Power Curve Analysis Of On-ground Airborne Wind

Energy Systems
Yashank Gupta, Jonathan Dumon, Ahmad Hably

To cite this version:

Yashank Gupta, Jonathan Dumon, Ahmad Hably. Power Curve Analysis Of On-ground Airborne
Wind Energy Systems. IEEE International Conference on Industrial Technology (ICIT 2019), Feb
2019, Melbourne, Australia. �hal-01973326�

HAL Id: hal-01973326

https://hal.archives-ouvertes.fr/hal-01973326
Submitted on 8 Jan 2019

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 Power Curve Analysis Of On-ground Airborne
Wind Energy Systems
Yashank Gupta, Jonathan Dumon, Ahmad Hably
Univ. Grenoble Alpes, CNRS, Grenoble INP∗ ,GIPSA-Lab,
F-38000 Grenoble, France.
∗
Institute of Engineering Univ. Grenoble Alpes
{yahsank.gupta, jonathan.dumon, ahmad.hably}@gipsa-lab.fr

Abstract—In this paper, an algorithm is proposed to control power curves with the dynamic simulations of the Enerkite AWE proto-
parameters of an on-ground airborne wind energy system for type EK30 and it concludes that the significant confidence can be placed
the operational wind speed range, considering different winch on the approach presented in the work for estimating the power curves
configurations and the related saturation of its components, for AWE systems. In [5] a simplified model is analyzed to estimate the
including electrical and hydraulic devices. Coupled with a static maximum feasible drag power for an on-board production system. In
model, this leads to a tool that can easily calculate expected power [6] an optimal control problem is discussed which is then used to obtain
curves for on-ground airborne wind energy system as a function power curves for a rotary kite AWE system. In the authors’ previous
of design parameters. The paper also presents the numerical work [7], a strategy to control the power production of a Magnus-
examples of the power curve for a Magnus-based system and based AWE system is proposed. In [8], the authors propose a 6-DOF
draws a comparison with the power curves of horizontal axis
mathematical model for the Magnus based AWE system validated by
wind turbines.
simulation, controlling the system in crosswind trajectories. A static
Index Terms—Airborne wind energy system, On-ground sys-
model of the full cycle is presented and compared to the dynamic
tem, pumping system, Power curve, Ground station design,
Magnus effect, Static modeling, High-level control, Hydraulic simulation. In this present study, based on a static model proposed and
stage validated in [8] and a structural analysis of ground station structure,
including electrical and hydraulic solutions, a generic static model of on-
ground AWES is proposed. A high-level algorithm is the developed to
I. I NTRODUCTION maximize the net output power of the system, taking into consideration
the necessary saturation of the system. This fast model can then be used
Airborne wind energy systems (AWE) aims at proposing a com-
to calculate power curves of on-ground AWES in function of different
plementary solution to the conventional wind turbines. AWE systems
design parameters. As authors are working specifically on Magnus-
harness the energy of high altitude winds by using an airborne platform
based on-ground AWE systems, numerical application for these type of
connected to the ground by tethers or conducting cables. AWE systems
systems is done to draw comparisons with conventional horizontal axis
have attracted a lot of interest in the last few decades (see [1] for a survey
wind turbines. The power curves calculated in this paper give a valuable
on soft kites and [2] for AWE systems in general).
insight into the potential of AWE systems and expose some advantages
Depending on the location of energy production, AWE systems can over conventional wind turbines.
be divided into two main classes: On-ground production and On-board The paper is organized as follows. Section II focuses on the modeling
production AWE systems. On-ground production systems consist of a of the different parts of the system. In section III, the control and the
ground-based generator connected to an airborne platform by tethers. optimization of the output power are presented. A numerical application
Whereas in the on-board production systems, the turbine is mounted on follows in section IV. The paper ends with some conclusions in section
the airborne platform. Currently, in the research community, there are V.
many innovative designs regarding the airborne platform. The choice
of the platform depends on the lifting capabilities of the aerodynamic II. M ODELLING
surface. Classically, kites and unmanned aerial vehicles (UAV) have
been used as the airborne platform. But with an increased interest in A. Model of Wind Profile
AWE usage of many innovative concepts are being explored such as In this paper, wind profile power law is considered to describe the
