[be_inspired]

Hi all,

My question is related to the utility simpleFoamResidual implemented into OF1.6ext. According to the code, it analyze the momentum equation. The perfect situation, I think, would be that uResidual is equal "0".

The attached convergence is OK but the maxium uResidual that I obtain is around 76 and the mean around 0.09. In other different mesh, the maximum uResidual obtained was around 2000.

Could anyone tell me how much is a high value of uResidual? Is possible to have a physical meaning about the uResidual? If I have higher values than 0.05, for example, does it tell me that it is require to refine the mesh?


I would appreciate any help. I would like to have a criteria to know if current mesh resolution is enough or not for my purposes without doing a mesh resolution analysis in detail. Thank you very much.

            Info<< "    Calculating uResidual" << endl;
            volScalarField uResidual
            (
                IOobject
                (
                    "uResidual",
                    runTime.timeName(),
                    mesh,
                    IOobject::NO_READ
                ),
                mag
                (
                    fvc::div(phi, U)
                  + fvc::div(turbulence->R())
                  + fvc::grad(p)
                )
            );

Time = 3000

DILUPBiCG:  Solving for Ux, Initial residual = 1.48121e-08, Final residual = 6.66887e-11, No Iterations 2
DILUPBiCG:  Solving for Uy, Initial residual = 1.09835e-05, Final residual = 1.32954e-08, No Iterations 2
DILUPBiCG:  Solving for Uz, Initial residual = 1.96974e-05, Final residual = 3.96294e-08, No Iterations 2
GAMG:  Solving for p, Initial residual = 0.00085226, Final residual = 4.08688e-05, No Iterations 8
time step continuity errors : sum local = 2.32144e-10, global = -6.39797e-11, cumulative = 2.45607e-07
DILUPBiCG:  Solving for epsilon, Initial residual = 2.62941e-07, Final residual = 2.54438e-10, No Iterations 1
DILUPBiCG:  Solving for k, Initial residual = 9.73059e-07, Final residual = 3.07595e-10, No Iterations 2
ExecutionTime = 7790.51 s  ClockTime = 7792 s

[akidess]

I don't think this tool can replace a mesh sensitivity study. All it tells you is that a solution is converged, not that it is converged and accurate.

[be_inspired]

I am trying to link Dr.Jasak Thesis about error estimation with my case and to know if it is needed to refine the mesh once convegence has been "obtained" (residuals).

Chapter 5 is dedicated to the adaptative mesh refinement to reduce the numerical error up to the desired level. There should be a criteria ( I hope that) about when the error (uResidual) has been reduced enough to minimize the numerical error.