Beruflich Dokumente
Kultur Dokumente
Sander Franco
Sander Franco
The author gives permission to make this master dissertation available for consultation and to copy
parts of this master dissertation for personal use. In the case of any other use, the copyright terms
have to be respected, in particular with regard to the obligation to state expressly the source when
quoting results from this master dissertation.
Sander Franco
January 18, 2016
Abstract
Hydraulic stability of rubble mound breakwaters in breaking wave conditions: a comparative study
of existing prediction formulas
Summary
Nowadays most breakwaters are built in shallow water. To enlarge the experimental basis of such a
conditions, some tests are carried out in the 30mx1.2mx1.2m wave flume in the ‘Laboratory of Ports
and Coasts’ of the ‘Polytechnic University of Valencia’. Each wave series attacking the rubble mound
breakwater scale model in shallow water consists of 1000 irregular waves with a fixed Iribarren number
in a fixed water depth. After each wave series, the damage is measured. When total destruction occurs,
the scale model is rebuilt and a new test series is initiated with different Iribarren number and/or in a
different water depth. The measured damage is compared with the damage predicted by formulas
based on the Shore Protection Manual (formula by Van der Meer (1988) and Medina (1994)) on the
one hand and predicted by 3 sets of formulas based on Van der Meer (1998) and Van Gent (2004) on
the other. Based on this comparison, some improvements to the original formulas are presented and
verified by a repetition test. Besides this quantitative analysis of the damage, also a qualitative analysis
is performed obtaining the different damage criteria.
Keywords
Rubble mound breakwater, damage, prediction formula, breaking conditions, irregular wave
Hydraulic stability of rubble mound
breakwaters in breaking wave conditions:
a comparative study of existing
prediction formulas
Sander Franco
Abstract-- Nowadays most breakwaters are The scope of this master thesis is to modify and
built in shallow water. To enlarge the experimental improve existing damage prediction formulas for
basis of such a conditions, some tests are carried rubble mound breakwaters in breaking wave
out in the 30mx1.2mx1.2m wave flume in the conditions based on the comparison between the
‘Laboratory of Ports and Coasts’ of the measured and predicted damage.
‘Polytechnic University of Valencia’. Each wave Two different approaches will be used. A first
series attacking the rubble mound breakwater approach is based on the ‘Shore Protection Manual’
scale model in shallow water consists of 1000 (SPM) [1] with starting point the Hudson formula [2],
irregular waves with a fixed Iribarren number in another approach is based on the formulas developed
a fixed water depth. After each wave series, the by Van der Meer (VDM) [3] and Van Gent (VG) [4].
damage is measured. When total destruction
occurs, the scale model is rebuilt and a new test EXISTING STABILITY FORMULAS
series is initiated with different Iribarren number
SPM-approach
and/or in a different water depth. The measured
damage is compared with the damage predicted by Both Van der Meer [3] and Medina [5] developed
formulas based on the Shore Protection Manual a formula to predict the damage of the armour layer of
(formula by Van der Meer (1988) and Medina a rubble mound breakwater under wave attack. The
(1994)) on the one hand and predicted by 3 sets of starting point of these formulas is the Hudson formula
formulas based on Van der Meer (1998) and Van [2]:
Gent (2004) on the other. Based on this 𝜌𝑟 𝑔𝐻 3
𝑊50 = (1)
comparison, some improvements to the original 𝐾𝐷 ∆3 cot 𝛼
formulas are presented and verified by a repetition 𝑊50 is the medium weight of an armour stone, 𝜌𝑟 the
test. Besides this quantitative analysis of the apparent rock density, 𝑔 the gravitational
damage, also a qualitative analysis is performed acceleration, 𝐻 the wave height, 𝐾𝐷 the stability
obtaining the different damage criteria. coefficient, ∆= 𝜌𝑠 /𝜌𝑤 − 1 the relative buoyant
density and 𝛼 the structure slope angle. Values for the
Keywords-- Rubble mound breakwater, stability coefficient 𝐾𝐷 can be found in the SPM [1].
damage, prediction formula, breaking conditions, Introducing the assumption stated by the SPM [1] that
irregular waves the highest of 10% of wave heights 𝐻1/10 = 1.27𝐻𝑠
should be used as the design wave height and by
INTRODUCTION introducing the stability parameter 𝑁𝑠 = 𝐻𝑠 /∆𝐷𝑛50
with 𝐷𝑛50 the nominal diameter of the armour stone
Most breakwaters nowadays are built in shallow
water, but only little research has been done to and 𝐻𝑠 the significant wave height, this gives:
improve the knowledge of predicting the damage of 𝐻𝑠 (𝐾𝐷 𝑐𝑜𝑡𝛼)1/3
= (2)
the armour layer caused by wave attack. One of the ∆𝐷𝑛50 1.27
reasons of this lack of research is the fact that such a
research is a challenging issue due to the uncertainties This formula only corresponds with a damage level of
in the different irregular wave parameters in shallow 0-5% (no damage). To overcome this limitation, table
water. 7.9 of the SPM [1] was used by Van der Meer [3] and
Medina [5] and by applying regression analysis,
following formulas were developed:
Van der Meer: It’s recommended to use plunging conditions for
𝐻𝑠 0.15 cot 𝛼 ≥ 4 irrespective of whether 𝜉𝑚 is smaller or
= 0.7(𝐾𝐷 𝑐𝑜𝑡𝛼)1/3 𝑆𝑆𝑃𝑀 (3)
∆𝐷𝑛50 larger than 𝜉𝑐𝑟𝑖𝑡 .
Medina:
𝐻𝑚0 For shallow water (ℎ < 3𝐻𝑠,𝑡𝑜𝑒 ), Van der Meer [3]
0.2
= 1.15 (cot 𝛼)1/3 𝑆𝑆𝑃𝑀 (4) recommended to use 𝐻2% instead of 𝐻𝑠 . With the
∆𝐷𝑛50
know relation 𝐻2% /𝐻𝑠 = 1.4, the formulas become:
𝑆 is the damage parameter in which the ‘SPM’ stands
for the fact that table 7.9 from the SPM [1] was used Plunging waves (𝜉𝑚 < 𝜉𝑐𝑟𝑖𝑡 ):
to develop the formulas. 𝐻𝑚0 is the significant wave 𝐻𝑠 𝑆𝑑 0.2 𝐻𝑠 −0.5
= 𝑐𝑝𝑙 𝑃0.18 ( ) 𝜉 (8)
height obtained from spectral analysis. ∆𝐷𝑛50 √𝑁 𝐻2% 𝑚
Surging waves (𝜉𝑚 > 𝜉𝑐𝑟𝑖𝑡 ):
VDM-approach
𝐻𝑠 𝑆𝑑 0.2 𝐻𝑠
The other approach used to predict the damage of = 𝑐𝑠 𝑃 −0.13 ( ) 𝑃
√cot 𝛼 𝜉𝑚 (9)
∆𝐷𝑛50 √𝑁 𝐻2%
the armour layer under wave attack is based on Van
der Meer [3]. Van der Meer developed some formulas, Based on his own test, Van Gent [4] modified
based on tests in a wave flume, taking into account the formulas (8) and (9) by using the spectral mean energy
number of waves 𝑁, the permeability of the wave period 𝑇𝑚−1,0 instead of the mean wave period
breakwater 𝑃 and the surf similarity parameter 𝜉𝑚 = 𝑇𝑚 for the calculation of the surf similarity parameter
tan 𝛼 /√2𝜋𝐻𝑠 /(𝑔𝑇𝑚2 ) with 𝑇𝑚 the mean wave period. 𝜉𝑠−1,0 . Also the coefficients 𝑐𝑝𝑙 and 𝑐𝑠 are modified,
For deep water (ℎ > 3𝐻𝑠,𝑡𝑜𝑒 ), these formulas are: these equal respectively 8.4 and 1.3. These formulas
are valid for both deep and shallow water:
Plunging waves (𝜉𝑚 < 𝜉𝑐𝑟𝑖𝑡 ):
𝐻𝑠 𝑆𝑑 0.2 −0.5 Plunging waves (𝜉𝑚 < 𝜉𝑐𝑟𝑖𝑡 ):
= 𝑐𝑝𝑙 𝑃0.18 ( ) 𝜉𝑚 (5)
∆𝐷𝑛50 √𝑁 𝐻𝑠 𝑆𝑑 0.2 𝐻𝑠
Surging waves (𝜉𝑚 > 𝜉𝑐𝑟𝑖𝑡 ): = 𝑐𝑝𝑙 𝑃0.18 ( ) 𝜉𝑠−1,0 −0.5 (10)
∆𝐷𝑛50 √𝑁 𝐻2%
𝐻𝑠 𝑆𝑑 0.2
= 𝑐𝑠 𝑃 −0.13 ( ) √cot 𝛼 𝜉𝑚 𝑃 (6)
Surging waves (𝜉𝑠−1,0 > 𝜉𝑐𝑟𝑖𝑡 ):
∆𝐷𝑛50 √𝑁
𝐻𝑠 𝑆𝑑 0.2 𝐻𝑠 𝑃
The values for respectively 𝑐𝑝𝑙 and 𝑐𝑠 are 6.2 and 1.0. = 𝑐𝑠 𝑃 −0.13 ( ) √cot 𝛼 𝜉𝑠−1,0 (11)
∆𝐷𝑛50 √𝑁 𝐻2%
The significant wave height 𝐻𝑠 is the incident wave
height at the toe of the structure based on time domain Van Gent also developed a new formula. It’s a more
analysis. The values of the permeability of the simple formula, not taking into account the wave
structure are given in Figure 1. period and the difference between plunging and
surging waves:
𝐻𝑠 1 𝐷𝑛50−𝑐𝑜𝑟𝑒 𝑆𝑑 0.2
= (1 + ) ( ) √cot 𝛼 (12)
∆𝐷𝑛50 0.57 𝐷𝑛50−𝑎𝑟𝑚𝑜𝑢𝑟 √𝑁
Cumulative damage
All formulas according to the VDM-approach are
limited to single storm events. To take into account
subsequent storm events, Van der Meer developed a
method which make directly use of the formulas of the
Figure 1: Values of permeability for different structures
VDM-approach based on the equivalence hypothesis
The distinction between plunging and surging waves (Figure 2) [6].
is based on the critical breaker parameter 𝜉𝑐𝑟𝑖𝑡 :
The procedure to assess the cumulative damage is as
1 follows:
𝑐𝑝𝑙 0.31 𝑃+0.5
𝜉𝑐𝑟𝑖𝑡 = [ 𝑃 √tan 𝛼] (7)
𝑐𝑠
-Calculate the damage 𝑆𝑑1 for the first wave Realized experiments
condition Every test series is initiated with a wave series of
-Calculate for the 2nd wave conditions how many 1000 waves with a wave height that doesn’t produce
waves would be required to give the same damage as breaking waves (8𝑐𝑚). By increasing the wave height
under the 1st wave conditions. This is denoted by 𝑁1′ . with 1 cm every step and by keeping the Iribarren
-Add this number of waves, 𝑁1′ , to the number of number constant, more and more waves are breaking
waves of the second wave conditions: 𝑁𝑡 = 𝑁2 + 𝑁1′ until destruction of the model is observed. Because
-Calculate the damage under the 2nd wave conditions the Iribarren number is kept constant in one test series,
𝑆𝑑2𝑡 with this increased number of waves 𝑁𝑡 . a corresponding peak period 𝑇𝑝 can be calculated for
-Calculate for the 3th wave condition how many each wave height. The different test series are
waves would be required to give the same damage as performed for different Iribarren numbers (𝐼𝑟 = 3 and
caused by the second wave condition etc. 𝐼𝑟 = 5) and water depths (ℎ𝑚𝑜𝑑𝑒𝑙 = 20, 30 and
40𝑐𝑚). One repetition test is carried out (𝐼𝑟 = 3 and
ℎ𝑚𝑜𝑑𝑒𝑙 = 20𝑐𝑚) to use later as a ‘blind test’ to verify
the improved and new developed formulas.
