A Continuous Damage Mechanics Model for Ductile

Fracture
Jean Lemaitre

To cite this version:

Jean Lemaitre. A Continuous Damage Mechanics Model for Ductile Fracture. Journal of Engineering
Materials and Technology, 1985, 107 (1), pp.83-89. �10.1115/1.3225775�. �hal-03609806�

HAL Id: hal-03609806

https://hal.science/hal-03609806
Submitted on 16 Mar 2022

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 1')

A Continuous Damage Mechanics Jean Lemaitre

Professeura l'Universite Paris 6,
Laboratoire de Mecanique et Technologie,

Model for Ductile Fracture E.N.S.E. T/Universite Paris 6/C.N. R.S.

94230 Cachan, France
A model of isotropic ductile plastic damage based on a continuum damage variable,
on the effective stress concept and on thermodynamics is derived. The damage is
linear with equivalent strain and shows a large influence of triaxiality by means of a
damage equivalent stress. Identification for several metals is made by means of
elasticity modulus change induced by damage. A comparison with the McC!intock
and Rice-Tracey models and with some experiments is presented for the influence of
triaxiality on the strain to rupture.

1 Introduction
The phenomenon of initiation and growth of cavities and Colloquim on "Damage Mechanics" held in Cachan (France)
microcracks induced by large deformations in metals and in 198 1). Models presented here are in the framework of that
called "ductile plastic damage," has been extensively studied thermodynamics which gives the possibility of identifying the
by means of micro-mechanics anlaysis. In the pioneer works damage by means of its coupling with elasticity.
of McClintock [l], Rice and Tracey [2] and subsequent
studies, defects are taken into account by analyzing their 2 Elements of Continuous Damage Mechanics
geometry in a continuous matrix using the procedure of the
As many developments of the theory of continuous damage
mechanics of continuous media.
mechanics have already been published [9, 10, 12], only the
At that microscale, a good representation of physical
principal features used to build a new model of ductile plastic
mechanisms can be introduced, but difficulties arise when
damage are summarized here.
these analyses have to be included in large scale structures to
predict ductile failures. The main reason is the lack of ac­ 2.1 Damage Variable. Consider a damaged body in which
curacy of local stress calculations for the microscale level. a volume element at macroscale level has been isolated (Fig.
Between that microscale level, say 10-3-10-2 mm, and the 1).
structure scale level, say 102 - 103 mm, there exists a Let S be the overall section area of that element defined by
macroseale level of constitutive equations for strain behavior. its normal n. In that section the microcracks and cavities have
The continuous damage mechanics approach deals within that intersections of different shapes, let SD be their total area.
macroscale defining a damage variable as an effective surface Let S be the effective resisting area taking into account this
density of cracks or cavity intersections with a plane. Of area SD, the microstress concentrations in the vicinity of
course, at that macroscale it is difficult to introduce much discontinuities and the interactions between closed defects.
physics but on the other hand this damage variable is easy to
S<S-SD
introduce in structural calculation, especially with the concept
of effective stress. This stress, written as the mean density of The concept of effective stress associated to the hypothesis
forces acting on the elementary surface that effectively resists, of strain equivalence described below avoids the calculation
has been introduced by Kachanov in 1958 to model creep of S and, by definition, the damage variable D associated
rupture [3]. This has been the starting point of continuous with the normal n is:
damage mechanics developed further for dissipation and low
cycle fatigue in metals (Lemaitre 1971 [4]), for coupling s-s
Dn = ­
between damage and creep (Leckie 1974 [5]), between damage S
and cyclic creep (Hult [6]), for high cycle fatigue (Chaboche
1974 [7]), for creep fatigue interaction (Lemaitre-Chaboche
1974 [8]) . . . Later, the thermodynamics of irreversible
processes provided the necessary scientific basis to justify
continuous damage mechanics as a theory (Chaboche 1978
[9], Lemaitre-Chaboche 1978 [ 10], Murakami [ 1 1] , Cor­
debois-Sidoroff [ 12] Krajcinovic [ 13]) with many develop­
ments (see for examples papers presented at the Euromech

Fig. 1 Damaged element

1
 From a physical point of view the variable D,, is the 2.2 Thermodynamics. In order to model elasticity,
corrected area of cracks and cavities per unit surface cut by a thermal effects, plasticity and damage within the hypothesis
plane perpendicular ton. of isotropy for the three phenomena, the following variables
From a mathematical point of view, as S approaches zero, have to be introduced [IO]:
then D11 is the corrected surface density of discontinuities in
the body relative to the normal n.
Observable Internal Associated
D11=0 corresponds to the undamaged state; variables variables variables

D,, = I corresponds to rupture of the element into two Elastic strain stress tensor
parts; tensor
------------ ----- a

characterizes the damaged state.

