Parallel or serial!!!

a.sarami · June 11, 2010, 13:48

Dear Sir/Madam
I used a parallel approach to solve a problem with CFX; and the problem was solved normally. But when I use serial method, my solution diverges without any specific result file. I am using version 12 and my problem is a transient one.
My next question is about making the mesh finner, I solved my problem and now I want to solve it with a finner mesh, but when I make it fine my solution diverges again.

I wanted to ask if it is possible to help me on these two issues.
Thank you.

liucheng860212 · June 13, 2010, 04:30

I have the same problem after i installed the HP-MPI service. I think maybe it has something to do with the MPI service. But i did not work out yet.
Anyone know what the problem is?

ghorrocks · June 14, 2010, 19:34

You may have run out of memory on the serial model, but the parallel one ran fine with distributed memory.

feizaghaee · June 17, 2010, 08:16

what do you simulate?
do you set everything up properly?

a.sarami · June 17, 2010, 10:40

As I previously told, my system is a multiphase gas-solid flow modeling. as I appreciated all the things are simulated truly because I have run it already. My problem in convergence is when I want to make my mesh a little finner.

feizaghaee · June 17, 2010, 15:01

divergence is related to several reasons. you meant that you can get reasonable result with coarse mesh but when you make it finer you can't get converged result?
some problem is related to قفل شکستگی and the only solution is simulating completely again.

June 11, 2010, 13:48	Parallel or serial!!!	#1
a.sarami Member Ali Sarami Join Date: Jun 2010 Posts: 34 Rep Power: 16	Dear Sir/Madam I used a parallel approach to solve a problem with CFX; and the problem was solved normally. But when I use serial method, my solution diverges without any specific result file. I am using version 12 and my problem is a transient one. My next question is about making the mesh finner, I solved my problem and now I want to solve it with a finner mesh, but when I make it fine my solution diverges again. I wanted to ask if it is possible to help me on these two issues. Thank you.

June 17, 2010, 10:40	Reply	#5
a.sarami Member Ali Sarami Join Date: Jun 2010 Posts: 34 Rep Power: 16	As I previously told, my system is a multiphase gas-solid flow modeling. as I appreciated all the things are simulated truly because I have run it already. My problem in convergence is when I want to make my mesh a little finner.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Poisson eq w setReference works serial diverges in parallel	tehache	OpenFOAM Running, Solving & CFD	5	August 29, 2012 10:41
Serial vs parallel different results	luca	OpenFOAM Bugs	2	December 3, 2008 11:12
Differences between serial and parallel runs	carsten	OpenFOAM Bugs	11	September 12, 2008 12:16
Help: Serial code to parallel but even slower	Zonexo	Main CFD Forum	4	May 14, 2008 11:26
Parallel convergence worse than serial!	Michael B	FLUENT	3	December 9, 2006 22:46

June 13, 2010, 04:30		#2
liucheng860212 New Member Cheng Liu Join Date: Apr 2010 Location: Beijing, China Posts: 4 Rep Power: 16	I have the same problem after i installed the HP-MPI service. I think maybe it has something to do with the MPI service. But i did not work out yet. Anyone know what the problem is?

June 14, 2010, 19:34		#3
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,850 Rep Power: 144	You may have run out of memory on the serial model, but the parallel one ran fine with distributed memory.

June 17, 2010, 08:16		#4
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	what do you simulate? do you set everything up properly?

June 17, 2010, 15:01		#6
feizaghaee Senior Member moein Join Date: Dec 2009 Posts: 132 Rep Power: 16	divergence is related to several reasons. you meant that you can get reasonable result with coarse mesh but when you make it finer you can't get converged result? some problem is related to قفل شکستگی and the only solution is simulating completely again.