Renewable Energy 31 (2006) 271–283

www.elsevier.com/locate/renene

Design, simulation, and testing of a novel

hydraulic power take-off system for the Pelamis
wave energy converter
Ross Henderson *
Ocean Power Delivery Ltd, Edinburgh, Scotland, UK
Available online 29 September 2005

Abstract
The economic viability of a wave energy converter depends largely on its power take-off system. Active
control of the power take-off is necessary to maximise power capture across a range of sea-states and can
also improve survivability. The high force, low speed regime of wave energy conversion makes it a suitable
application for high-pressure hydraulics.
This paper describes the hydraulic power take-off system employed in the Pelamis wave energy
converter. The process of the system’s development is presented, including simulation and laboratory tests
at 1/7th and fullscale. Results of efficiency measurements are also presented.
q 2005 Elsevier Ltd. All rights reserved.

Keywords: Hydraulic power take-off; Wave energy converter; Pelamis

1. Introduction

The extraction, by any wave energy converter (WEC), of useful energy from ocean waves
requires that the waves apply force to some form of responsive mechanism able to resist the
working force that the waves apply, and some form of reference against which that mechanism
can react. The mechanism by which energy is transferred between the waves and the WEC, and
subsequently or directly into useful form, is generally known as the power take-off (PTO).
Control of the response of a WEC can be achieved through active control of the PTO.
Control of the PTO is central to the economic viability of wave energy; a major sustainable
resource that has, as yet, remained untapped. It is important to show the potential of wave energy
at an early stage by firstly, ensuring that machines can survive under arduous conditions, and

* Tel.: C44 131 553 9495.

E-mail address: [email protected]

0960-1481/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2005.08.021
272 R. Henderson / Renewable Energy 31 (2006) 271–283

secondly, by absorbing and converting a respectable amount of energy. These goals can, in large
part, be achieved through control of the PTO.
To extract a maximum amount of power from the incident waves, the PTO should be capable
of applying a restraining force varying appropriately with time as the WEC responds. This
requires real-time measurement of response, and control of the PTO over the wave cycle.
Moreover, to operate effectively across different sea-states, the control and PTO must be able to
adapt such that power absorption is maximised in small seas and the risk of damage is minimised
in large seas.
For maximum power absorption, a WEC should respond in resonance with the waves such
that the exciting force and the response velocity are in phase [1]. While, with appropriate design,
a WEC can possess inherent dynamics such that its natural response frequency is in the correct
range to match the central excitation frequency of most sea-states, active control is necessary to
maximise power capture across a range of sea-states and in irregular (realistic) seas where a
range of component frequencies are present. Control can also play a vital role in improving
survival characteristics although, whilst crucially important, this aspect is not generally
expanded on in the existing literature.

2. Hydraulic PTO

Wave energy conversion can be considered a very suitable application for hydraulics. Waves
apply large forces at slow speeds and hydraulic systems are suited to absorbing energy under this
regime. Moreover, it is a simple matter to achieve short-term energy storage, necessary to
achieve the smooth electricity production required for a marketable machine, with the use of
cheap and available high-pressure gas accumulators.
Hydraulic systems have many favourable characteristics and many WEC proposals have
incorporated hydraulic PTO in their design [2–4]. However, the majority of literature on PTO
has focused on air turbines operating within oscillating water columns, such as employed in
existing shoreline plants [5,6]. This is perhaps the result of researchers gravitation to the
technology most deployed and recognised, rather than the ultimate suitability of air turbines as a
means of wave power conversion.
Salter et al. [7] and Artemis Intelligent Power Ltd [8] have developed a novel digital
hydraulic pump/motor, originally intended for a wave energy application, capable of offering a
continuously variable transmission of hydraulic power at much higher efficiencies than
conventional hydrostatic transmissions. While the technology is nearing commercial
application, it is still at the prototype stage and unsuitable for immediate deployment in a WEC.
Eidsmoen [9] described a hydraulic control system using a latching strategy, originally
suggested by Budal [10] and independently by French [11] to avoid the need for variable
transmission or reactive power.