Temperature entropy
T----- ----- s

Hypothesis of Isotropy. In the general case, cracks and

Accumulated Radius of yield
voids are oriented and D11 is a function of n. This leads to an
plastic strain surface: R ,. + R
intrinsic variable of damage which can be a second order
p R .
tensor [12) or a fourth order tensor [9)' depending upon the ·------------

hypothesis made. In this paper we restrict ourselves to Damage Damage strain

isotropic damage, the cracks and voids being equally energy release
distributed in all directions. D11 does not depend upon n and rate
the intrinsic damage variable is the scalar D. D y

·Concept of Effective Stress. If F is the load acting on the

section S of the element considered in Fig. I, T 1''/ Sis the
=
pis defined by:
usual stress vector which leads to the Cauchy stress tensor u
(u. n T). The quantity S = S (I - D) is the effective area
=

which effectively carries the load F. By definition:

- F T t" is the Euler-Almansi plastic strain tensor in large defor­

T=�= - mation theory, its rate being defined from the total strain rate
S 1-D
f by:

i/' = i: - i:"
is the effective stress vector which, since D is a scalar, leads to
the effective stress tensor (ii.n 'f)
=

Thermodynamic Potential. Taking the free-energy "1 as

u thermodynamic potential, it is assumed that it is a convex
ii= function of all observable and internal variables. Using the
1-D hypothesis that the elasticity and plasticity behaviors are
uncoupled gives:
Beside the hypothesis of isotropy, we assume here that the
mechanical effects of cavities and microcracks are the same "1= ,µ,, ( t", T,D) + "1,, ( T,p)
both in tension and compression. As it is generally .not the In order to obtain linear elasticity coupled with damage by
case in practice, this limits the applicability of the theory to means of the effective stress, "1e must be quadratic in t" and
those cases where "compressions" are small. linear in (I -D). If a is the fourth order tensor of elasticity
and p the density
Hypothesis of Strain Equivalence (4). It is assumed that the
I

�
strain behavior is modified by damage only through the ef­ l/Je= -- a: t'': t''(l -D)
fective stress: 2p

The strain behavior of a damaged material is represented by The damaged elasticity law is:
constitutive equations of the virgin material (without any
al/;,,
damage) in the potential of which the stress is simply u=p --- =a: te(I -D)
replaced by the effective stress. aE"
and the variable y associated with D, by the power dissipated
Examples: ( - y D) in the phenomenon of damage, is defined by:
-one dimensional linear elasticity of a damage material:
al/;.. I

Ee=
a =
a
y=p- =--a: t'': t''
aD 2
E (1-D)E
Damage Criterion. The density of elastic strain energy We
Ee being the elastic strain and E the Young's modulus. being defined as:
-Ramberg-Osgood equation for plastic strain-hardening
dW,. =u: dt"

(a)M [ ]M
evolution:

u
if we replace dt" by its value taken from the damaged
E = = elasticity law written for du 0 at constant temperature one
=

,, K (1-D)K

where E" is the plastic strain and K and M material coef­

ficients. 'The symbol ( : ) means the tensorial product contracted on two indices.

