Professional Documents
Culture Documents
Publications On Hydraulic and Geotechnical Problems
Publications On Hydraulic and Geotechnical Problems
- BfitlOTM'-ik
.O
: : r B
# * * *
-f Ti
-<**
# ? * *?$
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*
Rijkswaterstaat,
The Netherlands
Tel. +31-15-699111
Fax +31-15-611361
Telex 38043 NL
Subjects
3. Bank protection
5. Revetments
8. Contaminated soils
References
Preface
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
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.
3. Bank protection
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
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
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
Pilarczyk, K.W.,
Design Tools Related to Revetments incl. Riprap
8. Contaminated soils
References
PART X
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
1 Hoekstra
— dikes *'^\
sandy beaches and dunes .<' J
•«—_>
.T.T. erosion ateas. 1 to S m per year
desigmng execulion
problems
• in-site measurements
• models
• calculalions
• experience
solutions
governmenlaV conlractors
research consultants
manufacturers
instituüons
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.
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).
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.- . ,.' -
I '.
transfer
functions
or
field Iheorctical
fltodel
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'.
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}.
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.
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
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.
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
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.
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.
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) .
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.
12 Hoekstra
long
lermeffects
Fig. 9 Soil-Waler-Strueture Inleraction for Revelments funclional conslruclion/costs
requirémeals maintenance
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
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)
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).
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.
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.
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.
References
16 Hoekstra
Breusers, H and A, Raudkivi (1991), Scouring, IAHR Hydraulic Struc-
tures Design Manual, Published by A. Balkema, Rotterdam
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)
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
Keywords
Coastal structures
Dikes
Research
Design codes
Guidelines
TAW-reprint 89.09
Safety against inundation
the Dutch appröach
H.J. Verhagen
W,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
1 Veriiageh/Volkér
figure 1. Area of the Netherlands protected by dikes
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.
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'.
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
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
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
Siiecrecftt
Ol
> 32 at=3-3?5;
ai
=^2=
o
J3
IJt
2=
2 I 0 -1 -2 - j
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.
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.
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.
1. Introduction.
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
P = ƒƒ *B,q(.hfq) dh dq • (2)
where - .,'.,' ^ '
Dy = {h{h,q) > hi}.
=
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)
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.
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 V^i,. 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.
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).
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:
"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 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.
lock
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 ——^—*•
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
i Goidschalxoord
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
. 1 Öf . _, ui -,
A = -—s—5—p Dn e , d - —=— D
d öo* .-• - * öo* - 8tr
f = T - ( e cóty + o* ) sin0
g = x - o* sin*A
where
3 IMPLICIT INTEGRATION
••te'- x A | d t = | / x d t , - | ! |
where
fe;• f(£*+ D Ac)
R° = Q°- J*y B ? ° dV = 0
AR = - / y B Ao;1 dV , P » ƒ B D A£ P dV
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
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
Layer V y
o
c G
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
9 CONCLUDING REMARKS
REFEIïENCES
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
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.
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
and
1 tu-arlL u-aV
\s ss <; J) !> v // yy yy // ;? v / v / /s ss
\////////////////////////////////////. 6
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.
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.
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
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.
References
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
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 on^oing 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.
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.
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.
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).
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
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
©
M A N U A L CONTENTS DESIGN SUB - PROCESSES MAIN DESIGN PROCESS
[ BOUNDARY CONDITIONS
PHYSlCALSiTE
LJL L
Ff*
CONDfTlONS & ENVIRONMENTAL
DATA COLLECTION ASSESSMENT
__ J^.—J * ^
FUNCTIONAL
ANALYSIS
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
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
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
-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
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.
60 V^
/ ~pr
50
40-
30
20-
10-
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).
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
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:
(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
~*V—r ==#?^
*\ I =ÊÊ
i- • !- _. i ~Ï*,<V
*1% ' = f
0.0901-
2 2.5
—t
3 3.5 4
i — • —
4.5
_|
5
1—
5.5
j^.6
waveheightH[mI
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]
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.
totals
Pr 20 18 28 16 12 4 2
De 45 26 0 7 0 22 0
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)
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).
"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]
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.
30/60 mm 0.01 7 7 14 14
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).
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 .
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.
1.0 10
•ton* dlamftttr (ml
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.
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
£ 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
totats
V 90 80 120 70 50 20 10 440
quarry I quarry II
V 170 150 95 84
cro 1860 1680 540 460
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 fp^i.opti^isatiörLÓf^ 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).
r
Hydraulic Criteria Construction Criteria
window
2
T T
Repeat
Transmittèd Waveheight
Toplayer of slope
_L
window Dn50AL=Dn50A (y) (n) Dn50AL for Ht window
4a Dn50AL Norm.U.Cond. 4b
DnSOAi. design wave lee
Repeat Repeat
window 6a or
Design tóe structure window 6b
Repeat
Printing of results
Repeat
END
a*r
-fttsKtanueo. i jftflRjj,
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
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
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 ]
H, /AU,,» = 1.0 F013 (SJ/VN) 0 2 7(cota) £*; for surging waves (A2)
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:
tancr
k,, = cosct i/ ( l - ( )2) (A5)
tan<j)
with (|> being the friction angle of the rock (here assumed 35°).
