You are on page 1of 8

Floquet Modal Analysis

Karl Stol
National Revewable Energy Laboratory,
of a Teetered-Rotor Wind Turbine
National Wind Technology Center,
1617 Cole Blvd., Golden, CO 80401-3393 This paper examines the operating modes of a two-bladed wind turbine structural model.
e-mail: karl_stol@nrel.gov Because of the gyroscopic asymmetry of its rotor, this turbine’s dynamics can be quite
distinct from that of a turbine with three or more blades. This asymmetry leads to system
Mark Balas equations with periodic coefficients that must be solved by the Floquet approach to ex-
Department of Aerospace Engineering Science, tract the correct modal parameters. A discussion of results is presented for a series of
University of Colorado at Boulder, simple models with increasing complexity. We begin with a single-degree-of-freedom sys-
Boulder, CO 80309-0429 tem and progress to a model with seven degrees-of-freedom: tower fore-aft bending, tower
lateral bending, tower twist, nacelle yaw, hub teeter, and flapwise bending of each blade.
Gunjit Bir Results illustrate how the turbine modes become more dominated by the centrifugal and
National Renewable Energy Laboratory, gyroscopic effects as the rotor speed increases. Parametric studies are performed by
National Wind Technology Center, varying precone angle, teeter stiffness, yaw stiffness, teeter damping, and yaw damping
1617 Cole Blvd., Golden, CO 80401-3393 properties. Under certain levels of yaw stiffness or damping, the gyroscopic coupling may
cause yaw and teeter mode coalescence, resulting in self-excited dynamic instabilities.
Teeter damping is the only parameter found to strictly stabilize the turbine model.
关DOI: 10.1115/1.1504846兴

Introduction The governing system equations bring out the time-periodic terms
expected of a two-bladed turbine, whose dynamic interactions un-
The modes of a wind turbine are often determined with its rotor
dergo a periodic variation with each rotor revolution. Due to a
in a parked position. While such modes help validate a turbine’s
lack of symmetry in a two-bladed rotor, a multi-blade coordinate
structural properties, they do not capture the dominant centrifugal
transformation does not provide a time-invariant system, as with a
and gyroscopic effects associated with an operating turbine. Op-
turbine with three or more blades. We show that a conventional
erating turbine modes identify mechanisms and couplings that un-
eigenanalysis applied to the periodic equations yields erroneous
derlie critical loading conditions. If correctly identified, these
results. Therefore, we use the Floquet analysis to compute the
modes help in designing efficient controls that mitigate loads and
modal properties. Results are presented for four models. We start
improve stability. Because of the complexity of the dynamics of
with a single-degree-of-freedom model, which allows nacelle yaw
wind turbines, characterized by rotating components coupled with
only, and progress to the full 7-dof model that allows tower mo-
stationary components, only a limited number of researchers
tion, nacelle yaw, hub teeter, and blade flap. The authors have
关1– 8兴 have attempted computation of operating turbine modes.
presented observations on similar results at the 2000 ASME Wind
These attempts use time-intensive simulations, followed by post-
Energy Symposium 关13兴. In the present analysis, we use more
processing of time response data to extract the modal frequencies,
realistic turbine properties, add parametric studies for structural
and implicitly approximate the turbine as a time-invariant system.
stiffness and damping, and further explain modal behavior.
If the turbine has three or more blades and aerodynamics are
ignored, then the structural symmetry of its rotor ensures that a
multi-blade coordinate transformation 关9兴 is sufficient to trans-
form the system time-periodic equations into a set of time- Analytical Model
invariant equations. Therefore, a conventional modal identifica-
The full SymDyn model is capable of incorporating 8 ⫹ N b
tion technique may be used to compute the modes. An earlier
degrees-of-freedom, where N b is the number of blades. These in-
paper 关8兴 used this approach to compute the operating modes of a
clude tower fore-aft and lateral deflection, tower twist, nacelle
three-bladed wind turbine. However, due to lack of a turbine ana-
yaw and tilt, shaft rotation and compliance, hub teeter, and flap-
lytical model in an explicit form, modal computations required a
wise motion of each blade. The idealized wind turbine consists of
time-intensive post-processing of data from several simulations.
rigid bodies 共i.e., the tower, bed-frame, nacelle, hub, blades, low-
New wind turbine codes are under development 关10,11兴 that
speed shaft, and the generator lumped with the high-speed shaft兲
would provide comprehensive aeroelastic models of the turbine in
all interconnected with revolute joints. Each revolute joint allows
explicit forms required for direct modal analysis. It will be some
1 dof, which is a measure of the relative angular displacement
time before these codes are developed and validated. Meanwhile,
between the two adjacent components it joins. A discrete spring-
the authors have developed a simple model called SymDyn 关12兴
damper restrains each joint. The structure is capable of including
that idealizes the wind turbine structure as an assemblage of rigid
gravity and aerodynamic loads, but both are ignored in the present
bodies interconnected by springs and joints. This model is briefly
analysis. Complete modeling details and validation attempts are
described in the next section. Although the model is approximate
provided in Refs. 关12兴 and 关14兴.
in the sense that it lumps the distributed stiffness and damping
For the two-bladed, teetered-rotor turbine considered in this
properties into discrete spring and damper elements, it does retain
study, we assume that there is no tilt degree-of-freedom. We also
the dominant physics of a rotating wind turbine.
assume a constant rotor speed, which amounts to assuming a rigid
Our objective is the modal analysis of a two-bladed teetered-
rotor shaft and a constant-speed synchronous generator. Therefore,
rotor turbine, which we model as a seven-degree-of-freedom 共dof兲
we lock out the tilt and the rotor shaft compliance degrees-of-
system. This model is simple, but retains the essential physics.
freedom in the aforementioned SymDyn model. The resulting 7-
dof model is illustrated in Fig. 1. The prescribed rotor azimuth
Contributed by the Solar Energy Division of the American Society of Mechanical
Engineers for publication in the ASME JOURNAL OF SOLAR ENERGY ENGINEER-
position, ␺ , is zero when blade #1 is pointing skyward and has
ING. Manuscript received by the ASME Solar Energy Division, May 2001, final positive clockwise rotation when viewed from the nacelle. The
revision, March 2002. Associate Editor: D. Berg. nonlinear equations of motion for this model can be expressed as

