[hogglife]

Hi,

I want to perform a conjugate heat-flow simulation and thus needs to mesh multi regions, including fluid and solids.

A simplified case is attached which includes a chamber and a plate, as well as all needed files to run.

Strange cells are generated by snappyHexMesh at the interface between the chamber and the plate. These strange cells belong to the chamber and is a thin layer. In addition they seems not connected to the lower cells of chamber.

Meanwhile besides the /chamber/patch/chamber_to_plate, there is an extra patch with the name of "plate" under /chamber/patch/ in paraView. ParaView shows this patch is composed of double layers of surface but in original stl surface "plate" it is a single layer surface.

In the attached figures, all cliping is done at (0,0,0) with normal of (0,1,0).

The version of openFoam I used is v2206.

Any idea is appreciated!

Thank you in advance!

[hogglife]

The rest figures to be uploaded.

[hogglife]

Further investigation shows after the 1st step of snappyHexMesh, the original stl surface of plate becomes a flat patch, i.e. two layers of cells overlap with each other. This patch "plate" locates at the lower surface of original stl surface of plate. The two overlapped layers behaves as a boundary in cells of chamber, i.e. the cells of chamber adjacent the two layers in patch of plate do not connect to each other, and leads to the split in 2nd step of snappyHexMesh.

Can anybody give advice?

I guess the problem relates to the fact the region "plate" is completely in the region "chamber" which causes some issue when assigning cells to region.

[hogglife]

Trials found after changing the number of cell along x, y, and z in background mesh the problem vanishes, e.g.

original: (with issue at interface)

    xCells    172;
    yCells    172;
    zCells    115;

new: (without the mentioned problem)

    xCells    180;
    yCells    180;
    zCells    120;

Though changing numbers of cell along x, y, and z can solve problem in this simple geometry, for my complex geometry using this trial-and-error method may take long time.

Does anybody know the reason? Or is it a bug in snappyHexMesh?