2
can see that -yis one half of the variation of We due to an result! ( -yiJ) is the energy dissipated within the damage
infinitesimal increase of damage at constant stress and process for decohesion of the material. By analogy with the
temperature. This gives for -ythe name of "damage strain toughness criterion in fracture mechanics one may postulate
energy release rate" (as Gin fracture mechanics!) [9]. the following rupture criterion:
dWe
-y=
�
2 dD
)u 'T
The damage process gives rise to initiation of a macrocrack
for a critical value of -y, that value Ye being a charac­
teristic for each material.
The expressions ofyand We show that:
-y=ye - crack initiation
we
-y=
1- D
This corresponds to a critical value of the damage variable D
which can be calculated from the uniaxial case for the rupture
We is calculated as the sum of shear strain energy and volume conditions:
dilatation energy with the tensor of elasticity written in terms
of Young's modulusE and Poisson's ratiov, that is the
following relations between elastic strain deviator ee and
stress deviator s, the hydrostatic strain c'i-r = 1
/
3 tr(te) and
the hydrostatic stress aH = 13
/ tr(u):
v s
1+ 1-2v aH
e"=-- -­ c'i;=---­
E 1-D E 1-D aR == --
ITR
(2Eye ) "
==
''
is the condition of brittle
I -De fracture of Orowan.
We obtain
- 1 [l +v s:s 1-2v aH 2
y= 3
+
2 E (l -D)2 E (l -D)2 J is the critical value of damage
or with the Von Mises equivalent stress for plasticity at macrocrack initiation.
1/2
3
lleq = ( l S: )
S
Many experiments have shown that:
.2:5D e:S.8
ae[ 2
-y= / 2 2-0+11)+3(1-211( !_ ]
) _!!_!)
aeq
depending upon the materials.
2E(l -D) 3
This quantity can be calculated for an "equivalent" one­ Potential of Dissipation. In order to derive constitutive
dimensional case defined by its stress a*, giving the same equations for evaulation of dissipative variables, the existence
value ofy. of a potential of dissipation is assumed: a scalar convex
function of flux variables (EP, p , b, and the heat flux q) the
1
a H= - a* state variables acting as parameters [ 18].
3 e
<PW p , , D,q; t ,T,p,D)
Other equivalent potentials can be obtained by means of the
-y=
2E(l -D)2 Legendre-Fenchel transform, in particular the partial trans­
form changing iJ to its dual variabley;
Asyis the variable associated with D, it means that evolution
e
of D is governed by values of y; by analogy with the Von <P*<fPp
, y
, ,q; t ,T,p,D)
Mises equivalent stress for plasticity, the quantity: The constitutive equation for damage evolution D is given by
2
[2 a 112 the normality property of that potential:
a*= aeq
3
-(1+v)+3(1-2v(
) _!!__
lleq
)J . a<P•
D= -­
ay
is called damage equivalent stress and can act as a criterion
for damage just as aeq acts as a criterion for plasticity [14].
a• = a*/1- D) is the damage equivalent effective stress and: 3 Models for Ductile Plastic Evolution
a•z Restricting ourselves to isotropic plasticity and isotropic
-y=
2E damage, mathematical models are of a scalar nature. Ductile
platic damage, as plasticity, is a phenomenon which does not
It is interesting to note that a* is equal to the Von Mises depend explicitly upon time.
equivalent stre5s multiplied by a factor function of the Within these hypothesis the main features of ductile plastic
triaxiality ratio u:; I lle q which is very important for damage damage can be described by a potential of dissipation
evolution as shown by many experimental or theroetical restricted to three variables

<P*(y,fl,n
studies [15-17].

Dissipation, Rupture Criterion. Within the hypothesis of written as a power function ofyfor convenience and linear in
uncoupling between intrinsic mechanical and thermal pto ensure the non explicit dependency of D with time:
dissipations, the second law of thermodynamics imposes the
condition of mechanical dissipation being positive, <P*-
So ( -y so+ 1 .
p
)
(s0 + 1) S0
IT: iP-Rp-yD�O
where s0 and S0 are material and temperature dependent.
The processes of plasticity and damage may be independent, The complementary law of evolution of damage derives from
then we must have separately: <P* by
a: iP-Rp�O -yb�o . a<P• -y so+ 1
and
( -y) being positive iJ must be positive which is a trivial
D=--=( --
ay So
p ) •

3
 3.1 One Dimensional Models Written in Terms of Stress. In the expression for y:
It can be shown that several models proposed in the past from [ -(I 0
11 2

3 +P)-J-3(1-2P) (-- ) J
/ ae2
phenomenological considerations may be derived from the -y= 2E(l-D,
---- ,2
o"
"
above general constitutive equation:
In the one-dimensional case of monotonic loading defined
by the stress a and the plastic strain fP'' replace aeq by its value, taken from the Ramberg-Osgood
52
p.=E,. ,,-y=
hardening law coupled with damage and written for the three­
dimensional case:
2E .
With -[ M
ae q
] --=aeq
I
M

a=--- a
P-
(1-D)K or
1-D Kp
(1-D) Then
then follows K2 [ )2 (I+v) 3(1- 2v) ( - ) 2 ]p M) So p
D. = ( 2ES +
aH �

D.- ( --
-
a2
)
2ES0
"O
i"
o
This is the general constitutive equation for ductile plastic
aeq

Replacing i,, by its value taken from the Ramberg-Osgood damage.

law of hardening coupled with damage· by means of the ef­
Integration in the Particular Case of Radial Loading. In
fective stress:
most engineering applications the loading is such that the
(:)
M
directions of principal stresses (different in each point M) are
f" = or
constant with time through the process and may be described
_(2s0+M··I) by:
o
gives: D= --a
---- u(M,t> =cxu>ur(M) (a scalar)
Within this hypothesis, the triaxiality ratio 011 Ia"'1 is
K M
0
-(2ESo)'
M constant with time and it is possible, by integration, to obtain
or, with change of notation: a simple relation between the actual value of damage and D
the accumulated plastic strain p.
If PD is the damage strain
dD= ( ; )'1 da threshold:

This is the model proposed by Broberg (19). Usually a Integration yields:

damage threshold does exist such that:

� [2-o+v
D = ( 2ES0
a<aJJ-D=O

Then a more realistic model has been proposed by Dufailly 3 )

and Lemaitre 120 I: 2so �M 21 c ; M
'.� )
'o
dD- - ( S2(1-D) ) ·12 S2da
a- aJJ
- + 3 ( 1- 2v )
_
__
) (-'!!!_
Oeq
]) �-
M
( p
2 , - pn 2so +

with: (x)=xif>O , (x)=O if x�O This expression can be written in a simpler fashion by in­
Here aJJ, S2,
and s2 are material constants identified from
tension tests in which damage is derived from elasticity
troducing the rupture strain PR as a function of the triaxiality
ratio a11 I aeq corresponding to the intrinsic value of damage at
modulus change as explained in section 3.2. (Identification). failureDe which we assume to be a material property:
These models written in term of stress are difficult to apply
in the range of large strains, where usually ductile plastic
p=pR-D=De
damage occurs, because the stress does not vary very much as
K2 [ 32 +
De= ( 2ES0 (I v)
the material becomes almost perfectly plastic close to rupture.
Then a model written in terms of strains is more suitable
especially for metal forming calculations.

3.2 A Three Dimensional Model Written in Terms of

Strains.Ductile plastic damage generally occurs with large
deformations and in metal forming calculations. Large strain
theory must then be used. Dividing D by De yields:
Now f

a
is the Green Lagrange strain tensor,
is the Cauchy stress tensor,
D=D,. p ( 2so+M
M -pn
21·0 +M
M
t" is the elastic strain tensor defined with respect to
the unstressed state.
Due to the large strain hypothesis the damage is written as
function of total strain instead of plastic strain. Then:
PR
210 IM
M
2so +M
-pf) M )
1/2 In the range of large deformations in metals, the hardening
p= ( 32 ) t: t exponent M is usually very high (a perfectly plastic material
corresponds to M co ) . Otherwise, identifications of one­
=
dimensional models described in section I. show that the so 3.
Formulation. From the potential of dissipation already
chosen in section 3: coefficient is of order of magnitude of unity then
(2s0 + M) IM is of order unity.
-y "O
D= ( --
.

S ) p o
•
Pn and PR depend upon the triaxiality ratio but 1t 1s
physically admissible to assume that this dependence is the

4
 PoIPR Eo, ER, De
11.
same for both quantities and, that the ratio does not These models depend upon material constants for
depend upon triaxiality, and is equal to its value in the one­ damage properties and Poisson's ratio The first one which
dimensional case: also depends upon hardening exponent Mhas to be integrated

PR ER D(t).
in each particular case of history of loading represented by
aHlaeq (t) and p(t)
to obtain the damage evolution

where Eo
and ER
are the one-dimensional strain at damage
gives directly the damag;e D
The second, only valid for radial or proportional loading
as a function oft when aH aeq is I
threshold and at failure. known.

Identification of Parameters. Identification of such models

Eo, ER De
consists in the quantitative evaluation of the three coefficients
and characteristic of each material at each temp­

pR
erature considered. (Mis known from the uniaxial hardening

ER, Eo De
The accumulated strain at rupture can also be expressed as curve.)
a function of aH/aeq and of its value in the one­ and need a measurement of damage which is
dimensional case:

--=-31
somewhat difficult due to the fact that damage does not affect
aH very much any measurable quantity far from the rupture

D
condition.

