[gharek]

Hi cfd users!

I'm currently puzzled as to what kind of boundary conditions I should be imposing if i were to simulate a vertical channel, one side heated and the other side adiabatic.

I've tried with pressure boundary conditions at both inlet and outlet ends of the channel and the whole thing just exploded in 3 iterations. I presume this is because of the huge temperature jump at the inlet.

Hence, I've tried to extend the outlet and inlet so that I have an I-shaped computational domain. The solution is very unsteady and did not converge at all.

Any useful advise will be greatly appreciated.

Cheers

Ghar

[ztdep]

just impose inlet bc at the inlet , outflow or pressure outlet at the outflow.
please detail your algorithm to solve this problem.

[gharek]

The Navier Stokes equations are discretized by the control volume method. The transient term is taken into account, with an extra buoyancy term in the y-momentum equation. The vertical channel height is taken to be 2 m, while the distance between the heated and adiabatic sides is 0.1 m. So far, I have got 50 mesh points across the channel width, with a time step size of 0.01 s.

Since the flow is driven purely by natural convection, I dont know the inlet velocity, hence P=0 would seem to be an appropriate boundary condtion at the inlet and outlet. Velocity is solved by an initial guessed pressure, which I've set to 0.

[ztdep]

for natural convection prolbem , use pressure inlet at the inlet. i have used it for the natural convectin about a cylinder in an infinite space .