Proceedings of

Proceedings of the
the 20th
20th World
World Congress
Congress
Proceedings
The of the
The International
International 20th World
Federation
Federation of Congress
of Automatic
Automatic Control
Control
Proceedings
The of the
International 20th Worldof Congress
Toulouse,
Toulouse, France,Federation
France, July 9-14,
July Automatic
9-14, 2017
2017 Control
Available online at www.sciencedirect.com
The International Federation of
Toulouse, France, July 9-14, 2017Automatic Control
Toulouse, France, July 9-14, 2017
ScienceDirect
IFAC PapersOnLine 50-1 (2017) 9914–9919
Observer Based Blade-Pitch Control of
Observer
Observer Based Blade-Pitch
Basedoperating
Blade-Pitch Control
Control of
ofa
Wind
Wind Turbines
Turbines operating above
above rated:
Wind Turbines operating
preliminary above rated: a
study
rated: a
preliminary
preliminary study
study
∗ ∗∗ ∗∗
M.L.
M.L. Corradini
Corradini ∗ G. G. Ippoliti
Ippoliti ∗∗ G.
G. Orlando
Orlando ∗∗
M.L. Corradini ∗∗∗ G. Ippoliti ∗∗ ∗∗ G. Orlando ∗∗ ∗∗
M.L. Corradini G. Ippoliti G. Orlando ∗∗ ∗∗
∗
∗ Scuola di Scienze e Tecnologie, Università di Camerino, via
∗ Scuola di Scienze e Tecnologie, Università di Camerino, via
∗ Scuola di Scienze e Tecnologie, Università di Camerino, via
Madonna
∗
Madonna
Scuola didelledelle Carceri,
Carceri,
Scienze 62032 Camerino
62032
e Tecnologie, Camerino
Università (MC),
(MC), Italy, ee mail:
Italy,
di Camerino, mail:
via
Madonna delle Carceri, 62032 Camerino (MC), Italy, e mail:
letizia.corradini@unicam.it
letizia.corradini@unicam.it
Madonna delle Carceri, 62032 Camerino (MC), Italy, e mail:
∗∗ Dip. di Ingegnerialetizia.corradini@unicam.it
∗∗
Dip. di Ingegneria dell’Informazione,
dell’Informazione, Università
∗∗ Dip. di Ingegnerialetizia.corradini@unicam.it
∗∗ Università Politecnica
Politecnica delle delle
∗∗ Marche,
Marche, via
via Brecce
Brecce dell’Informazione,
Bianche,
Bianche, 60131
60131 Università
Ancona,
Ancona, Politecnica
Italy,
Italy, mail: delle
ee mail:
Dip. di Ingegneria dell’Informazione, Università Politecnica delle
Marche, via Brecce Bianche,
{gianluca.ippoliti,
{gianluca.ippoliti, 60131 Ancona, Italy, e mail:
giuseppe.orlando}@univpm.it
giuseppe.orlando}@univpm.it
Marche, via Brecce Bianche, 60131 Ancona, Italy, e mail:
{gianluca.ippoliti, giuseppe.orlando}@univpm.it
{gianluca.ippoliti, giuseppe.orlando}@univpm.it
Abstract: The
Abstract: The paper
paper focuses
focuses on on variable-rotor-speed/variable-blade-pitch
variable-rotor-speed/variable-blade-pitch wind wind turbines
turbines oper-oper-
Abstract:
ating
ating in theThe
in the paper
region
region of focuses
of high
high wind
windon variable-rotor-speed/variable-blade-pitch
speeds,
speeds, wherewhere blade
blade pitch
pitch and generatorwind
and generator torque
torqueturbines oper-
controllers
controllers
Abstract: The paper focuses on variable-rotor-speed/variable-blade-pitch wind turbines oper-
ating
are
are aimed in theat
aimed region
limiting of high
the wind speeds,
the turbine’s energy where bladeto
capture pitch
the and
the ratedgenerator
rated power
power value. torquePreliminary
controllers
ating in theatregionlimiting of high turbine’s
wind speeds, energy wherecapture
bladetopitch and generator value.
torquePreliminary
controllers
are
results
resultsaimedare
are at limiting
presented
presented the a
about
about aturbine’s
coupled
coupled energy technique
design
design capture to
technique of
of theobserver
an
an rated power
observer based
based value.
blade
blade Preliminary
pitch
pitch control
control
are aimed at limiting the turbine’s energy capture to the rated power value. Preliminary
results
input and are presented
and a generator
generator about a coupled
torque controller, design technique
neither of an requiring
of which
which observer based blade pitchof control
input
results a
are presented abouttorque controller,
a coupled design neither of
technique of an requiring
observer based the availability
the availability
blade pitchofcontrol wind
wind
input
speed and
speed a generatorClosed
measurements.
measurements. torqueloop controller,
convergenceneither of of
of thewhich
the overallrequiring
control the
control availability
system can
can be be of wind
proved.
input and a generatorClosed torqueloop convergence
controller, neither of overall
which requiring system
the availability proved.
of wind
speed
Preliminary
Preliminarymeasurements.
results
results about Closed
about the loop
the validation
validationconvergence
of
of the of the solution
the control
control overall control
using
using the system
55 −
−M Mcan
W be proved.
W three-blade
speed measurements. Closed loop convergence of the solution
overall control the
system can three-blade
be proved.
Preliminary
wind turbine
wind results
turbineresults
using the
using about
the the
National
National validation
Renewable
Renewable of the
Energycontrol solution
Laboratory using
(NREL) the 5
wind − M W
turbine three-blade
simulator
Preliminary about the validation of Energy
the controlLaboratory
solution(NREL)
using thewind 5 − turbine simulator
M W three-blade
wind
FASTturbine(Fatigue, using the National Structures,
Aerodynamics, Renewable Energy Laboratory (NREL) wind turbine simulator
FAST
wind (Fatigue,
turbine using the National Structures,
Aerodynamics, Renewable Energy and Turbulence)
and Turbulence)
Laboratory code.
code.
(NREL) wind turbine simulator
FAST (Fatigue, Aerodynamics, Structures, and Turbulence) code.
FAST
© 2017,(Fatigue, Aerodynamics,
IFAC (International Structures,
Federation of Automaticand Control)
Turbulence)Hosting code.
by Elsevier Ltd. All rights reserved.
Keywords: Wind
Keywords: Wind Energy
Energy Conversion
Conversion Systems,Systems, Pitch Pitch Control,
Control, High High Wind
Wind Speeds,
Speeds, Nonlinear
Nonlinear
Keywords:
Observers, Wind Generator Energy TorqueConversion
Control,Systems,NREL-FAST Pitch code.
Control, High Wind Speeds, Nonlinear
Observers,
Keywords: Generator
Wind Energy Torque Control,
Conversion NREL-FAST
Systems, Pitch code.
Control, High Wind Speeds, Nonlinear
Observers, Generator Torque Control, NREL-FAST code.
Observers, Generator Torque Control, NREL-FAST code.
1. INTRODUCTION
1. INTRODUCTION The most
The most common
common blade blade pitch
pitch control
control strategy
strategy is is aa feed-
feed-
1. INTRODUCTION The
back most
policycommon
based on blade
on the pitch
error control
betweenstrategythe ratedis apower
rated feed-
1. INTRODUCTION back
The policy
most based
common bladethe error
pitch between
control the
strategy is apower
feed-
back
