instability in tower at high wind idling with yaw error

Hi All,
I observe some increasing tower side-side vibrations at higher wind speeds and yaw error (DLC 6p2), see plot.

I m wondering if it is real or related to some numerical instability.

What is your opinion on that? Did anyone experienced a similar case?

Best Regards,

Florian Stache
morewind engineering solutions GmbH

Dear Florian,

Perhaps the problem you are seeing is similar to the (physical) instability discussed in the following forum topic and associated references: http://forums.nrel.gov/t/designing-for-yaw-errors-using-fast/1051/2?

Best regards,

Dear Jason,
thank you for you reply.
You wrote in the earlier post:
Yes, I’m familiar with the problem, as explained in the report you mentioned. I have talked to other modelers in Europe and heard that this problem has been seen in many different turbines and with different aero-elastic software. I’ve also met a PhD student at DTU in Denmark that researched the causes/solutions of this problem. My understanding is that the current belief is that the instability would likely not occur in the physical world and that the aero-elastic software only predict a problem due to simplifications in how the software treat the dynamics of deep stall. My understanding is that the industry’s current approach to dealing with this problem is to either (1) bypass it by choosing yaw errors that don’t result in the instability or (2) increase the structural damping in the blade edge / tower side-to-side mode until the instability goes away. For (1), instead of limiting the range of yaw errors, one may simply ignore the case that causes the instability (in your case this could mean dropping the 35-degree case).

This is my understanding until now as well and it is the way I was handling it so far. It is very interesting as it does not systematically occur on any turbine design. Probably there are critical values of wind speed for each type of turbine (size, tower eigenfrequency…). In the present case increasing damping on toward side helps. Anyhow it is not the most satisfying approach.

Best Regards,
Florian

Dear Jason,

Is it possible that these instabilities occure at a yaw-error of 90 ° with FAST v8.16.00a-bjj?

I’ve set the AeroDyn v15.03 parameters to
0 WakeMod
1 AFAeroMod

A normal turbulence model according to IEC Edition 3 with an average wind speed of 20 m/s is used.

For a not locked rotor the Tower-top fore-aft deflections look like:
TTDspFA.jpg

and for a locked rotor the instability isn’t less:
TTDspFA_Rotorlock.jpg

The maximum deflection Fore-aft at 90 ° yaw-error are for both examples almost the same.

Thank you for your fast response.

Best regards,
René

Dear René,

What pitch angles are you setting and what angles of attack are you seeing for simulation?

Best regards,

Dear Jason,

Sorry for my uncomplete post. The blades are feathered at 90 °. I have to output these angles of attack. After that I will give you an answer.

Best regards,
René

Dear René,

With a yaw error of 90 degrees and pitch angles of 90 degrees, I would expect angles of attack near 90 degrees. I have not seen aero-elastic instabilities at that condition, but I’d be curious as to what you find.

Best regards,

Dear Jason,

Thank you for your answer.
The angles of attack are about 90 degrees (over 90 degrees the angles are over -270 degrees). BαNβAlpha shows this:

Best regards,
René

Dear René,

Yes, these are the angles of attack I was expecting. Does a linearization analysis for steady wind at this condition show negative damping for a certain mode?

Best regards,

Dear Jason,

I haven’t done a linearization analysis yet, but I compared the results with using AeroDyn14 instead of AeroDyn15 and the instabilities are no more shown now. The result looks very good.
What is your opinion about this result?

Best regards,
René

#Did you mean “instabilities” instead of “stabilities” in your post from “Fri Aug 04, 2017 3:16 pm”?

Dear René,

Yes, I meant to say “instabilities”. I’ve updated my post above.

I would expect AeroDyn v14 and AeroDyn v15 to give the same result under these conditions because for this condition I would expect you to disable induction and disable unsteady airfoil aerodynamics, leaving only static airfoil data together with a geometric angle of attack. Is this how you’ve set up both models?

