed so far implies that the damage respectively, we have the proper equations of D and D as
state of Bt can be described by the linear transformation follows
K(G~l)T of equation (5). By introducing a new tensor ( I - D ) ( D - Dl)P=0 (13a)
to represent the tensor K(G~')T in the form of
(D-DI)v =0 (13b)
K(.G-1)T = I-T> (6a)
Substitution of equation (126) into equation (13a) gives
or
(D-JDI)[(F';)-7'?1 = 0 (14)
G = #[(I-D)7]-1=JK'(I-D)-7', (66)
Comparison between equations (14) and (136) furnishes the
equation (5) can be written as follows relation between principal directions of D and D:
v*dA* = (I-D)(vdA) (7)
v = (Fe)Tv/\\(Fe)TvW (15)
where I denotes the unit tensor of rank two. Thus the second
rank tensor D of equation (7) is an internal state variable where II II represents the norm of a vector in the Euclidean vec-
representing the general damage state (anisotropic damage) of tor space E.
materials, and will be called the damage tensor with respect to As mentioned already, damage states should be identified
the current damaged configuration B,. by the tensor D unless the elastic deformation is small.
However, the tensors D and D are related by equation (126),
2.3 Representation of a Damage State With Respect to the and coincide with each other particularly when the elastic part
Elastically Unloaded Damaged Configuration. The forego- F e of the deformation gradient F is small.
ing discussion is concerned with the configuration B, and,
therefore, the material is subject to damage and macroscopic 2.4 Restriction and Interpretation of Damage Ten-
deformation at the same time. Namely, the damage tensor D sors. We have shown that a damage state in general can be
defined by equation (7) depends also on the current state of described by the tensor D of equation (11). Let us now discuss
deformation. This means that the tensor D can no more the mathematical restrictions on D.
describe properly the internal state of damage in the case of The vector v*dA* of equation (11) is the equivalent area
large macroscopic deformation. vector in Bf transformed from vdA in B, as a result of the net
Concerning the damage state of the material element in the area decrease due to cavity distribution [Fig. 2 (d, e)], and thus
current configuration B,, only the irreversible change of inter- the scalar product of P*dA * and vdA should be positive, i.e.
nal structure caused in the process from B0 to B, is essential. (P*dA*)-(PdA)>0, [for V vdA e E] (16)
Thus we take the elastically unloaded configuration B, from
B, as shown in Fig. 2(d), and denote the elastic deformation Substitution of equation (13) into equation (16) gives
gradient from B, to B, by ( F e ) " ' . We further assume that the [(l-T5)(PdA)]-vdA)>0, [for V vdA€ E] (17)
load carrying net area of the area element PQR of B, is
decreased because of void formation, and that the net area is which means that the tensor ( I - D ) is a positive tensor.
mechanically equivalent to the area element P*Q*R* of the Let us further take the Cartesian decomposition of the ten-
corresponding fictitious undamaged configuration Bf [Fig. sor ( I - D )
2(e)]. By taking account of the equivalence between the con- ( I - D ) = ( I - D ) S + (I-D)- 4 (18)
figurations Bt and Bj, the elastic deformation gradient cor- where ( I - D ) 5 and ( I - D ) ^ s t a n d for the symmetric and the
responding to the unloading from Bj to Bf is given by (Fe) ~', antisymmetric part of ( I - D ) . In view of the contribution of
as before. (I - D)A in the transformation (11), we have the relation
According to a similar procedure as in the _preceding_sec-
tion, the relation between the area vectors vdA and v*dA* is (p*dA*)'(vdA) = [(I-D)- 4 (PdA)]-(PdA)
specified by introducing a fictitious deformation gradient G = -(PdA)-[(I-D)A(vdA)]
from Bt to Bf. Then, a similar operation as in equations (4)
= 0 (19)
and (5) furnishes the following relations
This relation implies that the tensor ( I - D ) ' 4 transforms the
dx* = Gdi, df* = Gdy (8)
area vector vdA of an arbitrary surface element in a damaged
P*dA* = (l/2)dx*xdy* material into the equivalent surface element P*dA* which is
= K(G-l)T(PdA), K=detG (9) always perpendicular to vdA independently of the direction of
vdA; i.e., this tensor ( I - D ) ' 4 is physically irrelevant as a
By introducing a new tensor D to represent the second rank transformation which represents the net area reduction due to
tensor K(G"l)T as follows damage. Therefore, the antisymmetric part (I - D)-4 of equa-
J?(G-') 7 ' = I - D , (10) tion (18) can be excluded from the following discussion, and
then we can assume symmetry of the damage tensor D. Thus
equation (9) leads to an alternative form the tensor D always has three orthogonal principal directions
v*dA*=(l-f>)(vdA). (11) n,- (/= 1, 2, 3) and the corresponding principal values Dh and
Unlike the tensor D, the tensor D represents the damage state can be expressed in the canonical form
of B, independently of the elastic deformation in Bt; i.e., it
3
represents the reduction of net area caused bythe development
of microscopic voids. Hereafter, the tensor D will be called a D= E Dja, <g> n, (20)
damage tensor or a damage tensor with respect to the elastical-
ly unloaded configuration B,. In order to examine the properties of D, let us take two sets
If we note the deformation from the configuration Bt to Bt, of principal coordinates systems Ox^x2x3 and 0*XjX2Xj of the
together with that from Bf to Bf, we have the following rela- tensor D which pass the points P, Q, R and P*, Q*, R* in B,
tions between G and G, and D and D: and Bf in Fig. 2. Then we construct two tetrahedra OPQR and

