Unsymmetric CG method failed

[bigmigfan]

Hello guys,

I am trying to run the cfd-thermal analysis. My model was working fine until, I changed the specific heat of the fluid and I started getting this error message.

"Unsymmetric CG method failed catastrophically"

I checked all the boundary conditions and constraints. They seem to be correct as the previous model was running fine. Only modification was the value of the specific heat.

Does anyone why does this occur and how to resolve this error?

 

[bigmigfan]

So nobody faced this ever before. The error message is not even documented in the ANSYS TROUBLESHOOTING GUIDE. I know it has to do something with convergence crieteria, I tried relaxing convergence criteria but it is giving me the same error. Any suggestions?

 

[Stringmaker]

Is your mesh too coarse? Is the flow turbulent but you have laminar specified? I'm not a fluids guy just throwing some stuff out there. I think without being able to reproduce the issue (i.e. you providing an input) this is a tough issue to diagnose via these forums. Have you had tech support look at the model?

 

[pja]

By changing the specific heat you must be decreasing the "nice" properties of the system of equations which you are trying to solve. I.E the condition number may be getting too large. The convergence or lack thereof of iterative methods is function of the conditioning of the system. Have you tried changing the type of iterative solver used? I don't use the fluid solver in ANSYS so I don't know if this is a problem. Also you might want to try reducing the specific to a level which you can get a solution and then perhaps that solution as initial conditions for the case which you are interested in.

 

[bigmigfan]

Thank you for your response. The flow is turbulent. I have tried changing the CFD solvers as well. I am sure that the reason the CG method fails is due to iterations don't converge. Only if I could find out how to untight the convergence criteria. I am playing with the relaxing parameters. Anyways, please let me know if you have any suggestions.

 

[cbrn]

Hi,
wouldn't it be possible that the new value of specific heat causes a physical nonsense somewhere, such as reaching negative absolute pressure or non-positive temperature (in K)? This could only be the case if there is a fixed-power heat sink or source somewhere, otherwise I can't see it...
Never used the fluid part of ANSYS, so I don't know, but would it be possible to solve the problem for small intermediate timesteps and see at what point it is running into troubles?

Regards

 

[bigmigfan]

Thank you for everyone's support. I was using the TDMA solver which specifies tight convergence criteria. I read slover guide and found that the PBCGM solver is the suitable solver for the unsymmetric conjugate gradient situtation and turbulence conditions where initial gradients are high. So changing the solver to PBCGM resolved the issue.