Data analysis
The wave characteristics (wave height, period,…)
are obtained by SwanOne, a model developed by
TUDelft in MatLab to simulate the evolution of a
wave spectrum starting from deep water to shore. It’s
capable to simulate interactions and transformations
Figure 2: Cumulative damage approach by Van der Meer of waves.
0 hs=30,Ir=5
Comparison VDM formula
0 3 6 9 12 15
15
R=0.78 Measured damage Sd[-]
hs=20,Ir=5
Predicted damage Sd[-]
12 rMSE=0.77
Figure 7: Comparison measured and predicted damage by
hs=30,Ir=5
9 improved formula of VDM
always lower than the measured damage. It can also 12 rMSE=0.59 hs=20,Ir=3
be observed that the lower the Iribarren number and 9 hs=30,Ir=3
the higher the water depth, the bigger the difference
between the measured and the predicted damage. So 6 hs=30,Ir=5
by introducing the water depth at the toe of the hs=40,Ir=3
3
structure ℎ𝑡𝑜𝑒 and the peak period 𝑇𝑝 (which
hs=20,Ir=5
influences the Iribarren number or surf similarity 0
parameter), the prediction is improved, as can be seen 0 3 6 9 12 15
Measured damage Sd[-]
on Figure 7. Also the 90% interval is drawn. For the
VDM formula modified by VG the same conclusions Figure 8: Comparison measured and predicted damage by
can be made. The 𝑟𝑀𝑆𝐸 decreases from respectively formula of VG
0.77 to 0.09 for the improved VDM formula and from
formulas should be verified and tested with other tests
Comparison improved VG formula
15 and more repetition tests (also in other laboratories) as
R=0.94
well as the qualitative analysis. These repetition tests
Predicted damage Sd[-]
rMSE=0.12 hs=20,Ir=5
12
are necessary because damage is a very sensitive
hs=40,Ir=3
9 parameter. The difference in measured damage for 2
hs=30,Ir=5 equal tests can be 30% [11].
6
REFERENCES .......................................................................................................................................... 54
1
Chapter 2: Literature study
In this chapter, an overview of the existing literature about the stability of rubble mound
breakwaters in breaking conditions will be given.
First of all, an important process that influences all experiments will be discussed, i.e. wave
reflection. Also some methods to overcome this problem will be given.
Next the phenomenon of wave breaking will be discussed. The different breaker types and the
methods to calculate the breaking wave height in both regular and irregular conditions will be
defined.
Further also the damage that waves cause on the armour layer of a breakwater is treated by
explaining the different failure modes that can damage a breakwater. Also the qualitative and
quantitative analyses of armour damage will be reviewed.
Finally the heart of the matter will be discussed, i.e. the different stability formulas. A historical
overview is given and the most important formulas will be discussed.
1. Wave reflection
a. Reflection coefficient
When waves attack a structure, 3 different processes will take place:
· Dissipation of energy on the permeable medium and transmission of a part of the
energy through the structure to the other side of the structure
· Breaking of the waves over the slope of the structure
· Reflection of the waves on the structure
The breaking of the waves will be introduced in calculations by adjusting coefficients or taking
into account the maximum breaking wave height. Also the permeability will be taken into
account by different coefficients in empirical formulas.
The amount of the wave energy that will be reflected on a sloping structure depends on the
slope, permeability and roughness of the structure. Also the wave steepness ܪȀ ܮand angle of
wave attack influence the reflected wave energy.
A measure to take these influences into account is the reflection coefficient which is equal to
the ratio of the reflected wave height and the incident wave height. Because of the proportional
relationship between the wave energy and the square of the wave height:
(1)
ܪ ܧ
ܴ ܥൌ ൌඨ
ܪ ܧ
The reflected, incident and total wave height and energy are related as follows:
ଶ
ܧ௧௧ ൌ ܧ ܧ ൌ ܪ௧௧ ൌ ܪଶ ܪଶ (2)
This gives following formulas for the incident and reflected wave height:
2
ͳ (3)
ܪ ൌ ܪ௧௧
ξͳ ܴ ܥଶ
ܴܥ (4)
ܪ ൌ ܪ௧௧
ξͳ ܴ ܥଶ
b. Separation methods
The problem is that this reflection coefficient is unknown. To obtain the incident or reflected
wave height, some methods are developed. Most of these methods are based on the linear wave
theory: it’s assumed that an irregular wave train can be represented as a superposition of a finite
number of regular waves of different amplitude, emphasis and frequency (Mansard and Funke,
1980). These methods will be discussed briefly and only the characteristics of these methods will
be given. An extensive discussion would lead us too far. If more information is needed, the
references can be consulted.
The classic method to separate the incident waves and reflected used in most laboratories is the
‘Two point method’ by Goda and Suzuki (1976) generalized using 3 sensors as the ‘Three least
squares method’ by Mansard and Funke (1980) and Gaillard et al. (1980). Each lab can have his
own variations and adaptations on these methods.
The method of Goda and Suzuki (1976) is based on earlier work of Kajima (1969) and Thornton
and Calhoun (1972). Goda & Suzuki introduce the ‘Fast Fourier transform (FFT)’ in the method
of Thornton & Calhoun to identify the linear wave component in each signal. This two point
method has some assumptions that limit its functionality:
· Linear dispersion: ߱ଶ ൌ ݃݇ ݄݇
· Only stationary waves are considered
· Linear superposition is assumed to calculate the irregular wave components
· No noise: the components of high and low frequency are eliminated previously
· Global estimation: It’s necessary to use the complete registration of the waves to
estimate the incident and reflected wave.
Later, the tow point method of Goda and Suzuki was extended to the Three point method.
As mentioned before, each laboratory has its own variation and adaptation on these methods.
The method that will be discussed more in detail is the one used in the Laboratory of Ports and
Coasts (LPC), i.e. the LASA method
c. LASA method
The LASA method (Local Approximation using Simulated Annealing) is a method developed by
Medina (2001) and improved by Figueres and Medina (2004) to separate reflected and incident
waves. The method is intended to overcome some of the limitations of previous methods, i.e.
the limitations of linearity and stationarity. To do so, the LASA method (Medina, 2001) is based
on a local approximation model considering linear and Stokes II non-linear components and uses
simulated annealing to calculate and optimize the model parameters of the local approximation
model. The main characteristics of this method are the possibility to take into account the data
of n sensors (n≥2), the analysis of both non-stationary regular and irregular waves and the
3
discretization of the wave analysis. The LASA method has been verified with the two point
method of Goda and Suzuki (1976), Kimura (1985) and others. It became clear that the LASA
method is very robust and consistent method in numerical and physical test both for regular as
irregular waves.
The general process of the LASA method to realize the separation of incident waves and
reflected waves can be divided in three steps:
· Elimination of the noise
· Establishment of the frames for the estimation of the central points
· Definition of a local approximation model
Figueres and Medina (2004) optimized the original LASA method, based on Stokes II
components, and developed ‘LASA-V’, a method that is based on a local approximation model
considering non-linear Stokes-V components. The LASA-V method can be used for waves with a
high steepness. This model allows the analysis of tests with non-linear and non-stationary
waves.
4
2. Breaking waves
a. General
When waves arrive in the surf zone, they start to break. Simply said, this is because when a wave
approaches a beach, its length ܮstarts to decrease and the wave height ܪincreases.
Galvin (1968) classified wave breaking in four categories: Spilling, plunging, collapsing and
surging breakers (Figure 1).
Spilling breakers produce a foamy water surface because of an unstable wave crest. Spilling
breakers are also characterized by their symmetrical wave contours. This type of breaker is
typical for very gentle beach slopes.
Plunging waves produce a high splash, coming from the crest that curls over the shoreward face
of the wave. The wave front is first very vertical, starts to curl and finally falls. A lot of energy is
dissipated during this process. This kind of breaking is observed on gentle to intermediate beach
slopes.
In very steep beaches, surging waves occur: the wave will not break. The front of the wave
arrives on the beach with minor breaking. The wave goes up and down on the slope only forming
a little bit of foamy water.
Collapsing waves are classified somewhere between surging and plunging waves. The crest is
not breaking, but the lower part of the shoreward face steepens up and falls. An irregular
turbulent water face is created.
The classification of the different breaking waves is based on the surf similarity parameter or
Iribarren number (Iribarren and Nogales, 1949). This number is proportional to the tangent of
the slope of the beach and inversely proportional to the root of the wave steepness:
ߚ ߚ (5)
ߦ ൌ ݎܫൌ ൌ
ܪ ʹߨܪ
ට ට
ܮ ݃ܶ ଶ
5
The values for the surf similarity parameter for the different breaker types (Battjes, 1974) are
given in Table 1.
Table 1: Values surf similarity parameter for different breaker types
Surging/collapsing ߦ ͵Ǥ͵
Plunging ͲǤͷ ൏ ߦ ൏ ͵Ǥ͵
Spilling ߦ ൏ ͲǤͷ
When a wave is moving into shoaling water, the limiting steepness which it can attain, will
decrease (CERC, 1984). Miche (1944) observed that the limiting wave steepness for waves in
depths less than ܮ Ȁʹ equal is to ͲǤͳͶʹ ሺʹߨ݄Ȁܮሻ, with ݄ the water depth.
A wave that moves from deep water to shoaling water will move towards the shore until it
breaks. The wave height at breaking is commonly defined by the breaker (depth) index ߛ
defined as the maximal wave height to depth ratio ܪȀ݄, with the subscript ܾ standing for the
breaking point.
ܪ ܪ ܪ (6)
ߛ ൌ ൬ ൰ ൌ
ܮ ݄ ௫ ݄
However, the latter induces a greater uncertainty in the prediction of ܪ according to present
authors (Camenen and Larson, 2007).
6
Regular waves
The first value of the breaker depth index for regular waves was estimated by McCowan (1891)
and equal to 0.78 for a solitary wave traveling over a horizontal bottom.
