Analysis of rubble mound breakwaters

J. D. METTAM, Scott Bertlin

SYNOPSIS. Working Group 12 was set up by Permanent

Technical Committee II of PIANC to consider the analysis of
rubble mound breakwaters with a view to achieving a better
understanding of safety aspects. This paper provides a
very brief introduction to the work of the group. A
practical way of evaluating safety has been developed (for
use particularly with Hudson's formula or Van der Meer's
formula for design of main armour stability) which enables
a theoretical design to be evaluated more realistically in
respect of varying probabilities of failure within speci-
fied design life. Its value has been demonstrated by
examples based upon data collected. Methods are outlined
for using partial coefficients to prepare preliminary
designs, and then use hydraulic model testing to refine the
preliminary design and provide the basis for final designs.

INTRODUCTION
1. After a considerable number of large breakwaters had
failed or suffered severe damage, a working group estab-
lished by PIANC's Permanent Technical Committee II produced
a report, published as a supplement to Bulletin No 48
(1985), on the stability of rubble mound breakwaters. This
summarised important details of design and construction of
a large number of rubble mound breakwaters and concluded,
inter-alia, that "it is not possible at the present time to
determine risk with a satisfactory degree of accuracy".
2. In late 1986, PTC II set up a further group, Working
Group 12, to consider the Analysis of Rubble Mound
Breakwaters. The terms of reference were as follows:

Terms of Reference
3. The immediate need for designers of breakwaters is to
achieve a better understanding of the overall safety
aspects in the design of rubble mound breakwaters. This is
important because the existing design tools are still very
incomplete.
4. Practical approach of the group
(i) Identify and list the parameters to be con-
Coastal structures and breakwaters. Thomas Telford, London, 1992 533

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 COASTAL STRUCTURES AND BREAKWATERS

sidered in the design of rubble mound break-

waters .
Evaluate the relative importance and the quality
of existing available knowledge related to these
parameters.
(ii) Evaluate the safety aspects (risk of failure)
and propose ways of dealing with the safety
problems in practical ways. This should include
a check of the proposed safety guidelines
against the behaviour of selected existing
breakwater structures.

MEMBERS OF THE WORKING GROUP

5. The working group comprised the following members:-

MR J D METTAM - Chairman UNITED KINGDOM

Scott Bertlin

PROF. H F BURCHARTH - Secretary DENMARK

Aalborg University

MR N W H ALLSOP UNITED KINGDOM

Hydraulics Research Limited

MR P BONAFOUS (retired 1990) FRANCE

Port Autonome du Havre

DIPL. ING M BORCHERT FEDERAL REPUBLIC OF GERMANY

Prof. Dr. Lackner & Partners

MR J E CLIFFORD UNITED KINGDOM

Consultant

DR T ENDO JAPAN
Applied Hydraulic Laboratory Nippon Tetrapod Co Ltd

PROF D FOSTER AUSTRALIA

Unisearch Ltd Tasmania

PROF L FRANCO ITALY

Politecnico di Milano, Dipartimento di Ingegneria Idraulica

PROF ENG P A HEDAR SWEDEN

PA Hedar AB

M. X LE BARS (from 1990) FRANCE

Port Autonome du Havre

MR H LIGTERINGEN THE NETHERLANDS

Frederic R Harris Holland BV

DR E P D MANSARD CANADA
National Research Council of Canada Hydraulics Laboratory

534

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 METTAM
MR C A R M PITA PORTUGAL
WW Consultores de Hidraulica E Obras Maritimes LDA
MR L STADLER ISRAEL
Consulting Engineer
IR D VANDENBOSSCHE BELGIUM
Haecon NV
MR L VAN DAMME BELGIUM
Ministry of Public Works, Coastal Department
DR JENTSJE VAN DER MEER THE NETHERLANDS
Delft Hydraulics

DR R W WHALIN U.S.A.
US Army Corps of Engineers, Waterways Experiment Station
MR JOOP A ZWAMBORN SOUTH AFRICA
Division of Earth, Marine and Atmospheric Science and
Technology, CSIR

