River Engineering 1-7

L.C.
van Rijn Lecture Notes
Principles of Sediment Transport in

Rivers, Estuaries, Coastal Seas and Oceans
Chapter 1 - 7
For those who like sediments
CONTENTS
1 INTRODUCTION ...................................... 1.1

1.1 - Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.
.1
1.2 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.3
1.3 Symbols and units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1.4
1.4 Characteristic parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5
2 FLUID VELOCITIES AND BED SHEAR STRESSES ............... 2.1
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.1
2.2 Currents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1
.
2.2.1 Current boundary layer . . . . . . . . . . . . . . . . . . . . . . . . . . .2.1
2.2.2 Hydraulic regimes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.2
2.2.3 Velocity distribution over the depth . . . . . . . . . . . . . . . . . . .2.4
2.2.4 Fluid mixing coefficient . . . . . . . . . . . . . . . . . . . . . . . . . .2.4
2.2.5 Bed-shear stress and bed friction . . . . . . . . . . . . . . . . . . . . . 2.5
2.3 Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.7
2.3.1 Near-bed orbital velocities . . . . . . . . . . . . . . . . . . . . . . . . . 2.7
2.3.2 Wave-boundary layer . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.9
2.3.3 Hydraulic regimes in waves . . . . . . . . . . . . . . . . . . . . . . . 2.10
2.3.4 Velocity distribution in wave boundary layer . . . . . . . . . . . . . 2.11
2.3.5 Bed-shear stress and bed friction . . . . . . . . . . . . . . . . . . . . 2.15
2.3.6 Breaking waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.18
2.3.7 Mass transport in non-breaking waves . . . . . . . . . . . . . . . . . 2.20
2.3.8 Mass transport by breaking waves . . . . . . . . . . . . . . . . . . . 2.28
2.4 Combined current and waves . . . . . . . . . . . . . . . . . . . . . . . . . . .2.28
2.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.28
2.4.2 Wave characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.29
2.4.3 Current velocities and bed-shear stresses . . . . . . . . . . . . . . . 2.29
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2.48
3 FLUID AND SEDIMENT PROPERTIES ...................... 3.1
3.1 Fluid properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1
3.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.1
3.1.2 Fluid density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.1
3.1.3 Fluid viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.1
3.2 Sediment properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.3
3.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.3
3.2.2 Density and porosity . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.5
3.2.3 Shape . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.7
3.2.4 Size . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.9
3.2.5 Particle fall velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.11
3.2.6 Angle of (natural) repose . . . . . . . . . . . . . . . . . . . . . . . . . 3.17
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3.19
4 . INITIATION OF MOTION ............................... 4.1

4.1 Initiation of motion in currents . . . . . . . . . . . . . . . . . . . . . . . . . . . .4.1
4.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1
4.1.2 Critical bed-shear stress . . . . . . . . . . . . . . . . . . . . . . . . . .4.1
4.1.3 Critical depth-averaged velocity . . . . . . . . . . . . . . . . . . . . . 4.14
4.1.4 Design of stable channels . . . . . . . . . . . . . . . . . . . . . . . . . 4.16
4.1.5 Examples and problems . . . . . . . . . . . . . . . . . . . . . . . . . . 4.19
CONTENTS (continued)
4.2 Initiation of motion in waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.23

4.2.1 Critical velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.23
4.2.2 Critical bed-shear stress . . . . . . . . . . . . . . . . . . . . . . . . . 4.23
4.3 Initiation of motion for combined currents and waves . . . . . . . . . . . . . 4.27
4.4 Initiation of suspension in currents . . . . . . . . . . . . . . . . . . . . . . . . 4.31
4.4.2 Critical depth-averaged velocity . . . . . . . . . . . . . . . . . . . . . 4.32
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.34
BEDFORMS ......................................... 5.1

5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1
5.2 Bed forms in unidirectional currents . . . . . . . . . . . . . . . . . . . . . . . . 5.1
5.2.1 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1
5.2.2 Shape and dimensions of bed forms . . . . . . . . . . . . . . . . . . . 5.7
5.2.3 Examples and problems . . . . . . . . . . . . . . . . . . . . . . . . . .5.22
5.3 Bed forms in non-steady currents . . . . . . . . . . . . . . . . . . . . . . . . .5.24
5.3.1 Non-steady river flow . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.24
5.3.2 Tidal flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5.28
5.4 Bed forms in waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5.32
5.4.1 Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.32
5.4.2 Shape and dimensions of bed forms . . . . . . . . . . . . . . . . . . 5.33
5.5 Bed forms in currents and waves . . . . . . . . . . . . . . . . . . . . . . . . . 5.43
5.5.1 Classification . . . . . . . . -. . . . . . . . . . . . . . . . . . . . . . . . 5.43
5.5.2 Shape and dimensions of bed forms . . . . . . . . . . . . . . . . . . 5.46
5.5.3 Examples and problems . . . . . . . . . . . . . . . . . . . . . . . . . .5.51
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5.53
6 EFFECTIVE BED ROUGHNESS ........................... 6.1

6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6.1
6.2 Current-related bed roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . .6.1
6.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6.1
6.2.2 Available methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6.1
6.2.3 Methods based on bed-form parameters . . . . . . . . . . . . . . . . . 6.1
6.2.4 Methods based on integral parameters . . . . . . . . . . . . . . . . . 6.11
6.2.5 Comparison of methods . . . . . . . . . . . . . . . . . . . . . . . . . . 6.15
6.3 Wave-related bed roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.18
. 6.3.1 Available method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.18
6.4 Bed roughness in combined currents and waves . . . . . . . . . . . . . . . . 6.23
6.4.1 Available method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.23
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6.26
7 BED MATERIAL SUSPENSION AND TRANSPORT IN STEADY

UNIFORM CURREmS .................................. 7.1
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.1
7.2 Bed-load transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7.2
7.2.1 Introduction and defiDitions . . . . . . . . . . . . . . . . . . . . . . . . 7.2
7.2.2 Characteristics of moving bed-load particles . . . . . . . . . . . . . . 7.4
7.2.3 Particle pick-up from the bed . . . . . . . . . . . . . . . . . . . . . . 7.13
7.2.4 Deterministic bed-load transport formulae . . . . . . . . . . . . . . . 7.20
7.2.5 Bed-load transport at low shear stress . . . . . . . . . . . . . . . . . 7.29
7.2.6 Bed-load transport at steep slopes . . . . . . . . . . . . . . . . . . . . 7.29
7.2.7 Bed-load transport of non-uniform material . . . . . . . . . . . . . . 7.33
7.2.8 Comparison of bed-load transport formulae . . . . . . . . . . . . . . 7.38
7.2.9 Stochastic bed-load transport formulae . . . . . . . . . . . . . . . . . 7.38
7.3 Suspended load transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.49
7.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.49
7.3.2 Mass balance equation suspended sediment . . . . . . . . . . . . . . 7.51
7.3.3 Fluid and sediment mixing coefficient . . . . . . . . . . . . . . . . . 7.53
7.3.4 Concentration profiles . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.55
7.3.5 Velocity profiles in lower regime . . . . . . . . . . . . . . . . . . . . 7.59
7.3.6 Reference concentration and reference level . . . . . . . . . . . . . 7.61
7.3.7 Suspended sediment size in case of non-uniform bed material . . . 7.65
7.3.8 Suspended load transport rates . . . . . . . . . . . . . . . . . . . . . 7.67
7.3.9 Stratification effects in high-concentration suspensions . . . . . . . 7.75
7.4 Total load transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.91
7.4.1 Prediction methods . . . . . . . . . . . . . . . . . . . . . . . . . . . .7.91
7.4.2 Comparison of methods . . . . . . . . . . . . . . . . . . . . . . . . . . 7.95
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7.105
8 BED MATERIAL SUSPENSION AND TRANSPORT IN WAVES ....... 8.1

8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8.1
8.2 Identification of transport processes . . . . . . . . . . . . . . . . . . . . . . . . 8.2
8.2.1 Non-breaking waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2
8.2.2 Breaking waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8.8
8.2.3 Vector presentation of fluxes . . . . . . . . . . . . . . . . . . . . . . 8.10
8.3 Analysis of measured concentration profiles and transport rates . . . . . . . 8.11
8.3.1 Instantaneous concentrations . . . . . . . . . . . . . . . . . . . . . . . 8.11
8.3.2 Time-averaged concentrations . . . . . . . . . . . . . . . . . . . . . . 8.15
8.3.3 Sediment load and transport rate . . . . . . . . . . . . . . . . . . . . 8.27
8.4 Computation of time-averaged concentration profiles . . . . . . . . . . . . . 8.35
8.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.35
8.4.2 Time-averaged convection-diffusion equation . . . . . . . . . . . . . 8.36
8.4.3 Particle size of suspended sediment . . . . . . . . . . . . . . . . . . 8.38
8.4.4 Sediment mixing coefficient for non-breaking waves . . . . . . . . 8.38
8.4.5 Sediment mixing coefficient for breaking waves . . . . . . . . . . . 8.46
8.4.6 Reference concentration in near-bed region . . . . . . . . . . . . . . 8.48
8.4.7 Methods for computation of time-averaged concentration profiles 8.5 1
8.5 Computation ofpdiment transport rates . . . . . . . . . . . . . . . . . . . . . 8.58

8.5.1 Introduktion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8.58
8.5.2 Sediment tramport models . . . . . . . . . . . . . . . . . . . . . . . . 8.58
8.5.3 Sediment transport formulae . . . . . . . . . . . . . . . . . . . . . . . 8.59
8.5.4 Influence of bed slope on bed-load transport . . . . . . . . . . . . . 8.67
8.6 Examples and problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.68
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.73
9 BED MATERIAL SUSPENSION AND TRANSPORT IN COMBINED
WAVES AND CURRENTS ................................ 9.1
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 9 . 1
9.2 Analysis of measured concentration profiles and transport rates . . . . . . . . 9. 1
9.2.1 Time-averaged concentration profiles . . . . . . . . . . . . . . . . . . 9.1
9.2.2 Sediment transport rates . . . . . . . . . . . . . . . . . . . . . . . . . .9.9
9.3 Computation of time-averaged concentration profiles . . . . . . . . . . . . . 9.16
9.3.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9.16
9.3.2 Comparison of measured and computed concentration profiles . . 9.17
9.4 Computation of sediment transport in non-breaking waves . . . . . . . . . . 9.24
9.4.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9.24
9.4.2 Comparison of measured and computed transport rates . . . . . . . 9.27
9.5 Computation of sediment transport in breaking waves . . . . . . . . . . . . 9.29
9.5.1 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9.29
9.5.2 Comparison of measured and computed transport rates . . . . . . . 9.33
9.6 Examples and problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.34
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9.37
10 BED MATERIAL TRANSPORT. EROSION AND DEPOSITION IN
NON-STEADY AND NON-UNIFORM FLOW ................... 10.1
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.1
10.2 Sediment transport in non-steady flow . . . . . . . . . . . . . . . . . . . . . . 10.1
10.2.1 River flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10.1
10.2.2 Tidal flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.1
10.3 Sediment transport in non-uniform conditions . . . . . . . . . . . . . . . . 10.10
10.3.1 General . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.10
10.3.2 Erosion and scour near structures . . . . . . . . . . . . . . . . . . . 10.14
10.3.3 Deposition in channels . . . . . . . . . . . . . . . . . . . . . . . . . 10.19
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.29
11 TRANSPORT OF COHESIVE MATERIAIS ................... 11.1

11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1
11.2 Cohesion. plasticity. viscosity and yield stress . . . . . . . . . . . . . . . . . 11.2
11.3 Flocculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.7
11.4 Settling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.9
. 11.4.1 Influence of salinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.9
11.4.2 Influence of concentration . . . . . . . . . . . . . . . . . . . . . . . . 11.9
11.4.3 Influence of water depth and flow velocity . . . . . . . . . . . . . 11.11
11.4.4 Influence of measuring instrument . . . . . . . . . . . . . . . . . . 11.12
11.5 Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11.15

11.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.15
11.5.2 Concentrations larger than 10 kgIm3 . . . . . . . . . . . . . . . . . 11.15
11.5.3 Concentrations from 0.3 to 10 kgIm3 . . . . . . . . . . . . . . . . 11.15
11.5.4 Concentrations smaller than 0.3 kglmf . . . . . . . . . . . . . . . 11.19
11.5.5 Critical bed-shear stress for deposition . . . . . . . . . . . . . . . 11.21
11.5.6 Deposition rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.22
11.6 Consolidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.22
11.7 Erosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.28
11.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.28
11.7.2 Consolidated hard deposits . . . . . . . . . . . . . . . . . . . . . . . 11.29
11.7.3 Consolidated soft deposits . . . . . . . . . . . . . . . . . . . . . . . 11.31
11.7.4 Erosion rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.34
11.7.5 Bed forms and roughness . . . . . . . . . . . . . . . . . . . . . . . . 11.34
11.8 Transport of mud by currents . . . . . . . . . . . . . . . . . . . . . . . . . . 11.35
11.8.1 (Quasi) steady flow . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.35
11.8.2 Non-steady (tidal) flow . . . . . . . . . . . . . . . . . . . . . . . . . 11.36
11.9 Transport of mud by waves . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.42
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11.46
12 MATHEMATICAL MODELS OF SEDIMENT TRANSPORT ......... 12.1
12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.1
12.2 Flow models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12.2
12.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2
12.2.2 Three-dimensional flow models . . . . . . . . . . . . . . . . . . . . . 12.3
12.2.3 Two-dimensional horizontal flow model for estuaries and seas . . 12.8
12.2.4 Two-dimensional vertical flow model . . . . . . . . . . . . . . . . 12.10
12.2.5 One-dimensional flow model for rivers or estuaries . . . . . . . . 12.11
12.3 Wave models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.14
12.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.14
12.3.2 Basic equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.14
12.3.3 Two-dimensional horizontal models for combined refraction.
diffraction. shoaling and dissipation . . . . . . . . . . . . . . . . . 12.16
12.3.4 Two-dimensional models for combined refraction. shoaling
and dissipation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.21
12.4 Sediment transport and morphological models . . . . . . . . . . . . . . . . 12.26
12.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.26
12.4.2 Basic equations of sediment transport . . . . . . . . . . . . . . . . 12.26
12.4.3 Three-dimensional models . . . . . . . . . . . . . . . . . . . . . . . 12.38
12.4.4 Two-dimensional vertical models . . . . . . . . . . . . . . . . . . . 12.42
12.4.5 Two-dimensional horizontal models . . . . . . . . . . . . . . . . . 12.50
12.4.6 One-dimensional models . . . . . . . . . . . . . . . . . . . . . . . . 12.55
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.58
13 MEASURING INSTRUMENTS FOR SEDIMENT TRANSPORT.

SETTLING VELOCITY AND WET BULK DENSITY .............. 13.1
13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13.1
13.2 Measuring facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.1
13.3 Measuring principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13.3
13.3.1 Suspended load transport . . . . . . . . . . . . . . . . . . . . . . . 13.3
13.3.2 Bed-load transport . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4
13.4 Measuring statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.6
13.4.1 General aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.6
13.4.2 Sampling site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13.6
13.4.3 Number of measurements for suspended load transport . . . . . 13.6
13.4.4 Number of measurements for bed-load transport . . . . . . . . 13.15
13.4.5 Sampling frequency . . . . . . . . . . . . . . . . . . . . . . . . . . 13.16
13.5 Computation of sediment transport . . . . . . . . . . . . . . . . . . . . . . 13.17
13.5.1 Suspended load transport per unit width . . . . . . . . . . . . . 13.17
13.5.2 Total load transport per unit width . . . . . . . . . . . . . . . . 13.22
13.5.3 Total load transport in cross-section . . . . . . . . . . . . . . . . 13.22
13.5.4 Tide-integrated total load . . . . . . . . . . . . . . . . . . . . . . 13.24
13.6 Instruments for bed-load transport . . . . . . . . . . . . . . . . . . . . . . . 13.25
13.6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.25
13.6.2 Trap sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.25
13.6.3 Bed-form tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.32
13.7 Instruments for suspended load transport . . . . . . . . . . . . . . . . . . . 13.33
13.7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.33
13.7.2 Bottle and trap samplers . . . . . . . . . . . . . . . . . . . . . . . 13.34
13.7.3 Pump samplers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.48
13.7.4 Optical and acoustical samplers . . . . . . . . . . . . . . . . . . 13.53
13.7.5 Comparison of suspended load samplers . . . . . . . . . . . . . 13.60
13.7.6 Selection of suspended load sampler . . . . . . . . . . . . . . . 13.63
13.8 Instruments for particle size and settling velocity . . . . . . . . . . . . . . 13.65
13.8.1 General aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.65
13.8.2 Sieve instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.66
13.8.3 Sedimentation methods . . . . . . . . . . . . . . . . . . . . . . . . 13.67
13.8.4 Coulter Counter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.72
13.8.5 In-situ Laser diffraction . . . . . . . . . . . . . . . . . . . . . . . 13.72
13.8.6 In-situ video camera . . . . . . . . . . . . . . . . . . . . . . . . . 13.73
13.8.7 Selection of instruments . . . . . . . . . . . . . . . . . . . . . . . 13.73
13.9 Instruments for bed material sampling . . . . . . . . . . . . . . . . . . . . 13.74
13.9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.74
13.9.2 Grab, dredge and scoop samplers . . . . . . . . . . . . . . . . . 13.75
13.9.3 Core samplers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.75
13.10 Instruments for in-situ measurement of wet bulk density . . . . . . . . . 13.76
13.10.1 General aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.76
13.10.2 Mechanical core sampler . . . . . . . . . . . . . . . . . . . . . . 13.76
13.10.3 Acoustic probe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.77
13.10.4 Nuclear radiation probe . . . . . . . . . . . . . . . . . . . . . . . 13.77
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.81
APPENDICES A: TRANSPOR-program; Computation of sediment transport in current

and in wave direction
B: Sand transport in closed conduits
C: Side-wall roughness correction method of Vanoni-Brooks
D: Pollution aspects of sediments
PREFACE
This book reflects the results of basic research and practical experience in sediment transport
and morphology in rivers, estuaries and coastal seas all over the world during a period of
about 15 years.
The purpose of this book is to give a unified view of sediment (sand and mud) transport over
a wide range of conditions; from quasi-steady river flow to the violent wave-breaking
processes in the surf zone of coastal seas. It was not the intention of the author to give a
complete overview of the overwhelming amount of literature available. On the contrary, the
emphasis is laid on the description and the application of those theories and formulae which
have proven to give realistic results based on the authors experience.
The application of refined theories consisting of many complicated equations is often not
justified, given the uncertainties of the input data like current velocity, wave height, bed
material composition, bed forms and roughness. An elegant way to overcome this problem
is to represent the ref& model by a much more simple parameter model or computer data
base model.
Chapter 2 presents an overview of the near-bed fluid velocities and shear stresses, both being
the driving agents of the sedim'entparticle motions. The current boundary layer as well as the
wave boundary layer are discussed. Basic wave properties are presented. Mass transport by
non-breaking and breaking waves is summarized.
Chapter 3 covers the fluid and sediment properties like density, porosity, shape, size and
settling characteristics of the sediment particles.
Chapter 4 presents the processes of initiation of motion and suspension in terms of the critical
velocities and bed-shear stresses. Special attention is given to the design of stable channels
which is an important aspect of irrigation projects.
Chapter 5 covers the characteristics of bed forms generated by currents alone, by waves and
by combined currents and waves. Bed form classification diagrams are given and the shape
and dimensions of the bed forms are discussed.
Chapter 6 presents information of the effective (grain and form) roughness of a sediment bed.
Methods based on bed-form parameters as well as methtds based on integral parameters
(velocity, depth, sediment size) are discussed.
Chapter 7 deals with bed load and suspended load transport in steady flow. Both deterministic
and stochastic approaches are discussed. The classical diffusion theory is used to describe the
distribution of the sediment concentrations over the depth. Special attention is given to
stratification effects of high-concentration suspensions.
The Chapters 8 and 9 cover the field of sediment transport by waves and by combined
currents and waves. The various transport processes related to breaking and non-breaking
waves are identified; high-frequency and low-frequency phenomena are discussed. Emphasis
is put on data analyis of concentrations and transport rates measured in flumes, tunnels and
in nature.
Chapter 10 is related to the transport of sediments in non-steady and non-uniform conditions.

The basic principles of erosion and deposition of sediment particles are presented.
Chapter 11 presents detailed information of the transport of cohesive sediment materials
(mud). Basic phenomena like cohesion, flocculation, settling, deposition, consolidation and
erosion which take place in a continuous cycle, are discussed.
Chapter 12 deals with the mathematical modelling of sediment transport and morphology.
Three-, two- and one-dimensional models of flow, waves, sediment transport and morphology
are presented.
Sediment transport cannot be studied without proper knowledge of measuring instruments.

The accuracy of the data is strongly related to the type of instrument applied. Chapter 13
presents a detailed overview of the available measuring principles, statistics, methods and
instruments. Simple mechanical and sophisticated optical, acoustical and nuclear instruments
are discussed.
The book ends with four appendices. Appendix A presents the TRANSPOR-program (avai-
lable on diskette) which is the sediment transport model of the author. Appendix B deals with
sand transport in closed conduits (pipelines). Appendix C presents a method to eliminate side
wall roughness which is necessary for narrow flumes and channels. Appendix D is related to
pollution aspects of sediments.
Many calculation examples of available methods and formulae are presented throughout the
book, which may help the reader to find a way through the many available equations. A 3.5"-
diskette (TRANSPOR-program) for computing sediment concentrations, transport rates and
bed-form dimensions in a current alone and in combined currents and waves is available to
help the reader to solve practical problems.
The present book has been written with a view to morphology of sediment beds. This latter
field of work will be described in a forthcoming book: "Principles of Morphology in Rivers,
Estuaries and Coastal Seas".
The author hopes that the present book and the TRANSPOR-diskette will serve as a useful
tool for students and graduates in civil engineering, earth sciences, physical geography and
oceanography.
Leo C. van Rijn

Oldemarkt
January 1993
ACKNOWLEDGEMENTS
The author wishes to acknowledge Delft Hydraulics and the Department of Physical
Geography of the University of Utrecht for the typing and the editing of the manuscript.
I am,grateful to the discussions and comments of all my national and international colleaguks,
without whom this book could never have been written.
I am also grateful to the following copyright holders for permission to reproduce figures:
Delft Hydraulics, American Society of Civil Engineers, North-Holland Publishing Company,
U.S. Geological Survey, Nedeco, D and A Instruments USA.
1. INTRODUCTION
Sediment is fragmental material, primarily formed by the physical and chemical desintegration
of rocks from the earth's crust. Such particles range in size from large boulders to colloidal
size fragments and vary in shape from rounded to angular. They also vary in specific gravity
and mineral composition, the predomiuant materiaf bekg quartz.
Once the sediment particles are detached, they may either be transported by gravity, wind
orland water.
When the transporting agent is water, it is called fluvial or marine sediment transport. The
process of moving and removing from their original source or resting place is called erosion.
In a channel the water flow erodes the available material in the banks and/or the stream bed
until the flow is "loaded" with as much sediment particles as the energy of the stream will
allow it to carry.
Usually, three modes of particle motion are distinguished:

rolling and/or sliding particle motion,
saltating or hopping particle motion,
suspended particle motion.
When the value of the bed-shear velocity just exceeds the critical value for initation of
motion, the bed material particles will be rolling and/or sliding in continuous contact with
the bed. For increasing values of the bed-shear velocity the particles will be moving along
the bed by more or less regular jumps, which are called saltations.
When the value of the bed-shear velocity begins to exceed the fall velocity of the particles,
the sediment particles can be lifted to a level at which the upward turbulent forces will be of
comparable or higher order than the submerged weight of the particles and as a result the
particles may go into suspension.
Usually, the transport of particles by rolling, sliding and saltating is called bed-load transport,
while the suspended particles are transport as suspended load transport.
The suspended load may also include the fine silt particles brought into suspension from the
catchment area rather than from the streambed material (bed material load) and is called the
wash load. A grain size of 50 pm is frequently used to make the separation between bed
material load and wash load. Sometimes a value of 63 pm is used (USA). Another method
of discrimination is given by Bagnold (1962), see Chapter 7. Based on energy considerations
it can be shown that all particles with a fall velocity smaller than 1.6 ii I (ii = depth-averaged
velocity, I = water surface gradient) can be transported in unlimited quantities, the latter
being a typical feature of wash load transport.
Bed load and suspended load may occur simultaneously, but the transition zone between both
modes of transport is not well-defined.
Sediment transport by flowing water is strongly linked to surface soil erosion due to rain.
Water seeping into the ground can contribute to landslides (subsurface erosion) which may
become major sources of sediments for rivers.
The whole process can be seen as a continuous cycle of:
soil erosion -r sediment transport -, sedimentation.
Soil erosion and sediment yield (in tonnes per km2 per year) strongly depend on the local
climatic (rain fall), soil, land (surface slope) and vegetation conditions. Values may vary from
50 to 500 tonnes per km2per year. Universal formulae are not available. Based on local data,
regional formulae have been developed.
A detailed discussion of soil loss estimation has been given by Kirkby and Morgan (Wiley,
New York, 1980).
The Yellow river in China is a dramatic example of a river carrying enormous quantities of
sediments, primarily eroded in the Loess valleys of the middle reach, to the sea.
Proper land use and management can substantially reduce the problems related to sediment
transport in rivers and estuaries.
The design of works (terracing, debris dams, slope fixation) can also help to control surface
erosion.
The following classification and definitions in accordance with the ISO-standards (IS0 4363)
are given:
moving as bed load
Total load
(origin)
Bed material
Bed material load
moving as suspended load
wash load moving as suspended load
:
I Total load
(transport)
The material, the particle sizes of which are found in appreciable

quantities in that part of the bed that is affected by transport.
Bed material load : The part of the total sediment transport which consists of the bed
material and which rate of movement is governed by the transport
capacity of the channel.
Suspended load : That part of the total sediment transport which is maintained in
suspension by turbulence in the flowing water for considerable periods
of time without contact with the stream bed. It moves with practically
the same velocity as that of the flowing water.
Bed load : The sediment in almost continuous contact with the bed, carried
forward by rolling, sliding or hopping.
Wash load : That part of the suspended load which is composed of particle sizes
smaller than those found in appreciable quantities in the bed material.
It is in near-permanent suspension and, therefore, is transported
through the stream without deposition. The discharge of the wash load
through a reach depends only on the rate with which these particles
become available in the catchment area and not on the transport
capacity of the flow.
1.2 History
Sediment transport problems related to irrigiation channels were studied and solved by trial
and error in ancient China, Mesopotamia, Egypt and in the Roman Empire.
One of the first researchers who studied theoretically and experimentally the effect of the flow
on a loose sediment bed was DuBuat (1734- 1809) in France. He determined the flow velocity
causing erosion of the bed; various soit materids -(clay to stones) were studied. DuBuat
developed the shear-resistance concept.
Hagen (1797-1884) in Germany and DuPuit (1804-1866) in France gave qualitative
descriptions of the motion of sediment particles along the bed and in suspension.
Brahms (1753) proposed an expression for the critical velocity of a bed consisting of stones.
Tulla (1770-1828) studied fluvial processes in the Rhine river.
The first bed-load type formula based on depth and slope was presented by DuBoys (1847-
1924) in France, who schematized the transport process as the movement of particles in a
series of layers.
Around 1900 the first movable bed models were built by Fargue (1827-1910) in France and
by Reynolds (1892-1912) in England. Basic reserach on sediment transport in laboratory
flumes was started seriously by Engels (1854- 1945) in Germany and by Gilbert (1843-1918)
in the USA. The data of Gilbert presented in the famous publication: "Transportation of
Debris by Running Waterw(1914) are still used by many researchers for calibration of bed-
load transport formulae.
Theoretical work was done by Forchheimer (1852-1933) and by Schoklitsch (1888-1969) in
Germany. The latter (1914) developed an equation for the critical bed-shear stress (initiation
of motion of a particle) resting on a longitudinal sloping bed. Similar work for a particle
resting on a transverse sloping bed (river bank) was done by Leiner (1912).
Empirical field research related to the design of irrigation channels (regime formulae) was
done by Kennedy in India (1895).
Shields (1936) made a significant contribution related to h e (critical) bed-shear stress for
initiation of motion of sediment particles. The curve proposed is well known as the "Shields"
curve.
Recent work to be mentioned is that of Kalinske (1946), Vanoni (1946), Einstein (1950) and
Bagnold (1966).
1.3 Symbols and units
The most important symbols used in this book are:
= reference level (m)

= Chkzy-coefficient (m0.5s-1)
= concentration (-)
= dimensionless particle parameter (-)
= particle diameter (m)
= force (N>
= friction factor (-)
= acceleration of gravity (m s2)
= wave height (m)
= flow depth (m)
= energy gradient (-)
= equivalent or effective bed roughness (m)
= volumetric flow discharge (m3s")
= volumetric total sediment discharge or transport rate (m3s-l)
= flow discharge per unit width (m2s")
= sediment discharge or transport rate per unit width (m2s4)
= relative density (pJp) (-)
= dimensionless bed-shear stress parameter (-)
= time (s)
= instantaneous velocities in x, y ,z direction (m S-')
= time-averaged velocities in x, y,z direction (m S-')
= peak orbital velocity near bed (m S-I)
= depth-averaged velocity (m S-')
= cross-section averaged velocitj; (m S-')
= bed-shear velocity (m S-l)
= particle fall velocity (m S-l)
= dimensionless suspension number (-)
= vertical coordinate (m)
= zero-velocity level (m)
= bed form length (m)
= bed form height (m)
= thickness ( 4
= mixing coefficient (m2s-')
= bed-shear stress (N m-2)
= fluid density 0% m-3)
= sediment density (kg m-3)
= dimensionless particle mobility parameter (-1
= dimensionless particle mobility parameter (-1
= dimensionless transport rate (->
= kinematic viscosity coefficient (m2s-l)
= dynamic viscosity coefficient (kg m-Is-')
= Von Karman constant (-1
Instantaneous variables (velocity) are represented by Capital letters (U). Time-averaged

variables are represented by normal letters (u).
The following subscripts are frequently used:
b = bed, bed-load, bottom
c = current, current-related
cr = critical, initiation of motion
cw = current and waves
f = fluid
m = mean, median, mixture
s = sediment, suspended load
t = total, total load
w = waves, wave-related
Examples are:
= current-related bed-shear stress

= bed-shear stress for current and waves-
= wave-related bed-shear stress
= critical bed-shear stress
= volumetric bed-load transport rate per unit width
= volumetric suspended-load transport rate per unit width
= volumetric total load transport per unit width
= current-related (volumetric) bed load transport rate per unit width
Units are according to the "Systeme International d'UnitCsN (SI-units), which has been
adopted by the International Organization for Standardization (10s).
Force is expressed in Newtons (N)
Mass is expressed in kilograms (kg)
Length is expressed in meters (m)
Time is expressed in seconds 6)
1.4 Characteristic parameters
The following dimensionless characterisic parameters are herein used:

particle diameter, P,
particle mobility parameter, 8
excess bed-shear stress parameter, T
suspension parameter, Z
transport rate, @
1. Particle parameter, D,
The particle diameter D, reflects the influence of gravity, density and viscosity and reads as:
in which:
d,, = median particle diameter of bed material
s = specific gravity (= p,/p)
v = kinematic viscosity coefficient
g = acceleration of gravity
2. Particle mobility parameter 8
Plane bed
The particle mobility parameter 8 is the ratio of the hydrodynamic fluid (drag and lift) force
and the submerged particle weight. The fluid force is proportional to p dZ(uJ2and the
submerged particle weight is proportional to ( p s - p ) g d 3 yielding
, a ratio of:
in which:
rb = overall time-averaged bed-shear stress due to currents and/or waves
u, = overall bed-shear velocity (rb = p ul)
s = specific density ( p , / p )
h = flow depth (m)
I = energy gradient (-)
Bed foms
When bed forms are present, the grain-related or effective bed-shear stress (sb in stead of
the overall bed-shear stress (r,,) should be used to calculate the particle mobility parameter.
The effective bed-shear s p a s represents that part of the overall bed-shear stress acting on the
bed material particles (rb = p t b with p = efficiency factor). Basically, it is assumed that
the form drag dissipated in the eddies generated by the bed forms is not fully effective in the
entrainment of sediment particles from the bed into the fluid.
3. Ezcess bed-shear stress parameter T
The excess bed-shear stress parameter T, is defined as:
in which:
t b , = critical time-averaged bed-shear stress according to Shields.
4. Suspension parameter, Z
The suspension parameter Z reflects the ratio of the downward gravity forces and the upward
fluid forces acting on a suspended sediment particle in a current and reads as:
in which:
w, = particle fall velocity in clear fluid
u, = overall bed-shear velocity
K = Von Karman constant
p = ratio of sediment and fluid mixing coefficient
5. Transport rate, @
The dimensionless transport @ usually is represented as:
Another dimensionless expression is:
in which:
qt = volumetric total sediment transport rate (m2/s)
g = acceleration of gravity (m2/s)
dm = median particle size of bed material (m)
w, = particle fall velocity of bed material (m/s)
s = specific density ( p s / p )
The volumetric sediment transport rate (qJ can also be made dimensionless with the specific
flow discharge (q), yielding the discharge-weighted concentration.
in which:
q, = volumetric total sediment transport rate (m2/s)
q = specific flow discharge (m2/s)
Sometimes the transport rate is given as a submerged mass per unit time and width. The mass
(above water) is equal to p,/(p, - p ) times the submerged mass.
2. FLUID VELOCITIES AND BED SHEAR STRESSES
2.1 Introduction
Sediment transport processes in currents and/or waves are dominant in the near-bed region.
Therefore, it is of essential importance to know the magnitude and direction of the near-bed
velocities and shear-stresses in the presence of bed forms (ripples, dunes) or in case of a flat
bed.
Detailed information of fluid velocities and bed-shear stresses is presented by Van Rijn, 1990
("Principles of Fluid Flow and Surface Waves in Rivers, Estuaries, Seas and Oceans").
In this Chapter a summary is given of experimental results and theoretical expressions. The
following subjects are covered:
currents,
waves,
currents and waves.
2.2 Currents
2.2.1 Current boundary layer
1. Equation of motion
The equation of motion for steady uniform (turbulent) flow is given by (see Fig. 2.2.1):
or using I = sir$
For z = 0 this yields the bed-shear stress:
By defdtion r, = p(u,,J2, yielding
2. Turbulent jlo w
Turbulence is a random fluctuating velocity field which interacts with and derives its energy
from the mean flow field. A turbulent velocity field can only be described by statistical
quantities such as root-mean-square values, amplitude distribution, correlations and spectra.
The amplitudes are generally normally distributed so that the root-mean-square deviation gives
a good estimate of the fluctuations.
According to the Reynolds' procedure, the shear stress in a turbulent flow at height z in a
steady uniform flow can be described as (Fig. 2.2.1) :
du -
t,= pv - - p u 1w I
dz
in which:
u = time-averaged fluid velocity at height z
u' = turbulent fluid velocity fluctation in horizontal direction
'
w = turbulent fluid velocity fluctuation in vertical direction
p = fluid density
Although the time-averaged vertical velocity w is equal to zero (w = O), the vertical turbulent
fluctuatio- not equal to zero (w' * 0 ) . Consequently, the turbulence shear stress
r, = - p u'w' c 0.
The turbulent shear stress t, is dominant in the major part of the flow depth. In case of a
smooth bottom the viscous shear stress t, becomes dominant close to the bottom because the
turbulent fluctuations u' and w' die out near the bottom and are equal to zero at the bottom
(u' = w' at z = 0). The layer where the viscous shear stress is dominant is called the
viscous sublayer (a,,).
Above the viscous sublayer the flow is turbulent. The most important turbulent sublayer is
the logarithmic sublayer. Between the viscous sublayer and the logarithmic sublayer there is
a transition sublayer, sometimes called the b u e r sublayer. Above the logarithmic sublayer
there is an outer sublayer (see also Figure 2.2.2).
2.2.2 Hydraulic regimes
The effect of the bottom (or wall) roughness on the velocity distribution in a turbulent flow
was first investigated for pipe flow by Nikuradse (1933). He used pipes covered with uniform
sand grains at the inside and he measured velocity distributions at different Reynolds'
numbers (Re), pipe diameters (D) and grain sizes (&). Based on these experiments,
Nikuradse introduced the concept of the equivalent sand grain roughness or Nikuradse
roughness (kJ as a standard for all other types of roughness elements (k).
The roughness elements mainly influence the velocity distribution close to the bottom,
because the roughness elements generate eddies (with a size of the order of the roughness
elements) which affect the turbulence structure and hence the velocities close to the bottom.
Further away, the eddies will rapidly be absorbed in the general existing turbulence pattern.
The type of flow regime can be related to the ratio of the Nikuradse roughness (kJ and a
length scale of the viscous sublayer (V/U,,~) in which v = kinematic viscosity coefficient
(m2/s) and u,, = current-related bed-shear stress (mls).
Based on experimental results, it was found:
Hydraulically smooth flow, for 4 -- -

