2005 Raaijmakers

Final Thesis Report
Supervisors:
Prof. Dr. Ir. M.J.F. Stive
Prof. Dr. Ir. F. Molenkamp
Dr. Ir. J. van de Graaff
Ir. K.G. Nipius
Author:
Tim Raaijmakers
in cooperation with TU Delft and Hydronamic

Delft, June 2005
Submarine Slope Development
of
Dredged Trenches and Channels
Final Thesis Report
Supervisors:
Prof. Dr. Ir. M.J.F. Stive
Prof. Dr. Ir. F. Molenkamp
Dr. Ir. J. van de Graaff
Ir. K.G. Nipius
Author:
Tim Raaijmakers
in cooperation with TU Delft and Hydronamic

Delft, june 2005
Preface
This report describes the research carried out as a graduation project in order to finish
the study of Civil Engineering and Geosciences, Section of Hydraulic Engineering, at Delft
University of Technology. This thesis attempts to bridge the gap between two separate
disciplines with respect to their design approach of side slopes of dredged navigation
channels and pipeline trenches. The focus will be on slope development in the first year,
generally the period of interest of the contractor. During this period the focus will
gradually shift from the geotechnical design approach, primarily based on initial slopes,
and the morphological design approach, primarily based on rough backfilling rates in the
long term. Therefore well established theories from both disciplines are described and
numerical computer programs have been used to create a set of guidelines for slope
design.
Because of the interdisciplinary approach some of the chapters may contain familiar
material, while other chapters may present totally new information, depending on your
particular theoretical background. Please do not fear the unknown, as one of the
purposes of this thesis is to bring separate disciplines to each other. The content should
be comprehensible to any civil engineer and should give insight in the main instability
mechanisms and the morphological behaviour of side slopes. Engineers of both
disciplines should take advantage of knowing the complete picture.
The quick reader is referred to Chapter 12 in which guidelines for slope design are
presented. If questions with respect to certain topics arise, one could easily study the
relevant chapter in more detail.
The very broad subject and the many aspects that are involved in submarine slope
design forced the author to study many different sources and researchers. Especially the
lack of theoretical knowledge on soil mechanics made this research time consuming. The
wide range of soils and hydrodynamic conditions forced me to constantly keep all
possibilities open. Besides, the application of four different, not always very transparent
computer models took a lot of time, because each program asks for a proper
understanding of the input parameters, a large number of simulations and a sound
processing of the results.
But, as a certain soccer phenomenon tends to say, every disadvantage has its
advantage. This broad subject gave me the opportunity to increase my knowledge on
multiple research fields. I think this thesis provides a useful overview of important
aspects of submarine slope design, but I realize that it is just a first attempt to combine
two disciplines in order to formulate a set of guidelines that should be helpful in
designing side slopes. It is obvious that every single subject deserves to be deepened,
although that was not possible in the framework of this thesis.
Of course I am most grateful to my supervisors of the examination committee, consisting

of Prof. Dr. Ir. M.J.F. Stive, Prof. Dr. Ir. F. Molenkamp, Dr. Ir. J. van de Graaff and Ir.
Karoune Nipius, for their advice and discussions on this topic and their remarks on this
report.
My gratitude goes also to various researchers of both disciplines, who were willing to
answer my many questions or to discuss the topic, as there are in arbitrary order, Theo
Stoutjesdijk, Maarten de Groot, Stan Schoofs and Piet Meijers of GeoDelft, Dirk-Jan
Walstra, Dick Mastbergen and Professor Van Rijn of WL|Delft Hydraulics, Jacob Jenssen
and Erik Madsen of the Technical University of Denmark, Jan Rik van den Berg of the
University of Utrecht, Annette Zijderveld of RIKZ, Professor Verruijt, Professor Stelling,
Ronald Brinkgreve and Mark Voorendt of Delft University of Technology, Chris Dykstra,
François Mathijssen and Peter Hendrickxs of Hydronamic, all other writers of the
reference sources, my colleagues of Hydronamic, my fellow graduate students and all
who I have forgotten to mention.
Delft, June 2005
Tim Raaijmakers
Submarine slope development
of dredged trenches and channels
T.C.Raaijmakers, June 2005
Table of contents
PREFACE ....................................................................................................................................................II
SUMMARY................................................................................................................................................ VI
1 INTRODUCTION................................................................................................................................. 1
1.1 GENERAL ......................................................................................................................................... 1
1.2 PROBLEM DESCRIPTION .................................................................................................................... 1
1.3 AIMS AND OBJECTIVES ..................................................................................................................... 2
1.4 OUTLINE OF THE REPORT .................................................................................................................. 3
2 MECHANISMS OF SLOPE INSTABILITY ..................................................................................... 4
2.1 KNOWN MECHANISMS OF SLOPE FAILURE IN LITERATURE ................................................................ 4
2.2 RELEVANT MECHANISMS OF SLOPE INSTABILITY IN TRENCHES AND CHANNELS ............................. 11
2.3 EXPERIENCES FROM DREDGING PRACTICE ...................................................................................... 12
3 REFERENCE CASES ........................................................................................................................ 18
3.1 DEFINITION OF REFERENCE SOILS ................................................................................................... 18
3.1.1 Classification of soil types ..................................................................................................... 18
3.1.2 Soil properties........................................................................................................................ 19
3.2 DEFINITION OF REFERENCE GEOMETRIES........................................................................................ 23
4 UNLOADED MACRO- & MICRO-INSTABILITY ....................................................................... 24
4.1 SHEAR FAILURE .............................................................................................................................. 24
4.1.1 Considerations on shear failure............................................................................................. 24
4.1.2 Unloaded shear failure .......................................................................................................... 25
4.2 LIQUEFACTION AND FLOW SLIDES .................................................................................................. 35
4.2.1 Static liquefaction .................................................................................................................. 35
4.2.2 Computations with SLIQ2D ................................................................................................... 41
4.2.3 Soil composites ...................................................................................................................... 46
4.3 RETROGRESSIVE BREACHING ......................................................................................................... 50
4.4 MICRO-INSTABILITY ON AN INFINITE SLOPE ................................................................................... 52
5 WAVE LOADS ................................................................................................................................... 54
5.1 DIRECT ELASTIC BEHAVIOUR ........................................................................................................ 54
5.1.1 Load or reduction of strength? .............................................................................................. 54
5.1.2 Water pressures below sea level ............................................................................................ 54
5.1.3 Water pressures inside seabed............................................................................................... 57
5.1.4 Total and effective stresses inside seabed.............................................................................. 59
5.1.5 Slope stability under transient wave action ........................................................................... 62
5.2 INDIRECT PLASTIC BEHAVIOUR...................................................................................................... 66
5.2.1 Dynamic liquefaction by wave loads...................................................................................... 66
5.2.2 Computations with MCycle for different slopes..................................................................... 74
5.2.3 Computations with MCycle for different storms .................................................................... 80
5.2.4 Computations with MCycle for multiple layers...................................................................... 84
5.3 OTHER LOADS ................................................................................................................................ 86
6 HYDRODYNAMICS IN TRENCHES AND CHANNELS............................................................. 88
6.1 CURRENTS ...................................................................................................................................... 88
6.1.1 Static wall flow....................................................................................................................... 88
6.1.2 Reynolds-averaged Navier-Stokes equations ......................................................................... 95
6.2 WAVES ........................................................................................................................................... 98
6.3 COMBINED ACTION OF CURRENTS AND WAVES............................................................................. 102
7 THRESHOLD OF MOTION ON A SLOPE .................................................................................. 103
7.1 CURRENTS .................................................................................................................................... 103
7.1.1 Threshold of motion on a plane bed..................................................................................... 103
7.1.2 Threshold of motion on a sloping bed: reduction coefficient............................................... 106
7.1.3 Threshold profile.................................................................................................................. 110
iii
7.2 WAVES ......................................................................................................................................... 112

8 SEDIMENT TRANSPORT IN TRENCHES AND CHANNELS................................................. 118
8.1 CHOICE OF SEDIMENT TRANSPORT FORMULA ............................................................................... 118
8.2 CURRENTS ONLY .......................................................................................................................... 118
8.2.1 Types of transport ................................................................................................................ 119
8.2.2 Bed load transport ............................................................................................................... 120
8.2.3 Dynamic Equilibrium Slopes ............................................................................................... 121
8.2.4 Suspended transport............................................................................................................. 124
8.3.1 Bed load transport ............................................................................................................... 125
8.3.2 Suspended transport............................................................................................................. 125
8.4 SEDIMENT MIXING COEFFICIENT ................................................................................................... 126
8.5 CHANNEL MORPHOLOGY .............................................................................................................. 127
9 SUTRENCH, MODEL DESCRIPTION......................................................................................... 130
9.1 INTRODUCTION............................................................................................................................. 130
9.2 MODEL DESCRIPTION ................................................................................................................... 130
9.3 RELEVANT INPUT PARAMETERS.................................................................................................... 135
10 RESULTS OF SUTRENCH-SIMULATIONS ........................................................................... 137
10.1 SENSITIVITY ANALYSIS ............................................................................................................ 137
10.1.1 Previous results.................................................................................................................... 137
10.1.2 Input parameters.................................................................................................................. 139
10.1.3 Results of sensitivity analysis............................................................................................... 140
10.2 SIMULATIONS WITH UNIDIRECTIONAL FLOW ............................................................................ 148
10.2.1 Introduction ......................................................................................................................... 148
10.2.2 Variation of initial side slopes ............................................................................................. 151
10.2.3 Variation of unidirectional flow velocity ............................................................................. 154
10.2.4 Variation of significant wave height .................................................................................... 158
10.3 SIMULATIONS WITH SIMPLE TIDAL FLOW .................................................................................. 161
10.3.1 Introduction ......................................................................................................................... 161
10.3.2 Variation of initial side slopes ............................................................................................. 162
10.3.3 Variation of tidal flow velocity............................................................................................. 168
10.3.4 Variation of significant wave height .................................................................................... 171
10.4 SIMULATIONS WITH A REAL TIDAL AND WAVE CLIMATE ........................................................... 175
10.4.1 Hydrodynamic input data..................................................................................................... 175
10.4.2 Different wave distributions................................................................................................. 180
10.4.3 Towards a representative wave height................................................................................. 183
10.5 QUALITATIVE COMPARISON OF RESULTS WITH FULL 3D-MODEL .............................................. 186
11 IMPROVEMENT OF TRENCH AND CHANNEL DESIGN .................................................. 188
11.1 CONSIDERATIONS ON SOIL MECHANICS AND MORPHOLOGY ..................................................... 188
11.2 RESULTS OF SIMULATIONS WITH IMPROVED SLOPE ................................................................... 190
11.2.1 Improved slopes in medium packed fine sand (MFS)........................................................... 190
11.2.2 Improved slopes in medium packed medium sand (MMS) ................................................... 193
12 GUIDELINES FOR SLOPE DESIGN ........................................................................................ 196
12.1 NON-COHESIVE SOILS ............................................................................................................... 196
12.1.1 No loads ............................................................................................................................... 196
12.1.2 Wave loads........................................................................................................................... 197
12.1.3 Morphology.......................................................................................................................... 199
12.2 COHESIVE SOILS ....................................................................................................................... 202
13 CONCLUSIONS AND RECOMMENDATIONS ...................................................................... 204
13.1 CONCLUSIONS .......................................................................................................................... 204
13.1.1 General ................................................................................................................................ 204
13.1.2 Applied computer models..................................................................................................... 206
13.2 RECOMMENDATIONS ................................................................................................................ 207
13.2.1 Future fundamental research............................................................................................... 208
iv
13.2.2 Applicability in dredging practice ....................................................................................... 208

13.2.3 Set-up of laboratory experiment .......................................................................................... 209
13.2.4 Set-up of calculation model.................................................................................................. 211
13.3 FINAL REMARKS ....................................................................................................................... 212
APPENDIX A: LIST OF REFERENCES .............................................................................................. 214
APPENDIX B: LIST OF SYMBOLS ..................................................................................................... 220
APPENDIX C: SOIL PROPERTIES OF REFERENCE SOILS......................................................... 224
v
Summary
In shallow areas like coastal seas or harbours the depth is not sufficient for present-day
navigation. Because vessel sizes are growing, existing harbours and approach channels
have to be deepened. Also the growth of the global population results in an increasing
world trade and a growing demand on new harbour areas. Furthermore, the ever
increasing energy consumption and the advancing technology of the oil industry together
make it possible to drill for oil in deeper water. Pipelines, connecting different oil
platforms with each other or an oil platform with the coast, become longer and longer.
Often it is necessary to bury these pipelines in a trench.
These two examples show the already big, but still increasing, importance of dredging
works in a marine environment. In both cases a sound determination of the required
depth and width has to be done. In addition to that, the angle of the side slopes has to
be determined. The designer should be able to give an estimation of the probability of
failure or the maintenance costs. Two rather separate disciplines, coastal morphology
and soil mechanics, have to be combined in slope design.
Although slope angles to a large extent determine on the one hand the slope stability and
on the other hand the volume to be dredged, not only at the beginning (capital
dredging), but also during maintenance dredging, it is still hard to predict whether a
submarine slope is stable and how a slope will develop in time. Therefore slopes are
often dredged as steep as geotechnical stability allows. However, slopes should not only
be geotechnically stable, but also morphologically, at least for some time or within
certain boundaries. So slope design is not only dependent on soil properties, but also on
hydrodynamic conditions (currents and waves).
The aims and objectives of this thesis are to gain insight in:
1. submarine slope development in dredged trenches and channels;
2. the relevant processes and failure mechanisms;
3. the importance of soil mechanics and morphology and the (possible) interaction
between them;
4. the behaviour of different types of soil (sand-silt-clay) with different particle
diameters and soil properties;
with the following restrictions:
a. the period of interest runs from the end of the dredging activities until the expire
date of the period of guarantee (up to one year).
b. the area of interest is situated at sea outside the sheltered area of a harbour
(breakwaters) in regions with significant current and wave action;
c. the depth of interest (= channel depth) will be between 5 and 20 m;
d. the soil types of interest are uncemented, homogeneous sand, silt and clay.
Research teaches that slope development can be divided into three categories:
1. macro-instability, which can be subdivided into macro shear failure, liquefaction
flow slide and retrogressive breaching;
2. micro-instability, (failure of the outer grains);
3. morphological development, which is caused by sediment transport due to
current and waves.
While macro-instability almost always can be interpreted as failure, micro-stability is
allowed under certain circumstances and morphological development simply is something
to deal with. A set of 15 reference soils was composed: five different grain sizes (d50 = 2,
50, 200, 500, 1000 μm) and three different porosities (n = 0,35; 0,40; 0,45), because
these two soil properties are most important regarding respectively morphology and soil
mechanics. All other soil properties are related to these properties and deduced from
literature or calculated from simple correlations. Also two reference geometries were
determined: a typical navigation channel and a typical pipeline trench.
Macro shear failure (‘slip circle analysis’) was studied with the computer program MSTAB.
For all reference soils Safety Factors were determined in drained (sand, silts and clays)
and undrained calculations (clays). Calculations with MSTAB by GeoDelft show that in
situations, where no external loads are present, stability in cohesive soils is governed by
macro-stability and in non-cohesive soils by micro-stability, unless weaker soil layers or a
water level somewhere through the slope are present. Then the slip circle is attracted to
these local weaknesses.
vi
Loosely packed sands are susceptible to liquefaction that can result in a flow slide.
Calculations for static liquefaction were executed with SLIQ2D for three different slope
heights and multiple porosities.
Retrogressive breaching is only likely to occur in slopes of medium to densely packed fine
sand or silt (d50 < 300μm) that are steeper than the angle of internal friction. Although
this instability mechanism controls the dredging process, it will not occur on the side
slopes of a navigation channel, which (after dredging) are often not steeper than 1:3.
Combinations of failure mechanisms ‘liquefaction flow slide’ and ‘breaching’ can occur, for
instance a breach that starts at a steep scar produced by a small flow slide and
retrogrades for many hours.
Although many possible external loads may occur in nature, only wave loads are
considered in this thesis. Many researchers distinguish between the direct, elastic effect
and the indirect, plastic effect. The first approach predicts the fluctuation of water
pressures and effective stresses within one wave cycle. This approach has no net effect
on the mean stresses and it is shown that this direct effect reduces Safety Factors up to
25%, but can cause no failure of a flat seabed. The second approach takes cyclic
compaction into account, which causes a gradual water overpressure. It depends on the
consolidation properties of the soil whether these excess pore water pressure (EPP) can
attenuate or will build up during a storm. Calculations were done in MCYCLE for different
reference soils, storms and 2-layer soil systems. Especially loosely to medium packed
soils can collapse. Improving the soil properties (e.g. vibro-compaction) appeared to be
more effective than flattening the slopes.
Currents and waves that cross a channel will adapt to the expansion in depth. Important
phenomena are flow deceleration due to continuity, flow acceleration due to equal driving
forces, gradual flow refraction, deviation from logarithmic velocity profiles, distortion of
the velocity profile and separation bubbles at steep upstream side slopes, ‘inversed’ wave
shoaling, wave refraction or reflection et cetera.
A sloping bottom affects the threshold of motion (‘Shields’ for currents and ‘Sleath’ for
waves); a reduction factor for all incoming flow angles, slope angles and angles of
internal friction was defined.
Also a so-called ‘threshold profile’ (no sediment transport anywhere along the slope) was
calculated that will develop under very mild hydrodynamic conditions, but also has an
indicative meaning for stronger currents, because sediment transport is based on the
difference between actual bed shear stresses and threshold values. The idea of ‘dynamic
equilibrium slopes’ is based on the idea that for currents perpendicular to the channel
axis the bed load transport is constant along the entire slope.
The width-averaged 2DV-program SUTRENCH developed by WL|Delft Hydraulics was

applied to model the morphological development of the ‘navigation channel’. Besides a
sensitivity analysis, simulations with increasing resemblance to field conditions were
executed. Variation of initial side slopes teaches that side slopes converge during the first
three months. It was shown that steepening of the upstream slope can occur, when a
significant hydrodynamic forcing is present and bed load transport has a large share in
total sediment transport.
Simulations with a real tidal and wave climate yielded three important conclusions.
Backfilling can well be predicted when using one representative wave height instead of a
full-year wave climate. This representative wave height is rather large. Slope
development is governed by the largest waves. A simulation over only the period in
which these largest waves occur yields accurate slope development. Furthermore it
appeared that the order of the waves throughout the year does not affect morphological
development, if one is interested in one or more years; of course some variations during
the year are visible.
Based on all findings of this research it was tried to improve slope design. Five different
slope geometries were evaluated on six criteria. It depends on the soil properties which
criteria should have the largest weights. Favourable morphological behaviour can be
attained as well as a reduction of the dredging costs or an increase of slope stability.
vii
It can be concluded that in this thesis a wide range of topics related to side slopes of
channels has been discussed. Simulations gained insight in slope instabilities and
morphological behaviour. Some easy applicable guidelines were presented, but
sometimes the theoretical background is still weak. Most important recommendations
concern a more accurate modelling of the hydrodynamics (current and waves combined)
at steep side slopes and the wave pressures in the water and in the seabed, especially
under non-sinusoidal waves. A possible set-up of a sophisticated computer model and of
a laboratory experiment are presented. In dredging practice dielectric measurements
should be applied more frequently to obtain in-situ porosities. Data from all channel and
trench projects should be collected in a database to increase knowledge and amount of
fitting data.
viii
1 Introduction
1.1 General
In shallow areas like coastal seas or harbours the depth is not sufficient for present-day
navigation. While vessel sizes are growing, existing harbours and approach channels
have to be deepened. Also the growth of the global population results in an increasing
world trade and a growing demand on new harbour areas.
Furthermore, the ever increasing energy consumption and the advancing technology of
the oil industry together make it possible to drill for oil in deeper water. Pipelines,
connecting different oil platforms with each other or an oil platform with the coast,
become longer and longer. Often it is necessary to bury these pipelines in a trench.
Although ‘pipe laying companies’ often have their own techniques, also dredging of
trenches (and landfalls in particular) occurs.
The two examples, mentioned above, show the already big, but still increasing,
importance of dredging works in a marine environment. Although the size and time
scale (pipeline trenches only have a temporary function; after the pipe is placed, the
trench may silt up or will be backfilled) of both examples are quite different, there are a
lot of similarities. In both cases a sound determination of the required depth and width
has to be done. In addition to that, the angle of the side slopes has to be determined.
The designer should be able to give an estimation of the life expectancy to gain insight
in the probability of failure or the maintenance costs.
This means many parameters have to be investigated, like current conditions, wave
climate, sediment transport and soil conditions. In this thesis the difficulties with respect
to slope design will be studied.
1.2 Problem description
It appeared that slope design forms an important part of the overall channel design, but
what exactly makes it so difficult? To answer that question first a short description of
the present engineering practice will be given.
When designing a navigation channel, usually attention is given to:

1. track and alternatives;
2. channel dimensions (width and depth) depending on vessel sizes, ship
movements, keel clearances, local regulations and sedimentological conditions;
3. capital and maintenance dredging volumes;
4. side slopes.
The fourth point of attention, the side slopes, brings many difficulties along. Although
slope angles to a large extent determine on the one hand the slope stability and on the
other hand the volume to be dredged, not only at the beginning (capital dredging), but
also during maintenance dredging, it is still hard to predict whether a slope is stable and
how a slope will develop in time.
Another remarkable thing is that research on slope development is conducted by either

the soil mechanics department or the morphological department. Nowadays the first
approaches between the two research groups are made, for example in Delft Cluster on
sand borrowing pits and modelling of turbidity currents.
In engineering practice it is found very difficult to give an estimation of the relevant

processes and on what time scale slope development takes place. Therefore slopes are
often dredged as steep as the geotechnical stability allows.
1
However, slopes should not only be geotechnically stable, but also morphologically, at
least for some time or within certain boundaries. Besides, the design of a slope will
influence the erosion/sedimentation pattern of the entire channel and therefore the
capital and maintenance dredging strategy. Predominantly, a steeper slope will decrease
the morphological impact (less sedimentation), because of decreasing width and
corresponding trapping efficiency. On the other hand, steeper slopes will be more
susceptible to instability mechanisms. Therefore the pros and cons have to be balanced
against each other to come to an optimal design. And this design isn’t only dependent
on the soil properties, but also on the hydrodynamic conditions: currents and waves and
their orientation to the channel.
Of course, the above considerations not only concern navigation channels; trenches for
pipelines or tunnels are subjected to similar points of attention, although other
processes and mechanisms will be normative.
Definition of the problem:

Designing slopes of trenches and channels is still too often based on dredging
experience or limited calculation methods. Besides, there is a lack of knowledge of slope
development in time, because most of the times the design is focused on the stability of
the initial slope.
1.3 Aims and objectives
The aims and objectives of this thesis are:

To gain insight in:
between them;
diameters and soil properties.
With the following restrictions:

a. the period of interest runs from the end of the dredging activities until the expire
date of the period of guarantee (up to one year). During this period significant
slope development will occur, but the focus will not be on maintenance
strategies, which have longer time scales, see Figure 1-1. Please note that the
‘final slope’ is of course just a snapshot; morphological development does not
just stop here;
Slope Failure
Initial Final
Slope Slope
Capital Period of Interest Maintenance

Dredging Dredging
Slope Development
Completion Completion of
of the slope entire project TIME
Figure 1-1: Schematization of ‘period of interest’
2
b. the area of interest is situated at sea outside the sheltered area of a harbour
(breakwaters) in regions with significant current and wave action;
c. the depth of interest (= channel depth) will be between 5 and 20 m outside the
breaker zone;
d. the soil types of interest are uncemented sand, silt and clay;
e. the soil will be completely homogeneous, so no separate layers are present.
Therefore first a literature study has been done to study the known instability
mechanisms. This knowledge is used to design initial stable slopes in several reference
cases; this design is mainly based on soil mechanics. Then the development of these
slopes will be studied with the help of a computer program (SUTRENCH). Conclusions
are drawn which channel geometry shows the least or favourable morphological
changes and meets the requirements best during the period of interest.
1.4 Outline of the report
In this thesis research is done on what the dominant processes in different situations
are. Therefore a literature study was done on all known instability mechanisms; a short
summary will be given in Chapter 2. From these mechanisms the relevant mechanisms
for channels and trenches are derived. Also in Chapter 2, the experiences of present-
day dredging practice will be sketched; how are slopes designed in engineering practice
and with what difficulties the contractors were confronted. In Chapter 3 the reference
cases of this thesis are presented. A set of reference soils and two reference geometries
will be composed that will be used throughout the calculations. These reference soils are
varying in grain size and density. The reference geometries represent a typical
navigation channel and a typical pipeline trench. Then, the instability mechanisms will
be treated in a logical order. At first the instability mechanisms from soil mechanics are
described. Chapter 4 discusses the unloaded macro-stability and micro-stability. Macro-
stability is subdivided into shear failure, liquefaction flow slides and retrogressive
breaching. In Chapter 5, submarine slopes are subjected to dynamic loads. In this
thesis of all possible loads only wave loads are considered.
From Chapter 6 morphology comes into play. First a description of hydrodynamics in
trenches and channels is given in Chapter 6. Then the threshold of motion under
currents and waves is determined for all reference soils in Chapter 7. If particles once
started to move, sediment transport has to be considered. Chapter 8 discusses the
application of the Van Rijn transport formulae and the morphology of a trench or
channel.
With a numerical model (SUTRENCH) some calculations will be done. In Chapter 9 this
model will be explained and in Chapter 10 the results of a sensitivity analysis and
simulations with unidirectional and tidal flow will be discussed. In Chapter 11 with the
newly gained knowledge it is tried to improve the slope design with a view to decreasing
maintenance dredging volumes, slope migration or flattening or increasing slope
stability. Some guidelines for slope design in engineering practice are presented in
Chapter 12. Finally conclusions and recommendations will be treated in Chapter 13.
3
2 Mechanisms of slope instability
2.1 Known mechanisms of slope failure in literature
Slope stability and failure has been the subject of intensive research for the last
decades. Since the rather pessimistic quote of Ralph B. Peck (co-author of ‘Soil
mechanics in engineering practice’ [Terzaghi et al.]) in 1967 a lot of progress is made:
“We have lost much of our confidence in our ability to predict the behaviour of a natural
hillside or in the results of our remedial measures….it is evident that nature was able to
outwit us and we fear she can and will do so on similar occasions in the future. This, I
submit, is the present state of the art.”
However, still many questions arise when a slope has to be designed. Especially in
marine conditions, where morphology can be just as important as soil mechanics,
theoretical foundations of a slope-design often are insufficient.
Subaqueous slope failures occur at various scales in different environmental settings.

Schwarz (1982) investigated 286 reported failures of subaqueous slopes covered with
uncemented sediment. It became clear that classification is very difficult and that the
important variables show a huge spread.
Slide structures
A slope failure can be defined as a gravitational deformation of sediment mass (or rock)
leading to a smaller slope angle. These slope failures can be divided into five groups
with increasing water content, see Table 2-1.
Table 2-1: Type of slope failures sorted by water content [source: Schwarz, 1982]
Slope Failures
Creep No visible break of particle contacts
<--- Water content
Fracturing / Folding Break or bending of (an)isotropic beds

Slide Block movement with restricted internal deformation
Mass Flow Viscous, mostly laminar flow
Turbidity Current Viscous, turbulent flow of concentrated suspension
Creep is a very slow type of granular flow and is measured in millimetres or centimetres
per year. Fracturing and folding mainly occurs in rocky beds and rocky soil is beyond
the reach of this thesis. (Sediment) mass flows can be divided into two classes with
increasing sediment concentration (Figure 2-1): (I) slurry flow is a moving mass of
water-saturated sediment and (II) granular flow is a mixture of sediment, air and water.
This flow does not have to be water-saturated and is supported by grain to grain
contact. In submarine conditions (saturated with water) of course only the first one will
occur. The well-known debris (loose, heterogeneous material coarser than sand) and
mud flows belong to this category of slurry flows.
4
Figure 2-1: Division of mass flows into slurry flow and granular flow
[source: www.earthsci.org/flooding]
Turbidity currents appear for instance during land reclamation and dredging activities
(breaching). In this thesis the focus will be on slides. Slides are the most common slope
failure mechanisms and can be divided into four subgroups, see Table 2-2. The most
common slide is the so-called slump, resulting in a more or less circular slip surface. In
this thesis the emphasis will be on this type of slides, which occurs in homogeneous,
isotropic soils, unlike the translational planar block slides. Plastic behaviour is typical for
a convolute slide.
Table 2-2: Types of slide structures [source: Schwarz, 1982]
Slide Structures
Shear or break of small single blocks out of
Minor Block Glide
undisturbed beds
Shear of sediment mass along a concave main shear
Rotational Block Slide
surface. Occurrence in homogeneous, isotropic
(=Slump)
deposits without or with weak influence of bedding
Translational Planar Shear of larger sediment blocks conformable to well-
Block Slide bedded (anisotropic) deposits
Shear of a well-bedded (anisotropic) sedimentary
Convolute Slide complex under plastic deformation along a curved
shear surface.
Another classification in grain-fluid mixture flows is adopted in Mastbergen et al. [1988]

with the help of two dimensionless hydraulic scale numbers (Table 2-3):
- Reynolds-number (Re) which is the ratio of the turbulent and the viscous shear
stress and distinguishes between laminar and turbulent flow;
uh q
Re = = Eq. 2-1
νm νm
in which ‘u’ is the average flow velocity, ‘h’ the water depth and ‘νm’ is the
kinematic viscosity of the soil-water-mixture.
5
- Bagnold-number (Ba) which is the ratio of the inertial and the viscous shear
stress and distinguishes between viscous and inertial flow. This is an indicator
whether fluid or grain properties are dominant.
ρ s d 2γ m Vg
Ba = Eq. 2-2
1 − Vg μ
in which ‘Vg’ is the volume of the grains, ‘d’ is a representative grain diameter,
‘γm’ is the shear strain rate of the soil-water mixture [unit: 1/s] and ‘μ’ is the
dynamic viscosity of the fluid.
A small Bagnold-number (Ba < 40) characterizes the regime of macroviscous flow,
where the viscous interaction with the pure fluid is important. For a large Bagnold-
number (Ba > 450) the flow is in the ‘grain-inertia regime’ where the grain-grain
interactions are dominant.
Table 2-3: Division of grain-fluid mixture flows related to Reynolds- and Bagnold-number
[source: Mastbergen et al., 1988]
Reynolds number (Re) ----------->

Laminar Turbulent
Concentration ---------------------->
Grain size --------------------------->
<-------------- Admission of water

Viscous
Relative grain density ---------->

Poiseuille or Turbulent flow
<-------- Bagnold number (Ba)
Couette flow Suspended sand

Mudflows transport
viscous
Macroviscous Turbidity currents

Macro
grain flows Hyperconcentrated

Flow slides suspension flow
Inertial grain flows
Rock avalanches
Inertial
Dry sand flow

Bed load sand
Bed load sand
transport
transport
Mixing length --------------------------->
Specific flow rate --------------------->
<--------------------------------- Viscosity
Of all mentioned grain-fluid mixture flows by Mastbergen et al. [1988], this thesis will
mainly focus on five of them, which will be only very briefly explained here to get an
idea of the connection between those types [De Groot, 2004]:
1. Grain flow
2. Turbulent density flow or turbidity current
3. Viscous density flow or flow slide
4. Bed load (sand) transport
5. Suspended (sand) transport
Grain flow (I) occurs when a small number of grains are falling from a breach or a slope
equal to or steeper than the natural slope. The grains are jumping and sliding over the
seabed, just like bed load transport (IV). In both cases the water pressure is
hydrostatic, but the driving force of a grain flow is only gravity, whereas bed load
transport is mainly controlled by hydrodynamic forces like current and waves. If more
grains are falling from the breach, they can entrain so much water that a turbulent
sand-water mixture comes into existence. This sand-water-mixture behaves as a
separate, dense fluid with only limited exchange with the surrounding water. The weight
of the sand grains is carried by excess pore pressure (which will be explained later on),
so no contact forces between the grains exist. This kind of flow is more similar to
suspended transport, because turbulence (in combination with the higher sand
concentration near the bed) keeps the grains in suspension, while gravity determines
the settling process. The most important difference between turbulent density flow
6
(turbidity current) and suspended transport is the sand concentration, which is much
smaller for the latter type of flow. This has a major consequence for the erosion
capacity of the flow: the larger the sediment concentration, the larger the bed-shear
stress, the larger the erosion.
The grain-fluid-mixture flow with by far the largest grain concentration is the viscous
density flow, which occurs during a flow slide and has about the same void ratio of the
not-yet collapsed (loose) soil. This flow is characterized by its viscous and slow process.
A special kind of grain-fluid flow that can develop in grain flow and turbulent density
flow under specific circumstances is uniform flow. As illustrated in Figure 2-2 these
specific circumstances are a constant flow velocity, layer thickness and concentration.
Because of the concentration-dependent erosion capacity, the slope angle of uniform
flow (βuniform) decreases for larger concentrations: a turbulent density flow of higher
concentration, and thus higher erosive capacities, is in equilibrium for smaller slope
angles.
z
tanβ
UNIFORM FLOW:
∂v ∂h ∂C
h =0 , =0 , =0
tanφ ∂x ∂x ∂x
0.5 0
0.4
v
0.3
0.2 SEDIMENTATION
=
EROSION x
0.1
sand discharge
0.1 1 10 kg/ms
Figure 2-2: Uniform flow; grain flow for sand discharge <1 kg/ms; turbulent density flow for
sand discharge >1 kg/ms [source: De Groot, 2004]
Once a grain-fluid flow of type II and III has come into existence it can sustain for a
rather long time because of a process of hindered settling. Whereas sand in low
concentrated flows have typical fall velocities of a few centimetres per second, this fall
velocity can be reduced to less than a millimetre per second in high sediment
concentrations.
7
Regions
An attempt was made to classify each failure using specific regions exposed to more or
less the same geomorphologic, hydrodynamic and depositional conditions. As can be
seen in Figure 2-3 most reported slope failures occur at continental slopes, canyons and
deltas. One should keep in mind that this figure is only a rough estimation, because in
the past there have been many slope failures which have not been reported.
Classification in regions
40
35
30
Percentage [%]
25
20
15
10
0
s s s
pe on lta
s ers tio
n
l fla
ts rds
ls
lo
an
y de , r iv v a a fjo
nta c es xc
a tid
ne lak le
nti ific
i a
co
art
Figure 2-3: Percentage of reported slope failures; classification in regions

[source: Schwarz, 1982]
Summarizing all of the reported slope failures, a topographic schematization has been
plotted in Figure 2-4. In this thesis we are especially interested in failures on the
continental shelf and marine deltas.
8
Figure 2-4: An overview of all kinds of slope failures

[source: faculty.etsu.edu/wallaces/MassW.pdf]
Slope angles
When analyzing slope angles at which failures occur, a very broad scattering is found
with values less than 0,5° (1:115) up to values of 50°-60° (1:0,8 to 0,6). Considering
large slope failures it becomes clear that even very flat slopes can collapse.
Classification in slope angles
40
35
30
percentage [%]
25
20
15
10
5
0
< 0,5º 0,5º - 3º 3º - 10º 10º - 20º > 20º
slope angle β
Figure 2-5: Percentage of reported slope failures; classification in slope angles

9
Release mechanisms
The transition from a stable slope to an unstable slope can have different causes. In
literature nine so-called release or ‘trigger’ mechanisms have been reported, varying
from gradual releasing causes to very sudden events. These nine causes, however
sometimes difficult to distinguish (or overlapping), are:
1. Accumulation
a. long-termed high sedimentation rate in case of slowly progressing delta
front areas or steep trench slopes. This mechanism is often combined
with an additional trigger effect;
b. short-termed heavy sediment supply because of large river floods;
c. overloading of insufficiently consolidated slope and near-slope areas
because of loss of buoyancy (tidal flats) or strong accumulation at river
mouths;
d. oversteepening of a depositional slope at a shelf break or at a trench
slope, possibly increased by tectonic movements;
2. Earthquakes, probably the most effective sudden trigger mechanism;
3. Tectonical reasons. In tectonically active zones favourable conditions for slope
failure can be created, although these processes normally are too slow to speak
of a ‘real trigger mechanism’;
4. Currents and waves
a. normal currents, just able to move some sediment, especially at former
periods of regression, the activity of currents and waves was
concentrated to a narrow seam near the shelf break, which could have
formed a spill of shelf sediment;
b. hurricanes and large waves in delta areas and at canyon heads;
c. tsunamis and seiches;
5. Emergence (sea level change). Dropping of the water level implies loss of
buoyancy and an increase of effective load on a potential shear surface. In tidal
flat areas or during Pleistocene low sea levels this can cause slope failure.
6. Change of certain physical-chemical properties
a. increase of sensitivity due to leaching (Dutch: uitlogen) of marine
cohesive sediments by freshwater influence is a well-known mechanism
responsible for the quick clay effect;
b. subterranean influx of freshwater into marine sediments results in a
weakening of the sediments;
c. chemical decomposition of organic or inorganic sedimentary components;
7. Pore water overpressure in an unconsolidated deposit diminishes its shear
resistance. This can be caused by different mechanisms, like tectonical stress,
rapid sediment accumulation, loss of buoyancy and development of gas
(methane) from decomposition of organic debris;
8. Erosion as a trigger mechanism means undercutting of slope deposits, for
example along river banks or tidal gullies;
9. Man-made slope failures especially occur during constructing of harbours, dams
or dredging works. In this respect also artificially dumped (waste) deposits are
mentioned.
10
Classification in release mechanisms

35
percentage [%] 30
25
20
15
10
5
0
t io
n
ke
s
on
s es nc
e t ie
s ure sion ilures
l a a s av e e r s s
u
um art h ical r
q u e a
nd
w
erg prop erpre ero e f a
c a e m l p
ac e
t on rr ent
s ica er o
v
e slo
c m t d
te cu he wa a
ico- c ore a n-m
ys p m
ph
Figure 2-6: Percentage of reported slope failures; classification in release mechanisms
One should keep in mind, that in many cases more than one of the above release
mechanisms are responsible for slope failure. So Figure 2-6 is only meant to get a
general idea. Some of the mentioned mechanisms are related to each other. Currents
and waves for example can cause gradual accumulation or erosion, but can also act as a
sudden trigger mechanism, like for instance a storm wave or tsunami.
2.2 Relevant mechanisms of slope instability in trenches and

channels
In the previous paragraph a lot of slope failures and trigger mechanisms were classified
in different categories. Although these failures are characterized by various time- and
space-scales, it appears that they all can be combined to a few basic instability
mechanisms, which are relevant in the time- (up to one year) and space-scales (few
hundred meters) of this thesis.
In Figure 2-7 an overview of these mechanisms is given. Usually a distinction is made

between micro- and macro-stability. With micro-stability the stability of the outer grains
is meant, in contrast to the stability as a whole, the macro-stability. The stability of the
outer grains is strongly depended on the working forces, like for example seepage or
the wave force by short waves. Normally in soil mechanics, the term ‘micro-stability’ is
only adopted when speaking of instability due to forces from inside the sand or clay
body, like ground water flow or water pressures due to seepage or passing waves.
When these forces are induced by waves or currents from outside the body and
sediment is picked up, we speak about ‘morphology’. Now, ‘instability’ means transport
of particles. As long as the channel profile stays within certain limits, sediment transport
is allowed. In the morphology-approach it is not the question whether the soil is ‘stable’
(no movement at all) but for how long the channel or trench will fulfil its purpose.
Unlike subaqueous slopes, slopes above or around the waterline like embankments or
dikes should be stable under all circumstances. Therefore often an armour layer is
applied to protect the underlying soil. As mentioned above, slopes of trenches and
channels may experience some profile changes as long as their functionality (required
profile) isn’t endangered.
As a rough design criterion, it is stated that:
11
- macro-instability should be prevented at all times, because large movements of

soil will almost certainly cause failure of the trench of channel;
- micro-instability can be allowed under certain circumstances, like for instance
storm waves. One should however be very careful, because small failure can
result in larger failure, as will be explained later on;
- transport of sediment will be allowed, although a sound determination of the
expected siltation and migration of the profile as a whole and the slope in
particular should give insight in improvement of design methods.
Slope Development
Macro-instability Micro-instability Morphology
macro-shear flow slide / retrogressive micro-shear seepage transport of

failure liquefaction breaching failure sediment
> threshold of
motion
increase of
pressure gradient
pore pressures
slope
geometry increase of bed
shear stress
soil
properties
waves currents
depth D < Drequired

width W (D > Drequired) < Wrequired
failure!
‘unwanted’ repair activities

or
‘expected’ maintenance dredging
Figure 2-7: Instability mechanisms
2.3 Experiences from dredging practice
To gain a first insight in occurred slope developments, some examples from dredging
practice will be treated. Of course a lot of parameters are involved, so guidelines cannot
12
be deduced from only a few experiences. Besides, understanding of the important

processes is far more useful when designing a submarine slope in the future.
Nowadays, in engineering practice the design of a slope generally is based on:

1) observed side slopes along existing channels in similar areas;
2) results of geotechnical investigations in the area concerned;
3) observed slopes during trial dredging and maintenance dredging.
Despite all calculation efforts, the third method still seems to be most reliable.
Furthermore, it is a useful tool to check the computations and gain knowledge for future
projects. Dredging of trial trenches however is rather expensive and only occurs when
the client has interest in reducing costs and/or risks.
The following examples of projects in the past give an idea of slopes, mechanisms and
problems to be expected.
Trial slopes and slope production cuts 1980

In 1980 some trial slopes were dredged and a slope stability test was executed by
Rijkswaterstaat, Svasek B.V. and Breejenbout Dredging Company to advise on
economical underwater slopes [Boehmer et al., 1983]. Routine soil investigations and
calculations resulted in uneconomical slopes of 1:10 to 1:15. Several tests have been
done, like box cuts, slope cuts and slope production cuts to reduce the uncertainties and
to demonstrate the feasibility of dredging steep slopes, despite the earthquake
intensities and loosely packed sand layers. These slope stability tests should reveal soil
properties which cannot be derived from field exploration or laboratory tests.
Because the first two of the above mentioned tests are also dredging methods that have
been and still are widely used techniques to create submarine slopes, they will be
explained briefly, see Figure 2-8.
Box cuts are dredged vertical slopes over a limited height which will result in slides
causing a sediment gravity flow and slopes that will be flatter than slopes calculated
from slip circle analysis (see Paragraph 4.1). Box cutting is the cheapest and easiest
way of dredging submarine slopes, because of the high ‘slope production’. Slope cuts
are applied when there are spatial restrictions. Steep slopes for example in harbours,
dry docks or in the vicinity of constructions are dredged by a trailing suction hopper
dredger. Production rates will be lower and the resulting slopes often fail after the
dredging has finished. This construction method is not considered to be very economic.
Sea
Seabed
Box Cut Slope Cut
Figure 2-8: Schematization of methods of ‘box cut’ and ‘slope cut’
The third testing method is a special slope stability test: in a slope production cut, a
flow slide is simulated which will be representative for e.g. earthquake or sedimentation
conditions. Slopes will flatten and the operational reliability of the harbour may be
harmed.
When the box cut tests were done, it appeared that the resulting slopes were of the
same order of the slopes calculated in slip circle analysis (1:1,6 to 1:3 in loosely packed
sand (RD = 20-30%) and 1:2 to 1:4,5 in medium packed sand (RD = 50-60%),
although it was expected otherwise as was mentioned before. Therefore, the slope cuts
have been omitted.
13
The slope production cuts were executed by supplying dredged spoil on top of the slope,
resulting in a turbidity current and much flatter slopes: 1:5-6. A significant liquefaction
flow slide wasn’t observed. The flatter slopes can only be explained by the large erosion
capacity of the sediment gravity flow.
The most relevant conclusions were:

1. slope stability tests give good insight in slope damages in case of calamities and
the corresponding repair and maintenance costs;
2. safety factors against failures appeared to be higher than calculated;
3. the steepness of the designed slopes was increased to 1:5, while first
calculations pointed at 1:10, because of the occurrence of earthquakes and the
low relative density.
Pipeline trench at Danish North Sea Coast 1982

In 1982 a 1600 m long trench was dredged across the surf zone at the exposed North
Sea Coast in Denmark to bring an oil and gas pipeline ashore [Mangor, 1986]. The
backfilling and the wave climate were monitored over a period of 16 months. The trench
was dredged up to 8 m deep in sand with a d50 of 200 μm. The theoretical design slope
was 1:5, but after dredging, the slope was about 1:2, so the design width of 28,5 m
was 50 m. Additional sediment trapping reservoirs were dredged to allow some
backfilling, see Figure 2-9. The wave induced longshore current dominated over tidal
and storm surge currents.
Table 2-4: Observed slopes of pipeline trench [source: Mangor, 1986]
at bars at troughs
slope just after dredging 1:1-4 1:1-4
windward leeward windward leeward
after unidirectional storm
1:3-5 1:20-50 1:10-20 1:20-50
after 'two-way' storm ~1:10 1:10-20
Because three more or less parallel surf zone bars were present, the wave breaking and
thus the sediment transport primarily occurred at these bars. This affected the
morphological development of the trench. The windward (or upstream) slopes moved
towards the centre. Especially in a unidirectional, rough wave climate rather steep
slopes can develop, while leeward (downstream) slopes become very flat, because of
the increased erosive capacities of the wave induced current. After a period with large
waves from both sides, the developed side slopes are somewhere in the middle. This
phenomenon will be discussed extensively in later chapters.
Figure 2-9: Cross-section of trench design [source: Mangor, 1986]
14
Puget Sound 1986

In February 1986 a large submarine flow slide occurred in Puget Sound near Seattle
during construction works of a water transfer system of treated wastewater directly into
Puget Sound, thereby restoring the water quality of the Duwamish River [Kraft et al.,
1992]. Two outfalls were constructed to water depths of 34m. The soil consisted of
Pleistocene silty clay, silt, sand and glacial till and a more recent upper layer of medium
packed sand (d50 between 160 and 260 μm; thickness of a few meters) with a relative
density of 35%, which is classified as a medium dense sand. The maximum slope at
water depths of 23-120 m was about 1:3,5, see Figure 2-10.
Based on these conditions there was no indication that a slide was likely. Further
research after the slide showed that in this area no slides have occurred for the last 50
years. However, a major slide occurred, which was very hard to explain. A possible
sequence of events causing a flow slide is described as follows:
“Failure of the dredged side slopes, face of the trench, or spoil piles adjacent to the
trench could result in material flowing down the slope or trench gradient, perhaps with
the assistance of the outgoing tidal current. This flow of suspended material has more
erosion and disturbance potential than water alone and could have eroded steeper
slopes in the deeper water to an oversteepened condition such that the near-surface
sediments could collapse, inducing pore-water pressures. With the decrease in effective
stresses resulting from the removal of only one meter or so of material in deeper water,
the excess pore water pressures might then be sufficient to cause deeper sliding, in turn
moving support to the upslope soils with the slide quickly regressing upslope or
shoreward, typical of a retrogressive flow slide, until an equilibrium condition is
reached.”
Figure 2-10: Cross-section of seabed at Puget Sound [source: Kraft et al., 1992]
Without explaining all of the above mentioned phenomena, which will be done later on,
this sequence of events clearly shows the complex character of the predictability of flow
slides. Often a number of unfavourable situations have to occur. From this case the
following can be learnt:
- Start all dredging activities in shallow water and proceed sequentially toward
deeper water, so work downslope rather than upslope, because any failure of
the side slopes or face of the trench and the resulting gravity flow of failed
15
material will remain inside the trench itself instead of flowing down
unobstructedly along the natural sloping bottom.
- Piles of spoil should not be sidecast in a loose condition next to the trench, but
have to be removed from the project vicinity. Although the design often
anticipates for cautious side casting, the procedures during construction are
most of the times less strict.
- Reduce or stop dredging activities during periods with very low tidal elevations,
because of the presence of excess pore water pressures in the less permeable,
denser deeper soil layers, resulting in a reduction of stability.
One can also question the use of loosely packed sands (waste material) in backfilling of
trenches, because of the high susceptibility to liquefaction (see Paragraph 4.2). Gravel
is much more suitable backfilling material. As a compromise discrete zones of gravel
can be placed within the trench, backfilled with loose dredge spoils.
Postma and his ‘equilibrium slopes’ 1987

Postma (1987) investigated several dredged pipeline trenches and found that the time
that went by to reach an equilibrium profile was strongly dependent on the weather
(wind waves) and that this equilibrium profiles showed wide variation, see Table 2-5.
Table 2-5: Equilibrium slope values of sandy bed material observed in dredged trenches for
pipelines [source: Postma, 1987]
Initial Equilibrium
Location Soil Data Dredging Depth
Slope Slope
Ameland, NL D50= 150 - 185 μm to 10m -NAP 1:3 1:13 or 1:14
Bacton, UK D50 = 570 μm to 10m -NAP 1:3 1:6 or 1:7
Great Yarmouth, UK D50= 260 -7500 μm to 30m -NAP 1:2 1:5 or 1:6
Hythe, UK silty fine sand (becomes lumpy) to 15m -NAP 1:3 1:4 or 1:5
Callantsoog, NL D50= 300 -350 μm to 15m -NAP 1:4 1:7 or 1:8
In this thesis, it will be shown that in fact there is no such thing as an ‘equilibrium
slope’, because the only equilibrium is the original seabed. Besides, unless there is a
completely symmetrical wave and (tidal) current pattern, one should distinguish
between the upstream, which is predominantly located at the upstream side of the
trench with respect to the current, and the downstream side slope.
It is however often useful to speak of an ‘equilibrium upstream/downstream slope’,
because under certain conditions slopes tend to develop to a specific slope, which
remains more or less constant during the sedimentation process, until this slope starts
‘interacting’ with the other slope or the trench bottom and will flatten.
Pusan 2002
Royal Boskalis Westminster NV has also dredged a lot of trenches and channels.
Because of the huge spread in natural conditions, soil properties, dimensions and design
purposes, only a few will be treated here.
In 2002 two trenches had to be dredged by trailing suction hopper dredger ‘Barent
Zanen’ in very soft clay and silt (2,5 million m3) as part of a harbour project in Pusan,
South Korea. Two breakwaters had to be constructed, but because of the weak clayey
and silty soil, a soil improvement had to be carried out. The life span of these trenches
was not very long, because they were filled up with sand, which was borrowed from an
area 105 km away, to form a solid foundation layer for the breakwater. The slope
heights were 15 m at most and the surrounding water depth was also approximately 15
m. There was hardly any tidal movement and wave heights were very small. The slope
development was mainly caused by soil mechanical aspects. This layer of clay and silt
had a very low undrained shear strength of about 5 kPa and an undrained shear
strength increase of 2,5 kPa/m.
Calculations were executed on this very poor input data. This resulted in slopes of 1:3
(14% probability of failure), 1:4 (8,8 % probability of failure) and 1:5 (6,4% probability
of failure). It was decided to dredge slopes of 1:4, because it was decided that the
probability of failure should not exceed 10% in case of very poor input data. The
dredging work has been executed and the slopes remained stable.
16
Limiting slopes at ebb-tidal shoals

Buonaiuto and Kraus [2003] investigated a wide range of tidal inlets and ebb-tidal
shoals on sandy coasts. They found that slopes of recently dredged entrance channels
ranged from 1:7 to 1:10, with stabilized inlets sustaining greater slopes than
unstabilized inlets. Wave-dominated environments sustain steeper slopes than tidally
dominated regimes. Channels that are maintained by dredging show somewhat steeper
slopes that typically decreased by 0,5-1 °/year.
Zuidwal 2004
In order to increase the accessibility of Gas Platform Zuidwal in the Wadden Sea an
entrance channel had to be deepened to MSL -7 m, while the water depth varied from 2
to 7 m. The side slopes were designed at 1:6, although no significant wave action was
present due to the favourable direction of the channel.
Dredging was performed in a two-layer operation. First the entire area was deepened to
MSL -5 m, then the remaining 2 m was dredged. Slopes just after dredging varied from
1:2,5 to 1:5 with small almost vertical parts. After some time the slopes flattened to
1:2,7-13. This flattening process was attributed to manoeuvring with barges. This
example clearly shows that especially during construction in shallow water the dredging
activities may be more harmful to the side slopes than the hydrodynamic conditions.
Figure 2-11: Two cross-sections of channel to Zuidwal Gas Platform; the steep right slopes
have just been dredged
All these experiences give some insight in the variety of conditions in engineering
practice and show that the knowledge of the underlying processes that cause slope
failure is still insufficient. In the past, only a few efforts have been made to determine
optimal side slopes under different conditions. But even if these processes are
understood, dredging of a trial trench will always be recommendable, combined with
extensive calculations, at least when there is enough budget.
17
3 Reference Cases
3.1 Definition of reference soils
3.1.1 Classification of soil types
There are different classifications of soil types; most use the grain size as governing
parameter. In this thesis the soils under investigation are sand, silt and clay, which are
classified by the Permanent International Association of Navigation Congresses (PIANC)
in Brussel [1984], see Table 3-1, in which the ‘white’ rows are the soil types under
investigation:
Table 3-1: Soil classification according to PIANC 1984; the high-lighted soils are treated in
this thesis
Particle identification
Main soil type Grain size [μm]
range of size
Boulders > 200.000
Cobbles 60.000 - 200.000
Coarse 20.000 - 60.000
Gravels Medium 6.000 - 20.000
Fine 2.000 - 6.000
Coarse 600 - 2.000
Sands Medium 200 - 600
Fine 63 - 200
Coarse 20 - 63
Silts Medium 6 - 20
Fine 2-6
Clays <2
Peats and Organic soils varies
As can be seen in the PIANC classification, grain size isn’t a good criterion to distinguish
between clays, peats and some silts; other characteristics are just as important. The
main difference between clay and silt is the plasticity: ‘Dry silt can easily be dusted off
fingers and dry lumps powdered by finger pressure, whereas dry clay sticks to fingers
and dry lumps do not powder, but shrink and crack.’ [PIANC].
The International Association of Dredging Contractors established a system for

identifying the parameters which determine the dredgeability. For the soil types at
issue, these are the following:
Sand
a. unit weight
b. water content w
c. specific gravity of grains
d. grain size
e. water permeability
f. frictional properties
g. lime content
h. organic content
In this thesis all above parameters but the lime and organic content are under
consideration.
18
Silt
a. unit weight
b. water content
c. grain size
d. water permeability
e. sliding resistance
f. plasticity
g. lime content
h. organic content
Silts are treated as a completely homogeneous material without any lime and organic
contents or plasticity. In literature often confusing definitions can be found. Most of the
time, silt as a soil classification is mixed up with ‘silt’ (Dutch: ‘slib’) as a deposited
sediment in harbour basins. Then ‘silt’ often represents a random mixture of organic
matter and inorganic sand and silt particles, sometimes called ‘mud’. It should be clear
that in this thesis silt is only treated as a type of soil, characterized by its grain size.
Clay
a. weight
b. water content
c. sliding resistance
d. consistency ranges (plasticity)
e. organic content
Clays show a huge spread in all above parameters. The very hard and stiff clays aren’t
interesting when considering slope stability; under almost all circumstances vertical or
nearly vertical walls will be stable. Therefore the focus will be on the very soft to firm
clays.
3.1.2 Soil properties
Now it is clear what is understood by sand, silt and clay, some reference soils will be
defined and sound assumptions of the various corresponding soil properties will be
made. These reference soils will be used throughout this thesis to investigate slope
stability problems, morphological behaviour and eventually to deduce guidelines for
slope design. In Table 3-2 can be seen that the in total 15 reference soils are divided
into grain size and porosity: 5 different grain sizes and 3 different porosities. For some
instability mechanisms porosity is the leading parameter; in morphology the grain size
of course is the most important soil property. Next some corresponding values for all
other relevant soil properties were deduced from literature [Das, 1994; Terzaghi et al.,
1996; Verruijt, 1999] and examples from engineering practice. Most of them will be
explained here.
Density solid material

All reference sediments are assumed to originate from weathered rock of 2650 kg/m3,
which means that the solid particle density is, of course, 2650 kg/m3. Also for clays, this
value is a good first approximation.
Porosity
The porosity is the quotient of the pore volume Vp and the total volume of a soil sample
Vt:
Vp
n= Eq. 3-1
Vt
19
Void ratio
The void ratio is also a measure for the porosity but with the (constant) volume of the
grain particles Vg in the denominator.
Vp
e= Eq. 3-2
Vg
Relative Density
This quantity is related to the void ratio and is an indicator for behaviour under shear
deformation (dilatant or contractant).
e max − e
RD = Eq. 3-3
e max − e min
The maximum and minimum void ratios, emax and emin, should be determined in
laboratory experiments. In this thesis, these values have to be assumed. The theoretical
values for nmax and nmin can be calculated by considering equally large spherical grains:
the loosest packing is the ‘cubic’ one with a porosity nmax of 1-π/6 = 0,4764 and the
densest packing is the ‘rhombic’ one (connection lines of grains form an regular
tetrahedron) with a porosity of nmin = 1-π/18 = 0,2595. Using the following relation
e=n/(1-n) results in emax = 0,9099 and emin = 0,3504.
According to the Japanese Society of Civil Engineering (JSCE) the maximum void ratio
emax of fine to coarse sands is about 0,95 (nmax = 0,49). When the sand contains silt,
this void ratio is about 0,90 (nmax = 0,47). In this thesis a maximum porosity of 0,48
will be used. The minimum porosity nmin is set to 0,30, so emin = 0,43.
Often the classification below is used to indicate loosely, medium and densely packed
sands and silts. For cohesive soils, like clays, this approach does not hold, because of
the complexity of the soil skeleton and the difficulties in determining the maximum and
minimum porosity.
RD < 0,33 : loosely packed sand / silt

0,33 < RD < 0,67 : medium packed sand / silt
0,67 < RD : densely packed sand / silt
Table 3-2: Reference soils and their soil properties (see Appendix C for complete table)
di
m
am
ed
et
ia
er
n
pa
at
pa
rti
rti
cl
hi
cl
e
un
ch
co
e
di
un
dr
90
fa
ns
am she hes
hy
de
sa
sa
ai
in
ll
%
ol
dr
ne
tri
ns
et
tu
tu
ve
id
sa
au
ns
by
er
ra
ra
d
ity
lo
at
re
tu
te
te
d
ic
lic
ci
of
io
w
so
ry
la
ra
fri
d
ty
pe
ei
n
co
to
tiv
c
lid
te
dr
ar
un
un
un
gh
at
tio
rm
ta
nd
vo
co
oe
d
e
y
it
it
it
st
m
po
ti
lw
T=
n
de
de
de
ea
id
uc
f
w
w
w
re
at
f
an
ro
ic
18
ei
ei
e
ei
ns
ns
ns
bi
ng
ra
t
fin
er
iv
ie
ig
si
g
gh
gh
io
gl
lit
°C
tio
ia
ity
ity
i ty
ity
er
nt
th
ht
ht
ty
n
e
y
t
t
l
Name Description ρs n e RD ρsat γsat ρdry γdry γunsat ϕ c su d50 d90 k κ cv ws

3 3 3 3 3 3 2
kg/m - - - kg/m kN/m kg/m kN/m kN/m ° kPa kPa μm μm m/s m m/s
LCS loosely packed coarse sand 2650 0,45 0,82 0,16 1908 19 1458 15 17 30 0 0 1000 1500 1,9E-04 1,0E-08 3,8E-01 0,1084
MCS medium packed coarse sand 2650 0,40 0,67 0,43 1990 20 1590 16 18 35 0 0 1000 1500 3,5E-05 1,8E-09 1,2E-01 0,1084
DCS densely packed coarse sand 2650 0,30 0,43 0,86 2155 22 1855 19 20 40 0 0 1000 1500 1,2E-05 5,5E-10 5,9E-02 0,1084
LMS loosely packed medium sand 2650 0,45 0,82 0,16 1908 19 1458 15 17 30 0 0 500 750 1,4E-05 7,5E-10 2,9E-02 0,0663
MMS medium packed medium sand 2650 0,40 0,67 0,43 1990 20 1590 16 18 35 0 0 500 750 8,8E-06 4,4E-10 3,1E-02 0,0663
DMS densely packed medium sand 2650 0,30 0,43 0,86 2155 22 1855 19 20 40 0 0 500 750 3,0E-06 1,4E-10 1,5E-02 0,0663
LFS loosely packed fine sand 2650 0,45 0,82 0,16 1908 19 1458 15 17 30 0 0 200 300 2,3E-06 1,2E-10 4,6E-03 0,0252
MFS medium packed fine sand 2650 0,40 0,67 0,43 1990 20 1590 16 18 35 0 0 200 300 1,4E-06 7,1E-11 5,0E-03 0,0252
DFS densely packed fine sand 2650 0,30 0,43 0,86 2155 22 1855 19 20 40 0 0 200 300 4,7E-07 2,2E-11 2,4E-03 0,0252
LSI loosely packed silt 2650 0,45 0,82 0,16 1908 19 1458 15 17 27 0 0 50 75 1,4E-07 7,5E-12 2,9E-04 0,0029
MSI medium packed silt 2650 0,40 0,67 0,43 1990 20 1590 16 18 32 0 0 50 75 8,8E-08 4,4E-12 3,1E-04 0,0029
DSI densely packed silt 2650 0,30 0,43 0,86 2155 22 1855 19 20 37 0 0 50 75 3,0E-08 1,4E-12 1,5E-04 0,0029
SOC soft (organic) clay, low plasticity 2650 0,75 3,00 1413 14 663 7 11 18 5 5 2 3 3,8E-09 2,7E-13 7,6E-07
MEC medium clay, medium plasticity 2650 0,60 1,50 1660 17 1060 11 14 18 7 20 2 3 9,0E-10 5,4E-14 4,5E-07
STC stiff clay, high plasticity 2650 0,40 0,67 1990 20 1590 16 18 20 10 50 2 3 1,4E-10 7,1E-15 1,4E-07
20
Hydraulic permeability
This property is dependent on both soil and fluid characteristics according to:
κγ n3
k= w in which κ = c K −C d2
(formula of Kozeny-Carman) Eq. 3-4
μ (1 − n) 2
The coefficient cK-C is taken 0,01, although it will show variations in reality. Normally the
hydraulic permeability will be determined in tests, which gives more accurate results
than the above formulas, but this approach yields values of the right order of magnitude
and clearly shows the influence of grain size and porosity.
Consolidation coefficient
The approximation of the consolidation coefficient of all reference soils is based on the
assumption that hydraulic permeability ‘k’ and the coefficient of compressibility ‘mv’
remain constant during the consolidation process and that the pore water is
incompressible (β = 0). The following formula can be deduced from the 1-dimensional,
vertical differential equation of consolidation.
k
cv = Eq. 3-5
γ w (mv + nβ )
The coefficient of compressibility mv is estimated according to the linear elastic stress-

strain-relation:
1
mv = Eq. 3-6
K + 43 G
in which the bulk modulus K and the shear modulus G can be expressed in terms of
Young’s modulus E and Poisson’s ratio ν:
E E
K= and G= Eq. 3-7
3(1 − 2ν ) 2(1 + ν )
The values of E and ν are taken from literature. Again this approximation is a rather
strong simplification of reality, but yields reliable input parameters.
Particle fall velocity

