[rembe]

Hello,
I'm new in this forum and I hope I get some hints helping me to understand my problem. first of all, sorry for my bad english, I'm german.

Using Comsol I simulate brine flow in salt caverns (50... 130m diameter, 50 .. 100m height). It is a laminar up to turbulent flow. At the surrounded wall there is salt dissolution. Salt disssolution causes a higher density of the fluid (rho=f(c,T)), so that mainly buoyancy flow is there. Of course, there is a fresh water input over a well and a brine extraction by a well, too. But the flow (about 10 ... 50 m³/h) is small andit isn't able to control flow regime.

I use FEM (Comsol 3.5a) for solving navier-stokes-eq. and the convection-diffusion-eq. There are many options to solve and tricks to get a solution, but there are some main problems I want to rule.

Buoyancy flow is considered by the Boussinesq-Appr. Fz=-g*delta_rho but it seems, that it isn't very numerical stable. I know there are limitation, delta_rho is about 1300-1000=300kg/m³.

Is there any better way to consider buoyancy flow? This seems to be the main problem.

Other problems are mass transport with a cell-peclet-number >10 ... >100, flow with a cell reynolds-number about 100.

Actual I use a laminar stationary solution, decreasing viscosity and diffusion step by step (restarting) and increasing g of the Boussinesq-Appr. Limit is the number of DOFs to solve for 3D! I use a DELL T7400 with 2 x QuadCore and 32GB RAM.

Is there anybody who works nearly on the same task?

Thank you for any answer!

Michael