Contact Initialization for General Contact in Abaqus/Explicit

You can specify whether initial overclosures should be resolved without generating stresses, strains, or momentum or treated as interference fits that are gradually resolved over the first step.

In addition, contact initialization controls for general contact in Abaqus/Explicit:

  • can be used to separate entangled double-sided surfaces;
  • can be used to model an initial gap between surfaces;
  • can influence which regions are initially bonded for cohesive contact;
  • can be used to identify an initially bonded node set in crack propagation analyses;
  • can be used to specify nondefault search zones that determine which nodes are affected by the contact initialization controls; and
  • should not be specified to correct gross errors in the mesh design.

This page discusses:

Defining and Assigning Contact Initialization Methods

You can define contact initialization methods that provide instructions for initializing contact, and then assign these methods to surface pairings. The default contact initialization method is in effect wherever you have not explicitly assigned a contact initialization method. For example, you may want to increase the tolerance for resolving initial deep penetrations or specify that certain gaps should be adjusted to a "just touching" status in some interface regions. Furthermore, some analyses call for initial overclosures to be treated as interference fits rather than resolved with strain-free adjustments.

Categories of characteristics associated with contact initialization methods include:

Two interfaces associated with contact initialization are available in Abaqus/Explicit. The preferred interface allows more general types of initialization. Usages for the preferred interface are included in the sections referenced above; usages for the alternative interface are described in Alternative Contact Initialization User Interface. Neither interface supports the use of edge-based surfaces.

You assign a name to each contact initialization method. This name is used in the assignment of a contact initialization method to specific surface pairings. Surface names used in the assignment of contact initialization methods to surface pairings do not have to correspond to the surface names used to specify the general contact domain. In many cases nondefault contact initialization methods are assigned to a subset of the overall general contact domain. Any contact initialization assignments for regions that fall outside of the general contact domain are ignored. The last assignment takes precedence if the specified interactions overlap. Special considerations for contact initialization assignments in the context of shell surfaces are discussed in Contact Initialization Details Specific to Shells.

Controlling Initial Gap or Penetration State

The desired initial contact state often differs from what can be interpreted from the specified mesh geometry. For example, regions with initial mesh overclosures are often intended to have surfaces treated as just touching with no initial overlap. Other regions with small gaps between parts according to the initial mesh may be intended to be overclosed by a distance associated with an interference fit. This section describes the default behavior and alternatives.

Default Initialization of Initial Gap or Penetration State

Abaqus/Explicit automatically adjusts the positions of surfaces or optionally stores penetration offsets (as discussed in Resolving Discrepancies between Mesh Geometry and Desired Initial Clearance or Penetration State) to remove small initial overclosures. In addition, if the search zone for contact initialization is extended to the gap region (as discussed in Contact Initialization Search Criteria), Abaqus/Explicit removes small gaps existing in the general contact domain in the first step of a simulation. Position adjustments are made with strain-free initial displacements. Figure 1 shows an example of repositioning nodes to resolve initial overclosures. This automatic adjustment of the initial penetration or gap state is intended to correct only minor mismatches associated with mesh generation and is done even if the interaction is defined through user subroutine VUINTERACTION.

Configuration of contact surfaces after strain-free adjustments to resolve overclosure.

Specifying an Initial Clearance Distance

By default, adjustments to the initial penetration or gap state adjust initial nodal positions such that surfaces are “just touching” (with zero penetration/separation). Alternatively, you can specify a single non-negative value representing a uniform clearance or a nodal distribution to define a clearance per node (see Distribution Definition). If a distribution is defined and the clearance is omitted for a secondary node, the clearance value is interpolated from the values at the main nodes. The secondary node is ignored if clearance values are omitted from the distribution for the secondary node and all of the nodes of the nearest main face. The adjustments occur only for regions that satisfy the search zone tolerances, as discussed in Contact Initialization Search Criteria. Mesh distortion can occur if large strain-free adjustments occur to achieve the specified initial clearance distance. Figure 2 shows an example in which a uniform initial gap is desired between concentric surfaces.

Specifying a uniform initial gap between concentric circular surfaces.

Modeling Interference Fits

The general contact algorithm in Abaqus/Explicit can optionally treat initial overclosures as interference fits. In this case the interference distance usually corresponds to the original mesh overclosure distance although exceptions are discussed in Specifying an Interference Distance. Subsequent resolution of this interference is effectively a form of loading within the first step and typically generates stresses and strains.