- u*,c 4 < 5
(2.2.5)
vIu*,c v
The roughness elements are much smaller than the thickness of the viscous sublayer and
do not affect the velocity distribution (Fig. 2.2.3).
Hydraulically rough flow, for 4 -- -
- u*sc 4 > 70
vlu*,c v
A viscous sublayer does not exist and the velocity distribution is not dependent on the
viscosity (v ) of the fluid (Fig. 2.2.3).
Hydraulically transitional flow, for 5 < '=*,c 4 < 70
-
v
The velocity distribution is affected by viscosity as well as by the bottom roughness.
--
z
Lx
Fl
/,///,//////////////
C
:
:
,/
,/
/
/7
TZ
P
Fo
Figure 2.2.1 Shear stress

water surface
F2
2
'EL H Tb +I
turbulent logarithmic layer

------------------------------
) I ------- -------
............................
transition layer
viscous sub. layer
\
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
bottom
\\\\
\
inner layer or
bottom layer
Figure 2.2.2 Current boundary layer

tf - ----- ---------
Figure 2.2.3 Velocity distribution in smooth and rough flow
Kh
-6
K u,h
Figure 2.2.4 Miring length and eddy viscosity distribution

2.2.3 Velocity distribution over the depth
A general expression for the velocity distribution over the depth for smooth and rough flows
is:
in which:
u,, = current-related bed-shear velocity ( d s )
K = constant of Von Karman (0.4)
z, = zero-velocity level (u = 0 at z = zJ, (m)
z = vertical coordinate (m)
v
Smooth flow regime : z, = 0.11 - for u * s c k , ~ 5 (2.2.9)
u*,c v
Rough flow regime : z, = 0.033 k, for u*sc4

v
2 70 (2.2.10)
Transition regime : z, = 0.11

v
- + 0.033 k, for 5 < -u-* c k, < 70 (2.2.1 1)
u*p v
The zero velocity level should be interpreted as a computation parameter without physical
meaning, because Eq. (2.2.8) is not valid close to the bottom. In case of a rough boundary
the position of the zero-velocity level is unknown. Usually, the bottom plane (z = 0) is
defined as the plane that is formed after smoothing of the roughness elements. For spherical
roughness elements this yields a value of about 0.75k above the underside of the spheres (k
= sphere diameter). A realistic valuelo$ the zero-velocity level (z = z), can be found by
plotting the values of u/u* against z-zdz, on sen)i-logarithmic scale and varying z, until the
best fit is obtained. This method yields values z, = 0.25k for sand and gravel particles.
Averaging of Eq. (2.2.8) over the depth, yields:
Neglecting the zJh-parameter in Eq. (2.2.12), the depth-averaged flow velocity does occur
at z = We = 0.37 h, in which e is the base of the natural logarithm (e = 2.72).
Applying Eq. (2.2.12) in Eq. (2.2.8), the velocity distribution can also be expressed as:
2.2.4 Fluid mixing coefficient
A turbulent flow field has a diffusive character. Gradients of momentum and scalar quantities
are rapidly diminished by this diffusive action.
The analogy of turbulent motion with the movements of molecules leads to the analogy given
by Boussinesq a n a e introduction of an eddy-viscosity concept for the apparent turbulent
' '
shear stress - p u'w = p sf hJaz (u'w are velocity fluctuations in horizontal and vertical
direction).
Applying a logarithmic velocity profile and assuming a linear shear stress distribution over
the depth, a parabolic fluid mixing coefficient is obtained (Fig. 2.2.4):
in which:
, ~fluid mixing coefficient related to current (m2/s)
E ~ =
Based on the experimental results of Coleman (1970), Van Rijn (1984) introduced a
parabolic-constant mixing coefficient distribution; parabolic in the lower half of the depth and
constant in the upper half of the depth:
sf,C= K ~ (- 1
z/h)~,,~ for z/h < 0.5 (2.2.15)
sfSc= 0.25 K u , , ~h for z/h 2 0.5 (2.2.16)
The main reason for applying Eq. (2.2.16) is that it produces a more realistic concentration
profile with finite concentrations at the water surface (Van Rijn, 1989).
< 2.2.5 Bed shear stress and bed friction
The overall time-averaged bed-shear stress is defined as:
in which:
h = water depth (m)
I- = energy line gradient (-)
u = depth-averaged velocity (m/s)
C = ChCzy-coefficient (CZ = 8g/fJ, (m0.5/s)
f, = friction factor of Darcy-Weisbach (-)
k, = effective bed roughness height (m)
p = fluid density (kg/m3)
g = acceleration of gravity (m/s2)
The friction factor and ChCzy-coefficient for laminar, hydraulic smooth, rough and
transitional flow are presented below:
I . Laminar flow
The friction factor for laminar flow in a wide open channel is:
in which:
Re = a v = Reynolds number (-)
2. Turbulent j7ow
The Chkzycoefficient (C) and the friction factor (fJ can be derived from Eq. (2.2.12).
ChCzy found empirically:
- which:
in
u = depth-averaged velocity (mls)
C = Chkzy roughness coefficient (mO.'/s)
h = water depth (m)
I = water surface slope or energy gradient (-)
Using u,, = (ghI)0.5in Eq. (2.2.12), yields
or using K = 0.4 and g = 9.81 m/s2:
Using Eqs. (2.2.9), (2.2.10) and (2.2.11), yields:
12 h
Hydraulic smooth flow : C = 18 log[
3.3 V/U,,~]
C =
)18: log; : (-
12 h
Hydraulic rough flow : C = 18 log[T)
Transitional flow
: C = l8l0g (I$+ %:Iu*,&)
c= 18 log jr + :ZVc,ii)
The friction factor follows from: f, -- -

8g
c2
yielding for hydraulic rough flow:
Equation (2.2.23) can be approximated by the Strickler formula (in the range of C = 40 to
70 m0.5/s):
Another widely used resistance equation is the Manning equation, which reads as:
in which:
n = 0.045 = Manning coefficient
h = waterdepth (m)
k = effective bed roughness (m)
It is noted that the waterdepth (h) should be replaced by the hydraulic radius R for narrow
channels.
2.3 Waves
3.3.1 Near-bed orbital velocities
Applying linear wave theory, the peak value of the orbital excursion (A,) and velocity@,)
at the edge of the wave boundary layer can be expressed as:
in which:
0 = 2x/T = angular velocity (s-I)
k = 2x/L = wave number (m-')
H = wave height (m)
L = (g'T2/2x) t.anh&h) = wave length (m)
T = wave period (s)
h = wave depth (m)
Based on field measurements in the near-bed region inside the surf zone (with a relative wave
height of H,/h in the range of 0.2 to 0.45) and outside the surf zone (H,/h = 0.14 to 0.27),
Van Heteren and Stive (1985) found that Eq. (2.3.2) applied to instantaneous water surface
elevations produced a maximum over-prediction of 25 % of measured horizontal rms-velocities
outside the surf zone and an underprediction of 5 % inside the surf zone.
Dean (1986) reports that linear wave theory provides a good prediction of near-bottom
kinematics for a wide range of relative wave heights and wave steepnesses.
An estimate of the near-bed peak velocities under the crest and trough of asymmetrical waves
in shallow water can be obtained by applying higher order wave theories:
According to second order wave theory of Stokes (which is valid for h r 0.01 g ~ ' ) , the
peak velocities under the wave crest and trough-are:-
This yields: 6 3 - = 1.1 for high wind waves (H/h = 0.2, h = 10 m, T = 7 s) and
1.7 for high swell waves (H/h = 0.2, h = 10 m, T = 14 s). This latter value may not be
realistic because it is outside the validity range.
Dean (1986), Koyama and Iwata (1986) and Rienecker and Fenton (1981) used a stream
function method to represent the higher order terms of the unsteady Bernoulli equation. Using
a suitable stream function, the non-linear free surface boundary condition can be applied in
a number of free surface points along the wave profrle in an iterative way. Klopman (1989)
made a numerical program based on the Fourier approximation method of Rienecker and
Fenton (1981).
Comparison of measured and computed peak velocities shows good agreement for the crest
peak velocities. The computed trough peak velocities are however much too small in shallow
water (h < 3 m), 'see Kroon and Van Rijn (1993).
Based on analysis of measurements in shallow water, Van Rijn proposes for shallow water
(h s 0.01 g ~ ' ) :
with a = 1 + 0. 3(H8/h)and 0, according to linear wave theory.

Another problem is the representation of the wave spectrum in a real situation. Assuming that
the higher waves contribute most to the sediment transport process, the significant wave
height (H, or H,,,) in combination with the peak period (T,)are herein assumed to be the
characteristic wave parameters for the sediment transport process.
2.3.2 Wave boundary layer
The wave boundary layer is a thin layer forming the transition layer between the bed and the
upper layer of irrotational oscillatory flow (Fig. 2.3.1). The thickness of this layer remains
thin (0.01 to 0.1 m) in short period waves (T s 12 s) because the flow reverses before the
layer can grow in vertical direction. The boundary layer thickness 6, can be defined as the
minimum distance between the wall and a level where the velocity equals the peak value of
the free stream velocity , see Fig. 2.3.1.
Figure 2.3.1 Wave boundary layer
Reynolds' number, 06As IV
Figure 2.3.2 Hydraulic regimes in oscillatory flow

In case of laminar flow the following values can be given:
Jonsson 1980 : 6, -- -
2Ir
P
4.6
Manohar 1955 : 6, = -
P
in which:
P = (X/VT)O.~ = length parameter of Stokes (m)
v = kinematic viscosity coefficient (m2/s)
T = oscillation period (s)
In case of turbulent flow Jonsson and Carlsen (1976) proposed):
30 6,
for 10 <
A*
- < 500
ks
Equation (2.3.7) can also be represented by:
Equation (2.3.7) is based on theoretical and experimental research. Artificial triangular

roughness elements have been used in wave tunnel experiments. The k-value of these
roughness elements was determined from water surface slope measurements in a steady
uniform flow yielding k,-values in the range of 2.5 to 4 times the maximum height of the
elements.
The theoretical results of Fredsere (1984) can be roughly (error f 20%) approximated by:
Although the wave boundary layer thickness is rather small, the generated shear stresses and
turbulence intensities are rather large and are important for the sediment transport processes.
2.3.3 Hydraulic regime in waves

Jonsson (1966) first presented a plot of the hydraulic regime in oscillatory flow in terms of
the Reynolds number and relative roughness, as follows:
Hydraulic regime = F (0;"

- ,+I
Figure 2.3.2 shows the graph of Jonsson (1966, 1980). The results of Kamphuis (1975) are
also shown.
The onset of turbulence in the wave boundary layer has been studied by many researchers
(Sleath (1974, 1988); Davies (1985)). The results of Sleath (1974, 1988) are probably the
most accurate results. Sleath investigated the onset of turbulence in oscillatory flow over flat
beds of sand, gravel and pebbles. Both horizontal and vertical velocity fluctuations were
measured using a laser Doppler velocity meter.
For practical purposes the following expression can be applied to determine the transition to
fully developed turbulent flow in case of a flat bed:
in which: d, = median particle size of bed material.

2.3.4 Velocity distribution in wave boundary layer
1. Eiperimental results
Detailed knowledge of the flow velocities in the boundary layer above rippled and flat beds
is essential to understand the sediment entrainment processes. Examples of velocity
distributions at maximum flow for a laminar and a turbulent case are shown in Fig. 2.3.1.
The fundamental difference is the vertical mixing effect in turbulent flow giving a more
uniform profile. Turbulent flow is the most interesting case for sediment transport processes
because the flow will be turbulent in case of a mobile rippled bed regime and a flat bed (sheet
flow) regime.
Several experimental studies have been performed in wave tunnels with rippled beds. The
most interesting data are presented by DuToit (1982). He used a laser-Doppler velocity meter
to measure the instantaneous horizontal velocities above the crest and trough of a sand ripple
(turbulent flow).
Figure 2.3.3 shows the horizontal velocity component above a sand ripple during half a cycle.
The essential features are, as follows:
the free stream is moving with maximum velocity from right to left and a well defined
vortex fills almost half the trough on the downstream side of the ripple,
the vortex increases in size as the free stream decelerates,
the vortex fills almost the whole trough,
the free stream reverses and the vortex is being ejected over the crest and a strong surge
of fluid (sweeping through the trough and spilling over the crest) is building up,
the flow has already started separating at the crest and a new vortex starts to form,
a jet (associated with the separation) shoots out over the trough and has almost reached the
trough.
Figure 2.3.4 shows the variation range of the peak velocity as a function of the height above
the crest for various test conditions and ripple dimensions. Both measurements in oscillating
u-tubes and oscillating tray rigs were used. The ripples were self-formed natural sand ripples
with &/A = 0.7 and AlA = 0.17. As can be observed, the peak velocity direct above the
crest can be about 1.5 to 1.6 times as large as the peak velocity outside the boundary layer.
Another informative experimental study related to oscillatory flow over a rippled bed has
been performed by Sato et al. (1984). The experiments were performed in an oscillatory
water tunnel. Velocities above symmetrical and asymmetrical ripples were measured with split
hot-film sensors under conditions of both sinusoidal and asymmetrical oscillating flow. The
artificial ripples, which consisted of cement mortar, were copied from self-generated sand
ripples.
x/A
(a) w t = 0''
+18
- - - - -
-
I I , I I 1 - I I I
t 4
0.0 0.5 1.O 0.0 0.5 1.0

( c ) wt = 60' x /A ( f ) wt = 150° x /A
Figure 2.3.3 Horizontal velocity field above rippled bed (DuToit, 1982)
1.8
06 = peak velocity outside
boundary layer
1.6
\\.-.-. i) = peak velocity at height z, in
7i
%
: ,-
--
I
0
I boundary layer
<?
1.4
J I--.\
-.---.- ------ ,- z, = height above crest
A = ripple length, P = ( w / ~ v ) O . ~
1.2
II
i;
-b--
----
II
-
11 crest
1.0
0 5 10 15
Pz1
Figure 2.3.4 Vertical distribution of peak velocity above ripple crest (DuToit, 1982)
Figure 2.3.5 shows the spatial distribution of the equiphase mean velocity, Reynolds stresses
and mean pressures in half a period. The flow is seen to be locally accelerated above a ripple
crest and decelerated above a trough. The Reynolds stress is large in the region of the lee
vortex. As the vortex is ejected upwards, the region of high turbulence moves with it and
then diffuses.
Figure 2.3.6 shows the time history of* m e a n x h i t y horizontal (u) and vertical (w)
- components and the turbulent quantities (u '3 and- (w and u 'wJin points at 0.03 m above
the ripple crest and the ripple trough (period = 4 s). The plotted values represent results
averaged over 30 periods. Coherent vortices were formed in the lee side of the ripple when
the velocity of the main flow reached its maximum. These vortices continued to develop until
they were ejected after flow reversal. The e j M v o r t i w e r e transported over the ripple
during the next half period. The values of (u')' and (wQ2show two peaks in each half
period, which correspond to the passageofo vortices created in the lee side of the nearest
and the neighbouring ripples. Values of u'w' above a crest show a strong peak according to
the passage of the first vortex but they donot show-a peak with the passage of the second
vortex. It appears that turbulence maintains a coherent structure just after the vortex ejection
and that, as the vortex moves upwards, turbulence decays and diffuses away.
Finally, the flow visualization experiments of Kaneko and Honji (1979) and Honji et al(1980)
are reported. Steady laminar vortices were generated above ripples in case&/h < 1
(standing vortices). Large separation vortices were found for & / A = 1. Similar results were
observed by Soulsby et al (1983) in field conditions using an underwater camera.
2. Theoretical models
Neglecting convective accelerations and vertical fluid velocities, the basic equation of motion
for a flat bed, reads as:
dt p a x p a z
in which:
U = instantaneous fluid velocity at height z
P = instantaneous fluid pressure at height z
s = instantaneous fluid shear stress at height z
Various researchers have applied Eq. (2.3.12) to determine the U-velocity. The basic problem
is to relate the shear stress to the gradient of the velocity (s = dU/&) using a turbulence
closure model. The boundary condition is U = U, at z = 6,. The pressure inside the
boundary layer is assumed to be equal to that just outside the boundary layer where the flow
is irrotational, giving: aP/& = p aU,/at
One of the first attempts to describe the turbulent boundary layer by a mathematical model
was done by Kajiura (1968). He adopted the eddy-viscosity concept in a three-layer model.
In the small inner layer the eddy viscosity E was taken as a constant. In the overlap layere
was assumed to vary linearly with the distance from the bed and in the outer layer e was kept
constant at a certain distance from the bed. By use of the equation of motion, Kajiura
obtained a rather laborious analytical-numerical solution. Brevik (1981) has simplified the
analysis by Kajiura by avoiding the inner layer, where E was kept constant.
Figure 2.3.5 Spatial distribution of velocity, Reynolds shear stress and pressure above a
rippled bed (Sato et al., 1984)
I trough I
Figure 2.3.6 Velocity and turbulence characteristics above ripple crest and trough
(Sato et al., 1984)
Kajiura and Brevik tested their analytical results against the experimental work done by
Jonsson (1966), see also Jonsson and Carlsen (1976), and obtained reasonable agreement.
However, three shortcomings in Kajiura's and Brevik's theory must be mentioned: (i) they
did not take into account that the eddy viscosity is a function of time as well as of the
distance from the wall, (ii) the thickness of the wave boundary layer was also taken as a time-
independent quantity, and finally (iii) the variation of the bed-shear stress was assumed to be
sinusoidal.
Jonsson and Carlsen (1976) used the momentum equation integrated over the boundary layer
thickness and the assumption of a logarithmic velocity profrle to determine the wave friction
factor for a rough wall. Their model does, however, not describe the time variation of the
shear stress and the phase lag of the maximum shear stress relative to the maximum velocity
in the outer layer.
Fredsse (1984) made further developments along the line of Jonsson and Carlsen, still using
the depth-integrated momentum equation and a logarithmic velocity-profile, assuming that the
flow field in each half period starts from rest.
Fredsse (1984) used the momentum equation to calculate the detailed variation of the
boundary layer thickness and the shear stress during the wave period.
The most recent developments of mathematial models for the wave boundary layer have
mainly been concentrated on the application of higher order turbulence models to describe the -
details of the flow. Bakker (1974) used a mixing length model to describe the wave boundary
layer and the suspended sediment concentrations under waves. A two-equation model was
used by Hagetum and Eidsvik (1986).
2.3.5 Bed shear stress and bed friction
Waves exert friction forces at the bed during propagation. The bed shear stress, which is
important for wave damping and sediment entrainment, is related to the friction coefficient
by:
in which:
Tb,w,t = instantaneous wave-related bed-shear stress (N/m2)
4" = friction coefficient (-)
U, = instantaneous fluid velocity just outside boundary layer (m/s)
The friction factor f, is assumed to be constant over the wave cycle and is determined from
the peak values as: fw = 2tbBW/(p 03.
The time-averaged (over half a wave cycle) bed shear stress is:
2. Laminar flow
In the laminar range @, Adv < 10', see Fig. 2.3.2) the friction coefficient can be
determined analytically f r ~ mEq. (2.3.12) applying t = pv dU/dz. This yields (Jonsson,
1966):
3. Smooth turbulent flow
In the hydraulic smooth regime (10' < 0, Adv < lo6 and A& > l@, zie Fig. 2.3.2)
Jonsson (1966, 1980) obtained from his model:
Equation (2.3.16) can be approximated by
Kajiura (1968) obtained from his model:
4. Rough turbulent flow
In the rough turbulent regime (0, &/v > 101 and A& < 100, see Fig. 2.3.2) Jonsson
(1966) proposed:
with ka = 0.3 for A& s 1.57.
Equation (2.3.19) can also be expressed as (Swart, 1976):
with fwa = 0.3 for A& r 1.57.

Equations (2.3.15), (2.3.16), (2.3.19) are shown in Fig. 2.3.7. Equation (2.3.20) is shown
in Fig. 2.3.8.
10' 2 5 01 2 5 *)L 2 5 ,I 2 5 X)' 2 5 0'
---+6 6 A 6 / V
Figure 2.3.7 Wavefnction co@cient for hydraulic smooth and rough conditions
(Jonsson, 1966)
Figure 2.3.8 Wave friction coejJicient for hydraulic rough conditions (Jonsson, 1966)
Little is known about the friction coefficient in the transition regime.
+
It is proposed to use Eq. (2.3.20) replacing k. by k. 3.3 v/u,, with u , , ~= (rdp)O5,
similar to Eq. (2.2.24).
2.3.6 Breaking waves
I . Limiting steepness
Waves break when the wave steepness (HIL) exceeds a critical value. A criterion for a
horizontal bottom has been given by Miche (1944):
yielding a maximum value of 0.14 for deep water.
Miche (195 1) also derived an expression for the limiting (deep water) wave steepness on a
sloping bottom assuming full reflection ( P = bottom slope angle):
2. Limiting wave height on a horizontal bottom
The limiting wave height corresponds to the maximum wave height at the crest just before
breaking. Miche (1944) computed the limiting wave height on a horizontal bed from a Stokes
wave solution by assuming that breaking occurs when the fluid velocity at the crest equals the
wave propagation velocity, yielding for shallow water (kh < < 1):
Since Stokes waves are symmetrical, Equation (2.3.23) represents a spilling breaking wave.
For a solitary shallow-water wave the theoretical limiting value is:
Based on the analysis of flume tests with regular and irregular waves over a horizontal
bottom, it was found that (Nelson, 1983):
3. Limiting wave height on a sloping bed

'
Nelson (1983) also found that bottom slopes as small as 0.01 (1:lOO) affect the wave
mechanics sufficiently to increase the H,n-parameter for individual waves.
The wave motion on a plane sloping bed is mainly determined by the bottom slope (tan P ),
the incident wave height (H) just before breaking, the wave period O and the gravity
acceleration (g) giving two dimensionless parameters: tan P and H(gTZ). The latter parameter
usually is expressed as HIL, with Lo = gTZ/2?r.
Waves reflect rather than break when fhe bed slope is rather large (full reflection against a
vertical wall). When the bottom slope P decreases beneath a certain critical value, the waves
become unstable and break.
A common breaking criterion is:
Generally, three types of breaking waves are distinguished:

surging breaker in case of flat incident waves and a large bed slope; the amount of
breaking is small and occurs close to the shoreline, reflection is large (2 < 5 < 2.3),
plunging breaker when the wave steepness is larger or the bed steepness is smaller; a
curling jet is generated plunging in the wave trough ahead accompanied by large turbulence
production and a forward marching bore after breaking (0.4 < 6 < 2),
spilling breaking in case of relatively steep waves and/or a relatively flat bed; the water
surface near the crest spills as the wave crest sharpens with the decrease in water depths;
breaking occurs at a greater distance offshore where the wave crest is still symmetric and
is accompanied by the generation of a foamy roller at the wave front while the wave trough
ahead is not visibly disturbed (5 < 0.4).
Plunging breakers show strong mixing properties. During breaking a jet is gene:ated which
strikes upon the forward slope of the wave generating a series of eddies that contribute to the
rapid decay of the wave. Each eddy moves forward and downward upto the bottom yielding
a strong mixing process. The forward velocity of the eddies is smaller than the wave (bore)
velocity and therefore the eddies move toward the back of the wave where they expand under
escape of the entrained air bubbles which rise to the surface (Miller, 1976).
This process eventually transforms into a bore that moves upto the shore. New waves may
be generated and go through the breaking process again, but closer to the shore.
Spilling breakers have less effective mixing properties because of the relatively small scale
eddies, which are confi~led to the near-water surface region and donot extend much below the
trough level.
Wave heights of individual breaking waves in field conditions (y = H,Jh) are given by Kana
(1979):
Spilling breakers : y = 0.55 - 0.65

Transitional breakers : y = 0.65 - 0.75
Plunging breakers : y = 0.75 - 0.90
Within each range the y-values were found to depend on the local bottom slope. A larger
bottom slope yields a larger breaker height.
In flumes the following values have been observed:
Spilling breakers : y = 0.65 for slope 1:20
y = 0.60 for slope 1:40
y = 0.55 for slope 1:60
Plunging breakers : y = 0.90 for slope 1:20
y = 0.80 for slope 1:40
y = 0.70 for slope 1:60
Battjes (1974) related the y-ratio to the f -parameter. Based on the analysis of field and flume
data, the following values are given:
y = 0.8 for 5 5 0.2

y =0.9 for 5 = 0.4
y = 1.0 for 5 = 0.6
y = 1.1 for 1 5 5 I2.
Based on the analysis of laboratory experiments with regular waves and plane sloping
bottoms, Weggel (1972) found:
1.56 - -43.8 (1 - LP) %

-
Y =
(1 + ,-19.5 tang
1 2x
-19
Lo
with L, = g 'IYLI2r = wave length at deep water.
Equation (2.3.27) yields y = 0.78 for tanp = 0.
Thornton - Guza (1982) and Sallenger - Howd (1989) state that in the surf zone y, (or y),
is approximately constant and independent of the offshore wave conditions because nearly all
waves are breaking in the surf zone (wave energy saturation).
Thornton and Guza (1982) report field measurements on a sloping bottom 150. They
observed a limiting breaker height (in terms of the rms wave height) of y, = 0.4 (y, =
0.55) for spilling breakers. Sallenger and Howd (1989) report y, = 0.3 ( y, = 0.45) for
a local bottom slope of tanp = 0.03 and y, = 0.4 (y, = 0.55) for tanp = 0.06.
Hotta and Mizuguchi (1980) did detailed wave height measurements in the surf zone with a
sloping bottom (varying from 1:100 to 150). Wave heights were measured simultaneously
over a distance of about 100 m with space intervals of 2 m. The breakerline (maximum wave
height) was about 100 m offshore. They found an almost constant ratio y, = H,,,Jh = 0.7
in the region where the water depth was continuously decreasing, indicating fully energy-
satured waves. The significant wave period T,was also constant in this zone.
Based on measurements of Kroon and Van Rijn (1992) performed in the inner surfmne (near
Egmond, The Netherlands) with bed slopes in the range of 0.03 to 0.1 and water depths in
the range of 0.4 to 1.4 m, the following results were found:
no breakers : y, 5 0.4
spilling breakers : ys = 0.4 - 0.6

plunging breakers : y, = 0.6 - 0.9
surging breakers (swash zone) : y, = 0.9 - 1.2
2.3.7 Mass transport in non-breaking waves
1. Non-viscous oscillatory flow
Stokes (1847) first pointed out that the fluid particles do not describe exactly closed orbital
trajectories in case of small-amplitude surface waves propagating in a perfect non-viscous
(irrotational) oscillatory flow. The particles have a second-order mean Lagrangian velocity
(called "Stokes" drift) in the direction of wave propagation. It results from the fact that the
horizontal orbital velocity increases with height (z) above the bed. Consequently, a particle
at the top of an orbit beneath a wave crest moves faster in the forward direction than it does
in the backward direction at the bottom of the orbit beneath a wave trough. By definition the
Lagrangian Stokes drift cannot be detected by taking measurements at a fixed point.
The instantaneous horizontal Stokes drift (U,) of a water particle with a mean position of x,
and z, is U,(x,+ +
a , z1 P ) where a, and fl are the coordinates of the particle position on
the trajectory. An approximation of Us is:
U,(xl + or, z, + p) = U(xl, z,) + (r + p-

au
az
Applying linear wave theory, and averaging over the wave period, yields the time-averaged
velocity (indicated by over bar), (Fig. 2.3.9):
i_nwhich:
Us = Stokes net velocity (ratio of net horizontal displacement and wave period)
o = 2xlT = wave frequency
k = 2x/L = wave number
z = vertical coordinate (positive downwards from mean water level)
- o kH2
At the bed (z =h) : Us =
8 sinh2(kh)
At the surface (z =0): eS =

o k H~ cosh(2kh)
8 sinh2(kh)
For waves propagating in a horizontally unbounded domain the depth-integrated volume flux
(m2/s) is:
in which:
c = (g/T)/(2x coth(kh)) = wave celerity
Equation (2.3.32) reduces to M, = o H2/8 for deep water (kh > 1).
For waves propagating in a horizontally bounded domain it is. appropriate to impose a
condition of zero volume flux at each location (x), yielding (see Fig. 2.3.9):
Equation (2.3.33) can be seen as the sum of the forward Stokes drift and an uniform return
flow. The generation of a positive volume flux near the surface and a negative flux near the
bottom requires the presence of a horizontal pressure gradient (shear stresses are absent in
a non-viscous flow) caused by a "set-up" of the free surface towards the coast (similar to
wind set-up).
The volume flux (m2/s) at a fixed location (x) in an unbounded domain can also be
determined in an Eulerian way, as follows:
in which:
U = instantaneous horizontal velocity at height z
q = water surface displacement from the mean surface level
Below the wave trough the time-averaged value of the horizontal velocities is zero. However,
in the region between the wave crest and the trough there is an asymmetry of the horizontal
velocity indicating that more fluid moves in the wave direction under the crest than in the
trough region.
For small-amplitude waves Eq. (2.3.34) yields:
The Eulerian and the Lagrangian method yield the same depth-integrated volume transport.
The vertical distribution of the time-averaged transport velocities, however, is different for
both methods.
Dalrymple (1976) has presented information of the vertical distribution of the Eulerian
velocities in the region between the wave crest and the trough, as shown in Fig. 2.3.10.
2. Viscous and turbulent oscilkatory flow
Longuet-Higgens (1953) has shown that for real fluids with viscosity v there is a time-
averaged net downward transfer of momentum into the boundary layer by vispous diffusion (v WaZ)
causing a mean Eulerian flow (UJ in addition to the Stokes drift (UJ . The mean Eulerian
flow can be seen as the mean velocity of thgorbit centers.
The total time-averaged transport velocity U(), is defined, as:
For laminar flow in the boundary layer Longuet-Higgens (see Russel and Osorio, 1957)
derived:
in which:
6 = ( 2 ~ / o )=
~ .thickness
~ of laminar boundary layer
Equation (2.3.38), shown in Fig. 2.3.1 1, has a maximum value
For z/6 -, = , Equation (2.3.38) yields:
in which:
fr, = peak value of orbital velocity just outside boundary layer
c = wave celerity (o/k)
Assuming zero volume flux over the full water depth, Longuet-Higgens (1953) derived:
The function F(z/h) is shown in Fig. 2.3.12 for some values of kh (see also Fig. 2.3.9).
Equation (2.3.41) can be seen as the sum of the forward Stokes drift (Eq. (2.3.29)) and a
parabolic velocity distribution. This latter distribution gives a forward flow at the bottom and
a backward flow at mid-depth.
Equation (2.3.4 1) is valid for H < 26, yielding a wave height range that is of little practical
interest. Based on a comparison with experimental results (Russel and Osorio, 1957), fairly
good predictions are obtained for 0.7 < kh < 1.5.
At the bottom (z =h) Equation (2.3.41) yields (see also Fig. 2.3.1 1):
water surface
.a-
*./-
../.
*./
w kh = 1
,./I Hlh = It3
.0* --
../*.
bottom
-0.8 -0.4 0 0.4 0.8
1.2 1.6 2.0 2.4 2.8 3.2
----+4 U m l ~ k ~ ~
Stokes drift velocity, Eq. (2.3.33), bounded domain
----Stokes drift velocity, Eq. (2.3.29), unbounded domain
-.-.-.total mass-transport of Longuet-Higgens, Eq. (2.3.41), bounded
-...-...- Eulerian mass-transport velocity of Craik, Eq. (2.3.46),unbounded
Figure 2.3.9 Net time-averaged transport velocities in non-breaking waves (Stokes 1847,
Longuet-Higgens 1953, Craik 1982)
z deep water 1 I I
shallow water
Ow = maximum horizontal velocity