METHOD OF WORKING
6. The working group decided to develop the practical
application of risk analysis in the design of rubble mound
breakwaters by using partial coefficients. Tasks were
delegated to six subgroups.
7. Subgroup chairmen and terms of reference were as
follows:-
Subgroup A Theories, Formulae
Chairman - Mr N W H Allsop
Identification of theories, formulae, empirical data from
model tests etc. which give the functional relationship
between the structural response and the environmental and
structural parameters. Evaluation of the related statisti-
cal uncertainty.
Subgroup B Environmental data and extreme events
Chairman - Prof H F Burcharth
Evaluation of uncertainty related to environmental data
including estimated extreme events.
Subgroup C Risk Analysis in Breakwater Design
Chairman - Dr J W van der Meer
Description and evaluation of practical ways of dealing
with safety aspects (risk of failure) based on the
uncertainties related to the parameters involved in the
design.
Level I methods (safety factors, partial coefficients)
Level II methods (safety index, normal distributions)
Level III methods (full consideration of probability
distributions)
Subgroup D Deviations from Design and Reliability of
Construct ion
535

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 COASTAL STRUCTURES AND BREAKWATERS

Chairman - Mr M Borchert
Identification (quantitative) of typical tolerances related
to construction procedures.
Subgroup E Investigation of Selected Cases
Chairman - Mr H Ligteringen
Selection of existing breakwater structures (of types as
given in Fig. 3.1) for which reliable data on environmental
conditions, structural parameters and structural response
(damage/no damage) are already available or can be made
available. (Sub group E also analysed all selected cases
using partial coefficients as far as possible).
Subgroup F Determination of Partial Coefficients
Chairman - Prof H F Burcharth
Develop and propose safety guidelines and check (calibrate)
these against the documented behaviour of exist ing break-
waters.
8. The Main Report of the Working Group will be
published by PIANC next year bound together in one volume
with the full texts of the six subgroup reports.

WORK OF THE SUBGROUPS

9. The amount of work carried out by the subgroups was
most impressive. Each subgroup chairman had help from a
small number of members of the Working Group, and in
addition was able to call on top experts from all over the
world who gave freely of their knowledge. This high level
of international co-operation is a feature of the work of
PIANC which is difficult to appreciate without first hand
experience.
10. The result will be seen in the reports. While the
work of each of the Subgroups A-D was designed to provide
inputs to the analysis carried out by Subgroup F, and
Subgroup E has served chiefly to provide a pract ical check
on the method developed by Subgroup F, each of the subgroup
reports has become a valuable document in itself.
11. The relationship between the work of the different
subgroups is very briefly described below and illustrated
in Fig 1.
12. Subgroup A provided a review of the state of art of
relevant analytical formulae. This has identified those
failure mechanisms which are sufficiently well described by
formulae to be used as a basis for application of the
partial factor approach. The uncertainty attached to each
of these formulae was also investigated and expressed where
possible in terms of standard deviations.
13. Subgroup B studied the uncertainty related to en-
vironmental data as a function of methods of data collect-
ion, including estimates of standard deviations for differ-
ent methods of determining significant wave heights and sea
level variations. The estimation of uncertainty on extrapo-
lation of design wave heights corresponding to extreme
return periods was also included.

536

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 METTAM

Theories, Formula
related uncertainty

B Uncertainty of
environmental data,
extreme events

Develop concept
of partial
Risk analysis, coefficient system
ways of dealing
with uncertainties
Optimisations for
calculation of
Deviations from partial coefficients
design, reliability
of construction

E Investigation of selected cases Analysis using

Data collection and analysis partial coefficents

MAIN REPORT Summaries Design implications

of the work Future work
of subgroups Conclusions and
recommendations

Fig 1 Method of working

14
Subgroup C reviewed different methods of determining
the risk of failure. This led to the selection of the
partial coefficient approach as a practical and relatively
simple method of breakwater design, and the decision t.o
derive values for the partial coefficients from higher
level calculations incorporating the uncertainties in
environmental data and construction obtained by Subgroups B
and D, as well as uncertainties in the design formulae,
derived by Subgroup A.
15. Subgroup D examined the importance of deviations from
the design which will be found in the breakwater as built
so that allowance for these uncertainties could also be
made in the higher level analyses carried out to determine
partial coefficients.
16
* Subgroup E investigated the safety level of a variety
of existing breakwaters with 10-20 year life for selected