3. Pelamis PTO

The Pelamis [12,13] is an offshore, floating, slack-moored wave energy converter consisting
of a set of semi-submerged cylinders linked by hinged joints. Ocean waves perform work on the
Pelamis by moving adjacent cylindrical sections relative to each other across two degree of
freedom joints. The two axes that comprise each joint are inclined to the horizontal to allow a net
inclined response to be induced by the PTO, which resists and reacts against the relative angular
motion of the joints. The inclined response offers an effective hydrostatic stiffness reduced from
 R. Henderson / Renewable Energy 31 (2006) 271–283 273

a vertical response, resulting in a natural frequency dependent on the slope. Thus, the machine
can be designed with a resonant response matched to the dominant frequency of the wave
climate in which it is to be installed, minimising the reactive power requirements of the PTO.
The level of excitation transmitted to the resonant response can be controlled via the PTO.
The Pelamis PTO consists of sets of hydraulic cylinders that pump fluid, via control
manifolds, into high-pressure accumulators for short-term energy storage. Hydraulic motors use
the smooth supply of high-pressure fluid from the accumulators to drive grid-connected electric
generators.
The Pelamis PTO can be examined and developed in two separate parts, which we will call
the primary and secondary transmission. The primary transmission, consisting of the hydraulic
cylinders and their controls, converts the work done by the waves on the structure to stored
energy. The secondary transmission, consisting of hydraulic motors coupled to electric
generators, converts the energy stored in the hydraulic accumulators into electricity transmitted
to shore. This separation, provided by the high-pressure accumulators, allows efficient
absorption over a large range of incident power.
The impeding joint moment provided by the cylinders must be varied throughout each wave
cycle such that the waves do the maximum amount of work on the machine as a whole. Real-
time control of the joint moment is achieved with the use of sets of electronically controlled
valves, which control the flow of fluid between the hydraulic cylinders and the accumulator and
reservoir. The valves are controlled to enable different combinations of chambers to contribute
to the applied joint moment.
Conventional hydrostatic transmissions, which use coupled variable displacement pumps to
deliver continuously variable pressure and flow, typically have a peak efficiency of around 60%.
Efficiency drops dramatically away from the ideal operating point. Furthermore, a conventional
system rated to absorb the full range of incident powers would be uneconomical and hopelessly
inefficient in low power regimes.
In the Pelamis PTO, whenever a chamber is pressurised it exchanges fluid directly with
storage accumulators; the only inefficiencies are associated with compressibility, bearing and
seal friction of the hydraulic cylinders, and flow losses through valves and pipes. Careful design
can set these combined losses at well under 20% over a wide range of operating conditions. The
Pelamis primary transmission can comfortably absorb incident powers ten times higher than the
average incident power to which the generation equipment should be economically matched.
High efficiency is maintained at low incident powers, which represent much of the available
wave energy throughout the year.

4. PTO simulation

A suite of numerical simulation software has been developed by Ocean Power Delivery Ltd
(OPD) to verify and develop the Pelamis design and allow prediction of annual average power
capture at specific sites [14].
Simulations and tank models initially assumed linear mechanical impedances at each axis
(i.e. a moment is applied to each axis proportional to angular position and velocity). Control of
this type is easily quantified and is straightforward to model both mathematically and
experimentally. Assuming linear control and linear hydrodynamics allows frequency domain
methods to be applied, offering very fast computation over a range of conditions. In tank tests,
the response of the Pelamis with linear impedances has been extensively studied over a wide
range of conditions, in terms of both power capture and survivability [15].
274 R. Henderson / Renewable Energy 31 (2006) 271–283

However, although modelling linear control gives great insight and continues to be useful, the
hydraulic PTO applies n output moment that is subject to a variety of physical effects. It is
therefore vital to gain an understanding of this quite different form of control and of the effects of
real hydraulic components on machine response, power capture, and survivability. To this end,
the existing time-domain simulation software was extended to include realistic control and PTO
models, thus allowing development and optimisation of control algorithms in the fast, risk free,
and relatively cheap virtual environment of computer simulations.

Fig. 1. Diagrams showing the response of the Pelamis WEC and the physical layout of the PTO.
 R. Henderson / Renewable Energy 31 (2006) 271–283 275

Hydraulic cylinders act primarily as pumps

Control
manifold
Motor Axis Centre of
High pressure Low pressure rotation

High-
Generator
pressure
accumulator
Control
manifold Low
pressure
reservoir

Hydraulic cylinder

Fig. 2. Much simplified schematic of the Pelamis PTO system.

