CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   domain decomposition (http://www.cfd-online.com/Forums/main/81310-domain-decomposition.html)

dshawul October 22, 2010 12:24

domain decomposition
 
Hello,
I am trying to implement a non-overlaping domain decomposition solution using a dirchlet-neumman algorithm for exchanging information at ghost cells.
The velocity prediction is solved correctly but I have a problem with the pressure solution. When I try to solve the cavity flow split in to two domains at the middle,
I get very big values for the left and right domain due to the all round Neumann boundary condition for one of the domains. Since it is incompressible solver,
only relative pressures are important. The way I did it before is to finish the solution with whatever values the solver comes up,
then decrement the value of the reference pressure during post processing. How to handle this in DD I don't know.

Also it seems to work with an SOR solver which gives value of p close to 0 (the initial guess). CG solver fails with big values in the first iterations.

I would like to avoid solving a coarse problem to coordinate solutions between different solvers, which is the reason why I am trying to use dirichlet-neumman algorithm (If I understood correctly).
I am new to this field and any hints or explanations of the algorithms are very welcome.
thanks in advance


All times are GMT -4. The time now is 11:03.