Renewable Energy 36 (2011) 3124e3132

Contents lists available at ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Selection of design power of wave energy converters based on wave basin

experiments
Luca Martinelli a, b, *, Barbara Zanuttigh b, Jens Peter Kofoed c
a

University of Padova, IMAGE, Via Ognissanti 39, 35129 Padova, Italy

University of Bologna, DICAM, Viale Risorgimento 2, 40136 Bologna, Italy
c
Wave Energy Research Group, Department of Civil Engineering, Aalborg University, Sohngaardsholmsvej 57, 9000 Aalborg, Denmark
b

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 16 July 2010
Accepted 10 March 2011
Available online 29 April 2011

Aim of this paper is to develop a method for selecting the optimal power generation capacity for which
a wave energy converter (WEC) should be rated. This method is suitable for the earliest stages of
development, when several studies are missing, including design of the Power Take Off (PTO) system,
and the rst economic considerations become essential for investment opportunities. It relies on the
availability of an experimental description of the maximum possible produced power under realistic
conditions, typically obtained by dummy PTOs. It consists of three steps: statistical characterisation of
the measured efciency; description of the energy production by means of a function of the design
capacity; application of a simple formula for cost benet analysis. The analyses here proposed are based
on the experimental results of 3D tests on two oating wave energy devices, named LEANCON and DEXA.
Limitations of this method essentially consist in the presence of scale effects related to the laboratory
investigations, where mechanical, aerodynamic, electrical losses are not accurately represented.
2011 Elsevier Ltd. All rights reserved.

Keywords:
Design
Statistics
Cost benet
Wave power
Capacity factor

1. Introduction
Recently, many concepts for wave energy converters (WECs)
have been developed or signicantly optimised (for instance PICO
[1], Pelamis [2]; WaveGen [3]; Ocean Power Technologies [4]; Wave
Dragon [5]; SeaBeavI [6]). This proliferation of new devices to be
tested (together with the main interest focused on the demonstration of improved efciency) is urging for the denition of an
efcient and common testing procedure.
Laboratory tests carried out at the initial stage of the design
usually provide a deterministic efciency or expected energy
production of the device, even if the irregular and random nature of
the wave attack is correctly simulated. Results may be based on
a more realistic power production expectation. The purpose of this
note is to show the importance of a stochastic description of the
experimental device efciency, in order to dene a more accurate
assessment of the power production.
It is well known that, after an initial experimental investigation,
the steps of the design process include many expensive and specic
studies, involving numerical simulations and large scale testing.
* Corresponding author. University of Padova, IMAGE, Via Ognissanti 39, 35129
Padova, Italy.
E-mail addresses: luca.martinelli@unipd.it (L. Martinelli), barbara.zanuttigh@
unibo.it (B. Zanuttigh), jpk@civil.aau.dk (J.P. Kofoed).
0960-1481/$ e see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.renene.2011.03.021

These studies aim at the development of the i) Power Take Off

(PTO) system, ii) the device geometry and eif appropriate- iii) the
mooring system:
i. The PTO, that converts an irregular oscillatory energy ux
into electricity, is obviously a key element in the device
economic performance ([7,8]), resulting in signicant barrier
to commercialising. Falco [9] rst provided a statistical
representation of the power production suited to a shore
xed Oscillating Water Column (OWC) device. He proposed
a method for random waves, by essentially deriving the
variance and the probability distribution of the PICO plant
through the transfer function by Evans [10], and by assuming
a linear relation between turbine ow rate and pressure head.
ii. the device geometry may be improved to benet from the
possible presence of the natural resonance which has been
theoretically and experimentally proved to be a very efcient
mechanism ([11,12]);
iii. the mooring design, and in particular the weathervaning
scheme under oblique waves, largely affects the oating body
movements ([13,14]) and presumably the energy production.
Economics of the wave energy sources are discussed by several
authors, among others by [15e18]. Optimisation is generally based
on availability of advanced studies of the device. In practice, an

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

3125

accurate selection of power capacity is needed at earlier stages, i.e.