Magnus cylinders, rotary kites, balloons etc. evolution of horizontal mean wind speed with altitude. This theoretical
Power curves serve as a tool to analyze the economic feasibility of model is discussed in detail in literature [9] and provides a good
any type of wind turbine. In literature, there are many studies discussing approximation of the wind speed for the altitudes between 100 to 2000
the power curves of conventional wind turbines. Over the years, the m. According to this model, the wind speed Vw at any altitude z can be
power curves have been validated and improved by incorporating the given by:
z
on-field data from the installed wind turbines. To evaluate the economic Vw (z) = Vw (z0 )( )α (1)
z0
viability of airborne wind energy systems, and to draw valid compar-
isons with the conventional turbines, there is a need to develop accurate Where, z0 represents the operating altitude, Vw (z0 ) is the known wind
power curves for AWE systems. Currently, there are very few working speed at altitude z0 , and α is an empirically derived coefficient that
prototypes of AWE systems and none of them is a fully functioning characterizes the surface. It depends on the stability of the atmosphere
commercial unit. Thus, the power curves for AWE systems are still an and is generally assumed to be equal to 0.143. To take into account the
open topic of discussion in the research community. In [3] a study is constant variation of the wind speed and to calculate the mean wind
presented discussing the family of power curves for different altitudes speed at a particular site throughout the year, the well know Weibull
derived from a model presented in [4]. The study also compares the distribution is used in this paper. The distribution basically tells at a
particular site how often the wind blows and how strong it is. Thus, it is
a good way to describe the wind speed variations and it is given by:
 k−1
k
k v v
f (v) = e− A (2)
a A
Where, f (v) is a probability to have v wind speed over the year, A
is the Weibull scale parameter expressed in m/s and is proportional to
mean wind speed, and k is the Weibull form parameter describing the
shape of the Weibull distribution with its value between 1 and 3. Smaller
values of k show very variable winds while larger values show constant
winds. Finally, v is the wind speed series whose probability distribution
is calculated.
Fig. 1. Magnus-based AWE system.
B. Model of a Horizontal axis wind turbine (HAWT)
For conventional wind turbines, the power produced depends on the
The component of the wind in the direction parallel to the tether, as
kinetic energy of the air and Cp the power coefficient of the turbine
experienced by the platform, can be expressed as:
which is smaller than its theoretical Betz limit 16/27:
Vt = Vw cos(θT ) cos(φT ) (6)
1
PHAW T = ρCp Aswept Vw3 (3)
2 Where, θT and φT represents the elevation and the azimuthal angle
Where, Aswept is the total surface swept by the blades. Power produced, of the tether with respect to the inertial frame. Figure 1 presents a 3D
PHAW T , is generally divided into four phases characterized by the model of a Magnus-based AWE system described using two frames of
design constraints of the system. The first phase is from zero wind references. Inertial frame, (xi , yi , zi ), centered at the ground station O,
speed to cut-in wind speed Vci , where a conventional turbine does not and a body frame, (xb , yb , zb ), centered at the center of gravity Cg of
produce any energy. The second phase is from Vci to the nominal wind the system. The tether length is presented by variable r which can be
speed Vnom , where the maximum wind power extraction occurs by expressed as: p
maximizing Cp coefficient. The third phase is from Vnom to the cut- r = x2 + y 2 + z 2 (7)
off wind speed Vco , and during this phase power is curtailed to nominal Where, (x, y, z) are the coordinates of Cg in the inertial frame. In the
power Pnom by reducing Cp coefficient. Finally, for any wind speed static analysis of [8], assuming straight taut tether, the traction force Fr
beyond Vco , the wind turbine is switched-off to avoid its mechanical developed in the tether is directly proportional to the resultant aerody-
and electrical degradation. Therefore, the power curves of a HAWT can namic force acting on the Magnus cylinder. Thus, the total mechanical
be produced by the following set of equations: energy available at the winch of the on-ground station is given by
 1
ρCpmax Aswept Vw3 if Vci < Vw ≤ Vnom P = Fr ṙ. Where, ṙ represents the reel-in or reel-out speed of the tether.
 2 1) Production phase: The power produced by any AWE during



a production phase is given by:

PHAW T = Pnom if Vnom < Vw < Vco (4)
Pprod = Fr ṙprod (8)




