ARTICLE IN PRESS

Ocean Engineering 35 (2008) 1015– 1028

Contents lists available at ScienceDirect

Ocean Engineering
journal homepage: www.elsevier.com/locate/oceaneng

2D Wave setup behind submerged breakwaters

M. Calabrese a, D. Vicinanza b,, M. Buccino a
a
Department of Hydraulic and Environmental Engineering, University of Naples ‘‘Federico II’’, Via Claudio 21, 80125 Naples, Italy
b
Department of Civil Engineering, Second University of Naples, Via Roma 29, 81031 Aversa (Caserta), Italy

a r t i c l e in fo abstract

Article history: In addition to reducing the incoming wave energy, submerged breakwaters also cause a setup of the sea
Received 19 April 2007 level in the protected area, which is relevant to the whole shadow zone circulation, including
Accepted 3 March 2008 alongshore currents and seaward flows through the gaps. This study examines such a leading hydraulic
Available online 13 March 2008
parameter under the simplified hypothesis of 2D motion and presents a prediction model that has been
Keywords: validated by a wide ensemble of experimental data. Starting from an approach originally proposed by
Submerged breakwaters Dalrymple and Dean [(1971). Piling-up behind low and submerged permeable breakwaters. Discussion
Hydraulic response note on Diskin et al. (1970). Journal of Waterways and Harbors Division WW2, 423–427], the model
Large wave flume
splits the rise of the mean water level into two contributions: one is due to the momentum flux release
Small wave flume
forced by wave breaking on the structure, and the other is associated with the mass transport process.
Wave setup
For the first time, the case of random wave trains has been explicitly considered.
& 2008 Elsevier Ltd. All rights reserved.

1. Introduction matter, research outcomes seem quite encouraging, seeing that a

number of reliable design tools have been proposed, including
In situations where landscape safeguarding is a major concern empirical formulae (van der Meer et al., 2005), neural networks
(Italian coasts are an important example), submerged breakwaters (Panizzo et al., 2003) and conceptual equations (Buccino and
are frequently employed as shore-protection measures, often in Calabrese, 2007). Nevertheless, other leading factors deserve to be
conjunction with beach-nourishment. Despite the fact that they researched in depth.
have long been employed, with different design philosophies, In analyzing the emblematic case of Palm Beach, Florida, where
either in Europe or in the United States or in Japan, substantial nearly 82,000 m3 of sand were lost (over three years) after the
uncertainties still remain on the predictability of shoreline placement of an experimental proprietary submerged breakwater
response to their placement; in fact a limited or deleterious effect of 1260 m length, Dean et al. (1997) reasoned that a key role in the
of structures on coastline equilibrium and nearshore profile has erosive process was played by longshore currents that transported
been often documented (Seji et al., 1987; Ranasinghe and Turner, sand from the landward of breakwater towards the heads. These
2006). transverse flows were most likely driven by a setup of the mean
For this reason, a number of research projects have recently sea level behind the structure, caused by both wave momentum
focused on how the structures affect the shadow-zone hydro- release and mass transport processes associated with wave
dynamics; among them the EU projects DELOS (Lamberti, 2005), overpassing. Just the alongshore variation of this setup evidently
which has also investigated the compatibility of breakwaters with allowed the water to be directed parallel to the breakwater and
marine environment, and Low Crested and Submerged Breakwaters then offshore.
in Presence of Broken Waves (Calabrese et al., 2002), which has Hence, we recognize the structure-induced variation of the
analyzed the hydraulic response of structures under extreme mean water level to be a very important flow parameter for
climatic conditions, must be mentioned. Following the sugges- engineering purposes, also considering its effect on segmented
tions coming from both field measurements and numerical systems, where any passages through the barriers have a seaward
simulations, a substantial amount of work has been directed to current that may jeopardize the safety of swimmers (Bellotti,
wave transmission across the breakwater, since it has been 2004). Nevertheless, the phenomenon has been little investigated
recognized as one of the fundamental mechanisms governing so far, although some attention has been drawn to the analogous
the long-term evolution of the protected beach. Regarding this case of coral reefs (see for instance Gourlay, 1996). Moreover, in
the past no research addressed the case of random waves, though
it is receiving now a growing attention, especially among the
 Corresponding author. Tel.: +39 3385059599; fax: +39 0815037370. Italian researchers (e.g. Cappietti et al., 2006).
E-mail addresses: calabres@unina.it (M. Calabrese), diegovic@unina.it This study focuses on the quantitative prediction of wave setup
(D. Vicinanza), buccino@unina.it (M. Buccino). behind submerged breakwaters with respect to the idealized case

0029-8018/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.oceaneng.2008.03.005
 ARTICLE IN PRESS

1016 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

Nomenclature kI wave number offshore the barrier

kII wave number inshore the barrier pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ac area of the breakwater cross-section KR reflection coefficient, HR/Hi or m p =m0i ffi
0Rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ar area of the surface roller in the vertical plane Kt transmission coefficient, Ht/Hi or m0t =m0i
B0 wave peakedness parameter L0 regular deep water wavelength
B breakwater crest width L regular wavelength
Beq breakwater equivalent crest width Ac/hc Lp peak wavelength
c phase speed m(0)i,R,t zero order moment of power spectrum; the suffixes
D(50)a,c nominal rock diameter (M(50)a,c/rr)1/3. The suffixes ‘‘a’’ ‘‘i’’, ‘‘R’’ and ‘‘t’’ stand for incident, reflected and
and ‘‘c’’ stand for armour and core, respectively transmitted, respectively
E wave energy M50 50% value of rock mass distribution curve
f friction parameter qin volume flux entering the protected area
g gravity acceleration R hydraulic radius
hI still water depth offshore the barrier Rc crest freeboard; it is negative when the crest is
hII still water depth inshore the barrier submerged
hb still water depth at incipient breaking s0 fictional wave steepness Hi/L0
hc breakwater height SJJ JJ component of the radiation stress tensor
hP height of the impermeable part of the breakwater T regular wave period
H0 deep water regular wave height Tp peak wave period
Hb wave height at incipient breaking UR Ursell number
Hi incident regular wave height V alongshore velocity at breakwater head
HR reflected regular wave height a,b correction factors
Ht transmitted regular wave height d total wave setup behind the barrier
pffiffiffiffiffiffiffiffiffiffi
H(rms)i,R,t energetically equivalent wave height 8m0 : The d0 difference between mean water levels behind and in
suffixes ‘‘i’’, ‘‘R’’ and ‘‘t’’ stand for incident, reflected front of the barrier
and transmitted, respectively dc continuity wave setup behind the barrier
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
H(m)i,R,t mean wave height 2p  m0 : The suffixes ‘‘i’’, ‘‘R’’ and dmf momentum flux wave setup behind the barrier
‘‘t’’ stand for incident, reflected and transmitted, rr rock density
respectively rw water density

of 2D motion. As a consequence of alongshore flow being As a further matter of novelty, the role of structure permeability is
impeded, an upper limit (or ‘‘potential’’ value) is of course investigated.
obtained, which, however, may be of significance for engineering In the concluding sections adaptation of the model to random
applications for several reasons. First 2D wave setup might be seas is addressed and its possible application to both practical
included within simplified circulation models, which aid en- engineering and research fields is discussed.
gineers at a preliminary stage of the design process (examples are
given at the end of the article); for another it may serve as a
reference for open flow situations, where, as argued by Lamberti 2. Review of existing models
et al. (2007), any effects due to the real 3D motion could be
separately calculated and then subtracted from the potential rise To the knowledge of the authors, the setup of the sea level past
of the mean sea level (e.g. Eq. (13.88) of the Lamberti et al. (2007) a submerged breakwater was first described in Homma and Sokou
paper). Finally, in the context of mathematical modelling, 2D wave (1959) and Homma and Hoikawa (1961). However, only a
setup gives a boundary value of the mean water level for zones of qualitative view of the phenomenon was provided there, without
computing domain, such as along a cross-shore symmetry axis any analysis on parameters that control its magnitude.
(Dalrymple, 1978), in which the motion is expected to be nearly At the end of the 1960s, Longuet-Higgins (1967) developed an
planar. analytical solution by considering the time-averaged flux of
The article is organized as follows: first, a detailed review of vertical momentum into a column of water included between
previous research is given to provide the reader with the the still water level and the free surface. Under the hypotheses of
entire spectrum of published predictive equations. Then, after a small amplitude waves and irrotational motion, which excludes
critical analysis of literature, the central part of the work deals any energy loss, either by breaking or friction or other means, the
with an alternative model based on the balance of momentum authors arrived at the following expression, which returns the
between two fixed vertical planes enclosing the structure cross- difference of mean water levels, d0 , between two regions of
section. Originally the model has been developed within uniform depth shoreward and seaward a submerged breakwater
Low Crested and Submerged Breakwaters in Presence of Broken (Fig. 1):
Waves (Calabrese et al. 2003, 2005), but it has never been
H2i ð1 þ K 2R ÞkI H2i K 2t kII
expounded properly. In the following, the case of a regular train of d0 ¼  . (1)
waves is considered first: all the hypotheses are widely discussed 8 sinhð2kI hI Þ 8 sinhð2kII hII Þ
and predictive power of the new method is compared with that In Eq. (1) Hi is the incident wave height, KR is the reflection
of the other equations; for this purpose a good deal of new coefficient (reflected to incident wave height ratio) and Kt is the
experiments carried out at the University of Naples ‘‘Federico II’’ transmission coefficient (transmitted to incident wave height
have been employed. In addition the model has been modified ratio); kI,II and hI,II, respectively, represent the wave number and
to make it more consistent from a physical point of view; the water depth offshore (subscript I) and inshore (subscript II)
modifications include accounting of wave reflection and a better the structures. It is easy to realize that Eq. (1) represents the
modelling of wave breaking occurrence onto the breakwaters. difference between the mean water levels at each of the uniform
 ARTICLE IN PRESS

