Modeling Discontinuities as an Enriched Feature Using the Extended Finite Element
Method
Modeling discontinuities, such as cracks, as an enriched feature:
is commonly referred to as the extended finite element method
(XFEM);
is an extension of the conventional finite element method based on the concept of
partition of unity;
allows the presence of discontinuities in an element by enriching degrees of freedom with
special displacement functions;
does not require the mesh to match the geometry of the discontinuities;
is a very attractive and effective way to simulate initiation and propagation of a
discrete crack along an arbitrary, solution-dependent path without the requirement of
remeshing in the bulk materials;
can be based on traction-separation cohesive behavior or based on the principles of linear
elastic fracture mechanics (LEFM);
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;
can also be used to perform contour integral evaluations for an arbitrary stationary
surface crack without the need to define the conforming mesh around the crack tip;
allows the application of distributed pressure loads or distributed heat fluxes to the
cracked element surfaces;
allows the output of some surface variables on the cracked element surfaces;
allows both material and geometrical nonlinearity; and
is available only for first-order stress/displacement solid continuum elements,
first-order displacement/pore pressure solid continuum elements, first-order
displacement/temperature solid continuum elements, first-order displacement/pore
pressure/temperature solid continuum elements, and second-order stress/displacement
tetrahedral elements.
Modeling stationary discontinuities, such as a crack, with the conventional finite element
method requires that the mesh conforms to the geometric discontinuities. Creating a
conforming mesh can be quite difficult. Modeling a growing crack is even more cumbersome
because the mesh must be updated continuously to match the geometry of the discontinuity as
the crack progresses.
The extended finite element method (XFEM) alleviates the
need to create a conforming mesh. The extended finite element method was first introduced by
Belytschko and Black (1999). It is an extension of the conventional finite element
method based on the concept of partition of unity by Melenk and Babuska (1996), which allows local enrichment functions to be easily
incorporated into a finite element approximation. The presence of discontinuities is ensured
by the special enriched functions in conjunction with additional degrees of freedom.
However, the finite element framework and its properties such as sparsity and symmetry are
retained. XFEM does not alleviate the need for sufficient
mesh refinement in the vicinity of the crack tip.
Introducing Nodal Enrichment Functions
For the purpose of fracture analysis, the enrichment functions typically consist of the
near-tip asymptotic functions that capture the singularity around the crack tip and a
discontinuous function that represents the jump in displacement across the crack surfaces.
The approximation for a displacement vector function with the partition of unity enrichment is
where are the usual nodal shape functions; the first term on the right-hand
side of the above equation, , is the usual nodal displacement vector associated with the continuous
part of the finite element solution; the second term is the product of the nodal enriched
degree of freedom vector, , and the associated discontinuous jump function across the crack surfaces; and the third term is the product of the
nodal enriched degree of freedom vector, , and the associated elastic asymptotic crack-tip functions, . For more details, see Nodal enrichment functions.
Accurately modeling the crack-tip singularity requires constantly keeping track of where
the crack propagates and is cumbersome because the degree of crack singularity depends on
the location of the crack in a nonisotropic material (see Nodal enrichment functions). Therefore, we consider the asymptotic singularity functions only when modeling
stationary cracks in Abaqus/Standard. Moving cracks are modeled using one of the two alternative approaches described below.
Modeling Moving Cracks with the Cohesive Segments Method and Phantom Nodes
One approach for modeling moving cracks within the framework of
XFEM is based on traction-separation cohesive behavior.
This approach is used in Abaqus/Standard to simulate both crack initiation and propagation. This is a very general interaction
modeling capability, which can be used for modeling brittle or ductile fracture. Other
non-XFEM–based approaches for modeling crack initiation
and propagation capabilities available in Abaqus/Standard are based either on cohesive elements (Defining the Constitutive Response of Cohesive Elements Using a Traction-Separation Description) or on
surface-based cohesive behavior (Contact Cohesive Behavior). Unlike these
non-XFEM–based methods, which require that the cohesive
surfaces align with element boundaries and the cracks propagate along a set of predefined
paths, the XFEM-based cohesive segments method can be
used to simulate both crack initiation and propagation along an arbitrary,
solution-dependent path in the bulk materials, since the crack propagation is not tied to
the element boundaries in a mesh. In this case the near-tip asymptotic singularity is not
needed, and only the displacement jump across a cracked element is considered. Therefore,
the crack has to propagate across an entire element at a time to avoid the need to model
the stress singularity.