10 kg 2
10 kg 200 kg 4
200 kg 16
Abstract
1. Introduction
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
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
.-•-••• — 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
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.
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.
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 •»
Figure 9
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
Reliability aspects
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
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.
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.
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]
+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
4.4 Results
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
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-
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
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
24&
Feasibility study of a two-barrier system in the Dutch sea water defence
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
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.
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.
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.
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.
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.
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
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.
The three effects leading to an open barrier situation will now be describes and
translated to a probability.
P ^ c = Pr{PMSL(MSWL)>CL}*C (4)
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 , ^ .
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:
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.
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).
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
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»
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.
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
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.
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.
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.
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
_ •^qr—s£: a/—T&77
hydraulic cylinder
A c
e
s sliding door
-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.
-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
-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.
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:
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.
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 ^ ' ^: ^au^'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).
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
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
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
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.
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.
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
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)
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).
K0 stability parameter at x = 0
Kx stability parameter at x
U0 reference velocity
U x flow velocity at x
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
(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).
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.
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.
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
Figure 9
6. CONCLUSIONS
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
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
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.
465
Q.67-R.28
466
Q. 67-R. 28
- 'MfKJuajaxixs
W?5?SWWrWWS353!5'S'^S?5S'S?
Sa
8b
OTWSSSSTSEWOTRWWSTOCWrasZS?
Fig. 1
467
Q.67-R.28
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.
468
Q. 67-R.28
«MJWJJMJUtf
^ss? •xywxyWxyAyKyK/xy>i>
.469
Q.67-R.28
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
Note : q = 1.7 H " for translation A into B Remarque : q = 1.7 H ' pour transfonper A
en B
Q.67-R.28
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) (h^AD > 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
Table 1
Review of indicative design criteria for threshold condition
Résumé des crirères de calcul pour la condition d'entrainemeni
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°*
In the following the design criteria for the succeeding flow regimes
during construction of a closure dam will be discussed.
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.
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].
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.
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
(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
"erwt
J. ;
Fig. 7
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
3.1. INTRODUCTION
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.
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
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.
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^=
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
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
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
ACKNOWLEDGEMENTS
REFERENCES
486
Q.67-R.28
SUMMARY
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
488
PART 3
Bank protection
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
December 1990
delft hydraulics
CONTROL OF BANK EROSION IN THE NETHERLANDS. STATE-OF-THE-ART
ABSTRACT
INTRODUCTION
CHOICE OF MEASURES
REFERENCES
CONTROL OF BANK EROSION IN THE NETHERLANDS. STATE-OF-THE-ART
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 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
dteaata»
(axeaadanoa
daatfft eandltlraa)
hTdranUa
Figure 2 Simplified fault tree
-5-
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.
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.
% 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
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.
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.
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):
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.
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).
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.
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.
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
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
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.
upstream downstream
V
c
>»
y
o
bank
The following conclusions could be drawn from the model investigations and the
field measurements:
- 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.
-2.0
-1.5
-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.
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.
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.
aandy leut
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.
.. summer polder
summer bed
winter bed
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.
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.
such design criteria that the amount of maintenance and construction costs of
new revetments is minimized.
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-
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)
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-
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.
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.
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 [-]
undi&turtied
hole lilling
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 = 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.
1
W 73
/
- bla clc eai-th
50
25
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).
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
__
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.
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
• 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.
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)
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.
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]
2
•f 1 L-ter Larer
initial sLope
dlstance (n)
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.
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.
" 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
4
Table 1 shows the four different types of riprap protection that have been installed.
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.
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)
0.45
0.25
0.00 0.10 0.20 0.30 0.40 0.50 0.60
distance (m)
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)
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.
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
WHM005
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.
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
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
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
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.
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 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.
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; _
— n
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
14
publication no. 464
November 1991
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.
-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.
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.
-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
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.
-3-
(a) (b) (c)
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)
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)
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-
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
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
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
STUDY AREA
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.
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
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.
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.
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
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).
DIRECT AND INDIRECT EFFECTS OF WAVE ATTACK ON HELOPHYTES 341
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
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
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.
DIRECT AND INDIRECT EFFECTS OF WAVE ATTACK ON HELOPHYTES 343
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.
DIRECT AND INDIRECT EFFECTS OF WAVE ATTACK ON HELOPHYTES 345
( b ) Freauency
1000
900
800 ^ M Site 1. I-U13.3cm
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.
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
Site 1 38 6 50 32 11 51 53 2 59
Site 2 31 2 41 28 1 29 32 5 38
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.).
DIRECT AND INDIRECT EFFECTS OF WAVE ATTACK ON HELOPHYTES 347
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
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.
