You are on page 1of 16

Coastal Engineering 57 (2010) 359–374

Contents lists available at ScienceDirect

Coastal Engineering
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / c o a s t a l e n g

A new formula for front slope recession of berm breakwaters


T. Lykke Andersen ⁎, H.F. Burcharth
Aalborg University, Department of Civil Engineering, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark

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

Article history: The front slope stability of breakwaters with a homogeneous berm was studied in a large number of two
Received 23 September 2009 dimensional model tests at Aalborg University, Denmark. The results are presented together with a new
Accepted 30 October 2009 formula for prediction of the berm recession which is the most important parameter for describing the
Available online 2 December 2009
reshaping. The formula has also been calibrated and validated against model test data from other
researchers. The significance of the new design formula is that it predicts berm recession much better than
Keywords:
the existing methods, especially in case of more stable structures.
Berm breakwaters
Front slope stability
© 2009 Elsevier B.V. All rights reserved.
Berm recession
Physical model tests

1. Introduction i.e. the top level of the core is not so critical as long as it is some distance
below SWL. Thus it can be expected that the present results can also be
A berm breakwater is a rubble mound breakwater initially used in cases where the core is extended into the berm as shown in
constructed with a large porous berm above or at still water level at Fig. 3. The predictions by the new recession formula have also been
the seaward side. During wave attack the berm breakwater will typically compared to test results for multi-layer berm breakwaters, cf. example
reshape into a s-shaped profile, cf. Fig. 1. Lately also more stable berm shown in Fig. 2. In case of multi-layer structures the front slope damage
breakwaters designed for very small deformations have been con- does not necessarily start in the top part of the berm because the largest
structed. Especially the multi-layered Icelandic berm breakwater stones are placed here but could be in a lower part of the front slope or
design, used for example in the Sirevag and Hammerfest breakwaters on the crest. The advantage of the multi-layered structure is that it can
is gaining more attention, Sigurdarson et al. (2003, 2007). The berm be designed to withstand very large design waves without significant
breakwater is preferably designed to make optimum use of the local recession and therefore much less problems with stone durability. The
yield from the quarry. This is obtained by sorting the quarry material added stability should however be compared to the extra cost of the
into more stone classes to be used in various parts of the structure. By more sorting and the more complicated construction method.
placing the largest stones in the most exposed part of the structure, i.e. An important and simple measure for the reshaping of a
on the top and in front of the berm, a more stable structure is obtained homogeneous berm breakwater is the recession of the berm (Rec)
compared to one with less separation of the material, cf. Fig. 2. as defined by Burcharth and Frigaard (1988), cf. Fig. 1. Failure is
Contributing to this is also that the more narrow gradations obtained by typically defined as Rec N B, where B is the berm width. The depth at
sorting have larger permeability. Large permeability is important for which the reshaped profile intersects with the initial profile (hf) is a
stability especially in the upper part of the berm and in the crest of the measure which in combination with the recession (Rec), can be used
structure, but is expected to be of minor importance for the rest of the to estimate the eroded volume. The extention of the profile at the sea
structure. The design and construction of berm breakwaters are bed ΔR can be important for design of the extension of the bedding
described in the PIANC (2003) report “State-of-the-Art of Designing layer and scour protection. The parameters Rec and hf are only good
and Constructing Berm Breakwaters”. measures for the reshaped profile when the berm is initially located
The present paper focuses mainly on the stability of berm break- above or at the still water level, SWL. If the berm is placed below SWL,
waters with only two stone classes, i.e. wide graded core material and damage can occur to the crest even when Rec = 0. For such cases it is
relative well sorted stones to form a homogenous berm, cf. Fig. 1. Lissev not possible to use Rec as a damage parameter.
(1993) and Lissev and Tørum (1996) studied the influence of the In addition to ensuring that Rec ≤ B for design conditions, the
material on the berm stability and found that the reshaped profiles were designer has to consider the longshore transport caused by oblique
almost identical for the four tested core configurations shown in Fig. 3, waves. Also, it is important to consider breaking and abrasion of
armour stones which might occur due to stone movements. This is
⁎ Corresponding author.
especially important for dynamically stable structures where frequent
E-mail addresses: tla@civil.aau.dk (T. Lykke Andersen), hfb@civil.aau.dk stone movements occur during storms. The problem of durability of
(H.F. Burcharth). the stone material is not dealt with in the paper.

0378-3839/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.coastaleng.2009.10.017
360 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

Fig. 1. Typical initial and reshaped profile for a berm breakwater with a homogeneous berm. hb is negative when the berm level is higher than the still water level.

Fig. 2. Example of a multi-layered berm breakwater.

Fig. 3. Core configurations tested by Lissev (1993).

2. Existing berm recession prediction methods and be the governing period for the reshaping, at least for large values of
stability classification H0.

The stability of a rubble mound is traditionally characterized by sffiffiffiffiffiffiffiffiffiffiffi


Hm0 g
the stability parameter Ns = Hm0/ΔDn. In the context of berm break- H0 T0 = · · T0;1 ð2Þ
Δ · Dn;50 Dn;50
waters it is commonly denoted

Hm0 T0,1 = m0/m1 is a spectral mean wave period calculated from the
H0 = ð1Þ
Δ · Dn;50 zero and the first order moments of the spectrum.
Berm breakwaters are classified into statically stable non-reshaped
where Hm0 is the incident significant wave height in front of the structures (H0 b 2–3, H0T0 b 20–40), statically stable reshaped (H0 = approx. 1.5
based on frequency domain analysis, Δ=ρs/ρw −1 is the relative reduced to 2.7, H0T0 = approx. 40 to 70) and dynamically stable reshaped
mass density, and Dn,50 = Volume1/3 is the equivalent cube length (H0 N 2.7, H0T0 N 70), cf. PIANC (2003).
exceeded by 50% of the armour stone units. The conventional linear For calculating the recession of the berm a number of methods
relation between wave height (Hm0) and the relative reduced mass exists, i.e. the numerical methods of Van der Meer (1992), Van Gent
density (Δ) expressed in Eq. (1) is assumed valid although the relation (1995) and Archetti and Lamberti (1996) and the simple regression
seems non-linear, Helgason and Burcharth (2005). models of Hall and Kao (1991), Tørum (1998), Tørum et al. (1999) and
The influence of the wave period on the breakwater stability can Tørum and Krogh (2000).
be introduced in different ways, but the Van der Meer (1988) The first three listed methods give the entire profile, but a computer
dimensionless parameter H0T0 as defined in Eq. (2) is very often used. program is really needed to perform the quite complicated calculations.
It contains the mean wave period which was found by Van der Meer to The last four methods are simple empirical formulae. However, the
T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 361

Fig. 4. Initial cross-section of breakwater. Note hb is negative when berm is above SWL.