just after proof of concept experiments.
In these experiments, i) the PTO response is only schematically
represented; ii) the device geometry is generally dened by the
inventor on the basis of intuitions and iii) the mooring system is
seldom realistic. Nevertheless in most cases the statistic nature of
the waves is respected, since its importance is considered essential.
This paper examines the stochastic nature of the power
performance based on existing 3D tests of two oating WECs,
specically LEANCON e an OWC device e and DEXA e a Wave
Activated Body (WAB) device.
Specic objectives of the paper are:

to identify the power generation capacity for which the device
should be designed. Such design value is necessarily greater
than the average power and lower than the highest possible
peak, certainly infrequent and therefore contributing very little
to the overall yearly power production;

to provide a preliminary tool to analyse different strategies as
regards the exploitation of extreme waves. On the basis of the
amount of energy harvested in these conditions, it can be
decided if it is worthwhile to keep the device running or to
switch it off, putting it in safe mode.
The paper is organized as follows. Section 2 describes the tests,
the WECs and their hydrodynamic functioning concept. The results
of power production and device efciency are summarised in
Section 3. Section 4 presents the statistical approach. A cost benet
analysis is then performed in Section 5 to provide basic criteria for
selecting the capacity factor. Finally conclusions are drawn in
Section 6.

2. The experimental dataset

2.1 The facility and tested wave attacks
3D hydrodynamic tests were performed in the directional wave
basin of the Hydraulics and Costal Engineering Laboratory at Aalborg University, DK. The basin is 15.7 m long, 8.5 m wide and 1.5 m
deep. The wave generator is a piston type paddle system composed
of 10 actuators with stroke length of 0.5 m, enabling generation of
short-crested waves. The software used for controlling the paddle
system to generate waves is AwaSys developed by the same laboratory (Aalborg University). Regular and irregular short-crested
waves with peak periods up to approximately 3 s, oblique 2D and
3D waves can be generated with good results.
The active absorption on the wave paddles was not used, but
passive absorption was placed at the rear end of the basin and at
both sides. The absorbing sidewalls were made of crates
(1.21 1.21 m, 0.70 m deep). The 1:4 sloping beach placed opposite
to the wave maker was made of concrete and gravel with
D50 5 cm.
Tested wave attacks are reported at prototype scale in Table 1
and reproduce the typical annual wave climate of the North Sea

Fig. 1. Picture showing the OWC device.

([19]). For each wave attack a 3D Jonswap spectrum was adopted

with directional spreading equal to 22.7.
Measurements were carried out with logging frequency of 25 Hz
and test duration was approximately of 30 min (corresponding to
more than 1000 waves in most cases) for each irregular wave state.
2.2. The OWC device: characteristics and measurements
The tested OWC is the LEANCON WEC [20], a multi-chamber
oating device working in near-shore and off-shore conditions
(details in [19]).
The model is formed by a oating V-shaped slender structure,
with two arms oblique 40 with respect to the incident wave front
(Fig. 1). The reference full scale device is 240 m wide, intended for
deployment in the Danish part of the North Sea. The model of the
device is realised in berglass, and it is 6 m wide (scale 1:40). In all
tests, water depth was kept constant at 0.73 m.
The oating beams are equipped, in the bottom submerged half,
with two rows of cylindrical chambers (each may be considered to
be an OWC). Fig. 2 shows a side view. In the model each chamber is
channelled via exible hoses toward two larger ducts, lying parallel
to the arm. One duct is kept at high pressure, the other one at low
pressure, separated by a dummy turbine (an articial pressure
drop). In full scale these ducts and hoses are integrated into the
structure. As the wave travels across the structure, the air present in
the chambers over the crest is pressurized and it opens the relative
non-return valves. The air is therefore forced to enter into the high
pressure duct and from here to the turbine. After the generator,
the air, at a pressure below the atmospheric one, is gathered at the
second duct and from here it is sucked out of the system by the
pipes placed at the wave troughs, again owing past non-return
valves. The structure beam is therefore divided into high pressure
zones, connected to chambers that redirect the air toward the main
inlet duct and from here toward the dummy turbine, and low

Table 1
Wave states representative of the North Sea climate, prototype scale (1:1).
Wave state

Hs [m]

Tz [s]

Tp [s]

Energy ux [kW/m]

Prob. occur. [%]

1
2
3
4
5

1.0
2.0
3.0
4.0
5.0

4.0
5.0
6.0
7.0
8.0

5.6
7.0
8.4
9.8
11.2

2.1
11.6
32.0
65.6
114.0

46.8
22.6
10.8
5.1
2.4

Fig. 2. Side view of the arm of the OWC device.

