overpimpledymfoam in openfoam v1912 unsteday case have not convergence

zkx123321 · July 15, 2020, 23:37

I have already simulated an elliptical airfoil translation and rotation motion in 20 periods by using overpimpledymfoam. But the lift force and drag force are not converge to the same value for the different periods. I want to know the reason why the difference between two periods would happen.

Tobermory · July 16, 2020, 12:38

Well, a first question to ask yourself is: is there actually a smooth, limit state pattern for this flow problem? And if so, is it indeed a pattern that has just one frequency of excitation? Are there other frequency modes? Indeed, is some form of chaotic motion expected?

I don't know what you are trying to simulate, or how you are simulating it (DES, RANS?), but these are some initial questions that I'd ask myself. Try also find examples online of others who have solved something similar, and compare.

Good luck.

zkx123321 · July 17, 2020, 11:16

Thank you for your reply Tobermory, the simulated case is period oscillation in platplane. Recording the relative papers validation, the simulation Cl and Cd present the period variation, rather than the simulation by myself.

bennn · July 17, 2020, 11:34

Hi,

A couple of things to check:

- The number of oscillations might not be sufficient. It is possible than more oscillations will bring convergence, due to the wake advection.
- The usual mesh convergence check : refinement, y+, wake etc...
- The number of inner iterations controlled by fvSolution:
nOuterCorrectors 2; // number of pimple iterations; 1 : PISO mode
nCorrectors 2; // number of p outter GAMG iteration; 0 : only pcorr
nNonOrthogonalCorrectors 1; // number of p and pcorr inner GAMG iteration
- Time scheme is fundamental in unsteady simulation. Backward is often very different from Euler or other 1st order schemes.

- Eventually as ToberMemory said, it's possible that the oscillations remain variable from one oscillation to the next. In that case you can try to average the last oscilations, and check if you're close to experimental data. Keep in mind that wind tunnel experiments are almost always averaged as well.

zkx123321 · July 17, 2020, 23:34

Dear bennn:
Your reply gives me a detailed description on the setting parmeters, than, I would try to test the case by following your suggestions. If there is any new progress, I will sort out the results and show them in here.

zkx123321 · September 8, 2020, 03:31

thank you very much bennn
My problem have already solved in one object moving oscillation, your advice is very useful in my problem.

July 17, 2020, 11:34		#4
bennn Member benoit paillard Join Date: Mar 2010 Posts: 96 Rep Power: 16	Hi, A couple of things to check: - The number of oscillations might not be sufficient. It is possible than more oscillations will bring convergence, due to the wake advection. - The usual mesh convergence check : refinement, y+, wake etc... - The number of inner iterations controlled by fvSolution: nOuterCorrectors 2; // number of pimple iterations; 1 : PISO mode nCorrectors 2; // number of p outter GAMG iteration; 0 : only pcorr nNonOrthogonalCorrectors 1; // number of p and pcorr inner GAMG iteration - Time scheme is fundamental in unsteady simulation. Backward is often very different from Euler or other 1st order schemes. - Eventually as ToberMemory said, it's possible that the oscillations remain variable from one oscillation to the next. In that case you can try to average the last oscilations, and check if you're close to experimental data. Keep in mind that wind tunnel experiments are almost always averaged as well. Tobermory likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Frequently Asked Questions about Installing OpenFOAM	wyldckat	OpenFOAM Installation	3	November 14, 2023 12:58
Some problems in naca 0012 V&V case of NASA TMR and DPW using OpenFOAM	chengdi	OpenFOAM Running, Solving & CFD	7	October 5, 2019 14:20
Suggestion for a new sub-forum at OpenFOAM's Forum	wyldckat	Site Help, Feedback & Discussions	20	October 28, 2014 10:04
New OpenFOAM Forum Structure	jola	OpenFOAM	2	October 19, 2011 07:55
Free surface boudary conditions with SOLA-VOF	Fan	Main CFD Forum	10	September 9, 2006 13:24

July 16, 2020, 12:38		#2
Tobermory Senior Member Join Date: Apr 2020 Location: UK Posts: 747 Rep Power: 14	Well, a first question to ask yourself is: is there actually a smooth, limit state pattern for this flow problem? And if so, is it indeed a pattern that has just one frequency of excitation? Are there other frequency modes? Indeed, is some form of chaotic motion expected? I don't know what you are trying to simulate, or how you are simulating it (DES, RANS?), but these are some initial questions that I'd ask myself. Try also find examples online of others who have solved something similar, and compare. Good luck.

July 17, 2020, 11:16		#3
zkx123321 New Member Join Date: Apr 2020 Posts: 4 Rep Power: 6	Thank you for your reply Tobermory, the simulated case is period oscillation in platplane. Recording the relative papers validation, the simulation Cl and Cd present the period variation, rather than the simulation by myself.

July 17, 2020, 23:34		#5
zkx123321 New Member Join Date: Apr 2020 Posts: 4 Rep Power: 6	Dear bennn: Your reply gives me a detailed description on the setting parmeters, than, I would try to test the case by following your suggestions. If there is any new progress, I will sort out the results and show them in here.

September 8, 2020, 03:31		#6
zkx123321 New Member Join Date: Apr 2020 Posts: 4 Rep Power: 6	thank you very much bennn My problem have already solved in one object moving oscillation, your advice is very useful in my problem.