Hi, I want to model natural convection processes inside a square enclosure(two dimensional) with a partial wall inside it attached to one of the edges and vertical. Now when i mesh the geometry and solve the partial dividing wall is not considered and the flow is seen ACROSS the wall, while actually the flow should change direction at the wall obstacle. I am new to CFD and fluent and would appreciate help in how to go about doing it. Any other considerations which should be kept in mind for such problems??? Thanks for the help. sandeep

Dear Sandeepmishra

It is no problem to construct dividing wall inside the cavity. if you do not assign the inner dividing wall as 'WALL', then Gambit consider it as 'INTERIOR', resulting in cross flow over the dividing wall. Check your dividing wall is considered as really 'WALL' by Gambit and by Fluent. Just before, I have tested your case, and I had no problem expect fairly difficult convergence. It takes only 10 minutes by my P-III PC.

Your case is very simple one, so that you should have good results. Good luck with you.

Recommendation : If you have both of Fluent 4.x and 5.x, I recommend that Fluent 4.x rather than 5.x for natiral convection problem. Convergence is very good by using 4.x.

Sincerely, Jinwook

Thanks ginwook, I will definitely give it a try by the way you suggested. Thanks again. sandeep

Dear Sandeepmishra ,

I have done some work in this area . There are a two of options available for your problem.

1. If you want to model the wall as conducting then follow Jin-wook's suggestion.

2. If you want to model the wall as adaibatic it is better to model just the fluid domain and leave the partition as empty.

The default BC at the domain bounadry i.e

adiabatic wall, will ensure that no flow occurs at the region.

Other suggestion -

Bias the mesh near all the walls

Use correct Boussinesq inputs to achieve quicker

convergence.