The controller and the power take-off models are programmed in separate subroutines with
information passing to and from the control subroutine in the same way as for a real controller.
This separation allows for the easy translation of control programmes between simulations and
actual hardware.
The control subroutine samples joint angles from the existing hydrodynamics subroutines just
as a real controller would sample transducer signals. A control algorithm is then applied and the
output is passed to the PTO model in a similar format to the real controller. The PTO subroutine
then models the physical hydraulic system and provides the resulting applied joint moment to the
rest of the program. Other useful signals such as chamber and accumulator pressures and flows
are also output for analysis.
Elements and effects of the PTO were treated as follows:

† Valves are modelled using conditional statements and functions relating pressure and flow.
Effects modelled include: passive and active operation, cracking pressures, fluid resistance
curves, and switching delays. Modelling the physical dynamics of each valve would be very
computationally expensive and not result in significantly different results.
† Flow losses in pipes and manifolds are included in the fluid resistance curves associated with
the valves. For example, an open valve between a chamber and the high-pressure
accumulator has an associated function of flow against pressure drop including the valve
cracking pressure and resistance, fittings, conduits, and piping between the chamber and the
accumulator. This function is applied to the derived flow to or from the chamber.
† Accumulators are modelled using gas laws with heat transfer accounted for by applying a
suitable polytropic index. This may be extended to a more general heat transfer model.
276 R. Henderson / Renewable Energy 31 (2006) 271–283

(a) Axis Moments for Sway 3

1
Moment (normalised)

Control Moment Physically Applied

0.5

0
0 10 20 30 40 50 60
–0.5

–1
Time (s)

(b) Absorbed and Generated Powers

800
700 Instantaneous absorbed Mean Absorbed Generated
600
Power (Kw)

500
400
300
200
100
0
0 50 100 150 200 250 300 350 400 450 500
Time (s)

Fig. 3. Example plots of simulation output for the Pelamis operating in a typical moderate sea (a) Time-series of the
moments at a single joint axis showing the moment demanded and assumed at the output of the controller along with
moment that is physically applied by the PTO. The physically applied moment includes the effects of delays,
compressibility, and fluid losses. The moment scale is normalised with respect to maximum moment; (b) An example
plot of total power absorbed and converted to electricity (generated) over a number of wave groups. The smoothing effect
of separate primary and secondary transmission is evident.

† Fluid compressibility is modelled within chambers as a volumetric spring based on a constant

bulk modulus and the chamber volume on closure. An effective bulk modulus, combining the
compressibility of fluid, entrained gas, and pipe work etc. is often used when describing
hydraulic systems.
† Each piston seal is modelled assuming coulomb and viscous friction with a gain to represent
pressure activation.
† Power generation equipment is modelled using relationships between pressure, flow, and
efficiency. The accuracy of the generation model does not affect that of the primary
transmission model (Figs. 1 and 2).

Fig. 3 shows some simulation output including moments from a single axis and total absorbed
and generated powers from the entire Pelamis operating in a moderate sea.

5. 1/7th scale model

Early in the Pelamis development programme, a need was identified for an intermediate scale
‘systems’ demonstrator with which to develop and prove the full-scale Pelamis hydraulic,
control and data acquisition systems. A scale of 1/7th was chosen as large enough for
functionally realistic systems to be tested while remaining small enough to be tested in existing
(large) wave tanks and avoid the need for specialist handling equipment.
 R. Henderson / Renewable Energy 31 (2006) 271–283 277

The 1/7th scale joint systems were designed to be functionally very similar to the full-scale
design with similar relative piston areas, control valves, and accumulator/reservoir sizes. This
enables control algorithms developed and tested at 1/7th scale to be transplanted into the full-
scale system without major changes. There are however some important differences between the
hardware implementations.
The lower flow and pressure regime at 1/7th scale allows direct-acting solenoid-operated
valves to be used without a piloting system as required at full-scale, where the solenoid valves
act like relays to provide the hydraulic pressure to open larger valves. This simplifies the system
hydraulically but makes no difference to the control system.
The 1/7th scale hydraulic cylinders are mounted externally rather than internally and connect
via flexible hoses to a single manifold that houses the complete set of valves and hydraulic
transducers and fixes directly to the high-pressure accumulator, reservoir, and filter. This
manifold is contained in an electro-hydraulic power pack including all instrumentation, data-
acquisition, control electronics and wiring.
Since, power scales with the scaling factor to the power of 3.5, the 1/7th scale prototype
absorbs w1000th of the power of the equivalent full-scale machine. No hydraulic motor-
generator set is included because accurately modelling these at 1/7th scale is neither feasible nor
necessary. Instead, the generation system is represented by a pressure compensated flow
regulator, which dissipates stored energy as heat.

