The Spring-Like Air Compressibility Effect in Oscillating-Water-Column Wave Energy Converters - Review and Analyses

Renewable and Sustainable Energy Reviews 112 (2019) 483–498
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser
The spring-like air compressibility effect in oscillating-water-column wave T

energy converters: Review and analyses
António F.O. Falcão∗, João C.C. Henriques
IDMEC/LAETA, Instituto Superior Técnico,Universidade de Lisboa, Av. Rovisco Pais 1, 1049-001, Lisbon, Portugal
ARTICLE INFO ABSTRACT
Keywords: The oscillating-water-column (OWC) wave energy converter with air turbine has been object of extensive re-
Wave energy search and development effort, including the deployment of floating and fixed-structure full-sized prototypes
Oscillating water column into the sea. It consists of a hollow (fixed or floating) structure, open to the sea below the water surface. Wave
Air turbine action alternately compresses and decompresses the air trapped above the inner water free-surface in a chamber,
Air chamber
which forces air to flow through a turbine coupled to an electrical generator. The spring-like effect of air
Compressibility
Thermodynamics
compressibility in the chamber is related to the density-pressure relationship. It is known to significantly affect
the power performance of the full-sized converter, and is rarely accounted for in theoretical modelling, and even
more rarely in physical model testing at reduced scale, as appears from the literature review. Three theoretical
models of increasing complexity are analysed and compared: (i) the incompressible air model; (ii) the isentropic
process model; (iii) and the (more difficult and rarely adopted) adiabatic non-isentropic process model in which
losses due to the imperfectly efficient turbine are accounted for. The air is assumed as a perfect gas. The hy-
drodynamic modelling of wave energy absorption is based on linear water wave theory. The validity of the
various simplifying assumptions, especially in the aero-thermodynamic domain, is illustrated by a case study
with numerical results for a fixed-structure OWC equipped with a Wells turbine. Results are shown for regular
and irregular waves, and for a theoretical simulation of model testing in wave tank at small scale.
1. Introduction large enough to accommodate tidal oscillations and to avoid ingestion

of green water by the air turbine under rough sea conditions. Typical
The waves are a major renewable energy resource for countries with design values of the air chamber volume divided by the area of the
coasts facing the open sea [1–3]. Among the wide variety of wave en- OWC free surface range between 3 and 8 m. The spring-like effect of air
ergy converters, the oscillating water column (OWC) is regarded by compressibility in the chamber is related to the pressure-density re-
many developers as the simplest and most reliable technology, and is lationship, and increases with chamber volume. Such effect is important
the one that has been object of the most extensive research and de- in a full-sized OWC converter, as shown for the first time in Ref. [8],
velopment effort, including the deployment of prototypes into the sea where a simplified isentropic model was used. This isentropic as-
[4–7] (Fig. 1). In an OWC, there is a fixed or floating hollow structure, sumption was adopted in many subsequent papers, in most cases in a
open to the sea at its submerged part, that traps air in a chamber above linearized way. A more realistic and sophisticated non-isentropic model
the inner free-surface. Wave action alternately compresses and de- was developed in Ref. [9] for an OWC equipped with a bidirectional
compresses the trapped air which is forced to flow through a turbine turbine and a relief valve. In Ref. [8], it was recognized that, for such
coupled to an electrical generator. In almost all OWCs, the air alter- effect to be adequately simulated in model testing (Froude linear scale ε
nately flows from the chamber to the atmosphere and back. Unless for the submerged parts of the converter), the ratio between the air
check valves are used, the turbines are self-rectifying, i.e. their rota- chamber volume of model and prototype has to be equal to the square
tional direction remains unchanged regardless of the direction of the air 2
, not the cube 3 , of the scale. This rule was implemented (possibly for
flow. This is assumed to be case here (see Fig. 2). An extensive review of the first time in OWC model testing) as part of the design work of the
OWC wave energy converters and self-rectifying air turbines can be Pico OWC plant [10,11], and recently in Ref. [12], but has been ignored
found in Ref. [6]. in almost all other model testing work. The compressibility effect was
The volume of the air chamber of the OWC converter should be taken into account in the theoretical modelling of OWCs, mostly in the
∗
Corresponding author.
E-mail addresses: antonio.falcao@tecnico.ulisboa.pt (A.F.O. Falcão), joaochenriques@tecnico.ulisboa.pt (J.C.C. Henriques).
https://doi.org/10.1016/j.rser.2019.04.040
Received 16 August 2018; Received in revised form 13 March 2019; Accepted 11 April 2019
Available online 12 June 2019
1364-0321/ © 2019 Elsevier Ltd. All rights reserved.
A.F.O. Falcão and J.C.C. Henriques Renewable and Sustainable Energy Reviews 112 (2019) 483–498
Nomenclature v air velocity

V air chamber volume
Roman letters V0 = V in the absence of waves
w turbine mass flow rate
a radius of OWC Z = V0/( a2 ) average height of air chamber
Aw regular wave amplitude
c wave speed Greek letters
cp specific heat at constant pressure
cv specific heat at constant volume α atmospheric density ratio
D rotor diameter = cp/ c v
E turbine work per unit mass excitation flow rate coefficient
G radiation conductance δ tank-to-sea water density ratio
h specific enthalpy (thermodynamics), water depth ε Froude scale
hr memory function η turbine efficiency
H radiation susceptance = KD/( at ) turbine coefficient
Hs significant have height wavelength
i imaginary unit turbine pressure ratio
k wavenumber dimensionless turbine power
K = / dimensionless linear turbine coefficient density
K sh shoaling coefficient dimensionless volume flow rate
l = L / w energy loss per unit mass dimensionless pressure head
L turbine power loss ω angular frequency
m mass of air in chamber rotational speed
OWC oscillating water column
p pressure oscillation in chamber Subscripts
pat atmospheric pressure
P complex amplitude of p at atmospheric
Pabs power absorbed from OWC fs full scale
Pavai power available to turbine is isentropic modelling
Pt turbine power output mod model
q OWC volume flow rate opt optimal
Q complex amplitude of q ref reference value
R = cp c v gas constant w water
s specific entropy
t time Superscripts
T absolute temperature (thermodynamics), wave period
Te energy period overline average
u specific internal energy dimensionless value
U internal energy
linearized isentropic version, in other papers [13–17]. testing in wave tank. The modelling of the hydrodynamic process of
In model testing of OWCs, except for relatively large scales (not less wave energy absorption is based on linear water wave theory.
than about 1/6th), the air turbine is in general simulated by a simple Basic thermodynamics is applied in Section 2 to model the spring-
device, usually an orifice or a porous plug. In most cases, this gives only like compressibility effect, including the contribution due to the im-
a rough representation of the pressure-versus-flow-rate relationship of a perfect efficiency of the turbine. Dimensional analysis is applied to the
real turbine [18]. Here, we are not concerned with how good this ap- air turbine in Section 3. This is followed, in Section 4, by the hydro-
proximation may be. dynamics of wave energy absorption based on linear water wave
The aero-thermodynamic process of air entering and leaving the theory, in the frequency and time domains. The theory underlying
OWC air chamber is similar to the charging and discharging a rigid- model testing, especially how it is affected by the air compressibility in
walled gas reservoir, the main difference being the piston-like motion of the chamber, is explained in Section 5. These issues are illustrated, in
the OWC inner free surface. This process is analysed in general in Ref. Section 6, by a case study involving the theoretical/numerical analysis
[19]. An important application is the fast filling process of a natural gas of the performance of a fixed-structure OWC converter with a Wells
or a hydrogen onboard vehicle's cylinder, modelled in Refs. [20,21]. turbine. Finally, conclusions are drawn in Section 7.
Three theoretical models of increasing complexity are analysed and
compared here: (i) the incompressible air model; (ii) the isentropic
2. Thermodynamics
process model; (iii) and the (more difficult and very rarely adopted)
adiabatic non-isentropic process model in which viscous losses due to
The air compression/decompression process in the chamber in-
the imperfectly efficient turbine are accounted for. The air is assumed
volves changes in air density that should not be neglected in a full-sized
as a perfect gas.
converter. This requires some thermodynamic modelling.
The validity of these various simplifying assumptions, especially in
The OWC converter is represented schematically in Fig. 2. The air
the aero-thermodynamic domain, is examined and discussed with the
chamber is limited (i) by its rigid walls, (ii) by the inner water free-
aid of a case study with numerical results for a fixed-structure OWC
surface, and (iii) by a surface S1 that conceptually separates the
equipped with a Wells turbine subject to regular and irregular waves.
chamber from the turbine space. Surface S2 separates the turbine space
The study also includes a numerical simulation of physical model
from the atmosphere.
484
Fig. 1. OWC prototypes deployed into the sea: (a) shoreline (Pico Island, Portugal), (b) U-OWCs integrated into breakwater (Civitavecchia, Italy), (c) spar-buoy
(Basque Country, Spain) (courtesy of IDOM), (d) backward-bent-duct-buoy BBDB (Galway Bay, Ireland) (courtesy of Kymaner).
Depending on converter design, its location (floating, bottom- 2.1. Air chamber
standing, incorporated into a breakwater), local water depth, sea state,
etc., the physical process that takes place in the air chamber and in the The chamber and the air it contains may be considered as an open
turbine can be variously complex. The eventual presence of water spray system. It is reasonable to consider the process as adiabatic. This means
in the chamber and turbine may be due to an air jet from the turbine that the heat exchanged across the chamber and turbine walls and
that impinges upon the inner free surface during inhalation, or may be across the inner water free-surface is assumed negligible compared with
due to water drops in the atmosphere (from rain or from wave the work done by the inner free surface displacement and with the work
breaking) that are swallowed by the turbine. Wave breaking produces done by the turbine. In this respect, it should be noted that the air is
an emulsion of air bubbles in water; if this occurs close to the converter, partly renovated at each inhalation and so the temperature cannot
some of the bubbles may raise inside the OWC structure and reach the differ much from the atmospheric temperature, even if the turbine is far
chamber, by-passing the turbine. Such situations are difficult to predict from perfectly efficient. Besides, the temperature oscillations in the air
and to model, and will not be considered here. One-phase flow in the chamber are relatively small and their time scale (a few seconds) is too
chamber and turbine will be assumed. The exchange of air between the short for significant heat exchanges to occur. It could be added that the
chamber and the atmosphere is supposed to take place exclusively flow through steam and gas turbines is in general modelled as an
through the turbine. No change of phase is supposed to occur in the adiabatic process, although large temperature differences may occur
chamber or in the turbine space: evaporation and condensation are (see e.g. Ref. [22]). The adiabatic assumption in OWC converters with
excluded. self-rectifying air turbine has been accepted by all, or nearly all, re-
searchers.
We denote by p + pat ( pat is atmospheric pressure), ρ, T, U, u, and
h = u + (p + pat )/ the pressure, density, absolute temperature, in-
ternal energy, specific internal energy and specific enthalpy of air in the
chamber, respectively. The time-varying mass of air in the chamber is m
and the mass flow rate across surface S1 (positive during exhalation) is
w = dm /dt . The time-varying volume of air in the chamber is denoted
by V.
Most self-rectifying air turbines are of axial-flow type and are
symmetrical with respect to a plane perpendicular to the rotational axis.
This is the case in general of Wells turbines and axial-flow impulse
turbines. Such turbines are insensitive to flow direction. In particular it
is p = 0 for w = 0 . Some self-rectifying turbines have been proposed
and tested for which this is not true. It is in particular the case of tur-
bines of radial-flow or mixed-flow type based on the original Wells
turbine concept [23–25] or on the impulse or reaction rotor concept
[26,27]: at zero flow rate, the centrifugal force induces a non-zero
pressure difference across the turbine.
Assuming, in general, the air density to be a function (x , y , z , t ) of
the spatial coordinates (x , y , z ) and of time t, the mass m (t ) may be
written as
Fig. 2. Schematic representation of an OWC converter with bidirectional air
turbine.
485
m (t ) = (x, y, z, t )dV.
V (1)
The equation of continuity of mass is
(x, y, z, t )
w (t ) = dV + ¯ (t ) q (t ),
t (2)
V
where ¯ (t ) = m (t )/ V (t ) is a spatially averaged density of air in the