From the expression of De with (2s0 +M) IM = 1 follows:

Ueq
D
Let us return to the definition of and the effective stress

[21].
concept applied to elasticity: it is possible to measure
through the variations of the elasticity modulus
Writing again the damage elasticity law:
a=Ete or a= E l ( -D )Ee
=ER= 2
[3(1+11)+3(1- 211)( 91 )] -so E being the Young's modulus of undamaged material, the

(1- :: )
----- quantity
K2 E(l-D) =E
E
can be considered as the elasticity modulus of the damaged

=ER [-(1+11
2 )+3(1-211)(
and

)2] -so
material. E being known and measured by a special

PR 3
a
__!_I_
technique described below, then the damage D is evaluated as:

( -"1:L)
D=l-­
:E
Ueq
Oeq
s0 being of the order of 1 as will be checked in section 4.1, the E
final equation is:

P [
D=D / _ �_ _(1_+_11) +
_ 3(1
_ -
_ 211
_
ER-ED
_)( )2] -Eo
_;_:_: _ _ _ _ - ) /
...........r-
I
---
---
- -

I
r -

\ I
I
/
//"

I
I
I
/

! ,'
I
I
In the particular case of one dimension
E-ED )
D=De ( --- E ,'
'

ER-ED I

PR D
which is very simple!

s0 = 1 (2s0 + e IM
With the above expression for the equation for can

1:
Fig. 2 Measure of damaged elasticity modulus
be written in a very simple way using and M)

K2(ER-Eo) E
=

De= -- K2 ER (PR-Po)= --
--
The damage elasticity modulus can be measured through
2ES0 p R 2ES0 tension tests but as damage is always localized in a very small

This allows to replace K212ES0 by Dcl(ER-Eo) in the [22].

region of the specimen some special precautions are needed.

2
They are described in reference For the static method,a
differential constitutive equation for iJ to obtain the final specimen of the shape given in Fig. is needed. Roughly,
results. ductile plastic damage begins when necking starts. As a

Differential model

D=�
ER-ED [�(1+11) +3(1-211)(
r/2
valid for any loading path
r ]p21Mp
p=( 32
aH

i: i
3 Ueq

Model valid for radial

1 D= ER�eEo (P[�o+11)+3(1-211>(�)2]-Eo)
P=( 32 ) /2
loading only

E: E
3 Ueq

5
 a (MPa) D

500
·: Ds .;;A l..L __________
_ �
400
I
300 I
.4
I
I

I
200
.2

I e:R.102

0 10 20 30 40 50 60 70 80 90 100 110 0 20
1
40 60 60 10 �
e:p0 =.35 l:pR = 1.07

(a) (b)
0

Fig. 3 Ductile plastic damage for 99,9 percent copper. T = 20 = C

consequence, rapid change of geometry occurs and the local 4.1 Ductile Plastic Damage Characteristics of Several
strain in the most damaged region must be measured through Metals. The method of identification described above has
very small strain gages, say .5 x .5 mm. As these gages have a been applied to several materials. The results are given below
maximum strain amplitude of 10 to 15 percent then, if the in Fig. 4 with the characteristic values of En, ER, and De. The
damage has to be measured up to strain of 50 or 100 percent linearity of D with E is again weJ! verified in these six exam­
or more, the gages have to be changed, which interrupts the ples.
tests. Last point, a better accuracy is obtained if E is measured
during unloading as shown in Fig. 2. With these precautions 4.2 Influence of Triaxiality on Strain to Failure. A way to
taken, a relative accuracy of 5 percent may be expected for D. check the model with regard to the triaxiality effect is to
When texture induces a variation of the elasticity modulus this compare the strain to failure predicted by p = PR when D
·is generally for small values of strain far below the damage reaches its critical value De with the one predicted by the
threshold. En is then defined by the value of E for which the Mcclintock or Rice and Tracey models for growth of voids
derivative dD!dE begins to be positive. The damage is of and with experiments.
course supposed to be zero for E < En· The strain to rupturepR for any value of the triaxiality ratio
An example is shown in Fig. 3 for copper (99.9 percent) at has already been calculated to derive the model. Dividing PR
room temperature, the variation of damage elasticity modulus by ER (the strain to rupture in the one-dimensional case) yields

) ]
is shown in Fig. 3(a) and the evolution of damage, deduced (withs0 1). =

1 - E/E, is plotted against the true strain Ev 2 -1

a
from D =

log(l + E) in Fig. 3(b).