and
and the policy
the actual
actual based
output
output on power,
the error butbetween
but the
the strongly the nonlinear
rated power re-
The back policy based on power,
the error betweenstronglythe nonlinear
rated power re-
The extraction
extraction of of wind
wind powerpower by by a a Wind
Wind Energy
Energy Conver-Conver- and the
lationship
lationship
and the
actual output
between
between
actual output pitch
pitchpower,
angle,
angle,
power,
but
wind
wind
but
the
the
strongly
speed
speed
strongly and
and nonlinear
rotor
rotor
nonlinear
re-
speed
speed
re-
The
sion extraction
sion System
System (WECS)
(WECS)of wind can
can power
be by a Wind
be divided
divided in Energyoperating
in different
different Conver- lationship between pitch angle, wind speed and rotor speed
The extraction of wind power by a Wind Energyoperating [Peng, 2010]
Conver- lationship
[Peng, 2010] suggests
suggests
between thatangle,
that
pitch aa coupled
coupled control
wind control
speed and design
design ofspeed
of
rotor pitch
pitch
sion System
regions (WECS)
associated withcanwindbe divided
wind speed, inmaximum
different operating
maximum allowable
sion System (WECS) can be divided in different operating [Peng,
regions associated with speed, allowable actuator
actuator
[Peng,
2010]
2010]
suggests
input
input and
suggests
that a coupled
and electrical
electrical
that a torque
torque
coupled
control
of the
of
control
design
the WT of
WT
design of
pitch
should
should
pitch
regions
rotor speedassociated
speed and rated with power.
rated wind speed, maximum allowable actuator input and electrical torque of the WT should
rotor
regions and
associated with power.
wind speed,In general,
In general,
maximum variable-rotor-
variable-rotor-
allowable actuator be
be pursued
pursued to
to achieve
input achieve effective results.
results.ofNonlinear
effective torque
and electrical the WToptimal
Nonlinear optimal
should
rotor speed and
speed/variable-blade-pitch
speed/variable-blade-pitchrated power. Wind
Wind In general,
Turbines variable-rotor-
(WT)
(WT) have
Turbinesvariable-rotor- have be pursued to achieve effective results. Nonlinear optimal
rotor speed and rated power. In general, control
control
be pursued approaches
approaches
to achieve have
have been
been
effective proposed
proposed
results. [Saravanakumar
[Saravanakumar
Nonlinear optimal
speed/variable-blade-pitch
three
three main
main regions
regions of of operation
operation Wind withTurbines
with respect
respect to to(WT)
wind
wind speed have control approaches have been proposed [Saravanakumar
speed
speed/variable-blade-pitch Wind Turbines (WT) have and and Jena,
Jena, 2015], also
2015], alsohaveadopting
adopting sliding mode
sliding mode control tech-
control tech-
three
[Johnsonmainet
et regions
al., of operation
2006]: i) a with respect
turbine that is tojust
wind
just speed control
starting approaches been proposed [Saravanakumar
three main regions of operation with respect to wind speed and
[Johnson al., 2006]: i) a turbine that is starting Jena,
niques.
niques.
and Jena, PI 2015],
PI also adopting
controllers
controllers
2015], also have been
have
adopting been sliding modefor
also used
also
sliding used
mode
control
for tech-
regulating
regulating
control tech-
[Johnson
up is et al., to
is considered
considered 2006]: i) a turbine that is1; just starting
up
[Johnson et al., to be operating
be
2006]: operating
i) a turbine in region
in region
that is1; justii) Region
ii) Region
starting 22 niques.
the
the pitch
pitch PI angle
controllers
PI angle
have been also
citepJonkman:09,
citepJonkman:09, used for
[Semrau
[Semrau et regulating
et al.,
al., 2015].
2015].
up
is is
an considered
operational to be
mode operating
with the in region
objective 1;
of ii) Region
maximizing 2 niques. controllers have been also used for regulating
is an
up operationaltomode
is considered with the in
be operating objective
region 1; of ii) Region 2 the
maximizing Though
Though
the
pitch
pitch
angle
effective
effective
angle
citepJonkman:09,
and
and easy
easy
citepJonkman:09, to
to [Semrau
implement,
implement,
[Semrau
et al.,
PI-based
PI-based
et al.,
2015].
con-
con-
2015].
is an operational
wind energy
energy capture; modeiii) within the objective
in region 3 (highof
3 (high maximizing
wind
is an operational capture;
modeiii) with region
the objective ofwind
wind speeds)
speeds) Though
maximizing trollers
trollers
Though
effective
rely
rely on
on
effective
and easy
loosely
loosely
and defined
defined
easy
to theoretical
to
implement, rules,
theoretical
implement,
PI-based
rules, require
require
PI-based
con-a
con- a
wind energy
the turbine
turbine capture;
must iii)
limitiii) in region
theincaptured
captured 3 (high
wind wind
power speeds)
so that
that trollers rely on and loosely defined
the
wind energymust limit
capture; the region 3 wind (highpower so
wind speeds) careful setting
careful
trollers setting
rely on and often
often
loosely need theoretical
need
defined corrective
theoretical
rules, require
corrective actions,
actions,
rules, usuallya
usually
require a
the
safe turbine
safe electrical
electrical must andlimit
and the captured
mechanical
mechanical loadswind
loads are not
are power
not so that careful setting and often need corrective actions, usually
exceeded.
exceeded.
the turbine must limit the captured wind power so that carefulbased
based on heuristics
onsetting and as
heuristics as
oftenwell. Also,
well.need anti-windup
Also,corrective
anti-windup mechanisms
mechanisms
actions, usually
safe electrical
Generator
Generator torque
torque and mechanical
control,
control, keeping loads
keepingloads the
the bladeare
blade not
pitch exceeded.
constant
constant based
pitchexceeded. on heuristics as well. Also, anti-windup mechanisms
safe electrical and mechanical are not have to be added
have toonbeheuristics
based to achieve
added toasachievewell. satisfactory
satisfactory
Also, anti-windup performances.
performances.
mechanisms
Generator
at
at an
an optimal torque control,
value for
for peak keeping
energy the blade pitch constant
optimal
Generator torque value
control, peak energy
keeping theextraction,
blade pitchis
extraction, usually
usually have to
isconstant be added to achieve satisfactory performances.
at an optimal
adopted
adopted in
in value2
region
region 2 for
[Pao
[Paopeak and
and energy
Johnson,
Johnson,extraction,
2011,
2011, is usually have
Corradini
Corradini The
The to be added
previous
previous to achieve
discussion
discussion do satisfactory
do indeed
indeed suggest
suggest performances.
