You are on page 1of 17

International Journal of Marine Energy 6 (2014) 1834

Contents lists available at ScienceDirect

International Journal of Marine Energy


journal homepage: www.elsevier.com/locate/ijome

Model-prototype similarity
of oscillating-water-column wave
energy converters
Antnio F.O. Falco , Joo C.C. Henriques
IDMEC, LAETA, Instituto Superior Tcnico, Universidade de Lisboa, 1049-001 Lisbon, Portugal

a r t i c l e

i n f o

Article history:
Received 4 April 2014
Revised 8 May 2014
Accepted 8 May 2014
Available online 27 May 2014
Keywords:
Wave energy
Oscillating-water-column
Dimensional analysis
Model testing
Air turbines

a b s t r a c t
Model testing in wave tanks or under sheltered sea conditions is an
essential step in the development of wave energy converters. The
paper focuses on the rules for geometric, hydrodynamic, thermodynamic and aerodynamic similarity in model testing of wave energy
converters of oscillating-water-column (OWC) type, with emphasis
on air compressibility effects in the air chamber and on air turbine
aerodynamics. It is shown that the correct volume scale ratio for the
air chamber is far from identical to the volume scale ratio for the
submerged part of the converter, and should take into account
the thermodynamics of the compressible ow through the air turbine or through the turbine simulator (orice or other). For those
cases when the model is large enough to be tted with a scaled
air turbine, dimensional analysis is applied to obtain ratios for turbine size and rotational speed, and also to establish relationships
between rotational speed control algorithms. A numerical example
is presented to illustrate the importance of appropriately simulating the air compressibility effects when testing at model scale.
2014 Elsevier Ltd. All rights reserved.

1. Introduction
The theoretical modelling based on linear water wave theory is an essential step in the development of wave energy converters. It provides insights and important information at relatively low
Corresponding author. Tel.: +351 919190017; fax: +351 218417398.
E-mail address: antonio.falcao@ist.utl.pt (A.F.O. Falco).
http://dx.doi.org/10.1016/j.ijome.2014.05.002
2214-1669/ 2014 Elsevier Ltd. All rights reserved.

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

Nomenclature
Latin letters
Aw
B
C
D
f
F
Fr
g
G
Hs
I
k
1
K wq1
at q
L
m
N
p
p0
P
q qe qr
qe
qr
Q,Qe,Qr
r p=pat
Re
s
Sf
Te
U
V
V0
w
Ye
Yt
Greek letters
c cp =cv

C
d qm =qF
 Lm =LF

g
m
N
P

q
r
U

v
W
X

incident wave amplitude


radiation susceptance
proportionality constant
diameter of turbine rotor
frequency
function (see Eq. (1))
Froude number
acceleration of gravity
radiation conductance
signicant wave height
rotational inertia
polytropic exponent
ow-rate/pressure-head ratio
characteristic length of device
mass of air in chamber
scale ratio (with subscript)
pressure oscillation in chamber
complex amplitude of p
power
volume ow rate displaced by OWC
excitation ow rate
radiation ow rate
complex amplitude of q,qe,qr
pressure ratio
Reynolds number
specic entropy
variance density spectrum of waves
energy period of waves
function (see Eq. (2))
air chamber volume
value of V without waves
turbine mass ow rate
electromagnetic torque on generator rotor
aerodynamic torque on turbine rotor
specic heat ratio of air
see Eq. (9)
water density ratio
length scale of device
turbine efciency
kinematic viscosity
dimensionless aerodynamic torque
dimensionless quantity
density
variance (with subscript)
dimensionless ow rate of turbine
proportionality constant (orice, porous plug)
dimensionless pressure head of turbine
rotational speed of turbine

19

20

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

Subscripts
avai
at
ch
F
imp
m
t
Wells
Superscripts

available to turbine (power)


atmospheric conditions
air chamber conditions
full-sized prototype
impulse turbine
model
turbine
Wells turbine
dimensionless value
time-average value

costs, in general in a relatively fast way. However, there are important non-linear effects that are not
accounted for by this kind of modelling, namely those associated with large amplitude waves, large
amplitude motions of the wave energy converter or oscillating-water-column, and real uid effects
due to viscosity, turbulence, and vortex shedding. Commercially available computational-uiddynamics (CFD) codes, usually based on the numerical integration of the Reynolds-averaged NavierStokes (RANS) equations, may be used to account for such effects. However, even such codes, apart
from being computationally demanding, require some experimental validation. Therefore physical
model testing in a wave ume or wave tank is normally the next step. The scales range between about
1:100th in small wave umes to about 1:10th in the largest wave tanks. Tests at larger scales (typically 1:4th scale) sometimes take place in sheltered sea locations. We may apply dimensional analysis
techniques to relate the conditions in model testing to those of the full-sized prototype in real sea
conditions.
The books on dimensional analysis by Bridgman [1], Langhaar [2] and Sedov [3] may be regarded as
classics. Most textbooks on uid mechanics include a chapter on dimensional analysis. Applications to
coastal engineering and to naval architecture can be found in [4] and [5], respectively. Dimensional
analysis in model testing of wave energy converters is addressed in the pioneer book by McCormick
[6] and more recently in [7,8]. Air turbines of oscillating-water-column converters are subject to the
same similarity laws as other types of compressible ow turbomachines, as analysed in [911].
Oscillating-water-column (OWC) devices, of xed structure or oating (Fig. 1), are an important
class of wave energy converters. A large part of wave energy converter prototypes deployed so far into
the sea are of OWC type [12,13].
In an OWC, there is a xed or oscillating hollow structure, open to the sea below the water surface,
that traps air above the inner free-surface. Wave action alternately compresses and decompresses the
trapped air which forces air to ow through a turbine coupled to a generator. Unless rectifying valves
are used, which is not practical except possibly in small devices like navigation buoys, the turbines are
self-rectifying, i.e. their rotational direction remains unchanged regardless of the direction of the air
ow. Several types of such special turbines have been developed. The axial-ow Wells turbine,
invented in the mid-1970s, is the most popular self-rectifying turbine, but other types, namely axialand radial-ow self-rectifying impulse turbines, have also been proposed, studied and used [14].
In OWC wave energy converters, the air in the chamber above the water column is subject to
oscillating pressure, and therefore its density is also time-varying according to some pressure-density
relationship. This spring-like effect, that affects the device performance, was theoretically modelled for
the rst time in [15] and shortly afterwards in [16]. In both papers, an isentropic relationship was
assumed. A more realistic model, in which variations in air entropy are related to viscous losses in
the air turbine can be found in [17]. Model testing of an OWC where for the rst time air compressibility
effects in the chamber are accounted for is reported in [18]. The representation of aero-thermodynamic
effects in small scale physical modelling of OWCs was addressed more generally in [19].
The present paper focuses on model testing of wave energy converters of OWC type. Dimensional
analysis techniques are employed to relate the results from the tests to the performance of the

