Applying Cohesive Material Concepts to XFEM-Based Cohesive
Behavior
Modeling discontinuities, such as cracks, as an enriched feature:
can be based on traction-separation cohesive behavior;
can be used in Abaqus/Standard to simulate both crack initiation and propagation;
is a very general interaction modeling capability;
can be used for modeling brittle or ductile fracture; and
can be simultaneously used with the surface-based cohesive behavior approach (see Contact Cohesive Behavior) or the Virtual
Crack Closure Technique (see Crack Propagation Analysis), which are
best suited for modeling interfacial delamination.
The formulas and laws that govern the behavior of XFEM-based
cohesive segments for a crack propagation analysis are very similar to those used for cohesive
elements with traction-separation constitutive behavior (Defining the Constitutive Response of Cohesive Elements Using a Traction-Separation Description) and those used for
surface-based cohesive behavior (Contact Cohesive Behavior). The similarities
extend to the linear elastic traction-separation model, damage initiation criteria, and damage
evolution laws.
Linear Elastic Traction-Separation Behavior
The available traction-separation model in Abaqus assumes initially linear elastic behavior followed by the initiation and evolution of
damage. The elastic behavior is written in terms of an elastic constitutive matrix that
relates the normal and shear stresses to the normal and shear separations of a cracked
element.
The nominal traction stress vector, , consists of the following components: , , and (in three-dimensional problems) , which represent the normal and the two shear tractions, respectively. The
corresponding separations are denoted by , , and . The elastic behavior can then be written as
The normal and tangential stiffness components will not be coupled: pure normal separation
by itself does not give rise to cohesive forces in the shear directions, and pure shear slip
with zero normal separation does not give rise to any cohesive forces in the normal
direction.
The terms , , and are calculated based on the elastic properties for an enriched element.
Specifying the elastic properties of the material in an enriched region is sufficient to
define both the elastic stiffness and the traction-separation behavior. For simplicity, we
assume that .
Damage Modeling
Damage modeling allows you to simulate the degradation and eventual failure of an enriched
element. The failure mechanism consists of two ingredients: a damage initiation criterion
and a damage evolution law. The initial response is assumed to be linear as discussed in the
previous section. However, once a damage initiation criterion is met, damage can occur
according to a user-defined damage evolution law. Figure 1 shows a typical linear and a typical nonlinear traction-separation response with a
failure mechanism. The enriched elements do not undergo damage under pure compression.
Damage of the traction-separation response for cohesive behavior in an enriched element is
defined within the same general framework used for conventional materials (see About Progressive Damage and Failure). However, unlike
cohesive elements with traction-separation behavior, you do not have to specify the
undamaged traction-separation behavior in an enriched element.
Crack Initiation and Direction of Crack Extension
Crack initiation refers to the beginning of degradation of the cohesive response at an
enriched element. The process of degradation begins when the stresses or the strains satisfy
specified crack initiation criteria. Crack initiation criteria are available based on the
following Abaqus/Standard built-in models:
the maximum principal stress criterion,
the maximum principal strain criterion,
the maximum nominal stress criterion,
the maximum nominal strain criterion,
the quadratic traction-interaction criterion,
the quadratic separation-interaction criterion, and
the three-dimensional LaRC05 criterion.
In addition, a user-defined damage initiation criterion can be specified in user subroutine
UDMGINI.
An additional crack is introduced or the crack length of an existing crack is extended
after an equilibrium increment when the fracture criterion, f, reaches
the value 1.0 within a given tolerance:
You can specify the tolerance . If , the time increment is cut back such that the crack initiation criterion
is satisfied. The default value of is 0.05. To improve performance, a separate tolerance can be specified to control the crack growth of an existing crack while is used to control the nucleation of an additional crack. If it is not
specified, the growth tolerance, is set equal to
Fracture of Multiple Elements in an
Unstable Crack Growth Analysis
For an unstable crack growth problem, sometimes it
is more efficient to allow multiple elements at and ahead of a crack tip to fracture without
excessively cutting back the increment size when the fracture criterion is satisfied. Abaqus/Standard activates this capability automatically if you specify an unstable growth tolerance, . In this case if the fracture criterion, f, is within
the given unstable growth tolerance:
where is the tolerance described earlier in this section, Abaqus/Standard immediately reduces the time increment size by default to a very small value, Reducing the time increment size allows more elements to fracture until for all the elements ahead of the crack tip. You can, however, optionally
specify the maximum number of cutbacks allowed, , to be controlled by the regular tolerance, , prior to the activation of the unstable growth tolerance in an increment.
