Publications On Hydraulic and Geotechnical Problems

.*! *' •" .
hydraulic and geotechnical

problems
- BfitlOTM'-ik
.O
: : r B
# * * *
-f Ti
"•* **• .-"j?s'

\A - --.
-<**
# ? * *?$
Vjf1,
t. • " k
Ministry of Transport, J2 V.i >. (.-lOJnatË;
fc\ Road and Hydraulic Engineering DiVtsTt»'* -
Directorate-Ceneral fo e " ' V'tics anr VtóSfeF Wa
«.f*
•' 'A " • •

(2* ex.)
Collected papers include the original contributions to

this volume, the ones submitted to some societies and the reprints
from some publications.
The views and conclusions in this book are those of the

authors and do not necessarily represent the views of the Rijkswater-
staat and other institutions involved. Neither the authors and their
organizations assume any liabilities with respect to the use of, or
for damage resulting from the use of, any information disclosed in
this book.
Limited copies of this book can be obtained at cost from

the Road and Hydraulic Engineering Division of the Rijkswaterstaat.
All correspondence and applications should be addressed to:
Rijkswaterstaat,
Road and Hydraulic Engineering Division,
Van der Burghweg 1,
P.O. Box 5044, 2600 GA Delft,
The Netherlands
Tel. +31-15-699111
Fax +31-15-611361
Telex 38043 NL
Reproduction of this publication, wholy or in part, is allowed provi-

ded that the source is duly acknowledged.
Publication nr. : I ^ BLIOTHEEK I

„„ „ r _ f Dienst Weg-en Waterbouwkunde »
W-DWW-93-253 I Van der Burghweg f
f Postbus 5044, 2600 GA Delft I
Collected Papers , 1989-1993
Subjects
1. Design codes and safety factors
2. Closures and storm surge barriers
3. Bank protection
4. Geotechnical stability of dikes
5. Revetments
6. Coastal protection and management
7. Scour and sand-water mixtures
8. Contaminated soils
References
Preface
The Public Works Department (Rijkswaterstaat = RWS) manages and

maintains roads, inland waterways and water defence structures. The
Road and Hydraulic Engineering Division (DWW) conducts research in the
fields of civil engineering and and the environment for the purpose of
advising the regional directorates of the Public Works Department on
construction, management and maintenance.
lts aims in so doing are as follows:
- to improve the quality of the public works infrastructure;
- to limit the costs of design, construction, management and
maintenance;
- to develop and apply systems for determining the technical
condition of public works structures;
- to develop models for studying the behaviour of structures;
- to apply the techniques of information science to civil
engineering;
- to find methods to describe the effects of civil engineering
projects on the environment (environmental impact reports).
The DWW is divided into four branches for the purpose of carrying out
its technical responsibilities. These are:
CONSTRUCTION MATERIALS,
ROAD ENGINEERING,
HYDRAULIC ENGINEERING and
ENVIRONMENT.
Hydraulic Engineering
The Netherlands has a tradition of hydraulic engineering stretching

back centuries as a consequence of its unceasing endeavours to protect
the land from the water and, from time to time, reclaim land that has
been lost. No matter how familiar the subject may be, it is seldom
realised that hydraulic engineering is a broad field of study covering
hydraulics, soil mechanics, the dynamics of solid and floating struc-
tures, groundwater flows, risk analysis and foundation techniques.
Research in these areas is carried out by the HYDRAULIC ENGINEERING
Branch. The Branch also advises on the design, construction, manage-
ment and maintenance of hydraulic engineering works and draws up
standards, guidelines and recommendations relating to hydraulic
engineering.
This book collects the selected publication over the period 1989 to
1993 and is meant to give an idea of the hydraulic and geotechnical
expertise of the Hydraulic Engineering Division.
Krystian W. Pilarczyk ( editor)

Contents
1, Design codes and safety factors
Hoekstra, A. and K.W. Pilarczyk,

Coastal engineering design eodes in the Netherlands
Verhagen, H.J. and W.F. Volker,

Safety against inundation; the Dutch approach
Niemeijer, J. and W.F. Volker,

Probability of overflow and overtopping for dike-rings
Brinkgreve, R.B.J. and H.L. Bakker,

Non-linear finite element analysis of safety factors
Termaat, R.J. and E.O.F. Calle,

Short term acceptable risk of slope failure of levees
Simm, J.D. and K.W. Pilarczyk,

CIRIA/CUR Manual on the use of rock in shoreline engineering:
objectives and contents
Leeuwestein, W., A. Franke, V. Hombergen and J.K. Vrijling,

Quarry based design of rock structures
2. Closures and storm surge barriers (see also P.7, Scour]
Janssen, J.P.F.M, and R.E. Jorissen,

Modelling storm surges and forecast effects in designing the Rotterdam
storm surge barrier
Helsloot, I.C.M. and J.P.F.M. Janssen,

Peasibility study of a two-barrier system in the Dutch sea water
defence
Janssen, J.P.F.M., A. van Ieperen, B.J. Kouwenhoven, J.M. Nederend,

A.F. Pruijssers and H.A.J. de Ridder,
The design and construction of the new waterway storm surge. barrier in
the Netherlands; technical and contractual implications
Jorissen, R.E., D.P. de Wilde and E. Berendsen,

Design of the bed protection of the Rotterdam storm surge barrier
Pilarczyk, K.W., H.J. Verheij and G.J. Akkerman,
Rackfill design criteria for overflow dams
3. Bank protection
Pilarczyk, K.W., H. Havinga, G.J. Klassen, H.J. Verheij, E. Mosselman

and J.A.A.M. Leemans,
Control of bank erosion in the Netherlands. State-of-the-art,
Boeters R.E.A.M., F.C.M, van der Knaap,

Behaviour of armour layers of riprap bank protections along navigation
channels
Boeters, R.E.A.M., H.J. Verheij and M. van der Wal,

Environment-friendly bank protections
Coops, H., R.E.A.M. Boeters and H. Smit,

Direct and indirect effects of wave attack on helophytes
Ivens, E.A.M.,
Penetration through geotextiles by shoots, roots and rhizomes of the
common reed (phragmites australis(cav.}trin.ex steudel): a pilot study
4. Geotechnical stability of dikes
Deirtekom, J.R. and R.J. Termaat,

Case history of an uplift probleirt
Dluzewski, J.M. and R.J. Termaat,

Consolidation by finite element method in engineering problems
Koehorst, B., B.P. The, J.A.M. Teunissen, M. Lojander and A. Naatanen,

The trial embankments in Vaasa, Finland: a simulation with the Adachi-
Oka creep model
Muijs, J.A. and G.A.M. Kruse,

Erosion and permeability of material for clay liners on dikes
Spierenburg, S.E.J., J.A.M. Tennissen and B.A.N. Koehorst, ,

Slope stability during infiltration
Termaat, R.J.,
Dams on soft soils: lessons from the Dutch Delta Projects
Tenaaat, R.J.,
Embankments on soft soils; application of the finite element method
Tonneijck, M.R.,
Berms at the inland side of river dikes
Weijers, J.B.A. and J.B. Sellmeijer,

A new model to deal with the piping ntechanism
Bakker, H.L. and J.R. Deutekom,

Probabilistic design method for cofferdams using FEM
5. Revetments
Pilarczyk, K.W.,
Dutch experience on design of dikes and revetments
Pilarczyk, K.W.,
Simplified unification of stability formulae for rock and other revet-
ments under current and wave attaek
Seijffert, J.J.W. and L. Philipse,

Resistance of grassmat to wave attaek
Stoutjesdijk, T., B.P. Rigter, and A. Bezuijen,

Field measurements on placed block revetments
Bezuijen, A. and M. Klein Breteler,

Oblique wave attaek on block revetments
Pilarczyk, K.W.,
Design Tools Related to Revetments incl. Riprap
6. Coastal protection and management
Verhagen, H.J. and K.W. Pilarczyk,

Coastal management in the Netherlands
Pilarczyk, K.W. and M. Stive,

Workshop document; Workshop on coastal and dredging problems on the
coast of Jastrzebia Góra cliffs to Hel peninsula. Poland
Roelse, P. and H.J. Verhagen,

Design of beach replenishment
Pilarczyk, K.W.,
Structural aspects and design procedure for sloping seawalls
Qiielerij, de, L., E. van Hijum and K.W. Pilarczyk,

Performance assessment and maintenance of coastal defenses
7. Scour and sand-water mixtures
Hoffmans, G.J.C.M, and R. Booij,

Two-dimensional mathematical modelling of local-scour holes
Hoffmans, G.J.C.M, and R. Booij,

The influence of upstram turbulence on local-scour holes
Jorissen, R.E. and J.K. Vrij 1ing,

Local scour downstream hydraulic construction
Konter, L.M. and R.E. Jorissen,

Prediction of time development of local scour
Winterwerp, J.C., M.B. de Groot, D.R. Mastbergen, and H. Verwoert

Hyperconcentrated sand-water mixture flows over flat bed
Winterwerp, J.C., W.T. Bakker, D.R. Mastbergen, and H. van Rossttm

Hyperconcentrated sand-water mixture flows over erodxble bed
Visser, P.J., H.J. Verhagen and J.K. Vrijling,

A field experiment on breach growth in sand-dikes
8. Contaminated soils
Wichman B., G. Greeuw , B. Thorborg and D.Vink ,

Development of finite strain consolidation model for optimising large
disposal sites for contaminated slurries
References
PART X
Design codes and safety factors
Hoekstra, A. and K.W. Pilarczyk,

Coastal engineering design codes in the Netherlands
Verhagen, H-J. and W.F. Volker,

Safety against inundation; the Dutch approach
Niemeijer, J. and W.F. Volker,

Probability of overflow and overtopping for dike-rings
Brinkgreve, R.B.J. and H.L. Bakker,

Termaat, R.J. and E.O.F. Calle,

Short term acceptatie risk of slope failure of levees
Simm, J.D. and K.W. Pilarczyk,

CIRIA/CüR Hanual on the use of rock in shoreline engineering:
objectives and contents
Leeiiwestein, W., A. Franke, V. Hombergen and J.K. Vrijling,

Quarry based design of rock structures
COASTAL ENGINEERING DESIGN CODES IN THE NETHERLANDS
Ammo Hoekstra* and Krystian W. Pilarczyk**
Abstract \'
Dutch research strategy oh water defences and the resulting coastal and
hydraulic engineering design codes (technical documents) are briefly
overviêwed. Special attention is paid to organizational and managerial
aspects of research and preparation of technical codes. An integrated
approach Is illustrated by the long-term activities on revetments.
Introduction
A large part of the Netherlands is situated below mean sea level; it is

protected by dikes, dams and dunes (Figure 1 ) , The country therefore
depends on good (safe) sea defences. Driven by the necessity to withstand
the water, Dutch engineers built up their knowledge on hydraulic engin-
eering over the centuries, particularly on constructing dikes and protec-
tion measures (revetments). However, the design of dikes and their revet-
ments was mostly based on rather vague experience than on calculatioh
methpds. Due to the incrèasing demand on rellable design methods, (i.e.,
as a result óf more "hard" safety regulrements) the Dutch Ministry of
Transport and Public Works (Rijkswaterstaat) and the Technical Advisory
Committee on Water Defences (TAW) have initiated a long term research
program on preparing guidelines fpr the design of sea and river defehcè
structures. Some öf these guidelines have been reported recently and sum-
marized by Pilarczyk (1987, 1990).
In this paper the main pöints of the research philosophy on the water
defences in the Netherlands will be outllnèd; both thé organizational and
* Head of the Hydraulic Engineering Division

** Head of the Research and Development Department
Rijkswaterstaat (Public Works Dpt.), R ° a d and Hydraulic Engineering
Division, P.O. Box 5044, 2600 GA Delft, The Netherlands
1 Hoekstra
— dikes *'^\
sandy beaches and dunes .<' J
•«—_>
.T.T. erosion ateas. 1 to S m per year
Fig. 1 Dutch Cöast. Erosionat Ateas
desigmng execulion
problems
fm sandy coasts (incL dunesjV

• grass/clay dikes
• rigid measures (groins)
• loose materials racuai.gnvd,
• pitched stone/concrete blocks J
• aspbatl
• mattresses/mats
va alternative measures
• in-site measurements
• models
• calculalions
• experience
solutions
governmenlaV conlractors
research consultants
manufacturers
instituüons
Fig. 2 Coastal Prolection-Integrated Apprbach
2 Hoekstra
the integrated têchnical aspects will be discussed (Figure 2 ) .
Special attention will be paid tö the management of research. An in-
tegrated approach (soil-water-structure interaction) will be illus-
trated by the long-term research on revetments.
Design philosophy of coastal defence structures
Coastal defences are constructed to protect the population and the econ-
omical values against storm surges. However, absolute safety is almost
impossiblé to realize. Therefore, it is more appropriate to speak about
the probability óf failure (or safety) of a certain defence system. To
apply this method, all possible causes of failure have to be analyzed and
consequences determined.
This method is actually undèr development in the Netherlands for dike and
dune design (TAW 1984, 1985). The 'Fault tree' is agoöd tooi for this
aim (Figure 4 ) . In the fault tree, all possible modes of failure of
elemehts can eventually lead to the failure of a dike sectiön and inuhda-
tion. Failure of an element can also degrade the behaviour of the revet-
ment even if properly designed.
Although all categories of events that may cause the inundation of a
polder are equally important for the overall safety, the engineer's res-
ponsibility is mainly limited to the têchnical and structural aspects.
In the case of sea-dikes the following main events can be distinguished
(see also Figure 3):
overflow of overtopping of the dike
erosion of the outer slope or loss of stability of the revetment
instability of the inner slope leading to progressive failure
instability of the foundation and internal erosion (i.e. piping)
instability of the whole dike.
For all these modes of failure, the situation where the forces acting are
just balanced by the strèngth of the construction is considered (the ul-
timate limit-state). In thé adapted concept of the ultimate limit-state
(Figure 5) , the probability-density function of the "potential threat"
(loads) and the "resistance" (dike strèngth) are combined. The category
"potential threat" contains basic variables that can be défined as
threatening boundary conditions for the construction, e.g., extreme wind
velocity (or wave height and period), water leveïs, and a ship's impact
(collision) . Thé resistance óf the construction is derived from the basic
variables by means of theoretical or physical models (e.g., theoretical
or semi-empirical stability-model of grains) . The relations that are used
to derive the potential threat from boundary conditions are called trans-
fer functions (e. g. to transform waves or tides into forces on grains or
other structural elements).
The probability of occurrence of this situation (balance) for ©ach têch-

nical failure mechanism can be found by applying mathematical and statis-
tical techniques. The safety margin bëtween "potential threat" and
"resistance" must guararitëe a sufficiently low probability of failure.
Three different philosophiès are currently available in construction
3 Hoekstra
micro instabilhy
flowslide
(fiquefactioa)
wave impact
(oe-and botfom-
d*^èé
scöur
Hees fcnfësliïMfê
A) Dike
Fig. 3 Övervréw of the failure
8) Saai;'
jinmtdafiOtt]
X
failure difcë
faflu iwiufe UIKC i . ïfailure
a n R ui&ed&é ._. Lfailurêdikel
rauwe u w
sectiont [ J T l . s e c t f o f e i , I T ~ ÜfefiflÉJL
explosion
sabotage «-"r^ 1
érosion intern al
acts erosion
of God' [slide plane revetment
overflow f
1
I
t overtopping
ï
'imfe„
etc.
wave > revetment
fléojf ^ dike wave slope attacfc strengtb ^
natup slability
w*ter , slope
pressure ' stability
Fig. 4 Simplified Fault Tree fóf a öike
B
l I
a pröbabiHty
!
B
1 of fatfarë -i 1 , - „ .1.- . ,.' -
o potenlial model resistance

X tfireat test t
1
i .
I '.
transfer
functions
or
field Iheorctical
fltodel
i' .... .*!

boundary materials
geometry
t ,
Fig. 5 TTte Concept of (be Ultimate Limit

State of Failure Mecbanism
4 Hoekstta
practice: (1) determihistic, (2) quasi-probabilistic and (3)
probabilistic.
Fora fully probabilistic approach, more knowledge must stil! be acquired
Goncerning the complete problems assoeiatêd with the use of theoretical
models reiating loads and streiigth. Also, improved know-ledge is needëd
urgently on the theoretical relationship between wave attack (induced
pressures) and the strength of the revetment, the probability of slope
stability rëlated to the various soil parameters, and also the theory of
interna! erosion. Studies on all these topics are still going on in the
Netherlands. The present Dutch guidelinës for dike and dunë design follow
a philbsophy that lies between the deterministic and thë quasi-
probabilistic approach (TAW 1985, CTJR 1987).
The ultimate potential threat for the Dutch dikes is derived from extreme
storm surgë levels with a verylow probability of exceédance (1% per cen-
tury for sea-dikes and 10% fór river dikes) equated with the average
resistance of the dike. Under these ultimate load conditions, probability
of failure of the dike should not exceed 10%.
Besides the ultimate limit-state, there are situations where the ever
continuing presence of a (frequent) load causes a deterioration of
constructional resistance in time, without any imminent danger of failure
(e.g. fatigue of concrete and steel, creep or erosiön of clay under the
revetment, clogging or U.V. deterioration of geotextile, corrosion of
cabling, unequal settleménts of deformations, e t c ) . However, this deter-
ioration of constructional resistance can cause an unexpected failure in
extreme conditions. These are the so called serviceability and fatigue
limit states which can also be considered as inspection and maintenance
criteria'.
As already mentioned, the fully probabilistic approach for dikes basëd

on the limit-state concept is rather cumbersome becaüse a theoretical
description for Various failure modes is not available yet- Tb overcome
this problem a scheme to simülate hearly all possible combinations of
natural boüiidary conditions in a scale model of the construction and to
correlaté the damage done to the boündary conditions can be developed
(black box approach).
Of course, field data of boündary conditions, resistance parameters and
damage are preferred as base for any correlatiori, if they are available
in sufficieht quantity.
It should also be stressed that having quantified (even roughly) the

fault tree, it is possible to pay extra attentionto thóse mechanisms
which contribute most to the overall probability of failure. Thus, this
approach is an important element in the quality control process of dike
design and dike execution. Moreover, the probabilistic approach can be
applied to spme important parts of the total defence structure (e.g.,
revetments) where the necessary input is already available from the
receiit invëstigations in the Netherlands (Burger 1988, Bezuijen 1987, De
Groot 1988, GÜR 1987, TAW/Rijkswaterstaat 1985). '
5 Hoekstra
J h e cfaullt t r e e i s a i s o a n i m p o r t a n t dtopl f p r p r e p a r i n g an fotegrated
•research program f o r w a t e r defences and f p r «he o b j e c t i w e j^^geinent of
p r i o r i t i e s (TW 1985}.
Mm complete d e s c r i p t i o n of t h e probabillts?t3tc approach f o r Jdalke d e s i g n

i s wellubeyond the scope p£ £his -paper,.„iHoweêr,,. jéemÉ&mê.mxË&mw^ém ean
'be. fotüEid i n t h e Butdh r e p o r t s and p ü b l i c a t i Q i ï s (see resf e r e p o e s ) . Knowl-
edge df t h e s e r e c e n t develppmehts c a n ;he r a t h e r 'ipi&gfcLt&tbffce,, f è s p e c i a l l y
f o r e s t i m a t i o n of p o s s i b l e r i s k s i n v o l v e d i n t h e r e a l d z e d p r o j e c t s and
f d r f i n d i n g t h e optimum between t h e r i s k s and (the investaaent ( i n c l u d ^ n g
r e s e a r c h - i n v e s t m e n t ) . The new Dutch approach t o t h e c o a s t a l defence p p l -
i c y ( d i k e s and sandy coasuts) r e l a t e d . , ampng o t h e r p a r a m e t e r , t o t h e p o s s -
i b l e s e a - l e v e l r i s e , i s d e s c r i b e d _by Werhagen (1988) > P i l a r q z y k (1990)
and i n «tüie R i j k s w a t e r s t a a t r e p o r t ( 1 9 9 0 ) .
Management and Organlzation
The main r e s e a r c h a c t i v i t i e s .on t h e «water öefenpes i n :£he i f e t h e r l a n d s a r e

t a k i n g p i a c e under s u p e r v i s i c m oi£ afche êqhnfical iAdV;ispry Coamiifctee on
Water P e ï e n c e s ("TAE) ^which was s e t >up ïfey lMfoe S f i n i s t e P wf fe^psport and
P u b l i c ¥ o r k s i n 1963. The Commiftee a ó V i s e s t h e M i n i s t e r on a l l t e c h n i c a l
m a t t e r s which may have a p e a r i n g on TÉhe pEpper p p n s t r u q t i o n and mainten-
ance of 'flood - b a r c i e r s o r on t h e s a f etsy ;o8E•-jÈhe ;a;reas - p r e t e c t e d >by them.
The studîés of tiheAdTisorgr Gommitteee i n c r e a s e ïfihe teipwledge Ün c p j i s t r u e -
t i p n , s t r e n g t h a n d environmenta;! -impact of ^ a r i o u s wa;ter tdefenoes,..
I n t h e Gommittee and i t s werking igroups., .natStpnal and 'ilpcal iauöhprit;ies
and e x p e r t s o ï TLaboratordes and t h e u i i t w e r s ; i t i e s a r e p r e s e n t e d (jFlgure
2) .
The ComsÜttee i s t e c h n i c a l l l y and a d ö ï i n i s t r a t ï i ^ e l y ö u p p p r t e d jby t h e
H y d r a u l i c E n g i n e e r i n g B i v i s i o n of t h e 'RüjjkswateEstaat (sloeated i n Belf t )
which a l s o p r o v i d e s r e s e a r c h management and budgets ;to r e s e a r c h i n -
s t i t u t e s and c o n s u l t a n t s .
No t e :
Spme f a s s o c i a t e d a c t i v i i t i e s i n t h e [ f i e l d of w a t e r jieffences t a k e allso
pïlace w i t h i n sone o t h e r o r g a n i z a t i o n s such as rfihe Butdh rCojpüittee on
'Management of waterways and t h e Bptch Gen£re ffpr GiviJl Êngineering
iReseiaorch, Codes and ;Specificat!ions (CUR). fiecause of ithe i m p o r t a n t
supplemerttary v a l u e , t h e CUR-sctiîtiies wïil'1 'be , d i s c u s s e d i b r i e f l y
n e a r zthe end of fthis ïpaper.
R e s e a r c h smanagement on w a t e r défeiices ; i s vCharacteriêd *by it-s . j i n t e g r a l

n a t u r e . ;Mirst, an i n t e g r a l apprpach t p w a t e r defences i s r e q u i r e d because
of t h e i n t e r r e l a t i o n s h i p between l a n d u s e , p r o t e c t i o n measures and t h e
d a i l y management and c o n t r o l .
Second, an i n t e g r a l approach i s r e q u i r e d s i n c e v a r i o u s d i s c i p l i n e s and
t e c h n i q u e s a r e inyal\5ed i n t h e a n a l y s i s of t h e coafftal problems and t h e i r
p o t e n t i a ! l s o l u t i o n s , ffor example, c o a s t a l e n g i n e e r i n g , r i v e r v e n g i n e e r i n g ,
economics, enwironmental iscsience, jBBathemafcical and sphysicïal m o d e l l i n g
techniques, monltoring teéhiïiques, e t c .
T h i r d , a c e r t a i n s p a t i a l i n t e g r a t i o n fis r e q u i r e d because of t h e p o t e n t i a l
6 Hoekstra
physical interactiöns between adjacent sections, for example, lower
reaches and/or mouth of rivers where the river-flood conditions should
be combined with a tidal intrusion and the storm conditions from the sea-
-side.
With respect to the diversity of the water-retaining structures, one may
distinguish the river-dikes, the sea-dikes, the natural or artificial
dunes and the storm-surge barriers of various design. The storm-surge
barriers in the Netherlands are mainly located in the mouths of rivers
and tidal estuaries (i.e., Haringvliet, Eastern Scheldt, e t c ) . With
respect to the storm-surge barriers, a wide range of the hydraulic engin-
eering scierice is involved. However, in this case, because of the speci-
fic technical matters, the management of the hydraulic research is
directly provided by the Rijkswaterstaat specialistic organizations.
Because of this integral nature of the water defences, the furiction of
the research management is not only to provide a proper integrated
research program but also to build a bridge between the research and the
practice (see also Figure 2). This last point is very important for imp-
lemèntation of the research-results.
The structure of research management.

The TAW-research is coordinated by 5 main permanent working groups, nl. :
A: Hydraulic Loading and Revetmênts
B: Geotechnical Aspects of Dikes
C: Sandy Coasts; dunes and beaches
D: Guidelines and Implementation
E: Safety Aspects and Risk Analysis
Under each of these permanent working groups a number of (time-limited)
project-teams (totalling more than 20) are working with specific terms
of reference for specific detailed problems, for example: wave run-up and
overtopping, wave impact, stability óf various revetmênts, vegetation,
settlemeht, piping, duné erosioh, various guidelines(blpck revetmênts,
asphalt, riverdikes, methods on choice of revetmênts, probabilistic
approach), safety philosophy, etc.
The working groups and project-teams have mostly a multidisciplinary cha-
racter and consist of not only researchers, but also designers, contrac-
tors and local managers (Figure 2). It helps to identify and to define
the problem, to create understanding for the choosen research strategy
and to implement the results in a quick way. This is one of the ways to
build a bridge between the research and the practice.
The main structure of the research program, its responsibility to the

political bodies providing research-funds, and the definition of the
research projects is derived from the general outline of the main problem
to be solved (i.e., protection against inundation) according to the
probabilistic technique (i.e., event/or fault tree), see Figure 6.
When the detailed problem (= project) has been defined, the detailed
research program for this project, with estimatión of the necessary
budget and time-schédule, will be prepared. Some of the problems involved
need a more fundamental approach bef ore the applied research can start
or practical products can be obtained.
When all these projects are combined horizontally (based on'the level of
development), another picture of the general research strategy can be
7 Hoekstra
political aim Law Against Inundation
(Verifkation)
i
i 1 1 '
objects dikes
1. . dunes \ |banks|
•C±]
' • ' 1 • • 1 |
design sUbility sUbility
elements |heigM| \ foundation | f ^ c (
oulerslope inner slope
1
•
1 >
1 1
failure water Ievel > wavenm-up/ 1
wave neight > gradiënt
mfffhwnisms dtkeheignt oveitoppiug sUbility armour slope/foreshore
projects/ statistic wave equüibrium

sUbDity gradients
producls water levels parameters armaur geotechnical
1 sUbility
1
cresl/slope stabüity /researcJiN
(grassmals) under V program J
overtopping/overflow
Fig. 6 Rrsass's ? f r . ™ ! - f e . h-4ilwees (PoHücsl Aim} and Rsseïrch-Projsels/Preducis
design Fig. 8 SOWAS '88 Concept

deteralniatic/
prooabillstlc
nethoda/nath. nodels
new conatructian aethods
(i.e. aand/rockfill closurea. dikea)
load- and strength criteria

(filters, toplayera. foundation)
basic/fundamental research
(procesaes. failure «echanisna.
physical underatanding; hydraulica 8 gsotechnics)
capacity/coat
Fig. 7 General Research Strategy Hydrauüc Engineering The Netheriands
(Directorale-General for Public Works and Water Management)
8 Hoekstra
presented (Figure 7). It gives an idea on the relationship between the
development-costs (inci. man-power and time) and the research products
and their appiicability. As seen from Figure 7, the basic (fundamental)
research is rather expensive.
Therefore, bef ore one starts with this kind of research, it is respons-
ible to work-öut the whole path of this research (from fundamental
research to the product), to be able to over look the whole necessary
budget and time, to ensure the continuity, to bring in some mile- stones
for new decision, and to make a proper judgement between costs and final
products.
As already mentioned, most research problems on water defences have
multidisciplinary character, specifically, ih the technical sense.
This is characterized by all relevant interactiöns between the element
soil, water and structure (so-called SOWAS-concept, Figure 8 ) , and may
lead to combined hydraulic-, geotechnical arid structural research.
Because the traditional research takesplace within the respective disci-
plines (hydraulic, geotechnic, structural, material technology, environ-
mental, e t c ) , it is not always easy to organize the combined
(multidisciplinary) research.
The separate disciplines are often faced by separate institutions (lab-
oratories, institutes, departments, etc.) with a different research cul-
ture and different management policy.
The Dutch practice has learned that the best way to perform an integrated
research is by organizing the working-groups or project-teams with inde-
pendent chairmans where the all institutions involved are able to par-
ticipate. The working group defines the total program and the involvement
of specific disciplines and institutions. It is evident that the concept
of soil-water-structure interaction (SOWAS) plays a very important role
in the Dutch Research Strategy on Water Defences.
With the hope that this concept will be more widely applied through the
world, the general introduction, as given by Van der Weide (SOWAS '88),
is briefly repeated below.
An introduction to the SOWAS concept.

Whén asked to define soil water structure interactiöns, the task seems
simple at first sight. However, when different people are asked, depend-
ing on their scientific or technological background, different answers
are obtained. Obviously, the answer calls for a more thorough analysis.
First, the problem may be analyzed at the technological level, where
practical problems are identified and Solutions have to be geherated for
this class of problems. Soil water structure interaction at this level
is defined as thë combined effect óf active and reactive forces, stemming
from mutual interaction between the structure and its physical environ-
ment.
Second, the problem may bé approached at the underlying scientific level,
where physical processes, relevant for the interaction, are isolated.
Process oriented research and development of numerical and experimental
tools are key words at this level. Soil water structure interaction at
this level is def ined in terras of the physics of the processes respons-
ible for thé interaction between the structure and its environment.
The objective of SOWAS is to bring together engineers and scientists with
different backgrounds to stimulate exchange of experience at and between
9 Hoekstra
both levels. Due to the complexity of the processes, experiments (e.g.,
model tests) play an important role in SOWAS research. Special attention
is therefore given to this aspect.
Even with the above objectives of SOWAS, different backgrounds still lead
to different definitions. It is feit that SOWAS will becomé a field of
research of increasing importance, and therefore, there is a
need to develop a general framework which can be used to integrate all

relevant aspects into one consistent picture.
The interactions described above may bé brought together in the diagram

shown on Figure 8. The outer circle represents the environmental or human
activities, responsible for loads on the system. The different elements
of this system - soil, water, structure - are represented by the inner
circles.
The external interactions between the elements, soil, water structure,

are shown by arrows, connecting the respective elements.
As described above, these primary interactions are in some instancés
accompanied by secondary interactions induced by the internal response
of the systems to loading. This response is represented by the central
box, and resultant interactions are again indicated by arrows.
Most of the subjects presented in SOWAS 88, cover one or more of these
interactions. The objective of this scheme is to identify which interac-
tion is treated and where connections with neighbouring problem areas can
be found.
It is feit that SOWAS may contribute substantially to the understanding
of the relevant processes shown in this scheme.
Centre for Civil Engineering Research and Codes (CUR) and its role in
Dutch research on Water Defences.
Collective research of national interest as well as work in connection

with codes and specifications for concrete and civil engineering (incl.
water defence structures) are the main activities of this centre (CUR).
Among others the activities concentrate on design methods, execution,
maintenance and management of hydraulic structures, and on codes and spe-
cifications for hydraulic engineering. The major hydraulic engineering
projects especially are characterized by broad-based multidisciplinary
studies and a lot of research. Therefore, the integral approach to pro-
jects and new developments is stimulated by the CUR.
In the CUR, experts from public authorities, contractors, industry, engi-
neering consultants, research institutions, and educational establish-
ments are acting in co-operation with one another on the same (collec-
tive) basis.
Finance is provided partly by industry, partly by individual members and
partly by government contributions. The procedure for each research sub-
ject and for each code to be drafted consists in setting up a committee
composed of experts in the field concerned.
Experimental research and special study projects are entrusted to yarious
Dutch laboratories and consultant firm commissioned to carry out these
investigations. New developments in technology for which there already
10 Hoekstra
exists substantial practical interest, but which are not yet considered
ready for specification in codes of standards, are the subject of CUR
recommendations. Sihce these are established through the existing commit-
tee structure and under the responsibility of the CUR, they carry a cer-
tain authoritative weight. This form of diffusion of technical Informa-
tion appears to be appreciated in practice.
The-list of some research projects related to the water defences which

are either in prögress or concluded by the CUR-publication is given
below.
Committee A27/A28: ,
Standards and behaviour models with regard to
management of the infrastructure in hydraulic engineering (mainly di-
kes and banks) incl. measuring techniques and inspection systems for
quality assessment (lst teport 1991)
Committee B 37:
Application of a l t e r n a t i v e materials in hydraulic engineering
Committee C 54:
Applied f i l t e r research (1991)
Committee C 56;
Artificial sand bodies constructed in water incl. Manual on Sand Clo-
sures (end 1991)
Committee C 57:
Geotextiles and Geomembranes
Committee C 58:
Concrete slab revetments for dikes and banks (1991)
Committee C 59:
Structural aspècts of environmentally acceptable banks (1990/1993)
Committee C 67:
Manual on Rock in Shoreline and Coastal Engineering (end 1991)
Committee C 68:
Structures on soft soil
Committee C 69:
Manual on sheet-pilirig structures
Committee C 70:
Concrete armour units for breakwaters
Committee C 74:
Guidelines on dimensióning of block revetments (1992)
- CUR report 119 (1984): Guide to concrete dyke revetments
CUR report 130 (1987): Manual on artificial beach nourishment
CUR report (1988): Jamuna Bridge, Bangladesh (Protection aspects)
CUR report 142 (1990): Guide to the design of river dikes
Examplé of integrated research on revetments
Numerous types of revetments have been developed in the past for shore
and bank protection of navxgation channels against erosiori by waves and
currents (i.e., rip-rap, blocks, asphalt, etc.)- The reason for this is
the increase óf thé problem with respect to the defence of the shores
(i.e. more rigid safety requirements for sea-dikës) and banks of naviga-
tion channels (i.e., increase of size and speed of motorvessels), as well
as the high cost and shortage of natural materials.
The fact that design rulés are still limited in quantity has stimulated
11 Hoekstra
investigations iti the area óf rip-rap, artificial blocks and bituminous
revetments as well, in the area of geotextiles.
Problems which arise due tó these deVelopnients require solutions which
often only can be föund by in-dêpth specific multidisciplinary studies
(Figure 9) .
In order to eontrol the future sea-dikes and bank-protection problems,

the. Dütch Ministry of Transport and Public Works (Rijkswaterstaat)
assigned the Delft Hydraulic Laboratory and Delft Soil Mechanics Labora-
to ry to carry out systematic research intö these areas. The research on
sea-dikes and revetments was carried-out under the joint responsibility
of the Rijkswaterstaat and the Technical Advisory Committee for Water
Defences.
The project was guided by two working groups with the common chairman-
ship. In this way it was possible to agree on the common integrated
research program.
On the basis of the analysis of practical design problems and the gaps
in the existing knowledge, the required research programmes had been de-
termined (Figure 10). This program follows the general SOWAS-concept as
it is outlined in Figure 9; it includes the intëfxation between two
fields (banks and dikes) as Well interdisciplinary integration (soil-
water-structure).
The basie programmes have been carried out by means of small-scale
models. However, it must be pointed out that a small-scale hydraulic
model for navigation purposes still nêeds a lot of space, For example,
in the scope of bank protection research programme, the hydraulic model
of an inland navigation fairway in scale 1:25 has been built in a 40 x
90 m shed to observe the induced water motions and their erosive effects
on the banks.
Since model research has certain inherent technical restrictions known

as scale effects, required additional information has been obtained by
means of prototype investigations, i.e., the Delta-Flume at the Delft
Hydraulics and some prototype locations in respect to the sea-dikes prob-
lems and the Hartel Canal (Rotterdam area) with test embankments with
respect tó bank- protection of navigation channels.
The result of the prototype tests, in combination with the model results
and the calculation methods (incl. mathematical model) developed in the
framework of the systematic research on dike protection (Delft Hydraulics
and Delft Geotechnics 1989) and systematic research on bank protection
(Delft Hydraulics 1988) extended with knowledge gained from practical
experience, led to preparation of guidelines for reliable dike and bank
protection designs. The aim of the total research was to develop such
design criteria to minimize the amount of maintenancê and construction
costs of new revetments.
Research aspects and research means.
In general, to be able to determine the dimensions of involved protection

layers of revetments, the follöwing design research-aspects have to be
taken into consideration.
12 Hoekstra
long
lermeffects
Fig. 9 Soil-Waler-Strueture Inleraction for Revelments funclional conslruclion/costs
requirémeals maintenance
Fig. 10 Sea-Dikes and Bank Protection Research Approach
Pinl(y.0
Overall Hydraulic
Hydraulic Hydraulic
Condiüons Conditions
Conditions at External at lateraal Response
HS.U Surface Surface of
Pex (y-t) Structure
r
int (y.0
Bed/Slope
Hydraulic
Geometry Propetties Mechanical
of Struclure Propetties
and Soil of Structure
Fig. 11 Schematic Presentation of the Three Transfer Funclions
13 Hoekstra
A) Research on characteristic/representative loads
Al - watermotion due to wind wavës, currents and ship-movement
A2 - geotechnical loads (gradients) due to the watermotion (Al)
B) Research on characteristic/representative strength

BI -strength of toplayer
-B2 - strength of a sublayers
B3 - geotechnical strength (i.e., sliding)
To solve problems involved in A) and B) the various research means are

available, nl:
a) Evaluation of past experiences (lessons from practice)
b) In site investigations on existing- and/or tëst-revetments, e.g.,
prototype measurements in the Hartel Ganal (Pilarczyk 1984)
c) Calculations and mathematical models (desk studies)
d) Small-scale physical models
e) Large-scale models, i.e., Delta Flume
The physical and mathematical models are very suitable for basic research
within a wide range of boundary conditiöns and for the devéloping of gen-
eral design rules, while in-site investigations, evaluation of experience
and large-scale (prototype) tests are needed for verifying the final
results (design rules). In the scope of the discussed research programmes
all of the above mentioned research-ways have been applied. The general
research strategy fór both programmes and their interactions is presented
in Figure 10.
The main difference in approach to the problem between these two investi-
gations lies in the fact that for the sea-dikes, the large Delta Flume
has been used for prototype tests while for the bank-protection problems
the 'in-situ' prototype tests have been done (large scale models for
bank-projection problems are not a reasonable solution). Another point
of difference is that the boundary conditiöns related tó banks cannot be
reproduced mechanically as it is the case for windwaves (i.e., wave gen-
erator) , but instead they have to be induced by ship movement. That is
the reason that such models need a lot of space.
Besides the difference in reproducing of the hydraulic load; both pro-
grammes involve some common aspects regarding stability of the toplayer
and the sublayer. The integration of both programmes took place by means
of the mathematical model which has to fuif il both requirements. This
mathematical model (called STEENZET, Burger 1987, Bezuijen 1987, Delft
Hydraulics/Delft Geotechnics 1989) might become an important tooi in the
design of revetments of dikes and banks of navigation channels.
The technical description of the projects and the results discussed above
can be found in references. However, to illustrate the SOWAS- concept on
the technical level, the integration of the soil-, water-, and structu-
ral-processes into one conceptual stability model will be repeated below
ace. to De Groot (1988).
General stability approach (SOWAS-concept).
The phenomena which may be relevant can be divided roughly'according to
14 Hoekstra
the three components of the system: water, soil and structure. The inter-
actiori between these components can be dëscribed using three Transfer
Functions (see Figüre 11):
I. The Transfer Function from the overall hydraulic cónditions, e.g. ,
wave height H, mean current velocity U, to the hydraulic cónditions
along the external surface, i. e. , the boundary between free water
and the protection or spil, e.g., external pressure P_
II.- The Transfer Function from the hydraulic cónditions along the ex-
ternal surface to those along the internal surface, i.e., the boun-
dary between protection and soil.
The hydraulic cónditions along thé internal surface can be dë-
scribed as the internal pressure P . .
III. The structuraï response of the protection to the loads along both
surfaces.
Information about these functions can be obtainéd by means of measure-

ments in nature and (scale) model tests. If quantitative knowledge of the
physical phenomena involved is available, or if there is enough practical
experience, then mathematical models or empirical fonnulae containing
Information are dealt with and are referred tö as "models" (De Groot
1988).
All three Transfer Functions can be dëscribed in one model, or individ-
ually in three separate models, depending on the type of structure and
the loading. The distinction between the three functions serves here
mainly as a framework to describe the different phenomena that are impor-
tant for the modelling.
International Cooperation
In general, it can be said that the Dutch Government and the institutions
involved in coastal research and management are very open for interna-
tional cooperation in the field of coastal engineering.
The Dutch reprësentatives are actively participating in the various
international organizations as PIANC, IAHR, ICÖLD etc. (see references).
In this way the research results obtainéd in The Netherlands are avail-
able f or the international community.
Also the bilateral cooperation between the countries is stimülated. An
example of such cooperation is the Manual on use of rock In shoreline and
coastal engineering, resulting from a joint project of CUR/-
Rijkswaterstaat and CIRIA, on behalf of the Netherlands and the U.K.,
respectiyely. A similar joint project with Poland has resulted in a
report on Evaluation óf Effectiveness of Coastal Protection Measures and
in the Short Course and Coastal Problems and Dredging Technology.
Dutch advances In coastal protection of lowlands (dikes, dunes) were
highlighted during the International Short Course on Coastal Protection
in 1990 (Pilarczyk, 1990 and ICCE, 1990).
The activities in the scope of the European harmonization of standards
(e.g. Eurocodes, CEN and ISO) arë coordinated by the Dutch Institute for
Standardization (NNI) and CUR.
To transmit the Dutch information to the international community, a num-

ber óf relevant Dutch technical documents have been translated and pub-
lished by CUR in English (see references).
15 Hoekstra
Because of the shortage of national researchr funds and the worldwide in-
terest and the complexity of the proper design and management of coastal
structures/systems, the international cooperation in this field should
be stimulated. It wil! not only save money, but it also will increase the
reliability of the design and in this way, it may guarantee more safety
for the population and the economical values to be protectëd all over the
wöfld. Special attëhtlöh shöüld hépaid tö the ëhviröhmëhtal impact of
the-coastal projects.
The international organisations such as PIANC, IAHR, ICOLD and the U.S.
Coastal Engineering Research Council (ASCE) can play an important role
in realization of this aim.
Conclusions and Recommendations
Based on the Rijkswaterstaat experience with management of research on

water defence systems one may draw the following general conclusions and
recommendations:
1. ' The êvent or fault tree can be a useful tooi for programming re-
search, and for defining the prioritiés and scheduling available
budget.
2. To increase the efficiency of research the whole path from basic
research to applied research (products) has to be outlined in the
early stage (incl. budget, manpower, time-schedule etc.)-
3. It is useful to define which projects or part of projects are uni-
disciplinary and which multidisciplinary. In the latter case special
attention should be paid td the organizational aspects of research
project; the multidisciplinary working group is preferable for this
purposë.
4. To obtain better research products, the integrated SOWAS-concept
should be propagated world-wide.
5. For subjects of more general/natiönal interest (applied research,
design codes) it is preferable to organize collective research pro-
jects where, for example, non-governmental institutions and industry
are also involved. it improves the content, increases the financial
possibility of the project and its effectiveness, and stimulates the
implementation.
6. Because of the shortage of national research-funds and complexity of
problëms, international cooperation, especially in the common fields
of problëms (research, design codes), should be further stimulated.
The international organizations as PIANC, IAHR, ICOLD and CERC
(U.S.), etc. can play an important role in realization of this aim.
References
Burger, A. , et al (1988), Design method for block revetnnent, 21st

Intern. Conf. on Coastal Engineering, Torremolinos, Spain.
Bezuijen, A., et al (1987), Design criteria for placed block revet-

ments and granular filters, 2nd Int. Conf. Coastal and Port Eng. in
Develop. Countries, Beijing.
Blaauw, H.G., et al (1984) , Design of bank protection of inland navi-
gation fairways, Conf. on Flexible Armoured Revetments inc.
Geotextiles, London •••.».
16 Hoekstra
Breusers, H and A, Raudkivi (1991), Scouring, IAHR Hydraulic Struc-
tures Design Manual, Published by A. Balkema, Rotterdam
Delft Hydraulics (1988). Technical Guidelines: End report on system-

atic research on bank protection, Report M 1115 (in Dutch).
CUR (1987), Manual on artificial beach nourishment, report 130,

-Centre for Civil Engineering Research and Codes (CUR), P.O. Box 420,
2800 AK Gouda, The Netherlands
CUR/CIRIA (1991), Manual on use of rock in shoreline and coastal

engineering
Delft Hydraulics and Delft Geotechnics (1989). Technical Guidelines:

End report on systematic research on block revetments for dikes,
Report M 1881 (in Dutch)
Groot, M.B. de, et al (1988), The interaction between soil, water and
bed or slope protection, Proc. Inter. Symp. on Modelling Soil-Water-
Structure Interactions, SOWAS '88, IAHR/ISSMFE/IUTAM, Delft (Editor
P. Kolkman et al).
ICCE (1990), Part IV: The Dutch Coast (special session) , 22nd Coastal
Engineering Conference, Vol. 2, Delft
ICOLD (1980), Dams and the environment, Bulletin no. 35 and no. 50
(1985) and no. 66 (1989)
Meer, J.W. van der and Pilarczyk K.W. (1984), Stability of rubble
mound slopes under random wave attack, 19th Intern. Conf. on Coastal
Eng. Houston, (also: Delft Hydraulics Publ. no 332, 378, 379 and
Bireakwaters Conf. Londón, 1985 and 1987)
Meer, J.W. van der (1988), Rock Slopes and gravel beaches under wave
attack, Doctoral Thesis, Technical University Delft (also: Delft
Hydraulics Communication N° 396)
PIANC (1985), The Stability of Rubble Mound Breakwaters in Deeper

Water, PIANC Bulletin no. 48
PIANC (1986), Disposal of Dregded Material at Sea, PIANC Bulletin

nop. 52
PIANC (1987), Guidelines for the design and construction of flexible

revetments incorporating geotextiles for inland waterways, PIANC
Bulletin no. 57, Secretariat: Boulevard S. Bolivar 30, B-1210
Brussels, Belgium
PIANC (1987), Risk consideration when determining bank protection

requiremènts, Bulletin nr. 58
PIANC (1989), Economie Methods of Channel Maintenance, Bulletin no.

67 '
17 Hoekstra
PIANC (1990), Inspection, Maintenahcè and Repair of Maritime Struc-
tures Exposed to Material Degradation caused by a Salt Water Environ-
ment, Bulletin no. 71
PIANC (1992), Guidelines for thê design and construction of flexible

revetments in marine environment (in preparation by the Working Group
nq. 21 of PTC II).
PIANC (1992), The performance and design parameters of rubble mound

breakwaters (in preparation by the Working Group no. 12 of PTC II).
Pilarczyk, K.W. (1984), Prototype tests of slope protection systems,

Inter. Conf. on Flexible Armoured Revetments inc. Geotextiles, Lon-
don.
Pilarczyk, K.W. (1987), Sea Defences: Dutch Guidelines on Dike Pro-

tection, Rijkswaterstaat, Road and Hydraulic Engineering Division,
Report WB-N0-87110, April 1987, the Netherlands (also: 2hd Int. Cönf,
Coastal and Port Eng. In Develop. Couiitries, Beijing)
Pilarczyk, K.W. editor (1990), Coastal Protection, Published by A.A.

Balkema, Rotterdam/Brookfield, 500 pages
Rijkswaterstaat (1990), A new coastal defence policy for the Nether-

lands, Rijkswaterstaat, Tidal Water Division
Technical Advisory Committee for Water Defences (TAW),

a. Guide to the assessment óf the safety of dunes in a sea défence
(1984), CÜR report 140 (1989)
b. Probabilistic design of flood defences, (1985), CUR report 141
(1990)
(both publications available at the CÜR Centre, P.O. Box 420,
Gouda, The Netherlands)
TAW/Rijkswaterstaat (1985), The use of asphalt in hydraulic engineer-

ing, Rijkswaterstaat Communications no. 35
Veldhuijzen van Zanten, R., editor (1986), Geotextiles and Geomem-

branes in Civil Engineering (Handbook), A.A. Balkema Publ., Rotterdam
Weide, J. van der (1988), An introduction of the S0WAS-concept, Proc.

Inter. Symp. on Modelling Soil-Water-Structure Interaction, SOWAS
'88, IAHR/ISSMFE/IUTAM, Delft, (Editor P. Kolkman et al)
Keywords
Coastal structures
Dikes
Research
Design codes
Probabilistic approach 18 Hoekstra
Guidelines
TAW-reprint 89.09
Safety against inundation
the Dutch appröach
TECHNISCHE ADVIESCOMMISIE VOOR DE WATERKERINGEN
H.J. Verhagen
W,F. Volker
Prèsented on the coastal zone 1989 Conference

Charleston, USA
safety against inundation
the dutch approach
Hendrik, J. Verhagen and Willem F. Volker*
abstract
The most urbanized part of the Netherlands is situated below
sea level and is protected by dikes. By act of parliamént
allowable inundation frequencies for each part of the coun-
try are fixed. In case of two dangers (river run-o ff and
storm surge) it is a problem to determine design values for
dikes. Guidëlines are developed to solye these problems.
Because econömical, social and cultural problems with the
dike improvement program, design heights óf dikes have to be
determined yery accurately. To ovércóme many problems, it
has been decided tp build a storm surge barrier of 380 m
wide and a threshold depth of 17 m below séa level in the
nayigation fairway to the port of Rotterdam.
introduction
The Netherlands are mainly located in a floodplain. Especi-
ally the industrialized and most urbanized part of the coun-
try is located in areas which are lying below high sea water
level. See figure 1. These areas (polders) are protected by
dikes and dunes. Dikes were constructed alpng the sea, along
rivers and along estuaries. The dikes should have such a
quality that inundation of the polders is prevented.
One of the most important parameters of the quality of a
dike is its height. Dikes can never be constructed in such a
way that the probability of inundation is zero. HoWever this
value should have such a low value that the inhabitants have
the feeling that they can live safely behind the dikes. In
this paper the philosophy to determihate the height of a
dike is discussed.
history
design practice for river dikes uhtil 1953

The height of diKes was in history always a problem. üsüally
*coastal engineers of the ministry of public works

(rijkswaterstaat), hydraulic division
p.o. box 5044; 2600 GA Delft; Thë Netherlands
1 Veriiageh/Volkér
figure 1. Area of the Netherlands protected by dikes
dikes were designed at a crest level of 0,5 m above the

highest known water level, with a surcharge for wave-run-up.
After a serious flood, much dikes in the coastal zone were
improved. The height was increased up to a level related to
the highest storm. However, because of bottom subsidence the
2 Verhagen/Volker
crést height of the dikes became also lower in respect to
leve! öf the sea. The river dikes in thé Netherlands were
also designed on a highest knöwn,,'Vwater-lèvéi., Alsö special
"green rivers" were in operation to divert water if the
water in thé river raised tp a too high level. Because after
1926 ttere wer^ ^
was less politica! pressure to improve the dikes than it was
iri the coastal area.: ..'After, the second world war engineers
and mathématicians warned that this appröach is hot correct,
and that an extreme value statistic should be "applied. The
consequencé of this apprpach would have considerable finan-
cial cohsequences, and no politica! decision was made.
the 1953 storm surae disaster

On february Ist, 1953 it störmed. The water-level raised tp
a level of 0.6 m higher than the highest obseirved stprm
surge (of 1894) r with as consequencé that 1365 km2 was inun-
dated and 1835 peöple were killed. 473°° houses were damag^
ed, the total damage to real estate was 160 million guilders
(in 1953 i ü.s. $ was appröx. 3.60 Dutch guilders). in
comparisoh, in 1916 6874 km2 and in 1894 306 km 2 was inun-
dated. Because „most Öf this area was aïso below meah sea
level, after passage of the storm, the polders were stil!
covered with water. Repairing 160 km of dike toOk more than
a year and it cost 380 million guilders. The total direct
costs of the disaster were 2000 miilion guilders, which was
14 % of the gross national product in 1952.
The main conclusion was: this should never happen again, and
a committee of spècialists (the delta-committee) was instal-
led to find a solution
solutipn of thé Delta-committee

The delta-committee concluded in its final report that dikes
should be designed on a design storm-surge levél with a
givön probability of occurrence. From economie considerati-
ons follöwed that storm-surge levels with a probability of
less than 1/10000 a year are thé optimum for the densely
populated central part Pf Holland. For the other provinces
this value is 1/4000 a year.
The 1/10000 storm surge leve! was determined from extrapola-
tipn pf all knpwn water-leveis at Hook pf Hplland (see fig;
2), and resulted in a design water-levél pf 5 m abpve meah
sea-leyei.,1' • . '•••.'''
This figure has to be correeted for the various locations
alóng the coast.
Thé Delta-comntittee advised that a dike should be designed
3;:",'.:;'"',:,;;,:v • ,,! Vérliagen/^plker,:'

in such a way that every cross-section can withstand this
water-level, with the accompanying wave-run up in such a way
that nó serious damage to the dike will occur. The number of
overtopping waves should be less than 2 %.
figure 2. dike-circles and inundation frequency
solution of the "Becht-coTOmittee" for river dikes

Around 1975 it was realizéd that also river dikes had to be
improved, in order tö guarantee an identical safety to the
people living behind river dikes. The "Becht-committee" was
formèd to determine the height of river dikes. This commit-
4 Verhagen/Volker
tee conqluded that inundation by rivers is less serious than
inundation by sea water. This is caused by the fact that:
* inundation with fresh water causes less problems
* the warning time for high-water run-off is longer than
for a storm surge from the sea;
* the poldersalong the rivers are mainly above the nor-
mal normaï waterlevel, andcönsëquëntly the water
flows Out of the polder after passing of the high-water
in the rivèr
* there is no tidal flow through the gap in the dike, and
repair is therefore more easy.
Based upon these considerations the Becht committee decided
that river dikes should be designed on a run-off with a
probability of occurrence of 1/1250 per year. The "Becht-
committee,, was influenced by the f act that the public dit
not like the works for raising dike-levels
probabilistic considerations
As discussed above, the design of a sea-defehse was in all
cases based upon a representative load (watër-level and
waves)with a defined probability of occurrence. The fact
that there is also a variation in this load, as well as a
variation in the strength of the sea-defense was neglected.
Studies were performed in order to find a probabilistic
method in which also these variations could be taken into
account. With a probabilistic approach it is possible to
achieve this. [Vrouwenvelder, 1987]
the dike-circle
Using the probabilistic considerations, mentioned above
allowed to give a probability of failure of dike section. In
fact a dike has also a given length. The inhabitant of a
polder area it is not interested what is the probability of
failure of a give dike section, but what is the probability
of getting wet feet or worse. The low lying part of the
Netherlands is devided in so-called dike-circles. A dike-
circle is a low-lying area, which is surrounded by dikes,
dunes and/or high grounds. Failure of one section of the
sea-defense usually results in the inundation of a whole
dike-circle. Each dike circle has a given allowable probabi-
lity of inundation. In figure 2 somé of the 40 dike-circles
of the Netherlands are presented with the allowable inunda-
tion freguencies.
The allowable inundation frequencies are laid down ' in the
Law on Sea Defense. The choiee of an allowable inundation
frequency is fundamentally a pQlitical decisionï
5
Vertiagen/Volker
More récent studies are performed to fihd the optimal allow-
able inundation freqüency, based üpon the economie value of
real estate and infrastructure in the dike circle. Hpwever,
the values of human life, natiiral environment, historica!
and cultural values; etc. made it impossible to define the
optimal value in an objective way.
It is interest ing to mention that the resul ts of these
studies indicate that the economically optimal values are in
the order of 10~ 4 to 1Ö"*5 per year. The frequencies proposed
in the new Law on Sea Defense are a factor 10 higher.
In this Law it is alsó stated that the boundary values (sUch
as water-levels) have to be recalculated every five year,
and that dike managing authorities have to certify every
five years that their dike still fuifills the reguirements*
So, they have to check the height of the dike, the quality
of the slope protectiön, etc. In this way it its tried to
prevent that the effect of climatic changes causes sur-
prises, and dikes have to be adapted to the new situations
regularly. This is the main reason that design water-levels,
e t c , are not given in regulation, but only their probabili-
ties'.
the problem of tidal rivers

Along tidal rivers there is a complicating factor. High
water levels can be caused by high river run-of f, but also
by storm surges entering from the sea. The problem of calcu-
lating the waterlevels is discussed by Van der Made [1969].
The step from a design Watërievél to an inundation freqüency
of a dike-circle is recently worked out for the guidelines
on de design of river dikes [TAW, 1989; TAW, 1990]. In
figure 3 and 4 the lócal water-level Is presented fqr two
stations at different ppints along the same dike circle. För
the locations of the stations Jaarsveld(J) and siiedrecht(S)
see figure 2. •.•
The water-level is a function of the water-level at sea
(Hook of Holland) and the run-off (Rhine discharge at
Lobith) . Both the water-level at Hook of Holland and the
river discharge at Lobith have there own probability of
occurrence. These variables are not fully independent. For
example at station Jaarsveld the height of the dike is 6.0
meters above mean sea level. This water-level can be achiev-
ed by a storm surge Of 5.5 m above mean sea level. and a
run-off of 8000 m 3 /s or by a sea water level of 3.3- m (once
in ten years storm) and a run-off of 15000 m /s.
üsing both the distribution of extreme storm surges as well
as extreme river discharges, the exceedance-freqüency for
6 Verhagen/Volker
all stations can be calculated. Examples are presented in
fig 3b and fig. 4b. From these diagrams follows that in
Jaarsveld a water-level of 6.0 m is exceeded with a proba-
bility of 1/4000. In Sliedrecht 3.75 m is exceeded with the
same frèquency.
Jaarsveld
2083 «e80 6609 8000 10880 12000 H880 '6800 18808
discharge of the Rhine at Lobith [mVsl
Jacrsveid
r z :s :
V)
2T
SBL
e 4£Z
ZZL
3Z T' '
a
:zz:
i I,I Mi
«U4.
•U+.
— -, x.
7^~
S_
42?
SXL •vu :
Ttr"
-3
exceedance-frequency fevents/yearl log10

figure 3. Waterlevels a t Jaarsveld
7 "' Verhagen/Volker
If for all stations along the dike these diagrams are con-
structed, for each dike section the watër-level with a pro-
bability of occurrence of 1/4000 per year can be determined.
Unfortunately this does not mean that if the dikes are con-
'• structed in such a way, the probability of inundation of the
dike-circle is also 1/4000 per year.
Slledrecht • R
200e *eee saaa sees leeee i28aa neee isaea ïeeea

discharge of the Rhine at Lobith [m 3 /sl
Siiecrecftt
Ol
> 32 at=3-3?5;
ai
=^2=
o
J3
IJt
2=
2 I 0 -1 -2 - j
exceedance-frequency [evenf-s/yearl loglO
figure 4. Waterlevels a t Sliedrecht
8 Verhagen/Volker
When a dike-circle has only fcwo section, both constructed in
such a way that they have each a probability of failure of
1/4000 per year, and they are attacked by fully uncorrelated
dangers, the probability of inundation of the dike-circle is
2 times 1/4000 per year.
In reality a dike-circle has many sections, and they are
attackedby partïy correïated dangers. Mathematica! tech-
niques are available to calculate also in those cases the
probability of inundation. This is in the order of three
times the probability of failure of a single dike section.
required accuracy; politics, ecology and social problems

In the Netherlands storm surges with a ten times less fre-
quency of occurrence are approx. three quartêr of a meter
higher. This implicates that if a dike is made approx. half
a meter higher, the safety increases approx. with a factor
5. (Wave run-up may change these figures) . The cost of
making a dike a little bit higher are generally not very
high, if the dike is situated in agricultural areas. Also a
few extra feet increase in height does not introducé ecolo-
gical and söcial problems as a rulë.
In build-up areas however, this is not true. The dikes cross
towns like Rotterdam and Dordrecht. E.g. in Dordrecht the
main shopping center is established in historical houses,
built on the dike. These buildings date from the 18th centu-
ry and have a high historical value. This implies that the
consequences of 10 cm extra height are considerable. Such a
difference costs sometimes 10 million US$ more for a dike
stretch of less than one kilometer.
the storm surge barrier

Because of the problems of constructing dikes in the tidal
area, it has been decided by parliament in march 1989 to
construct a storm surge barrier in the entrance channel to
Rotterdam. This will be a barrier with a free width of 3 60 m
(without any piers in the channel) and a threshold depth of
17 m below mean sea level and no limitations in height for
shipping traffic, in order not to disturb traffic to Rotter-
dam. (The harbor-basins for deep-drafting ships, more than
17 m, are situated seaward of the barrier) This barrier will
be closed only during storm surges occurring once in approx.
10 yeats.
üpstream of this barrier dikes in the tidal area will only
be attacked by river run-of f and not any more by storm
surges. The consequence is that dikes in the tidal area do
not need to be raised. In build-up areas this makes it more
9 Verhagen/Volkér
easy to find suitable solutions. Financially the construc-
tion of a storm surge barrier is more or less equal as
raising the dikes. The construction of the barrier will cost
a lot of money in the next few years (750 million US $), but
improving all the dikes in the tidal area will cost more
time (until 2020). Especially the long times required for
raising the dikes, and the social and ecological problems of
these works are politically unacceptable.
practical guidelines
In spring 1989 "Guidelines for the design of river dikes,
part 2, tidal rivers" has been published. In these guidelin-
es an operatiohal method is presented for the dike managing
authorities tó design dikes. This method consists of twó
parts. In the first part an approximation is given for the
height, using a table of water-level/wind combinations for
each place around the dike-circle. In the second step all
the dikes of a dike-circle are entered in a computer pro-
gram. This program calculates with the above mentioned meth-
od the probability of inundation of the area. By trial and
error the crest levels of the dikes may then be found in
such a way that the probability of inundation is correct and
the total "costs" (money + other problems) are minimal.
In this version "inundation" is defined as a situation in
which at at least one location around the dike-circle the
water-level and wave run-up exceed the crest height of the
dike. So in f act only failüre by overtopping is taken into
account. Failure by other causes (for example by instability
of the subsoil) is also taken into consideration. The design
should be in such a way that the probability of occurring of
such a failure is extremely small. For the time being it is
estimated that probability of failure is for 60 % determined
by overflow and overtopping (which is completely included
in the probabilistic approach), and for 40 % by all the
other mechanisms of failure.
It is the' intention that in the next vers ion of the guide-
lines also the other failure mechanism are included in the
calculation, so that really the probability of inundation is
calculated.
After running the program the authorities can see if the
system fulfills the legal requirements. If not, somewhere a
section of the dike has to be increased in height. The pro-
gram also indicates which sections contribute to the•insecu-
rity. These sections should be considered for improvement.
Again by trial and error the most effective improvement can
be found.
10 Verhagen/Volker
In the computer program the effect of a storm surge barrier
is fully included.
references
Remark: Many references can be given to reports of the Dutch
government on this subject. Although these reports are not
@öhfidéhtial, they are guitë difficult tö öbtain abroad, and
they are all in Dutch. In the following list only accessible
references in English are given. Interested readers may
contact the authors of this paper for background information
from the Dutch reports.
Made, J.W. van der; Design levels in the transition zone

between the tidal reach and the river regime reach, Associa-
tion Internationale d'Hydrologie Scientifigue, Bucarest
conference, 1969, pp 257-272
Technical Advisory committee on Waterdefenses [TAW]; Guide-

lines on the design of river dikes, part 1, non-tidal ri-
vers; to be published by CUR/Gouda in 1989 (translation of
the Dutch guidelines of 1984)
Tèchhical Advisory committee on Waterdefenses [TAW]; Guide-

lines on the design of river dikes, part 2, tidal rivers; to
be published by CUR/Gouda in 1990 (translation of the Dutch
guidelines of 1989).
Vrouwenvelder, A.C.W.M,; Probabilistic Design of Flood de-

fences, CUR/Gouda, 1987
11 Verhagen/Volkèr
PROBABILITY OF OVERFLOW AND
OVERTOPPING FOR DIKE-RlNGS
J. Niemeijer, W.F. Volker,
Ministry of Transportation and Public Works,
Road and Hydraulic Engineering Division,
P.O. box 5044, 2600 GA Delft, the Netherlands and
A.W.C.M. Vrquwenvelder,
Department of Structural Engineering, TNO,
P.O. box 49, 2600 AA Delft, the Netherlands.
Abstract.
The program DIJKRING enables the calculation of the probability of

failure by overtopping or overflow for a dike ring. The random
variables are the river discharge Q, the wind direction R, the sea
level H and the wind velocity V. The program is based on numerical
integration. The models for local water level calculation, the wave run
up and overtopping discharge are the same aö in conventional design.
This latter point is essential to have the program accepted as the
Standard. An important advantage of the dij kring -approach is that the
safety of the total system is eyaluated, where conventional design
always focuses on the safety of a single section.
1. Introduction.
Large parts of the Netherlands have to be protected by dikes against

inundation. The acceptable failure rate of thöse dikes is the outcome
of public debate. The resulting frequency, e.g. failure once in 10000
years, is taken for granted in this paper. After the storm disaster in
1953 it was undêrstood that therefore a dike height ought to be at
least 0.50 meter above the water level which is exceeded with the same
frequency. Those dike sections where overtopping of the crests by waves
was to be expected the design level was even more then 0.50 meter above
this water level.
Recently it was realised that this interpretatipn of failure rate
should be improved and that the resulting calculations could still be
done with the aid of a perspnal computer only. According to this new
interpretation failure is related to the probability that during bne
year, at least at one point of the dike-ring overflow or serious
overtopping occurs.
A dike-ring is here defined as a closed set of water retaining
elements, which together protect an area. Those elements may be dikes,
dunes, surge barriers etc. In the Netherlands there are about 50 of
those dike-rings, as shown in figure 1. For the purpose of the
calculatión every dike-ring is subdivided into a number of sections
(e.g, 100)
2. Overflow.
For the time being we will ignore overtopping. For a dike-ring like no
16 in figure 1 overflow may óccur because of a high sea level, a high
discharge by the river Rhine or a combinatiön of these. For given
values of both sea lèvel H and the Rhine discharge Q, local water
levels h along the branches in the Rhine delta area can be calculated.
Here we take this calculatión f óf granted. Overtopping at one place
occurs as soon as the local water level h exceeds the crest elevation
Failure = {h(Q,H) > ha} (1)
Let FEq(h,q) be the probability that during a arbritrary tidal period

in the winter H < h and Q,< q. (Italic letters like h and q are dummy
variables for the stochasts H and Q respectively.) And let f^ be the
corresponding joint probability density function.
The probability of local overflow, once Fgg is known, can be calculated
from the basic expression:
P = ƒƒ *B,q(.hfq) dh dq • (2)
where - .,'.,' ^ '
Dy = {h{h,q) > hi}.
Let fH(ii) and f^(.q) be the marginal probability density functions of

FQH(g,2i).
The frequency pf exceedance per year pf thè sea level at Hook pf

Holland is giveiï by the dottëd line in figure 4. This line is based pn
observatiöns. From this £s(h) can be calculated, when taking in mind
that there are rpughly 350 tidal peripds in a winter. We assume that
durii^g the supmier np high sea levels pccur.
Figure 5, alsp based pn pbservatipns, tells us the average number pf
days D(g) in a winter the discharge exceeds q. Therefpre alsp fQ(q) is
knpwn.
Because the pbservatipns dp npt disprove it, we assume that H and Q are
npt dependent. It fpllows that fHQCiJ,g) = feC^) fq(9)>
With these data it is now possible tP calculate the dpuble integral

given in formula 2. üsually we are more interested in the probability
that somewhere along the dike-ring overflow occurs; Again we may use
fPrmula 2, but the afea Dy in the q,h plane is npw larger in genera!.
3. Overflow and pvertppping.
Overtppping pf a ijike section occurs as the cpmbinëd result pf high

water level and wave run up (figure 3b). Most of thé dikes in the
Netherlahds are grass dikes, and failüre of those dikës by pvertppping
rèsults when the average discharge S pver the top óf the dike exceeds
s, a critical value. Typical values of s may vary from 10"4 m2/s for a
weak dike to 10"2 m2/s for a strphg pne. S depends óh the Ipcal water
level h and the height and dirëction of the waves at the toe of the
dike. (For some wind directions R there will be np waves at all.) Later
on we will show how, R and the wind velocity V being known, the waves
at the front of the dike and the average discharge S can be calculated.
SP overtopping of a dike section occurs if S(Q,H,R,V) > s.
As before failure of a dike section by overflow is defined as h(Q,H) >
Failure of a dike section by overflow or ovèrtopping is defined as
Failurè = {h(Q,H) > hd or SCQ,H,R,V) > s}. (3)
Again, failure of a dike ring by overflow and/or ovèrtopping occurs if

at least one dike section fails. Those combinations of q,h,r,v which
lead to failure we define again as the Dy domain.
Let FHQRV(ii,g,r, v) be the probability fuhction that during a tidal

përiod there is at least one hour that H < h, Q < q, R = r and V < v.
Let faQRvCA.g.r, v) be the correspoiiding density fuhction.
The probabirity of failure can be calculated by the expression:
F = Jjjj fHQRv(-ö,q,,r,v) dv dh dr dg. (4)

:
DF .. ' ' • ' ' ' :. . • .•
As stated in the previous paragraph, Q and H are independent. Further

we assume that the wind is independent of Q, becaüse only heavy
rainfall in Germany results in a high discharge of the Rhine in the
Netherlands and the time lag is a few days. Therefore:
=
fHQRV fHRV^»-^?") fQ(g). (5)
The sea level is a result of tides and wind set up. As a consequence
H,R and V are by no means independent. The function fgnyCh.r,v) was
fpund as follows:
By courtesy of the Royal Netherlands Meteorological Instituté we know
the wind speed statistics presented in figure 6 and fR(r). From these
data foll ows fgy(r, v). From figure 4 and f^(r) follows fng(jr,/z).
All these data are not enough to determine ffiRv ^ u t fortunately there
are more relevant data available at least f or the 7 wind directiohs,
S¥,WS¥, ...NNW an4 nörth which induce set up. A list of the highest
water level in a yeaf and the wind velocity at the same point in time
is available for each of these directions (60 combinations per wind
direction). Af ter inspection of these data we concluded that a wind
direction R and a water level H was not always accompanied with the
same wind velocity V. In statistical terms the conditional density
function fy|HR is a dome shaped curve with a non zero Standard
deviation.
It was feit that nearly any dome shaped curve was good enough, provided
that:
00
f
via(f1r) = jfv|HRM fl >-r> f
Hia(fclr) dh (6)
(As mentioned above fTO and fR a r e known and fV|R = fTO / f R .)

So f o r t h e 7 f u n c t i o n s Fv|HR(v"jh,r) we i n t r o d u c e d Gumbel f u n c t i o n s :
FV|HR(v-|A,r) = exp(-exp(-K R fv) + OR h + B R ) ) (7)
By trial and error we found for these wind directions the function
KR(v) and the constants ctR and % , using not only equation (6) but also
the maximum likelihood method for fitting the 60 combinations of V and
H.
4. The computer code.
A computer code was made to calculate the probability of failure P, as

given in equation (4). This code consists of two parts:
- the determination of the Dy space, i.e. those combinations of q,r,h,v
which lead to overflow ör övertopping.
- the integration with Dj as boundary.
The first part starts by calculating Vcri1:, for every dike section. Vcrit
is defined as the wind velocity V which produces the value of s, the
critical discharge by övertopping. Vcrit depends on the local water
level and R. Vcrit is calculated as follows. Subroutines calculate the
significant wave height, given the fetch lengths and water depths in
front of a dike sëction. These subroutines are based on Brets'chneiders
formula and result in the wave height, the wave period and the wave
direction (= r) at the toe of the dike. These results, together with
slèpes of the dike, dike height etc (see figure 3 ) , are input for the
subroutine calculating the discharge. By trial and error it is possible
to find Vcr:Lt.(h,r). These calculations are rather complicated and PC-
time consuming. We assüme that the effort is justified by the accuracy
of the formula used. Anyhow the output of the first part are files in
which combinations of a local water level and wind (h,Vcrit and R) are
stored. Each combination leads to overtopping with a critica! discharge
s or to overflow. (In the last case h = hj and Vcrit = 0.) One can state
that for every dike section the failure domain is determined by these
values of V crit (h,r).
The second part of the computer code calculates the fourfold integral.
The flow chart for the code is presented in figure 8. For each value of
Q,R,H the local water levels are calculated. ¥ith the aid of the files
just calculated follows Vî,. for every dike section of the dike ring
and consequentïy the minimum value qf all these values of ycvix{q,r,hj.
These minimum values determine the failüre space DF of the dike-ring.
The numerical integration over h and r is straight forward. However the
integration over q provides a complication treated in the next
paragraph.
5. Advanced time variability model.
In the program DIJKRING the above procedure is foliowed for river

discharges up to 30Ö0 m3/s. For high discharges, this procedure does
not meet our purpose exactly. So far we have restricted ourseives to
the probability that overflow and/or overtopping occurs during a tidal
period. However if up river a dike overflows, it happens during a
period of high discharges and probably during a few e.g 3 tidal periods
in succession. In that case we say that failure has happened only once
and not 3 times. As a consequence We are not interested in the
probability that overflow or overtopping occurs during a tidal period;
but we want to know the probability that overflow and/or overtopping
occurs at least once during a year.
Starting point is a model of the river discharge during a month. It is

assumed that there is one maximum value Qt every month. From the
observatioris, see figure 7, we deduce fQt(9t)» t n e density function for
the maximum discharges. Let m(.q\qz) be the number of days that in a
month Q >g. Here it is assumed that m(g|gt) is the same every month,
i.e. this function depends only on gt.
It was found that m can.be expressed reasonably well by:
30 qt - q 0.6
Mq\qz) * — — [ ] (8)
a(9t) Qz ~ 500
where
a(qt) .« 1; the exact value follows from D(g) and fQt(gt).
Let qm(ra,qz) be that value of Q which according to (8) is ovèrchargéd

during m days in succession. (m = 0 . 5 , 1.0, 1.5, 2.0, .... 30. )
With (4) P{F|g,m} is known, i.e. the probability that in a tidal period
overflow or overtopping takes place at one or more dike sections, given
the discharge qm. The desired result follows:
The probability PF that during 6 winter months one or more dike section
overflows or dvertops for discharges largér than 3000 rf/s:
PF - 6 J [1 - n [ 1 - P{F|Qm(m,gt)} ]fQt(gt) dgt (9)

3000 m
"where
m = 0.5, 1;0,...., m(g|qft)
For discharges lower than 3000 V / s (4) is used. The sum of (4) and (9)
gives the total probability that overflow or overtopping occurs.
6. Application.
As an application, consider the results for the dike-ring Hoekse Waard,

as indicated in figure 9. The total ring has been subdivided into 18
dike sections. In all cases 0.01 m2/s was considered as thé nomina!
value for the critical specifie discharge.
For each section the geometry of the dike and the fetch lengths have
been determined.
The resulting probability of failure is 1.5 * 10"3 /year. In addition
to this value, the program also gives the distribution of the difce
isections which have given the main contributiohs to the system failure
probability. In the present casei (see Tabïè 1) dïke sectiöh 10 has the
gréatest contribution, foliowed by the sections 2, 9, 11 and 12. So to
improve the system one should start by improving these sections. It is
not possible to teil in advance wether it will be sufficiënt to improve
these sections only.
As a matter of fact, the value of 1.5 * 10~ 3 is larger than the target
value, which equa,ls 2.5 10~4 /year. However, sihce last year, the Dutch
Government has decided to the construction of a storm surge barrier in
the Nieuwe Waterweg (see figure 1 ) . This barrier has a significant
influence on the calcülation óf thé local water levëls for given sea
level and river discharge. The influëricë, of course, depends oh the
design and closürë strategy or the barrier. Starting from the present
design, the calcülation for the Hoekse Waard can be repeated which
-4
results in a failure probability öf 1.8 lO /year. The effect of the
Storm surge barrier is that the dikes around the Hoekse Waard do not
nëed ahy further improvement.
Table 1: Contribution of the sections to overall failure probability
ar. Probability Contribution| nr. Probability Contribution
1 3.026E-0007 o.or 1 io 8.592E-0004 58.4Z

2 1.156E-0004 7.9Z 1 11 1.818E-0004 12.4Z
3 0.000E+0000 O.OZ | 12 1.253E-0004 8.5Z
4 6.158E-0016 O.OZ 1 13 1.013E-0007 O.OZ
5 5.763E-00Ö8 O.OZ 1 14 1.876E-0005 1.3Z
6 1.091E-0005 0.. 7Z 1 I5 O.OOOE+0000 O.OZ
7 6.790E-00Ö7 O.OZ 1 16 O.OOOE+OOOO O.OZ
8 2.909E-0007 • o.oz 1 17 0.OOÓE+OOÓO O.OZ
9 1.650E-0004 11.2Z 1 18 0.OO0Ë+O000 O.OZ
notation
a(g) : a function of the discharge

Df : area in H - Q - R - V plane where overflow or overtopping
occürs
D(q) : average number of days in the winter season that Q>g
f : probability density function
F : probability function
h : local water level
H .:' sea water lével
h,j : crest elevation
m(q) : number óf days in a month that Q>q
P : probability
Q : river discharge
Qt : maximum discharge in a month
R : wind direction
s : critical value of S
S : discharge over the crest of a dike due to overtopping
V : wind velocity
V crit : wind velocity f or which S = s
Italic letters denote dummy variables

figure 1 : dike-rihgs in the Netherlands
Figure 2: Schematic example of a dike ring
lock
Figure 3 : Mechanisms of overflow and overtopping
Figure 3a Mechanism of overflow.
Figure 3b Mechanism of overtopping.

