Non-zero mass flow rate for a confined region and bizarre vector field at wall
 

Hi everyone,

I am using Fluent17.1 to simulate the performance of a 2D water column and the results I get are very problematic. I will greatly appreciate it if you could kindly provide me with some suggestions.

The flow domain is rectangular(0.2m(horizontal)*3m(vertical)) and confined by four walls(left, right, top and bottom). The left wall is heated with a constant flux and the rest are adiabatic. The working fluid is water and Boussinesq approximation is used to include the buoyancy effect. The conjectured temperature change (<1 degree) should be low enough to keep the Boussinesq approximation valid. I ran the simulation in transient scheme and get some problematic results:

1. The confined domain has a non-zero mass flow rate, not even close to zero(somewhere around 0.05kg/s for the finest mesh I tried), while the mass flow rates across the boundaries are all zero. With CFL number kept constant (by changing time step accordingly with the mesh size), I used different meshes to run the same simulation and found that mesh refinement only made the results worse. I used both laminar and turbulence models to run the case and also tried to decrease the CFL number, but the problem persists.

2. The vector field near the top boundary shows some abnormal features. Since no-slip condition is applied to all the walls, I am expecting the velocities at the walls to be zeroish. However, in the vector plot, I got some vectors strangely pointing outside at the top boundary with abnormally large velocity magnitudes. There was no smooth transition between these vectors and their neighbours so I don't think it is the result of convection.

Could anyone give me some insight into the causes of these problems? Any help and advice will be greatly appreciated.

It is my first thread here and please let me know if you need any further information about my case. Thank you.

Regards,
appotoxin