Hello everyone

I am using fluent to simulate the ventilation in room. If the indoor air is mainly driven by buoyancy, which model I should use?

I have used standard k-e and RNG model. But it never gets to converge. I monitored the average temperature of the outlet. It began to fluctuate after about 800 steps.

My boundary condition is: velocity-inlet, outflow, and constant heat flux wall.

Could anyone give me some help?

Thanks

will

try boundary conditionsressure-inlet ,pressure-outlet,and the air is ideal-gas material

try boundary conditionsressure-inlet ,pressure-outlet,and the air is ideal-gas material

what is the boundary condition at your outlet ?

Try specifying velocity flow outlet

Hi,

If the indoor air is mainly driven by buoyancy?

I think the flow will be unsteady (note: this may again depend on the dimensions of your room). So, try out unsteady simulations.

I think your boundary conditions are ok. Outflow boundary condition is ok, if you are using bousinesqq assumption for modelling the density terms. ( note: bousinesqq assumption can only be used if the temperature difference between ur incoming air and the temperature of the walls or temperature inside the domain is very little). If the temperature difference is large then you have to enable incompressible ideal-gas law in the material properties and your outlet boundary condition should be pressure outlet. For more information please go through Fluent manual.

Coming to turbulence model now. If your case comes under plane wall jet (no lateral spreading) then I think RNG-K-e is ok( see Chen, Q., Comparison of different k- ε models for indoor airflow computations, Numerical Heat Transfer, 28, 353-369, 1995.)

If your case comes under three-dimesional wall jet then employing Reynolds Stress model will be appropriate,with RSM you will get pretty good results. But LES may also work well.

I am doing my thesis in indoor air flow , so if you have any more doubts contact me .

regards/

Pavitran D

I would guess your problem is of bouyancy, and you should model energy/ temperature. Pressure b.c.'s also seem like the way to go. If you have trouble with convergence try the k-epsilon realizable or spalart-allmarus.