To gain insight in the spread on the input value for the particle fall velocity three
formulae have been considered for the reference soils. Van Rijn [1993] distinguished
three ranges of particle diameters: d50 ≤ 100 μm, 100 ≤ d50 ≤ 1000 and d50 ≥ 1000 μm.
For diameters smaller than 100 μm he used the well-known Stokes equation:
( s − 1) gd 2
ws = Eq. 3-8
18ν
and for diameters between 100 and 1000 μm:
10ν ⎡⎛ 0,01( s − 1) gd 3 ⎞ 0,5 ⎤

ws = ⎢⎜⎜1 + ⎟⎟ − 1⎥ Eq. 3-9
d ⎢⎣⎝ ν2 ⎠ ⎥⎦
in which s is the specific density (s=ρs/ρw) and the kinematic viscosity ν [m2/s] is
dependent on the water temperature T [°C]:
ν = (1,14 − 0,031(T − 15) + 0,00068(T − 15) 2 ) ⋅ 10 −6 Eq. 3-10
21
This results in a kinematic viscosity of 1,31*10-6 m2/s in water of 10 °C and 1,05*10-6

m2/s in water of 18 °C.
Another well-known formula is developed by Delft Hydraulics in 1983 during

experiments in water of 10 °C and 18 °C. These experiments were only executed on
diameters varying from 50 to 350 μm.
log(1 / ws ) = A log 2 d + B log d + C Eq. 3-11
Table 3-3: Values of constants in equation 3-11 [source: Sistermans, 2002]
T=10°C T=18°C
A 0,476 0,495
B 2,18 2,41
C 3,19 3,74
The mentioned coefficients are for fresh water only. In Figure 3-1 the results of both
formula have been plotted for two temperatures; the formula of DH’83 in fact has been
extrapolated outside its range of 50 to 350 μm. The variations are quite small,
especially for the smaller particle diameters. Although the salinity of the water, the
shape or angularity of the particles and the sediment concentration have been
disregarded in this approach, they also have a small influence the particle fall velocity.
0,12
0,10
particle fall velocity ws [m/s]
0,08
0,06
0,04 Van Rijn; T=10°C

Van Rijn; T=18°C
0,02 DH'83; T=10°C
DH'83; T=18°C
0,00
0 100 200 300 400 500 600 700 800 900 1000
particle diameter d50 [μm]
Figure 3-1: Particle fall velocities in fresh water of 10°C and 18°C according to Van Rijn
[1993] and Delft Hydraulics [1983]
Grading
All reference soils are defined as homogeneous, which implies a uniform particle size. In
reality no such soils can be found. Therefore a narrow grading is assumed:
d 90
= 1,5 Eq. 3-12
d 50
With the definition of the reference soils, calculations can be done. It is however not
very useful to consider all fifteen soils in all cases. Macro stability calculations for
instance aren’t really sensitive to grain size, while transport calculations are extremely
sensitive to this quantity. On the other hand, morphology isn’t very much affected by
porosity; only the bed changes will proceed faster. Susceptibility to liquefaction is
extremely dependent on porosity as long as sand is under consideration; clays will not
be susceptible to liquefaction (quick clay which shows some similarities to flow slides
isn’t considered in this thesis). Therefore a selection of the appropriate reference soils
will be made at each subject. Sometimes the sensitivity to a single soil property is
investigated, when this particular property is of utmost importance.
22
3.2 Definition of reference geometries
Of course trenches and channels have different dimensions and it would be impossible
to derive guidelines for a dimensionless trench. Therefore a typical trench and a typical
navigation channel are defined, see Figure 3-3 for used symbols.
Trench
A trench often is only a few meters deep and not more than 10 meters wide. If trenches
are dredged instead of pipeline burial after the laying process, this often occurs in
shallow water, for instance outfalls. Because the trench should be located in the non-
breaking area, the depth will be 5 m, assuming that no waves larger than 2,5 m will
pass over the trench (either they do not occur or they are already broken). The depth of
the trench will be 5 m and the width is assumed to be 10 m, see Figure 3-2.
Navigation channel
Well-known examples of navigation channels are the ‘IJ-channel (Port of IJmuiden) and
the 54-km long ‘Euro-Maas’-channel (Port of Rotterdam). The depth of the former is
kept at 19m, navigable for ships with a draught of 16,5m; the depth of the latter is 24
m for ships with a draught of 22,5 m. The surrounding seabed is of course sloping
downward into the sea, but typical heights of side slopes of 10 m are not exceptional.
The width of the ‘Maasgeul’ (last 14 km) is rather larger, viz 400-600 m; in this thesis
the width is set to 200 m. Larger widths result in larger computational time and less
accuracy around the slopes; smaller widths prevent autonomous slope development
(slopes start interacting with each other).
Compared to a pipeline trench, not only the dimensions of the slopes are quite different,
also the morphological behaviour will differ a lot, because of the larger trapping
efficiency. In order to keep the amount of calculations/simulations within reasonable
limits and because the morphological development becomes more complex for very
small and sharp trenches, the morphological behaviour is only studied extensively for
the navigation channel. Most geotechnical calculations have been executed for slope
heights of 5, 10 and 15 or 20 m.
0 50 100 150 200 250 300 350 400
10
0
-10
-20
-30 Navigation Channel Pipeline Trench Sea Level

-40
Figure 3-2: Geometries of ‘navigation channel’ and ‘pipeline trench’ with side slopes 1:5
Figure 3-3: Schematization of some important parameters
23
4 Unloaded macro- & micro-instability

In this paragraph the focus will be on macro-instability. Unlike micro-instability, macro-
instability should be prevented in any case, because large movements of soil will
certainly endanger the functionality of the trench or channel. The resulting slope after a
failure will be flattened to such a degree, that the required depth cannot be guaranteed.
4.1 Shear failure
4.1.1 Considerations on shear failure
The most common instability mechanism for slopes is shear failure. Before calculating
slope stability factors, a number of aspects have to be considered to be able to make
sound assumptions on the input values.
- Drained and undrained conditions

One should consider the soil behaviour for each design condition. Generally this
behaviour is determined by the combination of the permeability of the soil and the
speed of the (dynamic) load or construction. In other words, can water pressures
dissipate during loading or construction? The permeability of sands is generally a factor
105 higher than that of clays.
As a rough guideline for design during and just after construction [Slope Stability
Engineering Manual, 2003], soils with values of permeability greater than 10-6 m/s
usually will be fully drained and soils with values of permeability less than 10-9 m/s will
be primarily undrained. This means that in static design calculations all reference
‘sands’ (see Paragraph 3.1.2) will show drained behaviour, while all ‘clays’ show
undrained behaviour. The permeability of the reference ‘silts’ lies in between these two
values. In this thesis it is assumed that ‘silt slopes’ of dredged trenches behave drained,
based on the following line of thought. In the first place, the sea bottom consists of
totally consolidated soil layers because of its age, in contrary to, for example, surcharge
loads, which will cause consolidation of the underlying soil and therefore extra water
pressures. In the second place, when dredging a trench or channel and extracting a
significant mass of soil, the seabed will be slightly overconsolidated and, as a
consequence, the seabed will relax, thus increasing pore volume and reducing water
pressures. Therefore a drained calculation seems appropriate for silts.
In the long-term condition, all soils will be fully drained by definition, regardless their
permeability. For clays, also this condition has to be considered: the less stable
condition (construction or long-term) should be normative for design.
Drained analyses use drained strengths related to effective stresses, which can be
obtained from consolidated-drained tests (CD) or consolidated-undrained triaxial tests
on saturated specimens with pore water pressure measurements (CU); undrained
analyses use undrained strengths related to total stresses, which can be obtained from
unconsolidated-undrained tests (UU).
The linear strength envelope according to the Mohr-Coulomb failure criterion becomes:
for sands and silts: τ = c'+σ ' tan φ , in which c’ = 0 for cohesionless soils;
for clays: τ = c = su , in which su is the undrained shear strength.
Although not considered in this thesis about purely theoretical, homogeneous and
isotropic cases, one should in real situations be aware of:
24
- Ductile and brittle stress-strain behaviour

For soils with ductile stress-strain behaviour (shear resistance does not decrease
significantly as strain increases beyond its peak value), like soft clays, loose sands and
clays compacted at water contents higher than optimum, the peak shear strength may
be used in slope stability calculations. For soils with brittle stress-strain behaviour
(shear resistance decreases significantly as strain increases beyond its peak value), like
stiff clays, dense sands and clays compacted at optimum water content or below, a
shear resistance lower than the peak shear strength should be used, because of the
possibility of progressive failure. This kind of failure occurs under conditions where
shearing resistance first increases and then decreases with increasing strain. As a result
the slope isn’t able to mobilize its peak shear strength at all points along the slip
surface.
- Strength anisotropy
The shear strength of soils may vary with orientation of the failure plane, particularly in
presence of fissures.
4.1.2 Unloaded shear failure
The unloaded (static) shear failure will be investigated using a slip-circle analysis
(Bishop-method). Although shear movements may occur across a zone of appreciable
thickness, shear failure is considered to occur along a discrete surface, a so-called slip-
circle. The idea is based on stability expressed as a safety-coefficient F, which is
strength divided by load. The load is determined by the weight of the soil of a certain
slip circle; the strength by the shear along that same circle. Therefore a slip circle is
divided into a lot of slices and for each of the slices the weight and the shear is
calculated. The stability coefficient F will be computed for a lot of circles and the lowest
value indicates the safety against failure: Fmin > 1 means that the slope profile is
theoretically stable, but often higher values for the stability coefficient are applied. In
this thesis, in case of macro-stability, a safety factor of 1,5 will be considered
unconditionally stable: the uncertainty in dredging works on one hand and the often
limited knowledge of the soil properties on the other hand require such a conservative
stability criterion. In case of micro-stability, a safety factor of 1,25 will be considered
sufficient safe, because of the smaller consequences, when failure occurs.
Please note that lower values of safety factors for slopes in existing channels are
acceptable, because one has been able to observe the actual performance of the slope
over a period of time. It is also recommended to apply lower safety factors for
temporary pipeline trenches.
Both Fellenius and Bishop follow this calculation method mentioned above, although
Bishop takes the reduction factor (F) and the fact that the slice surface makes an angle
(β) with the gravity force into account. Because this reduction factor isn’t known
beforehand, iteration is necessary. Unlike Fellenius, Bishop’s method preserves
equilibrium of vertical forces:
nlayers
bτ slip
R∑
cos β
F= 1
Eq. 4-1
M soil + M water + M load
R is the radius of the slip circle and τ is the shear strength along the slice bottom that
depends on the angle of the slice bottom and can be defined as:
c + σ v' tan φ
τ slip ;α = Eq. 4-2
F + tan β tan φ
in which underscore ‘α’ is used to express the angle of the bottom of a slice.
25
The three moments in the denominator are the driving moments. Msoil is the driving
moment caused by the mass of soil within the slip circle around its centre. This moment
increases if the unit weight of the soil increases. Mwater is the driving moment caused by
the water forces acting from the outside on the surface of the soil structure. External
loads can cause an additional driving moment Mload.
A lot of calculations have been made with the computer-program MStab by GeoDelft.
The two reference cases ‘trench’ and ‘navigation channel’ as well as some other slope
heights D have been investigated for different slope angles and soil characteristics.
Calculations under drained conditions have been done for all reference soils;
calculations under undrained conditions only for cohesive soils as was discussed in
Paragraph 4.1.1.
Drained conditions
One of the first things to keep in mind is that the water level does not play a role, as
long as the (static, flat) water level is situated completely above the slope; the safety
factors remain the same, when considering static stability. Later on in this paragraph,
the influence of a static water level under the top of the slope will be discussed.
In Table 4-1 can be seen that for non-cohesive soils, like sand and silt, the angle of
internal friction is the main parameter. In fact, when considering static macro-stability,
it is the only important variable. In a completely homogeneous soil, the safety factor
isn’t affected by the unit weight: the driving moment of the soil increases, when the unit
weight increases, but so is the resisting moment, which is dependent on the weight of a
slice. Heavier soils however are often more densely packed and are characterized by a
larger friction angle. This is the reason why a medium packed fine sand (MFS) is more
stable than a loosely packed fine sand (LFS). And the latter on his turn is more stable
than the equally loosely packed silt. In non-cohesive, homogeneous soils a sound
determination of this angle will be fruitful.
Table 4-1: Safety factors for different slope angles and reference soils; gray cells represent
safety factors larger than 1,5
γsat c φ tan φ Trench Navigation channel

3
kN/m kPa ° 1:x 1:1,5 1:2 1:2,5 1:3 1:1,5 1:2 1:2,5 1:3
LFS loosely packed fine sand 19 0 30,0 1,73 0,87 1,16 1,44 1,73 0,87 1,16 1,44 1,73
MFS medium packed fine sand 20 0 35,0 1,43 1,05 1,40 1,75 2,10 1,05 1,40 1,75 2,10 micro-stability
DFS densely packed fine sand 22 0 40,0 1,19 1,26 1,68 2,10 2,52 1,26 1,68 2,10 2,52
LSI loosely packed silt 19 0 27,0 1,96 0,77 1,02 1,27 1,53 0,77 1,02 1,27 1,53
MSI medium packed silt 20 0 32,0 1,60 0,94 1,25 1,56 1,88 0,94 1,25 1,56 1,88
DSI densely packed silt 22 0 37,0 1,33 1,13 1,51 1,89 2,26 1,13 1,51 1,89 2,26
SOC soft clay 14 5 17,5 2,42 2,70 2,95 3,19 1,54 1,82 2,04 2,26
slip c.
MEC medium clay 17 7 18,0 2,15 2,43 2,66 2,90 1,45 1,68 1,90 2,11
STC stiff clay 20 10 20,0 2,26 2,55 2,82 3,08 1,54 1,79 2,03 2,26
What also can be concluded from Table 4-1 is the fact that slopes steeper than the
corresponding ‘natural slope’ are of course not stable, but this instability mechanism can
better be characterized by micro-instability (see Paragraph 4.4), because failure occurs
at the slope surface with relatively small amounts of particles (in reality, grain particles
would ‘rain down’ along the slope, until the angle of friction is reached). The so-called
slip circles only slightly touch the slope surface. But it appears that in non-cohesive soils
even at much flatter slopes, the normative failure mechanism is micro-instability, in
contrary to cohesive soils.
The clays (SOC, MEC and STC) always collapse with nice slip circles in which a huge
amount of material is involved. Also can be concluded that cohesive soils are more
stable; even a steep slope of 1:1,5 can still be considered ‘safe’. Such thing as a
‘natural slope’ is hard to define for cohesive soils, because it also depends on the slope
height.
26
All values are plotted in Figure 4-1 for the “navigation channel”. It can be seen that
safety factors increase when slopes are flattened, but this increase is more significant
with non-cohesive soils. Cohesive soils are not only dependent on the friction angle, but
also on cohesion and unit weight.
3,0
2,5
Safety Factor SF [-]
2,0
1,5
1,0
0,5
0,0
1,0 1,5 2,0 2,5 3,0
slope 1:x [-]
DFS DSI MFS SOC STC MEC MSI LFS LSI
Figure 4-1: Safety Factors against critical slopes; navigation channel; all slip circles permitted
For slopes less steep than the ‘natural slope’ in non-cohesive soils it is also interesting
to know what the safety against macro-failure is, not only because macro-failure has
larger consequences, but also to compare static stability with dynamic stability.
Therefore a second series of calculations was done in which the tangent line was
restricted to depths of at least 0,5 m under the channel bottom and the centre point of
the slip circle was located above the highest point of the undisturbed surrounding bed
and between the toe and top of the sloping bottom, see Figure 4-2, in which the red
dots are all situated within the permitted area of centre points and the green lines are
all permitted tangent lines. The slip circle characterized by the lowest safety factor is
drawn.
Figure 4-2: Permitted tangent lines and centre points to create ‘nice’ slip circles; example slip
circle in soft clay
Constricting the permitted slip circles in this way resulted in larger slip circles with of
course slightly higher safety factors, see Table 4-2. It can be concluded that only steep
27
slopes (>1:2) of loosely packed sediment run some risk of collapse due to macro-
instability, but most of the times, when no external loads occur, when the soil is
completely consolidated and homogeneous, macro-stability will be guaranteed. Again
for the “navigation channel” the safety factors have been plotted, see Figure 4-3.
In the limiting case of infinite slopes, safety-factors of macro-stability and micro-

stability will coincide. Slope collapse due to macro-instability is most likely in the
presence of weaker soil layers or fluctuating water levels. That’s why in literature slope
failures of this kind have been reported, although this thesis seems to demonstrate that
submarine slope failures due to macro-instability are not very likely; it should be clear
that this is solely caused by the assumption of homogeneous soils that are completely
situated below sea level.
Table 4-2: Safety factors for different slope angles and reference soils with restricted slip
circles; grey cells represent safety factors larger than 1,5
γsat c φ tan φ Trench Navigation channel

3
kN/m kPa ° 1:x 1:1,5 1:2 1:2,5 1:3 1:1,5 1:2 1:2,5 1:3
LFS loosely packed fine sand 19 0 30,0 1,73 1,01 1,47 1,72 1,97 1,12 1,37 1,63 1,90
MFS medium packed fine sand 20 0 35,0 1,43 1,51 1,80 2,08 2,39 1,36 1,67 1,97 2,31
DFS densely packed fine sand 22 0 40,0 1,19 1,81 2,14 2,49 2,86 1,63 2,00 2,37 2,77
slip circles
LSI loosely packed silt 19 0 27,0 1,96 1,10 1,30 1,51 1,74 0,99 1,21 1,44 1,68
MSI medium packed silt 20 0 32,0 1,60 1,35 1,59 1,86 2,13 1,21 1,49 1,76 2,06
DSI densely packed silt 22 0 37,0 1,33 1,63 1,92 2,24 2,57 1,46 1,79 2,12 2,48
SOC soft clay 14 5 17,5 2,42 2,70 2,95 3,19 1,54 1,82 2,04 2,26
MEC medium clay 17 7 18,0 2,15 2,43 2,66 2,90 1,45 1,68 1,90 2,11
STC stiff clay 20 10 20,0 2,26 2,55 2,82 3,08 1,54 1,79 2,03 2,26
3,0
2,5
2,0
1,5
1,0
0,5
0,0
1,0 1,5 2,0 2,5 3,0
slope 1:x [-]
DFS DSI MFS SOC STC MEC MSI LFS LSI
Figure 4-3: Safety Factors against critical slopes with restricted slip circles (only macro-
stability!); navigation channel
When slopes emerge from the sea level, the situation gets less favourable, because of
loss of buoyancy as was mentioned earlier. In a marine environment, a large tidal
amplitude can cause this emergence. This phenomenon plays a role in regions like the
Wadden Sea.
To investigate this problem of fluctuating water levels, a rough approximation was done,
assuming flat water levels and fully consolidated soils. In the initial situation the soil is
saturated and consolidated. When the water level drops, the pore water pressure slowly
will dissipate resulting in consolidation until a situation has developed with a layer of
(consolidated) unsaturated (note: not completely dry) soil on top of a layer of saturated
soil. In case of a slow fluctuation of the water level and relatively permeable soils (like
most sands), pore water overpressures will not develop.
28
The stability is only worsened due to loss of buoyancy. Calculations have been done in
MStab for medium packed fine sand (MFS) and medium packed silt (MSI). The unit
weight of the soil above the freatic level was varied between the saturated and the
unsaturated value. Again distinction was made between macro-stability (with restricted
slip circles) and micro-stability.
It was found, that a water level of 0,4 times the slope height D (so 40% of the slope is
situated below sea level) was most unfavourable, when considering macro-stability. The
Safety Factors for macro-stability were reduced over 15% in case of the saturated soil
and around 13% in case of unsaturated soil with respect to the initial situation. The
saturated soil of course reduces stability more because of the larger driving moment
(γsat > γunsat), while the counteracting moment remains the same.
Furthermore it appeared that micro-stability isn’t normative with water levels

somewhere along the slope, to be more precisely: 0 D ≤ water level ≤ 0,75 D. For these
levels perfect slip circles with more or less equal safety factors were found. If the water
level is well under the channel bottom or above the top of the slope, micro-stability
becomes normative again. This phenomenon is clearly demonstrated in Figure 4-4 for
slopes of 1:3 in MFS and MSI. For water levels between 0 and 0,75D, there is a clear
dip in the safety factor, caused by failure due to macro-instability. Around this
unfavourable zone, the safety factors are governed by micro-instability (see Figure 4-1).
2,4
2,2
2,0
1,8
1,6
1,4
MFS (saturated)
1,2
MSI (saturated)
1,0
-0,50 -0,25 0,00 0,25 0,50 0,75 1,00 1,25 1,50
Relative height of water level wrt to slope height [-]
Figure 4-4: Safety Factors in case of fluctuating water levels in fully saturated MFS and MSI;
navigation channel; 1:3 side slopes
From the above observations it can be concluded that macro-stability is normative in

non-cohesive soils if slip circles can be attracted to weaker points in the slope, in other
words if the soil mass isn’t a completely homogeneous, regular slope. This means that
macro-stability will be normative if weak layers or unfavourable water levels are
present.
Another striking result is the apparently inconsistent fact that under drained conditions
soft clay is more stable than medium clay and that soft and stiff clay seems to be more
or less equally stable in these drained φ,c-calculations. To explore this problem, the
effect of variations of the governing variables is investigated.
In the drained φ,c-calculations, the three soil parameters γsat, φ and c have been
changed (see Table 4-3), starting from the properties of stiff clay: γsat = 20 kN/m3; φ =
20° and c = 10 kPa. The just stable slope angle (βSF=1) and the matching slope are
calculated.
29
Table 4-3: Variations in parameters φ, c and γsat
φ β SF=1 1:x c β SF=1 1:x γsat β SF=1 1:x

3
° ° - kPa ° - kN/m ° -
0 13,4 4,20 0 20,0 2,75 14 87,9 0,04
5 25,5 2,10 2 28,4 1,85 16 78,7 0,20
10 39,8 1,20 4 36,5 1,35 18 69,2 0,38
15 53,1 0,75 6 45,0 1,00 20 60,3 0,57
20 60,3 0,57 8 52,4 0,77 22 53,5 0,74
25 67,2 0,42 10 60,3 0,57
30 73,3 0,30
All values for the three variables are known from literature, but a certain combination of
φ, γsat and c can be very uncommon. In Figure 4-5 these variables are plotted, note the
different y-axis. On the left y-axis the friction angle φ and the saturated unit weight γsat
are indicated, successively the blue and brown line; on the right y-axis (on a different
scale) the cohesion c is indicated, the green line. The point where the three lines
intersect represents the ‘normal’ stiff clay (STC), which is the reference soil.
It can be concluded that:
- a decrease of cohesion means a decrease of stability. A non-cohesive soil (c = 0) is
completely governed by the angle of internal friction: 1/tan φ = 2,75;
- a decrease of the angle of friction means a decrease of stability. A purely cohesive soil
(φ = 0) is completely governed by cohesion: the critical slope depends on the slope
geometry. In this case a slope will be stable when flatter than 1:4,2;
- an increase of (saturated) unit weight means a decrease of stability. Heavier soil
results in a larger driving moment, which is only partly compensated by the larger
friction at the surface of the slip circle.
Because of these conclusions, it is possible that clay with a smaller cohesion, friction
angle and unit weight can be more stable. It just depends on the exact values of the
parameters.
30 15
25
friction angle φ [°] , unit weight γ sat kN/m ]
3
20 10
cohesion c [kPa]
15
10 5
0 0
0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5
slope 1:x [-]
angle of internal friction saturated unit weight cohesion
Figure 4-5: Critical slopes (SF=1) for varying friction angle, unit weight and cohesion
30
Undrained conditions
Comparing drained calculations with undrained calculations can only be justified if both
the drained and undrained soil properties of a certain soil are accurately known.
However, no definite relation between those properties exists; all depends on the
specific stress paths of a certain soil. Because a wild guess of the undrained shear
strength would make no sense, the following relation, not theoretically founded and
therefore only indicative, is used, based on the idea that the isotropic effective stress
will remain more or less constant and can be estimated (no contractant or dilatant
behaviour) as follows [Verruijt, 1999]:
cos φ sin φ
su = c + σ ' Eq. 4-3
1 − 13 sin φ 1 − 13 sin φ
0
The total (constant) isotropic stress is defined as (σxx = σyy):
σ 0 ' = 13 (σ zz '+2σ xx ' ) Eq. 4-4
When the stress-strain relationship is considered linear elastic, the ratio between
horizontal and vertical effective stress (under water) becomes:
ν
K'= Eq. 4-5
1 −ν
in which ν is the Poisson ratio.
undrained shear strength su [kPa]
0 20 40 60 80 100
0
SOC
5
MEC
10 STC
depth [m]
15
20
25
30
Figure 4-6: undrained shear strength su against depth
Together with the saturated unit weight of the reference clays, the undrained shear
strength is known over the entire depth, see Figure 4-6. As a rough criterion one often
uses that the shear strength increases in depth with 0,20 to 0,25 times the vertical
effective stress. In case of the reference soils, this value is a bit less for SOC (0,17) and
a bit more for STC (0,30) with MEC (0,21) in between.
In MStab a number of 36 calculations have been done for different slopes (1:1,5; 1:2;
1:2,5 and 1:3) and different slope heights (5, 10 and 20m), all completely situated
below sea level.
31
Figure 4-7: Bishop slip circle in three-layer seabed in soft clay (SOC)
In MStab the shear strength on top and at the bottom of a soil layer has to be entered.
Because of the varying thickness over the trench, this would result in a varying gradient
(δsu/δz). The shear strength at the trench bottom would increase faster with depth,
thereby increasing stability. Three soil layers were applied to give a good representation
of the undrained shear strength, see Figure 4-7.
The results of the undrained calculations (su) were compared to the results of the
drained calculations (c, φ) and plotted in Figure 4-8 and Figure 4-9; the solid lines
represent the drained, the dashed lines the undrained calculations.
2,5
2,0
1,5
1,0
0,5 SOC (drained) SOC (undrained)

MEC (drained) MEC (undrained)
STC (drained) STC (undrained)
0,0
1,0 1,5 2,0 2,5 3,0
Slope 1:x [-]
Figure 4-8: Safety factors for different initial slopes and reference soils under drained and
undrained conditions
It can be observed that in all reference clays rather steep slopes remain stable in both
drained and undrained calculations. Slopes of 1:2 with a height of 10 m are
unconditionally stable in all considered clays (knowing that these clays were not that
stiff or firm). In case of undrained calculations, the safety factors are somewhat lower,
especially for flatter slopes. Flattening the side slopes is a more effective measure if the
soil shows drained behaviour: the ‘drained’ lines increase faster.
32
Again the medium clay (MEC) was the weakest of the three, due to the combination of a
rather large saturated unit weight and small increase of the undrained shear strength
with depth (in the linear elastic relation dependent on cohesion, friction angle, Poisson
ratio and unit weight) compared to the soft clay (SOC). STC and SOC show similar
behaviour: the larger strength-related properties (c, φ and su) of STC are almost
completely counterbalanced by its large load-related property (unit weight). One should
keep in mind that this statement only holds for the unloaded static situation. Large
dynamic loads can have a larger impact on softer clays.
When varying the slope height (and keeping the angle of inclination constant at
1/tan(β)=2), it appeared that all clays show similar behaviour under drained and
undrained conditions, see Figure 4-9. In both cases, the safety factors are strongly
reduced for slope heights from 5 to 10 m, but this effect gradually starts to diminish for
larger slope heights. Furthermore, it can be concluded that undrained calculations yield
somewhat flatter critical slopes than drained calculations and this difference becomes
larger for higher slopes. Slope heights of 15 m or more become critical for SOC and MEC
in undrained conditions and 1:2-slopes.
25
SOC (drained) SOC (undrained)
20 MEC (drained) MEC (undrained)
STC (drained) STC (undrained)

slope height D [m]
15
10
0
1,0 1,2 1,4 1,6 1,8 2,0 2,2 2,4 2,6 2,8
Safety Factor [-]
Figure 4-9: Development of safety factor for varying slope height under drained and
undrained conditions; 1:2 side slopes
Another remark is made with respect to the consolidation properties of submarine

clayey slopes. When subjected to rapid sedimentation, the clayey material in the seabed
is in a state of underconsolidation. Towhata and Kim [1990] conducted several triaxial
compression tests in order to understand the undrained behaviour of underconsolidated
clay. The test results indicated that the undrained strength of the clay increases with
the degree of consolidation, while the cohesion and friction angle in terms of effective
stress are independent of the degree of consolidation. These results show that over the
entire depth (‘old’ and ‘new’ material) the same cohesion and friction angle can be
applied, but that the exact undrained shear strength can vary over depth. Just after
dredging of the channel, the channel bottom and side slopes will be overconsolidated,
because in fact an overburden has been removed. Freshly deposited sediments are in a
state of underconsolidation, which reduces the strength. The more rapid the
sedimentation takes place, the larger this effect will be. A more conservative slope
design is therefore advised when large sedimentation rates are to be expected.
33
Finally, in addition to the determination of the undrained shear strength, the following
line of thought will be briefly mentioned. Dolinar and Trauner [2000] claim that the
undrained shear strength in fully saturated clays is dependent on the quantity of free
water. The quantity of water in clay can be divided into free pore water between the
grains (‘intergrain’ water) and water that is strongly bound between the layers of clay
particles (‘interlayer water’). The quantity of intergrain water (we) is further subdivided
into free pore water (wef) and firmly adsorbed water around the grains (wea):
we = wea + wef = d a ASeC + wef Eq. 4-6
in which:
da = thickness of firmly adsorbed water [10-8 m]

ASeC = specific surface of clay [m2/g]
The quantity of free water is a good measure for the thickness of the water film around
the clay grains. Dolinar and Trauner found that the quantity of free water of different
soils is the same at the same undrained shear strength. According to this line of
thought, the undrained shear strength for different clay samples is the same at the
liquid limit (wL), because at this limit the water content of free pore water is the same.
The same conclusion can be drawn for clay samples at the plastic limit (wP):
at the liquid limit for all clays: su ( wL ) = 2, 66kPa Eq. 4-7

at the plastic limit for all clays: su ( wP ) = 266kPa Eq. 4-8
In most cases, however, the undrained shear strength will lie between both values.
Table 4-4: Test results on three clay samples; (I) well crystallized kaolinite; (II) poorly
crystallized kaolinite; (III) Ca-montmorillonite [source: Dolinar and Trauner, 2000]
liquid limit plastic limit

wL wP
quantity symbol unit I II III I II III
2
specific surface of clay sample ASeC m /g 10,05 23,50 97,42 10,05 23,50 97,42
undrained shear strength su kPa 2,66 2,66 2,66 266 266 266
intergrain water we % 40,0 50,9 110,8 25,8 29,8 50,0
free pore water wef % 31,9 31,9 31,9 23,1 23,3 23,1
adsorbed water around grains wea % 8,1 19,0 78,5 2,7 6,5 26,9
-8
thickness of adsorbed water layer da 10 m 0,81 0,81 0,81 0,27 0,27 0,27
Please note that purely cohesive soils (φ=0) can be still stable if the slope is a vertical
wall as long as the slope height stays within certain limits. Research on this critical slope
height was done by many researchers (Mohr-Coulomb, De Josselin de Jong, Pastor and
Fellenius). The upper and lower limit of the slope height under drained conditions can be
defined as follows:
3,64c 3,83c
≤ hc ≤ Eq. 4-9
γ γ
When dredging temporary shallow pipeline trenches, vertical cut slopes can be a good
solution, see Table 4-5.
Table 4-5: Upper and lower limits of vertical wall height
hc [m]
reference soil lower upper
SOC 1,29 1,36
MEC 1,53 1,62
STC 1,83 1,92
34
4.2 Liquefaction and flow slides
4.2.1 Static liquefaction
A liquefaction flow slide is a phenomenon in which a mass of sand on a subaqueous

slope suddenly starts behaving like a viscous fluid. Although the use of this term is
widely adopted, it is misleading:
- as sliding suggests downslope movement of a mass on a planar failure plane, with no
internal deformation, which is not the case in these slides;
- because it also suggests that liquefaction is the sand-supporting mechanism during
the whole event, whereas it only concerns the mechanism initiating the failure.
Liquefaction occurs when the framework of loosely packed fine-grained sand, saturated
with pore water, suddenly collapses as a result of some external triggering agent. This
is caused by contractant behaviour which means that the total volume of sand particles
tends to decrease when subjected to shear deformation. The pore water prohibits this
reduction in volume and an increase in pore pressure occurs, see Figure 4-10.
Figure 4-10: Schematization of (reversible) behaviour of densely and loosely packed grains
subjected to shear stress
The above mentioned sudden ‘triggering agent’ can appear in many guises, like cyclic
shear stresses during earthquakes, waves, dredging operations or small slides. With
(medium) fine sand in a layer of several meters thick, the drainage period is several
minutes. Thus, the load increase by the tide or the load increase by scour at the toe or
sedimentation at the top of a slope cannot be considered "sudden". However, these
loads can increase stresses, thereby reducing stability.
It is generally believed that a liquefied sand flow may accelerate and, if the slope is
steep and sufficient long, can become turbulent, because the thickness of the flow and
high velocity result in Reynolds numbers well above the criterion for turbulent flow.
In the relatively small space scales of navigation channels and pipeline trenches it is
however questionable if such a hyperconcentrated density flow can develop into a
turbidity current.
It should be noted that the understanding of liquefaction flow slides is based on larger
events like the channel bank failures in the Scheldt estuaries (Silvis, 1995), although
this kind of failures may occur more frequently in a smaller form in channels.
A rough criterion used in the Province of Zeeland tells that slopes steeper than 1:3 and
higher than 5 m are very susceptible to flow slides, although failures of flatter slopes
have been reported. Although, one may conclude that flow slides in trenches are
unlikely because of the small slope height, often less than 5 m.
Furthermore flow slides do not occur only in sand. They are also found in other
materials with a loose packing like in loose gravel, ore or coal deposits. Cohesive soils
with more than 20% of particles finer than 5 μm, or with liquid limit (LL) of 34 or
greater, or with the plasticity index (PI) of 14 or greater are generally considered not
35
susceptible to liquefaction [Slope Stability Engineering Manual, 2003]. In a somewhat

different form they are also found in clays with certain specific properties. Notorious are
the so-called quick-clay-slides in Norway, Sweden and Canada.
As first rough rules-of-thumb it can be stated that soils with the following conditions are
susceptible to liquefaction:
1. Soil conditions
a. low relative densities (see Paragraph 3.1.2):
RD < 60 %: risk of liquefaction is present;
RD < 30 %: risk of liquefaction is most likely;
b. poorly permeable sands (fine to medium sands; grain sizes of 50 – 1000 μm);
2. Slope geometry
slopes higher than 5 m and steeper than 1:3 (or up to 1:7 if slopes become
higher);
3. The presence of a trigger mechanism
like a small earthquake or a large wave;
Figure 4-11: Susceptibility to liquefaction (Ishihara, 1986) [source: Stoutjesdijk, 1994]
Many researchers state, that some kind of trigger mechanism will always occur.
Therefore attention should be given to the first conditions. If soils are susceptible to
liquefaction, design a more conservative slope!
Liquefaction is a prerequisite for the occurrence of a flow slide. Liquefaction takes place
if at least one sand element is in a metastable stress state. This means that the
undrained response to any quick change in load, however small it may be, consists of a
sudden large increase in pore pressure [Stoutjesdijk, 1998].
The slope geometry alone does not give enough information on the risk of a flow slide,
although in the past some empirical relationships have been developed. Better is to
integrate the degree of liquefiability of the soil and the slope geometry parameters.
This approach has been adopted by GeoDelft [Silvis 1988, 1995, Stoutjesdijk 1994,
1995, 1998] when developing a numerical model, which forms the base of a computer
program, SLIQ2D (static liquefaction 2-dimensional). This abbreviation clearly tells that
this program can only be used under static conditions. Loads due to dredging works or
cyclic loads, like waves and earthquakes, are not accounted for.
The actual process of and the excess pore pressures during a flow slide have been
investigated during research on the construction of sand fill dams [Bezuijen, 1988]. This
research showed that these excess pore pressures remained zero for a long time, but
suddenly rose to values up to 90% of the original vertical effective stress. Initial stable
slopes became unstable and sliding occurred and lasted till the excess pore pressures
disappeared. The time scale was in the order of minutes and the resulting slopes were
in the order of 1:10-30. After sliding, the porosity was only slightly decreased (up to 3
%), which means that in some cases the soil was still susceptible to liquefaction.
36
An important quantity which describes whether a certain soil will show contractant or
dilatant behaviour when subjected to a shear stress is the critical density. Distinction is
made between dry and wet critical density. Dry critical density is determined by
subjecting soil sample of different densities to a shear stress in a triaxial shear test
apparatus under drained conditions. The density at which neither a volume increase nor
a volume decrease occurs is called the ‘dry critical density’, see Figure 4-12, located in
the right-hand graph where the line crosses the y-axis.
Dilatation Curve Dry Critical Density
a
loosely
packed
b
densely c
deviator stress
packed
porosity n
d
a
e
b
volume volume c
decrease increase d
e
volume strain
maximum volume decrease εv ol;dm0
Figure 4-12: Dry critical density; ‘a’ up to ‘e’ represent soil samples with different porosity
On the other hand there is the wet critical density test under undrained conditions. Now
if a shear stress is put on the soil sample, the water isn’t able to flow out and the
volume of the sample cannot change. The result will be an increase in pore pressure.
The density at which the soil just starts to collapse is the wet critical density.
The dry and wet critical density are not the same, see Figure 4-13. If the sand is more
densely packed than the dry critical density, an increase of the deviator stress will result
in an increase of the isotropic stress, line A. Line B and C represent the situation of a
packing between the dry and wet critical density. Subjected to a shear stress, the sand
skeleton will decrease its volume and water overpressure occurs. This does not
necessarily result in liquefaction, as will be the case if the sand is more loosely packed
than the wet critical density: line D. Any small change of the shear stress may cause
liquefaction and therefore collapsing of the soil.
Deviator Stress
B
A
Isotropic Stress
Figure 4-13: Stress paths for different densities
37
It should be clear that the dry critical density is a safe upper boundary. Densities larger
than the dry critical density are unconditionally stable. A density between the dry and
wet critical density is indicated as ‘possible susceptible to liquefaction’ and a density
smaller than the wet critical density is ‘very susceptible to liquefaction’. It should be
noted that susceptibility does not necessarily have to result in a flow slide; the
geometrical properties have to be taken into account.
SLIQ2D considers dry critical densities and soils under drained conditions, but
introduces a small undrained change to investigate the susceptibility to flow slides. The
underlying concept is that the building up of soil layers is a very slow process (drained
behaviour), whereas a sudden change in load will cause undrained behaviour. Because
of this concept, this program is not applicable under dredging conditions or other quick
stress changes.
In SLIQ2D a lot of calculations have been done with Pluto (finite element method) to
obtain the initial soil stresses for different geometries. SLIQ2D uses this stresses and
determines whether a small change in shear stress under undrained conditions causes a
volume decrease. This can be expressed in a simple way:
1
dγ xy = dτ xy Eq. 4-10
λ
Parameter λ is the eigenvalue of the stiffness matrix describing the incremental
deformation of the soil elements in the critical zone of the slope as a function of a
sudden stress change [Molenkamp (1989) as described by Stoutjesdijk, 1994]. This
eigenvalue becomes negative when an increase of the shear strain γxy causes a decrease
of the shear stress τxy.
In SLIQ2D the slope geometry is defined and then the program successively:
1 reads the input;
2 divides the geometry in 500 points;
3 calculates the initial stresses starting from a very flat slope;
4 calculates the eigenvalue λ;
5 steepens the slope a little bit and repeats step 3 and 4.
The output contains for each grid point the flattest slope at which the eigenvalue
becomes negative. The program considers a certain slope geometry to be unstable if at
one point the eigenvalue becomes negative. In reality there will be some residual
strength, so the output will be conservative.
Input
The input consists of geometry, material and fit parameters. The geometry can be
represented as a straight slope or a bended slope with the steepest part at the toe of
the slope. In the latter case two slope heights and the slope angle of the upper part
have to be defined. Under the assumption that failure will occur in the lower part, only
this part is steepened.
The soil is represented by the unit weight under water, which is about 10 kN/m3, the
elastic Poisson ratio of about 0,3 and the non-elastic (plastic) Poisson ratio of 0,4.
To obtain reliable input for the fit parameters, relatively much effort has to be made.
Because of the sensibility to the different soil parameters, some tests have to be done:
- make borings to obtain soil samples;
- determine in-situ density with electrical conductivity measurements;
- examine drained triaxial test to determine the stress-strain curve
With the results, the input parameters can be calculated.
38
Figure 4-14 shows a cross-section of the grid points in the bed where the soil starts to
liquefy. This graph shows a cross-section through a side slope; the top of the slope is
located at the upper left corner. One should keep in mind that this graph is a distorted
view of reality. SLIQ2D uses 500 grid points with changing ‘real’ x- and z-coordinates, if
the bed profile is steepened. Therefore the peak points in reality are situated at much
larger real x-coordinates, because of the smaller angle of a just stable slope; the stable
points belong to a steep slope and have smaller x-coordinates.
Nevertheless, the above presentation is chosen, because it is able to give a quick
impression of the number of unstable points and the ‘rate of instability’ (= the height of
the peak). Moreover, the location of initiation of instability can be observed.
It is easy to get an impression of the area above a certain isoline (slope of 1:y) and to
determine the measurements to increase stability. For example, a large flat bump
means that stability can easily be reached by flattening the slope a little bit. Slender,
high peaks probably does not give severe problems because of the residual strength of
the neighbouring soil particles, but wide, high peaks demand serious measures. The soil
may have to be compacted. In any case, an additional research study is recommended.
Figure 4-14: An example of the output of a calculation with SLIQ2D; the plane is a cross-
section of a side slope of a trench
The accuracy of the results is expected to be limited; nevertheless the safety against
failure is considered sufficient. Because of the use of values of a dry triaxial test, the
results probably are conservative. Points of interest, however, are the spread in the
input parameters, the influence of thin layers of clay or mud and the determination of
in-situ stresses. On the other hand, some redistribution of soil stresses is likely, water
pressures can dissipate a little and liquefaction in a very small area (a few grid points)
probably does not cause failure.
The erosion length is defined as the length over which a flow side extends. Stoutjesdijk
developed a probabilistic expression based on data of flow slides collected by Wilderom
in 1979. This length Lflowslide is proportional to the slope height D:
L flowslide = K fs D Eq. 4-11
in which Kfs is normally distributed with mean μ = 2,63 and standard deviation σ = 1,69,
see Figure 4-15:
39
Normal Distribution of Erosion Length Lflowslide
0,25
0,20
Probability Distribution
0,15
0,10
probability of 95%
at Lflowslide / D = 5,4
0,05
0,00
0 1 2 3 4 5 6 7 8 9 10
Erosion Length Lflowslide / Slope Height D [-]
Figure 4-15: Normal distribution of the erosion length divided by the slope height
Knowing this erosion length, the erosion volume can be estimated by making a few
assumptions:
- the final flow side surface crosses the initial slope at half of the slope height D;
- the shape of the final eroded part of the geometry can be described by a square
root function, see Eq. 4-11 and Figure 4-16:
D 1
z ( x) = x for 0 ≤ x ≤ ( K fs + )D and 0 ≤ z ≤ 12 D Eq. 4-12
2 2 tan β
4 K fs + )
tan β
- the change in volume is negligible, because the decrease in porosity usually is
very small; so the ‘disappeared’ erosion volume on top of the slope is equal to
the sedimentation volume at the toe of the slope.
Distance along x-axis [m]
0
0 10 20 30 40 50 60 70 80 90 100
-5
Lflowslide
Depth wrt sealevel [m]
x
-10
erosion
½D z
-15
deposition β
-20
-25
Figure 4-16: Assumed square root profile of geometry after flow slide
40
Now, the erosion volume per unit width can with the help of some mathematics be
expressed as:
1⎛ 1 ⎞ 2
V fs = ⎜⎜ K fs ( μ ,σ ) + ⎟D Eq. 4-13
3⎝ 8 tan β ⎟⎠
For example, the erosion volume of a slope height of 10 m and an initial slope angle of
1:3 already becomes 100,7 m3/m, when using the mean value for the erosion factor Kfs;
considering the 95 % probability area, this erosion volume can increase up to 192,5
m3/m. The corresponding post-failure average slopes are respectively 1:8 and 1:14.
It should be noted that the above estimations show a huge spread, because of the wide
variation in erosion lengths and observed geometries. However one should understand
that erosion volumes this large almost always cause failure of the navigation channel,
i.e. the required depth and/or width cannot be guaranteed anymore.
4.2.2 Computations with SLIQ2D
With SLIQ2D a number of calculations have been done for slope heights of 5, 10 and 20
m and for soils with porosities of 0,42 (dry critical density) to 0,48 (maximum porosity).
The soil properties are primarily characterized by the relation between porosity and
maximum dilatant volume strain according to Figure 4-17.
48
46
44
porosity n
42
40
38
36
0,008 0,006 0,004 0,002 0
maximum volume decrease εvol;dm0
Figure 4-17: Relation between porosity and maximum dilatant volume strain, based on
assumptions for minimum and maximum porosity and dry critical density
As was mentioned above, also the exact stress-strain curve has to be known to
generate the input data. Of course this curve is different for every soil, but Figure 4-18
(which is a rotated version of Figure 4-12) shows a frequently observed curve, based on
relative strains and shear stresses, which will be used in the calculations with SLIQ2D.
The corresponding formula is:
Bs r
εvol d = εvol dm 0 ( As m − ) Eq. 4-14
s max − s
in which:
41
1 m m( s max − s 2 )
A= + and B=
s 2m s2 s2
s 2m (r − m + ) s 2r (r − m + )
s max − s 2 s max − s 2
εvold = dilatant volume strain

εvoldm0 = maximum dilatant volume strain
s = relative shear stress
m, r = ‘fitting constants’
s2 = relative shear stress at which εvoldm0 occurs
smax = asymptote of relative shear stress
The values of the parameters s2, smax, m, r, A and B are plotted in the graph. The
sensitivity of these parameters is investigated in the sensitivity analysis.
1,0
0,8
relative volume strain εvold / εvoldm0
0,6
0,4 s2
0,2
0,0
-0,2 s2 = 1,10
smax = 1,45
-0,4 m = 1,3
r = 4,0
-0,6
A = 1,083
-0,8 B = 0,053
-1,0
0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 1,1 1,2 1,3 1,4
relative shear stress (shear stress / isotropic stress)
Figure 4-18: Graph of relative volume strain against relative shear stress
A factor ‘v’ is introduced that discounts for the fact that the isotropic stresses in the field
can differ from the isotropic stress in the test (50 kPa) on which the above formulations
(dilatation curve) are based. A recommended value for this factor ‘v’ is 1,25;
The second curve that needs to be fit is the decompression curve, which describes the
volume strain as function of the isotropic stress:
dσ 'vol
d ε vol ;c = − Eq. 4-15
Ks
The decompression modulus ‘Ks’ is dependent on the isotropic stress, in other words the
stress-strain relationship is not linear. The following relation yields the last two SLIQ2D-
input parameters ‘Ks0’ and ‘u’:
u
⎛ σ' ⎞
K s = K s 0 ⎜ vol ⎟ Eq. 4-16
⎜σ ' ⎟
⎝ vol ;0 ⎠
in which Ks0 and σ’vol;0 respectively represent the decompression modulus and the
isotropic stress at the begin of decompression. Common values for both parameters
are: 40.000 ≤ Ks0 ≤ 60.000 kPa
0,5≤ u ≤ 1,0
42
Sensitivity Analysis
With this ‘schematized soils’ a number of calculations has been done. At first the
sensitivity of the most important input parameters has been investigated. In Table 4-6
can be seen that the ‘fitting parameters’ of the above mentioned stress-strain-curve
hardly have any influence on the slope stability. This is favourable, because
determination of this graph takes a lot of effort. On the other hand, it can be concluded
that a good estimation of the maximum volume strain and the decompression modulus
is rather important. The almost linear relationship between porosity and maximum
volume strain for porosities larger than the dry critical porosity can be helpful.
Table 4-6: Sensitivity of the most important input parameters

default
parameter description unit min max (1/tan β)default (1/tan β)min (1/tan β)max sensitivity
value
n porosity - 45 44 46 4,75 4,75 4,5 p
nmax maximum porosity - 48 47 49 4,75 4,75 4,75 0
nmin minimum porosity - 30 - - 4,75 - - -
evoldm0 maximum volume strain - 0,003 0,002 0,004 4,75 3,25 5,5 N
m fitting constant - 1,3 1 1,6 4,75 5 4 p
r fitting constant - 4 3,5 4,5 4,75 4,5 4,25 p
s2 relative shear stress at evoldm0 - 1,1 1 1,2
4,75 5 4 p
smax asymptote of shear stress - 1,45 1,35 1,55
v power of isotropic stress - 1,25 0,5 3 4,75 4,5 5,5 n
u power of decompression curve - 1 0,5 1 4,75 4,5 4,75 n
Ks0 decompression modulus kPa 50000 40000 60000 4,75 3,75 5,5 N
N = increase of parameter strongly influences stability negatively; n = little negative influence; 0 = hardly any influence;
p = little positive influence; P = strongly positive influence
Calculations for different heights

For sands with different porosities, the slope heights have been varied, see Figure 4-19.
On the right y-axis, the relative density is presented. Because this parameter is linearly
related to the void ratio, but not to the porosity, the scale is a bit distorted.
Two important conclusions can be drawn:
- the slope height is very important: doubling of the slope height requires
reducing the slope steepness with a factor 2;
- the in situ porosity has to be determined as accurately as possible if the soil is
susceptible to liquefaction.
This means that small variations between design and construction can reduce stability
significantly. Dredging of an overdepth of just 1 m has to be considered in the design
phase.
43
Critical slopes for different porosities 'n' and slope heights 'D'
48 0
UNSTABLE
47 7,3
Relative Density RD [%]

46 14,4
porosity n [%]
45 21,2
44 27,8
STABLE
43 34,1
42 40,2
1 2 3 4 5 6 7 8 9 10 11 12
critical slope 1:x [-]
slope height = 5 m slope height = 10 m slope height = 20 m
Figure 4-19: Critical slopes for different porosities (Relative Densities) and slope heights
Bended slopes
Also has been investigated whether improvements in the slope geometry can be applied
to reduce the dredging volumes maintaining the same stability, or to decrease the
required width. After all, instability originates in the lower half of the slope, so flatter
lower slopes and steeper upper slopes could me more stable.
Without explaining how these bended slopes can be implemented in SLIQ2D, the results
are presented in Table 4-7 and compared to the default run. Grey cells show an
improvement compared to the default run.
Table 4-7: Results of calculations with bended slopes compared to the ‘default’ run
bended profiles
default I II III IV
h1 m 10 5 5 5 2,5
tanβ 1 - 1:4,75 1:3,75 1:5 1:4 1:5
h2 m 5 5 5 5
tanβ 2 - 1:5,63 1:4,5 1:5,33 1:4,25
h3 m 2,5
tanβ 3 - 1:5,31
tanβ average - 1:4,75 1:4,69 1:4,75 1:4,67 1:4,70
3
Adredged m /m 238 258 231 250 286
Lslope, x m 47,5 46,9 47,5 46,7 47,0
All bended profiles are also plotted in Figure 4-20. Bended profiles I and III have
steeper upper parts and flatter lower parts. With these profiles the upper channel width
can be reduced, but the dredging volume of the slope increases. Bended profile II has a
flatter upper part and a steeper lower part, which reduces the total dredging volume,
but increases the upper width. ‘Outsider’ profile IV has its steepest part right in the
middle of the slope. The upper width is hardly reduced, while the total dredging volume
is increased. This profile seems to have no advantages at all, but this slope geometry
resembles the morphological profile after some time best. This profile is therefore
expected to ‘keep in (its initial) shape’. Depending on the hydrodynamic conditions in
combination with the sediment properties, this channel geometry may perform best in
the long term.
44
2
-10 0 10 20 30 40 50 60
0
Depth wrt surrounding seabed [m]
-2
3
Adefault = 238 m /m
Adredging of slope
AI = 258 m3/m
-4 A II = 231 m /m
3
3
A III = 250 m /m
3
-6 A IV = 286 m /m
-8 Default Profile
Profile I
Profile II
-10 Profile III
Profile IV
-12
Figure 4-20: Schematization of bended profiles and total slope dredging volume
Despite all good intentions, the differences of the upper channel width and slope
dredging volume between the default profile and the ‘bended profiles’ remain small. The
minimal obtained reduction in upper channel width or total dredging volumes does not
compensate for the extra effort to design and dredge these rather sophisticated profiles.
This measure of ‘bending slopes’ is therefore not recommended, unless some other
benefits can be gained. As will appear in later chapters, sometimes dredging of a
sedimentation reservoir at the side slopes prolongs the lifespan of a trench or channel.
Then, the measure of ‘bending profiles’ can be extremely useful, because the upper
width remains limited while the channel is provided with a reservoir at the same time.
Also it was mentioned that profile IV may have some morphological advantages.
Conclusions on SLIQ2D-calculations
- Soils with densities larger than the ‘dry critical density’ (n < 0,42) are
unconditionally stable;
- Soils with densities smaller than the ‘dry critical density’ (n > 0,42) can still be
stable, depending on the soil geometry;
- Slopes with heights smaller than 5 m are almost unconditionally stable. Only
very loosely packed soils (RD < 10) are susceptible to liquefaction.
- When considering dredging of an overdepth, it is dangerous to just increase the
slope height when the soil is susceptible to liquefaction, because of the strong
sensitivity to the slope height. A flatter toe of the slope can improve stability and
creates a more natural morphological profile.
- Slopes with steeper upper parts and flatter lower parts can be equally stable as
constant slopes, while the required navigation depth is reached earlier. The
necessary width can also be reduced, although the total dredging volume can
increase. But unfortunately, very large reductions of dredging volumes are not
to be expected (less than 5 %), but adaptation to a more morphological profile
is very well possible without giving concessions to slope stability or dredging
costs.
Soil porosity and the dielectric constant
Although the susceptibility to liquefaction is very sensitive to the maximum dilatant

volume strain and especially the porosity, it appeared that in dredging practice often
very little is known about these parameters.
45
From conversations with geotechnical engineers it became clear that they fear a limited
applicability of the research results. Therefore some supplementary research has been
done on possible correlations between porosity and other soil properties. It seems that
the most reliable method of soil investigation related to porosity deals with the electric
conductivity. Until recently the suggested formulations were rather complex and not
very useful for practical purposes. Kaya [2002] deduced a simple formulation, based on
his own test results and independent data from other tests on a wide range of soils. For
any soil-water mixture this formulation is:
n = 0, 0136ε + 0, 02 Eq. 4-17
in which
ε = the dielectric constant [-]
The dielectric constant is the relative dielectric permittivity of a certain medium (here:
soil-water mixture) and thus a measure of the relative ability of a material to store
charge for a given applied electric field. It can be determined form capacitance
measurements, in which the storage of charge in a certain medium between two
electrically charged plates is measured. Such an instrument can be employed in Cone
Penetration Tests. The dielectric constant can be obtained from the measured
capacitance ‘C’ as follows:
Cd
ε= Eq. 4-18
ε0 A
in which
C = capacitance [F] or [C/V]
d = distance between the plates [m]
ε0 = dielectric permittivity of vacuum = 8,854*10-12 F/m [F/m] pr [C/Vm]
A = surface of the plates [m2]
70
60
dielectric constant ε [-]
50
40
30
n = 0,0136ε + 0,02
20
R2 = 0,97
10
0
0 0,2 0,4 0,6 0,8 1
porosity n [-]
Figure 4-21: Dielectric constant against porosity [source: Kaya, 2002]
Formulation 4-13 and Figure 4-21 are valid at low MHz frequency range (13-50 MHz)
and regression analysis showed good agreement with test results: R2 = 0,97.
4.2.3 Soil composites
The liquefaction behaviour is extremely sensitive to the amount of other soil contents.
Most theoretical research, just as this thesis, is done on homogeneous soils, while small
amounts of contaminants can have a significant impact on the strength of the soil. Two
well-known mixtures will be mentioned briefly.
46
Silty sand
The effect of fines like silts in sand has often been studied, with very contradictory
results. Some researchers (Seed 1983, Pitman 1994) state that the liquefaction
potential is reduced by the presence of fines; others (Lade et al. 1997) state the
opposite. Most important difference between the research studies is the relative density
of all the tests. Some maintain the same void ratio of the original sand, filling the voids
with fines and thereby increasing the relative density, others maintain this absolute
density or use an equal natural settling process, which in my opinion is the most useful
method when considering submarine slopes. When submarine slopes are susceptible to
liquefaction, the sediment is often deposited under calm marine conditions.
Therefore only the results of Lade and Yamamuro (1997) will be briefly presented and
explained. Four different sands (Nevada 74-300 μm, Nevada 175-300 μm, Ottawa 74-
300 μm and Ottawa 74-250 μm) with increasing non-plastic fine contents (14-74 μm)
were investigated. It appeared that a so-called ‘dry funnel deposition method’ yielded
homogeneous soils which were very well comparable to alluvial or marine sedimentation
(n = 0,40-0,46). The tests were done at very low confining pressures (25 kPa), because
then static liquefaction is most prevalent. These low pressures are typical for sea beds
Their most important conclusion was:
All sands contaminated with fines are less resistant against liquefaction than non-
contaminated sand, even though their relative and absolute densities increase.
This conclusion is inconsistent with normal soil behaviour, because soil should show a
more dilatant behaviour at higher densities. So apparently void ratio and relative
density are no good criteria when considering the liquefaction potential of soil mixtures.
The explanation of this apparently inconsistent soil behaviour is based on the following
hypothesis: “in silty sands of low to moderate densities, a particle structure can develop
in the soil between the larger and smaller grains resulting in high volumetric
compressibility, which in turn results in static liquefaction at low pressures”.
Figure 4-22: Maximum and minimum void ratio and the ‘static liquefaction envelope’ against
percentage of fines [Lade et al. 1997].
The test results are demonstrated in Figure 4-22 for Nevada 50/200 sand (d=74-
300μm), which appeared to be susceptible to liquefaction for all fine contents, unlike the
other sands that appeared to be not susceptible to liquefaction for very small fine
contents. The maximum and minimum void ratio are plotted for different fine contents.
The line of the minimum void ratio, which is obtained when the soil is energetically
densified, shows that a small amount of fines increases the density by filling the void
spaces between the grains, leaving the larger grains in full contact with each other. By
enlarging the content of fines, the larger grains lose contact, thus reducing the density.
It appears that a fine content of approximately 20-30% is most effective in just filling
the voids, without pushing the larger grains from each other.
47
The line of the maximum void ratio is obtained by deposition of the soil under minimum
input of energy. During this deposition some, but not all fines will settle just in the
voids, causing a slightly smaller dip in this line, compared to the minimum void ratio.
With larger amount of fines, more fines start to occupy locations between the contact
points of the larger grains, thus reducing the relative density.
All test results are also plotted and the static liquefaction envelope was determined. It
can be seen that this line slowly shifts towards the minimum void ratio, indicating that
with increasing fine content, soils with higher relative densities are still susceptible to
static liquefaction, even up to relative densities of 60% for fines contents over 50%.
Although relative densities may increase at larger contents of fines, the overall
behaviour is not much affected, because during shearing, these fines will roll loosely
around in the voids between the larger grains. When the percentage of fines is further
enlarged, the original void spaces no longer exist and the fines start to dominate the
soil behaviour.
The soil behaviour during loading is schematized in Figure 4-23; the silt grains between
the larger sand grains will move towards the voids when compressed or sheared,
increasing contractant tendencies. Only if the shearing continues and the sand grains
come into better contact with each other, the soil behaviour can become dilatant, but
only for small percentages of fines.
Figure 4-23: Schematization of development of void ratio with increasing fines content
As was mentioned before many research studies on actual earthquake-induced

liquefaction events state that sands with a significant fines content should be more
resistant against cyclic liquefaction. It is however more likely that this apparent
resistance is caused by effects of stress history (overconsolidation) and aging (creep or
crementation); both effects are not significant in natural submarine slopes (or hydraulic
fills), which are characterized by low effective stresses and no overconsolidation.
Sediment is continually and slowly added to the slope, which tend to reduce the effect
of creep. And after all, most reported shallow submarine slope failures are situated in
silty sands.
48
Sand-gravel composites
Evans et al. [1995] investigated the liquefaction resistance of sand-gravel composites.
Undrained cyclic triaxial tests were performed on sand-gravel composites with gravel
contents (GC) of 0%, 20%, 40% and 60%, all with a relative density of the composite
of 40%. They found that maximum and minimum densities of sand-gravel composite
sediments increased significantly with increasing gravel content and reached peak
values at a gravel content of about 60%. This observation is in line with Lade and
Yamamuro [1997], when inversely reasoning: they found peak values for a fines
content of about 30%, so content of the coarser material (sand) of about 70%.
The liquefaction resistance of sand-gravel composites may increase considerably with

increasing gravel content. Or more precisely, the development of the pore pressure and
axial strain of the first two composites (GC=0% and 20%) under cyclic loading showed
classic liquefaction behaviour of moderately loose sand; the other composites (GC=40%
and 60%) showed more classic cyclic mobility behaviour of dense sand, see Figure 4-24
and Figure 4-25.
Figure 4-24: Pore pressure response of sand-gravel composite specimens with gravel
contents of (a) 0%, (b) 20%, (c) 40% and (d) 60%
Samples (a) and (b) show hardly no axial strain response during the first 10 stress
cycles, but then suddenly a large response occurs, indicating failure. The pore pressure
ratio becomes 1 and the grains loose contact. Before failure occurs, the amplitude of the
pore pressure ratio remains small, whereas samples (c) and (d) can undergo far larger
CSR’s according to the larger amplitudes of the pore pressure ratio. The axial strain
slowly increases, but is present from the beginning of loading; no ‘explosive’ behaviour
is observed.
49
Figure 4-25: Axial strain response of sand-gravel composite specimens with gravel contents
of (a) 0%, (b) 20%, (c) 40% and (d) 60%
Evans et al. [1995] also concluded that the relative density isn’t a good measure to
indicate liquefaction potential. When the gravel content is increased and the composite
relative density is kept constant, the relative density of the sand matrix is reduced. For
example, when the grain content is 60%, the relative density of the sand matrix is 0%
(at the loosest packing). However, the resistance against liquefaction increases.
Therefore a comparison was done between a clean sand (RD = 40%) and a sand-gravel
composite (GC=40%, RDsand = 40%, so RDcomposite = 52%).
It appeared that the cyclic stress ratio to cause 5% double amplitude strain in 10 load
cycles (CSR5%;10) increased from 0,15 to 0,36 for respectively the clean sand and the
composite and the behaviour transformed from loosely packed to densely packed.
Another test was done to find a clean sand equally stable as a sand-gravel composite
(GC=40%, RDcomposite = 40%, so RDsand = 29%). It appeared that a RD of 66% was
needed to obtain an equally stable clean sand, again proving that the addition of gravel
reduces the liquefaction potential. This test also proved that cyclic loading tests on
sand-gravel composites can be estimated by tests on clean sands at a representative
density.
Summarizing, both research studies prove that smaller particles between larger
particles can significantly increase liquefaction potential. An approach based on the
relative density only is not very meaningful. It should also be noted that soil tests on
homogeneous soils as well as on composites of two fractions are a strong simplification
of nature. A very widely graded soil can perform much better, because the unfavourable
particle structure of Figure 4-23 will not develop, because all fractions in between are
present.
4.3 Retrogressive breaching
Opposite of liquefaction slope failures, breach failures occur at medium to densely

