Advanced Series on Ocean Engineering — Volume 4

COASTAL BOTTOM
BOUNDARY LAYERS AND
SEDIMENT TRANSPORT

Peter Nielsen
World Scientific
ABOUT
THE BOOK
This book is intended as a handbook

graduate students in these fields. With its

emphasis on boundary layer flow and
basic sediment transport modelling, it
is meant to help fill the gap between
general hydrodynamic texts and descrip-
tive and coastal sedimentary processes.
The book commences with a review of

concept of eddy viscosity for theseflows

is discussed | in Sahl because of its

over flat bedsare discussed. Small scale

coastal bedforms and the
hydraulic roughness are described.
The motion ofsuspended sand particles
is studied in detail with emphasis on the

nisms in coastal flows. Sediment pickup

functions are provided for unsteady flows.
A new combined convection-diffusion
model is provided for suspended sedi-
ment distributions. Different methods of
sediment transport model building are

models.
 COASTAL BOTTOM
BOUNDARY LAYERS AND
SEDIMENT TRANSPORT
ADVANCED SERIES ON OCEAN ENGINEERING

Series Editor-in-Chief
Philip L-F Liu
Cornell University, USA

Vol. 1 The Applied Dynamics of Ocean Surface Waves

by Chiang C Mei (MIT, USA)
Vol. 2 Water Wave Mechanics for Engineers and Scientists
by Robert G Dean (Univ. Florida, USA) and Robert A Dalrymple
(Univ. Delaware, USA)
Vol. 3 Mechanics of Coastal Sediment Transport
by Jorgen Fredsoe and Rolf Deigaard (Tech. Univ. Denmark, Denmark)
Vol. 4 Coastal Bottom Boundary Layers and Sediment Transport
by Peter Nielsen (Univ. Queensland, Australia)

Forthcoming titles:
Water Waves Propagation Over Uneven Bottoms
by Maarten W Dingemans (Delft Hydraulics, The Netherlands)
Ocean Outfall Design
by lan R Wood (Univ. Canterbury, New Zealand)
Tsunami Run-up
by Philip L- F Liu (Cornell Univ.), Costas Synolakis (Univ. Southern California),
Harry Yeh (Univ. Washington) and Nobu Shuto (Tohoku Univ.)
Physical Modules and Laboratory Techniques in Coastal Engineering
by Steven A Hughes (Coastal Engineering Research Center, USA)
Advanced Series on Ocean Engineering — Volume 4

COASTAL BOTTOM
BOUNDARY LAYERS AND
SEDIMENT TRANSPORT

Peter Nielsen
The University of Queensland

Yo World Scientific
Singapore « New Jersey * London * Hong Kong
Published by
World Scientific Publishing Co. Pte. Ltd.
P O Box 128, Farrer Road, Singapore 9128
USA office: Suite 1B, 1060 Main Street, River Edge, NJ 07661
UK office: 73 Lynton Mead, Totteridge, London N20 8DH

Library of Congress Cataloging-in-Publication data is available.

COASTAL BOTTOM BOUNDARY LAYERS AND

SEDIMENT TRANSPORT

Copyright © 1992 by World Scientific Publishing Co. Pte. Ltd.

All rights reserved. This book, or parts thereof, may not be reproduced in any form
or by any means, electronic or mechanical, including photocopying, recording orany
information storage and retrieval system now known or to be invented, without
written permission from the Publisher.

ISBN 981-02-0472-8
981-02-0473-6 (pbk)

Printed in Singapore by JBW Printers & Binders Pte. Ltd.

To Felicity and the boys
 — ;

a 4 Woe re
‘ | = gstied
mts ™AAdte FA Ww ; ’ iad 7 i :

ths Bed Giger 1! 1 is

a,
“we
~
: ¢ : pe, SIS) SS bom Pee mw, VOM ‘ne ve

i -“" & @ =>). =| 7

is
s
a ; go 7 -

| 7 :
ain > pi
en —
~ i :
- } ie ~ - : a

tone” Galligan
gue iesealable;
= =

ai }
~ >= —
=) i

awiod
SS
Sitos yey:
id

, Take et
forge © Teaag Shutrinebindlie |
+ vipa somes’ ite bei,
© () 2 1)a,

iKsh. €2:.0)-M74
G8. 3S Qhey
 PREFACE

Coastal process modelling as an art and a science has progressed significantly

over the last four decades. The main areas of progress have been the development
of more realistic models of natural waves and the discovery of the wave thrust or
radiation stress which enables the quantitative description of phenomena like
longshore currents and wave setup.
Progress has been slower in the area of quantitative morphodynamic
modelling. The reason is that the quantitative link has been missing between the
main flow and the rate of morphological change. The required link consists of
efficient models of the boundary layer flow and the resulting sediment transport.
These are the topics of the present book.
The book has two main objectives. The first is to provide a review of coastal
bottom boundary layer flow and sediment transport by summarising the presently
available experimental data. The second objective is to provide the basic sediment
transport models which can be used as building blocks for comprehensive models
of coastal sediment transport.
In order to address both of these objectives, the treatment of major topics like
oscillatory boundary layer flow and sediment suspension are given in two parts
each. The first parts each attempt to review the experimental evidence without
reference to any particular theory. The second parts subsequently deal with state of
the art of modelling.
It is hoped that the book will be useful for Earth Scientists and Engineers
who work in the general area of coastal process modelling and to graduate students
in the areas of Marine Geology, Coastal Morphodynamics and Coastal
Engineering. For such students, the book should help bridge the gap in the
literature between general texts on Coastal Hydrodynamics and descriptive texts on
coastal processes. It is also hoped that the student’s supervisors will find
something new and useful in the book.
Indexing and many cross references are provided in order to make the book
more efficient as a handbook for practising professionals.
The structure of the book can be summarised as follows.
Section 1.1 gives a general introduction to coastal bottom boundary layers
and the flow above them. This includes a discussion of the equations of motion for
incompressible boundary layer flow. The major new contribution in this area is the
introduction of a set of equations which are similar to the classical Reynolds

vii
Preface

equations for turbulent, steady flow. The new equations, however, consider
explicitly the periodic velocity component as well as the time-average and the
random component. These equations, like the Reynolds equations for steady
turbulent flow, are very helpful in analysing the mechanisms of momentum
transfer in combined wave-current flows.
Section 1.2 presents a review of wave boundary layers. It is followed by a
discussion of oscillatory boundary layer models in Section 1.3.
Section 1.4 discusses the wave-generated boundary layer currents including
boundary layer streaming and the surf zone undertow.
Wave-current boundary layer interaction is the subject of Section 1.5.
Experimental data on wave energy dissipation and on wave boundary layer
structure in the presence of currents indicate that currents of typical, relative
strengths have very little effect on the wave boundary layer structure. On the other
hand, the current boundary layer structure is usually strongly influenced by the
presence of waves. A flexible, traditional modelling framework is suggested for
currents in the presence of waves. It is realised however, that the traditional eddy
viscosity based models are theoretically unsatisfactory. Therefore, a new type of
model is suggested in Section 1.5.9.
Chapter 2 considers the initiation of sediment motion and the quasi-steady
processes of sediment transport over flat sand beds under waves. Emphasis is
placed on discussing similarity and differences between these processes and steady
bed-load and sheet-flow sediment transport.
Small scale coastal bedforms are discussed in Chapter 3 together with the
hydraulic roughness of beds of loose sand under waves.
Chapter 4 deals with the motion of suspended sediment particles in coastal
flows. Emphasis is placed on the important mechanism of sediment trapping by
vortices. At the same time, it is shown that a pure wave motion has but negligible
effect on sediment settling.
Suspended sediment distributions is the topic of Chapter 5. Section 5.2
presents the nature of suspended sediment concentrations in coastal flows through
a review of the existing experimental data. The concept of pickup functions for
suspended sediment in unsteady flows is discussed in Section 5.3. Practical pickup
function formulae are derived, partly in analogy with pickup functions from steady
flow and, partly based on time-averaged suspended sediment concentrations in
unsteady flows. Section 5.4 discusses the mechanisms which distribute the
suspended sediment through the water column. It is shown that pure gradient
diffusion is an unsatisfactory model of suspended sediment distributions in most
natural flows. The reason is that the sediment distribution process often includes
mechanisms which have mixing lengths of the same order of magnitude as the

Viii
 Preface

overall scale of the concentration distribution. These mechanisms are referred to as

"convective". Because of the inadequacy of pure gradient diffusion as a model of
such large scale mixing, a new, quantitative framework is introduced for
convective processes. Recognizing that most natural sediment suspension
processes include elements of both convection and gradient diffusion, a new
combined convection-diffusion model is developed in Sections 5.4.5 through 5.4.8.
The new combined convection-diffusion model needs some calibration before all
combinations of flow type and bed sediment can be modelled quantitatively.
Qualitatively however, the new model can predict the observed differences
between sediment concentration profiles in different flows and between the
concentration profiles of different sand sizes in the same flow.
The topic of Chapter 6 is sediment transport model building. Models of
coastal sediment transport are seen as being essentially of two varieties. The
classical concentration-times-velocity integral models (cu-integral models) and the
particle trajectory models. It is found that the particle trajectory type of models are
generally simpler to construct and they provide more consistently accurate
predictions of coastal sediment transport rates. In particular, the surprisingly weak
grain size dependence of observed shore normal sediment transport rates over
rippled beds is predicted by the simplest particle integral models, the grab and
dump models.
Coastal sediment transport modelling is developing very rapidly. It is
therefore quite possible that todays state of the art models could be superseded
tomorrow. It is less likely however, that all of today’s unanswered questions will
be answered as quickly. Hence, an account of what is presently unknown may be
of more lasting value than a statement of what is known. Chapter 7 is therefore an
account of the most important unanswered questions in the subject area of this
book. It is hoped that this may provide some inspiration to other researchers in the
field.
Acknowledgements. I would like to thank my former and present employers,
The New South Wales Public Works Department, Dr Ian S F Jones of The
University of Sydney, and the Department of Civil Engineering, University of
Queensland for their support during the preparation of this book. Several
colleagues and friends, and most of all my wife Felicity have helped improve the
book by offering numerous valuable suggestions.

a Cp, Nabe
ix
 ‘a es la (ddwll weow yi MALL aie oy
esa on
a sake ee mae e |ai ie i
ubouink 2). deowton (Dewy tot 6. eae alm
» eons blind av Seta vaisingeee seer ae
oe antbauthibs hore} Los oaoyawihad deat to eisai abut saiest20207]
Pe apcny ; vro?Z pi trantsveds2 ‘abana ieu Nib a7
yeni
ior} view! f . met ‘hy Ww. es a lb-qcaagvpga pee ii

windy Pollan oA cap snsnuiem, Det

bie ae oh ene
nN opt oberg aS Tot Sey 3h) tavewor tos
oD ysengtiy. of, goliong ge mee Mites 03:
rol cement it 2aska Deseoaan mat bos iibed
vewn
; wh, .ssibiiel jokewe Fegan ‘[email protected] ite I
wecay <oei Wm glaianens eeted 2m Nee ote 1
bi (2 oh Ae yale abears
laps
; yotoskeay sianSina
Rag
Swi noo Segre shiva
Seni! I) ents aaa

uns ag, eae

a i ee arid) Et AED thee
| haaesn avy:

ia
| oO ans +403 body
be Mltnagen igre | aasiplaoep. 4, gk cle
phazs Klienbli Habouy.1g add 2o Abo ny
ee a tsb wen
‘iw’ sfaideoup. Lore wrtiaam alias 1Metalig hak
88 la 9
ous ienculdLee \ coiged SOO
tga
Yo
Bb Mh ”
9,ganeht
nett,
ce been
aegl sxor
Mslors Ly, I | dae mB ober ih ’ sg ne is a m

"it-erh erasures voy iles:2h Soeumugan Pian it List th ole voce

novels onthe eerenn gt Anala

“Ty « net 4 Zz ral A. oon
vile via, Anavini, Te u's
lajcccey! cond aid at it OP
mL Gyruigiy bratharth sonihd
fe = Ay wf ie on soi o

7 Ppa, u i |

oT Give Rs

iti ad
sol all oat,
“ Na
 CONTENTS

NOTATION XVii

1 BOTTOM BOUNDARY LAYER FLOW 1

1.1 THE BOUNDARY LAYER AND THE FLOW ABOVE IT, 1
LA. Introduction, 1
TD Natural flows, 2
1.1.3 Flow in laboratory models, 5
1.1.4 Equations of motion, 8
1.1.5 Reynolds equations for combined wave-current flows, 10

1.2 THE NATURE OF OSCILLATORY BOUNDARY LAYERS, 13

i I Introduction, 13
122 The shear stress distribution, 17
12.5 Turbulence structure in oscillatory flow, 18
1.2.4 Laminar oscillatory flow over a smooth bed, 19
125 The wave friction factor, 23
1.2.6 Energy dissipation due to bed friction, 27
E27 Boundary layer thickness for oscillatory flow, 29
1.2.8 Eddy viscosity in oscillatory flow, 31
Ex9 Eddy viscosity for sinusoidal flow, 34

1.3 OSCILLATORY BOUNDARY LAYER MODELS, 40

L3A Introduction, 40
|
esp Quasi-steady models of oscillatory boundary layers, 41
i 3.5 Velocity distribution models for oscillatory boundary layers, 42
1.3.4 Eddy viscosity based models of oscillatory boundary layers, 48
ieBo Higher order turbulence models for oscillatory boundary layers, 52

1.4 WAVE-GENERATED CURRENTS, 52

1.4.1 Introduction, 52
1.4.2 Eulerian drift in Stokes waves, 53
Contents

1.4.3 Lagrangian drift in Stokes waves, 54

1.4.4 Boundary layer drift, 54
1.4.5 Undertow, 60

1.5 WAVE - CURRENT BOUNDARY LAYER INTERACTION, 62

1.5.1 Introduction, 62
1.5.2 Resume of steady boundary layer concepts, 64
1.5.3. The eddy viscosity concept for combined flows, 68
1.5.4 Turbulence intensity in combined flows, 71 _
1.5.5 The influence of currents on the wave boundary layer, 73
1.5.6 Energy dissipation by waves on a current, 75
1.5.7. The influence of waves on current profiles, 78
1.5.8 Anempirical model for current profiles in the presence of waves, 84
Example 1.5.1: A simple estimate of F for fairly rough beds, 87
Example 1.5.2: Current profile from wave data and u(zr), 90
1.5.9 Wave-current interaction models with explicit consideration of aw, 91

SEDIMENT MOBILITY, BED-LOAD AND SHEET-FLOW 95

FORCES ON SEDIMENT PARTICLES, 95
2.1.1 Introduction, 95
2.1.2 Intergranular forces, 95
2.1.3 Fluid forces, 98

2.2 INCIPIENT SEDIMENT MOTION, 102

2.2.1 Introduction, 102
2.2.2 The mobility number, 103
2.2.3. The Shields parameter, 103
2.2.4 Skin friction and effective bed shear stress, 104
2.2.5 The grain roughness Shields parameter, 105
2.2.6 Critical Shields parameter and the Shields diagram, 107
2.2.7 Initiation of motion in combined flows, 108
2.2.8 The depth of closure, 109

2.3 BED-LOAD TRANSPORT IN STEADY FLOW, 109

2.3.1 Introduction, 109
2.3.2 What is bed-load?, 109

xil
 Contents

2.3.3 The amount of bed-load, 110

2.3.4 Steady bed-load and sheet-flow transport, 112
2.3.5 Sediment speed in steady bed-load and sheet-flow, 114

2.4 BED-LOAD AND SHEET-FLOW UNDER WAVES, 116

2.4.1 Introduction, 116
2.4.2 The amount of bed-load under waves, 117
2.4.3 Bed-load and sheet-flow under sine waves, 117
2.4.4 Bed-load and sheet-flow under skew waves, 121
Example 2.4.1: Application to King’s data, 126

BEDFORMS AND HYDRAULIC ROUGHNESS 129

INTRODUCTION, 129

COASTAL BEDFORM REGIMES, 129

BEDFORM DYNAMICS, 131

3.3.1 The continuity principle, 131
3.3.2 Bedforms migrating with constant shape, 132
3.3.3. Migration and growth of sinusoidal bedforms, 133

3.4 VORTEX RIPPLES, 135

3.4.1 Introduction, 135
3.4.2 Ripple length, 135
3.4.3 Ripple steepness, 138
3.4.4 Ripple height, 142

3.5 BEDFORMS IN COMBINED WAVE-CURRENT FLOWS, 143

3.6 HYDRAULIC ROUGHNESS OF NATURAL SAND BEDS, 145

3.6.1 Introduction, 145
3.6.2 Sand bed friction factors, 146
3.6.3 Sand bed roughness in steady flows, 149
3.6.4 Hydraulic roughness from oscillatory flow data, 152°
3.6.5 The roughness of flat sand beds under waves, 154
3.6.6 The roughness of rippled beds under waves, 155

xiii
Contents

4 THE MOTION OF SUSPENDED PARTICLES 161

4.1 INTRODUCTION, 161

4.2 SETTLING VELOCITY, 164

4.3 EQUATION OF MOTION FOR SUSPENDED PARTICLES, 167

4.4 ACCELERATING SEDIMENT IN A RESTING FLUID, 170

4.5 SEDIMENT PARTICLES IN ACCELERATED FLOW, 171

4.5.1 Introduction, 171
4.5.2 Small particles in accelerated flow, 171
4.5.3 Larger particles in homogeneous accelerated flow, 172
4.5.4 Settling through a homogeneous, oscillatory flow, 174
4.5.5 Sand particles in wave motions, 178

4.6 SEDIMENT TRAPPING BY VORTICES, 181

4.6.1 Introduction, 181
4.6.2 The kinematic trapping effect, 181
4.6.3 Secondary dynamic effects, 186

4.7 PARTICLES SETTLING THROUGH TURBULENCE, 189

4.7.1 Introduction, 189
4.7.2 The loitering effect, 190
4.7.3 Taylor’s dispersion model, 193
4.7.4 Dispersion with loitering effect, 195

SEDIMENT SUSPENSIONS 201

INTRODUCTION, 201
5.1.1 Sediment concentrations and the transport rate, 201
5.1.2 Mixing length and sediment diffusivity, 201
5.1.3. The bottom boundary condition, 205

5.2 THE NATURE OF SEDIMENT SUSPENSIONS, 206

5.2.1 What is suspended sediment, 206
5.2.2 Concentration measurements, 208
 Contents

aa. Concentration time series, 209

5.2.4 Time-averaged sediment concentrations, 215
32.9 Time-averaged concentrations under irregular waves, 218
5.2.6 Time-averaged concentration profiles under breaking waves, 219
beA Different sand sizes suspended in the same flow, 220

5.3 PICKUP FUNCTIONS, 222

=e | Introduction, 222
582 Pickup functions in steady flows, 223
33.3 Pickup functions in unsteady flows, 224
5.3.4 Pickup functions for sine waves over flat beds, 225
aeBe) Pickup functions for irregular waves over flat beds, 226
5.3.6 Pickup functions for waves over vortex ripples, 228
D3. Selective pickup from graded sand beds, 230

5.4 SUSPENDED SEDIMENT DISTRIBUTION MODELS, 233

5.4.1 Introduction, 233
5.4.2 Convection or diffusion, 233
5.4.3 Suspended sediment distributions due to pure convection, 238
5.4.4 Sediment concentrations due to pure gradient diffusion, 241
5.4.5 Sediment concentrations due to combined convection-diffusion, 245
5.4.6 Time-averaged concentrations in combined convection-diffusion, 248
Example 5.4.1: A simple case of combined convection-diffusion, 249
5.4.7 Time-dependent concentrations in combined convection-diffusion, 251
5.4.8 € , Wc and F(z) for combined convection-diffusion, 252
5.4.9 C(z) in combined convection-diffusion under non-breaking waves, 254
Example 5.4.2: c(z) over vortex ripples, 256
Example 5.4.3: c(z) over a flat bed, 259

SEDIMENT TRANSPORT MODELS 263

INTRODUCTION, 263

6.2 TRANSPORT MODELS: IN ESSENCE TWO VARIETIES, 263

Example 6.2.1: Particle trajectory approach to steady bed-load, 266

6.3 SHORE NORMAL TRANSPORT OVER VORTEX RIPPLES, 266

XV
Contents

6.3.1 What the experimental data show, 266

6.3.2 The gradient diffusion transport model, 269
6.3.3. A particle trajectory model for shore normal transport
over ripples, 270
6.3.4 The grab and dump model for shore normal transport
over ripples, 273
6.3.5 Comparison of the three models, 274

6.4 SHORE NORMAL SEDIMENT TRANSPORT OVER FLAT BEDS, 276

6.4.1 Introduction, 276
6.4.2 Shore normal sediment transport in the swash zone, 276
6.4.3 Shore normal sediment transport in the surf zone, 277
6.4.4 Shore normal sediment transport outside the surf zone, 278

6.5 SHORE PARALLEL SEDIMENT TRANSPORT, 280

6.5.1 Introduction, 280
6.5.2 Shore parallel sediment transport outside the surf zone, 281
Example 6.5.1: Shore parallel transport over a rippled bed, 282
Example 6.5.2: Shore parallel transport over a flat bed, 286
6.5.3 Longshore sediment transport in the surf zone, 291

7 LOOSE ENDS AND FUTURE DIRECTIONS 293

7.1 Introduction, 293
12! New models of wave-current boundary layer interaction, 293
13 Hydraulic roughness of flat sand beds under waves, 296
7.4 Instantaneous, effective shear stresses on sand beds under waves, 297
Te Bottom boundary condition for suspended sediment distributions, 297
7.6 Distribution models for suspended sediment, 298

REFERENCES 299
AUTHOR INDEX 309
SUBJECT INDEX 313

xvi
 NOTATION

Orbital amplitude of fluid just above the boundary layer.

Eulerian, turbulent scale ratio, p 197.
Reference sediment concentration, usually c(o).
Kajiura’s wave friction coefficient, p 23.
Drag coefficient, p 100 and Fig 4.2.1, p 164.
Instantaneous drag coefficient, p 167.
Added mass coefficient, p 7.
Wave celerity.
Volumetric suspended sediment concentration.
Time-average of c(z,t).
Bed-load sediment concentration.
Group velocity.
Group velocity seen by a fixed observer.
Group velocity relative to mean current.
Sediment concentration in the stationary bed.
Suspended sediment concentration, only in Section 5.2.1.
Water depth.
Wave energy dissipation rate, p 27.
Velocity defect function, see p 19.
Velocity defect function, p 20.
Velocity defect function, first harmonic, p 21.
Sediment grain diameter.
Median grain diameter.
Grain diameter exceeded by 10% by weight of sample.
Expected value of stochastic variable.
Wave energy flux per metre of wave crest.
Convective distribution function, p 237.
Frequency response function for t(0,t), p 23 and p 125.

xvii
Notation

Drag force, p 100.

Inertial force, p 7.
Lift force, p 101.
Pressure force, p 100.
Friction factor in steady flow, p 149.

Dissipation factor used by Carstens et al, p 27.

Energy dissipation factor used by Lofquist, p 27.
Wave energy dissipation factor, p 27.
Wave friction factor, p 23.
Grain roughness friction factor, p 105.
Acceleration due to gravity, g = (0,-g) .
Wave height.
Immersed weight longshore sediment transport rate.
Imaginary unit, V-1,
Imaginary part of complex number.
Permeability.
Wave number 27/L.
Dissipation factor used by Bagnold, p 27.
Wave length in Chapter 1, vertical convection scale in
Chapters 5 and 6.
Bed-load thickness, p 111.
Eulerian turbulent length scale.
Scale of suspended sediment distribution, p 217.
Thickness of wave-dominated layer, p 86.
Mixing length, p 201.
[L] Horizontal distance travelled by sediment particle.
[-] Solid fraction of a volume of bed sediment.
rey Amplitude of the n-th harmonic of p(t).
[-] Velocity distribution parameter, p 44.
p(t) (TA) Pickup function.
p (LT|] Time-averaged pickup function.
p(xy,z,t) [ML2T-] Pressure.

XVill
 Notation

(ML2T2] Time-averaged pressure.

[ML] Phase-averaged pressure, see Eq (1.1.17).
~
WD!
Ss [ML°T7] Random pressure component.
fiat’) Depth-integrated sediment transport rate, p 201.
Old ial Time-average of Q(t).
(er) Depth-integrated bed-load transport rate.
ier") Local sediment flux, g=cu = (Gx, dy, qz).
(LT|} Convective sediment flux.

(LT) Diffusive sediment flux.

[-] Reynolds number.

[L] Orbit radius.
[-] Real part of complex number.
Hydraulic roughness, Eq (1.5.9).
Shore normal wave radiation stress.
Relative density of sediment.
Wave period.
Wave period seen by fixed observer.
Eulerian turbulent time scale.
Lagrangian integral time scale.
Spectral peak wave period.
Period seen by observer who follows the mean current.
Time.
Time of entrainment event.
Depth-averaged current velocity in Section 1.5.6.
Amplitude of u(z,t), respectively USt).
Amplitude of 1.
Average bed-load velocity, p 114.
Dimensionless, relative velocity vector, p 173.
Water velocity vector, u = (u, Vv, w).
Relative sediment velocity vector, ur =Us-u, p 163.
Sediment velocity vector us = (Us, Vs, Ws).
Horizontal, velocity in direction of wave propagation.
Notation

esi) Friction velocity \t/p.

[ere First harmonic of u.
(LT) Defect velocity, Eq (1.1.3).
(LT) Eulerian time-averaged velocity.
(oT | Lagrangian time-averaged velocity.
[leis| Reference current velocity.
(ry Horizontal velocity of sediment particle.
[LT] Stokes drift, p 53.
[LT] Time-averaged horizontal velocity.
er 4 Periodic component of u, p 11.
(LT|] Random component of u.
[ET Average friction velocity, lu*l ux = (0)/p.

[LT] Peak wave friction velocity, = V 42 fy Aw.

[LT] Velocity just above the wave boundary layer.
[L?] Particle volume.
[L] Volume of sand picked up per unit area.
Variance of stochastic variable.
[LT] Horizontal velocity, perpendicular to wave direction.
i (LT|] Time averaged friction velocity in the y-direction.
[LT] Vertical water velocity.
[assy Time averaged, vertical velocity.
[LT] Periodic component of w, p 11.
~
er’) Random component of w.
VS
=
‘
(iy Root mean square value of w’ .
ee Profile shape (c(z)) parameter wo L/e€s, see p 254.
(LT Convection velocity of sediment, p 236.
Ls a Still water settling velocity, Wo = (0,-wo).
(ery Vertical sediment particle velocity.
[LT] Convection velocity of turbulence, p 19.
[L] Horizontal coordinate in the wave direction.
(L] Horizontal coordinate perpendicular to the x-direction.
[L] Vertical coordinate from flat bed or ripple crest level.

XX
 Notation

Zero-intercept level of log velocity profile, p 65.

See Fig 1.3.2, p 44.
Zero-intercept level of log velocity profile, p 80.
Bed level.
Entrainment level of suspended particle, p 237.
Level of reference current.
Constant, Eq (1.5.28), or Eq (4.3.8).
Constant, Eq (5.4.23).
Constants, Eqs (4.5.37) and (4.5.22).
Constant, Fig 4.2.1, pp 164 and 167.
Wave height to water depth ratio at the break point.
Boundary layer thickness, 54, 5j, 5.05, 5«, 5.01, p 30.
Laminar sublayer thickness, p 66.
Stokes Length V2 v/a .
Periodic delta function, p 230.
Perturbation parameter, p 163.
Sediment diffusivity, p 203.
Ripple length.
Linear sediment concentration, p 96.
Same as z.
Ripple height.
Time-averaged water surface elevation.
Shields parameter, Section 2.2.3.
Skin friction- or effective Shields parameter.
Critical Shields parameter, p 107.
Effective Shields parameter over ripples p 228.
Grain roughness Shields parameter, p 105.
See p 149.
von Karman’s constant (= 0.4).
Kinematic (laminar) viscosity.
Eddy viscosity derived from 1(z,f), p 34.
Eddy viscosity felt by u in combined flow, p 69.

XXi
Notation

Eddy viscosity felt by « in combined flow, p 92.

Eddy viscosity.
Turbulent wave eddy viscosity, p 33.
Eddy viscosity felt by ” in combined flow, p 69.
or in purely oscillatory flow, p 31.
2/21
Correlation coefficients.
Fluid density.
Sediment density.
Normal stress.
Time-averaged normal stress.
Effective normal stress, dispersive stress.
Turbulence intensities.
Horizontal shear stress in the x-direction (=T;).
Bed shear stress.
Time-averaged shear stress.
Periodic shear stress component, see Eq (1.1.17), p 11.
Peak bed shear stress under waves, p 23.
Lagrangian microscale.
Shear stress in the y-direction.
Turbulent shear stress, p 17.
Wind shear stress.

Dimensionless sediment transport rate, r Ey ree

Time average of ®(?).

Dimensionless bed-load transport rate.
Average of ®(t) through half a sine wave.
Phase angle, or angle between current and waves.
Phase shift between bedload transport and u..(f).
Dynamic angle of repose.
Static angle of repose.

XXil
 Notation

Or [-] Phase shift between bed shear stress and u.~(f).

Ov [-] Argument of complex eddy viscosity, p 35.
y [-] Mobility number, p 103.
Vv [-] Phase angle.
@ [ry Radian frequency, 2 %/T.
Wa a] Radian frequency seen by fixed observer.
Wp [T yy Spectral peak angular frequency.

General operators
yt Time-average, steady component.
~
Periodic component, Eq (1.1.17).
Random component, first derivative or skin friction.
>

Second derivative or form drag.

LE}

A
Peak value.

Subscripts
b Break point value.
c Pertaining to current in combined flow.
0 Deep water quantity.
rms Root mean square value.
s Pertaining to sediment.
w Pertaining to waves in combined flow.

XXili
 oe hire xemy tnodle bod peated iiidde
ot Cored
Broo ve)
the ellenae (Dip ‘olgmooto tasemagré
‘ Ll q ire anes : ai
-

—
i a“ng 084
cant Mid meee
Ir o.tot ee
wv bwatit ie nate Meee take 3S
mugenSze on
Ti bena
a
eens nite
pouraqo leans
i ae
Ls
REAR 1 Saas UE NEE -
a ‘ Ut
~ ANEAD) eee ‘ z. )
JvOra “Kne
a _ aol au nige
A 10 iinet ta . 1986 ig
ot, Trig ria
‘usin 6 s
i. LT }
iv hornets Sissengttic
ve) | i = cots sty sre th 3oe
Wee ' at | b l 2 abfeae wives,
vin intl”? ; Tike overaged ulio@ smep f
fy} (Mi. T* supe Saleh PM.
4 Mere) Pos 6 otthea
ese te
; Vy iogires ag! mictoon x = i s WT,

(al, "t ”) igha Cryst vies oS read

+Ml si | | tit : —
iM = ie
| ae bea
Pa) f i parry Meier

ct _ average ob @y2d,
¢ fq ~~“ aa och walege ¢

ire, } roteUt Gay |

9 °
4 } ~< Viviane ange
“ | a — Phinae white ‘es
Cy tieré
“Heat: ning
 COASTAL BOTTOM
BOUNDARY LAYERS AND
SEDIMENT TRANSPORT
 MOTIOd JATZAOO
CVACRIYAIYAAGVUOR
TAOIVART TAAMIGSe
 CHAPTER 1

BOTTOM BOUNDARY LAYER

FLOW

1.1 THE BOUNDARY LAYER AND THE FLOW ABOVE IT

1.1.1 Introduction
With respect to sediment transport modelling, the most important part of the
flow is the bottom boundary layer through which the main flow influences the bed.
The present chapter therefore describes the types of boundary layer flows
which occur in the coastal environment.
For simplicity, the bottom topography is taken for granted and considered to
be stationary. The effects of the boundary layer flow on mobile sand beds, and the
resulting topographical changes will be considered in the following chapters.
The bottom boundary layer is intuitively defined as the layer inside which
the flow is significantly influenced by the bed. There are various ways of defining
the thickness 6 of this layer in quantitative terms. However, in general terms the
boundary layer thickness obeys the formula

5 < WT (1.1.1)

where vz is the eddy viscosity and T is the flow period. Thus, for fixed eddy
viscosity, the boundary layer for tidal flow with a period of 12 hours will be
approximately sixty six times thicker than that of a ten second wave motion, and
while the tidal boundary layer thickness is often equal to the water depth the wave
boundary layer is generally only a small fraction of the depth. See Figure 1.1.1.
The ability of a certain flow component to transport sediment is mainly a
function of the shear stresses it induces at the bed. Therefore, since thinner
boundary layers mean larger shear stresses for a certain free stream velocity, the
Chapter 1: Bottom boundary layer flow

waves will tend to dominate over the tide with respect to sediment entrainment and
bedform formation. However, because of the “one step forwards and one step
backwards” nature of the wave motion, the currents may still be very important
transporters of wave entrained sediment unless the sediment concentrations vary
periodically in step with the wave motion.

wave induced
velocity
Figure 1.1.1: The tidal
boundary layer is much
thicker than the wave
boundary layer. So, even if
the tidal velocity Uide is
much larger than the
wave-induced velocity
amplitude U(z) near the
surface, the waves will
dominate the situation at
the bed.

1.1.2 Natural Flows

In natural situations the flow outside the boundary layer is a complicated
function of time which we often describe in terms of its harmonic components, for
example, a long term average, tidal components and a spectrum of wind-generated
waves. However, for sediment transport calculations at the present state of the art,
it is reasonable to simplify this picture considerably. Thus, in the following, we
shall generally consider the tides and even rip currents and longshore currents as
quasi-steady flows, and ignore effects of the Earth’s rotation. The waves will
generally be thought of as monocromatic with height H and radian frequency
W (=2 /T).
The treatment of the wave motion itself is outside the scope of this book, so
the wave-induced velocity u..(t) just above the bottom boundary layer will be

yD COASTAL BOTTOM BOUNDARY LAYERS

 1.1 The boundary layer and the flow above

taken for granted and generally expressed as a simple harmonic function of time.

Uol(t) = Usocos ot = AQMcos wt (42)

where A is the water particle semi-excursion. Wave-induced velocities vary only

slowly with the elevation z near the bed, see Figure 1.1.2, and for most sediment
transport purposes variation in the horizontal x and y _ directions can be
neglected.

N
Directi f
a ee WR ee

3) bed > Ma Ele xs

;
a >
ary, \
vai
ia tan Y
re ae \ L
Pa on
ka ee |
= wi :
pa i
th ral ‘
Figure 1.1.2: Water motion under a progressive wave in an inviscid fluid. In the
context of wave boundary layer flow the quasi-constant ( 0/dz ~ 0) velocity near the
bed is often referred to as ’the free stream velocity’ u(t).

Jonsson (1966) concluded from dimensional analysis that the structure of

oscillatory boundary layers depends mainly on the Reynolds number A’ w/v, and
on the relative bed roughness 7/A . It is therefore natural to ask the following
question at the start of our study: What are the likely ranges of the parameters
A’ w/v and r/A_ under natural conditions? and to keep the answers in mind
when choosing theoretical models and designing laboratory experiments.
Figure 1.1.3 represents an attempt towards answering this question in terms
of the presently available experimental data.
The indicated limit r/A ~ 0.5 for the use of horizontally uniform
descriptions: u = u(z,t) as opposed to more detailed descriptions of the form

AND SEDIMENT TRANSPORT 3

Chapter 1: Bottom boundary layer flow

10

K
Bs Kik
—s (oe) K

— u(x,z,t)
u=u(z,t) n::
oc
®::
aQ::
U:-
ox
::
° O::
@-:-:
O::
OF:
Ore

r/A
roughness,
Relative
0.01

0.001
103 104 105 106 107
Reynolds number, (A2 w/v)

Figure 1.1.3: An overview of the ranges of the Reynolds number A’ w/v and the
relative roughness r/A under likely field conditions and in some previous laboratory
experiments. The range shown for observed relative roughness values over loose
sand beds corresponds to all of the laboratory data of Carstens et al (1969) and
Lofquist (1986). If data from artificially flat beds at very low flow velocities are
excluded, the lower limit of r/A is 0.08. Legend, K: Kemp & Simons, flume (1982,
1983); S: Sleath, flume and tunnel (1982,1987); V: van Doom, flume and tunnel
(1981,1982); J: Jonsson & Carlsen, tunnel (1976); D: Jensen, tunnel (1989).

u=u(x, z,t) is based on considerations of the data of Kemp and Simons (1982,
1983) and those of van Doorn (1981, 1982). The former show pronounced
differences between the velocity structure over the bedform crests and over the
troughs, while the latter show reasonably uniform behaviour through most of the
boundary layer.

4 COASTAL BOTTOM BOUNDARY LAYERS

 1.1 The boundary layer and the flow above

The value of 0.5 is only meant as an indication. The choice of an upper limit
of r/A for the application of horizontally averaged models does of course, in the
end, depend on the amount of detail one needs to consider.
The observed range from laboratory experiments of the relative roughness of
natural sand beds, is derived from all of the dissipation data of Carstens et al
(1969) and the bed shear stress data of Lofquist (1986). Values of r/A are
obtained by applying Equation (1.2.22) "in reverse" to the authors’ friction factors.
Rippled sand beds generally gave r/A > 0.2, while values of r/A below
0.08 were found only for flat beds with very little sand movement, see Section 3.6.
The lower limit of observed r/A-values of 0.08 for natural (flat) sand beds is
surprisingly high. It corresponds to roughness values of the order a hundred grain
diameters, and there is some doubt regarding its interpretation in relation to
effective sediment transporting stresses, as discussed in Section 2.4. The roughness
of natural sand beds is discussed in detail in Section 3.6.
Very large relative roughness values may apply in nature over rocky areas
and over reef platforms.
The message from the presently available data for oscillatory flow over loose
sand is, in essence, that the range of Reynolds numbers and relative roughness
values likely to be found under field conditions are

A? w/v > 10°

corresponding to a limiting condition of, for example, (A, T) = (0.3m, 6s), and

rl/A > 0.08

From Figure 1.1.3 it can be seen that there is a serious shortage of

experimental data on boundary layer structure in this range.

1.1.3 Flow in laboratory models

Due to the great practical difficulties involved with good field measurements
a lot of experimental work on coastal hydrodynamics and sediment transport is
carried out in the laboratory with three different kinds of facilities.
The first and most common kind includes wave flumes and wave basins
where most aspects of prototype wave motion outside the boundary layer can be
modelled in accordance with Froude’s model law. However, because the size of

AND SEDIMENT TRANSPORT 5

Chapter 1: Bottom boundary layer flow

most wave flumes and basins is fairly limited it can be difficult to obtain
adequately large values of the Reynolds number A’w/v for modelling boundary
layer phenomena.
Therefore, a different type of apparatus was suggested by Lundgren and
Soerensen (1956), and later applied in many studies, namely the oscillating water
tunnel. This is essentially a large U-tube where the flow is driven by a piston in
one of the vertical legs, see Figure 1.1.4.
The horizontal test section of such tunnels can be several metres long so that
Reynolds numbers in excess of 10° can be obtained. The orbital motion in the test
section of the U-tube differs from real wave-induced flow by being totally uniform
in the x-direction (OUeo/dx = 0) and by having no vertical orbital motion, but these
dissimilarities are often of no concern.

RSS |

| Test Section |
Ree Se ee eee
| 2- 10m |

Figure 1.1.4: An oscillating water tunnel can be used to model many characteristics
of the wave boundary layer, but the vertical velocities of a wave motion and the
horizontal variation of the free stream are not modelled.

Due to the considerable cost of large wave flumes and tunnels a somewhat
simpler and cheaper type of facility was introduced by Bagnold (1946) and later
applied with modifications by many others. Instead of oscillating the bulk of the
water, Bagnold oscillated the bed in it’s own plane through otherwise still water,
see Figure 1.1.5.

6 COASTAL BOTTOM BOUNDARY LAYERS

 1.1 The boundary layer and the flow above

u=AW cos wt

Oscillating Plate

Figure 1.1.5: Oscillating plates or trays can be used for modelling oscillatory
boundary layers and sediment transport. However, distortion of the inertia/pressure
forces on sediment particles is of some concem.

The flow over such an oscillating bed is, for all practical purposes, similar to
the velocity defect

Ud(Z,t) = uot) — u(z,t) (1.1.3)

in a tunnel, but the two types are in some respects dissimilar as far as sediment
transport experiments are concerned.
The reason is that the forces resulting from fluid pressure gradients on a
resting sand particle are exaggerated on the oscillating bed. A grain at rest in the
reference frame which follows the tray with velocity —v..(t) will experience an
inertial force of magnitude

Fi = p(s+ Cw VE (1.1.4)
where p is the fluid density, s is the specific sediment density, V_ is the particle
volume and Cy is the added mass coefficient. However, the corresponding force
on a particle on a fixed bed in moving fluid is only

AND SEDIMENT TRANSPORT

Chapter 1: Bottom boundary layer flow

dice=
Fy = p(l+Cn)V— (1.1.5)
with typical values of Cy=0.5 and s = 26 the difference amounts to a factor
two. Thus, discrepancies can be expected for sediment transport phenomena
where the pressure force Fp is significant in comparison with the fluid drag force.
Further discussion of the forces on sediment particles is given in Section 2.1.

1.1.4 Equations of motion

The following section discusses briefly the background and the conditions of
applicability of the simplified equation of motion for bottom boundary layer flow.
As usual the starting point for the analysis of the fluid motion is the Navier
Stokes Equations, see e g Le Mehaute (1976), p 61. In relation to the bottom
boundary layer under unidirectional waves we shall consider only the equation for
the horizontal component of flow in the x-z plane

ou, ou, Ou 1 op fe a
= +i.
Qi i novon fw edz = =e
agent ro
where u and w are the velocities in the x and z directions respectively, p is the
fluid density, p isthe pressure, and v_ is the kinematic viscosity of the fluid.
The flow inside the boundary layer can often be considered to be essentially
horizontal (w = QO). The equation of motion can then be further simplified to

oor aau, au) _= aa

A. (E137)

where T is the viscous shear stress (t = — Vv du/dz).

This equation is however, still difficult to solve because of the non-linear,
convective acceleration term. It is therefore worthwhile discussing when that term
can be omitted. In broad terms this is possible when the velocity is horizontally
uniform ie, when u = u(z, f).
The first requirement for obtaining horizontal uniformity is that the free
stream velocity Uo is uniform. This condition is fulfilled exactly over oscillating
plates and in oscillating water tunnels. However, under real waves the variation
from wave crest to wave trough generates a convective acceleration of magnitude

OUlco 2nAw 2nA Juco

ane
lulco = Aw L = Fe
Ly taer (1.1.8)

8 COASTAL BOTTOM BOUNDARY LAYERS

 1.1 The boundary layer and the flow above

where L_ is the wave length. Hence the relative importance of the convective
term originating from uo is given by the factor 27A/L.
The second criterion for horizontal uniformity in the boundary layer is that
non-uniformities introduced by individual roughness elements should be restricted
to a layer which is considerably thinner than the boundary layer itself , see Figure
1.1.6.
Since the scale of the disturbances introduced by the individual roughness
elements is the bed roughness r, this may be expressed by 6/r>> 1 which
corresponds to A/r>>1, since 6/r_ isan increasing function of A/r.
More precise information about the boundary layer thickness relative to the
roughness height can be gained from Figure 28 of Sleath (1987). The Figure
shows, that over a bed of three dimensional roughness elements, the ratio of
boundary layer thickness to roughness size is given approximately by

80s/r = 0.26(A/r)””? (1.1.9)

corresponding to 50s=r for A/r=6.9 and 505=10r for A/r = 184. By

Sleath’s (1987) definition the top of the boundary layer (z= 50s) is where the
velocity defect amplitude becomes less than 5% of Aw.
When the criteria for horizontal uniformity are met, the non-linear,
convective acceleration term can be omitted, and the equation of motion becomes .

u=u(z),w=0O
Pi eel treet bre, vbr Pte ' OUTER LAYER

eee ee INNER LAYER

u=u (x,z),w#0

Figure 1.1.6: Even when the free stream velocity is horizontally uniform, the flow
near the bed may be non uniform throughout an inner layer of thickness similar to the
boundary layer thickness.

AND SEDIMENT TRANSPORT 9

Chapter 1: Bottom boundary layer flow

pat = = 22, (1.1.10)

This equation of motion for horizontally uniform flow can be simplified

further under the assumption of hydrostatic pressure distribution in the boundary
layer. That is when the vertical accelerations are negligible compared to the
acceleration of gravity. Then we can utilize the fact that the shear stresses vanish
outside the boundary layer so that

Ques
apalT_ _ Op
mde (1.1.11)

and rewrite Equation (1.1.10) in the form

p 2(u— ux) = (1.1.12)

1.1.5 Reynolds equations for combined wave current flows

In the following section we shall derive equivalent equations to the classical
Reynolds equations for combined wave current flows. That is, for flows which
contain a periodic component u as well as the familiar u+u’ for steady,
turbulent flows.
These equations are useful in the analysis of wave current boundary layer
interaction.
The classical Reynolds equations for a steady, turbulent flow are derived by
inserting (u,v,w) = (u+u’, v+v’, w+w’) into the Navier Stokes equations and taking
time-averages. The equation describing the flow in the x-direction can be written

a au? aw’ 2 _ Pea?

ofRitA a+ dam} o Te Ge vE
(1.1.13)
see, e g, Le Mehaute (1976) p 77. For a two dimensional flow in the xz-plane, this
can be written in the form

OG
ae Bho
ee 0 (1.1.14)

10 COASTAL BOTTOM BOUNDARY LAYERS

 1.1 The boundary layer and the flow above

where the total, time-averaged stresses (6,7) are given by

6%) = (p-pu? -pu’, pvt — pu W— pw’w’) (1.1.15)

Thus, one purpose of the Reynolds equations is to identify stress (or
momentum transfer) contributions from the different flow components. The shear
stress component -pu’w’ from the random velocity components is generally
referred to as the Reynolds stress.
We shall now derive the corresponding equations for a combined
wave-current motion where the velocity in each direction may include a periodic
component as well as the steady and the random components

(Uu,v,w) = (U+Ut+u’,V+V4+V,wtwtw’) (1.1.16)

The tilde denotes the periodic component, which is the phase average over
several (N) wave periods minus the time average
N
(zit) }= - Y' u(z, t+ JT) — u(z) C117)
fl
We note that with these definitions we have ¥=x'=x =O , and
S| 3! ?= xy’ = xy’ = xy’ = 0, while XY = Xy-—Xy. Inserting the expressions
—— wie —_—fos) into the “horizontal” Navier Stokes equation (1.1.6) gives

Load > re aK ? a ert ’ cer : Teut+ >

sere ) cearteeal’) See ) paeerisiey|aay, ) sek GenoD) ou ne u’)

é ~ 1p sep) + vV2Gititw’) (1.1.18)

In order to get the governing equation for the time-averaged velocity u, we
extract all non-trivial steady contributions from this equation and find

a et i eG, Oh. fp Do aihcs 8 oneOW

ry a 9 Bin
ly Bin e
IEMA ini y2 (1.1.19)
p ox

AND SEDIMENT TRANSPORT 11

 Chapter 1: Bottom boundary layer flow

This equation is now modified by using the continuity equation

2ay Grit)
(ut+u+u’) Deri) ++2ootiew’)
++Fada 57 w+w+w’) = 0 1.1.20
(1.1.20)

Thus, by adding u( - + spite

a )= to the first three terms in Equation

(1.1.19) and reorganising, these terms can be rewritten as Ps 5 caine

a +
ou. duv , duw
ao
By performing the analogous operations on the middle and the last three terms on
the right hand side of Equation (1.1.19) the complete, time-averaged equation
becomes

t)
Li + SGI) + 5 Gm + 27 +5, +5,
2 ii) + Su? + UV) +5 ew’)

= -1
—+ op on ou, ou |
vj —[+—+— 94.121)
yes & ay? az”
By collecting all terms which represent momentum flux in the vertical
direction under the label T this gives, by analogy with Equation (1.1.15),

Cs vst — puw — puw — pu’w’ (1:1.22)

This expression is of central interest with respect to the modelling of current

profiles in the presence of waves. The third term on the right hand side will
generally be dominant, except very close to the bed, and this has important
implications for the shape of u(z)-profiles with colinear waves superimposed, see
Section 1.5.9.
The corresponding expression related to a current perpendicular to the wave
motion is simply

Tz = pv — pvw — pu’w’ (1.1.23)

because V = 0.
If the flow is horizontally uniform we have w =0 by continuity, so that the
second term on the right hand side of Equations (1.1.22) and (1.1.23) disappear. In
order to derive the equivalent equation to Equation (1.1.21) for the periodic flow

12 COASTAL BOTTOM BOUNDARY LAYERS

 1.1 The boundary layer and the flow above

component, we take phase averages (with zero mean in accordance with Equation
(1.1.17)) on both sides of Equation (1.1.18) and make use of the continuity
equation as above. Then the following equation results

ou a Set
+22 Gm + Harri + 2 GH+TM

Ete
Se +5. Gin)4+7 ie)
Cig
+ Sl cae +S es
+ SV) We) (1.1.24)
wr an 2~ 2 2~
= ES Bal
ous ouHou
p ox ax" dy? az

Again we are particularly interested in the terms which represent vertical flux
of horizontal momentum, so we extract all such terms and put them under the
collective label T
_ ou
T= py) —puw -puw —p (un) — 9 (u’w’) ¢1.1.25)

The velocity product terms in Equations (1.1.22) and (1.1.25) play the same
role for wave-current motion as the Reynolds stresses —p u’w’ do for turbulent
steady flows. These velocity product terms will be used to discuss the total shear
stresses and the eddy viscosity concept for oscillatory flows and for combined
wave-current motions in Sections 1.2.8-1.2.9, and Section 1.5.3.

1.2 THE NATURE OF OSCILLATORY BOUNDARY LAYERS

1.2.1 Introduction
The wave boundary layer is intuitively defined as the layer close to the
bottom, where the wave-induced water motion is noticeably affected by the
boundary. This layer is normally very thin, generally a few millimetres over a
smooth, solid bed and a few centimetres over a flat bed of loose sand. Bedforms,
like ripples, may change the structure of the boundary layer by introducing strong

AND SEDIMENT TRANSPORT 13

Chapter 1: Bottom boundary layer flow

rhythmic vortices. Hence, the boundary layer over sharp crested ripples will extend
to a height of four or five ripple heights, or a total of about 50 centimetres under
field conditions.
Although the water motion induced by natural waves is not simple harmonic,
it is instructive and useful to study the simple harmonic, oscillatory boundary
layer, which corresponds to Uo = AW cos wt, and use it as an approximation to
natural wave boundary layers.
The total picture of the velocity variation in such a layer is at first sight very
complicated because both amplitude and relative phase change with the distances
from the bed, see Figure 1.2.1. However, in the following section we shall see that
a much clearer picture can be obtained by applying simple transformations to the
data.

TEST No. 1 cm TEST No. 2

-200 -100 0 100 200 -200 -100 0 100 200

Figure 1.2.1: Instantaneous velocities u(z,t) [cm/s] plotted against elevation from
top of the roughness elements. Numbers on the curves refer to the phase of the free
stream velocity Ue(z,t) = AW@cosM@t. The measurements were made in an
oscillating water tunnel by Jonsson and Carlsen (1976). Note that the velocity near
the bed tums before the free stream velocity and that the velocity amplitude is largest
in the range Scm<z<J0cm notat z3~-,

It is a typical feature of oscillatory boundary layers that the velocity close to

the boundary and the bed shear stress (o,f) are ahead of the free stream
velocity Uoo(t) as shown by Figure 1.2.1.
The bed shear stress in a simple harmonic, laminar flow is simple harmonic

14 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

and leads ue by 45°, but in a turbulent flow the variation with time is much more
complicated. The deviation from the simple harmonic behaviour increases with the
ratio between the bed roughness, r and the semi-excursion A. Thus, for small r/A
the variation of t(t) is still quite smooth and rather like a simple harmonic. This is
the case for the measurements of Jonsson and Carlsen (1976) Test 1 where r/A
was only 0.008, see Figure 1.2.2. In this case the phase shift between u.. and the
bed shear stress is still fairly well defined and we see that it is somewhat smaller
than the 45° of smooth laminar flow.

Ug (t)

Figure 1.2.2: Time variations of the bed shear stress t(o0, t) for rough turbulent flow
over relatively small roughness elements r/A=0.008. After Jonsson and Carlsen
(1976).

For flow over fully developed sand ripples, the ratio r/A is of the order of
magnitude one, and the flow near the bed is dominated by the rhythmic formation
and release of strong vortices. Lofquist (1986) measured instantaneous values of
t(o, t) under such conditions. Figure 1.2.3 shows some of his results and we see
that the behaviour is completely different from that of a simple harmonic and also
from that of sin(@t—)lsin(@t-@)| which has been assumed in several
“theoretical” studies.
Another typical feature of oscillatory boundary layers is the “overshoot” near
the bed. That is, there are elevations where the velocity amplitude U exceeds
Aq@ , see Figure 1.2.4.

AND SEDIMENT TRANSPORT 15

Chapter 1: Bottom boundary layer flow

Figure 1.2.3: Shear stress variation over fully developed sand ripples. For these
experiments A was fixed at 0.24m and the ripple length at 0.32m. u.(t) varied as
sin wt.

Figure 1.2.4: The local velocity

amplitude U(z) oscillates
around the free stream
value Aq@ reflecting the fact
that the velocity defect has the
nature of a dampened wave,
which propagates away from
VELOCITY AMPLITUDE U(z) the bed.

16 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

The velocity overshoot occurs because the velocity defect w..(t) — u(z, t)
has the nature of a damped wave which alternately adds to and subtracts from the
free stream velocity u..(t) , see Section 1.2.4.

1.2.2 The shear stress distribution

The simplicity of the equation of motion (1.1.12) is really somewhat
deceptive because the humble 1, which represents the shear stress, contains a
whole set of complexities of its own, which is indicated by Equation (1.1.25).
However complex, the total shear stress t(z, t) in a horizontally uniform
flow is nevertheless a measurable quantity in as far as Equation (1.1.12) is valid.
By integrating this equation and using the fact that t(o,t) = O we find

te: = | a (Uco — u)dz Gay

which says that the shear stress at the level z is equal to the fluid density times
the total acceleration defect above z.
Sleath (1987) discussed shear stresses and related quantities in great detail for
turbulent oscillatory flows. One of his most striking findings was that the total
shear stress calculated from Equation (1.2.1) was, for his experiments, about a
factor ten larger than the periodic, turbulent Reynolds stress defined by

TR(Z, 1) = —p(u'w’

where uv’ and w’ are horizontal and vertical, turbulent velocity fluctuations
respectively, see Figure 1.2.13.
Hence, as Sleath pointed out, the turbulent fluctuations u’ and w’ are “mere
spectators” to the oscillatory boundary layer processes. Their contribution to the
momentum transfer is totally overshadowed by the analogous contribution — p u w
from the periodic velocity components. This latter contribution in turn was found
to agree rather closely with the values of t calculated from the defect integral
ye 2 ig
For the shear stress amplitude |t (z, t)|_ the data of Sleath (1987), and of
several previous authors, show that it tends to decrease roughly exponentially with
increasing distance from the bed. The vertical decay scale is approximately the
boundary layer thickness 6 , ie,

AND SEDIMENT TRANSPORT 17

Chapter 1: Bottom boundary layer flow

It(z,H| = I1(0, tle’?

The magnitude of the bed shear stress is proportional to p(Aw) and to a

complicated function of the Reynolds number A’w/v and the relative roughness
r/A_ which will be considered in Section 1.2.5.

1.2.3 Turbulence structure in oscillatory flow

When discussing the turbulence intensity in oscillatory flows it is important
to distinguish between the time averages U'rms = Ci and W’rms = we)
and the phase-averaged (time dependent) (u’’)°> and (Ww),
With respect to u’rms and w’rms , Sleath (1987) found, for the relative
roughness range (0.03 < r/A < 0.25), that both decrease with increasing z. This is
in contrast to steady flows where the turbulence intensity practically constant
throughout the boundary layer, see Figure 1.5.7. For smaller relative roughness, a
layer of constant turbulence intensity becomes noticeable close to the bed. For
almost smooth beds (r/A < 0.0005) this layer extends almost throughout the
boundary layer indicating a similarity with steady flow, as in some of the
experiments of Jensen (1989).
The two components wu’yms and w’yms are of similar magnitude, and their
maximum value which occurs at the bed is approximately equal to the friction
velocity # = VA fy Aw, fw is the friction factor defined in Equation 1.2.18.
Sleath (1991) pointed out that the turbulence decay with distance from the
bed in fairly rough oscillatory boundary layers is analogous to the decay of grid
turbulence. He recommended the formula

a Stee re (1.2.2)
W rms
which is in good agreement with the data except within the above mentioned layer
of constant turbulence intensity immediately above the bed. The thickness and
importance of this constant-intensity-layer is generally small, but may be judged
from Figure 4 of Sleath (1987).
An example of the distribution of turbulence intensities in a turbulent,
oscillatory boundary layer is shown in Figure 1.5.7.
With respect to the time-dependent turbulence intensities ws and
(w’)", both Sleath (1987) and Jensen (1989) found that the amplitudes were
greatest near the bed and that the peak values occured later at higher elevations.

18 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

The general picture is one of parcels of turbulence propagating upwards from the
bed. On the basis of his data, Sleath derived the expression

@ S05
Wr = 221
(1.2.3)
for the speed of vertical convection of turbulence, where the boundary layer
thickness, 5,05 may be estimated from Equation (1.1.9).

1.2.4 Laminar oscillatory flow over a smooth bed

Although natural sand beds are never perfectly flat and natural flows tend to
be turbulent, it is worthwhile studying the case of smooth, laminar flow in detail
because many of its features are present in natural flows and because its structure
gives clues towards efficient methods of analysing natural flows.
We base the analysis on the linear equation of motion (1.1.12) which means
that the flow is assumed essentially horizontal and uniform in the x-direction. In
order to simplify the mathematical treatment we represent the free stream velocity
Uco(t) by the complex exponential

Uo(t) = Aw ei (1.2.4)
The real part A@ cos wt represents the physical velocity, see Figure 1.2.5.
For laminar flow the shear stress is proportional to the local velocity gradient
and the fluid viscosity

12,0) == pva”—ou (1.2.5)

EZe

so the equation of motion (1.1.12) can be written

F) du
Hit te) == VIG
=—(U-—Uo) Vz (1.2.6)
2G

Here we may introduce the non-dimensional velocity defect function

D(z, t) defined by

ud(z, t) = UoAt)—u(z,t) = AMD(z, t) (127)

in terms of which the equation of motion takes the form of the diffusion equation

AND SEDIMENT TRANSPORT 19

Chapter 1: Bottom boundary layer flow

Awcoswt

Figure 1.2.5: The complex velocity Aw e/®! has the constant modulus Aw and
moves around a circle with angular velocity @. The real part Aw cos wt which
represents the physical velocity oscillates between A@® and —AQ.

dD
Va =
a°D
> Ew (1.2.8)

This is easily solved by separation of variables and assuming that the velocity
defect function has the form
co

DG. Dien (1.2.9)

1
where D,(z) must then satisfy
dDn
inaDn = V5 (1.2.10)
This has solutions of the form

Di@) =. Ane Oa Be (1.2.11)

20 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

Since the velocity defect must vanish for z—> 2, Ay must be zero for all n, and
the boundary condition at the bed where the velocity itself vanishes

u(0,t) = Uso(t)-
Aw D(o, t) = Awe’ - AwY Bre” = 0 (1.2.12)
gives B, = 1 and By = 0 forn # 1. Hence, the complete solution is

u(z,t) = Awl —Dj (2) e™ (1.2.13)

= Aa[1 —exp LarZ__y) git (1.2.14)

The complex velocity defect function D(z) gives the velocity different
phases as well as different magnitudes at different elevations, see Figure 1.2.6.
The shear stress distribution is found by inserting the expression (1.2.14) for
the velocity field into Newton’s formula (1.2.5) which gives

t(z,t) = pvAw (1+i)V@/2v exp [-(1 +) —2Z—] e™

2v/@

This shows that the shear stress magnitude decreases exponentially away
from the bed with a decay length scale of V2v/w . The bed shear stress is
: cs I
1(0,t) = pvAa(1 +i) Va/2v & = pVav Awe*4) (1.2.15)
so we see that the bed shear stress in smooth, laminar oscillatory flow leads the
free stream velocity by m/4 radians or 45°.
The value 2 = p V@v Aw for the maximum bed shear stress shows that,
with the definition 7 = +P tw (Aw)? from Jonsson (1966), the wave friction
factor for smooth, laminar flow is

2
PRE aii
APSA
Ge (1.2.16)

Equation (1.2.15) for the bed shear stress corresponding to u(t) = AM Aes
is also valid for the individual harmonic components of an arbitrary free stream

AND SEDIMENT TRANSPORT 21

Chapter 1: Bottom boundary layer flow

Figure 1.2.6: Velocity variations with elevation in simple harmonic, oscillatory,

laminar flow over a smooth bed.
a: The defect function D(z) moves along a logarithmic spiral starting at 1 and
approaching 0 as z increases. Numbers on the curves refer to the non-dimensional
elevation zV@/2v = z/5s.
b: Corresponding variation of 1 —Dj(z) which is the complex ratio between u(z,t)
and u..(t). See Equation (1.2.13).
c: In the simple case of laminar flow over a smooth bed where u(z,t) is simple
harmonic u(z,t) = Aw[1—D;(z)] e we can construct u(z,t) geometrically by
using the circle from Figure 1.2.5 and the spiral above (b). The velocity at the bed
leads Uso by 45°.
d: The variation of the velocity amplitude U(z) = Aq@I1—Dj(z)I with elevation.
The maximum value (about 1.07A@ ) occurs for zV@/2v = 2.28.

