Coastal Engineering 101 (2015) 69–88

Contents lists available at ScienceDirect

Coastal Engineering
journal homepage: www.elsevier.com/locate/coastaleng

Crest modifications to reduce wave overtopping of non-breaking waves

over a smooth dike slope
Koen Van Doorslaer ⁎, Julien De Rouck, Sarah Audenaert 1, Valerie Duquet 2
Dept. Of Civil Engineering, Ghent University, Technologiepark 904, 9052 Zwijnaarde, Belgium

a r t i c l e i n f o a b s t r a c t

Article history: The formula to quantify the average wave overtopping discharge of non-breaking waves over a dike according to
Received 16 January 2014 the TAW-report (2002) and included in the EurOtop Manual (2007), only contains the influence of the roughness
Received in revised form 6 February 2015 of the dike slope and the obliqueness of the waves on wave overtopping. Unlike the formula for breaking waves,
Accepted 9 February 2015
the reductive effect of a storm wall or a berm is not included in this formula. Over 1000 scale model tests with
Available online 27 April 2015
non-breaking waves on a wide variety of structures have been carried out for this paper. The crest of the dike
Keywords:
slope has been modified with a storm wall, a storm wall and parapet, a stilling wave basin, a promenade, and a
Wave overtopping combination of a promenade and storm wall with or without parapet. For all these modified crests, a reduction
Reduction factors factor has been deducted to include in the wave overtopping formula of non-breaking waves. This paper presents
Storm wall the results of all investigated measures to reduce wave overtopping over a smooth dike slope in an easy and log-
Parapet ical way to serve as a guidance for use by designers. An example is given where all presented measures are com-
Promenade pared to each other, and to other well-known structures.
Stilling wave basin © 2015 Elsevier B.V. All rights reserved.

1. Introduction
Although the Belgian coast has only a length of 67 km, it is a densely
populated area of which every meter is intensively used. A similar trend
exists worldwide: coastal zones belong to the areas with high popula-
tion densities, even though these regions are the most vulnerable to
the risk of storm surges and overtopping. In addition, due to climate
change and sea level rise, an increase in the number and the intensity
of violent storms is expected. As a result, large amounts of water can
overtop which may lead to dike instability and flooding, a disaster for
safety of mankind and economy.
From this perspective, extensive research has been carried out in
order to improve the safety level of the coastal areas. In Belgium, the
Integrated Masterplan for Coastal Safety (IMCS) was approved by the
government in 2011. This masterplan states that the Belgian coastal
zones have to be protected against a storm with a return period of
1000 years and that measures have to be taken in vulnerable areas
(Mertens et al. (2009)).
⁎ Corresponding author at: Dredging, Environmental and Marine Engineering (DEME),
Scheldedijk 30, 2070 Zwijndrecht, Belgium. Tel.: +32 32505211.
E-mail addresses: van.doorslaer.koen@deme-group.com, koenvandoorslaer@gmail.com
(K. Van Doorslaer), j.derouck@hotmail.com (J. De Rouck), Sarah.Audenaert@imdc.be
(S. Audenaert), Valerie.Duquet@sbe.be (V. Duquet).
1
Present address: International Marine and Dredging Consultants (IMDC),
Coveliersstraat 15, 2600 Antwerp, Belgium. Tel.: +32 32709295.
2
Present address: SBE Engineering Consultants, Slachthuisstraat 71, 9100 Sint-Niklaas,
Belgium. Tel.: +32 37779519.
The typical geometry of a large part of the Belgian coastline consists
of a sandy beach under a mild slope (1:100–1:50) followed by a smooth
dike slope (1.5 ≤ cot(α) ≤ 3) and a quasi-horizontal part, further called
“promenade”. Just next to this promenade, apartment buildings are
present at several locations. Before 2004,3 during nearly every winter
storm, the water reached the sea dike in Ostend (Fig. 1), leading to a
high possible risk of wave overtopping and resulting damage. The
IMCS recommends both ‘soft’ and ‘hard’ measures to reduce wave
overtopping. The soft measures are beach nourishments, while the
hard measures can be new constructions on top of the smooth dike
(Fig. 2) slope or the modification of the crest of an existing sea dike.
Whatever measures are designed, they have to take the strict spatial
restrictions into account: increasing the height of the sea dike should
be limited due to visual implications for people living close behind the
coastline, while a landward expansion of the promenade is often impos-
sible due to the presence of apartments and buildings. The construction
of permanent or mobile storm walls with or without parapet, or the
integration of a so called stilling wave basin in the crest of the sea
dike are such measures. The influence of these hard measures on the
overtopping discharges in non-breaking wave conditions forms the
subject of this paper.
3
After 2004, the Flemish government decided not to wait for the IMCS to be published
and works to start, since the risk in Ostend was too high. A beach nourishment was carried
out as “emergency measure” before the improvements as suggested in the IMCS will be
implemented. This prevented waves overtopping the dike slope during normal winter
storms.

http://dx.doi.org/10.1016/j.coastaleng.2015.02.004
0378-3839/© 2015 Elsevier B.V. All rights reserved.
70 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88
Fig. 1. Sea dike in Ostend, at the Belgian coastline, during a winter storm before 2004.
Fig. 2. Artist impression of a coastal dike, after implementation of reductive measures
(promenade and storm wall) to reduce overtopping. Impression by ‘Coastal Division,
Flemish Government’ for IMCS.
Over 1000 tests have been carried out with and without overtopping
reducing measures. This happened during several test campaigns at
Ghent University between 2005 and 2012. The test matrix was based
on hydraulic boundary conditions present along the Belgian Coast, but
opened up to a broader spectrum of relevant parameters to have a
good base for analysis (e.g. from small to large wave periods, from
small to large dimensionless freeboards, etc.). Each year, the knowledge
of the authors increased, and new details and geometries were investi-
gated. All separate measures have been published on international
conferences and/or in local master theses at Ghent University (to be
downloaded from www.lib.ugent.be; see reference list for the indepen-
dent titles). This paper serves as the summary of all these different
reductive measures. The results are presented in the (standard) non-
dimensional log-linear graph, from which reduction factors ‘γ’ for
each individual measure are deducted. The existing formulae according
to EurOtop Manual 2007 are updated, and the data are again plotted in
the log-linear graph to visualize the increased performance due to the
reduction factor γ.
2. Literature study
Several methods can be found in literature to quantify the wave
overtopping discharge over a sea defense structure. The oldest method
is by means of experimental formulae, which have been set up by
testing a scale model of a structure in a wave flume or wave tank. In
the beginning only regular waves were investigated, while since the
1970s datasets using irregular waves were produced leading to a
more realistic simulation of sea states. A large number of data gathered
from different laboratories with many variation in hydraulic and
geometrical boundary conditions, were brought together in Van der
Meer and Janssen (1994). A dimensionless plot of all data led to a best
fitting trend line which allows calculating wave overtopping over a
sloping structure. Results for a wider variety of structures and hydraulic
parameters are published in TAW (2002) and EurOtop manual (2007).
The British guidelines for wave overtopping propose a slightly different
formula, based on a (smaller) data set by Owen (1980).
Besides the study of wave overtopping in an empirical way, also nu-
merical models and prototype measurements were used to investigate
wave overtopping (CLASH project, De Rouck et al., 2009). In this way,
complex structures and scale effects could be studied. Another method
to calculate wave overtopping is by using a neural network (NN), based
on the CLASH database (Fig. 3). Over 10,000 tests on many different ge-
ometries were collected and identified by 31 (geometrical and hydrau-
lic) parameters. By using such a neural network (Van Gent et al., 2007;
Verhaeghe et al., 2008) overtopping can be calculated for a wider range
of geometries and sea states.
In this paper a comparison with the existing empirical formulae of
overtopping over sloping dikes will be made, based on the new data
set of about 1000 tests with non-breaking waves.
2.1. Smooth dike slopes
According to the TAW-report, and included as such in the EurOtop
manual (2007), dimensionless average wave overtopping discharge q
(in m3/s/m) is expressed as a function of the dimensionless freeboard
of the construction. The formula giving the average overtopping
discharge for breaking waves on a sloping dike becomes:
!
q 0:067 R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffiffiffiffiffi
γb
ξm−1;0
exp −4:75
C