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

21

Fig. 1. Different types of OWCs: (a) bottom-standing structure (Pico plant); (b) Backward Bent Duct Buoy (BBDB); (c) spar-buoy
OWC.

full-sized prototype. Froude similarity is considered in Section 2. Particular attention is devoted, in


Section 3, to compressibility effects in the air chamber, known to be important at full size but frequently ignored in model testing. Thermodynamics and turbine aerodynamic efciency are shown
to play an important role here. Rules are derived for turbine size and rotational speed scale ratios
in Section 4. Dimensional analysis is extended to turbine rotational speed control in Section 5. A
numerical example is presented in Section 6 to illustrate the importance of appropriately simulating
the air compressibility effects when testing at model scale. Conclusions are presented in Section 7. The
results presented in Section 2 on Froude similarity are well known. They are presented here for completeness and because they are needed for the derivations in the more innovative sections 3 to 5
where new results are obtained.
2. Dimensional analysis in hydrodynamics of model testing
When performing model testing in waves, it is assumed that the wetted part of the model is an
exact geometric representation of the full-scale prototype. This geometric similarity is supposed to
apply also to the bottom and surrounding walls. This condition is assumed here to be fullled. It
should be noted that, in most cases in practice, there are limitations in the test facilities that prevent
perfect geometric similarity to be attained. This is the case of the presence of walls in tanks and
umes, or the impossibility of reproducing the sea bottom bathymetry.
We assume the incident waves to be represented by a given variance density spectrum
p
(Pierson-Moskowitz or other, see e.g. [20]) Sf Hs ; T e ; f , where Hs 4 m0 is the signicant wave
height, T e m1 =m0 is the energy period, and mn is the n-th moment of Sf with respect to the frequency f. Directional wave spread is ignored here for reasons of simplicity. In the case of regular
waves, Hs and Te are simply the wave height and period.
Let L be a characteristic length (this could be a diameter in the case of an axisymmetric device). We
denote by pat the atmospheric pressure and by pat pt the pressure in the air chamber of the OWC
converter; the pressure oscillation p(t) is related to the action of the power take-off system (PTO)
which is essentially an air turbine driving an electrical generator.
Let us consider in general a dimensional quantity a as a function of n dimensional independent
quantities a Fa1 ; a2 ; :::; an ; where the function F represents a denite physical law independent
of the choice of the system of units. In the case of a wave energy converter, this may have the form

P FL; Hs ; T e ; p; g; q; m;

where P is power (possibly power absorbed from the waves), g is the acceleration of gravity (which is
assumed as a physical constant), and q and m are the density and kinematic viscosity of water,

22

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

respectively. Here we ignored the surface tension of water, which is an acceptable assumption provided that the wavelength is larger than about 0.1 m [21]. Buckinghams theorem of dimensional analysis (see e.g. [2]) allows us to replace Eq. (1) by

P1 UHs =L; Fr; Re; Pp :

Here, U is a function, Fr L1=2 g 1=2 T 1


is a Froude number, Re L2 m1 T 1
is a Reynolds number,
e
e

Pp pL1 q1 g 1 is dimensionless pressure and P1 PL7=2 q1 g 3=2 is dimensionless power. (Note

that the reciprocal of the Froude number, T e Fr1 g 1=2 L1=2 T e , may be regarded as a dimensionless
wave period.) Identical relationships could be established by replacing P1 by other dimensionless
quantities. For example, this could be dimensionless volumetric ow rate P2 qL5=2 g 1=2 ; where q
is ow rate displaced by the motion of the OWC free surface. If the four dimensionless variables Fr,
Re, Hs =L and Pp take equal values in the model and the full-sized prototype, the same will be true
for P1 ; P2 ; :::
Note that, in the case of oating devices, mooring forces may be signicant, and, if so, should be
appropriately represented in model testing [22]. However, if the mooring system is adequately
designed, the effect of mooring forces on wave energy absorption is relatively small, as found in
[23]. Such effects will be ignored here.
Let the subscripts m and F denote the model and the full-sized prototype. The length scale is
dened as e Lm =LF . The constancy of the Froude number Fr and the Reynolds number Re cannot
be satised simultaneously, since this would require mm =mF e3=2 , a condition that is obviously
unachievable in practice if e is not close to unity. In model testing, the effects due to variations in Froude number are almost always much more important than those associated with changes in Reynolds
number. Following general practice, we will keep here the constancy of Froude number, and ignore
Reynolds number as a modelling rule.
The expression of dimensionless power P1 PL7=2 q1 g 3=2 shows that the scale ratio for power is
7=2
e (if variations in water density q are neglected). In model testing of wave energy converters in the
largest wave tanks, the length scale e in general does not exceed about 1:10th. This implies a maximum power ratio of about 1:3200. In the case of an OWC wave energy converter, this scale is too small
for the turbine to be simulated adequately by a mini-turbine. The usual procedure is to simulate the
turbine by an orice, if the turbine is of impulse type (Fig. 2), or by a layer of porous material, where
the ow is approximately laminar and simulates a linear turbine like the Wells turbine; the approximation provided by this testing procedure will be analysed in Sections 3, 4 and 5. Only in tests

Fig. 2. Model at 1:16th scale of a spar-buoy OWC developed at Instituto Superior Tcnico being tested at the large wave tank of
the National Renewable Energy Centre (NAREC), Blyth, England, in 2012. The turbine was simulated by an orice. The tank was
lled with sea water.