364 Õ Vol. 124, NOVEMBER 2002 Copyright © 2002 by ASME Transactions of the ASME

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


M 共 q 兲 q̈⫹ f 共 q,q̇ 兲 ⫽0 (1) two-bladed rotor also to yaw, and this induces a gyroscopic mo-
ment about the tower yaw axis. The important point to note is that
where this self-induced gyroscopic moment is a multiple of sin2␺ and,
therefore, shows 2p variation, where p denotes per rotor revolu-
q⫽ 关 ␶ 1 ␶ 2 ␶ 3 ␥ ␾ ␤ 1 ␤ 2 兴 T , tion. Rotors with three of more blades do not experience this
variation due to symmetry. All the other terms in the matrix rep-
dq d 2q
the degrees-of-freedom, q̇⫽ dt , q̈⫽ 2 , resent gyroscopic cross coupling with a 1p variation. We also note
dt that the flap and teeter motions are gyroscopically decoupled. This
is because both of these motions occur about parallel axes and are
M is the familiar mass matrix in terms of moments of inertia and
referenced to the same hub-fixed rotating frame.
angular positions, and f is the nonlinear function of remaining
terms. Floquet Approach
Linearization is performed about the steady operating state of
the turbine, q o p (t), with q(t)⫽q o p (t)⫹⌬q(t). For this study, we In this section, we will briefly outline the Floquet-Lyapunov
assume a steady operating state in which all the degrees-of- theory for periodic systems as it pertains to our modal calculation
freedom have zero values, or q o p (t)⫽0. This leads to a very method. For further details, the reader should refer to 关9兴 or 关15兴.
Consider the first-order state-space form of the equations of
simplified set of linearized equations wherein the matrices repre-
motion 共2兲:
senting mass, M , gyroscopic/damping properties, G, and stiffness
terms, K, can be readily identified. ẋ⫽A 共 t 兲 x (3)
M 共 t 兲 ⌬q̈⫹G 共 t 兲 ⌬q̇⫹K 共 t 兲 ⌬q⫽0 (2) where x⫽ 关 ⌬q ⌬q̇ 兴 T, the state vector, and
To obtain simpler models considered in this paper, we eliminate
from these matrices the rows and columns associated with the
degrees-of-freedom we wish to lock. Thus, if we do not want to
include the tower degrees-of-freedom, we simply eliminate the
冋 0
A 共 t 兲 ⫽ ⫺M 共 t 兲 ⫺1 K 共 t 兲
I

⫺M 共 t 兲 ⫺1 G 共 t 兲 ,

three rows and three columns corresponding to the three tower the time-periodic state matrix.
degrees-of-freedom from each matrix. This yields a 4-dof system. The solution of the linear time-varying system, 共3兲, can be
For this system, the corresponding equations are presented in the found in terms of the state transition matrix, ⌽(t,0):
Appendix. Note that the inertia, gyroscopic, and stiffness matrices
depend on the rotor azimuth position, ␺ , and on the turbine geo- x 共 t 兲 ⫽⌽ 共 t,0兲 x 共 0 兲 . (4)
metric, mass, stiffness, and damping properties. Also note that the The essence of Floquet-Lyapunov theory for periodic systems is
gyroscopic matrix is multiplied by ⍀, the rotor speed, and the that the state transition matrix can be decomposed into a periodic
stiffness matrix is multiplied by ⍀ 2 . For a constant rotor speed, matrix and the matrix exponential of a constant matrix:
␺ ⫽⍀t, and therefore the governing equations are periodic in time
with period T⫽2 ␲ /⍀. ⌽ 共 t,0兲 ⫽ P 共 t 兲 e Āt (5)
The physical interpretation of all terms appearing in the Appen- where P(t⫹T)⫽ P(t), P(0)⫽I, and
dix is outside the intent of this paper. However, a few observations
deserve mention here and in the analysis of selected results later. 1
All the gyroscopic terms, with one exception, cross couple the Ā⫽ ln共 ⌽ 共 T,0兲兲 (6)
T
degrees-of-freedom. This implies that, owing to rotor rotation, a
motion in 1 dof would induce a gyroscopic motion in another Since P(t) is bounded, the stability of x(t) is governed by the
degree-of-freedom. The exception is the direct gyroscopic term, eigenvalues of Ā, referred to as the characteristic exponents of
g ␥␥ . A yaw motion of the nacelle obviously causes the spinning A(t). The eigenvalue problem is