(hexagonally) packed fine sand and this failure type is restricted to very steep
submarine slopes. Thin superficial layers of sediment show dilatant behaviour, because,
when subjected to shear deformation, its volume expands, causing a negative pore
pressure with respect to the hydrostatic pressure, resulting in increased shear
resistance. This situation of underpressured sand is able to maintain a steep (steeper
than the angle of repose) slope, until pore water flows in and the slope will gradually
fail.
50
The necessary precondition for the initiation of a breach failure is a local steep slope
disturbance, such as that produced by dredging activities or scour by channel flow. In
conditions met in freshly deposited sand the difference between dilatant and contractant
behaviour is often very small. Therefore retrogressive breaching is frequently mistaken
for a liquefaction flow slide. A breach failure could also originate from the scar produced
by a small liquefaction slide; in that case both failure mechanisms may be present.
Two types of breaches can be distinguished, initiated by a supercritical (Fr > 1) or

subcritical (Fr < 1) flow. The first type is found in series of bed geometries formed at
high Froude numbers (Fr = u/√gh) that can be considered as an adaptation of chute-
and-pool bedforms to conditions in fine sand and silt. Large breaches of this type occur
at the bursting of an embankment (Visser, 1998). The second type of breaches can be
generated incidentally on steep, perhaps up to more than 5 m high submerged slopes
on bars or navigation channel banks. The interest is in the second type, which is often
mistaken for a liquefaction flow slide. The difference lies in the time span: a breach
retrogrades slowly and may be active for a large number of hours, producing a quasi-
steady turbidity current. Breach failure can occur when the top of the slope is steeper
than the angle-of-repose φ.
The other division in types of breaches is in ‘controlled’ and ‘uncontrolled’ breaching. In

dredging practice one speaks of controlled breaching. By increasing or decreasing
production rates a controlled breach can be maintained with a more or less constant
wall velocity. This wall velocity is defined as the retrograding velocity of the breach and
has an empirical relation to the hydraulic permeability (Vwall ≈ 25k). For the reference
sands under consideration this yields values in the order of 1 mm/s. When breaches
become higher and steeper, this grain-by-grain failure turns into larger failure in slices.
Breaches can also become self-maintaining and continue after the dredging is finished.
This slow and gradual failure mechanism of breaching is rather different from the faster
and intermittent liquefaction flow slides.
Figure 4-26: Flow from an active breach to a suction entrance with hydraulic jumps,
subcritical flow with sedimentation and supercritical flow with scour
In Overijssel research is done by Delft Cluster (GeoDelft and Delft Hydraulics) on sand
borrowing pits [WL|Delft Hydraulics, 2001, Mastbergen, 2002]. The growing demand on
sand resulted in the question whether existing pits could be deepened without risking
slopes instability. The former regulations of the Province of Overijssel appeared to be
too conservative and too general. The new approach takes grain diameters and depth of
the pit into account.
51
For illustration a graph is presented in which the required distance from the border to
the toe of the slope is drawn for a certain depth for two different grain diameters, see
Figure 4-27. Dredging ‘under the concession line’ is not allowed, even if it means that
the total amount of available sand cannot be extracted.
0
-10 d50 = 240 μm

d50 = 180 μm
-20
depth z [m]
-30
-40
-50
-60
0 100 200 300 400 500 600
distance x [m]
Figure 4-27: Concession line for sand borrowing pits in the Province of Overijssel for two
different grain sizes [source: WL|Delft Hydraulics, 2001]
It is noticeable that coarser sand allows for steeper slopes and that with increasing
depth of the sand borrowing pit, the maximum slope angle reduces. Furthermore, it can
be concluded that slopes gentler than 1:3 and not higher than 20 m are unconditionally
stable with respect to breaching.
Unlike controlled breaching, uncontrolled breaching is a natural process which is still not
commonly known as an instability mechanism (Van den Berg 2002, Mastbergen 2003).
This breaching process runs autonomously and can take several hours, in contrast with
the suddenly occurring flow slides. An initially steep slope is a prerequisite for
uncontrolled breaching, so this mechanism is not very likely in channels, because of the
limited slope height (about 10 m) and the relatively flat (flatter than the angle of
repose) slopes which will remain after dredging. This mechanism however can be of
major importance in large submarine canyons.
4.4 Micro-instability on an infinite slope
In this paragraph the stability of the outer grains of the slope will be discussed. The slip
circle analysis in paragraph already proved that in non-cohesive, homogeneous soils
micro-stability is the normative failure mechanism. This resulted in slip circles with large
radii that only slightly touched the slope surface. In this paragraph the unloaded slope
will be considered. Common loads that affect micro-stability are seepage and/or
pressure gradients, like for instance ground water flow through dikes between different
water levels or water pressure gradients after long-term high water in rivers or short-
term wave attack (wave troughs cause an overpressure inside the slope). Please note
that these wave forces are not to be confused with the erosive or stirring forces causing
instability of the outer grains, see Chapter 7 (threshold of motion) and 8 (sediment
transport).
The upper boundary of the slope angle (steepest possible slope), when it comes to
micro-stability, will be found when considering an infinite long slope without any
additional forces like seepage flow.
52
In Paragraph 4.1.2 a vertical wall of cohesive soil was considered; now an infinite slope
in non-cohesive soil without any currents and waves is considered. If we define a
coordinate system with the ξ-axis along the slope (positive upward) and the η-axis
perpendicular to the slope (positive downward), we obtain the following two equations
of equilibrium parallel and perpendicular to the slope:
∂σ ξξ' ∂σ ηξ' ∂p
+ + + γ sat sin β = 0
∂ξ ∂η ∂ξ
Eq. 4-19
∂σ ξη
'
∂p∂σ ηη
'
+ + − γ sat cos β = 0
∂ξ ∂η ∂η
When the groundwater is in rest, the water pressure can be assumed to be hydrostatic:
p = p0 − γ w z = p0 + γ wη cos β − γ wξ sin β Eq. 4-20
If the water level is assumed to be infinitely high, because the slope is assumed to be
situated completely below water level and the slope is infinitely long:
∂p0 ∂p0 ∂p ∂p
p0 → ∞ ⇒ , →0⇒ = −γ w sin β , = γ w cos β Eq. 4-21
∂ξ ∂η ∂ξ ∂η
This assumption does not fulfil reality, but this approach is useful to determine the
upper limit in case of very long, deep slopes.
Substituting Eq. 4-19 into Eq. 4-17 yields:
∂σ ξξ' ∂σ ηξ'
+ + (γ sat − γ w ) sin β = 0
∂ξ ∂η
Eq. 4-22
∂σ ξη
'
∂σ ηη
'
+ − (γ sat − γ w ) cos β = 0
∂ξ ∂η
On the assumption that the stresses are independent on the coordinate along the slope
(after all, the origin may be chosen anywhere along an infinite slope), these equations
can easily be integrated. The constants of integration must equal zero, because the
effective grain stresses equal zero at the slope surface:
σ ηξ' = −(γ sat − γ w )η sin β

Eq. 4-23
σ ηη
'
= +(γ sat − γ w )η cos β
If these two equations are combined with the Coulomb failure criterion for non-cohesive
soils, it gives:
σ ηξ' / σ ηη
'
< σ ηξ
'
/ σ ηη
'
max
⇒ tan β < tan φ Eq. 4-24
The above equation can of course be expressed in the stability or safety factor F:
tan β
SF = Eq. 4-25
tan φ
As was concluded in Paragraph 4.1, the unit weight of a non-cohesive soil does not
affect micro-stability in absence of groundwater flow. In absence of seepage a
submarine slope can be as steep as a slope above the water level. In reality, additional
forces occur, like waves.
53
5 Wave Loads
5.1 Direct Elastic Behaviour
5.1.1 Load or reduction of strength?
Until now, attention is only paid to unloaded conditions. Loads were only treated as a
sudden trigger mechanism. Of course, in practice some loads will occur. One can think
of loads from constructions, propellers of manoeuvring ships, earthquakes and of course
waves. In this thesis only wave loads are considered extensively, because waves are
common all over the world. Earthquakes however are also notorious for their
destructive capacities. These and other loads are described briefly in Paragraph 5.3.
Bottom pressure pulses due to ocean waves can be a major factor in initiating
submarine landslides. These waves produce pressure changes within the water below
the surface and pressure pulses on the surface of the sediment. Because of this
difference in pressure under crest and trough, the passage of a wave induces shear
stress in the soil. As the wave passes, soil at a particular point experiences cyclic
fluctuation in magnitude and direction of these induced stresses.
At this point it is very important to distinguish between the direct, elastic effect of
waves and the indirect, plastic effect. In this paragraph the direct effect is treated,
which causes a fluctuation of water pressures and effective stresses, but is assumed to
have no net effect: these transient stresses dissipate with each wave cycle. On the
other hand this fluctuation of effective stresses in the soil causes a tendency to
compact, resulting in gradual water overpressure (indirect effect). This overpressure
can cause liquefaction, which will be investigated in the next paragraph.
To investigate the problem of direct wave effects often many extensive simplifications
are assumed, which will be explained further on.
Often the rule of thumb is used that excess pore pressures (EPP’s) become negligible
when the depth of the water d becomes greater than half of the wavelength. Because of
the maximum depth of interest of 20m, excess pore pressures will always occur,
because wavelengths are easily larger than 40m.
The loss of stability under wave action is analyzed on the concept that failure is gravity
controlled, but that wave action can reduce the (undrained) strength of the soil. So in
fact waves do not act as loads, but more as a reduction of strength. The title of this
chapter in fact is wrongly chosen, when looking at submarine slopes, which are
completely situated below sea level. In the breaker zone wave forces can damage
revetments and can be spoken of ‘loads’.
5.1.2 Water pressures below sea level
There are several classical theories that describe the subsurface pressure fluctuations at
a given point due to the passage of a surface wave. Most researchers have used the
linear wave theory, but also second order Stokes’ wave theory and even higher order
wave theories can be applied. Which theory has to be used is of course dependent on
the wave characteristics: wave height H, wave period T and the depth of the water h.
These characteristics can be described by two well-known parameters H/gT2 and h/gT2
(LeMéhauté, 1976). The first one is a measure for the wave steepness (H/L), the second
one for the relative water depth (h/L).
54
In this thesis wave periods in the order of 5-15 s and wave heights in the order of 2,5-
10 m are dealt with in water depths varying from 5 to 20 m. Because we are interested
in the macro-stability of a slope, we look at the largest possible waves. Smaller, more
frequent waves are important when considering sediment transport which will be
treated later on.
Unlike smaller waves, which can be approximated by the linear theory, depth-limited
waves with a wave height of about half the water depth are in the ‘cnoidal theory
regime’. This means that the linear theory as well as the second order Stokes’ theory do
not apply and the more complex cnoidal formulations described by elliptic integrals for
wave pressures should be used, see Table 5-1.
Table 5-1: Application of the appropriate wave theory
T [s]
5 7 9 11 13 15
2 2 2 2 2 2 2 2 2 2 2 2
h [m] H [m] h/gT H/gT h/gT H/gT h/gT H/gT h/gT H/gT h/gT H/gT h/gT H/gT
0,020 0,010 0,010 0,005 0,0062 0,0031 0,0041 0,0021 0,0030 0,0015 0,0022 0,0011
5 2,5
cnoidal cnoidal cnoidal cnoidal cnoidal cnoidal
0,040 0,020 0,020 0,010 0,0123 0,0062 0,0083 0,0041 0,0059 0,0030 0,0044 0,0022
10 5
breaking cnoidal cnoidal cnoidal cnoidal cnoidal
0,080 0,040 0,041 0,020 0,0247 0,0123 0,0165 0,0083 0,0118 0,0059 0,0089 0,0044
20 10
breaking breaking cnoidal cnoidal cnoidal cnoidal
Most research studies however only consider linear and sometimes second order waves,
while especially in the upper layer of the water depth cnoidal theory shows different
behaviour. However, when approaching the bed, the difference in wave pressures
between cnoidal and linear wave theory decreases and an approximation according to
the linear wave theory is therefore considered sufficiently accurate:
ρ w gH cosh kz
p wave = cos(kx − ωt ) Eq. 5-1
2 cosh kh
For an impermeable rigid flat bed the amplitude of the wave pressure at the bottom
(z=0) can now be expressed as:
ρ w gH
pˆ wave; z =0 = p0 = Eq. 5-2
2 cosh kh
The assumption of a flat bed may be too favourable; in 1977 Mallard and Dalrymple
analyzed the effects of a deformable seafloor on the water pressures. They found that
these pressures are up to 15% higher for very soft cohesive sediments, but may be
ignored for most sands.
Liu (1973) cast doubt on the assumption of impermeability of the seabed and
investigated a permeable, rigid flat bottom of infinite thickness. This resulted in a
reconsidered expression for p0 :
ρ w gH 1
p0 = Eq. 5-3
2 K
cosh kh − sinh kh
Tν
Considering the reference soils, the second term in the denominator is only significant if
the intrinsic permeability κ is in the order of Tν (≈ 10-5), which is the case for very rough
gravel (e.g. filter layers); sand, silt and clay are less permeable for short waves.
Neglecting this second term introduces an error of at most 0,002 %.
55
It seems inconsistent that the wave period T is in the denominator; after all, the seabed
should be more permeable, if the wave period increases, but with increasing T, the
wavelength L also increases and the horizontal wave particle velocity over-all increases.
Therefore, longer waves experience a less permeable seabed.
Tsui et al. [1983] compared these theoretical computations with measured values from
experiments. First they investigated wave pressures (linear and second order wave
climate) on a completely impermeable, rigid and horizontal concrete bottom. The
predicted pressures lie within a confidence band of 0,02 kPa, while the measured
pressures varied from 0,069 to 0,207 kPa.
Then they placed a layer of Chattahoochee sand on the concrete bottom: two different
porosities were considered, a loose (n = 0,52) and dense (n=0,38) packing as well as
two different layer thicknesses 0,23 m and 0,33 m. According to Liu, the bottom could
still be considered ‘impermeable’. It appeared that indeed the wave pressures on top of
the dense sediment showed good resemblance to theory. However with increasing
thickness of the sediment layer and with increasing wave periods, the measured values
of the wave pressures became significantly smaller than the theoretical values.
Tsui et al. introduced the effective water depth h’. This effective water depth h’ is larger
than the real water depth h, but is chosen such that the measured wave pressure at
depth h is the same as the theoretical value at depth h’, see Figure 5-1 in which symbol
‘d’ is used for the water depth.
Figure 5-1: Ratio of effective water depth (d’) and real water depth (d) for two sediment
layers (ds = 0,23 m and ds = 0,33 m) on an impermeable layer [source: Tsui et al., 1983]
This effective water depth was found to increase with wave period T and thickness of
the sediment layer ds. Unlike Liu, they concluded that long period waves extend into the
entire thickness of the sand layer. The effective depth then equals the sediment
thickness and the wave-induced pressure on top of the sediment layer is smaller than
the theoretical value. The waves under consideration in this thesis, especially the larger
storm waves, have wave characteristics similar to the waves in these tests. So, the fact
that the seabed consists of permeable sediment instead of an impermeable layer,
makes it very likely that the actual water pressures at the sediment surface will be
somewhat lower that obtained from expression 5-3. This phenomenon however isn’t
widely adopted and still no analytical expression based on the soil stratification and
wave characteristics exists. It is therefore very hard to calculate with effective depths
and the expression of 5-3 will be considered as a safe upper boundary.
56
5.1.3 Water pressures inside seabed
Now these water pressures on top of the bed have to be translated into pore-water
pressures inside the bed, because they affect the resulting effective stresses in the soil.
Theoretically, both transient and residual pore-water pressures are generated in the
seafloor by wave loading. The transient pressures instantaneously result from the
travelling wave; the residual pressures are caused by the variation of dynamic wave
pressures in time and space. These pressures aren’t instantaneously related to the wave
loads, but depend on the intensity and duration of wave loading and the drainage
characteristics of the seafloor, see Paragraph 5.2.
Putnam (1949), who assumes the sand skeleton and the water to be incompressible and
hydraulically isotropic and only considers the transient pore-water pressures, gave one
of the first expressions. In his work he mainly investigated the wave damping. However,
implicitly he assumed the pore-water pressures to decay according to an inverse ‘e-
power’. Please note that the positive direction of the z-axis is upward:
cosh k (d s + z ) [e + kz + e − k ( 2 d s + z ) ]
pˆ wave;bed = p0 = p0 Eq. 5-4
cosh kd s [1 + e −2 kd s ]
in which z is the depth into the soil layer, k is the wave number (k=2π/L) and ds is the
thickness of the seabed on a rigid impermeable layer. Sleath extended this approach
and included hydraulic anisotropy. With kx and kz as the hydraulic permeabilities in
horizontal and vertical direction, the above equation changes into:
⎛ ⎛ k ⎞1/ 2 ⎞
cosh ⎜ k ⎜ x ⎟ (d s + z ) ⎟
⎜ ⎝ kz ⎠ ⎟
pˆ wave;bed = p0 ⎝ ⎠ Eq. 5-5
⎛ ⎛ k ⎞1/ 2 ⎞
cosh ⎜ k ⎜ x ⎟ d s ⎟
⎜ ⎝ kz ⎠ ⎟
⎝ ⎠
This equation has been validated in wave tank tests and despite a lot of scatter, the
experimental results are in line with theory, at least for sandy bottoms.
Most research studies are based on the uncoupled analysis: the coupling of the sand
skeleton and pore water in resisting the waves is ignored and the mechanical properties
of the sediment are neglected. Biot (1944) presented a coupled theory for a poroelastic
solid which accounts for elastic deformation of the porous medium, compressibility of
the pore fluid and Darcy flow (consolidation however is not accounted for). This theory
yields a relatively simple solution for a seabed of infinite thickness which is hydraulically
isotropic and under the assumption that the stiffness of water is much larger than the
stiffness of the sand skeleton. The amplitude of the wave pressure inside the seabed:
pˆ wave;bed = p0 e + kz Eq. 5-6
The above equation can also be obtained from Sleath’s formula with kx = kz and ds → ∞.
Under natural conditions the horizontal permeability often isn’t larger than 1,5 or 2
times the vertical permeability. Please note that here parameter ‘k’ represents the wave
number in the seawater, while ‘k’ with a subscript represents the hydraulic permeability
in the seabed! Normally, in both separate disciplines (fluid and soil mechanics) these
parameters do not interfere.
Experiments done by Tsui et al. [1983] confirmed the ideas behind the expressions:
1) the pressure ratio p/p0 decreases with increasing distance z into the sediment layer;
2) the pressure ratio p/p0 is larger in loose sands than in dense sands due to higher
permeability;
3) the pressure ratio increases with increasing period T.
57
The above conclusions are qualitatively in line with the theoretical expressions of
Putnam and Liu, but there are also some striking differences. The measured values for
loose sediment are more in agreement with both theories than the values for dense
sediment. Putnam and Liu do not take sediment properties (other than permeability)
into account and their formulas seem to be valid only for very permeable soils.
Tsui et al. [1983] found an increasing difference between theoretical values and test
results for decreasing permeability. So, wave-induced water pressures in loosely packed
sand, which happens to be most sensitive to wave action, can well be represented by
equation 5-6, but this equation may probably be not so applicable in dense sea beds.
Summarizing, the wave-induced pressure on top of the bed is mainly influenced by the
thickness of the sediment layer and hardly by permeability. The attenuation of wave
pressures in the bed is strongly dependent on permeability.
To illustrate the impact of wave-induced fluctuating water pressures on the hydrostatic

pressure, a sort of upper boundary is presented. Most unfavourable are shallow waters
with high, long waves. The influence on the water pressures can be in the order of 25 %
in case of depth-limited waves in shallow water. The value of 25% can theoretically be
explained if one considers the ratio between the amplitude of the wave pressure and the
hydrostatic water pressure at the surface of the sea bed. If the wavelength L increases
to infinity and the wave height is limited to half of the water depth:
ρ w gH
wave _ pressure 2 cosh(−kh) k ↓0 1
L → ∞ ⇒ k → 0 ∧ H = 12 h : = ⎯⎯→
⎯ Eq. 5-7
hydrostatic _ pressure ρ w gh 4
The above approximation is realistic. For instance, if the water depth equals 5m, the
wave height 2,5 m and the wave period 15 s, it means that the wavelength is 105 m
and the ratio between wave pressure and hydrostatic pressure already becomes 0,239.
An important additional phenomenon is the occurring time lag. Although the wave
pressures in the sediment also have a sinusoidal character, they lag behind the wave
pressures on top of the sediment surface. Tsui et al. found that this time lag could grow
as big as one-third of the wave period, see Figure 5-2.
Water pressure
water pressure at surface; p0

water pressure at depth z; pz
net upward pressure; pnet
Time
Figure 5-2: Wave-induced net upward pressure caused by time lag
Water pressures on top of the seabed and at a certain depth z inside the bed are plotted
for the unfavourable time lag of one third of the wave period. The net upward pressure
is represented by the green line and is largest under a wave trough.
Although some researchers warn against failure due to net upward pressures which may
occur under a wave trough, the risk may be doubted. The time lag increases with
increasing depth, but the effective overburden pressure (weight of the overlying soil)
also increases. Demars [1983] considers the effect of this phase shift negligible for a
flat seabed in practical purposes. However, in case of very long period waves, caution is
recommended, especially because this phenomenon isn’t fully understood. The risk of
slope failure due to net upward pressures is much more likely, especially for steep
slopes with micro-instability as normative failure mechanism.
58
5.1.4 Total and effective stresses inside seabed
Until now the development of cyclic pore water pressures has been discussed, but when
considering slope stability problems, effective stresses are more important (cf slip circle
analysis). The cyclic component of effective stress in the sea bed can be expressed as:
Δσ ' = Δσ − Δp Eq. 5-8
These cyclic effective stresses can be added to geostatic stresses if the seabed is
sufficiently rigid (unlike the plastic deformation described in the next paragraph) and
equilibrium is maintained. To obtain the total stress change, elastic theory on a
homogeneous, isotropic bed can be used. The vertical total stress change becomes:
Δσ z = p 0 (e kz − kze kz ) cos(kx) Eq. 5-9
and the horizontal total stress change is expressed by:
Δσ x = p 0 (e kz + kze kz ) cos(kx) Eq. 5-10
while the change in shear stress in the x-z-plane becomes:
Δτ xz = p 0 kze kz sin(kx) Eq. 5-11
These total stress changes caused by waves are presented in Figure 5-3. They have
been normalized by the wave pressure under a wave crest on top of the seabed. The z-
axis represents the dimensionless depth ‘2z/L’ (=kz/π), positive upward.
normalized stress and pressure

-0,5 0 0,5 1 1,5
0
dimensionless depth 2z/L
-0,25
-0,5
-0,75
-1
p^/p0 Δσz/p0 Δσx/p0
Figure 5-3: normalized sub-bottom stress and pressure profiles due to waves
The effective stress changes can be obtained by subtracting the pore-water pressure
changes from the total stress changes. It appears that the horizontal and vertical
effective stress changes have the same magnitude but a different sign:
59
Δσ ' z = − Δσ ' x = − p0 kze kz cos(kx) Eq. 5-12
Under a wave crest, the vertical total and effective stress reach their maximum value,
while under a wave trough the minimum value is reached. Under a wave node there are
no changes in horizontal and vertical total and effective stresses. The expressions of the
shear stress and effective stress are very similar; only the phase is different. Maximum
shear stresses occur under wave nodes.
In Figure 5-4 the vertical and horizontal effective stress changes under a wave crest are
presented; horizontal effective stress changes are negative under a wave crest and
positive under a wave trough. Furthermore it can be seen, that maximum values occur
at a depth of about one-sixth of the wavelength and that these stress changes are
about 0,37*p0. It was already mentioned that this fluctuating water pressure p0 cannot
exceed 25% of the hydrostatic pressure, so this change in effective stress does not look
very spectacular (less then 10% of the hydrostatic pressure). However, one should note
that just below the seabed, effective stresses are still small. This is a first indication that
wave-induced failure is a rather superficial phenomenon.
normalized effective stress

0 0,1 0,2 0,3 0,4 0,5
0
dimensionless depth 2z/L
-0,25
-0,5
-0,75
-1
Δσ'z/p0 = -Δσ'x/p0
Figure 5-4: normalized effective stress under a wave crest
The effective stress state for any element determines stability. In absence of waves this
stress state normally is represented by Mohr’s circle, see Figure 5-5. If the bed is
completely rigid, the excess pore pressures only generate from bed flow caused by
waves and the stress at any moment is the sum of the geostatic stress and wave-
induced increment.
60
effective shear stress
effective normal stress

no-wave wave crest wave trough failure line
Figure 5-5: Mohr’s Circle due to waves
In generally, the effective stress circle for the no-wave condition will increase
concentrically in diameter when loaded by a wave crest (‘purple’ line) and decrease
concentrically in diameter when loaded by a wave trough (‘red’ line), because the cyclic
stress is such that Δσz = -Δσx. An element is at a state of failure if Mohr’s circle for that
element contacts the failure line.
Calculation example
Neglecting potential time lag effects, this example will show that failure of a flat seabed
due to transient wave stresses is not very likely. Let’s consider a seabed at 10 m depth
of LFS with depth-limited waves (Hs = 5 m, T = 10 s, L = 93 m). At the seabed, the
hydrostatic pressure will be 100 kPa and the cyclic wave pressure (Eq. 5-2) will be 20,2
kPa. The maximum change in effective stress can be approximated by 0,37*p0 = 7,5
kPa, but this maximum only occurs at about 15,5 m below the seabed and at that depth
a significant (geostatic) effective stress already is reached.
The cyclic stress ratio (CSR = Δσ’z /σ’z), which can be obtained by dividing Figure 5-4 by
the linear geostatic effective stress, is plotted in Figure 5-6. Please note that at the
seabed (z=0) this CSR is not defined, because the effective stress is zero at the seabed:
Cyclic Stress Ratio (CSR)
Δσ' z/σ' z [-]
0,00 0,05 0,10 0,15 0,20
0
Depth below seabed [m]
-10
-20
-30
-40
Figure 5-6: Cyclic Stress Ratio against depth below seabed; calculation example
In this example the maximum CSR occurs at the bed surface and is about 0,15.
Theoretically, wave-induced bed failure can only occur if this ratio equals 1; this means
that under a wave crest, the effective stress is reduced to zero. In practical situations, it
can be shown that the CSR will never exceed 0,40 to 0,45. In Figure 5-7 the maximum
CSR’s for four different water depths h are plotted against wavelength. The solid part of
a line represents the wavelengths belonging to wave periods of 5 to 15s according to
linear wave theory. Wave heights are depth-limited (H=0,5h). So, maximum CSR’s
occur for larger wavelengths if the water depth increases. Please note that in more
shallow water this depth-limited wave will be more frequent, because a larger part of
the total incoming wave spectrum will be truncated down.
61
Unfavourable situations can occur when the saturated unit weight of the bed material
decreases or another wave breaking criterion is used, for example the solitary wave
breaking criterion (H=0,78h).
(γsat = 16 kN/m3, H=0,78h, CSR = 0,43).
0,30
h = 5m
h = 10m
0,25
h = 15m
h = 20m
CSR, Cyclic Stress Ratio [-]
0,20
0,15
0,10
calculation example
0,05
0,00
0 40 80 120 160 200 240
Wave Length L [m]
Figure 5-7: Maximum CSR’s for different water depths against wavelengths
Demars [1983] showed that failure of the seabed can only occur if the permeability in
vertical direction is much smaller than in horizontal direction: kx/kz > 7 or 8. Typical
values for this permeability-ratio in a seabed are 1,5 or 2, so transient seepage failure
is not very likely.
5.1.5 Slope stability under transient wave action
Until now, only the consequences of wave loading on water pressures and effective
stresses in the seabed have been discussed and it was shown that failure of a seabed in
the direct, elastic approach is not expected to occur. It is almost needless to state that
slope stability can be drastically reduced.
Macro-stability
Two failure modes are often suggested in literature:
I. Rotational sliding
II. Infinite slope sliding
The first one is very similar to the slip circle analysis, described in Paragraph 4.1,
although the calculations become much more complicated. Besides a number of slip
circles with different radii and centre points, also a number of waves with different
wavelength, wave height and location with respect to the slope, have to be considered.
These waves affect the effective stresses and shear strengths; driving moments can be
increased, resisting forces reduced. Besides iteration with centre point and radius of the
slip circle, also the wavelength and position of wave crest, trough and node with respect
to the slope have to be iterated. The wavelength, for which the maximum CSR occurs,
does not necessarily have to be most unfavourable; it is also dependent on slope height
and inclination angle. Because of the diminishing influence of waves with increasing
depth, it can be expected that the normative slip circle is less deep and in some way
related to the wavelength. Although the volumes that are involved in rotational sliding
due to waves can be smaller compared to unloaded shear failure, this does not mean
that the consequences are smaller. A sequence of such circular slides can result in
progressive down slope movement.
62
The second failure mode occurs when at some depth the shear stresses exceed the
shear strengths and sliding develops on a surface almost parallel to the seabed. This
failure mode can be expected at small slope angles in very soft clays/muds, which have
a strong stratification and therefore not the subject of this thesis.
Micro-stability
In Paragraph 4.4 submarine micro-stability proved to be dependent on the angle of
internal friction only: submarine slopes could be as stable as subaerial slopes. As was
mentioned before, the presence of a hydraulic gradient will influence stability. An
inward-directed hydraulic gradient, like for instance under a wave crest, will allow for
steeper slopes (β>φ); an outward-directed hydraulic gradient, like under a wave trough,
requires flatter slopes (β<φ). In practice, one would never count on positive gradients in
slope design, so only outward-directed hydraulic gradients will be considered.
In case of a slope completely situated below sea level, seepage flow will always be
directed perpendicular to the slope, if the seabed is an equipotential line. Van Rhee and
Bezuijen [1992] investigated two theories for the stability of the outer grains:
- continuum mode
- single-particle mode
The former theory is based on equilibrium of forces on a certain unit volume of soil at
the outer slope and is most widely adopted. The latter theory is based on the stability of
a single grain: the rolling criterion, determined by the total moment on the grain,
controls stability. For outward flow their test results showed that the continuum mode
determines stability, while the single-particle mode was normative in case of inward
flow. So only the continuum theory will be presented. Because micro-stability is the
normative failure mechanism for non-cohesive soils, only these soils will be under
investigation (c=0).
l
d
S
T
G N
β
Figure 5-8: Schematization of forces
The small rectangular unit weight (Figure 5-8) is just stable, if a well-known friction law
is fulfilled:
T = N tan φ Eq. 5-13
Forces T and N can be deduced from equilibrium of forces parallel and perpendicular to
the slope:
T = G sin β
Eq. 5-14
N = G cos β − S
Force G is the weight under water and can be described as follows:
G = ( ρ sat − ρ w ) gld = (1 − n)( ρ s − ρ w ) gld Eq. 5-15
The flow force S is calculated according to Darcy’s Law (i=1/ρwg δp/δx):
63
S = iρ w gld Eq. 5-16
The hydraulic gradient is positive for outward-directed flow. Combining equations 5-13,
5-14, 5-15 and 5-16 yields an expression for the maximum hydraulic gradient:
cos β tan φ − sin β sin(φ − β )

imax = (1 − n)Δ = (1 − n)Δ Eq. 5-17
tan φ sin(φ )
The maximum hydraulic gradients (imax) are plotted in Figure 5-9 for all non-cohesive
reference soils. This hydraulic gradient is a rather trivial quantity when dealing with
different piezometric levels, for instance during ebb in the tidal zone. The problem
becomes a little more complicated in presence of transient waves: the hydraulic
gradient will change in time and alter direction. The most vulnerable part of a slope to
wave loads will be the top of the slope (unlike for example static liquefaction), because
wave pressures diminish rapidly with depth. The hydraulic gradient due to waves will be
approximated by derivation of expression 5-1.
1 δp 0 Hk
i= =− sin( kx − ωt ) Eq. 5-18
ρ w g δx 2 cosh kh
Substituting equation 5-17 into 5-18 yields for the maximum wave height:
2(1 − n)Δ sin(φ − β ) cosh kh

H max = Eq. 5-19
k sin(φ )
1,2
1,0
hydraulic gradient i [-]
0,8
0,6
LSI
LCS LMS LFS
0,4
MSI
MCS MMS MFS
0,2
DSI
DCS DMS DFS
0,0
0 2 4 6 8 10 12 14 16 18 20
slope 1:x [-]
Figure 5-9: Maximum hydraulic gradients for all non-cohesive reference soils against slope
To gain insight in the impact of wave loads on micro-stability, the maximum waves
according to equation 5-19 have been calculated for LSI and DCS/DMS/DFS for different
water depths (h=5, 10, 15, 20m). It was already proved that the normative wavelength
(and corresponding wave period) is dependent on the water depth. So, from all
calculations the most unfavourable wavelength has been determined. The normative
wave heights are presented in Figure 5-10: the solid lines represent the maximum wave
heights for a certain slope in LSI (weakest reference soil) and DCS (=DMS, DFS;
strongest reference soil); these lines convert into dashed lines at the depth-limited
wave height. The most unfavourable wave periods for a certain water depth have been
used to determine a sort of upper boundary.
64
10
9
maximum wave height Hmax [m]
0
1,0 1,5 2,0 2,5 3,0 3,5 4,0
slope 1:x [-]
LSI (h=5m; T=5s) LSI (h=10m;T=6s) LSI (h=15m; T=8s) LSI (h=20m; T=9s)
DCS (h=5m; T=5s) DCS (h=10m; T=6s) DCS (h=15m; T=8s) DCS (h=20m; T=9s)
Figure 5-10: Maximum wave heights (upper boundaries) for weakest and strongest reference
soil
The primary conclusion to be drawn from Figure 5-10 is that the impact on micro-
stability of depth-limited waves does not exceed approximately 20%. This means that a
just stable ‘unloaded’ slope in non-cohesive soil has to be flattened by about 20% to be
able to resist the largest possible waves. The critical area of the slope is the upper part.
Inside trenches and channel not only will the depth be larger, due to wave shoaling the
wave impact will further decrease. Apparently the direct effect of waves remains rather
small and is often implicitly accounted for in the safety factor.
65
5.2 Indirect Plastic Behaviour
5.2.1 Dynamic liquefaction by wave loads
Static liquefaction, as explained in Paragraph 4.2, was determined considering a sudden

(relatively small) trigger. When these trigger force is a large, cyclic force (earthquakes,
storm waves, tsunamis) another approach is advised. In this thesis, this case will be
investigated for large storm waves. In Paragraph 5.1 the direct, elastic effect of waves
on the sea bed was explained: waves instantaneously affect pore water pressures and
total stresses, thus implicitly effective stresses. Although this reduction of strength can
result in stability problems, these elastic changes of water pressures and effective
stresses have no net effect over a full wave cycle. But as was mentioned before, there is
also an indirect, plastic effect. Every wave load cycle will cause compaction of the soil
due to shear stress variations, generating excess pore water pressures (EPP’s).
Depending on the consolidation properties of the soil these EPP’s fade out after every
wave cycle or cumulate during a storm.
In this thesis this process will be explained briefly and demonstrated with the help of
the rather simple, but therefore very clear computer program MCycle by GeoDelft,
which calculates the development of water overpressures in time under storm
conditions. However, when these calculations show a large pressure built-up, more
advanced analyses are recommended.
MCycle solves the 1-dimensional consolidation equation for the excess water pressure
‘u’:
∂ 2 u ∂u
cv 2 = −A Eq. 5-20
∂z ∂t
in which “pumping term” A represents the undrained internal generation of water
pressure (A = (δu/δt)undrained) which is dependent on the cyclic properties of the soil, the
wave characteristics (external load) and the extent to what these waves are felt inside
the slope. This parameter will decrease with depth and vary in time. Parameter ‘cv’
represents the consolidation coefficient according to equation 3-5. Because both the
consolidation properties and the pumping term vary with depth, the seabed is divided
into a number of soil layers.
The hydraulic permeability k is kept constant during the compaction process, whereas
the consolidation coefficient cv is defined at a certain soil stress and is proportional to
the square root of the effective stress. It is important to note that the consolidation
coefficient cv is not a constant, but varies with both the level of stress and degree of
consolidation. For practical site settlement calculations, however, it is sufficiently
accurate to measure cv relative to the loading range applicable on site and then assume
this value to be approximately constant for all degrees of consolidation (except for very
low values).
The approach adopted in MCycle is based on the following assumptions:

-) transient waves in one direction;
-) the number of waves needed to built up the water overpressure is relatively large
-) no construction present on the seabed;
-) compressibility of the pore water is small compared to compressibility of the grain
skeleton;
The first three assumptions are related to the 1-dimensional approach: the vertical
consolidation is predominant. Because of the large number of waves (and therefore
small influence of a single wave) the horizontal consolidation may be neglected. In case
of only one or a few very steep waves, the problem becomes 2-dimensional and far
more complicated. When considering the direct, elastic effect in the previous paragraph,
the variations in pore water pressures and effective stresses under a single wave were
determined. In that approach, no pressure generation effects were assumed to occur.
66
The prerequisite of ‘a large number of waves’ also has a more practical background: the
wave climate has to be described as a regular wave field. In the direct, elastic approach
one could suffice with the largest occurring wave; in this indirect, plastic approach a
representative wave height has to be chosen: the significant wave height Hs. This wave
height has to be chosen very carefully. If one takes a long storm duration and a lower
value of the significant wave height, a small group of larger waves can increase excess
pore pressures (EPP’s) considerably and cause failure. But storms also must contain at
least 50 waves to fulfil the above mentioned assumptions. A sound description of the
complete storm may ask for multiple storms with various significant wave heights.
The negligible compressibility of the pore water means that volume reduction due to
cyclic compaction under fully saturated conditions only occurs if pore water is pressed
out. If this pore water cannot be pressed out (undrained conditions), excess pore water
pressures arise: effective stresses are reduced. The tendency to compact will be larger
for larger variations in the shear stresses. Important to note is that not the absolute,
but the relative variation of these shear stresses is meant (like in all friction-based
calculations): Δτ / σv0’. Parameter σv0’ represents the initial vertical effective stress.
Because this effective stress decreases during a storm, the soil is subjected to an
increasing ‘load’. This behaviour is often mentioned as the ‘explosive character’ of cyclic
liquefaction. Therefore in practice often an upper limit of the excess water pressure is
defined:
u ≤ 0,5σ v 0 ' Eq. 5-21
although ‘real’ cyclic liquefaction occurs if effective stresses are reduced to zero.
The sensitivity of the cyclic response of sand to water pressures of 40-50% of the
effective stresses is presented in Figure 5-11 in which the excess pore pressure divided
by the effective vertical stress is plotted against the number of waves divided by the
total number of waves at which liquefaction will occur (Nl0). It can be observed that
during the first number of waves N, the excess pore pressure increases rather fast to a
value of about 30%, while quite a number of waves are needed to reach an excess pore
pressure of 50%, but then only a few extra waves can cause liquefaction.
1
relative excess water pressure
0,8
0,6
u/σ 'v0
0,4
0,2
0
0 0,2 0,4 0,6 0,8 1
relative number of wave cycles N/N l0
Figure 5-11: Relative excess water pressure against relative number of waves
Two special situations

When looking at the consolidation equation, one can distinguish two special situations:
I) the consolidation coefficient is very small: no noticeable consolidation occurs and the
increase in water pressure is equal to the generation term ‘A’: δu/δt = A
II) water pressure is constant in time: δu/δt = 0 and the equation 5-20 becomes:
67
∂ 2u
cv = −A Eq. 5-22
∂z 2
This represents the stationary situation that during every wave cycle the generated
water pressure disappears due to consolidation, so no net pressure built-up occurs. Both
situations are illustrated in Figure 5-12.
or very small cv cyclic liquefaction!
excess water pressure u
stationary situation for cv2

liquefaction
u = -σvo'
stationary situation for cv1
cv1 > cv2

NO cyclic liquefaction
time t
Figure 5-12: Excess pore pressure for two soils with different consolidation coefficients
Soil ‘1’ (green line) will never reach the liquefaction criterion (u = -σv0), whereas soil ‘2’
cannot reach the stationary situation, because the EPP becomes larger than the
effective stress and cyclic liquefaction occurs. Please note that in engineering practice
often the above mentioned more conservative liquefaction criterion (equation 5-21) is
applied due to the ‘explosive character’.
The stationary situation can be explained as follows. Due to drainage, volume reduction
occurs (increase of isotropic stresses) and this will result in a decreasing reduction of
the vertical effective stress during each wave cycle. As a consequence the excess water
pressure will decreasingly increase, until the stationary situation is reached. Now the
soil behaviour can be interpreted as ‘completely drained’.
However, under drained conditions the volume reduction per wave cycle will attenuate,
which can also be observed in drained cyclic triaxial tests. The consolidation/drainage
behaviour becomes predominant and the EPP starts to diminish, see Figure 5-13. So, in
fact the stationary situation will in reality never be reached.
68
excess water pressure u
or very small cv
u = -σvo' cyclic liquefaction!

NO cyclic liquefaction
stationary situation
only drainage
drainage and history effect
time t
Figure 5-13: Excess pore water pressure under undrained conditions and drained conditions
with and without history effect
The above described effect is often described as pre-shearing or ‘history effect’ (the
extent to what the soil has been subjected to cyclic loading before). The decrease in
porosity per wave cycle (Δn) is a good measure for this history effect. The fact that
‘virgin’ soils react different to storm waves than ‘mature’ soils can be explained by this
history effect.
The above described, mainly qualitative description of the behaviour of soils subjected
to wave loads has to be expressed in formulae. Therefore first the undrained cyclic
behaviour is taken as a starting point.
Seed en Rahman (1976, 1978) deduced a formula for the number of load cycles to
induce (undrained) cyclic liquefaction from a lot of laboratory tests:
0 , 351
⎛ 67,5 ⎞
log N l 0 = 3−⎜ (τ / σ v 0 ' ) − 12,25 ⎟ for Nl0 ≤ 1000
⎝ RD ⎠
0 , 351
⎛ 67,5 ⎞
log N l 0 = 3 + ⎜12,25 − (τ / σ v 0 ' ) ⎟ for Nl0 > 1000 Eq. 5-23
⎝ RD ⎠
This formula has been plotted in Figure 5-14 for different relative densities. Waves that
induce shear stresses of about 15 % of the vertical effective stress almost cause
immediate liquefaction in loosely packed soil (RD = 0,3), after 9 cycles in medium
packed soil (RD = 0,5) and after 50 cycles in densely packed soil.
69
0,40
0,35
RD = 0,3
0,30
RD = 0,5
0,25 RD = 0,7
τ / σv0' [-]
0,20
0,15
0,10
0,05
0,00
1 10 100 1000
number of load cycles for liquefaction in undrained conditions: Nl0 [-]
Figure 5-14: Development of relative shear stresses with number of load cycles for three
different relative densities
But are these relative shear stresses realistic when it comes to the reference geometries
and wave heights of this thesis? Therefore the relative shear stresses have been plotted
for depth-limited waves in water depths of 5, 10 and 20 m and wave periods of 5, 10
and 15 s with the remark that the maximum wave pressures for a certain water depth
occur for different wave periods (see Paragraph 5.1.4).
relative shear stress τ / σv0' [-]
0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14 0,16 0,18
0,0
h = 5m; T = 5s
0,5 h = 5m; T = 10s
h = 5m; T = 15s
1,0 h = 10m; T = 5s
h = 10m; T = 10s
1,5 h = 10m; T = 15s
h = 20m; T = 5s
h = 20m; T = 10s
2,0
depth z [m]
h = 20m; T = 15s
2,5
3,0
3,5
4,0
4,5
5,0
Figure 5-15: Development of wave-induced relative shear stresses as a function of sub-

bottom depth for different water depths and wave periods
So especially loosely to medium packed sands are susceptible to cyclic liquefaction.

Furthermore it can be observed that relative shear stresses faster diminish with depth
for smaller wave periods.
The relative excess pore pressure shows strongly non-linear behaviour, as was
mentioned before as the explosive character. Seed and Rahman delivered the following
expression that was plotted in Figure 5-11:
70
u / σ ' v 0 = 0,64 arcsin( N / N l 0 ) 0,714 Eq. 5-24
in which N is the number of wave cycles at a certain moment.
Pre-shearing or history effect

From test results obtained during the design of the Eastern Scheldt Storm Surge Barrier
Smits et al. [1978] introduced an expression for the number of wave cycles to induce
cyclic liquefaction after compaction of the sand skeleton due to drainage:
N l 0;hist = 10ε hist Δn N l 0 Eq. 5-25
in which Δn is the reduction of the porosity due to cyclic loading; εhist is a constant
representing the history effect. From the test results this constant could be estimated at
εhist = 830, but most of the times more conservative values are used, like εhist = 333,
because there is little knowledge of the ‘real’ behaviour in nature. There is however no
theoretical base for this value.
The introduction of pre-shearing causes a reduced sensitivity to generation of excess
water pressures. Because of the dependence on the reduction of porosity Δn this history
effect remains small in case of almost undrained conditions (small cv and large Hs),
because compaction does not occur.
Equivalent wave height and period

The irregular wave field in reality is in many hydraulic applications described with a
significant wave height Hs and wave period Ts. It is also desirable to be able to describe
the relative shear stress changes caused by an irregular wave field with a certain wave
height and wave period. To that purpose the impact of different wave groups has been
studied. When the so-called equivalent wave height He is equal to the significant wave
height Hs, it appeared that the equivalent wave period Te should be 2,3 times the wave
period belonging to the significant wave height Ts to produce the same pressure built up
during a storm. This equivalent wave period Te is only needed to calculate the total
number of waves within a certain period; the relative shear stress changes are
calculated with the ‘normal’ wave properties Hs and Ts.
This value of 2,3 (MCycle uses 2,0) can be explained by the fact that Hs is a rather large
wave; only 13,5% of the waves will exceed the significant wave height in case of a
Rayleigh-distributed narrow wave spectrum. So, the pressure built up will be governed
by the more frequent smaller waves or, and this approach has been adopted, waves
with a artificially increased wave period:
He = Hs
Eq. 5-26
Te = 2,3Ts
Pumping term A
The pressure generation or “pumping” term A can be obtained from the expression for
the relative excess pressure (equation 5-24) in which the quotient N/Nl0 is substituted
by a new expression, which combines the history effect and the efficient wave period:
N = t Eq. 5-27
N l 0;hist N l 0;hist Te
The expression for the excess water pressure becomes:
0,714
⎛ t ⎞
u (t ) = 0, 64σ 'v 0 arcsin ⎜ ⎟⎟ Eq. 5-28
⎜N
⎝ l 0;histTe ⎠
When this expression is differentiated to t, the undrained generation term A is obtained:
71
⎛ ∂u ⎞
A=⎜ ⎟ Eq. 5-29
⎝ ∂t ⎠undrained
Calculation method
The calculation method of MCycle by GeoDelft is roughly as follows:
- with the storm and wave input the wave pressures on top of the seabed and the
relative shear stress changes as a function of sub-bottom depth are calculated
according to the 2-dimensional direct, elastic approach of Paragraph 5.1 (Eq. 5-
11);
- with this relative shear stresses and the relative density the number of load
cycles is determined according to Eq. 5-23;
- together with the wave period, the undrained pumping term ‘A’ can be
calculated as a function of depth according to Eq 5-28 and 5-29;
- for every depth the 1-dimensional consolidation equation is solved, resulting in
the EPP as a function of depth and time.
The 1-dimensional approach is widely adopted, but there is some uncertainty about the
correctness of the uncoupled analysis of the 2-dimensional wave penetration and 1-
dimensional generation of EPP’s.
The computer model MCycle can calculate three different situations:

- water overpressure in a horizontal sea bed;
- water overpressure in an infinite under water slope that is entirely located below
sea level outside the breaker zone;
- water overpressure under a construction on the sea bed, assuming the length of
this construction is much larger than the width.
The second situation is solved just like the situation of a horizontal sea bed with pore
water flow perpendicular to the slope. Additionally, the stability of the slope will be
checked.
Restrictions
Some other restrictions of MCycle (some were already mentioned) are:
- this program is 1-dimensional. 2-dimensional effects are only accounted for in a
schematized way;
- the permeability of the soil is kept constant during compaction by waves;
- only five soil layers can be modelled;
- the number of waves needed to increase the water pressure is relatively large; in
other words generation term ‘A’ is very small;
- three storms with wave heights only varying in time can be prescribed;
- the compressibility of the sand skeleton is much larger than the compressibility of the
pore water
Numerical solution procedure

For the numerical solution of the consolidation equation a forward explicit integration
scheme is applied. To guarantee stability, the time step is restricted to:
(Δz ) 2
Δt ≤
2c v
This means that the maximum time step will be determined by the upper sublayer
(smallest thickness Δz) and the initial situation (t=0), because the consolidation
coefficient cv will decrease in time when water pressures increase and effective stresses
decrease (cv is related to root of the effective stress).
Stability under wave loads of an infinite long slope

The above mentioned approach for a flat seabed is adopted for wave loads on a slope.
At first the actual slope of the seabed is ignored: the consolidation equation is applied at
a line perpendicular to the seabed; the water depth (and the corresponding wave
loading) is of course dependent on the location on the slope. When the bottom is
72
sloping, in principal multiple calculations (for every point along the slope) with various
water depths and wave heights (which are depth-related) have to be done.
In the case of trenches and channels, which are in fact local ‘dips’ in the surrounding
seafloor, the surrounding water depth h0 and depth-limited wave height are normative.
Failure therefore means: a small, relatively shallow sliding of the upper part of the
slope, which can result in a large mass flow, downward the slope. At the beginning,
failure due to wave loads contains small amounts of sediment, but can eventually
become just as dangerous as slip circle failure or static liquefaction.
sinusoidal wave load

MSL
hnormativ e
h
z normativ e
channel bottom
z seabed
Figure 5-16: Schematization of calculations at two locations on the sloping bottom
After the consolidation equation in MCycle is solved numerically, the actual slope is
taken into account and the ‘stability check’ will be performed. The safety factor, under
the influence of wave loads, can be calculated, based on a simple schematization of
force equilibrium that is very similar to the approach of Paragraph 5.1.5. Instead of the
seepage force on the outer grains, now the difference in wave pressure (p0-p1) over a
certain depth is calculated. The maximum amplitude of this wave pressure is
represented by the variation of the effective stress (equation 5-12). The fact that the z-
axis isn’t really perpendicular to the slope surface is neglected. Another difference with
the ‘seepage approach’ is the addition of a term representing the excess water pressure
‘u’.
Force Equilibrium
-perpendicular to slope:
p0 + ρ w gh + ρ sat gz cos β = p1 + ρ w g (h + z cos β ) + u + Fg Eq. 5-30
in which Fg represents the total pressure by the grains from inside the slope.
-parallel to slope:
Ff = ( ρ sat − ρ w ) gz sin β Eq. 5-31
in which Ff is the friction force along the shear surface. The slope will be stable, if:
Ff < Fg tan φ Eq. 5-32
Combining these three equations yields an expression for the excess water pressure,
which can be interpreted as the slope stability check:
sin(φ − β )
u < ( p0 − p1 ) + (1 − n)( ρ s − ρ w ) z Eq. 5-33
sin(φ )
in which (p0-p1) is represented by:
73
ρ w gH
p0 − p1 = Δσ 'z = kzekz Eq. 5-34
2 cosh kh
If anytime during the calculation procedure instability occurs, the computer program
stops the simulation and the input data has to be adjusted.
5.2.2 Computations with MCycle for different slopes
Most important issue is of course the influence of the slope to the cyclic liquefaction
potential: how effective is flattening of the slope in increasing slope stability? As was
concluded before, see Paragraph 4.2.2, this was a very powerful measure when dealing
with static liquefaction.
A number of 36 simulations were done for all non-cohesive reference soils (see
Appendix C for the soil properties) for slopes (12 in total) of 1:3, 1:5 and 1:7 in a
surrounding water depth of 10m. The significant wave height was varied in steps of
0,25 m to find the maximum wave height which just does not cause slope failure due to
cyclic liquefaction with the restriction that the maximum wave height is depth limited to
half of the water depth: Hs=0,5h0.
In all simulations the same storm pattern is used. The total storm duration is 3 hours
according to a trapezoidal profile; see the little graph at the right bottom in the next
figures. The soil is again considered to be completely homogeneous, represented as a
rather thick soil layer of 20 m, divided into 40 sub layers. Some representative results
are presented, but because of the similar appearance of many of the graphs, not all
graphs will be drawn.
In Figure 5-17 for MFS can be seen that the excess pore pressures in the upper layers
increase rather fast (pink line), but when the storm progresses these pore pressures
diminish and spread out over the depth. The critical period is just after the storm
reached its maximum wave height. When the storm has ended, excess pore pressures
reduce very fast. If we consider the same seabed, but now in loosely packed fine sand
(LFS), see Figure 5-18, it becomes clear that the significant wave height is much
smaller (4 m against 1,44 m).
Excess Pore Pressure / Vertical Stress [-]
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Depth below sea bed [m]
Storm
5
Wave Height [m]
4
6
3
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
t = 1/4*Tstorm t = 1/2*Tstorm t = 3/4*Tstorm t = Tstorm
Figure 5-17: Excess pore pressure development in a slope of 1:5 in MFS
74
This is caused by the most important difference between LFS and MFS, the relative
density, which appears to be the most influential parameter: compare RDLFS=16% and
RDMFS=43%. Although the normative storm is much smaller, the excess pore pressures
pattern is almost similar for both soils, because the other important soil properties,
hydraulic permeability and consolidation coefficient, only slightly differ from each other.
It appears however that very loose soils increase their resistance to cyclic liquefaction
faster than denser soils. In fact every wave cycle compacts the soil, thereby increasing
this resistance. This can be observed from the somewhat lower excess pore pressures in
MLS for ½ and ¾ of the total storm period.
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
2
Wave Height [m]
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-18: Excess pore pressure development in a slope of 1:5 in LFS
From the above it can be concluded that soils with low relative densities in moderate
wave climates become more compacted, either by the often passing waves or already
during dredging activities. The before mentioned history effect, although difficult to
account for, is important. It seems relevant to neglect failure caused by waves that are
common in the area under investigation (i.e. small waves), because the soil will
automatically have adapted to this wave climate.
In DFS the slope even remains stable when subjected to depth-limited wave heights.
This situation is considered ‘unconditionally stable’. Dense fine sand (DFS) is stable for
all slopes and wave heights, because of its combination of a large relative density,
hydraulic permeability and consolidation coefficient, see Figure 5-19. Even waves of 5 m
are hardly capable of causing excess pore pressures. Another remarkable thing is the
fact that pore pressures keep increasing during the storm, while loosely packed soils
were mainly susceptible to cyclic liquefaction in the beginning of a storm period. This is
probably caused by the large number of wave cycles to induce liquefaction because of
the larger relative density.
75

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
6
Wave Height [m]

5
6
4
3
2
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-19: Excess pore pressure development in a slope of 1:5 in DFS
Until now, excess pore pressures were presented relative to vertical stresses. It is
however interesting to see where maximum absolute excess pore pressures occur to
demonstrate the differences between medium and densely packed soils, see Figure 5-20
and Figure 5-21.
Excess Pore Pressure [kPa]
0,0 1,0 2,0 3,0 4,0 5,0 6,0
0
6
10
Storm
5
12
Wave Height [m]
14 3
2
16 1
0
18 0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
20
Figure 5-20: Absolute excess pore pressures in a slope of 1:5 in MFS
In medium packed sand, maximum absolute excess pore pressures occur a few meters
below the seabed and halfway the storm period. Pore pressures tend to drain off to
deeper layers, where they can do less harm, because of the larger geostatic stresses. In
soils of higher density, excess pore pressures keep increasing in time and in depth
below the sea bed. Because of this large depth, these pore pressures are not very
dangerous. Even large long-lasting storms will not cause problems.
76
Excess Pore Pressure [kPa]

0,0 1,0 2,0 3,0 4,0 5,0 6,0
0
6
10
Storm
6
12
Wave Height [m]

5
4
14
3
2
16 1
0
18 0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
20
Figure 5-21: Absolute excess pore pressures in a slope of 1:5 in DFS
All 36 computations are summarized in Figure 5-22 with the remark that the densest
packed soils (DCS, DMS, DFS, DSI) are neglected, because no instability occurred for
maximum depth-limited waves. On the other hand, the loosest soils fail when subjected
to relatively small waves, which is an indication that such densities are not very likely in
a moderate wave climate.
5,5
4,5
Wave Height Hs [m]
3,5
2,5
1,5
0,5
2 3 4 5 6 7 8
Slope 1:x [-]
LCS MCS LMS MMS LFS MFS LSI MSI
Figure 5-22: Maximum wave heights for just stable slopes
The common medium packed soils are sensitive to storm waves, with the exception of
MCS, which is unconditionally stable (Hs > 5 m):
MSI: Hs ≈ 3,25 m
MFS: Hs ≈ 4,0 m
MMS: Hs ≈ 4,75 m
Please note that the abrupt bends in the lines are caused by the discrete steps in wave
height of 0,25m. The remarkable line of LCS has a steep part due to the good
consolidation properties and is then truncated down at the depth-limited wave height.
It becomes clear that due to the superficial influence of waves, the slope steepness isn’t
that important. Coarser sediments show some increase of stability if the slope is
77
flattened, but fine sediments, like silts, seem to be insensitive to slope flattening, see
also Figure 5-23. Clearly can be observed that maximum wave heights increase for
larger grains (read implicitly: better consolidation properties). This behaviour is less
pronounced with small waves, which can be understood when looking at Figure 5-14. At
very small relative shear stresses, the lines are almost flat around 300 load cycles
(t/2,3Ts = 7200/(2,3*10)=313 load cycles).
Especially medium packed fine and loosely packed coarse soils show some slope
dependency. One should however keep in mind that once a flow slide has been initiated
by large storm waves, the consequences are much larger for steeper slopes: the
resulting sediment gravity flow will be much more erosive on these slopes. This
computer program does not give any predictions about this phenomenon. One can
imagine that erosion of the toe of a 1:3-slope by such a gravity flow truly affects slope
stability; for example a circular slip circle may occur after cyclic liquefaction followed by
a turbidity current.
6
Maximum significant wave height H s [m]
1 RD = 0,2; tan β = 1:3 RD = 0,5; tan β = 1:3

RD = 0,2; tan β = 1:5 RD = 0,5; tan β = 1:5
RD = 0,2; tan β = 1:7 RD = 0,5; tan β = 1:7
0
0 200 400 600 800 1000
Grain size d50 [μm]
Figure 5-23: Maximum wave heights against grain sizes for different relative densities
We have seen that normative wave heights are quite different for the various reference
soils, but is there also a difference in the so-called ‘depth of influence’. In this thesis this
‘depth of influence’ is defined as the depth at which the excess pore pressures become
smaller than 10% of the total vertical stress. And what about the storm duration?
78

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
4
Wave Height [m]

6 3
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-24: Excess pore pressure development in a slope of 1:5 in MSI
When comparing Figure 5-24 (MSI) with Figure 5-17 (MFS), one can notice two
phenomena:
- the ‘depth of influence’ of MFS is about 5 m, whereas in MSI this depth reduces
to 2 m;
- in MFS the critical time is just after the wave height has reached its maximum;
after that time the excess pore pressures slowly diminish. MSI shows the
opposite behaviour: excess pore pressures keep increasing and only start to
fade when the wave height is decreasing.
These trends can also be observed when looking at MMS, Figure 5-25:
- the ‘depth of influence’ stretches to 8 m;
- the critical time is also in the beginning of the storm after which the excess pore
pressures decrease very fast.
Both observations can be explained by the variation in hydraulic permeability ‘k’ and
vertical compressibility ‘mv’, which are together determining the consolidation coefficient
‘cv’, remember Eq. 3-5, in which the compressibility of the fluid (β) was neglected:
k
cv = Eq. 5-35
γ w (mv + nβ )
- a larger hydraulic permeability means that excess pore water pressures can easy
dissipate to deeper layers, thereby increasing ‘the depth of influence’. Because
pore pressures are spread out over a larger area, bigger waves can be
withstood;
- the vertical compressibility (mv) determines the time in which compaction of the
grain skeleton and the matching decrease of pore volume takes place. If this
consolidation process progresses fast, the critical period will be at the beginning
of the storm. So, although compression results in a stiffer soil skeleton and
therefore a smaller susceptibility to cyclic liquefaction, too fast compression is
dangerous.
79