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

23

Fig. 3. Backward Bent Duct Buoy (1:4th of full scale) equipped with a Wells turbine being tested in Galway Bay, Ireland, about
2008 (courtesy of OceanEnergy).

performed under real sea conditions at scales not less than about 1:4th, is a real turbine tted to the
model. This was done in the sheltered waters of Galway Bay, Ireland, at scale 1:4th, with a Backward
Bent Duct Buoy (BBDB) tted rst with a Wells turbine and later with an impulse turbine [24], Fig. 3.
3. Aero-thermodynamic modelling of air chamber
The volume of the air chamber of the OWC converter should be large enough to avoid ingestion of
water by the air turbine under rough sea conditions. Typical design values of the air chamber volume
divided by the area of the OWC free surface range between 3 and 8 m. An increase in this ratio is not
necessarily detrimental to the efciency of the energy conversion. Obviously, if the volume increases
to very large values, the amplitude of the air pressure oscillations becomes very small, and the capability of the device to absorb wave energy vanishes. The spring-like effect of air compressibility in the
chamber increases with chamber volume, and is important in a full-sized OWC converter.
The aerodynamic and thermodynamic processes that take place in the air chamber and turbine of
an OWC converter are quite complex. It seems reasonable to assume that they are approximately adiabatic. Indeed, the temperature oscillations in the air chamber are relatively small and their time
scales (a few seconds) are too short for signicant heat exchanges to occur across the chamber walls
and across the air-water interface, in comparison with the energy ux in the turbine [17]. It should be
recalled that the compressible ow through turbomachines is usually modelled as adiabatic, even in
gas turbines where large temperature differences take place (see e.g. [11]).
However, even if the process is assumed as adiabatic, signicant changes in specic entropy occur
in the ow through the turbine, due to viscous losses. Such changes can be related to turbine efciency. We recall that the pressure in the chamber is pat p, where pat is the atmospheric pressure.
During inhalation, it is p < 0, and air with specic entropy s > sat is admitted to the chamber, where
a highly turbulent mixing process takes place. During exhalation, p > 0, air leaves the chamber
through the turbine in a process at approximately constant specic entropy for the air remaining in
the chamber. The inhalation and exhalation processes in an OWC were studied in some detail in [17].
We assume air as a perfect uid, with specic heat ratio c cp =cv 1:4. We may write, for the
pressure-density relationship of the compressible ow of a perfect gas through a turbine [11],

p2

qk2

p1

qk1

24

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

where subscripts 1 and 2 refer to conditions at turbine entrance and exit, respectively. The polytropic
exponent k is related to the turbine polytropic efciency (or small-stage efciency) g by

1

1  c1
c g

In turn, the turbine polytropic efciency g is related to the turbine (total-to-total) efciency g by (see
[11])
 c1=c

1  p2 =p1 g

c1=c

1  p2 =p1

Here, efciency is dened as for gas turbines: g Dh=Dhs ; where Dh is drop in specic enthalpy under
real conditions and Dhs is drop in specic enthalpy under isentropic conditions, for the same inlet and
outlet pressures. In the case of inhalation we are analysing here, we replace p1 by pat and p2 by pat p,
with p < 0. Eq. (3) becomes

pat p

qkch

pat

qkat

where qch is air density in the chamber and qat is density in the atmosphere.
We write r p=pat . From Eqs (4) and (5), we can express the polytropic exponent k as a function
of the turbine efciency g and the pressure ratio r. This is represented in Fig. 4 for c 1:4. It can be
seen that the polytropic exponent k is almost invariant with the pressure ratio r p=pat , even for values of r as large as 0.6 (pressure in the chamber equal to 40% of atmospheric pressure). For g 1 (perfectly efcient turbine), the ow is isentropic and k c 1:4, as expected. For g = 0, it is k = 1. In this
case, no work is done and therefore there is no change in specic enthalpy h. Since for a perfect gas it is
dh cp dT (here T is absolute temperature), the process is isothermal (note that this is not a result from
heat exchange). The case g 0; k 1 represents a throttling process and occurs if the pressure drop
through the turbine is simulated by an orice (which is frequently used to simulate a self-rectifying
impulse turbine) or, alternately, represents the ow through a plug made of porous material (where
the ow is approximately laminar and simulates a linear turbine like the Wells turbine).
Although the dependence of the polytropic exponent k on pressure ratio r can reasonably be
neglected as shown in Fig. 4, the same cannot be said of the dependence on turbine efciency g. During the inhalation, the ow rate through the turbine increases from zero to a peak value, and then
decreases to zero; correspondingly, the turbine aerodynamic efciency oscillates between a very small
value, at zero ow rate, and larger values at near design conditions. We now look for average values
for the efciency g and for corresponding average values for the polytropic exponent k.
An OWC wave energy converter may be regarded as a dynamical system having as input the wave
elevation at a given point, and as output the pressure oscillation p(t) in the air chamber. In a given sea

Fig. 4. Polytropic exponent k versus pressure ratio r p=pat for different values of turbine efciency g .

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

25

state, the wave elevation may be regarded with good approximation as a stationary Gaussian process
with a variance density spectrum Sf f (see e.g. [20]). If the system is linear, then the air pressure oscillation is also a stationary Gaussian process whose variance density spectrum Sp f can be obtained for
the pressure oscillation provided that the transfer function is known (see e.g. [25]). Then, the variance
(or mean-square value) rp , and the probability density function /p, of the pressure oscillation p, can
be obtained directly from Sp f . From /p, by integration, we may nd average values for turbine
power output and turbine aerodynamic efciency, provided that the instantaneous performance
curves of the turbine are known. Such procedures are described in detail in [26].
The Wells turbine is known to be approximately linear in terms of ow rate versus pressure head,
at xed rotational speed (see [14,27]). This means that an OWC wave energy converter equipped with
a Wells turbine may be regarded with fairly good approximation as a linear system. In self-rectifying
impulse turbines, the ow rate is approximately proportional to the square-root of pressure head [14],
which is a nonlinear relationship that makes the linear system assumption a rougher approximation;
as in [25], we accept this approximation at this point (and nowhere else in the paper) solely for the
purpose of obtaining typical values for the averaged efciency of self-rectifying impulse turbines of
OWC plants subject to random waves (see Fig. 5). Dimensionless results from model testing are presented in Fig. 5 for a Wells turbine and for a biradial impulse turbine [28]. In both cases, the tested
turbines may be regarded as state-of-the-art machines. The solid curves in Fig. 5 give the instantaneous efciency g of the turbine versus dimensionless pressure head W. The dotted curves represent
 versus the variance or root-mean-square rW of W. Here, it is
the average efciency g
2 2
W jpjq1
air X D , where qair is air density, X is rotational speed (in radians per unit time) and D is
turbine rotor diameter. These results indicate that, if a good turbine is employed and its rotational
 is
speed is adequately controlled to match the sea state, the average efciency of the turbine g
expected to be in the range 0.6 to 0.7.
 for the turbine, it seems reasonable to dene also an average
Having dened an average efciency g
value for the polytropic exponent k of the density-pressure relationship in the air chamber. Fig. 4 indicates this value to be in the range 1.2 to 1.25 for state-of-the-art air turbines.
We consider now the spring-like effect due to air compressibility in the chamber. Let qch , V and
m qch V be the instantaneous values of the density, volume and mass of air in the chamber. The mass
ow rate of air through the turbine (positive for inward ow) is