The general contact algorithm uses a shrink-fit method to gradually resolve the interference distance over the first step of the analysis as shown in Figure 3. Stresses and strains are generated as the interference is resolved. Large velocities can develop if the interference fit is resolved too rapidly. You should follow these general recommendations:

  • Do not apply other loads while the interference fit is being resolved.
  • Resolve the interference fit over a large enough number of increments such that the kinetic energy remains small.

Gradual resolution of contact interference fit.

Specifying an Interference Distance

By default, when the interference fit method is specified, the interference distance is implied by the initial overclosure of the mesh; alternatively, you can specify a uniform interference distance. In this case Abaqus/Explicit first makes strain-free adjustments of nodal positions such that initial overclosure in the adjusted configuration corresponds to the specified interference distance and then invokes the shrink-fit method discussed above, as depicted in Figure 4. Mesh distortion can occur if large strain-free adjustments are necessary to achieve the specified interference distance.

Treatment of a specified interference distance that differs from the interference implied by the original mesh.

Resolving Interference Fit over a Fraction of a Step Duration

Optionally, you can specify a fraction of the step time over which the interference fit is resolved. For example, if the interference fit is resolved during the first three-quarters of the step, some damping of dynamic effects occurs without additional loading associated with resolving the interference over the last quarter of the step. While this fraction does not affect contact initialization, it is specified within the contact initialization definition for convenience.

Resolving Discrepancies between Mesh Geometry and Desired Initial Clearance or Penetration State

Controlling Initial Gap or Penetration State discusses how to specify an initial gap or penetration state that may differ from that implied by the specified mesh geometry. This section discusses options available in Abaqus/Explicit to resolve descrepancies between the prescribed state and the specified mesh geometry.

Achieving Desired Initial Clearance or Penetration State with Strain-Free Adjustments

By default, Abaqus/Explicit automatically adjusts positions of surfaces to resolve discrepancies between gap or overclosure of the original mesh geometry and the desired initial gap or penetration state among general contact interactions of the first step. These adjustments are made with strain-free initial displacements to secondary node positions. This automatic adjustment of nodal positions is intended to correct only minor mismatches associated with mesh generation.

Output variable STRAINFREE (see Abaqus/Explicit Output Variable Identifiers) can be used to review strain-free adjustments. This output variable contains nodal vectors representing initial strain-free adjustments. By default, STRAINFREE is written to the output database (.odb) file for the original field output frame at zero time if any strain-free adjustments are made by Abaqus/Explicit.

Limiting Strain-Free Adjustments to One Surface in a Contact Initialization Assignment

Most interactions in Abaqus/Explicit are treated with a balanced main-secondary approach, such that the surfaces of an interaction act as both secondary and main (see Contact Surface Weighting). In this case, nodal position adjustments occur for both surfaces of an interaction. Optionally, you can limit strain-free nodal position adjustments to one surface of a contact initialization assignment even in such cases.

Achieving a Desired Initial Zero or Positive Clearance State with Contact Offsets Instead of Strain-Free Adjustments

Discrepancies between gap or overclosure of the original mesh geometry and a desired initial clearance state specified can optionally be resolved with contact offsets instead of strain-free adjustments. This option is not applicable to contact initialization methods involving interference fit. Contact offsets effectively modify penetration calculations for individual secondary nodes, such that the contact force is computed as f = k ( d c u r - d o f f s e t ) ; where k is the penalty stiffness, d o f f s e t is the contact offset distance, and d c u r is the current penetration distance. The contact offset distance remains constant for a given secondary node if you deactivate strain-free adjustments.

The use of contact offsets instead of strain-free adjustments (or repositioning parts in a preprocessor to avoid overclosures) can lead to confusion and nonphysical behavior for the following reasons:

  • The contact force is proportional to the modified penetration dmod=dcur-doffset, and this modified penetration differs from the penetration inferred by visual inspection of the current configuration of the model.
  • The use of offsets, especially large offsets, tends to degrade robustness of the contact algorithms. You can request initial overclosure and crossed surface diagnostics to diagnose some of these problems (see Contact Diagnostics in an Abaqus/Explicit Analysis).

