CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Pressure Dirichlet value in BiCG STAB (https://www.cfd-online.com/Forums/main/180462-pressure-dirichlet-value-bicg-stab.html)

jawapasu November 23, 2016 23:56

Pressure Dirichlet value in BiCG STAB
 
I am solving the pressure poisson equation in BiCG STAB, I am using finite element formulation. the matrix is like this AX=B. Where A is the matrix where each row represents the weight of the surrounding nodes of that node considered(can be taken as the center node which is unknown). X is the unknown node phi and the corresponding B is the known value of the grad times the provisional velocity.
Suppose if there are 9 nodes in 2D, the nodes be numbered from 1 to 9, let 9th node be the center of all other nodes, then the matrix A the row will look like,
[weight of node1 weight of node2 .. .. .. .. .. .. .. weight of node 8]
the X will be
[phi of 9]
and B will be
[grad times velocity]

The problem is when we consider the pressure nodes with known values(i.e., Dirichlet nodes) , then i remove the weight of the nodes from A and add that to B with the given dirichlet values multiplied with the weights of that nodes.
This makes the solution to diverge when i give any values other than zero(dirichlet), Can anyone help me how to give this constant dirichlet value in the matrix??

FMDenaro November 24, 2016 03:41

Are you using a library or is your own-made solver?

jawapasu November 24, 2016 03:46

Quote:

Originally Posted by FMDenaro (Post 626688)
Are you using a library or is your own-made solver?

its custom made solver, it works fine when pressure dirichlet value is given as zero, but when i give some numerical value it diverges, and the preconditioner is ILU for the solver

FMDenaro November 24, 2016 03:53

Check if your solver converges with a source term such that the solution is constant, for example 1


All times are GMT -4. The time now is 04:23.