The cohesive segments approach is based on the use of so-called phantom nodes, which are
superposed on the original real nodes, and are introduced to represent the discontinuity
of the cracked elements (see Figure 1). When the element is intact, each phantom node is completely constrained to its
corresponding real node. When the element is cut through by a crack, the cracked element
splits into two parts. Each part is formed by a combination of some real and phantom nodes
depending on the orientation of the crack. Each phantom node and its corresponding real
node are no longer tied together and can move apart. For more details, see Phantom node method.See Applying Cohesive Material Concepts to XFEM-Based Cohesive Behavior for more
information.
Modeling Hydraulically Driven Fracture
The cohesive segments method in conjunction with phantom nodes discussed above can also
be extended to model hydraulically driven fracture. In this case additional phantom
nodes with pore pressure degrees of freedom are introduced on the edges of each enriched
element to model the fluid flow within the cracked element surfaces in conjunction with
the phantom nodes that are superposed on the original real nodes to represent the
discontinuities of displacement and fluid pressure in a cracked element. The phantom
node at each element edge is not activated until the edge is intersected by a crack. The
flow patterns of the pore fluid in the cracked elements are shown in Figure 2. The fluid is assumed to be incompressible. The fluid flow continuity, which accounts
for both tangential and normal flow within and across the cracked element surfaces as
well as the rate of opening of the cracked element surfaces, is maintained. The fluid
pressure on the cracked element surfaces contributes to the traction-separation behavior
of the cohesive segments in the enriched elements, which enables the modeling of
hydraulically driven fracture.
Optionally, phantom nodes with pore pressure, temperature, and slurry concentration
degrees of freedom can be introduced on the edges of each enriched element, and phantom
nodes with displacement, pore pressure, and temperature degrees of freedom can be
superposed on the original real nodes. The temperature degrees of freedom are necessary
to model heat transfer due to thermal conduction and radiation across the cracked
element surfaces, as well as thermal convection within the cracked element surfaces. The
slurry concentration degrees of freedom are necessary to model proppant transport and
placement within the cracked surfaces. The temperature and/or slurry degrees of freedom
will be activated automatically if thermal and/or slurry material properties are
specified for the fluid flow within the cracked element surfaces.
Modeling Moving Cracks Based on the Principles of Linear Elastic Fracture Mechanics
(LEFM) and Phantom Nodes
An alternative approach to modeling moving cracks within the framework of
XFEM is based on the principles of linear elastic
fracture mechanics (LEFM). Therefore, it is more
appropriate for problems in which brittle crack propagation occurs either on reaching a
critical value of a fracture criterion or associated with fatigue crack growth; for more
discussion of fracture criteria and fatigue crack growth using the Paris law, see Fatigue Crack Growth Criterion, Linear Elastic Fatigue Crack Growth Analysis, and Low-Cycle Fatigue Analysis Using the Direct Cyclic Approach. The
XFEM-based LEFM approach
can be used to simulate crack propagation along an arbitrary, solution-dependent path in
the bulk material.
Similar to the XFEM-based cohesive segments method
described above, the near-tip asymptotic singularity is not considered, and only the
displacement jump across a cracked element is considered. Therefore, the crack has to
propagate across an entire element at a time to avoid the need to model the stress
singularity. The strain energy release rate at the crack tip is calculated based on the
modified Virtual Crack Closure Technique (VCCT), which
has been used to model delamination along a known and partially bonded surface (see Crack Propagation Analysis).