chamber, and q = dV /dt is the volume flow rate displaced by the
motion (relative motion in the case of a floating OWC converter) of the
inner free surface (positive for upward motion). The first term on the
right-hand-side of Eq. (2) represents the effect of air compressibility,
and its importance, or more precisely the amplitude of its oscillating
value, increases with the chamber volume.
In Eq. (2), the mass flow rate w through the turbine should be re-
lated to the instantaneous pressure p + pat in the chamber, to the tur-
bine performance curves and to the instantaneous rotational speed. This
has to do with the turbine aerodynamics and thermodynamics. On the
other hand, the volume flow rate q should be related to the incident
waves, to the time-record of the inner pressure oscillation p and, in the
case of a floating OWC, to the dynamics of the converter motion; this
should be modelled with the aid of hydrodynamics (and eventually
body dynamics).
From the First Law of Thermodynamics applied to an open system,
and assuming the process to be adiabatic, we may write, for the in-
halation (see Ref. [19]),
dU 1 Fig. 3. Entropy-enthalpy diagram for the turbine.

+ h1 + v12 w = (p + pat ) q,
dt 2 (3)
du 1 2
where v is the velocity of air, and subscript 1 denotes conditions on m = h1 + v1 u w + q (p + pat ).
dt 2 (6)
surface S1. Here, the pressure in the chamber is assumed spatially
uniform, and the gravitational potential energy of air in the chamber is Information on turbine performance is required to proceed with the
neglected; this is justified by the small density of air (compared with thermodynamic analysis of the process. Fig. 3 represents the enthalpy-
water) and by the limited vertical extention of the chamber (a few versus-entropy diagram (h, s ) as for a gas turbine (see e.g. Ref. [22]).
metres). The quantity on the right-hand-side of Eq. (3) is the rate of The diagram shows two constant pressure curves: p0,in is the stagnation
work done on the system by the motion of the inner free surface; this pressure at turbine inlet and pout is the static pressure at turbine exit.
may be positive or negative. During exhalation, it is p0,in = p + pat and pout = pat ; during inhalation,
In general, the temperature, density and specific internal energy of it is p0,in = pat and pout = p + pat . If the turbine were perfectly efficient
air in the chamber vary with position (x , y , z ) and with time t, and so it (isentropic expansion), the work per unit mass would be
should be Es = h 0,in hs,out . The work per unit mass done by the real turbine is
E = Es l , where the loss l = hout hs,out + vout 2
/2 includes viscous
dU d losses along the turbine in addition to the loss of the exit kinetic energy
= (x , y, z, t ) u (x , y , z , t )dV .
dt dt (4)
2
vout /2 . The turbine power Pt = |w|E is also given by Pt = Laero , where
V
is the rotational speed in radians per unit time and Laero is the aero-
During inhalation, and to a lesser extent during exhalation, the dynamic torque due to air action on the rotor blades and other rotor
process that takes place in the air chamber is very complex, with highly walls. If the rotational speed is not constant (as is usually the case in
turbulent flow. The theoretical modelling would involve the numerical an OWC plant subject to wave action), then Laero is different from the
integration, over the whole chamber and turbine spaces, of the shaft torque, and Pt is different from the power supplied to the electrical
Reynolds-Averaged Navier-Stokes (RANS) equations with an appro- generator, due to time variations in the kinetic energy of the rotating
priate turbulence model. This is out of the scope of the present paper. masses (rotor and shaft). The aerodynamic efficiency (or simply effi-
Here, as in Ref. [19], we assume that the mixing process in the chamber ciency) of the turbine is defined as = E /Es = 1 l/ Es . In the turbo-
is fast enough for the thermodynamic state of air to be approximately machinery literature, this is frequently named total-to-static efficiency
uniform at all times. In particular, for the internal energy, it is and denoted by ts . It should be noted that the power Pt becomes ne-
U = mu = V u , where u and ρ are only functions of time t. Equation (2) gative for very small flow rates, and so E as |w| 0.
becomes more simply During exhalation, it is
d 1 2
h1 + v1 = h
w= V + q.
(5) 2 (7)
dt
and Eq. (3) may be written as
In the absence of heat transfer, the process, during exhalation, is
approximately isentropic for the air remaining in the chamber. During du
= (h u)
1d
.
inhalation, the process is more complex, since the specific entropy of air dt dt (8)
at the turbine exit is larger that the atmospheric value sat due to viscous
During inhalation, it is
losses in the turbine, and the specific enthalpy h also changes due to
work performed by the turbine. l = h1 h1s +
1 2
v1 ,
If U = mu , Eq. (3) can be rewritten as 2 (9)
486
where h1s is the specific enthalpy of the thermodynamic state whose s > sat after some inhalation occurred. On the other hand, since the air
specific entropy is the same as in the atmosphere and the pressure is is renovated at each inhalation, s cannot increase indefinitely. After
equal to the chamber pressure. Equation (6) becomes some time (possibly a few wave periods), s is expected to oscillate about
q some average value sav > sat , with ds/dt being equal to zero during
du w L
= (u h1s ) + (p + pat ),
(10) exhalation, and taking negative, positive or zero values during inhala-
dt m m m
tion. It may be concluded that, over a large number of inhalation per-
where L = |w|l is the power loss. iods, the time-averaged value of ds/dt is zero, and the same must be
The instantaneous value of the mass m is related to the mass-flow true for the averaged value of the right-hand-side of Eq. (18). Note that
rate w by the power loss L in the real turbine is intrinsically positive. It may be
t concluded, as a first approximation, that
m (t ) = m (t 0 ) w (t ) dt ,
t0 (11)
cp w (T Ts ) L¯ , (19)
where t 0 is some initial time.
where the overline denotes here time-average during inhalation (re-
member that w < 0 during inhalation). Equation (19) shows that (at
2.2. Perfect gas
least in average) the power loss in the turbine produces an increase in
temperature in the chamber, compared with the isentropic situation, a
In what follows, the air is considered a perfect gas. The case of a
result that is not unexpected. We recall that cp dT = du , and so Eq. (19)
non-perfect real gas was considered in Refs. [28,29]. The perfect gas is a
may be written as w (u us ) L¯ (where the specific internal energy
reasonable assumption, since the pressure and temperature in the
us corresponds to the temperature Ts ), which means that, during in-
chamber and turbine are not substantially different from the corre-
halation, the power loss in the turbine results in an equivalent increase
sponding values in the atmosphere, and single-phase fluid flow is as-
in the flux of internal energy compared with the perfectly efficient
sumed. In this case, it is (p + pat )/ = RT , du = c v dT , dh = cp dT ,
turbine situation.
= cp/ c v , R = cp c v (cp , c v are the specific heat at constant pressure
The work done by the turbine per unit mass of air is E = Es . The
and at constant volume, respectively, and are assumed constant).
instantaneous efficiency η of the turbine is supposed to be known as a
The work per unit mass done by a perfectly efficient turbine is
function of the inlet stagnation conditions p0,in , T0,in , the outlet pressure
Es = cp (T0,in Ts,out ) or
pout and of the rotational speed.
( 1)/
pat
Es = cp T 1 (w > 0), 2.3. Perfectly and imperfectly efficient turbines
p + pat (12)
( 1)/
We consider now the ideal case of a perfectly efficient turbine
p + pat equipped with a perfect diffuser such that the kinetic energy at rotor
Es = cp Tat 1 (w < 0).
pat (13) exit has been totally recovered isentropically: the kinetic energy of air
at turbine exit is negligible. Since the whole process is assumed adia-
The time derivatives of specific entropy s, pressure p and density batic, it is also isentropic: the specific entropy has everywhere the same
are interrelated by value equal to sat . As above, we neglect spatial variations in air pressure
inside the chamber due to the motion of the inner free-surface and to
ds 1 dp d
= cv . gravity. It may be said that the kinetic energy of air is negligible ev-
dt p + pat dt dt (14) erywhere, except inside the turbine.
During exhalation (w > 0 ), it is ds/dt = 0 . Equation (8) leads to the From basic thermodynamics, it is known that the thermodynamic
well known isentropic relationship for a perfect gas state of the air may be defined by any two variables out of the variables
pressure, temperature, density and specific entropy. For a perfectly
1/
p + pat efficient turbine, we have just seen that, during exhalation, the specific
= .
at pat (15) entropy s and the pressure at turbine exit are equal to the corresponding
values in the atmosphere. So we can say that the thermodynamic
During inhalation (w < 0 ), the process is no longer isentropic. For a conditions of the air at turbine exit are identical to those of atmospheric
perfect gas, Eq. (10) becomes air. The same can be said, for the inhalation, about the air at turbine
dT w L q exit and the air in the chamber. It may be concluded that no irreversible
cv = (c v T cp Ts ) + RT ,
dt m m m (16) mixing processes occur in the chamber and in the atmosphere involving
masses of air with different thermodynamic conditions.
where
Since the whole process is adiabatic and isentropic, it follows, from
p + pat
( 1)/ basic thermodynamics, that
Ts = Tat
pat (17) d(U + Ut + Et,kin )
Pt = Pabs w (u2 uat ),
dt (20)
is the temperature of the thermodynamic state whose specific entropy
and pressure are sat and p + pat , respectively. Eliminating the volume where Pt is the turbine power (as defined above), Pabs is the rate of work
flow rate q between Eqs (16) and (5), we obtain, after some re- done by the motion of the inner free surface, U + Ut is the internal
arrangement, energy of the air contained in the chamber and turbine, Et,kin is the
kinetic energy of air inside the turbine and u2 is the specific internal
ds 1
= [L wcp (Ts T )] (w < 0). energy at surface S2 (turbine inlet/outlet, see Fig. 2). The term
dt mT (18)
w (u2 uat ) is the flux of internal energy into the system and is zero in
Equation (18) was derived for the first time in Ref. [9]. In the case of a the case of a perfect turbine since, as seen above, u2 = uat during in-
perfectly efficient turbine, it would be L = 0 and Ts = T , and hence halation and exhalation. In average, over a large time interval, it is
ds/dt = 0 , as should be expected. P¯t = P¯abs , which means that there is no power loss. The conclusion in
We consider a given sea state and assume that, at some initial time Ref. [13] that it could be otherwise, i.e. that the air compressibility
t 0 , the specific entropy s in the chamber is equal to sat . Since the whole effect induces a (small) loss of power even under the isentropic and
process is adiabatic, it cannot be s < sat for t > t 0 . More precisely, it is adiabatic assumptions, is presumably the result of the errors due to
487
linearizing approximations introduced in the isentropic relationship geometrically similar machines of different sizes, at various rotational
between air pressure and density, and in the definition of power speeds and with different fluid densities. Assuming the fluid to be a
available to the turbine in compressible flow (see below). perfect gas with = cp/ c v , we introduce the following dimensionless
In the case of a real, imperfectly efficient, turbine, we have, for the variables
power available to the turbine Pavai (which should be equal to the tur- p0,in |w| Pt
bine power output defined above as the aerodynamic torque on the = , = , = ,
pout D3 3 D5
(29)
rotor times the instantaneous rotational speed), 0,in 0,in
d(U + Ut + Et,kin ) 0,in D2 D