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
6
Wave Height [m]

5
6
4
3
2
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-25: Excess pore pressure development in a slope of 1:5 in MMS
What can be said with respect to the reference soils? Coarser soils generally have a
larger hydraulic permeability than finer soils. More loosely packed soils also tend to
have a larger permeability, but that is often accompanied by a larger compressibility, so
the consolidation coefficient for soils with equal grain sizes, but different porosities
usually is of the same order. However, the unfavourable small relative densities of
loosely packed soils result in a small number of load cycles to induce liquefaction (Figure
5-14) and therefore in a large pumping term (Eq 5-28 and 5-29), or in this approach,
where the maximum wave height is investigated, a small relative shear stress, because
the number of load cycles is determined by the storm duration.
Returning to the observed phenomena of depth of influence and critical time, it can be
explained that MFS (Figure 5-17) starts to benefit from its better consolidation
properties much faster than the equal dense MSI (Figure 5-24). Because EPP’s in MFS
faster drain off to deeper layers, the critical time is in the beginning of the storm. In
MSI with the poorer consolidation properties, every new wave will add a small
contribution to the EPP’s, because hardly any drainage occurs. Therefore the depth of
influence will be smaller and the critical time is at the end of a storm. Consequently the
maximum resistible wave height will be somewhat smaller.
When soils with comparable consolidation properties, but different relative densities are
compared (like MFS and LFS, Figure 5-18), the EPP development is rather similar, but
the main difference is the maximum resistible wave height: the number of load cycles is
determined by storm duration, but the relative shear stress has to be much smaller.
Please note that the unfavourable consolidation properties of pure silts are seldom
found, because in nature often mixtures occur. For instance, silty clays are very
impermeable, but have cohesive properties, which increase the resistance against
liquefaction.
5.2.3 Computations with MCycle for different storms
To consider the influence of different storm patterns, the following situations have been
simulated, all for 1:5-slopes in MFS:
A. 1 ‘trapezoidal’ storm with a wave height of 4,0 m and a duration of 3 h, similar
to Figure 5-17;
B. 1 ‘rectangular’ storm with the same wave height and duration;
80
C. 1 triangular storm with the same wave height and duration;

D. 1 smaller storm with a wave height of 2,5 m and a duration of 2 h, followed
after 1 h by storm a.
All storms are presented in the bottom right corner of the following graphs.
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
4
5
Wave Height [m]

6 3
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-26: Excess pore pressure development in a slope of 1:5 in MFS during storm B
It is based on Figure 5-26 and Figure 5-17 not easy to compare the influence of the
storm pattern, because of the presentation on certain moments. It is likely that the
‘pink line’ in Figure 5-26 does not represent the maximum excess pore pressures.
The fact that the ‘green and blue line’ are somewhat more positioned to the right in
Figure 5-17 can be explained by the fact that the consolidation process has less far
developed. Profound data analysis shows that maximum absolute excess pore pressures
are approximately the same (5,3 kPa), located at the same depth (2,37 m below the
sea bed). These maximum absolute EPP’s occur after 4557 s for storm A and after 2843
s for storm B, which shows that storm A has a delay on storm B of approximately a half
hour.
Maximum relative EPP’s (0,5) are found just below the sea bed. For storm A this
maximum occurs at 3560 s, while for storm B this happens to be at 1827 s. In both
cases maximum EPP’s appear to occur about half an hour after the wave height has
reached its maximum, which is consistent with the earlier conclusion that in medium
fine sands the situation in the beginning of a storm is normative.
81

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
4
5
Wave Height [m]

6 3
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-27: Excess pore pressure development in a slope of 1:5 in MFS during storm C
Storm c is less risky, because of the very slow increase of wave height. This is found in
the maximum absolute EPP (2,95 kPa at a depth of 1,73 m: less time for the waves to
penetrate into the seabed). Maximum relative EPP is 0,32. Both maxima occur just after
the peak wave height.
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40
0
2
Storm
4
5,0
4,0
Wave Height [m]
6 3,0
2,0
1,0
8
0,0
0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0
Time [hr]
10
Figure 5-28: Excess pore pressure development in a slope of 1:5 in MFS during storm D
One often assumes that smaller preceding storms have a favourable influence on the
stability of slopes. Although Figure 5-28 seems to confirm this assumption, it is hardly
true. Compare the blue line to the green line of Figure 5-17 (also halfway the main
storm) and there is no favourable effect noticeable. Although the EPP’s after the first
small storm have almost faded out, during the second storm EPP’s are reached similar
to storm A: maximum absolute EPP is 5,48 kPa and maximum relative EPP is 0,5.
Although not expected, the influence of smaller preceding storms appeared to be very
limited, at least for this combination of relative density and soil type.
82
It is however very well possible that longer and higher preceding storms (with only a
slightly smaller wave height than the main storm) have a more favourable effect.
Studying the development of the relative density in time revealed that this soil with a
porosity of 0,40 (RD=0,5) is only slightly compacted by waves. This is an indication that
preceding storms will have very limited effect. Therefore the same comparison is made
for a porosity of 0,42 (RD=0,4), see Figure 5-29 and Figure 5-30.
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
4
4
Wave Height [m]
6
2
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-29: EPP-development in a slope of 1:5 in MFS with n=0,42 during storm type A

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
4
4,0
Wave Height [m]
3,0
6
2,0
1,0
8
0,0
0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0
Time [hr]
10
Figure 5-30: EPP-development in a slope of 1:5 in MFS with n=0,42 during storm type D
It can be seen that the normative wave height is increased from 3 to 3,5 m when this
storm is preceded by a smaller storm. Maximum relative EPP’s are both 0,26 and occur
at nearly the same time, which is a half hour after the wave height has reached its
maximum.
83
It can be concluded that preceding storms are particularly favourable in case of loosely
packed soils, which can undergo a significant compaction (Δn). According to the same
line of thought, suddenly rising storms are far more dangerous than storms with
gradually increasing wave heights, especially for loosely packed soils.
5.2.4 Computations with MCycle for multiple layers
Another common phenomenon is the presence of multiple layers in the seabed. Freshly
deposited sediments are often less densely packed and therefore more susceptible to
liquefaction than older soil layers, but what is the effect on the normative storm wave
height?
Therefore two extra cases are investigated, which will be compared with a seabed of
homogeneous sand:
I a 5 m thick layer of loosely packed fine sand (MFS but with n=0,42) on top of
15m of sand with ten times smaller hydraulic permeability k (1,42*10-7 m/s) and
consolidation coefficient cv (0,0005 m2/s) and a relative density of 0,8;
II a 5 m thick layer of loosely packed medium sand (MMS but with n=0,42) on top
of 15 m of sand with ten times smaller hydraulic permeability k (8,8*10-7 m/s)
and consolidation coefficient cv (0,0031 m2/s) and a relative density of 0,8;
The excess pore pressure development of case I is presented in Figure 5-31 with the
remark that the y-axis depicts only the upper 10 m of the sea bed. The pore pressures
have difficulties in penetrating the denser layers. When case I is compared to Figure
5-29 (the same conditions but homogeneous MFS (n=0,42) over the entire depth), it
appears that such an impermeable lower layer does not affect the pore pressures of the
upper layer much. Because of the superficial wave-impact, the reduction of pore
pressure build-up in lower layers will not have any effect on slope stability: very deep
failures are not likely and the probability that a shallow failure occurs remains the same.
0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
4
4
Wave Height [m]
6
2
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-31: EPP development in a slope 1:5 with 2 layers: upper layer of MFS (thickness 5m)
and lower layer with ten times smaller k and cv (thickness 15m)
In Case II in the beginning of the storm similar behaviour can be observed: EPP’s are
less penetrating in the denser soil layer. During the storm another phenomenon can be
noticed, which was already mentioned at Figure 5-24. Excess pore pressures keep
increasing in the deeper parts of dense soils, because consolidation is a very slow
process.
84

0,00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40 0,45 0,50
0
2
Storm
5
Wave Height [m]

6
3
1
8
0
0 0,5 1 1,5 2 2,5 3 3,5
Time [hr]
10
Figure 5-32: EPP development in a slope 1:5 with 2 layers: upper layer of MMS (thickness 5m)
and lower layer with ten times smaller k and cv (thickness 15m)
It could well be possible that thinner layers of loose material (about 1-2m) have a larger
unfavourable effect. The consequences of various combinations of different layers have
not been studied.
It is advised that, as long as precise quantities of soil that will start to flow due to cyclic
liquefaction are still very hard to predict, acceptable probabilities of failure should be
based on thickness of susceptible soil layers, height of the slope and the slope
steepness, wave climate throughout the year from the moment the dredging works
have started and the possible (economic) damage.
Summarizing all of the above mentioned calculations, it can be stated that:

- an increase of the hydraulic permeability of the soil (k) results in a smaller pressure
build-up and therefore a reduction of the risk of cyclic liquefaction;
- a larger consolidation coefficient (cv) means faster diminishing excess pore water
pressures. Soils with a larger hydraulic permeability, like coarse sands, or a smaller
compressibility, like dense sands, therefore are far less susceptible to liquefaction due
to cyclic loading;
- although every wave cycle causes a certain compaction, soils with large relative
densities show less compressibility each wave cycle than soils with small relative
densities;
- the influence of waves fades out fast with increasing water depth, so the shallowest
part of a slope will be normative for the risk of cyclic liquefaction;
- due to the earlier mentioned large sensitivity of EPP’s between 0,5 and 1,0σ’v0 and
consequently very fast decrease of effective stresses around failure, the actual slope
steepness has little influence. Because cyclic liquefaction due to waves is caused by a
large number of waves, which all have a small contribution, slope failure is very
sensitive to the wave height; this means that steeper slopes in equal soils only can
withstand slightly higher waves. Flattening the side slopes is therefore not a quite
effective measure to increase slope stability. However, consequences of gravity flow
caused by this liquefaction are greater on a steeper slope;
- small preceding storms can increase the normative wave height of the main storm by
about 0,5 m, but this effect is only significant for loosely packed soils. Medium to
densely packed soils are not much influenced by preceding storms;
- although not mentioned before, compaction (increase of RD) that occurred during the
described storms was often in the order of 1 or 2% for loosely and medium packed
soils.
85
Measures to reduce the risk (and consequences) of cyclic liquefaction caused by waves
are:
-) adaptation of the channel geometry or alignment such that large sedimentation under
calm hydrodynamic conditions is prevented or is forced to certain predestined places,
like sediment pocket holes. This should prevent fast accretion of the side slopes,
resulting in loosely packed sediment;
-) selection of a ‘rough’ dredging method which triggers regulated liquefaction flow
slides or, if this does not occur, causes cyclic compaction of the soil, thereby increasing
safety against liquefaction;
-) construction of discrete zones with stable material like gravel;
-) (partial) (vibro-)compaction of the soil;
-) soil improvement.
All mentioned measures are aimed at improving soil properties (hydraulic permeability
k, consolidation coefficient cv and relative density RD) of the ‘danger zones’. Because of
the failure mechanism of wave-induced liquefaction increasing the relative density or
providing the side slopes with some drainage facilities are far better measures than
reducing the slope steepness.
Finally, the difference between static and cyclic wave-induced liquefaction will be
emphasized. While static liquefaction usually initiates at a location somewhere in the
inner lower part of a slope, wave-induced liquefaction takes place at a shallow location
at the top of the slope. The former is strongly related to the slope height, while the
latter is not; only the consequences of the generated turbidity current will be larger for
higher slopes, because the erosive capacities will increase when this turbidity current is
accelerated. While slope flattening is an effective measure to resist static liquefaction,
other measures are needed to deal with wave-induced liquefaction, aimed at improving
soil properties of the upper layer.
It is therefore very well possible that a certain slope is stable when considering static
liquefaction, but susceptible to wave-induced liquefaction. On the other hand, another
slope may be able to resist local storm waves, but is susceptible to static liquefaction.
Still, a large wave can of course act as a trigger mechanism.
5.3 Other loads
Already in Paragraph 2.1 earthquakes, tectonics and man-made loads, like dumping of
waste deposits, anchor forces and thrust produced by a ship propeller, were mentioned
as trigger mechanisms. None of these loads are unlikely in a navigation channel, but
they are not treated elaborately in this thesis. Some loads, like anchor or propeller
forces, are relatively small, but can however be just large enough to act as a releasing
mechanism for liquefaction flow slides. The design approach in this thesis was aimed at
avoiding slope geometries that are susceptible to such small forces under the
assumption that a certain trigger mechanism will always occur. Other loads occur at
totally irrelevant time scales, like tectonic movement.
Earthquakes however should not be neglected beforehand. Therefore some points of

special interest are mentioned briefly. Seismically induced liquefaction is similar in some
respects to the phenomenon of wave-induced liquefaction [Nataraja, 1983]. One can
however distinguish three differences:
1. ocean wave periods are longer than the periods of earthquake cycles of shaking.
Typical wave frequencies are about 0,1 Hz, whereas earthquake frequencies are in
the order of 1 Hz;
2. ocean wave storm durations (a few hours) are considerably longer than the seismic
shaking duration (about 1 minute);
3. ocean wave loadings are imposed at the surfaces of the seafloor rather than at some
lower boundary conditions as in earthquakes.
When designing the slope of a navigation channel, it depends very much on the location
whether one should take earthquakes into account. Because of the short timescale of
one year in this thesis, it does not seem relevant to consider heavy or even very heavy
earthquakes.
86
Probability of occurrence will be low and risk of human losses due to failure of the side
slopes of a navigation channel is almost negligible. In that case we’ll accept to do some
repair works.
Therefore it is advised to make calculations based on two reference earthquakes:

1) a light earthquake (4,0 on the Richter Scale) with an accelerations of 0,1 m/s2
of which a total of about 6200 take place annually on a global scale.
2) a moderate earthquake (5,5 on the Richter Scale) with an acceleration of 0,681
m/s2 of which throughout the world about 800 earthquakes take place every
year.
There are several ways to simulate earthquakes on slopes. In MStab earthquakes can
be modelled by introducing a constant horizontal or vertical acceleration (a constant
times g). This approach only accounts for the direct, elastic effect (cf wave loads). The
indirect, plastic effect asks for an approach which takes compaction and consolidation
into account. A free software package is Cyclic1D by Elgamal, Yang, Parra and Ragheb,
which can be downloaded from the internet (http://cyclic.ucsd.edu). Various
earthquakes from the past or a theoretical sinusoidal motion can be selected. A more
complete and expensive software package is for instance Liquefy Pro by CivilTech
Software which evaluates liquefaction potential and earthquake-induced settlement in
saturated soils (sand, silt and gravel).
Both programs give insight in the earthquake behaviour of the side slopes and a quick
investigation is therefore recommended in seismic areas.
In the past a lot of research has been done on earthquake-induced liquefaction,

especially in areas with frequent earthquakes like Japan and the United States of
America. Research in these countries usually is fundamentally different from the
approach of static liquefaction. Due to the frequent occurrence of earthquakes, cyclic
earthquake-induced liquefaction gets a lot of attention. At different void ratios the
steady state is determined (the bend in line C of Figure 4-13). When the void ratios of
different tests are plotted against the ‘steady state’ isotropic stresses, a logarithmic
relationship appears, see Figure 5-33 for an example. In-situ stresses and void ratios
are compared with this steady-state line to get an estimation of the susceptibility to
liquefaction. If in-situ points are above the steady-state line, there is a risk of
liquefaction.
0,9
susceptible to
0,8
void ratio e [-]
liquefaction
0,7 not susceptible

to liquefaction
0,6
0,5
0,1 1 10 100 1000 10000
effective stress [kPa]
Figure 5-33: Example of a steady-state line
From definitions can be deduced that steady state values are between dry and wet
critical densities, although this steady state is often confused with the wet critical
density, which isn’t well established abroad. An important advantage of this steady-
state line over the wet critical density is the fact that tests are done with multiple
isotropic stresses. Initial liquefaction is also often defined as the state in which either:
- the pore water pressure builds up to a value equal to the initially applied
confining pressure or
- axial strains of about 5% in amplitude occur.
87
6 Hydrodynamics in trenches and channels
6.1 Currents
When currents pass over a pipeline trench or navigation channel, there are two ways to
describe the flow velocity profiles, primarily depending on the bed level gradients. In
this thesis a flat surrounding seabed is assumed, where the flow is stationary and
friction dominated (‘wall flow’), so the channel geometry itself determines which
mathematical model of the flow velocity profiles inside the channel has to be applied.
The above mentioned bed level gradients are therefore represented by the side slopes
and the two mathematical models are:
- gradually varying flows: the flow velocity profiles can be represented sufficiently
well by logarithmic profiles;
- complicated flows: flow velocity profiles strongly deviate from the logarithmic
profiles and even flow reversal may occur on steep side slopes.
The properties of both models will be examined in the next two paragraphs.
6.1.1 Static wall flow
First, the situation of static wall flow will be considered. The flow velocities are
dominated by friction (Formula of Chézy), the velocity profile is logarithmic and there is
no flow separation. This situation is representative for slowly varying currents over
gentle slopes (approximately not steeper than 1:5-7, depending on the flow conditions
and slope height). The logarithmic velocity profile can be represented by:
ln⎛⎜ z ⎞
⎝ z 0 ⎟⎠
u ( z ) = u h ,e Eq. 6-1
ln⎛⎜ h ⎞
⎝ z 0 ⎟⎠
in which
z0 = zero-velocity level = 0,03 kr (Nikuradse, hydraulic rough bottom);
kr = equivalent bed roughness;
uh,e = equilibrium flow velocity at water surface (z=h).
This equilibrium flow velocity uh,e is related to the depth-averaged flow velocity, which is
equal to the flow velocity at about 0,4 times the water depth:
ln⎛⎜ h ⎞⎟
⎝ z0 ⎠
u h ,e = u Eq. 6-2
− 1 + ln⎜⎛ h ⎞
⎟
⎝ z0 ⎠
The bed-shear velocity (u*) is determined from the flow velocity computed at a small
height above the seabed (SUTRENCH: 0,05h), assuming a logarithmic profile in the
near-bed layer.
κ g
u*e = u≈ u Eq. 6-3
−1 + ln ⎛⎜ h ⎞⎟ C
⎝ z0 ⎠
in which
κ = constant of Von Karman ≈ 0,4
88
⎛ 12h ⎞
C = Coefficient of Chezy = 18 log⎜⎜ ⎟⎟
⎝ kr ⎠
When a current enters a channel, current velocities, directions and bed shear stresses
will change, depending on the alignment of the channel with respect to the current
direction. Subsequently a perpendicular, parallel and oblique incoming current will be
discussed.
Perpendicular current
In many cases, a current will be more or less perpendicular to the channel axis (e.g.
navigation channel perpendicular to a tidal current along the coast). If the water depth
increases, the flow will slow down, because of continuity (subscript ‘0’ means ‘at
surrounding bed’, while ‘1’ means ‘inside channel’):
h0
u1h1 = u0 h0 ⇒ u1 = u0 Eq. 6-4
h1
Parallel current
The opposite will occur with a current parallel to the channel axis. Because of the equal
water surface slopes ‘i’ inside and outside the channel, the driving force (=ρwgh∂h/∂x =
ρwghi) is the same. Because this driving force is balanced in the equilibrium situation by
the bottom friction force (ρw gv|v|/C2), the following set of equations can be obtained:
v0 v0 v0 v0 ⎫
τ b;0 = ρ w gh0i0 = ρ w g 2
⇒ i0 =⎪
C h0C02 ⎪
⎬ i0 = i1
0
Eq. 6-5
v1 v1 v1 v1 ⎪
τ b;1 = ρ w gh1i1 = ρ w g 2 ⇒ i1 =
C1 h1C12 ⎪⎭
(note: flow velocity ‘v’ is directed along y-axis, parallel to channel axis)
If the coefficient of Chézy is considered depth-dependent, the horizontal flow velocity

can be described as:
12h1
ln
h C h kr
v1 = 1 1 v0 = 1 v0 Eq. 6-6
h0 C0 h0 ln 12h0
kr
A first approximation, assuming a constant coefficient of Chézy, results in:

v1 h
= 1 Eq. 6-7
v0 h0
However, this coefficient of Chézy will increase inside the channel: at a larger depth, the
bed will be less ‘felt’, resulting in a larger coefficient of Chézy, representing a smoother
bottom. The above equation can well be approximated by a power law function like:
p
v1 ⎛ h1 ⎞
≈⎜ ⎟ Eq. 6-8
v0 ⎝ h0 ⎠
in which p is not longer ½ (Eq. 6-7) , but a constant between 0,61 and 0,65 for water
depths h0 between 5 and 20 m and bed roughness kr between 0,01 and 0,1 m. In Figure
6-1, this coefficient ‘p’ is 0,63 for h0 = 10 m and kr =0,03 m.
89
2,0
1,8
relative depth-averaged flow velocity
1,6
1,4
u1/u0 and v1/v0
1,2 approximation:
v1/vo = 0,993(h1/h0)0,627
1,0
R2 = 1,000
0,8
0,6
0,4
0,2
0,0
0,5 1 1,5 2 2,5
relative water depth expansion h1/h0
Flow perpendicular to channel axis 'u' Flow parallel to channel axis 'v'
Figure 6-1: relative flow velocities of perpendicular and parallel currents against relative
water depths
Due to the increasing flow velocity and, as will be explained later on, the sediment
transport capacity, parallel currents are theoretically capable of ‘self-cleansing’ of the
trench, but because of gravity effects, the side slopes will flatten and particles will settle
on the bottom of the channel. Under normal conditions self-cleansing will therefore not
be observed in nature.
Oblique current
Currents that are oblique to the channel axis give a more complex situation. The
perpendicular component is reduced due to continuity, the parallel component is
increased because of the equal water surface slope (or driving force) inside and outside
the channel. If both components are added as vectors, it will become clear that the
current will refract inside the channel. Because of continuity between two (refracted)
streamlines, the following relation holds:
V1 b0 h0
= Eq. 6-9
V0 b1 h1
This equation tells that if the depth ratio h1/h0 is larger than the ratio of the width
between two streamlines, b0/b1, the flow will slow down, otherwise the flow will
increase.
Roughly speaking, there are two fundamentally different methods to calculate the
current velocity inside the channel. The first method, which is applicable for large angles
α0, assumes a constant component of the current velocity parallel to the channel axis,
based on the idea that large velocity differences cannot exist over very short distances.
It can be deduced that:
2
⎛h ⎞
Method I: V1 = V0 cos α 0 + ⎜⎜ 0 ⎟⎟ sin 2 α 0
2
in which V = u2 +v2 Eq. 6-10
⎝ h1 ⎠
90
b0 trench
b0//
b1// surrounding
seabed
b0// = b1//
b1
surrounding
seabed
Figure 6-2: Schematization of refracted streamlines when crossing a trench
The second method which is applicable for small approach angles to the channel axis
assumes a constant water surface slope in the direction of the channel axis (longitudinal
component of current). The current velocity in the channel now becomes:
2
⎛ 12h1 ⎞
⎜ ln ⎟ 2
⎜ kr ⎟ ⎛ h1 ⎞ 2 ⎛h ⎞
Method II: V1 = V0 ⎜⎜ ⎟⎟ cos α 0 + ⎜⎜ 0 ⎟⎟ sin 2 α 0 Eq. 6-11
⎜ 12h0 ⎟ ⎝ h0 ⎠ ⎝ h1 ⎠
⎜ ln ⎟
⎝ kr ⎠
Method II yields higher values for the current velocity than Method I, especially for
small approach angles, but these angles actually are outside the range of Method I, see
Figure 6-3. With increasing approach angles both methods more and more coincide with
each other: they predict equal velocities in case of a perpendicular current. Approach
angles of 45-90° are best calculated by Method I. Experiments of Wallingford (1973)
and WL|Delft Hydraulics (1983) [Van de Graaff, 2003] confirmed the applicability of
Method I for large approach angles and Method II for small approach angles. The
transition between the two methods can not (yet) be described analytically and is just
bounded by the upper limit (Method II) and lower limit (Method I).
2,5
2,0
relative flow velocity V1 / V0
1,5
1,0
0,5
0,0
1,0 1,5 2,0 2,5 3,0 3,5 4,0
relative depth expansion h1 / h0
0 degrees, method I 0 degrees, method II 30 degrees, method I 30 degrees, method II

60 degrees, method I 60 degrees, method II 90 degrees, method I 90 degrees, method II
Figure 6-3: Current velocity inside channel against channel depth
91
In Jensen et al. [1999], a somewhat other approach is used. Also the different
development of the cross-channel and longitudinal component is considered. When a
current enters a channel, the depth-averaged velocity will be refracted to an equilibrium
velocity and flow angle. This refraction is caused by:
- an instantaneous decrease in the perpendicular ‘u’-component with increasing
depth
- a gradual increase in the parallel ‘v’-component
The slow development of the longitudinal velocity component is dependent on the angle
of the incoming current α0 and the relative depth increase h1/h0. Especially for large flow
angles the time to cross the channel is so small that the flow cannot adapt before
reaching the downstream side slope.
Not the assumption of equal longitudinal flow velocities or equal water surface slopes
inside and outside the channel is taken as a starting point, but equal driving forces
(δp/δy) in longitudinal direction outside the channel and inside the channel in the
equilibrium situation (fully adapted and refracted flow).
δp ⎫
τ by 0 = τ b 0 cos(α 0 ) = − h0
δ y ⎪⎪ τ beq cos(α eq ) h1
⎬ = Eq. 6-12
δ p ⎪ τ b 0 cos(α 0 ) h0
τ byeq = τ beq cos(α eq ) = −h1
δ y ⎪⎭
In the equilibrium situation with fully developed flow, the continuity equation in cross-
channel direction still holds:
V0 sin(α 0 )h0 = Veq sin(α eq )h1 Eq. 6-13
When equations 6-12 and 6-13 are combined with the expression for the bed shear
stress, according to Chézy’s Law (τ = ρwgV2/C2), the following analytical expression for
the equilibrium angle can be found, which can be substituted into Eq. 6-13 to obtain the
equilibrium flow velocity.
2
⎛ 1 ⎛ C ⎞ 2 ⎛ h ⎞3 sin 2 (α ) ⎞ 2 3
1 ⎛ C0 ⎞ ⎛ h0 ⎞ sin 2 (α 0 )
cos(α eq ) = ⎜ ⎜ ⎟ ⎜ ⎟
0 0 0
⎟ +1 − ⎜ ⎟ ⎜ ⎟ Eq. 6-14
⎜ 2 ⎝ C1 ⎠ ⎝ h1 ⎠ cos(α 0 ) ⎟ 2 ⎝ C1 ⎠ ⎝ h1 ⎠ cos(α 0 )
⎝ ⎠
The results of Eq. 6-14 are plotted for different incoming flow angles against the relative
depth expansion in Figure 6-4. It can be seen that currents with incoming flow angles
smaller than 60° eventually result in an increase of the flow velocity. For larger
incoming flow angles this depends on the relative depth expansion. A larger depth
expansion reduces the cross-channel component and increases the longitudinal
component.
92
relative equilibrium flow velocity Veq/V0 [-] 2,5
2,0
1,5
1,0
0,5
0,0
1,0 1,5 2,0 2,5 3,0 3,5 4,0
relative depth expansion h1/h0 [-]
α = 0° α = 30° α = 50° α = 60° α = 70° α = 80° α = 90°
Figure 6-4: The relative equilibrium flow velocity against incoming flow angles
The above approach yields an analytical solution, with only the assumption of the
empirical coefficient of Chézy. However, the acceleration of the longitudinal velocity
component proceeds so slow, that in most practical situations one of the earlier
mentioned phenomena will dominate:
- for large incoming flow angles the total depth-averaged flow velocity will
decrease due to continuity;
- for small incoming flow angles the total depth-averaged flow velocity will
increase;
To get an impression of the relevant time and space scales, the dimensionless timescale
for the adaptation process is presented without the full mathematical derivation [Jensen
et al., 1999]:
⎛ a v ⎞
⎜ + ⎟
h0 1 ⎜ e V0 ⎟
t= ln Eq. 6-15
V0 2 ae ⎜ a v ⎟
⎜ ⎟
⎜ e −V ⎟
⎝ 0 ⎠
in which
g cos(α 0 )
a=
C 02
gh0
e=
h1C12 1 + tan 2 (α )
The acceleration of the longitudinal component v is in fact an infinite process, but if we

are interested in the time it takes to reach, let’s say, 95% of the equilibrium value veq,
the expression becomes:
Δt = t (0,95 * v eq ) − t (v0 ) Eq. 6-16
93
Expression 6-15 is a function of longitudinal component v and related flow angle α. This
flow angle α changes throughout the flow refraction process. The upper estimate of the
timescale can be found, when angle α is taken just inside the navigation channel
(continuity taken into account (u1=u0h0/h1), but before longitudinal adaptation); a lower
estimate can be found when αeq is used. The ‘real’ timescale lies somewhere in between,
so the average of both angles will be used. The time Δt has been calculated for different
incoming flow angles (constant bed roughness kr=0,03, surrounding water depth
h0=10m). The results are very much in agreement with values reported by Jensen et al.
[1999], although they used the Colebrook-White friction formula and a depth-
independent bed roughness.
20000
18000
16000
time to reach v = 0,95veq [s]
14000
12000
10000
8000
6000
4000
2000
0
0 10 20 30 40 50 60 70 80 90
initial flow angle α0
V0 = 0,5 m/s V0 = 0,75 m/s V0 = 1,0 m/s
Figure 6-5: Time to reach 0,95veq plotted against incoming flow angle; h1/h0 = 2
In Figure 6-5 can be observed that, first of all, the longitudinal acceleration is a very
slow process and, second, that this adaptation time increases very fast for incoming
flow angles larger than 60°. When this graph is converted to the cross-channel distance
(Figure 6-6), it appears that the magnitude of the current velocity does not affect this
cross-channel distance. Therefore now, three relative depth expansions have been
plotted instead of different flow velocities. The influence of the relative depth expansion
however is minor.
5000
4500
cross-channel distance to reach v = 0,95veq [m]
4000
3500
3000
2500
2000
1500
1000
500
0
0 10 20 30 40 50 60 70 80 90
initial flow angle α0
h1/h0 = 1,5 h1/h0 =2 h1/h0 = 3
Figure 6-6: Cross-channel distance to reach 0,95veq plotted against incoming flow angle
94
These charts again demonstrate that currents with a large angle to the channel axis are
dominated by a decrease of velocity due to continuity, because the width of navigation
channels will be in the order of a few hundred meters. Currents with small incoming
flow angles, say 30°, will almost fully adapt to the equilibrium flow velocity and angle.
6.1.2 Reynolds-averaged Navier-Stokes equations
If side slopes become steeper (roughly 1:5) or one is particularly interested in the
detailed modelling of refraction of oblique currents, the full 3D flow problem needs to be
solved. The well-known computer model Delft3D developed by Delft Hydraulics solves
this problem under the shallow water assumption, which means that the vertical
momentum equation is reduced to a hydrostatic pressure equation. Sudden variations in
the bottom topography, like very steep side slopes, can still not be taken into account
correctly. Nevertheless, this model is much more capable to deal with non-stationary
problems and non-logarithmic velocity profiles. The fact that more complex velocity
profiles can be calculated does not necessarily result in a more accurate prediction of
sediment transport and morphological development, because sediment transport
formulae often are calibrated to logarithmic velocity profiles. Only when the complete
distribution of suspended sediment over the entire vertical profile can be obtained with
good resemblance to field measurements, this better description of the velocity profile
indeed yields better results in predicting sediment transport.
Nevertheless a very nice attempt to solve this problem has been made by several PhD-
students of the Institute of Hydrodynamics and Water Resources, Technical University of
Denmark [Jensen, 1999]. They solved the 3D-Reynolds-averaged Navier-Stokes
equations using a K-ε turbulence closure model (representing the transport equations
for turbulence kinetic energy K and dissipation rate ε) in a curvilinear coordinate system
(without the restriction of hydrostatic pressure), so they were able to describe the two
important flow features, when crossing a navigation channel obliquely:
- the flow will be refracted in the direction of the channel alignment, which can be
described by depth-averaged models;
- a secondary flow will develop caused by shear in the velocity profile.
Without discussing the mathematical equations some of the described, instructive

phenomena are presented. The above mentioned secondary flow is caused by non-
equilibrium between the pressure gradient and the centrifugal forces. These centrifugal
forces are unevenly distributed over the vertical velocity profile. The same phenomenon
is observed in river bends (helical motions, see River Dynamics).
With very steep side slopes of π/6:1 (about 1:1,9) the flow is represented in Figure 6-7
for incoming flow angles of 30° and 60°. Three streamlines, representing paths of three
fluid particles, are plotted: one just above the bed, one at mid-depth and one near the
water surface. Because of the formation of a separation bubble, also a fluid particle
within this separation zone is followed. Because this particle is trapped within this
bubble, it is carried downstream by the longitudinal velocity component in a corkscrew
pattern. Also it can be observed that the near-bed particles experience the largest
refraction.
95
Figure 6-7: Visualisation of refracted streamlines at three different depths and inside the
separation zone [source: Jensen, 1999]
When a current enters a navigation channel at the upstream slope, the near-bed fluid
particles experience a larger deceleration than near-surface particles due to the positive
pressure gradient on the upstream slope affecting low inertia particles more strongly
(note the logarithmic velocity profile in Figure 6-8) resulting in a larger refraction
(smaller flow angle) of the near-bed flow. When the current crosses the flat channel
bottom, this difference in flow angles is reduced due to shear in the velocity profile. On
the downstream slope, the current accelerates again (large negative δp/δx) and the
near-bed particles undergo a larger velocity increase than the near-surface particles:
now the current angle of the upper layers will be larger than the depth-averaged flow
angle.
Figure 6-8: Velocity profiles in trenches with steep side slopes
These observations are demonstrated in Figure 6-9 for two channel widths W=20h0 and
W=100h0. The variation of the current refraction starts at the side slopes and is reduced
at the channel bottom due to shear in the velocity profile. If the channel becomes wider,
the flow angle difference over the vertical will reduce to zero. Please note, firstly, the
flow angle’s tendency to decrease inside the channel due to adaptation of the
longitudinal velocity component and, secondly, the somewhat smaller flow angle (43
and 40°) after crossing the channel.
96
Figure 6-9: Flow angle development against cross-channel dimensionless distance for α0=45°,
h1/h0=2, side slopes 1:6, (a) W=20h0 and (b) W=100h0;
This variation of the current refraction over the vertical is considered as a secondary
motion on top of the depth-averaged flow. The difference between angles of bed-shear
stress and depth-averaged flow angle (Δα) may become as large as 20-30°, see Figure
6-10. In general, this difference is dependent on:
- steepness of side slopes, 1/tan(β)
- expansion in depth, h1/h0
- incoming flow angle, α
An increase of the above parameters means a larger secondary motion, with the remark
that currents with a flow angle of 80-90° however can experience separation bubbles,
but will show little secondary motion because of the small longitudinal velocity
component. Increasing the incoming flow angle, in fact, results in a steeper slope from
the point of view of the current. A steeper slope means the distance over which the
streamlines must bend will be shortened due to the larger pressure gradient δp/δx.
Figure 6-10: Difference between angles of bed shear stress and depth-averaged flow angle
for W=20h0 and side slopes 1:6
Furthermore, it can be concluded from Jensen [1999] that for channels with a width of
about 20h0 and side slopes of 1:6 an increase of the depth-averaged flow velocity only
occurs for incoming flow angles smaller than 30°. Otherwise the width is simply too
small to adjust to the longitudinal velocity component.
97
The above mentioned phenomena are typical for large-scale situations. Although not
much research has been done on very small channels, like pipeline trenches, some
things can be learned from the ‘larger-scale theory’. Currents crossing small trenches
cannot adapt to the longitudinal velocity component. If the width becomes too small and
the side slopes are very steep, the streamlines aren’t able to ‘reattach’ to the seabed,
because of the presence of a very turbulent recirculation zone. In this case the current
will be ‘blown’ over the trench and there will be hardly any sediment exchange between
the upper layer and the recirculation zone.
6.2 Waves
Waves were already mentioned as a trigger mechanism of static liquefaction flow slides
and as a cyclic load, or in fact as a reduction of strength, inducing dynamic liquefaction
flow slides. It was assumed that the largest occurring waves are depth-limited. This
means that the irregular wave field has become more complicated because the highest
waves are truncated down. On the basis of two dimensionless parameters (H/gT2 and
h/gT2), it was noticed that these large waves should be described by cnoidal wave
theory. Wave pressures at the seabed, however, could well be estimated by using linear
wave theory. Because eventually the impact of waves on sediment transport has to be
calculated, one is again bound to linear wave theory. On the other hand, the more
frequent smaller waves resemble sinusoidal waves better.
The most important phenomena that occur when a wave field passes over a channel are
shoaling, refraction and reflection, see Figure 6-11, depending on the incoming wave
angle.
Figure 6-11: Schematization of wave phenomena in channels
98
Perpendicular waves
When waves are propagating perpendicular to the channel axis, the most import effect
will be the adaptation of the wave height to the increased water depth. Most of the
times this will result in a small decrease in wave height, but that does not necessarily
occur for every wave period. The wave energy flux U will be equal inside and outside
the channel:
⎡1 kh ⎤ gT
U = Ec g = Enc = 18 ρ w gH 2 ⋅ ⎢ + ⋅ tanh kh Eq. 6-17
⎣ 2 sinh 2kh ⎥⎦ 2π
in which
E = wave energy per unit surface area
cg = wave group velocity
c = wave velocity
n = ratio between cg and c
The wave propagation velocity c will increase when the water depth increases, but the
wave group velocity can still decrease, depending on the wave period T and expansion
in water depth h1/h0. If the wave energy flux inside and outside the channel are
equated, the following expression for the ‘inversed’ shoaling factor is found:
⎛ 2k 0 h0 ⎞
tanh k 0 h0 ⎜⎜1 + ⎟⎟
KH =
H1
= ⎝ sinh 2 k 0 0 ⎠
h
Eq. 6-18
H0 ⎛ 2k1 h1 ⎞
tanh k1 h1 ⎜⎜1 + ⎟⎟
⎝ sinh 2 k 1 1 ⎠
h
This expression is plotted for different wave periods and depth expansions in Figure
6-12. One can notice that especially for small wave periods, an increase of the wave
height inside the channel is possible, but in most cases the wave height will be reduced
with 10-20%. This seeming inconsistent behaviour can be explained by the fact that
shorter waves are more looking like ‘deep’ water waves than longer waves. Shorter
waves will therefore experience only a minor velocity increase, because the depth
expansion is less felt, while the wave group velocity even decreases, resulting in a
larger wave height inside the channel.
1,2
h0 = 5; h1/h0 = 2
h0 = 5; h1/h0 = 3
1,1 h0 = 5; h1/h0 = 4
h0 = 10; h1/h0 = 1,5
h0 = 10; h1/h0 = 2
1,0
KH = H1/H0 [-]
0,9
0,8
0,7
0,6
5 6 7 8 9 10 11 12 13 14 15
wave period T [s]
Figure 6-12: ‘Inverse’ shoaling factor for different water depths and expansions against wave
period
99
At very steep side slopes wave reflection can occur as a secondary effect besides the
above mentioned “inverse” shoaling effect.
Parallel waves
Waves that are propagating parallel to the channel axis will primarily experience the
adaptation of the wave height to the increased water depth. At the side slopes wave
diffraction occurs due to the larger wave velocities inside the channel, resulting in
curved wave crests. This diffractional effects reduce wave energy at the side slopes.
Oblique waves
Waves directed oblique to the channel axis yield the most complex pattern. Waves will
experience “inverse’ shoaling and refraction effects, depending on the incoming wave
angle αw0, the wave period T, the surrounding water depth h0 and the relative depth
expansion h1/h0.
The underlying concept is conservation of wave energy flux between two wave rays:
Ub = Ec g b = Encb Eq. 6-19
At a way similar to perpendicular waves, the ratio between wave height inside and
outside the channel can be deduced:
H1 n0 c0 b0 b
= = KH 0 Eq. 6-20
H0 n1c1b1 b1
Snel’s Law is used to calculate the refracted wave angle inside the channel:
cos α w0 c0
= Eq. 6-21
cos α w1 c1
in which αw0 is the angle between the wave propagation and the channel axis. Because
the longitudinal distance between two wave rays remains constant, the following
relation of the (orthogonal) distances between two wave rays holds:
sin α w0 b0
= Eq. 6-22
sin α w1 b1
Just as with rays of light there is a certain critical incoming angle which forms the
transition between reflected waves and refracted waves. Waves entering the channel
under the critical angle will be refracted parallel to the channel axis and will therefore
not be able to cross the channel. This critical flow angle can be determined from Snel’s
Law, assuming a refracted wave ray that is completely parallel to the channel axis, so
cos(αw1)=1:
tanh k 0 h0
cos α w0;crit = Eq. 6-23
tanh k1 h1
In Figure 6-13 can be observed that this critical incoming wave angle shows a large
variation for different wave periods, water depths and relative depth expansions. Waves
with an approach angle larger than this critical wave angle will refract, others will
reflect.
100
70
Wave Refraction
60
50
αw0;critical [°]
40
30
Wave Reflection h0 = 5; h1/h0 = 2

20
h0 = 5; h1/h0 = 3
h0 = 5; h1/h0 = 4
10 h0 = 10; h1/h0 = 1,5
h0 = 10; h1/h0 = 2
0
5 6 7 8 9 10 11 12 13 14 15
wave period T [s]
Figure 6-13: Critical incoming wave angles for different water depths and relative depth
expansions
The ratio of wave heights inside and outside has been calculated for three wave periods
(T = 5, 10 and 15s) and for both the ‘navigation channel’, see Figure 6-14, and the
‘pipeline trench’, Figure 6-15. Please note the different y-axis: the left axis represents
the wave height ratio inside and outside the channel; the right axis represents the
refracted wave angle inside the channel or trench. Equal colours and markers represent
equal wave periods. If the angle of the incoming waves is only slightly larger than the
critical wave angle, the waves will seriously refract and the width between two
streamlines become so small that an increase of the wave height inside the channel can
be observed. Enlarging the wave angle always reduces the wave height ratio H1/H0, but
for a wave period of 5s, this ratio remains larger than 1. So for very short waves,
always a minor increase of the wave height occurs. Maximum wave height reduction of
up to about 10% occurs for medium to large period waves with large incoming flow
angles.
Wave height ratio and refraction angle in 'navigation channel'

1,4 90
1,3 80
Wave angle inside channel αw1 [°]
1,2 H1/H0; T = 5s 70
1,1 H1/H0; T = 10s 60
1,0 50
H1/H0 [-]
H1/H0; T = 15s
0,9 40
α(w1); T = 5s
0,8 30
α(w1); T = 10s
0,7 20
α(w1); T = 15s
0,6 10
0,5 0
0 10 20 30 40 50 60 70 80 90
Incoming wave angle αw0 [°]
101
Figure 6-14: ‘Solid’ wave height ratio and ‘dashed’ refracted wave angle for three different
wave periods in ‘navigation channel’
In case of the ‘pipeline trench’ similar behaviour is observed. Critical wave angles are
larger, although this critical angle of long period waves is less affected by the depth
reduction than the critical angle of short period waves. Wave height reductions are up to
15% for perpendicular waves as could also be concluded from Figure 6-12.
Wave ratio height and refraction angle in 'pipeline trench'

1,4 90
1,3 80
Wave angle inside channel αw1 [°]

1,2 H1/H0; T = 5s 70
1,1 H1/H0; T = 10s 60
1,0 50
H1/H0 [-]
H1/H0; T = 15s
0,9 40
α(w1); T = 5s
0,8 30
α(w1); T = 10s
0,7 20
α(w1); T = 15s
0,6 10
0,5 0
0 10 20 30 40 50 60 70 80 90
Incoming wave angle αw0 [°]
Figure 6-15: ‘Solid’ wave height ratio and ‘dashed’ refracted wave angle for three different
wave periods in ‘pipeline trench’
6.3 Combined action of currents and waves
If a current has a flow component in the wave propagation direction, i.e. if the waves
and current are not directed perpendicular to each other, the wave characteristics will
change. If the current is directed in the same direction of the waves, the wave height
will decrease and the wavelength increase. If the current has the opposite direction, the
wave height will increase and the wavelength decrease. For every current, this adapted
wave characteristics can be calculated with help of a fixed and moving coordinate
system. It is however rather cumbersome to calculate the adapted wave characteristics
for each discrete step of the tidal cycle and for every location inside the channel.
Besides, this effect will be significant for small angles between waves and current, but
almost negligible for situations with a longshore current and a cross-shore wave field,
which is often the case for channels and trenches. It was also assumed that the current
motion will be dominant over the wave motion. Therefore and for numerical-model-
considerations, this effect will be neglected from now on.
102
7 Threshold of motion on a slope

Opposite to sand borrowing pits or trenches in inland seas, offshore channels will almost
always be subjected to significant bottom shear stresses induced by currents and
waves. In case of seepage by waves (see Paragraph 5.1) the stability of the outer
grains could be determined; in case of currents and waves it is impossible to speak of a
fixed limit state; it is more useful to define a threshold of motion under various
conditions.
In this way a fast rough estimate can be made whether continuous movement of the
outer grains (sediment transport) will take place or not. If so, the situation would not
have to be catastrophic, it is just an indication that morphology has to be considered in
more detail. Instead of stability formulas of Shields and Sleath, transport formulas like
the formulae of Bijker or Van Rijn have to be considered, see Chapter 8.
7.1 Currents
7.1.1 Threshold of motion on a plane bed
In 1936 Shields developed a stability formula for particles in uniform flow (logarithmic
velocity profile) on a plane bed. With this formula the threshold of motion can be
determined. This ‘threshold’ can best be defined as ‘continuous movement of all grains’
or as the ‘start of sediment transport’. Van Rijn turned the well-known graph of Shields
into a more usable form. A mathematical approximation of this graph is presented in
Figure 7-1, in which the stability parameter ΨC is put along the vertical axis and the
dimensionless grain diameter d* along the horizontal axis:
τC u2
ΨC = = *C Eq. 7-1 and d * = d (Δg /ν 2 )1 / 3 Eq. 7-2
( ρ S − ρW ) gd Δgd
1,00
Ψc
Movement
0,10
No Movement
0,01
1 10 100 1000
2 1/3
d* = d (Δ g / ν )
Figure 7-1: Van Rijn's mathematical approximation of the Shields' curve
103
For different dimensionless parameters d* this graph can be described by the following
mathematical expressions:
−1
ΨC = 0,24d * for 1 ≤ d* ≤ 4
−0 , 65
ΨC = 0,14d * for 4 < d * ≤ 10
−0 ,1
ΨC = 0,04d * for 10 < d * ≤ 20 Eq. 7-3
ΨC = 0,013d * 20 < d * ≤ 150

0 , 29
for
ΨC = 0,055 for 150 < d * ≤ 1000
This Shields curve is normally used to find a diameter that is stable under the given
conditions. In this thesis it is the other way around. Sediment properties of the
reference soils are known, but we are interested under what flow conditions, still no
transport takes place. Therefore the hydrodynamics of Chapter 6 are used, i.e. the
formula for uniform wall flow, the formula of Chézy and the ‘smoothness’ coefficient of
Nikuradse-Colebrook for a hydraulic rough situation and with a Prandtl-Von Karman
logarithmic velocity profile:
u g
u* = Eq. 7-4
C
g ⎛ 12h ⎞
C= ln ⎜ ⎟ Eq. 7-5
κ ⎝ kr ⎠
Combining equations 7-1, 7-2, 7-4 and 7-5 yields an expression for the critical flow
velocity:
3 Δν g ⎛ 12h ⎞
ucrit = Ψ c d* ln ⎜ ⎟ Eq. 7-6
κ ⎝ kr ⎠
in which
κ = constant of Von Karman = 0,4 [-]
ν = kinematic viscosity [m2/s]
With this formula one can easily determine for which depth-averaged current velocities
transport will take place, if the sediment properties, the water depth and the bed
roughness are known. The stability parameter ΨC can be determined graphically from
Figure 7-1 or analytically from equation 7-3.
The bottom roughness is a very influential parameter. A lot of research has been done.
For a flat bed Engelund and Hansen (1973) found kr = 2 d65; Van Rijn proposed (1984,
1986) kr = 3 d90, which is about 4 to 5 d50. Lammers (1997) and Boutovski (1998)
found kr = 6 d50. Here a grain-related roughness of 5 d50 will be used.
If larger bed forms occur, this assumption is too low. A flat bed in offshore conditions
isn’t likely. In literature, values varying from 0,01 to 0,06 m often are used. Klein
[2003] got good results applying 0,03 m in a number of tests within the framework of
the Sandpit project for grain sizes varying from 100 to 300 μm in laboratory and field
experiments. As a first approximation therefore a roughness of 0,03 m is assumed
(comparable to ripples with a height of about 1 cm).
The above mentioned critical depth-averaged velocities are calculated for different grain
diameters on a plane bed, see Table 7-1. In the left column, the bed roughness is
assumed to be kr = 5 d50 (grain-dependent) and in the right column kr = 0,03 m
(constant for all grain sizes). Surprisingly, if the bed roughness isn’t dependent on the
grain diameter, small particles (d* < 4; d50 < 190 μm) are equally stable. The apparent
minimum at a particle diameter of 200 μm is only due to the mathematical
approximation (equation 7-3).
104
Table 7-1: Critical depth-averaged current velocities
ucrit [m/s]
h=5m h = 10 m h = 20 m
d50 [μm] kr = 5 d50 [m] kr = 0,03 [m] kr = 5 d50 [m] kr = 0,03 [m] kr = 5 d50 [m] kr = 0,03 [m]
50 0,423 0,259 0,447 0,283 0,470 0,307
100 0,399 0,259 0,423 0,283 0,447 0,307
200 0,372 0,257 0,396 0,281 0,419 0,304
350 0,391 0,284 0,417 0,310 0,442 0,336
500 0,407 0,307 0,435 0,335 0,463 0,363
1000 0,536 0,433 0,575 0,473 0,615 0,513
This approximation can be combined with equation 7-6 to obtain formulations for the
critical depth-averaged flow velocity. Assuming a constant bed roughness (kr =0,03 m)
a further simplification can be obtained:
3 Δνg 12h
u crit = 0,24 ln( ) ≈ 0,283 for 1 ≤ d* ≤ 4
κ kr
3 Δνg 12h 0,175
u crit = 0,14 ln( )d * ≈ 0,216d *0,175 for 4 < d * ≤ 10
κ kr
3 Δνg 12h 0, 45
u crit = 0,04 )d * ≈ 0,116d * 10 < d * ≤ 20
0 , 45
ln( for Eq. 7-7
κ kr
3 Δνg 12h 0, 645
u crit = 0,013 ≈ 0,066d * 20 < d * ≤ 150
0 , 645
ln( )d * for
κ kr
3 Δνg 12h 0,5
u crit = 0,055 )d * ≈ 0,136d * 150 < d * ≤ 1000
0,5
ln( for
κ kr
These critical depth-averaged flow velocities have been plotted both for a constant bed
roughness (kr =0,03 m) and for a grain-related bed roughness (kr = 5d50). In Figure 7-2
can clearly be observed that the critical flow velocity is constant to about d50 = 190 μm.
10,0
[m/s] crit
critical depth-averaged velocity u
Movement
1,0
No Movement
0,1
1 10 100 1000
dimensionless diameter d* [-]
Figure 7-2: Critical depth-averaged velocities for a constant bed roughness kr = 0,03 m;
water depth h = 10m
105
If the bed roughness is dependent on the particle diameter, the graph changes
somewhat and shows a minimum at d* ≈ 5, (d50 ≈ 235 μm). This is sometimes explained
by the fact that very small particles are completely situated in the viscous sub-layer,
where the bottom shear stress no longer is related to the square of the velocity.
10,0
[m/s] crit
critical depth-averaged velocity u
Movement
1,0
No Movement
0,1
1 10 100 1000
dimensionless diameter d* [-]
Figure 7-3: Critical depth-averaged velocities for grain-dependent bed roughness kr = 5d50;
water depth h =10 m
7.1.2 Threshold of motion on a sloping bed: reduction coefficient
When currents pass over a channel, the depth-averaged current velocity drops, but on
the other hand the grains on a slope are less stable, because gravity becomes an active
force, see Figure 7-4.
Figure 7-4: Forces on a slope: gravity force, flow force and counteracting friction force
106
If the slope angle β becomes larger than the angle of repose φ any load will induce
movement. Of course we are interested in slope angles smaller than the angle of
repose: β < φ. To take this reduction of strength into account while applying the Shields
formula, a reduction factor Kα,β will be deduced. This factor is dependent on the angle of
the flow force with the channel axis (α) and the slope angle (β).
The friction force (Ffriction = W cosβ tan φ) must balance the gravity force (Fg = W sinβ)
and the flow force (Fflow,α,β with a component along the channel axis (Fflow,α,β cosα) and a
component along the slope (Fflow,α,β sinα)). These vectors have to be added vectorially to
calculate the flow force.
( F flow,α , β cos α ) 2 + ( F flow,α , β sin α + W sin β ) 2 = (W cos β tan φ ) 2 Eq. 7-8
The flow force then becomes:
F flow,α , β = −W sin α sin β + W cos β tan 2 φ − cos 2 α tan 2 β Eq. 7-9
When dividing Fflow,α,β by the flow force on a normal plane bed (Fflow = Ffriction = Wtan φ),
the reduction factor Kα,β then becomes:
− sin α sin β + cos β tan 2 φ − cos 2 α tan 2 β

K α ,β = Eq. 7-10
tan φ
for β < φ and α is positive for flow on a downward slope. This rather complex formula
reduces for a plane bed (β = 0) to K = 1. For a sloping bed and a current perpendicular
to the channel axis (α = 90°) it becomes:
− sin β + cos β tan φ − sin β cos φ + cos β sin φ sin(φ − β )

K α =90D , β = = = Eq. 7-11
tan φ sin φ sin φ
This factor is also known as the Schoklitsch-factor. For a current parallel to the channel
axis (α = 0°), the reduction factor becomes:
cos β tan 2 φ − tan 2 β tan 2 β sin 2 β

K α =0D , β = = cos β 1 − = 1− Eq. 7-12
tan φ tan 2 φ sin 2 φ
The reduction factor Kα,β (Eq. 7-10, 7-11, 7-12) has to be substituted into equation 7-6
under the first square root, as it is a reduction of the stability.
In Figure 7-5 the reduction factor is plotted against slope angle β for different current
angles α. As expected, the most unfavourable situation occurs for a downward slope in
the flow direction. It can also be seen that according to this theory upward slopes can
increase the stability of a particle on a slope. In practice however one should be very
careful, because in most of the situations currents do not always approach the slope
from this favourable angle or at the same magnitude. In case of tidal flow a turn of the
current will occur. Reduction factors larger than 1 are therefore considered to be only
theoretical, but not useful in engineering practice.
To simplify calculations in offshore conditions most of the times flow directed

perpendicular to the channel will be considered. In reality, tidal currents will be more or
less perpendicular to the channel axis. This means that the lowest reduction factors (α
= 90°) are taken as a safe upper boundary. In inland navigation channels or tidal gullies
parallel flow is more likely (α = 0°) and slightly higher reduction factors can be applied.
107
2,0
1,8
α = -90°
1,6
α = -60°
1,4
Reduction factor Kα,β
1,2 α = -30°
1,0
α = 0°
0,8
α = 30°
0,6
0,4 α = 60°
0,2
α = 90°
0,0
0 5 10 15 20 25 30 35
Slope angle β
Figure 7-5: Reduction factors for different slope angles ‘β’ and flow angles ’α’
Of course, these reduction factors have to be combined with local flow velocities. When
looking at a gentle channel profile with side slopes of 1:5 (no flow separation), one may
assume that the velocity profiles stays logarithmic and because of continuity, the depth-
averaged velocity is inversely proportional to the depth. The situation at the top of the
slope will be normative in case of a uniform sloping bottom; the depth-averaged
velocity is largest here. When descending along the slope, the soil particles become
more stable.
In Figure 7-6 (perpendicular current) and Figure 7-7 (parallel current) one can easily
determine for a given flow condition the steepest possible slope, if one does not allow
transport. In this way, it is a quick aid to gain insight, when dredging a side slope, in
which situations (I) transport does not occur, (II) already occurs at the surrounding
seabed or (III) will occur when dredging a steep slope. In most offshore conditions
transport already occurs at the initial situation and dredging of a slope only stimulates
more transport. Most transport equations contain formulations for the threshold of
motion. The difference between the actual velocity and the critical velocity appears in
transport equations to the power of 3 to 5, so the determination of this critical velocity
is rather important. The morphology of trenches and channels will be discussed from
Chapter 8.
The two graphs below show some differences. With flow perpendicular to the channel
axis, the negative influence of the slope is already noticeable for relative gentle slopes,
while this influence with flow parallel to the channel axis is negligible for slopes gentler
than about 1:4. As explained above, there are no differences between small particles
(d50 < 200 μm) for a constant bed roughness. The steepest possible slope occurs of
course for the angle of internal friction and no current: tan φ = 1:1,73. Please note the
two scales on the y-axis, because of the relation u* = u√g/C.
Example: Particles with d50 = 500 μm are stable in a depth-averaged uniform flow of
0,30 m/s at a water depth of 10 m. If this flow is parallel to the slope, these particles
remain stable until the slope gets steeper than 1:3. If this flow is perpendicular to the
slope, the particles already become unstable at a slope of 1:9.
108
0,024
0,48
0,022
0,44
0,020
0,40
Depth averaged flow velocity [m/s]
0,018
0,36
Transport
0,016
Shear velocity [m/s]

0,32
0,014
0,28
0,012
0,24
0,010
0,20
d50 = 50 μm 0,008
0,16
No Transport d50 = 100 μm
0,12 0,006
d50 = 200 μm
0,08 d50 = 350 μm 0,004
d50 = 500 μm
0,04 0,002
d50 = 1000 μm
0,00 0,000
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Slope 1:x [-]
Figure 7-6: Critical slopes for different current velocities perpendicular to the channel axis;
water depth h =10 m
0,024
0,48
Transport 0,022
0,44
0,020
0,40
Depth averaged flow velocity [m/s]
0,018
0,36
0,016
Shear velocity [m/s]