M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028 1017

Dalrymple and Dean (1971) commented that the scatter

s.w.l. between the data of Diskin et al. (1970) and Eq. (2) could be
m.w.l.
explained by the authors not having included in the formula
either the transmission or the reflection coefficients, the role of
δ' δ which had been previously pointed out by Longuet-Higgins
(1967). Furthermore, they proposed a predictive procedure that
might be considered as the starting point of the conceptual
method we are going to present and discuss in the next sections.
The procedure (for submerged breakwaters only) assumes the
wave setup to be forced partly by the release of momentum flux
due to wave breaking, and partly by the need for a return current
to occur in order to compensate for the flux of water mass
entering the protected area over the crest of the structure. d is
αoff. αin. then thought as the sum of two contributions:

Fig. 1. Definition sketch of breakwater cross-section. d ¼ dmf þ dc . (5)

The ‘‘momentum flux’’ contribution, dmf, has been treated the
depth Regions I and II, calculated by the second order Stokes wave same as the wave setup at a beach. Starting from the conservation
theory (Longuet-Higgins and Stewart, 1962). Despite the irrota- of horizontal momentum (Longuet-Higgins and Stewart, 1962),
tional hypothesis being scarcely verified in practice, the formula under the shallow water hypothesis and assuming the wave
has the advantage of clearly highlighting the role of both height to depth ratio to be constant, the authors obtained
reflection and transmission processes; more reflection enhances
setup, whereas wave transmission acts in the opposite sense. dmf ¼ 0:15ðhb þ Rc Þ, (6)
Longuet-Higgins theory has been experimentally evaluated by where hb indicates the water depth at incipient breaking, which
Dick (1968), who performed regular wave tests on an imperme- can be estimated by the solitary wave breaking criterion
able rectangular breakwater model. The author found Eq. (1) to hb ¼ 1.28Hi.
greatly underestimate the experimental data. The ‘‘continuity’’ setup, dc, is calculated by imposing the net
Some years later, results of an extensive experimental research mass influx over the breakwater to be returned offshore over and
conducted at the Technion Israel Institute of Technology of Haifa through the barrier:
(Israel) were published by Diskin et al. (1970). The authors carried
out nearly 190 2D regular wave experiments on homogeneous qin ¼ qover þ qthrough . (7)
breakwater models, with fixed side slopes (tan aoff. ¼ tan ain. As far as the incoming water mass, qin, is concerned, the authors
¼ 1:3), rock size (D50ffi2.5 cm) and crown width (B ¼ 10 cm); supposed it to be a fraction of the incident waves Stokes drift:
wave height and period were varied within reasonably wide
ranges and both underwater and protruding configurations were E
qin ¼ b , (8)
considered. It was found out that experimental values of wave rw c
setup past the breakwaters, d, could be approximated, on an
where rw is the water density, E ¼ 18rw gH2i is the incident wave
average, by the following empirical relationship:
energy, g is the gravity acceleration and c is the phase speed of the
"   #
d Rc 2 incident (linear) waves, calculated at the offshore toe of the
¼ 0:60 exp  0:70  , (2) barrier, hI. The key factor b, less than unity, depends on the time
H0 H0
interval, over a wave period, in which the water surface exceeds
where H0 is the deep water wave height, which the authors the breakwater crest. Using the small amplitude wave theory,
calculated as the ratio between the measured incident one, Hi, and Dalrymple and Dean (1971) obtained:
the linear shoaling coefficient. We see that in the equation above,  
1 1
that holds for both submerged and emerged structures, the rise of b¼ cos1 ð2uÞ þ sin½2 cos1 ð2uÞ , (9)
p 2
the sea level is related to the crest freeboard by means of a bell-
shaped Gaussian curve. The maximum of the curve is at Rc/ where
H0 ¼ 0.7; that is to say, when the water level is just below the top  
Rc  ðdmf þ dc Þ=2
of the breakwater. For water levels above and below this value, d u¼ (10)
decreases and tends to zero. Eq. (2) should be used within the Hi
ranges: is the ‘‘effective’’ relative crest freeboard, including the setup
height. Obviously b ¼ 1 for up12, as in this case the crest of the
Rc
2:0o o1:5, (3) breakwater lies under the wave trough. Although a hypothetical
H0
Darcy type formula was suggested for calculating qthrough, the
authors observed, from a comparison with Diskin et al. (1970)
hI
0:10o o0:83. (4) data, that reasonable predictions of the total setup could be
H0 obtained by a simplified mass balance, in which qin was entirely
It is worth noting that wave period does not appear in the equated to the backflow over the structure. The latter has been
formula, although the authors recognized it to play some role in estimated according to the expression reported below:
the wave setup development. As expected, neither crown width, pffiffiffiffiffiffiffiffiffiffid þ dc 
nor armour unit diameter (and consequently permeability), nor qover ¼ 2gdc mf  Rc , (11)
2
side slopes are included in the predictive equation.
Two very interesting discussion notes followed the Diskin et al. which is based on the hypothesis that the hydraulic head dc is fully
(1970) paper: one by Gourlay (1971) and the other, of a more converted into kinetic energy. Clearly, the mass balance, Eq. (7),
quantitative nature, by Dalrymple and Dean (1971). must be iteratively solved for dc to obtain the total setup d.
 ARTICLE IN PRESS