282/Vol. 55, JUNE 1988 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 distributed cracks and cavities, the damage state correspond-
ing to arbitrary distribution of these voids, however com-
plicated it may be, can be described by the second rank sym-
dA, =(1-D,)dAi metric tensor D defined by equation (20), or by the tensor D of
equation (22); in these cases, equations (20) and (22) provide
general description of damage states. The canonical expres-
sions (20) or (22), on the other hand, imply that the tensors D
and D cannot describe the damage states which have more
-X2
complicated symmetry than orthotropy.
When Murakami and Ohno (1981) developed an anisotropic
damage theory of creep, they derived the second rank sym-
metric tensor D of equation (22) with regards to B,, by
representing the effects of microscopic grain-boundary
Fig. 3 Interpretation of the damage tensor D; (a) Elastically unloaded
damage configuration; (b) Elastically unloaded fictitious undamaged cavities in terms of a dyadic product of the unit normal vector
configuration to the relevant boundary. They discussed also the net area
reduction from the Cauchy tetrahedron in a damaged material
to that in the equivalent fictitious undamaged material, and
Q*p*Q*ji* ag s^QY^ i n pjg 2(a, b) which consist of the area showed the relation of equation (7). A similar damage tensor
elements PQR, P*Q*R* and the facets perpendicular to the was employed also by M. Kachanov (1980) to develop a con-
principal coordinates axes xlt x2, and x3. Substitution of tinuum model of cracked elastic materials. Contrary to
equation (11) into equation (20) furnishes Murakami and Ohno, Betten (1983), thereafter, started from
the transformation of a Cauchy tetrahedron in the current
*dA* = £} (l-DJdAiAj damaged configuration to that in the corresponding fictitious
undamanged configuration to define the damage variable, and
derived equation (22) by postulating the existence of a specific
= n, dA^ + ^dA^ + n^dA* (21a) plane in the material, which does not change its direction by
dAf= (l-Dj)dA, (no sum on /; / = 1, 2, 3) (216) damage. In the present paper, on the other hand, we employed
where dAt = v,dA and dA* = pfdA* (i = 1, 2, 3) signify the a less restrictive assumption that the damage states in
facet areas of the tetrahedra on the coordinate planes in B, materials can be characterized by the reduction of effective
and Bf, respectively; see Fig. 3(«, b). According to equation load-carrying area caused by the distributed microscopic
(20), the principal values D, of the tensor D can be interpreted cracks and cavities, and developed a systematic theory of
as the net area reductions or the effective void area densities general material damage caused by distributed cavities.
on the three principal planes of D in B, and Bf, as shown in
Fig. 3. 3 Definition of Effective Stresses
The preceding arguments apply also to the damage tensor D
in B,; namely, the tensor (I - D) is a positive tensor, and equa- 3.1 Effective Stress in Regards to Current Damaged Con-
tions (20) and (21) have the following forms in this case figuration. By use of the damage tensor D of equation (20)
[or tensor D related to D by equation (126)], as an internal
state variable for material damage, the constitutive equation
D D An/ ® n; (22)
and the evolution equation of damaged materials can be
established in the framework of the ordinary theory of in-
elasticity. In the usual procedure of continuum damage
v dA
* * = E (1-A)^4;n,. mechanics, however, the magnified effects of stress due to
damage are often represented by the notion of an effective
= n,fiL4f + n2GL42* + n3GL43* (23o) stress, i.e., the constitutive equation of a damaged material is
derived from that of the undamaged material (Lemaitre, 1985;
dA*= (1 - D , ) dAj (no sum on /; i = 1, 2, 3) (235) Lemaitre and Chaboche, 1978, 1985) by replacing the Cauchy
where dAt = vtdA and dA* = vfdA* designate the facet areas stress in the equation with the corresponding effective stress
of the tetrahedra in B, and Bf; see, Fig. 4 (b, c). on the basis of the hypothesis of strain equivalence (Lemaitre
So far it has been elucidated that, as far as the material and Chaboche, 1978, 1985). The effective stress is applied also
damage is characterized by the net area reduction due to to formulate the evolution equation of damage.