0,32
0,014
0,28
0,012
0,24
0,010
0,20
0,16
No Transport d50 = 50 μm 0,008
d50 = 100 μm
0,12 0,006
d50 = 200 μm
0,08 d50 = 350 μm 0,004
d50 = 500 μm
0,04 0,002
d50 = 1000 μm
0,00 0,000
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Slope 1:x [-]
Figure 7-7: Critical slopes for different current velocities parallel to the channel axis; water
depth h = 10 m
The above figures are only applicable for a water depth of 10m. To increase the
applicability of both graphs, another auxiliary graph is presented, which is based on the
relation between the bed shear velocity and the depth-averaged velocity according to a
logarithmic velocity profile (Eq. 7-4). The left y-axis of the above figures, in fact, has to
be multiplied by the correction factor obtained from Figure 7-8 for a certain water
depth.
109
1,2
1,1
Correction Factor Ku10
1,0
0,9
0,8
0,7
0,6
0 2,5 5 7,5 10 12,5 15 17,5 20 22,5 25 27,5 30
Depth h [m]
Figure 7-8: Correction Factor for other water depths than 10m
7.1.3 Threshold profile
If currents enter a navigation channel perpendicularly, the depth-averaged velocity will

decrease because of continuity. When the side slopes are straight, this means that
grains at the toe of the slope will be more stable than grains at the top of the slope. For
grains to be equally stable at the top and at the toe of a side slope, this slope can
become steeper towards the channel bottom. First the critical slope angle βcrit of the side
slope has to be defined. With help of equations 7-1, 7-2 and 7-6 first an expression for
the reduction factor Kα,β will be deduced:
2
q crit κ2
K α ,β = Eq. 7-13
⎛ 12h ⎞
ΨC d 50 Δgh ln ⎜⎜
2
⎟⎟
2
⎝ kr ⎠
in which
qcrit = discharge at which the grains at the surrounding (plain) seabed are at
the threshold of motion, qcrit = ucrit/h0;
ucrit = critical velocity calculated from equation 7-7;
ΨC = stability parameter calculated from equation 7-3;
h = water depth somewhere along the side slope; h > h0.
Assuming a current directed perpendicular to the channel axis (Eq. 7-11) the following
formula for the critical slope angle is obtained:
⎛ ⎞
⎜ ⎟
⎜ sin φqcrit
2
κ2 ⎟
β crit = φ − arcsin⎜ ⎟ Eq. 7-14
⎜ Ψ d Δgh 2 ln 2 ⎛⎜ 12h ⎞⎟ ⎟
⎜ C 50 ⎜ k ⎟⎟
⎝ ⎝ r ⎠⎠
This expression has been plotted in Figure 7-9. This graph clearly shows that a depth
expansion from 10 to 20 m allows for much steeper slopes. A further increase in
channel depth results in a slower increase of the slope steepness, eventually resulting in
the natural slope (1/tanβ = 1/tanφ) as an upper boundary.
110
10
9
just stable side slope 1:y [-]
8
d50 = 200 μm
7
kr = 0,03 m
6 h0 = 10 m
3
5 qcrit = 2,8 m /ms
4
3
2
1
0
10 12 14 16 18 20 22 24 26 28 30
water depth h [m]
Figure 7-9: Just stable side slopes for a channel in a surrounding seabed of MSL-10m
Because of this significant increase of just stable side slopes within the limits of a
navigation channel, it seems relevant to define a channel geometry with side slopes that
are just stable (or at the threshold of motion). This particular channel geometry will be
named the ‘threshold profile’. It is assumed that the side slopes first becomes steeper
until half of the slope height and than flatter again. In this way, when straight side
slopes are dredged, the eroded volume on top of the slope is equal to the accreted
volume at the toe of the slope. The upper half of the side slope is described by equation
7-14; the lower half of the slope is point symmetric to the upper half, see Figure 7-10.
cross-channel distance [m]
-300 -250 -200 -150 -100 -50 0 50 100
-5
-6
ebb current flood current
-7
3
-8 current discharge: q = 2,8 m /ms
grain size: d50 = 200 μm
-9
initial channel width at bottom: W = 200 m
-10
depth wrt MSL [m]
-11
-12
-13 erosion area erosion area
'threshold' profile during flood
-14 during ebb
-15
-16 initial profile with
-17 accretion area accretion area side slopes 1:3,0
during ebb during flood
-18 'threshold' profile
-19 after some time
-20
-21
Figure 7-10: ‘Threshold’ profile that is just completely stable during ebb and flood current
The ‘blue’ geometry is the initial channel geometry with side slopes of 1:3,0; the ‘red’
geometry will develop under very mild hydrodynamic conditions, not exceeding the
critical discharge: the top of the side slope will erode and roll down, until the threshold
profile is reached. In case of a flood current only the upper half of the upstream slope
will erode, because all other grains are more stable (milder slope or favourable influence
of gravity).
This ‘threshold profile’ is of course dependent on water depth and slope height, but not
on grain size, at least when the bed-form related bed roughness is used. After all, this
profile is defined at the critical discharge for a certain grain size.
111
For grain sizes smaller than 200 μm, this critical discharge is equal, see Figure 7-2 or
Figure 7-6. For larger grain sizes, the critical discharge increases to 3,07, 3,35 and 4,73
m3/ms for grain diameters of respectively 350, 500 and 1000μm for h0 = 10 m.
Because sediment transport is related to the 3rd to 5th power of the difference of the
actual flow velocity and the critical flow velocity, the difference between the actual
geometry and the ‘threshold geometry’ could also be a good measure to estimate the
development of the channel geometry. Because of the larger critical velocities of larger
grains, the morphological development will be much slower for larger grains.
As a final remark it is emphasized that, although critical flow velocities are rather small
inside the channel, there is some uncertainty of the resemblance of the velocity profile
to the logarithmic profile, especially at the steeper parts of the side slopes.
Nevertheless, the ‘threshold profile’ has a very natural appearance.
7.2 Waves
Not only currents induce bed shear stresses, but also waves appear to be very effective
stirring up sediment particles. Considering the water depths in this thesis varying from
5 to 20m, this subject is positioned in the ‘transitional water depth’-range judging by
Table 7-2. The smallest waves that satisfy the shallow water assumption and the largest
waves that satisfy the deep water assumption are calculated for three water depths.
Common waves will have wave periods between 5 and 14 s and wavelengths between
40 and 100 m, so are situated in the ‘transitional water depth’ range.
Table 7-2: Smallest waves satisfying ‘shallow water assumption’ and biggest waves satisfying
‘deep water assumption’
shallow water deep water

h 5 10 20 5 10 20 m
L >100 >200 >400 <10 <20 <40 m
Llimit 100 200 400 10 20 40 m
c 7,1 10,0 14,1 4,0 5,6 8,0 m/s
T 14,1 20,0 28,3 2,5 3,5 5,0 s
Although waves that are entering shallow water become more and more non-sinusoidal,
they still can quite reasonably be described by linear wave theory (see Paragraph 5.1).
The horizontal orbital velocity ‘u’ and excursion ‘a’ are given by:
ωH cosh k (h + z )
u= sin(ωt − kx) Eq. 7-15
2 sinh kh
H cosh k (h + z )
a=− cos(ωt − kx) Eq. 7-16
2 sinh kh
On top of the wave boundary layer at a small distance δ from the bed the amplitude of
the horizontal orbital velocity û 0 and the excursion â0 can be defined as:
ωH 1
uˆ 0 = Eq. 7-17
2 sinh kh
H 1
aˆ 0 = Eq. 7-18
2 sinh kh
112
The thickness of this wave boundary layer ‘δ’ is defined for turbulent flow by Fredsøe:
−0 , 25
⎛ aˆ ⎞
0 , 75 0 , 75
⎛ H ⎞ ⎛ H ⎞
δ = 0,15⎜⎜ 0 ⎟⎟ aˆ 0 = 0,15k r0, 25 ⎜ ⎟ ≈ 0,037⎜ ⎟ Eq. 7-19
⎝ kr ⎠ ⎝ 2 sinh kh ⎠ ⎝ sinh kh ⎠
for a wave-related bottom roughness kr of 0,03 m. As can be expected, the boundary

layer thickness increases with decreasing water depth, increasing wave height and
increasing wave period. A longer period means more time for the boundary layer to
grow. Jonsson found in 1966 the following relation for the amplitude of the wave-related
bed-shear stress:
τˆ w = 0,5ρf w uˆ 02 Eq. 7-20
in which the friction factor fw can be described as:
−0 ,194
⎡ aˆ ⎤
−5 , 977 + 5, 213 ⎢ 0 ⎥ ⎡ aˆ 0 ⎤
fw = e ⎣ kr ⎦
if ⎢ ⎥ ≥ 1,59
⎣ kr ⎦
⎡ aˆ 0 ⎤
f w = 0,30 if ⎢ ⎥ < 1,59 Eq. 7-21
⎣ kr ⎦
Now the situation becomes a bit more complicated. An increase of the wave period
means a larger wavelength and therefore a larger horizontal excursion a0 and boundary
layer thickness ‘δ’, which results in a smaller friction factor ‘fw’. On the other hand, the
horizontal velocity will increase, although slower and slower. As a result, if the wave
period increases, the shear stress will ‘decreasingly’ increase for larger water depths
(h= 15; 20m) and will decrease for small water depths (h= 5m), see Figure 7-11. The
apparently inconsistent phenomenon of larger bed-shear stresses in deeper water is
caused by the assumption of depth-limited waves (H=0,5h): in deeper water the
presented wave is higher in amplitude, larger in length and longer in period.
70
Maximum wave-induced bed-shear stress τw
60
50
40
[N/m2]
30
20 h=5m
h = 10 m
h = 15 m
10
h = 20 m
0
5 6 7 8 9 10 11 12 13 14 15
Wave period T [s]
Figure 7-11: Wave-induced bed-shear stresses for depth-limited waves against wave period
Now the bed shear stress is known, the threshold of motion for a certain grain diameter
has to be defined. Because of the wave boundary layer development that is completely
different form the situation with currents, the Shields’ curve was modified by Sleath to
represent stability of loose grains in non-breaking waves.
113
In the same way Van Rijn expressed the Shields’ Curve into mathematical formulations,
Sleath’s Curve can be formulated:
ΨC ;w = 0,1231d *−0, 6405 for 1 ≤ d* ≤ 20

ΨC ;w = 0,018 for 20 ≤ d * ≤ 30 Eq. 7-22
Ψ C ;w = 0, 0011d*0,8234 for 30 ≤ d * ≤ 120

ΨC ;w = 0,055 for 120 ≤ d * ≤ 1000
The above formulations are plotted in Figure 7-12. The Shields’ Curve for uniform flow
(dashed line) is shown as a reference.
0,1
Movement
Ψc;w
No Movement
Sleath's Curve
Shields' Curve
0,01
1 10 100 1000
2 1/3
dimensionless diameter d* = d (Δ g / ν )
Figure 7-12: Mathematical approximation of Sleath’s Curve (solid line) and Shields’ Curve
(dashed line);
Sleath used the same stability criterion as Shields:

τ C ;w
ΨC ;w = Eq. 7-23
( ρ s − ρ w ) gd
So for the critical wave-induced bed shear stress is obtained:
τ C ;w = ( ρ s − ρ w ) gdΨC ;w Eq. 7-24
If this expression is equated with Jonsson’s expression 7-20, the following relation for
the maximum amplitude of the orbital velocity at the bed is found:
2Δgd50 Ψ C ;w
uˆ0 = Eq. 7-25
fw
Combining this expression with equation 7-17 yields the formula for the critical wave
height:
114
2 2 sinh kh Δgd50 Ψ C ;w
H crit = Eq. 7-26
ω fw
This equation can only be solved by iteration, because the wave friction factor fw is
dependent on the wave height. Because the grain sizes of the reference soils are
maximum 1000 μm (d* = 21,24), they are almost completely situated at the ‘first’
branch (d ≤ 940 μm) of the mathematical approximation. This means that the part
“d50ΨC;w” in equation 7-26 can be substituted:
d50 Ψ C ;w = 2, 083*10−4 d500,3595 for d50≤ 940 μm Eq. 7-27
If also the following values are used for Δ=1,65, g=10 m/s2, ν=1,3*10-6 m2/s, equation
7-26 can be further simplified to:
T sinh kh
H crit = 2, 639*10−2 d500,18 for d50≤ 940 μm Eq. 7-28
fw
which is only dependent on the wave characteristics (T and k are related to each other),
the grain diameter, the bed roughness and the water depth.
Again the difference between bed-related or grain-related bed roughness can be made.
The critical wave height appears to be very sensitive to the bed roughness. To illustrate
this behaviour, iIn Figure 7-13 the critical wave heights are plotted for the reference
case “navigation channel”, based on bed-related roughness (kr = 0,03). The solid lines
represent the surrounding seabed and the dashed lines the channel bottom. Especially
the short period waves are less felt on the channel bottom.
1,0
0,8
Critical wave height Hcrit [m]
0,6
0,4
0,2
0,0
5 6 7 8 9 10 11 12 13 14 15
Wave period T [s]
d = 50 μm; h = 10m d = 200 μm; h = 10m d = 500 μm; h = 10m d = 1000 μm; h = 10m
Figure 7-13: Critical wave heights for different wave periods at surrounding seabed and at the
channel bottom, based on bed-related roughness (kr = 0,03m)
When the seabed is rough, very small waves of only a few decimetres are already
capable of stirring up some sediment. The same procedure for a grain-related bed
roughness (kr = 5d50) yields larger waves, see Figure 7-14. In both graphs, the
difference between the surrounding seabed and the channel bottom is clear.
115
3,0
2,5
Critical wave height Hcrit [m]
2,0
1,5
1,0
0,5
0,0
5 6 7 8 9 10 11 12 13 14 15
Wave period T [s]
Figure 7-14: Critical wave heights for different wave periods at surrounding seabed and at the
channel bottom, based on grain-related roughness (kr = 5d50)
Although there is no net transport over a wave cycle, it is very obvious that waves
combined with a current are very effective in reducing the threshold of motion and
therefore increasing sediment transport.
Although the action of currents and waves has been treated separately in the previous
paragraphs, in most cases both forces will be present. Bijker (1967) stated that, when
the current is dominant over waves, the influence of these waves can be taken into
account by adding the current and orbital velocity vectorially.
Two problems that occur are:

- at which depth should both forces be added;
- the current velocity is assumed stationary, whereas the wave motion is assumed
to be sinusoidal.
Bijker added both current- and wave-induced shear stress at a height zt (=ez0 in which
z0 is the zero-velocity level) and integrated the resultant over a full wave cycle. Without
giving the complete derivation, an expression is presented that is not longer dependent
on the wave direction:
1 ⎡ 1 ⎛ uˆ 0 ⎞ 2 ⎤
τ cw = τ c + τ w = τ c ⎢1 + ⎜ ξ ⎟ ⎥
ˆ Eq. 7-29
2 ⎣⎢ 2 ⎝ V ⎠ ⎦⎥
in which:
fw
ξ =C Eq. 7-30
2g
116
One should realize that this bed shear stress is a time-averaged value; during a wave
cycle larger stresses occur. However, when considering sediment transport (see Chapter
8), this time-averaged approach is legitimate.
Four different hydrodynamic situations have been calculated for the navigation channel:
q = 5 m3/ms; H = 1m; T = 8s;

q = 5 m3/ms; H = 2m; T = 8s;
q = 10 m3/ms; H = 1m; T = 8s;
q = 10 m3/ms; H = 2m; T = 8s;
The resulting bed shear stresses have been plotted to illustrate the decaying behaviour
with increasing depth, see Figure 7-15.
It appears that the wave-induced bed shear stress at the channel bottom is roughly
40% of the surrounding-bed-value. The larger water depth reduces the wave impact,
but also results in a larger wavelength. The net effect, however, is still a decrease of the
horizontal excursion a0 and horizontal particle velocity u0. Because the excursion is
reduced inside the channel, the wave friction factor fw increases. The overall effect on
the bed shear stress is a reduction inside the channel of 60%, whereas the water depth
has doubled.
The current-induced bed shear stress at the channel bottom is about 20% of the
surrounding-bed-value. This reduction is mainly caused by the halved current velocity,
which would have resulted in a four times smaller bed shear stress. Because also the
Chézy-coefficient increases (so friction decreases) a value of roughly 20% is obatined.
The combined response of current and waves to the expansion in depth is a reduction of
the bed shear stress up to 70%, see Figure 7-15. This calculation example clearly shows
that the stability of particles inside the trench can be much larger. Because the
sediment transport capacity is related to this bed shear stress, it is therefore obvious
that there will be a large gradient in sediment transport over the channel.
8
q = 5 m3/ms; H = 1m q = 10 m3/ms; H = 1m
7
q = 5 m3/ms; H = 2m q = 10 m3/ms; H = 2m
6
τcw [N/m2]
0
10 11 12 13 14 15 16 17 18 19 20
water depth h [m]
Figure 7-15: Bed shear stress due to the combined action of current and waves against water
depth; reference case “navigation channel”
117
8 Sediment transport in trenches and

channels
8.1 Choice of sediment transport formula
Before describing the characteristics of sediment transport in trenches and channels and
on sloping bottoms in particular, the pros and cons of a number of sediment transport
formulae will be balanced against each other, so the appropriate sediment transport
theory will be used. An extensive comparison of five well-known sediment transport
formulae by Camenen en Larroudé [2003] gave insight in the applicability of each
formula. The formulae under investigation were:
- Bijker total load formula [1968] which was developed in a river environment and
adapted to a coastal environment. It distinguishes between bed load and suspended
transport; waves are taken into account via an increase of the bottom shear stress;
- Bailard formula [1981] which is based on a formulation of energy for sediment
transport caused by waves;
- Van Rijn total load formula [1984, 1989] expresses total sediment also as bed load
and suspended load transport;
- Dibajnia and Watanabe total load formula [1992], which accounts for the
instantaneous velocity due to wave and current interaction and the resulting sediment
movement;
- Ribberink bed load formula [1998], a quasi-steady model of sediment transport using
instantaneous shear stresses.
The main conclusion of this research with regard to this thesis was that the Van Rijn
total load formula yielded reasonable results (which is good compared to the others) in
situations with prevailing currents, outside the breaker zone. However, Van Rijn is weak
at predicting sediment transport with prevailing waves, because in his approach waves
only increase the bottom shear stress and do not act as a transport mechanism. The
direction of the sediment transport will therefore always will be that of the current. As
long as the current velocity will be greater than the (oscillating) wave velocity, this error
remains limited.
Wave transport due to wave asymmetry isn’t taken into account, although waves in
coastal zones are substantially non-sinusoidal. To describe these waves well, 2nd order-
Stokes theory or even better cnoidal theory (see Chapter 5) should be used.
The important phase-lag-effect (the suspended sediment in the vertical lags behind on
the present hydrodynamic conditions), which becomes more significant with finer
sediment, causes large differences between calculated and observed sediment
transport, especially in tidal conditions.
The Van Rijn formula asks for long computation times compared to the other formulae
because the integral of the suspended load over the depth needs to be computed.
Taking all of the above into account it may be concluded that the Van Rijn formula is
preferable to the other sediment formulae for the conditions in this thesis (current-
induced transport outside the breaker zone, both bed-load and suspended transport).
Therefore Van Rijn’s line of thought will be used throughout this thesis.
8.2 Currents only
Backfilling of trenches will occur due to sediment transport, when the sediment capacity
is less within the channel than outside, and due to gravity effects, when a bed load
particle starts to move. The backfilling of dredged channels and trenches is predicted
successfully only if a detailed description of the flow and sediment transport process is
applied.
118
In the previous chapters, the hydrodynamics in channels and threshold of motion on

slopes were studied. Some basic assumptions had to be done regarding the current
velocity profile (logarithmic profiles).
If a certain channel or trench is exposed to such ‘serious hydrodynamic’ conditions, it is

not sufficient to consider the threshold of motion, but transport equations come into
play.
The natural backfilling mechanisms are partly levelling of side slopes due to bed load
movement and partly settling of suspended sediment carried across the channel by
currents. It is important to distinguish between the combined wave-current motion
offshore in the non-breaking area and the primarily wave-induced longshore sediment
transport. In the latter case, the transport will be concentrated on the breaker bars,
whereas in the first case the sediment transport will be more equally distributed and can
be better represented by a 2-dimensional approach. As was already mentioned, the
emphasis of this thesis will be on the non-breaking regime.
8.2.1 Types of transport
Transport of sediment can be split up into two different kinds of transport: bed load
transport and suspended load transport. The bed load transport will immediately be
affected by a change of the bed form, while the suspended load, which is completely
surrounded by water, needs some time to adapt. To contrast bed load with suspended
load one needs to define an artificial bed layer in which the bed load transport takes
place, because in reality such a sharp boundary does not exist. One usually
distinguishes three types of particle motion of which the first two types often are
considered as bed load transport (Van Rijn, 1984a):
I. rolling and sliding motion or combined
II. saltation motion (i.e. jumping over a maximum distance of a few particle
diameters)
III. suspended particle motion
In this bed-load layer the particle motion is gravity dominated and turbulent forces are
of little significance. According to Bagnold, particles come into suspension if the bed
shear velocity u* exceeds the particle fall velocity ws. For the reference soils, this means
that some sediment primarily is transported as bed load, like the coarse sands. Very
fine sediments (MSI), on the other hand, will always come into suspension as soon as
the critical bed shear velocity is exceeded. In between are the medium sands, which
sometimes just are transported in the form of bed load and sometimes also in the form
of suspended load, depending on the conditions. In Table 8-1 an example is presented
for a uniform flow in absence of waves to get an idea. Please note the difference
between the surrounding bed and the trench bottom. It is needless to say that in
presence of waves transport capacities increase considerably.
Table 8-1: Types of transport according to Bagnold for reference case ‘Navigation Channel’
Navigation Channel (h0=10m; h1=20m) u (depth averaged) [m/s] u (depth averaged) [m/s]
0,5 1
u*surrounding bed u*channel bottom u*surrounding bed u*channel bottom
soil type d50 [μm] ws [m/s] u*crit [m/s] 0,0245 0,0109 0,0489 0,0218
MSI 50 0,0029 0,0137 suspended no transport suspended suspended
MFS 200 0,0252 0,0135 bed load no transport suspended bed load
MMS 500 0,0663 0,0161 bed load no transport bed load bed load
MCS 1000 0,1084 0,0228 bed load no transport bed load no transport
Van Rijn [1984b] considered the Bagnold-criterion “as an upper limit at which a
concentration profile starts to develop” and suggested an intermediate stage “at which
locally turbulent bursts of sediment particles are lifted from the bed into suspension”.
This intermediate stage can for the grain sizes under consideration be defined as:
119
4 ws
u*transition = for 1 ≤ d * ≤ 10 Eq. 8-1
d*
Expressed as the depth-averaged current velocity, the above equation changes to:
4 ws C
u= Eq. 8-2
d* g
For the reference soils MSI, MFS and MMS for which this expression is valid, the depth-
averaged current velocities become:d50= 50, 200, 500μm, u = 0,22, 0,48 and 0,51m/s.
For the same situations of Table 8-1, the type of sediment transport has been
determined according to Van Rijn’s criterion. It can be observed that at the surrounding
seabed also some suspended transport can occur for MFS in a depth-averaged current
velocity of 0,5m/s and for MMS in a current velocity of 1m/s.
Table 8-2: Types of transport according to Van Rijn for reference case ‘Navigation Channel’
Navigation Channel (h0=10m; h1=20m) u (depth averaged) [m/s] u (depth averaged) [m/s]
0,5 1
soil type d50 d* u*crit u*transition u*surrounding bed u*channel bottom u*surrounding bed u*channel bottom
[μm] [-] [m/s] [m/s] 0,0245 0,0109 0,0489 0,0218
MSI 50 1,07 0,0137 0,0109 suspended no transport suspended suspended
MFS 200 4,27 0,0135 0,0236 suspended no transport suspended bed load
MMS 500 10,69 0,0161 0,0248 bed load no transport suspended bed load
When looking at slope development, this division in bed load and suspended transport
will turn out to be important. Now, both types of transport are described with special
attention to the influence of a sloping bottom, starting with bed load transport.
8.2.2 Bed load transport
Van Rijn describes the bed load transport due to currents as follows:
1, 5
⎡τ 'b ,c ⎤ ⎡τ 'b ,c −τ b ,cr ⎤
0,5
qb = 0,25ρ s d 50 d −0,3
⎢ ⎥ ⎢ ⎥ Eq. 8-3
⎣ ρw ⎦ ⎢⎣ τ b ,cr ⎥⎦
*
in which
qb = bed-load transport; [kg/ms]
τ’b,c = grain-related bed-shear stress due to current; [N/m2]
τ’b,crit = critical grain-related bed-shear stress; [N/m2]
The grain-related bed-shear stress can be deduced from the efficiency factor of the
current and the bed-shear stress as follows:
τ 'b ,c = μ cτ b ,c Eq. 8-4
in which
τb,c = critical bed-shear stress according to Shields;
μc = efficiency factor current, described by:
f 'c 0,24 log −2 (12h / 3d 90 )

μc = = Eq. 8-5
fc 0,24 log − 2 (12h / k r ,c )
120
Influence of sloping bottom on bed load transport
The influence of a sloping bottom was already proved to influence the critical bed shear
stress, see Paragraph 7.1.2. Downward sloping bottoms reduced the threshold of
motion, hence Kαβ was smaller than 1; upward sloping bottoms were favourably
influenced by gravity, so Kαβ > 1. Also, the bed load transport rates are affected by a
sloping bottom and should be multiplied by a slope correction factor. This so-called
Bagnold-factor is larger than 1 for downward sloping bottoms (β positive) and smaller
than 1 for upward sloping bottoms ((β negative) and can be described for the current
component directed perpendicular to the channel axis as follows:
tan φ
Kβ2 = Eq. 8-6
tan φ − tan β
So, sediment particles located at a downward sloping bottom not only are more easily
picked up by a current, but also more easily transported as bed-load transport.
Suspended sediment transport has no contact with the seabed and will not be affected
by a sloping bottom.
8.2.3 Dynamic Equilibrium Slopes
In Paragraph 7.1.3 a threshold profile was defined. It was concluded that this profile is
independent on grain size, because the critical discharge at the surrounding seabed was
used. Although the values of these critical discharges differed for different grain sizes,
the shape of the threshold profile was the same. According to the same line of thought,
the question arises whether an ‘equal sediment transport-profile’ can be defined. If such
a profile exists, this means that the entire cross-section of the channel is in equilibrium:
just as much sediment is transported at the surrounding seabed, at both side slopes
and at the channel bottom. In this situation no morphological changes occur, at least as
long as the hydrodynamic conditions remain the same.
To come to such a profile, one should first realise that it can only be deduced for bed-
load transport, which requires sufficient large particle sizes and sufficient small bed
shear velocities, see Table 8-1 and Table 8-2. Suspended transport is not taken into
consideration, because this type of transport isn’t able to adapt instantaneously to the
changing seabed. Furthermore, suspended transport is not just dependent on bed load
transport, but also on the height of the water column, as will be explained later on. This
implies that although the bed load transport at two different depths (and different
sloping sea beds) can be the same, but the suspended transport will differ due to a
different transport capacity at different water depths. So, the first assumption is that
only bed-load transport is allowed and therefore this calculation will be executed for a
grain diameter of 500 μm.
The second assumption contains logarithmic velocity profiles across the channel. The
influence of a sloping bottom is only represented by two slope correction factors:
Schoklitsch-factor Kαβ (threshold of motion) and Bagnold-factor Kβ2 (bed-load
transport).
Furthermore, a constant unidirectional current (u = 1m/s) and a water depth of 10 m is

assumed. At this surrounding seabed a certain equilibrium bed load transport can be
calculated according to formula 8-3:
1, 5
⎡τ ' −τ ⎤
qb = 0,005 τ 'b ,c ⎢ b,c b ,cr ⎥ Eq. 8-7
⎣⎢ τ b ,cr ⎦⎥
At a water depth of 10m, this results in qb;h0 = 54,0 g/ms. When a current enters a
navigation channel at the upstream side slope, this bed load transport can only remain
121
constant if the negative influence of the increasing water depth is counterbalanced by

the positive influence of the downward sloping bottom. This steepening effect is
demonstrated in Figure 8-1 for four discharges. Due to the additional Bagnold-factor,
the ‘equilibrium’ side slopes become somewhat smaller (=flatter) for larger discharges.
At infinite water depth, of course the angle of internal friction is reached.
side slope 1:x with constant Sb [-]
0 1 2 3 4 5 6 7 8 9 10
10
d50 = 500 μm upstream slope
11 d90 = 750 μm
12 kr = 0,03 m
h0 = 10 m I: Sb = 0 g/ms
water depth h [m]
13 φ = 30° II: Sb = 0,6 g/ms

2
14 qcrit = 3,35 m /s III: Sb = 12,7 g/ms
15 IV: Sb;h0;eq = 54,0 g/ms
IV: Sb;h1;eq = 0,2 g/ms
16
17
18
19 downstream slope
20
I: threshold of motion II: q = 5 m2/s III: q = 7,5 m2/s
IV: q = 10 m2/s; upstream IV: q = 10 m2/s; downstream
Figure 8-1: Side slopes, for which the bed-load transport is constant, against water depth
with perpendicular currents of 3,35 m2/s (threshold of motion on plane bed), 5,0 m2/s, 7,5
m2/s and 10 m2/s
As the water depth is increasing, the upstream side slope becomes steeper and steeper,
until the channel bottom is reached. The bed load transport inevitably has to adapt to
the local conditions. The equilibrium bed load transport at the channel bottom is: qb;h1 =
0,203 g/ms. This abrupt change in sediment transport would cause an enormous
accretion at the channel bottom near the toe of the upstream slope and this uncoupled
approach of a constant bed load transport at the surrounding bed and upstream side
slope and a constant bed load transport at the channel bottom is therefore not very
realistic. One would observe a gradually flattening lower half of the upstream side slope.
A true equilibrium profile cannot exist, because the lower part of the upstream side
slope ‘feels’ the channel bottom. As siltation proceeds, a larger and larger part of the
upstream slope starts feeling the bottom and the above described slope steepening
process becomes less pronounced.
As the current flows across the channel bottom, it becomes totally adapted to the
equilibrium bed load transport at the channel bottom, which is in fact very small. When
the current arrives at the downstream side slope an inverse process occurs: due to the
decreasing water depth and the corresponding increasing bed shear stress, the
sediment particles need a stabilizing gravity force to maintain the amount of bed load
transport. The downstream side slope becomes steeper, until the angle of repose is
reached. The side slope cannot respond to the increasing erosive capacity of the current
by increasing the gravity component anymore and erosion of the upper part of the side
slope is inevitable. This effect will extend over a very large distance downstream of the
trench, so this approach seems only valid for the upper part of the upstream slope and
the lower part of the downstream slope. The lower part of the upstream slope will be
subjected to large sedimentation and the upper part of the downstream slope to large
erosion, until the navigation channel has completely silted up and the initial, flat
equilibrium seabed is restored.
Before this discussion is proceeded, one should be aware that no influence of waves,
gravity effects or velocity profiles deviating from the logarithmic profile are taken into
account.
122
This brief explanation can be described in symbols, starting from an initial channel
geometry with side slopes βslope. The subscripts ‘sb’, ‘cb’, ‘eq’, ‘up’ and ‘down’
respectively refer to the surrounding seabed, channel bottom, equilibrium, upstream
and downstream slope. Please note that βup;eq and βdown;eq are functions of water depth.
In direction of the current, the following processes can be observed, starting from initial
straight side slopes:
⇒ top of upstream slope: ⇒ βslope > βup;eq

⇒ Sb;slope > Sb;sb;eq
⇒ erosion
⇒ βslope ↓ → βup;eq
⇒ slope flattening until βup;eq is reached
⇒ somewhere along upstream slope: ⇒ βslope = βup;eq

⇒ Sb;slope = Sb;sb;eq
⇒ no bed level or slope steepness change
⇒ toe of upstream slope: ⇒ βslope < βup;eq

⇒ Sb;slope < Sb;sb;eq
⇒ sedimentation
⇒ βslope ↓ → βup;eq
⇒ slope steepening until βup;eq is reached
⇒ channel bottom: ⇒ βslope = 0 < βup;eq

⇒ Sb;sb;eq > Sb;cb;eq
⇒ sedimentation until Sb;cb;eq is reached
⇒ toe of downstream slope: ⇒ βslope > βdown;eq

⇒ particles experience positive influence of
gravity
⇒ Sb;slope < Sb;cb;eq
⇒ sedimentation
⇒ βslope ↓ → βdown;eq
⇒ slope flattening until βdown;eq is reached
⇒ somewhere along downstream slope: ⇒ βslope = βdown;eq

⇒ Sb;slope = Sb;cb;eq
⇒ no bed level change
⇒ somewhere further along downstream slope:

⇒ βslope < βdown;eq ≤ φ
⇒ Sb;slope > Sb;cb;eq
⇒ erosion
⇒ βslope ↑ → βdown;eq
⇒ slope steepening until βdown;eq is reached
Eventually the required slope angle βdown;eq to maintain equal bed load transport can
become larger than φ. It is however not very likely that slopes steeper than the angle of
internal friction can sustain. Once put into motion, the particles will rain down along the
slope.
⇒ somewhere along downstream slope where βdown;eq ≥ φ

⇒ βslope ≤ φ ≤ βdown;eq
⇒ Sb;slope >> Sb;cb;eq
⇒ continuous erosion
⇒ βslope ↑
Once again, this is a very schematized approach, but it shows that slopes tend to a
certain slope steepness to minimize the gradients in sediment transport. After the side
slopes have reached their dynamic equilibrium, the sedimentation of the entire channel
still proceeds. The side slopes start feeling the channel bottom more and more and
123
finally the side slopes start feeling each other. In fact, one can not longer speak of side
slopes and a channel bottom, it is just a pit in the seabed.
This idea of dynamic equilibrium side slopes based on constant bed-load transport
shows some realistic ‘smooth’ slopes, but will initial straight slopes in reality develop to
these equilibrium slopes? The simulations in SUTRENCH in Chapter 10 will resolve these
questions.
8.2.4 Suspended transport
Once sediment particles are mixed over the vertical, one speaks of suspended sediment
transport. The suspended sediment transport is the product of the velocity distribution
and the concentration distribution, integrated from the reference level a to the water
depth h.
h
qb = ∫ ucdz Eq. 8-8
a
in which the velocity profile is described by the logarithmic Von Karman-profile and Van
Rijn describes the concentration profile at water depths larger than the reference height
as follows:
dc (1 − c) 5 cws
=− Eq. 8-9
dz ε s ,c
in which
c = concentration
ws = particle fall velocity, described in Paragraph 3.1.2
εs,c = sediment mixing coefficient for current only, described in Paragraph 8.4
As a boundary condition the reference concentration ‘ca’ at a height ‘a’ above the bed
will be used. In fact, the bed load transport is the product of the thickness of the bed
load layer (or saltation height) ‘δ’, the velocity of the bed-load particles and the bed
load concentration. To avoid large errors in the concentration profile, the effective
reference concentration at a somewhat larger height (reference height) above the bed is
used, although no physical basis exists.
q b = cb u b δ = c a u a a Eq. 8-10
This effective reference concentration can be expressed as:
1, 5
q d ⎡τ ' −τ ⎤
ca = b = 0,015ρ s d *−0,3 50 ⎢ b ,w b ,cr ⎥ Eq. 8-11
ua a a ⎢⎣ τ b ,cr ⎥⎦
In principle this bed load layer is a layer just above the bed in which the mixing is so
small that it cannot affect the particles, so it is impossible for them to get into
suspension. According to Van Rijn, this thickness of the bed load layer (‘a’) is the
maximum of the current-related and wave-related roughness, which are considered to
be equal throughout this thesis.
124
Although waves can cause a net transport sediment transport, they will be most
effective in stirring up sediment. In combination with a current, the sediment transport
will heavily increase. It was already mentioned that in this thesis currents and waves
act as a sediment pick-up mechanism and that only currents act as a convection
mechanism, see Figure 8-2.
Figure 8-2: Sediment transport mechanisms
8.3.1 Bed load transport
The formula of the bed load transport changes somewhat if waves are superimposed on
a current. A calibration factor is added and the (time-averaged) bed shear stress
represents the influence of both waves and current, according to Paragraph 7.3.
1, 5
⎡τ 'b ,cw ⎤ ⎡τ 'b ,cw −τ b ,cr ⎤
0,5
qb = 0,25α H ρ s d 50 d −0 , 3
⎢ ⎥ ⎢ ⎥ Eq. 8-12
⎣ ρw ⎦ ⎣⎢ τ b ,cr
*
⎦⎥
in which:
αH = calibration factor (=1-(Hs/h)0,5;

τ’b,cw = grain-related time-averaged bed-shear stress due to current and waves.
8.3.2 Suspended transport
The formula of suspended transport (Eq. 8-8) still holds, but the velocity and
concentration profile will change. The logarithmic velocity profile will be distorted by the
orbital velocity of the waves. To represent the real situation best, the instantaneous
transport should be calculated within a wave cycle and then averaged over the wave
period. The applied numerical model (SUTRENCH, see Chapter 9) is based on time-
averaged wave influence. This means that the logarithmic velocity profile of ‘current
only’ is used for the situation with current and waves. The concentration profile is
modified through a different sediment mixing coefficient, see the next paragraph.
125
8.4 Sediment mixing coefficient
The sediment mixing coefficient is related to the fluid mixing coefficient according to the
following expression:
ε s = βφε f Eq. 8-13
in which:
εs = sediment mixing coefficient [m2/s]
εf = fluid mixing coefficient [m2/s]
β = ratio sediment mass and fluid momentum mixing coefficients [-]
φ = turbulence damping factor [-]
Factor ‘β’ represents the difference in mixing of a fluid ‘particle’ and a sediment particle.
Often values larger than 1 have been found in experiments, especially for large u*/ws-
values. This can be explained by the fact that sediment particles are thrown out of the
turbulent eddies. For small u*/ws-values the sediment diffusivity tends to become
smaller than the fluid diffusivity, so β<1. Because factor β cannot be predicted with high
accuracy, in this thesis a value of 1 will be used.
Factor ‘φ’ represents the damping of turbulence on the velocity profile by the sediment
particles and the adaptation of the sediment diffusion coefficient to the sediment
concentration. The latter appears to be most important, however, this influence remains
limited for sediment concentrations smaller than 10 g/l. Therefore this influence will be
neglected in this thesis: φ = 1. So, the sediment mixing coefficient is only dependent on
the fluid mixing coefficient: εs = εf.
For gradually varying flows simple mixing coefficient distributions for equilibrium
conditions are applied:
Current alone
Instead of a parabolic distribution, a parabolic-constant distribution is used, see also
Figure 8-3:
η
⎛ 2z ⎞ z
ε s ;c = ε s ;c;max − ε s ;c;max ⎜1 − ⎟ for < 0,5
⎝ h ⎠ h
z
ε s ;c = ε s ;c;max = 0,25βκu*c h for ≥ 0,5 Eq. 8-14
h
in which:
η = coefficient which equals 2 for the situation of a current alone;
u*c = bed-shear velocity of a logarithmic velocity profile:
κ Q
u*c = Eq. 8-15
⎛ h ⎞ bh
1 + ln⎜⎜ ⎟⎟
⎝ 0,03k r ⎠
Figure 8-3: Current- and wave-related mixing coefficient
126
Waves alone
The wave-related mixing coefficient is constant for the upper half of the water depth.
Furthermore the sediment mixing coefficient is constant in the wave boundary layer. In
between there is a linear relation:
ε s ,w = ε s ,w,bed = 0,00065d *2α br δuˆ 0 for z ≤δ

hH s
ε s ,w = ε s ,w,max = 0,035α br for z ≥ 0,5h
T 's
⎛ z −δ ⎞
ε s , w = ε s , w,bed + (ε s , w,max − ε s , w,bed )⎜ ⎟ for δ ≤ z ≤ 0,5h Eq. 8-16
⎝ 0,5h − δ ⎠
in which:
αbr = breaking coefficient, representing the influence of breaking waves on the sediment
mixing process = 1, if Hs/h < 0,6;
d* = dimensionless particle diameter, according to Eq. 7-2;
û 0 = amplitude of orbital velocity at the sea bed, according to Eq. 7-17;
T’s = significant wave period (relative to moving coordinate system).
Current and waves

When a wave field is superimposed on a current, the overall sediment mixing coefficient
is obtained by linear addition of the wave-related and current-related mixing
coefficients:
ε s ,cw = ε s ,c + ε s , w Eq. 8-17
The wave-related mixing coefficient is assumed to still satisfy equation 8-16; the
current-related mixing coefficient is expected to adjust to the presence of waves. Power
‘η’ in equation 8-14, which for the current-alone case equals 2 (parabolic distribution),
is now assumed to be in the range of 1 to 2:
uˆ 0
η = −0,25 +2 for 0 ≤ uˆ 0 ≤ 4 u
u
η =1 for uˆ 0 > 4 u Eq. 8-18
8.5 Channel morphology
Some characteristic phenomena, that will be observed later on, will be explained in
more detail.
Direction of sediment transport

As was mentioned in Chapter 6 on hydrodynamics, the direction of the current and
waves to the channel axis determines the current and wave action inside a navigation
channel or pipeline trench to a large extent. Therefore, also sediment transport
capacities are affected by the channel alignment. If the sediment transport is directed
parallel to the channel axis, the transport capacity inside the channel will be larger than
outside the channel. The morphology however is governed by cross-channel transport,
although there isn’t any cross-channel component of the current velocity. The
deposition of side slope particles on the channel bottom occurs mainly due to the action
of gravity. This effect cannot properly be modelled and only currents with a significant
cross-channel component will be considered in this thesis.
127
Trapping efficiency
The trapping efficiency factor is defined as the relative difference of the incoming
suspended transport and the minimum suspended transport (somewhere) in the
channel:
b0 S s 0 − b1 S s1,min imum
e= Eq. 8-19
b0 S s 0
in which
b0 = width of approaching streamtube;

b1 = width of streamtube in channel;
Ss0 = incoming suspended load transport per unit width;
Ss1 = suspended load transport in channel per unit width.
The basic parameters which strongly determine this trapping efficiency factor are the
approach angle (α0), the approach bed-shear velocity (u*0), the particle fall velocity
(ws), the channel depth (h1-h0 = D), the channel width (W) and of course the channel
side slope (tan β). Other parameters are of less significance [Van Rijn, 1986a], such as
the approach velocity (u0), the relative wave height (H/h0) and the relative roughness
(kr/h0).
In the framework of this thesis, it is also relevant to define the trapping efficiency of the
side slope (eslope). While the general definition of the trapping efficiency is a measure for
the overall rise of the channel bottom and the maintenance costs to be expected, the
specific definition of the trapping efficiency of the side slope is important for the
development of these side slopes. If this trapping efficiency is small, suspended
transport does not affect slope development much and the bed load transport is
normative. If this trapping efficiency is large, bed load and suspended transport control
slope development. The above of course only holds if bed load and suspended transport
are of the same order, which is the case for fine sands under most hydrodynamic
conditions. Slope development in medium to coarse sands can be attributed to bed load
transport; in silts bed load transport is small compared to suspended transport.
b0 S s 0 − btoe S s ;toe
eslope = Eq. 8-20
b0 S s 0
in which
btoe = width of streamtube at toe of (upstream) side slope;

Ss;toe = suspended load transport at toe of (upstream) side slope per unit width.
Slope development and trench migration

Slopes will change due to sediment transport and gravity effects. The more a slope
differs from its dynamic equilibrium profile, the faster slope development will take place.
Trench migration will occur in case of unidirectional flow or in asymmetrical tidal flow
and can cause problems if the channel differs too much from the initial alignment. This
migration however is very hard to define in a satisfactory way. Most problems arise
from the fact that the slope geometry itself does not keep its initial shape. So, the
migration of which part of the slope or the channel has to be considered? More practical
or more theoretical definitions can be used. Theoretical definitions tend to describe the
migration of a mathematically formulated channel geometry. Practical definitions are
related to a simpler application in practice and to the actual functionality of a navigation
channel or pipeline trench, which is often based on providing a required depth over a
certain width.
128
In Jensen [1999] the migration of the centre of gravity of the channel geometry below
the surrounding seabed is used as a measure for channel migration. This is a very
sound theoretical measure, but is rather laborious and does not provide any information
on the functionality of the channel.
Walstra et al. [2002] use another mathematical description of the channel geometry to
express the channel development. A logarithmic function through the lower inflection
points on the side slopes is used to describe the channel geometry. This approximation
shows reasonable resemblance to actual profiles, but is also rather laborious and
provides little information on the slope steepness, which is an important parameter of
this thesis. So also, this method has on purpose not been adapted here.
But what are satisfactory measures of slope development and trench migration with
respect to the objectives of this research study on slope development? When dealing
with slope stability the steepest parts of the slope will be normative. However, if this
steepest parts only stretch over a limited height, the consequences will be minor, unless
a very erosive turbidity current develops. Besides, the location of the steepest part
along the slope is of importance. A steep toe is more susceptible to static liquefaction
than a steep top. Of course this also depends on the soil properties.
So a method to describe slope development and trench migration that is completely
legitimate in all cases and that is easily applicable in engineering practice perhaps does
not exist.
Therefore the steepest slope over a cross-channel width of 4 to 5 m (= 3 grid points) is

taken as a measure for the slope steepness. Often is found that this steepness extends
over a far greater height. This method, resulting in just one value, is complemented
with a graphical observation of the developed slopes.
The trench migration should of course be related to the navigability and is therefore
defined as the migration of that point at the upstream side slope where 90% of the
initial channel depth with respect to the surrounding seabed ‘h1-h0’ (which is equal to
the slope height ‘D’) is reached.
In this thesis not much attention is given to the morphological influence of the trench on
the surroundings, which was the case when considering large scale morphological
mining pits [Walstra et al., 2002]. The dimensions of navigation channels and pipeline
trenches are so small that this morphological influence will remain limited. As a
consequence the exact total volume of a trench is not as important as the useful volume
(required depth times the width at which this depth is guaranteed) for practical reasons
of navigability.
129
9 SUTRENCH, model description

9.1 Introduction
In rivers or narrow tidal inlets one can apply simple 1-dimensional (depth-averaged)
models to calculate the general hydraulic conditions. In seas and wide estuaries 2-
dimensional models have to be used, which are mostly depth-averaged. In Chapter 6 on
hydrodynamics was already shown that currents refract when entering the channel due
to the instantaneous effect of continuity and the more gradual effect of longitudinal
acceleration. In 2DV-models this effect is accounted for by an increase of the apparent
channel width. This increase is in reality a slowly developing process over the channel
width. Quasi-3D-models can capture the depth-averaged current refraction, based on
assumptions of hydrostatic pressure distributions and logarithmic velocity profiles.
Besides this depth-averaged current refraction, on the side slopes also a distortion over
the vertical velocity profile occurs; the bed shear stress will be more refracted on the
side slopes than the surface velocity. This effect will be more significant if side slopes
become steeper. Quasi-3D-models can only account for unidirectional horizontal
velocities. A full 3D-model is necessary to capture the complete process.
In this thesis the stability of a trench and its slopes will be treated as a 2-dimensional
problem (the ‘slices’ along the channel axis are all similar). This assumption to deal with
the problem in a 2DV-environment implies that primarily cross-channel flow will be
modelled and that oblique currents have to be changed into cross-channel currents with
adapted stream widths ‘b1’ and channel widths ‘W’. With this trick, sediment transport
can be reasonably modelled, although the difference in direction between the depth-
averaged velocity and the bed shear stress is neglected. Parallel currents or currents
with very small angles to the channel axis cannot be simulated. The width-averaged
2DV-program SUTRENCH by Delft Hydraulics will be applied to model the morphology of
a channel.
Before continuing with the model description, one should realize that at the present
state of research the application of numerical models that are not calibrated to specific
field conditions has a very limited predicting value, at least when it comes to absolute
values. Most numerical models, however, represent a satisfactory to good relative
behaviour over wide ranges of wave and current conditions, as was concluded by Davies
et al. [2002] within the framework of the SEDMOC Project. The TRANSPOR-model,
which is used in SUTRENCH, proved to yield results that are in the middle of the results
of all tested models. When compared to actual field measurements, more than half of
the calculated results lie within a confidence band of a factor 2 (and the resemblance is
even better for depths between 5 and 15 m and significant sediment concentrations,
which is the range of interest of this thesis. Moreover, the relative behaviour is assumed
to resemble nature very well.
This means that the time scales of the development of a trench and its side slopes in
particular may be somewhat inaccurate, but the occurring changes in profile and slopes
should be reliable, at least for channel geometries that are not too sharp-edged.
9.2 Model description
SUTRENCH (Van Rijn and Tan, 1985; Van Rijn, 1986) is a two dimensional vertical
mathematical model for simulation of bed-load and suspended load transport under the
conditions of steady state currents and wind-induced waves. It can be applied at space-
scales from 1 to 5 kilometres and at time-scales from 1 to 100 years in regions outside
the surf zone where wave breaking is limited. On the other hand, knowledge on
sediment transport in deeper water (>20 m) is still limited. Therefore in this thesis only
water depths between 5 and 20 will be considered.
130
In this thesis SUTRENCH will be applied at the lower limits of its space and time scales:
the emphasis will lay on space scales around 800-1200 m in order to minimize grid
spaces and on time scales up to one year in order to study the short-term development.
The SUTRENCH model is restricted to well-mixed gradually varying flow conditions over
a sediment bed consisting of fine particles with narrow size gradation (d90/d10 = ± 3). In
this thesis a gradation of d90/d10 = 1,5 (see Paragraph 3.1.2) is applied. The model is
applicable with or without waves. Tidal flow can be represented by schematizing the real
tidal period to quasi-steady flow periods. SUTRENCH uses the TRANSPOR-2000
sediment transport model. During the SEDMOC-project TRANSPOR-1993 has been
improved to this newer version.
Basic processes
Hydrodynamic
- Modification of velocity profile and associated bed-shear stress due to the presence
of waves;
- Modification of velocity and associated bed-shear stress due to the presence of
sloping bottom.
Sediment transport
- Advection by horizontal and vertical mean current;
- Vertical mixing (diffusion) by current and waves;
- Settling by gravity;
- Entrainment of sediment from bed due to wave- and current-induced stirring;
- Bed-load transport due to combined current and wave velocities (instantaneous
intra-wave approach);
- Slope-related transport components (bed-load);
- Effect of mud on initiation of motion of sand;
- Non-erosive bottom layers.
Basic simplifications
Hydrodynamic
- Logarithmic velocity profiles and associated bed-shear stress also in conditions with
waves: steep sided trenches and channels (steeper than 1:5) cannot be modelled
correctly;
- Shoaling and refraction of wind waves is not implicitly modelled, but can be taken
into account by the input data;
- Current refraction (veering) is not implicitly modelled.
Sediment transport
- Steady state sediment mass conservation integrated over the width of the flow
(stream tube approach);
- No longitudinal mixing (diffusion);
- No wave-related suspended sediment transport (no oscillatory transport
components)
- Uniform grain size (no mixtures or grain distributions).
Numerical
- Forward-marching numerical scheme (transport due to near-bed return currents can
not be modelled);
- Explicit Lax-Wendorff numerical scheme for bed level changes (smoothing effects).
Drawbacks
- SUTRENCH becomes unstable for very small waves and currents
- a frequent problem of ‘simple’ numerical transport models is also the
underestimation of sediment concentrations compared to field conditions. This is
attributed to a residual amount of suspended sediment in calm conditions (e.g. near
slack water) caused by a certain existing background level of turbulence which is not
included in the model formulation
131
Continuity equation for suspended sediment

The time-averaged sediment convection-diffusion equation in the 2DV-plane can be
described as follows:
∂c ∂ ∂ ⎛ ∂c ⎞ ∂ ∂⎛ ∂c ⎞
+ uc − ⎜ ε s ,cw ⎟ + (( w − ws )c) − ⎜ ε s ,cw ⎟ = 0 Eq. 9-1
∂t ∂x ∂x ⎝ ∂x ⎠ ∂z ∂z ⎝ ∂z ⎠
in which:
u = longitudinal velocity at height z above the bed;
c = sediment concentration;
w = vertical flow velocity;
If one assumes steady state conditions, neglects longitudinal mixing (usually an order of
magnitude smaller than the other terms) and applies a stream tube approach to
account for diverging or converging flows, this equation reduces to:
∂ ∂ ∂⎛ ∂c ⎞
buc + (b( w − ws )c) − ⎜ bε s ,cw ⎟ = 0 Eq. 9-2
∂x ∂z ∂z ⎝ ∂z ⎠
in which:
b = width of stream tube;
If the flow is in equilibrium and the width of the stream tubes is constant, equation 9-2
reduces to:
∂c
cws + ε s ,cw =0 Eq. 9-3
∂z
But most of the times the flow will be gradually changing and equation 9-2 has to be
solved numerically. Therefore the flow width b as a function of distance x must be
known a priori. Also the flow velocities, the sediment mixing coefficients and the particle
fall velocities should be known. All but the vertical flow velocity have been discussed
before. The vertical flow velocity can be computed from the (width-integrated) equation
of continuity of the fluid:
zb + z z +z
∂u 1 db b ∂u
w=− ∫
b dx zb ∫+ z0 ∂x
dz − udz Eq. 9-4
zb + z 0
∂x
Boundary conditions
All necessary boundary conditions will be mentioned briefly.
Flow domain
The initial bottom level ‘zb’, the water level ‘h’ and the flow width ‘b’ have to be defined.
Inlet boundary
The location of the inlet boundary should not experience any morphological changes.
This means that for unidirectional flow the inlet boundary can be close to the upstream
slope, while for tidal flow the inlet boundary also acts as the outlet boundary; then the
location should be as far away from the area of interest as possible. However, one
wants to describe the relative sharp channel geometry in a sufficient fine mesh (small
Δx). Due to a limited number of grid points and a constant grid size, this means that the
channel geometry has to be carefully positioned in calculation space for every
calculation. For these types of problems an adjustable grid size would be appropriate,
especially in tidal flow.
132
Assuming logarithmic velocity profiles, only the discharge ‘Q’ has to be defined for the
fluid. The sediment concentration c(z,t) is in this thesis considered to be in equilibrium,
although also measured or zero-concentration profiles can be set at the inlet boundary.
Outlet boundary
The location of the outlet boundary should be far away from the downstream slope,
taking the above considerations into account.
Water surface
No vertical sediment transport can take place through the water surface:
⎛ ∂c ⎞
⎜ ws c + ε s ⎟ =0 Eq. 9-5
⎝ ∂z ⎠ z = zb + h
Bed boundary
The vertical flow velocity at the zero-velocity level (z0 = 0,033kr) has to be zero: w=0 at
z=z0. For the sediment transport the equilibrium bed concentration ‘ca,e’ and the upward
sediment flux Ea have to be specified. The bed concentration can be calculated
according to formula 8-11 and is plotted in Figure 9-1. The sediment flux (gradient of
concentration profile at reference level ‘a’) can be deduced from equation 9-3:
d ⎛ τ ' −τ
1, 5
⎛ ∂c ⎞ ⎞
E a = ⎜ − ε s ⎟ = −0,015ws d *−0,3 50 ⎜⎜ b,cw cr ⎟⎟ Eq. 9-6
⎝ ∂z ⎠ a a ⎝ τ cr ⎠
Figure 9-1: Schematization of equilibrium concentration profile and bed boundary condition at
reference level ‘a’
Bed level changes

If the concentration profiles and the bed load and sediment transport are known at
every location, the bed level changes can be computed:
∂bz b 1 ⎡ ∂bhc ∂S tot ⎤

+ + =0 Eq. 9-7
∂t ρ s (1 − n) ⎢⎣ ∂t ∂x ⎥⎦
in which:
c = depth-averaged concentration
Stot = total sediment transport = Sb + Ss
The storage term ‘∂bhc/∂t’ can be neglected if flow conditions are considered quasi-
steady.
133
Numerical solution
The continuity equation for local sediment (Eq. 9-2) and the equation to compute bed
level changes (Eq. 9-7) have to be solved numerically.
Therefore a 2-dimensional grid (Δx in cross-channel direction and Δz in direction of

depth) is defined. The horizontal grid size is constant, while the vertical grid size
decreases towards the bed in order to obtain a greater resolution where largest
gradients exist. Some rough guide lines are:
Δx: at least 10 grid points over the characteristic length scale (side slope). In this thesis
cross-channel grid sizes of 2 to 2,5 m are used, which is rather small at locations where
flow conditions are constant, but may be rather large at the side slopes, especially when
the side slopes are very steep.
Δz: at least 10 grid points for current-alone conditions and at least 20 points if waves
are superimposed on a current. In this thesis values no larger than 15 are applied,
because there is a restriction to the total number of grid points. The formula of
maximum calculation space is:
5(8 N + 4 NV + N ⋅ NV ) ≤ 50.000 Eq. 9-8
in which:
N = total horizontal grid points;
NV = total vertical grid points.
This restriction to calculation space yields for a length scale of 1 km (with Δx = 2,5m) a
number of 400 horizontal grid points and thus maximum 16 vertical grid points. This
example demonstrates that it is very hard to fulfil all requirements.
The numerical scheme (Lax-Wendroff) that is used to compute the new bed level at
time t+Δt is presented to show how SUTRENCH handles with sharp transitions in bed
level, like the side slopes:
N t Δt
zbt +Δ
, x = zb , x −
t t
( S xt +Δx − S xt −Δx ) + 12 α s ( zbt , x +Δx − 2 zbt , x + zbt , x −Δx ) Eq. 9-9
2(1 − n) ρ s bΔx
in which
αs = bed-level smoothing coefficient

Nt = number of time steps between computation of two subsequent bed levels
zbt , x = bed level at a certain time t and location x
t
S x = total sediment transport at a certain time t and location x
This numerical scheme implies that due to numerical smoothing even with zero or very
small sediment transport bed level changes occur. During simulations of various
geometries and soils, numerical problems would arise. It appeared that a very clever
choice of smoothing coefficient ‘αs’, time step and calculation space for every different
situation had to be made to deal with numerical diffusion or instability.
If the problem is perfectly suitable for computation with SUTRENCH, the numerical
smoothing coefficient can be chosen very small. However, sometimes, especially in
unidirectional flow, instabilities can occur due to oversteepening of the upstream slope.
This oversteepening phenomenon of course is dependent on sediment transport and,
more particular, the transport gradient. In reality some smoothing processes on very
steep slopes are very likely to occur, so the application of a numerical smoothing
coefficient is in some way legitimate, although there is no physical meaning. This
artificial trick has some major consequences for the time step: if the time step is
reduced, the numerical smoothing process starts to prevail over the real sediment
transport process: every time step the bed level is smoothened with a constant rate,
134
while the sedimentation/erosion rates decrease, if the time step is reduced. So reducing
the time step should be accompanied by an equal reduction of the numerical smoothing
coefficient.
Concluding, it can be stated that numerical diffusion is smallest when:

- there is a significant sediment transport;
- the transport gradients are small, which is the case when suspended transport has a
significant share in total sediment transport;
- the flow is reversing, like tidal flow;
- the initial channel geometry is not too sharp (gentle side slopes).
This again shows that in this thesis SUTRENCH has to be used at its limits of
applicability to study the slope development of rather steep side slopes.
9.3 Relevant input parameters
Before discussing the results of simulations of SUTRENCH some input parameters will be
explained. The choice of the input values of these parameters will be discussed.
Dimensions of computation space

The number of horizontal and vertical grid points (‘N’ and ‘NV’ in SUTRENCH) together
with the horizontal grid size Δx and water depth h define the calculation space. The
water depth at the surrounding seabed is kept constant at 10m, the number of vertical
grid points is around 15 with decreasing Δy towards the seabed, automatically
calculated by SUTRENCH. The horizontal grid size is 2 or 2,5 m and the number of grid
points is 401 to 501, resulting in a horizontal calculation length of 800 to 1200 m. So,
the horizontal grid size is explicitly inserted by the user, while the vertical grid size is
implicitly calculated by SUTRENCH.
Dimensions of computation time

The computation time is defined rather cumbersome and consists of the product of the
number of tidal cycles (Ntidalcucles) and the duration of one tidal cycle (Σ(ΔtTS)), which is
the sum (from 1 to NTS) of all time steps (ΔtTS) during one tidal cycle:
TS = NTS
ComputationTime = N tidalcycles ∑
TS =1
ΔtTS Eq. 9-10
Numerical and physical parameters

- Reference level ‘a’ (‘ZA’ in SUTRENCH): the reference level at which the bed
boundary is applied is usually assumed to equal half the bed-form height. In this
thesis a = 0,05m;
- Acceleration of gravity ‘g’ = 9,81 m/s2;
- Constant of Von Karman ‘κ’ (‘RK’ in SUTRENCH) = 0,4;
- Kinematic viscosity of water (‘RNU’ in SUTRENCH) = 1*10-6 m2/s for fresh water and
1,3*10-6 m2/s for salt water;
- Angle of internal friction (‘PHI’ in SUTRENCH) is taken as 35°, representing medium
packed coarse, medium and fine sand and considered to be an averaged value of all
reference soils;
- Sediment density (‘RHOS’ in SUTRENCH) = 2650 kg/m3;
- Fluid density (‘RHOW’ in SUTRENCH) = 1000 kg/m3 for fresh water and 1030 kg/m3
for fresh water;
- The numerical coefficient αs (‘ALFA’ in SUTRENCH) is varied between 0,01 to 0,5,
depending on the type of simulation. SUTRENCH allows values in a range of 0,001 to
1,0;
- The sediment upward boundary condition can be inserted as an upward sediment
flux or as an equilibrium bed concentration, which results in a rapid adjustment of
the suspended load vertical to the equilibrium concentration. This was one of the
assumptions: incoming sediment transport is in equilibrium;
135
Wave data
The wave data can be defined variable in time (and thus constant over the entire
computation area) or variable over the computational area (and thus constant in time).
As was mentioned before, the variation of wave height over the expansion in depth due
to shoaling and refraction effects yielded only small variations in morphological
behaviour. Therefore a time-dependent wave-field will be defined: at first with constant
wave height, later various schematizations of varying wave heights.
Bottom profile
The bottom profile can be read from an input file (*.pol) or be filled in manually.
SUTRENCH interpolates between the inserted bottom profile coordinates.
Boundary conditions per time step

