https://linen.dev logo
Join Discord
Powered by
# meshing
  • f
    Therefore I set them to postive value, I also made additional baffles for the ones I do not want growth on (I rather not make them as baffles, but I thought this might help)
  • y
  • f
    The is the other ones
  • f
    Copy code
    coilInlet_to_background
        {
            nSurfaceLayers 0; 
            mergeFaces false;  
        }
        background_to_coilInlet
        {
            nSurfaceLayers 0; 
            mergeFaces false;  
        }

        coilOutlet_to_background
        {
            nSurfaceLayers 0; 
            mergeFaces false;  
        }
        background_to_coilOutlet
        {
            nSurfaceLayers 0; 
            mergeFaces false;  
        }
  • y
    alright, that's why
  • f
    It is a coil so both sides are the same
  • y
    don't define them in the layers dict
  • y
    if you explicitely set 0 layers, it prevents snappy to add layers on this patch. (including layers added on a patch next to it) If you don't define it, it won't grow layers on the patch, but it will preserve the layers added to the patch next to it. (so the layers won't collapse)
  • f
    Okay but when I did not define them as baffles (inlet/outlet) then it had the same results
  • f
    Let me try this.
  • y
    I mean remove these from your layers section
  • f
    sure! Result come in 20min
  • f
    Now I have some problems with stitching those baffles as the number of faces changed
  • f
    Copy code
    coilInlet_to_background
    {
        type            mappedWall;
        inGroups        List<word> 
3
(
wall
mappedPatch
baffles2
)
;
        nFaces          643;
        startFace       7706438;
        sampleMode      nearestPatchFace;
        sampleRegion    coil;
        samplePatch     background_to_coilInlet;
    }
    background_to_coilInlet
    {
        type            mappedWall;
        inGroups        List<word> 
3
(
wall
mappedPatch
baffles2
)
;
        nFaces          547;
        startFace       7707081;
        sampleMode      nearestPatchFace;
        sampleRegion    background;
        samplePatch     coilInlet_to_background;
    }
    coilOutlet_to_background
    {
        type            mappedWall;
        inGroups        List<word> 
3
(
wall
mappedPatch
baffles3
)
;
        nFaces          629;
        startFace       7707628;
        sampleMode      nearestPatchFace;
        sampleRegion    coil;
        samplePatch     background_to_coilOutlet;
    }
    background_to_coilOutlet
    {
        type            mappedWall;
        inGroups        List<word> 
3
(
wall
mappedPatch
baffles3
)
;
        nFaces          541;
        startFace       7708257;
        sampleMode      nearestPatchFace;
        sampleRegion    background;
        samplePatch     coilOutlet_to_background;
    }
)
  • f
    As we probably included those faces so a 
    stitchMesh coilInlet_to_background background_to_coilInlet -perfect -overwrite
    fails as cannot find points, and using`-partial` returns a segfault
  • y
    ah yeah if you need to stitch these patches you might have isues with the layers
  • f
    yes they should become internal. Have also tested using splitBaffles, mergeBaffles.
  • f
    (to later use splitMeshRegions)
  • f
    The only big reason I want this layer to be able to be extruded to the inlet patch is so I can use a flowRate boundary condition, as these cells led to weird normals which in the end results in then weird velocities (or in my case currents). I will try to add 1 layer to those inlet and outlets see what happends then, it will probably not be the best but it might be something.
  • y
    no way to avoid having to stitch the mesh?
  • f
    Nope, my simulation is something special
  • y
    I know, doesn't make things easier 😅
  • f
    Yeah I now have background, coil, conductor (which overlaps with the coil), fluid, solid(which overlaps with the conductor but not the coil). 🙂
  • f
    5 regions (In the final sim), but I can get away with using boundary layers also on the inlet
  • f
    But then I can just use the faceZone to create baffles etc (this how to do it in OF10 as you commented yourself in a post a few years ago, when i switch to of2506 then i can do it without)
  • l
    Has anyone had experience with polyDualMesh? Afaik tetra meshes give good surface and boundary layers but worse overall convergence, you can convert them to polyDualMesh but idk how long that step takes or if the surface would also be converted
  • l
    I mean look at this, Ive ran another 1k iterations and this is the velocity dist just above the surface. It looks like the flow cannot overcome the diamond cells https://cdn.discordapp.com/attachments/796072568385568808/1440008572183380081/image.png?ex=691c97fd&is=691b467d&hm=ecca543bf7a2069a8cdc3b77a384953e8e176f9a3686c6abb7d9771352b7d03d&
  • l
    Im going insane
  • f
    nice simulation of grain boundaries in solids, is there not some way to make the blockmesh conformal with the geometry.
  • m
    I have noticed before that simulation convergence for steady state differs based on cell type. What helped for me was using a lot of underrelaxation (talking 0.1). Less relaxation might work but didnt test that. Also I would set min iterations on a linear solvers to avoid some funky problems with foams residual definiton e.g. set min 5 iterations on all linear solvers.