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.
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
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).
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.
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
35
no
30
25
ZO--
15-•
10-•
i™i"i^+"4»*^*»f^f iiBiMiBi
i—>^I^I^I^I^I—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
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)
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
FREATIC LINE
2SSG
CLAY
y PEAT
CLAY TE7TT
SAND
-415-
•416-
-417-
jJPUFT HEAD (UMIJ LEVEL)
"? -418-
Time (hours)
RIVER
DURING UPUFT
L^
SAND
FREATIC LINE
JNCLiNOMETER
ISC sample
OBSERVATION
WELL
SAND V-t*?
INITIAL ULTIMATE
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.
N
Kn - Ko * ( 1 + (1)
1.96 * N + 4.93
N
Kn - Ko * ( 1 + (2)
3.18 * N + 8.39
2J0
1 1 1 r
N
Gn - Go * ( 1 + ) (3)
0.24 * N + 2.06
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
CONCLUSIONS
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
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
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
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,
dF
dt - K -&- * • K$- * (7)
To formulate the flow problem the continuation equation has been adopted [6]
in the following form:
(8)
w f
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
3. Vertical d r a i n a g e
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
-0.01
£
z
uj
-0.02
•0.03
•0.04
-0.05
400
TIME [days]
Ae (1-2W
m (16)
v " A<r " 2G(l-v)
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
(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
b j b-IOm
-
: • ; • . ' • .'•'.*••'second stage"/".-. •• • • • ' • ' j - / ' . 1.00 ra
1.00 m
6.00 m
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
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
1 1
1 1 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
TF~1 i I
— undrained deformation
1-st. loading step
cc
</>
UJ
CC
o.
cc
o
o.
t/i
TIME [days]
_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 • •
5. Conclusions
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
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.
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.
•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>
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_
with
m' = SL (9)
769
The parameters used in the formulae 7, 8 and 9 depict the following:
In MPLUTO the consolidation and creep process both are starting at the same time.
3 THE V A A S A TRIAL
EMBANKMENT.
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.
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 sub- soil 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-
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).
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 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).
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
774
•g 6 0
£>
ƒ> 5 ! s
^ ^
^ ^ ^ S s ^ ,
/ /
/
c ^
10 15
depih [ m ]
70
£-—
- 30
•20
fi*
/ /
^ ^ f e ^
«^ * ^ ^ _
k.
/
10
J 10 15 25
d»pth [ m ]
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
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.
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
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.
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.
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 .
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
9 CONCLÜSIONS
1219
Numerical Models in Geornechanics, Pande & Pietruszczak (eds) © 1992 Balkema, Rotterdam. ISBN 905410 088 5
ABSTRACT
1. INTRODUCTION
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
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.
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)
800
Wotef relention Re'oüve permeobiliw
analysis of infiltration k-h and e-h relations for a Standard sand were
used, see figure 1.
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:
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)
i.S
Figure 3a, (t« 6.7 h) Figure 4a, (t-3.5 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.
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.
3.6 Results
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.
803
• . • % N \\\\ ^ — - ^ — — _ _~__r\.
« • . * . ! . <U'«'.J"J** * • !
• * * >l * , >. - J . . <l •
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
4.3 Results
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 '
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.
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
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
LITERATURE
[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.
806
COMMISSION INTERNATIONALE
D E S GRANDS BARRAGES
Dix-septième Congres
des Grands Barrages
Vienne, 1991
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.
1709
Q. 66-R. 92
Fig. l
Aerial view of the Oesterdam (left) and the Markiezaatsdam (right)
Vue aérienne des barrages Oesterdam (a gauche) et Markiezaatsdam (a droite)
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
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)
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
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
\ 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
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
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.
-£ 1 1
ê #-•££
*. \ * *.. £•**£• s * ^M. &£*£ g
Fig. 12
Displacements fields from finite element calculations
Champs de déplacements calculés par la methode des éléments finis
1718
Q. 66-R. 92
*i dan design
•odification
design
deforaation analysls
yes
7~ . s t a b i l i t y factor
stop
Fig. 13
1719
Q. 66-R. 92
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
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»
Fig. 16
Induction of excess pore pressure by plasticity
Introduction de pressions interstiüeUes additionnelles du fait de la plasticite
1721
Q. 66-R. 92
sep'84 reconstTuction
Fig. 17
Cross section of the reconstructed dike
Coupe transversale de la digue reconstruite
1722
Q. 66-R. 92
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
Fig. 20
Iaduction of excess pore pressure by compaction
Introduction de pressions interstitielles additionnelles par compactage
5. CONCLUSIONS
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
RÉSUMÉ
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);
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.
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).
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.
sand
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 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)
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
RsadTtnom.DWW 6
2.2 Large lateral compression
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:
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
average
extreme
Fi+Tt+LFt-Lli
«Ii±A5
• • •• x • •
• ••• /_••• • •
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).
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.
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.
•* \. ^
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