Table 1 breakwaters. The test programme with a total of 636 tests covers the
Material properties. influence of berm width, berm elevation, sea state (wave height and
Armour 1 Armour 2 Armour 3 Armour 4 Core steepness), stone size, water depth, crest freeboard and crest width. The
W50 [kg] 0.175 0.088 0.020 0.0028 0.0069
crest freeboard and the crest width are two parameters which are
Mass density ρs 3000 2630 2610 2580 2700 expected to have little influence on the recession but large influence on
[kg/m3] overtopping which was also studied in the tests. The results on
Dn,50 [m] 0.0388 0.0323 0.0198 0.0103 0.0137 overtopping are presented in Lykke Andersen (2006).
fg = Dn,85/Dn,15 1.35 1.35 1.45 1.36 1.59
Due to the construction method used in the prototype, the initial
Length to width 1.98 1.96 2.01 1.88 2.04
ratio, l/b front and rear slopes of a reshaping breakwater are close to the natural
angle of repose of the material. Slopes between 1:1.1 and 1:1.5 are
typical. In all the present experiments the slopes were 1:1.25, which was
validity of these empirical formulae is questionable as found by Tørum close to the natural angle of repose for the armour materials tested. The
(1998) due to a large scatter. This scatter might partly be explained by crest level of the core was 2 cm below SWL, cf. Fig. 4.
missing parameters, e.g. berm elevation and front slope. In the present The properties of the materials are given in Table 1, and their grain
analysis the numerical method of Van der Meer (1992) and all of the listed size distribution curves are given in Fig. 5. Armour 3 corresponds in
simple regression models have been evaluated. almost all tests to a dynamically stable reshaping berm breakwater.
The largest armour stones, Armour 1, correspond to a non-reshaping
3. Objective of model tests berm breakwater. The ranges of parameters covered in the tests with
Armour 1–3 are listed in Table 2. 594 tests have been performed to
The objective of the physical model tests was to systematically cover the ranges of the listed parameters. For the study of scale effects
investigate the influence of a number of parameters related to berm additional 42 tests were performed using the smaller Armour 4. Of all
recession in order to establish a more reliable regression formula for the tests performed, 295 test are with statically stable berm break-
berm recession. waters (H0 b 2.7) and 400 are dynamically stable (H0 N 2.7). Armour
stones were placed with random orientation.
4. Tested structure designs and test programme The breakwater was only rebuilt when changing the initial
geometry or the wave steepness, i.e. a total of 150 times. All tests
The tested cross-section is shown in Fig. 4. The stone sizes in the were carried out with irregular waves generated from a JONSWAP
berm were varied in order to cover both statically and dynamically spectrum with a peak enhancement factor (γ) of 3.3 using a white
stable berm breakwaters as well as non-reshaping statically stable berm noise filtering method. The tested range of wave steepness (s0p) was

Fig. 5. Gradation curves.


362 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Table 2 The characteristic velocity is taken as g · Hm0 and the characteristic
Range of parameters in tests with Armour 1–3. length as Dn,50. ν is the kinematic viscosity. References to the tests with
Spectral peak wave steepness s0p =Hm0/(g/2π·Tp2) 0.010–0.054 regular waves of Dai and Kamel (1969) are often made when discussing
Wave height at toe of structure Hm0 [m] 0.064–0.164 the critical value of ReD for armour stability of conventional rubble mound
Water depth at toe of structure h [m] 0.24, 0.34 and 0.44 breakwaters. Dai and Kamel found that the critical Reynolds number for
Crest freeboard Rc [m] 0.08, 0.11, 0.14 and 0.17
armour stability was 3·104 as lower Reynolds numbers than this limit led
Crest width Gc [m] 0.17, 0.24, 0.31 and 0.38
Berm width B [m] 0.00, 0.20, 0.30, 0.40, 0.50, 0.65 to conservative results. Later on additional tests with irregular waves have
Water depth above berm hb [m] − 0.04, 0.00, 0.04 and 0.12 been performed by others and in most cases their results seem to be in
Front slope below berm cot(αd) 1.25 reasonable agreement with the work by Dai and Kamel. The critical value
Front slope above berm cot(αu) 1.25
of Dai and Kamel is typically exceeded in model tests with conventional
Rear slope cot(αr) 1.25
Stability number Ns = H0 0.96–4.86
rubble mound. However, for reshaping berm breakwaters the Reynolds
Stability index H0T0 16.8–163 numbers are typically lower due to the limitations of the size of most
Reynolds number for berm stones ReD Armour 1: 3.32·104 to 4.92·104 flumes on the scale and to the relatively smaller armour units used in
Armour 2: 3.13·104 to 3.91·104 berm breakwaters. A typical range of Reynolds numbers for small scale
Armour 3: 1.60·104 to 2.44·104
tests with homogenous berm breakwaters is 1·104 bReD b 5·104. As this is
somewhat lower than generally recommended for conventional rubble
mound breakwaters, it cannot be ruled out that viscous scale effects
influence the small scale berm breakwater results. In the present tests
with Armour 1–3, the Reynolds numbers are between 1.6·104 and
4.9·104. This means that for some tests the Reynolds number is lower
than recommended by Dai and Kamel.
Because it was not possible to perform large scale tests at Aalborg
University, it was chosen to check scale effects by performing additional
tests in even smaller scale using Armour 4, cf. Table 1. Tests with Armour
4 correspond to a length scale of approximately 1:1.9 of the tests with
Armour 3 leading to Reynolds numbers between 0.68·104 and 0.88·104.
However, comparing tests with Armour 4 to tests with Armour 3 gives
only a rough guess on the importance of scale effects. Due to the small
difference in grain size between Armour 4 and the scaled core material,
no core was present in the tests with Armour 4. The gradation factors for
Armour 3 and Armour 4 are not identical as Armour 3 contains more
large stones, cf. Fig. 5. The influence of this model effect on the berm
recession was not quantified.

6. Model test set-up and wave generation and analysis


Fig. 6. Tested sea states.
The model tests with head-on waves were performed in a wave flume
of dimensions 21.5 × 1.2× 1.5 m (length × width × depth) at Aalborg
0.01–0.054 with most of the tests close to 0.035, cf. Fig. 6, which University. Besides few metres of horizontal bottom the flume bed had
shows all the tested sea states. The wave height was in each test series a slope of 1:20, giving 0.60 m larger water depth at the wavemaker than at
gradually increased in steps with constant wave steepness. The the toe of the structure, cf. Fig. 7. Arrays of three resistance type wave
number of waves in each test was approximately 3000. gauges were placed both at the deep end and near the toe of the structure
to separate incident and reflected waves. Because the bottom is sloping in
5. Scale effects front of the toe and the waves therefore more non-linear, it is expected
that the separation based on linear theory is more reliable at the deep end
The armour stone Reynolds number (ReD) given by Eq. (3) is of the flume than near the toe. Nevertheless, the incident waves
commonly used to evaluate viscous scale effects in rubble mound calculated near the toe are used in the present analysis as large changes
breakwater model tests. in wave parameters due to breaking and shoaling could be expected.
The waves were generated by a hydraulically powered piston
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi wavemaker with steering signals generated by the AwaSys4 software
g · Hm0 · Dn;50 using the white noise filtering method, Aalborg University (2007a).
ReD = ð3Þ
ν Wave signals were filtered using an analog lowpass filter with a cut-off

Fig. 7. Longitudinal section of flume.