that
that the
the avail-
avail-
at an optimal value for peak energy extraction, is usually The previous discussion do indeed suggest that the avail-
adopted
et al.,
et in
al., 2013],
2013], region 2
while control
while [Pao
control and
ofand
blade Johnson,
pitch is 2011,
is typically
typically Corradini
used toto The ability
ability of
previous reliable
of reliable wind
wind do
discussion speed
speed
indeed measurements,
measurements,
suggest that or or
theits
its es-
es-
avail-
adopted in region 2 [Paoof blade pitch
Johnson, 2011, Corradiniused ability ofis reliable wind speed measurements, or itscon-
es-
et al.,
limit 2013],
power while
for control
turbines of blade
operating pitch
in is
region typically
3 [Peng, used to
2010]. timates,
timates, is a
a key
key factor
factor for
for improving
improving closed-loop
closed-loop con-
limit
et al.,power
2013],for whileturbines
control operating in region
of blade pitch 3 [Peng,used
is typically to ability
2010]. timates,
of reliable wind speed measurements, or its es-
is a key factor
limit
In power forpower
In particular,
particular,
limit power forpower
turbines
turbines
operating
regulation
regulation
operating in in region
in region
region
in region 3 3 [Peng,
3 using
using
3 [Peng,only2010].
only pitch trol
trol performances,
pitch timates,
2010]. performances,
is a key factor since for
since windimproving
wind
for speed
speed is
improving is closed-loop
both
both the
closed-loop
con-
the input
input
con-
In particular,
angle control power
exhibits regulation
some in region
limitations, 3 using
which only
are pitch
due to trol
for
for performances,
the
the control
control system
systemsince and
andwindthe
the speed is
parameterizing
parameterizingboth the input
variable
variable
angle control exhibits some limitations, which are due to trol performances,
In particular, power regulation in region 3 using only pitch for the control system and the parameterizing variable since wind speed is both the input
angle control
constraints
constraints onexhibits
on the amplitude
the some limitations,
amplitude and speed
and speed of which
of response
response are due to for
of the
of the for the
the dynamics
dynamics
control system that determine
that determine the operating
the
and the parameterizingoperating pointpoint
variableof
of
angle control exhibits some limitations, which are due to for the dynamics that determine the operating point of
constraints
pitch servos on the
[Bianchi
pitch servoson[Bianchi amplitude
et al., and
2004].
et al., 2004]. speed of
Nowadays
Nowadays responsethere of
is the
an
an for
there ofis the the
the WECS.
WECS. The
The measured
measured wind
wind speed
speed
the dynamics that determine the operating point of on
on the
the nacelle
nacelle is
is
constraints the amplitude and speed of response the WECS. The measured wind speed on the nacelle of is
pitch servos
increasing
increasing [Bianchi
interest
interest to et
reduce
to reduce al., 2004].
the Nowadays
effects
the effects of pitch there
of pitchthere is
actuators
actuators an unfortunately
unfortunately
the WECS. imprecise
imprecise
The measured and
and not
not
wind a
a good
good
speed representative
representative
on the nacelle of
is
pitch servos [Bianchi et al., 2004]. Nowadays is an
increasing
limitationsinterest
limitations
increasing by combining
by to reduce
combining
interest to reduce pitch
pitch the
the
effects
angle
angle andof
and
effects
pitch actuators
generator
ofgenerator
pitch actuatorstorque unfortunately
torque the
the rotor
rotor effective
unfortunately
imprecise
effective wind
wind speed
imprecise
and not
speed
and not
a good
[Soltani
[Soltani
a goodet representative
et al.,
al., 2013].
2013]. ToTo solve
representative
of
solve
of
limitations
control to by
shed combining
excess power pitchand angle
limit and
the generator
turbine’s torque
energy the
this
this rotor effective
problem,
problem, a
a wind speed
number
number of
of [Soltani et al.,
algorithms
algorithms 2013].dedicated
present
present To solve
dedicated
control to shed
limitations excess power
by combining pitchandangle
limitand thegenerator
turbine’s energy torque the this
rotor effective wind speed [Soltani et al., 2013]. To solve
problem, a number of algorithms
control
captureto
capture toshed
to the excesspower
the rated
rated power valueand limitin the turbine’s
in region 3 [Tangenergy
3 [Tang et
et al.,
al., thisestimators
estimators
problem, of the
of the wind speed
wind
a number speed affectingpresent
affecting
of algorithms the entire
the
present
dedicated
entire rotor,
rotor,
dedicated
control to shed excesspower
power valueand limit region
the turbine’s energy estimators of the wind speed affecting the entire rotor,
capture
2011].
2011]. to
The
The the rated
presented
presented power
control
control value
strategy
strategy in region 3
specifically
specifically [Tang et
addresses
addresses al., and
and a
a
estimatorsgood
good comprehensive
comprehensive
of the wind analysis
analysis
speed of
of
affecting these
these
the techniques
techniques
entire in
in
rotor,
capture to the rated power value in region 3 [Tang et al., and a good comprehensive analysis of these techniques in
2011].
the
the case
caseTheof
of presented
full
full load
load control
operation
operation strategy
in
in the
the specifically
so-called
so-called addresses
Regime
Regime 3,
3, given
given
and a in
in [Soltani
[Soltani
good et
et al.,
al.,
comprehensive 2013].
2013].
analysis of these techniques in
2011]. The presented control strategy specifically addresses given in [Soltani et al., 2013].
the case ofit
therefore full
shouldloadbe operation
intendedin
intended asthe so-calledof Regime 3, given in [Soltani et al., 2013].
therefore
the case ofit full
shouldloadbeoperation inas a
a solution
the solution
so-calledofRegime this
this part
part3,
therefore
of the WECS
of the WECS it should
control
control be intended
problem. as a solution of this part
therefore it should beproblem.
intended as a solution of this part
of the WECS control problem.
of the WECS control problem.
Copyright
2405-8963 ©
Copyright © 2017
2017 IFAC
© 2017, 10328
IFAC (International Federation of Automatic Control)
IFAC 10328Hosting by Elsevier Ltd. All rights reserved.
Copyright
Peer review©under
2017 responsibility
IFAC of International Federation of 10328
Automatic Control.
Copyright © 2017 IFAC 10328
10.1016/j.ifacol.2017.08.1631
 Proceedings of the 20th IFAC World Congress
Toulouse, France, July 9-14, 2017 M.L. Corradini et al. / IFAC PapersOnLine 50-1 (2017) 9914–9919 9915