Another similarity is the introduction of phantom nodes to represent the discontinuity of
the cracked element when the fracture criterion is satisfied (for more details. see Phantom node method). The
real node and the corresponding phantom node will separate when the equivalent strain
energy release rate exceeds the critical strain energy release rate at the crack tip in an
enriched element. The traction is initially carried as equal and opposite forces on the
two surfaces of the cracked element. The traction is ramped down linearly over the
separation between the two surfaces with the dissipated strain energy equal to either the
critical strain energy required to initiate the separation or the critical strain energy
required to propagate the crack depending on whether the
VCCT or the enhanced
VCCT criterion is specified.
Using the Level Set Method to Describe Discontinuous Geometry
A key development that facilitates treatment of cracks in an extended finite element
analysis is the description of crack geometry, because the mesh is not required to conform
to the crack geometry. The level set method, which is a powerful numerical technique for
analyzing and computing interface motion, fits naturally with the extended finite element
method and makes it possible to model arbitrary crack growth without remeshing. The crack
geometry is defined by two almost-orthogonal signed distance functions, and , as illustrated in Figure 3. For more details, see Level set method.
Defining an Enriched Feature and Its Properties
An enriched feature is a region of a model, specified with an element set, in which the
finite element interpolation functions are enhanced to include discontinuities. In other
words, an enriched feature is a region in the model where you can use the extended finite
element method (XFEM) to model a crack. You must specify an enriched feature and its
properties to use XFEM. Because there is a computational cost associated with modeling
discontinuities, limiting the extent of an enriched feature can enhance computational
performance.
An enriched feature can model a stationary crack or a propagating crack, but not both. You
must decide during model setup whether a particular enriched feature models a stationary
crack or a propagating crack. For a stationary crack, the enrichment functions include the
asymptotic elastic crack-tip fields (see Introducing Nodal Enrichment Functions). If you model a propagating crack, you must ensure that the enriched feature is large
enough to include all areas of the part where the crack could potentially propagate. You
should not include areas where the crack is unlikely to propagate in the enriched feature
(if this information is known ahead of time). Including these areas increases the
computational cost without having any impact on the accuracy of the model. Additional
details regarding the criteria for initiation and propagation of a crack are discussed in
Crack Initiation and Direction of Crack Extension.
The default XFEM implementation in Abaqus does not support interactions among multiple cracks or branching of a single crack. It is
best suited for problems where only a single crack is modeled within an enriched feature.
Real applications often require modeling multiple cracks within a single part. A part can
contain no preexisting cracks or one or more preexisting cracks; it can also contain one or
more cracks that initiate during the analysis. For such applications, you can use multiple
enriched features within a part (one for each crack) or use a single enriched feature to
model more than one crack, although the latter approach has limitations.
The choice of one approach versus the other depends on the ease of specifying multiple
enrichment features at the time of model setup. This, in turn, depends on the proximity of
the cracks to one another at initiation and during growth. You can choose a single enriched
feature in situations where you do not have any prior knowledge regarding the potential
locations of multiple cracks or when it is not straightforward to separate out two or more
distinct enriched features in a complex part. The following paragraphs provide some
guidelines to help you judge what types of situations you can model and to select the
correct modeling approach for your problem, starting with the important concept of a
level set conflict.
As discussed in Using the Level Set Method to Describe Discontinuous Geometry, an XFEM crack is represented in terms of nodal values of a level set function. A
requirement of the level set representation is that the level set function at any given node
must have a unique value. This requirement is violated when more than one crack cuts through
a single element or adjacent elements of a given node, resulting in an attempt to assign
multiple values of the level set function at the node. Such a condition is referred to as a
level set conflict. If a level set conflict occurs during an
analysis, the analysis ends with an error. Without any kind of prior experimental data,
paths of propagating cracks are often difficult to predict. This can cause uncertainty prior
to a simulation as to whether a level set conflict will occur if you model multiple cracks
in a single enriched feature.
If there are multiple preexisting cracks in a part, you can define them within a single
enriched feature as long as these cracks are relatively far away from one another to begin
with and do not come close to one another during the crack growth process. This restriction
ensures that a level set conflict does not occur at any node. Mesh refinement can help avoid
a level set conflict.
By default, crack initiation exhibits the following behavior to avoid level set conflicts:
Crack initiation in an enriched feature cannot occur until all existing cracks propagate
through the enriched feature boundary. As illustrated in Figure 4, Crack 2 does not initiate until Crack 1 propagates through the enriched
feature boundary.