T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 363

frequency of 8 Hz. Moreover, a digital filter with cut-off frequencies of 1/3 the influence of front slope, water depth and berm elevation. Moreover,
and 3 times the peak frequency (fp) was applied to the wave signals. The the influence of wave direction, number of waves, gradation and shape
method of Mansard and Funke (1980) was applied to calculate the of stones, and wave skewness are included in the formula. The formula is
incident wave spectrum, and the SIRW method of Frigaard and Brorsen calibrated mainly using the present data with Armour 1–3, but also data
(1995) was used to calculate the time domain incident wave trains to of other researchers have been used to cover ranges/parameters not
study incident wave height distribution and surface skewness. Analyses of covered in the present test programme. Examples of typical test results
wave signal were performed with the WaveLab2 software package, in terms of berm recession are shown in Fig. 8.
Aalborg University (2007b).
After each test the reshaped profile was measured in a grid spacing 7.1. Influence of slope, water depth and berm elevation
of 1 × 1 cm with a computer controlled non-contact laser profiler
developed at Aalborg University. The profiles were averaged over the The reshaped profile generally has the following characteristics:
flume width but 10 cm at each side were discarded due to wall effects.
For tests with significant damage the recession was close to constant • In case of dynamically stable profiles the final reshaped profiles are, for
over the width accept near the side walls. For lower recessions some the same volume of material, approximately independent of the initial
variation in the recession was observed as the damage is due to front slope below the berm. Therefore, for identical berm width (B)
movement of few stones. The recession was in all cases determined the recession is less for a flatter initial front slope (αd) as the volume in
from the averaged profiles and at the berm level. In cases with little the berm is larger. For structures with only very limited reshaping the
damage the recession is difficult to define as it is just due to influence of the front slope on the stability might be different.
displacement of few stones on the berm front. In cases where there is • If the breakwater is not located in very shallow water, the slope of the
a significant rounding of the berm front corner the recession is defined lower part of the front approaches the natural angle of repose. The
by extending the profile from a level a little below initial berm level. vertical distance from SWL to this transition is called the step height hs,
see Fig. 1.
7. Test results and development of recession formula • Between SWL and the step the profile is slightly curved being
steepest at SWL. The average slope is around 1:5 in this part of the
A new formula to calculate berm recession is presented in this profile in case of dynamically stable profile but can be much steeper
section. The recession formula is based on some observations regarding for more stable profiles.

Fig. 8. Examples of measured profiles for the three armour types for identical or almost identical sea states.
364 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

• The front slope taken from SWL and up to the horizontal berm Afterwards the recession formula for Situation 1 is calibrated to establish
approaches the natural angle of repose. This is a rough estimate close the final recession formula.
to SWL but quite good for higher berms. Note that in Situation 2 the reshaped profile from Situation 1 has
been used between the step and SWL. Above SWL and below the step
Based on the above observations the change in recession due to a slope of 1:1.05 has been assumed which is based on the general
change in front slope, water depth and berm elevation is investigated. profile observations listed above. The profile was then shifted
The two characteristic initial designs and related reshapings as shown horizontally so that the total volume was conserved. A formula for
in Fig. 9 are considered. Initially, an approximate formula for the the shifted distance is derived in the following.
difference in recession between the two situations is derived based on In Fig. 10 the initial and reshaped profiles in Situation 1 and
the observed geometry of the reshaped profiles listed above and some Situation 2 are shown in the same figure. In this figure the volume
geometrical considerations. This is done because the reshaping in conservation is fulfilled for Situation 1, but not for Situation 2. In the
Situation 1 can be described more easily due to the simple geometry. following it is calculated the horizontal distance the profile has to be
Thus a more generally valid recession formula which reflects the shifted in Situation 2 in order to fulfill the volume conservation
observations of the reshaped profiles given above can be established. criteria A1 + A2 = A3. These three areas are defined in Fig. 10 and can
for cot(αd1) ≈ 1.05, hb1 = 0 and, h1 = hs be calculated as:
The initial profile in Situation 1 has the following characteristics:
Area in slope:
• Water depth is equal to the step height (h1 = hs).
• Berm is initially located at SWL (hb1 = 0). 2
hs
• The initial front slope below the berm is equal to the natural angle of A1 = · ½cotðαd2 Þ−1:05 ð4Þ
2
repose (cot(αd1) ≈ 1.05).

Situation 2 has the following characteristics: Area in berm:

• The water depth is larger than or equal to the step height (h2 ≥ hs) 2
hb2
• The berm is initially above or at SWL (hb2 ≤ 0) A2 = −Rec1 · hb2 − · ðcotðαd2 Þ−1:05Þ ð5Þ
• Arbitrary initial front slope below the berm, αd2. 2
Area at bottom:
An important assumption in the following is that the step height (hs)
1 2
does not change from Situation 1 to Situation 2, which seems quite A3 = ðh2 −hs Þ · ½ΔR1 −ðcotðαd2 Þ−1:05Þ · hs − · ðh2 −hs Þ · ½cotðαd2 Þ−1:05
2
reasonable as the incident waves are the same. The idea is that
ð6Þ
compared to Situation 2 a much simpler formula for calculating the
recession of the berm in Situation 1 can be established as the initial
geometry is very simple in Situation 1. In the following a method to In general the reshaped profile is slightly curved between SWL and
correct the recession from Situation 1 to Situation 2 will be given. the step. If a straight line is used as an approximation, it follows that

Fig. 9. Considered designs and deformation characteristics in Situations 1 and 2.


T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 365

Fig. 10. Situations 1 (solid) and 2 (dashed).

the movement of the profile at the bottom in Situation 1 (ΔR1) is equal known. Situation 1 has an initial profile with cot(αd1) = 1.05, hb1 = 0
to the recession of the berm (Rec1). As in reality the movement at the and h1 = hs. Due to this simple geometry it is expected to be
bottom ΔR1 is slightly larger than Rec1 (see Fig. 9), the expression much easier to describe the recession in Situation 1 than in Situation
Eq. (7) is used, where c1 is a constant with a value around 1.2. 2. As Situation 1 has not been tested in the model tests, it is chosen to
calibrate the expression of the recession in Situation 1 (Rec1) using
ΔR1 = c1 · Rec1 ð7Þ recession data for Situation 2 and the formula given in Eq. (13). The
formula (13) is expected to give most reliable results for steep
The reshaped profile is now shifted forward or backwards until the structures or alternatively flatter structures with quite some
volume balance is fulfilled, which is an approximation as some reshaping.
compaction can occur. The volume conservation leads to: The dimensionless recession in Situation 1 is assumed to be a
function of the wave direction (β), the number of waves (N), the grading
ðRec2;SWL −Rec1 Þ · ðh2 −hb2 Þ = A3−A1−A2 ð8Þ of the armour stones represented by the gradation factor (fg =Dn,85 /
Dn,15) and a stability index (H0, T0 and s0m) and the wave skewness. s0m
where Rec2,SWL is the recession at still water level in Situation 2, and is the mean wave steepness calculated with the wave parameters at the
not at the top of the berm, cf. Fig. 9. After inserting the expression in structure but using the deep water dispersion equation, i.e. s0m =Hm0 /
Eqs. (4)–(7) is obtained: (g/2π ·T20,1). The combined effect of these parameters has been difficult
21 2 1 2 3 to assess due to limited data, but a product of five functions is used in the
c1 · ðh2 −hs Þ + hb2 hb2 − h2
Rec2;SWL −Rec1 = · Rec1 + ½cotðαd2 Þ−1:05 · 42 2 5 following:
h2 −hb2 h2 −hb2
Rec1
ð9Þ Dn;50
= f ðβ; N; fg ; H0 ; T0 ; s0m Þ = fβ ðβÞ · fN ðNÞ · fgrading ðfg Þ · fH0 ðH0 Þ · fskewness ðb1 Þ

