Fluids

Fluids

AMG solver divergence: compressible flow with species transport

    • cfd03
      Subscriber
      HellonI am solving a high speed flow with shocks with fuel air mixing.nDetails of simulation are:nDensity solver, AUSM, Implicit, Courant No 0.05, parallel simulation, SST k omega, y+ ~1nFirst simulation with air (density=ideal gas) is done and converged solution is obtainednOnce we start species transport on (no reaction) we get an errornDivergence detected in AMG solvernInternal error at line 1754 in file 'dbns/src/rp_mstage.c' on NodePlease suggest !nAny suggestions on the mixture material properties like viscosity/density/conductivity?n
    • RK
      Ansys Employee
      Hello, nCan you please refine the mesh, check the quality and then run the simulation again like the way you ran initially? Also what are the solution methods that you are using for the simulation?nRahuln
    • cfd03
      Subscriber
      Thanks for the suggestions. Mesh is already quite refined one 886060 nodesnMesh metrics:nStructured meshnMinimum Orthogonal Quality = 9.91381e-01nMaximum Aspect Ratio = 9.35133e+02 nStretching ratio max 1.2nAnything else need to check?nSolution methods:nDensity solver, Implicit, AUSM, Green-Gauss node based, flow-SoU, TKE- FoU, Omega- FoUnThe error occurs only when species is enabled (no reaction)n
    • Rob
      Ansys Employee
      A high cell count (which 900k cells isn't, add a few more 0's) doesn't mean the area of interest is well resolved: you may be introducing some very steep gradients with the species which is then causing an issue. n
    • cfd03
      Subscriber
      Fully agree. Sorry I should have added that the cell count is high due to refinement in all regions where high gradient is expected.nThe same simulation using OpenFOAM is done with 1200 x 250 cell count as mentioned in literature.nWhat diagnostics can I use to see where the mesh is causing problem?n
    • Rob
      Ansys Employee
      If it runs have a look for odd velocity values or poorly refined gradients a little before it fails. Chances are the shock/AFR/reaction rate are very tightly coupled. n
Viewing 5 reply threads
  • You must be logged in to reply to this topic.