Renewable Energy 88 (2016) 220e235

Contents lists available at ScienceDirect

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

Analysis of the SEA-OWC-Clam wave energy device e Part A:

Historical development, hydrodynamic and motion response
formulations & solutions
Norman W. Bellamy a, Andrea Bucchi b, Grant E. Hearn c, *
a
b
c

Emeritus Professor Coventry University, UK

School of Engineering, University of Portsmouth, UK
Fluid Structure Interaction Research Group, Faculty of Engineering and The Environment, University of Southampton, UK

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 10 January 2015
Received in revised form
8 October 2015
Accepted 8 November 2015
Available online xxx

The SEA-Clam wave energy device has undergone the metamorphic changes presented in the historical
development. Each variation leads to more complex hydrodynamic and motion responses. The most
recent novel Oscillating Water Column (OWC) form of the device has multiple internal free surfaces, the
motions of which are required to predict power take-off once a device specication with structural
integrity has been identied. Motion equations for structure and the large number of degrees-of-freedom
of the free surfaces modelled as massless plates are derived and presented in a very compact form. This
paper also proffers two simpler models consistent with standard offshore engineering calculations. These
are investigated within since structural integrity of device requires further renement; as demonstrated
in the companion paper. Quality checks of the hydrodynamic analyses are explained and applied to
justify the numerical investigations undertaken. Sea spectra for the possible operational site of South Uist
are used to generate motion transfer functions for associated wave frequency range. The peak pitch
response of this large annular shaped structure is a main concern regarding survivability. The analyses
undertaken reect conceptual rather than detailed design status of the device.
2015 Published by Elsevier Ltd.

Keywords:
Multiple degrees-of-freedom
Free-surface inuences
Hydrodynamic quality checks
Negative added mass
Fredholm integral equations
Boundary elements

1. Introduction

1.1. Historical development of the SEA-Clam device

Assessment of the OWC form of the SEA-Clam wave energy

device is by denition complex. Thorough explanation of the analyses undertaken and their justication necessitates the research to
be presented as two companion papers. Part A provides: historical
development of the device, an explanation of the determination of
the hydrodynamic characteristics, a full explanation of a complete
motion response formulation, choices regarding device modelling
and the presentation of hydrodynamic loads and the resulting
transfer functions. Part B addresses: dynamic structural loads and
resulting stress distribution calculations with identication of engineering hot-spots and device changes necessary for structural
integrity.

The SEA-Clam device has undergone a number of metamorphisms. Initially it had a long (275 m) terminator form
(1979e1984) with pressurised exible air bags facing the incident
waves. A one-tenth model in Loch Ness is illustrated in Fig. 1a.
Stability of device and wave direction independence led to the ring
form of Fig. 1b with the pressurised exible bags on the outer face
(1984e1992). Standard shear force and bending moment calculations can be readily applied to the long form of the device [1], but a
special analysis had to be developed for the ring form [2]. Various
specic design improvements took place in the period 1992e2008.
The device is essentially a regular dodecagon as illustrated in
Fig. 2a with one airbag per section. The initial 60 m outer diameter
of ring was increased to 80 m in 2008. Further development led to
the hybrid clam in which some of the ballast space in each of the
twelve sections was replaced by an N-shaped channel form of
oscillating water column (OWC) with an opening at the base of the
structure; compare Fig. 2b and c. That is, the outer surface extracted
energy through pressurised airbags, whereas the inner section

* Corresponding author. Faculty of Engineering & The Environment, University of

Southampton, Boldrewood, Southampton SO16 7QF, UK.
E-mail address: g.e.hearn@soton.ac.uk (G.E. Hearn).
http://dx.doi.org/10.1016/j.renene.2015.11.032
0960-1481/ 2015 Published by Elsevier Ltd.

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

221

Fig. 1. a Straight Clam Loch Ness tests (1982). b Circular Clam Loch Ness tests (1984e1992).

operated as an OWC. The next step was to replace each outer

pressurised bag with the J-shaped OWC channel of Fig. 2d. In each
case the draught is 6 m. The hybrid clam and the SEA-OWC-Clam
[3] exploit distinct methods of energy extraction. Within the bag of
the hybrid form, and the J-shaped channel of the OWC form,
extraction of energy is attributed to the surge degree-of-freedom,
whereas in both device forms the N-shaped OWC exploits the
heave degree-of-freedom. The natural surge and heave periods of
the J and N OWC are respectively 8 s and 12 s [4]. These periods are
quite distinct from overall structural natural periods, as predictions
in Section 4 will illustrate.
Whilst device efciency is not to be predicted in this paper, Fig. 3
[5] provides early estimates of individual capture efciency for each
OWC absorber and the combined capture efciency of the twin

OWC Clam. Essentially Fig. 3 is based on a combination of insights

deduced from Loch Ness testing of original SEA circular clam [6e8],
2D laboratory testing of the SEA-OWC-Clam partially addressed in
Ref. [9], and related behaviour of oscillating water columns in the
National Engineering Laboratory (NEL) [10]. Furthermore, it is
claimed in Refs. [11], that the surge OWC and heave OWC capture
powers are independent and have thus be added together. This
independence might not be observed in the three dimensional
form of the device which will be subjected to multi-directional seas
in contrast to ume tank unidirectional waves.
The capture efciencies of Fig. 3 are based on tabulated values of
South Uist signicant wave heights, crossing period, associated
power and sea state weightings provided in Ref. [12]. However,
detailed calculations leading to the claimed average efciency of

Fig. 2. a Plan view of SEA-OWC-Clam. b Transverse section of Circular Clam (1992). c Transverse section of Hybrid Clam (1994). d Transverse section of SEA-OWC-Clam (2008).

222

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

Capture Efficiency. (%)

Capture Efficiency v Wave Energy Period, Te

(s)

pressure loads must be determined.

120

2.1. Environmental representation

100
80

Surge OWC

60

Heave OWC
OWC Clam

40
20
0
1

9 10 11 12 13 14 15 16

Energy Period - Te (s)

Fig. 3. Predicted capture efciency of 2008 SEA-OWC-Clam.

59.4% and an associated annual average power capture of 1.54 MW

per device are not provided. Ref. [11] further claims that the earlier
hybrid 80 m steel and concrete devices only exhibited an average
power capture of 1.136 MW. These historical estimated changes in
efciency are used by Bellamy [9] to highlight evolution of the
device.
1.2. Paper structure
In Section 2 the basics regarding selected environment, hydrodynamic modelling and quality of associated predicted characteristics via Haskind [13,14] and the NewmaneTimman [15]
relationships are discussed. Thereafter, analytic solution of the
formulated motion responses and the procedure for determining
the resulting dynamic pressure distributions are presented.
Modelling of closed and open device forms with derivation of
their hydrostatic and inertial characteristics are provided in Section
3. Section 4 presents predictions and discussion of excitation and
radiation loads, and the motion responses in regular waves. Section
5 summarises principal observations made and establishes appropriateness of reported analysis for subsequent structural integrity
investigation [16].
2. Environmental, hydrodynamic, motion responses and
dynamic pressure modelling
Each of these distinct modelling aspects is addressed in turn.
Appreciation of possible operational environment assists in identication of appropriate wave frequencies, whereas motion equation formulation depends on number and form of degrees of
freedom. For ultimate structural integrity the resulting dynamic