40
30 -f
t-
ao
sea water wlndspeed
level H (m/sl
-ba) +NAP 10
1 10 * 10 io; a 10
1 10"* 10"" 10^ 10 frequency of exceedance per year - i — — * .
frequency of exceedance pep year
figure 4s Surge l e v e l a t a t i s t i e s figure 6» Wind speed s t a t i s t i c s
20000
20000
; •
,
- - ;-_-
15000
. " -; war / - 15000
- •
/ „•'"' ' mr • .*
- y ** - --
- - - / "' - ~
'
10000
-"' 10000 /
- - /
discharge - •
- maximum
** -
, * • "
lm3/?) discharge - -
sumner
5000 ' (mVs)
/ s
. - 5000
• •-
^ " -
^* <** _ -
/"V - / s
. i i -
3 8 I 0 -1 -2 (
10 10* 10 10 10 10
X<? 10 10 Ï0 S 102 10' 10° 10"' 10"s 1Ó'3 10"" 10"!
exceedance in days per year frequency of exceedance per year ——^—*•
figure 5: River discharge statistics figure 7i S t a t i s t i c s of maximum discharge

STAR!
Q - Q + 4Q
find V . f o r g i v e n h and E
ï - ÏIV > V c r l t ^ |B R Q)
F - P(f«llure|Q]
1 — 6/(X-n[l-P[faIlure|Q,Q]) f(Q) dQ
end
figure 8: s i m p l i f i e d flow c h a r t f o r t h e computer code DIJKRING
i Goidschalxoord
* Middelbarnis Hoeksche Waard
figure 9: The d i k e - r i n g the Hoekse Waard

R.B.J. Brinkgreve
Delft University of Technology, The Netherlands
H.L. Bakker
Public Works Department, The Netherlands
ABSTRACT: In this paper a robust method is proposed to determine the safety factor of
geotechnlcal constructlons in finite element computations. The method Is based on the
reduction of the strength parameters of soil, the friction angle £ and the coheslon c.
Three examples show the practical appllcation of the method.
1 INTRODÜCTION
In structural engineering a factor of

safety is always defined as the ratio of
the collapse load over the the working
load. The same definition is adopted in
foundation engineering, at least for
footings and piles. For soil bodies such as
road/river embankments and earthen dams the
situation is different. Here the dominating
load is not a direct external force, but
most of the load comes from soil weight.
Very cohesive soil bodies can be loaded to
collapse by increasing gravity, either
numerically or in centrifuge tests, but not
when the strength is dominated by friction.
Therefore other definitions of safety are
common in soil mechanics. The usual soil
mechanics definition of safety is:
safety factor = -f- = g *,£,gff

C C C
wheré S is the shear strength, further

def ined by the well known Mohr-Coulomb
er i ter ion. No te that c is the cohesion, <f>
is the friet ion angle and tr' is the normal
pressure at the plane considered. Both c
and 4> are effect ive strength parameters and
er' is an effect ive stress. We use the
subscript c to indicate critical strength
parameters; Just high enough to ensure
equilibrium. Indeed, for c^c and $*$ the
c c
safety factor becomes equal to unity. The
above definition of safety coincides with
the ciassical definition as used in slip
circle analysis on the condition that we
define:
c / tan^ -. a c / tan^ with cc = 1

C C
In this study the proport ionall ty with a«l

wilt be retained in order to remain
compatible to traditional slip circle
analysis, but the method allöws for Other
ratios of c and tan* .
This study concentrates on the cbmputation

of the above safety factor by use of an
elastic-plastic finite element method. In
stead of the usual incrementation of loads,
strength parameters will be decremented.
This technique was first proposed by
Zienkiewicz at al. (1975), but we make the
procedure robust by adding an arc-length
technique. A robust procédure was needed
for Implementing into the PLAXIS finite
element package. Finally the potentlal of
the method is demonstrated by considering a
nuinber of applications.
2 MOHR-COÜLDMB MODEL
Being interested in collapse loads rather

thanpreeise deformations, the
elastic-piastic Mohr-Coulomb model is
adopted. Hence
*-D ( è - èp) , è p "- X-^r
where D is the elasticity matrix and
. 1 Öf . _, ui -,
A = -—s—5—p Dn e , d - —=— D
d öo* .-• - * öo* - 8tr
The symboIs f and g are used to denote the
function respectiyely. Considering planar

deformatiohs only, with the x-y-eoordinates
in the plane and the z-axis normal to the
plane, we def Ine
f = T - ( e cóty + o* ) sin0
g = x - o* sin*A
where
T = -(o* -o* ) + o* , o* = -•-Co* + o* )

4 xx yy xy 2 xx yy
For the sake of convenience, it is assumed
that the out-of-plane stress o* Is in
zz
between the principal stresses o—r and O*+T.
3 IMPLICIT INTEGRATION
Finite element analyses involve finite

i -ements of stress and strain rather than
rates as considered above. Several
integration rules can be applied to pass
from rates to finite increments. One of the
most popular ones, as first proposed by
Vermeer C1979) and most recently by Borja
& Lee (1990), is the implicit integration
scheme. Here the direction of flow is
considered to be fully determined by the
state of stress at the end of a load
increment. Denoting finite increments by
the symbol A and t for time, we get
••te'- x A | d t = | / x d t , - | ! |
where
fe;• f(£*+ D Ac)
where <r is the stress at the beginning of

the load increment. Details are given by
Vermeer & Van Langen (1989). Having defined
the increment of plastic strain as a
function of the total strain increment, it
is easy to compute the stress increment
from:
Ao- = D (Ae -Aep)
4 FINITE ELEMENT FORMULATION
We consider a soil body subject to constant

gravity and constant external loads. These
are equilibrated by a known stress field
o;°. Hence, the finite element equllibrium
conditions give:
Fig. 1. Increments for strength-reductlon

Sy B V * dV - Q° , Sy B «r dV = Q°
where both body forces and external loads

are assembled in the vector Q for nodal
forces. We now apply a decrement of the
strength parameters, so that the yield
locus rotates as indicated in Fig.1.
Obviously, the new stresses dr (without
superscript) will be at the new yield
locus, at least for material points with
stresses or° above this line. Let us now
define
er • (r°+ Aar1* Air2 t Ao;2s D (AG -Ae p )
where £°+ Acr is the closes t po int

projection of 0;0 on the new yield surf ace.
Substitution in the equilibrium equation,
and using c = B u , gives:
K Au = R°+ AR + P
where K is the elastic stiffness matrix) and
R° = Q°- J*y B ? ° dV = 0
AR = - / y B Ao;1 dV , P » ƒ B D A£ P dV
In practice R° will not entirely vanish

because of numerical procedures; separation
from AR is important for later formulation
of an arc-length control procedure. P is a
pseudo load vector, as used within an
initial-stress iteration-procedure of the
type:
K Au k + 1 = R°+ AR + Pk,
for k=l,2, ..k and px= 0
nax *-
We adopt an initial stressprocedure since-
a tangent stiffness procedures tend to
break down in the fully plastic range dtie
to ill-conditioning.
5 INDIRECT DISPLACEMENT CONTROL
The above strength reduction procedure is

is far from being robust. Indeed, by a step
wise decrease of strength It Is possible to
approach collapse, but for some step the
strength will be reduced too much, i.e.
beyond critical, so that the iterative
procedure will not converge tp an
equilibrium state of stress. In this manner
the preeise critical values of tan0 and c
are hever obtained. Therefore indirect
displacement control is needed, or in other
Words are-lengthcontrol (Rheinboldt &
Riks, 1986). Within this approach we add a
load multiplier 0, and obviously an extra
scalar equation for solving 0. The system
of iteration equations is
K Auk•« R°+ 0kAR + P*"1

T k T
Au Au = Au Au
., — o • • — -o — o
The subscript o is used to denote results

f. i the previous load step. Instead of the
node displacement vector of the previous
step, other yectors can bë used to obtain a
linear scalar equation for 0. Similarly
non-linear equations are feasible.
6 STABILITY OF RIVER EMBANKMENT
The first practical application involves a

river embankment in the tidal zone. The
slope has a horizontal length of 10 m and a
height of 5 m, so the ratio is 1:2. The
most dangerous situation for a embankment
is at the onset of low tide. The waterlevel
outside the embankment is already low, but
the phreatic level in the embankment is
still high. The situation, which is
modelled is shown in Fig. 1. We consider a
h' ogeneous soil with material properties
as indicated in Table 1.
Fig. 1. River embankment at low tide
Table 1. Soil properties for embankment
Parameter Symbol Value
Dry weight rd 16.0 kN/m3

Wet weight . . . • 7
w
. . . , .
20.0 kN/m3
Frietion angle . " . ' • >
30.0 °
Cohesion C 5.0 kPa
Shear modulus G 1000 kPa
Poisson's ratio V 0.3
In the first part of the analysis we apply

the soil weight to introducé the stresses
for the situation of Fig. 1. Some Of the
Gaussian integration points appear to be in
plastic state already, but the embankment
nas not collapsed at all.
In the second part of the calculation we

want to analyze the safety of the
construction. Therefore we apply ten steps
in which the strength parameters of the
soil are stepwise reduced, according to the
theory as described before. When further
reduction of the strength parameters in not
possible anymore, the construction has
collapsed and the safety factor is
obtained. Fig. 2 shows the computed
strength-displacement curve for the creSt
point óf the embankment The value of the
safety factor for this particular problem
appears to be 1.20.
1.23- * !
Strength reduet ion

1 , ' 't ••
1 1
| f
1.20
f Ir ! i
1 /i
» • a s a s * • a ^ » • • « 4 * •'(•.••ga* • • ^mm£ • *£•» • • • • ! 1 —
1.15-
i f 't i i
! ƒ ! ! !
] ƒ • ! !
i ƒ l
1.05-
i i
i ƒ i i• «
i
i ƒ i j i
1.00 ,— i—J —4 'i , i
0 0.23 0.50 0.79 1.00 1.25
Displacements crest point * 0.1
FIg. 2. Strength-displacement curve
Fig. 3 shows a plot of the incremental

displacement contours at failure. In this
plot one can recognize a nice slip circle.
In the same plot the critical slip circle
according to Bishop'sraethodis plotted.
Note that we performed a drained analysis
both for the finite element approach and
the Bishop approach.
Fig. 3. Incremental displacement contours
Whilst the positions of the slip circles

are somewhat different, the safety factor
in both Bishop's method and the finite
element calculation is exactly the same.
This gives some validation of the proposed
methód. It might now be concluded that the
calculation of the safety factor in finite
element codes has no advantage in
comparison with conventlonal methods. In
this simple case this happens to be true,
but the finite element methód will also
detect non-circular slip surfaces and
a ociated factors of safety. The next
examples will show that in general the
finite element calculation of the safety
factor gives Information that cannot be
obtained with conventlonal methods.
7 STABILITY OF A SHEËT-PILE WALL
For the second applicatiön we consider the

safety of an excavated building trench. The
excavation depth is 21 m. The soil is
supported by 32 m long sheet-pile walIs,
which are strutted at the top. During the
excavation of the soil the water table
remains at a constant level, both inside
and outside the trench. Fig. 4 shows the
f inite element mesh of the trench. Special
line elements, as available in the PLAXIS
c puter code, are used to model the
Fig. 4. Mesh for building trench
sheet-pile wall (Bakker & Brinkgreve,

1990). The dashed lines in Fig. 4
indicate Joint elements to model reduced
friction at the soil-structure interface
(Van Langen & Vermeer, 1990).
The soil is mostly fine silty sand with

some clay layers. In fact it consists of 7
layers, which correspond to the element
layers in the mesh of Fig 4. The mater lal
properties for each layer are given in
Table 2. Layer No. 1 is the top layer and
layer Np. 7 is the lowest layer. For the
sheet-pile wall and the strut we used:
EI '« 1.65 x 1 0 6 kNia 2 /*
- sheet .
EA '« 8 . 0 4 x 1 0 6 kN/m
• 'sheet
EA / L••• 1 , 2 8 x 1 0 S kN/m 2
strut
Table 2. Properties for different layers
Layer V y
o
c G
kN/m3 w kPa kPa

3
kN/m
1 18.0 20.0 30.0 1.0 6000
2 16.0 16.0 25.0 8.0 3000
3 20.0 32.5 0.0 6000
4 20,0 32.5 0.0 8000
5 20.0 32.5 0.0 25000
6 18.0 27.5 4.0 9500
7 20.0 32.5 0.0 15000
The soil which is to be excavated, is

mode lied by extemal loads: horizontal
tractions simulating the horizontal
stressès at the sheet-pile wal 1 and
vertical tractions simulating the vertical
stressès at the bottom of the trench.
The calculation consists of 3 parts. The

first part is the introduction of the
initial stressès including the tractions on
the sheet-pile wall and the trench bottom.
The second part is the excavation of the

trench bystepwise removing the extemal
loads. Bending moments and strut forces
occür in this part of thé calculation.
There is a stress reduction directly behind
the sheet-pile wall mainly due to arching
(Bakker & Brinkgreye, 1990). The extreme
horizontal displacement of the sheet-pile
wall is about 100 mm. The extreme bending
moment amounts 2100 kNm/m and the strut
force is 500 kN/m. Fig. 5 shows a plot of
the incremental displacement contours at-
service state. A concentration of contour
lines can be regarded as the initialisation
of a shear band.
Fig. 5. Velocity contours at service state
The third part of the calculation Is the

determination of the safety factor by means
»»* d M C i i g k u t cuuwkxuu. nuCOi u i u g LU L Ü X S
calculation, the safety factor appears to
be 1.65. The development of the bending
moments during this calculation part is
'r:' ïrkable, as the clasping moment
disappears and the sheet-pile wall tends to
behave like a beam on two supports (Bakker
& Brinkgreve, 1991).
Fig. 6 shows a plot of the incremental

displacement contours at collapse. This
plot visualizes a slip line going from the
right-hand top, underneath the sheet-pile
wall to the middle of the trench bottom. In
comparison with the service state the slip
line has moved drastically downwards.
Fig. 6. Velocity contours at fallure

8 EMBANKMENT STABILITY WITH GEOTEXTILE
The third application involves the safëty

of a road embankment. The embankment has a
height of 6 m. In this example two
situations are modelled: Firstly an
Fig. 7. Geometry of road embankment
ordinary case where the embankment is build

Just on top of the underground and secondly
with the addition of a geotextile
underneath the embankment. The situation is
shown in Fig. 7.
The soil consists of four layers. The top

layer is the embankment fill materlal. The
underground consists of two clay layers
with a layer of soft peat in between. The
soil properties are given in Table 3.
Table 3. Propertles for different layers
r 0 c K
Layer o
0
3 w. kPa kPa
kN/m
3
kN/m
1 18.0 18.0 25.0 2. 0 500
2 18.0 18.0 24.0 5.0 500 0.60
3 11.0 11.0 15. 0 2.0 250 0.75
4 18,0 18. 0 24.0 5.0 500 0.60
Again the analysis consists of 3 parts. The

first part is the introduction of the
initial stresses in the underground. As the
underground is horizontal the initial
stresses can directly be derived from the
soil weight and the K -values given in
Table 3. °
The second part of the calculation is the
building of the embankment. At the end of
this part the top of the embankment shows a
settlement of 68 cm (without geotextile)
and 66 cm (with geotextile). The geotextile
seems to have little influence so far.
The^ third part is the calculation of the

safety factor. Now the dlfferences between
both situations are much more apparent. In
the case without geotextile the safety
factor is less than 1.20 while in the case
with geotextile the safety factor is over
1.30. In practlce, only the lat ter
situation is acceptable.
The difference in safety factor between the

two situations is due to the f act that the
fallure mechanisms are compïeteiy
different. Flg. 8 and 9 show plots of the
incremental displacement contours of both
situations. In the first case the f allure
mechanisra is a slip clrcle going through
the soft peat layêr. In the second case the
same mechanisra is initially developed, but
Ihls is resisted by the geotextile.
Collapse finally occurs in the embankment
itself.
Flg 8. Velocity contours without geotextile
Fig. 9. Velocity contours with geotextile

Considering the resul ts it would seem that
geotextiles have an effect on stability
rather than oh deformation, but this will
depend on the relative stiffness Of the
geotextile. For the present application we
used EA - 2500 kN/m.
9 CONCLUDING REMARKS
The proposed finite element method for the

determination of the safety factor by means
of strength reduction glves results which
agree well with those from Bishop's
slip-circle method, at least for simple
situations when a circular surface occurs.
The advantage of the method in comparison

with conventional methods is that it even
deals with the most complex kind of
geotechnical constructions. Besides one
does not have to predefine the fallure
mechanism.
Are-length control makes the procedure

robust since fallure need not be assoclated
with a non-converging iterative procedure.
Moreover the are-length procedure gives the
fuil strehgth-dlsplacement curves beyond
possible peaks.
REFEIïENCES
Bakker, K.J. & Brinkgreve, R.B.J. 1990. The

use of hybrid beam elements to model
sheet-pile behaviour in two dimensional
deformation analysis. In Proc. II Eur.
Spec. Cpnf. on Num. Meth. in Geotech.
Eng., p. 559-571. Santander, CEDEX
Bakker, K. J. & Brinkgreve, R. B. J. 1991.

Deformation analysis of a sheet-pile
wall, using a two dimensional model. In
Proc. X Eur. Conf. on Soil Mech. and
Foundation Eng., Florence
Borja, R.I. & Lee, S.R. 1990. Cam-Clay

plasticity, Part I: Implicit integration
of elasto-plastic constitutive relations.
In Comp. Meth. in Appl. Mech. and Eng,
Vol 78, p.49-72
Rheinboldt, W.C. & Riks, E. 1986. Solution
techniques for non-linear finite elements
equations. In State-of-the-art surveys on
finite element techniques, p. 183-223. New
York, Appl. Mech. Div. of ASME
Van Langen, H. & Vermeer, P.A. 1990. Finite

element analysis of a pile penetration
problem in clay. In Proc. II Eur. Spec.
Conf. on Num. Meth. in Geotech. Eng.,
p. 519-527., Santander, CEDEX
Vermeer, P.A. 1979. A módified initial

strain method for plasticity problems. In
Proc. III Int. Conf. on Num. Meth. in
Geomech., p.377-387. Rotterdam, Balkema
Vermeer, P.A. & Van Langen, H. 1989. Soil

collapse computatibns with finite
elements. In Ingenieur-Archiv 59, p.
221-236
Zienkiewicz, 0.C., Humpheson, C. & Lewis,

R.W. 1975. Associated and non-associated
visco-plasticity and plasticity in soil
mechanics. Géotechnique 25, No. 4,
p.671-689
f!
XIIIICSMFE, 1994, New Delhi, India / XIII C1MSTF, 1994, New Delhi, Inde
SHORT TERM ACCEPTABLE RISK OF SLOPE FAILURE OF LEVEES

RISQUE ACCEPTABLE DU GLISSEMENT DE TALUS DES LEVEES
RJ. TERMAAT, Hydraulic Engineering Division, Ministry of Public Works, Delft, The Netheriands
E.O.F. CALLE, Delft Geotechnics, Delft, The Neterlands
SYNOPSIS: The probability of failure during the stages of (re)consmiction levees on soft' soils depends on me strategy of
execution: e.g. loadsteps, waiting time between loadsteps, geometry and soil improvemenL In mis paper the acceptable
probability of failure from an economical point of view is discussed. The optimal target probability of failure is föund by
balancing the costs of diminishing tfae rist of slope failure against the gain in terms of reduction of expected cost of damage
repairë, using an insurance cost model. The present study shows that the acceptable probability of slope failure for short term
situations may be in the order of a few percentsfrom an economical point of view.
Introduction
A great number of levees along the main rivers in Holland have been or
are to be reconstructed, in order to bring up crest heights to statutory
regulated safety levels. Especially in the western part of the country,
sophisticated design of reconstruction stages is necessary, because of the
bad subsoil conditions concerning stability, consolidation and settlement
behaviour. The upper ten meters of the subsoil consists of peat and clay
layers. Nearby tillage and dwelling areas often limit tbe space available
for reconstruction. These aspects necessitate prudent strategies of
execution of reconstruction, in order to avoid damage due slope failure.
However, too prudent approaches would resült in siraall reconstruction
steps and long intermediate consolidation stages, which are unattractive
from an economical point of view. Sóme risk of slope failure is
considered acceptable and, in f act, actual or near failures occur now and
then.,
For example, in 1986 a slope failure over a Width of 60 m occurred in a
stretch of the Lek dike slightly before reconstruction would have been
completed. As part of a program to investigate the possible causes, a
probabilistic analysis of stability has been carried out. This analysis
revealed that the probability of slope failure during reconstruction , taking
into account the uncertainty about pore pressure dissipation in the clay
and peat layers, may have been in the order Of 0.5 percent. The
requirements for stability at the long term come down to probabilities in
the order of 0.001 percent or less per year. This raised the question about
what risk of slope failure might be acceptable. Though the water retaining
function has not been in jeopardy in this particular case, nor is it in most
other cases of observed slope failure, it was also questioned what the risk
is of a slope failure that would seriously affect the water retaining
capacity of a dike, leading to inundation.
This paper presents the results of a study, primarily addressing to the first
question. It should be pointed out that the matter of determing acceptable
probabilities of slope failure ceuld be appraoched in several ways. The
purpose of this study was to look at it from a point of view of economie
optimization of designing construction stages.
Outline of the procedure
The analysis is based on minimizing the total cost involved in the

execution of the reconstruction stages. Basically, two types of cost
components must be distinguished, namely the cost components of actual
execution of the reconstruction stages and cost components which reflects
a sort of "insurance premium'' to insure against the cost consëquences of
actual failure, if it occurs. Roughly speaking, if the acceptable risk of
slope failure is assigned to 1 percent for each stretch, one might expect
that in the course of execution of a dike improvement program real failure
occurs in one out of every hundred stretches at the average. The cost of
damage repair should be divided equally over hundred stretches, since
each one is equally likely candidate for failure.
The actual probability of slope failure can be reduced (or increased) by

allowing less (or more) critial conditions regarding porepressure
development during and in between the construction stages. For exmple
by choosing fewer or more construction steps, longer or shorter
intermittent consolidation periods or applying drainage systems.
Uncertainty about pore pressure development has a significant influence
on the probability of slope failure. In the analysis it will be assumed mat
monitoring of pore pressüres may signiflcantly reduce uncertainty,
provided that appropriate measure will be taken in practice when observed
pore pressure development tends to be less favorable than assumed
beforehand.
Precautions to reduce the risk of slope failure tend to increase the cost of
execution. On the other hand, reduced risk of failure tends to reduce the
amount of "insurance premium" mat has to be taken into account. From
an economical point of view the choice of an accepted (target) probability
of slope failure can be optimized by minimizing the total cost function.
Such an analysis requires the following steps:

- Estimation of the probabilities of slope failure for a set of potential
designs of the execution strategy of step by step reconstruction. In this
paper the result of such analysis is presented as the probability of
slope failure as a function of elapsed time of the reconstruction
period, since different strategies of step by step reconstruction mainly
influence the elapsed time needed for execution.
- Estimation of the costs of execution of these designs.

- Estimation of the actual cost of damage repair in case a real slope
failure oc;urs.
- Determination of the total cost fiinction and finding the probabilities
of slope failure which correspond widi minimal total costs.
Obviously, the analysis can be carried out for selected cases only. Two
cases from real practice have been considered in diis study (Best, 1992).
They have been selected, because in both cases real slope failure has
occurred, so realistic estimates of the cost of damage repair were
available. Though the cases were completely different, the analyses lead
to remarkable consistent results. In this paper, only the Streefkerk case
will be presented.
Method of probalistic stability analysis
The computation model, used for probabilistic analyses of stability is

based on the conceptual model discussed by Calle (1985).
Basic components of this model are:
- A Gaussian stationary random field description of spatial fluctuations
of shear strength of the soil..
- Circular failure modes and a Bishop type of equilibrium analysis,
however, adapted for second moment analysis of random variations of
shearing strength.
As a consequence, the stability factor (factor of safety) F is no longer a
deterministic single valued quantity, but instead a Gaussian random
function in the along slope direction. lts expeeted mean value, Standard
deviation and aütocorrelation function are detennined by the random field
statistics of the shear strength fluctuations. From this, estimates of the
probability of occurrence of a zone where F < 1 somewhere along the
slope axis and, if this occurs, an indication of the width of such zone, can
beobtained.
Apart from spatial variability of shearing strength parameters,
considerable uncertainty is involved in the estimation of excess pore
pressures in cohesive soil layers. Dissipation of excess pore pressures is
expressed in terms of the consolidation factor (w), the factor zero denoting
an undrained situation and die factor 1.0 complete consolidation.
Uncertainty is modeled by assuming that the consolidation factor is a
random variable with bèta type probability density function P„:
•^-f^ni^p^L. o<«<i a>:

y\b-ai \ b-a' b-a
a
and
1 tu-arlL u-aV
A normal distribution for the consolidation factor is considered less

appropriate, because the consolidation rate is physically limited. It cannot
exceed the upper bound b-i and the lower bound a=0. The parameters
q and r are detennined by matching the statistics of the bèta distribution
wim the predicted (calculated) consolidation factor and an estimated
Standard deviation, based on observed differences between predicted and
actually measured pore pressures. Since predicted pore pressures tend to
be accurate as the consolidation factor tends to its limiting values (0 or 1)
and less accurate in between, the Standard deviation (oj has been modeled
as a function of the predicted consolidation factor (u), using the following
relation:
The probability of slope failure (P7) is calculated by discrete integration of

the bèta distribution:
P,=£.7>. * P(u) (4)
In mis formula PAui denotes me conditional probability of failure given the

consolidation factor M,.
Total cost function
The economical optimum is the minimum sum of the construction costs

and me "insurance" costs for damage repair. The Jatter is the damage
costs in case of failure multiplied by the probability of failure. The
probability which corresponds with the minimum total costs, is the
optimal accèptable risk from an economical point of view. This approach
is illustrated with a case history, namely the reconstniction of a dike
stretch of 2.5 km near Streefkerk, 20 km east of Rotterdam. A typical
cross section is shown infigure 1.
\s ss <; J) !> v // yy yy // ;? v / v / /s ss
\////////////////////////////////////. 6
t reconstrucüon clay sand
%g$ existing dike peat
Fig. 1: Cross section of the dike reconstniction near Streefkerk
The construction costs depends of the strategy of execution, e.g. number

of loadsteps, waiting time between the loadsteps, geometry and
improvement of the soil. The alternatives which have been considered in
our analysis are summarized in table 1. The roman numerals indicate the
setected alternatives for presentation in this paper.
Table 1: Matrix of alternatives

drains no no yes yes
piezometers no yes no yes
berm (Standard reference I II III
width)
berm + 5 m - - - -
berm +10 m ' - - IV
Analysis of construction costs
The costs for construction exist of costs which do not and costs which do
depend on the choice number of construction steps and intermittent
consolidation periods. The first category contains expenses for
embankment material, installation of drainage systems and piezometers
etc. Table 2 resembles these costs for the yarious alternatives given in
table 1, relative to the costs involved for the reference case. The
estimated costs for the reference case amount 8 million dollars. The
second category contains the expenses for rent of containers, maintenance
and security, mobilization and demobilization of equipment, etc. These
costs have been estimated to be in die order of 3,000 dollar per month,
the costs of each mobilization of equipment are in the order of 6,000
dollar. To give an impression, a construction time of three years involves
additional costs which amount 132,000 dollars.
Table 2: Additional reconstruction costs

berm (Standard) 0$ 5,000$ 25,000$ 30,000$
berm + 5 m 120,000 $ 125,000 $ 150,000 $ 155,000 $
berm +10 m 240,000 $ 245,000 $ 275,000 $ 280,000$
Relation probabiiity qffailure versus elapsed construction time
When the construction time increases, the consolidation factor at the end
of a consolidation period decreases and consequently die probabiiity of
failure during the following construction step decreases too. The
construction costs on the other hand will increase. In figure 2 the
probabiiity of failure is plotted versus the construction time for each of
the four selected alternatives.
1000 1200 1400

construction time (days)
Fig. 2: Probabiiity of failure versus construction time
The method used for the calculation of probabiiity of failure is briefly

described in the previous chapter. For the Standard deviation a„ of tiie
consolidation factor a value of 0.10 is adopted at a reference level of 50%
consolidation. If pore pressures are mönitored by piezometer observations
the Standard deviation in our analysis was reduced to 0.05.
Costs of damage repair in case of failure
In the case of the dike reconstruction near Streefkerk failure occurred in

the final construction stage. The damage related to this failure has been
estimated to be 2.1 million dollar.
As mentioned previously, the instability in this case did not initiate breach
of the dike and inundation of the polder. A substantial part of die crest
was stil! intact and during failure and aftefwards no extreme waterlevei
occurred.
In case of an inundation die damage costs of damage repair would have
been much higher. Based on aivalable surveys it is estimated to be about
2.8 billion dollar. However, the real disaster would have been the
expected great number of casualties. The calculated risk of inundation,
triggered by mis slope failure, however, was extremely small.
The cost of damage repair which has to be taken into account, Q„: i.e.
the costs for "insurance", is approximately equal to the cost of damage
repair multiplied by die probability of failure:
c^ - cf*pf * c^* p, (5)
In mis formula PfM is me probability of an inundation triggered by the

slope failure event, Q, tiie cost of damage repair, effected by inundation
and Cf me cost of damage repair if inundation is not triggered.
The probability of inundation is extremely small, which means tiiat me
"insurance" cost for damage caused by inundation is also relative small.
Total cost function
In figure 3, the total costs, i.e. the sum of construction costs and expected
costs of damage repair, have been plotted as a function of target
probabilities of slope failure for each of the construction alternatives.
Acceptable probabilities of slope failure from an economical point of view
correspond with the minimal total cost values in this figure.
800+
60Of
§
400f
2004
a = total costs b = damage costs including

inundation
c = expected damage costs d. = construction costs
Fig. 3: Probability of failure versus cósts

In table 3 the minimal costs and corresponding elapsed construction
periods and probabilities of slope failure have been collected for the
construction alternatives. From this table and figure 3 it can been seen
that the optimal target probabilities of slope failure range in the order of
0.5 to a few percents. Further it shows mat a remarkable profit can be
gained from monitoring pore pressures, provided of course mat
appropriate measures are taken in case of non expected pore pressure
development. Also the investment in drainage systems to speed up
consolidation pays off. Stability control by overdimensioning, i.e.
applying wider berms than strictly necessary for the long term stability
requirements, is costly and does not seem to be of much help.
Table 3: Minimal total costs and corresponding elapsed construction

time and target probability of slope failure for tiie construction
alternatives
berm (Standard) 270,000$ 195,000 $ 220,000 $ 140,000$
1335 days 1335 days 505 days 505 days
4.8% 0.8% 4.8 % 0.8 %
berm + 5 m 310,000$ 240,000$ 260,000$ 225,000 $
1.0% 0.8 % 1.0% 0.8 %
berm +10 m 355,000 $ 320,000$ 335,000 $ 325,000 $
0.6% 0.7 % 0.6% 0.7 %
Concluding remarks
From case studies it was found that the probability of slope failure during
execution of reconstruction works on river dikes in Holland may well be
in the order of 0.5 percent. Though this may seem unacceptably high in
view of the long term stability requirements, it compares well with
investigated risks of failure during the construction stage of other
engineering structures, for example in the field of hydrauüc engineering.
The acceptable risk of slope failure, from an economical point of view,

has been determined for a real test case, in which the existing dike had to
be reconstructed to fulfil new requirements to the crest height. The
applied appraoch was to minimize the total cost of construction works.
Total costs include a fictitious cost item, called expected cost of damage
repair, which reflects the contribution to reservations to be made for
damage repair in case a real failure occurs. By applying this "insurance
cost" model it was found that optimal target probabilities of failure during
the construction period range in order of 0.5 to a few percents.
In the fïrst instance, potential breach of the dike foliowed by inundation
of the polder triggered by slope failure, was left Out of considëratiön.
Costs of damage repair included strictly repair of the dike itself. Later on
the probability of inundation, triggered by slope failure, has been taken
into considëratiön and, consequently, the enormous cost consequences of
inundation. It was found that circular failure modes which affect the crest
height of the dike and thus are likely to initiate inundation have an
extremely small probability of occurrence. As a consequence, taking
potential inundation into account did not significantly affect the optimal
target probabilities as found in the fïrst analysis.
It should be mentioned that overtopping waves in the case of damaged

inner slopes may cause further erosion and eventual breach of the dike.
This mechanisme has not been taken into considëratiön. However,
construction works are in general carried out during summer seasons, in
which the probability of river water levels, high enough to cause
overtopping, is small.
Several alternatives to construct the dike or to accompany the works, i.e.

ïnstalling vertïcal drainage and piezometers or överdimensioning by
appiying wider berms, have been considered. From the analysis results it
appeared that considerable gain can be expected from pore pressure
development control. Överdimensioning is costly and does not pay off.
References
Calle, E.O.F. (1985). Probabilistic analysis of stability of earth slopes.

Proc. of llth Inter. Conf. on Soil Mech. and Found. Eng., San
Francisco, Vol. 2, pp. 809-812.
Calle, E.O.F. (1992). The acceptable risk of slope failure during the
reconstruction of levees. Delft Geotechnics, Report No. Co-283811/25
and Co-282812/5 (in dutch)
MAFF Conference Of River & Coastal Enqineers - Loughborough, July 1991
CIRIA/CUR Manual on the Use of Rock in Shoreiine Engineering : Objectives and Contents
JDSimm, Robert West & Partners UK

KW Pilarczyk, Rijkswaterstaat, Netherlands
1. Introduction
After a two year project by CIRIA on behaif of the UK, and CUR (Centre for Civil
Engineering, Research and Codes) in the Netherlands» a Manual on the Use of Rock in
Coastal and Shoreiine engineering is now in publication CIRIA/CUR (1991). The manual
forms the first European guidelines on the subject and grew out of a recognition by both
nations of a continued and increasing use of quarried rock in the construction of new
coastal defences -such as breakwaters, sea walls and groynes - and in the upgrading of
existing works.
The mptivation for producing the manual in the UK came from two main directions. The first
of these was the frequently recurring complaihts from quarry producers and contractors
alike, the lack of standardisation in armöur stones specification. This was led to the quarry
producers being unablé to standardise their próduction processes for stonè abpve normal
aggregate sizes, nor to be able to stock pile armour stone or riprap for which they could
be guaranteed our future use. Criteria for excluding Weaker or lessdense rock have also
varied widely, leading in some cases to perfectly acceptable rock types being excluded on
arbitary grourids. Iri this context, a manual which standardises gradings and the approach
to their definition, and introduces criteria whereby durabjIJty of rock materials in the coastal
environment can be objectively assessed seemed highly desirable in assisting UK quames
to compete with Continental quarries and in lowering the end cost to the cliënt.
A second motivating fprce in the UK behind the production of the manual stertimed from
the feedback CIRIA received on its Report 61 éDesign of Riprap Slope Protection Against
Wind Waves" which was based on work by Hydrauiics Research Station, Wallingford, and
published in 1976. This report was widely welcomed at the time, but CIRIA has received
frequent requests from designers for clarificatión on the range of acceptable gradings.
The impetus in the Netherlands tor producing the manual, came frpm the completion of an
extensive series of tests on rock armour stability at Delft Hydrauiics Laboratory, which
extended the basis of CIRIA Report 61. The research, by Dr Jentsje van der Meer and his
colleagues, has now been written up and new formulae have been published for the
publication of rock armour stability in various situatiOns. Simultaneous development of
formulae on hydraulic response of rock structure (wave refteetion, transmission, run up, and
overtopping), has also taken place in the Netherlands, UK and in other countries. These
considerations, combined with developments in the understanding of the geotechnical
behavoir of coastal structures at Delft Geotechnics, combined with the extensive experience
gained recentty during the specification and use of rock in connection with the Eastem
Scheldt storm surge barrier, prompted the Dutch to suggest the inrtiative of producing a
manual on the use of rock in hydraulic engineering. In an atmosphere of tncreasing
European cooperation, it was therëfore appropriate for CIRIA and CUR to get together 01
order to distill the best of experience in both countries on the subject in one document
The project was funded by the Department of Public Works (Rijkswaterstaat), Ministiy Of
Economie Affairs and private companies in the Netherlands, and the Department of the
Environment and the Ministry of Agricufture, Fisheries and Food, in the UK. The work was
carried out in liaison with the British Aggregates Construction Materials Industries (BACMI),
in order to ensure that quarry producers views were properiy represented.
Thé project was supported by onôing research in the UK and the Netherlands including
i) research in the Geomaterials Unit of Queen Mary and WestReld College London,
on improving the integration of quarrying considerations into the design of rock
armour structures, funded by the Science and Engineering Research Council.
ij) the conclusion of research projects on rock stability and hydraulic response of rock
structures in the Netherlands
iii) research identified and funded specifically by CIRIA and CUR for the production
of the manual. In the Uk this included a limfted series of tests at Hydraulic
Research, Watfingford, on very wide graded rock armour layers, funded by CIRIA,
and work on the effects of waves in shallow water, also at Wallingford but
commissioned by CUR. In the Netherlands, the research included work on analysis
of wave transmission results by Dr van der Meer at Delft Hydrauiics and further
work on van der Meer*s formulae and program on the dynamic stability of berm
breakwaters, also carried out at Delft Hydrauiics, both the latter desk studies being
supported by CUR.
Project Objectives
The specific objectives of the Anglo-Dutch initiative in producing the manual were defined
as follows :
i) to coliate available research data and technical information, together with practical
experience gained by practitioners from the Netherlands and UK, in order to
facilitate better dissemination and use of existing knowledge.
ii) to describe the best at present design practice (ie state of the art), indicating the
limitations in present understanding of the processes involved, and the extent to
which empirical methods and engineering judgement is involved in present design
procedures.
iii) to set out a procedural framework of planning and design activities which guides
the practitioner in an integrated approach to the use of rock, yet allows an
appropriate site specific solution to be adopted, recognising that Standard solutions
do not generally exist in this field of engineering.
Approach to Production of Manual
Technical steering/working groups were set up in both the Netherlands and UK, representing
a wide range of interests and experience, to provjde technical advice and input through the
project, and to review contributiorts produced by nominated individuals/organisations (both
paid and unpaid) in both the UK and the Netherlands. A joint steering committee was set
up at an early stage comprising of ME Bramley (CIRIA, subsequently G Stephenson), J
Simm (Robert West and Partners, CIRIA Research Supervisor), KW Pilarczyk (Rikes Waterstat
and Chairman CUR steering group) and W Leeuwestein (CUR and technical secretary for
the project). It was also agrëed that J Simm would act as overall technical editor for the
manual (general reporter), and be shadowed in this task by W Leeuwestein.
Over a period of months commencing in eariy 1988, a contents list for the manual was
slowly evolyed and developed to quite an advanced stage by the end of 1988, prior to the
comrnissioning of the writing of any sections by indMduals or organisations. This ensured
in a situation wheremanyauthors were going to be Invorved, that a degree of consist^cy
across the manual would be achieved from a very early stage. Strict guidelines on the
approach to writing sections of the manual were also laid down and these generally
ensured that subsequent problems in integrating text were minimised.
Contents
It was realised in preparing this paper that many delegates to the conference would be
interested to know in detail whether particu!&r a'eas of persona! interest would bs covered
by the manual. The fmal contents list for the report has therefore been attached as an
appendix to this paper. This contents list enshrines within it an overview to the design
approach of rock structures in coastal and shoreline engineering which was developed in
the course of the project and is shown in the fiow chart in figure 1. From this it can be
seen that fundamental chapters on materials, hydraulic and geotechnical design, structure
specific design, and maintainance and monitoring are supported by chapters on the wider
view of the design process and definition on the hydraulic and geotechnical boundary
conditions around a rock structure, together with various appendices including a Standard
specification and series of tests for rock materials together with a Standard method of
measurement. It is considered that a major achievement of the manual was to synthesise
disparate pieces of information on a particular subject in a way that (in most cases) enabled
clear guidance (eg, one formulae or one graph) to be given instead of a range of
information presented in the form of a literature review. Subsequent sections of this paper
demonstrate specific items for which this has been achieved.
4.1 Materials
Rock materials properties were classified into three categories : intrinsic properties,
manufactured (ie as-produced) properties, and as-constructed properties. These
properties are related to one another and to the functional requirements for the
structure, as shown in Figure 2. A range of tests for the intrinsic and
manufactured properties were evaluated for their relevance to applications of these
materials in coastal and shoreline structure. As a result of this evaluation, a series
of Standard materials tests were drawn up. The tests were either new or were
modïfied from existihg tests used for aggregates, some from a draft series already
prepared in Holland and translated into English, and spme frorn those developed
öut of work at Queen Mary and Westfield college. The Standard tests are inciuded
as Appendix 2 of the Manual. The tests include detërmjnation of gradings, length
to thickness ratio, resistance to freezeAhaw, crushirig strength, water absorption^
presèncê of clay minerals, abrasion resistance, and a 'drop test breakage index'.
A good êxample of such a new test is the Queen Mary and Westfield College Mill
Abrasion Test, which enables assessment of rates of wear, (for example in the
abrash/e shingle environments around the UK coast), in order to assess whether
particular rock types are suitable for a particular application.
Èssentially, given the above categorisation, the manual proposes three levels of
specification in order to overcome the historie inconsistancies: common defïnitions
and approaches, güidance values for defined parameters, and methods of
achieving more economie use of availabte rock such as by Standard gradings.
Further information on the subject will not be given here, stnee a more extensive
paper on the subject is being presented at the Coastal Structures and Breakwaters
'91 Conference in November of this year (Sirrim, Latham and Orbell-Durrant, 1991).
However, in order to satisfy repeated queries on the likely Standard gradings to
be suggested, Figure 3 and Table 1 give an outline of the approach to defining
grading limits, and their recommended values, respectively.
4.2 Hydraulic Response and Stability
The manual presents many usefui syntheses of data of various matters of hydraulic
response including run up and run down levels, overtopping discharges, wave
transmission and reflections. A typical synthesised graph from a number of data
sets is shöwn for wave transmission in Figure 4 and represents a common
approach in the manual involving presentation of both the raw data and the
suggested design line. In relation to hydraulic stability, the manual of course
includes the extensively validated new formuiae by van der Meer (1987), but also
includes a number of other very usefui design graphs and formuiae (see for
example Figure 5). These stability formuiae include formuiae and models for
designing dynamieally stable slopes, (eg for berm breakwters) and for designing
scour protection.
4.3 Geotechnical Stabllitv
: As well as explaining the basis for various mathematica! models now being
developed to assess geotechnical stability in an essentially complex and
dynamically varying situation, simple guidance is also given on the subject of wave
penetration and the phreatic set up within mounds, siope stability, settlement,
liquification, filter rules and design aspects of geotextiles.
4.4 Design of Coast protection and Sea Defence Structures
The structures chapter of thé manual covers rubble mound bréakwaters, sea wails
and shoreline protection structures, dam face protection, gravel beachës and
rockfill in offshore engineering, but the sections on bréakwaters and sea wail
protection structures represent by far the largest part of the chapter.
Bréakwaters are considered first, principally because thé design of rock structures
has always begun with ah analysis of this caSe. thiS is because the incident wave
climate, if not a deep water condition, at least is generally unaffected by wave
breaking.
The section on Sea wajIS and shoreline protection structures then draws on the
fundamental principals Of design and construction of such bréakwaters and extends
and modifies them for the practical cases at the shoreline. Structures covered
ihclude cbhvehtional revetmejits or revetted mounds, scour protection to verticai
wails, gröynès, offshore bréakwaters, fishtari bréakwaters, L shaped and T shaped
bréakwaters, and cills Or submerged bréakwaters. Each structure is covered
both in térrhS öf its plan layout and its cross sectional desiga
The section on dam face protection has been included to assist water and dam
engineerS tö uSe the manual fof providing rock protection to the upstrearri faces
of earth or rock filled dams, recognising the peculiar constraints imposed by this
appircation. the involverneht of a reservoir safety panel engineer with the General
Reporter on this sectibn was greatly valued.
4.5 Maintenance & Monitoring
One great attractipn of rock structures to many coastal authorities, is the reduction
in maintenance costs which they often invplve, when compared w'rth for exampie
maintaining a timber groyne field. However the coastal authority must assess at
design stage whether pr not a maintenance commitment in terms of replacing
dispiaced pieces of rock armour for exampie, is going to be required, and make
appropriate provision at the commehcement of the project for materials, plant and
access for that plant, and finance. Even if a sufficiënt factor of safety has been
built into the design tp rendef maintenance an unlikely requirerhent, it would be
a foolish and irresponsible authority that did not monitor those structures on a
regular basis. Thus the manual not only gives guidance on maintenance
techniques but also on monitoring techniques and how these may be used to
decide when maintenance may be required, with specific guidance on such mattere
as survey technique (see Table 2).
Recommendations & Conclusions
5.1 Adoption of the design approach and design tools provided in the CIR1A/CUR
manual will enable a rational and standardised approach to the use of rock in
coastal engineering in the UK and throughout Europe. Further support for this
statement can be seen from the fact that the recommendations in the manual are
currently being considered by the appropriate committee on European
standardisation (CENTC154). Delegates to this conference are urged to encourage
the use of the manual to promote good design of rock structures and the
economie use of rock materials. Adoption of the guidelines and Standard
approaches set out and developed in the manual will also ensure that experience
and performance can be compared on a uniform basis from ohe authority to
another and one site to another.
5.2 Development of a centrally held databank (eg by NRA or MAFF), representing the
Itkeiy future need tor rock of various grades based on the capita! programmes df
the NRA and coastal prptection authorities, would enable UK producers to compete
much more effectively with continental producers and reduce end cost to clients
and the government.
6. Acknowledqements
Grateful acknowledgements are made to the funders of this project mentioned at the start
of this paper, and to a large number of talented and enthusiastic individuals who contributed
either as principal or supporting authors or as reviewers to the manual. Many individuals
could be named in this regard but space does not permit. Thanks is however given to Dr
JP Latham, Queen Mary and Westfield College, NWH Alisop of Hydraulics Research and JW
van der Meer of Delft Hydraulics for some of the examples presented in this paper.
7. References
CIRIA/CUR 1991 Manual dn the Use of Rock in Coastal and Shoreiine Engineering. In
publication.
SIMM, J.D., LATHAM, JP. and ORBELL-DURRANT, C. 1991 Standardising the approach
to specifying rock in coastal structures and breakwaters. Proc. Conf. Coastal Structures and
Breakwaters '91. Instn. Civ. Enors. London. November.
VAN DER MEER, J.W. 1987 Stability of breakwater armour layters - design formulae.
Coastal Engng., 11, 219-239.
Requirements Additional Information
for range of
Class ELCL LCL UCL EUCL effective mean Expected range for
Designation U<2) o<y<io 70<><100 97<V weight i.e. compliance with
(kg) excluding Standard gradings
where y is the % by weight pieces less
lighter on the cumulative plot than ELCL K • >eAS
Class Limit Definition by
weight, tyy (kg) '" -
HEAVY 3 00 - 1000 kg 200 300 1000 1500 540 — " 690 kg 595- 760 kg 2.3 - 3.8
GRADINGS 1000 - 3000 kg 650 1000 3000 4500 1700 - 2100 kg 1800-2200 kg 2.2 -3.6
3000 - 6000 kg 2000 3000 6000 9000 4200 - 4800 kg 42 00-4 80Ó kg 1.6 - 2.2
6000 - ïóoöo kg 4000 6000 10000 15000 7500 - 8500 kg 7500-8500 kg 1.4 - 1.8
LIGHT 10 - 60 kg 2 10 60 120 20' •-:• 35 kg 26 - 46 kg 3.2 - 7.7

GRADINGS 60 - 300 kg 30 60 300 450 13 0 - 190 kg 150 - 220 kg 2*8 - 6.0
10 - 200 kg(wide) 2 10 200 300 30 - 90 kg 70 - 130 kg 5.0 - 11.4
Average weight retaihed on LCL hole,
K <k9)
for rock density, pr (t/m3)
<2.5 2.5-2.9 >2.9
Class Limit Definition
by Square Hole (mm) min max min max min max
200 - 350mm 100 200 Not 350 20 40 25 45 25 50

3 50 - 650mm 200 350 def- 650 115 180 130 200 145 240
200 - 550mm (wide) 100 200 -ined 550 45 80 50 90 55 100
- • -
Note: The 10-60, 60-300 and 10-200 kg classes are equivalent to the 200-350, 350-650 and 200-550mm classes
respectively '
ASPECT OF STRÜCTURE STATE SURVEY TECHNTQJJE
(fuil details in Appendix A6)
LEVEL I LOCATION
2feö iö points on the strüeture Conventional survey techniqueis
itieasured in rélation to a (any settlement markers should
wèll-established grid and datum be installed during
levels. Redundant survey points construction)
needed if major movement
possiblé.
LEVEL II GEOMETRY
Outer surface descriptiön, -Conventional survey, using
related to level I survey profiling techniques recommended
points. in the model specification,
Appendix Al, but with profiles
at wider spacings say 20-3Om.
Fór underwater surveying,
bathymetric techniques may give
usefui Information.
LEVEL III COMPOSÏTION
Position and attitude of each Armour degradation inspection
piëCe Of rock armour, techniques
including unstable pieces. Comparative photography
Position and size of major Photogrammetry
voids and exposures of core For underwater surveying, side
or undërlayers. scan sönar techniques may be
used in co-ordinatioh with diver
surveys to identify features on
sonar traces
LEVEL IV : ELEMENT GOMPOSÏTION
Shupe and si2ê of armour rock, Armour degradation inspection
ing any fractufes technigues
Note: Level III and IV armour degradation inspection techniques

are difficult to implement on wide gradings and/or gradings with
D15 less than about 0.3m. The technigues, however, work
extremély Well on large "single sized" rock armour
(D15 > Ö.3m; D85/D15 in the range 1.2 to 1.4) including the
"heavy" Standard gradings in Appendix Al
Table 2. Measures of the state of a rock structure
©
M A N U A L CONTENTS DESIGN SUB - PROCESSES MAIN DESIGN PROCESS
PLANNING POL1CY PROBLEM

PLANNING & DESIGN1NG
•r DEFINITION IDENTIFICATION
MATERIALS OATA COLLECTION

T"
DEFINITION OF
[ BOUNDARY CONDITIONS
PHYSlCALSiTE
LJL L
Ff*
CONDfTlONS & ENVIRONMENTAL
DATA COLLECTION ASSESSMENT
__ J^.—J * ^
FUNCTIONAL
ANALYSIS
PHYSICAL PROCESSES DETERMiNATlON OF .

&DES1GN TÖÖLS
H
MATERIALS SOURCES
AND PROPERTIES u
GENERATION OF
ALTERNATIVE
h
n
SOLUTIONS
STRUCTURËS
L^ USE OF HYDRAULIC
AND GEOTËCHNICAL
DESIGN TOOftS
MAINTENANCE COMPARISON AND
n
SELECTION
STRUCTURE - SPECIF1C
•4 DESIGN METHODS
FINAL DESIGN
App.AI AND DETAILING
BENEFIT - COST OR
ROCK
MULTI - CRITERIA
SPECIFICATION
ANALYSIS
APB-A2
CONSTRUCTION
ROCK MATERIALS
STANDARDS ! 1 RISK ANALYSIS
r-
App.A3
METHÖD FOR
BILLS OF QUANT1TIES
1_. MATERIALS
SPECIFICATION ft MAINTENANCE
L
App.A4,A5
IN5TRUMENTATI0N
u PREPARATION & USE OF
BILLS OF QUANTITIES
REPAIR /
REHABIL1TATION
1 FOR DATA COLLECTION
App.A6
STRUCTURAL ABANDONMENT/
KEY
MONITORING DATA FLOW REMOVAL
ACTION FLOW
Figure 1. Lbgic diagram for Manual on Use of Rock in Coastal and Shoreline Engineering
Intrinsic Property Execution Induced; Functional Requlrements
Property Property
Colour
Armour
- Resist hydrodynarfiic loading (causing
Density Armour rocking, rolling, sliding, lifting)
Interlocking * Dissipate wave refleetions,
. Porosity - Reduce run-up, toe scour
Water Ab. - Layer thicknêss' - Environmental, recjreactional, aesthetic
/Porosity Roughness of profile - Strength to bear Iqads
— » • " Block Integrity
-r Bulk density
- Retention of functions with time
Discontinuities "
in-situ
bLock size t~ Block Weight eg •W 50 -*H
block shape
Weathering
i
Grading eg. Was AA/, 5 -*t~
^—)
Uriderlayer
Shear strength
internal friction
Underlayer
- Protect internal firje materials from
grade migration + piping
•
boundary friction - Resist interlayer sljding
Permeability - Regulate pore prössures
Intact strength Block shape eg. % l/d >Z1 -4-J Bulk density - Strength to bear Ipads
mineral fabric
resistanee to
Breakagè - Retention of functions with time
Abrasion Core
Dilatancy
- Shear strength
Gofie
Grain size internal friction
roughness of - Regulate wave transmission
boundary friction
fracture/block - Resist settlement and piping
Cohesion - Strength to bear loads
surfaces e
Permeability o
(3
o. 1) 1
S 1
J
c3
'
,
Bulk density -g)
CC
•
- Retention of functions with time
- <3-
ra
l
c
n) 1i |
3 1 C 'ft
a 'Ï
100
97 f- -
70
50
% by
weight
lighter examples of
grading curves
fulfilling ciass '
limit requirements
10
2
0
ELCL LCL UCL EUCL
weight of rock fragment V\fy (kg)
Figure 3. Explanation of the grading ciass limits for a Standard grading

! Transmission coëfficiënt K
i
CD
3
CO
3.
5" D
3 <"
1 CD
O
a ~>
I rr
CD.
yo
fa
• &
sa x
3
E o
§ o
^3
Toe stability
a 0 - 3 % DH
depth limited conditions x 3 - 1 0 % DH
3 - 1 0 % DHI
- * > 2 0 % OH
suggested design curve O > 2 0 % DHl
- -r - SPM <Hs)
- — SPM (H10)
.6-
J *
.4-
/ •
—T-
3
H S /AD
s / " u n(5 0
Figure 5. Toe stability
-f5
KANOAt, ON THE USE OF ROCK
IM COASTAL AND SHORELINË ENGINEERING
CONTENTS
Forewprd
Acknowledgements
Glossary
Notation
Chapter 1 Introduction
1. i . BacJcgróuhd use and approaéh
1.2 Use of the Mariual
1.2.1 Design process
1.2.2 Structure and contents of Manüal
1.3 Structure types covëred by Manual
Chapter 2 Planning and Designing
2.1 Design process
2.1.1 Pröblëm identificatión
2.1.2 Böundary conditiöhs
•Planning policy •Aóceptable risk
2.1.3 Functiönal Analysis
2.1.4 Gerieratioh of alternative sólutions
•Enyironniental Assesswënt «Materials
availability and properties •ünderstanding the
hydraulic and gëoteehnical pröcessës
•Structure-specific design method's
•Constructioh considèratións *Maintenance
considëratiöns
2.1.5 Coroparison and seièction
2.1.6 Final design and detailing
2.1.7 Cost Assessment
2.1.8 Quality Assurancé
•Quality Cöntröl
2.2 Structure types and failurè modes

2.2.1 Structure types
2.2.2 Principal failüre modes
2.2.3 Fault tree analysis
• D e f i n i t i o n of daaage «Fault t r e e systems and
mechanisms
2.2.4 Risk l e v e l assessment
2.3 Design approach
2.3.1 Deterministic approach
2.3.2 Probabilistic approach
•Level III methods •Level II methods «Examples
bf probabilistic calculations
2.4. Environmental assessment

2^4.1 Reasons for carrying out EA
2.4.2 EA as an intégral part of the design process
2.4.3 Method of assessment
2.4.4 Additional sources of inforroation
2.4.5 Assessing the impacts of the proposed works
•Geomorphology, landscape and existing physical
processes «Ecology «Social and socio-economic
aspects «Human sensory aspects
•Palaeontologicaï, archaeological, historial
and cultural aspects «Air, water and soil
guality
2.4.6 Impacts from the choice of materials and
construction
•Sources of material «Use of waste materials in
shoreline works •Construction
2.4.7 The Environmental Statement (ES)
Chapter 3 Materials
3.1 Rock types
3.1.1 General considerations
3.1.2 Evaluatipn of rock types at source
3.1.3 Alluvial, glacial and marine sources of rock
materials
3.2 Properties and functibns
3.2.1 Intrinsic properties
•Colour -Density «Water absorption and rock
p o r o s i t y • D i s c o n t i n u i t i e s «Weathéring grade
• I n t a c t f a b r i c strength
3.2.2 Production affected properties
«Block integrity «Block shape «Grading «Block
weight and size
3.2.3 Execution induced properties
• Interlocking and porosity «Interna! friction
•Roughness of layer topography «Implications
for porosity, dènsity and layer thickness
3.2.4 Durability and armourstone degradation rates
•Design Implications «Rock deqradation model
for armourstone
3.3 Testing and evaluation
3.3.1 Introduction
3.3.2 Size and weight gradations
• P a r t i e l e s i z e d i s t r i b u t i o n « P a r t i e l e weight
distribution
3.3.3 Slâpe
3.3.4 Petrographic examination
3.3.5 Density and water a b s ö r p t i ü n
•Oven dry mass d e n s i t y «Water Absorption a t
atiüospherib JcJressure
3.3.6 Iirtact s t r ë n g t h
• F r a c t u r e t o u g h n e s s • Inipact R ë s i s t a n c e
•Crüïshirig Resistance «Abrasion iResistance
• DëförTBabillty, hardness and s ö n i c v e l o c i t y
3.3.7 R e s i s t a n c e t ö weathering
•Sulphate sdundness t e s t »Freeze-thaw t e s t
•Breakdown due t o clay a i n e r a l s
3.3.8 Block i n t e g r i t y
Production a'nd hanüliing
3.4.1 Rësbürdë : appisaijsal 1
'«*erirëseria:i Sö**r=ce's '««arine S ö ü r c e s ' .
3.4.2 -PïröauCtiön
'•••lEgpës, o f i'güaifc'ty. ''•fj&rmbtïrssttone ;;Bl3a®t'-<ïe"&ign
• P r ë d i c t i o n öf,; 'tihe r ë s d l t s ali b i a s t ' i n g
•Sëlëdtiibn in Öië qua*r|r >;Handl4ng •StöCking
• Loading öf vëtvicles «Üöstt i m p l i c a t l d n s of
;pTröduotïdn
, , ,
,• '
3.4.3 ; i.-*FFari s p 'ottat-±:ön
•cdst impïicabïöns öf transpartoaciön
3.4.4 Gravel rësöurees
•Gravel drédgërs «Discharge of gravéls «vCos^fcs
bf gravel extraütiön
Othër materials
3.5.1 Concrete
3.5.2 ï n d ü s t r i a l by-rpirbdücts
3.5.3 Grbüts
• I n t r o d ü b t i o n «Asphalt g r o ü t « C o l l o i d a l cement
mbrtar
3.5.4 Gedtextiles
• I h t r ö d ü c t i ö n «Main funsctJions and a p p l i c a t i o n s
•Produdtidn tedhriology • C h a r a c t e r i s t i c s and
'prqper-tïes
3.5.5 Cdmposfte n r a t é r i a t s
• Gctbioiiis •"Mattrësses
Quality Central
3.6.1 Öbjective
•Armourstone Stocks «Quairisy f aces and
productiöh «iritëgrity of blöóks «Rock guality
tests «Bldck shape «Grading «UBdeflayer and
cöre mötdrials
3.6.3 Quality cohtrol during construction
•'Bulk placement df core matërials • Individual
placement of armdur
Chapter 4 Physical site Conditions and Oata Collection
4.1 Bathymetry and morphology
4.2 Hydraulic boundary conditions and data collection
4^2.1 Water levels
•Mean water level «Tides «Storm Surges «Wind
set-up «Wave set-up • Seichas • Tsunami • Flood
waves •Design extreme water level
4.2.2 Deep Water wave conditions
•Statistical properties of waves • Long term
distribution of wave height and period «Visual
wave observations •Wind statistics • Spèctral
description of waves «Sea development during
storm and wave forecasting «Wave forecasting . in
inland waters and reservoirs «Deep water wave
breaking and diffraetion
4.2.3 Transformation of waves in shallow water
•Refraction «Shoaling «Diffraction «Wave
breaking «Extreme wave heights in shallow water
•Reflection «Wave-current interaction
4.2.4 Sea bpttpm orbital velocities
4.2.5 Currents
4.2.6 Ship induced water movements
4.2.7 Joint probabilities
•Design with joint waves and water levels
•Design with joint waves and currents
4.2.8 Modelling
•Numerical modelling of water levels and
currents «Physical modelling of water levels
and currents «Numerical wave modelling
•Physical wave modelling
4.3 Geotechnical boundary conditions and data collection
4.3.1 General aspects of planning a soil
ihvestigation prpgramme
4.3.2 Desk studies
4.3.3 Preliminary soil invéstigations
4.3.4 Detailed soil invéstigations
Chapter 5 Physical processes and design tools
5.1 Hydraulic interactions

5.1.1 Governing parameters
•Hydraulic parameters « S t r u c t u r a l parameters
5.1.2 Hydraulic response
•Wave run-up and run-down «Overtopping
•Tfansmission «Reflections
5.1.3 Structural response
•Armour layers «Underlayers and filters «Crest
and rear armour {low-crested structures) «Crown
walIs «Tóe protection •Breakwatér head
•Dynamically stable slopes {berm breakwaters)
•Stepped and composite slopes •Composite
systems «Longshore transport «Combined current
4$
and wave attack • Scour
5.1.4 Numerical and physical model1ing
5. 2 Geotechnical interactions
5.2.1 Approach
5.2.2 Parameters
•Cohesion c ' « P o r o s i t y n • P e r m e a b i l i t y K
• F r i c t i o n angle <t> «Dilatancy
5.2.3 Scope of g e o t e c h n i c a l a s p e c t s
5.2.4 Physical background
• Loading and boundary c o n d i t i o n s •Geomechanic
p r i n c i p l e s «Porous flow « S l i p f a i l u r e
•Deformation • I n t e r n a l e r o s i o n / f a t i g u e
• Liquefaction «Dynamic e f f e c t s « C a v i t i e s
5.2-5 Hodéiiing and s i m u ï a t i o n
•Mathematical background of a v a i l a b l e models
• Methods and Models
5.2.6 Computer model types
• S o p h i s t i c a t e d models «Simplified (uncoupled)
models
5.2.7 Computer model a p p l i c a t i o n s
•Porous flow •Deformation « S t a b i l i t y
5.2.8 P r a c t i c a l formulae and e n g i n e e r i n g experience
•Wave p e n e t r a t i o n « I n t e r n a l p h r e a t i c s e t - u p
•Dynamic excess pore p r e s s u r e «Slope s t a b i l i t y
•Cover l a y e r s t a b i l i t y «Earthquake e f f e c t s
•Settlement of compressible s o i l •Settlement of
r o c k f i l l «Local s t a b i l i t y of r o c k f i l l
• L i q u i f a c t i o n p o t e n t i a l of seabed sand
• Squeezing of very s o f t s o i l « F i l t e r r u l e s
• Design a s p e c t s of g e o t e x t i l e s
5.2.29 Application of p r o b a b i l i s t i c a n a l y s i s
Chapter 6 S t r u c t u r e s
6.1 Rubble mound breakwaters
6.1.1 Definitions
6.1.2 Layout
•Influence of need • for berth protection •to
provide protection to access channel «to reduce
maintenance dredging costs
6.1.3 General design considerations for breakwater
cross-section
•Cross-section concept generation, selection
and detailing «Data collection and boundary
conditions «Materials availability
•Construction considerations
6.1.4 Structure-specific design aspects
•Conventional rubble mound «Rubble mound with
monolithic crown-wall «Berm of S-slope
breakwater «Reef breakwater •Low-crested and
submerged breakwaters «Caisson type breakwater
6.1.5 construction aspects
•Land based operation «Seaborne transport and
placing «Combination of land based and seaborne
operation «Placing tolerances
6.1.6 Cpst aspects and project optimisation
•Production aspects of the cpst Of breakwatèrs
• Cost aspects related to activities ta the site
of the structure •Examples of cost optinisation
in design and constructiön
Sea walls and shoreline protection structures
6.2.1 Definitions
6.2.2 Plan l a y o u t a??d o v e r a l l concept s e l e c t i o n
• F u n c t i ó n a l requirements «Position of s h o r e l i n e
• Layout opfcions and design «Effect of s e l ê c t e d
c o n c e p t / l a y o u t on beaches/coas1;lines
6.2.3 General c o n s i d e r a t i o n s for c r o s s - s e c t i o n design
of rock c o a s t a l s t r u c t u r e s
• F u h c t i o n a l requirements of c r o s s - s e c t i o n
• P h y s i c a l boundary conditions • M a t e r i a l s
a v a i l a b i l i t y «Material supply «Constructiön
c o n s i d e r a t i o n s «Maintenance c o n s i d e r a t i o n s
• s e l e c t i o n and design of annour l a y ë r
(revètment)
6.2.4 Structure-specific design aspects
•Conventiorial revetment or r e v e t t e d mouhd
(dike) «Scour p r o t e c t i o n of v e r t i c a l wall
•Bastion groyne «Offshorebreakwater
• F i s h t a i l e d breakwater «L-shaped and T-shaped
breakwatèrs « S i l l pr submerged breakwater
6.2.5 Constructiön aspects
•Handling and placing of annour s t o n e
Dam face p r o t e c t i o n
6.3.1 P r o t e c t i o n concept s e l e c t i o n
• P l a n n i n g - p o l i c y type boundary c o n d i t i o n s
•Material sources
6.3.2 General d e s i g n c o n s i d e r a t i o n s
• M a t e r i a l s - s p e c i f i c a t i o h and t e s t i n g
• P h y s i c a l s i t e conditions and d a t a c ö l l e c t i o n
6.3.3 Special design considerations
6.3.4 Special c o n s t r u c t i o n aspects
6.3.5 Measurëment and cbst aspects
6.3.6 Maintenance a s p e c t s
Gravel beaches
6.4.1 Design process
6.4.2 Coastal processes
6.4.3 Gravel beach profile
6.4.4 Construction and cost aspects
6.4.5 Monitoring and re-nourishmentl
6.4.6 Cost optimisation
•Influence of lpcation of borrow area and
grading of borrow material «Site conditions
during constructiön «Volume of gravel required
per m beaqh length
6.5 Rockfill in o f f s h o r e engineering
6.5.1 Principal design considerations
• D e s i g n appróach «Hydraulic s t a b i l i t y of
r o c k f i l l « G e o t e c h n i c a l s t a b i l i t y •Impact of
f a l l i n g o b j e c t s «Dragging a n c h o r s and f i s h i n g -
g e a r • Freespans «Upheaval b u c k l i n g
6.5.2 Construction aspects
• C o n s t r u c t i o n methods «Impact o f dumped rock
•Survey
6.5.3 Cost aspects
Chapter 7 Maintenance
7.1 Monitoring
7.1.1 Types o f Monitoring
7.1.2 Freguency o f Monitoring
7.2 Appraisal o f S t r u c t u r e Performance
7.3 Repair/Replaceraent C o n s t r u c t i o n Methods
Appendices
Al Model s p è c i f i c a t i o n f ö r quarried rock a p p l i c a t i o n s in
c o a s t a l and s h o r e l i n e e n g i n e e r i n g
A2 Standards f o r q u a r r i e d rock m a t e r i a l s a p p l i c a t i o n s in
c o a s t a l and s h o r e l i n e e n g i n e e r i n g
A3 Measurement o f q u a r r i e d rock i n c o a s t a l and s h o r e l i n e
engineering
A4 Hydraulic d a t a measurement and i n s t r u m e n t a t i o n
A5 Geotechnic d a t a measurement and i n s t r u m e n t a t i o n
A6 Structure monitoring technigues
A7 European, B r i t i s h and Dutch l e g i s l a t i o n / a u t h o r i t i e s /
d e s i g n a t e d s i t e s r e l e v a n t t o e n v i r o n m e n t a l a s s e s s m e n t of
p r o j e c t s i n v o l v i n g t h e u s e of rock i n c o a s t a l and
shoreline engineering
References
Index
International Workshop on Riprap, Colorado, USA, 1993
QUARRY BASED DESIGN OF ROCK STRUCTTTRES
by W. Leeuwestein (*) , A. Franke, V. Hombergen, J.K. Vrijling (**)
(.*) CUR, PO.Box 420, 2800 AK, Gouda, The Netherlands
(**) Rijkswaterstaat, civil Engineering Division, PO Box
20000, 3502 LA Utrecht, The Netherlands
1. introduction
Rock is a widely used material f or hydraulic structures in marine,
estuarine and riverine environments. The reason for that is mainly that
many types of rock can meet reguirements with respect to density, size and
strength and besides that this material is available in many places in the
world.
A Europian initiative2 was taken by two research organisations, CUR' in the
Netherlands and CIRIA in England, to support an optimal use of rock in
marine structures, by prodücing a Manual on the use of rock (CUR/CIRIA,
1991) . Construction and material use are the two principal variables for
design optimisation, of which material use is subject of this paper.
Characteristic control parameters are stone size (D) and grading (Dgs/D^) of
the rock.
In this paper it is demonstrated that there is a fair chance that by using
conventional procedures of optimizing the design of a rock structure, the
possibility of significant savings on material cost are ovèrlooked.
First a conventional lifetime-cost optimisation is presented with respect
to the size of the armour stone (demand-based design). This is done for two
examples, a breakwater and a river bank protection. Subsequently, an
optimisation is presented with respect to a given quarry production
(supply-based design). For the breakwater this is done using a optimisation
model developed in the Netherlands and described in this paper.
2. Materials, sources and production

Principal sources of stones are quarries and marine or river deposites of
gravel. Quarries are located on land, often in mountaneous regions, and
rock material is produced by blasting vertical slices from a rock or
móuntain. The result is often a more or less irregularly shaped material of
different sizes (D) and weight (W). Unless defined l/3
otherwise, in this paper
D is the nominal diameter, related to W by D=(W/pr) with pr being the rock
density. In practice, for D, the 50% nominal value D^ is used. The variety
in size of the stones produced is characterized by a stone size
distribution curve F(X)=Pr{D<X), which is known as the production curve or
so called Quarry Yield curve. In Figure 1 two examples are shown of which
only type I produces thé typical heavy armour stones. Alternatively,
corresponding density curves can be used (eg. see Figs. 5 and 7 ) .
By means of a blast design one may attempt to control the production.

