InterDyMFoam dynamic meshing in parallel fails under nonquiescent conditions

adona058 · August 22, 2008, 15:08

This bug report is in relation to my post in the CFD section: CFD Posting

The case is that of a bubble rising in a 10cm square channel. I have included a test case at time-step 1.015. Normally the error occurs at around 1.01687. The original mesh has a grid spacing of 20 x 20 x 120 (x,y,z). The test case was run without a gravity vector until the pressure, U, and gamma reached an equilibrium state. At that point (representing t=1), the gravity vector is set to (0,0,-9.81), and the simulation is continued.

The calculations are stable right up until the error is shown, hence the reason I am relatively confident that it is a grid problem with the dynamic mesh refinement operated in parallel.

The test case can be found here:

This is a tar.gz file. I appologize for the size of the file, unfortunately it is necessary to save you considerable computation time.

Any help that you may provide in overcoming this error is greatly appreciated.

Thanks

Adam

adona058 · August 22, 2008, 15:19

The file was not able to upload due to size constraints... I have placed it on my own personal site, so it can be downloaded. The link is here: Download File
the file is 23.4 mb.

Just to provide some additional detail while I wait for it to upload. The test case contains results at 0, 1, and 1.015. Everything is configured to run on 8 processors, with a start time of 1.015. It will take a few minutes for the error to appear.

If you make it to 1.07, then you've made it farther than I have.

Thanks

Adam

mattijs · August 26, 2008, 07:01

Thanks. I'll have a look.

mattijs · September 5, 2008, 06:22

Hi Adam,

finally fixed it (I think). There were some points not being removed when unrefining. Should be in the 1.5.x git now (in particular removeFaces.C)

Could you get back to me whether it works or not?

adona058 · September 5, 2008, 08:05

Thanks Mattijs,

I will try to get back to you either later today or early next week, depending on how quickly I can get the files updated on the cluster.

Adam.

adona058 · September 5, 2008, 12:47

Hi Mattijs

The modifications you made work great. The simulation is proceeding past the point where it previously had errors, and the cell merging appears to be working beside the processor boundaries.

Thank you for the quick turnarround.

Adam.

peter73 · November 18, 2008, 14:36

Adam,

How do you initialize the gamma in the directory 0/gamma?

Peter

peter73 · November 18, 2008, 14:58

Ignore the previous post. Should be more careful.

August 22, 2008, 15:08	This bug report is in relation	#1
adona058 Member Adam Donaldson Join Date: Mar 2009 Location: Ottawa, Ontario, Canada Posts: 37 Rep Power: 17	This bug report is in relation to my post in the CFD section: CFD Posting The case is that of a bubble rising in a 10cm square channel. I have included a test case at time-step 1.015. Normally the error occurs at around 1.01687. The original mesh has a grid spacing of 20 x 20 x 120 (x,y,z). The test case was run without a gravity vector until the pressure, U, and gamma reached an equilibrium state. At that point (representing t=1), the gravity vector is set to (0,0,-9.81), and the simulation is continued. The calculations are stable right up until the error is shown, hence the reason I am relatively confident that it is a grid problem with the dynamic mesh refinement operated in parallel. The test case can be found here: This is a tar.gz file. I appologize for the size of the file, unfortunately it is necessary to save you considerable computation time. Any help that you may provide in overcoming this error is greatly appreciated. Thanks Adam

August 22, 2008, 15:19	The file was not able to uploa	#2
adona058 Member Adam Donaldson Join Date: Mar 2009 Location: Ottawa, Ontario, Canada Posts: 37 Rep Power: 17	The file was not able to upload due to size constraints... I have placed it on my own personal site, so it can be downloaded. The link is here: Download File the file is 23.4 mb. Just to provide some additional detail while I wait for it to upload. The test case contains results at 0, 1, and 1.015. Everything is configured to run on 8 processors, with a start time of 1.015. It will take a few minutes for the error to appear. If you make it to 1.07, then you've made it farther than I have. Thanks Adam

August 26, 2008, 07:01	Thanks. I'll have a look.	#3
mattijs Senior Member Mattijs Janssens Join Date: Mar 2009 Posts: 1,419 Rep Power: 26	Thanks. I'll have a look.

September 5, 2008, 06:22	Hi Adam, finally fixed it (	#4
mattijs Senior Member Mattijs Janssens Join Date: Mar 2009 Posts: 1,419 Rep Power: 26	Hi Adam, finally fixed it (I think). There were some points not being removed when unrefining. Should be in the 1.5.x git now (in particular removeFaces.C) Could you get back to me whether it works or not?

September 5, 2008, 08:05	Thanks Mattijs, I will try	#5
adona058 Member Adam Donaldson Join Date: Mar 2009 Location: Ottawa, Ontario, Canada Posts: 37 Rep Power: 17	Thanks Mattijs, I will try to get back to you either later today or early next week, depending on how quickly I can get the files updated on the cluster. Adam.

September 5, 2008, 12:47	Hi Mattijs The modification	#6
adona058 Member Adam Donaldson Join Date: Mar 2009 Location: Ottawa, Ontario, Canada Posts: 37 Rep Power: 17	Hi Mattijs The modifications you made work great. The simulation is proceeding past the point where it previously had errors, and the cell merging appears to be working beside the processor boundaries. Thank you for the quick turnarround. Adam.

November 18, 2008, 14:36	Adam, How do you initiali	#7
peter73 New Member Peter Lian Join Date: Mar 2009 Posts: 12 Rep Power: 17	Adam, How do you initialize the gamma in the directory 0/gamma? Peter

November 18, 2008, 14:58	Ignore the previous post. Shou	#8
peter73 New Member Peter Lian Join Date: Mar 2009 Posts: 12 Rep Power: 17	Ignore the previous post. Should be more careful.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
InterDyMFoam dynamic refinement	ala	OpenFOAM Running, Solving & CFD	12	September 28, 2016 18:51
InterDyMFoam dynamic messing in parallel fails under nonquiescent conditions	adona058	OpenFOAM Running, Solving & CFD	5	August 19, 2010 11:47
Running interDyMFoam in parallel	sega	OpenFOAM Running, Solving & CFD	1	March 12, 2009 05:54
OF15 Dynamic Adaptive Meshing in Parallel	adona058	OpenFOAM Running, Solving & CFD	2	August 7, 2008 14:27
Serial run OK parallel one fails	r2d2	OpenFOAM Running, Solving & CFD	2	November 16, 2005 12:44