0 if Vw ≥ Vco or Vw ≤ Vci

Where, ṙprod refers to the reeling-out speed of the tether from the winch.
C. Static Model of Ground-based Airborne Wind Energy Sys- Fr refers to the traction force developed in the tether due to the resultant
tem aerodynamic forces. Considering the reel-out speed ṙprod , the power
produced during the production phase is given by:
The production cycle for any on-ground AWE system is divided into
two phases, namely, the production phase and the recovery phase. The 1 CL 2
Pprod = ρS(Vt − ṙprod )2 CL ( ) ṙprod (9)
production phase is characterized by reeling out the tether from the on- 2 CD eq
ground station while following a crosswind trajectory in the air. As a Here it is important to note that this static power equation first derived
result, a traction force is produced in the tether attached to the airborne in [4] and further refined in [10] is obtained by using equilibrium
platform which is then used to generate electricity. In the recovery phase, motion theory for crosswind maneuver. It does not take into account
the airborne platform is pulled back to the on-ground station. Thus, a the contribution of the centrifugal force acting on the airborne platform.
part of the power produced during the production phase is consumed In addition, it is assumed that the wind is parallel to the tether and the
during the recovery phase. In order to minimize the energy consumption AWE system is assumed in the static state. Thus, a detailed analysis
during this phase, the traction force has to be minimized. considering all the effects may yield a better expression for power
Aerodynamic lift and drag forces acting on any classical airborne produced by an AWE system. For the production phase, the maximum
platform (such as kite, rigid wing, UAV or Magnus cylinder etc.) can possible theoretical mechanical power that can be produced for a given
be expressed as: set of lift and drag coefficients CL and CD eq is studied in [4]. In [10]
1 1 it is proved that for maximum power the reel-out speed of the tether,
L= ρSVa2 CL , D = ρSVa2 CDeq (5) ṙprod , should be equivalent to V3w while the AWE system follows a
2 2
crosswind trajectory such as circles or eight-like figures perpendicular
Where, ρ is the air density, Va is the apparent wind velocity, S is to the wind direction. Thus, following this assumption the maximum
the projected surface of the airborne platform, CL is its aerodynamic theoretical mechanical power produced can be expressed by:
lift coefficient, and CDeq is equivalent aerodynamic drag coefficient
1 4 CL 2
of the system including the drag due to the tether and any other Pprod = ρ SVt 3 CL ( ) (10)
structural components of the airborne platform. Note that CDeq has to be 2 27 CD eq
expressed in function of tether length to consider additional drag effect However, this maximizes only the production phase and not the total
for each additional tether length. power produced during the full cycle, as discussed later in section III.
 2) Recovery phase: In the recovery phase, the airborne platform grid. The system is then able to provide a constant power during all
is returned to its initial position .i.e. the tether length is brought to the cycle. In this case, no power is taken from the grid, thus, ensuring a
its initial value. In this maneuver, a portion of the energy produced unidirectional connection to the grid. Figure 2 presents the configuration
during the production phase is consumed. The power consumed can be with two separate electric actuators for the two phases and a local
calculated as the product of the winding speed of the tether ṙrec and the storage device.
resulting drag force of the system through the tether Fdrag :
Prec = Fdrag ṙrec (11)
1 2
Prec = ρS(Vt + ṙrec ) CDrec ṙrec (12)
2
Where, CDrec is the resulting drag coefficient during the recovery
phase. Note that in this formulation, as ṙrec is negative and Fdrag is
positive, so Prec is negative.
There are several operational strategies proposed in the literature
to minimize the energy consumption during the recovery phase for
different types of AWE system. For instance, in case of soft kites based Fig. 2. Ground station using two separate electrical actuators and a storage
AWE systems to minimize the energy consumption, generally, the kite is device
flown to the zenith to the limits of the wind window before commencing
the reel-in of the tether. In an another approach, proposed in [11], use of a hydraulic stage
In case of a Magnus-based system, in order to minimize the power between the drum and the grid connection is considered. Figure 3 shows
consumption, the rotational speed of the Magnus cylinder, ωcyl is simply the configuration with a variable speed hydraulic motor/pump coupled
set to zero, leading to lift coefficient CL zero. Thus, minimizing the with a hydraulic accumulator and a fixed speed motor which produces
resulting drag. In [8], this type of maneuvers is performed in dynamic electricity which is then fed directly to the grid. Thanks to the emerging
simulations. high efficiency digital hydraulic devices, the authors expect substantial
Depending on the choice of airborne platform and recovery strategy, cost reductions in the cost of both actuators and storage devices. From
Prec might have to be modified to incorporate all the effects and a control point of view, as the same actuator is used for both the phases,
variables arising due to maneuvers performed for the recovery phase. this leads to Fprodmax = Frecmax and ṙprodmax = ṙrecmax . As
This model is adapted to a platform that can be configured with no or
very low lift coefficient during the recovery phase. It can be replaced by
another static model of the recovery phase, such as the one proposed in
[3] that includes a given set of lift and drag coefficients and a necessary
additional transition time.

