Linearization using ADAMS

Hello everybody,
My question is regarding the linearization of HAWT using ADAMS. It is assumed that wind turbine data set for ADAMS is generated using FAST-2-ADAMS preprocessor, as per description listed in FAST user manual. Actually I am interested in studying the flutter stability limit of WT blades at different operating conditions. And in this matter the interaction of flap-wise and torsional modes from Eigen analysis using ADAMS (as FAST currently does not possess the torsional degree-of-freedom) would be very helpful.

FAST theory manual shows that the linearization of HAWT, subjected to incident wind, can be carried out for both normal power production as well as parked conditions.

My question is:
Can the linearization of the turbine rotor spinning at various speeds and subjected to different winds,
be performed using current FAST-2-ADAMS and ADAMS framework as well?

Thanks
Hayat

Dear Hayat,

At this time, the FAST-to-ADAMS preprocessor and ADAMS-to-AeroDyn interface only support linearization of an ADAMS model with no gravity, rotor speed, damping, or aerodynamics. Unfortunately, the way the ADAMS-to-AeroDyn interface was developed prohibits ADAMS from accessing aerodynamic states or computing aerodynamic derivatives e.g. for calculation of Jacobians. Some advanced users of ADAMS have found work-arounds for this limitation – i.e. by computing the aerodynamic derivatives separately and using those within the ADAMS analysis, but these features are not directly built-into the current FAST-to-ADAMS preprocessor or ADAMS-to-AeroDyn interface.

By the way, NREL is making good progress on the development of our new BeamDyn module for FAST v8. This module will greatly enhance FAST’s ability to model blade structural dynamics. BeamDyn is based on geometrically exact beam theory (GEBT) and implemented using spectral finite elements. The model includes full geometric nonlinearity; bending torsion, shear, and extensional DOFs; anisotropic material couplings (full 6x6 mass and stiffness matrices); sectional offsets; and initially curved/swept blades. We are working through an extensive verification effort, will make it more user-friendly, will couple it to FAST, document it, and release it. Our goal is for the first public release in June, 2014.

Best regards,

Dear Hayat,

I’d like to add a point about linearizing with ADAMS, but first please understand that I am really just an amateur in this area and this is a recollection from long ago.

Many years ago, our former employee, Gunjit Bir, did a thorough analysis of how ADAMS linearizes and he and Mechanical Dynamics (since bought out by MSC.Software) agreed that their formulation does not work for spinning rotors. This problem was a fundamental feature of how ADAMS works, so it was unlikely that it could be fixed. I vaguely remember that it had something to do with linearizing trig functions. Maybe they have since fixed it, but I doubt it.

You may want to contact MSC.Software to verify this information. If they say it will work fine, please let me know so I can edit this statement. I would also then check with their wind-turbine expert just to make sure.

Marshall

Dear Marshall and Hayat,

Actually, MSC did fix the issue in ADAMS associated with linearizing a model with a spinning rotor. The solution was to linearize the system in terms of user-defined coordinates. There have been some publications on this, e.g.,:

Negrut, D. and Ortiz, J. “On an Approach for the Linearization of the Differential Algebraic Equations of Multibody Dynamics.” Proceedings of IDETC/MESA 2005, September 24-28, 2005, Long Beach, USA.

However, I never had the opportunity to test the updated code, and the FAST-to-ADAMS preprocessor and ADAMS-to-AeroDyn interface were never updated to take advantage of this improvement.

Best regards,

Dear Jason and Marshall,

Thank for your valuable comments. The news regarding BeamDyn module for FAST v8 sounds great. Really appreciate the effort done by NREL team.

From your comments, it seems to me that it would be more appropriate (off course would be cumbersome) to investigate the classical flutter behavior (i.e. interaction of flap-wise and torsional deformations) from time-domain simulation results from ADAMS analysis, using current FAST-to-ADAMS preprocessor and ADAMS-to-AeroDyn interface.

Your opinion and/or recommend reference document regarding the flutter analysis based on time-domain approach would be appreciated.

PS: I am not expert user of ADAMS, that’s why avoiding any modifications needed for linearization of spinning rotor using ADAMS.

Thanks,
Hayat

Dear Hayat,

I know that Don Lobitz of Sandia National Laboratories has done some time-domain-based stability analysis using ADAMS. See for example:

Lobitz, D. “Aeroelastic Stability Predictions for a MW-Sized Blade.” Wind Energy, 7:211-224, 2004; DOI: 10.1002/we.120.

Best regards,

Dear Janson,
Thanks for sharing link. It contains useful, but not explicit, information on time-domain flutter analysis.

The work of Lobitz describes the wind turbine rotor spinning at different speeds in the still air for time-domain flutter analysis.
Most probably the linearly rising rotation speed of the rotor in still air is modeled (using ADAMS-AeroDyn), as also mentioned in
other references as well like: Politakis, G., Haans, W. & Bussel, G.J.W. van (2008): Suppression of classical flutter using a ‘smart blade’, 46th AIAA Aerospace
Sciences Meeting and Exhibit,7 - 10 January 2008, Reno, NV, (pp. 2008-1301). Reno, NV.

My question is:
Is it possible to model the simulation for linearly rising rotation speed of the rotor in still air condition, by setting the
parameters of FAST input file?
There is an “Idling turbine case” special event simulation case in FAST, but, It does not
serve my purpose.

Regards,
Hayat

Dear Hayat,

By specifying constant generator torque with no aerodynamic torque, you should get constant generator acceleration, which would lead to linearly increasing rotational speed. The simplest way to specify constant generator torque is by setting VSContrl to 1, VS_RtTq to the desired constant generator torque, and VS_RtGnSp, VS_Rgn2K, and VS_SlPc to 9999.9E-9 (very small don’t care’s > 0.0).

Best regards,

Dear Jason,

I tried as per your advise, but, I could not model the linearly increasing
rotor speed in the still air, as you can see in the attached figure. The rotor speed
increases non-linearly up to 50 second, and then becomes almost constant as lift
and drag forces acting on the rotor are might be in a balance to each other.

Moreover, please also find “fast.fst” file in the attachment. The important
parameters used for analysis are as following :

VSContrl = 1, VS_RtGnSp, = 9999.9E-9, VS_RtTq = 43.093e3 Nm,
VS_Rgn2K = 9999.9E-9, VS_SlPc = 9999.9E-9
GenDOF = true, CompAero = true, steadywind = 25 m/sec

One more very basic question:
The rotor with speed linear increasing in still air means that i need to use steady uniform
wind and run-away rotor conditions in order to spin the rotor. Please
do correct me if I am wrong.
fast.fst.pdf (26.3 KB)
Figure.tif (1.57 MB)

Dear Khazar,

“Still air” to me means that the wind speed is 0 m/s. I think you meant to say steady and uniform wind.

At 25 m/s, there is enough aerodynamic torque (and the torque is impacted by rotor-speed changes) that a constant generator torque leads to a steady state generator speed instead of the desired continuously linear increasing speed.

I’m afraid in steady and uniform (nonzero) wind, there is no way through settings in the FAST input files to force a linear increasing generator speed. Instead, you’ll have to write your own UserGen() routine to calculate a generator torque, based on the aerodynamic torque, so as to obtain constant acceleration of the generator.

Best regards,