Other Cases Involving Contact Offsets

Contact offsets are used by default to treat discrepancies between the original mesh geometry and the desired initial contact state in the following cases:

  • Discrepancies remaining after strain-free adjustments, which may be due to conflicting adjustments from separate contact (such as when a secondary node that is part of a shell is pinched between two main facets), boundary conditions, tie constraints, coupling constraints, and rigid body constraints.
  • Discrepancies associated with contact interactions newly included in general contact after the first step if no contact was defined in the previous step.
  • Discrepancies associated with contact interactions for an Abaqus/Standard analysis imported into Abaqus/Explicit if the contact interaction is not defined with user subroutine VUINTERACTION.

In these cases, contact force is computed with f=k(dcur-doffset), as discussed in Achieving a Desired Initial Zero or Positive Clearance State with Contact Offsets Instead of Strain-Free Adjustments, but the contact offset distance, doffset, is reduced as surfaces separate. If dcur ever decreases below doffset, doffset is reset to dcur.

Cases with Undesired Initial Overclosures That Are Not Resolved

If the general contact domain is extended after the first step, Abaqus/Explicit takes no special actions, as part of initialization for the new step, to adjust nodal positions, store penetration offsets, or gradually resolve initial penetrations for the newly introduced interactions. Penalty contact forces are applied proportional to the current penetration, or the penetration may be ignored. Furthermore, contact initialization diagnostics are not available for these new interactions.

Contact Initialization Search Criteria

You can specify search distances to define search “zones” above and below the surfaces. Other aspects of the assigned contact initialization method are applied to secondary nodes that lie within these search zones. By default, the search zones exclude:

  • Initial overclosures into a solid surface greater than any contact thickness in effect plus the larger of approximately 15% of the dimension of facets attached to the secondary node and a specified interference fit distance.
  • Initial overclosures into a shell surface greater than the contact thickness in effect, except for situations discussed in Contact Initialization Details Specific to Shells for which a single-sided shell surface is used in the contact initialization assignment.
  • Initial gaps greater than the larger of a tiny fraction of the dimension of facets and a specified initial clearance distance.

For example, Figure 5 shows two solid elements with large initial penetrations that will not be detected during the default initial overclosure resolution procedure. Potential interactions associated with the excluded initial overclosures are not eligible for contact until validation criteria are met for newly established contact (such as if the two parts separated and then came back into contact during the simulation).

Undetected large penetrations of solid elements.

You can increase the overclosure threshold of the initialization search criteria by explicitly specifying a "search below" distance. If you specify a search below distance smaller than the default overclosure tolerance and assign this contact initialization method to interactions between surf1 and surf2, Abaqus/Explicit uses the default tolerance instead. If the default desired initial clearance of zero is in effect, increasing the search below distance to 0.2 causes the initial overclosures to be resolved as shown in Figure 6.

Resolution of large penetrations of solid elements.

Adjusting the nodal coordinates can do the following:

  • Degrade the mesh geometry by creating imperfections that were not initially present.
  • Reduce the element size and correspondingly the stable time increment size.
  • Cause elements to invert and prevent the analysis from continuing.
In such cases it may be preferable to bypass the nodal coordinate adjustments and specify the storage of a contact offset (as discussed in Resolving Discrepancies between Mesh Geometry and Desired Initial Clearance or Penetration State) or to modify part positioning in the preprocessor to avoid overclosures.

You can close small initial gaps between surfaces by specifying a "search above" distance.

Increasing the extent of the search zones for strain-free adjustments can potentially increase the computational cost of an analysis. It is not generally recommended that you specify a large search zone because this may cause mesh distortion when nodes are repositioned over large distances.

Using a Node Set as the Search Criterion or to Identify Initially Bonded Nodes

As an alternative to specifying search distances, you can specify a search node set containing the secondary nodes for which nondefault contact initialization instructions are meant to apply. However, the nondefault contact initialization instructions apply only for nodes of this set within a fixed distance from the main surface. The fixed distance is the maximum element size for solid elements or the thickness for structural elements (for example, shell elements) associated with the nodes.

You can also specify a search node set to identify which secondary nodes will be tagged as initially bonded in a VCCT crack propagation analysis. See Crack Propagation Analysis for more details.