In this paper, some preliminary results will be presented the tables in [NWTC, b] for the NREL 5 − M W reference
about a coupled blade pitch/generator torque controller wind turbine.
(in region 3) avoiding the need of wind speed measure- Analytical Approximation
ments. An observer-based blade pitch actuator input will 0.35
be designed such that the minimum of the error between β = 9°
rated and actual power is achieved, without any feedback 0.3 β = 12°
measurements of wind speed. Closed loop convergence β = 15°
0.25
of the overall control system is proved. The theoretical β = 18
°
development is supported by simulations using the three- 0.2 °
β = 21
blade NREL 5 − M W wind turbine using the FAST code

Cp
β = 23°
simulator [NWTC, a]. 0.15

2. WECS DYNAMICS 0.1

0.05
The system model here reported is inspired by the studies
[Zaragoza et al., 2011, Bianchi et al., 2007] and references 0
0 1 2 3 4 5 6 7 8 9 10
therein. As well known, wind energy is transformed first λ
into mechanical energy through the WT blades and, ulti- (a)
mately, into electrical energy through the generator. The
aerodynamic (mechanical) power that the wind turbine ex- Tabulated Values - FAST Simulations
0.35
tracts from the wind is expressed by the following equation β = 9°
[Zaragoza et al., 2011, Bianchi et al., 2007]: 0.3 β = 12
°