Munk (1949) defined the breaker height index ܪ Ȁܪ of a solitary wave as:
ͳ (8)
ȳ ൌ
͵Ǥ͵ሺܪ Ȁܮ ሻଵȀଷ
Subsequent observations and investigations by Iversen (1952), Galvin (1968), Goda (1970),
Weggel (1972), and others have shown that ȳ and ߛ depend on incident wave steepness and
beach slope. Thus for regular waves and uniform beach slope, next formulas are proposed:
Goda (1970)
ܮ ݄ (10)
ߛ ൌ ͲǤͳ ൬ͳ െ െͳǤͷߨ ൫ͳ ͳͷ ସȀଷ ߚ൯൨൰
݄ ܮ
Weggel (1972)
ܪ (11)
ߛ ൌ ܾ െ ܽ
݃ܶ;
Battjes (1974)
ߛ ൌ ͲǤͻ͵݉Ǥଵହହ ିܮǤଵଷ
(14)
ͳǤͳʹ (15)
ߛ ൌ െ ͷሾͳ ሺെͶ͵ ߚሻሿܮ
ͳ ሺെͲ ߚሻ
7
Rattanapitikon and Shibayama (2000)
Those formula were compared with a compiled data set (Camenen and Larson, 2007), but there
was not one best formula. The best results for the existing formulas and use of all data were
given by Battjes (1974), Goda (1970) and Ostendorf and Madsen (1979). For steep beaches, the
Weggel (1972) formula gave the best prediction. Also some general remarks were made. Weggel
(1972) and Singamsetti and Wind (1980) overestimate the breaker depth index and produce
considerable dispersion of the results while Smith and Kraus (1990) underestimate the breaker
depth index. A disadvantage of the Ostendorf and Madsen (1979) and Battjes (1974) formulas
are that no value can be calculated for a beach with a slope equal to zero.
A semi-empirical relationship for the breaker height index is derived from linear wave theory by
Gaughan and Komar (1974) and equals:
ଵ (17)
ି
ܪᇱ ହ
ȳ ൌ ͲǤͷ ቆ ቇ
ܮ
Finally another criterion is given by Rattanapitikon and Shibayama (2000) and Rattanapitikon et
al. (2003), with ܮ the wavelength when the wave breaks calculated using the linear wave
theory:
ܪ ଶ
ܪ Ǥଷହ (18)
ൌ ሺെͳǤͶͲ ߚ ͲǤͷ ߚ ͲǤʹ͵ሻ ൬ ൰
ܮ ܮ
Both Goda (1970) and Weggel (1972) included their formulas in a graph. These are given in
Figure 3 and Figure 4.
Figure 3: Graph showing formula by Goda (1970) Figure 4: Graph showing formula by Weggel (1972)
8
To calculate the breaking wave height ܪ from the breaker height index ȳ , The deep water
wave height ܪ has to be calculated if this is not known. This can be done using the shoaling
formula (Goda, 2000):
ିଵȀଶ (19)
ܪ ʹ݄݇
ܭ௦ ൌ ൌ ൬ͳ ൰ ݄݇൨
ܪ ʹ݄݇
Irregular waves
For irregular waves, the breaking point as well as the breaking wave height can start over a wide
zone, in contrast to the case of regular waves (Goda, 2000). In the zone where more or less all
waves are breaking (saturated breaking zone), the root-mean-square breaking wave height and
the zero-moment wave height depend on the local depth ݄ (Thornton and Guza, 1983):
ܪ௦ǡ ൌ ͲǤͶʹ݄ (20)
CIRIA (2007) states that the typical values for the breaker depth index are 0.5 to 0.6. These values
mainly depend on the Iribarren number ߦ , and they can reach 1.5 for individual waves. Data
from different authors are shown in Figure 5.
The critical wave height ܪ over which all the waves are broken can be found by applying an
energy flux balance and is given by (Battjes and Stive, 1985):
ͲǤͺͺ ߛ (22)
ܪ ൌ ቀ ݄݇ቁ
݇ ͲǤͺͺ
9
This formula was later modified by Nairn (1990):
ߛ ൌ ͲǤ͵ͻ ͲǤͷ ሺ͵͵ܮ ሻ (23)
A linear relationship between the product of the local wave number and the water depth was
found by Ruessink et al. (2003):
ܮ
ିଵǤହగ ್ ൫ଵାଵଵ ୲ୟ୬రȀయ ఉ ൯ (25)
ߛ ൌ ͲǤͳ ቆͳ െ ݁ బ ቇ
݄
The deep water wavelength ܮ is calculated using the period ܶଵȀଷ obtained from time-domain
analysis. This formula was verified starting with regular breaking waves for 6 different slopes
(including 0%).
As final remark, it can be stated that, as said before, the breaking point can’t be defined clearly.
Therefore, Goda (2000) proposes to use the peak value of the significant wave height ܪଵȀଷ within
the surf zone as an alternative to the breaker height. These values are depicted on Figure 6.
Figure 6: Index curves for the maximum value of the significant wave height
10
3. Damage
a. Failure modes breakwater
To be able to talk about the stability of a mound breakwater, it’s necessary to consider both the
individual stability of an armour unit as the stability of all units together. The loss of stability can
have different reasons which should be understood. Bruun (1978) is one of the authors who has
worked with it. He stated 11 different principal failure modes. They are summarized as follows
(Vanhoutte, 2009):
I. Loss of armour units from the principal armour layer (increasing porosity)
II. Rocking of the armour units; breaking occurs due to fatigue
III. Damage of the inner slope by wave overtopping
IV. Sliding of the armour layer due to a lack of friction with the layer below
V. Lack of compactness in the underlying layers, causing transmission of energy to
the interior of the breakwater; this might lift the cap and the interior layers.
VI. Undermining of the crone wall
VII. Breaking of the armour units caused by impact, simply by exceeding its
structural resistance or by slamming into other units
VIII. Settlement or collapsing of the subsoil
IX. Erosion of the breakwater toe or the breakwater interior
X. Loss of the mechanical characteristics of the materials
XI. Construction errors
These failure modes are depicted in Figure 7.
All these 11 different failure modes can be classified in 5 groups (Gomez-Martin, 2002):
· Unit stability: refers to the capacity of each armour unit to resist movement subjected
to the wave action. (I,II,III)
· Global stability: it’s the stability of the complete breakwater or more specific the
complete armour layer. (IV,V,VI)
· Structural stability: refers to the capacity of each unit to resist (without breaking) the
tensions caused by transport, construction, wave action, used granular and movements
caused by currents. (II, VII)
· Geotechnical stability: stability of the underground. It includes the carrying capacity and
the sensitivity to erosion of the breakwater toe.(VIII, IX)
· Construction errors (X,XI)
11
The global and unit stability are the stability that will be considered here. Together they can be
called the hydraulic stability. Burcharth (1992) also enumerate the different failure modes of a
breakwater (Figure 8).
In this thesis, most attention will go to the hydraulic stability of the armour units. This is also the
main mode of failure. It’s classified in the failure group of ‘Unit Stability’ according to Gomez-
Martin (2002). This failure mode can be caused by 2 different reasons: the simple extraction of
the units under wave action and the settlement caused by the heterogeneous packing. The
heterogeneous packing (HeP) failure mode is a failure mode that is significant in the case of
regular armour units (Gomez-Martin, 2006). The HeP failure mechanism reduces the packing
density of the armour layer near the SWL without extracting elements, generating zones with
low porosity and corresponding zones with high porosity. The impact of the HeP failure mode
depends on:
I. Type of armour unit
II. Difference between the initial porosity and the minimum porosity
III. Slope of the armour layer
IV. Friction coefficient between the armour layer and the secondary layer
b. Armour damage
There are two ways to quantify the damage: the qualitative and the quantitative way.
Qualitative analysis
To do a qualitative analysis of the damage, several stages of damage should been distinguished.
Losada et al. (1986) created three hydrodynamic criteria: incipient damage, Iribarren’s damage
and destruction. Vidal et al. (1991) further developed these criteria by adding a fourth criteria:
Initiation of destruction. Hence the four criteria are:
· Initiation of damage (IDa): certain number of units are displaced from their original
position to a new one at a distance equal to or larger than a unit length.
12
· Initiation of Iribarren damage (IIDa): wave action may extract armour units placed on
the lower armour layer. This can be defined as the moment when one stone and his
surrounding stones of the 2nd layer are visible.
· Initiation of destruction (IDe): small number of units (two or three) in the lower armour
layer are forced out.
· Destruction (De): more pieces of the secondary layer are removed and the filter layer is
visible.
A disadvantage of these criteria, is the fact that HeP is not taken into account, because only the
units extracted are considered.
Quantitative analysis
The damage in the quantitative analysis is measured by counting the number of displaced units
or by measuring the eroded surface profile of the armour slope (USACE, 2002). If the damage is
measured by counting the displaced units, which is mostly done in case of (complex structures
of) concrete armour units, the damage can be given as a percentage displaced units within a
reference area:
݊ݏݐ݅݊ݑ݈݀݁ܿܽݏ݂݅݀ݎܾ݁݉ݑ (26)
ܰௗ ൌ
ܽ݁ݎܽ݁ܿ݊݁ݎ݂݁݁ݎ݄݊݅ݐ݅ݓݏݐ݅݊ݑ݂ݎܾ݁݉ݑ݈݊ܽݐݐ
Another way to calculate the damage is making use of the eroded area. This is mostly done in
case of rock armour. One of the first who used this way of calculating were Iribarren (1938) and
Hudson (1959). Hudson defined the damage as the percent erosion of original volume:
݈݂ܽ݁݅ݎ݉ݎ݂ܽ݁ݎܽ݀݁݀ݎ݁݁݃ܽݎ݁ݒ (28)
ܦൌ
݈݂ܽ݁݅ݎ݈ܽ݊݅݃݅ݎ݁݃ܽݎ݁ݒ݂ܽܽ݁ݎ
13
mass of the stone. To improve this formula, Broderick (1983) deleted the bulk density and
defined the damage level parameter ܵௗ as follows :
ܣ (30)
ܵௗ ൌ ଶȀଷ
ܹ
ቀ ߩହ ቁ
The eroded area is depicted in Figure 9. It takes into account both settlement and displacement.
The eroded area can be seen as the number of squares with a side ܦହ that fits into the erosion
area. Another physical description of the damage of ܵௗ is the number of cubic stones with a side
of ܦହ eroded within a ܦହ-wide strip of the structure. The actual number of stones eroded
within this strip can be more or less than ܵௗ , depending on the porosity, the grading of the
armour rocks and the shape of the rocks. Generally the actual number of rocks eroded in a ܦହ-
wide strip is equal to 0.7 to 1 times the damage ܵௗ .(Van der Meer, 1998)
The slope angle of the structure has a big influence on the limits of ܵௗ . These limit values are
characterized as follows:
· Start of damage/initial damage: corresponding to no damage in the Hudson formula (see
later)
· Intermediate damage
· Failure, when the filter layer is visible
14
For design purposes of a double layer armour stone breakwater, these values are given in Table
2. For S-values higher than 15-20, the deformation of the structure results in an S-shaped profile
(Van der Meer, 1998).
Table 2: Design values of ܵௗ for a two-diameter thick armour layer
Start of Intermediate Failure (under
Slope
damage damage layer visible)
1:1.5 2 3-5 8
1:2 2 4-6 8
1:3 2 6-9 12
1:4 3 8-12 17
1:6 3 8-12 17
There are different ways to measure the eroded area. It can be measured by a surface profiler
(mechanic/laser profiler), by computing the planar eroded area on the outer layer of the armour,
using a digital image processing technique or by counting the removed armour stones settled
over the original armour layers (Vidal et al., 2006) . If the latter is used, the eroded area is given
by:
ܰௗ ܦହ Ϳ (32)
ܣ ൌ
ሺͳ െ ሻܴ
With ܰௗ the number of eroded stones, the porosity of the settled stones and ܴ the width of
the eroded area. This method is called ‘the visual counting method’.
The disadvantage of all formulas above is that they do not take into account the heterogeneous
packing. Therefore, a new method was necessary that takes into account the changes in
porosity. This method is called the new Virtual Net Method (Gómez-Martín and Medina, 2006).
