Mesh sensitivity when using phase change models
 

Hi,

I am having problems with the variation of solidification rate when solving different mesh sizes while using the Eularian multiphase VOF model with enthalpy-melting-solidification models enabled in CCM+.

I am solving the simplest model I could make (a 10x10x10mm cube) and using adiabatic surfaces on all boundary conditions bar two. The two faces that aren't adiabatic are set at different temperatures causing a delta T across the phase change material to induce melting. I am solving in unsteady state because I want to measure the rate at which the cube will melt.

The mesh size is causing large variances in the rate at which the cube melts. A finer mesh will cause the cube to react slowly and melting takes a long time. Where as a coarser mesh will cause the cube to react faster and melt quickly. I varied the mesh from 16 cells to around 37 million cells and I haven't been able to reach convergence on my mesh sensitivity test simply because I ran out of memory (96GB!) due to fact that the finest mesh I can create is too fine (100 million cells). Has anyone experienced such a large sensitivity in mesh size before?

Feel free to ask any questions you need to.