Why does the rotor speed oscillate at steady winds?

Hallo all,

I’m quite new to the wind energy technology in general and the simulation of wind turbines in particular.
To extend my knowledge about wind turbines I try to learn FAST. Therefore I currently work through the files and examples from the 2014 NREL Wind Turbine Modeling Workshop in Bergen.

In that context I was able to run the first two simulations and I also got pretty much the same results as provided in Demonstration_Jonkman.pdf (part of Simulations_NoOutput.zip)
However having a closer look on the results of Simulation 2, I was quite wondering about the behaviour of the rotor speed.
As in that simulation the turbine is subjected to a steady wind without shear. I expected a constant rotor speed, but in fact the rotor speed oscillates around the Initial or fixed rotor speed (RotSpeed) defined in the ElastoDyn primary input file.

Reading through the forum I found this topic. There it is mentioned that periodic loads on the rotor could arise from a non-zero shaft tilt, as this tilt leads to asymmetric aerodynamics even in steady winds without shear. Therefore I changed the parameter ShftTilt to 0 and rerun the simulation, but unfortunately the rotor speed still oscillate.

My second idea was that the oscillations may be induced through the motion of the blades. Therefore I disabled the blades DOF (FlapDOF1, FlapDOF2, EdgeDOF), but again that did not lead to a constant rotor speed.

As I now have no further ideas I would like to ask you:
Do you know where these oscillations could come from and why the rotor speed isn’t constant?

Thank you very much in advance.

Kind regards,
Paul Schünemann

Dear Paul,

Simulation 2 from that workshop involves the land-based NREL 5-MW turbine operating under steady wind with no control and a fixed generator speed. If you plot ElastoDyn output GenSpeed you’ll notice that indeed the generator speed is constant. However, this model also has the drivetrain-torsion DOF enabled (DrTrDOF = True in ElastoDyn). The turbine operation results in torques being applied to the shaft, which will tend to twist the shaft a bit. However, the initial twist of the shaft is not specified within the ElastoDyn input file along with the other initial conditions, and in fact is assumed to be zero by ElastoDyn. The oscillation you are seeing in rotor speed is the transient response seen as the twist is achieving equilibrium with the applied torques.

I hope that helps.

Best regards,

Dear Jason,

Thank you very much for your answer, which gives me a further insight in the physics of wind turbine.

As you wrote Simulation 2 is about a turbine with fixed generator speed, but in the input file we define a fixed rotor speed (RotSpeed).
So do I understand you right that the ElastDyn parameter (initial condition) RotSpeed - Initial or fixed rotor speed (rpm) in FAST actually rather defines the generator speed than the rotor speed.

Furthermore I could appreciate that the oscillations I am seeing are the transient response to “wrong” initial conditions. However if that’s the case shouldn’t these oscillations disappear after a couple of 10 sec, i.e. when the equilibrium is achieved? Unfortunately that’s not the case and the oscillations tend to never disappear.

Sorry for asking such may be simple questions, but I really would like to understand the basics before continuing with more complex simulations.

Kind regards,
Paul Schünemann

Dear Paul,

It is always good to start with simpler models and build complexity in steps.

The drivetrain model in FAST is initialized with rigid-body rotation. While the initial rotor speed is specified, the generator speed is initialized consistently as GBRatio*RotSpeed, where GBRatio is the gearbox ratio.

I would expect in the absence of applied aerodynamic torque oscillations for the torsional oscillations to decay; this may take a while if the damping is low.

Best regards,

Dear Jason,

I agree with you in the expectation that the torsional oscillations would decay after a while. To check that I just run a quite longer simulation (100 minutes), but there is only a decay in the first seconds and after that the rotor speeds oscillate between 8.997 rpm and 9.004 rpm for the rest of the simulation.

As you see the oscillation is very small, but it is existing. However due to the small amplitudes could it also be possible that the oscillations are just numerical artefacts?

