[boffin5]

My chtMultiRegionFoam case involves a radiator in a fairing duct, hung beneath a fuselage. Recently, with a complicated fairing geometry, I nearly got it to run, but at time step 1, it locked up. So I embarked on a series of simplified cases to try to find the source of the problem. Attached are an image of the geometry for the first case, along with a slice. Also a zip file of the case.

For this simplified geometry case, I couldn't even get it to time step 1, rather it gave me the message "plane normal defined with zero length." In this version, there is an exact face-to-face interface between the radiator and the duct. Apparently, OpenFoam is having trouble with the mesh interface between the two regions. I found a promising option in this thread: mapFields error: Plane normal defined with zero length

So I tried the suggestion of changing fvOptions, such that cellVolumeWeight is now mapNearest. The result of this was that the simulation froze at this point in the solver:

Creating finite volume options from "constant/fvOptions"

Selecting finite volume options model type constantHeatTransfer
    source solidTofluid
- selecting iter region mapping
Creating mesh-to-mesh addressing for solid and fluid regions
    using mapNearest

My next attempt with simplified geometries was to shrink the duct such that there is a slight overlap between the duct and the edge of the radiator. But it made no difference.

All these options are with surfaces open at the inner side, since I use mirrorMesh to create closed surface. In the version related above, the fairing surface is open at the top edge, where it abuts the fuselage surface. Other options will be with the top edge of the fairing closed, and one with the fairing and fuselage being all one surface. But I need to get the first one to run (or actually any of them, in case only one is valid). But importantly, a similar case to the first version ran successfully as a simpleFoam case, with no radiator.

At this point, I can only posit 2 theories as to the problem:
1. Perhaps in fvSolution there are some tolerances in the solver that I can change. I could try this, but it would be like shooting in the dark. I need some guidance on this.
2. Or, there may be an OpenFoam function that I need to use to relate the 2 regions together. But I don't have knowledge of this, and haven't seen anything in tutorials.

As ever, there is a solution for this problem, but after months of struggle, I haven't been able to find it. I would truly appreciate help from any of the community members who are far smarter than me.