D. Ground-based Airborne Wind Energy System design con-

straints
The ground station is a reversible winch that produces energy when
the airborne platform is pulling the tether, and during the recovery phase,
it produces a retraction force to reel-in the tether to the drum. Actuators Fig. 3. Ground Station using hydraulic stage
to be used will have saturations that are chosen to represent the following
parameters: electrical as well as hydraulic actuators has a yield that can vary with
• The nominal rotational speed ωnom
operating speed and torque, we can define for all theses configurations
• The nominal torque Γnom
common parameters ηp , yield of the generator (or pump) used during
• The efficiency η which depends on the actuator’s rotational speed
production phase, ηr yield of the motor used during recovery phase,
ω and torque Γ ηs yield of the storage device and ηg yield of the grid connection.
For hydraulic case, yield of grid connection includes hydraulic motor,
The saturation on traction force, Fmax , and reel-out and reel-in speed,
electrical generator and grid connection itself.
ṙmax , are dependant on the design of the ground station. They have to
1) Full production cycle: A given power cycle is thus defined
be considered in the control algorithm. Note that the system must be
by one complete production and recovery phase. The transitional phase
designed in such a way that Fmax = Rd Γnom , and ṙmax = Rd ωnom ,
between the production and the recovery phase is neglected in this
where Rd is the radius of winch’s drum. On the other hand, the tether
formulation. The time for each phase can then be expressed as:
and the airborne platform must be able to withstand the maximum
traction force Fmax . In this work, two configurations for the ground rmax − rmin ∆r
t1 = = (13)
station architecture are considered. In the first one, an electrical motor- ṙprod ṙprod
generator is used for both the phases .i.e. production and recovery phase. rmax − rmin ∆r
t2 = = (14)
In the second configuration, each phase uses a separate device .i.e. an ṙrec ṙrec
electrical generator with a gearbox optimized for production phase and Where, t1 and t2 represent the duration of the production and recovery
an electrical motor with a gearbox optimized for the recovery phase. phase respectively. Thus, the total time, t taken for one complete
In this case, saturation for the each phases has to be differentiated production cycle is:
as Fprodmax , ṙprodmax , Frecmax and ṙrecmax . Note that the same t = t1 + t2 (15)
differentiation has to be done if a single motor is coupled with two
different gearboxes. The net output power provided to the grid Pcycle can then be calculated
The winch-generator system then can be directly connected either to by the sum of power produced affected by its yield, power consumed
the grid or to a local storage device which can be used to provide the also affected by the inverse of its yield and the yield of storage device,
power required during the recovery phase and the mean power to the and additional power loss in the storage device Ploss . This is due to the
storage of the necessary energy that has to be provided to the grid during while considering this time that ṙrec as constant. The first order partial
recovery phase. Its expression is given in equation (16). derivative results in a polynomial function, and solving the system of
equation gives the optimal values of ṙrec and ṙprod :
Pcycle t2 (ηs−1 − 1)
Ploss = (16) 
t1 + t2 ṙrecopt = fs (Vw , ṙprod , M )
(22)
ṙprodopt = gs (Vw , ṙrec , M )
Hence, the mean power produced during one full production cycle can
be computed as: B. Optimization of reel-in speed when reel-out speed is set
ηp Pprod t1 + ηr−1 ηs−1 Prec t2 − Pcycle t2 (ηs−1 − 1) As presented before, the power produced by an AWE system in
Pcycle = ηg
t1 + t2 production phase is given by:
(17)
Where, replacing the values of t1 and t2 from equations (13) and (14) Pprod = Fr ṙprod (23)
into (17), the power produced in one complete cycle can be expressed
as: Due to design and structural constraints, there exists an upper limit for
ηp Pprod ṙrec + ηr−1 ηs−1 Prec ṙprod traction force Fr denoted by Fmax . This limit can be attributed to tether,
Pcycle = ηg (18)
ṙrec + ηs−1 ṙprod winch and/or kite structural limitations. When this limit is reached
during the production phase, and if cos(θT ) cos(φT ) and CL ( CCLeq )2
Note that from equation (18) it can be observed that in order to obtain D
are set at their maximum value, then ṙprod has to be set such that
maximum power from one complete cycle, there is a trade-off between
Fr = Fmax , and it can be calculated using equation (9):
ṙrec and ṙprod . Hence, the augmentation of ṙrec not only increases the
contribution of the production phase but also increases the value of the s
Fmax
power consumed Prec as given by equation (12). Similarly, increasing ṙprodf = Vt − 1 CL
(24)
ρSC L( C )2
ṙprod not only increases the power produced as expressed by equation 2 Deq

