3

S INGLE OVERTOPPING WAVE

IMPACT LOAD

Wave overtopping on a sea dike may pose a threat to people and property. However,
the overtopping features, in particular overtopping wave loads, are not well understood.
The aim of this study was to understand the overtopping process on a dike crest and to
develop an empirical formula for the resulting overtopping wave impact loads on a wall
as a function of the property of the incoming waves and dike geometry characteristics.
In this chapter, we propose a new descriptor (the overtopping momentum flux) in or-
der to predict the impact loads. To validate the proposed empirical function, a series
of physical scale model tests using regular waves was conducted. In these experiments,
we measured the overtopping wave loads on a vertical wall at different locations on a
dike crest, which were induced by broken waves. A correction coefficient for the wall ef-
fect on the initial flow depth, and an empirical initial flow depth coefficient for a broken
wave were determined. These empirical coefficients allowed for an interpretation of the
overtopping process of a broken wave from the dike toe up to the front of the wall on the
dike.

This chapter has been published in: Chen, X., Hofland, B., Altomare, C., Suzuki, T., Uijttewaal, W. (2015). Forces
on a vertical wall on a dike crest due to overtopping flow. Coastal Engineering 95, 94-104.

35
 36 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

Figure 3.1: The process of overtopping waves impacting on a building: ¬ Wind generating waves far away from
3 shoreline; ­ Offshore waves coming into the foreshore area, increasing wave height, decreasing wave length.
Finally, most waves breaking and wave energy dissipating in the form of turbulent bore. ® Turbulent bore
(broken wave) running up on the seaward slope of a dike and overtopping the crest of the dike; ¯ Part of the
overtopping wave continues propagating along the dike crest and the other part flowing back seaward; ° Over-
topping flow hitting the building, with some of it being reflected seaward, some passing through buildings; and
± Overtopping flow going landward.

3.1. I NTRODUCTION
Wave overtopping occurs when the dike crest is lower than the wave run-up height.
Avoiding or reducing overtopping rates under an acceptable limit, is nowadays recog-
nized as one of the design criteria for coastal structures, especially in those countries
characterized by low-lying coastal areas (e.g., the Netherlands, Belgium). In a densely
populated coastal town of these countries, a dike often has a wide crest which serves as
an urbanized frontage, see Fig. 3.1. It is important for designers and owners to recognize
key hazards (e.g., building damage) from overtopping for such buildings standing on
the dike crest. As a necessary design criterion for the buildings, overtopping wave load
has been considered by few studies (Allsop et al., 2005). Though the overtopping wave
characteristics have been investigated by Van Gent (2002), Schüttrumpf and Oumeraci
(2005), and Van der Meer et al. (2010) in the form of flow depth, front velocity and dis-
charge along a dike crest without structures, these characteristics are lacking direct rele-
vance to describe the overtopping wave load.
Chen et al. (2012), De Rouck et al. (2012) and Ramachandran et al. (2012) reported
overtopping wave loads on vertical structures on a dike crest using small and large scale
physical models. However, due to the complex flow characteristics of the highly aerated
overtopping wave, to accurately measure the overtopping wave features on a dike crest
is still a challenge (De Rouck et al., 2012). To date, the studies mainly describe the im-
pact of the overtopping wave phenomenon. A double peaked force evolution shape of a
single impact event is recognized and named as ‘dynamic impact force’ and ‘quasi-static
force’, which is similar to the proposed “church-roof” wave breaking impact evolution by
Oumeraci et al. (1993) for deeper water wave impacts on caissons and composite struc-
tures. However, the overtopping wave has no real sinusoidal wave shape, it is a bore-like
wave, which could qualitatively be considered as a bore (Cox and Machemehl, 1986).
Therefore, overtopping wave impact is outside the classification of the wave breaking
impact, as developed by Oumeraci et al. (1993). The mechanisms of impact and the
unique overtopping wave characteristics still need to be understood better.
In order to characterize the overtopping wave impacts against a building (simpli-
fied as a vertical wall in this paper) on a dike, the literature about two other prototype
3.1. I NTRODUCTION 37

problems with similar configurations was investigated: broken wave impact on a crown
wall of rubble mound breakwater and tsunami bore impact on a building inland. Over-
topping events at crown walls of a rubble mound breakwaters are very similar to the
overtopping of a dike: the incoming wave breaks on the seaward slope of the breakwa-
ter (processes ¬ to ® in Fig. 3.1) and the resulting broken wave overtops the crest and
impinges on the crown wall (processes ¯ to ° in Fig. 3.1). An important difference is
that the slope and crest are very rough and permeable, which will alter the approaching
flow. Many studies exist of wave loading on crown walls (e.g., Hamilton and Hall, 1992;
Pedersen, 1996; Martin et al., 1999). These studies reported that the horizontal force is
proportional to the incident wave height and wave period. Hamilton and Hall (1992) 3
found that lower crown walls yield smaller wave loads compared with a non-overtopped
high wall. Pedersen (1996) reported that a gentle seaward slope reduces wave loading.
Based on the argument from Cox and Machemehl (1986) about the similarity of a broken
wave and a bore , Martin et al. (1999) related the problem of wave impact on the crown
wall of rubble mound breakwaters to that of bore impact on a vertical wall and pointed
out the quasi-static force is most likely larger than the dynamic force. Kortenhaus and
Oumeraci (1998) mentioned that the order of magnitude of the broken wave forces is
the same as that of slightly breaking waves. Regarding the tsunami bore impact on ver-
tical walls (similar to processes ¯ and ° in Fig. 3.1), some pioneering analytical and
experimental researches can date back to Cumberbatch (1960) who provided an analyt-
ical solution for the impact pressure of an uniform steady flow striking on a wall. Cross
(1967) further developed Cumberbatch’s model by adding a hydrostatic pressure term
to calculate the total tsunami surge force. Ramsden (1996) investigated the interaction
of long waves, solitary waves, bores and surges with a vertical wall. The author observed
that the maximum measured total force is quasi-static but less than the hydrostatic force
computed by using the maximum measured run-up height of the wall. Arikawa (2008)
examined the impulsive tsunami loads on vertical concrete walls by using large scale
flume tests. Nouri et al. (2010) investigated the interaction between a hydraulic bore
and the impacted structures by using a dam-break approach experimentally. Several au-
thors also studied tsunami forces on various types and shapes of structures (e.g., Arna-
son, 2005; Arnason et al., 2009; Nouri et al., 2010). Since the possible similarity between
an overtopping wave and a bore, the existing knowledge of bore impact could be ap-
plied to the total overtopping wave force predictions. An important difference between a
tsunami bore and overtopping wave is that overtopping waves are induced by stochastic
storm waves. The overtopping wave impact is influenced by the presence of previously
reflected waves. This effect is also less for crown walls on rubble mound structures, as
the water of successive waves can penetrate into the rubble mound material.