In this method, the armour layers are divided into strips with each a width ܽ of ݊ times the
equivalent cube size ܦହ and a length ܾ. The number of armour units in every strip ܰ is
counted and with this number, the porosity of every strip after wave attack is calculated using
the formula below.
ଶ
ܰ ܦହ (33)
ൌ ͳ െ
ሺܾܽሻ
Next, the dimensionless damage in each strip ܦ can be calculated taking into account the initial
porosity :
ͳ െ (34)
ܦ ൌ ݊ ൬ͳ െ ൰
ͳ െ
By summing up the different damages over the different strips, the equivalent dimensionless
armour damage ܦ could be obtained:
15
4. Stability formulae
Until 1933 no methods to calculate a rubble mound breakwater existed. The breakwaters were
build using the experience obtained from the construction of older ones. Obviously, this
qualitative knowledge wasn’t sufficient for the construction of breakwaters. Also, the
complexity of the phenomena that were involved (wave characteristics, wave behavior,…)
impeded the development of this qualitative knowledge.
The first formula for the calculation of rubble mound breakwaters was formulated by Castro
(1933). Castro showed that the wave forces are the reason of the destruction of the breakwaters
and that the waves push the rock over the breakwater. However, this almost never happens
Iribarren (1938) published a new formula for the weight of the rocks resisting a certain wave
height for the principal layer of the breakwater. This formula would be the starting point for the
later developed formula by Hudson.
From 1949, a big development in the knowledge and of the formulas started. Also the
phenomena related with the water flow over the slope of the breakwater were studied. The
discussion of all the different formulas and modified formulas would lead us so far, so only the
most important and the formulas that have most to do with this master thesis will be discussed.
a. Hudson
One of the most known and used stability formula is Hudson`s formula (Hudson, 1959). It’s based
on model tests with regular waves on non-overtopped rock structures with a permeable core
and based on the pioneering work of Iribarren (1938) and Hudson and Moore (1951):
Iribarren (1938)
ߩ ݃ ܪܭଷ (36)
ܹହ ൌ
οଷ ሺ
ߙ െ ߙሻଷ
16
Hudson (1959)
ߩ ݃ ܪଷ (38)
ܹହ ൌ
ܭ οଷ
ߙ
17
This gives:
The formula of Hudson has its limitations. First, it can only be used for regular waves. Also the
wave period and storm duration are not taken into account. Besides these limitations, Hudson’s
formula can only be used for non-overtopped and permeable structures only. Finally there is no
description of the damage level, because the formula only corresponds with a damage level of
0-5% (no damage).
To overcome this last limitation, higher damage percentages have been determined as a
function of the wave height for several types of armour units. The values for armour stone are
given in Table 3 (CERC, 1984).
The notation ܵௌெ instead of ܵௗ is used in this case, because it’s not clarified in the SPM how
the damage is measured and the conversion from the damage ܦto the damage level parameter
ܵ is only an assumption. Also a slightly different conversion could be made.
Van der Meer (1988) modified equation (40) by using Table 3 (angular armour stone) and
applying regression analysis. This modified formula for the stability number ܰ௦ equals:
ܪ௦ Ǥଵହ (41)
ൌ ͲǤሺܭ ܿߙݐሻଵȀଷ ܵௌெ
οܦହ
According to the Rock Manual (CIRIA, 2007), the ܭ values for an impermeable and permeable
core for both breaking and non-breaking conditions are respectively 1 and 4, accepting that 5%
of the data will lead to a higher damage level than predicted. ܭ values of respectively 4 and 8
can be used to describe the main trend. This was concluded after comparing data used by Van
der Meer (1988) and Van Gent et al. (2004) with the equation above using different ܭ values.
Medina et al. (1994) proposes another empirical formula to include the data in Table 3 in
equation (40):
18
ܪ Ǥଶ (42)
ൌ ͳǤͳͷ ሺ
ߙሻଵȀଷ ܵௌெ
οܦହ
One of the reasons of the difference between this formula and formula (41) is because Medina
used a different conversion between the damage ܦand the damage level parameter ܵ. But
because the formula is still based on the data from the SPM (CERC, 1984), the notation ܵௌெ is
still used.
b. Van der Meer
Deep-shallow water
While the classification according to wave propagation of water waves (deep water waves,
transitional water waves and shallow water waves) is based on the relative depth criterion ݄Ȁܮ
(respectively ݄Ȁ ܮ ͲǤͷ , ͲǤͲͷ ൏ ݄Ȁ ܮ൏ ͲǤͷ and ݄Ȁ ܮ൏ ͲǤͲͷ) (USACE, 2002), the definition of
shallow water is defined by ݄ ൏ ͵ܪ௦ǡ௧ for the limit of the field of application of the Van der
Meer formulae (CIRIA, 2007). More recently, Van Gent developed even another classification,
based on the ratio of the significant wave height at the structure to the one observed offshore.
If the ratio is larger than 0.9 or smaller than 0.7, the structure is in respectively deep and very
shallow water (with a considerable amount of wave breaking). If the ratio is between 0.7 and
0.9, the structure is said to be in shallow water conditions and some (limited) wave breaking will
already occur (Van Gent et al., 2004).
Deep water conditions
Besides the modified Hudson formula, Van der Meer (1988) also developed also some formulas,
He based his study on the earlier work of Thompson and and Shuttler (1975). An extensive series
of model tests was conducted at Delft Hydraulics. This series include structures with a wide
range of core/underlayer permeabilities and a wide range of wave conditions.
The Van der Meer (VDM) formula takes, in contrast to the Hudson formula, into account the
effects of storm duration, wave period, the structure’s permeability and a clearly defined
damage level. VDM considers 2 types of breaking waves: plunging and surging waves. The
transition from plunging conditions to surging conditions is given by the critical breaker
parameter and depends on the structure slope ߙ:
ଵ (43)
ܿ ାǤହ
ߦ௧ ൌ ܲǤଷଵ ξ ߙ൨
ܿ௦
19
For surging waves (ߦ ߦ௧ ):
With ܪ௦ Significant wave height [m], ܪଵȀଷ of the incident waves at the
toe of the structure ሾ݉ሿ
ܿ 6.2
ܿ௦ 1.0
P Permeability of the structure [-]
N Number of incident waves at the toe of the structure [-]
ߦ surf similarity parameter based on the mean wave period ܶ
from time domain analysis and significant wave height ܪ௦ ሾെሿ
It is recommended to assume plunging conditions for
ߙ Ͷ irrespective of whether the surf-
similarity parameter ߦ is smaller or larger than ߦ௧ .
The values for the permeability of different structures are given in Figure 10. The lower boundary
is equal to 0.1. This is the case when the armour layer has a thickness of 2 diameters of the
armour unit and the layer under it is impermeable. This is often the case for seawalls and
revetment. The upper boundary is given by a homogeneous structure without filter and core. It
only consists of rock. The permeability is then equal to 0.6.
The considered range of parameters by Van der Meer (1988b) is given in Table 4.
20
Table 4: Range of parameters VDM formula
The VDM formulas are related to a single storm event. The maximum number of waves is equal
to 7500 because after this number of waves, the damage reaches more or less an equilibrium
This means that the damage for more than 7500 waves is found by using ܰ ൌ ͷͲͲ. For a
number of waves smaller than 1000, the formulas give a slight overestimation.
The limits of the damage level parameter ܵௗ mainly depend on the slope of the structure
If the Van der Meer formula is used to for design proposes, it’s required to do a sensitivity
analysis for all parameters. Also a sensitivity analysis of the constants should be done. Two
graphs are given as an example of such an analysis (Figure 11). The influence on the changes of
the damage level parameter and permeability on the design wave height for different breaker
parameters is analysed.
Figure 11: Sensitivity analysis for damage level parameter and permeability
21
Shallow water conditions
Some of the tests carried out by Van der Meer were in shallow water. Van der Meer (1988)
recommends to use the wave height exceeded by 2% of the waves ܪଶΨ instead of the significant
wave height ܪ௦ . This because the distribution of the wave heights deviate from the Rayleigh
distribution (due to wave breaking) so the stability of the armour layer in depth limited waters
is better described by the higher characteristic value of the wave height distribution ܪଶΨ than
by ܪ௦ . The known ratio ܪଶΨ Ȁܪ௦ is equal to 1,4 for a Rayleigh distribution, so the deep water
values of the coefficients ܿ and ܿ௦ should be changed to respectively 8.7 and 1.4 for shallow
water. The transition between plunging and surging waves is the same as equation (43) using
the coefficients ܿ and ܿ௦ valid for shallow water.
For plunging waves (ߦ ൏ ߦ௧ ):
ܪ௦ Ǥଵ଼
ܵௗ Ǥଶ ܪ௦ ିǤହ (46)
ൌ ܿ ܲ ൬ ൰ ߦ
οܦହ ξܰ ܪଶΨ
In conclusion, a remark should be made. A safer approach for design purposes is to use formulas
(44) and (45) with ܪ௦ . In that case the truncation of the wave height exceedance curve due to
wave breaking is not taken into account (Van der Meer, 1988).
The equations (44) and (45) will give the same results as (46) and (47) if the waves are Rayleigh
distributed due to the know Rayleigh-distribution-based ratio ܪଶΨ Ȁܪ௦ that was introduced
before.
To do the correct calculation, the actual ratio ܪଶΨ Ȁܪ௦ should be obtained, which is not always
possible because often only the wave heights based on energy (ܪ௦ is based on the zero-est
moment of the spectrum ݉ ) are obtained. This means that ܪ௦ (actually ܪଵȀଷ) is different from
ܪ and so it’s difficult to find a good estimation for ܪଶΨ. To overcome this issue, a good
approximation for ܪଶΨ based on ܪ is given by Battjes and Groenendijk (2000) who proposed
a composed Weibull-Rayleigh distribution for the waves. Van Gent et al. (2004) concluded that
both wave heights ܪ and ܪଵȀଷ can be used, leading to almost the same accuracy on average.
c. Van der Meer modified by Van Gent
Van Gent et al. (2004) modified, based on his own tests, the Van der Meer formula for shallow
foreshore. First, he uses the spectral mean energy wave period ܶିଵǡ instead of the mean wave
period obtained from time-domain analysis ܶ . Also the coefficients ܿ and ܿ௦ are recalibrated
and have a value of respectively 8.4 and 1.3. Thus the final formulas are:
For plunging waves (ߦ௦ିଵǡ ൏ ߦ௧ ):
22
For surging waves (ߦ௦ିଵǡ ߦ௧ ):
୲ୟ୬ ఈ
With ߦ௦ିଵǡ ൌ
మഏಹೞ
ඨ మ
షభǡబ
The formulas were developed using tests in the ranges given in Table 5 and are valid in both
deep and shallow water
d. Van Gent
Another (more simple) formula was developed by Van Gent et al. (2004), based on a earlier
series of tests with a 1:100 foreshore (Smith et al., 2002) and additional tests with a considerably
steeper foreshore (1:30). The structure slopes were 1:2 and 1:4. Both test series were combined
and analysed to develop the new formula.
In the new formula the permeability is directly related to a structure parameter (mean nominal
diameter of the core material ܦହି ). The influence of the period isn’t taken into account.
On the one hand the wave period influences the damage, but on the other this influence is small
compared to the amount of scatter in the data. For the same reason, there is also no difference
between plunging and surging waves. Besides this, the ratio ܪଶΨ Ȁܪ௦ also influences the damage,
but again, these influence is considered small. The Van Gent formula is given by:
The range of tests used to develop the formula is the same as mentioned above in Table 5. The
formula is valid in both shallow and deep water.
