This section describes some of the practices you should try to keep in mind while defining the properties of the contact conditions for your model.
Defining a proper mesh is critical to contact conditions. A well-defined mesh ensures accurate stress measurements at a contact region. Furthermore, a quality mesh is essential for nonlinear contact conditions in order to obtain an accurate solution. This is especially true for curved surfaces. Use local Mesh Controls, such as Proximity Controls and Contact Sizing controls to better ensure mesh quality. Review the Apply Mesh Controls and Preview Mesh section of the Help for more information on this topic.
Mechanical provides the following options for the Formulation property:
MPC (Multi-Point Constraint)
Formulation methods work in combination with the specified contact Types (Bonded, No Separation, Frictionless, Rough, Frictional, and Forced Frictional Sliding). The Augmented Lagrange method is the default Formulation property for all contact types.
However, you can use the Bonded and No Separation contact types with the Multi-Point Constraint (MPC) Formulation method. The examples listed below outline cases when this option is useful. Please see the Selecting a Contact Algorithm (KEYOPT(2)) section of the Mechanical APDL Contact Technology Guide for additional technical information about choosing contact formulations.
Workbench Mechanical considers the Bonded and No Separation contact types to be “linear contact.”
Generally, this means that if no other nonlinearities exist (plasticity, large deformation, or frictionless contact) a nonlinear solution is not required in order to obtain an accurate solution.
If a Formulation is not MPC-based, Mechanical constructs the input file to enforce a single iteration solution by issuing the NEQIT,1,FORCE command (in rare conditions this can result in an inaccurate solution, such as when a contact region is touching a constraint or a rigid body that has both a contact region and a remote displacement). In order to avoid this, you can use the MPC Formulation on the contact pairs to enable a truly linear solution or you can modify the boundary conditions to avoid contact overlap.
In a nonlinear analysis when convergence difficulties occur from Bonded/No-Separation contact situations, switching to MPC can be an attractive alternative compared to modifying the contact stiffness. A common example is where there is significant initial penetration. This is fine for a linear solution run but the presence of non-linear features can cause convergence issues. You can view NR residuals to help determine the proximity of convergence troubles.
During a Modal analysis, MPC can be employed to avoid spurious non-zero modes when gaps exist between curved surfaces. It is an inherent limitation of penalty based contact that is avoided by using an MPC based formulation.
Shell/Solid contact: When bonding shell edges to a solid, you need to make sure that the connection will properly constrain the two sides. The default (penalty-based) Formulation is not able to constrain rotational degrees of freedom that would create the possibility of a rigid body mode in cases such as a straight shell edge connected to a solid face. You can overcome this by using an MPC formulation that does provide options to constrain/couple the translation and rotation degrees of freedom.
To avoid contact conditions that overlap constraints, use the Bonded or No Separation contact types because you will see an overall correct solution, however, the reported reactions will be inaccurate.
This same phenomenon occurs in a less obvious way when you attempt to apply a Remote Displacement to a rigid body that also has bonded contact using a penalty based formulation.
The example illustrated below shows a remote constraint applied to a rigid body that is also has No Separation contact using a penalty formulation. In this example, the solution is correct, however, inaccurate reactions are obtained on the Remote Displacements because it is connected to the contact region via the MPC equations created. Using a remote displacement causes the solver to reorder the CE’s such that constrained node shares a CE with the bonded contact. This results in inaccurate reactions.
Using a General Joint instead of a remote displacement avoids the issue.
Regardless of the MPC formulation selection, MPC-based contact is used for Remote Boundary conditions. It is good practice to avoid having two or more MPC-based boundary conditions overlap. The solver does however attempt to negotiate and resolve the overconstraint conditions. The application issues a warning in this situation.
Intelligent use of Contact Trimming as well as the Pinball setting on remote boundary conditions can also be effective tools to mitigate this behavior.
In addition, MPC as well as other FE connections can be viewed via the Solution Information feature to help you graphically view the distribution of MPC equations in a model. These equations are generated from the MAPDL contact elements. See the Using Finite Element Access to Resolve Overconstraint tutorial for an example of an overconstraint situation along with steps to identify and correct it.
Properly choosing your source and target topology is also important. See the specific guidelines outlined in the MAPDL contact documentation. The default behavior is auto-asymmetric wherein the MAPDL solver determines the optimal source/target. Using a pure asymmetric behavior is suggested only for users willing to closely review each contact pair and able to determine the proper configuration.
Tip: Using the Initial Contact tool can help you determine which side the MAPDL solver chooses to keep in the analysis.
The Initial Contact Tool can be invaluable in determining that the contact is properly defined. It is also useful to determine the proper side for the source/target. Further the Initial Contact Tool can be useful to:
Make sure that the option Type property when contact conditions are touching and that all Rough/Frictional/Frictionless contact pairs that should be closed are, in fact, closed.or are selected for the
For nonlinear contact, check the amount of penetration (if any).
Even if nonlinear contact regions are in contact, make sure that more than one or two contact points are in contact, because if only one contact point is in contact, the condition may be unstable.
You can use NR residuals and result trackers to help obtain a fully converged analysis. For example:
Requesting three to four Newton-Raphson residuals under the Solution Information object before starting the solution allows you to graphically view the NR residuals so as to get a qualitative measure/indication for where convergence difficulties exist in the model.
Using Contact Result Trackers provides information during the solution, such as contact penetration, the number of elements in contact, contact stiffness values, as well as many other quantities. You can use these outputs to monitor the robustness of the solution and observe the trends occurring during a nonlinear incremental solution.
If there are a few nonlinear contact regions present and you are expecting the possibility of losing contact, you can also use the Results Tracker to add the number of contacting points for those contact regions.
If no convergence is achieved, check the NR residuals. If high residuals are present at contact regions, consider using aggressive automatic contact stiffness update or reducing contact stiffness by an order of magnitude.
While solving, if bisections occur (i.e., trouble converging), check Results Tracker to see if the number of contact points is decreasing (i.e., possible loss of contact).
Following the solution process, it is strongly recommend that you insert a Contact Tool to check penetration. Penetration units are the same as that of displacement - compared with displacements in local area.
For example, if local displacements are 2mm but penetration is 0.02mm, would a change in displacements by +/- 0.02mm influence overall results (including local stresses)? By comparing penetration to the results in local area (not maximum deformations of entire model), you can determine if penetration values are acceptable or not.
Caution: Do not assume that penetration values are always negligible because your solution converged. You need to verify this after the solution.
If you believe that penetration is excessive, modify the Penetration Tolerance (Augmented Lagrange), Normal Stiffness (Penalty or Augmented Lagrange), or use the Pure Lagrange formulation to reduce the penetration.