Pavai = Pabs (w < 0), Re = , Ma = .
dt (21) µ c0,in (30)
d(U + Ut + Et,kin ) v22 Here, subscripts “0,in” and “out” denote stagnation conditions at tur-
Pavai = Pabs w (u2 uat ) w (w > 0),
dt 2 (22) bine entry and static conditions at turbine exit, respectively, D is the
rotor diameter, c is the speed of sound, μ is the viscosity, is the
where v2 is the exit velocity on surface S2 . Note that, during exhalation
pressure ratio, is the dimensionless flow rate, is the dimensionless
(w > 0 ), it is v2 0 (imperfect turbine diffuser) and s2 > sat . Since
power output, Re is the Reynolds number and Ma is the Mach number.
p2 = pat , it follows that u2 > uat . It may be concluded that, over a long
Buckingham's theorem of dimensional analysis allows us to write (see
period of time, it is, in average, P¯avai < P¯abs .
Ref. [22])
2.4. Approximations , = function( , Re, Ma, ). (31)

The consideration, in the present analysis, of the Reynolds number and
Apart from the air incompressibility assumption, several approx-
the Mach number raises difficulties. This is due to the fact that ex-
imations have been proposed in the literature and used. Here we con-
perimental results for air turbines are in general obtained from model
sider those that concern: (i) the energy per unit mass available to the
testing in which both Mach number and Reynolds number are sig-
turbine between pressures p0,in and pout , and (ii) the relationship be-
nificantly lower than in full-sized machines operating under real sea
tween density and pressure p + pat in the chamber.
conditions. This is more critical in the case of Wells turbines: they are
Denoting, as before, the stagnation conditions at turbine inlet by the
known to be more sensitive to changes in Reynolds number and Mach
subscript “0,in” and the conditions at turbine outlet by the subscript
number than self-rectifying turbines whose rotor configuration derives
“out”, we may write (see Fig. 3)
from conventional axial-flow and radial-flow turbines of impulse or
Es = h 0,in hs,out . (23) reaction types. On what concerns the Reynolds number effect, we as-
sume here that the data to be used are taken from testing sufficiently
In the case of a perfect fluid, Eq. (23) becomes
large models at sufficiently high rotational speeds. Significant air
|p|
( 1)/ compressibility effects in the flow through self-rectifying turbines are
Es = cp T0,in 1 1 known to take place at Mach numbers that rarely have been reproduced
p0,in
(24) in laboratory testing. For these reasons, the effects of Mach and Rey-
nolds numbers are ignored here. Equation (31) is replaced by
or, introducing the binomial series,
= f ( ), = f ( ), (32)
|p| 1 |p| p2
Es = 1+ +O 2 . where
0,in 2 p0,in p0,in (25)
|p|
If the pressure oscillation |p| is small compared with the inlet stagnation = 2D 2
0,in (33)
pressure p0,in , then Eq. (25) becomes approximately
is the dimensionless pressure head, and = 1.4 for air. Besides, for the
|p|
Es . efficiency, we may assume, approximately, as in incompressible flow,
0,in (26)
In the limiting situation, this becomes the available energy per unit = = f ( ).
(34)
mass in incompressible flow. The relative error in Eq. (26) is
1 |p| 4. Hydrodynamics
Es .
2 p0,in (27)
The core of the present paper is thermodynamic modelling.
In an isentropic process of a perfect gas, the relationship between
However, it should not be forgotten that the overall goal is to in-
density and pressure is given by Eq. (15), or, with the binomial series,
vestigate the effect of air compressibility upon the converter perfor-
1 p p2 mance, an essential element of which is the hydrodynamic process of
=1+ +O 2 . wave-to-pneumatic energy conversion.
at pat pat (28)
Basically, there are two approaches to the hydrodynamic modelling.
Frequently authors neglect the terms of second order and higher in Historically, the first one is based on linear water wave theory and on
Eqs (25) and (28) (see e.g. Refs. [13,14]); this may introduce non- hydrodynamic coefficients. This was firstly developed for oscillating-
negligible errors in energetic sea states when |p| reaches values that are body wave energy converters (model (i)) (the basic paper is [31]) and is
not small compared with the atmospheric pressure pat . This question still frequently applied to OWCs, as an approximation, by replacing the
has been discussed in Ref. [30]. inner free-surface by a rigid piston (see e.g. Refs. [32,33]). A more
realistic model (model (ii) pioneered in Ref. [34]) assumes that the air
3. Air turbine pressure on the inner free-surface of the OWC is spatially uniform and
accounts for the deformation of the inner free-surface. Models (i) and
It is usual to apply dimensional analysis to the performance study of (ii) are described in detail in Falnes book [35]. In both models, hy-
turbomachinery [22]. This has the advantage of extending the applic- drodynamic coefficients are defined: (i) added mass, radiation damping
ability of the results from the testing of a given machine to coefficient and excitation force coefficient, in the oscillating-body
488
model, and (ii) radiation susceptance, radiation conductance and ex- (G and H are real), where G is the radiation conductance and H is the
citation flow rate coefficient, in the OWC model. Model (i) is more radiation susceptance (see Ref. [35]). The hydrodynamic coefficients G,
frequently used, even for OWCs, because commercial and open-source H and depend on the frequency ω and on the geometry of the sub-
codes (normally based on the boundary-element method) are widely merged part of the structure. Note that G cannot be negative.
available for the numerical evaluation of the coefficients for arbitrary Assuming the air as a perfect gas, Eq. (5) can now be written as
geometries; this can hardly be said of model (ii). In most cases, the 1/
application of model (ii) employs the analytical integration of Laplace's w p V dp
qe + qr = 1+ + .
equation for the velocity potential function based on the method of at pat (p + pat ) dt (39)
separation of variables, which in practice is feasible only for relatively
This may be linearized (i) by assuming that |p| is much smaller than the
simple geometries (see e.g. Refs. [36–42]). Relationships between the
atmospheric pressure pat , (ii) by replacing the chamber volume V (t ) by
two sets of hydrodynamic coefficients can be found in Ref. [6].
is value V0 in the absence of waves, and (iii) by considering a linear
The methods mentioned above, based on linear water wave theory,
turbine defined by = K , where K is a dimensionless proportionality
are unable to model non-small wave amplitudes and real fluid dis-
constant, and and are defined as in Eqs. (29) and (33), with 0,in
sipative effects that may be important especially in the more energetic
replaced by at . The result is
sea states and close to resonance. The alternative approach is the nu-
merical integration of the Reynolds-averaged Navier-Stokes (RANS) V0 dp
qe + qr = p+ ,
equations with an appropriate turbulence model. It is computationally pat dt (40)
much more demanding (especially if calculations in irregular waves are
where
to be of statistical value). Such methods have been employed to model
OWC converters in two-dimensional configurations (the so-called 2-D =
KD
numerical wave tank, see e.g. Refs. [43,44]) and more rarely in three- at (41)
dimensional geometries [15].
is a turbine coefficient ( 1
is proportional to turbine-induced
Here we adopt the model based on the linear water wave theory,
damping).
more specifically model (ii) (see above) in which the air pressure on the
inner free-surface is assumed as spatially uniform (the rigid piston as-
4.1.1. Regular waves
sumption is excluded). Floating OWC converters can be modelled by
In the case of regular waves of frequency ω, Eq. (40) becomes
combining model (i) for the structure motions and model (ii) for the
water column oscillations (see Ref. [35]). V0
1
The effects of air compressibility are important in both fixed- P= +i Q,