The objective of this study is to develop an empirical formula to predict the over-
topping wave load on a vertical wall on a wide crested dike, such as found in low-lying
countries. In Section 3.2, a new descriptor, named as overtopping momentum flux, to
characterize the total overtopping wave load on a vertical wall is proposed. An empiri-
cal formula based on the proposed descriptor is calibrated by a series of physical model
tests with regular waves in a shallow water condition. The test set-up and the results are
provided in Section 3.3 and Section 3.4. Finally, in Section 3.5 and 3.6, a discussion and
conclusions are given.
 38 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

3.2. OVERTOPPING MOMENTUM FLUX

3.2.1. WAVE MOMENTUM FLUX
Hughes (2004a,b) proposed a method of using maximum depth-integrated wave mo-
mentum flux (M F,max ) to describe the interaction between nearshore waves and a coastal
structure, see Eq. (3.1):
Z η(x)max
p d + ρu 2 d z
¡ ¢
M F (x, t )max = (3.1)
−h
3 where M F (x, t )max is maximum depth-integrated wave momentum flux per unit of width
at moment t ; x is horizontal coordinate; z is vertical coordinate, positive upward with
z=0 at still water level; t is time; p d is instantaneous wave force (or dynamic pressure) on
the wall; u is horizontal water particle velocity at the same position; ρ is water density;
h is water depth in front of the dike toe; η(x) is sea surface elevation at location x. For
convenience, the “max" subscript will be removed with following: M F .
Hughes (2004a) made a simple physical argument: the horizontal velocity u is zero
when a perfect reflection occurs on an impermeable vertical wall. Then Eq. (3.1) simply
becomes the integral over the water depth of the dynamic pressure, which is also the
total instantaneous wave force on the wall. And the maximum depth-integrated wave
momentum flux M F can be determined for any surface wave if the velocity and pressure
field are specified.
M F is a useful concept for the application to both the periodic and transient wave
types (Hughes, 2004b), the argument of the relevance of M F and wave force on struc-
tures becomes practical. Therefore, M F is assumed to be very useful to describe the
overtopping wave load as well.

3.2.2. D ERIVATION OF OVERTOPPING MOMENTUM FLUX

Fig. 3.2 gives a definition sketch of an overtopping wave propagating on a wide crested
dike with boundaries setting as dike toe x t oe and the location of wall x w , covering three
zones: Zone 3 (run-up), Zone 4 (overtopping) and Zone 5 (impacting), referring to Fig. 3.1.
In the following sections, the overtopping momentum flux is derived for Zone 3 to 5.

I NITIAL OVERTOPPING MOMENTUM FLUX IN Z ONE 3

In Zone 3, the incoming wave strikes onto the dike slope and runs upward. Fig. 3.3 a
illustrates the moment of the incoming wave arriving at the point of maximum virtual
wave run-up. The maximum depth-integrated momentum flux (M F ) of the incoming
wave before it reaches the dike toe is proportional to the weight of the water contained
in the wedge area ABC (W(ABC ) ) (Archetti and Brocchini, 2002; Hughes, 2004a). Here the
authors defined the overtopping momentum flux (M Fov ), which is proportional to the
water mass contained in the wedge area CEF (W(C E F ) ) above the dike transition point
(x t r ), based on the same physical argument made by Hughes (2004a), i.e.,

K P M Fov = K o W(C E F ) (3.2)

where K o is an unknown constant of proportionality and K P is a reduction factor to ac-

count for slope porosity which was defined by Hughes (2004a) (K P = 1 for the imperme-
3.2. O VERTOPPING MOMENTUM FLUX 39

❸ ❹ ❺

z
Zonetr Zonede
t  to
vf Overtopping
wave front

Wall
x (a)

z
vf
Spray
splashing
x (b)
3
z
Reversed flow vf

x (c)

z t  ti
Air bubble
Reversed flow vf

x (d)
xtr xde xim xw
xtoe
Overtopping wave front
Propagated overtopping flow
Reversed overtopping flow

