I'm running an Abaqus/Standard implicit dynamic analysis with some realistic material damping in a flexible elastic part and some speed-dependent viscous damping in a HINGE connector that also has connector stops to limit its rotation.

The model is driven through some snap-through transitions by a frictionless sliding contact with a rigid body. The contact is intermittent because of the snap-through transitions. The rigid body actuator moves at a constant velocity and sometimes the elastic part gets ahead of it because of a snap-through. Kinetic energy and element velocities are generally small, but since KE isn't always totally negligible, I'm running the analysis with application=MODERATE_DISSIPATION rather than application=QUASI_STATIC.

Most of the time, the behavior is plausible and the energy balance is okay. Energy in ALLSE is dumped to ALLVD dissipation during the snap-through.

But there are intermittent strong dissipation events where ALLVD sharply jumps up. That energy does not come out of stored elastic energy, external work, or kinetic energy. It just appears from nowhere as a sudden increase in ALLVD and therefore there's a corresponding unphysical jump in ETOTAL.

I switched from "hard" normal behavior to exponential on the sliding contact and it seemed to improve the results, suggesting that work done by contact forces was not being calculated correctly. This is discussed in section 6.3.2 of the Analysis User's Guide under "Intermittent contact/impact." I decided to resolve it with exponential contact in the sliding part instead of changing the analysis parameters.

But this only reduced the frequency of the jumps in ALLVD. It did not eliminate them entirely.

I suspect that I'm having a similar problem with the work done by the hard connector stops, but I'm not sure how to check it and whether or not there's an option to soften those stops. I might try a nonlinear elasticity in the connector that stiffens a large amount near the stop instead of the hard connector stop, but other ideas are appreciated.

Thanks,

Dan