FVM vs FEM for natural convection in square cavity
 

Hello all,

I am dealing with a problem of natural convection in a square cavity with two side faces being adiabatic and top face at room temperature and bottom face has time varying temperature BC (ISO curve).

I have analysed this problem in FLUENT using transient analysis using Boussinesq approximation. After certain time the velocity and temperature contours lost symmetry in the cavity and we believe this is due to onset of turbulence. But when we solve the same problem with in house developed FEM code, the velocity is losing symmetry much later in time than it did in FLUENT. Although, FLUENT results are much closer to experimental, we are trying to figure out what are the reasons for the differences between FEM and FLUENT results.

Any inputs from the experts would be appreciated.

Thanks