Questions about buoyancy driven flow
 

I'm doing a flow diagnostic for a combustion lab. A ventilating hood is suspended above the burner and the hood is connected to a pipe systems where there is a fan at the alternative end.
The hood is like a rectangular nozzle that its smaller cross section is connected to the pipe line as mentioned before. The measured velocity at this cross section is at an average of 5.45m/s.
If the burned gases can successfully reach to the top of the hood(the smaller cross section), they will be ventilated.
It seems that at working condition, the burned gas may escape to the surroundings rather than go up to the top of the hood and then be sucked off.
I'm performing a 2-D axisymmetric simulation now, but doubts arose frequently.
Provided by inlet velocity 10m/s at 800K
The simulation is carried out with B.Cs as following:
velocity inlet: 10m/s, turbulence parameter: intensity 5%, length scale 0.07*D=0.0007m, thermal:800K
Wall-hood: no slip
Wall-free: no shear, slip wall
pressure outlet: gauge pressure 0
Backflow direction: normal to the boundary
average pressure specification: ticked
Target mass flow rate: 0.054398 (0.1m diameter with the measure velocity(5.45m/s)
Turbulence: backflow turbulence intensity: 5% and backflow hydraulic diameter: 0.1m
thermal: backflow total temperature: 500k

Are there some mistakes in the boundary conditions?
Because there were terrible reversed flow at the pressure-outlet and I tried to extend the exit but to no avail.
Could you pls give me some suggestions on this?
And I intended to use Boussinesq Model to account for buoyancy effect; however, in the user's guide, it states that the limitation of this model is large temperature difference. In my case, is it with a large temperature difference?
If this model can be used, how can I determined those inputs for Boussinesq parameters, namely the operating temperature, operating density, thermal expansion coefficient. Moreover, is this operating density the same as that in Material section where the thermal expansion coefficient can be inputted.

I will be so grateful if you could help me on this!!!!!!
And if anything I did not specify or not clearly stated, pls let me know.
Here are the schematic drawings; hope they could make the scenario more clear

http://www.facebook.com/photo.php?fb...type=1&theater

https://www.facebook.com/photo.php?
fbid=701577913285491&set=pcb.701577973285485&type= 1&theater