CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Transient Simulation (http://www.cfd-online.com/Forums/fluent/37653-transient-simulation.html)

Cathy Lee August 22, 2005 12:24

Transient Simulation
 
For a transient simulation, the residual outcomes showed fluctuations and did not decrease down- Droped down to convergence criteria and increased and do iteration again-and droped down to the criteria....it kept cycling like this and did not stop. Would you please help me on this?

helper August 22, 2005 15:09

Re: Transient Simulation
 
Try to use less aggressive (smaller) under-relaxation factors or CFL number. Or, see if reducing time-step size helps.

Cathy Lee August 22, 2005 16:55

Re: Transient Simulation
 
Thank you so much for your advice. However, initially, I set the under-relaxation factors 0.3 for pressure and 0.7 for momentum/ 20 for the time-step size. I changed to 0.3 for pressure and 0.2 for momentum and kept the time-step size;however it did not help at all. Are there any other potential reasons cause these problems?

Thanks again.

helper August 22, 2005 18:44

Re: Transient Simulation
 
Try a smaller time-step size or keep reducing time-step size, as I said.

Cathy Lee August 23, 2005 12:16

Re: Transient Simulation
 
I reduced the time-step size from 20 to 5. It hit the convergence criteria once and a message telling "it is converged". However, it did the next iteration and the redisuals increased again; it does not stop and keeps running . Can I use the outcome for postprocessing? Or I have to change something and try again. If I have to change, would you please guide me what else I can do? Also, I would like to know whether the accuracy changes if I reduce under-relaxation factor and take smaller time-step size.

Suvash August 24, 2005 02:04

Re: Transient Simulation
 
I think you should keep the default under-reluxation factor and decrease the time step size. Time step size 5 is also very large for some cases. try it for 1(one) or less than 1(one)..

edi August 24, 2005 04:29

Re: Transient Simulation
 
On accuracy and URFs: this issue was faced many times in this forum, you can also find analytical explanations in the manual. (Very) roughly speaking:

-low URFs: less convergence problems, more iterations required for a "converged" solution;

-higher URFs: growing convergence problems, less iterations required.

In theory, once you get your solution "converged", it doesn't matter what URFs you chose, so the goal should be finding the higher URFs that allow a converged solution... . The moral of the story is that you better keep on with your default URFs...

On your time step: time step 5 is very large for MOST of the cases!! I agree with Suvash: try it for less than 1. To speed up your calculation you should consider the NITA algorythm, but the time must be further reduced and it affects accuracy in some cases...

Hope this can help

Good luck

Edi.

breathless August 24, 2005 12:32

Re: Transient Simulation
 
i have the same problem with breathing, i inhale and then later i exhale, ... it keeps repeating ... inhale followed by exhale ... what should i do?

the code is doing what it should ... converging at each time-step, advancing time, and then converging again at the next time step.

Cathy Lee August 24, 2005 13:30

Re: Transient Simulation
 
Many thanks to all!!

Swarup August 27, 2005 11:42

Re: Transient Simulation
 
There is no problem in this. The code is trying to obtain converged solution per time step & obtaining it. Residuals increase suddenly because they are calculated using values between two time steps. Some problems converge periodically. If you are using transient simulation for time accurate predictions, then you must achieve convergence per time step, otherwise you can afford to obtain convergence at higher times, when steady state is achieved. There are some problems which have NO steady state solution as is the case with natural convection heat transfer.

Swarup.

alac1407 September 29, 2010 21:49

Hi everyone! :)
I have a problem about simulating venturi tube
In my simulation, Inlet velocity is sin function and it varies on over time.
So should I use Steady State or Transient !
And if I use Transient model, please help me set parameters in Transient tab!
Thank you very much in advance !


All times are GMT -4. The time now is 23:36.