After this limit the time increment size is recovered automatically to a larger value, , where:
Minimum time increment allowed
Time increment size prior to the unstable crack growth
(default 0.5), (default 2.0), and (default 0)
Scaling parameters
If you do not specify a value for the unstable growth tolerance, the default
value is infinity. In this case the fracture criterion, f, for unstable
crack growth is not limited by any upper bound value in the above equation.
Specifying the Crack Direction
When the maximum principal stress or the maximum principal strain criterion is specified,
the newly introduced crack is always orthogonal to the maximum principal stress/strain
direction when the fracture criterion is satisfied. However, when one of the other Abaqus/Standard built-in crack initiation criteria is used, you have to specify if the newly introduced
crack will be orthogonal to the element local 1-direction or orthogonal to the element
local 2-direction (see Conventions) when the
fracture criterion is satisfied. By default, the crack is orthogonal to the element local
1-direction. If a user-defined damage initiation criterion is specified, the normal
direction to the crack plane or the crack line can be defined in user subroutine UDMGINI.
Maximum Principal Stress Criterion
The maximum principal stress criterion can be represented as
Here, represents the maximum allowable principal stress. The symbol represents the Macaulay bracket with the usual interpretation (that is, if and if ). The Macaulay brackets are used to signify that a purely compressive
stress state does not initiate damage. Damage is assumed to initiate when the maximum
principal stress ratio (as defined in the expression above) reaches a value of one.
Maximum Principal Strain Criterion
The maximum principal strain criterion can be represented as
Here, represents the maximum allowable principal strain, and the Macaulay
brackets signify that a purely compressive strain does not initiate damage. Damage is
assumed to initiate when the maximum principal strain ratio (as defined in the expression
above) reaches a value of one.
Maximum Nominal Stress Criterion
The maximum nominal stress criterion can be represented as
The nominal traction stress vector, , consists of three components (two in two-dimensional problems). is the component normal to the likely cracked surface, and and are the two shear components on the likely cracked surface. Depending on
what you specify (see Specifying the Crack Direction above), the
likely cracked surface will be orthogonal either to the element local 1-direction or to
the element local 2-direction. Here, , , and represent the peak values of the nominal stress. The symbol represents the Macaulay bracket with the usual interpretation. The
Macaulay brackets are used to signify that a purely compressive stress state does not
initiate damage. Damage is assumed to initiate when the maximum nominal stress ratio (as
defined in the expression above) reaches a value of one.
Maximum Nominal Strain Criterion
The maximum nominal strain criterion can be represented as
Damage is assumed to initiate when the maximum nominal strain ratio (as defined in the
expression above) reaches a value of one.
Quadratic Nominal Stress Criterion
The quadratic nominal stress criterion can be represented as
Damage is assumed to initiate when the quadratic interaction function involving the
stress ratios (as defined in the expression above) reaches a value of one.
Quadratic Nominal Strain Criterion
The quadratic nominal strain criterion can be represented as
Damage is assumed to initiate when the quadratic interaction function involving the
nominal strain ratios (as defined in the expression above) reaches a value of one.
Larc05 Three-Dimensional Criterion
The LaRC05 three-dimensional criterion can be applied
generally to polymer-matrix fiber-reinforced composites. This criterion considers four
different damage initiation mechanisms: matrix cracking, fiber kinking, fiber splitting,
and fiber tension. For detailed information on the damage initiation criterion, see LaRC05 Criterion.
