While using the thermo-mechanical solver in DAMASK, we (me and Tias) figured that if we specify some thermal source, we need an initial deltaT for convergence. For example, even if we have a bicrystal and we give zero deltaT then the calculation fails even at the first increment because the 2-PK is zero.
However, if we start even with a minute deltaT (1K or so..) we get convergence. But if we have a single crystal this situation doesn’t arise. This is something to keep in mind, and possible requires delving into.