[shock77]

Hi,


I have currently modelling a hypersonic free stream. I have issues using flux limiters. Whenever I use sth. besides upwind, the simulation crashes:

at ??:?
[276] #1  Foam::sigFpe::sigHandler(int) at ??:?
[276] #2  ? in "/lib64/libc.so.6"
[276] #3  Foam::hePsiThermo<Foam::psiThermo, Foam::pureMixture<Foam::sutherlandTransport<Foam::species::thermo<Foam::hConstThermo<Foam::perfectGas<Foam::specie> >, Foam::sensibleInternalEnergy> > > >::calculate() at ??:?
[276] #4  Foam::hePsiThermo<Foam::psiThermo, Foam::pureMixture<Foam::sutherlandTransport<Foam::species::thermo<Foam::hConstThermo<Foam::perfectGas<Foam::specie> >, Foam::sensibleInternalEnergy> > > >::correct() at ??:?
[276] #5  ? at ??:?
[276] #6  __libc_start_main in "/lib64/libc.so.6"
[276] #7  ? at ??:?
[bcn1602:16130] *** Process received signal ***
[bcn1602:16130] Signal: Floating point exception (8)
[bcn1602:16130] Signal code:  (-6)
[bcn1602:16130] Failing at address: 0x769600003f02
[bcn1602:16130] [ 0] /lib64/libc.so.6(+0x36400)[0x2aaaafe06400]
[bcn1602:16130] [ 1] /lib64/libc.so.6(gsignal+0x37)[0x2aaaafe06387]
[bcn1602:16130] [ 2] /lib64/libc.so.6(+0x36400)[0x2aaaafe06400]
[bcn1602:16130] [ 3] /sw/eng/OpenFOAM/4/skl/gcc/OpenFOAM-4.1/platforms/linux64GccDPInt32Opt/lib/libfluidThermophysicalModels.so(_ZN4Foam11hePsiThermoINS_9psiThermoENS_11pureMixtureINS_19sutherlandTransportINS_7species6thermoINS_12hConstThermoINS_10perfectGasINS_6specieEEEEENS_22sensibleInternalEnergyEEEEEEEE9calculateEv+0x2c1)[0x2aaaaccfeb01]
[bcn1602:16130] [ 4] /sw/eng/OpenFOAM/4/skl/gcc/OpenFOAM-4.1/platforms/linux64GccDPInt32Opt/lib/libfluidThermophysicalModels.so(_ZN4Foam11hePsiThermoINS_9psiThermoENS_11pureMixtureINS_19sutherlandTransportINS_7species6thermoINS_12hConstThermoINS_10perfectGasINS_6specieEEEEENS_22sensibleInternalEnergyEEEEEEEE7correctEv+0x2a)[0x2aaaacd2186a]
[bcn1602:16130] [ 5] rhoCentralFoam[0x43328e]
[bcn1602:16130] [ 6] /lib64/libc.so.6(__libc_start_main+0xf5)[0x2aaaafdf2555]
[bcn1602:16130] [ 7] rhoCentralFoam[0x435aba]
[bcn1602:16130] *** End of error message ***
--------------------------------------------------------------------------
Primary job  terminated normally, but 1 process returned
a non-zero exit code. Per user-direction, the job has been aborted.
--------------------------------------------------------------------------
--------------------------------------------------------------------------
mpirun noticed that process rank 276 with PID 16130 on node bcn1602 exited on signal 8 (Floating point exception).
--------------------------------------------------------------------------

My Inlet-BCs are totalPressure, totalTemperature and pressureInletOutletVelocity. I use a CFL of 0.3, tried also 0.2. I have tried, minmond, vanLeer and vanAlbada.


Here are my schemes and solution dicts:

fluxScheme      Kurganov;

ddtSchemes
{
    default         Euler;
}

gradSchemes
{
    default         Gauss linear;
}

divSchemes
{
    default         Gauss linear
    div(tauMC)      Gauss linear;
}

laplacianSchemes
{
    default         Gauss linear corrected;
}

interpolationSchemes
{
    default         linear;
    reconstruct(rho) vanAlbada;
    reconstruct(U)   vanAlbadaV;
    reconstruct(T)   vanAlbada;
}

snGradSchemes
{
    default         corrected;
}

solvers
{
    "rho.*"
    {
        solver          diagonal;
    }

    "U.*"
    {
        solver          smoothSolver;
        smoother        GaussSeidel;
        nSweeps         2;
        tolerance       1e-9;
        relTol          0.0;
    }

    e
    {
        $U;
        tolerance       1e-10;
        relTol          0.0;
    }
}

Any suggestions?




Kind regards,
shock77

[dlahaye]

Run with upwind, inspect residual fields, refine mesh in regions of large residuals, and rerun with other schemes for the convective terms?

[shock77]

The residuals are very small (order of 1e-15) and I also varried the mesh resolution from 1 mm to 0.2 mm and it had no effect.

[dlahaye]

Residual norm (as in real number) or residual fields (as in something with spatial distribution)?

D.

[shock77]

Hi,


the normal residuals you get while computing.

[dlahaye]

Possibly you gain more information by looking into the spatial distribution of the residual using

1.1/ In system-folder/controlDict-file

…
writeInterval 10;
…

functions
{
#includeFunc residualFieldDict
}

1.2/ In system-folder/residualFieldDict-file

type solverInfo; // specific for version-1906
libs ("libutilityFunctionObjects.so");

writeControl runTime; // timeStep;
writeInterval 10; // make sure this to be equal as the writeInterval specified in the controlDict file

writeResidualFields yes;

[shock77]

Hi,


I think I have those instabilites because of the strong expansion of the gas.
The simulation crashes while solving the energy equation, although in the previous timestep the residuals are ok.


So I think the solution is to optimize the schemes.


I am trying this now as stable solution:

fluxScheme      Kurganov;

ddtSchemes
{
    default         Euler;
}

gradSchemes
{
    default         faceLimited Gauss linear 1;
}

divSchemes
{
    default         none;
    div(tauMC)      Gauss linear;
    div(phi,U)     Gauss linearUpwindV grad(U);
}

laplacianSchemes
{
    default         Gauss linear uncorrected;
}

interpolationSchemes
{
    default         linear;
    reconstruct(rho) Minmod;
    reconstruct(U)   MinmodV;
    reconstruct(T)   Minmod;
}

snGradSchemes
{
    default         uncorrected;
}

wallDist
{
    method meshWave;
}

Any other suggestions to make it more stable?