pat (42)
structure OWCs and in floating OWC converters. These effects are il-
lustrated in a case study in Section 6, where a fixed-structure converter where P and Q are complex amplitudes, see Eq. (36). Equation (42)
is analysed. shows that the pressure oscillation p (t ) is not in phase with the flow
rate q (t ) displaced by the inner free surface motion except if V0 = 0 : the
4.1. Frequency domain analysis air chamber introduces reactive power in the wave to pneumatic energy
exchange. Equation (42) may be rewritten as
We consider a wave energy converter consisting of a single fixed- 1
V0
structure OWC subject to unidirectional incident waves. The linear- P= +G + i +H Qe,
ization allows us to decompose the volume flow rate q (t ) displaced by pat (43)
the motion of the inner free-surface as
which relates the air pressure oscillation amplitude P to the excitation
q (t ) = qr (t ) + qe (t ). (35) flow rate amplitude Qe .
In this linearized approximation, the power available to the turbine
Here qr is the radiation flow rate due to the pressure oscillation p in the is Pavai = pw/ at , which, for the linear turbine considered here, becomes
chamber in the absence of incident waves, and qe is the excitation flow
rate due to the incident wave field if p = 0 . We assume that there is a Pavai (t ) = (p (t ))2. (44)
functional relationship between the flow rate q (t ) and the pressure Its time-average is
oscillation p (t ) associated to the aero-thermodynamic process that
takes place in the air chamber and in the turbine. P¯avai = |P|2 ,
We consider first that the incident waves are regular, with frequency 2 (45)
ω and amplitude Aw . Besides, we assume that a linear relationship holds or, taking into account Eq. (43),
between the flow rate q (t ) and the pressure oscillation p (t ) in the sense 1
2
that, if one of these variables is a sinusoidal function of time t, so is the V0
P¯avai = ( + G )2 + +H |Qe |2 .
other, possibly with a phase difference between them. Since linear 2 pat (46)
water wave theory is assumed, the wave-to-pneumatic conversion
system is linear, and we may write, in the frequency domain analysis, Equation (46) appeared, firstly and in a different form, in Ref. [8], and
later in Refs. [38,39] and elsewhere. In the case of a given OWC geo-
(q, qr , qe , p) = (Q , Qr , Qe, P )ei t , (36) metry and if the incident wave amplitude Aw and frequency ω are fixed,
where Q, Qr , Qe and P are, in general complex, amplitudes. Here, and then Qe , G and H are also fixed. Equation (46) shows that the time
whenever a physical quantity is equated to a complex expression, it averaged power available to the turbine (or equivalently the time
should be understood that the real part is to be taken. The excitation averaged power absorbed from the waves) P̄avai is maximum for
flow rate coefficient is defined as pat H ( )
V0 = if H ( ) < 0
=|Qe |/ Aw . (37) (47)
As usual in the frequency domain analysis, we further write, for the or for V0 = 0 if H ( ) > 0 (note that H may be negative, zero or positive
radiation flow rate, depending on the OWC geometry and on the wave frequency ω, see
Section 6. This shows that, if H < 0 , the compressibility effect of the air
Qr = (G + iH ) P (38) chamber may improve the wave-to-pneumatic energy conversion up to
489
an optimum value given by Eq. (47); beyond that, P̄avai will decrease to V 1 dp
X= .
zero as V0 . The turbine that maximizes P̄avai in Eq. (46) is char- q (p + pat ) dt (52)
acterized by
Condition Xmod = Xfs (see above) implies that
2 1/2
opt = G2 +
V0
pat
+H Vmod q
= mod
pmod + pat,mod ( )
dp
dt fs
.
(48) Vfs qfs pfs + pat,fs
( )
dp
dt mod (53)
and, for fixed ω, depends on the chamber volume V0 . Note that Eq. (48)
gives the ratio D/ that maximizes the energy available to the turbine Froude similarity requires that (see Ref. [18])
(or the energy absorbed from the waves). The computation of the se-
parate values of D and that also maximize the energy produced by the qmod
= 5/2,
pmod
= ,
( )
dp
dt mod
= 1/2,
turbine requires the knowledge of the efficiency curve of the turbine. qfs pfs
( )
dp
dt fs (54)
4.2. Time domain analysis where = w,mod / w,fs and w is water density; it is = 1/1.025 or = 1
depending on whether the model is tested in fresh water or in sea water.
In general, the relationship between the displacement volume-flow If Froude scale ε is to apply approximately to the air chamber volume,
rate q and the pressure oscillation p is not linear. The non-linearity may more precisely if Vmod/ Vfs = 3 , then it must be
be introduced by the air chamber thermodynamics and by the aero- pat,mod = pat,fs . (55)
dynamic performance curves of the turbine. In such cases, the fre-
quency domain analysis is no longer applicable and has to be replaced If, in reduced model testing, a smaller, geometrically similar, air
by a time domain analysis. In particular, the radiation flow rate qr can turbine is to be employed, then the appropriate rotor diameter ratio
be expressed by a convolution integral [9,35] Dmod / Dfs and rotational speed ratio mod / fs are to be determined. The
effects of different Reynolds number and Mach number are ignored
here. It should be mod = fs and mod = fs for the dimensionless flow
t
qr (t ) = h r (t ) p ( )d . (49)
rate and dimensionless pressure head, respectively, as defined by Eqs
The memory function hr is related to the radiation susceptance G (29) and (33), which may be written as
through the Fourier transform [9,35] 3
wmod at,mod mod Dmod
= ,
h r (s ) =
2
G ( )cos( s ) d .
wfs at,fs fs Dfs (56)
0 (50)
2 2
pmod at,mod mod Dmod
= .
pfs at,fs fs Dfs (57)
5. Model testing
On the other hand, Froude similarity for the hydrodynamic process of
Theoretical/numerical modelling should be the first step in the wave energy absorption requires that pmod / pfs = and qmod / qfs = 5/2
development of a wave energy converter. Testing the device at full- for the pressure oscillation p and the volume flow rate q displaced by
scale under real sea conditions is an expensive and risky exercise. This the OWC motion. If the compressibility effect is to be correctly simu-
is in general preceded by model testing in wave tank at scale between lated, it should be qmod / qfs = 1wmod/ wfs (where = at,mod / at,fs ) for the
about 1/100th and 1/10th, and possibly testing at scale about 1/4th in mass flow rate w through the turbine. These equalities yield
a confined sea area where the water depth and the representative wave Dmod 1/4 1/4 mod 3/4 3/4 1/2.
= , =
length approximately reproduce at the same scale the corresponding Dfs fs (58)
values at the ocean/sea location where the prototype is to be deployed
These results, with = 1 ( at,mod = at,fs ), were obtained in a different,
[18,45].
less general, way in Ref. [18].
In model testing, Froude similarity criterion should be met for the
Condition (55) is practically unfeasible if the Froude scale ε is not
part of the structure subject to wave action, but not for the air chamber
close to unity. The alternative is to neglect the pressure oscillations pfs
size and for the turbine or devices used to simulate it at reduced scale.
in the chamber of the prototype in comparison with the atmospheric
These issues are analysed here.
pressure pat,fs . Assuming pat,mod = pat,fs , the scale for the chamber vo-
We consider a given OWC converter being model tested at (Froude)
lume becomes
scale < 1. Equation (5) represents the air mass balance in the
chamber. If dynamic similarity is to be kept, then the ratio between the Vmod 1 2.
=
two terms on the right-hand-side of the equation should take equal Vfs (59)
values in model and prototype, i.e. Xmod = Xfs , where If, in addition, the atmospheric density is equal in model and prototype,
V d and the turbine prototype is to be simulated by a smaller, geometrically
X= identical turbine, then Eq (58) may be applied with = 1.
q dt (51)
Condition (59) (with = 1) was established for the first time in Ref.
and subscripts “mod” and “fs” stand for model and full-size, respec- [8] and shortly afterwords in Ref. [46]. The scaling in model testing of
tively. In the absence of heat transfer and if atmospheric conditions OWC converters was studied in detail in Refs. [18,30]. This rule was
remain unchanged, the Second Law of Thermodynamics determines implemented (possibly for the first time in OWC model testing) as part
that the specific entropy s in the chamber cannot be lower than sat in the of the design work of the Pico OWC plant, at scale = 1/35 at the
atmosphere. It is s = sat in the ideal case of a perfectly efficient turbine. National Civil Engineering Laboratory, Lisbon, Portugal [10] and at
For a real turbine and a given sea state, s oscillates about some value scale = 1/25 at the Hydraulics & Maritime Research Centre, University
above sat during inhalation and remains constant during exhalation (see College Cork, Ireland [11]. Experimental investigations of the com-
numerical results for a test case in Section 6). Here we neglect those pressibility effect with an additional air reservoir are reported in Ref.
oscillations in specific entropy and assume that the thermodynamic [47].
process in the chamber is approximately isentropic. If air is a perfect If pat,mod = pat,fs = pat , and if pmod = pfs as required by Froude
gas, Eq. (51) becomes scaling, then the second fraction on the right-hand-side of Eq. (53) may
490
be written as 6. Case study
pmod + pat,mod 6.1. OWC geometry and hydrodynamics

