[Diro7]

Hello to everyone,

I'm trying to simulate conjugate heat transfer with the chtMultiRegionSimpleFoam solver.
My case consists in 2 fluid channels that exchange heat in counterflow through a solid matrix (you can see a section view of the solid region in the first picture).
The channels are small in section (diameter is about 2-3 mm) but very long (2.5 m).

The inlet and outlet BCs for the fluids are quite straightforward to assess, as well as the ones for the 2 fluid/solid coupling surfaces.

Regarding temperature in the solid region, since the heat exchanger is supposed to consist in a very large number of channels, I set zeroGradient BCs on the left/right faces (due to pattern symmetry) and cyclicAMI BCs on the top/bottom faces (due to the pattern periodicity).
The only "arbitrary" choice is the BC on the front/back solid faces (i.e. the ones where the fluid inlets and outlets lie). I couldn't find any reasonable boundary condition (from the physical modelling viewpoint), so I decided to use zeroGradient also here.

The simulation after 1500/2000 steps seems to reach convergence (residuals are very low for each quantity) and most of the results seem to be in agreement with predictions (fluid T differences, total heat transfer rates), but something is not convincing:

1) In the second picture you can find a T plot over an axial line in the solid region (at the exact center of the section; you can see the point in the first picture): the temperature derivative along the axial direction at the surfaces is not 0, in contrast with the zeroGradient BCs!
I tried different meshes (I use the Sweep method of the Fluent meshing tool) with different axial refinements but, as far as I know, the results are always quite the same.

2) Also, as you can see, the four corners show temperature values that are quite probably wrong (much higher/lower than the rest of the section).

Did anyone experience the same issues, or does know a possible explanation?
Should I use different BCs, or try to improve the mesh?

Thank you very much in advance for the help.

Andrea

PS: if you are wondering, the temperature field in the first picture is taken from a section exactly at half the channel length.

EDIT1: I added a picture of the mesh over the face.

EDIT2: it seems that the strange values at the corners were only a matter of visualisation in ParaView. I'm not an expert on how it works, but I realised that when plotting the original Cell Field rather than the interpolated (or smoothed, I don't know how to say it) Point Field the corner values are ok. Probably - and reasonably - in order to show correct values ParaView should account for the cyclic-ness of the boundary conditions, but it can not and some plotting error occurs.

[Diro7]

No hints?