Damage Evolution Law
Damage Evolution Law
Damage Evolution Law
Related terms:
FAILURES OF MATERIALS
JEAN LEMAITRE, RODRIGUE DESMORAT, in Handbook of Materials Behavior
Models, 2001
(13)
where is the plastic multiplier and S and s are two material- and temperature-depen-
dent parameters. The normality rule also defines the evolution of the state variable
r associated with R as
(14)
(15)
(15.5)
The proposed evolution law for the cohesive model can also be expressed in terms
of the rate of change of Kc and then discretized in time steps, as follows:
(15.6)
The superscripts i and i + 1 stand for loading steps i and i + 1, respectively. As shown
in the above-mentioned relations, during the reloading phase, the cohesive stiffness
at each material point along the cohesive zone gradually decreases in proportion to
the increment in displacement of the crack opening. This proportionality factor
evolves with the number of cycles N and thus gives a measure of the total damage
accumulated during the degradation process.
> Read full chapter
[126]
where the damage evolution law dI(w) is defined in terms of normal relative displace-
ments. For a bilinear constitutive law, the expression for dI(w) is given by
[127]
The strain energy release rate associated with mode I delamination is defined by the
area underneath the stress–relative displacement, which in turn is defined by the
bilinear constitutive law. That is,
[128]
where wf is the relative displacement in which the interfacial stress in mode I is equal
to zero (complete decohesion). From eqn [128], wf can be written in terms of the
strain energy release rate as follows:
[129]
For a linear-polynomial constitutive law, the expression for damage evolution law
dI(w) is written as
[130]
with
[131]
The derivation of the damage law for a linear-polynomial constitutive law enforces
the areas under the stress–relative displacement defined for both bilinear and linear
polynomials to be the same.
[132]
(8.11)
where is the identity state defined in (3.1). A useful example of a damage evolution
law is given by the following model:
(8.12)
where f (e, ) is a nondecreasing function of e, , and the maximum is taken over all
times up to t. Assume for simplicity that for a given motion, is a nondecreasing
function of time. Observing from (8.11) that
(8.13)
differentiating (8.12) with respect to time yields an equivalent damage evolution law
in terms of the rate:
(8.14)
(8.15)
(8.16)
provided is nondecreasing. A specific case of such a damage model is bond breakage
in tension, in which
(8.17)
where H is the Heaviside step function and is the prescribed bond breakage extension
for the bond . In this case,
Alternatively, the same damage evolution law can be defined in terms of the rate
through (8.16) with
MULTIPHYSICS BEHAVIORS
PIERRE LADEVÈZE, in Handbook of Materials Behavior Models, 2001
10.6.3.2.1 Remarks
• Two damage variables are used to describe the damage associated with matrix
microcracking and fiber-matrix debonding. They seem to account for all the
proposed damage kinematics, including that starting from an analysis of the
microcracks. Many works have established, experimentally or theoretically, a
relation between the microcrack density and our damage variable d, which can
be very useful for the identification of a damage fatigue model.
• What we call the single layer is the assemblage of adjacent, usual elementary,
plies of the same direction. The damage forces, being integral values through
the thickness of the single layer, can be interpreted as energy release rates. It
follows that the damage evolution law of the single layer is thickness-depen-
dent. For single layers which are not too thick, such damage evolution laws
include results coming from shear lag analyses. Consequently, the size effects
– observed, for example, in tension – are produced by both the single-layer
model and the interface model through a structure problem. This theory,
which is very simple, works very well for most engineering laminates; however,
it cannot be satisfactory for rather thick layers. A first solution is to modify the
damage evolution law, using the thickness as a parameter.
• The damage variables are active for [0°, 90°]n laminates even if the apparent
modulus does not change. The model predicts this hidden damage [15].
• For fatigue loadings, we introduce:(6)where ds and d's are the quasi-static
part of the damage defined by Eq. 4 or 5. dF and d'F denote the fatigue part
characterized by the following fatigue evolution laws:(7)where a, a' are two
material functions and [.] denotes the maximum value over the cycle.
2.7 Conclusion
In this chapter, constitutive models of polycarbonate were presented. A dam-
age-based elastic–viscoplastic constitutive model was developed for polycarbonate
within the framework of irreversible thermodynamics and CDM. It was able to char-
acterize the mechanical behavior and damage evolution of the material over a wide
range of strain rates and temperatures. The constitutive equations, specifically the
viscoplastic flow rule and damage evolution law, were derived from the Helmholtz
free energy and the damage-coupled plastic flow potential. To realize the numerical
implementation of the proposed model, a coupled elastic-damage predictor and
viscoplastic corrector algorithmic scheme was introduced and further simplified
into a single-scalar Newton–Raphson scheme. The model was implemented into
LS-DYNA by using the UMAT subroutine. Finally, numerical examples, including
both the single-element and the cylinder specimen, were given. The good correlation
between the numerical predictions and the experimental data has demonstrated
the capabilities of the proposed model in capturing the mechanical behavior of
polycarbonate over a wide range of strain rates and temperatures.
(6.5)
where is a principal stretch measure within the RVE (Upadhyaya et al., 2011), and 0
and m are material constants that are assumed to be dependent on environmental
conditions but independent of the applied strain rate. These material constants are
actually evaluated in this chapter by performing fracture experiments using DCB
specimens in conjunction with DIC, as described in Section 6.5.
All layer-wise calculations presented in this paper are based on a modified classical
laminate theory with the structural law written as
(6.15)
By setting the coupling matrix to zero, the specimen geometry and the unsym-
metrical layup are considered (Koch, 2010). The differences in calculated stress
distribution in the inner and outer layer for tube specimens in tension loading
comparing three different mathematical approaches is displayed as an example in
Figure 6.9. The stress distribution in the undisturbed area is taken to be comparably
within the context of this paper, and the fastest method (CLT mod.) is used.