1 β = 15°
Pa = ρπr2 Cp (λ, β) V (t)3 (1) 0.25
2 β = 18°
where ρ is the air density, r is the wind turbine rotor Cp 0.2 β = 21°
radius, V is the wind speed and the power coefficient β = 23
°
Cp (λ, β) represents the turbine efficiency to convert the 0.15
kinetic energy of the wind into mechanical energy [Bianchi
0.1
et al., 2007]. This coefficient is a function of both the blade
pitch angle β and the tip speed ratio λ which is defined 0.05
ω
as [Qiao et al., 2009] λ = r, where ω is the WT angular
V 0
shaft speed. The introduction of the expression of λ in Eq. 0 1 2 3 4 5 6 7 8 9 10
λ
(1) gives:
(b)
Ka rωCp (λ, β)
Pa = V (t)2 (2)
λ Fig. 1. (a) Analytical approximation with function (4) ;
def
with Ka = ρ π2 r2 .
As a consequence, the torque that the (b) Tabulated values obtained from FAST simulations
wind turbine extracts from the wind is given by: [NWTC, b]. β is in degrees.
Ka rCp (λ, β)
Ta (t) = V (t)2 . (3) The mechanical equation governing the turbine can be
λ simplified as follows [J.M. Jonkman and M.L. Buhl , 2005]:
The power coefficient Cp (λ, β) is a nonlinear function J ω̇(t) = −Kω(t) + Ta (t) − Ng T̃e (t) (6)
[Monroy and Alvarez-Icaza, 2006, Siegfried, 1998], and where ω(t) is the rotor angular speed, K is the coefficients
depends on blade aerodynamic design and WT operating of viscous friction of the low-speed shaft, Ng is the gearbox
conditions. In [Zaragoza et al., 2011], the following equa-
ratio, T̃e (t) is the electrical torque of the generator, which
tion is proposed to approximate the power coefficient:
  can be imposed designing currents and voltages of the
Cp (λ, β) = c1 k1 γ + k2 β + k̄3 exp(k4 γ) (4) generator stage. For convenience the following definition
  def
1 0.035 is introduced, to be used hereafter Te (t) = Ng T̃e (t).
γ= − 3 (5)
λ + 0.08β β +1
The coefficients c1 , k1 , k2 , k̄3 , k4 depend on the shape of 3. VARIABLE ROTOR SPEED AND VARIABLE
the blade and its aerodynamic performance [Beltran et al., BLADE PITCH WT REGIME
2009, Zaragoza et al., 2011]. The following coefficient
values have been considered in this work: c1 = 6.909; In the paper, operation in the region of high wind speed
k1 = 7.022; k2 = −0.04176; k̄3 = −0.3863; k4 = −14.52. will be considered, and the objective of the control system
These coefficients have been obtained fitting the Eq. (4) will be defined as that of maintaining the captured wind
to the Cp tables for the NREL 5 − M W wind turbine power at the rated value P̄a .
generated using the NREL code.
3.1 Problem statement
Following [Georg et al., 2012], the behavior of the func-
tion (4) with the proposed coefficients has been reported As pointed out, the control objective in high wind speeds
in Fig. 1(a) for different pitch angles, compared to the is to maintain the captured wind power at the rated value
corresponding plots (see Fig. 1(b)) of the curves based on P̄a , i.e. to minimize the following squared error:

10329
Proceedings of the 20th IFAC World Congress
9916
Toulouse, France, July 9-14, 2017 M.L. Corradini et al. / IFAC PapersOnLine 50-1 (2017) 9914–9919

 2 4. THE OBSERVER-BASED BLADE PITCH AND

e(β, λ) = Pa (β, λ) − P̄a (7)
GENERATOR TORQUE CONTROLLERS
by regulating the blade pitch actuator input and the
generator torque Te . In order to pursue such objective,
an “optimal value” β ∗ of β(t) will be sought satisfying Based on the results reported so far, control laws should be
designed ensuring the tracking of both the optimal value
∂e(β, λ) β ∗ (V ) (i.e. that minimizing the squared error (7)) by the
=0 (8)
∂β blade pitch angle β, and the reference rated aerodynamic
in correspondence of the rated value ω ∗ of the rotor power, for achieving power regulation. It should be noted
speed ω(t) corresponding to the desired performance of that the reference variables β ∗ (V ) depends on wind speed
the generator. A control law guaranteeing the tracking of V , whose measurements are assumed not available. To
such optimal value β ∗ is aimed to be designed. overcome this problem, an auxiliary variable V̂ (t) will be
here introduced, with dynamics to be designed in order to
solve the control problem. Define the following variables:
3.2 Characterization of the optimal blade pitch angle β ∗ ω(t)
β̂ ∗ (t) = β ∗ (V̂ (t)); λ̂(t) = r ; Ĉp (t) = Cp (λ̂(t), β ∗ (t))
In view of the complex expression of the power coefficient, V̂ (t)
computing analytically the derivative (8) of the error 1 0.035
γ̂(t) = −
variable with respect to pitch angle and solving it with ∗ ∗
λ̂(t) + 0.08β̂ (t) β̂ (t)3 + 1
respect to β is a difficult task, particularly due to the P̂a (t) = Pa (β̂ ∗ (t), V̂ (t)) = Ka Ĉp (t)V̂ (t)3 (11)
coupled dependence of the solution also on the wind speed
V . The approach taken here will consists of searching where V̂ (t) is a variable to be sought such that the
numerically the expression of β ∗ = β ∗ (V ) such that estimated aerodynamic power equals the real one, i.e. such
∂e(β, λ) that a sliding mode is enforced on the following sliding
= 0 in correspondence of the rated rotational
∂β surface:
speed ω in order to achieve an expression of β ∗ (V )
∗
s(t) = Pa (t) − P̂a (t) = 0 (12)
depending only on the independent variable V . More in
detail, consider the derivative of the error with respect to The achievement of a sliding motion on such a sliding
the blade pitch angle β: surface guarantees that the estimated aerodynamic power
∂e(β, λ) ∂Pa (β, λ) ∂γ is equal to the actual one, and ultimately, to the rated aero-
= · dynamic power, provided that the tracking of the reference
∂β ∂γ ∂β
  signal β̂ ∗ (t) is achieved. It should be also recalled that the
= Ka c1 exp(γk4 ) k1 + k4 (k1 γ + k2 β + k̄3 ) V 3 requirement about the tracking of the rated aerodynamical
 