537

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 COASTAL STRUCTURES AND BREAKWATERS

breakwater types. Although data collection started, with

the work of Subgroups A-D, at an early stage in the work of
the group the results from Subgroup E were not, like those
of Subgroups A-D, inputs to the work of subgroup F, but
were designed to provide a means of "calibrating" the
theoretical development of partial coefficients. Results
are limited to the stability of main armour.
17. Subgroup F carried out an impressive amount prob-
abilistic computation to develop and optimise the system of
partial coefficients. This is described fully in the
accompanying paper by Professor H F Burcharth.
It is therefore necessary only to state here that this work
constitutes a fundamental advance in the design of rubble
mound breakwaters, and to acknowledge the considerable
input of time of staff and computers from the Universities
of Aalborg and Delft without which it would not have been
possible to conclude the work; financial aid from the Dutch
Rijkswaterstaat and the Danish Technical Scientific Council
is also gratefully acknowledged.

BRIEF DESCRIPTION OF PROPOSED DESIGN PROCEDURE

18. Within the scope of this paper it is possible only to
give the very briefest account of the nature of the
proposed design procedure. This will be described in
relation to Hudson's formula for main armour.
19. Written in its original format, but using the
notation of the Working Group 12 report, Hudson's formula
is: n TT3

M = ^-5 (i)
K D A 3 Cot a
(Notations are listed at the end of this paper)
This can be reorganised into a response function:
M
g = ZA ( ^^ j' -H (2)
Failure occurs for g<0 . The right hand term H is a load
parameter, while the other term comprises resistance
parameters.
20. In our analysis the parameter Z is a stochastic
variable representing the reliability of the formula.2
has a mean value of 1.0. and normalised standard deviation
'z = 0.18. K D is treated as a deterministic coefficient of
damage while all the other parameters involved are taken as
stochastic variables.
21. Having decided for simplicity to introduce only two
partial coefficients Y z and Y H (both greater than unity) the
response function can be developed into a design equation
Z A fM K c
A I " TH H (3)
J
3
or G = . A D n <KD cotCC) - yH . H (4)

Jz

538

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 METTAM
22. The design condition is satisfied by G >0. It will
be seen that the product of the two partial factors
Yz Y H represents a simple factor of safety in the design
cquation.
23. As described in Professor Burcharth's paper, formulae
have been selected to calculate Yz a n ^ YH . These include
coefficients whose optimum values for various design
formulae have been determined by higher level analysis
using standard deviations for all the stochastic variables.
24. Adopting those values the partial factors for any
design can be calculated in terms of probabilities of
failure within a given design lifetime. Preliminary
designs may then be prepared using Hs values for storms of
the appropriate return period, together with published K D
values such as those in the Shore Protection Manual.
Examples of this system, applied to the case of Torshavn
Breakwater are shown in Table 1.

Case 1 2(a) 2(b) 3 4

Life in years 25 50 50 50 50

Damage level Moderate Moderate Moderate Moderate None

S-Van der Meer 6 6 6 2
Kp-Hudson 3.4

Failure probability Pf 0.4 0.7 0.7 0.1 0.1

Partial factors
YHs 1.10 1.00 1.00 1.31 1.37
Yz 1.02 1.01 1.01 1.06 1.06

Armour size
Dn50-m 1.20 1.35 1.35 1.64 2.14
Av, Mass-t 4.8 7.0 7.0 12.3 27.4

25. The preliminary design must then normally be subjected

to thorough hydraulic model testing, both to refine the
main armour design and to check aspects not covered by the
design formula. At this stage it will be necessary to
consider the effect on the test procedure of having
introduced partial factors in the design equation.
26. Current practice is to build a scale model represent-
ing the intended design as accurately as possible. With no
partial factors this means that the response function is
being tested directly and that the degree of damage
represented by the chosen K Dvalue should be expected to be
reproduced when testing under the design wave.
27. With a design incorporating partial factors the
armour size Dn will have been increased by the factor of
safety Yz YH - The stability is thereby increased and the
design wave will not produce the damage corresponding to
the chosen K D value. To confirm that the design has the

