Leithead, W.E. and Neilson, Victoria and Dominguez, S. and Dutka, Arkadiusz (2009) A novel approach
to structural load control using intelligent actuators. In: 17th Mediterranean Conf. on Control and
Automation, 24-26 June 2009, Thessaloniki, Greece.
http://strathp
rints.strath.a
c.uk/28425/
Strathprints is designed to allow users to access the
research output of the University of
Strathclyde. Copyright © and Moral Rights for the papers on this site are retained by the
individual authors and/or other copyright owners. You may not engage in further
distribution of the material for any profitmaking activities or any commercial gain. You
may freely distribute both the url (http://strathprints.strath.ac.uk
) and the content of this
paper for research or study, educational, or not-for-profit purposes without prior
permission or charge. You may freely distribute the url (http://strathprints.strath.ac.uk)
of the Strathprints website.
Any correspondence concerning this service should be sent to The
Strathprints Administrator: eprints@cis.strath.ac.uk
A novel approach to structural load control using
Intelligent Actuators
W.E.Leithead & V.Neilson
Dept. of Electrical and Electronic Engineering
University of Strathclyde
Glasgow, UK
w.leithead@eee.strath.ac.uk
S.Dominguez & A.Dutka
MLS
Glasgow, UK
sdominguez@mls-control.com
Abstract— The recent trend towards large multi-MW wind
turbines resulted in the role of the control system becoming
increasingly important. The extension of the role of the
controller to alleviate structural loads has motivated the
exploration of novel control strategies, which seek to maximise
load reduction by exploiting the blade pitch system. The
reduction of blade fatigue loads through individual blade pitch
control is one of the examples. A novel approach to reduction
of the unbalanced rotor loads by pitch control is presented in
this paper. Each blade is equipped with its own actuator,
sensors and controller. These local blade control loops operate
in isolation without a need of communication with each other.
The single blade control approach to regulation of unbalanced
rotor loads presented in this paper has an important advantage
of being relatively easy to design and tune. Furthermore, it
does not affect the operation of the central controller and the
latter need not be re-designed when used in conjunction with
the single blade controllers. Their performance is assessed
using BLADED simulations.
Keywords-control;individual pitch;unbalanced rotor loads;
I. INTRODUCTION
Wind turbine technology has undergone a rapid
development over the last three decades, largely driven by
concerns over the environment. In the last decade, the power
rating and size of wind turbines has increased rapidly, and
turbines rated at 5MW with 120m rotor diameter are
commercially available today. With the increased turbine
size and structural flexibility, greater demands have been
placed on the control system to alleviate asymmetric loads
on the rotor.
As a wind turbine blade sweeps through the wind-field, it
experiences loads caused by the rotational sampling of the
wind-field. These
0
n
loads are concentrated at integer
multiples (n) of the rotor speed (
0
) and consist of both
deterministic and stochastic components. The stochastic
component largely arises from the turbulence of the wind.
The deterministic loads largely arise from wind shear, tower
shadow and blade imbalance [1]. In combination the
rotationally sampled blade loads result in an unbalanced
rotor loads which not only impact on the blades and rotor but
on the rest of the wind turbine structure and the drive-train.
The most significant components of these loads are typically
those at
0
1
,
0
2
and
0
3
.
Individual blade pitch control has demonstrated great
potential to alleviate rotor loads in above rated wind speed
operation [2, 3, 4]. While these reported algorithms may
differ in structure or implementation details, all aim to
reduce the asymmetric loads by varying the pitch angle of
each blade individually in response to some suitable
measurement such as blade bending moments. Improvements
in sensor technology are now making these individual pitch
control algorithm a practical possibility [2]. In Bell et al [2]
significant reductions in fatigue equivalent loads on the
blade, the main shaft and the yaw bearing are reported. In
previously reported approaches to individual pitch control,
individual pitch control is realised through the wind turbine
central controller. The loads on each blade are measured,
communicated to some controller which then determines the
pitch angle demand for each blade using all the
measurements. The direct-quadrature (d-q) axis
transformation [3] is central to this procedure. In this paper a
novel approach to reducing the unbalanced rotor loads by
individual pitch control is presented. Each blade has its own
pitch control system operating in isolation from the wind
turbine central controller. The objective for this SISO
feedback loop is chosen so that only the contribution to rotor
imbalance is regulated. An incremental adjustment to the
pitch demand from the collective pitch demand from the
central controller is determined for the blade using only the
measurement of the load on that blade. The instrumentation
required for each blade is bending moment sensors, typically
optical fibre sensors, and linear and angular acceleration
sensors to determine the tower motion.
