Heat equation with phase change: poor convergence

[mzenker]

Hi,

I want to simulate the electric heating of a conductive material containing water, including the phase change (evaporation) of the water at 100 °C.
I am using the Heat Equation and the StatCurrent solvers, with phase change activated. The results are looking reasonable in principle, but I observe that when the phase change occurs, the convergence becomes very poor, the iteration often reaches the maximum number of steps without converging. So I am not sure if I can really trust the results, and if the simulation is already sufficiently stable. Later, I want to simulate more complicated geometries with the same setting when it works reliably with the simple geometry I have used here. And in a next step, I would also like to try to simulate the shrinking of the material as it dries out.

I have tried with smaller timesteps, but without real improvement. Adaptive timestepping does not work for me (I have started another thread on this subject, without response so far).

Due to my lack of experience and knowledge in FEM, I do not really know where to start to improve the situation. Should I tune the solver parameters, use still much smaller timesteps, a finer grid, another solver (the ModelsManual says that the internal phase change solver is perhaps not the best choice for abruptly changing peoperties, which may be the case here)?

Furthermore, in the tests directory, there is one example for phase change with the Heat Equation solver. There I have seen the following keywords, for which I did not find an explanation in the ModelsManual:

In the equation section:
Check Latent Heat Release

In the materials section:
Phase Change Intervals

Did I miss something, or are those keywords undocumented? What are they doing exactly? Could they be useful here?

I attach my test case and would be very grateful if someone with more experience than I have could have a look at it and give me some hints what I could do in order to proceed with this problem.

Thank you very much in advance,

Matthias