dAi=(1-D,)dA.

•tdA=0(udA)
(OY(U*dA*)
UdA

dA, dA*=(1-D2)dA2 u*dA"=(I-D)(udA)

Fig. 4 Effective stress tensors with respect to current and initial configurations; (a) Initial undamaged
configuration; (b) Current damaged configuration; (c) Fictitious undamaged configuration

Journal of Applied Mechanics JUNE 1988, Vol. 55/283

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 According to the notion of effective stress of equation (3), U)
(b)
the effective stress with reference to the current damaged con-
figuration B, (i.e., effective stress for the Cauchy stress a) for
the general states of damage can be defined as a stress a*
which would be induced in the tetrahedron 0*P*Q*R* in Bf
[Fig. 4(c)], when the surface element P*Q*R* is subject to the
indentical force vector t w dA to that on the element PQR in
Bt, Fig. 4(b). This stress a* corresponds exactly to the first
Piola-Kirchhoff stress tensor concerning the reference con-
figuration Bj, and is given as follows
o*=K~1Go (24)
In view of equation (6), this equation can be written in an
alternative form Fig. 5 Cracked elastic material and stress path; (a) Two-dimensional
a* = K~lK[(l-D)~T]a elastic material with randomly distributed microscopic cracks;
(b) Stress path
= (I-D)-'cr (25)
where we have employed the symmetry of D discussed in the
preceding section. Figure 4(c) shows a mechanical interpreta- Therefore, it should be noted that these stresses do not satisfy
tion of a*. the equilibrium condition in the relevant configuration. 3
The effective stress a* of equation (25) is asymmetric. Since
it is inexpedient to formulate constitutive and evolution equa-
tions of damaged materials by using the asymmetric tensor a*,
4 Application to Problems With Crack Closure
we need to symmetrize it in some appropriate way. One possi-
ble way is to adopt the symmetric part of the Cartesian decom- 4.1 Extension of Effective Stress. In the case of small
position of a* elastic deformation, the states of material damage can be
S=[a*] s = (l/2) [ ( I - D ) - ' ( 7 + (7[I-D)" 1 ] (26)2 described by a second rank symmetric tensor D, and the
magnified effect of stress is expressed by the effective stress
So far we have assumed that the mechanical state of the tensors S of equations (26) and (27).4 So far, it has been
material in the current damaged configuration B, is equivalent assumed that microscopic cavities characterizing damage do
to that in the fictitious undamaged configuration Bf. Thus we not transmit force through their surfaces.
can interpret that the mechanical effect of the Cauchy stress a
However, when damage is brought about by a set of
in B, is magnified to the effective stress S of equation (26) as a
microscopic penny-shaped cracks in the material, and inelastic
result of the effective area reduction due to damage.
deformation as well as elastic deformation is small, the effects
3.2 Effective Stress in Regards to Initial Undamaged Con- of crack closure under compressive stress are often important
figuration. Similarly to the second Piola-Kirchhoff stress and not negligible (Horii and Nemat-Nasser, 1983; Kyoya, et
