When a degree of freedom is subjected to multiple constraints, overconstraint occurs. Overconstraint can be caused by the combination of the following constraints within the model:
Overconstraint may result in convergence difficulties or inaccurate solutions. In order to prevent this problem, the program tries to detect and eliminate overconstraints automatically. You can verify the eliminated constraints during postprocessing.
At the beginning of a contact analysis, warning messages may appear outlining the cause of the potential overconstraints. Issue CNCHECK,DETAIL to identify the overconstraints indicated by the messages. Manually remove as many unnecessary constraints as possible.
Some overconstraints occur only during the solution due to new deformed configurations or due to changes in contact status. The program eliminates a limited set of overconstraints detected during solution. Appropriate messages are issued when this occurs. For MPC-based contact pairs, the contact statuses (contact result item STAT) are set to negative values if one or more contact constraints are removed. STAT is set to -3 for bonded contact. STAT is set to -2 for no-separation contact. Be aware that, due to complexities, certain overconstraints may not be easily detected.
The way the program removes the overconstraint is not unique. The order of the redundant constraints will influence how they are removed (the first encountered is kept, the next is removed). You should always verify the modified model carefully. You can list and display the contact status via the STAT contact-result item (PLESOL and PRESOL). For example:
PLESOL,CONT,STAT
Example 10.1: Effect of Order on Overconstraint Detection
In this example, the order of constraint equations has a significant effect on the results. All the surface-based constraints are defined as force-distributed constraints (see (a) in the figure below), and the center pilot node is defined as the last node (node 21618 of pair ID 25 shown in (b) below). The contact nodes of contact pair 25 are the pilot nodes of the other surface-based constraints. A boundary condition is applied to the center pilot node.
Due to the nature of force-distributed constraints, this configuration is over-constrained. And because the last-defined constraint ties the other constraints together, the overconstraint elimination logic is not able to find a good resolution. The program removes the boundary condition on the center pilot node, which results in no deformation at all (see (c) below).
When the center pilot node is not defined as the last node (node 21609 of pair ID 21 shown in figures (d) and (e) below), the overconstraint elimination logic can successfully resolve the overconstraint issue and produce correct deformation results as shown by (f).
The relaxation method is another tool used to eliminate overconstraint. When MPC-based surfaced-based constraints or rigid bodies are subjected to overconstraint, this method relaxes the constraint between contact-generated internal constraint equations and other constraint equations or Lagrange multipliers.
To activate the relaxation method, set KEYOPT(11) = 1 on the target element (TARGE169 or TARGE170). You can also specify translational and rotational relaxation coefficients (similar to penalty stiffness for the contact penalty method) through contact element real constants FKN and FKT, respectively, and corresponding tolerance values through real constants FTOLN and TNOP.
Example 10.2: Using the Relaxation Method to Eliminate Overconstraint
In the previous example, the optimum order of constraints is not intuitive. It is not obvious that the center pilot node must be defined first before defining the other surface-based constraints.
The relaxation method provides an alternative solution that enables you to use a normal workflow when defining the constraints. You can then activate relaxation at the center pilot node to eliminate overconstraint and achieve valid results, as shown by this deformation plot:
Overconstraint most likely occurs when a rigid beam/line or joint element (MPC184) connects to a rigid surface constraint or a rigid body. To eliminate potential overconstraints, you can issue the CNCHECK,OVER command in /PREP7 or right before the first solve command.
By default, theCNCHECK,OVER will perform:
When any rigid beam or weld joint (MPC184) connects to a rigid surface constraint (MPC formulation only: KEYOPT(2)=2 of contact element type) or rigid body, the rigid beam or weld joint will be converted automatically to a contact/target element as a part of the rigid constraint definition. The conversion is performed only if both density and thermal expansion coefficient are not defined for the rigid beam.
If you set Val1
=1 as in
CNCHECK,OVER,,,,,1 , the operation will perform:
When any rigid beam or weld joint (MPC184) connects to a rigid surface constraint (MPC formulation only: KEYOPT(2) = 2 of contact element type) or rigid body, the rigid beam or weld joint will be converted automatically to a contact/target element as a part of the rigid constraint definition. The conversion is performed only if both density and thermal expansion coefficient are not defined for the rigid beam.
When any Lagrange multiplier based joint (MPC184) connects to a surface rigid constraint or a rigid body, the penalty-based option (KEYOPT(2) = 1 of MPC184) will be set automatically.
When a rigid link (MPC184) connects to a surface rigid constraint or a rigid body, the relaxation option (KEYOPT(11) = 1 of the target element type) will be set automatically for the rigid constraint definition.