Hap COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

velocity u(t), so the frequency response function for the bed shear stress (o,f)
from input u.0(f) is

F@) = pVove'™ (1.2.17)

1.2.5 The wave friction factor

The water in streams and under waves interacts with the bed sediment mainly
through the bed shear stress (o,f) . Hence the determination of (o,f) is a
crucial step in all sediment transport calculations. Therefore, considerable effort
has been put into the study of ‘t(o,f) under waves. Both Jonsson (1966) and
Kajiura (1968) developed semi-empirical and theoretical formulae based on their
early flow models.
Jonsson defined the wave friction factor fy in relation to the maximum of
t(o,t) by

A = 5Pw (Aw)* (1.2.18)

This definition may be applied to natural flows where (o,f) is not necessarily
simple harmonic. Kajiura, on the other hand, considered only the fundamental
(sinusoidal) mode of the flow and wrote the bed shear stress in the form

1(0,t) = pC; (Aw)’e!™ (1.2.19)

He accounted for the phase shift @ between bed shear stress 7 and the free
stream velocity u by allowing C; tobecomplex Ci; = IC\l e'?.
Kamphuis (1975) presented very comprehensive measurements of % on flat
beds of glued-down sand which are summarized in Figure 1.2.7.
Jonsson (1966) showed from dimensional analysis that the wave friction
factor can be expected to depend on the Reynolds number A*w/v_ and on the
relative bed roughness r/A
2

fw = fw oe 4] (1.2.20)

AND SEDIMENT TRANSPORT 23

Chapter 1: Bottom boundary layer flow

0.1

0.01
Numbers refer to A/d 99

x Smooth

0.001

Figure 1.2.7; Measured values of f, from oscillatory flow over flat beds of fixed
sand grains, after Kamphuis, 1975. The Nikuradse roughness for these sand beds was
taken to be 2do0,

Somewhat simpler formulae are adequate for the fully developed, rough
turbulent regime where A’w/v — © while r/A_ is finite, and for smooth
conditions (r/A — QO).
In Figure 1.2.7 the crosses correspond to experiments with a smooth bed. It
can be seen that, for Reynolds numbers up to about 3- 10° the smooth bed wave
friction factor is well described by

D
(i= ot (1.2.16)
VA2w/ Vv
which is the theoretical result for smooth laminar flow. The range
310° < A’w/v < 610° isa transition zone where fw increases. It is followed
by the fully developed, smooth turbulent regime where fy decreases again,
although at a slower rate than in the laminar regime. Justesen (1988) suggests

fu = 0.024 (A*w/vy” Sifor 10° < A*w/v'= 16° 92.21)

for the wave friction factor in the fully developed, smooth turbulent regime.

24 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

Kamphuis’ rough bed data are arranged in groups with fixed relative
roughness in Figure 1.2.7, the numbers referring to the ratio doo/A. . Note that
data points towards the upper right hand corner tend to fall on horizontal lines
corresponding to constant fy for constant do9/A.. We say that these data, for
which fy = fw(r7A), are in the fully developed, Rough Turbulent Regime. Several
authors have proposed formulae for fw in this regime, of which the most
commonly used is
\ 0.194
Sw = exp [5.213 g = 597 7h) (1.2.22)

This formula was suggested by Swart (1974) as an explicit approximation to an

implicit, semi-empirical formula given by Jonsson (1966).
Kajiura (1968) and Jonsson (1980) have both suggested upper limits for the
value of fw. Kajiura suggested 0.25, Jonsson 0.30. These limits were, however,
inspired by observations of the energy dissipation factor, fe (defined in Section
1.2.6) by Bagnold (1946). Bagnold recommended a constant limiting value of fe
of 0.24 for large relative roughness. However, the recent measurements by Sleath
(1985) yielded fe-values in excess of 0.5, and Simons et al (1988) measured
fe-values far in excess of 1.0.
It is well known that fe can be expected to be somewhat smaller than
Jw for very rough beds (r/A = 1) as illustrated by Sleath (1984) p 200. The
reason is, that for large r/A aconsiderable part of t(o0, t) is due to the pressure
gradients acting on the individual roughness elements. These pressure gradients are
in quadrature with u.. and therefore do not contribute to the energy dissipation rate
which is given by De = 1(0,f) UooA?).
Figure 1.2.8 shows the presently available fy and fe data for rough,
turbulent flow over beds with known Nikuradse roughness. The full line
corresponds to Swart’s formula (1.2.22) and we see that it tends to over predict
fw for small r/A. On the basis of this data set it might be justified to adjust the
coefficients in Swart’s formula to
r 0.2

fw = exp [5.5 q - 6.3) (1.2.23)

which corresponds to the dotted curve.

The fully developed, rough turbulent regime in steady flow is bounded by the
condition uxr/v > 70, where ux is the friction velocity (see Section 1.5.2). In

AND SEDIMENT TRANSPORT 25

Chapter 1: Bottom boundary layer flow
1

0.5

0.1

0.05

0.01

0.005
0.001 0.01 0.1 1
Figure 1.2.8: Observed wave friction factors for rough turbulent flow. + : Riedel
(1972) with r = 2d90, 0: Kemp and Simons (1982), Sleath (1987), AJensen (1989),
e: Jonsson and Carlsen (1976).

analogy with this, Kamphuis (1975) suggested a criterion of the form

is r/v = Vfw/2 Awr/v > constant (1.2.24)

for oscillatory flow. However, he found that a single constant would not apply
throughout the whole range of relative roughness. Instead he suggested two values,
namely 200 for fairly rough beds (r/A 2 0.01) while the value 70 from steady
flow applies asymptotically for r/A — 0.
While the peak value * is reasonably described by the formulae above, the
time dependence of the bed shear stress is not so well understood in general.
Two examples of the variation of (o,f) with time were given in Figures
1.2.2 and 1.2.3 for two cases of turbulent flow with different relative roughness.

26 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

These show how (o,f) is fairly similar to the smooth laminar solution (1.2.16) for
small r/A. That is, the variation is almost sinusoidal but the phase lead, relative
tO Uco , is less than the “laminar” 45°. Jonsson and Carlsen’s Test 1 shows a
phase lead of about 25°.
Lofquist (1980, 1986) reported comprehensive measurements of t(0,f) over
natural sand beds. The sequence shown in Figure 1.2.3 corresponds to a constant
value of the velocity amplitude Aw with the peak acceleration Aw growing
from top to bottom. The curves show that larger Aw” leads to a greater number
of more pointed peaks in (0, t). Lofquist explained how the peaks are related to
the growth and release of lee vortices.

1.2.6. Wave energy dissipation due to bed friction

The time-averaged rate of energy dissipation due to bed friction is given by

De = o,f) Uoo(t) (1.2.25)

and Jonsson (1966) defined the energy dissipation factor fe by

=.= 2 3
De = 3 Pfe (Aw) (1.2.26)
1.2.26

Thus Jonsson’s fe is related to Kajiura’s C, , defined in Equation (1.2.19 ), by

fe = Fic cos@ = SERefC, } (1.2.27)

Bagnold (1946), Carstens et al (1969) and Lofquist (1986) all used different
definitions and terminology. Their energy dissipation factors k, f and fi; are
related to fe by

fe = 3k = 4f = aii fi (1.2.28)

The two friction related coefficients fy and fe are different according to

their definitions. However, the experimental scatter of measurements of one or the
other over natural sand beds is so large that for practical purposes fw and fe can
be assumed equal. This is illustrated by the data in Figure 1.2.9 which shows
fw plotted against fe.

AND SEDIMENT TRANSPORT 27

Chapter 1: Bottom boundary layer flow

Tu.
fe "ah pias

0-06 o-1 o-2 o-4 o-6 10

Figure 1.2.9: The friction factor fy plotted against the energy dissipation factor
fe from measurements by Lofquist (1986) over rippled sand beds.

The range of measured fe values is quite large. Laboratory measurements by

Bagnold (1946), Carstens et al (1969), Kemp & Simons (1981), Sleath (1985),
Lofquist (1986) and Simons et al (1988) range from 0.03 to 40 while field
measurements by Bretschneider (1954), and Iwagaki & Kakinuma (1967) range
from 0.02 to 2.46.
It should be noted that other mechanisms of energy dissipation than friction
may be present in natural flows. These include viscous dissipation in mud bottoms
and losses due to percolation through a sand bed. For a review of these effects, see
Dean & Dalrymple (1991) and Sleath (1984).

28 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

1.2.7 Boundary layer thickness for oscillatory flow

While the qualitative meaning of the term boundary layer is straightforward,
the opinions about the most appropriate quantitative definition are varied.
Sleath (1987) intuitively defined the top of the boundary layer as the position
where the amplitude of the velocity defect has dropped to a certain small fraction
of Aw. He chose |D!=0.05 which, for a smooth laminar boundary layer flow
corresponds to ID; (50s) = exp (os Vw/2v) = 0.05 (see Equation 1.2.14) or
305 = 3V2v/m = 35s, where 5s is the Stokes length,

Jonsson (1966) used a different type of definition. He defined the top of the
boundary layer as the minimum elevation where u(z, ft) equals u(t) when the
latter is maximum. Jonsson’s boundary layer is quite thin i e, his definition
corresponds to Oj = V2v/@ for smooth laminar flow or approximately half
of 805.
Kajiura (1968) worked with the displacement thickness defined as
1 co

dd = aa Max J(Uco — u) dz (1.2.29)

oO

This again is a fairly thin boundary layer since for smooth laminar flow it
corresponds to S¢=Vv/w andhence ID; (84)l = exp (- V2/2) = 0.49.
However, the displacement thickness has the advantage that it is related in a
simple way to the other important boundary layer parameter, namely, the wave
friction factor. Their interrelation stems from the fact that the above definition for
da_ is very similar to the integrated momentum Equation (1.2.1) which in turn
defines the friction factor through Equation (1.2.18). Combining these equations
leads to

bad = +fv A (1.2.30)

This formula is exact for simple harmonic flows with the form
u(z, t) = [1—D,(z)] u(t) (Equation 1.2.13). However, it also provides a useful
estimate of 5g in general. This is important because the two major data sets of
oscillatory boundary layer flow over natural sand beds, Carstens et al (1969) and

AND SEDIMENT TRANSPORT 29

Chapter 1: Bottom boundary layer flow

Lofquist (1986) provide only fw (or fe) but no details of the velocity distribution.
Therefore, 5g (or any other vertical scale for the boundary layer) cannot be
determined directly from the data, only via Equation (1.2.30).
The practical limit for measuring boundary layer structures with present day
technology lies around the level where the velocity defect is one percent of the free
stream velocity amplitude. This level, 501 relates to the other measures of
boundary layer thickness as follows

So. = 15505 = 38 = 4585 = 6484 (1.2.31)

The definitions and interrelations are also illustrated in Figure 1.2.10.
Another boundary layer length scale which has the advantage of being related
to the friction factor is

Ss = tx /@ = Vfw/2A (1.2.32)
which, for flow of the form u(z,f) = [1—Dy (z)] u(t) , is related to the
displacement thickness by 5« = 64 Gi72y and since fy is generally of the
order 0.2 orless we see that 65x is fairly large compared to the other 5-values.

Aw

Figure 1.2.10: Comparison of the different definitions of boundary layer thickness

for the case of smooth, laminar flow with Aw*/v = 10°. The laminar length
scale V2v/q@_ is called the "Stokes length".

30 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

1.2.8. Eddy viscosity in oscillatory flow

The eddy viscosity concept, which was introduced by Boussinesq more than
a century ago, is a useful tool for obtaining simple flow models. However, as we
shall see in the following section, some unusual eddy viscosities are required for
the description of oscillatory boundary layer flows.
The simplified equation of motion for horizontally uniform flow

p2 (u-ux) ae (1.1.12)

contains two unknowns namely u and _ tT. It can therefore only be solved when the
relation between these two is known or assumed to have a certain form. We know
the relationship for laminar flows where it is given by Newton’s formula (1.2.5)
but for turbulent flows it is not well understood.
Many schemes (turbulence closure schemes) have been suggested for getting
around this problem and the simplest of these involves the use of the eddy
viscosity concept, which was first introduced by Boussinesq. It is defined by
analogy with Newton’s formula for laminar shear stress i e

tT = pv (122.33)

which for a steady, uniform turbulent flow corresponds to

= a +V (1.2.34)
az
in terms of the Reynolds stress -puw and it is a general feature of steady
turbulent flows that the first term dominates over the molecular viscosity except
inside the laminar sublayer.
For a uniform, oscillatory flow with zero net flow (u = u+u’), the
analogous expression for the eddy viscosity becomes (in accordance with Equation
f12))

Vt
_= &
En =
_ =-(aw) on-'w) + V (1.2.35)

Paz oz

where the meaning of the “~” operator is as defined by Equation (1.1.17).

We note that by this definition v; is not a purely turbulent quantity. It

31
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

contains a dominant, deterministic contribution —(#w)_ which was found by

Sleath (1987) to be an order of magnitude larger than — (u’w’). This caused Sleath
to express concern about its use. However, Equation (1.2.35) is the most useful
definition of v; because it enables us to write the equation of motion (1.1.12) in
the linear form

dt _ dio | A (v al (1.2.36)
are of are or
and it is the definition which has been used in all existing eddy viscosity based
models of oscillatory boundary layer flow.
Except for laminar flow, and for one particular class of turbulent flows,
which will be discussed in Section 1.2.9, the eddy viscosity for oscillatory
boundary layers is a function of the distance from the bed and it can generally not
be considered constant (at least not a real valued constant) over time for fixed z.
The reason is that the shear stress and the velocity gradient tend not to be zero at
the same phase see Figure 1.2.11.

Figure 1.2.11: Time dependence of shear stress and velocity gradient, both phase
averaged. From Jonsson & Carlsen (1976) Test 1, elevation above ripple crest: 4.5em.

This was realised previously by Horikawa & Watanabe (1968), Jonsson &
Carlsen (1976) and Sleath (1987). An example of the corresponding variation of
the eddy viscosity derived directly from the definition (1.2.33) is shown in Figure
1.2.12. This eddy viscosity takes on negative and even infinite values.
Sleath (1987) studied the turbulent stress component

32 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

°8

° S

)
Eddy
(cm?/sec
Viscosity

0 60 120 180 240 300 360

wt (deg.)
Figure 1.2.12: Eddy viscosity derived from velocity measurements using Equation
(1.2.33), as function of wave phase. After Horikawa & Watanabe (1968).

ani) (1.2.37)

as well as the total given by Equation (1.2.35). He found that the turbulent stress
component contributed only about ten percent of the total and it showed a very
different variation with phase, see Figure 1.2.13. In contrast, the
contribution —(u w) from the phase-averaged flow corresponded rather closely to
the total shear stress derived from the momentum equation (1.2.1).
Sleath (1987) also studied the time average of ze and of the total v;. He
found that Vzr was “firmly negative” over several roughness heights near the bed,
then went positive and continued to increase away from the bed.
For the time average v; of the total eddy viscosity he found that it was
generally positive and tended to increase with distance from the bed. However, the
general magnitude was significantly smaller than the equivalent for steady flow.
That is, in terms of the average friction velocity u* = (\q|/ p)°° he found

Vi = Ku«z , 0.10<K<0.13 (1.2.38)

where K_ is analogous to von Karman’s constant « (= 0.4) for steady flow, but
is seen to be smaller by a factor of 3 to 4. These results were obtained for relative
roughness of the order r/A = 0.01.

AND SEDIMENT TRANSPORT 33

Chapter 1: Bottom boundary layer flow

0.016
~

Ae

1/2p(Aw)?
: ua!
0.008 4X: 1/2 (Aw)2

-90 0 90 180 270

wt (degrees)

Figure 1.2.13; Reynolds stress -p(uw’) compared to the total shear stress T
calculated from Equation (1.2.1). After Sleath (1987), Test 4.

1.2.9. Eddy viscosity for sinusoidal flow

In order to obtain manageable, analytical descriptions for oscillatory
boundary layer flow, it is generally deemed acceptable to consider only the
fundamental harmonic component u; (z,t) of the horizontal velocity and the
corresponding shear stress component 7; (z,/). When this simplification is applied,
the eddy viscosity also becomes somewhat simpler than for the general situation,
described above.
In terms of the first harmonics u(z,f) and 7 (z,t) we may define an eddy
viscosity v; by

Var On, (1.2.39)

34 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

which, with the velocity gradient and the shear stress written in the complex forms

Our _ OM) 4 i (@t+ 9)

— (1.2.40)
oz
and

1 = Ile tort) (1.2.41)

gives

vi = Wle™® = el (¥- 9) (1.2.42)

r

Thus, treating the oscillatory boundary layer flow as simple harmonic with a
correspondingly simple harmonic shear stress (or analysing only the fundamental
harmonic modes) leads to an eddy viscosity .(z) which is a function of z but
not of ¢.
However, making provision for the shear stress to be out of phase with the
velocity gradient requires v, to be complex with an argument @y equal to the
phase shift between 1 and du;/0z.

z/6q

Figure 1.2.14: Eddy

viscosity magnitude based
on the fundamental modes
uy (z,t) and 1%(z,t) of
Jonsson & Carlsen’s data.
Elevations are measured
from the level of the
theoretical bed, and the
displacement thickness 54
and the friction velocity
ue are defined by
Equations (1.2.30) and
(1.2.24) respectively. x:
Test 1, 0: Test 2.

0 0.1 0.2 0.3 0.4

35
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

We note that if one chooses to use real valued cosine functions instead of the
complex exponentials in (1.2.40) and (1.2.41) then, the result of the phase shift
between shear stress and velocity gradient is a real valued eddy viscosity which
varies strongly with time and takes both negative and infinite values as indicated
by Figure 1.2.12.

z/byq

Figure 1.2.15: Argument

variation for the complex eddy
viscosity v1. Data from
Jonsson & Carlsen (1976). Test
I: x, T=84s, A=2.85m, ri/A =
: 0.008, Test II: 0, T=7.2s,
-20° 20° 40° 60° 80° 100° A=1.79m,
r/A = 0.035.
Examples of the variation of lv,l and Arg (vi) with z are shown in Figures
1.2.14 and 1.2.15.
The behaviours of vi! and of Arg(v:) are fairly coherent and well defined
for the two experiments represented in Figures 1.2.14 and 1.2.15. It can be noted
that the initial trend for the magnitude of v; is reasonably close to

vil = Kak z (1.2.43)

with Kk = 0.4, which corresponds to steady flow.

Figure 1.2.15 shows that, apart from a thin layer (z < 54), the argument of v;
is positive which means that in Jonsson & Carlsen’s experiments the shear stress 7
tended to be ahead of the local velocity gradient. This behaviour makes it difficult
or indeed impossible to explain v, as generated by local, instantaneous
parameters, as done for steady flows, and serves as a reminder of the complicated

36 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

Z (cm)

0.5 1.0

Figure 1.2.16: Variation of lvil/de for a range of flow conditions. The straight
line corresponds to Equation (1.2.43). x : Jensen (1989) test 10, (A, T, r, 505 ) =
(3.1m, 9.72s, 0.0mm, 60mm); + : Jensen (1989) test 12, (A, T, r, 5.05 ) = (1.58m,
9.72s, 0.84mm, 50mm); e: Jonsson & Carlsen test 2 (A, T, r, 505 ) = (1.79m, 7.2s,
63mm, 150mm) ; 0 : van Doorn (1982) M-series (A, T, r, 505 ) = (0.33m, 2.0s,
21mm, 26mm); «: van Doorn (1982) S-series (A, T, r, 5.05 ) = (0.10m, 2.0s, 21mm,
14.5mm).

AND SEDIMENT TRANSPORT 37

Chapter 1: Bottom boundary layer flow

nature of v; which is indicated by Equation (1.2.35). The convective behaviour of

the turbulence described by Sleath (1987, Figure 36) may be an explanation for
the complicated behaviour of the eddy viscosity.
The shape of the distribution of v, for oscillatory flows depends mainly on
the Reynolds number, A’w/v while the magnitude of v, also depends on the
bed roughness via the peak friction velocity ie see Figure 1.2.16.
Figure 1.2.16 shows that up to Reynolds numbers of the order / 6-10°
(Jensen, 1989, test 10) v, maintains its maximum value throughout the upper
part of the boundary layer while for the largest Reynolds numbers (> 3- 10°), a
decline becomes visible far from the bed. Thus, for small and moderately large
Reynolds numbers, the observed v,-distributions agree qualitatively with the
distributions suggested by Brevik (1981) and by Myrhaug (1982).
This constant-lv,|-range corresponds to the situation discussed by Sleath
(1991) where the turbulence intensity decays roughly as z/ in accordance with
Equation (1.2.2), while the mixing length (Iv;|/w’rms ) grows as 0.1z . This leads to

vil = 0.0025A”7r2 (1.2.44)

see the discussion following Equation 1.3.14.
The initial growth pattern for lv, near the bed corresponds to the usual
steady flow formula (1.2.43), but only for very high Reynolds numbers and small
relative roughness, as in Tests 10 and 13 of Jensen (1989), is this expression valid
for a substantial fraction of the boundary layer thickness.
For some of the experimens represented in Figure 1.2.16, the initial growth
region for v; is only a vanishing fraction of the boundary layer thickness, and is
restricted to those levels closest to the roughness elements where the assumption of
horizontally uniform flow is violated. For these flows, the assumption of a
constant, real-valued v; provides a reasonable description of the flow.
With a constant, real-valued v, the flow structure (u;(z,t)) is analogous to
smooth laminar flow with the only difference being that the laminar viscosity
v is replaced the much larger constant v; .
In order to interpret the structure of oscillatory boundary layers with constant,
real-valued v1, we draw on the mathematical framework derived for smooth
laminar flow in Section 1.2.4. In Section 1.2.4 it was found that the velocity defect
function defined by

ui(z,t) = [1—D; (2)]

Aw e™ (1.2.13)

38 COASTAL BOTTOM BOUNDARY LAYERS

 1.2: The nature of oscillatory boundary layers

was given by
v4
D(z) = exp[-(1 +) ] (1.2.45)
V2vV/@
for smooth laminar flow. Thus, the real and imaginary parts of the complex
logarithm In(D;) = In |ID,| + i Arg Dj, are identical:

Z
In ID\| ArgD, = 7a, (1.2. 46)

100

MILLIMETRES
IN

ELEVATION

TOP OF ROUGHNESS ELEMENTS

7
om 0:2 0:4 1-0 2 4 10
Figure 1.2.17: ArgD, and In\D,i for Test SOORA from van Doom (1982).
A=10cm, r=2.1cm, T=2.0s. The identity of ArgDi and /n|Djl and their
shared proportionality with z shows that the boundary layer structure is analogous
to smooth laminar flow. A deviation from this "smooth laminar analogy" is visible
below z = 3mm, but at these elevations, the assumption of horizontal uniformity (u =
u(z,t)) is not valid anyway.

AND SEDIMENT TRANSPORT 39

Chapter 1: Bottom boundary layer flow

Hence, it is interesting to plot experimental values of ArgD; and In |D\!

together against z for turbulent flows in order to get a comprehensive picture of
the behaviour of u(z, t) and especially of its similarities and dissimilarities with
the smooth laminar solution.
This has been done in Figure 1.2.17 for a set of very detailed velocity
measurements over a fairly rough bed.
The data show that, except for the first half millimetre above the roughness
crest, the identity ArgD; = InlD,! holds in analogy with smooth, laminar flow.
Furthermore, both quantities are linearly proportional to z, so we may write

Zz —
v4
ArgD, = In ID\| — ia Z1 pe \2v,;/@ (1.2.47)

Hence, the structure of this rough, turbulent boundary layer is completely

analogous to that of smooth, laminar flow, as far as the first harmonic u;(z,f) is
concerned. The eddy viscosity v; is therefore a real valued constant, independent
of both z and ¢. Its value is, for the case shown in Figure 1.2.17,
V1 = 28-10°%m’/s, or vi = 0.5@z7 with z1=0.0042m,
and w=3.14rad/s
The relative roughness range over which the “constant-
eddy-viscosity-model” applies is found, in the following section, to be 0.06< r/A
<1. General guidelines, (Equation 1.3.14) for calculating vi = 0.5 zp under
such conditions will emerge from Figures 1.3.4 and 1.3.5.

1.3 OSCILLATORY BOUNDARY LAYER MODELS

1.3.1 Introduction
The existing models for oscillatory boundary layers fall into two broad
physical categories, namely horizontally uniform models where u = u(z,f) and
models which take into account the horizontal variability of u(x,z,t) between
crests and troughs of the bed roughness elements. The latter group is by far the
smallest although realistic modelling of flow over the commonly observed sand
ripples obviously calls for models which can describe localised vortex formation.
Longuet-Higgins (1981) developed an essentially inviscid model based on
conformal mapping of a sharp crested ripple profile and the discrete vortex
method. The model yields a reasonable description of (o,f), but the results are

40 COASTAL BOTTOM BOUNDARY LAYERS

 1.3: Oscillatory boundary layer models

numerical only. Also, because of the assumed sharp ripple crest the vortex
shedding is continuous throughout each half cycle while observations of flow over
natural ripples tend to show that separation only occurs during flow deceleration.
A different numerical model was developed by Sleath (1982). His ripple
profiles had rounded crests and separation did not always occur. This model also
reproduced (o,f) quite well, but the fact that it is laminar calls for some caution
when extrapolating the results.
Recently, a different type of numerical solution was presented by Blondeaux
and Vittori (1990) which gives a good reproduction of the vortex shedding and the
resulting sediment clouds over vortex ripples.
Horizontally uniform models are much simpler. They can however, only be
literally valid at elevations which are well clear of the top of the roughness
elements, i e for z>>r. Hence, unless 8 >> r they are at most relevant as
descriptions of the horizontal average of the flow. Estimates of the ratio 5/r can
be found from Equation (1.1.9).

1.3.2 Quasi-steady models of oscillatory boundary layers

Horizontally uniform models of rough turbulent oscillatory flow can be
further subdivided into three main categories.
The first category consists of the quasi-steady models which assume that the
velocity distribution is at all times logarithmic throughout a boundary layer
thickness which may be constant or time-dependent. The classic model of Jonsson
(1966) is quasi-steady in the above mentioned sense.
This model provided surprisingly good estimates (essentially Equation
(1.2.22) ) of the wave friction factor, see Figure 1.2.8, page 26.
The good performances of this formula for fy at large relative roughness
was unexpected and, in fact, the more recent, quasi-steady model of Fredsoe
(1984) fails to predict fy as accurately for r/A > 0.03. This limit for the use
of “logarithmic velocity models” was predicted by Kajiura (1968) and has been
experimentally verified by the work of Jensen (1989).
The viability of the assumption of a logarithmic velocity distribution at all
phases of the flow depends on the relative roughness r/A. Some velocity
distribution data which may give an impression are those of Jonsson & Carlsen
(1976). They correspond to relative roughness of 0.008 and 0.035 respectively
and are shown in Figure 1.2.1. These data show that the assumption of a constant
shape of the velocity distribution is not very realistic. Better agreement is shown
by the smooth bed data of Jensen (1989), see Figure 1.3.1.

AND SEDIMENT TRANSPORT 41

Chapter 1: Bottom boundary layer flow

wt (degr.) =
0° 15° 30° 45° 60° 75° 90°
y ‘ Sv hy lob lewtermn %
+ —p

x “Pe ° r
1000 ° % ‘5 ape

100 §‘ 2 €
ott. A ° ; ” 4
10 / / 7 js 7 / /
ut

1
0: 5".0" 5-0 (e) () fe) 0

Flow reversal at wt=165°

105° 120° 135° 150° 160° 170°

7 i

05 050 ce) =5i20585

Figure 1.3.1: Dimensionless velocity profiles for smooth turbulent oscillatory flow at
different phases of the free stream velocity Awsin @t ; A = 3.1m, T 9.725,
yt =z. /v, ut = ii(z,t)/i(t). After Jensen (1989).

1.3.3 Velocity distribution models for oscillatory boundary layers

The second category of horizontally uniform models seeks empirical
expressions for the velocity distribution u(z,f) in more or less close analogy with
smooth, laminar oscillatory flow. Kalkanis (1957, 1964), Sleath (1970) and
Nielsen (1985) all followed this line of approach.

42 COASTAL BOTTOM BOUNDARY LAYERS

 1.3: Oscillatory boundary layer models

Drawing on analogies with smooth, laminar oscillatory flow, which is simple

harmonic everywhere, these models are only designed to describe the fundamental
harmonic mode of the flow.
The physical similarities between turbulent flows (rough or smooth) and the
smooth, laminar solution become particularly striking when the data are plotted in
the appropriate way for the purpose. Hence we take a closer look at the
mathematical structure of the smooth laminar solution

u(z,t) = Aw[1—D4(z)] e™ (1.2.13)

Di(z) = exp Be z—}

irre (1.3.1)

From this expression we note that In|D,| and Arg Dy, are identical
quantities and both proportional to the distance from the bed

In\Di! = Re{lnD;} = oe (1.3.2)

ArgDi = Im{InDi} = -y>—— (1.3.3)

The vertical scale of the velocity distribution is thus 5s = V2v/@ which is

called the Stokes length.
These mathematical relationships prompted Nielsen (1985) to study the
quantities In |D;| and ArgD, (subscript 1 referring to the fundamental harmonic
component) of measured velocities and to plot them as function of the elevation z.
Examples of such plots are shown in Figures 1.2.17 and 1.3.2.
Obviously the two quantities In |D;| and Arg Dj are all but identical almost
throughout the boundary layer. Hence, it seems reasonable to apply a description
of the form (1.2.13) with

D, = exp[-(1+/) F(2)] (1.3.4)

where F(z) is a real valued function of z. However, F(z) is not necessarily a

linear function as in the smooth laminar case. If one accepts a straight line in
Figure 1.3.2 as a reasonable approximation, we get an expression of the form

AND SEDIMENT TRANSPORT 43

Chapter 1: Bottom boundary layer flow

JONSSON AND CARLSEN (1976)

TEST I
200

Dw){e)

ELEVATION
MILLIMETERS
IN

2 4 6 10 20 40 60 100

Figure 1.3.2: The analogies between smooth, laminar and rough, turbulent
oscillatory flows become visible when real and imaginary parts of In D; are plotted
together against z. Data from Jonsson & Carlsen (1976) Test 2, (A,T,r) = (1.79m,
7.28,63mm).

ui(z,t) = Aw [1 —D,(z)] &™ (1.2.13)

with

D, = exp[-(1+i) ie (1.3.6)
where the two parameters z; and p are derived as shown in the figure above.

44 COASTAL BOTTOM BOUNDARY LAYERS

 1.3: Oscillatory boundary layer models

From presently available data it appears that Equations (1.2.13) and (1.3.6)
describe w(z,t) reasonably well for turbulent and transitional oscillatory
boundary layers with relative roughness greater than about 0.01. We may note
here, that all the available measurements of friction factors over beds of loose sand
by Carstens et al (1969), and Lofquist (1986) indicate relative roughness values
well within this range, see Section 3.6.
For smooth and almost smooth turbulent flows, like those investigated by
Jensen (1989), there is no longer an identity between Arg D; and In\DjI, see
Figure 1.3.3.

100

60

40

0:1 0:2 04 06 1:0 2 4 6

Figure 1.3.3: Arg D and In\D\\ derived from the measurements of Jensen (1989)
Test 13, A = 3.1m, T = 9.7s, r = 0.84mm. Even for these, almost smooth flow
conditions the formula (1.3.11) predicts the defect magnitude ID;! with great
accuracy. For thes experiments :0.09 Vr A = 0.0046m.

AND SEDIMENT TRANSPORT 45

Chapter 1: Bottom boundary layer flow

For very low relative roughness values Arg D; is smaller than /n |D| down
to 40%, but the two quantities do behave in a very similar fashion throughout the
boundary layer.
As mentioned in connection with Figure 1.2.17, there is a very close analogy
between u1(z,t) from fairly rough turbulent flows (r/A > 0.06) and the smooth
laminar solution. For such flows one finds p = / and hence

D; = exp[-(1 ay (1.3.7)
and as mentioned in Section 1.2.9 this flow structure corresponds to a constant
eddy viscosity vr = 0.5 zi.

107!
Arg D, <In |D,] ; Arg: In |D,|

-2 2

10 z,=0-09VrA

10-
10-4 103 102 107) 10

Figure 1.3.4: The simple formula (1.3.10) is valid over the full range where
horizontally uniform (u = u(z,t)) velocity models make sense. G : van Doorn (1981),
strip roughness, *: Sleath (1987), sand roughness, @ : Jonsson & Carlsen (1976) strip
roughness, 0: Jensen (1989), sand roughness.

According to the dimensional analysis of Jonsson (1966) it should be possible

to prescribe the two parameters z; and pas functions of the relative bed
roughness and the Reynolds number i e

46 COASTAL BOTTOM BOUNDARY LAYERS

 1.3: Oscillatory boundary layer models

Z1
ss= 21olited
ARNE (1.3.8)

>
Nid ~ CL 3.9)
>|> <|2
SS

where the dependence on A’/v is expected to be weak or absent for most

practical cases. This was attempted by Nielsen (1985) who found that z; can be
predicted by

Zz) = 0.09VrA (1.3.10)

for the complete range 0.01 < r/A <0.5 where Equation (1.3.6) is applicable,
see Figure 1.3.4.
In fact, this.formula can also be used for prediction of the magnitude of the
velocity defect for almost smooth turbulent flows where Arg D; and InID,I are
no longer identical. Thus, for such flows we still have

Dil = exp (eat oreil (1.3.11)

see Figure 1.3.3.

For smooth turbulent oscillatory flows, the data of Jensen (1989) indicate

ID,l = oy Cee p= (1.3.12)

ie , the parameter z; maintains its "laminar value", the Stokes length, for smooth
oscillatory flows, even when they become fully turbulent.
With respect to the second parameter p the message from the experimental
data is not quite as clear, see Figure 1.3.5.
However, it can be seen that the power p varies from unity for very rough
flows to about //3 for smooth turbulent flows. It was noted by Nielsen (1985) that
the p-value of approximately 1/3 for smooth turbulent flows corresponds to
maximum energy dissipation for fixed 2}.
The fact that p=1 for r/A2 0.06 indicates similarity with laminar flow,
and hence that the assumption of a constant, real-valued eddy viscosity
(vi = 0.5 @ zi) provides a reasonable model in this range. See also page 40.

AND SEDIMENT TRANSPORT 47

Chapter 1: Bottom boundary layer flow

0)
NOs 1073 102 0"! 10

Figure 1.3.5: Best fit p-values (Equation 1.3.11) from experiments with different
roughness types. Legend as for Figure 1.3.4.

1.3.4 Eddy viscosity based models of oscillatory boundary layers

The third category of horizontally uniform models contain the eddy viscosity
based models which have all so far been based on the equation of motion (1.1.10)
in the form

= (u-u60) = —(VI—) (1.3.13)

and, with the exception of the model by Trowbridge and Madsen (1984) all of the
existing eddy viscosity models for oscillatory boundary layers assume that the
eddy viscosity is a function of z but not of ¢.
Kajiura (1968) developed a very detailed model with very few oscillatory
flow data at hand and consequently had to rely very heavily on experimental
information from steady flows. He suggested the three layered eddy viscosity
distribution which is shown in Figure 1.3.6.
The mathematics involved in deriving the velocity distribution from Kajiura’s
model are somewhat overwhelming and various attempts have been made to
simplify it. Thus, Brevik (1981) omitted the inner layer while Grant and Madsen
(1979) assumed vy = Kix z for the complete range 0<z<oe,
The latter assumption can at best be valid for almost smooth conditions and

48 COASTAL BOTTOM BOUNDARY LAYERS

 1.3: Oscillatory boundary layer models

the continued growth of v; for z—> co seems particularly unrealistic since it is

well known that there is virtually no turbulence above a few roughness heights
from the bed.
Nevertheless, the resulting velocity distributions look quite reasonable even
with this unrealistic eddy viscosity distribution. This is due to the very forgiving
nature of the governing equation (1.3.13) with the boundary conditions
u (r/30,t) = O and du/dz > 0 for z3&.

Outer Layer

d
Overlap Layer
r
2
|nner Layer 0-185 K Gyr Vi

Figure 1.3.6: The eddy viscosity distribution applied by Kajiura (1968). The overlap
- layer, which is analogous to the logarithmic part of a steady boundary layer, was only
expected to exist for almost smooth beds. Kajiura suggested that it would probably
disappear completely for r/A > 1/30. The friction velocity is defined by
te = Vfw/2 Aw, and the thickness d of the wall layer is given by d = 0.05 ix /0.

In order to be satisfactory, an oscillatory boundary layer model must be able

to predict both magnitude and phase of the simple harmonic velocity u1(z,t).
An efficient way of testing this ability is to plot the complex velocity defect
function D(z) in terms of InID;l and ArgD; against z together with
corresponding data. The defect function is defined by Equations 1.3.2 and 1.3.3.
This has been done in Figure 1.3.7 to compare the models of Grant & Madsen
(1979), Brevik (1981) and Myrhaug (1982) to the measurements of Sleath (1987)
Test 4.

AND SEDIMENT TRANSPORT 49

Chapter 1: Bottom boundary layer flow

eo Grant & Madsen (1979)

oe Brevik (1981)
100
Myrhaug (1982)

in. rs)
millimetres

Elevation
Z

0-1 0:2 0:5 1 2 5

Figure 1.3.7: Comparison of the models of Grant & Madsen (1979), Brevik (1981)
and Myrhaug (1982) with data fromm Sleath (1987), Test 4. A = 0.45m, T = 4.58s,
r=2d=3.26mm. x: In|D\| from measurements, 0: Arg D from measurements.

The comparison in Figure 1.3.7 shows that all three models predict the
magnitude of the velocity defect, and hence In |DI quite well, but only Myrhaug’s
model performs well with respect to the phase (Arg D). The performance generally
improves with increasing model complexity.
The fact that Myrhaug’s model fares best among the ones tested in Figure
1.3.7 corresponds to the fact that his assumed eddy viscosity distribution
corresponds rather closely to the empirical values shown in Figure 1.2.16 for the
relevant Reynolds number. Sleath’s Test 4 has a similar Reynolds Number to
Jensen’s test 12. Myrhaug’s vydistribution may be less suited to model some of
the other cases shown in Figure 1.2.16.
It is not surprising that the agreement between these theories and
measurements is fairly unimpressive. They all assume v; to be a real-valued

50 COASTAL BOTTOM BOUNDARY LAYERS

 1.3: Oscillatory boundary layer models

function of z only, while most data show a general tendency for the local shear
Stresses to be out of phase with the local velocity gradients, see Figure 1.2.11.
These phase shifts make it necessary for v; to be either a complex function of z
or to be strongly time-dependent, see Sections 1.2.8 and 1.2.9.

VAN DOORN 1982, MIORAL

T(z,t)= : p(Kz) 2 [oul] du

Dzlesaez

40
- £n 1D,1

fe)

ELEVATION
IN
MILLIMETRES

jl
0.1 0.2 0.4 1.0 20 40 10.0
Figure 1.3.8: Comparison of the Bakker & van Doom mixing length model with data
from van Doom (1982), Test M10 RAL. We see that the model predicts fairly
reasonable values of InID,| while the predicted values of Arg Dj are clearly
different and further removed from the experimental values. The identity between
Arg D; and In|DjI is very clearly shown for these conditions, (T, A, r) = (2s, 0.33m,
0.021m).

The only case where it seems well justified to assume that v; is real-valued
and independent of time is for fairly rough beds, r/A > 0.06 where vy is
independent of z as well and given by

AND SEDIMENT TRANSPORT 51

Chapter 1: Bottom boundary layer flow

vi = 05wz? = 0.50(0.09VrA )* = 0.004 arA (1.3.14)

see Section 1.2.8 and Figure 1.2.16. It will be noted that this formula is not in
general agreement with Equation (1.2.44) which was suggested by Sleath (1991).
The reason is that the two formulae are not really predicting the same thing.
Sleath’s formula is aimed at the outer layer only, where v;= constant even for
very low values of r/A (see Figure 1.2.16). Equation (1.3.14) is meant to predict a
global value of v; for fairly rough conditions (7/A > 0.06). The two formulae give
identical results for r/A = 0.4.

1.3.5 Higher order turbulence models for oscillatory boundary layers

Higher order turbulence closure models have been tried as well for the
modelling of oscillatory boundary layers. Bakker & van Doorn (1978) applied
Prantl’s mixing length concept in a numerical model which was developed further
by van Doorn (1983). However, like some of the models discussed in Section
1.3.4, this model neglects some fundamental characteristics of the oscillatory
boundary layer, at least for fairly rough beds, as shown in Figure 1.3.8. Thus, the
identity between Arg D; and InID,! , which is quite clearly shown by this data
set, is not predicted by the numerical model.
Justesen (1988) applied a «-e scheme which gives reasonable results for
nearly smooth beds. At the extreme end with respect to computational effort
Spalart & Baldwin (1987) solved the complete Navier-Stokes equations for the
case of oscillatory flow over a smooth bed.

1.4 WAVE-GENERATED CURRENTS

1.4.1 Introduction
Wave motion is often thought of as purely oscillatory. However,
measurements of the velocity field under real waves show the existence of time-
averaged velocity components almost everywhere.
The magnitude of these steady flow components is generally much smaller
than that of the oscillatory components. However, because their effect is
cumulative, their contribution to the net sediment transport may well be
significant.

52 COASTAL BOTTOM BOUNDARY LAYERS

 1.4: Wave-generated currents

1.4.2 Eulerian drift in Stokes waves

Imagine a velocity probe positioned just above the mean water level of a sine
wave. This probe will only be wet during part of each wave period but for all of
this time, it will be recording positive horizontal velocities. Hence, the time-
averaged Eulerian velocity at this point will be non-zero and directed in the
direction of wave propagation. Similarly, a probe anywhere between the MWL and
the wave trough will also record a positive mean velocity.
The Eulerian net flow between the crest and the trough amounts to gH’/8c,
and this is the total, because a probe positioned anywhere below the trough level
will measure zero average velocity under a sine wave.
Now, in most cases, the presence of a beach or the end of the wave flume
prevent net transport of water in the shoreward direction. The zero net transport
thus required, is most simply obtained by superposition of a uniform, negative,
steady velocity

Ustokes = (1.4.1)

stokes 0

Figure 1.4.1: Eulerian (left) and Lagrangian (right) net flow velocities resulting from
superimposing a uniform drift on a sine wave in order to obtain zero flow of water.
These velocities are based solely on continuity considerations. That is, the forces
required to set up such a flow pattern in a real fluid are not considered.

AND SEDIMENT TRANSPORT 53

Chapter 1: Bottom boundary layer flow

onto the sine wave velocity field. This uniform, seaward drift velocity is generally
referred to as the Stokes drift. The considerations behind it are based on continuity
only, ie nothing is said about the forces which would be needed to drive it with a
real (viscous) fluid.

1.4.3 Lagrangian drift in Stokes waves

Similarly to the Eulerian net velocity described above, a pure sine wave will
also result in a positive, Lagrangian mean velocity at all levels. The Lagrangian net
velocity is qualitatively due to two properties of the sine wave velocity field.
Firstly, a fluid particle in a sine wave will move with larger forward
velocities at the top of its orbit than the backward velocities at the bottom.
Secondly, the particle moves with the wave during its forward motion and
against it during its backward motion, and will therefore spend more time moving
forewards than backwards.
The Lagrangian net velocity thus obtained is often referred to as the mass
transport velocity and its distribution over the depth is given by

2
uy = oO cosh 2kz (1.4.2)

A zero net flow can again be obtained by superimposing a suitable, uniform

velocity. This leads to

ila (Aw)” sinh 2kD

in, = (cosh2kz — ee ) (1.4.3)
8c 2kD
See Figure 1.4.1

1.4.4 Boundary layer drift

While the inviscid velocity distributions discussed above might lead to the
expectation of seaward net velocities near the bed, most measurements under
non-breaking waves actually show a positive or shoreward net velocity close to the
bed.
An explanation for this forward boundary layer drift was given by
Longuet-Higgins (1953, 1956). It occurs because the horizontal and vertical
velocities in a wave motion with a viscous bottom boundary layer are not exactly

54 COASTAL BOTTOM BOUNDARY LAYERS

 1.4: Wave-generated currents

ninety degrees out of phase, as they would be in a perfectly inviscid wave motion. _
This gives rise to finite time-averaged stress terms of the form -—p uw
which are analogous to the familiar Reynolds stresses —p u’w’.
These stress terms grow from zero at the bed (where w = 0), towards an
asymptotic value of

—— (HW) = p (Aw)?k 85/4 (1.4.4)

where k is the wave number and $s is the Stokes length V2v/m. See Figure
1.4.2. For the case of constant eddy viscosity as well as for laminar flow, the
distribution of —p uw is given by

Figure 1.4.2: Distribution of —p i ina wave boundary layer with constant eddy
viscosity. By comparing the expression (1.4.4) with the expression t= pVav Aw
for the peak bed shear stress due to a purely oscillatory flow, we see that the
= kA
asymptotic value of —p uw amounts to we

55
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

p(w) = p >(Aw)k 8s [1 — > ( 2cos& — e > + 2EsinE) ] (1.4.5)

where € = z/V2v/w. This distribution is shown in Figure 1.4.2.
These "Reynolds like" stresses influence the u-distribution through the
equation of motion (1.1.21) which for a two dimensional case with uniform flow

conditions Cs=() can be written

Iey tir
oe pe ene)
uw’) = ay DRC
+ =
a: uw (1.4.6)

which may be integrated once to yield

ob
Mans op
uw =e
= ae a A5Ho) (1.4.7)

or

Vt ou
Be _1lo
p ax =~
z+uwt _1-
p T(0) (1.4.8)

This can be integrated further using, u(o) = 0, to give

a aft
* Vt
Ps
ox
iw +p4H) \ae (1.4.9)
ve

For the laminar case Longuet-Higgins (1956) suggested the solution

22
Up = oe [3+ et (—4cos§ + 2sinE + ve 2Esin§ + 2Ecos&)]

(1.4.10)
which is based on the assumptions of ge= 0 and 70) = -—p (iWon.
We see that according to this solution, the Eulerian mean velocity tends
towards an asymptotic value of magnitude 3(Aw)*/4c for z>> ds. The
corresponding mass transport velocity, i e the asymptotic time-averaged,

56 COASTAL BOTTOM BOUNDARY LAYERS

 1.4: Wave-generated currents

Lagrangian velocity is 67% larger at 5(Aw)*/4c.

The velocity distribution (1.4.10) corresponds to laminar flow or to a
turbulent flow with constant eddy viscosity, but Longuet-Higgins showed that the
asymptotic velocities above the boundary layer are the same for any distribution of
the eddy viscosity, as long as it is time-independent,
v; = v,(z).
It should be kept in mind however, that the solution (1.4.10), for the drift
velocity, relies on the condition

(0) = tw-pgD dy
ai
_ Sx = -P(UW)oo .
arte
If that is changed, the velocity distribution changes accordingly. See Figure 1.4.3.
The behaviour of —p uw and hence of uz, in turbulent boundary layers in
general, is at present neither well documented nor well understood. However, the
results in Section 1.3.3 concerning the structure of turbulent oscillatory boundary
layers indicate that the laminar solution (with v replaced by v;) may have fairly
wide applicability. It was found in Section 1.3.3 that turbulent oscillatory boundary
layers in the roughness range 0.06 < r/A < 0.5 seem to correspond to a constant
eddy viscosity with V2v;/® = 21 = 0.09 VrA. Hence, the stress distribution
(1.4.5) with 8 = 0.09 VrA may provide reasonable estimates for these flows.

ae

Figure 1.4.3: The bed shear stress must in general balance the sum of wind stress on the surface,
the radiation stress gradient and the pressure force due to mean surface slope.

57
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

The driving mechanism behind the velocity distribution (1.4.10) is also

present under standing waves and partially standing waves. In that case however,
the wave conditions vary significantly in the x-direction so, several more terms
from Equation (1.1.21) must be retained. In that case, the result is a series of
counter rotating cells with the velocities at the bed directed from the surface nodes
towards the surface antinodes.

Figure 1.4.4a: The wave maker is tured on and sand is entrained under the surface
nodes where the near-bed velocities are strongest.

Figure 1.4.4b; The suspended sediment is convected by the boundary layer drift
towards the surface antinode.

58 COASTAL BOTTOM BOUNDARY LAYERS

 1.4: Wave-generated currents

These cells generate a clearly observable sediment transport pattern resulting

in bar formation, as pointed out by Carter et al (1973), and illustrated by the
photographs in Figures 1.4.4 a through d. The first three (a through c) show how
the sediment suspension is re-established when the wavemaker is turned on after a
period of quiescence. Sand is picked up only under the surface nodes and then
convected by uw towards the surface antinodes and then towards the surface.

Figure 1.4.4c: Some sand is rising towards the surface under the antinode.

Figure 1.4.4d: Eventually, the sand which settles under the antinodes forms bars.

AND SEDIMENT TRANSPORT

59
Chapter 1: Bottom boundary layer flow

Figure 1.4.4d shows the resulting bar pattern.

The water depth in these experiments was 0./2m, the wave period was 0.5s
and the median sand size was 0.08mm .
For a detailed discussion of drift patterns over flat and rippled beds under
standing waves as well as progressive waves see Sleath (1984).

1.4.5 Undertow |
The shoreward boundary layer drift indicated by the solution (1.4.10) is
usually not observed in the surf zone. In the surf zone, the u-picture tends to be
dominated by the so-called undertow, a seaward mean velocity between the bed
and the wave trough level, see Figure 1.4.5.
The undertow is a gravity driven current related to the phenomenon of wave
setup as illustrated very lucidly by the experiments of Longuet-Higgins (1983). It
occurs because the radiation stress gradient dSx,/dx is not uniform over the depth
under breaking waves while the opposing pressure gradient dp/dx = pg dn/dx
from the wave setup is (nearly) uniform, and therefore dominates near the bed.
Quantitatively, the undertow is described by the time averaged equation of
motion (1.1.21) which for the two dimensional case can be written

undertow

Figure 1.4.5: The undertow occurs in the surf zone because the radiation stress
gradient is concentrated near the surface, while the opposing pressure gradient due to
the setup slope is essentially uniform over the depth.

60 COASTAL BOTTOM BOUNDARY LAYERS

 1.4; Wave-generated currents

2 0=
+
ay
earns
+ —.
5, uw (1.4.11)

and assume hydrostatic mean pressure @ = Ne a6

Bet this can be rewritten as

a.
rm Vel dy
Ao _Si Ee:
MAM, + Io,
ay + IF
5H
— AG
+ ay Ae
+ 5, Hw (1.4.12)

which can be integrated once to give

ou
val _= oa
oles oO « Be
(Be 09 +, 972 + OEuo)
au * de

(1.4.13)
u
where the boundary condition vc oe | = T(o) has been used.
Oz z=0

The bed shear stress (0) may include a wind stress as well as the radiation
dS
stress contribution — aa and the setup contribution —pgD A See Figure 1.4.3.
To obtain the undertow distribution, Equation (1.4.13) is integrated once
more with the use of the non-slip boundary condition u(o) = 0. This leads to

Z Zz =

uz) = Jab? + ax + ae"

‘\ae+ + oe + dz
(0)

(1.4.14)
From this expression it is obvious that in order to model the undertow, one
must consider the complete system of turbulence from wave breaking (to get the
eddy viscosity Vci), the wave setup 1, and the process of wave transformation and
decay across the surf zone. For examples of recent models see Svendsen et al
(1987), Roelvink & Stive (1989) and Deigaard et al (1991).

61
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

1.55 WAVE-CURRENT BOUNDARY LAYER INTERACTION

1.5.1 Introduction
The hydraulic conditions which cause problems of siltation or erosion on the
coast are nearly always mixed in the sense that the velocity field has both a steady
component u and a periodic component u which may have very different
relative magnitudes and different directions.
We are, in the present context, mainly interested in the sediment transport
which results from these combined flows and hence we shall concentrate on the
flow structure near the bed where most of the sediment transport occurs. The
changes in wave height and direction which occur when waves travel through a
variable current field are considered outside the scope of this text. For a recent
review of these phenomena see Jonsson (1990).

ELEVATION
MM
IN

COMBINED
ALONE
ALONE
O
O 1@) 20 30
VELOCITY (cm/s)
Figure 1.5.1: Profile of a pure current u (+), the velocity amplitude lid in a pure
oscillatory flow (x), and both u and luli (e) from the combination of the two
flows. Measurements in an oscillating water tunnel by van Doorn (1982).

62 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

An example of velocity profiles from corresponding pure current, pure wave

and combined wave-current flows are shown in Figure 1.5.1.
These data show that while the structure of the oscillatory flow is unchanged
by the addition of the current, the addition of the waves changes the current profile
considerably.
In essence the effect of the waves is to suppress the current gradients and in
turn the current strength inside the wave boundary layer, an effect which is
generally attributed to increased wave-induced mixing near the bed.
The current profile u (z) in a combined flow is commonly split conceptually
into three parts which are more obvious in Figure 1.5.2. than in Figure 1.5.1.

e RIPPLE TROUGH
+ RIPPLE CREST

ELEVATION
MILLIMETRES
IN

ELEMENTS

5 Le) I5 20 25 30
VELOCITY, cms!
Figure 1.5.2: Measurements of the steady component u_ and the amplitude U 1 of
the primary oscillatory component from a combined flow (following current) in a
wave flume. Data from van Doom (1981) Test V20 RA+RB.

AND SEDIMENT TRANSPORT 63

Chapter 1: Bottom boundary layer flow

There is the wave-dominated layer near the bed. Then there is a logarithmic
layer and finally an upper layer which covers sixty to eighty percent of the water
depth.
Since most of the sediment transport normally occurs in the two lower layers
ie, the wave-dominated layer and the logarithmic layer, the following sections are
focussed on those.

1.5.2. Resume of steady boundary layer concepts

This section gives a brief summary of those steady flow concepts, which will
be used in the description of wave-current boundary layer interaction.
The steady or quasi-steady flows involved in coastal sedimentary processes
are driven by either gravity, through a surface slope or density gradients, by wind
stresses on the surface, or by wave momentum flux.
The equation of horizontal motion reads

ou op at
essere? et
which becomes

ot
5 _gee
Op (1.5.2)

for steady flow.

If the pressure can be assumed hydrostatic, so that the pressure gradient is
constant over the depth and given by

op
ay PS on
> (1.5.3)
then, it follows from Equation (1.5.2), that the shear stress gradient is also constant
over the depth, and we have: In a steady flow with hydrostatic pressure
distribution, the shear stress distribution must be linear. See Figure 1.5.3.

BE
a v=eh reno
P&SOM
>, (1.5.4)
The bed shear stress t(0) defines the typical turbulent velocity at the bed

ux = Vt(0)/p (1.5.5)

64 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

Figure 1.5.3: In a steady flow

with hydrostatic pressure
distribution, the shear stress is
linearly distributed. The bed
shear stress T(o) defines the
typical turbulent velocity at the
bed: ux = Vt(0)/p.

which is called the friction velocity.

It is well established experimentally that the velocity profile in a steady,
uniform, turbulent boundary layer can be described by

a = = da (1.5.6)
Ux K Zo

where X is called von Karman’s constant and has a value of approximately 0.4,
see Figure 1.5.4.

; th z

| ~ factor e

Figure 1.5.4: Geometric properties of the logarithmic velocity distribution.

AND SEDIMENT TRANSPORT 65

Chapter 1: Bottom boundary layer flow

This logarithmic velocity law which is also called the “law of the wall” is not
a good approximation to the velocity distribution near the water surface, see for
example Figure 1.5.2, but it does provide a good and workable model where most
of the sediment transport, in steady flows, occurs.
The logarithmic velocity profile given by Equation (1.5.6) goes to zero at a
finite elevation z) above the bed. This is of course not physically realistic, but it
reflects the fact that Equation (1.5.6) is a horizontally uniform description, which
does not attempt to model the three dimensional flow details close to the roughness
elements. See Figure 1.1.5.
If the bed is perfectly smooth, a thin bottom layer exists within which the
flow is laminar. It is called the /aminar sublayer, and its thickness is generally
taken to be

Siam = 11.6 V/ux (1.5.2)

where v is the laminar viscosity.

If the irregularities of the bed surface are larger than jam, so that they
penetrate the laminar sublayer, we say that the flow is rough.
Nikuradse (1933) found that the logarithmic velocity profile over a bed of
closely packed spheres of diameter d_ goes through zero at

1
Z = 30 d (1.5.8)

when the flow is fully turbulent, and this has been observed to be the case for
rux/V > 70. Based on this observation we define the equivalent Nikuradse
roughness of other surface geometries by

r = 3020 (1.5.9)
where Zo is determined experimentally by fitting a logarithmic curve of the form
(1.5.6) to measured velocities.
The choice of origin for the z-axis: the theoretical bed level, is not obvious,
but it is generally defined as the level of the origin (z=0) which leads to the best fit
by Equation (1.5.6) to measured current profiles.
The equivalent roughness of many different surface types has been
investigated by Schlichting (1979), and the corresponding, theoretical bed level has
been discussed in detail by Jackson (1981).

66 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

In analogy with Newton’s equation

iii ge (1.5.10)
for shear stresses in laminar flow, we can define a turbulent eddy viscosity by

tT = pv; (1.5.11)
Then, from the logarithmic velocity distribution (1.5.6) and the
corresponding shear stress distribution

Tz) = t(o)(1-z/D) (1.5.12)

where D is the water depth, we find that the eddy viscosity in steady, uniform
flow with no surface stress is parabolic

vz) = Kuxz(1-—2z/D) (1.5.13)

Figure 1.5.5: The eddy viscosity

distribution which corresponds to
the logarithmic velocity profile is
parabolic. However, for sediment
transport calculations, the linear
approximation near the bed is often
sufficient.

AND SEDIMENT TRANSPORT 67

Chapter 1: Bottom boundary layer flow

This is a simple and useful formula, but since it is based on the law of the
wall, Equation (1.5.6), it is not reliable near the free surface.
If the layer of interest is very thin compared to the total flow depth, we often
apply the so-called constant stress assumption. That is, the last terms in the
expressions for shear stress, Equation (1.5.12) and for eddy viscosity, Equation
(1.5.13), are left out. Thus, in the constant stress layer we have 7 = T(o) and
Vt = K Ux Z.
The velocity distribution in the constant stress layer is still logarithmic and
given by Equation (1.5.6), and the eddy viscosity corresponds to the straight line in
Figure 1.5.5.

1.5.3 The eddy viscosity concept for combined flows

The simplest way of defining the relationship between the velocity field and
the shear stresses in a turbulent fluid is by using the eddy viscosity concept.
However, we found in Section 1.2.8 that the eddy viscosity for a purely oscillatory
flow is much more complicated than that of a steady flow, and this holds even
more so for combined wave-current flows.
In order for the eddy viscosity to be useful in the sense that it enables the
greatest simplification of the equations of motion, it must be based on the total
transfer in the vertical direction of horizontal momentum, see Section 1.1.5.
For a steady, turbulent flow this leads to

Wile ce bY (1.2.34)
du du
dz dz
where the term —uw_ vanishes if the flow is horizontally uniform, because then
w =0 by continuity.
For oscillatory flows the concept of eddy viscosity is less trivial as discussed
in Section 1.2.8. However, an analogous expression can be derived for a turbulent,
purely oscillatory flow (u = 0)

t/, ~iiw — uw’

w
Vv = er
~ i, at
= ay a i, ap (1.2.35)
0z Oz
where the meaning of uw _ is defined in connection with Equation (1.1.17).
Let us now consider the eddy viscosity concept for a combined wave-current
motion.

68 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

Based on the expressions (1.1.22) and (1.1.25), for the steady and the
periodic components respectively of the total momentum transfer, we obtain the
following eddy viscosities

at/p hw Siw <a’

c
Vv, = di = = sey (1.5.14)
dz dz
and

Vv w =
Wp
a
_ -uw- iw-iw-ww’
= ai= = Ley (1.5.15)
Oz Oz

where Vc is the eddy viscosity felt by the steady flow component u(z) and Vy is
the one felt by the periodic component 1(z,t). The interesting thing about these
two expressions is that they are not similar at all. Judging from the appearance of
these expressions, it would be quite surprising if vc and Vy turned out to be
identical or just similar.
From (1.5.14) it is clear that vc must be constant in time, but vw may well
be a function of time as well as of z. As for the pure wave case, which was
discussed in Sections 1.2.8 and 1.2.9, it is quite possible for the expression (1.5.15)
to take both negative and infinite values.
Coffey & Nielsen (1984, 1986) showed examples of corresponding values of
Vc and Vw from wave flume data (van Doorn 1981 Tests V10 and V20) and from
tunnel data (van Doorn 1982, Test S10). They found that in both cases vc was
typically three to four times greater than vy .
A similar data set derived from the velocity data of van Doorn (1982) Test
M10 is shown in Figure 1.5.6. The plotted values of vy are calculated as lv|I
from the first harmonic of 4(z,f) in accordance with Equation (1.2.42) from
Section 1.2.9.
The plotted values of Vc are derived from

45 (z)/p Ux (1 z/D)
Ve = Se ( lJs
3) 16 )
du du

with “« determined from the logarithmic part of the u -distribution.

AND SEDIMENT TRANSPORT 69

Chapter 1: Bottom boundary layer flow

Elevation,
cm

1 2 3 4 5 6 tj 8 9
Eddy viscosity, cm

Figure 1.5.6: The magnitudes of the eddy viscosities vc (0) and Vw (+), based on
Equations (1.5.14) and (1.5.15), may well be very different. Data from oscillating
water tunnel, van Doom (1982) Tests M20, (A, T, r, <u>) = (0.33m, 2s, 0.0021m,
0.2m/s). The line corresponds to Vc = 0.4 ux z, see Figure 1.5.5.

Despite the scatter in the values of Vc it is quite clear that the magnitudes
of Ve and vw are quite different. Typically, vc is three to four times greater than
Vw.
While this conclusion is perfectly in line with those previously drawn by
Coffey & Nielsen (1984, 1986), it is acknowledged, that the determination of ux
through a log-curve fit introduces an element of arbitrariness when the extent of
the "logarithmic layer" is poorly defined.
The definitions (1.5.14) and (1.5.15), of Vc and vw are the ones applied,
with more or less explicit acknowledgement, in previous eddy viscosity based
models of wave current boundary layer interaction. There are however, other
options which may turn out to be more appropriate. i
It may for example be reasonable to give the dominating term — uw of the
total momentum transfer expression (1.1.22) explicit consideration in the equation

70 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

of motion for u(z) instead of hiding it under the collective label of T(z). This
approach is outlined in Section 1.5.9 page 91, and in Section 7.2 page 293.
With respect to the eddy viscosity concept, the advantage of this approach is
that the expression for the current eddy viscosity vc then becomes analogous to
that from steady turbulent flow as the term — Zw is removed. Compare the
formulae (1.2.34) and (1.5.14). The only difference will be due to the difference in
turbulence intensity.
With the definition (1.5.14) however, vc is a function of the angle between
the current and the direction of wave propagation. The term —iw , which is
positive or zero (see Figure 1.4.2) makes a positive contribution to vc for a
following current and a negative contribution for an opposing current.
Sleath (1987) found, for purely oscillatoryflow in an oscillating water tunnel,
thatuw was typically ten times greater than u’w’. It is therefore natural to expect
— uw to be the dominant term in ve defined by Equation (1.5.14). The magnitude
of —uw will be greater under real progressive waves than in a tunnel where the
contribution described by Longuet-Higgins Equation (1.4.5) is absent. The
corresponding difference between u-profiles in wave flumes and in tunnels can be
seen by comparing Figures 1.5.1 and 1.5.2. For more details, see Section 7.2.