Figure 3.2: Sketch of the propagation of overtopping wave on dike crest

able slopes) as well. W(ABC ) in Fig. 3.3a is given by

ρg R u2 tan β
µ ¶
W(ABC ) = −1 , (3.3)
2 tan β tan θ

where β is the dike slope, θ is an angle between the run-up water surface and the still
water level, R u is the fictitious maximum wave run-up height. The weight of water per
unit of width contained in the area CEF is proportional to W(ABC ) , thus
¶2
Ru − Rc
µ
W(C E F ) = W(ABC )
Ru
ρg (R u − R c )2 tan β
µ ¶
= −1 (3.4)
2 tan β tan θ
ρg 2
(R u − R c ) cot θ − cot β ,
¡ ¢
=
2
where R c is the distance between the dike crest level to still water level (or freeboard).
Note that θ depends on the dike slope and wave steepness (Yamamoto and Horikawa,
1992), but the whole term “cot θ − cot β" can be a function of dike slope and needs to be
determined empirically (Hughes, 2004a). Substituting Eq. (3.4) into Eq. (3.2) and rear-
 40 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

z
C

G Ru
E  zcrest
 F R
c

A B D
h
 ztoe
x
R
Length AD  u
tan 
3 Length BD 
Ru
tan 
1 Ru  tan  
2
(a) Area ABC    1
2 tan   tan  

z
Length FG  d 0

G
E d0 zcrest
 F R
c

A B
h
(b)  ztoe
xtoe xtr xwall x

Figure 3.3: (a) Definition sketch of maximum wave run-up on a smooth impermeable plane slope presented
by Yamamoto and Horikawa (1992) firstly, adapted from Hughes (2004a), (b) schematic sketch of overtopping
wave in real situation.

ranging yields an equation for the initial overtopping momentum flux, i.e.,

K o W(C E F )
M Fov =
KP
(3.5)
ρg K o
(R u − R c )2 cot θ − cot β .
¡ ¢
=
2 KP

The maximum initial overtopping wave depth d 0 at x t r (see Fig. 3.3b) is proportional
to the difference between R u and R c (Schüttrumpf, 2001; Van Gent, 2002; Schüttrumpf
and Oumeraci, 2005; Van der Meer et al., 2010), and can be expressed as

d 0 = C 1 (R u − R c ) , (3.6)

where C 1 is an empirical coefficient.

Substituting Eq. (3.6) into Eq. (3.5), it yields

ρg K o
M Fov = d 2 f (β), (3.7)
2 K P C 12 0

where f (β) is a function of the dike slope, and needs to be determined empirically. After
3.2. O VERTOPPING MOMENTUM FLUX 41

rearranging, Eq. (3.7) can be rewritten as

M Fov = C 2 f (β)ρg d 02 (3.8)

where C 2 is an unknown empirical coefficient.

OVERTOPPING FLOW IN Z ONE 4

When the overtopped flow front drops on the dike crest, the overtopping wave experi-
ences a kind of nappe impact first (Fig. 3.2b), then propagates horizontally to the wall
(Fig. 3.2c), and finally impacts on the wall (Fig. 3.2d). 3
Based on different mechanisms of propagation of the overtopping wave, Zone 4 has
been subdivided into two parts: transition zone (Z one t r ) and decay zone (Z one d e ).
Within the transition zone, part of overtopping wave returns back to the sea. The over-
topping momentum flux loses a certain part here. Due to the complexity of the momen-
tum exchange process, an unknown factor C 3 is introduced to describe this reduction:

M F0 = C 3 M Fov = C 4 f (β)ρg d 02 , (3.9)

where M F0 is the maximum depth-integrated momentum flux at x d e . x d e is the location

where the overtopping wave particle velocity transfers horizontally, and the flow behaves
as a progressive wave. C 4 is an unknown coefficient need to be determined experimen-
tally.
The maximum overtopping wave depth, d x at a specific location x along the dike
crest, follows an exponential decay function (Van Gent, 2002; Schüttrumpf and Oumeraci,
2005; Van der Meer et al., 2010, and so on), and is expressed as

dx x − xt r
= exp(−λ0 ), (3.10)
d0 Bc

where B c is the width of the dike crest, and λ0 is an empirical coefficient to describe the
overtopping depth decay rates. Since M F0 is a function of d 0 , it would be reasonable
to regard M F (x) at the specific location x on the dike following an exponential decay as
well. Thus M F (x) can be predicted by

M F (x) x − xt r
= exp(−λ1 ), (3.11)
M F0 L

where L is local wave length at the dike toe, calculated by L = C T . T is the local wave
period, and C is the incoming wave local phase celerity, λ1 is an empirical dimensionless
coefficient for the decay rate. Herein, the authors used the local wave length L instead of
B c in Eq. (3.10) due to the irrelevance of dike crest width for the overtopping wave impact
event, when a wall appears on the dike.

OVERTOPPING FLOW IN THE IMPACT ZONE (Z ONE 5)

In the impact zone (x i m É x < x w , as seen in Fig. 3.2), the dominant process of the over-
topping wave changes from energy dissipation into a violent impingement. The appear-
ance of the vertical wall on the dike suddenly alters the features of the incoming over-
topping wave. However, the flow features within the tip region are assumed constant
 42 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

(Cross, 1967) before the tip impacts on the wall. Then going back to the argument made
by Hughes (2004a), the maximum depth-integrated momentum flux M F is a physically
relevant descriptor of the wave force on a structure with a force per unit of wave width.
Thus the maximum total impact force on the wall is F t = α1 K R M F (x w ), where M F (x w )
is the maximum momentum flux near the wall and K R is a reduction factor to account
for the wall height (K R = 1 for high vertical wall which means there is no momentum
loss at x w ), and α1 is an empirical coefficient. Combining the previous equations, the
maximum total impact force on the wall F t can be expressed as:

3 Ft
= exp(−λ1
B
), (3.12)
C 5 f (β)ρg d 02 L

in which B = x w − x t r is the boulevard width, and C 5 is an empirical coefficient.