23
e. Cumulative damage
All previous formulas are limited to single storm events. Some studies were performed to take
into account the phenomenon of progressive damage due to subsequent storm event. 2
different methods taking into account subsequent storm events will be discussed: the method
developed by Melby (2001) and the method developed by Van der Meer (2000).
Melby method
The cumulative damage is evaluated by
ହ
ܰ௦ǡ (51)
ܵௗ ሺݐ ሻ ൌ ܵ௧ ሺݐ ሻ ͲǤͲʹͷ
ሺݐ െ ݐ ሻ
ܶǡ
Permeability ܲ <0.4
24
· Calculate the damage ܵௗଵ for the first wave condition
· Calculate for the second wave conditions how many waves would be required to give
the same damage as under the first wave conditions. This is denoted by ܰଵᇱ .
· Add this number of waves, ܰଵᇱ to the number of waves of the second wave conditions:
ܰ௧ ൌ ܰଶ ܰଵᇱ
· Calculate the damage under the second wave conditions ܵௗଶ௧ with this increased
number of waves ܰ௧ Ǥ
· Calculate for the third wave condition how many waves would be required to give the
same damage as caused by the second wave condition etc.
25
Chapter 3: Experimental setup
In this chapter, the test equipment and experimental model will be discussed. Also the realized
experiments and calculations will be explained.
First of all, the wave flume will be described including the wave generating system, the system
that dissipates the energy, the sensors that measure the wave elevations and the visual
equipment.
Next the design of the foreshore and model will be set out together with the properties of the
used quarry stone.
Furthermore, the characteristics of the realized experiments will be given.
Finally this chapter will be concluded by the different steps taken to analyse the data. The
different methods to obtain the wave characteristics will be explained, the porosity
measurement method will be discussed together with the damage calculation method. Also it
will be explained how the original formulas will be improved.
1. Test equipment
All tests are performed in the 2D wave flume in the Laboratory of Ports and Coast of the
Polytechnic University of Valencia(LPC-UPV)(Figure 13). The length of the wave flume is 30
meters, the height is 1.2 meters and the width is 1.2 meters. There is a false bottom of 25 cm
for the circulation of the water. The wave generator is installed at one side of the wave flume.
The system to dissipate the energy of the waves is installed at the other side. The breakwater
model is placed in front of the energy dissipater. A part of the walls of the wave flume is
transparent to give the researchers the possibility to see what happens inside the wave flume.
26
a. Wave generator
As mentioned before, the wave generator is installed at one end of the wave flume. It consists
of and metal plate that is installed vertical and connected to a piston which generates the
translation of the plate. This piston is driven by an electrical servomotor (Figure 14).
The wave generating system is controlled by an Active Wave Absorption Control System
(AWACS). This is a digital control system that absorbs the reflection of the waves and generates
the desired waves. This is needed to take into account the reflected waves from the structure
on the other side of the wave flume. Since incident waves on the model are reflected, they could
re-reflect on the wave maker, which results in an uncontrollable, and undesirable nonlinear
distortion of the desired waves impinging on the test structure, because the wave generator
keeps on having the same movement. So the reflected waves from the breakwater are taken
into account by the AWACS. The principle of the AWACS is as follows. It measures the surface
elevation by 2 wave gauges integrated in the paddle front. This measured surface elevation is in
fact the superposition of the desired wave and the reflected wave. The reflected wave is
identified by the digital recursive filter of the AWACS and absorbed by the wave maker. The
specific type of AWACS installed in LCP-UPV is DHI AWACS2, from Denmark.
The wave making system has 3 options. First of all, it can reproduce previous generated waves.
Secondly, it can produce regular waves (Stokes 1st order). Last it can also generate irregular
waves. The parameters that has to be given in into the system are (depending if regular or
irregular waves will be generated):
· Scale: it’s used to make the conversion from the parameters of the prototype to the
model
· Water depth: water depth in front of the wave paddle.
· Wave height and period
· Spectrum(only for irregular waves)
· ܨ and ܨ௫ (only for irregular waves)
· Skewness (only for irregular waves)
· Duration
27
b. Energy dissipation system
The energy dissipation system is placed at the end of the wave flume, on the opposite side of
the wave generating system. It consists of five groups of three grooved metal frameworks and a
plastic plate which is perforated. The metal frameworks have 3 different porosities: 70%, 50%
and 30%. The frameworks are placed in such a way so the framework with the highest porosity
is the closest to the approaching wave. The first group of 3 metal frameworks with a porosity of
70% is followed by 2 groups with a porosity of 50%. The porosity of the last 2 groups is 30%. The
voids between the 3 frameworks of this last group are filled with quarry stone (Figure 15).
Figure 15: Energy dissipation system and schematically overview of the frameworks
c. Wave measurement
The wave flume is equipped with a series of wave gauges and run-up sensors. Because there is
no run-up during the test carried out for this master thesis, only the wave gauges will be
discussed.
The wave gauges measure the surface elevation. They consist of two vertical electrodes. When
they are submerged in the water, the sensors measure the conductivity of the volume of water
between the electrodes. The conductivity changes as a function of the surface elevation
between the electrodes.
The gauges has to be calibrated every time before the tests to overcome any errors caused by
for instance changes in water level, caused by leaks and evaporation.
The wave gauges are placed in different groups: One group close to the wave paddle to measure
the wave height generated by the wave generating system in deep water (S1, S2, S3 and S4),
one group close to the model to measure the wave height at the toe of the structure (S11 and
S12). Between these 2 groups, there are placed some other sensors along the slope of the beach
(Figure 16).
28
Figure 16: Position of the sensors, dimensions in meter
Mansard and Funke (1980) defined some criteria for the distance between 3 wave gauges, with
݀ଵ and ݀ଶ the distances between the 3 wave gauges:
ܮ (52)
ۓ ݀ଵ ൎ
ۖ ͳͲ
ܮ ܮ
ۖ ൏ ݀ଵ ݀ଶ ൏
͵
݀ ۔݀ ് ܮ
ۖ ଵ ଶ
ͷ
ۖ ͵ܮ
݀ ەଵ ݀ଶ ് ͳͲ
Using these criteria, a distance between the waves gauges is chosen for all the wave periods
that will be used, so it wouldn’t be necessary to change them every time after one test.
d. Cameras
The wave flume is equipped with different hardware for visual registration. Both digital pictures
and digital videos are made. They are not only used to have a good view of them in the office
during the test, but also to take the pictures used to calculate the damage.
29
2. Experimental design
a. Foreshore slope
A foreshore slope is created to make the waves break (Figure 16). The slope of the first one
located 545 cm in front of the wave paddle and with a length of 625cm is 4%. Behind this slope,
another foreshore slope of 2% is constructed. Its length is 1100 cm and the model is built on the
end of this slope. The water level varies according to the test series between 61.7 and 81.7 in
front of the wave paddle with steps of 10 cm. This corresponds with a water level between 20
and 40 cm at the toe of the breakwater model.
b. Model
The used model (scale 1:60) is a rubble mound breakwater model with a slope of 3/2 at the
side exposed to the waves (Figure 17) . The armour layer of the breakwater is composed out of
2 layers of quarry stone with a porosity of approximately 37% without a toe berm. The model
has 3 different parts: The core, a filter layer and an armour layer. The characteristics of the 3
will be briefly discussed.
30
Filter
The size of the stones of the filter used in the model have a nominal diameter of ܦହ ൌ ʹܿ݉.
The distribution for a sample of 20 filter layer stones is given in Figure 19.
Armour layer
The mean nominal diameter of the stones used is equal to ܦହ ൌ ͵Ǥͳͺܿ݉. The distribution of
a sample of 25 stones is given by Figure 20.
The initial porosity of the armour stones has to be approximately 37% (CERC, 1984) before the
test series start. The model with this initial porosity is depicted on Figure 21.
Figure 21: Rubble mound model constructed with an initial porosity of approximately 37%
31
3. Realized experiments
The realized experiments on the rubble mound breakwater model are part of a 2 years long
ongoing project called ‘ESCOLIF’ (Estabilidad hidráulica de los mantos de escollera, cubos y
Cubipodos frente a oleaje limitado por el fundo/ Hydraulic stability of cube, Cubipod and rubble
mound breakwaters in depth limited conditions) which is carried out at the LPC-UPV. The goal of
the project ESCOLIF is to improve the knowledge of the hydraulic stability of single- and double-
layer Cubipod armours (an armour unit developed by LPC-UPV) in depth limiting conditions.
Another goal of ESCOLIF is to increase the experimental basis of double-layer cubes and quarry
stone in depth limiting conditions:
· ͲǤ͵ ൏ ܪ௦ Ȁ݀ ൏ ͲǤͺ
· ͲΨ ൏ ߠ݊ܽݐ൏ ͶΨ
· ͲǤͲʹ ൏ ܪ௦ Ȁܮ ൏ ͲǤͲ
The initial testing conditions (foreshore slope of 0, 2 and 4%) were modified by adding also a
foreshore slope of 10% to the tests because a high influence of the slope of the foreshore was
observed.
The experiments discussed in this master thesis are carried out with 1000 irregular waves with
a JONSWAP spectrum (ߛ ൌ ͵Ǥ͵). The foreshore for all test is equal to 2%. The typical breaker
type under such a foreshore slope is spilling breakers.
Every test series is initiated with a wave series of 1000 waves with a wave height that doesn’t
produce breaking waves (ͺܿ݉). By increasing the wave height with 1 cm every step and by
keeping the Iribarren number constant, more and more waves are breaking until destruction of
the model is observed. Because the Iribarren number is kept constant in one test series, a
corresponding peak period ܶ can be calculated for each wave height. The different test series
are performed for different Iribarren numbers ( ݎܫൌ ͵ and ݎܫൌ ͷ) and in different water depths
(݄ௗ ൌ ʹͲǡ ͵Ͳ and ͶͲܿ݉). One repetition test is carried out ( ݎܫൌ ͵ and ݄ௗ ൌ ʹͲܿ݉) to
use later as a ‘blind test’ to verify the improved and new developed formulas.
An overview of the input data for the different test series is given in Table 7 to Table 11
32
16 1.44 61.70 20 9.6 11.16 37.02 12
17 1.48 61.70 20 10.2 11.50 37.02 12
18 1.53 61.70 20 10.8 11.84 37.02 12
19 1.57 61.70 20 11.4 12.16 37.02 12
ࡴ࢙ ሾࢉሿ ܶ ሾݏሿ ݄ௗௗ ሾܿ݉ሿ ݄ௗ ሾܿ݉ሿ ܪ௦ ሾ݉ሿ ܶ ሾݏሿ ݄ሾ݉ሿ ݄ௗ ሾ݉ሿ
33
Table 10: Test data for Ir=5 and h=30cm
ࡴ࢙ ሾࢉሿ ܶ ሾݏሿ ݄ௗௗ ሾܿ݉ሿ ݄ௗ ሾܿ݉ሿ ܪ௦ ሾ݉ሿ ܶ ሾݏሿ ݄ሾ݉ሿ ݄ௗ ሾ݉ሿ
ࡴ࢙ ሾࢉሿ ܶ ሾݏሿ ݄ௗௗ ሾܿ݉ሿ ݄ௗ ሾܿ݉ሿ ܪ௦ ሾ݉ሿ ܶ ሾݏሿ ݄ሾ݉ሿ ݄ௗ ሾ݉ሿ
34
4. Data analysis
a. Wave analysis
The wave characteristics (wave height, period…) have to be obtained to estimate the damage
using the different prediction formulas. These values can be obtained using different methods.
