[sabago]

Dear OFers,

I have a case that I know is ok in terms of geometry as I have run other solvers on it and checkMesh is OK.

I made a change somewhere (in the charge density equations) in my solver and it compiled fine, however, I get an error after one iteration at a different point from where I made the edit (e.g in the diffusion equations). Below is the error

#0  Foam::error::printStack(Foam::Ostream&) at ??:?
#1  Foam::sigFpe::sigHandler(int) at ??:?
#2   in "/lib/x86_64-linux-gnu/libc.so.6"
#3  double Foam::sumProd<double>(Foam::UList<double> const&, Foam::UList<double> const&) at ??:?
#4  Foam::PCG::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const at ??:?
#5  Foam::fvMatrix<double>::solveSegregated(Foam::dictionary const&) at ??:?
#6  Foam::fvMatrix<double>::solve(Foam::dictionary const&) at ??:?
#7  
 at ??:?
#8  
 at ??:?
#9  __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#10  
 at ??:?
Floating point exception (core dumped)

My fvScheme is as follows

ddtSchemes
{
    default         Euler;
}

gradSchemes
{
    default         Gauss linear;
    grad(p)         Gauss linear;
}

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

laplacianSchemes
{
    default         none;
    laplacian(DH2,H2) Gauss linear corrected;
    laplacian(DT,H2O) Gauss linear corrected;
    laplacian(DH2Oeff,H2O) Gauss linear corrected;

    laplacian(Tussle,phi) Gauss linear corrected;
    
}

interpolationSchemes
{
    default         linear;
    interpolate(HbyA) linear;
}

snGradSchemes
{
    default         corrected;
}

fluxRequired
{
    default         no;
    p               ;
}

and my fvSolution

solvers
{H2O
    {
        solver          PCG;
        preconditioner  DIC;
        tolerance       1e-06;
        relTol          0;
    }
    
    H2
    {
        solver          PCG;
        preconditioner  DIC;
        tolerance       1e-06;
        relTol          0;
    }
    phi

    {
        solver          smoothSolver;
	smoother	GaussSeidel;
	//smoother	DICGaussSeidel;
        tolerance       1e-06;
	nSweeps 	1;
        relTol          .99999999999999;//.999969;//.99995;//.99999;
    }

}
SIMPLE
{
    nNonOrthogonalCorrectors 0;
}

I changed the fvSolution to this (below) for the calculation of H2O diffusion where I was getting the error but I still get the DICPCG error! even when I change the fvSolution!

H2O
    {
         solver           GAMG;
        tolerance        1e-12;
        relTol           0.1;

        maxIter         20;    

        smoother         DIC;
    
        nPreSweeps       1;
        nPostSweeps      1;
        nFinestSweeps    2;

        cacheAgglomeration true;

        nCellsInCoarsestLevel 1000;
        agglomerator     faceAreaPair;
        mergeLevels      1;
    }

I have spent days trying the suggestions elsewhere on the forum like checking mesh, reducing timeStep to reduce Courant number and BC (case works fine with one line of code changed in the solver) and tried playing with the fvSchemes and fvSolution; what I'm I doing wrong/missing?

Best,
Sandra

[sabago]

Hi again OFers,

At this point I just want the changes that I make in fvSolution to take!

I changed from PCG to GAMG solver but I'm still getting a "Foam::PCG::solve" error.

Any ideas will be much appreciated.

Best,
Sandra

[wyldckat]

Hi Sandra,

Originally Posted by sabago

I made a change somewhere (in the charge density equations) in my solver and it compiled fine, however, I get an error after one iteration at a different point from where I made the edit (e.g in the diffusion equations). Below is the error

#0  Foam::error::printStack(Foam::Ostream&) at ??:?
#1  Foam::sigFpe::sigHandler(int) at ??:?
#2   in "/lib/x86_64-linux-gnu/libc.so.6"
#3  double Foam::sumProd<double>(Foam::UList<double> const&, Foam::UList<double> const&) at ??:?
#4  Foam::PCG::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const at ??:?

The problem likely has nothing to do with the parameters you defined in "fvSolution". The major problem is that "sigFpe" refers to this: https://en.wikipedia.org/wiki/Unix_signal#SIGFPE - the usual reason is a division by zero, but given that it occurred in the operation "sumProd", my guess is that there is one value that is either NaN (not-a-number) or INF (infinite).

You need to retrace your steps back to the change you made in the source code. Whatever you did, no matter how innocent it might have seemed, it's likely the origin of this problem.

When in doubt, old-school-debugging is one way to isolate the breaking point. Quoting myself from another post:

Originally Posted by wyldckat

First thing you should do: isolate and conquer. There is a wiki page on the topic of debugging: http://openfoamwiki.net/index.php/HowTo_debugging
And recently I wrote a post with an example on how to do old-school debugging, namely to output as much information as possible, to see what's truly happening: http://www.cfd-online.com/Forums/ope...tml#post512967 post #9

I don't have much time to ask for more details, so the best i can do is to tell you for you to look for the details yourself
More specifically, try to see what exact values do the more relevant fields have before the crash. For example:

Info << "flu_: " << flu_ << endl;

Best regards,
Bruno

[fabian_roesler]

From your log I can see that your simulation is running serial an not parallel. However, I sometimes have this kind of error when some boundaries or processor patches have no value defined. Since this happened a couple of times I always set the boundary values as initial condition.

myPatch
{
   type    someType;
   value   $internalField;
}

Cheers

Fabian

[sabago]

Thanks for your input, Bruno and Fabian,

I'll try your suggestions.

Best,
Sandra

[xoitx]

Hello Sandra
Did you find out the cause?
I have the exact same error message.. I guess it's a division by 0 somewhere can't figure out where.
Thanks

[ Moderator note: This forum member later started a new thread on this topic here: http://www.cfd-online.com/Forums/ope...erdymfoam.html ]