CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Compressible external flow (https://www.cfd-online.com/Forums/fluent/32304-compressible-external-flow.html)

 Rob October 10, 2003 18:30

Compressible external flow

Hello,

I am trying to simulate an external compressible flow (M=0.5) over a hemisphere on a flat plate in Fluent 6. I am starting off by simplifying the problem to just a 2-D cylinder atached to a flat plate. So the flow goes along the flat plate then over the 2-d cylinder then along the flat plate again (just for a visual). I can run the case no problem with the air being incompressible, but as soon as I change the density to ideal-gas I get an error message that reads"Error: divergence detected in AMG solver: temperature" and the continuity and energy residuals jump off the charts 10^15. I have tried both the velocity inlet and mass flow rate as a inlet boundary condition and I have the outlet boundary condtion set as a pressure outlet. My grid is a structured grid that is fine, I checked its quality in gambit and the skew is less than 0.1 for all elements. I also tried to run the outter boundaries as pressure far field however since the outter boundaries actually meet the edges of the flat plate that did not work either. Any suggestions on how to get this case to converge would be great. Thank you for your time and I hope to hear from some of you CFD gurus!

Rob

 Venkatesh V October 15, 2003 03:58

Re: Compressible external flow

Hi Rob,

I hope you are solving with default Courant Number. For initial few iterations reduce the courant number.

Venkat

 Rob October 15, 2003 12:55

Re: Compressible external flow

Hello Venkat, Thank you for the response. I have discovered that reducing the courant number will keep my solution from diverging with using the pressure far field boundary conditions. However the convergence is real slow and the residuals oscillate with a large amplitude. So far I have completed 100,000 iterations and the lowest the residuals have come down to is 5*10^-3. The residuals oscillate between 1*10^-2 and 5*10^-3. So I guess it will be slow going. However if it is this slow with a 2-D case, is it realistic to attempt a 3-D case?