1018 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

More recently, Loveless et al. (1998), in the frame of a research For these reasons, the Dalrymple and Dean (1971) procedure
program funded by the UK Ministry of Agriculture Fisheries and (D2 hereafter) should be preferred in principle, as it accounts for
Foods, carried out at the University of Bristol a substantial number both the breaking induced reduction of the wave momentum flux
of experiments on submerged and low-crested breakwaters with (radiation stress) and the mass balance. However, the following
homogeneous cross-sections. Based on regular wave tests, the model shortcomings must be highlighted:
authors proposed the following formula, which is valid for both
positive and negative clearances:  Eq. (6) returns no momentum flux setup, dmf, for the very
 2 "  2 # frequent case in which waves break at the top of the barrier
d Hi L=phI T Rc (hb ¼ Rc), and this sounds physically questionable.
¼ 1:23 exp 20 , (12)
B gD50 hc  Since the momentum flux release continues while waves travel
across the breakwater crown, dmf should also increase with B;
where besides the symbols already introduced, L is the local in contrast, the crest width is not included in Eq. (6).
incident wavelength, T is the incident wave period and hc is the  The procedure does not consider either the transmission or the
structure height. reflection coefficients, although the authors themselves under-
The Loveless et al. (1998) idea is that the breakwater cross- lined the relevance of these parameters.
section functions like a weir and the wave height serves as  The high transport velocities, together with the roughness of
hydraulic head; owing to the wave damping, the head at the materials which most of the breakwaters consist of, make the
inshore side of the barrier is smaller when compared with the one backflowing process hardly to occur without any dissipation. In
at the offshore side, and the water discharged into the shadow this regard, it should also be mentioned that the thickness of
zone during the first half period is not balanced by an equal the structures might reach considerable values, as in the case
outflow during the second half period. This misbalance is also of the ‘‘artificial reefs’’ frequently adopted in Japanese practice.
amplified by the circumstance that the small head constraints the  The procedure does not apply to random waves.
return current to take place (partly) through the barrier, increas-
ing the backflow resistances. Eventually, the mass flux difference
To cope with these problems, an alternative calculation
allows the water to ‘‘pile up’’ shoreward the barrier, until an
method is presented and discussed in the following. The general
equilibrium state is reached.
structure of D2 (Eq. (5)) is maintained, but the way of computing
Consistent with the above statements, Eq. (12) models d as the
both the momentum flux and the continuity setup contributions
hydraulic head required to drive back by seepage, the volume flux
is changed. In particular, dmf is calculated by applying the
discharged by the incident waves over a half-period. For linear
horizontal momentum balance to a control volume surrounding
waves this flow rate just equals
the structure, that allows including KR, Kt and, indirectly, B in the
Z þT=4 Z 0 computation. As far as the continuity contribution is concerned, it
2 Hi L
QW ¼ dt u dz ¼ . (13) is modelled as the hydraulic head that is necessary to equilibrate
T T=4 dI pT
the flow resistances due to the interaction between the undertow
Since the motion is assumed to be rough-turbulent, the compensating the mass influx, and the structure. Finally, by use of
hydraulic gradient at the left-hand side of Eq. (12), d/B, is irregular wave experiments, the ‘‘new’’ method is adapted to
proportional to the square of the transport velocity Vw ¼ QW/hI, random seas.
and it is inversely proportional to D50. According to Diskin et al.
(1970), a bell-type function has been chosen to globally represent
3.2. The momentum flux contribution
the effects of Rc, which influences both the amount of inflowing
water and the magnitude of backflow resistances.
To calculate dmf, the time-averaged horizontal momentum
Yet, the maximum of the curve is now located at Rc ¼ 0 instead
balance, per unit of span, has been applied to the control volume
of Rc ¼ 0.7H0 (Eq. (2)); the authors explained such a behavior
shown in Fig. 2, which is included between the sea bottom, the
as an effect of the breakwater permeability, which in their
free surface, and the outer profile of the structure. The balance
investigation was at least 40% larger than that of Diskin et al.
equation reads
(1970).
Sxx;II  Sxx;I  DP þ PII  P I ¼ 0, (14)
where Sxx,I–II are the xx components of the radiation stress tensor
3. An alternative calculation method (Longuet-Higgins and Stewart, 1962) at the vertical planes I and II,
DP is the net (horizontal) force exerted by the structure on the
3.1. Conclusions from literature analysis and description of the volume of fluid and PI–II represent the external hydrostatic thrusts.
adopted approach In calculating previous terms, some rough hypotheses are
introduced. First, we assume the wave period to be an invariant of
Apart from Eq. (2), which is purely empirical, in all the the process, that is, generation of high frequency free waves in the
prediction methods presented earlier, the wave setup is inter- sheltered area (Grue, 1992) is not considered. It might be of interest
preted as the force counteracting a structure-induced lack of that the writers have recently observed this assumption to be
momentum balance. The difference among them is on which force reasonable when heavy plunging breakers occur at the crest or at
is considered to be the primary cause of the disequilibrium. Thus the seaward slope of a barrier (Calabrese et al., 2006). Furthermore,
Longuet-Higgins focused on the flux of wave momentum, while in a partial standing wave field (PSf hereafter) is thought to take place
Loveless et al. (1998) formula, the mass flux is dominant and the in Region 1 of Fig. 2, while simple progressive waves are supposed
shear stress related to the backflowing process is the main force. for Region 2. For the radiation stress components, the following
However, neither of the two approaches seem to be entirely approximate expression is used at plane I:
convincing; the former, in fact, disregards wave breaking, as well
Sxx;I ¼ Sxx;i ð1 þ K 2R Þ
as any dissipation source, while the weir-like scheme proposed by

Loveless et al. (1998) appears more suitable for emerged break- 1 4kI hI
¼ rw gH2i 1 þ ð1 þ K 2R Þ, (15)
waters. 16 senhð2kI hI Þ
 ARTICLE IN PRESS

M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028 1019

ΔΠ

Fig. 2. Scheme for calculation of the momentum flux setup.

which is based on averaging the wave momentum flux over a (2003) limit wave height:
wavelength, neglecting the shift of the phase constant between  
2pjRc j
incident and reflected waves; note that for the limit cases of simple Hcr:
b ¼ 0:095L0 tanh , (18)
L0
progressive waves (KR ¼ 0) and standing waves (KR ¼ 1), Eq. (15)
returns the theoretical expressions found by Longuet-Higgins and where L0 is the deep water wavelength. Eq. (18) comes from
Stewart (1964). Clearly, at plane II we have specific experiments on wave breaking at the crown of a

submerged obstacle, though the model used in the Hur et al.
1 4kII hII
Sxx;II ¼ Sxx;i K 2T ¼ rw gH2i 1 þ K 2. (16) (2003) research was smooth and impermeable. We see that
16 senhð2kII hII Þ T
under the feasible hypothesis of shallow water, Eq. (18) returns
In estimating DP, we consider, as a first gross approximation, only wave height to depth ratios of about 0.6. This corroborates
terms related to the mean hydrostatic pressures; that means findings of Sawaragi (1995), who analyzed the behavior of small
assuming all the other forces (mean dynamic pressures as well as and medium scale models of Niigata and Nishiki submerged
average Reynolds stresses related to the through-passing) to vanish breakwaters and found that wave breaking took place for Hi/|Rc|
in their integral values along the breakwater, or to be negligible. around 0.625. Note that if we had simply used the Iwata and
For the computation, we additionally suppose that the surf Kiyono formula for progressive waves (KR ¼ 0) we would have
zone extends from the breaking point to the inshore toe of the found Hi/|Rc|ffi0.87, that is quite larger. This explains why we
barrier and that wave setup linearly increases across it (Fig. 2). For employed two different formulae for modelling wave breaking
estimating the still water depth at incipient breaking hb, the occurrence.
following procedure is adopted. It is assumed that waves collapse Finally, if incident wave height is less than the right-hand side
onto the front slope of the breakwater where the incident wave of Eq. (18), the waves will travel across the structure without
height equals the limiting value proposed by Iwata and Kiyono breaking and dmf as well as the total setup is set equal to zero.
(1985): As far as hydrostatic thrusts at planes I and II are concerned,

  they can be readily calculated from the triangular pressure
1 1  KR h
Hi ¼ 0:218  0:076 Lb tanh 2p b , (17) distributions displayed in Fig. 2; in this regard, we make the
1 þ KR 1 þ KR Lb
further assumption of neglecting the wave setup in I, though in
where Lb is the wavelength at incipient breaking, calculated the offshore zone the mean water level actually fluctuates, owing
through the linear dispersion relationship. to wave reflection.
Iwata and Kiyono (1985) criterion has been chosen, because it Altogether, under the hypotheses enumerated earlier, Eq. (14)
explicitly includes effects of reflection, which dominates the wave results in a simple second order equation in dmf. It has a unique
motion seaward the breakwater crown. The formula returns the positive solution, which for the simplest case of flat bottom
well-known Miche breaking criterion for progressive waves (hIhIh) reads:
(KR ¼ 0) and the Daniel’s (1952) formula for standing waves qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2
(KR ¼ 1). It is noteworthy that in deriving Eq. (17) wave shoaling dmf ¼ 0:5 b þ ðb  4cÞ , (19)
on the front slope has been neglected; this is because it has been
reasonably assumed the breakwater face to be so short and steep where (Fig. 2)
that waves have no time to change their height according to the
abrupt variation of water depth (see Appendix A for further b ¼ ð2h  AÞ, (20)
discussion). Additionally the front face of the breakwater has been 


thought to be a quasi-antinode of the PSf taking place in the x þB h þ Rc
A¼ 1þ b hc  xb b , (21)
Region 1 of Fig. 2. Ls Ls
If from Eq. (17), a value of hb less than |Rc| is obtained, then

waves will not collapse onto the front slope; indeed the breaking 1 4kh
c ¼  H2i 1 þ ð1 þ K 2R  K 2T Þ. (22)
might occur at the breakwater crest if Hi exceeds the Hur et al. 8 senhð2khÞ
 ARTICLE IN PRESS