Furthermore it’s clear that (like you said) the torsional oscillations will only decay if there is no applied aerodynamic torque. The simulations of the considered workshop are done with FAST v8.08 and AeroDyn v14.02. With that configuration is it possible to output the aerodynamic torque? (I have read that it should be possible with AeroDyn v15 but didn’t find anything about AeroDyn v14.)

Kind regards,
Paul Schünemann

Dear Paul,

You implied that you uploaded a figure, but it didn’t seem to be get uploaded. Regardless, I wouldn’t expect numerical issues to result in a small oscillation, although there may be numerical round-off in the output.

Have you disabled all structural DOFs along with eliminating the shaft tilt? The tower deflection would also induce a small amount of shaft tilt, which would induce some oscillation. Unless you’ve disabled this feature in AeroDyn, this model also has a tower-influence model from potential flow enabled, which will induce a 3 per-rev oscillation (3P) in aerodynamic torque. I would expect you’d need to disable all these features to remove all oscillations in rotor peed.

In AeroDyn v15 it is possible to output the pure aerodynamic torque, but this was not possible with earlier versions of the code. Regardless, an oscillation in rotor speed will result in an oscillation in aerodynamic torque; it is usually a “chicken-and-egg” problem to identify the root cause.

Best regards,

Dear Jason,

Thank you very much for your helpful answers and advices. I think, no I get it (at least a bit). Inspired by you last post, I did some further simulations and it turned out that in fact it is the tower potential flow, which induces the oscillations I have seen.

However these simulations left me with some further questions.

1. You said that the tower potential flow will induce a 3P oscillation. Therefore I did a simulation based on simulation 2 of the workshop with the following configuration:
• wind: as in simulation 2 (constant wind speed (8m/s) without shear)
• Initial Conditions: fixed rotor speed (9 rpm)
• ShftTilt = 0
• All Blade-DOF are disabled (FlapDOF1, FlapDOF2, EdgeDOF = FALSE)
• Tower potential flow is enabled (in AeroDyn: TwrPotent = TRUE)

By doing this I get the following results for RotSpeed and RotTorq (after reaching the “equilibrium”):

As far as I understood the oscillations seen in these figures are 6P oscillations, isn’t it? So these results do not correspond with your statement that the oscillations should be 3P ones. Do you have an idea what’s the reason for that or do I interpret the figure in a wrong way?
(If I additionally to the above configuration also disable the tower potential flow, all oscillations disappear.)

1. Furthermore you said that “The tower deflection would also induce a small amount of shaft tilt, which would induce some oscillation.” To verify this, I did another simulation with the following configuration:
• wind: as in simulation 2 (constant wind speed (8m/s) without shear)
• Initial Conditions: fixed rotor speed (9 rpm)
• ShftTilt = 0
• All Blade-DOF are disabled (FlapDOF1, FlapDOF2, EdgeDOF = FALSE)
• Tower potential flow is disabled (in AeroDyn: TwrPotent = FALSE)

What I see in the results (see figure below) are no oscillations in RotSpeed, but some oscillations in RotTorq and RotPwr.

As you said RotSpeed and RotTorq are closely connected, which I understood as oscillations in one quantity should result in oscillations of the other one. Do you have an idea why that’s here not the case? Could it be possible that the changes in Rotortorque are to little to accelarate the whole rotor because of the rotors inertia?

Kind regards,
Paul Schünemann

Dear Paul,

1. I see a combination of 3P and 6P in your results. In most cases (due to nonlinearities, nonsinusoidal excitation, etc.), 3P excitation will also result in excitation at the harmonics of 3P (6P, 9P, etc.).

2. It looks like the torque oscillation is very small (changes in the 4th digit of precision), and my guess is the rotor speed oscillation is even smaller and not visible with only 4 digits of precision in the FAST output. You could increase the precision of the FAST output (e.g. by outputting in binary format) if this is a concern.

Best regards,

Dear Jason,

thank you very much for your time and all of your answers to my sometimes trivial questions.

That helped me a lot and I learned many new aspects of the physics and the simulation of wind turbines. Thanks for that.

Kind regards,
Paul Schünemann