539

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 COASTAL STRUCTURES AND BREAKWATERS

intended margin of safety it will be necessary either to

increase the wave height (load parameter) or to reduce the
armour size Dn, or some other resistance parameter, in
proportion to Y 2 . YH . Various options will be discussed
in the report of the Working Group. The simplest solution
for any particular case will vary depending, for example,
whether waves are depth limited or not.
28. Having completed model testing the response function
for the specific conditions of the breakwater may be
modified and used to revise the design equation for the
final design.

CONCLUSIONS
29. The more rational assessment of probabilities made
possible by the partial coefficient system will be a
valuable tool, but it can of course only be applied to
failure mechanisms for which design formulae are available.
It will not take into account other forms of failure such
as geotechnical failure or seabed scour leading to in-
creased wave attack.
30. Improvements are proposed in the statistical treat-
ment of uncertainty in predicting extreme wave heights;
this will help the design but will not obviate the need for
further improvements in wave recording and use of wind data
to supplement inadequate observations.
31. A design prepared using the proposed method of
partial coefficients will result in a more objective
assessment of probabilities of failure of the structure
during its design life than has previously been possible.
This can be used to assess economic damage levels and will
not necessarily result in more expensive design although
this will depend upon the nature and consequences of
failure.
32. As will be emphasised in the Report to be published
by PIANC next year, the results which have been achieved
should not be regarded as a code of practice for design of
breakwaters. Much more work is required to reach that
stage; nevertheless the new system is seen as a major step
forward, and engineers are urged to use the method to
examine existing breakwaters and new designs. This will
give useful insights and, with further research, will
enable more logical design codes to be developed in future.
33. One problem, which has perplexed engineers using the
Shore Protection Manual to prepare designs using the Hudson
formula, has been the recommendation in the 1984 edition to
use Hi/io instead of Hs. In conjunction with K D values
previously adopted with Hs this resulted in the effective
introduction of a factor of safety of 1.27, doubling the
required weight, of armour. Considered in relation to the
partial coefficient system proposed by Working Group 12
this can be seen as a prudent response to the recent
breakwater failures but it is now considered more rational
to revert to the use of Hs as the design wave height (H in

540

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 METTAM
equation 14) and use the system of partial coefficients to
calculate the acceptable probabilities of failure.
34. With armour units which can break leading to rapid
and progressive failure it is likely that it will be
economic to design for significantly lower probabilities of
failure than have been adopted in recent years and the new
method may lead to results at least as conservative as the
use of H 1/10 However in many simpler cases the partial
factors will be found to be little more than unity. For
example, with simple rock armouring, with no great diffi-
culties in repair, the economic design may not differ
significantly from the earlier designs using Hudson's
formula with Hs.
35. The British Code (BS 6349 Part 7 ) , which has been
circulated for comment and has recently been published, has
adopted a different approach by recommending the use of Hs
and adopting K D values from the Shore Protection Manual for
rock armour but introducing more conservative K D values for
certain types of concrete armour as a method of introducing
a margin of safety. This is a more rational approach than
using H l/iofor all types of armour but it should be seen as
a temporary expedient to be replaced by the partial
coefficient system when this has been further developed.
NOTATION
l
Nominal armour size, defined (M/pa) 3
G Design equation
g Failure or response function
H Wave height, trough to crest
H
s Significant wave height, average of
highest 1/3
Mean height of highest 1/10 waves
Empirical damage coefficient in
Hudson's formula
M Mass of armour unit
Z Stochastic variable signifying the
r e l a i b i l i t y of a response function
a (alpha) Breakwater front slope, angle to
horizontal
y (gamma) Partial coefficient
YHS YZ Partal coefficients on Hs and Z
pa (rho) Mass density of armour units
pw Mass density of seawater
A (delta) Relative density (pa/pw -1)

541

Downloaded by [ RICE UNIVERSITY] on [12/09/16]. Copyright ICE Publishing, all rights reserved.