=1 ,
pfs + pat,fs (60)
The theory developed above is illustrated by a case study to more
where clearly show the influence of the air compressibility effect upon the
wave energy conversion process.
1 We consider a fixed-structure OWC wave energy converter (Fig. 5).
= pat
1+ pfs (61) The water depth is h. The submerged part of the OWC structure is of
vertical cylindrical geometry with radius a. In calm water, the sub-
is the error introduced by neglecting the pressure oscillation pfs in mergence is b. The thickness of the walls is assumed negligible.
comparison with the atmospheric pressure. This error may be sig- The dimensionless hydrodynamic coefficients G (radiation con-
nificant during inhalation in highly energetic sea states, especially in ductance) and H (radiation susceptance) are defined as
some types of OWC converters as is the case of the sparbuoy OWC
wg
device (see Ref. [48]). {G , H } = {G, H },
(65)
a2
In practice, condition (59) may be implemented by connecting the
air chamber of the model to a rigid-walled reservoir of appropriate where w is the water density, g is the acceleration of gravity and ω is
volume, as shown schematically in Fig. 4. The reservoir may be partly the wave radian frequency. For an axisymmetric OWC, the excitation
filled with water as a practical way of adjusting the volume of air, as flow rate coefficient is related to G by (see Ref. [35])
seen in the figure. The connecting duct should be such that the pressure 2D (kh ) G 1/2
2 wg
loss it introduces remains small compared with the amplitude of the = ,
oscillating pressure head available to the turbine. Let ε be the Froude k (66)
scale, V1 the volume of the model air chamber and V2 the volume of the
where k is the wavenumber and
additional air reservoir plus the connecting duct. For simplicity, the
density of water in the tank is assumed equal to the density of sea water. 2kh
D (kh) = 1 + tanh kh,
It should be (V1 + V2)/ Vfs = 2 , where Vfs is the chamber volume at full- sinh 2kh (67)
scale. In the absence of waves (denoted by the subscript 0), it is
which may be written in dimensionless form as
V0,1 = 3V0,fs and so V2 = ( 2 3) V . For simplicity, the model is sup-
0,fs
posed to be equipped with a linear turbine (or a linear turbine simu- 2 D (kh) G
lator) characterized by a constant coefficient = w/ p , where w is mass = ,
a kh (68)
flow rate and p is pressure head. Since volume V2 is invariant, the mass
flow rate w2 through the connecting duct is due to variations in air where = g 1/2a 3/2 and a = a /h . In the equations above, kh may
density, and hence in pressure, inside that volume. To evaluate w2 , the be expressed as kh = c a , where = a/ g is a dimensionless
pressure loss at the connecting duct is assumed negligible in a first wave frequency, and the dimensionless wave speed c = c / gh is the
approximation, to be confirmed subsequently in the calculation. The solution of the dispersion relationship
whole process is assumed approximately isentropic, i.e.
a
c = tanh .
1/ c a (69)
p
= at 1+ ,
pat (62) Equation (50) for the memory function may be written in dimensionless
form as
which is linearized as
1 p
= at 1+ .
pat (63)
The time-varying mass of air in volume V2 is m2 = V2 . The mass flow

rate of air through the connecting duct (positive for air entering the
additional reservoir) is w2 = V2 (d /dt ). This may be rewritten as
V2 dw
w2 = ,
cat2 dt (64)
where cat is the speed of sound at atmospheric temperature. In regular

waves, both w and w2 are sinusoidal functions of time, with phase
difference equal to /2. The pressure loss due to mass flow rate w2
through the duct, including the exit loss, will depend on w2 , duct geo-
metry and flow regine (or Reynolds number); Eq. (64) can be used to
check whether that pressure loss may be neglected or not compared
with the oscillation amplitude of the pressure head available to the
turbine.
In a fixed-structure OWC, the presence of the additional reservoir
and connecting duct is not expected to significantly affect the hydro-
dynamic similarity of the wave energy absorption process. Naturally
this may be untrue in the case of a floating OWC, because the con-
necting duct could significantly modify the dynamics of the floating Fig. 4. Schematic representation of an OWC model with additional air reservoir
device. Such a case is reported in Ref. [49]. that may be partly filled with water to adjust the volume of air.
491
power absorbed from the waves by 31%. Fig. 9 also shows that, for
T < 5.8 s, R exceeds unity and increases with chamber height Z.
It seems from Fig. 9 that all curves cross each other at a point of
ordinate equal to unity and abscissa T 5.8 s (that approximately co-
incides with the zero T = 5.80 s of the radiation susceptance function
H (T ) ). The conclusion that, for wave period T 5.8 s, the averaged
absorbed power P̄abs is (at least approximately) independent of chamber
height Z requires further investigation. Fig. 10 is an enlargement of
Fig. 9 for = ref and shows that the seven curves in fact do not in-
tersect each other at a single point, although the intersection points are
relatively close to each other. This shows that, in regular waves, the
sensitivity of the converter performance to the size of the air chamber is
minimal (but does not vanish) for wave period T close to the root of
H (T ) = 0 .
Fig. 5. Bottom-fixed cylindrical OWC converter.
6.3. Irregular waves and time domain results
2
h r (s ) = G ( )cos( s ) d ,
0 (70) The Pierson-Moskowitz variance density spectrum was adopted for
where deep water [50]
h r (s ) = 1 1h
(71)
SPM ( ) = 262.6Hs2 Te 4 5exp( 1052 Te 4 4)
(72)
wa r (s )
and s = (g / a)1/2s . (SI units), where Hs and Te are the significant wave height and the
Values of G and H for the geometry shown in Fig. 5 were ob- energy period, respectively. It is assumed here that the energy flux per
tained analytically by Evans and Porter [37]. In what follows, we as- unit wave crest length remains unchanged as the waves travel from
sume the special case a = h , (a = 1), and b = a/2 = h/2 . Functions deep water to the OWC converter location where the water depth is h
G ( ) and H ( ) for this geometry are plotted in Fig. 6. The memory (this assumes negligible dissipation and no wave breaking, as well as
function hr (s ) is plotted in Fig. 7. rectilinear wave crests parallel to lines of constant depth). The fre-
The dimensions a = h = 2b = 8 m were adopted for the full-sized quency-dependent shoaling coefficient is K sh ( ) = Aw / Aw, , where
OWC. The functions G ( ) and H ( ) were obtained directly from the subscript stands for deep water. It is (see Ref. [51])
curves for G ( ) and H ( ) in Fig. 6, together with the definitions in K sh ( ) = (cg, / cg )1/2 , where cg is the group velocity. The shoaling
Eq. (65) and the values w = 1025 kg m 3 for sea water density and coefficient K sh ( ) is plotted in Fig. 11 for water depth h = 8 m and
g = 9.8 m s 2 for the acceleration of gravity. The following numerical g = 9.8 m s 2 .
values were adopted: = 1.4 and R = 287 Jkg 1K 1 for air, The resulting spectrum S ( ) = SPM ( ) K sh 2
( ) was discretized into
pat = 1.013 × 105 Pa for atmospheric pressure and Tat = 288 K for at- 225 equally spaced ( = 0.01 rad/s ) sinusoidal harmonics in the range
mospheric absolute temperature.
6.2. Regular waves and frequency domain results
Regular waves and a linear turbine, together with the linearized

isentropic model, are assumed first, which allows the frequency domain
analysis to be employed. Equation (46) gives the time-averaged power
P̄avai available to the turbine (which is also the averaged power absorbed
from the OWC motion).
Curves of the averaged available power ratio, P¯avai/ A w2 , versus wave
period T are plotted in Fig. 8, for chamber height Z = V0/( a2 ) = 0 and
8 m (Z = 0 represents incompressible air), and turbine coefficient
= ref , where ref is a reference value that is taken equal to the op-
timum value opt , as given by Eq. (48) for wave period T = 8 s and
chamber height Z = 8 m. It can be found that, for the OWC geometry
adopted here, it is H > 0 for T > 5.80 s and H < 0 for T < 5.80 s . Fig. 8
shows that the air compressibility effect enhances the wave energy
absorption in the range of wave periods T for which H < 0 (i.e. for
T < 5.80 s ), the opposite being true for T > 5.80 s . This confirms the
finding in section 6.2 where it was shown that, in the range of wave
periods within which H < 0 , the compressibility effect may enhance the
wave energy absorption.
Curves of the averaged available power ratio
R = P¯avai (T , Z )/ P¯avai (T , 0) versus wave period T are plotted in Fig. 9, for
several values of the chamber height Z = V0/( a2 ) and three values of
the turbine coefficient = ref , = 0.5 ref and = 2 ref . Fig. 9 shows
that the ratio R decreases (and the compressibility effect increases)
with increasing chamber height Z (as should be expected), with in-
creasing wave period T (up to some point) and with decreasing turbine
coefficient . In the most extreme condition in Fig. 9 (for = 0.5 ref
Fig. 6. Curves of G and H versus for the OWC converter.
and Z = 10 m), it is R = 0.69 , i.e. the compressibility effect reduces the
492
Fig. 7. Curve of hr versus s for the OWC converter.
Fig. 8. Curves P¯avai/ A w2 versus wave period T, for chamber height