Their registered values can’t be used directly because these values consist of 2 different
influences, namely the incident wave and the reflected wave by the structure. The
characteristics of the incident wave are the characteristics that will be used to do all calculations.
There are 3 different methods to obtain the characteristics of the incident waves with each its
advantages and disadvantages: SwanOne, measurements by wave gauges in canal without
model and measurements by wave gauges in canal with model. Each one will be discussed.
SwanOne
SwanOne is a model developed by TUDelft in MatLab to simulate the evolution of the wave
spectrum starting from deep water to shores. The model is capable to simulate interactions and
transformations of waves (TUDelft, 2015).
The different input parameters are:
· Bottom profile
· Current
· Wave direction
· Water level
· Wind velocity and direction
· Boundary conditions: arbitrary spectrum file or definition input diameters (significant
wave height based on spectral analysis ܪ[m] and the peak period ܶ [s]).
Also the output locations should be provided, i.e. the locations along the profile where you want
to obtain the output parameters.
The output parameters are:
· Significant wave height ܪ [m]
· Root-mean-square wave height ܪ௦ [m]
· Peak period ܶ [s]
· Mean absolute period ܶଵ [s]
· Mean period ܶଶ [s]
· Mean energy wave period ܶିଵǡ [s]
· Significant wave height calculated using the Battjes and Groenendijk method ܪ௦ [m]
· Wave height exceeded by 2% of the waves ܪଶΨ [m]
· Mean wave height of highest 1/10 fraction of waves ܪଵȀଵ [m]
No breakwater model is used in SwanOne thus also no reflection by the model is possible.
Therefore the obtained parameters are the incident characteristics.
35
Measurements by wave gauges in canal without a model
A second method to measure and calculate the wave characteristics is using directly the surface
elevations measured by the wave gauges in the wave flume. To avoid reflection by the model,
no model is placed in the canal. Theoretically the waves will not be reflected by the end side of
the canal since the energy of the waves is dissipated there by the energy dissipating system.
The wave characteristics are obtained by the software developed in the LPC and is called LPCLab
2.0. LPCLab 2.0 calculates different wave heights and period spectra and moments for later
calculations using the surface elevations measured by the wave gauges.
The wave characteristics are analysed by LPCLab 2.0 in both the time-domain and frequency
domain and generates information about all relevant parameters and gives also some graphs.
i. Time-domain analysis
In the time-domain, each individual wave is defined by the downward crossing of the zero-line
by the surface elevation (zero down-crossing). The mean wave height ܪ is calculated from
time-series of individual waves.
ii. Frequency domain analysis
In addition to the time domain analysis , the wave spectrum of the realized test is calculated
using the discrete Fourier transform (DFT) of the measured surface elevation.
The total registered waves are divided in different time-windows. The fast Fourier transform
(FFT) is done over the different windows and the results are summed at the end. The width of
the time-window, the number of points to represent the spectrum and the percentage of
overlap between the different windows can be chosen by the user. Another important
parameter to put in LPCLab 2.0 is a range of frequencies so useless frequencies can be
eliminated. A screenshot of LPCLab 2.0 is given in Figure 22.
36
Measurements by wave gauges in canal with model.
LPCLab 2.0 can also be used to process the surface elevations that are measured by the wave
gauges in the wave flume when the model is present. The only problem is that the model will
reflect the waves so LPCLab 2.0 won’t generate the incident wave characteristics, but only the
registered wave characteristics. To overcome this problem, the LASA method can be used.
However, the LASA or LASA-V method can’t be used in breaking wave conditions, so the LASA
method can only be applied on the wave gauges in front of the wave paddle, where the waves
aren’t breaking. Hence the incident and reflected waves are known for the sensors in front of
the wave paddles and only the total registered waves are known for the rest of the sensors.
Starting from the calculated reflection coefficient in front of the wave paddle (mean value of the
3 wave gauges in front of wave paddle) obtained from the LASA method, the height of the
incident and reflected waves can be calculated using equations (3) and (4) starting from the total
measured wave height by the other wave gauges where the waves are breaking. The problem
with this method is that the assumption is very rough (it’s assumed that the reflection coefficient
in front of the wave paddle is equal for the whole canal). Of course this is not the case.
Comparison between SwanOne and measurement in canal without model
A comparison between the significant wave height obtained from SwanOne and the significant
wave height obtained from the measurements in the canal without the model is depicted in
Figure 23.
It can be stated that the values of the significant wave height obtained from SwanOne are
slightly higher than the ones obtained from the measured water elevations in the canal without
model.
For this master thesis, the wave characteristics are obtained from SwanOne.
20,00
15,00
Hm0, t (SWAN)[cm]
10,00
5,00
0,00
0,00 5,00 10,00 15,00 20,00
Hm0,t (without model) [cm]
Figure 23: Comparison between significant wave height obtained from SwanOne and measurements in the canal
without the model
37
b. Porosity measurement
The porosity has to be measured for 2 purposes. First, the initial porosity has to be measured to
verify if this one is about 37% as prescribed (CERC, 1984). Also, the porosity after each test
should be measured, because this porosity is used to calculate the damage.
The porosity , equal to the ratio of the area of the voids ܣ௩ and the total area ܣ௧௧ , is calculated
counting the number of stones. The counting of the stones is done using the software AutoCAD.
First a virtual net is drawn over the picture of the breakwater. Different strips are drawn. This is
done because the Virtual Net Method requires this (which is used for other test in the ESCOLIF
project, but doesn’t has a real meaning for this master thesis). Then each stone is marked (Figure
24). Finally, by the command PRICAPAXYZ, developed by the LPC-UPV, the marks are counted
(ܰ) and saved. Finally the porosity can be calculated:
ଶ
ܣ௩ ܰܦହ (53)
ൌ ൌͳെ
ܣ௧௧ ܣ௧௧
c. Damage calculation
The damage of the armour layer is calculated according to the visual counting method. This
method is discussed in the literature study and makes use of equations (31) and (32). The
number of eroded stones is calculated by counting the difference of stones between the initial
state and the state after wave attack. To do so, again AutoCAD is used as described above.
To do the qualitative analysis of the damage, the damage criteria are distinguished as follows:
· Initiation of damage (IDa): 5 or more units are displaced from their original position to
a new one at a distance equal to or larger than a unit length
· Initiation of Iribarren damage (IIDa): One stone and his surrounding stones of the 2nd
layer are visible
· Initiation of destruction (IDe): 2 or more stones in the lower armour layer are forced out
· Destruction (De): The filter layer is visible
38
d. Comparison measured damage with predicted damage
Once the damage is calculated, the measured damage will be compared with the predicted
damage. This predicted damage will be calculated using 2 approaches: The SPM-approach and
the VDM approach.
SPM-approach
The first approach is the one based on the SPM (CERC, 1984) by Van der Meer (1988) and Medina
et al. (1994):
Van der Meer (1988) ܪ௦ ଵ (54)
ൌ ͲǤሺܭ ܿߙݐሻଷ ܵௗǤଵହ
οܦହ
ͳ ܪ௦ ଵȀǤଵହ
֜ ܵௗ ൌ ൬ ൰
ͲǤሺܭ
ߙሻଵȀଷ οܦହ
All parameters are explained in the literature study. For the Van der Meer (1988) formula, the
value for the stability coefficient ܭ is taken to be equal to 8 because the goal is to predict the
damage so the value for an permeable core describing the main trend is chosen.
The damage will be predicted using both prediction formulas for 5 different wave heights. First
the significant wave height obtained from spectral analysis will be used because this is also the
wave height proposed by the authors of the formulas. But because the tests are performed in
breaking conditions, the damage will also be calculated using the breaking wave height ܪ .
There are different methods to calculate the breaking wave height for regular waves, but for
irregular waves the waves can break in a wide range as discussed before so there is not one right
value. Because of this, the breaking wave height is calculated in 4 different ways. The first 3 are
based on the method described in the Shore Protection Manual(CERC, 1984). For this method,
Figure 3 based on Goda (1970) can be used. It’s not clearly defined which period should be used
to calculate the deep water wave steepness, so the breaking wave height ܪ is calculated 3
times based on the mean absolute period ܶ ൌ ʹߨ݉ Ȁ݉ଵ, the mean energy wave period
ܶିଵǡ ൌ ʹߨ݉ିଵ Ȁ݉ and the peak period ܶ with all periods obtained from SwanOne. As 5th
used wave height, the wave height proposed by Goda (2000) for irregular conditions is used, i.e.
the peak value of the significant wave height ܪଵȀଷ in the surf zone which can be read from Figure
6.
Next the predicted and measured damage will be compared for each formula and each used
wave height. Furthermore 5 new formulas will be presented using the 5 different wave heights
based on all the tests except for the repetition test ( ݎܫൌ ͵ and ݄ௗ ൌ ʹͲܿ݉).
39
The new formula will be developed by calculating the values of the constant coefficients ܥଵ and
ܥଶ of:
ܪ (56)
ൌ ܥଵ ሺ
ߙሻଵȀଷ ܵௗ మ
οܦହ
The coefficients will be calculated using the Solver function in excel obtaining the lowest relative
mean squared errorܧܵܯݎ. The ܧܵܯݎis used to evaluate the extent to which measured and
predicted damage are similar. It’s equal to:
ଶ (57)
σ൫ܵௗǡௗ௧ௗǡ െ ܵௗǡ௦௨ௗǡ ൯
ܧܵܯݎൌ ଶ
തതതതതതതതതതതതതതത
σ൫ܵ ௗǡ௦௨ௗ െ ܵௗǡ௦௨ௗǡ ൯
Also the correlation coefficient ܴ will be calculated, but this is only a measure of the linear
correlation between the measured and predicted damage, so won’t be really used in the
discussion of the comparison between the predicted and measured damage..
The repetition test will be used to verify the new formulas as a blind test. In conclusion the best
of the 5 different formulas will be chosen to predict the damage.
No cumulative damage is taken into account because the formulas from the SPM-approach do
not include the number of waves. So it can be assumed that the damage of a lower wave height
which is already present before the initiation of the wave series with a higher wave height also
will be caused by a certain (lower) number of waves with the higher wave height than the
previous waves.
VDM-approach
The second approach is based on Van der Meer (1988) and Van Gent et al. (2004). The damage
for each test will be predicted (taking into account cumulative damage as discussed before) and
compared to the measured damage (the data points with a measured damage higher than 15
were deleted, because it doesn’t make sense to take them into account; destruction will already
have occurred at such a high value). This is also shown in Table 2. The prediction of the damage
will be done 3 times. Once according to the Van der Meer formula (in deep or shallow water
according to the conditions), once according to the Van der Meer formula modified by Van Gent
and finally once according to the formula by Van Gent. A summary of all formulas is given in
Table 12.