g
H 3m0 tanα Hm0 ξm−1;0
γb
γ f
γβ
γ v
ð1Þ
with a maximum for non-breaking waves:
!
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:6
C
: ð2Þ
g
H 3m0 Hm0 γ f
γ β
Overtopping over a smooth impermeable dike with a simple slope
and perpendicular wave attack can be calculated by setting all γ-factors
equal to one. The reductive effects that certain geometrical (γb: berm in
the dike, γf: roughness of the slope, γv: vertical wall on top of the dike)
and hydraulic (γβ: wave obliqueness) variations bring along, can be
accounted for by using the individual values of γ b 1 as defined in the
TAW-report.
The transition from the formula for breaking waves to the one
for non-breaking waves occurs for a wave breaker parameter ξm−1,0
(calculated with spectral wave period Tm−1,0) with a value of about
1.82. In the current dataset, only non-breaking waves (formula (2))
are considered, since the boundary conditions of the model tests are
based on the geometry of the Belgian sea dikes before the improve-
ments according to the IMCS-report were implemented, leading to
non-breaking wave conditions.
Note that new research by van der Meer and Bruce (2013) has im-
proved the Eqs. (1) and (2) mainly for small freeboards Rc/Hm0 b 0.5–
1. For larger freeboards, the updated formulae by Van der Meer and
Bruce lead to similar results as formulae (1) and (2), so they can be
maintained. In our research, mainly the range Rc/Hm0 N 0.5–1 is present.
Consequently our results will be compared with the approach by
EurOtop (2007).
The formula for non-breaking waves (2) does not contain a γv or γb
reduction factor. The fact that the overtopping reducing effect of a
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88
Fig. 3. Parameters from the NN and CLASH database to quantify wave overtopping discharges over coastal structures.
vertical wall or a promenade cannot be taken into account in formula (2)
forms the starting point for the research carried out for this paper.
EurOtop (2007) mentions an alternative overtopping formula devel-
oped by Owen in 1980, based on a smaller dataset than formulae (1) and
(2) (and therefore a smaller range of validity). The average overtopping
according to Owen can be calculated with formula (3):
!
q Q0 RC
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ pffiffiffiffiffiffiffiffi
exp −b
pffiffiffiffiffiffiffiffi :
g
H3m0 s0;m HS s0;m
The coefficients b and Q0 are listed in Owen's paper (1980) and
depend on the slope of the dike.
This formula (3) shows three remarkable differences compared to
formula (2) for the non-breaking waves:
- There is no distinction between breaking waves and non-breaking
waves in the Owen formula (3). This formula overestimates
overtopping for steep slopes.
- Wave overtopping discharge is influenced by the wave period
by means of the wave steepness s0,m according to Owen, while
formula (2) for non-breaking waves shows no dependency of the
wave period.
- The dike slope angle is another active parameter according to Owen
(different coefficients b and Q0 for different slopes) while formula (2)
shows no dependency for non-breaking waves.
Apart from Owen, also the final report of the Dike-3D project
(Kortenhaus et al. (2006)) mentions a possible influence of the wave
period, and they include a correction factor γs0 to account for the
wave steepness in the non-breaking formula. In (2) and confirmed by
many international researchers, the wave period does not influence
the wave overtopping discharge for non-breaking waves on slopes
cot(α) = 2 to 3. Also a recent work by Victor (2012) confirms this; he
states that there is a small influence of slope angle and wave period
for non-breaking waves, but this influence is insignificant within the
range tested in this study (see further, Table 1 and others).
Not the presentation by Owen, but the international standardized
presentation as in formula (2) will be used throughout the paper.
Nevertheless, the influence of the wave period and slope angle on wave
overtopping for non-breaking waves will be checked in the current
data set, and conclusions will be given for each geometry.
2.2. Storm wall and parapet
EurOtop (2007) includes a section on the effect of wave walls on
sloping structures. It is stated that the knowledge on this structure
is limited and only a few model studies were available. Therefore the
71
use of the neural networkis recommended for more reliable calcula-
tions. Nevertheless, a method to calculate the reduction factor γv is pro-
posed in the manual, to include in formula (1), not in (2). A first step is
to calculate the average slope by changing the vertical wall by a 1:1
slope. This allows calculating the wave breaker parameter ξm−1,0 in
order to use the breaking or non-breaking overtopping formula. Next,
EurOtop (2007) mentions a reduction factor γv = 0.65 when a wall is
present, and γv = 1 when no wall is present. This leads to the following
ð3Þ interpolated formula:
γv ¼ 1:35−0:0078
αwall ð4Þ
where αwall is the angle of the steep slope in degrees (between 45° for
a 1:1 slope and 90° for a vertical wall). Even though no factor γv is
included in formula (2) (in contrast to formula (1)), it will be verified
in our data if formula (4) is a good method to predict overtopping of
non-breaking waves over a smooth dike slope with storm wall.
Kortenhaus et al. (2001) discusses Storm Surge Protection (SPP)
wall, which is similar to our dike slope with promenade and storm
wall. In the current study, the promenade is above the SWL, while in
the SPP-project there is a considerable water depth on the promenade,
which is a different hydraulic situation. Instead of an overtopping bore
(like in the current study), waves can face the SPP wall. The focus is
mainly on impact forces on the wall, but the paper by Kortenhaus
et al. also deals with the overtopping reduction shortly. It is mentioned
that a so-called overtopping reducer, which is comparable to what we
call parapet, is an efficient method to further reduce wave overtopping,
especially for dimensionless freeboards Rc/Hs N 1.2. Another reductive
measure is the implementation of “underwater barriers” on the prome-
nade. This looks a bit like the stilling wave basin (SWB), but again with
the difference that the SWB in our study is located above the SWL. One
of the conclusions by Kortenhaus et al. (2001) is that a certain horizon-
tal space should be available between the underwater barrier and the
SPP wall, to allow wave energy dissipation.
A last important finding in the SPP project is that the wave period
has a significant influence on the overtopping rates for the tested ge-
ometries (dike slope with promenade and storm wall). Also Pearson
et al. (2004), which studies the effectiveness of parapets on vertical
walls, claim to see a clear dependency of the wave period on the
overtopping discharges. However, both papers don't work out a ge-
neric method to account for this dependency. As mentioned in the
previous section, this will be verified in the current data set.
Recently, new work by Tuan (2013) was published which gives an
approach to calculate the reduction due to crown-walls on low sea
dikes. Tuan states that a reduction factor for the influence of a vertical
wall on top on the slope has to be included in the non-breaking
formula (2) just as in formula (1). He also introduces a reduction factor
for a promenade, but the length of his promenade is based on typical
Vietnamese coastal dikes and is much smaller (1 m–2 m prototype)
72 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88
and therefore out of the range of this study. He proposes a reduction
factor for a vertical wall as follows:
1 smooth slopes with and without storm wall with parapet. Nevertheless,
γ wall ¼
hwall 1
1 þ 1:60

Rc ξ0m
where hwall is the wall height, and Rc⁎ the difference between the base of
the storm wall and the still water level, which is not in agreement with
the international standard.4 In the current paper, Rc will respect the
international interpretation by defining Rc as the difference between
the highest point of the structure and the still water level. Tuan
(2013) also changes the Rc-value in formula (2) by his Rc⁎.
Furthermore, his study advises to use the actual slope angle α instead
of the equivalent slope angle as is done in the EurOtop manual. The data in
the current paper are also analyzed by always using the actual slope angle.
Tuan (2013) states that the total reduction is always achieved by
multiplying all contributing reduction factors. That is because he only
has data of the combined geometry promenade with storm wall
and no independent data for the promenade alone; the Vietnamese
promenades are simply too small for that. However, making a general
statement that reduction factors can always be multiplied to account
for the combined effect is not correct. The current research shows that
it can be an overestimation of the reductive effect: the final γ by multi-
plication of the independent values can be lower than the real value of
γ, leading to an underestimation of the wave overtopping discharge.
For the dike slope with storm wall, both the EurOtop approach as
Tuan's approach will be compared with the current data set.
When a parapet is available, EurOtop (2007), Cornett et al. (1999),
Kortenhaus et al. (2003), Pearson et al. (2004) only give a reduction
related to vertical walls (caisson, quay wall, …) and not to sloping
structures. Cornett et al. (1999) investigated a caisson with 3 angles of
overhanging wall geometry: 30°, 45° and 60°. They found a γ-factor
starting from 0.9 up to 0.7. A value of γv = 0.7 means a reduction in
wave overtopping up to 95% for Rc/Hm0 = 1.67. Even a relatively modest
overhang inclined 30° with respect to vertical can reduce overtopping
flows by a factor of 10 or more. The extent of the decrease in
overtopping discharges at the overhanging wall was found to be highly
variable, depending on the water level and wave conditions. No generic
method is proposed in Cornett et al. (1999) to calculate the reduction
factor. The other references (Kortenhaus et al., 2003; Pearson et al.,
2004) did develop a generic method to determine a k-factor defined
as the ratio of overtopping discharge with parapet to the overtopping
without parapet. This is unfortunately not applicable here since our
study always has a sloping dike which does not fit the parameters
used by Kortenhaus et al. (2003).
Some studies exist with a storm wall with or without parapet on top
of a rubble mound breakwater, which is a sloping structure. Coeveld
et al. (2006) define a parameter Q' which is similar to the k-factor by
Kortenhaus et al. (2003): the ratio of overtopping discharge of a break-
water with crest element to one without crest element. This ratio Q'
seems to be dependent on a number of parameters:
- An exponential decreasing trend for increasing Rc/Hm0 was found.
- No relationship between Q' and the wave period was found.
- A decrease of the ratio Q' was observed for an increase in nose
length.
- A decrease of the ratio Q' was observed for an increase in crest width.
The ratio Q' is defined as
0
−4
hwall
− 0:4 B NL
− 2H
Q ¼ 1:55
e H m0 H m0 m0
4
Symbols used in Tuan's paper have been changed to avoid misunderstandings.
where hwall is the wall height, B the crest width and NL the length of the
parapet nose. The authors believe that the ratio filters out the roughness
of the rubble mound breakwater and is therefore also applicable on
ð5Þ
Coeveld et al. (2006) state that their data show bad comparison with
the TAW prediction. A better comparison was found to compute the
overtopping over the rubble mound breakwater without storm wall
with the Neural Network approach, and then using formula (6) to
include the effect of the storm wall and parapet. This approach will be
tested on our data.
By adding a storm wall or a parapet on the slope, the slope virtually
increases. The extreme boundary of a virtually increased slope would be
a vertical wall. New analysis of vertical wall data by van der Meer and
Bruce (2013) shows that for small freeboards (Rc/Hm0 b 0.91) the for-
mula by Allsop et al. (1995) and for large freeboards (Rc/Hm0 N 0.91)
the formula by Franco et al. (1994) should be used. This line, with bend-
ing point at Rc/Hm0 = 0.91, will be plotted along with the data of a slop-
ing dike with storm wall and parapet (Fig. 10). Also the line of a steep
slope cot(a) = 1, according to the data analysis by Victor (2012) and
formulae mentioned in van der Meer and Bruce (2013), will be added
to this graph for reasons of comparison.
2.3. Promenade
In this paper, the (quasi) horizontal part at crest level is called
promenade. Keep in mind that it contains a gentle slope of 1% to 2% to
stimulate draining from rainfall and overtopped water towards the
sea. A promenade is different than what is meant by the term “berm”
in the EurOtop manual. A berm is a (quasi) horizontal part in the dike
slope and often located around the design water level, to reduce
the average slope and thereby reduce wave overtopping discharges.
The promenade in this study is at crest level, clearly above SWL.
Thereby, the reduction coefficient as presented in EurOtop (2007)
might not be the best prediction tool. Nevertheless, it will be checked
with our data.
EurOtop (2007) also acknowledges that if the overtopping is not
measured at the end of the slope, but a few meters backwards the
hazard effect of overtopping will reduce. As a rule of thumb, a reduction
qeffective = qcrest/x where x is the backward distance between 5 and 25,
is mentioned. This is however not based on measurements, but just
an indication that the promenade can significantly reduce wave
overtopping effects. It will be further studied in the present data set.
2.4. Stilling wave basin
This paper introduces a number of other crest modifications,
such as a storm wall at the end of a (fairly long) promenade, or a
stilling wave basin. Some literature is available on these topics,
like Geeraerts et al. in the Dike-3D project, but it all belongs to
the research project of Ghent University on the reduction of
wave overtopping, and is thereby included in this paper. On the
stilling wave basin, some independent literature was found on
overspill basins in Italian marina breakwaters (van der Meer and
Bruce (2013)). It shows that a storm wall and an overspill basin
can have similar behavior in terms of reducing average
overtopping discharges, but the overspill basin acts twice as
good as the storm wall regarding individual wave overtopping
events. Both measures can decrease the average discharge up to
a factor of 2 under the tested conditions. Further, it is stated that
the volume of the overspill basin plays a role, but the freeboard re-
ð6Þ
mains the most important parameter. A last important statement
is that the water in the basin should be allowed rapid drainage in
order to maintain the efficiency of the basin. This latter is con-
firmed by Burchart and Andersen (2006).
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 73

Fig. 4. Position of wave gauges in the 2D flume of Ghent University (distances in mm).

