I overcame the problem of "Divergence in Solution" by setting up my grid in cartesian co-ordinates instead of body-fitted co-ordiantes in preBFC. The solution converged well in Fluent 4.4 with water being the flow medium. I then went a step further to perform a 2 phase modelling using the Eulerian(granular) model. I followed the steps in the Fluent manual and when trying to iterate, I came across the same nightmarish message "Rapid Divergence Detected *Abort*". If anybody is familiar with 2 phase modelling, please help.

You are in a good path! I am trying the same with oil-water stratified flow within a circular pipe, using Fluent 4.4.7.

Do not use Eulerian Multiphase, unless you have dispersions in your flow. If you do not have, then you can try the VOF free-surface Method, without donor-acceptor scheme activated.

Supposing that you have patched correctly the cells for your 2 phase flow, proceed with careful set up of the underellaxation parameters. Generally, those should not be greater than 0.1-0.2.

Also your grid may need further refinement. Try the refinement across the Plane, wrt which the two phases move.

If you do all the above, then you would not have any problem, however be prepared for very slow convergence. My current run has reached 25000 iterations after 21hrs......

If you are dealing with cartesian grid formulations, why do you do it in PreBFC?