3.2.3. I NITIAL OVERTOPPING WAVE DEPTH

Eq. (3.12) is the derived empirical equation for the overtopping wave load, in which C 5 ,
f (β) and λ1 need to be determined by experiments. The next step is to connect d 0 in
Eq. (3.12) with the local incident wave conditions. Martin et al. (1999) predicted the
broken wave load on a crown wall for a rubble mound breakwater by estimating d 0∗ as

Rc
d 0∗ = α2 H (1 − ), (3.13)
Ru

where d 0∗ is an imaginary overtopping wave depth at the dike transition point x t r , thus
the effect of the wall on the real overtopping wave depth is not taken into account; α2 is
an empirical coefficient. In Martin et al. (1999)’s study, α2 = 1 is determined empirically
from Fig. 7 of Yamamoto and Horikawa (1992). However the value of α2 = 1 is not suit-
able for the present study, it needs to be determined by experiment. For regular wave
run-up height R u on a infinite beach slope, Eq. 3.14 is empirically used:

R u = ξ0 · H , (3.14)
p
where ξ0 is Iribarren number (= tan β/ H /L 0 ). For large values of ξ0 , R u is only related to
the incident wave height, and the ratio of R u /H tends to a constant value. The constant
is within a range of 2 to 3 (e.g., Hunt, 1959; Battjes, 1974; Peregrine and Williams, 2001).

3.3. E XPERIMENT SET- UP

3.3.1. FACILITIES
Physical model tests were performed in a 4 m wide, 1.4 m deep and 70 m long wave
flume at Flanders Hydraulic Research, Antwerp, Belgium. A piston-type wave genera-
tor with a stroke length of 0.6 m was used generating monochromatic, multi-chromatic
and random waves, using a passive wave absorption system. The wave flume was split
into four sections (approx 1 m for each), as shown in Fig. 3.4. One reason of this de-
sign was to test the configurations with and without a wall simultaneously, the other
one was to reduce reflected waves by installing passive wave absorption baffles in the
3.3. E XPERIMENT SET- UP 43

outer sections. The reflection coefficient of the passive wave absorbers is 0.16. By lim-
iting the time series of each experiment to the traveling time of the reflection towards
the wave generator, the reflections would not affect the measurements. Incident wave
conditions were determined by measuring the waves without a dike in the outer section
(Fig. 3.4b); Section A was used to measure the unobstructed overtopping wave features
along the dike crest and Section B was used to measure the impact force of overtop-
ping wave. In Section A, the overtopping wave was collected by an overtopping tank
behind the dike model, named as unobstructed flow, see Fig. 3.4c. In Section B, the over-
topping wave was reflected back seawards after its impingement, named as obstructed
flow, see Fig. 3.4d. The shallow foreshore profile and the dike models were made of con- 3
crete and wooden plates respectively. The main parameters are depicted in Fig. 3.4.

0.8 m
Outer section:foreshore Wave absorbers

Dike slope

Dike crest
Wave maker

1m
Section A: foreshore+dike

4m
Section B: foreshore+dike+wall

1m
0.8 m
Wave absorbers B
(a)
0 11 45 45.3 45.8 70
z
Outer section
Z (m)
Wave maker

H o , To H ,T
x h
h0 (b)
Foreshore
0 11 45 70
Section A
Z (m)
Wave maker

d A0

Rc Dike
Foreshore xtoe xtr (c)
0 11 45 45.3 70
Wall
Z (m)

Section B
Wave maker

dB0

B Dike
xtoe xtr xw (d)
Foreshore
0 11 45 45.3 45.8 70
X (m)

Figure 3.4: Wave flume in Flanders Hydraulic Research (Antwerp, Belgium). (a) is a top view of the flume, (b),
(c) and (d) are the respective sections: ‘outer section’, A and B.

Two different dike slopes of cot β = 3 and cot β = 6 were tested which covers a typical
range of coastal dike slopes representative for low-lying countries, see Schüttrumpf and
Oumeraci (2005). For the steep slope (cot β = 3), tests were conducted with B = 0.25, 0.5
and 0.75 m to investigate the spatial effect of overtopping wave impact force. For the
gentle slope (cot β = 6), only B = 0.5 m was tested. The foreshore slope was 1:35 and dike
height 0.1 m. The foreshore slope created a depth-limited condition for incident waves
and thus only two still water depth h 0 = 0.96 and 1 m were tested.
In order to calibrate the empirical formula proposed in Section 2, which is derived
from the overtopping process of a single wave, the test program was restricted to regular
waves only. The ranges of the parameters varied in the tests are listed in Table 3.1 and
 44 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

Table 3.2. The symbols in the tables are illustrated in Fig. 3.4. For each water depth,
regular waves with offshore wave steepness (H0 /L 0 ) from 0.003 to 0.021 were generated.
This leads to the incident waves breaking on the foreshore slope. The resulting broken
wave (bore) heights at the dike toe are within a range of H = 0.03 − 0.1 m in model scale.

Table 3.1: Ranges of structure dimensions for dike model

R c (m) tan β (-) B (m) R c /B (-)

0.012;0.052 1:3 0.25; 0.5; 0.75 0.016; 0.21
3 0.012;0.052 1:6 0.5 0.024; 0.1

Table 3.2: Ranges of wave conditions

h 0 (m) h (m) H0 (m) T0 (s) H (m)

0.96; 1 0.05; 0.09 0.1; 0.2; 0.3 3; 4; 5 0.04-0.076
T (s) ξ0 H /h R c /H h/L
3-5 2.3-10 0.6-1.1 0.16-1.3 0.01-0.025

The key parameters investigated include incident wave height near the toe of the
dike H , incident wave period near the toe of the dike T , overtopping wave depth near
x t r , d A0 in Section A and d B 0 in Section B. All these parameters were measured directly
or indirectly by wave gauges. Home-made resistance type wave gauges recorded the
water surface elevation at a sampling rate of 20 Hz throughout the entire duration of 80
seconds in each test. The number of waves per test varied between 10 and 15, depending
on the input wave period. These short test durations avoid the influence of lacking active
wave absorption. The locations of measured wave or flow surface elevation can be found
in Fig. 3.4 and Table 3.3, in which H0 is offshore wave height and H is local wave height
at the dike toe.
In total 118 regular wave tests were conducted and 39 unique tests. All of the tests
with B = 0.5 m, cot β = 3 and h 0 = 1 m were repeated at least 2 times. The test repeata-
bility was investigated in a series of tests with the same configuration (h 0 = 1 m, H0 = 0.2
m, T0 = 4 s, B = 0.5 m and cot β = 3) with 36 repeated test runs. Good test repeatability
resulted in an averaged 0.24% standard deviation of overtopping wave depth d B 0 . Based
on the good repeatability and the limited test time, no extra repeated tests were done for
other configurations. Therefore, the results of the present work are based on the evalua-
tions of 39 different test runs.