tensor with respect to the initial undamaged configuration B0 al., 1985; Ladeveze and Lemaitre, 1984). This phenomenon is
observed, e.g., in the differences in tensile and compressive
Z = J¥-la(¥-i)T, J=detF (28) properties in fracture strength and elastic moduli of brittle
we can define the effective stress concerning B0. materials such as concrete and rocks, or in the mean-stress
The magnified effect of stress in B, has been given by the ef- dependence of fatigue strength of metals. The
fective stress S. We now introduce a stress vector corres- damage theory discussed so far can be extended to these
ponding to S as shown in Fig. 4(b) problems if we define a new effective stress by incorporating
the effect of closure of microcracks.
i *dA=ST(i>dA) (29)
As was done by Ladeveze and Lemaitre (1984), let us first
Then, we will make a similar assumption to that for the stress decompose the Cauchy stress a as follows
E, and postulate that the surface element P 0 Qo^o m #o °f Fig-
(33a)
4(a) is subject to the force vector transformed from equations
(29) (a,) 0 0
tw 0*dA0-- :F~'t. « *dA
(30) Ml 0 (o-2> 0 (336)
Denoting the effective stress concerning the configuration B0
by T, we have Cauchy's formula for B0 0 0 (a3)
T
t\*dA0 = T (v0dA0) (31)
By using Nanson's formula together with equations (30) and
(31), the effective stress concerning B0 is given by the follow-
'Effective stresses of S and T discussed in Sections 3.1 and 3.2 represent the
ing relation magnified effects of stress induced by the load-carrying net area due to the
T = JF~1S(F~l)T (32) damage. Strictly speaking, therefore, these effective stresses should be related to
local decohesion effects such as damage and fracture. Deformation of material
which is similar to equation (28). in general depends not only on the cavity distribution in specific planes but also
The effective stress S and T are fictitious stresses which on the spatial distribution of cavities. Thus, as regards deformation, it is
sometimes more reasonable to define effective stresses other than S and T. This
represent the magnified effect of stress due to damage. problem was discussed by Murakami and Imaizumi (1982), Murakami and
Ohno (1981) and Murakami, et al. (1986).
2
Other ways of symmetrization are possible. Cordebois and Sidoroff (1982) The damage tensor D of equation (22) and the effective stress S of equations
for example, proposed the following effective stress (26) and (27) have been applied mainly to damage problems of metals; i.e., the
anisotropic damage of creep (Murakami and Ohno, 1981; Murakami, et al.,
S = (I-D)" 1 / 2 ff(I-D)" 1 / 2 (27) 1986); the effects of prior plastic damage on the subsequent creep damage pro-
cess (Murakami and Sanomura, 1986); creep crack growth under nonpropor-
which gives the elastic potential of damaged materials by replacing the Cauchy tional loading (Murakami, Kawai, and Rong, 1988); the change of elastic
stress a in the corresponding elastic potential of undamaged materials. moduli caused by plastic damage (Cordebois and Sidoroff, 1982); etc.

