We are seeing what it looks like a yaw dependent instability in OpenFAST that appears mainly around yaw = +30 deg / -30 deg, and we were wondering whether others have run into something similar.
For the parked cases, we are using:
Blade pitch = 87 deg
GenDOF = False
CompServo = 0 so the controller is disabled
UA_Mod = 0
So these are intended to be fully parked cases with no active control action.
What we see is that most yaw angles look reasonable, but +/-30 deg tends to produce a much stronger oscillatory response, in some channels almost looking like the onset of an instability:
What is interesting is that we sometimes see something similar in idling cases too. In those cases, however, we are not disabling GenDOF or CompServo, so the rotor, generator and controller dynamics are still present. Even there, the +/-30 deg yaw cases tend to stand out:
This makes us wonder whether yaw around +/-30 deg is a known sensitive condition for parked/idling simulations, especially with blades feathered to 87 deg. So our questions are the following ones:
Have you ever seen this kind of behavior before?
Is +/-30 deg yaw known to be especially problematic or sensitive in parked/idling OpenFAST simulations?
What would you recommend testing for this specific cases?
You are running into a known blade edgewise / tower side-side instability that happens under parked/idling conditions under high winds and sizeable yaw error. This issue is not unique to your turbine model and not unique to OpenFAST.
The issue has been discussed in other topics on our forum and the OpenFAST repository issues/discussions. I suggest reviewing the following for explanations, workarounds, and solutions:
I want to ask you which wind turbine you are using as well as the corresponding type of substructure ? I am asking so in order to reproduce this instability.
The wind turbine model I am using is a proprietary model.
Regarding the substructure, if you are referring to offshore substructures, we are not using any, since the simulations are being carried out for an onshore wind turbine. In this case, the substructure is assumed to be fully rigid.
I made a comparison of the blade tip edgewise deflection between the NREL 5 MW and the IEA 3.4 MW onshore wind turbines under the same seed of stochastic wind speed field having V_mean = 50 m/s and a nacelle yaw of +25 degrees. The rotor speed is set to zero and the blades are feathered (blade pitch angles equal to +90 degrees).
Please kindly find below the figure showing this comparison:
I was able to reproduce the edgewise instability in the NREL 5 MW. However, it looks like that the IEA 3.4 MW onshore wind does jot suffer from this instability.
Is there any explanation for this difference in behavior of the blade tip edgewise deflection ?
I would expect all wind turbines to suffer from an edgewise instability for an idling rotor subject to high winds and a yaw error, however, the exact yaw errors may differ between turbines; the azimuth angle may also play a role. I don’t have much personal experience with the IEA Wind 3.4-MW RWT, but can you get this model to be unstable at different yaw or azimuth angles?
I have ran 360 simulations covering nacelle yaw (or yaw misalignment error) from -179 degrees to +180 degrees for the IEA 3.4 MW onshore wind turbine. The wind speed field is stochastic and has a mean of 50 m/s. The azimuth angle was kept intact. All the DoFs were enabled except for nacelle yaw and the six degrees of freedom of the substructure. The inital conditions for the azimuth, the rotor speed are zero. The blades are feathered (i.e. blade pitch equals to 90 degrees). Due to the enormous number of load cases, i decided to proceed as follows: I computed the PSDs of:
tower top fore-aft displacement
tower top side-side displacement
blade tip flapwise deflection
blade tip edgewise deflection
I order to get a global view, i plot the heatmap of the PSD as function of the frequency and nacelle yaw as shown in the following figures:
When looking at the heatmap of blade tip edgewise deflection, one could see that the value of the PSD is the highest in the range of [-45 deg:+45 deg]. The PSDs of tower top displacement in both fore-aft and side-to-side are the highest at 0.4 Hz which is the frequency of the tower 1st bending mode. It is more pronouced for tower top fore-aft displacement.
I think the critical range to be compared with the NREL 5 MW is [-45 deg:+45 deg].
I’m not really following your interpretation of the heat map for blade-tip edgewise deflection. Also, are you seeing any signs of instability in any of the cases?
Looking briefly at the IEA Wind 3.4-MW RWT repository: IEA-3.4-130-RWT/openfast at master · IEAWindSystems/IEA-3.4-130-RWT · GitHub, I presume you are using the default model set-up with ElastoDyn for modeling the blade structural dynamics. I noticed in the ElastoDyn blade input file (IEA-3.4-130-RWT/openfast/IEA-3.4-130-RWT_ElastoDyn_blade.dat at master · IEAWindSystems/IEA-3.4-130-RWT · GitHub) that the structural damping is quite a bit higher than other reference models (BldFlDmp(1-2), BldEdDmp = 3% in the IEA Wind 3.4-MW RWT compared to ≈0.5% in the NREL 5-MW baseline and IEA Wind 15-MW RWT). I’m not sure why the damping is so much higher, but I suspect that explains why this model is not seeing the classic edgewise instability seen in other models. If you reduce the blade structural damping to ≈0.5%, do you then see signs of an instability?
I changed the damping ratio for both flapwise (1st and 2nd bending modes) and edgewise (1st bending mode) directions in the ElastoDyn_blade.dat of the IEA 3.4 MW onshore wind turbine.
As you have expected, the behavior now is quite similar to that of the NREL 5 MW onshore wind turbine where the edgewise instability appears clearly.
Please find below a figure that shows a comparison of the balde tip edgewise deflection between two vales of damping ratio: 0.48 % and 3 %.
Yes, recent experience tells me that the 0.5% structural damping is more realistic.
But a common solution to avoid the edgewise instability–believing that a real turbine likely has other sources of damping not represented in the numerical model, such as aerodynamic damping from unsteady aerodynamics in deep stall–is to increase the structural damping specifically for DLC 6.2.
I would like to ask whether an edgewise structural damping value of 2.0% would be considered appropriate for DLC 6.2, or if you would recommend a different value.
I have tested this damping value and the differences in the results are clearly noticeable comparing them to the initial 0.5% I was using. Since increasing the damping can help account for damping sources not represented in the numerical model, I would like to know whether 2.0% is a reasonable assumption, or if another damping value would be more suitable.
I am working with an OpenFAST model and I am trying to justify the structural damping values used in ElastoDyn for the blade edgewise and tower bending modes for the particular case of DLC 6.2.
I performed a sensitivity study with four damping configurations:
0.5% structural damping for all blade/tower modes;
2.0% for the blade edgewise mode, keeping the other values at 0.5%;
2.0% for the blade edgewise mode and 1.0% for the tower fore-aft and side-to-side modes;
1.0% for the tower fore-aft and side-to-side modes, keeping the blade damping at 0.5%.
The best overall response, in terms of reduced oscillations in blade-root, tower-base and low-speed shaft moments, was obtained with:
In a previous forum answer, you mentioned that, in the absence of more specific information, you typically assume structural damping ratios of 2–3% for composite blades and 0.5–1.5% for steel towers:
Would you consider the use of 2% structural damping for the blade edgewise mode and 1% structural damping for the tower fore-aft and side-to-side modes to be appropriate and defensible and realistic for DLC 6.2 simulations?
In this context, would you recommend validating these damping values through any method?
Indeed, those were the default values I had used for many years. More recently, our experience is that structural damping is less for composite beams, on the order of 0.5%.
Regardless, increasing the structural damping is common in DLC 6.2 to account for terms that are missing in the aerodynamics model in deep stall.
