AnsweredAssumed Answered

Manually reducing iteration step...?

Question asked by Colter Cederlof on Jul 21, 2008
Latest reply on Jul 22, 2008 by Bill McEachern

Like • Show 0 Likes0
Comment • 3

Hello,

I'm attempting a fairly simple flow problem with just a heat generating source in the middle of a box. There is a fan facing directly at the object and vents on the opposite side. The goal is to insure the object doesn't reach a maximum temperature.
I'm attempting to match experimental data with the FloWorks model to later optimize the layout. The data I have spans 15 minutes = 900 seconds. However, when I run FloWorks in the necessary time-dependent mode, each iteration only covers a physical time span of ~.04 seconds. Some quick math indicates there will be ~30,000 iterations required which is time consuming.

Thus, my question is: Is there a way to set the number of iteration for a given time span (ie 1 iteration covers .5 seconds)... or simply some way to cut down on the number of iterations for the allotted physical time span? I don't have a problem sacrificing accuracy.

Thanks for your assistance.
CC

No one else has this question

Outcomes

Rich Bayless Jul 22, 2008 1:23 PM

Hi Colter,

only 0.04 sec per iteration? pretty good! I've been running a transient analysis and it runs at 8 seconds per iteration. My target is 1800 seconds, I'm up to 1500 seconds, so do the math...

to reduce time per iteration, keep experimenting with number of cells, and be sure to not specify anything extra, like heat conduction is solids, humidity, etc. keep your analysis as simple as possible.

Warning! If you accidentally restart your anaylsis, you will lose all your work! So make backup copies!! This happened to me at 1200 seconds...... and no, I didn't have a backup.

They do not have a 'are you really, really sure?' button.

Best regards, Rich.

Actions

Colter Cederlof @ Rich Bayless on Jul 22, 2008 2:31 PM

Hello Rich and everyone,

Thanks for your reply. I'm doing my best to keep is extremely simple.
I don't think I explained my situation completely however...
The calculations run at about 5 seconds per iteration. What I meant by "each iteration only covers a physical time span of ~.04 seconds" is that the program is calculating what the fluid does at each instant in the time-dependent mode of 15 minutes. When it completes an iteration, only .04 seconds out of the total 15 minutes is estimated...
In short, to calculate what the fluid will do in 15 physical minutes, it will take 30,000 iterations at 5 seconds per iteration = ~2.5 days.

The good news is that I did find the solution I was looking for!
Click on "FloWorks" in the command bar
Select "Calculation Control Options"
Select "Advanced" tab
Click the "Manual Time Step" box and alter the time that you want each iteration to cover for a time-dependent calculation.

Funny thing... I did try this earlier, but I guess I set the value too high so I though it was for something else.

Thanks and happy analyzing.
CC

Actions

Bill McEachern Jul 22, 2008 2:35 PM

The time step is auto-adaptive it will lengthen on its own. It works pretty well. I have done analysis that had times of upwards of 700 seconds with fairly dense mesh. I had cell counts of say over 100k cells and the analysis took about 10-12 hours. CFD takes awhile.....for any code

You can always put in whatever time step you like but you do run the risk of having your solution diverge along some unintended non physical path.

Actions

3 Replies

Rich Bayless Jul 22, 2008 1:23 PM
Mark Correct
Correct Answer
Hi Colter,

only 0.04 sec per iteration? pretty good! I've been running a transient analysis and it runs at 8 seconds per iteration. My target is 1800 seconds, I'm up to 1500 seconds, so do the math...

to reduce time per iteration, keep experimenting with number of cells, and be sure to not specify anything extra, like heat conduction is solids, humidity, etc. keep your analysis as simple as possible.

Warning! If you accidentally restart your anaylsis, you will lose all your work! So make backup copies!! This happened to me at 1200 seconds...... and no, I didn't have a backup. They do not have a 'are you really, really sure?' button.

Best regards, Rich.
Like • Show 0 Likes0
Actions
- Colter Cederlof @ Rich Bayless on Jul 22, 2008 2:31 PM
  Mark Correct
  Correct Answer
  Hello Rich and everyone,
  
  Thanks for your reply. I'm doing my best to keep is extremely simple.
  I don't think I explained my situation completely however...
  The calculations run at about 5 seconds per iteration. What I meant by "each iteration only covers a physical time span of ~.04 seconds" is that the program is calculating what the fluid does at each instant in the time-dependent mode of 15 minutes. When it completes an iteration, only .04 seconds out of the total 15 minutes is estimated...
  In short, to calculate what the fluid will do in 15 physical minutes, it will take 30,000 iterations at 5 seconds per iteration = ~2.5 days.
  
  The good news is that I did find the solution I was looking for!
  Click on "FloWorks" in the command bar
  Select "Calculation Control Options"
  Select "Advanced" tab
  Click the "Manual Time Step" box and alter the time that you want each iteration to cover for a time-dependent calculation.
  
  Funny thing... I did try this earlier, but I guess I set the value too high so I though it was for something else.
  
  Thanks and happy analyzing.
  CC
  Like • Show 0 Likes0
  Actions
Bill McEachern Jul 22, 2008 2:35 PM
Mark Correct
Correct Answer
The time step is auto-adaptive it will lengthen on its own. It works pretty well. I have done analysis that had times of upwards of 700 seconds with fairly dense mesh. I had cell counts of say over 100k cells and the analysis took about 10-12 hours. CFD takes awhile.....for any code

You can always put in whatever time step you like but you do run the risk of having your solution diverge along some unintended non physical path.
Like • Show 0 Likes0
Actions

Manually reducing iteration step...?

Outcomes

Related Content

Recommended Content