This study was made to have a better understanding of thermal Rayleigh-Benard instabilities. To understand this we made a first 2D approach.  The following results werwobserved, for a 2*1cm box, with a 1.5K temperature difference between the two walls. The fluid used is water, with a Boussinesq hypotheses. The mesh is made up of 40*20 cells, and the convergence is ontained in 87 iterations, with an initialisation at a zero speed.

We also tried out with a special initializing to obtain three rolls, in 56 iterations :
 
 

This showed the influence of the initial conditions on the final result. This first approach also confirmed the possibility of simulating Raleigh-Benard instabilities with Fluent.

This initial study led us to bring out two objectives :

• Try to simulate a 2D Raleigh-Benard instability in a free surface liquid in order to observe the height change of the surface of the liquid.
• Try to simulate a 3D instability to observe time variating instabilities.