ð14Þ
It should now be taken into account that the recession is measured at
the top of the berm and not at SWL as assumed in Eq. (9). Based on the
These five functions and the function for the step height hs needed
previous mentioned observations of the profiles the slope of the reshaped
for Eq. (13) are discussed in the following.
profile above SWL is assumed to be close to the natural angle of repose
here taken as 1:1.05. With this assumption the change in recession due to
7.2. Influence of wave direction
the higher measuring point could for hb2 b 0 be expressed as (hb1 =0):

Rec2 −Rec2;SWL = hb2 · ½cotðαd2 Þ−1:05 ð10Þ The influence of the wave direction has not been studied in the
present tests, but Van der Meer (1988) assumes that all profile
parameters except the crest height are reduced by a factor cos(β),
Combining Eqs. (9) and (10) leads to:
where β is the angle of wave attack (β = 0 for head-on waves). Based
21 2 1 2 3
hb2 − h2
on the DHI (1995) data set, covering the range of β from 0 to 45°, the
c1 · ðh2 −hs Þ + hb2 4
Rec2 −Rec1 = · Rec1 + ½cotðαd2 Þ−1:05 · 2 2 + hb2 5 relationship (15) was found valid for the recession.
h2 −hb2 h2 −hb2

ð11Þ fβ = cosðβÞ ð15Þ

which could be rewritten as: The equation is valid for a trunk section where a supply of stone
material balances the longshore transport. Eq. (15) is in good
ð1 + c1 Þ · h2 −c1 · hs 1
Rec2 = · Rec1 + ½cotðαd2 Þ−1:05 · ½hb2 −h2  agreement with the assumptions of Van der Meer (1988).
h2 −hb2 2

ð12Þ 7.3. Influence of number of waves

The recession is made dimensionless with the stone size by Reshaping of berm breakwaters is expected to develop much faster
dividing by Dn,50 on both sides: than deformation of armour in conventional rubble mound break-
waters due to the smaller initial resistance to wave action. Van der
Rec2 ð1 + c1 Þ · h2 −c1 · hs Rec1 ½cotðαd2 Þ−1:05 Meer (1988, 1992) found that the reshaped profile parameters are
= · + · ½hb2 −h2 
Dn;50 h2 −hb2 Dn;50 2 · Dn;50 proportional to f(N) = N0.07 for step height (hs) and step length (ls),
ð13Þ and f(N) = N0.15 for crest height in case of dynamically stable
structures (H0 N 5). This indicates that changes in the crest are slower
As mentioned earlier c1 is a constant equal to approximately 1.2. than in the rest of the profile and continues even after a fairly long
With Eq. (13) it is possible to calculate the recession in Situation 2, duration of wave attack. For conventional rubble mound breakwaters
when the recession in Situation 1 (Rec1) and the step height (hs) is f(N) = N0.5 is found for the eroded area, Van der Meer (1988).
366 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

As Eq. (13) describes the relationship between equilibrium the recession formula.
profiles, it is expected that the influence of the number of waves is 8
not the same in Situations 1 and 2, as the eroded volume can vary <1 for fg ≤ 1:5
significantly. Limited data are available on the combined influence of fgrading = 0:43 · fg + 0:355 for 1:5 b fg b2:5 ð17Þ
: 1:43 for fg ≥ 2:5
slope, stability number, water depth and the number of waves.
Therefore, only an approximative equation describing the influence of
the number of waves is proposed. However, the influence of the The increase in stability for fg N 2.5 observed by other researchers is
number of waves (N) is relatively small for N larger than approxi- not taken into account. This is expected to be on the safe side, but it
mately 1000. Based on the data of Van der Meer (1988) the following should be noted that for a very wide grading the recession can be
is used: much more non-uniform along the breakwater.

8 −0:046 · H + 0:3 7.5. Step height


>
> N 0
>
< 3000 for H0 b5
fN =  0:07 ð16Þ To apply Eq. (13) a formula to calculate the step height (hs) is
>
> N
>
: for H0 N 5 needed. The step is not always as easy to define as shown in Fig. 1.
3000 Therefore, it has been chosen to define the step height in Fig. 11 as the
point where the slope gets flatter than 1:2. Curve fitting by hand to the
The reason for using 3000 waves as the reference value is that this present experiments gave the following expression:
value is used in the present tests to which the formula is fitted. During
−0:3
the model tests it was visually observed that after 500 waves only hs = 0:65 · Hm0 · s0m · fN · fβ ð18Þ
minor changes in the profiles occur when the profile is dynamically
stable while for more stable profiles the reshaping continues longer. The assumption that the step height is independent on the initial
The formula (16) is in good agreement with this. geometries is quite good as demonstrated in Fig. 11, where the
expression is evaluated against the present data.
7.4. Influence of stone gradation and shape
7.6. Influence of incident stability indices
The shape of a stone can be characterized by the ratio between the
length of the longest and shortest side (l/b). Frigaard et al. (1996) tested Van der Meer (1992) found for values of H0 larger than five that
four types of stone shapes, round (l/b = 1.0–1.5), normal (l/b = 1.5–2.5), the governing parameter is H0T0, which implies equal influence of
flat (l/b = 2.5–3.5) and a mix of these three (l/b = 1.0–3.5). The four wave height and period for given stone material. However, for smaller
stone types had exactly identical gradation curves. No measurable values of H0 it was observed that the period may have less influence
difference in the profiles for the four stone types was observed when the than given by the H0T0 parameter. Kao and Hall (1990) found that for
breakwater is infinite long or infinite material is available. However, in statically stable berm breakwaters the period had only little influence
oblique waves the longshore transport rates were 3–5 times higher for on berm recession.
the flat stones compared to the round stones. A breakwater with a finite The influence of the stability numbers is evaluated by inserting the
length exposed to oblique waves thus require a much higher supply of expression for Rec1 given by Eq. (13) into Eq. (14), which after
material for the flat stones compared to the round stones even though rearranging leads to:
the reshaped profiles are identical. " #
Hall (1991) performed physical model tests to study the influence Rec2 ½cotðαd Þ−1:05 h−hb 1
fH0 = − · ½hb −h · ·
of stone gradation and the percentage of rounded stones. He tested Dn;50 2 · Dn;50 ð1 + c1 Þ · h−c1 · hs fN · fgrading · fskewness