Depending on the demand by the structure, large size stones or large
quantities of the smaller sizes may be wanted. The percentages of the
various size categories neëdëd are gouverned by stability requirements and
filter rules discussed below. Blast design is mainly a matter of specifie
experience, but some guidance nas been developed using placement and
spacing of boreholes. Methöds to assess the production by means of
distribution functions, F(X), are presented by Rosin-Rammler and Schumann
(CUR/CIRIA, 1991). The link with blasting practice is provided by
parameters, which are taken from the blast design in the quarry.
^•Centre for Civil Engineering Research and Codes

2
Construction Industry Research and Information Association
quarry-based design, W. Leeuwestein et.al. 1

100
90
80
Y
7TTT1
70
quarrylf
m quarryl
60 V^
/ ~pr
50
40-
30
20-
10-
100.0 1000.0 10000.0 100000.0

stone weight, X Ikgl
Figure l Example Quarry Yiéld curves

For practical reasons, quarry owners and contractors preferably use a set
of Standard gradings, definedby (at least) an upper and lower limiting
weight (or size). In the following secticns a set of gradings are used,
which follow the standards presented by CUR/CIRIA.
3. Structure characteristics
The stractures considered contain or consist entirely of rock material. At
the wave and/br current exposed (usüally sloping) face, this material
extends froiu the crest down to the toe and sometimes even further.
At the crest, the rock material can be extended to a lëe-side slope or
(horizontally) over a backfill area.
In this paper a conventional rubble möund breakwater and a river bank
protection are considered, with crest height and slope gradiënt being the
main characteristic geometrical design parameters.
In general, armour layers (primary and secondary), filter layers and a core
are distinguished, all mainly charaterized by a stone size (D). These stone
sizes are determined by hydraulic stability. Regarding to wave attack, a
structural porosity factor (P) is defined (see Appendix). A commonly used
armour layer thickness (ta) relates to D according to
t. = 2 D„, (3)
The same relation is also used for other layers. The size of the material
for filter and core are derived from requirements related to internal
stability and filter functions (filter rules). The latter is commonly
formulated in terms of a size ratio of the upper (armour) and the lower
(filter) material (D/D,) , for example in the overall filter rule by
Terzaghi (1922):
tWDsof = 5 (4)
The above, simplified, requirement is used in this paper. This is valid for
uniformly sized material, whereas for wide graded rock values up to 20*60
apply (CUR/CIRIA, 1991).
4. Functions and requirements

The function of rock in structures such as breakwaters and river bank
protection is protection against exteijial actipn by waves and/or currents.
When extended to the sea or river bed in front of the toe, the fünctioh can
be bëtter described as scour or bottöm protection. When extended on an
inner slope or a backfill area, the function is to resist ov'ertopping.
In all cases, the rock protection supports the basic function of the
structure such as reduction of wave disturbance (breakwater) and slope

protection (river bank protection), whereby a certain accepted degree of
damage to the protection may bè inherent to an economie design aiming to
minimize the cost of investment and expected damage.
As an example, Figure 2 shows a conventional optimisation with respect to
armour size (Da) under wave action, with wave conditions from section 5.
Damage is calculated as the displaced rock volume (Sd) according to Van der
Méer's formula, see appendix (CUR/CIRIA, 1991).
1 1.2 1.4 1.8

Annour, Size, Da {m]
Figure 2 Conventional optimisation of breakwater armour

investment vs. damage
lay-out
A varièty of functions determines the alignment and length (breakwater) and
and extent (toe, bottom protection) or situation (river bank protection) of
rock structures. Also local bathymetry may play a role.
The bank protection discussed in this paper is designed for a braiding
river. This implies that river channëls and flood plains show a rather
irregular pattern, which chahges from year to year under influencë of flood
waves. As a consequence, scour and the moving chaiinels largely determine
the lay-out of the bank protection.
cross section
Functional requirements play an important role in the design of the cross
section of structures. Most important for breakwaters are wave
transmission, wave run-up and overtopping. The crest height is largly
determined by these parameters. Besides stability requirements (geotechic
and hydraulic) often determine slope gradients and the extent of a toe
protection. In this paper, the height of the transmitted wave (H,) and the
combined flood level (zf) and run-up are the determining requirements for a
breakwater and a bank protection respectively.
In this paper the crest level of the breakwater is determined by the
requirement that the maximum wave height in the area protectéd by the
breakwater be limited to H ^ = 1 m. Construction further requires that the
crest width be B = 5 m. The height of the river bank considered is
determined by the requirement that crest is no't overtopped during the
design flood conditions. The slope is 1:3.5.
For both structures the depth of expected scour at the toe determines the
construction (toe) depth.
hydraulic stability
Much emphasis has been put in the past on reduction of stone size for a
cover or armour layer, given a design wave or current. Proven'stability
requirements are given by Shields (193 6), based oh shear stress and
applicable for currents, waves or combinations of these and by Isbash

(1959), based on velocxty and applicable for currénts only. On both
requirements, roughly summarized herè as the stability criteria ^ w = 0.03
(Shields) and U/2gAÖ' = 1.4 (Isbash) reference can be found in Pilarezyk's
paper. Here it is only useful to mention that experience has shown that in
fact the above given threshold values are only coraraon values. The observed
variations can be indicated as 0.03 * 0.05 and 0.7 * 1.4 respectively,
which can be partly explained as to be caused by differences in: a)
exposure or b) probability of initial displacements (damage) or c}
deff initions.
current attack
The threshold velocity (U) can bé written in terms of Shields' bottom shear
stress (i[t) by:
U 2 /2g C2
= *,. — (3)
A Do 2g
where C is Chezy's friet ion cpefficient, note that C=f(B,..) and kal is the
slope factor (see Appendix) . An unambigious definition of damage for
current attack is still not available. However, in this context damage may
be roughly estimated by using TJTCT as a damage parameter (a role comparable
to Van der Meer's Sd, see appendix). By fitting of data by Gessier (1965)
and interpreting his probability of displacement as a damage fraction (SJ ,.
a practical relationship between ya and s„ has been found to be:
S u = 3.5 ($„ - 0.0198)"-5 (4)
(In the range of ^ between 0.02 and 0.10 the deviations from Gessiers'
data in terms of damage are about 0.05).
wave attack
For stability of sloping rock under wave actión Van der Meers* formulae
have been gradually introduced since 1984 (CUR/CIRIA, 1991) as an
alternative for the well known Hudson formula (1965>. The main xmprovements
are that account is given to wave period (T), storm duration (N> and
structure porosity (P) and that a clear definition of damage (Sd) is
provided.
5. Design conditions
breakwater
In this paper the design conditions for the breakwater are wave height (H)
and period (T) and for the bank protection f lood level (fa), wave height (H)
and current velocity (tl). The design values are derived from fitting of
prototype data into a suitable long-term wave height distribution function
(Figures 3) .
1= s s s
I Ss.
••^S^ , M „ j
» ... '.( ..~~*—, ;; 'i;,,,,;1, =
. 'X. "i - :I :
• ^
f £
^ «V
0.1=
:~—:.
• . . .
I^2^ï H
1 i
=c===ï L~ """-Si ^r.i.-::-:n-rj::;ryrn: =

ï *.», ! I-
*^ != ""î
^ < ; ï
~*V—r ==#?^
*\ I =ÊÊ
i- • !- _. i ~Ï*,<V
*1% ' = f
ë 0.001: =— -4-r-r • f \ sunwner

r.-a :,.!,-
=
' i
• • f1' i, , ! '*
0.0901-
2 2.5
—t
3 3.5 4
i — • —
4.5
_|
5
1—
5.5
j^.6
waveheightH[mI
Figure 3 Example distributions of significant wave height

The brèakwater is designed for the 1:50 wave conditions, which iropliës a
significant wave height of Hs = 4.8 m and the corresponding meari wave
period is assumed to be T m = 9 s. The design wave height at the lee side is
H, = 1 ïu- The number of wavés in the design storm is N = 5000 and the
accepted damage (Sd) is based on a preliminary optimisation përformed in
séction 4; Tides can be disregarded.
bank proiection
For the bank protection, the 1:100 years design river discharge is Q =
20,300 m3/s. Corresponding flood level and current velocity are zf = 7.8 m
and U = 2 m/s. The average water level is zav = 1.6 m and the lowest water
level is only slightly beneath this value. The stage curve with
corresponding current velocities and the discharge distribution are shown
in Figure 4.
Ünder 1:100 years wind conditions the wave height and period are Hs = 1.0 m
and T m = 3.5 s respectively. Due to (local) S C O U T the bed level under
design f lood conditions is at zb == -34 m.
The expected damage due to the current velocity (ü), on the lpwer part of
the slope, is evaluated using the Shields parameter (i|»CT) as a damage
parameter.
0.1 8-
V
\ N
water level, z [m]
v
*- -
•N.s •v
ntVelc)city
0
\
N
1
3
o
-5 4-
I1 °
2-
. • ' •
i — i
vetocrty, U [m/sj
0.01 i
1&5 17.0 17.5 18.0 18.5 19.0 19.5 20.0 20.5 1- i
Dttcharga. 0(1000 m37«] 18.5 17.0 17.5 18.0 1&5 19.0 19.5 20.0 20.5
Discharge, Q [1000 m3/s]
Figure 4 Example discharge distribution (left) and stage arid velocity

curves (right)
6. Demand-based design
The more conventional cross-sectiorial design can be characterized as a
"demand design", which is basically described below. Lay-out and alignment
are not further considered in this paper.
preliminary optimisation of brèakwater armour size
Using a simplified approach forthe wave transmission (CUR/CIRIA, 1991), it
can be found that for a crest height of hc = 15 m the relativé freeboard is
RyHj = 1.04 with an associated transmission coëfficiënt of C, = 0.15 (see
Appendix). The resulting transmitted wave height is H, = 0.7 m, which
remains well within the given requirement (H^) ,
For the slope a preliminary gradiënt of 1:2 is chosen for both the front
and lee side. However, in section 7 these slope angles are adjusted to
optimise the design. Van der Meer's porosity factor is P = 0.4 (see
appendix).
Expected damage (Sd) is used as a design criterion for which values
exceeding Sd = 10 are not acceptable since these imply exposure and
consequently (progressive) damage to the secondary and other layers.
quarry-basëd design, W. Leeuwestein et.al. 5

Material cost ratés for construction and repair are assumed as given in the
appendix. The structure's lifetime (TL) is T^ = 25 years.
Based upon these considerations and calculations for a series of (cover
layer) stone diameters (D) a preiiminary optimum armour size appears to be
D, = 1.5 m approximately (Figure 1). The corresponding damage during design
conditions is Sd = 6 approximately. Applying a similar procedure to the lee
side slope (with H, = 1.0 m) an armour size of D,L = 0.3 m is found.
evaluation of rejgöiretf s t o n e v o l u m e s f o r b r e a k v a t e r
In order to facilitate the evaluation of stone volumes and a further
iterative design of the breakwater cross section a design model has been
used which is briefly described in section 9. This model enables evaluation
of stone volumes associated with parameterized Standard geometries,
including toe structure.
The initial design was based upon the conditions described in section 5
with Sd=6 as the damage criterion. The results are summarized in Table 1,
listing stone volumes (per running meter) needed for the primary and
secondary armour layers and for the corè. Evaluation of costs is done by
using the material cost rates given in the Appendix.
part of stone size volume grading cost

breakwater D Did V Cm3] lieg]
prïü!. annour 1.52 90 •6/10 t 1440
idem, leeside 0.34 3 10/200 12
sec. armour 0.30 95 10/200 380
core 0.06 120 50/150 mm 240
sub-totals = 310 = 2070
prim. arm. toe 0.86 30 1/3 t 480
sec. arm. toe 0.29 7 10/200 28
fi tter toe < 0.07 45 50/150 rtm 90
totals ~ 390 = 2670

Table 1 Required minimum stone volumes (per running m) for the breakwater
obtained with the model
Armour stability together with filter requirements determine to a large

extent the demand curve for the rock material to be used in the structure.
Using the production from the quarry characterized by I in Figure 1, the
actual approximate blasted rock volumes can be determined. The produced
(Pr) and demanded (De) relative volumes (in %) are listed in Table 2.
totals
<10 10/200 '0.2/lt 1/3t 3/6t 6/10t >10t
Pr 20 18 28 16 12 4 2
De 45 26 0 7 0 22 0
Pr/De 0.45 0.69 (O 2.27 <D 0.18 co
E •:. 0 3.30 4.07 5.53 5.09 0 5.53
R 5.00 4.50 7.00 4;00 3.00 .. T • • • ' • 0.50
V 450 400 620 350 270 90 50 , 2230
cm 900 1600 10000 5700 4300 1400 700 24600

Table 2 Quarry production and cost per runningrafor design-based breakwater,
including toe as evaluated with the model
The results for the matching ratio (Pr/De) show that the heavy stones (6/10
t) have the lowest valuë of Pr/De and thus are the determining size.
Defining the production multiplication factor as:
fP = 1 / min{Pr/De} (5)
this involves a production of fp=22/4=5.5 times all grading volumes, to

correspond to the determining volume, Pd = 90 m3 of heavy stones (6/10 t) .
The relative excess production, defined for each grading as
E==(fP Pr - De)/Pr, (6)
the product ion ratio (R) of each relative grading volume (Pr) with the
determining volume (Pd) , actual produced volumes (V, in m3) and the the
consequent material production cost (CJ are shown in Table 2 from which it
can be seen that the total cost amounts to 9.2 times the theoretical
minimum cost of 2670 from Table 1. A practical measure for an economie
design from, regarding material use, is the cost (efficiency) ratio (P^}
between thé actual cost associated with a design and theoretical cost of
the minimum design (here, ^ = 9 . 2 ) . Similarly, a volume (efficiency) ratio
is defined as R v (heré R v =5.7).
In this example case most of the excess production cost is made for the
three successive gradings from 200 kg to 6 t. The (mis)match of demand and
supply volumes for the considered gradings is shown in Figure 5 (left).
10 1000 6000 200 3000 10000

grading upper limit Ikg) grading upper limit [kg]
Figure 5 Matching of relative demand and production

preliminary (left) and optimal (right) design
river bank protection

The crest level is primarily determined by the design flood level (zf = 7 . 8
m, see section 5 ) . Since serious wave action does not coincide with the
flood season overtopping is no additional consideration with regard to the
crest level. Provisionally it is assumed that the slöpe is 1:3.5 and that
under design conditions the bottom level at the toe, including scour, is at
quarry-based design, W. Leeuwestein et.al.

zb = -34 m. (disregarding othêr alternatives 3) .
F6r the upper part of the slope pratection, waves are the deterraining
ïöading while for the lower part the current velocities are. The design
formulae used dre Van der Meer's (Appendix) and eq.(3) respectively. In
order to account for Wave run-down, the transition is chpsen at 1.5 m (=1.5
H,) below thé average water levél (so at z, = 0.1 m ) .
As a result, the upper part is designed against the wave (with H5 = 1 m)
described in séction 5, whereas for the lowèr part (from zt = 0.1 ra down to
. a s ~ -34 m) the giveh velocityj of u = 2 m/s is used with an average water
depth for the slope, say h = 15 m.
With the material cost rates used in this paper (see Appendix) the cost of
damageare sraall relativë to the constructiön cost, at least for stone
sizes D > 0.04 m. Using the above given design conditions and the damage
curve accordihg to eq. (4) , any damage can be avoided by chosing D = 0.05 m
(see Figure 6).
"X N n
X •
i
Oata Gessier
X V.
I . v
appr.eq.(4)
i sL
• i. - - H
t
f ^ ^
0- ï
10 15 20 25 30 35 40
i' 45
i 50
i 55
Stone stee, D [mm]
Figure 6 relativë damage of bank protection due to the design flood
Applying eq.(1) for the layer thicknesses the material volumes and the
associated cost can be calculated as listed in Table 33. Total required
stone volume and Cost of this minimum design are 34 m and 88 respectively.
Grading Diameter Volume Cost

D [ml V[ms]
upper lower upper lower 2 V upper lower 2Cm
30/60 mm 0.01 7 7 14 14
0.3/10 kg 0.04 0.05 8 9 17 16 18 34
10/200 kg 0.22 10 10 40 40
totals: 18 16 34 56 32 88
Table 3 Required st one vol Line•s (per rurining m) f sr the min jmum design b,ank protec t i o n
3
An pption is to provide the stone for the lowest part as a falling
apron. Then usually a surplus of material is placed (CUR, 1993).

Produced volumes and assocxated costs are listed in Tablè 4. Comparing wxth
the minimum design from Table 3, it becomes clear that this design for the
bank protection dóes not match the given quarry yield curve at all. Volume
and cost efficiency ratios are Ry = 4.7 and B^ - 20.2.
stone class Ckg] totals
<0.3 0.3/10 10/200 0.2/1t 1/3t 3/6t 6/10t >10t
Pr 10 10 18 28 16 12 4 2
De 20 50 30 0 0 0 0 0
Pr/Dè 0.50 0.20 0.60 » *> « » m .
E 3.00 0 3.33 5.00 5.00 5.00 5.00 5.00
R 1.00 1.00 1.80 2.80 1.60 1.20 0.40 0.20
V 16 16 29 45 26 20 6 3 161
Cm 30 30 120 720 420 310 100 50 1780
Table 4 Quarry production and cost (per running m) for design-based bank protection
The figures in this example show that the determining grading is 0.3/10 kg
and that considerable excess volumes are blasted of all gradings above 200
kg.
7. Supply-based design
A basically different approach is to start with the quarry and to taylor
the design to the size distribution of the supply. A striking example of
this is the development of the concept of (dynamically stable) berm
breakwaters, allowing a considerable reduction of armour size (see Figure
7 ) . In this Figure demand and supply (of two alternative quarries) are
shown as production densities, f(X). It is obvious that the production
curves match the grading (1) better for the berm breakwater. However in
this paper it is shown that also for conventional breakwaters this approach
will often pay off.
dossical rubble mound berm breakwater

1 * eutcr ormaur
' '2wtaa.
3»»Mondary omtour
iso
X/m \ 4 * cara
5-ntar
B«bOMfnttr
« \ —• Cortnto guarry
\ \ —RI>lMUr»taot
1.0 10
•ton* dlamftttr (ml
Figure 7 Rock demand and supply for a conventional vs. berm

breakwater (Vrijling et.al. 1990)
breakwater
In order to facilitate a supply-based approach within an iterative design
procedure for the breakwater cross section, a model has been developed in
the Netherlands as a spinn off from the CUR/CIRIA Manual. This model
includes the design of the toe and has been used to make the following
evaluations of material costs. The model is briefly discussed in section 9.

Using the systematic design procedure of the model the breakwater design is
further optimised to match the given quaixy yield curve as good as
possible. For this paper, optimisation of geómetry have been sought in:
- front slope (cotg ctF)

- crest height (hc)
By considering the calculated armour size (D„) and transmission coëfficiënt
(€,) is has been checked wheter the basic funcfcional requirements stated in
section 3 are still be met. Starting with the preliminary design from
section 6 the cross section has been varied as shown in table 5.
cross-sectional parameter mimmun reference maximum
front slope, cotg aF [-] i:3 1:2 1:1.5
crest height, hc W 14.0 15.0 15.0

Table 5 variation of cross-sectionat parameters
Evaluation of the quarry yield curve and the required volumes (V)
calculated with the model shows that still cönsiderable savings can be
achieved, The cost associated with a numbex of alternatives are indicated
in Figure 8".
70i
ie C M H <fr^t) H, W |
9 2 2 ,5
60- 1 Z 2 IS
Z 1.5 1.5 15
50-
4 3 1.5 15
5 1.5 1.5 1*.5

3T An
•o 40- 6 Z 1:S K.5
8 § 7 2.5 1.5 14.5
£ 30"
9 2.5 1.5 »
§& &i 5SS SSJE5
- ^1 1 1 1 1
o- in 11MHIl1^1
0 1 2 3 4 5 6
alternatives
7
il8 9
Figure 8 Cost comparison of alternative designs
It is obvious that application of a relatively mild front slope of cotg(aF)

= 3, while maintaining the preliminary lee-side slope with cotg(aL) = 1.5,
enables an optimum use of the quarry. The total volume of blasted rock is
only 440 m3 with associated cost of 4700. Relative to the minimum design
(Table 1) the volume and cost ratios for this design are only Ry=l.l and
Rc=1.8. All of the other alternatives except one show cost ratios between
R v = 8.4 and 9.0. One exception is the one with a steep front slope of
cotg(aF) = 1.5. This design requires an armour size of Da = 1.76 m, which is
not produced in sufficiënt relative quantities (demand=23%, production=2%).
To prpduce the armour stone, blasted volumes in excess of the needed result
in a cost ratio of Rp = 25. For comparison, the production data of the
optimum design are listed in Table 6 and placed next (right) to those of
the preliminary design (left) in Figure 5.

•• "'
totats
0.3/10 10/200 0.2/1t 1/3t 3/6t 6/1 Ot >10t
'. E 0 1.11 2.38 1:98 0.46 2.38 2.38
R 1.00 0.90 1.40 0.80 0.60 0.20 0.10
V 90 80 120 70 50 20 10 440
Gm 200 300 1900 1100 800 800 300 4700

Table 6 Production data of optimum breakwater design obtained with the model
river bank protection

Given the function of flood protection, the crest height is deterrained by
the design f lood level (zf = 7.8 m) and the possibilities for variations in
the cross section are confined to the revètment slope angle (cotga).
Using eq.(3) for the lower and and Van der Meer's formulae for the upper
part of thé protection, stability calculations have been made for slopes of
1:4 and 1:2 (which is assumed ppssible regarding geotechnical stability).
The calculated stone sizes are D, = 0.20 ahd 0.30 ra fespectively, but this
hardly affeets thé demand per girading (appr. 20/50/30% for the threë finest
gradings) . Regarding the requij-ed volumes, the indirect effect of an
increase in layer thickness (1:2 comparing to a 1:4 slope) largely
compensates for the direct effect of volume reduction, see Tablé 7.
The figurës show that savings achieved by the 1:2 slope are only 10%.
In general the values for Rv and R,, emphasize that, to comply with th
production of quarry I, these bank protection designs are highly
inefficiënt. Further, by variation of the slope hardly any savings cari be
achieved for this type of rock structure, as long as the required volumes
of the determining grading (here 10/200 kg) are not efféeted. However, if
instead a quarry with yield curve II is available, in the first place the
efficiency in general is improved considerably while, besides, possible
savings achieved by slope optimisation become interesting (Table 7 ) .
quarry I quarry II
1:4 1:2 1:4 1=2
V 170 150 95 84
cro 1860 1680 540 460
Rv 5.0 4.4 2.8 2.5
Rc 21.1 19.1 6.1 5.2

Table 7 Effect of bank slope f o r t w o quarries
8. Comparison of demand vs. supply-based design

In contrast with the conventional demand-based design a supply-based design
considerably reduces the excess cost of rock production in the quarry.
This holds in particular for rock structures with many degrees of freedom
with regard to the design of the cross section, such as for breakwaters-
Applying the model to the example case of a breakwater the cost for
materials could be roughly reduced down to 20% of those of thé demand-type
design. A practical criterioh for comparison is the relative excess
material loss (E), which is a measure for the amount of stone produced but
not actually used in the structure. This loss (E) can be quickly reduced by
adjusting the design by using the present model.
For structures like the riveir bank protection the possibilities to vary the
cross section are limited and practically confined to the slope gradiënt
(stability) and crest width (overtopping). However, it has been
demonstrated that for a type II quarry a supply-based design can be useful
as well for structures as river bank protection for a final optimisation of

'the, slopë.
9. Model description
A brief description is given here on the model used to optimise wave
exposed breakwater cross sections. The experierice that an economie design
of breakwaters and other rock stxnctures is partly based upon efficiënt use
of quarry output>• was the reason to initiate the development of a
comEuterized system fpî.optiîsatiörLÓf^ the design ofjhyd^üiic^staMctures,
ïike breakwaters, seawalls^ dikes and bottom protection. In this context
and as a spin off from the product ion of both CUR/CIRIA Manual On rock, the
Dutch Department of Public Works and Transport (Rijkswaterstaat) has
developed a model for a supply-based optimisation model for breakwaters.
The model structure diagram is shown in Figure 9, which is largély self
explaning.
describing the stone supply
Beihg a major materials design conditioh, a basic input for the model is
'thé quarry production, schematized by a yield curve. Depending on whether
an existing pröducing quarry is coneërned (with a known yield curve) or
rather that a dedicated quarry may be opéned (with little information on
the expebted production) two options are presented to the user:
- providé the available quarry supply volumes;
- give the parameters for the théprétical prediction curves (eg. Rosin-
Rammlèr, Schumann, section 2).
quarry-based design, W. teeuweStein et.al. 12

intro
Identification and maximum settings window 1
r
Hydraulic Criteria Construction Criteria
window
2
T T
Repeat
Transmittèd Waveheight
Toplayer of slope
Hydraulic Crït. Construct. Crit-

Norm. weather Cond. P,S,Slope
Transmittèd Waveh. window 3
Secondary Layer Repeat

Core material
Design Both sides identical ?
_L
window Dn50AL=Dn50A (y) (n) Dn50AL for Ht window
4a Dn50AL Norm.U.Cond. 4b
DnSOAi. design wave lee
Repeat Repeat
choose toe type window 5 choose toe type
Standard toe Dredged Toe Séa-side Sea-side

Both side both sides Standard toe dredged toe
Lee-side Lee-side
Lighter design lighter design
Repeat T Repeat
! <
window 6a or
Design tóe structure window 6b
Repeat
Plot Breakwater Cross-section plot 1
Set Standard gradings window 7

Repeat
Calculation of required volumes window 8

<——
window 9 window 10
1
Rosin-Rammler Distribution Curve Manual input availability
T Repeat
Plot Bars available and required plot 2
Printing of results
Repeat
END
Figure 9 Model structure diagram

In both cases a series of productIon volumes can be provided, each
concerning a grading or class, defihed by lower and upper limiting weights.
These gradings Can: be either Standard gradings (usually cheaper) or
specially defined gradings (usually more cöstly). In the latter case the
model will generate the supply volumes for each defined grading. By
providing a the lehgth of the structure, the model evaluates the total
volumes involved with the entire structure (instead of per running meter).
At the end of each iterative cycle, plots can be prpduced shpwing the
breakwater cross section with principaï results regarding required stone
weights and a cpmparison (per grading or class) of quarry supply and demand
of the design made so far (FigÜres 10 and 11).
iCflQOB Uersion v e r s i o n 3,Q CÓ7col ( 3>
a*r
-fttsKtanueo. i jftflRjj,
u m i u m u n u u; m m u m n n u m u m > u m u i p u n i u i i 11 m m n i i m i 111 j 1111 n t e 11 u i

Om. 5a. 10». 15». 20». 23». 30». 33». «fc. • « • . 90a. 99a. «0». 6Sm. TOm. Tim. 80». 85». 90». 9Sk
ntntstru, al Transpart, Public Uorks and UaMr ttanmqamtm

Dtr«cter«t» Otntral For Public Udrkz and UMar Hanaomtm
Civil Ehojnnring OiTision
^jdraulic Efrgi'n—ring Branch ^ -
Figure 10 Example plot of breakwater cross section
determining the rock demand

Rock stability against waves is determined and used as a design criteria.
This can be done for two optional design principles (Figure 12)s
- both sides subject to the same design conditions (so basically a
symmetrie design);
- different wave exposure at both sides (lëading to asymmetrie design).
Additional options for the user concerning the basic initial cross-section
are the crest width and the toe structure. with respect to the latter the
user has two options, a Standard toe or a "dredged" toe (Figure 12)..
Principal hydraulic boundary conditions to be provided by the user are the
design wave height and period and design high and low water levels (see
section 5). Principal structural design input parameters (for definitions
see sections 2 and 4 and Appendix) are porosity (P), slope angle (tga),
damage (Sd), stone size ratio or filter rule (Dj/Dj) of successive layers,
relative crest width (B/DJ and transition levels. In each iterative design
cycle the user may dëcide to adjüst each of these parameters.

CADOB Uersion version 3.0 C67col ( 3>
. Y-todss Percemaqe RequirsdxAvaUabicNGN-vxcMdence
ioaoo..| 6rjdin Q P»**,? —i- I 5 1 <l 7i
90.005: • Cl 2s 1Q>200 ka
a 3: 0.2^1 t
O t i / S t
8Q.0<Kf• Cl S: 3s& t
Cl «I 6^10 t
70.005; " a 7i ia>2o t
5E3 1E4
Ueight tkg] ->
mnistry of Transport, Public Uorks and Uater' nanageaem
«Yailable r:-:*:*:::-::---; Plrectorat» General For Public Uorks and Uatar nanageaent
Ciyil Engineering Qivision
Required &&&£& Hudraulic Engineering Braneh
Figure 11 Example plot of stone supply vs. demand
construction stage
Besides, the model also includes an option to designing against damage
during the construction stage.
Due to storm during construction serious damage may occur to the finer
(non-armour) layers, leading to unacceptable delays and cost overrruns.
Since an exposed filter in this stage is particularly susceptible to wave
action, the filter may be damaged. This can be avoided by proper
dimensioning of the second layer (filter or secondary armour), using a
second set of design conditions for construction. For construction stages
reaching into known periods of increased storm frequency, these design
conditions will be stronger (eg. than for construction in a quit "summer"
season (see Figure 1 ) .
Both, for the design of completed and construction stage the designer can
adjust the accepted risk of damage through the damage parameter Sd
(Appendix) by calculation of the cost associated with replacement of the
lost stone volumes.
Pn50AL-Pn50A
DnSOSL-DnSOS
Boduxidcs idartical
DnSOA
Boch. «ide* different
Figure 12 Options of symmetrical and asymmetrical cross section

Smtad toe dolen
Figure 13 Options of tóe

10. Summaxy and conciusions
It has been demonstratéd that a supply-based design approach of rock
structures wil1 easiïy leads to significant savings of material cast. This
holds in particular for breakwater, since this type of structure allows
rélatively iaany cröss-sectiöhal variatibns. Total investment costs of
structures as far as they are related io material use may thus be reduced
cpnsiderably.
The present state of the art of design méthods for hydraulic structures
built with rock allows for the development of routine design procedures,
able to prevent already in the early stage of a project a too costly
design. In this respect, the presented model developed in the Netherlands
is a usefül tooi to arrive at a more economie use of rock in hydraulic
engineering.
Allpwing for further optimisation by considering replacement of determining
gradings with substitutes (eg. earthfill, geotextiles or concrete units)
the present model can be a practical tooi to achieve even further
reductions on material cost.

refGrenc&s
CUR/CIRIA, "Manual o n t h e u s e o f Rock in Coastal and Shoreline
Engineering", C U R Report 1 5 4 , CIRIA Special Publicatibn 8 3 , Gouda, T h e
N e t h e r l a n d s , 1991.
Gessier, J., ''The beginning of Bedload Movement of M i x t u r e Investigated a s
Natural Armouring in Channels", Truninger P u b l . , Zürich, 1965.
V r i j l i n g , J.K., Nooy vd. Kolff, A . H . , "Quarry Yield and Breakwater Demand",
IAEG, B a l k e m a P u b l . , Rotterdam, 1990.
notation
Ag Wave-induced erosion volume of rock slope
B Berm width m]
C Chezy friction coëfficiënt m'Vs]
Cm Material production cost
C, Wave transmission coëfficiënt - ] ' ' • ' • ' :
D Stone diameter
Dn Nominal stone diameter m]
D-, Stone diameter not exceeded by i% by weight m]
De Rock demand m3]
E Relative excess production -]
F Distribution function (loading parameter, stohe size)
f Density function (loading parameter, stone size)
fp Production multiplication factor -]
h Water depth m]
hc Crest level m]
N Number of waves in design storm -1
P Porosity factor for wave-exposed rock slopes -]
Pr Rock production m3]
Pd Determining production volume
Q River discharge
m3]
R Production ratio
m3/s]
~RC Crest freeboard (R^ = h c - h ) -]
R,, V o l u m e efficiency factor -]
Rv Cost efficiency factor -]
Sd Damage parameter for wave-exposed rock slope -]
Su Damage parameter f o r current-exposed rock -]
T W a v e period s]
t L a y e r thickness m]
U Current velocity m/s]
V Rock volume m3]
W W e i g h t o f stone kg]
z W a t e r level relative t o datum m]
a Slope angle of -]
A Relative density o f stone ^1
<j> F r i c t i o n angle (here chosen a s 3 5 " ) . -]
^f S h i e l d s ' b e d shear parameter -]
\ Surf similarity parameter -] 3
pr Rock density kg/m ]
guarry-based design, W . Leeuwestein et.al. 17

Appendix
rpcJt stability urider wave action
Based on earXier work of Thompson and Shuttler (1975) an extensive series
of model tests was conducted by Van der Meer (1988) on structures with
covering wide ranges perraeabilities and wave conditions. Two formulae were
derived for plunging and surging waves respectively, describing the
stability in terms of a ratio of wave height (H) over stone size (D) or
H/AD. Note that for H and D the significant wave height (HJ and 50%
nominal diameter (D,,») should be substituted:
H. /AD,,» = 6.2 P018 (S..//N)0-2 lm ; for plunging waves (Al)
H, /AU,,» = 1.0 F013 (SJ/VN) 0 2 7(cota) £*; for surging waves (A2)
The transition from plunging to surging waves can be calcülated using a

critical value of £ m :
^ = [6.2 P0-31 7(tana)]"(P+0-5) (A3)
In these formulae, er is the slope angle, A is the relative submerged rock

density, $ m is the surf-similarity parameter (with respect t o the me an wave
period) , N is the momber of waves in the design storm, Sd is the damage
parameter, defined as the dimensipnless erosion volume A,/D2 and P is the
notional permeability factor. Further details can be found in the CUR/CIRIA
Manual (1991). Here it is only npted that Sd=2 to 3 and Sd=8 to 17
correspond to "ho damage" and "failure" respectively (actual values
depending on sippe angXe a) and that P varies from P=0.1 (stones on
impermeable slope) to 0.6 (homogéneous rubble mound).
wave transmission
Based upon evaluation of a large data base performéd for the Manual on Rock
(CUR/CIRIA, 1991) wave transmission formulae has been found to answer
approximatly the formula:
C, = 0.46 - 0.3 R./H, (A4)
where Rc is the crest freeboard, defined as Rc = hc - h.

The formulae is valid for a RyH, range of -1.13 to 1.2, while for higher
crests (R/H, = 1.2 to 2.0) and lower crests (Rc/Hs = -1.13 to -2.0) the
limiting values are C, = 0.1 and C, = 0.8 respectively.
slope factor for rock stability under current attack

For rock or stones on a river bank the following slope factor should be
included in tha hydraulic stability analysis.
tancr
k,, = cosct i/ ( l - ( )2) (A5)
tan<j)
with (|> being the friction angle of the rock (here assumed 35°).
material cost rates

With regard to overall material cost rates for armour, filter and core
material the following figures are used:
weight class cost rate

Cm"3]
minimum maximum
10 kg 2
10 kg 200 kg 4
200 kg 16

PART 2
Closures and storm surge barriers (see also Part 7, Scour)
Janssen, J.P.F.M, and R.E. Jorissen,

Modelling storm surges and forecast effects in designing the
Rotterdam storm surge barrier
Helsloot, I.C.M. and J.P.F.M. Janssen,

Feasibility study of a two-barrier system in the Dutch sea
water defence
Janssen, J.P.F.M., A. van Ieperen, B.J. Kouwenhoven, J.M.

Nederend, A.F. Pruijssers and H.A.J. de Ridder,
The design and construction of the new waterway storm surge
barrier in the Netherlands; technical and contractual
implications
Jorissen, R.E., D.P. de Wilde and E. Berendsen,

Design of the bed protection of the Rotterdam storm surge
barrier
Pilarczyk, K.W., H.J. Verheij and G.J. Akkerman,

Rockfill design criteria for overflow dams
MODELLING STORM SÜRGES AND FORECAST
EFFECTS IN DESIGNING THE ROTTERDAM
STORM SURGE BARRIER
J.P.F.M. Janssen and R.E. Jorissen,
Ministry of Transport and Public Works,
P.O. Box 5044, 2600 GA Delft, The Netherlands
Abstract
In 1989 the Dutch government decided to build a storm surge barrier in

the New Waterway near Rotterdam being a good and cheap altemative for
the necessary strengthening of dikes along the lower regions of the
Dutch rivers Rhine and Meuse. To be feasible the barrier had to meet
several goals. The most important ones being :
* closing frequency less then once every 10 years now and once every
5 years after 50 years from now (including 25 cm sea-level rise)
* prescribed reduction of design water-levels at two representative
locations, being Rotterdam and Dordrecht.
These and other parameters are calculated by means of a probabilistic
calculation method. This method involves a mathematical open-channel
network model of which the results are combined with the statistical
properties of input parameters. A risk-analysis of the performance of
the barrier is included. Finally the model is adapted to study
operational aspects.
1. Introduction
The project area is situated in the southwest of the Netherlands as

shown in figures 1 and 2. In figure 2 the dikes are shown, which were
to be reinforced as a consequence of the Delta plan. This plan was
229
developed after the flood of 1953. It
included shortening of the coast line by
closing a number of tidal inlets and to
strengthën remaining dikes. By the time
of 1985 most of this plan has been com-
pleted : still about 200 km of dikes had
to be strengthened. A re-examination of
design water-levels (DWL) led to higher
Fig. 1
values than those originally calculated.
This implied very costly construction
works in densely populated or otherwise
complicated areas. Therefore, in 1987 the
Dutch government initiated a study to
consider a storm surge barrier near
Rotterdam. This barrier should meet two
main goals. Firstly the DWL should be
Fig. 2 Project atea
reduced significantly to avoid problems
related to dike reconstruction works. The prescribed reductions of DWL
are based on the situation in the areas, which would be most affected
by the reconstruction works. These areas are the cities of Rotterdam
and Dordrecht. Secondly the presence of Rotterdam harbour doesn't allow
the barrier to be closed too often. The closing frequency of the
barrier should (on average) be less than once every 10 years now and
less than once every 5 years after 50 years from now. The latter
situation includes a sea-level rise of 25
centimetres. From several predesigns a
final choice was made at the end of 1989.
&
The selected barrier design features two
sea v_ river semi-circular doors. The barrier is
V
closed by rotating the floating doors
into the river (figure 3) and lowering
the entire construction to the river
Fig. 3 Barrier principle bottom by filling ballast tanks with
water. This paper deals with hydraulic
aspects of this barrier and the influence of reliability of the barrier
on DWL and operational aspects.
230
2. Hydraulic aspects
2.1 The hydraulic system
projected river In figure 4 the hydraulic system is shown

barrier discharge
in a schematic form. Basically there are
two river branches with several
connections. The southern branch runs into
the large Haringvliet estuary; This
= = river dike estuary is separated from the sea by a
seadike
barrage with large discharge sluices. The
\\\\\ polder
Fig. 4 The system northern branch runs freely into the sea
via the city of Rotterdam and the New Waterway. Tidal movement and
therefore storm surges enter the system through this northern branch.
Close to the river-mouth the water-
level is determined completely by
tidal movement combined with storm
surge effects. Figure 5 shows that
the water-level at the river-mouth
can be constructed as a combination
of astronomical tide and a storm
surge. The maximum sea water-level
(MSWL) is influenced by five para-
meters: astronomical tide, duration,
Tim ItUrsl
shape and height of the storm surge
and phase difference between storm
surge and astronomical tide. This
number of variables can be reduced.
The variation of the astronomical
'« '« 7WU7 « nou» « '«* «•
tide at Hoek van Holland is so Tim fiers)
small, that a mean tidal curve will Figure 5

be used. Secondly the shape of the storm surge is considered to be con-
stant as shown in figure 5.
The incoming tidal wave is damped travelling upstream and phase dif-
ferences occur, as shown in figure 6. In this intermediate region the
water-levels are determined by both tidal movement and river discharge.
231
f »».»
•l.OO
1.04
5 «4.50
5 -o.»
: -o.7i
P-i1.00 a a u a
TOOIOO naai « TOOIOO TOOIOO TOOIOO « 7(0107 " 70010» 70010» 7stue » 700107 <2 TOOIOO 700109
Location Hoek van Holland Location BnttBPdan Location Dorareent
.-•-••• — V 1 3 j
l L
•/ ""
J\J. i I.L
-0.07
700100 " 7W107 « 700100 « 70010» 700100 « 700107 « 700100 •* 700100 7O01OO " 700107 « 70010» » 700100
r/ow litofx/ o» ttow Atom* o» n— (*»rsl —o»

Figure 6
The relation between MSWL, river discharge and maximum river water-
level (MRWL) is shown in figure 7. This figure shows lines of equal
MRWL at Rotterdam and Dordrecht as a function of river discharge and
MSWL. In these graphs the values of duration and phase difference of
the storm surge are kept constant at 29 and 4.5 hours.
Location Rotterdam Location Dordrecht

6 I I I I | ' i ' | I I I | I r I | I I
4 8 12 16 k 8 12
'River Discharge (103 m3/s) River Discharge HO3 m3/sl
Figure 7
The hydraulic conditions in the project area are governed by four
variables. Three variables determine the boundary condition at the
river-mouth (MSWL). These variables are duration (s), height (m) and
phase difference (p) of the storm surge. The fourth variable is the
river discharge ( q ) , which in combination with the MSWL, détermines the
MRWL. The river discharge varies on a much larger time scale than storm
2331
surges do. Therefore this variable can be treated as a constant during
a storm surge. The four governing variables are considered to be
independent. This method of determining water-levels is an extension of
the usual approach, which is based on only two governing variables:
storm surge height and river discharge. The parameters storm surge
duration and phase difference are assumed to be constant (29 and 4.5
hours). For the situation without a barrier the results of both
calculation methods don't differ very much.
2.2 The effect of the barrier
The effects of the barrier on water-levels in the hydraulic system are

twofold. One effect is the reduction of water-levels because the storm
surges can't enter the system any more. On the other hand there is an
increase of water-levels in the system because of the accumulation of
river discharge. To calculate the total balance of effects the hydrau-
lic system has been modelled by a mathematical open-channel network
model. The schematization of the system is conform figure 2 and con-
sists of about 200 branches and nodes. This model has been in use for a
long time to predict water-levels on a daily basis. The barrier should
reduce the DWL at Rotterdam and Dordrecht. The prescribed DWL are 3.60
meter above NAP for Rotterdam and 3.00 meter above NAP for Dordrecht.
In order to reach this goal the barrier has to be closed whenever the
predicted maximum sea water-level (PMSL), given the actual river dis-
charge, would result in a exceedance of the DWL at one of the loca-
tions. In order to have some freeboard regarding the effects of in-
accuracy of the PMSL (see paragraph 3) a critical water-level (CL) is
introduced. This level is somewhat lower than the DWL. The margin
between DWL and CL is 40 cm for the location of Rotterdam and 10 cm for
the location of Dordrecht. The reason for these different margins is
the fact, that Dordrecht is about twice as far from the North Sea as
Rotterdam and therefore less sensitive to sea-level variation. This
means, that the margin, which is strongly influenced by the accuracy of
the PMSL, can be somewhat smaller at Dordrecht. It should, be kept in
mind, that both DWL and CL are fixed water-levels at the locations Rot-
terdam and Dordrecht. Figure 6 can be used to transform the CL into a
233
river discharge dependent criterion CL(q)
for the location Hoek van Holland for
which also the PMSL is given.
In figure 8 the effect of the barrier is
780108 8 12 18 7MM7 8 12 18 7WM '
shown for the locations of Rotterdam and Tlae Uaysl •»
Dordrecht for a specific storm surge and

Location Rotterdam
river discharge. In figure 9 the effect »2..47..
of the barrier on the relation between

MSWL, river discharge and MRWL is shown.
In figure 9 the thick line represents the
•2 ' • 750107 « « U 780108 » « * 780105
closing criterion CL(q). Below this line Time Uaysl •»
figure 9 is exactly the same as figure 7.

Above this line the barrier is closed and Location Oordrecnt
i-M.
this reduces the MRWL significantly
compared with figure 7. Only for very
high river discharges the effect of the
5 -0-50
780108 8 12 18 780107 8 12 18 78010* 8 12 U 78010»
barrier is less significant.
Ti* (Uaysl •»
Figure 8
Location Rotterdam Location Oordrocnt
0.00 *0.50 +1.00 tl.30 0.00 +0.50 +1.00 +1.50

RiYtr discharge HO3 op/al •> Rirer Oiacltarga fis3 a?/sl •»
Figure 9
2.3 Calculation of statistical aspects
So far only hydraulics have been discussed. The link with statistics
however has to be made because of the probabilistic design procedure of
the barrier and the definition of DWL. A DWL is a water-level which has
a prescribed frequency of exceedance. The DWL is the most important
factor in designing a dike. The prescribed frequency varies along the
234
country depending on the type of threat (sea/river/lake), population
density and economical activities. For Rotterdam this frequency is
1/10000 per year whilst for Dordrecht this frequency is 1/4000 per year
for the situation without a barrier and 1/2000 per year for the
situation with a barrier. This increased design frequency is based on
the reduction of the threat by storm surges.
By means of the hydraulic network-model, mentioned in the previous
paragraph, the probability distribution functions of the boundary con-
ditions are transformed into probability distribution functions of the
water-levels in the hydraulic system. From these functions the DWL can
be calculated easily. This method can also be applied to calculate
design discharges, velocities and hydraulic head of the barrier. The
transformation of probability distribution functions can be determined
by a numerical solution of the following equation :
P{MRWL->X)=\\\\ f{m,s,p,q)dmdsdpdg
J J J JMRWL=X
In this equation f(m,s,p,q) is the combined probability distribution

function of the stochastic variables m, s, p and q. The probability
distributions of river discharge and storm surge duration are directly
derived from recordings during the last century. The probability of
storm surge height and the phase difference are determined in such a
way, that the calculated probability distribution of the MSWL shows a
good agreement with the historical data and its extrapolation.
Reliability aspects
The beneficial effect of

PSHL yes barrier yes barrier yes HRHL yes
—».
above CL? closed? — » • sustains? <DHL? -O the barrier on MKWL is
no ) no C nol Pb strongly influenced by
1
functional and structural
HRHL yes HRHL yes
<DHL? " -O <DHL? -O
1 HRHL
1 <DHL? ^b reliability of the bar-
nol nol nol no rier. To investigate
\
+ IL>DHL|
P
2 P
3
+
4 - F MMRI
P
these influences a risk-
F ig. 10 Bas ie evei nt tr ee analysis has been carried
235
out. In figure 10 an event tree is shown with all possible branches
leading to a MRWL exceeding the DWL. At the top of the figure the
'normal' sequence of events is shown. That is a good prediction, a
correct decision, a properly functioning barrier, which is strong
enough to withstand the forces, mostly generated by the hydraulic head.
This leaves four points where something can go wrong:
* a PMSL lower than the CL(q), so the barrier is not closed, but an
actual MRWL higher than the DWL at either Rotterdam or Dordrecht.
* a PMSL higher than the CL(q), but the barrier is not closed and as
a consequence MRWL can be higher than DWL.
* a PMSL higher than the CL(q), the barrier is closed, but collapses
due to the enormous (hydraulic) load. As a consequence the MRWL can
be higher than DWL.
* even if everything functions well, there is still a possibility
that .the MRWL is higher than DWL.
4. Design conditions/calculations
4.1 Calculation method
For a numerical integration the probability distribution functions of

the boundary conditions have to be sampled. Combining the four sampled
probability distribution functions, yields a total number of hydraulic
calculations to be made. The total number of calculations is set at
6900, which is sufficiënt for accurate results. The results of these
calculations are stored in a database. This database contains the
following information : the boundary condition and its probability
density, the MSWL, the MRWL at each node of the network model, the
hydraulic head and maximum discharges in specific branches of the
network model. Two of these databases are made. One database contains
information for the situation without a barrier. The other database
contains information for the situation with a barrier, which is being
closed for every storm surge. By reading both databases and determining
if the barrier should have been closed [by comparing the MSWL and the
CL(q)] a choice for the 'open' or 'closed' data can be made. After
reading both databases the probability distribution functions of all
236
data can be constructed. Because of this separation between calculation
and selection, several values for CL(q) and other reliability para-
meters (see next paragraph) can be used without having to make an
extensive calculation. This is very convenient for a sensitivity
analysis.
„4.2 Integration of reliability aspects
The separation between calculation and selection also offers the possi-
bility of integration of the reliability aspects. By integrating the
event tree of figure 10 into the selection program, a probability for
each situation ('open' or 'closed') can be calculated. This probability
is determined by the CL(q), the accuracy of the PMSL, the unreliability
of the barrier (C) and the probability of collapse of the barrier (Pb).
If the barrier collapses, the data for the 'open' situation is assumed
to be valid. The unreliability C is the probability of not closing the
barrier given the fact that the PMSL exceeds the CL(q).
It is clear that the parameters CL(q), the accuracy of the PMSL, the
unreliability C and the probability of collapse Pb play a central role
in determining the effect of the barrier in terms of DWL and closing
frequency. The presented calculation method also provides the pos-
sibility of setting targets considering closing criterion, accuracy of
PMSL, unreliability and structural strength.
4.3 Sensitivity analysis
Already in an early stage of the project the permissible probability of

collapse was fixed at 10"6 in any one year. Considering the frequency of
DWL for Rotterdam (10~4 per year) the effect of Pb can be neglected. The
closing criterion CL(q) is set to such a level that the closing
frequency is tolerable. This leaves the accuracy and the unreliability
as variables. A sensitivity analysis has been carried out considering
these two parameters. Initially, the accuracy was set to the properties
determined by an analysis of predicted MSWL during the last decades.
This analysis showed that the difference between predicted and measured
237
MSWL could be assumed to have a normal
probability distribution with an average
(|x) of -20 cm and a Standard deviation
(er) of 25 cm. This means that the pre-
dictions were on average on the safe
side. In figure 11 the sensitivity of the
DWL at Rotterdam for the o- is shown. This
line shows a very large sensitivity to a. T T"
5 10 ~r- 15
By using von Kalman filtering, based on Sigma PHSL 'cm]
on-line measurements during the passage

Figure 11
of t h e
*4 25
storm at the British east coast and
f platforms in the North Sea, one beiieves
that er can be reduced tq 15 cm. This
value is used for design calculations of
+3.50J •&
the storm surge barrier. The sensitivity
of the DWL at Rotterdam for the unrelia-
bility is shown in figure 12. This figure
Unreliability . shows a sharp increase of sensitivity for
Figure 12 values of C larger than 10"3. Therefore

the maximum value of C was set to 10"3.
4.4 Results
In the figure 13 the results of a com-

plete calculation for the situation
without a barrier are shown. In this
figure the probability of exceeding
certain water-levels at the locations
Hoek van Holland, Rotterdam and Dordrecht
is shown. The DWL for these locations are
5.15, 4.80, 3.78 meters. In figure 14 the
same lines are shown for the situation
10 10 10 10 10 10 10 10 10 10
with a storm surge barrier. The DWL for Frequency of exceeüance fl/year] ^
Rotterdam and Dordrecht are reduced to

F i g u r e 13
3.52 and 3.15 meters.
238
These results show that the prescribed
"r
+7.00.f.jo HOK «antollana
reduction of DWL for the location
*6.00i-)A3onraiit
Dordrecht is not achieved, according to
*5.00..
the calculation method with four
i variables. However, for the calculation
n.ooJ.
of DWL the method with two variables has
I
»3.00J
been applied for all dike reconstruction
•2.00J
works up till now. For reasons of
continuity this method will be applied
+1.00 U.i' m
J
10** 10" 10° 10"' 10"* 10" 10"* 10"* 10"* 10"'
Fremency of exceeaance tl/yearl w for the final stage of the dike
Figure 14
reconstruction works as well. Another
calculation with only the variables storm surge height and river dis-
charge reduces the DWL for the location Dordrecht to 2.90 meter above
NAP. For the location Rotterdam the DWL is reduced to 3.46 meter above
NAP. For the situation with a barrier the difference between the
results of both calculation methods are increased. Especially for
locations, which are relatively far away from the sea. These locations
suffer the most from the negative effect of the barrier.
5. Operational aspects
5.1 Specific problems related to operational use of the barrier
During operation decisions have to be made about the closing and

opening of the barrier. For both decisions criteria have been set. To
close the barrier the PMSL must be higher than the CL(q) and a closing
condition (CC) must be actually reached. In the current design the CC
depends on the river discharge. For low discharges the CC is a water-
level of 2.00 meter above NAP and for high discharges the CC is the
local zero inflow condition. It is obvious that these conditions can be
met several times during a storm. To prevent high water-levels by
unnecessary long closures of the barrier (accumulation of discharge)
the closure starts at the last CC before the sea water-level exceeds
CL(q).
To open the barrier a opening condition (OC) must be reached and the
PMSL may not exceed the DWL any more. The OC is defined to be the
239
moment when the water-levels on both sides of the barrier are equal.
Non-exceedance of the DWL is guaranteed if the PMSL does not exceed the
CL(q) any more.
The involvement of predicted water-levels causes some problems. If
predictions were precise it would be very easy to piek the right moment
to close and open the barrier. In practice however predictions are in-
accurate (increasing with time) and also only available for a limited
time ahead (at the most 18 hours). This implies that every occurrence
of CC or OC in a specific storm is a potential closing or opening
moment. Which moment is chosen depends on the outcome of the predicted
water-levels. The use of predicted water-levels may even cause the
barrier not to be closed at all at the last CC; for example if the
predicted (last) CC does not occur in reality. Since this is thought to
be unacceptable an extra possibility for closing the barrier is
introduced at
' Hoek van Holland
U6END: """""
behind barrier the lowest
water-level
J \ oc . after the last
CLIql
^
CC. Figure 15
\ "
illustrates four
-
•
possibilities
-2.00 i 1 1—
-20 -W 0
time (hours)
for a single
storm: so-
5,0» "
4.00-
* called
3.00" J \ .oc realizations. In
CUq) 1^ \ Z J-— » *
this case there
i.oo •
V \ "
0.00 are two CC that
1.00-
7 00- ' i ""- — T -n

can be combined
-20 -10 0 -20 -10 0
time (hours} — time (hours) -

with two OC.
Figure 15 Combination c)
illustrates the
optimum choice. The model described in the previous paragraphs only
includes the optimum closure and opening of the barrier [combination c)
in figure 15]. Since other closure or opening moments may.generate dif-
ferent MRWL and loads on the barrier a model is developed to include
the effect of all possible realizations. The calculation scheme is
described in the next paragraph.
240
5.2 Adaption of the calculation method
The calculation method to determine DWL and loads

is roughly the same as described in paragraph
4.1. The hydraulic model is used to calculate the
water-levels. For every storm all realizations
are included. The procedure to generate all
realizations for a storm is shown in figure 16.
For every realization the resulting MRWL and
loads are determined and stored in a database.
The statistical calculation to derive DWL and
design loads is based again on numerical inte-
gration of all possible combinations. However the
important difference with the method described in
paragraph 4.1 is that for a single storm more
realizations are possible and this requires an
extra integration over realizations. The
Figure 16
conditional probability of the realization (for a
given storm) is calculated. At the current CC the probability of sa-
tisfying the criteria for closing is calculated and the same procedure
is foliowed for the OC with respect to the criteria for opening. In
principle more than one predicted high water is considered in
calculating this probability. An analysis of predictions during the
last decades showed that these predictions were independent. The
conditional probability of a realization will depend on the CL(q)
level, the height of individual high waters and the accuracy of the
predicted water-levels. The probability of a realization is calculated
by multiplying the conditional probability with the probability of the
boundary conditions. Initially a very fast and simplified open channel
network model was used because of calculation time. It appeared that by
taking into account realizations the number of calculations increased
with a factor 3.5. In the future the more complex model described in
section 2.2 will be used. The preliminary results in the next paragraph
are based on calculations with the simplified model.
241
5.3 Preliminary results
The analysis with the model described in section 5.2 was focused on the
influence of the operational decisions on DWL and design loads. When a
CT of 0.25 meter was used for the accuracy of water-level predictions
the following results were found.
* an 302 increase in closing frequency resulting from more potential
closures for every storm
* no influence on DWL at the representative locations compared to the
previous analysis (also including ar = 0.25 meter)
* no influence on the positive design head over the barrier
* an increased negative design head over the barrier (202) resulting
from closures at an earlier moment and/or opening at a later
moment.
Influences at other locations and for other parameters could not be
determined with the simplified model. This will be analyzed with the
complex model.
6. Conclusions
* The use of large numerical models does not necessarily rule out the
use of probabilistic design methods.
* The accuracy of the predictions of water-levels is of vital
importance for the performance of the storm surge barrier.
* The calculation method of DWL should be based on all governing
variables. Reducing the number of variables is only tolerable after
a sensitivity analysis. Based on such an analysis the significant
value of the variable can be used in the calculation of DWL.
* The effect of actually operating the storm surge barrier should be
included in the design process, also in the phase of determining
the boundary conditions.
243.
7. Acknowledgements
It is obvious that the study presented in this paper involves an

enormous number of computations. Most of this work is done by mr. Henk
de Deugd also from Rijkswaterstaat. The authors are duly grateful for
this.
Abbreviations and symbols used in this paper
C - unreliability NAP = reference water-level

CC = closing condition OC = opening condition
CL = critical water-level at p = phase difference
Rotterdam and Dordrecht Pb =» probability of collapse
CL(q) = critical water-level at PMSL = predicted maximum sea
Hoek van Holland. water-level
DWL = design water-level <j " Standard deviation of
MRWL » maximum river water- error in PMSL
level q = river discharge
MSWL =* maximum sea water-level s = duration of the
m = storm surge height storm surge
fi. = average of error in PMSL
24&
Feasibility study of a two-barrier system in the Dutch sea water defence
I.C.M. Helsloot" and J.P.F.M. Janssen2)

Ministry of Transport, Public Works and Water Management
The Netherlands.
Institute for Inland Water Management and Waste Water Treatment,

Ministry of Transport, Public Works and Water Management,
Van Leeuwenhoekweg 20, 3316AV Dordrecht, The Netherlands.
Road and Hydraulic Eng. Division
Ministry of Transport, Public Works and Water Management,
PO box 5044, 2600GA Delft, The Netherlands.
Abstract
In the southwest of the Netherlands a storm surge barrier is under construction in the
waterway that gives access to the port of Rotterdam. A feasibility study into the effect of an
extra harbour entrance to the sea is being performed. The entrance requires a second
barrier. Combined with dikes along the rivers, the barriers have to provide the required
safety against flooding. The study is concentrated on the influence of the second barrier on
design water levels for dikes which were already set by the first barrier. The frequency of
closure for both barriers has to be minimized, because of harbour interest. With this
restriction, no significant increase in design water levels is allowed. A hydraulic network
model is used to calculate the effect of the barriers on water levels. To calculate design
water levels an efficiënt probabilistic calculation procedure is developed. Reliability aspects
related to the barriers are integrated into the calculations.
1 Introduction
The project area covers the lower river reaches in the southwest of the Netherlands, as
shown in figures 1 and 2. Figure 2 represents the Rhine and Meuse Delta network, in
which water levels will be influenced by operating the storm surge barrier to be built in the
New Waterway (S w ). Combined with the dikes along the river branches this barrier will
provide safety against flooding.
The main southern branch (Haringvliet and eastward) was separated from the southera
delta by the Volkerak dam in 1969 and from the sea by the Haringvliet sluices in 1970.
These sluices are closed during high tide to prevent sea water from entering the delta region
during low tide only when
the river discharge is low.
In this way the water is
forced to discharge through
the northern branches, to
stop the salt intrusion.
The conditions for control
of the northern branches
are different. The northern
branches give access to the
port of Rotterdam and
therefore have to render
free transit for navigation
whenever possible. From
this infrastructural point of
view the plan was drawn
up to create an extra har-
bour entrance to the sea. In
1990 a feasibility study was
started into the effects of
the project. One of the
aspects incorporated was
the safety against flooding,
which has to remain gua- _-~ ) »«Bcii»lv__^C*MIW»» \
ranteed. Since the entrance
again will connect the
rivers to the sea it was
obvious that another (storm
surge) barrier, located so-
x^.
' ' ^ ^ ^ ^^-#£VI10B«UK
mewhere in the Hartel-
channel (figure 2, SH), was
needed. The feasibility
study was concentrated on
the influence of this Hartei Figures 1 and 2 The project area
barrier on the design water levels of dikes. This level strongly determines the safety against
flooding and can be affected by barrier closure. However, on economical grounds, origi-
nating from port and navigational interest, the closure of the Hartei barrier has to be
minimized, in concordance with the closure of the barrier in the New Waterway. The
frequency of closure of the Waterway barrier is restricted to once every ten years. For the
Hartei barrier, which is not situated in the main entrance, once every one to three years,
seems to be acceptable.
This paper deals with the approach, chosen to meet the requirements of both safety
against flooding and minimal closure of the system of two barriers. To this end an integral
safety concept has been developed, where both the combined effect of storm surges and
river floods and the performance of the two-barrier system are handled probabilistically and
integrated to assess safety conditions in the whole Rhine and Meuse delta region.
In section 2 the project and the hydraulic system will be described. Next the calculati-
on procedure will be outlined (section 3). In section 4 the results of an indicative sensivity
analysis are discussed. Finally some conclusions are drawn. 4tY
2 Project description and impact on the hydraulic system
2.1 Project description
Water levels in the lower riverreaches in the Rhine and Meuse delta area in the
southwest of the Netherlands are determined by discharges of both rivers and by sea levels.
When the storm surge barrier in the New Waterway will be in full operation the extreme
water levels in this area will be influenced by using it.
The western part of the area to be
influenced is shown schematically
in figure 3. The combined action
of both river discharges and sea
levels is indicated. The figure also
shows the location of the storm
surge barrier in the New Water-
way (Sw) and the main channels
through the port of Rotterdam
where the extra harbour entrance
is projected. It involves the remo-
val of the Beerdam (BD), which
will create a second connection
with the sea. Tidal movement and
storm surges will enter the delta
region through both the New
Waterway and the Hartelchannel.
Water from the rivers Rhine and _. _ _, . .
Meuse will then discharge through ^ë*™ 3 Schematic view of the system
these two branches and at high
discharges also through the Haringvliet sluices that are to be opened (figure 3).
In the project even a third connection to the sea is considered. This involves the
removal of the Rozenburg lock (RL). Studying the effects of this altemative, it is obvious
that the Hartel barrier has to be located in the Hartelchannel east of the Rozenburg Lock.
The strategy of control of the two-barrier system has to be settled, whereby it has to
comply with two demarcations defined in the project. First it is required that the safety
against flooding remains guaranteed. Secondly, the closure of the Waterway barrier is
restricted to once every ten years. These requirements make demands on the performance
of the Hartel barrier within the two-barrier system.
2.2 The hydraulic system
Safety against flooding will be expressed by Design Water Levels (DWLs), a local
water level with a fixed frequency of exceedance. Closing the barriers very frequently
would give a large reduction of the DWLs, but would not comply with the requirements.
From the characteristics of the hydraulic system of the delta area it is possible to describe
the effects of operating the barriers on the DWLs qualitatively, given these requirements.
The effects of operating a barrier on River Water Levels (RWLs) are twofold. By closing
the barrier RWLs are reduced because storm surges are prevented from entering the
system. On the other hand RWLs are increased because of the accumulation of river
discharge. In this respect the delta area upstream of the barriers, where water levels are
influenced by operating them, can be dèvided into two regions as follows:
The 'tidal region' where storm surges which have access to the area due to the
restriction in frequency of closure are dominant.
The 'transition region' where water levels attained by accumulation of river discharge
are dominant.
Taking into account the characteristics of the hydraulic system these regions can be
indicated roughly. The exact extension of the regions is dependent on various aspects, e.g.
the management of the barriers and the river regime.
The tidal region is located in the vicinity of the barriers. Roughly this covers the
western part of the northern branch with the New Meuse and most part of the Old Meuse
(fïgure 2). It is obvious that water levels in the area just upstream of the Waterway banier
will be determined by storm surges that are allowed through by not closing this banier. A
similar reasoning concerning the area upstream of the Hartel banier is legitimate. Generally
spoken the closure of the Waterway barrier has a far more pronounced effect on RWLs
then closing the Hartel barrier, because the cross-section of the New Waterway is larger.
The characteristics of the tidal region can explain the fact mat the Hartel barrier will
have to be closed more frequently than the barrier in the New Waterway. Here DWLs are
determined by the storm surges allowed through by the Waterway barrier. An extra
connection with the sea will result in an increase of RWLs. Closing both barriers as often
as was intended for the Waterway barrier in the present geometrical situation will lead to an
increase in DWLs. Because the frequency of closure of the Waterway barrier is limited,
reduction of the DWLs to the required safety Standard can be accomplished by closing the
Hartel barrier more frequently.
The transition region is located at sufficiënt distance from the barriers. It covers the
river branches on the eastern part of the northern side and the whole southern branch
(fïgure 2). The upstream boundary can be drawn where the barriers have virtually no effect
on DWLs. From the characteristics of the transition region some conclusions can be drawn
concerning effects on DWLs. Since accumulation of river discharge is dominant in this area
RWLs have to be minimized when the barriers are to be closed during periods of high
discharge. Therefore, at high discharges, it is important to:
drain water through the barriers (during closure) whenever possible,
drain water through the Haringvliet sluices whenever possible,
minimi7& inflow into the area during closure of the barriers.
It can also be concluded that:
phenomena such as changes in storm surge or river regime (due to climatic changes),
resulting in an increase in water levels at sea and discharges on the rivers cannot be
fully reduced by a change in the frequency of closure of the barrier(s),
reduction of the now large storage capacity in the area (e.g. Haringvliet estuary) will
directly lead to an increase in DWL in this region.
3 Safety system; calculations
First the system and the safety calculations will be described for a one-barrier system.
This will enable a good understanding of the calculation procedures foliowed. Then the
procedure will be extended to the two-barrier system.
3.1 The one-barrier system in general
A schematic representation of the one-barrier system is given in fïgure 3. The system

consists of two major parts: the barrier (Sw) that is used to prevent the sea from coming up
the river, and the dikes to protect the land against the
water. Since it is not necessary to close the banier for all mean
tide
„,P. _
*l *
storm surges a criterion for closure (CL) is defined. For
this criterion a water level will be used. If the Predicted
Maximum Sea Level (PMSL) exceeds the CL the barrier A A A-.'
has to be closed. A very low CL would cause the largest
reduction of the RWLs but also many barrier closures.
VVV/V/
i
Since the frequency of closure of the Waterway barrier is
restricted the CL has to be chosen carefully.
1
storm |
surge __
The safety against flooding depends on the height of
the dike with respect to the water levels in front of it.
Even with the barrier in operation, dikes will still be
necessary. In the safety system that describes the protecti-
on of the land against flooding the barrier is only one but
/T\,
L J
vital part. In dike design the water levels are represented storm
profüe
by the DWL. Dikes are constructed in such a way that
there is a fixed safety margin as long as the DWL is not
exceeded. The barrier reduces the influence of the sea.
The water levels behind the barrier are therefore to some
s^ "
extent man-made. This also implies that the reliability of
the barrier, its operation and human errors are of influen-
ce. A probabilistic approach that integrates functional and
Flgure 4 5 m
structural reliability is therefore indispensable if an overall ^ Pt^Heteni
assessment of the safety system and the related DWLs is PP
to be achieved. This is described in the next sections.
3.2 Probabilistic calculations for the one-barrier system
In the Netherlands it is common practice to calculate DWLs using extrapolated

hydraulic boundary conditions and a numerical model to simulate the water levels. The
altemative, extrapolating measured water levels, cannot be applied when major changes
(like adding a barrier) are made to the hydraulic system.
To simulate the effect of the barrier on the RWLs a one-dimensional mathematical
model is used. Boundary conditions are water levels at sea and river discharge. Sea water
levels result from a superposition of the astronomical tide and a storm surge. Three
parameters are used for characterization of the storm surge (figure 4): the maximum storm
surge height (m), the storm surge duration (s) and a phase difference (p). A mean tidal
curve is used for the astronomical tide. One storm profïle (a combination of m, s and p)
has a Maximum Sea Water Level (MSWL). The river discharge (q) is treated as a constant
during the storm since it varies on a much longer time scale than storm surges. One
combination of boundary condition parameters leads to a Maximum River Water Level
(MRWL) at every location along the river. The hydraulic model is used as a transfer
function from boundary conditions to MRWLs at various locations.
Using the transfer function to solve the integration boundaries, the probability of
exceedance of the MRWL can be calculated from the following integration:
P{MRWl>mrwï)=\\\\fm^tAdmdsdpdq (1)
MRWL>mrwl
in which f denotes the probability density and d the differential. The frequency distributions
of all boundary condition parame-
ters are known by extrapolating
data from long periods of registra-
tion. Assuming statistical indepen-
dency, which is realistic, the
probability density function of
parameter combinations (4^,,,) is
also known. The integration pro-
cedure is illustrated schematically
in figure 5 for the simple case of
only two boundary condition
parameters, the discharge ( q ^ )
and the MSWL. The figure shows
the probability density functions of
the boundary conditions (MSWL,
q ^ ) , the probability of the com-
bination (iVp,), and lines of equal F i g u r e 5 Schematic integration procedure
MRWL at a certain location,
which is a result of the transfer function. The hatched area indicates all combinations that
lead to MRWLs above the level MRWL=4. Integrating the probability density function
over this area results in the frequency of exceedance for this level.
3.3 Integrating risk analysis
In section 3.1 it was explained that the effect of the barrier is influenced strongly by
the functional and structural reliability of the barrier. To describe these influences more
quantitatively a risk analysis has been carried out. Figure 6 shows the event tree with all
possible branches that may lead to a MRWL exceeding the DWL. The same principle
applies to the exceedance of any other water level or any other result. The upper part of the
figure shows the cause of events if the system functions properly. There are four paths
where things can go wrong:
1 PMSL < CL, so the barrier is not closed, but MRWL > DWL
2 PMSL > CL, but the barrier is not closed, and MRWL > DWL
3 PMSL 5: CL, the barrier is closed but is not strong enough to withstand the loads
from wind and water; it collapses, and MRWL > DWL
4 the system functions well but still MRWL > DWL
The first two paths reflect the functional reliability, the third path reflects the structural
reliability and the last path reflects fate. The four probabilities related to the paths (indica-
ted as PI to P4) should be calcu-
lated for every boundary condition
combination. However, in the PSWL barrier barr i er
es
MRWL | * y
calculation procedure a more above CL? closed? sustams? <DWL?
general and efficiënt approach is fC^3

foliowed. This is discussed now.
MRWL yes MRWL yes MRWL
For path 1, 2 and 3 the water
HO HO Ho
r r
<DWL? «DWL? «DWL?
levels are calculated with an open
barrier. In principle the resulting
P = PCMRWL>DWL5
A
water levels of path 3 will be
lower than in case of an open
barrier. However, to simplify Figure 6 Basic event tree
matters it was assumed that in case of collapse the hydraulic system would react as if no
barrier was present. This means that for each hydraulic boundary condition two hydraulic
states are relevant: the barrier is open or it is closed. Reliability can then simply be
integrated by performing the integration from formula (1) for both open and closed barrier
situations, taking into account the probabilities of the barrier being open or closed. This is
done by weighing the probability function f with the probabilities of each state ( P , ^ and
Pck«d)- m formula for some parameter X (e.g. water levels):
^X>xHSS!f^^*Popend^^l * jJSJfmjsJ>A*PehsJmdsdpdq (2)

X >X X >X
Bpen üoSJ
The (conditional) probability of the barrier being open depends on the PMSL, the failure of
not closing and the failure due to coËapse. The relations are described in section 3.4.
In the calculation procedure the hydraulic model is run for all boundary condition
parameter combinations, once with the barrier open and once with the barrier closed. All
parameters of interest (water levels, hydraulic head etc.) are stored in databases. These
databases are used to perform the integration (2). With this procedure it is very easy to
perform a sensitivity analysis on parameters related to the functional and structural
reliability since it only involves the weighing factors ( P , ^ and P^^J) and the databases
remain unchanged.
3.4 LfStermining tiie proauuiuty oj tne state oj <x ucomet
The three effects leading to an open barrier situation will now be describes and
translated to a probability.
Open barrier due to inaccuracv of the PMSL

Predicted water levels are often inaccurate. The difference between the PMSL and the
MSWL has a stochastic nature. In case of the PMSLs at Hoek van Holland this difference
can be modelled with a Gaussian distribution (mean /t, Standard deviation o). Then PMSL
only depends on MSWL, /t and o. If the PMSL is below the CL the barrier will not be
closed. Now for every MSWL (i.e. combination of boundary conditions) the contribution to
Popen caa te calculated from:
PcemPMSL = Pr{PMSL(MSWL) < CL} (3)
This probability is easily derived from the normal distrubution.
Open barrier due to not closing

If the PMSL exceeds the CL then the barrier should be closed. However, the barrier
may not be closed because of human or technical errors. The probability of these errors is
schematized with a constant value C. The contribution to T'opCB then becomes:
P ^ c = Pr{PMSL(MSWL)>CL}*C (4)
Open barrier due to collapse

If the barrier is really closed and the loads on the barrier become too high it collapses.
The barrier's strength (r) is a stochastic variable. Normally a lognormal distribution is
assumed. Comparing each hydraulic load (1) on the barrier with the strength r results in a
probability of collapse. The contribution to P then becomes:
PcpoiPb = Pr{PMSL(MSWL)>CL}*(l-C)*P(l>r) (5)
The total probability of collapse (PJ results from intgrating P(l > r) over all hydraulic
boundary conditions. Since P b is a design parameter which in this project is set to a very
low target value (10 5 to 10"*/year) it is easily seen that the influence of collapse on DWLs
with a frequency of exceedance of 10^/year or more will be negligible.
For each boundary condition P , ^ is determined as the sum of the contributions des-
cribed above. The probability of the barrier being closed is then simpïy Pcloeed = 1 - P , ^ .
3.5 Extending the procedure to a two-barrier system
In the two-barrier system the area is protected by two barriers: one barrier in the New
Waterway and one in the Hartelchannel. In principle the procedure to calculate the
hydraulic effect of two barriers is the same as foliowed for the one-barrier system.
However, a few differences have to be mentioned. First the event tree illustrated in figure 6
now applies to both barriers. It should also be realized that functional reliabilities may be
correlated. For example: the PMSL for both barriers, used to decide on a closure, will
originate from the same source. Secondly, instead of two, there are now four possible states
of the system. Each barrier may be open or closed and four combinations are possible.
Like with the one-barrier system, the hydraulic effects are calculated with the hydraulic
model, which is now extended with the Hartel barrier. The results of all four states are
stored in databases. A typical example of these results is shown in figure 7. Here the
MRWL in Rotterdam is shown for all four states as a function of only two boundary con-
dition parameters (MSWL and river discharge). The figure illustrates the different results
obtained from the four states of the system for the same boundary condition (e.g. H,, Qj).
If the probabilities of the four states are marked P„ P b , P c , Pd respectively then the
integration procedure is described by:
P(X>x) = £ [{fff^'P, dmdsdpdq (6)