dm
dq
V ch qch q;
dt
dt

where q dV=dt is the volume ow rate displaced by the motion of the OWC free surface (in a structure-xed frame of reference for a oating device). The rst term on the right-hand-side of Eq. (7), proportional to dqch =dt, represents the spring-like effect of air compressibility in the chamber. If dynamic
similarity is to be respected, then the ratio between the two terms on the right-hand-side of Eq. (7),

Fig. 5. Aerodynamic efciency curves for a Wells turbine (left) and a biradial impulse turbine (right). Solid curves:
 versus variance (or rms)
instantaneous efciency g versus dimensionless pressure head W. Dotted curves: average efciency g
rW of W. From [28].

26

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

V dqch
;
qch q dt

should take equal values under corresponding conditions in model and at full-scale. We saw above
that pressure and density of air in the chamber, p pat and qch ; may be related to pressure and air density in the atmosphere, pat and qat ; by a polytropic relationship (6), where the polytropic exponent k is
 of the turbine as shown in Fig. 4. From Eqs. (6) and (8), we nd
related to an averaged efciency g

V
dp
:
kqp pat dt

For perfect Froude similarity, the scale ratios are e5=2 for q, e d for p, (d qm =qF is water density ratio),
e1=2 for time, and e1=2 d for the time derivative of pressure dp=dt.
We consider rst that the geometric similarity is extended to the part of the device located above
water level, i.e. V m =V F e3 . This would allow realistic representation of water motion inside the
chamber and provide information on how to avoid green water from reaching the air turbine. Eq.
(9) shows that full dynamic similarity, i.e. Cm CF , requires km kF (equally efcient turbines at both
scales) and pat;m =pat;F e d. The latter condition, concerning atmospheric pressure, is obviously impossible to satisfy in practice if the length scale e is not close to unity.
Now we assume more realistically that the atmospheric pressure is the same at both scales
pat;m pat;F . We further assume that the pressure oscillation p is much smaller than the atmospheric

Fig. 6. Model testing of a cylindrical xed-structure OWC in a wave ume (Instituto Superior Tcnico, Lisbon, 2013). To
appropriately reproduce the air compressibility effect in the chamber, the top of the tube is connected by a pipe to an air
reservoir placed above.

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

27

pressure, which would exclude the more energetic sea states at full scale. If the Froude scale ratios for
ow rate q, pressure oscillation p and its time derivative dp=dt are as above, we are left with

Cm V m kF 2

e d:
CF V F k m

10

The correct representation of air compressibility effects in the chamber requires Cm CF . Then, we
obtain for the air chamber volume ratio

V m km 2 1

ed :
VF
kF

11

In model testing at large scale (say about 1:4th to 1:3rd) performed in the sea, it is d 1 (equal
water density). If in addition an appropriately scaled-sized turbine is used to realistically simulate
the full-sized machine, it is km kF (if the model turbine is not too small). Then, from Eq. (11), we have

Vm
e2 :
VF

12

A result identical to Eq. (12) was rst obtained in [15], based on a frequency-domain analysis of the
hydrodynamics and a linearized isentropic assumption. It later appeared in other papers [16,18,19].
It shows that the scale ratio for air chamber volume should be e2 , rather than e3 . Failure to meet this
condition may result in substantial errors in the conversion to full scale of the experimental data at a
smaller scale (see the numerical example in Section 6).
In testing at scales smaller than about 1:8th, the turbine is likely to be simulated by an orice or by
a window covered by porous material. In such cases, as mentioned above, we have a process at constant temperature, and the polytropic exponent is km 1: Eq. (11) becomes

Vm
1
e2 d1 :
kF
VF

13

If the model testing takes place in wave tank or ume lled with fresh water, which is the most frequent situation at such scales, the water density ratio is d qm =qF 0:97 .
If the OWC device is of xed structure (possibly bottom-standing), it is not difcult to satisfy condition (13) at small model scale. One simple way of achieving that is to connect the air chamber of the
model to a rigid-walled reservoir of air of appropriate volume. This procedure was adopted in the
model testing of the bottom-standing OWC installed in 1999 on the island of Pico, Azores, Portugal,
as reported in [18], and was also adopted in the model testing shown in Fig. 6. If however a oating
OWC device is to be modelled at small scale, this could introduce difculties because the reservoir is
likely not to be small compared with the devices size, and possibly is a lot larger. If the reservoir is
xed to the oating device, the stability and the dynamics may be severely affected. The alternative
is to have a exible pipe connection, which however would introduce undesirable forces and
moments.
It may be interesting to investigate whether the failure to meet condition (13) in small-scale testing
could be compensated by an appropriate design of the turbine simulator. We will do that in a simplied way, and consider only regular waves. Besides, we assume small wave amplitude and a linear system, so that frequency-domain analysis may be employed. The assumedly linear turbine is
represented by the relationship w=qat Kp, where K is a real positive proportionality constant. Taking into account the polytropic pressure-density relationship (6), the time derivative dqch =dt that
appears in Eq. (7) may be written as

dqch
qch dp qat dp

:
dt
kp pat dt k pat dt

14

Then, Eq. (7) becomes

Kp

V 0 dp
q;
kpat dt

15

28

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

where the chamber air volume V was replaced by its value V 0 in the absence of waves, and qch was
replaced by qat .
Since the system is linear and the incoming waves are regular, we may adopt the frequency domain
analysis and write p p0 eixt and q Q eixt , where p0 and Q are complex amplitudes and x is radian
frequency. Equation (14) takes the form