3126

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

Fig. 3. Picture showing the OWC device moored with a spring board and the position of the wave gauges.

pressure aspiration zones, where the chambers suck air from the
outow duct. In an ideal condition, two rows of 60 pipes blow air
into the pressurized inlet duct, and two rows of 60 pipes suck the
air from the outow duct (Fig. 2). During every wave period, each
pipe completes a cycle and blows air when the waves rise and sucks
air when the wave falls.
The oating model is anchored in a single point, at the tip of the
V by means of a spring board (Fig. 3).
A parameter which is believed to strongly inuence the device
efciency is the draft. Tests considered here are those described in
Ref. [19] in oating conguration with what is believed the most
realistic draft, on the basis of previous testing.
Resistive wave gauges were used to measure surface elevation:
two arrays of 4 gauges were used, and placed in front of one of the
arms. One array was positioned in the usual way, along the wave
direction, aiming at the measure of incident wave condition and
reection. The other array was rotated in order to be perpendicular
to the wave direction (and to the wave maker) in order to assess the
decay of the wave along the arm.
In order to (indirectly) measure the absorbed power in time,
the pressure and the ow were monitored. More specically, the
pressure was measured in the inlet pe(t) and outlet pi(t) ducts, just
before the pressure drop, whereas the air velocity v(t) was
measured only in the latter. The measured produced power P(t) in
time is obtained as: P(t) (pe(t)  pi(t))$v(t) and is given in W.
Pressure measurements were carried out by means of extremely
sensitive cells (/ 10 Pa), covering the range 0e500 Pa. Velocity

was acquired by means of a thermal ow meter working in the

range (0e10 m/s), calibrated in the section of duct under exam. The
ow meter appeared to have a much lower accuracy at low speed
(10e15%), and therefore the low values of measured wave power
production are rather uncertain.
The load on the PTO is changed by varying the pressure drop in
the dummy turbine.
2.3. The WAB device: characteristics and measurements
The DEXA device [21], of the Wave Activated Body (WAB) type,
consists of two rigid pontoons, hinged together in the center so that
each pontoon is allowed to pivot in relation to the other. A power
take off system, based on Aquagear (i.e. a low pressure power
transmission technology, based on water) is placed in between.
The model realised in the laboratory in 1:20 scale is 2.10 m long
(cross-shore direction), 0.81 m wide (long-shore direction), and has
a global weight of 22 kg. Above the devices the measuring system
and two additional weights (sand bags) were placed to reach the
required draft. View of the model is given in Fig. 4. In all tests, water
depth was kept constant at 0.65 m.
The PTO system consists of a metal bar with an elongated hole,
a wire welded at the two ends of the hole and a small electric
engine with a wheel (Fig. 5). The bar is connected to one half of the
device through the wheel and to the other half through a load cell
(strain gauge equipped bone, 10 mm thick). The wire is coiled
around the wheel that is forced to rotate while translating along the

Fig. 4. View of the DEXA device.

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

3127

Fig. 5. Power take off for the DEXA device and measuring system of pontoon displacements (by means of the bone) and of device horizontal displacements (by means of the
ultrasonic sensor).

