Convergence problems are very problem specific, but often arise due to poorly formed elements. Run the mesh checker (in classic); if you have more than 10% "red" elements, you need to smooth or reformulate the mesh. Sometimes, the autostepping can get bogged down working on a small area in the model. You can fall back to the traditional user-specified stepping in which you grow the steps using a multiplier of 1.1 to 1.3 and start from a very small value, like 1e-6. Also, do not let your steps grow too large. In transient models with a lot of stress variation, user-stepping can get very complicated. When using user stepping, you must check the mass balance is ok and that the local balance is ok in areas of concern (because user stepping may not yield optimal results). Experimentation is necessary.
Pete