7.1.2 Contact iterations

Product: ABAQUS/Standard  

References

Overview

Contact iterations can be used instead of regular severe discontinuity iterations to speed up computations. Contact iterations:

  • are effective for the solution of large, geometrically linear, small-sliding, frictionless static problems with many severe discontinuity iterations;

  • avoid global stiffness matrix assemblage and factorization during displacement correction solution; and

  • preclude the use of certain features in the model.

Using contact iterations

Regular severe discontinuity iterations involve costly assemblage and factorization of the global stiffness matrix. When the number of contact elements changing status from open to closed and vice versa is small, it may be more effective to perform contact iterations. During such an iteration the exact displacement correction solution is obtained by performing forward and back substitution on several global right-hand sides. The number of right-hand sides to solve is proportional to the number of contact status changes. The already factorized global stiffness matrix from the last regular iteration is used. The structural stiffness and right-hand-side contributions are constant during contact iterations.

Contact iterations are usually effective only for relatively large problems. During a contact iteration, the global stiffness matrix assemblage and factorization are skipped, and the computational expense is dominated by the right-hand-side solutions. Such solutions involve heavy disk access; therefore, their speed can be machine dependent. Contact iteration is effective if several right-hand-side solutions without global matrix factorization can be obtained faster than regular solution that includes matrix factorization and one right-hand-side solution. For each global system matrix there is an upper limit on the number of right-hand sides to solve for the contact iteration to be effective. ABAQUS/Standard automatically chooses a reasonable limit, but you can specify a correction factor to override this default. You can also specify n_max, the maximum number of contact iterations allowed before new global matrix assemblage and factorization. The default value is 30.

ABAQUS/Standard stores the right-hand-side solutions obtained during contact iterations. Some of these solutions may be used during the next contact iterations. The number of contact status changes from the reference state as of the last factorization can, thus, be larger than the number of right-hand-side solutions allowed per contact iteration. Each contact iteration involves a solution of an internal linear system of equations, and the size of this system is the number of contact status changes.

In the status file the severe discontinuity iterations column is marked with an asterisk if successful contact iterations are present during the time increment. The whole sequence of successful contact iterations is counted as a single severe discontinuity iteration.

Input File Usage:           
*SOLUTION TECHNIQUE, TYPE=CONTACT ITERATIONS
correction_factor, n_max

ABAQUS/CAE Usage: 

Step module:

Create Step: General: any static step type except coupled temperature-displacement: Other: Apply contact iteration solution technique: Adjustment factor for the number of solutions in any iteration: correction_factor, Maximum number of contact iterations: n_max

Create Step: General: Coupled temp-displacement: Other: Solution technique: Contact Iterations, Adjustment factor for the number of solutions in any iteration: correction_factor, Maximum number of contact iterations: n_max


Limitations

Contact iterations:

  • can be used only with the static solution procedure;

  • require the use of the direct sparse solver;

  • cannot be used with coupled analyses with the separated solution technique; and

  • cannot be used with sparse format matrix input and assemblage.

Contact iteration requests that are not valid are switched off automatically.

The following features are not treated by contact iterations:

  • Contact with friction.

  • Points that are permitted to violate contact conditions.

  • Points that are released from contact by overconstraint checks.

  • Debonded contact.

For example, if there are frictional and frictionless contact pairs in the model, the state of the frictional contact pairs will be kept frozen together with the structural right-hand side and stiffness while contact iterations are performed on the frictionless contact pairs.

In geometrically or materially nonlinear problems, in contact problems with finite sliding, and in problems with features not treated by contact iterations, a nonzero residual force is expected when contact iterations are completed. Contact iterations may be not effective in such cases.

Contact iterations are abandoned if the internal linear system of equations is singular, which may happen due to overconstraints in the model.