Ok, In an implicit dynamic simulation, in the first step I apply a force to tension a fastener assembly.  The force is applied at a reference node which is also the control node for an mpc which includes the back face of the tension member.

In the second step, i remove the force and replace it with a bc, constraining the ref point from translating in any direction and only allowing rotation about one axis.   In this step a curved analytical rigid surface is translated upward into the assembly to simulate wrapping the assembly around a drum.

Everything seems to work well, except that the bc that I added in the second step to replace the pull force does not hold position.  

Am I missing something?  I have not been able to find and reference to issues like this when using bc's in an implicit dynamic simulation.

Attached: pic of overall assembly    and       overlay plot of tensioned condition vs beginning of bend condition.