[Adam_K]

Can the compressible::turbulentTemperatureCoupledBaffleMix ed boundary condition between two regions handle no resistance between two regions of the same thermal conductivity? I am using the chtMultiRegionFoam solver on OpenFoam 5.

I want to simulate heat flow through a composite slab that is comprised of cylinders (regularly spaced). The analytical solution for a slab of a given thermal diffusivity is known (slab initially at 300K and then for t>0s, the temperature on the two walls is 310 K). So I thought that would be a good limiting case to test. When the inclusions have the same thermal diffusivity as the matrix and there is no interfacial thermal resistance, the response should be the same as a homogeneous slab.

Using four different meshes, my results are surprising as the temperature inside a fiber at the center of the slab is lower than the matrix. With no thermal resistance at the interface and the same conductivities, this doesn't make sense. (figure)

When I include a thermal resistance, the results make more sense. For high thermal resistance (green), the fibre temperature (dashed line) lags behind that of the matrix (solid line). As the resistance decreases, the fibre temperature better follows the matrix temperature, a logical explanation. (figure)

I spoke with a colleague and we're thinking that the zero resistance interface between to regions of the same material may be a poor limiting case to check as it's not really their destined application. Is this a known issue with Open Foam 5?

[pete20r2]

Try with many more loops per timestep, that BC tries to match the T value and flux and the patch from 2 regions explicitly. This is quiet slow.
My point is you may not be converged.

[Adam_K]

Originally Posted by pete20r2

Try with many more loops per timestep, that BC tries to match the T value and flux and the patch from 2 regions explicitly. This is quiet slow.
My point is you may not be converged.

Normally, this would require changing the PIMPLE algorithm options or the tolerance, correct?

PIMPLE
{
    nNonOrthogonalCorrectors 0;
    nCorrectors          0;
    nOuterCorrectors    0; 
}

It seems like the PIMPLE options are more useful in fluid flow cases where the momentum/pressure equations are updated. As I am only working with solids, then I should just use much lower tolerances.

solvers
{
    h
    {
        solver           PCG;
        preconditioner   DIC;
        tolerance        1e-7;
        relTol           0;
    }

    hFinal
    {
        $h;
        tolerance        1e-7;
        relTol           0;
    }
}