The initiation criterion that first reaches a value of 1.0 determines the damage
initiation.
User-Defined Damage Initiation Criterion
User subroutine UDMGINI provides a general
capability for implementing a user-defined damage initiation criterion.
You can define several damage initiation mechanisms in user subroutine UDMGINI. You represent each damage
initiation mechanism by a fracture criterion, , and its associated normal direction to the crack plane or the crack
line. Although you can define several damage initiation mechanisms, the actual damage
initiation for an enriched element is governed by the most severe damage initiation
mechanism:
Damage is assumed to initiate when f, as defined in the expression above, reaches a value
of one.
You must specify any material constants that are needed in user subroutine UDMGINI as part of a user-defined
damage initiation criterion definition.
Limiting the Crack Propagation Direction
When the maximum principal stress, maximum principal strain, or user-defined damage
initiation criterion is specified, you can limit the new crack propagation direction to
within a certain angle (in degrees) of the previous crack propagation direction. The
default is 85°.
Local Calculations of the Stress and Strain Fields Ahead of the Crack Tip
An accurate and efficient evaluation of the stress/strain fields ahead of the crack tip is
important for both evaluating the crack initiation criterion and computing the crack
propagation direction when needed. Abaqus/Standard offers several options for computing these fields.
Centroidal Values of Stress and Strain
By default, the stress/strain computed at the element centroid ahead of the crack tip is
used to determine if the damage initiation criterion is satisfied and to determine the
crack propagation direction. See Figure 2.
Computing the Stress and Strain Fields at the Crack Tip
With a sufficiently refined mesh, the centroidal approximation is accurate and
economical. However, if the finite element mesh in the vicinity of the crack tip is coarse
relative to the gradients in the stress/strain fields, the default centroidal
approximation may not be sufficient. In such cases you can use the stress/strain
extrapolated to the crack tip to determine if the damage initiation criterion is satisfied
and to determine the crack propagation direction. See Figure 2.
Combining Crack Tip and Centroidal Calculations
You can also choose to combine the two previous alternatives: you can use the
stress/strain values extrapolated to the crack tip to determine if the damage initiation
criterion is satisfied, and you can use the stress/strain values at the element centroid
to determine the crack propagation direction.
Nonlocal Averaging of the Stress/Strain Fields and Smoothing of the Crack Surface Normals
to Improve the Accuracy of Crack Propagation Directions
The three options for evaluating the stress and strain fields discussed above are local
calculations in the sense that the evaluated fields are local to the single element ahead of
the crack tip. In the case of coarse and/or unstructured meshes a nonlocal averaging of the
stress and strain fields ahead of the crack tip can lead to a more accurate evaluation of
those fields, which can improve the accuracy of the computed propagation directions. In
addition, a moving least-squares approximation by polynomials is used by default to obtain
more accurate crack propagation directions. The least-squares approximation further smooths
out the normals of the individual crack facets in elements along the crack front that
satisfy the damage initiation criterion.
Specifying the Region of the Model Used for Nonlocal Averaging and Smoothing
To control the range of elements used for nonlocal averaging and smoothing in the crack
direction calculations, you can specify a radius, , within which the elements ahead of the crack tip are included (see
Figure 3).
The default radius is three times the typical element characteristic length in the
enriched region.
Smoothing the Stress/Strain Fields before Averaging
To further improve the nonlocal averaging, you can request an initial smoothing of the
stress/strain fields ahead of the crack. In this case Abaqus/Standard averages the field values to element nodes and then interpolates the smoothed fields to
the integration points. Once smoothing is complete, the nonlocal averaging is applied. No
smoothing is applied by default.
Weighting Schemes for Nonlocal Averaging
Abaqus/Standard offers a number of weighting schemes for field smoothing that provide additional
control over nonlocal averaging. For example, you may want to give a higher weighting to
elements close to the crack tip. You can specify a weight function, , to compute the average stress/strain based on the distance from the
element integration points to the crack tip, . By default, a uniform weighting is applied to all elements used for
averaging; alternatively, you can use a Gaussian function or a cubic spline function. You
can also define a weight function with a user subroutine.