40
Table 12: Summary formulas VDM-approach
ࡴ࢙ ࡿࢊ Ǥ ିǤ
ൌ ࢉ ࡼǤૡ ൬ ൰ ࣈ
Deep οࡰ ξࡺ
VDM Plunging (58)
water ࡿࢊ ࡴ࢙
֜ ൌ ቆ ࡼିǤૡ ࣈ
Ǥ
ቇ
ξࡺ ࢉ οࡰ
ܪ௦ ିǤଵଷ
ܵௗ Ǥଶ
ܵௗ
ൌ ܿ௦ ܲ ൬ ൰ ξ
ߙ ߦ ֜
Deep οܦହ ξܰ ξܰ
VDM Surging ହ
(59)
water ͳ Ǥଵଷ ି ܪ௦
ൌ ൬ ܲ ߦ ξ ߙ ൰
ܿ௦ οܦହ
ܪ௦ ିǤଵଷ
ܵௗ Ǥଶ ܪ௦
ൌ ܿ௦ ܲ ൬ ൰ ξ
ߙ ߦ
Shallow οܦହ ξܰ ܪଶΨ
VDM Surging ହ (61)
water ܵௗ ͳ Ǥଵଷ ି ܪଶΨ ܪ௦
֜ ൌ ൬ ܲ ߦ ξ ߙ ൰
ξܰ ܿ௦ ܪ௦ οܦହ
The 3 formulas will be compared using again the ܧܵܯݎand in a smaller extent the correlation
coefficient ܴ and some improvements to the original formulas will be presented. Finally the
modified original formulas will be verified with a blind test.
41
Chapter 4: Results
In this chapter the results of the comparison between the measured and predicted damage are
presented. The comparison is based on all test but the repetition test series. This repetition test
is used later as a blind test to verify the new and improved formulas.
The comparison is done for the 2 approaches as discussed before, i.e. The SPM-approach and
the VDM-approach.
To quantify the error between the measured and predicted damage, the relative mean squared
error (ܧܵܯݎሻ is calculated for each case.
Finally a new formula or improvement to the original formula is proposed based on the
comparison between measured and predicted damage. The repetition test is used to verify these
new and improved formulas (blind test).
1. Wave data
As mentioned before, the used wave data is obtained from SwanOne at the toe of the model
(unless otherwise mentioned). The most important parameters are listed in Table 13, Table 14
and
Table 15. All data from SwanOne at the toe of the structure is listed in Appendix A.
Table 13: Summary of the SwanOne data at the toe of the structure (݄௧ ൌ ʹͲܿ݉ሻ
42
Table 14: Summary of the SwanOne data at the toe of the structure (݄௧ ൌ ͵Ͳܿ݉)
Table 15: Summary of the SwanOne data at the toe of the structure (݄௧ ൌ ͶͲܿ݉)
43
2. Measured damage and porosity
The measured damages and porosities for the initial state and after wave series are listed in
Table 16 to Table 19
The initial porosity of each test series was 36% or 37%. So this is in line with the requirements
(CERC, 1984).
Also the different damage criteria are identified. This is based on the photos taken after each
wave series. The photos are listed in Appendix B. The average damage parameter ܵௗ for each
damage criteria is calculated and the following values are obtained:
· Initiation of damage: ܵௗ ൌ ͲǤ
· Initiation of Iribarren damage: ܵௗ ൌ ʹǤ
· Initiation of destruction: ܵௗ ൌ ͷǤͻ
· Destruction: ܵௗ ൌ ͳͳǤͲ
Table 16: Measured damage, porosity and qualitative damage analysis for ݄௧ ൌ ʹͲܿ݉
44
Table 17: Measured damage, porosity and qualitative damage analysis at ݄௧ ൌ ͵Ͳܿ݉
ࡵ࢘ ࡴ࢙ ࡿࢊ ࡵ࢘ ࡴ࢙ ࡿࢊ
Q.A. Q.A.
ሾെሿ ሾࢉሿ ሾെሿ ሾെሿ ሾെሿ ሾࢉሿ ሾെሿ ሾെሿ
initial 0.0 36% initial 0 36%
8 0.0 36% 8 0.1 36%
9 0.0 36% 9 0.6 37% IDa
10 0.3 37% 10 0.9 38%
11 0.7 37% IDa 11 3.8 44% IIDa
12 1.7 40% 12 7.8 53% IDe
3 5
13 3.8 44% IIDa 13 13.4 64% De
14 5.0 47% IDe 14 21.6 82%
15 7.0 51%
16 11.0 59%
17 17.4 73% De
initial 0.0 36%
Table 18: Measured damage, porosity and qualitative damage analysis for ݄௧ ൌ ͶͲܿ݉ and Ir=3
ࡴ࢙ ࡿࢊ
Q.A.
ሾࢉሿ ሾെሿ ሾെሿ
initial 0.0 36%
8 0.0 36%
9 0.0 36%
10 0.3 36%
11 0.8 37% IDa
12 1.3 38%
13 3.0 42% IIDa
14 6.7 50% IDe
15 10.6 58% De
16 21.0 80%
45
Table 19: Measured damage, porosity and qualitative damage analysis for the repetition test (݄௧ ൌ ʹͲܿ݉, Ir=3)
ࡴ࢙ ࡿࢊ
Q.A.
ሾࢉሿ ሾെሿ ሾെሿ
initial 0.0 36%
8 0.0 36%
9 0.1 36%
10 0.2 36%
11 0.6 37% IDa
12 0.9 38%
13 2.0 40%
14 2.6 41% IIDa
15 3.0 42%
16 4.7 46%
17 5.2 47% IDe
18 7.3 51%
19 10.5 58% De
46
Comparison Medina, VDM and New formula Comparison Medina, VDM and New formula
based on Hm0 (SwanOne) based on Hbr (Goda, 2000)
2,0 2,0
Medina(X=0.2) Medina (X=0.2)
VDM (X=0.15) VDM (X=0.15)
1,6 1,6
1,2 1,2
0,8 R=0.96
0,8 R=0.92
rMSE=0.18 rMSE=0.21
R=0.97 R=0.93
0,4 rMSE=0.73
0,4 rMSE=0.25
R=0.97 R=0.99
rMSE=0.07 rMSE=0.14
0,0 0,0
0,0 0,4 0,8 1,2 1,6 2,0 0,0 0,4 0,8 1,2 1,6 2,0
Measured damage Sd^X[-] Measured damage Sd^X[-]
Figure 25: Comparison formulas based on ܪ Figure 26: Comparison formulas based on the ܪ (Goda,
2000)
Comparison Medina, VDM and New formula Comparison Medina, VDM and New formula
based on Hbr (Tm) (CERC, 1984) based on Hbr (Tm-1) (CERC, 1984)
2,0 2,0
Medina (X=0.2) Medina (X=0.2)
VDM (X=0.15) VDM (X=0.15)
1,6 New Formula (X=0.17) 1,6 New Formula (X=0.17)
Predicted damage SSPM^X[-]
1,2 1,2
Figure 27: Comparison formulas based on ܪ ሺܶ ሻ Figure 28: Comparison formulas based on ܪ ሺܶିଵǡ ሻ
Comparison Medina, VDM and New formula Verification new formulas by repeated test
based on Hbr (Tp) (CERC, 1984)
2,0 2,0
Medina (X=0.2) Hm0 SWAN
VDM (X=0.15) Hb (Tm), SPM
1,6 New Formula(X=0.17)
Predicted damage SSPM^X[-]
Hb (Tm-1), SPM
Predicted damage SSPM^X[-]
Hb (Tp), SPM
1,5 Hb, Goda
1,2
0,8 R=0.92
rMSE=0.91 1,0
R=0.93 rMSE=0.05
0,4 rMSE=0.33 rMSE=0.10
R=1.00 rMSE=0.12
rMSE=0.14 rMSE=0.15
0,0 rMSE=0.16
0,5
0,0 0,4 0,8 1,2 1,6 2,0
0,5 1,0 1,5 2,0
Measured damage Sd^X[-] Measured damage Sd^X[-]
Figure 29: Comparison formulas based on ܪ ሺܶ ሻ Figure 30: Verification new formulas (blind test)
47
Table 20: Proposed new formulas based on SPM-approach
ܪ
ܪ ൌ ͳǤͳሺ
ߙሻଵȀଷ ܵௗǤଵସ (65)
οܦହ
ܪ
ܪ ൌ ሺܪଵȀଷ ሻ ൌ ͳǤʹሺ
ߙሻଵȀଷ ܵௗǤଵ (66)
οܦହ
ܪ
ܪ based on ܶ ൌ ͳǤ͵ʹሺ
ߙሻଵȀଷ ܵௗǤଵ (67)
οܦହ
ܪ
ܪ based on ܶିଵǡ ൌ ͳǤ͵ሺ
ߙሻଵȀଷ ܵௗǤଵ (68)
οܦହ
ܪ
ܪ based on ܶ ൌ ͳǤͶሺ
ߙሻଵȀଷ ܵௗǤଵ (69)
οܦହ
The ܧܵܯݎof all formulas using all the different wave height are given in Table 21. It can be
concluded that for the existing formulas the Medina formula using the significant wave height
ܪ is the best formula to predict the damage ( ܧܵܯݎൌ ͲǤͳͺሻ. Also the VDM formula using
breaking wave height ܪ gives a low ܧܵܯݎ, which is equal to 0.19 using both ܶ or ܶିଵǡ.
The best fitting proposed new formula is equation (65) using ܪ. The ܧܵܯݎequals 0.07. All
proposed new formulas were verified by a blind test (Figure 30), and it can be concluded that
also here, the best fitting formula is equation (65) with a ܧܵܯݎof 0.05.
48
b. VDM-approach
The comparison between the predicted and the measured damage calculated by the original
Van der Meer formula is depicted in Figure 31. As can be seen on the figure, it’s clear that all
predicted damage is lower than the measured damage. It can also be observed that the lower
the Iribarren number and the higher the water depth, the bigger the difference between the
measured and the predicted damage. That is why the original formula will be improved by
introducing the water depth at the toe of the structure ݄௧ and the peak period ܶ which
influences the Iribarren number. All waves are plunging waves so only the formulas for plunging
waves can be improved. The power to which the new dimensionless parameter is raised and the
value of the constant are calculated using the solver function in excel obtaining the smallest
ܧܵܯݎ:
For shallow water:
ିǤଶ
ܪ௦ כ
ʹߨ݄௧ ܵௗ Ǥଶ ܪ௦ ିǤହ (70)
ൌ ܿ ቆ ቇ ܲǤଵ଼ ൬ ൰ ߦ כ
ܿ ൌ ͶǤͳ
οܦହ ݃ܶଶ ξܰ ܪଶΨ
By introducing the dimensionless parameter that takes into account the influence of the water
depth and Iribarren number, the ܧܵܯݎdecreases from 0.77 to 0.09 which is a very big
improvement (Figure 32). Also the correlation coefficient improves. The interval in which 90%
of the values are located is also indicated on Figure 32.
12 12
Predicted damage Sd[-]
hs=30,Ir=3
9 hs=30,Ir=5
9
hs=40,Ir=3
6 hs=20,Ir=3 6
hs=20,Ir=3
3 hs=30,Ir=3 3
hs=30,Ir=5
0 hs=40,Ir=3 0
0 3 6 9 12 15 0 3 6 9 12 15
Measured damage Sd[-] Measured damage Sd[-]
Figure 31: Comparison measured and predicted damage Figure 32: Comparison measured and predicted damage
by formula of VDM by improved formula of VDM
The same improvement was executed for the VDM formula modified by VG. The same
observations and conclusions can be made: Almost everywhere the measured damage is higher
than the predicted damage, and the lower the Iribarren number & higher the water depth, the
higher the difference between measured and predicted damage (Figure 33). Again the same
improvement is made by introducing a dimensionless parameter.