Regional wind generated wave data, in terms of signicant

wave height and crossing period, is captured in seasonal scatter
diagrams. These diagrams also provide an indication of the
probability of such wave parameter combinations. Sea spectra,
based on the cited combinations of crossing period and signicant
wave height, indicate the distribution of wave energy with wave
frequency.
In classical waveestructure interaction, the incident waves are
considered to be regular and harmonic with wave direction and
wave frequency specied. The product of the square of the transfer
function, H(u,q) of the calculated (measured) structural quantity of
engineering interest (a motion, a stress et cetera) and the sea
spectra S(u,q) provide the corresponding response spectra R(u,q).
In this case a transfer function H(u,q) is essentially independent
of the wave direction q, due to the axisymmetric geometry of the
device.
Mathematically the usual expression for response spectrum will
simplify as shown in Eq. (1):

Ru; q Su; qHu; q2 Suf qHu2 :

(1)

The wave data collected to generate the spectra are usually

considered unidirectional and wave direction dependence can be
accommodated via an appropriate spreading function, f(q), which
may have a simple cos2q dependence or the more complicated
Mitsuyasu et al. form [17]. In the case of the investigation of scaled
models of the Clam in Loch Ness, special wave direction monitoring
systems were developed at British Ship Research Association
(BSRA) to assist assessment of efciency of operation in the prevailing wave conditions [18].
With South Uist specied as a likely installation site [12] available numerical data [19] has been used to generate the monthly
variation of sea spectra at this site, see Fig. 4. Fully developed sea
states are usually modelled mathematically using a Pierson-Moskowitz one parameter formulation [20] with more identication
arguments presented in Ref. [21]. Two parameter formulations
echo the Bretschneider approach [22].
Water depth of 50 m is assumed for South Uist. To ensure
motion responses are available for the associated spectra, the
individual wave frequencies addressed in the hydrodynamic
analysis are presented in Table 1 together with an indication of
the water depth treatment in the associated hydrodynamic
analysis. The kernel of the hydrodynamic integral equations
solved must reect deep water or nite depth water as appropriate. The software used automatically identies water depth
conditions and undertakes the appropriate required kernel
(Green function) evaluations.
2.2. Hydrodynamic modelling

Fig. 4. Environmental characteristics captured as spectral density versus wave frequency (rad/s).

Water is assumed to be inviscid and incompressible and its ow

is considered irrotational. The governing elliptic partial differential
equation (pde) is Laplacian. Consequently boundary conditions
must be applied over a closed boundary [23]. These boundary
conditions express continuity of velocity and pressure across the
free surface, impermeability of the seabed and the structure(s)
being analysed. At large distances from the structure, the Sommerfeld radiation boundary condition [14,23] ensures the waves
attenuate as they travel away from the structure, since waves may
not be reected at innity.
Automatic satisfaction of the governing equation and all

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

223

Table 1
Wave characteristics and indication of hydrodynamic analysis performed.
Frequency (rad./s)

Depth

0.100
0.200
0.300
0.400
0.465
0.500
0.522
0.550
0.620
0.645
0.669
0.717
0.750
0.798
0.870
0.878

Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite
Finite

depth
depth
depth
depth
depth
depth
depth
depth
depth
depth
depth
depth
depth
depth
depth
depth

Wavelength (m)

Frequency (rad./s)

Depth

Wavelength (m)

1379.72
672.07
428.23
300.50
244.52
219.81
205.81
189.49
154.89
144.39
135.11
118.70
108.90
96.51
81.36
79.90

0.878
0.930
0.975
1.000
1.020
1.035
1.055
1.068
1.085
1.100
1.125
1.150
1.240
1.375
1.500
1.760

Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Finite depth
Deep Water
Deep Water
Deep Water
Deep Water

79.90
71.25
64.83
61.63
59.24
57.54
55.38
54.05
52.36
50.94
48.70
46.61
40.09
32.61
27.39
19.90

Fig. 5. Reference system.

boundary conditions, except the structural impermeability

requirement, can be satised by appealing to Green identities [24]
to transform the pde formulation into a Fredholm integral equation,
which is solved using a boundary element formulation. The
mathematical forms of the kernels used with the integral equation
formulation are computationally efcient [25]. The deep-water
mathematically equivalent forms of Green function were published some years ago [25], whereas the efcient nite water depth
forms, also used in this study, remain essentially unpublished.
The boundary element method (BEM) provides the unknown
radiation and diffraction velocity potentials on the wetted surface
of the structure(s). That is, the resulting Fredholm integral equation
[14,23] method uses the minimum solution domain required to
provide the reactive and wave excitation forces and moments
determined from the associated wetted surface pressure distributions. The hydrodynamic simulations were performed using the
MATTHEW software [26] designed to analyse an arbitrary
arrangement of arbitrarily shaped structures undergoing arbitrary
harmonic rigid body motions.
The structures may be distinct structures, interacting with one
another, or, articulated structures or a combination of both. The
rigid body motions are those indicated in Fig. 5. The translational
motions of surge, sway and heave are positive along the positive
coordinate axes, with right-hand rule rotations in Fig. 5 dening
positive roll, pitch and yaw.
Mathematical analysis of the uid structure interaction (FSI) of
the SEA-OWC Clam can be readily achieved by treating each distinct
internal free surface as a single moveable mass-less plate. This
concept was applied previously by Hearn and Katory [27] in a 2D

analysis of an OWC device and by Hearn et al. [28] in the 3D analysis

of a damaged ship with multiple internal free-surfaces.
The complexity of the associated FSI of the OWC clam device is
a function of the number of distinct degrees-of-freedom associated with each OWC free surface and the structure itself. Since
potential theory only acknowledges continuity of normal velocity
across any boundary surface, it makes no sense to consider massless plate degrees-of-freedom tangential to the undisturbed internal free surface; namely the surge, sway and yaw motions
dened in Fig. 5.
Each free surface has three distinct vertical plane motions of
heave, roll and pitch and hence there are 72 degrees-of-freedom
associated with the 24 internal free surfaces. This and the inclusion of the usual 6 rigid-body degrees-of-freedom of the actual
Clam structure would necessitate formulation and solution of 78
coupled equations of motion; once all the associated hydrodynamic
reactive (radiation) and wave excitation loads have been determined. Recognition of the existence of two anti-slosh plates per
OWC would increase the internal surface degrees-of-freedom from
72 to 216; each original internal free-surface being partitioned into
three distinct free-surfaces.
Clearly replacement of the actual internal free surfaces with
judiciously located structural plates would permit limited internal
uid ow within the OWC channels and treatment of the whole
structure as a regular 6 degree-of-freedom offshore structure. This
model is designated the open device (Fig. 6a). Replacement of the
bottom and side openings of the J & N-shaped conduits by closure
plates will be referred to as the closed device (Fig. 6b). A subsequent parametric analysis of closure plate thickness would allow
appreciation of the effect of the openings on the overall stress
distribution [16]. The closed device greatly simplies both hydrodynamic and structural analysis.
To appreciate complexity differences in the uid structure
interaction the motion response formulation will be provided for
each possibility.
In the FSI analysis undertaken the dynamic radiation and
diffraction pressures are determined using a 3D velocity potential
based boundary element procedure [28], with subsequent structural integrity analysis undertaken using a matched 3D nite
element model [16].
Provided hydrodynamic reactive coefcients of added mass &
added inertia and the mass & inertia characteristics of the structure
are available an estimation of the natural frequencies of the device
can be readily determined. Heave natural frequency is a function of
device water plane area, whereas pitch and roll are functions of
wetted area rst moments and the assumptions made concerning

