CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Solver Failure due to unrealistic high Mach number (http://www.cfd-online.com/Forums/cfx/27139-solver-failure-due-unrealistic-high-mach-number.html)

Andreas March 13, 2009 04:29

Solver Failure due to unrealistic high Mach number
 
Hello everyone,

I have a strange problem. I am simulating slightly different geometries using pretty much the same unstructured mesh every time. Especially the wall distances are the same in every mesh, first node about y+ = 50 to 100 - just as required by the CFX-manual. My inflow air velocity is 36 m/s, so it is far from being transonic.

Strange thing is: Sometimes the solver will crash due to an Mach number that tends towards the hundreds. I could localize the region of the high velocities to a region close to a wall where separation most likely occurs.

Does anyone has an idea where this problem comes from and it can be solved? I am using SBF=0.75 and a physical timestep that is about five times higher than the automatically calculated one (in order to avoid the simulation of transient osciallations). The solver precision is set on double. The element quality as calculated by ICEM should be alright - worst element about 0.01, but only a handful maximum. And as I said: This setup works in 19 of 20 cases!!!

I am grateful for any ideas, maybe there is a expert parameter that could solve my problem?

Regards,

Andy

Bart Prast March 13, 2009 06:58

Re: Solver Failure due to unrealistic high Mach nu
 
Your time scale is too large. Start with local time step factor. And use the high resolution schema with automatic blend factor (but this is your choice ofcourse). The fact that your setup works in 19 of 20 cases means your were lucky (just inside stability range of your code/problem).

Success,

Bart


All times are GMT -4. The time now is 06:10.