https://linen.dev logo
Join Discord
Powered by
# meshing
  • b

    bananaboi

    11/11/2025, 3:13 PM
    ofc
  • f

    finn

    11/11/2025, 3:13 PM
    So thencyou dont go larger
  • b

    bananaboi

    11/11/2025, 3:13 PM
    thats another way to think
  • f

    finn

    11/11/2025, 3:13 PM
    But if it is efficient to make it in all directions bigger probably not
  • b

    bananaboi

    11/11/2025, 3:14 PM
    so this kinda of mesh is not good?
  • f

    finn

    11/11/2025, 3:15 PM
    No independent of the mesh itself
  • f

    finn

    11/11/2025, 3:15 PM
    I thoughtbyou were talking about the size of the field itself
  • f

    finn

    11/11/2025, 3:15 PM
    LxLxL
  • f

    finn

    11/11/2025, 3:16 PM
    Size of the blockmesh, domain size
  • b

    bananaboi

    11/11/2025, 3:21 PM
    yeah
  • b

    bananaboi

    11/11/2025, 3:21 PM
    thats what im referring to?
  • b

    bananaboi

    11/11/2025, 3:21 PM
    lemme reframe my question
  • b

    bananaboi

    11/11/2025, 3:24 PM
    im studying an mesh that is fitted to the object so let's say im doing flow around a spheroid, my outer fluid domain mesh is also a spheroid thats scalled up but usually, the length from the inlet to the body is shorter than the body to the outlet, but by scaling, all the distances will be equal and my concern is this a waste of resources? should i truncate the intial length somehow?
  • b

    bananaboi

    11/11/2025, 3:25 PM
    turns out in the paper, they used 30D everywhere, my mesh logic is also incorrect, since they use a spherid of ratio of almost 1 for the outer fluid domain since an obvisous scaling of the inside 6:1 spheroid would make no sense
  • f

    finn

    11/11/2025, 3:30 PM
    I have an idea, you can do a oval shaped. And then slice the mesh this way the inlet gets closer
  • b

    bananaboi

    11/11/2025, 3:30 PM
    i dont get you?
  • f

    finn

    11/11/2025, 3:33 PM
    Instead of a sphere create it more like an oval so elongated along the flow, and then cut part on one of the elongated sides
  • f

    finn

    11/11/2025, 3:33 PM
    You will get an egg pretty much
  • b

    bananaboi

    11/11/2025, 3:35 PM
    ohhh
  • b

    bananaboi

    11/11/2025, 3:35 PM
    yeh i get it
  • f

    finn

    11/11/2025, 3:35 PM
    New paper idea 😮
  • b

    bananaboi

    11/11/2025, 3:35 PM
    lol
  • f

    finn

    11/11/2025, 3:37 PM
    I think the sphere itself is really nice where the object itself becomes a source term. Like cooling/heating sphere
  • o

    otaolafr

    11/12/2025, 6:40 AM
    Open a new file create a cube in geom, go to smesh define gmsh with all params you want compute it the go to file/dump file and save a python script there you will see how to define all parameters.
  • f

    finn

    11/12/2025, 6:59 AM
    How about the line I create in the geometry, that Netgen wants to follow
  • o

    otaolafr

    11/12/2025, 7:29 AM
    you can not skip it is the seam of the cad model, you can set a different 1D refinement so you get elements near the edge similar to the rest. use a same max size as min size and you will get a more or less, equlateral mesh with equal triangle sizes.
  • b

    bazz

    11/12/2025, 10:17 AM
    I suck at meshing how to get good
  • f

    finn

    11/12/2025, 10:18 AM
    use AMR
  • b

    bazz

    11/12/2025, 10:18 AM
    all my homies hate AMR
  • l

    lennygo

    11/12/2025, 11:49 AM
    Wanted to ask about something: I am currently running the following: cartesianMesh (cfmesh) which uses all cores to save the mesh into the constant folder Then I have to decompose the case and only the can I run parallel mesh checks to check if the mesh is good. The decompose step takes a long time and I would like to skip it if I can. I have not been able to figure it out as Im just constantly getting errors thrown at me. Does anyone know if you can run cfmesh in parallel to immedietely decompose the mesh into the subdirectories?