284/Vol. 55, JUNE 1988 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 0
[<-*>] =
(3*/2)T0. (37c)
Substitution of equations (35) and (37) into equation (34) gives
the components of effective stress Sy in the coordinate system
Oxxx2. Applying further a coordinate transformation to S,y,
we have the component of the effective stress S in the system
Ox\X2 as follows
[S] = (*r 0 /4)
"1/(1 -D)- 3/(1 -CD) 1/(1 -D) + 3/(1 -CD)
(38)
1/(1 -D) + 3/(1 -CD) 1/(1 -D)- 3/(1 -CD)

If we employ the hypothesis of strain equivalence

0.5
(Lemaitre, 1985; Lemaitre and Chaboche, 1985), the elatic
deformation of the cracked body is furnished by Hooke's law
r/(r0/G)
by replacing the Cauchy stress in it with the effective stress of
Fig. 6 Elastic behavior of cracked material under combined equation (38) as follows
hydrostatic pressure and shear
7 / ( T 0 / G ) = (1/4)[1/(1 -D) + 3/(1 - CD)](T/T0) (39)
(-a,) 0 0 The solid line in Fig. 6 shows the results of equation (39) for
the case of D = 0.4, C = 0 . 2 . The dashed line and the chain line
[<-»>] = 0 <-a2> 0 (33c) in the figure, furthermore, are the results of the similar
analyses for the stress paths @ and (3) in Fig. 5(b). As
0 0 <-a 3 }J observed in Fig. 6, in the case of the proportional stress path
(T), both the closed cracks and the open cracks exist
where ( ) is the Macauley bracket, and 07 (/= 1, 2, 3) repre- throughout the whole process of loading, and the final state P
sent principal values of a. Because of the effects of crack is attained. For path (2), a larger number of cracks than for
closure, the effective area reduction due to D under com- path (T) remain open up to point Q, and thus we have larger
pressive stress is smaller than that under tensile stress. strain than (T). However, after point Q, cracks start to close
Denoting this relation by a parameter C(0 < C < 1) and, in view with the increase of hydrostatic pressure and the shear strain
of equation (33), equation (26) can be extended as follows decreases. For stress path (3), on the other hand, only
hydrostatic pressure acts up to point R, and no shear deforma-
S = (l/2)[(I-D)- 1 (<T> + (a)(I-D)- 1 ] tion occurs. For the subsequent path RS, since all cracks re-
- ( l / 2 ) [ ( I - C D ) - ' ( - < r > + (-<r>(I-CD)- 1 ] (34) main closed, we have the smallest rate of strain increase
among these paths. Finally, after point S, part of cracks start
where C = 0 and C = 1 correspond to the case of perfect to open and the increase of strain is accelerated. Similar results
closure of cracks (unilateral condition) and the case without to Fig. 6 have been obtained by Horii and Nemat-Nasser
crack closure (bilateral condition), respectively. (1983), when they analyzed the overall elastic moduli and the
4.2 Elastic Deformation of Elastic-Brittle Materials Con- load-induced anisotropy of an elastic body containing
taining Distributed Microcracks. As an application of equa- uniformly distributed microscopic cracks by means of self-
tion (34), let us now analyze the two-dimensional elastic defor- consistent methods of micromechanics. 5
mation of a cracked body as shown in Fig. 5. The elastic body
is in the state of plane stress and subject to combined 5 Conclusions
hydrostatic pressure/? and shear stress T according to the stress
paths of Fig. 5(b). If the body contains a number of randomly A continuum theory was developed for the systematic
distributed microscopic penny-shaped cracks, the damage modeling of general three-dimensional anisotropic states of
state is isotropic and can be described by a damage tensor material damage.
Directional features of material damage are especially im-
D=Z>I = Z>[n1(g)n1+n2(x)n2] (35) portant in many engineering problems, ranging from creep
Let us first consider the stress path (T) in Fig. 5(b). The damage, fatigue damage and spall damage in metals to elastic-
components of the stress tensor with respect to the coordinate brittle damage in geological materials and concrete. The pres-
systems Oxlx2 and Ox,x2 are ent paper provides a consistent framework to analyze these
problems from a continuum damage mechanics point of view.
'-p T1 I" - (k/2)r0 kr0
Acknowledgment
-(*/2)T0.
The present paper is the outcome of discussions made dur-
0 < f r < l ; k = r/T0 (36) ing the author's stay at Universite Paris 6 from April through
July, 1986. The author is grateful to his host, Prof. Jean
Lemaitre of Universite Paris 6, for his hospitality and his
(k/2)r0 0 stimulating discussions. This work was supported in part by
= [ < * > - < - * > ] (37fl)
m= 0 -(3£/2)T0J
the Ministry of Education of Japan under the Grant-ih-Aid
for Scientific Research A (No., 61420024) for the fiscal year of
1986.
Xk/2)To 0
[(&)] =
0 0 *• ' 5
They assumed a two-dimensional state of plane strain.

