I have observed puzzling behavior in a nonlinear solution. In my output file, after a converged substep at TIME = 39.8846, a next step with NEXT TIME INC = 0.15928E-01 is attempted, but fails to converge due the DOF limit being exceeded. The program then says it bisects, but the time step size does not change: NEW TIME INCREMENT= 0.15928E-01. In this bisected solution, the incremental plastic strain computed is larger than the criterion of 15%, leading to another bisection. This time, the time increment changes only slightly: NEW TIME INCREMENT= 0.15359E-01. With this setting, the substep converges.

Why does the timestep size not change at all in one bisection, and change only slightly in the other bisection?


This behavior becomes less puzzling when one realizes how the gradual changes to the Bisection feature have given the term `Bisection` an evolving meaning in ANSYS. Over the years, heuristic improvements to automatic nonlinear solution controls have expanded the meaning of `Bisection` in the ANSYS program, so that one could say that `Bisection` now means `automatic restart with automatically changed solution controls.` Timestep size is just one solution control that the program can change automatically in such a restart. The objective in the current program design is to try to achieve the most rapid solution possible, by automatically striving for a balance between reliable convergence and maximum timestep size.

In the output that you sent, after TIME = 39.8846, a non-converged solution `Bisected,` or was automatically restarted, with a changed solution control: the Predictor automatically turned off. No other solution control changes took place in this restart -- including the time step size, which remained unchanged. This automatically restarted solution converged, but had a plastic strain increment slightly above the limit of 15%, so another `Bisection` took place, in which the timestep size was reduced slightly, from0.15928E-01 to 0.15359E-01. The time-step size reduction in this instance was based on the ratio of maximum plastic strain increment to the plastic stain increment limit.





Show Form
No comments yet. Be the first to add a comment!