224

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

Fig. 6. a SEA-OWC-Clam: Open cross-section. b SEA-OWC-Clam: Closed cross-section.

the location of the interior water ballast. Given the interior mass of
water below the undisturbed free surface cannot exceed the displaced mass of water minus the material mass of the structure
(appealing to Archimedes Principle), the extent to which the underwater part may be ooded has to be determined to match
preferred device draught. Device draught was initially specied by
device inventor without mass distribution and engineering details
being specied. Archimedes principle is satised as discussed in
Section 3.3.
2.2.1. Hydrodynamic characteristics
The hydrodynamic analysis assumes that weight-buoyancy
equilibrium exists. Satisfaction of this requirement is addressed in
Section 3. The interaction of incident waves with the structure,
assumed xed, gives rise to diffraction waves, whereas the radiation waves result from the structure oscillating in any one of its
rigid body degrees-of-freedom in otherwise calm water.
2.2.2. Exploiting geometric planes of symmetry
The origin of the Cartesian hydrodynamic coordinate system
will be located in the undisturbed free surface at a horizontal
location that permits geometric symmetry to be exploited. Selection of such an origin leads to additional mechanical coupling in the
motion response equations because of the location of the centre of
gravity. Clearly an origin that allows two planes of symmetry to be
exploited reduces the size of the full matrices formulated to a
quarter of the complete device formulation.

The device quarter explicitly dened can be located in any

quadrant; here it is positioned in the rst quadrant with positive xand z-values. The hydrodynamic characteristics of input structure
and the 1st, 2nd and 3rd images of Fig. 7 are determined without
images being explicitly generated. Table 2 provides the symmetry
properties for each degree-of-freedom for each image. The
diffraction velocity potential has no general symmetry properties
and so the mathematical formulation must facilitate solution
determination in all four quadrants. All six degrees-of-freedom are
addressed to allow hydrodynamic quality checks.

2.2.3. Hydrodynamic quality checks

The radiation potentials provide the reactive coefcients of
added mass (inertia) and uid damping, Akj & Bkj, determined in
accordance with Fkj Akjsj  Bkj s_j , subject to:

8
>
< r
Akj
>
:Sja u

Z
SW

9
8
>
>
=
<r
j
fIm nk dS and Bkj 
>
>
;
:Sja

Z
SW

9
>
=
j
fRe nk dS :
>
;
(2)

That is the force/moment Fkj in the kth direction due to the

motion in the jth direction is expressed in terms of the unknown
acceleration sj and the unknown velocity s_j. Using Green's identities
[14] one can readily prove the symmetry relationship that Akj Ajk
and Bkj Bjk for all k & j, irrespective of structural geometric form
[14]. For the clam device the differences of cross coupling terms
were very small. To avoid introducing unrealistic asymmetries in
the motion response equations the average value of corresponding
cross terms is used.
Through Green's identities one can establish the mathematical
equivalence of the direct method of calculating wave excitation
loads using incident and diffracted velocity potentials and the
Haskind relationship [14] based on incident wave velocity potential
and the radiation velocity potential associated with the degree-of-

Table 2
Single rigid structure velocity potential sign changes of image structure relative to
Input section.

Fig. 7. Designation of images of explicitly dened clam structure.

Motion

Input section

1st image

2nd image

3rd image

Surge
Sway
Heave
Roll
Pitch
Yaw

1
1
1
1
1
1

1
1
1
1
1
1

1
1
1
1
1
1

1
1
1
1
1
1

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

225

Fig. 8. BEM discretization a & b used with closed and open form of SEA-OWC-Clam.

freedom corresponding to the direction of the required wave

loading. That is, one uses the mathematically equivalent wave
excitation expressions:

FkDirect
Excitation

r4I 4D 
SW

and

FkHaskind
Excitation

"
r 4I

SW

v4k
dS exp  iut
vn

v4
v4
 4k I
vn
vn

(3)

#
dS exp  iut

This requirement has been checked for all wave frequencies.

These checks provide condence in the appropriateness of the
calculated hydrodynamic properties and justify the sizing and
distribution of the selected boundary elements.
2.2.4. Alternative discretisation of the SEA-OWC-Clam
The device geometry presented in Fig. 2d requires many
degrees-of-freedom to be addressed. Given the device details
provided are conceptual rather than detailed design based, two
alternative simpler models are used for reasons justied in Sections
2.2 and 2.4.
In the open model, at plate closures are introduced within the
J and N OWC channels as illustrated in Fig. 6a. Their location can

The closed structural model is illustrated in Fig. 6b. Ideally an

opening within the structure should be strengthened, as addressed
in Fig. 4 of Part B [16]. However, such ne structural support elements makes the hydrodynamic model excessively large. Therefore
the closure plates of the closed model have variable thickness in
the structural model [16].
Different BEM discretisations are required for the closed and
open versions of the SEA-OWC-Clam. In the closed case 270
quadrilateral boundary elements explicitly dened a module;
hence 810 elements are used to dene the quarter structure of
Fig. 8a. The corresponding open discretisation requires 480
quadrilateral elements for each module and hence 1440 elements
dene the quarter structure of Fig. 8b. Here the representative
length scale of the largest boundary element used, in each case, was
less than 5% of the shortest wavelength considered.
On those parts of the structure common to both idealisations,
the BEM discretisation is identical to allow direct comparisons of
the closed and open device forms in these areas.
2.3. Generalised equations of motion for the SEA-OWC-Clam
For wave energy extraction performance the motion analysis must
include the inuences of the three possible degrees-of-freedom
associated with the J and N shaped OWC free surfaces. The surge,
sway and heave translational motion equations of the device are:

12
12


i X
i

X
X h SJ J
X h SN N
SJ J
SS _S
S S
S S
S iut
i _Ni
S
MS ASS
A1ji sji B1ji s_ji 
A1j i sj i BSN

s
11 s1 B11 s1 M xg s6  M yg s5 F1 e
j
1j

12
12


i X
i

X
X h SJ J
X h SN N
SJ J
SS _S
S S
S S
S iut
i _Ni
S
MS ASS
A2ji sji B2ji s_ji 
A2j i sj i BSN

s
22 s2 B22 s2 M yg s4  M zg s6 F2 e
2j j

i1 j3;4;5




SS _S
SS S
S S
S S
S
MS ASS
33 s3 B33 s3 C33 s3 M zg s5  M xg s4

i1 j3;4;5

i1 j3;4;5
12
X
X h SJ J
F3S eiut 
A3ji sji
i1 j3;4;5

represent emergency safety closure-plates activated when operation of OWC wave extraction is not considered prudent in terms of
device survivability. The plate thickness is consistent with the rest
of the structure.

i1 j3;4;5
12
i X
X h SN N
SJ J
A3j i sj i
B3ji s_ji 
i1 j3;4;5

i _Ni
BSN
s
3j j

(4a)
i

The additional moments of the mass on the left hand side of Eq.
(4a) are associated with the centre of gravity (xg, yg, zg ) being
distinct from the origin of the Cartesian reference system presented
in Fig. 5. The symmetry of geometry and mass distribution will

226

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

imply that xg zg 0. The corresponding general roll, pitch and

yaw device motions in a relatively simple form satisfy:

12
12


i X
i

X
X h SJ J
X h SN N
SJ J
S
SS _S
SS S
S S
S S
S iut
i _Ni
S
I44
A4ji sji B4ji s_ji 
A4j i sj i BSN
ASS

s
44 s4 B44 s4 C44 s4 M yg s2  M xg s3 F4 e
4j j

12
12


i
i

X
X
X h SJ J
X h SN N
SJ J
S
SS _S
SS S
S
S
S iut
i _Ni
S
I55
A5ji sji B5ji s_ji 
A5j i sj i BSN
ASS

s
55 s5 B55 s5 C55 s5 M yg s1  M zg s3 F5 e
5j j

i1 j3;4;5

i1 j3;4;5

i1 j3;4;5

i1 j3;4;5

(4b)

12
12


i X
i

X
X h SJ J
X h SN N
S
SS _S
S S
S S
S iut
i Ji
i _Ni
S
I66
A6ji sji BSJ
A6j i sj i BSN
ASS

s_ 
s
66 s6 B66 s6 M xg s1  M zg s2 F6 e
6j j
6j j
i1 j3;4;5

, and the uid damping, BMN

, provide
Here the added mass, AMN
kj
kj
measures of the reaction in the kth direction on the Mth structure
or substructure, resulting from the Nth structure or substructure
undergoing a motion of unknown amplitude in the jth degree-ofMN denotes the hydrostatic restoration terms
freedom. Similarly Ckj
acting on a structure or substructure; there is no hydrostatic
Archimedean restoration in the horizontal plane. Clearly superscript S, J and N indicate physical structure and massless free surface plates associated with the J-shaped and N-shaped OWC. The
motion related displacement, sN
j , now indicates that the Nth
structure or substructure moves in jth degree-of-freedom subject
to all other substructures remaining stationary in the associated
hydrodynamic analysis. This approach permits identication of
cause and effect inuences. The excitation load in the kth direction
on the Nth structure or substructure due to the incident and diffracted waves is designated FkN . The mass of the structure, M, has a
superscript S and the moments of inertia, I, have a double subscript
corresponding to the associated degree-of-freedom of the pertinent motion equation. Velocity and acceleration are indicated using
classical mathematical notation.
In the case of geometric symmetry not implying symmetry of
mass distribution, the products of inertia will be non-zero. Hence
there is structural borne coupling of the rotational degrees of
freedom [29]. When non-symmetry of mass exists the rst pure
moment of inertia dependent term in each of Eq. (4b) are replaced
in accordance with Eq. (5)

k k
k k
k k
s3k B33
A33
s3k F3k eiut 
s_3k C33

J J

J J

J J

12
X

X h

i1 j3;4;5

k k
k k
k k
s4k B44
s4k F4k eiut 
s_4k C44
A44

J J

J J

J J

isk
12
X

X h

i1 j3;4;5

k k
k k
k k
s5k B55
s5k F5k eiut 
A55
s_5k C55

J J

J J

J J

isk
12
X

X h

i1 j3;4;5
isk

i1 j3;4;5

S S
S S
S S
S S
I44
s4 /I44
s4 I45
s5 I46
s6
S S
S S
S S
S S
I55 s5 /I54 s4 I55 s5 I56
s6 :
S S
S S
S S
S S
s6 /I64
s4 I65
s5 I66
s6
I66

(5)

It is implicit within the device and free surface reactive loads

that the forcing in the kth direction on the Mth structure due to the
Nth structure oscillating in the jth degree-of-freedom, subject to all
other substructures remain stationary, can be expressed in terms of
added mass (inertia) and uid damping terms as:
MN
MN
N
Fkj
AMN
kj sj  Bjk

_N
kj sj :

(6)

Since these reactive forces are a function of the unknown motions, this mathematically equivalent way of expressing the reaction terms permits their correct inclusion in the dynamic motion
equations. Essentially the radiation forces and moments are
resolved in a right handed reference system recognising that velocity and acceleration are orthogonal.
There are 72 coupled free surface equations of motion for the
massless plates modelling the 24 free surfaces and their 3 associated degrees-of-freedom. That is, each free surface for each OWCshape is identied according to k 1, 2, , 12, with j 3, 4 & 5
denoting the heave, roll and pitch motions. The 72 newly derived
motion equations for the J related and the N related free surfaces
are readily expressed in the compact form of Eqs. (7a) and (7b),
namely:

12
i X
i
X h JN N
J J J
J J J
J N
i
A3jk i sji B3jk i s_ji 
A3jk i sj i B3jk i s_N
j
i1 j3;4;5
12
i X
i
X h JN N
J J J
J J J
J N
i
A4jk i sji B4jk i s_ji 
A4jk i sj i B4jk i s_N
j
i1 j3;4;5
12
i X
i
X h JN N
J J J
J J J
J N
i
A5jk i sji B5jk i s_ji 
A5jk i sj i B5jk i s_N
j
i1 j3;4;5

(7a)

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

Nk Nk _Nk
Nk Nk Nk
Nk iut
k Nk Nk
AN

33 s3 B33 s3 C33 s3 F3 e

12
12
i X
X
X h N J J
N J J
A3jk i sji B3jk i s_ji 
i1 j3;4;5

Nk Nk _Nk
Nk Nk Nk
Nk iut
k Nk Nk

AN
44 s4 B44 s4 C44 s4 F4 e

12
X

X h

i
N J J
N J J
A4jk i sji B4jk i s_ji 

Nk Nk Nk
C55
s5

F5Nk eiut

12
X

isk
12
X

X h

k Ni Ni
k Ni _Ni
AN
sj B N
sj
4j
4j

i1 j3;4;5

isk
k Nk _Nk
BN
55 s5

i
X h NN N
k Ni _Ni
A3jk i sj i BN
sj
3j

i1 j3;4;5

i1 j3;4;5

k Nk Nk
AN
55 s5

227

(7b)

isk

X h

AJ5jk Ji sJji

BJ5jk Ji s_Jji

12
X

i
X h NN N
k Ni _Ni
A5jk i sj i BN
sj
5j

i1 j3;4;5

i1 j3;4;5

isk

isk

Inclusion of the structural bafes would require generalisation of

the proposed equations over each partitioning of the free surface.
The increased cross coupling requires summation over free surface
subdivisions designated outer, middle and inner, say, and generalisation of Jk and Nk to Jkl and Nkl with the superscript indicating the
appropriate affected or causal portion of the free surface. Device
performance analysis would need inclusion of power take off
characteristics to reect method and physical changes associated
with energy extraction.
Energy conversion is to be based on a closed system. The
amount of power extracted by the device clearly depends upon the
movement of the internal free surfaces relative to the motion of the
structural ring enclosing the J and N-shaped channels. These motions will generate compressed air that is pushed into and stored in
the single high pressure reservoir (designated HP in Fig. 2d). The HP
reservoir is used to drive a unidirectional turbine which is coupled
to an AC generator. The exhausted air from the turbine is fed into a
low pressure (LP) duct that is used to re-establish the initial conditions of the extraction process.
Theoretical prediction of overall efciency requires the fuller

efcients depends upon detailed knowledge of the engineering of

the energy extraction process.
Realistic physical measurement of the wave extraction device
requires a prototype of an appropriate scale. Investment in such a
structure requires justication of survivability of the device, as
addressed in Ref. [16]. Currently the design team estimates of efciency are limited, as explained in discussion of Fig. 3.
2.4. Reduced motion equations and analytic solutions
The number of distinct velocity potentials to be determined to
evaluate the reactive added mass & uid damping coefcients and
the wave excitation loads in the generalised equations of the previous section is very computationally demanding and makes the
initial design process unnecessarily over rened when structural
details are preliminary estimates at best. That said, simplication of
the initial more exact motion response to standard oating offshore
analysis seems more sensible, even if viewed as more pragmatic.
For either simplied single rigid body the equations required
take on the form:



 S

M A11 s1 B11 s_1 A15 s5 B15 s_5 M S xgs6  M S ygs5 F1S eiut


 S

M A33 s3 B33 s_3 C33 s3 MS zgs5  MS xgs4 F3S eiut




S
I55
A55 s5 B55 s_5 C55 s5 A51 s1 B51 s_1 M S yg s1  MS zg s3 F5S eiut

generalised motion analysis to incorporate power extraction terms

Ji
J
i
_Ni
_s
of the form BPTO
$s_ji  s_s3 or BN
PTO $sj  s3 to address relative air
volume changes in each OWC; notation is explained in Section
2.2.4. However, specication of power extraction damping co-

 S

M A11 s1 B11 s_1 A15 s5 B15 s_5  M S ygs5 F1S eiut
 S

M A33 s3 B33 s_3 C33 s3 F3S eiut
S

I55 A55 s5 B55 s_5 C55 s5 A51 s1 B51 s_1 M S yg s1 F5S eiut

(8a)

Superscripts 0 ss0 on the reactive hydrodynamic coefcients are

now unnecessary since there is just one oating structure.
Assuming both geometric and mass distribution symmetry implies
xg 0 & zg 0 then Eq. (8a) simplify to:

(8b)

228

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

Author experience in offshore analysis shows that mooring inuences, often included as additional stiffness terms in Eq. (8), has
less inuence on the structural excursions than the structural excursions have on the mooring line analyses. Hence mooring effects
are not addressed here since only rst order inuences are being
modelled [Chapter 9 of 18, [30]].
Whilst it is quite common to solve the motion responses
numerically, using Gaussian elimination, the axi-symmetric nature
of the SEA-OWC-Clam device geometry permits analytic solution of
the uncoupled heave response and the coupled surge-pitch
responses.
Since all the hydrodynamic and hydrostatic quantities produced by the MATTHEW software are per unit uid density, it
follows that upon dividing the motion equations by the uid
density r then the calculated hydrodynamic and hydrostatic data
can be used directly and the mass term MS replaced by the
calculated displaced volume V. Hence the solution of the heave Eq.
(8b) yields:

s1R

s3I

S
F3I

S
F3R
uB33

C33  u V A33
h
i2
C33  u2 V A33 uB33 2

(9)

o
:

It should be noted that the analytic solutions for the coupled

surge and pitch require the effective pitch moment of inertia:
E
I55

S
I55
1
W
I Steel I55
;
r 55 r

(10)

Steel is determined from the structural model of the device

where I55
W is the moment of the ballast water of unit density derived in
and I55
Sections 3.3.1 and 3.3.2.
Required analytic solution yields:

S
S
F5I
C55  u2
uB15  F1R

DR 2 DI 2
i
  
o
 
E
S
S
I55
A55  F1I
uB55  F5R
u2 V yg  A15 DR

DR 2 DI 2
n
h

i
  
o
 
S
S
E
S
S 2
F1R
C55  u2 I55
uB55  F1I
A55  F5R
uB15  F5I
u V yg  A15 DR
n

s5R

and

n
h

i
  
o
 
S
S
E
S
S 2
F1R
C55  u2 I55
uB55  F1I
A55  F5R
uB15  F5I
u V yg  A15 DI
n

s1I

s3R

n h
o
i
S
S
F3R
C33  u2 V A33  F3I
uB33

h
i2
C33  u2 V A33 uB33 2

S
S
F5I
C55  u2
uB15  F1R

DR 2 DI 2
i
  
o
 
E
S
S
I55
A55  F1I
uB55  F5R
u2 V yg  A15 DI
DR 2 DI 2

(11)

 
n
h

o
i
 
S
S
S
S
F1I
u2 V A11  F5I
uB51 F5R
uB11  F1R
u2  V yg  A51 DR
n

DR 2 DI 2
 
h

o
i
 
S
S
S
S 2
F5R
u2 V A11  F1R
uB11 F1I
uB51  F1I
u  V yg  A51 DI

n
s5I
n


S
S
F5R
uB11 F1I

DR 2 DI 2
 

o
i
 
S
S 2
u2 V A11  F1R
uB51  F1I
u  V yg  A51 DR

DR 2 DI 2
 
h

o
i
 
S
S
S
S
F1I
u2 V A11  F5I
uB51 F5R
uB11  F1R
u2  V yg  A51 DI
DR 2 DI 2

(12)

The denominator terms are:

 
h

i


  

 
 
E
DI uB11 C55  u2 I55
A55  u3 V A11 B55 u3 B15  V yg  A51 u3 B51 V yg  A15
 

i
  


h
 
 
E
 V yg  A51  u2 B51 B15 :
A55 u2 B11 B55 u4 V yg  A15
DR u2 V A11 C55  u2 I55

(13)

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

Since the vertical centre of gravity coordinate is negative it

can be replaced by its modulus. The hydrostatic terms C33 &
C55 and the velocity potentials per unit displacement are
determined by the MATTHEW code. Hence the motions and
structural loadings may be investigated upon noting that the
hydrodynamic analysis is determine using two planes of geometric symmetry.

229

3. Device modelling
Whatever the actual structural arrangement, closed or open,
the mass-inertial characteristics of the structure must be evaluated
to be consistent with the Archimedean state of balance; this is
addressed next for both open and closed structural models.
3.1. SEA-OWC-Clam geometric models

2.5. Dynamic pressure distribution

The ABAQUS [31] FEA based structural analysis requires
explicit specication of the dynamic pressure distribution over the
entire structure. This is facilitated through the following steps:
 Using known symmetry properties of the radiation velocity
potentials, specied in Table 2, the velocity potentials over the
whole structure of Fig. 7 can be readily determined.
 Using analytical solutions of the rigid body motions the dynamic
pressure acting on the structure for the selected wave frequency
and heading at a generic point of the structure (x,y,z) is determined from:

vF

vt
 iur4Re x; y; z; u; q i4Im x; y; z; u; qeiut :

pStructure  pFluid px; y; z; t r

3.2. Hydrostatic characteristics

(14a)
 The most general resulting velocity potential is determined in
accordance with:
6
 X

Fx; y; z; t a Fincident FDiffracted
FSj

S
radiation sj

j1

12
X
X

Fjk radiation sjk

k1 j3;4;5

12
X
X

The initial idealisation considered is the closed structure. The

closure plates highlighted in Fig. 6b are initially the same thickness
as the whole structure. Thereafter they are reduced so that the
changes in stress levels in the remainder of the structure can be
assessed [16].
Lower experienced loads will be demonstrated [16] through
removal of the closure plates and analysing the open structure of
Fig. 6a.
The open structure thus consists of essential members that
contribute to the structural integrity of the device and nonessential members that serve to direct the water through the
openings in a simpler and more restrictive manner than the
intended J-tube and N-tube paths.

k
FN
sNk
j radiation j

k1 j3;4;5

(14b)
Here a is the amplitude of the incident wave propagating in the
0
direction
x
and
described
analytically
by
0
ag coshkyd
Fincident u
expikx

ut.
F
is
the
change
in
Diffracted
coshkd
the incident wave potential due to the presence of the structure,
assumed to be xed. FDiffracted is determined in each quadrant
automatically by the MATTHEW code.
For the simplied single body analysis one simply sets each sJjk
k
and sN
to zero and reduces the number of structural degrees-ofj
freedom to surge, heave and pitch; since loading is effectively independent of wave heading (see Section 4.1).