1020 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

In Eq. (21), xb is the distance between the breaking point and As far as the computation of qin is concerned, we already
the seaward edge of the structure crown, which is known from mentioned that Dalrymple and Dean (1971), and Loveless et al.
Eqs. (17) and (18). Note that xb is always positive. (1998) as well, somehow assumed the barrier to behave like a
For the sake of simplicity, we take xb ¼ 0 as the waves break on weir, that seems more appropriate for emerged breakwaters.
the crown (Eq. (18)), although several studies (e.g. Hattori and On the other hand, the authors noted, by inspection of a good
Sakai, 1994) indicate wave collapse to initiate somewhat forward. deal of video recordings (see Pasanisi et al., 2006 for details), that
breaker profiles on a submerged breakwater do not differ
substantially from those observed on sloping beaches; for well
3.3. The continuity setup
submerged barriers and steep waves, spilling–plunging breakers
take place on the crest (Fig. 3(a)), while, for longer periods, bore-
As widely reasoned in the literature, a leading point of 2D wave
like waves with an evident surface roller seem to occur (Fig. 3(b)).
setup behind submerged breakwaters is the process of mass
Finally, as soon as submergence reduces, plunging or collapsing
transport related to the wave–structure interaction. When waves
breakers may take place on the front face, depending on both
travel across the barrier and eventually break, an amount of water
slope angle and permeability (Figs. 3(c) and (d)).
is pumped forward, that must be conveyed back, by a return
With this in mind, and owing to a substantial lack of
current, to ensure the mass continuity.
knowledge for the specific case of submerged breakwaters, the
Unlike D2, we assume the interaction between return current
general theory of mass transport in the surf zone has been
and barrier, even if it is impermeable, to cause a shear stress that
referred to for calculating the net influx.
must be compensated by an increase of the hydrostatic thrust in
In general, according to Svendsen (1984), the total mass drift of
the sheltered area. In this approach, a hydraulic resistance law is
a breaker of height H, including both the orbital motion and the
chosen for calculating the ‘‘continuity setup,’’ dc, that is
surface roller contributions, can be expressed as
q2in rffiffiffiffiffi  
dc ¼ 2
Beq . (23) g 2 Ar h
f jRc j10=3 qin ¼ H B0 þ 2  , (24)
h H L
Eq. (23) is known as the ‘‘Gauckler–Strickler formula’’ for
uniform turbulent flows; in this expression, qin is the volume where h* is a generic water depth, L* represents the wavelength at
influx to be balanced and f represents a friction parameter, inverse depth h* and Ar is the area of the surface roller in the vertical
of the Manning’s roughness factor, which has the dimensions of a plane; B0 is a shape factor that for sine waves equals 0.125 and
(length)1/3 divided by a time. diminishes as soon as the peakedness of the waves increases
The formula assumes the backflow to dominantly occur (Buhr-Hansen, 1990). Regarding the surface roller contribution,
between the crest of the breakwater and the still water level; Okayasu (1989) proposed setting Ar ¼ 0.06  H  L*, that leads to
accordingly |Rc| is used as hydraulic radius. Furthermore, the qffiffiffiffiffiffiffiffi  
 H
actual trapezoidal shape of the cross-section is replaced by an qin ¼ gh H B0  þ 0:06 . (25)
h
equivalent rectangular one, the width of which, Beq, equals the
ratio between the area of the cross-section, Ac and the breakwater When introducing Eq. (25) into Eq. (23), the water depth, h*, is
height, hc. approximated with the submergence |Rc| and the average wave

Fig. 3. Breaker types at a submerged breakwater: (a) spilling–plunging, (b) bore, (c) collapsing, and (d) plunging on the front slope.
 ARTICLE IN PRESS

M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028 1021

height along the structure, H ¼ Hi ð1 þ K t Þ=2, is used for H. As far as the friction factor f is concerned, predictions are even
The latter because as the wave decays across the barrier, the more complicate, mainly because of the lack of knowledge about
mass transport reduces as well and the mean return current the magnitude of the shear stress exerted by a submerged
(undertow) that locally balances the mass influx, induces a breakwater on the neighbouring fluid. However, with the aim of
unitary shear stress on the structure, which also decreases along deriving a first rough estimate, a number of calculations have
the breakwater. Therefore, since Eq. (23) basically deals with the been tentatively performed, using different approaches. Results
integral (resultant) value of the shear stress, the mean wave are summarized in Table 1. Rows 1–3 refer to empirical formulae
height is used in the computation at a first approximation level. calibrated on natural channels with coarse bed material and no
B0 and f are left as unspecified parameters to be calibrated from significant vegetation (Lang et al., 2004). In these formulae, which
experimental data; accordingly their values will smooth the are not consistent from a dimensional point of view, R is the
effects of the crude hypotheses we introduced earlier. However, hydraulic radius and Dx represents the diameter of bed material,
because they play a key role in the proposed model, a general for which x% of the diameters are equal or smaller. The fourth
discussion about their order of magnitude could be of course column of the table reports the maximum D50 used for the
useful. This issue is addressed in the following section. calibration of the formulas.
Rows 4–5 refer to a more specific coastal engineering context.
In those research, the unitary shear stress at the bottom, tb, has
3.4. Expected values of B0 and f
been modelled by the expression 12 rlU 2b , where l is a non-
dimensional friction coefficient and Ub is the bottom velocity.
It has been already stated that B0 is a wave shape factor, which
Reference values for l have been also provided by the authors.
parameterizes the peakedness of waves. Although no theory exists
Now, it is noteworthy that in this study we have implicitly
for predicting its value in the surf zone, we may obtain some
assumed tb ¼ rgRSw, where Sw is the slope of the energy grade
useful indication through the cnoidal wave theory, which gives B0
line; consequently, by using the Gauckler–Strickler formula for
as a function of the Ursell parameter UR. Here the following
calculating Sw and by equating the two expressions of tb, we
approximate expression can be employed:
readily come to the relationship between l and f reported in
 
11:40 Table 1. Clearly, it holds only in the frame of a fully turbulent flow.
B0 ¼ 0:125 tanh pffiffiffiffiffiffi , (26)
UR In calculating the expected values of f, it has been obviously
assumed R|Rc|; moreover, since we are looking for the order of
which is accurate within 72% for URo2000. Buhr-Hansen (1990)
magnitude of friction factor, it has been realistically set |Rc|1/6E1
found the equation above to furnish a reliable estimate of B0 at the
and |Rc|/DxE1, for whatever x.
breaking point; hence we might reason it could be used as a
Altogether the inspection of Table 1 reveals that order of
reference for narrow crested breakwaters, where wave transfor-
magnitude of f should be likely 10 m1/3/s. Note that this value
mation after breaking is expected to be small.
refers to prototype conditions; it must be properly scaled in
If we assume the breaking point to be located not far from the
laboratory experiments. In this regard we note the since the actual
crest, then Ursell factor can be written as:
dissipative processes associated with the shear stress are of a
HL2 Hi ½gT 2 ðHi þ jRc jÞ purely turbulent nature, they should be in similitude in an ideal
UR  3
ffi Froude-scaled geometrically undistorted model. Consequently, the
h jRc j3
 2   scale ratio of f should be N1/6
L (NL is the prototype to model length
2p Hi H
¼ 1þ i , (27) scale ratio), though we cannot exclude, in principle, the presence of
s0 jRc j jRc j
transitional boundary layers in the small-scale experiments.
where it has been set L ¼ [g(Hi+|Rc|)]0.5T and s0 ¼ Hi/L0. If the However, to avoid any confusion and facilitate the application
latter varies in the realistic interval 0.01–0.04, B0 should not be of the model, in the following all the experimental results are
larger than 0.039–0.071, for Hi/|Rc| is hardly less than 1 in practice. discussed with reference to the prototype conditions; a discussion
It is also clear that as soon as the structure thickness becomes on possible scale effects is given at the end of the paper.
large, previous values may be no longer valid; this because wave Before concluding this section, it should be also highlighted
height evidently undergoes a deep transformation over the crown, that in our approach f accounts the energy loss due to the
the effects of which on B0 are very difficult to foresee. interaction between the return current and the structure, in a

Table 1
Expected values of the friction factor f

Author(s) Formula. Valid for Max D50 (m) Expected for (m1/3/s)

Limerinos 1:16 þ 2:0 logðR=D84 Þ Natural channels 0.253 10

f ¼
0:113R1=6

Griffiths 0:760 þ 1:98 logðR=D50 Þ Natural channels 0.301 7

f ¼
0:113R1=6

Phillips and Ingersoll 1:46 þ 2:23 logðR=D50 Þ Natural channels 1.181 13

f ¼
0:113R1=6
sffiffiffiffiffiffiffiffiffiffiffiffi
Nelson (1996) 2g Coral reefs – 10–14
f ¼ ; l ¼ 0:1  0:2
lR1=3

sffiffiffiffiffiffiffiffiffiffiffiffi
Lamberti et al. (2007) 2g LCBs Qualitative suggestion 7–9
f ¼ ; l ¼ 0:25  0:35
lR1=3
 ARTICLE IN PRESS

