CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   periodic boundary conditions (http://www.cfd-online.com/Forums/main/67520-periodic-boundary-conditions.html)

mranji1 August 18, 2009 01:07

periodic boundary conditions
 
Hi,
Iam trying to simulate one microorganism locomotion in a low reynolds number fluid using IBM.The microorganism is described as a sine wave.i applied periodic boundary conditions at the left and right for velocity and pressure and no gradient boundary condition at the top and bottom of the square computational domain.iam using SOR as the solver for velocity and pressure.My problem is the SOR solver for pressure is not converging because of periodic boundary condition.Can anyone suggest me how to overcome this.even i tried with by making average velocity to zero after every iteration.it also does not work.Reg.this i need help and suggestions from any one.
Thanking you
Sincerely,

aatman August 24, 2009 12:11

Quote:

Originally Posted by mranji1 (Post 226639)
Hi,
Iam trying to simulate one microorganism locomotion in a low reynolds number fluid using IBM.The microorganism is described as a sine wave.i applied periodic boundary conditions at the left and right for velocity and pressure and no gradient boundary condition at the top and bottom of the square computational domain.iam using SOR as the solver for velocity and pressure.My problem is the SOR solver for pressure is not converging because of periodic boundary condition.Can anyone suggest me how to overcome this.even i tried with by making average velocity to zero after every iteration.it also does not work.Reg.this i need help and suggestions from any one.
Thanking you
Sincerely,

How are the grid cells? are regular in all three dimension? or they are stretch and looks like a sheet? In this case maybe can be helpfull to have implicit solution on the streched direction.
How many cycle of SOR do you use? You need really a lot of iteration with just sor to have convergence, because sor dumps very well high frequency disturbance but not the low one, maybe you can add a multigrid solution to your sor.
Then you say that you are using the IBM. Try to plot cell by cell your divergence, maybe it is not converging in your cell closed to the immersed surface.

mnabi August 24, 2009 18:49

SOR converges very slowly for pressure correction equation. I think, it is better to solve the pressure by Krylov space methods (like CG) or Multigrid.
If your boundary conditions for pressure are all Neuman and periodic, you will have a singular problem. to avoid this problem, you need to fix the pressure by a constant (i.e. zero) in one point.

mranji1 August 24, 2009 23:03

Hi,
Thanks for your valuable suggestions.Actually I modified the same problem by applying periodic boundary conditions in X and Y directions for both pressure and velocity.Iam using uniform grid system.Still Iam facing the problem of SOR convergence for pressure.Also the continuity is not satisfying.I found that the the pressure and velocities are becoming high at the 4 corner points ofthe boundary.Iam wondering how it is happening?Because we have boundary conditions there.Near the Immersed boundary there is no problem......can anyone guess what is happening and how I can overcome this problem.

aatman August 24, 2009 23:45

Quote:

Originally Posted by mranji1 (Post 227367)
Hi,
Still Iam facing the problem of SOR convergence for pressure.Also the continuity is not satisfying.

The Poisson eq. for pressure has exactly the role to force the continuity.
Which is the residual of your poisson solver? how many cycle are you using? if you use just sor you can require also several hundred of pressure cycle to obtain the solution


All times are GMT -4. The time now is 12:42.