CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > CD-adapco

Solution method for transient problems

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   November 17, 2000, 13:31
Default Solution method for transient problems
  #1
Roberto Ciardulli
Guest
 
Posts: n/a
Can anybody tell me how the values of the following parameters affect the solution?

maximum number of corrector stages

reduction in residuals for corrector stages

under-relaxation for pressure correction

max residual tolerance

Thanks everybody,

Roberto
  Reply With Quote

Old   November 18, 2000, 04:02
Default Re: Solution method for transient problems
  #2
A.Hassaneen
Guest
 
Posts: n/a
Roberto, The values of the corrector stages and the residuals works like a criteria for convergence of the inner iteration for each time step. i.e. the solution terminate the inner iteration when any of these two parameter fulfilled. The under-relaxation of the pressure is very important like any other relaxation. Any body correct me if I'm wrong.
  Reply With Quote

Old   November 18, 2000, 10:45
Default Re: Solution method for transient problems
  #3
X.Ye
Guest
 
Posts: n/a
Hi,

Because of Calculation high transient flow with Star CD, I just want to ask: If you calculate it with a smaller time step, do you need only a smaller number of iteration for each time? For example, If the time step is 100.0e-6, you need 60 iterations, but if the time step is 10.0e-6, do you need only 6 iteration?

I ask this question, because I want to compare the calculation efficiency with the explicit method. StarCD has the implicit method.

X. Ye
  Reply With Quote

Old   November 19, 2000, 04:11
Default Re: Solution method for transient problems
  #4
A.Hassaneen
Guest
 
Posts: n/a
I don't have definite answer for your question, but you may try it and look at the file (case.info) to see how the solution is going.
  Reply With Quote

Old   November 22, 2000, 10:13
Default Re: Solution method for transient problems
  #5
J. Y. Luo
Guest
 
Posts: n/a
The maximum no. of correctors is very similar to the maximum no. of iterations for steady flow calculations - if the equations converges earilier, this no. is then unimportant: the most important criterion is how much the residuals of all equations are reduced, normally PISO will converge within 2 to 4 correctors, so the default 20 correctors are quite irrelevant.
  Reply With Quote

Old   November 27, 2000, 04:13
Default Re: Solution method for transient problems
  #6
Roberto Ciardulli
Guest
 
Posts: n/a
So is it ok if I use a very high value for MAX NUMBER of CORRECTOR STAGES (ex. 100) and a very low value for REDUCTION in RESIDUAL for CORRECTOR STAGES (ex. 0.1)

I still get the neg. density error during transient state runs and I would like to know if these parameters affect the solution's stability.

Thanks everybody,

Roberto
  Reply With Quote

Old   November 27, 2000, 06:23
Default Re: Solution method for transient problems
  #7
J. Y. Luo
Guest
 
Posts: n/a
Hi, Roberto,

What you described is certianly Ok. You may need to use under-relaxation factor for PISO (this is used to under-relax the pressure correction) that would help in your problem. The under-relaxation factor would not affect the final results at each time step, but may need a few more correctors for PISO to converge. Typically a value of 0.5 is common.

  Reply With Quote

Old   November 27, 2000, 10:10
Default Re: Solution method for transient problems
  #8
Roberto Ciardulli
Guest
 
Posts: n/a
I've set the following values:

Max number of corrector stages = 1000

Reduction in residual for corrector stages = 0.1

Under-relax. for press. correction = 0.1

but I got the NEGATIVE DENSITY error even before the first time step!!!!!

What can I do??? I really need to solve this problem!!

Thanks,

Roberto
  Reply With Quote

Old   November 27, 2000, 13:51
Default Re: Solution method for transient problems
  #9
John C. Chien
Guest
 
Posts: n/a
(1). I found that the questions and answers are very interesting, that is, it is not going anywhere. (2). I think, the general rule is that, each CFD simulation run, including all of the parameters defined, is a "special case". (3). For this reason, it is important to provide as much information as possible in the question, if one is interested in getting the real help. (4). A commercial CFD code does not provide the solution, unless it also knows the user's problem ahead of the time and set the corresponding parameters accordingly. Unfortunately, such automatic codes are not available yet (may not be possible at all). (5). My suggestion to the one who asked the question is: (a). try to learn as much as possible the related methods used in the commercial codes, even if it is a black box and does not give adequate information. In this way, you will be able to ask the right question, (b). try to follow the sample tutorial cases closely first. These cases were worked out by the vendor's engineers, and supposed to have the repeatable solutions. (c). if your problem is in one of the sample tutorial cases, then the chances of getting your solution are rather high, because you don't have to change too much the control parameters. (d). when your problem is outside the sample tutorial cases, then you are on your own. It is your responsibility to put together all of the necessary steps and control parameters and find out whether this black box will generate a good solution for you or not. This is the common understanding of the commercial CFD codes. (6). AS matter of fact, these problems and issues must be considered long before one start running a code. Bye the way, running the code blindly will only make the life miserable. If I were you, I would first check into the listing of the sample tutorial cases, if the problem of interest is there, then go ahead and follow the sample. Otherwise, "No One In THE WHOLE WORLD KNOWS WHETHER YOUR PROBLEM HAS A SOLUTION OR NOT." A "general CFD code" does not mean that it will generate a solution for any CFD problem. It simply say that the "formulation" is general enough that, you can try it on other problems beside the sample tutorial cases. So, if you read this message thoroughly, and understand it, it will save you a lot of emotional problems.
  Reply With Quote

Old   November 27, 2000, 16:47
Default Re: Solution method for transient problems
  #10
Sreenadh Jonnavithula
Guest
 
Posts: n/a
Hi Roberto,

John's right - you need to tell us more about the actual problem to allow someone to figure out whats going on. My guess is that the problem has nothing to do with corrector stages or residual tolerance, but something either in the setup or the mesh. For example, if you have a large outflow velocity imposed, you might cause the cells next to it to "empty out" over your time step, with the code reporting it as negative density. What exactly are you trying to simulate? I would also encourage you to contact user support directly.

  Reply With Quote

Old   November 28, 2000, 05:38
Default Re: Solution method for transient problems
  #11
Ahmed Hassaneen
Guest
 
Posts: n/a
Roberto, As I said before, your problem is related to either one of these four reasons:

1. the length of the time step (make it smaller) 2.an event of cell deletion or activation which is not properly set up. 3.your mesh size 4.initial or boundary values.

Could you give us an idea about your boundries and initial values??
  Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Transient solution Amir Jourak CFX 0 July 9, 2007 16:27
TRANSIENT SOLUTION CONVERGE MARK Main CFD Forum 3 November 2, 2006 11:08
Transient Run solution diverges Narmin CD-adapco 1 February 15, 2005 06:31
How to divide the transient Solution file Narmin B Hushmandi CD-adapco 4 January 19, 2005 09:00
Transient solution Elyyan FLUENT 2 December 2, 2003 04:26


All times are GMT -4. The time now is 17:08.