1.5.4 Turbulence intensity in combined wave-current flows

The majority of existing models for wave-current boundary layer interaction
are based on the philosophy that the interaction and the observed changes to the
current profiles in particular are due to wave-induced changes in the turbulence
intensity rather than to terms like uw. We shall, therefore, take a look at some
measurements of the turbulence intensity in combined flows.
As a_measure of the turbulence intensity some measured values of
w’rms = (w’”)°* are shown in Figure 1.5.7. This data was recorded in an
oscillating water tunnel under five different sets of conditions: a pure wave motion
(A,T) = (0.10m, 2.0s); two cases of pure current with a depth average of 0.1m/s
and 0.2m/s respectively; and for the wave motion superimposed on each of the
currents. The strip-roughness bed had a hydraulic roughness of 2/mm.
The pure wave case is the one for which the boundary layer structure data is
shown in Figure 1.2.17. From that figure it may be inferred that this flow is
analogous to a smooth laminar oscillatory flow with z = V2vw/@ = 4.2mm
corresponding to Vw = 2.8:10-°m’/s.
The data in Figure 1.5.7 show a clear difference between the cases with
waves and those without.

AND SEDIMENT TRANSPORT Jl

Chapter 1: Bottom boundary layer flow

millimetres
in
Elevation

0.5 1.0 1.5 2.0wi

ms (cm/s)

Figure 1.5.7: Root mean square vertical velocity fluctuations in five cases; pure
current 0.Jm/s (*) and 0.2m/s (0) depth average, a pure wave motion (A, T) =
(0.10m, 2.0s) (x) and the waves superimposed on each of the currents (®) and (0).
Data from oscillating water tunnel, van Doorn (1982).

For all cases with waves, w’rms peaks at the top of the roughness elements
(z = 2mm ) with a maximum value roughly equal to 0.5 Ux .
For z < zj there is no apparent increase in turbulence intensity due to the
addition of currents, but for z > zy the data do not contradict the simple

iP? COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

superposition rule suggested by Sleath (1991). He suggested that the turbulence

intensity of combined flows could be estimated simply by adding the values for the
corresponding pure cases : W’rms,cw = W’rms,w + W’rms,c-
For the pure current cases, the turbulence intensity is almost independent of z
and proportional to the current strength, w’rms,c = 0.5 ux.
We note that the thickness of the layer with considerable wave-generated
turbulence is, in this case, of the order 5 z,; = 5-0.09 VrA ~ 0.5VrA . For this
experiment that coincides roughly with the magnitude of ix /@ and ofr.

1.5.5 The influence of currents on the wave boundary layer

The present section deals with the possible effects of a superimposed current
on the wave boundary layer structure. More precisely, the question is as follows.
Assume that the wave motion just above the boundary layer is unchanged and
given by

tao (t) = Awe (1.5.17)

How will the wave boundary layer structure then change if a steady current
flow is superimposed?
Most of the existing theoretical models, e g Grant & Madsen (1979),
Christoffersen & Jonsson (1985) and Myrhaug & Slaattelid (1989), suggest that the
addition of uu will increase the turbulence intensity and the eddy viscosity
applicable to the wave motion. Hence the wave boundary layer thickness should
increase aS VVw/@. However, the experimental evidence indicates that the effect
is somewhat weaker than the models predict. For example, the data in Figure 1.5.8
show no change to the wave boundary layer structure, and also the Figures 6 and 7
of Myrhaug & Slaattelid (1989) show that their model over-estimates the wave
boundary layer thickness compared to the measurements. That is, their predicted
increase of Vw seems to be exaggerated.
This may be understood by considering the expression (1.5.15) for Vw

ROE WS jae iw (1.5.15)

ou ou
oz dz

and remembering that Sleath (1987) found, for purely oscillatory flow, that the
term — iw totally over-shadowed the turbulence term -uv’w’. If the turbulent

73
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

contribution —w’w’ to vw is generally insignificant it is unlikely that

current-induced turbulence will have much impact on vw in a combined wave-
current flow. The lack of sensitivity to the superposition of currents is also
indicated by the turbulence intensity measurements shown in Figure 1.5.7, in
particular for z < 2}.

© NO CURRENT
x U,/U,*.51
2 e =.

ROUGHNESS
ELEMENTS -£n 1D)
1 0.2 04 O6 08 10 20 40 60 80100

Figure 1.5.8: Measured values of the dimensionless velocity defect Dj(z) for tests
VOORA, VIORA, and V20RA from van Doom (1981). Even for superimposed
currents as strong as iix/ttx = 0.79 there is no evidence of change to the wave
boundary layer structure.

The absence of experimental evidence for current-induced changes to the

wave boundary layer structure prompted Coffey & Nielsen (1986) to put forward a
simpler wave-current interaction model which ignored current-induced changes to
the wave boundary layer. It is a clear practical advantage to be able to ignore the
effect because then, the wave component of a combined boundary layer flow can
be calculated as if the waves were there alone.

74 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

It must of course, be expected that with very strong currents, such as some rip
currents, the wave boundary layer structure can no longer be assumed unchanged.
However, the presently available data show no significant effect for the parameter
range 0 < Ux/ix < 0.79, see Figure 1.5.8.
The data in Figure 1.5.8 correspond to colinear waves and currents, but
recent measurements of the perpendicular case by Sleath (1990) also show little or
no change to the wave boundary layer structure due to superimposed currents of
moderate strengths, ux/A @< 0.08.

1.5.6 Energy dissipation by waves superimposed on a current

Several authors (Kemp & Simons 1983, Asano et al 1986 and Simons et al
1988) have observed that waves travelling along a wave flume will lose height
more rapidly if they travel on an opposing current and more slowly if travelling on
a following current. It has been suggested that this is due to large changes of the
wave boundary layer structure and a resulting change in energy dissipation rate.
This is however, an interpretation which depends on the definitions applied.
In the following, it will be demonstrated that the above mentioned effect can
be attributed mainly to the change in wave group velocity (which together with the
wave height determines the wave energy flux), rather than to a change of the wave
boundary layer structure. This agrees with the measurements by Sleath (1990)
which showed very little change of the wave friction factor due to superimposed
currents of moderate strengths (u*/A w < 0.08).
By applying the definitions for energy dissipation and shear stresses
suggested by Christoffersen & Jonsson (1985) and by assuming that the waves and
the current have separate energy budgets the interpretation of the wave height
attenuation data comes into line with the evidence presented in the previous
section. That is, the data indicate that the influence from a current on the wave
boundary layer structure and hence on the wave energy dissipation rate is weak.
When experiments with extremely large relative roughness (r/A > 10) are
excluded, the presently available data does not indicate significant changes to the
effective wave energy dissipation rate or to the wave boundary layer structure for
relative current strengths ( u«/ ie ) up to about /0. See Figure 1.5.9.
The energy dissipation factors in Figure 1.5.9 are based on the following
definitions. It is assumed that the waves and the current have separate energy
budgets so that the energy balance for waves in a flume with uniform cross section
reads

75
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

er (1.5.18)

where Ef is the wave energy flux per unit length of wave crest and Dz is rate of
wave energy dissipation per unit bed area, see Section 1.2.6. The energy flux is -
according to linear wave theory - given by

Ef = 5PR? (Cpr + U) (1.5.19)

where U is the depth-averaged current velocity and Cgr is the group velocity of
the waves in the frame of reference which moves with the current. This group
velocity can be calculated from the usual formula

2kD
Op anB 4n
le anit Dear sinh 2D?
(1.5.20)

100

0.2

0.1
1 eZ 2 S 10 20 S16)
Figure 1.5.9: Wave energy dissipation factors derived, through Equation (1.5.29),
from the wave height decays measured by Simons et al (1988). The symbols *, 0, 0,
A correspond respectively to the following relative current strengths (ux/ttx S70)
1.3-2.1, 4.1-9.8, 5.5-10.5. The curve corresponds to Swart’s formula, Equation
(E222);

716 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

applied in the frame of reference which travels with speed U. That is, Ty is the
wave period in this moving frame of reference and the wave number must be
determined from the dispersion relation
2
kDtanhkD = —>D = alt = | (1.5.21)

where Tq is the wave period seen from the laboratory frame of reference.
For the wave energy dissipation we apply Jonsson’s (1966) definition

2
De = 3 pfe(A@g)° (1.5.22)
so we have

Gala 2
me Epg H* (Cor + »)hee 0 fe (A@a)’ (1.5.23)

where Cygr+U is a constant when the depth is fixed and the near bed velocity
amplitude is given by

7H
A@q = Tq sinh kD (1.5.24)

Then the energy dissipation equation can be reduced to

pa fed ites SteeysreeBU febsionsd

> i (1.5.25)
dx |H 397? (Cgr + U) sinh” kD

which, with the starting conditions H(xo) = Ho has the solution

H
Hx) = ; i (1.5.26)
1
Lot Hy igreTS; ao)
3gTa (Cer + U) sinh” kD
In most previous studies, the observed wave height variation has been fitted
by an exponential which corresponds to viscous or laminar energy dissipation, i e

H(x) = Hx) oO -*2) (1.5.27)

AND SEDIMENT TRANSPORT 77

Chapter 1: Bottom boundary layer flow

and the measurements have been discussed in terms of the dissipation factor

a= -—— (1.5.28)

For the real wave height variation under turbulent conditions, which is given
by Equation (1.5.26) there is no constant corresponding to @ but for small
energy dissipation rates we have

8"
Wo a 3ee fe (1.5.29)
3 g Ta(Cgr
+U) sinh® kD

This shows how, for fixed fe, the observed o-values will decrease for
increasing U and vice versa. This effect of the mean current on dH/dx was also
noted by Simons et al (1988).
The fe-values in Figure 1.5.9 which are based on measured o-values and on
the analysis above show that fe and presumably the general structure of the wave
boundary layer are practically unchanged by the superposition of currents. This
agrees with the friction factor measurements of Sleath (1990) and with the findings
related to Figure 1.5.8 about the wave velocity defect function.
Hence, the data indicate that, for practical purposes, the wave boundary layer
structure and the wave energy dissipation factor for a combined flow can be
calculated as if the current was not there. That is of course, after the near bed
velocity amplitude Aq@ _ has been calculated with due respect to the current by
using the dispersion relation (1.5.21). The conclusion above should not be applied
uncritically to conditions outside the range of conditions represented by the data in
Figures 1.5.8 and 1.5.9.

1.5.7 The influence of waves on current profiles

In relation to the modelling of coastal processes the most important aspect of
wave-current boundary layer interaction is the change of current velocity
distributions due to the superposition of waves.
The scene can be set by either of the following two questions

a) Assuming that the average bed shear stress and hence ux is kept
unchanged, what is the change to the current profile due to the
superposition of waves? See Figure 1.5.10, left-hand side.

78 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

b) What are the changes to the current profile when waves are added in
such a way that the reference current velocity uy = u(zr) is kept
unchanged? See Figure 1.5.10, right.

Inz Inz

Figure 1.5.10: Two ways of considering the problem of wave-induced changes to

current profiles. Left) The friction velocity ux is considered known and fixed. right)
The velocity at a certain level zy is considered known and fixed.

The first approach which was taken by Lundgren (1972) is the most rational
because it leads to logically straight model building, while the latter approach leads
to iterative thinking, as applied in the model of Grant & Madsen (1979).
In the following we shall follow Lundgren’s line of thought where u« is
taken to be known and fixed, - for example determined by a fixed mean surface
slope

ix = (1(0)/p)°° = (-gD di/dx )°°

Then, the wave-induced changes to u(z) are as outlined by Figure 1.5.11.
Let the undisturbed current profile be given by

uz) = raIn= (1.5.30)

AND SEDIMENT TRANSPORT 79

Chapter 1: Bottom boundary layer flow

Wave induced
mixing

Ve

Figure 1.5.11: The wave-induced changes to current profiles may be simplistically

described as follows. Close to the bed (z < /) the current gradients will be
suppressed due to wave-induced mixing. Further from the bed, wave-induced
turbulence is weak and the current profile is logarithmic, but its zero-intercept will be
at za instead of zo corresponding to an apparent roughness increase

where Zo = r/30 as usual.

With waves superimposed, there will be a layer of thickness / close to the
bed with direct wave influence on u(z). That is, inside this layer the current
gradient du/dz is suppressed due to wave-induced mixing, see Figures 1.5.1 and
Pew
Judging from the turbulence measurements shown in Figure 1.5.7, the
thickness / of the layer with predominantly wave-generated turbulence should, in
that particular case, be of the order

bow 5 21.5 . 009Nr A. =. 05 Ved (1.5.31)

see Equation (1.3.10), or alternatively,

l= tx /@ (1.5.32)
However, the relationship between / and the various boundary layer thickness
values (Section 1.2.7) may depend on the relative roughness r/A , the relative

80 COASTAL BOTTOM BOUNDARY LAYERS

 1.5; Wave-current boundary layer interaction

current strength ux/ ux and possibly on the angle @ between the current and the
direction of wave propagation

tt = Fi(y>
I Ux
,Q). (1.5.33)
Zo A@

The shape of the current profile for z</ seems to be as much like a linear
function of z as any other simple shape including the logarithmic shape suggested
by Grant & Madsen (1979), Fredsoe (1984), and by Christoffersen & Jonsson
(1985), at least for moderate values of r/A, see Figure 1.5.12.
Outside this layer, the wave-generated mixing is relatively weak, so the law
of the wall may be assumed to apply in which case the current distribution must be
logarithmic and can be written as

= Ux vE
u(z) = a Lape forez¢> (1.5.34)

where 2g is analogous to, but larger than Zo = 7/30.

The existence of a logarithmic layer corresponding to Equation (1.5.34) is not
always equally obvious in the experimental data, compare Figures 1.5.1 and 1.5.2,
and in some cases the fitting of a logarithmic curve is somewhat arbitrary.
However, the description above, which is essentially due to Lundgren (1972), is
reasonably adequate considering the presently available experimental details.
The apparent roughness increase 7Zg/Zo and the corresponding velocity
reduction

AuOMT = —In—
hg
K Zo
(1.5.35)
can be expected to depend on the relative current strength, the relative roughness,
and on the angle between the current and the direction of wave propagation

4 ~ F(—,-,9) (1.5.36)
oO *

The presently available data indicate that, the influence of @ on Za/Zo is

strongest under real waves where it stems from the term — uw , as discussed in
Section 1.5.9. In tunnels and on oscillating plates where the main part of this
momentum transfer term is absent, the angle @ seems to have little impact on
Za/Z. Thus, the "@ = 90°data" of Sleath (1990) measured over an oscillating plate,
follow the same trend as the tunnel data of van Doorn (1982), see Figure 1.5.13.

81
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

1000

100

10

1.0

Grant and Madsen (1979)

U [cm/s]

0-16 5 10 15 20
Figure 1.5.12; Even for fairly small relative roughness (r/A), the current profile in the
inner, wave-dominated layer is not logarithmic. Data from Sleath (1990), A =
0.141m, T = 1.76s, r= 1.49mm, ux = 1.9cml/s (x) or 2.3cm/s (0).

There is some evidence however, which suggests that opposing currents

(@ = 180°) are affected in a different way than following currents (9 = 0°). Thus
the "opposing current data" of Kemp & Simons (1983) plot somewhat higher than
the "following current data" from 1982 in Figure 1.5.13. That is, the apparent
roughness increase is greater for opposing currents than for
following currents.
This can be explained by considering explicitly the “w-terms in the equation
of motion (1.1.21) instead of making them part of T as done in the present,
simplistic approach. See Section 1.5.9.
Based on the data, which was then available, Coffey & Nielsen (1986)
suggested that it might be possible to express the apparent roughness increase as
function of a single parameter, namely the friction velocity ratio tux. They
suggested the formula :

z
—Zo = 140.06 ig
ifs (1.5.37)
Ux

82 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

100

40

20

10

1 2 4 10 20 40 100 200 400

Figure 1.5.13: Apparent roughness increase as function of relative current strength

and relative roughness. The curve corresponds to the theory of Sleath (1991). o:
Kemp & Simons 1982 flume @ = 0, e: Kemp & Simons 1983 flume 9 = /80°, e:
Asano et al 1986 flume @=0°, A: van Doom 1981 flume 9 = 0°, o: van Doom
1982 tunnel ,x,+: Sleath 1990 p= 90°

However, Sleath (1990) found that this formula seemed inadequate for data
with smaller relative roughness ( r/A < 0.J ). Subsequently Sleath (1991)
developed a model which leads to the expression

ET: fe
*
ue (1.5.38)
Zo

which is shown together with the presently available data in Figure 1.5.13.
Several field studies, e g Cacchione & Drake (1982), Grant et al (1983),
Coffey (1987), Lambrakos et al (1988) and Slaattelid et al (1990), provide useful

83
AND SEDIMENT TRANSPORT
Chapter 1: Bottom boundary layer flow

data on zq . However, since the corresponding values of the hydraulic roughness r

(and hence of zo=r/30 ) are very hard to estimate, it is impossible to make firm
conclusions about the detailed behaviour of zg/zo from field data at present. There
is still an urgent need for data on wave-current boundary layer interaction over
beds with well known roughness.

1.5.8 An empirical model for current profiles in the presence of waves

Several models of combined wave-current boundary layer flows have been
presented over the last two and a half decades. However, because detailed data is
only now starting to become available, most of these models have been based on
theoretical adaptation of steady flow concepts such as the law of the wall and
Prandtl’s mixing length model for momentum transfer in steady boundary layers.
With the exception of those of Bijker (1967) and Fredsoe (1984) all the
previous models apply the eddy viscosity concept, and it seems reasonable enough
to apply this simplistic approach at the present state of the art.
Some of the previously assumed eddy viscosity distributions are shown
qualitatively in Figure 1.5.14.
With the exception of that of Sleath (1991), all of these eddy viscosity
distributions are discontinuous at z=/, and in all cases it was assumed that the
periodic velocity component “% would feel the same eddy viscosity as the current
component WU, Ve = Vw.
The eddy viscosity concept will also be applied in the following model
development but with one major difference from previous models. We shall not
assume that the wave component u(z,t) feels the same eddy viscosity as the steady
component u(z). Instead, we shall acknowledge the empirical fact that the eddy
viscosity Vc which is felt by u(z) is generally about four times larger than vw
which is felt by “(z,t)

Ve = 4Vw (1.5.39)

see the discussion in Section 1.5.3, page 70.

For the sake of simplicity, it shall also be assumed that “(z,f) can be
calculated from (A, @,r) alone, as if the current was not there. This approach
seems justified by the available experimental data, which indicate that the wave
boundary layer structure, represented by u(z,t)/A @, changes very little even with
fairly strong currents superimposed (see Sections 1.5.4 and 1.5.5). The velocity
defect data in Figure 1.5.8 show no effect from superimposed currents with

84 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

Figure 1.5.14: Some of the eddy viscosity distributions assumed by previous authors
for combined wave-current flow (shown qualitatively). G&M: Grant & Madsen
(1979), C&J: Christoffersen & Jonsson (1985), M&S: Myrhaug & Slaattelid (1989),
Sleath (1991). In all of these studies, it was assumed that the waves felt the same
eddy viscosity as the current.

strength Ux/ its up to 0.79. In addition, the wave energy dissipation data in
Figure 1.5.9, page 76, indicate no change to the wave friction factor for much
higher values of the relative current strength uU+/ be .
The essence of the problem is then to specify the distribution of the eddy
viscosity Vc which governs the distribution of u(z) .
Following the line of thought of Lundgren (1972) we assume that an outer
layer exists, where the eddy viscosity felt by the current u(z) is given by

Ve = KUKeZ for z>/

just as in a pure current boundary layer which is thin compared with the flow depth
(a constant stress layer), see page 68.

AND SEDIMENT TRANSPORT 85

Chapter 1: Bottom boundary layer flow

Inside the lower, wave-dominated layer, we assume that Vc is constant and,

in order to obtain continuity of vc at z=/, we must then have

Ve = Kuxl for z<1 (1.5.40)

see Figure 1.5.15.

According to the constant-stress layer assumption (page 68) which
corresponds to the eddy viscosity distribution (1.5.39), the steady component of the
shear stress is uniformly given by t=pus« and hence the current gradient is
given by
Ux
ey fOleeZ!
du
— = (1.5.41)
dz zx
i for z>l/
Kz

Ku, 2@ U,/K

Figure 1.5.15, left: The assumed distribution of the eddy viscosity vc which is felt
by the steady component u(z) of a combined wave-current boundary layer flow.
Right: The resulting velocity distribution.

Starting with “=O at z= 0 we get the linear current distribution

86 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

u(z) =
||
x
4 for z</ (1.5.42)
in the inner layer.
In the outer layer the current distribution is logarithmic and may be written

& a 4
u(z) = PsIn D5 On *2>"] (1.5.43)

where matching with the lower-layer solution is obtained when

ie a ee (1.5.44)
At the present state of the art, it does not seem possible to derive a general
expression for / which is valid for all flow conditions (see the discussion related
to Equation 1.5.36). As more experimental information becomes available, for
example in the form of Equation (1.5.36), it will be straightforward to make use of
it using the framework of Figure 1.5.15 with ] and zg determined from

Oa he gD ae)

Example 1.5.1: A simple estimate of F for fairly rough conditions

It seems futile at present to speculate too much about the detailed form of the
function F i =e @) since so many aspects of the underlying physics are
*

unknown. However, for fairly large relative roughness, 0.06 < r/A < 0.5, where
the structure of the wave boundary layer corresponds to a constant, real-valued
eddy viscosity (see Section 1.2.9, page 40), a simple estimate of F can be
obtained if the influence of @ is neglected.
For such wave boundary layers, the eddy viscosity can be estimated by

Vw = 05az¢ =~ 050(0.09VrA )* = 0.004m@rA (1.3.14)

see page 52. With this estimate of vw and with the observation that vc = 4 Vw
(see Section 1.5.3, page 70), the following estimate for vc is obtained

Ve = 0.016@MAr_ for z<!l (1.5.46)

AND SEDIMENT TRANSPORT 87

Chapter 1: Bottom boundary layer flow

Then, since the current eddy viscosity in the lower layer is written as
Vc = Kul (see Figure 1.5.15) , this gives the following expression for /

1 = 0.016 Aw , (1.5.47)
K Ux

and hence with zg = el, «=0.4, and zo = 1/30

100

40

20

10

1 2 4 10 20 40 100
Figure 1.5.16: Apparent roughness increase as function of the relative current
strength for the same data which was shown in Figure 13.13, The straight line
corresponds to Equation (1.5.48). <

88 COASTAL BOTTOM BOUNDARY LAYERS

 1.5: Wave-current boundary layer interaction

F=— = 044— (1.5.48)

This expression for F corresponds to the line in Figure 1.3.16. The

agreement is fair, except for the data of Sleath (1991) (+), for which zg/zo is
systematically underpredicted. These data are, however, outside the validity range
of the assumed eddy viscosity formula (1.3.14), page 52, so this is not surprising.
The validity range of Equation (1.3.14) is approximately 0.06 < r/A < 0.5, while
the mentioned data correspond to r/A = 0.011 or r/A = 0.020.

Although the model in the example above was developed following

Lundgren (1972) i e, the time-averaged friction velocity was assumed known and
fixed, it can still be applied to problems where current friction velocity ux is not
given.
Consider the situation where data from a current meter at z=z,y above the
wave boundary layer are available, so that we know the values of [A, , u(zr), Q].
Assume further that the bed roughness and hence Zo is somehow known.
Information about current distribution in general and about the time averaged
friction velocity in particular can then be derived in the following way.
We assume that zr is within the logarithmic part of the current profile.
Hence, we have, in accordance with Equation (1.5.34)

Ux Zr Ux Zr
u(zr) = nae eee ore (1.5.49)
0 Ux > A’

where the unknown is the time-averaged friction velocity ux. The form of the
function F is assumed known, for example, given by Sleath’s (1991) expression
(1.5.38) or by Equation (1.5.48).
For the purpose of iterative solution, it will generally be convenient to rewrite
this Equation (1.5.49) in the form

(1.5.50)

which will normally converge rather rapidly.

89
AND SEDIMENT TRANSPORT
 Chapter 1: Bottom boundary layer flow

When the current friction velocity u* has been found from the iteration, the
value of F can be calculated and so can Zqg = Zo F and / =e'z )F. All
components are then available for drawing the velocity distribution

mk for z<l
Kk /
u(Zz) = (1.5.51)
Ux Z
in for z>/
K Za

Example 1.5.2: Current profile from wave data and reference current
To illustrate the use of the model above for deriving details of the current
distribution u(z) from the current velocity at a single point plus the wave data and
the bed roughness, consider the numerical example [A,7,r,u(1.0m)] =
[0.9m, 85, 6cm, 0.45m/s].
Inserting these data into Equation (1.5.50) with F given by Equation
(1.5.48) gives

ea wistae
ie Sea.
le finn”
“ngEs' teeP
pet Ou 0 tire (M
: a
0.06/3
which converges rapidly to the friction velocity value u« = 0.0425m/s . This
corresponds to F = 0.44 “2=7.32, and hence za= zo F =0.0146m, and | = zq
2

e = 0.040m. The corresponding velocity distribution, given by Equation (1.5.51), is

shown in Figure 1.5.17.
Alternatively, if Sleath’s (1991) expression (1.5.38) is used for F in
Equation (1.5.50), we get

SESE Ot
EOSONG 90 S ee ee Co
Mast 0.18
Ux =
1.0 0.9: =2n/8 [0.9
0.06 ) 6.22 -In(1+
0.52
= )
In 0.06/30 ~ In (1 +0.19

(1.5.53)
which equally rapidly converges to a friction velocity of ux = 0.048m/s. With this
value, the model of Sleath (1991) gives the velocity distribution which is shown by
the broken line in Figure 1.5.17.

90 COASTAL BOTTOM BOUNDARY LAYERS

 1.5; Wave-current boundary layer interaction

Figure 1.5.17: Velocity profiles obtained from the data set [A,7,r, u(1.0m)] =
[0.9m, 8s, 6cm, 0.45m/s], with two different wave-current boundary layer models.
The full line corresponds to the model defined in Figure 1.5.15 with F = za/zo = 0.44
A@/ux , while the broken line corresponds to the model of Sleath (1991).

1.5.9 Wave-current interaction models with explicit consideration of uw

As mentioned above, a detailed theoretical prediction of the function Zg/zo =

F A® r ) will not be attempted in the present text. However, one particular

Ux’ A’
aspect of the available experimental data, which is not addressed directly by the
simplistic approach above, is easily addressed in theoretical terms. That is the

AND SEDIMENT TRANSPORT 91

Chapter 1: Bottom boundary layer flow

indication by the data that opposing currents (@= 180°) experience a larger,
apparent roughness increase Za/Zo than following currents, or in other words,

ee 130°) > ee ) (1.5.54)

see Figures 1.5.13 and 1.5.16.

To get an explanation for this we must give explicit consideration to the term
aw in the time-averaged equation of motion (1.1.21) instead of hiding it under the
collective label of "total shear stress" as done in the simplistic approach above.
For a two dimensional, uniform case (2 =0) Equation (1.1.21) can be

written

One -Old <s5 L)ODe ecdvs

Ocinabeo
UVa ee
Ninehusmesieal
WY =
amanda i
1.4.

which may be rewritten in terms of an eddy viscosity defined by

—wWww’
Val == V + aa
— (1.5.55)
ES:

az
(which is different from the definition used in the simplistic discussion above, see
Equation (1.5.14)), and integrated once to yield

ou _map lop,
Velcazs fig ai =,
ot 1)
nw’ (1.5.56)

For a hydrostatic mean pressure gradient due to the mean surface slope a

where the mean bed shear stress equals —pgD a. Equation (1.5.56) can be
rewritten as

du
Vala = ¢ (-D) 1
2 +iw = = -gp0-5a Tigo
+ UW (1.5.57)

which, for a thin bottom boundary layer where z<<D, can be conveniently
approximated by

Nels uw = Iuelix + aw (1.5.58)

p
92 COASTAL BOTTOM BOUNDARY LAYERS
 1.5: Wave-current boundary layer interaction

where the friction velocity is defined by

ln. = eS & 9D a
The current distribution is then obtained by integration and use of the
non-slip boundary condition u(o) = 0

7 luslite + iw
Z ———

u(z) = | err (1.5.59)

oO

With u positive in the direction of wave propagation, the term aw is likely

to be always negative, see page 55. The effect of iw on the current distribution
(with fixed ux) is therefore to reduce the magnitude of u(z) for following currents
and increase it for opposing currents.
However, when Zq is found for each profile by extrapolation of the upper,
logarithmic part of the current profile, one finds larger zg/zo for the opposing
currents than for the following currents, see Figure 1.5.18. This agrees with the
data of Kemp & Simons (1982, 1983), see Figure 1.5.13, page 83.
In addition, Equation (1.5.57) shows that due to uw being negative, the
current gradient (du/dz), for a following current, will become negative close to
the free surface. This was observed by Kemp & Simons (1982) and can be seen
from the wave flume data of van Doorn (1981) in Figure 1.5.2, page 63. wad
For perpendicular currents and for currents in an oscillating water tunnel “w
is much smaller because the the part of it, which is equivalent to the laminar
expression (1.4.5), page 57, is absent. Thus, the logarithmic part of the u-profile
shows the "correct" slope, and du/dz does not turn negative, see Figure 1.5.1.
The situation is demonstrated qualitatively by the example in Figure 1.5.18
where the current distribution has been calculated from Equation (1.5.59) with the
eddy viscosity distribution

ee 5s for 2/85 < 4

Vel WS
0.25 luxl z for z/8s = 4 (1.5.60)

and with uw based ona constant eddy viscosity of Vw = 0.5 @ 5s.

The magnitude of (uw)! under progressive waves, relative to the
maximum wave-induced bed shear stress 7 may be estimated roughly on the basis
of their respective values for boundary layers with constant eddy viscosity:

— (iw) *Ya (Aw) k 8s 1

PA ae itescyl kA (1.5.61)
Tt /p fw(Aay 22
AND SEDIMENT TRANSPORT 93
Chapter 1: Bottom boundary layer flow

10

ZO

ea 2 4 6 8 10 12
Figure 1.5.18: Current distributions for a pure current and for a current with
following, opposing and perpendicular waves superimposed according to Equation
(1.5.59) with the eddy viscosity distribution (1.5.60). The relative "drift strength"
| (GW) oo/un | is, for both following and opposing waves, equal to 0.5. Legend: x:
pure current, *: following waves , +: perpendicular waves or tunnel, rectangle :
Opposing waves.

since the wave friction factor for constant eddy viscosity amounts to
2/V\Re = 2Vv/(A’@) = V2 8s/A, see Equations (1.2.16) and (1.4.4).
It seems rather natural to incorporate uw into new models of wave-current
boundary layer interaction, along the lines suggested above. For further incentive
see also Section 7.2, page 293. However, some questions need to be addressed
regarding the general nature of “w in turbulent boundary layers over rough beds.
Longuet-Higgins’ expression (1.4.5) was derived for laminar flow over a
plane bed and the only vertical velocities involved are due to the thickening and
thinning of the boundary layer under a progressive wave. Over a rough bed
however, there will be additional periodic, vertical velocities caused by the bed
geometry, and the significance of their contribution to “uw must be considered.

94 COASTAL BOTTOM BOUNDARY LAYERS

 CHAPTER 2

SEDIMENT MOBILITY, BED-LOAD

AND SHEET-FLOW

2.1 FORCES ON SEDIMENT PARTICLES

2.1.1 Introduction
For the purpose of sediment transport modelling, it is necessary to consider
three types of forces which govern the behaviour of cohesionless sediment
particles whether they are resting at the bed or moving around in a slurry or a thin
suspension. These are: the gravity force Fg = Mg ; intergranular forces related to
collisions or continuous contact; and the fluid forces which may be due to surface
drag or fluid pressures.

2.1.2 Intergranular forces

The intergranular forces are well understood as far as resting (non-shearing)
grains are concerned.
For resting grains the static angle of repose @s is determined by the
frictional coefficient, i e the ratio between the effective normal stress Oe and the
maximum sustainable shear stress Tmax , see Figure 2.1.1.
The angle @s applies to dry sand or to sand which is entirely under water.
Sand which is wet but not saturated may stand at a much steeper angle because the
negative porewater pressure increases the effective normal stress Oe.
For most sandy materials the static angle of repose is between 26 and 34
degrees, being greatest when the material is most densely packed. For moving
(shearing) materials of near maximum concentration, the frictional coefficient.
Hence the dynamical angle of repose, @da_is of similar magnitude to Qs.

AND SEDIMENT TRANSPORT 95

Chapter 2: Sediment mobility, bed-load and sheet-flow

Figure 2114) ihe

frictional coefficient
(t/ Oe)max equals tan
Qs, where @s is the
static angle of repose.

Hanes & Inman(1985b) suggested a typical value of 3] degrees for beach sand.
When a horizontal sand bed is exposed to a fast, steady flow, a finite top
layer of sand will start to move with the flow, partly as bed-load and partly in
suspension. The fact that the moving layer is of finite thickness is significant
although seemingly trivial, because it shows that the moving sand has increased
the strength of the sand below.
Since the shear stress is not decreasing downward, the top layer of immobile
sand is able to withstand the shear stress which eroded the top layers when the
flow was started. This is due to the fact that the moving sand is transferring at least
part of its weight to the bed as effective stresses and thereby increasing the
effective normal stress in the bed. The effective nomal stress transferred by the
moving sand is generally referred to as the dispersive stress.
Bagnold (1954,1956) studied the normal and tangential stresses in granular
flows and suggested that they be given as functions of the shear rate du/dz and of
the linear sediment concentration 2 which is the relative surface proximity
between sediment particles, see Figure 2.1.2. The linear sediment concentration is
related to the volumetric concentration c by

1
= oo 2 ii
(CmaxZ0, eed Ge)
where Cmax is the maximum concentration corresponding to grain contact. The
linear concentration 4 increases drastically as c approaches cmax, see Figure
aah

96 COASTAL BOTTOM BOUNDARY LAYERS

 2.1 Forces on sediment particles

k—
d —

0) 0.2 0.4 0.6

Figure 2.1.2: Bagnold’s linear sediment concentration A becomes infinite as the
volumetric concentration c approaches its maximum value cmax which is generally
of the order 0.65.

Bagnold considered two different regimes in which different types of

interactions dominate the behaviour of the fluid-grain-mixture. For small, light
grains in a very viscous fluid the interactions are dominated by viscosity and
Bagnold termed this the macro-viscous regime. For large, dense particles at high
shear rates the interactions are dominated by particle collisions and this is called
the inertial regime. The dimensionless parameter which separates the different
regimes is
sf Vn &
B ore Sat sae (2.1.2)

which has been termed the Bagnold number. A purely inertial regime is found for
B > 450, and a purely macro-viscous regime corresponds to B < 40.
In the inertial regime the stresses are proportional to the linear sediment
concentration and to the shear rate squared. Bagnold obtained the following semi-
empirical formula

rn
Syren Oe =idee
55 ps(Ad )
Fis for B > 450 (2.1.3)

AND SEDIMENT TRANSPORT 97

Chapter 2: Sediment mobility, bed-load and sheet-flow

where the dynamical angle of repose was found to be about 18 degrees

corresponding to tan @¢d = t/Oe = 0.32. Bagnold’s experimental results, which
were obtained with neutrally bouyant particles have generally been confirmed by
more recent studies involving heavy particles, e g Hanes & Inman (1985b).
Theories have also been proposed which take more of a “basic principles
approach” to the problem than Bagnold did. However, while these theories may be
more satisfactory from a theoretical point of view, they are generally inferior to
Bagnold’s formulae with respect to accuracy, especially at high sediment
concentrations. For a review, see e g Hanes and Inman (1985a).
For the viscous regime Bagnold found

T du
Oe ae Oe = 13(1+A)(1+A/2)
pv i for..8 < 40.2.1)

with a somewhat greater stress ratio than in the inertial regime, namely
tan Qd = T/6e = 0.75.
With typical values (s, d, 4, du/dz, v) = (2.65, 2:10%m, 1, 100 s!,10° m/s),
corresponding to B = 10.6, sheet flow conditions under waves will generally be in
the macro viscous regime.

2.1.3 Fluid forces

The fluid forces on sediment particles are of two kinds. Namely surface drag
forces and pressure forces resulting from pressure gradients in the fluid.
The total pressure force which is determined as the surface integral of the
pressure is by Green’s theorem equal to minus the volume integral of the pressure
gradient, Vp = (ce
ae Bay Hence, for the situation in Figure 2.1.3 where the
)
pressure along a vertical is hydrostatic and there is a constant horizontal pressure
gradient, the total pressure force on the body is

Fp = ve
Er Mek
ax (2.1.5)
a dz paV

where the vertical component is the familiar bouyancy force corresponding to

Op/dz = —pg. V_ is the particle volume.

98 COASTAL BOTTOM BOUNDARY LAYERS

 2.1 Forces on sediment particles

Pg

dp
Ox
Figure 2.1.3; The total pressure force is, by Green’s theorem, equal to the volume
times the pressure gradient when the pressure gradient can be considered constant
over the body.

If the shear stresses in the fluid are relatively small (and in particular for
inviscid fluids), the equation of motion (1.1.10) may be used to write the horizontal
pressure force in terms of the horizontal fluid acceleration

ye
Va _ pV du
oa (2.1.6)
When calculating the pressure force Fp ona body which is held fixed while
the fluid is accelerating past it, an extra mass oCywV_- must be added
corresponding to the volume of surrounding fluid which the body keeps from
accelerating, so for a fixed body in a horizontally accelerated fluid we get

Fpx = p(1+Cw) v (2.1.7)

Correspondingly, the force required to give a sediment particle with volume V and

AND SEDIMENT TRANSPORT 99

Chapter 2: Sediment mobility, bed-load and sheet-flow

du ' He
relative density s the acceleration ee through aresting fluid is

d
F = p(s+Cw) Vi (2.1.8)
where p(s+Cm) V_ is called the virtual mass of the body, and pCmV_ is called
the added hydrodynamic mass. The added mass coefficient for a sphere is 0.5, for
a long cylinder itis 1.0, see e g Lamb (1936).
For a particle which is fixed in a wave motion with the homogeneous velocity
field u=A@sin wt Equation (2.1.7) gives

Fp = p(1+Cmu)
VA cos wt (2.1.9)
Apart from the pressure forces described above, which can be evaluated on
the basis of inviscid flow theory, a particle exposed to a viscous or turbulent flow
will in addition feel drag forces. Drag forces occur in two varieties: skin friction
and form drag. Skin friction contributes most of the drag on slender, streamlined
bodies like kajaks, while form drag dominates for plump shapes like spheres and
most sediment grains. The drag force is normally given on the form

ti pAColulu (2.1.10)

where A (~1d’/4) is the cross sectional area facing the flow, and Cp is the
drag coefficient which depends on the sediment shape and on the Reynolds
number, dlul/v. When the flow becomes laminar the drag force is actually
proportional to the flow velocity and, for small (d lul/v < 1) spherical particles, the
drag force is given by Stokes’ law

Fp = 3npvdu Qainty

which corresponds to a drag coefficient of

24
Co = dlul/v
(2.1.12)
Drag coefficients for spherical particles are plotted in Figure 4.2.1, page 164.
By comparing the expressions (2.1.9) and (2.1.10) for the pressure force and
the drag force respectively on a spherical particle in a wave motion, we see that the
force amplitude ratio

100 COASTAL BOTTOM BOUNDARY LAYERS

 2.1 Forces on sediment particles

Femax ise (1+Cu) A

P64 2
sibtach 1 Ff
R 2c Aa = (2.1.13)

is proportional to d/A which is called the Keulegan Carpenter number.

A sand particle in a flat sand bed will on the average have to carry a drag
force of magnitude To d° but the instantaneous force experienced by individual
particles will be highly variable because no two particles will be equally exposed
to the flow and because the flow at any point of the bed constantly changes with
the formation of high velocity streaks and low velocity streaks, see Kline et al
(1967) or Lian (1990).
A particle on the bed will also experience a lift force Fy, which is due to the
curvature of the stream lines in ineflow over the top of it. Similarly to the force on
a stone on a string which is Mi? /r ,the lift force is also proportional to the square
of the fluid velocity and inversely proportional to the orbit radius which in this
case is of the order of d. The force on the sediment particle with volume of the
order then is
2
Fy = pCa = pCivd (2.1.14)
Seepage or infiltration, i e a flow perpendicular to the sand surface may have
a stabilising or destabilising effect on the sand because the vertical fluid drag
changes the effective normal stress.

piss Bie EER S| ee eee sca

Figure 2.14: An _ outflow

velocity w corresponds to a lift
force of —x? g per unit volume,
where K et
is permeability.

AND SEDIMENT TRANSPORT 101

Chapter 2: Sediment mobility, bed-load and sheet-flow

This is the mechanism which causes the formation of quicksand. Ground

water flows through sandy materials are generally described in terms of Darcy’s
Law

we nx (te
FF ) (2.1.15)

where K is the permeability which has the dimension of velocity, or

1 op
( = —K| .P8& % (2.1.16)
w
—FP1 op 44
pg dz

The water velocity u = (u, w) is in this definition the equivalent clear water
velocity, i e, the flow rate per unit area. The actual velocities of the water in the
pores are somewhat greater.
For the situation in Figure 2.1.4, where the seepage rate corresponds to the
vertical, equivalent clear water velocity w, the total vertical pressure gradient in the
pore water must be

OD. yiSevens
retin dbaWw.
pondiy (2.1.17)

corresponding to a bouyancy force of p g (1+ a) per unit volume. Thus, in order

to lift a sediment particle with density s p a vertical outflow velocity (flow rate per
unit area) of magnitude (s - /)K is required.

2.2 SEDIMENT MOBILITY AND INCIPIENT MOTION

2.2.1 Introduction
The initiation of sediment motion under steady flows and under waves has
attracted considerable interest in the past because it is a philosophically appealing
concept. In practical terms however, it is a very difficult concept to deal with.
Firstly, because “initiation of motion” is difficult to define - is it when one in a
thousand grains moves or when one in a hundred moves? Secondly, because the

102 COASTAL BOTTOM BOUNDARY LAYERS

 2.2 Sediment mobility and incipient motion

complicating variables in a natural situation is very large. For example, the sand
bed is never left perfectly smooth from previous events. Relict bed forms will be
present and initiation of motion will occur sooner near the crest of these bedforms
due to local enhancement of the bed shear stresses.
In addition, biological activity will also complicate the micro-topography and
excretions from animals may tend to glue the sand particles together. Nevertheless,
we shall consider a few classical approaches to the description of incipient
sediment motion in the following.

2.2.2 The mobility number

A simple, yet useful, dimensionless measure of the fluid forces on a sediment
particle under waves is the mobility number w which will be defined in the
following.
For sand size particles (d~0.2 mm) under waves with typical semi-excursions,
A of the order 0.1m -2m, the Keulegan Carpenter number d/A is very small and
hence the drag force will tend to dominate over the pressure force, see Equation
(2.1.13), page 101.
Hence, the total disturbing force on a sand particle at the bed is
approximately proportional to the square of the velocity amplitude A@, and the
ratio between this disturbing force and the stabilising force due to gravity is
reasonably described by the mobility number

_ Ao)

2.2.3 The Shields parameter

A different measure of the balance between disturbing and stabilising forces
on sand grains at the bed was suggested by Shields (1936) in a study of the
incipient sediment motion in steady flow,
2
Ae t(OVGs +hSt 9netoi
mean lielvg dotwec(sst nad 22)
Accordingly, @ is known as the Shields parameter.
This parameter is particularly convenient to use in connection with steady
flow because there, the steady bed shear stress, (0) and hence the friction

AND SEDIMENT TRANSPORT 103

Chapter 2: Sediment mobility, bed-load and sheet-flow

velocity %* are quantities which are easily measured, t(0) = p g DJ, where D is
the flow depth and / is the hydraulic gradient.
In connection with wave motion, the Shields parameter (corresponding to
total stress) is generally defined in terms of the peak bed shear stress

q _ WfwAoy _ 1
=usip (ele dol a Dédala wale aa i ow
where fw is the wave friction factor, defined on page 23, and this is the notation
which will be used in the following. |

2.2.4 Skin friction

The total bed shear stress t may be seen as consisting of two contributions
namely, the form drag 1’’ and the skin friction 1’. The significance of each of
these for the sediment transport is quite different as described by Engelund &
Hansen (1972).
The form drag is generated by the difference in pressures between the
upstream and the downstream sides of bedforms, and it does not directly affect the
stability of individual surface sediment particles. The main disturbing influence
to the surface grains is generally considered to come from the skin friction tT’. The
corresponding skin friction Shields parameter,

,
= peDed
Tv’
228
is therefore frequently used to predict initiation of motion and the magnitude of
moving sediment concentrations.
Correspondingly, 1’ is often referred to as the effective stress in connection
with sediment transport.
A comprehensive discussion of the concept of effective stress in steady flows
is given by Engelund & Hansen (1972).
If the bed is flat, the form drag is absent so, from that point of view, 7 = T
and @’= 80.
There is, however, some indication that the total shear stresses on flat sand
beds under waves may not be totally effective with respect to transporting
sediment, see Section 2.4.4, page 121.

104 COASTAL BOTTOM BOUNDARY LAYERS

 2.2 Sediment mobility and incipient motion

2.2.5 The grain roughness Shields parameter.

Flat beds of loose sand under waves as well as under steady flow may offer
considerably more resistance to the flow than sand paper with the same grain size.
This is a consequence of the momentum transfer by moving sand from the
flow to the bed, see Section 3.6, page 145.
It is, however, difficult to estimate this momentum transfer and the amount of
related data is very limited. So a generally accepted method for calculating the skin
friction on a bed of highly mobile sand under waves is not yet available (an
empirical formula will be presented in Section 3.6.6, p 155 ff).
On the other hand, the mobility number y is not a totally adequate measure
of the sediment mobility because it neglects the dependence on the ratio d/A of
the force exerted by waves on sediment particles, see for example Figure 3.4.6,
page 141.
It was therefore suggested by Madsen & Grant (1976) that the sediment
mobility be estimated in terms of a grain roughness Shields parameter, and this
approach has since been quite popular.
Following Engelund & Hansen (1972) and Nielsen (1979), we shall adopt the
value 2.5ds0 for the grain roughness of a flat bed of sand with median size d50,
and correspondingly operate with a grain roughness Shields parameter 625
defined by

5 =
Yoened
fps p (Aw) 175 W (2.2.5)

where the special grain roughness friction factor, f 2.5 is based Swart’s (1974)
formula (1.2.22) and a roughness of 2.5d50
0.194
fos = exp [5.213 ee Sty AMA (2.2.6)

The relationship between 025 and 9 for both rippled and flat sand beds in
oscillatory flows is illustrated by the data of Carstens et al (1969) and of Lofquist
(1986) in Figure 2.2.1.
The data show that for rippled beds 9 is generally an order of magnitude
greater than 625 with no systematic trend between different grain sizes.
For flat beds, @ (= 6’) is also considerably larger than 025 , by a factor five
or so, when the activity level is high (625 2 0.3). For a flat bed at low activity
level, one would expect to find 8 = 0’ = 025.

AND SEDIMENT TRANSPORT 105

Chapter 2: Sediment mobility, bed-load and sheet-flow

10

0.1
0.01 0.1 1
Figure 2.2.1: Relationship between the "total Shields parameter" @ and the grain
roughness Shields parameter 62,5 for the presently available oscillatory flow data.
Legend bar : Lofquist 0.55mm sand, + : Lofquist 0.J8mm, * : Carstens et al
0.19mm, rectangle : Carstens et al 0.30mm, X : Carstens et al 0.59mm . All of the
above correspond to rippled beds, while the triangles correspond to flat beds,
Carstens et al 0.19mm and 0.30mm.

While the data of Carstens et al and of Lofquist consistently show high

energy dissipation rates over flat sand beds, these values do not immediately agree
with the available information on corresponding sediment transport rates.
Thus, while the available friction/energy-dissipation data indicate that the
total bed friction and the energy dissipation on a flat sand bed correspond to a
stress five times as large as 125, a different picture is presented by the sediment
transport data examined in Section 2.4, page 116. They indicate that the effective
stress for moving sediment over a flat bed under waves is of the order 125, rather
than 5%25. The writer knows of no satisfactory explanation for this at the present

106 COASTAL BOTTOM BOUNDARY LAYERS

 2.2 Sediment mobility and incipient motion

time. Wave friction factors and the corresponding hydraulic roughness of sand
beds under waves are discussed in detail in Section 3.6, page 145.

2.2.6 The critical Shields parameter and the Shields diagram

The critical Shields parameter Oc is the effective Shields parameter (6’)
at which sediment movement starts.
The value of @c is a function of the sediment size and density, of the fluid
density and viscosity, and of the flow structure. Typical O¢-values for sand in
water are of the order 0.05. For sand in air, they are somewhat lower, usually in
the range 0.01<®;-<0.02, see Allen (1982). In both air and water, 9 - becomes
much larger in the silt range of grain sizes (d<0.063mm).
To obtain a simple description of the behaviour of 9¢, Shields (1936) noted
that both the drag force Fp and the lift force Fy , on a bed sediment particle, are
proportional to ux , and to functions of the grain Reynolds number, ux d/v. He
therefore plotted observed values of 6c, against the the grain Reynolds number.
The resulting diagram is called the Shields diagram.

0.60
0.40

0.20

0.10

0.06
0.04

0.02
dv (s-1)gd /4V
0.01
0.4 1 2 4 10% 2820700840 100 200
Figure 2.2.2: The Madsen-Grant diagram for the initiation of sediment motion in
oscillatory flow and on oscillating trays. Data from Manohar (1955) and from
Carstens et al (1969).

AND SEDIMENT TRANSPORT 107

Chapter 2: Sediment mobility, bed-load and sheet-flow

A slightly different diagram was suggested by Madsen & Grant (1976).

Instead of the grain Reynolds number they used dN(s—1)gd/4v as the abscissa.
An example of such a diagram is shown in Figure 2.2.2.
For typical beach sand with d = 0.2mm and s = 2.65, the value of
dN(s-l)gd/4v is approximately //. Around this value the data show no
significant trend. Hence, the use of an all round value of

@¢ = 0.05 (2.2.7)

for the critical Shields parameter seems justified in most practical cases.
The effective shear stress Te which corresponds to O¢ [Tc = P (S—1) gd Oc]
is called the critical shear stress for the particular sediment.
The Shields criterion ®¢ = @c(u«d/v) is but one of many criteria for the
initiation of sediment motion. Many others have been suggested. For a
comprehensive review, see Hallermeier (1980).

2.2.7 Initiation of motion in combined wave-current flows.

It is hard enough to agree on a sharp criterion for the initiation of motion in
pure currents or pure wave motions, but when it comes to combined wave current
flows, it becomes even harder. The number of possible governing parameters
increases and the amount of experimental data is more limited, e g the laboratory
studies of Natarajan (1969) and Hammond & Collins (1979), and the field study of
Amos et al (1988).
Based on their observations, Amos et al recommended the "Shields type
criterion"

T+?
Glia 0.04 (2.2.8)

see Figure 3.5.1, page 144.

In this formula T is the time-averaged bed shear stress estimated by
t = ’%p 0.003 Wie” and too is the mean current velocity one metre above the
bed.
Amos et al used the formulae of Madsen & Grant (1976) to estimate the peak
wave-induced bed shear stress 7 , but very similar values are obtained from
Equations (1.2.18) and (1.2.22) with r = 2.5ds50.

108 COASTAL BOTTOM BOUNDARY LAYERS

 2.3 Steady bed-load and sheet-flow

2.2.8 The depth of closure

A related concept to the initiation of motion is the depth of closure on a beach
profile. The depth of closure is the depth beyond which sand level changes
between seasonal surveys become unmeasurable or insignificant.
The depth of closure is a prerequisite for the use of the Bruun rule (Bruun
1962) which provides a simple estimate of the shoreline recession in response to
sea level rise under a number of simplifying assumptions, see e g Bruun (1983,
1990).
Practical estimates of the depth of closure are of the order 3.5 times the
annual maximum significant wave height. For a review of the concept and related
formulae see Hallermeier (1981).

2.3 STEADY BED-LOAD AND SHEET-FLOW

2.3.1 Introduction
The following section summarises current experimental facts about steady
bed-load transport, mainly in the light of the theory of Bagnold (1956), and with
emphasis on concepts and formulae which are transferable to oscillatory flows and
combined wave current flows.
Most of the steady bed-load formulae are of the form ® = ®(6’). That is,
they state direct relationships between effective bed shear stresses and
dimensionless transport rates without considering underlying details such as, the
amount of moving sediment, and the typical speed with which that sediment is
moving.
It is necessary, however, to know about these details in order to adapt steady
flow models to unsteady flow situations. Sections 2.3.3 and 2.3.5 therefore attempt
to extract information about the amount of bed-load and about the typical speed
with which the bed-load moves in steady flows.

2.3.2 What is bed-load ?

The total load of moving sediment is generally seen as composed of three
parts: the wash-load, the suspended load and the bed-load.

AND SEDIMENT TRANSPORT 109

Chapter 2: Sediment mobility, bed-load and sheet-flow

The bed-load has been defined in different ways depending on the context. In
relation to measurements, it is often defined as that part of the total load which
travels below a certain level or (very pragmatically) as the part which gets caught
in bed-load traps. For modelling purposes however, it is more convenient to apply
the definition of Bagnold (1956).
Bagnold defines the bed-load as that part of the total load which is supported
by intergranular forces. The rest, i e, the suspended load and the wash load are
supported by fluid drag.
Obviously a given grain may well be supported partly by intergranular forces
and partly by fluid drag and hence contribute to both the suspended load and the
bed-load. This makes the bed-load practically unmeasureable in situations where
suspension is present as well, and this is of course of some concern. We shall
however stay with Bagnold’s definition in order to make use of the advantages it
gives with respect to rational discussion and modelling.

2.3.3 The amount of bed-load

Making use of Bagnold’s definition of bed-load, it is fairly easy to estimate
the weight of material which will be moved as bed-load under a certain effective
stress T’.
The bed-load must, due to its immersed weight, deliver an effective normal
stress
co

ce = p(s-l)g J calz) dz (2.3.1)

0

onto the top-most layer of the immobile bed; cg(z) is the volumetric concentration
of bed-load.
Assuming then, that the yield criterion for the top layer of immobile grains is

Tmax = Te + Oe tan Os (2.3.2)

see Figure 2.3.1, we see that the amount of bed-load which is in equilibrium with
tT’ is given by

TT -T%
cp(z) dz
0
Pp (s-1) g tangs Se

110 COASTAL BOTTOM BOUNDARY LAYERS

 2.3 Steady bed-load and sheet-flow

Topmost
non-moving
layer

zh pona T=T.+O0,tanyp,

Figure 2.3.1: For the equilibrium bed-load transport rate, the dispersive stress Oe
must satisfy the yield criterion t = T+6e¢ tangs.

Here it is convenient to introduce the maximum concentration Cmax , which is

the volumetric concentration of solid sediment in the immobile bed. In terms of
Cmax , the vertical scale of the bed-load distribution is then defined by
1 co

Lp = .. Jca(z) dz (2.3.4)
0

Introducing this expression for Lg into Equation (2.3.3) we see that the
vertical distribution scale measured in grain diameters is

Lp 0’ -@-
Te SE ee CRT YEREST OEM
d Cmax tan Qs ( :

Bagnold (1956) gave tangs = 0.63 as a typical value for fairly rounded
grains corresponding to a maximum concentration of the same value,i € Cmax =
0.63 [vol/vol], and he noted that the product cmax tan @s is fairly constant at
about 0.4 for different grain shapes.
Hence, as rules of thumb we have

Lg = 2.5 (0 -8c)d (2.3.6)

and

AND SEDIMENT TRANSPORT 111

Chapter 2: Sediment mobility, bed-load and sheet-flow

Cmax Lp = PhS) (6’ = 6c) d Cmax (2397)

Lg is the equivalent thickness-at-rest of the bed-load, and Cmaxlg is the

corresponding solids volume per unit area of the bed .

2.3.4 Steady bed-load and sheet-flow transport

We cannot, at present, claim that the bed-load transport rate

Qzp = Jcat) Us (z)dz = Cmax Lp Us (2.3.8)

oO

is well understood because neither the sediment velocity distribution wUs(z) nor
the concentration distribution cg(z) through “the bed-load layer” are well
understood. We can, however, still predict Qs empirically with reasonable
confidence for steady flow because it has been measured directly in a large number
of experiments. Most of the data support transport formulae of the form

@z (0’—6,) VO’. (2.3.9)

see Figure 2.3.2. The dimensionless flux ®g is defined by

@, =i Qs
unease @.3:10)

The plotted ®-values correspond to the total, measured sediment transport,

but the contribution from suspended load amounted only to a small fraction.
Engelund (1981) estimated at most 20%, even for Wilson’s high-stress data. The
general trend of the data is closely matched by the classical bed-load formula

®z = 8(0’-6,)!° (2.3.11)
of Meyer-Peter & Muller (1948), except that for high stress values (@ > 1), the
numerical constant 8 is too small and a value of about 72 seems more
appropriate. The slightly different formula

® = 12(6’ -6,) Ve”. (2.3.12)

also represents a reasonable approximation to the data.

112 COASTAL BOTTOM BOUNDARY LAYERS

 2.3 Steady bed-load and sheet-flow

100
2 SYMBOL MATERIAL d(mm) §S SOURCE i009
Gravel 28.65 268
Sand 5.20 2.68
Lignite Breeze 5.20 1.25 [ Meyer-Peter
Baryta 5.20 4.22
Sand 0.79 2.68 Gilbert
Sand 0.69 2.67 Wilson :
100 ae 100

10
12 (6'-0.05)

Meyer-—Peter

0.1

0.01 0.01
0.01 0.1 1 10
Figure 2.3.2: Total sediment transport rate under steady flows over flat beds. Due to
the fact that the sediment was fairly coarse (0.7mm), suspended load contributed little
to the transport rates for the shown "high stress data" (0 > 0.5).

AND SEDIMENT TRANSPORT 113

Chapter 2: Sediment mobility, bed-load and sheet-flow

2.3.5 Sediment velocity in steady bed-load and sheet-flow

The approximate proportionality of the bed-load (and sheet-flow) transport
rate to (@’—@,) V6’ can be briefly rationalised in the following way. The
transport formula

® = 12(6’-0,) Ye” 23:12)

can be rewritten in the form

Q = Cmax Lp Up = D5) (0’ — 0c) d Cmax 4.8ux (2.3.13)

where the expression (2.3.7), page 112, has been inserted for the amount of
bed-load per unit area Cmax Lg . This reveals that the typical sediment velocity
defined by

1
oo

Oz
CmaxLe Ny~Cmax 25( 66) d poo spy
6"— Oa pag3.1al4
w ~ 750

is of the order 4.8 times the friction velocity

Up = 4.8 ux (2.3.45)

This is an interesting experimental fact because it is incompatible with the

constitutive relationships (2.1.3) and (2.1.4) which are due to Bagnold (1956). That
is, Bagnold’s constitutive formulae, together with the expression (2.3.7) for the
amount of bed-load, lead to typical sediment velocities Ug which are
proportional to higher powers of ux,
This in turn leads to sediment transport formulae which increase too rapidly
with ux (or @’).
In order to obtain Ug ~ u« (where "~" is short for "proportional to"), from
the definition (2.3.14), for a layer of thickness proportional to Lg , the averaged
velocity gradient must be proportional to u«/Lz:

nla
eecaok es (2.3.16)
That is, for a given friction velocity us, the velocity gradients in the
bed-load layer are inversely proportional to the weight of bed-load. This weight
being proportional to Lg.

114 COASTAL BOTTOM BOUNDARY LAYERS

 2.3 Steady bed-load and sheet-flow

From the form of Equation (2.3.16), it can be seen that, if the constitutive
relationship for the granular flow is written in terms of an effective viscosity, Ve

2 dus
T/p = Uk = Ve<—>
- ail
this viscosity must be proportional to ux Lg

Ve ~ U«xLp (2:3.17)

As mentioned above the result (2.3.15) for the typical sediment velocity and
the corresponding result (2.3.16) for the averaged velocity gradient disagree with
Bagnold’s constitutive relationships from Section 2.1.2. Thus, Bagnold’s formula

/p = uk = 0.0135 nay? for B > 450 (2.3.18)

obviously corresponds to

dus 1
92
epee
dz
which is in conflict with Equation (2.3.16), and with Lg ~ 0’ it leads to

d
Qa ~ CmaxLle Ug = Cmax Lg (La <*>) ~ 9” (2.3.19)

A model of this type was developed and discussed in detail by Hanes &
Bowen (1985).
If Bagnold’s formula (2.1.4) for the hyper-viscous regime is used instead of
(2.3.18) above, the result for the transport rate is

Gite ior (2.3.20)

which is even further from the experimental trend of the data in Figure 2.3.2.
If data with fine sand at high effective stresses (0’2 1) are included in a
plot like Figure 2.3.2, they may indicate a more rapid increase of the total transport
rate than 0’). This is however, most likely due to the rapidly increasing rate of
suspended transport, and that is another matter.
The discussion above is only related to the constitutive relationships for the
bed-load layer.

AND SEDIMENT TRANSPORT 115

Chapter 2: Sediment mobility, bed-load and sheet-flow

2.4 BED-LOAD AND SHEET-FLOW UNDER WAVES

2.4.1 Introduction
The following section describes the processes of bed-load and sheet-flow
sediment transport under waves. Or in other words, the quasi-steady processes of
sediment transport which occur under waves over effectively flat sand beds.
Under sheet-flow conditions, the sediment transport rate is not entirely due to
bed-load transport in Bagnold’s sense, see Section 2.3.2. However, the assumption
of quasi-steadiness may be applied to the total transport rate under certain
conditions. That is, when the thickness of the sediment transporting layer is small
compared to the typical distance woT settled by a suspended sediment particle
during one wave period.
Such conditions can be observed at moderate flow intensities (025 S 1)
over artificially flattened beds before vortex ripples form, and for high flow
intensities where vortex ripples are naturally absent, see Section 3.4, page 135 ff.
The approach taken is essentially to adapt the steady flow sediment transport
models from the previous section to oscillatory flows and to combined
wave-current flows through quasi-steady considerations.
The quasi-steady approach is used because, it seems fairly reasonable to
consider the processes of bed-load and sheet-flow sediment transport as quasi
steady. There are however two main problems with this transfer of technology
from steady flows to oscillatory and combined flows.
Firstly, steady flow transport formulae are generally of the form ® = ®(6’),
i e, they assume knowledge of the effective (skin friction) stress 6’, and our
knowledge about the effective Shields parameter 9’ for waves over sand beds is
sparse. See Figure 2.2.1, page 106, and Section 3.6, page 145 ff.
The most commonly attempted way around this problem is to replace 6’ by
the more "calculable" 625 and we shall see that this approach is quite justifiable in
the sense that it leads to good predictions.
It should be kept in mind, however, that 625 and 6’ are conceptually
different as discussed in Sections 2.2.4 and 2.2.5.
The second main problem with transferring steady flow sediment transport
formulae to coastal conditions, is to deal with those effects of unsteadiness which
cannot be ignored.
This problem is addressed in detail in Section 2.4.4, page 121, and the
conclusion which is reached is in essence as follows.
It may well be possible to model bed-load transport over flat beds under

116 COASTAL BOTTOM BOUNDARY LAYERS

 2.4 Bed-load & Sheet-flow under waves

waves by quasi-steady expressions of the form Q(t) = Q(t[t]), but it is not, in

general, possible to use expressions of the form Q(t) = Q(u.t]).
The reason is that the bed shear stress which occurs at a certain time is not
just a function of the instantaneous free stream velocity, but depends strongly on
whether this wu. has been achieved through rapid or very gradual acceleration.