Z = V0/( a2) = 0 (representing incompressible air) and Z = 8 m , and turbine
coefficient ref .
0.1 6 2.24 rad/s. (The irrational number 6 ensures the non-

periodicity in the time series). The phases at t = 0 were made equal to Fig. 9. Curves of ratio R = P¯avai (Z )/ P¯avai (0) versus wave period T for various
random numbers in the interval (0,2 ) . values of chamber height Z = V0/( a2) and turbine coefficient = 0.5 ref ,
The volume of the air chamber in the absence of waves was taken = ref , = 2 ref . Z = 0 represents incompressible air.
equal to V0 = 1650 m3 , average height Z = V0/( a2 ) = 8.2 m.
The OWC converter is equipped with a Wells turbine without guide
vanes that was model-tested at the University of Siegen, Germany
[52,53], where the geometry is code-named rotor B, Fig. 12. The rotor
is a monoplane with five blades. As is usual in model testing of air
turbines for wave energy conversion, the experiments were performed
at Mach numbers not high enough for compressibility effects to be
significant; for this reason, the performance curves appear in terms of
the dimensionless flow rate and pressure head . The performance
curves used here were taken directly from Refs. [52,53] (see also [25]),
with a significant modification: the approximately linear relationship
( ) was replaced exactly by = 0.45 (i.e. K = 0.45). In the defini-
tions of , and , the reference air density 0,in was made equal to the
atmospheric air density during inhalation and to the chamber air den-
sity during exhalation. Fig. 13 shows the curves ( ) and ( ) . The
power and efficiency curves versus pressure head exhibit the sharp drop
due to aerodynamic stalling at the rotor blades that is characteristic of Fig. 10. Enlargement of part of Fig. 9.
Wells turbines in general. The dimensionless power loss
3D5) is given by .
L = L/( 0,in L = outside the scope of the present paper whose focus is on the compres-
The choice of the turbine size and of the rated power of the elec- sibility effect of the air in the chamber.
trical generator, together with the rotational speed control strategy, are A diameter D = 2.5 m was adopted for the turbine rotor. This is not
important issues in the converter design process [54]. Such choices very different from the sizes of the Wells turbines of the Pico plant
affect the amount of produced electrical energy and also the costs of (2.3 m) and the LIMPET (2.6 m) (see Ref. [6]). The rotational inertia is
equipment. Naturally, the levelized cost of energy (LCOE) is a major supposed to be large enough for the rotational speed to remain ap-
factor in the final decision (see Ref. [55]). Such detailed analyses are proximately constant during a given sea state.
493
Fig. 11. Shoaling coefficient K sh ( ) for water depth h = 8 m .
Fig. 12. Geometry of the Wells turbine rotor (from Refs. [52,53]).
Fig. 14. Specific entropy s sat , pressure oscillation p and air chamber volume
V versus time t, for non-isentropic model.
equal to 0.02 s . The interval of integration in the convolution integral of

Eq. (49) was taken equal to 30 s .
Results are shown for three different theoretical models: (i) non-
isentropic model, in which turbine losses are accounted for in the air
compressibility effects; (ii) isentropic model, in which turbine losses are
ignored; (iii) incompressible flow model, in which the air in the
chamber is assumed incompressible. In model (iii), the size of the air
chamber obviously does not affect the results.
Fig. 14 shows the time variation of the specific entropy s, pressure
oscillation p and air volume V for the non-isentropic theoretical model.
Initially, there is an ascending trend of the specific entropy, from initial
atmospheric conditions, followed by oscillations related to individual
waves and to wave groups. Fig. 15 is a magnification of part of Fig. 14.
It may be seen that s remains constant during exhalation, p > 0 , and
Fig. 13. Dimensionless representation of the performance curves of power ( ) may increase or decrease during inhalation, p < 0 .
and efficiency ( ) versus pressure head of the Wells turbine (from Refs. For a perfect gas, it is
[25,52,53]).
( 1)/
s sat T p + pat
= ln .
The numerical results shown here were computed for = 150 rad/s, cp Tat pat (73)
Hs = 3 m and Te = 8 s.
The integro-differential equations were numerically integrated in If we compare homologous points in Figs. 14 and 15, defined as points
the time domain from prescribed initial conditions, with a time step with p = 0 , Eq. (73) shows that the temperature T in the chamber
494
here, because the process is assumed adiabatic and isentropic without

