[flavio_galeazzo]

Hi,

I am currently developing a solver for incompressible (but with variable density), multicomponent, steady state flow based on simpleFoam. It solves a equation for the concentration, which is used to calculate the density of the mixture. I want now to add the effects of buoyancy, due to density gradients created by the mixture, and I found two ways of doing it:

1. Modifying the pressure equation, as in buoyantSimpleFoam;
2. Adding the "weight" rho*g in the equation of velocity, as in reactingFoam.

With the first approach I don't get convergence. With the second one it converges, but to a wrong solution.

My questions are: What is the difference between these two approaches? There is one that is more suitable to my problem?

[sradl]

Hi,

buoyantSimpleFoam is based on a compressible fluid solver. Hence, you have to be very careful when using this as you might get problems with stability due to the additional equation for the density!

What I can suggest is to throw away the equation for the density and compute it locally from temperature (and your mixture fraction). Then couple this with a standard incompressible solver and underrelax the density calculation - you should have no problems with convergence.

Second, your wrong result described under 2. maybe due to the fact that the boussinesq approximation you are using does not hold (
http://en.wikipedia.org/wiki/Boussin...%28buoyancy%29

, i.e., you temperature differences are too big.

cheers
Stefan

[flavio_galeazzo]

Thanks for the reply, Stefan!

You are right, my maximum density difference is about 7:1, what is too large for the Boussinesq approximation.

I will follow your advice and use the approach from buoyantSimpleFoam. I will post my impressions here, as soon as get good results with it.

Flavio

[AlanR]

You could also look at buoyantBoussinesqSimpleFoam, which is an incompressible solver.

Alan