6. The 1/7th scale test rig

The 1/7th scale hydraulic system was developed and tested using a laboratory rig. The 1/7th scale
PTO test rig, initially actuated by hand, was later adapted for actuation by a ball-screw operating
under closed-loop control to perform the role of the waves. Pre-prototype hydraulic circuit and
component test assemblies were designed and constructed for ad hoc experimentation (Fig. 4).
This work culminated in a design applied to build seven ‘power pack’ units, six for the joint
axes of the 1/7th scale Pelamis and a spare lab test unit. Fig. 5 shows the rig being used to test
and develop control software using one of the finished power packs from the 1/7th scale
prototype.
The 1/7th scale test rig was used for a set of tests designed to demonstrate the operation of the
Pelamis PTO, test implementations of basic control algorithms, and to verify the mathematical
model developed for computer simulation. Pressure transducers were fitted to each ram chamber
volume to allow measurement of pressure at every point in the system (a valuable approach later

Fig. 4. Photographs of the 1/7th scale model joint during a sea-trial and the whole machine in the narrow towing tank at
I’Ecole Centrale de Nantes.
278 R. Henderson / Renewable Energy 31 (2006) 271–283

Fig. 5. (a) Photograph of a fully assembled power pack positioned upside down from the orientation in which it is
installed. The hydraulic cylinders are connected to the manifold via flexible hoses fitted to ports extending through the
hatch; (b) The 7th scales test rig, fitted with a ballscrew actuator under position control, being used to test a 7th scale
power pack.

adopted at full-scale). Signals used by the control system, such as accumulator and reservoir
pressure, and angular position, were also measured using a data acquisition system.
Figs. 6 and 7 show a time series of moment applied by the 1/7th scale PTO. Moment is also
plotted against angular velocity to show the effective impedance applied by the PTO.
Each power pack has an independent controller, which when installed in the 1/7th scale
Pelamis, links to a common centralised hub mounted in the nose of the Pelamis. The hub acts as
the communication bridge between an external shore-based control PC and the individual joint
controllers via an external Ethernet or radio modem link, as is the case for the full-scale design.
Initially, all real-time control was carried out locally by the controller mounted in each power
pack. This was later extended to allow real-time centralised control from the hub.

1
Moment

0 1 2 3 4 5 6

–1
Time

Fig. 6. A graph showing the applied moment derived by summing individually measured chamber pressures. The scale is
normalised with respect to maximum moment.
 R. Henderson / Renewable Energy 31 (2006) 271–283 279

(a) 1 Test Rig (b) Model in waves

Moment

–1
Angular velocity Angular velocity

Fig. 7. (a) A plot of the impedance applied by the PTO system, derived from the applied moment in Fig. 6 above and
plotted to the same moment scale; (b) The same function applied by the 7th scale Pelamis model during tank tests, also
plotted on the same moment and velocity scale. The maximum moment reached is lower due to a lower operating
pressure but the slope and width of the function is the same.

The 1/7th scale Pelamis first underwent sea-trials in the Firth of Forth off Edinburgh and was
later tank tested at I’Ecole Centrale de Nantes, France. The results proved both the novel control
and PTO concept and its implementation at the hardware and software level.

7. Full-scale test rig

After making use of a laboratory test rig at 1/7th scale, it was decided to apply a similar
approach to the full-scale PTO prior to deployment on a full-scale Pelamis. A single power
module, representing one of the universal joints of the Pelamis, was constructed using the same
component assemblies planned for use in the full-scale prototype Pelamis. A 1 MW rated
hydraulic actuation system was mounted externally to perform the role of the waves (Fig. 8).
Initially, the control system of the primary transmission was transplanted from the 1/7th scale
system. After changing the appropriate parameters, the full-scale system operated as expected
under the control algorithms developed at 1/7th scale. The control set-up was later swapped for a

Fig. 8. A photograph of the full-scale joint test rig.