−0.08 0.1β 2 power P̄a imposes that the electrical torque Te should
· + =0 (9) be straightforwardly designed as follows [Jonkman et al.,
(λ + 0.08β)2 (1 + β 3 )2
2009]:
Inserting the rated value ω ∗ and solving numerically the
previous equation, it can be verified that the following P̄a
Te (t) = (13)
expression for β ∗ (V ) provides a good estimate of the ω(t)
optimal value minimizing the squared error (7):
Finally it should be recalled that, according to [Stotsky
def
β ∗ (V ) = aβ V + bβ = 1.2998V − 8.2824 [deg] (10) et al., 2013], the blade pitch actuator can be modeled as a
first order lag:
β ∗ (V ) depending only on the wind speed V expressed in
[m/s] (see Fig.2). 1
β̇(t) = − β(t) + u(t) (14)
τ
22
Approximation of BetaStar vs. V
where u(t) is the control input, i.e. the torque supplied by
the blade pitch actuator.
20
Remark 1. Inherent physical limitations of real WECSs
18 associated to the specific operation region addressed imply
that bounded ranges of values exist for wind speed, rotor
16
speed and blade pitch angle. Inferior and superior limits
Pitch angle [deg]

14
for the range of admissible values for a variable x will be
denoted hereafter by xm and xM respectively.
12
Assumption 1. It is assumed that for the WECS under
10
operation it holds J ω̈(t)  0 and ω(t)  1. Such approxi-
mations do usually hold for large size WTs.
8
Theorem 4.1. With reference to the WECS model (6)
6
13 14 15 16 17 18 19 20 21 22 23
under Assumption 1, consider an auxiliary variable V̂ (t)
Wind speed V [m/s]
governed by the following dynamics:
 
˙  
Fig. 2. Linear best fit approximation of the optimal speed V̂ (t) = ρ1 G1 (β, V̂ (t))−1 G2 (β, V̂ (t)) sign [s(t)] (15)
β ∗ (V ) according to (10).
where ρ1 > 1, and consider the following definitions

10330
 Proceedings of the 20th IFAC World Congress
Toulouse, France, July 9-14, 2017 M.L. Corradini et al. / IFAC PapersOnLine 50-1 (2017) 9914–9919 9917

def
G1 (β, V̂ (t)) = Ka c1 F1 (β, V̂ (t))V̂ (t)2 exp(k4 γ̂(t)) (16) Aerodynamic Power
8
def
G2 (β, V̂ (t)) = J ω̇(t)2 + 2Kω(t)ω̇(t) + Ka c1 F2 (β, V̂ (t))·
7
· V̂ (t)2 exp(k4 γ̂(t)) (17)
def H(β, V̂ (t))rω(t) 6

P a (t) [M W ]
F1 (β, V̂ (t)) = − +3
(rω(t) + 0.08β V̂ (t))2
0.105β(t)2 5
+ aβ H(β, V̂ (t)) + aβ k2 V̂ (t) (18)
(β(t)3 + 1)2
4
def   
H(β, V̂ (t)) = k1 + k4 k1 γ̂(t) + k2 β + k̄3 V̂ (t) (19)
˙ 3
˙ rω̇(t)V̂ (t) − rω(t)V̂ (t) + 0.08β V̂ (t)2
γ̂(t) =−  2
rω(t) + 0.08β V̂ (t) 2
0 20 40 60 80 100
t [s]
˙
0.105β̂ ∗ (t)β(t)2 (a)
+ (20)
(β(t)3 + 1)2 Rotor Sp eed ω r
The following observer-based law for the blade pitch con- 14
troller:
1  
˙
u(t) = β(t) + aβ V̂ (t) − ρ2 sign β(t) − β ∗ (V̂ ) ; ρ2 > 0 13.5
τ
(21)

ω [r pm]
13
ensures the closed loop asymptotic tracking of both the
optimal pitch angle reference (10) and the rated aerody-
namic power P̄a , without any feedback measurement of 12.5
wind speed.
12
The proof is omitted for the sake of brevity.
11.5
5. SIMULATION TESTS 0 20 40 60 80 100
t [s]

