Is there any output I can enable (e.g. nodal residual force) that I can then contour to determine where in my model is causing problems with convergence?
My model has nonlinear material and some significant displacements, but I don't expect any buckling, as precalculated by the linear buckling solver. There is no contact in the model, but there are some CBUSH spring-damper elements. The last converged subincrement shows no plastic strain, and no high stresses.
I believe I've found what I was looking for. The output quantity Total Summed GPForce seems to show the residual force vector, which indicates which nodes have either unconverged or difficult-to-converge degrees of freedom. If someone can confirm that it is indeed the residual force (i.e. the R vector from the itrerative nonlinear formulation F = KX + R) that would be nice.
This pointed to some rod elements which are loosely constrained to simulate a lifting condition which is not important in this analysis. I will tighten up the stabilizing CBUSH element there and it should work.*
*After removing the rod element "pyramid" the model converged right away.