GI = eulerian velocity averaged w e r wave period
G2 = eulerian velocity CI~~raged
over occurrence period
Figure 2.3.10 Net time-averaged transport velocities in non-breaking waves

(Dalrymple, 1976)
For oscillatory flow in an unbounded domain the Eulerian time-averaged transport velocity
(due to viscosity effects) can be described by (Craik, 1982):
Equation (2.3.46) is shown in Fig. 2.3.9 for kh = 1.-

The depth-integrated volume flux is:
Longuet-Higgens also showed that Eq. (2.3.45) can be used to describe the time-averaged
transport velocity just outside the boundary layer in case of turbulent smooth oscillatory flow.
This was confirmed by Johns (1970). For turbulent conditions the velocity distribution inside
the boundary layer is different from Eq. (2.3.38) because the turbulent (eddy) viscosity differs
from the molecular viscosity. Information of the time-averaged transport velocities in
turbulent rough conditions is rather scarce. Computed results of Brnker Hedegaard (1985) are
shown in Fig. 2.3.13.
Finally, some measuring results are presented. Figure 2.3.14 shows measured Eulerian net
velocities for oscillatory flow over a smooth bottom in a laboratory channel (Borghei, 1982).
The measurements were made in fixed points using a Laser Doppler velocity meter.
Comparison with the net velocities of Longuet-Higgens shows reasonable results.
Figure 2.3.15 presents measured residual Eulerian velocities for sinusoidal oscillatory flow
over a symmetrical ripple, showing a pair of circulation cells. These patterns seem to be
strongly influenced by the formation of lee vortices.
Bijker et a1 (1974) confirmed the presence of drift velocities as proposed by Longuet-Higgens.
However, the measured velocities were found to be considerably smaller than the theoretical
values. Bijker et al concluded that the near-bed net velocities on a sloping bottom (in the
range of 1: 10 to 1:40) are determined by the local depth rather than the slope angle. Their
results also show a considerable reduction of the near-bed net velocity when ripples are
present on the bottom.
- total mass-transport velocity, Eq. (2.3.38)

- Eukrian mass -transport velocity
Figure 2.3.11 Net time-averaged velocities in wave boundary layer of non-breaking waves
(Longuet-Higgem, 1953)
Figure 2.3.12 Vertical distribution of net velocities in nun-breaking waves (Longuet-
Higgens, 1953)
laminar flow (6/ks=8 )

turbulent rough flow
(A6 lks = 1000)
-
---+ uc/Oa2
Figure 2.3.13 Eulerian net velocities in non-breaking waves (B. Hedegaard, 1985)
8
p =(wl2v)"=
I
I T =1.4s
6 .I I
C H =0.0754 m
a \ h =0.3m
N
cl -
,\;. Uma, = 0.0253 mls
t4
~2
f
a L\ measured
- total mass-transport of
Longuet - Higgens
-,
,,aulerian mass- transport of
#'
0 / Longuet - Higgens
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
----+ 616max
Figure 2.3.14 Measured net velocities in non-breaking waves (Borghei, 1982)

Figure 2.3.15 Measured net (Eulerian) velocities over a rippled bed in non-breaking
sinusoidal waves (Sato et al, 1984)
' C A K S)gp
B (CASE 1)
& f
-----.---p -.-------
p --------- p
-
---------
--------
--.------- p ----,----S
P
..
-- -- .----5
*
. .
--."----
.
5
--'.,..-.-...
- .--.--.
P
p
--.------
s ----------5
" ..
9
..
..
9
. . SWL
6
Horizontal component, G.
Figure 2.3.1 6 Net velocities in breaking waves (Nadaoka et al., 1982)
wave propagation t
onshoro h
mean bed levd
Figure 2.3.17 Net velocities in breaking waves

2.3.8 Mass transport by breaking waves
Breaking waves generate a net current in longshore direction (longshore current) and in
offshore direction (undertow).
Herein, the attention is focussed on the undertow. The measurements of Nadaoka et al(1982)
and Stive (1980, 1983) using a Laser Doppler velocity meter in flume conditions clearly show
the generation of offshore-directed velocities inside the breaker zone, see Fig. 2.3.16. The
measured velocities of Nadaoka et id represent time-averaged-values of 70 wave cycles. The
measured net velocities above the wave trough are based pn the assumption of zero-velocity
during the period that the measuring location was above thC? water surface. They also analyzed
the time variations of the velocities showing a strong asymmetric motion with relatively large
velocities of short duration in onshore direction and relatively small velocities of long
duration in offshore direction.
Above the trough level there is a net volume flux in onshore direction. According to a first
order approximation, the volume flux above the trough level-can be estimated as:
Applying c = (gh)'.' in shallow water, it follows that:
Assuming no net flow over the depth, the time-averaged return flow, which is also called the
undertow, below the trough level is given by (see Fig. 2.3.17):
Taking h, = 0.8h, it follows that:
- = 0.15 g05 h-1" ~2

urnoff
2.4 Combined current and waves
2.4.1 Introduction
When waves propagate into shallow waters near coasts they may encounter relatively strong
currents which affect the wave characteristics, the current velocities and the bed-shear
stresses.
Herein, the main attention is focussed on the latter two phenomena. Modification of the wave
height by the current effect is not studied.
2.4.2 Wave characteristics
Especially, opposing currents have a significant influence by steepening the waves even to
the point of breaking (Brevik and Aas, 1980). A following current enlarges the wave trough
and thereby the wave length and reduces the wave height (Brevik and Aas, 1980).
Linear wave theory can be applied with respect to a coordinate system moving with the
current velocity (Jonsson -et al, 1970;.Jonsson, 197%).In that case the wave length remains
constant and the wave period is altered: larger for a following current 6 > 0) and smaller
for an opposing current 6 < 0).
A general expression for the wave period (TJrelative to the moving coordinate system is:
-u which:
in
= depth-averaged current velocity
4 = angle between current and wave direction ( 4 = 0 for following, and
@ = 180" for opposing)
L' = wave length in presence of waves from [L '/T - 7i cos+r = [(& '/2x) tanh(2xh/l')]
c = wave propagation velocity
T = absolute wave period
2.4.3 Current velocities and bed-shear stresses
1. Enperimental results
Laboratory experiments show a distinct influence of the waves on the current velocity profile
(Bakker and Van Doorn, 1978; Kemp and Simons, 1982, 1983; Nieuwjaar and Van der
Kaaij, 1987; Nap and Van Kampen, 1988; Van der Stel and Visser, 1985; Visser, 1986).
Kemp and Simons (1982, 1983) measured mean velocities in regular waves with following
and opposing currents over a smooth bed and over a rough bed (5 mm high triangular strips
spaced at 18 mm). The experiments were performed in a wave flume (length = 14.5 m,
width = 0.46 m, height = 0.70 m) with wave generation by a bottom-hinged paddle.
Considerable attention was given to the establishment of an optimum in- and outflow of the
current. The still water depth was 0.2 m and the wave period was 1 sec. The relative wave
heights (Hlh) were in the range of 0.1 to 0.25. Figure 2.4.1 presents the results of Kemp and
Simons. Some results were only reported qualitatively.
The following phenomena can be observed:
following current:
smooth bed increased velocities near the bed
reduced velocities near the water surface
larger effect for higher waves
rough bed reduced velocities near the bed and water surface
increased velocities at intermediate depths
opposing current:
smooth and rough bed reduced velocities near the bed
increased velocities near the surface
larger effect for higher waves
Kemp and Simons also determined the current-related bed-shear stress (rbSc) and the effective
roughness (kJ by plotting the velocities measured in-the lower layer on a semi-logarithmic
scale. The current-related bed-shear stress for a combination of waves and current was a
factor 2 to 3 larger than for a current alone. The effective roughness for a current with
opposing waves was 0.18 m, which was as large as the water depth(!), while that for a
current alone was 0.025 m. The bed roughness in combined current and waves should be
interpreted as an apparent bed roughness (kJ because it largely reflects flow resistance due
to pressure forces generated by the strong vortices moving forward and backward over the
roughness elements. Therefore, this k-value cannot be related to the geometrical properties
of the roughness elements.
Nieuwjaar-Van der Kaaij (1987) and Nap-Van Karnpen (1988) measured time-averaged
velocities using an electromagnetic velocity meter in irregular waves over a (natural) rippled
bed with following and opposing currents.
The experiments were performed in a wave flume (length = 45 m, width = 0.8 m and
depth = 1 m). The height and length of the sand ripples were resp. 0.02 m and 0.15 m. The
wave heights were in the range of H,/h = 0.15-0.35. The peak period of the waves was about
T, = 2.5 s. The fluid velocities in each point were averaged over time (= 256 s) and over
space (= 0.6 m).
Space-averaging was performed by moving the measuring carriage on top of the flume slowly
forward and backward during the measuring period.
The results of Nieuwjaar-Van der Kaaij are shown in Figure 2.4.2.
The following phenomena can be observed.
following currents:
reduced velocities near the bed (zlh < 0.2)
reduced velocities near the water surface (z/h > 0.6)
increased velocities at intermediate depths
most pronounced effect in case of a weak current and high waves
opposing current:
reduced velocities near the bed (zlh < 0.4)
increased velocities near the water surface (z/h > 0.5)
most pronounced effect in case of a weak current and high waves.
Visser (1986) measured time-averaged velocity profiles in a wave-current basin with an angle
of 90" between the current direction and the wave direction (regular waves). A smooth bed
and a gravel bed (d, = 0.005 m) was used. The current velocities were measured with an
immersible Laser Doppler velocity meter. The velocities were averaged over 100 s. Relative
wave heights were in the range of 0.25 to 0.5. The wave period was 1 s.
WAVES AND FOLLOWING CURRENT
,-• mean fluid vglocity, ul6
-
---
current alone, ii= 0.18 mls,
current and waves, fi =O.18mls, Hlh=O.lO
-
---
current alone
current and waves
. current and waves, 6 ~0.17mls, Hlh=0.22 (G and H not given)
A. SMOOTH BED (h = 0.2 m) B. ROUGH BED (h =0.2m)
WAVES AND OPPOSING CURRENT
------+ maan fluid wlocity, ulG
-
-
current alone
current and waves
-
---
current alona, u'=0.10 m h
current and waves, i i = O . l l mls, H/h=0.14
(G and H not givan) -.-. current and waves, 6 = 0.125 mk, Hlh =0.30
A SMOOTH BED (h=0.2m) 6. ROUGH BED (h =0.3m)
Figure 2.4.1 Influence of waves on current velocity profile; waves following and opposing
the current (Kemp and Simons, 1982, 1983)
=
-
o H, 0.75 rn
x H, =
0.10 m
v H, 0.12 m
H, 0.15 rn
current and waves
------ current alone
H, = significant wave height
U
, = depth-mean velocity
Figure 2.4.2 Influence of waves on current velocity profile; waves following and opposing
the current (2Vieuwjaar and Van der Kaai~',1987)
The mean current velocity was in the range of 0.1 to 0.2 mls.
Some results are presented in Figure 2.4.3, showing reduced velocities in the near-bed region
and increased velocities in the near water surface region.
Havinga (1992) measured time-averaged velocity profiles over a rippled sediment bed in a
wave-current basin. Relative wave heights were in the range of 0.2 to 0.35. The mean current
velocity was in the range of 0.1 to 0.3 mls. The angle between the wave and current direction
was 60°, 90" and 120" respectively. The largest reduction of the near-bed velocities was
observed for an angle of 90".
Summarizing all results, it can be concluded that the current velocities in the near-bed region
are reduced by the wave-induced vortices in the wave boundary layer. This effect is most
pronounced in case of a relatively weak current and relatively high waves. In case of an
opposing current the reduction of the near-bed velocities is somewhat larger than in case of
a following current. When the wave direction is perpendicular to the current direction, the
reduction of the near-bed velocities is largest. The current velocities in the upper layers are
increased in case of an opposing current and in case-the waves are perpendicular to the
current. When the waves are in the same direction as the current (following), the velocities
near the surface are reduced.
2. Mathematical models
Most theoretical models have some common features which can be summarized as follows:
the wave motion is a potential flow over the entire water depth except in the thin wave
boundary layer,
the wave potential flow does not interact with the turbulence of the steady flow,
the steady flow outside the wave boundary layer can be treated independently from the
wave motion.
Inside the wave boundary layer turbulence generated by the wave and the current motion will
affect the velocity profile of the mean current. Due to the non-linear interaction of the mean
current and the wave boundary layer all theoretical models predict an increase of the flow
resistance for the mean current. The effect can be described as an apparent wave roughness
(kJ, which is larger than the physical bed roughness (kJ. The presence of the waves will
therefore cause a larger bed-shear stress for a given flow rate or a reduced flow rate for a
given bed shear stress. The models for combined wave and current boundary layers can be
divided in two main groups: models based on a combination of time-constant eddy viscosity
profiles and models based on higher-order turbulence closures.
One of the first models was the engineering method of Bijker (1967). It is based on the
mixing length concept. Bijker considered a certain point (z' = e z,) above the bed at which
a line emanating from the bottom (z=O) is tangential to the current velocity profile. The
wave-induced orbital motion in the boundary layer is assumed to have a logarithmic profile
and thus a straight line can be drawn emanating from the bed to be also tangential to the wave
velocity profile. Therefore, once the velocity at the level z' = e z, is described, the velocity
gradient at this level is automatically known and the shear stress can be determined from the
mixing length theory.
The model of Lundgren (1972) includes a description of the conditions in the wave boundary
layer. The eddy viscosity distribution inside the boundary layer was estimated from measured
values averaged over a wave period (see Fig. 2.4.4).
The mean current velocity profile is assumed to have the normal logarithmic shape above the
wave boundary layer, but the increased eddy viscosity in the wave boundary layer will give
a higher flow resistance than corresponding to the natural bed roughness alone.
smooth bed
-----+currant wlocity (mls)
r
1441 T 401
maowrod
cornputd
HI (m)
-
a068
,
,
,
X
,
,
0
T (5) 2
6s (mls) 0.21
ii (mls) 0.105 a082
0 P) 90
smooth bed
- currant wlocity (mls)
Figure 2.4.3 Influence of waves on current velocity profile; waves perpendicular to current
(Visser, 1986)
The characteristics of the wave boundary layer are determined from the wave parameters
alone, without taking the current into account. The model is therefore only valid in situations
with a relatively weak current.
The model for combined wave and current motion by Smith (1977) is based on a linear eddy
viscosity distribution in the wave boundary layer (see Fig. 2.4.4). This eddy viscosity
distribution is continuous at z = 6,, but it is hardly realistic that the eddy viscosity over the
entire flow depth receives a contribution from the wave-induced eddy viscosity. It must be
expected that the contribution from the wave-generated turbulence gradually decays with the
distance from the boundary layer.
Grant and Madsen (1979) presented a model which in some respects is similar to that of
Smith (1977). They also apply a time constant eddy viscosity.
The model has been improved to cover an arbitrary angle between the waves and the current.
The eddy viscosity varies linearly with the distance from the bed. Inside the wave boundary
layer the slope is determined by the largest shear velocity from the waves and the current (see
Fig. 2.4.4). Outside the wave boundary layer, the eddy viscosity is determined by the mean
shear stress. The eddy viscosity is discontinuous at z = 6,. The mean current is given as two
logarithmic velocity profiles intersecting at z = 6,.
Some modifications to the model of Grant and Madsen (1979) have been introduced by
Christoffersen and Jonsson (1985). They apply a constant eddy visocity in the wave boundary
layer. The eddy viscosity distribution is given in Fig. 2.4.4.
The model of Fredsse (1984) does not apply the eddy viscosity directly in the description of
the phenomena. The basic assumption is that the time scale for production and decay of
turbulent kinetic energy is small compared to the wave period. Based on this, the formation
of the wave boundary layer can be treated independently for each half wave period, and the
velocity profile in the wave boundary layer can be assumed to be logarithmic, corresponding
to instantaneous equilibrium in the turbulence energy.
The development of the wave boundary layer during each half wave period is described by
the momentum equation integrated over the wave boundary layer thickness. The current
velocity profile outside the wave boundary layer is used as a boundary condition in the
momentum equation. This outer profile is described by a logarithmic velocity profile,
determined through the apparent wave roughness (Q analogous to other models, e.g. Grant
and Madsen (1979).
The eddy viscosity concept is not used directly in this model, but the (time-variant) eddy
viscosity can be estimated from the bed-shear stress and boundary layer thickness calculated
by the model.
The model of Bakker and Van Doorn (1978) applies a mixing length turbulence model; the
eddy viscosity is a function of the local flow conditions. The eddy viscosity varies with time
and with the distance from the bed. It is therefore interesting, that a comparison made by
Bakker and Van Doorn between their theory and that of Lundgren shows results of the same
order of magnitude, though Lundgren assumes a time constant eddy viscosity.
Davies et al. (1988) have presented a model based on a one equation turbulence model (k-
equation). A two-equations model has been presented by Hagatum and Eidsvik (1986).
Similar models were presented by Myrhaug et al (1990) and Huynh Thanh et al (1991).
The flow resistance predicted by the models of Lundgren (1972), Bakker and Van Doorn
(1978), Grant and Madsen (1979) and Fredsse (1984) was compared (by Davies et al), where
it was possible to make the calculations based on the information given in the papers.
A. Lundgron (1972)
B. Smith (1977)
-
C. Grant Madsen (1979)
D. C h r i s t o f f ~ r s ~(1982)
n
E. .Myrhoug (1982)
Figure 2.4.4 Eddy viscosity distributions
.........-
TCtCOR*:
ORANT L MIDSEN
200- -
loo - ilk 100
.
n
Y, =- ...' 01
YLISURCYINTS'
*0 SenlZS:
*
- MY.
T :- . +
.
USCY
+
USCHI PEW L e v t L s ~
5
--- - AMCLITUOL
.
-.-a- BANKER AND W R N (1978) LCOC*O
O.*Am 0 I7m
2 - 4
/
0 W A N T AND MADSEN Il9fOI
FPEDSBE I19821 - 10.
IS*
r*
A
0
0
Figure 2.4.5 Comparison of apparent roughness according to various models

Figure 2.4.6 Apparent roughness for following waves (4 = 0") and waves perpendicular
(4 = 90") to current according to Freabe (1984)
instantaneous
current velocity
velocity
---d
Figure 2.4.7 Vertical distribution of current velocity and wave velocity profile
The results are shown in Fig. 2.4.5 depicting the increase in the apparent roughness (k)
acting on the mean current outside the wave boundary layer versus the ratio between the
maximum wave orbital velocity @,) and the mean bed shear velocity from the current (u,,~).
Soulsby et al (1991) presented an intercomparison of various numerical models. They found

a broad similarity between theories in many cases.
The model. of Fredsae was found to give reasanable results. This is also concluded by Visser
(1986) comparing his experimental results (waves perpendicular to currents) with the results
of the Fredslere model.
Finally, Figure 2.4.6 is presented showing basic results of the Fredslere-method for 4 = 0"
(waves following the current) and 4 = 90" (waves perpendicular to the current).
3. Engineering methods
Bijker estimated the resulting bed-shear stress of combined wave-current conditions by

considering the near-bed flow structure and by assuming the validity of the logarithmic
distribution.
current
at level q = e z,, it follows that (see Fig. 2.4.7):
waves
The instantenous bed-shear stress is defined:
Assuming an instantaneous logarithmic velocity distribution near the bed, it can also be
defined (at level q = e z,), that:
Define: Uw,,= p 0, sin(ot)

From Eqs. (2.4.4), (2.4.5) it follows that:
(current direction)
--
\Ir.
1 Ga (wave direction)
Figure 2.4.8 Vector addition of instantaneous wave and current velocity m level q = e z,
I- - EXPLICIT APPROXIMATION I
-- --- NUMERICAL SOLUTION
0.7
0.6-
0.5-
0.4 -
0.3 -
0.2 -
0.1 - -- 1
OD8 -- -
-- 0.7
OD6 -
- -0.5
ODL- -
- - 0.3
0.02 - -0.2
I 1 1 1 1 1 1 I I I I I I l l O , ,
0.1 0.2 0.3 0.5 0.8 1 2 3 4 5 6 7 8 9
[;,5/i
Figure 2.4.9 -co@cient as a functr'on of 2 o& according to

current and waves
Assume that there is an angle @ between the wave and current direction (see Fig. 2.4.8). The
instantaneous resulting velocity at level q = e z, can be expressed as:
cV,,I2 = ( u ~ , , )+~ (vW,,)2 + 2 U , , u,, cosd
The instantaneous bed-shear stress is:
Substitution of U , , = p 0, sin(ot)-in Eq. (2.4.7) yields:
The value of the time-averaged bed-shear stress vector can be found by averaging over the
wave period of Eq. (2.4.9), yielding:
in which:
b,cw = time-averaged value of absolute bed-shear stress for combined wave-current flow
Z blC = time-averaged current-related bed-shear stress = (118) p fc ii2
T
,b,w = time-averaged value of absolute wave-related bed-shear stress = (1/4) p ,f 0;
u = depth-averaged current velocity
fc = 8glC2 = 0.24 log'2(12h1kJ = current friction factor
C = 18 log(l2h/kJ = ChCzy-coefficient
k, = effective bed roughness of Nikunadse
f,= exp[-6 +
5.2 ( A d l ~ ) ~ . =
' ~ wave
] friction factor (f, 5 0.3)
IJ,peak value of near-bed orbital velocity (= fi,~/2n)
=
A, = peak value of near-bed orbital excursion
T = wave period
6 = thickness of wave boundary layer
Equation (2.4.12) expresses the time-averaged value of the absolute bed-shear stress vector
and is found to be independent of the angle @ between the wave and current direction.
This parameter is most important for the sediment transport process because it governs the
agitation (stirring) of the bed particles.
The most important parameter for resisting the mean current (bed-shear stress experienced
by the current) is the time-averaged value of the component of the bed-shear stress parallel
to the current direction (r,,?J. Its effect on the current is reflected by the slope (outside
wave boundary layer) of the tlme-averaged current velocity profile, as shown in Fig. 2.4.10.
The rk,m,parameter cannot be obtained by analytical integration. Bijker presented an
approxlmatlon function based on numerical integration.
in which:
t b,carg = time-averaged bed-shear stress parallel to current-direction.
fl = 0.75 + 0.45 (( o ~ v ) ~for. ~4 ~< 70'
C = (f, c2/2g)lP = ~ ( f ~ f ) "
For 4 = 90" it can be derived that:
Yoo (1989) presented a graphical form of the P -coefficient (Eq. 2.4.13) for 4 = 0°, 45",
90", as shown in Fig. 2.4.9.
The major defect of the Bijker-method is the neglect of the current velocity reduction in the
near-bed region due to wave-current interaction. O'Connor and Yoo (1988) introduced a
velocity reduction factor a to account for this effect. Figure 2.4.9 shows fl -values for a =
0.38, 0.58 and 0.78.
Tolman (1992) studied the effect of the current on the wave-related friction and found little
effect in the practical range of interest.
Roelvink (1990) modified the Bijker-approach and represented the wave-current interaction
effect by calibration using experimental data.
He proposed:
Coffey and Nielsen (1986) proposed an engineering model based on a logarithmic velocity
profile. The wave effect on the steady current is represented as a shift of the velocity profile
yielding an apparent roughness k, (Fig. 2.4.10). The ratio k,/k, was obtained from
experimental results and related to the ratio of the wave-related and current-related bed-shear
velocities (u.,,/u+,,).
A disadvantage of this model is that iterative computations are required.
C. Van Rijn (1989):
Van Rijn proposed a straightforward method for the rough flow regime, which is based on
the application of logarithmic velocity profiles.
The velocity profile for a current in the presence of waves is described as a two-layer system
(see Fig. 2.4.10):
logarithmic velocity profile affected by bed-form roughness (kJ inside the near-bed mixing
layer (8), which is assumed to be equal. to three~thesthe thickness of the wave boundary
layer,
logarithmic velocity profile affected by apparent roughness (u outside the near-bed mixing
layer (8).
k, /30
"6
----+relative velocity, v , / i
Figure 2.4.10 Influence of waves on current velocity profile

The expressions for rough turbulent flow are:
with : vrs8-- % ln(3081kJ

-1 + In(30h/k,)
in which:
v,, = modified current velocity at height z above bed '2 0 5
v, = magnitude of depth-averaged velocity vector = + v ]
6 = thickness of wave-related near-bed mixing layer (= 3 8 2 , (6 > 0.033 4)
6 , = thickness of wave boundary layer, Eq. (2.3.8)
v*,, = current velocity at z = 6
k, = bed roughness
k, = apparent bed roughness
The overall resistance (in current direction) experienced by the current due to friction and
vortices-related pressure effects as modified by the wave motion, follows from Eq. (2.4.15)
yielding:
D e f ~ rb,,
g = 118 p fcwfiR)', it follows that:
in which:
5b,rn9 = time-averaged shear stress experienced by the current in combined wave-current
conditions
The current-related shear stress (v,, or tbJ acting on the bed is reduced due to the reduced
current-velocities in the near-bed region due to the wave-current interaction.
From Equations (2.4.16), (2.4.17), it follows that:
The bed-shear stress in a current alone (no waves) is:
From Eqs. (2.4.21), (2.4.22) it follows that:

with
= bed-shear stress reduction factor ( 5 1)
The a,-coefficient is considerably smaller than one. m case of relatively high waves combined
with a weak current.
The time-averaged value of the absolute bed-shear stress (Eq. 2.4.12) acting on the bed
material particles, now becomes:
with rbScand r, according to Eq. (2.4.12) and a, according to Eq. (2.4.24).
The time-averaged shear-stress experienced by the current in the presence of waves is given
by Eq. (2.4.19). For application in mathematical models, it is proposed to use:
-
in which:
v, = depth-averaged velocity vector = [t2+ V 2 ] l P
f,, = current-related friction factor = 0,24[10g(12h/kJ]-~
The apparent roughness (k3 was determined by fitting of logarithmic profiles to time-
averaged velocity profiles (outside the wave boundary layer) measured in the presence of
waves. The experimental data of Nieuwjaar-Van der Kaaij (1987), Nap-Van Karnpen (1988)
+
and Havinga (1992) representing a data set with -values of 0°, 60°, 90°, 120" and 180"
(4 = angle between wave and current direction).
Basically, the ratio k,/k, should be a function of:
the ratio of the peak orbital velocity and depth-averaged velocity 0JVR,
the ratio of the peak orbital excursion and the bed roughness, A&,
the angle + between the current and wave direction.
Thus,
The oJVR ratio represents the relative strength of the wave and current motion. For
U6/VR + 1 it follows that lg/k, -. 1. The Ab/ks ratio represents the influence of the wave
period because at the same ~~/V,-valuea decrease of the A,-value means a smaller wave
period (faster oscillations) which probably results in an increase of the apparent roughness
4. The mathematical model results of Fredsse (1984) show a rather weak influence of the
A&-ratio for rough flow in the practical range of interest (see Fig. 2.4.6). Therefore, this
parameter was neglected.
Based on analysis of the experimental data of flumes with rippled beds (see Fig. 2.4.1 I), it
was found that:
The y -coefficient was found to be dependent on the angle between the current and the waves,
as follows:
$ = 0" (following waves) : y = 0.75 f 0.4
$ = 60" : y =1.1 f0.4
$ = 90" (perpendicular waves) : y = 2.1 f 0.9
$ = 120" : y =1.7 f 0.4
$ = 180" (opposing waves) : y=1.1 50.4
A reasonable estimate of y can be obtained by:
in which:
$ = angle between current and wave direction
(in radians between 0 and x; 112 x = 90°, x = 180")
The experimental data show that the largest &-values and hence the largest values do
occur for a wave-current angle of 4 = 90'. Figure 2.4.12 shows values as a function
of T ~ , ,-values for $ = 90" and 0" according to the method of Van Rijn. Both parameters
were made dimensionless with r b c + ib,w,where rbSc= current-related bed-shear stress (no
waves) and ibW = peak wave-related bed-shear stress (no current). The largest cbby-values
do occur for 4 = 90'. For example, the 7,-value for O6/G = 1 and $ = 90 1s about
30% larger than that for $ = 0".
This latter behaviour is opposite to the mathematical model results of Fredsnre (1984) for
plane beds who found smaller &-values for $ = 90" than for $ = 0".
Figure 2.4.13 shows a comparison of computed values according to six different
methods for $ = 0" , Adzo = lo3, zo/h = lo4. The results of the numerical models of
Fredsere (1984), Davies et al (1988), Myrhaug et a1 (1990) and Huynh Thanh et al (1991)
were presented by Soulsby et al (1991). The Bijker-model yields the largest values
because the velocity reduction in the near-bed region due to wave-current interaction has been
neglected resulting in an overprediction of the value.
The method of Myrhaug et ai yields relatively large values for rb,J((rb + ibSw)> 0.5. The
engineering method of Van Rijn yields the smallest rbSm-values.
Computed velocity profiles based on Eqs. (2.4.15), (2.4.16), (2.4.17), (2.4.29) and (2.4.3 1)
are shown in Fig. (2.4.3) for a current in a wave field (4 = 90") as measured by Van der
Stel and Visser (1985). The computed current profiles are in good agreement with the
measured values.
Figure 2.4.11 Ratio of apparent and physical roughness (k,kJ
Figure 2.4.12 Computed r b,m-valuesfor 4 = 0' and 90' according to method of Van Rijn
- Van Rijn
----- Bijker (1966, 1986)
-.-.- Fredsae (1984)
-..-..- Davies et al (1 988)
--- Myrhaug et ol (1990)
Huynh
.-'------. Thanh e t al (1991)
Figure 2.4.13 Comparison of computed T,,-values for 4 = 0"