any dissipative losses, the values of Pavai and Pabs averaged over a suf-
ficiently long time interval must be equal. The instantaneous difference
between Pavai and Pabs is related to the rate of change of internal energy
alternately stored in, and released from, the air in the chamber (see
Section 2.3).
The lower part of Fig. 17 shows the time variation of Pavai for the
incompressible and isentropic models. Obviously, the broken curves
(incompressible model) are identical in the upper and lower parts of
Fig. 17. For the isentropic model, the curves are quite different (unlike
Pavai , Pabs is negative during part of the time), but the values of Pavai and
Pabs averaged over a long time interval are equal (see above). The lower
part of Fig. 17 clearly shows that, in the case analysed here, the spring-
like effect of air compressibility (accounted for by the isentropic model)
results in less absorbed energy from the waves, and also in less energy
available to the turbine. It should be pointed out that there may be
special situations (see Section 6.2) in which the compressibility effect
increases the amount of energy absorbed from the waves.
A comparison between the results from the isentropic and the non-
isentropic models is illustrated in Fig. 18. The upper part of the figure
compares the increase in temperature T Ts due to turbine losses with
the oscillations T Tat in temperature with respect to atmospheric
temperature. The first value is somewhat smaller than the amplitude of
the oscillations of T Tat , but is not negligible. Obviously, it would be
T Ts = 0 for the isentropic model.
The power absorbed from the waves depends only on the pressure
oscillation p (t ) . The middle part of Fig. 18 shows that the difference
p pis (between the pressure oscillation p given by the non-isentropic
Fig. 15. Detail of Fig. 14.
model and the corresponding value pis for the isentropic model) is (in
terms of oscillation amplitude) about 1% of p.
A similar comparison, in the bottom part of Fig. 18, for the power
increases exponentially with the specific entropy s: an ascending or Pavai available to the turbine, reveals that the differences between the
descending trend in s will be accompanied by a similar trend for the two models are also of the order of magnitude of 1%. Care should be
temperature T. The figures indicate that the “average” temperature of taken when analysing and making use of these results. In the non-
air in the chamber (i.e. temperature for conditions p = 0 ) will be higher isentropic model, Pavai (t ) is the power output (aerodynamic torque times
than atmospheric temperature. The curve s sat versus time t in Fig. 14,
together with Eq. (73), gives for this “average” temperature difference
T Tat a value about 2 °C.
Fig. 16 shows curves of pressure oscillation p and flow rate q dis-
placed by the OWC free-surface motion, versus time t, for the isentropic
and the incompressible models. The spring-like effect of air compres-
sibility introduces a phase shift between pressure and flow rate, which
does not occur if the air is assumed incompressible. It also produces a
substantial increase in the amplitude of the flow rate q. The latter effect
may be important, since an increase in flow rate amplitude q is expected
to aggravate the viscous losses in water, especially by vortex shedding
at the submerged edge of the cylindrical structure, and hence to ne-
gatively affect the hydrodynamic efficiency of the wave-to-pneumatic
energy conversion. Such losses are not accounted for in the present
model based on linear water wave theory.
The power absorbed from the motion of the OWC free surface (in-
deed the power absorbed from the waves in our analysis based on linear
water wave theory) is Pabs = pq in all three models. The time variation
of Pabs is represented in the upper part of Fig. 17 for the isentropic and
incompressible models. The phase shift between p and q, visible in
Fig. 16, results in Pabs being alternately positive and negative in the
isentropic model; this occurrence of reactive power does not happen in
the incompressible model.
The instantaneous power Pavai available to the turbine is defined as
the power output of a perfectly efficient turbine, with mass-flow rate w,
operating isentropically: (i) with inlet stagnation conditions pat , Tat , and
exit pressure p + pat , during inhalation (p < 0) , and (ii) with inlet
stagnation conditions p + pat , T and exit pressure pat , during exhalation
(p > 0) (see Fig. 3). Obviously it is Pavai = Pabs for the incompressible Fig. 16. Curves of pressure oscillation p (solid lines) and flow rate q displaced
model. by the OWC free-surface motion (broken lines), versus time t. Above: isentropic
In the case of the isentropic model, it is in general Pavai Pabs . But model results. Below: results from incompressible air model.
495
Fig. 17. Power Pabs absorbed from the OWC (above) and power Pavai available to
the turbine (below). Solid lines: isentropic model; broken lines: incompressible
model.
rotational speed) of the real turbine plus the power loss L (t ) . In the
isentropic model (perfectly efficient turbine), the two values – available
power and power output – are equal. In a strict analysis, it is thermo-
dynamically inconsistent to obtain the power output as the difference
Pavai,is (t ) L (t ) , between the power available to the perfectly efficient
turbine and the power lost in the real turbine. However, since the dif-
ference between the values, given for the available power by the two
models, is relatively small, this procedure may be acceptable for most Fig. 18. Comparisons between isentropic and non-isentropic model results
practical purposes. This agrees with the findings in Ref. [9]. versus time t. Top: T Ts (solid line) and T Tat for non-isentropic model
The value, P̄abs , of the power absorbed from the OWC free surface (broken line). Middle: (p pis ) × 100 (solid line) and p (broken line). Bottom:
motion, averaged over the same time interval of 980 s, is as follows: (Pavai Pavai,is ) × 100 (solid line) and Pavai,is (broken line). Subscript “is” denotes
80.2 kW for the isentropic model, 80.2 kW for the non-isentropic model isentropic.
and 103.4 kW for the incompressible model. The difference between the
first two values is negligible, which confirms the findings in Ref. [9]. pat , which may be untrue in the more energetic sea states (see Section
The value given by the incompressible model substantially exceeds the 6.3).
other two by a factor 1.29. In time-average, the power available to the This point is addressed here by means of a comparison based on the
turbine is equal to the power absorbed from the OWC motion in the isentropic model as described above in Section 6.3. The results for the
isentropic and incompressible models (see Section 2.3). This is not true wave power absorbed by the full-scale device are compared with results
for the non-isentropic model: here the power P̄avai available to the tur- for a model at scale = 1/25, with amod = a for the radius of the OWC,
bine was found to be 79.6 kW, 0.75% less than the power P̄abs = 80.2 kW Hs,mod = Hs for the significant wave heigh and Te,mod = 1/2Te for the
absorbed from the OWC. It should be noted that these are not values of energy period, as required by Froude similarity. In addition, it is
the turbine output, which is substantially smaller due to turbine losses. V0,mod = 1 2V0 for the chamber volume (see Eq. (59)). The model is
assumed to be equipped with a smaller, but otherwise geometrically
similar, turbine rotating at a higher velocity, with Dmod = 1/4 D and
6.4. Model testing
mod =
3/4 1/2 (see Eq. (58)). Both prototype and model are subject
to the same atmospheric pressure, with the same atmospheric tem-
It was seen in Section 5 that, if an OWC wave energy converter is to
perature and density ( = 1).
be model tested at Froude scale < 1 (including the air chamber vo-
The results for the power absorbed from the OWC motion, averaged
lume), the correct reproduction of the compressibility effect requires
over corresponding time intervals (980 s at full size and 196 s at model
the atmospheric pressure to be multiplied by , where = w,mod / w,fs
scale), are as follows: P̄abs,fs = 79.93 kW for the full-sized converter and
is the water density ratio (see Eq. (55)). This is obviously unpractical.
P̄abs,mod = 0.9990 kW for the model. Taking into account that the Froude
An alternative, although less rigorous, approach consists in multi-
scale for power is 7/2 (see e.g Refs. [18,45]), the following value is
plying the volume V0 of the air chamber by the factor 1 2 (rather than
obtained for the model power converted into full-scale:
3
) (see Eq. (59)). This is valid as far as, in the prototype, the pressure
abs,mod = 80.00 kW, which differs by about 0.1% from the value
1 7/2P̄
oscillations p are negligible compared with the atmospheric pressure
496
even in the case of a far from perfectly efficient Wells turbine. The
difference between the power available to the turbine as given by the
isentropic model and the non-isentropic model was found to be quite
small (of the order of 1% ). This justifies the practice of computing the
turbine power output as the power available to the turbine, as given by
the isentropic model, minus the power losses in the real turbine.
In the case of physical model testing at Froude scale < 1 (for the
part of the OWC structure subject to wave action), theoretical con-
siderations give for the air chamber volume ratio (in the absence of
waves) the value V0,mod/ V0,fs = 1 2 , where = w,mod / w,fs is the tank
water to sea water density ratio. If this rule is kept, then it was found
that model testing in wave tank may provide a good simulation of the
air compressibility effects.
In physical model testing, the turbine is almost always simulated
either by the pressure drop across an orifice (in most cases assumed a
Fig. 19. Absorbed power Pabs versus time t. Solid line: full scale; broken line: quadratic function of the flow rate and used to simulate a self-rectifying
model at scale = 1/25 converted to full scale.
air turbine of impulse type) or more rarely by the pressure drop across a
porous plug (assumed to simulate a linear Wells turbine). Such devices
79.93 kW directly computed for the full-scale converter. This difference do not provide better than a rough (or even very rough) approximation
is much smaller than the experimental error and than the error in- to the damping introduced by a full-sized air turbine operating and
troduced by the imperfect simulation of the turbine by an orifice or a controlled under real conditions. Such errors introduced into the energy
porous plug, which shows that the rule V0,mod = 1 2V0 is adequate for conversion chain are likely to be considerably larger than the errors due
model testing. to imperfect simulation of the air compressibility effect if the appro-
Fig. 19 shows a comparison between curves for the power Pabs,fs priate value is adopted for the volume of the model air chamber.
absorbed by the prototype and the power Pabs,mod 1 7/2 absorbed by It is well known that, in model testing in wave tank, Reynolds
the model converted into full-scale. The differences are quite small. number equality cannot in practice be achieved. This is especially re-
They are the consequence of the non-linearity of the equations gov- levant in very small scale testing, where real fluid (viscous) effects in
erning the aerodynamic process. They would vanish if the frequency water may be over-evaluated. Since the hydrodynamic model adopted
domain analysis (based on linearization) had been adopted instead (see here is based on linear water wave theory, such effect could not be
Eqs (43) and (46)). taken into consideration.
7. Conclusions Acknowledgements
The effect of air compressibility in the air chamber of an OWC This work was partly funded by the Portuguese Foundation for
converter is important at full scale, but has been frequently disregarded Science and Technology (FCT) through IDMEC, under LAETA, project
in theoretical/numerical modelling and almost always ignored in phy- UID/EMS/50022/2019. Author JCCH was supported by FCT researcher
sical model testing at reduced scale. Such effect is in general detri- grant No. IF/01457/2014.
mental to the wave energy absorption and increases with the volume of
the chamber; only in special cases it may be beneficial. In regular References
waves, the sensitivity of the fixed-structure converter performance to
changes in air chamber volume is minimum (but does not vanish) for [1] Mackay EBL. Resource assesment for wave energy. In: Sayigh A, editor.
Comprehensive renewable energy. Ocean energy, vol. 8. Oxford: Elsevier; 2012. p.
frequencies ω close to the zero of the radiation susceptance H ( ) .
11–77.
The effect of air compressibility in the chamber of an OWC con- [2] Reguero BG, Losada IJ, Méndez FJ. A global wave power resource and its seasonal,
verter upon the conversion efficiency was theoretically studied here interannual and long-term variability. Appl Energy 2015;148:366–80.
from linear water wave theory and basic aero-thermodynamics. It is [3] Zhang CW, Wang Q, Li CY. An overview of medium- to long-term predictions of
global wave energy resources. Renew Sustain Energy Rev 2017;79:1492–502.
acknowledged that losses due to real fluid effects in water were not [4] Falcão AFO. Wave energy utilization: a review of the technologies. Renew Sustain
accounted for. Such effects may be important, especially in energetic Energy Rev 2010;14:899–918.
sea states, but this was not the primary object of the study. Another [5] Heath TV. A review of oscillating water columns. Phil Trans Royal Soc A Math Phys
Eng Sci 2012;370:235–45.
important simplifying assumption is the spatial uniformity of the [6] Falcão AFO, Henriques JCC. Oscillating-water-column wave energy converters and
thermodynamic conditions in the chamber. The lack of information on air turbines: a review. Renew Energy 2016;85:1391–424.
the Mach number dependence of the performance of air turbines is a [7] Delmonte N, Barater D, Giuliani F, Cova P. Review of oscillating water column
converters. IEEE Trans Ind Appl 2016;52(2):1698–710.
limitation to the study of the air chamber compressibility effects. [8] Sarmento AJNA, Falcão A F de O. Wave generation by an oscillating surface-pres-
Numerical results were obtained for a large air chamber volume and sure and its application in wave-energy extraction. J Fluid Mech 1985;150:467–85.
an energetic sea state; this choice was made to emphasize the effects of [9] Falcão A F de O, Justino PAP. OWC wave energy devices with air flow control.
Ocean Eng 1999;26:1275–95.
the air compressibility.
[10] Sarmento AJNA. Model-test optimization of an OWC wave power plant. Int J
The results show that the air incompressibility assumption in gen- Offshore Polar Eng 1993;3:66–72.
eral leads to substantially unrealistic overvaluation of the converter [11] Holmes B, Hurley J, Lewis T, Sarmento AJNA, Brito-Melo A, Falcão A F de O. 1:25
performance. This kind of unrealism occurs frequently in model testing scale testing of the Azores pilot plant. Proc. 2nd European wave power conference,
Lisbon, Portugal. 1995. p. 148–55.
of OWC converters in wave tank in which the air chamber is reproduced [12] Viviano A, Naty S, Foti E. Scale effects in physical modelling of a generalized OWC.
at the same Froude scale as the submerged parts of the device. It also Ocean Eng 2018;162:248–68.
occurs in many cases of theoretical/numerical modelling in the litera- [13] Sheng W, Alcorn R, Lewis A. On thermodynamics in the primary power conversion
of oscillating water column wave energy converters. J Renew Sustain Energy
ture. 2013;5:023105.
The effect of the turbine imperfect efficiency may raise the tem- [14] Sheng W, Lewis A. Wave energy conversion of oscillating water column devices
perature in the chamber by a few degrees Celsius in comparison with including air compressibility. J Renew Sustain Energy 2016;8. No. 054501.
[15] Elhanafi A, Macfarlane G, Fleming A, Leong Z. Scaling and air compressibility ef-
the isentropic model. However, such effect upon the averaged power fects on a three-dimensional offshore stationary OWC wave energy converter. Appl
absorbed from the OWC motion was found to be practically negligible,
497
Energy 2017;189:1–20. [36] Evans DV, Porter R. Hydrodynamic characteristics of an oscillating water column
[16] Dimakopoulos AS, Cooker MJ, Bruce T. The influence of scale on the air flow and device. Appl Ocean Res 1995;17:155–64.
pressure in the modelling of oscillating water column wave energy converters. Int J [37] Evans DV, Porter R. Efficient calculation of hydrodynamic properties of OWC-type
Marine Energy 2017;19:272–91. devices. J Offshore Mech Arctic Eng Trans ASME 1997;119:210–8.
[17] Simonetti I, Cappietti L, Elsafti H, Oumeraci H. Evaluation of air compressibility [38] Martin-Rivas H, Mei CC. Wave power extraction from an oscillating water column
effects on the performance of fixed OWC wave energy converters using CFD mod- along a straight coast. Ocean Eng 2009;36:426–33.
elling. Renew Energy 2018;119:741–53. [39] Martin-Rivas H, Mei CC. Wave power extraction from an oscillating water column
[18] Falcão AFO, Henriques JCC. Model-prototype similarity of oscillating-water-column at the tip of a breakwater. J Fluid Mech 2009;626:395–414.
wave energy converters. Int J Marine Energy 2014;6:18–34. [40] Lovas S, Mei CC, Liu Y. Oscillating water column at a coastal corner for wave power
[19] Kestin J. A course in thermodynamics. Waltham: Blaisdell; 1966. extraction. Appl Ocean Res 2010;32:267–82.
[20] Deymi-Dashtebayaz M, Farzaneh-Gord M, Nooralipoor N, Rastgar S. The full si- [41] Rezanejad K, Bhattacharjee J, Soares CG. Stepped sea bottom effects on the effi-
mulation of rapid refuelling of a natural gas vehicle on-board cylinder. J Nat Gas Sci ciency of nearshore oscillating water column device. Ocean Eng 2013;70:25–38.
Eng 2014;21:1099–106. [42] Rezanejad K, Bhattacharjee J, Soares CG. Analytical and numerical study of dual-
[21] Deymi-Dashtebayaz M, Farzaneh-Gord M, Nooralipoor N, Niazmand H. The com- chamber oscillating water columns on stepped bottom. Renew Energy
plete modelling of the filling process of hydrogen on board vehicle cylinders. Braz J 2015;75:272–82.
Chem Eng 2016;33:391–9. [43] Luo YY, Nader JR, Cooper P, Zhu SP. Numerical simulation of an oscillating water
[22] Dixon SL, Hall CA. Fluid mechanics and thermodynamics of turbomachinery. sixth column device using a code based on Navier-Stokes equations. Renew Energy
ed. Amsterdam: Elsevier; 2010. 2013;64:255–65.
[23] Moisel C, Carolus TH. A facility for testing the aerodynamic and acoustic perfor- [44] Teixeira PR, Davyt DP, Didier E, Ramalhais R. Numerical simulation of an oscil-
mance of bidirectional air turbines for ocean wave energy conversion. Renew lating water column device using a code based on Navier-Stokes equations. Energy
Energy 2016;86:1340–52. 2013;61:513–30.
[24] Moisel C, Carolus TH. Novel mixed flow air-turbine for bidirectional operation in [45] Heller V. Development of wave devices from initial conception to commercial de-
OWC wave energy converters. Proc. 11th European wave tidal energy conference, monstration. In: Sayigh AA, editor. Comprehensive renewable energy. Ocean en-
Nantes, France. 2015. ergy, vol. 8. Oxford: Elsevier; 2012. p. 79–110.
[25] Falcão AFO, Henriques JCC, Gato LMC. Self-rectifying air turbines for wave energy [46] Jefferys R, Whittaker T. Latching control of an oscillating water column device with
conversion: a comparative analysis. Renew Sustain Energy Rev 2018;91:1231–41. air compressibility. In: Falcão A F de O, Evans DV, editors. Hydrodynamics of ocean
[26] Takao M, Itakura K, Setoguchi T, Kim TH, Kaneko K, Thakker A. Performance of a wave energy utilization. Berlin: Springer; 1986. p. 281–91.
radial turbine for wave power conversion. Proc. International offshore and polar [47] Howe D, Nader JR, Macfarlane G. Experimental analysis of the effects of air com-
engineering conference, Kitakyushu, Japan. 2002. p. 562–7. pressibility in OWC model testing. Proc. Asian wave and tidal energy conference,
[27] Lopes BS, Gato LMC, Falcão AFO, Henriques JCC. Test results of a novel twin-rotor Taipei, Taiwan. 2018.
radial inflow self-rectifying air turbine for OWC wave energy converters. Energy [48] Falcão AFO, Henriques JCC, Gato LMC, Gomes RPF. Air turbine choice and opti-
2019;170:869–79. mization for floating oscillating-water-column wave energy converter. Ocean Eng
[28] Medina-López E, Moñino-Ferrando A, Clavero-Gilabert M, del Pino C. Note on a real 2014;75:148–56.
gas model for OWC performance. Renew Energy 2016;85:588–97. [49] Benreguig P, Vicente M, Dunne A, Murphy J. Modelling approaches of a closed-
[29] Medina-López E, Moñino A, Clavero M, Borthwick AGL. Thermodynamics of an circuit OWC wave energy converter. J Mar Sci Eng 2019;7(2):23.
OWC containing real gas. Energy 2017;85:709–17. [50] Goda Y. Random seas and design of maritime structures. Singapore: World
[30] Weber J. Representation of non-linear aero-thermodynamic effects during small Scientific; 2000.
scale physical modelling of OWC WECs. Proc. 7th European wave and tidal energy [51] Holthuijsen LH. Waves in oceanic and coastal waters. Cambridge: Cambridge
conference, Porto, Portugal. 2007. University Press; 2007.
[31] Evans DV. Theory for wave-power absorption by oscillating bodies. J Fluid Mech [52] Starzmann R. Aero-acoustic analysis of Wells turbines for ocean wave energy con-
1976;77:1–25. version Doctoral thesis, Universität Siegen Germany: VDI Verlag, Düsseldorf; 2012
[32] Evans DV. Oscillating water column wave-energy device. J Inst Math Appl [53] Starzmann R, Carolus T. Model-based selection of full-scale Wells turbines for ocean
1978;22:423–33. wave energy conversion and prediction of their aerodynamic and acoustic perfor-
[33] Gomes RPF, Henriques JCC, Gato LMC, Falcão AFO. Hydrodynamic optimization of mances. Proc IME Part A J Power Energy 2014;228:2–16.
an axisymmetric floating oscillating water column for wave energy conversion. [54] Falcão AFO, Henriques JCC, Gato LMC. Rotational speed control and electrical
Renew Energy 2012;44:328–39. rated power of an oscillating-water-column wave energy converter. Energy
[34] Evans DV. Wave-power absorption by systems of oscillating surface pressure dis- 2017;31:253–61.
tributions. J Fluid Mech 1982;114:481–99. [55] Falcão A F de O. Stochastic modelling in wave power-equipment optimization:
[35] Falnes J. Ocean waves and oscillating systems. Cambridge: Cambridge University maximum energy production versus maximum profit. Ocean Eng 2004;31:1407–21.
Press; 2002.
498

