Sieberling2013_Chapter_FlightGuidanceAndControlOfATet
Sieberling2013_Chapter_FlightGuidanceAndControlOfATet
Sören Sieberling
1 Introduction
Harvesting wind energy is being investigated for several decades with many inter-
esting outcomes as a result, [6]. One of the concepts introduced by [3] in the early
eighties, describes a technique using a tethered glider that drives a generator on the
ground through tension in the tether. [3] refers to this concept as lift power, also
known as pumping-kite power. Today the pumping-kite system represents a subset
of airborne wind energy, which has been a growing industry in the past decade with
the sole objective of outperform conventional wind turbines in terms of cost per unit
of energy.
The operational principle of the pumping kite system is discussed in Sec. 2, which
is followed by Sec. 3 providing the physical framework. Sec. 4 presents the flight
Sören Sieberling
Ampyx Power B.V., Lulofsstraat 55 - 13, 2521AL Den Haag, The Netherlands
e-mail: [email protected]
338 S. Sieberling
Fig. 1 The pumping kite system during power generation (left) and during reset (right).
control system of the pumping kite system split into longitudinal and lateral control.
Sec. 5 demonstrates some results from test flights, followed by the conclusions in
Sec. 6.
1
Pw = ρ Vw3 (1)
2
A specific device can generate more power than another by being located in an
area with more wind (higher wind power density). The operational altitude of the
pumping kite system is between 400 and 500 m. When assuming that the wind pro-
file increases logarithmic with altitude [7, 4], wind power at for example 80 m hub
Flight Guidance and Control of a Tethered Glider 339
height compared to 400 m doubles (above agricultural land). In terms of wind speeds
this translates roughly into 25% wind speed increase.
A second advantage is that wind turbines are not expected to grow in the near
future [5]. Consider that power is cubically related to airspeed. With airspeed over a
wind turbine blade increasing linearly, it follows that most power in a wind turbine
is generated by its blade tips. When assuming that the power generated by a section
of the wind turbine blade is only a function of airspeed, hence neglecting the effect
of airfoil and chord variation, almost 60% of the power would be generated by the
outer 25% of the blades. In effect the pumping kite system could be compared to
this outer section of the wind turbine blades. The difference to wind turbines is that
it does not require a large tower and can operate with only one ’blade’, thereby
removing one of big upscaling obstruction of conventional wind turbines.
3 Physical Framework
The operation of the pumping kite system and the possible power output as a func-
tion of the wind speed is best described by basic aerodynamic equations and deriva-
tives thereof, [3]. For simplicity the influence of mass on the system is neglected
in these derivations. The glider is assumed to have no roll angle compared to the
tether. And the wind is assumed to have a constant velocity, (Vw ), parallel to the
ground plain.
The lift generated by the glider is given by Eq. 2.
1 2
L = ρ VTAS SCL (2)
2
With L the lift, VTAS the true airspeed of the glider, S the wing surface area and CL
the lift coefficient. The drag of the system consists of two components, glider drag
and tether drag. The drag of the glider is given by Eq. 3.
1 2
Dac = ρ VTAS SCD (3)
2
With Dac the glider drag and CD the drag coefficient, approximated by
CL2
CD = CD0 + (4)
π ARe
Where CD0 represents the zero lift drag coefficient and the remaining term the in-
duced drag with e the span efficiency factor and AR the aspect ratio. The tether drag
is approximated by assuming that the tether is straight, [2]. Since the lift of the glider
is about one order of magnitude larger than the tether drag, the resulting errors are
negligible. The effect of the true wind on the tether drag is also neglected. The drag
of an infinitesimal section of the tether ds is given by Eq. 5.