Crack initiation cannot occur near a preexisting crack (because this immediately causes
a level set conflict).
More than one crack can initiate at nonpredetermined locations within an enriched
feature only if two or more nonadjacent elements satisfy the damage initiation criterion
in the same time increment and one of the following conditions is satisfied:
There are no preexisting cracks in the enriched feature.
All preexisting cracks in the enriched feature have propagated through the feature
boundary.
A level set conflict occurs if the elements in which cracks initiate share nodes.
A newly initiated second crack cannot approach or enter an already cracked element.
You can use one of the following methods to change the default behavior described above:
Use multiple enriched features.
Use a single enriched feature that is capable of initiating more than one crack at
different locations and at different time increments before existing cracks propagate
through the feature boundary.
The sections that follow outline these two approaches.
Multiple Enriched Features
To use multiple enriched features, you must define the element sets belonging to each
feature at the time of model setup. In other words, you must be able to determine the
boundaries among multiple enriched features at the time of model setup. Figure 5 shows a part with two holes that result in stress
concentrations. For the loading shown, two cracks would initiate and grow near these
holes. For this part, it is easy to define multiple enriched features during model setup.
If both cracks in Figure 6 are preexisting cracks and do not come close to each other as
they grow, you can also use a single enriched feature. However, with a single enriched
feature, performance suffers if most of the enriched feature remains uncracked. From a
performance point of view, the better choice might still be to use two separate enriched
features in the areas where the cracks grow.
Single Enriched Feature Capable of Initiating Multiple Cracks before Existing Cracks
Propagate through the Feature Boundary
You might not be able to clearly define two or more enriched features at the time of
model setup in all situations. Figure 6, which shows a plate with two closely spaced holes, illustrates
an example of this situation. The potential crack initiation and growth paths are
difficult to anticipate ahead of time. It might take some experience and modeling
iterations to properly choose the boundaries that separate multiple enriched features.
Using a single enriched feature might be more appropriate for this situation. XFEM cannot
model cases in which cracks should coalesce during a simulation.
Another example where a clear separation of enriched features is difficult (maybe
impossible) involves a composite panel with a hole and with +45/−45 plies. In this panel
many parallel matrix cracks could nucleate and propagate in each ply, as illustrated in
Figure 7. A single enriched feature per ply is more appropriate for this
problem.
For the examples shown in Figure 6 and Figure 7, it is recommended that you use a single enriched feature (per ply), but
activate the capability to initiate more than one crack at different locations and at
different time increments even before preexisting cracks propagate through the feature
boundary. This nondefault approach has fewer limitations compared to the default approach
discussed earlier. During the model setup phase in this nondefault approach, you
explicitly specify that multiple cracks can initiate within a single enriched feature. You
also specify a small relative radius, , to define a crack initiation suppression
zone near a crack tip within which the initiation of another crack is not allowed (to
prevent a level set conflict). Multiple cracks are allowed to initiate in an enriched
feature, but only outside the crack initiation suppression zone for each individual crack
tip (see Activating and Deactivating the Enriched Feature).
However, a level set conflict could still occur if two cracks propagate close to each
other during the crack growth. If this happens, further mesh refinement in the region
where the cracks approach each other might help run the analysis to completion.
Specifying the Enrichment
Abaqus activates the enriched degrees of freedom only when a crack cuts through an element.
You can associate only stress/displacement, displacement/pore pressure,
displacement/temperature, or displacement/pore pressure/temperature solid continuum
elements with an enriched feature.
Defining the Type of Enrichment
You can choose to model an arbitrary stationary crack or a discrete crack propagation
along an arbitrary, solution-dependent path. The former requires that the elements around
the crack tips are enriched with asymptotic functions to catch the singularity and that
the elements intersected by the crack interior are enriched with the jump function across
the crack surfaces. The latter infers that crack propagation is modeled with either the
cohesive segments method or the linear elastic fracture mechanics approach in conjunction
with phantom nodes. However, the options are mutually exclusive and cannot be specified
simultaneously in a model.