Fig. 1 Reduced-order SymDyn model showing degrees-of-freedom and geometric


parameters

Journal of Solar Energy Engineering NOVEMBER 2002, Vol. 124 Õ 365

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


Table 1 Turbine models analyzed
Ā⫽V⌳V ⫺1 (7)
Model Degrees-of-freedom
where ⌳ is a diagonal matrix containing the characteristic expo-
nents, ␭ i . Now consider the eigenanalysis of ⌽(T,0), sharing the 1-dof 共 ␥兲
same modal matrix, V: 3-dof 共␾, ␤ 1 , ␤ 2 )
4-dof 共␥, ␾, ␤ 1 , ␤ 2 )
7-dof ( ␶ 1 , ␶ 2, ␶ 3 , ␥, ␾, ␤ 1 , ␤ 2 )
⌽ 共 T,0兲 ⫽V⌺V ⫺1 . (8)
Here, ⌺ is the diagonal matrix of eigenvalues, termed the char-
acteristic multipliers ( ␴ i ) of the system. From 共6兲, the relationship these proposed techniques in our future applications of the Flo-
between 共7兲 and 共8兲 is quet theory. The models we are analyzing in this paper have, at
1 the most, 7 dof. Therefore, we can apply the classical Floquet
⌳⫽ ln⌺ (9) approach directly.
T
which suggests a convenient method to calculate the characteristic Results and Discussion
exponents: We first consider a simple turbine model with only the yaw
degree-of-freedom; this is the simplest model we can have that
1
T 冉
1
␭ i ⫽ ln兩 ␴ i 兩 ⫹j tan⫺1
T
Im共 ␴ i 兲
Re共 ␴ i 兲 冊
⫹2 ␲ k ⫽ ␨ i ⫹j␻ i , (10)
exhibits periodicity. Subsequently, we use the Floquet approach to
compute the modal characteristics of three other turbine models of
increasing complexity, starting with a three-degree-of-freedom
where j⫽ 冑⫺1. The unknown integer, k, appears because there model that allows only hub teeter and flapping of the two blades,
are infinite branches in the logarithm of a complex number. Note and progressing to the full seven-degree-of-freedom model 共see
that when ␴ i is complex, then ␭ i is complex with a corresponding Table 1兲. We used the structural and configuration parameters
complex conjugate 共since ⌽ is real兲. However, when ␴ i is real, listed in Table 2 unless otherwise stated.
then ␭ i is, in general, complex and there need not be a conjugate 1-dof Model: Yaw Only. For this model, the rotor is spinning
to form a complex pair. This behavior of periodic systems is de- at a constant rate with the blades attached rigidly to the hub.
tailed further in 关9兴 and will become apparent in later results. Assuming zero yaw stiffness and damping, and locking all
In view of the foregoing development, the Floquet approach for degrees-of-freedom except yaw, expressions in the Appendix pro-
modal analysis comprises the following four steps: vide the single governing equation:
1. Compute the transition matrix after one period, ⌽(T,0).
From 共3兲 and 共4兲, the transition matrix can be computed by m ␥␥ 共 ␺ 兲 ⌬ ␥¨ ⫹g ␥␥ 共 ␺ 兲 ⌬ ␥˙ ⫽0. (14)
integrating
The yaw inertia, m ␥␥ , and the gyroscopic damping, g ␥␥ , show
⌽̇共t,0兲 ⫽A 共 t 兲 ⌽ 共 t,0兲 (11) a 2 ␺ periodicity. Because ␺ ⫽⍀t, the azimuth represents non-
with linearly independent initial conditions; ⌽(0,0)⫽I, the dimensionalized time. Researchers unfamiliar with periodic sys-
identity matrix. tems may attempt a conventional eigenanalysis with the intent to
2. Compute the characteristic multipliers, ␴ i , by eigenanalysis compute time-varying eigenvalues. Such an analysis provides two
of ⌽(T,0). eigenvalues. One has zero real and imaginary parts, implying an
From 共8兲, the eigenanalysis of ⌽(T,0) computes both the undamped rigid-body yaw mode. The other eigenvalue has a zero
modal matrix, V, and matrix of characteristic multipliers, ⌺. imaginary part, implying zero frequency, and an azimuth-
3. Compute damping and frequency for each mode. dependent real part, implying a time-varying damping. Figure 2
From 共10兲, the modal damping coefficient is given by shows the azimuth-dependent real part from this analysis, normal-
ized with respect to the rotor speed. 共Henceforth, all the eigenval-
1 ues will be shown normalized with respect to the rotor speed.兲 The
␨i⫽ ln兩␴i兩, (12) yaw damping shows a 2p variation. In this form, we cannot de-
T
duce information about the stability of the system, and therefore,
and the modal frequency is given by the conventional eigenanalysis is not appropriate. The actual

冉 冊
mechanism involved is conservation of yaw angular momentum,
1 Im共 ␴ i 兲 while the total system energy is not conserved. Consider the yaw
␻i⫽ tan⫺1 ⫹2 ␲ k . (13) response shown in Fig. 3 due to a unit velocity initial condition.
T Re共 ␴ i 兲
The yaw rate exhibits a 2p variation to conserve the angular mo-
The unknown, k, can be determined by time response of the mentum about the yaw axis given the 2p variation in inertia, m ␥␥ .
system and subsequent frequency identification. The system The instantaneous total energy is not conserved due to a periodic
is stable when ␨ i ⬍0 for all modes. transfer of energy between the generator and the nacelle. This is
4. Identify the modes. made possible by the constraint that rotor speed is constant.
The modal matrix, V, computed in step 2 contains the eigen- The Floquet approach provides the correct modal analysis of
vectors for all modes. Comparing the entries in each mode- the system. This approach yields two zero characteristic exponents
shape helps to identify the dominant motion. Note that the implying two rigid-body modes, both undamped. The first mode is
actual modeshapes are computed from P(t)V and, therefore, a constant yaw-position rigid-body mode and the other is a con-
are periodic in time. Since P(0)⫽I, the modal matrix pro- stant yaw-velocity mode modulated by a time-periodic modal am-
vides the system modeshapes at t⫽0, i.e., blades vertical. plitude. The modal amplitude for this second mode is contained in
the periodic modeshape. The Floquet analysis result is consistent
A key step in the application of the Floquet analysis is the com- with the yaw response in Fig. 3, which shows zero damping
putation of the transition matrix, ⌽. This poses no problem for a behavior.
system with a small number of degrees-of-freedom. However, if
the number of degrees-of-freedom exceeds around 50, computa- 3-dof Model: Teeter and Flap Only. For this model, all joint
tional cost can become overwhelming. A few researchers 关16 –19兴 damping values are assumed to be zero and there is no teeter
have proposed innovative extensions of the Floquet theory to deal spring. The sixth-order linear equations of motion for this system
with large-order systems and improve accuracy. We will exploit are time-invariant 共not periodic兲 because all degrees-of-freedom