3. Test set-up combinations of a promenade with a wall with or without parapet, are
related to this reference situation. A sketch of this reference geometry
The objective to predict the reduction on average overtopping is given in Fig. 5, the arrow with indication ‘q’ shows where the
discharge has been accomplished by means of experimental research overtopping is measured.
carried out in the wave flume (L = 30.00 m, W = 1.00 m, H = 1.20 m) To determine the reduction factors for all different geometries, the
of the Coastal Engineering Department of Ghent University. Waves are measured data are not compared to formula (2), but to a reference
generated using a piston type wave paddle, and the steering of this paddle formula which is determined out of 80 new tests on a smooth dike
features active wave absorption. Each tested time series contained approx- slope as shown in Fig. 5. In this way, the test set-up, measuring devices
imately 1000 waves, in order to obtain reliable average overtopping and -techniques are the same for the reference case as for cases with
discharges. reductive measures, which guarantee good comparison and exclude
Waves are measured using resistance type wave gauges, positioned model effects. The range of parameters of these 80 tests is summarized
as shown in Fig. 4: two in front of the wave paddle (on behalf of the in model values in Table 1.
active wave absorption), three at deeper water, and three in front of All test results are plotted in a semi-logarithmic diagram with the
the structure (at a distance of about 0.4 L from the structure). By dimensionless freeboard (Rc/Hm0) on the horizontal axis and the
 qffiffiffiffiffiffiffiffiffiffiffiffi
means of these groups, incident and reflected wave conditions can be
dimensionless overtopping discharge q= gH3m0 on the vertical
separated from each other and the incoming wave height can be deter-
mined, using the method by Mansard and Funke (1980). The height of axis. An exponential trend line is fitted through the data and gives the
the foreshore is 27 cm, its length is 2 wavelengths. The water depth reference formula to calculate the average overtopping discharge on a
on the foreshore is large enough to avoid wave breaking, see details smooth dike slope under non-breaking wave conditions.
for each geometry. The wave spectra at the deeper water and on the
foreshore are very comparable, no loss of energy takes place.
Wave overtopping is captured by a tray on top of the smooth dike
slope, and lead to a 30 liter basin that is constantly weighed on a bal-
ance. When the basin is full, water is pumped back to the wave flume
in order to maintain the correct water level in the flume during the
test. Total overtopping volume can be deducted from the balance's
weight registration in time. A part of the tests also had individual
overtopping measurements, but this is not further treated in this paper.
Non-breaking wave conditions were tested on a smooth dike slope
(γf = 1) with perpendicular wave attack (γβ = 1). Both a dike slope
1 V:2 H and 1 V:3 H were tested. A few of the test results with ξm−1,0
just above 1.82 gave results that could not be identified as “non-breaking
waves” with full confident. Therefore, in the current data set, the limit
was set on ξm−1,0 N 2.1 to clearly define a test as non-breaking.
The majority of the tests were performed with a JONSWAP 3.3 spec- Fig. 5. Smooth dike slope.
trum while some were performed with a Pierson–Moskowitz spectrum,
both single peak spectra. No influence of the spectrum on the overtopping
volumes could be noticed for the range of dimensionless freeboards and Rc/Hmo
1.0E+00
tested spectra in the current data set. 0 0.5 1 1.5 2 2.5 3 3.5

1.0E-01
4. Results
1.0E-02
q/(g ·Hmo³)1/2

4.1. Smooth dike slopes

1.0E-03
y = 0.2e-2.28x
R² = 0.94
The smooth dike slope serves as a reference situation. The reduction
factors which take into account the influence of overtopping reducing 1.0E-04
measures such as a wall, a parapet, an SWB, a promenade and EurOtop average trendline
5% lower boundary
1.0E-05 y = 0.2e-2.6x
Table 1 5% upper boundary
Summary of the characteristics of the tests (non-breaking waves) on a smooth dike slope Smooth dike slope: measured data
(scale model values). 1.0E-06 Exponential (smooth dike slope: measured data)

Slope angle of the smooth dike slope cot(α) 2 and 3

Fig. 6. Reference data set of non-breaking waves on a smooth dike slope.
Mean spectral wave period Tm−1,0 0.91–2.45 s
Dimensionless freeboard Rc/Hm0 0.83–3.15
Freeboard (top of structure to SWL) Rc 0.12–0.27 m
Spectral wave height Hm0 0.07–0.21 m  
q R
Water depth at toe of the structure d 0.35–0.49 m qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c : ð7Þ
Wave breaker parameter ξm−1,0 2.10–4.90 g
H3m0 Hm0
74
It can be noticed that the scatter around this trend line is small.
When taking the exponential coefficient as normally distributed sto-
chastic variable, the mean value is 2.28 with a standard deviation of
0.15. Additionally, almost all results are located above the average
trend line in EurOtop, 2007 (formula (2)) resulting in a slightly higher
trend line and thus a smaller coefficient in the exponent: 2.28 instead
of 2.6.
If the CLASH dataset is used, but limited to the data with similar
geometry
- β = 0° for perpendicular wave attack,
- γf = 1 for smooth slopes,
- 1.5 ≤ cot(α) ≤ 3.0 for the dike slope,
- Rc N 0 for positive crest freeboards,
- B = 0 for the absence of berms,
- and Gc = 0 for the absence of promenades at crest level; overtopping
is measured direct at the end of the dike slope
- ξm−1,0 N 2.1 like in our own database, in order to only work with
non-breaking waves.
only 472 of the 10,532 tests remain. When also reliability factor RF is set
to one, 255 data points remain, which are shown in Fig. 7. The average
trend line through these 255 tests has a coefficient 2.29, very close to
the reference line of the data from Fig. 6. The data from Fig. 6 are plotted
together with the considered CLASH data in Fig. 7, and they are in line
with each other, which means that the new data are comparable to
what was found in other laboratories with similar boundary conditions.
The CLASH data have larger scatter, probably due to the fact that they
are found in different laboratories all over the world.
The coefficient 2.6 in Eq. (2) is the average coefficient (normally dis-
tributed with a standard deviation of 0.35) of all different geometries
and hydraulic conditions for non-breaking waves over sloping
structures. For reasons of comparison, all reductive measures further
in the paper will be referred to formula (7) since all experiments are
run in the same wave flume. The internationally accepted exponential
coefficient 2.6 in Eq. (2) is not being called into question to predict
wave overtopping of non-breaking waves over smooth dike slopes.
Hence we propose to use the reduction coefficients as proposed later
in this paper in combination with formula (2) with coefficient 2.6 to cal-
culate average overtopping discharges. For deterministic design,
EurOtop (2007) manual advises to use coefficient 2.3. This will be
further explained in Section 5.
As mentioned in Section 2, the new dataset was subjected to a
detailed analysis on the influences of slope angle and wave period on
the mean overtopping discharge. Fig. 8 contains 29/80 data points on
the smooth dike slope where the focus was put on the slope angle and
Rc/Hmo
1.0E+00
0 0.5 1 1.5 2 2.5
1.0E-01
1.0E-02
q/(g ·Hmo³)1/2
1.0E-03
1.0E-04
EurOtop average trendline
5% lower boundary
1.0E-05
5% upper boundary
new data UGent
CLASH limited 255 tests
1.0E-06 exponential (data of this paper)
exponential (CLASH limited 255 tests)
Fig. 7. Comparison between new data and CLASH database on a smooth dike slope.
K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88
Rc/Hm0
1.0E+00
0 0.5 1 1.5 2 2.5 3 3.5
1.0E-01
1.0E-02
q/(g ·Hm0³)1/2
1.0E-03
EurOtop average trendline
1.0E-04 5% lower boundary y = 0.2e-2.28x
5% upper boundary R² = 0.94
Smooth dike slope 1:2 Tm-1,0 = 1.64
1.0E-05 Smooth dike slope 1:2 Tm-1,0 = 2.36 y = 0.2e-2.6x
Smooth dike slope 1:3 Tm-1,0 = 1.64
Smooth dike slope 1:3 Tm-1,0 = 2.36
1.0E-06 Exponential (smooth dike slope: all measured data)
Fig. 8. 29 data points on a smooth dike slope split up by wave period and slope angle α.
wave period, where the red trend line is the same one as in Fig. 6
based on all 80 points.
- Slope angle: when comparing the slope angles cot(α) = 2 versus
cot(α) = 3 the mildest slope is slightly more overtopped since
this slope reflects less energy and allows more overtopping. Another
explanation is that the layer thickness of the “tongue” of the
overtopping wave is somewhat larger for milder slopes, giving a
larger overtopping discharge (Bosman et al., 2008). The difference
between both slopes is however almost negligible in the obtained
data, from which it can be concluded that the slope angle only has
very limited influence within the used range of tested parameters.
This finding is confirmed by Victor (2012); wave overtopping has a
weak dependency on the slope angle for mild dike slopes
(1.5 ≤ cot α ≤ 3).
- Wave period: on a slope 1:2 there is no influence of the wave period
on the average overtopping discharge, while on slope 1:3 a minor
difference exists: an increase of wave overtopping occurs for
increasing wave period. This is in line with the findings by Victor
(2012); an (minor) influence of the wave period on wave overtopping
exists, but its influence is limited compared to the effect of the relative
crest freeboard.
Due to the minor differences in overtopping discharge for tests in the
current data set with different wave periods or different slope angles,
and due to confirmation in literature (Victor (2012)), it can be concluded
that parameters Tm−1,0 or α do not need to be included in the non-
breaking overtopping formula for smooth mild dike slopes with
1.5 ≤ cot(α) ≤ 3.
3 3.5 4
4.2. Smooth dike slope with storm wall
Wave overtopping can be reduced by placing a vertical storm wall
(with height hwall) on the dike slope. In this way incoming waves are
projected upwards. Note that the crest level of the dike slope (dashed
line in Fig. 9) can be maintained by placing the storm wall seaward on
the slope (Fig. 9 left), or can be increased by placing the storm wall on
y = 0.2e-2.28x
R² = 0.9435 top of the original crest (Fig. 9 right). The developed reduction factors
y = 0.2e-2.29x for wave overtopping are of course generally applicable. The only differ-
R² = 0.866
ence is the situation on the right in Fig. 9 has a larger freeboard (larger
y = 0.2e-2.6x
X-value on the log-linear graph) in comparison with the X-value of
the original smooth dike slope. The situation on the left of Fig. 9 has
the exact same X-value and allows for direct comparison. In our exper-
iments, the situation on the left in Fig. 9 was built in the wave flume, to
be able to keep the overtopping collector at the same level.
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 75

Fig. 9. Smooth dike slope with a vertical wall with the definition of the freeboard Rc.

