Hi everyone,

I'm working with a 3D bubble column where the gas amount is so significant that I have to use the Euler-Euler method. My problem is the sudden change in gas fractions at the surface, which causes numerical unstabilities.

As a test case I fill half the column with water, and the rest with air. As long as the gravity is turned off everything looks nice, with a sharp cut for the surface. Turning on the gravity causes the surface to grow thick, and in a few iterations the case will diverge.

I have tried using higher order schemes, increased number of sweeps and decreased the underrelaxation with the same results. I have also tried to set the gravity to only 0.1 to decrease the shock on the system, but it does not help.

I have also written my own code using the well known 1. order upwind scheme. Here I solved this problem by using harmonic averaging instead of arithmetic. This made it possible to get nice, smooth surfaces even for heavily stirred tank reactors.

My questions are:

1. Is it possible to implement something like this in Fluent 4.5?

2. Have anyone any experience on how to best simulate such surfaces? (I want to have the outlet boundary in the gas phase, not in the two-phase area).

Regards, PC

Hi, unfortunatly you don't give enough information on the things you do. Therefore, some of my hints might be obsolete: - you perform an unsteady simulation and hence you should use the scheme geometric reconstruction to get a sharp interface. - what is your time step. My experience is to decrease it to 0.001 s or even smaller. - use the first-order time stepping.

I hope this helps a bit. I would be interested in any success or failure.

Greetings Detlef