Hi, kengwit.
I encountered something similar in attempting to model high impact forming with subsequent steps.
I found it useful to add in an extra step with nominal time. Applying full motion constraints to all nodes I was able to remove the k.e.
I was then able to continue with my...