Results here considered refer to wave attacks in Table 1. The

devices were tested by adopting the best load conditions, i.e. the
rigidity of the PTO that was found to optimize power production
during preliminary tests performed with regular waves.
Fig. 6 shows the curves giving the probability distribution of the
measured power for each wave state. The order of magnitude of
the generated power is several Watts, which is something quite
appreciable and the concept capability of harvesting energy is thus
proved behind doubts. For each sea state, it can be observed e as
expected e that:

the exceeding probability decreases as the measured power
increases;

as the severity of the sea states increases, the average measured
power increases.
Fig. 7 compares the efciency measured by the different sea
states. The efciency given in Fig. 7 is obtained as the ratio between
the measured and available power: h <PM>/<PW>, where <PM>
is the measured average power and <PW> is the incident energy
ux per unit length multiplied by the structure width W (wings
aperture for LEANCON, long-shore width for DEXA). The group
celerity is evaluated on the basis of Tm1, i.e. the spectral period
obtained by rst and second moments.
The device is meant to be placed at a depth of the order of half
the wave length, and therefore close to off-shore conditions.
This simple way to evaluate h does not allow a fair comparison
between devices characterised by different width, giving for
instance the possibly false image that DEXA is more efcient than
LEANCON in Fig. 7. Other descriptions of the efciency are possible,
with marginal inuence for the following analysis.

1
0.9

LEANCON
Exceedance probability

3. Analysis of power production

In Fig. 7, a progressive reduction of efciency is seen as the sea

states become more energetic.
For the LEANCON, this is partially related to the chosen
dimensions of the chambers and partially to the dynamic behaviour

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-2
10

-1

10

10

10

10

Measured power (W)

1
0.9

Exceedance probability

bar hole. The load on the PTO is modied by varying the current in
the engine and therefore the resistance of the wheel rotation, so
that the body rigidity is changed.
The calibrated instrumentation for power measurement includes
a non-contact ultrasonic displacement sensor with a plate for the
signal reection.
An array of 3 resistive wave gauges was placed between the
wave maker and the device along the wave direction, aiming at the
measurement of incident and reected waves.
Tests considered here are those described in [22] with spring
mooring.

DEXA

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
-4
10

-2

10

10

10

10

Measured power (W)

Fig. 6. Measured power production for each wave state (from left to right from wave 1
to wave 5) versus its exceedance probability, top: LEANCON, bottom: DEXA, laboratory
scale.

3128

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

Fig. 7. Measured efciency of the power conversion for each wave state.

of the oating structure, that tends to ride on the waves for long
wave periods. For the DEXA, the effect is due to the designed length,
which is tuned to have maximum efciency with the second wave
state.

4. Statistical analysis
The key of the method lies on the availability of the experimental statistical distribution of the produced power. Unfortunately in most cases only the average extracted energy is available.
For the tests described above, we could indirectly measure the
time history of the produced power either by multiplying
the pressure drop and the air ow rate (LEANCON) or by multiplying the forces and the velocities (DEXA). The exceeding probability of the produced power was therefore derived from the time
history. After a proper interpolation on sufciently ne and regular
steps, the experimental occurrence was also obtained.
The results were up-scaled to prototype based on the assumption that efciency is scale-independent, and consequently power
scales with l3.5, l being the length scale.
Fig. 8 shows the occurrence probability of the produced power
at prototype scale for the two analysed devices. The shape of the
two curves is rather different. In both cases it is appreciable
a threshold value of order 0.02e0.04 W at model scale due to a limit
in the resolution of the measurement system. For the DEXA, which
is a typical device responding on a wave by wave basis, as expected
the highest occurrence probability corresponds to the lower
threshold. For the LEANCON, that is a multi-chamber device that
aims at producing a steady energy ux, the highest occurrence
probability corresponds to a signicantly high value (105 W).
The expected produced energy per year is simply obtained by
the integral of power in time, or similarly by the product of power
and occurrence (expressed in seconds/year). Fig. 9 shows how the
expected produced energy is distributed for each level of power
(ranging from zero to innite).
In practice, the PTO has an upper threshold in the maximum
amount of harvestable power. Usually the cost of the PTO is
proportional to such limit, which represents one of the main design
parameters, and will be termed rated or design power Pd. It is of
great interest to correlate the Pd of the PTO to the actual amount of
produced energy.
The Capacity Factor (CF) is the ratio between the (annual)
average extracted energy by a PTO system with given rated power,
and the rated power itself, both expressed in W:

CF hEM i=Pd

(1)

Typical values of CF in the wind industry are in the range

30e50%.