Assigning a Name to the Enriched Feature
You must assign a name to an enriched feature, such as a crack. This name can be used in
defining the initial location of the crack surfaces, in identifying a crack for contour
integral output, in activating or deactivating the crack propagation analysis, and in
generating cracked element surfaces.
Identifying an Enriched Region
You must associate the enrichment definition with a region of your model. Only degrees of
freedom in elements within these regions are potentially enriched with special functions.
The region should consist of elements that are presently intersected by cracks and those
that are likely to be intersected by cracks as the cracks propagate.
Defining a Crack Surface
As a crack propagates through the model, a crack surface representing both facets of
cracked elements is generated on those enriched elements that are intersected by a crack
during the analysis. You must associate the name of an enriched feature with the surface
(see Assigning a Name to the Enriched Feature above).
Defining Contact Interaction of Enriched Cracked Element Surfaces
Contact interaction of cracked element surfaces, including thermal interaction and pore
fluid interaction, is allowed based on a small-sliding formulation or on a finite-sliding
formulation within the general contact framework.
Specifying the Initial Location of an Enriched Feature
Because the mesh is not required to conform to the geometric discontinuities, the initial
location of a preexisting crack must be specified in the model. The level set method is
provided for this purpose. Two signed distance functions per node are generally required to
describe a crack geometry. The first describes the crack surface, while the second is used
to construct an orthogonal surface so that the intersection of the two surfaces gives the
crack front (see Initial Conditions).
The first signed distance function must be either greater or less than zero and cannot be
equal to zero. If an initial crack has to be defined at the boundaries of an element, a very
small positive or negative value for the first signed distance function must be specified.
Activating and Deactivating the Enriched Feature
You can activate or deactivate the crack propagation capability within the step definition.
You can specify a small relative radius, , around the crack tip (as shown in Figure 8) to define a crack initiation suppression zone within which the elements in
the enriched feature are excluded from crack nucleation. However, multiple crack nucleations
are allowed to occur when the damage initiation criterion is satisfied in the elements lying
outside the crack initiation suppression zone in the enriched feature. The default radius
value is five times the typical element characteristic length in the enriched region.
Contour Integral
When you evaluate the contour integrals using the conventional finite element method (Contour Integral Evaluation), you must define the crack front explicitly and specify
the virtual crack extension direction in addition to matching the mesh to the cracked
geometry. Detailed focused meshes are generally required and obtaining accurate contour
integral results for a crack in a three-dimensional curved surface can be cumbersome. The
extended finite element in conjunction with the level set method alleviates these
shortcomings. The adequate singular asymptotic fields and the discontinuity are ensured by
the special enrichment functions in conjunction with additional degrees of freedom. In
addition, the crack front and the virtual crack extension direction are determined
automatically by the level set signed distance functions.
Specifying the Enrichment Radius
Although XFEM has alleviated the shortcomings associated
with defining the conforming mesh in the neighborhood of the crack front, you must still
generate a sufficient number of elements around the crack front to obtain path-independent
contours, particularly in a region with high crack-front curvature. The group of elements
within a small radius from the crack front are enriched and become involved in the contour
integral calculations. The default enrichment radius is six times the typical element
characteristic length of those elements along the crack front in the enriched area. You
must include the elements inside the enrichment radius in the element set used to define
the enriched region. Elements with a large aspect ratio should be avoided along the crack
front.
Procedures
Modeling discontinuities as an enriched feature can be performed using any of the
following:
Initial conditions to identify initial boundaries or interfaces of an enriched feature can
be specified (see Initial Conditions).
Boundary Conditions
Boundary conditions can be applied to any of the displacement, temperature, or pore
pressure degrees of freedom (see Boundary Conditions).
Loads
The following types of loading can be prescribed in a model with an enriched feature:
Concentrated nodal forces can be applied to the displacement degrees of freedom (1–3)
or the pore pressure degree of freedom (8); see Concentrated Loads.
Distributed pressure forces or body forces can be applied; see Distributed Loads. The
distributed load types available with particular elements are described in Abaqus Elements Guide.
Concentrated heat fluxes.