3.3.2. WALL MODEL AND SETUP

The wall model consisted of two parts: a force-measuring portion and a fixed wall. The
force-measuring portion consists of an aluminium plate of 1 cm thickness, supported
by an aluminium framework. Narrow gaps (1 mm) separated the aluminium plate from
the fixed wall. The wall part was made of PVC board, fixed into Section B. Two-load
cells of model series Tedea-Huntleigh 614 were used to determine the total overtopping
wave force with sampling rate of 1000 Hz, which were mounted with the aluminium
plate. The geometry of the wall model, supporting framework and positioning of the load
3.3. E XPERIMENT SET- UP 45

Table 3.3: Locations of wave gauge group for steep slope tests

Wave measurement
Near wave maker x (m)
H0 7.85
8.45
9.32
10.42
Outer section x (m)
H 44.96 3
Flow depth measurement
Section A x − x t r (m)
d A0 0.025
Section B x − x t r (m)
dB 0 0.025

Figure 3.5: Wall model and force measuring system including (a) load cell (b) positioning of load cells and
supported frameworks and (c) front view of wall model and locations of pressure sensors

cells are shown in Fig. 3.5. Next to the force-measuring portion, four pressure sensors
were mounted on the face of the wall. They are sampled at a rate of 1000 Hz in order to
detail the local pressure evolution and to help understand the impact mechanism of the
overtopping wave. The lowest sensor (P 1 ) was mounted 3.5 cm above the dike crest and
the space between each two sensors’ center was 4 cm.

3.3.3. D ATA PROCESSING PROCEDURES

I NCIDENT WAVE CONDITION
After the experiments, the wave and force data were processed. Firstly, incident wave
height (H ) and period (T ) at the dike toe were extracted from the wave surface elevations
 46 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

collected by the wave gauges installed at the dike toe in the outer section, using the zero-
down-crossing method.

OVERTOPPING FLOW DEPTH

0.15
dB0(t)
First wave dA0(t)
Overtopping flow surface elevation(m)

3 0.1

Incident overtopping flow peak Reflected overtopping flow peak

0.05

dB0 d A0

40 42 44 46 48 50 52
Time(s)

Figure 3.6: An example of overtopping wave surface elevation near the edge of dike crest, 2.5 cm to x t r .

The peaks of overtopping wave depth along the dike crest in both Section A and Sec-
tion B were picked from the time series of the flow surface elevation. Fig. 3.6 illustrates
an example of time series of overtopping wave surface elevations in both of Section A
and B. The dashed line indicates the flow time series in Section A and the solid line in-
dicates that in Section B. Due to the reflection from the wall, the overtopping wave in
Section B shows a two-peak time series signal (e.g., incident flow peak and reflected flow
peak). Only the incident flow peak (the d A0max and d B 0max ) is chosen for further anal-
ysis. The subscript “A" refers to Section A and “B" to Section B. For convenience, the
subscript “max" was removed from the maximum flow depth yielding d A0 and d B 0 in
this study. As the first wave generated is higher than the subsequent ones, which doesn’t
represent a wind-wave overtopping event, see Fig. 3.6. Thus, the first overtopping event
was removed from the processed database. The final data of d A0 and d B 0 used for anal-
ysis were the averaged value from three overtopping waves which gave the largest three
total impact loads on the wall of each test.

F ORCE DATA
The signals of the load cells were digitally filtered. A cut-off frequency of 50 Hz was cho-
sen for the load cells, in order to keep the characteristics of force signal and at the same
time reduce noise. The forces measured by load cells are compared with the integrated
forces from the pressure sensors, see Fig. 3.7a. The two-peak force signal of a single im-
pact event was observed. The first peak is called dynamic impact force and the second
one is called quasi-hydrostatic force. The second peak is generally higher than the first
peak in the low frequency force measurement, thus most important for the overall sta-
bility of large structures like houses and walls. Therefore, the analysis focuses on the
3.4. R ESULTS 47

(a) Time series of force (b) Pressure distribution at t=49.9 s

90
Ft (t) 0.18 Pressure sensors
Quasi-static force Fpt (t) Linear fit
80
0.16
70
Overtopping flow force(N/m)

0.14
60 Dynamic impact force 0.12
50

Wall (m)
0.1
40
0.08
30
0.06
20
0.04
10

0
0.02 3
0

49 50 51 52 53 54 55 0 1 2 3 4 5 6 7 8 9
Time(s) Pressure (100N/m2)

Figure 3.7: An example of overtopping wave force measurement. (a) a comparison of flow force (F t ) by load
cells (straight line) and flow force (F pt ) by pressure sensors (dashed line). F pt is obtained by integrating the
raw pressure data following the method of Ramachandran et al. (2012). (b) pressure distribution at t = 49.9 s.

second peak. Pressure sensors work well for the first peak with short impact duration,
and both the pressure sensors and load cells give a similar good performance on the
quasi-hydrostatic force measurement. Also the large force inferred from the pressure
sensors is probably due to the fact that a local high pressure is attributed to a large area.
Fig. 3.7b illustrates the linear pressure distribution at the moment t = 49.9 s, which pro-
vides a proof of quasi-hydrostatic nature of the second peak force. Furthermore, the data
of maximum total impact force F t were determined by averaging the values of the three
highest measured quasi-hydrostatic forces of each wave train by load cells. Based on the
measured hydrostatic pressure at the time of occurrence of the second force peak, the
point of application of the measured force will be at a distance of z a = 13 2F t /ρg from
p

the bed. The localized flow impact mechanism and its dynamic impact force are not
included in the current chapter.