366 Õ Vol. 124, NOVEMBER 2002 Transactions of the ASME

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


Table 2 Properties of the analyzed wind turbine

Parameter Description Value


⍀0 Nominal rotor speed 57.5 rpm
␤0 Blade precone angle 7°
d t1 Fore-aft tower hinge offset 9.332 m
d t2 Tower fore-aft hinge to tilt axis distance 15.468 m
d n1 Tilt axis to shaft axis distance 0m
d n2 Yaw axis to teeter axis longitudinal distance 2.388 m
d h1 Teeter axis to flap hinge longitudinal distance ⫺0.375 m
d h2 Shaft axis to flap hinge distance 5.783 m
cb C.o.m. distance of blade, from flap hinge 2.864 m
ch C.o.m. distance of hub, from teeter hinge toward nacelle 0.139 m
mb Mass of each blade 173 kg
mh Mass of hub 共including the non-flapping sections of both blades兲 1 804 kg
I yaw Moment of inertia about the yaw axis 共excludes rotor兲 26 453 kgm2
I hx Moment of inertia of hub about shaft axis 7536 kgm2
I hy Moment of inertia of hub about lateral axis 30 kgm2
I hz Moment of inertia of hub about teeter hinge 7561 kgm2
Ib Moment of inertia of blade about flap hinge 1 469 kgm2
I b-long Moment of inertia of blade about pitch axis 0 kgm2
K ␶ 1, K ␶ 2 Tower fore-aft and lateral hinge stiffness 1.14⫻108 Nm/rad
K ␶3 Tower twist hinge stiffness 4.35⫻107 Nm/rad
K␥ , K␾ Nacelle yaw and hub teeter hinge stiffness 0 Nm/rad
K␤ Blade flap hinge stiffness 3.998⫻105 Nm/rad
C ␶1 , C ␶2 , C ␶3 , Hinge damping 0 Nms/rad
C␥ , C␾ , C␤

are in the rotating frame. Therefore, both a standard eigenanalysis imply that all modes are undamped. The following vectors and
and Floquet analysis produce the same results. At the nominal their conjugates represent the corresponding modeshapes. The ve-
rotor speed, ⍀ 0 ⫽57.5 rpm, the three pairs of complex conjugate locity components are removed.

冋 册冋 册冋 册
eigenvalues 共characteristic exponents兲 are: 兵0⫾3.227j, 0⫾6.364j,
0⫾j0.993其. Teeter 0⬔0° 0.26⬔0° 1.0⬔0°
As in the rest of the paper, these modes are normalized with
respect to the rotor speed. The zero real parts of the exponents Blade # 1 flap 1.0⬔0° 1.0⬔180° 0.002⬔180°
, , .
Blade # 2 flap 1.0⬔0° 1.0⬔0° 0.002⬔0°

The entries in each vector represent the amplitude and phase


contributions from each degree-of- freedom. The first mode is the
collective-flap mode, wherein the blades flap in-phase and with
equal amplitude. The second mode is the differential-flap mode
and here there is a significant coupling of the rotor differential
motion with the teeter motion. The third mode is predominantly a
teeter mode with very little contribution from the rotor differential
motion.
The fan-plot in Fig. 4 shows that all modal frequencies increase
with the rotor speed. The collective mode frequency is predomi-
nantly dictated by the centrifugal stiffening effect associated with
the k ␤␤ term in the stiffness matrix 共see Appendix兲. Due to com-
plete decoupling of flap deflection in this mode we can in fact

Fig. 2 Yaw modal damping variation from conventional modal


analysis

Fig. 3 Yaw response due to unit velocity initial condition Fig. 4 Fan-plot for the 3-dof model „⍀ 0 Ä57.5 rpm…

Journal of Solar Energy Engineering NOVEMBER 2002, Vol. 124 Õ 367

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


calculate the normalized modal frequency as the square root of The vectors are identical except that in the first, corresponding
(k ␤␤ /m ␤␤ ). The frequency asymptotically approaches 1.70p as to the stable teeter mode, the yaw motion is out of phase with
the flap spring stiffness disappears from k ␤␤ : teeter motion. In the unstable mode, the yaw motion is in-phase
with teeter motion. Recall that the actual system modeshapes are

冉 冊 冑
periodic, and that those calculated here correspond to a single
␻␤ k ␤␤ rotor orientation, ␺ ⫽0°, when the teeter axis is perpendicular to
lim ⫽ lim
⍀→⬁ ⍀ ⍀→⬁ m ␤␤ the yaw axis. At this instant the gyroscopic moment transfer from
the rotor to the nacelle is at a maximum. This explains why, al-