A total of 117 tests were performed for a range of storm conditions research by Tuan, which is only based on small wall heights hwall/
and different heights of the storm wall in order to investigate its reduc- Rc b 0.5. Since none of the above methods are completely satisfying,
tive capacity. Table 2 provides a summary of the parameters of the test and don't always offer an approach which is possible to combine with
program on this geometry. the expression of Eq. (2), new reduction factors are proposed in this
In Fig. 10, the overtopping results are grouped by the height of the paper.
wall and plotted in different symbols and colors. The different wall Similar to what is done in Fig. 8 for smooth dike slopes, the influence
heights (2 cm to 8 cm) are shown in the legend. The reference formula (7) of the slope angle and the wave period is studied in the data set of
is plotted in a full red line. A clear reduction in overtopping volume can be
noticed since all the data are located below the reference line. The lines 0 0.5 1 1.5 2 2.5 3 3.5
for a vertical wall and a smooth dike slope cot(α) = 1 are also added 1.00E+00
Rc/Hmo smooth dike slope + wall 2cm
to Fig. 10 for reasons of comparison, even though both are not applicable smooth dike slope + wall 4cm
for the Belgian Coast. It is clear that the storm wall on top of a sloping dike smooth dike slope + wall 5cm
1.00E-01 smooth dike slope + wall 6cm
does not increase the virtual slope that much that it can be considered smooth dike slope + wall 8cm
as a vertical structure, but it still is highly reductive compared to the Vertical structure
smooth dike slope low freeboards
reference line from this research. The purple data points in Fig. 10 1.00E-02
steep slope 1:1
(overtopping over a smooth dike slope with storm walls of 8 cm)
q/(gHmo³)1/2

y = 0.2*exp(-2.28*x)
comes close to the steep dike slope cot(α) = 1. However, an increased
1.00E-03
virtual slope 1 V:1 H is not steep enough for Rc/Hm0 b 1, and too steep
for Rc/Hm0 N 1.5. Calculating overtopping over a smooth dike slope with
a storm wall as if it was overtopping over an increased virtual steeper 1.00E-04
slope does not give the best results. y = 0.2e-2.28x
Note that the crest freeboard was not increased during the new tests,
1.00E-05
only the wall height was increased by moving the wall forward on the
slope (Fig. 9 left). This allows for direct comparison between tests,
Fig. 10. Data set on a smooth dike slope with storm wall — measured values.
since the value on the X-axis does not change: a lower overtopping
discharge is found for a higher wall.
At first, the data are corrected by reduction factors as found in liter-
ature (formula (4) by EurOtop, (5) by Tuan and (6) by Coeveld), to see if
they move closer to the black reference line. The approach by EurOtop in
green symbols in Fig. 11 doesn't work, since only one general value is 0 0.5 1 1.5 2 2.5 3 3.5
used and Fig. 10 already shows that there is a difference between the 1.00E+00
Rc/Hmo/γ v
different wall heights which is not present in the γv according to
smooth dike slope + wall COEVELD
EurOtop. The approach by Coeveld in red symbols in Fig. 11 is also not
smooth dike slope + wall EUROTOP
a good approach, probably due to the fact that this approach might Vertical structure
1.00E-01
not be valid when only a storm wall is present. The approach by Tuan y = 0.2*exp(-2.28*x)
in Fig. 12 works satisfying, but is less accurate for the highest walls
where the overtopping discharge is overpredicted. That was to be
expected, since those were outside the range of parameters from the 1.00E-02
q/(gHmo³)1/2

Table 2
1.00E-03
Summary of the characteristics of the tests on a smooth dike slope with storm wall (scale
model values).

Slope angle of the smooth dike slope cot(α) 2 and 3

Mean spectral wave period Tm−1,0 1.10–3.00 s 1.00E-04
Dimensionless freeboard Rc/Hm0 0.60–2.60
Wall height hwall 2, 4, 5, 6 and 8 cm
Freeboard (top of structure to SWL) Rc 0.05–0.29 m
Dimensionless wall height hwall/Rc 0.08–1.60 1.00E-05
Spectral wave height Hm0 0.07–0.18 m
Water depth at toe of the structure d 0.36–0.57 m
Fig. 11. Smooth dike slope with storm wall corrected by formula (5) EurOtop (green) or
Wave breaker parameter ξm−1,0 2.20–4.80
formula (7) Coeveld (red).
76 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88

0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5

1.00E+00 1.0E+00
Rc/Hmo/γ v TUAN Rc/Hm0/γ v

1.0E-01
1.00E-01

q/(g·Hm0³)1/2
1.0E-02
y = 0.2e-2.28x
q/(gHmo³)1/2

1.00E-02

1.0E-03

1.00E-03 Smooth dike + W2: corrected values

smooth dike slope + wall 2cm TUAN
smooth dike slope + wall 4cm TUAN
1.00E-04
smooth dike slope + wall 5cm TUAN
smooth dike slope + wall 6cm TUAN
smooth dike slope + wall 8cm TUAN
Vertical structure
1.00E-05 y = 0.2*exp(-2.28*x)
1.0E-04 Smooth dike + W4: corrected values
Smooth dike + W5: corrected values
Smooth dike + W6: corrected values
Smooth dike + W8: corrected values
1.0E-05 Reference situation
Fig. 14. Data set on a smooth dike slope with storm wall — corrected values.

Fig. 12. Smooth dike slope with storm wall corrected by formula (6) Tuan.
hwall/Rc. It can be seen that the decrease of γ-value (or increase of reduc-
tion) slows down towards higher dimensionless wall-heights, which is
better expressed by an exponential relationship than by a linear one. For
smooth dike slope with storm walls. Only the conclusions are given
hwall/Rc N 1.24 the reduction coefficient becomes constant. There is no
below:
extra reduction when the wall is larger than the freeboard (i.e. water
- Slope angle: the difference between slope cot(α) = 2 and cot(α) = 3 level is above the base of the wall).
is again minor, but in contrast to the dike slope without storm wall, The reduction factor γv is defined in formula (9).
the steepest slope gives the largest overtopping discharge. The reason
for this is that on a steeper slope, the vertical velocity component of
 
the run-up is larger compared to a milder slope, which leads to slight- h hwall
ly larger overtopping discharges. Nevertheless, data analysis shows γv ¼ exp −0:56
wall for b1:24
Rc Rc
that the difference is again negligible, and no component α will be ð9Þ
hwall
γv ¼ 0:5 for ≥1:24:
included in the formula. Rc
- Wave period: no difference between small and large wave periods
was distinguished this time.
The data of Fig. 10 are corrected by means of formulae (8) and (9)
To take the reducing effect of the storm wall into account, formu- and plotted in Fig. 14. This leads to a much better prediction of wave
la (7) is adjusted to formula (8) by introducing an influence factor γv, overtopping over smooth dike slopes with a storm wall, compared
which is independent of slope angle and wave period as stated above. to the presentation in Fig. 10. All corrected points are now close to the
γv is calculated for every single data point, by isolation from formula (8). reference line, which demonstrates the efficiency of formula (8). The
This value means for every data point how much it would have to be exponential coefficient 2.28/γv is taken as a normally distributed
shifted on the X-axis of Fig. 10 to be exactly on the reference line and stochast with mean value 2.82 and standard deviation 0.41. The relative
thus give perfect prediction of the overtopping discharge by using standard deviation becomes 0.15 which has the same order of magni-
formula (8). tude as the relative standard deviation on formula (2) as mentioned in
EurOtop (2007): 0.35/2.6 = 0.13.
 
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c
ð8Þ 4.3. Smooth dike slope with storm wall and parapet
g
H3m0 Hm0 γv

Wave overtopping can be further reduced, without increasing the

γv is plotted versus a dimensionless parameter to find a best fitting height of the wall, by adding a “nose” to the vertical wall. This is further
curve. The dimensionless parameter plotted on the horizontal axis in referred to as parapet. Due to the presence of the parapet, waves are not
Fig. 13 is the height of the storm wall divided by the crest freeboard only projected upward, but also back towards open sea. A sketch of the
tested geometry as well as the definition of the used parameters in the
1.2
formulae are given in Fig. 15.
175 tests have been carried out on a smooth dike slope with storm
1 y = e-0.56x
R² = 0.46
wall and parapet, divided in 2 phases. In a first phase the influence of
0.8 the geometrical parameters such as the height of the wall and the
0.6 nose (hwall and hn) and the angle ε of the parapet were investigated in
γv

0.4 order to find an optimal geometry. The ranges of the parameters of

0.2 trend line the 92 tests from phase 1 are given in Table 3.
hwall/Rc new data In Van Doorslaer and De Rouck (2010) the results of phase 1 are
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
analyzed in full detail. The main conclusions are given here. The data
were plotted for the tested angles ε, and the different height ratios λ.
Fig. 13. Calculated γv as a function of hwall/Rc (left) and comparison of measured and The reduction of wave overtopping was found to be strongly dependent
calculated reduction factor γv. on both parameters.
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 77

Fig. 15. Sketch of a smooth dike slope with storm wall and parapet and definition of the used parameters.

A reduction factor γε describes the influence of the angle ε (Fig. 16 For hwall/Rc ≥ 0.25:
top left). One symbol in this graph represents the average of all data
grouped by their nose angle (e.g. the 2nd red cross from the left is the γpar ¼ 1:80
γ ε
γ λ
average of all data points with wall 8 cm and parapet ε = 30°, their with
average γε-value is 0.61). The larger the angle ε, the better the waves −4 2 −2  
γ ε ¼ 1:53
10
ε − 1:63
10
ε þ 1 if 15 ≤ ε ≤ 50
are projected towards the sea resulting in smaller overtopping 
γ ε ¼ 0:56 if ε ≥ 50
discharges. But the reduction doesn't increase linearly. An optimal γ λ ¼ 0:75 − 0:20
λ if 0:125 ≤ λ ≤ 0:6
angle with respect to reduction of wave overtopping and not too high
ð11Þ
uplift forces (which is outside of the scope of this paper) was found
for ε-values of 30° to 45°.
For hwall/Rc b 0.25:
Although the parameter ε is the dominant geometric variable, wave
overtopping also decreases when the nose of the parapet hn is more
prominent, and thus when λ increases. A small descending trend is no- γpar ¼ 1:80
γ ε
γλ −0:53
ticeable in Fig. 16 top right. Data were plotted versus λ, and a reduction with
 
ð12Þ
factor γλ was deducted. Best reduction was achieved for λ ≥ 0.3. γε ¼ 1 − 0:003
ε if 15 ≤ ε ≤ 60
To determine γε and γλ, the data were grouped according to ε-value γλ ¼ 1 − 0:14
λ if 0:1 b λ ≤ 1
and λ-value separately. However, a storm wall with parapet always has
both ε and λ as parameters, which makes independent separation of the
γλ in (11) is not equal to one when λ is zero, and γpar in (11) or (12) is
data impossible: γε is influenced by λ and vice versa. Consequently, the
also not one for γε or γλ equal to one. The formula (11) and (12) cannot
multiplication of γε and γλ leads to an overestimation of the reduction.
be used outside the mentioned intervals. In case λ or ε is really small,
A curve fitting between the measured data and γε∙γλ was drawn (see
the parapet is not much more beneficial than a vertical wall. In that
Fig. 16 bottom left), and formulae (11) and (12) are the final formulae
case, the formula (7) for a vertical wall is recommended for a conserva-
to predict the extra reducing effect of adding a parapet to a storm
tive design approach.
wall. Remark that γpar only takes into account this extra reducing
In the second phase of the research on parapets, the influence of the
effect of the parapet, as shown in Fig. 16 bottom right. Consequently
wave period and slope angle is investigated on two optimal parapet
γpar always has to be combined with γv of the vertical wall!
geometries, based on 83 tests. These optimal parapets haveε = 30° or
ε = 45°, keeping λ constant at 0.375. This leads to γpar = 0.79 respectively
!
q R 1 γpar = 0.70 for hwall/Rc ≥ 0.25 according to formula (11). A summary of
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c
ð10Þ the test program from phase 2 is given in Table 4.
g
Hm0 3 H m0 γv γpar