four different gradings with fg = Dn,85/Dn,15 in the range 1.35 to 5.4 ð19Þ
with 30% rounded stones. In addition Hall tested two gradations with
respectively 0% and 15% rounded stones. Hall found increasing where Rec2 is the recession in Situation 2, i.e. the actual measured
recession of the berm with increasing grading factor (fg b 3) and recessions. By calculating the right hand side of Eq. (19) for each
increasing percentage of rounded stones. A wider grading typically
decreases the porosity resulting in less energy dissipation in the berm.
Another reason for the decreased stability for a wide gradation could
be the redistribution of material along the slope where the larger
stones are deposited at the lower part of the profile. For identical Dn,50
this might lead to a smaller Dn,50 on the part of the profile close to the
water level for a wide gradation compared to a narrow gradation. For
a very wide grading (fg N 3) Hall found increasing stability for constant
Dn,50, which he explained by the very large stones dominating over
the decrease in energy dissipation.
Lissev and Daskalov (2000) performed physical model tests with
regular waves and 5 different stone gradations (fg) in the range 1.2 to
3.92 and found no difference in recession for the two stone gradations
fg = 1.2 and fg = 1.5, while for higher values of fg the recession was
significantly larger with maximum for fg ≈ 2.5.
Van der Meer (1988) found almost no difference in the profiles
when comparing fg = 1.25 and fg = 1.50, whereas the wide grading
with fg = 2.5 gave a much longer profile below SWL.
DHI (1996) performed tests with the two gradings fg = 1.38 and
fg = 1.79 and found more recession for the wide grading.
Thus all available data on the influence of the stone gradation factor
seem to be in good agreement, and the following factor is introduced in Fig. 11. Evaluation of Eq. (18) against present data.
T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 367

Fig. 12. Variation of fH0 with H0 for present data with hb ≤ 0. Fig. 14. Evaluation of Eq. (20) against present data with hb ≤ 0. Note that wave
skewness has not been taken into account in this plot leading to some deviations for
large values of H0.

For H0 N 5 the recession is assumed to be a function of H0T0. Based


on the only data for very large values of H0T0 (van der Meer (1988)
and Burcharth and Frigaard (1990)) is proposed:

fH0 = 0:05 · H0 T0 + 10:5 for H0 N ≈5 ð21Þ

It is important to note that H0 N 5 is outside the typical range used


for berm breakwaters. Between H0 = 3.5 and H0 = 5 there seems to
be more scatter than for lower and higher values. It is nevertheless
proposed simply to use the intersection between the two formu-
lae (20) and (21) as the transition point, which can be expressed
analytical as:

 
7:08
19:8 · exp − · s−0:5
0m −10:5
⁎ H0
T0 = ð22Þ
0:05 · H0

Eq. (20) is used for T0 ≥ T0⁎ and Eq. (21) is used for T0 b T0⁎, yielding:
Fig. 13. Variation of fH0 with H0T0 for present data with hb ≤ 0.
8  
>
< 19:8 · exp −7:08 · s−0:5 for T0 ≥ T0⁎
experiment and plotting these values against the corresponding H0 H0 0m
fH0 = ð23Þ
and H0T0 values, the variation of fH0 is evaluated, cf. Figs. 12 and 13. In >
:
0:05 · H0 T0 + 10:5 for T0 b T0⁎
this analysis the step height hs is calculated from Eq. (18) and fskewness
is disregarded and thus set equal to unity. The effect of the wave
skewness will be taken into account later. Instead of using the
combined effect of wave height, wave period and stone size as given 7.7. Influence of detailed wave characteristics
by H0T0, a different approach is taken as it was observed that a better
fit could be obtained. The recession is found to be proportional to In addition to the influence of the Hm0 wave height and the mean
s−
0m
0.5
for H0 b 3.5 and could be expressed as an exponential function of wave period (Tm) included in Eq. (23), there seems to be an influence
1/H0: of the more detailed characteristics of the waves, i.e. spectral width,
wave skewness and wave height distribution as discussed in Lykke
  Andersen (2006).
7:08 −0:5
fH0 = 19:8 · exp − · s0m for H0 b ≈ 3:5 ð20Þ Breaking waves seem to have an effect on the stability of a berm
H0 breakwater as there are several indications of more damage for
breaking waves, Lykke Andersen (2006). Hedar (1960) came to more
or less the same result from tests with regular waves on rock fill slopes
Fig. 14 shows the evaluation of Eq. (20). In the upper end of H0
as he found that waves breaking around half a wave length from the
more scatter is observed, which could partly be explained by wave
structure gave significantly more damage. This influence is included in
breaking. This is discussed and taken into account later in the paper by
the recession formula by fskewness given by:
inclusion of fskewness. It can easily be seen by comparing Figs. 13 and 14
that less scatter is observed especially for small recessions when using
2
Eq. (20) instead of a function of H0T0. fskewness = expð1:5 · b1 Þ ð24Þ
368 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

where b1 is the skewness of the surface elevation defined as: 8. Proposed recession design formula

N
 3 A summary of the proposed recession formula and the included
1 η −η
b1 = ∑ i ð25Þ factors is given below.
N i = 1 σðηÞ

" #
Rec 2:2 · h−1:2 · hs ½cotðαd Þ−1:05
After the inclusion of this factor in the recession formula, there = fhb · fH0 · · fβ · fN · fgrading · fskewness − · ½h−hb 
Dn;50 h−hb 2 · Dn;50
seems not to be any systematic scatter on the errors of the calculated
ð29Þ
recessions with respect to spectral width, wave height distribution
and wave groupiness.
If no information on the wave skewness is available, the Ursell where
number (Ur) might be used to predict the wave skewness. The Ursell 8
number is defined as: >
> hb
>
<1 for ≤ 0:1
Hm0
fhb =   ð30Þ
>
> h hb
H H · L2 >
: 1:18 · exp −1:64 · b for N 0:1
Ur = = ð26Þ Hm0 Hm0
2 · h · ðkhÞ2
8 · π2 · h3
fβ = cosðβÞ ð31Þ
Note that the Ursell number is sometimes defined without the
8 −0:046 · H + 0:3
factor 8π2. For irregular waves is here used the Hm0 wave height and >
> N 0
>
< for H0 b5
the peak wave length (Lp) calculated from the linear dispersion 3000
fN =  0:07 ð32Þ
relation. Ur N 1 indicates strong non-linearities. A correlation analysis >
> N
>
: for H0 N 5
resulted in Eq. (27) to estimate wave skewness. 3000

0:54 −0:3
b1 = 0:54 · Ur ð27Þ hs = 0:65 · Hm0 · s0m · fN · fβ ð33Þ

Hm0
Fig. 15 shows the evaluation of Eq. (27) against the present data. H0 = ð34Þ
Δ · Dn;50
7.8. Influence of a low berm sffiffiffiffiffiffiffiffiffiffiffi
g
T0 = · T0;1 ð35Þ
When the initial berm is located below SWL (not really a berm Dn;50
breakwater) the recession is much lower. This is taken into account by
the reduction factor given in Eq. (28).  
7:08
19:8 · exp − · s−0:5
0m −10:5
⁎ H0
8 T0 = ð36Þ
> hb 0:05 · H0
>
> ≤ 0:1
<1 for
Hm0
fhb =   ð28Þ 8  
>
> h hb >
< 19:8 · exp −7:08 · s−0:5 for T0 ≥ T0

>
: 1:18 · exp −1:64 · b for N 0:1 0m H0
Hm0 Hm0 fH0 = ð37Þ
>
:
0:05 · H0 T0 + 10:5 for T0 b T0⁎

