[Astan]

Hi guys, i would kindly ask for an advice, i don't know how to solve this stability problem.

I'm running an adjoint simulation on a car.

For the first 500 iteration, the simulation goes in the " right " direction, meaning that the all residuals decrease and the sensitivity does not exhibit any " crazy " behaviour.

From the 500th iteration on, the adjoint pressure residual does not decrease anymore and starts to increase in a monotonic way and so does the continuity error, too.

The checkmesh is ok, the maximum non orthogonality of the mesh is 64.44°.

I would kindly ask if you have any suggestion to solve this problem, the fvscheme and fvsolution are the following:

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  5                                     |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      fvSchemes;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
// Adj

ddtSchemes
{
    default         steadyState; 
}

gradSchemes
{
    default         cellLimited Gauss linear 1;
    grad(Ua)        cellLimited Gauss linear 1;
}

divSchemes
{
    default         none;
    div(-phi,Ua)    Gauss upwind;
    div((nuEff*dev2(T(grad(Ua))))) Gauss linear;
}

laplacianSchemes
{
    default         Gauss linear limited 0.777; 
}

interpolationSchemes
{
    default         linear;
}

snGradSchemes
{
    default         limited 0.777;
}

fluxRequired
{
    default         no;
    p;
    pa;
}


// ************************************************************************* //

// ************************************************************************* //

wallDist
{
   method meshWave;
}
// ************************************************************************* //

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  5                                     |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      fvSolution;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

solvers
{
    "(pa)"
    {
        solver          GAMG;
        tolerance       1e-6; // up 900
        //tolerance       1e-8; up 4300
        //tolerance       1e-9;
        relTol          0.001; 
        smoother        GaussSeidel;
        cacheAgglomeration true;
        nPreSweeps      0;
        nPostSweeps     2;
        nCellsInCoarsestLevel 500;
        agglomerator    faceAreaPair;
        mergeLevels     1;

    }

    "(Ua)"
    {
        solver          PBiCGStab;
        preconditioner  DILU;
        tolerance       1e-8;// up 4300
        //tolerance       1e-9;
        relTol          0.0;
    }

}


SIMPLE
{
    nNonOrthogonalCorrectors 	2;
    pRefCell        		0;
    pRefValue       		0;

    residualControl
    {
       pa               1e-8; 
       Ua               1e-8; 
    }

}

relaxationFactors
{
    fields
    {
        pa                 0.05;
    }
    equations
    {
        Ua                 0.05; // lowering it can solve solver instability at startup
    }
}

// ************************************************************************* //

Astan.

Thanks you for the time spent in this thread, any suggestions are really appreciated