1022 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

‘‘global manner’’. Then also permeability is expected to affect its Calabrese et al. (2005), whereas 64 are discussed here for the
value, although it has been formally ignored when the model has first time. The flume where experiments have been carried out is
been derived. In general, the larger the permeability, the lower the 23.50 m long, 0.5 m wide and 0.75 m high; the wave generation
setup is expected to be; this because of the increase of the flow system, capable of generating both regular and irregular waves, is
section available for backflow, that reduces transport velocity and, provided with an active wave absorption system. Two models of
accordingly, the shear stress. This effect should also prevail on the homogeneous rubble mound submerged breakwaters have been
expected increase of wave influx due to more transmission that, tested with waves of different heights and periods. The models
oppositely, would increase the wave setup. differed by crown width (0.25 and 0.35 m, respectively) and by
In short, the variation of structure permeability leads crest freeboard (0.065 and 0.055 m, respectively). The front slope
to a variation of the shear stress induced by the structure on was 1:2 and the rear one was 1:1.5 for both the structures. To
the surrounding fluid; in the model, this effect is accounted investigate the effects of the permeability, tests with the shorter
through a variation of friction parameter f, which would increase model have been repeated after having inserted a wooden
(less friction) with increasing permeability. More details on this impermeable plate in the body of the barrier. Two different
matter are given in the subsequent sections. heights of the plate have been employed, namely hc and hc/2.
In all the experiments, passive wave absorbers were located at
the end of the flumes and wave profiles were acquired in front and
4. Experiments
at the rear of the structures. The wave setup in the sheltered area,
d, has been obtained by time averaging wave profiles in the
With the aim of calibrating and validating the new model, as interval where the mean water level was seen to be constant.
well as for investigating its applicability to random seas, four data Actually, before this stationary condition is reached, the mean sea
sets, coming from experiments conducted at three different level progressively ramps up due to a temporary misbalance
laboratories, have been gathered. The whole database sums 362 between influx and backflow.
hydraulic model tests. For Bristol’s regular wave tests, incident and reflected wave
The Bristol small-scale experiments (scale 1:20) have been heights were separated by using a single probe mounted on a
carried out at the Hydraulics Laboratory of University of Bristol’s trolley and moved offshore the breakwater along nine positions
Civil Engineering Department. The experimental study has been over half the incident wavelength. For Bristol’s random experi-
completed for the UK Ministry of Agriculture, Fisheries and Food, ments and UoN, the Mansard and Funke (1980) technique was
and it is described in details in Loveless and Debski (1997). The applied; as far as GWK tests are concerned, the presence of wave
flume used to perform the tests is 15 m long, 1.5 m wide and 1.1 m breaking on the foreshore was thought to significantly affect the
high. Altogether, eight different models were tested, including reflection analysis, and consequently KR were not evaluated. Here
variations in crest width, still water level, slope angles and rock the incident wave height was estimated (spectrally) by repeating
size. All the cross-sections were homogeneous. Most of the the tests in the absence of structure, and by retaining only the part
experiments were conducted with regular waves, but some of the power spectrum at frequencies larger than half the offshore
irregular wave tests were also run. Results from regular wave peak one. Transmitted waves were measured at a single location
experiments were used for calibrating the Loveless et al. (1998) for Bristol and UoN and at three different points for GWK; for
formula (Eq. (12)). Although both positive and negative clearances these tests, both the wave setup and the transmitted wave height
were tested, only the experiments conducted on submerged (calculated from the high frequency spectrum) were defined by
breakwaters have been used here. averaging results coming from the three probes.
Large-scale GWK experiments (approximate scale 1:2) were Table 2 provides a summary of the main characteristics of the
conducted by the writers at the Grosser WellenKanal of Hannover data sets. The suffix ‘‘i’’ stands for ‘‘incident’’, while Lp and Hrms,
(Germany), in the frame of the EU project Submerged and Low- respectively, represent the peak wavelength
Crested Breakwaters in Presence of Broken Waves (Calabrese et al., pffiffiffiffiffiffiffiffiffiffi and the ‘‘energeti-
cally equivalent’’ wave height Hrms ¼ 8m0 , where m0 is the zero
2002). GWK flume is 307 m long, 5 m wide and 7 m deep, and order spectral moment. For UoN, Hrms has been used for
water waves up to a height of 2 m can be simulated under a approximating the crest to trough wave height, H, either incident,
‘‘quasi-prototype’’ condition. The facility is also equipped with a or reflected, or transmitted. For Bristol regular wave experiments,
‘‘double loop’’ online active absorption system. The structure a zero crossing method has been apparently used for estimating
cross-section consisted of a core and an armour layer; two crest both the incident and the reflected wave heights, whereas Hrms
widths were considered, namely 1 and 4 m (2 and 8 m at a has been used for the transmitted one.
prototype scale) and both the front and rear slopes were 1:2. The
models were subjected to random waves ranging from prebreak-
ing to broken. Although both submerged and subaerial config- 5. Performances of the new method under regular waves
urations have been tested, only underwater structures have been
considered for the scopes of the present work. 5.1. Permeable (conventional) breakwaters
Supplementary small-scale regular wave tests (scale 1:20)
were conducted at the University of Naples ‘‘Federico II’’ (Pasanisi The new model has been initially best-fitted to Bristol regular
et al., 2006; Di Pace, 2006); they will be referred to as UoN tests wave experiments and the values of 0.02 and 6 m1/3/s have been
hereafter. Twenty of them have been already presented in estimated for B0 and f, respectively. On the whole the value of

Table 2
Database characteristics (prototype values)

Data set (wave type) D50,a (m) D50,c (m) Front slope Rear slope hI/L(p)i Rc/H(rms)i B/H(rms)i Nr.

Bristol (reg.) 0.7–1.0 – 0.33–0.5 0.5–0.67 0.10–0.42 0.3–3.3 1.1–13.2 216

Bristol (ran.) 0.7–1.0 – 0.33–0.5 0.5–0.67 0.08–0.23 0.5–3.4 3.6–10.6 37
GWK (ran.) 0.44 0.14 0.5 0.5 0.06–0.13 0.3–1.1 1.5–10.9 25
UoN (reg.) 1.2 – 0.5 0.67 0.09–0.23 0.4–1.1 2.0–5.3 84
 ARTICLE IN PRESS

M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028 1023

roughness parameter is consistent with our previous estimates, 0.50

while B0 is somewhat lower than what we expected. However, New Method
such a value of wave shape factor is not unrealistic, seeing that in 0.40
Loveless et al. formula
analyzing variation of B0 across sloping beaches, Basco and Perfect agreement
Yamashita (1986) found a value of B0 very close to 0.02 for
plunging breakers (at the breaking point). 0.30

δcalc. [m]
Predicted and measured values of d are compared in Fig. 4,
which reveals a reasonable agreement, although some residual 0.20
scatter exists; the latter might be partly due to the aforemen-
tioned non-homogeneity in the wave height definition.
0.10
However, when compared with Loveless et al. (1998) formula,
which has been also calibrated on these data, the new method
seems only slightly more reliable, giving a standard error, SE 0.00
(standard deviation of difference between measured and pre- 0.00 0.10 0.20 0.30 0.40 0.50
dicted values), of 0.080 m vs. 0.087 m and an R2 ¼ 0.84 vs. 0.82. δmeas. [m]
Nevertheless, when UoN data are dealt with (homogeneous
breakwaters only without wooden plate; 42 data), it can still Fig. 5. The new model and Loveless et al. (1998) formula compared with UoN data.
provide good predictions, with the same values of parameters,
while Loveless et al. (1998) formula returns significant under-
estimates (Fig. 5). Table 3
In Table 3, the performances of all the prediction methods Reliability of prediction methods (database size: 320 data)
reviewed in Section 1, as well as of our approach, are compared.
Prediction method Data BIAS (m) SE (m) R2
Together with SE and R2, also the bias, mean of differences
between measured and predicted values of wave setup, is New method UoN 0.006 0.035 0.89
considered. This is to measure the tendency of formulae at All data 0.004 0.076 0.84
underpredicting (positive bias) or overpredicting (negative bias) Diskin et al. (1970) UoN 0.045 0.073 0.65
All data 0.042 0.105 0.75
the experimental data. Whether UoN or all data are considered, D2 UoN 0.020 0.050 0.78
the new formula seems the best performing, as it is practically All data 0.007 0.089 0.74
unbiased, with the minimum SE and the maximum R2. As Loveless et al. (1998) UoN 0.068 0.058 0.78
previously observed, the equation of Loveless et al. (1998) is All data 0.009 0.089 0.79
Diskin et al. (1970) with Hi UoN 0.024 0.049 0.81
probably too empirical, while D2 seems to have a reasonable
All data 0.021 0.088 0.79
predictive power that confirms the effectiveness of the authors’ New method with no reflection UoN 0.002 0.038 0.87
approach. All data 0.001 0.076 0.81
As far as Diskin et al. (1970) formula is concerned, we note it to
be more reliable when the incident wave height, Hi, is used (see
‘‘Diskin et al. (1970) with Hi’’, in Table 3) instead of the deep water (recently some indications have been provided by Zanuttigh and
one, H0. It is also worth noting that the quality of predictions van der Meer, 2006), wave reflection might be neglected in
seems to be little influenced by wave reflection. This is clear when practical applications.
reliability indexes of Table 2 last row are compared with those of Before proceeding with discussion it is necessary to remark
the ‘‘complete’’ model (first row). The reason why this happens is that present work has not commented on the model proposed by
because KR acts on two terms of the momentum equation, which Bellotti (2004), which uses the structure induced wave setup as a
tend to balance. From a side the larger the reflection coefficient, variable to address the problem of the rip currents through the
the larger the shoreward oriented force at the vertical plane I due gaps of a segmented system of submerged breakwaters. This is
to radiation stress (Eq. (15)); otherwise, with increasing reflection, because, as a consequence of having employed the shallow water
the breaking point moves seaward (Eq. (17)), causing an additional equations, Bellotti’s method includes a friction term (shear stress),
thrust, offshore directed, exerted by the structure on the which vanishes as soon as the structure length increases. This
volume of fluid. Consequently, by also considering the lack of makes the model inherently inadequate for 2D flow conditions
reliable equations for calculating KR at submerged breakwaters (i.e. when the structure length is theoretically infinite), where it
has been seen to give heavy underpredictions. However, the
1.20 authors are aware that Bellotti is currently improving his method,
Perfect agreement just to cope with this problem. Once the new model is published it
1.00 data will be extremely interesting to compare its predictions against
our method.
0.80
δcalc. [m]

