Hi, I'm running a time-dependent simulation, using electric currents and solid heat transfer physics. I'm ramping up the current with time, which heats the material. I have included an effect, which makes the current increase with temperature. At one point, the temperature and current will have a positive feedback effect, which should increase the temperature to infinity, and thus I get non-convergence at some timestep. This is expected. I need the solution before the non-convergence.
Now, when doing regular simulations (without parametric sweep), the solution is kept, but when doing a parametric sweep (e.g. over a range of geometries), the solutions are not kept when non-convergence occurrs.
Currently I'm doing every simulation without a parametric sweep, find the timestep when it doesn't converge and, create stop-conditions for those timesteps and then run the parametric sweep to get all the data conveniently. This is inconvenient and basically means I have to simulate everything twice.
Any ideas? How to keep the non-converged solutions in parametric sweeps?
Now, when doing regular simulations (without parametric sweep), the solution is kept, but when doing a parametric sweep (e.g. over a range of geometries), the solutions are not kept when non-convergence occurrs.
Currently I'm doing every simulation without a parametric sweep, find the timestep when it doesn't converge and, create stop-conditions for those timesteps and then run the parametric sweep to get all the data conveniently. This is inconvenient and basically means I have to simulate everything twice.
Any ideas? How to keep the non-converged solutions in parametric sweeps?