- Discharge Q [m3/s] at upstream boundary, positive for flood discharge and negative
for ebb discharge;
- Water depth h with respect to reference level at x=0. Because the water level is
assumed horizontal the water level is known over the entire computation area;
- Particle fall velocity ws [m/s] of suspended material, which can be rather different
from the bed material, especially for wide gradations;
- Particle diameter d50 [m] which follows from the definition of the reference soils;
- Particle diameter d90 [m] which was defined at 1,5d50, see Paragraph 3.1.2;
- Porosity n, which varies from 0,35 to 0,45 according to the definition of the
reference soils, but is kept constant throughout the simulations at a value of 0,40,
representing the ‘average reference soil’;
- The duration of this particular time step (‘DT’ in SUTRENCH), which varies between
1 to 6 hours depending on the flow schematization;
- Correction factor of bed-load concentration (‘CA’ in SUTRENCH), which determines
suspended sediment transport. If measured sediment verticals are available, this
coefficient can be varied to fit calculated sediment transport at the upstream
boundary. In this thesis no such data are available, so CA=1;
- Correction factor of bed-load transport (‘SB’ in SUTRENCH), which determines bed-
load sediment transport. The same holds for this correction factor, so SB=1;
- Coefficient β (‘BF’ in SUTRENCH) was mentioned before and considered to equal 1;
- The angle between current and wave direction (‘RPHI’ in SUTRENCH) is set to 90°
for all simulations, except the simulations with oblique flow.
Per time step also flow refraction effects can be taken into account by varying the width
of the stream tube across the channel. In case of perpendicular flow this flow width
remains constant and therefore the width equals 1.
The current- and wave-related bed roughness can also be varied across the channel.
Sometimes this roughness is considered to be depth-dependent, primarily because this
allows for nice mathematical formulations (water depth drops out of the friction
coefficient). Other researches like Davies et al. [2002] mention “the self-regulating
nature of the seabed roughness, whereby rippled beds formed at low (orbital) flow
stages (small waves) tend to enhance sediment transport, while plane beds formed at
high flow stages tend to inhibit the transport”. Not only the flow stage determines the
bed roughness, but De Boer (1998) also mentions the packing of the seabed: “the
transport on a densely packed bed, which is relatively smooth, is larger than on a loose
bed.” This sediment density however is not constant in time.
Representing the bed roughness as a constant value is easily applicable, but not
completely legitimate. On the other hand describing the bed roughness as a function of
depth, flow stage, time, sediment diameter and angularity, is almost impossible,
especially because fitting data are very scarce. Current- and wave-related bed
roughness are therefore kept constant at 0,03m, see Paragraph 7.1.1.
136
10 Results of SUTRENCH-simulations
Although many parameters can be changed, many hydrodynamic conditions can be

investigated, many channel and trench geometries should be considered and multiple
schematizations of reality should be verified, some choices had to be made.
At first an extensive sensitivity analysis was executed to gain insight in the sensitivity of
18 important input parameters. Then the “navigation channel” is subjected to a
unidirectional current with a constant perpendicular wave field. Initial side slopes,
magnitude of the current and wave height will be changed in order to find out what the
consequences for the upstream and downstream slope steepness will be. The same
procedure is followed for schematized tidal flow, represented by a block function. Now,
the distinction between upstream and downstream slopes has disappeared. In order to
obtain a better resemblance to real hydrodynamic conditions a real tidal and wave
climate of the North Sea has been used. The wave climate will be schematized in
different ways to investigate the influence of this schematization and to come to a wave
height that is representative for a full year.
All of the above simulations are based on currents perpendicular and waves parallel to
the channel axis, because this situation is representative for most navigation channels
perpendicular to the coastline with a longshore tidal current and a wave field that has
refracted towards the coast. As was discussed in great detail in Chapter 6, current
velocities and wave heights are affected by an expansion in water depth and this effect
is largely dependent on the angle to the channel axis.
Van Rijn [1986a] investigated the situation with waves directed opposite to the current,
because then the reduction of the wave height inside the channel is largest (up to
20%). Although he found a further reduced sediment transport capacity due to the
smaller sediment pick-up and wave-related mixing, the overall effect on bed level
changes remained limited to only a few percent. This adaptation of the wave height is
therefore considered to be of minor importance. Current refraction will be of
significance, but will not be investigated in this thesis due to time restrictions. However,
it is interesting to compare situations without current refraction, with instantaneous
refraction solely due to continuity and with slowly adapting current refraction with each
other. One should note that the effective length of a side slope increases if a current
enters the channel under an angle: a current that has an incoming flow angle of 45°
over a side slope of 1:5 ‘feels’ this slope as it was 1:7. This effect is favourable with
respect to resemblance to a logarithmic velocity profile.
10.1 Sensitivity Analysis
10.1.1 Previous results
To gain some insight in the sensitivity of the calculated results to small variations in the
input parameters a sensitivity analysis was done. To be able to compare the results with
previous researches, the conclusions of two of these researches are presented.
I. In his thesis Jenniskens [2001] studied large numbers of data on sediment

transport and deduced a strongly simplified approach formula of the original total load
formula of Van Rijn (1993). Although it is advised not to apply this formula in transport
calculations, it tells a lot about the sensitivity to variation of the input parameters. An
important drawback is the fact that only five input parameters are included, but these
parameters are all within the range of this thesis. This formula reads:
137
S total ;VanRijn _ approach = 2 ⋅ 10 −6 H s1,7T p0, 2 h −1, 2 u 2,7 d 50−1,75 Eq. 10-1
with 0,5 ≤ Hs ≤ 3 [m], 5 ≤ Tp ≤ 9 [s], 3 ≤ h ≤ 15 [m], 0,2 ≤ u ≤ 1,2 [m/s] and 125 ≤ d50 ≤
500 [μm] on condition that Hs / h ≤ 0,6.
In this thesis the current velocity and water depth are related via the discharge q, which
remains constant when passing over a trench. Therefore the above formula will be
modified to:
S total ;VanRijn _ approach = 2 ⋅ 10 −6 H s1,7T p0, 2 h −3,9 q 2,7 d 50−1, 75 Eq. 10-2
in which the water depth h has become the most influential input parameter, because
this parameter affects the wave and current stirring action on the bed and the transport
capacity of the current. On the trench bottom (h=h1) not only the waves are ‘felt’ to a
lesser degree, the current velocity has also dropped, which reduces bottom shear
stresses and transport.
Furthermore it can be seen that the wave height Hs is far more influential than the wave
period Tp. The bottom shear stress induced by waves is quadratic proportional to the
wave height, so the increased ‘stirring effect’ is easily accounted for, but when it comes
to transport the situation gets far more complicated and SUTRENCH does not well
represent nature anymore. In presence of waves the velocity profile becomes distorted
and if waves are getting bigger (and so are the horizontal orbital velocities), this profile
cannot be presented like an ideal logarithmic profile anymore. Sediment transport is
calculated as the depth integrated product of concentration and velocity. As soon as one
of both profiles does not resemble reality, the calculated total transport becomes
inaccurate.
Another important aspect is the before mentioned ‘phase lag’, i.e. the situation that
small particles that are stirred up by waves in positive direction can be transported in
negative direction due to the small particle fall velocities. This effect reduces with
increasing wave period; a small positive relation of the sediment transport to the wave
period is often noticed in experiments (Dohmen-Janssen (1999) as described by
Camenen and Larroudé [2002]). The appearing ‘coincidentally’ power of 0,2 in Eq. 10-2
however, cannot be caused by this effect, because Van Rijn does not take sediment
transport by waves into account. The small positive influence has to be explained in a
somewhat different way. On the one hand, larger wave periods mean longer waves and
therefore larger horizontal excursions and velocities; on the other hand, the wave
boundary layer has more time to develop, thereby reducing the friction factor fw. The
overall net effect is that bed shear stresses first increase for larger wave periods, but
eventually become smaller if the wave period further increases. This effect is also
depending on the water depth, as described in Paragraph 7.2.
Jenniskens found a significant negative relation between sediment size and sediment
transport (power of -1,75), which seems logical and is unconditionally true for
unidirectional flow. However, because of the above mentioned ‘phase-lag phenomenon’,
sediment transport can increase if sediment particles become larger, because these
particles are subjected to the phase-lag to a lesser degree, because of their larger
particle fall velocities. That’s why sediment transport under waves may increase with
increasing particle diameters. Again, Van Rijn does not account for this effect and
according to his formula sediment transport always decreases with increasing particle
diameters.
II. When Klein [2003] verified SUTRENCH on some laboratory and field
experiments, he also executed a sensitivity analysis of ten different input parameters on
both the (suspended) sediment transport and the sediment transport gradient. The
most important conclusions were:
With respect to sediment transport:

- the results are extremely sensitive to variation of wave height Hs and wave-related
bed roughness height ‘kr;w’;
138
- the discharge ‘Q’, the water level ‘h’, the particle fall velocity ‘ws’ and the reference
level ‘a’ have a significant influence on the magnitude of sediment transport;
- the current-related bed roughness height ‘kr;c’, particle sizes ‘d50’ and ‘d90’ and wave
period ‘Tp’ have little influence.
These conclusions are to a large extent compatible with the approach formula of
Jenniskens. The conclusions of Klein on the influence of the water depth and the
discharge, which are the most influential parameters according to Jenniskens, are
justified if one adopts the idea that water depth and discharge can be measured quite
accurately and are more or less regularly varying in time (with the tide), while the wave
height and bed roughness are strongly varying in time and/or difficult to determine,
especially bed forms during storms.
With respect to sediment transport gradient (erosion / sedimentation):

- the particle fall velocity ‘ws’ has the largest influence on the transport gradient;
- the particle size ‘d50’ and ‘d90’, the reference level ‘a’ and the wave period ‘Tp’ hardly
have any influence;
- the other parameters have significant influence.
So it appears that if an expansion in water depth is present and the sediment transport
capacity changes, the particle fall velocity becomes the governing parameter.
10.1.2 Input parameters
Are the above conclusions also found in the current sensitivity analysis? Besides, eight
other input parameters are studied. Three parameters are related to the sediment
properties, viz the sediment density ‘ρs’, the sediment porosity ‘n’ and the angle of
internal friction ‘φ’. These parameters are related to the slopes that will develop and are
therefore included in this analysis. Two parameters are related to the current-wave-
characteristics, viz kinematic viscosity ‘ν’ and the angle between current and waves
‘αc-w’. Three parameters are numerical coefficients normally used to calibrate the
results: a numerical smoothing coefficient ‘αs’, a correction coefficient ‘CA’ for the bed
load concentration ‘ca’ (bed boundary condition for suspended transport) and a
correction coefficient ‘SB’ for the bed load transport ‘Sb’; the latter two equal 1 in
theoretical equations, but can be useful when adjusting the calculated concentration
profile to a measured profile. Because in this thesis the calculations cannot be calibrated
with values from experiments, the sensitivity of these parameters is examined.
The variation of the different parameters is based on expected difficulties in

determination of the exact value of a certain parameter in nature. Therefore the rate of
variation is not the same for all input parameters. The three numerical parameters are
varied between values that are common in previous calculations with this model. For
instance, Klein found values for ‘CA’ and ‘SB’ between 0,5 and 1,15 for a roughness of
0,03. This implies that the default value of 1 might be an overestimation.
Of course all results have been compared to a reference case which is a calculation with
all mean values as input parameters, see Table 10-1.
The bed geometry that is considered in this sensitivity analysis is the reference case
‘navigation channel’ with a width of 200 m, so the upstream and downstream slope will
not be influenced by each other, because of total adaptation of flow and sediment
transport to local conditions. Because of this large width, both slopes can develop
autonomously. The water depth inside the channel is 20 m and the surrounding water
depth is 10 m.
139
Table 10-1: input parameters of sensitivity analysis; variation around mean values
input par. description symbol unit var +/- mean min max
ZA reference level of bed-form height a m 40% 0,05 0,03 0,07
2
RNU kinematic viscosity ν m /s 18% 1,10E-06 9,00E-07 1,30E-06
PHI angle of repose of bed material φ degr 15% 35 30 40
RHOS sediment density ρs kg/m
3
2% 2650 2600 2700
ALFA pseudo-viscosity factor; numerical smoothing coefficient αs - 90% 0,5 0,1 1
Hs significant wave height Hs m 20% 1,5 1,2 1,8
Tp peak period Tp s 15% 7 6 8
3
Q discharge; flood is positive; ebb is negative Q m /s 10% 10 9 11
H water surface level w.r.t.datum/reference level h m 5% 10 9,5 10,5
WS particle fall velocity suspended material: 0,01-0,05 ws m/s 25% 0,0252 0,0189 0,0315
D50 50 % particle diameter of weight of bed material d50 μm 25% 200 150 250
D90 90 % particle diameter of weight of bed material d90 μm 25% 300 225 375
P porosity of bed material n - 13% 0,4 0,35 0,45
CA correction coefficient of bed load concentration CA - 30% 1 0,7 1,3
SB correction coefficient of bed load transport SB - 30% 1 0,7 1,3
RPHI angle between wave and current direction αc-w ° 20% 90 72 108
RC current related bed roughness kr;c m 67% 0,03 0,01 0,05
RW wave related bed roughness kr;w m 67% 0,03 0,01 0,05
10.1.3 Results of sensitivity analysis
Output parameters
The output parameters on which the separate runs will be compared are the bed-load,
suspended and total sediment transport (Sb, Ss and St) on a plane bed (at 50 m from
the upstream boundary) as well as the minimum gradient of bed-load, suspended and
total transport on the upstream ((dSb/dx)min, (dSs/dx)min, (dSt/dx)min) slope as a
measure for the maximum sedimentation that occurs somewhere along the slope. The
transport gradient is a good measure for slope development, because not the sediment
transport itself determines whether the seabed erodes or accretes, but the change in
sediment transport.
A qualitative (visual) analysis of the morphological development after 1 month will be
done for the most relevant input parameters to demonstrate the long-term effects of
small variations in the input parameters.
The above mentioned period of 1 month (350 time steps of 2 hour = 29,17days) is
considered to be representative, because slopes have adapted, but ‘silting up problems’
have not occurred yet. Of course, this relatively small period is related to the rather
strong constant unidirectional current with a constant velocity of 1m/s and a relatively
large (constant) significant wave height of 1,5m.
Sensitivity of sediment transport and gradient

In Table 10-2 the results of 37 sensitivity runs with SUTRENCH are presented. In design
practice an engineer is likely to deal with variations of this magnitude. Because one
parameter can better be estimated or measured than another, the resulting variation of
each input parameter may differ significantly.
If grain mixtures are very well graded, a problem arises when applying SUTRENCH and
errors may rise excessively: the suspended sediment is likely to differ from the bed
sediment. Even under changing hydrodynamic conditions the grain distribution of the
suspended sediment may constantly adapt itself.
Variations in bed-load, suspended and total sediment transport are presented in
percentages. A negative value means that sediment transport is reduced. For instance
when the wave height is increased by 20% (Hs from 1,5 to 1,8 m), the total sediment
transport is increased by 42%, which can be mainly contributed to the suspended
sediment transport, which is increased by 45%. Larger waves are extremely effective in
stimulating bed particles to leave the bed-load layer and getting mixed throughout the
sediment vertical. These values for the variations in sediment transport are all
determined at the flat surrounding seabed (equilibrium situation), unlike the sediment
transport gradients. These gradients are the largest negative gradients, occurring
somewhere along the upstream slope. This is in fact a measure for the largest
sedimentation to be expected.
140
Table 10-2: Output parameters of sensitivity analysis [%]

sediment transport minimum gradient (=max sedimentation)
Sb Ss St (dSb/dx)min (dSs/dx)min (dSt/dx)min
symbol variation min max min max min max min max min max min max
a 40 0,0 0,0 40,2 -16,3 36,1 -14,6 0,0 0,0 51,0 -23,0 40,4 -18,6
ν 18 5,4 -12,4 7,9 -16,1 7,7 -15,7 5,0 -11,6 4,7 -8,9 4,7 -9,9
φ 15 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,8 -0,4 0,6 -0,6
ρs 2 2,4 -2,3 2,2 -2,1 2,2 -2,1 2,0 -2,0 2,7 -1,9 2,5 -2,5
αs 90 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0
Hs 20 -14,6 17,8 -30,9 44,9 -29,3 42,2 -11,0 13,2 -14,4 16,3 -13,7 15,5
Tp 15 -13,7 9,7 -8,6 4,4 -9,1 5,0 -8,5 5,3 -8,2 5,1 -8,1 5,0
Q 10 -27,2 35,3 -41,0 64,1 -39,7 61,2 -27,4 35,3 -29,2 37,0 -28,9 36,6
h 5 23,2 -18,2 40,3 -27,9 38,6 -26,9 26,0 -20,0 27,2 -20,6 27,0 -20,5
ws 25 0,0 0,0 116,1 -45,4 104,5 -40,8 0,0 0,0 -4,3 -0,4 -3,7 -0,6
d50 25 -12,0 2,4 -16,6 10,5 -16,1 9,7 -12,4 3,1 -7,0 1,2 -8,1 1,6
d90 25 -14,5 13,2 -6,2 5,1 -7,0 5,9 -13,5 12,2 -5,8 5,1 -7,8 6,2
n 13 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0 0,0
CA 30 0,0 0,0 -30,0 30,0 -27,0 27,0 0,0 0,0 -30,0 30,4 -23,9 23,9
SB 30 -30,0 30,0 0,0 0,0 -3,0 3,0 -30,1 30,1 0,0 0,0 -6,2 5,9
αc-w 20 22,5 14,8 0,4 0,1 2,6 1,6 19,5 12,4 0,8 0,4 4,3 2,5
kr;c 67 26,5 -13,5 -5,6 1,5 -2,4 0,0 24,2 -12,7 17,1 -10,1 18,0 -10,6
kr;w 67 -10,8 5,4 -22,7 18,1 -21,5 16,8 -10,9 5,3 -20,2 15,6 -18,3 13,4
In Table 10-2 variations in sediment transport or transport gradient exceeding 20% are
plotted in bold. Some conclusions can be drawn:
- Reduction of the reference height ‘a’ introduces serious errors. This parameter
should certainly not be chosen too small;
- The angle of internal friction ‘φ’ of course does not affect sediment transport on a
flat bed in equilibrium. The gradient, which is determined at the upstream slope, is
slightly affected, although the gradient of the bed-load transport remains the same.
This does not imply that the bed-load transport at the slope remains the same; this
will increase for smaller friction angles;
- The solid sediment density ‘ρs’ has little effect (about 2%);
- The numerical smoothing coefficient ‘αs’ does not affect initial sediment transport
rates. However it does have great influence on morphological changes and will
certainly become more important in the long run. Almost the same explanation
holds for the bed porosity ‘n’, which affects bed level changes and therefore
eventually sediment transport;
- The main wave parameter is the significant wave height: both bed-load (wave
stirring) and suspended (wave mixing) transport will increase for larger wave
heights. The wave period only affects bed-load transport due to larger accelerations
near the bed;
- The flow velocity (Q/h) has a strong positive correlation with sediment transport and
transport gradient, because sediment pick-up, convection and mixing are involved;
- The particle fall velocity usually is related to the particle diameter, but in SUTRENCH
both properties have to be entered separately. An increase of the particle fall
velocity reduces suspended (and total) sediment transport and has no effect on bed-
load transport, while an increase of the grain size increases both bed-load and
suspended transport due to the larger particle volume. In reality both properties are
related and the overall effect will be a negative correlation;
- The correction factors for bed-load (SB) and suspended (CA) transport need no
further explanation. The variation in input (30%) results in an equal variation in
output, which is an indication that the SUTRENCH-results are sound;
- The angle between waves and currents primarily affects bed-load transport.
Furthermore it appears that a wave field that is directed perpendicular to the current
yields least sediment transport (‘min’ and ‘max’ value both have positive sign);
- Sediment transport is more sensitive to the wave-related bed roughness than to the
current-related roughness.
141
Approximation and/or sensitivity formula

The influence of the most important input parameters can also be expressed as a
power-lay function of the total sediment transport, comparable to the approach of
Jenniskens [2001].
S t = Cx1p1 x 2p2 x3p3 x 4p4 x5p5 Eq. 10-3
When considering one input parameter at a time, two approximations can be calculated:
one power-law function through the reference case and the minimum value of the input
parameter and one power-law function through the reference case and the maximum
value of the input parameter:
S t = Cx p through ( x min , S min ) and ( x mean , S mean ) yields ( p min , C min )

S t = Cx p
through ( x max ,S max ) and ( x mean , S mean ) yields ( p max , C max )
in which:
xmin, xmean, xmax = minimum, reference and maximum input variable;
Smin, Smean, Smax = minimum, reference and maximum output of sediment
transport;
pmin, pmax = power in approximation formula;
Cmin, Cmax = constant in approximation formula;
The following relations can be obtained:
⎛S ⎞ ⎛S ⎞
log⎜⎜ min ⎟⎟ log⎜⎜ max ⎟⎟
p min = ⎝ S mean ⎠ , p max =
⎝ S mean ⎠ , C min =
S mean S mean
, C max = p
⎛ 100 − VAR( X ) ⎞ ⎛ 100 + VAR( X ) ⎞ pmin
x mean x mean
max
log⎜ ⎟ log⎜ ⎟
⎝ 100 ⎠ ⎝ 100 ⎠
in which
VAR(X) = the variation of the input variable in %.
If both approximations are averaged, the following ‘powers’ and ‘constants’ are
obtained:
p min + p max C min + C max

p= and C=
2 2
The values for all input parameters are presented in Table 10-3. A value of ‘0’ means no
influence, a value of ‘1’ means a linear relation.
The most important sediment and hydrodynamic parameters will be used in the
approximation formula:
S t ;sensitivity _ analysis = 3,4 ⋅ 10 −2 H s1,7T p0,5 h −6, 4 q 4,9 ws−2, 4 d 500,5 Eq. 10-4
which is centred around Hs = 1,5 m; Tp = 7 s; h = 10 m; q = 10 m3/ms; ws = 0,0252

m/s; d50 = 200 μm and yields for this reference case St = 0,506 kg/ms. Please note that
the constant of 3,4*10-2 has a rather strange unit of about kg s m-4,2 and has no
physical base. When comparing this formula with the formula found by Jenniskens, it
appears that all signs are similar and the power of the wave height is exactly the same.
The sensitivity to water depth and discharge is somewhat larger for this case. The
influence of the sediment size is not so transparent, because of the presence of the
particle fall velocity. However, if one assumes a linear relation between ws and d50 (see
Figure 3-1) the power of d50 becomes -1,9, whereas Jenniskens found -1,75.
142
Because all ‘powers’ resemble ‘Jenniskens’ powers’ reasonably, the application of his
formula should yield values for total sediment transport of at least the same order:
St;Jenniskens= 1,10 kg/ms, which differs about a factor 2.
Although Jenniskens has calibrated his formula with over 2500 results and expression
10-4 is only calculated around the reference case, it still yields a rather good sediment
transport formula for this particular case and demonstrates the sensitivity to the various
input parameters well. It should be evident that it makes no sense applying this
strongly schematized formula in other conditions.
Table 10-3: Powers of sensitivity of total sediment transport to 15 input variables [-]
power p [-]
symbol var [%] Sb Ss St
a 40 0,0 -0,6 -0,5
ν 18 -0,5 -0,7 -0,7
φ 15 0,0 0,0 0,0
ρs 2 -1,2 -1,1 -1,1
αs 90 0,0 0,0 0,0
Hs 20 0,8 1,8 1,7
Tp 15 0,8 0,4 0,5
Q 10 3,1 5,1 4,9
h 5 -4,1 -6,7 -6,4
ws 25 0,0 -2,7 -2,4
d50 25 0,3 0,5 0,5
d90 25 0,5 0,2 0,3
n 13 0,0 0,0 0,0
CA 30 0,0 1,0 0,9
SB 30 1,0 0,0 0,1
αc-w 20 -0,1 0,0 0,0
kr;c 67 -0,2 0,0 0,0
kr;w 67 0,1 0,3 0,3
Morphological development
Until now the sensitivity of initial sediment transport and transport gradients (t=0) to
some input parameters was discussed. This approach says little about actual
morphological development of the channel or trench in time. Therefore some relevant
parameters are studied in more detail, complete with a graph to explain the differences
in morphological development.
Particle diameters d50 and d90

Because of the occurrence of mixtures of different grain sizes and the difference in
sediment particles at different locations (upstream or local) and at different depths (at
the initial bed surface or at the bottom of the trench), the spread was assumed quite
large (25%). It appears that the particle diameter itself has very little influence, see
Figure 10-1. As was explained above, a larger sediment diameter causes a faster
backfilling of the trench. This is solely due to the fact that the sediment transport is
somewhat larger because of the larger particle volume, while the particle fall velocity is
kept constant.
143
Variation of particle diameter 'd50'

initial profile and three profiles after one month in unidirectional flow
1
-1
depth wrt surrounding bed level [m]
-3
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800
distance x [m]
initial d50 = 150 μm d50 = 200 μm d50 = 250 μm
Figure 10-1: Morphological development caused by variation of d50 with 25%
Particle fall velocity ws

Related to the grain size is the particle fall velocity. A variation of the grain sizes with
25%, also leads to a variation in particle fall velocities of about 25%.
Variation of particle fall velocity 'ws'
1
-1
-3
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800
distance x [m]
initial ws = 0,0189 m/s ws = 0,0252 m/s ws = 0,0315 m/s
Figure 10-2: Morphological development caused by variation of ws with 25%
In Figure 10-2 one can see that this parameter is very influential. The smaller the
particle fall velocity, the more sediment particles will stay in the water column. When
this larger amount of suspended sediment transport travels over the trench, most of the
sediment will have enough time to settle and a rapid silting up of the channel occurs.
Also can be seen that the deposited particles with a large particle fall velocity stay close
to the upstream slope, leaving the initial bottom of the trench almost unaffected. Only
shifting of the upstream slope occurs. At the downstream slope, the accelerating flow
causes less erosion.
144
The particles with a small particle fall velocity are spread over a wide area and the
navigable depth is reduced over the entire channel width. The net sedimentation (within
the initial channel boundaries) differs between 483 and 1138 m3/m.
A sound determination of the particle fall velocity therefore seems to be inevitable, but
will cause some problems in engineering practice because of the mixtures of sediment
and differences between local sediment particles and sediment particles carried from
further on.
Porosity of bed material n

Often little is known about the porosity of the initial bed, but even more questions exist
about the porosity of the sedimented material. This is extremely dependent on the
conditions, especially on the wave conditions, under which sedimentation takes place.
When varying the porosity ‘n’ between 0,35 and 0,45, the only observed changes in the
bed geometry after one month, is the vertical location of the bed. On these locations
where net sedimentation takes place (upstream slope and trench bottom), the bottom
elevation is largest for n = 0,45; on these locations where net erosion takes place (the
upper half of the downstream slope) the drop in bottom level is largest for n = 0,45. At
the upstream slope at a trench depth of about 7 m, the three profiles coincide, because
no net erosion or sedimentation occurs and the porosity ‘n’ is of no interest.
Variation of porosity 'n'
-1
-3
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800
distance x [m]
initial n = 0,35 n = 0,40 n = 0,45
Figure 10-3: Morphological development caused by variation of n with 12,5%
In Paragraph 9.3 it was mentioned that the porosity also influences sediment transport.
More densely packed sediment means a smoother seabed hence more sediment
transport. Sediment transport formulae, however, do not consider this phenomenon
implicitly. It can only be taken into account by variation of the bed roughness. In
practice both porosity and bed roughness are very hard to determine. In this thesis the
porosity is only a measure for the rate of bed level changes; the bed roughness is kept
constant.
In the future a relation between bed roughness and porosity may be developed. One
could for instance distinguish between the ‘old, more densely packed bed’ (with water
depths larger than the initial profile) and the ‘new bed’ (with depths smaller than the
initial profile).
145
Significant wave height Hs

Variation of the significant wave height affects sediment transport. Larger waves stir up
more sediment, mix it better throughout the vertical and still have impact on particles
located deeper in the channel. As a result an increase of the wave height has a
somewhat larger effect than a decrease, see Figure 10-4.
Variation of significant wave height 'Hs'

1
-1
-3
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800
distance x [m]
initial Hs = 1,2 m Hs = 1,5 m Hs = 1,8 m
Figure 10-4: Morphological development caused by variation of Hs with 20%
Although it seems that smaller waves cause steeper upstream and downstream slopes,
this conclusion cannot be drawn based on this graph. The morphological development of
the case with larger waves is just ahead of time compared to the cases with smaller
waves. During this process steeper slopes may have occurred. This phenomenon of
slope development in time will be studied in the next paragraphs.
Numerical smoothing coefficient αs

This parameter causes numerical smoothing at sharp discontinuities. Because the initial
profile shows some sharp upper corners, without numerical smoothing ‘oversteepening’
of the upstream slope may occur, resulting in unrealistic steep slopes, even steeper
than the natural slope. In Figure 10-5 one can see, that the influence of variation of this
parameter mainly affects the steepness of the slopes, while the trench volume or depth
barely changes. Also the total erosion and sedimentation on the upstream and
downstream slope is almost the same, although the amount of local
erosion/sedimentation shows some variation. Considering the three profiles, an αs of 0,5
results in realistic geometries: not too sharp, not too slanted.
146
Variation of parameter 'αs'

1
-1
-3
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800
distance x [m]
initial αs = 0,1 αs = 0,5 αs = 1,0
Figure 10-5: Morphological development caused by variation of αs with 90%
Because variation of this parameter does not affect sediment transport and the time-
scale in which the channel silts up, slopes can be compared at any moment. Many other
parameters, as was explained at the section about the significant wave height, influence
the morphological time-scale, which makes it impossible to compare slope development
at a certain time. Moreover, the smoothing coefficient will turn out to be of great
importance when considering slope development and will therefore be studied in more
detail. Another graph-format is chosen to present the slope of the seabed, see Figure
10-6.
Slope of seabed (tan β) in cross-section of channel
0,40
0,35 initial slopes
slopes after 90 days; αs = 0,1

0,30
slopes after 90 days; αs = 0,5
slope of seabed tan β
0,25 slopes after 90 days; αs = 1,0
0,20
0,15
0,10
0,05
0,00
50 150 250 350 450 550 650 750
distance x [m]
Figure 10-6: Slope of cross-channel seabed for various numerical smoothing coefficients
147
The initial side slopes are the ‘bumps’ with height 0,2, because the reference case
“navigation channel” consisted of a flat surrounding seabed, flat channel bottom and
two straight 1:5 side slopes. Because for geotechnical stability the direction of the slope
does not matter, all slopes are represented by their absolute values.
It can clearly be observed that a smaller smoothing coefficient results in sharp peaks;
the maximum upstream slope is even steeper than 1:3 for αs = 0,1. At the middle of
the navigation channel the smoothing effect has faded out and all profiles show equal
bed slopes. So, although the numerical smoothing coefficient does not change the
backfilling predictions (the main purpose of SUTRENCH), slope development is
extremely sensitive to the value of this parameter. The only way to deal with this
parameter properly is to treat it as a sort of dustbin-parameter which accounts for all
morphological processes that SUTRENCH cannot model. This is perhaps the most
important shortcoming of SUTRENCH when one is particularly interested in
morphological development of the side slopes.
Conclusions
The sensitivity analysis revealed that all relevant input parameters show logical
behaviour, which is in line with previous studies like Jenniskens [2001] and Klein
[2003]. Regarding sediment transport, the most influential parameters are the
discharge, water depth, significant wave height, particle fall velocity and the wave-
related bed roughness. Morphological (particular slope) development is also heavily
affected by numerical smoothening processes.
10.2 Simulations with unidirectional flow
10.2.1 Introduction
To study the slope development in unidirectional flow, a lot of calculations for all
reference soils have been done with varying discharges (current velocities), wave
heights and initial slopes. The results of these calculations should answer the following
questions:
- What is the influence of the current velocity on slope development and backfilling of
the trench for different soil types?
- What is the influence of the (significant) wave height on slope development and
backfilling of the trench for different soil types?
- Does the initial slope affect slope development and backfilling ratio?
- Is there something like a ‘dynamic equilibrium slope’?
- When do the steepest slopes occur? Is there a greater risk on instability in time?
- What is, starting from the idea that initial slopes are completely straight, the best
design strategy for all reference soils.
- How reliable are the presented time scales of backfilling and slope development?
All executed calculations are presented in Table 10-4, complete with simulation names.
‘F-types’ are simulations with different initial side slopes; ‘E-types’ are simulations with
different current discharges; ‘H-types’ are simulations with different wave heights and
periods. The different simulations will be judged on the following criteria based on the
geometries of both side slopes and the total channel. In this situation of unidirectional
flow, it is useful to distinguish between upstream and downstream slopes:
- Upstream slope after 90 days;

- Downstream slope after 90 days;
- Maximum depth reduction [%] defined as the maximum depth after 90 days divided
by the initial channel depth;
148
- Width reduction [%] defined as the width after 90 days at which 90% of the initial
design channel depth is present divided by the initial channel width at which 90% of
the initial design channel depth is present. This depth of 90% of the initial depth is
chosen such that very small siltation and gravity effects on the grain particles on a
slope will not affect the results too much. In reality there will always be some initial
overdepth, so 90% seems reasonable.
- Remaining channel volume [%] according to Figure 10-7 defined as the channel
volume after 90 days within the initial channel, which can be significantly smaller
than the total channel volume after 90 days due to channel migration. The depth
above the downstream slope increases because of slope flattening, but this extra
depth will not be useful: a large area with a relatively small depth increase.
These criteria are formulated from a practical point of view, as was considered in
Paragraph 8.5.
The above mentioned criteria together will give a good idea on problems to be expected
with slope stability, dredging frequency and effective measurements to improve the
channel geometry.
However, one should note that, because of the above mentioned logarithmic velocity
profile, the model can yield slightly too steep upstream and downstream slopes if the
initial slopes are steeper than 1:5, as was found in validation of SUTRENCH with
laboratory experiments by Havinga (1992) [Walstra et al., 2000]. The acceleration on
the downstream slopes causes higher near bed velocities than is predicted by the
logarithmic velocity profile. The sediment picking up capacities will be somewhat larger,
resulting in flatter slopes. Nevertheless steeper slopes are modelled in this thesis to be
able to compare the magnitude of these differences.
3
slope migration + slope erosion +
slope steepening not useful depth increase
depth wrt to surrounding seabed [m]
Useful
-1
Volume:
-3 remaining
volume
within initial
-5 trench
boundaries
-7
-9
-11
0 100 200 300 400 500 600 700 800 900 1000
distance along x-axis [m]
Initial Profile after 30 days after 60 days after 90 days
Figure 10-7: Schematization of remaining volume between initial channel boundaries (output
simulation ‘E04’)
149
Table 10-4: Variation of different parameters; ‘grey’ cells illustrate the changes
Initial Current Wave Wave

Name
Slope Discharge Height Period
- m3/s/m m s
F06 1:2 5 1,5 6
Medium packed Coarse Sand F10 1:3 5 1,5 6
F14 1:4 5 1,5 6
F02 1:5 5 1,5 6
E03 1:5 7,5 1,5 6
E04 1:5 10 1,5 6
H03 1:5 5 0,5 5
H04 1:5 5 2,5 7
F18 1:6 5 1,5 6
F22 1:7 5 1,5 6
F26 1:8 5 1,5 6
F07 1:2 5 1,5 6
Medium packed Medium Sand
F11 1:3 5 1,5 6

F15 1:4 5 1,5 6
F03 1:5 5 1,5 6
E05 1:5 7,5 1,5 6
E06 1:5 10 1,5 6
H05 1:5 5 0,5 5
H06 1:5 5 2,5 7
F19 1:6 5 1,5 6
F23 1:7 5 1,5 6
F27 1:8 5 1,5 6
F05 1:2 5 1,5 6
F09 1:3 5 1,5 6
Medium packed Fine Sand
F13 1:4 5 1,5 6

F01 1:5 5 1,5 6
E01 1:5 7,5 1,5 6
E02 1:5 10 1,5 6
H01 1:5 5 0,5 5
H02 1:5 5 2,5 7
F17 1:6 5 1,5 6
F21 1:7 5 1,5 6
F25 1:8 5 1,5 6
F08 1:2 5 1,5 6
F12 1:3 5 1,5 6
F16 1:4 5 1,5 6
Medium packed Silt
F04 1:5 5 1,5 6

E07 1:5 7,5 1,5 6
E08 1:5 10 1,5 6
H07 1:5 5 0,5 5
H08 1:5 5 2,5 7
F20 1:6 5 1,5 6
F24 1:7 5 1,5 6
F28 1:8 5 1,5 6
150
10.2.2 Variation of initial side slopes
To study the existence of some sort of ‘equilibrium slope’, simulations with different
initial slopes have been done for four different reference soils: MCS, MMS, MFS and MSI.
Will these slopes evolve to the same slope? All other input parameters were kept
constant, but the initial slopes were varied from 1:8 to 1:2. By way of illustration, the
channel development in MFS is plotted in Figure 10-8. As can be seen, the slopes are
slightly flattening and the water depth at the middle of the channel bottom remains
constant. The downstream slope clearly starts to become flatter than the upstream
slope and the entire channel starts to migrate in downstream direction.
5
3
depth wrt surrounding bed (-10m)
Current
1
-1
Upstream Slope
-3
Downstream Slope
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800 900 1000
initial geometry after 30 days after 60 days after 90 days
Figure 10-8: Developing slopes in unidirectional flow (u = 0,5 m/s) in MFS (output simulation
‘F01’)
As can be seen from Figure 10-9, where slope development is plotted for different initial
slopes in MCS, the upstream slopes are converging, but with a slight tendency to flatter
slopes. At t = 0 days, all initial slopes are plotted on the y-axis. After 90 days in
unidirectional flow the upstream slopes are all in the order of 1:6 to 1:8. The
downstream slopes, which have not been plotted show a similar behaviour with slopes
after 90 days in the order of 1:8 to 1:10. Remarkably all slopes flatter than 1:4 first
show steepening behaviour, which can be important for slope stability.
151
Development of upstream slopes in time for different initial slopes in MCS
7
Upstream slope 1:y [-]
0
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-9: Slope development in unidirectional flow in MCS
When the grain size and the particle fall velocity are reduced (MMS), it appears that the
upstream slopes become somewhat steeper, see Figure 10-10
Figure 10-10: Slope development in unidirectional flow in MMS
. After 90 days most upstream slopes are in the order of 1:6 to 1:7, downstream slopes
1:8 to 1:10.
Development of upstream slopes in time for different initial slopes in MMS
7
0
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-10: Slope development in unidirectional flow in MMS
Slope development in MFS shows similar behaviour: the upstream slopes become
somewhat steeper (1:5 to 1:6), the downstream slopes somewhat flatter (1:9 to 1:11),
see Figure 10-11. Again the upstream slopes will steepen, at least in the beginning
except the ‘1:2 initial slope’.
152
Development of upstream slopes in time for different initial slopes in MFS
7
0
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-11: Slope development in unidirectional flow in MFS
The finest sediment under consideration, MSI, is transported in such large amounts
under these flow conditions that both side slopes soon start to ‘feel’ each other. Because
of the large sedimentation over the entire trench bottom, due to the very small fall
velocity of the silt particles, slope flattening occurs very fast and is accelerated (at
about t = 60days) when the toes of both side slopes start to coincide, see Figure 10-12.
As a result the upstream slope has become 1:22 and the downstream slope 1:33 with a
remaining channel depth of only 3,6 m, which can be considered as ‘failure of the
channel’.
Development of upstream slopes in time for different initial slopes in MSI
25
20
15
10
0
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-12: Slope development in unidirectional flow in MSI
Summarizing it can be concluded that under these flow conditions steepest upstream
slopes are to be expected in fine sands (MFS). This soil combines the property of easily
being picked up by current and waves and therefore transported in significant amounts
and the property of a relatively large particle fall velocity, so grains start settling on the
slope as soon the current passes over the channel, unlike silt which settles over the
entire channel width.
153
10.2.3 Variation of unidirectional flow velocity
The development of side slopes in the previous paragraph was considered under just
one hydrodynamic condition. Will this development be affected by other current
velocities or wave heights? Now, simulations with three depth-averaged current
velocities (0,5, 0,75 and 1,0 m/s) will be done. All of the current velocities under
investigation are able to pick up at least some sediment and are representative for
marine conditions. Larger flow velocities do occur, but mainly in tidal inlets, estuaries or
closures. Eight simulations have been done, all with initial slopes of 1:5, and have been
compared with the corresponding runs from the previous paragraph.
In Figure 10-13 can be seen that upstream slopes remain rather steep, while
downstream slopes flatten significantly. In this MFS some reduction of the maximum
channel depth and a little shift in downstream direction take place, which is an
indication that a fair amount of total transport consists of suspended transport, which
settles down not until the flow has reached the channel bottom. The upstream slope
after 30 days seems to be steeper than the other plotted slopes, while the slope height
is only slightly reduced.
Furthermore, the backfilling within the initial channel boundaries is plotted with respect
to the initial channel geometry, which gives a good idea of the locations where
sedimentation takes place.
Bed geometry and backfilling with depth-averaged current velocity of 0,75 m/s in MFS
11
Depth wrt to surrounding bed (-10 m) [m]
9
7
5
3
1
-1
-3
-5
-7
-9
-11
0 100 200 300 400 500 600 700 800 900 1000
initial profile after 30 days after 60 days after 90 days backfilling after 90 days
Figure 10-13: Development of bed geometry in time with depth-averaged current velocity of
0,75 m/s in MFS.
From Figure 10-14 can be concluded that an increase of the current velocity influences
slope stability unfavourably in case the sediment diameter is in the order of 1 mm
(MCS). Upstream slopes get steeper, if this velocity becomes larger than 0,75 m/s. One
should note, that resulting slopes steeper than 1:2 may not be realistic, because of the
existence of turbulent eddies behind the upstream slope and the already mentioned
effect of increased bed-shear stresses on steep side slopes (see 10.2.1).
Besides, other instability mechanisms may already have caused slope failure, which
SUTRENCH does not take into account. This observation is in line with observed failures
in nature, where large supply of sediment by a river can cause slopes of its delta to
collapse.
154
Slope development in time for MCS under different current velocities
16
14
12
10
Slope 1:y [-]
0
0 10 20 30 40 50 60 70 80 90
Time [days]
u = 0,5 m/s; upstream u = 0,75 m/s; upstream u = 1,0 m/s; upstream
u = 0,5 m/s; downstream u = 0,75 m/s; downstream u = 1,0 m/s; downstream
Figure 10-14: Slope development in unidirectional flow in MCS for different current velocities
The same behaviour can be observed with smaller diameters, see Figure 10-15 and
Figure 10-16: for higher current velocities upstream slopes become steeper and
downstream slopes flatter.
Slope development in time for MMS under different current velocities
16
14
12
10
Slope 1:y [-]
0
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-15: Slope development in unidirectional flow in MMS for different current velocities
155
Slope development in time for MFS under different current velocities
16
14
12
10
Slope 1:y [-]
0
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-16: Slope development in unidirectional flow in MFS for different current velocities
The dip in the angle of inclination of the upstream slope for 1,0 m/s in Figure
10-16Error! Reference source not found. is caused by infilling of the channel, which
reduces the channel depth. This infilling becomes so large for MSI that it is only
meaningful to speak about independent slope development in the first 30 days, see
Figure 10-17. Because of the small fall velocity of silt, sediment is spread over the
entire trench, which means that slope steepening is never to be expected.
Slope development in time for MSI under different current velocities
20
18
16
14
Slope 1:y [-]
12
10
8
6
4
2
0
0 10 20 30 40 50 60 70 80 90
Time [days]
u = 0,5 m/s; upstream u = 0,75; upstream
u = 0,5 m/s; downstream u = 0,75; downstream
Figure 10-17: Slope development in unidirectional flow in MSI for different current velocities
A plot of the maximum slopes that do occur during this period of 90 days against grain
diameters (Figure 10-18) shows that upstream slopes get steeper for larger current
velocities and that the grain diameter for which the steepest slope occurs increases for
increasing current velocities. This can be explained by the fact that weak currents
hardly move any large sediment, but stronger currents are able to move more and more
sediment, because of the strong nonlinear relationship between velocity and transport.
156
Maximum upstream slopes against grain diameters

for different current velocities
6
within a period of 90 days 1:y [-]
5
Maximum upstream slope
0
0 100 200 300 400 500 600 700 800 900 1000
Grain diameter d50 [μm]
u = 0,5 m/s u =0,75 m/s u = 1,0 m/s
Figure 10-18: Maximum upstream slopes within a period of 90 days against grain sizes
While sediment transport is increasing, the fall velocity remains the same, resulting in
large sedimentation as soon as the current enters the channel and slows down, and
consequently building up steep slopes, see Figure 10-19 for the large difference in
sediment transport. The gradient of this graph is a measure for erosion and
sedimentation. The sediment transport capacity is reduced to zero at the upstream side
slope for the largest sediment (MCS), because suspended transport has a negligible
contribution to total sediment transport, which therefore reacts instantaneously to the
changing bed level. MFS shows a rapid drop in total sediment transport at the upstream
side slope, caused by changing bed load and suspended transport, and a more gradual
decrease at the channel bottom, caused by the remaining suspended sediment
transport.
0,6
0,5
total sediment transport St [kg/ms]
0,4
MFS; q = 5 m2/s
0,3 MFS; q = 10 m2/s
MCS; q = 5 m2/s
0,2 MCS; q = 10 m2/s
0,1
0,0
0 100 200 300 400 500 600 700 800 900 1000
Figure 10-19: Total sediment transport for different current discharges in MCS and MFS
157
10.2.4 Variation of significant wave height
The same procedure is followed for variation of the wave height. When wave stirring
increases, sediment transport increases. As can be concluded from Figure 10-20, Figure
10-21 and Figure 10-22 downstream slopes for all sediments become flatter as the
wave height increases.
Slope development in time for MCS under different wave conditions
12
10
8
Slope 1:y [-]
0 10 20 30 40 50 60 70 80 90
Time [days]
Hs = 0,5 m; upstream Hs = 1,5 m; upstream Hs = 2,5 m; upstream
Hs = 0,5 m; downstream Hs = 1,5 m; downstream Hs = 2,5 m; downstream
Figure 10-20: Slope development in unidirectional flow in MCS for different wave heights
The development of the upstream slopes depends on the wave height and soil type.
Wave heights larger than approximately 2 m cause significant slope steepening,
especially for grain sizes between 200 and 500 μm.
Slope development in time for MMS under different wave conditions
14
12
10
Slope 1:y [-]
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-21: Slope development in unidirectional flow in MMS for different wave heights
158
Slope development in time for MFS under different wave conditions
18
16
14
12
Slope 1:y [-]
10
0 10 20 30 40 50 60 70 80 90
Time [days]
Figure 10-22: Slope development in unidirectional flow in MFS for different wave heights
For MSI opposite behaviour is again observed: larger wave heights cause rapid siltation
and therefore flatter slopes. Another remarkable conclusion is that in absence of large
waves, the development of upstream and downstream slopes is very similar for all soil
types under consideration: all side slopes tend to develop to a slope of about 1:7,5.
These phenomenon is coincidental, however can be explained. In absence of waves
slope development is only governed by the unidirectional current. The upstream slope
develops slowly to a relative flat slope, because of sedimentation and trench migration,
while the downstream slope is hardly eroded, because the grains on the downstream
slope experience a constant favourable influence of gravity without the cyclic wave-
induced bed-shear stresses.
Slope development in time for MSI under different wave conditions
35
30
25
Slope 1:y [-]
20
15
10
0 10 20 30 40 50 60 70 80 90
Time [days]
Hs = 0,5 m; upstream Hs = 1,5 m; upstream
Hs = 0,5 m; downstream Hs = 1,5 m; downstream
Figure 10-23: Slope development in unidirectional flow in MSI for different wave heights
159
The above four graphs are again summarized in one graph, see Figure 10-24. For most
wave heights a slow flattening of the side slopes occurs (please note that the drawn
lines represent the steepest upstream slope, which can be the initial slope of 1:5), but
larger wave heights can result in steeper slopes. In the case of wave heights larger than
2,5 m, probably even steeper slopes can develop in sediments coarser than 500 μm,
although the impact of bigger waves will become deeper and deeper, resulting in
erosive activities along the side slopes.
Maximum upstream slopes against grain diameters

for different wave heights
6
5
Maximum upstream slope
0
0 100 200 300 400 500 600 700 800 900 1000
Hs = 0,5 m Hs = 1,5 m Hs = 2,5 m
Figure 10-24: Maximum upstream slopes within a period of 90 days against grain sizes
160
10.3 Simulations with simple tidal flow
10.3.1 Introduction
In most natural situations tidal flow will occur instead of unidirectional flow. In this case
trench migration will be much smaller and the difference between the so-called
upstream and downstream slopes will be less significant and can be zero if the tidal flow
is completely symmetrical. In this paragraph the tidal flow will be schematized to a
simple block function, see Figure 10-25.
Tidal current velocity
-2 0 2 4 6 8 10 12 14
Time t [hr]
Figure 10-25: Block function of tidal current velocity [dimensionless]
Because of the equal absolute ebb and flood current velocities, there will be no
difference between upstream and downstream slopes and no trench migration. The
morphological changes will also be smaller under tidal conditions due to the turning
current and sediment transport; therefore the simulation period is extended to 300 days
(which is almost the entire period of interest of this thesis). The slope development
during this period will be observed every 30 days to find out what the critical stage for
every condition and sediment is.
Again calculations for different side slopes (1:3, 1:5 and 1:7), different (tidal) current
velocities (u = ± 0,5, 0,75 and 1,0 m/s) and different wave heights (Hs = 0,5, 1,5 and
2,5 m) will be done for four non-cohesive reference soils (MCS, MMS, MFS and MSI).
As was explained in the sensitivity analysis, ‘numerical smoothening’ is important in

case of very steep profiles or sharp discontinuities; otherwise instabilities will occur in
the bottom profile or sediment transport, which can hardly be prevented by decreasing
the time step Δt. In unidirectional flow, the problem of ‘oversteepening’ of the upstream
slope asked for a rather large smoothening coefficient (α = 0,5). In tidal flow the
morphological development of the trench is much more symmetrical, allowing the
application of less numerical smoothening. A lot of calculations with different space and
time steps and smoothening coefficients showed that with space step Δx = 2m, time
step Δt = 3 hr and α = 0,01 no instabilities occur and the representation of the
development of the trench profile is sufficient accurate: smaller numerical smoothening
causes instability, while the profile shows hardly any changes, when the ‘instability
peaks’ are neglected. The main part of the side slopes is not very sensitive to a
(further) decrease of the smoothening coefficient; the steepness also remains the same.
Only at the top and toe of the slope small differences can be observed, which are not
relevant when considering slope stability.
161
Table 10-5: Variation of different parameters in simulations with simple tidal flow
Initial Current Wave Wave

Name
Slope Discharge Height Period
- m3/s/m m s
DC1 1:5 5 1,5 6
Medium packed DC2 1:3 5 1,5 6
Coarse Sand DC3 1:7 5 1,5 6
DC4 1:5 7,5 1,5 6
DC5 1:5 10 1,5 6
DC6 1:5 5 0,5 5
DC7 1:5 5 2,5 7
DM1 1:5 5 1,5 6
Medium packed
Medium Sand
DM2 1:3 5 1,5 6

DM3 1:7 5 1,5 6
DM4 1:5 7,5 1,5 6
DM5 1:5 10 1,5 6
DM6 1:5 5 0,5 5
DM7 1:5 5 2,5 7
DF1 1:5 5 1,5 6
Medium packed
DF2 1:3 5 1,5 6

Fine Sand
DF3 1:7 5 1,5 6

DF4 1:5 7,5 1,5 6
DF5 1:5 10 1,5 6
DF6 1:5 5 0,5 5
DF7 1:5 5 2,5 7
DS1 1:5 5 1,5 6
Medium packed
DS2 1:3 5 1,5 6

DS3 1:7 5 1,5 6
Silt
DS4 1:5 7,5 1,5 6

DS5 1:5 10 1,5 6
DS6 1:5 5 0,5 5
DS7 1:5 5 2,5 7
10.3.2 Variation of initial side slopes
A number of 12 simulations have been done with initial side slopes of 1:3, 1:5 and 1:7
and a constant depth-averaged tidal flow velocity of 0,5 m/s and a wave height of 1,5
m. Because of the turning current the side slopes also will undergo a turning
development, for example during flood steepening and during ebb flattening of the
slope. Slope geometries are therefore expected to develop slower compared to
unidirectional flow and slope angles will be somewhere between the upstream values
(upper boundary) and downstream values (lower boundary) of the previous paragraph.
All simulations were done with Δx = 2 m, Δt = 3 hr and α = 0,01, except for the 1:3
and 1:5 side slopes in MMS. These simulations showed small signs of instability at the
top of the slope. Therefore the time step was halved to 1,5 hr. The results then showed
slopes very similar to the initial simulations with the larger time step, but the
instabilities disappeared.
It appears that the channel development within one year remains limited to the side
slopes for sediments larger than 200 μm (MCS and MMS) under these hydrodynamic
conditions. The channel depth remains constant, because adaptation of the sediment
profile is restricted to the slopes because of the large particle fall velocity and the small
total sediment transport, which is mainly in the form of bed load transport (about 80%
for MMS and over 99% for MCS).
162
In Figure 10-26, it can be seen that there is some sedimentation (±16 cm) on the
channel bottom in MFS and only minor slope development. At first sight, it can be
observed that all initial slopes are moving towards a dynamic equilibrium slope: the
1:3-slope is flattening, the 1:7-slope is steepening and the 1:5-slope stays more or less
the same, indicating that a side slopes of 1:5 is approximately the dynamic equilibrium
slope for these conditions (water depth, tidal current and wave height). This
morphological behaviour can completely be explained by the distribution of the total
sediment transport into bed load and suspended transport. The bed load transport
diminishes when passing over the slope. At the toe of the slope, there is only little bed
load transport; when the seabed becomes flat at the channel bottom, there is only
suspended sediment transport left, resulting in a small sedimentation on the channel
bottom.
development of channel geometry within 300 days in MFS for different initial slopes
1
depth wrt to surrounding seabed (MSL-10m) [m]
-1
-3
-5
-7
-9
-11
200 250 300 350 400 450 500 550 600
initial slope 1:3 initial slope 1:5 initital slope 1:7
Figure 10-26: Development of channel geometry for three initial slopes in tidal flow in MFS
Smaller grain particles like MSI show complete silting up of the channel within one year.
Because of the small particle fall velocity, slopes start to flatten instantaneously and the
sedimentation pattern extends over the entire width. Sediment transport is almost
100% in the form of suspended transport.
When varying the side slopes in MCS, very slow slope development was observed due to
the rather small sediment transport capacity (Stot = 0,69*10-2 kg/ms), see Figure
10-27. Extrapolating the three lines results in an equilibrium slope of about 1:4,5 to 5.
Required channel depth is guaranteed for years; slope stability problems are not to be
expected. After all, morphological timescales are large, sedimentation remains limited,
so good wave compaction can be reached and liquefaction problems are not to be
expected, because of the large grain sizes, good permeability and consolidation
properties and large relative densities. Rapid slope steepening does not occur and shear
failure is not to be expected. Steep initial slopes like the 1:3-slope will slowly be
flattened by waves and currents, so the initial situation will be normative.
163
Development of side slopes in time for different initial slopes in tidal flow in MCS
6
Side slopes 1:y [-]
0
0 50 100 150 200 250 300
Time [days]
initial slope 1:3 initial slope 1:5 initial slope 1:7
Figure 10-27: Slope development for initial slopes 1:3, 1:5 and 1:7 in tidal flow (MCS)
In medium sands (MMS) the situation becomes slightly different:

- sediment transport capacity is almost doubled ((Stot = 0,013 kg/ms);
- not only bed load transport, but also suspended transport (± 20% at the
surrounding seabed) occurs;
- particle fall velocities become smaller.
This results in a faster morphological development judging by the faster converging
lines in Figure 10-28. Also the ‘width of influence’ extends: the sediment concentration
vertical is ‘emptied’ over the entire slope. The main consequence is a small flattening of
the slope to values of 1:6 to 7. The sedimentation on the channel bottom remains
limited to a few dm’s on the sides to zero in the middle. Again the normative situation is
the initial slope. Medium sands become more susceptible to (cyclic) liquefaction,
primarily depending on relative density.
Development of side slopes in time for different initial slopes in tidal flow in MMS
6
Side slopes 1:y [-]
0
0 50 100 150 200 250 300
Time [days]
Figure 10-28: Slope development for initial slopes 1:3, 1:5 and 1:7 in tidal flow (MMS)
164
When the grain sizes become smaller (MFS), the following can be observed:
- sediment transport capacity is increased (Stot = 0,024 kg/ms);
- particle fall velocities become smaller;
- a significant part of the total sediment transport is suspended transport (72%).
Following the same line of thought, a further slope flattening would be expected. The
simulations, however, show a different behaviour: slopes in MFS are steeper than in
MMS. Equilibrium slopes are in the order of 1:4, even steeper than in MCS. Slope
development is dependent on the gradient of sediment transport. So, although the
sedimentation width extends over the entire upstream side slope and the channel
bottom, due to the high contribution of suspended transport, the local gradient at the
upstream slope still is larger than for MCS and MMS because of the larger sediment
transport at the surrounding bed.
Development of side slopes in time for different initial slopes in tidal flow in MFS
6
Side slopes 1:y [-]
0
0 50 100 150 200 250 300
Time [days]
Figure 10-29: Slope development for initial slopes 1:3, 1:5 and 1:7 in tidal flow (MFS)
165
A further decrease of the grain size to 50 μm (MSI) again shows different behaviour,
see Figure 10-30. The silting up proceeds so fast that it is almost impossible to draw
conclusions on autonomous slope development. Due to the high contribution of
suspended transport to total sediment transport, the actual angle of inclination of the
side slopes hardly affects sediment transport.
Development of channel geometry in time in tidal flow in MSI
5
Depth wrt surrounding bed (MSL-10m)
-1
-3
-5
-7
-9
-11
200 300 400 500 600
initial geometry after 30 days after 60 days after 90 days after 150 days after 300 days
Figure 10-30: Morphological development of channel geometry in MSI
In Figure 10-31 (please note the different y-scale compared to the other reference soils)
can be seen that during the first three months, the simulations with different side slopes
show some differences, but after about 90 days an accelerated slope flattening takes
place. From that moment slope heights are drastically reduced and both side slopes
start to interfere with each other.
Development of side slopes in time for different initial slopes in tidal flow in MSI
20
19
18
17
16
15
14
Side slopes 1:y [-]
13
12
11
10
9
8
7
6
5
4
3
2
1
0
0 50 100 150 200 250 300
Time [days]
Figure 10-31: Slope development for initial slopes 1:3, 1:5 and 1:7 in tidal flow (MSI)
166
In MSI, it is completely irrelevant to speak of dynamic equilibrium slopes and the design
of an appropriate slope should be based on other considerations than morphological
development of the side slopes:
- which slope angle is safe from geotechnical point of view, while taking into
consideration the rather short period this slope will have to persevere under local
hydrodynamic conditions?
- which channel geometry is able to reduce overall sedimentation within the channel
boundaries and to maintain required navigable depths as long as possible?
In this respect it is useful to consider trapping efficiencies of the three channel

geometries, see Table 10-6. Flatter slopes result in larger upper widths of the trench, if
the bottom width remains the same, and therefore larger trapping efficiencies and initial
capital dredging volumes; starting from side slopes of 1:5, an increase of the slope
steepness to 1:3 brings along a decrease of 8% in capital dredging volume, a decrease
of the slope steepness to 1:7 results in an increase of 8%. These percentages of course
rise if required channel bottom widths get smaller. So, designing flatter slopes improves
slope stability, but increases construction costs.
Table 10-6: Sediment transport and trapping efficiency of three channel geometries
initial sediment transport at minimum sediment transport trapping average rise of sedimentation on
slope surrounding seabed inside channel efficiency channel bottom channel bottom
tan(β) Stot(0) Sb(0) Ss(0) Stot(0) Sb(0) Ss(0) e after 30 days after 30 days
[-] [kg/ms] [%] [%] [kg/ms] [%] [%] [%] [m] [%]
1:3 2,12E+00 0,06% 99,94% 1,87E+00 0,00% 100,0% 11,9% 2,84 94,0%
1:5 2,12E+00 0,06% 99,94% 1,81E+00 0,01% 99,99% 14,8% 2,55 86,3%
1:7 2,12E+00 0,06% 99,94% 1,76E+00 0,02% 99,98% 17,1% 2,37 80,5%
From this table and also from Figure 10-32 can be concluded that a larger trapping
efficiency does not inevitably result in a larger rise of the channel bottom. This can be
explained by the fact that gentle slopes, which extend over a larger width, act
themselves as sediment trapping reservoir: the actual sedimentation on the channel
bottom as a percentage of the total sedimentation within the channel boundaries can be
reduced from 86,3% to 80,5%, when flattening the side slopes from 1:5 to 1:7. On the
other hand, the capital dredging volume is increased by 8% and the real advantage in
time is only about 4 days. Variation of the side slopes is therefore not recommended for
fine sediments, see Chapter 11. Keeping the upper width as small as possible and
dredging of an overdepth seems the best construction method for sediments smaller
than 200μm.
Time [days]
0 50 100 150 200 250 300
0
Channel bottom level wrt surrounding seabed (MSL-10m)
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
tan β = 1:3 tan β = 1:5 tan β = 1:7
Figure 10-32: Average location of channel bottom for different initial side slopes
167
It can be seen that sediment transport in MSI is mainly suspended transport. Inside the
channel there is hardly any bed load transport. The fact that bed load transport on more
gentle slopes has a larger share in total sediment transport depends on the actual water
depth where this minimum transport takes place; this depth is smaller for more gentle
slopes, so current and wave impact on the seabed is felt better.
10.3.3 Variation of tidal flow velocity
The simulations of the previous paragraph were done for only one tidal current velocity.
Now, the influence of the current velocity is investigated by varying the constant tidal
current of Figure 10-25: utidal = 0,5; 0,75; 1,0 m/s.
The coarse sand (MCS) shows an interesting behaviour. For a tidal current velocity of
0,75 m/s, a small bend is located at a channel depth of 5 m (water depth of 15 m), see
Figure 10-33; when the tidal current velocity is further increased this bend has
disappeared and the bed geometry is smooth again (straight middle part and rounded
top and toe of the slope). This phenomenon probably has something to do with depths
of influence of both current and waves. In comparison to a depth-averaged tidal current
of 0,5 m/s, bed-load transport is tripled and suspended transport is even multiplied by
142! Nevertheless, all morphological development occurs on the side slopes due to the
still high particle fall velocity. It is likely that the immediate reacting bed-load transport
and the somewhat lagging suspended transport create such an interesting profile.
1
-1
depth wrt surrounding bed (-10m)
-3
-5
-7
-9
-11
220 230 240 250 260 270 280 290 300 310 320
initial geometry after 100 days after 200 days after 300 days
Figure 10-33: Slope development in time of MCS with a tidal velocity of 0,75 m/s
A further increase of the tidal current velocity up to 1,0 m/s raises the share of
suspended transport to 50%, while the bed load transport reaches to the channel
bottom. The transition from hardly any sediment transport to a considerable transport
complicates drawing conclusions. However, it is certain that slope angles will not alter
too much, see Figure 10-34.
168
Development of side slopes in time for different tidal current velocities in MCS
4
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
u = 0,5 m/s u = 0,75 m/s u = 1 m/s
Figure 10-34: Slope development in tidal flow in MCS for different current velocities
The medium sand (MMS) shows a familiar behaviour: an increase of the current velocity
results in steeper side slopes, see Figure 10-35.
Development of side slopes in time for different tidal current velocities in MMS
4
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
u = 0,5 m/s u = 0,75 m/s u = 1 m/s
Figure 10-35: Slope development in tidal flow in MMS for different current velocities
In MFS the same slope steepening process is observed, until the channel starts to silt
up. In the largest tidal current (1,0 m/s), this silting up-process is already noticeable
after 60 days, see Figure 10-36, resulting in a rapid slope flattening.
169
Development of side slopes in time for different tidal current velocities in MFS
10
7
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
u = 0,5 m/s u = 0,75 m/s u = 1 m/s
Figure 10-36: Slope development in tidal flow in MFS for different current velocities
As was mentioned before, homogeneous silts are that easily movable by even mild
currents that one cannot speak of autonomous slope development. After respectively
30, 60 and 90 days, the channel is almost completely silted up, see Figure 10-37.
Development of side slopes in time for different tidal current velocities in MSI
30
25
20
side slope 1:y [-]
15
10
0
0 50 100 150 200 250 300
time [days]
u = 0,5 m/s u = 0,75 m/s u = 1 m/s
Figure 10-37: Slope development in tidal flow in MSI for different current velocities
Summarizing all of the above simulations leads to the following conclusions, see Figure
10-38:
- It is demonstrated that larger tidal current velocities cause larger slope steepening
just like in unidirectional flow, although this steepening process progresses more
slowly and is less extreme than in unidirectional flow;
- Fine to medium sands are most susceptible to this slope steepening. Coarse sands
are hardly transported and very fine sands and silts are transported in large
amounts, primarily as suspended transport. The sedimentation pattern is regularly
spread out over the entire channel bottom and one cannot speak of autonomous
slope development.
170
Maximum side slopes against grain diameters for different current velocities
6
5
Maximum side slope
0
0 100 200 300 400 500 600 700 800 900 1000
u = 0,5 m/s u =0,75 m/s u = 1,0 m/s
Figure 10-38: Maximum side slopes within a period of 300 days against grain sizes
10.3.4 Variation of significant wave height
Also the influence of the (constant) significant wave height has been examined for all
four reference sediments. The wave height was varied as follows: Hs = 0,5;1,5;2,5 m.
The morphological development in coarse sand proceeds very slowly, unless the wave
height is increased to 2,5 m, see Figure 10-39. Instead of a small slope steepening as
was the case in unidirectional flow, a little slope flattening occurs. However, slope
development remains limited and shouldn’t have to be taken into account in slope
design.
Development of side slopes in time for different wave heights in MCS
4
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
Hs = 0,5 m Hs = 1,5 m Hs = 2,5 m
Figure 10-39: Slope development in tidal flow in MCS for different wave heights
171
Increasing the wave height to 2,5 m in MMS has a steepening effect (Figure 10-40).
Strange enough, little waves show a small slope steepening, while medium waves
create more gentle slopes, at least for this tidal current. In the previous paragraph it
was already shown that larger tidal currents cause steeper slopes.
Development of side slopes in time for different wave heights in MMS