5.2. Effects of permeability on friction parameter f

0.60

0.40 As mentioned earlier in our approach the friction parameter is

expected to become lower (that means more friction) as soon as
0.20 the structure permeability reduces. This basically accounts the
increasing of the shear stress exerted by the structure on the
0.00 water because of the reduction of the flow section available for
0.00 0.20 0.40 0.60 0.80 1.00 backflow. In other words, when permeability of the structure
δmeas. [m] decreases, instead of reducing the hydraulic radius in Eq. (23), we
decrease the friction coefficient f, leaving the hydraulic radius
Fig. 4. The new model compared with Bristol’s regular wave data. (which has been set equal to |Rc|) unchanged. The relation
 ARTICLE IN PRESS

1024 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

between the roughness factor and the hydraulic conductivity of evidently not guaranteed. Hence, a modification of model
the barrier has been partially studied by using the two series of parameters might be required to counteract this effect. On the
UoN experiments where two wooden plates were installed into other hand, it is of interest that Loveless and Debski (1997),
the cross-section of the breakwater. The heights of the plates, hP, comparing results of the Bristol regular and random wave
were equal to the breakwater height, hP/hc ¼ 1, and to half of the experiments, concluded that ‘‘the results of regular wave tests
breakwater height, hP/hc ¼ 0.5, respectively. When the formulae may be used with reasonable accuracy to predict the setup due to
have been best-fitted to the data, the values f ¼ 5.0 and 3.8 m1/3/s random wave series of average wave height equal to the regular
have been found for the semipermeable and impermeable wave height’’.
structure, respectively. This confirms that the reduction of As a consequence of the above statements, two ‘‘equivalent
hydraulic conductivity leads to an increase of wave setup. wave heights’’ have been considered, namely Hrms and the average
Predictions of the ‘‘re-calibrated model’’ are compared with wave height, Hm; p note
ffiffiffiffiffiffiffiffi that under the hypothesis of Rayleigh
measured values in Fig. 6(a), while the friction parameter is distribution Hm ¼ p=4 Hrms . Clearly, the equivalence applies to
reported in Fig. 6(b), as a function of hP/hc. both the incident and the transmitted wave heights.
Despite the very good agreement shown in Fig. 6(a), which, Regardless of the selected wave height statistics, the peak
however, underlines the model capability of fitting the experi- period, Tp, has been used, since, according to our hypothesis, it
mental data, previous results must be considered but general remains nearly constant across the structure. The transmission
indications that are of a purely qualitative nature. This is mainly coefficient, defined as the square root of the transmitted to
because the permeability has been here modulated without incident wave energy ratio has also been treated as an invariant of
changing the porosity; oppositely in most of the practical cases, the problem, and finally, wave reflection effects have been
the reduction of hydraulic conductivity is accompanied by a neglected for the sake of simplicity.
reduction of voids volume. Thus, more experiments are recom- However, even after the above assumptions have been
mended to develop relationships with an acceptable level of introduced, the problem still remains to define the ‘‘breaking
robustness. point’’ of the representative wave. In this sense, it is widely known
that for a random wave parameter (e.g. the significant wave
height), the locution ‘‘incipient breaking condition’’ holds but in
6. Adaptation of the new model to random seas the sense that when a given amount of waves in the sea-state have
physically broken, the wave descriptor starts to decay in a way
Since this study deals with an engineering approach to quite similar to a regular wave. This is shown in Thompson and
computing wave setup, the adaptation of predictive equations to Vincent (1985) and, above all, in Kamphuis (1991). Now the sense
random waves will be pursued to seek a single wave descriptor of looking for a random breaking criterion to introduce in the
(wave height and period), which is representative of the entire momentum balance Eq. (15) is that we suppose the wave setup
sea-state both seaward and leeward of the breakwater. In this (and all the forces related to it) to be statistically negligible before
regard, Thompson and Vincent (1985) argued that there is no the equivalent wave height has commenced to reduce.
intrinsic relationship between regular and irregular wave para- As far as the quantitative indications are concerned, Kamphuis
meters, and accordingly their equivalence depends on which wave (1991) noted the ‘‘breaking point’’ of Hrms to be the same as for the
feature may be considered as the primary concern. In the opinion significant wave height and proposed the following limiting
of the authors, the choice of the representative wave quantities for criterion for straight slopes:
the present matter can only be established on an empirical basis.
   
This is because the wave setup development in the protected area Hrms 1 Hs
¼ pffiffiffi ¼ 0:40 expð3:5mÞ, (28)
has been seen to depend on both wave energy dissipation and h b 2 h b
mass
p ffiffiffiffiffiffiffiffiffiffitransport processes; consequently, if, for instance, Hrms ¼
8m0 is chosen, an equivalence between regular and irregular where m represents the bottom slope. Because of the lack of
wave energy is ensured, but the equivalence of mass transport is knowledge for the specific case of a submerged breakwater,

0.40 8.00
hp/hc = 0.5
hp/hc = 1
0.30 Perfect agreement 6.00
f [m1/3/s]
δcalc. [m]

0.20 4.00

0.10 2.00

0.00 0.00
0.00 0.10 0.20 0.30 0.40 0.00 0.30 0.60 0.90 1.20
δmeas. [m] hP/hc

Fig. 6. Effect of permeability on friction parameter, f: (a) comparison between the new method and data, and (b) f as a function of structure permeability.
 ARTICLE IN PRESS

M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028 1025

Eq. (26) has been employed as a breaking criterion for Hrms; we set 7. Discussion and conclusions
m ¼ tan aoff for the front slope and m ¼ 0 when wave collapsing
onto the breakwater crown is considered. Regarding Hm, since it is This study has presented a new method for calculating the 2D
only weakly smaller than Hrms, we may reasonably suppose its wave setup behind a submerged breakwater. Starting from a
breaking point to be the same. This leads to conceptual scheme originally proposed by Dalrymple and Dean
  rffiffiffi  (1971) (D2, Eq. (5)), the method decomposes the whole setup into
Hm p Hrms two parts, namely the ‘‘momentum flux’’ contribution, dmf, and
¼ ¼ 0:35 expð3:5mÞ. (29)
h b 4 h b the ‘‘continuity setup’’, dc; the former is related to the reduction of
wave momentum flux forced by wave breaking, whereas the latter
Application of the new procedure has shown that when the is associated with mass transport process. The momentum flux
energetically equivalent wave height is used, a slight recalibration contribution has been calculated by applying the balance of the
of the friction parameter is needed, its best-fit value being equal to horizontal momentum to a control volume surrounding the
8.5 m1/3/s. This was partly expected. Otherwise, when the mean structure; two different breaking criteria have been used,
wave height is chosen, it returns good estimates of wave setup, depending on whether the wave collapse occurs at the front face
without changing any coefficients. That basically confirms the or at the crown of the barrier, and both the transmission and the
findings of Loveless and Debski (1997). In both cases, a very reflection coefficients have been explicitly included in the
satisfactory agreement is detected, SE being around 0.024 m and predictive equations (Eqs. (19)–(22)). For ‘‘continuity setup,’’ dc
R2 close to 0.95 (Figs. 7 and 8). This is somewhat surprising, given (Eq. (23)), it has been interpreted as an additional specific
the supplementary and rather crude hypotheses we had to hydrostatic thrust required to counteract the shear stress due to
introduce. the interaction between the return current and the structure.

0.40
Bristol
0.35
GWK
Perfect agreement
0.30

0.25
δcalc (m)

0.20

0.15

0.10

0.05

0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
δmeas (m)

Fig. 7. Adaptation of the new method to random seas using Hrms. f ¼ 8.5 m1/3/s.