280 R. Henderson / Renewable Energy 31 (2006) 271–283

Full-scale joint test rig moment time series

1
Moment

–1
0 2 4 6 8 10 12 14
Time (s)

Fig. 9. Example moment time series from the full-scale joint test rig undergoing a test with sinusoidal excitation. The
moment scale is normalised.

more flexible and robust distributed system based on the 1/7th scale technology. Figs. 9–11 show
some results from the full-scale test rig.
The efficiency of the primary transmission was measured using the full-scale test rig by
comparing the average power absorbed by the hydraulic cylinders with the constant power
dissipated from the storage accumulator through a relief valve. The moment applied by the
hydraulic cylinders was measured via pressure transducers in each chamber. The absorbed
power at the axis could then be calculated by multiplying measured moment by angular velocity.
Fig. 11 shows the instantaneous and average absorbed power of a single axis during a typical test
at moderate power.
The flow rate into the high-pressure accumulator was measured by treating the hydraulic
cylinders as displacement flow meters. This takes volumetric efficiency into account since only
flow into the accumulator at high-pressure is measured. The high-pressure fluid transferred into
the accumulator, and therefore available to the secondary transmission, was dissipated through a
throttle valve at a constant rate. This power available to the secondary transmission was
calculated as the pressure drop across the throttle valve multiplied by the mean flow out of the
accumulator, which at steady state is equal to the mean measured flow into the accumulator.
The fluid efficiency (combined volumetric and flow efficiency) of the primary transmission,
between the hydraulic cylinders and the storage accumulators, could then calculated by dividing

1
Moment

–1
Angle Angular velocity

Fig. 10. Applied moment from Fig. 9 plotted on the same scale against angular position and velocity. The slope of the plot
on the left gives an indication of stiffness (in this case there are no mass terms); the slope of the plot on the right
approximates damping.
 R. Henderson / Renewable Energy 31 (2006) 271–283 281

Absorbed Power
120 axis instantaneous power
100 axis mean power
Power (kW)

80
60
40
20
0
–20 0 2 4 6 8 10 12 14
Time (s)

Fig. 11. Plot of power absorbed at a single axis during the same test at moderate power on the full-scale joint test rig as
illustrated in Fig. 9. The instantaneous absorbed power can be seen to be negative over part of the cycle as reactive power
is returned from the storage accumulator. The mean absorbed power is also plotted.

the power dissipated from the accumulator by that absorbed at the rams. The efficiency was
found to be between 88 and 94% depending on the control settings, the level of peak power, and
the amount of reactive power applied. These figures include the effects of fluid compressibility
and flow losses through valves, manifolds, and hoses. Efficiency is expected to drop with
increased reactive power as a greater proportion of the fluid flow does not transfer any net power.
The losses are dominated by pressure drops across valves, manifold galleries, and lengths of
hose. Additional losses due to seal friction in the hydraulic cylinders were not measured but have
been estimated to be around 5%.

Fig. 12. Comparison of results between simulation (solid) and experiment (dashed) for (a) the 7th scale; and (b) the full-
scale PTO systems running under their respective test rigs. The pressures of two chambers of a hydraulic cylinder are
plotted showing the effects of real components on the chamber pressures, which ultimately determine the moment applied
at the axis.
282 R. Henderson / Renewable Energy 31 (2006) 271–283

The efficiency of the primary transmission can be significantly improved by careful design of
hose connections and the selection of valves for reduced flow losses. Efforts are ongoing in this
regard.
Electrical generation equipment (the secondary transmission) was also tested using the full-
scale rig and performed as expected with conversion efficiencies in excess of 80% over a
representative range of conditions. Results are not given in this paper.

8. Simulation verification

The PTO simulation was verified using angle and control data from experiments conducted
on both the 1/7th and full-scale test rigs. The simulated PTO system outputs the applied moment,
flows, pressures etc. resulting from the time-series of angle and of control output recorded during
the test.
Results from the computer simulation were compared with corresponding experimental
results from the 1/7th scale and the full-scale test rigs. Fig. 12 shows chamber pressure time-
series from the PTO simulation plotted alongside experimentally measured values. The
agreement can be seen to be good. This verification of the PTO model, along with similar
verifications of the hydrodynamics simulation, gives confidence in the use of simulations to
predict power capture and to further develop control algorithms.

