A NEW WIND-POWER GENERATION METHOD

EMPLOYED WITH HIGH ALTITUDE WIND

Pim Breukelman1, Michiel Kruijff1, Hironori A. Fujii2, Yuusuke Maruyama3
1
Ampyx Power
2
Kanagawa Institute of Technology
3
Meda Corporation
SUMMARY: A new method of wind-power generation is introduced in this paper. The technology, developed by
AmpyxPower, generates energy by flying a tethered glider-plane attached to a ground-based generator following a cross-wind
pattern as the tether unwinds under high tension (plane spirals away from the generator), and rewinds under near-zero tension
(plane glides back to generator). The benefits, drawbacks and decision rationales of major design choices are discussed:
crosswind operation, rigid plane concept, ground-based generator. The development plan is shared and performance claims
are supported by prototype tests and extrapolations based on validated dynamic simulation, providing confidence that a
highly competitive utility-scale solution can be provided. Keywords: renewable energy, wind power, airborne wind energy
INTRODUCTION
Ampyx Power develops the PowerPlane, a novel wind energy
technology that will eventually allow sustainable production of power
at lower costs than fossil-fueled alternatives. It thus has the potential
to trigger a paradigm shift in the electricity sector.
A NEW WIND-POWER GENERATION METHOD
Concept
PowerPlane systems convert wind power into mechanical power by
having an autopilot-controlled glider plane creating pull on a tether
by flying repetitive cross-wind patterns at an altitude of 200-400
meters (Figure 1). Conversion to electrical power happens in a
ground mounted generator from which the tether is extracted. Once
the tether has been extracted to full length, the glider plane is
controlled to dive/glide to lower altitude, during which phase the
tether is retracted. During this reel-in phase, tether tension is minimal
and power consumption is only a fraction of the power produced
during the reel-out phase.

offshore applications, where the PowerPlane generator can simply be

placed on a floating barge.
With synchronized systems positioned closely together, our
commercial system design (AP-4) is expected to eventually achieve a
park level energy density of about 15-20 MW/km 2 at optimum LCoE
(Levelized Cost of Energy). For prudence reasons (obtaining flight
hour statistics), initial parks will operate at about a third of that
density. Note that if total park output rather than kWh price is the
design driver, higher densities can be achieved at sub-optimal LCoE
(in analogy with conventional wind turbines).
Design choices
Figure 2 highlights the benefits and drawbacks of the main design
choices.
Ampyx Power has opted for voluntary compliance with stringent
civil aviation airworthiness certification standards for its PowerPlane
systems (ED-79/ARP 4754, CS23.1309/ARP 4761, ED-12/DO-178,
safety management system, operations manual, pilot training a.o.).
In order to make it possible for a small company to provide full
design traceability and support process-based engineering to
aeronautical standards, a Project Data & Process Management
Platform has been custom-built to the needs of AWE design by the
company Xignum, based on (space) satellite development experience
(Ref. 7).

Figure 1: Principle of operation

In the case of conventional wind turbines, only the tips of the blades
move at the maximum velocity and generate the vast majority of the
power (which is proportional to air speed cubed), the remainder of
the turbine (blade roots, mast) is virtually dead weight. Moreover, the
fact that the wind force acts on the top of a tall mast contributes to a
massive torque that demands further increase of the mast weight and
a heavy and deep foundation. In contrast, the full wing span of the
PowerPlane is exposed to the high speed of the air flow that is
obtained by the cross-wind trajectory of the aircraft [Ref. 1],
therefore the use of structure is maximally effective. There is no
torque, so the foundation can be minimal. This is particularly true for

Figure 2: Design choices

Approval status
Our prototypes already comply with stringent civil aviation safety
standards (Figure 3). Autonomous flight operations are performed at
our test site under approval and exemption provided by the The
Dutch Airspace Authorities.

