CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Time step for unsteady problem (http://www.cfd-online.com/Forums/cfx/20087-time-step-unsteady-problem.html)

Jackie January 8, 2004 03:48

Time step for unsteady problem
 
I'm simulating an unsteady problem and have a problem in setting time step. Some value made anerror and cause the program can't create tha result file at the final step. How should we know the suitable value of time step for the unsteady problem ( Grid generation for my domain are the unstructure mesh so delta X , Y and Z are not equal)

Sincerely,

Jackie

Jan Rusås January 8, 2004 07:30

Re: Time step for unsteady problem
 
I only have experience with 5.5.1, I don't know if you are using 5.6. Please in the future always specify which verison you are using. The question about selecting time step for transient problems have been raised a couple of times. Try to do a search also at the cummunity pages. If you have an idea about the period of your transient problem, then you could select a timestep that is 30 -100 smaller than the period. Pay some attention to the number of iterations within each timestep. If you have convergence problem - try to increase the number of iteration within each timestep, it is default 3 and/or decrease the timestep Are you sure that your problem is related to the timestep? Jan

Jackie January 9, 2004 01:19

Re: Time step for unsteady problem
 
Thank a lot Jan.

Bob January 16, 2004 13:58

Re: Time step for unsteady problem
 
Hi Jan and Jackie, Jan said that if convergence is a problem that you should increase your number of iterations, however I've herd that it is better to reduce your time step rather than increase iterations. Why ? well I'm not 100% sure, may be robin could answer that, but I got the information from the Help desk some time ago. Really you should do a time step dependency study, in the same way that you should do a grid dependency study. A good site to look at is the flow mania web site where they look at DES simulations and URANS simulations (which is what you are interested in). the address is: http://cfd.me.umist.ac.uk/flomania/index2.html

Have a look at the publications, particularly one on "Persistance and Effect of 3 dimensionality in URANS sims of 2D Bluff Bodies" there is some useful info their - heighlighting the effects of mesh and time step.

Bob

jackie January 19, 2004 00:51

Re: Time step for unsteady problem
 
Thank you for your kind answer Bob.


All times are GMT -4. The time now is 18:04.