X>x
The results of this approach are discussed in he next section.
4 Effects of introducing a second storm surge barrier on water Ievels
4.1 Introduction
From the characteristics of the hydraulic system, some basic principles of the effect on
water Ievels have become clear without having made one computation (see section 2). This
section will elaborate on this, using some preliminary results of the safety analysis. These
results are emanating from a sensivity analysis carried out on the base of the calculation
procedure as described in section 3, but with two instead of four hydraulic boundary
condition parameters (MSWL and q). In the characterization of the storm surge two para-
meters (storm duration and phase difference) were set to representative values (29 respecti-
vely 4.5 hours). The results are used in an indicative sensivity analysis which will be
extended carrying out a detailed study. This is in preparation now.
The analysis incorporates the characteristics of both regions and refers to thesituation
without the extra harbour entrance, with only a barrier in the New Waterway.
S w and S H open. S w open and S H closed.
River dachargs [ITTVSÏ
Sy, closed and SH open. S w and SH closed.
Figure 7 Iso-level Unes for Rotterdam
When figures are shown this concerns comparison of the situation with both Beerdam and
Rozenburg lock removed (BD+RL OPEN) to the reference situation (REFERENCE).
4.2 Maximum reduction on design water levels by closing the barrier(s)
A first insight in the sensitivity analysis can be gained by depicting the situation when
both barriers (or one in case of the reference) are closed 100% effectively at each high
water period. In this way uncertainties in closure are excluded. The results show - with
respect to the situation without a barrier - the maximum reduction on DWLs. In figure 8
the iso-level line of the DWL is represented for a characteristic location in each of the two
regions. Rotterdam is situated in the tidal region and Hellevoetsluis is located along the
southern branch in the transition region (figure 2). For each location the iso-level line is
shown for the reference situation and for the situation when two barriers are present and
closed each high water period. In section 3.2 is explained that the form and size of the area
above the iso-level line is representative of the frequency of exceedance of the water level.
With the aid of equation (1) - of course reduced to two variables - the frequency of
exceedance line of water levels can be calculated. The result is illustrated in figure 8 below
the iso-level lines, for each location for the two concerned situations. The point correspon-
ding with the iso-level line in the figure above is explicitly indicated. In both figures it can
be seen that the maximum reduction on DWLs will be less by creating a second connection
with the sea. For both locations the area above the iso-level line is larger and thus, the
frequency of exceedance of this water level is larger.
Isolines of DWL Rotterdam Isolines of DWL Hellevoetsluis
'\ :
1 - ^ ^ " ^ •
I •»•£» : ^^""^
ï aoo
s
1JOO
5000 10CO0 5000 10O00 15000
R i v w discharge faf/sl Rmwr discharge [m*/sl
Frequency o f e x c e e d a n c e Frequency of e x c e e d a n c e
Water levels Rotterdam Water ievels Hellevoetsluis
B O P L OPEN
S» and ^ cfoeod
•g 4JOO
| 4.00 I
H • | MUL
ti
...j^^ss
i i
11
_
.
«r* icr* 10- 10-* «r* ter7 ia* 10" 10* IO- 1 io-« icr»
Freqjency of exesedanoe [1/yeor]
Figure 8 Isolines and frequenties of exceedance of two locations for one (KEFERENCE)
and two barriers (BD+KL OPEN) closed.
The difference in the two locations can be found in the exact magnitude of the frequen-
cy of exceedance of the DWL, again: in a situation with (a) closed barrier(s). In the case of
Rotterdam this frequency is very small: by closing the barrier(s) the water level at this
location is reduced substantially to a frequency that lies far below the required safety. For
Rotterdam the frequency of exceedance of the safety Standard amounts to 1-lCTVyear. As
was stated before: the conditions when both barriers are closed do not contribute to
situations that determine the height of the DWL.
This is not the case for Hellevoetsluis. The frequency of exceedance of the safety
Standard amounts 2.5-10^/year. As can be read from figure 8 this safety Standard is almost
achieved under conditions with (a) closed barrier(s). Indeed, as was stated before, DWLs
for locations in the transition region are determined by accumulation of river discharge,
with (a) closed barrier(s). These characteristics are important to realize: with a second
connection to the sea the increase in frequency of exceedance of the DWL with closed
barriers has to be minimized. Every significant increase does not comply with safety
requirements and cannot be compensated for by closing the barrier(s) more often.
4.3 Requirements for criteria for closure
For every location in the delta area iso-level lines can be made for the four states of
the system (see figure 7). Because DWL is fixed for each location it is possible to derive
some requirements for closure of the barriers. A reasonable assumption is made that
maximum water levels anywhere in
Envelope of isolines o f DWL
the delta area are equal or higher Sw snö SH closed - - Dordrecht Moerdi|<
when both barriers are open, com- Sw open, SH closed Spykenisse Dordrecht Streefkerk
Sw and SH open Spykenisse Dcrdrecht Streefkerk
pared with the case when only the
Hartel barrier is closed. Secondly, \
;\
the requirement is given that the \
'. \
frequency of closure of both bar- % 4.00
Z
riers has to be minimized. This --^ "* -
5 3.00
implies that the Hartel barrier has ^ ^ — - T ^ ^= "~"*">--~
to be closed more often (section 2). 2.0O
Having represented the iso-level

lines of DWL for each location in 0 5000 10000 15000
3
the area in a situation when the River discharge [m /s]
barriers are still open, the envelope

of the lowest lines (i.e. of conditi- F i g u r e Q Envelope of condiüons requiring the closu-
ons most occurring) will indicate re of Hartel barrier and Waterway barrier.
the conditions when the Hartel bar-
rier has to be closed (figure 9). A storm surge barrier cannot be closed 100% effectively.
By quantifying the uncertainties in the closing procedure however one can derive a crite-
rion for closure (the CL). For this the CL has to be somewhat lower than the DWL enve-
lope. A similar reasoning concerning the hydraulic conditions that require the Waterway
barrier to be closed is legitimate. In this derivation however, one has to focus on conditions
that only the Hartel barrier is closed. This relationship is also represented in figure 9.
Locations that are part of the relationships are dominant in determining safety require-
ments. Figure 9 shows that in case of high discharges these locations are situated in the
transition region (Dordrecht and Streefkerk). For lower discharges the tidal region is
dominant (Spijkenisse). Notice that in case of high discharges both relationships do not
differ much.
The envelope of conditions that lead to DWL somewhere in the delta area when both
barriers are closed is illustrated by the upper line in figure 9. Above this line DWL
exceedance cannot be avoided by closing barriers more often.
4.4 Effects ofchanging the strategy ofoperating the barriers
By studying the effects of closing one or two barriers on RWLs much of the sensitivi-
ties in the system have become clear. This can also be used in indicating points of
significance in the formulation of an operational control of the barrier. The most important
operational aspect regarding safety is the moment at which the barrier has to be actually
closed. To close the barrier the PMSL must be higher than the CL and a Closing Condition
(CC) must be actually reached. This section will deal with the choice of a CC.
In formulating a strategy of operational use, one can optimize for many aspects. Within
the scope of this paper, only aspects tiiat affect safety against flooding will be dealt with.
Optimization of e.g. technical aspects of the barrier are clearly left aside.
In the transition region the situation with closed barriers is dominant for DWLs (because of
the accumulation of river discharge). From figure 8 can be derived that in the case of
Hellevoetsluis, safety endangering conditions are most likely to appear at periods of high
discharge. At discharges of approximately 7000 m3/s or less DWL will only occur (with
closed barriers) in combination with very severe storm surges (MSWL of at least NAP +
4.5 m). The combined probability of occurrence of these conditions is so limited, that it
does not contribute to the frequency of the safety Standard. At high discharges however it is
essential that only minimal inflow during closure of the barriers is allowed into the delta
area. This can be accomplished by closing the barriers at the last turn of the tide (with
minimal inflow) before the storm surge rises. In this way minimization of MRWLs and by
that of DWLs is achieved. The iso-level lines in figure 8 are indeed based upon computati-
ons wherein both the Waterway barrier and - if applicable - the Hartel barrier are being
closed at the turn of the tide (local zero inflow) at discharges equal or greater than 6000
m3/s. For lower discharges a part of the storm surge is allowed into the delta area and the
barriers are closed at a (local) fixed water level. Comparing both iso-level lines of
Hellevoetsluis it can be seen that here the lines are deviating from each other. The probabi-
lity added by the lower iso-level line is very limited (compare the change in frequency of
exceedance) so it barely affects DWL. Applying the same strategy at extreme high
discharges would certainly increase DWL.
For DWLs in the tidal region the CC is of much lesser significance. In this region it is
more important that the top of a storm surge that will cause exceedance of DWL with (an)
open barrier(s) is cut off to some extent. As safety requirements regarding the closing
condition at discharges less than 6000 m3/s are not as strict, it is possible to optimize for
other aspects. For example a CC of local zero inflow may reduce bottom protection near
the barrier and a CC at a high water level may reduce hydraulic head over the barrier.
5 Conclusions
To meet the requirements of both safety against flooding and minimal closure of the
barriers the use of an integral safety concept is indispensable. In this concept both
hydraulic modelling and the functional and structural reliability of the two-barrier
system are to be handled probabilistically and integrated to assess safety conditions.
Characterization of the hydraulic system of the Rhine and Meuse delta area is very
important in understanding the overall effectivity of the barriers in protecting an area
from flooding. In this way aspects that affect the safety system can be assessed.
6 Acknowledgements
It is obvious that the study presented in this paper involves an enormous number of
computations. Most of this work is done by mr. Henk de Deugd and by mr. Ton Visser
also from Rijkswaterstaat. The authors are duly grateful for this.
Abbreviations and symbols
BD = Beerdam Pb = probability of collapse

C = probability of not closing PMSL = predicted maximum sea level
CC — closing condition q = river discharge
CL = criterion for closure RL = Rozenburg lock
DWL = design water level s = duration of the storm surge
m = storm surge height SH = Hartel barrier
MRWL = maximum river water level Sw = New Waterway barrier
MSWL = maximum sea water level P- = average of error in PMSL
NAP = reference water level a = Standard deviation of error in
p = phase difference PMSL
THE DESIGN AND CONSTRUCTION OF THE NEW WATERWAY STORM
SURGE BARRIER IN THE NETHERLANDS
technical and contractual implications

by
J.P.F.M. JANSSEN, Rijkswaterstaat, Road and Hydraulic Division
P.O. Box 5044, 2600GA, Delft
A. van IEPEREN and B.J. KOUWENHOVEN
Rijkswaterstaat, Building Division
P.O. Box 72, 3150AB, Hook of Holland
J.M. NEDEREND, A.F. PRUIJSSERS and H.A.J. de RIDDER
B.M.K.Banier Design and Construction Group
P.O. Box 63, 3150AB, Hook of Holland
(The Netherlands)
1. Introduction
For the south-western part of the Netherlands the safety against flooding is prescribed by the Delta Act.
This Act was made after the severe flood disaster in 1953 when nearly 2000 people drowned. The Delta Act
prescribed the shortening of the coast line by closing off many of the existing tidal inlets and the
strengthening of the remaining dikes. It also proclaimed that the New Waterway and the Western Scheldt
had to remain open because of the harbours of Rotterdam and Antwerp. The dikes along these waterways
and along the waterways connected with it had to be strengthened. In 1985 a reexamination of the design
water levels led to even higher values than those considered in the Delta Act. This meant another 200
kilometres strengthening of river dikes including costly protection works in densely populated or
infrastructurally complicated areas. See figs 1 and 2 for the project area.
The work involved with the strengthening of dikes would
not be finished before the year 2020. Therefore, in 1987,
the Dutch Government initiated a study to reconsider a
movable storm surge barrier in the New Waterway. To
be feasible the barrier had to meet several goals. The
most important ones being:
* closing frequency of the barrier less then once every
10 years now and once every 5 years after 50 years of
operation (due to 25 cm sea level rise)
* prescribed reduction of design water levels (a local
water lever with a fixed frequency of exceedance) at
two representative locations, the cities of Rotterdam
and Dordrecht
The prescribed maximum closing frequency shows the
strong emphasis on the open character of the Rotterdam
harbour. Also other goals reflect that the Dutch
Government highly appreciated the economie importance
of the Rotterdam harbour.
After an extensive study the feasibility and the
effectiveness of a movable barrier was shown and 5
contractor consortia were asked to make predesigns for
the barrier. At the end of 1989 a definite choice was
made for the sector gate barrier designed by the B.M.K.
Barrier Design and Construction Group and a design and
construct contract was granted. The BMK barrier turned
out to be an economical alternative for the original
programme in which the dikes had to be improved. In Rg j (md 2. T^ project ah
concordance with this decision the Delta Act had to be
-1-
changed. Clearly, also after the barrier is completed and is operating, dikes will be necessary and still some
will have to be strengthened although to a much lesser extent. This was taken into account when making the
decision in favour of the barrier.
In a previous contribution [1] the preconditions for the barrier design were briefly described. The three
main predesigns were also outlined in that contribution, among them the BMK barrier. At that time the final
choice had not been made yet.
This contribution concentrates on the next phases. It describes the tender philosophy and the procedure of
selecting the frnal barrier design (par. 3), the design and construct contract with all its specifïc problems and
advantages (par. 4), the details of the current barrier-design which is slightly different from the original one
(par. 5) and also a description of the design process itself (par. 6), the hydrodynamic problems that strongly
influenced the final geometry of the barrier (par. 7) and last but not least the management of the design
process (par. 8). Finally some conclusions will be drawn. However, fïrst the project area and its hydraulic
features are described to get a better understanding of the preconditions imposed on the barrier-design.
2. Project description and hydraulic system
The project area covers the lower river reaches in the south-western part of the Netherlands, as shown in
figures 1 and 2. Figure 2 represents the Rhine and Meuse Delta network, in which water levels will be
influenced by operating a storm surge barrier in the New Waterway. Combined with the dikes along the
river branches the barrier will provide safety against fiooding.
In figure 3 the hydraulic system is shown in a schematic
form. Basically there are two river branches with several river discharge —,
1 projcctcd barrier
connections. The southern branch runs into the large 1- •-":•• , , , , , , , , , , : . - - . : , ; , , , , : S:;:
Haringvliet estuary. This estuary is separated from the ' • / 4- N Now W,itcrw,iy •4—
sea by a barrage with large discharge sluices. These
sluices are closed during high tide to prevent sea water
from entering the delta region. During low tide they are
_1.
Sea 1
/LJff
^=========f=^^
closed only when the river discharge is low. In this way ( Haringvliet ^ ^ ^
the water is forced to discharge through the northern
\
branches, to stop the salt intrusion from the sea.
The northern branch runs freely into the sea through the r i v e r rijke
1 1 polder
city of Rotterdam and the New Waterway. Tidal ^—— scadikc
movement and storm surges enter the system through this
northern branch. Close to the river mouth the water level p}„ 3. fhe SYStem
is determined by the tidal movement and the storm surge.
Travelling upstream the incoming wave is damped and
the phase is shifted. The water levels in this intermediate region are determined by both sea-water levels and
river discharge.
In the hydraulic system the effect of the barrier on water levels is twofold. One effect is the reduction of
water levels because the storm surges cannot enter the system any more. On the other hand there is an
increase of water levels because of the accumulation of river water behind the barrier.
To calculate the total balance of effects the hydraulic system has been modelled by a mathematical open-
channel network model. The schematization of the system is conform figure 2 and consists of about 200
branches and nodes. This model has been in use for a long time for prediction of water levels on a daily
basis and to determine design water levels for the situation without a barrier. To study the effects of a
barrier the model has been extended with a weir structure. Boundary conditions for the model are the river
discharge and the combined effect of tide and storm surge.
Due to shipping strict requirements have been imposed upon the barrier-design. The most important ones
being the maximum closing frequency of the barrier and the restriction of operation induced
translation-waves within specifïed limits. Also the space required for the passage of ships is prescribed. The
minimum width above Mean Sea Level (MSL) -10 m is 360 m and the minimum sill depth is MSL -17 m.
Since the performance of the barrier is measured in statistical quantities (e.g. design water level reduction
and closing frequency) probabilistic calculations are necessary. The hydraulic model is used to translate
boundary conditions into local water levels. The probability density functions of the boundary conditions are
-2-
used to derive the probability of exceedance of the local water levels by a probabilistic full integration
method. In the probabilistic calculations also the reliability of the barrier is taken into account. The
reliability of the barrier concerns the probability that the barrier is not closed due to:
- inaccurate water level prediction
- human or technical errors
- collapse (insufficiënt strength)
The performance of the barrier is also influenced by the characteristics of the barrier and its operation, e.g.
moment and duration of both closing and opening.
Since design water levels are a Governmental responsibility it seems logical to set technical requirements to
the characteristics of the barrier that influence the design water levels. On the other hand the contractor
should have as much freedom of design as possible.
The technical requirements have been established by a sensitivity analysis for the barrier performance. In
this analysis the influence of the reliability, the characteristics and the operation (strategy) of the barrier is
established. Aiming for adequate performance an optimum strategy and realistic reliability targets (especially
with human errors involved) can be described.
The optimum strategy is:
- closure at a sea water level of MSL +2.00 m for river discharges less than 6000 m3/s and at slack water
for higher discharges,
- opening at equal water levels on both sides of the barrier,
- discharging water through the barrier between two high waters at sea (i.e. when water levels on the
riverside of the barrier are higher than on the seaside) with a minimum wet surface of 1000 m2 net that
has to be realised within 20 minutes.
The reliability targets are:
- probability of not closing due to human or technical errors less than 10'3 on demand,
- probability of collapse less than 10"6 in any year,
- probability of not opening due to human or technical errors less than 10 4 on demand.
For the barrier-characteristics limits have been targeted. Within these liinits the barrier performs well. The
targeted barrier-characteristics are:
- full closure in less than 2.5 hours and 80% closed within 1.5 hours,
- full opening in less than 2.5 hours and 20% open within 1 hour,
- average retaining level MSL +5 m,
- leakage area through the closed barrier less than 100 m2 net,
- up to 4000 m3/s river discharge it is allowed to reduce the hydraulic head over the barrier by letting in
water through a limited opening in the barrier.
If the contractor satisfies the technical requirements, then the design water levels are sufficiently reduced.
Since it is not described how to fulfil the requirements, this procedure gives the contractor maximum
freedom of design.
3. Basic tender philosophy and selection procedure
The construction of the Eastem Scheldt storm surge barrier (1978-1986) induced considerable negative
publicity because of it's budget overruns (approx. 30% in total). Therefore, the intention was to realise the
storm surge barrier in the New Waterway following the basic principles of market mechanism philosophy.
The market (designer-contractors) should be given the opportunity to show their skills as opposed to the
usual designer-role of the Rijkswaterstaat of the Dutch Ministry of Public Works. Moreover this philosophy
was enhanced by political aims to diminish activities of governmental bodies in favour of a stronger market
approach. The result was that a "design and construct contract" was put out for tender according to
European rule. This tender philosophy was new to the Rijkswaterstaat. It should result in a project where
the owner would surely get:
- a storm surge barrier for a predetermined price,
- value for money through competitive designing and bidding by Europe's most outstanding
designers/contractors,
- a technical state of the art work that complies with the specifications,
- design and construction in one hand.
-3-
top view cross section
"floating"
l
"ballasted"
^
1 ^^z
sector gate (BMK)
^JWJWJWJW<^JWJWJwJ&
segment door
_/:» floating body
_ •^qr—s£: a/—T&77
pneumatic flap gate
hydraulic cylinder
A c
hydraulic flap gatc
e
s sliding door
Fig. 4: Schematic representation 5 pre-designs
-4-
Therefore, different European contractor consortia were invited to:
- draw up a preliminary design of a storm surge barrier, based on a limited number of rather abstract
technical (operational - and design boundary conditions) requirements,
- present a lump sum price for the design and construction of the barrier including maintenance for a
period of 5 years.
To do this the contractors were given a period of three months. The technical requirements were
intentionally formulated on a high level of abstraction to encourage innovative, bright and economical
solutions.
Initially 6 contractor consortia applied for the tender. After a first selection (considering the contractor's
design experience), 5 consortia were invited to prepare the conceptual design and to offer a fixed price
within 3 months.
During these 3 months the Rijkswaterstaat prepared the technical requirements for the next phase of the
selection. This resulted in the basic requirements for reliability, operational management and characteristics
of the barrier as described in the previous section.
To select the most appropriate technical and financial solution Rijkswaterstaat feit the need to acquire
sufficiënt knowledge of the specifïc "snags" in the design process. Therefore, they elaborated a number of
conceptual designs of the barrier to such a level, that all major problem areas and other areas of interest
were understood. This work was carried out by a team of specialists during the same three months in which
the contractors prepared their design and bid offer. This resulted in a clear understanding of what was
possible and what was not. Reliability and maintainability were the key criteria to select the designs on,
besides of course cost and design quality.
After the three month period, the 5 barrier-designs were evaluated. In fig 4 the barriers are shown
schematically. Based on the criteria mentioned above the sector gate and the segment door were chosen for
further competition. An important difference between the two designs is that the sector gate is floated into
the river and then sunk to the bottom and the segment door is riden into the river. In this final phase of the
competition the two consortia did hydraulic model testing on their designs at the Delft Hydraulics
Laboratory to eliminate all remaining uncertainties. The Rijkswaterstaat monitored these testings. Then the
consortia refined their bids. After a technical evaluation the Rijkswaterstaat finally selected the BMK sector
gate on the basis of cost. The BMK consortium was granted the contract amounting approximately 700
million guilders in total.
Main technical reason for selecting the BMK design is the simplicity of the technical concept. Moreover the
structure is easy to maintain, mainly in dry conditions with only limited parts remaining under water.
The selection procedure almost took one year. The use of a step by step selection procedure guaranteed a
continuing market mechanism throughout the procedure and induced the best quality.
4. Contract considerations
The contract involves both the total design and the construction of the storm surge barrier. Besides, the
contractor is responsible for the maintenance of the barrier during the first five years of operation. This
combination of responsibilities (risks) should give the contractor an incentive to search for an overall
economically sound solution for all design problems.
However, the New Waterway storm surge barrier is a structure to defend one of the most densely populated
parts of the Netherlands against floods. Defending the country against floods is one of the responsibilities of
the government which of course cannot be transferred to a private organisation. If the structure would fail
under design conditions this would mean a major national disaster. Moreover it is hardly possible from an
economical point of view to have the design insured against risks of this magnitude. Therefore the Dutch
government takes over the risks of faulty design work on delivery.
Another problem to bear in mind is the fact that the design Ioads (with very low frequencies of occurrence)
on the barrier are not likely to occur. It is thus impossible to run tests on delivery to determine whether the
barrier meets its basic requirements.
The three considerations mentioned above are directive to the role the owner should play in the process. It
has therefore been decided that a team of experts from the Rijkswaterstaat monitors the design and
-5-
construction. To enforce this role, the contract contains a so called "procedure of acceptance". Each part of
the design or construction that has influence on the barrier's performance has to be accepted by the team of
the Rijkswaterstaat. If no acceptance is given, the contractor is not allowed to release that particular part for
further engineering or construction.
To ensure a clear phase-to-phase working procedure, the contract states that a number of documents have to
be produced in a certain order. In this way fast tracking is eliminated for most of the work. Again,
acceptance of these documents by the owner is necessary in order to continue parts of the project.
Successively, basic design documents, engineering documents, specification documents, quality control
documents and construction plans have to be produced. All of this work is done following the requirements
imposed by the ISO-9000 standards. These quality assurance standards are rather new to the practice of civil
engineering in the Netherlands. Subsequently all parties involved (both owner and contractor(s)) had to put a
lot of energy into the design and the implementation of quality assurance (QA) and quality control (QC)
procedures and also to create and maintain a quality minded workforce.
This procedure has now been foliowed for about three years. The experiences have been mostly positive. In
the beginning of the project differences had to be overcome mainly concerning the new roles of both
contractor and Rijkswaterstaat. The contractor was used to building and now had to include designing, the
Rijkswaterstaat was used to designing and now had to refrain from doing this. The following exaggeration
will further clarify this: the staff engineers of Rijkswaterstaat feit that they were the only really experienced
designers; the contractor engineers feit that they had to go through the ordeal of some design work before
the real job, being the construction, could start. Rather soon after some initial "misunderstandings" both
parties found out that the only way to fulfil the assignment was by working together as the complementary
components of a team and to respect one another.
5. Description of the final design and barrier operation
The lay-out of the sector gate barrier

is shown in fig 5. On each side of ^*LjS-:
the channel an abutment is
constructed between the groynes. As
such, the space available to shipping
and water flow remains virtually
unaltered.
Inshore of the abutment a steel sector
gate is parked in a dry doek. A
sector gate consists of a circular 1. Steel sector gate
shaped retaining wall, which is 2. Sill and bottom protection
3. Parking dry doek -
connected to a hinge point by means 4. Abutment _^j
of two steel truss works. Due to the 5. Control centre ::=
circular shape, the rotation of the 6. Ball joint foundation block

sector gate is hardly influenced by 7. Locomobile and guidance tower
currents or hydraulic head. Fig. 5: Lay-out BMK sector gate
Therefore, the traction of the sector
gate can be simple and small, especially because the gates are floating during motion. As such, the motion
of the gates is similar to the motion of a ship.
The traction consists of a traction work -the locomobile- which generates the horizontal force. The traction
of the locomobile is transferred to a rail construction on top of the gate. Thus the locomobile rides on top of
the retaining wall. While the locomobile remains stationary the gate moves horizontally. As shown in fig 6,
the connection between the locomobile and the support near the doek must allow a vertical motion, while a
horizontal displacement is prevented. Once in position the floating doors are ballasted with water and thus
sunk tó the bottom.
In immersed position the sector gate rests on a sill consisting of large concrete blocks. On both sides of the
sill a conventional granular scour protection is situated. The structural integrity and, dimensions of the
barrier are largely influenced by the hydraulic head over the barrier.
-6-
The concept of the barrier is, however, mainly .uprxirttn» towcr
determined by the strict requirements concerning
obstruction obstruction to shipping during
construction and maintenance of the barrier.
The barrier has to be built to last for 100 years. In
its operational lifetime, closure of the barrier is
expected to occur once or twice in ten years.
Management of the structure focuses on inspection
and maintenance because, during its operational
lifetime, closure is expected to occur only once or
twice in 10 years. Therefore a doek was used to
store the gates when not operational. The doek can
be set dry to achieve maximum accessibility of all
vital components. As such, maintenance of the
sector galc
gates is as simple as the maintenance of a ship.
Moreover, during construction the doek is used to
assemble the prefabricated sections of the gate.
Construction and maintenance of the steel truss
works can be carried out onshore, thus anticipating
the requirements related to the prevention of
environmental pollution. The back bone of the
structure is the hinge. Due to the ship alike Fig 6: Schematic function Locomibile
behaviour the hinge must allow rotation in all
directions. Therefore a ball joint construction was
used. This rotation can even occur under the
ultimate loading condition as a result of large
waves. Due to the large dimensions the load
caused by the hydraulic head can mount up to 370
-^1 steel truss work
MN for each of the gates. The ball joint transfers

this large force to the sub soil by means of a
concrete foundation structure which is situated
TïjBrication
inshore of the abutment on a highly compacted ' film
sand-fill. This is shown in fig 7. The abutment ~__ ball of
consists of a sheet-pile wall (combiwall type) with steel plates
a sand-fill. Apart from the gates and the """ 5 rk radius
locomobiles the abutment comprises buildings for
foundation structure
the operation team and the power supply. The
abutment also protects the vital parts of the barrier
against the large colliding forces from shipping Fig. 7: Cross-section ball joint
accidents.
A global outline of an anticipated barrier operation can be described as follows:

- Initially the barrier is at rest in the doek; the barrier control centre is unmanned but the automatic control
system is alert
- The Storm Surge Warning System predicts a storm surge level that exceeds the critical level
- The crew of the barrier (decision team and operators) is waraed by the automatic control system; if the
decision team is not able to reach the barrier in time the automatic system is able to continue unmanned;
it will only ask for authorization of crucial decisions
- Preparations are started (start of energy generators and rilling of docks)
- The ship-traffic is stopped 2 hours before closing the storm surge barrier
- The dock-doors are opened (20 minutes)
- The automatic control system asks for authorization of the actual closure
- The floating barrier is driven out into the river by the horizontal moving system (30 minutes)
- The valves are opened and the barrier immerses (120 minutes)
- The barrier rests on the sill with approximately 45 MN pre-tension
-7-
- The moment of equal water levels at the up- and downstream side of the barrier is predicted
- Some ballast tanks are emptied to reduce the vertical pre-tension to zero (45-120 minutes)
- When the water levels on both sides of the barrier are equal the barrier is floated by emptying the ballast
tanks (120 minutes)
- The barrier is driven into the docks (30 minutes)
- The dock-doors are closed (20 minutes) and the doek water level is lowered till the gate rests on the
doek floor (supports)
- The operation ends by demobilisation
In fig 8 the operational timeline of the barrier is illustrated. Together with the water levels on both sides of
the barrier the most important activities are shown.
Fig. 8: Operational timeline
6. Required performance of the storm surge barrier and how to deal with it
Required Performance
The required performance of the storm surge barrier can roughly be divided in two sets of requirements: (1)
functional requirements describing the operational use of the storm surge barrier and (2) given the
operational use, the overall performance.
Functional requirements
In order to reduce the water levels in the hinterland adequately, barrier-characteristics are defined (see par.
2). The most relevant characteristics are:
- closing time not more than 2 1/2 hour;
- opening time not more than 2 1/2 hour;
- opening speed initially more than 8 m/hour (related to the requirement to discharge water whenever the
sea water level is lower than the river water level)
- operation induced translation-waves must be controlled within specified limits.
These requirements form the basis for the design of the horizontal and vertical moving system. To be able
to verify and dimension the systems, several simulation programs, including hydraulic models, have been
-8-
developed.
Overall performance
For the storm surge banier, when operating according to the functional requirements in the specified
environment, overall performance requirements were assessed for the relevant utilisation phases:
- probability of not closing: 10"3 on demand

- probability of collapsing: 10"6 per year
- probability of not opening: 10"4 on demand
The probability of not closing can be divided into three contributions:

- probability of human failure during closing operation : 9.10"4
- probability of ship-collision prior to closing operation : 8.10"5
- probability of technical failure during closing operation : 2.10"5
Lifetime
The required life-time of the Banier is 100 years. Parts of the banier that are practically irreplaceable
should be designed for this life-time. Therefore special attention had to be paid to conservation, cathodic
protection, concrete and monitoring.
How to deal with the performance requirements

Obviously the above requirements cannot be used for design without further elaboration. In order to
decompose the abstract requirements in appropriate requirements for design work, three basic analyses have
been performed: (1) risk analyses to transform the overall failure requirements into failure requirements for
(sub-)systems, (2) human failure analysis to determine the influence of human intervention on operations and
(3) analysis of ship collision.
Risk-analvsis
The defined probabilities of failure for the three
relevant phases have been elaborated with
fault-tree techniques. In that way all possible
failures at elementary level can be summarized
systematically in order to determine the overall
failure. In fïg 9 this is illustated for the case of
£
banier collapse.
Each elementary failure has its own quantitative
contribution to the overall failure. This
£
* Failure rates per year
contribution was distributed initially by the design
manager using both technical and economical 1 = Barrier faüs (9.1 E-7)
criteria. In principle expensive or difficult sub- 2 = Steel construction fails (6.1 E-7)
systems will have a large contribution, cheap and 3 = Hinge construction fails (2.2 E-7)
relatively simple sub-systems will have a small 4 = Threshold construction fails (7.3 E-8)
contribution. 5 = Sheet pile walls fails (-)
6 = Seadikesfail (1.0 E-8)
In the ideal siruation the distribution of sub-system 7 = Miscellaneous causes (-)
failure contribution to the overall failure of the 8 = Doors fail (2.1 E-7)
system is interactively assessed. In practice the 9 = Arms fail (4.0 E-7)
initially established distribution was hardly
chansed ^ ' ^: âu^'tree for bartier collapse
Systems Iike the ballasting-system or the horizontal
moving system were analyzed by a failure mode and effect analysis which resulted in the probability of
failure of this system. If the probability of failure exceeded the maximum allowed contribution, redundancy
or alternative solutions were considered. For some cases the allowed contribution had to be enlarged at the
expense of other sub-systems of the banier.
A full probabilistic design approach was used to establish the dimensions of the sub-systems of the barrier.
For this approach it was necessary to determine the probability density functions 'of both loads and
-9-
resistance
The most important step in the probabilistic approach is the assessment of the loads induced by the use of
the barrier in the environment during storm surges. Therefore the lower Rhine delta was modelled by a
mathematical open-channel network model and by fuU integration the probability density fünctions of
relevant loads were established. The following environmental variables were used: (1) storm surge level, (2)
tidal phases, (3) storm duration and (4) river discharge. One of the outcomes is the probability-function of
the hydraulic head as is illustrated in fïg 10.
For the determination of the element resistance
(strength, stability), initially an attempt was made
8
to distribute probability density fünctions of loads
to the designers in order to enable a full 'S'
probabilistic design. It became apparent that such a CS
probabilistic methodology did not work for this D
si
complex project. Therefore in most cases a more o
4
practical method was applied, based on a
conversion of the probabilistic results into design
I •—,
loads and partial safety-factors. In that way the -o >>
designers could use the standards. si
-> .
10-2 10-4 10-6 1 0 -8
Human failure
All activities with human invoivement before and exceeHance frennp.ncv H / v r )
during a barrier-operation were analyzed using a Fig. 10: Probability-function for hydraulic head
proven technique. Each activity was specifïed and
analyzed on: (1) basic failures, (2) effects of failures and (3) possibilities to retrieve.
With the help of quantitative data from earlier research for these categories of human behaviour, an
appropriate estimate of the probability of failure has been made. For some operations suitable measures
were recommended to improve the performance (e.g. monitoring by a supporting computer).
Ship collision
The navigation on the New Waterway is very dense. Each year about 80,000 ships pass the barrier-location
(10 per hour). Both historical data and reliability analyses showed that a ship-collision may be expected once
every 20 years. Therefore, a comprehensive study was performed to evaluate the consequences of a
ship-collision. Special navigational and damage-models were developed to quantify the effects of a collision
with respect to the fünctions of the barrier. In order to meet the performance requirements, parts of the
barrier had to be adjusted. Also appropriate repair-procedures and provisions were developed.
7. Model testing
During the tender-phase an extensive model testing program was performed at Delft Hydraulics Laboratory.
In order to investigate hydraulic effects and in particular the dynamic behaviour of the main gates under
operational conditions a 1:60 scale model of approximately 3 kilometres of the New Waterway was built.
However as a result of contract negotiations the design was altered significantly.
At that time the gates of the tender-design were provided with 40 sluice openings with a total area of 950
m2. The sluice openings could be closed by lifting gates. The cross section of the tender-phase barrier
concept is shown in fig 11a. The lifting gates had three fünctions: (1) discharge river water between two
tidal high waters or between two major storm surges when they are too close to open the sector gate
completely, (2) minimize heads and associated pressure-fluctuations under the gates when the barrier is
lowered to the bottom during closure and (3) reduce the maximum hydraulic head over the barrier by
allowing water-inlet under certain conditions.
During contract negotiations the design concept was changed. It was decided to skip the sluice openings
with the lifting gates since this reduced the costs significantly. Of the three sluice-functions mentioned above
two (discharge water and head reduction) can be realized by lifting the sector-gate(s). The time needed to
lift the sector gates to the required opening is almost the same as it is for opening the lifting gates. However
the decision to skip the sluices implied that a large part of the model testing had to be repeated. In particular
-10-
Tig. 11: Cross sections barrier (a): pre-design (b): contract design (c): present design
the decision to skip the sluices implied that a large part of the model testing had to be repeated. In particular
the closing and opening phase of the sector gate had to be investigated again, because the earlier model tests
were performed with the lifting gates open. The cross section of the barrier without lifting gates is shown in
fig 11b.
It was intended to repeat the tender-phase model tests with the new barrier-design. The following hydraulic
conditions (design-values) were tested again:
S AH + AH"
[m] [m] [m]
7 1.40 0.45
4 2.75 0.70
1 4.00 1.25
0 4.50 0.90
where S is the distance between the barrier and the sill, AH+ is the positive head (water level at the seaside
higher than the water level at the riverside) and AH" is the negative head (visa versa).
After re-building the model positive head tests were performed. When the distance between the sector gate
and the sill was between 1 and 4 metres a strange phenomenon was encountered that had not been
recognised in the earlier model tests.
Both gates made an almost perfect out of phase cyclic heave movement with a one minute period. At the
same time downstream of the barrier a transverse standing wave developed between the abutments. The
period of the movement perfectly matched the wave-formula: T = L/c = 2*360/12 = 60 seconds, where T
= wave period, L = wave lenghth and c = wave celerity. The sector gates foliowed the downstream
standing wave, which can be explained by considering the small (10 seconds) eigenperiod of the gates. The
amplitude of the movement could be more than 3 metres. A similar phenomenon was found with the
negative head tests. Due to the geometry of the abutments now a pitch-movement was found with a period
of 30 seconds .
The cause of the phenomenon was not conceived immediately. The first attempts to improve the stability of
the barrier can be characterized as trial and error. It was found that the stability improved much by
adjusting buoyancy at the seaward side of the barrier.
Unfortunately when considering pessimistic sediment profiles at the sill the stability decreased drastically
again. In this phase however the phenomenon was understood. It can be explained by a rather simple model
[2]:
Assume an infïnitesimal water level rise at the downstream side. As a result the flow decreases due to
the decrease of the head. The downstream buoyancy of the gate increases (Archimedes), which results in
a greater opening causing an increase of the flow. If the extra flow due to the infïnitesimal rise is
-11-
positive, the initial rise will be foliowed by an extra rise due to the extra flow. Such a situation can be
qualifïed as unstable.
Following this model it turned out that the stability of the gates was strongly influenced by the bottom side
geometry of the gates. The model was used to pilot further optimisation of the barrier-geometry. It was a
major problem to design a geometry suiting both positive and negative head. After some minor
complications a solution was achieved that was stable in all operational conditions. The fïnal geometry is
shown in fig l i c . The entire operation of re-testing and solving dynamic problems however caused a serious
delay in the final design (approximately 1 year).
8. Organizational aspects and control of design
The most relevant requirements have been discussed in the previous sections. At the start of the project the
barrier concept has been decomposed in 20 sub-systems. The design work needed to develop these sub-
systems was described in CTR task units (Cost, Time, Resources) with fixed input and output. This was
done with the perception that the precontractual concept would not be subject to significant changes anymore
and hence the design work would be limited to the engineering of a few details. However, it became
apparent that this initial perception of the necessary design work was not correct. In particular it appeared
that the large number of requirements together make it difficult to control design activities. Therefore, a few
centralized coordination tools are used to link the performance of the storm surge barrier with the
contractual requirements.
The fïrst tooi was an overall system analysis in order to observe the behaviour of the system in its environ-
ment. Emphasis is laid on operational aspects for both Standard and non-standard use.
Given the probabilistic context of the requirements a full probabilistic design philosophy has been adopted.
In this philosophy the three fault trees as discussed briefly in the previous section play a central role. Using
the fault tree technique it is possible to determine the individually required probability of failure for the sub-
systems.
Due to the difficult interaction between the prescribed use of the barrier and the complex environment it is
necessary to have appointed specialists to obtain an appropriate distribution of loads. These specialiste also
have a consultancy task to convert the probabilistic loads into more familiar deterministic design loads and
to assist designers with full probabilistic calculations whenever necessary.
Due to the complexity of the barrier system (the sub-systems are strongly interrelated) two coordinating
systems are used to monitor interrelations. The fïrst one is the tolerance system which covers all possible
fitting problems. The second one is an overall interface control system. Initially the last system did not work
satisfactory due to the strong interrelation of the 20 structural sub-systems.
For this reason the 20 sub-systems have been grouped into five working clusters. For the formation of the
working clusters two criteria were used: (1) minimum number of relations with sub-systems outside the
cluster together with a maximum number of relations within the cluster and (2) an easy change-over of
engineering clusters into construction clusters. As
the project is rigorously divided into a steel
construction part and a civil engineering part, the BMK
managing director
last criterion was rather predominant in the
clustering philosophy. staff
groups
The effect of the clustering on interface control is
positive. Nevertheless some problems at the
BMK design BMK
interfaces have remained unsolved. This is due to steel department civ. eng.
the fact that a cluster leader experiences solving coordinating
problems within the cluster to be more important staff
than problems at the interfaces with other clusters.
In order to solve these remaining interface
problems 4 special target groups have been
installed. The organisational structure is
F
'MM
clusters
CTR": mmm
schematically given in fig 12. & 12: Organisation of interface control
-12-
9. Conclusions
The conceptual design of the storm surge barrier in the New Waterway was chosen due to the simple load
transfer system. The two defined aims in the development process were to use conventional methods of
construction and to minimize research efforts. These aims have satisfactory been reached, although a
substantial amount of research was needed due to the unacceptable dynamic behaviour of the sector gates in
floating conditions during closure and opening.
This has led to an extensive research programme performed by Delft Hydraulics Laboratories. During these
research activities which lasted about 2 years the exact geometry of the gate could not be fixed. Since the
geometry of the barrier has relations to almost all surrounding construction elements, the instability of the
changed concept caused severe design problems.
The barrier concept is very complex when considering the number of relations between elements and the
number of functions of elements. In consequence the design work required substantial coordination efforts.
Interface management and confïguration control has turned out to be key functions in this project. During
the project the organisation structure has been adjusted to improve the coordination and to implement these
functions.
The storm surge barrier project is the fïrst project of this size to be developed under a total QA/QC scheme
according to the ISO-9000 standards. QA/QC has proven to be a very learning experience for both the
contractor and the cliënt. Implementation of quality management procedures throughout the organisation
demands a great deal of effort especially in showing people that to have a single point responsibility towards
quality means that one has to monitor the quality of one's work. This implies that one should be able to
inform management about wavers and deviations without having the feeling that it means that one has done
a "bad job".
The tendering procedure, developed by the Ministry of Public Works, has resulted in a conceptual design to
be developed and realized with a fixed price plus escalation fïnancial contract. Unexpected phenomena, such
as the instability of the gates, have had a large impact on the design process causing difficulties of both
technical and organizational nature. An adequate check on the stability of the changed concept of the gates
prior to the contract negotiations could have minimized disruptions in the execution of the project.
After some difficulties of both technical and organisational nature the project is now well on its way. Due to
unforeseen problems the targeted delivery date has been moved from mid 1996 to mid 1997. In a quality
(and budget) driven project this is the only realistic escape to prevent quality loss.
10. Literature
1. Huis in't Veld, J.C. et al, A Storm Surge Barrier on the New Waterway, the Netherlands, 27th
International Navigation Congress, Osaka, May 1990.
2 Jongeling, T.H.G., Kolkman, P.A., Instable Behaviour of a Floating Sector-Gate Barrier, 113th
A.S.M.E. Winter Annual Meeting, Anaheim, California, USA, nov.8-13, 1992.
-13-
Projet et construction d'un barrage anti-tempête dans Ie Nieuwe Waterweg, aux Pays-Bas
(Specifications techniques et implications contractuelles)
Résumé:
Après Ie raz-de-marée de 1953 la loi Delta fut lancée pour assurer la protection du Sud-Ouest des Pays-Bas
contre les innondations. Cette loi prévoit la fermeture de nombreux bras de mer. De même, la loi tient
compte d'une vois ouverte entre la mer et Ie Nieuwe Waterweg, lequel donne acces a 1' important port de
Rotterdam (fig. 1 et 2). Les digues Ie long du Nieuwe Waterweg et ses liaisons fluviales devaient être
renforcés. En raison des incidences considérables sur ces régions, caracterisées par une population a densité
elevée et un aménagement du territoire complexe, ce programme de renforcement des digues est coüteux.
C'est pourquoi en 1987 L'Administration des traveaux publics et d'aménagement du territoire
(Rijkswaterstaat) lan$a une étude de faisabilité sur la réalisation d'un barrage mobile dans Ie Nieuwe
Waterweg. Parmi les conditions essentielles auxquelles doit satisfaire Ie projet figurent une reduction
déterminé du niveau des eaux du cours inférieur du fleuve et une fréquence de fermeture de 1/5 a 1/10 par
an pour maintenir 1'accèssibilité du port de Rotterdam. Au niveau de remplacement, Ie fleuve mesure 360
m de large et 17 m de profondeur. En fin de compte Ie barrage s' avere moins coüteux que Ie reforcement
des digues.
Cinq groupements d'entreprises furent invites a présenter dans les trois mois qui suivent un avantprojet sur
la base d'un contrat a prix fixe, comprenant étude et construction. Sur les cinq avantprojets (fig. 4) les deux
en are de cercle furent sélectionnés pour élaborer leurs projets respectiefs a partir de normes fonctionelles
détenninées. Finalement Ie projet fut confié au groupe BMK a la fin de 1989, en raison de la simplicité de
la construction et du prix. Le montant du contrat s'éleve a quelque 700 milions de florins (y compris les
cinq premières années d'entretien).
Le barrage se compose de deux portes flottantes en are de cercle, liées a des pivots (fig. 7) de part et
d'autre du cours d'eau par des bras en treillis (fig. 5). Les portes sont maintenues a flot pendant la
fermeture, puis abaissées par balastage sur un seuil de bloes en béton amenagé au fond du cours d'eau. Le
barrage s'ouvre par délestage des portes et parcage dans des bassins creusés de part et d'autre du Nieuwe
Waterweg. Ainsi, les portes reposent dans des bassins qui peuvent être fermés et mis a sec.
A cause du grand interêt public, l'Administration a tenu de suivre de prés le développement du projet et sa
réalisation. Dans le cadre de 1'assurance de la qualité, des normes de qualité (ISO-9000) et des procedures
d'acceptation sont utilisés.
Les normes de fonctionnement du barrage comprennent des specifications pour la retenue des eaux (vitesse
de déplacement, geometrie etc.) et des normes de probabilité d'échec. L'entrepeneur doit donc traduire ces
normes abstraites en une structure réelle. L'analyse du risque (fig. 9) et les calculs des probabilitées ont
servi a dimensionner les différents éléments structurales et les charges correspondantes a considerer, a partir
des normes établis de probabilité d'échec. Pour maltriser ce projet complexe, caractérisé par de nombreux
points de contacts, BMK a bati un système de controle autour de cinq unités d'étude (fig. 12).
L'avant-projet de BMK comprenait des portes de barrage avec des vannes coulissantes (fig. 11a), qui
pouvaient être utilisées pour 1'écoulement des eaux entre deux hauts niveaux de 1'eau. Même pendant la
fermeture, les vannes resteraient ouverts. Pour des raisons économiques, ces vannes ont été enlevées du
projet avec la supposition que 1'écoulement des eaux pouvait aussi être réalisés en soulevant les portes de
leur seuil (fig. 11b). Pendant les recherches de vérification du comportement sur un modèle reduit de 1:60
de Delft Hydraulics il apparut résulter de grandes instabilitées pendant 1'abaissement et le soulèvement des
portes. Le mouvement était exité par des lames transversales entre les rives. La geometrie des portes y
jouait un grand röle. En combinant un modèle mathématique au modèle reduit de recherche pour ce
phénomène, une solution optimale a put être trouvée (fig. lic). Ces problèmes ont générés un retard d'un
an.
Le projet d'étude et de réalisation du barrage dans le Nieuwe Waterweg dure maintenant prés de quatre ans
et, après quelques problèmes de demarrage, est bien en cours. On prévoit la fin des traveaux vers le milieu
de 1'anée 1997.
-14-
DESIGN OF THE BED PROTECTION OFTHE ROTTERDAM STORM SURGE BARRIER
R.E. JORISSEN11, D.P. DE WILDE11 AND E. BERENDSEN21
Ministry of Transport, Public Works and Water Management, The Netherlands
11
Civil Engineering Division, PO box 20000, 3502 LA Utrecht, The Netherlands.
21
Road and Hydraulic Engineering Division, PO box 5044, 2600 GA Delft, The
Netherlands.
ABSTRACT
In 1989 the Dutch government decided to build a storm surge barrier in the New
Waterway near Rotterdam. This barrier will be closed if a storm surge level is
predicted that might threaten the safety of an economically very important region
of the Netherlands. Contrary to earlier similar projects the design has largely been
contracted out to the builder of the barrier. Only some components of the design,
like the bed protection, are solved by a joint effort of the builder and the Ministry
of Transport and Public Works.
The design of the bed protection is based on two conditions to be fulfiiied : the
permissible probability of failure and the required lifetime of the construction.
Based on model tests a relation between the stability of the toplayer of the bed
protection and the hydraulic loads has been developed. This relation has been
integrated into a model, that describes the probability distributions of the hydraulic
loads. As a result of this integration the probability distribution of the required
toplayer diameter can be calculated.
1. INTRODUCTION
The project area is in the southwest of the

Netherlands as shown in the figures 1 and 2.
Figure 2 shows the dikes, which were to be
reinforced as a consequence of the Delta
plan.This plan was developedafter the flood
of 1953. It included shortening of the coast
line by closing several tidal inlets and
s-i ,/"" Belgium 50km
strengthening remaining dikes. In 1985 most
of this plan has been completed, when a re-
examination of design conditions led to higher
Figure 1
values than those originally calculated. This
implied very costly construction works in densely populated areas. Therefore, in
1987 the Dutch government initiated a study to consider a storm surge barrier near
Rotterdam. This barrier should significantly
reduce design conditions to avoid problems
related to dike reconstruction works. The pre-
scribed reduction of design conditions is
based on those areas, which would be most
affected by the reconstruction works. These
areas are within the cities of Rotterdam and
Dordrecht. From several predesigns a final
choice was made at the end of 1989. The
selected barrier design features two semi- Figure 2
circular doors with a length of about 220 meters.
The barrier is closed by rotating the floating doors
é.
into the river as shown in figure 3 and lowering
"X T the entire construction to the river bottom by filling
ballast tanks with water. This operation is
•^
sea river
completed in 2.5 hours. If closed, the doors will
rest on a threshold, that consists of concrete
X^\ elements placed on a riprap filter construction. On

both sides of the barrier a riprap bed protection is
necessary to prevent scouring in the immediate
Figure 3
environment of the barrier and threshold. Based
on the operational behaviour of the barrier the bed protection had to be designed
to withstand hydraulic loads during the closure operation (flood) and the opening
operation (ebb), If the doors are entirely closed, the hydraulic loads on the bed
protection are negligible.
This paper deals with the design of the bed protection and the reiated hydraulic
aspects. The design of the bed protection is based on two main principles : the per-
missible probability of failure and the required lifetime of the construction. The
contribution of the bed protection to the total probability of failure of the barrier is
set at a value of 10 8 per year. Because the required lifetime is 200 years, the use
of geotextiles as a essential part of the bed protection is not considered. Both the
filter and the toplayer of the bed protection will be made of riprap.
2. MODEL INVESTIGATIQNS
The stability of the riprap was investigated in two scale models, both a section (2D)
model and a 3D model. The section model was scaled 1:30 and the 3D model was
scaled 1:60. All numbers presented in this paper are based on the prototype.
Based on the operation of the barrierthe

gap [m] head [m] bed protection had to be designed to
withstand hydraulic loads during the
10.0 0.40 closure operation (flood) and the
7.50 1.40 opening operation (ebb). The barrier will
be closed as soon as a certain water
5.00 2.90
level has actually been exceeded and
3.00 4.10 the predicted maximum sea level
exceeds a critical level. Due to the
1.00 5.80
duration of the closing process (2.5
hours) a head will gradually develop.
This head will be influenced by the
development of the water levels in front
of the barrier and by wind effects. The head will, together with the effects of waves,
cause a hydraulic load on the bed protection at the riverside of barrier. To give an
impression of the order of magnitude of these hydraulics loads the head at 10"8 per
year conditions is shown in table 1 as a function of the barrier gap h s . The effect
of these loads were investigated in the model tests.
The barrier will be opened if the water levels on both sides of the barrier are equal
and the predicted maximum sea level does not exceed the critical level any more.
Mostly due to the duration of the opening process (2.5 hours) again a head will
develop. However this head is much smaller than the heads mentioned in table 1.
A typical design head for this situation is about 1 meter.
In closed condition the hydraulic loads on the bed protection are negligible. Therefore
the model tests were focused on the stabiiity of the toplayer of the bed protection
at severai stages of the closing and opening operation. In table 2 an overview of
the tests, which were carried out, is shown.
The stabiiity of the riprap was tested by

rneasuring the number of displaced Toet 2D/3D h
• 'S
Tml
l." 'J
flood
stones from each section of the bed /ebb
protection. The sections correspond
T403 3D 1.0 flood
with the planned sections in the
prototype. T 3 2D 1.0 flood
T 6 2D 1.0 flood
At the beginning of each test a small
T408 3D 1.0 ebb
head was applied. This head was
increased step by step until the design T 19 2D 1.0 ebb
head was exceeded or the riprap
T405 3D 3.0 flood
construction had failed. The head was
kept constant during a period of 2 hours T405F 3D 3.0 flood
(prototype). After each step the total
T420 3D 3.0 flood
number of displaced stones was
counted. These data were collected in T4 2D 3.0 flood
a relation between head and the total
T 5 2D 3.0 flood
damage.
T409 3D 3.0 ebb
Based on these relations the critical T409F 3D 3.0 ebb
head for each investigated geometry has
been determined. The critical head is T 17 2D 3.0 ebb
defined as the head at which the dama- T406 3D 6.0 flood

ge to the toplayer exceeds the number
T410 3D 6.0 ebb
of 100 stones per 1000 m 2 during 2
hours. T411F 3D 10.0 flood
Table 2
Also the streamprofiles during the mentioned phases were investigated. The results
of this part of the model tests proved to be very useful, because of the limited
number of stability tests which were carried out.
The values for the critical heads were used to determine the value of K in the
following stability formula. This formula is based on the well-known Shields relation
with a correction factor K to account for the distorted velocity profile (see Franken,
Jorissen and Klatter, [1]).
" C2*Y
(2)
UQ = ^/2*flr*(V/fe)
A relative density riprap [-] = (p s -pj/p w

Dn nominal diameter riprap [m] = V(M 50 /p s )
U0 reference velocity [m/s]
K correction factor or stability parameter [-]
C roughness parameter [m 05 /s] = 25*(h x /D n ) 1/e
4> damage parameter [-]
h., water level upstream [m]
h2 water level downstream [m]
hx local flow depth (see figure 5) [m]
The damage parameter MJ is taken to be 0.056 at the described failure conditions.
In figure 4 the results of the model tests are shown as values of K as a function
of the distance x from the barrier and the gap between the floating doors and the
threshold (hs). In this figure the results for the riverside (flood) and the seaside (ebb)
of the barrier are both shown.
The lacking information at some distances is clearly visible in figure 4. Also the
situation with a gap of 10 meters was not investigated for the riverside (flood) of
the barrier. This situation was only investigated for the seaside (ebb).
seaside (ebb) riverside (flood)
2.00- • m 2.00
h s = 1.0 m D
h,= 1.0 m
D
3.0 m •
h< = 3.0 m
1.50
'S hs = 6.0 m h s = 6.0 m
£
ra
8. 1.00
hs = iÖ.Öm
&•
V-..
<b o
« 8 - , 1.00
>.
* -"i» B B^P,
Si
« 0.50 • 0 . .50 K
500 400 300 200 100 100 200 300 400 500
•4 distance x (m) distance x (m)
Figure 4
The values of K turn out to be a function of barrier gap (h s in figure 5) and the
distance x from the barrier. It is believed that this behaviour of K is dominated by
the behaviour of the velocity U x . At x = 0 this velocity equals more or less the
reference velocity U0. This led to the assumption that if the local velocity U x is used
in relation (1) K becomes more or less a constant. This assumption is shown in
equation (3).
Kx* U0 = Ko* Ux (3)
K0 stability parameter at x = 0
Kx stability parameter at x
U0 reference velocity
U x flow velocity at x
The further analysis of the test results is be based on this assumption.

3. FURTHER ANALYSIS OF TEST RESULTS
The results of the model tests are further analyzed to investigate whether the
assumption (3) is applicable. If the result is satisfactory it can be applied to fill in
the lacking data on the stability parameters K. For this reason the streamline data
has been used to construct a momentum balance between the cross-section at the
barrier (x = 0) and a other cross-section (x). In this balance the effect of energy
dissipation due to friction has been included. Based on this balance a relation betw-
een the reference velocity U 0 and a velocity U x can be derived.
sea nver
t U ma<
'U(z)
Figure 5
The available streamline information is shown in figure 5. From the model tests the
parameters h x and Bx could be expressed as follows. The suffix s stands for seaside
(ebb), while r stands for riverside (flood).
x
(4)
S 1 2
(216-0>x
B„ = 216 - (5)
325
Ber = 90 + 9*hs (6)
(216-BJ** (7)
Bxs = 216 -
75
B„ = 180 (8)
The momentum balance between the cross-section at the barrier (x = O) and another
cross-section (x) yields equation (9).
Q*g*hjx Q*g*hT0
(9)
+Q*Qx*Ux+FfHcdon= +Q*Q0*U0
If the total water depths downstream of the barrier hTX and hT0 are assumed to be
equal, equation (9) can be written as equation (10).
B0*hs*UZ=Bx*hx*U;+Y,{x *BX*AX) (10)

o
Combined witha friction law, equation (10) canbe written as equation (11) or (12).
x
B0*hs*Uo=Bx*hx*Ux+E(k*Ux*Bx*AX} d 1)
o
BQ*hs
un (12)
Bx*hx+Il(\*Bx*AX)
\
If the assumption (3) and equation (12) are combined, a relation between Kx, K0,
U0 and x has been derived.
BQ*hs
Kx - KQ*
(13)
Bx*hx+T(X*BX*AX)
\
A friction parameter
B0 f l o w w i d t h at x = 0 ( = 2 1 6 m, see figure 5)
B x flow width at x
h s flow depth at x = 0
h x flow depth at x
In this relation only one variable (A) remains, because K0/ B0, hs, B x , h x are known
from the model tests and U0 represents the hydraulic load. This remaining variable
has been fitted to the data. The results of this procedure are shown in figure 6 and
table 3.
gap hs [m] Ko A situation
10.0 2.2 0.05 ebb
6.00 2.0 0.05 ebb + flood
3.00 1.8 0.05 ebb + f lood
1.00 1.6 0.05 ebb + flood
Table 3
Regarding to the values for K0 and A the following remarks can be made.
First of all it seems that the direction of the flow doesn't influence the value of K0.
This is logical, because immediately downstream of the barrier the flow direction
doesn't influence the stability of the riprap. It is only further away from the barrier
that due to a different horizontal velocity profile the stability of riprap will be
influenced differently. This means that the test results for the ebb situation can be
used to predict stability of riprap at the riverside of the barrier or vice versa.
The value of A turns out to be constant. The value of A can be related to the hydraulic
roughness parameter with equation (14).
A. = -3- (14)
For a A of 0.05 the hydraulic roughness is about 15 m° 5 /s, which is rather low. But
based on the very large diameter of riprap and the small values of h x immediately
downstream of the barrier this value seems to be reasonable. Further downstream
this vaiue of C is too iow, but the infiuence of this error in the cumulative type of
formula like equation (13) is relatively small.
4. DESIGN APPROACH
Based on the model tests a relation between the stability of the toplayer of the bed
protection and the hydraulic loads has been derived. This relation has been integrated
into a model, which generates the probability distributions of these hydraulic loads.
The procedure of determining these distributions has been treated by Janssen and
Jorissen [2, 3]. In this paper only a brief summary of this procedure will be given
based on the calculation scheme shown in figure 7.
start integral The hydraulic conditions in the project area are

calculation
determined by two boundary conditions : the storm
surge level at the river entrance and the upstream
select boundary river inflow. These boundary conditions have a
conditions (bc)
stochastic nature. The probability distribution
calculate comhined
functions of these boundary conditions are known.
probability of bc
Also, the occurance of storm surge can supposed
to be independent of the occurrence of a river
start hydraulic
computation f lood.
/ /
The hydraulic model is used as a transfer function

calculate water
levels, discharges f rom the boundary conditions mentioned above to
/ /
1 hydraulic conditions at variouslocations. Foreach
calculate reauired
diameters (D n ) relevant combination of storm surge and river
/ /
discharge a hydraulic computation has been made.
no During that computation not only the time
development of water levels and discharges are
yes
calculated, but also the required diameter for each
store maximal values section of the bed protection. This calculation is
for(D n )and
combined probability based on the equations (1) and (13). Of each
combination and each section the maximum
no
v^. all bc tre;ated ? J> required diameter is stored together with the
probability of occurrence of the combination of
calculate pdf storm surge and river discharge. As a result of this
for (D n )
procedure the probability distribution of required
toplayer diameter can be calculated. From this
end of integral
calculation distribution the design diameter, which has a
permissible probability of failure, can be
Figure 7 determined.
The bed protection consists of various sections in which different riprap diameters
are appiied. The question remains if these sections can be assumed to be correlated
or not. If they can be assumed to be correlated, each section is allowed to have
a probability of failure equal to the total probability of failure. If the required diameter
in the various sections are independent, the total probability of failure has to be
divided by the total number of sections.
From the design calculations it showed, that the hydraulic loads {•*• K*U) on each
section of the bed protection were completely correlated. This and the fact that the
Standard deviation of the stability relation is large compared t o the Standard deviation
of the actual strength ( = A * D ) , made the assumption of a complete correlation be-
t w e e n all sections acceptable. Also because of the relativeiy large Standard deviation
of the hydraulic loads it was decided t o treat the strength parameter AD as a
deterministic parameter. This rneans that the stochastic variables are the value of
K x and the reference velocity U 0 . For Kx relation (13) has been used. This relation
is s h o w n in figure 6 as the collection of solid lines. This relation has been included
in the hydraulic model mentioned earlier. The reference velocity U 0 is determined
w i t h equation (2) based on data directly derived from the hydraulic model.
5. RESULTS
If the probability distribution

riverside (flood)
functions of the required riprap
diameter Dn are known, the final
design values for the diameter in
each section can easily be
determined by looking for that
diameter, which has a frequency of
exceedance that is equai to the
demanded probability of failure.
1 0 ' 1 0 " 10"* 10"' 10" 10"7 1 0 * 10~9
frequency of exceedance
In figure 8 the calculated frequency
exceedance curves for the required Figure 8
diameter at various distances x are shown for the riverside only. These resuits show
that immediately downstream of the barrier the required diameter is about 1.20
metres. Also it shows, that the required diameter will decrease rapidly with increasing
distance from the barrier.
In figure 9 both the theoreticai and

riverside (flood)
1.4 the practical design of the bed
3 - 6 tons
1.2 Theoreticai design protection at the riverside are shown.
1.0 1 - 3 tons Practical design This practical design is based on
\ 0.8 available riprap, various loads during
O 300- lOOOkgs
S 0.6
V construction phases, filter rules and
60 - 300 kgs
Jo.4 minimum section width, construction
10-60kgs
0.2 method.
H 1 1 f-
40 80 120 160 200
distance x (m)
Figure 9
6. CONCLUSIONS
The calculation of various hydraulic loads on a bed protection within a hydraulic

model is a relatively small task. The results of such an addition prove to be very
useful if the hydraulic model is used for an overall probabilistic design. In the case
of this project similar additions to a hydraulic model have been made, like the
development of a scour hole and the hydraulic forces due to head, velocities and
waves.
If model tests are carried out to investigate the stability of riprap, it is essential to
acquire reliable velocity profiles. These prove to be necessary if a relation between
the hydraulic loads and the strength of the bed protection has to be derived. Such
a relation will contribute to the development of stability models and will reduce the
number of mode! tests to be carried out.
7. REFERENCES
[1] Design rules for the use of riprap in closure works, A. Franken, R.E. Jorissen
and H.E. Klatter, to be presented at The International Riprap Workshop, Fort
Collins, Colorado USA, 12-16 july 1993.
[2] Modelling storm surges and forecast effect in designing the Rotterdam storm
surge barrier, J.P.F.M. Janssen and R.E. Jorissen, workshop STORM'91,
Hamburg, 1991.
[3] Integrating forecast effects and operational behaviour in designing the Rotterdam
storm surge barrier, J.P.F.M. Janssen and R.E. Jorissen, 3rd International
Conference on FLOODS AND FLOOD MANAGEMENT, Florence, Italy, 24-26
November 1992.
J
COMMISSION INTERNATIONALE
DES GRANDS BARRAGES
Dix-septième Congres
des Grands Barrages
Vienne. 1991
ROCKFILL DESIGN CRITERIA FOR OVERFLOW DAMS (*)
K. W. PILARCZYK
R. & D. Manager, Rijkswaterstaat
Road and Hydraulic Engineering Division, Delft
H. J. VERHEY
Project Manager
G. J. AKKERMAN
Head of Decision Support Section
Delft Hydraulica, Delft
THE NETHERLANDS
1. INTRODUCTION
1.1. GENERAL
Rockfill is one of the most widely used materials for protection works
and dam construction. Extensive research into the surface layer stability of
rockfill in slope protection, closure dams and overflow dams have been
carried out over the world in the past decades. Most studies, however, relate
to specific structures and flow situations. Moreover, there are differences
in research approach, model facilities, scales used and the way of presen-
tation of the results. Due to all these factors a comprehensive, more
systematic insight into stability aspects of rockfïlled structures is still
absent.
(*) Critères de projet relatifs aux enrochements des barrages déversants.
465
Q.67-R.28
In this paper an attempt is made towards a unifïcation of stability

approaches for rock in various applications, with emphasis on closure
dams. A more consistent picture for various rockfill applications under
different hydraulic conditions, including flow and wave attack, has been
made by introducing new stability parameters, and processing the available
data from former studies.
1.2. DESIGN PHILOSOPHY
TT aiî-iwicuiiiiig a u uCiuica a i c u€sigxiCu LU n . w p uit» rr<iiv.i UJ huw I/HJIU

and away from inhabitated areas. The long tradition of dike construction
in the Netherlands, as well as the history of dam disasters shows that full
safety is unattainable. Because of that, a method to assess the probability
of failure of a system of water-retaining structures has been developed [15].
Good engineering practice requires that all possible modes of failure
of the structure under design should be considered. This is a common
approach in the design of concrete or steel structures. In the design of
water-retaining structures, such as dams and dikes in the Netherlands, this
approach is gaining ground, especially in combination with probabilistic
reasoning. This is one of the spin-offs of the design of the Storm-Surge
Barrier in the Eastern Scheldt.
A brief outline of the failure mechanisms of dikes or dams is given in
Fig. 1. Within the scope of the structural design all possible structural
failure mechanisms of the structure and all other possible exteraal causes
of failure (human errors, management error, equipment defect, and so on)
have to be determined. Then the relationship between the failure mecha-
nisms and the external causes on the one hand and the ultimate conse-
quence of a flood or complete drawdown on the other hand has to be
analyzed. The fault tree is a very helpful tooi for solving this problem.
The probability of failure of the various mechanisms has to be
determined by means of probabilistic calculations [13]. A full descriptïon
of the probabilistic approach for dam and dike design goes far beyond the
scope of this paper. However, it is important for the designer to take the
current knowledge on this approach into account so as to be sure that the
whole structural design is treated in a consistent way. On the other hand
the design philosophy should evolve from a purely technical level to a more
integrated approach that takes into account social, environmental and
economical considerations.
As can be seen in Fig. 1, a dike or closure dam may fail because of
various phenomena. From these mechanisms only Types 1 to 3 and 7
(" external " damage) are discussed in this paper. In the next sections
466
Q. 67-R. 28
- 'MfKJuajaxixs
W?5?SWWrWWS353!5'S'^S?5S'S?
Sa
8b
OTWSSSSTSEWOTRWWSTOCWrasZS?
Fig. 1
Failure mechanisms and causes

Modes de rupture et causes
(i) Overflow-crest (1) Dérersement - crête

(2) Overflow-inner slope (2) Dérersement - talus aval
(3) Seepage-inner slope (3) Percola1ion - talus aval
(4) Partial sliding (4) Glissement partiel
(5) Total sliding (5) Glissement total
(6) Settiement (6) Tassement
(7) Wave attack (7) Attaque par la hoitle
(8 a]1 Scour hole (8 al Fosse d'érosion
(8 b) Bottom protection instability (8 b) Instabilité de la protection de fond
(9) Liquefaction (9) Liquefaction
damage behaviour and/or damage margin after exceedance of the thres-

hold conditions of stability will be discussed. Types 4, 5 and 6 are typical
soil mechanical stability problems and are beyond the scope of this study.
The Type 7 failure mechanism is similar to the problem of seawall or
breakwater stability. Types 8 a, 8 b and 9 prevent the adjacent cohesionless
bed aga'inst erosion: the design of the corresponding bed protection top
layer is dealt with briefly. For more detailed information on scour and
scour protection the reader should refer to [10]. When the subsoil has a
reasonable bearing capacity Types 4 to 6 will not be as important as Types 1
to 3 and 8 and 9. However, soil mechanical properties (investigation) will
influence the design when compact profiles are built up (with steep slopes)
or when the subsoil is relatively soft.
467
Q.67-R.28
2. DESIGN CRITERIA FOR CLOSURE DAMS
2.1. GENERAL
River and tidal basin closures have been carried out around the world
for many centuries, mostly at a limited scale while mainly based on
experience and trial-and-error. However, the increasing land reclamation,
flood protection and freshwater reservoir requirements at the beginning of
this century stimulated an enlargement in scale of the closure works. This
was made possible because of the improved understanding of tidal hy-
drodynamics and the development of large seif-propelled equipment.
During closure the dumping of large stones, concrete cubes or other
flow-resistent elements into the gap reduces ïts cross-section. At first, the
total flow of water is hardly reduced; consequently the flow velocity
increases more or less proportionally with the decrease in cross-section,
thus requiring larger units of material in the later stages of closure. In
addition, the stability of the adjacent seabed is endangered (provided it
consists of erodable material) and there is a need for bottom protection so
as to ensure a stable foundation for the closure dam. After the dam crest
has emerged the core can be filled up with finer materials, e.g. sand or
gravel, to reduce the permeability. Finally, covering layers are applied to
seal the slopes of the dam and to provide protection against wave attack
and wave overtopping.
Rivers or tidal gaps can be closed by constricting the waterway

vertically, horizontally, or a combination of both. The dam types tested at
Delft Hydraulics and/or mentioned in the literature are summarized in
Fig- 2 [I], [2], [7], [8]. The definition or classification of dams is shown in
Table 1. Design criteria for the vertical method (or frontal dumping or
horizontal tipping method) as far as they are related to the overflow dams,
will be discussed in the following sections. A brief review of the available
literature is given; a substantial part of this literature is the result of model
investigations into the various older closure schemes of the Delta Project
in the Netherlands. The results presented are also applicable to overflow
dams.
More consistent design criteria were obtained by introducing new
stability parameters. In addition, the result of the analysis of model
investigations recently carried out by Delft Hydraulics (commissioned by
the Rijkswaterstaat), will be summarized. They provide a better physical
basis for the proposed design criteria.
More detailed information can be found in reference [3]. The design
criteria as presented enable an easy indicative assessment of the stone
468
Q. 67-R.28
«MJWJJMJUtf
^ss? •xywxyWxyAyKyK/xy>i>
Typical dam types

Coupes types de barrages
.469
Q.67-R.28
stability at various closures stages; in particular, critical situations of the

closure operation can be identified.
2.2. STABILITY RELATIONSHIPS FOR VERTICAL CLOSURES

DURING CONSTRUCTION
2.2.1. General
In this section the recently developed design criteria will be discussed
for the identified flow regimes during a closure. Before that, the relevant
parameters will be presented, as well as the flow regimes. Finally, some
remarks will be made about indicative and detailed designs and the damage
margin.
Parameter definitions
Parameter definitions (Fig. 3) have been elaborated for the outline
design of rockfill closure dams. These provide practical stability criteria for
a Iarge variety of dam types, Fig. 4. It should be noted that Fig. 4 is based
on all the available data from tests carried out at Delft Hydraulics, while
Table 1 in combination with Fig. 5 also includes information from litera-
ture. Therefore, Table 1 and Fig. 5 are most useful for indicative designs.
Fig. 3
Definition sketch
Définition des paramètres
(B) Crest width (B) Largeur de crête
(D) Nominal stone diameter (D) Diamètre nominal de l'enrochement
(H) Upstream water depth relative to the (H) Hauteur d 'eau amoni au-dessus de la crête
dam crest du barrage
(d) Dam height (d) Hauteur du barrage
(h) Downstream water depth (h) Hauteur d'eau aval
(hh) Tailwater elevation relative to the dam (hh) Hauteur d'eau aval au-dessus de la crête
crest du barrage
(q) Total unit discharge (q) Débit total unitaire
(a) Inner slope angie (aj Angle du talus
(A) Relative density (A) Densitê relative
470
»
a A
<i r£
fs
i 6 3
%
ir*"
8f
•'Sf
•f
"3
HO^«
».3**
j&^
'
*«
%
•9
'h 4 •» 1 *
-1 \ 1 i < t < l > ii i • i
-*. Nj/ilD
2
1 t
ffl
•f
3" ! 1 1 J
'I*
9
1 !2
'f' 1
er
. .*
*
/
/ *
•
W
e»?
V
d
f*
f. 1
°$ !
3
» %"V N,
X
5' t
*£ ff , • l I 1 < • i >«
• K,/AO
Fig. 4
Stability plot overall results rockfill closure dams

Ensemble des résultats de stabilité pour les barrages en enrochement
(A) Overtopping height parameter (A) Paramètre de hauteur de submersion
(1) Mean value with confidence Iimits (1) Valeur moyenne avec limires de confiance
H/AD - h„/AD + 2 (—1 < h„/AD < 10)
H/AD - I (— 3 < h„/AD < - 1)
(B) Discharge parameter (B) Paramètre de débil

(2) Mean value with confidence Iimits (2) Valeur moyenne avec limites de confiance
q/(g"5 [AD]") = 0.55 (h„/AD + 2 ) " (h„/AD > - 1)
Note : q = 1.7 H " for translation A into B Remarque : q = 1.7 H ' pour transfonper A
en B
Q.67-R.28
The independent parameter is the tailwater parameter, hb/AD

[hb = tailwater eievation relative to the dam crest (instead of the tailwater
depth, up to now commonly used), D = nominal stone diameter (size of
a cube, D = (M50/ps)033 with M50 = mean stone mass, ps = density of
stone), A = relative density of stone].
Basically, two stability parameters have been introduced :
• the overtopping height parameter H/AD (H = upstream water
depth relative to the dam crest, " overtopping height "); .
• the discharge parameter q/(g05(AD)15) (q = total unit discharge,
g = gravitational constant).
Flow tegimes
Four typical flow regimes have been identified, which depend on
hb/AD (see Table 1 and Fig. 5) :
Fig. 5
Flow regimes
Régimes d'écoulement
(A) Low dam flow : (A) Écoulemeni sur barrage -faible chute :
(h,/AD > 4) (hÂD > 4)
(B) Intermediate flow : (B) Écoulement intermediaire:
(— 1 < hs/AD < 4) (- 1 < h*/AD < 4)
(C) High dam flow : (C) Écoulement sur barrage - chute élevée .
(hb/AD < — 1 and H > 0) (h*/AD < — landH> 0)
(D) Through flow : (D) Écoulement a travers:
(H<0) (H<0)
All
Q.67-R.28
• low dam flow (hb/AD > 4) : drowned flow, no influence of

porosity;
• intermediate flow (— 1 < h b /AD < 4) : free flow, flow penetration
into the porous crest;
• high dam flow (hb/AD > — 1 and H > 0) : submergence of down-
stream crest line, rough shute flow on the inner slope;
• through flow (H < 0) : the full discharge passes through the dam
body, outflow on the inner slope.
Table 1
Review of indicative design criteria for threshold condition
Résumé des crirères de calcul pour la condition d'entrainemeni
Flow regime H-criterion q-criterion
low dan- low mean fit through data. Fig. 4

sharp crest : H ~ h b - 2 - 3 at
AD
increasing ht/AD
narrow/broad crest : 1.5 — 2 Shields :.k. u/(gAD) 0S - C-iy/Sp*
round crest :2
very broad crest :2 — 3 (k. - 1. V - 0.04. h b in C)
intermediate mean fit through data. Fig. 4 mean fit through data, Fig. 4
flow
high dam flow raean fit through data. Fig. 4 mean fit through data. Fig. 4
(note the dominating influence Knauss (tan a - 0.33 - 0.5) :
of porosity D/d) ,.H . - 1 18 + 0 5 1» l 87 tin fi
conversion of q into H : g 0 i (AD)' 5
q - m-I.7 H u withm provisional design curve for
from [11 e.g. Knauss [f>\ U n a - 1:2 to 1:12
JL » i ^ » (i.49 - 1.87 sin a ) 0 6 7
AD m°*
through flow Prajapati (adapted to H) : Prajapati (tan ot — 0.8

H/AD - 2.78 + 0.7ï hb/AD and D/d - 0.02 - 0.051 :
"
g0S(AD)lS
«/hr2
• i M\AD^
2.2.2. Design criteria for flow regimes
In the following the design criteria for the succeeding flow regimes
during construction of a closure dam will be discussed.
/•ÖVV dam flow

• In the low dam flow situation, a uniform flow approach, e.g. Shields
with \y = 0.04, fits the mean data well, so both Shields and the mean data
curve (Fig. 4) can be taken for rough design. Because of the relationship
between H and hb in drowned flow conditions the use of H/AD is not
recommended. Instead, one should use (H — hb)/AD, which appears to be
more or less constant at varying hb/AD (Table 1).
473
Q.67-R.28
Intermediate flow
After the free flow situation is reached, i.e. the intermediate flow
situation, raising the dam will lead to an increasing flow attack on the
downstream part of the crest and the inner slope, even though the discharge
does not increase significantly. This is caused by flow penetration into the
porous crest, causing an increase of the local flow velocity up to values in
excess of the critical velocity at the onset of free flow.
Both the H and the q parameters can be used for stability prediction
(Fig. 4); from this figure the decreasing stability in this flow range
(— 1 < hb/AD < 4) as the dam is raised (equivalent to lowering the
tailwater elevation) is obvious.
Note that the Isbash criterion [4], for well-embedded stones (for broad
crested dams only), is highly dangerous at low tailwater elevations (i.e. at
about crest level or lower) when the theoretical critical velocity (0.67 gH05)
is substituted. In contrast, assuming the velocity is equal to the theoretical
discharge 1.7 H15 divided by the actual tailwater elevation hb, this will
compensate, to some extent, for the underestimation of the actual flow
velocity, provided a water-depth correction equal to D is added to account
for the flow penetration.
High dam flow

After the emergence of the downstream crest height, the porosity of
the rockfilled dam is still such that there is positive overtopping height :
high dam flow situation. In this situition a characteristic flow velocity at
the inner slope cannot be clearly defined, bccause of the extremely rough,
aerated type of flow, comparable to rough shute flow on rockfilled spill-
ways and upper river reaches.
The stability of the inner slope, with a potential damage region near
the intersection with the tailwater level, turns out to be described fairly well
by the Knauss relationship [6] for steep chute flow, provided the total
discharge (over and through the dam) is taken (Table 1 and Fig. 5). The
corresponding slöpe angle is in the range 1:2 to 1:3. At steeper slope angles
the Knauss relationship seems to be a too conservative approach.
The q-criteria mentioned above can be transferred into H-criteria using
q = m 1.7 H15 (m = free flow discharge coëfficiënt, Table 1).
The assessment of the discharge characteristics is important in this
high-dam flow region, because of the dominant porosity influence (D/d,
d = dam height); a simple computational procedure, as outlined in [1],
often yields too many deviations to be practically useful. It is envisaged that
discharge measurements in a scale model will be needed for the typical dam
type under design.
474
Q.67-R.28
Through flow
Normally, the through flow situation (H < 0) will be stable, if the
inner slope is not too steep, because of the highly reduced discharge (no
overtopping). For a dam with a very steep slope, e.g. nearly at an angle of
repose » 1:1.25, a stability criterion has been obtained from the experimen-
tal results of Prajapati [12], Table 1. Note that in this case the actual
tailwater depth h appears and not hb. Conversion of this into a H-criterion
leads to the expression shcwn in Table 1. These criteria are valid for
D/d = 0.02 to 0.05, thus for dams of relatively fine materials.
To conclude this section, it should be noted that during all flow
regimes the structure can be attacked by wind waves. This additional wave
attack (magnitude of wave amplitude smaller than overtopping height) can
roughly be taken into account by adding one third of-the significant wave
heigHt to the overtopping height; for the stability analysis the resulting
equivalent overtopping height Heq ( = H + 0.33 Hs) can be considered as
the effective overtopping height.
In addition, one should note that in the case of a multi-crested dam
layout the lower crest stability can be appraised in a way analogeous to the
stability of the highest crest. This can be done by referring the tailwater
elevation to the lower crest height under consideration and the overtopping
height to the highest crest. This approach is discussed in [1].
2.2.3. Detailed design and damage margin