Figure 3: AP-2A aircraft and 90 kW generator

PERFORMANCE SUBSTANTIATION
The claimed performance of the PowerPlane has been substantiated
by a multipronged approach:
1. Simulation using detailed modelling of aircraft flight
characteristics (incl. Simulated deflection of all control
surfaces) and controller.
2. Prototype flight jusing AP1 and AP2 (Figure 3), matching
closely the results of step 1.
3. Fast simulation using validated point-mass dynamics for
sizing and optimization purposes.
4. Ampyx sizing toolbox, Figure 11, used to size the
commercial system (AP-4).
AD 2. RESULTS OF PROTOTYPE TEST
Fully autonomous flight tests (launch-climb-entry-pattern-reelin-etc.)
are currently being performed with two near-identical registered
aircraft (AP-2), Figure 3. There is capability for unlimited flight (onboard power generation) and for propulsion (compact landing and
take-off). Autonomous flight without intervention has been
demonstrated for flights of over 2 hours. Flight data of the previous
generation AP-1 (2011) has been made available (Figure 4, Ref. 2),
from which the match with the advanced simulation model (item 1.
above) has been established, and significant net power generation is
showcased (including in fact feeding net power into the grid).

Figure 4: AP-1B test results (2011) showing significant net power

generation.
AD3. SOME RESULTS OF FAST DYNAMIC SIMULATION
For the purpose of optimization within the Ampyx system sizing
toolbox, three fast models have been independently developed:
1. Dynamic path tracker of a pointmass with lift, drag and angle of
attack and roll (no-side-slip assumption). Optimal tension
determination (per time step, max. power). Tether force vector
model. Allows for power capping by angle-of attack control.
2. Dynamic waypoint tracker (target ahead of current position). It
includes physically realistic transition to reel-in.
3. Quasi-static simulation. Force balance including centripetal force.
The three simulations produce with percent points identical results
The exact pattern followed has little influence on power generation,
Results are reported from Simulator 1 (Figure 5) for fairly high wind
speed. At lower wind speed, the tension will be more irregular and
below its maximum, such as shown in Figure 4.

drag of a realistic tether has on performance. With a thin tether, the

maximum tension is reached already at low wind speed, and the
power gain with higher wind speed becomes linear through most of
the regime. It would seem more favorable to have a thicker tether,
such that the cubic growth of power with wind speed is stretched
deep into the high-wind regime. However, such a tether has a large
drag and at low wind speeds, it is not used to its potential (not fully
loaded), and a lot of unnecessary tether drag is suffered. The result is
that the power curves, though cubic, are so flat initially that the
benefit of the thick tether only becomes apparent at very high wind
speeds, that are exceedingly rare.
When the power curves are multiplied with the wind statistics
distribution (e.g. a Weibull), the optimal tether thickness becomes
clear as a function of wing area (Figure 7, Figure 8). It seems that
approximately, the optimal tether thickness would lead to transition
to the linear regime around the average wind speed.
Primary factors that have a significant impact on power production
have been analyzed to be:

Figure 5: Mechanical power (Green) and equivalent max. cycle

power (Red, based on 20 m/s reel-in) - for 35 kg, C L=2.5, S=3 m2
PowerPlane at 11 kN tension, 18 m/s wind speed, resulting airspeed
about 48 m/s. Resp. reference (horizontal 8 pattern) pwoercapped at
80 kW and 10x15 deg circular pattern.
AD 4. SIZING OF A COMMERCIAL SYSTEM
The tools and methodology of Figure 11 has been applied for a
preliminary sensitivity analysis to understand the primary cost factors
and support design designs and first order sizings.

Reel-in speed: increase from 20-30 m/s raises power by 10%

Capping of maximum power to limits of generator: relatively small
effect 50% capping has only 5% effect on yearly power. May be
worth to have generator sized not to peak load.
System efficiencies: With respect to the mechanical real-out power in
ideal conditions (point mass), an overall 50-60% efficiency is
considered a fair target to account for control and power conversion
losses.
Lift of wing profile: can be improved by multi-element wings, flow
control etc. to 2.5 and possibly higher.
Roll angle:A system with large mass or short tether needs a large roll
angle to fly the steep curves of the pattern, longer tether or wider
pattern.
Drag of tether: can be improved by 30% with e.g. latex coating, and
by having a thinner cable although it would have to be replaced more
often (Ref. 3, experiences with Dyneema material in Ref. 6 & 7).
Sizing: Though heavily dependent on the mass model, Figure 8
appears to suggest that a larger aircraft generates more energy per
unit wing area per year, so there is a benefit to scaling up.

