Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Journal of Physics: Conference Series

PAPER • OPEN ACCESS You may also like


- Normal values of offline exhaled and nasal
Model-free control of the dynamic lift of a wind nitric oxide in healthy children and teens
using chemiluminescence
turbine blade section: experimental results A Menou, D Babeanu, H N Paruit et al.

- Improving Silicon-Based Composite


Electrodes by Chemical Grafting
To cite this article: Loïc Michel et al 2022 J. Phys.: Conf. Ser. 2265 032068 Claudia Ramirez-Castro, Cédric Martin,
Olivier Crosnier et al.

- (Keynote) To Honor Michel Armand:


Developments on New Electrode and
Electrolyte Materials for Batteries at the
View the article online for updates and enhancements. IMN
Dominique Guyomard, Florent Boucher,
Thierry Brousse et al.

This content was downloaded from IP address 87.89.223.13 on 03/06/2022 at 19:51


The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Model-free control of the dynamic lift of a wind


turbine blade section: experimental results
Loı̈c Michel1 , Ingrid Neunaber2 , Rishabh Mishra2 , Caroline Braud2 ,
Franck Plestan1 , Jean-Pierre Barbot3 , Xavier Boucher4 , Cédric
Join5,7 , Michel Fliess6,7
1
Centrale Nantes-LS2N, UMR 6004 CNRS, Nantes, France
2
LHEEA lab. - CNRS - Centrale Nantes, 1 rue de la Noë, 44100 Nantes
3
Centrale Nantes-LS2N, UMR 6004 CNRS, Nantes, and ENSEA-Quartz Laboratory EA 7393,
France
4
Université du Québec à Trois-Rivières-LSSI, Trois-Rivières, Québec, Canada
5
CRAN (CNRS, UMR 7039), Université de Lorraine BP 239, 54506 Vandœuvre-les-Nancy,
France
6
LIX (CNRS, UMR 7161), École polytechnique 91128 Palaiseau, France
7
AL.I.E.N., 7 rue Maurice Barrès, 54330 Vézelise, France
E-mail: loic.michel@ec-nantes.fr

Abstract. This work addresses the problem of developing control algorithms for the control of
the aerodynamic lift of wind turbine blades using air injection, taking into account disturbances
caused by turbulent perturbations. For this, a test bench is used where the lift of a 2D blade
section in a wind tunnel can be controlled by a set of micro-jets close to the trailing edge.
Through a continuous, local identification of the lift variations a model-free control that does
not need any prior knowledge of the system is proposed. It allows the control of the flow of
the micro-jets and stabilizes the lift around a tracking reference. The ability of the proposed
control algorithm to track the lift reference when subjected to external perturbations, i.e.,
gusts, is discussed. In particular, this work demonstrates that the lift can be set to particular
values using the proposed control strategy, and can be re-stabilized to pre-gust lift conditions.
Experimental results illustrate globally the feasibility of such a control.

1. Introduction
Wind turbines are exposed to strong inflow shears of different scales due to the atmospheric
conditions in which they operate [1], rotors misaligned with the inflow, and wakes of upstream
turbines. This is even more significant for offshore wind turbines whose large rotors face locally
sheared inflows along the rotor-sweep area and even along the blades. Consequences are strongly
varying loads and reduced power production. For more efficiency, the pitch control [2] therefore
needs to be complemented by local aerodynamic controllers.
Proving the interest of such local control strategies for wind energy applications is a
challenging task as most of available models/solvers (BEM, aeroelastic, CFD solvers, etc.) are
limited to accurately describe the control effect, and many challenges have to be tackled before
being able to put such strategies on realistic field tests (e.g. actuator/sensor developments
and implementations). However, two field tests have demonstrated the potential of such local
control to reduce the Levelized Cost Of Energy (LCOE) for wind energy applications: the work

