14.2. Troubleshooting a Springback Analysis

Springback analyses are sometimes numerically difficult to solve. Some typical causes of convergence difficulties and actions you can take to resolve them are listed below.

If you have checked all of the above conditions and the analysis is still not converging, try the following methods (in the order listed):

14.2.1. Springback Stabilization

If convergence difficulties persist, you should activate springback stabilization using the SPSCALE and MSCALE arguments of the RIMPORT command. The stabilization method is analogous to attaching artificial springs to all unconstrained degrees of freedom. The stiffnesses of these springs are directly proportional to the stiffnesses of the elements.

In each substep, ANSYS will assign the artificial springs large default stiffnesses to simulate a total restraint of the motion of the model at the beginning of each load step. The artificial stiffness is scaled down exponentially with the progress of iterations. (The picture below illustrates rapid decline of artificial spring stiffness over an iterative cycle.) The substep is considered to be converged only when the equilibrium convergence tolerances are met (as in a normal nonlinear analysis) and the artificial stiffness has reached an insignificantly small value. This may slow down the conversion of the solution in some cases; however, the benefit is more robust behavior of the model.

The SPSCALE parameter on the RIMPORT command is a scale factor that you input to scale (up or down) the default initial stiffness of the springs. (You must input SPSCALE to activate springback stabilization.) You can also specify an acceptable small stiffness value using the MSCALE parameter on the RIMPORT command.

The springback stabilization method is valid only when the geometric nonlinear option is active (NLGEOM,ON).


Release 18.2 - © ANSYS, Inc. All rights reserved.