Data of the 2nd phase on dike slope 1:2 are plotted in Fig. 17, together
with the data of phase 1 (green triangles in Fig. 17, which are the same
green triangles as in Fig. 16 bottom right). Some conclusions can be
drawn from this graph:
- Data of phase 2 are amongst the data of phase 1, which is expected
Table 3
Summary of the test program on a smooth dike slope with storm wall and parapet (phase
since apart from a larger wave period the range of parameters in
1, scale model values). phase 2 is similar to phase 1.
- All data of both phase 1 and 2 are clearly below the reference line. In
Slope angle of the smooth dike slope cot(α) 2
Mean spectral wave period Tm−1,0 1.10–1.45 s
some tests with small freeboards, even lower overtopping discharges
Dimensionless freeboard Rc/Hm0 0.60–2.40 then vertical structures with equal dimensionless freeboards are
Wall height hwall 2, 5 and 8 cm noticed. This indicates that a storm wall with parapet is a very good
Freeboard (top of structure to SWL) Rc 0.09–0.18 m measure to reduce wave overtopping for non-breaking waves over
Dimensionless wall height hwall/Rc 0.11–0.90
smooth dike slopes.
Spectral wave height Hm0 0.08–0.15 m
Water depth at toe of the structure d 0.44 m–0.53 m - When looking at the data of slope angle 1:2 and ε = 30° (blue and red
Wave breaker parameter ξm−1,0 2.47–2.92 data points), the blue squares with the largest wave period clearly
Height of the nose hn 1, 2 and 3 cm show more wave overtopping than the red diamonds. For the data
Height ratio parapet (hn/hwall) λ 0.125–1 slope angle 1:2 and ε = 45° (orange and purple data points) the
Nose angle parapet (in degrees) ε 15°, 30°, 45° 60°
same observations are made: the orange circles have the largest
78 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88

Fig. 16. Explanation of the analysis of γpar for the example case hwall/Rc ≥ 0.25.

wave period and give more wave overtopping than the purple
triangles.
For slope angle 1:3 no such comparison could be made, because all
tests with wave period Tm−1,0 = 1.36 s lead to breaking waves and
are therefore excluded in the current analysis. Based on the data on
slope angle 1:2, there is a clear influence of the wave period for the
geometry with parapet. This confirms what was found in
Kortenhaus et al. (2001); Pearson et al. (2004). It is visually observed
in Pearson et al. (2004); Van Doorslaer (2008) that long waves, who
have a larger volume of water under the crest of a wave, first “fill” the
space underneath the parapet's nose, after which it acts as a normal
storm wall which is more easily overtopped than a parapet.
- When comparing data sets with the same slope angle and the same
wave period, such as red diamonds versus purple triangles
(cot(α) = 2, Tm−1,0 = 1.64 s) or blue squares versus orange circles
(cot(α) = 2; Tm−1,0 = 2.36 s) or black cross versus pink plus
(cot(α) = 3; Tm−1,0 = 2.36 s), it can be seen that ε = 45° reduces
a little bit more than 30°. This confirms what was found in phase 1
in Fig. 16 top left.
- When comparing data sets with the same nose angle ε and the same
wave period, for example black cross versus orange circle (ε = 45°,
Tm−1,0 = 2.36 s), there is hardly any difference. The mildest slope is
overtopped the least since the run-up on the mildest slope
has more horizontal velocity and less vertical velocity to overcome
the structure. The influence is nevertheless again too small and the
number of different slopes in this data set too limited to deduct an
influence factor for the slope.

Table 4
Summary of the test program on a smooth dike slope with storm wall and parapet 0 0.5 1 1.5 2 2.5 3
1.00E+00
(phase2, scale model values).
R c /Hm0
Slope angle of the smooth dike slope cot(α) 2 and 3 1.00E-01
Mean spectral wave period Tm−1,0 1.50–2.30 s⁎
Dimensionless freeboard Rc/Hm0 1.25–2.26
1.00E-02
Wall height hwall 8 cm
q/(g.Hm0³)1/2

Freeboard (top of structure to SWL) Rc 0.16–0.29

Dimensionless wall height hwall/Rc 0.28–0.50 1.00E-03
Spectral wave height Hm0 0.09–0.18 1/2 par 30° T = 2.36
Water depth at toe of the structure d 0.36–0.51 m 1.00E-04
1/2 par 30° T = 1.64
1/2 par 45° T = 2.36
Wave breaker parameter ξm−1,0 2.20–4.61
1/2 par 45° T = 1.64
Height of the nose hn 3 cm 1/3 par 30° T = 2.36
Height ratio parapet λ 0.375 1.00E-05 1/3 par 45° T = 2.36
Nose angle parapet (in degrees) ε 30° and 45° y = 0.2*exp(-2.28x*x)
data phase 1
1.00E-06
⁎ Notice that the wave period in phase 2 was chosen larger in order to investigate the Vertical Structure

influence of this parameter on the overtopping discharge over the optimal parapet
geometry. Fig. 17. Data set on a smooth dike slope with storm wall and parapet — measured values.
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 79

0 0.5 1 1.5 2 2.5 3 3.5

Table 5
1.00E+00
Rc/Hm0/γ v / γ par / γ s0,par Summary of the characteristics of the tests on a smooth dike slope with promenade (scale
model values).
1.00E-01
Slope angle of the smooth dike slope cot(α) 2 and 3
Mean spectral wave period Tm−1,0 1.1–2.22 s
1.00E-02 Dimensionless freeboard Rc/Hm0 0.85–2.68
q/(g.Hm0³)1/2

Length promenade B 33.3, 66.7 and 100 cm

Slope promenade – 1% and 2%
1.00E-03 Freeboard (top of structure to SWL) Rc 0.10–0.28 m
Dimensionless promenade length B/Lm−1,0 0.045–0.5
1/2 par 30° T = 2.36 corrected Spectral wave height Hm0 0.07–0.17 m
1.00E-04 1/2 par 30° T = 1.64 corrected Water depth at toe of the structure d 0.28–0.53 m
1/2 par 45° T = 2.36 corrected Wave breaker parameter ξm−1,0 2.2–4.2
1/2 par 45° T = 1.64 corrected
1.00E-05
1/3 par 30° T = 2.36 corrected
1/3 par 45° T = 2.36 corrected

1.00E-06 y = 0.2*exp(-2.28x*x)
data phase 1 corrected
Fig. 18. Data set on a smooth dike slope with storm wall and parapet — corrected values.
Summarizing for the storm wall with parapet, it can be concluded
that together with γv to include for the effect of the height of the
storm wall, and γpar to include the reducing effect of the nose angle ε
and location of inclination λ, also a correction factor γs0,par has to be
included to account for the larger wave overtopping for larger wave
periods, by means of the dimensionless wave steepness sm−1,0 (formu-
la (14)). Formula (10) is now adapted to formula (13) to calculate the
reduction through a parapet.
! duction overpredicted), bringing the green data points too high above
q R 1 the reference line. This shows that a new reduction factor especially
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c

g
Hm0 3 H m0 γv
γpar
γs0;par
γs0;par ¼ 1:33−10
sm−1;0
Fig. 18 shows that by introduction of the correction factors γv, γpar
and γs0,par, the data can be well predicted by using formula (13). The
exponential coefficient 2.28/(γv · γpar · γs0,par) is taken as a normal
distributed stochastic variable, with a mean value of 3.62 with standard
deviation 0.65. This gives a relative standard deviation of 0.18.
4.4. Smooth dike slope with promenade
of the construction and the still water level. The slope of the promenade
is included in Rc. The slope of the promenade in the test set-up was 1% or
2% to stimulate drainage on the promenade of overtopped water and
rainfall back towards the sea (wave flume).
A total of 62 tests were performed on this geometry. The range of the
parameters of the test program is given in Table 5.
As mentioned in Section 2, EurOtop (2007) proposes a γb reduction
factor to account for a berm around still water level. Eventhough this is
not the case here, the data on the smooth dike slope with promenade
are corrected by the γb from EurOtop (2007) in Fig. 20. In blue the
data as measured, where it can be seen that all of them are
located below the reference line, showing the reductive capacity of a
promenade. In green, the same data are corrected by means of γb ac-
cording to the EurOtop, 2007 formula. γb from EurOtop is too low (re-
ð13Þ for smooth dike slopes with promenades at crest level should be
deducted.
The blue data are analyzed, with extra attention paid to the wave
ð14Þ period and slope angle. Similar conclusions as for the reference case
(smooth dike slope) are found:
- Slope angle: a minor difference in overtopping, where the mildest
slope is overtopped slightly more due to the thicker layer thickness
of the incoming wave on milder slopes.
- Wave period: on slope 1:2 there is no difference, and on slope 1:3
there is a minor difference with the most overtopping for large
wave periods.

The typical geometry of a sea dike along the Belgian Coast consists
of a dike slope with a wide promenade at crest level (Fig. 19). The free- Overall, both influences are neglectable and therefore not further
board Rc is defined as the difference in height between the highest point considered here.

Fig. 19. Smooth dike slope with promenade and used parameters.
80 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88

0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5

1.00E+00 1.00E+00
R c /H m0 / γ b EUROTOP R c /Hm0 / γ prom
1.00E-01

1.00E-01
q/(g.H m0³)1/2

1.00E-02

q/(g.H m0³)1/2
1.00E-02
1.00E-03

1.00E-04
y = 0.2*exp(-2.28*x) 1.00E-03
data corrected by EurOtop2007 form 5.29
1.00E-05 data original

Fig. 20. Data on a smooth dike slope with promenade (blue) corrected by EurOtop (2007) 1.00E-04
y = 0.2*exp(-2.28*x)
reduction factor for a berm (green).
smooth dike slope + B0.33
smooth dike slope + B0.66
1.00E-05
smooth dike slope + B1.00

The data in Fig. 21 show that the presence of a promenade has a Fig. 22. Smooth dike slope with promenade — corrected values.
reducing effect on wave overtopping, slightly increasing with the length
B of the promenade, but not as strong as a storm wall.
A reduction factor γprom is fitted through the data with the dimension-
less promenade length as a parameter. Formula (7) is then transformed to proves that Eq. (16) describes the overtopping reduction accurately.
(15) in case of wave overtopping over dike slopes with a promenade The exponential coefficient 2.28/γprom is taken as a normally distributed
clearly above SWL: stochastic variable and has a mean value of 2.55 with standard deviation
! 0.19. This gives a relative standard deviation of 0.07.
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c
ð15Þ
g
Hm0 3 H m0 γprom 4.5. Smooth dike slope with promenade and storm walll