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

of Gonzales et al [3] on a 4MW wind turbine rotor which is using an active flap device actuator
and the work of Matsuda el al [4] on a 1.7MW turbine which is using plasma actuators.
Other actuator types exist and were tested at lower scales using wind tunnel facilities and
non-rotating blade sections (e.g. slats [5] or micro-jets [6]). Depending on the actuator system
used, the authority of the control is more or less important and the control action is more or less
fast. For instance, Leroy et al [7] show that micro-jets have a superior authority compared to
plasma actuators; however, the repetition rate of the control action is slower. Also, for Active
Flow Control strategies, local sensors were developed to be used with actuators in closed-loop
control (e.g. e-penons [8]). For closed-loop implementations, actuator and sensor candidates
need to be combined with control algorithms sufficiently robust to be used in wind turbine
operating conditions such as atmospheric flows or gust events. However, few contributions have
been made so far in that direction.
At first, simple experimental closed-loop test benches were explored with the aim of controlling
flow separation of simple geometrical objects such as a hump [9], a backward facing step [10]
or a bump [11], etc in small wind tunnel facilities with simplified perturbations. More complex
geometries, such as airfoils put in complex inflows, need different controllers; they were first
developed for NACA profiles with objectives towards aeronautic applications without inflow
perturbations (see e.g. [12]). For wind energy applications, Bartholomay et al [13] has recently
developed a feed-forward controller to alleviate the lift fluctuations imposed by a sinusoidal
inflow using trailing edge flap controllers. Jaunet et al [6] proposed a proportional controller
to reduce lift fluctuations acting on a NACA64(4)421 profile using micro-jet actuators close to
the trailing edge. The reference signal was given by a small number of pressure taps around the
airfoil. The control was tested in a constant shear inflow mimic at the blade scale by blade pitch
oscillations. The oscillations of the lift have been significantly damped around a constant value;
however, some fluctuations remained. The current paper will extend the work of Jaunet et al [6]
towards controllers more suited for strong inflow perturbations found in situ at the blade scale,
simulated by the chopper system [14].
Taking into account the difficulty to model phenomena in aerodynamics, and, in particular,
to model the flow around the blade environment, non-model-based alternative methods have
been proposed e.g. Extremum-Seeking Control (ESC) associated with sliding mode control;
such an approach has been implemented in the framework of aerodynamic drag reduction to
optimize some physical parameters and the control parameters online [15]. The ESC method
has been also experimented as a real-time optimization strategy to cope with high-lift problems
in the context of flow instability at the blade section scale through active flow control [12]. This
strategy is encouraging for the closed-loop lift control but could be difficult to implement since
it is based on a modulation procedure including filter design.
Similarly, the model-free control law [16] can be considered as an alternative to the standard PI
and PID controls as it does not need any prior knowledge of the plant, and it is straightforward
to tune, contrary to the commonly used classic PID controllers whose tuning usually depends
on trial and error methods. Its usefulness in many situations, including severe non-linearities
and time-varying properties, has been demonstrated through successful applications since it is
computationally efficient, easily deployable even on small embedded devices, and can be imple-
mented in real time since it requires very light computations (see, e.g., [17]).

In this work, a model-free control technique [18] has been implemented to control the lift
using micro-jets installed at the trailing edge of a 2D blade section that is mounted in a wind
tunnel (see Figure 1). The test bench used to test the efficiency of the control to alleviate
aerodynamic load fluctuations induced by turbulence includes a perturbation system producing
a large mean flow variation with turbulence superimposed on it (described in Section 2). Only
global aerodynamic force measurements are used as feedback for the control.

2
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Figure 1: (a) Aerodynamic wind tunnel of LHEEA: A gust generator, the chopper, can optionally
be used. (b) Example of gust and energy spectrum of flow generated by the chopper. (c) Lift
curves of the airfoil in dependence of the Reynolds number. (d) setup used for active flow control
measurements: a tube with wholes is installed at the airfoil’s trailing edge and pressurized air
can be blown out of the wholes.

2. Experimental setup
The main purpose of this experiment is to highlight the feasibility to use advanced control
algorithms in a simplified flow configuration. Simplifications stands in the Reynolds number,
the blade shape and the 2D section (no rotation and no transverse flow). This means that
the flow transition and the flow separation location are certainly different. However, these flow
properties exist in the present blade and the controller is dealing with them, while the robustness
is evaluated through strong inflow perturbations, making this model-free approach realistic for
other flow configurations.
The experimental closed-loop bench is composed of a wind tunnel with its perturbations
system, a 2D aerodynamic blade profile, lift and drag sensors and micro-jet controllers that are
detailed in this section.