⫽ 冑 1
共共 I ⫺I 兲共 c 2 ⫺s 2 兲 ⫹c ␤ m b c b d h2 兲
I b b b long ␤ ␤
though we are analyzing a teeter mode, the yaw amplitude is
about 13 times the teeter amplitude. The large angular momentum
associated with the spinning rotor induces large gyroscopic mo-
⫽1.7048 using the properties in Table 2. ments about the yaw axis with even a small change in the rotor
plane. The mechanism underlying gyroscopic moment transfer is
The differential mode frequency is determined by the hub os- actually somewhat more complicated because the teeter-hinge di-
cillation resisted by the centrifugal pulling of the two blades. One rection alters periodically as the rotor spins.
need not be inordinately concerned with the differential mode. Its We have shown that by adding the yaw degree-of-freedom the
high frequency implies that it would require substantial energy for system is made unstable. In practice, free-yaw teetered-rotor wind
sustenance, and even a small amount of flap damping would eas- turbines are stable, due to large aerodynamic forces that oppose
ily suppress it 共as will be shown in the next test case兲. The teeter structure deflection and velocity. Without provision for aerody-
mode frequency follows the 0.993p trend for all speeds. In fact, it namics in our model, we now investigate the effect of structural
would also be 0.993p if the flap spring were made rigid, owing to stiffness and damping properties on stability.
the weak blade-flap coupling. This property allows us to predict Adding blade damping at the flap hinges has the expected result
the normalized teeter frequency very accurately using the square of increased damping factors in both the collective and differential
root of 共 k ␾␾ /m ␾␾ ) from the Appendix. flap modes; particularly the latter, which has a higher natural fre-
quency. For example, with a flap damping of C ␤ ⫽10 kNms/rad
4-dof Model: Yaw, Teeter, and Flap Only. This model rep- the collective and differential flap characteristic exponents be-
resents an eighth-order periodic system. At the nominal rotor come 兵⫺0.565⫾3.177j, ⫺2.373⫾5.913j其, respectively. The yaw
speed, the Floquet analysis yields the following characteristic ex- and teeter modes are not noticeably affected at all, due to weak
ponents: 兵0, 0, 0⫾3.227j, 0⫾6.371j, ⫺0.0956⫺1j, 0.0956⫹1j其. blade-flap coupling.
The first two exponents correspond to the same yaw modes that
were calculated in the first test case; two rigid-body modes. Nei-
ther teeter nor flap motions participate in these modes. The second
and third pairs of exponents represent the rotor collective and
differential-flap modes and are hardly affected by the yaw degree-
of-freedom.
The last two eigenvalues represent a split teeter mode, with the
first stable and the second unstable. The stable mode has a decay
rate of 0.0956p and the unstable mode has a growth rate of equal
magnitude. However, both modes have exactly the same fre-
quency of 1p. The stable mode pumps energy into the unstable
mode. Because there is no external source of excitation, this is a
self- excited instability. The corresponding modeshapes for these
two modes are:

冋 册冋 册
Yaw 1.0⬔180° 1.0⬔0°
Teeter 0.076⬔0° 0.076⬔0°
, .
Blade # 1 flap 0.0003⬔0° 0.0003⬔0°
Fig. 6 Effect of precone and teeter stiffness on teeter mode
Blade # 2 flap 0.0003⬔180° 0.0003⬔180° stability

Fig. 5 Effect of teeter damping on teeter and differential-flap


mode stability Fig. 7 Effect of yaw stiffness on modal damping

368 Õ Vol. 124, NOVEMBER 2002 Transactions of the ASME

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


Including teeter damping alone has no effect on the yaw and Thus, we see a yaw frequency of ␻ ␥ locked in with a teeter
collective flap modes but does stabilize the teeter and differential frequency of 1p-␻ ␥ . When K ␥ exceeds 10.1 kNm/rad both the
flap modes, as shown in Fig. 5. Recall that a positive real part of yaw and the teeter modal damping levels fall to zero 共critical
a characteristic exponent indicates instability. Here we see that the stability兲. The yaw spring effect dominates over the gyroscopic
differential mode and one of the teeter modes are stable for all effect and the yaw frequency increases as expected. The teeter
values of teeter damping, while the second teeter mode is only frequency converges to the 0.993p value as in the case when yaw
stable for damping values of 31 kNms/rad or greater. Meanwhile, was fixed.
the teeter mode frequencies remain locked at 1p, while the differ- Finally, we examine the effect of yaw damping on the structure.
ential mode frequency decreases slightly from 6.371p to 6.348p The yaw rigid-body mode and collective flap modes are unaf-
over the teeter damping range 共not plotted兲. fected, as expected. With increased yaw damping the differential
Up until this point, the blades have been set at the nominal flap becomes strictly stable and its modal frequency decreases
precone of ␤ 0 ⫽7° and teeter stiffness has been zero. The next slightly. The 共already兲 stable teeter mode continues to remain
study investigates what happens to teeter stability when the pre- stable with frequency constant at 1p. The second 共unstable兲 teeter
cone angle is changed between 0° and 10°, and the teeter stiffness mode and second yaw mode vary as shown in Figs. 9 and 10.
is changed between 0 kNm/rad and 20 kNm/rad. All the damping Similar to the case when yaw stiffness was varied we notice that
values and the yaw stiffness are still assumed zero. The variation the teeter and yaw modes coalesce beyond a yaw damping, C ␥ , of
of teeter mode stability is illustrated in Fig. 6. In fact, two teeter 98 kNms/rad. However, as yaw damping is increased the modal
modes exist but we only plot the unstable one—the stable mode damping converges to but never reaches zero and therefore the
has a real part of equal magnitude but opposite sign. For the case system is always unstable. Once again, we see frequency locking
when the teeter stiffness is zero, we see that the teeter mode is between the teeter and yaw modes during coalescence 共Fig. 10兲.
unstable for all precone angles examined. There is, however, a The teeter modal frequency converges to 0.993p as before.
point at which the exponential growth rate is minimal. This occurs
at a precone angle of 1.59°. With this knowledge, wind turbine 7-dof Model. We now introduce the three tower degrees-of-
designers could maximize natural stability of the structure. As the freedom to obtain the 7-dof turbine model. We eliminated the yaw
teeter spring stiffness is increased beyond approximately 4 kNm/ and teeter springs and set all other parameters to the nominal
rad an increasing range of precone angles exist at which the teeter values listed in Table 2. Results show that the yaw, teeter, and
mode is critically stable 共real part of characteristic exponent is rotor differential modes remain relatively unaffected by the inclu-
zero兲. Increasing teeter stiffness to at least 17 kNm/rad provides sion of the tower degrees-of-freedom. At the rotor nominal speed,
critical stability at the nominal precone angle of 7°. the rotor collective, tower fore-aft bending, tower lateral bending,
Next, we study the effect of yaw stiffness on the turbine modal and tower torsion modes are 兵0⫾3.282j, 0⫾1.081j, 0⫾1.081j,
behavior 共with the precone angle kept fixed at 7°兲. As expected, 0⫾169.0j其, which indicate zero damping. The tower has identical
the rotor collective and differential modes remain relatively unaf- properties in the fore-aft and lateral dimensions, which gives rise
fected. We therefore focus our attention on the yaw and teeter to the identical natural frequency of 1.08p. The corresponding
modes. Figure 7 shows the modal damping variation with yaw modeshapes are:
stiffness, and Fig. 8 shows the modal frequency variation 共note the
two vertical axes in the latter plot兲. Here we show only the un-
stable modes — the stable modes simply trace reflections in the
horizontal axes. At zero yaw stiffness, we notice the unstable tee-
ter mode with frequency 1p and the rigid-body yaw mode with
zero damping and frequency, as identified earlier. As the yaw stiff-
ness increases within the range 0⬍ K ␥ ⬍7.5 kNm/rad, the teeter
damping coefficient decreases rapidly, whereas the yaw mode
damping coefficient increases slowly. The teeter and yaw mode
frequencies remain fixed, indicating the dominance of gyroscopic
loads over the yaw spring force. When K ␥ ⫽7.5 kNm/rad, the
teeter and the yaw modes coalesce and remain so over the range,
7.5⬍ K ␥ ⬍10.1 kNm/rad. The fixed-frame observed yaw fre-
quency, ␻ ␥ , is seen as a 1p ⫾ ␻ ␥ frequency by the rotor in the
rotating frame. The 1p-␻ ␥ motion of the nacelle, as seen by the
rotor, couples with the hub teeter motion and gets locked with it.