6
4
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
Hs = 0,5 m Hs = 1,5 m Hs = 2,5 m
Figure 10-40: Slope development in tidal flow in MMS for different wave heights
In fine sands, the side slopes become steeper if waves become larger, although this
effect remains small, observe Figure 10-41. Due to the increasing share of suspended
transport with increasing wave height, rapid sedimentation occurs, resulting in flatter
slopes.
Development of side slopes in time for different wave heights in MFS
12
10
8
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
Hs = 0,5 m Hs = 1,5 m Hs = 2,5 m
Figure 10-41: Slope development in tidal flow in MFS for different wave heights
In MSI for the first time autonomous slope development is observed, if the wave stirring
is very small (Hs = 0,5m), see Figure 10-42. In sediment like silt almost 100% of the
sediment transport moves in suspension. Grains that are picked up by currents and
waves immediately are mixed through the vertical.
172
Trapping efficiencies are small, even for this wide navigation channel. So sedimentation
takes place over the entire channel and isn’t confined to the side slopes. To maintain
the initial side slopes, this sedimentation needs to be distributed evenly over the entire
channel profile. However, the hydrodynamic forcing diminishes with increasing depth.
Especially wave impact is reduced on the channel bottom. It is therefore not strange
that particularly small waves (Hs = 0,5m) cause a more regular sedimentation pattern,
because the influence of the water depth will be less. Also more gentle side slopes will
result in a more regular sedimentation pattern, as was already concluded in Paragraph
10.3.2: the sedimentation on the more gentle side slopes (1:7) was far more than on
the steeper side slopes (1:3).
Development of side slopes in time for different wave heights in MSI
14
12
10
side slope 1:y [-]
0
0 50 100 150 200 250 300
time [days]
Hs = 0,5 m Hs = 1,5 m Hs = 2,5 m
Figure 10-42: Slope development in tidal flow in MSI for different wave heights
So, channel development in MSI is most influenced by variation of the wave height.
Large waves are able to almost completely silt up this navigation channel within a few
days, while small waves cause a gradual siltation. If protection against waves isn’t
possible, the alignment of the channel itself seems the only appropriate measure to
withstand storm waves. A small angle between current and channel axis generates the
before mentioned self-cleansing effect. Sediment trapping reservoirs are only useful for
sediments in the range of 100-200μm; smaller sediments remain too long in suspension
due to a small particle fall velocity and therefore won’t settle in such reservoirs.
Again summarizing all of the above simulations with different wave heights, the
following conclusions can be drawn:
- The influence of varying wave height is less pronounced in tidal flow than in
unidirectional flow. This can be explained by the fact that the flow is turning and
slope steepening only proceeds when a side slope is acting as an upstream slope;
- The influence of varying wave height is less obvious than the influence of varying
current discharge. This can be explained by the fact that an increase of the wave
height not only increases wave stirring at the surrounding seabed, but also
increases the ‘depth of impact’. Bigger waves start reshaping the side slopes and
disturb the gradual sedimentation process by the current;
The above explanation can be observed in Figure 10-43. Small waves (H=0,5m) only
stir up sediment at the surrounding seabed, which is transported by the tidal current to
the upstream slope. Medium waves also stir up sediment (in larger amounts), but have
a deeper impact and interact with the side slopes at intermediate depths, resulting in
somewhat flatter side slopes. With big waves the first described process is dominant.
173
The most important conclusion should therefore be: waves affect the morphological
development of the channel as a whole to a large extent, but the consequences for the
side slopes are not that obvious. Fine to medium sands are most likely to show some
slope steepening, but due to the turning character of the flow, side slopes probably
would not become steeper than 1:4.
Maximum side slopes against grain diameters for different wave heights
5
Maximum side slope
0
0 100 200 300 400 500 600 700 800 900 1000
Hs = 0,5 m Hs = 1,5 m Hs = 2,5 m
Figure 10-43: Maximum side slopes within a period of 300 days against grain sizes for
different wave heights
174
10.4 Simulations with a real tidal and wave climate
10.4.1 Hydrodynamic input data
In order to resemble nature more and more, now some simulations with real
(schematized) tidal flow and wave climate will be done. In fact any location can be
chosen, as long as all hydrodynamic input data can be obtained at that location. These
hydrodynamic input data consist of the vertical tidal elevation, the horizontal tidal
velocity and the wave data over a complete year. The data are obtained from Noordwijk
Meetpost (52º16'26" Northern Latitude, 04º17'46" Eastern Longitude). This measuring
station is located about 10 km off the mainland coast in a water depth of MSL-18m, see
Figure 10-44.
Figure 10-44: Measuring stations in the North Sea
The (vertical) tidal elevations of the year 2003 at Meetpost Noordwijk are obtained from
www.getij.nl and plotted in Figure 10-45. In blue the highest water levels in a tidal
cycle, in red the lowest water levels are plotted. The horizontal lines represent the year-
average values.
200
150
tidal elevation wrt MSL [cm]
100
50
HWaverage = 94,5 cm
0 HW
LW
LWaverage = -63,5 cm
-50
-100
-150
1-jan 10-feb 21-mrt 30-apr 9-jun 19-jul 28-aug 7-okt 16-nov 26-dec
date in year 2003
Figure 10-45: Tidal elevations at Meetpost Noordwijk in the year 2003
175
Because in SUTRENCH only one representative tidal cycle can be entered (and not a full
year of tidal data), a tidal cycle has been selected with the average high and low water:
December 6th from 12:03u, plotted as the ‘blue’ line in Figure 10-46. In this way the
small oscillations throughout the year due to the interaction between moon and sun
(like neap and spring tide) are neglected.
The asymmetrical tide (longer ebb period) is partially caused by the small distance to
the amphidromic point (about 60km), where the influence of the moon is weakened and
other higher order components become more significant.
100 0,8
Tidal elevation at December 6th 2003, Meetpost Noordwijk
80
Average tidal current velocity at Meetpost Noordwijk 0,6
60
0,4
Tidal elevation wrt MSL [cm]
Tidal current velocity [m/s]

40
0,2
20
0 0,0
-20
-0,2
-40
-0,4
-60
-0,6
-80
-100 -0,8
0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00
Standardized time [hr:min]
Figure 10-46: Representative tidal cycle and average current velocity against normalized time
The Nautilus Digital Current Atlas (Dutch: Nautilus Digitale Stroomatlas by RIKZ)
supplies the horizontal tidal data (current velocities). The data are related to the time of
high water (HW) and have to be shifted in time to be compatible with the vertical tidal
data, see the ‘red’ line in Figure 10-46. It appears that horizontal and vertical tide are
almost in phase, especially during flood.
The representative tidal cycle, based on the above vertical and horizontal tidal data, has
to be further schematized into discrete periods, see Figure 10-47.
1,00 1,00
0,75 0,75
0,50 0,50
Tidal elevation wrt MSL [m]
Tidal current velocity [m/s]
0,25 0,25
0,00 0,00
-0,25 -0,25
-0,50 -0,50
Tidal elevation
-0,75 Tidal current velocity -0,75
-1,00 -1,00
0:00 1:00 2:00 3:00 4:00 5:00 6:00 7:00 8:00 9:00 10:00 11:00 12:00
Standardized time [hr:min]
Figure 10-47: Schematizations into 6 discrete periods for tidal elevation and current velocity
176
In this way a full (representative) tidal cycle can be represented by 6 discrete periods
with a certain water level and discharge and 2 discrete ‘transition’ periods without any
tidal elevation or current discharge between every flood and ebb period to ensure that
the simulation time and actual time are equal, see Table 10-7.
Table 10-7: Schematization of representative tidal cycle
water level discharge time step

3
period h0 [m] Q [m /ms] dt [s]
1 10,40 3,64 3600
flood
2 10,85 7,05 7200

3 10,60 5,30 5400
transition 10,00 0,00 5400
4 9,70 -3,88 5400
ebb
5 9,50 -5,70 7200

6 9,40 -3,76 7200
transition 10,00 0,00 2940
total tidal cycle 12hr19min = 44340
The wave data at Meetpost Noordwijk are obtained from www.golfklimaat.nl for a period
of ten years (1992-2001) and plotted in Figure 10-48. As wave height ‘H1/3’ has been
used, because Hs ≈ H1/3. From now on, we will only speak of significant wave height,
although the real wave data were obtained as ‘H1/3’.
Because SUTRENCH cannot process large wave input files (restricted to 20 periods with
constant (significant) wave height), a schematized set of wave input data has to be
constructed. Although it is inevitable that some information is lost during this process, it
is very important that in some manner the same amount of wave energy is put into the
morphological model. Therefore the following approach has been adopted.
Figure 10-48: Significant wave heights in a period of 10 year (1992-2001) at Meetpost

Noordwijk
First all waves of the wave data set over 10 years have been categorized into smaller
wave bins. For each bin the significant wave period and the probability have been
determined, see Figure 10-49. Now, it is questionable which value for the wave height
has to be used for each wave bin. One could use the middle of the bin (H = 0,25; 0,75;
1,25;...5,75m), but due to the distribution of the waves inside each bin towards the
peak probability somewhere inside the second bin, the wave height would be slightly
overestimated for most wave bins. This error of course can be reduced by decreasing
the bin width, but then the resulting input becomes to extensive for SUTRENCH.
Another option is to calculate the average wave height inside each bin for the 10-year
period.
177
This method neglects the fact that increasingly large waves have a more than linearly
2
increased impact. Therefore the ‘average square wave height ( H i )’ has been
calculated for each wave bin ‘i’. This approach ensures conservation of wave energy
inside each wave bin.
40%
35%
30%
Probability [%]
25%
20%
15%
10%
5%
0%
0-0,5 0,5-1,0 1,0-1,5 1,5-2,0 2,0-2,5 2,5-3,0 3,0-3,5 3,5-4,0 4,0-4,5 4,5-5,0 5,0-5,5 5,5-6,0
Significant wave height Hs [m]
Figure 10-49: Probability distribution of wave bins
Together with the probability of each wave bin a number of 12 ‘storms’ with different
durations have been constructed, see Table 10-8. If this combination of storms is used
in morphological simulations, approximately an amount of wave energy is involved that
is equal to a 10-year-averaged wave climate at Noordwijk Meetpost. Please note that
the wave heights were determined as the average of the square wave heights, while the
wave periods were averaged in the usual way. This method is quite in line with the
approach formula, deduced in Paragraph 10.1.2, based on the powers of most
important input parameters. There a power of ‘1,7’ for the wave height and ‘0,5’ for the
wave period was found.
Table 10-8: Schematization of wave height, wave period and ‘storm duration’
Wave bin Hs T Duration

cm m s hr
0 - 50 0,36 4,6 2084,1
50 - 100 0,71 5,0 3176,8
100 - 150 1,17 5,5 1798,9
150 - 200 1,62 6,0 908,2
200 - 250 2,11 6,6 446,4
250 - 300 2,58 7,2 186,5
300 - 350 3,09 7,7 100,1
350 - 400 3,56 8,2 32,4
400 - 450 4,08 8,7 17,4
450 - 500 4,70 9,3 6,3
500 - 550 5,13 9,5 2,4
550 - 600 5,51 10,8 0,6
Now, the only remaining question is the order of the storms throughout the year. One of
Postma’s conclusions [1987], see Paragraph 2.3, was that the morphological
development to an ‘equilibrium profile’ of a trench was strongly dependent on the
weather conditions; the sooner storm conditions occurred, the faster the side slopes
developed towards this ‘equilibrium profile’. It seems evident that it makes a big
difference whether a freshly dredged channel profile can slowly adapt to deteriorating
weather conditions or when the same initial profile is instantaneously swept by a storm.
- But is morphological development really influenced by the wave distribution

throughout the year?
178
When no extensive wave data are available or one just wants to make a quick
computation, it will be valuable if one can apply just one wave height instead of a
(schematized) full-year wave climate.
- But which wave height has to be chosen in a morphological simulation to be

representative for a full-year wave climate?
Soulsby (1987) gives some indication: he concluded that “the most important
contributions to the long-term sediment transport are made by fairly large (in relation
to water depth) but not too infrequent waves, combined with tidal currents between
mean neap and maximum spring tide. Weak currents and low waves in relation to water
depth give a small contribution, because their potential for sediment transport is low,
although their frequency is high. Extreme conditions also are relatively unimportant,
since their frequency is too low, although their transport potential is high.”
Based on Soulsby’s statement a wave height of about 1,5 to 2,0 m would be expected
to be representative for the entire year.
To investigate both underlined questions the following wave distributions are proposed:
I: an ever increasing wave climate, lower boundary of morphological development;

II: an ever decreasing wave climate, upper boundary of morphological development;
III: first increasing and then decreasing wave climate, most realistic case;
IV: constant year-averaged wave height;
V: constant year-averaged square wave height;
VI: constant wave height that will be fitted to case III.
Cases I and II give insight in the maximum variation in morphological development due
to the wave distribution throughout the year. Case II can be considered as a maximum
for bed level changes, because the largest waves occur when the profile is still sharp,
while in case I the storm waves occur when the channel profile is already more gentle.
Case III is considered to be most realistic. Often dredging works are executed under
calm weather conditions (summer). After about a half year the storms have intensified
(winter) and then the wind will die down again. Cases IV, V and VI should demonstrate
the validity of calculations with a constant wave height and should yield a value for this
particular wave height that approximates case III best. The first three cases are plotted
in Figure 10-50.
6
Case I
5 Case II
Case III
4
Wave height H [m]
0
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Time [hr]
Figure 10-50: Schematization of first three cases of wave distribution
179
One should keep in mind that morphological development remains a probabilistic

process. The applied schematization of the wave climate, however averaged over 10
years, is still just a random indication: there is a large variation in wave height between
different years. For instance the averaged significant wave height in 1997 was 91 cm,
while in 1998 this was 117 cm.
10.4.2 Different wave distributions
Before the final simulations were done, first the well-known problem of numerical
diffusion was examined. Because the tidal flow was reversing and relatively mild, the
need for numerical smoothening was expected to be limited. Indeed, it appeared that
even with a very small value for coefficient ‘αs’ the simulations remained stable and
showed natural behaviour. All simulations of these paragraph were executed with αs =
0,001[-].
The first question of the previous paragraph that needs to be examined was: “is the
morphological development significantly influenced by the wave distribution throughout
the year?” Therefore the morphological development within 1 year is presented for
cases I, II and III in Figure 10-51. It can be observed that after one full year of inserted
wave energy, the difference between the cases is hardly noticeable and certainly much
smaller than the uncertainty in the wave distribution, the sediment transport, the
hydrodynamics et cetera. Furthermore it appears that both slopes show some upstream
and downstream slope behaviour; the top of the slopes are flattened and rounded due
to erosion (downstream slope), while the middle and lower part remained steep and
shifted somewhat towards the centre of the channel (upstream slope). The channel
width is only slightly reduced, while the depth reduction on the channel bottom does not
exceed 1m.
0
-1
-2
depth wrt surrounding bed [m]
-3
-4
-5
-6
-7
-8
-9
-10
-11
50 150 250 350 450 550 650 750 850 950
Initial Profile Case I after 1 year Case II after 1 year Case III after 1 year
Figure 10-51: Morphological development within 1 year for cases I, II and III
Also it can be seen that the morphological development is slightly asymmetric.

Obviously the ebb current is slightly dominant over the flood current, although this is
also dependent on the tidal period in which the largest storm waves occur. Case II with
the decreasing wave height has its peak storm during flood (at t = 0), which can be
observed from the better developed left slope (acting as the upstream slope during
flood). For case I holds the opposite: storm occurs during ebb (at t = 365 days) and the
right slope has accreted more compared to case II. The described differences are
however very hard to notice on this scale.
180
So, now let us look at the slopes in more detail. In Figure 10-52 the slopes of the initial
seabed and the seabed after 1 year are plotted for the three cases. The rectangular
bumps represent the initial situation; the more slender peaks represent the situation
after 1 year. The most important conclusions are:
- maximum slope steepness does not change much and remains in the order of 1:5;
- the steepest parts occur over a smaller height which is favourable for stability,
although the density of the accreted sediment at both slopes is very hard to predict;
- both side slopes can develop autonomously, until the slopes have shifted so close to
each other that the toes of the slope start to intersect. If the side slopes become
steeper and flow separation occurs, the location of the reattachment point of the flow
determines whether side slopes can develop autonomously.
0,25
Flood Ebb
0,20
0,15
0,10
0,05
0,00
150 250 350 450 550 650 750 850
Initial Slopes Case I after 1 year Case II after 1 year Case III after 1 year
Figure 10-52: Slope of seabed (tan β) along the x-axis
So, it is proved graphically that the differences in morphological development due to

different wave distributions throughout the year remain limited and the small deviations
could be explained. But what about Postma’s statement [1987] about the time that the
side slopes needed to reach their (dynamic) equilibrium and how this time was
dependent on the time that elapses till the first storm? To explore this question slope
development throughout the year is presented in Figure 10-53. This graph looks rather
chaotic at first sight, a maze of apparently arbitrary lines, but they can be explained.
The largest slope steepening or flattening occurs during the stormy period. The steep
parts in Figure 10-53 correspond to the cases with different wave distribution. For
instance, case III has the stormy period halfway the year.
The slope development is most pronounced, when a slope is acting as an upstream
slope. This could also be concluded from the simulations with simple tidal flow; both
side slopes remained rather steep and the downstream behaviour of slope flattening
was hardly noticeable. So left slopes develop mainly during flood, while right slopes
develop mainly during ebb. Because the water level is higher during flood, only the
biggest waves are felt at the seabed. During ebb, the water level is lower and the fast
slope development is caused by a larger number of waves. This can be observed in
Figure 10-53: the quick changes of the left (flood) slopes occur in a period of 30 days,
while these changes take about 60 days for the right (ebb) slopes. Comparing these
periods with the wave distribution shows that the flood slopes are mainly developed by
wave heights of 2,11 m and larger, while ebb slopes develop in a wave climate of waves
with a minimum height of 1,62 m. These seemingly exact values only follow from the
applied classification in wave bins, as was explained above and are therefore only
indicative.
181
5,2
5,1
5,0
slope of seabed cot β
4,9
4,8
4,7
4,6
4,5
4,4
0 30 60 90 120 150 180 210 240 270 300 330 360 390
time [days]
case I; left slope case II; left slope case III; left slope
case I; right slope case II; right slope case III; right slope
Figure 10-53: Slope development of three cases throughout the year
Both the flood slope and the ebb slope develop towards their own dynamic equilibrium
value. For the flood slope this dynamic equilibrium slope (DES) is about 1:5,1, while the
ebb slope steepens somewhat to 1:4,6.
Remarkable is the fact that flood slopes steepen under smaller waves (H < 2,11m),
whereas ebb slopes tend to remain approximately 1:5 for smaller waves (H < 1,62m).
Then for the larger waves flood slopes flatten and ebb slopes steepen. This effect is
probably caused by the combination of wave height and water depth. During flood
(large water depth) the smaller waves do not affect the slopes, but only provide a
considerable amount of sediment, while during ebb these waves are involved in the
slope shaping process.
So, it seems that there is a certain wave height of influence, but a sound formulation of
this wave height is difficult: the water level is constantly changing, the wave period is
not constant, the wavelength is adapting at the side slopes et cetera. A rough criterion
for the wave height of influence can be obtained from the maximum orbital velocity at
half-depth ((h0+h1)/2). The minimum wave heights that influence slope development
induce orbital velocities at the seabed of 0,32 - 0,44 m/s at a water depth of 15 m. This
results in the following design criterion:
Hinfluential ≥ 0,13T sinh( 12 k (h0 + h1 )) Eq. 10-5
Of course, this is a very rough relation, not based on many simulations, but it explains
the observed behaviour. The location of ‘half-depth’ is based on the idea that the
steepest part of the side slope usually is situated at half-depth.
The observed values of the steepness of both side slopes should only be interpreted in a
qualitative way; the quantitative values are strongly dependent on the combination of
wave height, water depth and (as will be shown in the next paragraph) the wave
distribution.
182
10.4.3 Towards a representative wave height
The second question that rose in Paragraph 10.4.1 concerned the idea whether a full-
year wave climate could be represented by a single representative wave height. It was
suggested to run simulations with the average wave height (H=1,02m), the average of
the square wave height (H=1,23m) and finally to search for a representative wave
height by trial and error. This last simulation will be compared with case III which was
considered to be the most realistic case (construction works during summer, storms
during winter).
What strikes one first, when observing Figure 10-54, is the fact that both the simulation
with the average wave height (case IV) and the average square wave height (case V)
result in too little morphological development.
After the conclusion of the previous paragraph that the major morphological
development occurred for waves larger than 1,62 m (ebb slopes), this was to be
expected.
In case VI the wave height was varied until an acceptable graphical resemblance to case
III was obtained. This constant wave height is 1,8 m. This wave height falls in the wave
bin of 1,5-2,0 m, which has an occurrence of 10%. Wave bins representing larger wave
heights together account for 9%. This means that the representative significant wave
height of 1,8 m is not exceeded in 81% to 91% of the time.
1
0
-1
-2
-3
-4
-5
-6
-7
-8
-9
-10
-11
50 150 250 350 450 550 650 750 850 950
Initial Profile Case III after 1 year Case IV after 1 year Case V after 1 year Case VI after 1 year
Figure 10-54: Morphological development within 1 year for cases III, IV, V and VI
To get a better estimate of the exceedance in time, we have to return to the rough data
of the wave heights during 10 year at MPN. It turns out that in 87,4% of the time, the
significant wave height (in fact H1/3) does not exceed 1,8m. So, indeed a rather large
wave height has to be applied to represent a full-year wave climate. On the other hand,
this wave height is still far smaller than the design wave height for dykes or structures.
The next question that arises is whether morphological simulations with a

representative wave height yield similar results for slope development. Simulating with
one wave height instead of a complete wave distribution may be practical for the
prediction of backfilling, but inevitably some information is lost during this process. How
significant this loss in geometrical accuracy is, can be observed in Figure 10-55. The
location of the side slopes is quite similar for cases III and VI, but case VI yields steeper
side slopes, especially the ebb slope (1:3,4).
In earlier simulations was already found that side slopes can steepen significantly for a
large wave height, depending on the sediment size. Now, it turns out that a varying
wave height suppresses this behaviour. This can be explained by the fact that the very
influential, large waves reshape the side slopes due to the deep impact of these waves.
183
0,30
Flood Ebb
0,25
0,20
0,15
0,10
0,05
0,00
150 250 350 450 550 650 750 850
Initial Slopes Case III after 1 year Case IV after 1 year Case V after 1 year Case VI after 1 year
Figure 10-55: Slope of seabed (tan β) along the x-axis
Finally, slope development in time for the different cases will be studied, see Figure
10-56. Cases IV, V and VI show familiar slope behaviour: slopes steepen in time and
only the right slope of case IV then flattens again. For larger wave heights the steepest
slope will not be reached within a year. The ebb slopes become somewhat steeper than
the flood slopes. Case IV (smallest wave height) only shows steeper ebb slopes in the
first 90 days.
Anyhow, the cases with a uniform wave height (IV, V, VI) show a gradual morphological
development, which can certainly not be said for case III. Both the flood and ebb slope
change rather suddenly during respectively 30 and 60 days, i.e. 8 and 16% of the time.
As a result the side slopes of cases III and VI (representative wave height) after one
year are rather different. Geotechnical instabilities are for this situation not likely to
occur in reality (full-year wave distribution; case III), whereas the simplified procedure
of simulating morphological development (representative wave height; case VI) predicts
a significant slope steepening of especially the ebb slope.
4,9
slope of seabed cot β
4,5
4,1
3,7
3,3
0 30 60 90 120 150 180 210 240 270 300 330 360 390
time [days]
case III; left slope case IV; left slope case V; left slope case VI; left slope
case III; right slope case IV; right slope case V; right slope case VI; right slope
Figure 10-56: Slope development of cases III, IV, V and VI throughout the year
184
From this paragraph some important conclusions can be drawn:

- SUTRENCH can only handle a schematized hydrodynamic climate. Both waves and
tidal current have to be discretized. Although some information is lost during this
process, there is some confidence that the input data are a reliable reflection of the
field conditions.
- If one is particularly interested in the morphological development over one or more
years, the sequence of the wave heights throughout the year (case I, II and III) is
not that important, as long as the inserted wave energy is conserved and the wave
distribution function is guaranteed. In other words, it does not make that much
difference if storm waves occur early or late in a year, although it was expected that
large waves would have a larger impact on a newly dredged profile.
- However, if one is interested in the morphological development throughout the year,
it is extremely important at what time during the year storm waves occur. The
largest waves that only occur about 10-20% of the time govern slope development.
This ‘time of influence’ is dependent on the combination of wave height and water
depth. During ebb more waves are felt at the side slopes, so ebb slopes will develop
during a longer period of time.
So the slope development can well be approximated if one only considers the largest
waves (satisfying the wave distribution). In fact a full-year simulation is then
replaced by a simulation over a shorter period of time with only the largest waves.
Equation 10-5 gives an idea for a minimum value for the wave height. It is then
easy to determine which part of the wave distribution in time has to be applied. One
should note that in this approach the backfilling is strongly underestimated, because
the smaller waves, that do transport some sediment, are neglected. Besides, this
approach is only permitted if one knows for sure that both slopes can develop
independently throughout the ‘real’ simulated time.
- The application of a uniform wave height, that is representative for a full-year wave
climate, yields good results, if one is particularly interested in the backfilling within
the channel boundaries, but less interested in the exact locations of the sedimented
material. A wave height based on significant wave heights that are exceeded only
87,6% of the time (H=1,8 m in this case) showed best graphical resemblance.
So, hopefully the difference in determining total backfilling and slope steepness is made
clear.
Backfilling: full-year simulation with uniform, representative wave height

Slope steepness: simulation over only the influential period of the wave distribution
185
10.5 Qualitative comparison of results with full 3D-model
Although it wasn’t possible to run a full 3D-model that solves the complete Navier-
Stokes equations for flow, such as was described in Paragraph 6.1.2, a qualitative
comparison can be made between both results to gain insight in the most spectacular
differences and the limits in between the SUTRENCH-results of this thesis are valid.
First, the results of the comparison of different sediment transport models, as was
executed by Jensen et al. [1999], are presented, see Figure 10-57. The water depth
and the sediment transport are made dimensionless; the side slope is π/6 ≈ 1:1,9,
which is rather steep. The most clear distinction between their sophisticated 3D-model
and the other models is the (total) sediment transport at the upstream slope. The
relative sediment transport first increases at the top of the upstream slope and then
approximates the quasi-3D model. When one only considers bed-load transport, it can
be observed that the bed-load transport shows a rapid decay on the upstream slope.
This effect becomes larger for steeper slopes, because the flow looses contact with the
seabed. The model of Mayor-Mora is based on a gradual decay of the sediment
transport and only suitable for calculations of total sedimentation inside a trench or
channel.
Figure 10-57: Results of different sediment transport models [Jensen et al., 1999]; (a)
relative bed-load transport, (b) relative total sediment transport from -5x/D0 to 5x/D0, (c)
relative total sediment transport from -10x/D0 to 100x/D0, (d) relative total sediment
transport from -500x/D0 to 3000x/D0
The differences between the sophisticated 3D-model and the quasi 3D-model become
smaller when the side slopes become flatter. It should be noted that these models do
not include the effect of waves.
The appearing inconsistent increase of sediment transport at the channel bottom is
caused by the incoming flow angle of 60°. As was explained in Paragraph 6.1.2, the
slow longitudinal acceleration of the refracted flow increases sediment transport.
In Figure 10-58 the dimensionless transport of four reference soils is presented.

Whereas the relative transport in MSI is only reduced by 15% in the navigation channel,
the sediment transport in MMS and MCS is reduced to zero, when the toe of the slope is
reached (at x/h0=15). This of course is caused by the very low content of suspended
transport and the large particle fall velocity. The reduction of sediment content of the
flow when the seabed consists of MFS is about 85% on the upstream slope.
186
The suspended transport further reduces over the channel bottom; at the toe of the
downstream slope only a few percent of the equilibrium sediment transport at the
surrounding bed is left.
Please note that all sediment transport rates are relative to their own equilibrium value:
transport in MSI is a factor 320 larger than transport in MCS.
1,2
upstream channel downstream
slope bottom slope
1,0
relative total sediment transport St/St0
0,8
0,6
0,4 MSI
MFS
MMS
0,2
MCS
0,0
5 10 15 20 25 30 35 40 45 50
dimensionless distance x/h0 [-]
Figure 10-58: Relative total sediment transport for MSI, MFS, MMS and MCS
When the gradient of sediment transport is determined, it becomes clear that large
sedimentation occurs at the top of the upstream slopes at t=0 days. This is partially
caused by the sharp upper corners of the side slope. When the channel geometry
changed into a more gentle profile after 90 days, the transport gradient is more
gradual. Also can be seen that especially in MFS a downstream shift of the upstream
slope has occurred. The downstream slope has flattened and shifted more, which can be
concluded form the ‘wide-spread’ and flat gradient.
0,0020
0,0015
dSt/dx (erosion / sedimentation) [kg/ms]
0,0010
upstream Erosion
0,0005 slope
0,0000
-0,0005
MFS; t = 0 days
Sedimentation MMS; t = 0 days
-0,0010 MCS; t = 0 days
MFS; t = 90 days
-0,0015 MMS; t = 90 days
MCS; t = 90 days
-0,0020
5 10 15 20 25 30 35 40 45 50
dimensionless distance x/h0 [-]
Figure 10-59: Sedimentation/erosion pattern inside channel
Summarizing can be concluded that the shape of the graphs of sediment transport
primarily deviate at the top of the upstream slope. As long as side slopes become not
too steep (steeper than 1:5), the errors remain limited.
187
11 Improvement of trench and channel design

11.1 Considerations on soil mechanics and morphology
As was mentioned earlier, channel geometries could well be adapted to expected

morphological changes and possible instability problems. With the knowledge gained in
the previous chapters one can easily cope with the relevant instability problems or
backfilling behaviour. In this thesis the focus will be on the optimization of the side
slopes. The most elegant slope design creates a balance between increasing slope
stability and reducing backfilling volumes. Advantageous side-effects can be a reduction
of the capital dredging costs (for a great deal dependent on dredging volumes) or a
reduction of the channel upper width in case of restrictions in space. A disadvantageous
side-effect can be the more difficult dredging method. The ability to control the
breaching mechanism, when box-cutting (see Paragraph 2.3), determines to a large
extent the applicability of variation in slope steepness. Besides, such measures are only
profitable if the slope height is considerable and the dredging is performed in multiple
layers.
When coping with backfilling problems, the most common solution is overdredging. It is
very important to distinguish between two definitions. Some consider overdredging as a
tool to deal with uncertainties inherent in channel dredging operations, but in this thesis
by overdredging is meant “an effective strategy to increase side slope stability or to
reduce dredging frequency and total maintenance costs”. Both types of overdredging
can be applied simultaneously. Also, overdredging does not necessarily have to be
performed at the channel bottom. Dredging of flatter side slopes than necessary is in
fact also a form of overdredging. Sometimes dredging of pocket holes in the side slopes
at intermediate depths is considered. This measure can be extremely efficient if large
travelling sand waves occur.
The appropriate method to reduce backfilling is strongly dependent on hydrodynamic

conditions and sediment properties, of which grain size and particle fall velocity are
most important. Sediment sizes larger than 500 μm turned out to stay very close to the
side slopes and the morphological development is characterized by slope migration,
while hardly any slope flattening occurs. The channel bottom remains unaffected, so no
significant reduction in depth takes place. Finer sediments (100 μm ≤ d50 ≤ 500 μm)
are more and more transported in suspension, so sediments spread over a larger width.
Often a small depth reduction is observed as well as some accretion at the side slopes.
Silty, but still non-cohesive sediments ((50 μm ≤ d50 ≤ 100 μm) are almost solely
transported in suspension. As a result the sediment will settle down over the entire
channel width and the current will blow still a rather large amount of sediment over the
channel. The above considerations lead to the following proposition of slope
improvement with respect to morphology:
500 μm ≤ d50 ≤ 1000 μm: (small) side slope shift, no overdredging at channel bottom;
100 μm ≤ d50 ≤ 500 μm: side slope shift or mid-depth pocket-holes, little overdredging
at channel bottom;
50 μm ≤ d50 ≤ 100 μm: minimum channel upper width to reduce trapping efficiency,
significant channel-wide overdredging, in fact over-dimensioning of the entire channel,
because sedimentation is almost evenly distributed over the width.
The above measures need to take possible tidal asymmetries into account. One could
think of shifting just one side slope or shifting one side slope over a larger distance than
another. Typical downstream behaviour is difficult to reduce, but also rarely
unfavourable for slope stability or navigable depth.
Besides these measures for slope improvement, it can be interesting to observe the
morphological behaviour of an initial channel geometry that is more or less identical to
its dynamic equilibrium shape. Will this lead to a more gradual and widely spread
sedimentation?
188
Although measures to increase geotechnical stability are to a lesser extent determined

by grain size only, also some remarks can be made. The susceptibility to static
liquefaction is reduced by flatter lower parts, while the susceptibility to dynamic
liquefaction by waves is reduced by flatter upper parts. Slope failure due to unloaded
micro-instability is most likely to occur at the steepest parts, while macro-stability (in
non-cohesive soils) only comes into play, if weaker soil layers or unfavourable water
levels are present.
Combined with the morphological considerations, the following improved slopes are
proposed, see also Figure 11-1. All-embracing names could not be created, so one of
the main characteristics of a certain slope design is chosen to name the alternative:
I: slope migration: straight side slopes at some distance from the required
profile to allow for some slope migration;
II: static liquefaction: flatter lower part and steeper upper part, compared to
slope I. This shape is implicitly equipped with a sediment
trapping reservoir;
III: volume reduction: a significant reduction in capital dredging volume is
obtained and the part of the slope that is most attacked
by storm waves is flattened;
IV: dynamic equilibrium: a side slope with a steep middle part and a flatter toe and
top resemble the dynamic equilibrium shape best;
V: sediment trapping berm: this division into two smaller slopes is easy to dredge and
satisfies the static liquefaction criterion of slopes not
steeper than 1:3 and not higher than 5 m (see Paragraph
4.2.1). This shape also is equipped with a considerable
sediment trapping reservoir.
cross-channel distance wrt channel centre [m]
75 85 95 105 115 125 135 145 155 165 175
1
surrounding seabed
-1
depth wrt surrounding seabed [m]
-3 Required Volume of
Volume overdredging
-5 required profile
I: slope migration
-7 II: static liquefaction
III: volume reduction
-9 IV: dynamic equilibrium
channel bottom
V: sediment trapping berm
-11
Figure 11-1: Schematization of required profile and improved side slopes
In Figure 11-1 also the required profile is plotted. This rectangular fairway forms the
required space for navigation and is set to a minimum water depth of 19 m over a
minimum width of 190 m. This navigable depth asks for a channel bottom 9 m below
the surrounding seabed. This required profile needs to be guaranteed for a full year.
The hydrodynamic conditions are again obtained from MPN (see Paragraph 10.4.1).
Because of the almost symmetrical morphological development, the same slopes are
applied at both sides. Simulations will be done with MMS and MFS for all five improved
slope designs. The coarser sand (MCS) is assumed to behave similar to MMS, only at a
189
slower rate. Silt will be transported in such large amounts that every channel profile
would experience a rather fast backfilling.
11.2 Results of simulations with improved slope
11.2.1 Improved slopes in medium packed fine sand (MFS)
The first five simulations with improved slopes in MFS are presented in Figure 11-2; this
figure zooms in on the ebb slope to be able to distinguish between the different slopes.
It can be concluded that the shapes of the initial profiles are smoothened by current and
waves, but some characteristics are still visible.
0
-1
-2
-3
-4
-5
-6
-7 required profile Case I after 365 days
-8 Case II after 365 days

Case III after 365 days
-9
Case IV after 365 days
-10
initial channel profile of case Case V after 365 days
-11
550 575 600 625 650
Figure 11-2: Morphological development of improved slopes; zoomed in on right (= ebb)

slope
To come to a comparison of all alternatives first some aspects will be investigated in

more detail. In Paragraph 10.4 wave-induced slope development throughout the year
was discussed. The largest waves primarily contribute to slope development. Which
‘improved profile’ will be most affected by storm waves?
190
6,5
Case I; left Case I; right
Case II; left Case II; right
6,0
Case III; left Case III; right
Case IV; left Case IV; right
5,5 Case V; left Case V; right
Steepest slope 1:y [-]
5,0
4,5
4,0
3,5
3,0
2,5
0 30 60 90 120 150 180 210 240 270 300 330 360 390
Time [days]
Figure 11-3: Steepest slopes of improved profiles throughout the year
When observing the (again) rather chaotic Figure 11-3, one can conclude that especially
case II is sensitive to the stormy period of the year; case IV and case I to a lesser
degree, while side slopes of profiles III and V only slightly flatten. This can of course be
explained by the fact that steeper upper parts of the side slopes are more easily
affected by larger waves. The more stable profile III has a relative flat upper part; the
steeper lower part is situated deep enough to be out of reach of the larger waves.
Profile V shows a more complex behaviour (Figure 11-4). The steep upper half is spread
out over the berm (slope of 1:10), while the steep lower part can persevere (1:4). So
Figure 11-3 shows an unjust favourable slope steepness of case V.
0,35
Initial Profile
0,30
after 365 days
0,25
0,20
0,15
0,10
0,05
0,00
150 250 350 450 550 650 750 850
Figure 11-4: Slope development of improved profile V in one year
Another criterion is channel backfilling behaviour. Therefore some numbers of dredging

volumes and backfilling rates are presented in Table 11-1.
191
Table 11-1: Backfilling behaviour of ‘improved’ profiles in MFS
Profile I Profile II Profile III Profile IV Profile V

3
initially dredged channel volume m /m 2500 2500 2400 2450 2500
3
volume of overdredging m /m 790 790 690 740 790
3
sedimentation within initial channel boundaries m /m 284 294 273 282 291
3
sedimentation within required profile m /m 6,2 7,5 10,1 9,1 5,8
3
remaining channel volume m /m 2216 2206 2127 2168 2209
3
remaining volume of overdredging m /m 506 496 417 458 499
sedimentation/initial channel volume % 11,4 11,8 11,4 11,5 11,6

sedimentation/volume of overdredging % 36,0 37,2 39,6 38,2 36,8
Because this navigation channel has a rather large width and particle fall velocities are
significant, almost all suspended particles will settle down inside the channel. Most time
of the year, bed load transport will be practically zero at the channel bottom. This
means that almost all incoming sediment will be trapped inside the channel. Measures
to reduce the channel width can be very effective if grains are mainly transported as
suspended sediment and have small particle velocities, but in this case these measures
will not have any effect.
Therefore the amount of trapped sediment can best be reduced by reshaping the
channel profile in such a way that little extra sediment is picked up inside the channel.
The discussion on threshold profiles teaches that the upper part should not be too
steep. Deeper inside the channel the side slopes can become steeper, because flow
velocities are reduced. So, channel profiles II and V will induce most sedimentation,
while profiles III and IV induce least sedimentation inside the channel. Sedimentation is
thus determined by intraslope transport and not so much by the total channel
geometry.
The criteria to distinguish between the various profiles are the following:
- capital dredging volume: this volume determines the initial costs;
- sedimentation within initial channel boundaries; this volume determines
maintenance costs;
- slope development: which channel profile shows largest slope development during
the ‘rough’ period of the year;
- static liquefaction: which profile runs the largest risk on liquefaction throughout the
year? The stability of the toe of the side slope and the amount of sedimentation,
which is generally loosely packed, at the lower half of the slope;
- cyclic liquefaction: the risk on cyclic liquefaction increases if the upper part of the
side slope is steep;
- risk on reduction of navigable width: if the lower part of the slope approaches the
required profile, there is a risk that the navigable width cannot be guaranteed.
The scores can be ‘+’, ‘-‘ or ‘0’ and are relative to each other. Depending on other soil
properties and specific goals, the appropriate slope design can be chosen. For instance,
if the soil is susceptible to static liquefaction, the fourth criterion will be of more
significance. Otherwise profile II is not recommended, see Table 11-2.
Table 11-2: Multi-criteria analysis of improved profiles in MFS
Criteria Profile I Profile II Profile III Profile IV Profile V

capital dredging volume - - + 0 -
sedimentation within initial channel boundaries 0 - + 0 -
slope development 0 - + 0 0
static liquefaction 0 + - 0 0
cyclic liquefaction 0 - + + 0
risk on reduction of navigable width + 0 - 0 -
total 0 --- ++ + ---

Profiles III and IV score well because of their morphological behaviour, while profile I is
still an attractive, simple alternative. If the channel depth is only limited and slope
heights remain small, the risk on slope instabilities, like static liquefaction, is reduced.
192
11.2.2 Improved slopes in medium packed medium sand (MMS)
The same procedure can be followed for improved slopes in MMS, see Figure 11-5.
1
0 Profile I after 365 days

-1
Profile II after 365 days
-2
Profile III after 365 days
-3
-4 Profile IV after 365 days
-5 Profile V after 365 days

-6
-7 required profile
-8
-9
-10
initial channel profile I
-11
550 575 600 625 650
Figure 11-5: Morphological development of improved slopes; zoomed in on ebb slope
In medium sand the channel depth remains 10 m and morphological development is

restricted to the side slopes. Sediment transport is primarily bed load transport, which
adapts instantaneously to the depth expansion at the side slopes. The improved slopes
are still visible after 365 days. Slope development is less pronounced than in MFS, see
Figure 11-6. Profiles I and II react strongly to the midyear storm, while profiles III and
IV show least variation in slope steepness.
5,5
5,0
4,5
Steepest slope 1:y [-]
Profile I; left Profile I; right

Profile II; left Profile II; right
4,0 Profile III; left Profile III; right
Profile IV; left Profile IV; right
Profile V; left Profile V; right
3,5
3,0
2,5
2,0
0 30 60 90 120 150 180 210 240 270 300 330 360 390
Time [days]
Figure 11-6: Steepest slopes of improved profiles throughout the year
The proportions of sedimentation rates between the different profiles are more or less
the same, although the absolute amounts of sediment are less, see Table 11-3. In MFS
the channel bottom entered the (prohibited) required profile, but this does not occur in
193
MMS. It will be interesting to determine the life span (time until bed level intersects
required profile) of all improved profiles, but that is not the aim of this thesis.
Table 11-3: Backfilling behaviour of improved profiles in MMS
Profile I Profile II Profile III Profile IV Profile V

3
initially dredged channel volume m /m 2500 2500 2400 2450 2500
3
volume of overdredging m /m 790 790 690 740 790
3
sedimentation within initial channel boundaries m /m 25 26 23 24 32
3
sedimentation within required profile m /m 0,0 0,0 0,0 0,0 0,0
3
remaining channel volume m /m 2475 2474 2377 2426 2468
3
remaining volume of overdredging m /m 765 764 667 716 758
sedimentation/initial channel volume % 1,0 1,1 1,0 1,0 1,3

sedimentation/volume of overdredging % 3,2 3,3 3,3 3,3 4,1
Again the profiles will be compared to each other on the same 6 criteria (Table 11-4).
Some small deviations in the scores are noticeable. Because the contribution of bed load
transport to morphological development is larger for larger sediment, the resemblance
to a threshold profile (Paragraph 7.1.3) or a dynamic equilibrium slope (Paragraph
8.2.3) pays off. Profile III resembles these profiles most and scores well at the other
criteria.
Table 11-4: Multi-criteria analysis of improved profiles in MMS
Criteria Profile I Profile II Profile III Profile IV Profile V

capital dredging volume - - + 0 -
sedimentation within initial channel boundaries 0 - + 0 -
slope development - - + + 0
static liquefaction 0 + - 0 0
cyclic liquefaction 0 - + + 0
risk on reduction of navigable width + 0 - + -
total - --- ++ +++ ---
Finally, the morphological behaviour of profile IV can be observed in Figure 11-7. Only
very minor slope development takes place.
1
0
Initial Profile
-1
-2 Profile after 365 days
-3 Required Profile
-4
-5
-6
-7
-8
-9
-10
-11
300 400 500 600 700
Figure 11-7: Initial profile IV and profile after 365 days
194
The above mentioned multi-criteria analysis does not contribute any weights. It depends
on the soil properties which criterion is most important. Furthermore, one should
constantly realize that these simulations are based on logarithmic velocity profiles.
Steep slopes should therefore be regarded suspiciously.
195
12 Guidelines for slope design
Throughout this thesis research it became more and more clear that a straightforward
and easily applicable set of design graphs that cover a wide range of hydrodynamic
conditions, channel geometries and sediment properties probably cannot be composed.
For example, tidal flow creates totally different slopes compared to unidirectional flow.
Between these two opposites, all kinds of asymmetric tidal flow exist. Channel
geometries strongly interfere with flow conditions. This interference becomes more
significant, if side slopes become steeper, if the channel bottom becomes narrower and
if the relative depth expansion increases. Only for gradually changing bed levels, the
schematization with logarithmic velocity profiles in equilibrium is completely
satisfactory. In the morphological simulations only the non-cohesive sediments were
considered, assuming a rather narrow grading. Wider gradings ask for a ‘multiple-
fraction approach’. A reasonable approximation can be obtained if the grain size of the
suspended sediment is assumed to be smaller than the bed-load sediment, but there is
often a lack of precise data. Moreover, the composition of the suspended sediment
changes with the hydrodynamic conditions.
Hopefully, one is aware of the above mentioned drawbacks, which are only a few
amongst many others, when applying the presented guidelines, as some guidelines can
indeed be deduced from this research as presented in the previous chapters. So in a
way this chapter can be considered as a summary of the most important conclusions
regarding slope design. At the same time this chapter has a generalizing nature. Simple
guidelines cannot be represented, while taking all relevant parameters into account.
When some guidelines seem obscure, the reader is referred to the relevant chapter.
12.1 Non-cohesive soils
12.1.1 No loads
As long as the soil shows no cohesive behaviour, the unloaded stability is governed by
micro-stability (φ), see Table 4-1. Slip circle analysis showed that macro-stability is not
likely to occur, unless weaker soil layers or a water level somewhere through the slope
are present. Then the slip circle is attracted to these local weaknesses.
The governing parameter is therefore the angle of internal friction, which is usually
larger for denser soils. Silts have an angle of internal friction that is about 3° smaller
than sands of the same porosity.
When the instability mechanism is forced into a slip circle (macro-stability), safety
factors are typical 10-40% larger (Table 4-2), depending on the slope height and the
soil type, but the consequences of macro-instability are also larger. When the slope
height goes to infinity macro- and micro-stability coincide; this occurs faster for weaker
soils.
It was shown that when the water level drops and the soil behaves drained, safety
factors are reduced about 15%, if the water level is located somewhere between the toe
of the slope and 0,75D. Besides, the normative failure mechanism changes from micro-
instability to macro-instability. If the side slopes contain a relative impermeable layer,
this strength reduction will be far more significant.
Flow slides can be caused by static liquefaction, when the sand is loosely packed. An
easy to use criterion is based on the dry critical density which is for most sands about
n=0,42. Soils with densities larger than the critical density (n < 0,42) are
unconditionally stable. Soils with densities smaller than the dry critical density (n >
0,42) can still be stable, depending on the soil geometry (Figure 4-19). Slopes with
heights smaller than 5 m are almost unconditionally stable, unless the soil is very
loosely packed (RD < 10%). For other combinations of slope height and soil properties
the reader is referred to Paragraph 4.2.
196
One should keep in mind that these results can be a bit conservative due to the
assumption that liquefaction is prohibited anywhere in the slope.
By far the most important parameter is the relative density (or maximum, minimum
and in-situ porosity). Especially the determination of the in-situ porosity used to cause
difficulties. Nowadays very good correlations between the dielectric constant and
porosity are available (n = 0,0136ε +0,02 for any soil type, see Paragraph 4.2.2.).
The slope height is very important when dealing with static liquefaction; doubling of the
slope height requires a reduction of the slope steepness with a factor 2. When
considering dredging of an overdepth, it can be dangerous to just increase the slope
height if the soil is susceptible to liquefaction. A flatter toe of the slope can improve
stability and simultaneously creates a more natural morphological profile.
Variation of the slope geometry, for instance slopes with steeper upper parts and flatter
lower parts, did not yield very large reductions of capital dredging volumes (less than
5%, which can increase for smaller channel widths), but adaptation of the side slopes to
a more morphological channel profile is very well possible without giving concessions to
slope stability or dredging costs. Small channel upper width reductions are also feasible.
Due to the limited advantages on slope stability with respect to liquefaction, ‘custom-
dredged’ slopes are only recommended if some combined advantageous effects can be
obtained (sediment reservoir, width reduction, overdepth etc).
Although these guidelines are based on research on homogeneous soils, some remarks
can be made on soil composites, like silty sand or sand-gravel-mixtures. Various
researchers drew opposite conclusions, mainly caused by different assumptions of the
relative density: some maintain the relative density of the original soil (neglecting the
‘contaminant’), some of the soil composite. However, when a constant settling process
is applied (which is representative for submarine conditions), all composites are less
resistant against liquefaction. It should be noted that these results are built on two-
composite mixtures. It is very well possible that widely graded soils will be more
resistant against liquefaction.
Breaching is only likely to occur in slopes of medium to densely packed fine sand or silt
(d50 < 300μm) that are steeper than the angle of internal friction. These soils have a
very low permeability and are therefore able to maintain a situation of underpressured
sand for some time. Although this instability mechanism controls the dredging process,
it will not occur on the side slopes of a navigation channel, which (after dredging) are
often not steeper than 1:3. Often for 10 m high slopes the following design rules are
used: coarse sand slope of 1:2,5; medium sand 1:3; fine sand 1:3,5. A good idea can
also be obtained from Figure 4-27.
Combinations of failure mechanisms ‘liquefaction flow slide’ and ‘breaching’ can occur,
for instance a breach that starts at a steep scar produced by a small flow slide and
retrogrades for many hours.
12.1.2 Wave loads
Waves appear in many guises when submarine slope development is discussed. Waves
were already mentioned as a trigger mechanism for static liquefaction. They can also
play a more influential role. Although wave forces are very common around sea level,
waves below sea level act more as a reduction of strength. Many researchers distinguish
between the direct, elastic effect and the indirect, plastic effect. The first approach
predicts the fluctuation of water pressures and effective stresses within one wave cycle.
This approach has no net effect on the mean stresses. The second approach takes cyclic
compaction into account, which causes a gradual water overpressure. It depends on the
consolidation properties of the soil whether these excess water pressure (EEP) can
attenuate or will build up during a storm.
Direct effect
Every soil experiences this direct effect. It was shown (Paragraph 5.1.3) that the
amplitude of the wave pressure has an upper boundary of about 25% of the hydrostatic
197
pressure, based on the assumption of linear wave theory and therefore decreasing
water pressures with depth according to a cosh-function. In the direct approach these
wave pressures fade out into the seabed according to an ‘e-power-decay’. Maximum
changes in effective stress occur at a depth of about 1/6 of the wavelength and are
about 37% of the amplitude of the wave pressure at the bed. At this depth the
overburden is already significant and the critical zone will be situated much shallower,
where effective stresses are still small.
Simple calculations for realistic (depth-limited) storm waves showed for depths varying
from 5 to 20 m that the cyclic stress ratio (CSR= wave-induced amplitude of effective
stress divided by geostatic effective stress) will usually be in the order of 15-20% and
will never exceed 40-45%, even if the unit weight is very small. Failure of the seabed
can occur if the horizontal permeability is approximately 7 to 8 times larger than vertical
permeability.
Although failure of a flat seabed caused by this direct effect is not very likely, safety
against slope failure will decrease in presence of waves. The Mohr-Coulomb criterion
should be adapted to effective stresses under wave crests and troughs. It asks for
numerous calculations to define the critical wave (length, period, height and orientation
to the slope). Such calculation tools do not exist yet. A safe compromise would be a
reduction of the effective stresses (about 20-25%), so an increase of the required
safety factor with the same percentage. Although still many questions exist on a
completely sound theoretical description of wave-induced pressures and stresses inside
the seabed, it seems that in engineering practice, where slope design usually not will be
governed by micro-stability and slopes are often in the range of 1:3 or flatter, the risk
of wave-induced slope failure implicitly is accounted for in the safety factor, as long as
the indirect effect (see below) does not cause failure, which generally is the case for
denser, more permeable soils.
Indirect effect
Every load cycle will induce compaction of the soil skeleton, no matter how dense the
soil is, but it depends on the combination of load and strength whether the soil will be
susceptible to cyclic liquefaction. The load is represented by the wave height, wave
period, storm duration and water depth, whereas the strength is represented by the
consolidation properties (k, mv) and the relative density (RD).
Coarse, permeable and dense soils will have such good consolidation properties that
they won’t experience wave pressure build-up. Finer, impermeable, more loosely
packed soils should be checked on this instability mechanism.
As a first design criterion Figure 5-14 can be used to determine the number of load
cycles for a given wave condition (safe upper boundary). This graph is based on
undrained response, which means that for better consolidation properties and/or pre-
shearing conditions the situation will be more favourable. If local wave-induced relative
shear stresses are in the order of 0,10 or larger, generally there is some risk on cyclic
liquefaction. Then drainage and pre-shearing should be taken into account. In many off-
shore conditions such stresses will be present as was shown in Figure 5-15.
Numerous calculations with MCycle made it clear that the dense reference soils
(RD=0,8) are unconditionally stable under (depth-limited) waves (H=5m). The loosely
packed reference soils (RD=0,2) all collapsed in a very mild wave climate (H=1,00-
1,75m), which is an indication that such loosely packed soils are not very likely in a
moderate wave climate. The common medium packed soils liquefied, when subjected to
storms with waves of 3,25 to 4,75m.
Slope flattening was not very effective (see Paragraph 5.2.4). Instead one can achieve
great results with improving soil properties. Relevant soil properties are:
- (hydraulic) permeability: an increase means that excess pore water pressures can
easy dissipate to deeper layers, thereby increasing ‘the depth of influence’;
- vertical compressibility: a decrease means that compaction of the grain skeleton
and the matching decrease pore volume takes more time, which is favourable for
drainage of EPP’s;
198
- relative density: an increase considerably raises the number of load cycles until
cyclic liquefaction occurs or the maximum resistible relative shear stress. In other
words, longer storms or storms with bigger waves can be withstood.
Of course all of the above soil properties are related to each other and manipulating one
of them will certainly affect the others.
More permeable soils will spread the EPP’s over a larger depth and the critical time will
be at the beginning of a storm (typical after half an hour). Relative impermeable soils
keep building up EPP’s during the entire storm and these EPP’s are more concentrated in
the upper layer. So soils with a larger hydraulic permeability, like coarse sands, or a
smaller compressibility, like dense sands, therefore are far less susceptible to
liquefaction due to cyclic loading, see Figure 5-23.
Preceding storms appear to be primarily favourable for soils with small relative densities
(RD<0,5). The maximum wave height can increase with 0,5m.
Not many combinations of various soil layers have been investigated. A two-layer
system (5 m thick upper layer on a 15 m thick ten times less permeable lower layer) did
not yield spectacular results on maximum wave heights. Very thin, weak layers on top
of an impermeable, dense layer can be more unfavourable. This has to be investigated.
Measures to reduce the risk and consequences of wave-induced cyclic liquefaction are:
-) adaptation of the channel geometry or alignment such that large sedimentation under
calm hydrodynamic conditions is prevented or is forced to certain predestined places,
like sediment pocket holes. This should prevent fast accretion of the side slopes,
resulting in loosely packed sediment;
-) selection of a ‘rough’ dredging method which triggers regulated liquefaction flow
slides or, if this does not occur, causes cyclic compaction of the soil, thereby increasing
safety against liquefaction;
-) construction of discrete zones with stable material like gravel;
-) (partial) (vibro-)compaction of the soil;
-) soil improvement
Most effective measure certainly is compaction, because the relative density is the most
influential parameter. Although slope flattening is not considered to be an effective
measure to increase slope stability, slope steepness can be important, once a wave-
induced flow slide has occurred. The erosive capacity of the resulting turbidity current is
strongly dependent on slope steepness and slope height.
So whereas the direct, elastic wave effect was implicitly accounted for in the safety
factor, the plastic, indirect effect can ask for serious measures!
Other loads
Not much attention was given to other than wave loads. Loads like earthquakes,
tectonics and man-made loads (anchor forces, thrust of a ship’s propeller) can act as a
trigger mechanism to induce static liquefaction, but can also act in a more active, cyclic
way. Seismically induced liquefaction shows many similarities to wave-induced
liquefaction. Sophisticated software packages can be applied or the more
straightforward approach of steady-state lines (Figure 5-33). Centre of fundamental
research is located in Japan and the USA, because of the high earthquake frequencies.
12.1.3 Morphology
Often the initial channel profile will develop under the influence of hydrodynamics. In
this thesis, (tidal) currents and wind waves are investigated. Once the bed-shear
stresses, induced by these hydrodynamic forces, exceed a certain threshold value,
sediment transport will reshape the channel profile. Especially in the presence of waves,
almost all reference soils will come into motion. In special conditions slopes will develop
towards less favourable shapes with respect to geotechnical instabilities; most of the
time side slopes will flatten and slope development will be governed by morphology.
199
When currents approach a navigation channel there are two extreme possibilities.
Currents that are directed perpendicular to the channel axis will slow down due to
continuity, while currents parallel to the channel axis will accelerate due to equal water
slopes inside and outside the channel, see Figure 6-1.
Oblique currents give a more complex situation and the relative flow velocity (V1/V0) can
decrease as well as increase inside the channel, depending on the relative depth
expansion, incoming flow angle, channel width and bed roughness outside and inside
the channel. The perpendicular component will decrease almost instantaneously, while
the longitudinal component needs some time to adapt. In practice this means that
currents with incoming flow angles smaller than 30° will almost reach the equilibrium
situation of fully refracted flow (Figure 6-4), whereas currents with incoming flow angles
larger than 60° will slow down, unless the channel width is very large (>1500m,
depending on the relative depth expansion), see Figure 6-6.
Analysis of a full 3D-Reynolds-averaged Navier-Stokes model taught that besides the

current refraction, also a secondary flow will develop caused by shear in velocity profile.
This means that the bed-shear stress on the upstream slope is refracted more than the
depth-averaged current velocity. This difference in flow angle can grow as large as 30°
for incoming flow angles of 60-80° (Figure 6-10). At very steep side slopes (around 1:2)
also flow separation can occur. There is hardly any sediment exchange between the
separation bubble and the current.
When waves are directed perpendicular to the channel axis, the wave heights inside a
navigation channel generally are 80-110% of the incoming wave height, depending on
the wave period and the relative depth expansion. Based on continuity of wave energy
flux, the opposite of wave shoaling occurs, see Figure 6-12.
Parallel incoming waves will also experience this effect, but due to curved wave crests
the wave energy will be reduced at the side slopes.
Oblique waves either will be reflected or will be refracted, depending on the wave period
and the relative depth expansion. The transition between refraction and reflection
concerns wave angles of 20-60°, see Figure 6-13. Waves with smaller incoming angles
will totally reflect, while large-angle-waves will refract. Especially incoming wave angles
around this critical wave angle will result in an increase of the wave height inside the
channel, see Figure 6-14.
The threshold of motion for the reference soils is about 0,3-0,5 m/s in terms of depth-
averaged flow velocity. For grain sizes smaller than 500 μm this threshold value is
nearly constant. Critical flow velocities decrease for downward sloping bottoms. In case
of perpendicular flow this effect is noticeable for slopes steeper than 1:10, see Figure
7-6. Based on logarithmic profiles, the formula for the reduction factor (Eq. 7-10) and
equally stable grains at the surrounding seabed and the top of the upstream slope, a
threshold profile was deduced. This channel profile will develop under very mild
hydrodynamic conditions, but also has an indicative meaning for stronger currents,
because sediment transport is based on the difference between actual bed shear
stresses and threshold values.
Fine to medium sediments are very easily stirred up by waves. Wave heights of a few
dm are able to move some sediment. In combination with a current this can result in
actual sand transport, because under the assumption of linear, non-breaking waves,
waves themselves cause no net transport.
Simulations with SUTRENCH, a computer model developed by WL|Delft Hydraulics, gave