REFERENCES
Bakker, W.T., 1974. Sand Concentration in an Oscillatory Flow. Proc. 14th Conf. Coastal
Eng., Copenhagen, pp. 1129-1148.
Bakker, W.T. and Van Doorn, Th.,1978. Near-Bottom Velocities in Waves with a Current.
Coastal Eng. Conf., Paper 110, Hamburg, West-Gemany.
Battjes, J.A., 1974. Surf Similarity Parameter. Proc. 14th Conf. on Coastal Eng.,
Copenhagen, Denmark.
Borghei, S.M., 1982. Oscillatory Boundary Layer over Fixed Rough Beds. Univ. of
Nottingsham, Dep. of Civ. Eng., England.
Brevik, I. and Aas, B., 1980. Flume Experiments on Waves and Currents, Rippled Bed.
Coastal Engineering Vol. 3, p. 149-1 77.
Brevik, I., 1981. Oscillatory Rough Turbulent Boundary Layers. Journal of Waterways,
Harbours and Coastal Eng., ASCE, Vol. 107, W 3 .
Broker Hedegaard, I., 1985. Wave-generated Ripples and Resulting Sediment Transport in
Waves. Inst. of Hydrodynamics and Hydraulic Engineering, Technical University of
Denmark, Lyngby, Denmark.
Bijker, E.W., 1967. Some Considerations about Scales for Coastal Models with Movable
Bed. Dissertation, Delft University of Technology, Delft, The Netherlands.
Bijker, E.W., 1986. Coastal Engineering, Vol. 11. Lecture Notes, Delft Univ. Technology,
Delft, The Netherlands.
Bijker, E.W., Kakwijk, J.P. and Pieters, T., 1974. Mass Transport in Gravity Waves on
Sloping Bottom. Coastal Eng. Conf., pp. 447-466, Copenhagen, Denmark.
Christoffersen, J.B. and Jonsson, I.G., 1985. Bed Friction and Dissipation in Combined
Current and Wave Motion. Ocean Engineering, Vol. 12, No. 5.
Coffey, F.C. and Nielsen, P., 1986. The Influence of Waves on Current Profie. Coastal
Eng. Conf., Taipeh, Taiwan.
Coleman, N., 1970. Flume Studies of the Sediment Traqfer CoqJicient. Water Resources
Research, Vol. 6, No. 3, USA.
Craik, A.D.D., 1982. The Dri! Velocity of Water Waves. Journal of Fluid Mechanics, Vol.
116, pp. 187-205.
Dalrymple, R.A., 1976. Wave-induced Mass Transport in Water Waves. Journal of the
Waterways, Harbours and Coastal Engineering Div., Vol. 102, No. WW.2.
Davies, A.G., 1985. Observations of the Stability of Oscillatory Flow above the Seabed and
of Sand Ripple Fomation. Continental Shelf Research, Vol. 4, No. 5.
Davies, A.G., Soulsby, R.L. and King, H.L., 1988. A Numerical Model of the Combined
Wave and Current Bottom Boundary Layer. Journal of Geophysical Research, Vol.
93, NO. Cl, p. 491-508.
REFERENCES (continued)
Dean, R.G., 1986. Intercomparison of Near-Bottom Kinemutics by Several Wave Theories
and Field and Laboratory Data. Coastal Engineering, 9, pp. 399-437.
Dutoit, C.G., 1982. Wave-Induced Velocities close to a Rippled Bed. Proc. Conf. Coastal
Engineering, p. 312-327, Cape Town, South Africa.
Fredsse, J., 1984. Turbulent Boundary Layer in Wave-CurrentMotion. Journal of Hydraulic
Engineering, Vol. 110, No. 8.
Grant, W.D. and Madsen, O.S., 1979. Combined Wave and Current Interaction with a
Rough Bottom. Journal of Geoph. Research, Vol. 84, No. C4, pp. 1797-1808.
Hageturn, K. and Eidsvik, K.J., 1986. Oscillatory Turbulent Boundary Layer with
Suspended Sediment. Journal of Geophysical Research, Vol. 91, No. C l l .
Havinga, F., 1992. Sediment Transport in Non-Breaking Waves combined with a Current.
Dep. of Civil Engineering, Coastal Eng. Dep., Delft University of Technology, Delft,
The Netherlands
Honji, H., Kaneko, A. and Matsunaga, N., 1980. Flow above Oscillatory Ripples. Sedi-
mentology, Vol. 27, p. 225-229.
Hotta, S. and Mjzuguchi, M., 1980. A Field Study of Waves in the Surf Zone. Coastal
Engineering in Japan, Vol. 23, 1980.
Huynh Thanh, S. and Temperville, A., 1991. A Numerical Model of the Rough Turbulent
Boundary Layer in Combined Wave and Current Interaction. Euromech, 262, Bakernu,
Rotterdam, The Netherlands.
Iwata, K. and Sawaragi, T., 1982. Wave Defonnat'on in the Surf Zone. Memoirs of the
Faculty of Engineering, Nagoya Univ., Vol. 34, No. 2, Nagoya, Japan.
Johns, B., 1970. On the Mass Transport Induced by Oscillatory Flow in a Turbulent
Boundary Layer. Journal of Fluid Mechanics, Vol. 43, part 1, pp. 177-185.
Jonsson, I.G., 1966. Wave Boundary Layer and Frim'on Factors. Proc. 10th Con$ Coastal
Eng., p. 127-148, Tokyo, Japan.
Jonsson, I.G., Skougaard, C. and Wang, J.D., 1970. Interaction between Waves and
Currents. Proc. 12th Con$ on Coastal Eng., p. 489-508, Washington, USA.
Jonsson, I.G. and Carlsen, N.A., 1976. Experimental and Theoretical Investigations in
an Oscillatory Rough Turbulent Boundary Layer. Jounuzl Hydr. Research, Vol. 14,
. NO. 1, p. 45-60.
Jonsson, I.G., 1978. Energy Flux and Wave Action in Gravity Waves Propagating on a
Current. Journal of Hydraulic Research 16, No. 3.
Jonsson, I.G., 1980. A new Approach to Oscillatory Rough Turbulent Boundary Layers.
Ocean Engineering, Vol. 7, pp. 109-152, Pergamon Press.
Kajiura, K.A., 1968. A Model for the Bottom Boundary Layer irr Water Waves. Bulletin
Earthquake Research, Inst., No. 45, pp. 75-123.
Kamphuis, J.W., 1975. Friction Factor under Oscillatory Waves. Journal of the Waterways,
Harbours and Coastal Eng. Div., Vol. 101, No WW2.
Kana, T.W., 1979. Suspended Sediment in Breaking Waves. Techn. Report No. 18-C.R.D.,
Coastal Res. Div., Dep. of Geology, Univ. of south Carolina, Columbia, USA.
Kaneko, A. and Hoqji, H., 1979. Double Structures of Steady Streaming in the Oscillatory
Viscous Flow over a Wavy Wall. Journal of Fluid Mechanics, Vol. 93, p. 727-736.
Kemp, P.H. and Simons, R.R., 1982. The Interaction between Waves and a Turbulent Cur-
rent: Waves PropagQting with .the Current*Journal of Fluid Mechanics, Vol. 116,
p. 227-250.
Kemp, P.H. and Simons, R.R., 1983. The Interaction of Waves and a Turbulent Current:
Waves Propagating against the Current. Journal of Fluid Mechanics, Vol. 130, p. 73-
89.
Klopman, G., 1989. MAW-program, Delft Hydraulics, Delft, The Netherlands.
Koyama, H. and Iwata, K., 1986. Estl'mution of Water Partr'cle Velocities of Shallow Water
Waves by a Modified Traqfer Function Method. Coastal Engineering Conference,
Taipeh, Taiwan.
Kroon, A. and Van Rijn, L.C., 1993. Suspended Sediment Fluxes in the Nearshore Zone
at Egmond, The Netherlands. Dep. of Phys. Geography, Univ. of Utrecht, The
Netherlands.
Longuet-Higgens, M.S., 1953. Mass Transport in Water Waves. Roy. Soc. Phil. Trans., Vol.
245 A.903, pp. 535-581.
Lundgren, H., 1972. Turbulent Currents in the Presence of Waves. Coastal Eng. Conf., p.
623-634, Vancouver, Canada.
Manohar, M., 1955. Mechanics of Bottom Sediment Movement due to Wave Action. Techn.
Memo. No. 75, Beach Erosion Board, Corps of Eng., USA.
Miller, R.L., 1976. Role of Vortices in Surf Zone Prediction: Sedimentation and Wave
Forces. Symp. Beach and Nearshore Sedimentation, Dallas, USA.
Miche, R., 1944. Mouvements Ondulatoires des Mers en Profondeur Constante ou Decrois-
,
sante. Annales des Ponts et Chausstes.
Myrhaug, D. and Slaattelid, D.H., 1990. A Rational Approach to Wave-Current Friction

Coeficientsfor Rough, Smooth and Transitional Turbulent Flow. Coastal Engineering,
14, p. 265-293.
Nadaoka, A. and Kondoh, T., 1982. Laboratory Measurements of Velocity Fluid Structure
in the Surf Zone. Coastal Engineering in Japan, Vol. 25.
Nap, E. and Van Kampen, A., 1988. Sediment Transport in Irregular Non-Breaking Waves.
Coastal Eng. Dep., Delft Univ. of Technology, Delft, The Netherlands.
Nieuwjaar, M. and Van der Kaaij, Th., 1987. Sediment Transport in Irregular Non-
Breaking Waves. Coastal Eng. Dep., Delft Univ. of Technology, Delft, The
.. Netherlands.
Nikuradse, J., 1933. Stromungsgesetze in Rauhen Rohren. Forschung Arb. Ing. Wesen
No. 361, Gennany.
Nelson, R.C., 1983. WaveHeights in Depth Limited Conditions. Sixth Australian Conference
on Coastal and Ocean Engineering, Gold Coast, Australia.
O ' C o ~ e r ,B.A. and Yoo, D., 1988. Mean Bed Friction of Combined Wave-Current Flow.
Coastal Engineering, 12, p. 1-21.
Rienecker, M.M. and Fenton, J.D., 1981. A Fourier Approximation Method for Steady
Water Waves. Journal of Fluid Mechanics, Vol. 104, pp. 119-137.
Roelvink, D. 1980. Mean Shear Stress in Current Direction due to a Combinutian of Current
and Waves. Internal Note, Delft Hydraulics, Delft, The Netherlands.
Russel, R.C.H. and Osorio, J.D.C., 1957. An Experimental Investigation of Drift Profiles
in a Closed Channel. Proc. 6th Coastal Eng. Conf., Florida, USA.
Sallenger, A.H. and Howd, P.A., 1989. Nearshore Bars and Break-Point Hypothesis.
Coastal Engineering, Vol. 12, p. 301-313.
Sato, S., Mimura, N. and Watanabe, A., 1984. Oscillatory Boundary Layer Flow over
Rippled Beds. Proc. 19th Coastal Eng. Congress, Houston, USA.
Soulsby, R.L., Davies, A.G. and Wilkinson, R.H., 1983. The Detailed Processes of
Sediment Transport by Tidal Currents and by Sulfate Waves. Instr'tute of
Oceanographic Sciences, Report, No. 152, Taunton, England.
Sleath, J.F.A., 1974. Velocities above a Rough Bed in Oscillatory Flow. Journal of
Waterways, Harbours and Coastal Eng. Div. 100, WW4, p. 287-304.
Sleath, J.F.A., 1974. Mass Transport over a Rough Bed. Journal of Marine Research.
Sleath, J.F.A., 1984. Sea Bed Mechanics. Wiley Interscience, New York.
Sleath, J.F.A., 1988. Transition in Oscillatory Flow over Rough Beak. Journal of Waterway,
. Port, Coastal and Ocean Engineering, ASCE, Vol. 114, No. 1.
Smith, J.D., 1977. Modelling of Sediment Transport on Continental Shelves. In Goldberg,

E.D. et al, The Sea, Vol. 6, Wiley Interscience, New York.
Soulsby, R., Davies, A.G., Fredsee, J., Myrhaug, D., Simons, R.R., Temperville, A.,
1991. Intercomparison of Wave-Current Interaction MAST-workshop, Edinburgh,
Scotland.
Stive, M.J. F., 1980. Velocity and Pressure Field of Spilling Breakers. Proc. 1 7th Conf. on
Coastal Eng ., Sydney, Australia.
Stive, M.J.F., 1983. A Model for qffshore Sediment Transport. Delft Hydr. Lab., Note
M1882, Delft, i%e Netherlands.
Swart, H., 1976. Predictive Equations Regarding Coptal Transports. Proc. 15th Conf.
Coastal Eng., Honolulu, Hawaii.
Stokes, G.G., 1847. On the Theory of Oscillatory Waves. Trans. Cambridge Phil. Soc.
Vol. 8, pp. 441-455.
Thornton, E.B. and Guza, R.T., 1982. Energy Saturation and Phase Speeds Measured on
a Natural Beach. Joumal of Geophysical Research, Vol. 87, No. C12, p. 9499-9508.
Tolman, H.L., 1992. An Evaluation of Expressions for Wave Energy Dissipation due to
Bottom Friction in the Presence of Currents. Coastal Engineering, 16, p. 165-179.
Van der Stel, J. A and Visser, P.J., 1985. Model Investigation of the Eflect of Pe~endicular
Waves on the Longshore Current (in Dutch). Coastal Eng. Dep., Civil. Eng., Delft
Univ. of Technology, Delft, m e Netherlands.
Van Heteren, J. and Stive, M.J.F., 1985. Kinematics and Directionality of Waves in the
Suf Zone. Note WKKZ-84.SO13, Rijhaterstaat, Hellevoetsluis, The Netherlands.
Van Rijn, L.C., 1984. Sediment Transport, Part II: Suspended Load Transport. J o u m l of
Hydraulic Engineering, ASCE, No. HYll .
Van Rijn, L.C., 1989. Handbook of Sediment Transport by Currents and Waves. Report
H461, Delft Hydraulics, Delft, The Netherlands.
Visser, P.J., 1986. Wave Basin Experiments on Bottom Friction due to Currents and Waves.
Coastal Eng. Conference, Taipeh, Taiwan.
Weggel, J.R., 1972. Maximum Breaker Height. Journal of Waterways, Harbours and Coastal
Eng. Div., ASCE, Vol. 98, No. WW4.
Yoo, D., 1989. Explicit Modelling of Bottom Friction in Combined Wave-Current Flow.
Coastal Engineering, 13, p. 325-340.
3. FLUID AND SEDIMENT PROPERTIES
3.1 Fluid properties
3.1.1 Introduction
All real fluids have certain measurable characteristics or properties such as density, viscosity,
compressibility, capillarity, surface tension etc. Some properties are combinations of other
properties. For example, kinematic viscosity involves dynamic viscosity and density.
Herein, the two most basic properties being the density and 'the viscosity are given.
3.1.2 Fluid density
The density of fresh water varies with temperature, as follows:
The density of sea water (= 1025 kg/m3) can be determined from the following expression:
p = lo00 + 1.455 CL - 0.0065 (Te - 4 + 0.4 (3.1.1)

in which:
CL = Chlorinity (in %o)
Te = temperature (in "C)
p = fluid density (in kg/m3)
The Chlorinity follows from:
in which:
S = Salinity = total quantity of dissolved salt in grammes per kilogramme of sea water
(in %o, weight ratio).
3.1.3 Fluid viscosity
The kinematic viscosity coefficient v is defined as:
in which:
q = dynamic viscosity coefficient (Ns/m2)
The kinematic viscosity coefficient can be approximated by:
The dynamic viscosity coefficient is influenced by the sediment particles. For dilute
suspensions (c < 0.1) Einstein (1906) found: .
in which:
q, = dynamic viscosity coefficient of fluid-sediment mixture
q = dynamic viscosity coefficient of clear-water
c = volumetric sediment concentration
Based on experiments with volume concentrations in the range of c = 0.1 to 0.6, Bagnold
(1954) found:
in which:
A = [(0.74/~)'/~
-I]-' = dimensionless concentration parameter
Equation (3.1.6) is shown in Fig. 3.1.2.
An empirical relationship was given by Do Ik Lee (1969), which reads as:
with a = -(2.5 + 1.9 c + 7.7 c2).
An excellent review related to the effect of the sediment particles on viscosity was presented
by Savage (1984).
The kinematic viscosity coefficient in a fluid-sediment mixture is given by:
in which:
q, = dynamic viscosity coefficient in fluid-sediment mixture
pm = fluid-sediment mixture density = p (1 + (s - 1) c)
c = sediment concentration (volume)
s = p,/p = specificdensity
3.2 Sediment Properties
3.2.1 Introduction
Sediment is fragmental material, primarily formed by the physical and chemical desintegration
of rocks from the earth's crust. Such particles range in size from large boulders to colloidal
size fragments and vary in shape from rounded to angular. They also vary in specific gravity
and mineral composition, the predominant materials-being quartz mineral and clay minerals
(kaolinite, illite, montmorillonite and chlorite). The latter hqve a sheet-like structure, which
can easily change (flocculation) under the influence of ele~t~ostatic forces (cohesive forces)
in a saline environment. Consequently, there is a fundamental difference in sedimentary
behaviour between sand and clay materials.
Sediments can be classified according to their genetic origin:
Lithogeneous sediments, which are detrital products of disintegration of pre-existing rocks
Biogeneous sediments, which are remains of organisms mainly carbonate, opal and calcium
phosphate
Hydrogeneous sediments, which are precipates from seawater or from interstitial water
Descriptive sediment classifications can also be used and are related to characteristics like
color, texture, grain size, organic content, etc. For example, a mixture of sand and clay is
classified as a sandy clay when the percentage of sand is between 25 % and 50%. Similarly,
clayey sands, gravelly sands, sandy gravels, clayey gravels, and gravelly clays are
distinguished.
Sediment particles larger than 62 pm and smaller than 2000 pm (see Section 3.2.4) are
usually refered to as sand particles.
Based on mineral and chemical composition, three types of sands can be distinguished:
silicate sands;
carbonate sands;
gypsum sands.
Silicate sands mainly consist of quartz and feldspar minerals, which are extremely insoluble
in water.
Carbonate sands consist of calcite and aragonite, which are two different crystalline forms
of calcium-carbonite (CaCO,), originating from shell and coral fragments (coral sands). The
percentage of carbonate in a sample usually is larger than about 80 % . Carbonate sands are
much more soluble in fresh water than silicate sands. In sea water (especially in the tropics)
which is already supersaturated with carbonate it is hardly soluble. Carbonate sands usually
exhibit some degree of cementation: weakly-cemented or well-cemented, which means that
the fragments cannot be manually broken.
Gypsym sands consist of crystal forms of gypsum (CaSO4.2H,O), which is a moderately
soluble mineral that can survive only in arid regions.
The sediment properties herein presented are: density, porosity, shape, size, fall velocity and
angle of repose.
1
t :TEMPERATURE ( .C )
~,OIU~
'
1 -.
,
0.5 J Q ~
10 15 20 25 30 35
-- TEMP (t)
Figure 3.1.1 Viscosity comcient
1 o0
10-3 2 4 6810-2 2 3 4 6810-1 2 3 4 6810°
---+volume concentration, c
Figure 3.1.2 Viscosity coejjicient as a function of concentration according to

(1954)
3.2.2 Density and porosity
The density of quartz and clay minerals is approximately equal to p, = 2650 kg/m3. The
density of carbonate material may be somewhat smaller (p, = 2500 to 2650 kg/m3). The
specific gravity is defined as the ratio of the sediment density and the fluid density, s = p,lp
= 2.65.
The dry sediment density is the dry sediment weight per unit volume (= concentration) and
is equal to:
in which:
p,, = dry sediment density (kg/m3);
p, = sediment density (kg/m3);
p = porosity factor.
The wet density or vohrme weight of deposited material (assuming total saturation) is the
weight of water and sediment per unit volume (sometimes called the wet bulk density) and
is equal to:
The porosity of sediment material is often related to the deposition history of the sediment
bed. Loose packing occurs when sediments settle from suspension in still water. Basically,
four packing arrangements are possible for spherical particles. The most unstable arrangement
is the cubic arrangement with the sphere centres forming a cube yielding a porosity of 48 %.
The Rhombohedral arrangement with the spheres in the hollows of each other yields the most
stable packing and the smallest porosity of 26 % . Random packing of spheres yields porosity
ranges from 36 % to 40 % . Natural sediments with particles of various sizes have relatively
small porosity values because the smaller particles can occupy the large void spaces. A pourly
sorted (many sizes) coarse sand has a porosity of about 40%. A well sorted (almost uniform)
fine sand has a porosity of about 45 %. The porosity of coral sand (mixture of coral and shell
fragments) has been found in the range from 0.5 to 0.65 (Van der Meulen, 1988).
Deposits consisting of clay, silt, sand and organic material are called mud deposits and can
have a large porosity factor (upto 80 %).
The dry sediment weight of mixtures can be estimated from empirical relationships. Based
on the analysis of samples from the toplayers (recent deposits) of reservoirs, Lane and
Koelzer (1953) found:
in which:
p, = fraction of sand particles (d > 50 pm).
Lara and Pemberton (1963) analyzed reservoir samples always submerged (A) and submerged
50 % of the time (B). They found:
in which:
p, = dry sediment density (kg/m3)
pd = fraction of sand particles (d 1 62.5 pm)
p, = fractionofsiltparticles (4 < d < 62.5 pm)
p,,, = fraction of clay particles (d I 4 pm)
Murthy and Banerjee (1976) analyzed toplayer samples (sampling depth = 0.4 m) from
Indian Reservoirs, which were about 50% of the time empty (exposed to the sun) and found:
in which:
p,, = fraction of sand particles (20 pm 5 d I 200 pm)
pa, = fraction of silt particles (2 pm < d < 20 pm)
pchy = fraction of clay particles (d I2 pm)
Equations (3.2.3) to (3.2.6) express values representing the initial (0 to 1 year) dry weight
of the deposits. The dry weight increases with time due to consolidation.
Based on analysis results of a large amount of soil samples, Allersma (1988) proposed:
in which:
a = consolidation coefficient ( a in the range from 0 to 2.4 with a mean value of a = 1.2).

The presence of organic material such as Cellulose, Lignin, Pectin, Coal (with densities in
the range of 1200 to 1500 kg/m3) can give a substantial reduction of the density of the
mixture.
The wet sediment weight ( p ,,J or wet bulk density (assuming total saturation) can be
expressed as:
Other soil parameters are (see Fig. 3.2.2):
sediment concentration (c or pe) = ratio of dry solid mass (MJ and total volume Or) =
MJV,
solids content (sc) = raiio of the mass of dry solids (Ms) and the total
mass (MJ = Ma/Mt
water content (wc) = ratio of the water mass (M,) and the solid mass (MJ
= MJM,
void ratio (e) = ratio of the void volume (V,) and the solid volume
= V,/V,
sturation degree (sd) = ratio of water volume (V,) and void volume (V,) =
VWN"
Relations are:
porosity @)
solid content (sc) - ( P , / P ) P,

Ps + [(PSIP) - I] P a n
void ratio (e)
water content (wc)
3.2.3 Shape
Most of the sand particles on the face of the Earth are more or less rounded because their
edges and corners are smoothed by abrasion as running water or wind moves the sand
particles from their origin (source) to their final resting place. Roundness is a function of
abrasion induced by transport and it increases slowly with distance. Thousands of kilometers
of transport in a river are required to achieve even moderate rounding. Beaches where sand
moves in and out with each wave are ideal places for rounding of sand particles if they stay
there for any length of time.
sand (%) silt (%)
------+ pd, (1000kglm3)
Figure 3.2.1 Sediment densities according to Allersma (1988)

Figure 3.2.2 Conversion values between dry density, wet bulk density, moisture content and
voids ratio (ps = 2650 kg/m3)
The shape of particles generally is represented by the Corey shape factor, defined as:
in which:
a = length along longest axis perpendicular to other two axes
b = length along intermediate axis perpendicular to other two axes
c = length along short axis perpendicular to other two axes.
The SF-factor for natural sand is approximately 0.7.

The shape factor is essentially a flatness ratio and does not take into account the distribution
of the surface area and the volume of the particle. For example, a cube of a given length and
a sphere of a diameter equal to the length of the cube have the same shape factor (SF = 1).
To .overcome this, another shape factor is also applied, defined as:
SF, =
ds
SF -
dn
in which:
SF, = shape factor accoding to Eq. (3.2.10)
d, = diameter of a sphere having the same surface area as that of the particle
d,, = diameter of a sphere having the same volume as that of the particle
SF, approaching unity implies increasing sphericity of the particles (sphericity is ratio of
surface area of a sphere and surface area of the particle at equal volume). A behavioural
measure of shape is expressed by the rollability parameter (Winkelmolen, 1971). The
rollability is a functional shape property measured by the time it takes for grains of equal size
and density to travel the length of a cylinder revolving with its axis inclined at an angle of
2.5 " to the horizontal.
Grain fabric is the attitude in space and degree of preferred orientation displayed by the
grains during sedimentation. Shape fabric has a physical cause and is therefore an indicator
of hydraulic conditions.
3.2.4 Size
Usually, sediments are refered to as gravel, sand, silt or clay. These terms refer to the size
of the sediment particle. The following Table 3.1 presents the grain size scale of the
American Geophysical Union. This scale .is based on powers of 2 mm, which yields a linear
logarithmic scale via the phi-parameter defined as 4 = - 210g d (with d in mm).
Various methods are available to determine the particle size. Cobbles can be measured
directly with a ruler. Gravel, sand and silt are analyzed by wet or dry sieving methods
yielding sieve diameters. Clay materials are analyzed hydraulically by using settling methods
(Van Rijn, 1986) yielding the particle fall velocity from which the standard fall diameter is
computed. Clay materials can also be analyzed with various electronic techniques such as the
Coulter counter and the Laser Diffraction technique (Van Rijn, 1986). Thus, the size of a
sediment particle is closely related to the analysis method.
Typical "diameters" are:
sieve diameter which is the diameter of a sphere equal to the length of the side
of a square sieve opening through which the given particle will
just pass,
nominal diameter which is the diameter of a sphere that has the same volume as the
particle,
standard fall diameter which is the diameter of a sphere that has a specific gravity of
2.65 and has the same fall velocity as the particle in still,
distilled water of 24" C.
A natural sample of sediment particles contains particles of a range of sizes. The size
distribution of such a sample is the distribution of sediment material by percentages of
weight, usually presented as a cumulative frequency distribution (see Fig. 3.2.3).
The frequency distribution is characterized by:
median particle size d, which is the size at which 50% by weight is finer,
mean particle size d,,, = C (pi d,)/100 with pi = percentage by weight of each grain
size fraction 4 ,
standard deviation +
ad = C pi(&-Q2/100or ad = 0.5(ds,/dI6 dM/dm)which is a
measure based on graphic values.
Often the phi-scale is used for size distribution representation:
where d is the particle diameter in millimeters.

Class Name Millimeters Micrometers Phi Values
B~uldu~ > 256 < -8
Cobbles 256 - 64 -8 to -6
Gravel 64 - 2 -6 to -1
Very coarse sand 2.0 - 1.0 2000 - 1000 -1 to 0
Coarse sand 1.0 - 0.50 1000 - 500 0 to +1
Medium sand 0.50 - 0.25 500 - 250 i - 1 tn +2
F i e sand 0.25 - 0.125 250 - 125 +2 to +3
Very fine sand 0.125 - 0.062 125 - 62 +3 to +4
Coarse silt 0.062 - 0.031 62 - 31 +4 to +5
Medium silt 0.031 - 0.016 31 - 16 +5 to +6
Fine silt 0.016 - 0.008 16 - 8 +6 to +7
Very fine silt 0.008 - 0.004 8 -4 +7 to +8
Coarse clay 0.004 - 0.0020 4 -2 +8 to +9

Medium day 0.0020 - 0.0010 2 - 1 +9 to +10
Fine clay 0.0010 - 0.0050 1 - 0.5 +10 to +11
Very fine clay 0.0005 - 0.00024 0.5 - 0.25 +11 to +12
Colloids < 0.0024 < 0.24 > +12
Table 3.1 Grain size scale American Geophysical Union
Characteristic values are the mean phi-value ,@, the median-value, @, the standard
deviation (sorting) 04, the skewness (asymmetry) and kurtosis (peakedness):
standard deviation : a+ = 0.5(@@- @,J
skewness
kurtosis
Ideally, a sample has a normal (Gaussian) distribution in terms of the @-scale(log-normal

distribution), yielding:
with:
a, = O.S(dJd,, + d,/d,)
Figure 3.2.3 shows particle size curve for a log-normal distribution on a logarithmic-linear
scale and on a logarithmic-probability scale.
Grain size can be used as an indicator of energy conditions. Fine grains usually are dominant
in low-energy conditions near river banks, on tidal.flats, in protected or sheltered basins.
Coarse grains are found in high-energy conditions near breaker bars along the coasts and in
the deeper channels of rivers and estuaries, where the finer grains cannot easily survive
(strong currents). The main grain size decreases wite distance-from the source known as:
"fining down the transport path", due to abrasion effects and deposition of fines in quiescent
conditions. Near the source the size range usually is relatively wide (well sorted); while a
narrow size range (poorly sorted) is found far away from the source. The coarsest particles
are transported as bed load; the fmest particles as suspended load.
The relationship between grain size, sorting and hydraulic conditions has been used to
distinguish between different sedimentary environments like wave-dominated, current-
dominated; eroding or depositing and paths between them.
3.2.5 Particle fall velocity
1. Sphere f U n g in a stiU fluid
Basically, the fall velocity is a behavioural property. The terminal fall velocity (wJ of a
sphere is the fall velocity when the fluid drag force = 1/2(% p w3 1/4(rr d2) on the
particle is in equilibrium with the gravity force = 1/6(p, - p)gd3, giving:
in which:
w, = terminal fall velocity of a sphere in a still fluid
d = sphere diameter
s = specific gravity (= 2.65)
C, = drag coefficient
The drag coefficient C, is a function of the Reynolds number Re = wsd/v and shape factor
(see Fig. 3.2.4). In the Stokes region (Re < 1) the drag coefficient is given by: C, = 24/Re,
yielding:
Outside the Stokes region there is no simple expression for the drag coefficient. The C,-value
decreases rapid1 outside the Stokes region (Re < 1) and becomes nearly constant for
lo3 < Re < Id, yielding w, proportional to d0'.
The effect of temperature on the fall velocity is taken into account by the kinematic viscosity
coefficient v . The largest effect occurs for the smallest sphere diameters.
Figure 3.2.3
- 2 4 6 8 103
diameter (+m)
4 6 8 103
diometer (w)
Particle size curve.far a log-normal distribution on-logarithmic-linear scale

and on logarithmic-probability scale
Figure 3.2.4 Drag comcient as a function of Reynolds number for differentshape factors
(Albertson, 1953)
2. Non spherical particles
The expressions valid for a sphere cannot be applied for a natural sediment particle because
of the differences in shape. The shape effect is largest for relatively large particles (>
300 pm) which deviate more from a sphere than a small particle.
Experiments show differences in fall velocity of the order to 30% for SF in the range from
0.5 to 1.
The terminal fall velocity of non-spherical sediment particles can be determined from the
following formulae:
d v
r' -11 for 100 < d < 1000 pm (3.2.22)
for d 2 1000 pm (3.2.23)

in which:
d = sieve diameter
s = specific gravity (= 2.65)
Figure 3.2.5 shows fall velocities according to the Stokes Equation (3.2.21). Figure 3.2.6
shows fall velocities as given by the U.S. Inter-Agency Committee on Water Resources
(1957) and is largely based on experimental results. Equation (3.2.22) is also shown.
The fall velocity of coral sand (particles larger than 300 pm) may be considerably smaller
than that of quartz sand (Van der Meulen, 1988). Figure 3.2.8 shows fall velocities for both
materials at a temperature of 20°C. The differences are mainly caused by differences in
shape. Coral sand particles are more angular and have, therefore, a smaller fall velocity. The
density of coral sand may also be somewhat smaller (= 2500 to 2650 kg/m3).
3. Effect of sediment concentration
The fall velocity of a single particle is modified by the presence of other particles. A small
cloud of particles in a clear fluid will have a fall velocity which is larger than that of a single
particle. Experiments with uniform suspensions of sediment and fluid have shown that the fall
velocity is strongly reduced with respect to that of a single particle, when the sediment
concentration is large. This effect, known as hindered settling, is largely caused by the fluid
return flow induced by the settling velocities. A state of fluidization may occur when the
vertical upward fluid flow is so strong that the upward drag forces on the particles become
equal to the downward gravity forces resulting in no net vertical movement of the particles.
According to Richardson and Zaki (1954), the fall velocity in a fluid-sediment suspension can
be determined as:
w,, = particle fall velocity in a suspension

w, = particle fall velocity in a clear fluid
c = volumetric sediment concentration (-)
y = coefficient (-)
Figure 3.2.5 Fall velocities for particle sizes smaller than 100 pm according to Stokes
particle fall velocity, w, (m/s)
Figure 3.2.6 Fall velocity for particles sizes larger than 100 pm according to US Inter-
Agency Committee (195 7J
Figure 3.2.7 Influence of sediment concentrm'on on fall velocity (Re small)
- diameter (mm)
Figure 3.2.8 Fall velocity of coral sand

The y - coefficient varies from 4.6 to 2.3 for Re = w, dl v increasing from 10' to lo3. For
particles in the range of 50 to 500 pm under normal flow conditions the y -coefficient is
about y = 4.
Figure 3.2;7 shows Equation (3.2.24) and the experimental results of Oliver (1961) and
McNown-Lin (1952). As can be observed, Eq. (3.2.24) yields w,,,-values which are 20% to
30 % too large for small concentrations.
The formula of Oliver which reads as:
yields good results over the full range of concentrations (see Fig. 3.2.7).
4. Influence of osciUatory flow
The fall velocity of a single sediment particle in flowing water is generally assumed to be
equal to its terminal fall velocity (wJ in still water. Various researchers have investigated the
possible reduction of the terminal fall velocity in case of an oscillatory flow. A review and
new information have been presented by Hwang (1985) and Nielsen (1979, 1984). These
studies show that the major mechanism governing the fall velocity reduction in an oscillating
flow is the drag non-linearity effect. According to Hwang, the fall velocity reduction can be
expressed as:
in which:
V, = effective fall velocity
Vf,max
-
- peak value of fluid velocity
w, = terminal fall velocity in still water
d = particle diameter
Equation (3.2.26) is presented in graphical form by Hwang (1985). The ratio v,/w, decreases
for increasing values of w,d/v and vf-/w,. The results of Hwang suggest a rather large
reduction of the fall velocity, (vs/ws = 0.5 for vf /ws = 10). The results of Hwang are
somewhat suggestive, because the influence one oscillation period is not shown.
Information of the influence of the oscillation period can be obtained from the results of Ho
(1964) given in terms of (see Hwang, 1985):
in w.hich:
o = 2x/T = angular oscillation frequency.
Taking a 100 pm sediment particle, an oscillation period of T = 10 s, and vf,, = 0.1 mls,
it follows that v,/w, = 1.
Similar values were also found by Nielsen (1984), who concluded that the reduction of the
terminal fall velocity is negligible in a pure oscillating motion (7' 2 1 s).
Turbulence is a special type of (random) oscillating motion dominated by high frequencies.
It has been shown by Murray (1970) that the particle fall velocity can be considerably reduced
by isotropic turbulence effects due to drag non-linearities. Another mechanism may be
intensive eddy production close to the bed inducing vertically upward motions which may
reduce the fall velocity until the eddies dissolve at higher levels.
Jobson and Sayre (1970) reviewing all available information conclude that the turbulent
motions may slightly increase the particle fall velocity.
Ludwick and Domurat (1982) have simulated the movements of 100 pm and 200pm
sediment particles in a turbulent velocity field and found that the settling of fine sand is not
significantly reduced when the vertical turbulent fluid velocities have a symmetric
distribution. The basic question is what type of vertical velocity distribution is present in
natural conditions? Analysis of (turbulent) velocity measurements near the bottom where most
of the sediment is transported, shows the presence of bursting processes characterized by lift-
up of low-momentum fluid (bursts) and a down-rush of high-momentum fluid to the bed
(sweeps). This indicates an asymmetric fluid motion in vertical direction with relatively high
(short duration) downward velocities,- which may result in a slight increase of the fall
velocity. This is in agreement with the findings of Jobson and Sayre (1970).
3.2.6 Angle of (natural) repose
The angle of (natural) repose is a behavioural property of sand particles.