1
dD = ρ Vds2 CDc tds (5)
2
340 S. Sieberling
With Vds the speed of an infinitesimal section of the tether, t the tether thickness
and CDc the tether drag coefficient. By assuming a straight line (the speed changes
linearly with position on the tether) the sectional speed is approximated by Eq. 6
s
Vds = VTAS (6)
l
With l the tether length. Substituting Eq. 6 into Eq. 5 and equating moments around
the generator of the tether drag (integrated over the tether length) on one hand and
a resulting force on the glider, the resulting tether drag is derived, Eq. 7, [2].
l l
1 sVTAS 2 1 2 2
Dc l = sdD = s ρ CDc tds = ρ VTAS tl CDc (7)
0 0 2 l 8
The lift to drag ratio (G) of the pumping kite system is then obtained by dividing
Eq. 2 by Eq. 3 and Eq. 7 resulting in Eq. 8.
CL
G= tlCDc
(8)
CD + 4S
Where the right hand component of the denominator will also be referred to as
effective tether drag coefficient. In a massless system, the system lift to drag ratio
equals the ratio of forward and upward airspeed (Fig. 2), or in other words the lift
to drag ratio describes the forward speed of the glider as a function of wind speed,
tether angles and tether speed, Eq. 9.
With Vp the forward speed, Θ the in the wind plane rotation of the tether (Fig. 3),
Φ the out of the wind plane rotation of the tether (Fig. 3) and Vt the tether speed.
Note that this equation holds only for a massless system, or in other words when the
system is in equilibrium.
For rigid wing kites, the difference between Vp and VTAS becomes negligible due
to the high system lift to drag ratio, Eq. 10.
VTAS ≈ Vp (10)
Furthermore, when assuming a high lift to drag ratio, the tether tension is approxi-
mately equal to the lift. With the tension then given, the resulting power during the
power generating phase is computed by Eq. 11.
P = ηm TVt (11)
With ηm the mechanical efficiency of the motor and other winch components. In the
derivation above the tether speed has not been specified. [3] however demonstrates
that for maximum power production the tether speed, should be 1/3 of the wind
speed component perpendicular to the glider, thus Vw sin Θ cos Φ /3.
Flight Guidance and Control of a Tethered Glider 341
Vw sinθ cosφ
Xz Θ
Φ
Vp
VTAS
T y x
D
Xk
z
Fig. 2 Force and speed diagram during Fig. 3 Definition of tether angles with re-
power production for forces in equilib- spect to wind direction blowing along the
rium, showing identical ratios between X-axis. Θ in the wind plane rotation and
L/D and Vp /Vw sin Θ cos Φ . The dashed Φ out of the wind plane rotation.
lines indicate the kinematic reference
frame (X pointing in the direction of the
airspeed).
Note that these derivations are a simple means to analyzing the pumping kite
systems characteristics. The biggest assumption is that the system is massless. In
reality the glider is constantly maneuvering to stay inside the wind window and thus
constantly accelerating hence not being in equilibrium. The true motion is governed
by the equations of motion conventional to aircraft, Eqs. 12.
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤
u̇ X − sin θ p u
⎣ v̇ ⎦ 1⎣ ⎦
= Y + g ⎣ sin φ cos θ ⎦ − ⎣ q ⎦ × ⎣ v ⎦ (12a)
ẇ m Z cos φ cos θ r w
⎡ ⎤ ⎛⎡ ⎤ ⎡ ⎤ ⎧ ⎡ ⎤⎫⎞
ṗ L p ⎨ p ⎬
⎣ q̇ ⎦ = J −1 ⎝⎣
M ⎦ − ⎣ q ⎦ × J⎣q⎦ ⎠ (12b)
⎩ ⎭
ṙ N r r
⎡ ⎤ ⎡ ⎤⎡ ⎤
θ̇ 1 sin φ tan θ cos φ tan θ p
⎣ φ̇ ⎦ = ⎣ 0 cos φ − sin φ ⎦ ⎣ q ⎦ (12c)
sin φ cos φ
ψ̇ 0 r
⎡ ⎤ ⎤θ ⎡
⎡ cos cos θ
⎤
VN u VwindN
⎣ VE ⎦ = Tnb ⎣ v ⎦ + ⎣ VwindE ⎦ (12d)
VD w VwindD
Fig. 4 The lift to drag ratio of a clean free flying glider vs angle of attack compared to the
lift to drag ratio of a tethered glider with a fixed tether length of 400m vs. angle of attack.
from the body to the inertial reference frame, which can be found in any common
book on aircraft dynamics.