2.1. Wind tunnel facility and gust generator


The aerodynamic wind tunnel of LHEEA is a recirculating tunnel. The test section has a cross-
section of 500 × 500 mm2 and a length of 2300 mm, see Figure 1. The turbulence intensity of the
wind tunnel is below 0.3%. In the present study, a grid is installed at the inlet of the test section
to generate turbulent inflow with a turbulence intensity of 3%. This bypasses the laminar-to-
turbulent transition occurring at low Reynolds numbers and low angles of attack (AoA) for this
blade geometry, cf. the linear part of the lift curve in Figure 1(c).
The inlet of the test section is additionally equipped with a system which enables the
generation of a sudden variation of the mean flow with turbulence superimposed on it. This
system is called ”chopper” and consists of a rotating bar that cuts through the inlet of the test

3
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

(a) Control board (b) Wind tunnel setup

Figure 2: Experimental setup within the wind tunnel.

section (Figure 1(a)). An example of a gust produced by the chopper and the energy spectrum
is shown in Figure 1(b). The flow generated the chopper has been characterized in [14].

2.2. Aerodynamic profile


A NACA 654 − 421 profile with a chord length of c = 9.6 cm was installed in the test section.
It is a thick profile with two changes of the lift curve corresponding to a first boundary layer
separation at the trailing edge of the profile and a second flow separation at the leading edge,
causing stall (see Soulier et al [8] for more details on the blade aerodynamics). In the present
study, the AoA was set to α0 = 20◦ as preliminary open-loop tests have been found to produce
optimal control results at this AoA.

2.3. Lift and drag measurements


Two Z6FC3 HBM bending beam load cell sensors were used on each side of the blade support
to measure the lift (Y1 , Y2 ) and drag forces (X1 , X2 ; see Figure 1(a)). They were calibrated in
situ using calibrated weights from 0 − 5 kg in steps of 0.5 kg.

2.4. Micro-jets
To control the flow around the airfoil, holes of 1 mm diameter with equidistant 8 mm spacing
were placed at x/c = 0.8 from the airfoil’s trailing edge, along the entire spanwise direction.
They were connected to a plenum chamber, itself fed with pressurized air at 6 bar. The air
circuit was connected to solenoid valves that acted as On/Off switches, so that pulsed micro-jets
can be generated with a repetition rate of up to 300Hz, cf. Figure 1(d).

2.5. Control hardware


The control is managed by a STM32 Nucleo board H742ZI2 allowing a 16-bit ADC acquisition
as well as the possibility to monitor the signals in real-time on the computer (Figure 2 presents
the control hardware where the Nucleo board is mounted on a fully featured PCB): the lift
force measured by the force balances (Y1 , Y2 ) is used as input signal. It is acquired with a
sampling frequency of 20 kHz using the Nucleo board. The signal is filtered using a fourth order
Butterworth filter with a cut-off frequency of 20 Hz. The control updates at 20 kHz and drives
the valve at 200 Hz in response to the input from the force balances.

4
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

3. Principles of model-free control


Full details on model-free control are given in [18]. Its usefulness in many situations, including
compensating severe non-linearities and time-varying reference signals, has been demonstrated.
The corresponding intelligent controllers are much easier to implement and to tune than standard
PID controllers which are today the main tool in industrial control engineering (see, e.g., [19]).

3.1. The ultra-local model


In the current application, the unknown description of the plant is restricted to a SISO (single-
input single-output) system because the objective is to control only the lift y (output) thanks
to the pneumatic actuator (input). The unknown description of the SISO plant is replaced by
an ultra-local first order model

ẏ = F + βu (1)
where: the control and output variables are, respectively, u and y; the notation ẏ designates the
time-derivative of y; the time-varying quantity F subsumes the unknown internal structure and
the external disturbances; the constant β ∈ R is chosen by the practitioner such that ẏ and βu
are of the same magnitude. Therefore, β does not need to be precisely estimated.
Equation (1) is only valid during a short time lapse that must be continuously updated: it
implies that F is estimated via the knowledge of the control and output variables u and y.

3.2. Intelligent P controllers


The control law reads as the intelligent P controller, or i-P controller [16]