Grains piled up on each other have an equilibrium slope which is called the angle of natural
repose (4,).
This parameter appears to be a function of size, shape and porosity. The angle increases with
decreasing roundness. Values from the literature are in the range of 4, = 30" to 40" for
sand sizes from 0.001 to 0.01 m. Observations in nature on the avalanche lee slope of desert
dunes and river bed dunes also show values in the range of 30" to 40".
The angle of repose (4) also refered to as the angle of internal friction is a characteristic
angle related to the particle stability on a horizontal or sloping bed (see Figs. 4.1.1 and
4.1.9). The angle of repose ( 4 ) may differ from the angle of natural repose (4,). Usually,
the angle of repose is determined from initiation of motion experiments for horizontal and
sloping beds (see section 4.1). The critical bed-shear stress for a particle on a sloping bed can
be expressed, as:
,
b = critical bed-shear stress on a sloping bottom
critical bed-shear stress on a horizontal bottom
.)a,o = angle of repose
p = angle of (longitudinal) bottom slope
For a given P -value the r,, and the rb, -values can be determined from the measured
variables. Equation (3.2.28) can then be isid to determine the 4-value.
Table 3.2 shows 4 -values reported in the literature.
The @-values of carbonate sands were determined from results of triaxial tests (direct shear
testing) .
The 4 -values of the silicate sands determined from Eq. (3.2.28) using results of initiation of
motion experiments express the angle of repose of the particles of the top layer of the bed
surface. This value should approximately be equal to the angle of natural repose ($3. The
values observed (see Table 3.2) are in the range of 40" to 50" and are much larger than the
observed angles of natural repose (30" to 40"). The reason for this is not fully clear; most
probably it is related to the validity of Eq. (3.2.28). The degree of packing of the entire bed
layer should not have much influence on the angle of repose of the top layer. Most probably
the top layer will always have a loose packing.
Sourct Type of sediment Si & ( pm) Shape Packing Angle of repose 4
White (1940) silicate sand. 900 45

5600 45
Lysne (1969) silicate sand -
1000 4000 rounded loose -
38 50
Lugue (1974) silicate sand 900 rounded deme -
40 45
1800 rounded loose 40 - 45
Van der Wal (1991) silicate sand 210 rounded lqose 45 - 55
Delft Hydr. (1984) silicate stone 36000 angular loose 41 - 47
Hudson (1988) carbonate sand -
450 670 angular loosede& -
38 50
AireY (1988) carbonate sand 300 - 500 angular loosedense -
38 50
Hull (1988) carbonate sand 200 - 1000 angular loosedense 38 - 50
silicate sand 200 - 1000 rounded loose -
38 41
Table 3.2 Angle of repose for silicate and carbonate sands
For the design of stable channels the following conservative values are recommended (see
Table 3.3).
Size d, Angle of npose (4)
Rounded Angular
(m)
S 0.001 30' 35O
0.005 32" 37O
0.01 35O 40°
0.05 37O 42'
2 0.1 40° 45 O
Table 3.3 Angle of repose for stable channel design (silicate muterial)
Airey, D.W. and Randolph, N.F., 1988. The Strength and Sti@tess of Two Calcareous
Sam&. Engineering for Calcareous Sediments, Balkema, Rotterdam.
Albertson, M.L., 1953. Eflect of Shape on the Fall Velocity of Gravel Particles. Proc. 5th
Hydr. Conf., Univ., of Iowa, Bull. no. 34, USA.
Allersma, E., 1988. Composition and Density of Sediments (in Dutch). Report 271.03, Delft
Hydraulics, Delft, The Netherlands.
Bagnold, R.A., 1954. Fxperiments on a Gravity-Free Dispersion of Large Solid Spheres in

a Newtonian Fluid under Shear. Proc. Royal Society of London, Vol. 225A, London,
England.
Einstein, A., 1906. Eine neue Bestimng der Molekiil dimensionen. Annalen der Physiek,
Leipzig (4),19.
Delft Hydraulics, 1984. Stability of Rubble Mound Breakwaters. Report S467, Vol. 111, Delft
Hydraulics, Delfr, The Netherlands.
Do Ik Lee, 1969. The Viscosity of Concentrated Suspensions. Trans. of the Soc. of Rheology,
13 n ~ 2,
. pp. 273-288.
Hudson, M.J. et al, 1988. Properties of near Sulfate Bass Strait Soils. Engineering for
Calcareous Sediments, Balkema, Rotterdam.
Ho, H.W., 1964. Fall Velocity of a Sphere in a Field of Oscillating Fluid. Thesis, State
Univ. of IOWA, USA.
Hull, T.S., Poulos, H.G., Alehossein, Ha, 1988. The Static Behaviour of Various
Calcareous Sediments. Engineering for Calcareous Sediments, Balkem, Rotterdam.
Hwang, P.A., 1985. Fall Velocity of Particles in Oscillating Flow. Journal of Hydraulic
Jobson, H.E. and Sayre, W.W., 1970. Vertical Trawer in Open Channels. Journal of the
Hydraulics Division, ASCE, Vol. 96, HY3.
Ludwick, J.C. and Domurat, G.W., 1982. A Deterministic Model of the Vertcal Compo-
nent of Sediment Motion in a Turbulent Fluid. Marine Geology, Vol. 45, p. 1-15.
Luque, R.F., 1974. Erosion of Bed Load Transport. Thesis, Delft Univ. of Technology,
Lysne, D.K., 1969. Movement of Sand in Tunnels. Journal of the Hydraulics Division,
ASCE, No. HY 6.
McNown, J.S. and Lin, P.N., 1952. Sediment Concentration and Fall Velocity. Proc.
Second Mid Western Conf., Fluid Mechanics, Ohio State Univ., USA, p. 401-411.
Murray, S.P., 1970. Settling Velocities and Vertical Dzmion of Particles in Turbulent
Water. Journal of Geophysical Research, Vol. 75, No. 9, pp. 1647-1654.
Murthy, B.N. and Banerjee, B.K., 1976. Initial Unit Weight of Deposited Sediment in
ReserVoirs with Considerable Drawdown. Diamond Jubilee Symp., Vol. 1, paper A6,
Poona, India.
Lane, E.W. and Koelzer, V.A., 1953. Density of Sediments Deposited in Reservoirs. Report
No. 9, St. Paul Eng. District, St. Paul, Minnesuta, USA.
Lara, J.M. and Pemberton, E.L., 1963. Initial Unit Weight of Deposited Sediments. Proc.
Federal Interagency Sedirnentation Conf., Publ. No. 970, p. 818-845, US Dep. of
Agriculture, USA.
Nielsen, P., 1984. On the Motion of Suspended Sand Particles. Journal of Geophysical
Research, Vol. 89, No. Cl.
Oliver, D.R., 1961. The Sedirnentation Suspension of Closely-Sized Spherical Particles.

Chem. Eng. Science, Vol. 15, p. 230-242.
Richardson, Y .F. and Zaki, W.N., 1954. Sedimentation and Fluidization, Part I. Trans.
Inst. Chem. Eng., Vol. 32, p. 35-53.
U.S. Inter-Agency Committee on water resources, 1957. Some Fundamentals of Particle
Size Analysis. St. Anthony Falls Hydraulic Lab., Minneapolis, Minnesota.
Van der Meulen, A., 1988. Fall Velocity of Coral Sand (in Dutch). Report Geopro 1988.01,
Dep. of Physical Geography, University of Utrecht, Utrecht, 77ze Netherlands.
Van der Wal, M., 1991. Determination of Stability Parameter of Course Sand (in Dutch).
Report H869, Delft Hydraulics, Delft, lhe Netherlands
Van Rijn, L.C., 1986. Manual Sediment Transport Measurements. Report S304, Delft
Hydraulics, The Netherlands.
Savage, S.B., 1984. The Mechanics of Rapid Granular Flows, Advances in Applied
Mechanics, Vol. 24, pp. 288-366.
White, C.M., 1940. The Equilibrium of Grains on the Bed of a Stream. Proc. Royal Society
of london, Series A, No. 958, Vol. 174, pp. 322-338.
Winkeholen, A.M., 1971. Rollability, A Functional Shape Property of Sand Grains.
Journal Sedimentary Petrology, Vol. 41, pp. 701- 714.
4. INITIATION OF MOTION AND SUSPENSION
4.1 Initiation of motion in currents
4.1.1 Introduction
Particle movement will occur when the instantaneous fluid force on a particle is just larger
than the instantaneous resisting force related to the submerged particle weight and the friction
coefficient. Cohesive forces are important when the bed consists of appreciable amounts of
clay and silt particles.
The driving forces are strongly related to the local near-bed velocities. In turbulent flow
conditions the velocities are fluctuating in space and time which make together with the
randomness of both particle size, shape and position that initiation of motion is not merely
a deterministic phenomenon but a stochastic process as well.
Early work on the initiation of motion was done by Brahms (1753), who proposed a sixth
power relationship between the flow velocity and the required weight of a stone to be stable
and by Dubuat (1786) who introduced the concept of critical bed-shear stress.
4.1.2 Critical bed-shear stress
The fluid forces acting on a sediment particle resting on a horizontal bed (see Fig. 4.1.1)
consist of skin friction forces and pressure forces. The skin friction force acts on the surface
of the particles by viscous shear. The pressure force consisting of a drag (FD)and a lift force
(FJ is generated by pressure differences along the surface of the particle.
The forces acting on a sediment particle resting on a horizontal bed surface are depicted in
Fig. 4.1.1.
The angle of repose ( 4 ) is defined as the angle between the line through the particle centre
and the point of contact and the line through the particle centre normal to the bed surface.
Particle movement will occur when the moments of the instantaneous fluid forces FDand FL
with respect to the point of contact are just larger than the stabilizing moment of the
submerged particle weight G, yielding (see Fig. 4.1.1):
For reasons of simplicity, the ratio of the lift force and the submerged particle weight is
assumed to be relatively small (a 1). Physically, this is not necessarily true. The lift force
depends, however, on the same variables as the drag force and therefore the effect of the lift
force is automatically taken into account by the empirical coefficients which will be
introduced.
This yields:
/
/ a1 = (bl +b2) cos 4J
a2= b2 sin 4J
03= b3 sin O
9 = angle of repose
I
Figure 4.I . 1 Forces on' a sediment particle (horizontal bed)
------+ Shields parameter, O

Figure 4.1.2 Transport rates at low bed-shear stresses
The value a , = b,/(b, + b,) depends on the local Reynolds number. At a high Reynolds
number the pressure force will be much larger than the (viscous) skin friction force and the
resulting fluid force FDwill act through the centre of the particle (b, = 0 and thus a, = I),
(White, 1940). At a low Reynolds number the viscous friction force at the top of the particle
will dominate yielding b, > 0 and thus a , < 1.
Equation (4.1.4) was first derived by White (1940) based on the equilibrium of the horizontal
forces. The parameter G tan4 can be seen as a stabilizing friction force with p = tan@ as
the friction coefficient.
The effect of both the (viscous) skin friction force and the pressure force, usually, is
expressed as:
in which:
FD = drag force
CD = drag coefficient
p = fluid density
u, = fluid velocity at particle centre
The drag coefficient (C,) is known to be a function of the local Reynolds number (hdlv).
The fluid velocity (u3 at the centre of the particle can be expressed as:
The a ,-coefficient depends on the local Reynolds number when the flow regime is hydrau-
lically smooth; u, = ( r d p)O.' = bed-shear velocity.
The fluid force can now be expressed as:
The a rcoefficient (= a: n CJ8) is a coefficient depending on the local Reynolds number.
The submerged particle weight G can be expressed as:
The a ,-coefficient (= x/6) depends on the shape of the particle.
Substitution of Eqs. (4.1.7) and (4.1.8) into Eq. (4.1.4) yields:

in which:
2
u* - b' = mobility (Shields) parameter
8 =
(s-l)gd (ps-p)gd
4 a,
8, = as tan+ = = critical Shields parameter
3 a: C,
s = P$P = relative density
The 8,-factor depends on the hydraulic conditions near the bed, the particle shape and the
particle position relative to the other particles. The hydraulic conditions near the bed can be
expressed by the Reynolds number Re, = u,dlv. Thus 8, = f(Re,).
Many experiments have been performed to detemine the 8, as a function of Re,. The
experimental results of Shields (1936) related to a flat bed surface are most widely used to
represent the critical conditions for initiation of motion. Herein, the median particle diameter
(d,,) is used as the characteristic particle diameter in the case of non-uniform sediment
material.
Figure 4.1.3 shows the Shields curve in terms of 8 and Re,. Shields (1936) measured
transport rates at bed-shear stresses just larger than the critical bed-shear stress ((rJ). As
his definition of the critical bed-shear stress for initiation of motion ((t,)) Shields used the
value of the bed-shear stress at which the extrapolated transport rate was zero (Fig. 4.1.2).
Details of the extrapolation method were not given by Shields (1936).
The experimental range of Shields roughly is:
8, 2 0.035 for Re, 5 5, hydraulic smooth regime
0.03 I 8, I 0.04 for 5 I Re, 1 7 0 , transitional regime
0.04 < 8, r 0.06 for Re, 1 70, hydraulic rough regime
A Reynolds number smaller than about 5 means that the particle diameter is smaller than the
thickness of the viscous sublayer (6, 2 d). The 8,-value for laminar flow (Re, < 1) was
found to be equal to 0.2 (White, 1940).
The value of 8, from Eq. (4.1.10) can be estimated for Re, in the range of 10 to 100. Assu-
ming a , = 0.5 to 1, a , = 5 to 10, C, = 0.5 and +
= 30°, it follows that 8, = 0.01
to 0.06, which is in reasonable agreement with the Shields curve (see also Fig. 4.1.3).
The Shields curve in terms of 8 and Re, is not practical because the r ,,-value can only be
obtained by iteration.
Bonnefille (1963) and Yalin (1972) showed that the Shields curve can be expressed in terms
of the dimensionless mobility parameter 8 and the dimensionless particle diameter D, (see
Figure 4.1.4).
Applying these parameters, the Shields curve can be represented as:
8, = 0.24 D*-' for 1 < D, I 4
8, = 0.14 D : ~ for 4 < D, I 10
8, = 0.04 D?' for 10 < D, I 20
8, = 0.013 DO." for 20 < D, 5 150
8, = 0.055 for D, >I50
in which:
0, = r,,J[(p,-p)gQ = critical Shields parameter
D* = [(s - l ) g / ~ ~ ] ~= ~particle
d ~ parameter
= time-averaged critical bed-shear stress.
Figure 4.1.5 shows the critical bed-shear stress (t,) acqrding to Shields as a function of
the particle diameter (d) for a temperature of T = 100, 20'" and 30°C, ( p , = 2650 kg/m3,
p = 1000 kg/m3). The influence of temperature is only significant for a particle diameter
smaller than 600 pm. For a particle diameter larger than 7000 pm (7 mm) the r ,,-value is
independent of temperature (viscosity).
no motion
Figure 4.1.3 Initiation of motion for a current over a plane bed, 0 = f (Re*)Shields (1936)
- particle parameter, D,
- particle diamater, d&lO

-3
(m) at T,=I~.C
Figure 4.1.4 Initiation of motion and suspension for a current over a plane bed,
0 = f @ J , Van Rijn (1989)
-10' 2 3 4 6 102 2 3 4 6 lo3 2 3 4 6 10' 2 3 4 6 10'
-----+ particle diameter, d (km)
Figure 4.1.5 Initation of motion for a current over a plane bed, r ,, = f (d) Shields (1936)
------+ ~ * d 5 0IV
measured
1 occasional particle movement at some locations
2 frequent particle m o w m n t at some locations
3 frequent particle m w e m n t at many locations
4 frequent particle movement at nearly all locations
5 frequent particle movement at all locations
6 permanent particle movement at all locations
7 general transport (initiation of ripples)
Figure 4.1.6 Initiation of motion and suspension for a current over a plane bed, 8 = f (Re,),
Delft Hydraulics (1972)
1. Influence of criterion
The complexity of defining a critical bed-shear stress for initiation of motion is mainly caused
by the stochastical character of the driving fluid forces and the stabilizing resisting forces and
by the lack of an unambiguous definition of initiation of motion.
The most commonly used definitions (criterions) for the critical bed-shear stress are:
the zero transport rate after extrapolation of measured transport rates (Shields, 1936), see
Fig. 4.1.2,
the number of particles displaced per unit area and time (Neil1 and Yalin, 1969; Graf and
Pazis, 1977), see Fig. 4.1.7,
the qualitative transport stage based on visual observation (Kramer, 1932; Delft Hydraulics,
1972), see Fig. 4.1.6.
The extrapolation method of Shields is depicted in Fig. 4.1.2. Paintal (1971), Delft
Hydraulics (1972), Graf and Paris (1977) and others have found measurabIe quantities of
transported particles at much smaller bed-shear stresses than the Shield-values. Figure 4.1.6
shows that the Shields curve represents permanent grain movement at all locations of the bed
surface (seven modes of particle movement were distinguished by Delft Hydraulics, 1972).
It is noted that the transport rates (q,) are extremely sensitive to changes in the 8 -parameter
at low shear stresses (just beyond the critical shear stresses). Paintal (1971) reported
qb = 8 16. Extrapolation will easily lead to large errors in 8, using this method.
Grass (1970) introduced a stochastic approach to define initiation of motion. Particle motion
will occur when the probability functions of both the driving bed-shear stresses and the
resisting stresses overlap each other (see Fig. 4.1.8). A minimum overlap region must be
present before initiation of motion is generated. Grass (1970) defined the minimum overlap
region by means of a multiplication factor n for the standard deviation of the bed shear stress:
T + n u b = T ,- n a ,, with T ,= space and time-averaged critical bed-shear stress (constant
for each type of sediment material) and a, = standard deviation of critical bed-shear stress.
Using a / t = 0.4 and a ,/ t ,= 0.3, Grass found n = 0.625 for critical values corresponding
to the Shields curve, which means that 7,,&,& = 0.65 t ,. Lavalle and Mofjelt (1987) also
question the existence of critical conditions for initiation of motion. Given adequate time,
movement of particles somewhere will always be observed because turbulent flow basically
is a stochastic process and there will always be a bed-shear stress large enough to move a
particle.
Although a critical stage below which no single grain is moving, may not really exist, a
critical stage is necessary for practical design purposes.
2. Influence of shape, gradation and size
Experiments with particles of different shapes show that the €),-parameter is not much
affected by the shape of the particles when the nominal diameter (diameter that yields the
same volume) is used as the characteristic parameter. Very flat particles have larger 0,-
values (factor 1.5 to 2).
Gradation has an effect when the size range is rather wide (&Id, > 3), because the larger
particles will be more exposed, while the smaller particles are shielded by the larger particles.
Egiazaroff (1965) proposed a method to determine the critical bed-shear stress of each
fraction of the bed material, as follows:
N = number of particles moving per unit ama (m2)
Figure 4.1.7 Initiatzon of motion for a current over a plane bed, Grqf and Paris, 1977
4 motion
incipient transport
2\
-
2
&
50
bed-shear stress
Figure 4.1.8 Initiation of motion for a current over a plane bed according to stochastic
approach (Grass, 1970)
in which:
= critical bed-shear stress of fraction i with size di
*b,a = critical bed-shear stress of the average diameter d,
6 = [ log( 19)
log (19 di/dJ
$ = exposure coefficient
Equation (4.1.13) yields: 6 = 5 for d,/$ = 0.2 and ti = 0.4 for di/& = 5. Thus, particles
smaller than d,,, have a much larger s,,, because they are shielded between the large
particles, whereas the particles larger than 4, have a smaller s,,, because they are more
exposed.
Armouring will occur when the bed-shear stress is not large enough to move the largest
particles of the bed mateiral. When there is no supply of smaller particles from upstream, all
smaller particles will eventually be eroded, and the coarser particles will form an armour
layer preventing further scour. Based on experimental results, the median diameter (&) of
the armour layer will approximately be equal to the $, of the initial bed material. This
phenomenon has been observed downstream of weirs.
Mantz (1977) studied the initation of motion of fine cohesionless flaky sediments with particle
sizes in the range of 10 to 100 pm and proposed the following critical bed-shear stress (see
Fig. 4.1.4):
'b,cr = 0.1 for 0.03 < -

(P,-p)g d, v
The data presented by Miller et al (1977) also indicate that 6, of Shields is too large for fine
sediments (D, < 10). A better representation of the data presented by Miller et a1 is given
by (see Fig. 4.1.14):
0, = 0.11 D
": for D,<10 (4.1.15)
Flume tests on loose (not cemented) calcareous sediments (coral sands) showed a similar
behaviour as quartz sediments of the same size (Delft Hydraulics, 1972).
3. Influence of bed slope
a. Longitudinal slope P
A sediment particle resting on a longitudinal sloping bed, will be set in motion (see Fig.
4.1.9), when the sum of the critical fluid force (assuming a , = 1, Eq. 4.1.4) and the gravity
force component is just equal to the stabilizing force (= N tan+), yielding:
The critical fluid force for a horizontal bed (F,,,,,) is given by Eq. (4.1.4), giving ( a ,= 1):
Combining Eqs. (4.1.17) and (4.1.18), the critical fluid force (F,,) on a sloping bed can be
related to the critical fluid force on a horizontal bed (FD,,,,), as follows:
in which:
h ( 4 P)
+
for upsloping flow
kP= sin4
Equation (4.1.16), first presented by Schoklitsch (1914), can also be expressed in terms of
the bed-shear stress, giving:
Whitehouse and Hardisty (1988) have shown that Eq. (4.1.19) and (4.1.21) yield realistic
results compared with experimental data.
\
\
\
Figure 4.1.9 Forces on a sediment particle (longitudinal bed slope)

Figure 4.1.1 0 Forces on a sediment particle (transverse bed slope)
b. Transverse slope y
A sediment particle resting on a transverse sloping bed will be set in motion (see Fig. 4.1.10)
when the resulting driving fluid force is equal to the stabilizing fluid force, giving:
Combining Eqs. (4.1.24) and (4.1.18), it follows:

in which:
[
k., = cosy 1 - -
:;r
This factor was first presented by Leiner (1912)!
For a combination of a longitudinal and a transverse bed slope, it follows that:
4. Influence of bed forms
As soon as the sediment transport process is established, ripples and dunes are formed on the
bed. The critical bed-shear stress for initiation of motion on a bed form is different from that
for a flat,bed. The bed shear stress (rJ over a bed consisting of bed forms,,is composed of
a part (t,) related to (skin) friction over the bed surface and another part (r,) related to the
non-uniform pressure distribution over the bed form crest and eddy region (see Chapter 6 ) .
A sediment particle resting on the surface of a bed form will be set in motion by the fri$ion
force (s,,,) or by the turbulent fluctuations in the eddy region downstream of the crest(r,,,).
This means that the critical bed-shear stress (t,,) is always larger when bed forms are
present than when the bed is flat. The initiation of motion on a bed form has been observed
to occur at the downstream end of the eddy region, where the flow reattaches to the bed.
Despite the fact that the velocity and shear stress are small near the reattachment point, the
longitudinal turbulence intensity is relatively large at that location.
5. Influence of cohesive material

When the bed consists of silty and muddy materials, cohesive forces between the sediment
particles become important. These forces cause a distinct increase of the strength of the soil
against erosion. Depending on the type of clay minerals, this effect may be more or less pro-
nounced. Flume experiments show that an amount of 25% of clay in a sand bed caused a
dramatic reduction (factor 30) of the sand concentrations generated by wave action (Van Rijn,
1985).
Biological activity at the bed may also influence the critical values for inititation of motion,
especially in muddy and silty environments.
An important factor governing the erodibility of cohesive soils is the rate of consolidation.
Fresh mud deposits have a very loose texture of mud flocs which already have a low density
themselves. The wet bulk density of such a deposit may be within the range of 1050 to 1100
kg/m3 of which 95% or more consists of water. In this stage the cohesive forces in the
deposit are still very low, and erosion can occur easily. If the deposits are not eroded again,
its density gradually increases as interstitial water is pressed out of the fresh soil by the
weight of the deposit itself. This process of consolidation initially goes relatively fast but
gradually slows down. With the compaction of the soil the erosion resistance rapidly
increases. It is well known that old compact clay soils are highly resistant against erosion.
A sand bed with small percentages of silts and clays (silty or clayey sand) already shows a
distinctly increased resistance against erosion. Little is known of the erodibility of cohesive
soils in a quantitative sense.
Some information can be obtained from flume tests performed at Delft Hydraulics (1989) on
bed samples from the North Sea with particle sizes (&) in the range of 100 to 200 pm and
mud-silt percentages in the range of 2 to 20%. The measured critical bed shear stress was
related to the critical bed-shear stress according to Shields, yielding:
in which:
p, = percentage of fines (mud, silt) smaller than 50 pm (in %, minimum value = 1%)
4.1.3 Critical depth-averaged velocity
The earliest studies were related to critical velocities of stones (Brahms, 1753 and Sternberg,
1875). They studied the critical near-bed velocity and found that it was related to the particle
diameter, as follows:
The near-bed velocity is, however, not very well defined and therefore it is prefered to use
the critical depth-averaged velocity (ia) as the characteristic parameter.
One of the most well-known relationships between the critical depth-averaged velocity and
the particle diameter (in graphical form) is given by Hjulstrom (1935). The influence of the
relative bed roughness (ratio of water depth and particle diameter, h/d) was, however,
neglected by Hjulstrom.
The critical depth-averaged velocity can be derived from the critical bed-shear stress using
the ChCzy-equation. Assuming hydraulic rough flow conditions (u,\/v > 70), the critical
depth-averaged flow velocity (ia) for a plane bed can be expressed as:
-
in which:
u = depth-averaged critical velocity
u.,, = 8; [(s-1)g ddo5 = critical bed-shear velocity (mls)
h = = water depth (m)
= a & = effective bed roughness of a flat bed (m)
a = coefficient ( a = 1 for stones d, r 0.1 m and a = 3 for
sand and gravel material)
Equation (4.1.32) can be expressed as:
Using Eq. (4.1.33), k, = 3d, and d, = 2d, and the Shields curve, the critical depth-
averaged velocity for sand particles in the range of 0.0001 to 0.002 m can be expressed as:
-u, (4.1.34)
= 0.19 (dm,".' log for 0.0001 S d, S 0.0005 m
-
u, = 8.50 (dm)" log for 0.0005 < d, s 0.002m
in which:
d, = median particle diameter (m)
d, = 90% particle diameter (m)
Equations (4.1.34) and (4.1.35) are shown in Figure 4.1.11 for a plane bed.
Many expressions for the critical depth-averaged velocity of (coarse) gravel and stone
material (d, 1 0.002 m) can be found in the Literature. Herein, the expressions of Neill
(1968) and Maynord (1978) are given:
Neil1 : u-
, = 1.4 [(s-l)g d m y
Maynord : -,
u = I:[
1.3 [(I-1)g d m y
Y6
.--- Gcr = depth-averaged flow velocity for initiation of motion

G,,,, = depth-averaged flow velocity for initiation of suspension
Figure 4.1.11 Critical depth-averaged velocities for a plane bed

4.1.4 Design of stable channels
1. Introduction
The design and maintenance of stable channels is extremely important for irrigation systems.
The design methods can be divided into two categories:
the (empirical) regime method,
the tractive force method.
The regime method originating from irrigation channel design in India and Pakistan, is based
on a set of empirical equations which specify the water depth (h), width (b), slope (I) and
mean velocity @ as a function of discharge (Q) and particle diameter (&), (Lacey , 1930;
Lane, 1955; Blench, 1957). Thus:
Transport of sediments is allowed as long as there is no net annual scour or deposition in

each channel. In that case the channel is considered to be in regime (equilibrium). The regime
method is most suitable for large-scale irrigation systems with a wide range of discharges in
soils of silts and fine sands. The equations may only be used within their validity ranges (no
extrapolation!).
The tractive force method is based on the principle of allowing no or negligible motion of bed
material, which may be the natural bed material or dumped (protective) material like gravel,
stones or riprap. The tractive force method is most suitable for small-scale irrigation systems
with small discharges in soils of coarse material (d, 2 500 pm).
Herein, the tractive force method is presented. Information of the regime method is given by
Lacey (1930), Lane (1955) and Blench (1957). A critical review is given by Stevens and
Nordin (1987).
2. Tractive force method
As shown in section 4.1.2 (see Figs. 4.1.6 and 4.1.7), the 8,-values of the Shields curve
represent conditions with frequent particle movements at all locations of the bed. Therefore,
the Shields curve can not be recommended for the design of a stable channel. A better
approach is the application of the experimental results of Paintal (1971) who measured bed-
load transport rates close to initiation of motion (particle size d, = 0.0025 m, 0.008 m and
0.022 m). The following dimensionless transport rates were measured:
4 = lo-' for 8 = 0.01
4 = for 8 =0.02
4 = lo-' for 8 = 0.025
4 = lo4 for 8 = 0.03
4 = 10" for 8 =0.04
in which: 3 0.5
4 = nb/[P.((s-l)g &o) ] = dimensionless transport rate (-)
qb = = bed load transport rate (kglsm)
S = P,/P = specific density (-)
d, = = median particle size (m)
8 = r,/[(p,- p)g dm] = dimensionless mobility parameter (-)
The experimental results of Paintal(1971) have been used by the author to compute the bed-
load transport rate (in kg per m width per day) and the corresponding number of moving
stones (per m width per day) for stone sizes of 0.05, 0.1 and 0.3 m at small 8 -values of
0.02, 0.025 and 0.03 (all below the Shields curve), see Table 4.1.
A value of 8 = 0.02 appears to be a safe value yielding a number of moving stones much
less than one per m width per day.
Applying Eq. (4.1.33) and assuming a = 1, d, = 2d,,; the critical depth-averaged velocity
for stones is:
8 = 0.03: -,
u = 1.0 [(s -1)g dS0Ylog
Equations (4.1.39) and (4.1.40) are valid for uniform river flow. For regions with additional
turbulence production (downstream of structures; weirs, bridge piers) a reduction factor a,
must be taken into account, yielding:
in which:
a, = 1.45/(1+3r) = reduction factor related to additional turbulence
r = uu/u = relative turbulence intensity (r = 0.15 for uniform flow
r = 0.35 for region dowstream of hydraulic jump)
= standard deviation of time-averaged velocity
u = time-averaged velocity
stone diameter qb = bed load transport in kg per m width per day

N = number of stones per m width per day
dm (m)
8 = 0.02 8 = 0.025 8 = 0.03
0.05 qb = 0.1 qb = 1 qb = 10
N = 0.6 N = 6 N = 60
0.10 q = 0.3 qb = 3 qb = 30
N = 0.2 N = 2 N = 20
. 0.30 q, = 1.5 qb = 15 qb = 150
N = 0.04 N = 0.4 N = 4
Table 4.1 Bed load transport close to initation of motion according to Paintal (1971)
3. Schematimtion of cross-section
The most practical cross-section that can be constructed is a trapezoidal cross-section.
Analysis of computed and measured bed-shear stresses along the cross-section of trapezoidal
channels (with a ratio of the surface width and the depth larger than 5) has shown the
following results (Olsen and Florey, 1952; Lane 1955):
= 0.75 pghI = 0.75 td, for side slope 1:2

I,,, = 0.85 pghI = 0.85 tdae for side slope 1:3
I,,, = 0.90 pghI = 0.90 tdae for side slope 1:4
Z bsids = 0.95 pghI = 0.95 tda, for side slope 1:6
in which:
z bpldme =
=b&de =
maximum bed-shear stress in middle part of channel
maximum bed-shear stress in side part (with slope vertical to horizontal)
The bed material will be stable when the prevailing maximum bed-shear stress is smaller than
the critical bed-shear stress. Thus,
Middle part is stable if : tbpla~ ' 'b,a,o
Side part is stable if : ' k, 'b,a,o
in which:
= critical bed-shear stress for horizontal middle part (N/m2)
= cosy [I -m2y/m24]0.5 = reduction factor (-)
- = angle of side slope (")
X
I
-
-
- = angle of repose (o)
= surface slope (-)
h - = water depth (m)
The angle of repose 4 (see section 3.2.6) is given in Table 4.2).

The shape of an ideal stable channel with the same critical conditions along the cross-section
can be determined by assuming (see Fig. 4.1.12):
pghI AY = =, cosy
*y or r, = pgh~cosy
and
tb = at y =O (middle of channel)
Thus,
At critical conditions, it follows that:

-
'b = =b,a 'b - ky 'b,a o ~ d i =, kr,yielding
h = 6 11-ta12~/tan~41~.~
Finally, a consinusoidal shape of the cross-section is found:
in which:
h = water depth at y from the middle
6 = water depth at y = 0 (middle)
For 4 = 35", the following parameters can be computed:
surface width 2b = 4.5 &
area A = 2.9 (h)2
perimeter P = 5 6
hydraulic radius R = 0.6 6
critical discharge Q = AC(R1)O.' = P.5
Recent information on this topic was presented by -Dipla5 and Vigilar, 1992.
Particle size d, Angle of repose 6,
(m) Rounded material Angular material
5 0.001 30" 35 "

0.005 32" 37"
0.01 35" 40"
0.05 37" 42 "
1 0.1 40" 45 "
Table 4.2 Angle of repose 6,
Figure 4.1.12 Ideal stable channel shape

4.1.5 Examples and problems
1. A wide open channel with a plane sloping bed has a water depth h = 2 m, slope I = 0.5
-
lo", bed material characteristics d, = 0.002 m, d, = 0.005 m, p, = 2650 kg/m3,
water temperature Te = 20°C, and water density p = 1000 kg/m3.
Is there movement of bed material according to the Shields curve?
The critical bed-shear for d, = 0.002 m is (Fig. 4.1.5) : s,,, = 1.3 N/m2.
The applied bed-shear stress is : t, = pghI =9.8N/m2.
Thus, t, > t,,, and movement of bed material.
2. A wide open channel with a plane sloping bed has a slope I = l o 5 , bed material
characteristics d, = 0.0002 m, d, = 0.0003 m, p, = 2650 kg/m3, water temperature
Te = 20°C, kinematic viscosity v = 1 10" m2/s, water density = 1000 kg/m3.
What is the maximum discharge per unit width without movement of bed material
(according to Shields curve)?
The critical bed-shear stres for
d,, = 0.0002 m is (see Fig. 4.1.5) : t,,, = 0.18 N/m2
The water depth h follows from : t,,, = pghI = 0.18, yielding h = 1.83 m
The discharge per unit width is
given by (see Eq. 2.2.19) 9 = ii h = c h1.5 p.,
Assuming flow in the hydraulic
transition regime, the dhtzy
equation is (see Eq. 2.2.24) :C
The effective bed roughness of

a plane bed is (see Eq. 4.1.32) : k, = 3 d, = 0.0009 m
The bed-shear velocity is : UI = (rblp)" = (0.18/1000)~.~= 0.0134 QI
Thus, the ChCzy-coefficient is : C = 77.1 mO.'/s, (u&/ v = 12)
The maximum permissible

discharge is : 9 = 0.6 m2/s.
One of the most uncertain parameters in this computation is the estimation of the effective
bed-roughness of a plane bed (see Eq. 4.1.32). Taking k, = 1 d,, it follows that C =
82.8 rn0.'/s and q = 0.65 m2/s. Taking the largest value k, = 3 d,, a safer (more
conservative) value of the maximum discharge q = 0.6 m2/s is obtained.
3. A wide open channel has a water depth of h = 5 m. The bed is covered with sand dunes.
The bed material characteristics are d, = 0.0003 m, d, = 0.0005 m, p, = 2650 kg/m3.
The overall ChCzy-coefficient is C = 50 m0.5/s.The water temperature is 20°C. The
kinematic viscosity coefficient is v = 10" m2/s. The fluid density is p = 1000 kg/m3.
What is the depth-averaged current velocity at initiation of motion at the upsloping part
of the dunes and what is the overall bed-shear stress?
Using the Shields curve

(Fig.4.1.5)fordm =0.0003m : tb,= = 0.2 N/m2
The grain-related bed-shear I

stress (see Chapter 6 ) is : T,, = ~gculc?~
The grain-related ChCzy

coefficient is : C' = 18 log[12h/(3 v ] 83 moa5/s
=
At initiation of motion : tb
I -
- 'b,a
The overall bed-shear stress is : tb = pgGr./q2 = 0.57 N/m2
4. A wide open channel with a plane sloping bed has a water depth of h = 1.7 m, a depth-
averaged velocity ii = 2.5 mls, bed material characteristics d, = 2 d,, water
temperature Te = 20°C, kinematic viscosity v = 1 lod m2/s, fluid density = 1000
kg/m3, sediment density p, = 2650 kg/m3.
What is the minimum size of the bed material to obtain a stable bed (no movement)?
Assuming hydraulic rough flow,
the bed material size can be
obtained from Eq. (4.1.33) :u
,
- = 5.75 [(s-l)gd,PJ 8; log(?)
\ - 4
The effective bed roughness

is estimated to be : k, =1d,=2d,
The critical 8 -value is taken
as (safe value) : 8, = 0.02
using s = 2.65 and

g = 9.81 m/s2, it follows that
- = 3.27 & log
: ucr
Using i, = 2.5 m/s and
h = 1.7 m yields : d, = 0.2 m
Check hydraulic regime of rough
flow which requires
u, k,/v > 70, yielding *
. -* - ii 11 = 1.02 105
- -go5
v C v
Taking a critical value of 8, = 0.03 yields a much smaller size of d, = 0.1 m.