QF
xF V 0;F
i
;
p0;F
kF pat

KF

16

where the subscript F indicates full-sized prototype. We recall that K F is real positive for the full-sized
turbine, and therefore Q F =p0;F is complex and there is a non-zero phase difference vF argQ F =p0;F
between ow rate qF displaced by the OWC and pressure oscillation pF , except if V 0;F 0; as should
be expected. This phase difference should be the kept unchanged in the model, if hydrodynamic similarity is to be conserved.
We recall that the Froude scale ratios for q, p and x are e5=2 ; e d and e1=2 , respectively. Then, we may
write, for the model,

Km

Qm
xm V 0;m
Q
xF V 0;F
i
e7=2 d1 F  ie1=2
:
p0;m
k pat
p0;F
kF pat

17

From Eqs. (16) and (17), it is not difcult to show that the proportionality constant K m for the turbine
simulator cannot be real positive except if

V 0;m km 2 1

ed :
V 0;F
kF

18

Not surprisingly, this is similar to condition (12) obtained above. If a different ratio is to be adopted,
namely V 0;m =V 0;F e3 for perfect geometrical similarity, dynamical similarity requires a mechanism
exhibiting an appropriate phase difference between pressure head and ow rate, which, although conceivable, would be a much more complex device than an orice or a plug of porous material.
4. Aerodynamic modelling of air turbine
We assume now that the full-sized turbine and its model are geometrically similar. Dimensional
analysis applied to the compressible ow through a turbomachine (gas turbine, compressor) requires
that three dimensionless quantities take equal values between the model and the full-scale prototype
for full aerodynamic similarity: Reynolds number, Mach number and pressure ratio (see [11]). In our
case, during inhalation, the pressure ratio is

p pat
:
pat

19

We saw that Froude similarity requires that pm =pF ed, where e is length scale and d qm =qF is water
density ratio. Then, it is Km KF only if pat;m =pat;F ed, a condition involving atmospheric pressure
that cannot be satised in practice if the length scale e is not close to unity.
Since full aerodynamic similarity is not compatible with Froude similarity, we look for an approximation in turbine dimensional analysis. To do that, we assume, as above, that oscillations |p| in air
pressure are small compared with atmospheric pressure pat, and so changes in air density may be
neglected. This is equivalent to ignoring Mach number effects. In the case of approximately incompressible ow, the application of Buckinghams theorem yields the following relationship for the ow
through the turbine (see [11])

U F U W; Ret ;

g F g W; Ret ; N F N W; Ret ;

20

where

p
2

qat X D

qat XD

Yt
2

qat X D

Ret

XD2

mat

21

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

29

Here, w is mass ow rate through the turbine, qat and mat are, respectively, density and kinematic viscosity of air under atmospheric conditions, X is the rotational speed (in radians per unit time), D is the
turbine rotor diameter and Y t is the aerodynamic torque on the turbine rotor. In Eqs. (20) and (21), W
is dimensionless pressure head, g is aerodynamic efciency, U is dimensionless ow rate, N is dimensionless aerodynamic torque, and Ret is a Reynolds number. Functions F U , F g and F N are the same for
model and prototype, since these are assumed geometrically similar.
Note that bearing friction torque is in general relatively small and does not follow aerodynamic
similarity laws. For these reasons it will be ignored here.
We assume rst that Wm WF , for pressure head coefcient, and Um UF , for ow rate coefcient.
Later we will investigate whether these conditions are compatible with Ret;m Ret;F .
We introduce the notations N X Xm =XF and N D Dm =DF for the scale ratios for rotational speed
and rotor diameter, respectively. We assume that atmospheric conditions are the same for the model
and the prototype.
From Um UF we have

NX N 3D

wm
:
wF

22

Eq. (7) may be rewritten as w C 1qat q, where q is the volume ow rate displaced by the oscillating-water-column motion and C is a dimensionless quantity, dened by Eq. (8), that is related to the
spring-like air compressibility effect in the chamber. If this effect is adequately simulated in the
model, then Cm CF and wm =wF qm =qF . We recall that the conditions for this to apply are dened
in Section 3. The Froude scale ratio for the volume ow rate q is e5=2 : Eq. (22) becomes

NX N 3D e5=2 :

23

Condition Wm WF , combined with pm =pF e d for Froude similarity, gives

N2X N 2D e d:

24

This, together with Eq. (23), yields

Dm
e d1=4 ;
DF

25

Xm
e1=2 d3=4 :
XF

26

Now we easily nd

Ret;m
e3=2 d1=4 :
Ret;F

27

Since d is equal, or close, to unity, no equality in terms of Reynolds number is possible in the air ow
through the turbine. Eqs. (25) and (26) give the scale ratios for turbine size and rotational speed if a
geometrically similar turbine is used in the model testing of the OWC converter.
It may be of interest to investigate the effects of variations in Reynolds number on the efciency of
self-rectifying air turbines, particularly variations between prototype and model. We consider rst the
case of impulse turbines, that are geometrically and aerodynamically more similar to conventional
turbines than Wells turbines are. Typical values for the blade tip speed XD=2 of full-sized air turbines
of impulse type in various sea states range between about 50 and 100 m/s. Typical rotor diameters D
of full-sized impulse turbines are between about 0.7 and 2.0 m. Taking mat 1:47  105 m=s at 15 C,
we nd, for the Reynolds number Ret XD2 =m, the values 4:7  106 for a small turbine in mild seas,
and 2:7  107 for a large turbine in energetic seas. There is extensive information in the literature on
the effects of the Reynolds number on the performance of conventional turbomachines; only a few
results are presented here. In centrifugal pumps, the efciency becomes independent from the Reynolds number for values of XD2 =m greater than about 4  106 , and is affected by a factor of about
0.95 or 0.75 if the Reynolds number is reduced to 4  105 or to 4  104 , respectively [9]. In the case