F − ẏ ∗ + KP e
u=− (2)
β

where y ∗ is the output reference trajectory; e = y ∗ − y is the tracking error; KP is a usual tuning
gain.
The i-P controller (2) is compensating the poorly known term F . Controlling the system
therefore boils down to the control of an elementary pure integrator. The tuning of the gain KP
becomes thus quite straightforward.
To numerically estimate the derivative of y, recent algebraic parameter identification
techniques [20] or homogeneous semi-implicit differentiators [21] can be used. A comparison
of some kind of differentiators can be found in [22] as well as discrete time differentiator based
sliding modes with polynomial corrector terms with respect to the sampling period [23] and
differentiators with variable exponent to improve the rejection of the noise [24].
Practically, as shown in the diagram of the control depicted in Figure 3, the control drives
the signal sent to the solenoid valves to release or block the pressurized air; thus, the micro-jets
act on the airfoil surface which, in turn, influences the lift. The measured lift is tracked in the
closed loop and the valve is driven using a variable duty-cycle (conversion of the output of the
control u as a duty-cycle) at the constant frequency of 200 Hz.
The PWM (pulse width modulation) is a TTL (transistor-transistor logic) signal that allows
driving directly the valves from the control signal that adjusts the mass flow injected to the
system. The actuator frequency is much larger than mechanical ones in the systems, and much
larger than the perturbation imposed to the system, however, it may be of the order of magnitude
of some frequencies of the system. The targeted objective is to get rid of perturbations /
vibrations imposed to the system.
Remark: In real operation, it is assumed that the micro-jets must be spatially controlled
according to the local measured profile of the wind (speed, pressure, flow separation) along the
blade.

5
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Figure 3: General closed loop scheme to track the lift.

4. Experimental results
The experiments are conducted considering a constant wind speed of 20.3 ms−1 and an AoA of
the blade set to α0 = 20◦ ; the chopper is used to create a periodic reduction of the air flow that
allows evaluating the robustness of the model-free control under (strong) perturbations of the
lift.
In equation (2), the coefficients are set to β = 5 · 105 and Kp = 1. The efficiency of the lift
tracking is evaluated for several scenarios. Before starting the control, an identification procedure
of the lift is performed to determine the range of the lift variations that can be compensated by
the micro-jets actuator system. It is identified by a simple succession of opening and closing of
the valves (the identification procedure is illustrated in Figure 4 before the vertical green line).
In the sequel, each figure displays the experimental results when the i-P controller is applied to
the system, following different scenarios.
• Scenario 1: variation of reference trajectories. Figures 4 and 5 exhibit good control
performances for sine and constant reference trajectories, despite flow perturbations that
occur when the reference is changing.
• Scenario 2: external periodic perturbations while starting the control. Figure 6
depicts the tracking of the lift considering a perturbation produced by the chopper. The
control starts and the chopper is switched on at the same time to produce a 0.03 Hz-periodic
decrease of the lift highlighting some robustness towards maintaining stability of the closed-
loop during the first transient of the control. The chopper induces strong ”effort” larger
than the perturbation rejection capability of the system thus saturating the duty-cycle to
’1’; in this case, the tracking fails and the duty-cycle is equal to one thus providing the
maximum of air injection. However, when this huge perturbation disappears, the tracking
is efficient again.
• Scenario 3: external static perturbations while starting the control. Figure 7
depicts the tracking of the lift considering a limited static lift perturbation introduced at
the very beginning of the control process highlighting again some robustness properties
of the closed-loop during the transient of the control. The static perturbation is then
”amplified” at 58 sec to introduce an additional decrease of the lift, thus inducing more
reaction from the control to maintain good tracking performances.
• Scenario 4: variation of reference trajectories and external static perturbation.
Figure 8 depicts the tracking of the lift considering reference variations and a limited static
lift perturbation introduced at the very beginning of the control. The static perturbation is

6
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Figure 4: Tracking by the lift of a sine reference. Top. Measured lift and reference lift versus
time (sec); the vertical yellow line marks where the initialization of the control ends and the
actual test begins. Bottom. Duty-cycle versus time (sec).

then ”amplified” at 68 sec to introduce an additional decrease of the lift, thus inducing more
reaction from the control to maintain good tracking performances despite the variations of
the reference.

To additionally show in which frequency range the control has the biggest impact on the lift,
the energy spectral density E(f ) of the lift is plotted in Figure 9. Up to a frequency of f ≤ 5 Hz,
the control manages to dampen frequency modes that may generate mechanical vibrations of
the blade.

5. Conclusion
In this work, a model-free control algorithm for the manipulation of the lift force acting on
an airfoil, using micro-jets, has been experimented. Different scenarios of experimental tests
have been carried out to illustrate the robustness of the control and promising results show that
good tracking performances have been obtained despite the strong perturbations induced by
the chopper and the limited range of control of the lift using the micro-jet actuator strategy.
Further investigations will include aerodynamic sensors such as remote wall pressure sensors or
remote flow separation sensors.

