[anthonygay1812]

Currently getting the error

--> FOAM FATAL ERROR: 
The search for an edge-connected face and tet-point failed.

    From function void Foam::particle::changeFace(Foam::label)
    in file particle/particle.C at line 320.

FOAM exiting

I'm trying to make a 1D sod shock simulation and solve it with a modified DSMC solver. I'm pretty new to OpenFoam and CFD as a whole and dont have much of an understanding about whats happening in the particle class to cause this.

Currently I'm running a modified DSMC Initialise function from the DSMCCloud class along with a modified dsmcInitialiseDict and blockMeshDict. All the other dictionaries are from the DSMC: freeSPacePeriodic tutorial.

DSMC Initialise class:https://github.com/Anthony-Gay/HPC_w...DSMCCloudMod.C
blockMeshDict: https://github.com/Anthony-Gay/HPC_w.../blockMeshDict
DSMCInisialiseDict: https://github.com/Anthony-Gay/HPC_w...InitialiseDict

My current hypothesis is that its how my tracking data (td) in the DSMCCloud::evolve function is being constructed since the error comes from line 1046

// Move the particles ballistically with their current velocities
    Cloud<ParcelType>::move(*this, td, mesh_.time().deltaTValue());

However I dont really know how to get the actual tracking data to print to my logfile.

When I dig into the particle class i find that its this loop thats triggering the message

tetPti_ = -1;
    forAll(mesh_.cells()[celli_], cellFaceI)
    {
        const label newFaceI = mesh_.cells()[celli_][cellFaceI];
        const class face& newFace = mesh_.faces()[newFaceI];
        const label newOwner = mesh_.faceOwner()[newFaceI];

        // Exclude the current face
        if (tetFacei_ == newFaceI)
        {
            continue;
        }

        // Loop over the edges, looking for the shared one. Note that we have to
        // match the direction of the edge as well as the end points in order to
        // avoid false positives when dealing with coincident ACMI faces.
        const label edgeComp = newOwner == celli_ ? -1 : +1;
        label edgeI = 0;
        for
        (
            ;
            edgeI < newFace.size()
         && edge::compare(sharedEdge, newFace.faceEdge(edgeI)) != edgeComp;
            ++ edgeI
        );

        // If the face does not contain the edge, then move on to the next face
        if (edgeI >= newFace.size())
        {
            continue;
        }

        // Make the edge index relative to the base point
        const label newBaseI = max(0, mesh_.tetBasePtIs()[newFaceI]);
        edgeI = (edgeI - newBaseI + newFace.size()) % newFace.size();

        // If the edge is next the base point (i.e., the index is 0 or n - 1),
        // then we swap it for the adjacent edge. This new edge is opposite the
        // base point, and defines the tet with the original edge in it.
        edgeI = min(max(1, edgeI), newFace.size() - 2);

        // Set the new face and tet point
        tetFacei_ = newFaceI;
        tetPti_ = edgeI;

        // Exit the loop now that the tet point has been found
        break;
    }

    if (tetPti_ == -1)
    {
        FatalErrorInFunction
            << "The search for an edge-connected face and tet-point failed."
            << exit(FatalError);
    }

From a conceptual standpoint I really dont know whats going on inside this loop. I have tried running a chekmesh but everything came back cosher and I have used my current geometry with the original solver and it worked fine.

Any insight is appreciated! Thank you!

[HPE]

More info is needed for forum dudes, so that people can replicate ur issue - and help u eventually.

[anthonygay1812]

Thank you HPE, I've added more information. I think my real issue is I'm not yet effective with debugging in C++ on WSL (I've spent several years working with just Matlab) and still need to work on my skill toolbox.

[anthonygay1812]

I managed to find the issue. Once again being new to C++ in general and not knowing much about compilers, I was compiling everything as a super user. I accidentally ran a case as a super user, meaning that when I ran my all clean script as a regular user, files in my 0 directory weren't being deleted/overwritten.