On asking about the parallel version

2009-03-10
2013-04-25
  • Sungwook Lee
    Sungwook Lee
    2009-03-10

    Dear Developper of OpenFVM,

    I'm still struggling with the code for my purpose(making Overset grid solver). Here is my question for the parallel version of OpenFVM.

    1. I've got serious problem when the velocity,especially the cell which has PROCESSOR b.c., was corrected with the pressure field obtained from the continuity equation. There is a inconsitency at the cell with PROCESSOR b.c.. Because of my purpose, the coefficient matrix for continuity eq. was solved with asymmetric matrix solver(disabled symmetric condition for continuity matrix). The result were check with tecplot (vertex value) because I could not see the strange trends with cell centered value. In addition, I've already check single block overset solver works fine (good consistency and conversency).
    Whenever I decomposed the backgroud grid and body fitted grid, the situation is getting worse.

    With the regards to the above question, I've check pressured field but both results gave almost same results. Therefore, I'm doubting that the problem was occured in the calculation of gradient of the pressure to correct the velocity at cell center(Especially, the cell with PROCESSOR faces).

    Do you have any idea to fix or get more better gradient ?

    Actually, I'll try to calculate the gradient with the least square approach to obtain the gradient at the cell with PROCESSOR b.c..

    P.S.: I wanted to attach my strange result to discuss with you. But I couldn't know how to attach these. If you know it, inform me.

    Thanks in advance

     
    • Sungwook Lee
      Sungwook Lee
      2009-03-10

      I've forgot describing what the calcuation case was.

      I've tested lid-driven cavity with Re=1000. At this moment, I've disabled all other equation solvers except the momentum and contiuity.

      The test cases are as follows :

      1. Multi-block case : 4 background grids with 4 overset grids(body fitted).
         The length of overset grid is half of the cavity length and located at (0.5,0.5). Both grid(background and overset) have same cell size each other.
      At least, three cells layer was overlapped each other.

      2. Sing case : 1 background grid and 1 overset grid. Others are same with the multi-block case.

      As reference, all matrices for the momentum and continuity are strongly coupled with the relationship between the donor cells and inter-communication cells.

      Thanks again,

       
    • Billy Araujo
      Billy Araujo
      2009-03-10

      Hi,

      Can you send me the input files?

      email: x-flow - a.t - users.sourceforge.net