Acknowledgements
The authors would like like to thank Dr. Pierre Molinaro who designed and built the electronic
control board in the experimental setup.
This work has been partly carried out within the research project ASAPe, grant no. 2018
ASAPe, funded by WEAMEC program, Région Pays de la Loire, France.

7
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Figure 5: Tracking of the lift with a constant reference. Top. Measured lift and reference lift
versus time (sec). Bottom. Duty-cycle versus time (sec).

Figure 6: Tracking of the lift including a chopper perturbation starting from the beginning of
the control. Top. Measured lift and reference lift versus time (sec). Bottom. Duty-cycle
versus time (sec).

8
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Figure 7: Tracking of the lift considering a small perturbation introduced at the beginning of the
control. Top. Measured lift and reference lift versus time (sec). Bottom. Duty-cycle versus
time (sec).

Figure 8: Tracking of the lift considering a small perturbation introduced at the beginning of
the control under reference variations. Top. Measured lift and reference lift versus time (sec).
Bottom. Duty-cycle versus time (sec).

9
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

Figure 9: Energy density of the lift with the control switched on (closed loop) and the control
switched off (open loop).

References
[1] Schepers J et al. 2021 Tech. rep. IEA Wind TCP Task 29, Phase IV: Detailed Aerodynamics of Wind Turbines
[2] Bossanyi E 2000 Wind Energy 3 149–163
[3] Gonzalez A G, Enevoldsen P B, Akay B, Barlas T K, Fischer A and Madsen H A 2018 Journal of Physics:
Conference Series 1037 022039
[4] Matsuda H, Tanaka M, Osako T, Yamazaki K, Shimura1 N, Asayama M and Oryu Y 2017 International
Journal of Gas Turbine, Propulsion and Power Systems 9
[5] Elhadidi B, Elqatary I, Mohamady O and Othman H 2015 International Journal of Mechanical and
Mechatronics Engineering 9
[6] Jaunet V and Braud C 2018 Renewable Energy 126 65–78
[7] Leroy A, Braud C, Baleriola S, Loyer S and P Devinant S A 2016 Journal of Physics: Conference Series 753
(2) 022012
[8] Soulier A, Braud C, Voisin D and Podvin B 2021 Wind Energy Science 6 409–426
[9] Allan B G, Juang J, Raney D L, Seifert A, Pack L G and Brown D E 2000 Closed-loop separation control
using oscillatory flow excitation Tech. rep. NASA/CR-2000-210324, ICASE Report 2000-32
[10] Becker R, Garwon M, Gutknecht C, Barwolf G and King R 2005 J. of Process Control 15 691–700
[11] Shaqarin T, Braud C and Coudert S 2013 Experiment in fluids 54
[12] Becker R, King R, Petz R and Nitsche W 2007 AIAA Journal 45 1382–1392
[13] Bartholomay S et al. 2021 Wind Energy Science 6 221–245
[14] Neunaber I and Braud C 2020 Wind Energy Science 5 759–773
[15] Mariette K 2020 Contrôle en boucle fermée pour la réduction active de traı̂née aérodynamique des véhicules
Ph.D. thesis INSA of Lyon, France.
[16] Fliess M and Join C 2021 Int J Robust Nonlinear Control 1–13
[17] Lafont F, Balmat J F, Join C and Fliesś M 2020 International Journal of Circuits, Systems and Signal
Processing 14 1181–1191
[18] Fliess M and Join C 2013 International Journal of Control 86 2228–2252
[19] Åström K J and Murray R M 2008 Feedback Systems (Princeton University Press)
[20] Fliess M and Sira-Ramirez H 2008 Identification of Continuous-time Models from Sampled Data Advances

10
The Science of Making Torque from Wind (TORQUE 2022) IOP Publishing
Journal of Physics: Conference Series 2265 (2022) 032068 doi:10.1088/1742-6596/2265/3/032068

in Industrial Control ed Wang H G L (Springer) pp 362–391


[21] Michel L, Selvarajan S, Ghanes M, Plestan F, Aoustin Y and Barbot J P 2021 IEEE Journal of Emerging
and Selected Topics in Industrial Electronics 2 227–236
[22] Othmane A, Rudolph J and Mounier H 2021 European Journal of Control 62 113–119
[23] Barbot J P, Levant A, Livne M and Lunz D 2020 Automatica 112 108633
[24] Ghanes M, Barbot J P, Fridman L, Levant A and Boisliveau R 2020 IEEE Trans. on Automatic Control

11

You might also like