Fig. 8. Measured annual power distribution up-scaled at prototype conditions North

Sea climate, top: LEANCON, bottom: DEXA. Note the threshold of the measurement
system that at model scale is 0.035 W for the LEANCON and 0.025 W for the DEXA.

In order to predict the amount of energy that can be produced

with a limited Pd, it is necessary to make an assumption on the PTO
behaviour when the available power exceeds the rated one. Two
simple hypotheses can be considered:

hypothesis 1, no energy is produced when the available power
exceeds Pd,

hypothesis 2, energy production when the available power
exceeds Pd is equal to Pd multiplied by the exceeding duration.
Fig. 10 shows in continuous black line the loss of energy which
corresponds to the rst hypothesis, i.e. for a given upper threshold
Pd 0.5 MW, the energy that would not be converted is A 3 TJ,
and in dashed grey line the energy regained according to the
second hypothesis, i.e. for the same threshold Pd the energy that
would be converted is B 2.2 TJ.
Fig. 11 shows the energy loss versus Pd for the DEXA and the
LEANCON under both hypothesis 1 and 2. Even if the devices are

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

3129

12

x 10

8
7

Energy (J)

6
5
4
3
2
1
0

10

Design Power of PTO (W)

15
5

x 10

Fig. 10. Energy loss (solid black line) in case no energy is produced above the PTO
design value Pd, and energy regained (grey dashed line) for the same Pd in case the
energy production is equal to the upper limit Pd multiplied by the exceeding duration.

The economic consequences of two simple extreme strategies

are considered, which correspond to hypothesis 1 and 2 described
in Section 4:
1) the system is switched off, i.e put in safe mode, after
a threshold coincident to the PTO design power Pd is exceeded;
2) the PTO absorbs a power equal to Pd (i.e. the maximum
possible) when the available power is greater than Pd.

Fig. 9. Distribution of the measured annual wave energy versus power, data up-scaled
at prototype conditions, North Sea climate, top: LEANCON, bottom: DEXA.

different, the two curves are very similar and show a logarithmic
trend. It is therefore possible that this result has a general validity.
The relevance of this novel result is highlighted in the next Section,
which will compare the costs of the whole device (usually the PTO
cost increases linearly with Pd) with the loss of benet (energy
decreases logarithmically).
Fig. 12 shows the energy loss in terms of overall efciency when
the rated power is limited and equal to Pd, non-dimensionalised by
the maximum overall efciency (i.e. for an innite Pd), for DEXA and
LEANCON. The overall efciency is obviously monotonically
increasing, the hypothesis 2 is more effective. Fig. 12 also shows the
capacity factor CF as dened by Eq. (1).

5. Cost benet analysis

Selection of the design power Pd of the PTO has important
economic effects: if Pd is too small, the cost is low but the WEC will
not produce sufcient energy to provide a benet. On the other
hand, if the PTO is designed for extreme wave conditions the WEC
will always produce energy but the total cost may be excessive.

The selection is based on the statistical distribution of the

measured extracted power and on simple considerations in terms
of yield, i.e. benet-cost.
In general, the yield should include all the different aspects of
interest: direct and indirect costs, maintenance, failure risk, costs
for dismissal, upgrade, duration, etc. on the one hand and direct
and indirect benets, including employment opportunities, advertisement, increase of knowledge, etc. on the other.
In this investigation, the following simplied assumptions are
considered:

the cost of the device is proportional (with proportionality
constant a) to Pd and the other costs (Co) are independent from
the actual choice of the PTO:

C aPd Co

(2)

the benet B is proportional (b) to the annual expected

produced energy E:

B bEPd

(3)

Since this investigation cannot enter into the details of the real
values of a and b, the study will assume values of a, b, and Co
derived from private communications of inventors and available
literature [7]. The designer experience should allow a more accurate case-specic evaluation.
Further assumptions are that:

the ratio a/b is such that leads to a null yield when Pd is equal to
a reference value P*. In practice, the inventor knows by deterministic computations the value P* that will give some sort of

3130

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

12

x 10

LEANCON

Energy loss (J)

7
6
5
4
3
2
1
0 0
10

10

10

10

10

Design Power of PTO (W)

12

4.5

x 10

DEXA

Energy loss (J)

3.5
3
2.5
2
1.5
1
0.5
0
0
10

10

10

10

10

Design Power of PTO (W)

Fig. 11. Expected energy loss (A) under hypothesis 1, solid black line, and expected
energy loss (A  B) under hypothesis 2 versus design power of PTO Pd, data up-scaled
at prototype conditions, North Sea climate, top: LEANCON, bottom: DEXA.

small but positive yield (P* is easier to assess than a and b). In
the following, an additional criterion will be used to choose P*.

