Boundary conditions buoyancy driven flow
 

Dear all,

I have been struggling for the past couple of weeks to find appropriate BCs for my problem. I am solving the buoyancy driven flow around a heated cylinder confined by an adiabatic ceiling (see fig1). The domain is opened to the atmosphere at the bottom and on the sides. The main issue is the side boundary, which is the outlet but where fluid should be free to enter as well.*Solver is buoyantPimpleFoam, with perfectGas as thermo model.

Here are the BCs which I am using right now:
sides
U: inletOutlet (inletValue 0)
T: inletOutlet (inletValue 300)
p_rgh: zeroGradient

cylinder
U: noSlip
T: fixedValue (between ~305-320)
p_rgh: fixedFluxPressure

topWall
U: noSlip
T: zeroGradient
p_rgh: fixedFluxPressure

bottom
U: pressureInletVelocity
T: fixedValue (300)
p_rgh: fixedValue (1e05)

With these conditions, the fluid is unusually accelerating in part of the domain as it flows toward the exit (fig2). In this region, the flow should be strictly decelerating in my opinion. I guess the problem is that neither p_rgh nor U is imposed on the boundary, but that is how my problem seems to be. I tried with totalPressure for p_rgh, but this creates some kind of blocage*at the exit in the top part of the domain.

So would anyone have a suggestion on what to impose here?

Also note that, physically, neither p_rgh nor T should have zeroGradient at the exit since T should be strictly*decreasing as the fluid moves away from the cylinder. I was thinking of something like extrapolating the values of both T and p_rgh to make the fluid exit smoothly, but I have not found any such BC...

Thanks!

So the main issue was the geometry, which supposed an infinitely large top wall, allowing fluid to simultaneously enter and exit the domain from the sides. In reality (experiments), the ceiling has a finite width and the flow exits vertically by buoyancy.

I enlarged the domain while keeping the width of the top wall fixed and now the flow exits rather smoothly the domain from the top left and right (see attached fig3). It's not prefect, but the residuals are good and the outlets conditions do not seem to affect anymore the flow near the cylinder, which is the region of interest. The flow is significantly different with the new geometry and BCs, as observed when comparing the two figures.

The new top outlets have the following BCs
U: inletOutlet
T: zeroGradient
p_rgh: zeroGradient

Both the bottom and sides have totalPressure for p_rgh, and fixedValue (300) for T.