9. Conclusions

A novel hydraulic power take-off system (PTO), designed for use in the Pelamis WEC, has
been demonstrated. First at 1/7th scale, then at full-scale. In each case, a laboratory test rig was
used for experimentation and development. The aim of this staged development was to
maximise learning and minimise risk. Each successive demonstration of the technology relied on
knowledge gained and technology developed in the previous stage.

1. The PTO allows an arbitrary moment to be applied at each joint axis of the Pelamis as
demanded by the control algorithm in real-time. Reactive power capability is provided to
enable impedance matching to maximise power capture.
2. Separation of the primary and secondary transmissions with energy storage allows smooth
output power to be generated efficiently from a highly variable input power. The generating
equipment need only be rated to match a suitable range of mean incident power. The primary
transmission is not practically limited in the level of instantaneous power it can capture and
store.
3. A numerical model of the Pelamis hydraulic PTO has been created using conditional
functions to simulate valve operation, thus requiring less computational effort than a
conventional dynamic flow model using differential equations to model each element. The
numerical models were verified using experimental data from the test rigs at 1/7th and full-
scale.
4. Experiments have shown that performance of the PTO is as expected with combined
efficiency of the primary transmission in excess of 80% at full-scale over a representative
range of operating conditions. This efficiency can be improved upon through reducing flow
losses by design.
 R. Henderson / Renewable Energy 31 (2006) 271–283 283

Acknowledgements

The work outlined in this paper forms part of an industrial programme of development carried
out by Ocean Power Delivery Ltd. Edinburgh Designs Ltd created hardware and software
infrastructure for this programme. Projects were supported by the UK Department of Trade and
Industry.

References

[1] Falnes J. Principles for capture of energy from Ocean waves. Phase control and optimum oscillation, Technical
Report, Institutt for fysikk, Norway; 1995.
[2] Salter S. Power conversion systems for ducks. Proc. International conference on future energy concepts, vol. 171.
London: IEEE Pub.; 1979. pp.100–108.
[3] Salter S, Lin C. The sloped IPS wave energy converter. Proc. Second European Wave Energy Conference, Lisbon
Portugal, November 1995.
[4] Blake TE, Chaplin RV. The P.S.FROG: Latest developments and model testing, Proc. Third European Wave
Energy Conference, Vol. 1, Patras Greece, September 1998.
[5] Heath T, Whittaker TJT, Boake CB. The design, construction and operation of the LIMPET wave energy converter.
Proc. 4th European Wave Energy Conference, Alborg Denmark, 2000.
[6] Falcao A.F. de O. The shoreline OWC wave power plant at the Azores. Proc. 4th European Wave Energy
Conference, Alborg Denmark, 2000.
[7] Salter S, Rampen WHS. 11. The wedding cake multi-eccentric radial piston hydraulic machine wih direct computer
control of displacement applied to wave energy devices, 1993 European Wave Energy Symposium (ISBN
0-903640-84-8), Edinburgh Scotland.
[8] http:/www.artemisip.com.
[9] Eidsmoen H. Tight-moored amplitude-limited heaving-buoy wave-energy converter with phase control. Appl
Ocean Res 1998;20.
[10] Budal K, Falnes J. Optimum operation of improved wave power converter. Mar Sci Commun 1977;3.
[11] French M. A generalised view of resonant energy transfer. J Mech Eng Sci 1979;21.
[12] http://www.oceanpd.com.
[13] Yemm R, Henderson R, Taylor C. The OPD pelamis WEC: current status and onward programme. Proc. 4th
European Wave Energy Conference, Alborg Denmark, 2000.
[14] Pizer D, Yemm R, Retzler C. The OPD Pelamis: Experimental and numerical results from the hydrodynamic work
program. Proc. 4th European Wave Energy Conference, Alborg Denmark, 2000.
[15] Retzler C, Pizer D, Henderson R, Ahlqvist J, Cowieson F, Shaw M. PELAMIS: Advances in the numerical and
experimental modelling programme. Proc. 5th European Wave Energy conference, Cork, Ireland, 2003.