Mass model - The advanced mass model is not discussed in this

paper, but results in a mass dependency not unlike the following:
m = S/3 * (35 + (Tmax - 11000) / (1000 + 3 S2 )
m in [kg], S wing area in [m2], Tmax the max. operational cable
tension in [N]. This approximation has been used in most of the
below analyses (where not mentioned otherwise).
Power curves - Typical power curves for different maximum cable
tension are shown in Figure 6. Looking also at Figure 1, three
regimes can be distinguished:
1. Cubic law regime. This is the regime where Lloyd's estimation
works well, at least as a trend (Ref. 1). The power rises with the cube
of the wind speed.
2. Linear regime. The maximum cable tension has been reached, and
the cable is operated at maximum tension. The reel-rate increases
with the wind speed to limit the tension.
3. Flat regime. The max. reel rate or generator power has been
achieved. E.g . through angle of attack control the power is limited.
The lesson learned from this picture is the enormous impact that the

Figure 6: Typical power curves as function of maximum operational

cable tension, for 16 m2 wing area

Figure 9: Qualitative results of AmpyxPower commercial sizing

exercise and AP-4A target sizing.

Figure 7: Wind statistics for test site windprofile (all at pattern

altitude)

Figure 8: Yearly power production for test site statistics, as a

function of maximum operational cable tension, 20 m/s reel-in speed
(above), for various wing areas.

OPTIMAL WING AREA LCoE

Optimal power production however is not necessarily the optimum
we are looking for (Figure 11). Eventually we are optimizing for
minimal levelized cost of energy (LcoE). From a sensitivity analysis
on LCoE (including an advanced load-case based structural mass
model.), based on variation of all discussed design parameters, there
appears to be an optimal wing area for minimal LcoE (Figure 9).
Figure 10 illustrates the road to such an optimum (AP-4), through the
existing prototype (AP-2) and the functionally representative and
certifiable, yet more modestly dimensioned AP-3 currently under
development.

Figure 10: PowerPlane generations AP2 (left, in operation), AP3

(middle, under development), AP4 (commercial version)

Figure 11: Ampyx Power toolbox and methodology for system sizing
to minimal LCoE
CONCLUSIONS
The AmpyxPower Powerplane concept generates energy by flying a
tethered glider-plane attached to a ground-based generator following
a cross-wind pattern as the tether unwinds under high tension (plane
spirals away from the generator), and rewinds under near-zero
tension (plane glides back to generator). AmpyxPower currently
operates 3 prototype PowerPlanes (incl. AP-2A1 and AP-2A2, about
20 kW net power production demonstrated) in a test-field in The
Netherlands. The fAP-3 is currently under development following
stringent aeronautical design processes, airworthiness and safety
standards. The AP-4 (2 MW class) shall operate at a LcoE well below
that of conventional wind energy, as predicted by extrapolation of
flight data based on validated dynamic simulation over a range of
wind speeds for typical sites, coupled to a structural mass model and
cost model including capital and operational aspects.
REFERENCES
[1] M.L. Loyd, Crosswind Kite Power, J. Energy. Vol. 4, 1980, pp.
106-111.
[2] R. Ruiterkamp and S. Sieberling, Description and preliminary
results of a 6 DOF rigid wing pumping system, Airborne Wind
Energy edited by U. Ahrens e.a., Springer, 2013.
[3] T.P. Jung, Wind Tunnel Study of Drag of Various Rope
Designs, proceedings of 27th AIAA Applied Aerodynamics
Conference, AIAA 2009-3608.
[4] Bootle, W. J., Forces on an Inclined Cylinder in

Supercritical Flow, AIAA Journal, 9(3):514-516,

December, 1970.
[5] S.R. Norris and D. Andrisani II, Longitudinal equilibrium
solutions for a towed aircraft and tow cable, AIAA 2001-4254.
[6] R. Bosman, Engineering guidelines for Airborne Wind Energy
tethers with HPME fibers, Airborne Wind Energy edited by U.
Ahrens e.a., Springer, 2013.
[7] M. Kruijff, Tethers in space, a propellantless propulsion in-orbit
demonstration, ISBN 978-90-8891-282-5, 2011.
COPYRIGHT
The copyright belongs to the authors. The Grand RE2014 Organizing
Committee has a right to publish Abstract books and proceedings for
this conference.