For a detailed design two approaches are recommended :
• further analysis of relevant data;
• additional model tests focussed on the most critical building stages
and flow situations.
Both approaches enable the optimization of the structure before the
construction of the closure work really starts, as well as during the construc-
tion itself (monitoring). Then, the influence of, for instance, small devia-
tions in the dam geometry can be studied.
From the designer's point of view it may be important to assess the
margin between the threshold condition and a total or partial collapse of
the dam -structure. During investigations at Delft Hydraulics a lot of effort
was put into establishing the margin between threshold damage (about
1 stone/m dam length) and " extensive " damage (some 10 stones/m dam
length) for the actual construction stage of the dams under investigation.
The corresponding damage margin proved to be strongly dam-configura-
tion dependent. Fig. 6 shows this dependency clearly. Data on the actual
475
Q.67-R.28
collapse situations are relatively scarce and tend to be indicative only. Some
quantitative data can be found in [1].
Special emphasis must be given to the critical overtopping height
concept for monitoring the dam stability during closure. The advantage of
the overtopping height method is that only two water levels, one at each
side of the dam, have to be known. For computation of the two water levels
only general data on the discharge characteristics are forwarded. In
addition, for monitoring the stability during the closure operation the water
levels can easily be measured. In contrast, the discharge criterion requires
detailed knowledge of the discharge distribution over the entire dam
alignment. In practice this means that the critical overtopping height
method (depending on the tailwater level only at a certain closure stage)
is a very suitable tooi for controlling the closure operation.
2.3. PROTECTION AGAINST WAVE ATTACK INCL. OVERTOPPING
In the final stage of a closure work the top layers of the crest and the
slopes of the dam have to be protected against wave attack and overtop-
ping. A general empirical formula for slope protection systems is [11] :
_I
r AmD > vü' (p (cos er') H, g for cot cc > 2
with : Cz = tan et (H s /Lo) -a5 = 1.25 T r H s _ a ï - tan a (Cz = breaker similarity)
in which : \yu = upgrading factor (i)/u = 1.0 for riprap as a reference and
vyu > 1 for other revetment systems), cp = stability factor or stability
function for incipient of motion defined at |z = 1, Hs = significant wave
height, Tz = average wave period, Lo = wave length; Lo = gT*/2 n,
D = specific size or thickness of protection unit, a = slope angle, Am = re-
lative density of a system-unit, b = exponent related to the interaction
process between waves and revetment type (roughness, porosity/permeabi-
lity, and so on; 1 < b < 0.5; for rough and permeable revetments as riprap
b = 0.5, for smooth and less permeable placed-block revetments it can be
close to b = 1. The value b = 2/3 can be treated as a representative
common value for other systems (i.e. more open blocks and block-mats,
mattresses of special design, and so on).
D and Am are defined for specific systems as :
• rock : D = D„ = (M5o/ps),/3, D„ = 0.85 D50, Am = A = (ps - pw)/p„
• blocks : D = thickness of block and Am = A
• mattresses : D = d = average thickness of mattress and Am
= (1-n) A, (n = bulk-porosity of fül material, A = density of fill material;
for common quarry stone (1-n) A » 1, for stone mattresses min.
d = 1.8 Dn).
476
Q.67-R.28
too
5.0]
2.5
ft, "fijl-ttf*
-25 O 23 5.0
/ I
-*. - 2 .
AD
-25
Fig. 6
Influence of crest width for a trapezoidal dam

Injluence de la largeur de crête pour un barrage trapezoidal
(1) Mean curve broad-crested : (1) Courbe movenne pour une crête large :
(B/d » 1) (B/d - 1)'
x D/d = 1/15, A = 1.7 x £ W = 7 / 7 5 . 4 - 7.7
+ D/d = 1/15, A = 2.1 + D/</ = 7 / 7 5 . 4 = Z7
• D/d = I/10,A - 1.7 • D/d~ 1/10. A = 1.7
(2) Mean curve narrow-crested (2) Courbe movenne pour une crête élroite :
(B/d - 0.33) (B/d - 0.33)
• D/d - 1/15, A - 1.7 • D/d= 1/15. 4 = 1.7
o D / d - 1/15,A - 2 . 1 O 0 / t f - 7 / 7 5 , 4 = 2.7
O D/d - 1/10, A - 1.7 o D/d= 1/10. A =1.7
(3) Mean curve sharfHcrested : (3) Courbe movenne pour une crête aiguè :
(B/d - 1) (B/d « 1)'
D D/d - I/1I,A - 1.7 a D/d= 1/11. A - 7.7
Note : tan a = 0.5 Remarque : tang. a = 0,5
For rock and |* > 3, the sizes calculated at §* = 3 can still be applied.
<p == 2.25 for incipient motion of rock on less permeable core as sand
and clay (revetments).
<p = 3.0 for maximum tolerable damage for 2-layer system on granular
filter (damage-depth less or equal to 2 DJ and less permeable core.
477
Q. 67-R. 28
tp « 3.0 for incipient motion; high permeable core/breakwaters,

closure dams).
If the structure is overtopped, even by minor splash, then its stability
can be affected. Overtopping can : (a) erode the area above or behind the
revetment, negating the structure's purpose; (b) remove soil supporting the
top of the revetment, leading to the unraveling of the structure from the top
down; and (c) increase the volume of water in the soil beneath the structure,
contributing to drainage problems. The effects of overtopping can be
limited by choosing a higher crest-level or by amouring the bank above or
behind the revetment with a splash-apron. For a small amount of over-
topping a grass-mat on clay can be adequate. The splash apron can be a
filter blanket covered by a bedding layer and, if necessary to prevent scour
by spiSSn, nprap or pavernent of concrete units or aspuait.
There are no generally valid recommendations for acceptable levels of
overtopping for seawalls and/or dikes and dams. In Standard practice in
the Netherlands a safe value 0.002 m3/s for grassed crest and rear slope is
recommended per unit dike length. Recent experience provides an indi-
cation that this value can be increased to 0.005 or even to 0.01 m3/s for
" good " quality grassmat on clay-sublayer. The information on a proper
Clay specification for a grassmat can be found in guidelines [14].
No definite method for designing against overtopping is known; this
is due to the lack of a proper method on estimating the hydraulic loading.
"erwt
J. ;
Fig. 7
Definition of splash area

Définition de la zone de projection
(1) Still water levei (1) Plan d'eau stagnant

(2) Plane slope (2) Pente plane
(3) Reclamation area (3) Zone de colmaiage
(4) Dike/dam (4) Digue/barrage
478
Q. 67-R. 28
Pilarczyk [l 1] proposes the foUowing indicative way for the design of the
splash area (Fig. 7) :
The critical wave height follows from :
H / A D n - 2 V f' § - " cos o, (1 - R./RJ-'
The width of protection U in the splash-area can be roughly assumed
equalto :
U = 0.2vi/ T -T]/g(R n -Rc) > L™
The admissible discharge follows with the Knauss formula as elabor-
ated in [1] :
qadm = 0.625 g 05 (AD„)' 5 (1.9 + 0.8 q>p - 3 sin a ; )
The criterion is :
q»dm ^ q max.eff.ovenopp. — ömax /Uu
B mai = 0.1 (cot ai)'- (Rn — Re)2 for R„ > Re and t ^ < 0.5 T
5
In the above formulae symbols not earlier defined are defined as :

qadm = maximum admissible discharge, et; = inner slope angle, cpp = stone
packing-factor (0.6 for dumped rockfill, 1.1 for placed rock), Re = crest-
height above still water level, Rn = wave run-up on plane slope, T = period
of representative wave, Bmax = volume of water during run-up above a
certain location ( = crest) per unit width, teff = effective time of overtop-
ping, \yT = importancy coëfficiënt (i|/T ^ 1), Lmin = three times the
dimensions of the slope units.
The overview of slope stability criteria against flow attack for various
revetment systems is given by Pilarczyk [11].
3. CASE STUDIES ON RECENT FAILURES OF CLOSURE DAMS
3.1. INTRODUCTION
Two recent prototype failures, which occurred in the Netherlands [I],

are used here to illustrate the applicability of the stability criteria defined
in this paper :
— 20 February, 1981 : failure of the inner slope of an overflow weir
of a storage basin (salt/fresh water separation) for the Krammer Sluices
under construction (the " Hoge Bekken overflow weir "), see Fig. 8.
— 11 March, 1982 : failure of a closure dam under construction (the
" Markiezaatskade "), this closure dam being part of the secondary closure
scheme of the Delta Project, Fig. 9.
479
Q.67-R.28
The failure of iae Hoge Bekken overflow weir occurred at the fïrst high
tide after closure of the adjoining closure gap, during which the storage
basin was filled by the overflow over the weir section. The overflow
imposed such a heavy flow attack that the rockfill top layer at the inner
slope as well as the concrete block pitching at the weir crest suffered heavy
damage and failed consequently. The onset of the damage was observed
at 7.15 p.m. when ticking noises were heard; the failure is assumed to have
taken place after 7.30 p.m. when the tailwater level was at MSL. At 8.10 p.m.
the tailwater level was equal to the upstream level, completing the filling
sequence.
The failure of the Markiezaatskade, a closure dam 800 m long, was
initiated by failure of the southern abutment adjacent to the finished
southern dam section. Of interest in this particular case study is the large
deformation of the closure dam section under construction, north of the
abutment, which failed; this can be characterized as a nearfailure situation
(a small part of the crest remained unaffected). This deformation was
induced by approximately two-dimensional flow so, in this context, this
situation may be looked upon as a typical vertical closure failure. The
failure was caused by a severe storm surge with a maximum water level of
M.S.L. + 3.7 m. The closure dam was in its fïnal stage of construction; the
crest height of the dam section under consideration was at M.S.L. + 2.25 m.
No observations were available of the location where the damage started,
or the time when the stones lost their stability.
3.2. STABILITY ANALYSIS OF THE HOGE BEKKEN OVERFLOW WEIR FAILURE
Because of the simplïcity and convenience of the critical overtopping

height criterion, it will be applied here.
The upstream water level was measured, whereas the tailwater eleva-
tion was assessed by a simple storage basin computation with a starting
level at M.S.L. — 0.8 m (Fig. 8) and an estimated value for the discharge
coëfficiënt m of 1.0.
From the plot of the overall results in Fig. 4 (mean line through all
data) the critical overtopping height H can be directly appraised as a
function of the tailwater elevation, providing an indicative value for
exceedance of the threshold condition of stability. Dividing the instanta-
neous critical value of H for threshold condition by the actual H (= Ha)
gives a safety coëfficiënt F for threshold damage for the original dam
profïle.
In Fig. 8 F is plotted against time, during the hour when failure
occurred, for two mean stone weights (the actual mean stone weight is
480
Slaak t>e«fc«n I H09V BcMivn
*&'
Ftanjan iOLt4St: « m c a r n .
7.00 MS T3Ö Fis a00 "5ÏS
4 4 fa*w« 11 M M I W ra iiiiiiat»
ttc»n«*q n w n i of w »ton««
]
C
"*"^"—»i.^ /
/
1 (r<»
4 > t '»S
0
00 T-. 13 7 30 7 4S * tX I 4: S
S:lS
Fig. 8
Stability analysis of " Hoge Bekken " overflow weir failure

Analyse de la perte de stabilité pour Ie barrage " Hoge Bekken
(A) Section (A) Coupe transversale

(B) Water levels (B) Niveaux d'eau
(C) Stability for low stone mass (C) Stabilité pour enrochements légers
(D) Stability for high stone mass (D) Stabilité pour enrochements lourds
(1) Overall data curve (Fig. 4) (1) Enveloppe de toutes les données (Fig. 4)
(2) Knauss[6] (2) Knauss[6]
(3) Brogdon and Grace [2] (3) Brogdon et Grace [2]
Note : F »» H„,,„/H4C1„i Remarque : F — H,„,myH,„i
F — 1 : threshold damage F = I : condition d'enlrainemenl
Q. 67-R. 28
assumed to be in between these two values). It should be noted that for low
value of hb/AD (e.g. Iower than — 2), the accuracy is greatly reduced
because of the lack of data in Fig. 4; furthermore, the value of F is then
very sensitive for small deviations in H and Ha because of the small absolute
values involved. This is the main reason for the exceptionally low values
of F at low hb/AD. On the other hand, the more reliable application of the
overall results curve close to 8.00 p.m. also shows unsafe values of F, say
0.8 to 0.9, indicating that the threshold condition of motion was exceeded
at that time.
A second method has been tried, applying the Knauss-formula and
substituting H for q according to q =» m 1.7 H! 5. Tne critica! overtopping
height, based on the Knauss formula, is obtained from :
H/AD = m- 067 (2.74 — 3.44 sin a) 067
with : m » 1.0 as an arbitrary estimate for the mean value for non-porous
dams and sin a = 0.24 (slope angle 1:4) follows H/AD = 1.55 or
H = 0.51 m for a mean stone weight of 20 kg and H = 0.61 m for a mean
stone weight of 35 kg.
The curves for the Knauss criterion are also presented in Fig. 8. For
even the lower stone weight there are indications that the damage threshold
was not exceeded. However, because of the very low tailwater level at the
instant of failure the Knauss prediction was expected to converge to the
Delft Hydraulics data when the critical discharge was taken into consider-
ation. The unsafe prediction with Knauss may be explained by the
underestimate of the value of m for the low tailwater depths involved
and by the presence of a smooth crest for the actual weir, consisting of
concrete block pitching. The fïrst cause is clearly indicated in Fig. 8 (very
high range of tailwater level).
A third hindcast is based on the results of the study by Brogdon and
Grace [2], since their access-type overflow embankment highly resembles
the present overflow weir (smooth and broad crest, slope angle 1:4).
However, their data range goes up to hb/AD values of about — 1 and,
therefore, does not relate to the failure shortly after 7.30 p.m.
The Brogdon and Grace stability line indicates a significant loss of
stability and coincides fairly well with the overall results stability curve
shown in Fig. 4. Taking into account the smooth crest at the corresponding
porosity (D/d), this would have lowered the F-curve even more.
3.3. STABILITY ANALYSIS OF THE MARKIEZAATSKADE DAM FAILURE
The critical overtopping height criterion has also been applied in this
case. The water Ievels on both sides of the closure dam were measured by
482
•xromax.n.j. n Q. 67-R. 28
3
J
4 *
-a^^*^
2
I '
>
2
a
t
^_
f 0
—-
t0**a»j£r^=
3:00 330 *00 *-30 500 5:30 *;00

-» Mwxn ft, t992 • urnt a m .
3 c
f, ^
"**-*^ " ~
^^^
\
0
100 J:J0 «00 *-JO 5:00 SM SOO
3
0
A J^-
4 .2.0
n
iOO *J0 *O0 *:J0 S:00 3:30 SS
Fig. 9
Stability analysis of " Markiezaatskade " dam failure

Analyse de la pene de stabiliié pour Ie barrage ' Markiezaatskade
fA) Section (A)Coupe transversale
(B) Water levels (B)Nireaux d'eau
(C) Stability without waves (C)Stabiliié saus houle
(D) Stability with waves (D)Stabiliié avec houle
(1) Overall data curve (Fig. 4) (1)Enveloppe de toutes les données (Fig. 4)
(2) Knauss[6] (2)KnaussfóJ
(3) Detail curve broad-crested dam (Fig. 6) (3)Détail de la courbe pour des barrages de
crëte large (Fig. 6)
Note : F - H„,,«,/H«,ral Remarque : F'«= H,„,„„/Hî
F — 1 : threshold damage F= 1 :condition d'entrainement
Q.67-R.28
selfrecording staff gauges and are shown in Fig. 9. The initially damaged
portion of the dam consisted of rockfill 300-1 000 kg (M» = 615 kg and
A = 2.0), see Fig. 8. During the storm surge the significant wave height was
about 1 m. The influence of this wave height on the rockfill stability can
be roughly taken into account by adding one third of the significant wave
height to the overtopping height.
The overall results stability curve shown in Fig. 4 is applied. Com-
putations (Fig. 9) yield a minimum safety coëfficiënt of about 1.0 without
wave influence and about 0.8 with wave influence for the original dam
profile. So, taking into account the wave influence, the threshold condition
for damage is distinctly exceeded.
Because of the steep slope angle (1:1.5) of the original dam profile the
prediction, based on the overall results for a slope angle of 1:2, may be
somewhat overoptimistic. On the other hand, after some deformation the
slope would be flattened whereas a substantial part of the crest still
remained unaffected. Therefore, the prediction, based on the overall
results, may be suitable for the situation after some initial deformation had
occurred.
The Knauss formula was reconsidered and adapted to the critical
overtopping height and with insertion of m = 1 and sin a =• 0.55, resulting
in H/AD = 0.85 giving H - 1.00 m for 300-1 000 kg stones used (A =• 2.0,
D = 0.59 m). Because the tailwater depth is larger than roughly half of the
dam height, the Knauss prediction may be considered as somewhat conser-
vative. A minimum value of F of 0.7 is found without waves and a value
somewhat below 0.6 with waves. In contrast to the overflow weir crest, the
closure dam crest is fully rough and the prediction will not be overopti-
mistic from this point of view.
Apart from the steep slope angle in the initial situation, the present
dam profile resembles a broad-crested dam type with uniform porosity
(D/d ~ 1/10) as presented in Fig. 6. The application of this stability curve
is considered to provide the most reiiable prediction. The minimum F
nearly coincides with the value found with the overall results curve (note
that in the latter, data with slope angles 1:4 have also been incorporated),
resulting in a value of about 1.0 without waves and 0.8 with waves. From
this it can be concluded that, even after some flattening of the inner slope
to an angle of about 1:2, the situation was not stable, and further flattening
was unavoidable, causing the subsequent erosion of the observed dam
profile.
Computations, as illustrated in the above, although somewhat indi-
cative, provide the construction manager as well as the design engineer a
quick insight into critical situations which the closure dam will have to
withstand. Based on this insight detailed model tests, focusing on the
484
Q.67-R.28
critical situations or constructive measures, can be initiated, thus reducing

the chance of failure.
From the designer's point of view it is difficult to maintain a reliable
safety margin due to the inaccuracies, uncertainties and stochastic scatter
involved in the stability analysis. Therefore, recently a probabilistic design
approach has been developed. The present design criteria can be used in
such an approach.
4. CONCLUSIONS AND RECOMMENDATIONS
The information contained in this present paper may help in the design
of rockfill structures and especially closure dams with a compact profile,
e.g. with stable rockfill material. However, the design criteria are generally
indicative only and should not be applied for " limit design **, unless the
geometry of the closure dam under consideration closely resembles one of
the dam types which has already been studied.
The design of rockfill structures with transportable material like gravel
and sand are beyond the scope of this paper. This interesting field of
application, though widely used in river closure design (see for instance
[4], is rarely practised for estuarine closures (only sand closures are widely
applied, see [10]. The use of closure dams of non-compact profile may be
very interesting for regions with a potential lack of stable closure material.
This paper is mainly based on experimental studies. To date there are
many uncertainties in a theoretical approach. This does not alter the fact
that a rigorous development of the involved physics, must be pursued since
this will lead to the best results ultimately.
A number of recommendations, focused on the lack of knowledge
which was experienced during this study, are summarized below :
a. Further study on two-dimensional discharge characteristics, espe-
cially in the intermediate and high dam flow range. It is anticipated that
sufficiënt iiterature is available for the theoretical assessment of the
discharge in the through-flow situation.
b. Three-dimensional flow effects, viz. flow-contraction phenomena
at abutments etc, have rarely been studied (as far as could be traced) in
the case of combined vertical/horizontal constrictions. It is stressed that
this gap should be filled; an illustrative approach is the investigation of
Kindsvater and Carter, for discharge characteristics with horizontal
constrictions [5].
c. In addition, it is recommended that the stability aspects for com-
485
Q.Ó7-R.28
bined vertical/horizontal closures, being an extension of the horizontal

closure experiments by Naylor [9], should be studied.
d. Further study on the stability of through-flow dams, including
failure mechanisms.
e. Assessment of the suitability of present design criteria by :
Phase 1 : deterministic accuracy analysis (safety factor analysis).
Phase 2 : further development of the probabilistic computational ap-
proach, with emphasis on the failure mechanism.
f. The wave-overtopping criteria should be further studied and veri-
fied especially concerning the definition of a representative discharge as
well as a further specifïcation of the coefficients in Pilarczyk's formula for
wave attack.
The limited scope of this paper does not include a full evaluation of
all available Dutch data on dam-and-dike design and construction. The
problem is too broad and complicated. The details and/or background data
can be found in the Guidelines reports as mentioned in the references.
Nevertheless, this brief evaluation is expected to be useful for design and
construction engineers involved with design and monitoring of closure
dams.
ACKNOWLEDGEMENTS
Thanks are due to the Government Department of Transport and

Public Works, of the Netherlands (Rijkswaterstaat), which commissioned
these studies.
REFERENCES
[1] Akkerman G. J., Konter J. L. M. (1985), Hydraulic design criteria for

rockfill closure of tidal gaps, vertical closure méïhod. Delft Hydrau-
lics, report Ml741 part IV, Delft, The Netherlands.
[2] Brogdon, N. J. and Grace J. L. (1964), Stability of riprap and discharge
characteristics of overflow embankments. U.S. Army of Eng. W.E.S.
Vicksburg, Tech. Rep. 2-6500.
[3] CUR (1991), Handbook on use of rock in hydraulic engineering. Centre
for Civil Engineering Research and Codes and Specifications (CUR),
P.O. Box 420, Gouda, The Netherlands.
486
Q.67-R.28
[4] Isbash S. V., Khaldre K. Y. (1970), Hydraulics of river channel clo-

sures. Butterworths, London.
[5] Kindsvater C. E., Carter R. W. (1954), Tranquil flow through open
channel constrictions. ASCE, Hydr. Div., Vol. 80, Separate no. 467.
[6] Knauss J. (1979), Computation of maximum discharge at overflow
rockfill dams, 13th ICOLD Congress, New Delhi, Q. 50, R. 9.
[7] Linford A., Saunders D. H. (1967), A hydraulic investigation of
through and overflow rockfill dam, B.H.R.A., RR 888.
[8] Lysne D. K., Tvinnereim K. (1971), Scour protection for submerged
rockfill sills. River and Harbour Auth., Norway, Buil. 13E.
[9] Naylor A. H. (1976), A method for calculating the size of stone needed
for closing end-tipped rubble banks in rivers, CIRIA report 60,
London.
[10] Huis in't Veld J.C. (ed.) (1984), The Closure of Tidal Basins, Delft
University Press, The Netherlands.
[11] Pilarczyk K. W. (1990), Design of seawalls and dikes incl. revetments.
In " Coastal Protection ", Balkema Publishers, Rotterdam, The
Netherlands.
[12] Prajapati J. J. (1981), Model studies on through-flow rockfill structu-
res, XIX IAHR Congress, New Delhi, Subject D, paper 12.
[13] TAW (1987), Probabilistic design of sea defences. Dutch Technical
Advisory Committee on Water Defences (TAW), published by CUR
(1990), Gouda, The Netherlands.
[14] TAW (1990), Guide for design of river dikes. Published by CUR,
Gouda, The Netherlands.
[15] Vrijling J. K. (1985), Probabilistic Design of Water retaining Structu-
res. In : Proceedings of the NATO Advanced Study Institute, Confe-
rence on the Engineering Reliability and Risk in Water Resources,
Tucson, Arizona, USA.
SUMMARY
Stability aspects of rockfill applied in various structural components

of closure dams are summarized on the basis of a Iiterature evaluation and
investigations by Rijkswaterstaat and Delft Hydraulics. Useful design
criteria are presented, covering a wide variety of dam geometries, of
487
Q.67-R.28
permeability effects, of flow and wave attack, from the initial stage up to
the final stage of closure. The applicability of the criteria is illustrated by
analysis of two prototype failures in the Netherlands.
RÉSUMÉ
Les aspects de stabilité des enrochements utilïsés dans diverses zones

d'un barrage ont été résumés a partir d'informations fournies par la
littérature et de recherches récentes effectuées par Rijkswaterstaat et Delft
Hydrauiics. Les critères présentés s'appiiquent aux barrages a divers stades
de construction et couvrent une grande variété de géométries. Ils tiennent
compte aussi de 1'effet de la perméabilité et de Fattaque par Ie courant et
la houle. L'application de ces critères est illustrée par 1'analyse de deux
ruptures qui se sont produites sur des prototypes aux Pays-Bas.
488
PART 3
Bank protection
Pilarczyk, K.W. , H. Havinga, G.J. Klassen, H.J. Verheij,

E. Mosselman and J.A.A.M. Leemans,
Control of bank erosion in the Netherlands. State-of - the-art.
Boeters R.E.A.M., F.C.M, van der Knaap,

Behaviour of armour layers of riprap bank protections along
navigation channels
Boeters, R.E.A.M., H.J. Verheij and M. van der Wal,

Environment- friendly bank protections
Coops, H., R.E.A.M. Boeters and H. Smit,

Direct and indirect effects of wave attack on helophytes
Ivens, E.A.M.,
Penetration through geotextiles by shoots, roots and rhizomes of
the cornmon reed (phragmites australis (cav.) trin. ex steucüel) :
a pilot study
publication no. 442
Control of bank erosion in the

Netherlands.
State-of-the-art
Extended abstract presented at the 3rd National

Conference on Hydraulic Engineering of the ASCE,
New Orleans, U.S.A., August 14-18, 1989
K.W. Pilarczyk, H. Havinga, G.J. Klaassen,

H.J. Verhey, E. Mosselman, and J.A.A.M. Leemans
December 1990
delft hydraulics
CONTROL OF BANK EROSION IN THE NETHERLANDS. STATE-OF-THE-ART
ABSTRACT
INTRODUCTION
MANAGEMENT, ORGANIZATION AND GENERAL PHILOSOPHY
BANK EROSION DUE TO RIVERINE PROCESSES

General
Erosion problems
Mechanics of river-bank erosion
Meandering rivers
Braided rivers
BANK EROSION DUE TO NAVIGATION
General
Flow patterns in groyne fields
Erosion and sedimentation in groyne fields
Conclusions
ENVIRONMENTAL ASPECTS OF RIVER BANK EROSION

General
Abiotic diversity
Biotic diversity
Protection and restoration
Conclusions
INTEGRATED RESEARCH ON REVETMENTS

General
Research aspects and research means
General stability approach (SOWAS-concept)
Overview of practical design formulae
CHOICE OF MEASURES
SHORT OVERVIEW OF CURRENT RESEARCH
CONCLUSIONS AND RECOMMENDATIONS
REFERENCES
CONTROL OF BANK EROSION IN THE NETHERLANDS. STATE-OF-THE-ART
By K.W. Pilarezyk1, H. Havinga2, G.J. Klaassen3, H.J. Verhey3, E. Mosselman1*

and J.A.A.M. Leemans 5 ,
ABSTRACT: In the Netherlands protection against flooding and inland navigation

have always been extremely important. Many problems related to these items,
among them erosion and the protection of banks and dikes, were and are solved
within the scope of various specific studies supported by experience.
Recently, much more attention is paid to environmental aspects of projects and
applied measures. The increased demand on reliable design methods has resulted
in development of various calculation methods on hydraulic loading, morpholo-
gical changes and erosion prediction, and on the other side, in preparing a
set of design methods and guidelines for protection structures. The develop-
ments in these areas in the Netherlands are discussed in this paper.
INTRODUCTION
The Netherlands is a flat, low-lying country, the western part of which lies
below sea level and has gradually been sinking over the years. The land is
protected against flooding by dikes and kept dry by drainage stations. For
this purpose, a network of canals has been built in addition to the already
existing waters. Both systems are also used for water transport, irrigation,
water supply and recreation. The total length of the navigable waterways is
about 4370 km; rivers 841 km and canals 3529 km. Because of the navigational
requirements the main Dutch waterways have been regulated already in the early
stage by river structures and/or canalized. Only a small percentage of the
waterways and some estuaries remain in a natural state.
Inland navigation has always been extremely important (Figure 1). During the
last decades, the sizes and engine powers of inland motor vessels have
increased, particularly since the introduction of push-tow units. This induced
more attack on banks, more erosion and the necessity of heavier protection
became evident.
1
Rijkswaterstaat, Road and Hydraulics Engineering Division, P.O. Box 5044,
2600 GA Delft, The Netherlands
2 Rijkswaterstaat, Directorate Gelderland, P.O. Box 9070, 6800 ED Arnhem, The
Netherlands
3 Delft Hydraulics, P.O. Box 152, 8300 AD Emmeloord, The Netherlands
*• Technical University Delft, Faculty of Civil Engineering, P.O. Box 5084,
2600 GA Delft, The Netherlands
s Institute for Applied Landscape Ecology, Groesbeekseweg 20, 6524 DB
Nijmegen, The Netherlands.
-2-
A long-term research program on these subjects has been carried out. A variety
of design rules and calculation methods have been derived from small-scale
model tests and were verified by full-scale tests in canals.
There has been increased public discussion on more natural waterway develop-
ment in recent years. The need for that is generally recognized. Evidence of
this trend underlines an initiative by the government (in 1986) to start a
national project on, aptly named, "Environment-sensitive solutions for bank
protection". The aim of this project is to stimulate the use of biological
methods (if possible in combination with other methods) for reinforcing the
waterway banks, and to prepare design methods including a maintenance stra-
tegy.
As a result of the trends mentioned above various studies on the mechanics of
bank erosion and bank protection have been carried out or are still going on.
They include such subjects as morphological changes of natural rivers, influ-
ence of river structures and navigation on river regime and morphology, ship-
induced water movement and its interaction with banks, erosion and protection
of estuary-shores, stability criteria for various protection methods (reed,
cellular blocks allowing vegetation, riprap, blocks, mattresses, asphalt, in-
direct protection by offshore breakwaters/sills, e t c ) .
Previous work is concisely summarized in this paper but major attention is
given to reporting on current investigations.
The state-of-the-art on past re-

search and the general approach
to erosion control of navigation
channels have been published
(Blaauw et el, 1984; Van de Kaa
et al, 1985; PIANC, 1987a), as
well as on the strategy to ero-
sion control of Dutch estuaries
(Leewis, 1983; Pilarczyk, 1986),
dike protection (Pilarczyk, 1987)
and mathematical tools for river
engineering (Schilperoort,
Wijbenga and van der Zwaard,
1985). In all these publications
attention is also paid to envi-
ronment-friendly solutions and
the ehoice of the proper con-
struction.
Regarding the treatment of coas-
tal erosion problems useful re-
ference can be made to the Manual
on Artificial Beach Nourishment
(CUR, 1987).
The mechanics of bank erosion and

the stability of protective
structures (incl. vegetation)
subject to hydraulic loading are 0000000000000000 £V
complex problems. The
III
understanding of erosion
processes and failure mechanisms I of II
of structures is still in a
rudimentary stage, and it is not Figure 1 Main navigable waterways
yet possible to describe many in The Netherlands
-3-
important phenomena and their interactions by theory. While laboratory inves-

tigations and practical experience provide means for gaining further under-
standing of interaction processes, the solution of many practical engineering
problems cannot wait until complete understanding of these processes is ob-
tained. Therefore, existing knowledge on this subject (though limited) should
be systematized and made available for designers and managers in this field.
MANAGEMENT, ORGANIZATION AND GENERAL PHILOSPHY
Management on water defences incl. bank protection is characterized by its

integral nature. Firstly, an integral approach to water defences is required
because of the interrelationship between land use, protection measures and
daily management and control. Secondly, an integral approach is required since
various disciplines and techniques are involved in the analysis of coastal
problems and their potential solutions; for example, coastal engineering,
river engineering, economics, evironmental science, mathematical and physical
modelling techniques, monitoring techniques, etc.
Because of this integral nature of the water defences incl. bank protection,
the function of the management is not only to provide a proper integrated
research program but also to build a bridge between research and practice.
This last point is very important for the implementation of the research
results.
In the Netherlands, Rijkswaterstaat (Public Works Department) plays a very
important role in research into and management of erosion control for shores
and banks. Especially by initiating research, managing funds and evaluating
and implementation of results. Research is carried out by specialized services
of Rijkswaterstaat and by external institutes (i.e. Delft Hydraulics, Delft
Geotechnics, Delft University) mainly financed from government sources.
Because of the integrated nature of water defence and associated erosion pro-
blems and also the involvement of other bodies/authorities (in addition to
Rijkswaterstaat), for the co-ordination and funding of research in specific
fields several task committees have been created. These committees are bud-
getted by the Ministry of Public Works and may commission research to specia-
lised institutes and Rijkswaterstaat.
- The Committee for Applied Research in Hydraulic Engineering (TOW) (actu-
ally, under direct supervision of specialised divisions of Rijkswater-
staat). One of the objectives is to analyse processes and to develop cal-
culation methods for the prediction of coastal behaviour and river mecha-
nics.
- The Technical Advisory Committee for Water Defences (TAW). The studies of
this Committee increase the knowledge on environmental impacts, strength
and construction of several kinds of water defences (river- and sea-dikes).
The final results are presented as guidelines for design.
Example of projects:
- Guidelines on a method to select revetment materials for dikes and
shores.
- Guide to concrete Dyke Revetments.
- Guidelines for River Dykes.
- The Dutch Committee on Management of Waterways (CVB). One of the objectives
is to prepare the guidelines on various aspects involved in the management
of waterways (i.e. nautical guidelines, guidelines on bank protection,
etc).
Project example: - Design recommendations for bank protection of navigation
fairways (CVB, 1988).
-4-
- The Dutch Centre for Civil Engineering Research, Codes and Specifications
(CUR). The main objective of this centre is to stimulate the collective
research in the field of civil engineering. In the CUR, experts from public
authorities, contractors, consultants and research institutes are acting on
the same collective-basis.
Example of projects: - Structural aspects of environmentally acceptable
banks (Manual).
The Dutch practice has learned that the best way to perform an integrated
(multi-disciplinary) research project is to do so by organizing working groups
or project-teams and appoint independent chairmen when all institutions invol-
ved are able to participate. Most research carried out under the above-
mentioned task committees is realized by the working groups not only con-
sisting of researchers but also of designers, contractors and local managers.
It helps to identify and to define the problem, to create understanding for a
chosen research strategy and to implement the results in an efficiënt way.
This is one of the ways of building bridges between research and practice.
For solving some complex problems, the common Dutch practise is to apply a
probabilistic approach and specifically the "event or fault tree" (Figure 2 ) .
To apply this method, all possible causes of failure have to be analysed and
consequences determined. It has to be stressed that having quantified (even
roughly) the fault tree, it is possible to pay extra attention to those mecha-
nisms which contribute most to the overall probability of failure. Thus, this
approach is an important element in the quality-control processes of the
design and execution of projects.
The fault tree is also an important tooi for preparing the integrated research
program for erosion control projects and for objective judgement of priori-
ties.
Taking knowledge of these recent developments can be rather beneficial,
especially for the estimations of possible risks involved in realized projects
and for finding the optimum between risks and investment, including research-
investment.
fallura rrtar
trsiniAC wwrfc
macro
tnatabUity inaUbtlitr
datormatlan «tablllly faüur* pari fallur» uafc:

problama problaza of raYatmant tranaiUon avattta laflnaao
(•atUamant)
lUdinc flow Plptn* azaeuttaa laautfletaai

alida taotta
dteaata»
(axeaadanoa
daatfft eandltlraa)
loei loeal plplnc UqvataotiAO laatabfnty

•Udlnt upUfttzx (aarta quaka) alamaata ehaaleai
ooiwr lapar of flttar laflia
matarlaj
hTdranUa
Figure 2 Simplified fault tree
-5-
BANK EROSION DÜE TO RIVERINE PROCESSES
General
Land-use planning in alluvial river valleys and the choice of locations for
bridges and hydraulie structures require predictions of future river planform
changes and, consequently, knowledge of river-bank erosion and riverine pro-
cesses. Of particular interest is the problem of stabilizing a river planform
by eonstructing protection works at certain carefully selected locations only,
allowing for solutions attractive from an economie but also an environmental
point of view, as natural banks provide biotic and abiotic diversity, which
are very important for riverine eco-systems (see Section Environmental aspects
of river bank erosion). For this reason, some canalized streams in Germany
have been changed back into more natural ones by partial removal of bank
protection works (Keiler and Brookes, 1983; Kern and Nadolny, 1986). For
rivers in the Netherlands, De Bruin et al. (1986) have developed similar
ideas, parts of which have been incorporated in the Dutch government's policy
for urban and rural planning.
The ünderstanding of the mechanisms of bank erosion along rivers is imperative

for assessing the need for protective works and for the design of these works.
This ünderstanding should comprise both the riverine morphological processes
that determine the erosive forces, and the soil-mechanical properties of the
river banks that characterize their ability to withstand erosion.
In order to gain insight into the interactions between river-bank erosion,
river flow and bed topography, and in order to develop tools for the predic-
tion of river planform changes and their morphological consequences, research
is performed at the Delft University of Technology and at Delft Hydraulics. At
present this research is coneentrating on the riverine processes considering
both meandering and braided rivers.
Erosion problems
Bank erosion shows up in different forms: gradual recession of the banks and a
consequent loss of bank vegetation or cliff development (the often steep tran-
sition between the river bank and the groyne field beach). Sometimes serious
erosion occurs during floods. Factors governing erosion are, amongst others,
the frequency and duration of high current velocities caused by navigation and
flood.
An analysis of erosion of Dutch river banks in reaches with important push-tow

navigation based on a comparison of airial photographs taken between 1975 and
1985, resulted in the following conclusions (see Table 1 ) : almost half of the
river banks in the reaches observed are protected; one-sixth of the unprotect-
ed banks do not appear to be eroded; two-third shows erosion up to 1 metre/
year and one-sixth shows severe erosion of more than 1 metre/year. No conclu-
sions can be drawn from the airial photographs about the causes of erosion. In
fact, a number of factors may be responsible: the level of the lower bed, the
geometry of the reach, the discharge regime, navigation. In order to decide
upon the cause(s) in a particular case and to prepare the most suitable
designs for erosion control measures, the river reach and its banks have to be
studied as a part of the river system with its dynamic variation in water
levels and water and sediment discharge.
-6-
total protected unprotected

Class: m/yr length banks banks
% km % km % km
1: 0 52 168 44 142 8 26
2: 0 -1 40 131 5 17 35 114
3: 1 -2 6 22 1 5 5 17
4: over 2 2 6 0 0 2 6
total 100 327 50 164 50 163
Table 1 Bank erosion rates on reaches with push-tow navigation
For instance, the following phenomena may contribute to bank erosion:
In the middle and upper reaches of alluvial rivers the lower bed may
degrade because of, for example, mining activities, training works and
navigation. In such cases the groyne field beach will also degrade. The
slope of the beach is very small (1:30 to 1:100) because of the small size
of the sediment which has settled particles there (low flow velocities). In
fact, a slight degradation of the bed may eventually cause severe bank ero-
sion.
- During floods, large quantities of water and sediment are exchanged between
the lower bed and the flood plain. The water exchange occurs with large
velocities, especially where there are sudden changes in the bed topo-
graphy, for example at the banks. During floods groynes function like
weirs, and after floods local erosion can often be observed downstream and
sedimentation upstream of the groynes.
If navigation becomes important on a river, the impact of large and heavy

powered ships on the bed will increase and inevitably bank erosion will
grow. The propellor race may cause bed erosion, which can indirectly affect
the banks. In contrast the bow and stern waves of fast-sailing ships attack
the banks directly. The immense water displacement of large ships causes
heavy currents near the banks, generating large sediment transport.
The qualitative description above only indicates possible causes of bank ero-
sion. Generally speaking only the total erosion can be observed in the field,
the causes seldom being obvious.
Mechanics of river-bank erosion

River-bank erosion is a complex phenomenon in which many factors play a role:
flow (discharge flow, groundwater flow, wind waves, ship waves), bed topo-
graphy (undermining), sediment transport (e.g. removal of slump debris after
mass failure) and bank properties (bank material weight and texture, shear
strength including cohesive properties, electrochemical properties, bank
height and cross-sectional shape, groundwater level and permeability, strati-
graphy, tension cracks, vegetation and constructions).
At present, the numerical models under development only account for river-bank
erosion due to shear stresses exerted by discharge flow. According to Blon-
deaux and Seminara (1985) these shear stresses can be represented well by the
longitudinal near-bank shear stress, as continuity implies that the vertical
-7-
component of the flow field close to the banks is driven by the perturbation
of the longitudinal component, and is relatively small with respect to the
latter.
Cohesive banks usually erode by mass failure, which implies that bank geometry
fluctuates during bank retreat. Following mass failure slump, debris accumu-
lates at the bank toe. The debris is removed by lateral erosion prior to
further bank oversteepening or bed degradation generating further mass
failures. These periodical bank geometry changes cause apparent variations in
bank erodibility, thus complicating erosion laws.
However, when assuming that the debris is removed immediately from the toe,
though still taking complex bank failure mechanisms into account, the time-
average migration rate can be well characterized by a critical shear stress
and an erodibility coëfficiënt of the bank material at the toe (Osman and
Thorne, 1988). This complies with field observations of Hickin and Nanson
(1984), who found the relationship between grain sizes at the outer bend toe
and bank migration resistance to be very similar to Shields' diagram. They
conclude that bank migration is primarily determined by fluvial entrainment of
basal sediments, after which cohesive upper sediments erode by the collapse of
cantilevered overhangs.
The considerations above justify the use of a simple bank erosion law in the
models, similar to relations for the erosion rate of cohesive soil samples
(Ariathuri and Arulanandan, 1978):
for x > T (2.1)

! • « • ( . . • V c
f"» for T < x

c
(2.2)
in which dB/dt is the migration rate, E is an erodibility coëfficiënt, T is

the shear stress exerted by the near-bank longitudinal flow component and x
is a critical shear stress below which no erosion occurs.
Arulanandan et al (1980) give relations to determine the erodibility coëffi-
ciënt and the critical shear stress of a cohesive soil. Osman and Thorne
(1988) consider the approach of Arulanandan et al to be one of the most promi-
sing of the currently available methods, because calculation of erodibility
and critical shear stress is based on the electrochemical properties of the
soil, pore water and eroding fluid.
The presented bank erosion law has been incorporated in a mathematical model
for river morphology. A linear analysis of this model with the approach of De
Vriend and Struiksma (1983) indicates that the input of bank erosion products
decreases transverse bed slopes, but hardly influences the wave lengths and
damping.lengths of flow and bed topography in natural rivers. A more important
effect of an increase in bank erodibility seems to be widening of the bed and
associated shallowing. This change of width-to-depth ratio leads to longer
wave lengths and less damping. Ultimately, when the banks are extremely
eroded, the river may become braided.
Meandering rivers
For meandering rivers tools are being developed to predict changes in plan-
form. At the Delft University of Technology research is going on, in combina-
tion with Delft Hydraulics, to develop a mathematical model for large-scale
changes in planform (Crosato and Struiksma, 1989). The 1D model, MIANDRAS,
combines the approach of Johannesson and Parker (1985) with the insight gained
in the Netherlands on the water movement and bed topography in curved channels
-8-
(Struiksma et al, 1985; Struiksma, 1985). Also the understanding of the occu-
rence of natural cut-offs of meanders is improved via analytical modelling and
exploring experiments (Klaassen and van Zanten, 1989). This phenomenon is the
controlling mechanism, causing that the length of a meandering river does not
increase, in the long run (Figure 3).
Furthermore, also at the Delft University of Technology the two-dimensional
depth-averaged model RIPA is being developed. This model deals with river
planform changes on a smaller scale and in fact is an extension of the work of
Olesen (1987), in which more attention is being paid to the bank erosion and
its effects on the two-dimensional bed topography in river bends.
Figure 3 Example of actual and predicted planform of river
An arbitrary river geometry with a curved centre line, a non-uniform width and
eroding banks requires a boundary-fitted coordinate system ('grid'). Such a
coordinate system for which each segment of the boundary of the physical
region coincides with some coordinate line is very convenient for the repre-
sentation of boundary conditions and, therefore, has been chosen for implemen-
tation in the numerical model. After a certain amount of bank migration, a new
grid must be generated, which is adapted to the new river geometry. Non-
homogeneous bank erodibility can be accounted for by attributing bank
properties to each bank grid-point separately. In order to retain these pro-
perties at the proper locations when generating a new grid, bank points are
not allowed to shift along the banks. They only move perpendicularly to the
local bank lines. This restriction implies that it is not possible to generate
a purely orthogonal grid (Thompson et al, 1985).
Summarizing, the coordinate system to be implemented is boundary-fitted, cur-

vilinear and non-orthogonal. The set of equations is solved in the transformed
space, which is inherently rectangular with a square grid. Due to the trans-
formation some terms have to be added to the equations. These terms account
for rates-of-change of grid spacing and for grid skewnesses. An economical
grid generator has been developed. The transformation of equations is current-
ly being performed (progress has been made, but not completed yet).
-9-
Braided rivers
For braided rivers an extensive study of the erosion along the Jamuna (= lower
Brahmaputra) River in Bangladesh was recently carried out by Delft Hydraulics.
This study used satellite images over successive years. It was found that bank
erosion along such a braided river is far more difficult to predict due to the
much more difficult processes. Bank erosions occur both along curved channels
and along straight reaches, The bank erosion along curved channels is to some
extent similar to what was observed by Hickin and Nanson (1984) for meandering
rivers, notably an increase of the celerity of bank erosion with decreasing
relative radius of curvature. Results are, at present, only available in
limitedly distributed reports within the framework of the Jamuna Bridge
Appraisal Study (RPT/NEDECO/BCL, 1989), for which the study was carried out.
Another interesting aspect is the substantial reduction in outer bend scour in
the Jamuna River due to the contribution of bank erosion products. Bank ero-
sion is, on average, about 300 m/year, but can be up to 1000 m/year. The maxi-
mum depth in the outer bends is only about twice the average channel depth and
independent of the relative radius of curvature of the bend.
BANK EROSION DUE T0 NAVIGATION
General
For many Dutch rivers (Figure 4), the lower bed is restricted by groynes. The
bed in the groyne field between two groynes may be considered as a beach,
visible during low water periods. The beach and bank in some groyne fields, is
often eroded. A reason for such bank erosion may be navigation, especially
when large, fast ships like push-tows use the river. Therefore, it is neces-
sary to understand the erosion process when considering the introduction of
(six-)barge push-towing. With this knowledge predictions of the extent of ero-
sion can be made and designs prepared for "environment-friendly" river-bank
protection instead of traditional bank protections consisting of layers of
stones or blocks laid on geotextiles. With respect to traditional bank protec-
tion the past research and the general approach to erosion control of naviga-
tion embankments have already been treated extensively in recent publications
(Blaauw et al, 1984; Pilarczyk, 1984; van de Kaa et al, 1985; PIANC, 1987a).
Therefore, only the new developments related to the possible introduction of
six-barge push-towing are discussed.
Figure 4 The Dutch sections of the River Rhine

-10-
Push-towing is a possible cause of bank erosion, of which, however, the physi-

cal processes are not fully understood. Therefore, this type of bank erosion
has been investigated in field studies and in laboratory experiments. The 1986
six-barge push-tow trials were used to obtain field data of the flow field and
sediment transport in groyne fields. Small-scale hydraulic tests were carried
out at Delft Hydraulics to study the flow field and bank erosion which could
be caused by push-towing. This research was aimed at predicting whether or not
six-barge push-tows units would result in greater erosion than four-barge
push-tows (Brolsma, 1988).
Flow patterns in groyne fields
The poor correlation between clearly defined examples of bank erosion and its
possible causes among which navigation, resulted in carrying out a study to
the erosion processes caused by push-tows, rather than to actually measure the
erosion itself. With respect to the full-scale tests, the investigations were
concentrated on two groyne fields, one in a straight reach and one in a bend.
In a small-scale model at Delft Hydraulics on length scale 1:25, the flow
pattern was systematically studied in a large and a small groyne field.
Figure 5 shows the flow pattern without navigation in a large groyne field
(200 x 50 m) at a river discharge of 1450 m3/s (mean flow in the River Waal).
A vortex (eddy) develops immediately downstream of the groyne where the main
current cannot make a sharp bend into the groyne field. Further downstream the
main current does enter the groyne field and consequently the stream width is
larger. A second eddy develops at the upstream face of the second groyne where
the outflow is hampered by this groyne.
There is a continuous exchange of water and sediment across the streamline be-
tween the main flow and the flow between the groynes. The current velocities
in the groyne field (0.3 m/s or less) are smaller than the velocities in the
main current (1 m/s). Sediment carried by the incoming flow of water will tend
to settle in the groyne field, except in the eddy downstream from the upstream
groyne where the flow velocities are about 0.4 to 0.5 m/s.
Down-stream Up-stream
Figure 5 Flow pattern in a groyne field, no navigation (dimensions in metres)

-11-
In a shorter groyne field the incoming flow is strongly reduced and the down-
stream eddy does not develop and so the groyne field is dominated by the up-
stream vortex.
Before discussing ship-induced water movement in groyne fields, some observa-
tions should be made about the water movement around a ship in a fairway with-
out groynes (Figure 6). When a ship is loaded the return current and the
supply flow are the dominant phenomena for banks.
A. front wave
B. water level depression
C. transversal stern wave
D. return flow
E. supply flow
F. secondary waves
G. propeller race
Figure 6 Water movement around a ship in fairway without groynes
The direction of the return current and the supply flow are parallel to the
fairway axis. The return current is in the opposite direction to the ship's
sailing direction, while the supply flow is in the same direction. The aft
boundary of the water level depression caused by the return current is called
the transversal stern wave and here the change in flow direction between the
return flow and the supply flow occurs. Normally the propeller race has no
direct impact on bank stability. The size of the secondary waves depends on
ship speed and thus are only important in relation to bank erosion, when ships
sail relatively fast, which is not the case with loaded push-tow units.
The effects of push-towing on the flow patterns in fairways with groynes is

schematically presented in Figure 7. The discontinuities in the cross-section
formed by groynes cause flow concentrations near the groynes and flows perpen-
dicular to the fairway axis.
Three important stages can be distinguished during the passage of a push-tow
unit sailing upstream. The return current is at a maximum immediately when the
bow passes a groyne. The return current is furnished by water from the up-
stream groyne field and the groyne field alongside. An eddy develops at the
groyne head and the small vortex at the downstream end of the groyne field
apparently disappears entirely. As the push-tow passes by the supply flow
refills the groyne field. When the stern of the push-tow passes the particular
groyne field, the supply flow is forced to flow perpendicular to the axis of
the fairway, out of the groyne field at the upstream groyne. The eddy immedia-
tely downstream of the groyne is transported downstream by the main current.
•12-
upstream downstream
V
c
>»
y
o
bank
Figure 7 Flow pattern in a groyne field during passage of a push-tow unit

-13-
The following conclusions could be drawn from the model investigations and the
field measurements:
- Push-towing induces large velocities, especially immediately downstream of

the groynes. The magnitude of the velocities depends on ship dimensions,
snip speed, distance between push-tow unit and groyne, and barge draught.
- The maximum current velocity in small groyne fields are less than in large
groyne fields, because of the smaller effect of the supply flow.
- When the river discharge increases, the river cross-section also increases
and this causes a reduction of the return current and supply flow and thus
a reduction of the flow velocities.
- Field experiments in 1983 indicated that, whereas push-tow units tended to
produce a considerable increase in the flow velocities in the groyne field,
the largest self-propelled ship (about 2000 tons) had very little effect.
Erosion and sedimentation in groyne fields
A thorough understanding of the ship-induced water movements in groyne fields

is essential for predicting bank erosion. However, since the relationship be-
tween flow and sediment transport is highly non-linear, and in curved flow,
the direction of the flow does not coincide with the direction of the sediment
transport, measurements of sediment transport are also necessary.
Therefore, field observations have been carried out in the two pilot groyne
fields as well as tracer experiments in the small-scale model. The findings of
these erosion studies were as follows:
- Four and six-barge push-tow units induce increased sediment transport out
of the groyne field, especially immediately downstream of the groyne head,
caused by the increased intensity of the eddy at the moment the stern
passes.
- The amount of sediment transported out of the groyne field, as a result of

the passage of a six-barge push-tow unit, was 1.5 to 3 times larger than
that with a four-barge push-tow unit. Of the transported material very
little returns. The rate of return is larger for a four-barge push-tow unit
than for a six-barge push-tow unit.
- Calculations with a conceptual mathematical model based on the insight

obtained indicated small erosion rates for the groyne beach, both in the
case of four-barge and six-barge push-tow units. As is to be expected the
six-barge push-tow unit produces a somewhat larger erosion rate. Other
erosion-producing phenomena appear to be dominant because of the small flow
velocities. This is a continuing process which is only disrupted by passing
push-tow units.
- Field observations made in the large groyne field indicated an aggradation

(or recovery) capability of about 0.1 m per year, assuming no navigation,
floods and low-water bed degradation. In contrast, the shorter groyne field
does not have much recovery capability.
-14-
-2.0
-1.5
-*. Diatance bow - groyne (after paaaage)

Puah-tow fortnatton :
Draught 3 m.
1x2 o
2x2 o
3x2 7
-2.0
Passing distance 100 m
-1.5 v <7-
i*
o
'S
-1.0
f ^z \
>
o
N,
-0.5 ^ .
*v
AA
v
-o o o„
^Sx
<)•:^ 8^D
126 250 375 500
-*. Oistance bow - groyne (after paaaage)
Draught 3 m.
Figure 8 Flow v e l o c i t i e s , push-tow unit passing a t 50 m and 100 m

from groyne field
-15-
Conclusions
Push-tow navigation causes erosion in groyne fields. The erosion rate, how-
ever, is low. Six-barge push-tow units almost certainly will cause more ero-
sion than four-barge push-tow units. The erosion starts immediately downstream
of the groyne heads, and will extend to the river bank. Eventually the bank
will collapse. The rate of bank erosion due to push-towing depends also on the
reeovery capability and local sediment transport of which more information is
required. Consequently, the rate of bank erosion cannot be predicted yet.
ENVIRONMENTAL ASPECTS OF RIVER BANK EROSION
General
In the long history of river bed regulation, the main rivers in the Nether-
lands have been transformed from rivers that meander ed freely in a broad f lood
plain to one fixed in a single channel with less sharp bends and contained in
a flood plain reduced in width. This also holds for the branches of the River
Rhine, which is one of the main navigable waterways. The river flows in an
almost flat landscape and consequently, builds up its own flood plain by
sedimentation. The annual fluctuations are moderate compared to other rivers.
Most of the discharge comes from precipitation in the catchment areas in the
upper river reaches in Germany, particularly in winter and spring. In surmner
the Alpine region contributes with meltwater.
-41*- plalatocana eovar tand arna
oialatneana csvw aand rot

ptalatocia « • puaiwd ridgaa
aandy leut
«'.. ' | «and and gravni

dry graaland
«nan high watnr laral
awan grond walar !•»•)
Figure 9 Cross-section of the Rhine,

1000 AD (above) and 1900 AD (below)
Every reach of the river has its own environmental features with corresponding
plant and animal communities composed of general and specific species. The
loss of natural environment has been considerable, sinee centuries ago human
acitivities started to influence the river (Figure 9). Up to the 19th century
this influence only consisted of diking activities and river 'training works in
order to prevent flood disasters and to improve navigation. Nowadays, severe
water pollution and intense navigation also play a role. The result of this
-16-
all is that in the flood plain natural forests of willow and poplar were re-
placed by pastures, crests of high natural levees were flattened and banks
were smoothed.
Abiotic diversity
In the present situation the low-water channel has sufficiënt capacity to con-
tain the river water most of the time without inundating the pastures on the
higher grounds between the dikes. High discharges now flood this restricted
flood plain only once or twice a year. Summer dikes have been constructed in
most areas to prevent summer inundation. These dikes also prevent polluted
water entering the summer polder. The low-water channel is bordered by
beaches, groynes and other artifical bank protection structures composed of
bitumen, concrete, riprap, blocks and geotextiles. At high discharges the
border is formed by the natural levees or winter dikes (Figure 10).
Geomorphological features such as old meanders, steep slopes, old and young
natural levees, shallow and deep water pits and man-made clay pits, dikes and
other structures create a very complex, continuously changing landscape.
In, places where old natural levees of flood plains are exposed to the eroding
forces of the current, steep slopes are created. This is a natural phenomenon
that can be seen in most river landscapes. However, in the present Dutch river
landscape the natural phenomenon of steep and high slopes on river banks is
almost totally absent.
groyne clay pit summer dlke winter dik»
channel beach river bank clay pit old channel
.. summer polder
summer bed
winter bed
Figure 10 Schematized section across an active flood plain
In the past the river banks were more varied and to a certain extent covered
with vegetation. Nowadays, however, the overall picture is one of almost bare
sand beaches, except in the transition zone to the estuarine, where there is
still a marsh land vegetation.
The young soils of the river banks are to some extent calciferous, in particu-
lar the soils in front of the summer dikes. Their composition varies from
coarse sand and gravel in the deeper subsoil, to alternating layers of sand
and clay and also thick clay packets.
Higher areas are dryer and more ripened and homogenized by the activity of
soil fauna and show less hydromorphic features because of lower groundwater
levels. Former river beds or cut-off meanders, have slowly turned into marches
with organic soils. Hydrological factors play a very important role in the
distribution of organisms in the river bank system. Compared with dry ecosys-
tems the variations in time and place are great. Most dramatic, for example,
is the periodic inundation of the river bank which only a s'pecialized selec-
tion of organisms can survive. In addition, the water quality of the River
Rhine deteriorated to a minimum. Recently, a slight recovery has taken place.
The effects of pollution are severe because many polluting agents, such as
-17-
heavy metals and polycyclic aromatic organic substances, become bound to clay
mineral deposits and stay in the system for a long time. In the short term,
the accumulation of poisonous substances in the organisms of the food chain
are more serious.
Biotic diversity
Nowadays, the low-water channel borders are almost completely without vegeta-
tion, because of the turbulent water motion generated by navigation and the
sand movement. The higher grounds are less exposed to eroding forces and carry
a permanent vegetation cover. The impact of the water regime on nature is very
decisive and governs the presence or absence of organisms in the flood plain
and on the river banks.
In early spring when most organisms hibernate, the effect of flooding is mini-
mal, but in summer, flooding can be disastrous for many plants that are
flowering or setting seeds and some animals species can suffer under oxygen
depletion.
There is a relation between the organism and the mean duration of annual
flooding. At the water line, species can be found which are resistant to high
flow velocities (reed, willow). Some species are tolerant to inundation be-
cause they possess aerial canals in their sterns (reeds, bulrushes). Others
have short life cycles to avoid the flooding season. Organisms most sensitive
to flooding are found on the crest of the natural levees and on dikes.
Biotic diversity is also related to soil diversity. Clay soils do not dry out
quickly since they contain very fine particles and retain ground water better
than sandy soils. Permanently water-saturated soils are less aerated or un-
aerated and so only marsh land species can grow.
The mean duration of flooding is a measure for expressing the sensitivity of
living organisms or the chance they can be found at a particular place. It is
also an excellent way to describe different environments with specific biotic
compositions. On this basis the environment of the river bank is divided into
five more or less separate zones with their own features corresponding to pos-
sible life forms and abiotic phenomena and processes:
zone 1: permanently under water
zone 2: frequently flooded in summer
zone 3: sometimes flooded in summmer
zone 4: very seldomly flooded in summer
zone 5: never flooded in summer.
Protection and restoration
A new environmental equilibrium had to be reached along the rivers
The activities up to now already caused a substantial loss of habitat varia-
tion, which in turn has reduced the number of species which a river can sup-
port. A more recent negative influence is caused by the intensification of
agriculture, which has resulted in the increased application of biocides and
the transformation of extensively used land cropped for hay and rich in
species to agro-industrial pasture with a poor species composition. Thornbush-
hedges which have contributed enormously to the ecological infrastructure of
the river landscape have, at the same time, been replaced by barbed wire.
Cattle water holes rich in amphibian and insect life have lost their function
and disappeared. In addition cattle treading has damaged the sandy soils of
beaches and dunes rich in plant and animal species.
All these aspects have contributed to the growing awareness of the grave
situation of the environment and governmental and public willingness to take
action.'
-18-
The contents of some pollutants, for instance, have already been decreased.
The willingness of the Dutch Government to take action is there and plans are
being developed and have been partially worked out already. Consultation with
other Rhine-countries has resulted in agreements to limit discharge of wastes
into the river, to improvements in industrial installations and to prohibit or
alter industrial processes as much as possible. At present, inventories of the
river banks are being made. This will give information on the current abiotic
and biotic diversity and the spatial differentation and quality of types. A
monitoring network will gather information on fast or slow changing ecological
processes, identify external influences and assist the interpretation of the
experimental results.
The aim for the near future is to clean the river to such an extent that re-
introduction and spontaneous repopulation of organisms takes place.
Restoration of the abiotic environment so as to improve the possibilities for
living creatures that belong in a river (bank) ecosystem is a more distant
aim. Within the restrictions of river management and costs, preparations are
being made and ideas have been presented to restore the variation and quality
of the biotic and abiotic environment, which can keep itself in equilibrium.
A first stage is the construction of nature-restoring types of bank protec-
tion. Little is known about appropriate methods for protecting river banks
and, at the same time, for giving nature a chance. Experimental structures
properly managed and maintained, are needed to evaluate possible solutions.
Growth of willow forests at river banks and creation of less dynamic shelter
areas between groynes seem possible. More toleration of land loss could dimi-
nish the amount of the artificial solutions. Adequate connection of the river
with cut-off bends or isolated sand-pits could restore spawning places for
fish and offer new habitats to macro fauna and marsh vegetation.
Conclusions
A substantial loss of plant and animal habitat has taken place, due to river
bank erosion, enforced wave action on unprotected beaches and artificial bank
protection constructions that suppress natural processes.
To prevent further degradation of the natural environment with its specific
flora and fauna elements and the particular river landscape, measures are
necessary. The willingness of government and public is growing and has already
resulted in plans to enlarge natural developments, in measures to reduce pol-
lution and attention for biotechnical bank protection solutions.
INTEGRATED RESEARCH ON REVETMENTS
General
Numerous types of revetments have been developed in the past for shores and
for banks of navigation channels to prevent erosion by waves and currents
(i.e. riprap, blocks, asphalt, e t c ) . The reason for this was the increase of
the problem in respect to the defence of the shores (i.e. more rigid safety
requirements for seadikes) and banks of navigation channels (i.e. increase of
size and speed of motor vessels), as well as the high cost and shortage of .
natural materials. The fact that design rules are still limited in quantity
has stimulated new investigations in the area of riprap, artificial blocks and
bituminous revetments, as well in the area of 'geotextiles.
Problems which arise due to these developments, require solutions which often
only can be found by specific in-depth multidisciplinary studies (Figure 11).
-19-
In order to control the future seadikes and bank protection problems, the
Dutch Ministry of Transport and Public Works (Rijkswaterstaat) assigned Delft
Hydraulics and Delft Geotechnics to carry out a systematic research into these
areas. On the base of the analysis of practical design problems and the gaps
in the existing knowledge, the required research programme had been determined
(Figure 12). This programme follows the general SOWAS-concept as outlined in
Figure 11; it includes the integration of two fields (banks and dikes) as well
interdisciplinary integration (SOil-WAter-Structure).
The basic programmes have been carried out by means of small-scale models.
However, it has to be pointed out, that a small-scale hydraulic model for
navigation purposes needs still a lot of space. In the scope of the bank
protection research programme, for example, the hydraulic model of an inland
navigation fairway on scale 1:25 has been built in a 40 x 90 m 2 shed to
observe the induced water motions and their erosive effects on the banks.
Figure 11 Soil-water-structure Figure 12 Sea-dikes and bank

interaction for protection research
revetments approach
Since model research has certain inherent technical restrictions known as

scale effects, required additional information has been obtained by means of
prototype investigations i.e. the Delta-Flume and some prototype locations in
respect of the sea-dikes problems, and the Hartel Canal (Rotterdam area) with
test embankments in respect to bank protection problems of navigation
channels. The result of the prototype tests, in combination with the model
results and the caleulation methods (incl. mathematical models) developed
within the framework of the systematic research on dike protection (Delft
Hydraulics and Delft Geotechnics, 1989) and systematic research on bank pro-
tection (Delft Hydraulics, 1988) extended with knowledge gained from practical
experience, had led to the preparation of guidelines for reliable dike and
bank protection designs. The aim of the total research effort was to develop
-20-
such design criteria that the amount of maintenance and construction costs of
new revetments is minimized.
Research aspects and research means
In general, to be able to determine the dimensions of involved protection

layers of revetments the following design research-aspects have to be taken
into consideration:
A) Research on characteristic/representative load
A1 - water motion due to wind waves, currents and ship movements
A2 - geotechnical load (gradients) due to the water motion (A1)
B) Research on characteristic/representative strength

B1 - strength of toplayer
B2 - strength of sublayers
B3 - geotechnical strength (i.e. sliding)
To solve problems involved in A) and B) various research means are available,

viz.:
a) Evaluation of past experiences (lessons from practice)
b) On-site investigations on existing and/or test revetments, i.e. prototype
measurements in the Hartel Canal (Pilarczyk, 1984)
c) Calculations and mathematical models (desk studies)
d) Small-scale physical models
e) Large-scale models, i.e. Delta Flume.
The physical and mathematical models are very suitable for basie research
within a wide range of boundary conditions and for the development of general
design rules, while on-site investigations, evaluation of experience and
large-scale (prototype) tests are needed for verifying the final results
(design rules). In the scope of the discussed research programmes all the
above-mentioned research methods have been applied. The general research
strategy for both programmes and their interactions is presented in Figure 12.
The main difference in approach to the problem between these two investiga-
tions lies in the fact that for the seadikes the large Delta Flume has been
used for prototype tests while for the bank protection problems the 'in-situ'
prototype tests have been carried out. (Large-scale models to help solve bank
protection problems are not a reasonable solution). Another point of
difference is that the boundary conditions related to banks cannot be
reproduced mechanically as is the case for wind-waves with a wave generator,
but they have to be induced by ship movement.
Besides the difference in reproducing of the hydraulic load both programmes
involve some common aspects regarding stability of the toplayer and the
sublayer. The integration of both programmes had taken place by means of
mathematical models which had to fulfil both requirements. These mathematical
models, called STEENZET, (Bezuyen et al, 1987; Burger, 1988) and DIPRO (CUR,
1989) might become important tools in the design of revetments of dikes and of
banks of navigation channels.
The technical description of the projects and the results discussed above can
be found in the references. However, to illustrate the SOWAS-concept on the
technical level, the integration of the soil, water, and structural processes
into one conceptual stability model will be repeated below (de Groot et al,
1988).
-21-
General stability approach (SOWAS-concept)
The phenomena which may be relevant can be divided roughly according to the
three components of the system: water, soil and structure. The interaction
between these components can be described using three Transfer Functions (see
Figure 13).
I. The Transfer Function from the overall hydraulie conditions (e.g. wave
height H, mean current velocity U) to hydraulie conditions along the ex-
ternal surface, i.e. the boundary between free water and the protection
of soil (e.g. external pressure P e x ) .
II. The Transfer Function from the hydraulie conditions along the external
surface to those alöhg the internal surface e.g. the boundary between
protection and soil. The hydraulie conditions along the internal surface
can be described as the internal pressure Pint-
III. The structural response of the protection to the loads along both sur-
faces .
HYORAULIC HYORAULIC
OVERALL
CONDITIONS CONDITIONS
HYDfUULIC
AT EXTERNAL AT INTERNAL
C0NOITI0NS
SURFACE SURFACE
H, U
Pex <y.t) Pint <V.t)
BEO/SLQPE HYORAULIC KECHANICAL

GEOHETRY PROPERTIES PROPERTIES
OF STRUCTURE OF STRUCTURE
ANO SOIL
Figure 13 Schematic presentation of the three transfer functions
Information about these functions can be obtained by means of measurements in

nature and (scale) model tests. If quantitative knowledge of the physical
phenomena involved is available, or if there is enough experience at hand,
then mathematical models or empirical formulae containing information are
referred to as "models".
All three Transfer Functions can be described in one model, or individually in
three separate models, depending on the type of structure and loading. The
distinction between the three functions serves here mainly as a framework to
describe the different phenomena that are important for the modelling.
Based on the extensive past research on various protection systems it is pos-

sible to present the load-strength relationships of these systems in the more
uniform way as it will be presented in the next sections.
-22-
Overview of practical design formulae
a) current attack, civil engineering structures (Pilarczyk, 1989)

A general formula:
AD = , K 0^35 K -1 11
m n ^ T 4>er h s 2 g°
in which: D n = thickness of protection unit [m], a = relative density of
protection unit [-], $ = stability factor [-], K T = turbulence factor [-], K h
= depth factor [-], K = slope factor [-], 41 = critical shear stress
parameter [-], ü = mean velocity [m/s], g = acceleration of gravity [m/s 2 ].
The strength parameters a en D can be calculated with:

for rock: Dn= (M5()/PS)O.33 or D n = 0.85 D 5 0 and a m = a (Pg- pj/p^
for mattresses (gabions, sand-mattresses, e t c ) : D n = d (= average thickness

of mattress) and a = (1-n)a
with M C Q = 50? value of the mass ISistribution of the stones [kg], p = speci-
fic weight of rock [kg/m 3 ], p = specific weight of water [kg/m 3 ], n = poro-
sity of stone or sand [-] (approximately 0.4).
The different K-factors follow with:

K, = 2/(log 12h/k ) 2 for a logaritmic velocity profile or
K. = (h/D )-o.2 f o r a not-fully developed velocity profile
K = cos S (1-tan2 a/tan2 9)°-s = (1-sin2 a/sin2 8)°- 5
K™ = 0.67 (low turbulence, uniform flow); K T = 1.0 (normal turbulence,
rivers); K T = = 2.0 (high turbulence, local disturbances, outer bends),
with h = water depth [m], k = D (smooth units, i.e. concrete blocks) or k =
2D n (rough units, i.e. rock), a = slope angle [°], e = angle of interval fric-
tion [°].
The following remarks can be made. Firstly, the factor K^ can be neglected if
in stead of ü the bottom velocity ufa is substituted. Secondly, the factor K s
only holds for bank and shore slopes. For a bed slope a different value should
be used. Finally, the value of 2 for K T should only be applied if due to
difficulties in defining the local mean velocity, the average mean velocity is
applied.
The following values of the stability factor $ are recommended:
$ = 1.25 for exposed edges of loose units

cj> = 1.0 for exposed edges of block-mats and/or mattresses
<t> = 0.75 for continuous protection of loose units
ij> = 0.50 for continuous protection of block-mats and/or mattresses
with i|»er = 0.035 for rock and 4>er = 0.06 * 0.10 for gabions.
Examples of exposed edges are: bed protection at scour holes (particularly in
the case of two-directional current i.e. ebb and flood), edges of a toe pro-
tection, transitions between adjacent revetment systems, connections between
mats or mattresses.
-23-
b) wave attack, eivil engineering structures (Pilarczyk, 1989)
The general formula is:

H
cosa s . , cosa
w
*' "V *° Ir
with: E = tan ct/(H ?L )°-s = 1.25 . T /H°-s . tan a
z s o z s
in which: <J>„ = 4> = stability factors [-], H = significant wave height [m],
T = average wave period [s], L = wave lengch [m], D = thickness of protec-
tion unit [m], a = slope angle [°], A = relative density [-].
The general equation is valid especially for placed/pitched stones and blocks,
more in particular: the left side for free blocks/stones (<t>f) and the right-
hand side for old revetments (<t> ) with natural friction and/or interlocking.
The following values of the parameters <t>f and * are recommended:
y type of revetment
f 2?5 poor quality (irregular) pitched stone
2.0
3.0 3.5 good quality (regular) pitched stone
3.5 4.0 natural basalt and less permeable (closed) blocks
4.0 5.0 relatively permeable blocks (open area 5 * 20%)
5.0 6.0 closed blocks on good quality and smooth clay-surface
It should be emphasized that in the case of the first four types of revetment
a granular sublayer is required. In the case of blocks on a geotextile on a
sandy subsoil, the applied wave height should be less than 1.5 m, because of
the danger of liquefaction.
For block-mats and grouted/interlocked systems on a granular sublayer the

right-hand side of the general equation can be applied. Values for <t> are:
4> = 3-5 — blocks connected to geotextile by pins

4> = 4.0 - grouted blocks connected by geotextile
$ - 4.5 - cabled closed blocks
<j> = 5.0 - cabled open blocks; grouted concrete prism (basalton)
4> > 6.0 - grouted cabled blocks; properly designed mechanically interlocked
blocks
The edges of the adjacent block-mats, if not properly connected, should be

treated as free blocks (<)> « 3 ) .
In all cases, experience and sound engineering judgement play an important
role in applying these design rules, or else mathematical or physical testing
can provide an optimum solution.
For other revetment systems, such as riprap, grouted aggregates (riprap,

bitumen grouting), gabions (incl. gabion mattresses), open stone asphalt,
fabric (geotextile) mattresses the general wave-design equation has been
changed slightly:
H
m n z
with ip = upgrading factor (ip = 1 . 0 for riprap and tp * 1.0 for other revet-
ments systems).
-24-
Depending on the type of revetment different values for A and <t> should be
substituted. For Dn should be applied a value based on Men or in the case of
mattresses the value of d. The value of i|> can vary between 1.5 and 3.0 depen-
ding on the type of revetment and for $ a safe value is 2.25.
Also more sophisticated formulae have been developed. For instance, for
plunging waves (E < 2.5) attacking slopes protected with loosely materials
taking into accont duration of wave attack, damage level and permeability of
sublayer and subsoil (van der Meer and Pilarczyk, 1984):
1
H /AD < 6.2.p,°-l8.(S2 /N)°"
s n b b «i
with p. = permeability factor with p, = 0.1 for the practically impermeable
core (ï.e. sand-/clay-body), and p. = 0.5 for the permeable (granular) core
(i.e. breakwaters), N = number of waves [-], S. = damage level [-]
A physical description for S, is the number of cubical stones with a side of 1

x D , eroded over a width of 1 x D . The "no-damage" criterion is generally
applicable when S. is between 1 and 3 stones eroded.
c) biotechnical revetments (Klein Breteler et al, 1988).
Due to the recent trend to environment-friendly solutions for bank protection,

research is started on combined systems (artificial protection consisting of
cellular revetment blocks allowing vegetation), which may be applied in the
situations with hydraulic loadings exceeding the value of the natural
resistance of particular vegetation especially in the first stage of growing
(just after planting).
Cellular revetment blocks combine the merits of a closed revetment and a
revetment of loosely packed materials: permeable enough to prevent the occur-
rence of high uplift pressures, and the individual blocks support each other
against severe wave attack. Thus, the advantages are high permeability, a high
stability against wave attack, a low block weight, and last but not least, the
holes allow the growth of vegetation (Figure 14).
The aim of the ongoing research is the derivation of equations predicting the
erosion of hole fillings, because for the stability of the slope protection
and the establishment of a vegetation it is of vital importance that the
material in the holes does not erode completely. Studied aspects are water
motion, geometry of holes and characteristics of hole filling.
celluldr btock failure m e c h a n b n u
undi&turtied
hole lilling
Figure 14 Cellular concrete block revetment