2.4.2 The amount of bed-load under waves

The amount of bed-load, quantified by the maximum value Lgmax of the
equivalent thickness at rest Lg , was measured by Sawamoto & Yamashita (1986)
in an oscillating water tunnel. Their data are plotted in Figure 2.4.1 versus the
grain roughness Shields parameter 025s.
The line corresponds to the steady flow formula (2.3.6) with 6’ replaced by
02:5:

Lgemax = 2.5 (@25-—0.05)d (2.4.1)

The remarkably close agreement with the directly adapted steady flow
formula is unexpected because the data in Figure 2.2.1, page 106, indicate that the
total Shields parameter on a flat sand bed is generally much greater than 625. If
the data in Figure 2.2.1 are to be trusted, in the sense that measured energy
dissipation rates are true indications of the effective sediment transporting stress,
this effective stress should be of the order 5625 when @25 = 1. Hence the
expected magnitude of Lgmax , based on Equation (2.3.7) and assuming "perfect
quasi-steadiness", would be Lgmax =~ 2.50’d = 2.5(5625)d = 12.5 5d.
That is however, about five times more than what was observed.
The writer can offer no clear explanation for this at present, but the observed
amounts can of course still be reasonably predicted by Equation (2.4.1), or maybe,
a fine-tuned version of it, using a power of about 3/4 rather than / as suggested
by Sawamoto and Yamashita.

2.4.3 Bed-load and sheet-flow under sine waves

We saw in the previous section that the amount of bed-load moving under a
sine wave can be predicted by the directly adapted steady flow formula Equation
(2.4.1). That is, by a steady flow formula with 6’ replaced b 625 and, possibly,

AND SEDIMENT TRANSPORT 117

Chapter 2: Sediment mobility, bed-load and sheet-flow

0.1 0.2 04 06 1.0 2 4 6

Figure 2.4.1: Measured values of the peak amount of bedload on a flat bed under
"sine waves" in an oscillating water tunnel. The data are from Sawamoto &
Yamashita (1986) and include relative sediment densities in the range /.58<s<2.65,
and grain sizes in the range 0.2mm<d<1.6mm. The curve corresponds to Equation
(2.4.1).

with slightly adjusted numerical coefficients. Similarly, we shall see in the

following, that the average sediment transport rate under half a sine wave can be
estimated by a directly adapted version of Equation (2.3.11) or Equation (2.3.12).
Due to its symmetry, a complete sine wave must of course produce zero net
sediment transport, but the average transport rate Q7/2 during half a sine wave is
of some theoretical interest, The corresponding, dimensionless ®74 is plotted
against 025 in Figure 2.4.2.

118 COASTAL BOTTOM BOUNDARY LAYERS

 2.4 Bed-load & Sheet-flow under waves

Pr).

10(8.; — 0.05)1.5

0.1 0.2 0.4 06 41.0 2 4 6 10

Figure 2.4.2: Averaged, dimensionless sediment transport rates under half sine waves
as function of 92.5. Data from Sleath (1978), Horikawa et al (1982), Sawamoto &
Yamashita (1986) and King (1991). Sediment densities were in the range
1.14<s<2.66, grain sizes in the range 0./35mm<d<4.24mm and periods in the range
0.5s<T<J2s. The crosses correspond to d<0.25mm, the squares to d>0.25mm.

AND SEDIMENT TRANSPORT 119

Chapter 2: Sediment mobility, bed-load and sheet-flow

The curve corresponds to

@y, = 3 (025-0.05)!> (2.4.2)

which will be rationalised below as a direct adaptation of the Meyer-Peter formula
(2.3.11) from steady flow.
At high flow intensities (6252 1), the data points for the fine sand lie
significantly above the curve, but this is probably due to the increasing importance
of suspended transport for these sediments, and agrees with the trend of the steady
flow data in Figure 2.3.2, page 113.

Sleath (1978) studied the instantaneous bed-load transport rates under sine
waves as well as ®7% . He found that ®(t) varied approximately as sin*(ot + sp)
with a phase shift @g of the order ten to twenty degrees ahead of the free stream
velocity, ie

O(t) = 5O15sin‘(cot + @g) (2.4.3)

This time-dependence may be rationalised in the following way. Assume that
the instantaneous, effective bed shear stress varied approximately as a sine
function with amplitude @25 and a phase shift of @ relative to the free stream
velocity:

O’(t) = 25 sin(@t + @r) (2.4.4)

Assume further that the instantaneous sediment transport rate is given by the
adapted Meyer-Peter formula

M(t) _= J8(@(t)-0.05)!° for @’>0.05 2.4.5

: for 6’ <0.05 —

The instantaneous bed-load transport rate will then have essentially the shape of
Sleath’s sin‘-expression (2.4.3), and since sin*t = 38 it leads to the expression
(2.4.2) for the averaged transport rate.
We note that since Equation (2.4.5) includes no phase lag between @7(t) and
@(t), we get sz = Qr. In reality however, it would be reasonable to expect some
time lag of @(t) behind 6’(¢) and hence, in general @g < @r.

120 COASTAL BOTTOM BOUNDARY LAYERS

 2.4 Bed-load & Sheet-flow under waves

2.4.4 Bed-load and sheet-flow under skew waves.

In order to model bed-load transport under irregular waves of arbitrary shapes
(arbitrary u(t) ) it seems reasonable, at this stage, to make use of the assumption
of quasi steadiness in some form.
The simplest quasi-steady transport models may be written in the form

Oz(t) = F(u.—[t] ) (2.4.6)

expressing that the bed-load/sheet-flow transport rate is at all times determined by

the instantaneous velocity above the boundary layer.
Several models of this type, have been suggested, e g by Bailard (1981) and
by Ribberink & al Salem (1990), and they lead to transport formulae of the type

Op = Const: \Uco(t)!" Uoo(t) (2.4.7)

-1 0 1 2 3 4 2, 6 ei

Figure 2.4.3: Sawtooth waves with zero mean flow and opposite asymmetry. These
waves would both generate zero net bed-load transport according to formulae of the
type (2.4.7).

AND SEDIMENT TRANSPORT 121

Chapter 2: Sediment mobility, bed-load and sheet-flow

While such formulae have the advantage of simplicity, they have the
disadvantage of not being able to model the net sediment transport rate which
results from certain types of wave asymmetry.
That is, asymmetry which are described essentially by the velocity moments
such as that of a 2nd order Stokes wave can be represented, while "sawtooth
asymmetry" is overlooked. Thus, the two "opposite" sawtooth waves in Figure
2.4.3 would lead to exactly the same transport rate according to formulae of the
type (2.4.7).
This is important to note, because several experiments, e g those of King
(1991), clearly show that the sawtooth wave with the steep front tends to generate
shoreward bed-load transport while the one with the steep rear generates seaward
bed-load transport. See Figure 2.4.4.
King’s observations show that sawtooth skewness matters, and hence that the
approach which leads to bed-load transport formulae of the type (2.4.7) is too
simplistic. The "next one up" from Equation (2.4.6) as a starting point is to assume
that the instantaneous transport rate is determined by the instantaneous effective
bed shear stress, ie

Qz(t) = F(t’[o,t]) (2.4.8)

Figure 2.4.4: Measured net

bed-load transport for half
sawtooth waves with
0.1 Opposite steepness. * : Steep
front, rectangle : steep rear.
Data from King (1991)
including quartz sand_ with
diameters 0.135mm, 0.44mm
and 1.Jmm, and "wave"
0.01 7 i periods in the range

0.01 0.1 { 10
3s<T<l1ls.

122 COASTAL BOTTOM BOUNDARY LAYERS

 2.4 Bed-load & Sheet-flow under waves

where the findings in Sections 2.4.2 and 2.4.3, page 117 ff, seem to indicate that
the effective stress t’ can be calculated as the grain roughness stress 125.
Basing the transport rate on 1’(0,f) instead of u.(f) gives the potential of
capturing the effect of sawtooth asymmetry because the bed shear stress under a
‘sawtooth wave is likely to have the shape shown in Figure 2.4.5, with higher
(absolute) stress values under the rapidly accelerated half cycle.

free stream velocity

Figure 2.4.5: Bed shear stress under a sawtooth wave for a boundary layer which is
laminar or has a constant (independent of z and ft) eddy viscosity. Although the
shoreward and seaward peak velocities are of equal magnitude, the shoreward peak
shear stress is considerably larger than its seaward counterpart. Qualitatively, the
reason is that the boundary layer has had less time to grow during the rapid shoreward
acceleration.

The important question is then: How can we estimate instantaneous, effective

bed shear stresses under waves of arbitrary shape? - A hard question since we
know very little about instantaneous stresses on movable beds at all. What has
been measured are total stresses under sine waves, and the discussion of sediment
transport rates in Section 2.4.3 indicates that the effective stresses may be
considerably smaller than the total, even on flat beds.

AND SEDIMENT TRANSPORT 123

Chapter 2: Sediment mobility, bed-load and sheet-flow

The influence of the steady flow component on the bed shear stress is
difficult to estimate from a time series of u(t) at a single level, and if u is
relatively large as in some rip currents or in a strong undertow, this calls for special
consideration, see e g Section 1.4.5 and Example 1.5.2.
If the steady flow component is weak however, in the sense that its influence
on the bed shear stress is small, it seems reasonable to try and derive t(0,t) from
Uco(t) by means of a simple transfer function based on our knowledge from simple
harmonic boundary layer flows.
For an oscillatory boundary layer with constant eddy viscosity (independent
of z and ¢ ), the bed shear stress corresponding to uo(t)=AM@e™
iot is given by

t(0,t) = pYav; Awelrt™ (2.4.9)

see Section 1.2.4, or in terms of the friction factor

10.) = pf Aa) eorr™ (2.4.10)

For the sake of developing a transfer function (digital filter) we rewrite this
equation in terms of the free stream velocity and its derivative given as real-valued
functions

t(0,)) = 2PpfwA (cos™ @ Un(t) + sin a=) (2.4.11)

The analogous expression for an arbitrary (forward) phase shift 7 is

t(0,f) = 5P fwA (CcosQt @ Uoo(t) + sings ae) (2.4.12)

The phase shift @r is likely to depend on the Reynolds number A’w/v and
on the relative bed roughness _ r/A.
At present, it seems reasonable to approximate the instantaneous, effective
bed shear stress under waves of arbitrary shape (arbitrary uco[t]) by an adaptation
of Equation (2.4.12), for example

, ; d Uco
t(0,t) = eP f2.5 Arms ( COSOt Wp Uoo(t) + sings (2.4.13)
dt
Here, the friction factor f25 should be based on the representative semi-excursion
Anns = N21Urms/Wp , where Wp is the spectral peak angular frequency (2 7% Sp):
The bed shear stress time series corresponding to a given Ueo-record may then

124 COASTAL BOTTOM BOUNDARY LAYERS

 2.4 Bed-load & Sheet-flow under waves

be constructed, either by using forward and inverse Fourier transformation, or

alternatively, by using a simple digital filter in the time domain.
For the spectrum transformation approach, one might suggest the simple
frequency response function

1 5

F(@) = > P f2.5 Arms Wp e'?

based on the expression (2.4.13), in analogy with Equation (1.2.17), page 23.
As a simple digital filter to generate the time series 1’(0,tn), on a flat bed,
from Ueo(tn) one might correspondingly suggest

T’(0,tn) = +P S25 Arms ( COS@t Wp Uoo(tn) + singr

Uco(tn+1) 25— Uoo(tn—-1) )
t

(2.4.14)
The corresponding time series for the instantaneous effective Shields parameter is
then given by
1
= fr5 Arms =
’ roe? ; Uoo(tn+1) Uco(tn—-1)
O'(tn) = eed:
ah ( COS@q Wp Uoo(tn) + singr Sate ity 5;age ge )

(2.4.15)
As indicated by the review of bed-load transport under sine waves in Section
2.4.3, page 117 ff, a respectable estimate of the instantaneous bed-load transport
rate is given by the Meyer-Peter formula with the effective Shields parameter
based on 62.5, see Equations (2.4.4) and (2.4.5), page 120.
In order to incorporate information about the changing direction of the bed
shear stress, Equation (2.4.5) must now be augmented to

O(t) = 8(e'(t) - eos rary for 1@’(t)| > 0.05 (2.4.16)

0 for |6’(HI < 0.05

The factor 8, which corresponds to the Meyer-Peter formula for bed-load

transport, may not be appropriate for fine sand at high flow intensities as indicated
by the data in Figures 2.3.2 and 2.4.2. For sand with dso <0.25mm at ®252 1.0
a factor /2 might be more appropriate.
For half sine waves, Equation (2.4.16) is identical to (2.4.5) which is in close
agreement with instantaneous bed-load transport rates observed by Sleath (1978).

AND SEDIMENT TRANSPORT 125

Chapter 2: Sediment mobility, bed-load and sheet-flow

For the half cycle average ©74, Equation (2.4.16) leads to results very similar
to Equation (2.4.2) which corresponds to the curve in Figure 2.4.2 with minor
deviations depending on the relative magnitude @25/@c of the peak effective
stress.
The expressions (2.4.15) and (2.4.16) are easily applied to any u..-record.
However, since the derivation of the model involves several assumptions and
simplifications, which cannot be checked completely at present, the model should
be calibrated as more data become available.
Application of the sediment transport model consisting, of Equations (2.4.15)
and (2.4.16) to skew waves over flat beds is illustrated in the following example.

Example 2.4.1: Application to King’s data.

King (1991) performed a comprehensive set of experiments on bed-load
transport under half sine waves and under half sawtooth waves with some pairs
mirror imaged. King’s sawtooth waves had the shape of the first halves of the two
curves in Figure 2.4.3.
For the half sawtooth waves he found typically that of a pair of mirror
images, the one with steep front would transport of the order /.7 times more
sediment as the one with the steep rear.
The reason for this was given qualitatively in connection with Figure 2.4.5. A
quantitative example is given in the following by applying the model outlined
above to a pair of King’s sawtooth waves.
The example corresponds to the average of runs 473-485 and 478-520, which
had the following characteristics: (T, urms, d50, ®™%) = (5s, 0.89m/s, 0.44mm, 1.09)
for the steep-front-case, and (5s, 0.88m/s, 0.44mm, 0.60) for the steep-rear-case.
The two cases were modelled as one complete (periodic) wave with the shape
shown in Figure 2.4.6 corresponding to both halves of the wave having the
rms-velocity 0.89m/s.
The calculation of the ingredients for the formula (2.4.15) and hence (2.4.16)
goes as follows.
The peak period is 5s corresponding to the main harmonic component, and
hence ®p = 2n/Tp = 1.25757.
From the rms-velocity and @p one finds Arms = V2 urms/ Wp = 0.706m ,
and with this inserted for A in Equation (2.2.6), page 105, we find fos = 0.0112.
The instantaneous dimensionless transport rates were then calculated using
Equations (2.4.15) and (2.4.16) with four different values of the phase shift @r :
20°, 30°,40°, and 45°. The results are shown in Figure 2.4.6.

126 COASTAL BOTTOM BOUNDARY LAYERS

 2.4 Bed-load & Sheet-flow under waves

Figure 2.4.6: Free stream velocity and calculated, dimensionless sediment transport
rates corresponding to the conditions in runs 473-485 and 478-520 of King (1991).
Legend: &: free stream velocity, * : ®(t) with gr = 20°, rectangle: ®(t) with
Qt = 30°, x: (ft) with or = 40°, -: O(t) withgr = 45°,

The calculations give the correct general magnitudes for ®74. Thus, for the
steep-front-halves (0<t<2.5s) the calculations give 7.09, 7.13, 1.18, and 1.19 in
order of increasing ©. The corresponding numbers for the last half of the waves
(2.5s<t<5s) are -1.04, -0.985, -0.855, -0.764. King’s measured values are /.09
and -0.60 respectively.
Inspecting the ratios between corresponding values reveals that the right
order of transport skewness is achieved by the model only when the phase shift @r
is set the maximum value of 45 °.
For a phase shift of 20° which corresponds roughly to the observations of

AND SEDIMENT TRANSPORT 127

Chapter 2: Sediment mobility, bed-load and sheet-flow

Sleath (1978) the predicted transport rate ratio is only /.05 as compared to the
measured value of /.7.
A complete explanation for this is not possible with the available
information, but it may be that the bed-load transport actually lags a bit behind the
bed shear stress, so that when Sleath observed a bed-load phase shift of @g = 20° ,
the bed shear stress might have been a bit more ahead of u.o(f) .

128 COASTAL BOTTOM BOUNDARY LAYERS

 CHAPTER 3

BEDFORMS AND HYDRAULIC

ROUGHNESS

3.1 INTRODUCTION
The sea bed is very rarely flat. On the contrary, it tends to be covered by
sedimentary structures with a large range of sizes and of many different shapes,
and these structures: bars; dunes; ripples and animal mounds interact with the flow
in different ways.
The larger sedimentary structures such as bars modify the main flow pattern,
ie they make the incoming waves refract, diffract or break, and they reflect part of
the wave energy back into deep water. They may also determine the positions of
rip currents.
The small scale structures, such as ripples, have no immediate impact on the
main flow patterns, but they strongly influence the boundary layer structure and
the turbulence intensity near the bed. Hence, they have great influence on the
sediment transport.
The interaction between the large scale topography and the main flow will
not be considered here, but the dynamics and geometry of the bedforms (mainly
vortex ripples) will be discussed in detail together with their influence on the
boundary layer structure. Finally, an attempt is made to quantify the influence of a
given bed topography on the flow by a single linear measure, namely the
equivalent Nikuradse roughness.

3.2 COASTAL BEDFORM REGIMES

The type of bedforms which prevail in a certain area at a certain time depends
mainly on the flow strength at the time.

AND SEDIMENT TRANSPORT 129

Chapter 3: Bedforms and hydraulic roughness

If the flow is too weak to cause appreciable sediment motion (62.5 $ 0.05),
the bed topography will be dominated by relict bedforms from previous more
vigorous events and if no such events have occurred recently, the topography will
be dominated by bioturbation.
Under flows of intermediate strength (0.05 S @25S 1.0 ) the bed will be
active and will be covered with bedforms which are more or less in equilibrium
with the flow conditions. However, the shape of these bedforms will depend on
the detailed nature of the flow. If the flow is purely oscillatory and almost
symmetrical, the bed will be covered by regular, long crested vortex ripples, which
are so named because a vortex is formed twice every wave period in the lee of their
crests.

Shore
break

Reformed Swash
| Breaker zone waves

No wave induced | Wave induced

sediment motion |sediment motion

Figure 3.2.1; Example of bedform distribution on an accreting, barred beach. Note

that the situation may be very different under storm conditions and that special types
of bedforms exist in rip channels.

130 COASTAL BOTTOM BOUNDARY LAYERS

 3.3 Bedform dynamics

The regular, symmetrical flows required for the formation of vortex ripples
occur in nature seaward of the breaker zone and in bar troughs where the waves
have reformed after breaking on the bar, see Figure 3.2.1.
Vortex ripples are rarely seen in the breaker zone. Here, the bed tends to be
either flat or covered by megaripples. Megaripples or lunate megaripples (Clifton,
1976), are irregular features with typical lengths of the order one to two metres and
heights of the order ten to twenty centimetres, and with rounded crests. There is no
regular, rhythmic vortex shedding associated with megaripples, but occasionally a
large plume of turbulent, sediment laden water rises from a certain spot due to
complicated instabilities. These sand plumes can be of the order one metre high.
Vortex ripples are sometimes found superimposed on megaripples (Southard
et al 1990).
Under very vigorous flows (825 2 1.0) vortex ripples cannot exist even under
perfectly regular, oscillatory flows. However, megaripples have been found to
exist under such flow conditions (82.5 up to about 2.4) both in large wave flumes
and in oscillating water tunnels, see e g Ribberink & Al-Salem (1989).

3.3 BEDFORM DYNAMICS

3.3.1 The continuity principle

The relationship between gradients of the sediment transport rate Q and
changes to the bed level zp is derived by expressing the conservation of sediment
volume, see Figure 3.3.1.

Figure 3.3.1: If there is

| more sediment entering
than there is leaving a
certain control volume, the .
bed level must go up and
vice versa.

AND SEDIMENT TRANSPORT 131

Chapter 3: Bedforms and hydraulic roughness

We quantify the sediment transport rate Q as the volume flux et sediment

per unit width of the channel and the dimension of Q is therefore L*/T, and the
units usually mls.
The volume of sediment which enters the control volume in Figure 3.3.1 per
unit time is therefore Q(x) per unit width and the volume that leaves is

Q(x) + a dx, and hence, the accumulated volume of solid sediment is — < dx

per unit time.

A bed level change 5zp corresponds to a sediment volume of n 5zp per unit
area, where n is the volume of solid grain material in a unit volume of the bed; 7
is usually about 0.7.
Hence in order to conserve the volume of sediment we must have

dp
nh _ _aQ
.e (3.3.1)

Note that this equation is only strictly valid when sediment storage in the
water column is insignificant. See Equation 5.3.1, page 223.

3.3.2 Bedforms migrating with constant shape

The continuity principle can be used very easily to derive the sediment
transport rate Q(x, t) from the shape and speed of bedforms if these are known to
be moving without change of shape. The argument is as follows.
Consider bedforms which move with constant shape at speed c so that the
local sand level can be described by

Zp(x,t) = f(x —ct)

Inserting this on the left hand side of the continuity equation, we get

ak f py ’(x -ct)
n (-c) aie +

which we integrate with respect to x and get

Ox, = Qo+ ncef(x-ct) = Qo + nc zH(x,t) (3.3.2)

where the constant of integration Qo represents the sediment transport rate

132 COASTAL BOTTOM BOUNDARY LAYERS

 3.3 Bedform dynamics

through cross sections where zp(x,t) = 0. The result (3.3.2) shows that if the
bedforms move downstream (c > 0) with constant shape, the sediment transport
rate must be maximum over the crest and minimum over the trough, ie Q varies
in step with zp, and the difference in transport rates between the crest and trough
sections is

Qmax- Qmin = nc (Zcrest - Ztrough ). (3.3.3)

For bedforms which move upstream (e g antidunes) with constant shape,

Equation (3.3.2) says correspondingly that the maximum transport rate occurs over
the trough, but Equation (3.3.3) is still valid (c<0).

3.3.3 Migration and growth of sinusoidal bedforms

Consider sinusoidal bedforms with length A and height 2A, which are
migrating with speed c, so that the local, instantaneous sand level can be given by

zpo(x,) = Acos( au —ct)) = Acosk(x -ct) (3.3.4)

or by introducing the complex exponential e'7 = cosz + isinz and only

attaching physical meaning to the real part

zp(x,t) = A elk — ct)

Figure 3.3.2: If the peak in the sediment transport rate is shifted dx away from the
crest, the bedforms will be growing or diminishing at a rate determined by 6x/A.

AND SEDIMENT TRANSPORT 133

Chapter 3: Bedforms and hydraulic roughness

Assuming then, that the sediment flux Q varies in a similar way to zp ie

Ole a
OG = 0.
where the shift 5; is the downstream distance between the bedform crest and the
point of maximum sediment flux, see Figure 3.3.2.
Under these conditions, the continuity Equation (3.3.1) gives

nAik(-c)e#*®—) = -~Qy ik gre che)

where we can cancel the common factor — ike *%—) and solve for the speed of
migration to get

Cos Fe thts = 2a [eos kdx — isin kdx)] (3.3.5)

This expression is most easily ee when 6x=0 or 5,=A/2 so that

c is real-valued and equal to Qj/nA and to -Qj/nA respectively. That is: the
bedforms move downstream with speed Q7/nA when Q peaks over the crest
and, they move upstream with the same speed if Q peaks over the trough.
We can however also give a physical interpretation of the complex c-values
which result when 6, + (0, A/2). If we insert the expression (3.3.5) for the speed
of migration c into z(x%,f) = A eis —ct) we get

zp = A exp fe Qi sinksx +] exp [ik (x— Or costs yy

zp = AekliCht pik(x-Richt) (3.3.6)

where R{c} and I{c} denote respectively the real part and the imaginary part of
c. The last expression shows that when c becomes complex, the imaginary part
measures the rate of exponential growth, while the real part measures the speed of
migration.
The model above, which is a brief synthesis of the models of Kennedy (1963)
and Engelund (1970), gives a simple mathematical framework which could
possibly be used for predicting the geometry of bedforms via stability analysis.
Unfortunately however, the model does not apply to vortex ripples because their
length is known to grow with their height while the model above assumes a fixed
wave length during the growth process.

134 COASTAL BOTTOM BOUNDARY LAYERS

 Section 3.4: Vortex ripples

Secondly, it is hard to predict the shift 5, between the bedform crest and the
point of maximum transport.
For a comprehensive discussion of the formation and growth of vortex ripples
see Lofquist (1978) and Sleath (1984).

3.4 VORTEX RIPPLES

3.4.1 Introduction
Vortex ripples are of special interest for coastal sediment transport studies
because their influence on the boundary layer structure and the sediment transport
mechanisms is very strong. That is, over vortex ripples, the suspended sediment
distribution will scale on the ripple height, while for other bedforms like
megaripples and bars, the suspension distribution will scale on the flat bed
boundary layer thickness which is much smaller than the height of those bedforms.
The shape and size of vortex ripples was first studied in detail by Bagnold
(1946), who also described the flow and the sediment transport mode above them.
Inman (1957) investigated their natural occurrence in a comprehensive field study,
and their development from a flat bed and their adaptation to new flow conditions
have been studied comprehensively by Sleath (1984) and Lofquist (1978)
respectively. All of these sources provide excellent illustrations of ripple shapes
and associated flow patterns. The occurrence of different overall ripple patterns, i e
neatly two-dimensional versus confused three-dimensional has been discussed by
Carstens et al (1969), Nielsen (1979) and Sleath (1984).

3.4.2 Ripple length

Vortex ripples are unique to the wave environment, and their scaling is
closely tied to the wave motion. With respect to sedimentary structures the most
important difference between waves and unidirectional flows is that wave flows
have a well defined horizontal scale namely the wave-induced water particle
excursion 2A, see Figure 3.4.1.
It turns out that, under an important range of flow conditions, the ripple
length A is a constant fraction of 2A

AND SEDIMENT TRANSPORT 135

Chapter 3: Bedforms and hydraulic roughness

Figure 3.4.1; The size of vortex ripples is closely linked to the orbit length 2A of
the wave-induced fluid motion near the bed.

XK = 133A for y< 20 (3.4.1)

see Figure 3.4.2.

Under more vigorous flow conditions, the relative ripple length 2/A tends
to be smaller than /.33, but the details of the mechanisms which determine the
ripple length in this regime are not well understood. Nielsen (1981) suggested the
following simple formula

ze a3) = 0345 Wo for 2<y<230 (3.4.2)

which describes the behaviour of 4/A_ reasonably well for a large range of wave
periods, grain sizes and sediment densities as long as the waves are regular.
The validity of Equation (3.4.2) has recently been questioned (Ribberink &
Al-Salem, 1989) because bedforms with lengths of the order 2A have been
observed in tunnels and large wave flumes at very high flow intensities
(y = 1000).
However, those bedforms are not vortex ripples, they tend to have rounded
crests like the megaripples which are sometimes found in the breaker zone in the
field (Clifton 1976 and Figure 3.2.1, page 130). They do not shed vortices
regularly like voriex ripples, and they are in fact some times superimposed with
vortex ripples with lengths of the order predicted by Equation (3.4.2).
Wave irregularity tends to result in smaller ripples in the sense that A/A is
smaller than estimated by Equation (3.4.2) when A_ is based on for example,
significant wave height or the rms wave height.

136 COASTAL BOTTOM BOUNDARY LAYERS

 Section 3.4: Vortex ripples

2.0

1.0

0.40

0.20

0.10F get DATA

AINMAN (1957) ,Hg
+ DINGLER (1974) ,Hg=V2 Hime
0.04F0 Nn " » Single Wave
© MILLER & KOMAR, d=0.165mm Ve
eat "d= 0.287mm§ A~V2 Hrs

2 4 10 20 40 100 200 400 1000

Figure 3.4.2: Field data of relative ripple length versus mobility number based on
significant wave heights for field data. The curve corresponds to Equation (3.4.3) .

Based on the field data of Inman (1957), Dingler (1974) and Miller & Komar
(1980), Nielsen (1981) suggested

y
ep |£23-=.0.37 In 8u fae
>> 1000 + 0.75 In’y
for ripples under field conditions, i e, under irregular waves.

AND SEDIMENT TRANSPORT 137

Chapter 3: Bedforms and hydraulic roughness

3.4.3 Ripple steepness

The height to length ratio 1/A of vortex ripples is limited by the angle of
repose @ of the bed sediment when the flow is not too vigorous
(0.05 < 02.5 < 0.2).
That is, the maximum steepness occuring along the ripple profile is
approximately equal to tang. If the ripples were of (equilateral) triangular shape,
this. criterion corresponds to a height to length ratio of O.5tang. A parabolic
profile with maximum, local steepness tang has a height to length ratio of
0.25tang, see Figure 3.4.3.

Parabola

Figure 3.4.3: If the maximum slope on the ripple profile is assumed equal to tan 9,
the height to length ratio must for geometric reasons be O.Stan @ for a triangular
ripple, and 0.25 tan @ for a parabolic ripple.

It turns out that vortex ripples have a maximum height to length ratio (n/A)
which falls within the range corresponding to these two idealised geometries, see
Figure 3.4.4. The data indicate a maximum steepness, at vanishing flow speed, of

When the grain roughness Shields Parameter 625 exceeds 0.2, the flattening
effect of the contracted flow over the crest tends to increasingly overpower the
constructive effect of the lee vortices which build up the crests by scooping sand
towards them.
As a result, n/A becomes a decreasing function of @25, see Figure 3.4.5,
and eventually the ripples are flattened completely. This happens at a 025-value of
approximately /.0.

138 COASTAL BOTTOM BOUNDARY LAYERS

 Section 3.4: Vortex ripples

0.50

0.40

0.30

0.20

0.15 LEGEND d(mm) s tang

X HORMKAWA & WATANABE 0.54 127/0'32
@ CARSTENS et AL 0.585 2.65 0.56
O NIELSEN (1979) 0.082 2.65 0.59
O " 7] 0.55 2.65 0.63

0.10
0.05 0.10 0.15 0.20 0.30

Figure 3.4.4: The maximum value of n/A for a given bed material lies in the range
0.25 tang <(N/A)max < 0.5 tan , where the two limiting values correspond to an
equilateral triangle and to a parabola respectively, both with a maximum local
steepness of tang, see Figure 3.4.3.

The trend of the laboratory data (regular waves) shown in Figure 3.4.5 is
reasonably described by

: ro Tee 246 (3.4.5)

AND SEDIMENT TRANSPORT 139

Chapter 3: Bedforms and hydraulic roughness

0.40

0.20

0.10

0.04
LEGEND T(sec) d(mm) s
+ DHI TUNNEL 7.7-13.9 0.082 2.65
on " 4.4-9.0 0.2+0.135 2.65
0.02 |4 CARSTENS et AL 3.55 0.19 2.65
A " " 3.55 0.30 2.65 | FLAT BED |
O YALIN & RUSSELL 1.82 0.48 1.19 | +0+++4 |
V KENNEDY & FALCON 1.8-2.8 1.00 1.035 | |
O NIELSEN (1979) 1.70 0.082 2.65 |

0.04 0.10 0.20 0.40 1.0 2.0

Figure 3.4.5: Under gentle flow conditions, the ripple steepness N/A is only limited
by the angle of repose, but for 62.52 0.2 it becomes a decreasing function of 02.5.

For field conditions (irregular waves), the ripple steepness tends to be smaller
than for regular laboratory waves. On the basis of the field observations of Inman
(1957) and of Dingler (1974), Nielsen (1981) suggested the formula

~ = 0,342 — 0.344V025 (3.4.6)

for irregular waves, with 625 based on the significant wave height.
It would be convenient if the concept of the nominal grain roughness Shields
parameter @25 and the corresponding friction factor f25 could be avoided. In that
case, the most likely replacement for 625 would be the mobility number w, and
this was tried by Dingler (1974).

140 COASTAL BOTTOM BOUNDARY LAYERS

 Section 3.4: Vortex ripples

Dingler suggested, on the basis of his field observations, that the ripples
disappear when y reaches a value of approximately 240, and that n/A could, in
general, be described as a function of the mobility number y.
However, while this may be true, when only quartz sand is considered, it
does not hold in general. This can be seen by comparing Figures 3.4.6 and 3.4.5.
In Figure 3.4.6 1/A has been plotted as a function of w for the same data
as in Figure 3.4.5 and it is quite obvious, that ripples formed by sediments of
different densities form different trends in this plot.

0.40

DINGLERS FLAT BED LIMIT |

0.04

T(sec) d(mm)
0.02 | + DHI TUNNEL 7.7-13.9 0.082
eon " 4.4-9.0 0.2 0.135
f CARSTENS et AL 3.55 0.19
4 " " 3.55 0.30 O
O YALIN & RUSSELL 1.82 0.48
0.01 | SO KENNEDY & FALCON 1.8-2.8 1.00
O NIELSEN (1979) 1.70 0.082

4 10 20 40 100 200

Figure 3.4.6: Ripple steepness as function of the mobility number for sediments of
different densities.

AND SEDIMENT TRANSPORT 141

Chapter 3: Bedforms and hydraulic roughness

3.4.4 Ripple height

Figure 3.4.6 shows that the mobility number cannot be used as a predictor for
ripple steepness when widely different sediment densities are considered. For field

0.40

0.20

0.10

0.04

0.02

LEGEND d(mm) T(sec)

CARSTENS et AL (1969) 0.19 3.55
0.01 " " " 0.30 3.55
DINGLER (1974) 0.17 1.7-5.0
" " 0.177 1.7-4.0
NIELSEN (1979) 0.082 1.0
0.082 1.7
0.004 0.36 1.7
0.17 Weré
0.082
©+070
©08xpe
0.002
4 10 20 40 100 200
Figure 3.4.7; Relative ripple height as function of the mobility number for laboratory
data (regular waves) with quartz-density sediments.

142 COASTAL BOTTOM BOUNDARY LAYERS

 Section 3.5: Bedforms in combined flows

conditions however, where most sediments will have densities close to that of
quartz (s ~ 2.65), it is quite reasonable to describe n/A as a function of the
mobility number.
Consequently, since the relative ripple length X/A, is also reasonably well
described as a function of y (cf Figure 3.4.2, page 137) one might expect that the
relative ripple height 1/A can be described as a function of y. Figure 3.4.7
shows that this is indeed the case.
Nielsen (1981) suggested the formula

Dives 10275 — 0,022 y for w< 156 (3.4.7)

A 0 for w> 156 _

to model the trend of the regular wave data in Figure 3.4.7.

For irregular waves, the formula

Ae2y!® (w>10) (3.4.8)

was suggested, where y and A should be based on the significant wave height.

3.5 BEDFORMS IN COMBINED WAVE-CURRENT FLOWS

Bedforms generated by combined wave-current flows occur in a number of
different coastal sub-environments: inner shelf subjected to waves and tides, rip
channels; rip feeder channels; surf zones with longshore currents and finally in the
mouths of estuaries.
They have been studied in the laboratory by Bijker (1967), Natarajan (1969)
and Arnott & Southard (1990), and in the field by Amos & Collins (1978), Coffey
(1987) and Amos et al (1988).
Qualitatively, all observers agree that when co-directional currents are
superimposed, the ripples tend to migrate in the direction of the current (while the
net sediment transport may be either with or against the current) and they become
asymmetrical with steeper downstream faces.
When perpendicular or near-perpendicular currents are superimposed, two
ripple systems may coexist with one or the other being better developed according
to the relative current strength, see Figure 3.5.1 and Amos et al (1988).

AND SEDIMENT TRANSPORT 143

Chapter 3: Bedforms and hydraulic roughness

Figure 3.5.1: Ripple types mapped

in terms of wave and current
Shields parameters. Open triangle:
wave ripples, filled square: wave
ripples with subordinate current
ripples superimposed, 0 : wave
ripples and current ripples
coexisting, open square: linguoid
current ripples, e : current ripples
with subordinate wave ripples
superimposed, + : poorly developed
ripples, x : flat bed. The curve
corresponds to the initiation of
motion criterion 025+9 = 0.04.
After Amos et al (1988).

0.01 On 1

From their field study (weak to moderate flows in 22m depth on the shelf off
Sable Island), Amos et al identified eight different bedform regimes and mapped
them in terms of the grain roughness Shields parameter 625 versus a current
Shields parameter given by

ae 14 p 0.003 uioo
p (s-l) gd

where Uioo is the mean current velocity measured one metre above the bed. See
Figure 3.5.1.
The bedforms in strong rip currents are not well researched for the good
reason that such rips are extremely dangerous work environments. However,
qualitative observations indicate that the bedforms tend to be current-dominated,
e, they have steeper offshore faces and migrate offshore.
In fairly deep bar troughs over which waves have reformed after breaking on

144 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

the bar, very regular wave ripples are often found. One such case was reported by
Nielsen (1983) and Wright et al (1986). The depth was 1.2m to 1.5m, the
significant height and period of the reformed waves were 0.5m and 7s
respectively, and the depth averaged longshore current was approximately 0.5m/s.
In shallower bar troughs or rip feeder channels however, the conditions are
often quite obviously current-dominated, with the bedforms resembling current
dunes. This may be due to relatively weaker wave activity, Put,it may also be due
to the shallower depth itself via the current Froude number Vv’/gD. The stability
analysis of Engelund (1970) indicates that current dunes are generally formed at
Froude numbers between 0.5 and 2.0, but are unlikely to form at Froude
numbers below 0.25.
With respect to the megaripples which are sometimes found in oscillatory
flows of high intensity (@ 2 1, see Section 3.4.2) , Arnott and Southard (1990)
found that they generally disappeared if a moderate current (<u> > 0.05m/s) was
superimposed. Their experiments were carried out in an oscillating tunnel with
0.09mm sand and a fixed wave period of 8.5s.
Coffey (1987) observed bedforms, current profiles and sediment suspension
profiles in depths between one and two metres in the mouth of the Port Hacking
estuary south of Sydney and found ripples superimposed on the shoals. These
ripples were in the relative height and length ranges : 50 < n/d < 200 and 300 <
A/d < 1540, which corresponds to the expected ranges for current ripples. The
relative current strengths ux/ ux for these observations were in the range [0.15;
0.72).

3.6 HYDRAULIC ROUGHNESS OF NATURAL SAND BEDS

3.6.1 Introduction
In order to formulate simple models of natural flows, it is generally necessary
to apply a simplified description of the bed geometry, and in the extreme, we often
try to summarise the bed geometry in terms of a single length. Most commonly, the
chosen length is the equivalent Nikuradse roughness or briefly the hydraulic
roughness, r.
The only bed geometry for which the definition of the roughness is obvious is
a layer of densely packed spheres for which the roughness equals the grain
diameter, r=d. For all other geometries, the definition is indirect.

145
AND SEDIMENT TRANSPORT
Chapter 3: Bedforms and hydraulic roughness

That is, the roughness is determined from the structure of a purely steady
flow above the bed, see Equation (1.5.9), page 66.
The following section deals with the concept of the equivalent Nikuradse
roughness of natural sand beds, particularly those exposed to oscillatory flows. It is
found from the available friction and energy dissipation data that the roughness of
such beds are generally one to two orders of magnitude larger than that of sand
paper with the same sand size.
This great roughness is in many cases obviously due to bedforms which
generate roughness of the order of their height (r= 1 ).
However, also flat, mobile sand beds dissipate wave energy at a high rate and
thus, in this sense, appear very rough, particularly at high flow intensities where a
substantial layer of sediment is in motion. Thus, flat beds in oscillatory sheet-flow
(252 1.0) generally exhibit roughness values (based on total friction or on total
energy dissipation rates) of the order 100 to 200 grain diameters.
This is somewhat surprising because the corresponding roughness in steady
sheet-flows are generally one order of magnitude smaller. It therefore prompts the
question whether a substantial part (more than half) of the energy dissipation
measured by Carstens et al (1969) and Lofquist (1986) on flat sand beds, could
have been due to other mechanisms than skin friction. If that is the case, the
obvious candidate for an additional dissipation mechanism under waves is
percolation, see e g Sleath (1984) or Dean & Dalrymple (1991).
The suspicion above is reinforced by the observations of sediment transport
rates over flat beds under waves, which indicate that the effective stress, with
respect to sediment transport, is considerably smaller than the bed shear stress
measured either directly or via the rate of energy dissipation.
Thus, the bed-load data analysed in Sections 2.4.2 and 2.4.3, page 117 ff,
indicated that the effective sediment transporting stress corresponds to a roughness
of only about 2.5d50, while energy dissipation measurements under similar
conditions indicate roughness values of the order 100d50.
This difference between steady and oscillatory flows remains unexplained,
except for the possibility that it may be due to percolation.

3.6.2 Sand bed friction factors

The hydraulic roughness is closely related to the rate of momentum transfer
at the bed and hence to the friction factor f.
Measured bed shear stresses and energy dissipation rates for sand beds
exposed to oscillatory flows are presently available from two main sources:

146 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

Carstens et al (1969) and Lofquist (1986).

Carstens et al measured the rate of energy dissipation, while Lofquist
measured the total drag in terms of the total pressure gradient minus the inertia of
the fluid.

0.1

0.01
0.01 0.1 1
Figure 3.6.1: Measured wave energy dissipation factors from Carstens et al (1969)
and from Lofquist (1986). Legend: bar: Lofquist 0.55mm sand, +: Lofquist 0.18mm
sand, *: Carstens et al 0.19mm, rectangle: Carstens et al 0.30mm, x: Carstens et al
0.59mm. All of the above correspond to equilibrium ripples while the triangles
correspond to artificial flat beds where measurements were taken before ripples had
time to form (Carstens et al ).

By assuming that alternate dissipating mechanisms than surface friction were

negligible in the experiments, the results of Carstens et al (1969) and of Lofquist
(1986) have been translated (Equation 1.2.26) into the energy dissipation factors

AND SEDIMENT TRANSPORT 147

Chapter 3: Bedforms and hydraulic roughness

(fe) which are plotted in Figure 3.6.1. fe is expected to be smaller than fw, but the
differences are generally small compared to the scatter see Figure (1.2.9).

0.1

0.01

0.01 0.1 1 10
Figure 3.6.2; Steady flow friction factors over natural sand beds with median sand
sizes in the range [0./9mm; 0.93mm] and depths in the range [0./4m, 0.33m]. Data
from flume experiments by Guy et al (1966). Legend: filled rectangle: plane bed and
transition., + : rippled beds, *: dunes, open rectangle: antidunes, x : standing
waves, triangle: chute and pool flow.

Roughly speaking, these friction factors for sand beds in oscillatory flows are
an order of magnitude larger than those found in steady flows; namely between
0.04 and 0.4, compared to a range of about [0.005; 0.04] for steady flows, see
Figures 3.6.1. and 3.6.2.
When data from artificially flattened beds at low flow intensities (02.5 < 0.25)
are excluded, the range of observed energy dissipation factors for sand beds under
oscillatory flows is

148 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

0.06 < fe < 0.5 (3.6.1)

In order to provide a steady flow comparison, the steady flow friction factors
in Figure 3.6.2 are derived from the flume data of Guy et al (1966) through the
definition

t== 5 Pf <u>
2 3 =
t=pgDI

The abscissa @@ is an effective Shields parameter derived from measured

transport rates via Equation (2.3.12), page 114, which is a reasonably reliable
relationship between Shields parameter and sediment transport rates over flat beds
see Figure 2.3.2, page 113.
At high rates of sediment transport, the momentum transfer between the flow
and the bed is partly due to sediment particles receiving momentum from the flow,
when they are picked up and accelerated, and transferring it to the bed, when they
are deposited.
In this case the friction coefficient contains a contribution fs from the moving
sediment as well as the usual ff from the form drag and the surface drag on a
fixed bed.
Correspondingly, the roughness of a mobile sand bed may be seen as
including a "fixed bed component" rf and acomponent rs, which accounts for
the momentum transfer by moving sediment.
We shall see in the following, that the fixed bed contributions are similar in
steady and oscillatory flows over rippled beds; while the moving sediment
contributions, which become dominant over flat beds may be much larger in
oscillatory flows with similar grain sizes and Shields parameters.

3.6.3 Sand bed roughness in steady flows

The following contains a brief discussion of the hydraulic roughness of sand
beds exposed to steady flows. Since the main aim of the section is to provide a
basis for comparison with subsequent results for sand beds under waves, the
approach is somewhat different from previous treatments such as those of
Engelund & Hansen (1972), van Rijn (1984) and Wilson (1989).
Nikuradse (1933) noted that if the formula

u(z) = = In= (3.6.2)

AND SEDIMENT TRANSPORT 149

Chapter 3: Bedforms and hydraulic roughness

was fitted to steady flows over beds of densely packed sand grains, at sufficiently
large grain Reynolds numbers, then the zero intercept level zo was approximately
one thirtieth of the sand size (cf Schlichting 1979)

vy Ss OY RY for ux d/v > 70

Correspondingly, if a value zo has been found by fitting a distribution of the form

(3.6.2) to a measured velocity profile, the bed geometry is said to have the
equivalent Nikuradse roughness

i, == SUZ (3.6.3)

Figure 3.6.3: Hydraulic roughness derived through Equation (3.6.5) for sand beds
with different grain sizes in a variety of steady flows. Same legend as for Figure
S102.

150 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

If not the whole velocity distribution, but only the depth-averaged velocity
<u> together with the depth D and the hydraulic gradient J has been
measured, a crude estimate of the roughness can be obtained based on a total-flow
relationship of the form

<i> = SJ twee, ux =VgDI (3.6.4)

K Zo
Zo
which leads to

Te 3029 30D exp [ VeDI 1] (3.6.5)

A sample of roughness values derived via Equation (3.6.5) from a variety of

flow regimes over beds with different sand sizes are shown in Figure 3.6.3.
These data show that for a plane bed and for standing waves, the
roughness is of the order one to ten grain diameters, while for rippled beds, it is
typically of the order J00ds50. The latter corresponds to the typical height of
current ripples.
Note that the depth range of the shown data was only about [0./4m; 0.33m]
and that deeper flows may form dunes and antidunes of greater heights,
corresponding to greater values of r/d59. For a general discussion of the size and
shape of steady flow bedforms, see e g Allen (1982) or Sleath (1984).
Roughness values extracted with Equation (3.6.5) should not be interpreted
too closely for at least two reasons. Firstly, most of the flows are highly
non-uniform with depth and boundary layer structure varying strongly from
bedform crest to bedform trough. Secondly, the logarithmic velocity distribution
on which Equation (3.6.5) is based, is not an accurate description over the full
depth - only the bottom twenty to thirty percent of the velocity distribution is well
described by the logarithmic law of the wall.
Nevertheless, these crude roughness values are of interest in relation to the
roughness values for sand beds under waves, which will be discussed in the
following.
For steady-sheet flow in closed conduits, Wilson (1989) found remarkable
agreement between his own (1966) data and the simple formula

r= 50d (3.6.6)

AND SEDIMENT TRANSPORT 151

Chapter 3: Bedforms and hydraulic roughness

corresponding to r = 2Lg , where Lp _ is the equivalent bed-load thickness

which is defined in Section 2.3.3, page 111.

3.6.4 Hydraulic roughness from oscillatory flows

For oscillatory flows, similar procedures can be applied in order to extract the
hydraulic roughness of a sand bed from velocity or friction measurements.
If the velocity distribution has been measured so that the velocity defect D
can be plotted against z, then a value for z; (see Figure 1.3.2, page 44) can be
found. Subsequently, the roughness can be found from Equation (1.3.10), page 47.

1000

10

0.01 0.1 1
Figure 3.6.4: Hydraulic roughness corresponding to the fe-data in Figure 3.6.1. For
rippled beds the roughness is generally in the range [100ds0; 1000dso0], and if the
trend of the limited flat bed data can be extrapolated beyond 62,5 = 1, the bed
roughness under oscillatory sheet-flow is also of the order J00ds0 or more. Same
legend as in Figure 3.6.1, page 147.

152 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

2
r=Settee ti:A
0.0081 (3.6.7)

If only the peak bed shear stress f or the time-averaged wave energy
dissipation D ee has been measured, the hydraulic roughness may be inferred via
the wave friction factor fy or the energy dissipation factor fe, Thatis, fw or fe
is found from Equation (1.2.18) or Equation (1.2.26) respectively, and the
roughness is then found by applying a wave friction factor formula like Equation
(1.2.22) in reverse ie
Sls)
Inf + 5.977
r= a[ 5.213 (3.6.8)

The roughness values derived with this formula from the energy dissipation
factor data in Figure 3.6.1 are shown in Figure 3.6.4.

1000

Figure 3.6.5: Roughness

of flat sand beds in
oscillatory flows (flat bed
10 data from Carstens et al).
The full line corresponds
to Wilson’s (1989)
formula (3.6.6) for steady
sheet-flow, and the dotted
line corresponds to
Equation (3.6.9)

0.1
AND SEDIMENT TRANSPORT 153
Chapter 3: Bedforms and hydraulic roughness

We see that, except for (artificially) flat beds at low flow intensities
(25S 0.20), this roughness is of the order 100d50 to 1000d50.
This is interesting because, the steady flow data in Figure 3.6.3 indicate
roughness values as low as 2ds5q occurring in the upper (steady) flow regimes,
and Wilson’s (1989) formula (3.6.6) predicts only r= 5d __ for Shields
parameters of the order unity, see Figure 3.6.5.

3.6.5 The roughness of flat sand beds under waves

The roughness values in Figure 3.6.5 indicate that, in terms of the Shields
parameter 0 (= 6’), the energy dissipation over flat, mobile sand beds in oscillatory
flows corresponds to a roughness of the order

r = 70V0d for 02 0.5 (3.6.9)

This is of the order ten times more than predicted by Wilson’s (1989) formula
for steady flows.
No explanation has been given so far to this large difference between the
roughness values of sand beds under steady and oscillatory sheet-flows, except
that part of the energy dissipation or total drag under waves may be due to
percolation.
In terms of the grain roughness Shields parameter 025 the roughness,
corresponding to total drag on flat sand beds under oscillatory sheet-flow, may be
expressed as

r = 170 [email protected] d (3.6.10)

see Figure 3.6.6.

Grant & Madsen (1982) derived a different expression based on a model for
saltation in air by Owen (1964), but their result does not compare as well with the
presently available data when r is derived from f via Equation (3.6.8). For
details, see Nielsen (1983).
The friction factors in Figure 3.6.1 and the corresponding roughness values in
Figure 3.6.4 show that sand beds in oscillatory flows are always fairly rough. That
is, the range of relative roughness values, including those of artificially flat beds
with 625>0.3 is

154 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

Figure 3.6.6: Relative

roughness of flat sand
beds in oscillatory flow as
function of 625. The
curve corresponds to
Equation (3.6.10). The
roughness values are
obtained from Equation
(3.6.8) applied to the
energy dissipation factors
of Carstens et al (1969).

0.085 “<L7A <<. 1.15 (3.6.11)

If this is a valid indication of the boundary layer structure in general over flat
beds of loose sand, it has important implications for the relevance to flow over
natural sand beds of many of the existing models of oscillatory boundary layers.
The reason is that many of these have been developed to match experimental
evidence from flows with a relative bed roughness of less than 0.03, and are not
suited for flows with r/A>0.1, see Section 1.3, pp 40-52.

3.6.6 The roughness of rippled beds under waves

For sand beds covered by vortex ripples, the shape of which varies little for
62.5 < 0.25 (see Figures 3.4.4 and 3.4.5, pp 139 and 140), it is natural to expect the
hydraulic roughness to be closely related to the ripple height. This is confirmed by
Figure 3.6.7.

AND SEDIMENT TRANSPORT 155

Chapter 3: Bedforms and hydraulic roughness

0
9
8
7
6
5
4
3 Figure 3.0.75 The
hydraulic roughness of
rippled beds is between
2 one and three _ ripple
heights as long as the flow
1 is not too strong
(62.55 0.5). Same legend
as in Figure 3.6.1.
0
0 0.2 0.4 0.6 0.8

Indeed , r/1_ is seen to vary very little up to 625= 0.5 even though the
ripple steepness generally starts to decrease from about 825=0.25. See Figures
3.4.5 and 3.4.7.
The momentum transfer between a sediment laden flow and the bed can be
seen as consisting of two components: The usual drag contribution, which would
also be there for a fixed bed with the same geometry, and a component due to the
transfer of momentum by accelerating and subsequently crashing sediment
particles.
Correspondingly the roughness may be seen as consisting of a fixed bed part
rf and of a moving sediment part rs:

r = Mft+rs (3.6.12)

The magnitude of the fixed bed contribution rg for rippled beds may be

156 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

0.01 0.1 1
Figure 3.6.8: Relative ripple roughness as function of the grain roughness Shields
parameter for oscillatory flow data. Same legend as in Figure 3.6.1.

inferred from some of the experiments which have been carried out with ripple-like
fixed beds.

It has generally been assumed that the roughness of a rigid ripple profile
would be proportional not only to the ripple height, but also to the steepness and
hence relations of the form r= const-n 1/A or rA/n* = const have been
sought.
Thus, Bagnold’s (1946) sharp crested, parabolic ripples correspond to r Wr
=20.3, while the triangular concrete ripples of Jonsson & Carlsen (1976)
correspond to r WA =10.9 and 7.4 for TestI and Test II respectively.
Figure 3.6.8, where r Wn? is plotted versus 25 for the available movable

AND SEDIMENT TRANSPORT 157

Chapter 3: Bedforms and hydraulic roughness

bed data, indicates that a value for r Wn? of about 8 is appropriate for vortex
ripples at low flow intensity.
In order to account for the roughness contribution (momentum transfer) from
the moving sand over ripples, we may add a term corresponding to the expression
(3.6.10), page 154, for moveable flat bed roughness and get the following estimate

r = 8772/0 + 170V025-0.05 d (3.6.13)

Estimated
(given
ripples)
[cm]
r

10
Observed r [cm]
Figure 3.6.9: Predicted versus measured roughness values corresponding to the
Se-data of Carstens et al (1969) and of Lofquist (1986). The predictions are based on
Equation (3.6.13) with measured ripple data inserted.

158 COASTAL BOTTOM BOUNDARY LAYERS

 3.6: Hydraulic roughness of natural sand beds

where the ripple geometry may be measured or estimated from Equations (3.4.5,
3.4.7) or (3.4.6, 3.4.8).
Figure 3.6.9 shows a comparison between this formula and the roughness
values derived via Equation (3.6.8), page 153, from the fe-data of Carstens et al
(1969) and Lofquist (1986). The figure shows reasonable agreement.
We note that the nature of the first term makes it impossible to estimate the
roughness by a simple formula of the form r/d = F(@25) or r/d = F(w) as
suggested by Raudkivi (1988). The reason is that the ratio 17/Ad_ varies strongly
with the wave period for a fixed value of 25 or of wy, see the discussion of
Nielsen et al (1990).
While, according to Figure 3.6.9, energy dissipation rates based on Equation
(3.6.13) are in good agreement with experiments, there is reason to suspect that the
roughness determined from (3.6.13) and (3.6.10) for flat beds are too large for the
estimation of the vertical scale of suspended sediment profiles, see Section 5.4.9,
page 255.
Also, the discussion of bed-load transport rates and corresponding effective
bed shear stresses for flat beds under waves in Section 2.3 indicates that bed shear
stresses determined from a roughness given by (3.6.13) or (3.6.10) for flat beds are
too large.
The optimal choice of roughness for both bed-load transport rates and
suspension distributions over flat sand beds under waves lies closer to Wilson’s
steady flow formula (3.6.6), page 151, with 6 replaced by 025.
Thus, for purposes other than estimation of frictional dissipation, it might be
appropriate to replace Equation (3.6.13) by an equivalent formula with the
"moving grain contribution" given by Wilson’s expression for example

r = 8n’°/A + 505d (3.6.14)

AND SEDIMENT TRANSPORT 159

Chapter 3: Bedforms and hydraulic roughness

160 COASTAL BOTTOM BOUNDARY LAYERS

 CHAPTER 4

THE MOTION OF SUSPENDED

PARTICLES

4.1 INTRODUCTION
Transport of suspended sand is of great practical importance in coastal and
fluvial engineering. Therefore, the mechanics of suspended sediment transport
have been the object of substantial research efforts through several decades.
However, there is still much to be learned about the concentration magnitude and
about the distribution of suspended sediment in waves, in currents, and in
combined wave-current flows.
In order to describe suspended sediment distributions we must first seek to
understand the behaviour of suspended sand grains in different types of flows.
That is the object of the following sections.
The most important flow structure in connection with suspended sediment is
that of a vortex with horizontal axis because such vortices are able to trap sand
grains and carry them along over considerable distances, see Figure 4.1.1 (right).
In contrast, a wave motion where the fluid orbits are also circles or ellipses,
has no trapping capability, see Figure 4.1.1 (left). The mechanism of sand trapping
in vortices was first pointed out by Tooby et al (1977), but we shall discuss it in
detail in Section 4.6, p 181, and refer to some empirical evidence of its importance.
In coastal sediment transport the nearly homogeneous oscillating flow field
induced by surface waves is, of course, important, and there have been many
attempts to derive a suspension-maintaining mechanism from this flow structure.
For example, the delay distance idea of Hattori (1969) and the time lag hypothesis
of Bhattacharya (1971).

AND SEDIMENT TRANSPORT 161

Chapter 4: The motion of suspended particles

The former of these has little physical substance but the latter will be proven
and quantified in Section 4.5.
A third mechanism, which might influence the sediment settling rate under
waves, is the possible settling velocity reduction due to nonlinearity of the drag
force. We shall see in Section 4.5.5, through an analytical approach to this
mechanism, that the changes in settling rate due to the wave motion itself are either
purely oscillatory, and therefore without net effect, or they are so small that they
have no practical importance.

Wave Vortex

sediment
water ath
particle

U, = Wo + Wo (€7) u,=Oo+w,O(e)

Figure 4.1.1; While vortices and wave motions are similar in so far as both have
circular or elliptical fluid orbits, they differ greatly with respect to their influence on
suspended sediment particles. Sand will settle almost unhindered through the wave
motion, but it will tend to get trapped in vortices.

The analysis of the motion of suspended particles in the following sections

takes a stepwise, analytical approach.
We start with the simplest case, where the fluid accelerations are negligible
compared to the acceleration of gravity. In this case, the relative velocity between
sand and water is everywhere equal to the still water settling velocity wo.

162 COASTAL BOTTOM BOUNDARY LAYERS

 4.2: The settling velocity

The second step takes terms of magnitude ewe into account, where € is the
relative magnitude of the fluid accelerations

1 du | (4.1.1)
metieruceulelt

Finally, as a third step, effects of the order Wo are considered. Hence the
sediment particle velocity vector us is built up as a series of the form

Us = U+Wottritunt...

= ut vol
(SJreuneet +. (4.1.2)

where uw is the fluid velocity vector, see Figure 4.1.2.

The justification for this approach lies very much in its usefulness. That is, in
the fact that the most important suspension maintaining mechanism, namely
trapping in vortices, can be derived from the zero order approximation

us = U+Wo (4.1.3)

Secondly, the fluid accelerations induced by waves are generally much smaller
than the acceleration of gravity so that the perturbation parameter € is
conveniently small.

Us

Figure 4.1.2: The sediment particle velocity vector, us is described in terms of a

perturbation series.

AND SEDIMENT TRANSPORT 163

Chapter 4: The motion of suspended particles

4.2 THE SETTLING VELOCITY

By the sediment settling velocity wo we understand the terminal velocity of a

single sediment particle which settles through an extended, resting fluid.
In that situation the fluid drag on the particle balances exactly the force of
gravity. For a spherical particle with relative density s and diameter d_ this can be
expressed by

L5 P Rete + otk S=oae p(s-1) gs

4d Cows e (4.2.1)
which corresponds to

Wo = V A(s-1) gd (4.2.2)
3CpD

STOKES
Cp=24/R

Figure 4.2.1: Drag coefficients for spherical particles.

164 COASTAL BOTTOM BOUNDARY LAYERS

 4.2: The settling velocity

The drag coefficient Cp is a function of the particle Reynolds number

R = Wo d/V and of the particle shape. The relationship for spherical particles is
shown in Figure 4.2.1. The inclined asymptote

Cp. = aceian Choy

B Wo d/vV Gees)

corresponds to the linear drag law, Stokes law

Fp = 3npvdwo (4.2.4)

which applies for very low Reynolds numbers.

Balancing the linear drag force given by equation (4.2.4) with gravity leads
to the alternative expression

_ (s-lgd Ps
am 18v

which is valid for very small (R = wo d/v <1), spherical particles.

Gibbs et al (1971) provided the following empirical formula for wo(d)

-3v+ V9 Vv" + 2d (s—1)(0.003869 + 0.02480 d)

ae 0.011607 + 0.07440 d (4.2.6)

where wo is measured in cm/s, d incm, and the viscosity v in cm’/s. This

formula is applicable for spheres in water for the diameter range [0.0063cm;
1.0cm).
Natural sand particles are of course not spherical, but more or less angular or
even disc shaped in the case of shell hash. For such natural grain shapes the
formula (4.2.6) should represent the upper limit for the settling velocity as function
of size. An example of measured settling velocities as function of “sieve size” for a
sample of natural beach sand is shown in Figure 4.2.2.
Many particles which settle together in a thick suspension will settle
somewhat slower than single particles if the suspension is homogeneous. The
reason is, that the downward motion of one sediment particle will generate a
compensating upward flow elsewhere, which will delay the downward motion of
the sediment particles.
On the other hand, a dense cloud of sediment in an otherwise clear fluid will

AND SEDIMENT TRANSPORT 165

Chapter 4; The motion of suspended particles

settle faster than the typical settling velocity of the individual particles. This effect
may be the reason for the unexpectedly large, measured settling velocities for the
smaller ( d<0.04 cm.) particles in Figure 4.2.2.
These small, angular sand grains may have been settling in clouds.
Alternatively they may have obtained abnormally large averaged settling velocities
by hitch-hiking during parts of their trip in the wakes of larger particles.
These effects must be considered when settling velocity measurements from
settling tubes, where many sand sizes are dropped together, are interpreted.

w(cm/s)

0 0.05 0.10 0.15 0.20

Figure 4.2.2; Settling velocities (w0, ws0, w90) determined with a settling tube for
narrow sieve fractions of a sediment sample from the swash zone at Palm Beach,
Sydney, Australia. This sample contained a large fraction of shell hash, so the
measured settling velocities lie considerably below the curve given by Gibbs et al for
quartz spheres, Equation (4.2.6). The relatively higher, measured values for the
smaller grain sizes are probably due to the particles settling as a cloud rather than as
individual particles.