Journal of Applied Mechanics JUNE 1988, Vol. 55/285

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 References Leckie, F. A., and Onat, E. T., 1981, "Tensorial Nature of Damage Measur-
ing Internal Variables," Physical Non-Linearities in Structural Analysis, Hult,
Betten. J., 1983, "Damage Tensors in Continuum Mechanics," Journal de
J., and Lemaitre, J., eds., Springer, Berlin, pp. 140-155.
Me'canique Theorique et Applique'e, Vol. 2, pp. 13-32.
Lemaitre, J., 1985, "Continuum Damage Mechanics Model for Ductile Frac-
Chaboche, J. L., 1981, "Continuous Damage Mechanics-A Tool to
ture," ASME Journal of Engineering Materials and Technology, Vol. 107, pp.
Describe Phenomena Before Crack Initiation," Nuclear Engineering and
83-89.
Design, Vol. 64, pp. 233-247.
Chaboche, J. L., 1984, "Anisotropic Creep Damage in the Framework of Lemaitre, J., and Chaboche, J. L., 1978, "Aspect Phenomenologique de la
Continuum Damage Mechanics," Nuclear Engineering and Design, Vol. 79, pp. Rupture par Endommagement, : ' Journal de Me'canique Applique'e, Vol. 2, pp.
309-319. 317-365.
Cordebois, J. P., and Sidoroff, F., 1982, "Damage Induced Elastic Lemaitre, J., and Chaboche, J. L., 1985, Mecanique des Materiaux Solides,
Anisotropy," Mechanical Behavior of Anisotropic Solids, Boehler, J. P., ed., Dunod, Paris.
Martinus Nijhoff, The Hague, pp. 761-774. Murakami, S., 1983, "Notion of Continuum Damage Mechanics and Its Ap-
Hayhurst, D. R., and Storakers, B., 1976, "Creep Rupture of the Andrade plication to Anisotropic Creep Damage Theory," ASME Journal of Engineer-
Shear Disk," Proceedings of the Royal Society of London, Series A, No. 349, ing Materials and Technology, Vol. 105, pp. 99-105.
pp. 369-382. Murakami, S., Kawai, M., and Rong, H., 1988, "Finite Element Analysis of
Horii, H., and Nemat-Nasser, S., 1983, "Overall Moduli of Solids With Creep Crack Growth by Local Approach," International Journal of Mechanical
Microcracks: Load-Induced Anisotropy," Journal of Mechanics and Physics of Sciences, (in print).
Solids, Vol. 31, pp. 155-171. Murakami, S., and Imaizumi, T., 1982, "Mechanical Description of Creep
Hult, J., 1979, "CDM-Capabilities, Limitations and Promises," Mechanisms Damage State and Its Experimental Verification," Journal de Me'canique
of Deformation and Fracture, Easterling, K. E., ed., Pergamon, Oxford, pp. The~orique et Appliquie, Vol. 1, pp. 743-761.
233-247. Murakami, S., and Ohno, N., 1981, " A Continuum Theory of Creep and
Kachanov, L. M., 1958, "On Rupture Time Under Condition of Creep," Creep Damage," Creep in Structures, Ponter, A. R. S., and Hayhurst, D. R.,
Izvestia Akademi Nauk SSSR, Old. Tekhn. Nauk, No. 8, pp. 26-31 (in eds., Springer, Berlin, 1981, pp. 422-444.
Russian). Murakami, S., and Sanomura, Y., 1986, "Analysis of the Coupled Effect of
Kachanov, L. M., J986, Introduction to Continuum Damage Mechanics, Plastic Damage and Creep Damage in Nimonic 80A at Finite Deformation,"
Martinus Nijhoff, Dordrecht. Engineering Fracture Mechanics, Vol. 25, pp. 693-704.
Kachanov, M. L., 1980, "Continuum Model of Medium with Cracks," Jour- Murakami, S., and Sanomura, Y., and Saitoh, K., 1986, "Formulation of
nal of the Engineering Mechanics Division, Proceeding of ASCE, EM5, pp. Cross-Hardening in Creep and Its Effect on the Creep Damage Process of Cop-
1309-1051. per," ASME Journal of Engineering Materials and Technology, Vol. 108, pp.
Krajcinovic, D., 1983, "Constitutive Equations for Damaging Materials," 167-173.
ASME JOURNAL or APPLIED MECHANICS, Vol. 50, pp. 355-360. Onat, E. T., and Leckie, F. A., 1988, "Representation of Mechanical
Krajcinovic, D., 1984, "Continuum Damage Mechanics," Applied Behavior in the Presence of Changing Internal Structure," ASME JOURNAL OF
Mechanics Reviews, Vol. 37, pp. 1-6. APPLIED MECHANICS, Vol. 55, pp. 1-10.
Kyoya, T., Ichikawa, Y., and Kawamoto, T., 1985, " A Damage Mechanics Stolz, C , 1984, "Etude des Milieux a Configuration Physique et Applica-
Theory for Discontinuous Rock Mass," Numerical Methods in Geomechanics, tion," Comptes Rendus Academie des Sciences Paris, Serie II, Vol. 299, No. 17,
Kawamoto, T., and Ichikawa, Y., eds., A. A. Balkema, Rotterdam, pp. pp. 1153-1155.
469-480. Rabotnov, Yu. N., 1969, Creep Problems of Structural Members, North-
Ladeveze, P., and Lemaitre, J., 1984, "Damage Effective Stress in Quasi- Holland, Amsterdam.
Unilateral Conditions," The 16th International Congress of Theoretical and Vakulenko, A. A.,and Kachanov, M. L., 1971, "Continuum Model of
Applied Mechanics, Lyngby, Denmark. Medium with Cracks,' Mekhanika Tverdogo Tiela, No. 4, pp. 159-166.

286/Vol. 55, JUNE 1988 Transactions of the ASME

Downloaded From: http://appliedmechanics.asmedigitalcollection.asme.org/ on 02/19/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use