strange phenomenon for compressible gas in closed volume
 

Before sorry for my bad english.

Dear colleagues, I need help for relatively simple task on Fluent 15.0.

There is closed 2d-volume with gas. Gas is compressible (ideal gas or Redlich-Kwong, it doesn't matter). Gravity operates on y-direction. As Fluent requires for compressible gases with buoyancy, I set reference operate density as 0. The task is transient. No heat transfer, no velocities, no sources, only gas in state of rest. Coupled scheme is used as Solution Method.

Further I encounter very strange phenomenon.

Velocity vectors appears on bottom cells in the volume (see fig). They directs down on gravity direction. For different turbulence models (I have seen laminar, k-e standart and SST) the behavior very different. For k-e model velocity is small (about 1.6 m/s) and stable during calculation process. For laminar model velocities is most great and increase during calculation. Therewith temperature and density in bottom cells also change. SST model gives intermediate results with other models.

If set operating density to average density in the volume, the vectors disappears and velocity in the volume becomes 0. If I increase operating density, the vectors changes its direct to reverse direction.

As I understand, operating density (reference density) appears in term (rho-rho_op)*g and its value (which I impose) affects on convergence (see fig) but it is not to affect on field variables. I fail to see what reason in addition can account for this phenomenon.

I'd be glad any help.

I have solved this problem having established Body Force Weighted for Pressure Spatial Discretization in Solution Methods tab.

Anybody can explain me how this method operate?

Rhie chow type dissipation is added in flux computation that contains body forces.

Coeff * ( averaged body force at face - interpolated body force at face).

Thanks for your reply. I will study the theory