0.40

Bristol
0.35
GWK
0.30 Perfect agreement

0.25
δcalc (m)

0.20

0.15

0.10

0.05

0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
δmeas (m)

Fig. 8. Adaptation of the new method to random seas using Hm. No recalibration of parameters.
 ARTICLE IN PRESS

1026 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

Accordingly, the influence of breakwater permeability has been behind a submerged breakwater that completely closes the bay
accounted through the ‘‘global’’ friction parameter f. where it is located, other physical interpretations may be possible.
Although derived for regular waves, the method has been We observe that under the hypotheses of normally incident
heuristically extended to random seas by simply using appro- waves and straight parallel depth contours, the middle section of a
priate sea-state descriptors, namely Hrms (or Hm) and Tp, and a breakwater system, of whatever (but finite) length, may be
random wave breaking criterion (Eqs. (28) and (29)). More than considered as a symmetry axis of the problem; accordingly, the
350 experiments from three different laboratories have been used flow motion will be there nearly 2D and, if effects of diffraction
to calibrate and verify the model, which seemed to fit well the can be considered negligible, wave setup should approach d2D.
data, often with a very satisfactory approximation (Figs. 5, 6(a), 7 Since the rise of the mean water level at the breakwater heads
and 8). However, three additional points deserve to be discussed; (or at gaps) is quite low, or possibly slightly negative, we see d2D
two of them concern model features, while the third is related might represent an estimate of the potential alongshore variation
to how present results may be used in the future research of the mean water level by which a given amount of water is
developments. conveyed parallel to the breakwater and then offshore; such a
First, we have to note that model application is inherently conclusion is consistent with the aforementioned field observa-
limited by the value of crest freeboard, Rc. As soon as the crest tions by Dean et al. (1997).
level approaches, and possibly overcomes the still water level, the Hence, the authors believe that d2D might be used as a variable
structure tends to behave more or less like a weir and the basic of simplified design equations, which aid engineers in calculating
undertow scheme we adopted for dc fails; note that for Rc ¼ 0, Eqs. important flow parameters such as the alongshore velocity behind
(19)–(22) return an infinite wave setup. Hence, we recommend the structure or the flow speed through the gaps. The equations to
extreme caution when applying the model out of the experi- be developed, which require a proper experimental calibration
mental ranges reported in Table 1, and particularly for Hi/Rc and validation, could be both empirical formulae and simplified
(or H(rms)i/Rc when random waves are considered) larger than circulation models.
about 3.5 in absolute value. It may be of interest to confirm the To get an example of possible application the reader could refer
above statement looking at the four experimental data reported in to the aforementioned Lamberti et al. (2007) paper, where a
Table 1 of the Diskin et al. (1970) research. The comparison with simple circulation model is proposed, based on the generalized
the new method is reported in Table 4, assuming a 1:20 model Bernoulli theorem to calculate the rip current velocity through the
scale and using d’Angremond et al. (1996) to calculate Kt. gaps of a segmented breakwater system (Eq. (13.92) of the paper).
Predictions of new method are shortly referred to as d2D. As In that model d2D could be used as a potential value of the
expected, the model is seen to give proper estimates until the pressure behind the structure, ruling a potential rip current flow
absolute value of Hi/Rc is slightly larger than 3; for Hi/Rc ¼ 4.27, it speed.
gives a heavy overprediction. A further and more detailed example is given in Fig. 9, where
Another point to be commented is about scale effects. All the the case of a single submerged breakwater is considered. If we
experiments we used in this study come from 1:20 models of assume that d2D =hII 51 and that the structure is short enough in
submerged breakwaters except GWK, being at a quasi-prototype the alongshore direction to render the energy loss by friction
scale. Yet the latter includes multilayered cross-sections, whereas negligible, the maximum velocity of the longshore current flowing
in small-scale experiments homogeneous barriers were em- behind the structure, V, might be obtained from the momentum
ployed. It is then a little surprising that experimental data in balance in the y direction, applied to the volume of fluid shown in
Figs. 7 and 8 compare well with model predictions, though a the graph. By invoking the symmetry condition at the centreline,
unique value of friction parameter f was used (6 m1/3/s). Actually, it can be expressed as
as the permeability of GWK structures is likely smaller than in
Bristol, we expected, according to Fig. 6(b), a lower value of f for rw V 2 hII ¼ rghII d2D  Syy;D þ Syy;t ¼ 0, (30)
those experiments. There are at least three possible explanations
for this behavior. The first is that reduction of structure porosity where Syy,D;t represents the yy component of the radiation stresses
(from nearly 0.4 for homogeneous breakwaters to likely 0.3 for a associated to the diffracted and transmitted waves, respectively.
barrier with a core) causes a variation of permeability much They may be roughly calculated by multiplying the incident
smaller than in the case of an impermeable sheet inserted into the radiation stress by the square of the local diffraction and
breakwater cross-section. The second is that a transitional transmission coefficients, KD and Kt. This also implies that, as
boundary layer took place in the Bristol tests, giving rise to some mentioned before, effects of diffraction are negligible for y ¼ 0.
scale effects. A third possible explanation may be the roughness of Under the further hypothesis of shallow water, we obtain the
the armor units to have been significantly different between the following equations where V is directly related to d2D:
two experimental series. Unfortunately, no definitive conclusions " #0:5
 
can be drawn about this major matter, which needs be deeply V d 1 Hi 2 2
pffiffiffiffiffiffiffiffi ¼ a 2D  ðK D  K 2t Þ , (31)
analyzed in the future research works. ghII hII 16 hII
Finally, it is worth discussing possible applications of the
model, both in the practical engineering and research domains. where a is an empirical correction coefficient (to be experimen-
Although it returns in principle the rise of the mean water level tally calibrated) accounting for all the neglected physics. Note that
in developing a circulation model for barred beaches, Dalrymple
(1978) used the same boundary conditions, that is, 2D motion at
Table 4
the central part of the bar and no setup at the gaps.
New method vs. Diskin et al. (1970) data
Note Eq. (31) holds only if the right-hand side of Eq. (30) is
Test number H0 (m) T (s) H(i *) (m) Rc Hi/Rc K(t**) dmeasured (m) d2D (m) positive; if not, the current will be directed towards the shadow
zone and a salient possibly forms. Note that in accordance with
1 2.82 5.50 2.59 2.00 1.29 0.66 0.14 0.16 the simplified scheme of Fig. 9, a net influx QIN ¼ VhIIl is required
2 3.22 6.71 3.12 1.00 3.12 0.53 0.58 0.63
3 2.12 7.24 2.14 0.5 4.27 0.51 0.48 1.04
to close the mass balance, where l is the width of the zone where
4 2.54 5.46 2.60 0 N 0.35 0.80 N the water flow takes place; Diskin et al. (1970) suggested lE4–6hc.
Moreover, it should be noted that if the flow at the middle section
 ARTICLE IN PRESS

M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028 1027

x
no set-up
y

(ρhIIV2+SYY,D)l (ρghII2D+SYY,T)l
V
l

S
I
M
M
QIN=V hII L
E
T
R
Y
WAVES

Fig. 9. Longshore current behind a single submerged breakwater.

9 1.2

6
H1/3 bc [cm]

H1/3 bc [m]

0.8

3
No shoaling
T = 1s
T =1.5s No shoaling
T = 2s GWK data
0 0.4
0 3 6 9 0.4 0.8 1.2
H1/3 i [cm] H1/3 i [m]

Fig. A1. Wave transformation on the front slope of submerged breakwaters (data at model scale). (I) UoN data: solid symbols refer to permeable breakwater, open symbols
refer to hP/hc ¼ 1. (II) GWK data

could not be considered 2D, then dmf instead of the entire d2D Appendix A. Wave transformation across the front slope of a
might be used. submerged breakwater
Despite the above circulation model being extremely crude, it
clearly shows that the findings of the present study can be used in In deriving our model it has been assumed that if waves did
developing more powerful design tools for submerged break- not break onto the front slope of the breakwater, wave height at
waters. the outer edge of the structure crown equalled approximately the
incident one. In principle, this means that wave shoaling on the
breakwater face is negligible or, more precisely, that effects of
Acknowledgments shoaling and reflection somehow compensate each other. Using
experimental data at our disposal, an effort has been directed to
The large-scale tests in the large wave channel (GWK) of the verify this hypothesis.
Coastal Research Centre FZK have been supported by European In some of UoN experiments (regular wave tests) a wave probe
Community under the Access to Research Infrastructures Action of was mounted at the centre of the barrier crest; since the model
the Human Potential Programme (contract HPRI-CT-1999-00101). was rather narrow (B ¼ 0.25 m) results from this gauge have been
The authors thank Dr. J.H. Loveless (University of Bristol) for assumed to be the same as at the beginning of the crest. To make
supplying the data. G. Festa and L. Gesuele are also acknowledged the analysis more reliable only experiments where the waves
for their support in tests and data analysis. were not seen to break have been considered. In GWK random
 ARTICLE IN PRESS