It is important to note that when the berm is located below SWL


Hm0
(hb N 0) the breakwater could suffer severe damage much before s0m = g 2
ð38Þ
Rec = B. · T0;1

2
fskewness = expð1:5 · b1 Þ ð39Þ
8
<1 for fg ≤ 1:5
fgrading = 0:43 · fg + 0:355 for 1:5 b fg b2:5 ð40Þ
: for fg ≥ 2:5
1:43

Dn;85
fg = ð41Þ
Dn;15

In absence of wave skewness information the surface skewness


can be predicted from the Ursell number by:

0:54
b1 = 0:54 · Ur ð42Þ

Hm0
Ur = ð43Þ
2 · h · ðkp hÞ2


kp = ð44Þ
Fig. 15. Correlation between wave skewness (b1) and Ursell number (Ur). Lp
T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 369

where Lp is the peak wave length determined from the dispersion Kao (1991) and Tørum and Krogh (2000) on the basis of the new test
relationship for linear waves. results, cf. Fig. 16. The figure shows that the present recession formula
In some cases the recession formula predicts negative recession. fits the data very well and much better than the other three models.
This however should be interpreted as zero recession. The negative The largest deviations for the present formula, both above and below
value just indicates that the breakwater can resist even more severe the prediction line, is found for tests with the berm some few stone
wave attacks without recession of the berm. diameters below SWL, which is not a typical berm breakwater. When
Note that one of the assumptions when deriving the recession the berm is above SWL or far below SWL, the predictions seem very
formula was that hs ≤ h. If this condition is not fulfilled, then the reliable for all three armour configurations. This is expected also to
formula might be less accurate. include the Icelandic type of berm breakwater which are designed for
The water depth above the toe could be used for h in Eq. (29) if the small recessions. With the present method it is actually possible to
toe is wide enough to support the entire reshaped profile. This calculate the start of damage within very few stone diameters (cf.
procedure is probably valid unless the toe is very high. If the toe is not Fig. 17), however the scatter might be too large for using this method
wide enough to support the entire profile, then one can use a value for for designing multi-layer berm breakwaters as for this type only very
h between the water depth above the toe and the water depth without limited reshaping is allowed. An alternative approach for these
the toe. This could be based on an estimation of the extension of the structures might be to use the method of Van der Meer (1988) for
profile at the bottom, and then use the mean water depth over the statically stability to calculate the eroded area.
distance the profile extents. If the structure has no berm, the recession
is measured at the crest, hence hb = −Rc should be used in the
formula, as hb is negative when the berm is above SWL. 9.1. Comparison with the Van der Meer prediction method

9. Evaluation of recession formula Van der Meer (1988) performed an extensive experimental work on
dynamically stable profiles including gravel and rock beaches and berm
The performance of the present prediction method was compared breakwaters, which led to a calculation procedure for estimating the
to the performance of the methods by Van der Meer (1992), Hall and reshaped profile. Based on that work Van der Meer (1992) presented a

Fig. 16. Comparison of measured and calculated recession for present data for four different prediction methods.
370 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

Fig. 17. As Fig. 16 but focus on small Rec/Dn,50 values.

modified method that focused on dynamically stable berm breakwaters Van der Meer slightly overpredicts the deformation (see e.g. Fig. 8,
only. Test no. 563) but for the more resistant Armour 1 the predictions are
For each test the measured profile was plotted together with the very inaccurate as the damage is highly overpredicted (see e.g. Fig. 8,
profile calculated by the method of Van der Meer (1992) and the Test no. 443–448). This limitation is also mentioned by Van der Meer
recession calculated by the present proposed formula. Examples of as his method was developed for dynamically stable profiles, i.e. H0T0
such plots are given in Fig. 8, while a complete set of plots is available larger than approximately 70. However, the method of Van der Meer
in Lykke Andersen (2006). The method of Van der Meer (1992) (1992) gives the second best fit to the present data.
assumes conservation of the total volume of the breakwater.
However, it has been observed both during the present test
programme and by other researchers that some compaction can 9.2. Comparison with the Tørum and Krogh method
occur. Moreover, with the method of Van der Meer it is possible to use
the reshaped profile as input and then continue from this profile with Based on data from different laboratories Tørum has proposed a
another sea state. However, in the present study it was observed that number of recession formulae were the dimensionless recession is
such procedure resulted in less damage profile and worse agreement represented as a polynomia of H0T0 and in some cases also other
with measurements than if calculations always start with the initial. parameters. The one included in the present study is the formula by
Unless the water level varies significantly it is therefore suggested Tørum and Krogh (2000), which is the one included in PIANC (2003). The
always to start with the initial profile in the calculations, as done for data included in the fitting of their formula are from homogeneous berm
results in Figs. 8, 16 and 17. breakwater tests with the main part of the data being from the study of
In tests with Armour 3 the method of Van der Meer (1992) gives Andersen and Poulsen (1991), but also data from DHI (1995), Lissev
very reliable predictions of the reshaped profile (see e.g. Fig. 8, Test (1993), Tørum (1998) and various other SINTEF projects were included.
no. 510). In most tests with Armour 3 the breakwater corresponds to a The formula gives the dimensionless recession as a third order
dynamically stable breakwater. In tests with Armour 2 the method of polynomia of H0T0 together with correction factors for the influence of
T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 371