Fig. 9 Effect of yaw damping on modal damping

Fig. 8 Effect of yaw stiffness on modal frequencies Fig. 10 Effect of yaw damping on modal frequencies

Journal of Solar Energy Engineering NOVEMBER 2002, Vol. 124 Õ 369

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


冤 冥冤 冥冤 冥冤 冥
Tower fore-aft 0.01⬔180° 1.0 0.40⬔90° 0
Tower lateral 0 0.40⬔90° 1.0 0
Tower twist 0 0 0 1
Yaw 0 0.01⬔⫺90° 0.003 1⬔180°
, , , .
Teeter 0.007 1.0⬔90° 0.40⬔⫺90° 0
Blade # 1 flap 1.0 0.70 0.19⬔90° 0
Blade # 2 flap 1.0 0.68 0.17⬔90° 0

The collective mode, represented by the first modal vector, modal properties. Efforts are underway to include the unsteady
shows the rotor collective motion coupled with the expected out- aerodynamics in our model, and we will be using this model for
of-phase tower fore-aft motion. The modal frequency is almost more elaborate aeroelastic stability analyses in the future. The
identical to the 4-dof case. Teeter motion also participates in this methodology described in this paper would be directly applicable.
mode. This is because the fore-aft motion of the tower tends to
oscillate the rotor disk plane, but the rotor resists this change
owing to its gyroscopic inertia. In the second modal vector, we see
Appendix
that the tower fore-aft motion couples with all the motions except

冋 册
the tower twisting. Coupling with the teeter motion is particularly The linearized equations for the 4-dof turbine model are:
strong due to gyroscopic inertia of the rotor as in the previous
m ␥␥ m ␥␾ m ␥␤ ⫺m ␥␤
mode. Interestingly, the tower fore-aft motion also induces a sig-
nificant tower lateral motion, resulting in an elliptical path traced m ␥␾ m ␾␾ m ␾␤ ⫺m ␾␤
out by the tower top when viewed from above. The third modal m ␥␤ m ␾␤ m ␤␤ 0 ⌬ q̈
vector shows that the dominant lateral tower motion is quite
strongly coupled with tower fore-aft bending, nacelle yaw, and ⫺m ␥␤ ⫺m ␾␤ 0 m ␤␤

冋 册
hub teeter motions, as in the previous mode. The fourth vector
represents a pure twisting of the tower. Because there is no yaw g ␥␥ g ␥␾ g ␥␤ ⫺g ␥␤
spring to transfer the tower twist motion to the nacelle, the
nacelle-rotor substructure tends to stay fixed in space. This pro- g ␾␥ g ␾␾ 0 0
duces an apparent out-of-phase yaw motion of the nacelle with ⫹⍀ g ␤␥ 0 g ␤␤ 0 ⌬ q̇
respect to the tower top. The high modal frequency 共169.0p兲 is
⫺g ␤␥ 0 0 g ␤␤

