[bokong]

I was investigating the mass conversation problem of implicitly solving particle volume fraction due to stiff particle pressure. And I built a case of a vertical periodic channel for twoPhaseEulerFoam with fixed gas flow rate. I made the volume fraction not uniform, with higher particle volume fraction on the top. The three following pictures are the initial condition, after 2 seconds and after 5 seconds. I used foam 24x here, instead of 30x due to the change of the pressure definition. And as you can see, I decomposed the domain to 4 processors.

init.jpg

2sec.jpg

5sec.jpg

Here is the case file, with the body force fvoption to keep gas flow rate constant.

verticalChannel_1.tar.gz

This is probably caused by the inconsistency of pphiPp on the processor boundary.

What is truly significant is that in this 5 seconds, the amount of particles increased 2%, which is a lot. That is due to the Laplace term in alpha equation, but even scheme was changed to "corrected", the mass still increases, which is a big problem for my current homogenous DNS simulations using my own solvers. If anybody has a good idea on how to fix it, I will be very grateful.

[ssss]

If you consider this problem as a bug, you might want to report it on the OpenFOAM's mantis bug platform.

Do you have the same problem in serial?

If I remember correctly there were important changes for twoPhaseEulerFoam from 2.4.x to 3.0.x