Mixed convection with large Buoyancy force - cannot converge
 

Hi
i am simulating the flow and heat transfer in a rectangular cavity, i want to get the temperature distribution in the steady state.
It's size is 6m(length)*10m(width)*8m(height), the top surface (6m*10m) is considered as velocity inlet with constant and uniform velocity and temperature of V = 0.023m/s and T = 137 K, the four side walls are considered as constant and uniform temperature of Tw =177K. The outlet is located at the bottom of the cavity. Nitrogen is considered as the working fluid, and the boussinesq approximation for density is used.
Due to symmetry, i modeled 1/4 of the cavity. The grid size is tested to get the Y+ between about 30 and 70.
The Reynolds number and Grashof number for the above condition is calculated as about 10^4 and 10^14, respectively. RNG-ke model is used for turbulence. "Standard wall functions" is used for near wall treatment.
Under the above boundary conditions, i can get the converged solution, as the residual for all variables decrease to less than 10^-4.
Then i want to get the results for other inlet velocities, but i cannot get the correct results.
1. When i increase the inlet velocity to 2V or even higher, the residuals keep fluctuating at the level about 10^-1, and no converged solution can be get.
I tried to increase the grid number, but it still does not work.
2. I decreased the inlet velocity to V/2, i can get the converged solution. But the results are obviously not correct, as the results show the average temperature in the cavity becomes a little smaller.
This problem has confused me for a few weeks, as each calculation lasts for 2 days. i don't know if the models and procedure i used for such parameters are correct, or the problem itself is not stable?
please give me some advise if you can help. Thanks a lot.

I think the CFL number is too high when you increase your velocity to 2V. Try to change the time step and the mesh size to keep CFL number below 1.

The post was made an year ago. :D