3.4. R ESULTS
3.4.1. D ETERMINE THE UNOBSTRUCTED FLOW DEPTH d A0
In the current study, the incident wave breaks on the foreshore, the resulting broken
wave continues running up to dike slope and generating the overtopping wave. The
range of the calculated ξ0 is from 2.3 to 10. A regular wave run-up formula R u proposed
by Schüttrumpf (2001) was applied which covered a wide range of ξ0 from 0.9 to 9.6.
The measured unobstructed flow depth d A0 was compared with Eq. (3.13), and the
authors get
Rc
· ¸
d A0 = 0.77H 1 − , (3.15)
Ru
where R u = 2.25H tanh(0.5ξ0 ). 0.77 is the empirical coefficient α2 in Eq. (3.13), deter-
mined by fitting measured d A0 with H (1−R c /R u ). Fig. 3.8 plots the estimates of overtop-
ping wave depth in Section A with the measurements. Eq. (3.15) works well for providing
the good trend but with much scatter. Some of the scatter was caused by the limitation
 48 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

of the measurement using wave gauges, whereas some of it came from the choice of α2 .
A further discussion on α2 , is given in Section 3.5.

0.06
cot β=3,
0.05 cot β=6,

3 0.04
cot β=6,

Best fit
 m

0.03
d A0

y = 0.77x
0.02 R² = 0.89

0.01

0.00
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
 Rc 
H 1    m
 Ru 

Figure 3.8: Measured d A0 versus Eq. (3.13) with α2 = 0.77.

3.4.2. D ETERMINE FLOW DEPTH CONSIDERING THE ABSENCE OF WALL

Since d A0 is commonly used with the assumption of no wall on the dike, the authors
defined a coefficient C t r to correlate the unobstructed flow depth (d A0 ) and obstructed
flow depth (d B 0 ) by considering the influence of the wall,

d B 0 = C t r d A0 . (3.16)

where C t r is obtained experimentally. In the present study, d A0 was measured at 2.5 cm

behind x t r in Section A, and d B 0 was measured at the same location but in Section B, see
Fig. 3.6. Fig. 3.9 shows the comparison of the results d B 0 /d A0 , and an empirical formula
for C t r given by the best-fit curve,

B
µ ¶
C t r = 0.33 · l n + 1.86. (3.17)
L

In Fig. 3.9, even though there is a lot of scatter, an interesting trend is observed: C t r
crosses two regions: C t r < 1 and C t r > 1. When C t r < 1, it indicates that the d B 0 would be
overestimated by using d A0 to predict F t . It is a fact that the effective overtopping water
mass contribution to the impact process is a function of the relative location of the wall
3.4. R ESULTS 49

1.8
y = 0.33ln(x) + 1.86
1.6 R² = 0.57
1.4

1.2
d B 0 d A0

1.0

0.8 B=0.25, cot β=3

0.6
B=0.5, cot β=3
B=0.75, cot β=3
3
0.4 B=0.5, cot β=6
Best fit
0.2 dA0=dB0
0.0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
B L

Figure 3.9: Influence of wall on initial overtopping wave depth, d A0 is measured at 2.5 cm behind x t r in Section
A and d B 0 is measured at the same location but in Section B.

B /L. Only part of overtopping wave of one overtopping event will finally impact on the
wall. And when C t r > 1, d B 0 would be underestimated by using d A0 to predict F t because
of the residual water layer left by the previous overtopping event which would increase
the level of d B 0 .

3.4.3. D ETERMINE OVERTOPPING MOMENTUM FLUX

In this section, the empirical coefficients (e.g., C 5 , f (β), and λ1 ) as proposed in Eq. (3.12)
for the maximum total impact force F t are determined by fitting the equation to experi-
mental data.
Fig. 3.10 shows the results from cot β = 3 with three different boulevard widths B
(from x t r to x w ). d B 0 is measured at 2.5 cm behind x t r in Section B. From the best-fit
curve, 4.9 represents the factor C 5 f (β) and 8.86 represents the factor λ1 with a regression
coefficient R 2 = 0.84. There is some scatter mainly for the case B = 0.25 m. A reasonable
explanation is that the wall is located within or close to the transition zone, where the
characteristics of the overtopping wave are complex, see Fig. 3.2b.
Data of slope cot β = 6 is compared with those of slope cot β = 3 for the same B = 0.5
m, shown in Fig. 3.11. The solid squares indicate the tests of cot β = 3 and the open
squares indicate the tests of cot β = 6. It is found that a gentle slope gives a smaller over-
topping wave load comparing to a steep slope when B /L is between the range of 0.1 to
0.15 in accordance with finding by Pedersen (1996).
Table 3.4 shows the results of empirical coefficients for both dike slope tests. Sub-
sequently an empirical dike slope function f (β) = cot β and λ1 = λ2 cot β can be de-
termined based on the results shown in Table 3.4, where λ2 is an empirical coefficient.
Then the proposed overtopping wave load relationship given by Eq. (3.12) can be fitted
to the test results for slopes cot β = 3 and 6 with the dike slope function f (β) = cot β, see
 50 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

6
y = 4.90e-8.86x B=0.25, cot β=3
5 R² = 0.84
B=0.5, cot β=3
B=0.75, cot β=3
4 Best fit
Ft  gd B 0 2

3 3

0
0.00 0.10 0.20 0.30
B L
Figure 3.10: Test results for cot β = 3.