The physical process of a wave hitting a wall is different when a

B
γprom ¼ 1−0:47
: ð16Þ promenade is present in between the top of the dike slope and the
Lm−1;0 wall. For this reason, it is not possible to just multiply the influence
factors γprom (formula (16)) and γv (formula (9)) to account for the
All results of a smooth dike slope with promenade, grouped by the combined effect of a promenade and a wall. Therefore 136 model tests
length of the promenade, are plotted together with the reference were performed on geometries with both a promenade and a wall. A
formula in Fig. 21 (B0.33 means a promenade length of 33.3 cm etc.). sketch is given in Fig. 23. Rc again includes the slope of the promenade
In Fig. 22 these values are corrected by means of formulae (15) and and the height of the storm wall.
(16). All points are close to the reference line with little scatter. This The range of the parameters of the test program is listed in Table 6.
Again no clear influence of the wave period or the slope angle is no-
ticed in the analysis:
0 0.5 1 1.5 2 2.5 3 3.5 - Slope angle: mildest slope 1:3 is slightly more overtopped compared
1.00E+00
to 1:2, due to the thicker water layer. The influence is however too
R c /H m0 weak to include in the reduction factors.
- Wave period: for both slopes there was just a little more overtopping
1.00E-01 measured for the longest wave periods. This is due to the larger layer
q/(g.H m0³)1/2

1.00E-02

1.00E-03

1.00E-04
y = 0.2*exp(-2.28*x)
smooth dike slope + B0.33
smooth dike slope + B0.66
1.00E-05
smooth dike slope + B1.00

Fig. 21. Smooth dike slope with promenade — measured values. Fig. 23. Sketch of a smooth dike slope with promenade and storm wall.
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 81

0 0.5 1 1.5 2 2.5 3 3.5 4

Table 6
1.0E+00
Summary of the characteristics on a smooth dike slope with promenade and storm wall Rc /H mo
(scale model values).

Slope angle of the smooth dike slope cot(α) 2 and 3 1.0E-01

Mean spectral wave period Tm−1,0 1.10–2.25 s
Dimensionless freeboard Rc/Hm0 0.80–2.50
Freeboard (top of structure to SWL) Rc 0.10–0.28 m 1.0E-02

q/(g ·H mo³)1/2
Spectral wave height Hm0 0.075–0.17 m
Length promenade B 33.3, 66.7 and 100 cm
Slope promenade – 1%–2% 1.0E-03 y = 0.2e-2.28x
Dimensionless promenade length B/Lm−1,0 0.05–0.41
Wall height hwall 2, 4, 6 and 8 cm
Dimensionless wall height hwall/Rc 0.07–0.80
1.0E-04
Water depth at toe of the structure d 0.36–0.55 m
Wave breaker parameter ξm−1,0 2.26–4.80

1.0E-05

Smooth dike slope: measured data

1.0E-06 smooth dike slope + promenade + wall
thickness of the water on the promenade as a consequence of the
larger volume of water under the crest of longer waves. Also here, Fig. 25. Smooth dike slope with promenade and storm wall — measured values.
the influence will not be included in the reduction factors since the
difference in overtopping measured was too small.
The reducing effect of the combination of a wall and a promenade is
stronger than the multiplication of both influences separately. γprom_v
is smaller than γprom × γv, leading to smaller wave overtopping dis-
charges, which makes this geometry a very efficient measure to 0 0.5 1 1.5 2 2.5 3 3.5 4
reduce wave overtopping (Fig. 24). 1.0E+00
A new influence factor γprom_v is introduced, which can be calculated Rc /Hmo /γ prom_v
by formula (18), and has to be incorporated in formula (17).
1.0E-01
!
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c
ð17Þ q/(g ·H mo³)1/2
g
Hm0 3 H m0 γprom v 1.0E-02

γ prom v ¼ 0:87
γ prom
γv ð18Þ 1.0E-03

with γprom as defined in formula (16) and γv in formula (9). Note that 1.0E-04
this formula is only usable when both a promenade and a storm wall
are present. When one of both is missing, formula (17) makes no
1.0E-05
sense and (15) or (8) should be used. Smooth dike slope: measured data
Because 10 different combinations of a promenade (33.3 cm,
66.6 cm and 1 m) and a vertical wall (2 cm, 4 cm, 5 cm, 6 cm and smooth dike slope + promenade + wall corrected
1.0E-06
8 cm) at the end of the promenade were tested in order to investigate
the combined effect, all the results are plotted in the same color, Fig. 26. Smooth dike slope with promenade and storm wall — corrected values.
together with the reference data in Fig. 25. In Fig. 26 the blue values
are corrected by means of formula (17) and (18), which leads to a
high correlation to the trend line meaning that a good prediction is ob-
tained. The exponential coefficient 2.28/γprom_v has a mean value of 3.42
with standard deviation 0.46. A relative standard deviation of 0.13 is
found.

1 4.6. Smooth dike slope with promenade, storm wall and parapet

0.8 Even though the combination of a promenade and a vertical wall is

already a very efficient measure, wave overtopping can be further
reduced without increasing the height of the wall by adding a parapet
prom_v

0.6
to the wall. This combined effect has been investigated by means of
0.4 y = 0.87x 100 tests (Table 7). A sketch of this geometry is given in Fig. 27.
It is not possible to simply multiply γprom_v and γpar since that would
0.2 overestimate the actual reduction in wave overtopping (the calculated
reduction factor γprom_v × γpar is lower than the measured one).
0 This doesn't mean that adding the parapet will not further
0 0.2 0.4 0.6 0.8 1 reduce the overtopping discharges; it means that the new
γ prom × γ v parameter γprom_v_par is not as effective as the multiplication of γprom_v
and γpar. A parapet intrinsically functions best directly on a slope,
Fig. 24. Deduction of γprom_v as a function of γprom and γv. since it takes benefit of the upward motion of the water to reflect it
82 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88

1
Table 7
Summary of the characteristics of the tests performed on a smooth dike slope with prom-
enade, storm wall and parapet (scale model values). y = 1.19x
0.8
Slope angle of the smooth dike slope cot(α) 2 and 3
Mean spectral wave period Tm−1,0 1.25–2.25 s
0.6

prom_v_par
Dimensionless freeboard Rc/Hm0 0.7–1.9
Freeboard (top of structure to SWL) Rc 0.08–0.24 m
Spectral wave height Hm0 0.08–0.17 m
0.4
Length promenade B 33.3, 66.7 and 100 cm
Slope promenade – 1%–2%
Dimensionless promenade length B/Lm−1,0 0.04–0.40
0.2
Wall height hwall 4, 6, 8 cm
Dimensionless wall height hwall/Rc 0.17–0.80
Height ratio parapet ε 30°, 45°
0
Nose angle parapet (in degrees) λ 0.25–0.375
0 0.2 0.4 0.6 0.8 1
Water depth at toe of the structure d 0.40–0.55 m
Wave breaker parameter ξm−1,0 2.15–4.77
γ prom_v × γ par

Fig. 28. Deduction of γprom_v_par as a function of γprom_v and γpar.

back towards the sea. Following reduction factor can be concluded from wall and parapet. The hydraulics on the geometry dike slope with prom-
Fig. 28: enade, storm wall and parapet are different, and make the overtopping
over this geometry not strongly dependent on the wave period. The
γprom v par ¼ 1:19
γprom v
γpar : ð19Þ storm wall with parapet at the end of the promenade reflects the incom-
ing water layer equally for long as for short waves with only very little
Combining Eq. (18) with Eq. (19), the reduction factor can also be difference due to the larger layer thickness on the promenade of long
calculated as waves overtopping the dike. This is similar to the geometry dike slope
with promenade and storm wall.
γprom v par ¼ 1:03
γprom
γv
γpar : ð20Þ The formula for overtopping over a dike slope with promenade,
storm wall and parapet becomes:
The underestimation of γprom_v × γpar almost neutralizes the
overestimation of γprom × γv, and the final reduction factor γprom_v_par !
q R 1
seems to be 3% less efficient than the product of all individual qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c
ð21Þ
g
Hm0 3 H m0 γprom v par
measures. This shows that reduction factors can't be just multiplied
with each other without detailed study, especially in situations like
these where the physics change between a wave overtopping a
with γprom_v_par as defined in formula (20).
structure, and an overtopping bore on the promenade overtopping a
For this geometry, 10 different combinations of a promenade and a
storm wall.
storm wall with parapet at the end of the promenade were tested to
When one or more of the above parts are missing in the geometry,
investigate the combined effect. Therefore, all results on the geometry
like no parapet or no berm, formula (19) or (20) cannot be used!
smooth dike slope with promenade, storm wall and parapet are plotted
The user should then use the correct geometry as mentioned in earlier
in the same color (blue), together with the reference data (red) in
paragraphs in this paper.
Fig. 29. In Fig. 30 the blue values are corrected by means of formula (20)
The influences of slope angle and wave period on the overtopping
of this geometry have been studied. The same conclusions as for the
promenade with storm wall are valid: no influence of slope angle or
wave period is clearly noticeable on dike slope with promenade,
storm wall and parapet, despite the parameter γs0,par which accounts
for the influence of the wave period on the geometry dike with storm
0 0.5 1 1.5 2 2.5 3 3.5
1.00E+00
Rc /Hmo

1.00E-01
q/(g·H m0³)1/2

1.00E-02

1.00E-03

1.00E-04

smooth dike slope

1.00E-05 smooth dike slope + promenade + wall + parapet

Fig. 27. Sketch of a smooth dike slope with promenade, storm wall and parapet. Fig. 29. Smooth dike slope with promenade, storm wall and parapet — measured values.
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 83

0 0.5 1 1.5 2 2.5 3 3.5

4.7. Stilling wave basin (SWB)
1.00E+00
Rc/Hmo/γ prom_v_par
A last measure proposed in this paper to reduce wave overtopping
by modifying the existing crest of dike slopes, is the so-called stilling
1.00E-01
wave basin (Beels, 2005; Geeraerts et al., 2006). The SWB is made up
of a seaward wall, a basin and a landward wall (Fig. 31). The seaward
q/ (g·H m0³)1/2
wall is partially permeable to allow the evacuation of the water in the
1.00E-02
basin. It may consist of a double row of shifted walls (Fig. 32) or a single
wall with some gaps. This innovative crest design is based on the
principle of energy dissipation: the incoming wave hits the seaward
1.00E-03
wall and is projected upward, then drops in the spilling basin before
hitting the landward wall. At that moment, most of its energy is already
dissipated. Consequently, the landward wall is overtopped less in
1.00E-04
comparison with an unmodified crest, even though the crest height
smooth dike slope
has not been increased.
smooth dike slope + promenade + wall + parapet: Many geometrical variations of the SWB have been tested, with over
1.00E-05 CORRECTED 300 tests with non-breaking wave conditions. The range of hydraulic
parameters and geometric variations is listed in Table 8, and is illustrated
Fig. 30. Smooth dike slope with promenade, storm wall and parapet — corrected values.
in Fig. 32 and 33.
The front wall of the SWB varied in height from 48 mm to 144 mm,
in which the 48 mm above the SWB floor was kept constant over
all variations tested. In Fig. 32 a total wall height of 96 mm is
and (21). As can be seen in Fig. 30, a good prediction is obtained. The shown; it's the 48 mm below the SWB floor which has been varied be-
exponential coefficient 2.28/γprom_v_par is taken as a normally distribut- tween 0 mm and 96 mm. At a Froude scale of 25, the fixed 48 mm
ed variable, and has a mean value of 4.13 with standard deviation 0.59. upper part becomes 1.2 m which is a perfect height to lean on, like a
The relative standard deviation is 0.14. railing.

