[immortality]

I have set tolerances like this(1e-9):

U
    {
        solver          smoothSolver;//smoothSolver
        //preconditioner  DIC;
        smoother        DICGaussSeidel;
        nSweeps         1;
        tolerance       1e-9;
        relTol          0;
        maxIter        50000;
    }

    h
    {
        $U;
        //tolerance       1e-10;
        //nSweeps         10;
        //relTol          0;
        //maxIter         25000;
    }

is it sufficient for variables that their residuals to reach 1e-9 generally for unsteady compresible cases?or lower can be acceptable?(because it takes too long for starting the first iteration)

[wyldckat]

Hi Ehsan,

After some searching, I found the following threads:

• http://www.cfd-online.com/Forums/ope...vsolution.html
• http://www.cfd-online.com/Forums/ope...d-solvers.html
• http://www.cfd-online.com/Forums/ope...nvergence.html
• http://www.cfd-online.com/Forums/ope...-problems.html
• http://www.cfd-online.com/Forums/ope...am-solver.html

After reading parts of these threads, my conclusion is this:

1. Depends on your case: the mesh, properties of the flow and so on.
2. Either you have experimental data for comparison, or you study the effect of changing the tolerance values.
3. A tolerance value too low can lead to too much computing time iterating, which might lead to no improvement.
4. A value too high, will lead to a bad solution and possibly it will eventually diverge/crash.
5. From these threads, "good enough" values could be found scattered around for steady-state cases, but I did not find any indication for unsteady compressible flow.

Best regards,
Bruno