The proposed controller has been tested by intensive (b)

simulations using the NREL FAST code. The FAST code
is a comprehensive aeroelastic simulator of two- and three-
bladed horizontal-axis WT [NWTC, a, J.M. Jonkman and
M.L. Buhl , 2005]. It is well assessed that the structural
model of FAST is of high fidelity and it is suitable for
wind turbine design and certification [A. Manjock, 2005,
M.L. Buhl and A. Manjock, 2006]. The main parameters of
the considered WECS have been derived from [Jonkman
et al., 2009] for the NREL 5 MW wind turbine. Validation
tests have been performed using the FAST wind data
shown in Fig. 5(b) [NWTC, a], with turbulence 10% and
mean value 20 m/s. Initial conditions have been set as
ωr (0) = 12 rpm, β(0) = 22 deg.
Some of the performed tests have been reported in the
following pictures. In particular, Fig. 3(a) and 3(b) show
the aerodynamic power and the rotor angular speed,
respectively. It can be noted that, as expected, the power
is maintained close to the rated value P̄a = 5.29661M W .
Fig.4(a) shows the blade pitch angular position. The
performance achieved in terms of efficiency of transferred
energy is testified by results reported in Fig.4(b), showing
that the tip speed ratio is far from its optimal value
because of the high wind speeds. Finally, Fig.5(a) shows
the behaviour of the auxiliary variable V̂ (t).
For the sake of comparison, the presented simulation re-
sults have been compared with the FAST baseline con-
troller [J.M. Jonkman and M.L. Buhl , 2005], in order to
evaluate performances of the proposed control algorithm.
Fig. 3. (a) Aerodynamic Power; (b) Rotor angular speed.
Fatigue histograms produced by the FAST baseline con-
troller have been reported in Figs- 7(a)-7(b). The aerody-
namical power obtained using the FAST baseline controller
is also reported in Fig. 7(c). Using the IAE criterion, i.e.
the integral of the absolute value of the power tracking
error, can used to numerically evaluate the above simula-
tion results (see Tab. 1). The integral has been computed
starting from t = 10 s, to avoid the initial transient phase
associated to the observer warming-up. The previous table
Table 1. Performance comparison: IAE [M W ].
FAST baseline controller Presented approach
12.22 11.57
shows that a performance improvement of nearly 6% is
obtained with the approach proposed in this paper with
respect to the standard baseline controller provided by
FAST, though no feedback measurements of wind speed
are used. In addition, it worths noticing that the proposed
controller benefits also from the availability of a rigorous
methodological proof ensuring closed loop stability.
6. CONCLUSIONS
In this paper, some preliminary results have been pre-
sented about the power regulation above rated wind speed

10331
Proceedings of the 20th IFAC World Congress
9918
Toulouse, France, July 9-14, 2017 M.L. Corradini et al. / IFAC PapersOnLine 50-1 (2017) 9914–9919

Pitch Angle β Auxiliary variable V̂ (t)

22 20

20 19

18
18

V̂ (t)
[deg ]

17
16
16

14
15

12 14
0 20 40 60 80 100 0 20 40 60 80 100
t [s] t [s]
(a) (a)

Tip Sp eed Ratio Wind Sp eed

6 24

5.5 22

V (t) [m/s]
5 20

4.5 18

4 16

3.5 14
0 20 40 60 80 100 0 20 40 60 80 100
t [s] t [s]

(b) (b)

Fig. 4. (a) Blade pitch angular position; (b) Tip Speed
Ratio.
in coupled blade pitch/generator torque controlled WECS.
The need of wind speed measurements is avoided by an
observer based controller tracking the optimal blade pitch
angle which minimizes the squared error of the aerody-
namical power with respect to the rated value. Closed loop
convergence of the overall control system is proven and an
extensive simulation study has been reported to support
the theoretical development.
REFERENCES
A. Manjock. Design Codes FAST and ADAMS for Load
Calculations of Onshore Wind Turbines. Technical Re-
port 72042, Germanischer Lloyd WindEnergie GmbH,
2005. URL http://wind.nrel.gov/designcodes/
papers/GL_Report.pdf.
B. Beltran, T. Ahmed-Ali, and M. Benbouzid. High-order
sliding-mode control of variable-speed wind turbines.
IEEE Trans. Ind. Electron., 56(9):3314 –3321, 2009.
F. D. Bianchi, H. N. De Battista, and R. J. Mantz.
Wind Turbine Control Systems: Principles, Modelling
and Gain Scheduling Design. Springer-Verlag, Berlin,
2007.
F.D. Bianchi, R.J. Mantz, and C.F. Christiansen. Power
regulation in pitch-controlled variable-speed WECS
above rated wind speed. Renewable Energy, 29(11):1911
– 1922, 2004.
Fig. 5. (a) Auxiliary Variable V̂ (t); (b) Wind Inflow.
M.L. Corradini, G. Ippoliti, and G. Orlando. Fully sensor-
less robust control of variable-speed wind turbines for
efficiency maximization. Automatica, 49(10):3023–3031,
2013.
S. Georg, H. Schulte, and H. Aschemann. Control-oriented
modelling of wind turbines using a takagi-sugeno model
structure. In 2012 IEEE International Conference on
Fuzzy Systems (FUZZ-IEEE), pages 1–8, June 2012.
J.M. Jonkman and M.L. Buhl . FAST User’s Guide. Tech-
nical Report NREL/EL-500-38230, National Renewable
Energy Laboratory, 2005. URL http://wind.nrel.
gov/public/bjonkman/TestPage/FAST.pdf.
K.E. Johnson, L.Y. Pao, M.J. Balas, and L.J. Fingersh.
Control of variable-speed wind turbines: standard and
adaptive techniques for maximizing energy capture.
IEEE Contr. Sys. Mag., 26(3):70 – 81, 2006.
J. Jonkman, S. Butterfield, W. Musial, and G. Scott. Def-
inition of a 5-MW Reference Wind Turbine for Offshore
System Development. Technical Report NREL/TP-500-
38060, National Renewable Energy Laboratory, 1617
Cole Boulevard, Golden, Colorado, February 2009. URL
http://www.nrel.gov/docs/fy09osti/38060.pdf.
M.L. Buhl and A. Manjock. A Comparison of Wind
Turbine Aeroelastic Codes Used for Certification. Tech-
nical Report NREL/CP-500-39113, National Renewable
Energy Laboratory, 1617 Cole Boulevard, Golden, Col-
orado, January 2006. URL http://www.nrel.gov/