To achieve the required draught ballast water must be added.

Trivially, geometric and mass symmetry means the structural
centre of gravity must satisfy xg zg 0; whereas yg 2 m is a
proposed operational condition. Tables 3 and 4, automatically
generated by the hydrodynamic analysis, specify displaced volume,
wetted surface area and hydrostatic coefcients for the closed and
open structures respectively.
The presented restoration coefcients (Cij) are the translational
and rotational hydrostatic stiffness coefcients, determined from
integration of the appropriate sums and moments of the hydrostatic pressure. GM is a measure of the transverse and longitudinal
intact stability [32] of the oating structure. GM must be positive.
Integrity of the hydrodynamic boundary element model of the
device is established by ensuring wetted surface area and displaced
volume are consistent with the actual structure being modelled.
3.3. Inertial characteristics
The buoyancy force is balanced by the total weight of the
structure. Hence for a specied draught the total mass of the
structure is readily determined. The moment of inertia however
requires some thought. Analytic expressions are developed in
preference to applying numerical methods.

Table 3
Determined characteristics of closed form of SEA-OWC-Clam.
Restoration coefcients per unit uid density

Volume, GM and surface area (SA)

V determined within hydrodynamic analysis is 1.05380D04

GMT 1.04344D02
GML 1.04343D02
SA 4.22864D03

C33 1.74851D04
C43 0.00000D00
C53 0.00000D00

C34 0.00000D00
C44 1.07868D07
C54 0.00000D00

C35 0.00000D00
C45 0.00000D00
C55 1.07868D07

Table 4
Determined characteristics of open form of SEA-OWC-Clam.
Restoration coefcients per unit uid density

Volume, GM and surface area (SA)

V determined within hydrodynamic analysis is 7.97483D03

GMT 1.15213D02
GML 1.15213D02
SA 7.84871D03

C33 1.50786D04
C43 0.00000D00
C53 0.00000D00

C34 0.00000D00
C44 9.01349D06
C54 0.00000D00

C35 0.00000D00
C45 0.00000D00
C55 9.01347D06

230

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

3.3.1. Moment of inertia of closed structure

The SEA-OWC-Clam steel mass, Msteel, and the ballast water
within the structure, Mwater, must equal the displaced mass of
water, Mdisplaced, that is:

Msteel Mwater Mdisplaced rwater V:

(15)

Here V is the calculated displaced volume of Table 3. Sea water

density of 1025 kg/m3 is assumed. Msteel is determined within the
structural nite element code ABAQUS [16].
In offshore structure analysis cylindrical elements are usually
modelled using a polygonal representation of appropriate radius
identied to provide good estimates of wetted surface area and
displaced volume. Here, a similar logic is applied for identication
of the ballast water region. The ballast water within the dodecagonal device is located between appropriate inscribed and circumscribed circles.
The required amount of ballast water is Mwater Mdisplaced 
Msteel rwater Vwater. Mwater is an important component of the pitch
moment of inertia Ixx and can be determined by considering an
annulus of internal water whose radii can be evaluated in two
different ways:
 Initially the SEA-OWC-Clam is considered a dodecagon inscribed
in an external circle of radius Ro 40.0086 m, and a smaller
internal circle of radius Ri 31.7266 m; these two radii values
give a mean radius value of Rm 35.8676 m as shown in Fig. 9a.

moment of inertia of the water inside the structure i.e.

dry
water . The value of I dry I Steel corresponding to the
Ixx Ixx
Ixx
xx
55
structural elements is provided by ABAQUS. In the standard motion analysis formulation of Section 2.4 the effective moment of
inertia dened in Eq. (10) is used; in this case it corresponds to
6,648,588.601 m5.
It should be noted that in this case, in order to ensure that the
closed form of the device oats, the underwater section was only
lled to 6 m and this volume of water represents 87.85% of the
displaced underwater volume.

3.3.2. Moment of inertia of open structure

For the open case we have two cylinders of water to allow for
the location of the openings and the simplied ow paths. Otherwise the process is similar in terms of inscribed and circumscribed
cylinders, see Fig. 10a and b, but the existence of internal water and
openings xes certain radii values.
Table 5 summarizes the representative calculations.
In this case the internal water occupies 80.34% of the displaced
underwater volume.
With the equations of motion derived together with an indication of the structural, hydrostatic and hydrodynamic characteristics
required, the results of the hydrodynamic and motion response
analyses are addressed next.

Assuming the water is spread a distance twater either side of the

mean radius then the annulus has an outer radius of (Rmtwater)
and an inner radius of (Rmtwater). Hence it readily follows that
for draught h:


1
Mwater rwater Vwater rwater 2htwater 2p Rm twater
2

Rm  twater  rwater 4pRm htwater

(16)

 Next the SEA-OWC-Clam is considered a dodecagon circumscribed to an external circle of radius Ro 38.6454 m, and a
smaller internal circle of radius Ri 30.6454 m; these two radii
values give a radius mean value of Rm 34.6454 m as shown in
Fig. 9b.
Clearly for each of the two strategies the mean radius Rm is
different. In fact using the average value of these two mean radii
leads to Rm 0:535:8676 34:6454 35:2565 m. Hence rearranging Eq. (16) and utilising said radius yields:

twater

Mwater
3:48 m:
4prwater hRm

(17)

Hence the required mass moment of inertia of the ballast water

about the device diameter is:
water W
Ixx
I55


2 
2

1
Mwater Rm twater Rm  twater
4
1
Mwater h2

12

(18)

water value of 5,983,731,696 kg m2.

yielding an Ixx
The moment of the inertia of the entire SEA-OWC-Clam is the
summation of the moment of inertia of the dry mass and the

Fig. 9. a Plan view of closed SEA-OWC-Clam as inscribed dodecagon. b Plan view of

closed SEA-OWC-Clam as circumscribed dodecagon.

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

231

Fig. 11. Modulus of surge excitation force versus wave frequency.

Fig. 12. Modulus of heave excitation force versus wave frequency.

Fig. 10. a Plan view of open SEA-OWC-Clam as inscribed dodecagon. b Plan view of
open SEA-OWC-Clam as circumscribed dodecagon.

4. Presentation and discussion of predictions

All hydrodynamic and hydrostatic based quantities are per unit
uid density for consistency with the motion response arguments
of Section (2.4). A subscript indicating a related direction in the
preceding mathematics is no longer present in the graphical presentation of these quantities. The hydrodynamic characteristics will
indicate the likely resonant frequencies to be observed in the motion responses of open and closed models.
4.1. Excitation and reactive hydrodynamic characteristics
For the ringed form of the Clam (irrespective of specic congurations) and for the majority of wave frequencies considered

Fig. 13. Modulus of pitch excitation moment versus wave frequency.