1028 M. Calabrese et al. / Ocean Engineering 35 (2008) 1015–1028

wave experiments a wave probe was placed just at the beginning Di Pace, P., 2006. Riflessione di onde regolari in presenza di opere a gettata
of the breakwater crest that is excellent for our scopes. (in Italian). Ph.D. Thesis, University of Naples Federico II.
Gourlay, M.R., 1971. Piling-up behind low and submerged permeable breakwaters.
Unfortunately, there were no tests in which waves did not break. Discussion note on Diskin et al. (1970). Journal of Waterways and Harbors
To cope with this problem we employed only the breakwater Division WW1, 219–222.
models with a 0.4 submergence. This is because in those tests Gourlay, M.R., 1994. Wave transformation on a coral reef. Coastal Engineering 23,
17–42 (Elsevier).
most of the waves broke on the crown, many of them landward Gourlay, M.R., 1996. Wave set-up on coral reefs. 1. Set-up and wave generated flow
the probe. on an idealised two dimensional horizontal reef. Coastal Engineering 27,
Results of the analysis are shown in Fig. A1, where the incident 161–193.
Grue, J., 1992. Nonlinear water waves at a submerged obstacle or bottom
wave height (Hi) is compared with that at the breakwater crest topography. Journal of Fluid Mechanics 244, 455–476.
(Hbc). Note that the zero up crossing significant wave height has Homma, M., Hoikawa, K., 1961. A study on submerged breakwaters. Coastal
been used as wave parameter, because it should better account Engineering in Japan 4, 85–102.
Hattori, M., Sakai, H., 1994. Wave breaking over permeable submerged break-
non-linearity effects associated with the wave–structure interac-
waters. In: Proceedings of the International Conference of Coastal Engineering,
tion (Thompson and Vincent, 1985). Altogether experimental data ASCE, Kobe, Japan, pp. 1101–1114.
seem to corroborate our hypothesis, at least on average; this apart Homma, M., Sokou, T., 1959. An experimental study on the submerged break-
from a couple of points belonging to UoN data (they are circled in waters. Coastal Engineering in Japan 2, 103–109.
Hur, D.S., Kawashima, N., Iwata, K., 2003. Experimental study of the breaking limit
Fig. A1I) where a surprising decay of wave height has been of multidirectional random waves passing over an impermeable submerged
detected. This is possibly due to damping effects caused either by breakwater. Ocean Engineering 30, 1923–1940.
seepage or by a slight spilling not surveyed at naked eye. For UoN Iwata, K., Kiyono, H., 1985. Breaking of standing two component composite
and irregular waves. Coastal Engineering in Japan 28, 71–87 (Institution of
the ratio H1/3bc/H1/3i (i.e. the shoaling ratio) varied between 0.71 Engineers, Australia, 2004).
and 1.35, with a 1.05 mean. Total 80% of points lay in the interval Kamphuis, J.W., 1991. Incipient wave breaking. Coastal Engineering 15,
0.9–1.2. As far as GWK random wave experiments are concerned 185–203.
Lamberti, A., 2005. Editorial paper on DELOS project. Coastal Engineering 52,
the shoaling ratio resulted between 1.01 and 1.16, with a 1.06 815–818 (Elsevier).
mean. On the whole present results agree with findings of Gourlay Lamberti, A., Martinelli, L., Zanuttigh, B., 2007. Prediction of wave induced water
(1994) who found, for a coral reef, the shoaling coefficients were flow over and through the structure, of set-up and rip-currents. In:
Environmental Design Guidelines for Low Crested Coastal Structures, Section
included between 0.87 and 1.18 with a mean just over 1.0. 13.5. Elsevier, Amsterdam.
Lang, S., Ladson, T., Anderson, B., 2004. A review of empirical equations for
References estimating stream roughness and their application to four streams in Victoria.
Australian Journal of Water Resources 8 (1), 69–82.
Longuet-Higgins, M.S., 1967. On the wave induced difference in mean sea level
Basco, D.R., Yamashita, T., 1986. Towards a simple model of the wave transition between two sides of a submerged breakwater. Journal of Marine Research 25,
region in the surf zone. In: Proceeding of the 20th International Conference on 148–153.
Coastal Engineering, ASCE, Taipei, pp. 955–970. Longuet-Higgins, M.S., Stewart, R.W., 1962. Radiation stress and mass transport in
Bellotti, G., 2004. A simplified model of rip currents systems around discontinuous gravity waves, with application to surf beats. Journal of Fluid Mechanics 13,
submerged barriers. Coastal Engineering 51, 323–335. 481–504.
Buccino, M., Calabrese, M., 2007. Conceptual approach for the prediction of wave Longuet-Higgins, M.S., Stewart, R.W., 1964. Radiation stress in water waves:
transmission at low crested breakwaters. Journal of Waterways, Port, Coastal a physical discussion with applications. Journal of Deep Sea Research 11,
and Ocean Engineering 133 (3). 529–563.
Buhr-Hansen, J., 1990. Periodic waves in the surf-zone: analysis of experimental Loveless, J.H., Debski, D., 1997. The design and performance of submerged
data. Coastal Engineering 14, 19–41. breakwaters. Report completed for UK Ministry of Agriculture, Fisheries and
Calabrese, M., Vicinanza, D., Buccino, M., 2002. Large scale experiments on the Food, Contract no. CSA 2606, University of Bristol, UK, p. 137.
behaviour of low crested and submerged breakwaters in presence of broken Loveless, J.H., Debski, D., McLeod, A.B., 1998. Sea level set-up behind detached
waves. In: Proceedings of the International Conference on Coastal Engineering, breakwaters.In: Proceedings of the International Conference on Coastal
ASCE, Cardiff, UK, pp. 1900–1912. Engineering, ASCE.
Calabrese, M., Vicinanza, D., Buccino, M., 2003. Wave set-up behind low crested Mansard, E.P., Funke, E.R., 1980. The measurement of incident and reflected spectra
breakwaters. In: Proceedings of the International Conference on Offshore end using a least squares method. In: Proceedings of the 17th Coastal Engineering
Polar Engineering, Honolulu, USA, May 2003. Conference, ASCE, New York, pp. 154–172.
Calabrese, M., Vicinanza, D., Buccino, M., 2005. Verification and re-calibration of an Nelson, R.C., 1996. Hydraulic roughness of coral reef platforms. Applied Ocean
engineering method for predicting 2D wave setup behind submerged break- Research 18, 265–274.
waters. In: Proceedings of the 2nd International Coastal Symposium, Hoepn, Okayasu, A., 1989. Characteristics of turbulence structures and undertow in the
Iceland, June 2005. surf zone. Ph.D. Dissertation, University of Tokyo.
Calabrese, M., Pasanisi, F., Buccino, M., 2006. Sull’interazione tra moto ondoso e Panizzo, A., Briganti, R., van der Meer, J.W., Franco, L., 2003. Analysis of wave
barriere sommerse (in Italian). In: Proceedings of the XXX Convegno Nazionale transmission behind low-crested breakwaters using neural network. In:
di Idraulica e Costruzioni Idrauliche, Rome, Italy, September 2006. Proceedings of the Coastal Structures, Portland, USA, 2003.
Cappietti, L., Clementi, E., Aminti, P., Lamberti, A., 2006. Piling up and filtration at Pasanisi, F., Buccino, M., Calabrese, M., 2006. Observation of breaker types at
low crested breakwaters of different permeability. In: Proceedings of the 30th submerged breakwaters. In: Proceedings of the 2nd Short Course on Coastal
International Conference on Coastal Engineering, ASCE, San Diego, USA, and Port Engineering, Cosenza, Italy, May 2006.
pp. 4957–4969. Ranasinghe, R., Turner, I.L., 2006. Shoreline response to submerged structures:
Dalrymple, R.A., 1978. Rip currents and their causes. In: International Conference of a review. Coastal Engineering 53, 65–79 (Elsevier).
Coastal Engineering, ASCE, Hamburg, pp. 1414–1427. Sawaragi, T., 1995. Coastal Engineering Waves, Beaches, Wave–Structure Interac-
Dalrymple, R.A., Dean, R.G., 1971. Piling-up behind low and submerged permeable tions. Elsevier, Amsterdam.
breakwaters. Discussion note on Diskin et al. (1970). Journal of Waterways and Seji, M., Uda, T., Tanaka, S., 1987. Statistical study on the effect and stability of
Harbors Division WW2, 423–427. detached breakwaters. Coastal Engineering in Japan 30 (1), 131–141.
d’Angremond, K., van der Meer, J.W., de Jong, R.J., 1996. Wave transmission at low- Svendsen, I.A., 1984. Mass flux and undertow in the surf zone. Coastal Engineering
crested structures. In: Proceedings of the International Conference of Coastal 8, 347–365.
Engineering, ASCE, Houston, USA, pp. 2418–2426. Thompson, E.F., Vincent, C.L., 1985. Significant wave height for shallow
Daniel, P., 1952. On the limiting clapotis. Gravity waves: circular no. 521, National water design. Journal of Waterways, Port, Coastal and Ocean Division.,
Bureau of Standards, pp. 35–38. 828–842.
Dean, R.G., Chen, R., Browder, A.E., 1997. Full scale monitoring study of a van der Meer, J.W., Briganti, R., Zanuttigh, B., Wang, B., 2005. Wave trans-
submerged breakwater, Palm Beach, Florida, USA. Coastal Engineering 29. mission and reflection at low-crested structures: design formulae,
Dick, T.M., 1968. On solid and permeable submerged breakwaters. CE Research oblique wave attack and spectral change. Coastal Engineering 52 (10–11),
Report No. 59, Queens University, Ontario. 915–929.
Diskin, M.H., Vajda, M.L., Amir, I., 1970. Piling-up behind low and submerged Zanuttigh, B., van der Meer, J.W., 2006. Wave reflection from coastal structures. In:
permeable breakwaters. Journal of Waterways and Harbors Division WW2, Proceedings of 30th International Conference on Coastal Engineering, vol. 5.
359–372. ASCE, San Diego, USA, pp. 4337–4349.