(9) but also the power consumed.

In this case, if ṙprod reaches ṙprodmax then it can be saturated, but Fr
III. C ONTROL AND O PTIMIZATION OF OUTPUT POWER has to be maintained equal to Fmax . This can be  done 
by reducing other
2
parameters such as cos(θT ) cos(φT ) or CL CCDLeq . Thus, in this
As explained in the section II, for a ground-based AWE system, the
case in order to maximize Pcycle the reel-in speed has to be optimized
net power produced in one cycle depends on physical parameters, design
for Fr = Fmax and ṙprod given by equation (24). The optimal reel-in
parameters, and controllable parameters. All controllable parameters of
speed ṙrecopt1 can be found by:
Pprod described in equation (9) must maximize Pprod , except reel-
out speed ṙprod which also influences the whole cycle through Prec
 . ∂Pcycle
2
= f1 (Vw , ṙprod , rrec , Fr , M ) (25)

This leads to maximization of cos(θT ) cos(φT ) and CL CCDLeq . ∂ ṙrec
Similarly, all controllable parameters of Prec described in equation (12)
has to minimize it, except reel-in speed ṙrec which also influences where, Pcycle is calculated from equation (18) with a saturation on
the whole cycle through Pprod . In this phase, cos(θT ) cos(φT ) and traction force and is calculated as:
CD rec has to be minimized. Thus, to obtain the maximum mean power, ηp Fmax ṙprod ṙrec + ηr−1 ηs−1 Prec ṙprod
it is necessary to obtain optimal values of ṙprod and ṙrec . In this Pcycle = ηg (26)
ṙrec + ηs−1 ṙprod
section, a high level algorithm is described, using all control parameters
available for on-ground AWE systems in presence of saturation. An and, the optimal reel-in speed rrec˙opt1 is given by:
optimization technique is discussed for three separate situations which
characterizes different typical constraints due to on-ground winch and ṙrecopt1 = fs1 (Vw , ṙprod , Fr , M ) (27)
airborne platform design, namely:
• Optimize both ṙrec and ṙprod Similarly, if ṙprod reaches ṙprodmax before the saturation of traction
• Optimize ṙprod when ṙrec is set force Fr then ṙrecopt1 is calculated from equation (27), using ṙprod =
• Optimize ṙrec when ṙprod is set ṙprodmax and Fr are calculated from eqs. (8) and (9).