the water depth and the gradation of the armour stones. The influence with the measurements of Van der Meer with some scatter mainly on
of the front slope and the berm elevation are not included. the conservative side.
Fig. 16 shows that the method of Tørum and Krogh (2000) The structure tested by Burcharth and Frigaard (1990) is a
underestimates the recession for the main part of the present data. homogenous structure (no core) with a straight 1:1.5 front slope
This might partly be explained by a number of important parameters without a berm. The tests were carried out for very high stability
not included in their formula. This was actually found to be the case numbers (3.6 b H0 b 7.2), which is outside the typical range used for
for many of the data sets used in the present analysis, Lykke Andersen berm breakwaters. These data are used to fit the proposed recession
(2006). formula for high stability numbers, and as expected the agreement is
very good.
9.3. Comparison with the Hall and Kao prediction method Andersen and Poulsen (1991) performed during their diploma
thesis a parametric study on stability of berm breakwaters and carried
Hall and Kao (1991) investigated the influence of armour stone out approximately 100 model tests at DHI. In each test they
gradation and the percentage of rounded stones. Based on these simultaneously tested two different cross-sections separated by a
model test data they proposed a simple regression formula which plate in the middle of the flume. The test programme covered the
includes the influence of the stability number H0, the armour stone influence of wave height and steepness, berm width, berm elevation,
gradation (fg) and the percentage of rounded stones. The formulae of crest freeboard and the water depth. Waves were measured at the toe
Hall and Kao (1991) and Tørum and Krogh (2000) give in many cases of the structure and approximately 27 m from the structure. The
similar results and are both somewhat on the unsafe side for the waves were measured by three wave gauges at the deeper part for
present data, cf. Fig. 16. However, a little less bias and scatter for the reflection analysis and at the toe of the structure with no reflection
Hall and Kao method compared to the method of Tørum and Krogh analysis. It is not clear if it is the wave parameters at the toe or at
are observed. deeper water that are given in the report, but it is assumed to be those
at the toe as no reflection coefficients are given. The agreement with
9.4. Scale effects the present formula is quite good for most tests.
Hall (1991) tested a berm breakwater with a homogeneous berm
There is a tendency that slightly more recession is measured for with a front slope 1:1.25. Hall focused on the influence of the stone
Armour 4 compared to Armour 3, but only minor differences exist, cf. gradation and wave groupiness. For this data set it can be seen that in
Fig. 16. The difference could be due to a scale effect or the model effect case the stone gradation factor (fg) is smaller than 3, the agreement
that Armour 3 contains more large stones than Armour 4, cf. Fig. 5. with the proposed formula is very good, while a conservative bias is
Therefore, it could only be concluded that the results indicate only observed for fg = 5.4. This was also expected as the increase in
small or no scale effects. stability for fg N 2.5 was not taken into account in Eq. (17), as strong
variations along the breakwater might occur.
9.5. Comparison with data from other researchers The data set of Lissev (1993) needs special attention as a large bias is
observed not only for the present recession prediction method, which
The existing formulae and the present proposed formula were predicts much larger recession than was measured, but also for the other
evaluated against not only the present data, but also 11 other data sets methods tested. It was not possible to find any explanation for this
from other researchers. The proposed recession formula takes into deviation. The cross-section tested by Lissev is very similar to that of the
account also the wave skewness. However, information on this is only present tests with Armour 3, but in a 1.7 times larger scale. However, the
available for the present data. For all other data Eqs. (42)–(44) were deviations are much too large to be explained by viscous scale effects
used to estimate the wave skewness. In case of multi-layer berm only. This is also because results of other tests in the same range of
breakwaters the largest stone size (class I) is applied in the formulae if Reynolds numbers are well predicted by the proposed formula.
the cross-section is of a similar type as that indicated in Fig. 2. Aalborg University (1995) tested a very steep (1:1) straight rubble
The comparison to the data of other researchers is given in Fig. 18 mound structure with a rather high toe and a rather impermeable core.
and Table 3. In Table 3 the standard deviation (σ) is calculated as: Despite quite different from a normal berm breakwater design the
agreement with the present proposed formula is surprisingly good.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!2 DHI (1995) carried out three dimensional basin tests using long-
u
u 1 N Rec
meas −Reccalc crested head-on (β = 0°) and oblique waves (β = 15°, 30° and 45°).
σ=t ∑ ð45Þ
N i=1 Dn;50 The tested berm breakwater had initially a steep 1:1.1 slope below the
i
berm and a homogeneous berm. The present prediction method
predicts the recession in these tests with very good accuracy also for
For some of these data sets the waves have not been measured at the very oblique attack.
the toe, but instead on deeper water. In these cases the wave DHI (1996) tested berm breakwaters with a homogenous berm
parameters at the toe of the structure have been estimated from the and berm breakwaters of different multi-layered types, all having an
SWAN model, Holthuijsen et al. (2004). initial slope below the berm of 1:1.1. The agreement with the
Instanes (1987) and Tørum et al. (1988) tested berm breakwaters proposed recession formula is quite good for both types of structures.
with a homogenous berm and a 1:1.5 initial slope below the berm. The The data set of Porarinsson (2004) is for a multi-layer berm
SWAN method was used to estimate the wave parameters at the toe as breakwater where only very small recessions were studied. It can be
for both data sets the waves were not measured here but at deeper seen that the proposed formula predicts the start of damage very well,
water only. This might partly explain why the data set of Tørum et al. but further research is needed to investigate if that will be the case for
(1988) shows more scatter than most of the other data sets. multi-layered berm breakwaters in general.
The data set of Van der Meer (1988) covers many different kind of
structures, but only few tests were done with berm breakwaters, all
with an initial front slope of 1:1.5. In addition to the normal tests 10. Uncertainty of recession formula
where the berm is above SWL, some tests were carried out with a
berm below SWL. Up to 10,000 waves were used with intermediate In this section, the uncertainty of the proposed formula is
profiles measured at several steps. The predictions made by the evaluated. Uncertainty of design formulae has to be taken into account
present proposed formula are in most cases in very good agreement in probabilistic design.
372 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

Fig. 18. Comparison of calculated recession from the present formula and measured recession data from other researchers.

Table 3 shows the standard deviation of the dimensionless to quite similar conclusions as seen from Table 3, except for the Van
recession predictions for the four methods tested and for the der Meer (1992) method, which gives large relative errors as this
individual data sets. It seems that the proposed method is better method is not valid and give bad predictions for cases with little
than the three other methods. A study of the mean relative error led damage.
T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374 373

Table 3
Standard deviation between calculated and measured dimensionless recession (Rec/Dn,50) for five methods and for different data sets.

Data Present formula Hall and Kao (1991) Van der Meer (1992) Tørum and Krogh (2000) Tørum (1998)

Present data 1.91 5.73 3.79 7.18 4.41


Instanes (1987) 2.36 2.84 2.79 5.31 2.79
Tørum et al. (1988) 6.12 6.39 6.15 8.27 6.60
Van der Meer (1988) 3.53 9.67 4.12 18.83 15.43
Burcharth and Frigaard (1990) 2.89 21.09 2.23 79.68 51.04
Andersen and Poulsen (1991) 3.38 4.83 4.31 4.09 2.63
Hall (1991) 4.06 2.61 2.68 6.12 5.56
Lissev (1993) 9.60 8.96 9.40 8.46 10.56
Aalborg University (1995) 3.96 1.72 – 7.20 3.51
DHI (1995) 1.84 2.93 3.04 2.72 2.66
DHI (1996), homogeneous 2.56 1.91 3.80 3.31 2.67
DHI (1996), multi-layer 1.21 1.53 2.15 3.35 1.59
Porarinsson (2004) 0.16 0.25 0.50 0.67 1.48
Overall 2.98 5.60 3.96 9.53 6.69

The confidence band shown in Fig. 19 was found by calculating the As the derived recession formula is based on averaged profiles, it
standard deviation of the residuals within 7 intervals of Recmeas/Dn,50. should be noted that there could be a non-uniform distribution of
Assuming Gaussian distributed residuals, the upper and lower 95% damage along the breakwater, which also should be taken into account.
confidence levels are 1.645 times the standard deviation. This resulted This is expected especially to be important for small recessions or wide
in the following equations for the 90% confidence band on all data, i.e. graded armour stones.
only 10% of the data are outside this band:
11. Conclusions
Rec95%
calc Rec50%
calc
= 1:080 · + 4:00 ð46Þ
Dn;50 Dn;50 A new empirical formula for calculating berm recession, which was
found to be superior to any other method tested, has been presented.
The formula has been calibrated for a large interval of stability numbers
Rec5%
calc Rec50%
calc
= 0:909 · −2:95 ð47Þ ranging from stable to dynamically stable structures.
Dn;50 Dn;50
The new formula includes a number of parameters not included in
existing formula among a parameter describing wave skewness. This
For the present data the following much more narrow 90% was included as breaking waves were found to give significantly more
confidence band is observed: damage than non-breaking waves.
The presented formula performs very well for the main part of the data
95% 50%
Reccalc Reccalc sets used to evaluate the formula. However, the scatter on the data from
= 1:028 · + 2:38 ð48Þ
Dn;50 Dn;50 some few researchers was found to be much greater than for the main
part of the data sets. The reason for this scatter could not be explained.
Rec5% Rec50% The uncertainty of the formula has been documented and should
calc calc
= 0:961 · −2:10 ð49Þ be taken into account in design.
Dn;50 Dn;50
Scale effects on reshaped profiles seem not to be a major problem
and are most likely negligible for ReD N ≈ 3 · 104 as found for static
stable rubble mound breakwaters.