166 COASTAL BOTTOM BOUNDARY LAYERS

 4.3: Equation of motion for suspended particles

4.3 EQUATION OF MOTION FOR SUSPENDED PARTICLES

The following section briefly discusses the various terms in the equation of
motion for suspended sediment particles. Subsequently, the equation is brought
into the most convenient form for deriving a perturbation solution in the form of
Equation (4.1.2), page 163.
Consider a quasi-spherical particle with diameter d and relative density s
which moves under the action of gravity and pressure gradients and drag from the
surrounding fluid.
We neglect the history term of Basset (1888) which accounts for changes in
the fluid drag due to changes in the flow structure around the particle, and neglect
any effects of the particles rotation such as the Magnus effect (Magnus 1853). The
reader is referred to, for example, Clift et al (1978) for a more detailed discussion.
With those simplifications applied, the equation of motion reads

ps Ef St = pede - e# Vp + es Cplu= Us|(uU— Us)

+ =od? Cu es(u— us) (4.3.1)

The variable drag coefficient Cp_ is defined by

~y
Co _ bie; (4.3.2)
Wo

in accordance with the variation of Cp with changing particle Reynolds number,

see Figure 4.2.1.
From Figure 4.2.1, we see that y = 1 for small particles which move in
accordance with Stokes Law, and y = 0 for large particles which settle under
fully developed turbulent conditions.
The usefulness of C “ has been proven by the numerical study of Ho (1964)
but the introduction of ‘e's and y in the present context also enables us to
develop general expressions which cover both of the important, special cases y= 1
and y=0.

AND SEDIMENT TRANSPORT 167

Chapter 4: The motion of suspended particles

The last term in Equation (4.3.1) corresponds to fluid pressure on the added
hydrodynamic mass, and the coefficient Cy may be assumed to have a value
close to 0.5 which is the theoretical Cy-value for a sphere.
By introducing the pressure gradient vector in the form

Vp = pe - pm (4.3.3)
and rearranging, the equation of motion becomes

(s+ Cw = (s-l)g + (1+Cw) Bi nee lus—ul(us—u) (4.3.4)

dt 4d
We wish to write the equation of motion in terms of the relative sediment
particle velocity uy = us—u_ and for this purpose we note the following about
the total derivative

dus _ dus
i ae
= £ (usw) + (utur)-V(u+ur)

a ou
= ar + =e + u-Vu + (ut+uy)-Vuy + urVu

_BA du,asdale
duy om Vu (4.3.5)

where we have applied

dur Our
“aks = 5 + (u + ur)-Vur (4.3.6)

With Equation (4.3.5) inserted, the equation of motion (4.3.4) becomes

d ,

orodur +a) == (s—l)g + (1-s) el

ie Seen
He lus — ul(us — u)

(4.3.7)
and with

168 COASTAL BOTTOM BOUNDARY LAYERS

 4.3: Equation of motion for suspended particles

we get
dur + Cp
NEARLY MOIan (4.3.9)
diy arpdcd(sveiGi) hintvo Gesoawn
It is convenient to eliminate the drag coefficient Oy from this equation. To
this end, we apply the definition of C,, , Equation (4.2.1), page 164, which gives

3Co _ (-Ng (4.3.10)

ted 8 ie
and the definition, Equation (4.3.2), of a . By inserting these into Equation
(4.3.9) we find

I-¥
du u
ae «a(t Wied Oey 5 me pugLE (4.3.11)
6)

This general differential equation for the relative sediment particle velocity
ur is the basis for the analysis of the motion of suspended particles in the
following sections.
The nature of the solutions is found to depend in a very important way on the
third term u;-Vu , which, in essence, represents the effect of the small scale flow
structure.
This term is negligible in a pure wave motion (for woT/L << 1), and this is
the reason why a pure wave motion has very little effect on the sediment settling
rate, see Figure 4.1.1, page 162, and Section 4.5.5, page 178.
On the other hand, for a sediment particle inside a vortex with diameter of the
order of ten centimetres, the term u,-Vu is very significant, and it provides the
vortex trapping mechanism which is discussed in Section 4.6, pp 181-189.
Equation (4.3.11) is linear for y=1, and exact analytical solution is
therefore possible, for small particles for which y= 1, see Figure 4.2.1.
For larger sediment particles, which have y-values in the interval [0;/],
perturbation solutions of the form

Us =~ U+uUr = U+wWotn
t+ unt.

are provided in Section 4.5 for homogeneous flows and in Section 4.6 for particles
in vortex flows.

AND SEDIMENT TRANSPORT 169

Chapter 4: The motion of suspended particles

4.4 ACCELERATING SEDIMENT IN A RESTING FLUID

Consider the simplified equation for the relative sediment velocity

d
et
OO 2 =e 2 (4.4.1)
(a)

which results from Equation (4.3.11) when the fluid is at rest (or moving with a
steady, uniform velocity), and the particle is so small that the flow around it is
laminar and Stokes law applies (y= /).
This equation is equivalent to

i (ur—-Wo)
—, = eR
wa (ur ae— Wo) (4.4.2)

where the vector g has been written as gwo/Wo, and this equation has the solution

Uur(t) = Wo + [ur (to) — Wo] & %8 (F—0)/o (4.4.3)

From this we see that deviations from the terminal settling velocity wo decay
exponentially with a time scale of wo/ag which, for beach sand with relative
density 2.65 and settling velocity 0.02m/s, is approximately 4 - / Os.
This means that, starting from rest, such a sediment particle will be moving at
99% of its terminal settling velocity after only 0.002 seconds.
In other words, this means that deviations from the first approximation

us = U + Wo (4.1.3)

are normally very small.

Note that the discussion above is only meant as an order of magnitude
estimate. A detailed analysis of situations where the magnitude of uy = us—u
changes considerably should include the history term (Basset 1888), but that would
preclude analytical solution.
A quantitative discussion of the importance of the Basset term is given in
terms of numerical solutions by Hardistry and Lowe (1991).

170 COASTAL BOTTOM BOUNDARY LAYERS

 4.5; Sediment particles in accelerated flow

4.5 SEDIMENT PARTICLES IN ACCELERATED FLOW

4.5.1 Introduction
In the previous section we saw that sediment particles in a resting fluid will
attain a relative velocity approximately equal to the terminal settling velocity after
only about one hundredth of a second.
However, if the fluid motion is accelerated, deviations of the order ew, will
remain where

€ = 1 du | (4.1.1)
g at
The nature of these deviations is the subject of the following section.

4.5.2 Small particles in accelerated flow

Most of the important characteristics of sediment motion in accelerated flow
are essentially the same for large and small particles. That is, their general nature
does not depend on the drag force being linear or non-linear. Consequently, the
general nature of these effects is quite adequately described by the linear equation
which can be derived from Equation (4.3.11) under the “small-particle-
assumptions”.
When the particles are so small that the drag force is linear and given by
Equation (4.2.4), the equation of motion (4.3.11) is linear. Under this assumption
of linearity and with the relative sediment particle velocity written in the form

Ur = WortUri (4.5.1)

we find the following linear equation for uy1

dur uri du
dt og—
A Wo + rl un:Vu = —-Aa—
— WoVu
i) (4.5.2 )
dt

This equation is particularly simple to solve if the flow field may be assumed
homogeneous so that the gradient matrix is zero, Vu = 0.
As an example, a solution (Equation 4.5.27, page 176) is given in Section
4.5.4, for settling through homogeneous, oscillatory flow.

AND SEDIMENT TRANSPORT 171

Chapter 4: The motion of suspended particles

4.5.3 Larger particles in homogeneous, accelerated flow

The non-linearity of the drag force for larger (y< 1) sediment particles gives
rise to a reduction of the settling velocity when the particles settle through an
accelerated flow.
To assess the importance of this effect for suspended sediment we must
determine its magnitude and that is the subject of the present section.
In order to obtain analytical solutions we shall express the relative sediment
1 Idul
particle velocity uy as a perturbation series in € = ai and formally assume

Grom.
For sediment particles moving in a homogeneous, oscillatory flow, it can then
be shown that all effects of order € Wo are analogous to the results for small
particles derived from the linear Equation (4.5.2). These effects are all purely
oscillatory, so there is no settling velocity reduction of the order € Wo. The settling
velocity reduction due to non-linear drag is of the order Wo .
We introduce the following dimensionless variables

U = u/R®

Uy = ur/Wo

T = ot (4.5.3)
where R is the typical scale of the water motion and @ its radian frequency.
Then Equation (4.3.11) for particles in a uniform flow becomes

Wr 08 yi-ry, _ 28 _,,Rw dU
aT WoW WoW a Wo® dT See,

or

dU.
eT en " a eg
dT Wo® WoW WoW dT oe

Wh . : ° :
where €=Ro'/g. We seek solutions to this equation in the form of a
perturbation series

172 COASTAL BOTTOM BOUNDARY LAYERS

 4.5; Sediment particles in accelerated flow

ur = Wotunt+unt... (4.5.6)

and the corresponding dimensionless Ur = ur/wo

Up = Up eu te Up F

Ur ib= “)
(0 Un
+ (wn) + 2(Un
| oe (4.5.7)

Then the magnitude of Uy; is given by

(UAH CUA+...) + (14 Wate Wnt.) (4.5.8)

and for the linearisation of the drag term in Equation (4.5.5) we find

Ur’ = 1- e(1-7)Wn+ on Uetack

e mH Wry o— «e7.(1-) Woe +) (4.5.9)

after using the binomial expansion

(l4xP? = 14+ px + Le ee (4.5.10)

We can now find a hierarchy of equations of motion corresponding to rising

order in € by inserting the expansion (4.5.9) into Equation (4.5.5), and separating
terms of equal order in €.

Order 1: For the order 1 solution we find the trivial equation

. i 7 waiEy) (4.5.11)
WoW —] WoW | -8

because we have set the order-1-solution to be wo

AND SEDIMENT TRANSPORT 173

Chapter 4: The motion of suspended particles

Order ¢: For the order € we find

UM,) ,, Sasi
dF \Wn ried nN WwW
(4.5.12)
dT Wo® (2-y)Wr1 WoW dT

corresponding to the dimensional

Auf rte eek hagl t t hey pe k Beeler (4.5.13)

dt |wri Wo | (2-Y) wri dt|w

Order ¢”: For the order €” we find

a(U si y ‘ (-yUn Wn
d (Un no.ge( Urs E\patiocons. Te) ud
aT | Wo lop] WoW = U2, + Peas Ww,
(4.5.14)

with the corresponding dimensional equation

ag urn ag (1- y)uniwri

d(u
dt i 7 ales a, =
alee we a ays7 a (4.5.15 )
- wei + 2 37+

4.5.4 Settling through a homogeneous, oscillatory flow

Substantial efforts have been made to find out if a homogeneous, oscillatory
water motion has any net effect on the settling velocity of sand. One possible cause
of a retardation is the non-linearity of the drag force.
Ho (1964) made numerical calculations of this non-linear drag effect, and
they compared favourably with experiments. However, no analytical solution was
given to provide an easy estimate of the effect.
In the following, we will give a simple analytical solution and show that the
settling velocity reduction due to non-linear drag in a homogeneous oscillatory
flow is of the order of magnitude (R w*/ g)’Wo where Rw” is the typical vertical
acceleration of the fluid motion.
Let us consider a flow field similar to the one used by Ho (1964) and others

174 COASTAL BOTTOM BOUNDARY LAYERS

 4.5: Sediment particles in accelerated flow

in experiments. The water is oscillated vertically as a rigid body

+ a 0
u= ut) = rs cea (4.5.16)

We need only consider the vertical fluid motion

w(t) = Racosot (4.5.17)

and the corresponding vertical sediment velocity

ws(t) = w()—-Wotwrntwrt... (4.5.18)

To find the first order solution wri, we use the vertical part of Equation
(4.5.13) which becomes, in this case

Latah obs (2-Y)wn = -@ us(R @ cos ot) (4.5.19)

dt 0 dt

oan oes (2-Y)wn = OR w” sin wt (4.5.20)

oO

This equation has the simple harmonic solution

Ro Wo, : =i
Wri = sin (wf — tan Bz) (4.5.21)
Sent ok oP
+B?
where

Wo ®
Bz = ws
Si
(Q-) agSo (4.5.22 )

We note that for sand size particles in water the value of Bz is of the order
of magnitude 10-*. Hence, for practical purposes, the expression (4.5.21)may be
approximated by

Wri = Ro” Wo sin wt (4.5.23)

aki d

AND SEDIMENT TRANSPORT 175

Chapter 4: The motion of suspended particles

and hence with reference to Equation (4.5.17)

Wr = __Wo _dw (4.5.24)

(2-y)g dt
Note that the magnitude of wr relative to the still water settling
velocity w, is the ratio between fluid accelerations and gravity

Wri 1 du.— 4.5.25

Wo im dt i ( )

[(2-y) is of the order unity].

According to the approximation (4.5.24) the sediment particle velocity,
correct to order € , has the form

Wo du
Us A u + Wo oe eS
Q-ve dt (4.5.26 )

which we can write as

Us = u(t-d:) + Wo (4.5.27)

This is in accordance with the time lag hypothesis of Bhattacharya (1971),

and shows that his assumed time lag is real and has the value

5; i eee (4.5.28)
(2-y)g
or approximately wo/g, which is of the order 0.00]/s to 0.0]s for beach sand.
We note that the solution (4.5.21) has the time average zero so there is no
settling velocity reduction of magnitude ew.
There is however a net reduction of the settling velocity of order €°Wo when
the drag force is non-linear. We find its magnitude by inserting the first order
solution (4.5.21) into the second order equation (4.5.15).
In order to simplify the algebra however, let us write the first order solution
(4.5.21) in the form

Wr = 7 sin ot! (4.5.29)

Then the vertical part of the second order equation reads

176 COASTAL BOTTOM BOUNDARY LAYERS

 4.5: Sediment particles in accelerated flow

da
te Zot: 8 02) we
a
res 2-3y+7 rt sin? w t (4.5.30)
dt oO we 2
or

da
Tt Ww
=

t
= mg 2+7 21 4 oso @ 2!) (4.5.31)
Wo Wo D 2

We are, at the moment, looking for a net (time-averaged) settling velocity

reduction, so we need only consider the time-averaged solution wr2 corresponding
to the time average of Equation (4.5.31). We find

a
wr2 =
ey
aes
2
rj (4.5.32 )

or with the full expression (4.5.21) inserted

2
—— _ _l-y Ro 1 4.5.33
ne eal g ve Sie

Wr,
/Wo
-
]
Figure 45.1: Settling
velocity reduction for a
Imm brass sphere in
vertically oscillating water
———— Ho’s Numerical Solution Wo = 0.39mls, @ =
Equation (4°5:33) 22.7rad/s. |Measurements
and numerical solution by
Ho (1964).

AND SEDIMENT TRANSPORT 177

Chapter 4: The motion of suspended particles

This shows that the settling velocity reduction due to non-linear drag has the
order of magnitude €’wo. The solution (4.5.33) is compared to measurements and
numerical results from Ho (1964) in Figure 4.5.1.
Note that even though the derivation was based on the formal assumption of
€ << 1, the agreement with Ho’s experiments is good up to € = 3.

4.5.5 Sand particles in wave motions

It is of considerable interest with respect to coastal sediment transport
modelling to analyse the behaviour of sediment particles in a pure wave motion.
That is the purpose of the following Section.
We shall find that deviations from the simple description

Us = U + Wo (4.1.3)

are either purely oscillatory, and therefore without a time-averaged effect, or they
are for all practical purposes negligibly small.
In most practical cases the water motion above the bottom boundary layer
under waves can be considered quasi-uniform as far as the motion of suspended
sediment is concerned.
More precisely, this is the case when the distance settled by a sediment
particle over one wave period is small compared to the vertical scale on which the
wave induced velocities change. That is when

kwoT << 1 (4.5.34)

where k is the wave number 2x%/L. When this condition is fulfilled together
with the assumption that kRz << 1, where Rz is the amplitude of the vertical
water particle motion due to the waves.
These assumptions mean that the term uyVu_ in the equation of motion
(4.3.11), page 169, can be neglected. When that is done, the problem becomes
analogous to homogeneous flow problem which was considered in the previous
section.
For a wave motion simplified in accordance with the assumptions above, the
fluid motion near the bed is given by

uote he
WR; cos wt
4.5.35
(4.5.35)

178 COASTAL BOTTOM BOUNDARY LAYERS

 4.5: Sediment particles in accelerated flow

We now use Equation (4.5.13), page 174, to find the horizontal and vertical
components of the first order solution uy1, [us(t) = u(t)+wotun(t)+...]. In
analogy with the solution (4.5.21) for purely vertical motion, we find

ao Wo : cos (wt — tan'Bx)

ur) =
g M+ (4.5.36)

sete
Ro Wo_
ie sin (wt
— tan 'B:Bz)

where

oe _ (4.5.37)
and 8, is given by Equation (4.5.22), page 175.
Note that the time average uy; =0 . Therefore, as indicated by Figure 4.1.1
page 162, there is no net drift induced by the waves at order ewo.
The solution is simplified considerably when the particle is so small that
y= 1 in which case we will also have Bx, Bz << 1. Then we get

2
Wo COS Wt
ur) = : ee Wog ales
du
dt 4.5.38
(4.5.38)

Wo Sin @t
and hence

Wo du
Hes ul) Wes = eins
= (4.5.39)

rae w(t?) + Wo (4.5.40)

This result proves and quantifies Bhattacharya’s (1971) time lag hypothesis
for small particles in the two-dimensional case. For larger particles (y=0) the
time lag for the vertical component is smaller, namely wo/(2-)g instead of
Wo/, g.

AND SEDIMENT TRANSPORT 179

Chapter 4: The motion of suspended particles

Another instructive interpretation of ui is given in Figure 4.5.2 from which

it is realised that uy1 represents a centrifugal effect.

ud
i gg Figure 4.5.2: In a
homogeneous flow field
where the water particles
follow elliptical orbits, the
component u,i of the
sediment velocity is
directed away from the
orbit centre and _ thus
represents a _ centrifugal
effect.

For larger sediment particles (y~0) there is a net delay of settling wy due
to the non-linearity of the drag force. It can be evaluated by inserting the first order
solution (4.5.36) into the second order Equation (4.5.15), page 174, and taking the
time average.
The solution is analogous to Equation (4.5.33). Hence, the settling velocity
reduction Wr is ofthe order €’wo.
In most practical cases the reduction is insignificant. For example, for a
typical wave motion with vertical semi-excursion Rz = 0.1m and @ = 1s! we
find from Equation (4.5.33).

wn = 6:10 wo (4.5.41)

which is a totally insignificant reduction of the settling velocity.

180 COASTAL BOTTOM BOUNDARY LAYERS

 4.6: Sediment trapping by vortices

4.6 SEDIMENT TRAPPING BY VORTICES

4.6.1 Introduction
One of the most important mechanisms for entraining and suspending
sediment is the trapping of sediment by vortices with horizontal axes. See Figures
4.1.1, page 162, and Figure 4.6.1.
This mechanism was first illustrated experimentally by Tooby et al (1977).
The reader is referred to Tooby et al’s magnificent photographic illustration. It
shows a heavy particle as well as air bubbles trapped on circular paths inside a
water filled cylinder which rotates around its horizontal axis.
In the following we shall develop an analytical description of the
phenomenon along the same lines as used in the previous sections. That is, the
sediment particle velocity is written in the form

Us = U+WotUnt+urt.... (4.1.2)

= u t+ wo]
[Steuart |

where uw is the fluid velocity vector and wo _ is the still water settling velocity. The
perturbation parameter € is the ratio between fluid accelerations and gravity

er =t—-r4 (4.1.1)

It will be shown that the essence of the trapping mechanism is purely

kinematic and can be explained by the zero-order solution

us = U+Wo (4.6.1)

4.6.2 The kinematic trapping effect

In the following we shall see how the zero order solution for the sediment
particle velocity

Us(x,z,t) = u(x,z,t) + Wo (4.6.2)

AND SEDIMENT TRANSPORT 181

Chapter 4: The motion of suspended particles

provides adequate approximations to the sediment paths in simple flow structures.

That is, on the basis of these approximations we can explain important sediment
suspension mechanisms such as the sediment trapping by vortices.
Consider the water motion of a forced vortex with the velocity field

u(x,z) = (7) (4.6.3)

x

which is that of a rigid body rotating around the origin with angular velocity
w, see Figure 4.6.1. Applying Equation (4.6.2), we find the approximate sediment
particle velocity

-Z 0 4
us = U+ Wo = of) + ee = (sre) (4.6.4)

This equation is identical to Equation (4.6.3), which describes the water

motion, except for a horizontal shift of magnitude wo/.

Figure 4.6.1: Sediment particle with settling velocity wo moving in a forced vortex
with velocity field u given by Equation (4.6.3). The particle paths corresponding to
u+Wo are circles and thus the time-averaged velocity of a sediment particle will (at
this level of approximation) be zero.

182 COASTAL BOTTOM BOUNDARY LAYERS

 4.6: Sediment trapping by vortices

Hence, the sediment paths are analogous to the fluid paths except for a
horizontal shift of wo/@. That is, since all circles around (0, 0) are fluid particle
paths, any circle around (wo/@, 0) is a possible sediment path, see Figure 4.6.1.
Note that the sediment paths in a pure, deep water wave motion are not
closed, although the water particles move in circles like in the vortex, see Figure
4.1.1, page 162, and Section 4.5.5, page 178 ff.
It is important to understand the difference between wave motion and vortex
motion with respect to sediment suspension. It stems from the fact that the wave
motion is essentially homogeneous while the vortex motion is not.
Quantitatively, in terms of the equation of motion (4.3.11), page 169, the
difference is that the term u;-Vu is approximately 0 for the wave motion while
it is significant for sediment particles inside vortices.
Tooby et al (1977) showed experimentally that small sediment particles in a
vortex do, in fact, follow circular paths very closely and only very slowly spiral
away from them. This spiralling process is discussed in Section 4.6.3, page 186.
The fact that the settling is strongly delayed in a vortex flow field has been
noticed by Reizes (1977), who found the phenomenon in a numerical study. He
concluded that the sediment particle must tend to spend more time in the upward
moving parts of the flow than in the downward moving parts.
This is indeed the case. The sediment path shown in Figure 4.6.1 lies entirely
in the “upward moving” part of the vortex.
For buoyant particles or air bubbles the centre of the “sediment path” would
lie on the negative x-axis. Thus, they would spend the majority of the time within
downward moving fluid.
The described trapping mechanism will work for all sand grains with settling
velocity smaller than the maximum velocity in the vortex.
The next question to be asked is whether the trapping is a feature of the rather
unnatural, forced vortex only.
The answer is no for the following reason. In general, the velocity field of a
circular vortex with angular velocity @ can be written

u(x%,z) = WF(X’+7) | (4.6.5)

and the first approximation for the sediment velocity is then

usAY = U + Wo 0 = OFOC+7)| 7]x 4+ uy

—Wo (4.6.6)

AND SEDIMENT TRANSPORT 183

Chapter 4: The motion of suspended particles

For the components us and ws of this velocity field, we therefore have the
following symmetries

Us (X,-Z) = —Us (X, 2) (4.6.7)

and

Ws(x,-2) Ws (x, ra) (4.6.8)

This means that any particle path which crosses the x-axis twice must be
closed. Since, as a result of the symmetry, a particle that has travelled along the
curve P;P2 in Figure 4.6.2 must travel back to P; via the mirror image of PiP2.

Figure 4.6.2: Ina vortex, where the flow field has the form given by Equation (4.6.5),
any particle path given by us = 4+ Wo must be closed if it crosses the x-axis twice.

Hence, closed sediment paths, and the corresponding sediment trapping

capability, are general features of vortex flow of the general form (4.6.5).
In particular, this includes the irrotational vortex with the fluid velocity field
2
WRo (-z
ES eee E) (4.6.9)

184 COASTAL BOTTOM BOUNDARY LAYERS

 4.6: Sediment trapping by vortices

Therefore, the trapping capability of a vortex is not conditional on the flow being
rotational.
A fair model of many natural vortices is the Rankine vortex in which the
fluid velocity field is given by

ie Saal Mamet A
BO) = GRE ORY ie cael)
It is characteristic for this vortex model that the core rotates as a rigid body,
with the velocity being proportional to the distance from the centre, while
u_ becomes inversely proportional to this distance farther away.
In this vortex a sand grain with settling velocity wo can theoretically be at
rest at the two singular points in Figure 4.6.3. At these two points we have
u+ws = 0, and their coordinates are (R’w/2w, + V(R? @/ 2wo)” =e 0) The
circle which is shown, is given by the equation

z/R

Figure 4.6.3: Particle paths corresponding to us = &U+Wo in a Rankine vortex

with the velocity field (4.6.10). In the inner region, where u_ is essentially
proportional to the distance from the origin, we get closed sediment orbits that are
very similar to the circles inside the forced vortex, see Figure 4.6.1.

AND SEDIMENT TRANSPORT 185

Chapter 4: The motion of suspended particles

ZZ 2 2
fe
= ae ra | = ey = (4.6.11)
R 2Wo R 2Wo

and it is the locus of all points with ws= 0. Sand grains will move
upward (ws >0) in the interior and downward (ws <0) outside the circle.
Figure 4.6.3 also shows the sediment particle paths corresponding to
Us = u+ws. Some of these are closed and could thus keep sand grains trapped.
Trapping is only possible if the maximum upward water velocity in the
vortex exceeds the sediment settling velocity which, for the Rankine vortex, is the
case when Wo < WR/2.
The equation of the sediment path through a point (Xo, 0) is

pep eae (Rx. exp Ee ae 4

Now it could be argued that, since the trapping paths are closed, sediment
particles are no more likely to get onto them than to get off. Hence the trapping
mechanism might not be effective. However, the situation in practice is that the
sediment particles get into the vortex during its formation. This process is easy to
observe with the vortices behind ripples and dunes.

4.6.3 Secondary dynamic effects

In the previous section it was shown that a wide class of vortices, rotational
and irrotational, have the potential for trapping sediment particles which move in
accordance with

Us = uU + Wo (4.6.1)
This equation is only strictly valid however, if the flow is steady and uniform,
and that is never the case for vortex flow. Vortex flow is always accelerated, and
the accelerations will cause deviations from the closed sediment particle paths
described above. In the following we shall describe these acceleration effects and
show that they consist of a spiralling effect and a steady horizontal drift.
Qualitatively the spiralling effect is familiar and easy to understand. The

186 COASTAL BOTTOM BOUNDARY LAYERS

 4.6: Sediment trapping by vortices

vortex is a centrifuge. Heavy particles will spiral outwards and light particles or air
bubbles will spiral inwards towards its orbit centre.
The steady horizontal drift is best understood by considering a sediment
particle at rest at the point (wo/q@, 0) in a forced vortex with angular velocity
@, see Figure 4.6.1, page 182.
Resting at this position is possible for the sediment particle under the
assumption (4.6.1). However, a fluid particle whose orbit passes through this point
is travelling on a circular path with radius ws/@ and radian frequency «. It is
therefore accelerated towards the vortex centre with acceleration @ wo.
The horizontal pressure gradient which provides the fluid particle with this
acceleration will tend to push the “resting” sediment particle towards the centre of
the vortex. This describes the driving mechanism for the steady horizontal drift
which is not only present at the point (Wo/@ ,0) but everywhere in the vortex.
In order to obtain a quantitative description of the drift, consider the sediment
particle in Figure 4.6.4 which moves around the circle

Ros ot + Wo/®@
= 4.6.1
rs) R sin ot ) kAi6:12)
in a forced vortex in accordance with the zero order solution us = u+Wo.

Figure 4.6.4: The acceleration

required for the sediment
particle to stay on the circular
| »4 path is directed towards the
centre of the circle. However,
the pressure gradients in the
undisturbed flow are directed
towards (0,0). Hence, an excess
pressure force will make the
particle drift towards (0, 0).

AND SEDIMENT TRANSPORT 187

Chapter 4: The motion of suspended particles

A water particle on the fluid orbit, which passes through the sediment
particles position, would be travelling on a circular path with radius rs(t) and
radian frequency @ and hence would have the acceleration

d
= = -w'rs (t) (4.6.13)

dulg i 2 » {cos ot ie Wo @
6.14
dt oF Sra 0 ) a )

The velocity field of the forced vortex is given by

(1)- of3)
Wey =z
an
and with this and Equation (4.6.14) inserted, the equation of motion (4.5.2) for a
small particle reads

du
<a + Cea — @wr = awRcos@t + (a-1)Wo @
dt Wo

dw. ag ;
—T + Wri + Our = aorR sin wt (4.6.16)
dt Wo

These equations have the solution

a
QR
uri) _ nege ‘Yas cos (wf-tan™ 28) A (1-8) “ee
Wri Vi + 4p? sin (wt—tan! 2B) . 1+f? -B

(4.6.17)

where B = woW/ag is a quantity which is typically of the order 10! or less.

Hence, the order € contribution u;; to the sediment particle velocity can be
closely approximated by

Uri) _ @WR_
2
(cos at 1
bes strat Die © (ises — (-a) v6 (0| (4.6.18)

188 COASTAL BOTTOM BOUNDARY LAYERS

 4.7: Particles settling through turbulence

or

2
ur) = a Wor (1) = ad = Q)Wo B (| (4.6.19)

which shows that the first term represents a centrifugal effect with a spiralling time
scale
=i
ip = Eal => (4.6.20)

while the second term represents a horizontal, steady drift towards the z axis.

4.7 PARTICLES SETTLING THROUGH TURBULENCE

4.7.1 Introduction
In order to understand sediment suspension in turbulent flows it is first
necessary to analyse the influence of turbulence on a single particle which is
settling through it.
Experiments by Murray (1970) indicate that the turbulence has two main
effects on settling particles, which may be quantitatively described in the following
way.
Consider an ensemble of many identical particles which are dropped into a
water filled jar where a certain level of turbulence is maintained by some kind of
stirring, see Figure 4.7.1.
The distances L; , which the particles settle during the interval ¢, are
measured and the corresponding time averaged velocities wj=Lj/t are
calculated. It may be observed then that the ensemble average wj;_ is less than the
still water settling velocity

Wi < Wo (4.7.1)

and that the variance Var{wj} is roughly proportional to: the turbulence intensity
w as to the Lagrangian integral scale of the turbulence Ty ; and to //t. That is

AND SEDIMENT TRANSPORT 189

Chapter 4: The motion of suspended particles

: 0

Wot

STILL WATER
z

Figure 4.7.1: Sediment particles settling through turbulent water will, on the average,
settle more slowly than they would through still water.

Var{wi} « w2TL/t for t >TL (4.7.2)

The second result can be understood in terms of G I Taylor’s classical model
for diffusion by continuous movement (Taylor 1921), which will be outlined in
Section 4.7.3.
The first effect however, which is equally important for the maintenance of
sediment suspension, is not as well understood. It is however closely related to the
trapping mechanism described in Section 4.6, page 181 ff, by which heavy
particles or air bubbles can be trapped for a long time inside a vortex. Obviously, a
sediment particle which has been trapped for some time in a few vortices on its
way down will be delayed considerably. The same applies to an air bubble which
rises through a turbulent flow field.

4.7.2 The loitering effect

It is important to note, that the settling delay described above does not require

190 COASTAL BOTTOM BOUNDARY LAYERS

 4.7: Particles settling through turbulence

a specific, highly organised vortex structure. Trapping by vortices is just one

manifestation of the general “Loitering effect” which is illustrated in Figure 4.7.2.
The term “Loitering effect’ refers to the fact that a settling or rising particle
in a steady, non-uniform flow field will spend relatively more time (loiter) in those
parts of the flow which move against its natural settling/rising velocity. By this
mechanism, a non-uniform flow field with zero spatial mean velocity (<u> = 0)
will delay the settling of sediment particles and the rising of air bubbles.

X
Figure 4.7.2: A sediment particle which settles through this arrangement of vortices
will experience a considerable delay compared with settling through still water
because it spends more time in the upward moving parts of the fluid than in the
downward moving parts. Note that the spatial average of the particle velocity is wo .

AND SEDIMENT TRANSPORT 19 —

Chapter 4: The motion of suspended particles

To quantify the effect, consider the special flow pattern in Figure 4.7.2. For
simplicity, let the flow structure be such, that the particle velocity varies with z as

ws(z) = Wo (1+Acos 5. (4.7.3)

where D is the vortex diameter and A <1. Then the resulting settling velocity is

Ws = = (4.7.4)

where T2p_is the time it takes to settle past two complete vortices

2D
r dz
4.7.5
Ce JWs(Z)
0

2»
dz
RPS JWo (1+A cos 1 z/D)
0

2D
Typ2D = ——a
wo dA

Hence, the settling velocity is reduced from its still water value wo to

Ws = Wo(1-A’)°> (4.7.6)
This expression becomes meaningless for A > 1 but in compliance with the
physical “reality” of Figure 4.7.2, we may extend it to

Ws = Fy(A) = yest Oe 1 (4.7.7)

2)0.5

This corresponds to the fact that when the maximum upward water velocity
along the symmetry line in Figure 4.7.2 exceeds wo there will be positions where
u+Wo = 0 at which the particle will stop and could be trapped forever.

192 COASTAL BOTTOM BOUNDARY LAYERS

 4.7: Particles settling through turbulence

In practice, the time limit for trapping is set by the typical lifetime of local
flow structures Tg which is called the Eulerian time scale for the turbulence.

4.7.3 Taylor’s dispersion model

Taylor (1921) derived a description for the dispersion of a cloud of particles
in turbulence. That model has since formed the basis for most research into
turbulent diffusion and into some of the more fundamental aspects of turbulence.
The main results of Taylor’s theory can be summarised as follows.
Consider the one-dimensional case of spreading in the z-direction due to
turbulent velocities w with zero mean and variance 06%, (wv, w?) =5(0; on):
The spread of a cloud of particles, originating at z=o may then be quantified
by the ensemble average zi or the expected value E{z 71 of the positions of these
particles after a certain time f¢.
Noting that

ad 4
£7?he = 2zw 4.7.
(4.7.8)

we get
t t
E{2(0} = E{2) w(t) |w(e) dr dr} (4.7.9)
oO oO

In terms of the Lagrangian autocorrelation coefficient

Pww(Ar) = +Ad)}/ow
E{w(to) w(to (4.7.10)

Equation (4.7.9) can be written as

t >1

E(2(0} = 20% J J pw —f) ddr (4.7.11)

00

The autocorrelation coefficient Pww(As) is generally expected to have the

shape shown in Figure 4.7.3. By definition the value at zero time lag is unity and
because Pww(Ay)_ is expected to be a smooth and even function, the derivative at
A;=0 should be zero, Pyw’(0) = 0.

AND SEDIMENT TRANSPORT 193

Chapter 4: The motion of suspended particles

TL Th
Figure 4.7.3: Typical shape of the Lagrangian autocorrelation coefficient.

Measurements of Pww(Ay are quite scarce, but they tend to confirm the
general picture presented above, see e g Snyder & Lumley (1971) or Sato &
Yamamoto (1987). Also, the integral

Tr = Jpw at (4.7.12)
0

normally exists and it is called the Lagrangian integral scale.

The short time scale ty is called the Lagrangian microscale and is defined by

. =) \2/pw©) (4.7.13)
In terms of these parameters, two asymptotic approximations for the spread
given by Equation (4.7.11) can be derived, which are valid for relatively short and
relatively long times respectively

194 COASTAL BOTTOM BOUNDARY LAYERS

 4.7: Particles settling through turbulence

E{7} = we for t<% (4.7.14)

E{z?} = 20 Tit for t >T;, (4.7.15)

Since, for t1< tz, the inner integral in Equation (4.7.11) is approximately equal
to t, , while for large t; , itis approximately 7, .
Considering a heavy particle with settling velocity wo we may, as a first
approximation, use the approximation ws = w—wo for its vertical velocity. The
accuracy of this approximation was discussed in Section 4.5.
Using this simple superposition law, Taylor’s dispersion model can be used
to estimate the variance of the settling velocities of the particles settling through
the turbulence in Figure 4.7.1. If the settling time is large compared to the Integral
time scale, we can use the approximation (4.7.15) and get

Var{wi} = Ef{(z—-wot)}/? = 2owTt/t (4.7.16)

Thus Taylor’s model enables us to model the variance of the measured
average velocities, wj/t from the experiment in Figure 4.7.1. However, it says
nothing about a reduction of the ensemble average settling velocity, wj due to
trapping in vortices or loitering in general.
To predict a delay, a model must in some way capture the trapping by
vortices or the loitering effect. To do this, some account must be taken of the flow
structure.
Taylor did not explicitly consider details of the flow structure because his
approach was purely Lagrangian ie, w = w(Fro, t). That is, after noting the
particle’s starting position, there is no further consideration of position. Time is the
only independent variable.

4.7.4 Dispersion with loitering effect

Taylor (1921) developed the concept of diffusion by continuous movements
in analogy with a simple, discrete random walk model where the correlation
between the velocities wj and wi+: in successive steps was given by

p = exp [-6,/7z] (4.7.17)

AND SEDIMENT TRANSPORT 195

Chapter 4: The motion of suspended particles

where 6; is the time step.

The model which is developed in the following, is designed to predict the
loitering effect for non-neutral particles. The necessary difference from Taylor’s
model, which enables this, is that the correlation between velocities in successive
steps must depend on the magnitude luj-1! of the instantaneous velocity.
The correlation formula (4.7.17) corresponds to the relation

Elsie Ege
Or} 8 = Tot (4.7.18)
In order to capture the loitering effect we must consider the velocity derivative in
its Eulerian form

= DAR. pennies 400: (4.7.19)

and then in analogy with Equation (4.7.18) write

E{\8.l} = eiow +| Wat

ow. Yay
dw + ow Js =_ ow

(4.7.20)
for the unconditional expected value of the absolute velocity increment.
Wishing to model the loitering effect, the essence of which is that small
velocities correspond to slow changes of velocity, see Figure 4.7.2, we consider the
conditional expected value of [dy

E{ ldwl | uj-1} =e iG + Uj-1 oeVin} 3 + Wi-1 oe }|Uj-1,Vi-1, wi-t} dr

(4.7.21)
The value of this expression will depend on the possible correlations between the
Me ow dw ow
velocity components (u,v,w) and the partial derivatives wer
ax oy’ Oz
However, knowing nothing about these correlations it seems coe ire as the
simplest option, to assume zero correlation so that

196 COASTAL BOTTOM BOUNDARY LAYERS

 4.7: Particles settling through turbulence

Ef \8wi jui-i} =
(4.7.22)

5 [Er2sao + Win ELSMy» +r 62bl + win (2 }) eee

After applying the Eulerian equivalents of (4.7.18):

E(\SA}
ow
= >
ns Ow
(4.7.23)
and

OW, OW, . OW... Ow

E(is) = E(x) = etsy = 7 (4.7.24)
we find

E(\5yll ua} = 1 + ABLCY + Ow

eC”Ow
4725)
where the parameter Ag which measures the time scale against the spatial scale of
the turbulence is given by

Ow TE
AE = iF (4.7.26)

Based on Equation (4.7.25), we see that the replacement for Taylor’s constant
correlation

p = exp[-d/Tz) (4.7.17)

is (under the assumption 4.7.22)

Uj-112 Vi-1 2, Wi-1,2 0.5 |

pin = exp{- Ae e i oogs
eis cas Ee 9) (4.7.27)
Ow Ow

For a non-neutral particle with still water settling velocity wo which moves
in accordance with ws = w - Wo, the analogous expression is

AND SEDIMENT TRANSPORT 197

Chapter 4: The motion of suspended particles

Ui-1,2 , Vi-1,\2 , Wi-1-Wo 2 0.5

pi- = exp - FCI + Az[( a, +( am ine v1 y | 14.728)

The last term in this expression generates the loitering effect because it will be
small if wo is nearly balanced by w;-1 and hence j-; will then be closer to
unity.
The loitering effect expressed by Equation (4.7.28) is much weaker for three
dimensional flow than it would be for purely vertical flow. This is because of the
first two terms inside the square bracket whose randomness will smooth things out.
However, the loitering effect may still be significant.
Figure 4.7.4 shows results of a numerical simulation for a particle with still
water settling velocity wo settling through three-dimensional turbulence with
u’ =v" =w’ =o". The model is described by

0.8

0.6

0.4

® Wo = 1 cm/s
0.2 | X Wo= 2cm/s
© Wo = 3cm/s
A Wo = 4cm/s
0.1 02 04 1.0 2.0 4.0 10

F igure 4.7.4: Measured settling velocities from grid turbulence by Murray (1970)
and simulated values based on the random walk model described by Equations
(4.7.28), (4.7.29), and (4.7.30) with Ag = J.

198 COASTAL BOTTOM BOUNDARY LAYERS

 4.7: Particles settling through turbulence

Wi = Pri uer + OVI-phi Cui

Vi = Piven + OVI-pr Gy

Wi = Pi-1wi-l + OVI — ph Cw,i (4.7.29)

and

Ws,i = Wi-Wo (4.7.30)

where (Cu,Cv,Gw) are independent, normalised, normal variates and pj-1 is

calculated from Equation (4.7.28).
The model predicts a settling velocity reduction which depends on the scale
ratio Ag , as well as on the relative strength o/w, of the turbulence.
For Ag21 and o2w, the model predicts a settling velocity reduction of
about 20%, which is generally less than what was measured by Murray (1970).
The fact that Murray measured greater delays is probably due to his
turbulence containing an element of organised vortices rather than being totally
random, and thus creating the possibility of occasional trapping. This is likely to be
the case for most types of natural turbulence.
Murray suggested that the observed delays might be due to non-linearity of
the drag force, but he also showed in his Figure 5, that with the typical eddy radius
assumed equal to the grid bar diameter (/cm), a frequency of 25-40Hz would be
required in order to generate settling delays of the order 40%.
Such high frequencies seem incompatible with the recorded rms-velocities
which were only a few centimetres per second.
The analytical expression (4.5.33) for the non-linear drag ettect gives for R
= Icm and @=Omax/[lcm] =3.4 rad/s, a reduction of only 10° SW: Thus the
effect of possible non-linear drag is insignificant.
In the special case of Ag =O which corresponds to LE — -, the model
described above becomes identical to Taylor’s model, and there is no delay. We
may call this case the strongly homogeneous case because it corresponds to a
situation where the turbulence changes in step everywhere so to speak. The
magnitude of the velocity increment is then independant of the instantaneous
velocity magnitude. (Taylor showed that in order for the turbulence to be a
stationary stochastic process, the velocities and the velocity increments must be
uncorrelated: p(w jeey = 0 , but this does not automatically mean that the

AND SEDIMENT TRANSPORT 199

Chapter 4: The motion of suspended particles

absolute values must be uncorrelated).

On the other hand when Ag > 0 _ there is a positive correlation between
luj-7| and luj-uj-7| in the model above. For Ag=/ we find

p(luj-il,lwi-wi-ul) = 0.13 > O (4.7.31)

The actual value of this correlation coefficient for various types of natural
turbulence is presently unknown.
This means that the validity of the assumption (4.7.22) is basically unknown.
If the correlation turned out to be zero for some flow, it would mean that there was
no settling delay due to loitering in this flow.
If the correlation turned out to be negative, a random turbulence field would
tend to enhance the settling velocity.

200 COASTAL BOTTOM BOUNDARY LAYERS

 CHAPTER 5

SEDIMENT SUSPENSIONS

5.1 INTRODUCTION

5.1.1 Sediment concentrations and the transport rate

A considerable portion of the sediment transport in coastal areas is due to
sediment which moves in suspension. It is therefore necessary to develop models
for sediment suspensions so that concentration distributions can be calculated.
These may in turn be combined with the sediment velocity distributions to give the
transport rate. The sediment transport rate is usually given in the form

OOg= Jcan Us(Z,t) dz (5.1.1)

O

for the x-direction and correspondingly for the y-direction.

The suspended sediment concentration c(z,t) is, in the following, a
dimensionless quantity calculated as the volume of solid sediment divided by the
total volume of water-sediment mixture. The horizontal velocity us(z,t) of a
sediment particle can generally be assumed equal to the horizontal velocity of the
immediately surrounding fluid, see Section 4.5.5, page 178.
The sediment concentration being dimensionless, the transport rate per unit
width Q(t), defined by Eguation (5.1.1), has the dimension of L’T! and the
corresponding SI units of m*/s.

5.1.2. Mixing length and sediment diffusivity

Since the sediment is heavier than water, it has a natural tendency to sink and
settle out unless a compensating upward sediment flux is somehow created to

AND SEDIMENT TRANSPORT 201

Chapter 5: Sediment suspensions

balance the settling rate cws. The vertical sediment velocity ws is practically
equal to the local, vertical fluid velocity minus the still water settling velocity,
Ws ~ W— Wo, See Section 4.5.5, page 178.
Stationary sediment suspensions do exist. Hence, the upward fluxes which
balance settling must be present. They occur in turbulent flows because water
parcels which travel upward through a given plane generally contain larger
sediment concentrations than water parcels which travel downward through the
same plane. In other words, the upward sediment flux required to balance the
settling rate in a steady situation, is generated by vertical mixing. This type of
process is called gradient diffusion.
Random vertical mixing can generate an upward sediment flux provided the
average sediment concentration decreases with height above the bed.
To obtain a quantitative description of gradient diffusion, consider the simple
mixing process which is illustrated by Figure 5.1.1.

VM MMI

NS BSS Ss

SUL Sd hoe
Lm process of gradient
diffusion may be
quantified in terms of
dz { IN QS \ ekJae exchange

Assume that the sediment concentration varies linearly with z and that
mixing is provided by exchanging layers of thickness dz and separation Jm once
every ¢ seconds.
Since the concentration difference between such a pair of exchanging layers
; dc
is lm aE each exchange corresponds to transport of the volume — dx dy dz lm dc
dz
upward through an area dx dy of an intermediate plane, say z = zo . The minus is

202 COASTAL BOTTOM BOUNDARY LAYERS

 5.1: Introduction to sediment suspensions

elie mo 92 ,
due to the fact that, if a positive, the exchange results in a net downward

: dew: ;
transport of sediment. In most natural cases, a negative and the exchange
process provides the upward transport required to balance settling.
Summation of the contributions from all exchanges which intersect the plane
Z = Zo, gives a total transport of — dx dy Im = through the area dx dy.
With the exchanges occurring at the rate of one every ft seconds, the vertical
transport rate per unit area is therefore

lm de
Epes rae 51.2)

This is generally written in terms of a sediment diffusivity e¢s as

dc
qz = —&5—
es (5.1.3)
el

corresponding to
2
Es = bm (5.1.4)

Thus, sediment diffusivity has the dimension of Lee like viscosity, and the
SI units are m/s.
The equation which expresses a time-averaged balance between settling and
the diffusive flux, given by Equation (5.1.3), reads

WoC oF 3, & = ( (5.1.5)

dz

To obtain values of the sediment diffusivity from measured concentrations,

under the assumption of a pure diffusion process, Equation (5.1.5) may be
alternatively written as

(5.1.6)

Gradient diffusion can be used to describe the upward sediment flux if the
mixing length Jm is small compared to the overall height of the concentration
profile. However, if the two lengths are of the same order of magnitude, a different

AND SEDIMENT TRANSPORT 203

Chapter 5: Sediment suspensions

model must be applied. See the discussion in Section 5.4.2, page 233 for details.
Nevertheless, gradient diffusion has been used exclusively for the modelling
of suspended sediment distributions since the Nineteen Thirties without critical
assessment of its validity as a description of the actual physical process. Worrying
experimental data like those of Coleman (1970) and of Nielsen (1983) have
generally been bypassed by ad hoc modifications to the sediment diffusivity, see
Figure 5.1.2.
Coleman’s data show that, in order to model the distribution of suspended
sediment in steady, open channel flows as pure gradient diffusion, the sediment
diffusivity €s must generally be of a very different magnitude than the eddy
viscosity Vr, (Vt =~ Kuxz(1-z/p) for z<D/2). Furthermore, the magnitude of
€s must be a strongly increasing function of the relative settling velocity wo/ux.

/U.D
Es
° 5
e a
@
tc) oO
® B
4 8
A +
v x

(0) 0-1 02 03 O4 O5 O6 07 O8 O9 1:0

z/D

Figure 5.1.2: Sediment diffusivities derived through Equation (5.1.6) from measured
concentration profiles c(z) under the assumption of pure gradient diffusion.
Different sediment sizes (different values of the relative settling velocity wo/u)
give very different values for the diffusivity €s. Data from Coleman (1970).

204 COASTAL BOTTOM BOUNDARY LAYERS

 5.1: Introduction to sediment suspensions

The data of Nielsen (1983) and McFetridge & Nielsen (1985) from
oscillatory flow over ripples (Figures 5.2.11 and 5.2.12, p 222-223) give further
evidence that pure gradient diffusion is inadequate as a model of suspended
sediment distributions. In order to explain their measurements, the magnitude of €s
would have to be different for different grain sizes, as for Coleman’s
measurements. More importantly, however, the distribution of €s(z) would have
to be significantly different for different sand sizes in the same flow.
This would be very unsatisfactory. Using gradient diffusion as a model
makes sense only if the same diffusivity can be used for all particles and if this
diffusivity is closely related to the diffusivity of momentum, i e, the eddy viscosity.
In recognition of the increasing weight of experimental evidence against pure
gradient diffusion as the sole distribution mechanism, Section 5.4 is devoted to a
new modelling framework which acknowledges the presence of large scale or
convective mixing mechanisms.
The new, combined convection-diffusion model is discussed qualitatively in
general terms and developed quantitatively for the special case of pure,
non-breaking waves over rippled beds. For that case, the new model leads to
improved agreement with observations, but the details of the model outlined in
Section 5.4.8 should not be seen as definitive. As more detailed data become
available, the details of the model should be adjusted as appropriate.
The main point at present is, that a combined convection-diffusion model
with the general features outlined in Section 5.4.5 has the potential of vastly
improved modelling of sediment suspensions, including aspects which could not
possibly be accounted for by pure gradient diffusion.
It is hoped that the combined convection-diffusion approach will make it
possible to reconcile the concepts of sediment diffusivity and eddy viscosity, so
that €s ~ v;. This is not possible under the assumption of pure gradient diffusion,
as illustrated by Coleman’s data in Figure 5.1.2.
Before the development of the new model, an illustrated introduction to
suspended sediment distributions in coastal areas is given in Section 5.2, page 206.

5.1.3 The bottom boundary condition

The description of the physical processes at the bottom end of the suspended
sediment distribution presents many difficult problems. In mathematical terms that
is. The specification of the bottom boundary condition for the differential
equations, which describe the distribution of suspended sediment, present no lesser
problems than the formulation of the equations themselves.

AND SEDIMENT TRANSPORT 205

Chapter 5: Sediment suspensions

For flows that are both steady and uniform, it is reasonable to assume
instantaneous equilibrium between near-bed sediment concentrations and the flow
conditions (represented for example by the Shields parameter). However, this is
not possible if the flow is unsteady or non-uniform. In unsteady or non-uniform
flows, the near-bed concentrations are often considerable when the instantaneous
bed shear stress is zero because sand is arriving from above.
The alternative approach, which will be discussed in detail in Section 5.3
(page 222), is to consider entrainment and deposition separately and to try and
specify the pickup rate p(t) in terms of the instantaneous, local flow parameters. It
is acknowledged, however, that a complete description of the sediment pickup
process under waves may not become available for some time.

5.2 THE NATURE OF SEDIMENT SUSPENSIONS

5.2.1 What is suspended sediment

In Section 2.3 the bed-load was defined as that part of the total sediment load
which is supported by intergranular forces while the suspended load is carried by
fluid drag. The same criterion will be used here to distinguish suspended load from
bed-load.
The wash load is considered as part of the suspended load. It consists of fine
sediment which is not in equilibrium with sediment in the local bed because the
capacity of the flow to entrain these fine particles exceeds their settling rate at the
bed. Thus, wash load is, in a sense, similar to water vapour over a dry or drying
surface. Wash load sediment may be carried to the coastal environment by rivers,
but the general conditions on beaches are such that the wash load will move
offshore and get deposited there. Hence, beaches away from river entrances are
generally free from wash load. The water looks clear apart from distinct clouds of
suspended sand.
At elevations of more than a centimetre or so above the bed, the suspended
concentration equals the total sediment concentration and is therefore directly
measurable. Close to the bed however, where a considerable part of the moving
sediment is bed-load, it is difficult to get information about the suspended part of
the sediment concentrations.
To obtain suspended sediment concentrations very close to the bed is a
challenge to the present measuring techniques and is, to some extent, a matter of
definition. The measurable quantity is the total concentration which, with

206 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

Bagnold’s (1956) definitions, may be partly bed-load and partly suspended load
near the bed. Thus, after measuring the total concentrations near the bed, the
problem of separating bed-load and suspended load remains.
As an example, this separation will be attempted with the total concentration
measurements by Horikawa et al (1982), using Bagnold’s definitions.
Bagnold’s definition of bed-load can be written as
co

p(s-1) g Jeoler) da = oel2) (5.2.1)

Z

where Ge(z) is the dispersive stress which supports the bed-load that is travelling
above the level z.
Solving this equation with respect to the bed-load concentration gives

-1 doe
(5.2.2)
sihealed <6 oe
Consider Case 1-1 of Horikawa et al (1982) at the phase 90 degrees which
corresponds to maximum free stream velocity. The general experimental
conditions are described by (A, 7, d,s) = (0.72m, 3.6s, 0.2mm, 2.65).
The linear sediment concentration A, derived from the total, volumetric
concentration in accordance with the definition (2.1.1) page 96, is of the order
unity, and the velocity gradient is of the order / 00s close to the bed. Hence, the
Bagnold number, defined by Equation (2.1.2), page 97, is about 20. Therefore, the
relevant formula for the dispersive stress according to Bagnold (1954) is

Oe ae. lol (14+Ay(142/2) pv (5.2.3)

See Section 2.1.2 pp 95-98.

By applying Equations (5.2.2) and (5.2.3) to the measured values of A and
du/dz we can estimate the bed-load concentration cp . Subtracting this from the
total concentration then gives an estimate of the suspended concentration (cs =
Ctotal - Cb) . The distributions of total, bed-load and suspended concentrations thus
derived are shown in Figure 5.2.1.
It can be seen that the suspended concentration is indistinguishable from the
total concentration for z>5 mm _ and, that the distribution of cs(z,t) is nearly
exponential in the range 5mm < z < 25mm. The nominal value cs(o) obtained by
extrapolating the exponential part of the distribution to the level of the undisturbed
bed (z=o) is of the order 0.]cmax .

AND SEDIMENT TRANSPORT 207

Chapter 5: Sediment suspensions

Between the outer, "exponential region" of pure suspension and the level of
the undisturbed bed, the concentration gradients are considerably greater. Through
this layer, which for these experiments has a thickness of five millimetres, the total
concentrations drop by a factor 100.

e Measured total
x Estimated bed-load
O Suspended

x
x

UNDISTURBED

0.001 0.01 0.1 1.0

Figure 5.2.1; Measured values of the total, relative sediment concentrations and
estimated values of the contributions cz(z,t) and cs(z,t) in oscillatory sheet flow.
Data from Horikawa et al (1982), Case 1-1.

In the moving layer below the level of the undisturbed bed, the
concentrations decrease rather slowly with elevation, from cmax in the stationary
bed to between 50% and 80% of cmax at the level of the undisturbed bed.

5.2.2 Concentration measurements

Many different devices are now available for measuring suspended sediment
concentrations.
The oldest and simplest type of device is the suction sampler. These have the

208 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

advantage of being simple and reliable and their output is easy to interpret. The
concentrations, determined from drying and weighing the captured samples, are
independent of such factors as grain size, shape, colour and grading. These factors
influence the outputs of other devices to varying extents.
When using suction samplers, consideration must be given to nozzle
orientation and intake velocity because these may influence the amount of sand
that gets captured. If the suction speed is low compared to the ambient flow and to
the sediment settling velocity, and if the grains have to turn sharp corners to get in,
the larger grains will tend to escape.
For steady flow the criterion is simple. The intake velocity should match the
flow velocity with respect to both direction and magnitude. This is of course not
possible in a wave motion so compromises must be made and corrections may be
required.
Bosman et al (1987) investigated the efficiency of suction samplers in
oscillatory flows. They found that for nozzles perpendicular to the main flow
oscillation the capture rate increased with increasing intake velocity up to
Uintake ~ A @. For higher intake velocities, the capture rate was fairly constant at
around 0.8.
In order to capture instantaneous suspended sediment concentrations, many
other types of concentration gauges have been developed over recent years. Most
of these rely on the absorption or back-scatter of sound (Jansen, 1978), or
electromagnetic radiation (Sleath 1982, Downing et al 1981). The output of these
types of instruments depends strongly on such sediment characteristics as grain
size, shape, grading and colour which vary with both location and height. Precise,
quantitative information is therefore difficult to derive. Nevertheless, even
qualitative information such as correlations between velocities and concentrations
is valuable at the present state of the art.
The complications mentioned above are less severe for devices which are
based on X-ray or y-ray absorption. The absorption of these types of radiation
depends almost exclusively on the density of the absorbing medium.
Also the dependence of the electric conductivity on sediment concentration
has been used to measure sediment concentrations, see e g Horikawa et al 1982.

5.2.3 Concentration time series

The concentration of suspended sediment is in general a function of all three
space coordinates and of time c = c(x,y,z,t), and the amount of detail involved in a
complete description is overwhelming. It is therefore appropriate to start with less

AND SEDIMENT TRANSPORT 209

Chapter 5: Sediment suspensions

detailed descriptions.
Theoretical considerations often deal with the horizontally averaged quantity
c(z,t). It is, however, important to acknowledge the difference between this
theoretical quantity and conveniently measurable quantities such as the average
concentration along a sensor beam or over a small measuring volume. While the
time series of the real horizontal average c(z,t) may be fairly well behaved, the
time series of the measured quantities may look very noisy due to the "spotted
carpet effect". That is, since natural sediment concentrations are horizontally
non-uniform, the "carpet of concentrations" at a certain level c(x,y,zo,t) is spotted
and these spots are convected past. the sensor. This may result in much high
frequency variation which is irrelevant to the modelling of the horizontal average
c(z,t).
Figure 5.2.2 shows suspended sediment concentrations under irregular waves
measured with y-absorption along a 30cm shore parallel path. The bed was
covered by rounded megaripples which do not shed vortices regularly as vortex
ripples do. The measured concentrations seem random to a great extent although
some correlation obviously exists with the measured near-bed velocities.

u (0.20,t) [m/s]

c (0.04,t) BY VOLUME

0 1 2 3 4 5 6an
MINUTES
LER ENNEN eee ceemeeeeeeee eS eS
0 102030 SECONDS

Figure 5.2.2: Field measurements of suspended sediment concentrations 4cm above

irregular megaripples. During the first three minutes, several concentration events
occurred. During the last three minutes, the concentrations dropped to the
“background level” even though the wave conditions were essentially unchanged.
The difference is due to subtle changes in the bed topography in the vicinity of the
probe. Data from Nielsen (1984).

210 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

Sometimes, a certain area of a sand bed with megaripples will generate large
suspended sediment concentrations for a given set of flow conditions. At other
times, it may not.
Thus, individual entrainment events are not entirely predictable in terms of
the free stream velocity history. They result from complicated flow instabilities
which also depend on subtle details of the micro-topography.
Nadaoka et al (1988) described an important mechanism for the creation of
large clouds of suspended sediment in the outer surf zone, where the bed is
commonly covered by megaripples. They attributed the formation of these clouds
to obliquely descending vortices which are able to lift considerable amounts of
sand into suspension after they reach the bed. Some of the large concentration
peaks in Figure 5.2.2 may be associated with this type of sediment clouds.
The concentration variations in Figure 5.2.2 result not only from clouds of
sand rising from the bed or settling from above, but also from clouds being
convected horizontally past the probe.
More details about concentration time series under irregular waves are given
by, for example, Hanes & Huntley (1986) and Vincent & Green (1990).
The picture is somewhat less confused if the wave motion is regular,
especially under sheet-flow where the conditions are essentially constant in a

Figure 5.2.3: Phase-averaged

sediment concentrations ¢c+C
over a flat bed under regular,
symmetric oscillatory flow. One
peak occurs each half-period at
all levels. Near the bed, it occurs
shortly before the peak of the
free stream velocity (w@t=1/2).
Data from Horikawa et al
(1982).

AND SEDIMENT TRANSPORT 211

Chapter 5; Sediment suspensions

horizontal plane, c ~ c(z,t). Such conditions were investigated by Horikawa et al

(1982) and Staub et al (1984). An example of the smoothed, phase-averaged
behaviour is shown in Figure 5.2.3. Pronounced concentration peaks are present at
all levels. Near the bed they occur fairly shortly before the maximum free stream
velocity (at wt = 7/2). At higher elevations they occur progressively later.
Nakato et al (1977) and Sleath (1982) investigated c(x,z,t) over vortex
ripples under regular, oscillatory flow conditions. Over vortex ripples the
concentration distribution is dominated by dense, travelling sediment clouds.
These are formed when sand is swept over the ripple crest and into the lee vortex.
Subsequently, they are transported in a fairly coherent form by the lee vortices
which are released at the time of free stream reversal. See page 160.
At each ripple crest there will be two vortices released each period but the
phase-averaged concentration records at most levels show more than two peaks.
See Figure 5.2.4.

— @—
C+C —— _z=1.5mm
<<a -- z=14mm
—-— z=26mm

2x10°

Figure 5.24: Suspended

sediment concentrations over
vortex ripples in a symmetric,
regular oscillatory flow. In
contrast to the situation over a
flat bed (Figure 5.2.3), the
phase-averaged concentrations
show more than two peaks per
period. Data from Nakato et al
(1977).

22, COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

This is because the sand clouds remain identifiable for a considerable part of
a wave period. Hence, other sand clouds than the latest one generated at the nearest
ripple are also making an organised imprint. In some cases a sensor may also see
the same sediment cloud twice.
When the sand bed has high-profile features like vortex ripples, the
suspended sediment concentrations cannot be expected to be horizontally uniform
and the velocity field will likewise be non-uniform in a horizontal plane.
Consequently, the time-averaged vertical flux of sediment wsc_ will vary
greatly in a horizontal plane. In fact, the upward transport of sediment over vortex
ripples occurs through quite narrow spatial corridors which originate near the
ripple crest and follow the typical upward paths of the released lee vortices, see
Figure 5.2.5.
The existence of special corridors for upward sediment movement is very
obvious from the observations of Nakato et al (1977). They found that the time-
averaged, vertical sediment transport rate cCws was clearly different from zero
almost everywhere in the two vertical sections which they studied. One above the

& L

<<
<—
~<—
<—
<<
>
«ee
<—

Figure 5.2.5: The upward transport of sand over vortex ripples is very organised.
The sand is carried upwards by the lee vortices which are released at the free stream
velocity reversals. The arrows in this figure indicate the local, time-averaged, vertical
sediment flux c Ws.

AND SEDIMENT TRANSPORT 213

Chapter 5; Sediment suspensions

ripple crest and one above the trough. In general, they found that wsc was
negative everywhere in their two vertical sections except very close to the ripple
crest, see Figure 5.2.5. At the same time, the stationarity of the situation obviously
requires zero net vertical transport on the average (averaged over time and over
any horizontal plane).
Nakato et al (1977) estimated the average upward sediment flux as
(w-—wo)¢+wc in accordance with the assumption that the sediment velocity
equals the fluid velocity plus the still water settling velocity of the sediment
Us =U +Wo. The limitations of this assumption are insignificant for the present
purpose, see Sections 4.5 and 4.6 in general and Equation (4.5.26) page 176 for a
simple estimate.

ie)

nm

ELEVATION
CENTIMETRES
IN

4 6 10-5 2 4 6
Figure 5.2.6: Vertical distributions of the time-averaged concentration ¢ and the
root-mean-square of the periodic Crms. The values are obtained by averaging the
values given by Nakato et al (1977) for two vertical sections, one over the ripple crest
and one over the trough.

214 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

In relation to the discussion of sediment distribution models (Section 5.4,

page 233) it is of interest to compare the decay with increasing elevation of the
steady concentration component c(z) and of the periodic component C(z,t). In
purely convective descriptions, the decay rates are essentially the same. Diffusion
models will, however, predict significantly faster decay for the oscillatory
component, compare Equation (5.4.21) page 240 and Equation (5.4.38) page 244.
Figure 5.2.6 shows a comparison of these decay rates based on the average of
the crest-section values and the trough-section values of ¢ and Cyms from Nakato
et al (1977), Test 2. It can be seen that the decay rates are essentially identical in
agreement with the obviously convective nature of the sediment entrainment
process over vortex ripples.

5.2.4 Time-averaged sediment concentrations

It is comparatively easy to measure time-averaged sediment concentrations at
"a point" for example with a simple suction sampler. Consequently, there is a
substantial amount of experimental data available for c(x,z).
Given a flat sand bed, the time-averaged sediment concentrations should not
vary much in a horizontal plane.
For rippled beds, however, there are subtle differences between the
time-averaged concentration profiles above the ripple crests and over the troughs.
Nielsen (1979) found that the crest sections had larger concentrations close to the
bed while the trough sections had larger concentrations higher up. As a result, the
total amount of suspended material in each vertical section was fairly constant.
Figure 5.2.7 shows four different c(z)-profiles measured over vortex ripples
in a large oscillating water tunnel.
This sequence of four c(z)-profiles with fixed sediment size show the typical
change of profile shape with increasing wave period. That is, for the shortest wave
period, the profile is upward convex.
For the two intermediate periods, the profile is practically linear over the first
4 ripple heights (0 < z/n < 4). Hence, this major part of the profile is appropriately
described by the simple exponential relationship

C2yranCge k's (5.2.4)

For the longest wave period, the c(z)-profile is upward concave from the start.
A similar sequence of profiles is seen in Figure 5.2.8, for time-averaged
concentrations over flat beds.

AND SEDIMENT TRANSPORT 215

Chapter 5: Sediment suspensions

10
ae:
gv
aot
3 7(3)
E
S6
5
5
3
bea
5
ue
jam

2
1

E05 0.0001 0.001 0.01 0.1

Averaged concentration

Figure 5.2.7: Time-averaged sediment concentrations c(z) measured over vortex

ripples in an oscillating water tunnel. In all cases the sand size was 0.2mm. Legend:
square : T=1s, AW =0.5mls; +:T = 2s, A® =0.5m/s; * : T=4s, A®=0.3m/s; x:
T=10s, A@ = 0.3m/s. Data from Bosman (1982) and Delft Hydraulics (1989).

A similar, gradual change in concentration profile shape with increasing T

for fixed d, which is seen in Figures 5.2.7 and 5.2.8 is also seen in Figure 5.2.12.
However, in Figure 5.2.12, where different sand sizes in the same flow are
considered, the wave period is obviously fixed and the change from upward
convex to upward concave happens with increasing settling velocity.
These profile changes are predicted by the combined convection-diffusion
model developed in Section 5.4.5, page 245. The dimensionless parameter which
determines the profile shape is found to be the dimensionless settling velocity
Wo L/€s , where Wo is the sediment settling velocity, L is the vertical scale of the
convective mixing process, and €s is the sediment diffusivity.
The dimensionless settling wo L/es velocity is found to be roughly
proportional to wo T/ VAr , where r is the bed roughness, see page 254 for

216 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

Figure ~ 5.2.8: C(z)-

profiles over flat beds in
an oscillating water
tunnel. While (A,d) =
(0.72m, 0.2mm) in all
tests, the period was
varied from 3.6s to 6s. As
the wave period increases,
the profiles become flatter
and change from being
upward convex to being
upward concave. Data
from Horikawa et al
(1982).

details. Since the shape parameter wo T/NAr is proportional to both wo and T,

the same change in the shape of the c(z)-profile is observed when the settling
velocity is increased and when the wave period is increased while the other
parameters are kept constant. See Figures 5.2.7, 5.2.8 and 5.2.12.
The simple exponential model

Gz) = Coe" (5.2.4)

which is a reasonable description for c(z) over rippled beds within a fairly wide,
intermediate range of the shape parameter w ,L/€s, has been studied in
considerable detail by Nielsen (1979, 1984, 1986, 1990). The reference
concentration Co [ = c(o)] can be estimated from Equation (5.3.10), page 228, and
the vertical scale Ls is closely related to the ripple height n for sharp crested
ripples. Nielsen (1990) suggested

iene Kt ee (5.2.5)

AND SEDIMENT TRANSPORT 217

Chapter 5: Sediment suspensions

5.2.5 Time-averaged sediment concentrations under irregular waves

Wave irregularity influences the magnitude and distribution of suspended
sediment concentrations. For irregular waves with parameters (Tp, Hrms) the
concentration magnitude will decrease more rapidly with elevation than under
regular waves with T = Tp and H = Hrms , see Figure 5.2.9.

35

30

25

20

15
elevation
Local
[cm]

8.01 0.1 1 .
Averaged concentration [g/l]

Figure 5.2.9: Time-averaged concentration profiles from regular and irregular

oscillatory flows with the same d50 , Tp and Ucorms , data from Delft Hydraulics
(1989), oscillating water tunnel.

The main reason for the difference in concentration profiles is the difference
in ripple size. Thus, for the two cases shown in Figure 5.2.9, the regular flow
formed ripples with an average height of /5cm, while the ripples in the irregular
flow had an average height of only 4cm.
As a consequence of the bed roughness (~ ripple height) being smaller, the
shape parameter wo T/NArr will be larger for the irregular waves and the profile
shape therefore tends to be more upward concave.

218 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

5.2.6 Time-averaged concentration profiles under breaking waves

Wave breaking has profound and highly diverse effects on the suspended
sediment concentrations. However, except for extreme cases of plunger jets hitting
the bed, the pickup rate at the bed and hence the sediment concentrations very
close to the bed will be unaffected by the breaking. The main effect of the
turbulence from wave breaking is a vertical stretching of the sediment
concentration profiles. See Figure 5.2.10.

20

*)
160.°2 4- 10,529 “4 10°2 4 10°2 4
c(m*/m?)
Figure 5.2.10: Time-averaged concentration profiles under non-breaking (x) waves
and under spilling breakers (0) of the same height. Wave flume data from Nielsen
(1979).

With respect to the modelling of sediment concentration profiles under

breaking waves, it may be possible to treat the case of gently spilling breakers as
"simply" a case of increased diffusivity compared to the non-breaking case. For
plunging breakers however, the distribution process seems to be dominated by
large scale convection, with large amounts of sediment travelling straight to the
surface with the entrained air behind the plunging breakers. See Figure 6.5.4, page
290 and Nielsen (1984).

AND SEDIMENT TRANSPORT 219

Chapter 5: Sediment suspensions

5.2.7 Different sand sizes suspended in the same flow

It is of interest to examine concentration profiles for different sand sizes
suspended in the same flow. Similarities and differences between such profiles
provide information about the mechanisms which distribute the sediment. A couple
of such mechanisms, namely the classical gradient diffusion and a simple
convective process will be discussed in detail in Sections 5.4.2 through 5.4.4.
Two comprehensive data sets with concentration distributions of different
sand sizes in the same flow, are presently known to the writer. One set of field
data, Nielsen (1983), and one set of laboratory data, McFetridge & Nielsen (1985).
Both data sets were obtained with suction samplers over rippled beds under
non-breaking waves.
The laboratory experiment was duplicated nine times and the complete data
set for two sieve fractions (the finest and the coarsest) is shown in Figure 5.2.11.
Despite the scatter, it is quite clear that the two sieve fractions display different
trends. For the fine material, the trend is convex upward. For the coarse material it
is concave upward.

@ d=0.42mm - 0.60mm
+ d=0.06mm - 0.11mm

ELEVATION
ABOVE
(cm)
RIPPLE
CREST,
z

fe)
10° 10° 1007
CONCENTRATION BY VOLUME
Figure 5.2.1]: Concentration profiles for two sieve fractions of sand suspended in
the same flow (waves over natural ripples). The profiles are not similar. The trend
for the fine sand is convex upward while the trend for the coarse sand is concave
upward. (h, T, H, dso, A, nN) = (0.3m, 1.51s, 0.13m, 0.19mm, 0.078m, 0.01 1m).

220 COASTAL BOTTOM BOUNDARY LAYERS

 5.2: The nature of suspension profiles

The difference is significant because it shows that the entrainment process

can neither be described as purely diffusive or as purely convective. The details
behind this conclusion will be given in Sections 5.4.3 to 5.4.5, page 238 ff.
Figure 5.2.12 shows the hand-drawn trends of the data from McFetridge &
Nielsen (1985) for all six sieve fractions. The curves show a continuous transition
from the upward convex for the fine sand fractions to the upward concave for the
coarse ones. The same transition is shown by the field data of Nielsen (1983) in
Figure 5.4.7, page 256.
This transition can be explained in terms of the distribution mechanism being
partly diffusive and partly convective, see Section 5.3.4, page 245 ff. The diffusive
characteristics are displayed by the finer fractions while the coarse fractions are
more influenced by the convective mechanisms. The parameter which describes
the profile shape change is wo T/VA r, see also page 216, and Section 5.4.8, page
252.

Z(cm)

SQ
-42-0.6 0.3-0.42 0.18-0.3 0.15-0.18 0.11-0.15 0.06-0.11

10 1077 10° 10° 10° 105 10° 140° 10°

TIME AVERAGED CONCENTRATION
Figure 5.2.12: Hand-drawn concentration profiles for the data from McFetridge &
Nielsen (1985). The curves show a continuous transition from upward convex for
fine sand to upward concave for coarse sand. The numbers on the curves indicate the
grain size interval in millimetres.

Different grain sizes are not picked up in direct proportion to their abundance
in the undisturbed bed. Small particles have a greater chance of becoming
suspended than large particles. Details of this phenomenon are discussed in Section
5.3.7, page 230.

AND SEDIMENT TRANSPORT 221

Chapter 5: Sediment suspensions

5.3. PICKUP FUNCTIONS

5.3.1 Introduction
At the fundamental level, the modelling of suspended sediment transport
presents two challenges. One is to describe the mechanisms which distribute the
sand through the water column. The other is to describe the pickup (entrainment)
process at the bed. The present section deals with the latter process.
The traditional approach to sediment suspension modelling in steady,
uniform flows has been to assume equilibrium between bed shear stresses and
near-bed sediment concentrations. Under this assumption, it should be possible to
relate sediment transport rates directly to the bed shear stress. It is however, fairly
obvious that this equilibrium assumption is unrealistic if the flow is either
non-uniform or unsteady.
In an unsteady situation, where the bed shear stress varies with time, the
near-bed sediment concentrations may well be considerable at times when the bed
shear stress is zero because suspended sand is arriving from above. Similarly, for
steady but non-uniform flows, like most scour problems, there is no local
equilibrium between shear stresses and near-bed suspended sediment
concentrations.

c (x,y,z,t)
> —*

q, q,+ al dx
d (x,y,t) p (x,y,t)
ae a Sn a ee eRe ah ie cer
eee eee

CB OM YS

OeSeerie:
Sree a am Se) eer isoo
6) eecayenne
fel ew a tae UG:
eer Meer
Meee ee eee ater
OS CC6 feria Gretta:
RAC SEIS oie at he PAC OTee OSES ao
Sere
Oo
eenee ete
Shel te
SY ehOLaKaTe te Wh 6 ale te gere wiinlets OE bk Le
Shh 75 5
Sire ‘ea etas
(rsSko
STO gue
BawahaS pele ve ter gtae
tererscatc
Ouin terol lame,

Figure 5.3.1: The imbalances between local pickup rate p(x,y,t) and deposition rate
d(x,y,t) = Wo C(x,y,zr,t) through the near bed reference level z, generate the changes
in bed level.

D2? COASTAL BOTTOM BOUNDARY LAYERS

 5.3: Pickup functions

Thus, unsteady and non-uniform flows call for a different modelling

approach, capable of dealing with non-equilibrium situations. The alternative
approach, which has been suggested by Nielsen et al (1978) for oscillatory flows,
and by van Rijn (1984) for steady flows, is to consider sediment entrainment and
deposition as independent processes.
The deposition rate down through a level zy is then taken as c(x,y,Zr,t) Wo .
The reference level z; is usually taken as the bed level, corresponding to zr = 0,
but other choices are equally possible.
The instantaneous rate at which sand is picked up through the level z, is
given in terms of a pickup function p(x,y,t). The pickup function is a non-negative
function with the dimension of sediment flux, i e, sediment concentration times
velocity. Similarly, for the deposition rate d(x,y,t) = c(x,y,2zr,t) wo, see Figure
5.3.1. The SI units for both pickup functions and deposition rates are thus m/s.
The rate of morphological change quantified as the rate of bed level change
dzp/dt is related to the pickup and deposition rates and to the horizontal sediment
fluxes gx and qy through the conservation of sediment expressed by the
conservation equation

)
Sy bot d-p
hee. no. (abhlG of0ofBen+ Gay
oq
Ra I(+ y |e (5.3.1)
a)

where n is the solids volume contained in a unit volume of bed material.

5.3.2 Pickup functions in steady flows

Van Rijn (1984) performed experiments on scour rates in steady flows and
on the development of the sediment concentration profile in a situation similar to
Figure 5.3.2. From the experiments, the pickup function was determined
directly for a wide range of flow velocities and grain sizes (0.13mm< d < 1.5mm,
0.06 < @’< 1.0).
Van Rijn recommended the following formula for the pickup function in
steady flow
1.5 0.1
ag
<= aT 0.000330-8)”
0. ) ((s-lI) gd iio
(5.3.2)

AND SEDIMENT TRANSPORT 223

Chapter 5: Sediment suspensions

COO :
SSCS nar ae sity tte ®

Figure 5.3.2: Van Rijn (1984) studied pickup functions in steady flow via the scour
rates and the build up of the concentration profile over a sand bed following a fixed
concrete bed.

Van Rijn compared his experimental results to previous pickup functions

which are implicit in some earlier work on steady bed-load transport. He found
that with the exception of that of Fernandez-Luque (1974) these implicitly
assumed pickup functions did not match his data at all. The dependence of the
pickup function p(t) on the bed shear stress was generally too weak.

5.3.3. Pickup functions in unsteady flows

For the estimation of pickup functions in unsteady flows we have two options
at present. The first, is to adapt van Rijn’s expression (5.3.2) or a similar steady
flow formula. This requires some knowledge about the effective Shields parameter
6’(t) as a function of time. The other option is to infer the pickup function from
measured near-bed sediment concentrations in unsteady flows.
The time-averaged pickup function, p may be inferred from the time-
averaged near-bed sediment concentration Co by the following argument. The
averaged deposition rate is assumed to be woCo and, if the process is stationary,
the average pickup rate must equal the average deposition rate, i e,

224 COASTAL BOTTOM BOUNDARY LAYERS

 5.3: Pickup functions

P = woCo (5.3.3)
A considerable amount of information is available about Co [=c(o)] under
waves, see e g Nielsen (1986).
Alternatively, we can try to derive an expression for p(t) from van Rijn’s
formula (5.3.2) (or a similar steady-flow formula) if we assume that this
steady-flow formula can be applied instantaneously to an unsteady flow. With a
slight rearrangement van Rijn’s formula reads:

Q’(1)-0-) 1.5> (c—1)0% 9% 08

p(t) = 0.00033 aaa (a 2yo2 d for 0’ >6- (5.3.4)
(45

In order to apply this formula it is, of course, necessary to know how the
effective Shields parameter 6’(t) varies with time. Attempts to estimate p(t) for
three special cases are presented in the following sections.

5.3.4 Pickup functions for sine waves over flat beds

For sine waves over flat beds (sheet-flow), it might be justified to make the
assumption that the effective Shields parameter 0’(t) is simple harmonic with
amplitude 025 anda phase shift @z relative to the free stream velocity

0’(t) = @25 cos (MH Or) (G3:5)

and to apply the general value @,~0.05 for the critical Shields parameter. This
model for 6’(f) was found to work well for estimation of bed-load transport rates
over flat beds in Section 2.4.4, page 121.
Hence, we shall try and insert these expressions for 9’(f) and Qc into van
Rijn’s pickup formula (5.3.4) and compare the result with experimental data.
For the purpose of comparing van Rijn’s formula (5.3.4) with p-values
determined from concentration measurements via Equation (5.3.3), we note that
the last factor in Equation (5.3.4) is almost proportional to the settling velocity
within the relevant range of quartz grain sizes. Thus, the expression

AND SEDIMENT TRANSPORT Q25

Chapter 5: Sediment suspensions

na (ae ge3
Wo = yo

is within twenty percent of Gibbs et al’s (1971) expression (4.2.6) for quartz
spheres in the diameter range 0.3mm < d < 1.5mm.
Inserting these approximations for 6’(f) and wo into van Rijn’s formula
(5.3.4) we find

P(t) ~ 0.017 Wo (825 — 8c)!* |cos*(@t+

Or)! for O25>>0c¢ (5.3.6)

Then, noting that | cos at i; ae

3n we find the time-average

DP ~
uv 0.007 wo (025-0.05)!° (5.3.7)
Combined with Equation (5.3.3), this formula corresponds to

Co = 0.007 (€25-0.05)!° (5.3.8)

which is shown by Figure 5.3.3 to agree reasonably with measurements from

oscillatory sheet flow.
Thus, quasi-steady application of van Rijn’s pickup function with the
instantaneous, effective Shields parameter given by Equation (5.3.5) leads to
reasonable agreement with observed Co-values for oscillatory sheet-flow.

5.3.5 Pickup functions for irregular waves over flat beds

Not much is known at present about sediment concentrations and
corresponding pickup rates under irregular waves of arbitrary shape. However, if
the bed is flat (no sharp crested ripples), it seems reasonable to apply the quasi-
steady approach outlined in the previous section. That is, instead of the "sine wave
expression" (5.3.6) we might use

PulsWo | 0’(tn) — Oc It for @’(tn) > Qc (5.3.9)

P(tn) = {
for 0’ (tn) S Oc

226 COASTAL BOTTOM BOUNDARY LAYERS

 5.3: Pickup functions

Figure 5.33: Equation

(5.3.8) compared to data
from oscillatory sheet flow
by Horikawa et al (1982),
(J) and by Staub et al
(1984), (S).

with the instantaneous, effective Shields parameter, at time step tn estimated by

O'(tn) =
V4Coed
fr.s Arms (COSt Uco(tn+1) — Uco(tn—1)
Wp Ucoo(tn) + sinGr as ay gaat

(2.4.15)

on the basis of measured or simulated values of the free stream velocity Ueo(t). Wp
is the spectral peak angular frequency for the velocity record and
Arms = V2 (Uco)rms/Wp. The grain roughness friction factor 25 is calculated

AND SEDIMENT TRANSPORT 227

Chapter 5: Sediment suspensions

from Equation (2.2.6), page 105, and @r is the assumed phase shift of the bed
shear stress ahead of the free stream velocity at the peak frequency. Details are
given in Section 2.4.4, page 121.

5.3.6 Pickup functions for waves over vortex ripples

Over vortex ripples under waves, the shear stress variation with time is very
complicated, see Figure 1.2.3, and not well understood. It is therefore difficult to
devise a formula for the effective, instantaneous Shields parameter 0’(f) to insert
into van Rijn’s pickup function formula (5.3.4) to find p(t) for rippled beds.
However, the simple estimate (5.3.3) can still be used to find the time
average p on the basis of measured (extrapolated) Co-values.
On the basis of measured c(z)-profiles and the simple exponential profile
model

(2) = Coe 4s (5.2.4)

Nielsen (1986) suggested the formula

Co = 0.005 62 (5.3.10)
for the near-bed reference concentration, which through Equation (5.3.3)
corresponds to

D = 0.005 wo @ (5.3.11)
These formulae for Co and p apply reasonably well for both rippled and flat
beds, see Nielsen 1986. The modified effective Shields parameter 6, is given by

es 025
= ——-_ 53ah2
(l—- n/a)? ee
where the grain roughness Shields parameter 625 is calculated as described in
Section 2.2.5, page 105. The correction factor (1 —m N/A)’ is the square of the
velocity correction suggested by Du Toit & Sleath (1981) for the flow
enhancement near the crest of vortex ripples.

228 COASTAL BOTTOM BOUNDARY LAYERS

 5.3: Pickup functions

For irregular waves, Equations (5.3.10) and (5.3.11) may be used with the
regular wave height H_ replaced by Hrms_ or, correspondingly with Aw
replaced by V2 (ucc)rms in the calculation of 025.
With respect to the time dependence of p(t) over rippled beds, Nielsen
(1988) noted that most of the resulting sediment transport over vortex ripples
occurs above the ripple crest level and that the input of sand into this domain
happens virtually as instantaneous puffs at the time of free stream velocity
reversals. Thus, the pickup function must vary with time as shown qualitatively in
Figure 5.3.4. It has two distinct peaks at the times (7 and ¢“ where the free stream
velocity changes direction. These are the times when the lee vortices with their
clouds of sand move upwards into the main flow.
Quantitatively, these pickup functions may be described in terms of functions
of the form
u(t)

p(t)

E
= ~

Figure 5.3.4: The sediment pickup function p(t) under regular waves has two peaks
per wave period. For rippled beds the peaks are very sharp and correspond to the
release of the lee vortices. For flat beds (sheet-flow) the peaks are flatter and occur
closer to the extreme velocity in either direction, see Figures 5.2.3 and 5.2.4.

AND SEDIMENT TRANSPORT 229

Chapter 5: Sediment suspensions

as" a _t4) cos”” ae )

= 4 5. y*#*—__—_ OS)
Po) vi 2m © 2m W \
cos 2 t cos 2 t

where m is a positive integer, see Nielsen (1979). Alternatively, a train of delta

functions can be used

p(t) = V4 80-01% + V4 57(t-1") (5.3.14)

The periodic delta function 67(t-to) has the dimension T!. It is periodic with
period T and is further specified by

dr(t-to) = O for t# to+nT (5.3.15)

and 8 = 1s’. It has the convenient Fourier series

Ski-f)ecad + Yx 2cos 2<T (t-to) (5.3.16)

n=1

The coefficients V4 and V in Equations (5.3.13) and (5.3.14) are then seen
to represent the total amount of sand (per unit area) which is picked up in each
pickup event. They have the dimension of length.
The sum of V@ and V“ must therefore, if the process is stationary, balance
the total amount of sand which settles out in one wave period , ie

Vt neuawveiCo Tt (5.3.17)
Their relative magnitude V/V" must reflect the relative ability of the forward and
the backward velocities to entrain sand.

5.3.7 Selective pickup from graded sand beds

Natural sand beds always consist of a mixture of different sand sizes with
correspondingly different settling velocities. Hence, measured near-bed

230 COASTAL BOTTOM BOUNDARY LAYERS

 5.3: Pickup functions

concentrations and pickup rates are generally compounded by contributions from

the different grain sizes which are present in the bed.
For most practical purposes, it is sufficient to consider the bulk quantities
(total concentrations and total pickup rates), and try to model them in terms of the
median sediment parameters. However, it is sometimes of interest to consider the
concentrations and pickup rates for the individual size fractions, for example in
relation to sediment sorting across a beach profile.
Intuitively, one would expect that the finer fractions of a graded sand bed
would have a greater tendency to go into suspension than the coarser fractions.
This expectation is confirmed by the data in Figure 5.3.5.

6) 1 2 3 4 5 6
e : w,lcom/s],x :dx10f[mm]

Figure 5.3.5: Variation of the average grain size and the average settling velocity
with elevation above the ripple crest. The abrupt decrease in both quantities from z
=0 to z=0.01m indicates that most of the sorting happens in the pickup process.
After that, the size distribution remains fairly constant. Data from Nielsen (1983).

AND SEDIMENT TRANSPORT 231

Chapter 5: Sediment suspensions

Noting that the Shields parameter, see Equation (2.2.2), p 103, is proportional
tod” fora given bed shear stress. One might expect the relative pickup rate for
sand with size d to be proportional to dso/d. It might also be argued that, the
pickup rate of a sand fraction with settling velocity w should be proportional to
u,/w . Hence, the pickup rate relative to that of the median size should be
proportional to ws0/w. On the other hand, small particles are sometimes sheltered
against the eroding effect of the flow by larger grains (the armouring effect), and
this makes things more complicated.
There is precious little information available about the selective entrainment
of different sand sizes under waves. However, the data in Figure 5.3.6 indicate that
the simple formula

fraction in bed d
fraction in near—bed suspension ~ dso (5.3.18)
can be used as a rule of thumb.

6 % in bed
in nearbed suspension

Figure 5.3.6: relative

abundance of different
sediment size fractions
in the bed and in
near-bed _ suspension.
Rippled beds under
waves, Nielsen (1983)
(x), and McFetridge &
Nielsen (1985) (0).

232 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

Equation (5.3.18) corresponds to the time-averaged pickup rate for sand with
parameters (d,w) being given by

= d "
D(dw) = i * fraction in bed * p(dso) (5.3.19)

where p(dso) is given by a suitable formula for the bulk pickup rate. For example
Equation (5.3.11), page 228.

5.4 SUSPENDED SEDIMENT DISTRIBUTION MODELS

5.4.1 Introduction
In the previous section the processes of sediment pickup and deposition were
considered, i e the processes by which sediment goes from a state of rest to a state
of movement and vice versa. The present section deals with the processes which
take the sand upwards from the immediate vicinity of the bed.
These processes are under one referred to as distribution processes but two
distinct categories aie defined and aiialysed in detail. These are convective
processes and gradient diffusion.
A quantitative framework is suggested for describing convective entrainment.
The characteristic behaviour of sediment concentrations c(z,t) resulting from this
model are compared to that resulting from the gradient diffusion description.
It is shown that natural entrainment processes, in general, contain elements of
both convection and diffusion. Therefore, a combined convection-diffusion
description for the distribution of suspended sediment is developed in Sections
5.4.5 through 5.4.9.

5.4.2 Convection or diffusion

The process of sediment distribution may be considered as an orderly
convective process or as a disorganised, diffusive process or, most commonly, as a
combination of these two.
In terms of the mixing length Jm, defined in Section 5.1.2, p 201, the
distinction between convective and diffusive processes is made as follows. If the
mixing length is large compared to the overall scale of the sediment distribution,
the process is convective. Conversely, if the mixing length is small compared to

AND SEDIMENT TRANSPORT 233

Chapter 5: Sediment suspensions

CONVECTION DIFFUSION

Figure 5.4.1: In a convective process the distance of interest is travelled in a smooth,

organised way while in diffusive processes it is covered in a large number of more or
less random steps.

the overall scale, the process may be described as diffusive, see Figure 5.4.1.
For both types of processes, and indeed for combined convection-diffusion
processes, the conventional means of describing concentration profiles of
suspended sediment is the conservation equation for the volume of sediment which
may be written as

ener div (us C) (5.4.1)

which expresses that a divergence of the sediment flux field g = c us must result
in a change of the local sediment concentration.
For the sake of simplicity, we shall in the following, consider a horizontally
uniform sediment concentration field c = c(z,t) and a correspondingly uniform
sediment velocity field us = us(z,t). The conservation equation (5.4.1) is then
simplified to

Oc Ne Nias Cl pea
Be 1Ta inigyll WS) tara) RAD

234 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

where qz denotes the total sediment flux (transport rate per unit area) in the
z-direction.
It is generally considered too complicated to describe the vertical sediment
velocity ws in detail in a turbulent flow. So instead, a broader approach is
generally applied to the sediment flux. In a horizontally averaged description, the
vertical sediment flux gz is considered to consist of a downward component -woc
due to gravitational settling, and an upward flux which can be of convective
(subscript C) or diffusive (subscript D) nature or a combination of both. The total,
vertical sediment flux is thus written as

q, = ~Wolt+Gnt+ |e (5.4.3)

and hence, the conservation equation (5.4.2) is written

dc _ yy 9C _ Op _ 94e (5.4.4)
see Figure 5.4.2.

‘c i

fe}°

Figure 5.4.2: For a horizontally uniform situation, conservation of sediment is

expressed in terms of the three vertical flux components: The settling flux -woc; the
diffusive flux g,; and the convective flux q,.

AND SEDIMENT TRANSPORT 235

Chapter 5: Sediment suspensions

Diffusive sediment flux is generally described in terms of gradient diffusion

Qo = —€sVec (5.4.5)

ie, the diffusive flux vector is directed from larger towards smaller concentrations.
It is proportional to the concentration gradient vector Vc and to the sediment
diffusivity ¢€s. The diffusivity has dimension length squared per unit time (or
velocity times length), the same as kinematic viscosity and eddy viscosity which
can both be considered as diffusivities of momentum. The diffusivity may be
expressed in terms of a typical mixing length as outlined in Section 5.1.2, p 201.
The concept of gradient diffusion was developed in connection with the
kinetic theory of gasses and statistical mechanics where macroscopic distances are
covered in many random steps between which the velocity changes due to
collisions. In that case the diffusivity has a clear physical meaning. It is simply the
typical step length (the mixing length) squared divided by the typical time between
collisions. Thus, diffusion is a useful concept when the motion considered is of this
random walk nature.
However, gradient diffusion cannot describe details of a mixing process on a
smaller scale than the mixing length. Processes where the mixing length is of the
same magnitude as the overall scale of the sediment distribution can therefore not
be described in terms of gradient diffusion.
Examples of such convective processes are the entrainment of sediment
from rippled sand beds under waves by travelling vortices, and the lifting of sand
straight from the bed to the surface by the rising plumes generated behind plunging
breakers. In these processes, the sediment flux is obviously not necessarily related
to the concentration gradient as expressed by the diffusion equation (5.4.5).
For the purpose of modelling natural sediment suspension processes, which
generally contain elements of both diffusion and convection, we must first develop
a quantitative description of the convective sediment flux. To this end, the vertical
convective sediment flux is written in the form

qc(z, t) = pt- 7)F(z) (5.4.6)

where the pickup function p(t) is a non-negative function describing the
instantaneous pickup rate at the bed. The pickup function has the dimension of
sediment flux (velocity times concentration) while we is the average vertical
velocity with which the sand is convected upwards, see Figure 5.4.3.
Quantitatively, the sediment convection velocity we may be assumed similar
to the speed w; with which bursts of turbulence travel upward from the bed.

236 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

to

Figure 5.4.3: In the simple convection model considered here, a sediment particle
which is picked up from the bed at time ¢o will travel upwards with speed we until it
reaches its entrainment level ze at time to + ze/wc. After that, it is assumed to settle
with its still water settling velocity wo.

Sleath (1987) found that in oscillatory boundary layers

5
We = Wt = ee G23)

see Section 1.2.3, page 19.

The dimensionless convective distribution function F(z) in Equation (3.4.6)
determines the fraction of the entrained sand which travels (convectively) beyond
the level z. Thus, F(z) is a probability function expressing the probability that the
entrainment level ze reached by a given particle is higher than z

P{ze>z} = F(z) (5.4.7)

Hence F(z) must be a positive function with F(o) = 1] if the process is

assumed purely convective, and in general F(z)>0 for zo.
Correspondingly, the probability that a given particle is entrained to the level z is

AND SEDIMENT TRANSPORT 237

Chapter 5: Sediment suspensions

expressed in terms of the derivative F’(z) = dF/dz

P{z<ze<ztdz} = —F’(z)dz (5.4.8)

where the minus is due to the ">" in the definition (5.4.7) of F(z), which is
opposite to the standard definition of distribution functions in probability texts.

5.4.3 Suspended sediment distributions due to pure convection

While natural concentration profiles of suspended sediment will normally
result from a combined convection-diffusion process, it is instructive to consider
the case of purely convective entrainment. That is, a process in which the sediment
distribution can be described by the continuity equation (5.4.4) with zero diffusive
flux, dp = 0

a, SE (5.4.9)

and with the convective flux described in accordance with Equation (5.4.6) and
Figure 5.4.3. With the expression (5.4.6) inserted for g * Equation (5.4.9) becomes

3 = Mog ty oc ey
P't-| FO - pt-F FO
? a | ae ’
(5.4.10)

: d
where the brief notation p’ = oh and F’ = has been used for the derivatives.

The time-average of this equation, which describes the time-averaged

concentration profile c(z), is considerably simpler

dc se
Wo = PF) (5.4.11)
and is easily integrated to yield

Wo C(Z) — p F(z) = const (5.4.12)

The constant of integration is found by considering this equation at the bed level,
Z=0

238 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

Wo C(0) — F(o)p = woCo - p = n— (5.4.13)

where woCo is the time-averaged rate_of deposition , see Figure 5.3.1, page 222.
Hence, for a stationary situation with dzp/dt = 0, we get

Wo C(0) = p (5.3.3)

That is, the average pickup rate p must balance the average settling rate at the
bed.
For such a stationary situation, Equation (5.4.12) takes the form

Wo C(z) - pF(z) = 0 (5.4.14)

which, with wo c(o) substituted for p , shows that the relative concentrations at
different levels, i e the concentration profile shape, is the same for all settling
velocities, wo, and given directly by the convective distribution function F(z)

UG, Wo)” d> C(Z)e *

Fe ay (5.4.15)

Solutions to the time-dependent Equation (5.4.10) are, in general, somewhat

more complicated. However, since Equation (5.4.10) is linear, and since most
coastal sediment transport problems can be described in terms of Fourier series, it
is natural to consider the solution which corresponds to a simple harmonic pickup
function with amplitude Py, and angular frequency

prt) = Pre (5.4.16)

which generates the convective flux

dent) = Pn elt 2/"o) F(z) (5.4.17)

upward through the level z in accordance with Equation (5.4.8).
Assuming that the solution has the form

AND SEDIMENT TRANSPORT 239

Chapter 5: Sediment suspensions

ent) = Ca foe (5.4.18)

with fn(o) = 1, we find the following differential equation for fn(z)

=—WoCifrot ind Cn fa .= inine e712O2/We F(z) — Py e~1O2/We F(z)

Cc

(5.4.19)
This equation is linear with constant coefficients. It can therefore be solved
analytically for most realistic functions F(z), but it has particularly simple
solutions if the distribution function F(z) is an exponential

F@) = e7 (5.4.20)
In this case the solution is

Cni2t) =°"C, ene (5.4.21)

where
Pn 1+ in@L/we
= 5.4.22
Cn Wo 1+inol/we + inoL/wo ( )
and

Bn = Pues (5.4.23)
We

The fact that Cyn is complex with a small negative argument shows that the
concentration cy(o,f) at the bed lags behind the pickup function py, . The
imaginary part of By, makes this phase lag grow with distance from the bed. In
terms of real-valued functions, the solution (5.4.21)-(5.4.23) can be written

Ca(2f) = ICnl 27 cos[n@(t— z/wc) + Arg{Cn}] (5.4.24)

This solution, which corresponds to the pure convection process, differs from
the pure gradient diffusion solution, which will be discussed below, in two
important respects. Firstly, the magnitude of all harmonic concentration
components decay as e © since Re{Bn} = J for all m. Secondly, the time lag
zlwe-Arg{Cn}/n @® grows at the same rate with z for all frequencies.
The expression (5.4.24) shows that, when the distribution process is
convective, a concentration peak will travel upwards with the convection speed we

240 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

as expected. Hence, wc can be inferred from measurements of c(z,t). Such

measurements were made by Homma et al (1965) and Nakato et al (1977) over
rippled beds, and by Horikawa et al (1982) and Staub et al (1984) under sheet-flow
conditions.

5.4.4 Sediment concentrations due to pure gradient diffusion

For a pure diffusion process (zero convective flux), the conservation equation
(5.4.4), page 235, for suspended sediment, becomes

ac
ae
_ Mee
dc _ 9p
5 (5.4.25)

Ip = ~8555 (5.4.26)
we get

Cer
or =
Ee 27.
Wo a7 ap -e (Es a
OC (5.4.27)

For the steady concentration component c(z) Equation (5.4.27) becomes

Wo Dcuinds fice.
am SF ay (Es 7) = 0 (5.4.28)

This time-averaged equation can immediately be integrated once to yield

Wot + ate = 0 (5.4.29)

since the constant of integration is zero when there is no sediment flux at infinity.
The time-averaged Equation (5.4.29) can also be written as

feine (5.4.30)
dz Es

and is therefore seen to have the general solution

AND SEDIMENT TRANSPORT 241

Chapter 5: Sediment suspensions
Zz

Bea
C(z) = Clo)e”?Sés (5.4.31)

The constant c(o) ie, the concentration at the bed, may be given in terms of
the time-averaged bed shear stress if the flow is uniform.__
Alternatively, if the bed-level is stationary dep _ 0), c(o) can be
dt —
expressed in terms of the time averaged pickup rate p , since the deposition rate
Wo C(o) must be balanced by p. See Figure 5.3.1. This leads to the expression
Z
= dz
Cz) = a e-wo] (5.4.32)

for the time-averaged sediment concentration.

For graded sediments this shows that, in contrast to the pure convection
solution (5.4.21)-(5.4.23), the shape of the c-profile depends strongly on the
settling velocity in a pure gradient diffusion process. The concentrations of coarse
sand decay faster with elevation than the concentrations of fine sand. Still, the
profiles for different grain sizes in the same flow are similar in the sense expressed
by Equation (5.4.30). See Figure 5.4.4.

Figure 5.4.4: In a pure gradient diffusion process, the time-averaged concentration

profiles for different grain sizes are different but similar in the sense described by
Equation (5.4.30). That is, they are stretched horizontally in proportion to the settling
velocity w.

242 COASTAL BOTTOM BOUNDARY LAYERS

 5.4; Distribution models

The traditional approach, from steady flow, of prescribing the sediment

concentration at the bed level as a function of the bed shear stress, does not work
for time-dependent suspension problems, because there is no instantaneous
equilibrium between bed shear stress and suspended sediment concentration. The
concentration at the bed may well be large when the stress is zero because dense
clouds of sediment are settling from above.
The alternative is to express the upward sediment flux at the bed which, in
nates ' ee deus ; ;
the diffusion terminology is written as —€5 a? terms of a pickup function

dc
—Es 7 pm p(t) for z=0 (5.4.33)

as suggested by Nielsen et al (1978).

As for the pure convection case considered in Section 5.4.3, page 238, the
general form of the time-dependent sediment concentrations c(z,t) , in a pure
gradient diffusion process, can be investigated by considering a Fourier component

cn(z,t) = Crfalz) em (5.4.34)

of c(z,t), which is generated by the Fourier component

prt) = Pre™™ (5.4.35)

of the pickup function.
This leads to the following ordinary, linear differential equation fn(z)

lee tre
— 3_
Es Es
in = 0 (5.4.36)

which corresponds to Equation (5.4.19), page 240, for the pure convection
problem. In order to get an impression of similarities and differences between the
two descriptions, consider the special case of constant diffusivity. For constant
diffusivity, Equation (5.4.36) is reduced to

ee eae iat in® Soak wll (5.4.37)

Es Es

With the boundary condition (5.4.33) and the simple harmonic pickup function

AND SEDIMENT TRANSPORT 243

Chapter 5: Sediment suspensions

(5.4.35), Equation (5.4.37) has the solution

Pn —OnZWo/Es pinwt
cn(z,b) ———————
Woh é tena 2 (5.4.38)

where

cniéceugibl
2 4
wlan.
INWE,
Wo
(5.4.39)

The nature of the complex coefficient O, is illustrated by Figure 5.4.5. In terms

of real-valued functions, this pure gradient diffusion solution may be written as

Cr(z,t) = Pn ew Re{On}wo2/€s cos nent — Im{On} ~2

Es
z— Arg{On})
WolQnl

(5.4.40)
Im (An)

ao,
ASYMPTOTE

Figure 5.4.5: The variation of the complex coefficient o» of the pure diffusion
solution. It differs from the corresponding B, (Equation (5.4.23)) of the pure
convection solution in that Re{a,} is a function of @ and, Im{an} is not
proportional to n®. This means that the different harmonic components of the
diffusion solution decay at different rates with z, and their time lags grow at
different rates with z.

244 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

from which we see that, in contrast with the convection solution, the rate of decay
with elevation of individual Fourier components cp(z,t) increases with frequency.
In the pure convection solution (5.4.24), all Fourier components of c(z,t) decay at
the same rate.
Furthermore, the time lags of cp(z,t) relative to the pickup function grow at
different rates with z. In the pure convection solution, all the time lags grow at the
same rate, namely, z/wc. This means that in a convection-dominated process,
concentration peaks will show up with very similar shape in records from different
elevations. In a diffusion-dominated process, the peaks will become blurred more
rapidly.
The length scale L of the convection solution is replaced by €s/wo in the
pure gradient diffusion solution.

5.4.5 Sediment concentrations due to combined convection-diffusion

Natural sediment suspension processes contain elements of both convection
and diffusion. This is obvious from physical inspection of the processes. Most
sediment distribution processes involve mixing mechanisms with mixing lengths
lm (see Section 5.1.2, p 201) of quite different magnitudes. For some of the
mechanisms the typical mixing length is small compared to the overall scale of the
sediment distribution. For others the mixing length is relatively small.
When /m is small compared to the overall scale of the suspended sediment
distribution, gradient diffusion may provide a satisfactory description. However,
when /» is relatively large, a different approach is required. Recall that the term
"convective" is used in the present text for processes where the mixing length is
not small compared to the overall distribution scale.
Thus, convective sediment entrainment in rivers is exemplified by the “boils”
of sediment laden water which sometimes appear at the surface. They are
generated by travelling vortices formed in the lee of dune crests , see e g Jackson
(1976). Due to the trapping mechanism described in Section 4.6, page 181, these
vortices can carry suspended sediment directly from the bed to the surface.
In the surf zone, the air which is entrained by plunging breakers will generate
strong upward flows which can carry large amounts of suspended sand directly to
the surface, see Figure 6.5.4, page 290, and Nielsen 1984.
The travelling vortices which carry suspended sand in a fairly organised
manner over rippled beds under waves are also an example of convective
entrainment, see page 160.
At the same time it is fairly obvious that in all of these environments, small

AND SEDIMENT TRANSPORT 245

Chapter 5: Sediment suspensions

scale turbulence is present. The role of the small scale turbulence is to smooth
concentration differences by gradient diffusion.
The combined nature of the entrainment process is also evident from
measured time-averaged concentration profiles of different grain sizes in the same
flow. As discussed in Section 5.2.3, page 209, the distributions of different sand
sizes are not always similar, as they should be in any one of the pure processes, see
Equation (5.4.15) page 239 and Figure 5.4.4 page 242.
For a combined convection-diffusion process, the behaviour of the
horizontally averaged concentrations c(z,t) can be described by the conservation
equation (5.4.4) page 235 with both convective and diffusive fluxes included and
given by Equations (5.4.6) page 238, and Equation (5.4.26) page 241 respectively

dc Or ee S + oa “a vEt 0G
Finlay iis we? errr z era ah dz (Esey
(5.4.41)
By taking time-averages, this equation is reduced to

WOaide)
Pe ileal ee cies (5.4.42)
which describes the time-averaged suspended sediment concentrations in a
horizontally uniform, combined convection-diffusion process. This equation can
immediately be integrated once with respect to z

au = dc
WoC — pF(z) + es on = const (5.4.43)

where the constant of integration must be zero if all concentrations and

concentration gradients are assumed to vanish far from the bed. In that case the
equation can be written

WoC + e, 82 = p F(z) (5.4.44)

By comparing this equation with the corresponding time-averaged Equation

(5.4.29) page 241 for the pure gradient diffusion case, we see that the
homogeneous version of the combined Equation (5.4.44) is the time-averaged
diffusion equation. Hence, the pure gradient diffusion solution is always part of the
combined solution (with a suitable multiplying factor).
We shall now turn to the bottom boundary condition for the suspended
sediment distribution in the combined convection-diffusion process.

246 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

The bottom boundary condition in the combined convection-diffusion system

would logically have to be a combination of

0z
Wo (0,1) — p(t) Fo) = nSa
from the pure convection case, and

Wo C(0,t) + Es oC =n oe

from the pure gradient diffusion case. That is

dc
Wo C(0,t) — p(t) F(o) + esa, =p =
2b
LEW

The second and the third terms on the left-hand side are minus the convective
and diffusive fluxes upward from the bed. Together, these terms amount to the
total upward sediment flux from the bed, i e , to the pickup rate

C]
p(t) F(o) - €s Ee = p(t) (5.4.45)
or

~ e541, = [1-F
Thus, for a given pickup rate p(t), the concentration solution depends on the
chosen value F(0) of the convective distribution function at the bed. In the
following we shall mostly be using F(0)= 1 which through Equation (5.4.46)
corresponds

~€5 a es (5.4.47)
That is, the diffusive sediment flux is assumed to vanish at the bed. If the
diffusivity €s tends smoothly towards zero at the bed, then the diffusive flux
vanishes smoothly. However, it vanishes abruptly if €s is not tending towards zero
at the bed. Thus, the general validity of the choice of F(0) = 0 is perhaps
debatable. It is a convenient choice however, and it leads to reasonable agreement
with measurements, see Example 5.4.2, page 256.
For the time-averaged concentration we always have the simple bottom

AND SEDIMENT TRANSPORT 247

Chapter 5: Sediment suspensions

boundary condition

WoC(o)
-p =n Ai

see Figure 5.3.1, page 222 which for a stationary situation with no net erosion or
deposition becomes

Wo C(0) - p = 0 (5.3.3)

5.4.6 Time-averaged concentrations in combined convection-diffusion

The particular solution to the time-averaged conservation Equation (5.4.43),
which takes the value c(zr) at the reference level zy can be written

yak

=,c(z) .=e 5-62) |JPope

ED F(6)e7*” GO) dC + -Cc(zr) (5.4.48)

or, with c(zr) = D/Wo

ateC(z) nit PAprGOn)

me e Jlemon
a) F(Q)e" G(o) dG +1 (5.4.49)

where the function G(z) can be any integral of wo/€s(z)

Ga = feWooe (5.4.50)
as long as the same function is used both inside and outside the brackets.
The first term inside the brackets of Equation (5.4.49) grows very rapidly
with the settling velocity wo. Thus, while the pure diffusion solution, which
corresponds to the last term, may dominate for the finer sand fractions, the first
term will become dominant for the coarser sand fractions. The first term of the

248 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

solution accounts for the convective entrainment, and for vanishing diffusivity, it
tends towards the pure convection solution

C(z) = c(o) F(z) = 2 F@ (5.4.15)

This may not be immediately obvious from the form of the solution (5.4.49).
It will however become evident from the Examples 5.4.1 and 5.4.3.
The fact that the shape of the convective distribution function F(z) can be
expected to be displayed by the coarsest sand through Equation (5.4.15) is
important. It shows that information about F(z) can be extracted directly from
measured concentration distributions of the coarsest sand fractions.

Example 5.4.1: A simple case of combined convection-diffusion

The following example illustrates the roles of convection and diffusion in the
combined system with F(zr) = F(o) = 1 in general. In particular, it illustrates the
asymptotic behaviour of the combined convection-diffusion distribution model in
the limits of negligible and dominant diffusion.
Consider the time-averaged solution (5.4.49) for the simple situation where
the convective entrainment function is a simple exponential with length scale L

Fa) = et
and where the diffusivity €s is a constant. In this case, the integral (5.4.50) may be
taken as

G@ = aa

and the time-averaged solution (5.4.49) with zr=O0 reads

az) = 2 ewor/ts cE eW/Lewol/es at 4 1 )

Wo Es O

which can be rewritten as

AND SEDIMENT TRANSPORT 249

Chapter 5: Sediment suspensions

2rye D 1
gee A = sae liu = 1 ned —Wo2/E
Ea (5.4.51)
tee ESSE
WoL WoL

This concentration distribution is shown in Figure 5.4.6 for selected values of

the relative diffusivity &s/woL.
It can be seen that the solution tends towards the pure convection solution

te F(z) (5.4.15)
Co
for vanishing relative diffusivity, €s/wol — 0

elevation
relative
(z/L)

.01 0.1 4
relative concentration
Figure 5.4.6: Relative time-averaged sediment concentrations corresponding to the
combined convection-diffusion solution (5.4.49) for different values of the relative
diffusivity €s/woL. For small values of €s/wol the solution approaches the pure
convection solution (5.4.15).

250 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

5.4.7 Time-dependent concentrations in combined convection-diffusion

This section contains a comparison of the solution to the time dependent
Equation (5.4.41), page 246, for the combined convection-diffusion.process with
the cases of pure convection and pure diffusion considered in Sections 5.4.3. and
5.4.4. All of the processes have been assumed horizontally uniform for simplicity.
Consider the combined convection-diffusion equation (5.4.41) for the special
case of a constant sediment diffusivity €s and a simple exponential convective
distribution function F(z) = Reta Applying the simple harmonic pickup function
Pn(t) = Pne’”™ , and assuming that the concentration solution has the form

cn(z,t) = f(z) nm (5.4.52)

we find the following differential equation for the complex function f;,(z)

Wo in® —BnP n e Pr2/L

tn i, eeag!tn| ale
commas 3
Ameer tn = be, (5.4.53 )

where, as for the pure convection solution (p 240)

Bn = ieee (5.4.23)
We

Note that the homogeneous version of Equation (5.4.53) is identical to the

corresponding Equation (5.4.37) for the pure diffusion problem which has the
solution (5.4.38) - (5.4.39), p 244. Hence, the complete solution to Equation
(5.4.53) for combined convection-diffusion is a constant times the diffusion
solution plus a particular integral to Equation (5.4.53). It is easily verified that

f(z) = Bne Prt (5.4.54)

with
Bits 02) 1
Pn See Gs 5.4.55
Bn Wo 1 — Bats , ; nol ( )
WoL BnwWo

is a solution to Equation (5.4.53). Thus, the complete solution is

cn(z,t) = (AnenonWo2/€s 4B, @Bnt/ 7 ein (5.4.56)

where the constant A, is determined by the boundary condition (5.4.46), p 247.

AND SEDIMENT TRANSPORT 251

Chapter 5: Sediment suspensions

5.4.8 Determining €s, we and F(z) for combined convection-diffusion

In order to make use of the combined convection-diffusion model, for
predictive purposes, it is of course, necessary to be able to predict the sediment
diffusivity €s, the typical convection velocity wc, and the convective distribution
function F(z). The physical meanings of the convective process parameters Wc
and F(z) are explained in connection with Figure 5.4.3, p 237.
The prediction of all three parameters is a major task because a large number
of cases would need to be given special consideration. That is, suspended sediment
distributions under waves may well be different from those in steady flows and
wave breaking is likely to have a profound effect. It is also possible that the
unsteady components of c(z,t) correspond to a different diffusivity than c(z) just
like the steady and unsteady flow components of a combined wave-current flow
seem to feel different eddy viscosities, see Section 1.5.3.
For the special case of pure non-breaking waves it seems reasonable to
expect that €s, we and F(z) will all be closely related to the bottom boundary
layer parameters. It turns out that a simple model based on this philosophy is in fair
agreement with the available c(z)-data for flat beds as well as rippled beds.
With respect to the sediment diffusivity €s, it seems reasonable to relate it to
some eddy viscosity. The phrase "the eddy viscosity" is consciously avoided
because of the complicated nature of the eddy viscosity concept relating to
oscillatory boundary layers and combined wave-current flows. See Sections 1.2.8,
1.2.9 (pp 31-40) and Section 1.5.3 p 68.
A priori, a few expressions for €s might be suggested, e g, Equations
(1.2.43), (1.2.44) and (1.3.14). However, in view of the experimental evidence, as
discussed below, the most pragmatic choice at the moment seems to be

€s = 4v. = 40577] = 4[050(0.09VrA)*] = 0.016mrA

(5.4.57)
The factor 4 corresponds to the factor of approximately 4 between the
diffusivities of time-averaged momentum (u) and of oscillatory momentum (7)
in a wave-dominated boundary layer. See Equation (1.2.38) and Figure 1.5.6, page
70. The eddy viscosity expression applied in Equation (5.4.57) is Equation
(1.3.14), page 52. This eddy viscosity estimate should be a reasonable choice for
the large relative roughness values (0.085<r/A<1.15) which are generally
exhibited by sand beds under waves. See Sections 1.3.3, p 42, and 3.6.4, p 152.
The vertical speed of sediment convection we should be closely related to
the speed w; with which boundary layer turbulence propagates upwards in

252 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

oscillatory boundary layers. Sleath (1987) found, from purely oscillatory flow that

w 5.05
Wr = 2919)
(1.2.3)
where 605 is the boundary layer thickness which corresponds to a dimensionless
velocity defect IDI of 0.05 at the top of the boundary layer.
Assuming the simple boundary layer structure which corresponds to constant
eddy viscosity, the defect magnitude |DI decays exponentially in accordance with
Equation (1.3.7). Then Equation (1.2.3), for the convection velocity can be written

We SY) EA gts GPP (5.4.58)

DOH
where the factor "3" comes from 0.05 ~ e~? and the boundary layer length scale
z1 isfound from z; = 0.09VrA , see Equation (1.3.10), page 47.
It should also be possible to observe actual values of the sediment convection
velocity wc as the speed of upward propagating concentration peaks. Provided of
course, the convective (the second) term of the solution (5.4.56) is dominant.
Remaining to be modelled is the convective distribution function F(z).
Assuming a close link between advection of boundary layer turbulence and of
suspended sediment, it should be possible to derive information about F(z) from
measured turbulence intensity distributions. The vortex trapping described in
Section 4.6 (pp 181-189) provides the close link because it enables the trapped
sediment to travel with the turbulence (the vortices). However, the derivation of
F(z) from turbulence intensity distributions offers the extra complication of
modelling the decay of turbulence.
Therefore, a simpler and more directly empirical approach will be applied in
the following. The general nature of the sediment distribution function F(z) will
be inferred directly from the distributions of coarse, suspended sediment. In this
context, the term "coarse" corresponds to large values of the relative settling
WoL 2 ; , /
velocity, w* = 5 >> 1, where L is the vertical scale associated with F(z).
Ss

See Example 5.4.1, page 249 ff.

As discussed in relation to the time-averaged convection-diffusion solution
(5.4.48), p 248, and as indicated by Example 5.4.1, the distribution of relative
concentrations for very coarse suspended sand (w* >> 1) is essentially F(z) as
for the pure convection case.

edre ] (5.4.59)
c(o) Es

AND SEDIMENT TRANSPORT 253

Chapter 5: Sediment suspensions

Hence, considering the ¢-distribution for the coarsest sand in Figure 5.2.12
we see that, in this semi-logarithmic presentation, F(z) has an upward concave
shape which is typical of functions of the form (J +z/L)" where L is an
appropriate vertical scale. Furthermore, F(z) is seen to decrease by roughly a
factor 10° through nine centimetres of height for those experimental conditions.
It seems natural to chose L = z1, where z; is the equivalent of the laminar
Stokes length in an oscillatory boundary layer with constant eddy viscosity, see
Equation (1.3.7), page 46. For the experiments reported in Figure 5.2.12, page 221,
Z, was approximately three millimetres. Hence, the elevation of 9cm corresponds
to z/L = z/z, ~ 30 anda power of -2 _ is thus required to give the observed
concentration decrease of the order 10°? over 9cm. Hence we may suggest

F(z) = (1+2/z)° (5.4.60)

Although the derivation above is very case specific, the expressions (5.4.57)
and (5.4.60) lead to fairly reasonable predictions of c(z) when inserted into the
convection-diffusion solution (5.4.49). This is shown for large vortex ripples and
for flat beds (sheet-flow), in Examples 5.4.2 and 5.4.3 in the following section.
Nevertheless, the expressions (5.4.57) - (5.4.60) should only be seen as first
estimates of €s and F(z) and improved versions should be sought as soon as the
necessary data become available. For this purpose, c(z)-data of the type presented
in Figure 5.4.7 are essential. That is, separate concentration profiles for the
different sand fractions. No data of this type are presently available from sheet-
flow conditions or from long period waves over fairly small ripples.
Note that when the diffusivity is proportional to wz/, as in Equation
(5.4.57) and when the vertical scale L_ is proportional to z; as assumed in the
derivation of Equation (5.4.60). Then, the relative settling velocity or profile shape
parameter w" = Wol./€s becomes proportional to the wave period
W'~ Wo/(W 21) ~ Wo T/VrA as observed in Figures 5.2.7 and 5.2.8, pp 216-217.

5.4.9: Convection-diffusion modelling of c(z) undernon-breaking waves

The time-averaged concentration profiles under non-breaking waves over
both vortex ripples and flat beds can be successfully modelled by the combined
convection-diffusion model which was developed in Section 5.4.6. That is, with
appropriate choices of the time-averaged pickup function p, the sediment
diffusivity €s, and the convective distribution function F(z), the concentration
magnitude and the general profile shape are reasonably predicted.

254 COASTAL BOTTOM BOUNDARY LAYERS

 5.4: Distribution models

Quantitatively, the combined convection-diffusion model consists of the

following formulae from the previous sections:
v4

=n)C(z) = P_
ae -Gle) |©
| _Wo_ F(C)e"’ Go) do + 1 (5.4.49)

o dz
see page 248, where G(z) = lee and the reference level z; is generally
S
chosen to be the bed level, ie, zy = 0. The elevation z is measured from the
ripple crest level if the bed is rippled. The time-averaged pickup function is
predicted by

D = WoCo = 0.005 wo OP (5.3.11)

see page 228. The sediment diffusivity is given by

&s = 40.5 mz] = 4[0.5@(0.09VrA)"] =~ 0.016mrA (5.4.57)

see page 252, and the convective distribution function by

F(z) = 5.4.60
oe (1+2/z1) is
see page 254. The optimal choice of the bed roughness _r seems to be

r = 81’/d + 5025 dso (3.6.14)

(page 159). If Equation (3.6.13) is used instead, for sheet flow conditions, the
resulting values of z; = 0.09VrA seem to be too large to be used with Equation
(5.4.60).
With the expressions (5.4.57) and (5.4.60) for €s and F(z), and with the
reference level taken at the ripple crest level, zr=0, the time-averaged solution
(5.4.49) becomes
Zz
= Wol/Es
a¢(z) = Pe = WoZ/Es eta
w é
Neca dG+1) (5.4.61)
Wo Es 0 Z1

or, with the dimensionless elevation § = C/z; and the dimensionless settling
velocity w* = WoZ1/E€s

AND SEDIMENT TRANSPORT 255

Chapter 5: Sediment suspensions

+ 1) (5.4.62)

Example 5.4.2: c(z) over vortex ripples

In this example, the combined convection-diffusion model is applied to the
suspended sediment distribution under non-breaking waves over a rippled sand
bed.
Consider the field situation studied by Nielsen (1983) and described in
further detail as Tests 57-60 by Nielsen (1984). The measured concentration
profiles for the different sediment sieve fractions are shown in Figure 5.4.7.

d(mm) w(ms)
—-— 0-125-0-18
seseseees O18 -0-25
0-25 -0-35
0-35 -0-50
IO LOO A
—-— 0:71-1-0 0-106

o-0O1 o-10 1-0

Figure 5.4.7; Time-averaged concentration profiles for different sand size fractions
over vortex ripples under non-breaking waves. Field data from Nielsen (1983).

256 COASTAL BOTTOM BOUNDARY LAYERS

 5.4; Distribution models

The distributions of the different grain sizes show the characteristic

behaviour which calls for a combined convection-diffusion description rather than
the traditional, pure gradient diffusion description. That is, the profiles plotted with
z against Inc do not show the similarity predicted by the pure diffusion solution,
see Figure 5.4.4, p 242. The profiles for fine sand tend to be upward convex while
the profiles for coarse sand are upward concave. This is the same pattern as shown,
by the laboratory measurements in Figures 5.2.11 and 5.2.12, pp 220-221.
The relative (the factor p(wo)/wo omitted) concentrations corresponding to
the combined convection-diffusion solution (5.4.61) are shown in Figure 5.4.8 for
a few values of the dimensionless settling velocity w* = woz1/€s.

elevation
Relative
(Z/Z1)

0
0.001 0.01
Relative concentration
Figure 5.4.8: Relative sediment concentrations calculated from the combined
convection-diffusion model, Equation (5.4.61). The numbers on the curves
correspond to the dimensionless settling velocity w* = woz1/€s.