Fig. 31. Simple smooth dike slope (left) compared to a dike slope with SWB built in the crest (right).

Fig. 32. Side view of stilling wave basin (dimensions in mm), Geeraerts et al. (2006).

Table 8
Summary of the characteristics of the tests performed on a smooth dike slope with SWB (scale model values).

Slope angle of the smooth dike slope cot(α) 2; 2.5; 3

Mean spectral wave period Tm−1,0 1.16 s–2.33 s
Dimensionless freeboard Rc/Hm0 0.56–2.7
Freeboard (top of structure to SWL) Rc 0.10–0.27 m
Spectral wave height Hm0 0.08–0.18 m
Length basin Lbasin 48, 36, 24 and 12 cm
Slope basin – 2%
Wall height hfront wall 48, 72, 96, 120 and 144 mm
Distance between front walls – 4 cm
Water depth at toe of the structure d 0.30–0.52 m
84 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88
Fig. 33. Plan view of stilling wave basin (dimensions in mm), Geeraerts et al. (2006).
- The blocking coefficient, which is the ratio between the open and the
closed part of each row of shifted walls, has an important influence
on the wave overtopping over the landward wall. An optimum be-
tween inflow (as low as possible) and outflow (as high as possible)
was a subject of the study. A blocking coefficient of 50% for the most
seaward wall, and 65% of the 2nd row wall has been found optimal.
To avoid that the wave flows directly into the basin, 20% of each wall
part of the first row overlaps with a wall part from the second row.
To encourage the drainage back towards the sea, the basin has
been given a 2% slope.
Fig. 34. Stilling wave basin as constructed in the city of Ostend. Picture was taken during low tide ©airmaniacs.be.
The two separate overlapping walls may be replaced by one wall
with small gaps just above the floor of the basin. This has also been
constructed in the city of Ostend, where the engineered and archi-
tectural design go hand in hand (Fig. 34).
- The height of the front wall and the length of the basin have
been studied. While the effect of γprom is smaller than the
effect of γv (see Section 4.4), a similar conclusion can be
drawn for the SWB: the variation of height of the front wall is
dominant, while the effect of the basin's length is present but
less pronounced.
- The slope angle and the wave period have a minor influence on
the reduction in wave overtopping. As for the vertical wall
(γv) and promenade above SWL (γb), both influences are not
strong enough to be included in the formula of the reduction
coefficient.
Since so many variations in the geometry of the SWB are
possible, one uniform reduction formula has not been determined. The
blocking coefficient, the distance in between the double row walls, the
slope near the landward wall, the length of the basin and the height of
the front wall all have their influence on the reduction of wave
overtopping. The basic geometry (Lbasin = 48 cm, hfront wall = 96 mm)
with the optimal blocking coefficient of 50% (1st row) and 65% (2nd
row) has been tested in full detail. A reduction factor of 0.48 is found
for this geometry, and can be used to quantify wave overtopping over
a dike slope with SWB. In case a specific geometry is required, the au-
thors suggest determining the reduction capacity by means of scale
model tests.
 
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:28
c
ð22Þ
g
H3m0 Hm0 γSWB
γSWB ¼ 0:48 ð for the selected geometryÞ ð23Þ
Landward wall constructed
as steps
Frontwall with gaps
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 85

5. Application of formulae in a case study

The reductive effect of each measure is dependent on the hydraulic boundary conditions and the geometry of the structure. For example a
promenade above the SWL will reduce the small overtopping volumes (large Rc/Hm0) better than the large overtopping volumes (small Rc/Hm0).
Large overtopping volumes are best reduced by a combination of promenade and storm wall with or without parapet, or by an SWB.
The SWB combines the effect of a promenade and a storm wall, but is capable of reducing the incoming energy even further by means of the
double row of front walls and a spilling basin. This reflects in the low reduction coefficient γSWB = 0.48 for the presented geometry, without
increasing the crest height, whereas the combination of promenade and storm wall the original crest level increases with the height of the
storm wall.
An example is worked out below to demonstrate the reductive capacity for all proposed measures under 3 different wave heights. The other
parameters, such as Rc, Tm−1,0 and cot(α) remain the same throughout the whole example. Based on the deterministic design approach according
to EurOtop (2007), the basic formula to calculate wave overtopping discharge over a smooth dike slope for non-breaking waves is

 
q R
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3
C ð24Þ
g
H3m0 Hm0

where the average discharge from formula (2) is increased by about a standard deviation as recommended in EurOtop (2007). For all reduction
measures further on in this case study, the standard deviations deducted in this paper are not included. It is up to the designer which level of safety
they want to achieve.

cot(α) 2
SWL mTAW 7.00
Crest level mTAW 9.00
Rc m 2.00
Tm−1,0 s 8.2
Hm0 m 3.00 2.00 1.00
Rc/Hm0 – 0.67 1.00 2.00
q l/m/s 702.47 177.64 6.30

Wave overtopping over a smooth dike with crest level at +9.00 mTAW and water level at +7.00 mTAW is calculated by means of formula (24). A
mean overtopping discharge of 6.3, 177.6 resp. 702.5 l/m/s is found for a storm with wave height 1.00, 2.00 resp. 3.00 m at the sketched smooth dike
slope.
When a storm wall is added to the slope of the dike, without increasing the freeboard Rc, a reduction factor γv has to be included in the exponen-
tial part of formula (24). The formula now becomes

 
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3
C
:
g
H3m0 H m0 γv

With γv calculated using formula (9). The mean overtopping discharges reduces to 0.92, 67.75 resp. 369.46 l/m/s which is 6.9, 2.6 or 1.9 times less
than without the storm wall and the same crest freeboard.

Smooth dike slope with storm wall

hwall = 1.25 m; hwall/Rc = 0.625

γv – 0.703 0.703 0.703

q l/m/s 369.46 67.75 0.92
Ratio – 1.90 2.62 6.87

Further, a parapet nose is added to the same storm wall with ε = 45° and λ = 1/3. Again, no change in crest freeboard Rc. The average discharge
can be calculated by means of formula (26).

!
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3
c
ð26Þ
g
Hm0 3 H m0 γv
γpar
γs0;par

With γv calculated using formula (9), γpar using formula (11) and γs0,par using (14). A reduction ratio of 17.5, 5.7 and 4.1 is achieved in comparison
to a smooth dike slope under the same hydraulic conditions. The effect of a parapet is most prominent for larger dimensionless freeboards, which was
86 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88

also concluded by Kortenhaus et al. (2001) in the SPP-project. Reductions up to a factor 10 and higher are possible, just like the study by Kortenhaus
et al. (2003) and Pearson et al. (2004).

Smooth dike slope with storm wall and parapet

hwall = 1.25 m; hwall/Rc = 0.625
ε = 45°, λ = 1/3

γv*γpar*γs0,par – 0.521 0.569 0.617

q l/m/s 171.55 31.07 0.36
ratio – 4.09 5.72 17.47

When a promenade at crest level is taken into account, it is explained in Section 4.4 to include a reduction factor γprom in the exponential part of
the formula, which now becomes

!
q Rc 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3

:
g
H3m0 Hm0 γprom

γprom is calculated by using formula (16). The promenade in this exercise is 20 m wide and has a 2% slope, leading to an increased crest height
+9.04 mTAW. To maintain the Rc constant at 2 m (for direct comparison), the water level in this (theoretical) exercise is also increased by 4 cm.
The table below shows the reduced mean overtopping discharges. The effect of a promenade is much lower than the effect of other measures.
Nevertheless, the overtopping is reduced by a factor of 1.73, 1.38 resp. 1.28.

Smooth dike slope with promenade

Promenade width = 20 m, slope 2%

γprom – 0.91 0.91 0.91

q l/m/s 549.40 128.79 3.64
ratio – 1.28 1.38 1.73

When a storm wall of 1.25 m high is present at the end of the promenade, the crest height of the structure is increased to +10.29 mTAW. To
maintain the same crest freeboard Rc = 2 m for reasons of comparison, the water level is also increased up to +8.29 mTAW. The mean overtopping
discharge is now calculated using the formula
!
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3
c

g
Hm0 3 H m0 γprom v

with γprom_v according to formula (18).

Smooth dike slope with promenade and storm wall

Promenade width = 20 m, 2% slope
Storm wall 1.25 m high. Rc = 2 m, hwall/Rc = 0.625

γprom_v – 0.56 0.56 0.56

q l/m/s 189.71 26.13 0.15
Ratio – 3.70 6.80 42.05

This geometry is capable of reducing the wave overtopping discharge to a maximal level so far, with reduction ratio of 42.1, 6.8 and 3.7 compared
to the discharge over smooth dike slopes with the same freeboard Rc. Therefore, this geometry is applied a lot at the Belgian coastline to reduce wave
overtopping. The storm wall can be constructed as a mobile wall which is only set up when there is a flood risk. When the high tide and storm surge
have passed, this mobile wall can be deconstructed and the promenade regains its original function as a touristic promenade, without disturbing the
open view at the sea.
The discharge can be reduced even further, by adding a parapet to the above structure, with ε = 45° and λ = 1/3. The formula becomes

!
q R 1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3
c

g
Hm0 3 H m0 γprom v par
 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88 87

with γprom_v_par according to formula (19).

Smooth dike slope with promenade and storm wall

Promenade width = 20 m, 2% slope
Storm wall 1.25 m high. Rc = 2 m, hwall/Rc = 0.625
Parapet (ε = 45°, λ = 1/3)

γprom_v_par – 0.47 0.47 0.47

q l/m/s 113.96 12.17 0.03
Ratio – 6.16 14.60 193.98

To conclude, also an SWB with the standard geometry as presented in Fig. 32 is included in this (theoretical) comparison. The average overtopping
discharge now has to be calculated by using
 
pffiffiffiffiffiffiffiffiffi
q ffi
3
¼ 0:2
exp −2:3
HRm0
c

γ1 :
g
Hm0 SWB

Stilling wave basin

Lbasin = 12 m
hfront wall = 2.4 m

γSWB – 0.480 0.480 0.480

q l/m/s 133.42 14.70 0.04
Ratio – 5.27 12.08 145.96

The boundary conditions such as wave conditions, space, crest height, etc. will have to decide what kind of crest modification can be applied to
reduce wave overtopping.
To conclude this example, overtopping over a rubble mound breakwater and a vertical caisson breakwater is added to this comparison. Since
the γ-factor for the roughness of the rubble mound comes close to the values of reduction factors for SWB or dike slope with promenade and wall
with/without parapet, similar reduction in wave overtopping is achieved. The vertical wall on the other hand reduces more than a dike slope with
wall or parapet, since the slope is much steeper, but doesn't reduce wave overtopping as affective as an SWB or a dike slope with promenade and
storm wall does.