=
pR
__
ER
= [ 2
- (1+v)+3(1 - 211)
aeq
( _!!_
In the range of normal scatter D is linear in E as predicated
3
by the model. Let us now calculate the same ratio with the McClintock or
En is the value of E where the best fit straight line of ex­ Rice and Tracey models (1, 2]. These models predict the
perimental data cuts the E axis. growth of a void of radius R in a plastic matrix by
De is the value of D for the strain at rupture ER. aH
dR =
R
B exp C
aeq
( ) dp
4 Applications
where B and C 1.5 are coefficients determined by the
=

The two main properties of the model is the linearity of D theory.

with strain and the influence of triaxiality given by the factor Assume an initial value of the radius R0 which grows after a

[� (l+v)+3(1-2v ) ( :: r] threshold value of the equivalent strainpn is reached.

p<pn - R=Ro
The first set of applications will check the first property and Assume that failure occurs when a critical value of the radius
the second application will give some elements of the second. Re is reached (23]
D
D
D
Al alloy AU4G1
013 steel xc 38
Steel E 24
De 9)2_ __ __ _

.R.c�-- ---
0,1 0,1

q2 0,4 q6 0)8
�b� c���
D

Q-��8�- --- - -
D D

Q13 Steel 30CD4 qo

Jlo.e:�L _ -

q2 q4 �6 0,8
fb� ���JQ�
Fig. 4 Ductile damage evolutions (from J. Dufailly, D. Nouailhas, B.
Ghatoufi, B. Abdouli·l.M.T. Cachan France).

6
 R=Rc - p=pR
and integrate the differential equation for constant ratio
[ +(l+v)+3(1-2v)( :: r J
aHI aeq between these two limits. 5 Conclusion
Rc aH
Log --
Ro
= B (pR-PD exp C --
�q
) ( ) The integrated model of ductile plastic damage developed
on a thermodynamic and effective stress concept basis is
or linear in strain and shows a very strong effect of triaxiality as
do the McClintock and Rice and Tracey models. Its range of
Rc= R0 exp B pR-pD exp c� ([
aeq
] ( )) validity is limited by the hypothesis of isotropy of damage and
isotropy of plasticity, and also by the hypothesis of constant
Dividing by the same expression written for the one­ triaxiality ratio during loading, that is radial loading in the
dimensional case: sense of plasticity (approximately constant principal direc­
tions of stresses), which is a very common case in metal
Rc =R0exp B cR -ED exp ([ ] �) forming. In more general cases of loading, the differential
model written in terms of a continuum mechanics variable of
after some mathematical manipulations, we obtain: damage is easy to apply together with plasticity equations
coupled with damage in any type of structural step by step

PR PD
1- �
0R
calculations such as the finite element method to predict the
state of damage and ductile fracture.
= + __________
References
aH
(
exp c - _ _ -
aeq
�)
3 I McClintock, F., "A Criterion for Ductile Fracture by the Growth of
Holes," ASME Journal ofApplied Mechanics, June 1968.
2 Rice, J., and Tracey, D., "On Ductile Enlargement of Voids in Triaxial
The corresponding values of PRIER for C 1.5 are plotted = Stress Fields," Journal of Mechanics Physics of Solids, Vol. 17, 1969.
against the triaxiality ratio aHlaeq in Fig. 5. The two dashed 3 Kachanov, L. M., "Time of the Rupture Process Under Creep Con­

lines correspond to extreme values of the ratios PD/ER and

ditions," IVZAkad Nauk, S.S.R., Otd Tech Nauk, No. 8, 1958.
4 Lemaitre, J., "Evaluation of Dissipation and Damage in Metals Sub­
EDIER that experiments on metal may show. That is: mitted to Dynamic Loading," Proceedings I.C.M.1, Kyoto, Japan, 1971.

PD ED
5 Leckie, F., and Hayhurst, D., "Creep Rupture of Structures,"
Proceedings of R. Soc., London A, Vol. 240, 1974, p. 323.
= 0 (no threshold)
ER ER 6 Hult, J., "Creep in Continua and Structures," Topics -in Applied Con­
tinuum Mechanics, Springer-Verlag, Vienna, 1974.
7 Chaboche, J. L., "Une Joi differentielle d'endommagement de fatigue
PD ED
:::: = .2 avec cumulation non lineaire," Revue Francaise de Mechanique, No. 50-51,
ER ER 1974.

