Uy not converging in natural convetion peudo 1D melting case
 

Hello Forum,

Problem:
I want to simulate the melting of a magnesium ingot in a crucible, so I chose OF8 with its fvOptions "solidificationMeltingSource" to model the 2 phases in 1 fluid. The testcase is a slice (yellow marked area) out of a litte crucible as can be seen in the attached picture.

Convection is important so I want the flow to exit and enter the domain with cell value, as if z would be infinite. As of my understanding this leads to:
- "zeroGradient" on the velocity at top and bottom
- top and bottom: fixedFluxPressure as in chtMultiRegionFoam tutorial heatExchanger.

Edit:
This leads to jumping Uy residuals. The first timestep does not converge, then it runs for several steps just to diverge from 0.027s onwards and converging again from 1.3s onwards.

Question:
Is it crucial that all timesteps converge to a low residual? The pressure gradient in y is marginal and the absolute values of Uz are of magnitude e-09, so i think it's not easy for the numerics?

Attachment 82692
Attachment 82701

The material properties in fvOptions were wrong, hence the higher velocity. With the right velocity I encountered a new error and edited the original post to suit it. And the testcase isn't empty anymore :rolleyes:

Hey,

Have you solved your problem? I am running with the same trouble too. It seems like it is not easy to reach convergence using sMS.

Regards,
Mike

Just try to solve using the enthalpy porosity method.
Instead of using solidificationmeltingsource just add the source term to the temperature equation and update it according to the enthalpy-porosity or any update scheme.
You can even try to explicitly converge both temperature and liquid fraction at each update using a loop.

Hello,

Thanks for your response. What you said is another option too, though it seems that the fvOption sMS is programmed well. Anyway, do you have some test cases? I would be very glad to see and run it. Thanks!