Dear OpenFAST team,
I´m currently simulating DLC 1.3 with Extreme Turbulent wind, and I´m appreciating a transient at low wind speeds (100–200 s): AoA becomes erratic and LSS loads oscillate. DLC 1.2 with the same controller does not show this behavior, as can be seen below:
Some observations:
-
With AeroDyn BEM_Mod=2, I get some errors when the simulation is running:
-
Disabling UA does not change that transient.
-
Switching BEM_Mod from 2 → 1 makes the AoA and LSS loads well behaved and the problem dissapears:
From what I’ve read in the forum, these warnings often occur when relative velocity approaches zero. My case seems consistent with that, but I’m not sure what the recommended mitigation is when using BEM_Mod=2.
So my question is the next one:
- Are there recommended settings to keep BEM_Mod=2 in this Design Load Cases and avoid this numerical error? If not, do you recommend to use BEM_MOD=1 in this specific situation?
Happy to share the OpenFAST (4.1.1) and the Aerodyn (v15) versions in order to provide you more information.
Thanks!
Dear @Kepa.Cueto,
I’m a bit surprised the sensitivity is tied to BEM_Mod, which refers to the coordinate system with the momentum balance takes place. Can you clarify what other AeroDyn settings you’ve enabled with BEM_Mod = 2? Is the issue tied to the settings of Skew_Mod, DBEMT_Mod, or UA_Mod?
Best regards,
Good morning, Jason. Thank you very much for your reply.
I have performed another test with the parameters I gave you yesterday, and the same thing continues to happen. Below are two simulations, where the only difference is the BEM_Mod, as you can see below:
- In Blue: BEM_Mod=2, DBEMT_Mod=2, UA_Mod=4 and Skew_Mod=1.
- In Orange: BEM_Mod=1, DBEMT_Mod=2, UA_Mod=4 and Skew_Mod=1.
I have also realized that changing from BEM_Mod 2 to 1 makes a change in the AD.sum file (in the BlCrvAng), as follows:
Could it be this parameter that is making the difference when it comes to stabilising the angles of attack or not?
Thank you and kind regards,
Kepa
Dear @Kepa.Cueto,
Thanks for the additional information. BlCrvAng changes with BEM_Mod because of the change in internal coordinate systems used for the momentum balance.
Does the use of BEM_Mod = 2 stabilize if you use DBEMT_Mod = 0, UA_Mod = 0, and/or Skew_Mod = 0? I’m curious if one of these submodels is causing the problem with BEM_Mod = 2, and if so, which one?
Best regards,
Dear Jason,
I’ve run the tests you mentioned, and I’ve noticed that keeping BEM_Mod at 2 and setting DBEMT_Mod=0 does improve stability:
As additional information, and in case it helps, setting the BEM to 2 and disabling the UA or Skew_Mod makes no difference. In fact, in the case of Skew_Mod, it makes things even worse:
Could we therefore say that DBEMT_Mod at 0 provides sufficient stability to maintain BEM=2? Or would you like to see any further changes?
I would be very grateful if you could explain the reason for these changes.
Kind regards,
Kepa
Dear Kepa,
If you don’t mind, could you add the following quasi-steady AeroDyn nodal outputs: BEM_k_qs, BEM_kp_qs, AxInd_qs. TnIn_qs, Phi. My guess is that BEM_Mod=2 introduces some jumps in these values which triggers some instabilities, potentially due to a change of flow angle Phi.
After that, maybe you can try the following simulations:
-
BEM_Mod=1, DBEMT_Mod=2 (rerun with extra nodal outputs)
-
BEM_Mod=2, DBEMT_Mod=2 (rerun with extra nodal outputs)
-
BEM_Mod=2, DBEMT_Mod=0 (rerun with extra nodal outputs)
-
BEM_Mod=2, DBEMT_Mod=1 with a tau1_const value that is reasonable. To see if it has to do with the variation of the time constant
-
BEM_Mod=2, DBEMT_Mod=2, halving the time step, to see if has to do with the time integration in DBEMT.
Let us know if you find some weird jumps in the BEM quasi steady nodal outputs, if keeping a constant time constant helps, or if changingthe time step helps.
Thanks a lot,
Emmanuel
Hello Emmanuel,
I have done the tests you suggested and here they are the results with the extra nodal outputs. For a better understanding, I am comparing them in pairs, in the order you gave me:
As far I can see, the most numerically stable options with the extra nodal output that you have proposed to me are:
- BEM=1 and DBEMT=2
- BEM=2 and DBEMT=0
I consider it necessary to mention that UAMOD=4 and SKEWMOD=1 have been used for all simulations performed here.
Looking at these two, I would therefore say that numerically, the first option would be the best. I would be very grateful if you could confirm this for me. Once again, thank you very much for your attention.
Kind regards,
Kepa