Table 3.4: Results of λ1 and c 5 f (β) for cot β = 3 and 6

cot β 3 6
λ1 11.41 20.3
C 5 f (β) 6.43 12.63
R2 0.85 0.92

Fig. 3.12. A best-fit of Eq. (3.12) to the overtopping wave data leads to:

Ft B
= 1.7 cot β exp(−3.08 cot β ) (3.18)
ρg (d B 0 ) 2 L

where (1.7 cot β) represents the empirical function C 5 f (β) and (3.08 cot β) represents
the decay coefficient λ1 in Eq. (3.12). Note that the dike slope function f (β) = cot β
is determined empirically in the present study, so it is only applicable for the range of
cot β = 3 − 6. A discussion on the dike slope function will be provided in Section 3.5.
For practical reasons, d B 0 used in Eq. (3.18) is not obtainable if there were no wall
present on the dike. Thus, there is a limitation of using Eq. (3.18) to predict F t at specific
locations on the dike. Therefore, including C t r to correct d B 0 , a final equation can be
written as:
Ft B
= 1.7C t2r cot β exp(−3.08 cot β ). (3.19)
ρg (d A0 )2 L
3.5. D ISCUSSION 51

4.5

4 y1 = 6.43e-11.41x
R² = 0.85
3.5
y2 = 12.63e-20.30x
3 R² = 0.92
Ft  gd B 0 2

2.5

2 3
1.5 B=0.5, cot β=3
B=0.5, cot β=6
1
Best fit (y1)
0.5 Best fit (y2)
0
0.00 0.05 0.10 0.15 0.20 0.25
B L
Figure 3.11: Test results of different dike slope for the same B = 0.5 m condition: solid square indicates the case
for cot β = 3 and hollow square indicates the case for cot β = 6.

The predicted force (F cal ) using Eq. (3.19) is compared with the measured F t , showing a
good performance with a regression coefficient R 2 = 0.95, see Fig. 3.13.
Finally the authors use Eq. (3.15) and Eq. (3.19) to estimate the overtopping wave
force, shown in Fig. 3.14. A qualitative comparison between Fig. 3.13 and Fig. 3.14 shows
an increased scatter with substituting Eq. (3.15) into Eq. (3.19) as shown in Fig. 3.14, com-
pared to the good performance of Eq. (3.19) by using measured d A0 shown in Fig. 3.13.

3.5. D ISCUSSION
Prior work has documented the characteristics of overtopping waves on a dike (e.g., flow
depths, flow velocities and discharges) and the impacts of bore-like waves on a vertical
wall (e.g., broken wave impact on a rubble mound breakwater and tsunami bore impact
on a wall). However, these studies have been lacking either direct relevance to describe
the overtopping wave load or similar configurations.
In this chapter, a simple formula for overtopping wave load on a vertical wall placed
on top of a wide crested impermeable dike was derived, based on two arguments: one:
is the weight of water contained in the upper part of a run-up wedge above the dike crest
level is proportional to the initial maximum depth-integrated overtopping momentum
flux, and the other: the overtopping momentum flux is following an exponential decay.
A slope function and two key coefficients were determined by physical model tests using
regular waves.
The overall satisfactory performance of the proposed crude formula for the measure-
ments, demonstrates that applying the maximum depth-integrated wave momentum
 52 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

1.6
y = 1.70e-3.08x B=0.5, cot β=3
1.4
R² = 0.92 B=0.25, cot β=3
 cot   1.2 B=0.75, cot β=3
1 B=0.5, cot β=6
Best fit
0.8
2
  gdB0

3 0.6

0.4
Ft

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2
cot   B L

Figure 3.12: Test results of different dike slopes and different B with considering dike slope function cot β.

100
B=0.25, cot β=3
B=0.5, cot β=3
80 B=0.75, cot β=3
B=0.5, cot β=6
X=Y

60
 N m
Ft

40

20

R 2  0.95
RMSD  5.12
0
0 20 40 60 80 100
Fcal  N m
Figure 3.13: Measured F t versus Eq. (3.19).
3.5. D ISCUSSION 53

100
B=0.25, cot β=3
B=0.5, cot β=3
80 B=0.75, cot β=3
B=0.5, cot β=6
X=Y

60
 N m

3
Ft

40

20

R 2  0.88
RMSD  8.1
0
0 20 40 60 80 100
Fcal  N m
Figure 3.14: Measured F t versus predicted flow force using Eq. (3.15) and Eq. (3.19).

flux method proposed by Hughes (2004b) for overtopping wave. The author also defined
a transition zone at the beginning of the dike crest and a decay zone. The former one is
used to connect the wave run-up zone with the latter one. The overtopping wave in the
decay zone is assumed similar to a progressive wave, and is following an exponential de-
cay shape. This exponential decay shape of the predictive formula is in good agreement
with the spatial distribution of the overtopping discharges, see Pullen et al. (2007). From
the results, the authors found that there is much scatter for the tests with a smaller value
of B /L for cot β = 3 condition. This could suggest the differences of flow characteristics
between the transition zone and the decay zone; Meanwhile Fuchs (2013) reported the
importance of the dike slope for the overtopping wave features near the beginning of the
dike crest, but not far inland. This gives a reasonable proof of the classification of the
two zones. However more work still needs to be done to confirm the boundary between
the transition and the decay zone in the future.
The overtopping momentum flux (M Fov ) introduces an empirical dike slope function
f (β) to the overtopping wave load formula with a physical relevant meaning. Though
Hughes (2004a) has already established a series of empirical slope functions for the run-
up height of regular waves, irregular waves, and solitary waves, it is still lacking for reg-
ular broken waves. Interestingly, Hughes’ (2004a) dike slope function for regular wave
run-up height contains a simple term of tan β, and it works well in a wide range of
cot β = 2 − 10. The overprediction for the case with cot β < 2 indicates the limitation
 54 3. S INGLE OVERTOPPING WAVE IMPACT LOAD