30

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

of centrifugal compressors, the critical Reynolds number was again found to be about 4  106 [29]. For
axial-ow gas turbines, the critical Reynolds number is about 2  105 based on blade chord and ow
velocity, or roughly Ret;crit XD2 =m 2  107 for a typical axial-ow gas turbine; for lower Reynolds
numbers, the loss 1  g should be corrected by a factor proportional to Re1=5
according to the Ainley
t
and Mathieson criterion (see [11]). It seems reasonable to conclude that, in the range of Reynolds
numbers within which full-sized impulse-type air turbines operate, the efciency is practically insensitive to variations in Reynolds number. Since, in model testing, Ret is proportional to e3=2 (assuming
d 1, see Eq. (27)), Reynolds number effects may signicantly affect the performance of the impulse
turbine model if the scale e is small, say less than 1:10th.
Compared with impulse turbines, and for identical applications, Wells turbines are characterized
by signicantly larger values of the rotor diameter D and the rotational speed. The following ratios
were derived in [14] from theoretical considerations: X DWells =X Dimp 2:3, DWells =Dimp 1:4,
Ret;Wells =Ret;imp 3:2. This shows that the Wells turbine is characterized by higher Reynolds number
than turbines of impulse type. On the other hand, Wells turbines, because of their special aerodynamic
conception, are known to be much more sensitive to changes in Reynolds number than more conventional turbines, as explained in [14], and to perform poorly in small model testing (and small ow
velocities), more so than impulse turbines [14]. A review of experimental rigs for testing air turbines
can be found in [14]. Most of the more reliable results for Wells turbines as well as impulse turbines
were obtained with model rotor diameters about 0.6 m and Reynolds numbers about 6  106 to
12  106 . It is worth mentioning that the highest efciency values for Wells turbines (about 0.75) were
recently measured in a test rig with rotor diameters about 0.4 and rotational speeds up to 4000 rpm
[27]. Although experimental evidence is scarce, we may say that, at small scales (say e < 1=6, Reynolds number effects are likely to be more marked on Wells turbines than on impulse turbines.
5. Turbine rotational speed control
The turbine rotational speed should be controlled to match the energetic level of the sea state. It
should be noted that variations in rotational speed affect the turbine aerodynamic efciency. Besides,
such variations also affect the damping effect of the power take-off system on the OWC and so, indirectly, affect the hydrodynamic efciency of the wave energy absorption process. The instantaneous
rotational speed is controlled through the electromagnetic torque on the generator rotor. This is
implemented by the programmable logic controller (PLC) of the plant by acting on the power
electronics.
The following control algorithm for irregular waves

Y e a Xb :

28

was proposed, based on numerical simulations of different OWC converters, air turbines and sea states
[28,30,31]. Here, Y e is the electromagnetic torque on the electrical generator rotor, a is a proportionality constant and b is an exponent. The optimized values a and b were found to depend on device and
turbine geometry, but not (or only weakly) on sea state. Optimized values of the exponent b were
found to range between 2 and 3.
Since the scale factors for X and Y e are respectively e1=2 d3=4 and e4 d3=2 , we nd easily

am
e8b=2 d63b=4 :
aF

29

The dimensionless exponent b is unchanged by the conversion.


If the bearing friction torque is ignored, the time averaged values of the pneumatic power XY t and
of the electromagnetic power XY e are equal. The instantaneous power difference XY t  Y e is equal to
the rate of change of the kinetic energy stored in the coupled rotors of the turbine-generator set operating as a ywheel. We may write, from basic dynamics,

Yt  Ye I

dX
;
dt

30

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

31

where I is the rotational inertia of the turbine and generator rotor set. For dynamic similarity, we easily nd, from the scale factors for torque, rotational speed and time,

Im
e5 d1=4:
IF

31

When, in OWC model testing, the turbine is simulated by a simple device like an orice or a porous
plug, it is assumed that the device provides a reasonably good representation of the pressure-versusow-rate curve of the real turbine. It is well known [14,27] that Wells turbines are approximately linear, i.e. we may write W C 1
Wells U, where C Wells depends on the turbine geometry but not on its size,
rotational speed or uid density. It is also known that, for self-rectifying impulse turbines, of axial2
ow and radial-ow types, the relationship is approximately quadratic [14,28], i.e. W C 1
imp U . A plug

made of porous material may exhibit a linear pressure-versus-ow-rate relationship pm v1
plug wm
provided that it is designed in such a way that the ow through the porous material is laminar. The
2
relationship for an orice is approximately quadratic (i.e. pm v1
orif wm if the Reynolds number is
not too small. For these reasons, in model testing in wave tank or wave ume, a porous plug or an orice are frequently used to simulate a Wells turbine or an impulse turbine, respectively.
It should not be forgotten that, in most OWC plants, the rotational speed of the turbine is not constant: it varies with sea state and also over the wave-to-wave time scale, depending on rotational inertia and on the rotational speed control strategy and algorithm, as seen above. So it is important to
examine how variations in turbine rotational speed affect the pressure-versus-ow-rate relationship.
From the denitions of U and W (see Eqs. (21)), we may write for impulse turbines

p
1

;
w2 C imp qat D4

32

which means that the relationship between pressure head p and ow rate w is not signicantly
affected by changes in rotational speed. The same is not true in the case of Wells turbines, for which
we nd

p
X

:
w C Wells D

33

We may conclude that, while the damping provided by an orice, such that vplug C imp qat D4 ; may satisfactorily simulate a real impulse turbine damping independently of variations in rotational speed, a
given porous plug can only simulate a Wells turbine at a xed rotational speed. In the latter case, it
should be vplug C Wells DX1 :
6. Numerical example
It may be interesting to investigate the magnitude of the error introduced if results from model
testing of an OWC device are converted into full-size prototype values when, as is frequently done,
the compressibility effect of the air in the chamber is not appropriately simulated. We illustrate this
with the real case of the bottom-standing OWC plant installed in 1999 on the island of Pico, Azores,
Portugal, and still operational. The plant was tested in wave tank in 1993, at scale 1:35, at the National
Civil Engineering Laboratory, Lisbon [18], and again, in 1994, at scale 1:25, at the University College
Cork, Ireland. In both cases, the air compressibility was simulated by connecting the air chamber of
the model to an air reservoir of appropriate volume.
The plant is shown in cross section in Fig. 1.a, and is described in detail in [32]. The chamber crosssection is square with 12 m  12 m, and the air volume, in the absence of waves and under mid-tidal
level conditions, is V 0 1050 m3 : The plant is equipped with a Wells turbine, with guide vanes, of
rotor diameter D 2:3 m: Model tests of the turbine indicated that it is approximately linear and characterized by U C Wells W or w C Wells D X1 p, with C Wells 0:680 .
We use the frequency domain analysis, with regular waves of radian frequency x and amplitude
Aw ; and assume the pressure on the inner free-surface as spatially uniform, rather than modelling

