CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Software User Forums > ANSYS > FLUENT

Ensure convergence at eachtime step

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

Reply
 
LinkBack Thread Tools Search this Thread Display Modes
Old   March 9, 2016, 02:15
Default Ensure convergence at eachtime step
  #1
New Member
 
Amit Ray
Join Date: May 2014
Location: Chennai, India
Posts: 5
Rep Power: 12
amitray is on a distinguished road
Friends I am new to Fluent and I am doing a transient simulation in Fluent. I just need a small help. I want to know how we can ensure that at each time step the simulation converges. I mean as manually I can check that at certain time step the solution converged but suppose there are 2000 time steps and it is not possible to check convergence at each time step manually, so is there any method which can give me the data that at certain time step the solution didn't converge.
amitray is offline   Reply With Quote

Old   March 9, 2016, 04:49
Default
  #2
Senior Member
 
Cees Haringa
Join Date: May 2013
Location: Delft
Posts: 607
Rep Power: 0
CeesH is on a distinguished road
Hi Amit,

You can set convergence criteria (on velocity, temperature, ..., and the residuals of course). If these criteria are met, the simulation will jump to the next timestep. So, set the criteria such that you are sure the relevant parameters are sufficiently converged inside a timestep, and choose the number of iterations per timestep such that the convergence criteria are virtually always met before the number of iterations criterion is fullfilled. That should ensure you are safe.
CeesH is offline   Reply With Quote

Old   March 9, 2016, 05:48
Default
  #3
New Member
 
Amit Ray
Join Date: May 2014
Location: Chennai, India
Posts: 5
Rep Power: 12
amitray is on a distinguished road
Quote:
Originally Posted by CeesH View Post
Hi Amit,

You can set convergence criteria (on velocity, temperature, ..., and the residuals of course). If these criteria are met, the simulation will jump to the next timestep. So, set the criteria such that you are sure the relevant parameters are sufficiently converged inside a timestep, and choose the number of iterations per timestep such that the convergence criteria are virtually always met before the number of iterations criterion is fullfilled. That should ensure you are safe.
Thanks for the reply. I have already set the criteria and have set number of iterations per time step to 200. Initial few iterations it got converged in between 150 to 180 iterations but in between some steps it didn't converge. So I wanted to know if there is any inbuilt method or an UDF that can tell exactly which time step converged and which didn't.
amitray is offline   Reply With Quote

Old   March 9, 2016, 06:03
Default
  #4
Senior Member
 
Cees Haringa
Join Date: May 2013
Location: Delft
Posts: 607
Rep Power: 0
CeesH is on a distinguished road
Hi Amit,

In any case I would reduce your timestep size a bit. Although i don't have the reference at hand, I do recall reading somewhere that transient simulations should target at converging in 20-50 iterations per step to ensure all features are captured (if someone disagrees or has the reference, please throw it in!)

On what basis do you conclude convergence by the way? I often set the targets quite strict, so that even when the criteria are not met, the situation is likely acceptable. Of course it should be possible to write a UDF; the macro N_ITER gives the iteration number so if you export that at the end of each timestep, you know the number of iterations per timestep.

But, you can also simply export a monitor each iteration rather than each timestep (for example, average velocity). You can easily see if that converges per timestep then, in excel, matlab, etc.
CeesH is offline   Reply With Quote

Old   March 10, 2016, 02:09
Default
  #5
New Member
 
Amit Ray
Join Date: May 2014
Location: Chennai, India
Posts: 5
Rep Power: 12
amitray is on a distinguished road
Quote:
Originally Posted by CeesH View Post
Hi Amit,

In any case I would reduce your timestep size a bit. Although i don't have the reference at hand, I do recall reading somewhere that transient simulations should target at converging in 20-50 iterations per step to ensure all features are captured (if someone disagrees or has the reference, please throw it in!)

On what basis do you conclude convergence by the way? I often set the targets quite strict, so that even when the criteria are not met, the situation is likely acceptable. Of course it should be possible to write a UDF; the macro N_ITER gives the iteration number so if you export that at the end of each timestep, you know the number of iterations per timestep.

But, you can also simply export a monitor each iteration rather than each timestep (for example, average velocity). You can easily see if that converges per timestep then, in excel, matlab, etc.
Thanks again. Actually I have set convergence criteria for all residuals to 0.001. I set a time step of 0.001s which I think will capture all details. And why I took 200 iterations because I felt initial iterations will converge in more iterations and later it may converge in less number of iterations. That was my idea as the project involves two phases water and air and is a bit complex. I may also have made a mistake in boundary conditions. By the way thanks for the time you gave. I will do some detailing and then retry the problem.
amitray is offline   Reply With Quote

Old   March 11, 2016, 06:46
Default
  #6
New Member
 
ami
Join Date: Dec 2015
Posts: 8
Rep Power: 10
Ami24 is on a distinguished road
Hello,

To have convergence between each time step, you have to switch off convergence criteria in the monitors section as shown in the picture.
Attached Images
File Type: png Convergence criteria.png (20.0 KB, 107 views)
Ami24 is offline   Reply With Quote

Reply

Tags
convergance, fluent, time steps

Thread Tools Search this Thread
Search this Thread:

Advanced Search
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 Off
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
Ensure UDF runs only once a time step PranjalNewton Fluent UDF and Scheme Programming 1 October 18, 2015 07:06
PimpleFoam: check for convergence within time step Begineer OpenFOAM Running, Solving & CFD 3 April 21, 2014 18:42
Time step size & convergence absolute criteria yuitsang FLUENT 5 April 15, 2013 05:27
same geometry,structured and unstructured mesh,different behaviour. sharonyue OpenFOAM Running, Solving & CFD 13 January 2, 2013 23:40
Force can not converge colopolo CFX 13 October 4, 2011 23:03


All times are GMT -4. The time now is 17:18.