冋 册
dependent only on the tower-twist spring stiffness and the tower
longitudinal moment-of-inertia.
We also performed a Floquet analysis on this model at other k ␥␥ k ␥␾ k ␥␤ ⫺k ␥␤
rotor speeds. The three tower modal frequencies remained virtu-
ally unaffected, owing to the weak coupling with centrifugal and 0 k ␾␾ k ␾␤ ⫺k ␾␤
gyroscopic loads. ⫹⍀ 2 0 k ␾␤ k ␤␤ 0 ⌬ q⫽0
0 ⫺k ␾␤ 0 k ␤␤
Concluding Remarks
We have examined the modal behavior of a teetered-rotor tur- where ⌬q⫽ 关 ⌬ ␥ ⌬ ␾ ⌬ ␤ 1 ⌬ ␤ 2 兴 T.To simplify terms, we use the
bine using simple models, ranging from one to seven degrees-of- following notation: s ␤ ⫽sin ␤0 and c ␤ ⫽cos ␤0 . Other symbol
freedom. We showed that the governing equations are periodic definitions can be found in Table 2.
and that a Floquet analysis must be used to correctly predict the The entries in the mass matrix are:
modal behavior, in particular the damping and therefore stability.
Frequency results illustrate the dominant centrifugal loads affect- m ␥␥ ⫽I yaw ⫹sin2 ␺ 共 I hz ⫹2I b 兲 ⫹cos2 ␺ 共 I hy ⫹2s ␤2 I b ⫹2c ␤2 I b long 兲
ing flap motion as rotor speed is increased. Gyroscopic coupling is
evident when both yaw and teeter degrees-of-freedom are present. ⫹m h 共 d n2 ⫺2c h 兲 2 ⫹ 2m b 共共 d n2 ⫺d h1 兲 2 ⫹d h2 sin2 ␺ 共 d h2
In fact, this coupling causes the system to be unstable without
⫹2c ␤ c b 兲 ⫹2s ␤ c b 共 d n2 ⫺d h1 兲兲
stiffness or damping present at the yaw and teeter hinges. This is
ordinarily provided by aeroelasticity. m ␥␾ ⫽sin␺ 共 I hz ⫹2I b ⫹m h c h 共 c h ⫺d n2 兲 ⫹2m b 共 d 2h1 ⫹d 2h2 ⫺d h1 d n2
Parametric studies were performed through variation of struc-
tural stiffness and damping properties. We found that yaw damp- ⫹c b 共 2c ␤ d h2 ⫹s ␤ 共 d n2 ⫺2d h1 兲兲兲兲
ing alone cannot stabilize the system but enough yaw or teeter
stiffness can provide critical stability. Teeter damping was the m ␾␾ ⫽I hz ⫹m h c 2h ⫹2I b ⫹2m b 共 d 2h1 ⫹d 2h2 ⫹2c b 共 c ␤ d h2 ⫺s ␤ d h1 兲兲
only parameter that was found to strictly stabilize the turbine m ␥␤ ⫽sin␺ 共 I b ⫹m b c b 共 c ␤ d h2 ⫹s ␤ 共 d n2 ⫺d h1 兲兲兲
model. At certain levels of yaw stiffness or damping, the teeter
and yaw modes coalesce, wherein modal damping is identical and m ␾␤ ⫽I b ⫹m b c b 共 c ␤ d h2 ⫺s ␤ d h1 兲
frequencies are locked together. A study of blade precone angle
variation showed that improvements in stability could be made in m ␤␤ ⫽I b
the turbine design phase. Active feedback control could poten- The entries in the gyroscopic/damping matrix are:
tially improve other stability characteristics.
Due to the rigid-body assumptions in the SymDyn model, g ␥␥ ⫽sin2 ␺ 共 I hz ⫺I hy ⫹2c ␤2 共 I b ⫺I b long 兲
higher order modes of the tower and blades are ignored. The in-
clusion of these is unlikely to alter the results substantially. The C␥
⫹2m b d h2 共 d h2 ⫹2c ␤ c b 兲兲 ⫹
addition of aerodynamics, however, will have a big effect on the ⍀

