|
[Sponsors] |
Unable to correctly restart a simulation that has stopped. |
|
LinkBack | Thread Tools | Search this Thread | Display Modes |
June 12, 2023, 14:48 |
Unable to correctly restart a simulation that has stopped.
|
#1 |
New Member
Join Date: Jun 2023
Posts: 3
Rep Power: 3 |
Hello everyone!
My 2D simulation is a cylinder forced to move vertically, it is attached with springs and has a flow coming horizontally. I am using overPimpleDyMFoam. When the simulation stops, I am unable to correctly restart it. The simulation restarts but doesn't calculate the right thing. As an example, I ran 2 simulations with exactly the same conditions, and I stopped one at 1.92 sec, the other was not stopped. You can see on the attached pictures (0sec, 1.92sec, 2.5sec, 4.5sec) that the cylinders are not following the same movements. I would like to be able to restart the simulation from where it stopped because they take a lot of time to run. I already solved a problem, the file 'points' used to restart the simulation was the initial one. I just have to copy the last saved 'points' file to the folder 'constant' so the cylinder is at the good position for the restart. I wonder if my problem comes from the forces or the displacement of the cylinder that is not updated for the restart? I would really appreciate if someone could help me. Let me know if you need any information more to be able to help me. Thank You! |
|
June 14, 2023, 08:24 |
|
#2 |
Member
Vinzenz Götz
Join Date: Mar 2021
Location: Munich, Germany
Posts: 31
Rep Power: 5 |
Could you give a bit more detail?
Have you tried setting the "startTime" in controlDict to "latestTime" and then just running the solver? |
|
June 14, 2023, 11:24 |
|
#3 |
New Member
Join Date: Jun 2023
Posts: 3
Rep Power: 3 |
Hi Vinzmann,
I tried to put 'startForm latestTime'. The simulation restarts but I can clearly see on paraview that the movement of the cylinder changes. The U field changes also. I have the feeling that something is missing to restart the simulation correctly and the code is calculating a new state which is not continuous with the previous one. 5 files from the simulation are attached to this message (3 from the folder 'system' and 2 from the folder 'constant'). I can put other files if you need. Thank you very much for your help! |
|
June 15, 2023, 03:35 |
|
#4 |
Member
Vinzenz Götz
Join Date: Mar 2021
Location: Munich, Germany
Posts: 31
Rep Power: 5 |
Hello Feuille,
you said in your first post that you copied the points folder over into constant, what happened before you did that? Did it throw an error? Did you do something else except stop the simulation and then start it again? How do you start the simulation? Do you use a script like: Code:
#!/bin/sh cd "${0%/*}" || exit # Run from this directory . ${WM_PROJECT_DIR:?}/bin/tools/RunFunctions # Tutorial run functions #------------------------------------------------------------------------------ runParallel $(getApplication) -parallel Code:
overPimpleDyMFoam Best regards, Vinz |
|
June 15, 2023, 17:17 |
|
#5 |
New Member
Join Date: Jun 2023
Posts: 3
Rep Power: 3 |
Hello Vinzmann,
Before doing anything, when I restarted the simulation where it stopped, the cylinder was back at the initial position (center of the block). However, there were no error or warning in terminal. I was able to see there is a problem with the deltaT that went from 5e-5 to 3e-9. (deltaT is adjustable so the courant nb is under 1) To start and restart a simulation, when I use 3 cores for example, I use: Code:
renumberMesh -overwrite | mpirun -np 3 renumberMesh -overwrite -parallel overPimpleDyMFoam | mpirun -np 3 overPimpleDyMFoam -parallel Code:
overPimpleDyMFoam After I copied points over into constant, when the simulation restarts, the cylinder seems to be at the right position, the position where simulation stopped. I noticed that the pointDisplacement file refers to the restarted postion for the zero point. For example, before simulation stopped, at 1.92sec, points of the cylinder are displaced like (0 -0.0382993 0). After the simulation restarts, at 1.94sec, these points are displaced like (0 -0.000827018 0). (I read that in the files pointDisplacement for each time step) I hope this is clear to understand . It makes me think that it can affect the refAttachementPt and the anchor of the springs that are set to the center of the block. Thank you for your help Vinzmann! |
|
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Restart a dynamic refined simulation from a different time-step | Andrea_85 | OpenFOAM | 6 | November 29, 2018 09:28 |
a transient cfx simulation suddenly stopped writing .out and then .bak while running | mona.li | CFX | 1 | March 5, 2018 05:15 |
How to restart a simulation in MPI mode | hamedhamed | AVL FIRE | 4 | April 26, 2013 14:27 |
Simulation Time in restart runs | Elina | CFX | 3 | January 21, 2009 04:57 |
Restart of FSI simulation | V. Kumar | CFX | 3 | July 20, 2006 14:23 |