The Gaussian function is represented by:
The cubic spline function is represented by:
Smoothing the Normals of Individual Crack Facets Using Least-Squares
Approximation
After the predicted crack propagation direction is obtained based on the nonlocal
stress/strain averaging, a moving least-squares approximation by polynomials is used by
default to further smooth out the crack normals. The least-squares approximation is
applied to the normals of the individual facets in elements along the crack front that
satisfy the damage initiation criterion, as highlighted in Figure 4. This
approximation provides a smoother crack surface (as shown in Figure 5), leading to
a more accurate crack propagation direction.
You can use linear, quadratic, or cubic polynomial approximation for the moving
least-squares approximation to smooth out the crack normals. You specify the number of
terms in the polynomial. You can also suppress the least-squares approximation. In this
case, the predicted crack propagation direction is determined based only on the nonlocal
stress/strain averaging.
Limiting the Elements Involved in Crack Normal Smoothing
At the beginning of the analysis, you can choose to include or exclude the preexisting
crack facets in elements from the moving least-squares approximation to obtain the crack
propagation direction. During the analysis, you can also limit the elements involved in
the least-squares approximation. You can set the maximum allowed difference (in degrees)
below which the normals of the crack facets are included in the moving least-squares
approximation. The default is 70°.
Damage Evolution
The damage evolution law describes the rate at which the cohesive stiffness is degraded
once the corresponding initiation criterion is reached. The general framework for describing
the evolution of damage is conceptually similar to that used for damage evolution in
surface-based cohesive behavior (Contact Cohesive Behavior).
A scalar damage variable, D, represents the averaged overall damage at
the intersection between the crack surfaces and the edges of cracked elements. It initially
has a value of 0. If damage evolution is modeled, D monotonically
evolves from 0 to 1 upon further loading after the initiation of damage. The normal and
shear stress components are affected by the damage according to
where , , and are the normal and shear stress components predicted by the elastic
traction-separation behavior for the current separations without damage.
To describe the evolution of damage under a combination of normal and shear separations
across the interface, an effective separation is defined as
Use with User-Defined Damage Initiation Criterion
A separate damage evolution law should be specified for each damage initiation criterion
defined in user subroutine UDMGINI. Each combination of a
damage initiation criterion and a corresponding damage evolution law is referred to as a
failure mechanism. Damage will accumulate for only one failure mechanism per element,
corresponding to the mechanism whose damage initiation criterion was achieved first.
Use with LaRC05 Criterion
You can specify four separate damage evolution laws, one for each of the four initiation
mechanisms. Alternatively, you can specify fewer than four damage evolution laws. In this
case, the initiation mechanisms that do not have a corresponding evolution law use the
specified damage evolution law with the smallest failure index. Damage accumulates for
only one failure mechanism per element, corresponding to the mechanism whose damage
initiation criterion was achieved first.
Viscous Regularization in Abaqus/Standard
Models exhibiting various forms of softening behavior and stiffness degradation often lead
to severe convergence difficulties in Abaqus/Standard. Viscous regularization of the constitutive equations defining cohesive behavior in an
enriched element can be used to overcome some of these convergence difficulties. Viscous
regularization damping causes the tangent stiffness matrix to be positive definite for
sufficiently small time increments.
The approximate amount of energy associated with viscous regularization over the whole
model is available using output variable
ALLVD.
Output
Whole element variables:
STATUSXFEM
Status of the enriched element. (The status of an enriched element is 1.0 if the
element is completely cracked and 0.0 if the element contains no crack. If the element
is partially cracked, the value of
STATUSXFEM lies between 1.0 and
0.0.)
Surface variables (available only for propagating cracks modeled with first-order solid
continuum elements):
CRKDISP
Crack opening and relative tangential motions on cracked surfaces in enriched
elements.
CSDMG
Damage variable on cracked surfaces in enriched elements.
CRKSTRESS
Remaining residual pressure and tangential shear stresses on cracked surfaces in
enriched elements.