Error during initialization of "rhoSimpleFoam"
Hello everyone,
I am running a "rhoSimpleFoam" case and got an error message while it just initialized the case (see below). At my walls I am using: "alphatWallFunction", "compressible::epsilonWallFunction", "compressible::kqRWallFunction", and "mutkWallFunction". I can't see where I did a mistake, does anybody know what's wrong? Best regards, Frank The error message: Code:
/*---------------------------------------------------------------------------*\ my "thermophysicalProperties" file: Code:
/*--------------------------------*- C++ -*----------------------------------*\ my "fvSolution" file: Code:
/*--------------------------------*- C++ -*----------------------------------*\ |
Hello,
you could try to start your simulation without using a turbulence model. When the solver doesn't blow up, you should check the physical plausibility of the initial conditions for the turbulence model. Regards Thomas |
Thanks, Thomas! I did it and it actually ran through without a turbulence model. So it seems something's wrong with my turbulence settings. Do you mean the k and Epsilon initial values with initial conditions for the turbulence model?
|
Hello,
yes, I meant k and epsilon (or which values your turbulent model is modelling). There is a very usefull tool on cfd-online to estimate k, epsilon, omega and nut. Follow this link. Consider the turbulent length scale! In Fluent it is 7% (I am not Sure about this value) of a length, that is characteristic for your Problem. There is also a hint on the URL above. Since I initialize with the "Fluent-definition", my Simulation is not/rarely blowing up because initial problems. Regards Thomas |
Oh wow, that's neat! Thank you so much! I already calculated k, eps and mut but my values seem to be unrealistic low. I'll try this calculator :)
|
Alright I got reasonable values for k&eps but it's still the same error message... I've found THIS site where it's explained where "sigfpe" originates from and it seems that it can only be a "division by 0 from having an initial field set to 0" issue.
I actually do have some initial fields set to 0 (pressure, alphat, mut) but that shouldn't be a problem, right? |
Hmm ..ok, there seems to be another mistake. I often set pressure to zero. Did you run checkMesh? Maybe there are bad elements?
You can also unset the Floating Point Exception control by using this command "unset FOAM_SIGFPE". Sometimes, the calculation became stable after a few Iteration steps. |
Yes, I did checkMesh and there are actually some high aspect ratio cells. But I did some incompressible simpleFoam calculations before with this mesh and I never had with those before. I also changed the inlet- and outlet types of k/eps but the error message remains the same.
It's interesting that the message occurs even before the first iteration started, does the solver even check the plausibility of the bc valued in this state? If I unset sigFpe it actually runs through to the first iteration. The Ux/Uy/Uz residuals are pretty reasonable but my h-initial-residual is already at 10^-6 at the first iteration. And every following one (T, k, eps) are "nan". In a different post I found this if the "nan" error occurs: Quote:
In the T-file I just defined a fixedValue at the inlet and a inletOutlet at the outlet. In the pressure-file I defined a totalPressure inlet and a fixedValue outlet. And in the U-file I defined a pressureInletVelocity inlet and a zeroGradient outlet. The simulation with the U- and p-bcs worked fine with simpleFoam. my k-file: Code:
/*--------------------------------*- C++ -*----------------------------------*\ my eps-file: Code:
/*--------------------------------*- C++ -*----------------------------------*\ my mut-file: Code:
/*--------------------------------*- C++ -*----------------------------------*\ my alphat-file: Code:
/*--------------------------------*- C++ -*----------------------------------*\ |
On a first view, I didn't see a mistake in your conditions (although the values for k and epsilon seems to be realy high). Could you send me a PM with a zip-file from your case? Then I will try to find out whats wrong.
Thomas |
All times are GMT -4. The time now is 13:47. |