[hjxy2012]

Is there anyone writing cfd codes? I wrote the pressure outlet boundary for dolfyn(an open source cfd code) but it just got divergence. The mass conservation equations could not get convergence. Is there any tricks for this problem? The theory guide is the book by Ferziger Peric--"Computational Methods for Fluid Dynamics". Thank you!
https://www.dolfyn.net/index_en.html

[arjun]

Originally Posted by hjxy2012

Is there anyone writing cfd codes? I wrote the pressure outlet boundary for dolfyn(an open source cfd code) but it just got divergence. The mass conservation equations could not get convergence. Is there any tricks for this problem? The theory guide is the book by Ferziger Peric--"Computational Methods for Fluid Dynamics". Thank you!
https://www.dolfyn.net/index_en.html

Cases with pressure outlet also can diverge in commercial code too. So if your calculation diverged on off then i am not sure what your question is.

If your question is that it diverged all the time, then makes sense.

There is nothing special in pressure-outlet condition. In fact it is preferred condition because it adds Dirichlet condition and makes linear system converge faster.

I have written iNavier in past and now writing wildkatze FVUS and pressure outlet works good. Don't have complain so far.

[hjxy2012]

Originally Posted by arjun

Cases with pressure outlet also can diverge in commercial code too. So if your calculation diverged on off then i am not sure what your question is.

If your question is that it diverged all the time, then makes sense.

There is nothing special in pressure-outlet condition. In fact it is preferred condition because it adds Dirichlet condition and makes linear system converge faster.

I have written iNavier in past and now writing wildkatze FVUS and pressure outlet works good. Don't have complain so far.

Thank you. I am checking my code now. And it will be better if there is some codes to refer to. The source code of iNavier can not be downloaded now. I am using Windows 10 and the exe can not execute. Would you please send me the source code of iNavier if it is proper? Thank you very much.

[hjxy2012]

Originally Posted by arjun

Cases with pressure outlet also can diverge in commercial code too. So if your calculation diverged on off then i am not sure what your question is.

If your question is that it diverged all the time, then makes sense.

There is nothing special in pressure-outlet condition. In fact it is preferred condition because it adds Dirichlet condition and makes linear system converge faster.

I have written iNavier in past and now writing wildkatze FVUS and pressure outlet works good. Don't have complain so far.

Oh, I found the problem. The coefficient of the pressure correction equations was wrong. It worked well after fixing it. Anyway, thank you for helping me. The pressure outlet bc was just as easy as you said.

[arjun]

commercial solvers also add dynamic head in case of reversed flow. This discourages reversed flow that can make solver instable. This is optional in FVUS (and in starccm), in fluent I am not sure if you can switch it off but by default it is on there.

Here is code that does in FVUS wildkatze

void
PBasedFlowModelEquation::
BoundaryConditionTerm::
addDynamicHead(
    FVFaces& fvFaces, PhaseContinuum& phase, StorageAndMemoryManager& stManBnd,
    ObjectOptions& objOptions)
{

  StorageAndMemoryManager * stMan  =  getSimulation().getStorageAndMemoryManager( phase, getFiniteVolume());

  double *ax = fvFaces.areaX() ;
  double *ay = fvFaces.areaY() ;
  double *az = fvFaces.areaZ() ;

  double *xVel    = stMan->getDoubleVar("x-velocity");
  double *yVel    = stMan->getDoubleVar("y-velocity");
  double *zVel    = stMan->getDoubleVar("z-velocity");
  double *pres    = stMan->getDoubleVar("pressure"  );
  double *density = stMan->getDoubleVar("eff-density");

  int const nfaces  = fvFaces.size      ();
  int const    *CL  = fvFaces.cellsLeft ();
  int const    *CR  = fvFaces.cellsRight();

  int    cl   = 0;
  int    cr   = 0;

  for (int f = 1; f <= nfaces; f++)
    {
      cl     =     CL[f];
      cr     =     CR[f];

      double AdotVel    =    xVel[cr] * ax[f] + yVel[cr] * ay[f] + zVel[cr] * az[f];

      if (AdotVel < 0.0)
        {
          pres[ cr ]    -=    0.5*density[ cr ]*( xVel[cr] * xVel[cr] + yVel[cr] * yVel[cr] + zVel[cr] * zVel[cr] );
        }
    }
}