The other costs Co are equal to x times the cost of the PTO
designed for such power P*, with x 1 for simplicity, although
may be a factor 3 may be more appropriate.
The expression for evaluating the yield Y as function of P*
therefore reads:

YP*

bE
bE
BP*
1
1 0
1
*
CP*
a1 xP
2aP *
P*

P*

(4)

and the ratio between the proportionality coefcients is

b
2P *
2P *

 *
a E P*
E

(5)

The approach here described is generic, since it can be applied also

when a deeper knowledge of the costs is available, leading to
a more precise quantication of Co and b/a.

Fig. 12. Reduction of overall efciency (compared to maximum) versus rated power of
the PTO, and Capacity Factor. Data up-scaled at prototype conditions, top: LEANCON,
bottom: DEXA.

The yield for a generic choice of Pd becomes:

YPd

bEPd
BPd
2P * EPd


1
1
1

CPd
*
E* P P *
a Pd P
d

(6)

where for the rst and second hypothesis, the produced energy E is
given by

EPd Emax  APd Hypothesis 1

(7a)

EPd Emax  APd BPd Hypothesis 2

(7b)

The selection of Pd is based on the condition that the yield is

maximum.
For our case it is also cautiously assumed that under hypothesis 1
the maximum yield is null (at rst stage the maximum yield will not
be very high anyway). Since for null yield we also require that Pd P*
(see requirements above), this requirement poses a condition on P*,
i.e. P* corresponds to the maximum yield for hypothesis 1.
With this academic simplication, we require that the costs are
not so high that a decit is expected for any choice of the rated PTO

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132

Table 2
Iterations used to dene P* [W].
P* [W]

LEANCON

Tentative 1
Tentative 2
Tentative 3
Tentative 4
Tentative 5
Final choice

1.0  104
1.0  105
5.0  105
1.0  106
1.3  106
1.3  106

DEXA
1.0  107
5.0  107
5.0  106
1.0  106
1.3  106
1.3  106

1.0  104
1.0  105
2.1  105
2.6  105
2.7  105
2.7  105

1.0  107
5.5  105
3.0  105
2.7  105
2.7  105

power Pd, and there is still the possibility to obtain a positive yield
by using hypothesis 2 (provided that the optimised Pd in hypothesis
2 is lower than P* Pd in hypothesis 1, and that there is no need to
revise the other costs Co).
The procedure used to evaluate P* is merely iterative. A rst
guess for P* is selected and inserted in Eq. (7a). The maximum of Eq.
(7a) is then used as the next tentative value of P*. Table 2 shows our
iterations for LEANCON and DEXA. In order to be sure that the
procedure is convergent, the procedure is repeated twice, starting
with the extreme low and extreme high values of P*.
Fig. 13 shows the yield as function of the power used for PTO
design under hypothesis 1 and 2. It can be observed that
a maximum is found when P* equals 1.3 and 0.27 MW respectively
for LEANCON and DEXA.