insight in the morphological behaviour of channels and side slopes. The amount of
sediment transport, and thus the time scale of morphological development, is most
sensitive to wave height, current discharge, water depth, particle fall velocity and wave-
related roughness.
The minimum sediment gradient, which is a measure for local sedimentation and
therefore (upstream) slope development, is most sensitive to current discharge, water
depth and rather sensitive to wave height and current- and wave-related bed
roughness.
200
Sediment transport for the reference case ‘navigation channel’ can be approximated by
formula 10-4. Although it can be very dangerous to apply this formula in a wide range
of conditions, it shows the sensitivity to the important hydrodynamic parameters.
The current discharge and direction have a large influence on slope development. Two
extreme conditions have been investigated: unidirectional and symmetric tidal flow. In
the former case, two completely different side slopes will develop, hereafter named
upstream and downstream slope. In the latter case both side slopes show a more or
less equal morphological development. In between all kinds of asymmetric tidal currents
are possible and the side slopes will show intermediate behaviour.
In unidirectional flow, downstream slopes will always flatten and stretch out over a
considerable width. Upstream slopes will migrate and can show steepening as well as
flattening behaviour. This behaviour is also time dependent. It was shown that initial
upstream side slopes flatter than 1:4-5 first show some slope steepening and then a
gradual development towards dynamic equilibrium slopes. Typical values of slope
steepness are 1:5 to 1:7. Variation of initial side slopes taught that slopes converge
during the first three months.
Very fine sediments did not show autonomous slope development, because backfilling
proceeds very fast.
If the sediment transport capacity is increased by increasing current discharge or wave
height, slopes can steepen significantly. Steepest upstream slopes seem to occur for
increasing sediment sizes and hydrodynamic forcing (see Figure 10-18 and Figure
10-24), because the fast adapting bed load transport has a large share in total
transport. A depth-averaged current velocity of 1 m/s resulted for sediment sizes larger
than 200 μm in slopes even steeper than the angle of friction. Although such slopes are
not realistic (slope failure will occur) and cannot be modelled properly with SUTRENCH,
the morphological behaviour is in line with observations in nature. Large (constant)
sediment supply, e.g. in river deltas, can cause oversteepening and, consequently,
slope failure.
In tidal flow both slopes develop similarly. Upstream slope behaviour dominates over
downstream slope behaviour. Because of the turning current, morphological
development proceeds more slowly, slope migration and steepening are less
pronounced and failure is not likely to occur. Slopes after one year are typical 1:4 to
1:6, unless backfilling processes start to interact with slope development; then fast
slope flattening occurs.
If the hydrodynamic forcing is increased again some slope steepening can be observed
(see Figure 10-38 and Figure 10-43). Currents are more effective in causing steeper
slopes. Larger waves also increase sediment transport, but bigger waves start to
interact with particles on the side slopes. Fine to medium sediment (d50 =200-500 mm)
combine the properties of being transported in large amounts and having a large
sediment fall velocity. As soon as the current enters the channel, grains start settling
down.
Simulations with a real tidal and wave climate yielded two important conclusions.
Backfilling can well be predicted when using one representative wave height instead of a
full-year wave climate. This representative wave height is rather large. When wave data
are obtained as significant wave heights over small periods, a significant wave height,
that will not be exceeded 80-90% of the year, has to be chosen as representative wave
height. Although backfilling predictions are rather accurate, slope development is not
well predicted. Slope development is governed by the largest waves. The minimum
wave height that affects fast slope development is dependent on the water depth (see
Eq. 10-5 for an approximation). A simulation over only the period in which this
minimum wave height is exceeded yields accurate slope development.
Furthermore it appeared that the order of the waves throughout the year does not
affect morphological development, if one is interested in one or more years. So, the
effect of a storm, that occurs just after dredging has finished, is hardly noticeable after
one year, see Figure 10-53.
Four different ‘improved slopes’ in fine and medium sand were compared to straight
side slopes, see Figure 11-1. Slope development was evaluated on six different criteria,
201
varying from capital dredging volume to susceptibility to static liquefaction. Depending

on the soil properties, tailor-made slope design seems advisable, unless suspended
transport has a major share in total transport. For example, improved slope geometry
‘dynamic equilibrium’ shows least slope development and has lowest backfilling rates
(see Paragraph 11.2). Geometry ‘static liquefaction’ scores worse on all criterions, but
one: susceptibility to liquefaction. In loosely packed soil, this criterion will certainly
outweigh the other criteria. It should be noted that the technique of improved soils will
only be advantageous if slope heights are significant, bed load transport exceeds
suspended transport and the designer has some interest in reducing dredging volumes,
increasing stability or reducing slope development.
12.2 Cohesive soils
Cohesive soils received less attention than cohesionless soils in this thesis; partly
because a number of stability mechanisms are not very likely to occur when some
cohesion is present and partly because the underlying theories are not well established
(research is often site-/soil-specific).
No loads
When no external loads are present, failure will always occur along (deep) slip circles.
Micro-stability is prevented by cohesion. Because of the low permeability of cohesive
soils (k<10-9 m/s) undrained calculations based on total stresses (undrained shear
strength) have to be done. Besides, the long-term condition for fully drained soils
should be checked. A completely satisfactory relation between drained soil properties (c,
φ) and undrained soil properties (su) does not exist. Stress history and (over)
consolidation ratio influence the undrained shear strength. In practice, determination of
both properties is recommended.
Calculations based on a rather simple relation resulted in slightly lower Safety Factors
for undrained calculations. Slope flattening appeared to be less effective if the soil
behaves undrained (Figure 4-8). The sensitivity to slope height also becomes larger for
undrained clay (Figure 4-9). The over-all conclusion however is that unloaded slope
failure is not very likely in cohesive soils, because just stable slopes (SF=1,5) are in the
order of 1:2 for slope heights of 10 m and in the order of 1:2,5-3,0 for slope heights of
20-30m. One should take into account that the reference clays under investigation are
all relatively weak; in reality also much stiffer clays occur.
For small periods of time vertical walls can persist if the slope height does not exceed
values of 1-2 m (Table 4-4).
Wave loads
Waves can liquefy cohesive soil just as cohesionless soil. Due to the even stronger
dependency on soil properties, the large variation of these properties in cohesive soil,
the lack of a computer program and the lacuna in theory of slope instability caused by
wave-induced liquefaction, it was chosen not to investigate wave-induced slope failure
in cohesive soils. Some researchers investigated this problem (see De Wit [1995]) and
got some very nice results, although conclusions are only soil-specific (and thus site-
specific).
De Wit found that waves with a wave height exceeding a certain threshold value were
able to liquefy cohesive soils. This wave height increases with the consolidation period.
Once a layer of fluid mud is present on the seabed, turbulence intensities in a current
tend to decrease. The fluid mud layer behaves as a viscous fluid.
Morphology
Although in recent years great efforts were made at the ability to predict morphological
behaviour of cohesive sediment, this subject received very little attention in this thesis.
This lack of attention doesn’t mean that this subject is not important, but it takes a
rather different approach of sediment transport. Besides, the variability in soil
properties of clays that are relevant to sediment transport makes it almost impossible to
create a universal set of guidelines. With cohesive sediments, the bulk properties of the
202
mixture determine sediment behaviour, whereas non-cohesive sediments could well be

represented by just the particle diameter (and the related particle fall velocity).
At present state-of-the-art it is still not possible to predict cohesive sediment behaviour
from its physical and chemical properties alone. Most research therefore is aimed at
laboratory tests in which hydrodynamic field conditions are simulated and cohesive
sediment is taken from the field. A disadvantage of this method is the rather strong
relation to the site. Additional fundamental research is needed.
The relevant processes and sediment states are illustrated in Figure 12-1. The four
processes are erosion and transport, deposition and consolidation. Especially the last
mentioned process is typical for cohesive sediment. Consolidation is the gradual
expulsion of interstitial water by the self-weight of the sediment and cyclic stresses (due
to waves) resulting in an increase in both the density of the bed and its strength
(resistance against erosion) with time. Deposition is influenced by flocculation of
particles caused by a change of the physico-chemical properties.
Figure 12-1: Schematization of cohesive sediment transport [source: Whitehouse et al., 2000]
The four states in which cohesive sediment can occur are mobile suspended sediment, a
high concentration near bed layer (fluid mud), a freshly deposited or partially
consolidated bed and a settled or consolidated bed. Thus, the separation between the
seabed and the transported sediment particles is less pronounced than in non-cohesive
sediment.
Due to the highly concentrated (erosive) fluid mud layers near the bed, slope
development probably will be harder to predict than in non-cohesive soils. Once these
fluid mud layers have come into existence, side slopes may become much flatter than
non-cohesive slopes under equal hydrodynamic conditions.
In future years cohesive sediment transport and slope development in particular will
slowly develop from site-specific research in theoretical research.
SUTRENCH is equipped with a module for mud transport, but this module hasn’t been
tested. Even sand-mud mixtures can be introduced.
203
13 Conclusions and recommendations
13.1 Conclusions
13.1.1 General
In the previous chapter guidelines for slope design were presented. In a way these
guidelines can be interpreted as conclusions of this research. In this paragraph not the
results of the various simulations are discussed, but the findings on achieved goals,
present state-of-knowledge, the reliability of the results, the applicability of the present-
day computer models and future developments. This will eventually lead to a set of
recommendations for future research, described in Paragraph 13.2.
The main objective of this paragraph is of course to investigate to what extent this
research has been able to satisfy the aims and objectives of Paragraph 1.3. Does this
thesis report answer all the stated questions regarding slope design in engineering
practice? This thesis should gain insight in:
between them;
diameters and soil properties.
1. Development of submarine slopes was considered due to morphodynamics. If

this ‘development’ is caused by geotechnical processes, this was considered failure,
because usually large masses are wasted. Simulations with SUTRENCH (to be
mentioned more extensively in the next paragraph) gained insight in morphological
development. The focus was aimed at the influence of initial side slopes, current
discharge, wave height and wave schematization throughout the year.
2. Especially Chapter 2 gave a broad overview of all known failure mechanisms for
a very wide range of slopes. It was shown that subaqueous slope failures do not only
occur at steep slopes; even slopes of only a few degrees can collapse. The classification
into (subaqueous) regions showed that most slope failures occur at continental slopes,
while numerous release mechanisms were mentioned, which can act individually or in
combination. Examples from dredging practice showed the state-of-the-art and
problems that occurred in real cases.
3. In present engineering practice the initial slope is designed from a geotechnical

point of view. Morphological development of side slopes usually receives little attention;
computations are mainly aimed at predicting backfilling and expected life time of a
navigation channel or pipeline trench with respect to maintenance costs. So the focus
shifts in time from geotechnical towards morphological aspects. Interaction becomes
important if due to morphological development the slope stability is reduced. Two
possible causes are:
- the slope steepens, while the slope height does not decrease too much;
- soil properties, like porosity or permeability, worsen, e.g. because of large
sedimentation under calm weather conditions.
Simulations proved that the first cause can occur under specific circumstances. The
current has to be (almost) unidirectional and the wave height has to be limited: wave
stirring on the surrounding seabed with limited wave action on the side slopes. This
situation of a rather large, constant, unidirectional sediment supply of grains with large
particle fall velocities primarily occurs at river deltas. In most cases this unfavourable
combination will not be present.
The second cause sounds very logical, but is very hard to predict, because little is
known about the soil skeleton during the submarine settling and consolidation process.
204
The soil strength increases in time, waves cause cyclic compaction and stirring at the
same time. Soil stratification on the side slopes may occur, because larger grains will
settle faster than smaller grains. If interaction should be considered, side slopes tend to
be steep, which means that the current and wave pattern is rather complicated.
Modelling of these processes is still very difficult, if not impossible. However, one could
estimate the risks on this type of failure due to interaction. Most susceptible are slopes
in fine to medium sands (no cohesion) in unidirectional flow with little wave compaction
and large slope heights.
4. Not all soil types received equal attention. Cohesive soils were only discussed
briefly; partly because a number of instability mechanisms are not very likely to occur,
when some cohesion is present, and partly because the underlying theories are not well
established (research is often site-/soil-specific). In Paragraph 13.2.1 some brief
remarks on cohesive sediment transport are made.
Non-cohesive sediments were studied for a wide range of soil properties. Every failure
mechanism has its own important soil parameters, which are often not related.
Sometimes, simple assumptions had to be made.
So, although all aims and objectives have been studied, the problem could not
completely be solved.
This thesis research tended to bridge the gap between theoretical research and practical
engineering tools. But not only this gap appeared to be very important throughout this
study, also the gap between two fields of expertise turned out to be very evident,
namely ‘soil mechanics’ and ‘coastal morphology’. In literature specific problems were
treated separately; the same symbols were used for totally different quantities in both
disciplines; most researchers were either geotechnical or hydraulic/morphological
engineers. Gradually, some mutual efforts on the interface between soil and water are
made by representatives of both disciplines, for instance in Delft Cluster on breaching
processes.
Starting from the development of both research fields, this division can be understood,
but it is to be expected that the interface of soil and water will play a more and more
important role. An increasingly larger area of land is reclaimed from the sea. The
demand on reclamation soils increases, which was already shown in Singapore. The
submarine slopes determine to a large extent the required amount of soil.
As world trade and harbours grow, vessel sizes will increase and navigation channels
have to be deepened. Due to a lack of space, side slopes need to be steeper. As a
consequence of steeper side slopes and larger relative depth expansions, the
hydrodynamics are more and more influenced by the presence of a channel and start to
interact with the seabed. Soil mechanics, fluid mechanics and morphology cannot be
treated separately anymore. They are constantly involved in a process of dynamic
interaction.
Due to the very broad set-up of this thesis, it was impossible to study every single
instability mechanism in great depth. This was not only caused by the limited time or
theoretical knowledge of the author, whose educational background is morphological,
but also due to the already large gap between the most sophisticated theories and the
much more straightforward approach in practice. Therefore this thesis research was
characterized by a constant search for balance between theoretical depth, which is a
prerequisite for a good master’s thesis, and practical applicability, which is of course the
express wish of the customer, Royal Boskalis. As a consequence every specialist on a
certain subject will perhaps not be dazzled by striking innovations or completely new
theoretical ideas, but on the other hand he will learn something from the other topics
that have been discussed.
Because of the complexity of many of the governing processes, it appeared to be very

difficult to make sound guidelines for slope design right at the interface of both
disciplines, because on the one hand the lack of theoretical background and on the
other hand the fact that known theories had to be applied at their boundaries of validity.
For instance, wave pressures inside the seabed are based on a flat seabed, or velocity
205
profiles based on logarithmic velocity profiles. Nevertheless some universal guidelines

were deduced and presented in Chapter 12.
In engineering practice only a few, but certainly not all parameters needed for good
sediment transport predictions are available. Nevertheless some calculations have to be
executed that have some predicting value. Qualitative behaviour sometimes can
reasonably be estimated, but quantitative behaviour and exact timescales are a lot to
ask for. A reliable method to improve the accuracy can be obtained by dredging a trial
trench or channel.
As was mentioned before almost all calculations methods are based on schematized
sinusoidal waves, whereas in coastal regions these waves should be described by the
cnoidal wave theory. Because of the lack of mathematical formulations on cnoidal waves
it is impossible to introduce cnoidal waves in mathematical models. In numerical models
this can be done, but most sediment transport formulations and wave pressure
penetration formulas are based on sinusoidal waves.
When considering the indirect plastic effect, more attention should be given to the wave
pattern, because the occurrence of a group of very large waves is far more effective in
causing flow slides than a more regular wave pattern. Also the history of the sea bed is
very important, although it is almost impossible to describe it accurately, because of the
porosity which is dependent on the current and wave conditions throughout the year. A
sudden storm in an otherwise calm hydrodynamic region has far more impact than a
slowly increasing storm in a rough climate throughout the year.
13.1.2 Applied computer models
In this thesis various computer programs have been used, some of which are available
on the market, some of which are for internal use only. All used computer programs will
be discussed with respect to the validity of the computed results and the applicability in
engineering practice.
M-Stab
This 2-dimensional slope stability program by GeoDelft is easy to use for both drained
and undrained calculations. The possibilities to include loads are limited; only constant
earthquake loads can be taken into account in a very schematized way. For
groundwater flow additional calculations can be done with M-Seep, but a satisfactory
method to model waves still does not exist.
SLIQ2D
This 2-dimensional program calculates the just stable slope with respect to static
liquefaction. The results probably are a bit conservative due to the criterion that
liquefaction is prohibited everywhere in the slope. The input asks for extensive soil tests
to obtain no less than 11 soil input parameters, which is already a strong improvement
compared to earlier versions of this program. In this thesis a sensitivity analysis was
executed to reduce the large number of input parameters to only a few, more familiar
input parameters. Indeed, it can be concluded that it is possible to obtain reasonable
results when, besides of course the slope properties (height and shape), only the
porosity (or relative density) and the maximum volume strain are known. Still, even
these two properties are hard to predict in field conditions. Therefore a reliable
correlation on the porosity and the dielectric constant, which can be measured easily
with simple adjustments to the well-known CPT-equipment, was presented. One big
advantage of this correlation was the independence on soil type. It is recommended to
increase the efforts on electric permittivity measurements.
The development of SLIQ2D has stopped in the mid 90’s and future improvements are
not to be expected. SLIQ2D can only handle homogeneous soils and is based on a
number of PLUTO-stress-calculations and therefore not very transparent. The more
widely used program PLAXIS may be more suitable for development of a special module
that calculates susceptibility to liquefaction.
206
MCYCLE
This is another computer program developed by GeoDelft that is based on the
uncoupled approach of direct, elastic stress-strain behaviour within a wave cycle and
the indirect, plastic behaviour over a large number of wave cycles. Many
schematizations were needed to develop a model that is able to predict whether excess
pore pressures in the seabed will build up during a certain storm period. The results
probably are reasonable for a flat seabed, but will deviate from reality if the seabed is
sloping or in the presence of structures, like caissons or breakwaters. Many researchers
consider wave pressure built-up as one of the most difficult problems. Too little is
known about the exact wave pressures below the water surface (especially larger waves
deviate from linear sinusoidal waves), the state of the seabed, the interaction between
wave pressure at the seabed (p0) and the composition of the seabed itself, et cetera.
SUTRENCH
SUTRENCH developed by WL|Delft Hydraulics is a user-friendly model with a less user-
friendly interface. The model yields reasonably good results and can still compete with
more sophisticated models like Delft3D. In fact if some calibration data are available, it
produces more realistic morphological development of trenches and channels than
Delft3D, as was investigated by Klein [2003]. Because SUTRENCH is primarily suitable
for gradually varying flows, DELFT3D should probably yield better results for steeper
slopes and more severe hydrodynamic conditions, although this program is also
bounded by hydrostatic pressure distributions. A model that solves the full Navier-
Stokes equations developed by Jensen et al. was described. This model appears to
produce nice results, but this very sophisticated model has the strong limitation that
waves can still not be implemented.
A disadvantage of SUTRENCH concerns the very schematized input data. Tidal currents
have to be discretized in relatively large ‘blocks’. The time step to compute bed changes
is dependent on the duration of such a tidal ‘block’. A wave climate can only contain 20
periods of a constant wave height.
Another disadvantage is the constant grid size. On the one hand a certain adaptation
length must be guaranteed, asking for a large grid size, while on the other hand one
desires a small grid size at the location of the side slopes, where the bed level and flow
are changing rather fast.
The output was not aimed at producing the steepness of the side slopes, which is
understandable, because this program is not intended to simulate slope development.
Besides, the definition of slope steepness is arbitrary and should be based on the
instability mechanism one is interested in.
Because of the relative steep side slopes, it is expected that the results are reasonable
accurate with respect to the amount of sedimentation within the channel boundaries,
whereas it is questionable whether the exact sedimentation/erosion pattern on the side
slopes is correct. The results seem realistic and can explain some of the phenomena
observed in practice. So, there is some confidence in the qualitative behaviour, but the
quantitative behaviour and the representative timescales may be less reliable.
13.2 Recommendations
Based on the idea that this thesis should be the bridge between theoretical research and
dredging practice, the recommendations are subdivided into two categories. In
Paragraph 13.2.1 the lacunae in current theories are discussed. These lacunae need to
be filled to improve the calculation methods. In Paragraph 13.2.2 some steps to fill the
gap between slope design guidelines, in fact this thesis, and the application in practice
are explored.
207
13.2.1 Future fundamental research
Wave pressure build-up in water and in layered soil

Little is known on pressures caused by non-sinusoidal waves. Some remarks were made
on differences between dense and loose sea beds. Pressure development inside slopes
will also deviate from a flat seabed. Still, many questions have to be answered;
Wave pressure modelling in finite element method

Waves are still not implemented in finite element methods, like PLAXIS. Some
additional research on wave pressures inside the seabed probably has to be done, but
with present-day knowledge nice predictions can be done;
Morphological modelling of slopes in cohesive sediment

As was mentioned in Paragraph 12.2, progression is achieved in modelling of sediment
transport in cohesive soil. On steeper side slopes the layer of fluid mud makes
predictions on slope development very difficult. Once a layer of fluid mud comes into
existence, slopes can flatten significantly and cause failure of the navigation channel;
3D-hydrodynamic and morphological modelling at steep side slopes

Jensen et al. [1999] made a very nice attempt, but they still not included waves, see
Paragraph 6.1.2 for a possible model set-up. It was mentioned many times that the
representation of the current velocity with logarithmic profiles introduces errors that
increase for steeper side slopes. As a first indicative step one could compare measured
bed shear velocities with the ‘logarithmic-based’ bed shear velocity to get an idea of the
magnitude of this error. Probably, some more laboratory data with various current and
wave input over steep side slopes are needed. At the same time, wave pressures on top
of and inside the seabed can be measured.
13.2.2 Applicability in dredging practice
Correlations between soil properties

The porosity and the maximum dilatant volume strain appeared to be the most
important soil parameters that determine the susceptibility to liquefaction. In dredging
practice often no data on porosity are available. Therefore in Paragraph 4.2.2 a method
based on the dielectric constant was presented, which showed reliable results. In the
future this method will become invaluable in regions that are susceptible to liquefaction
and an investment on such a device is recommended.
Probabilistic slope optimization for economic purposes

In the future, tender forms like ‘design, construct & maintenance’ will become more
common. Contractors will be responsible for the functionality of the channel or trench
for a certain period. The trend towards maintenance contracts will also shift the period
of interest of a contractor. The transition from optimal slope design in the period of
capital dredging to optimal slope design over the entire life span of the navigation
channel (capital and maintenance dredging) asks for a somewhat different approach.
The focus shifts from the side slopes and their stability during construction towards the
morphological behaviour of the entire channel geometry. In fact, the design of the side
slopes is not longer based on stability calculations and steepening/flattening processes,
but on the consequences for the backfilling of the channel as a whole. During this
research it already turned out to be rather difficult to treat the side slopes as separate
parts of a navigation channel. As time went by, the channel started to behave more and
more as a whole of bottom and slopes.
A very nice 1D-attempt was made for the Mississippi by Moser et al. [1995]. Of course
in offshore conditions, the situation is far more complicated. Risk-based planning has to
be involved. Some very interesting questions will arise. Will the contractor be
responsible for a ‘1:10.000-year storm’? What will happen if coastal measures in the
neighbourhood change the sediment pattern? Which risks is the client willing to take
and which risks are for the contractor?
208
Database of navigation channels

Because few fitting data are available, many models are calibrated with the same data.
To enlarge the range of such models, a database of all navigation channels over the
world would be of enormous value. During this research it appeared to be very difficult
to obtain input data from old project files. If a database on navigation channels is
available, new projects can easily be added to the database. One can think of
hydrodynamic data, soil properties, survey data et cetera.
13.2.3 Set-up of laboratory experiment
Already many shortcomings of theoretical knowledge have been mentioned before.

Some additional laboratory experiments are sugested to answer some of the remaining
questions. In Figure 13-1 a possible model set-up is proposed that should gain insight in
the following:
- wave pressures are often assumed to satisfy linear wave theory, although especially
in coastal regions waves become non-linear. What water pressures in a vertical
column of water are induced by cnoidal or second order waves? What is the
difference with linear waves? Especially the orbital pressures at the seabed are
important;
- some researchers already stated that the porosity of the seabed influences the wave
pressure at the seabed. The (hydraulic) permeability seems to be important. Is this
phenomenon always important for every wave type?
- how will wave pressures attenuate inside the seabed, considering three different
relative densities (RD = 0,3; 0,5; 0,7), two different grain diameters (d50 = 200;
300 μm) and two different layer thicknesses;
- is there some phase lag noticeable? It was already mentioned that a phase lag could
cause failure, but it is questionable whether such failures do occur in practice.
Figure 13-1: Schematization of laboratory set-up to measure wave pressures outside and
inside a flat seabed for different wave types, sands, relative densities and thickness of the
seabed.
After all measurements on a certain soil are carried out, some soil tests are executed to
obtain the in-situ soil properties (unit weight, porosity, relative density, permeability
and angle of internal friction). These soil tests will be done in the area that will be
excavated to ‘dredge a channel’.
Then the same waves as the previous experiment will be generated, see Figure 13-2.
The left-hand piezometers and stress cells serve as a control system. These measuring
instruments shouls be located so far from the side slope that the test results should not
be affected by the presence of the sloping seabed. The right-hand instruments record
the wave pressures at a steep side slope. It will be studied which waves (H, T, L) cause
slope failure. These failures will be registered on film to answer the following questions:
209
- where do failures start;

- how deep is the failure plane located (macro/micro-instability?);
- what are the consequences of failure (sliding/slumping, liquefaction flow slide,
turbidity current);
- what does the resulting geometry look like? How much material is involved.
different wave types (H, L, T);

similar to previous experiment
water level
piezometers
fresh/salt (water pressure)
water
side original seabed
slop
e: 1
sand :2, 1
total stress cells :3 , 1:4 excavated channel
d50 = 200 / 350 m
RD = 0,3 / 0,5 / 0,7 & piezometers
(pore water pressure)
varying bed thickness to

investigate its influence
Figure 13-2: Schematization of laboratory set-up to measure wave pressures outside and
inside a flat seabed for different wave types, sands, relative densities and thickness of the
seabed.
The following experiments deal with sedimentation behaviour. A unidirectional current

will be generated from left to right. Can the observed velocity pattern well be predicted
by a numerical model? Is this also the case for very steep side slopes, when separation
bubbles appear?
This current supplies large amounts of sediment. Can the sediment transport process
well be predicted, using a numerical model?
Besides the no-wave situation, also small waves are considered. How will the
morphological development of the side slopes be influenced by these waves? What are
the differences in settling behaviour? What is the relative density of the freshly
deposited sediment? Based on the observed morphological development, a certain
moment in time will be chosen at which the side slope may be more susceptible to
failure than the initial situation. Large waves that have proven to be able to cause slope
failure will be generated. Again failure will be monitored. Will this failure mechanism be
equal to the initial situation? Is failure caused by an unfavourable slope geometry
(steeper side slopes) or an unfavourable relative density (more loosely packed
sediment)?
The above described sediment should gain insight in the soil-water-interaction at steep
side slopes. Wave loads act as a sudden release mechanism and a more gradual
transport and compaction mechanism. Based on the experimental data, the numerical
models that predict pore water pressure, consolidation, current velocity profiles and
sediment transport can be improved. Eventually, the ultimate goal is a complete
numerical model that describes submarine slope development in time and that can
handle morphological development as well as slope failure. A set-up of such a model is
described in the next paragraph.
210
13.2.4 Set-up of calculation model
In order to be able to model the interaction between morphology and soil mechanics a
coupled approach is advised. The set-up of a possible model is presented in Figure 13-3.
(initial) channel geometry: current: waves:

h0, h1, D, β, W u(z), αc Hs, T, αw
hydrodynamics: flow properties:

K-ε turbulence closure model κ, ν, ρw
morphology: sediment properties:

sediment transport model to d50, d90, ws, ρs
compute bed level changes
model for gradual loading

in-situ soil properties:
n, RD, φ, c, su, k, cv, E
soil mechanics:
finite elements model to compute
wave compaction:
soil stresses and pore pressures
Hs, T
model for sudden loading
stability check! liquefaction check!

λ<0
failure!
Figure 13-3: Schematization of calculation model set-up
The biggest difficulty that will be encountered when setting up such a sophisticated
interactive model will be the different stage in theoretical foundation of all components.
Therefore some additional research was recommended. Another difficulty is the range of
different time scales one has to cope with when running such a model. One can think
of:
211
- small timescale for turbulent flow and sediment transport;

- medium timescale for adaptation of bed level;
- medium to large timescale for consolidation of (sedimented) layers under the
combination of time and wave action;
- large timescale for stability-checks (slope collapse, slumps or liquefaction flow
slides). Some loads, like earthquakes or storm waves, can be exerted on the side
slopes. For each specific load a safety factor F or even better a probability
distribution of failure can be defined. In this way the behaviour of a navigation
channel throughout time (instead of just the initial stability) can be predicted. An
example of its applicability is for instance a navigation channel that is dredged
during summer under mild hydrodynamic conditions and that will be subjected to
large sediment supply during winter and that will experience severe storms or
earthquakes. The engineer will gain insight in the most critical stage of the channel
during its lifespan and may be able to adapt the current design.
Or course such a complete computer program asks for a lot of effort and cooperation of
two still rather separated disciplines. Besides, many fitting data have to be available and
because only very few projects have been monitored so extensively that all relevant
input parameters are available, laboratory experiments are needed for supplementary
data. These experiments can be done under fully regulated conditions, while
observations and measurements are relatively easy, compared to real projects,
especially because failure mechanisms cannot be controlled.
13.3 Final remarks
Most critical remarks of the readers are implemented in this thesis. My appreciations go
out to all who pointed me at errors, indistinct formulations or newer theories. However,
not all remarks are implemented due to time limitations or differences of opinion.
Relation between drained and undrained soil properties

The estimation formula of the undrained shear strength appeared to be out of date.
Henkel included a correction factor for the second term to account for the effect of over-
or underconsolidation. Because the relation was used for indicative purposes, it was
decided not to execute all undrained MSTAB-calculations again. Besides, the conclusions
are not expected to change.
Ratio between horizontal and vertical earth pressure

The ratio between horizontal and vertical earth pressure was obtained from a linear
elastic stress-strain relationship (equation 4-5). Another well-known expression for the
earth pressure at rest is that of Jacky (1944, as described by Keskin et al. [2004]):
K 0 = 1 − sin φ Eq. 13-1
This equation is more commonly used for determination of K0 in normally consolidated

soils. Keskin et al. [2004] summarized a number of expressions for K0 and conducted a
lot of oedometer-tests. They stated that this factor K0 changes depending on relative
density, stress-history, overconsolidation ratio (OCR), plasticity index and similar soil
properties. Because it is strongly recommended to measure the undrained shear
strength and not to deduce this soil property from cohesion and friction angle, the
calculations have not been executed once again.
212
213
Appendix A: List of references
This list contains all used references and some recommended literature. In the thesis
text, references are indicated by ‘[…]’- signs.
Alfrink, B.J., Van Rijn, L.C., Two-equation turbulence model for flow in trenches. Delft:
WL|Delft Hydraulics, 1983.
Ambrož, J., Slope stability of embankments and excavations. Proceedings Sixth

International Congres of International Association of Engineering Geology. Amsterdam:
A.A.Balkema, 1990.
Asilbekov, D.A., A new approach to solution of slope stability problems. Proceedings

Sixth International Congres of International Association of Engineering Geology.
Amsterdam: A.A.Balkema, 1990.
Atigh, E., Byrne, P.M., Liquefaction flow of submarine slopes under partially undrained
conditions: an effective stress approach. Canadian Geotechnical Journal 41: 154-165,
2004.
Attewell, P.B., Farmer, I.W., Principles of engineering geology. London: Chapman and
Hall Ltd, 1976.
Barends, F.B.J., Thabet, R.A.H., Ground water flow and dynamic gradients. Delft:
Proceedings of the International Symposium on soil mechanics research and foundation
design for the Oosterschelde Storm Surge Barrier, volume 2, October 9-12, 1978.
Barends, F.B.J., Ruygrok, P.A., The cyclic liquefaction potential of a submerged

cohesionless sand bed; a practical method. Boss '97 Eighth International Conference on
the behaviour of offshore structures, pp 71-84, 1997.
Berg, J.H. van den, Gelder, A. van, Mastbergen, D.R., The importance of breaching as a
mechanism of subaqueous slope failure in fine sand. Sedimentology 49, 81-95, 2002.
Bezuijen, A., Mastbergen, D.R., On the construction of sand fill dams – part 2: soil
mechanical aspects. Rotterdam: A.A.Balkema, 1988.
Biscontin, G., Pestana, J.M., Nadim, F., Seismic triggering of submarine slides in soft
cohesive soil deposits. Marine Geology 203 341-354, 2004.
Bliek, A.J., Svasek, J.N., On slope stability of dredged trenches. Amsterdam: CEDA
Dredging Days 1987.
Boehmer, J.W., Borst, W.G., Svasek, B.V., Bras A., Van Raalte, G.H., Slope stability and
slope production tests. A new tool in harbour design and dredging practice. Singapore:
World Dredging Congress 1983.
Buonaiuto, F.S., Kraus, N.C., Limiting slopes and depths at ebb-tidal shoals. Coastal
Engineering 48 51-65, 2003.
Camenen, B., Larroudé, P. (2003), Comparison of sediment transport formulae for the
coastal environment. Journal of Coastal Engineering 48, 111-132
Clough, G.W., Iwabuchi, J., Rad, N.S., Kuppusamy, T., Influence of cementation on
liquefaction of sands. Journal of Geotechnical Engineering, Vol 115, No 8 pp. 1102-
1117, 1989.
Coppoolse, R.C., Schoonees, J.S., Maintenance dredging at Richards Bay: Sensitivity

analysis of a predicitive model and the location of an offshore dump site. Amsterdam:
CEDA-PIANC Conference 1991.
214
CUR, Artificial sand fills in water, report 152. Stichting CUR Centre for Civil Engineering
research and codes. Gouda: 1992.
CUR 162, Construeren met grond. Constructies op en in weinig draagkrachtige en sterk

samendrukbare ondergrond. Gouda, 1992.
Das, B.M., Principles of Geotechnical Engineering. 3rd Edition. Boston: PWS Publishing
Co., 1994.
Davies, A.G., Van Rijn, L.C., Damgaard J.S., Van de Graaff J., Ribberink, J.S.,
Intercomparison of research and practical sand transport models. Coastal Engineering
46, p1-23, 2002.
Delft Cluster, Samenvatting dijkdoorbraakprocessen. Beschrijving initiërende

faalmechanismen doorbraakprocessen en reststerkte, 2002.
Demars, K.R., Transient stresses induced in sandbed by wave loading. ASCE Journal of
Geotechnical Engineering Division. Vol 109 No 4 1983, p 591-602, 1983.
Dhat, N.R., Chhatre, M.V., et al., Wave and earthquake induced liquefaction potential of
silty sand form sea bed. Rotterdam: A.A.Balkema, 1988.
Dolinar, B., Trauner, L., Undrained shear strength in dependance on the quantity of free
water and firmly absorbed water in fully saturated clays. Maribor, University of Maribor,
2000.
DWWwijzer, nummer 66, Zettingsvloeiingen, 1995.
Evans, M.D., Zhou, S., Liquefaction behaviour of sand-gravel composites. Journal of

Geotechnical Engineering, Vol 121 No 3, pp. 287-298, 1995.
Fredsøe, J., Deigaard,R., The state of the art of predicting natural backfilling of sand in
navigation channels. In: Proceedings of XIth World Dredging Congress 1986, Brighton
UK, WODA, 1986.
Fredsøe, J., Deigaard,R., Mechanics of coastal sediment transport. Singapore: World

Scientific, 1995.
Gebruikershandleiding Mcycle. Grondmechanica Delft, 1990.
GeoDelft, Mstab user manual, release 9.7. September 2002.
Graaff, J. van de, Coastal morphology and coastal protection. Lecture notes ct5309.
Delft: 2003.
Grondboek Boskalis
Groot, M.B. de, den Adel, H., Stoutjesdijk, T.P., van Westenbrugge, C.J., Risk of dike
failure due to flow slides. Coastal Engineering 26 241-249, 1995.
Groot, M.B. de, Stoutjesdijk, T.P., Undrained stress path of loose sand predicted from
dry tests. Canadian Geotechnical Journal 34: 131-138, 1997.
Groot, M.B. de, Liquefaction, flow and settlement of sand in dredging, hydraulic fill and
flow slides. Proc. 3rd International Conference on Civil and Environmental Engineering
2004, University of Hiroshima, Dep Soc. Environ., 2004.
Hassona, F.A.K., The effect of fines on the liquefaction behaviour of cohesionless soils.
Proceedings Sixth International Congres of International Association of Engineering
Geology. Amsterdam: A.A.Balkema, 1990.
215
Havinga, F.J., Sediment concentrations and sediment transport in case of irregular non-
breaking waves with a current. Delft: WL|Delft Hydraulics, 1992.
Heezen, F.T., Van der Stap, A.C.M., An engineering approach to under water dumped
sandbodies. Rotterdam: A.A.Balkema, 1988.
Herbich, J.B., Handbook of dredging engineering. New York, McGraw-Hill Inc, 1992.
Ishihara, K., Yamazaki, A., Wave-induced liquefaction in seabed deposits of sand.

Seabed mechanics, 1983.
Jenniskens, M.J.J., Comparison of the sediment transport formulae according to Bijker

and Van Rijn. Thesis. Delft: Delft University of Technology, 2001.
Jensen, J.H., Madsen, E.Ø., Fredsøe, J., Oblique flow over dredged channels, I: Flow
description, II: Sediment transport and morphology. Journal of Hydraulic Engineering,
Vol. 125, No. 11, p.1181-1198, 1999.
Kaya, A. Evaluation of soil porosity using a low MHz Range Dielectric Constant. Izmir:
Dokuz Eylül University, 2002.
Keskin, S.N., Tekinsoy, M.A., Uzundurukan S., The effects of over consolidation ratio
and effective stresses to the earth pressure at rest at clay soils. Digest 2004, p947-961,
2004.
Kessel, T. van, Kranenburg, C., Wave-induced liquefaction and flow of subaqueous mud
layers. Coastal Engineering 34, 109-127, 1998.
Klein, M.D., Large scale sandpits. Thesis. Delft: Delft University of Technology, 1999.
Klein, J.A., Verification of morphodynamic models on channels, trenches and large scale
mining pits in water depths greater than 20 m. Thesis. Delft: Delft University of
Technology, 2003.
Köhler, H., Druckwechselbelastung an Wasser-Boden-Grenzflächen. Z. angew. Geol. 47,

2001.
Kolkman, P.A., Lindenberg, J, Pilarczyk, K.W., Modelling soil-water-structure

interactions, SOWAS 88 Delft. Rotterdam: A.A.Balkema, 1988.
Kraft Jr., L.M. ; Gavin, T.M. ; Bruton, J.C., Submarine flow slide in Puget Sound. In:
Journal of Geotechechnical Engineering Vol. 118 No. 10, pp. 1577-1591, 1992.
Lade,P.V., Yamamuro, J.A., Effects of nonplastic fines on static liquefaction of sands.

Canadian Geotechnical Journal 34, pp 918-928, 1997.
Lean, G.H., Estimation of maintenance dredging for navigation channels. Oxfordshire,

Hydraulics Research Station Wallingford, 1980.
Ledden, M. van, Sand-mud segregation in estuaries and tidal basins. Ph.D. Thesis.
Delft: Printpartners Ipskamp BV, 2003.
Liam Finn, W.D., Siddharthan, R., Martin, G.R., Response of seafloor to ocean waves.
ASCE Journal of the Geotechnical Engineering Division. Vol 109 No 4 1983, p 556-572,
1983.
Loman, G.J.A., Prediction of channel infill: a concerted survey and modelling approach.
Amsterdam, CEDA-PIANC Conference 1991.
216
Mangor, K., Dredging a trench across the surf zone at an exposed coast. In:
Proceedings of XIth World Dredging Congress 1986, Brighton UK, WODA, 1986.
Mastbergen, D.R., Berg, J.H. van den, Breaching in fine sands and the generation of
sustained turbidity currents in submarine canyons. Sedimentology 50, p625-637, 2003.
Mastbergen, D.R., Stoutjesdijk, T.P., Overheid scherpt regelgeving aan voor diepe
zandwinputten om oeverbressingen tegen te gaan. Optimum, Informatiemagazine voor
de GWW sector, Nr.9, p6-9, 2002.
Mastbergen, D.R., Winterwerp, J.C., Bezuijen, A., On the construction of sand fill dams
– part 1: hydraulic aspects. Rotterdam: A.A.Balkema, 1988.
Mikkelsen, L., Mortensen, P., Sorensen, T., Sedimentation in dredged navigation

channels. Copenhagen: Danish Hydraulic Institute, 1980.
Mitchell, R.J., Tsui, K.K., Sangrey, D.A., Failure of submarine slopes under wave action.
Vancouver: Proceedings of the Thirteenth Coastal Engineering Conference, 1972.
Molenkamp, F., Static and dynamic analysis of offshore gravity structures. PhD Thesis.
Manchester: University of Manchester, 1977.
Molenkamp, F., Liquefaction as an instability. Artikel ICSMFE, 1989
Molenkamp, F., Choobbasti, A.J., Heshmati, A.A.R., Modelling of liquefaction and flow of
water saturated soil. Manchester, Imperial College Press, 1998.
Moser, D., L. Skaggs, J. Lund, S.J. Ratick, H. Morehouse Garriga, W. Du, and R.
Klimberg, Risk-based planning and management of maintenance dredging: development
and application of the reliability-based dynamic dredging decision (RBD3) model. PIANC
Bulletin No 87, 1995.
Nataraja, M.S., Gill, H.S., Ocean wave induced liquefaction analysis. ASCE Journal of
Geotechnical Engineering, Vol 109 No 4, p 573-590, 1983.
Nieuwenhuis J.D., Geest, J.M. van, Molenkamp, F., Interaction forces between piers and
sill structure. Delft: Proceedings of the International Symposium on soil mechanics
research and foundation design for the Oosterschelde Storm Surge Barrier, volume 2,
October 9-12, 1978.
Pitman, T.D., Robertson, P.K., Sego, D.C., Influence of fines on the collapse of loose
sands. Canadian Geotechnical Journal 31, pp 728-739, 1994.
Postma, H., Slope stability evaluation based on practical experience. Amsterdam: CEDA
Dredging Days 1987.
Raalte, G.H. van, Can dredging industry rely on geotechnics? from Geotechnical
Engineering for Transportation Infrastructure, Volume 2. Rotterdam: A.A. Balkema,
1999.
Schwarz, H.U., Subaqueous slope failures - Experiments and modern occurrences.
Stuttgart: E. Schweizerbart'sche Verlagsbuchhandlung, Contributions to Sedimentology,
1982.
Seed, H.B., Idriss, I.M., Simplified procedure for evaluating soil liquefaction potential.
Proceeding ASCE Vol 97 SM 9, 1971.
Seed, H.B., Idriss, I.M., Arango, I., Evaluation of liquefaction potential using field
performance data. In: Journal of the Geotechnical Engineering Division Vol 109 No 4,
1983.
217
Silvis, F., Lindenberg, J., van Heteren, J., A numerical model to describe the initiation of
flow slides in under water sand slopes. Rotterdam: A.A.Balkema, 1988.
Silvis, F., Groot M.B. de, Flow slides in the Netherlands: experience and engineering
practice. Canadian Geotechnical Journal, No 32, p1086-1092, 1995
Sistermans, P.G.J., Graded sediment transport by non-breaking waves and a current.

Ph.D. Thesis. Delft: Printpartners Ipskamp BV, 2002.
Smits, F.P., Andersen, K.H. Gudehus, G., Pore pressure generation. Proceedings of the
International Symposium on Soil Mechanics Research and Foundation Design for the
Oosterschelde Storm Surge Barrier, Foundation Aspects of Coastal Structures, October
9-12, paper II-3, 1978.
Southern California Earthquake Center, Recommended procedures for implementation

of dmg special publication 117. Guidelines for analyzing and mitigating liquefaction
hazards in California. 1999.
Spierenburg, S.E.J., Seabed response to water waves. Phd Thesis. Delft: University of
Technology, 1987.
Stam, F., Stability of underwater slopes in loose sand during and after dredging.
Brighton: XIth World Dredging Congress 1986.
Stark, T.D., Olson, S.M., Liquefaction resistance using CPT and field case histories.
Journal of Geotechnical Engineering, december 1995, pp 856-869, 1995.
Stelling, J.G., Maintenance dredging in the Netherlands; efficiency in relation to

morphological, technical and environmental aspects. Amsterdam: CEDA Dredging Days
1987.
Stoutjesdijk, T.P., De Groot, M.B., Handboek zettingsvloeiing. Delft: Grondmechanica

Delft, 1994.
Stoutjesdijk, T.P., Heemstra, J. et al.., Computermodel helpt stabiliteit van taluds onder
water te beoordelen. Land en Water, jaargang 35, No. 10, p41-45, 1995.
Stoutjesdijk, T.P., De Groot, M.B., Lindenberg, J., Flow slide prediction method:
influence of slope geometry. NRC Canada, Canadian Geotechnical Journal 35, p43-54,
1998.
Terzaghi, K., Peck, P.B., Mesri, G., Soil mechanics in engineering practice. New York:
Wiley, 1996.
Towhata, I., Kim, S.R., Undrained strength of underconsolidated clays and its
application to stability analysis of submarine slopes under rapid sedimentation. Soils
and Foundations, 30(1990)1, March, pp. 100-114, 1990.
Tsui, Y., Helfrich, S.C., Wave-induced pore pressures in submerged sand layer. ASCE
Journal of the Geotechnical Engineering Division. Vol 109, No 4, p603-618, 1983.
Tyshing, R., Sustainable dredging. PIANC Congress, Sydney, 2002.
US Army Corps of Engineers, Slope Stability Engineering Manual. Washington, 2003.
Veen, R. van der, How harbour construction can influence maintenance dredging –
results of some research for the Port of Delfzijl. Amsterdam, CEDA-PIANC Conference
1991.
Rhee, C. van, Bezuijen, A., Influence of seepage on stability of sandy slope. Journal of
Geotechnical Engineering. Vol 118, No 8, 1992.
218
Rijn, L.C. van, Sediment transport, part I: Bed load transport. Journal of Hydraulic
Engineering. Vol 110 No 10, 1984a.
Rijn, L.C. van, Sediment transport, part II: Suspended load transport. Journal of
Hydraulic Engineering. Vol 110 No 11, 1984b.
Rijn, L.C. van, Sediment transport, part III: Bed forms and alluvial roughness. Journal
of Hydraulic Engineering. Vol 110 No 12, 1984c.
Rijn, L.C. van, Tan, G.L., SUTRENCH-model; Two-dimensional vertical mathematical

model for sedimentation in dredged channels and trenches by currents and waves. Den
Haag: Rijkswaterstaat Communications No 41, 1985.
Rijn, L.C. van, Sedimentation of dredged channels by currents and waves. In: Journal of
Waterway Port Coastal and Ocean Engineering Vol 112 No 5, 1986a.
Rijn, L.C. van, Mathematical modelling of suspended sediment in nonuniform flows.

Journal of Hydraulic Engineering. Vol 112 No 6, 1986b.
Rijn, L.C. van, Principles of coastal morphology. Amsterdam: Aqua Publications, 1998.
Rijn, L.C. van, General view on sand transport by currents and waves: data analysis and
engineering modelling for uniform and graded sand (TRANSPOR-2000 and CROSMOS-
2000 models. Delft: Report Z2899.20/Z2099.30/Z2824.30, WL|Delft Hydraulics, 2000.
Rijn, L.C. van, Davies, A.G., Van de Graaff, J., Ribberink, J.S., SEDMOC Sediment
transport modelling in marine coastal environments. Amsterdam: Aqua Publications,
2001.
Rijn, L.C. van, Walstra, D.J.R., Morphology of pits, channels and trenches. Part I:
Literature review and study approach. Delft: 2002.
Vellinga, T., The origin and dispersal of silt and measures to reduce the volume of
dredging. Amsterdam: CEDA Dredging Days 1987.
Verruijt, A., Bruijn, V.M.P.J. de, Muralidhar, B., Liquefaction around buried pipe lines.
Boss '97 Eighth International Conference on the behaviour of offshore structures, pp 99-
108, 1997.
Verruijt, A., Grondmechanica. Delft: Delft University Press, 1999.
Walstra, D.J.R., Van Rijn, L.C., Aarninkhof, S.G.J., Sand transport at the lower
shoreface of the Dutch coast. Delft: WL|Delft Hydraulics, 2000.
Walstra, D.J.R., Van Rijn, L.C., Hoogewoning, S.E., Aarninkhof, S.G.J., Modelling of
sedimentation of dredged trenches and channels under the combined action of tidal
currents and waves. Delft: WL|Delft Hydraulics, 2000.
Whitehouse, R., Soulsby, R., Roberts, W., Mitchener, H., Dynamics of estuarine muds.
London: Thomas Telford Ltd, 2000.
Whitely, B., Integrating geophysical and geotechnical technologies for improved site
assessment of ports and harbours. Sydney: 30th PIANC-AIPCN Congress 22-26
september 2002.
Wit, P.J. de, Liquefaction of cohesive sediments caused by waves. Delft: Delft University
Press, 1995.
WL| Delft Hydraulics, Taludinstabiliteit en veiligheid bij diepe zandwinningen in

Overijssel. Delft: 2001.
219
Appendix B: List of symbols

Symbol Description unit
Latin
a wave-induced particle excursion m
or thickness of bed load layer, reference level m
c cohesion kPa
cK-C coefficient in formula of Kozeny-Carman; 0,01 -
cv consolidation coefficient m2/s
d distance between plates in capacitance measurement m
ds thickness of the seabed m
d50 particle diameter at which 50% by weight is finer m
d90 particle diameter at which 90% by weight is finer m
d* dimensionless particle diameter -
e void ratio -
or trapping efficiency -
emax maximum void ratio -
emin minimum void ratio -
fw friction factor for waves -
g constant of gravity m/s2
h water depth m
h' effective water depth of Tsui et al. m
h0 surrounding water depth m
h1 water depth in trench/channel m
hc critical wall height of purely cohesive soil m
i water surface slope -
k wave number; 2π / L rad/m
or hydraulic conductivity, permeability m/s
kr effective bed roughness m
kr;c current related bed roughness m
kr;w wave related bed roughness m
kx permeability in x-direction m/s
kz permeability in z-direction m/s
m fit constant of dilatation curve (SLIQ2D) -
2
mv coefficient of vertical compression m /MN
n porosity -
nmax maximum porosity -
nmin minimum porosity -
p (pore) water pressure kPa
p0 amplitude of wave pressure at seabed kPa
r fit constant of dilatation curve (SLIQ2D) -
s specific density (ρs/ρw) -
smax asymptote of relative shear stress (SLIQ2D) -
su undrained shear strength kPa
s2 relative shear stress at which εvoldm0 occurs (SLIQ2D) -
u excess pore pressure (EPP) kPa
or horizontal flow veloctiy in x-direction, perpendicular to channel axis m/s
or power of decompression curve (SLIQ2D) -
uh flow velocity at water surface m/s
220
uh;e equilibrium flow velocity at water surface m/s

u* bed shear velocity m/s
v horizontal flow veloctiy in x-direction, parallel to channel axis m/s
or power of isotropic stress (SLIQ2D) -
w vertical flow veloctiy in z-direction m/s
wL liquid limit %
wP plastic limit %
ws particle fall velocity m/s
x horizontal axis, perpendicular to trench/channel axis; positive in flow direction m
y horizontal axis, parallel to trench/channel axis; positive in flow direction m
z vertical axis; positive upward m
z0 zero-velocity level m
za reference level of bed form height m
zb bed level m
Latin Capitals
A parameter which represents internal generation of water pressure due to waves
or surface of plates in capacitance measurement m2
or auxiliary parameter of dilatatant volume strain -
B auxiliary parameter of dilatatant volume strain -
Ba Bagnold number
C Chézy coefficient m0,5/s
or capacitance F=V/C
CSR Cyclic Stress Ratio (=Δσ'/σ') -
D trench/channel depth D = h1 - h0 m
Dinit initial trench/channel depth m
D90 trench/channel depth after 90 days m
Dreq required depth of trench/channel m
E compression modulus, bulk modulus MPa
EEP excess pore pressure kPa
F stability factor of slope stability -
Fg force exerted by grains kN
Fr Froude number
Fw friction force kN
G shear modulus MPa
GC gravel content of soil %
H wave height m
He equivalent wave height (relative shear stress changes) m
Hs significant wave height m
K Young's modulus MPa
Kfs normally distributed factor of erosion length due to a flow slide -
Ks decompression modulus (SLIQ2D) kPa
Ks0 decompression modulus at begin of decompression (SLIQ2D) kPa
Kα,β slope reduction factor for grain stability as function of flow angle and slope angle -
L wavelength m
Lflowslide erosion length due to flow slide m
N number of horizontal grid points in SUTRENCH -
Nl0 number of load cycles to cause cyclic liquefaction in undrained conditions -
221
Nl0;hist number of load cycles to cause cyclic liquefaction after a certain history (compaction -
due to drainage)
Nt number of time steps -
NV number of vertical grid points in SUTRENCH -
Q discharge m3/s
RD relative density based on minimum and maximum void ratios -
Re Reynolds number
T wave period s
or (water) temperature °
Te equivalent wave period (relative shear stress changes) (Mcycle) s
Tp peak wave period s
Ts wave period belonging to significant wave height s
V flow velocity m/s
Vg volume of the grains in a soil sample m3
Vp volume of the pores in a soil sample m3
Vt total volume of a soil sample m3
W width of the trench/channel m
Wreq required width of trench/channel m
Greek
α0 approach angle of current to channel axis °
α1 angle of current inside channel °
αc-w angle between current and wave action °
αbr breaking coefficient representing influence of breaking waves on the sediment -
mixing process
αeq 'equilibrium' angle of current inside channel °
αc-w angle between current and wave action °
αs numerical smoothing coefficient for bed level changes in SUTRENCH -
αw0;crit critical wave angle to channel axis for which waves are refracted parallel to the °
channel axis
β ratio sediment mass mixing and fluid momentum mixing coefficients -
or compressibility of water
βinit angle of side slope °
βdown;90 angle of downstream side slope after 90 days °
βup;90 angle of upstream side slope after 90 days °
δ thickness of wave boundary layer near bed m
ε dielectric permittivity of medium in capacitance measurement F/m=C/Vm
ε0 dielectric permittivity of vacuum = 8,854*10^-12 F/m F/m=C/Vm
εf fluid mixing coefficient m2/s
εhist constant representing 'history effect' of cyclic loading -
εs sediment mixing coefficient m2/s
εvold dilatant volume strain -
εvoldm0 maximum dilatant volume strain -
γdry unit weight of dry soil kN/m3
γm shear strain rate of soil-water mixture 1/s
γsat unit weight of wet saturated soil kN/m3
γunsat unit weight of wet unsaturated soil kN/m3
γw unit weight of water kN/m3
γxy shear strain -
η coordinate perpendicular to slope; positive into the soil -
222
or coefficient of sediment mixing coefficient; 2 for current alone; 1-2 for current with -
waves
κ constant of Von Karman -
or intrinsic permeability m2
λ eigenvalue of stiffness matrix (static liquefaction) -
μ dynamic viscosity kg/ms
ν kinematic viscosity m2/s
or Poisson's ratio -
φ angle of repose °
or turbulence damping factor -
ξ coordinate along slope -
or auxiliary parameter representing friction caused by currents and waves -
ρs density of solid sediment material kg/m3
ρsat density of completely saturated soil kg/m3
ρunsat density of wet, but not completely saturated soil kg/m3
ρw density of water kg/m3
σ total stress kPa
σ' effective grain stress kPa
σd single amplitude cyclic axial stress in triaxial test kPa
σ'0 initial effective confining stress in triaxial test kPa
σ'v0 initial effective stress in simple shear test kPa
σ'vol;0 istropic stress at begin of decompression kPa
τb bed shear stress kPa
τc current-induced bed shear stress kPa
τcw bed shear stress caused by the combined action of current and waves kPa
τh double amplitude cyclic shear stress in simple shear test kPa
τw wave-induced bed shear stress N/m2
τxy shear stress kPa
ω wave frequency (=2π/T) rad/s
Greek Capitals
Δ relative density -
Ψc stability parameter of movement -
223
di
am m
ed
ia et
n er
pa at pa
rti
cl w rti
e hi
un c h co cl
e
un d di
a 9 n fa
de sa sa ra
i m 0 %
hy i n s o ll
ns tu tu n e e d r tr l i
s a ra ra d t by a u i ns d a
ve
ity re tu te te
er
ic
lo e
la d
dr
y d f
sh of w lic tio ci
so
lid tiv
ra
te u r i c ea t o ei c p e n ty Po las
tic
e d
un dr
y
un ni t rs ta gh on rm co at is
m
m po vo
i
it
w
it t w ion coh lw ti du e ef so
at ro d de de de w tre s ct ab fic
T= n od
er s ra ns ns ei
g ns ei
g
ei
g
an es n g
ei
g f in i v i i e
18 ra ul
ia tio h it h h gl
e
io
n t h h e it li t n ° C ti o us
l
ity ity ity
t y t t t r y y t
Name Description ρs n e RD ρsat γsat ρdry γdry γunsat ϕ c su d50 d90 k κ cv ws ν E
3 3 3 3 3 3 2
kg/m - - - kg/m kN/m kg/m kN/m kN/m ° kPa kPa μm μm m/s m m/s - MPa
LCS loosely packed coarse sand 2650 0,45 0,82 0,16 1908 19 1458 15 17 30 0 0 1000 1500 1,9E-04 1,0E-08 3,8E-01 0,1084 0,30 20
MCS medium packed coarse sand 2650 0,40 0,67 0,43 1990 20 1590 16 18 35 0 0 1000 1500 3,5E-05 1,8E-09 1,2E-01 0,1084 0,35 35
DCS densely packed coarse sand 2650 0,30 0,43 0,86 2155 22 1855 19 20 40 0 0 1000 1500 1,2E-05 5,5E-10 5,9E-02 0,1084 0,40 50
LMS loosely packed medium sand 2650 0,45 0,82 0,16 1908 19 1458 15 17 30 0 0 500 750 1,4E-05 7,5E-10 2,9E-02 0,0663 0,30 20
MMS medium packed medium sand 2650 0,40 0,67 0,43 1990 20 1590 16 18 35 0 0 500 750 8,8E-06 4,4E-10 3,1E-02 0,0663 0,35 35
224
DMS densely packed medium sand 2650 0,30 0,43 0,86 2155 22 1855 19 20 40 0 0 500 750 3,0E-06 1,4E-10 1,5E-02 0,0663 0,40 50
LFS loosely packed fine sand 2650 0,45 0,82 0,16 1908 19 1458 15 17 30 0 0 200 300 2,3E-06 1,2E-10 4,6E-03 0,0252 0,30 20
MFS medium packed fine sand 2650 0,40 0,67 0,43 1990 20 1590 16 18 35 0 0 200 300 1,4E-06 7,1E-11 5,0E-03 0,0252 0,35 35
DFS densely packed fine sand 2650 0,30 0,43 0,86 2155 22 1855 19 20 40 0 0 200 300 4,7E-07 2,2E-11 2,4E-03 0,0252 0,40 50
LSI loosely packed silt 2650 0,45 0,82 0,16 1908 19 1458 15 17 27 0 0 50 75 1,4E-07 7,5E-12 2,9E-04 0,0029 0,30 20
MSI medium packed silt 2650 0,40 0,67 0,43 1990 20 1590 16 18 32 0 0 50 75 8,8E-08 4,4E-12 3,1E-04 0,0029 0,35 35
DSI densely packed silt 2650 0,30 0,43 0,86 2155 22 1855 19 20 37 0 0 50 75 3,0E-08 1,4E-12 1,5E-04 0,0029 0,40 50
SOC soft (organic) clay, low plasticity 2650 0,75 3,00 1413 14 663 7 11 18 5 5 2 3 3,8E-09 2,7E-13 7,6E-07 0,20 2
MEC medium clay, medium plasticity 2650 0,60 1,50 1660 17 1060 11 14 18 7 20 2 3 9,0E-10 5,4E-14 4,5E-07 0,30 5
STC stiff clay, high plasticity 2650 0,40 0,67 1990 20 1590 16 18 20 10 50 2 3 1,4E-10 7,1E-15 1,4E-07 0,40 10
Appendix C: Soil properties of reference soils
225

2005 Raaijmakers

Uploaded by

Copyright:

Available Formats

2005 Raaijmakers

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2005 Raaijmakers

Uploaded by

Copyright:

Available Formats

Final Thesis Report

in cooperation with TU Delft and Hydronamic

Dredged Trenches and Channels

Final Thesis Report

in cooperation with TU Delft and Hydronamic

Of course I am most grateful to my supervisors of the examination committee, consisting

7.2 WAVES ......................................................................................................................................... 112

13.2.2 Applicability in dredging practice ....................................................................................... 208

The width-averaged 2DV-program SUTRENCH developed by WL|Delft Hydraulics was

1.2 Problem description

When designing a navigation channel, usually attention is given to:

Another remarkable thing is that research on slope development is conducted by either

In engineering practice it is found very difficult to give an estimation of the relevant

Definition of the problem:

1.3 Aims and objectives

The aims and objectives of this thesis are:

With the following restrictions:

Capital Period of Interest Maintenance

Figure 1-1: Schematization of ‘period of interest’

1.4 Outline of the report

2 Mechanisms of slope instability

2.1 Known mechanisms of slope failure in literature

Subaqueous slope failures occur at various scales in different environmental settings.

Fracturing / Folding Break or bending of (an)isotropic beds

Another classification in grain-fluid mixture flows is adopted in Mastbergen et al. [1988]

Reynolds number (Re) ----------->

<-------------- Admission of water

Relative grain density ---------->

Couette flow Suspended sand

Macroviscous Turbidity currents

grain flows Hyperconcentrated

Dry sand flow

Figure 2-3: Percentage of reported slope failures; classification in regions

Figure 2-4: An overview of all kinds of slope failures

Figure 2-5: Percentage of reported slope failures; classification in slope angles

Classification in release mechanisms

2.2 Relevant mechanisms of slope instability in trenches and

In Figure 2-7 an overview of these mechanisms is given. Usually a distinction is made

As a rough design criterion, it is stated that:

- macro-instability should be prevented at all times, because large movements of

Macro-instability Micro-instability Morphology

macro-shear flow slide / retrogressive micro-shear seepage transport of

depth D < Drequired

‘unwanted’ repair activities

Figure 2-7: Instability mechanisms

2.3 Experiences from dredging practice

be deduced from only a few experiences. Besides, understanding of the important

Nowadays, in engineering practice the design of a slope generally is based on:

Trial slopes and slope production cuts 1980

Box Cut Slope Cut

Figure 2-8: Schematization of methods of ‘box cut’ and ‘slope cut’

The most relevant conclusions were:

Pipeline trench at Danish North Sea Coast 1982

Figure 2-9: Cross-section of trench design [source: Mangor, 1986]

Puget Sound 1986

Postma and his ‘equilibrium slopes’ 1987

Limiting slopes at ebb-tidal shoals

3.1.1 Classification of soil types

The International Association of Dredging Contractors established a system for

This results in a kinematic viscosity of 1,3110-6 m2/s in water of 10 °C and 1,0510-6