[dlahaye]

What does

"rho.*"
{
solver nthn;
}

in e.g. tutorials/compressible/rhoPimpleFoam/RAS/squareBendLid/system/fvSolution refer to?

Thx, Domenico.

[dlahaye]

Anybody? Thx!

[tomf]

Hi Domenico,

I think it is just a placeholder in the sense that the rhoPimpleFoam solver needs a solver entry for rho or rhoFinal, but internally it uses always the diagonal solver. I just changed the nthn solver into xnthn and it also runs.

Using a grep nthn from my OpenFOAM installation folder only gives this output:

~/OpenFOAM/OpenFOAM-v2012>grep -r nthn .
./tutorials/compressible/rhoPimpleFoam/RAS/externalCoupledSquareBendLiq/system/fvSolution:        solver nthn;
./tutorials/compressible/rhoPimpleFoam/RAS/squareBendLiq/system/fvSolution:        solver nthn;

With diagonal or GAMG I can at least find references in code using grep.

Best Regards,
Tom

[dlahaye]

Dear Tom,

Thank you for your reply.

My understanding is that rhoPimpleFoam is a pressure-based solver and that the density is updated from the pressure using the equation of state implemented in the thermodynamics as explained e.g. in this post [1].

A solver for density thus seems to be superfluous. Hence my confusion.

Could you please elaborate? Thanks, Domenico.

[1] How rho_ and psi_ are calculated in compressible solvers of OpenFOAM?

[tomf]

Dear Domenico,

I have to say I only tried to find any reference to the nthn part, but I just looked at the solver which reads for v2012:

rhoPimpleFoam.C:

            if (pimple.firstIter() && !pimple.SIMPLErho())
            {
                #include "rhoEqn.H"
            }

Based on the Make/include file I think that this rhoEqn.H is from

src/finiteVolume/cfdTools/compressible/rhoEqn.H:

{
    fvScalarMatrix rhoEqn
    (
        fvm::ddt(rho)
      + fvc::div(phi)
      ==
        fvOptions(rho)
    );

    fvOptions.constrain(rhoEqn);

    rhoEqn.solve();

    fvOptions.correct(rho);
}

and this solves the continuity equation for rho. I think this means that we will have mass conservation in cases there is a net influx or outflux into a cell, or is there is some source in fvOptions that creates or destroys mass. As the call to phi is explicit (fvc:div(phi)), I think this will result in a simple vector equation, or a pure diagonal matrix.

As to why this is necessary, I would have to spend more time than I have available I am afraid. I can imagine it has something to do with mesh motion at least and mass conservation as described above.

Best Regards,
Tom

[dlahaye]

Dear Tom,

This is very valuable information to receive. Sincere thanks!

Domenico.

[Bloerb]

I am pretty sure nthn is short for nothing. Hence no solver specified.
An easy way to see if you need to define something is to comment out the entire block:

/*

"rho.*"
    {
        solver nthn;
    }
*/

If you run your solver now you'll either get an error message that a definition for rho and maybe rhoFinal is missing, or nothing needs to be specified.
You can also try

"rho.*"
    {
        solver banana;
    }

Since there is definitively no banana solver, you'll get a list of all possible solvers.