The optimal value of P* is much lower than the maximum

possible. As expected, the hypothesis 2 gives a lower optimal value
than hypothesis 1 since part of the energy loss is re-gained.
In order to analyse the effects of a greater initial cost Co or of
different ratios a/b on the choice of Pd, an analytical expression is
searched.
The maximum yield is given by

0
1
v
2P * @ EPd
vEPd
1


A
YPd 0 *

2 vP
vPd
E
d
Pd P *
Pd P *

EPd
Pd

P*

(9)

12

10

x 10

Hypothesis 1
Hypothesis 2

8
Hypothesis 1
Hypothesis 2

Produced energy (J)

Yield

vEPd
vPd

In the range of interest, the curves (7a) and (7b) can be approximated by logarithmic functions (i.e. we assume that the Pd value
occurs in the logarithmic part of the A and B curves):

0.4

(8)

from which one derives

0.2

3131

LEANCON

-0.2
-0.4

LEANCON

7
6
5
4
3
2

-0.6

1
-0.8
-1
2
10

0
4
10
3

10

10

10

10

0.4

DEXA

Yield

x 10

Produced energy (J)

0.2

-0.2

-0.4
-0.6

10

10

Design Power of PTO (W)

10

Design Power of PTO (W)

Hypothesis 1
Hypothesis 2

10

11

Hypothesis 1
Hypothesis 2

DEXA

6
5
4
3
2

-0.8

-1
2
10

10

10

10

10

10

Design Power of PTO (W)

Fig. 13. Yield of the project as function of the power used for PTO design. Top:
LEANCON, bottom: DEXA.

0
3
10

10

10

10

10

Design Power of PTO (W)

Fig. 14. Produced energy as function of the power used for PTO design. Top: LEANCON,
bottom: DEXA.

3132

L. Martinelli et al. / Renewable Energy 36 (2011) 3124e3132


EPd Eo Ls $lnPd E P * Ls lnPd  ln P *

(10)

where Eo and Ls can be dened by the slopes of the A or B curves

and Eq. (7a) or (7b).
By solving Eq. (9) the following trascendental equation that
gives the optimal design power is obtained:

Eo Ls lnPd
L


 s
P
*
d
Pd P

Acknowledgements
The support of the Danish Council for Strategic Research
through the Structural Design of Wave Energy Devices project
(SDWED, www.sdwed.civil.aau.dk) is gratefully acknowledged.
Notations

(11)

Fig. 14 shows the tting of Eo and Ls in Eq. (10) to Eq. (7b), for the
LEANCON and DEXA devices. The tting for LEANCON gives E(P*)
8.23  1012 [J], and Ls 2.46  1012 [J/ln(W)] (so that Eo 2.62
 1013 [J]) and can be based on just two points of Eq. (7b). The
application of Eq. (11) determines Pd 6.7  105 (W), in perfect
agreement to the results of Fig. 13. For DEXA, E(P*) 8.89  1011 [J]
and Ls 2.75  1011 [J/ln(W)] so that Eo 2.55  1012 [J]). These
data are sufcient to dene through Eq. (11) the optimal design
power, which is Pd 1.6  105 (W).
It is of course expected that the two kinds of experiments are
characterised by different generation capacities after scaling with
different factors (1:40 and 1:20).
6. Conclusions
This paper addresses the problem of dening the most advantageous value of the power generation capacity for which a PTO
should be designed, and proposed a method based on small scale
experiments.
Two oating devices named LEANCON and DEXA were
considered. They were tested against a wave climate representative of the North Sea, where the devices might be
installed. The produced power was evaluated by a combination
of measurements of pressure drop and ow rate or velocity and
force.
The initial analysis dened the average frequency distribution of
the extracted energy, and of the associated power. During this
analysis, there was no limitation on the possibility to convert the
available energy (i.e. even the highest wave contributed to the
power production).
Subsequently, a statistical approach was proposed to evaluate the amount of energy that would be extracted by a PTO
with an upper limit in the generation capacity. When the
available power exceeds such limit, two operating hypotheses
were examined: the PTO was bypassed or produced its possible
maximum.
The dependence of the energy loss on the power generation
capacity suggests, at least for the DEXA case, that an optimal design
value is probably far from the measured maximum, i.e. the power
exceeded once per year. In order to dene the optimal possible
design value, the obtained energy curves may be used in the
framework of a cost benet analysis.
A brief example of cost benet analysis is performed providing
for both devices an optimal design value of order 600 kW and
150 kW for LEANCON and DEXA. In the absence of better assessments, this remains an important starting point for future design
considerations, that have also to account for many other components, such as the presence of a real mooring system, the effects
due to multidimensional waves, detailed denition of the PTO
system and of its latching strategy.
Limitations of this method essentially consist in the presence of
all scale effects related to the measured energy conversion in the
laboratory. Anyway this same approach could be used also at full
scale, on the basis of prototype measurements.

