[derekm]

When using SemiImplicitSource in FVoptions on p, li { white-space: pre-wrap; } chtMultiRegionSimpleFoam solver with field h, I became aware its effect was dependent on the timestep length.
e.g.

    scalarSemiImplicitSourceCoeffs
        {
            volumeMode      specific;
            injectionRateSuSp
            {
                h           (  20000 0 ); // 2e4 
            }
        }

injects 2E4 joules per m3 per deltaT thus if you double the deltaT you will halve the Watts per m3

Please can someone confirm that this is the correct interpretation of the solveSolid.h and the resulting action is the intended one.

{
    for (int nonOrth=0; nonOrth<=nNonOrthCorr; nonOrth++)
    {
        fvScalarMatrix hEqn
        (
            (
                thermo.isotropic()
                ? -fvm::laplacian(betav*thermo.alpha(), h, "laplacian(alpha,h)")
                : -fvm::laplacian(betav*taniAlpha(), h, "laplacian(alpha,h)")
            )
          ==
            fvOptions(rho, h)
        );

        hEqn.relax();

        fvOptions.constrain(hEqn);

        hEqn.solve();

        fvOptions.correct(h);
    }
}

thermo.correct();

Info<< "Min/max T:" << min(thermo.T()) << ' ' << max(thermo.T()) << endl;

[derekm]

ignore the above ... it turns out to be slow convergence in the solid regions
now at least partially fixed by changing nNonorthogonalCorrectors
and relaxationFactors in fvSolutions.