-25-
In the case of a bank protection along a navigable waterway the main attack is
caused by ship waves. As a first result the following equation could be
derived (Figure 15):
Y / G = C (z
max )0 * 5 ' l 0 S (N+1) (D5Q « 0.15 mm)
with c = 0.7 for sand and c = 0.5 for black earth, G = hole width [m], Y =
maximum erosion [m], z = height of ship wave [m], N = number of ship pas-
sages [-].
Along sea dikes wind waves cause the main attack. Depending on the size of the
holes the failure mechanism is outflowing water (small holes) or overflowing
water (large holes). Of course, the loads on sea-dikes will be too high for a
vegetation, however erosion of hole filling is not allowed because of instabi-
lity of subsoil. The tests resulted in the equations:
G > 100 mm: Y/G =0.2 (H/D_n )o.3 3 (D 5Q * 5-25 mm)

t>0
G < 100 mm: Y/G = 25 (H/D5Q)o.33 . ( D /G)i-*
G = hole width [m], Y = maximum erosion [m], H = wave height [m], D c n = dia-
meter hole filling [m]. ->"
Note: The first equation is also valid for small holes above the still water
line.
The main conclusion of the 200

research carried out up to now, HOLE NR. 10
175 H - 0.13 m
is that cellular blocks can be G - 0.06 m
applied along inland waterways
and coasts. A safe design ap-
proach is to apply blocks with a
£150 sand
g 125
hole width less than half the 2
block thickness. S ïoo
n
o
«
1
W 73
/
- bla clc eai-th
50
25
0 500 1000 1500 2000 2500 3000 3500 4000

»- NUUBER OF WAVES (-)
Figure 15 Erosion depth vs. number of waves

CHOICE OF MEASURES
In general the choice of a proper measure for a particular case is related to

the functional requirements and the cost-benefit optimalization. The latter
needs preparation of various technical alternatives satisfying the functional
requirements.
Regarding the bank protection, to predict when and which measure is desirable,
it is essential to have a proper understanding of the hydraulic loading
(waves, currents, water stages, sediment transport e t c ) , the current
morphological processes and the possible interactions/response due to possible
measures including ecological response.
-26-
The actual state of knowledge on all these aspeets it not always sufficiënt
for solving these problems. However, due to the gradually increasing knowledge
on the measures already used and the morphological and ecological response to
them, the possibilities of better prognosis and choice of proper measures are
also gradually increasing.
It is important that already at an early stage of the preparation of hydraulic
schemes, the main view point be formulated that purely civil engineering
measures are not always necessary. Taking into consideration the knowledge of
natural processes and availability of materials, it must be possible in most
cases to build appropriate ecological shore/bank protection ai.u to control the
erosion processes.
By involving biological elements (vegetation) in the solution some reduction
in cost may also be achieved. In the case of sandy shores, the dynamic prin-
ciples should be used instead of fighting against nature (e.g. sand replenish-
ment).
The choice of protective measures today also depends on a weighing-value of
functions of the area considered. Such functions are recreation, fishery,
nature (biological value), navigation, etc. Thus, before choosing a measure
the following aspeets have to be analysed (Pilarczyk, 1986):
- what does society want to do with this area (destination of the area)?
- what kind and what amount of erosion is likely to happen without taking
measures? (not taking any measures is one of the basic alternatives)
is doing nothing responsible/desirable?
- if not, how can the erosion be fought or stopped?
To arrive at a choice it is necessary to proceed systematically. In The
Netherlands the so-called policy-analysis is often recommended for this pur-
pose. The principle of this method is that no decision is taken during ana-
lysis; only the pro and contra arguments for different solutions are collected
and properly weighed against each other.
The final choice is left to the proper authorities.
The example of this approach for Dutch estuaries is given by Pilarczyk (1986).
CIMtS _A«ES ESTjAO'CS LAXES .ESTUARIES
ctrannel foreshore banu
Figure 16 Basic forms of bank protection in The Netherlands

-27-
The method to select the revetment materials for covering dikes and banks is
prepared by TAW (1988). Requirements have been formulated which, in general,
have to be met by various materials. Selection criteria have been derived from
these requirements for different local situations. By including the costs of
eonstruction and maintenance in the selection procedure, a conscious decision
concerning the materials (incl. environmental aspects) to be used can be made,
based on qualitative as well as economie grounds.
It is recommended to make the choice of protective systems by a group of the
involved specialists so that the subjective aspect can play the least possible
part. For the different aspects weighing factors can be made so that a more
objective choice may be possible.
The basic foras of banks in the Netherlands are shown in Figure 16. In the
case of wide rivers and channels with no space-restriction, the common way of
protection is an indirect protection by groynes and sills parallel to the
bank. The crest level of a sill is determined by the water level and the
transmission of waves related to the type of vegetation on the land-side. The
most common vegetation for larger rivers is reed. For smaller rivers there is
a larger diversity of water plants because of lower hydraulic loading or
better water quality.
However, the cross-section of I
most navigation channels in the
Netherlands is only restricted (often a road)
for navigation purposes and
defined by the nautical
__
guidelines (classes of traffic,

see PIANC (1987a)), while the
•wet-pile
land along the channels is in

private hands. %
In such cases the realization of 'Mür-Ljâv y/i^f v
m
the environment-friendly
solutions needs a special
attention and can often be more
costly than a Standard solution.
An example of such a situation
with some alternatives is given
in Figure 17.
In general the installation costs

of the environment-friendly
measures are lower than the
Standard measures but the
maintenance costs are higher and
need another approach. However,
recent Dutch studies have
indicated that, on average, the
environment-friendly solutions
need not be more expensive than
the traditional solutions, if
both installation- and
maintenance costs over the life-
time of construction are properly
taken into account. A legal
problem can arise when, for Figure 17 Examples of environment-
example, the installation costs friendly solutions for
are subsidized by some central restricted waters
-28-
authorities and the maintenance costs are not. In such a case the local
authorities will prefer a more expensive traditional
(rigid) solution with less maintenance.
SHORT OVERVIEW OF CURRENT RESEARCH
The following studies are being carried out (and their end date):
• Reed as bank protection. Damping of wave attack (1989 )'*
• 2DH-numerical model of morphological changes and river bank erosion (1991)H
• Influence of soil type on erosion through cellular blocks (1989)2
• Desk study on scour at groynes (1989)1'2
• Geotechnical aspects of slope protection (1990)3
• Influence of berm on hydraulic loading on a bank (1989)2
• Prototype inventarization on growth of vegetation in open block-mats
(1990)i
• Physical description of hydraulic loading between the groynes (1990)1'5
• Guidelines for clay specifications for grassmats (1990)1
• Guidelines for vegetation for river-dikes (1991)1'5
•. Conceptual models on behaviour (and failures) of dikes and banks (1990)5
• Rational maintenance and management of river and channel banks (1990)5
• Computer program on design of navigation embankments (1989)2'5
• Computer handbook on scour in hydraulic engineering (1989)1»2
• Handbook on rock in hydraulic engineering (incl. river training works)
(199D 5
• Manual on constructive aspects of environment-friendly (biotechnical) bank
protection (199D 5
• Evaluation of vegetation requirements for bank protection (1990)s
• Manual on filter constructions in hydraulic engineering (1990)5
• Manual on application of alternative materials in civil engineering incl.
environmental aspects (1991)5
• Numerical tools for maintenance and management of hydraulic structures
(1989)2
• Decision models for operation management (1989)2
• Conceptual model on erosion rate of unprotected banks along navigable
waterways (1991)2'5
C0NCLUSI0NS AND REC0MMENDATI0NS
• The current knowledge and available (mostly 1D) mathematical tools for
treatment of bank erosion are still not adequate for solving more compli-
cated practical treatment of problems. Still more research is needed on
that especially related to two-and three-dimensional treatment of problems.
• The present mathematical tools are important for sensitivity studies and
for making a proper choice from alternative solutions.
• Because of the complexity of the river morphology and the mechanics of bank
erosion for some specific problems, physical scale modelling is still the
only investigation tooi.
1 Rijkswaterstaat, 2 Delft Hydraulics, 3 Delft Geotechnics,

4 Delft University, 5 Cooperative research incl. international cooperation.
-29-
• The current knowledge on erosion and sedimentation processes due to inter-

action of natural currents, waves and water stages and protective struc-
tures is still in a rudimentary stage and it is not yet possible theoreti-
cally to describe many phenomena and their interactions.
• The natural resistance of various types of vegetation against hydraulic
loading is still not sufficiently evaluated, especially regarding the soil
and the conditions after planting, and the methods of their (temporary or
permanent) protection.
• The "technical optimum" goal for a bank protection should be transformed
into a "social-economic optimum". It should include not only technical and
economie aspects but also nature, landscape, environment, recreation, em-
ployment, etc.
This means an evolution from just a bank protection to a multi-functional
water-land transition.
• A proper balance between construction, inspection and maintenance, in terms
of direct and indirect costs and effects, should be aimed at.
• The costs of environment-friendly solutions/measures seem often high com-
pared with the maintenance costs of rigid protective structures. However, a
cost-benefit analysis, including capital investment, will often turn out in
favour of environment-friendly solutions.
• A review of the available data on erosion control indicates the lack of a
central (international) data-bank, both for laboratory as on-site data. In
general, the monitoring-data from the existing/realized schemes is rather
scarce.
• Nowadays, ship-induced water motion can approximately be predicted for dif-
ferent traffic situations for schematized waterways. The same holds for the
erosion of cover layers and filters for the common protective materials.
The state-of-the-art on these subjects was already published. However, for
more complicated geometries of waterways and banks supplementary research
incl. physical modelling is still needed.
• A probabilistic approach for the design of a bank protection is important
to obtain insight into the different "limit states" of structural stabi-
lity. The state-of-the-art was published by PIANC (1987b). The event or
fault tree can also be a useful tooi for programming research and defining
the priorities, and scheduling available budget.
• Computer models are under development to describe the water motion and to
check the stability of banks. Existing techniques are only able to solve
schematized situations satisfactorily.
• For research subjects of more general (national or international) interest
it is preferable to organize collective research projects where all the
proper authorities are involved. It increases the financial possibility of
the project, its effectiveness and stimulate the implementation.
• It is important that the existing (international) knowledge on erosion con-
trol be systematized and made available for designers and managers in this
field. This existing knowledge seems to be rather limited from a national
point of view but it can be much more than we expected when it is inte-
grated internationally into one homogeneous unit.
• Because of the complexity of the problems and the shortage of national
research-funds, international co-operation, especially in the common fields
of problems, should be further stimulated. International organizations as
IAHR, PIANC or ASCE Task Committee on River Mechanics can play an important
role in the realization of this aim.
REFERENCES
Ariathurai, R. and Arulanandan, K. (1978),

Erosion rates of cohesive soils,
J. of Hydr. Div., ASCE, Vol. 104, No. HY2, February 1978.
Arulanandan, K., Gillogley E. and Tully R. (1980),
Development of a quantitative method to predict critical shear stress and
rate of erosion of natural undisturbed cohesive soils,
Rep. GL-80-5, US Army Engineers, Waterways Exp. Stn.,
Vicksburg, Miss.
Bezuijen, A. , Klein Breteler, M., and Bakker, K.J. (1987),
Design criteria for block revetments and granular filters,
2nd Inter. Conf. on Coastal and Port Eng. in Dev. Countries, Beijing,
China.
Blaauw, H.G., Knaap, F.C.M, van der, Groot, M.T. de, Pilarczyk, K.W. (1984),
Design of bank protection of inland navigation fairways,
Inter. Conf. on Flexible Armoured Revetments Incorp. Geotextiles,
London (also Delft Hydraulics, publ. No. 320).
Blondeaux, P. and Seminara G. (1985),
A unified bar-bend theory of river meanders,
J. of Fluid Mech., vol. 157.
Brolsma, J.U. (coord.), (1988),
Six-barge push-tow trials,
PIANC Bulletin No. 62, Brussels.
Bruin, D. de, Hamhuis, D., Nieuwenhuijzen, L. van, Overmars, W. Sijmons, D.
and Vera, F. (1986),
Ooievaar, De toekomst van het rivierengebied (The future of the river
landscape from an environmental point of view).
Uitgave Stichting Gelderse Milieufederatie, 1987.
Burger, A., Klein Breteler, M., Banach, L., Bezuijen, A. (1988),
Design method for block revetments,
21st Int. Conf. on Coastal Engineering, Malaga, Spain.
Crosato, A. and Struiksma N. (1989),
Analysis of a 2D bed topography model for rivers (to be published),
A.G.U., Washington DC, USA.
CUR/Centre for civil engineering research, codes and specifications,
Rijkswaterstaat and Delft Hydraulics (1987),
Manual on artificial beach nourishment,
report 130, Gouda, The Netherlands.
CUR/Centre for civil engineering research, codes and specifications (1989),
DIPRO, Computer design program for bank protection in navigation fairways
(in Dutch),
CUR, report Q614 Delft Hydraulics, Gouda.
CVB/Dutch Committee on Management of Waterways (1988),
Design recommendations for bank protection of navigation fairways
(in Dutch),
Rijkswaterstaat, The Netherlands.
Delft Hydraulics (1988),
Technical guidelines for riprap and block revetments along navigation
fairways (in Dutch).
Final report on systematic research on bank protection, Report M1115
part XIX, Delft, The Netherlands.
Delft Hydraulics and Delft Geotechnics (1989),
Technical guidelines: Final report on systematic research on block revet-
ments for dikes (in Dutch),
Report M1881/M1795, Delft, The Netherlands.
REFERENCES (continued)
Groot, M.B. de, Bezuijen, A., Burger, A.M., Konter, J.L.M. (1988),
The interaction between soil, water and bed or slope protection,
Proc. Inter. Symp. on Modelling Soil-Water-Structure Interactions
(SOWAS '88), Delft, The Netherlands.
Hickin, E.J. and Nanson G.C. (1984)
Lateral migration rates of river bends,
J. of Hydr. Eng., ASCE, Vol. 110, No. 11.
Johannesson, H. and Parker, G. (1985),
Computer simulated migration of meandering rivers in Minnesota,
Report no 242, St. Anthony Falls Hydraulic Lab., Univ. of Minnesota,
Minneapolis.
Kaa, E.J. van de, Groot, M.T. de, Hijum, E. van, Pilarczyk, K.W., Stuip, J.
and Verhey, H.J. (1985),
Erosion control of navigation embankments,
26th Inter. Navigation Congress, PI ANC, Brussels.
Klaassen, G.J. and Zanten, B.H.J. van (1989)
On cut-off ratios of curved channels,
Proc. XXIIIth IAHR congress, Ottawa, Canada.
Keiler, E.A. and Brookes A. (1983),
Consideration of meandering in channelization projects:
selected observations and judgements,
River Meandering, Proc. of the Conf. Rivers '83, New Orleans,
Louisiana.
Kern, K. and Nadolny I. (1986),
Naturnahe Umgestaltung ausgebauter Fliessgewasser (in German),
Institut für Wasserbau und Kulturtechnik, Mitteilungen, Heft 175,
Karlsruhe.
Klein Breteler, M., Laboyrie, J.H. and Verhey, H.J. (1988),
Erosion of sediment through cellular blocks applied as slope protection
along coasts and inland waterways,
Proc. Inter. Symp. on Modelling Soil-Water-Structure Interactions
(SOWAS '88), Delft, The Netherlands.
Leewis R.J. et al (1983),
Shore Protection - a tension field between two types of conservation,
Wat. Sci. Tech. Vol. 16, Rotterdam (printed in Great Britain).
Meer, J.W. van der, and Pilarczyk, K.W. (1984),
Stability of rubble mound slopes under random wave attack,
19th Inter. Conf. on Coastal Eng., Houston,
(also: Delft Hydraulics Publ. no. 332).
Olesen, K.W. (1987)
Bed topography in shallow river bends.
Communications on Hydr. and Geotechn. Eng., Rep. No. 87-1,
Faculty of Civil Eng., Delft Univ. of Technol., The Netherlands.
Osman, A.M. and Thorne C R . (1988),
Riverbank stability analysis, I: Theory,
J. of Hydr. Eng., ASCE, Vol. 114, No. 2, 1988.
PIANC (1987a),
Guidelines for the design and construction of flexible revetments incorp.
geotextiles for inland waterways,
Supplement to PIANC Bulletin, no. 57, Brussels.
PIANC (1987b),
Risk consideration when determining bank protection requirements,
Supplement to PIANC Bulletin no. 58, Brussels.
REFERENCES (continued)
Pilarczyk, K.W. (1984),

Prototype tests of slope protection systems,
Inter. Conf. on Flexible Armoured Revetments Incorp. Geotextiles, London.
Strategy to erosion control of Dutch estuaries (state-of-the-art),
3rd Symposium on River Sedimentation, Jackson, Mississippi, USA.
Sea Defences: Dutch guidelines on dike protection,
Rijkswaterstaat, Report WB Mo. 87110, The Netherlands.
Also: 2nd Int. Conf. on Coastal and Port Eng. in Develop. Countries,
Beijing.
Design of coastal protection structures (in short course on Design of
Coastal Structures),
Asian Institute of Technology, Bangkok.
RPT/NEDECO/BCL (1989),
Jamuna Bridge Appraisal Study, Feasibility Report and Design Report River
Training Works,
Rendel Palmer and Tritton/Nedeco/Bangladesh Consultants Ltd.
Schilperoort, T., Wijbenga, A. and Zwaard, J.J. van der (1985),
Mathematical tools and their growing importance for river engineering,
Delft Hydraulics, Separate No. 85/08.
Struiksma, N., Olesen, K.W., Flokstra, C. and Vriend, H.J. de (1985),
Bed deformation in curved alluvial channels,
Journ. of Hydr. Res., IAHR, Vol. 23 no. 1.
Struiksma, M. (1985),
Prediction of 2-D bed topography in rivers,
J. of Hydr. Eng., ASCE, Vol 111, no. 8.
TAW/Technical Advisory Committee for Water Defences (1988),
Guideline on a method to select revetment materials for dikes and shores,
(in Dutch, English translation in 1990),
Rijkswaterstaat, The Netherlands.
Thompson, J.F., Warsi, Z.U.A. and Mastin, C.W. (1985),
Numerical grid generation,
Elsevier Science Publishing Co., New York.
Vriend, H.J. de, and Struiksma N. (1983),
Flow and bed deformation in river bends,
River Meander ing, Proc. of the Conf. Rivers '83, New Orleans,
Louisiana.
Behaviour of armour layers of riprap bank protections along navigation channels
R.E.A.M. Boeters Directorate-General for Public Works and Water Management, Road
and Hydraulic Engineering Division, P.O. box 5044, 2600 GA Delft,
The Netherlands
F.C.M, van der Knaap - Delft Hydraulics, De Voorst Branch, P.O. box 152, 8300 AD
Emmeloord, The Netherlands
H.J. Verheij Delft Hydraulics, De Voorst Branch, P.O. box 152, 8300 AD
Emmeloord, The Netherlands
Abstract
Model tests have been carried out to determine the damage that may occur to a riprap armour layer
along a navigation channel, for instance because of the increase in ship-induced hydraulic loads
compared to the initial design bad. The executed model tests at Delft Hydraulics have been set up
with the assumption that the occurrence and development of damage to armour layers caused by
ship-induced waves are similar to the behaviour of breakwater armour layers under wind waves.
In a European Class IV waterway sections with different riprap sizes have been tested. Profile
measurements have been carried out several times after a number ofship passages, the maximum
of which being 300. The analysis of the results has shown that the development of damage to an
armour layer along a navigation channel can indeed be predicted with a mathematical model similar
to that for breakwaters. The derived model can be used to predict the behaviour of riprap bank
protections as a function of time.
1. Introduction
Armour layers of riprap protecting banks along fairways, are often designed by taking into account
a maximum characteristic hydraulic load and calculating a stone size that ensures that the applied
riprap protection will be absolutely stable. Design rules developed by Iribarren and Hudson are used
in this respect.
Still, damage of the riprap protection occurs. Possible reasons for this are the inccease in hydraulic
loads and the displacements of smaller stones, inducing instability of the larger ones.
1
The ways in which damage to riprap armour layers of breakwaters develops, caused by the attack
of wind waves and the number of these waves, has been extensively studied by Van der Meer (Van
der Meer, 1988). Secondary waves induced by shipping are very much similar to wind waves.
Therefore the idea arose to investigate the possibility of using the formulas of Van der Meer to
describe and predict the damage to riprap bank protections, caused by ship-induced hydraulic loads,
as a function of the number of loads exerted.
When indeed the formulas of Van der Meer can be applied to riprap protections along fairways, the
opportunity occurs to predict damage to these protections as a function of time, since there is a
direct dependence between the number of ship-induced loads and the number of passages, during
a year for instance. The prediction of damage as a function of time can be very useful when
planning maintenance of a protection.
In this paper the study of the applicabiiity of Van der Meer's formulas to riprap bank protections
is described. The investigations (Van der Knaap e.a., 1992) consisted of a series of model tests on
a length scale of 1 to 10.5 and an analysis of the characteristics of secondary ship waves in
comparison to wind waves.
2. Aims of the research and approach

The main goal of the research was to develop a predictive model for time-dependent development
of damage to different types of riprap bank protections along fairways. Because of the similarity
between armour layers of quarry stone of bank protections and of breakwaters on the one hand,
and between wind waves and secondary ship waves on the other, it was considered that the most
efficiënt way to develop such a predictive model was to focus on the predictive models derived by
Van der Meer, which are valid for breakwaters attacked by wind waves. This means that the
already existing formula of Van der Meer, giving the relationship between development of damage,
properties of the protection, and the number and magnitude of waves, provided the framework for
the investigations to be carried out. The general form of Van der Meer's formula is as follows:
HJÏ c po.18 {_§_)* (D

A0* {N
In which
Hs significant wave height [m]
f tan a IV (Hs/L), surf similarity parameter [-]
a angle of the slope [°]
L wave length [m]
A relative mass density [kg/m 3 ]
Dn nominal stone diameter [m]
C coëfficiënt [-]
P permeability coëfficiënt of the armour layer [-]
S damage level, A/(D n60 ) 2 [-]
N number of waves [-]
A erosion area [m2]
An explanation of the damage level S is given in figure 1.
2
•f 1 L-ter Larer
initial sLope
prof i te ofter 3000 waves
dlstance (n)
Figure 1. Damage level S, based on erosion area A
In order to check the validity of this type of formulation of damage prediction for bank protections
surcharged by ship-induced waves, a physical model was installed in a flume, in which all the
relevant variables could be varied. This meant that different riprap protections, varying in stone size
(Dn) and permeability (P), had to be applied. The waves were generated by a small, fast sailing ship.
The sailing speed of a ship determines the height and length of the waves that are generated, so
by varying the speed, different combinatipns of wave height (H) and length (L) could be achieved.
Each run with a specific sailing speed was repeated a great many times, in order to get a large
number of waves (N). By measuring the damage profiles of the different protections for each
combination of wave height and length, at regular intervals of the number of ship passages, sets
of data were obtained that could be fitted in the form of Van der Meer's formula. When this should
prove to be possible, one might conclude that the assumptions about the validity of Van der Meer's
formulation for the development of damage to a riprap bank protection along a fairway were correct
and a similar mathematical description could be derived.
Of course, it was realised that there are differences between wave loads caused by wind and
generated by ships, and between the circumstances under which damage occurs to a breakwater
and to a riprap bank protection.
First of all, wind waves exist almost continuously, while ship-induced waves only occur when a
ship passes by. In a wave field caused by a storm, there is a rather large variety in the wave
heights and lengths that occur, and the properties of such a wave field have to and can be
statisticaliy described by a Rayleigh distribution. The secondary waves induced by a ship show a
more or less fixed pattern, with one or two high waves, foliowed by four to six decreasing waves,
strongly intercorrelated. Therefore it was decided that a statistical analysis had to be made of the
ship-induced secondary wave field, in order to make a comparison with the Rayleigh distributed
wind waves and to be able to choose a characteristic waveparameter, like the significant wave
height, which is used in Van der Meer's original formula.
Secondly, damage to a breakwater will only occur in a storm with a certain probability of
occurrence, which depends on the technical life-time of the structure, for example once in a
hundred years.
3
In navigation channels, the possibility exists every day that the design load occurs and
consequently, that damage appears. This means that there is the question of order of succession
in the occurrence of wave loads and, subsequently, in the development of damage. In other words,
is the occurring distortion of the riprap armour layer sensitive to the order of appearance of
different heights of wave loads, or is it possible to superimpose the effects of different classes of
wave loads, neglecting their sequence, in order to predict the same distortion?
This problem was investigated by using both sides of the model. These were provided with exactly
the same riprap protections. One side however, was put into its originai profile after each series
of tests with a specific combination of wave height, length and number of passages. The other side
was left untouched for several of the combinations mentioned above. By comparing the damage
level of this side with the superimposed damage levels of the other side, that was reconstructed
after every series of tests, it should be possible to draw conclusions about the validity of the
principle of superposition for the development of damage.
3. Description of the model and testseries

At Delft Hydraulics a model has been built of a European Class IV canal, on a length scale of 1 to
10.5. A transsection with prototype dimensions is depicted in figure 2.
!!!!!!!!!!!!!!!!!,,- • ..illllllllllllllll
lil
'Il|| ,>'"
Figure 2. Transsection of the modelled navigation channel (prototype dimensions)
The total length of the model was 35 meters, plus at each side 20 meters of acceleration and
deceleration zone for the model vessel. In the center of the model, at each side, four sections of
5 meters wide were provided with riprap protections with different stone sizes and layer
thicknesses. Almost in the middle of each 5 meters wide section, a space of 1 meter wide was
prepared to execute measurements of the displacements of the stones. Figure 3 gives a bird's eye
view of the model.
2*4 test sections
" I I I
1.0 15 2.0
I
L4
I
L1 L2 L3
1
|
salllngdlrection
R1 R2 I R3
I R4
M
« wave height meter distances In meters
• current veiocity meter
HJrneasuring area
Figure 3. Bird's eye view of the model
4
Table 1 shows the four different types of riprap protection that have been installed.
Table 1. Riprap protections applied in the model.
Section layer t h i c k n e s s Dsoin Da, in t h e D 8 5 /D 1 5

prototype (m) model (mm) (model)
1 1*Dso 0.20 19.51 2.10
2 2'Dso 0.12 11.50 2.21
3 2'Dgo 0.08 8.94 2.04
4 10'Dgo 0.06 6.46 1.72
The model vessel "Rixt", used in the test, had the shape of a small tugboat. lts dimensions (in
meters, prototype):
Length 12.20
Width 3.15
Draught 1.20
This vessel was pulled through the water, it had no power installed aboard.
The course of the vessel was in the centre of the simulated channel, thereby ensuring that at both
sides the attack by the ship-induced waves was the same.
In table 2 a summary is given of the velocities with which the ship was pulled through the water
and the number of passages after which the profiles of the test sections were measured.
Table 2. Sailing velocities of the model ship and number of passages
Series Vessel Vessel number of passages between measurements of Total

speed speed profile number
(m/s) (m/s) of
(model) (proto first second third fourth fifth sixth passages
-type)
T1 1.30 4.20 10 20 30 40 30 130
T2 1.40 4.55 5 5 10 20 60 60 160
T3 1.45 4.70 5 15 60 220 300
T4 1.50 4.85 10 30 60 100
The sections on the left side were restored after each test series. The sections on the right side
were restored after series T1 plus T2 and after series T3 plus T4, in order to investigate the validity
of the principle of superposition of damage levels.
Measurements of the distortion of the armour layers were executed by means of a profile follower,
mounted in a specially designed frame, that could be placed over the measuring section, each time
in the exact and same position. Measurements were taken in eight rows per section. To be able
to use the profile follower, the waterlevel had to be raised temporarily to the toplevel of the sides
of the model.
5
4. Testresults
In figures 4 and 5 two examples are shown of the damage profiles that were the result of the runs
with the model vessel. In each figure the test section and the number of passages are mentioned.
0.45
^V
0.40 : ^
^ :
\
0.35 J^J
' ^
o
.a
x>
0.30
• water level
• initial profiie
• profiie after 20 runs in test section L3
0.25
0.00 0.10 0.20 0.30 0.40 0.50 0.60
distance (m)
Figure 4. Damage profiie in test section L3 after 20 runs
0.45
0.25
0.00 0.10 0.20 0.30 0.40 0.50 0.60
distance (m)
Figure 5. Damage profiie in test section L4 after 100 runs

As already mentioned, eight profiles were measured in each testsection. To derive the values of
the damage level S, t w o methods were applied:
1. By composing an average damage profile out of the registered eight profiles, and calculating
the damage level for this profile (S1).
2. By calculating the damage levels for all eight measured profiles and taking the average of
these values (S2).
As it appeared, there was little difference between S1 and S2. In the analysis, S1 has been used.
Atseveral places in the model, waves were registered. See figure 3 forthe exact locations. Figure
6 shows an example of such a registration.
0.06
0.04 -
- 0.02
.5?
0.00
o
-0.02
-0.04
-0.06
time (s)
Figure 6. Example of a measured secondary wave pattern
Table 3 summarizes the maximum wave heights and average wave periods that have been
registered in each test series, at the toe of each slope. The values are transiated into prototype
wave heights and periods.
Table 3. Registered wave heights and periods
Series sailing sailing Model Prototype

speed speed
(m/s) (m/s)
max. wave wave max. wave wave
(model) (prototype) height period height period
(m) (s) (m) (s)
Tl 1.30 4.20 0.055 0.68 0.58 2.20
T2 1.40 4.55 0.065 0.69 0.68 2.24
T3 1.45 4.70 0.074 0.74 0.78 2.40
T4 1.50 4.85 0.084 0.78 0.88 • 2.53
7
5. Analysis
5.1 Secondary ship waves compared to wind waves
To make a statistical analysis of the secondary ship-induced waves, a number of five runs have
been made with the model vessel having a fixed speed. This has been done for five different speeds
(1.2 to 1.6 m/s). All registrations have been combined into one long series of wave registrations
and this has been statisticaliy anaiysed, using a same method as appiied to wind waves. Figure 7
shows the combined registrations and table 4 gives the result of the analysis.
0.08
0.06
S 0.04
„ 0.02
2
-0.02
TÊW "f*
0.00 *|jlf[Wl'llW^#^*'"''^Wfr» •ll.l.i...
"•Ml""'
.1 llltlj
11*
ijlU
|Prtp*
-0.O6
-O.0B
20. 00. too.
TIM (S)
-O.08
o.oa
o.os
S o.oo lilU.Art? Mn] T | ^ # » Ut^m-^m

mffffl
r^Wî^l] jLfMfft-^4 f\ hwW»w
«
x
-o.oa
m
| -0.04
-0.06
-0.08 1
240. 260. 300. 330. 380.
T I K (s)
Figure 7. Combined registrations of secondary waves, generated by the model vessel
Table 4. Statistical parameters of the secondary wave field
WHM005
number of waves 437
Wave height H1/3 (m) 0.0310
Wave height H 1/10 (m) 0.0568
Wave height Hm0 (m) 0.0420
Average wave period (s) 0.806
Peak period (s) 0.890
Spectral density (m2/Hz) 0.213
8
It appears that secondary ship waves have different ratios between the characteristic wave heights
that describe a wind-generated wave field, which represent the Rayleigh distribution, H 1/3 , H 1/10 and
Hm0. So one may conclude that secondary ship waves do not fit the Rayleigh-distribution. This is
confirmed by the graph in figure 8, that shows the probability of exceedance, derived for the
registered waves, plotted against the probability according to a Rayleigh-distribution.
0.10
0.08
0.06
f f'
CD
> <£ 'p
| 0.04 $
V
0.02
0.00
100 90 50 20 10 5 2 1 0.5 0.1
exceedance percentage (%)
Figure 8. The probability of exceedance of secondary ship waves plotted against the same
probability according to a Rayleigh-distribution
Van der Meer has shown that in case of broken waves on a shallow foreshore, which are also not
Rayleigh-distributed, distortion of riprap armour layers of breakwaters could be predicted by using
H 2% . Based on this and on the graph in figure 8, one may conclude that in a predictive model for
damage caused by secondary ship waves, some kind of H2% should be used, meaning more or less
the maximum wave height.
5.2 Derivation of the predictive model

In order to get a proper description of a model that predicts damage to a riprap bank protection,
caused by secondary ship waves, a kind of "short cut" analysis has been carried out. Because of
the similarity, it was assumed that the model should have the same form and terms as Van der
Meer's formulation for breakwaters attacked by wind waves, see section 2, formula (1).
The above means that the analysis of the data could be constrained to four items:
a. Wave parameters (H\/f)
b. Number of waves (N)
c. Permeability (P)
d. Coëfficiënt (C)
a. Wave parameters
Earlier investigations (Verheij and Bogaerts, 1988) have resulted in a set of equations with which
9
the maximum secondary wave height (H), the average wave period (Ti) and waveiength (L,) can be
calculated. The equations are:
s,-°-33 , K 4 (2)
h V Jgh
v (3)
T. = 5.1 -l£
0
7? (4)
With:
s distance between the ship's side and the secondary waves (m)
h waterdepth (m)
Vs sailing speed (m/s)
g acceleration of gravity (m2/s)
In figure 9 a comparison is made between maximum wave heights and average wave periods, as
measured during the test series and calculated with the given equations. The resembiance is very
good.
0.12 1.1
'0.10 1.0
i
0.08 0.9
/ 11
0.06 >f x> j T 0.8

i
0.04 0.7 i <^t •

X.
0.02
X > mra sured 0.6
** > measured
• - calc jlcrted •s
— — • calculaied
i
•N U, 1 1 1
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9
shlp spe«d ( m / s ) wm- shlp speed ( m / s )
Figure 9. Comparison between measured and calculated secondary wave heights and periods
(model data)
As stated in section 5 . 1 , the choice of the maximum wave height to be applied in the damage
prediction model, looks very appropriate. Since the given equations can be applied to a large variety
of ship-fairway combinations, it was decided to use in the predictive model the wave parameters
that follow from these equations.
10
b. Number of waves
By choosing the maximum wave height as a governing parameter in the model, the number of
waves can be directly translated into the number of passages. Furthermore, as can be seen in
figure 7, secondary waves have a more or less fixed pattern, with one or two high and four to six
smaller waves. So with a known number of passages, the number of waves is also known.
Now the question remains whether the reiationship between the measured damage levels and the
number of passages is a function of the root of the latter.
In the graph of figure 10 t w o exampies are shown of this reiationship.
^ j
9
8
-
— ! X T2 , I L 2 , par
' M l '
= 4,22
7 —
6 — - O T1 , L3, par= 4,16
5 — . Q T1 , R3, par= 4,16
4 —
• T2 , L3 , par = 5,42
•&
<b-&
^
A T2 , R3, par= 5,42
S = 2,2 Vn ^s*
vttlS
'^y^
<
3.
1
10
A
ir <^>1
1 K n / ^
[f
i» *\ •<S>«£
5 —
S =4,0 O ° _* 'T
corr. T^.Ri /
• <
> \ < < S =0,35Vn "
*
irT
K N
K \
I M-t 13 1
C
i
m c
1
| H
R
2
4 5 6 7 8910 4 5 6 7 S910 2 3 4 5 6 789103
Figure 10. Reiationship between S and n, test series T1 and T2
Based upon an analysis of the graphs of this kind for all test series, it could be concluded that the
root of the number of passages is highly correlated with the damage levels S.
c. Permeability
Van der Meer has given an overview of different types of riprap armour layers and connected
values for the permeability P, see figure 1 1 .
11
OnioA = nomina! diameter of armour stom
DnsoF = nominol diameter of filter material
OnsoC = nomina! diameter of core
Figure 1 1 . Permeability coëfficiënt P
Based on this, a value of 0.1 has been chosen for test sections 2 and 3, and 0.5 for sections 4.
Sections 1 showed hardly any damage, and have not been taken into account.
d. Coëfficiënt
Having chosen all the relevant parameters and using the same functional relationship between these
as in Van der Meer's formulation, the coëfficiënt C can be determined.
Table 5 gives a summary of the result.
Table 5. Values used in determining C
Series Sections Dn P Hs/f SA/n C

(m/s) (m) (m) (-) (mm) (-) AD n (-) (-)
T1 13, R3 1.3 0.05 0.73 1.09 7.42 0.1 4.16 0.35 7.8
T1 L4, R4 1.3 0.05 0.73 1.09 5.36 0.5 5.76 0.4 7.8
T2 L2 1.4 0.068 0.84 1.00 9.55 0.1 4.22 0.35 7.9
T2 L3, R3 1.4 0.068 0.84 1.00 7.42 0.1 5.43 2.2 7.0
T2 .L4, R4 1.4 0.068 0.84 1.00 5.36 0.5 7.51 1.95 7.4
T3 L2, R2 1.45 0.078 0.90 0.97 9.55 0.1 4.76 0.57 8.2
T3 L3, R3 1.45 0.078 0.90 0.97 7.42 0.1 6.12 1.58 8.7
T3 L4, R4 1.45 0.078 0.90 0.97 5.36 0.5 8.47 2.5 8.0
T4 L3, R3 1.5 0.089 0.97 0.94 7.42 0.1 6.88 1.34 9.9
T4 L4, R4 1.5 0.089 0.97 0.94 5.36 0.5 9.53 2.5 9.0
C = 8.2
12
So for C an average value of 8.2 could be determined (Standard deviation 0.8). It is interesting in
this respect to mention the value of the same coëfficiënt, found by Van der Meer in his formuia for
damage levels caused by the already mentioned H2% (see section 5.1), being 8.7.
5.3 The validity of superposition

Proof of the fact that superposition of damage, caused by different levels of hydraulic loads, is
ailowed, can be detected from the results of test sections 3 and 4 after series T1 and 12. Figure
12 shows testresults for sections 4 after T1 and T2.
7 —- O T1 , L4, par = 5,7 5

6 — • T1 , R4, par= 5,7
5 — A T2, L4, par = 7,5
!i p^.
4 —- A T2 , R4, par = 7,5 ft/* 0 ^
S = 1,95\/n >vN
\ k^-^
A
6
10
1
•
\J
1
INS<N
'*
'ft
A
'
^ - 4 R
f- ~ ~- — "v^
'2'fH
Ie? \ s = 0,4\/n
' i
S }
m
1! o
< I
t I
10
0
1 I
s; _
1 2 3 4 5 6 7 8 910 1 2 3 4 5 6 7 8 910 2 2 3 4 5 6 7 8 910 3
— n
Figure 12. Test results for sections 4 after T1 and T2
The lower graph gives the relationship between damage levels S and the number of passages n,
for both the left as well as the right side of section 4, as a result of test series T 1 . After this series,
the left side has been put in its originai state. The right side however, has been left in its distorted
state. Before the start of T2, both profiles have been measured to act as a reference for the
damage caused by the wave loads, generated in this series. In order to get the upper graph of
figure 12, the damage levels measured on the already distorted right side have been corrected by
adding the damage level, found after T1 (S = 4.6). Also the number of passages has been
corrected, by adding a fictive number of passages, that has been derived from the damage level
after T 1 , as though it has been the result of test series T2 (n = 5.5). As it appears, the thus
corrected damage levels, measured during test series T2 at the right side, agree rather well with
the damage levels measured at the left side. (The same applies to testsections 3, where an identical
correction has been carried out for the damage levels measured at the right side, see figure 10.)
Based upon this agreement, one may draw the conclusion that an already existing distortion of the
riprap protection, casu quo damage, caused by a specific number of wave loads, does not affect
13
the damage, caused by a number of higher wave loads. This implies that the effects of different
combinations of wave loads and numbers of waves, on the damage to a riprap bank protection are
independent of each other, which ailows the superposition of the separate, corresponding damage
levels.
6. Conclusions
By means of tests in a physical scale model, the assumption has been verified that Van der Meer's
description of a mathematical model that predicts damage to stone armour layers of breakwaters,
attacked by wind waves, is also applicable to riprap bank protections along navigation channels,
surcharged by secondary ship waves. The assumption proved to be correct.
A statistical analysis of a set of secondary ship waves showed that these have an energy density
spectrum which is in shape similar to that of wind waves. However, the statistical distribution of
secondary ship waves differs from the Rayleigh distribution, which is valid for wind waves.
The analysis indicated that the use of the maximum secondary wave height, generated by each
passing ship, and the corresponding length and period as governing parameters in the predictive
model for damage to riprap bank protections, is appropriate.
The predictive mathematical model that could be derived from the test results and earlier research
by Van der Meer, describes the development of damage as a function of the properties of the riprap
armour layer, the maximum secondary wave height and length for each ship passage, and the
number of passages.
Test results indicated that damages caused by different combinations of wave loads and number
of waves, casu quo passages, may be superimposed.
The mathematical model describes the behaviour of riprap armour layers of bank protections along
navigation channels. It can be used to predict the maintenance state of the protection as a function
of time.
To be able to do this, shipping on the fairway has to be well known, in terms of number of
passages per year and average sailing speed and course of each specific type of ship that
frequently sails the channel.
References
J.W. Van der Meer, 1988
Rock slopes and gravel beaches under wave attack (Thesis); Delft Hydrauiics Communication No.
396
F.C.M. Van der Knaap, D.H. Wilkens, H.J. Regeling, 1992

Models to predict the behaviour of riprap bank protections; report of model investigations (in Dutch)
H.J. Verhei/, M.P. Bogaerts, 1988

Ship waves and the stability of armour layers protecting slopes; Paper presented at the nineth
International Harbour Congress, Antwerp
14
publication no. 464
Environment-friendly bank protections
To be presented at the International Symposium

on Environmental Hydraulics, Hong Kong,
December 16-18, 1991
R.E.A.M. Boeters, H J . Verheij and M. van der Wal
November 1991
v^^.^:? delft hydraulics

ENVIRONMENT-FRIENDLY BANK PROTECTIONS
R.E.A.M. Boeters, H.J. Verheij* and M. van der Wal*
Rijkswaterstaat, Road and Hydraulics Engineering Division,

Delft, The Netherlands
* Delft Hydraulics, Emmeloord, The Netherlands
ABSTRACT: Embankments along navigation canals and rivers usually require a

protection against the erosive forces of currents and waves. Up to now, in most
cases riprap, concrete blocks or sheet-pilings have been applied. However, the
increased public awareness about more environment-friendly solutions has been
leading to greater attention for design criteria for protective structures
allowing the presence of vegetation. After surveying several types of environment-
friendly protections the recently-developed design rules for two particular
structures will be presented. In addition, we will also present a summary of the
expected outcome of research projects being carried out presently.
1. INTRODUCTION
In the Netherlands much attention is focussed on building environment-
friendly protections along embankments of navigation canals, levees and summer
dikes along rivers [1]. These protections are designed in such a way that plants
and animals can grow and reproduce, whereas the protection itself blends nicely
into the surrounding environment. Thus, they do not only provide a habitat for
plants and animals, but they are also attractive for recreational purposes. The
environment-friendly structure should resist the occurring hydraulic forces
induced by currents, wind-waves and ship-induced water motions. In some situations
the vegetation will have a protective function. In that case the sterns and leaves
are expected to reduce the hydraulic loads (active role of the vegetation), while
the roots are expected to improve the stability of the subsoil against erosion
(passive role of the vegetation). However, in other situations the vegetation will
only play an aesthetical role. In this paper results will be presented of physical
model tests and other studies dealing with the current environmental bank
protection research in the Netherlands.
2. TYPES OF ENVIRONMENT-FRIENDLY PROTECTIONS

Several types of environment-friendly protections can be distinguished. The
difference between the various types is determined by the function of the
vegetation in relation to the hydraulic loads (aesthetical or protective role). In
Fig. 1 the following environment-friendly types of bank protection are shown
schematically:
A) Embankments without any protection, resulting in the most natural environment
for animals and plants. This type can only be applied in the case of low
hydraulic loads and/or a very large available construction width.
B) Partially protected embankments, i.e. the lower part of the embankment is
protected against (ship-induced) currents, the upper part is protected by the
vegetation. This type is suitable for smaller canals and rivers in which
relatively low hydraulic loads occur. The vegetation then has to be temporarily
protected against incoming ship or wind waves during the first stage after
construction.
C) Structures with a shallow pool behind a fixed protection. In the pool plants
and animals find their habitat. The water in the pool is refreshed through gaps
in the fixed protection. The fixed protection absorbs all the hydraulic loads.
This type may be applied in the case of medium to heavy hydraulic loads and
sufficiënt construction width.
-1-
D) Protections through which plants can grow and develop. For these structures
block mattresses can be used, but also gabions and specially designed
protections consisting of open riprap layers. This type of structure is
suitable in the case of little available width and medium hydraulic loads.
Fig. 1 Types of environment-friendly bank protections
In the Netherlands various research projects are carried out yet with respect to
environment-friendly bank protections. Among these protections, one of them
concerns a combined system, such as cellular concrete revetment blocks allowing
vegetation. Another project deals with the water exchange between a navigable
fairway and a shallow pool with vegetation behind a fixed protection along the
fairway. In the next sections design criteria for both structures will be pres-
ented.
3. EROSION OF SEDIMENT THROUGH CELLULAR BLOCKS

A promising structure with respect to providing a natural habitat for plants
and animals' living along the water, blending into the surrounding landscape is the
application of cellular concrete revetment
blocks, particularly in the case of limited
available space for constructing a bank pro-
tection (Fig.2). The cells in this type of
structure allow vegetation to grow through
or between hard elements. The blocks are
permeable enough to prevent the occurrence
of high uplift pressures while, on the other
hand the individual blocks support each
other against severe wave attack. The ero-
sion of the cell fillings is decisive for
the stability of the slope protection and the Fig. 2 Cellular concrete block
growth of the vegetation in the first stage. revetment
-2-
As soon as vegetation has established itself, erosion will be prevented by the
roots and leaves of the plants.
In order to establish design criteria studies have been carried out at Delft
Hydraulics in a small-scale model [2]. The objective of the project was to assess
the rate of erosion of soil material out of the cells. Fig. 3 shows the cellular
concrete block revetment schematically.
U.IO
1 sand,
wave height 0.25m
clay,
-""" ... wave height 0.40m
1 0.08
*^
'S
1 0-04
0 10 20 30 40 50
numbsr op ship passages (xtOO)
cell fllllng
Fig. 3 Schematized cellular block Fig. 4 Typical test results of

erosion of cell fillings
For wave loads attacking around the undisturbed water level, the following
design equation was found after the tests:
OJ
Y/G = c H log (N + 1) (1)
in which: c = coëfficiënt (nr0-5) , G = cell width (m), Y - maximum erosion (m), H=
wave height (m), N = number of waves or ship passages (-). The coëfficiënt c has a
-0.2
value of 0.7 for non-cohesive cell fillings and 0.25 for cohesive fillings
with a very low compaction (p - soil fraction with particles smaller than 0.002
mm). A specific result is presented in Fig. 4.
The main conclusion after having carried out the research project is that
cellular blocks can be applied along inland waterways. A safe design approach is
to apply blocks with a cell width G less than half the block thickness D for
cohesive and non-cohesive cell fillings and to allow maximum wave heights not
larger than 0.4 m. In the case of a cohesive cell filling values for the wave
height up to 1.0 m are allowed, whereas the cell width G may be less or equal to
D. The erosion of cohesive material predicted by equation (1) can be reduced
considerably by increasing the compaction of this material in the cell.
Meanwhile, a prototype inventarization is carried out to the establishment of
vegetation on cellular block revetments. For that purpose real test embankments
are examined. The results of this inventarization will become available soon.
4. WATER EXCHANGE BETWEEN POOL AND WATERWAY

An environment-friendly bank protection along inland waterways that is of ten
applied consists of a fixed protection, border ing a shallow pool with varying depth.
This shallow pool provides a habitat for all kinds of plants and animals improving
the appearance of the bank protection while befitting the surrounding environment.
Fig. 5 shows several varieties of the structure mentioned above.
-3-
(a) (b) (c)
Fig. 5 Protective structures with a shallow pool
The fixed protection may consist of a sheet-piling (a) , a riprap or block mat-
tress protection (b) or a combination of the structures mentioned before (c) . A very
important aspect of this type of structure is the exchange of water between waterway
and pool in order to keep the quality of the water in the pool on a sufficiënt level.
This exchange is possible by wave-overtopping, but more importantly, it can be real-
ized by gaps in the fixed protection. Another advantage of the application of gaps
is that fish and other animals use these gaps to get in and out of the pool.
The distance between these apertures as well as the effective area largely
govern the rate of water exchange, and, consequently, fluctuations of the water level
in the pool. In order to make a proper design, in terms of the dimensions of the gaps
and their distance, a numerical model has been developed.
The model is based on the following assumptions:
* Water exchange is possible by the water level depression caused by a passing ship.
* During the passage time of a ship, the water level in the waterway is lowered
and due to the head difference between pool and waterway, water flows out of the
pool.
* After the ship's passage, the original water level in the waterway is restored,
resulting in an inverse flow, since the water level in the pool has decreased
during the stage of outflow.
* The process can be described using the one-dimensional partial differential equa-
tions, that describe non-stationary flow in open channels, combined with suitable
discharge-head relations for the applied gaps.
In Fig. 6 the system of waterway, permeable protection and shallow pool has been
schematized. Each gap is defined as some kind of weir or spillway, and the distance
embankmem s lope L between two gaps is divided into
sections of equal length Ax. The
1T 1 1 1T 1 1 1T y -i i i i i i i i i 1
p*o t I 1 1 1 1 1 1 1 1
water level depression caused by
1 1 > 1 1 1 t t 1
1
1
the passing ship starts at t - 0 s
1 t 1
1
1
1
I i r" ""i i i i i at the first aperture, and
lf\ f I xed p r o t e c t 1 o n \
T
at - L/V.^, s at the next (V^ is
^<.Cio \ 0 a p L H,Ct3 the speed of the ship in m/s). For
-*- —»•
each section the water movement is
snip
govemed by the equations for con-
servation of mass and conservation
of momentum. By applying dis-
charge -head relations at the boun
Fig. 6 Schematization of the waterway, daries (both gaps) , the set of equa
permeable protection and shallow tions can be solved.
pool
The outcome of the model is govemed by a selected volume of water that will be
exchanged during the passage of a characteristic vessel. This volume is expressed
relative to the original volume of water in the shallow pool. This criterion is
dependent on the number of characteristic ships that pass the structure daily. For
a waterway with a relatively low intensity of shipping traffic, about ten per cent
of the original volume Is regarded to be sufficiënt.
In the graphs below, some results of the model are shown. They apply to the
-4-
depicted situation. In this case, the relation between discharge and head for a
sharp-crested weir has been applied at the gaps [3]:
Q(t) =4ca1/2g mbH(t) (2)
in which: c, = coëfficiënt for sübmerged flow (-), m = discharge coëfficiënt (-), b

= gap width (m) , g - acceleration due to gravity (m2/s) , Q(t) - discharge (m3/s) , H(t)
= water level (m).
water lava I -
Oeprosslon 0 . 4 0 i
gop C"«'01t 0.50 m,
w l d t h 0 . 5 0 , 0 . 7 5 , 1.0 m r e c p Q
20 30 40 50
distance between gaps, L (m)
Fig. 7 An example of the use of the derived mathematical model
The example shows that in order to exchange a relative volume of about ten per
cent per passage, for the smallest gap a distance of 20 meters is necessary, while
for the largest gap a distance of 40 metres suffices.
By using the model, the proper distance between the gaps and their geometry can
be assessed, related to a selected level of water movement in the shallow pool.
Besides structures using all sorts of sheet-piling, the model can handle riprap or
block mattress protections with lowered crests (Figure 5 (b)) or pipes (culverts) as
aperture, simply by adapting the discharge-head relations. Also the water exchange
via a riprap protection without any gap can be calculated. In this situation water
flows through the pores between the stones, which can be described using Forchhei-
mer's equation for combined turbulent and laminar flow through stones [4].
The establishment and development of vegetation in the pool, and the use of the
pool as habitat or breeding place by fish or other animals depends largely on the
quality of the water and the intensity of water movements inside the pool. With the
model described above, a proper design of the structure and the pool can be realized.
5. RECENT AND FUTURE DEVEL0PMENTS

The trend to environment-friendly sol-
utions for protective structures has original profils
resulted in many other research projects in
the Netherlands. For example, at this moment
research is carried out in order to develop
conceptual models for the behaviour of non
or less protected banks. The influence on a
non-protected embankment of the hydraulic
loads under normal daily conditions will be
a gradual erosion of the original profile
(Fig. 8) eventually resulting in an equili-
brium profile. Small-scale model tests to- Fig. 8 Schematized gradual erosion
gether with theoretical considerations of a non-protected embankment
-5-
should allow to develop a mathematical model describing the time-dependent erosion
of the embankment profile. The results of the studies will also be important for
determining maintenance strategies.
Another research project deals with the properties of geotextiles as a filter
layer in which the question is: does geotextile allow vegetation? If the openings in
a particular geotextile do not allow the vegetation to grow through the geotextile
vegetation will not develop in the way it was expected. Obviously, a compromise has
to be found between the traditional requirements, viz. water permeability and soil
tightness, and the properties of the geotextile. Ten geotextiles are investigated of
which one is made of the natural fibre jute.
planting width
Fig. 9 Vegetated berm as Fig. 10 Cross-section of a test section with reed

bank protection for wave damping tests
A third project deals with the damping of waves by plants like reed or bulrush
on a horizontal berm at the embankment (Fig. 9). For this purpose a special facility
was built at Delft Hydraulics next to a wave basin, in which in fourteen test sec-
tions reed and bulrush on a substrate of sand or clayey sand are growing in the open
air (Fig. 10). During the growing season and in wintertime waves will be generated
40 in the basin and the wave damp ing caused by vegeta-
tion will be studied. In addition, the influence of
the roots of the plants on the prevention or retar-
iU
dation of the substrate will be studied as well.
Results of earlier tests are presented in Fig 11.
20 The results of all the research projects
described above will be included in the mathemat-
10
ical design program DIPRO (DImensioning PROtec-
tions) [5]. This program yet contains the design
formulas for traditional bank protections. In the
2 4 6
near future it will also allow the determination of
wktth of vegetation (m) stable environment-friendly protections. DIPRO will
Fig. 11 Previous results of then be of particular interest for consultants,
wave damping charac- contractors and management and maintenance author-
teristics of reed ities of waterway systems.
REFERENCES:
[1] Pilarczyk, K.W., Havinga, H., Klaassen, G.J., Verheij, H.J., Mosselman, E. and
Leemans, J.A.A.M. (1989). "Control of bank erosion in The Netherlands. State-of-
the-art", 3rd Nat. Conf. on Hydr. Eng. of ASCE, New Orleans.
[2] Klein Breteler, M. , Laboyrie, J.H. and Verheij , H.J. (1988). "Erosion of sediment
through cellular blocks applied as slope protection along coasts and inland
waterways", Proc. Symp. on Mode11ing Soil-Water-Structure Interactions, Delft,
The Netherlands.
[3] Chow, Ven te (1959). "Open Channel Hydraulics", McGraw-Hill.
[4] Forchheimer, P. (1930) "Hydraulik", Teubner Verlag, Leipzig.
[5] Delft Hydraulics (1991). "DIPRO, a computer aided design system for bank
protections along navigation fairways", Delft, The Netherlands.
-6-
Aquatic Botany, 41 (1991) 333-352 333
Elsevier Science Publishers B. V., Amsterdam
Direct and indirect effects of wave attack on

helophytes
Hugo Coopsa, René Boetersb and Henk Smitc

"Laboratory ofAquatic Ecology, Catholic University, Toernooiveld, 6525 ED Nijmegen. Netherlands
b
RoadandHydraulicsEngineeringDivision, P.O. Box 5044, 2600 GA Delft, Netherlands
Hnsütute for Inland Water Management and Waste Water Treatment, Van Leeuwenhoekweg 20,
3315 AVDordrecht, Netherlands
{Accepted 14 March 1991)
ABSTRACT
Coops, H.. Boeters, R. and Smit, H., 1991. Direct and indirect effects of wave attack on helophytes.
Aquat.Bot.. 41: 333-352.
Along the shorelines of larger water bodies, wave action affects littoral emergent vegetation directly
through mechanical action on plants, and indirectly by interference with the plant's environment. A
study was conducted to reveal the effects of wave exposure on four helophyte species, Phragmites
australis (Cav.) Trin. ex Steudel, Typha anguslifolia L., Scirpus lacustris L. and Scirpus maritimus L.
in the northern Delta area in the Netherlands. The species differed with respect to the wave exposure
of their locations. Typha angustifolia was restricted to sheltered sites, whereas the other species were
present on sheltered and moderately exposed shores.
In a gradiënt from sheltered to exposed sites, wave exposure appeared to be related to soil texture
and organic matter content. In a cultivation experiment with the four species on transplanted soil
originating from three sites (exposed, intermediate and sheltered), only the growth of T. angustifolia
was affected by the soil types. At the sites, transplanted rhizomes of the helophyte species studied did
not survive the wave attack in the growing season at the exposed site. while at the completely sheltered
site these plants produced the highest biomass.
Wave attack on the sites within the growing season was determined by means of the wave-forecast-
ing model HISWA (Hindcast Shallow Water Waves). Maximal wave heights and orbital velocities
were concluded to be key factors in the decreased growth rates of plants at exposed sites.
INTRODUCTION
Several types of emergent vegetation can be distinguished in the littoral

zone of freshwater areas. Much of this variation within a water body can be
attributed to variation in exposure to waves at the shoreline (Pearsall, 1920;
Bernatowicz and Zachwieja, 1966; Spence, 1967; Hutchinson, 1975). In eu-
trophic waters, the emergent macrophyte belt is mostly dominated by one or
a few species of large, monocotyledonous helophytes. Such dominants are, for
example, Phragmites australis (Cav.) Trin. ex Steudel (common reed ),£«>-
0304-3770/91/S03.50 © 1991 Elsevier Science Publishers B.V. All rights reserved.

334 H.COOPSETAL.
pits lacustris L. (bulrush), Scirpus maritimus L. (sea clubrush) and Typha

angustifolia L. (narrow-leavedcattail).
Exposure to waves affects plant establishment, growth, survival and disper-
sal directly as well as indirectly (Jupp and Spence, 1977; Keddy, 1982; Spence,
1982; Chambers, 1987). Direct exposure effects are due to the mechanical
action of wind and waves, resulting in physical stress imposed on plants and
sediments (Raspopov et al., 1988). Wave attack on the shoreline is spatially
differentiated along water bodies, depending on size and morphology, shore-
line depth profile and the wind climate in the region (Geis, 1985). Indirect
influences of exposure are sometimes less conspicuous, but may nevertheless
be of great importance to emergent plants. Soil textural and chemical char-
acteristics are influenced by soil elutriation and partiele sorting through the
action of waves on the shore (Hakanson and Jansson, 1983).
Both direct and indirect processes are likely to affect plant growth condi-
tions. In various studies, distribution patterns of helophytes along water bod-
ies have been attributed to indirect wave effects (Burgermeister and Lacha-
vanne, 1980; Weisner, 1987; Breen et al., 1988). The question remains,
however, to what extent the occurrence of helophyte species is limited by these
factors. The distribution patterns of different species along water bodies may
varv due to the conditions they require (Wilson et al., 1985; Breen et al.,
1988).
Along large waters and intensively used waterways, wave attack on the
shoreline may be strong. The wave energy regime may limit the development
and maintenance of emergent vegetation (Binz-Reist, 1989; Stark and Dienst,
1989; Sukopp and Markstein, 1989). A reed belt itself may be of protective
value to the adjoining land due to wave extinction and sediment stabilization
(BacheandMacaskill, 1981; Bonham, 1983).
This paper discusses the relationship between wave exposure and the growth
of helophytes. The direct and indirect effects of wave exposure on four species
were investigated and analysed by measurements at the borders of natural
stands of P. australis, S. lacustris, S. maritimus and T. angustifolia in the Har-
ingvliet-Hollandsch Diep basin in the Netherlands, a transplantation exper-
iment in an exposure gradiënt at the island of Tiengemeten, and a sediment
transplantation experiment to separate between the direct and indirect effects
of wave action.
STUDY AREA
The Haringvliet-Hollandsch Diep-Biesbosch basin is situated in the

southwestern part of the Netherlands (Fig. 1). Prior to 1970, it formed one
of the estuarine outlets of the Rivers Rhine and Meuse. Since 1970 it has been
closed off from the sea, while a large sluice works at the seaward end of the
basin regulates the water discharge of the area. At present, the basin is almost
OF WAVE ATTACK ON HELOPHYTES
41°15 41°20
Fig. 1. Geographical situation of the northem delta in the Netherlands, and the study sites on
the isle of Tiengemeten and its surroundings.
1
336 H.COOPSETAL.
Wat er-level (cm +NAP)

(a)
100
80
k. '
60 -ijui ifflLi i
40
"WNF lÜü
20 I
i I
1 2 3
Period
WIND DIRECTION (b)
N N
N
\ ,
7|
3
< k
\
7
s S
110%
© ® © period
MAXIMAL WIND SPEEC)

N N N
Ni/\ \/
/\
\ /
\
llOm s~ 1
7fs ;
/
© (D © period
Fig. 2. (a) Water levels in the central Haringvliet basin (hourly recordings of the station at
Middelharnis. 51 °46'N, 4° 11'E) during the 1989 growing season. Long-term mean high water
level + 65 cm NAP, long-term mean low water level + 34 cm NAP. (b) Frequency of occurrence
of wind directions and maximum wind speeds from each wind direction (Geulhaven station
51°53'N. 4°19'E. 20 km inshore) between 19 May and 16June (Period l).between 17June
and26Juiy (Period 2), and between 27 July and 6 September (Period3).
a still water body for most of the time, although varying water levels occur
due to river discharge, sluice operation and reduced tides.
The brackish coastal marsh vegetation gradually disappeared after 1970,
while the reed marshes which were on higher grounds were invaded by terres-
trial plant species. Nowadays, the lowest semi-emergent vegetation zone is
situated at mean high water level. Serious erosion and shoreline retreat has
occurred below this zone (Fergusonand Wolff, 1983).
The experimental sites were located at the western end of the Island of
Tiengemeten in the central Haringvliet area ( 51 ° 45' N, 42 ° 15' E), where the
shoreline is largely without vegetation after the disappearance of the
S. maritimus belt in the 1970s. There is a gradual slope towards deep water.
The mean low water line is at + 34 cm NAP (where NAP is the Normal Am-
DIRECT AND INDIRECT EFFECTS O F WAVE ATTACK ON HELOPHYTES 337
sterdam Level (Dutch Standard level)) and the mean high water line is at
+65 cm NAP. An irregular, longer-period fluctuation is superimposed on the
small tidal fluctuation, as shown in Fig. 2 (data from the 1989 growing
season).
The bare slope of the west shore of Tiengemeten (Site 1) consists of a sandy
substrate alternating with layers of clay; it has been subjected to heavy ero-
sion, mainly during winter storms. Towards the northeast (Site 2), the shore
becomes more sheltered due to the prevailing southwest and south winds; the
soil has a higher silt and organic matter content than Site 1. Sheltered condi-
tions are found in creeks that are connected with the open water (Site 3),
where sedimentation and accumulation of organic matter take place.
MATERIALS AND METHODS
Exposure ofhelophyte stands
The occurrence of stands of T. angustifolia, S. lacustris, S. maritimus and

P. anstralis bordering on the open water in the Haringvliet-Hollandsch Diep-
Biesbosch basin was surveyed in 1989.
The effective fetch (Fe) of localities was calculated for each of 12 compass
segments
F e (/) =
/ r (/) + F(/+30")cos(30 !3 )+F(/+60 3 )cos(60°)+F(/-30 o )cos(30 !: )+/'(/-60 = )cos(60')

l+2[cos(30=)+cos(60°)]
where F(i) is the fetch in the direction of/ degrees (measured as the distance
of open water to the opposite shore). As a measure of wave exposure, mean
weighted effective fetch (MWEF) was calculated as follows
Ï 1 T O
M W E F . "
12
where W-t is a weighting factor (ranging from 0.44 (120 °) to 1.9 5 (240 °)) to
correct for the relative long-term occurrence of winds from each direction, as
well as for the mean wind speed from each direction (data from Wieringa and
Rijkoort, 1983). Measurements were carried out in a large number oïPhrag-
mites stands in the area, as well as in all encountered stands of T. angustifolia
and of both Scirpus species. The occurrence of interspecies differences in ex-
posure (expressed as MWEF) was tested by non-parametric analysis of vari-
ance (Kruskall-Wallistest).
338 H. COOPS ET AL.
Effects ofwaves on sediment composition
At 20 locations situated at a level of +25 cm NAP on the west shore of

Tiengemeten, samples were taken from the top 5 cm of the soil. Each sample
consisted of five cores, taken at random from a 1 m2 quadrat, using a 10 cm
diameter PVC pipe. The samples were analysed for soil size fractions (by
sieving and sedimentation) and organic matter content (percent loss on ig-
nition of dried soil). The silt fraction (soil partiele size less than 16 /zm) and
organic matter content were related to the calculated measure of exposure
(MWEF) for each location. Associations between exposure and soil para-
meters were assessedby calculating Kendall's T (rankcorrelation coëfficiënt).
Wave attack on sites in an exposure gradiënt
Data on wind directions and wind speeds, obtained from the nearest me-
teorological station at Geulhaven (51 °53'N, 4° 19'E), and water level data
from Middelharnis (51°46'N,4°H'E) were used for the calculation of wave
height frequency distributions at the experimental sites at Tiengemeten by
means of the wave estimation model HISWA (Hindcast Shallow Water Waves;
Booij et al., 1985). This numerical model was developed for the estimation
of wave parameters in coastal and estuarine areas and lakes, for stationary
wind, bottom topography and currents. Modelling wave attack on the emer-
gent vegetation is complicated by the unpredictable nature of wind phenom-
ena over a time interval. In this case, it was further complicated by fluctuating
water levels, the effects of which on wave height are apparent in the zone near
the shoreline. Therefore, various wind direction, wind speed, and water level
classes were incorporated and assumed to be stationary over periods of 1 h.
Wave parameters were calculated using two HISWA models successively. The
first model used a rather coarse grid that contained a large portion of the
topography of the Haringvliet near Tiengemeten. It provided the boundary
conditions for the second model that used a much finer grid, describing the
coastline of the west part of the island.
Significant wave height (i.e. the mean of the highest one-third of all waves
during a period with stationary wind conditions) and wave period were cal-
culated for grid points that matched the experimental sites as closely as
possible.
To construct frequency distributions of hindcasted wave heights for each
experimental site over the experimental periods, the following steps were car-
ried out.
(a) The water levels measured at Middelharnis were averaged over periods
of 1 h. The water levels at Middelharnis are considered to be representative
of those at the boundaries of the HISWA models.
DIRECT AND INDIRECT EFFECTS OF WAVE ATTACK ON HELOPHYTES 339
(b) Combined wind speed and wind direction values were treated in the
same way.
(c) Four water level classes were distinguished: 30-50 cm +NAP, 50-70
cm +NAP, 70-90 cm +NAP, and over 90 cm +NAP.
(d) For each class of water levels, a matrix was constructed containing
classes of wind speed (with a width of 4 m s~') and wind direction (sections
of 30°), of which each cell contained the number of occurrences of a partic-
ular combination of wind speed and direction. Cases in which the wind was
coming from the directions 75-135° (approximately east to southeast) were
disregarded, since there was no fetch for these directions.
(e) Calculations were made for each matrix cell using the class middle val-
ues, which resulted in values for the significant wave height at each of the
experimental sites. The significant wave heights represent the wave condi-
tions for periods of 1 h. Stationary wind conditions can safely be assumed
because 1 h periods were used. Furthermore, the waves can fully develop dur-
ing 1 h, which means that they were not influenced by the waves generated in
the previous hour.
(f) Finally, frequency distributions of the significant wave heights were
composed by combining the results of the HISWA rans and the number of
occurrences of each set of water level, wind speed and wind direction values.
Site 3 was located at a shallow creek. Owing to the grid size of both models,
a proper schematization of this situation was not possible, so Site 3 could not
be incorporated in the calculations. Wave heights at Site 3 can be assumed to
be negligible, which was affirmed by the observation of very small wave heights
during strong winds from the west.
The HISWA model also allows calculation of the root mean square value of
the maximum orbital velocities near the bottom, which occur under the cal-
culated wave conditions. The orbital velocity describes the movement of water
particles during the passage of a wave. It is strongly dependent not only on
wave height, but also on wave period. In the case of non-broken waves, drag
forces on the sterns of plants are proportional to the square of the orbital ve-
locity. The orbital velocity has been calculated in order to compare the effects
of the highest waves at each site during the periods inbetween the measure-
ments of the plants at Tiengemeten (Period 1: 19 May-16 June; Period 2: 16
June-26 July; Period 3: 26 July-6 September).
Transplantation experiments
Four helophyte species were used in a transplantation experiment. They

were planted directly at the three sites on Tiengemeten and in soil from these
sites which had been put into containers in an open air lay-out near the labo-
ratory at Dordrecht.
Rhizome sections (about 10 cm long, cut off rhizomes bearing roots and
340 H.COOPSETAL.
vegetative tillers; Fig. 3) of P. australis, T. angustifolia, S. lacustris and S.

maritimus were collected from stands in the Haringvliet-Hollandsch Diep
area in May 1989 and stored for 1 week in a shallow pond until the day of
planting, either in the field or in the containers.
At the Tiengemeten sites, rhizome sections were planted on 19 May. Four
rhizomes were buried just below the soil surface in the centre of each
0.5 mx0.5 m plot. The plots were positioned in a row at the +25 cm NAP
level perpendicular to the shoreline slope. At each site 40 plots, consisting of
ten plots per species, were randomly assigned to the row.
At the end of three periods, on 16 June (Period 1), 26 July (Period 2) and
Phragmites australis
Scirpus maritimus
Typha angustifolia
Scirpus lacustris
Fig. 3. Rhizome sections with vegetative tillers of T. angustifolia. P. australis, S. lacustris and 5.
maritimus (t=tillers, r=rhizome).
6-7 September (Period 3), the numbers of tillers and tiller lengths of each
plot at the field sites were determined. Tiller length was defined as the dis-
tance between the stem base and the tip of the highest leaf. The dry biomass
of each tiller was calculated from its length using regression equations derived
from plants harvested separately from the study plots (Table 1). These bio-
mass values were used in calculating relative growth rates (RGR) in the sam-
pling intervals.
RGR=(ln W2-In Wx)lt (day- 1 )
where Wx is grams of biomass at the start and W2 is grams of biomass at the
end of an interval of / days.
In the laboratory, 36 containers (801) were layered with 20 cm of surface
soil from the sites and brimmed with river water. Eight rhizome sections were
planted per container; each species-substrate combination was replicated three
times. The containers were randomly positioned in the open air at the Dor-
drecht laboratory site from 20 May until 1 September 1989.
Harvesting was carried out on 1 September (3 containers of each species
for each sediment type) and 6-7 September (five plots of each species at each
site at Tiengemeten). Because the root systems of the original clusters were
completely entangled below ground, clusters were not separated into their
TABLE1
Regression equations for the non-destnictive estimation of dry weight of above-ground biomass (XV,
g) by measurement of tiller length (L, cm) determined monthly for the four species
Species Month Regression equation n R2
Phragmites australis May In ( W) =2.223 In (L)- 9.124 21 0.871

June In (W) = 1.848 In (L)- 7.505 17 0.919
July In ( « 0 = 2.152 In (L)- 9.229 25 0.877
August In ( W) =2.047 In (L)- 8.552 20 0.848
Scirpus lacustris May In ( W) = 1.894 In (L)- 8.461 26 0.856

June In(WO = 2.0541n(L)- 8.289 30 0.941
July In (W) = 1.878 In ( L ) - 8.343 32 0.582
August In (WO = 2.309 In (L)-10.776 24 0.862
Scirpus maritimus May In (W) = 1.9941n ( L ) - 8.107 26 0.943

June In (WO = 1.694 In ( L ) - 6.701 20 0.882
July In (W) = 1.888In (L)- 7.519 26 0.897
August In (WO = 1.750In ( L ) - 6.591 27 0.908
Typha angusiifolia May In (W) = 2.888 In ( L ) - 12.414 15 0.973

June In ( « 0 = 3.309 In (L)-15.029 12 0.856
July ln(WO = 2.9191n (L)-12.881 14 0.610
August In ( W) = 3.219 In (L) - 14.535 20 0.658
342 H.COOPSETAL.
separate sections. Plant contents were dug up, and the roots and rhizomes
were separated from soil by rinsing and sieving. The dry weight biomass of
above-ground and below-ground parts was determined after drying at 105°C
for 48 h.
Differences in final above-ground and below-ground biomass, and num-
bers of tillers, between treatments in the container and field experiment were
tested by the least significant difference at the 5% level.
RESULTS
Exposure ofhelophyte stands
The four helophyte species had significantly different Qf2=48.0, P< 0.001)
distributions of exposure (MWEF) at their locations. Figure 4 shows the dis-
tribution of the species over six MWEF classes (class interval 500 m). In
particular, maximum exposure of T. angustifolia was limited to the lowest
fetch class: MWEF does not exceed 370 m. P. australis was also predomi-
nantly present at the lowest fetch class. The range of MWEF, however, ex-
tends to very high values. Both Scirpus species occurred more commonly at
higher MWEF classes.
Phragmites australis Typha angustifolia
é Scirpus lacustris
é Scirpus maritimus
11
I II III IV V VI
A
I II
w
III IV V VI
MWEF-class
Fig. 4. Distribution of stands of T. angustifolia («=26, median= 107 m), P. australis (« = 324,
median = 227 m), S. lacustris(«=25, median = 538 m) andS. maritimus (« = 27. median=534
m) in classes of MWEF in the northern Delta area, surveyed in 1989. The MWEF classes are
500 m wide.
ïnfluence of wave exposure on sediment composition
Sites 1, 2 and 3 at Tiengemeten form an exposure gradiënt. Differences in

wave exposure between the sites were reflected by differences in soil silt frac-
tions and organic matter (OM) contents of the sediments. The sheltered Site
3 (MWEF= 10 m) represents an environment where the deposition of silt
and organic matter takes place (13% OM, 49% silt). Site 2 (MWEF=918 m)
is intermediate (9% OM, 20% silt). Site 1 (MWEF = 1342 m) is an eroding
shore (6% OM, 18% silt).
Based on 20 samples from the western shore of the island of Tiengemeten,
the soil organic matter content of the shore zone topsoil showed a significant
negative rank correlation ( T = 0 . 4 3 , P<0.01) with MWEF; the sediment silt
fraction and MWEF were also negatively correlated ( T = 0 . 4 9 , P<0.01)
(Fig.5).
Wave attack on experimental sites
The wind and water Ie vel conditions in the Haringvliet area over the 1989
growing season are represented in Fig. 2. Differences in wave exposure be-
tween Sites 1 and 2 are to a large extent due to the exclusion of south and
30 f A
«
20 -
•#
10 * **
1 10 100 1000
B
60
I
50 •
40 •
*•
30
20 •
10
10 100 1000
MWEF (m)
Fig. 5. Relationship between MWEF and (A) organic matter content (loss on ignition), (B)
soil size fraction less than 16 ,«m of the shoreline topsoil at +20 cm NAP on the western shore
of the isle of Tiengemeten.
344 H.COOPSETAL.
southwestera fetches at Site 2 (Fig. 6(a)). The data from the Geulhaven sta-
tion show the wind was blowing from the directions south to southwest for
12% of Period 1. During Periods 2 and 3, these percentages were 17 and 42,
respectively.
For the growing season as a whole, there was a distinct difference in wave
attack between Sites 1 and 2 (Fig. 6 (b)). Differences in wave attack between
the three experimental periods were also apparent. At Site 1, the trend in av-
erage wave height was upward, at Site 2 it was downward (Fig. 6(c)).
Table 2 shows the maximum significant wave heights for Sites 1 and 2 in
site 1
300
240
180
Wind speed (m s- ) 1
10 120 Wind direction (degrees)
60 | | above 25cm
*% I | 20-25cm
15-20cm
10-15cm
5-10cm
below 5 cm
site 2
Hs (cm)
Hs (cm)
30 \ -
20
a
10"]
o-'
18 180
Wind sp«;ed (m s- 1 ) Wind direction (degrees)
Fig. 6. Hindcasted wave heights calculated using HISWA. (a) Dependence of wave height at
Sites 1 and 2 on wind direction and wind speed when the water level is +60 cm NAP. (b)
Frequency of occurrence of wave heights at Sites 1 and 2(19 May-6 September 1989); class
middle values of wave heights are given. (c) Frequency of occurrence of wave heights in three
periods (1: 19 May-16 June 1989; 2; 17 June-26 July 1989; 3: 27 July-6 September 1989) at
Sites 1 and 2.
( b ) Freauency
1000
900
800 ^ M Site 1. I-U13.3cm
700 I I Site 2, H» 8.2 cm
600
500
400
300
200
100
0 £U3_
(C)
500
Site 1
450
• H Period 1. H>11.6cm
400
I I Period 2 . H-12.7cm
350
WMX Period 3, H"15.1em
300
250
200
150
100
50
0
(C)
500
Site 2
450
^ H Period 1, H»9.9cm
400
I I Period 2, H«8.8cm
350
Period 3, H-6.4cm
300
250
200
150
100
50
0
2 7 12 17 22 27 3 2 3 7 4 2 4 7 52cm
W a v e heights
Fig. 6. Continued.
each period, as well as the number of occurrences. In addition, the maximum

orbital velocities at the bottom which occurred under these wave conditions
are given for each period.
Transplantation experiments
In the container experiment, sediments from the three sites affected neither
biomass nor the number of tillers at 1 September of P. australis, S. lacustns
and S. maritimus at a significant level (Fig. 7). In T. angustifolia, differences
346 H.COOPSETAL.
TABLE:
Significant wave height (i/ s ,max) and number of occurrences of the highest waves, and the orbital
veiocities (Ub) under these waves at Sites I and 2 in 1989, hindcasted by HISWA
Period 1 Period 2 Period 3