10332
 Proceedings of the 20th IFAC World Congress
Toulouse, France, July 9-14, 2017 M.L. Corradini et al. / IFAC PapersOnLine 50-1 (2017) 9914–9919 9919

Blade 1 in−plane moment (moment caused by in−plane forces, (kN·m) Normalized) at the blade root Blade 1 in−plane moment (moment caused by in−plane forces, (kN·m) Normalized) at the blade root
3.5 3.5

3 3
Nr of cycles(log10): 2031.5 (5.5 from half−cycles)

Nr of cycles(log10): 1355.5 (4.5 from half−cycles)

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.5

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Histogram of "rainflow" amplitudes Histogram of "rainflow" amplitudes

(a) (a)
Tower base roll (or side−to−side) moment (moment caused by side−to−side forces, (kN·m) Normalized) Blade 1 in−plane moment (moment caused by in−plane forces, (kN·m) Normalized) at the blade root
3.5 3

3
2.5

Nr of cycles(log10): 613.5 (4.5 from half−cycles)

Nr of cycles(log10): 1410 (4 from half−cycles)

2.5
2

1.5

1.5

1
1

0.5
0.5

0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Histogram of "rainflow" amplitudes Histogram of "rainflow" amplitudes

(b) (b)
6 FAST Baseline: Aerodynamic Power
Fig. 6. Fatigue Load Histogram: (a) Tower base roll x 10
7
moment; (b) Blade 1 in-plane moment at the root.
6
docs/fy06osti/39113.pdf.
A. Monroy and L. Alvarez-Icaza. Real-time identification 5
of wind turbine rotor power coefficient. In 45th IEEE
P a (t) [W ]

Conf. Dec. Contr., pages 3690 –3695, 2006. 4

NWTC. NWTC Information Portal (FAST). https://
3
nwtc.nrel.gov/FAST, a.
NWTC. NREL 5-MW reference turbine - CP, CQ, CT 2
Coefficients. https://wind.nrel.gov/forum/wind/
viewtopic.php?f=2&t=582, b. 1
L.Y. Pao and K.E. Johnson. Control of wind turbines.
IEEE Contr. Sys. Mag., 31(2):44 –62, 2011. 0
0 20 40 60 80 100
Guo Peng. Nonlinear feed-forward pitch controller for t [s]
wind turbine based on rotor aerodynamic characteris- (c)
tics. In Proc. 9th int. Conf. Machine Learning and
Cybernetics, pages 879–883, 2010.
Fig. 7. Baseline FAST controller: (a) Fatigue Load His-
Wei Qiao, Liyan Qu, and R.G. Harley. Control of IPM
togram of Tower base roll moment; (b) Fatigue Load
synchronous generator for maximum wind power gen-
Histogram of Blade 1 in-plane moment at the root;
eration considering magnetic saturation. IEEE Trans.
(c) Aerodynamical Power.
Ind. Appl., 45(3):1095 –1105, 2009.
R. Saravanakumar and D. Jena. Validation of an in- A. Stotsky, B. Egardt, and O. Carson. Control of wind
tegral sliding mode control for optimal control of a turbines: a tutorial on proactive perspectives. In Proc.
three bladed variable speed variable pitch wind turbine. 2013 Am. Control Conf., pages 3429–3436, 2013.
International Journal of Electrical Power and Energy Choon Yik Tang, Yi Guo, and J.N. Jiang. Nonlinear
Systems, 69(4):421–429, 2015. dual-mode control of variable-speed wind turbines with
G. Semrau, S. Rimkus, and T. Das. Nonlinear systems doubly fed induction generators. IEEE Transactions on
analysis and control of variable speed wind turbines Control Systems Technology, 19(4):744–756, July 2011.
for multiregime operation. J. Dyn.Sys., Measur. and J. Zaragoza, J. Pou, A. Arias, C. Spiteri, E. Robles, and
Contr., Trans. ASME, 137(4), 2015. S. Ceballos. Study and experimental verification of
Heier Siegfried. Grid integration of wind energy conversion control tuning strategies in a variable speed wind energy
systems. John Wiley & Sons Ltd, 1998. conversion system. Ren. Ener., 36(5):1421 – 1430, 2011.
M.N. Soltani, T. Knudsen, M. Svenstrup, R. Wisniewski,
P. Brath, R. Ortega, and K. Johnson. Estimation
of rotor effective wind speed: A comparison. IEEE
Transactions on Control Systems Technology, 21(4):
1155–1167, July 2013.

10333