The paper is structured as follows. In section II, the
conventional individual pitch control design based on the d-q
axis transformation is discussed, followed by an introduction
of the single blade controller in section III. In Section IV, the
dynamic model for a single blade is presented. This includes
the fact that the blade is coupled to the dynamics of the
whole wind turbine. Also, the required modification to the
dynamics of the blade in terms of fictitious forces dependent
on measured accelerations is introduced. The single blade
model is validated in section V. The design issues associated
with Individual Pitch Control system based on single blade
17th Mediterranean Conf. on
Control and Automation
control is discussed in Section VI and conclusions drawn in
Section VII.
II. C
ONVETIONAL INDIVIDUAL PITCH CONTROL
The reasons behind a drive towards individual pitch
control have its origins in blade loads being dependent on the
azimuth and wind conditions as seen by the blade. Rotating
blades sample the uneven wind-field resulting in significant
load variation. These include deterministic components such
as tower shadow and wind sheer, and stochastic components
that result from the turbulence. Most of these loads are
concentrated around multiples of rotor speed. The most
significant blade loads are concentrated at 1Ω0 frequency, as
seen in Figure 1.
Figure 1: Rotor load imbalance spectrum
The rotor load imbalance may be tackled through the
individual pitch control. It is intended to be used during
above rated operation, where loads are the highest. The
results reported so far demonstrate great potential to reduce
asymmetric loading on the rotor. The individual pitch control
algorithms have previously been embedded in the central
controller, which necessitate careful tuning to the specific
turbine. These control methods employ d-q transformation
that has its origins in three-phase electrical machine theory
[3]. It is the co-ordinate system transformation from the
three-vector,
T
cba
XXX ][
, to the two-vector,
T
qd
XX ][
, such that
c
b
a
q
d
X
X
X
X
X
)sin()sin()sin(
)cos()cos()cos(
3
2
3
4
3
2
3
4
3
2
The inverse transformation is
q
d
c
b
a
X
X
X
X
X
)sin()cos(
)sin()cos(
sincos
3
4
3
4
3
2
3
2
The d-q transformation enables a degree of separation of
the design of two controllers that apply to orthogonal
directions of imbalance on the rotor. The tuning of the
controller depends on full wind turbine dynamics including
the interaction between the blade and the rest of the flexible
structure. The central controller shown in the Figure 2
contains the d-q transformation and outputs three blade
position setpoints.
Figure 2: Individual Pitch Control concept
III.
SINGLE BLADE CONTROLLER: THE INTELLIGENT
ACTUATOR
The individual pitch control approach taken in this paper
requires each blade to have its own local pitch controller.
The central controller sets the average demand for the pitch
angles as required to control the speed; the blade controllers
make incremental adjustments to this average. The overall
concept is illustrated in Figure 3. Note that bending moments
M1-3 are not sent to the central controller, but are utilised by
the local controllers. This results in the central controller
providing only collective control for the pitch that is aimed at
regulating the rotor speed. Local controllers for each blade
will act upon variation of the bending moments and regulate
these loads at a desired level.
Figure 3: Single Blade Control concept
The conventional actuator has task of driving the blade
pitch angle β
a
, to demanded pitch, β
b
. The central controller
contains integral action that drives the difference between the
demanded generator speed ω
d
and actual ω
a
to zero. The
requirement for, what is called a cascade control loop is that
the inner loop is markedly faster than the outer loop.
However, this requirement is not always sufficiently fulfilled
with the bandwidth of the outer loop in the region of 1 rad/s
and the pitch actuator of the bandwidth of a region between 4
1Ω
0
frequency
rad/s to 9 rad/s. That may sometimes result in an undesirable
coupling between loops and special care should be taken
during the design of both loops.
Figure 4: Single Blade Controller – conventional actuator
The controller structure used by the Intelligent Actuator is
shown in Figure 5. The feedback loop in the actuator acts on
blade root bending moment with the central controller
enclosed in an outer feedback loop. The moment actuator
sets the actual moment M
a
to the demanded moment M
d
.
Figure 5: Single Blade Controller – Intelligent Actuator
The moment disturbances shown in Figure 6 are rejected
by the inner loop controller through pitch manipulation. The
pitch actuator plus blade feedback loop can be considered to
be a modified actuator. The rotor speed disturbances are
attenuated by the central controller that forms the outer loop.
Figure 6: Single Blade Controller, disturbance rejection loops
The design of the two loops may be carried out
independently if there is a sufficient loop bandwidth
separation and there is no interaction between controlled
system dynamics in the inner and outer loop. This
unfortunately is not the case here. Pitch actuator constraints
pose a limitation for the inner loop bandwidth that aims at
achieving the bandwidth of roughly 2Ω
0
, where Ω
0
is of
about 2 rad/s. Outer loop aims to regulate rotor speed with a
bandwidth around 1 rad/s and tower loads close to tower
frequency, at about 2 rad/s. It is also clear that blade
dynamics interact with the dynamics of the wind turbine,
therefore inner and outer loop plant dynamics are not
independent. In the next section a method of decoupling
blade dynamics from the rest of the turbine will be outlined.
