[Hkp]

Hello,

I'm working on developing a model of plasma jet flow for plasma cutting. I've attached figures of temperature distribution at the same time. They are the results of parallel and serial run.

As you can see the figures, the unphysical temperature is seen at the interface between subdomains for parallel run. As time goes on the simulation by parallel run blows up. For the results of one processor run(serial run), the temperature distribution looks good.

Scotch method is used for decomposition. I already checked the mesh quality but it is fine. I don't know what causes this problem. Do you have any idea about this problem?



HK

[carstenf]

hi, i'm relatively new to OpenFoam and I didn't thought that there are problems in this way. interesting to know.
Did you play around with the hierarchicalCoeffs?
for example (2 2 1) which can be better than (4 1 1)
Maybe there are problems if the borders of the subdomains are on positions with high gradients

[alexeym]

Hi,

Originally Posted by Hkp

I'm working on developing a model of plasma jet flow for plasma cutting. I've attached figures of temperature distribution at the same time. They are the results of parallel and serial run.

Can you be more specific? Did you modify solver code (and BTW what solver?)? What modifications you did? What version of OpenFOAM do you use?

I can only guess that in your model you modify certain volume field with = operator and then forget to call correctBoundaryConditions method. Though the real reason can be elsewhere.