The engineering constants E1, E2, G12, 12 and strength values R|| , R|| , R , R
and R|| of the undamaged i-UD layer (Table 6.2) have been derived from the inverse
laminate theory and quasi-static tests until damage initiation and total failure (Böhm
et al., 2011). The remaining values are adapted from typical values from literature
and comparable in-house material data.
Table 6.2. Engineering constants and strength values, i-UD layer GF-MLG/EP
The identification of the remaining parameters for damage growth ci, coupling qi
and tension–compression asymmetry hi are derived iteratively from the elaborated
stiffness degradation curves of the cross-ply-reinforced tube specimens (type S) only.
In a first step the stiffness degradation curve for loading path 1 (tension–tension
fatigue) and here the characteristic constant stiffness drop is focussed. In the frame-
work of the pursued top-down approach, this stage of stiffness degradation can be
dedicated to damage in the 0° layer only. Here damage due to inter-fibre failure in
the 90° layer reached saturation already. Hence the damage evolution law Eqn (6.5)
can be idealised to the damage growth in fibre direction
(6.16)
Hence the parameters of the growth function || can be elaborated by analysing
the constant stiffness drop in pulsating tension loading. It is proposed to analyse
the constant stiffness drop of several tension fatigue experiments on different load
levels. By fitting the chosen damage growth function to the experimentally derived
dependence of damage increment and material effort for all results, the parameters
are elaborated more precisely.
For the parameter identification of the first stage of stiffness degradation in pulsat-
ing tension fatigue is focused. Following the top-down approach in this stage, two
failure modes are active and the damage evolution law can be rewritten to
(6.17)
Hence with the given damage growth function || the first-stage stiffness degra-
dation can be used for the identification of the parameters of . The final failure of
the material under pulsating tension loading is driven by damage localisation and
instable crack growth. The beginning of damage localisation is modelled with the
help of a critical damage value Dc1 where the calculation is stopped.
(6.18)
Here the first and second stage of stiffness degradation of pulsating torsion tests is
used. The results of often recommended ±45° specimens under pulsating tension
loading cannot be used because of parallel-acting damage modes such as transverse
tension and compression. In Table 6.3 the identified parameters for the damage
growth of GF-MLG/EP are displayed.
Ds 0.4
C3 12
Dc2 0.4
C5 18.46
0.11
Dc1
To capture the stiffness evolution, it seems reasonable to follow the physically found-
ed and very simple idea of Zinoviev [101], who suggested that the effective transverse
and shear stresses in the damaged ply do not exceed the level at which cracks
initiates. The model of Zinoviev proved to be one of the five most successful models
of the World-Wide Failure Exercise (WWFE) I [102]. Another successful degradation
model of WWFE, the model of Puck [103], employs a similar post-cracking assump-
tion. It requires reducing the elastic constants in a way that keeps the failure initiation
index at a constant level. Upon active loading, both the models exhibit a stress–strain
response resembling the curve of an ideal plasticity. Both the damage models do
not require any material constants other than those needed for the failure initiation
prediction.
where E and G are Young's and shear moduli, and d2, d12 degradation of the elastic
constants, index ‘0’ corresponds to intact material, coordinate ‘1’ denotes the fibre
direction, 1–3 is the plane of the crack defined by means of the Mohr-Hashin-Puck
failure initiation criteria [105]. Without contradicting available experimental data, it
is postulated that all the cracks in a segment have the same orientation [106]. All the
other elastic constants remain unaffected by cracking. The symmetry of the stress
tensor in the damaged media leads, as described by Zako et al. [46], to the following
relation between degradation of the Young's and shear moduli:
(2.7)
The damage factors d2 and d12 are nearly proportional at small values, and the
transverse degradation is slightly higher at the advanced damage levels. Following
the continuum damage models [107], we employ the energy release rate Y12 as a
factor driving damage accumulation. It is introduced as the derivative of the elastic
energy of the damaged material with respect to the degradation parameter:
(2.8)
where operator denotes the maximum over the time period t and reflects the fact
that the damage state is non-healing. The square root is used for convenience, to
make it proportional to strain. The elastic energy may be written as a function of
the damage parameters d2, d12, the elastic stiffness matrix of intact material C0,
and the strains averaged over the segment volume . This allows differentiating the
expression explicitly, which results in the following definition:
(2.9)
In the case of a closed crack (transverse compression), the normal stress traction
is assumed to be continuous across the crack face. Hence, the degradation factor
d2 = 0 and . Zinoviev's and Puck's assumptions are in good correspondence with the
following damage evolution law:
(2.10)
where indexi reflects the value of a variable at the moment of failure onset and .
Equation (2.10) matches the shear diagram assumed in [101] – Figure 2.8(a), as it is
reduced to in pure shear. The stress–strain responses in other loading situations are
obtained from the solution of Eqns (2.8) and (2.10) relative to d12. Figure 2.8(b) shows
predicted response in uniaxial transverse tension. Right after the damage onset, the
stress drops and before settling down at a value of about 80–90% of transverse
strength.
Figure 2.8. Stress–strain diagrams of an elementary unidirectional fibre bundle in
(a) pure shear (input), (b) pure transverse tension
Such a law has an analogy with classical fracture mechanics, where energy released
through crack propagation is linked to the crack length. The meaning of Eqn (2.10)
is that the stiffness reduction is proportional to the energy, which can be released
through damage accumulation.