A. No saturation on actuators: Optimization of reel-in and C. Optimization of reel-out speed when reel-in speed is set
reel-out speeds In some situations, reel-in speed is required to be saturated to
To maximize the net power production during one complete cycle, ṙrecmax , before other variables like traction force or reel-out speed
ṙprod and ṙrec has to be optimized simultaneously. In the considered reach their upper limit. As optimal reel-in speed can be several times
optimization procedure, Pcycle is derived with respect to each ṙrec and higher than the reel-out speed, especially in the case when the ground
ṙprod while considering the other variables as constant. station uses the same actuator for production and recovery phases .i.e.
without a two-stage gearbox. In such situations, the reel-in speed can
∂Pcycle reach its upper limit before the traction force. Thus, in this case the
= f (Vw , ṙrec , ṙprod , M ) (19)
∂ ṙrec optimal ṙprodopt2 has to be calculated using equation (22) and ṙrecmax :
while considering ṙprod , Vw and the vector of parameters M as con- ṙprodopt2 = gs2 (Vw , ṙrecmax , M ) (28)
stant. Vector M is given by:
Here it is important to note that in this case it is mandatory to check other
M = [CL , CDeq , CDrec , φT , θT , ηp , ηr , ηs , ηg , ρ, S] (20)
variables for saturation. Thus, if ṙprodopt > ṙprodmax then ṙprodopt has
Similarly, to be set to ṙprodmax . On the other hand, by using equation (9) if the
∂Pcycle traction force is found to be greater than its maximum i.e. F r > Fmax .
= f (Vw , ṙrec , ṙprod , M ) (21) Then ṙprod has to be calculated by using equation (24).
∂ ṙprod
D. Control by using altitude C. 90m span MGAWES parameters
Control of working altitude can be done by controlling rmin and The parameters considered are listed in table II. Aerodynamical
rmax parameters i.e. minimum and maximum tether length. This can parameters are taken from [14], and for the sake of comparison, some
be done in order to find an altitude that maximizes power produced other parameters are designed to get the same power curve as for the
or to curtail production by lowering wind speed when it is necessary. considered HAWT for wind speeds varying from Vci to Vnom . Also, for
Thus, there is a maximal effective wind speed Vtmax beyond which the same reason same cut-in and cut-out wind speeds are used.
Pcycle starts to decrease despite of the augmentation of wind speed. Two types of ground stations as described in section II-D are considered.
This is due to the fact that at Vtmax , due to all the design constraints, One with two separate electrical actuators, and other with hydraulic
the production phase saturates while the power required for the recovery stage with their apecifications also listed in table II.
phase continues to grow with the increase in wind speed. Seeking an
Kite parameters Variable Value
altitude where the wind is lower is useful when the wind is greater than Span [m] L 90
the optimal wind speed of the system. Radius [m] R 9
Magnus Surface [m2 ] S 1620
 2  2
E. Control of elevation angle θT or azimuthal angle φT Maximum CL CCL for X = 3.6 CL CCL 69.44
Deq Deq