Rubble mound breakwater: γf = 0.50

q l/m/s 151.60 17.81 0.06
ratio – 4.63 9.97 99.48

Caisson breakwater (vertical wall)

q l/m/s 127.52 24.04 0.12
Ratio – 5.51 7.39 54.60

6. Conclusions
In order to investigate the reduction in wave overtopping over
smooth dike slopes, over 1000 tests with non-breaking waves have
been carried out on 7 different types of geometry, with many variations
in both geometrical and hydraulic parameters for each geometry. The
analysis of this large data set shows that each proposed reduction mea-
sure can be included in the wave overtopping formula to predict wave
overtopping. Several reduction factors have been proposed in this
paper, each of them applicable for a specific geometry and within the
ranges of parameters given in the tables. Each individual reduction fac-
tor should be included in the formula (31) to calculate the average
overtopping discharge for non-breaking waves. According to EurOtop,
2007, the deterministic design approach with exponential coefficient
2.3 is used:
 
q R 1 storm wall without parapet at the end of a promenade.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:2
exp −2:3
C
: ð31Þ
g
H3m0 Hm0 γreduction
The different reduction factors are:
- γv (formula (9)) defines the reduction due to a storm wall on the
crest, related to the height of the storm wall.
- γpar (Formula (11) and (12)) gives the extra reduction that can
be obtained by adding a parapet nose to the storm wall on the
crest. This influence factor is a function of the parapet angle ε and
height ratio λ. It should always be combined with γv. The parapet
is the only geometry where a clear influence of the wave period
on the overtopping discharges was noticed. Therefore, also γs0,par
(formula (14)) should be combined with γpar and γv.
- γprom (Formula (16)) is the reduction factor due to a promenade
at the crest level. γprom is related to the length of the promenade,
which is expressed dimensionless by means of the wave length.
- γprom_v (Formula (18)) represents a storm wall at the end of a
promenade. This is a very efficient reductive measure, which
reduces wave overtopping more than multiplying their separate
influence factors γprom and γv.
- γprom_v_par (Formula (19)) is the reduction factor when a prome-
nade with storm wall and parapet are present. A parapet is less
efficient compared to when it's positioned at the dike slope, but
still reduces the average overtopping discharge more than a
- A stilling wave basin is also an efficient way to reduce wave
overtopping without increasing the crest level. The combined
effect of (a double row shifted) storm walls, a promenade and
an energy dissipating basin leads to a low reduction coefficient
γSWB. Due to the multiple geometric variables, a generic reduc-
tion formula was not deducted, but increasing the height of the
front wall and choosing an optimal blocking coefficient turned
out to be the most efficient ways to optimize the reduction of
88 K. Van Doorslaer et al. / Coastal Engineering 101 (2015) 69–88
wave overtopping over the landward wall, which is in fact the
crest of the original dike slope.
While analyzing the data, for each geometry it was studied whether
the slope angle of the dike slope and/or the wave period have their
influence on the measured wave overtopping.
- In general there was a minor influence of the slope angle 1:2 versus
1:3: for a normal dike slope or a dike slope with a promenade, the
mildest slope was overtopped slightly more than steeper ones
(least reflection on mildest slope and the thickness of the runup/
overtopping water layer is a bit larger). For dike slopes with an
SWB, a storm wall or parapet at the crest, the steepest slope was
overtopped more (steeper slope have more upward velocity in the
runup, which allows more overtopping). For dike slopes with a
promenade in combination with a storm wall with or without para-
pet, the same influence as without storm wall was found: slope 1:3 is
slightly more overtopped than slope 1:2. However, the influence of
the slope angle was in any of the 7 geometries too small to include
in the overtopping formulae.
- In most of the geometries the largest overtopping discharges were
measured for tests with the largest wave period. For all geometries,
except for the one with the parapet directly placed at the crest of the
dike slope, the influence was however too small to include in the
overtopping formula. For the case dike slope with storm wall and
parapet, an extra parameter γs0,par is included in the overtopping
formula, since the effect of the wave period was clearly noticeable
in the results: long waves have the tendency to fill the space under
the parapet's nose with water, after which the parapet acts as a
normal storm wall and is more easily overtopped compared to
short waves who encounter a parapet. When a promenade is present
in front of the parapet, this effect disappears.
Acknowledgments
The authors acknowledge the assistance of the technical staff of the
Coastal Engineering Laboratory at Ghent University. A large data set
with many geometric variables was only possible to obtain with their
effort. Master thesis students Tobias Boderé, Gilles Vanhouwe (Boderé
and Vanhouwe (2010)), Sarah Audenaert and Valerie Duquet
(Audenaert and Duquet (2012)) performed a great effort in data analy-
sis during and after completion of their thesis. At last, discussions with
prof. A. Kortenhaus and Dr. Ir. D. Vanneste are also acknowledged.
References
Allsop, N.W.H., Besley, P., Madurini, L., 1995. Overtopping performance of vertical walls
and composite breakwaters, seawalls and low reflection alternatives, Final report of
Monolithic Coastal Structures (MCS) project. University of Hannover.
Audenaert, S., Duquet, V., 2012. Golfovertopping over zeedijken - krachten op stormmuur
(Wave Overtopping Over Sea Dikes; Forces On a Storm Wall). (Master thesis). Ghent
University.
Beels, 2005. Experimenteel onderzoek naar de reductie van golfoverslag bij dijken
(Experimental Research to Reduce Wave Overtopping Over Smooth Dike Slopes).
(Master thesis). Ghent University.
Boderé, T., Vanhouwe, G., 2010. Reductie van golfoverslag bij een gladde dijk: combinatie
van een berm en een stormmuur (Reduction of Wave Overtopping Over Smooth
Dikes: Combination of a Berm and a Storm wall). (Master thesis). Ghent University.
Bosman, G., van der Meer, J.W., Hoffmans, G., Schüttrumpf, H., Verhagen, H.J., 2008.
Individual overtopping events at dikes. ASCE, Proceedings of ICCEpp. 2944–2956.
Burchart, H.F., Andersen, T.L., 2006. Overtopping of rubble mound breakwaters with front
reservoir. ASCE, Proceedings of ICCEpp. 4605–4615.
Coeveld, E.M., Busnelli, M.M., van Gent, M.R.A., Wolters, G., 2006. Wave overtopping of rub-
ble mound breakwaters with crest elements. Proceedings of ICCE 2006pp. 4592–4604.
Cornett, A., Li, Y., Budvietas, A., 1999. Wave overtopping at chamfered and overhanging
vertical structures. Proceedings International Workshop on Natural Disasters by
Storm Waves and Their Reproduction in Experimental Basinsp. 14 (Kyoto, Japan).
De Rouck, J., Verhaeghe, H., Geeraerts, J., 2009. Crest level assessment of coastal
structures —General overview. Coast. Eng. 56 (2), 99–107.
EurOtop, 2007. Wave Overtopping of Sea Defences and Related Structures: Assessment
Manual. www.overtopping-manual.com.
Franco, L., de Gerloni, M., van der Meer, J.W., 1994. Wave overtopping on vertical and
composite breakwaters. Proc. 24th Int. Conf. on Coastal Engineeringpp. 1030–1045
(ASCE).
Geeraerts, J., De Rouck, J., Beels, C., Gysens, S., De Wolf, P., 2006. Reduction of wave
overtopping at seadikes: stilling wave basin (SWB) ASCE. Proceedings of ICCE
2006pp. 4680–4691.
Kortenhaus, A., Haupt, R., Oumeraci, H., 2001. Design aspects of vertical walls with steep
foreland slopes. Proceedings of ICE 2001pp. 221–232.
Kortenhaus, A., Pearson, J., Bruce, T., Allsop, W., van der Meer, J.W., 2003. Influence of
parapets and recurves on wave overtopping and wave loading of complex vertical
walls. ASCE, Proceedings of Coastal Structures 2003.
Kortenhaus, A., Geeraerts, J., Hassan, R., 2006. Wave Run-up and Overtopping of Sea Dikes
With and Without Stilling Wave Basin Under 3D Wave Attack (DIKE-3D), Final
Report, Braunschweig, Germany.
Mertens, T., De Wolf, P., Verwaest, T., Trouw, K., De Nocker, L., Coudere, K., 2009. An
integrated master plan for Flanders future coastal safety. ASCE. Proceedings of ICCE
2008pp. 4017–4028.
Owen, M.W., 1980. Design of seawalls allowing for wave overtopping. Hydraulics
Research, Wallingford, Report No. EX 924, UK.
Pearson, J., Bruce, T., Allsop, N.W.H., Kortenhaus, A., van der Meer, J.W., 2004. Effectiveness
of recurve wave walls in reducing wave overtopping on seawalls and breakwaters.
ASCE, Proceedings of ICCE 2004pp. 4404–4416.
TAW, 2002. Technical Report. Wave Runup and Wave Overtopping at Dikes. Technical
Advisory Committee on Flood Defence, Delft.
Tuan, T.Q., 2013. Influence of low sea-dike crow-walls on wave overtopping discharge.
JSCE. Coastal Engineering Journal 55 (4), 1350013 (23 pages).
Van der Meer, J.W., Janssen, J.P.F.M., 1994. Wave run-up and wave overtopping at dikes.
ASCE. In: Kobayashi, N., Demirbilek, Z. (Eds.), Wave Forces on Inclined and Vertical
Wall Structures, pp. 1–27.
Van Doorslaer, K., 2008. Reductie van golfoverslag over dijken door middel van een parapet
(Reduction of wave overtopping over dikes by means of a parapet). (Master thesis).
Ghent University.
Van Doorslaer, K., De Rouck, J., 2010. Reduction of Wave Overtopping on a Smooth Dike
by Means of a Parapet. ASCE. Proceedings of ICCE 2010.
Van der Meer, J.W., Bruce, T., 2013. New physical insights and design formulas on wave
overtopping at sloping and vertical structures. ASCE Journal of Waterway, Port, Coast-
al & Ocean Engineering ASCE, ISSN 0733-950X.
Van Gent, M.R.A., Van den Boogaard, H.F.P., Pozueta, B., Medina, J.R., 2007. Neural network
modelling of wave overtopping at coastal structures. Coast. Eng. 54 (8), 586–593.
Verhaeghe, H., De Rouck, J., van der Meer, J., 2008. Combined classifier–quantifier model:
a 2-phases neural model for prediction of wave overtopping at coastal structures.
Coast. Eng. 55 (5), 357–374.
Victor, L., 2012. Optimization of the Hydrodynamic Performance of Overtopping Wave
Energy Converters: Experimental Study of Optimal Geometry and Probability
Distribution of Overtopping Volumes. Ghent University (PhD Manuscript).