[okstatecheme]

I'm simulating a injection of a chemical species (CO2, H2S and some others) into an airflow to model dispersion. I'm using LTSReactingFoam.

The flow domain is set up like a wind tunnel with a velocity inlet (zeroGradient Pressure) BC at one end a pressure outlet (zeroGradient Velocity) BC at the other. Air enters the domain at the inlet an flows over a geometry (Tankd with vent stack) with another velocity inlet (zeroGradient Pressure) BC that injects the chemicals into the flow domain. The sides, bottom surface and geometry are modelled as walls [U (0 0 0)]. See "domain" picture.

On accident I didn't have the gravity field set up correctly [g ( 0 0 0))] and got results that looked as expected. The outlet boundary condition was set as totalPressure. See "velocity" and "CO2".

When I corrected the gravity field [g (0 0 -9.81)] there was a significant amount of back flow at the top of the outlet boundary into the flow domain and back out of the bottom of the flow domain. See "velocity with gravity". The this recirculation zone causes the chemical plume to tail down. See "CO2 with gravity". Because CO2 is heavier than air I would expect the buoyant effect to settle at ground level, but it appears that the dispersement is being affected by convection.

Does anybody know of an alternative boundary condition that would eliminate this back flow? I've tried prghPressure as well with similar results. Or do I need to locate the outlet boundary far enough away from the plume I'm interested in looking at?

Thanks

[huttata]

Hi okstatecheme,

currently I'm heading exactly the same problem in a backward facing step validation case. I use the reactingParcelFoam.

Same BCs, same strange backflow at the outlet when I turn on gravity, also with inletOutlet BC. Did you figure out what was the problem in your case??

Thanks