Values of P RI ER given by the model described in this paper are

8 Lemaitre, J., and Chaboche, J. L., "A Non Linear Model of Creep­

v .33.
also plotted on the figure for the two values of
=
.25 and v, v =
Fatigue Damage Cumulation and Interaction," Proceedings of l.U.T.A.M.
Symposium on Mechanics of Visco-elastic media and bodies, Gothenburg,
Sweden, Springer-Verlag, 1974.

P. 9 Chaboche, J. L., "Description Thermodynamique et phenomenologique

de la visco-plasticite cyclique avec endommagement," These O.N.E.R.A. et
I [.R UniversiteParis 6, 1978.
IO Lemaitre, J., and Chaboche, J. L., "AspectsPhenomenologiques de la
1,5 \
'1
1,
Rupture par Endommagement," Journal de Mecanique Appliquee, Vol. 2, No.

"'
\\
3, 1978.

'
11 Murakami, S., "Effects of Cavity Distribution in Constitutive Equations
\ of Creep and Creep Damage," EUROMECH Colloquium on Damage
Mechanics, Cachan France, 1981.
12 Cordebois, J. P, and Sidoroff, F., "Endommagement anisotrope en
elasticite et plasticite," Journal de Mecanique Thtiorique et Appliquee, No.
Special, 1982.
13 Krajcinovic, D., and Fonseka, G. U., "The Continuous Damage Theory

0,5
of Brittle Materials," ASME Journal ofApplied Mechanics, Vol. 48, 1981, pp.
809-824.
14 Lemaitre, J., and Baptiste, D., "On Damage Criteda," Workshop NSF
on "Mechanics of Damage and Fracture," Atlanta, U.S.A., 1982.
15 Rousselier, G., "Contribution a l'etude de la rupture des metaux dans le
Ott domaine de l'elasto-plasticite," These, EcolePolytechnique et UniversiteParis
_

o �---�----�2-----===3,__ __.,_a.q
6, 1979.
16 Needleman, A., and Triantafyllidis, N., "Void Growth and Local
Fig. 5 Influence of triaxiality on strain to rupture•• A508 steel Necking in Biaxially Stretched Sheets," Report of Brown University, No.
@@ H Steel 16421/1, 1977.

ZZ-£.fi
17 Twergaard, V., "Influence of Voids on Shear Band Instabilities Under
Domaine cover by McClintock-R.T. model
Plane Strain Conditions," Report of Technical University of Denmark, No.
.IIZ7I Domaine cover by present model. 159, 1979.
18 Germain,P., Cours de mechanique des milieux continus, Masson, 1973.
Except for very small values of the ratio aHlaeq • the domain 19 Broberg, H., "A New Criterion for Brittle Creep Rupture," ASME
Journal ofApplied Mechanics, Vol. 41, 1974.
of this model (between the two solid lines) covers the domain
20 Lemaitre, J., and Dufailly, J., "Modelisation et identification de l'en­
of the McClintock or Rice and Tracey models (between the dommagement plastique des metaux," 3e
me Congres Frarn;ais de Mecanique,
two dashed lines). This strong influence of triaxiality ratio is
also in accordance with results of reference and Also [16] [17]. Grenoble, France, 1977.
21 Lemaitre, J., Cordebois, J. P., and Dufailly, J., "Sur le couplage en­
dommagement-elasticite," Compte-rendu a l'Academie des Sciences, Paris,

results from reference [23]

plotted in Fig. 5 are some points representing experimental
obtained with notched specimens
of A508 steel at room temperature and from reference [24]
B.391, 1979.
22 Lemaitre, J., "Evaluation of Damage by Means of Elasticity Change
Measurements,'' To be published, 1984.
obtained with deep-drawing of sheets of "H" steel. 23 Mudry, F., "Elude de la rupture ductile et de la rupture par clivage
d'aciers faiblement allies," These de !'Ecole des Mines, Universite de
Considering the simplicity of the model and the scatter of
Technologie de Compiegne, 1982.
tests, correlation may be considered to be good, and provides 24 Grumbach, M., and Sanz, G., "Influence de quelques parametres sur les
a proof of the validity of the triaxiality factor: courbes limites d'emboutissage," Revue de Metallurgie, Avril 1972.