32

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

the free-surface as a rigid piston. In linear theory, we may write (see [33]) q qr qe , where q is, as
above, the ow rate displaced by the inner free-surface motion, qr is the radiation ow rate (induced
by the air pressure oscillations in the absence of incident waves) and qe is the excitation ow rate
(induced by the incident waves if the chamber air pressure were equal to the atmospheric pressure).
In the frequency domain, we write

fp; q; qr ; qe g fp0 ; Q ; Q r ; Q e g eixt ;

34

where p0 , Q , Q r and Q e are complex amplitudes. We also write Q r G iBp0 , where G and B are the
frequency-dependent radiation conductance and radiation susceptance, respectively, and jQ e j RAw ;
where R is an excitation ow rate coefcient. Values of the hydrodynamic coefcients G, B and R were
computed for the Pico OWC plant with the aid of a boundary-element-method code [34] and are plotted versus frequency x in [30].We may nd (see [26])


p0

C Wells D

qat X


Gi

xV0
kpat

 1
B
Q e;

35

where, as above, k is the polytropic exponent. In our linearized analysis, the instantaneous power Pavai
available to the turbine is given by the pressure head p times the volume ow rate w=qat . This may
1 2
be written as P avai C Wells Dq1
at X p . Taking into account Eq. (34), we nd, for the time-average
value of Pavai ,

C Wells D
Pavai
2qat X

"
C Wells D

2 
2 # 1
xV0
G
B
R2 A2w :
qat X
kpat

36

Note that this is power available to the turbine, not turbine power output.
 1;avai P
 avai L7=2 q1 g 3=2 versus dimensionless
Fig. 7 shows curves of the dimensionless power P
1=2 1=2

wave frequency x x L g
. At full scale, the size of the chamber cross-section is LF 12 m
and the wave amplitude is Aw;F 1 m. In Fig. 7, the solid line represents the real prototype, with polytropic exponent k = 1.25 and rotational speed XF 80 rad=sec (764 rpm). The chain line represents a
model at 1:4th scale, assumed to be tested in sea water, with a scaled Wells turbine (rotor diameter
Dm 0:575 m, rotational speed Xm 160 rad=sec, polytropic exponent k 1:25) and a chamber air
volume equal to V 0;m V 0;F e3 1050  43 16:4 m3 . Finally, the dotted line represents a model at
1:35th scale, tested in fresh water, with a porous plug simulating a linear turbine (polytropic exponent
k 1:0) and a chamber air volume equal to V 0;m V 0;F e3 1050  353 0:0245 m3 . Note that in
both models, the chamber volume was chosen as V 0;m V 0;F e3 , rather than as dened by Eq. (11)
for appropriate representation of the air compressibility effect.

 1;avai P
 avai L7=2 q1 g 3=2 available to the turbine, versus dimensionless wave
Fig. 7. Dimensionless time-averaged power P
frequency x x L1=2 g 1=2 . Solid line: full-sized prototype. Chain line: model at 1:4th scale in sea water, equipped with model
turbine, and with air chamber scaled as the submerged structure. Dotted line: model at 1:35th scale in fresh water, with porous
plug simulator, and with air chamber scaled as the submerged structure.

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

33

The results plotted in Fig. 7 show that scaling the air chamber in the same way as the submerged
part of the structure (i.e. V 0;m V 0;F e3 may introduce signicant errors into the results for absorbed
power. The errors are larger for the smaller scale 1:35th, but the representation is only marginally better for the much larger scale 1:4th. It should be noted that the errors may change sign depending on
the wave frequency, which can be explained by the phase difference, due to the air compressibility
effect, between the air ow rate displaced by the OWC motion and the air pressure oscillation in
the chamber.
7. Conclusions
Relationships were derived for application in model testing of OWC converters in wave tank or in
sheltered sea waters. The new results concern the aero-thermodynamics of the air chamber and the
aerodynamics of the air turbine.
If the air chamber is to be geometrically scaled as is the submerged part of the device (i.e. the
chamber volume scale ratio is equal to the cube of length scale ratio e), exact dynamic similarity of
the spring-like effect of air compressibility in the chamber would require the atmospheric pressure
(or room pressure) during the model testing to be much smaller that the atmospheric pressure at full
scale, which is unpractical or even impossible. For the more realistic case when the atmospheric pressure is the same, then approximate dynamic similarity can be obtained provided that the air chamber
volume scale ratio is equal to km =kF e2 d1 , where d is water density ratio, and k (km in model and kF in
full-sized prototype) is a polytropic exponent that is equal to c 1:4 for a perfectly efcient air turbine, is in the range of about 1.2 to 1.25 for good air turbines, or is simply equal to unity if the laboratory device used to simulate the turbine is a simple orice or a plug made of a porous material. It
should be noted that in many published papers reporting OWC model testing these similarity rules
were simply ignored (the scale ratio for chamber volume was simply taken equal, or approximately
equal, to e3 , the consequence being that substantial errors may have been introduced and overlooked
in the conversion of results from the tested model to the full-scale prototype. That such errors may be
substantial was conrmed by a numerical example. It was found that, if the chamber volume scale
ratio is taken equal e3 ; approximate dynamical similarity can still the achieved, but this would require
the real turbine to be simulated in laboratory by a mechanism exhibiting a phase difference between
air pressure head and ow rate.
If geometrically similar turbines are to be used at both scales (which in practice would require
model testing at a relatively large scale, possibly about about 1:4th), then the linear scale ratio for
the turbine should be the same as the length scale ratio e for the submerged structure of the wave
energy converter, and the rotational speed scale ratio should be e1=2 (with small corrections if the
water density is not the same).
Finally, scale ratios were derived for application in rotational speed control of the turbine-generator set.
Acknowledgements
This work was funded by the Portuguese Foundation for Science and Technology through IDMEC,
under LAETA Pest-OE/EME/LA0022 and contracts PTDC/EME-MFE/103524/2008 and PTDC/EME-MFE/
111763/2009, and by Project Offshore Test Station, KIC InnoEnergy, European Institute of Technology.
The authors want to thank the anonymous reviewers for their constructive comments.
References
[1]
[2]
[3]
[4]
[5]
[6]

P.W. Bridgman, Dimensional Analysis, Yale University Press, New Haven, 1922.
H.L. Langhaar, Dimensional Analysis and Theory of Models, Wiley, New York, 1951.
L.I. Sedov, Similarity and Dimensional Analysis in Mechanics, 10th ed., CRC Press, Boca Raton, Florida, 1993.
S.A. Hughes, Physical Models and Laboratory Techniques in Coastal Engineering, World Scientic, New Jersey, 1993.
J.N. Newman, Marine Hydrodynamics, MIT Press, Cambridge, Massachusetts, 1977.
M.E. McCormick, Ocean Wave Energy Conversion, Wiley, New York, 1981.