When using the derivation as a qualitative means to describe the pumping kite
system scaling, two observations can be made, [8]:
(A) From the discussion above it is concluded that when flying at higher lift to
drag ratios the power output grows. Since not only the glider lift to drag ratio is
of concern but the lift to drag ratio of the system, higher lift to drag ratio within
operational flight envelopes almost always correspond to higher lift and therefor
higher angle of attack, Fig. 4.
As is also stated by Eq. 9, higher lift to drag ratio also corresponds to higher
true airspeed, which results in the unconventional flight characteristic for a tethered
glider that the airspeed is raised by pitching up instead of pitching down.
(B) For free flying glider the lift to drag ratio provides a means to compare glider
performance because it is dimensionless. For tethered glider the system lift to drag
ratio is however not independent of dimension, meaning that system lift to drag
ratios of differently sized systems should not be compared without corrections.
The scaling dependency of the system lift to drag ratio is caused by the tether di-
mensioning. When expressing the scaling of glider in terms of wing surface, twice
as much wing surface results in twice as much lift and drag, hence tension. Twice
the amount of tension requires twice the amount of tether, hence the tether cross-
sectional surface doubles. Since the system lift to drag ratio does not depend on
tether cross section but on tether diameter, the effective tether drag coefficient will
grow with the square root of the tether cross section, hence proportional to the square
root of the wing surface. Therefore the relative contribution of the tether drag be-
comes smaller.
Fig. 5(a) illustrates the system lift to drag ratio of differently sized pumping kite
systems having identical aerodynamic characteristics in terms of glider lift and co-
efficients. Furthermore the tether length is identical for different sizes. Sizing of
Flight Guidance and Control of a Tethered Glider 343
Fig. 5 System lift to drag ratio and total drag coefficient as a function of glider dimension
the tether thickness is based on the tension at a fixed airspeed and lift. In reality
the tether length will grow slightly for increasing systems as will airspeed, which
would have a softening effect on the differences in system lift to drag ratio. Fig. 5(b)
illustrates the glider and tether drag coefficients, which illustrates that the system lift
to drag ratio grows for aerodynamically identical glider as the system is scaled up,
thus tether drag is a bigger problem for smaller systems.
FI
Ψz ψz
θy
-Θy
φx
Φx
φtx FB
FT ψtz
θty
Fig. 6 Reference frame transformation summary. Reference frames are indicated by the cap-
ital letter F, with the subscript indicating the specific frame. Superscripts are used to indicate
the axis of rotation.
With Cti the direction cosine matrix mapping inertial coordinates into tethered co-
ordinates and the other way around by taking the transpose, Cit = CtiT . When intro-
ducing the conventional axis transformation from the inertial reference frame to the
body fixed reference frame by means of the Euler angles (roll (φ ), pitch (θ ) and yaw
(ψ )) as Cbi , the direction cosine matrix to the body frame from the tethered frame is
given by Eq. 14.
Cbt = CbiCit (14)
The tethered Euler angles are then derived consequently as in Eqs. 15. The transfor-
mations are summarized in Fig. 6.
φt = tan−1 Cbt(2,3) /Cbt(3,3) (15a)
θt = sin−1 −Cbt(1,3) (15b)
ψt = tan−1 Cbt(1,2) /Cbt(1,1) (15c)
Fig. 7 Block diagram indicating the functioning of the longitudinal control scheme during
power generation
the lift to drag ratio of the system (Fig. 4), which lowers the airspeed (which again
lowers the tension). Raising the winch speed, lowers the effective wind at the plane,
which lowers the airspeed and thereby the tension. It is not hard to understand that
both control inputs in effect influence the same parameters, which can easily grow
into instabilities. What makes this control problem more complicated is that commu-
nication between winch and glider is over radio suffering from transmission delays
and that the winch has implementation delays orders of magnitude larger than the
glider implementation delays.
