I’m modeling ice deformation by dislocation creep with the base of the ice frozen to the rock beneath it. The ice deforms such that the right boundary (initially almost perpendicular to the base) contacts the rock. It should then freeze onto it, so essentially there should be adhesion of the ice to the rock. I’m able to get a contact model both with (attached) and without adhesion to converge, but only up to ~200 years, and then it reaches a singularity and doesn’t converge. I think that part of the reason for this is likely the geometry, since the two sides of the contact pair are again almost perpendicular at that point. Is there any thing I can try to improve the likelihood of convergence for this sort of geometry? One idea I've had is to decrease the mesh size, but I've had some problems with decreases in mesh size making a similar model not converge for as long so I wasn't sure that would help.
↧