34

A.F.O. Falco, J.C.C. Henriques / International Journal of Marine Energy 6 (2014) 1834

[7] G. Payne, Guidance for the Experimental Tank Testing of Wave Energy Converters, SuperGen Marine Rep., Univ. Edinburgh,
2008. Available at <http://www.supergen-marine.org.uk/drupal/les/reports/WEC_tank_testing.pdf>.
[8] V. Heller, Development of wave devices from initial conception to commercial demonstration, in: Sayigh, A. (Ed.),
Comprehensive Renewable Energy, vol. 8, Ocean Energy. Elsevier, Oxford, pp. 79110.
[9] G.T. Csanady, Theory of Turbomachines, McGraw-Hill, New York, 1964.
[10] S.A. Korpela, Principles of Turbomachinery, Wiley, Hoboken, New Jersey, 2011.
[11] S.L. Dixon, C.A. Hall, Fluid Mechanics and Thermodynamics of Turbomachinery, seventh ed., Elsevier, Amsterdam, 2014.
[12] A.F.de.O. Falco, Wave energy utilization: a review of the technologies, Renew. Sustain. Energy Rev. 14 (2010) 899918.
[13] T.V. Heath, A review of oscillating water columns, Philos. Trans. R. Soc. A 370 (2012) 235245.
[14] A.F.O. Falco, L.M.C. Gato, Air turbines, in: A. Sayigh, A. (Ed.), Comprehensive Renewable Energy, vol. 8, Ocean Energy,
Elsevier, Oxford, 2012, p. 111149.
[15] A.J.N.A. Sarmento, A.F.de.O. Falco, Wave generation by an oscillating surface-pressure and its application in wave-energy
extraction, J. Fluid Mech. 150 (1985) 467485.
[16] R. Jefferys, T. Whittaker, Latching control of an oscillating water column device with air compressibility, in: D.V. Evans, A.F.
de O. Falco (eds.), Hydrodynamics of Ocean Wave Energy Utilization, Springer, Berlin, 1986.
[17] A.F. de O. Falco, P.A.P. Justino, OWC wave energy devices with air-ow control, Ocean Eng. 26 (1999) 12751295.
[18] A.J.N.A. Sarmento, Model tests optimisation of an OWC wave power plant, Int. J. Offshore Polar Eng. 3 (1993) 6672.
[19] J. Weber, Representation of non-linear aero-thermodynamic effects during small scale physical modelling of OWC WECs,
in: Proc. 7th European Wave Tidal Energy Conf, Porto, Portugal, 2007.
[20] L.H. Holthuijsen, Waves in Oceanic and Coastal Waters, Cambridge University Press, Cambridge, 2007.
[21] J. Lighthill, Waves in Fluids, Cambridge University Press, Cambridge, 1978.
[22] S.W. Hong, J.-H. Kim, Experimental study of a compliant mooring system for a oating OWC device, in: Proc. 14th Int.
Offshore Polar Eng. Conf., Toulon, France, 2004, vol. 1, pp. 225231.
[23] F. Cerveira, N. Fonseca, R. Pascoal, Mooring system inuence on the efciency of wave energy converters, Int. J. Marine
Energy 34 (2013) 6581.
[24] F. Thiebaut, D. OSullivan, P. Kracht, S. Ceballos, J. Lpez, C. Boake et al., Testing of a oating OWC device with movable
guide vane impulse turbine power take-off, in: Proc. 11th Wave Tidal Energy Conf., Southampton, UK, 2011.
[25] A.V. Oppenheim, A.S. Willsky, S.H. Nawab, Signals and Systems, second ed., Prentice Hall, Upper Saddle River, New Jersey,
1997.
[26] A.F.de.O. Falco, R.J.A. Rodrigues, Stochastic modelling of OWC wave power performance, Appl. Ocean Res. 24 (2002) 59
71.
[27] R. Startzmann, T. Carolus, Model-based selection of full-scale Wells turbines for ocean wave energy conversion and
prediction of their aerodynamic and acoustic performances, Proc. Inst. Mech. Eng. Part A: J. Power Energy 228 (2014) 216.
[28] A.F.O. Falco, J.C.C. Henriques, L.M.C. Gato, R.P.F. Gomes, Air turbine choice and optimization for oating oscillating-watercolumn wave energy converter, Ocean Eng. 75 (2014) 148156.
[29] D. Japikse, N.C. Baines, Introduction to Turbomachinery, Oxford University Press, Oxford, 1994, p. I-48.
[30] A.F.de O. Falco, Control of an oscillating water column wave power plant for maximum energy production, Appl. Ocean
Res., 24 (2002) 7382.
[31] J.C.C. Henriques, J.C. Chong, A.F.O. Falco, R.P.F. Gomes, Latching control of a oating oscillating water column wave energy
converter in irregular waves, in: Proc. 33rd Int. Conf. Ocean Offshore Arctic Eng., San Francisco, 2014, paper No.
OMAE2014-23260.
[32] A.F. de O. Falco, The shoreline OWC wave power plant at the Azores, in: Proc. 4th European Wave Energy Conf., Aalborg,
Denmark, 2000, paper No. B1.
[33] J. Falnes, Ocean Waves and Oscillating Systems, Cambridge University Press, Cambridge, 2002.
[34] A. Brito-Melo, T. Hofmann, A.J.N.A. Sarmento, A.H. Clment, G. Delhommeau, Numerical modelling of OWC-shoreline
devices including the effect of surrounding coastline and non-at bottom, Int. J. Offshore Polar Eng. 11 (2001) 147154.

You might also like