The Spring-Like Air Compressibility Effect in Oscillating-Water-Column Wave Energy Converters - Review and Analyses

Uploaded by

Copyright:

Available Formats

The Spring-Like Air Compressibility Effect in Oscillating-Water-Column Wave Energy Converters - Review and Analyses

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Spring-Like Air Compressibility Effect in Oscillating-Water-Column Wave Energy Converters - Review and Analyses

Uploaded by

Copyright:

Available Formats

Renewable and Sustainable Energy Reviews 112 (2019) 483–498

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

The spring-like air compressibility effect in oscillating-water-column wave T

ARTICLE INFO ABSTRACT

1. Introduction large enough to accommodate tidal oscillations and to avoid ingestion

Nomenclature v air velocity

The equation of continuity of mass is

where ¯ (t ) = m (t )/ V (t ) is a spatially averaged density of air in the

dU 1 Fig. 3. Entropy-enthalpy diagram for the turbine.

d(U + Ut + Et,kin ) 0,in D2 D

2.4. Approximations , = function( , Re, Ma, ). (31)

The effects of air compressibility are important in both fixed- P= +i Q,

be written as 6. Case study

pmod + pat,mod 6.1. OWC geometry and hydrodynamics

The time-varying mass of air in volume V2 is m2 = V2 . The mass flow

where cat is the speed of sound at atmospheric temperature. In regular

6.2. Regular waves and frequency domain results

Regular waves and a linear turbine, together with the linearized

Fig. 7. Curve of hr versus s for the OWC converter.

Fig. 8. Curves P¯avai/ A w2 versus wave period T, for chamber height

0.1 6 2.24 rad/s. (The irrational number 6 ensures the non-

Fig. 11. Shoaling coefficient K sh ( ) for water depth h = 8 m .

equal to 0.02 s . The interval of integration in the convolution integral of

here, because the process is assumed adiabatic and isentropic without

You might also like