of using the argument of the triangular shape of water wedge for a steep dike, because
the run-up water surface is not straight, but concave (Hughes, 2004a). In the present
study, the empirical slope function was determined in terms of cot β, and was covering
the range of cot β = 3 − 6. This is in line with the basic shape of Hughes’(2004a). There-
fore, the dike slope function in the present study is reasonable, and could be expanded
to the range of cot β = 2 − 10, but still needs to be confirmed. Herein, the authors would
like to note that the dike slope function contains the dike slope β and the run-up water
surface angle θ. The reason of empirically determining f (β) is the difficulty of estimating
the run-up surface angle θ. Yamamoto and Horikawa (1992) proposed a method for esti-
3 mating θ, which depends on the incident wave steepness and dike slope. Yamamoto and
Horikawa’s (1992) θ function provides a potential of combining Hughes’(2004a) work to
develop a uniform empirical dike slope function for different wave types in future stud-
ies.
In the present study, the dike slope function f (β) was introduced in the decay coeffi-
cient λ1 of the proposed overtopping wave load formula. The final load formula includ-
ing f (β) in the decay coefficient fits the experimental data well. In order to explain this
good fit, a simple physical argument of the exponential decay function is given. A clas-
sical exponential decay function contains two constant terms, one is an initial quantity
and the other one is a constant decay coefficient. The ratio of these two constants deter-
mines the uniqueness of an exponential decay function. In the present study, the initial
amplitude of the proposed exponential function for overtopping wave load includes a
dike slope function, left hand side of Eq. (3.12). The incident wave conditions were the
same for different dike slope tests, and the position of the dike toe was kept constant
in Section A and B. Thus it is reasonable to include the same dike slope function in the
decay coefficient λ1 in Eq. (3.12).
From the test results, the gentle slope reduces the overtopping wave load which is in
line with the observations from Pedersen (1996). This reduction is expected, due to the
longer travel distance of overtopping waves along the gentle slope which causes more
energy loss. Whereas from the decay coefficient point of view, a gentle slope gives a
faster drop of the overtopping wave load.
In this study, the initial overtopping wave depth measured in Section B (with a wall)
was used to develop the flow load formula. Considering the effect of the wall on the
overtopping wave features, and the availability of an overtopping wave depth formula
in literature, a coefficient C t r was defined to correct the obstructed overtopping wave
depth in Section B. If the wall is too close to the seaward edge of dike crest, the real over-
topping wave volume in front of the wall would be overestimated in comparison with
the situation without considering the wall, whereas it would be underestimated with ig-
noring the water volume still present from the previous overtopping wave. Though the
scatter of C t r weakens the performance of the proposed overtopping wave load formula,
it provides a reasoning to consider the effect of the wall on the flow load. Increased scat-
ter shows up when B /L is larger than 0.15. It is undoubtedly caused by the variation
in the overtopping wave depth measurement when the reflected flow collides with the
successive flow near the position of wave gauge.
When estimating the initial overtopping wave depth by using local wave conditions,
an empirical coefficient α2 = 0.77 was used for both dike slopes of cot β = 3 and 6. Ya-
3.6. C ONCLUSIONS 55

mamoto and Horikawa (1992) proposed a theoretical equation for α2 , and verified it by
some experiments and field measurements. When comparing the α2 value obtained in
the present study with that of Yamamoto and Horikawa (1992), α2 = 0.77 agrees well
with Yamamoto’s theoretical value for cot β = 3, but overestimates his theoretical value
for cot β = 6. But the theoretical value of α2 underestimates their measurement when
the dike slope is gentle (cot β = 6 − 10). Yamamoto’s measured data points lie scattered
above the theoretical curve, but have an upper limit around 0.8, which is close to 0.77
for cot β = 6 in this study. Note that the obtained α2 = 0.77 currently is only compared
with what is calculated or measured within a similar range of wave steepness. There-
fore, for estimating the initial depth of overtopping wave induced by a broken wave, Ya- 3
mamoto and Horikawa’s (1992) method is recommended (Martin et al., 1999), and with
cot β = 3 − 6, α2 = 0.77 is suitable for an impermeable dike slope. The authors would like
to point out that Eq. (4) and (7) of Schüttrumpf and Oumeraci (2005) also work well for
the prediction of d A0 in this study, but Eq. (3.15) is relatively simple.
When comparing to a hydraulic bore generated by a dam-break approach, there is
no well-defined equivalent depth (a 0 ) for the wave overtopping situation to describe the
flow characteristics (Ryu et al., 2007). Only limited research work has been reported to
make the connection between these two types of flow/wave. Ryu et al. (2007) found
that dam-break flow theory can be used to predict the overtopping wave field on a ship
deck in deep water, and pointed out that the importance for the application is to deter-
mine the a 0 for the overtopping wave. However, their a 0 was determined empirically by
one wave height in deep water condition, which doesn’t work for the current situation.
The authors think similarity exists between the dam-break wave and overtopping wave
within the decay zone (Z one d e ) in Fig. 3.2. The proposed empirical formulas in this pa-
per aim to give a solution for overtopping wave using wave conditions at the dike toe.
Therefore, the proposed empirical formula can’t be directly used for the hydraulic bore,
for example just assuming a 0 = d A0 .

3.6. C ONCLUSIONS
In this chapter, overtopping momentum flux was used to develop an empirical formula
to predict overtopping wave load. The proposed empirical formula Eq. (3.19) was veri-
fied by experimental data, demonstrating good performance. A dike slope function f (β),
a correction coefficient for obstructed flow depth C t r , and an initial overtopping wave
depth coefficient α2 were determined by experiments. These empirical coefficients al-
lowed for an interpretation of the overtopping process of a broken wave from dike toe
up to the front of wall on the dike. In this chapter, the work only focused on a single
overtopping wave in a series of regular waves. In the following chapter, more work were
carried out by other types of waves (e.g., random waves) to investigation the influence
of the interaction processes within a wave group on the overtopping wave load. These
would give an insight the occurrence of extreme overtopping events and impact loads,
improve the formulation for C t r with a wider parameter range or statistical values, and
provide statistical tools for the extreme load prediction.