To curtail output power or reduce effective wind speed, the trajectory Minimum drag coefficient CDrec 0.5
Minimum elevation angle [deg] θT 25
of the airborne platform can be controlled, in order to modify elevation Cut-in wind speed [m/s] Vci 3
angle θT or azimuthal angle φT . This gives an additional control Cut-out wind speed [m/s] Vco 22.5
variable in order to operate on-ground AWES. Maximal traction force [kN] Fmax 2405
Maximal strength [N/m2 ] σ 1485
Working altitude [m] z 160
F. Summary Of The High-Level control algorithm Aspect ratio AR 5
Reynolds number for Va = 10m/s Re 10.9e6
A high-level control strategy to maximize net power production Case 1: Ground station with 2 electrical actuators
Pcycle in presence of multiple actuator saturation for all operating wind Electric generator nominal power [MW] Pprodmax 10
speed range presented in this section can be summed as the following. Nominal Power for grid connection [MW] Pgrid 5.91
Reel-in and reel-out speed has to be optimized, taking into account Electric motor nominal power [MW] Precmax 5.56
system saturations. With both reel-in and reel-out are saturated, control Yield of generator ηp 0.92
Yield of motor ηr 0.88
of altitude z , elevation angle θT , and/or azimuthal angle φT can be used Yield of storage device ηs 1
to maintain output power to its nominal value. Yield of grid connection ηg 0.98
Maximal reel-in speed [m/s] ṙrecmax 14.7
Maximal reel-out speed [m/s] ṙprodmax 4.16
IV. NUMERICAL APPLICATION Case 2 : Ground station with hydraulic stage
Electric generator nominal power [MW] Pgrid 10.9
In this section, different set of parameters of Magnus-based on- Hydraulic motor/pump nom. power [MW] Pmax 40
ground airborne wind energy system (MGAWES) are considered. Power Yield of motor/Pump for production ηp 0.92
curves, annual production, and capacity factor are compared with those Yield of motor/Pump for recovery ηr 0.88
Yield of storage device ηs 1
of similar size horizontal axis wind turbine (HAWT).
Yield of grid connection ηg 0.98
Maximal reel-in and reel-out speed [m/s] ṙmax 16.65
TABLE II
A. HAWT parameters PARAMETERS OF THE 90 M SPAN MGAWES.
VESTAS V 150 − 4.2M W T M , designed for IEC IIIB wind class,
is chosen as HAWT system to be compared with on-ground AWES.
Corresponding parameters presented in table are taken from [12] or D. Results
deduced from power curve data set in [13].
Power curves of HAWT V150 both with theoretical model and dataset
are plotted in figure (4). Power curves derived using Pcycle for a 90m
Variable Value
Nominal Power [MW] Pnom 4.2 span MGAWES for case 1 and case 2 ground station configuration are
Cut-in wind speed [m/s] Vci 3 also plotted in figure (4). As case 1 represents a more relevant example
Cut-out wind speed [m/s] Vco 22.5 of different configurations of saturation expressed in section III, thus,
Rotor diameter [m] DHAW T 150 for this case Pprod and Prec are also plotted in figure (4). To describe
Tower height [m] h 160 the 4 different phases of the power curve, denoted by I to IV, reel-in and
Power coefficient Cp 0.45
reel-out speed, traction forces and elevation angle are shown on figure
TABLE I (5).
PARAMETERS OF HAWT VESTAS V 150 − 4.2M W T M .
As it can be observed from figure (4) phase II of HAWT and MGAWES
are very similar. This is because in this phase, just like HAWT,
MGAWES is a function of Vw3 , and an optimal amount of kinematic
energy is extracted from the wind. During phase III, the traction force
B. Wind parameters is saturated, but Pprod continues to increase as reel-out speed increases
with an increasing wind speed. In phase IV, once the maximum Pprod is
Wind is considered at an altitude at the tower height i.e. 160m reached then only Prec increases as with an increase in the wind speed
with mean annual wind speed 7.5m/s. For annual production, Weibull the drag also increase. Thus, more and more power is required to recover
distribution is used with parameters A = 8, 47m/s and k = 2, which the airborne platform. This leads to a reduction in the net power output
leads to class IIIB. Pcycle as represented by dotted line in figures (4) and (5). Finally, solid
lines in phase IV shows the use of elevation angle θT control, to maintain
15 80
the effective wind Vt at its maximal value in order to the maintain output

reel-in, reel out speed (m/s)

Elevation angle T (deg)

power at its maximum. In table II, case 2 represents the case when there 60
10
are no set limitations on the winch actuators. This set of parameters are
40
chosen to illustrate hydraulic configuration because for both the phases 5
I II III IV

actuator is same. So, the winch actuators have to at least produce the 20

high force Fmax and high speed rrec˙max in the same time. This extends 0 0
the phase III to maximum cut-out value Vco . The maximum output 0 5 10 15
(V w) wind speed (m/s)
20 25

power is then produced at this maximum wind speed, and has to be 106
2.5
used to size the grid connection Pgrid . Note that the same power curve

Traction Forces (N)

can be produced with a single 40M W electric motor/generator coupled 2

I II III IV
with a single gearbox. Using these theoretical power curves and wind 1.5

distribution described in subsection IV-B, the theoretically expected 1

energy produced during one year can be computed for HAWT V150 and 0.5

different ground-station configurations for the same 90m span Magnus 0

0 5 10 15 20 25
effect kite. In particular, it is interesting to note that just for the sake (V w) wind speed (m/s)
of comparison, using figure (4), the use of a generator and a motor that
saturates at the same wind speed as HAWT (i.e. Pprodmax = 6.7M W
and Precmax = 4M W ) will lead to a similar power curve and Fig. 5. High level control variables of case 1, 90m span MGAWES. Top: reel-
similar annual production. This configuration is considered as case 0. out speed during production phase (black) and reel-in speed during recovery
Table III summarizes the different systems considered and associated phase (red) in function of wind speed. Dotted line represents variables when
size of actuators. It also gives the corresponding theoretical annual control of elevation angle is not used. Down: Traction force during production
production and capacity factor, computed by annual production divided (black) and recovery (red) phases.
by 8760Pgrid .