370 Õ Vol. 124, NOVEMBER 2002 Transactions of the ASME

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms


关2兴 Malcolm, D.J., and James, G.H., 1996, ‘‘Identification of Natural Operating
g ␥␾ ⫽cos␺ 关 I hz ⫹I hy ⫺I hx ⫹2m h c h 共 c h ⫺d n2 兲 ⫹2s ␤2 共 2I b ⫺I b long 兲 Modes of HAWTs From Modeling Data,’’ Proc. of 15th ASME Wind Energy
Symp., pp. 24 –31.
⫹2c ␤2 I b long ⫹4m b 共 d 2h1 ⫺d h1 d n2 ⫹s ␤ c b 共 d n2 ⫺2d h1 兲兲兴 关3兴 Lobitz, D.W., and Sullivan, W.N., 1980, ‘‘VAWTDYN: A Numerical Package
for the Dynamic Analysis of Vertical Axis Wind Turbine,’’ Technical Report
g ␥␤ ⫽cos␺ 共 s ␤2 共 2I b ⫺I b long 兲 ⫹c ␤2 I b long ⫹2s ␤ m b c b 共 d n2 ⫺d h1 兲兲 SAND-80-0085, Sandia National Laboratories, Albuquerque, NM.
关4兴 Carne, T.G., Lobitz, D.W., Nord, A.R., and Watson, R.A., 1982, ‘‘Finite Ele-
g ␾␥ ⫽cos␺ 共 I hz ⫺I hy ⫹I hx ⫹2c ␤2 共 2I b ⫺I b long 兲 ⫹2s ␤2 I b long ment Analysis and Modal Testing of a Rotating Wind Turbine,’’ Presented at
the AIAA/ASME/ASCE/AHS Structures – Structural; Dynamics and Materials
⫹4m b d h2 共 d h2 ⫹2c ␤ c b 兲兲 Conf., New Orleans, LA.
关5兴 Carne, T.G., Martinez, D.R., and Ibrahim, S.R., 1983, ‘‘Modal Identification of
C␾ a Rotating Blade System,’’ Technical Report SAND82–2115, UC–32, Sandia
g ␾␾ ⫽ National Laboratories, Albuquerque, NM.
⍀ 关6兴 James, G.H., Carne, T.G., and Lauffer J.P., 1993, ‘‘The Natural Excitation
Technique 共NExT兲 for Modal Parameter Extraction From Operating Wind Tur-
g ␤␥ ⫽cos␺ 共 c 2␤ 共 2I b ⫺I b long 兲 ⫹s 2␤ I b long ⫹4c ␤ m b c b d h2 兲 bines,’’ Technical Report SAND92–1666, UC–261, Sandia National Labora-
tories, Albuquerque, NM.
C␤ 关7兴 Yamane, T., 1987, ‘‘Coupled Rotor/Tower Stability Analysis of a 6-Meter Ex-
g ␤␤ ⫽ perimental Wind Turbine,’’ Proc. of 6th ASME Wind Energy Symp., Dallas,
⍀ TX, pp. 59– 66.
The entries in the stiffness matrix are: 关8兴 Bir, G.S., and Butterfield, C.P., 1997, ‘‘Modal Dynamics of a Next-Generation
Flexible-Rotor Soft-Tower Wind Turbine,’’ Proc. of 15th Int. Modal Analysis
Conf., Orlando, FL, pp. 76 – 84.
K␥
k ␥␥ ⫽ 关9兴 Johnson, W.J., 1980, Helicopter Theory, Princeton Univ. Press, Princeton, NJ.
⍀2 关10兴 Bir, G.S., and Robinson, M., 1999, ‘‘Code Development for Control Design
Applications 共Phase I: Structural Modeling兲,’’ Proc. of 18th ASME Wind En-
k ␥␾ ⫽sin␺ 关 I hx ⫺I hy ⫹2 共 I b ⫺I b long 兲共 c ␤2 ⫺s ␤2 兲 ⫹m h c h 共 d n2 ⫺c h 兲 ergy Symp., Reno, NV, pp. 114 –123.
关11兴 Hodges, D., and Patil, M., 2000, ‘‘Multi-Flexible Body Analysis for Applica-
⫹2m b 共 d 2h2 ⫺d 2h1 ⫹d h1 d n2 ⫹c b 共 s ␤ 共 2d h1 ⫺d n2 兲 ⫹2c ␤ d h2 兲兲兴 tion to Wind Turbine Control Design,’’ Proc. of 19th ASME Wind Energy
Symp., Reno, NV, pp. 110–120.
关12兴 Stol, K., 2001, ‘‘Dynamics Modeling and Periodic Control of Horizontal-Axis
k ␥␤ ⫽sin␺ 共共 I b ⫺I b long 兲共 c ␤2 ⫺s ␤2 兲 ⫹m b c b 共 s ␤ 共 d h1 ⫺d n2 兲 ⫹c ␤ d h2 兲兲
Wind Turbines,’’ Ph.D. thesis, Univ. of Colorado at Boulder, Boulder CO.
关13兴 Bir, G., and Stol, K., 2000, ‘‘Modal Analysis of a Teetered-Rotor Wind Turbine
k ␾␾ ⫽I hx ⫺I hy ⫺m h c 2h ⫹2 共 I b ⫺I b long 兲共 c 2␤ ⫺s 2␤ 兲 ⫹2m b 共 d 2h2 ⫺d 2h1 using the Floquet Approach,’’ Proc. of 19th ASME Wind Energy Symp., Reno,
NV, pp. 23–33.
K␾ 关14兴 Stol, K., and Bir, G.S., 2000, ‘‘Validation of a Symbolic Wind Turbine Struc-
⫹2c b 共 s ␤ d h1 ⫹c ␤ d h2 兲兲 ⫹ tural Dynamics Model,’’ Proc. of 19th ASME Wind Energy Symp., Reno, NV,
⍀2 pp. 41– 48.
关15兴 Nayfeh, A.H., and Mook, D.T., 1979, Nonlinear Oscillations, John Wiley &
k ␾␤ ⫽ 共 I b ⫺I b long 兲共 c ␤2 ⫺s ␤2 兲 ⫹m b c b 共 s ␤ d h1 ⫹c ␤ d h2 兲 Sons, New York, NY.
关16兴 Sinha, S. C., and Wu, D. H., 1991, ‘‘An Efficient Computational Scheme for
K␤ the Analysis of Periodic Systems,’’ J. Sound Vib. 151, pp. 91–117.
k ␤␤ ⫽ 共 I b ⫺I b long 兲共 c ␤2 ⫺s ␤2 兲 ⫹c ␤ m b c b d h2 ⫹ 关17兴 Bauchau, O.A., and Nikishkov, Y.G., 1998, ‘‘An Implicit Floquet Analysis for
⍀2 Rotorcraft Stability Evaluation,’’ Proc. of 54th AHS Annual Forum, Washing-
ton, DC, pp. 674 – 683.
关18兴 Holley, W.E., and Bahrami, M., 1981, ‘‘Periodic Linear Systems Forced by
References White Noise With Wind Turbine Applications,’’ Proc. of 15th Asilomar Conf.
关1兴 James, G.H., 1994, ‘‘Extraction of Modal Parameters From an Operating on Circuits, Systems, and Computers, Pacific Grove, CA, pp. 117–122.
HAWT using the Natural Excitation Technique 共NExT兲,’’ Proc. 13th ASME 关19兴 Peter, D. A., 1994, ‘‘Fast Floquet Theory and Trim for Multi-Blade Rotor-
Wind Energy Symp., pp. 227–232. craft,’’ J. Am. Helicopter Soc. 39-4, pp. 82– 89.

Journal of Solar Energy Engineering NOVEMBER 2002, Vol. 124 Õ 371

Downloaded From: http://solarenergyengineering.asmedigitalcollection.asme.org/ on 07/20/2014 Terms of Use: http://asme.org/terms

You might also like