CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Convergence problems - steady/unsteady (http://www.cfd-online.com/Forums/main/275-convergence-problems-steady-unsteady.html)

Jonathan Zerihan October 27, 1998 07:06

Convergence problems - steady/unsteady
 
Hi. I'm a PhD student looking at wings in ground effect, comparing my experimental results with CFD. My code gives a n oscillating response using second and third order in space , but 'converges' for first order, when using a steady state solution. With the problem set up as unsteady, a single solution is converged upon which is not unsteady. This leads me to think that I should be able to get proper converge with the problem set up as steady, using third order.

Help - can anybody please give me some advice as to how I can get the steady problem to converge/any advice greatly appreciated.

John C. Chien October 27, 1998 11:02

Re: Convergence problems - steady/unsteady
 
(1). use " first order " to get solution to converge, then (2). continue on with " second and higher " scheme, (3). watch carefully the nature of the oscillation and especially the location of oscillation. (4). use under-relaxation factors to slow down the change between iterations. (5). refine your mesh in the area where you have oscillations. (6). Maybe the higher order schemes you use simply are less stable than the first order scheme. (6). oscillation is not a bad thing because most of the time, it is trying to tell something. So you can do something about it.

Jonathan Zerihan October 28, 1998 11:35

Re: Convergence problems - steady/unsteady
 
Many thanks - I am a beginner in this field, but am learning quickly!!

John C. Chien October 28, 1998 11:51

Re: Convergence problems - steady/unsteady
 
Learning is one simple way to keep your brain young and healthy. To keep ourself alive, we all have to keep learning. My answers to your question are simply suggestions only. ( There is no guarantee that it's the right answer. )

Andrew Chernousov October 30, 1998 13:00

Re: Convergence problems - steady/unsteady
 
The poor convergence to steady state of higher-order schemes with limiters can be significantly improved, see ref.:

Venkatakrishnan V. Convergence to steady state solutions of the Euler equations on unstructured grids with limiters. // J. Comput. Phys. 1995. v. 118. N 1. P. 120-130.


All times are GMT -4. The time now is 01:18.