5. A trapezoidal channel has a water depth h and a bottom width b and side slopes of
1 to 2 ( y = 27"); the water surface slope is I = 3.5 lo4. The discharge is Q = 160
m3/s. The bed material consists of rounded stones with d, = 0.04 m and d, = 0.08 m. -
The water temperature is 20°C; the kinematic viscosity is v = 1 10" m2/s; the densities
are p = 1000 kg/m3 and p, = 2650 kg/m3.
What is the water depth (h) and the bottom width (b) to obtain a stable bed?
Take a safe 8 -value : 8, = 0.03
The critical bed-shear stress

at the middle part is : 'b,a,o = 0.03(pS-p)gd, = 20 N/mZ
The critical bed-shear stress

at the side part is : 'b,a = 1C b,=,o = 20
Using 4 = 36" (see Table 4.2)
and side slope angle y = 27 " ,
yields : $ = cosy [l-tan2y/tan24]03 = 0.64
yielding . : Tb,u = 20 0.64 = 12.8 N/mZ
The middle part is stable, if

This yields
: eb
h
= pghI = tb,=.
= 20/(pgI) = 5.8 m
The side part is stable, if : 3, = 0.75 pghI = 12.8, see Eq. (4.1.43)
h = 12.8/(0.75 pghI) = 5 m
The smallest water depth is

decisive, thus :h =5m
The cross-section area is :A = bh + 2h2 = 5b 50+
The wetted perimeter is :P =b + +
2t2.23 h) = b 22.4
The hydraulic radius is :R + +
= (5b 50)/(b 22.4)
The ChCzy coefficient is :C = 18 log(12RkJ
The effective bed roughness is : k, = d, = 0.08 m
The discharge is :Q = AC(RI)O.S= 160 m3/s
A bottom width b = 10 m yields :A = 100 m2, P = 32.4 m, R = 3.1 m
C = 48 m0.5/sand Q = 158 m3/s.
Thus, water depth is h = 5 m and the bottom width is b = 10 m.

6. A wide open channel with a plane sloping bed has a bed surface slope of I = 104. The
bed material characteristics are d, = 0.003 m and d, = 0.006 m. Other data are: p =
1000 kg/m3, ps = 2650 kg/m3, v = 1 lo6 m2/s.
What is the maximum discharge (per unit width) without movement of bed material?
Solution: q = 1.9 m2/s for k, = 3 d,
q=2.2m2/s for &=Id,
7. The bottom of a wide open channel is protected with uniform round stones (diameterd).
The stones have a mass of 30 kg and a sediment density of p, = 2800 kg/m3. The water
depth is h = 4 m. Other data are: p = 1000 kg/m3, v = 1 lo6 m2/s.
What is the critical depth-averaged velocity using 8, = 0.03?
Solution: = 4.9 m/s for k, = d
8. A trapezoidal channel has a water depth (in middle) of h = 2 m, bottom width b = 15
m and side slopes 1 to 2 ( y = 27"). The bottem is covered with rounded stones d, =
0.05 m, k, = 0.05 m.
What is the maximum bottom slope and the maximum discharge to obtain a stable bed
(use Shields)?
Solution: 1, = 2 10"
Q, = 100 m3/s
4.2 Initiation of motion in waves
4.2.1 Critical velocity
In oscillatory flow there is no generally accepted relationship for initation of motion on a
plane bed. Many equations have been proposed. Silvester and Mogridge (1970) present 13
different equations collected from the Literature. One of the more popular equations is that
of Komar and Miller (1975):
for d,, < 500 pm
@ad2
(s- l)gd,
= 1.45 [2tzr for dm 2 500 pm
in which:
IJ, =,critical peak value of orbital velocity near the bed
A , , = critical peak value of orbital excursion near the bed
Analysis of Eqs. (4.2.1) and (4.2.2) shows a weak increase of 0 , , for an increase of the
wave period T.
Various sets of previously published data: Bagnold (1946), Manohar (1955), Rance and
Warren (1968), Silvester and Mogridge (1970), Dingler (1974), Bosman (1981) and Davies
(1985), were analyzed by Van Rijn (1989) to describe the critical velocity o, as a function
of the particle size d, and the wave period T. Only experiments with 'sand particles
(p, = 2650 kg/m3) and wave periods in the range of 4 to 15 seconds were considered. The
data represent a particle range from 90 to 3300 pm.
Figure 4.2.1 shows the peak value of the critical near-bed orbital velocity as a function of the
wave period and sediment size. The average inaccuracy of the curves is about 25 %.
Detailed analysis of the data shows a clear influence of the wave period. For most data the
critical stage (in terms of the peak orbital velocity, 0,,,) increases with the wave period,
which is consistent with Eqs. (4.2. I), (4.2.2). The experimental results of Silvester and
Mogridge, however, show an opposite trend.
The experimental data presented in section 4.2.1 can also be expressed in terms of the Shield
parameter using the time-averaged bed-shear stress (tb,,), yielding (Van Rijn, 1989):
in which:
b,w,a
= 0.25 pfW(~,,)2 = time-averaged (over half period) wave-related bed-shear
stress, Eq. (2.3.14)
f, = wave friction factor
',,a = peak value of critical near-bed orbital velocity according to linear wave theory,
Eq. (2.3.2)
-------+ wave period, T (s)
measured: -
o 100 150 pm -
r 4 0 0 800pm
-
150 200 pm -
o 8 0 0 1600pm
v 2 0 0 - 4 0 0 pm 8 1600 - 3200 pm
Figure 4.2.1 Initiation of motion for waves over a plane bed based on critical velocity
-------+ particle parameter, D*

Figure 4.2.2 Initiation of motion for waves over a plane bed based on critical bed-shear
stress
The time-averaged bed-shear stress and not the maximum bed-shear stress was used by Van
Rijn (1989) because the Shields curve, originally proposed for unidirectional flow, is based
on time-averaged parameters.
To determine the (time-averaged) wave-related bed-shear stress, the friction factor must be
known, which means that the grain roughness (k$ and the kinematic viscosity (v) must also
be known. The k-value is assumed to be equal to a d , with a in the range of 1 to 3 (Chapter
6). In some cases the d,was not reported in the Literature. To overcome this, a value equal
to d, = 1.5 d, was assumed. When the water temperature was not reported, a value of 20"
was assumed.
In the case of turbulent smooth conditions the wave friction factor was computed from Eq.
(2.3.17). In the case of turbulent rough conditions Eq. (2.3.20) was applied.
For the transitional regime (in analogy with unidirectional flow) the wave friction factor was
determined from the equation for turbulent rough conditions (Eq. 2.3.20) representing the
wave-related roughness by:
The a-coefficient was assumed to be in the range of 1 to 3 (see Chapter 6). To determine the
critical bed-shear stress for the transitional regime, an iterative solution method was used
because the &-parameter is a function of the critical bed-shear stress (Eq. 4.2.4). Figure 4.2.2
shows the dimensionless critical bed-shear stress as a function of the dimensionless D,-
parameter for the selected data. The Shields curve which is valid for unidirectional flow data
only, is also shown. A vertical bar is used to express the influence of the wave period and
the a-value (1 to 3) for each particle size. The variation between the results of different
investigators is mainly caused by the definition problem for initiation of motion. For example,
Rance and Warren (1968) define the critical stage as the stage when one or two particles are
dislodged and moved ove a small distance, while Bosman (1981) defines the critical stage as
that when about 10% of the surface particles is in motion resulting in a relatively large
critical bed-shear stress (see Figure 4.2.2).
Based on these results, it seems justified to conclude that the Shields curve can also be
applied as a criterion for initiation of motion for oscillatory flow over a plane bed. The curve
represents a critical stage at which only a minor part (say 1 to 10%) of the bed surface is
moving.
Figure 4.2.2 is only valid for plane bed, which is a rare phenomenon in natural conditions.
Usually, small-scale ripples are present and in that case the critical velocities for initiation
of motion are considerably smaller due to the generation of ~ortex~motions. For example,
Carstens et al (1969) report a 50 %-reduction of the peak velocity (U,,,) for a rippled bed.
1. The water depth in a coastal sea with a plane bed is h = 5 m. The wave period is T =
7 s. The bed mateiral characteristics are d, = 0.0002 m, d, = 0.0003 m, p, = 2650
kg/m3. The water temperature is Te = 20 "C. kinematic viscosity coefficient v = 1.10"
m2/s, fluid density p = 1025 kg/m3.
What is the wave height at initiation of motion?
Method 1 using Fig. 4.2.1
The wave length can be computed

from : L = ( g ~ ~ / tanh(2xh/L)
2~)
yielding L =46m
Using Fig. 4.2.1 and d, = 200 pm,

the critical peak orbital velocity
can be obtained : q ,= 0.23 m/s
The critical wave height can 1
be obtained from Eq. (2.3.2) : H, = - fT8,Tsinh(2ah/L) = 0.38m
R
Method 2 using Shields curve Fig. 4.2.2
The dimensionless particle size is : D, = [cs ;:)g]Vl d, = 5
The dimensionless Shields value is : 0, = 0.14 D

:~ = 0.05
The critical bed-shear stress is : rb,~ = 0.05 (p, - p)gd, = 0.16 N/m2
The mean wave-induced bed-shear

stress is given by (Eq. 2.3.14) : =b,w = -
P fw@,)2
4
At initiation of motion : Tb,w - =b,= OT -
lM5 fw@,,.J2 = 0.16
4
Assuming a transitional hydraulic
regime, the friction coefficient is
(Eq. 2.3.20) : fw = exp
Taking 4 = 0,T/2n, k, = 3 d,
and u , , = (rd p ) 0 5 = (0.16/1025)~~
= 0.0125 m/s, yields : f = exp[-6 + 1.4@,,a)4.19]
~ h u srbSw= rbSayields
4
1025 @,,a)2 eW[-6 ' ~0.16
+ 1.4@,,~)~= ]
: @,,J2 exp[-6 + 1.4@,,2'19] = 0.00062
yielding : o , , = 0.19 m/s

1
The critical wave height is : H, = -
7r
fT,, T sinh(2trh/L) = 0.31 m
Method 3 using Eq. (4.2.1) of Kornar
Critical peak orbital velocity at
the bed
using A,, = O,,T/2n
yielding : ',pa = 0.17 m/s
2. A coastal sea (plane bed) has a water depth h = 5 m. The peak wave period is T, =
7 s. The significant wave height is H, = 1 m. Other data are: p = 1025 kg/m3, p, =
2650 kg/m3, v = 1 lo4 m2/s, d, = 2 d,, k, = 1 &.
What is the d, of the bed material which is just stable at the given conditions (use
methods of Shields, Komar and Van Rijn, Fig. 4.2. I)?
solution: d, = 0.0026 m (Shields)
d, = 0.0038 m (Komar)
d, = 0.0032 m (Van Rijn)
4.3 Initiation of motion for combined current and waves
Information of initiation of motion for combined unidirectional and oscillatory flow is rather
scarce. Some results were presented by Larsen et al (1981) for three continental shelf sites
in the United States and in Australia, and by Hammond and Collins (1979) for flume wndi-
tions. The data of Larsen et al included measurements of current velocity and direction at
1 m above the sea bed, mean bottom pressures, photographs of the sea bed and bed material
samples. The bed material at the USA-sites was characterized as sandy silt (d,= 35 pm and
42 pm). The bed material at the Australian site was characterized as fine sand (d,= 170
pm). The water depths at the sites were in the range of 75 to 90 m. Information of the local
bed forms (ripples or plane bed) was not reported by Larsen et al.
The data were used by Van Rijn (1989) to determine the initiation of motion parameters
according to the Shields method.
The resulting time-averaged total bed-shear stress (tbpm) is computed from Eq. (2.4.25),
,,,
neglecting the wave-current interaction. Thus, t , , = t,, + s .The (time-averaged) wave-
related bed-shear stress ( 7 ,), is determined from the peak value of the bed-shear stress (ib,,)
reported by Larsen et al(l g8 I), applying s, ,= l/2 f , ,. The current-related bed-shear stress
is computed from the measured velocity (u3 at a height (z,) of 1 m above the bed, assuming
a logarithmic velocity distribution in the near-bed region, as follows:
in which:
k, = effective roughness height (Eq. 4.2.4).
The sea bed was assumed to be flat. The water temperature was assumed to be 15°C.
Applying the above-given approach, the computed total time-averaged bed-shear stresses
(tb,,) were somewhat (15 %) larger than those reported by Larsen et al, because the wave-
current interaction was neglected by Van Rijn (1989). The dimensionless parameters for
initiation of motion are presented in Fig. 4.3.2, showing critical values that are smaller than
the Shields values, especially for the two sandy silt sites (d,= 35 pm and 42 pm). The
results are in better agreement with the curve proposed by Mantz (1977).
------, -
current related bed- shear velocity, u.,+c (mls)
Figure 4.3.1 Critical bed-shear velocitiesfor combined current and waves over a plane bed
(Hammnd and Collins, 1979)
-------+ particle parameter, D,

Figure 4.3.2 Initiation of motion for combined currents and waves over a plane bed,
e = f(D*)
The data of Hammond and Collins (1979) were measured in a flume (length = 3.7 m, width
= 0.3 m, depth = 0.3 m). In the flume, a perspex carriage with sediment material (dm =
142, 363, 771 and 1134 pm) was installed, which could be set into an oscillating motion.
Time-averaged velocities were measured at 0.02 m above the sediment bed by using a
miniature propeller-type current meter. The test results of the two finest sediments (142 and
163 pm) were used by Van Rijn (1989). The test results of the two coarsest materials (771
and 1134 pm) were not used because they may have been influenced by relatively large
inertial forces acting on the grains produced by the horizontal acceleration of the sediment
bed (on the carriage).
Figure 4.3.1 shows the critical combinations in terms of the wave-related and the current-
related bed-shear velocities computed from Eqs. (2.3.14), (4.2.4) and (4.3.1) .As can be
observed, larger critical current velocites (u,,J are generally achieved for small wave periods
than for large periods at a given u, ,-value. Probably, this is the result of a relatively large
reduction of the near-bed current velocities of short-period waves for which the wave-induced
vortex motions near the bed are relatively intense (see also section 2.4). Consequently, the
imposed current velocity has to be larger to initiate motion for a short period wave than for
a large period wave at the same wave height.
Figure 4.3.2 shows the experimental results in terms of the Shields parameters. The resulting
critical bed-shear stress is computed from Eq. (2.4.25) as the sum of the wave-related and
the current-related bed-shear stress ( t b , = r b , + t,,). The vertical bar indicates the
influence of the wave period and the error in the estimated &-value (1 to 3 &). Generally,
the longer period wave produce the smallest critical mobility parameters. As can be observed,
the mean critical values are close to the Shields curve. The extreme vaues are about 40%
smaller and larger than the mean values.
1. The water depth in a coastal sea with a plane bed is h = 5 m. The wave period is T =
7 s. The depth-averaged current velocity is 0 = 0.3 m/s. The angle between the wave and
current direction is 60". The bed material characteristics are d, = 0.0003 m, d, =
0.0005 m, p, = 2650 kg/m3. The water temperature is Te = 20°C, the kinematic
viscosity coefficient is v = 1. lo6 m2/s. The fluid density is p = 1025 kg/m3.
What is the wave height at initiation of motion?
The wave length follows from
(Eq. 2.4. I), : [(L'm- ii cos$)y = ( g '/2x)
~ tanh(2xh/L?
yielding : L' = 47m
The relative wave period is T

(Eq. 2.4.1) : Tr = = 7.2 s
1-(ii T cQS4)/L '
The particle size parameter is : D, = 7.5
The 'critical Shields parameter is : 8, = 0.037 (see Fig. 4.2.2)
The critical bed-shear stress is : Zb,~ = 0.037 (p, - p) gd, = 0.18 N/m2
The mean bed-shear stress exerted

by combined wave and current
(Eq. 2.4.25) is : =~,cw - 'b,~ + 'b,~
The current-related bed-shear
stress (assuming transitional
flow and k, = 3 &) : =b,c - P~<V
yielding : =b,c = 0.14 N/m2
The wave-related bed shear stress

(Eq. 2.3.14) : =b,w = 0.25 p fw @,12 =
At initiation of motion
Taking U,,w = (rb,w/p)o'5 = 0.00639
v = m2/s, yields : 256.2 @,aexp[+ + = 0.04

1.6@b9~Q19]
0 6 ,= 0.066 m/s
1
The critical wave height is : Hu = -
X
O,,T, sinh(2rrh/L') = 0.11 m
2. A coastal sea (plane bed) has a water depth of h = 5 m. The significant wave height is
H, = 1 m. The peak wave period is T, = 7 s. The depth-averaged current velocity is;
= 1 mls. The angle between the wave and current direction is 90". Other data are: p =
1025 kg/m3, ps = 2650 kg/m3, v = 1 lod m2/s, d, = 2 d,, k, = 3 d,.
What is the d, of the bed material which is just stable at the given conditions (use
Shields)?
solution: d, = 0.012 m
4.4 Initiation of suspension in currents
For increasing values of the bed-shear velocity, the particles will be moving along the bed
by more or less regular jumps (saltations). When the value of the bed-shear velocity becomes
comparable to that of the particle fall velocity, the sediment particles may go into suspension.
Bagnold (1966) stated that a particle only remains in suspension when the turbulent eddies
have dominant vertical velocity components which exceed the particle fall velocity (wJ.
Assuming that the vertical velocity component (w') of the eddies are of the same order of
magnitude as the vertical turbulence intensity (I?), the critical value for initiation of
suspension can be expressed as:
Studies on turbulence phenomena in boundary layer flow provide detailed information of the
vertical distribution of the turbulence intensity (Hinze, 1975). The vertical turbulence intensity (I?)
has a maximum value of the same order as the bed-shear velocity (b),both for hydraulic
smooth and rough flow conditions. Thus: I? = u*.
Using the above-mentioned values of the vertical turbulence intensity, the criterion for
initation of suspension becomes:
which can be expressed as:
in which:
u , , = critical bed-shear velocity for initiation of suspension (mls)
w, = particle fall velocity in clear (still) water (mls)
dm = particle diameter (m)
p, = density of sediment (kgIm3)
p = density of fluid (kg/m3)
Another criterion for initiation of suspension was given by Engelund (1965). Based on a
rather crude stability analysis, he derived:
Equations (4.4.3) and (4.4.4) are shown in Figure 4.1.4.
An experimental investigation was carried out at Delft Hydraulics (Delft Hydraulics, 1982)
to determine the critical flow conditions for initiation of suspension defined as the stage of
flow at which the particles perform a jump length larger than about 100 particle diameters.
Visual observations during the (critical) flow conditions showed instantaneous upward particle
motions (turbulent bursts) with jump heights in the range of 0 to 100 particle diameters at
various locations of the bed.
The experiinental results, which are shown schematically in Figure 4.1.4 can be presented
by:
Summarizing, it is suggested that the criterion of Bagnold may define an upper-limit at which
a concentration profile starts to develop, while Equation (4.4.5) may define an intermediate
stage at which locally turbulent bursts with sediment particles are lifted from the bed into
suspension.
4.4.2 Critical depth-averaged velocity
Equation (4.4.5) was used to compute the critical depth-averaged velocity for a plane
bed. The grain roughness was computed as k, = 3 d, with d, = 2 d,. The critical depth-
averaged velocity follows from (see Eq. (4.1.33)):
- = 5.75 [(s-1) g d5,Y :0 log

4.4.3 Examples
1. A wide open channel with a plane sloping bed has a water depth of h = 5 m. The bed
material characteristics are d, = 0.0003 m, d, = 0.0005 m, p, = 2650 kg/m3. The
fall velocity is w, = 0.04 mls. The water temperature is Te = 20°C, the kinematic
viscosity coefficient v = 10" m2/s, the density p = 1000 kg/m3.
What is the depth-averaged current velocity at initiation of suspension?
According to Bagnold : 8aa = 0.33
According to Van Rijn

The depth-averaged current velocity
at initiation of suspension follows
from : tb = pg6102 =
Assuming transitional flow : C = 18 log

3 ,&t + 3.3 vlu* I
Thus, Bagnold
Van Rijn
Bagnold, R.A., 1946. Motion of Waves in Shallow Water; Interaction between Waves and
Sand Bottom Proc. Royal Soc. London, Ser. A, 187, p. 1-15, England.
Bagnold, R.A., 1966. An Approach to the Sediment Transport Problemform General Physics
Geological Survey Prof. Paper 422-1, Washington.
Blench, T., 1957. Regime Behaviour of Canals and Rivers Butterworth, London.
Bonneffle, R. (1963). Essais de Synthese des Lois de Debut d 'entrainment des Sediments
sous 1 'action d'un Courant en Regime Unifomte Bull. du CREC, No. 5, Chatou.
Bosman, J., 1981. Bed Behaviour and Sand Concentration under Oscillatory Water Motion
Delft Hydraulics Laboratory, Report M169.5-I, Delft, m e Netherlands.
Brahms, A., 1753. Anfangsgriinde der Deich- und Wasserbaukunst, Aurich.
Brewers, H.N.C. and Schukking, W.H.P., 1971. Inititation of Motion of Bed Material (in
dutch) Delft Hydraulics Laboratory, Report S159-I, Delft, The Netherlands.
Carstens, M.R., Neilson, F.M. and Altinbilek, H.D., 1969. Bed F o m Generated in the
Laboratory under an Oscillatory Flow U.S. Corps of Eng., Coastal Eng. Res. Center,
Tech. Memo, No. 28, USA.
Davies, A.G., 1985. Field Observation of the Threshold of Sediment Motion by Wave Action
Sedimentology, Vol. 32, p. 685-704.
Delft Hydraulics, 1972. Systematic Investigation of Two-Dimensional and Three-Dimensional
Scour. Report M648/M863, Delft, The Netherlands.
Delft Hydraulics, 1982. Initiation of Motion and Suspension, Development of Concentration
Profiles in a Steady, Uniform Flow without Initial Sediment Load. Report M1531-111,
Delft, m e Netherlands.
Delft Hydraulics, 1989. Nogat O$&shore

Pipeline, Report on Erodibility tests, Report H1050,
Dingler, J.R., 1974. Wave-Formed Ripples in Nearshore Sands Ph.D. Thesis, University of
California, San Diego, USA.
Diplas, P. and Vigilar, G., 1992. Hydraulic Geometry of Treshold Channeki. Journal of
Hydraulic Engineering, Vol. 118, No. 4.
Dubuat, L.G., 1786. Principes d'Hydrauliques et de Pyrodynamique, Paris Delft Hydraulics,

1992 Scaling Tests for Calcareous Sediments Report H1576, Delft, The Netherlands.
Egiazaroff, P.I., 1965. Calculation of Non-Uniform Sediment Concentration Journal of
Hydraulics Division, ASCE, no. 44.
Engelund, F., 1965. A Criterion for the Occurence of Suspended Load La Houille Blanche,
No. 8, France.
Graf, W.H. and Pa&, G.G., 1977. Les Phenomenes de Deposition et DYErosiondans un
Canal Alluvionnaire. Journal of Hydraulic Research, Vol. 15, No. 2.
Grass, A.J., 1970. Initial Instability of Fine Sand Bed Journal of the Hydraulic Division,
ASCE, Vol. 96, No. HY3.
Hammond, T.M. and Collins, M.B., 1979. On the Threshold of Transport of Sand-Sized
Sediment under the Combined Influence of Unidirectional and Oscillatory Flow
Sedimentology, Vol. 26, p. 795-812.
Hinze, J.O., 1975. Turbulence. McGraw Hill Book Company.
Hjulstrom, I?., 1935. Studies of the Morphological Activity of Rivers as illustrated by the
River Fyris Bulletin, Geological Institute of Upsala, Vol. 25, Upsala, Sweden.
Komar, P.D. and. Miller, M.C., 1975. On the Comparison between the Threshold of
Sediment Motion under Waves and Unidirectional Currents with a Discussion of the
practical Evaluation of the Threshold. Journal Sedimentary Petrol., Vol. 45, p. 362-
367.
Kramer, H., 1932. Modellgeschiebe und Schleppkrsft. Mitt. der Preusz. Versuchanstaltfur
Wmserbau und Schzmau, Berlin, Hefr 9.
Lacey, G., 1930. Stable Channels in Alluvium. Proc. Inst. Civil Engrs., London, Vol. 229,
p. 259-384.
Lane, E.W., 1955. -Design of Stable Channels. Trans. ASCE. Vol. 120, p. 1234-1279.
Larsen, L.H., Stemberg, R.W., Shi, N.C., Marsden, M.A.H. and Thomas, L., 1981.
Field Investigation of the Threshold of Grain Motion by Ocean Waves and Currents.
Sedimentary Dynamics of Continental Shelves, Elsevier, Amrterdam.
Lavelle, J.A. and Morjeld, H.O., 1987. Do Critical Stresses for Incipient Motion and
Erosion Really Exist. Journal of Hydraulic Engineering, Vol. 113, No. 3.
Leiner, O., 1912. Zur ErJorschung der Geschiebe und Sinkstofl Bewegung. Zeitschrift fiir
Bauwesen 62, p. 490-515.
Mantz, P.A., 1977. Incipient Transport of Fine Grains and Flakes by Fluids-mended
Shields Diagram. Journal of the Hydraulics Division, ASCE, Vol. 103, HY8.
Manohar, M., 1955. Mechanics of Bottom Sediment Movement due to Wave Action U.S.
Army Corps Engineers, Beach Erosion Board, Tech. Memo No. 75.
~ a y n o r d S.T.,
, 1978. Practical Riprap Design. USWES, Vicksburg, paper H-78-7, USA.
Miller, M.C., McCave, I.N. and Komar, P.D., 1977. Threshold of Sediment Motion under
Unidirectional Current Sedimentology, Vol. 24, p. 507-527.
Neill, C.R., 1968. A Re-examination of the Beginning of Movement for Coarse Granular Bed
Materials. HRS Wallingford, England.
Neill, C.R. and YALIN, M.S., 1969. Quantitative Definition of Beginning of Bed
Movement. Journal of the Hydraulics Division, ASCE, Vol. 95, HYl.
Olsen, O.J. and Florey, Q.L., 1952. Sedimentation Studies in Open Channels U.S. Bureau
of Reclamation Laboratory Report, No. Sp-34, August 5.
Paintal, A.S., 1971. Concept of Critical Shear Stress in Loose Boundary Open Channels
Journal of Hydraulics Research Vol. 9, No. 1.
Rance, P.J. and Warren, N.F., 1968. The Threshold of Movement of Coarse Material in
Oscillatory Flow. Proc. 11th Conf. Coastal Eng., Vol. I, p. 487-491, London, England.
SchoMitsch, A., 1914. ~ b e Schleppkraft

r und Geschiebebewegung. Leipzig und Berlin, W.
Engelmun.
Schields A., 1936. Anwendung der kjznlichkeitsmechnik und der Turbulenz Forschung aufdie
GeschiebebewegungMitt. der Preuss. Versuchsamst.fir Wasserbau und Schiflau, Heft
26, Berlin, Deutschland.
Silvester, R. and Mogridge, G., 1970. Reach of Waves to the Bed of the Continental Shelf.
Proc. 12th Conf. Coastal Eng., Vol. 11, p. 651-667, Washington, U.S.A.
Sternberg, H., 1875. Untersuchungen iiber das Langen und Querprofil GeschiebefZhrender
Fliise Zeitschrift Bauwesen, vol. 25.
Stevens, M.A. and Nordin, C.F., 1987. Critique on the Regime Theory for Alluvial
Channels. Journal of Hydraulic Engineering, ASCE, Vol. 113, No. 11.
Van Rijn, L.C., 1982. Equivalent Roughness of Alluvial Bed. Journal of the Hydr. Div.,
ASCE, No. HY 10, 1982.
Van Rijn, L.C., 1984. Sediment Transport, Part III: Alluvial Roughness Journal of
Hydraulic Engineering, ASCE, Vol. 110, No. 12.
Van Rijn, L.C., 1985. The @ect of Waves on KaolinitelSand Beds. Report M2060, Delft
Hydraulics, Delft, l%e Netherlands.
Van Rijn, L.C., 1989. Handbook of Sediment Transport by Currents and Waves. Report
H461, Delft Hydraulics, Delft, The Netherlands.
White, C.M., 1940. The Equilibrium of Grains on the Bed of a Stream. Proc. of Royal
Society of London, Series A, No. 958, Vol. 174, pp. 332-338.
Whitehouse, R.J.S. and Hardisty, J., 1988. Experimental Asesment of Two Theories for
the Efect of Bed Slope on the Threshold of Bed Load Transport. Marine geology, Vol.
79, p. 135-139.
Yali., M.S., 1972. Mechanics of Sediment Transport. Pergamon Press.
5. BED FORMS
5.1 Introduction
Bed forms are relief features initiated by the fluid oscillations generated downstream of small
local obstacles over a bottom consisting of movable (alluvial) sediment materials.
Herein, the bed forms are considered from a hydraulic point of view and not from a
sedimentary point of view. This means that the type, shape and the dimensions of the bed
forms are presented; the detailed internal sedimentary processes are not considered.
Information of these latter processes is presented by Allen (1968, 1982).
Bed forms under the following conditions are discussed:

unidirectional currents
non-steady currents (river and tidal flow)
waves
current and waves.
5.2 Bed forms in unidirectional currents
5.2.1 Classification
I . Sand-bed rivers
Many types of bed forms can be observed in nature.
-.When the bed form crest is perpendicular (transverse) to the main flow direction, the bed
forms are called transverse bed forms, such as ripples, dunes and anti-dunes.
Ripples have a length scale smaller than the water depth (see Figure 5.2.3, whereas dunes
have a length scale much larger than the water depth (see Figure 5.2.7). Figure 5 -2.1 shows
ripple and dune-type bed forms as observed in rivers (Simons and Richardson, 1966). The
crest lines of the bed forms may be straight, sinuous, catenary, linguoid or lunate (see Figure
5.2.2).
Ripples and dunes travel downstream by erosion at the upstream face (stoss side) and
deposition at the downstream face (lee side; see Figure 5.2.3). Antidunes travel upstream by
lee side erosion and stoss side deposition (Figure 5.2.3).
Bed forms with their crest parallel to the flow are called longitudinal bed forms such as
ribbons and ridges.
The morphological regimes for unidirectional currents over a sand bed (dm I 2000 pm) can
be classified into (see also Fig. 5.2.1):
lower transport regime with flat bed, ribbons and ridges, ripples, dunes and bars,
transitional regime with washed-out dunes and sand waves,
upper transport regime with flat mobile bed and sand waves (anti-dunes).
Many attemps were made to explain the type of bed forms generated under given flow
conditions by use of (kinematic) instability analysis (Kennedy, 1963; Engelund, 1970). The
instability analysis indicates the conditions at which an initial perturbation of the bed will
grow, but it does not give information of the equilibrium dimensions of the bed forms. The
most reliable classification of bed form types is based on the analysis of bed forms as
observed in flume and field conditions.
In the literature, roughly four groups of classification methods for sand beds are presented.
Engelund (1967) used the Froude number as a classification parameter, while Liu (1957)
described the type of bed forms in terms of a suspension parameter and a particle-related
-
- water surface
-
-
water surface
.............................................................................................................................
.............................................................................................................................
..............................................................
...............................................................
E Plane bad
A Typical ripple p a t t e r n
-F - 1
weak boil
-
-C-A
............................... F Antidune standing waves