It can be seen, in qualitative agreement with the data in Figure 5.4.7, that the
concentration profiles for the finer sand are upward convex while those for the
coarser sand are upward concave.
In detail, the calculations behind the c(z)-profiles in Figure 5.4.8 are as

AND SEDIMENT TRANSPORT 257

Chapter 5: Sediment suspensions

follows. Based on the measured wave and_ sediment parameters

(T, D, Hrms, ds0, N, A) = (7.28, 1.3m, 0.34m, 0.46mm, 0.09m, 0.53m) and linear
wave theory, the following near-bed flow parameters A= 0.52m, Aw = 0.65m/s
are found, corresponding to "the rms wave". The hydraulic roughness is then
calculated in accordance with Equation (3.6.14), page 159, which gives

r = 0.12m

and hence, z; = 0.09VrA = 0.023m. The sediment diffusivity is then calculated

from (5.4.57), €s = 40.5@z ~ 1.2:10°m’/s.
With these values of (zi, €s), the sand fractions represented in Figure 5.4.7,
which have settling velocities (0.0145m/s, 0.025m/s, 0.039m/s, 0.057m/s, 0.080m/s,
0.106m/s) correspond to the following values of the dimensionless settling velocity
w* = woz1/€s : (0.28, 0.48, 0.75, 1.09, 1.53, 2.03).
By comparing the measured distributions in Figure 5.4.7 with the calculated
ones in Figure 5.4.8 for appropriate w*-values, it can be seen that the model gives
reasonable predictions.
It may be noted, that the predicted concentration profiles for the finer sand
fractions (w’ = wWozi/es < 0.25) in Figure 5.4.8 are not as strongly upward
convex as the corresponding, measured profiles in Figure 5.4.7.
The predicted concentration profiles, for the finest sand, tend towards the
pure diffusion solution, for constant diffusivity

C(Z,Wo) = Co e Wo2/Es

which would plot as a straight line in Figure 5.4.8. The deviations from this
straight line, very close to the bed, are due to the form of the bottom boundary
condition (5.4.47) applied in the combined convection-diffusion model.
In order to obtain upward convex concentration distributions for the fine sand
fractions, it is necessary to assume a diffusivity distribution for which the pure
diffusion solution (5.4.32) is upward convex. This is obtained when €,(z) is a
decreasing function of z. There are good physical reasons for suggesting that the
sediment diffusivity is a decreasing function of z over ripples under non-breaking
waves, but this fine tuning of the model will not be attempted at the present stage.
At present, the main point is, that the combined convection-diffusion model
is capable of explaining the qualitative difference between the distributions of
different grain sizes suspended in the same flow. That is not possible within the
framework of either pure gradient diffusion or pure convection.

258 COASTAL BOTTOM BOUNDARY LAYERS

 5.4; Distribution models

The distribution of total concentrations (the "sum" of the size fraction profiles
discussed above) over a bed of fairly well graded sediment is not necessarily itself
given by Equation (5.4.61) with any particular value of w*. Nevertheless, it is
often attempted to predict the distribution of total concentrations with a formula
like Equation (5.4.61) and using the mean (or median) settling velocity of the bed
sediment.
For the case corresponding to the data in Figure 5.4.7, that leads to the total
c(z)-profile which is shown in Figure 5.4.9. The reference concentration
Co = c(O) = p/wo is calculated from Equation (5.3.10), page 228.

0.4

©w sees ccc cce secs ee sence cccseccrsnnessececee semen

eee eee Nene n ene ee ncn sen ene een sesccceweeeecesanncecceceemencesscsecssescece

© DS)

0.1 Ss a wcencenceeceesesees = on cnn saccscsseansscnsnsccesnsesexes

above
elevation
ripple
[cm]
crest

{E06 1E-05 0.0001 0.001 0.01

time averaged concentration
Figure 5.4.9: Measured (*) total concentrations compared to the distribution (5.4.61),
page 255, calculated for the mean sediment size ( d, wo )= (0.46mm, 0.061m/s) of
the bed sediment (+). The straight line corresponds to the simple exponential model
given by Equations (5.2.4), (5.3.10) and (5.2.5) page 217.

Example 5.4.3 ¢(z) from combined convection-diffusion over a flat bed

As an example of sediment suspension in purely oscillatory flow under sheet-
flow conditions consider Test 26 of Delft Hydraulics (1989).
The experiment was performed in a large oscillating water tunnel, and the

AND SEDIMENT TRANSPORT 259

Chapter 5: Sediment suspensions

flow and sediment parameters were (T, AQ, dso, Wo) = (85, 1.50m/s, 0.21mm,
0.033m/s ) corresponding to a grain roughness Shields parameter of 025 = 2.43
>> 1. Hence, the bedforms which were present (n, A) = (0.17m, 2.5m) would
have been rounded megaripples rather than sharp crested vortex ripples.
The boundary layer parameters and the concentration profile therefore are
calculated as for a flat bed. Then we find the hydraulic roughness r = 0.0026m
from Equation (3.6.14), page 159, leading to z; = 0.09VrA = 0.0063m, and the
diffusivity es; = 2@zi = 8.2:10°m’/s, see Equation (5.4.57), p 252. The
reference concentration is found from Equation (5.3.10), p 228

Co = p/Wo = 0.005 035 = 0.072.

The combined convection-diffusion solution (5.4.61) based on _ these
parameters, and the pure convection solution

0.3

Ogi salen.Netee ee Vie ose _ieee

= : i i
[iS
xe
o
3o :
O:sfiing
a.calpeteactigts
t sats Nese AS feseseeeeececececeeeceeennnnnesees ;1S ee eng

0 : ; ;
1E-05 0.0001 0.001 0.01 0.1
time averaged concentration

Figure 5.4.10: Measured c(z)-values from oscillatory sheet-flow compared to the

pure convection solution (5.4.15) and to the combined convection-diffusion solution
(5.4.61). Data from Delft Hydraulics (1989), test T26.

260 COASTAL BOTTOM BOUNDARY LAYERS

 5.4; Distribution models

COE BCG EC) = Co

(5.4.15)
(1 +.2/z1)
are compared with the measured concentrations in Figure 5.4.10.
The agreement is reasonable considering that the used expressions for the
diffusivity €s and for the convective distribution function F(z) (Equations 5.4.57
and 5.4.60) are the same as applied in the previous example for very different flow
conditions. These formulae were derived in Section 5.4.8 from concentration data
measured over vortex ripples. Nevertheless, these expressions for €s and F(z) are
temporary. Improvements should be sought when detailed measurements of c(z)
for different sand sizes in the same oscillatory sheet-flow become available.
The difference between the combined convection-diffusion solution (5.4.61)
and the much simpler, pure convection solution (5.4.12) is fairly small for this case
(Figure 5.4.10). This corresponds to the fairly large value of the dimensionless
Wo Z1
settling velocity w* = oad 2.5, compare with the sequence of concentration
S
profiles in Figure 5.4.8.
Assuming that the diffusivity and the convective distribution function can be
estimated from Equations (5.4.57) and (5.4.60) as above for oscillatory sheet-flow,
it seems likely that the relative settling velocity will be quite large

we= MOZl > 9

Es
for most sheet-flow conditions. Hence, using the pure convection solution (5.4.12)
instead of the complete convection-diffusion solution (5.4.61), may be acceptable
for sheet-flow under non-breaking waves in general.
It should be remembered however, that the use of Equations (5.4.57) through
(5.4.60) for sheet-flow conditions has not been verified in detail. The formulae
were derived from measurements over rippled beds. The verification will have to
be done in relation to measured concentration distributions of different sand sizes
in the same oscillatory sheet-flow, when such data become available.

AND SEDIMENT TRANSPORT 261

Chapter 6: Transport models

Figure 6.1.1: The modelling of coastal sediment transport processes still presents many
interesting, unresolved problems. The challenges of the swash zone are particularly daunting.

262 COASTAL BOTTOM BOUNDARY LAYERS

 CHAPTER 6

SEDIMENT TRANSPORT MODELS

6.1 INTRODUCTION

Natural beaches are constantly changing in response to changes in wind and

wave conditions and sometimes also in response to human interference, e g the
construction of breakwaters and groynes.
Such changes of the beach morphology are usually modelled iteratively with
each iteration involving three main steps. First the main flow pattern is calculated
on the basis of the incoming waves and the existing topography. Secondly, the
local sediment transport rates are calculated from the main flow pattern and the
sediment characteristics. Thirdly, the rates of morphological change are calculated
from the local sediment transport rates.The focus of the present text is the second
step in this modelling process.
Chapters 1 through 5 contain a discussion of the basic sediment transport
mechanisms for simplified topographical conditions and with the hydrodynamic
conditions, basically the free stream velocity u(t) taken for granted. The aim of
the present chapter is to provide a selection of manageable sediment transport
models each applicable to a certain combination of flow type and bed topography.
These simple models can then be used as building blocks in comprehensive
sediment transport models.

6.2 TRANSPORT MODELS ARE IN ESSENCE OF TWO VARIETIES

Sediment transport models are, in essence, of two kinds. Namely, cu-integral

models, and particle trajectory models. These two types are described in the
following.
First the traditional cu-integral (concentration times velocity integral) type of

AND SEDIMENT TRANSPORT 263

Chapter 6: Sediment transport models

models. The sediment transport rate, through a unit width of a vertical plane
perpendicular to the x,u-direction, can be calculated as
D

Q(t) = Jcet) usle,t) dt (6.2.1)

Z=0

where c(z,t) is the local, instantaneous sediment concentration and us(z,t) is the
instantaneous, horizontal sediment velocity. In most cases we are mainly interested
in the time-averaged transport rate, Q which, for a periodic process can be
calculated as
t+TD
Q= =) Jcz) u(z,t) dz dt (6.2.2)
tz=o

where T is the wave period. In terms of the steady, periodic and random
components, this may be written as

O = |\(cu+cut+cw)az (6.2.3)
Z=0
see Equation (1.1.17), page 11.
The last term c’w’ in this:expression is generally expected to be small. The
first two terms may be of similar magnitude or either of them may be dominant,
when transport in the wave direction is considered.
Transport perpendicular to the wave direction will, of course, be dominated
by cv_ which is the equivalent of the first term in Equation (6.2.3).
By far the majority of the existing sediment transport models are based on
this "cu-integral approach". However, the alternative models, using the "particle
trajectory approach” will, in some cases, lead to simpler and more accurate models.
The basic idea of the particle trajectory approach is illustrated in Figure
6.2.1. It can be seen that the time-averaged transport rate Q can be expressed
very simply in terms of the time-averaged pickup rate p , and the average distance
I, travelled by sediment particles.
In a short time interval 6; around f! , an amount p(t’) 5: of sediment is
picked up per unit area. During the time interval from this pickup to the settling of
the last grain, these grains will make a total transport contribution of J p(t) 5r.
This corresponds to the time-averaged transport rate

264 COASTAL BOTTOM BOUNDARY LAYERS

 6.2 Two kinds of sediment transport models

ror:
2 ES ES GCE JRaCR SS PEDO BOPEORe) wo Gina TRO ia SSR ORe ans ORIORaAT icmkaealn Kaas
JSS TAT
eiran ei e ESE
ees nae oh figeMirRPO
a es Sig Mi PRT
MgriAl vwERT sg,Sr aan
PTR ROB
g YS iaiea 18g
Mel PT RT
eo Ye
os” IPT
>
eiSig,IP Pe PS > IWael
Oni e recog! TaharMeee
RP PeSig Bieta

Figure 6.2.1: Ina short time interval 8; the amount of sand picked up per unit area
is p. If the average distance travelled by the moving sediment is J/,, the
corresponding sediment transport through one unit width of the plane A is seen to be
lx,Pp &: ‘

Q = kp (6.2.4)

a remarkably simple result.

The required ingredients to this type of model are obviously the pickup
function, which was discussed in Section 5.3, page 222 ff, and an estimate of the
average jump length /;. While the individual particle trajectories may be very
complex and highly variable it is generally not too difficult to make a
simple description and extract an adequately accurate value of the average jump
length /,.
This was done by Nielsen (1988a) for the case of non-breaking waves over
rippled beds. Two particle trajectory models including different amounts of detail
were developed for this phenomenon and compared with experimental data. In
addition, a classical cu-integral model based on the diffusion equation (5.4.27) was
considered.
In the simplest particle trajectory model, the "grab and dump model", it was
assumed that /, was simply plus or minus the near-bed water particle semi-
excursion (+A ).
It turned out that this simple model was the most consistently accurate of the
three. All three transport models will be discussed briefly in the following section.

AND SEDIMENT TRANSPORT 265

Chapter 6: Sediment transport models

Example 6.2.1: The particle trajectory approach to steady bed-load

The simple particle trajectory transport formula (6.2.4) is not restricted to
suspended sediment transport. It applies equally well to bed-load transport in
steady flow. It can therefore be used to find the average particle jump length /x
which corresponds to the Meyer-Peter formula (p 112) for steady bed-load

Dp = 8(0’-6)'> (2.3.11)

and to van Rijn’s pickup function

adAte
ae = ) (Ee
= 00003ieee z | 32
(5.3.2)

Rewriting these two formulae in the forms

Oz = 8(0’-6)'!>
Vs-l gd d
where the definition (2.3.10), page 112, for ®g has been used, and

el et Pie
p = 000039") ee | V(s-l) gd

one finds, for the typical values of (@c¢ , d) = (0.05, 0.2mm),

pres seope na
p
Thus, for sand which is being picked up at the rate given by van Rijn’s
pickup formula to yield the bed-load transport rate given by the Meyer-Peter
formula, the grains must, on the average, jump a distance equivalent to 44/ grain
diameters in each jump. Interestingly, this result is independent of the Shields
parameter @’.

6.3 SHORE NORMAL TRANSPORT OVER VORTEX RIPPLES

6.3.1 What the experimental data show
Vortex ripples are common in coastal areas where the flow is wave-
dominated and not too violent, see Section 3.4, page 135. They are also the

266 COASTAL BOTTOM BOUNDARY LAYERS

 6.3 Shore normal transport over ripples

predominant type of bedforms in laboratory flumes. The ripples themselves and the
sediment transport above them have therefore been studied in considerable detail.
Sediment transport over ripples is interesting because it is at first sight
counter intuitive. That is, the net sediment transport tends to be in the opposite
direction to the strongest flow velocities, see Figure 6.3.1. Thus, the net sediment
transport over vortex ripples under Stokes waves without boundary layer drift
(Uoo(t) = Uicos mt + U2 cos 2t ) is always in the offshore direction, see the
tunnel data of Sato (1986). Correspondingly, Inman & Bowen (1963) found that
when a current was superimposed on fairly sinusoidal waves over rippled beds the

@
~

E
=
Oo
2
®
> R 4 5 6 f§ 8 9 10 11

vm 6
~
N

E 2
iS
= -2

Sa -6

= .
- -10 A

oa Test 12
@®

=-14 — — Test24
"O Spiny ohha YS eT:

Figure 6.3.1: Measured net sediment transport rates over rippled beds for three
different bed sediments under essentially the same flow conditions. Data from
Schepers (1978).

AND SEDIMENT TRANSPORT 267

Chapter 6: Sediment transport models

net sediment transport tended to be in the opposite direction of the superimposed

current.
These general features are shown by the data in Figure 6.3.1 which also show
another interesting fact. Namely, that for almost identical flow conditions, the
magnitude of the net sediment transport rate over ripples seems to be practically
independent of the median size (d50) of the bed sediment. This again is counter
intuitive and interesting, particularly in connection with the "CERC Formula
Paradox" discussed in Section 6.5.3, page 291.
For the data in Figure 6.3.1 the flow conditions (top graph) were very similar
in the three experiments, (T, D) = (1.5s, 0.3m), and the velocity amplitudes U7
and U2 and the time-averaged velocity u, all measured 0.Jm above the bed,
were within the shown envelopes. The parameters (d50, wo) of the bed sediment
were as follows: Test 12 (0.125mm, 0.010m/s); Test 24 (0.25mm, 0.028m/s); Test
36 (0.465mm, 0.060m/s).
The corresponding, measured sediment transport rates (bottom graph) show
at least two interesting facts. Firstly, even though the measured current u(0.1m) is
always positive (shoreward), the sediment transport rates are predominantly
seaward. Thus, the net current velocity measured a few ripple heights above the
bed gives no indication of the net, shore normal sediment transport rate.
Secondly, the measured sediment transport rates have very similar magnitudes
despite the considerable differences in sediment size.
The fact that the resulting transport tends to be in the direction opposite to the
strongest velocities can be understood when the mode of transport is considered
more closely. The sand cloud cannot travel very far with the velocities that
entrained it from the bed. It gets trapped in the lee vortex behind the ripple crest
until the free stream velocity changes direction. Then the vortex moves into the
free stream with the sand and travels in the direction opposite to the velocity which
entrained the sand.
The ability of vortices to trap sand is discussed in Section 4.6, page 181 ff.
Nielsen (1988a+b) discussed three different models of sediment transport
over ripples (under non breaking waves) and discussed their relative ability to
reproduce the behaviour shown by experiments similar to those in Figure 6.3.1.
All of the transport models were developed for the case of an exponential
distribution of the time-averaged suspended sediment concentrations, (page 215).

Cz) = Coe = a e/Ls (5.2.4)

(3)

which would be a reasonable approximation under those experimental conditions,

and using the pickup function (p 230)

268 COASTAL BOTTOM BOUNDARY LAYERS

 6.3 Shore normal transport over ripples

p(t) = V4 dr(t-14) + V" Sr(t-24) (5.3.14)

For details of the sediment amounts V4 and V“ see page 230.
The delta functions which have zero width are preferred to functions with
wider peaks because of mathematical convenience. This has little effect on the
final results however, because only the first one or two harmonics make significant
contributions to Cu since the higher harmonics of the near-bed velocity are very
small.

6.3.2 The gradient diffusion transport model

The first model developed by Nielsen (1988a) was based on the assumption
that pure gradient diffusion was the sole sediment distribution mechanism as
discussed in Section 5.4.4, p 241.
The sediment diffusivity which corresponds to the assumed _ c-distribution,
Equation (5.2.4), is the constant es = Lswo . In this model, the resulting
sediment transport rate is calculated from the usual cu-integral expression

D
DT+t

0 = Ji
s§
@aeae + FJ Jiten cen arae
Ont

which, with the velocity field given by

u(z,t) = u(z) + u(t) = uz) + ¥ Uncos nat

—tn)
n=1

and, with the expressions (5.2.4), page 215, and (5.4.38), page 244 for the
sediment concentrations inserted, leads to the diffusion based transport formula
a)
Gy
9 a —_
aps JWZ)
aps ay —2z/Ls
Taz

+ 2 Ls >) is cos(not@ + 2argOn — n@tn)

WoT pe lon

+ a Ls ahAe cos(nat” + 2argon — nWtn) (6.3.1)

WoT jal om

AND SEDIMENT TRANSPORT 269

Chapter 6: Sediment transport models

where v4 = pT=CoWoT and the ratio v4 Vv" was simplistically taken as

(UTI ye , inspired by the form of the pickup function (5.3.11), page 228.

6.3.3 A particle trajectory model for shore normal transport over ripples
The second model, the convective transport model, discussed by Nielsen
(1988a+b) was of the "particle trajectory type" and based on the assumption that
the sediment was distributed vertically by a simple convection process similar to
the one described in Section 5.4.3, page 238. For simplicity however, the vertical
convection velocity we was assumed very large, so that all sand was assumed to
reach its entrainment level ze instantaneously when the lee vortex was released. |
In this description the sand which is entrained at time f° and which totals Vv
per unit area, contributes to the total sediment distribution by

ct) = VF@ = Ves

immediately after the time of entrainment. This contribution subsequently
decreases with time as the distribution settles uniformly with speed wo. Hence, at
a later time ¢, when the distribution has settled the length wo(t-t') the
concentration contribution is reduced to

(zt) = Vi F(ztwolt-t]) = Viele DLs tor tof 6.3.2)

The basic philosophy of particle trajectory models, as outlined in Section 6.2,
page 263, is to express transport contributions as the amount of sand entrained
times the average distance travelled and then divide by the corresponding time
interval to get the average transport rate.
In the case of vortex ripples under periodic waves, where the amounts v4
and V“ respectively, are entrained at the two Uso(t)-reversals, see Figure 5.3.4,
page 229, this corresponds to

Q = 7Ve ey) (6.3.3)

where ee and /x” are respectively the average distance travelled by sand
entrained at 1” and ¢“. These lengths are calculated as follows.
A sand particle which is entrained to the level ze will, while it settles at
speed wo be carried the horizontal distance

270 COASTAL BOTTOM BOUNDARY LAYERS

 6.3 Shore normal transport over ripples

It!ze) = J u(ze—wolt-t!],t)dt (6.3.4)

tt
This sand particle hits the bed at the time t= t!+ zelwo. .
The average distance travelled by particles entrained at time ¢’ is therefore

ke = is F’ (Ze) Kg Ze) dze = { F’ (ze) J U(Ze—Wolt— o t) dt dze

os fe) tot!
(6.3.5)
where -F’(Ze) represents the probability density of entrainment to the level ze ,
see Equation (5.4.8), page 238.
With a velocity field of the form u(z,t) = u(z) + u(t) , ie, with the boundary
layer for the oscillatory velocity component & assumed very thin compared to
the sediment distribution, Equation (6.3.5) can be reduced to

—| = FO xii(z) dz - |F'(ze)
& Jixcartie dt dze
D D Wo

(6.3.6)
zZ=0 Ze=O tet

where the first term has been transformed using integration by parts.
While this may be a fairly complicated expression, its evaluation is straight
forward compared to solving the time-dependent diffusion equation with even a
moderately complicated €;-distribution.
To illustrate the basic principles of transport modelling with models of the
particle trajectory type, we shall evaluate Equation (6.3.6) on the basis of a
particularly convenient expression for the velocity,

u(z,t) = u(z) + U; sin w(t-t) (6.3.7)

where the sign "-" is used in connection with the zero down-crossing event (at t =
t*) and "+" is used in connection with the zero up-crossing event (at t=t"), see
Figure 5.3.4, page 229. With this expression for the velocity inserted into Equation
(6.3.6), we find

AND SEDIMENT TRANSPORT DTM

Chapter 6: Sediment transport models

D D Ze/Wo

pe ee JF@i@ dz + (eee s fun sin at’ dt’ dze

Wo Ls
Z=0 Ze=0 re)

D D
1
mete
Bits Fe =a U1
u(z) dz +Birt
an. Je—Ze/Ls (1lL-— cos wo
eo dz e
Z=0 Ze=0

which for D/Ls >>1_ can be reduced to

; ols
= w
f= +frei@at
Wo
“1,
Wo
—~— Sed,
(6.3.8)
=O Lots)
0)

Inserting this into the general transport formula (6.3.3) gives

; als
~ V4"
2. == pri FOIu
—— +. emi
v4 EL = Woaa | 6.3.9 OO
2=0 1 + OE.
(4)

which, since V44V" = CoWol = pT can be simplified to obtain the convection

based transport formula

D
oLs
—_

0, = Co]
F@)ue) dz + oti L sia
ageaatlt
oat 4020)
z=0 . 1+ Cyan

This result is very similar to the result (6.3.1) for the diffusion based model,
although the model philosophies are very different. The actual transport rates are
practically identical, see Figures 6.3.2 and further examples given by Nielsen
(1988a). It would therefore seem logical to use the convection model rather than
the diffusion model since Equation (6.3.10) is much simpler than Equation (6.3.1).
The particle trajectory approach, which is used in the convective transport
model, has a further advantage however, namely that the expression (6.3.6) can be

Die, COASTAL BOTTOM BOUNDARY LAYERS

 6.3 Shore normal transport over ripples

15

r
®
we 5
1=

S 0) Sates ~:,Gee i ee | fe | a

5
: +5

3 -10

-15
grab and dump
-20
3 4 5 6 v 8 9 10 11 12
distance from wavemaker, [m]

Figure 6.3.2: Estimated sediment transport rates corresponding to the diffusion

based transport model (6.3.1), the convective transport model (Equation 6.3.10), and
the simple grab and dump model, Equation (6.3.11). For this sand size (d50 =
0.125mm), all of the models give very similar results, and the agreement with the
measured transport rates is good. Data from Schepers (1978) Test 12, For the
hydrodynamic details see Figure 6.3.1, page 267.

evaluated for almost any convective distribution function F(z) while an analytical
solution for the diffusion model is impossible for all but a few €s-distributions.
Furthermore, the convective transport model can, with minor modifications,
be used on a wave by wave basis for irregular waves, while the analytical solution
for the diffusion based transport model relies on the assumption of periodicity.

6.3.4 The grab and dump model for shore normal transport over ripples
With the present level of accuracy of coastal bottom boundary layer
calculations, it is difficult to justify very complicated sediment transport formulae.
Hence, it is worthwhile checking if further simplification is possible, and indeed it
is. It is, in fact, possible to model the shore normal sediment transport over rippled

AND SEDIMENT TRANSPORT 273

Chapter 6: Sediment transport models

beds under non-breaking waves without even considering the concentration

distribution. Only the amounts v2 and V“ of sand entrained by the released lee
vortices at each Uo-reversal are needed.
The grab and dump transport model is developed as follows. SS sand is
entrained ("grabbed") in two parcels each wave period, at the times f° and t” of
free stream reversal, see Figure 5.3.4, page 229. In each case the grabbed sand is
then transported the average distance A, in the direction opposite to that of the
velocities which picked it up, and dumped. A is the water particle semi-excursion
just above the boundary layer.
With the two parcels containing the amounts v7 and V“ respectively, the
corresponding transport in one wave period is A( VV = corresponding to an
average transport rate of

ae 1
VT = 5a a es (6.3.11)
OF =“
for the grab and dump model. This is a remarkably simple expression which,
according to Nielsen’s (1988a) comparisons with Schepers’ (1978) laboratory
experiments, turns out to be more accurate than Equations (6.3.1) and (6.3.10)
from the more complicated models above.
The result (6.3.11) does not include a transport contribution carried by the
steady flow component u, but such a contribution may be included by adding the
usual Jem u(z) dz provided that u(z) and c(z) are known.

6.3.5 Comparison of the three models

In Nielsen’s (1988a) comparison of the three models of shore normal
sediment transport over rippled beds the transport contribution carried by u was,
in all cases, ignored. Mainly because the distribution of u(z) including the
boundary layer drift near a rippled bed is virtually unknown, see Section 1.4.5,
page 60. The diffusion based solution (6.3.1) was calculated with the use of the
first two harmonics of C(z,f) and U(z).
It was found that the three models performed equally well for the finer sand
sizes, see Figure 6.3.2. However, only the grab and dump model predicted the right
magnitude of the sediment transport rate for the coarsest sand. Thus, for Test 36
shown in Figure 6.3.1, p 267, the diffusion based model and the convective
transport model both predicted a range of transport rates which was less than one
tenth of the measured range while the grab and dump model predicted the right
range, see Figure 6.3.3.

274 COASTAL BOTTOM BOUNDARY LAYERS

 6.3 Shore normal transport over ripples

sediment
net
rate
transport

3 4 5 6 7 8 9 10 11 12
distance from wavemaker, [m]

Figure 6.3.3: Estimated and measured sediment transport rates as function of distance from the
wavemaker in a wave flume. For this coarse sand (ds50 = 0.465mm) only the simple grab and
dump model predicts the correct magnitude of the transport rates. Data from Schepers (1978) Test
36. For further details see Figure 6.3.1, page 267.

To understand this, consider the three formulae (6.3.1), (6.3.10) (both with the
c u-contribution ignored) and (6.3.11) in slightly rewritten and simplified forms. If
only the contribution from the first harmonic of u(z,t) is included in the diffusion
based transport formula (6.3.1), the three formulae can be written as follows

asia
nO) obs
a8 Ig SF ols - (6.3.12)

Pompe
oe ols
ete (6.3.13)

Se atercerigli (6.3.14)
pe pa, UiS

AND SEDIMENT TRANSPORT DES

Chapter 6: Sediment transport models

where S is a skewness function which determines the relative magnitudes of VA

and V“. Thus, S should be zero if the motion is completely symmetrical like a
sine wave, and generally non-zero if the motion is asymmetrical.
Nielsen (1988a) used a skewness function of the form S = S (U*,U). This
may be adequate for Stokes waves where the transport asymmetry is reflected by
the ratio U*/U between the velocity extremes. However, waves with sawtooth
asymmetry may have U*= U' while t™ #7 as discussed in connection with
King’s (1991) experiments in Section 2.4.4, page 121. For such waves it would be
necessary to express the transport skewness as a function of the two extreme bed
shear stresses, S = S(tt, i
pa»
The three transport expressions above show that for fixed p(=Cowo) and
fixed S,ie, for fixed V7 and V4 , the transport rates Q 4 and @Q‘ for the first
two models will both decrease rapidly with increasing particle size because of the
Ls
factor . This factor is absent in the grab and dump expression, making this
Wo T
model less dependent on grain size. The data in Figures 6.3.1, 6.3.2 and 6.3.3 show
that this weak grain size dependence of the grab and dump model is a true
characteristic of shore normal sediment transport by non-breaking waves over
rippled beds.
Note that weak grain size dependence is not a unique feature of shore normal
transport over ripples in small wave flumes. Also field measurements of the total
longshore transport rate or littoral drift show remarkably weak grain size
dependence. See Section 6.5.3, page 291.

6.4 SHORE NORMAL SEDIMENT TRANSPORT OVER FLAT BEDS

6.4.1 Introduction
Shore normal sediment transport over flat beds may occur on the beach face,
ie, in the swash zone, throughout or in parts of the surf zone and outside the surf
zone if the waves are large and the sand is fine.

6.4.2 Shore normal sediment transport in the swash zone

While it is generally recognised that the swash zone contributes a major part
of the total littoral drift, and while it is obviously necessary to model the shore

276 COASTAL BOTTOM BOUNDARY LAYERS

 6.4 Shore normal transport over flat beds

normal sediment transport through the swash zone in order to model beach change,
it is beyond the present state of the art to model swash zone sediment transport.
Too little is known, at present, about the boundary layer flow and the
corresponding shear stresses in the swash zone to even attempt a description of the
basic sediment transport mechanisms in this area. See also Figure 6.1.1, page 262.
Furthermore, swash zone sediment transport includes the extra complication
of significant flows of water perpendicular to the sand surface. Experience has
shown, that increasing the inflow by pumping from the ground water in the beach,
tends to increase the rate of beach accretion.
Thus, experiments show that the flow perpendicular to the sand surface in the
swash zone affects the rate of beach accretion. The details of the mechanisms are,
however, not understood.

6.4.3 Shore normal sediment transport in the surf zone

In the surf zone the bed is often flat or with respect to sediment transport
practically flat. That is, the megaripples which are sometimes present have no
sharp crests that might cause rhythmic vortex formation, and the boundary layer
structure over the megaripples is therefore practically the same as over flat beds.
However, while the bed topography in the surf zone may be simple, the flow
structure is certainly not simple. The immediate vicinity of bars offers special
problems, see e g Hedegaard et al (1991) and the entrainment under plunging
breakers is at present beyond detailed modelling. Even the simplest of hypothetical
surf zones, namely a two-dimensional one with constant bed slope and spilling
breakers with constant wave height to water depth ratio, causes problems. The
main difficulties lie in the modelling of the important seaward bottom current, the
undertow, see Section 1.4.5, page 60.
While detailed, quantitative modelling of shore normal sediment transport is
very difficult, some aspects, like the position of bars, have been predicted
successfully by Boczar-Karakiewicz & Davidson-Arnott (1987) and by
Boczar-Karakiewicz & Jackson (1991). They use a fairly simple model of the
waves and the shore normal sediment transport which bypasses the intricacies of
wave breaking.
The broad question of whether a beach is going to accrete or erode when
exposed to a certain set of wave conditions may be answered with a certain amount
of confidence according to Kraus et al (1991). Based on a comprehensive review
of field and laboratory data they suggested the criterion

AND SEDIMENT TRANSPORT D8

Chapter 6: Sediment transport models

== 3
H.
Beaches erode if —° < 0.00070 te e )

(6.4.1)

Beaches accrete —*
Lo

where Ho and Lo are the deep water wave height and wave length respectively.
___ They found the best correlation by using the mean offshore wave height
(Ho). For a Rayleigh distribution, the mean wave height is related to the root mean
square and significant heights by H = 0.886 Hrms = 0.626 Hs.

6.4.4 Shore normal sediment transport outside the surf zone.

If the waves are big enough (82.5 2 1, see Section 3.4) the sea bed outside the
surf zone will be free from vortex ripples and thus, with respect to sediment
transport, practically flat. In such cases, it should be possible to model the
sediment transport rate with the sheet-flow formulae which were developed in
Section 2.4, page 116.
A simpler formula was presented by Ribberink & Al Salem (1990)

QO = 0.00018ua (6.4.2)
where the total net transport Q is measured in m/s and ut) in m/s.
This formula was, however, only designed for 0.2mm quartz sand, and it
may be under predicting Q systematically for shorter wave periods than the ones
used by the authors (6.5s and 9./s). This was mentioned by the authors in relation
to the data of Horikawa et al (1982), and is not unexpected since shorter period
means larger d59/A and hence a larger friction factor for fixed uyms, see Section
1.2.5, page 23. It should also be remembered that formulae of this type fail to
predict the net sediment transport due to "sawtooth wave asymmetry" as discussed
in Section 2.4.4, page 121.
Not all of these limitations apply to the alternative sheet-flow sediment
transport formula (2.4.16), which was developed in Section 2.4.4, page 125. It

278 COASTAL BOTTOM BOUNDARY LAYERS

 6.4 Shore normal transport over flat beds

should be equally valid for all (non-cohesive) sediments and for waves of all
shapes and periods. However, neither of the two formulae applies, without major
modifications, to flows with large steady flow components such as rip currents or
in surf zones with a strong undertow.
In such flows, the contribution to the bed shear stress from 4u(z) must be
given special consideration. As a first, rough approximation, this problem might be
approached as if there was a linear transfer function between near-bed velocities
and the bed shear stress. The procedure would then be to subtract the time-average
from the near-bed velocities and apply Equation (2.4.15), page 125, to the residual
u(zr,t) = u(Zr,t) — Uu(Zr) ie

1
5 S25 Arms as
at Tire as ; U(Zr,tn+i1) — U(Zr,tn—1)
0’[u(tn)] = ea ad; (cos@r Wp H(zr,tn) + sings ao iea wt )

(6.4.3)
To this expression, a contribution due to the time-averaged bed shear stress (0)
must then be added in order to get the total, instantaneous effective Shields
parameter

"n)
O(n) OIF
= Olli] + gateed
UO)Fins (6.4.4)

This total value is then inserted into the sheet-flow sediment transport formula

M(t) = 8 [6’(t) —0.05]!> wa for 10’(t)l > 0.05 (2.4.16)

0 for 18’(¢)| < 0.05

The factor 8, which corresponds to the Meyer-Peter bed-load formula, may be

teplaced by a factor of the order /2 for fine sand at high flow intensities in
accordance with the trends of the data in Figure 2.3.2, page 113, and Figure 2.4.2,
page 119.
With respect to obtaining the mean bed shear stress t(0) from velocity data
there is basically two options. If ” is known only at a single level zy a rough
estimate may be obtained with a simple wave-current interaction model as done in
Example 1.5.2, page 90. If a profile of u(z) is available, an estimate of the
friction velocity uw and hence T(o) may be obtained from a log-curve fit. In
both cases however, new methods:should be developed to account for the influence

AND SEDIMENT TRANSPORT 279

Chapter 6: Sediment transport models

of the momentum transfer term p (iw)oo which was discussed in Section 1.5.9,
page 91. For an example of shore normal sediment transport calculations over a
flat bed see Example 2.4.1, page 126.

6.5 SHORE PARALLEL SEDIMENT TRANSPORT

6.5.1 Introduction
Sediment transport in the shore parallel direction or more precisely, the
direction perpendicular to the wave motion, is simpler to model than the transport
in the wave direction because V = 0 and the equivalent formula to Equation
(6.2.3), page 264, for the time-averaged sediment transport rate therefore becomes

Oy = \(ev+ev)d (6.5.1)
Z=0

where it seems reasonable to assume that the last term c’v’ is insignificant so that
the time-averaged sediment transport rate is given simply by
D

Oy = Joe) vWe) az (6.5.2)

Z=0

While this formula is valid in principle, both inside the surf zone and outside,
the details of the calculations are quite different for the two cases. This is because
of the strong influence on both c and v_ from the extra turbulence and from
additional convective mechanisms which are associated with wave breaking.
The presence of steady flow components (u(z), v(z) ) superimposed on the
wave motion i..(t) will increase the effective bed shear stresses and hence the
magnitude of c(z) compared with a pure wave motion. However, direct
experimental data which quantify this effect are not yet available.
As an educated guess however, it might be adequate to base the estimation of
the reference concentration Cg on a pure-wave formula like Equation (5.3.10),
page 228, with 025 replaced by an augmented effective Shields parameter
analogous to Equation (6.4.4). For example

ul & ne Tx(0) 2 Ty(0) 70.5

) {(@2.5(u) + heya
ly eee as b Aiie
(Dea?
Died } (6.5.3)

280 COASTAL BOTTOM BOUNDARY LAYERS

 6.5 Shore parallel sediment transport

A similar concern arises with respect to the sediment concentration

distribution. That is, the additional turbulence due to the presence of
u(z) and/or v(z) will stretch the sediment concentration profiles upwards but
again, direct experimental quantification of the effect is lacking.
This problem may, in theory, be addressed in fairly simple terms. The natural
approach is to replace the pure-wave sediment diffusivity by an eddy viscosity for
the combined wave-current flow when calculating the Cc(z)-distribution from the
combined convection-diffusion Equation (5.4.49), page 248.
However, care must be taken when deriving sediment diffusivities from eddy
viscosities in combined wave-current flows. There are various choices of eddy
viscosities, see Section 1.5.9, page 91, and their respective relations with the
sediment diffusivity are not well understood.
For rippled beds, the increased near-bed sediment concentrations and the
increased vertical distribution scale due to current turbulence will be partly
balanced by the expected decrease in bedform size and steepness due to the
presence of the current, see Section 5.2.5, page 218.

6.5.2 Shore paralle! sediment transport outside the surf zone

Outside the surf zone, time-averaged sediment concentrations may be
estimated either with the simple exponential model described by Equations
(5.2.4)-(5.2.5), pp 215-217, for rippled beds, or in general, with the more detailed
combined convection-diffusion model developed in Sections 5.4.5-5.4.9, p 245 ff.
The time-averaged velocities perpendicular to the wave direction are
considerably simpler to model than those in the wave direction because Vw = 0.
This simplifies the equation of motion considerably compared to that for the wave
direction, (Equation 1.1.21, p 12), where the corresponding term ww _ plays a
significant role, see Section 1.5.9, page 91ff.
Thus, the shore parallel current distribution v(z) and the shore parallel bed
shear stress Ty(0) = plv*lv* should be adequately estimated by the method
outlined in Example 1.5.2, page 90, with the apparent roughness increase

Za Awr ,t
= —.
— +—
Zo Nee ey

determined in agreement with data from perpendicular waves and currents, and
from tunnel data where iw = 0, see Figures 1.5.13 and 1.5.16, pp 67 and 69.

AND SEDIMENT TRANSPORT 281

Chapter 6: Sediment transport models

Example 6.5.1: Shore parallel sediment transport over a rippled bed

Consider the problem of estimating the shore parallel sediment transport rate
at a position outside the surf zone (no breaking waves). Velocity data from a single
current meter at elevation zy = 1m above the bed is available together with a
sample of the bed sediment.
Consider the example of (¥(zr), Urms, Tp, d50, Wo, 5) = (0.3mIs, 0.355mis, 8s,
0.8mm, 0.10m/s, 2.65), which correspond to an equivalent velocity amplitude
Aw = V2 firms = 0.502mls and Arms = tirmsTp/2m = 0.639m.
Using the relevant formulae from Section 2.2, page 102 ff, we find the
corresponding parameters: (y, f2.5, 925) = (19.5, 0.014, 0.136), from which we
can estimate the ripple geometry with formulae from Section 3.4, page 135 ff.
The relative ripple height is found to be

i = 21w!*® = 0.087 (3.4.8)

corresponding to an estimated ripple height of 0.056m, and the ripple steepness is

* = 0:342= 0:34 4VOss 290.136 (3.4.6)

The sediment distribution parameters (Co and Ls), corresponding to the

simple exponential distribution model, Equations (5.2.4) and (5.2.5), page 217, can
then be estimated. The reference concentration is
3
02°5
Co = 0.0050; = 0.003 ——"—— = 0.00035 (5.3.10)
(l-nn/A)

and the distribution length scale is

AQ@
Ls = 0.075 pres be 0.021m (5.25)
oO

Hence, the estimated distribution of the time-averaged suspended sediment

concentrations is given by

az) = Coe “NE “0000s e (6.5.4)

where z is measured in metres above the ripple crest level. Note that this simple
exponential model for c(z) is only valid for a certain range of conditions which
may be inferred from the Figure 5.2.7, page 216.

282 COASTAL BOTTOM BOUNDARY LAYERS

 6.5 Shore parallel sediment transport

The current velocity profile is determined in analogy with Example 1.5.2,

page 90. Based on a hydraulic roughness of

r = 817/X+ 505d = 0.061m (3.6.14)

and Zo = r/30 = 0.0020m: The steady friction velocity v« is determined from

a
ig oS KV
eee (1.5.50)
eo = oe = 42
Zo i eae?

With the apparent roughness increase given by the simplistic formula (page 89)

AMO r ,v AO
ie)
ary 5 eg e
0.44 25:
(1.5.48)

and «=0.4 and with other values inserted Equation (1.5.50) becomes

axles 0.4-0.3 a 0.12

it eel. 0502 7.73+ Inv«
n 0.0020 ~ In 0.44 5

This formula converges rapidly to the friction velocity value v« = 0.029m/s,

corresponding to F = 0.44 4° = 7.62 , and hence (Equation (1.4.5) page 87)

*

2a=
F 79=0.015m, and | = e'zq = 0.041m.
With these values inserted, the current velocity distribution given by
Equations (1.5.42) and (1.5.43), page 87, becomes

re = 1.77z [m/s] for z<0.041m

Wz) = Se (6.5.5)
Vx Z Z
7 In goose 0.073 In 0015 [m/s] for z>0.04lm

This current distribution is shown in Figure 6.5.1 together with the sediment
concentration profile (Equation 6.5.4), and the distribution of the product c v.

AND SEDIMENT TRANSPORT 283

Chapter 6: Sediment transport models

elevation
[m]

Figure 6.5.1: The sediment concentration distribution (6.5.4) and the current
distribution (6.5.5) together with the distribution of the product Cv ( not to scale).
The estimated ripple height is 0.056m.

The time-averaged sediment transport rate is then calculated from

D 1 ra
Co Ve 3 Cie fn
ae aR
Oy = Jem v@ az = orl temas’ age ppd Lapel
i) 1
4)

which can be approximated (within 3% for I/Ls > 0.01) by

2
Ve gol9(Ls) 0.79
Oy = —K Co Ls {1 _ (6.5.6)
For //Ls 2 2, as in the present example, the second term can be neglected
completely so, we find the simple formula

Oy =
Ve
x ©? ;
B for I/Ls = 2 (6.5.7)

284 COASTAL BOTTOM BOUNDARY LAYERS

 6.5 Shore parallel sediment transport

With the values from above inserted, this gives a net sediment transport rate
of Oy =2.7- 107 'm’/s.
This transport rate seems to be too small since the Meyer-Peter formula
(Equation 2.3.11, page 112), with the effective Shields parameter 0’ based on the
current friction velocity alone, gives the higher sediment transport rate

— att 1.5
Qy = ee - 005| Vs-l) gd d = 13-10 °m’/s (6.5.8)

This ought to be a lower estimate, since the additional bed shear stress contribution
of the waves is neglected.
Augmentation of the effective bed shear stress in accordance with Equation
(6.5.3), page 280, and keeping the bedform shape unchanged leads to an increase
of about 36% for the expression (6.5.7). But that is not enough to make it realistic.
Nevertheless, Rasmussen & Fredsoe (1981) used a very similar method to
that which lead to Equation (6.5.7) above, and they found good agreement with
laboratory experiments. The difference is, that their sand was fine (d59 = 0.18mm)
compared to the sand in the example above (d59 = 0.80mm).
Thus, we get a similar picture to that for the shore normal sediment transport
over rippled beds discussed in Section 6.3, p 266 ff. That is: traditional cu-integral
formulae like Equation (6.5.7) seem to work for fairly fine sand. For coarse sand,
however, the process is a very organised "grab and dump" process and is probably
better described by a "grab and dump" model.
A "grab and dump" model for longshore transport over ripples under non-
breaking waves may be constructed as follows.
Sand is picked up twice every wave period and the total amount picked up is
V = CowoT. The average distance ly travelled in the direction along the ripple
crests by that sand would be about 7/2 times the velocity at a certain near-bed
level, say one ripple height. Making use of the velocity distribution (6.5.5), that
leads to

L sioPiaxAVe

and hence, (see Figure 6.2.1, page 265) to the transport contribution

Ve
Vly = Cowol — abT

AND SEDIMENT TRANSPORT 285

Chapter 6: Sediment transport models

from sand entrained in one wave period, or an average grab and dump sediment
transport rate of

taist? ve 1)
Oy, = CowoT x 21 (6.5.9)

With values from above inserted, this gives Oye = 14-10°m’Is.

A fine tuning of Equation (6.5.9) and a serious discussion of the limits of
applicability for the formulae (6.5.7)-(6.5.9), is hardly possible with the presently
available data. However, the discussion in Section 6.3, pp 266 ff, where similar
models were compared with laboratory data of shore normal transport over ripples,
and the experience of Rasmussen & Fredsoe (1981), indicate that models of the
type (6.5.7) are applicable for grain sizes up to 0.25mm, while the grab and dump
models perform reasonably over a wider experimental range, at least 0.085mm <
dso < 0.5mm.

Example 6.5.2: Shore parallel sediment transport over a flat bed

Consider a situation outside the surf zone with the same sand and wave
parameters as in Example 5.4.3, page 259, but with a shore parallel current of
0.3m/s measured at zy = 1.0m,ie, (v(zr), A, T, dso, 5) = (0.3m/s, 1.50m/s, 8s,
0.21mm, 1.65 ).
For these conditions it was found in Example 5.4.3 that the time-averaged
sediment concentrations c(z) were reasonaly described by the pure convection
solution

:
6)... = Co F(z). = Commman OSa (5.4.15)
(1+ 2/2)

with Co = 0.072 and z; = 0.0063m. The hydraulic roughness was found from
Equation (3.6.14) tobe 0.0026m corresponding to zo = r/30 = 0.000085m.
As in the previous example and Example 1.5.2, page 90, the velocity
distribution is derived from the information above by first finding the current
friction velocity from

be K V(z
7. = (2r) (1.5.50)
inze — nH 42 © 42
Zo vx’ A’ 2

286 COASTAL BOTTOM BOUNDARY LAYERS

 6.5 Shore parallel sediment transport

which, with the simplistic formula for the apparent roughness increase (page 89)

Z
cay | ( | 5|= 0.44 es (1.5.48)

and with the values above inserted, yields

<7 0.4: 0.30 0.12

1.0 £50,500 >9:79-+"ln Ve
In 5000085 ~ ™ (0.44 Vx

This converges rapidly to the friction velocity value vx = 0.020m/s. Based on

this value, we find

za = 04452 7 = 0.0028m
*

and

1 = e!zq = 0.0076m
From these parameters, the velocity distribution is found by inserting into
Equations (1.5.42) and (1.5.43), page 87

Ft = 6,58z[m/s] for z<0.0076m

Wo t= (6.5.10)
Vet.
Ke za
UL yiosoin
EO
== — tl ||for 2> 000 76m
0.0028 :

The net shore parallel sediment transport rate can now be found from
D
Oy = Javea
4)
which becomes
my if A D Stn s
—_— Vx Z. a
= Co—s |ea + |— az
2 Big J
3)
(142/21) | (142/21)

AND SEDIMENT TRANSPORT 287

Chapter 6: Sediment transport models

_ o D/z\ : =

Oy = Cony
— Vx Z1
2 tntt- “7
1
1+ 2
14+//7
Z1
+ J—™a +
Vz i+
nx
*)
Za
+e

(6.5.11)
which for D/z; > °° and within the range 0./< I/z; < 80 may be approximated
by
In=
ne Bs
0.52 + 0.38 cos (0.8In=> ra = Sea:
=e Vx
Oye aCe
Z1
(6.5.12)
see Figure 6.5.2. The contents of the brackets in Equation (6.5.12) generally have
magnitude about / so, for order of magnitude estimates it is reasonable to use

a v*
Oy = on 21
Goues 2 (6.5.13)
for the shore parallel sediment transport rate under non-breaking waves over a flat
bed.

x etary Lae uaa

0.6 betes Mi feedendecacbdadeecrneneen eee

Sebeeeeentee rte: SOetienny Eernee Sener Sant Gee Gan ae Se Gein, Cn |

0.2 Segoe fete pee ereeeres Sor sustwvcneesy

0. it H oe SO
0.1 1 10
100
Figure 6.5.2: The first two terms (x) inside the brackets of Equation (6.5.11) can be
approximated by the corresponding terms in Equation (6.5.12), (the curve).

288 COASTAL BOTTOM BOUNDARY LAYERS

 6.5 Shore parallel sediment transport

With the estimated parameter values inserted, Equation (6.5.12) yields

Qy = 2.9. 10°m’/s
If the bed shear stress contribution from the current is included, using
Equation (6.5.3),page 280, in the calculation of Co from Equation (5.3.10), one
finds Co = 0.074 _ instead of 0.072 or an increase of 2.8% which is
insignificant.
On the other hand, the effect of the current on the concentration distribution
through increased diffusivity may be significant. Thus, if the diffusivity €s(z) in
the combined convection-diffusion solution (5.4.49), p 248, is assumed equal to
the eddy viscosity on which the current velocity distribution (6.5.10) is based, ie

pate Ve for z</ (6.5.14)

K Ve Z for z>/

then, the concentration profile is stretched vertically, compared to the pure wave
solution, as shown in Figure 6.5.3. The net shore parallel sediment transport rate,

0.3

[m]
elevation
0.1 edocesenesascasnsaceacaccnesessees ee eee eee ne Wyereenenee et eeaccceweneccenccencccnsccescscese! sucwsesceutesvandabaec=ccbsadeunscl

pure convection

{Eos 0.0001 0.001 0.01 0.1

time averaged concentration
Figure 6.5.3: Estimated time-averaged sediment concentrations based on the pure
convection solution (5.4.15) for pure waves and the complete solution (5.4.49)
with the diffusivity €s(z) given by Equation (6.5.14). See also Figure 5.4.10, p 260.

AND SEDIMENT TRANSPORT 289

Chapter 6: Sediment transport models

TRV ABA

Figure 6.5.4: Sediment suspension in large plunging waves at Whale Beach, Sydney,
Australia. Large amounts of sand are brought to the surface near "the plunge point"
and spread horizontally along the surface.

290 COASTAL BOTTOM BOUNDARY LAYERS

 6.5 Shore parallel sediment transport

which now has to be evaluated numerically, is increased roughly by a factor 2 to

about 6: 10°m7/s.
The fact that the complete solution "convection+diffusion" in Figure 6.5.3,
which is based on combined wave-current flow, agrees almost perfectly with the
experimental data shown in Figure 5.4.10, page 260, is coincidental. There was no
current present in those experiments.

6.5.3 Longshore sediment transport in the surf zone

The modelling of shore parallel sediment transport in the surf zone contains
numerous new challenges for a few decades into the future.
The daunting variability in sediment concentrations from point to point in
space and from wave to wave in time was illustrated by the field studies of Kana
(1979) and Nielsen (1984).
Some significant progress has been made with respect to the modelling of
turbulence due to spilling breakers (Peregrine & Svendsen 1978, Svendsen &
Madsen 1984 and Stive 1988), and this has been incorporated in gradient diffusion
models for suspended sediment by, for example, Deigaard et al (1986).
However, convective (or large scale) distribution mechanisms, such as the
strong upward flows which are generated by entrained air behind plunging
breakers, see Figure 6.5.4, are obviously important in many surf zones. Another
large scale (convective) mixing mechanism in the outer surf zone is provided by
the obliquely ascending vortices described by Nadaoka et al (1988). Where these
large scale mixing processes are important, a pure gradient diffusion approach will
not be adequate.
Comprehensive models of surf zone sediment transport must also include the
large contribution from the swash zone, see Kamphuis (1991).
While detailed models are thus still being developed, simpler approaches to
longshore sediment transport in the surf zone have been in use for some time.
The most famous of these is the so-called CERC formula which expresses the
total, longshore sediment transport rate as a constant times the longshore wave
energy flux at the break point. With the use linear shallow water wave theory this
can be written
rl
p(s-l)g|Qydx = 1 = cael Hf sin(20n)
7 = vd K ¥x¥ ay) 5
(6.5.15)
Xb

where the subscript "b” refers to the break point, y is the wave height to depth

AND SEDIMENT TRANSPORT 291

Chapter 6: Sediment transport models

ratio at the break point, a» the breaker angle, and = "ri" stands for the runup
limit. The quantity J is called the immersed weight longshore sediment transport
rate.
The value of the constant K for use in connection with the root mean square
breaker height for field data is about 0.77 (Shore Protection Manual 1984).
Recommended values of K and its scatter have recently been discussed by Bodge
& Kraus (1991) in relation to the presently available data.
One of the interesting aspects of the CERC formula is that it seems to work
quite well without considering the size of the sand - "The CERC Formula
Paradox" (Nielsen 1988b).
The weak grain size dependence was confirmed by Kamphuis (1990) who,
based on dimensional analysis and consideration of both field and laboratory data,
recommended
rl

J, te * ~1.25 fs 0.25
=
Hbrms/Tp
26-109 ||
Lop
m8 |EL
dso
sin™(208)
(6.5.16)
where mp is the beach slope at the break point, Lop is the deep water wave
length corresponding to the peak wave period Tp, and Hb,rms is the root mean
square breaker height. ©
This weak grain size dependence of the total longshore sediment transport
rate is interesting in relation to the observations of shore normal transport over
rippled beds discussed in Section 6.3.1, page 266, and to the corresponding shore
parallel transport rates discussed in Example 6.5.1, page 282. In both cases it was
found that sediment transport rates over rippled beds seems to be less sensitive to
grain size than classical transport models predict.
Classical transport models based on gradient diffusion and the assumption of
a flat bed, e g Deigaard et al (1986), predict a very strong grain size dependence,
for a given topography.
The most likely reason for the observed weak grain size dependence for the
total longshore sediment transport rate seems to lie in the topographical difference
between beaches of coarser and finer sand. Beaches of fairly coarse sand
(dso2 0.4mm) tend to develop topographies which are more conducive to
sediment transport. An example is shown in Figure 6.5.4. If the sand on that beach
had been finer, the beach profile would have been flatter and the waves would
have broken as spilling breakers with much less sediment entrainment capability.

292 COASTAL BOTTOM BOUNDARY LAYERS

 CHAPTER 7

LOOSE ENDS AND FUTURE

DIRECTIONS

7.1 Introduction
It has been the aim of the previous chapters to point out some of the
remaining, unanswered questions in coastal sediment transport modelling as well
as to summarise our present knowledge. Identifying loose ends or gaps in our
knowledge is the first step to any serious research project. Therefore, the following
sections summarise some of the presently unresolved problems and indicate areas
in which new insights are urgently needed.

7.2 New models of wave-current boundary layer interaction

The discussion of the concept of eddy viscosity for combined wave-current
flows in Section 1.5.3 (pp 68-71) leads to the conclusion that it is time to introduce
a new generation of eddy viscosity based models for these flows. The reason is
that, with the presently used definitions, the eddy viscosity which applies to the
current component of combined wave-current flows, has a number of undesirable
characteristics and is unlikely to be related to the sediment diffusivity.
Most of the presently available models which use the eddy viscosity concept
are based on the equation of motion

du
arerage (7.1)
7.1

for the current component. However, the eddy viscosity vc which is defined by
this equation has some rather unfortunate characteristics. This can be seen by
considering the expression

AND SEDIMENT TRANSPORT 293

Chapter 7: Loose ends and future directions

= pvtZ — puw — piw - pww’ (1.2.22)

for the time-averaged shear stress in a two-dimensional wave-current flow of the
form (u,w) = (U+utw, wt+wtw’), see page 12.
If this expression for 7 is inserted into Equation (7.1) it can be seen that the
eddy viscosity vc, which applies to the current, contains the following terms

ve + Vv (7.2)

The first deterministic term, (— uw), in this expression may often be ignored
since w_must, by continuity, be zero for horizontally uniform flows. The second
term, (—i w ), however, will often be dominant and this gives the eddy viscosity Vc
a problematic nature.
Hence, vc is not a "turbulent eddy viscosity", because the leading
contribution (iw) is deterministic.
It is not isotropic, but depends strongly on the direction of the current relative
to the direction of wave propagation.
___ The eddy viscosity given by Equation (7.2) may in fact become negative if
—uw is dominant and of the form, derived by Longuet-Higgins (1956), (see pages
55 and 56), while a is negative. This corresponds to the commonly observed
situation which is shown in Figure 7.1.
___ The bed shear stress is positive but smaller than the asymptotic value of
—u w. See also Figure 1.4.2, page 55.
The steady flow velocity is positive close to the bed, but starts to decrease at
the elevation where

-—puw = Xz) (7.3)

Above this point, the eddy viscosity vc, is negative because T and du/dz
have opposite signs.
A negative eddy viscosity is unsatisfactory in itself and even more so in
relation to sediment diffusivity. It cannot be avoided however, if the time-averaged
equation of motion is written in the commonly used form (7.1).
These conceptual problems with the eddy viscosity are, however, avoided if

294 COASTAL BOTTOM BOUNDARY LAYERS

 Chapter 7: Loose ends and future directions

_—Oo

(oe)

relative
z/5s
elevation

0
=1.9 -1 -0.5 0 0.5 1 1.5

Figure 7.1: In situations where the time-averaged bed shear stress is positive but
smaller than the asymptotic value of —iuw , the shear stress T(z) and the current
gradient a have opposite signs in the upper part of the flow. This corresponds to
negative values of the eddy viscosity defined by Equation (7.1).

the momentum transfer term -“w _ is considered explicitly rather than

anonymously as part of Vc é. When that is done, the equation of motion for the

current becomes

du UZ 7
Vel ‘de = asta] + Uw (7.4)

With this equation it is easier to explain the situation shown in Figure 7.1.
The new eddy viscosity, Vc1, which is given by (for uw = 0)

ee ee (7.5)

is, unlike vc, basically a turbulent eddy viscosity and it is unlikely to be strongly

AND SEDIMENT TRANSPORT 295

Chapter 7: Loose ends and future directions

anisotropic. It is also more probable that vc1 can be meaningfully related to the
sediment diffusivity. The latter is important with respect to sediment transport
modelling.
With the time-averaged equation of motion written in the form (7.4) it is also
possible to explain the decrease in current velocity near the suface which can be
seen in Figure 1.5.2, page 63. The situation in Figure 1.5.2 corresponds to a
positive bed shear stress (0) of the order —1.5p (UW). According to
Equation (7.4), with

uz) = to) (1—-z/D) (7.6)

the velocity gradient du/dz therefore turns negative at z =~ D/3 = 100mm.

The development of a wave-current boundary layer interaction model based
on Equation (7.4) for the current distribution is outlined briefly in Section 1.5.9
(pp 91-94), but the details are left for the future.

7.3 Hydraulic roughness of flat sand beds under waves

We have at present no direct information, in the form of detailed velocity
measurements, about the hydraulic roughness of flat sand beds under waves, and
two types of indirect experimental evidence are apparently conflicting. That is,
measured bed-load quantities and bed-load transport rates over flat sand beds
under waves indicate much smaller effective bed shear stresses than do energy
dissipation measurements. See Section 2.4.2 and 2.4.3 (pp 117-120).
Correspondingly, it was found in Section 3.6.3 and 3.6.4 (pp 149-154), that
the roughness values, found by Wilson (1989), for flat sand beds in steady flows
were much smaller than the roughness derived from energy dissipation over flat
sand beds under waves. Wilson found typical roughness values of the order ten
grain diameters while the energy dissipation data of Carstens et al (1969)
correspond to roughness values of the order one hundred grain diameters.
Thus, energy dissipation measurements under waves indicate much greater
hydraulic roughness for flat, movable beds than other types of experiments.
No direct measurements are available of the wave boundary layer flow over
flat, movable sand beds. It is therefore an open question as to whether the large
roughness inferred from energy dissipation experiments or the smaller roughness
corresponding to sheet-flow sediment transport rates should be applied in
modelling the structure of these flows.
It is possible, that the larger roughness indicated by the energy dissipation
experiments can be explained in terms of the energy dissipation due to percolation

296 COASTAL BOTTOM BOUNDARY LAYERS

 Chapter 7: Loose ends and future directions

under waves, which is not directly related to the effective bed shear stress. It
would, however, be very interesting to see some direct measurements of the flow
structure over flat beds of loose sand under oscillatory flows.

7.4 Instantaneous, effective shear stresses on sand beds under waves

As mentioned in the previous section, we have at present no direct
measurements of oscillatory boundary layer flow structure over movable sand beds
and the various types of indirect evidence are apparently conflicting. It is therefore
impossible to estimate instantaneous, effective bed shear stresses for these flows
with much confidence.
The instantaneous, effective bed shear stresses are, however, essential for
detailed sediment transport modelling. Hence, there is an urgent need for
experimental data in this area.
It might be possible to obtain the necessary boundary layer structure data
with acoustic velocity probes. Alternatively, the method applied by Lofquist
(1986) to measure instantaneous bed shear stresses over rippled beds might be used
for flat beds as well.

7.5 The bottom boundary condition for suspended sediment distributions

The bottom boundary condition for suspended sediment distributions presents
another urgent problem in sediment transport modelling.
It is by now fairly widely agreed that the description of suspended sediment
distributions in non-uniform or unsteady flows requires a different bottom
boundary condition than just a quasi-steady version of

c(o) = c(0’) (7.7)

This equation, which is the usual type of boundary condition for steady,
uniform flow, expresses equilibrium between suspended sediment concentrations
at the bed and the effective Shields parameter 9’. Such an equilibrium does,
however, not exist in non-uniform or unsteady flows, see Section 5.3.1 (pp
222-223). A different boundary condition has therefore been suggested by Nielsen
et al (1978) and van Rijn (1984).
Their approach is to treat deposition and entrainment separately, and then try
to relate the entrainment rate, or pickup rate p(t), directly to the effective,
instantaneous Shields parameter 9’(t).
The pickup function model is, however, still a simplified, formal description

AND SEDIMENT TRANSPORT 297

Chapter 7: Loose ends and future directions

rather than an explicit, physical description of the sediment entrainment process. It

is, therefore, not obvious how it should be incorporated into the combined
convection-diffusion model of sediment suspensions as discussed on page 247.
To resolve this type of problem, a clear physical description is needed for the
processes in the layer between the immobile bed and elevations where all sediment
moves in suspension according to Bagnold’s definition. For the situation shown in
Figure 5.2.1, page 208, that is for the elevation range -Smm < z < 7mm.

