Restarting with pimpleFoam

deinstein · June 30, 2010, 11:00

Hi,

I have been using pimpleFoam lately and am very satisfied with its performance. Lately, however, when network issues caused my simulation to die I noticed that restarts are problematic. Perhaps someone can shed some light on this for me.

Specifically, I was running with a variable time steps and a maxCo of 0.9. The simulation was terminated (due to a network glitch not any fault of pimpleFoam) at t=0.03 with a maxCo of 0.8 and a time step of 0.002. When I restarted it from t=0.03, the maxCo went to 1.8 and the time step (logically) went down to 0.0004. Ok all good. But from there it basically never recovered. The maxCo hovers at 0.9 and the timestep is now order 1E-05. If I start from scratch (t=0), the simulation sails (accurately) through this time slice with a large time step an a reasonable maxCo. In other words, basically the restart di not work. Any clues?

Kaskade · November 26, 2013, 09:22

Did you find a solution? I am running a simulation using pimpleDyMFoam, after I kill it and restart the time step drops from 4e-4 to 7e-9.

Can some provide a solution and/or an explanation. I need to restart since I want to run LRR as turbulence model.

timo_IHS · November 26, 2013, 13:40

Hi,

can you try a fixed timestep!?
And do you see a pressure jump?

Best

Kaskade · November 26, 2013, 13:45

I tried a fixed time step, which resulted in a maximal courant number of 551 and the abrupt end of the run.

timo_IHS · November 26, 2013, 13:53

In the first timestep?

Kaskade · November 26, 2013, 13:58

In the first time step after the restart, yes.

Could there be a problem with the DynamicMesh?

timo_IHS · November 27, 2013, 05:59

There might be a problem with combination of PIMPLE and dynamic mesh!?

Kaskade · November 27, 2013, 08:42

Next time I restart, I will set the writeInterval to 1. Maybe pimpleDyMFoam rotates the mesh too far upon restart. Should be visible in Paraview.

timo_IHS · November 27, 2013, 09:23

which version do you use?
There was a bug in 2.0.? or 2.1.? with AMI...

Kaskade · November 27, 2013, 09:53

I use 2.2.2. According to github there have been no changes to pimpleDyMFoam in the last couple of months.

June 30, 2010, 11:00	Restarting with pimpleFoam	#1
deinstein New Member Daniel Einstein Join Date: Mar 2009 Posts: 22 Rep Power: 17	Hi, I have been using pimpleFoam lately and am very satisfied with its performance. Lately, however, when network issues caused my simulation to die I noticed that restarts are problematic. Perhaps someone can shed some light on this for me. Specifically, I was running with a variable time steps and a maxCo of 0.9. The simulation was terminated (due to a network glitch not any fault of pimpleFoam) at t=0.03 with a maxCo of 0.8 and a time step of 0.002. When I restarted it from t=0.03, the maxCo went to 1.8 and the time step (logically) went down to 0.0004. Ok all good. But from there it basically never recovered. The maxCo hovers at 0.9 and the timestep is now order 1E-05. If I start from scratch (t=0), the simulation sails (accurately) through this time slice with a large time step an a reasonable maxCo. In other words, basically the restart di not work. Any clues?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
pimpleFoam Body Force	mdjames	OpenFOAM Running, Solving & CFD	24	November 4, 2021 21:59
Running stably with a large Courant number using pimpleFoam	mlawson	OpenFOAM	11	November 29, 2014 11:37
Problem running IDDES with pimpleFoam	charlie	OpenFOAM Running, Solving & CFD	7	July 11, 2013 12:45
tried with pimpleFoam and a big CFL = 5	ivan_cozza	OpenFOAM Running, Solving & CFD	9	June 4, 2013 19:14
PimpleFoam problem	Vadims.geza	OpenFOAM	8	December 7, 2011 09:54

November 26, 2013, 09:22		#2
Kaskade Senior Member Onno Join Date: Jan 2012 Location: Germany Posts: 120 Rep Power: 15	Did you find a solution? I am running a simulation using pimpleDyMFoam, after I kill it and restart the time step drops from 4e-4 to 7e-9. Can some provide a solution and/or an explanation. I need to restart since I want to run LRR as turbulence model.

November 26, 2013, 13:40		#3
timo_IHS Member Timo K. Join Date: Feb 2010 Location: University of Stuttgart Posts: 66 Rep Power: 16	Hi, can you try a fixed timestep!? And do you see a pressure jump? Best

November 26, 2013, 13:45		#4
Kaskade Senior Member Onno Join Date: Jan 2012 Location: Germany Posts: 120 Rep Power: 15	I tried a fixed time step, which resulted in a maximal courant number of 551 and the abrupt end of the run.

November 26, 2013, 13:53		#5
timo_IHS Member Timo K. Join Date: Feb 2010 Location: University of Stuttgart Posts: 66 Rep Power: 16	In the first timestep?

November 26, 2013, 13:58		#6
Kaskade Senior Member Onno Join Date: Jan 2012 Location: Germany Posts: 120 Rep Power: 15	In the first time step after the restart, yes. Could there be a problem with the DynamicMesh?

November 27, 2013, 05:59		#7
timo_IHS Member Timo K. Join Date: Feb 2010 Location: University of Stuttgart Posts: 66 Rep Power: 16	There might be a problem with combination of PIMPLE and dynamic mesh!?

November 27, 2013, 08:42		#8
Kaskade Senior Member Onno Join Date: Jan 2012 Location: Germany Posts: 120 Rep Power: 15	Next time I restart, I will set the writeInterval to 1. Maybe pimpleDyMFoam rotates the mesh too far upon restart. Should be visible in Paraview.

November 27, 2013, 09:23		#9
timo_IHS Member Timo K. Join Date: Feb 2010 Location: University of Stuttgart Posts: 66 Rep Power: 16	which version do you use? There was a bug in 2.0.? or 2.1.? with AMI...

November 27, 2013, 09:53		#10
Kaskade Senior Member Onno Join Date: Jan 2012 Location: Germany Posts: 120 Rep Power: 15	I use 2.2.2. According to github there have been no changes to pimpleDyMFoam in the last couple of months.