System PGrid Generator Motor Prod. Capacity of different phases, each phase corresponding to the different control
[MW] [MW] [MW] [GWh/yr] factor variables, and illustrates the high flexibility of on-ground airborne wind
Vestas V150 4.2 4.2 - 18.2 0.49 energy system.
AWES Case 0 4.2 6.7 4 18.2 0.49 The strategy presented can be used to evaluate other types of AWE
AWES Case 1 5.9 10 5.56 21.9 0.42
AWES Case 2 10.9 40 - 24.8 0.26 systems and draw some valid comparisons with conventional wind
TABLE III turbines.
C ONSIDERED ACTUATORS CONFIGURATIONS AND CORRESPONDING
THEORETICAL ANNUAL PRODUCTION AND CAPACITY FACTOR R EFERENCES
[1] M. Ahmed, A. Hably, and S. Bacha, “High altitude wind power
systems: A survey on flexible power kites,” in 2012 XXth International
Conference on Electrical Machines, Sept 2012, pp. 2085–2091.
106
[2] A. Cherubini, A. Papini, R. Vertechy, and M. Fontana, “Airborne
12
VESTAS V150 theoretical
VESTAS V150 dataset
wind energy systems: A review of the technologies,” Renewable and
10
P Cycle MGAWES system, case 1 Sustainable Energy Reviews, vol. 51, pp. 1461–1476, 2015.
case 1, P prod
case 1, P rec [3] M. Ranneberg, D. Wolfle, A. Bormann, P. Rohde, F. Breipohl, and
8 case 1, without angle modulation
P Cycle MGAWES, case 2
I. Bastigkeit, “Fast power curve and yield estimation of pumping
Power (W)

airborne wind energy systems,” vol. AWEC, 2017.

6
[4] M. Loyd, “Crosswind kite power,” J. ENERGY, vol. 4(3), 1980.
4 [5] F. Bauer, R. M. Kennel, C. M. Hackl, F. Campagnolo, M. Patt, and
I II III IV R. Schmehl, “Drag power kite with very high lift coefficient,” Renewable
2
Energy, vol. 118, pp. 290 – 305, 2018.
0
[6] J. De Schutter, R. Leuthold, and M. Diehl, “Optimal control of a rigid-
0 5 10 15
(V w) wind speed (m/s)
20 25 wing rotary kite system for airborne wind energy,” 03 2018.
[7] A. Hably, J. Dumon, and G. Smith, “Control of an airborne wind
energy system with a Magnus effect,” in The 2016 American Control
Fig. 4. Power curves: Power production comparison between HAWT Vestas Conference, Boston, United States, Jul. 2016. [Online]. Available:
V150 and case 1 and case 2 of MW-size MGAWES. Pcycle is decomposed https://hal.archives-ouvertes.fr/hal-01272253
into production and recovery contributions for case 1. Dotted line denotes [8] Y. Gupta, J. Dumon, and A. Hably, “Modeling and control of a magnus
Pcycle when control of elevation angle is not used effect-based airborne wind energy system in crosswind maneuvers,”
IFAC-PapersOnLine, vol. 50, no. 1, pp. 13 878–13 885, 2017.
[9] C. L. Archer, “An introduction to meteorology for airborne wind energy,”
in Airborne Wind Energy, 2013, pp. 81–94.
V. C ONCLUSION [10] I. Argatov, P. Rautakorpi, and R. Silvennoinen, “Estimation of the me-
In this work, a static model is proposed as a simple tool that can be chanical energy output of the kite wind generator,” Renewable Energy,
vol. 34, pp. 1525–1532, 2009.
used to predict the performance of the on-ground AWES in relation to [11] M. A. Tattersfield, J. F. Atwater, and C. P. Houle, “Method and system
the main design parameters. The proposed algorithm gives a strategy for harnessing wind energy using a tethered airfoil,” Jun. 2 2015, uS
to maximize the energy produced by this kind of system for different Patent 9,046,072.
configurations of actuator saturation under the static assumption. By [12] VESTAS, “https://www.vestas.com/en/products/turbines,” 2018.
coupling the static model with the high-level algorithm, a way to [13] windparka, “http://www.windparka15lingewaard.nl,” 2018.
calculate power curves is presented. This approach is used to study the [14] M. Milutinović, M. Corić, and J. Deur, “Operating cycle optimization for
a magnus effect-based airborne wind energy system,” Energy Conversion
effects of design parameters on performances and can be used directly and Management, vol. 90, pp. 154–165, Jan 2015.
to control a system in real-time. The resulting power curves consist