Acknowledgements

The study was partly financed by the CLASH project EVK3-CT-


2001-00058 within the EESD programme of the Fifth Framework
Programme of the EU. The financial contribution of the European
Community is acknowledged.

References
Aalborg University, 1995. Heraklion Airport. Hydraulic Model Tests of Slope Protection.
Department of Civil Engineering, Aalborg University, Denmark.
Aalborg University, 2007a. AwaSys 5 homepage. http://www.hydrosoft.civil.auc.dk/
AwaSys.
Aalborg University, 2007b. WaveLab 2 homepage. http://www.hydrosoft.civil.auc.dk/
wavelab.
Andersen, F.J., Poulsen, C., 1991. MAST Berm Breakwaters. Danmarks Ingeniørakademi,
Lyngby, Denmark. Thesis carried out at DHI.
Archetti, R., Lamberti, A., 1996. Parametrizzazione del profile di frangiflutti berma. Proc.
Congresso AIOM, Padova.
Burcharth, H.F., Frigaard, P., 1990. On 3-dimensional Stability of Reshaping Break-
waters. Proc. 22th International Conference on Coastal Engineering, Delft.
Burcharth, H.F., Frigaard, P., 1988. Reshaping breakwaters. On the Stability of Roundhead
and Trunk Erosion in Oblique Waves. Berm Breakwaters: Unconventional Rubble
Mound Breakwaters, Ottawa, Canada.
Dai, Y.B., Kamel, A.M., 1969. Scale Effect Tests for Rubble Mound Breakwaters. U. S. Army
Engineer Waterway Experiment Station, Corps of Engineers, Vicksburg, Mississippi,
Fig. 19. Evaluation of present recession formula against all available data. Research Report H-69-2.
374 T. Lykke Andersen, H.F. Burcharth / Coastal Engineering 57 (2010) 359–374

DHI, 1995. EU MAST II Berm Breakwater Structures. Report on the Three-Dimensional Mansard, E.P.D., Funke, E.R., 1980. The Measurement of Incident and Reflected Spectra
Model Tests. Draft Report. Danish Hydraulic Institute (DHI). Using a Least Squares Method. Proc. 17th Coastal Engineering Conference, Sydney,
DHI, 1996. EU MAST II Berm Breakwater Structures. Influence of Permeability and Stone Australia.
Gradation. Draft Report. Danish Hydraulic Institute (DHI). PIANC, 2003. PIANC MarCom Report of Working Group No 40. State-of-the-Art of
Frigaard, P., Brorsen, M., 1995. A time-domain method for separating incident and Designing and Constructing Berm Breakwaters. International Navigation Associa-
reflected waves. Coastal Engineering 24. tion. PIANC General Secretariat, Brussels, Belgium.
Frigaard, P., Jensen, M.S., Hald, T., 1996. Berm Breakwater Design – Influence of Rock Porarinsson, J. S., 2004. Porlákshöfn-harbour. Aalborg University. MSc. Thesis.
Shape. Hydraulics & Coastal Engineering Laboratory, Aalborg University. Sigurdarson, S., Jacobsen, Smarason, Bjordal, Viggosson, Urrang, Torum, 2003. Sirevåg
Hall, K.R., 1991. Report #1 – Hydraulic Model Test Results – A Sudy of the Stability of Berm Breakwater, design, construction and experience after design storm. Coastal
Berm Breakwaters. Queen's University Coastal Engineering Laboratory. Structures 2003. ASCE.
Hall, K.R., Kao, S., 1991. A study of the stability of dynamically stable breakwaters. Sigurdarson, S., van der Meer, J.W., Burcharth, H.F., Soerensen, J.D., 2007. Optimum
Canadian Journal of Civil Engineering 18, 916–925. Safety Levels and Design Rules for the Icelandic-type Berm Breakwater. Coastal
Hedar, P.A., 1960. Stability of Rock-Fill Breakwaters. Akademiförlaget-Gumperts, Structures, Venice. ASCE.
Göteborg. Tørum, A., 1998. On the stability of berm breakwaters in shallow and deep water. Proc.
Helgason, E., Burcharth, H.F., 2005. On the use of high-density rock in rubble mound 26th International Conference on Coastal Engineering. ASCE, Copehagen, Denmark,
breakwaters. International Coastal Symposium, Höfn, Iceland. pp. 1435–1448.
Holthuijsen, L.H., Booij, N., Ris, R.C., Haagsma, I.G., Kieftenburgh, A.T.M.M., Kriezi, E.E., Tørum, A., Krogh, S.R., 2000. Berm Breakwaters. Stone Quality. SINTEF Report No. STF22
Zijlema, M., Van der Westhuysen, A.J., 2004. SWAN User Manual — SWAN Cycle III A00207. SINTEF, Civil and Environmental Engineering.
version 40.31. Delft University of Technology. Tørum, A., Næss, S., Instanes, A., Vold, S., 1988. On Berm Breakwaters. Proc. 21st Coastal
Instanes, A., 1987. Molostabilitet. Institut for Marin Byggteknikk, Trondheim, Norway. Engineering Conference, Spain, 1988.
Kao, J.S., Hall, K.R., 1990. Trends in Stability of Dynamically Stable Breakwaters. Proc. Tørum, A., Krogh, S.R., Bjørdal, S., 1999. Design Criteria and Design Procedures for Berm
22th International Conference on Coastal Engineering, Delft. Breakwaters. Proc. of Coastal Structures ’99.
Lissev, N., 1993. Influence of Core Configuration on the Stability of Berm Breakwaters. Van der Meer, J.W., 1988. Rock Slopes and Gravel Beaches under Wave Attack. Delft
Experimental Model Investigations. Report No. R-6-93. Department of Structural Hydraulics Communication No. 396.
Engineering, University of Trondheim, The Norwegian Institute of Technology. Van der Meer, J.W., 1992. Stability of seaward slope of berm breakwaters. Coastal
Lissev, N., Daskalov, K., 2000. Berm Type Breakwater — An Alternative Solution for New Engineering 16, 205–234.
East Breakwater for Port of Burgas. Varna conference, 2000. Van Gent, M.R.A., 1995. Wave Interaction with Permeable Coastal Structures. Delft
Lissev, N., Tørum, A., 1996. Influence of the core configuration on the stability of berm University of Technology.
breakwaters. Proceedings of the 25th International Conference on Coastal Engineering,
Orlando, Florida.
Lykke Andersen, T., 2006. Hydraulic Response of Rubble Mound Breakwaters. Scale
Effects – Berm Breakwaters. PhD Thesis, Department of Civil Engineering, Aalborg
University, Denmark.

You might also like