Fig. 8 Block diagram indicating the functioning of the longitudinal control scheme during
system reset
controlled to maintain a specific flight path angle that is derived from a sink demand
setting, which is scheduled against wind speed.Since the winch speed in this phase
is set, so is the ground speed. Which implies that the true airspeed, and thus the
drag, rises with increasing wind. The scheduling is therefore chosen to have the
reset phase consist of a steeper dive for higher wind speeds such that gravity com-
pensates for the glider drag as much as possible, yet never makes it exceed the tether
speed.
Fig. 9 Block diagram indicating the functioning of the longitudinal control scheme during
transition into the power generation phase
Fig. 10 Block diagram indicating the functioning of the lateral control scheme
Fig. 11 Scheduled waypoints during the power generation phase in tether angles
Table 1 Pumping kit states used in power generation and corresponding criteria to complete
the task of that state and trigger transition into a next state
2500 8
2000
4
−2
1000
−4
−6
500
−8
0 −10
140 150 160 170 180 190 200 210 140 150 160 170 180 190 200 210
Time (s) Time (s)
40
True airspeed (m/s)
40
35
30 45
25 50
20 55
140 150 160 170 180 190 200 210
Θ (rad)
Time (s)
60
15
Airflow angles (deg)
10 65
5 70
0 75
−5
140 150 160 170 180 190 200 210 −20 −10 0 10 20
Time (s) Φ (rad)
(c) True airspeed and angle of attack (d) Position on spherical and waypoints
60 40
Roll Roll
Pitch Pitch
40
20
20
0
Attitude (deg)
Angle (rad)
0
−20
−20
−40
−40
−60
−60
−80 −80
140 150 160 170 180 190 200 210 140 150 160 170 180 190 200 210
Time (sec) Time (s)
Fig. 12 Test results of one power cycle at 7 m/s wind speed, measured at 6.5 m reference
height. Starting point of the graphs is the reset phase, followed by power generation and
ending in another reset phase.
350 S. Sieberling
6 Conclusions
Through choosing the tether based reference frame and carefully selecting control
variables, simple control schemes are well capable of flying the highly aggressive
patterns of the pumping kite system. In lateral direction the glider is controlled by
means of a waypoint controller that produces track demands. Tracking errors gen-
erate roll angle demands, which again are used to create roll rate demands each
through linear controllers. The longitudinal motion is controlled by fixing the eleva-
tor and thereby fixing the lift coefficient and system lift to drag ratio in combination
with a tension controller determining the tether velocity.
References
1. Diehl, M., Houska, B.: Windenergienutzung mit schnell fliegenden Flugdrachen: eine Her-
ausforderung für die Optimierung und Regelung - Wind Power via Fast Flying Kites: a
Challenge for Optimization and Control. at-automatisierungstechnik (2009)
2. Houska, B., Diehl, M.: Optimal Control for Power Generating Kites. In: Proc. 9th Euro-
pean Control Conference (2007)
3. Loyd, M.L.: Crosswind Kite Power. J. of Energy (1980)
4. Manwell, J.F., McGowan, J.G., Rogers, A.L.: Wind Energy Explained Theory, Design and
Application. John Wiley and Sons, Ltd. (2002)
Flight Guidance and Control of a Tethered Glider 351
5. Thresher, M., Robinson, M., Veers, P.: To Capture the Wind. IEEE Power and Energy
Magazine (2007)
6. Williams, P., Lansdorp, B., Ockels, W.: Optimal Cross-Wind Towing and Power Genera-
tion with Tethered Kites. J. of Guidance, Control and Dynamics (2008)
7. Wortman, A.J.: Introduction to Wind Turbine Engineering. Butterworth (1982)
8. Sieberling, S., Ruiterkamp, R.: The PowerPlane an Airborne Wind Energy System Con-
ceptual Operations. In: Proc. 11th AIAA ATIO Conference (2011)