6 Dunes and 5uperpO5ed ripples
C Dunes
G Antidune breaking waves
---
D Washed-out dunes o r transition H Chuta and pool
Figure 5.2. I Bed form types in rivers according to Simons and Richardson (1966)
straight sinuous catenary linguoid lunate
Figure 5.2.2 Crest line patterns

flow
/---
------ --
a
'
DUNES IN LOWER REGIME
,----
- flow
deposition
ANTI-DUNES IN UPPER REGIME
Figure 5.2.3 Bed form migration in lower and upper regime

Reynolds number. Simons and Richardson (1966) used the stream power (r,,i) and the
median fall diameter as basic parameters. Van Rijn (1984) and Van den Berg et al. (1989)
used a dimensionless -bed-shear stress parameter and a dimensionless particle parameter to
classify the bed form types.
Herein, the classification diagrams for sand beds of Liu, Simons-Richardson, Van den Berg -
Van Gelder and Van Rijn are presented (see Figs. 5.2.4 to 5.2.7).
Liu (1957) : BF type = F

U*,c
(
~*.cdm
~ . ~ )
Simons-Richardson (1966) : BF type = F(rbSc&d,)
Van den Berg - Van Gelder (1989) : BF type = ~ ( 8 D,)

'~
Van Rijn (1984, 1989) : BF type = F(T, D,)
in which:
u*p = bed-shear velocity related to current
' b,c = overall current-related bed-shear stress

ws = particle fall velocity of bed material
dso, 40 = particle diameters
= kinematic viscosity coefficient
-uv = depth-averaged velocity
= relative density
= water depth
= particle mobility parameter
= bed-shear stress parameter
= grain-related bed-shear stress

= grain-related ChCzy-coefficient
= particle diameter parameter
The diagrams of Liu (1957) and Simons-Richardson (1966) are mainly based on flume data.
Van den Berg - Van Gelder (1989) and Van Rijn (1984, 1989) used flume and field data to
develop their classification diagrams.
According to Van Rijn, the asymmetrical dune-type bed forms with a length scale much
larger than the water depth (A >> h) are the dominant features for T I 15. Mega-ripples
and mini-ripples may be superimposed on the dunes for 3 I T 5 10 and D, I 10. Mega-
ripples are ripples with a length scale of the order of the water depth X = h); mini-ripples
have a length scale of the order of the near-bed turbulence length scale (X a h) . Mini-ripples
are the dominant features for T < 3 and D, < 10.
The upper regime which is defined to occur for T 2 25, is characterized by a dominating
suspended load transport. The characteristic bed forms are nearly symmetrical sand waves
,
with a length scale much larger than the water depth ( A h), which occur in large rivers
at relatively small Froude numbers, (Fr a 0.8), whereas plane bed and sand waves (also
called anti-dunes) are generated in small rivers at relatively high Froude numbers (Fr > 0.8).
The actual onset of the generation of anti-dunes cannot be predicted from Figure 5.2.7,
because this is mainly governed by the interaction of generated free surface waves with the
stream bed (at Froude numbers > 0.8). A small discontinuity in the bed will cause a
stationary surface wave which will increase in amplitude for a Froude number approaching
unity. Fully established anti-dunes have a length scale equal to the length scale of the surface
waves (in phase).
The type of bed forms in the transition regime (15 < T < 25) is somewhat obscure. It may
range from that typical of the lower regime (disappearing dunes) to that typical of the upper
regime (disappearing sand waves) depending mainly on the preceding flow conditions (rising
or falling stage).
Summarizing the following classification is proposed by Van Rijn:
Particle size
Transport regime
1 S D, 5 10 D, > 10
Lower O I T I 3 mini-ripples dunes
3<T<10 mega-ripples and dunes dunes
10 < T I 15 dunes dunes
Transition 15 < T < 25 washed-out dunes, sand waves
Upper T 1 2 5 , Fr < 0.8 (symmetrical) sand waves
T 1 2 5 , Fr 2 0.8 plane bed and/or anti-dunes
Table 5.1 Classification of bed forms according to Van Rijn

2. Gravel-bed rivers
Classification diagrams for gravel-bed rivers are not available. The bed materials of gravel
rivers usually have a broad range of grain sizes from sand particles to large boulders. As a
result selective transport processes and armouring of the bed surface may occur locally. The
most common regime is the lower transport regime; the transition regime (with plane bed)
is a rare event. Very regular bed-features such as mega-ripples and dunes have been observed
in laboratory flumes and small-scale channels (Tsubaki and Shinohara, 1959). Various types
of three-dimensional bar forms (refered to as alternate bars, cresentic bars, transverse bars,
deltaic bars, see Fig. 5.2.8) have been observed in depositional regions of the river
(Richards, 1982).
---D median fall diameter d f (rnrn)
Figure 5.2.4 Bed form class@cation Figure 5.2.5 Bed form classz@cationdiagram
diagram of Liu (1957) of Simons-Richardson (1966)
. Ripples (R)
- Lower stage plane bed /LPl --
10.4 I I I I I I I I t I 1
10.' 2 4 6 8 10' 2 4 6 8 10' 2 4 6 8 10'
Particle parameter, D r
Figure 5.2.6 Bed form classification diagram for currents Van den Berg and Van GeMer
(1989)
a
0.1
- 0.2
plane bod (no motion)

0.3 0.40.5
d f i 0 ~lo3(m) a t
o
1.o
Te = 15'~
2.0
washed -out dunes (transition)

30 4.d
A miniripplos plane bed

A megaripples and dunes I anti -dunes (standing waves)
o dunes anti -dunes (breaking waves)
w e 5.2.7 Bed form classification diagram for currents, Van Run (1984, 1989)
cresen t ic bar transverse bars deltaic bars
Figure 5.2.8 Three-dimensional large-scale bar forms
5.2.2 Shape and dimensions of bed forms

1. Flat bed, lower regime
A flat immobile bed may be observed just before the onset of particle motion, while a flat
mobile bed will be present just beyond the onset of motion. The bed surface before the onset
of motion may also be covered with relict bed forms generated during stages with larger
velocities.
2. Ribbons and ridges, lower regime
Small-scale ribbon and ridge type bed forms parallel to the main flow direction have been
observed in laboratory flumes and small natural channels, especially in case of fine sediments
(d, < 100 pm) and are probably generated by secondary flow phenomena and near-bed
turbulence effects (burst-sweep cycle) in the lower and transition flow regime. These bed
forms are also called parting lineations because of the streamwise ridges and hollows with a
vertical scale equal to about 10 grain diameters and are mostly found in fine sediments (say
50 to 250 pm). The width scale is of the order of 100 vlu,.
3. Ripples, lower regime
When the velocities are somehwat larger (10-20%) than the critical velocity for initiation of
motion and the median particle size is smaller than about 500 pm, small (mini)ripples are
generated at the bed surface. Ripples that are developed during this stage remain small with
a ripple length much smaller than the water depth (Ar<<h).Yalin (1985) report values in the
range of:
The characteristics of mini ripples are commonly assumed to be related to the turbulence
characteristics near the bed (burst-sweep cycle). Current ripples have an asymmetric profile
with a relatively steep downstream face (lee side) and a relatively gentle upstream face (stoss
side). As the velocities near the bed become larger, the ripples become more irregular in
shape, height and spacing yielding strongly three-dimensional ripples (Fig. 5.2.9 A). In that
case the variance of the ripple length and height becomes rather large. These ripples are
called lunate ripples when the ripple front has a concave shape in the current direction (crest
is moving slower than wing tips) and are called linguoid ripples when the ripple front has a
convex shape (crest is moving faster than wing tips), see Fig. 5.2.2. The largest ripples may
have a length upto the water depth and are commonly called mega-ripples. Figure 5.2.10
shows the relative height (A/h) of mega-ripples observed in Pakistan irrigation channels
(Mahmood et al, 1984) as a function of a dimensionless bed-shear stress parameter (T). The
relative height of the mega-ripples varies in the range of 0.02 to 0.06. A clear influence of
the T-parameter cannot be detected. The length of the mega-ripples varies in the range of 0.5
to 1 h.
Herein it is assumed that the ripples will completely disappear for T = 10 (see Fig. 5.2.7).
Tentative expressions for the relative height and length of mega-ripples are given by:
-
*r
= 0.02 (I-eaT) (10-T) for 1 r D, s 10 and 3 r T r 10 (5.2.3)
h
Equation (5.2.3) is shown in Figure 5.2.10.

Analysis of bed material samples has shown that coarser particles are concentrated in the
trough areas of the ripples. Samples of the trough area show larger median particle diameters
and are less sorted than samples from the crest area (Harms, 1969).
0
50 80 70 80 SO 100 110 120 130 140 150 SO 170 180
A. RIPPLES (water dapth s 6 feet1
3, I
0Jo ~b i o i o i o 160 iio i i o 140 40 140 140 GO J.0

B RIPPLES AND DUNES (water depth I 11 hat1
OL 40 i o do 100 iio iio 130 iio 150 a 170 do

C DUNES (water dopth = 5foot)
-
0. WASHED -OUT DUNES (Ironsition) (water dapth
distance (foot1
:8
Figure 5.2.9 Bed forms in Pakistan Irrigation Channels, Mahmood et a1 (1984)

foot
h1
dunes
-----+ bed-shear stress parameter, T
Figure 5.2.10 Relative bed form height, Pakistan irrigation channels
4. Dunes, lower regime
Another typical bed form type of the lower regime is the dune-type bed form. Dunes have
an asymmetrical (triangular) profile with a rather steep leeside and a gentle stoss side (Fig.
5.2.9C). A general feature of dune type bed forms is leeside flow separation resulting in
strong eddy motions downstream of the dune crest.
Experiments were carried out at Delft Hydraulics to get information of the flow field along
a dune (Delft Hydraulics, 1988). Artificial dunes with a height of 0.08 m and a length of 1.6
m were made in a flume (width = 1.5 m, length = 50 m). The leeside slope of the dunes
was 26". A layer of sand grains (d,= 1600 pm, & = 1900 pm) was attached to the
upperside of the dunes. The average water depth was 0.33 m. The overall mean velocity was
0.5 1 m/s. The surface slope was 9.5 lo4. The overall Froude number was 0.29. A Laser-
Doppler velocity meter was used to measure the instantaneous horizontal and vertical
velocity.
Figure 5.2.1 1A ?shows mean horizontal (u) and mean vertical (w) velocities at various
locations along the dune. A recirculation zone can be observed downstream of the dune crest.
Reattachment and acceleration of the flow does occur on the stoss side (luff) side of the dune.
The presence of the flow separation zone can also be observed at the water surface in the
form of large boils rising up to the surface.
Figure 5.2.11B shows the turbulence intensities. Maximum values can be observed in the
recirculation zones.
The length of the dunes is strongly related to the water depth with values in the range of 3
to 15 h. Extremely large dunes with heights of the order of 7 m and lengths of the order of
500 m have been observed in the Rio Parana river at water depths of about 25 m, velocities
of about 2 m/s and bed material sizes of about 300 pm.
The formation of dunes may be caused by large-scale oscillations as described by Yalin
(1972). Due to the presence of large (low frequency) eddies, there will be regions at regular
intervals with decreased and increased bed-shear stresses, resulting in the local deposition and
erosion of sediment particles.
When the bed material is non-uniform, vertical sorting takes places, which means that the
coarser particles accumulate in the dune trough region.
The water temperature was found to have a substantial influence on the dune dimensions
(Colby and Scott, 1965). A drop in water temperature of 20°C (summer to winter) at the
same discharge resulted in a decrease of the dune height and an increase of the dune length
and hence a significant decrease of the friction factor. The sand transport rate increased by
a factor 2 from summer to winter.
The most detailed theoretical analysis of dune migration was given by Fredsnre (1980, 1982)
based on the continuity equation for the sediment
in which:
z, = bed level
q, = sediment transport rate (volume)
p = porosity factor
t = time
The migrating velocity (a) of a dune is given by
in which:
q, = amount of sediment per unit width and time deposited at the dune front
A, = dune height
The magnitude of q, depends on the total sediment transport rate at the dune crest (Q& and
the relative amount of suspended load (q$ and bed load transport (g).
The bed load transport consists of grains which move in almost continuous contact with grains
in the bed, and will be deposited at the front where flow separation occurs. The suspended
sediment grains move above the bed without contact and will not settle immediately. A
suspended grain will contribute to q, and the dune migration if it by settling or diffusion
moves into the separation zone before it is carried past the separation zone by the flow.
Whether the grain is deposited at the front or not will thus depend on its distance from the
bed when it passes the dune crest.
The total sediment transport is expressed as a function of the depth-averaged velocity and the
depth: = f(ii, h).
distance from dune-crest in downstream direction, x (m)
o 0.5 1.0
t--t---l
velocity scale (m/s) mean horizontal velocity (3
distance from dune-crest in downstream direction, x (m)
o
m
0.05 0.1 velocity scale (m/s) mean vertical velocity (4
Figure 5.2.IIa Mean horizontal and vertical velocities along a dune (Delfr Hydraulics,
1988)
----b distance from dune-crest in downstream direction, x (m)
x = horizontal turbulence intensity 4(u )7
0 0.1 0.2 velocity scale (m/s)
M = vertical turbulence intensity
-
o 0.5 1.0
distance from dune-crest
-u9*w' (10-2m2/s2)
in downstream direction, x (m)
A = covariance u'w'
Figure 5.2.11b Turbulence intensities along a dune (Delft Hydraulics, 1988)

-
The local change of the bed level near the crest can be expressed, as follows:
- - a- - - - % -
Geometry : - -% az,
at itx ( 1 - p ) ~dx
~
Since &@x = - ah/&, the dune height A, can be determined from Eqs. (5.2.7), (5.2.8)
yielding:
The sediment transport attains a maximum at the crest of the dune. Immediately after the
crest the flow separates and the dune ends. At situations with dominant bed load transport the
sediment transport is in local equilibrium with the hydraulic conditions, and the point of
maximum transport coincides with the maximum in bed-shear stress (7,). The bed-shear stress
has a local maximum at a distance of about 16Ad downstream of the preceeding crest. At
dominant bed load transport the dune length (h,Jis thus given by:
Fredsare included the effect of gravity in his analysis. The gravity on the moving grains will
tend to retard the bed-load transport moving up the dune, modifying the transport relation
which gives a predicted dune length larger than Eq. (5.2.10). The effect of gravity is
important at low transport intensities.
The suspended load transport cannot adjust immediately to changes in the hydraulic
conditions, as it takes time for the suspended sand grains to settle out from the concentration
profile. The maximum in the suspended load will therefore have a lag distance (L,) behind
the maximum in the bed-shear stress. The dune length in case of both suspended and bed-load
transport is expressed as:
The dune dimensions according to Fredsse (1982) are given in Fig. 5.2.12.
Based on Figure 5.2.12, the maximum dune length for particle sizes smaller than 1000 pm
is approximately 3.7 h, which is rather small compared with measured values.
- O'crest
Figure 5.2.12 Dune dimensions according to Fredsge (1982)
Based on the analysis of flume and field data, Van Rijn (1982, 1984) proposed the following
relationship for the dune height and length:
in which: h = water depth from water surface to mean bed-level (at half the bed form height,
see Fig. 6.2.2).
The dune characteristics according to Eqs. (5.2.12) and (5.2.13) are shown in Figure 5.2.13.
The dune height increases weakly with particle diameter for a given water depth. The dunes
are assumed to be washed out for T 2 25 (see Figure 5.2.7). Equation (5.2.12) yields a
maximum dune height in the range of 0.1 to 0.2 h for T = 5. Figure 5.2.10 shows measured
and computed dune heights for some Pakistan Irrigation channels (Mahmood et al, 1984). The
generation of mini or mega-ripples on the stoss sides of the dunes is also a typical
phenomenon of the lower flow regime. The influence of the water temperature (viscosity) on
the dune dimensions was not found by Van Rijn.
Other relationships for dune dimensions were proposed by:
Ad Ad
Tsubaki-Shinohara (1959) : -
h'
-
I,
= ~(0')
Yalin (1972)
Ranga Raju-Soni (1976) : -

Ad
= F(o/,Fr)
dm
-
Ad
= F(0" d,, Fr, h)
dso
Allen (1968)
Fredsse (1980, 1982) . -

A d Ad -
- - ~(06,d5&
. h' h
dune height
dune length
particle mobility parameter related to grains
grain-related bed-shear stress for currents
water depth
critical bed-shear stress (Shields)
Froude number
median particle diameter
power coefficient of velocity in sediment transport formula
(n = 3 to 6)
dune shape coefficient (a = 0.5 to 0.7)
The results of Tsubaki-Shinohara (1959), Ranga Raju-Soni (1976) and Fredsse (1980, 1982)
were presented in graphical form.
-------,transport parametar. T
------, transport parameter. T
flow velocity flow depth particle size temperature

source
(m/s) h (m) dso ( F m) ('C)
o Guy e t al 0.34- 1.17 0.16-0 22 190 -
8 34
x Guy e t al 0.41 - 0.65 -
0.1 4 0.34 270 8-34
A Guy e t a1 0 4 7 - 1.15 0.16-032 280 8 - 34
b Guy e t al 0.77-0 98 0.16 330 8-34
flume Guy e t al 0.48- 1.00 0.10-0.25 450 8-34
doto Q Guy e t a1 0.53- 1.15 0.12-0.34 930 8-34
Williams 0.54- 1.06 0.15-0.22 1350 -
25 28
/d Delft Hydr 0.45 -0.87 0.26-0 49 790 12-18
6 Stein 0.52 -0.95 0.24-0 31 400 -
20 26
d Znamenskayo 0.53 - 0 80 0 11 -0.21 800 -
Dutch Rivers 0.85 - 1.55 4.4- 9 5 490 - 3600 5 - 20
field ,d Rio Parana 1.O 12.7 400 -
data + Japanese Channels 0.53 - 0 89 0.25- 0 88 1100 - 2300 -
Mississippi River 1.35 - 1.45 6 - 16 -
350 550 -
Figure 5.2.13 Dune dimensions according to Van Rijn (1984)

0.5 1.O
w maan flow valoci ty, Tj (mls)
van Rijn
-
Tsubaki Shinohara
- -
, ,
Yalin
,,
, ,, Rangu Raju Soni -
,, Allan .
, -
o Fradwa, 1980
-,--
o Fradsw, 1982
Figure 5.2.14 Comparison of computed dune heights

Several dune height predictors were used by Van Rijn to compute the dune height for a sand
bed with a median particle diameter of d, = 600 pm (d,= 1500 pm), mean water depths
of h = 1 and 10 m and mean flow velocities of = 0.5, 1.0 and 1.5 m/s.
The results are shown in Figure 5.2.14. As can be observed, the methods of Ranga Raju-Soni
(1976), Tsubaki-Shinohara (1959) and Fredsm (1980) show an increasing trend, the latter two
yielding extremely large values. The method of Yalin produces values which approach to
A, = 116 h. The equation of Van Rijn (1984) and Fredsm (1982) predict a decreasing bed-
form height for increasing velocities (washed-out dunes). For a small depth (h = 1 m) the
methods of Van Rijn and Fredsere (1982) show good agreement, while for a large depth
(h = 10 m) the method of Fredsere yields values which are about twice as large as according
to the method of Van Rijn. The method of Ranga Raju-Soni yields remarkably small bed-form
heights in the case of large flow depths, probably because only flume data were used for
-calibration.
Bars, lower regime

:,3.
The largest bed forms in the lower regime are sand bars (such as alternate bars, side bars,
braid bars and transverse bars), which usually are generated in areas with relatively large
transverse flow components (bends, confluences, expansions). Alternate bars are features with
their crests near alternate banks of the river, see Fig. 5.2.15. Braid bars actually are alluvial
"islands" which separate the anabranches of braided streams. Numerous bars can be observed
distributed over the cross-sections. These bars have a marked streamwise elongation.
,Transverse bars are diagonal shoals of triangular-shaped plan along the bed. One side may
be attached to the channel bank. These type of bars generally are generated in steep slope
channels with a large width-depth ratio. The flow over transverse bars is sinuous (wavy) in
plan. Side bars are bars connected to river banks in a meandering channel. There is no flow
over the bar. The planform is roughly triangular. Special examples of side bars are point bars
and scroll bars.
6. Washed-out dunes and sand waves, transition regime
It is a well-known phenomenon that the bed forms generated at low velocities are washed out
at high velocities. It is not clear, however, whether the disappearance of the bed forms is
accomplished by a decrease of the bed form height, by an increase of the bed form length or
both. Flume experiments (Termes, 1986) with sediment material of about 450 pm show that
the transition from the lower to the upper regime is effectuated by an increase of the bed
form length and a simultaneous decrease of the bed form height, as shown in Figures 5.2.16
and 5.2.17. Ultimately, relatively long and smooth sand waves with a roughness equal to the
grain roughness were generated (see Figure 5.2.17).
-
,In the transition regime the sediment particles will be transported mainly in suspension
(u,/w, 2 5). This will have a strong effect on the bed form shape. The bed forms will
become more symmetrical with relatively gentle lee-side slopes. Flow separation will occur
less frequently and the effective bed roughness will approach to that of a plane bed.
Based on Figure 5.2.7, the transition regime will occur for T 2 15. Julien (1992) has studied
bed form generation in large rivers. Large-scale bed forms with a relative height (A/h) of 0.1
to 0.2 and a relative length (X/h) of 5 to 15 were present in the Mississippi river at T-values
upto 50 (Froude numbers smaller than 0.4). The shape of the bed forms was not reported.
Probably, more or less symmetrical sand waves were generated because the effective bed
roughness was rather small at high velocities (see Chapter 6).
According to the author, the bed forms in the transition regime which will most likely occur,
are washed-out dunes and (symmetrical) sand waves. Both types may occur simultaneously.
bank
through /
,
.-----.s
/?<.
O*;halreg \
bar crest
bar crest'\.
throughN----- \
\
bank
Figure 5.2.15 Alternate bars in a river
T = 9.3
P distance (m)
Figure 5.2.16 Bed form development in the lower and transition regime
(T = 9.3 to 24.9; d,, = 450 pm; maximum bed form height = 0.I5 m;
water depth = 0.3 to 0.35 m; Termes, 1986)
C
bed material q0: 45Oprr

-- 18
C
--I6 2
C
--I2 - Y
--2
A' k,lh
i I I I I 1 I I I I L A 1 I
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 :
- bed-shear stress parametrr .T
Figure 5.2.17 Bed form development in transition regime
1.5 fl I I/ I I -
/
/ tl
- Antidunes / /
Transition -
O /
Flat Bed 0
1.3 - 0,' 0 0 -
0
- b 0 0 0 0 -
0 0
Geometric Standard Deviation
Figure Bed f o m in transition and upper regime for non-uniform sediments

1991)
The dimensions of the washed-out dunes are described by Eqs. (5.2.12) and (5.2.13); these
bed forms will fully disappear for T = 25. Hence A, = 0 for T = 25.
The dimensions of the (symmetrical) sand waves are described by:
for T 1 15
As, = 10 h for T 2 15
7. Plane bed and sand waves, upper regime
Two sub regimes can be distinguished in the upper transport regime:

subcritical upper transport regime : T 2 25 and Fr < 0.8
supercritical upper transport regime : T 2 25 and Fr 2 0.8
In the subcritical upper regime the bed forms will be symmetrical sand waves with
dimensions according to Eqs. (5.2.23) and (5.2.24).
In the supercritical upper regime the bed form types will be plane bed andlor anti-dunes. The
latter type of bed forms are sand waves with a nearly symmetrical shape in phase with the
water surface waves.
The anti-dunes do not exist as a continuous train of bed waves, but they gradually build up
locally from a flat bed. Anti-dunes move upstream due to strong lee-side erosion and stoss-
side deposition (Figure 5.2.3). Anti-dunes are bed forms with a length scale of about 10 times
the water depth (A = 10 h).
When the flow velocity further increases, finally a stage with chute and pools may be
generated (Figure 5.2.1).
Winterwerp et al(1992) did experiments in a tilting flume with sand of 120 pm and observed
violent sub and super-critical flow over asymmetrical sand dunes with lenghts between 2 and
5 m and heights between 0.1 and 0.2 m. The length was related to the sand transport rate
(A = 2 t o 3 m f o r F =0.1t00.2andqt=5t020kg/sm; A = 3 t o 5 m f o r F =0.2to
0.4 and q, = 20 to 50 kglsm). Just upstream and downstream of the dune crest the flow was
super-critical with velocities between 1 and 2 mls resulting in heavy erosion. Just downstream
of the trough a hydraulic jump was generated giving a transition to a region with sub-critical
flow where heavy deposition was observed (velocities from 0.5 to 1 mls, water depth =
0.1 m). Downstream of this region the flow was accelerating again towards the crest region
ending in super-critical flow.
This regime can be observed near pipeline outlets of sand-water mixture flows during closure
of channels.
Most data are confined to nearly uniform sediment materials. The sediment transport process
and hence the bed form development in non-uniform materials are different from those in
uniform materials. Selective sediment transport may occur in non-uniform materials at low
velocities resulting in the formation of an armoured layer. Chiew (1991) studied the influence
of the geometric standard deviation of the bed material (0, = d,/d?) on the type of bed
forms in the transition and upper regime. In all experiments the median particle size was
600 pm. The 0,-value was varied from 1.2 to 5.5. Figure 5.2.18 shows the observed bed
form types as a function of the Froude number and the geometric standard deviation. For
uniform material the dunes were fully washed-out at Fr = 0.8 and anti-dunes were generated
at Fr = 1. For non-uniform sediments dune-type bed forms were present upto Fr = 1 and
anti-dunes did not form at Froude numbers upto 1.5 and o, > 2.
5.2.3 E x ~ l e and
s problems
1. A wide open channel has a mean water depth h = 3 m, a mean bed slope I = 1.5 lo4,
a mean velocity ii = 1 m/s, the bed material characteristics are d, = 350 pm, & =
1000 pm, sediment density ps = 2650 kg/m3, fluid density p = 1000 kg/m3, fluid
temperature Te = 20°C ( v = 1.O1 lo4 m2/s).
What type of bed forms are generated?
What are the dimensions of the bed forms?
Bed fonn type
Liu (1957) : 'b,c = pghI = 4.4 N/M2
u*,~ = (ghl)O.' = 0.066 m/s
W, = 0.048 m/s (Fig. 3.2.6)
U*,C/WS = 1.4
u * , dM/v
~ = 23
-. plane bed
Simons-Richardson (1966) : r b ,u
- = 4.4 N / s ~
+ dunes
Van Rijn : C' = 18 log(12h/3&) = 73.4 mH/s
I
b,c
~ = 1.82 N/m2
= p g m ?2
-rb,a = 0.21 N/m2 (Fig. 4.1.5)

1
T = (r,, - rb,J/sb, = 7.7
D* = ( ( s - l ) g / ~ ~d,) ~=
~ 8.8
+ mega-ripples and dunes

Bedfom dimensions
Gill (1971) : Takingn = 4, a = 0.62
Fr = ii/(gh)OS = 0.184
Allen (1968) : A, = 0.086 h 1+19= 0.32 m
A*
- h1.6 = 5.8 m
Van Rijn : Am = 0.02 h (1-e4.1T) (10-T) = 0.074 m
A, = 0.5 h = 1.5 m
Fredsde (1982) : estimate Ad = 0.1 h = 0.3 m
hcrrcn = 3 - 0.15 = 2.85 m

~~
-
/
= 18 log(12h/3&) = 73 mln/s
U-t = (312.85); = 1.05 m/s

I
U*,C& = ( g 0 . 5 /')~ c, = 0.045 1x11s
~~ I
= ( ~ ! , d ~ / ( ( s ds)
- l ) ~= 0.36
Ad = 0.13 h = 0.39 m (Fig. 5.2.12)
A,& = 0.052 (Fig. 5.2.12)
Ad = 7.5 m
2. A wide open channel width h = 3 m, ;= 1.5 mls, I = 3. lo4. Other data as given in
example 1.
What are the type of bed forms and the dimensions of the bed forms according to the
method of Van Rijn?
Bed fonn type : C' = 73.4 m1121s

I
tb , ~ = p g ( G / ~ ' )=
~ 4.1 N/m2
t b,ct = 0.21 N/m2

/
T = (qc - rb,d/rb,= = 18.5
D, = 8.8, Fr = 0.28
+ washed-out dunes and symmetrical sand waves in the
transition regime
Bedfonn dimensions : A, - 0.11 h (%r ( 1 - e 4 T (25-T) = 0.14 m
Both types may occur simultaneously.

3. A wide open channel with h = 3 m, u = 2 mls, I, = 4. lo"; other data as given in
example 1.
What are the type of bed forms and the bed form dimensions according to the method
of Van Rijn?
Bed form type : symmetrical sand waves in the (sub-critical) upper transport
regime.
4. A wide open channel has a water depth of h = 2 m, mean current velocity i = 0.6
mls, the ChCzy-coefficient is C = 63 m112/s.The bed material characteristics are d, =
150 pm, d, = 300 pm.
Other data are: p = 1000 kg/m3, p, = 2650 kg/m3, v = 1 lod mz/s.
What type of bed forms can be expected according to the methods of Liu, Simons-
Richardson and Van Rijn?
What are the dimensions of the ripples?
Solution:
Liu : dunes (close to ripples)

Simons-Richardson : ripples
Van Rijn : mini-ripples (close to dunes)
Ripple dimensions : A, = 0.0075 to 0.03 m
A, = 0.075 to 0.15 m.
5.3 Bed forms in non-steady currents
5.3.1 Non-steady river flow
In river flow the discharge and the water level stage vary as a function of time (rising and
fallhg stages) depending on seasonal and climatological conditions. Small-scale bed forms
like ripples respond rapidly to a new situation but large-scale bed forms like dunes have a less
rapid response and there may be a considerable phase lag between the establishment of the
new flow conditions at time t, and the establishment of the new dune dimensions at time b.
This lag behaviour of dunes was studied by Simons-Richardson (1962), Gee (1973), Allen
(1976), Freds+e (1979), (198 I), Wijbenga-Klaassen (198 I), Tsujimoto-Nakagawa (1983),
Iseya (1984), Fournier (1984), Goodwin (1986), Termes (1986) and Van Rijn (1989).
The frrst laboratory experiments with respect to the non-steady behaviour of bed forms were
carried out by Simons and Richardson (1962).
Allen (1976) developed a stochastic model describing the creation and destruction of the
dunes. After a dune has traveled a certain assigned distance it is destroyed and a new dune
is created. At the moment of creation the dune dimensions correspond to the prevailing flow
conditions assuming steady flow. During the life of the dunes the dune height can adjust to
changes in the flow conditions, but the dune length remains constant.
Gee (1973) and Goodwin (1986) performed flume experiments to study the non-steady effects
of dune transformation.
Fredsde (1979), (1982) proposed analytical relationships for the initial change of the dune
height which is based on flow and sediment transport parameters at the dune crest. The dune
shape is essentially constant.
Wijbenga and Klaassen (1981) also performed flume experiments to study the changes of
dune dimensions for unsteady flow conditions (sudden increase and decrease of discharge and
depth). Fig. 5.3.1 shows changes in ChCzy-coefficient, dune height and dune length for a
sudden change in water depth.
Comparison of the experimental results and the model results of Allen and Freds4e did not
give satisfactory agreement.
Tsujimoto and Nakagawa (1983) performed flume experiments and proposed semi-empirical
relationships for the change in time of the dune height and the dune length. Various cases
were studied: dune development from an initially flat bed under a constant discharge, dune
transition under a suddenly decreasing discharge and dune transition under a gradually
varying discharge.
Iseya (1984) performed flume experiments related to the development of dunes from an
initially flat bed. The bed material size was 570 pm. Flume runs with varying and constant
discharges were carried out. Figure 5.3.2 shows the variation of the dune characteristics for
a gradual change of the discharge. During the fall of the discharge curve the dune length was
still increasing, because the run was started from a flat bed. Thus, at the start of the
experiment, the bed form dimensions were not in equilibrium with the prevailing flow
conditions.
Fournier (1984) also performed flume experiments and proposed expressions for the dune
height change and the transition time scale. According to Fournier, the dune length first
adjusts itself in a relatively short period to the new flow conditions, while the dune height
- remains essentially constant. After this period ("coalescence" time) the growth in dune height
becomes significant.
Levey et al (1980) measured bed level changes in a bend of the Congaree river during
extreme and average discharges. The bed material was in the range of 250 to 500 pm. The
height.of the bed forms (with wave lengths from 5 to 30 m) were largest during maximum
discharge. The adjustment period to return to a lower bed form height at an average discharge
was about six days.
According to Van Rijn (1989), the transition period T,, during which the dune dimensions
change from those of stage 1 to those of stage 2, is related to the ratio of the change in cross-
sectional area of the dune and the average bed-load transport in the transition period.
INCREASE OF WATER DEPTH
DECREASE OF WATER DEPTH
watardapth (m) Chaty coafticiant (m%

31
1: 28
27
26
25
-
24
23
dune haight (m)

0.12
tim (hours) 22
21
0 1 2
duna kngth (m)

-
3 4 5 8 7
time (hours)
8 9 1 0
1a:
a00
a08
407
am
005
0 1 2 3 4 5 6 7 d Q l O 0 1 2 3 4 5 6 7 8 9 1 0
dtime (hOurs) +tima (hours)
Figure 5.3.1 Variation of C h k q comcient, dune height and dune length for a sudden
change in water depth (Wijbenga and Klaassen, 1983)
upper: water depth changes from 0.2 to 0.4 m
lower: water depth changes from 0.4 to 0.2 m
Iv
I
I II I
I 111
I
- . , %
0
E 89
E 400-
G300-
a
3200- I%~P*%
e (3)
8 loo- ".&w*+l *'.& 4"
i 0:
,
,o-0 O . 0 o., 0-80, .
.@..,.. 0
Elapsed Run Time , h~
Figure 5.3.2 Variation of dune characteristics for a grudual change of discharge

