|November 3, 2012, 19:11||
3D Finite Element for incompressible flow
Join Date: Sep 2012
Posts: 4Rep Power: 6
I am writing a finite element code for solving incompressible navier-stokes equation. to solve the system of linear equations I am using Bi-conjugate gradient algorithm as it works with unsymetric system. I am not getting a converged solution and the residuals becomes 'nan' in 2nd or 3rd iteration. Although I can not post my code here, I would like to know few things if possible.
1. Does the global stiffness matrix have to be diagonally dominant ? In my case it is not. so in that case I am making a mistake in basic formulation.
2. If diagonally dominant is not a requirement for N-S equations, what preconditioner is best suited for unsymmetric system ? I am using ILUpreconditioner. Just wondering if the problem is in using preconditioner.
3. I do not have any zeros on the diagonal. What are the general scenarios when FEM solution diverges or becomes 'nan' ? (Its hard to answer without the code but I would appreciate answers if anyone has faced this problem before)
Firstly I thought that problem could be in mixing the pressure and velocity unequal interpolation function. SO I removed all the pressure terms and continuity equation. It remain like a 3D heat equation. Still the problem persists. I doubt that its the preconditioner which is causing it.
|Thread||Thread Starter||Forum||Replies||Last Post|
|How can be calculate a surface integral in the finite element method?||HectorRedal||Main CFD Forum||3||March 13, 2015 09:25|
|Control Volume Finite Element Method||gerardosrez||Main CFD Forum||0||March 16, 2011 15:49|
|finite volume VS finite element||solomon||Main CFD Forum||0||March 19, 2002 20:04|
|Equations for finite difference, element & volume||W.L.C.||Main CFD Forum||9||April 3, 2001 12:08|
|fluid flow fundas||ram||Main CFD Forum||5||June 17, 2000 21:31|