With that the design of the single blade controller only
depends on the dynamics of a single blade. This has the
advantages of being structurally simple and easy to
implement and tune.
IV.
FULL BLADE MODEL
As was already mentioned, the blade motion will interact
with the rest of the turbine. The coordinate system associated
with the blade is not inertial. The dynamics in a non-inertial
frame are the dynamics in an inertial reference frame plus
fictitious forces proportional to the relative acceleration of
the reference frames. The non-inertial reference frame moves
linearly with tower head, rotationally with the nacelle and
rotates with the rotor. To be able to compute fictitious forces
acceleration measurement of the movement of non-inertial
reference frame is required. Accelerometers will measure
acceleration resulting from the movement of the turbine and
the earth’s gravity. Consequently, the contribution of the
gravitational force is included in the fictitious forces.
The full non-linear model [5] of the blade including the
coupling to the rest of the wind turbine dynamics is:
R
R
T
T
A
A
R
R
FEFE
FEFE
R
R
M
M
J
M
M
J
cscs
cssc
11
222222
222222
R
R
FEFE
FEFE
PO
PI
cscs
cssc
J
M
M
222222
222222
/
/
with the fictitious forces:
yR
zR
B
B
b
T
T
J
a
a
lm
M
M
R
R
3
2
The in-plane and out-plane angles of deflection of the
blade are
R
and
R
and
is a pitch angle. J is blade inertia
and
E
and
F
are blade and flap frequencies, respectively.
The in-plane and out-of-plane blade root bending moments
are denoted as
PI
M
/
and
PO
M
/
.
R
A
M
and
R
A
M
are the in-
plane and out-of-plane aerodynamics moments.
2B
a
is the
acceleration of the centre of rotation of the blade
perpendicular to the blade in the plane of rotation and
3B
a
is
the acceleration of the centre of rotation of the blade
perpendicular to the plane of rotation.
zR
and
yR
are
rotational accelerations measured at the origin of the rotor
plane (hub),
B
m is the blade mass, l is the distance between
the blade’s centre of mass and the centre of rotation of the
rotor.
More appropriately for the purpose intended here, this
model can also be expressed in terms of the namely in-plane
and out–of-plane moments:
RR
RR
TA
TA
PO
PI
PO
PI
MM
MM
A
M
M
A
M
M
/
/
/
/
PO
PI
R
R
M
M
A
J
/
/
1
1
where
222222
222222
cscs
cssc
A
FEFE
FEFE
222222
222222
22
1
1
sccs
cscs
A
FEFE
FEFE
FE
The derivation of this model is explained in [5].
Now, the control system should be modified by
subtracting the contribution of the fictitious forces from the
measured bending moment as shown in Figure 7. The
fictitious forces are derived directly from measured
accelerations.
Figure 7: Feedback Control System for the Blade Model with Tower
Dynamics Deducted
V. BLADE MODEL VALIDATION
To be able to validate the blade model a comparison
between the simulation results obtained from Bladed and the
response of a single blade model was carried out. The main
problem associated with this comparison is inability of
obtaining the magnitude of the aerodynamic moment that is
seen by the blade. The procedure undertaken here is as
follows. Fictitious forces are calculated using accelerations
of the non-inertial reference frame extracted from Bladed.
The hub wind speed is also extracted from Bladed and then
fed into the single blade model and used to calculate the
blade bending moments therein. The point wind speed is
modified by the model shown in Figure 8 to generate the
effective wind speed as seen by the blade.
Figure 8: Wind model
The effective wind speed is augmented by 1 Ω
0
and 2 Ω
0
cyclic components prior to calculating the out-of-plane and
in-plane aerodynamic moments in the usual manner. The
blade model described in the previous section is
implemented in Simulink along with the spatial filter and
fictitious force models.
The simulation results obtained in Simulink are compared
with Bladed results. Comparison of bending moment spectra
is shown in Figure 9. A good match for frequencies range of
interest is achieved, which confirms that the model may be
used for the control design.
Figure 9: Blade model validation
60
80
100
120
140
160
Magnitude (dB)
10
-2
10
-1
10
0
10
1
10
2
-225
-180
-135
-90
-45
0
Phase (deg)
Bode Diagram
Frequency (rad/sec)
Figure 10: Bode plot of linearised blade model
The validated non-linear model needs to be linearised to
extract the local characteristic that will be used for the linear
control design. The Bode plot of the system is shown in
Figure 10. The model exhibits dominant 2-nd order system
behaviour.
VI.
CONTROL SYSTEM DESIGN
The controller for the blade is designed to achieve the
following objectives:
Frequency (rad/s)
Out-plane blade load at mean wind speed = 18m/s
Bladed
Simulink
Bending
Moment
R
R
z
y
;
yR
zR
Controller
Blade
Model
Pitch
Angle
-
Fictitious Load
Dynamics
-
0