49
ିǤଶ Ǥଶ
ܪ௦ כ
ʹߨ݄௧ ܵௗ ܪ௦ ିǤହ (72)
ൌ ܿ ቆ ቇ ܲǤଵ଼ ൬ ൰ ߦ כ
ܿ ൌ ͶǤ͵
οܦହ ݃ܶଶ ξܰ ܪଶΨ
Again the comparison is improved by introducing the dimensionless parameter. The ܧܵܯݎ
decreases from 0.49 to 0.09. The interval in which 90% of the values are located is indicated on
the figure.
Comparison VDM formula (modified by VG) Comparison improved VDM formula (modified
by VG)
16 15
R=0.77 R=0.97 hs=20,Ir=5
hs=20,Ir=5
rMSE=0.49 rMSE=0.09
12 12
hs=40,Ir=3
hs=30,Ir=5
9
8 hs=30,Ir=5
hs=20,Ir=3
6
4 hs=30,Ir=3 hs=20,Ir=3
3
hs=40,Ir=3 hs=30,Ir=3
0
0 4 8 12 16 0
Measured damage Sd[-] 0 3 6 9 12 15
Measured damage Sd[-]
Figure 33: Comparison measured and predicted damage Figure 34: Comparison measured and predicted damage
by formula of VDM modified by VG by improved formula of VDM modified by VG
If the comparison between the measured and predicted damage by the VG formula is observed,
there can’t be really made a distinction between the different test series. The only conclusion
that can be made is that the predicted damage is almost always lower than the measured
damage (Figure 35). That’s the reason why only the constant coefficient is changed:
By changing the coefficient from 0.57 to 0.67, the results are improved from a ܧܵܯݎof 0.59 to
0.12 (Figure 36). The interval in which 90% of the values are located is indicated on the figure.
50
Comparison VG formula Comparison improved VG formula
15
15 R=0.94
R=0.94 hs=20,Ir=3 hs=20,Ir=5
rMSE=0.12
rMSE=0.59 12
12
hs=30,Ir=3 hs=40,Ir=3
9 9
hs=30,Ir=5 hs=30,Ir=5
6 6
hs=40,Ir=3 hs=30,Ir=3
3 3
0 hs=20,Ir=5 hs=20,Ir=3
0
0 3 6 9 12 15
0 3 6 9 12 15
Measured damage Sd[-] Measured damage Sd[-]
Figure 35: Comparison measured and predicted damage Figure 36: Comparison measured and predicted damage
by formula of VG by the improved formula of VG
Finally, the improved formulas of VDM, VDM modified by VG and VG can be verified by a ‘blind
test’ using the repetition test. The comparison between the measured damage and the
predicted damage by the 3 new formulas is given in Figure 37. To make the improvement clearer
also the predicted values by the original formulas are depicted in the graph.
It can be concluded that the new formulas give a good improvement of the original formulas.
51
Chapter 5: Conclusions
The aim of this thesis was to compare and improve different prediction formulas. 2 different
approaches were used: the approach based on the SPM (CERC, 1984) and the approach based
on Van der Meer (1988).
First of all, a qualitative analysis was executed determining the Initiation of damage, Initiation
of Iribarren damage, Initiation of destruction and destruction. The average values based on the
tests are given in Table 22.
Initiation of damage ࡿࢊ ൌ Ǥ ૠ
Destruction ܵௗ ൌ ͳͳǤͲ
If the (deep water) wave characteristics are not known, these can be obtained using SwanOne.
The best formula to use these wave characteristics based on the executed tests is the new
proposed formula using the significant wave height ܪ:
ܪ
ൌ ͳǤͳሺ
ߙሻଵȀଷ ܵௗǤଵସ
οܦହ
If the deep water characteristics are known, the breaking wave height can be calculated and
used to predict the damage according to the SPM approach using the new proposed formula
with the mean period ܶ :
ܪ
ൌ ͳǤ͵ʹሺ
ߙሻଵȀଷ ܵௗǤଵ
οܦହ
Also the new proposed formula using the peak value of the significant wave height in the surf
zone can be used if the deep water characteristics are known:
ܪ
ൌ ͳǤʹሺ
ߙሻଵȀଷ ܵௗǤଵ
οܦହ
Using the VDM formula, the breaking wave height based on ܶ or ܶିଵ gives the best prediction
of the damage while using the Medina formula the significant wave height ܪ at the toe of the
structure obtained by SwanOne gives the best approximation.
For the approach based on Van der Meer (1988), the improved VDM formula (deep & shallow
water) and improved VDM formula modified by VG for plunging waves and improved VG formula
are given respectively by:
ିǤଶ Ǥଶ
ܪ௦ כ
ʹߨ݄௧ Ǥଵ଼
ܵௗ ܪ௦ ିǤହ
ൌ ܿ ቆ ቇ ܲ ൬ ൰ ߦ כ
ܿ ൌ ͶǤͳ
οܦହ ݃ܶଶ ξܰ ܪଶΨ
52
ିǤଶ כ
ܪ௦ ʹߨ݄௧ ܵௗ Ǥଶ ܿ ൌ ʹǤͻ
כ
ൌ ܿ ቆ ቇ ܲǤଵ଼ ൬ ൰ ିǤହ
ߦ
οܦହ ݃ܶଶ ξܰ
ିǤଶ
ܪ௦ כ
ʹߨ݄௧ ܵௗ Ǥଶ ܪ௦ ିǤହ כ
ܿ ൌ ͶǤ͵
ൌ ܿ ቆ ቇ ܲǤଵ଼ ൬ ൰ ߦ
οܦହ ݃ܶଶ ξܰ ܪଶΨ ௦ିଵ
53
References
Battjes, J. A. (1974). Surf Similarity. 14th International conference on Coastal engineering,
Copenhagen, Denmark.
Battjes, J. A. and M. J. F. Stive (1985). "Calibration and Verification of a Dissipation Model for
Random Breaking Waves." Journal of Geophysical Research: Oceans 90(C5): 9159-9167.
Camenen, B. and M. Larson (2007). "Predictive Formulas for Breaker Depth Index and Breaker
Type." Journal of Coastal Research: 1028-1041.
CERC (1977). Shore Protection Manual, Washington (D.C.) : US government printing office.
CERC (1984). Shore Protection Manual, Washington (D.C.) : Government printing office.
CIRIA (2007). The Rock Manual : The Use of Rock in Hydraulic Engineering, London : CIRIA.
Figueres, M. and J. R. Medina (2004). Estimating Incident and Reflected Waves Using a Fully
Nonlinear Wave Model: 594-603.
Gaillard, P., et al. (1980). "Method of Analysis of Random Wave Experiments with Reflecting
Coastal Structures." Coastal Engineering Proceedings(17).
54
Goda, Y. (1970). "A Synthesis of Breaker Indices." Transactions of the Japan Society of Japan
Engineers 2(Part 2): 227-230.
Goda, Y. (2000). Random Seas and Design of Maritime Structures, Singapore : World scientific.
Goda, Y. (2010). "Reanalysis of Regular and Random Breaking Wave Statistics." Coastal
Engineering Journal 52(01): 71-106.
Goda, Y. and T. Suzuki (1976). "Estimation of Incident and Reflected Waves in Random Wave
Experiments." Coastal Engineering Proceedings(15).
Hudson, R. Y. and L. F. Moore (1951). "The Hydraulic Model as an Aid in Breakwater Design."
1951(1).
Iribarren, C. R. (1938). Una Fórmula Para El Cùalculo De Los Diques De Escollera. Pasajes, Spain.
Kajima, R. (1969). "Estimation of Incident Wave Spectrum in the Sea Area Influenced by
Reflection." Coastal Engineering in Japan 12: 9-16.
Kimura, A. (1985). "The Decomposition of Incident and Reflected Random Wave Envelopes."
Coastal Engineering in Japan: 59-69.
Losada, M., et al. (1986). "Stability of Blocks as Breakwaters Armour Units." Journal of
Structural Engineering(112): 2392-2401.
55
Mansard, E. P. D. and E. R. Funke (1980). The Measurement of Incident and Reflected Spectra
Using a Least Squares Method. B. L. Edge. Sydney, Australia, American Society of Civil
Engineers: 154-172.
Medina, J. R. (2001). "Estimation of Incident and Reflected Waves Using Simulated Annealing."
Journal of Waterway, Port, Coastal, and Ocean Engineering 127(4): 213-221.
Medina, J. R., et al. (1994). "Breakwater Armor Damage Due to Wave Groups." Journal of
Waterway, Port, Coastal, and Ocean Engineering 120(2): 179-198.
Munk, W. H. (1949). "The Solitary Wave Theory and Its Applications to Surf Problems." Annals
of the New York Academy of Sciences 51: 376-462.
Nairn, R. (1990). Prediction of Cross-Shore Sediment Transport and Beach Profile Evolution.
London, UK, Imperial College.
Rattanapitikon, W., et al. (2003). "A Proposal of New Breaker Height Formula." Coastal
Engineering Journal 45(1): 29-48.
Ruessink, B. G., et al. (2003). "Calibration and Verification of a Parametric Wave Model on
Barred Beaches." Coastal Engineering 48(3): 139-149.
56
Smith, E. R. and N. C. Kraus (1990). Laboratory Study on Macro-Features of Wave Breaking
over Bars and Artificial Reefs / by Ernest R. Smith, Nicholas C. Kraus. Vicksburg, Miss. :, Coastal
Engineering Research Center
U.S. Army Engineer Waterways Experiment Station.
United States.
Smith, G. M., et al. (2002). Paper No: 131 Rock Slope Stability with Shallow Foreshores. Solving
Coastal Conundrums.
Stokes, G. G. (1880). On the Theory of Oscillatory Waves. Mathematical and Physical Papers,
Cambridge University Press. 1: 199-229.
USACE (2002). Coastal Engineering Manual. Washington, D.C., U.S. Army Corps of Engineers.
Van der Meer, J. W. (1988). Rock Slopes and Gravel Beaches under Wave Attack, Delf
University of Technology.
Van der Meer, J. W. (1998). Application and Stability Criteria for Rock and Artificial Units.
Seawalls, Dikes and Revetments. K. W. Pilarczyk. Balkema, Rotterdam.
Van der Meer, J. W. (2000). Design of Concrete Armour Layers. Proc 3rd dint con. coastal
structures, Santander, Spain, ASCE, New York, USA.
Van Gent, M. A., et al. (2004). Stability of Rock Slopes with Shallow Foreshores. Portland: 100-
112.
Vidal, C., et al. (1991). "Stability of Mound Breakwater's Head and Trunk." Journal of
Waterway, Port, Coastal and Ocean engineering: 570-587.
57
Vidal, C., et al. (2006). Wave Height Parameter for Damage Description of Rubble-Mound
Breakwaters, Elsevier B.V.
Weggel, J. R. (1972). "Maximum Breaker Height." Journal of the Waterways, Harbors and
Coastal Engineering Devision 98(WW4): 529-548.
58
Appendix A: Summary data SwanOne
59
Appendix B: Photos model after wave
action
hs=20cm and Ir=3
60
hs=30cm and Ir=3
61
hs=30cm and Ir=5
62
hs=20cm and Ir=3 (repetition test)
63
64
Hydraulic stability of rubble mound breakwaters in
breaking wave conditions: a comparative study of existing
prediction formulas
Sander Franco