I have a very simple geometry:a rectangular configuration with walls everywhere and three windows. One window in the ground is set as mass flow rate, one in the ceiling is a pressure outlet and another one in the front wall as pressure inlet. I am modeling a very simple gas dispersion problem. My mass flow rate is 0.037 g/s of helium. In the first second I am guetting more helium gas in the interior volume than I should really have . Using the same configuration but with a larger mass flow rate : 3.7 g/s , i have everything correct , there is no mass imbalance. I should say that I have an incompressible flow at room temperature and pressure. Please help.

Is your model converged?

yes it converged using the segregated solver with 2nd order discretization scheme. I even reduced the convergence criterion (up to 10e-5) for some parameters but the mass of the gas delivered is more than double what I should really get in 1 second. I worked on many flow problems using fluent , this never happed to me. If you have any idea, it will be greatly appreciated.

are u sure that ur physical model is correct

well, I tried laminar (i have very low reynold number) and RNG-ke . Both gave similar results. What I noticed however is the results are very dependant on the pressure inlet/outlet I have. If I have 0 pressure on both pressure inlet/outlet, huge mass imbalance occurs. If I put some values on these inlet/outlet the result is different but still some imbalance occurs when the convergence is obtained.

r u using some well posed problem or just trying to see the solution behaviro in fluent . do u have some analytical or exerimental results

Thank you for trying to help. It is a validation problem and I have some experimental results.

Check flow direction at the outlets or openings. I did have a case in which the mass flux report was not correct. I modelled a flow domain without inlet and outlet, the flow was driven by a fan. I found at an opening, the flux report gave a different direction from the flow vectors.

tanx a lot WX. You gave me I hint. I usaully set the direction of the flow as being NORMAL TO THE BOUNDARY, which I changed now by specifying the direction vector (x=0, y=1, z=0) in my case. And Eureka, everything works well. Thanks again.