7.6 Distribution models for suspended sediment

A framework for the modelling of sediment suspensions in coastal flows as a
combined convection-diffusion process was developed qualitatively in Sections
5.4.5-5.4.9 (pp 245-261).
The new model was found capable of modelling observed differences
between sediment distributions in different flows and between the distributions of
different grain sizes in the same flow. However, many quantitative details need to
be filled in.
One of the advantages of the new combined convection-diffusion approach is
that the sediment diffusivity €s, which is only responsible for the truly diffusive
part of the distribution process in this description, may be closely related to "the
eddy viscosity" of the flow. That is often not the case when a natural sediment
distribution is modelled in terms of pure gradient diffusion.
Inverted commas are used in relation to "the eddy viscosity" because, the
eddy viscosity may be defined in various ways in a combined wave-current flow.
Of the two eddy viscosities discussed in Section 7.2, the second, Vci, seems a far
better model for the sediment diffusivity €s.
However, even if equality, €s = Vci, may be assumed, the diffusivity is
known no better than the eddy viscosity, and for flat sand beds that is not very
well. The problem of properly describing the sediment diffusivity is therefore
linked to the uncertainty about the hydraulic roughness of flat sand beds under
waves mentioned in Section 7.3 above.
The most direct way to obtain some conclusive information about these
matters, is to measure the concentration profiles of different sand sizes together
with the flow structure. This is most urgently needed for oscillatory flows over flat
sand beds and for combined wave-current boundary layer flows.
After the present manuscript was finished, the paper by Dick & Sleath (1991)
appeared, which contains considerable new insights into the problems mentioned
in Sections 7.3, 7.4 and 7.5.

298 COASTAL BOTTOM BOUNDARY LAYERS

 REFERENCES

Allen, J R L (1982): Sedimentary structures: Their character and physical basis. Elsevier,
Amsterdam.
Amos, C L & MB Collins (1978): The combined effect of wave motion and tidal currents
on the morphology of intertidal ripple marks: the Wash. U K J Sedimentary Petrology
Vol 48, No 3 , pp 849-856.
Amos, CL, A J Bowen, D A Huntley & C F M Lewis (1988): Ripple generation under the
combined influences of waves and currents on the Canadian continental shelf.
Continental Shelf Res, Vol 8, No 10, pp 1129-1153.
Amott, R W & J B Southard (1990): Exploratory flow duct experiments on combined flow
bed configurations and some implications for interpreting storm event stratification. J
Sedimentary Petrology, Vol 60, No 2, pp 211-219.
Asano, T, M Nagakawa & Y Iwagaki (1986): Changes in current profiles due to wave
superimposition. Proc 20th Int Conf Coastal Engineering, Taipei, pp 925-940.
Bagnold, R A, (1946): Motion of waves in shallow water: Interaction of waves and sand
bottoms, Proc Roy Soc Lond, A 187, pp 1-15.
Bagnold, R A (1954): Experiments on a gravity-free dispersion of large solid spheres in a
newtonian fluid under shear. Proc Roy Soc Lond A 225, pp 49-63.
Bagnold, R A, (1956): The flow of cohesionless grains in fluids. Phil Trans Roy Soc Lond,
No 964, Vol 249, pp 235-297.
Bailard, J A (1981): An energetics total load sediment transport model for a plane sloping
beach. J Geophys Res, Vol 86, No C11, pp 10938-10954.
Bakker, W T, & T van Doorn, (1978): Near bottom velocities in waves with a current. Proc
16th International Conference on Coastal Engineering, ASCE, Hamburg, pp 1394-1413.
Basset, A B, (1888): On the motion of a sphere in a viscous fluid. Phil Trans Roy Soc Lond,
Ser A, Vol 179, pp 43-63.
Bhattacharya, P K, (1971): Sediment suspension in shoaling waves, Ph D thesis, University
of Iowa, Iowa City, Iowa.
Bijker, E W (1967): Some considerations about scales for coastal models with movable
beds. Delft Hydraulics Lab Pub No 50.
Blondeaux, P & G Vittori (1990): Oscillatory flow and sediment motion over a rippled bed.
Proc 22nd Int Conf Coastal Engineering, Delft, pp 2186-2199.
Boczar-Karakiewicz, B & R D L Davidson-Arnott (1987): Nearshore bar formation by
non-linear wave processes - A comparison of model results and field data. Marine
Geology, Vol 77, pp 287-304.
Boczar-Karakiewicz, B & L A Jackson (1991): Beach dynamics and protection measures in
the Gold Coast area, Australia. Proc 10th Australasian Conf on Coastal and Ocean
Engineering, Auckland, pp 411-415.

AND SEDIMENT TRANSPORT 299

 References

Bodge, K R & N C Kraus (1991): Critical examination of longshore sediment transport rate
magnitude. Proc " Coastal Sediments ’91", AS CE, pp 139 - 155.
Bosman, J J (1982): Concentration measurements under oscillatory water motion. Delft
Hydraulics report M1695 part 2.
Bosman, J J, ET J M van der Velden & C H Hulsbergen (1987): Sediment concentration
measurements by transverse suction , Coastal Engineering, Vol 11, pp 353-370.
Bretschneider , C L (1954): Field investigation of wave energy loss of shallow water ocean
waves, Beach Erosion Board, Tech Memo 46.
Brevik, I, (1981): Oscillatory rough turbulent boundary layers, J Waterway Port Coastal
Ocean Div. ASCE, 107 , pp 175-188.
Bruun , P (1962):Sea level rise as a cause of shore erosion. Proc A SC E,, Vol 88, WW1, pp
117-130.
Bruun, P (1983): Review of conditions for uses of the Bruun Rule of erosion. Coastal
Engineering , Vol 7, pp 77-89.
Bruun, P (1990): Port Engineering, IV edition, Vol 2, Gulf Publishing Company.
Cacchione, D A & D E Drake (1982): Measurements of storm generated bottom stresses on
the continental shelf. J Geophys Res, Vol 87, No C3, pp 1952-1960.
Carstens, M R, FM Neilson & H D Altinbilek (1969): Bedforms generated in the laboratory
under an oscillatory flow, CE RC Tech Memo 28.
Carter T G, P L-F Liu, and C C Mei (1973): Mass transport by waves and offshore
bedforms. J Waterways Harbours & Coastal Division, A S C E, Vol 99, WW2, pp
165-184.
Christoffersen, J B & I G Jonsson (1985): Bed friction and dissipation in combined current
and wave motion, Ocean Eng, Vol 12, No 5, 387-423.
Clift, R, R J Grace, & M E Weber, (1978): Bubbles, drops and particles. Academic Press,
New York
Clifton, H E (1976): Wave-formed sedimentary structures - A conceptual model. In Davis &
Ethington ed: Beach and nearshore sedimentation. S EP M special publication 24, pp
126-148.
Coffey, F C (1987): Current profiles in the presence of waves and the hydraulic roughness
of natural sand beds. Ph D Thesis, Univ of Sydney, 252 pp.
Coffey, F C & P Nielsen (1984): Aspects of wave-current boundary layer flows. Proc 19th
Int Conf Coastal Engineering, Houston Texas, pp 2232-2245.
Coffey, F C & P Nielsen (1986): The influence of waves on current profiles. Proc 20th Int
Conf Coastal Eng, Taipei, pp 82-96.
Coleman, N L, (1970): Flume studies of the sediment transfer coefficient, Water Resource
Res, 6(3), pp 801-809.
Dean, RG & R A Dalrymple (1991): Water wave mechanics for engineers and scientists.
World Scientific, Singapore.
Deigaard, R, J Fredsoe & I B Hedegaard (1986a): Suspended sediment in the surf zone. J
Waterway, Port, Coastal and Ocean Eng, ASCE, Vol 112, No 1, pp 115-128

300 COASTAL BOTTOM BOUNDARY LAYERS

 References

Deigaard, R, J Fredsoe & I B Hedegaard (1986b): A mathematical model for littoral drift. J
Waterway, Port, Coastal and Ocean Eng, ASCE, Vol 112, No 3, pp 351-369.
Deigaard, R, P Justesen & J Fredsoe (1991): Modelling undertow by a one-equation
turbulence model. Coastal Engineering, Vol 15, pp 431-458.
Delft Hydraulics (1989): Bedforms, near-bed sediment concentrations and sediment
transport in simulated regular wave conditions. Report No H840.
Dick, JE & J FA Sleath (1991): Velocities and concentrations in oscillatory flow over beds
of sediment. J Fluid Mech, Vol 233, pp 165-196.
Dingler, J R (1974): Wave formed ripples in nearshore sands. Ph D thesis Univ of
California, San Diego, 136 pp.
Downing, J P, R W Sternberg & C R B Lister (1981): New instrumentation for the
investigation of sediment suspension processes in shallow marine environments. Marine
Geology, Vol 42, pp 14-34.
Du Toit, CG, & JF A Sleath, (1981): Velocity measurements close to rippled beds in
oscillatory flow, J Fluid Mechanics 112, pp 71-96
Engelund,F (1970): Instability of erodible beds. J Fluid Mech, Vol 42, pp225-244.
Engelund, F (1981): Transport of bed load at high shear stress. Progr Rep 53, Inst
Hydrodynamic and Hydraulic Eng, Tech Univ Denmark, 31-35.
Engelund, F & E Hansen (1972): A monograph on sediment transport in alluvial streams.
Teknisk Forlag, Copenhagen.
Fernandez-Luque, R (1974): Erosion and transport of bed-load sediment. Dissertation,
KRIPS Repro BV, Meppel, The Netherlands.
Fredsoe, J (1984): Turbulent boundary layer in wave-current motion, J Hydr Eng, ASC E,
Vol 110, HY8, 1103-1120.
Gibbs, R J, M D Mathews & D A Link (1971): The relationship between sphere size and
settling velocity. J Sed Petrology, Vol 41, pp 7-18.
Gilbert, G K (1914): The transportation of debris in running water. U S Geol Survey, Prof
Paper 86.
Grant, W D, & O S Madsen, (1979): Combined wave and current interaction with a rough
bottom, JGeophysical Res, 84, 1808
Grant, W D, & O S Madsen, (1982): Movable bed roughness in unsteady oscillatory flow. J
Geophys Res, Vol 87, pp 469-481.
Grant, W D, J A Williams, S M Glen, D A Cacchione & D E Drake (1983): High frequency
bottom stress variability and its prediction in the Code Region. Woods Hole Oceanogr
Inst, Tech Rep 83-19.
Guy, H P, DB Simons, & E V Richardson (1966): Summary of alluvial channel data from
flume experiments, 1956-1961. U S Geological Survey, Prof paper 462-I, Washington
DC.
Hallermeier R J (1980): Sand motion initiation by water waves: Two asymptotes. J
Waterway, port,coastal and ocean Div, A SC E, Vol 106, 299-318.
Hallermeier R J (1981): A profile zonation for seasonal sand beaches from wave climate.
Coastal Engineering, Vol 4, pp 253-277.

AND SEDIMENT TRANSPORT 301

 References

Hallermeier, R J (1981b): Seaward limit of signficant sand transport by waves. CETA 81-2,
CERC, Ft Belvoir, Va.
Hammond, T M & MB Collins (1979): On the threshold of transport of sand-sized sediment
under the combined influence of unidirectional and oscillatory flows. Sedimentology,
Vol 26, pp 795-812.
Hanes, D M (1990): The structure of events of intermittent suspension of sand due to
shoaling waves, in The Sea, Vol 9 Part B, John Wiley & Sons, pp 941-954.
Hanes, D M & A J Bowen (1985): A granular fluid model for steady intense bed-load
transport. J Geophys Res, Vol 90, No CS, pp 9149-9158.
Hanes, D M & D A Huntley (1986): Continuous measurements of ‘suspended sand
concentration in wave dominated nearshore environment. Continental Shelf Res, Vol 6,
No 4, pp 585-596.
Hanes, D M & D L Inman (1985a): Observations of rapidly flowing granular-fluid
materials. J Fluid Mech, Vol 150, 357-380.
Hanes, D M & D L Inman (1985b): A dynamic yield criterion for granular-fluid flows. J
Geophys Res, Vol 90, No BS, 3670-3674.
Hansen, J B & I A Svendsen (1986): Experimental investigation of the wave and current
motion over a longshore bar. Proc 20th Int Conf Coastal Eng, Taipei, pp 1166-1179.
Hardistry, J & J P Lowe (1991): Experiments on particle acceleration in stationary and
oscillating flows. J Hydraulic Engineering, ASC E, Vol 117.
Hattori, M, (1969): The mechanics of suspended sediment due to standing waves, Coastal
Engineering Japan, 12, 69-81
Hedegaard, I B, R Deigaard & J Fredsoe (1991): Onshore/offshore sediment transport and
morpho- logical modelling of coastal profiles. Proc Coastal Sediments ’91, Seattle, pp
643-657.
Ho, H W, (1964): Fall velocity of a sphere in an oscillating fluid. PhD thesis, University
Iowa, Iowa City, Iowa
Homma, M, K Horikawa & R Kajima (1965): A study of suspended sediment due to wave
action. Coastal engineering in Japan, Vol 8, pp 85-103.
Horikawa, K, & A Watanabe, (1968): Laboratory study on oscillatory boundary layer flow,
Coastal Engineering Japan 11, 13-28
Horikawa, K, A Watanabe & S Katori (1982): Sediment transport under sheet flow
condition. Proc 18th Int Conf on Coastal Eng, Capetown, pp 1335-1352.
Inman, D L (1957): Wave generated ripples in nearshore sands. Beach Erosion Board, U S
Army Corps of engineers, Tech Memo 100.
Inman, D L & A J Bowen (1963): Flume experiments on sand transport by waves and
currents. Proc 8th Int Conf Coastal Engineering, A S C E, pp137-150.
Iwagaki Y & T Kakinuma, (1967): On the bottom friction factors off five Japanese coasts,
Coastal Eng Japan, Vol 10, 13-22.
Jackson, P S (1981): On the displacement height in the logarithmic velocity profile. J Fluid
Mech Vol 111, pp 15-25.

302 COASTAL BOTTOM BOUNDARY LAYERS

 References

Jackson, R G (1976): Sedimentological and fluid-dynamic implications of the turbulent

bursting phenomenon in geophysical flows. J Fluid Mech, Vol 77, pp 531-560.
Jansen, R H J (1978): The in situ measurement of sediment transport by means of ultrasound
scattering. Delft Hydraulics Lab Publication No 203.
Jensen, B J (1989): Experimental investigation of turbulent oscillatory boundary layers.
Series Paper 45, Institute of Hydrodynamics and Hydraulic Engineering (ISVA),
Technical University of Denmark.
Jonsson, I G (1966): Wave boundary layers and friction factors, Proc 10th Int Conf Coastal
Eng, Tokyo, 127-148.
Jonsson, I G (1980): A new approach to oscillatory rough turbulent boundary layers, Ocean
Engineering 7, 109-152
Jonsson, I G (1990): Wave current interactions, in The Sea, Vol 9 Part A, John Wiley &
Sons, pp 65-120.
Jonsson, IG & N A Carlsen, (1976): Experimental and theoretical investigations in an
oscillatory turbulent boundary layer, J Hydraulic Res, 14, 45-60
Justesen, P (1988): Turbulent wave boundary layers, Series Paper 43, Inst Hydrodynamic
& Hydraulic Eng , Tech Univ Denmark,
Kajiura, K, (1968): A model of the bottom boundary layer in water waves, Bulletin
Earthquake Res, Institute, 46, 75-123.
Kalkanis, G (1957): Turbulent flow near an oscillating wall. Beach Erosion Board, Tech
Memo 97.
Kalkanis, G (1964): Transportation of bed material due to wave action. US Army CERC,
Tech Memo 2.
Kamphiis, J W, (1975): Friction factors under oscillatory waves, J Waterway Harbours
Coastal Engineering Division, ASCE, 101, 135-144.
Kamphuis, J W (1990): Littoral transport rate. Proc 22nd Int Conf Coastal Eng, Delft,
ASCE, pp 2402- 2415.
Kamphuis, J W (1991): Alongshore sediment transport rate distribution. Coastal Sediments
"91, ASCE , pp 170 - 183.
Kana , T W (1979): Suspended sediment in breaking waves. Tech rep 18-CRD, Department
of Geology, University of South Carolina.
Kemp, P H, & R R Simons, (1982): The interaction between waves and a turbulent current:
waves propagating with the current, J Fluid Mechanics 116, 227-250.
Kemp, PH, & R R Simons, (1983): The interaction between waves and a turbulent current:
waves propagating with the current, J Fluid Mechanics 130, 73-89.
Kennedy J F (1963): The mechanics of dunes and antidunes in erodible-bed channels. J
Fluid Mech, Vol 16, pp 521-544.
Kennedy, J F & M Falcon (1965): Wave-generated sediment ripples. M J T, Hydrodynamics
Laboratory Report No 86, 55pp.
King, D B Jr (1991): Studies in oscillatory flow bedload sediment transport. Ph D Thesis,
University of California, San Diego (Scripps), 183pp.

AND SEDIMENT TRANSPORT 303

 References

Kline, S J, W C Reynolds, F A Schraub & P W Rundstadler (1967): The structure of

turbulent boundary layers. J Fluid Mech, Vol 30, 741-777.
Kraus, N C, M Larson & D L Kriebel (1991): Evaluation of beach erosion and accretion
predictors. Proc "Coastal Sediments 91", A S C E, pp 572-587.
Lamb, H (1936): Hydrodynamics, 6th ed, Cambridge Univ Press
Lambrakos, K F, D Myrhaug & O H Slaattelid (1988): Seabed current boundary layers in
wave-plus-current flow conditions. ASCE, J Waterway Port Coastal and Ocean Eng, Vol
114, No2, pp 161-174.
Le Mehaute, B (1976): An introduction to hydrodynamics and water waves. Springer.
Lian, Qi Xiang (1990): A visual study of the coherent structure of the turbulent boundary
layer in flow with adverse pressure gradient. J Fluid Mech, Vol 215, 101-124.
Lofquist K E B (1978): Sand ripple growth in an oscillatory flow water tunnel. C E R C tech
paper 78-5.
Lofquist, K E B (1980): Measurements of oscillatory drag on sand ripples, Proc 17th Int
Conf Coastal Eng, Sydney, 3087-3106.
Lofquist, K E B (1986): Drag on naturally rippled beds under oscillatory flows, Misc Paper
CERC-86-13.
Longuet-Higgins, M S (1953): Mass transport in water waves. Phil Trans Roy Soc Lond, Vol
245 A, pp 535-581.
Longuet-Higgins, M S (1956): The mechanics of the boundary-layer near the bottom in a
progressive wave, Proc 6th Int Conf Coastal Eng, Miami, 184-193.
Longuet-Higgins, M S (1981): Oscillating flow over steep ripples. J Fluid Mech, Vol 107,
pp 1-35.
Longuet-Higgins, M S (1983): Wave setup, percolation and undertow in the surf zone. Proc
Roy Soc Lond, Vol A 390, pp 283-291.
Lundgren, H, (1972): Turbulent currents in the presence of waves, In Proc. 13th Conference
on Coastal Engineering, Vancouver, ASCE, 623-634.
Lundgren, H & T Soerensen (1956): A pulsating water tunnel. Proc 6th Int Conf Coastal
Engineering, Miami.
McFetridge, W F & P Nielsen (1985): Sediment suspension by non-breaking waves over
rippled beds. Tech Rep UFL/COEL-85/005, Coastal & Oceanographical Engineering
Dept, Univ of Florida, Gainesville.
Madsen, OS & W D Grant (1976): Sediment transport in the coastal environment. Report
No 209, Palph M Parsons Lab, MIT.
Magnus, G, (1853): Poggendorfs Annalen der Physik und Chemie, Vol 88, No 1.
Manohar, M (1955): Mechanics of bottom sediment movement due to wave action. Tech
Memo 75, Beach Erosion Board, U S Army corps of engineers, Washington D C.
Meyer-Peter, E & R Muller (1948): Formulas for bed-load transport. Proc Int Ass Hydr
Struct Res, Stockholm.
Miller, M C & P D Komar (1980): A field investigation of the relationship between ripple
spacing and near-bottom water motions. J Sed Petrology, Vol 50, pp 183-191.

304 COASTAL BOTTOM BOUNDARY LAYERS

 References

Murray, S P (1970): Settling velocities and vertical diffusion of particles in turbulent water,
J Geophysics Res. Vol 75, No 9, 1647-1654.
Myrhaug, D (1982): On a theoretical model of rough turbulent wave boundary layers.
Ocean Eng, Vol 9, No 6, pp 547-565.
Myrhaug, D & O H Slaattelid (1989): Combined wave and current boundary layer model for
fixed, rough seabeds. Ocean Engineering Vol 16, No 2, pp 119-142.
Nadaoka, K, S Ueno & T Igarashi (1988): Sediment suspension due to large eddies in the
surf zone. Proc 22nd Int Coastal Eng Conf, Torremolinos, pp 1646-1660.
Nakato, T, F A Locher, J R Glover, and J F Kennedy (1977): Wave entrainment of sediment
from rippled beds, Proc. ASCE, 103 (WW1), 83-100.
Natarajan, P (1969): Sand movement by combined action of waves and currents. Ph D thesis,
University of London.
Nielsen, P, (1979): Some basic concepts of wave sediment transport, Ser. Paper 20, Inst
Hydrodyn Hydraul Eng, Tech Univ Denmark, 160 pp.
Nielsen, P (1981): Dynamics and geometry of wave generated ripples. J Geophys Res, Vol
86, No C7, pp 6467-6472.
Nielsen, P (1983): Entrainment and distribution of different sand sizes under water waves. J
Sedimentary Petrology, Vol 53, No 2, pp 423-428.
Nielsen, P (1983): Analytical determination of nearshore wave height variation due to
refraction, shoaling and friction. Coastal Engineering, Vol 7, pp 233-251.
Nielsen, P (1984a): On the motion of suspended sand particles. J Geophys Res, Vol 89, No
C1, pp 616-626.
Nielsen, P (1984b): Field measurements of time-averaged suspended sediment
concentrations under waves. Coastal Engineering, Vol 8, pp 51-72.
Nielsen, P, (1985): On the structure of oscillatory boundary layers, Coastal Engineering 9,
261-276.
Nielsen, P (1986): Suspended sediment concentrations under waves. Coastal Engineering,
Vol 10, pp 23-31.
Nielsen , P (1988a): Three simple models of wave sediment transport. Coastal Engineering,
Vol 12, pp 43-62.
Nielsen, P (1988b): Towards modelling coastal sediment transport. Proc 21st Int Conf
Coastal Eng, Torremolinos, pp 1952-1958.
Nielsen, P (1990): Coastal bottom boundary layers and sediment transport. In P Bruun ed
Port Engineering (4th edition), Vol 2, pp 550-585.
Nielsen, P, I A Svendsen & C Staub (1978): Onshore-offshore sediment transport on a
beach. Proc 16th Int Conf Coastal Eng, Hamburg, pp 1475-1492.
Nielsen, P, N R Sena & ZJ You (1990): The roughness height under waves. J Hydraulic
Res, Vol 28, No 5, pp645-647.
Nikuradse, J (1933): Stromungsgesetze in glatten und rauhen rohren. V D J Forschungsheft
361, Berlin.

AND SEDIMENT TRANSPORT 305

 References

Owen, P R (1964): Saltation of uniform grains in air. J Fluid Mech, Vol 20, No 2, pp
225-242.
Peregrine, J H & I A Svendsen (1978): Spilling breakers, bores and hydraulic jumps. Proc
16th Int Conf Coastal Eng, Hamburg, pp 540-551.
Rasmussen, P & J Fredsoe (1981): Measurements of sediment transport in combined waves
and current. Inst Hydrdyn & Hydraul Eng (ISVA), Tech Univ Denmark, Progress report
53, pp 27-30.
Raudkivi, A J (1988): The roughness height under waves. J Hydraulic Res, Vol 26, No 5,
pp569-584.
Reizes, J A, (1977) A numerical study of the suspension of particles in a horizontally
flowing fluid, paper presented at 6th Australasian Hydraulics and Fluid Mechanics
Conference, Adelaide .
Ribberink, J S and A Al-Salem (1989): Bed forms, near-bed sediment concentrations and
sediment transport in simulated regular wave conditions. Delft Hydraulics Report H840,
part 3.
Ribberink, J S and A Al-Salem (1990): Bed forms, sediment concentrations and sediment
transport in simulated wave conditions. Proc 22nd Int Conf Coastal Engineering, Delft,
pp 2318-2331.
Riedel, H P (1972): Direct measurement of bed shear stress under waves. Ph D Thesis, Dept
Civ Eng, Queen’s University, Kingston, Ontario.
Roelvink, J A & M J F Stive (1989): Bar generating cross shore flow mechanisms on a
beach. J Geophys Res , Vol 94, No C4, pp 4785 - 4800.
Sato, S (1986): Oscillatory boundary layer flow and sand movement over ripples. Ph D
thesis, Dept of Civ Eng, Univ of Tokyo.
Sato, Y & K Yamamoto (1987): Lagrangian measurement of fluid particle motion in an
isotropic turbulent field, J Fluid Mech, Vol 175, 183-199.
Sawamoto, M & T Yamashita (1986): Sediment transport rate due to wave action. J of
Hydroscience and Hydraulic Eng, Vol 4, No 1, pp1-15.
Schepers , J D (1978): Zandtransport onder invloed van golven en een eenparige stroom bij
varierende korreldiameter. M Eng Thesis, Delft Univ of Technology.
Schlichting, H (1979): Boundary layer theory, 7th ed, McGraw-Hill, New York.
Shields, A (1936): Anwendung der Aehnlichkeitsmechanik und Turbulenzforchung auf die
Geschiebebewegung. Mitt Preuss Versuchsanstalt fur Wasserbau und Schiffbau, No 26,
Berlin.
Shore Protection Manual (1984), U S Army Coastal Engineering Research Centre,
Vicksburg Mississippi.
Simons, R R, A J Grass & A Kyriacou (1988): The influence of currents on wave height
attenuation. Proc 21st Int Conf Coastal Engineering, Malaga.
Slaattelid, OH, D Myrhaug & K F Lambrakos (1990): North Sea bottom steady boundary
layer measurements. J Waterway, Port, Coastal and Ocean Engineering, Voll16, No 5,
pp 614-633.

306 COASTAL BOTTOM BOUNDARY LAYERS

 References

Sleath, J F A (1970): Measurements close to the bed in a wave tank. J Fluid Mech, Vol 42,
pp 111-123.
Sleath, J F A (1978): Measurements of bed-load in oscillatory flow. Proc A SC E, Vol 104,
No WW4, 291-307.
Sleath, J F A (1982): The suspension of sand by waves. J Hydraulic Res, Vol 20, No 5, pp
439-452.
Sleath, J F A (1984): Sea Bed Mechanics, Wiley Interscience.
Sleath, J F A (1985): Energy dissipation in oscillatory flow over rippled beds, Coastal Eng,
Vol 9, 159-170.
Sleath, J F A (1987): Turbulent oscillatory flow over rough beds, J Fluid Mech, Vol 182,
369-409.
Sleath, J F A (1990): Bed friction and velocity distributions in combined steady and
oscillatory flow. Proc 22nd Int Conf Coastal Eng, Delft, pp 450-463.
Sleath, J F A (1991): Velocities and shear stresses in wave-current flows. J Geophys Res,
Vol 96, No C8, pp 15237-14244.
Snyder, W H & J L Lumley (1971): Some measurements of particle velocity autocorrelation
functions in a turbulent flow, J Fluid Mech, Vol 48, 41-71.
Southard, J B, JM Lambie, D C Federico, H T Pile, & C R Weidman (1990): Experiments
on bed configurations in fine sands under bidirectional purely oscillatory flow, and the
origin of hummocky cross-stratification. J Sedimentary Petrology, Vol 60, No 1, pp 1-17.
Spalart, PR & B S Baldwin (1987): Direct simulation of a turbulent oscillating boundary
layer, NASA Tech Memo 89460, Ames Res Centre, Moffett Field, Ca.
Staub, C, IG Jonsson & I A Svendsen (1984): Variation of sediment suspension in
oscillatory flow. Proc 19th Int Conf Coastal Eng, Houston, pp 2310-2321.
Stive, M J F (1988): Cross-shore flow in waves breaking on a beach. Dissertation, Delft
University of Technology.
Svendsen, I A & P A Madsen (1984): A turbulent bore on a beach. J Fluid Mech, Vol 148,
pp73-96.
Svendsen, I A, H A Schaffer & J Buhr Hansen (1987): The interaction between the
undertow and the boundary layer flow on a beach. J Geophys Res, Vol 92, pp
11845-11856.
Swart , D H (1974): Offshore sediment transport and equilibrium beach profiles. Delft Hydr
Lab Publ No 131.
Taylor, G I (1921) Diffusion by continuous movement, Proc Lond Math Soc, Vol 20,
196-211.
Tooby, P F, G L Wick, & J D Isacs, (1977): The motion of a small sphere in a rotating
velocity field: A possible mechanism for suspending particles in turbulence, J
Geophysical Res, 82 (15), 2096-2100.
Trowbridge, J & O S Madsen (1984): Turbulent wave boundary layers I. Model formulation
and first order solution. J Geophys Res, Vol 89, No C5, pp 7989-7997.
van Doorn, T, (1981): Experimental investigation of near-bottom velocities in water waves
without and with a current. TOW Report M 1423 part 1, Delft Hydraulics Laboratory .

AND SEDIMENT TRANSPORT 307

 References

van Doorn, T, (1982): Experimenteel onderzoek naar het snelheidsveld in de turbulente

bodemgrenslaag in een oscillerende stroming in een golftunnel, TOW-ReportM 1562-1a,
Delft Hydraulics Laboratories.
van Doorn, T, (1983): Computations and comparisons with experiments of the bottom
boundary layer in an oscillatory flow. TOW-Report M 1562-2, Delft Hydraulics
Laboratories.
van Rijn, L C (1984): Sediment pickup functions. J Hydraulic Eng, Vol 110, No 10, pp
1494-1502.
van Rijn, L C (1984): Sediment transport, Part III: Bedforms and alluvial roughness. J
Hydraulic Eng, Vol 110, No12, pp 1733-1754.
van Rijn, L C (1986): Applications of sediment pickup function. J Hydraulic Eng, ASC E,
Vol 112, No 9, pp 867-874.
Vincent, C E & M O Green (1990): Field measurements of the suspended sand
concentration profiles and fluxes and of the resuspension coefficient Yo over a rippled
bed. J Geophys Res, Vol 95, No C7, pp 11591-11601.
Wilson, K C (1966): Bed-load transport at high shear stress. J Hydraulics Div, A S C E, Vol
92, No HY6, pp 49-59.
Wilson, K C (1989): Mobile bed friction at high shear stress. J Hydraulic Eng, AS C E, Vol
115, No 6, pp 825-830.
Wright, L D, P Nielsen, N C Shi & J H List (1986): Morphodynamics of a bar-trough surf
zone. Marine Geology, Vol 70, pp 251-285.
Yalin, M S (1977): Mechanics of sediment transport, 2nd Ed. Pergamon Press, London, 312
pp.
Yalin, S & RCH Russell (1962): Similarity in sediment transport due to waves. Proc 8th
Int Conf Coastal Eng, Mexico City, pp 151-167.

308 COASTAL BOTTOM BOUNDARY LAYERS

 AUTHOR INDEX

A\-Salem, A 121, 131, 136, 278 35, 36, 37, 41, 44, 46, 48
Carter, TG 59
Allen, JRL 151
Carstens, MR 4, 5, 27, 28, 29, 45,
Altinbilek, H D 4, 5, 27, 28, 29, 45,
105, 106, 135, 139, 140, 141, 142,
105, 106, 107, 135, 139, 140, 141,
146, 147, 152, 153, 156, 157, 158,
142, 146, 147, 152, 153, 156, 157,
159, 296
158, 159, 296
CERC 291, 292
Amos, CL 108, 143, 144
Christoffersen, J B 73, 75, 81, 85
Arnott, R W 143, 145
Clift, R 167
Asano, T 75, 83, 88
Clifton, HE 131, 136
Coffey, FC 69, 70, 74, 82, 83, 143,
Bagnold, R A 6, 25, 27, 28, 96, 97, 145
98, 109, 110, 111, 114, 115, 116, Coleman, N L 204
135, 157, 207 Collins, M B 108, 143
Bailard, JA 121
Bakker, W T 51, 52
Dairymple, R A 28, 146
Baldwin, BS 52
Davidson-Arnott ,R DL 277
Basset, A B 167, 170
Dean, R G 28, 146
Bhattacharya, PK 161, 176
Deigaard, R 61, 277, 291, 292
Bijker, E W 84, 143
Delft Hydraulics 140, 141, 142,
Blondeaux, P 41
216, 218, 259
Boczar-Karakiewicz, B 277
Dick, JE 298
Bodge, KR 292
Dingler, JR 137, 140, 141, 142
Bosman, J J 209, 216
Downing, J P 209
Bowen, AJ 108, 115, 143, 144, 267
Drake, DE 83
Bretschneider, C L 28
Du Toit, CG 228
Brevik, I 38, 48, 49, 50
Bruun, P 109
Engelund, F 104, 105, 112, 134,
Cacchione, DA 83
145, 149
Carlsen, N A 4, 14, 15, 26, 27, 32,

AND SEDIMENT TRANSPORT 309

Author index

Falcon, M 140, 141 140, 267

Isacs, JD 161, 181, 183
Federico, DC 131
Iwagaki, Y 28, 75, 83, 88
Fernandez-Luque, R 224
Fredsoe, J 41, 61, 81, 84, 277, 285,
286, 291, 292 J ackson,L A 277
Jackson,PS 66
Jackson, RG 245
Gibbs, RJ 165, 166, 226
Jansen, RH J 209
Gilbert, GK 113 Jensen, BJ 4, 18, 26,37, 38, 41,42,
Glen, SM 83 45, 46, 47, 48
Glover, JR 212, 213, 214, 215, 241 Jonsson, I G 3, 4, 14, 15, 21, 23, 25,
Grace, R J 167
26, 27, 29, 32, 35, 36, 37, 41, 44,
Grant, W D 48, 49, 50, 73, 79, 81, 46, 48, 62, 73, 75, 77, 81, 85, 212,
82, 83, 105, 107, 108, 154 241
Grass, AJ 25, 28, 75, 76
Justesen, P 24, 52, 61
Green, MO 211
Guy H P 148, 149
Kajima, R 241
Kajiura, K 23, 25, 29, 41, 48
Hatlermeier, R J 108, 109
Kakinuma, T 28
Hammond, TM 108
Kalkanis, G 42
Hanes, DM 95, 96, 98, 115, 211
Kamphuis, J W 23, 24, 25, 26, 291,
Hansen, E 104, 105, 149
292
Hansen, JB 61
Kana, TW 291
Hardistry, J 170
Katori, S 119, 207, 208, 209, 211,
Hattori, M 161
212, 217, 241, 278
Hedegaard, I B 277, 291, 292
Kennedy, JF 134, 140, 141, 212,
Ho, HW 167, 174, 177, 178
213, 214, 215, 241
Homma, M 241
Kemp, P H 4, 26, 28, 75, 81, 82, 83,
Horikawa, K 32, 33, 119, 139, 207,
88,93
208, 209, 211, 212, 217, 241, 278
King, D B Jr 119, 122, 126, 127, 276
Hulsbergen, CH 209
Kline, SJ 101
Huntley, D A 108, 143, 144, 211
Komar, PM 137
Kraus, N C 277, 292
Igarashi, T 211, 291 Kriebel, D L 277
Inman, D L_ 95, 96, 98, 135, 137, Kyriacou, A 25, 28, 75, 76

310 COASTAL BOTTOM BOUNDARY LAYERS

 Author index

Lamb, H 100 Nadaoka, K 211, 291

Lambie, JM 131 Nagakawa, M 75, 83, 88
Lambrakos, K F 83 Nakato, T 212, 213, 214, 215, 241
Larson, M 277 Natarajan, P 108, 143
Le Mehaute, B 8, 10 Neilson, FM 4, 5,27, 28, 29, 45,
Lewis, CFM 108, 143, 144 105, 106, 107, 135, 139, 140, 141,
Lian, Q X 101 142, 146, 147, 152, 153, 156, 157,
Link, DA 165, 166, 226 158, 159, 296
List, JH 145 Nielsen, P 35,42, 43, 47, 69, 70, 74,
Lister, CR B 209 $2, 105, 135; 136; 137,'1399140,
Liu, P L-F 59 141, 142, 143, 145, 154, 159, 204,
Locher, FA 212, 213, 214, 215, 241 205, 21052159 217,.219; 2209223,
Lofquist, K EB 4, 5, 15, 27, 28, 29, 225, 228/229, 230; 2317232)243,
45, 105, 106, 135, 146, 147, 152, 245, 256, 265, 268, 269, 270, 272,
156, 157, 158, 159, 297 274, 276, 291, 292, 297
Longuet-Higgins, MS 40, 54, 56, Nikuradse, J 66, 149
57, 60, 94, 294
Lowe, JP 170
Owen, PR 154
Lumley, JL 194
Lundgren, H 6, 79, 81, 85, 89
Peregrine, JH 291
Pile, HT 131
McFetridge, W F 205, 220, 232
Madsen, OS 48, 49, 50, 73, 79, 81,
82, 105, 107, 108, 154 Rasmussen, P 285, 286
Madsen, P A 291 Raudkivi, AJ 159
Magnus, G 167 Reizes, JA 183
Manohar, M 107 Reynolds, WC 101
Mathews, MD 165, 166, 226 Ribberink, JS 121, 131, 136, 278
Mei, CC 59 Richardson, E V 148, 149
Meyer-Peter, E 112, 113, 266, 279, Riedel, H P 26
285 Roelvink, JA 61
Miller, MC 137 Rundstadler, P W 101
Muller, R 112, 113, 266, 279, 285 Russel, RCH 140, 141
Murray, S P 189, 198, 199
Myrhaug, D 38, 49, 50, 73, 83, 85,

AND SEDIMENT TRANSPORT si

Author index

Sato, S 194, 267 Taylor, GI 190, 193, 195, 196, 199

Sawamoto, M 117, 118, 119 Tooby,
PF 161, 181, 183
Schaffer, H A 61 Trowbridge, J 48
Schepers, JD 267, 273, 274, 275
Schlichting, H 66, 150 Ueno, § 211, 291
Schraub, FA 101
Sena, NR 159
Shi, NC 145 Van der Velden, ETJ M 209
Shields, A 103, 107 van Doom, T 4, 37, 39, 46, 48, 51,
Simons, DB 148, 149 52, 62, 63, 69, 70, 71, 72, 74, 81,
Simons, RR 4, 25, 26, 28, 75, 76, 83, 88
81, 82, 83, 88, 93 van Rijn, LC 149, 223, 224, 225,
Slaattelid, OH 73, 83, 85 266, 297
Sleath, J FA 4, 9, 17, 18, 25, 26, Vincent, CE 211
28, 29, 32, 33, 34, 38, 41,42, 46, Vittori, G 41
48, 49, 50, 52, 60, 71, 73, 74, 75,
78, 81, 82, 83, 84, 85, 88, 89, 90, Watanabe,A 32, 33, 119, 139,
91, 119, 120, 125, 128, 135, 146,
207, 208, 209, 211, 212, 217, 241,
151, 209, 212, 228, 236, 253,
278
298
Weber,ME 167
Snyder, WH 194
Weidman,
CR 131
Soerensen, T 6
Wick, GL 161, 181, 183
Southard, JB 131, 143, 145
Williams, JA 83
Spalart, PR 52
Staub, C 212, 223, 241, 243, 297
Wilson, K C 112, 113, 149, 151,
153, 154, 159, 296
Sternberg, R W 209
Wright, LD 145
Stive, MJ F 61, 291
Svendsen, 1A 61, 212, 223, 241,
243, 291, 297 Yalin,MS 140, 141
Swart, DH 25, 76, 105 Yamashita, T 117, 118, 119
Yamamoto, K 194
You, Z-J 159

312 COASTAL BOTTOM BOUNDARY LAYERS

 SUBJECT INDEX

Absorption instruments 208-209 vertical scale 111-112

waves 116-128, 296
Accretion 131, 277-278
Bed roughness, see Hydraulic
Acoustic velocity probe 297
roughness
Added mass 7, 99
Bed shear stress
Alongshore sediment transport
combined flows 57, 89-90, 279,
280-292
280, 283, 286, 289
Angle of repose 95, 138
effective 104, 107-109, 115, 123,
Antidunes 133
297
Angular frequency 2
instantaneous 120, 124, 279
Armouring effect 232
oscillatory flow 15-16, 21, 23-27
steady flows 57, 64-65
Backscatter instruments 208-209 peak value 15-16, 21, 23-27
Bagnold number 97 time-average 12, 56-57, 61,
Bar formation 58-60 89-90, 279-280, 289
Bars 58-60, 129, 135, 277 wave-current 57, 89-90, 279
Basset term 167, 170 Bedforms 129-160,
Beach accretion/erosion 277-278 see also Ripples and Megaripples
Bed friction, see Bed shear stress combined flow 143-145
Bed-load 109-128 growth rate 134-135
amount of 110-112, 117, 296 migration rate 132-135
concentration 111, 207-208 regimes 129-131
definition 109-110, 206-207 Biological activity 103, 130
flux 112 Bioturbation 103, 130
Meyer-Peter formula 112, 266, Boundary layer 1-94
279, 285 combined wave-current 62-94
steady flow 112-115 current 52-62, 64-68, 78-93
transport rate, steady 112-115 current, influence on wave
transport rate, unsteady 116-128, boundary layer 73-75
296 drift velocity 53-60
velocity 114-115 eddy viscosity based models
48-52, 70, 84-87, 293

AND SEDIMENT TRANSPORT 313

Subject Index

higher order turbulence models 52 260, 286-289

laminar sublayer 31, 66 over ripples 228, 256,
length scales 1, 29-30, 38, 43, 47, Concentration measurements 208
32, 25 0. Concentration of sediment
mixing length models 51 bed-load 111, 207-208
models 40-52 linear 96
oscillatory 1, 13-52, 62, 236-237 see also Suspended sediment
overlap layer 49 concentrations
overshoot 15-17 Concentration profiles
quasi-steady models 41-42 breaking waves 219, 290,
steady currents 52-6, 64-68, different grain sizes 220
78-93 flat beds 208, 217,
streaming 54 irregular waves 218
thickness 1, 9, 29-30 ripples 214, 216,
turbulent 42-48, 57 shape parameter 216-218, 254
velocity distribution models 42-48 vertical scale 217, 254
wave 1, 13-52, 62, 236-237 Concentration time series 209
waves, influence on current Conservation of sediment 131-134,
boundary layer 78-93 223, 234-235, 238, 241, 246
wave-current 62-94 Constant stress assumption 68,
Bruun rule 109 92-93
Buoyancy 98, 102 Constant stress layer 68
Continuity equation
for fluid 12
Centrifugal force 180, 187, 189
for sediment 131-134, 223,
CERC formula 291-292
234-235, 238, 241
CERC formula paradox 268, 292
Convection process 205, 211, 219,
Combined convection-diffusion
221, 233-241, 245-250, 252-253,
distribution model 205, 245-261
270
Combined wave-current flow
Convection-diffusion distribution
the structure of 62-94, 293-296
model 205, 245-261
bed shear stress 108, 279
Convection velocity
bedforms 143-145
sediment 237, 253, 270
initiation of motion 108
turbulence 19, 236
sediment transport 279-289
Convective distribution function
Concentration magnitude
237, 239, 253-255, 261, 298
over flat beds 208, 217, 226, 228,

314 COASTAL BOTTOM BOUNDARY LAYERS

 Subject Index

Convective distribution model 205, Diffusion equation 19, 203, 236, 241
238-242, 244-245, 249-251, Diffusive distribution model
253, 260-261 241-245, 248, 257-258
Convective sediment flux 234-236, Diffusive sediment flux 202-203,
238-239, 241, 246-247 234-236, 241, 243, 246-247
Convective transport model 270-276 Diffusivity 201-205, 236, 241, 243-
Current 245, 252-255, 258, 269, 289, 298
boundary layer concepts 64-68 Dispersion 190, 193-200
eddy viscosity 36, 67-68, 84-94, Dispersive stress 96, 111, 207
293-296 Displacement thickness 29-30
Eulerian drift 53-54, 56-57 Dissipation factor, see Energy
influence on wave boundary layer dissipation factor
73-75 Distribution function 237, 252-253
Lagrangian drift 54-60 Distribution models 233-261, 298
mass transport 54-60 Drag coefficient 100, 165
profile 62-64, 78-94, 293-296 Drag force 7, 98-104, 107, 110, 147,
shear stress distribution 64 162, 164, 167, 169, 171-172, 180
undertow 60-61 Drift velocity
wave-current boundary layer Eulerian 53-54, 56-57
62-94 Lagrangian 54-60
wave-generated 52-61 Dunes 186
Critical Shields parameter 107-108,
144-145
Eddy viscosity
Critical stress 108
based models 48-52, 70, 84-87,
293
Darcy’s law 102 combined flow 68-71, 92,
Decay, wave height 77 293-296
Defect function 19-22, 29-30, oscillatory flow 31-40, 57, 68, 71
38-40, 43-47,49-52, 74 relation with sediment diffusivity
Defect velocity 7, 19, 29-30, 152 204-205
Deposition 131, 222-223, 239 steady flow 36, 67-68
Depth of closure 109 turbulent component 32-34
Diffusion 190, 193-205, 219, 221, wave-current flow 68-71, 92,
233-236, 241-261 293-296
Diffusion-convection distribution Effective normal stress 95-98,
model 205, 245-261 110-111

AND SEDIMENT TRANSPORT 315

Subject Index

Effective shear stress 104, 107-109, settling 235

11.5$c12352255297 Force
Energy dissipation factor 25, 27-28, buoyancy 98, 102
75, 147-148, 153 centrifugal 180, 187, 189
Energy flux 76 disturbing 103
Entrainment 211, 221-233, drag 7, 98-104, 107, 110, 147,
245-246, 297, see also 162, 164, 167, 169, 171-172, 180
Convection, Diffusion and form drag 100, 104
Pickup function gravity 107
Entrainment level 237 inertial 7
Equation of motion intergranular 95-98, 110
for the fluid 8-10, 12, 61, 293 lift 101, 107
for suspended particle 167-169, pressure 8, 57, 98-100, 103
171, 173, 188 stabilising 103
Equivalent Nikuradse roughness 24, Forced vortex 182
66, 129, 145, 150, see also Form drag 100, 104
Hydraulic roughness Free stream velocity 3, 14
Eulerian drift velocity 53-54, 56-57 Frequency response function 23, 125
Eulerian time scale of turbulence 193 Friction factors 23-27, 146-149
Erosion 131 grain roughness 105-106
laminar 21
oscillatory flow 21, 23-30, 41,
Fall velocity, see Settling velocity
124, 147
Flow regimes
rippled beds 158
smooth 66
sand beds 146-149
smooth laminar 19
smooth beds 21, 24
rough turbulent 25-26, 66
steady flow 149
smooth turbulent 24
wave 21, 23-30, 41, 124, 147
transition 24
Friction velocity
Flux
current 64-65, 93, 103-104, 151,
bed-load 112
279, 283, 286
convective 235-236
oscillatory flow 18, 25-26, 30, 35
diffusive 203, 235-236
Frictional coefficient 95-96
momentum 12-15, 55-57, 91-94
Froude number 145
sediment 131-134, 201-203, 213,
231, 234-236,238-239,241, 243,
245-247

316 COASTAL BOTTOM BOUNDARY LAYERS

 Subject Index

Gamma ray absorption 208-209 Initiation of motion 102-109

Intergranular forces 95-98, 110
Grab and dump transport model
Inviscid velocity distributions 53-54
265, 273-276, 285, 292
Irrotational vortex 184
Gradient diffusion 201-205, 236,
see also Diffusion
Gradient diffusion distribution Keulegan Carpenter number 101,
model 241-245, 248, 257-258 103
Gradient diffusion transport model Kinematic viscosity, see viscosity
269-270, 273-276, 291
Grain Reynolds number 107
Lagrangian drift velocity 54-60
Grain roughness friction factor
Lagrangian integral scale 189,
105-106
194-195
Grain roughness Shields parameter
Lagrangian microscale 194-195
105-107, 138-140
Laminar
Ground water 102
flow 14, 19-23, 29-30, 38-39, 43
length scale 30
Hydraulic roughness 3-5, 84, sublayer 31, 66
145-160, see also Equivalent velocity defect function 38-39
Nikuradse roughness Law of the wall 65-66, 84
apparent increase 78-79, 81-84, Lift force 101, 107
89, 93 Linear sediment concentration 96
combined wave-current flow 84 Littoral drift 276, 289
components 149 Logarithmic velocity distribution
flat sand beds 154-155 65-66, 149-151
from velocity profile 66 Loitering effect 190-193, 195-200
grain roughness 105 Longshore sediment transport
Nikuradse 24, 66, 129, 145, 150 280-292
oscillatory flow 4, 146, 152-160 Lunate megaripples 131
rippled sand beds 155-159
steady flow 66, 146, 149-152
Macroviscous regime 97-98
Hydrodynamic mass 100
Mass transport velocity 54-60
Mean surface slope 57
Incipient motion 102-109 Megaripples 130-131, 135, 145,
Inertial force 7 210-211, 277
Inertial regime 97-98

AND SEDIMENT TRANSPORT

B17
Subject Index

Meyer-Peter formula 112, 120, 266, Phase average 11

279, 285 Phase lag (see also time lag)
Mixing length 38, 52, 84, 201-203, for bed-load transport 120, 128
216-217, 233-234, 236 for pickup function 227
Mobility number 103 for suspended concentration 211,
influence on ripples 136-137, 212 240, 244, 251
140-143 Phase shift (see also phase lag)
Momentum flux 12-15, 55-57, 91-94 for bed shear stress 14, 120, 124,
Momentum transfer 11-15, 55-56, 127
91-94, 146, 149, 156, 158, Pickup function 205, 222-233,
293-295 236, 239, 297-298
aw 55-56, 91-94, 293-295 different sizes from graded bed
230-233
irregular waves 226-227
Navier Stokes equation 8, 11, 52
rippled beds 228
Newton’s formula 19, 31, 67
sine waves 225-226
Nikuradse roughness 4, 24, 66, 84,
steady flow 223-224
129, 145, 150, 153, 218, see also
unsteady flows 224-225
Hydraulic roughness
van Rijn’s 223, 266, 297
Normal stress 11, 95-98, 101,
vortex ripples 228-230
110-111, 207
waves over flat beds 225-228
Pore water 95
Oscillating water tunnel 6, 14 Prandtl’s mixing length 52, 84,
Oscillating plate 7 Pressure force 8, 57, 98-100, 103
Oscillatory boundary layers 1-2, Pressure gradient 60, 64, 92
13-40, 40-52, 62, 236-237 Progressive waves 60
models 40-52
Outflow velocity 102
Quasi-steady boundary layer
Overlap layer 49
models 41-42
Overshoot 15-17
Quasi-steady sediment transport
models 116-117, 121-128
Particle trajectory model 263-266, Quicksand 102
270
Percolation 28, 146, 154, 296
Radian frequency 2
Periodic component 11
Permeability 101-102
Radiation stress 57, 60-61

318 COASTAL BOTTOM BOUNDARY LAYERS

 Subject Index

Random components 11 243, 245-247

Rankine vortex 185-186 initiation of motion 102-109
Rayleigh distribution 278 mobility 102-109
Relative bed roughness 3, 5 velocity 163
Relative density 164 Seepage 101-102
Reynolds equation 10-13 Settling flux 235
Reynolds number 3, 4, 6, 23, 38, Settling rate 222-223, 235
100, 108, 124 Settling velocity
grain 107 terminal 164-166, 170-171
particle 165 reduction 161-162, 174, 176-178,
Reynolds stresses 11, 17, 34, 55, 68 180, 199-200
Ripple decay of deviations 170
growth 133-135 Setup 60-61
height 14, 142-143, 218 Shape parameter for suspended
length 135-137 sediment distributions 216, 254
megaripples 130-131, 145, Shear stress
210-211 amplitude 17-18, 225
migration 133-135 bed, see Bed shear stress
roughness 155-160 combined flow 12, 57, 89-90,
steepness 138-141 108, 279, 280, 283, 286, 289,
vortex 130-131, 135-143, 160, 293-294
212-216, 228-230, 256-259, constant stress assumption 68,
266-276 92-93
Roughness, see Hydraulic roughness critical 108, 225
distribution 1, 17-18, 19, 21,64,
67, 294-296
Scour rates 223-224
effective (bed) 104, 107-109,
Sediment - see also Suspended
115,423,297
Sediment, Bed-load, Sheet-flow,
estimation of 90, 158, 279
Transport models and Entrainment
form drag 100, 104
“deposition 222-223 frequency response function 23,
diffusivity 201-205, 236, 241,
125
243-245, 252-255, 258, 269,
instantaneous, effective 120, 124,
289, 298
279
distribution models 233-261, 298
laminar 67
flux 112, 132-134, 201-203, 213,
231, 234-236, 238-239, 241,

AND SEDIMENT TRANSPORT 319

Subject Index

Shear stress (continued) grain roughness 105-107, 138-140

maximum 15-16, 21, 23-28, 55, skin friction 104
104, 108 time series 125, 279
oscillatory flow 15-16, 21, 23-27 total 103, 106, 232
over vortex ripples 16, 40-41, Simple harmonic flows 2-3, 14-15,
147, 228 19, 295355
peak 15-16, 21, 23-28, 55, 104, Skin friction 104
108 Sorting 230-233
periodic component 13 Specific density 164
phase shift 14-15, 21, 32, 51, Standing waves 58, 60
91-93 Steady flow
skin friction 104 boundary layer concepts 64-68
steady flows 57, 64-65 eddy viscosity 36, 67-68
time-averaged 12, 56-57, 61, Eulerian drift 53-54, 56-57
64, 89-90, 104, 108, 279-280, influence on wave boundary
283, 286, 289 layers 73-75
time series 124-125, 279 Lagrangian drift 54-60
vortex ripples 16, 40-41, 147, 228 mass transport 56
wave-current flow 12, 57, 89-90, profiles 62-68, 78-84
108, 279, 280, 283, 286, 289, profiles in the presence of waves
293-294 62-64, 78-94
Shear velocity, see Friction velocity shear stress distribution 64
Sheet-flow 109-128, 225, 296 undertow 60-61
steady 112-113 wave-generated 52-61
unsteady 116-128, 225, 296 Stokes drift 54, 153
sediment velocity 114-115 Stokes law 100, 165
Shields criterion 108 Stokes length 29-30, 43, 47, 55
Shields diagram 107-108 Stokes waves 53-60
Shields parameter 103-104 Streaming 54
amplitude 225-229 Stress, total mean 11, see also
combined flows 144-145, 279- Dispersive stress, Effective
280 normal stress, Effective shear
critical 107-108, 144-145, 223, stress, Normal stress, Reynolds
225-226 stress, Shear stress
effective 104, 125, 223, 225-228, Suction samplers 208-209
297

320 COASTAL BOTTOM BOUNDARY LAYERS

 Subject Index

Surf zone sediment transport 249-251, 253, 260-261

277-278, 291-292 convective sediment flux
Surface slope 57 234-236, 238-239, 241, 246-247
Suspended particles (motion of) convective transport model
161-200 270-276
accelerated flow 171-174 decay rates 215
accelerating in resting fluid 170 definition 206
dispersion model 193-200 deposition rate 222-223, 239
entrainment level 237 diffusion 190, 193-205, 219,
equation of motion 167-169, 171, 221, 233-236, 241-250, 261
173, 188 diffusive distribution model
loitering effect 190-193, 195-200 241-245, 248,257-258
oscillatory flow 174-180 diffusive sediment flux 202-203,
resting fluid 170 234-236, 241, 243, 246-247
settling 189-200 diffusivity 201-205, 236, 241,
time lag 176 243-245, 252-255, 258, 269,
trapping 181-189 289, 298
turbulence 18-19, 89-200, 236 dispersion 190, 193-200
vortices 181-189 dispersion model 193-200
Suspended sediment concentrations distributions 206-221, 239
201-261 distribution function 237, 252-253
bottom boundary condition 208, distribution mechanisms, see
222-223, 247, 297-298 Diffusion and Convection
breaking waves 219 distribution models 233-261, 298
combined flow 280-281, 289-291 entrainment 211, 221-233,
conservation equation 223, 234- 245-246, 297
235, 241, 246 entrainment level 237
convection 205, 211, 219, 221, exponential profile model 228
233-241, 245-250, 252-253, 270 flat beds 208, 215, 217, 228,
convection-diffusion model 205 259-261, 286-291
245-261, flux 131-134, 201-203, 213, 231
convection velocity 237, 253, 270 234-236, 238-239, 241, 243,
convective distribution function 245-247
237, 239, 253-255, 261, 298 irregular waves 211, 218, 226,
convective distribution model 229-230
205, 238-242, 244-245, measured distributions 208,

AND SEDIMENT TRANSPORT 321

Subject Index

215-221, 226-227 Transport models 263-292

Suspende sediment concentrations convective transport model
(continued) 270-276
measuring equipment 208-209 cu-integral models 263-264
megaripples 210-211 diffusion based model 269-270,
mixing length 201-203, 233-234, 273-276, 291
236 grab and dump model 265,
models 233-261, 298 273-276, 285, 292
oscillatory sheet flow 208, 217, particle trajectory models
226-227 263-266, 270
pickup functions, see Pickup Transport rate
functions bed-load 112, 116, 117, 121,
profile 206-221, 239, 282-286 126, 223, 266, 279, 285, 296
286-289 flat bed 276-280, 286-291
profile shape parameter 216-218, rippled bed 266-276, 282-286
254 sheet-flow 112-115, 116-128,
reference concentration 217, 278-280, 296
224-228, 230, 250, 258, 260-261 shore normal 266-280
regular waves 211-215, 218, 228 shore parallel 280-292
settling rate 202, 216-217 surf zone 277-278, 291-292
time series 209 suspended 201, 203, 213,
vertical scale 216-217, 234-236, 263-265, 266-280,
vortex ripples 212-216, 218, 280-292
220, 228-230, 256-259, 266-276 swash zone 262, 276-277
wave-current flows 280-289 Trapping 181-189
Swash zone sediment transport 276- Turbulence
277; 291 boundary layer 42-48, 57
decay 18, 38
Taylor’s dispersion model 190, Eulerian length scale 197
193-195 Eulerian time scale 193
intensity 18, 38, 71-73
Theoretical bed level 66
Lagrangian integral scale 194
Tidal boundary layer 2
Lagrangian micro scale 194
Time lag for sediment particle 176,
179 models for oscillatory boundary
layers 52
Total concentrations 259

322 COASTAL BOTTOM BOUNDARY LAYERS

 Subject Index

Turbulence (continued) Viscous dissipation 28

wave-current boundary layers Viscous regime 97-98
61,71-73 Viscosity (kinematic)
Turbulent effective (granular flow) 115
fluctuations 17, 71-73 laminar 8, 19, 97, 236
eddy viscosity 31-40, 67-71 see also Eddy viscosity
von Karman’s constant 33, 65
Undertow 60-61, 277 Vortex 162, 181
forced 182
uw 55-56, 91-94, 293-295
irrotational 184
Rankine 185-186
Van Rijn’s pickup function 223, Vortex ripples 130-131, 143
266 formation and growth 135
Velocity defect 7, 19, 29-30, 152 height 142-143
Velocity defect function 19-22, hydraulic roughness 155, 157-158
29-30, 38-40, 43-47, 49-52, 74 length 135-137
Velocity distribution pickup function over 228-230
boundary layer drift 54-60 steepness 138-141
Eulerian drift 53-54, 56-57 suspended sediment over 135,
Lagrangian 54-60 212-216, 256-259, 282-286
logarithmic 65-66, 149-151 Vortex trapping 181-189
mass transport 54-60
oscillatory flows 19-23, 42-48,
Wash load 109, 206
62-63
Wave boundary layer, see Boundary
steady flow 62-68
layer and Oscillatory boundary
steady component of combined
layer
flow 62-64, 78-94
Wave-current flows
undertow 60-61
bed shear stress 108, 279
wave-current flows 62-64, 78-94,
bed forms 143-145,
293-296
initiation of motion 108
Velocity distribution models for
sediment transport 279-289
oscillatory flow 42-48
structure of 62-94, 293-296
Velocity gradient 32
Wave energy dissipation 27, 75
Velocity overshoot 15-17
Wave energy dissipation factor 25,
Vertical decay scale 17
27-28, 75, 147,-148, 153
Virtual mass 100

AND SEDIMENT TRANSPORT

323
Subject Index

Wave friction factor 23-27, 28, X-ray absorption 208-209

29-30, 41, 153
Wave heights 278
Wave height attenuation 75-78 Yield criterion 111
Wave momentum flux 64
Wave number 55
Wind shear stress 57, 61, 64

324 COASTAL BOTTOM BOUNDARY LAYERS

, <
J 7
Lvl 2h-Id : i
¥ =, / -—"

seston. 471% _—
= - a
ABOUT THE AUTHOR

Peter Nielsen is a Senior Lecturer in

Civil Engineering at The University of
Queensland, Brisbane, Australia. He
received his M of Eng from The Technical
University of Denmark in 1976 and his
Ph.D. in Coastal Engineering from the
same institution in 1979. Since then, he
has been working as a lecturer and a
researcher at The University of Sydney
and The University of Florida.
His current research interests range
from basic hydrodynamics through
theoretical and experimental coastal
hydrodynamics to coastal sediment
transport processes.
Advanced Serieson Ocean Engineering — Vol.3
MECHANICS OF COASTAL SEDIMENT TRANSPORT.
by Jorgen Fredsoe & Rolf Deigaard (Tech. Univ. Denmark)
The main objective of the book is to describe from a rministic point
ofview the sediment transport in the general wave-current situation. For
this purpose, the bock is divided into two major parts:
The first part of the book is related to the flow and turbulence in com-
bined wave-current. This part covers the turbulent wave boundary layer,
bed friction in combined wave-current
motion, turbulence
in the surf
zone, andwave-driven
currents in the long- and cross-shore direction.
The second part treats the sediment transport as a result of the wave-
current action. This part includes an introduction to basic sediment trans-
port concepts, distribution of suspended sediment in the sheet flow regime,
description of bedforms formed by current and waves, and their in-
fluence on sediment transport pattern. Finally, the modelling of cross-
and long-shore sediment transport is described. This book is useful for
students with a background in basic hydrodynamics.
Contents: Basic Concepts of Potential Wave Theory; Wave Boundary
Layers; Bed Friction and Turbulence in Wave-current Motion; Waves in
the Surf Zone; Wave-driven Current Velocity Distribution in the Surf Zone;
Basic Concepts of Sediment Transport; Vertical Distribution of Suspended
Sediment in Waves and Currents over a Plane Bed; Current-generated
Bed Waves; Wave-generated Bed Forms; Cross-shore Sediment Trans-
port and Coastal Profile Development; Long-shore Sediment Transport

ISBN 981 - 02 -0472-8