Table 5
Interim calculations of moment of inertia for open structure.
Cylinder

Water Mass (kg)

Inertia of water (kg m2)

Steel inertia (kg m2)

Total inertia (kg m2)

Effective inertia (m5)

Inner
Outer
Inner
Outer
Totals

3711837.293
2855259.457
3711837.293
2855259.457
6567096.750

2057536810
2071532324
1932379954
1926279810

1027947000

5157016134

5031235.253

1027947000

4886606765

4767421.234

Means

5021811449

4899328.243

232

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

Fig. 14. Variation of pure surge added mass versus wave frequency.

Fig. 17. Variation of pure heave uid damping versus wave frequency.

Fig. 15. Variation of pure surge uid damping versus wave frequency.

Fig. 18. Variation of pure pitch added mass versus wave frequency.

the heave excitation is insensitive to wave direction. That is,

incident wave travelling along the common radius of two neighbouring modules or striking the at side of a module show
negligible difference in wave excitation loads for most wave
frequencies.
Figs. 11e13 provide the variation of magnitudes of the wave
excitation loads, for unit wave amplitude, with incident wave frequency (u) for the closed and open forms of the SEA-OWC-Clam
device for surge, heave and pitch. Whilst the magnitudes of
closed and open forms are quite comparable this does not mean

Fig. 16. Variation of pure heave added mass versus wave frequency.

they have the same phase relation with the incident wave since
their displaced volumes and wetted surface areas of Tables 3 and 4
are quite distinct. The corresponding changes in the hydrostatics
(particularly the pitch restoration coefcient) will inuence the
phasing relationship. Openings within the structure allow some
internal ow and this reduces the loading at longer wavelengths
(higher periods). This will ultimately inuence the stress levels
predicted.
Figs. 14e19 provide the pure surge, heave and pitch reactive
(radiation) coefcients of Akk & Bkk: k 1, 3 & 5 for both geometries.

Fig. 19. Variation of pure pitch uid damping versus wave frequency.

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

233

Fig. 20. Variation of pitch induced surge added mass (A15 A51) versus wave
frequency.
Fig. 22. Modulus of surge motion per unit wave amplitude versus wave frequency.

Fig. 21. Variation of pitch induced surge uid damping (B15 B51) versus wave
frequency.

Negative pitch added inertia for the closed structure, negative surge
and heave added mass for the open structure, are not a readily
appreciated concept for those who think in terms of entrained
mass. However, in an offshore engineering context it is expected
e.g. negative added mass values for submerged cylinders [33] and a
oating torus [34]; clearly of relevance to this device.
Only the surge-pitch cross-terms presented in Figs. 20 and 21
are non-zero. For both geometries the numerical differences are
not readily detectable in these graphs.

Fig. 23. Modulus of heave motion per unit wave amplitude versus wave frequency.

The behaviour of Fig. 23 is dramatically different. As u tends to

zero the motion response per unit amplitude tends to unity from
below. The peaks that occur are initiated from very low responses
and return to low responses over a short range of frequencies
(0.75e1.25). However, it is in keeping with the heave predictions of
a simple circular cross section torus analysed by Newman [34]. At

4.2. Rigid body motion responses

Figs. 22e24 provide the magnitude of the surge, heave and pitch
motions per unit wave amplitude respectively.
The surge motion, Fig. 22, is not particularly sensitivity to the
differences of geometric form or amount of water ballast and hence
inertia characteristics.
The heave response in Fig. 23 is quite different from what is
usually expected in a conventional heave transfer function for
ships. In the latter case we would expect heave amplitude per unit
wave amplitude to tend to unity (from above) as u tends to zero,
otherwise the heave motions will gradually increase to a resonance
value and then quickly drop away as frequency increases. Above a
certain threshold frequency the amplitude will appear to be frequency independent, with an amplitude small compared to the
incident wave amplitude. Short wavelengths have little impact.

Fig. 24. Modulus of pitch motion per unit wave amplitude versus wave frequency.

234

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

the natural frequency the term [C33  u2 (V A33)] of Eq. (9) will
equal zero. Inspection of detailed related computational les shows
that zeroing of this term and the observed peak response occurs
near 0.95 rad/s and 1.0 rad/s for the open and closed forms
respectively in Fig. 23. Approximate hand calculation of this situation is possible using Table 3 or 4 for appropriate geometric form
and use of added mass plots presented in Fig. 16.
There is very little difference in the heave response of the closed
and open forms of the device.
For the closed form of the device pitch (Fig. 24) is equally
dramatic for a frequency close to 1 rad/s. For the open device the
pitch motion is relatively normal with a small secondary peak at
1 rad/s.
Operation of the closed form of the device in a sea state with
excessive energy at a period of 6 s could prove interesting in terms
of resulting vertical motion (vector sum of heave and pitch).
In offshore engineering analysis apparent over predicted resonances can be addressed by introducing a viscous damping
correction based on the peak uid damping coefcient. For
example, the pitch uid damping term B55 s_5 can be replaced in the
_ _
pitch motion equation by the nonlinear term B55 a Bpeak
55 js5 js5 ,
peak
where a B55 js_5 j is treated as the equivalent viscous damping
term. The scaling term alpha is assigned a value up to, but not
usually exceeding 0.2.
In practice the motion equations are solved as originally
formulated. Thereafter the pitch solution is used to assign js_5 j and
for a selected alpha value the viscous damping term is evaluated to
permit solution of the nonlinear formulation as a linear equation.
Iteration will then yield a steady convergent value of pitch motion.
The viscous damping correction will only affect responses in the
region of the resonance peak.
To assess the sensitivity of the linear predicted pitch to viscous
damping we will rst address the pitch response spectra generated
using Eq. (1) and the spectral data of Fig. 4 for the higher energy
March and November observation for the South Uist location.
Fig. 25 shows that the high peak resonance of closed form in Fig. 24
is prominent in the generated response spectrum. Fig. 26 investigates the inuence of viscous damping terms for different
values of alpha.
For an alpha value as small as 0.05 the reduction in peak
amplitude for both March and November is 36%.

Fig. 25. Variance response spectral density versus frequency (rad/s).

Fig. 26. Increasing viscous damping inuences on peak pitch response of closed SEAOWC-Clam.

5. Closure
Hydrostatic and inertial properties have been deduced in each
model to provide preferred draught and centre of gravity so as to
ensure device is capable of oating at required draught.
As an operational wave energy device with some condence in
the structural integrity the more complex 222 degrees-of-freedom
hydrodynamic and motion response analysis is required together
with the indicated form of power take-off effects. At the current
preliminary design two simpler hydrodynamic models have been
investigated. The closed form structure is expected to exhibit
higher stresses than the open model, which in turn will experience
higher stresses than the operational full J & N tube based OWCs.
The quality of the hydrodynamics has been checked for numerical stability, in terms of conditioning numbers provided by the
principal author's MATTHEW software suite, together with checking of equality of radiation cross terms of added mass (inertia) and
uid damping coefcients and agreement between direct and indirect Haskind calculations of wave excitation.
Sensitivity of excitation to wave heading is negligible and hence
a head sea wave with formulation of coupled surge, heave and pitch
is sufcient to permit representative motion response analysis;
although all six degrees-of-freedom were investigated for hydrodynamic quality purposes.
Therefore the hydrodynamic velocity potentials and motion
responses are of sufcient quality to provide the complex variable
based dynamic pressure loads necessary to proceed with the
structural integrity enquiry pursued in the companion paper.
The distinct pitch response of the simpler closed form model
illustrated in Fig. 24 has signicant impact within the response
spectrum of Fig. 25. In reality the responses of interest are the
resultant vertical motion (acceleration) dependent upon combined
heave and pitch amplitude and phase information. For mooring
considerations the surge and pitch inuence upon horizontal
excursion would be studied in greater depth, although viability of
the moored device would ultimately be addressed in the time
domain with inclusion of integral memory terms, since actual
response at any instant is as much a function of previous levels of
excitation and response as current wave loading.
An ad hoc procedure fort adding peak responses through the