(19May-16June) (16June-26July) (26 July-6 September)
Hs, max Times Ub i/ s , max Times Ub H%, max Times Ub

(cm) (ons-') (cm) (cms"') (cm) (cms"')
Site 1 38 6 50 32 11 51 53 2 59
Site 2 31 2 41 28 1 29 32 5 38
Aboveground Beiowground Number of

biomass biomass shoots
g DW- -N-
Typha 150
4 0 0
angustifolia 100
200 50
r-r—n —0
Phragmites 150
australis 400
100
200 50
0 r+r—Tl -rh —0
Scirpus 150
lacustris 400
100
200 +T+ 50
0 —0
Scirpus 150
maritimus 400
100
200 50
0 —0
1 2 3 1 2 3 1 2 3 Site
Fig. 7. Mean above-ground and below-ground biomass, and numbers of tillers ( = shoots) of T.
angustifolia, P. australis. S. lacustris and S. maritimus grown on transplanted sediments in con-
tainers in the 1989 growing season. Sediments originate from exposed (1), intermediate (2)
and sheltered (3) sites at Tiengemeten, respectively. Each bar represents three containers
(mean±S.E.).
in biomass production on the various sediments occurred, showing signifi-

cantly higher values at Site 3 than at those from the other two sites.
In Fig. 8, biomasses and numbers of tillers of the Tiengemeten plantings at
6-7 September are shown. At Site 1, all the plants were washed away at the
end of the growing season. At Site 3, the plants showed high values for all
parameters. Differences between 5*. maritimus plants on Sites 2 and 3 were
not significant. The other species at these two sites differed significantly in all
parameters.
Table 3 shows mean relative growth rates of the biomass per site, obtained
by regression from the tiller length-tiller dry weight biomass relationship, in
the time intervals between measurements. No significant differences in plant
growth were found between Sites 2 and 3 in early summer (Period 1); in the
subsequent period, only T. angustifolia growth was significantly lower at Site
2 than at Site 3. In late summer (Period 3), the growth of all species at Site I
was affected; at Site 2, the biomass per plot decreased, while it was still in-
Aboveground Belowground Number of

biomass biomass shoots
gDW-
Typha 150
4 0 0
angustifolia 100
200
50
— 0
Phragmites
400 150
austraüs
100
200 50
O — 0 J=£L
Scirpus 150
lacustris 400
100
200 50
0 —0
Scirpus 150
maritimus 400
100
200 50
0 r^l- -L^Ül —0
1 2 3 1 2 3 1 2 3 Site
Fig. 8. Mean above-ground and below-ground biomass, and numbers of tillers (=shoots) of T.
angustifolia, P. australis. S. lacustris and S. maritimus on Sites 1 (exposed). 2 (intermediate)
and 3 (sheltered) at Tiengemeten at the end of the 1989 growing season. Each bar represents
five replicate plots (mean ± S.E.).
348 H.COOPSETAL.
TABLE 3
Relativegrowth rates (day"', ±standard deviation, n=5) of T. angustifolia, P. australis, S. lacustris

and 5. maritiinus at Sites 1, 2 and 3 in three periods of the 1989 growing season (Period 1:19 May-
16 June; Period 2: 16 June-26 July; Period 3: 26 July-6 September). The relative growth rates are
based on biomasses calculated from relationships between tiller length and dry weight of tiller biomass
Sitel Site 2 Site 3

(exposed) (intermediate) (sheltered)
Typ/ia angustifolia
Period 1 (-0.160) 0.056 ±0.029 0.052 ±0.027
Period 2 0.037 ±0.011 0.064 ±0.019
Period 3 -0.057 ±0.046 0.026 ±0.014
Phragmites australis
Period 1 -0.031 ±0.064 0.064 ±0.024 0.059 ±0.019
Period 2 -0.094 ±0.045 0.026 ±0.018 0.029 ±0.008
Period 3 -0.026±0.014 0.028 ±0.008
Scirpus lacustris
Period 1 -0.024 ±0.022 0.073 ±0.012 0.092 ±0.012
Period 2 -0.098 ±0.016 0.037 ±0.007 0.048 ±0.012
Period 3 -0.026±0.014 0.010± 0.006
Scirpus maritiinus
Period 1 -0.020 ±0.044 0.101 ±0.009 0.091 ±0.030
Period 2 -0.101 ±0.030 0.021 ±0.006 0.022 ±0.009
Period 3 -0.007 ±0.024 0.027 ±0.013
creasing at the sheltered Site 3. Maximum relative growth rates were found in
early summer; the general decrease in relative growth rate follows the growth
pattern of the species with a biomass peak in summer.
DISCUSSION AND CONCLUSIONS
The distribution of stands of the four species P. australis, T. angustifolia, S.

lacustris and S. maritimus within the northern delta area is related to wave
exposure. Most of the investigated stands are remnants of former brackish
tidal marshes in the area, so the study demonstrated persistence on locations
rathef than establishment of helophytes. In view of the increased wave energy
at the shoreline as a result of the reduction of the area occupied by the inter-
tidal zone in the past, and the long period since this reduction occurred (20
years), the present distribution of species probably reflects the species' ranges
of wave exposure accurately. The degree of wave exposure to which a helo-
phyte stand is subjected can be expressed on a relative exposure scale, in this
case MWEF. Other measures of exposure, however, would be equally suit-
able, because different measures are usually well correlated (Keddy, 1982).
The value of MWEF gives only a relative description of the wave exposure at
a site, since it does not involve the unpredictable, temporally varying nature
of wave attack. Apart from the wind climate, the actual wave exposure of the
sites depends on the size and morphology of the water body, shoreline profiïe,
water depth fluctuations, and the presence of wave-damping structures such
as submerged macrophytes. Wave action can be quantified by measurements
in situ, or by calculating wave heights with the aid of wave-forecasting models.
By using HISWA, it was possible to relate exposure to actually occurring wave
heights. Maximal wave forces exerted on the sterns can be attributed to the
drag caused by the movement of water under waves. Forces under unbroken
waves are proportional to the square of the orbital velocity. Under broken
waves, wave forces are related to the wave propagation speed (Ippen, 1966).
Drag forces under waves result in physical stress on the bottom and on sub-
merged plant structures.
Erosion and sedimentation in the shoreline zone are essentially dependent
on the occurrence of wave heights above or below a certain limit. At lake
shores, unconsolidated fine sediments are transported and eroded at current
velocities of 20-50 cm s - ' (Hakanson and Jansson, 1983). Orbital velocities
under the highest waves at the exposed sites in this study are within this range.
Erosion of sediments at exposed sites, as well as the transport and deposition
of sediments and organic detritus at sheltered sites, represent indirect effects
of wave action (Keddy, 1982, 1984;Spence, 1982). They act on the environ-
ment of the plant mainly via the rooting substratum, by affecting the available
nutrients, organic matter and soil texture. In several studies, emergent macro-
phyte occurrence was shown to be hardly affected by a wide range of these
conditions.
Scirpus lacustris grows on fine silty to sandy soils, and on mineral as well as
organic substrates (Seidel, 1955; Haslam et al., 1975). Phragmites australis
is found on soils poor to rich in nutrients and on a variety of textures (Has-
lam, 1972,1973; Haslam et al., 1975; Burgermeister and Lachavanne, 1980).
Granéli (1985) found no strong effect of nutriënt additions on the standing
erop of Phragmites. On soils with very low fertility, however, biomass pro-
duction is reduced and shoot morphology changes (Haslam, 1972); the bio-
mass reduction is more pronounced in Typha species than in Phragmites
(Szczepanska and Szczepanski, 1976). In addition, in oxbow Iakes in the
Netherlands T. angustifolia is restricted to sites consisting of non-cohesive
mud (van Donselaar-ten Bokkel Huinink, 1961).
The results of the experiment in which the species were grown in containers
on sediments originating from the sites showed that the production of P. aus-
tralis, S. lacustris and S. maritimus did not respond clearly to the sedimenta-
tion effects of wave exposure within the gradiënt under study, whereas direct
wave attack does induce a significant response. In contrast, T. angustifolia
350 H.COOPSETAL.
seemed to be affected strongly by both factors, which explains its restricted

occurrence.
The exposure of the experimental sites covers the distribution of the species
at extremely exposed sites (Site 1, MWEF= 1342 m), at moderately exposed
sites (Site2,MWEF=918m) and at sheltered sites (Site3,MWEF=10m).
Ho wever, comparison of the results at the experimental sites with the effects
of wave action on established stands in the region is not possible because of
the susceptibility to waves of stands in the first years after planting. The poor
performance of the plantings at Site 1 coincides well with the near absence of
stands of the four species at locations with MWEF> 1500 m. At these loca-
tions, like Site 1, maximal orbital velocities greater than 50 cm s~' within the
growing season occur. At Site 2, all species performed better, but no autumn
or winter storm occurred during the experiment. Average wave heights at Site
1 showed an increase during the growing season, those at Site 2 a decrease
(Fig. 6(c)). This can partly be ascribed to an increase of winds coming from
the west and the southwest (Fig 2 (b)). During the first period, average wave
heights at Sites 1 and 2 show a minor difference. Yet, the relative growth rates
of all species at Site 1 are negative, while those at Site 2 are positive and com-
parable to those at Site 3. Average wave height does not pro vide enough in-
formation about the impact of wave attack. Maximum wave height and the
number of occurrences of the highest waves appear to be more relevant
parameters.
The largest growth differences between plants at Site 2 and Site 3 occurred
in Period 3. As shown in Table 2, relatively high wave attack on Site 2 oc-
curred in Period 1 and, more frequently, in Period 3. The absence of signifi-
cant differences in growth rate between the plants at Site 2 and at Site 3 in
Period 1 suggests that the young tillers are less affected by waves in spring
than the fully developed tillers are in autumn. Another explanation is that the
capacity for regrowth of damaged plants decreases during the course of the
growing season.
The results indicate that wave height is not the only decisive parameter in
determining the mechanical effects of waves on plants. Both wave height and
wave period contribute to the orbital velocity, and to the forces exerted on
submerged plant parts. The frequency of occurrence of critical forces proba-
bly determines the overall effect on plant growth.
Wave breaking has not been considered so far. Clearly, wave breaking does
occur more frequently at Site 1, owing to higher waves. As a result of break-
ing, increased turbulence and much higher velocities occur under waves. The
differences in plant performance between Site 1 and the more sheltered sites
are probably partly due to this phenomenon.
Wave exposure influences plant growth on the shore directly by the effects
of wave force, which damages and uproots plants, and transports seeds and
plant parts. In this study, only one season's growth was examined. Within-
DIRECT AND INDIRECT EFFECTS OF WAVE ATTACK. ON HELOPHYTES 351
season variations in the wind climate may affect growth and establishment;
the timing of wave exposure probably interacts with the plant's develop-
mental stage. In late autumn and winter, when wave attack is usually most
severe due to strong winds and higher water levels, the impact on the plants
will be reduced since no living above-ground parts remain. In that case, sur-
vival will probably depend on the stability of the soil surface.
ACKNOWLEDGEMENTS
This study was part of a project initiated by the Institute for Inland Water
Management and Waste Water Treatment, and the Road and Hydraulics En-
gineering Division of the Department of Transport and Public Works in the
Netherlands.
We thank Peter Boekestein, Yvonne Godefroy, technical personnel of the
Department of Transport and Public Works for their assistance during the
field work, and Plony Cappendijk, Peter Veelo and Bas Kranenborg for their
help in determining wave heights. We are indebted to Gerard van der Velde,
Egbert van Nes, Geert-Jan Verkade and Hans Janssen for their valuable com-
ments on the manuscript.
REFERENCES
Bache. D.H. and Macaskill, LA., 1981. Vegetation in coastal and streambank protection. Land-
scape Plann.. 8: 363-385.
Bernaiowicz. S. and Zachwieja, J., 1966. Types of littoral found in the lakes of the Mazurian
and Suwalki lakelands. Ekol. Pol., 28: 519-545.
Binz-Reist, H.-R.. 1989. Mechanische Belastbarkeit natiirlicher Schilfbestande durch Wellen.
Wind und Treibzeug. Veröffentlichungen des Geobotanisches Institutes der ETH, Stiftung
Rübel, Zürich, no. 101, 536 pp.
Bonham, A.J.. 1983. The management of wave-spending vegetation as bank protection against
boat wash. Landscape Plann., 10: 15-30.
Booij, R , Holthuijsen. L.H. and Herbers, T.H.C., 1985. The shallow water wave hindcast model
HISWA. Tech. Univ. Delft, Dep. Fluid Mechanics Rep. No. 6-85.52 pp.
Breen, C.M., Rogers, K.H. and Ashton, P.J., 1988. Vegetation processes in swamps and flooded
plains. In: J.J. Symoens (Editor), Vegetation of Inland Waters. KJuwer, Dordrecht, pp. 223-
247.
Burgermeister, G. and Lachavanne, J.-B., 1980. Les macrophytes du Pfaffikersee. Ber. Schweiz.
Bot. Ges., 90:213-243.
Chambers, P.A., 1987. Nearshore occurrence of submersed aquatic macrophytes in relation to
wave action. Can. J. Fish. Aquat. Sci., 44: 1666-1669.
Ferguson, H.A. and Wolff. W.J., 1983. The Haringvliet-Project: the development of the Rhine-
Meuse estuary from tidal inlet to stagnant freshwater lake. Water Sci. Technol., 16: 11-26.
Geis, J.W., 1985. Environmental influences on the distribution and composition of wetlands in
the Great Lakes Basin. In: H.H. Prince and F.M. D'Itri (Editors), Coastal Wetlands. Lewis,
Chelsea, pp. 15-31.
352 H.COOPSETAL.
Granéli. W., 1985. Biomass response after nutriënt addition to natural stands of reed. Phrag-
mites australis. Verh. Int. Ver. Limnol., 22: 2956-2961.
Hakanson, L. and Jansson. M., 1983. Principles of Lake Sedimentology. Springer. Berlin. 316
pp.
Haslam, S.M.. 1972. Phragmites communis Trin.; Biological Flora of the British Isles. J. Ecol..
60:565-610.
Haslam. S.M.. 1973. Some aspects of the history and autecology of Phragmites communis Trin.,
a review. Pol. Arch. Hydrobiol., 20: 79-100.
Haslam, S.M., Sinker, CA. and Wolseley, P.A., 1975. British water plants. Field Stud.. 4: 243-
351.
Hutchinson. G.E.. 1975. A Treatise on Limnology, III. Limnological Botany. Wiley. New York,
660 pp.
Ippen. A.T., 1966. Estuarine and Coastline Hydrodynamics. McGraw-Hill, New York. 744 pp.
Jupp, B.P. and Spence, D.H.N., 1977. Limitations of macrophytes in a eutrophic lake, Loch
Leven. II. Wave action, sediments and waterfowl grazing. J. Ecol., 65:431-446.
Keddy, P.A., 1982. Quantifying within-lake gradients of waveenergy: interrelationships of wave
energy, substrate partiele size and shoreline plants in Axe Lake, Ontario. Aquat. Bot., 14:
41-55.
Keddy, P.A., 1984. Quantifying a within-lake gradiënt of wave energy in Gilfillan Lake. Nova
Scotia. Can. J. Bot., 62: 301-309.
Pearsall. W.H., 1920. The aquatic vegetation of the English lakes. J. Ecol. 8: 163-201.
Raspopov, I.M.. Slepukhina, T.D., Vorontzov. F.F. and Dotzenko, O.N.. 1988. Wave effects on
the bottom biocoenoses in the Onega Lake bays. Arch. Hydrobiol.. 112:115-124.
Seidel. K., 1955. Die Flechtbinse, Scirpus lacustris L. Die Binnengewasser, 21, 216 pp.
Spence. D.H.N., 1967. Factors controlling the distribution of freshwater macrophytes with par-
ticularreference to the lochs of Scotland. J. Ecol., 55: 147-170.
Spence, D.H.N., 1982. The zonation of freshwater plants. Adv. Ecol. Res., 12: 37-125.
Stark. H. and Dienst. M.. 1989. Dynamics of lakeside reed beits at Lake Constance (Untersee)
from 1984to 1987. Aquat. Bot., 35: 63-70.
Sukopp. H. and Markstein, B.. 1989. Changes of the reed beds along the Berlin Havel. 1962-
1987. Aquat. Bot.. 35: 27-39.
Szczepanska. W. and Szczepanski. A., 1976. Growth of Phragmites communis, Typha latifolia
and Typha angustifolia in relation to the fertility of soils. Pol. Arch. Hydrobiol. 23: 233-
248.
Van Donselaar-ten Bokkel Huinink, W.A.E., 1961. An ecological study of the vegetation of three
former river beds. Wentia, 5: 112-162.
Weisner. S.E.B.. 1987. The relation between wave exposure and distribution of emergent vege-
tation in a eutrophic lake. Freshwater Biol., 18: 537-544.
Wieringa, J. and Rijkoort, P.J., 1983. Windklimaat van Nederland. KNMI, De Bilt. 263 pp.
Wiison. S.D.. Keddy, P.A. and Randall. D.L.. 1985. The distribution of Xyris difformis along a
gradiënt of exposure to waves: an experimental study. Can. J. Bot.. 63: 1226-1-30.
PENETRATION THROUGH GEOTEXTILES BY SHOOTS, ROOTS AND RHIZOMES OF THE COMMON
REED (PHRAGMITES AUSTRALIS (CAV.) TRIN. ex STEÜDEL): a pilot study
ir. Eric A.H. Ivens

Road and Hydraulic Engineering Division
P.O. Box 5044, 2600 GA, Delft, The Netherlands
Keywords: bank vegetation, Phragmites, geotextiles.
ABSTRACT
Geotextiles are widely used in bank protection works as a filter construc-
tion to prevent the basic material from being washed out. However, to
fulfil this function geotextiles through their specific design may affect
bank vegetation growth. A pilot study was carried out to obtain an indica-
tion of the penetration possibility of ten different types of geotextiles
by an emergent macrophyte (common reed).
From the experiments can be concluded that:
(1) only cocomat, jute, filament mat and woven tape showed a considerable
number of shoot penetration.
(2) rhizome penetration and shoot penetration in significant numbers
occurred in the same types of geotextiles.
(3) a considerable number of root penetration occurred in the types of
geotextiles in which no rhizome penetration of importance occurred.
(4) it is assumed that the number of yarns per 10 cm affect the penetra-
tion possibility in woven fabrics.
KURZFASSÜNG
Geotextilien werden heutzutage vielfach als Erosionschutz der Uferböschun-
gen eingesetzt. Die charakteristischen Eigenschaften des Geotextils können
aber die Entwicklung einer Ufervegetation beeinflussen. Ziel der Studie war
es, die Durchwachsung der zehn verschiedenen Geotextilien durch Röhricht zu
untersuchen.
Die Ergebnisse, lassen folgende SchluBfolgerungen zu:
(1) Sprossen durchwachsen am besten Kokosmatte, Jute, dreidimensionale
Nylon Strukturmatte und Bandchengewebe.
(2) Geotextilien, die die Sprosse durchwachsen lassen, garantieren auch
ein gutes Durchwachsen der Rhizome.
(3) Die Durchlassigkeit für Wurzeln war bei solchen Textilen am besten,
die nur eine geringe Durchdringung der Rhizome gezeigt haben.
(4) Es wurde angenommen da/3 die Fadenzahl pro 10 cm die mögliche Durch-
wachsung des Gewebes beeinf lulït.
A INTRODUCTION
At present geotextiles are used on a large scale in geotechnical enginee-
ring, e.g. bank protection works. The word geotextiles means that they are
essentially made from fibres and filaments. There is a great variety of
geotextiles but the most common ones are woven and non-woven fabrics.
When incorporated in bank revetments, geotextiles are principally used as
filters to prevent soil particles from being washed out, with drainage and
separation being secondary functions. Geotextiles can also be used as a
protective layer.
Their chief advantage is that they are able to deform with the structure
and still remain continuous. This is essential when they provide a filte-
ring function (Veldhuijzen van Zanten 1986).
Applied as a filter in bank protection works geotextiles are laid directly
on the subsoil and must guarantee the soil tightness of the structure and
be sufficiently water permeable to allow groundwater to pass through.
Soil tightness reguires a limited pore size of the geotextile. However, in
ecologically sound bank protection works in the Netherlands one of the main
objectives is to provide bank vegetation development (Anonymus 1990). To
penetrate through the geotextile the vegetation reguires a large pore size
of this geotextile, especially rhizomatous riparian plants important in
erosion control against boat wash.
The application of natural fibre mats and webbings, which have been used by
man since early times, is an alternative raeans but these materials have
several disadvantages, including low durability and unreliable quality.
Generally they are only used for short term projects or where it is ad-
vantageous to have a biodegradable fabric (PIANC 1987). Therefore, in
protection works where a persistent filter is needed geotextiles made from
polymeres are most commonly used, but through their specific design they
may affect bank vegetation growth.
Poschadel (1991a,b) noticed that mainly woody species (Salix spec.) and a
few herbs were able to penetrate through a non-woven by roots with a mean
diameter of 1-2 mm. Also Bureau Waardenburg (1990) had noticed that Salix
alba, S. viminalis, S. cinerea, Alnus glutinosa and Phragmites australis
penetrated through woven geotextiles by roots. However, it is hard to teil
whether other species had penetrated because of the difficulty to determine
in the field. As geotextiles are usually covered with loose stone, concrete
blocks or soil one has to lift the entire construction to detect penetra-
tions.
The aim of this paper is to examine the penetration possibility of ten
different types of geotextiles by a clonal eraergent macrophyte (common
reed).
B MATERIALS AND METHODS

The ten different geotextiles used in this study were selected as they (i)
represent the range of available types of geotextiles and (ii) are the most
commonly used geotextiles in the Netherlands. Table 1 gives a prescription
of these ten types (further on described as A, B,...J) and their main
characteristics.
In March 1991 42 glass containers were placed outdoors in a small artifi-
cial pond (4,5 x 13 x 1 m ) . Each container consisted of two parts. The
larger part had a size 50 x 50 x 30 cm, the smaller part 50 x 50 x 10 cm.
After filling with soil and placing the rhizome segments in experiment A
these two parts were fit together with a geotextile in between them (fig.
1). Two holes at the base provided enough moisture supply.
On 12 April 1991 two experiments (both in duplicate) were started:
(A) ten containers were filled with a 30 cm sandy-clay soil mixture and
were each covered with a different geotextile. In each container 5 rhizome
segments were planted under the geotextile to test penetration by shoots.
These containers are further on described as A3, A4,...,J3, J4.
(B) another ten containers were filled with a 30 cm sandy-clay soil
mixture. In each container 5 rhizome segments were then placed in a 5 cm
layer of soil covering the geotextiles. In this experiment the penetration
by roots and rhizomes was tested. These containers are further on described
as Al, A2,...,Jl, J2.
Two glass containers (ref. 1 and ref. 2) each containing 5 rhizome segments
and no geotextile served as references.
Each rhizome segment consisted of two internodes and three buds. The
rhizomes were collected in April at a commercially reed stand in the
Veluwemeer, the Netherlands.
The containers from experiment A were placed on wooden platforms in the
pond so that all rhizome segments were on the same level above the water
level and thus had equal moisture conditions. The water level was kept 5 cm
below the rhizome segments during the experiment. All other plants than
reed growing in the containers were removed to avoid interference. Aphids
were manually removed from the reed leaves.
On 9 October 1991 -after 26 weeks- the geotextiles were removed from the
containers. Throughout the experiment the emergent shoots were monitored
and marked weekly. Measurements of shoot number, shoot diameter, shoot
length, number of rhizomes, rhizome diameter, root number, root diameter
were made at the end of the experiment. The numbers of shoot-, root- and
rhizome penetrations were also recorded on 9 October 1991.
C RESULTS
After three weeks of treatment shoots had not emerged in any of the 42
containers. The first emerging shoots were observed in week 4 (references
and exp. B) and week 5 (exp. A ) . Due to a cold June with mean temperatures
of 12,7 °C only a few new shoots were detected in week 8-10. In experiment
A shoot development stopped in week 23. In experiment B development of
short, thick shoots continued till the end of the experiment (week 26). In
addition, containers A3, F3, 13, 14, J3 and J4 showed no above-ground
growth after 26 weeks.
Density of reed shoots was higher in experiment B containers than in
experiment A (tab. 2 ) . The greatest density was recorded in the reference
containers. No shoots have emerged through geotextile I and J in experiment
A where as in C, E, H and D the largest numbers of shoots were recorded. On
the other hand in experiment B the largest shoot numbers were noticed in
geotextile F, G and I.
In general shoot diameter in experiment A was comparable to the references
(tab. 2). A maximum diameter of 6 mm was recorded in geotextile H. In
geotextile A, B and F a maximum diameter of 4 mm and in G a maximum of 3 mm
was determined. Shoot diameter in experiment B was slightly larger than the
references with a maximum of 9 mm in geotextile B and I.
In general the mean shoot height was greater in experiment A than in
experiment B, because a lot of short, thick shoots developed in experiment
B (tab. 2). The maximum height recorded was 140 cm in both experiments.
Flowering shoots were noticed in experiment B only.
The rhizome and root diameter are shown in table 2. In geotextile A, B, C,
D, H and J the mean diameter of newly formed rhizomes below the geotextile
was larger in experiment B than experiment A. In geotextile H a maximum
rhizome diameter of 14 mm was recorded with a mean diameter of 11 mm. In
geotextile F a maximum diameter of 3 mm was determined. Root diameters
recorded in both experiments were comparable to reference containers (tab.
2).
Figure 2a and 2b show the numbers of shoot- and rhizome penetration
respectively. In geotextile I and J no penetrations occurred during the
experiment and in A, F and G a small penetration number was counted in
comparison with geotextile C, D, E, H and B. As can be seen in figure 2a
and 2b shoot- and rhizome penetration in significant numbers occurred in
the same geotextiles.
Root penetration was observed in all tested geotextiles (fig. 2c), especi-
ally in geotextile G and I. Only a few number of root penetration was
recorded in geotextiles C, E, H, F and J.
D DISCUSSION
Geotextiles can be classed as one, two or three dimensional (Veldhuijzen
van Zanten 1986). Two dimensional products include the fabric sheet-like
materials as used in this study: wovens, non-wovens and other types like
composites and filament mats. The pore sizes of these materials are ranging
from 0.06 mm (geotextile A) to 1.5 mm (geotextile B) (tab. 1). In table 2
is shown that the mean shoot-, rhizome- and root diameter in the reference
containers is 3.5 mm, 8 mm and 1.7 mm respectively, which means that they
are larger than the pore sizes of the tested geotextiles. Feyen (1971)
reported that roots are able to exert forces of 10-20 kg/cm2 on their
environment. Once these roots have penetrated the geotextile they can
stretch the material by secondary growth. Poschadel (1991a,b) and Bureau
Waardenburg (1990) both made a vegetation survey at sites where non-wovens
and wovens respectively were incorporated in bank protection works. Most
species they recorded were annual herbs and grasses, which have only thin
(up to 2mm) roots and regenerate by seed. Also a few woody species were
noticed: Salix spec. and Alnus glutinosa. These species do Have secondary
growth and are therefore able to stretch the geotextile material and
penetrate by roots with a diameter up to 25 mm. Common reed however, is a
rhizomatous species like most other riparian plants e.g. Phalaris arundina-
cea, Scirpus lacustris subsp. lacustris, Iris pseudacorus, Typha spec,
Acorus calamus and Carex spec.. These species regenerate mainly by vegeta-
tive reproduction. In bank protection works where geotextiles are applied
it is of importance for these riparian species to penetrate through the
material by rhizomes and reach the underlying subsoil. However, monocots
like most aguatic plants, macrophytes and grasses do not have secondary
growth and stretching the material is much more difficult for them.
As is well-known reed shoots can penetrate through asphalt pavement, but
not much is known about the ability to do so. Further study is needed at
this point.
The geotextiles tested in this study can be classified in four groups:
(i) group 1 is formed by the natural webbings cocomat and jute. Both shoots
and rhizomes are able to penetrate easily through these materials, due to a
relatively large deformation capability and pore size (tab. 1). It is
striking however, that just a few root penetrations was recorded as much
thicker rhizomes (up to 14 mm) penetrated easily. This is easily explained
by the fact that most roots develop from the rhizomes which have already
penetrated the material. This phenomenon was also seen in geotextile E
(group 2) and D (group 3).
(ii) group 2 is formed by the filament mat (geotextile E). Although the mat
is filled with bitumen and therefore horizontal deformation is neglegible a
large number of shoot- and rhizome penetrations was recorded at the end of
the experiment. This can be explained by the high percentage f ree space
(25%) of this material, which makes it easier for sterns and rhizomes to
penetrate through this material, though the maximum rhizome diameter (8 mm)
was evidently smaller than recorded in group 1 (max. 14 mm).
(iii) group 3 is formed by the needlepunched non-woven (geotextile A ) . As
can be seen from table 1 it has a very thight structure with a pore size <
60 jna. This could explain the difficulty for shoots and rhizomes to
penetrate through this material, although one rhizome with a diameter of 12
mm had succeeded to do so. On the other hand, a large number of roots with
mean diameter of 1.5 mm and a maximum diameter of 3 mm was able to penetra-
te through the non-woven. Apparently it is possible to stretch the material
easily up to 1.5 mm. Poschadel (1991a,b) reported the same results from a
survey at sites along the river Aller and the Mittellandkanal in Germany,
where a needlepunched non-woven (with pore size 70-120 pm) was used in the
bank protection work. He noticed that several species, mainly woody plants
(Salix spec.) penetrated through the non-woven by roots with a diameter up
to 2 mm and a maximum of 25 mm. It is thought that this type of geotextile
has a lot of pores which can be easily used by thin roots (< 2mm) to reach
the underlying subsoil.
(iv) group 4 is formed by the wovens (geotextile B, D, F, G, I and J). In

general the woven geotextiles showed a small number of shoot- and rhizome
penetrations in comparison to the above mentioned groups. In geotextile I
and J containers no shoots had penetrated at all throughout the experiment,
although the buds on the rhizome segments have sprouted. The number of root
penetrations however, exceeded those recorded in groups 1-3, because of the
small rhizome penetration number as explained before (group 1). Although
these materials have a large pore size and permeability (see table 1) it
seems almost impossible for shoots and rhizomes to stretch the material as
f ar as they can penetrate through, except for geotextile D and to some
extent B. This could indicate that the penetration ability is negatively
affected by the number of threads per 10 cm, as these are lowest in
geotextiles D and B (tab. 1). The difference between D and B in number of
penetrations could be either explained by difference in type of y a m (tape
versus monofilament) or the fact that B is coated and therefore stretching
is more limited than in case of D. As this pilot study was carried out in
duplicate further research is needed to get statistical evidence at this
point.
The highest density of shoots in experiment B was recorded in geotextiles

F, G and I, while these types of geotextiles also showed the lowest rhizome
penetration number. Apparently resources were allocated towards the above-
ground organs. Hence, flowering shoots were only recorded in experiment B.
Also a considerable number of short (< 5 cm), thick shoots was determined
at the end of experiment B. It is thought that these shoots can store
resources during winter time and start growing again early next season.
As this pilot study lasted just one growing season, it is hard to predict
reed development in the next few years. Experiments lasting more than one
year have to be carry out.
E CONCLUSION
These results show that a well developed reed stand can be expected in
ecologically sound bank protection works where a type of geotextile is
incorporated like the cocomat, jute, filament mat, woven tape and woven
coated monofilament as both shoots and rhizomes are able to penetrate
through these materials. For this reason it is not recommended to apply the
needle-punched non-woven, woven monofilaments and woven/non-woven composite
in ecologically sound protection works.
F LITERATUKE CITED
Anonymus (1990): Milieuvriendelijke Oevers. Project Milieuvriendelijke
Oevers, rapport nr.3, Rijkswaterstaat Dienst Weg- en Waterbouwkunde, Delft
Bureau Waardenburg bv. (1990): Begroeiing van betonblokkenmatten. Project
Milieuvriendelijke Oevers, rapport nr.16, Rijkswaterstaat Dienst Weg- en
Waterbouwkunde, Delft
Feyen, J. (1971): Wortelontwikkeling op zand- en leembodems. Dissertatie
Katholieke Universiteit, Leuven
PIANC (1987): Guidelines for the design and construction on flexible
revetments incorporating geotextiles for inland waterways. Supplement to
Permanent International Association of Navigation Congresses, Bulletin
No.57, Brussels
Poschadel, K. (1991a): Vegetationsentwicklung über geotextilen Filtermatten
der Firma Naue-Fasertechnik unter besonderer Berücksichtigung der Durchwur-
zelung. Planungsbüro Ökologie, Lehrte/Ahlten
Poschadel, K. (1991b): Untersuchungen an einem Hochstauden-Röhricht-Saum,
der über einer geotextilen Filtermatte wachst. Planungsbüro Ökologie,
Lehrte/Ahlten
Veldhuijzen van Zanten, R. (1986): Geotextiles and geomembranes in civil
engineering. A.A. Balkema/Rotterdam/Boston
Figure 1. Glass container with a geotextile in between (photograph by R.

Spaan, April 1991).
Figure 2. Number of penetrations through the ten types of geotextiles (A-J)

by (a) shoots, (b) rhizomes and (c) roots.
Table 1. Types of geotextiles and their main characteristics.

var. = variable; pe » polyethylene; pet » polyester; pp = polypropylene
pa = polyamide.
Table 2. Summary of the data collected at the end of experiment A and B.

Geotextiles A1,A2...Jl,J2 refer to experiment B, geotextiles A3,A4—J3,J4
to experiment A. Ref. 1 and 2 are the reference containers.
shoot penetration
40
35
no
30
25
ZO--
15-•
10-•
i™i"i^+"4»*^*»f^f iiBiMiBi
i—>ÎÎÎÎÎ—i—i—i—i
A3 A4 83 34 C3 C4 03 04 E3 E4 F3 F4 G3 G4 H3 H4 13 14 J3 J4
type of geotextiIe
rhizome penetration
no .
25
i
0 ll,]*l*l l * " l ^ l " l " l " l K l t—1 h"t*1 1 (•(••»
Al A2 B1 82 C1 C2 01 02 E1 E2 F1 FJ G1 G2 H1 H2 11 12 J1 02
type of geotextiIe
root penetration
250
no
200--
100
50
llllllllllllllll
l™t™l»l'"l™l™|M l™l™l™l™l™l»]Ml™l™lMlMl™l
A1 A2 B1 BZ C1 C2 01 02 E1 E2 F1 F2 G1 G2 H1 H2 11 12 J1 OZ
type of geotextiIe
weight tensile elongation pore permea- no, rau
strength at break si ze bility yarns material
<9/m2) (kM/m) (X) (P»> U/m 2 s) (lOcm')
type of geotextile uarp weft uarp weft uarp weft Harp weft
A needlepunched non-woven 360 28 37 45 50 63 93 pet

B uoven coated monofilament 200 29 29 18 18 1500 400 45 45 pet pet
C cocomat 500 0.35 0.35 25 25 var. 20 coco
D woven tape 300 55 55 13 11 250 28 40 50 PP PP
E filament mat 20000 2.5 2.5 60 60 350 45 pa
F woven monofilament 235 45 45 25 25 180 50 145 100 pe pe
G uoven monofilament 210 50 40 30 20 450 600 173 123 pe pe
H jute 400 25 25 5 5 700 200 68 60 jute jute
I uoven monofilament 195 42 37 30 20 1200 700 80 70 pe pe
J woven/non-woven composite 670 55 50 15 10 300 20 60 55 PP PP
geotextile no. mean mean no. no. mean mean max. max.
shoots shoot shoot root rhizome root rhizome shoot rhizome
diameter length penetra- penetra- diameter diameter diam. diam.
(mm) <cm> tion tion (mm) (mm) (mm) (mm)
A 1 38 4.8 56.6 113 1 2 6 8 6

A 2 47 4.9 63.2 118 5 1.6 9 8 12
A 3 0
A 4 4 3.3 63.5 1.7 6 4 7
B 1 44 3.9 61.8 90 16 2.2 8 7 11
B 2 49 4.3 72.4 140 0 2.2 8.7 9 9
B 3 8 3.3 80.0 1.7 7.5 4 9.5
B 4 1 4.0 100.0 0.7 5 4 6
C 1 32 3.8 58.8 60 15 2 10.5 6 12
C 2 54 3.5 63.4 75 30 2.2 9.5 6 12
C 3 7 3.7 53.9 1.5 6 5 6
C 4 38 3.6 71.4 2 9 6 11
D 1 26 3.9 76.5 65 16 1.4 8.8 5 9
D 2 46 3.6 60.5 115 17 1.7 10 5 12
D 3 17 3.9 103.9 2 9 5 12
0 4 9 4.2 102.7 1.5 7 5 8
E 1 17 2.5 40.8 44 8 1.5 4.8 4 7
E 2 28 3.5 61.1 55 14 2 6 5 8
E 3 23 3.5 73.0 2 8.5 5 13
E 4 34 3.1 63.1 2 7.5 5 10
F 1 45 3.4 55.4 15 1 0.7 3 5 3
F 2 69 3.4 50.3 25 0 0.7 6
F 3 0
F 4 1 4.0 101.0 1.0 5.3 4 8
G 1 62 3.7 49.8 250 0 1.7 6
G 2 47 3.8 59.8 130 0 2 6
G 3 3 1.7 45.7 1 4 2 4
G 4 1 3.0 92.0 0.7 5 3 6
H 1 20 4.3 65.9 25 10 2.5 11 6 12
H 2 43 3.9 61.6 50 15 2 11 6 14
H 3 14 3.9 89.9 2 9.5 6 11
H 4 26 2.5 40.2 1.7 5.5 4 8
I 1 61 3.8 65.4 140 0 2 8
I 2 48 3.9 64.4 150 0 2.2 9
I 3 0
I 4 0
J 1 46 3.7 48.8 75 1 2.2 5 5 5
J 2 42 4.6 68.9 65 1 2.2 6 6 6
J 3 0
J 4 0
Ref. 1 74 3.5 58.9 1.7 8 6 10
Ref. 2 67 3.3 52.6 1.5 8 5 12
PART 4
Geotechnical Stability of dikes
Deutekom, J.R. and R.J. Termaat,

Case history of an uplift problem
Dluzewski, J.M. and R.J. Termaat,

Consolidation by finite element method in engineering problems
Koehorst, B.A.N., B.H.P. The, J.A.M. Teunissen, M. Lojan der and

A. Naatanen,
The trial embankments in Vaasa, Finland: a simulation with the
Adachi-Oka creep model
Muijs, J.A. and G.A.M. Kruse,

Spierenburg, S.E.J., J-A.M. Teunissen and B.A.N. Koehorst,

Termaat, R.J.,
Daros on soft soils: lessons from the Dutch Delta Projects
Termaat, R.J.,
Embankments on soft soils; application of the finite element
method
Tonneijck, M.R.,
Berms at the inland side of river dikes
Weijers, J.B.A. and J.B. Sellmeijer,

A new model to deal with the piping mechanism
Bakker, H.L. and J.R. Deutekom,

Probabilistic design method for cofferdams using FEM
J.R. DEUTEKOM» & R.J. TERMAAT*
«Project Specialist, Delft Geotechnics, Delft
ïHead of Geotechnical Research, Ministry of Public Works, Delft
Case History of an Uplift Problem
SYNOPSIS Deformation and related stability problems of dikes in the

western parts of Holland were for many years ascribed to creep and excess
pore water pressures only. However in the spring of 1988 during a high
water period in the river Rhine the so called uplift mechanism was for the
first time measured in practice. It was then recognised that in some cases
the problems have been caused by the uplift phenomenon. For a few years
the pore water pressures and the deformations were observed in a test
site. With the results of the observations a procedure was forroulated for
design and reconstruction.
INTRODUCTION AND LOCATION OF TEST SITE
In the western parts of Holland the ground level lies

usually lower than the mean river level. For that reason
the land is separated from the rivers by water retaining
structures called dikes or levees. These dikes exists
usually of impermeable heavy clay and which are construc-
ted on a relatively impermeable and weak subsoil of peat
and clay. The pore water pressure in the sand beneath
this weak layer is related to the water level of the
river, in case the riverbed is in direct contact with
this sand layer, as illustrated in figure 1.
This paper describes a dike section near the town Bergam-

bacht (20 km east of Rotterdam). After reconstruction 12
years ago the road on top of that dike showed severe
damage due to deformations. Maintenance was necessary
each yea'r. Also the toe of the dikeraovedlandwards.
In the beginning of 1988 the local authorities of that

area decided to investigate the origin of the problems.
Therefore Delft Geotechnics carried out a small investi-
gation programm including cone penetration tests, raea-
surement of pore water pressures and stability analysis
based on Bishops method.
The first result of the investigation was a geotechnical
profile of the concerning dike near the town Bergambacht
(figure 1).
FREATIC LINE
2SSG
CLAY
y PEAT
CLAY TE7TT
SAND
Fig. 1. Geotechnical orofile of test site.
The second result was the pore water pressure distribu-

tion in the profile. In the relatively impermeable top-
layers a water overpressure was measured. The overpres-
sure was at that time ascribed to creep. 'The pore water
pressure in the sand layer showed the expected relat ion
with the tide of the river.
The last result was the Bishop stability analysis showing

the expected low stability.
At that time with the results of the investigation it was

concluded that the probleras were caused by creep in
combination with low stability.
OBSERVATION AND DESCRIPTION OF UPLIFT MECHANISM
After finishing the above raentioned investigation pro-

grarame the measurement of pore water pressure was conti-
nued. In the spring 1988, during an extreme high water
period in the river Rhine, it was suddenly observed that
pore water pressure measurements in the sand reached an
ultiraate value while the river level was still increasing
(figure 2 ) .
At that time it was recognised that the deformation

problems were not caused by creep, but by uplift of the
weak toplayer at the passive zone of the dike. The mea-
surement of the pore water pressures had reached the same
value as the weight of the weak toplayer (figure 3 ) .
During uplift conditions there will be a zone between the
weak toplayer and the sand layer underneath, where no
shear strength is present. Even before the uplift condi-
tion is reached the shear strength will reduce strongly.
This phenoraenon is of great iraportance for the evaluation
of the stability of water retaining structures.
Uplift raeans a change in stress distribution. The shear
stress along the sand layer is reduced and can even be
zero, while the initial horizontal stress in the weak
layer increases (figure 4 ) . This causes relatively large
horizontal deformations of the weak toplayer and tempora-
ry instability of the water retaining structure.
-415-
•416-
-417-
jJPUFT HEAD (UMIJ LEVEL)
"? -418-
Time (hours)
Fig. 2. Observation of the uplift mechanism.
Fig. 3. The uplift mechanism with initial and ultimate

pore water pressures.
UPUFT ZONE
RIVER
DURING UPUFT
L^
SAND
Fig. 4. Shear stresses before and during uplift condi-

tions.
UPLIFT AND HORIZONTAL DEFORMATIONS
After the observation of the ultimate pressure at Bergam-

bacht the Dutch Public Works Department in association
with Delft Geotechnics decided to start a monitor pro-
grarame, which includes monitoring of the horizontal
deformations and pore water pressures for a long period
(figure 5).
FREATIC LINE
JNCLiNOMETER
ISC sample
OBSERVATION
WELL
SAND V-t*?
INITIAL ULTIMATE
Fig. 5. Instruraentation of t e s t s i t e at Bergambacht.

The ma in result of the monitor programs» was the relat ion
between the water level of the river which is related to
the pore water pressure in the sand and the horizontal
deformation of the toplayer of the dike (figure 6 ) .
15 ra»
14 River level
f'
s
\
A J; ho.»
HW,
S 4 A /
o •as s
E 8
V
£ 7 r*-±—
>
V / 0)
Q 6 _J
/ .as ,_
ö 5H
C4 \ ^ i^ N / -
£T
hü.4.
>
O
.ü 3
I2 l
•aj
i— • aa
II -i i
(M
Time (days)
D [1 M-Surface] + [3 M—Surface]
• [5 M-Surface] A [7 M-Surface]
Fig. 6. Relat ion between tide of the river and horizon-
tal deformation of the toplayer of the dike.
Clearly can be seen the increase of horizontal deforma-

tion velocity with increasing water level of the river
and decreasing horizontal deformation velocity with
decreasing water level of the river. When the water level
in the river reduces, consequently the water pressure in
the sand reduces too. At a level of 70X of the uplift
pressure the velocity reduces to zero. The effective
stress between weak toplayer and sand will probably
increase and consequently the horizontal stress in the
weak layer will reduce. This implies that the observed
horizontal deformations of the weak toplayer behind the
dike were caused by uplift and not by creep.
STABILITY AND DEFORMATION ANALYSIS
In case of uplift or nearly uplift the classic stability

analysis can not be used. The slip surface will partly
stay at the boundary of the weak layer and sand. Not only
the stability approach is different, but the deformations
have to be considered too. In the Dutch Guidelines two
methods are recommended (Termaat, 1991).
The first method is based on a non circular slip surface
method, i.e. Morgenstern & Price. In this method firstly
the stability is checked. When the stability is suffi-
ciënt then the lateral compression is deterroined. In
this, only the horizontal part of the critical slip
surface is considered. This horizontal part is analog to
a beara which is loaded by compression. The load boundary
conditions follow froro the stability analysis, actually
the interlamel force in the horizontal part.^For each
lamel the deformation is calculated by Hooke's lav assu-
ming undrained conditions and neglecting the shear force
and the excentricity. The deformations are calculated for
additional changes in river level, because the deforma-
tions due to the normal water level can be equalised
immediately after complet ion of the construction. The
undrained conditions are used for both the stiffness and
the shear strength, because the additional increase of
water level span mostly a short time period.
The second method, which is used for calculation the

effects of lateral compression, is a finite element
approach with a linear elasto-plastic stress strain
reiation. The advantage of this method is that the defor-
mation and stability are coupled (Vermeer, 1990).
In both methods the stiffness is an important parameter.

Special attention was paid to this parameter as described
in the next paragraph.
PERMANENT INCREASE OF HORIZONTAL STRESS AND STIFFNESS

AFTER ÜPLIFT
After each uplift or semi uplift period the stiffness of

the weak toplayer (the compression beam) will increase.
In the field the horizontal stiffness was measured by
Pressuremeter tests. In the laboratory the increase of
stiffness was simulated in an Independant Stress Controll
Cell (ISC-Cell).
Four cubic samples were taken from the weak toplayer

behind the dike. In the ISC-Cell the three directions of
the samples were kept under the same conditions as in the
field during uplift conditions. This means that the
vertical direction of the sample was kept under plain
stress conditions simulating a constant vertical stress.
The horizontal direction parallel to the dike axes was
kept under plain strain conditions simulating a two-
dimensional situation. In the other horizontal direction
the uplift mechanism was simulated by increasing the
horizontal stress under undrained conditions from the
stress before uplift to the stress during uplift. After
the undrained increase of horizontal stress the horizon-
tal deformation was fixed and from that moment drainage
was allowed until the end of consolidation and
relaxation. The test programme had three cycles.
The horizontal stress index observed in the four ISC

tests, after 1, 2 and 3 loading cycles shown a permanent
increase of horizontal stress index ratio after each
uplift period. Figure 7 shows a typical test result. From
this figure the increase of the horizontal stress index
can be derived. In the same figure a square hyperbolie
fitting is given together with the 90X reliability inter-
val.
This has resulted in an avarage formula for the horizon-
tal stress index Kn after N uplift periods, dependant on
the initial horizontal stress index Ko and the number of
uplift periods N:
N
Kn - Ko * ( 1 + (1)
1.96 * N + 4.93
For design purposes the 51 lover bound formula has to be

used:
N
Kn - Ko * ( 1 + (2)
3.18 * N + 8.39
2J0
1 1 1 r
Number of Uplift Periods N (-)

Data from ISC + Avarage Fitting
» 5% Lower Bound A 5% Upper Bound
Fig. 7. Typical test result and square hyperbol ie curve

fitting of stress index ratio and number of
uplift periods.
Field observations with the Pressuremeter shown an index
ratio of Kn/Ko of 1.23 between the uplift zone and the
undisturbed landward zone. This means that in the field 2
uplift periods have occurred according to the avarage ISC
formula .
The results also shown a permanent increase of horizontal

stiffness. The increase of stiffness can be expressed by
the stiffness ratio Gn/Go. The observed ratio's in the
four ISC tests after 1, 2 and 3 loading cycles are plot-
ted in figure 8 together with a square hyperbolic fitting
and a 90ï reliability interval.
Fig. 8. Test results and square hyperbolic curve fitting
of shear stiffness ratio and number of uplift
periods.
This has resulted in an avarage formula for the horizon-
tal stiffness Gn after N uplift periods dependant on the
initial stiffness Go and the number of uplift periods N:
N
Gn - Go * ( 1 + ) (3)
0.24 * N + 2.06
For design purposes the 51 lower bound formula has to be

used:
N
Gn - Go * ( 1 + • ) (4)
0.51 * N + 2.33
With Pressuremeter tests a stiffness ratio of Gn/Go of
1.79 was measured. This means that in the field 2 uplift
periods have occurred according to the avarage formula.
IMPLEMENTATION OF RESULTS IN PRACTICE
With the results of the field observations and the labo-

ratory testing a plan for the reconstruction of the river
dike was made, taking into account the houses at the toe
of the dike. A soil improvement was realised within the
inner side of the dike in combination with a drainage
system (figure 9). This improvement makes sure that there
will be a limit stress increase and consequently a limit
deformation of the weak toplayer. After the reconstruc-
tion the monitoring of horizontal deformations still
continue, üntil now no further deformations were observed
(figure 9).
Fig. 9. Cross section of reconstruction and measured

deformations.
CONCLUSIONS
After the field observations, the laboratory testing and

the lessons from the reconstruction three conclusions
were formulated:
1) Be aware of uplift.
2) Calculation of both stability and deformation are
required under uplift conditions.
3) The increase of stiffness after each uplift or semi
uplift period plays an important role in the history
of the dike and must be taken into account in case of
reconstruction.
REFERENCES
Deutekom, J.R. (1991), "Case History of an Uplift Ero-
blera", Delft Geotechnics, Delft.
Termaat, R.J. (1991), "Daras on Soft Soils: Lessons from
the Dutch Delta Projects", International Coramission on
Large Daros, Vienna.
Vermeer, P.A. (1990), "Users Manual Plaxis", Balkema
Publishers, Rotterdam.
18 - 20 September, 1990
2nd. EURGPEAN SPECIALTY Santander, Spain
CONFERENCE ON NUMERlCAl
METKODS iN
GEOTECHNICAL ENGINEERING
CONSOLIDATION BY FINITE ELEMENT METHOD IN ENGINEERING PROBLEMS
J.M. DLUZEWSKI, Warsaw University of Technology, Poland

R.J. TERMAAT, Department of Public Works, The Netherlands
Abstract
The aim of the paper is an analysis of practical approach towards
engineering consolidation problems with the help of advanced software based
on finite element method. In many cases engineers face the problem of
foundation of a construction on w a t e r - s a t u r a t e d clay-type subsoil, where the
consolidation can takes months or even years. The realistic prediction of the
redistribution of the pore pressure and settlement gives the opportunity to
design the stepwise building up of an embankment on soft soil or the answer
how long we have to wait for consolidation when vertical drains are used.
Though t h e elasto-plastic consolidation has been established and worked out,
in many practical cases the mixed elasto-plastic undrained calculations with
elastic consolidation analysis can be sufficiënt.
The problems analyzed herein where calculated with the help of PLAXIS
package which gives the possibility to mix the elastic, plastic and
consolidation calculations.
1. I n t r o d u c t i o n
Though the three-dimensional Biot's theory of consolidation has been

established in the forties [1], it still gives difficulties in practical
applications. The engineers often use the simpler Terzaghi's one dimensional
consolidation theory or Kjellman's formula for radial consolidation, which
can be successfully applied without the use of a computer, but in the
simplest cases only. In this paper the attention is focused on two practical
problems. The first one is the vertical drainage problem where the sand
drains a r e placed into a clay subsoil. The analysis was done with the finite
element program as well as by hand with the formulas of Terzaghi and
Kjellman. The staged embankment is the second practical problem analyzed
herein. The embankment is built up in three stages. Each increase of
embankment creates plastic zones in the subsoil which disappear during
consolidation process.
2. Basic e q u a t i o n s
The governing equations has been adopted after Biot's self-consistent

theory [1],[2],[3]. The small strain formulation is considered. Darcy's low
for fluid and elasto-plastic behaviour for skeleton is assumed. The
formulation of the consolidation in the finite element method can be found
for example in [4],[6]. For consistency of the paper we rewrite the basic
equations of finite element method for consolidation problem. According to
Terzaghi:
0* = O9 - mp , (1)
where <r is the total stress tensor written in the vector form, C* is the
effective s t r e s s tensor, p is pore pressure and m is equal to unity for
the normal s t r e s s components and zero for the shear stress components. The
constitutive equation takes the following form:
aV = D de, (2)
T
where D represents the constitutive tangent material matrix, de is the

total strain of the skeleton. Applying finite element approximation with the
notation after [5]:
u = N ü, p = N p, C = B u, (3)
where u and f> a r e the nodal displacement and pore pressure vectors, u and p
are displacement and pore pressure, N and N are matrices of the shape
functions, B is the geometrical matrix and starting from the incremental
equilibrium equation,
|BTd<r dv = ( V d f dv + | N T d t ds, (4)
where f is a body force due to volumetric weight, t surface loading, we

arrive to nodal equilibrium equation:
**-&- - «--af- - -£-•

where
KT- ƒ,B'D B dv, L = [B T m N dv, (6)
T
dF
dt - K -&- * • K$- * (7)
To formulate the flow problem the continuation equation has been adopted [6]
in the following form:
(8)
w f
where k is the permeability matrix, n is the porosity and K is the bulk

modulus of fluid compressibility. Applying finite element discratisation with
the help of Galerkin procedure [4] and incorporating prescribed boundary
conditions we arrive to
dp LT du Q (9)
H p - S L y
"dt "dt^ '
where
H = J{V N) T - 7 N dv, S = J*NT - £ - N dv, (10)
Q = fNTq ds (11)
To sum up, the complete set of equations is constituted by (5) and (11).
Having the following set of equations,
_ " df '
0 0 u K -L u
+ d dt (12)
dt —
0 H . P . -L -S . P . Q .
the simple step-by-step integration procedure was applied. The following set
of equation was developed in program PLAXIS by the f i r s t author of the paper.
_ " df "
K -L Au O O u dt
+ At, (13)
-L -S+ocAtH Ap O -AtH . p .t Q .
At
In the calculations the time integration coëfficiënt a=l. The consolidation

time-stepping process was starting from the minimum time step according to
Vermeer and Vermijt 17].
The problems described in the next sections were approached in the
following way. First the loading was applied during elastic or elasto-plastic
undrained analysis. Next the obtained pore pressure in Gauss points were
transferred to the nodes. Such prepared distribution of the excess pore
pressures based on nodal values was improved by prescribing the excess of
pore pressure to be zero a t the chosen boundary lines. Next the elastic
consolidation process was started. During each step of consolidation process
the yield condition was checked. Then time was stopped and the loading step
was applied in the elasto-plastic program in undrained conditions. The second
consolidation process was activated.
The described above approach can be successfully applied in that cases
where the decay of the excess pore pressure will not cause the creation of
plastic zones in the subsoil. That means that the embankment problem can be
successfully analyzed, but the approach will fail in excavation problems.
3. Vertical d r a i n a g e
The subsoil is composed of one clay layer only. A sand embankment is

planed t o be risen. To speed up the consolidation process vertical drains
a r e applied in a triangle pattern with a center to center distance of two
meters. The situation was simplified to a single drain analysis with
axi-symmetric conditions. The loading, material data and boundary conditions
are shown in the Fig.1. The loading was applied in one step in undrained
conditions. Tfte following two situations were analyzed: 1 - with vertical
drains, 2 - without vertical drains. The time-displacement curves (for point
A of Fig.1) for both cases a r e depicted in the Fig.2. The time needed for the
consolidation in the clay subsoil is over one year in this case. Introducing
the vertical drains the consolidation process can be completed five times
faster. The results obtained from finite element analysis were compared with
hand calculations. Terzaghi's formula was used for the calculations of the
one dimensional consolidation process (only vertical flow).
t = -»lX. , (14)
v
where
k
C = — . (15)
v m y
V W
20 kN/mz
ï X
X
y y y y y y V •»
8
perraeable
8
f;
h
CLAY
G = 300kN/n/
v=0.33
1
k x =2.5.ld 4 m/day
1
ai ^
k s 25.10*'' m/day
1
E"?-:-
*n SAND
G«300kN/m 2
v=0.33
Sri--
I
kj,= 1 m/dag
1
1
k s 1 m/dag
1
j impermeable
!.15. 0.65
llD
H
Fig. 1 Vertical drain Fig. 2. Distribution of excess

pore pressure
-0.01
£
z
uj
-0.02
•0.03
•0.04
-0.05
400
TIME [days]
Fig.3 Settlement-time comparisons

The m -value is related to the shear modulus G and the Poisson's r a t i o v.
V
This relation can be found with the Hooke's low;
Ae (1-2W
m (16)
v " A<r " 2G(l-v)
For the radial consolidation Kjellman's formula was applied:
t =
D'
8C
In
m- _3_
4
In
100
100 - U
(17)
where D is the center to center distance, d is the diameter of the drain and
, (18)
C = ±-
h m y
The coupled effect of horizontal and vertical consolidations is calculated

with Carillo's formula:
(100 - U ) (100 - U )
U e 100 - v h (19)
c 100
In the above problem a one layer system was analyzed. In case of a multilayer
subsoil, it is hardly possible to calculate the one dimensional consolidation
process with the scope of classic theories. Then the finite element analysis
shows to be a helpful tooi in designing process.
4. Staged embankment
As second problem an embankment on a layered subsoil is analyzed, Fig.

4. The subsoil is composed of three layers: sand, clay and peat respectively.
The sand layer begins at a depth of 10 m and is stiff and very permeable. The
attention is focused on the building up the embankment in three stages with
a consolidation periods in between. At the very beginning the initial
stresses are introduced. In the first stage the embankment is built up to
the height of 1.75 m. Then the construction is left 200
a=20 m
b j b-IOm
."*.•..'"."".'• '••' '••"third"stage'.'. '.•]•','.- ".1v\ 1.00 171
-
: • ; • . ' • .'•'.*••'second stage"/".-. •• • • • ' • ' j - / ' . 1.00 ra
/?••'''\' '•'•'.' •."•'."• "•!"•-"•'"• * ' r s * s

* a 9 e . " ' ' * . ' - . ' ' ' . " •".".•'.ii • / - " . " • 1.75 ra
1.00 m
6.00 m
Figure 4 Embankment on multilayer subsoil
days to allow consolidation in the subsoil. After 200 days the embankment is
increased by 1 m and the consolidation process is engaged. After 500 days the
embankment is increased by 1 m again and t h e consolidation process is
continued until 99 7. of consolidation. Each stage of loading is done so fast
t h a t both the peat and clay behave undrained during t h e actual construction.
In t h e Figure 6 the calculated load-settlement curve is shown. Increasing the
embankment height up to failure during the first undrained stage a maximum
3.75 mx20kN/m 3 = 75kPa
permeable
JL
3
Y = " kN/m c = 5 kPa k x =2.10'3 m/day
^ peat G = 200kPa 9=25° ky = UO"3 m/day 4m
v = 0333 tprO"
8^
B Y= 16 k N / m 3 c =3kPa k x = 1.10 m/day
clay G=700kPa <P =20° k y = 1.M)4 m/day 6m
1 V = 0.333 «|> = 0 °
zï'
r
J.
permeable
. 10 m , 10 m , 20 m
Figure 5. Material data, boundary conditions and loading

height of the embankment was found to be 1.80 m. This maximum height is in
excellent agreement with the stability slip circle analysis with the
undrained condition. The effect of the consolidation process of the
embankment height can be easy analyzed in the Fig.6. The settlement versus
1 1
1 1 consolidation
3-fh s fage elasfo -plastic

undrained
2-nd stage elasfo-ptastic /

V
consolidation /
undrained f
ultimate he gt of embankment
\ /
consolidation/ . / v i t h o u l tonsuüiWiuu iu> uu
z:
te
z
<
2:
T
A ^
\ i - s r . ST« ge elasfo-[ilasfic
undrained
-0.1 •02 •0.3 -0A -0.5 -0.6 -0.7

SETTLEMENT OF POINT A [m]
Figure 6. Load-settlements curve for point A of Figure 4
TF~1 i I
— undrained deformation
1-st. loading step
ui « 1- dim. consolidation (Terzaghi)

_l
I— l-st. loading step
I—
UJ
250 500 750 1000 1250 1500 1750

TIME I d a y s l
Figure 7. TUne-settlement curve for point A of Figure 4

o
o.
cc
</>
UJ
CC
o.
cc
o
o.
t/i
250 500 750 1000 1250 1500 1750
TIME [days]
Figure 8. Excess pore pressure at point B of Figure 4
_j
n D Q D o a a a D
Q
n° n
• D D D
D
• • D a tP
D n D
O
°n • D tf • a
n • • DQ • •
Fig. 9 Plastic points in the subsoil
Fig.10 Distribution of the excess pore pressure in the subsoil
time is presented in the Figure 7. The one dimensional settlement calculated

with the Terzaghi's formula is compared with the finite element results. The
elastic deformations, calculated by hand, are larger, which can be explained
by the stress distribution in the subsoil. The excess pore pressure (point B
Fig. 4) in time is depicted in the Figure 8. The increase of the embankment
was simulated by the elasto-plastic undrained analysis. Then after each
increase of the embankment the consolidation process was activated. During
building up the embankment, plastic zones appear in the subsoil and then
vanished in the first time step due to increase of the effective mean
stresses.
5. Conclusions
Finite element analysis of consolidation process gives a helpful support

in engineering designing. Herein the elasto-plastic undrained calculations
were mixed with an elastic consolidation analysis. The main advantage of such
approach is the simplicity. The only one thing what is needed is a procedure
to transfer the pore pressures from the Gauss points to the nodes. On the
other hand such analysis is correct in t h a t cases only where the decay of
excess pore pressure will not create plastic zones. So, the check of yield
condition is necessary every time step. Two problems were approached with the
finite element analyses and engineering calculations which show the
consistency, differences as well as limitation of both approaches.
Acknowledgment
The development of the PLAXIS program was partly sponsored by the Road and
Hydraulic Division of Rijkswaterstaat what is duly acknowledged.
References
1. Biot M.A. "General Theory of Three-Dimensional Consolidation", Journal of
Applied Physics, Vol. 12, 1941.
2. Biot M.A. "Theory of Elasticity and Consolidation for Porous Anisotropic
Solids" Journal of Applied Physics, Vol. 26, 1955.
3. Biot M.A. "General Solutions of the Equations of Elasticity and
Consolidation for Porous Material" J. Applied Mechanics, Vol. 23, 1956.
4. Huyakorn, Pinder "Computational Methods in Subsurface Flow" Academie Press
New York 1983.
5. Zienkiewicz O.C. "The Finite Element Method" , London, 1977.
6. Zienkiewicz O.C., Humphenson C , Lewis R.W. "A Unified Approach to Soil
Mechanics Problems Including Plasticity and Viscoplasticity" Swansea, 1975
7. Vermeer P.A., Verruijt A. "An Accuracy Condition for Consolidation by
Finite Elements" Int. J. Num. Analy. Meth. in Geomechanics, Vol. 5, 1981.
8. Kjellman W. "Accelerating consolidation of fine grained soils by means of
card board wicks" 2-nd Int. Conf. Soil Mech. and Found. Eng., Rotterdam
9. Barron R.A. "Consolidation of fine grained soils by drain wells Soil
Mechanics ASCE June.1947
lO.Terzaghi K., "Theoretical Soil Mechanic", New York, J Willy & Sons, 1943
NumericalModels in Geomechanics. Pandeé Pietruszczak (eds) @ 1992Balkema, Rotterdam. ISBN905410 0885
The trial embankments in Vaasa, Finland: A simulation with the Adachi-Oka

creep model
B. A. N. Koehorst & B. H. P.The
Ministry of Public Works & Transport, Rood and Hydraulic Engineering Division. Delft. Netherlands
IA. M.Teunissen
Delft Geotechnics, Netherlands
M. Lojander & A. Naatanen
Helsinki University ofTechnology, Espoo, Finland
ABSTRACT: In this paper comparative calculations have been made for a well moni-
tored embankment on soft clay in Vaasa, Finland. For the constitutive model time
independent (Cam Clay) and time-dependent (Adachi-Oka) models have been used. The
settlements and exeess pore pressures can be accurately simulated. The settlements
calculated with the Adachi-Oka model are less than the settlements computed with Cam
Clay.
1 INTRODUCTION
After the year of 1953, the year in which a great part of the south-west of the Nether-
lands has been flooded, the Delta works has been started in order to prevent such a
disaster. New dikes have been built and existing water defences have been strengthened.
The most of the reconstructions have been designed using classical design methods and
theories such as Bishop slip surface analysis, Hooke's law (linear elastic stress strain
relationship) and Terzaghi's theory on consolidation. Because of the uncertainties in the
design methods a conservative approach is normally used, resulting in the design of large
profiles. Due to the reconstruction of a dike a great amount of soil has been put on the
soft and almost impermeable subsoil. This loading results in large excess-pore pressures.
The generation of those excess-pore pressures reduces the development of the effective
stresses and therefore the stability of the dike. Large settlements in time can occur. In
some cases it seems to be that also creep effects are involved. Experience shows that the
classical design methods sometimes fails to predict the correct behaviour of soil construc-
tions on soft soil. This is due to the fact that the design methods are based on the
assumptions of a linear elastic behaviour of the subsoil. It's known however that soil is
behaving strongly non-linear and non-elastic.
In the past decade the Dutch Ministry of Public Works, in cooperation with the Delft
University of Technology and Delft Geotechnics has been working on new design tools
for the engineering practice. Two of the tools, PLAXIS and MPLUTO, are computer
codes based on the fmite element method. PLAXIS has been developed for the engineer-
ing practice at the Delft University of Technology [1], whereas MPLUTO has been
developed at Delft Geotechnics [2] for research purposes. Both codes have been ported
from the main frame computer to a personal computer and are made 'user-friendly' by an
interactive pre- and postprocessing program. Both models incorporate the Mohr Coulomb
767
constitutive law and the Biot consolidation theory. At the moment MPLUTO also
incorporates a modified Cam-Clay model and the Adachi-Oka creep model.
Attention is now focused on the validation of these models. The Road and Hydraulic
Engineering Division of Rijkswaterstaat is therefore interested in long term measurements
of test embankments to validate the Adachi-Oka creep model. In the Netherlands it's
very difficult to find such a location. For the future RWS is looking for international
cooperation in order in order to collect in-situ measurement data from test sites. At the
moment RWS has some contacts with the University of Helsinki about the Vaasa trial
embankment in Finland. In Florence 1991 the Finnish have presented a simulation of the
behaviour of the embankment with the Cam Clay model using the FEM code Crisp. The
results have been compared with the measured settlements and excess pore pressures.
One of the conclusions has been that the differences found in measured and calculated
settlements may be due to creep.
In the next sections of the paper a Cam Clay calculation and some Adachi-Oka creep
calculations with the Mpluto code are presented. The data presented by A. Naatanen et al.
has been used [3]. The simulation of the Vaasa embankment is part of the validation
program of MPLUTO.
2 DESCRIPTION FEM CODE MPLUTO
In this section a short description is given of the two constitutive models within the FEM
code MPLUTO used for the calculations presented in this paper. One model is based on
modified Cam Clay. The other is a modified Cam Clay model with a creep extension as
proposed by Adachi-Oka [4]. In MPLUTO the equations for groundwater flow and the
constitutive relations are put in one matrix and solved together (coupled approach). Six-
noded triangular elements were used.
2.1 The Cam Clay model
The plastic behaviour is determined by the flow rule
•P-Cir,+ 1
f ?^2n ^ (1)
jw *"H ; n T —rv > ii (Ï-TZXA-K)
2
M P-e
Where
fw = Camclay flow rule (Wet side, cap) [-]
pc = preconsolidation pressure [kN/m2]
q = deviatoric stress [kN/m2]
c = cohesion [kN/m2]
n = porosity H
K = swelling index H
X = compression index H
M = slope of the critical state line H
i^ = cumulative plastic volumetric strain [-]
768
The plastic deformations are described by the plastic potential gw. The index w denotes
"wet side".
g =ƒ (2)
The behaviour at the dry side is described by the Drucker Prager failure criterium.
fd = q-M(p+ccot$) (3)
with f,j is the flow rule for the dry side. The plastic potential at the dry side is non-
associative, which allows no isotropic plastic volumetric strain.
id - ff <4>
with ga is the plastic potential for the dry side.
2.2 The Adachi-Oka model.
The model of Adachi-Oka (1982, see [4]) describes the time dependent soil behaviour
(creep). This model is a so called overstress model and is based on the Cam Clay model.
The viscoplastic strain is determined by
(5)
è*p = <r>e<> —
with
f.
£ f r\e • iffw > 0 (6)
l
0 (ƒƒ„ ^ 0
The two creep parameters are the viscosity r\ [time1] and the coëfficiënt a [-]. These
creep parameters have a relationship with the classical parameters Ca (coëfficiënt of
secondary compression). The viscosity r? is given by:
(7)
2.3t_
The parameter a is given by:

k K
a = ~ (8)
(l+é)m'
with
m' = SL (9)
769
The parameters used in the formulae 7, 8 and 9 depict the following:
C„ coëfficiënt of secondary compression H

Cc compression index (CC=2.3X) H
X compression index H
K swelling index H
e voids ratio H
t„ after this period the creep process ends [time]
In MPLUTO the consolidation and creep process both are starting at the same time.
3 THE V A A S A TRIAL
EMBANKMENT.
3.1 The geometry, instrumentation

and construction
In the paper presented by A. Naatanen

(University of Helsinki, Finland) and
P. Vepsalainen et al. (YS-Engineering)
[3] predictions have been made of the
behaviour of the trail embankment in
Vaasa (Finland). The main purpose of
the construction was the testing and
development of the calculation
methods for the planning of road
A-A
LI U L3 L4
embankments. The trial embankment
has a rectangular shape (figure 1.).
^11^ T X $0.9
£ > j?w ÏC.0 The height of the embankment is 2
2 BriPI meters and the gradiënt of slopes is
i . 33B2 1:1.5. The width of the embankment
6 83
P3 is 15 m. and the length 25 m. The
& L= Settlemeni p l a t e s instrumentation of the trial embank-
B4K
10 P5 P- Piezometer t i p s , o p e n
ment is also presented in figure 1.
B= Piezometer t i p s . c l o s e d
E= Extensometer
Settlements at the base level of the
14' •P&
embankment are measured with settle-
Figure 1. The t r i a l ombankaent in Vaasa. Geometry ments plates, and deeper in the ground
and instrumentation. with magnetic extensometer. The pore
pressures are measured using open and
closed piezometer tips installed at different depths. The construction of the embankment
took 3 days. Furthermore, extensive laboratory soil tests using the triaxial apparatus and
the oedometer have been done in order to determine the soil parameters.
770
3.2 Ground conditions
The subsoil consists of a over 40 m. thick layer of soft silty clay with high organic
content and sulphur. The clay layer is homogeneous and the most of the laboratory
investigations are concentrated only to 15 m. depth. The ground water level is 0.5 m
below the original surface. The parameters used for the calculations are obtained from
the results of oedometer tests. They are presented in table 1. and table 2.
Table 1. Soil parameters.
layer K» [m/d] K^ [m/d] "H *H X[-]

1 1.167D-5 1.167D-5 0.08 0.060 0.600
2 6.000D-6 6.000D-6 0.08 0.045 0.500
3 6.000D-6 6.000D-6 0.08 0.045 0.500
4 6.000D-6 6.000D-6 0.08 0.045 0.500
5 1.333D-5 1.333D-5 0.07 0.045 0.500
6 2.633D-5 2.633D-5 0.15 0.045 0.500
7 8.700D-5 8.700D-5 0.22 0.035 0.450
8 8.700D-4 8.700D-4 0.22 0.010 0.100
9 8.700D-4 8.700D-4 0.22 0.010 0.100
Table 2. Soil parameters (cont.)
Layer *[°] c [kPa] n[-] «H 1/ij [d]

1 26.0 6.0 0.35 0.016 4.43D5
2 26.0 6.0 0.35 0.014 4.43D5
3 26.0 6.0 0.35 0.014 4.43D5
4 26.0 6.0 0.35 0.014 4.43D5
5 26.0 6.0 0.35 0.007 5.10D5
6 25.0 6.0 0.35 0.012 4.61D5
7 30.0 6.0 0.35 0.009 7.69D5
8 30.0 6.0 0.35 0.007 7.69D5
9 30.0 6.0 0.35 0.007 7.69D5
In table 1 and 2 the values for l/rj are related to too=30 years.
771
3.3 Calculations
3.3.1 Introduction
Two different types of calculations have been made with the FEM code MPLUTO:
1 Cam Clay calculations using the same parameters as the CRISP calculations [3].
2 Adachi-Oka calculations with different values for the viscosity i\. From Eq. 7 follows
that the value for t\ depends on the choice of the time U in which the creep process
occurs. In the calculations presented in this paper Xx has been chosen to be 10, 30
and 100 years. For the calculations with
MPLUTO the Vaasa embankment and the
centre One of embankment
subsoil have been schematised using the
q- 45.6 IcN/m mesh presented in figure 2. The mesh
39S-
iillll,. LATEB
consists of 260 six-noded triangular
elements. The embankment is schematised
as a vertical distributed surface load. The
265- \ \ \ \ \ \ \ \ \ \ \ \ \
calculations have been done in plane strain
3ZS-
\ \ \ \ \ \ \ \ \ \ \ \ \ conditions. At first the stress distribution
\ \ \ \ \ \ \ \ \ \ \ \ \ S due to the body weight of the subsoil is
19.5 - calculated (drained). The load due to the
17*" ^ S \ ^ ^ : ^ A S \
2A*
embankment has been applied in an
undrained state, foliowed by consolidation
\ \ \ \ \ \ \ \ \ \ \ \ \ and creep. The calculations have been made
using the Cam Clay and the Adachi-Oka
\ \ \ \ \ \ \ \ \ \ \ \ \ creep model as implemented in the
MPLUTO code. Three sets of calculations
were made using the Adachi-Oka model.
\ \ \ \ \ \ \ \ \ \ \ \ \
2SJ
—1-
Figure 2. FEM mesh. 3.3.2 The Cam Clay computations
In the Cam Clay computations the input parameters listed in table 1 and 2 except from the
creep parameters (a and i\) has been used. The determination of the parameters was done
by extensive laboratory testing in Finland. The same set of parameters has been used in
the CRISP-computations. The overconsolidation ratio OCR in the CRISP and MPLUTO
calculations is 1.45. The main difference between the CRISP and MPLUTO code is the
different type of element (resp. 3-noded and 6 noded elements).
3.3.3 The creep computations (Adachi-Oka)
The same parameters as in the MPLUTO Cam Clay calculations has been used. The two
creep parameters a and TJ are determined with the formulas 7,8 and 9. Cc and C„ have
been determined by laboratory tests in Finland. So in these calculations a is considered as
a material constant. In the creep computations presented in this paper the parameter 17 has
been varied. Three sets of calculations has been made, one with t m = 10 years, t ^ 30
years and t„= 100 years. The values for -q are lying within a range of 10"5 and 4.3*103
[day1].
772
3.4 Results
The results are presented in two different types of figures.
The time vs. settlements curve (fïgure 3):
The calculated initial settlement is 114 mm (that means after the construction of the
embankment), the measured one is 100 mm. Figure 3 shows that the calculated settlement
is quite the same as the measured one (10% overrating).
Compared with the CRISP-calculation the results of the MPLUTO-computation suites
much better with the measured settlements. Probably the main reason for the difference is
the dissimilar way of setting-up the calculation. For instance, to calculate a proper initial
stress distribution (first part of the computation) different values for v has been used in
the second part of the computation (construction embankment and consolidation with the
values of v determined in the laboratory).
0 200 400 600 800 1000 1200 UO0

