Physically wrong heat fluxes during phase change




I'm relatively new to Simcenter and I have weird results when modelling Phase Change.

The geometry of my model is pretty simple. It's a 2D approximation of wax between two aluminum fins.

I coupled the fins and the wax with a thermal coupling and a total conductance of 10000 W/K to model extremly good contact conductance (I also tried different values, perfect contact, and merged nodes, but the pictures are for the 10k variant).

No matter what I try I keep getting a weird result: Although all temperature gradients are pointing in the same direction, one heat flux (at the melting front) always points in the wrong direction, contrary to the temperature gradient. And that's not how heat flow is supposed to work.

It's not even a very small heat flux which you might expect from numerical inaccuracies. For HTC = 10000 W/K the ratio between correct and bad fluxes is about 2 and for merged nodes e.g. the ratio is about 4.

HeatFlux_CloseUp_HTC_10000.pngHeat FluxTgrad_CloseUp_HTC_10000.pngTemperature Gradient


The coupling resolution is "one-to-one", the box "Only Connect Overlapping Elements" is not checked, Temperature Dependence uses "Average Temperature" and the Distribution is "Per Element". Both meshes have the same mesh size.


Additionally: When I used a fine mesh for the aluminum and a coarse mesh for the phase change material and made the aluminum the primary region (which is the recommended way), this resulted in sign oscillation in the heat flux.

Bad_TC_10000_Flux_Color.pngHeat flux with sign oscillation

Is it "supposed" to work like this? Did I design my model badly? Did I define the couplings badly? What other properties should I check?


Thanks a lot!