N.W. Bellamy et al. / Renewable Energy 88 (2016) 220e235

introduction of a viscous damping correction has been presented

and shown to have impact on peak pitch response spectrum for low
alpha values.
The frequency approach adopted is deemed to be of sufcient
quality and detail to permit investigation of initial structural
integrity for the device. The resulting stress analyses, with investigation on how to improve device survivability, are provided in the
companion paper [16].
Acknowledgements
The authors gratefully acknowledge assistance of Dr. Mingy Tan
in the preparation of the original Carbon Trust research contract
that has been substantially expanded in this paper.
References
[1] G.E. Hearn, E. Donati, I.K. Mahendran, Prediction, measurement and comparison of uid-structure interaction using mathematical and experimental
results, Appl. Math. Model. 7 (February 1983) 41e47.
[2] G.E. Hearn, K.C. Tong, The Hydrodynamic, Motion and Structural Strength
Analysis of a Ring Shaped Clam Wave Energy Device, Consultancy contract
report for Coventry (Lanchester) Wave Energy Group sponsored by the
Department of Energy, University of Newcastle upon Tyne, 1986, pp. 1e36.
[3] N.W. Bellamy, The SEA Clam Wave Energy Converter. Figure 26 of Section 4.2
& Figure 28 of Section 4.3, Carbon Trust Marine Energy Accelerator (MEA)
Application, 2008.
[4] N.W. Bellamy. Ibid Figure 22 of Section 3.4 and Page 51 of Section 4.3.
[5] N.W. Bellamy. Ibid Figure 33 of Section 4.3.
[6] N.W. Bellamy, A.M. Peateld, Design and performance of the circular sea clam
wave energy converter, in: 3rd International Symposium on Wave, Tidal,
OTEC, and Small Scale Hydro Energy, Brighton (UK), May 14e16, 1986.
[7] L.J. Duckers, F.P. Lockett, B.W. Loughridge, A.M. Peateld, M.J. West,
P.R.S. White, Offshore wave energy devices, in: 1st World Renewable Energy
Congress, Reading (UK), September 23e28, 1990.
[8] L.J. Duckers, F.P. Lockett, B.W. Loughridge, A.M. Peateld, M.J. West,
P.R.S. White, Optimisation of the clam wave energy converter, Renew. Energy
5 (1994) 1464e1466.
[9] N.W. Bellamy. Ibid Figures 29e32 of Section 4.3.
[10] T.W. Thorpe, A Review of Wave Energy, ETSU Report for The Department of
Trade and Industry, December 1992. UK, ETSU-R 72.
[11] N.W. Bellamy. Ibid Page 52 of Section 4.3.
[12] N.W. Bellamy. Ibid Figure 16 of Section 2.4.
[13] M.D. Haskind, The Exciting Forces and Wetting of Ships in Waves. Izvestia
Akademii Nauk S.S.S.R., Otdelenie Technicheskikh Nauk, no. 7, 1957, p. 65 (In
Russian). (English Translation is available as: David Taylor Model Basin
Translation No. 307, 1962).

235

[14] A.Y. Odabasi, G.E. Hearn, Seakeeping theories: what is the choice? Trans.
NECIES 94 (1978) 1e53.
[15] T. Timman, J.N. Newman, The coupled damping coefcients of a symmetric
ship, J. Ship Res. 5 (1962) 1e7.
[16] A. Bucchi, G.E. Hearn, Analysis of the SEA-OWC-clam wave energy device e
part b: structural integrity analysis, J. Renew. Energy (2015) (?:???-???
[Submitted for review]).
[17] H. Mitsuyasu, F. Tasai, T. Sumara, S. Mizuno, M. Ohkuso, T. Honda, K. Rikiismi,
Observations of the power spectrum of ocean waves using a cloverleaf buoy,
J. Phys. Oceanogr. 10 (1980) 286e295.
[18] G.E. Hearn, A.V. Metcalfe, Spectral Analysis in Engineering: Concepts and Case
Studies, Elsevier, Oxford, 1995.
[19] British Oceanographic Data Centre (BODC) https://www.bodc.ac.uk/data/
online_request/waves/.
[20] W.J. Pierson Jr., L. Moskowitz, A Proposed spectral form for fully developed
wind seas based on the similarity theory of S. A. Kitaigorodskii, J. Geophys.
Res. 69 (1964) 5181e5190.
[21] W.J. Pierson Jr., L. Moskowitz, A Proposed Spectral Form for Fully Developed
Wind Seas Based on the Similarity Theory of S. A. Kitaigorodskii. Technical
Report Prepared for U. S. Naval Oceanographic Ofce, 1963. N62306 e1042:26.
[22] C.L. Bretschneider, Wave Variability and Wave Spectra for Wind- Generated
Gravity Waves. Tech. Memo. 113, Beach Erosion Board, US Army Corps of
Engineers, 1959.
[23] J.V. Wehausen, E.V. Laitone, Surface waves, Encycl. Phys. e Handbuch Physic 9
(1960) 446e778.
[24] O.D. Kellogg, Foundations of Potential Theory, Springer, Berlin, 1929.
[25] G.E. Hearn, Alternative methods of evaluating the green's function in threedimensional ship-wave problems, J. Ship Res. 21 (1977) 89e93.
[26] G.E. Hearn, J.R. Chaplin, The uid structure interaction of wave energy devices: some old and some new theoretical and experimental challenges, in:
The Royal Institution of Naval Architects Conference on Marine Renewable
Energy, London, 2008.
[27] G.E. Hearn, M. Katory, Application of hydrodynamic analysis to wave power
generators. Institute of fuels' golden jubilee conference on advancing energy
technology, J. Inst. Fuel 119 (June 1978) 119e126.
[28] G.E. Hearn, D. Lafforgue, E. Perdriset, D. Saydan, The hydrodynamics and dynamic motion analysis of a damaged ship, Int. J. Marit. Eng. Trans. RINA 150
(June 2008) 14e36.
[29] E.V. Lewis (Ed.), Principles of Naval Architecture 2nd Revision (Volume III
Motions in Waves and Controllability), The Society of Naval Engineers and
Marine Engineers (SNAME), New York, 1989, pp. 45e46. Section 3.3.
[30] G.E. Hearn, K.C. Tong, F.A. Ramzan, Wave drift damping coefcient predictions
and their inuence on the motions of moored semisubmersibles, in: Offshore
Technology Conference (OTC Paper 5455), Houston (USA), April 1987.
[31] Abaqus 6.9, Analysis User's Manual, in: Elements, vol. IV, 2009.
[32] Principles of naval architecture, in: J.P. Comstock (Ed.), Article on Intact Stability by Moore, CS, The Society of Naval Engineers and Marine Engineers
(SNAME), New York, 1967.
[33] P. McIver, D.V. Evans, The occurrence of negative added mass in free-surface
problems involving submerged oscillating bodies, J. Eng. Math. 18 (1984) 7e22.
[34] J.N. Newman, The motion of a oating slender torus, J. Fluid Mech. 83 (1977)
721e735.