Time [days]
- G - Mplufo (Com Clay) -jfc- ObswvwJ - * - Crisp

-X-^pluto (Oka, lOy) -S-i^lulo (Oka 30 y) »*tufo (Oka 100 y)
Figure 3. Measured and calculated settlement at the centre of the embankment
For the Adachi-Oka calculations with t„,= 100 years the time dependency is very large.
The settlements are much lower than in the Cam Clay calculations, as can be seen in
figure 3. The soil reacts stiffer because of the time dependent behaviour of the cap. In
Cam Clay the cap reacts instantaneously. The viscoplastic strain in the Adachi Oka creep
calculations is smaller than the plastic strain in the Cam Clay calculation . In the creep
calculations the elastic strain is dominant. This implies that the soil reacts stiffer than in
the Cam Clay calculation. It's for this reason the settlements calculated by the Adachi-
773
Oka model are smaller than the settlements computed with the Cam Clay model. It seems
that the results of a Cam Clay calculation are a limit state of the results of a creep
calculation. By changing the loading rate the creep results probably will be closer to the
Cam Clay results.
Furthermore it can be noticed that in general by applying a creep computation an extra
deformation (creep) is expected. This is not the case with the Adachi-Oka model as extra
creep deformation is not included in the formulation. In fact the Adachi-Oka model is a
Cam Clay model which is extended with a time dependent cap.
The time vs. excess pore pressures curve (fïgure 4, 5, 6 and 7):
The measured excess pore pressures correspond quite well with the Adachi-Oka calcula-
tions as can be seen in figure 4, 5, 6 and 7.
The Cam Clay calculation shows the highest excess pore pressures compared with the
Adachi-Oka results. This is because of the delayed behaviour of the cap. The difference
between the Adachi-Oka calculations for different t» are very small. The line which
corresponds to the calculation with t„= 100 years shows the lowest excess pore press-
ures. It's rernarkable that in the top layers the excess pore pressures disappear very quick
(5 days), while in the layers between 7 and 11 meters the excess pore pressures are still
increasing until t= 1 year. After this period the excess pore pressures remain the same.
Note that the excess pore pressures in the layers between 20 and 25 meter do not change.
For the three Adachi-Oka calculations the differences between the development of the
excess pore pressures at the centre of the embankment are negligible.
70
- Cam Oay " >«i Oka. 1 0 y»ori - Oka, 30 y«

- Cfea. l O O ymarm -afc—
Figure 4. Measured and calculated excess pore pressures at t=5 days.
774
•g 6 0
£>
ƒ> 5 ! s
^ ^
^ ^ ^ S s ^ ,
/ /
/
c ^
10 15
depih [ m ]
- C o m CJay • Oko» 1 O ymarm • Oka( 3 0 ymarm

• O k a , I O O ymarm
Figure 5. Measured and calculated excess pore pressures at t=180 days
70
£-—
- 30
•20
fi*
/ /
^ ^ f e ^
«^ * ^ ^ _
k.
/
10
J 10 15 25
d»pth [ m ]
- Carr» Ctoy -3W Oka, I O yw - Oka, 50 y*

- O k o . 1 OO y » 1 A M*aaur*<t
Figure 6. Measured and calculated excess pore pressures at t = l year
- Oka, 1 O ymarm • Oka. 3 O y*an - Oka. 1 OO ymarm
Figure 7. Calculated excess pore pressure at t = 3 years
775
4 CONCLUSIONS
Considering the settlements it can be concluded that the initial stress distribution has a
large influence on the results. The measured settlements can be simulated accurately with
the Cam Clay model. In this case the settlements were approximately 10% overrated.
Furthermore it has to be kept in mind that the measured settlements consist of primary
settlements and secondary settlements due to creep.
A simulation with the Adachi-Oka model is more complicated. The settlements calculated
with the creep model are less than the measured settlements and the Cam Clay (Mpluto)
results. Probably the loading rate is an important factor and/or the Adachi Oka model
needs to be improved (an extra deformation due to creep is not taken into account; in fact
the Adachi-Oka model is an extension of the Cam Clay model with the time dependent
behaviour of the cap). Further research is required.
Considering the excess pore pressures at the centre of the embankment the calculated
excess pore pressures correspond very well with the measured ones (10% accuracy).
Notice that the excess pore pressures in the top layers are decreasing while the excess
pore pressures in the deeper layers are still increasing. Regarding the excess pore
pressures the differences between the three creep calculations are negligible.
For the future it will be interesting to study the influence of following a different stress
path on the calculational results.
REFERENCES
1 Vermeer P. et al, 1991. PLAXIS, fïnite element code for soil and rock plasticity,
AA. Balkema.
2 Teunissen J., 1988. Weak time-dependent mechanical processes in soil, viscoplastic
behaviour (part 3), Delft Geotechnics, report CO-290511/6.
3 Vepsalainen,P., Arkima O., Lojander M., Naatanen A. The trial embankments in
Vaasa and Paimio, Finland. Proc. of the tenth European conference on soil mech-
anics and foundation engineering Florence, Vol II, pg. 633-640, AA. Balkema
4 Adachi,T and Oka, F, dec 1982. Constitutive equations for normally Consolidated
clay based on elasto-viscoplasticity. Soils and Foundations Vol 22 no 4, Japanese
Society of Soil Mechanics and Foundation Engineering.
5 Adachi T., Mimura., Oka F. Descriptive accuracy of several existing constitutive
models for normally Consolidated clays. Proc. 5th Int. Conf. Num. in Geomechanics,
Vol I pg. 259-266, Nagoya.
6 Bjerrum L., 1967. Engineering geology of Norwegian normally-consolidated marine
clays as related to the settlements of buildings. Geotechnique Vol 17, pg. 81-118.
7 Olszak W., Perzyna P., 1966. The constitutive equations of the flow theory for the
non-stationary yield condition. Applied mechanics, Proc. llth Int. Congr. Applied
Mechanics, pg. 545-553, Springer Verlag.
8 Perzyna P., 1966. Fundamental problems in viscoplasticity. Advances in Applied
Mechanics, (9), pg 243-277, Academie Press.
9 Teunissen J., 1986, Weak time-dependent mechanical processes in soil, non-linear
consolidation (part 1), Delft Geotechnics, report CO-286640/6.
10 Teunissen J., 1986, Weak time-dependent mechanical processes in soil, time-
independent deformations, elasto-plasticity (part 2), Delft Geotechnics, report CO-
286640/7.
776
6th International IAEG Congress/6ême Congres International de AIGI, © 1990Balkema, Rotterdam. ISBN 9061911303

Sur 1'érosion et la perméabilité de materiau argileux des revêtements de digues
J.A.Muijs
Ministry ofTransportation and Public Works, Roods andHydratdics Engineering Division, Netherlands
G.A.M.Kruse
ABSTRACT: The Ministry of Transport and of Public Works of The Netherlands is

presently developing new criteria for the evaluation of the suitability of
available clay soils for liner material on dikes. The focus of the research has
been on erosion behaviour and infiltration characteristics. The paper describes
elements of the background of the presently proposed procedure for evaluation of
clay soils. This evaluation procedure is based on observations implying that
soil structure dominates erodibility and infiltration characteristics of clay
soils in the unsaturated zone. The macropores and other discontinuities of the
soil structure, determine largely the infiltration characteristics and the
erodibility of clay soils. Water content and water content changes are
important elements for the development of soil structure in clay liners. The
soil water suction and water content in liners on dikes have been determined.
Physico- Chemical properties of clay soil were found to be important soil
characteristics for watercontent and watercontent changes. Estimates on physico-
chemical properties of clays can be obtained from Atterberg limits of the clays.
General recommendations were given for evaluation of the development of civil
engineering behaviour of clay soils, based on Atterberg limits and sand content.
RESOME: Le Ministère des Travaux Publics des Pays-Bas actuellement développe des
critères nouveaux pour estimer la qualité des sols argileux comme revêtements
sur digues. Les recherches principalement adressent le comportement d erosion et
les caractéristiques de infiltration. Cette publication décrit la procédure des
investigations pour 1 evaluation des sols argileux. La procédure est basée sur
des observations qui indiquent que la structure pédologique domine les
caractéristiques d'érosion et d infiltration dans la zone non- saturée: Les
interstices et autres discontinuités de la structure pédologique. La teneur en
eau et les variations de cette teneur constituent des paramètres importants en
ce qui concerne le développement d'une structure pédologique en sol argileux
appliqué comme revêtement sur des digues. La pression de succion de 1 eau et la
teneur en eau dans des revêtements ont été déterminé. On a pu conclure que les
propriétés physico-chimiques sont importantes pour la teneur en eau. Estimations
de la valeur de ces propriétés peuvent être déduites des limites d Atterberg.
Des recommendations généraux, basées sur les limites d Atterberg, ont été
développées sur le comportement des sols argileux dans les constructions
civiles.
1 INTRODUCTION - substrate for vegetation or for a

revetment cover
Clay is less easy to handle and rather - soil mechanical stability of the slope
more expensive than sand in The also under tread (cattle)
Netherlands. Consequently, many dikes - the material must be durable, workable,
consist of a sand core covered with a and available at reasonable prices.
suitable clay liner. The clay of these The function of the clay liners
liners has to meet several requirements, necessitates good estimates to be made of
to be mentioned are: the civil engineering properties which
- erosion protection under wave attack; clay soils acquire in the unsaturated zone
- reduction of infiltration during on dikes.
inundation
1213
Most clay soil in The Netherlands is civil engineering properties
illite dominated. The Ministry of of c l a y l i n e r s f o r d i k e s
Transportation and Eublic Works of The erasion resistance infiltration rate
Netherlands is presently developing new
criteria for the evaluation of the üeteriined for Dutctt dikes related to Atterberg l i i i t s
suitability of available clay soils for effects of:
liner material on dikes. The research - cliiate phys.- cnei. properties
presented here is part of this \ / - solids
development. The investigations have been regetation - salid sirface
reported in a series of reports by Delft gratmthater - dissolved latter
Geotechnics, Delft (Reports CO- 275920, surface «ater - adsorbed «atter
CO- 275921, CO- 275923/29 and /33, C0- dilte construction type
275925, CO- 287750).
water content changes
The focus of the research has been erosion
behaviour and infiltration
characteristics. soil structure
The current evaluation of clay soil for

use on dikes is based largely on grainsize erosion
distribution characteristics and organic infiltration - type
matter content. The relation between these - rate
parameters and civil engineering
properties of clay soils in the
unsaturated zone is very indirect, and
therefore often not consistent with Figure 1: Ambiant conditions of clay
observed behaviour of clay soil. This liners on dikes and physico-chemical
paper describes elements of the background properties of the clay determine the
of the presently proposed procedure for development of soil structure, through the
evaluation of clay soils. It is based on influence of notably water content
observations implying soil structure to changes. Watercontentof liners and many
dominate erodibility and infiltration physico-chemical properties are correlated
characteristics of clay soils in the with Atterberg limits.
unsaturated zone.
The development of soil structure in clay

soil and the civil engineering properties, of hard to evaluate parameters, the
are to a large extent the result of the research is based on the assumption that
influence of environmental circumstances many variables are more or less
such as climate, weather and vegetation, correlated. It has been tried to have the
on the interaction of solid matter and correlated sets of variables represented
water. The complex interaction of by a few parameters (see also figure 1).
environmental factors with the physico- In general the investigations did comprise
chemical properties of the solid matter three major tracks. The first track dealt
and dissolved substances, determine water with the mechanisms involved in
content and water content changes of the infiltration and erosion of clay in the
soil (Yarif & Cross-1979). The water unsaturated zone. This track also included
content changes cause volume changes, the determination of reference values for
which result in shrinkage and swelling infiltration and erodibility of existing
fractures in clay soil (White, 1967). The clay liners and for the environmental
macropores and discontinuities resulting conditions for the liners. The second
from the soil deformation together with track dealt with the soil properties
biogenic factors, determine largely the determining or influencing these
infiltration characteristics and the mechanisms. The third track did involve
erodibility of clay soils. Figure 1 gives the correlation of classification test
a general overview of the elements and parameters with the soil properties which
relations assumed to be involved in the dominate the mechanisms influencing the
development of civil engineering civil engineering behaviour. In this
properties of clay soil in the unsaturated manner a consistent picture could be
zone. obtained of the functioning of clay
liners, and of the way in which proposed
test parameters are related to this
functioning. This picture enables
2 METHODS
adaptations of criteria for special
conditions to be made, and also provides a
Since the known interaction of
basis for possible further refinements or
environmental influences on the behaviour
modifications.
of clay already comprises too large a set
1214
soil structure types, and strike, dip,
size and density of fractures were
measured, as well as density and diameter
of biopores. The observations were
completed with determination of the build
up of the dike and its subsurface,
including groundwater level.
The relation between physico- chemical

characteristics and water retention
characteristics was investigated, as well
as the relation between these soil
properties and classification test
parameter values. Notably the relation of
Atterberg limits with water retention and
physico- chemical parameters has been
investigated.
Figure 2: Sketch of the in situ
infiltration measureraent set up for a The final steps did involve the
sufficiently large sample area on steep correlation of the selected classification
slopes (1:3). tests with the civil engineering behaviour
of the liners.
3 INFILTRATION RATES
Existing dikes in the saline and brackish
coastal environments, and in riverine The results of the infiltration tests do
areas in The Netherlands have been not show significant variation. Almost all
investigated in order to determine measured infitration rates are in the
expected values for erosion and range between 10—» tot 10—* m/s.
permeability. Also the relevant Consistent seasonal variation within this
environmental conditions for these dikes range was found on some locations, but was
were determined as well as a range of found to be of the same magnitude as local
physico- Chemical, textural, and variation on each of the sites.
classification parameters. Measureraent of the development of the
infiltration rate of a newly applied liner
Infiltration rate of the clay liners was of recent clay, did show the infiltration
determined using an infiltrometer designed rate to increase from 10-' m/s immediately
to make in situ measurements of a suitably after construction to 10—* m/s within one
large sample area on steep slopes (see summer season.
figure 2). At 7 locations clay liners were
tested in early spring, summer and autumn, An in situ infiltration test using a dye,
at several other locations single demonstrated most of the infiltrated water
measurements were made. to have run through fractures and
biopores. Coloration from the dye
Erodibility was determined in laboratory penetrated only a few millimeters in the
tests on undisturbed samples (diameter 60 soil matrix. These observations are
mm, height appr. 100 mm) from test pits. consistent with reported infiltration
The test were carried out with a small behaviour for agricultural soils (Bouma
laboratory rotational erosion device. The 1981).
test allows for comparison of erodibility
of soil samples, and correlates well with
field observations and large scale test in Breliminary calculations with the Hagen
flumes. Poiseuille- equation for flow in pipes and
cracks indicates that already one average
sized worm hole can account for three
The influence of dike construction type
times the observed infiltration rate
and geometry, weather and climate on water
during the infiltrometer tests. Since
content and water content changes of the
several tens of worm holes usually occur
investigated dikes were determined.
in the area covered by the infiltrometer,
Samples for water content were collected
it is apparently the discontinuity of the
in early spring, summer and autumn,
macropore system which gives an upperbound
together with nuclear probe measurements
limit to the permeability of clay liners.
of soil water content profiles to depths
of over 4 m.
The findings all demonstrated infiltration
in unsaturated clay soil to be largely
Detailed observations on soil types and determined by the soil structure,
soil structure were made in more than 50 including biopores.
test pits in clay liners on dikes. Overall
1215
The observed variation in clay liner The clay liners frequently did show
infiltration rates on dikes, however, was fossilized remnants of a very coarse
considered not significant enough to be prismatic structure with cracks up to over
taken into account in the development of 20 mm. This structure is a remnant of
criteria for general appraisal of the clay strong shrinkage after application of clay
soils for use as clay liner. soil with too high water contents. The
large cracks persisted from the sand core
up to about 0.4 m below the surface, from
4 ERODIBILITY where bioturbation and intense swelling
and shrinking had annihilated the cracks.
Detailed macroscopic and microscopic These cracks often gave the bottom part of
observations of samples eroded in clay liners the appearance of a crudely
laboratory set ups, and field evidence, laid pavement of clay tiles.
indicated erosion of clay soils to take
place by soil loss through three Vertical cracks associated with the
mechanisms, notably: prismatic structure, were observed to
- Dispersion of soil particles through develop within 6 month after application
lack of cohesion; of a clay liner. The crumbly structure was
- drag on individual grains, mainly developed within 2 years after application
concerning sand sized particles; of a clay liner of heavy clay.
- drag on small and large soil aggregates.
The first mechanism did not produce The development of soil structure and
significant soil loss during erosion and bioturbation was observed to have
slaking tests on undisturbed soil samples annihilated the densification effects of
from clay liners. The second mechanism was compaction in 2 liners of heavy clay to a
only significant in sandy clay soil depth of about 0.6 m, within a period of
samples. The third mechanism did cause about 4 years after application.
most soil loss during erosion tests on
roost of the samples. The abiotic soil structure results mainly
from shrinkage and swelling associated
Observations on eroded samples, and of the with changes in water content
weight loss of samples during erosion (White 1967). Initial shrinkage upon
tests, indicated that fractures related to application of too wet soils on dikes
the fine soil structure, and interlock of results in a coarse vertical prismatic
soil aggregates determine the erodibility structure. Repeated and irregular swelling
of most of the clay soil samples. These and shrinkage results in a highly
findings where corroborated by a few fractured clay as observed in the liners.
observations of larger scale erosion of The shrinkage results in vertical cracks,
clay dikes, and of large scale flume whereas swelling results in stresses
tests. Soil structure is therefore a causing fractures at angles of about 40»
dominating factor for erodibility of many to the principal stress. Figures 3 give
types of clay soils from the unsaturated results of the measurements of dip and
zone. orientation of fractures in the liners.
The fractures develop when plastic and

6 SOIL STRUCTURE viscous deformation cannot cope with the
deformation induced by changes in water
Soil structure can be considered a most content. In general soil structure will
important factor in the civil engineering develop, therefore, when the soil has
behaviour of clay soil in the unsaturated becorae sufficiently rigid through inter
zone. The observations in the test pits partiele cementation bonds and high water
show the soil structure in clay liners to suction pressure. The reaction of the clay
exhibit a strong vertical gradiënt, in soil to changes in water content are
general stronger than is common in clay important for the changes volume and also
topsoils (see for examples Scheffer & for the stiffness of the soil matrix.
Schachtschabel 1970). The upper
0.1 to 0.2 m, the grass roots zone, Volume change behaviour associated with
generally has a fine crumbly structure. changes in water content was determined
Dependent on the type of material the next for a large number of samples. In figure 4
few decimeters have a medium to coarse the results of these experiments are
blocky and prismatic structure, except for illustrated. The changes in volume of the
very sandy clays which have only an often samples are plotted against the changes in
weakly developed prismatic structure. At water content related to changes in water
some locations remnants of a platy suction in the range between 10 kPa and
structure were found, which could be 1 MPa. In general the volume change in
traced to result from compaction efforts percentages is about half the change in
during application of the liner. water content in percentages (m/m).
1216
N-talud
+
- + +
+
• +
+
+
+
0 10 20 30 tO SO
watergehalteverandering, gew.%
Figure 4: The change in volume, in
percentages, related to the change in
water content in the range from 10 kPa to
1 MBa suction. Each mark gives water
content change and volume change for a
change of suction of 10E(14 x n ) , n- 2 to 8
Figure 3: Equal area plot of dip and
orientation with respect to the dike of
normals to planar voids in clay liners on
dikes. The percentages in the key refer to
relative count numbers in 1 Z surface area
of the plot. at 250 kPa and the specific surface area
of various types of clay soils, with
relatively low Sodium and Pottassium
adsorbtion ratios (non- marine areas). The
study also did find relations between
water content at various suction levels
In general, (i) the water content at and salt content respectively the
application and (ii) subsequent seasonal, adsorbtion complex of clays for saline and
or more frequent, water content changes brackish environments.
due to changes in ambiant conditions, are
important for soil structure development
of clay soils in the unsaturated zone. The amount of water the soil will hold
during summer conditions (suction level of
100 kPa) can in the same manner be
7 WATER CONTENT, PHYSICO- CHEMICAL estimated from physico- chemical
properties of the soil. This allows
PROPERTIES AND ATTERBERG LIMITS
estimations to be made of the approximate
change of the water content of a clay soil
Environraental circumstances such as
when it is applied as a liner on a dike.
climate, weather, groundwater level and
vegetation, determine the water suction in
the unsaturated zone. The average soil It could be demonstrated for claj' soils in
water suction in summer conditions in clay saline and brackish environments, that the
liners of dikes in The Netherlands, was physico- chemical properties of soil also
determined to be about 100 kFa, and varied correlate with changes in water content at
from over 1 MFa to about 10 kPa. higher suction levels. The number of
samples and the accuracy of the
The properties of the solid matter of the experiments and the sampling was
soil, and of the dissolved substances insufficiënt, however, to determine this
determine the amount of water the soil can relation for all sites.
hold at specific suction levels. The
geometry of the pores is important at In view of the above indications, the
lower suction pressures (< appr. 50 kPa). physico- chemical properties of clay soil
A comparison of the amount of water in the were considered to be appropriate to
soil at higher suction levels with the estimate the development of soil structure
specific surface area of clay soils in a clay after application as liner on a
suggests the water to be distributed in dike. Since such properties are not
films of frequently only several tens of commonly used and determined in civil
water molecules thick, or even less. At engineering practlze, suitable substitutes
higher suction levels (> appr. 100 kPa) were looked for. Atterberg limits show
the physico- chemical properties of the correlations with the same
soil determine the amount of water in the physico- chemical properties as the water
soil. In figure 5 this is illustrated by content at higher suction levels. In
the correlation between the water content
1217
•0.30
water content a t 250 kPa of s o i l s w i t h
low sodium-potassium adsorbtion r a t i o
• 0.25
o
LU
O
«0.
10.00 20.00 30.00 40.00
56.23 70.79 89.13 112.20 141.25 plasticity index
specific surface area (log scale) Figure 6: Plasticity inde:: and surface
charge (CEC/specifie surface area) of the
Figure 5: Water content at s o i l water investigated clay liners for which
suction l e v e l s of 250 kPa and specific Atterberg limits could be determined.
surface area of the i n v e s t i g a t e d clay
l i n e r s with low Sodium and Potassium
adsorbtion r a t i o s .
figure 6 the correlation between

plasticity index and surface charge is
given as an example.
The liquid limit indicates at what

dilution level the bonds between soil
particles become too weak to withstand the
dynamic ,'oading in the test. A soil which
looses this strength at a high water
content will have stronger bonds between
water molecules among each other and with
soil particles, than a soil which looses
strength at low water content. Said bonds
will also hold water in the soil against 2 20.00 40.00 60.00 80.00
suction stresses induced by ambiant liquid limit
conditions. The strength and amount of
bonds between water molecules and the soil
Figure 7: Water content of clay liner
solid matter surface, therefore, determine
samples at soil water suction of lOOkPa
both the liquid limits, and the amount of
and the liquid limits of the sample
water the soil will retain at high suction
material.
levels.
The correlation of physico- Chemical

properties with both, Atterberg limits and
with soil water content at higher suction properties of the soil suggest that the
levels results in a correlation of limits can also indicate the development
Atterberg limits with soil water content. of soil structure after a clay has been
In figure 7 water content at 100 kPa is applied as a liner. Suitable quantitative
plotted against the liquid limit of information on soil structure is very hard
several types of soil. The Atterberg to obtain, and the reiation could
limits can be used directly to obtain therefore not be determined directly. A
indications on the average water content reiation between erodibility of clay liner
of a clay soil after it has been applied samples, which is determined by the fine
as a liner on a dike. soil structure, and the Attterberg limits,
could be demonstrated however. In figure 8
the Atterberg limits of samples of the
8 ATTERBERG LIMITS AND ERODIBILITY clay liners are plotted in a plasticity
chart, and the erodibility of undisturbed
The reiation of Atterberg limits with samples of clay liners is indicated.
water content and with physico- chemical Samples with a high liquid limit, and
1218
The ambiant conditions for clay liners on
dikes with a grass cover in The
Netherlands were determined. Soil water
suction levels during suramer condition
were on average about 100 kEa. Physico-
chemical properties of clay soil are
important soil charaeteristics for the
water content
Estimates on physico- chemical properties

of clays can be obtained from Atterberg
limits of the clays. General
recommendations on evaluation of the
development of civil engineering behaviour
of clay soils, based on Atterberg limits
were developed, and are soon to be
20 40 60 80 published. This includes estimation of
liquid limit average summer condition water content of
clay liners on dikes.
Figure 8: Plastieity index and liquid
limit of the investigated clay liners, and
the erodibility of samples from the ttErnRENCES
liners. Samples vith a lov liquid limit,
or plotting belov the A- line are in Bouma, J. 1981: Soil morhology and
general susceptable to erosion. Highly preferential flow along macropores.
erosion resistant samples are indicated Agric. Water Manag.,3, pp. 235-250.
with a circle. Scheffer, F. & P. Schachtschabel 1970:
Lehrbuch der Bodenkunde. Enke Verlag,
Stuttgart, 448 pp.
White, E.M. 1967: Soil age and texture in
subsoil genesis. Soil
which do not plot below the A- line in the Science,104,4,pp.288-298.
chart, are highly erosion resistant. Yarif, S. & H. Cross 1979: Geochemistry of
Samples of clay soils plotting below the colloid systems for Earth Scientists.
A- line did erode auickly in the erosion Springer, New York, 450 pp.
test. These samples had a very pronounced
fine structure of aggregates with little
interlock, and erosion was merely the
removal of the loosely packed individual
aggregates. Most most samples with a
liquid limit less than about 452 were not
very erosion resistant.
'ihe erodibility of the samples was found

also to be related to sand content of the
clay. Samples containing over about 40 %
sand (>63u) did erode very quickly in
general. These samples did show erosion by
individual grains as mentioned earlier.
9 CONCLÜSIONS
Based on the assumptions that clay soil

behaviour in the unsaturated zone is
determined largely by the soil structure,
the influences on the development of soil
structure in clay liners on dikes have
been investigated. The assumption was
confirmed for erosion behaviour of clay
soil (except for very sandy clay soils)
and permeability of clay liners.
Water content and water content changes

are important elements for the development
of soil structure in clay liners.
1219
Numerical Models in Geornechanics, Pande & Pietruszczak (eds) © 1992 Balkema, Rotterdam. ISBN 905410 088 5

S. E. J. Spierenburg & J.A.M.Teunissen
B.A.N. Koehorst
Ministry of Public Works & Transport, Road and Hydraulic Engineering Division. Delft, Netherlands
ABSTRACT
Slope stability is generally based on the evaluation of a total failure

mechanism. Surface run-off induced by overtopping waves results in
infiltration in the top part a dyke accompanied by deformation.
Infiltration during overtopping was studied using a finite element code for
the Richard's equation. Deformations and stability during infiltration have
been calculated using an uncoupled method.
The calculations how that under overtopping conditions unsafer situations
are possible than the stability of the total embankment.
1. INTRODUCTION
Generally the design of a dyke, in particular the geometry of the

profile, is based on the evaluation of a total failure mechanism using a
(Bishop) slip circle analysis. Records from dyke failures, e.g. during the
disaster in 1953 in the Netherlands, show that in a lot of cases failure is
initiated by local failure of the inner slope near the crest. Instead of
damage on the outward side by wave attack or a deep seated (total) slip
circle, failure occured on the inner slope. Typical damage patterns that
have been observed are vertical cracks in longitudinal direction in the
upper part of the inner slope near the crest or local failure of the inner
slope.
This type of failure is initiated by surface run-off of water along the
crest and the inner slope. The surface run-off is induced by overtopping
waves. Part of this water infiltrates into the dyke accompanied by
deformation.
The Ministry of Public Works in the Netherlands has initiated a

research project, which includes: theoretical analysis and experimental
verification. The present paper describes some of the results of the
computational work on the effects of infiltration on the stability of a
slope.
2. METHODOLOGY: UNCOUPLED APPROACH
Generally infiltration increases stability as fluid flow is directed

vertically downwards, resulting in higher effective stresses. Initially the
study focussed therefore on the calculation and interpretation of
infiltration during overtopping conditions. The geometry was reduced to a
homogeneous slope. The progress of infiltration from an initial situation
up to saturation of the slope was studied for various boundary conditions.
799
In particular the length of the zone of Infiltration was varied. Not every
wave will induce enough overtopping to cover the complete surface of the
slope. The results show that if infiltration occurs over a part of the
surface only (i.e. the crest), an area of the dyke remains unsaturated.
This region acts as an impermeable boundary. Groundwater flow concentrates
around it and therefore the unsaturated region is potentially unstable.
This effect was studied with a deformation analysis using a model for
saturated material. An unsaturated region was created by defining areas of
low permeability. Finally, a calculation was made using an uncoupled method
where fluid pressures from an infiltration analysis were imposed on a
deformation calculation.
3. INFILTRATION
3.1 Richard s equation
The simplest model to describe the process of infiltration is to

consider conservation of mass for the (incompressible) pore fluid only,
leading to:
ff = V . ( k V $ ) U)
where 8 is the volumetric moisture content and $ the total potential. The
potential is defined as the sum of the elevation z and the pressure head h,
which is the ratio of the pressure p and the volumetric weight y.
<)> = z + h = z + p/y (2)
Assuming that the permeability k and water content 9 are functions of the
pressure head h only, the following equation is obtained:
= V ( k(Ö)
8h 9t " ' h) +
~iz~^ (3)
The result is a non-linear differential equation which is generally known

as Richard's equation. Within this concept the process of infiltration is
modelled in a restrictive way. The flow of air is neglected and the air
remains at constant atmospheric pressure. Moreover, deformations coupled to
the flow of water, i.e. (unsaturated) consolidation effects, are not taken
into account. For some cases an analytical solution has been derived, see
[1].
3.2 Material properties: k-h and 6-h relations
The non-linear relations of the permeabilty and water content as a

function of the suction head have to be determined from laboratory tests.
Standard curves for different soil types can be found in literature. Dyke
structures may be build-up from different materials. In the Netherlands,
river revetments are predominantly constructed with clay while in sea-
defense structures a sand core is overlayed by a clay cover. The profile is
in most cases inhomogeneous. The permeability of clay can be characterised
as low, resulting in little or no infiltration. However, the upper region,
where the infiltration occurs, is situated above the freatic surface and
therefore the clay in this part is cracked. In-situ investigations have
shown cracks progressing deep into the core. The 'overall permeability of
such material corresponds with the permeability of sand. A water retention
curve (9-h relation) for cracked clay is not yet available. For the
800
Wotef relention Re'oüve permeobiliw
Figure 1, Material properties.
analysis of infiltration k-h and e-h relations for a Standard sand were
used, see figure 1.
3.3 Boundary conditions
As a result of wave overtopping a volume of water spreads over the

crest and along the inner slope. Water is transported by surface run-off.
The thickness of the layer is relatively small which means that on the
surface the pressure is reduced to zero, no pressure caused by the
thickness of the layer. Therefore the infiltration will occur under the
influence of gravity and suction.
The time dependent aspect, caused by the fact that not every wave will
induce some overtopping combined with a possible tidal variation of the
mean sea level, is as a first approximation neglected. The overtopping
condition of zero pressure along the surface has been supposed as constant.
The geometry is reduced to a single slope, gradiënt 1 : 2. Two
configurations have been studied:
CASE I: overtopping on the crest only,

CASE II: overtopping on the crest and along a inner slope.
These boundary conditions have been indicated in figure 2.
3.4 Initial conditions

As the governing differential equation is non-linear the time
dependent solution will depend on the assumed initial conditions. The
initial freatic line in a dyke at the onset of overtopping is difficult to
predict. The position will depend on the the rate of build-up of the water
level on the outward (sea or river) side. Additionally, an inhomogeneous
profile of the dyke influences the response of the freatic surface on
fluctuations of the sea-(or river) level. The supposed initial conditions
are shown in figure 2.
The initial freatic surface is assumed 0.5 m below the lower soil
surface. Above the freatic surface a linear increasing pressure head is
assumed until a suction head of 1.0 m. Above this level a constant suction
is assumed and consequently a constant water content corresponding to this
801
P=0.0
0.0 X
-4.5
Initial conditions (pressure head and dimensions in meters)
Figure 2, Initial and boundary conditions.
i.S
Figure 3a, (t« 6.7 h) Figure 4a, (t-3.5 h)
Figure 3b, (t=18.5 h) Figure 4b, (t=9.5 h)
Bressure head (in meters) during infiltration, CASE I and CASE II.
suction. The initial water content follows from the adopted soil
characteristics, in particular the 6-h relation.
3.5 Solution method
For the solution of these problems a general purpose finite element

code (DIANA) was used, see [2], Time steps are obtained using an implicit
scheme. A difficulty is to find the optimal time-step. In [3] a criterion
is presented for infiltration problems, which is based on the ratio of
element size and maximum fluid discharge. Initially the performance of
802
-»r-h*-h'-ht-h*-hi-h'-ht-
i 'IVI.VI.VI
v. \> \ \v\ 0 \ y \ vv
\ 0 \ 0 \ \>s \X. sS. W
Figure 5, Discharges.
DIANA was tested against an analytical solution for one-dimensional

infiltration and results with a different FE code, which was available on
personal computer, see [4].
3.6 Results
It appears that for the given material infiltration occurs practically

vertical, which is not surprising for sand. The gravitational force exceeds
over the absorption component. Typical infiltration patterns for the two
cases at some monents are given in figures 3 to 4.
CASE I
At first the infiltration developes vertical (figure 3a) untill the
front has advanced up to the freatic line. From that moment the
groundwaterflow is directed towards the lower soil surface (figure 3b). The
region immediately under the inner slope remains unsaturated while the
saturated flow concentrates around this area. This region acts as an
impermeable boundary and saturation of this part develops "from the bottom
to the top" reducing the size untill the complete profile is saturated.
From a stability point of view, this unsaturated region is potentially
unstable. The groundwater flow around it exerts a tangential force while at
the bottom effective stresses are reduced and so the stability.
CASE II
The infiltration front in this configuration progresses more or less
parallel to the crest and the slope (figure 4a, 4b). Near the tow of the
slope the flow is directed to the lower surface. Saturation of the dyke
develops "from the top to the bottom".
From a stability point of view this flow pattern will increase the
stability.
4 DEFORMATIONS AND STABILITY DÜRING INFILTRATION
4.1 Mechanical model
For the analysis of the plane strain deformation a general purpose

finite element programme (PLDTO) developed within Delft Geotechnics, see
[5], was used. The mechanical model is based on:
the mass balance equations for a consolidating, fully-saturated

material (Biot equations), leading to:
3e 9p (4)
GV*u + (K + 1/3 G)
3x 3x
803
• . • % N \\\\ ^ — - ^ — — _ _~__r\.
« • . * . ! . <U'«'.J"J** * • !
• * * >l * , >. - J . . <l •
Figure 6, Incremental deformation.
GV*w + (K + 1/3 G) 7oz + Ioz2 (5)
V2p (6)
where K and G are the buik and shear modulus of the material, The
permeability is denoted by k and the specific weight by y. The
dispacement in horizontal x and vertical z direction are denoted
by u and w, while the volumetric strain E is defined by:
3u 9w
(7)
3x 3z
a constitutive model: in this study the Mohr-Coulomb failure

condition was adopted.
4.2 Method of analysis
Initially the effect of an unsaturated region was studied by defining

areas of low permeability. The shape of the region was reduced to a
triangle. Failure was induced by stepwise increasing the volumetric weight
as well as the volumetric weight of the pore fluid. In this way the
volumetric effective weight was kept constant. Strength properties were
chosen cohesion = 5 . 0 kN/m2 and internal friction $ = 20°.
A comparison was made with a true uncouple method where fluid

pressures from an infiltration analysis are imposed on a deformation
calculation.
4.3 Results
The unsaturated region is reduced to a triangular shape. The

calculated discharges for this case are shown in figure 5. On the crest and
the lower soil surface fluid pressures are prescribed as zero.
In figure 6 calculated incremental deformations at failure are shown.
Failure occured in this case (corresponding to CASE I) for a volumetric
fluid weight of 7.0 kN/m3 and soil weight 11.2 kN/m3. The unsaturated
region becomes unstable as a result of:
the tangential force exerted by the saturated fluid flow,

the effect of a reduction of effective stresses on the bottom of
this region, which induces a reduction in strength.
804
J—L-L J. J. -L-]
'SS / s s / r f i i i t ' i'
s/ s / r i t i i i i'
s s s s / / / / i ti'
S S S * S / f f > t l '
Figure 7a, Incremental deformation.
4.4 Comparison with a fully saturated analysis
A similar calculation was made for the same geometry only under the
boundary condition of constant overtopping. The final, steady state
situation for the groudwaterflow under constant overtopping is the
situation where the crest and the inner slope act as the freatic surface.
Failure was induced in the same manner as in previous analysis. Failure for
this configuration occurs for a soil weight 16.0 kN/m3 and a pore fluid
weight 10.0 kN/m3, with the same strength characteristics. Incremental
displacements near failure and fluid pressures are shown in figure 7a and
7b. The stability for this loading situation is much greater than in the
previous case. It is clear that this failure mechanism resembles a slip
circle which would be evaluated in a Bishop analysis.
Furthermore fluid pressures near the tow calculated for CASE II correspond
to the fully saturated pressure field. In fact, comparison of figures 4 and
7 shows that, the gradients in the fully saturated case are a maximum
compared to CASE II. Therefore the stability of a slope during infiltration
according to CASE II is greater than the stability of a fully saturated
slope.
4.5 Uncoupled analysis
The same finite element mesh was used in the infiltration as in the
deformation analysis. Fluid pressures calculated with the
unsaturated/saturated flow model (DIANA) were imposed as nodal pressures in
the calculation of deformations with PLUTO.
Deformations for a soil weight 11.6 kN/m3 and fluid weight 7 kN/m3 are
Figure 7b, Fluid pressures.
805
Figure 8, Incremental deformation.
shown in figure 8. The calculations are made for CASE I, conforming to the
pressure field in figure 3b. A similar failure mechanism deveiops in the
calculation with the approximated (triangular) unsaturated region in
paragraph 4.3.
5. CONCLÜSIONS
Infiltration on the crest of a dyke only results in local failure.

This failure mechanism predominates over a total mechanism under conditions
of overtopping on the total surface of the slope. The results is a lower
factor of safety during infiltration. It can be concluded that slopes that
are stable based on a total slip circle analysis may fail during
overtopping. These conclusions are confirmed by records of case histories.
LITERATURE
[1] Philip, J.R.

Theory of infiltration
Advances in Hydrosciences, Vol. 5, 215 - 296 (1969).
[2] DIANA 4.1, User Manual, TNO Building Research, 1992.
[3] Zaradny, H.
Boundary conditions in modeling waterflow in unsaturated soils
Soil Science, Vol. 102, no. 2, 75-82 (1978).
[4] Zaradny, H.
User Manual HAZARSFRI, Polish Academy of Sciences, Gdansk, 1991.
[5] Teunissen, J.A.M.

User Manual PLUTO, Delft Geotechnics, 1991.
[6] Alonso, E.E., Gens, A. and D.W. Hight

Behaviour of partially saturated soils,
Special Lecture, Ninth Eur. Conf., Int. Soc. SMFE, Dublin, 1987.
806
COMMISSION INTERNATIONALE
D E S GRANDS BARRAGES
Dix-septième Congres
des Grands Barrages
Vienne, 1991
DAMS ON SOFT SOILS : LESSONS FROM THE DUTCH DELTA

PROJECTS (*)
R.J. TERMAAT
Senior Geotechnical Engineer
Rood and Hydraulic Engineering Division
Ministry of Transport and Public Works
T H E NETHERLANDS
1. INTRODUCTION
In total eleven new dams were built, including the Eastern Scheldt
storm-surge barrier and a few hundred km's dike were reconstructed within
the framework of the Dutch Delta plan. In some projects stability and
deformation problems arose caused by the soft subsoil. The lessons we
learnt from those cases are discussed in this paper.
The following three items are selected for presentation :
— Failure by large plastic deformations,
— Large lateral compression of the subsoil at the passive side of the
construction and,
— Pore water pressure introduced by plasticity.
Each item is illustrated with a case history.
The experience with the construction of dams on soft soil, as outlined
in this paper, is implemented in additional design rules, which are recorded
in the dutch guidelines on the design of river dikes, part 2, tidal rivers.
(*) Barrages sur sols compressibles : enseignements tirés du Plan Delta.
1709
Q. 66-R. 92
2. FAILURE BY LARGE PLASTIC DEFORMATIONS

During the construction of an earth dam on a subsoil with a high
plasticity and a low shear strength, large plastic deformations can occur-
red. In practice those large deformations will be interpreted as failure,
while theoretical the failure surface is not fully developed [7].
Above described phenomenon was observed during the construction
of the Markiezaatsdam, one of the two secondary dams in the Easter
Scheldt Estuary (Fig. 1).
Fig. l
Aerial view of the Oesterdam (left) and the Markiezaatsdam (right)
Vue aérienne des barrages Oesterdam (a gauche) et Markiezaatsdam (a droite)
The geometry of the Markiezaatsdam just before failure is given in

fig. 2a. First a sill of sand was built by hydraulic fill to mean sea level
(MSL). Because no surrounding dikes of stony material were used, the
slope of the sill varied from 1:25 to 1:30. For the actual dam (above MSL)
a steeper slope was constructed by using bunds of sand. The sill and the
dam were constructed in a period of two weeks, due the high capacity of
the equipment, so that the situation can be considered as undrained.
1710
Q. 66-R. 92
The instabilities were observed at the moment that the dam had a
level of about 2 00 m + MLS. The dam failed by subsidence of the crest,
whilst the adjacent sill slightly rosé. Before failure cracks developed at the
surface of the sand sill, which widened after the failure. About 4 or 5
days after the first instability a dark strip of blue soft clay could be seen
on the edge of the sand sill (Fig. 2b).
Fig. 2
a. Cross section of the Markiezaatsdam a. Coupe transversale du barrage Markie-

shortly before failure zaatsdam peu de temps avant la rupture
b. Cross section of the Markiezaatsdam after b. Coupe transversale du barrage après la
failure rupture
Field observations indicated failure by large plastic deformation,

resulting in settlement rates of the crest equal to or exceeding the surcharge
rates. Stability analysis with Bishop's method showed an overall of safety
of 1.15, which was considered as acceptable because :
— for the construction stage this value was within our scope of
experiences,
— the shear strength data were based on the Dutch cell tests which
are generally conservative, because in the test procedure the deformations
are limited [6], •
Nevertheless the sability analysis indicated the probability of large
deformations, because the slip circles with nearly the same stability factor
as the critical circle covered a large part of the clay layer (Fig. 3). In fact.
a large region of the subsoils is near failure and consequently intensively
deformed.
A better result was obtained with finite element analysis, in which an
simple elasto-plastic stress-strain relation was applied [7], [10]. The power
1711
Q. 66-R. 92
Fig. 3
Area with nearly equal factor of safety based on slip circle analysis
Zone ayant presque Ie même coëfficiënt de sécurité (calcul suivant la methode des cercles
de glissement)
of elasto-plastic computation is that it is a coupled approach to defor-

mation and stability.
The calculated settlement versus the hight of the embankment is
depicted in Fig. 4. The effective height above mean sea level is the height
of the fill minus the settlement of the sill, the bold curve in Fig. 4. This
curve has speak value of 2.3 m for the eflFective hight. In the unstable
post-peak regime, the settlements simply exceed the speed of construction.
According to the calculations, a plastic zone develops very soon, the
deep clay becomes more and more plastic (Fig. 5). This gives the strong
non-linearity in Fig. 4. Finally, the strong peat layer becomes plastic.
The plastic zone calculated with the finite element method corresponds
very well with the area of critical slip circles in Fig. 3.
The computational results predict a very abrupt settlement at an
effective height of 2.3 m. In fact, an abrupt settlement was observed at
2.0 m, so that the stiflfhess of the peat and/or clay was slightly overesti-
mated. A good fit has been found for G = 40 Cu, Cu being the undrained
shear strength and G the shear stiffness. This corresponds very well with
literature data (Fig. 6), in which the ratio between shear stifness and shear
strength is related to the plasticity index of the clay. In this situation the
clay had a plasticity index of 88 ± 22 %.
The above described experience with the construction of Markiezaats-
dam was used in the design and construction of the second secondary
dam, the Oesterdam, located parallel to the Markiezaatsdam (Fig. 1) [8].
1712
Q. 66-R. 92
3
1• H ,
"" S=0.3
„-.--'" 1
' " -
—"'" ^
r: ir
1
> 2.3
.0 •
effect
t o t a l heighr=H t *S
7Scu
G =
' '-
°c 01 0.2 0.3 0i4 0:S
setrlemenf (m)
Fig. 4
Effective surcharge height versus settlement

Hauteur effective de surcharge en fonction du tassement
14 m 52m
1—
freatic line ..^llllü
~~~~~~~~——^.;——-^ "^N\
9m
~~~~"—----~- i ^ - ^ ~• $ ^—-~-_^___ ] •^——______ | •
\ ^ i \ i \
~~^~——-^j ——~"
•
• — —
•
.
' ' ^~~\- ' ~~"^^ i
• " * • —
, , - — _
;
J ,
^^ kk
2.3m
rrrTTT*0?mm0MSfe?Zm
Wl
peak
af e f f e c t i v e heighf
2.1m
post-peak
limit equilibrium
—^^^m,
Fig. 5
Finite element mesh and development of plastic zones (shaded), note the scale distortion
Maillage des éléments finis et développement des zones plastiques (hachurees) — La
distorsion d'échelle est a noter
1713
Q. 66-R. 92
500
1 ° O P o r t s m o u t h cl ay
• Bo s t o n c l a y
•
from
250 1 - + B a n g k o k clay
• Q M aine o r g a n i c clay > Foott &
• A gs c l a y Ladd
n . A A f c h a f a l a y a clay
esterdam ,
100
75
k /
50
•
A
25
0 10 20 30 40 50 60 70 80 90 100
p l a s t i c i t y index
Fig. 6
G/C„ ratio versus plasticity indec
Rapport G/Ca en fonction de l'indice de plasticité
3. LARGE LATERAL COMPRESSION
In a number of cases we were confronted with deformation problems

caused by large lateral compression of the subsoil. This phenomenon
appears when the subsoil at the toe of the construction is relatively weak.
Passive resistance, required for equilibrium, is only possible after large
deformations. Those deformations can increase dramaticly when the soft
subsoil is uplifted by high water pressure in the permeable layer below
this soft subsoil (Fig. 7).
\ W
-^v sanü
^-•••f-'-ïf
uplift pressure
Fig. 7
The mechanism of lateral compression
Mécanisme de la compression laterale
1714
Q. 66-R. 92
The water pressure or uplift pressure reduces the effective stress and
consequently the shear stress at the interface between sand and soft soil,
both stresses can even become zero when the water pressure reach a value
equal to the weight of the above soil layers.
Two methods are used to check the importance of this mechanism
and to calculated the value of the lateral compression. The first method
is based on a non circular slip surface method, i.e. Morgenstern & Price.
In this method firstly the stability is checked. When the stability is sufficiënt
than the lateral compression is determined. In this only the horizontal
part of the critical slip surface is considered (Fig. 8). This horizontal part
is analog to a beam which is loaded by compression. The load boundary
conditions folio ws from the stability analysis, actuaily the interlamel force
in the horizontal part. For each lamel the deformation is calculated with :
AF;
A/, = '- l,
AGd
This formula is found by Hooke's low assuming undrained conditions

and neglecting the shear force and the eccentricity. The deformations are
calculated for additional changes in reservoir, river or sea level, because
the deformations due to the normal water level can be equalised immedia-
tely after completion of the construction.
average
Ij
extreme
Fig. 8
Schematisation of the combined deformation and stability analysis
Schématisation du calcul combine de la deformation et de la stabilité
1715
Q. 66-R. 92
The undrained conditions are used for both the stiffness and the shear
strength, because the additional increases of the water level span mostly a
short time period.
Delft Geotechnics has implemented a modified version of above
method in a computer program ([2] and Dutch Guidelines). The advantages
of this program are the combination of stability and deformation analysis
and a simple procedure for finding the critical slip surface.
The second method, which is used for calculation the effects of lateral
compression, is a finite element approach with a liriear elasto-plastic stress-
strain relation ([10] and Dutch Guidelines). The advantage of this method
is that the deformation and stability are coupled.
The application of both methods is illustrated with the foliowing case
history. A dike section near Bergambacht (20 km east of Rotterdam) was
troubled for several years by maintenance problems, after reconstruction
ten years ago (Fig. 9).
Based on an intensive geotechnical survey it was concluded that the
deformation of the crest was introduced by lateral compression of the
subsoil. The periods with increased deformation rate corresponded with
the period of high water caused by an increased discharge of the river. In
the same period the water pressure below the soft soil layer became nearly
equal to the weight of this layer, as illustrated in Fig. 10. It will be clear
that the water pressure can not exceed the weight of the soft soil layer.
Fig. 9
Maintenance problems dike section Bergambacht
Problèmes d'entretien (la digue a Bergambacht)
1716
Q. 66-R. 92
In this case the lateral compression had to be reduced and conse-

quently the horizontal force on the subsoil behind the construction. This
can be achieved by the construction of a berm. In this case this will be
very expensive, because great number of houses at the toe of the dike had
to be pulled down. So, an alternative solution was chosen in which the
crest height is partly reduced (Fig. 11). This solution is decreasing the
effective weight of the dike and consequently the horizontal force on the
compression zone.
Cross section of the existing situation with water pressure raeasurements

Coupe transversale de la situation existante avec les mesures des pressions interstitv " s
10in
• • 1• • I
Fig. 11
Cross section after reconstruction
Coupe transversale après la reconstruction
1717
Q. 66-R. 92
The existing and the new situation had been analysed with both
methods, the modified slip surface method of Delft Geotechnics (M-Lift)
and the finite element code PLAXIS. The calculation results for a extreme
water level in the river (1/4 000 year) are summarized in the Table below.
Stability Hor. deformation point A
(Fig. 10 & 11) [m]
Existing situation M-Lift 1.15 1.10
PLAXIS 1.15 1.19
New situation M-Lift 1.34 0.29
PLAXIS 1.33 0.18
The displacement fields from the finite element calculations (Fig. 12)
show cleariy the positive effect of the reconstruction.
The above described reconstruction will be completed at the end of
1990. So in a few years we will know the real effect of the reconstruction.
In the dutch guidelines on the design of river dikes, it is recommended
to check above described mechanism of large lateral compression, when
the stability factor is within certain limits, as is illustrated in the flow
diagram in Fig. 13.
A Extreme displacement 1.19E+00 units
-£ 1 1
B Extreme displacement 1.79E-01 units
ê #-•££
*. \ * *.. £•**£• s * ^M. &£*£ g
Fig. 12
Displacements fields from finite element calculations
Champs de déplacements calculés par la methode des éléments finis
a. Existing situation a. Situation existante

b. New situation b. Situation nouvelle
1718
Q. 66-R. 92
*i dan design
staDility analysls with

in the coapreeslon zone
a lateral stress ratio
eaual to Ko
•odification
design
staoility analysls with

passive stress ratio in
the coopression zone
deforaation analysls
yes
7~ . s t a b i l i t y factor
stop
Fig. 13
FIow diagram which indicated when lateral compression has to be checked

Schema de calcul indiquant Ie moment de Ia vérification de la compression laterale
4. PORE WATER PRESSURE INTRODUCED BY PLASTICITY
In general an embankment on soft soil has to be built in several steps.

Between each step the soft soil can consolidated. During this consolidation
•process the excess pore water pressure can dissipated and the effective
stress will increases. Consequently the shear strength will increased too,
which has a positive effect on the stability. The next step of the construc-
tion can start when the stability has reached a required level. Monitoring
the increase of the shear strength, the effective stress or the reduction of
the excess pore water pressure is of major importance. In the Netherlands
usually the excess pore water pressure is monitored at the toe of the
1719
Q. 66-R. 92
construction. The interpretation is based on a combination of elastic stress

distribution and required shear strength determined by stability analyses.
The required shear strength is transformed with Mohr circles to the requi-
red effective mean stress. This method gives for a particular piezometer
the allowable water pressure, as schematic is presented in Fig. 14.
embankment
/required stage 3_ ^j/
effective stage 3 \
/stress — — —..
Y x
stage 1
v.
ooservea
Mater pressure
height of embankment
Fig. 14
Interpretation of pore pressure measurements
Interpretation des mesures des pressions interstitielles
In this philosophy the plasticity is neglected. We have learnt that this

can let to mis-interpretation of the actual condition of the construction,
as shown in the next case history.
A dike section near Streefkerk (close to the dike section discussed in
the previous chapter) failed during the last stage of reconstruction (Fig. 15).
Large cracks appeared in the middle of the dike and half the road sank
nearly 1 m. At the passive side large horizontal deformations were obser-
ved up to 0.50.
An important conclusion of the geotechnical evaluation was, that the
probability of failure could be recognized one year before, from pore
pressure observations, when plasticity had been taken in account in the
implementation [4]. Plasticity introduces additional excess pore pressure,
as illustrated in Fig. 16. During loading the ground behaviour will be
undrained, which means that the effective mean stress (p) can not change
and the increase of the water pressure becomes equal to the increase of
the total stress. The total stress increases elastic, but as soon as the effective
deviatoric stress (q) had reached the failure line plasticity or local failure
occurs and the total stress is increased as far as necessarily for equilibrium.
This introduces extra water pressure.
1720
Q. 66-R. 92
Fig. 15
The crest of the dike after sliding
La crête de la digue après glissement
J1-T3 MoWCOUfcCT*
tawetr»
ESP - «ffecfi» stress patfi U* - por» water prassur» («tetel

TSP'toM stress pari UP - por* water pressure (ptartd
Fig. 16
Induction of excess pore pressure by plasticity
Introduction de pressions interstiüeUes additionnelles du fait de la plasticite
1721
Q. 66-R. 92
Above phenomenon was also observed in Streefkerk one year before

failure (Fig. 18). A local failure or plasticity will not let necessarily let to
failure of the construction. In that case the plastic zone has to be limited.
The position of the plastic zone can be located with piezometer observa-
tions, as outlined above. The allowable dimensions of the plastic zone can
be determined with a finite element code with a elasto-plastic stress-strain
relation [5]. When the load increases the plastic zone will grow, starting
from the active side. This because the higher stiffness of the soil in and
under the construction and the shorter stress path needed for reaching
plasticity conditions (Teunissen e.a. 1986 and Fig. 16). With the installa-
tion of piezometers at two locations, it will be possible to check wether
or not the plastic zone has reached his critical dimension (Fig. 19).
sep'84 reconstTuction
Fig. 17
Cross section of the reconstructed dike
Coupe transversale de la digue reconstruite
height of embankment » tine

Fig. 18
Observed excess pore water pressure
Pressions interstitielles additionnelles observées
1722
Q. 66-R. 92
Delft Geotechnics has developed a alternative method, which is based

on the same principle as discussed above. The advantage of this method
is that the extra water pressure induced by compaction causes by shear
deformation (Fig. 20) can be taken in account too [1].
embankment
plasticrty just allowed
plasttciiy not allowed
plastic zone
Fig. 19
Monitoring system to locate the plastic zone
Dispositif de controle pour localiser la zone plastique
AUd AOP
U°= por* water pranur» (compaction)
Fig. 20
Iaduction of excess pore pressure by compaction
Introduction de pressions interstitielles additionnelles par compactage
5. CONCLUSIONS
In this paper three phenomena are discussed, which can be of major

importance for the construction of dams on soft soils : failure by large
plastic deformations, large lateral compression and pore water pressure
1723
Q. 66-R. 92
introduced by plasticity. For each item two design methods are available,
which are briefly summarized in this paper. Remarkable is that for all the
three cases the finite element approach with a elasto-plastic stress-strain
relation can be used. This has motivated us to put much effort in the
modification of this finite element code, i.e. to make it suitable for the
consulting engineer.
Furthermore this paper shows that local failure or plasticity in soft
soils can be located with pore pressure measurements, which can be helpful
for monitoring the actual stability conditions of a construction.
REFERENCES
[1] BAUDUIN C.M.H, and M O E S C.J.B. (1987). Excess pore water pressure
measurement as a method for embankment stability control. Proc.
IXth ECSMFE, Dublin.
[2] BAUDUIN C.M.H., M O E S C.J.B, and VAN BALEN M. (1989). The
infiuence of uplift water pressures on the deformations and stability
of flood embankments. Proc. Xllth ICSMFE, Rio de Janeiro.
[3] CALLE E.O.F. (1985). Probabilistic analysis of stability of earth slopes.
Proc. XIth ICSMFE, San Francisco.
[4] DEKKER J. and BAUDUIN C.M. (1988). Evaluation of the failure of
an important dike. Proc. IInd Conf. on Case Histories in Geotech-
nical Engineering, St. Louis.
[5] DLUZEWSKI J.M. and T E R M A A T R.J. (1990). Consolidation by finite
element method in engineering problems. Proc. IInd Eur. Conf. on
Numerical Methods in Geot. Eng. Santandor, Spain.
[6] HEIJNEN W.J. and VAN D U R E N F.J. (1979), The Dutch cell test ;
comparison of results of cell tests and triaxial tests on clay. Proc.
VII ECSMFE, Brighton, U.K.
[7] TERMAAT R.J., VERMEER P.A. and VERGEER G. (1985). Failure by
large plastic deformation. Proc. XIth ICSMFE, San Francisco. .
[8] TERMAAT R.J., C A L L E E.O.F, and PETSCHL R.O. (1988). The pro-
bability of failure of an in stages constructed embankment on soft
soil. Proc. IInd Conf. on Case histories of Geotechnical Enginee-
ring, St. Louis.
[9] Technical Advisory committee on Waterdefences (1989), Guidelines on
the design of river dikes, part 2, tidal rivers. (English translation
will be published in 1991).
[10] VERMEER P.A. E.A. (1990). Users manual plaxis. Balkema Publis-
hers, Rotterdam.
1724
Q. 66-R. 92
SUMMARY
In this paper three phenomena are discussed, which can be of major

importance for the construction of dams on soft soils : failure by large
plastic deformations, large lateral compression of the subsoil at the passive
side of the construction and excess pore water pressure introduced by
plasticity. This three phenomena can be obtained with finite element
analysis, in which an simple elasto-plastic stress-strain relation is applied
or can be obtained with a modification of the existing design methods
based on slip surfaces.
The phenomenon local failure or plasticity can be located with pore
pressure measurements, which can be helpful for monitoring the actual
stability conditions of a construction.
RÉSUMÉ
Le rapport traite de trois phénomènes qui peuvent être d'importance

primordiale pour la construction des barrages sur sols compressibles :
— rupture par grandes deformations permanentes (plastiques) ;
— grande compression laterale de la fondation du cöté passif du
barrage, et
— augmentation des pressions interstitielles résultant de la plasticité
dans le sol.
Les trois phénomènes peuvent être analyses a l'aide de la methode
des éléments finis en utilisant une relation contrainte-déformation, ou a
l'aide des methodes classiques de calcul de la stabilité avec cercles de
glissement.
Le phénomène de la rupture locale ou de la plasticité peut être localisé
au moyen des mesures des pressions interstitielles, ce qui peut servir au
controle de la stabilité d'un ouvrage.
1725
EMBANKMENTS ON SOFT SOILS
AfpüaDoo of the Fime Element Method
R.J. TERMAAT
Raid md Hydraubc Engineering Dnrisx»
Mimstty of Trinsport md Public Woria
1 Introduction
The Netherlands is a low-lying country and consequently more than 50% of the sub-
soil consist of very soft clay and peat Iayers with an average thickness of 10 m. The
construction of roads and dikes on such soft sous introduces particular problems:
- Stability during construction;
- In case of reconstruction the stability and deformations of the existing
construction;
- Large horizontal deformations which can effect buildings and other constructions
close to the embankment;
- Maintenance causes by creep.
A two phase (water and soil) finite element approach with a simple elasto-plastic
stress-strain relation shows to be an important design tooi for embankments on soft
soils. In particular in the following cases:
- Stability problems which cannot be described with circular slip surfaces (thin
weak Iayers, uplift, large differences in stiffhess etc);
- Deformation problems of constructions close to the embankment;
- Staged construction; in combination with excess pore pressure measurements;
- The efficiency of soil improvement techniques (geotextiles, vertical drainage,
stone columns etc);
To illustrate the application of a finite element code in engineering problems three

case histories are presented in chapter 2.
The quality of a prediction depends not only form the finite element code, but also
from the schematization of the problem and the detennination of the soil
parameters.
The quality of the finite element program can be check with baüchmarking-
RuodTemw*, DWW 1
problems. For the determinations of the parameters the Dutch Technical Advisory
Committee on Waterdefences will developed a guideline. The schematisation is still
an engineering probiem.
The next step in the development is the introduction of the cam-clay model. The
advantage of this model is a better description of the deformation behaviour, both
in compression and deviatoric conditions. On this moment the cam-clay model is
validated with case histories, to evaluate the importance for practical applications.
Future develops are creep and anisotropic behaviour. Modelling of creep is needed
for optimalisation of maintenance and long term stability. Probably peat behaviour
is dominated by anisotropy. Validation must show Üie practical importance.
2 Application of FEM in engineering problems
2.1 Failure by large plastic deformations
During the construction of an earth dam on a subsoil with a high plasticity and a
low shear strength, large plastic deformations can occurred. In practice those
large deformations will be interpreted as failure, while theoretical the failure surface
is not fully developed (Termaat e.a 1985).
Above described phenomenon was observed during the construction of the

Markiezaatsdam, one of the two secondary dams in the Easter Scheldt Estuary (fig.
1).
The geometry of the Markiezaatsdam just before failure is given in fig. 2a. First a
sill of sand was built by hydraulic fill to mean sea level (MSL). Because no
surrounding dikes of stony material were used, the slope of the sill varied from 1:25
to 1:30. For the actual dam (above MSL) a steeper slope was constmcted by using
bunds of sand. The sill and the dam were constructed in a period of two weeks,due
tiie high capacity of the equipment, so Üiat the situation can be considered as
undrained.
The instabilities were observed at the- moment that the dam had a level of about
2.00 m + MSL. The dam failed by subsidence of the crest, whilst the adjacent sill
Rmid TOTOTI, D WW 2
slightly rosé. Before failure cracks developed at the surface of the sand sill, which
widened after the failure. About 4 or 5 days after the first instability a dark strip of
blue soft clay could be seen on the edge of the sand sill (fig. 2b).
Figure I. Aerial view of the Oesterdam Oeft) and the Markiezaatsdam (right).
2.00m
•0.75a1.00m
0 21*5 a30_ , M.S.L.
variable ;' SOm 26m 5Om '\ variable

5-, ,
sand
Figure 2. a. Cross-section of the Markiezaatsdam shorriy before failure.

b. Cross-section of the Markiezaatsdam after failure.
Ruud Tenmt. D WW 3
Field observations indicated failure by large plastic deformation, resulting in
settlement rates of the crest equal to or exceeding the surcharge rates. Stability
analysis with Bishop's method showed an overall factor of safety of 1.15, which
was considered as acceptable because:
- for the construction stage this value was within our scope of experiences,
- the shear strength data were based on the Dutch cell tests which are generally
conservative, because in the test procedure the deformations are limited (Heynen
& van Duren, 1979).
Nevertheless the stability analysis indicated the probability of large deformations,
because the slip circles with nearly tihe same stability factor as the critical circle
covered a large part of the clay layer (fig. 3). In fact. a large region of the subsoil
is near failure and consequently intensively defonned.
A better result was obtained with finite element analysis, in which an simple elasto-
plastic stress-strain relation was applied (Tennaat e.a., 1985 and Vermeer
e.a.,1990). The power of elasto-plastic computation is that it is a coupled approach
to deformation and stability.
The calculated settlement versus the hight of the embankment is depicted in fig.4.
The effective height above mean sea level is the height of the fill minus the
settlement of the sill, the bold curve in fig. 4. This curve has s peak value of 2.3
m for the effective hight. In the unstable post-peak regime, the settlements simply
exceed the speed of construction.
[sand]
Figure 3. Area with neariy equal factor of safety based on slip circle analysis.
Rond T e m t , DWW 4
According to the calcuiations, a plastic zone develops very soon, the deep clay
becomes more and more plastic (fig. 5). This gives the strong non-linearity in fig.4.
Finally, the strong peat layer becomes plastic.
The plastic zone calculated with the finite element method corresponds very well
with the area of critical slip circles in fig. 3.
The computationai results predict a very abrupt settiement at an effective height of

2.3 m. In fact, an abrupt settiement was observed at 2.0 m, so that the stiffness of
the peat and/or clay was slightly overestimated. A good fit has been found for G =
40 Q, Q being the undrained shear strength and G the shear stiffness. This
corresponds very well with literature data (fig. 6), in which the ratio between shear
stiffness and shear strength is related to the plasticity index of the clay. In this
situation the clay had a plasticity index of 88 +. 22 %.
The above described experience with the construction of Markiezaatsdam was used
in the design and construcüon of the second secondary dam, the Oesterdam, located
parallel to the Markiezaatsdam (fig. 1 and Termaat e.a., 1988).
0.2 0.3
settiement (m)
Figuie 4. Effective surcharge height versus settiement
Rund TL rm—, DWW 5

lUm S2m
9m
2.3m
^ r r ^ Z m m ^ ^ Ë Ë M
p.lm
post-peak
limit equilibrium
Figure 5. Finite element mesh and deveiopment of plastic zones (shaded), note the scale distortion.
500
O OP o r t s m o u f h cl ay <
• B oston ctay
_ +
B angkok clay from
250 aine organic clay >
• Foott &
gs clay Ladd
• . A A rchafalaya clay
*• Ao esterdam ,
100
75
fc /
50 •
•
.25
10.
) 10 20 30 ^0 50 60 70 80 90 100
plasticity index
Figure 6. G/C^ ratio versus plasticity index
RsadTtnom.DWW 6
2.2 Large lateral compression
In a number of cases we were confronted with defonnation problems caused by

large lateral compression of the subsoil. This phenomenon appears when the subsoil
at the toe of the construction is relatively weak. Passive resistance, required for
equilibrium, is only possible after large defamations. Those deformations can
increased dramaticly when the soft subsoil is uplifted by high water pressures in the
penneable layer below this soft subsoil (fig. 7).
The water pressure or uplift pressure reduces the effective stress and consequently
the shear stress at the interface between sand and soft soil, both stresses can even
become zero when the water pressure reach a value equal to the weight of the above
soil layers.
Two methods are used to check the importance of this mechanism and to calculated
the value of the lateral compression. The first method is based on a non circular slip
surface method, i.e. Morgenstern & Price. In this method firstly the stability is
checked. When the stability is sufficiënt than the lateral compression is determined.
In this only the horizontal part of the critical slip surface is considered (fig. 8). This
horizontal part is analog to a beam which is loaded by compression The load
boundary conditions follows from the stability analysis, actually the interlamel force
in the horizontal part. For each lamel the defonnation is calculated with:
This formula is found by Hooke's low assuming undrained conditions and

neglecting the shear force and the eccentricity. The deformations are calculated for
additional changes in reservoir, river or sea level, because the deformations due to
the normal water level can be equalised immediately after completion of the
construction.
The undrained undrained conditions are used for both the stiffness and the shear
strength, because the additional increases of the water level span mostly a short time
period.
Delft Geotechnics has implemented a modified version above method in a computer

program (Bauduin e.a., 1989 and Dutch Guidelines). The advantages of this program
are the combination of stability and defonnation analysis and a simple procedure for
Rood Tonaa, DWW 7

finding the critical slip surface.
The second method, which is used for calculation the effects of lateral compression,
is a finite element approach with a linear eiasto-plastic stress-strain relation
(Vermeer e.a. 1990 and Dutch Guideünes). The advantage of this method is that the
deformation and stability are coupled.
extreme
soil
~VVv • • =
• " /• sand ; - '. \ : .
9 9
• » * . • •
uplift pressure
Figure 7. The mechanism of lateral compression.
average
extreme
Fi+Tt+LFt-Lli
«Ii±A5
• • •• x • •
• ••• /_••• • •
Figure 8. Schemansatjon of the combined deformation and stability anaJysis.
RaadTonaK, DWW 8
The application of both methods is illustrated with the following case history.
A dike section near Bergambacht (20km east of Rotterdam) was troubled for several
years by maintenance problems, after reconstruction ten years ago (fig.9).
Based on an intensive geotechnical survey it was concluded that the deformation of

the crest was introduced by lateral compression of the subsoil. The periods with
increased deformation rate corresponded with the period of high water caused by
an increased discharge of the river. In the sam period the water pressure below the
soft soil layer became nearly equal to the weight this layer, as illustrated in fig. 10.
It will be clear that the water pressure can not exceed the weight of the soft soil
layer.
In this case the lateral compression had to be reduced and consequently the
horizontal force on the subsoil bebind the construction. This can be achieved by the
construction of a berm. In this case this will be very expensive, because great
number of houses at the toe of the dike had to be pull down. So, there was chosen
for an alternative solution in which the crest hight is partly reduced (fig. 11). This
solution is decreasing the effective weight of the dike and consequently the
horizontal force on the compression zone.
The existing and the new situation as been analyzed with both methods, the
modifies slip surface method of Delft Geotechnics (M-Lift) and the finite element
code PLAXIS. The calcularion results for a extreme water level in the river (1/4000
year) are summarized in the table below.
Stabihty Hor. deformation point A

(fig. 1 0 & l l ) [ m }
Existing situation M-Lift 1.15 1,10
PLAXIS 1,15 1,W
New situation M-Lift 1,34 0,29
PLAXIS 1,33 0,18
The displacement fields from the finite element calculations (fig. 12) show cleariy
the positive effect of the reconstruction.
The above described reconstruction will be completed at the end of 1990). So in a
few years we will know the real effect of the reconstruction.
Ruud Tennnt, DWW 9

WÊÊ%ÈÈÈÈÊÊ?£ÊÈ
•* \. ^
Figure 9. Maintenance problems dike section Bergambacht.
u p l i f r h e a d dimif' leyeü
-0.1-T"
u im) : JlOcm
13 -0-3^
1
f" " * ' ' f I I "' | r « " " - "TT1-"f T-
29 30 31 1 2 3 i 5 i 7 I
msrch «pnt
sand
10m
• • • • •
Figure 10. Crosc-iecrjon of the existing situation with water pressure measuremeno.
sand
10m
CEMZXZC
Figure 11. Cross-section after reconstniction.
RnaiTeraaB, DWW 10

Publications On Hydraulic and Geotechnical Problems

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Publications On Hydraulic and Geotechnical Problems

Uploaded by

Copyright:

Available Formats

.*! *' •" .

hydraulic and geotechnical

"•* **• .-"j?s'

•' 'A " • •

Collected papers include the original contributions to

The views and conclusions in this book are those of the

Limited copies of this book can be obtained at cost from

All correspondence and applications should be addressed to:

Road and Hydraulic Engineering Division,

Van der Burghweg 1,

P.O. Box 5044, 2600 GA Delft,

Reproduction of this publication, wholy or in part, is allowed provi-

Publication nr. : I ^ BLIOTHEEK I

1. Design codes and safety factors

2. Closures and storm surge barriers

4. Geotechnical stability of dikes

6. Coastal protection and management

7. Scour and sand-water mixtures

The Public Works Department (Rijkswaterstaat = RWS) manages and

The Netherlands has a tradition of hydraulic engineering stretching

Krystian W. Pilarczyk ( editor)

1, Design codes and safety factors

Hoekstra, A. and K.W. Pilarczyk,

Verhagen, H.J. and W.F. Volker,

Niemeijer, J. and W.F. Volker,

Brinkgreve, R.B.J. and H.L. Bakker,

Termaat, R.J. and E.O.F. Calle,

Simm, J.D. and K.W. Pilarczyk,

Leeuwestein, W., A. Franke, V. Hombergen and J.K. Vrijling,

2. Closures and storm surge barriers (see also P.7, Scour]

Janssen, J.P.F.M, and R.E. Jorissen,

Helsloot, I.C.M. and J.P.F.M. Janssen,

Janssen, J.P.F.M., A. van Ieperen, B.J. Kouwenhoven, J.M. Nederend,

Jorissen, R.E., D.P. de Wilde and E. Berendsen,

Pilarczyk, K.W., H. Havinga, G.J. Klassen, H.J. Verheij, E. Mosselman

Boeters R.E.A.M., F.C.M, van der Knaap,

Boeters, R.E.A.M., H.J. Verheij and M. van der Wal,

Coops, H., R.E.A.M. Boeters and H. Smit,

4. Geotechnical stability of dikes

Deirtekom, J.R. and R.J. Termaat,

Dluzewski, J.M. and R.J. Termaat,

Koehorst, B., B.P. The, J.A.M. Teunissen, M. Lojander and A. Naatanen,

Muijs, J.A. and G.A.M. Kruse,

Spierenburg, S.E.J., J.A.M. Tennissen and B.A.N. Koehorst, ,

Weijers, J.B.A. and J.B. Sellmeijer,

Bakker, H.L. and J.R. Deutekom,

Seijffert, J.J.W. and L. Philipse,

Stoutjesdijk, T., B.P. Rigter, and A. Bezuijen,

Bezuijen, A. and M. Klein Breteler,

6. Coastal protection and management

Verhagen, H.J. and K.W. Pilarczyk,

Pilarczyk, K.W. and M. Stive,

Roelse, P. and H.J. Verhagen,

Qiielerij, de, L., E. van Hijum and K.W. Pilarczyk,

7. Scour and sand-water mixtures

Hoffmans, G.J.C.M, and R. Booij,

Hoffmans, G.J.C.M, and R. Booij,

Jorissen, R.E. and J.K. Vrij 1ing,

Konter, L.M. and R.E. Jorissen,

Winterwerp, J.C., M.B. de Groot, D.R. Mastbergen, and H. Verwoert

Winterwerp, J.C., W.T. Bakker, D.R. Mastbergen, and H. van Rossttm

Visser, P.J., H.J. Verhagen and J.K. Vrijling,

.! ' •" .