Body fluxes and distributed surface fluxes.
Node-based film and radiation conditions.
Average-temperature radiation conditions.
Element- and surface-based film and radiation conditions.
For more information on heat fluxes, film conditions, and radiation conditions, see Thermal Loads.
Predefined Fields
The following predefined fields can be specified in a model with an enriched feature, as
described in Predefined Fields:
Nodal temperatures (although temperature is not a degree of freedom in
stress/displacement elements). The specified temperature affects temperature-dependent
critical stress and strain failure criteria.
The values of user-defined field variables. The specified value affects
field-variable-dependent material properties.
Material Options
Any of the mechanical constitutive models in Abaqus/Standard, including user-defined materials (defined using user subroutine UMAT) can be used to
model the mechanical behavior of the enriched element in a crack propagation analysis. See
Abaqus Materials Guide. The inelastic definition at a
material point must be used in conjunction with the linear elastic material model (Linear Elastic Behavior), the porous
elastic material model (Elastic Behavior of Porous Materials), or the
hypoelastic material model (Hypoelastic Behavior). Only isotropic
elastic materials are supported when evaluating the contour integral for a stationary crack.
Elements
Only the following elements can be associated with an enriched feature:
For an incompatible mode element, Abaqus/Standard discards the contribution due to the incompatible deformation mode immediately after the
element is fractured under a tensile loading. Therefore, the stress level at the cracked
element may not return completely to its originally unloaded state even when this cracked
element is unloaded completely and the contact of the cracked element surfaces is
reestablished.
Output
In addition to the standard output identifiers available in Abaqus (Abaqus/Standard Output Variable Identifiers), the following
nodal, whole element, and surface variables have special meaning for a model with an
enriched feature.
Nodal variables:
PHILSM
Signed distance function to describe the crack surface.
PSILSM
Signed distance function to describe the initial crack front.
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.)
LOADSXFEM
Distributed pressure loads applied to the crack surface.
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.
Use of Unsymmetric Matrix Storage and Solution
When the pore pressure degrees of freedom are activated in the enriched elements,
matrices are unsymmetric; therefore, unsymmetric matrix storage and solution may be needed
to improve convergence (see Matrix Storage and Solution Scheme in Abaqus/Standard).
Visualization
A crack can be visualized through the iso-surface for the signed distance function
PHILSM.
If a crack cuts through a very tiny corner of an enriched element, the displacements
along the crack front in the enriched element may be distorted in rare cases in the Visualization module of Abaqus/CAE (Abaqus/Viewer) when displaying the contours. The distortion, however, is not present when viewing
only the deformed shape.
When an element is cut through by a crack, the cracked element splits into two parts,
each part formed by a real domain and a phantom domain, as illustrated in Figure 1. Contour plot integration point values for cracked elements consider contributions from
only the real domains in both parts of the cracked elements. However, when you probe
cracked elements, only the contribution from the part of the elements containing the real
domain and the phantom domain is reported.
When evaluating the contour integrals in a stationary crack, additional integration
stations are introduced internally in the elements enriched with singular asymptotic
crack-tip fields. However, visualization of the element output variables in those
additional integration points is not supported in the Visualization module of Abaqus/CAE (Abaqus/Viewer).
Limitations
The following limitations exist with an enriched feature:
An enriched element cannot be intersected by more than one crack.
A crack is not allowed to turn more than 90° in one increment during an analysis.
Only asymptotic crack-tip fields in an isotropic elastic material are considered for a
stationary crack.
Adaptive remeshing is not supported.
Composite solid elements are not supported.
Import analysis is not supported.
Input File Template
The following is an example of modeling crack propagation with the
XFEM-based cohesive segments method:
Belytschko, T., and T. Black, “Elastic Crack Growth in Finite
Elements with Minimal Remeshing,” International
Journal for Numerical Methods in Engineering, vol. 45, pp. 601–620, 1999.
Melenk, J., and I. Babuska, “The Partition of Unity Finite
Element Method: Basic Theory and Applications,” Computer Methods in Applied
Mechanics and Engineering, vol. 39, pp. 289–314, 1996.