Contact Initialization Details Specific to Shells

Contact initialization involving parts modeled with shells may require extra attention due to the lack of a unique outward direction. For example, Figure 7 shows two shells separated by some distance in the initial mesh geometry, along with two possible adjustments to initially touching (zero gap) configurations. The case shown on the right preserves Surface 1 above Surface 2 before and after the nodal position adjustments, whereas the case shown on the left has Surface 2 below Surface 1 after nodal position adjustments. These cases differ in the following initialization settings to achieve the respective adjustments:

  • Left side: "Search below" tolerance set to at least as large as the overclosure distance shown on the top left of Figure 7. Single-sided shell surfaces (created with SPOS or SNEG side identifiers explicitly specified) corresponding to the top of Surface 1 and the bottom of Surface 2 are used in the contact initialization assignment. Single-sided surfaces used in contact initialization assignments must have consistent surface normal orientations for adjacent faces.
  • Right side: "Search above" tolerance set to at least as large as the gap distance between the shells in the initial mesh configuration, as discussed in Contact Initialization Search Criteria. Double-sided shell surfaces (created without side identifiers specified) can be used for Surface 1 and Surface 2 in the contact initialization assignment because the initialization algorithm by default preserves Surface 1 remaining on the top side of Surface 2. Alternatively, using single-sided surfaces corresponding to the bottom of Surface 1 and the top of Surface 2 also leads to the adjusted configuration shown on the bottom right of Figure 7.
Both of these cases call for retaining default settings of the desired initial gap distance (zero) and the method (nodal position adjustments) used to achieve the desired initial gap distance.

Two interpretations of adjusting nearby shells to "just touching."

Figure 8 shows a side view of a case with secondary nodes on different sides of a shell main reference surface. Four nodes of the original mesh geometry are initially within the region spanned by the shell thickness. The figure also shows the default corrected positions of secondary nodes to achieve zero initial clearance at secondary node locations while keeping each secondary node on the same side of the reference surface as in the original mesh geometry.

Default resolution of initial overclosure for tangled surfaces.

Secondary nodes that are trapped on opposite sides of a double-sided main surface often lead to serious problems involving nonphysical behavior during a simulation. A secondary surface initially crossing the main shell reference surface may or may not involve some initially overclosed nodes in the original mesh geometry, so the STRAINFREE output variable may or may not bring attention to these regions. Warning messages are provided for cases with adjacent secondary nodes trapped on opposite sides of a double-sided main surface. Similar to the discussion of Figure 7, the use of single-sided surfaces (created with SPOS or SNEG side identifiers) in contact initialization assignments provides a mechanism to detangle surfaces that are crossed in the initial mesh geometry. For example, the use of single-sided surfaces in the contact initialization assignment along with an increase in the "search below" distance leads to all secondary nodes adjusted to the top side of the main surface, as shown in Figure 9.

Improved resolution of initial overclosure for tangled surfaces.

Alternative Contact Initialization User Interface

The alternative contact initialization user interface applies to a subset of functionality of the preferred contact initialization user interface and focuses on specifying contact clearances. The alternative contact initialization user interface cannot be used to specify contact interference. With the alternative method, you specify contact clearance methods and assign contact clearance methods to interactions. Only single-sided surfaces can be used in the assignment of contact clearance methods, whereas single- or double-sided surfaces can be used for assignments with the preferred user interface.

Specifying an Initial Clearance Distance (Alternative Interface)

In general, the concepts of specifying an initial clearance distance with the alternative interface are the same as with the preferred interface (see Specifying an Initial Clearance Distance); however, the syntax for specifying the initial clearance distance is different.

Achieving Desired Initial Clearance with Strain-Free Adjustments (Alternative Interface)

In general, the concepts of achieving an initial clearance distance with strain-free adjustments using the alternative interface are the same as with the preferred interface (see Achieving Desired Initial Clearance or Penetration State with Strain-Free Adjustments); however, the syntax for specifying this method is different.

Contact Initialization Search Criteria (Alternative Interface)

In general, the concepts of modifying contact initialization search criteria using the alternative interface are the same as with the preferred interface (see Contact Initialization Search Criteria); however, the syntax for specifying this method is different.

Using a Node Set as the Search Criterion or to Identify Initially Bonded Nodes (Alternative Interface)

In general, the concepts of using a node set as the search criteria or to identify initially bonded nodes using the alternative interface are the same as with the preferred interface (see Using a Node Set as the Search Criterion or to Identify Initially Bonded Nodes); however, the syntax for specifying this method is different.