Convergence Problems with contact simulation

[SwimBikeRun4342]

Hello Everyone


I know this is a common problem in ABAQUS (or at least it is for me). I'm trying to run an impact simulation of a basic bumper impacting a discrete rigid barrier. Clearly this is a dynamic model with surface-to-surface contact interaction. To avoid convergence problems, I allow for an automatic time step (ABAQUS defined time steps). However, the time steps are ridiculously small (around 1E-6 to 1E-5) and this would cause my simulation to take weeks to complete. However, if I define the time steps as more reasonable increments (around 0.01) there are convergence problems.

Previously I would always use automatic time steps, but these increments are far too small. Is there any method or trick I can use to make the time increments larger and the simulation faster? This is a real problem if I have to use time steps of 1E-6. I know contact simulations typically yield convergence issues, so there must be some methods to deal with this problem in a reasonable way. Any ideas?

More info on my simulation: A deformable bumber impacting a discrete, rigid barrier in a dynamic explicit analysis. Surface-to-surface contact is used. Time steps are set to automatic, but are way too small.


Any insight would be greatly appreciated. LEt me know if you need for information.

thanks

 

[burningDiesel]

1E-5 is not that small. How long does the impact take? About 1 sec? With a very large model (500000 DOF) this would run about 8h on a decent workstation (what is called an "overnight" run). If you are only modeling the bumper, this is more a 100000 DOF problem and would take 2h, which is not bad for an explicit calculation.

Are you sure you are not having e.g. memory issues (using slow swap memory) or I/O problems?