Best regards,

Dear Jason,

Yes, I’ve set up both models this way. Let me show you my AeroDyn Input files:

AeroDyn v14.04:

STEADY StallMod - Dynamic stall included [BEDDOES or STEADY] (unquoted string) USE_CM UseCm - Use aerodynamic pitching moment model? [USE_CM or NO_CM] (unquoted string) EQUIL InfModel - Inflow model [DYNIN or EQUIL] (unquoted string) NONE IndModel - Induction-factor model [NONE or WAKE or SWIRL] (unquoted string) 0.005 AToler - Induction-factor tolerance (convergence criteria) (-) PRANDtl TLModel - Tip-loss model (EQUIL only) [PRANDtl, GTECH, or NONE] (unquoted string) PRANDtl HLModel - Hub-loss model (EQUIL only) [PRANdtl or NONE] (unquoted string) NEWTOWER TwrShad - Use NEWTOWER feature True TwrPotent - Calculate tower potential flow (flag) False TwrShadow - Calculate tower shadow (flag) "AeroDyn14_Tower.dat" TwrFile - Tower drag file name (quoted string) True CalcTwrAero - Calculate aerodynamic drag of the tower at the ElastoDyn nodes. TwrPotent 1.225 AirDens - Air density (kg/m^3) 1.48571E-05 KinVisc - Kinematic air viscosity [CURRENTLY IGNORED] (m^2/sec) 0.005 DTAero - Time interval for aerodynamic calculations (sec)

AeroDyn v15.03:

====== General Options ============================================================================ False Echo - Echo the input to "<rootname>.AD.ech"? (flag) "default" DTAero - Time interval for aerodynamic calculations {or "default"} (s) 0 WakeMod - Type of wake/induction model (switch) {0=none, 1=BEMT} 1 AFAeroMod - Type of blade airfoil aerodynamics model (switch) {1=steady model, 2=Beddoes-Leishman uns 1 TwrPotent - Type tower influence on wind based on potential flow around the tower (switch) False TwrShadow – Calculate tower influence on wind based on downstream tower shadow? (flag) True TwrAero - Calculate tower aerodynamic loads? (flag) False FrozenWake - Assume frozen wake during linearization? (flag) [used only when WakeMod=1 and w ====== Environmental Conditions =================================================================== 1.225 AirDens - Air density (kg/m^3) 1.48571E-05 KinVisc - Kinematic air viscosity (m^2/s) 335 SpdSound - Speed of sound (m/s) ====== Blade-Element/Momentum Theory Options ====================================================== [used only w 1 SkewMod - Type of skewed-wake correction model (switch) {1=uncoupled, 2=Pitt/Peters, 3=coupled} [ True TipLoss - Use the Prandtl tip-loss model? (flag) [used only when WakeMod=1] True HubLoss - Use the Prandtl hub-loss model? (flag) [used only when WakeMod=1] true TanInd - Include tangential induction in BEMT calculations? (flag) [used only when WakeM True AIDrag - Include the drag term in the axial-induction calculation? (flag) [used only whe True TIDrag - Include the drag term in the tangential-induction calculation? (flag) [used onl "Default" IndToler - Convergence tolerance for BEMT nonlinear solve residual equation {or "default"} 100 MaxIter - Maximum number of iteration steps (-) [used only when WakeMod=1] ====== Beddoes-Leishman Unsteady Airfoil Aerodynamics Options ===================================== [used only w 3 UAMod - Unsteady Aero Model Switch (switch) {1=Baseline model (Original), 2=Gonzalez’s True FLookup - Flag to indicate whether a lookup for f' will be calculated (TRUE) or whether b

To rule out modelling differences in AeroDyn (e.g. Airfoil-Data, …) I’ve compared AeroDyn v14 with v15 for steady wind too. I think these results look good.

Thank you for your support.

Best regards,
René

Dear René,

Indeed, your input files look like I would expect them to be for this load case. I’m surprised if these settings are yielding different overall FAST responses. But I’d have to look at both full models to understand why you are seeing differences between them.

Best regards,

Dear Jason,

It seems to work with FAST v8.16.00a-bjj with AeroDyn v15.04 (instead of AeroDyn v15.03).

Best regards,
René