d
g
Hs
Lo
p
P
PW
so
Tm1
Tp
v
t

OWC draft
gravity acceleration
signicant wave height
wave length based on Tm-1
pressure
power produced by the WEC
wave power
wave steepness based on Tm1
spectral wave period
peak wave period
air velocity
time

References
[1] Falco AF de O. Design and construction of the Pico OWC wave power plant.
In: Dursthoff W, Diamantaras K, editors. Proc.of the third European wave
energy conference; 1998.
[2] Pelamis Wave Power. http://www.pelamiswave.com.
[3] WaveGen. WaveGen, http://www/wavegen.co.uk.
[4] Ocean Power Technologies. http://www.oceanpowertechnologies.com.
[5] Kofoed JP, Frigaard P, Friis-Madsen E, Sorensen HC. Prototype testing of the
wave energy converter wave dragon. Renewable Energy 2008;31(2):181e9.
[6] Elwood D, Solomon C, Prudell J, Stillinger C, von Jouanne A, Brekken T, et al.
Design, construction, and ocean testing of a taut-moored dual-body wave
energy converter with a linear generator power take-off. Renewable Energy
2010;35:348e54.
[7] Masuda Y, Kuboki T, Lewis T, Liang X, Sun P. Prospect of economical wave
power electric generator by the terminator backward bent duct buoy (BBDB).
In: Proc.of the twelfth international offshore and polar engineering conference, Kitakyushu, Japan, May 26e31; 2002. p. 607e13.
[8] Henderson R. Design, simulation, and testing of a novel hydraulic power takeoff system for the Pelamis wave energy converter. Renewable Energy 2006;
31(2):271e83.
[9] Falco AFde O. Control of an oscillating water column wave power plant for
maximum energy conversion. Applied Ocean Research 2002;24(2):73e82.
[10] Evans DV. Wave power absorption by systems of oscillating surface-pressure
distributions. J Fluid Mech 1982;114:481e99.
[11] Boccotti P. On a new wave energy absorber. Ocean Engineering 2003;30:
1191e200.
[12] Arena F, Filianoti P. Small-scale eld experiment on a submerged breakwater
for absorbing wave energy. J. Waterway Port Coastal and Ocean Engineering
2007;133:161e7.
[13] Johanning L, Smith G, Wolfram J. Measurements of static and dynamic
mooring line damping and their importance for oating WEC devices. Ocean
Engineering 2007;34(14e15):1918e34.
[14] Martinelli L, Ruol P, Zanuttigh B. Wave basin experiments on oating breakwaters with different layouts. Applied Ocean Research July 2008;30(3):
199e207.
[15] Petroncini S, Yemm RW4th European Conference on Wave Energy,
Denmark. Introducing wave energy into renewable marketplace; 2000:
33e41.
[16] Mueller MA, Baker NJ, Spooner E4th European Conference on Wave Energy,
Denmark. Electrical aspects of direct drive wave energy converters;
2000:235e42.
[17] Leijon M, Bernhoff H, Berg M, A gren O. Economical considerations of
renewable electric energy productiondespecially development of wave
energy. Renewable Energy 2003;28:1201e9.
[18] Falco AF de O. Stochastic modelling in wave power-equipment optimization:
maximum energy production versus maximum prot. Ocean Eng 2004;31.
[19] Kofoed, J.P. and P. Frigaard. Hydraulic evaluation of the LEANCON wave
energy converter. DCE Technical Report No. 45. ISSN1901-726X. Dep. of Civil
Eng., Aalborg University, Oct. 2008.
[20] Leancon Wave Energy. http://www.leancon.com/.
[21] Dexa Wave Energy. http://www.dexawaveenergy.co.uk/.
[22] Kofoed JP. Hydraulic evaluation of the DEXA wave energy converter. DCE
Technical Report No. 57. Dep. of Civil Eng., Aalborg University; 2009. 23pp.