(Iseya, 1984)
Thus,
Assuming a first order adjustment process, the changes in dune height and dune length can
be expressed as:
in which:
A, A = dune height, length at time t
A1, A1 = equilibrium dune height, length at stage 1
A,, X, = equilibrium dune height, length at stage 2
cg,,, %2 = equilibrium bed-load transport rates at stage 1, 2
P = porosity of bed material
Td = dune transition period
a, P, y = coefficients
The experimental results (T25, T26) of ~ijben~a- laa ass en (1981) were used to determine the
a, p and y -coefficients. The tests were performed in a large scale flume (length = 100 m,
width = 1.5 m, depth = 1 m) with bed material of 770 pm. The dune height and length were
measured as a function of time after a sudden increase of the discharge. The water depth was
changed from h = 0.2 m to h = 0.4 m.
Applying the measured values, Van Rijn (1989) found that a = 4, P =3 and y = 1.
A value of p = 1 was also found by Fournier (1984).
The above given expressions can be applied to obtain an order of magnitude estimate of the
dune transition parameters in unsteady river flow. The equilibrium dune dimensions are given
by Eqs. (5.2.12) and (5.2.13).
Finally, the detailed literature review of Wijbenga (1990) is mentioned. Based on his work,
it can be concluded that:
the time lag is larger for larger bed forms,
the time lag of the bed form length is larger than that of the bed form height,
the time lag depends on the variation of the discharge with time (shape of the hydrograph),
the time lag is relatively small during increasing discharge (rising stage) and relatively
large during decreasing discharge (falling stage),
the time lag is most importantly related to variations in flow velocity; variations in grain
size and water depth are less important,
the time scale of bed form changes can be represented by the change of the bed form
volume divided by the sediment transport rate.
5.3.2 Tidal flow
Bed form behaviour in tidal flows in estuaries in Germany was studied by Nasner (1974,
1976, 1978, 1983), Stehr (1975); in the Netherlands by Terwindt (1970a), by Bucx and
Tobias (1986) and by Voogt et al (1991); and in the USA by Salsman et al (1966).
1. Germany
A very detailed study of bed form behaviour in the tidal Elbe and Weser rivers in Germany
was made by Nasner (1974, 1976, 1978, 1983). Echo soundings of the bed were carried out
at four locations in the lower Weser river, at one location in the lower Elbe river (near
Hamburg) and at one location in the outer Elbe river (near Scharhorn). The characteristic data
are presented in Table 5.2.
Most of the bed forms in the Elbe and Weser river are asymmetrical and are herein classified
as sand dunes (see also Fig. 5.3.3). The length of the sand dunes shows much less variation
than the height of the sand dunes (see Table 5.2). The height in the lower Weser river and
the lower Elbe river was found to be related to the fresh water discharge. Maximum sand
dune heights (about 2.4 m) were observed at high river discharges (1200 m3/s), see Fig.
5.3.4. The extreme height and length of the sand dunes were found to be equal to
aproximately twice the mean values. The oberserved net migration rates (mlday) were:
Weser : 0.1 - 0.3 mlday in ebb direction at low river discharge
1.0 - 1.5 mlday in ebb direction at high river discharge
Elbe (Hamburg) : 0.05 - 0.1 m/day in flood direction at low river discharge
0.1 - 0.3 mlday in ebb direction at high river discharge
Elbe (ScharMrn) : 0.05 - 0.3 rnlday in flood direction.
Analysis of velocity measurements above the crests and the troughs of the bed forms showed
a depth-averaged velocity above the crest of about 1 m/s during all conditions (independent
of the river discharge). Thus, the sand dunes are eroded at high discharges until the depth-
averaged velocity above the crest is about 1 mls. The sand dune heights increase at low
discharge until the depth-averaged velocity above the crest is again about
1 m/s.
The effect of dredging on the regeneration of the bed forms was studied in the Elbe river near
Hamburg (Nasner , 1983). The vertical irregularity of the bed after dredging was about
0.5 m. After about one week the original sand dune dimensions (height = 1.2 m, length =
25 m) were observed.
Stehr (1975) also observed asymmetrical sand dunes with a length of about 45 m, a height
of about 3 m in the Elbe river. The water depths were about 15 m. Particle sizes were in the
range of 300 to 500 pm.
Peak flood velocity Peak ebb velocity

Particle sizes (m/s) @Is)
Location dm d, Low river High river Low river High river
discharge discharge discharge discharge
Olm) Orm)
Weser 1 480 1020 0.8-1.0 0.6-0.8 0.9-1.1 1.0-1.2
Weser 2 490 1080 0.8-1.0 0.6-0.8 1.0-1.2 1.1-1.3
Weser 3 350 500 0.7-0.9 0.5-0.7 0.9-1.1 1.0-1.2
Weser 4 450 760 0.7-0.9 0.5-0.7 0.9-1.1 1.0-1.2
Elbe (near 420 810 1.0-1.2 0.8-0.9 0.6-0.8 0.9-1.2
Hamburg)
Elbe (near 320 620 0.7 - 1.1 0.6 - 0.7
Scharhorn)
Mean depth h Mean bed form height Mean bed form length
Location (m) (m)
(m)
minimum maximum minimum maximum
Weser 1 11.1 1.1 2.6 49 63
Weser 2 11.1 1.2 2.8 47 63
Weser 3 10.9 0.7 2.0 30 50
Weser 4 11.6 1.3 2.3 38 57
Elbe (near 14.4 0.8 1.2 20 24
Hamburg)
Elbe (near 22 2.8 3.3 109 119
Scharhorn)
Table 5.2 Bed form dimensions of Weser and Elbe river, Nasner (1974)
5.29
Chart
Datum Nominal Depth
SKN -8.OL m
Sound~ngo f SKN
Figure 5.3.3 Echo sounding recordings of the Weser river bed, Nasner (1978)
Figure 5.3.4 Mean bed form height in relation to the discharge (QJ of the Weser river,
Nasner (1978)
2. The Netherlands
Terwind (1970a) studied bed forms in tidal channels and in tidal fluvial channels (lower Rhine
river) in The Netherlands. The bed material sizes were in the range of 200 to 500 pm. Mega-
ripples, sand dunes and sand waves were observed in sandy areas; no bed forms were
observed in areas with a mud content larger than 15%. The bed form dimensions varied in
the range:
Tidal channels : A = 0.3 to 1.5 m, X = 7 to 4 0 m

Fluvial tidal channels : A = 0.3 to 1 m, X = 7 to 20 m
The bed form height showed the largest variability (0.5 m per day).
The height of the bed forms in the tidal fluvial channels was related to the local water depth;
the relative height was in the range A/h = 0.1 to 0.15. Bed forms with a height smaller than
1 m showed asymmetry changes with the turn of the tidal currents. Bed forms with a height
larger than 1 m did not show this behaviour.
Bucx and Tobias (1986) have analyzed a series of echo soundings carried out over a length
of 20 km in the eastern (landward) part of the Westerschelde estuary in the Netherlands. The
total track length of the bed forms in the various ebb and flood channels was about 100 km.
The water depths in that part of the estuary were in the range of 7 to 25 m. The tidal range
was 4 to 5 m. The current velocities were in the range of 1 to 1.5 mls. The ebb currents
were slightly dominating in most channels. Vertical salinity stratification did not occur. The
size of the bed material was in the range of 200 to
300 pm.
Analysis of the echo sounding tracks shows the presence of mega-ripples in most channels.
The heights of the mega-ripples were in the range of 0.2 to 1 m with a mean value of about
0.6 m. The lengths were mostly in the range of 1 to 20 m with a mean value of about 10 m.
About 20% of the bed forms had a length larger than 20 m. Large asymmetrical sand dunes
with a length of 100 m and a height of 2 m were present in some deep channels (depthE
25 m). Most (80%) of the mega-ripples were slightly asymmetrical in the direction of the
dominating current. About 10%of the mega-ripples had a symmetrical shape and about 10%
of the bed forms were more or less plane (steep slope areas, dredging areas, muddy areas).
Voogt et al (1991) observed asymmetrical mega-ripples with a height of about 0.5 m and a
length of about 10 m in water depths of about 10 m, bed material sizes of about 250 pm and
mean velocities up to 1.75 mls. These asymmetrical mega-ripples change into symmetrical
sand waves with a reduced height of about 0.4 m and an increased length of 30 m for
increasing velocities up to 2.5 mls. The tidal period was too short to generate a fully plane
bed even at the highest velocity of 2.5 mls.
3. USA
Salsman et al (1966) observed mega-ripples with a height of about 0.5 m and a length of
about 20 m in a tidal bay (St. Andrews Bay, Florida) with depths of about 10 m. The median
bed material size was about 135 pm. The tidal velocities were in the range of 0.2 to 0.4 mls.
Net migration velocities of 0.01 to 0.1 mlday were observed.
Summarizing, the bed forms most frequently found in tidal currents in estuaries are mega-
ripples, asymmetrical sand dunes and symmetrical sand waves. Tidal currents may have an
asymmetrical variation depending on the channel and shoal system. Mega-ripples have a
height of the order of 0.5 to 1 m and a length of the order of the water depth (5 to 20 m).
Generally, they are generated in the main flood and ebb channels with bed material sizes
smaller than 300 pm. The lee slope of the mega-ripples is approximately equal to the angle
of repose (avalanche angle). The shape is typically asymmetrical in the direction of the main
current. Thus, reorientation takes place with the reversal of the tide. The sand dunes and sand
waves have lengths of several times the water depths. These lengths, however, are smaller
than those of the sand dunes in steady river flow because the tidal period is not large enough
for the generation of equilibrium values. The largest bed forms do occur in areas where the
river discharge is dominating the tidal discharge. Generally, the sand dunes and sand waves
are found in the deeper wider tidal channels with bed material sizes larger than about 300
pm. Sand dunes are typically asymmetrical in the direction of the dominating current velocity.
Sand waves are typically symmetrical with their crests approximately midway between the
troughs. The lee side angles are much smaller than the angle of repose and seldom exceed
10". This means that flow separation will not occur yielding small effective roughness values.
Sand waves are typically found in areas with symmetrical tidal currents generating an equal
amount of sediments transported backward and forward over the crest on each phase of the
tide.
The dimensions of the tidal bed forms in relation to water depths (h) are, as follows:
mega-ripples : A = 0.05 to0.1 h and A = 0.5 t o 2 h

in shallow channels
(d,, 5 300 pm)
sand dunes, sand waves : A = 0.1 to 0.2 h and A = 3 to 6 h

in deeper channels
(d,, = 300 - 600 pm)
1. A wide open channel has a water depth h, = 3 m, mean current velocity u,

= 1 mls.
The bed material characteristics are d,, = 600 pm, d, = 1000 pm, p = porosity = 0.4.
The bed is covered with dunes. The bed load transport is q,,, = 0.1 kglsm.
A flood wave is passing, giving a new hydraulic situation with h2 = 4.5 m, 4 =
1.5 m/s and q,,, = 0.3 kglsm. Other data are: p = 1000 kg/m3, p, = 2650 kg/m3,
v = 1 m2/s.
What is the dune transition period T,?

What is the dune length after 25 hours during the new situation?
Solution: T, = 40 hours
A, = 28 m after 25 hours.
2. A tidal channel in an estuary has a depth of 5 m near the landward boundary and a depth
of 15 m near the seaward boundary. The maximum current velocity is 1.2 m/s. The tidal
range is 4 m. The bed material characteristics are d,, = 350 pm, d, = 600 pm.
What types of bed forms may exist and what are the bed form dimensions?
Solution: Mega-ripples near landward boundary : A = 0.25 to 0.5 m,

A = 2.5 to 10 m
Sand dunes and waves near seaward boundary : A, = 1.5 to 3 m,
A, = 45 to 90 m
5.4 Bed forms in waves
Two typical regimes can be observed in nature:

lower regime with flat immobile bed, ripples and bars,
upper regime with flat mobile bed (sheet flow).
A typical transition regime does not occur.
Figure 5.4.1 shows a classification diagram given by Allen (1982) which is based on 648 data
sets.
Ripples occur where the near-bed peak orbital velocity is about 1.2 times the critical velocity
for initiation of motion.
Ripples are washed out when the parameter v = is larger than about 200
to 250 (Dingler-Inman, 1976; Horikawa et al, is refered to as the sheet
flow regime.
According to Wilson (1989), the sheet flow regime is present for
Thus,
Ripple regime for g < 250 or 8 < 0.8
Sheet flow regime for \Ir 2 250 or 8 2 0.8
Surf zone bars may be generated near the breaker line and typically reflect transport processes
related to breaking waves.

I . Ripples
Wave ripples are formed once the oscillatory motion is of sufficient strength to cause general
movement of the surface particles. The height and length of the ripples grow until a stable
ripple is obtained depending on the prevailing conditions. Wave-generated ripples show an
almost symmetrical and rounded profile. Wave ripples occur in a wide range of
environments: flooded overbank areas of rivers, flood plains, lake bottoms, intertidal flats,
deep-sea bottom, shelf bottoms and nearshore sea bottoms. Wave ripples have been observed
in depths upto 200 m. In deep water wave ripples are probably formed by internal density
currents rather than by surface waves.
Bagnold (1946) defined two types of ripples: two-dimensional ripples related to rolling grains
and three-dimensional ripples related to eddy motions. According to Bagnold, the rolling
grain ripples are stable at velocities smaller than two times that of initiation of motion.
Rolling grain ripples have a small height to length ratio. They are generated during oscillatory
flow at low Reynolds number. This latter type of motion over an initially wavy bed results
in the formation of steady recirculating cells. Close to the bed the flow is from the trough to
the crest. The sediment particles will tend to be transported to the crests and the ripples tend
to grow. As the ripples get steeper, the opposing effect of the gravity force component giving
a downward motion will be more and more important. When fully developed, rolling grain
ripples are generally two-dimensional, regular and have a sinusoidal shape. At larger
velocities the flow is separated from the ripples and strong lee-side eddies are generated
which can sweep the particles from the troughs to the crests and vice versa. Separation starts
when the orbital diameter exceeds the ripple length. At high velocities the sediment particles
are carried away from the bed (suspension). At increasing velocities the ripples will
eventually be washed-out (sheet flow regime).
The time scale of ripple growth from flat bed to equilibrium values can vary from about 1
hour in case of relatively small waves (Davies, 1985) to about 1 minute in case of relatively
large waves (Dingler, 1975). In deeper water where the wave action generally is too weak
to move the sand particles, the bed may consist of relict ripples formed under earlier more
intense wave action.
Sedimrnl parliele d i a m t e r ( m l
NO BED-MATERIAL MOVEMENT
I 1 1 I d
I
0 OOOOI 00002 0-3 00004 0 0 0 0 5 0 0006 00007 00008 00009 0.001 00011 0 0012
Sadimrnt particle diameter ( m l
Figure 5.4.1 Bed fonn classification diagram for waves, Allen (1982)
A conceptual model for wave-formed sedimentary structures was given by Clifton (1976). For
waves of uniform height and period propagating normal to a straight shoreline over a gentle
sloping bottom, Clifton assumes that the bed configuration is related to%, Ao,, T and d,
where 0, = peak orbital velocity at bed, AO, = orbital velocity asymmetry, T = period
and d particle diameter.
The AU, -parameter is a measure of the velocity asymmetry and it applies to fully oscillatory
flow and>o oscillatory flow superimposed by a current.
Thus, AU, = fT_ - oOB + u,. The main reason for wave-induced asymmetry is the shoaling
process. As a wave begins to shoal, the crest elevates and steepens and the trough shallows
and becomes flatter. The volume of water cxried forward (in wave direction) under the crest
must be equal to the volume of water carried backward under the trough. Because the
steepened crest passes a given point in shorter time than the broader trough, the velocity
under the wave crest will be larger than that under the trough. Clifton used second-order
wave theory to estimate AO,. The transition from symmetric to asymmetric ripples will
occur for AO, > 0.05 m/s caused by asymmetric oscillatory motion with or without a
superimposed longshore current.
Mobile asymmetric ripples generally migrate in the direction of their leeside slope. According
to Clifton, three types of ripples can be distinguished depending on the variables &, A, =
peak orbital excursion, and X, = ripple length:
Orbital ripples (2&d, < 1000), which form under short period waves, the length depends
directly on the length of the orbital diameter;
Suborbital ripples (1000 < 2AddH)< 5000) which form under longer period waves, the
length increases with increasing grain size but decreases with increasing orbital diameter;
Anorbital ripples (2Addm > 5000) which form under waves with very large orbital
diameters, the length depends on grain size but is independent on orbital diameter.
Observations in coastal zones with medium to coarse-grained sand (250-750 pm) indicate that
asymmetric bed forms develop in a consistent pattern, as shown in Figure 5.4.2 (Clifton,
1976). The most significant features as:
Ripples become increasingly irregular for increasing energy conditions (longshore bar and
upper shoreface);
Irregular ripples may grade into cross-ripples with increasing wave heights and orbital
asymmetry; cross-ripples consist of two sets of ripples both oriented oblique to the
oscillatory flow, one set tends to be long-crested and the other set is composed of shorter
ripples in the troughs of the longer ripples; cross-ripples are not detectibly related to
longshore currents or to waves approaching from different directions,
Cross-ripples may grade into lunate mega-ripples for increasing orbital asymmetry; these
mega-ripples have a length scale of about 1 m and are common in medium to coarse-
grained sand (250-750 pm) in conditions of intense asymmetric orbital motion
( ~ 0 >, 0.25 m/s) generated by long-period waves.
Shipp (1984) used the classification of Clifton (1976) to describe the bed form types observed
in a single-barred coastal system at Long-Island, New York. The results of Shipp are valid
for fair weather conditions; longshore currents were not present. The most interesting features
are (see also Figure 5.4.2B).
- inactive -bad
symmatric
land
asymmetric
linear ripplas -
irrPgular rippks cross rippks
w
lunate
magarippks f lot bads
A. SEQUENCE OF BED FORMS IN COASTAL ZONE (Clifton, 1976)
INNER SHELF
I SHORE FACE
I
- -10 -,
C
b,
u
.:,:.... . . .
... ... .....
J-1s
6 . TYPE OF BED FORMS IN COASTAL ZONES DURING FAIR WEATHER CONDITIONS
(Shipp, 1984)
Figure 5.4.2 Bed f o m in the coastal zone according to Clifton (1976) and Shipp (1984)
Upper shore face : linear ripples, asymmetric ripples, flat bed (sheet flow)
Longshore trough : linear ripples (A, = 0.7 m, A, = 0.15 m)

Landward slope of bar : cross ripples, irregular ripples and linear ripples (from top to
bottom)
Longshore bar crest : irregular and cross ripples for low-energy conditions lunate
mega-ripples (A, = 0.7 m, A, = 0.15 m) for higher energy
conditions
Seaward slope of bar : cross-ripples and linear ripples
Transitional zone : linear ripples of fine sand (200 pm);
locally coarse-grain deposits (600 pm) forming linear mega-
ripples (A, = 0.7 m, A, = 0.15 m)
ocffshore : linear ripples of fine sand (150-200 pm)
Figure (5.4.3) shows ripple characteristics along the bed (d,= 270 pm), as measured by
Sakakiyama et al (1985) in a large scale wave flume with regular waves.
Case 3-2 shows a rippled bed in the offshore and onshore zone, while a flat bed can be
observed near the breaker point (plunging breakers). In case 3-4 ripples can also be observed
near the breaker point, probably because spilling breakers are present which cannot wash out
the bed forms.
In the literature many equations are available to determine the dimensions of wave-generated
bed forms (ripples). Important contributions were made by Inman (1957), Mogridge and
Kamphuis (1972), Dingler (1975) and Nielsen (1981).
Based on the analysis of laboratory and field data, Nielsen concluded that:
the size and shape of the ripples are influenced by the irregularity of the waves; for
irregular waves .the ripples are shorter and flatter than for regular waves,
field data conform best with laboratory data when the field wave parameters are based on
the significant wave height.
For laboratory conditions Nielsen proposed:
For field conditions Nielsen proposed:
A r = 21 g-1.85
- for I# 2 10
A,
in which:
A = ripple height
= ripple length
4, = peak value of orbital excursion near the bed
U, = peak value of orbital velocity near the bed
I# = mobility parameter = @,)2/((s- l)gdm)
I B.P. (plunging)
1
- X(m) X (4
( a ) CASE 3-2 (Ho=1.05 m. T -6.0 s). (b) C A S E 3 4 ( H o w 1.62 m. T ~ 3 .s l) .
flat waves Ho/Lo=O.O1 9 steep waves Ho/Lo=O.l 0 8
Figure 5.4.3 Ripple characteristl'cs along the bed in a waveflume, SakQkiyama et a1 (1985),
q = ripple height, h = ripple length, do = orbital diameter
Source
DM 4 4 b
Orm) (m/s) ( 4 (m)
Inman (1957) 250-460 0.2-0.4 0.05 -0.15 0.3 -0.9
Dingler (1975) 150-210 0.3-0.7 0.003-0.007 0.07-0.09
Boyd et al (1988) 100-120 0.1-0.5 - 0.06-0.24
Table 5.3 Ripple data for field conditions

Van Rijn (1989) analyzed 140 sets of ripple data for regular waves and 50 sets of ripple data
for irregular waves. The data were selected from laboratory and field experiments with sand
beds consisting of particle sizes in the rage of 100 to 500 pm under non-breaking wave
conditions. The field data are those of Inman (1957) and Dingler (1975), see Table 5.3.
Recent field observations of Boyd et al (1988) are also presented in Table 5.3. Long-crested
(2D) ripples were observed during persistent low-energy conditions; short-crested (3D) ripples
were observed during high-energy conditions. Ripple orientation responded rapidly to changes
in the wave propagation direction. Ripples were observed to migrate both in the direction of
wave propagation (positive) and in the opposite direction (negative). Positive migration
predominated.
Based on the analysis of Van Rijn (1989), the rgple height and length were found to be
related to the peak value of the orbital excursion (A,) and a particle mobility parameter ($) ,
as follows:
in which:
Figure 5.4.4 shows the ripple height for regular and irregular waves. Figure 5.4.5 shows the
ripple steepness (AJXJ . The relationships of Nielsen (Eqs. 5.4.1 to 5.4.4) for regular and
irregular waves are also shown.
Comparing the experimental results of regular and irregular waves, it appears that the ripple
height and steepness are smaller for irregular waves than for regular waves when the $ -
parameter is larger than about 25. For these latter conditions the ripple generation is
increasingly dominated by suspended sediment transport processes which have a more
diffusive character resulting in a smoothing of the ripples. This was also reported by Nielsen
(1981). The relationships of Nielsen for irregular waves seem to give a ripple height that is
somewhat too small.for $-values in the range of 20 to 60. Further, it is noted that the field
data of Dingler (1975) are relatively small compared with the laboratory data for $-values
in the range of 40 to 80. The reason for this is not clear. It may be caused by the limited
accuracy of the ripple height measurements in field conditions, especially when the ripple
heights are small. More field experiments are necessary to verify the results of Dingler
(1975). Van Rijn (1989) proposed the following relationships for irregular waves:
for $ I 10
A r = 2.8
- 10'13(250 - $)5 for 10 < $ < 250
A,
-
Ar = o for $ 2 250
A8
Ar
- = 0.18 for $ I10
r
A r = 2.10"(250 -
- 4 r 1 2 a 5 for 10 < $ < 250
r
-
A, = o for I# 2 250
4
The upper regime with sheet flow conditions is assumed to be present for $ 2 250.
DATA REGULAR WAVES,
-----+
mobility paramatar f 62 ~ ( ( -1
s ) g d so)
DATA IRREGULAR WAVES

0.8
=
4"
4
0.7 x
o
Ribbarink at al, 1987 (210 pm)
Niauwjaar fzt al, 1987 ( 2 0 0 pm)
Van Rijn, 1987 (190 p m )
-
b 0.6 v Inman, 1957 (260 4 6 0 pm)
2 o Dinglar, 1975 (130 210,pm) -
-L
.P
-----+ mobility paramatar f 6 2 I(( s -1 ) g dso)

Figure 5.4.4 Ripple height in waves, Van Rijn (1989)
-------+ mobility parameter iJa2 /Us-1 )gdSo)
Figure 5.4.5 Ripple length in waves, Van Rijn (1989)
From Eqs. (5.4.6) and (5.4.7, it can be derived that:
A, = 1.22-A, for 9 i 10 (5.4.8)
which is close to the value A, = 1.3 Ab reported by Dingler and Inman (1976), Miller and
Komar (1980) and Nielsen (1981).
The proposed expressions are valid for non-breaking wave conditions. In case of breaking
wave conditions the mobility parameter ( 9 ) will, in gene&, be larger than 250 yielding sheet
flow over a flat bed. In spilling breaking waves this may be realistic. However, in plunging
breaking waves the interaction of the waves with the bed is so vigorously that a rather
irregular bed surface may be generated.
2. Sheet cflow regime
Based on Figure 5.4.4, the sheet flow regime with a plane mobile bed will occur for
dd
9 = O ( ( S - ~ ) ~ > 250. According to Wilson (1989), sheet flow conditions will be
generated for 0 = rJ((p, - p)g dN) > 0.8.
Generally, sheet flow conditions are assumed to be present in the surf zone where breaking
waves are dominant. Kroon and Van Rijn (1989), however, did not observe a fully plane bed
in the surf zone. The bed was always irregular with bumps and holes of the order of 0.02 m
high and about 1 m long.
3. Surf zone bars or longshore bars
These type of bars have their orientation (crests) parallel to the coastline and are found in the
surf zone near the breakerline (Figs. 5.4.2 and 5.4.3). The basic mechanism may be the
generation of net onshore-directed velocities seaward of the breakerline and net offshore-
directed velocities (undertow) in the surf zone. In case of high-energy coasts consisting of
fine sediment material (200-300 pm) two or more parallel bars are generated, while no bars
are generated in case of low-energy coasts of relatively coarse sediment (> 500 pm).
Longshore variations in bar shape and dimensions may be affected by the interaction of swell
waves and edge waves yielding beach cusps and wing-type bars connected to the beach in
case of high-energy reflective beaches.
1. A coastal sea'-has a water depth of h = 5 m. Irregular waves with a peak period of

T, = 7 s are present. The bed material characteristics are d, = 300 pm, & = 500 pm.
Other data are: p = 1025 kg/m3, p, = 2650 kg/m3, v = 1 lo4 m2/s.
What is the significant wave height at initiation of sheet flow?
@612
Initiation of sheet flow: = 250
(s-l)g d,
Wave length is L, = 46 m (h = 5 m, Tp = 7 s)
Thus, H, = 1.8 m
2. A coastal sea has a water depth of h = 5 m. The (irregular) wave characteristics are
H, = 1 m, Tp = 7 s. Ripples are present. Other data: see example 1.
What are the ripple dimensions according to the m e t h d of Nielsen and Van Rijn?
Nielsen : A, = 0.0044 m X, = 0.088 m
Van Rijn : A, = 0.028 m X, = 0.36 m
5.5 Bed forms in currents and waves
Ripples are the dominating type of bed forms in the nearshore and surf zone. Two general
types of currents exist: cross-shore return flow and longshore currents. Tidal currents are also
important, especially in periods with calm weather.
The cross-shore return flows refer to a general seaward flow or to a channelized narrow
- seaward rip current. General seaward flow is most significant close to the shore under
breaking wave conditions and accounts for seaward-facing bed forms. In rip currents which
are opposing the waves, seaward-facing megaripples have been commonly observed (Clifton,
1976).
Longshore currents refer to the current in the zone between the longshore bar and the
shoreline. They feed the rip currents at breaks in the longshore bar. Usually, the wave
propagation direction is almost normal to the current direction. Little is known of the type
of bed forms which are generated under these conditions. Some information is given by
Nielsen (1983), who reports the presence of wave-generated ripples with their crest lines
parallel to the shore in the littoral drift zone, showing no influence of the (weak) longshore
current of about 0.3 mls. Shipp (1984) reports the generation of cross-ripples and irregular
3D-ripples in the surf zone (see Fig. 5.4.2). Complex ripple patterns (sometimes called
interference ripples) are found in areas where wave and currents cross at a certain angle, rip
currents with waves or cross-waves reflections with longshore currents.
Bed forms in the offshore zone are generated by tidal currents superimposed by waves. The
currents may be following, opposing or oblique to the waves. Bed forms in tidal seas are
related to the peak current velocities, water depth, sediment diameter and the availability of
- sediment.
The bed forms in combined current and wave conditions fall into four categories:
symmetrical wave-induced ripples with their crest almost perpendicular to the wave
direction in case of weak tidal current velocities (see Fig. 5.5.2).
asymmetrical current-induced ripples and large symmetrical sand waves with their crest
perpendicular to the tidal current direction in case of a strong tidal current and weak orbital
velocities (see Fig. 5.5 -3).
wave-current ripples in a honeycomb pattern in case of equal strength of the current and
peak orbital velocities (see Fig. 5.5.2).
longitudinal furrows, ribbons, ridges and banks with their crests and troughs almost parallel
to the peak tidal current direction (see Figs. 5.5.4 and 5.5.5).
The ripple data of Nieuwjaar-Van der Kaaij (1987), Nap-Van Kampen (1988) and Havinga
(1992) yield the following results (see Fig. 5 S.2):
I
symmetrical wave-induced ripples (2D, 2 %D) for U,,~IU*,~ 2 0.25
I
wave-current ripples (honeycomb pattern, 3D) for 0.25 < U,,,/U,,~ < 2
I
asymmetrical current-induced ripples (3D) for U*,~IU*,~
2 2
According to Amos and Collins (1978), who analyzed field measurements in the intertidal
zone of a sand flat, ripples can be classified as follows:
wave-dominated ripples for U,/U,,> 10
current-dominated ripples for 6,/~*,~
< 1
in which: u,,, = overall current-related bed-shear velocity.

The bed forms generated by combined currents and waves bear some features of both
hydraulic effects. Where the wave component dominates, the bed forms are similar to fully
developed wave-related bed forms. As the current component gains in strength, the bed forms
become more asymmetrical and larger in height and length, especially in case of an opposing
current. The influence of the waves is that the bed form crest will become more rounded.
Based on the available data, Van Rijn proposes a classification diagram for bed forms under
combined current and wave conditions, which is shown in Fig. 5.5.1. The basic parameters
are the current-related and the wave-related mobility parameters defined as follows:
in which:
I
u,, = (0.125 f,)1 0.5 u
'
= current-related effective bed-shear velocity
I
uas = (0.25 f;)O.' U, = wave-related effective bed-shear velocity
:f = 0.2~g(12h/3d,d]" = current-related friction factor

I
f, = wave-related friction factor
= exd-6+5.2(~k~3d,.J~.'~]
= depth-averaged velocity
= peak orbital velocity at bed based on relative wave
period
= peak orbital excursion at bed based on relative wave
period (0, = o A,)
= water depth
= particle diamaters
= relative density
1' I
/
0
4
0
I I IIiive
8 9
wave 1 wave 'flat'
-
0
0 no motion
I I
ripples(2D)
I 1 I l l
-ipples(3D
I I I
(sheet
I
flow)
I I Y
i-3 2 3 4 5 6 8,o-Z 2 3 4 5 6 8i0-' 2 3 4 5 6 8iO0 2 3 4 5 6 810'

wave-related mobility, (~',,~)*/(s-l)g d50
o angle waves-current = 0' (following)

0 angle waves-curren t = 90' (perpendicular)
a angle waves-current = 180' (opposing)
20 regular long-crested bed forms in wove direction or in current direction

2.50 regular short-crested bed forms in wave direction or in current direction
3D irregular short-crested bed forms in wave direction in current direction or
in both directions
Figure 5.5.1 Bed form classflcation for current and waves, Van Rijn
The generation of ripples (length smaller than the water depth) in flumes with following and
opposing waves with respect to the current was studied by Harms (1969), Tanaka and Shuto
(1984), Nieuwjaar-Van der Kaaij (1987) and Nap-Van Kampen (1988). The wave-induced
ripples are approximately symmetrical, but become almost immediately asymmetrical when
a current is generated in the flume.
Generally, the ripples in a current opposing the waves are less asymmetrical than in a current
following the waves. This is caused by the fact that there is a relatively strong (wavy-indyced)
reduction of the near-bed current velocities in case of an opposing current. For U , , ~ / U , , >
1.5 i 2 the asymmetry of the ripples is about equal to that for ripples in a current alone.
A considerable increase (50%) of the ripple length was observed when a current was
generated in the flume, especially in case of a current opposing the waves. The ripple height
showed a minor increase (20%) when a current was generated.
The generation of ripples in case of current and waves under an angle was studied by
Havinga (1992). The observed ripple patterns are shown in Fig. 5.5.2. The current-induced
ripple height was about 0.02 to 0.03 h. The wave-induced ripple length (0.2 to 0.3 h) was
about twice as large as the wave-induced ripple length. The ripple height and length were not
noticeably affected by the wave-current angle (60°, 90" and 120").
Wave-current ripples with lengths between 0.1 and 1 m have been observed on intertidal flats
(Reineck and Wunderlich, 1968). These types of ripples have more rounded crests than
current-ripples. The ripple steepness lies between those of wave ripples ( a 0.15 to 0.2) and
those of current ripples ( = 0.05 to 0.15).
-
Current-induced mega-ripples are found to be the dominant features in the surf zone. They
are also generated as secondary features on the back of sand waves in the offshore zone.
Their shape is asymmetrical with a steep lee side slope and they respond rapidly to local
currents and waves. They are generated easily, but they are also easily washed out at higher
velocities.
2. Transverse sand waves
Field observations in shelf areas indicate that the presence of symmetrical and asymmetrical
sand waves (Fig. 5.5.3) with their crest perpendicular to the main current direction
(transverse) is related to the tidal current conditions. Stride (1982) states that the lower limit
is a peak tidal current of 0.65 rnls for fine to medium sediments (< 500 pm). They are found
in areas with available sediments. They could have grown up from a flat bed with only small-
scale bed forms. Sand waves are restricted to tidal environments occuring in estuaries and in
shelf seas.
At high current velocities the crests of the sand waves are eroded, similarly to that of dunes
in the upper river flow regime. Sand waves can be symmetrical with an almost trochoidal
shape or slightly asymmetrical with a gentle stoss side and a steeper lee side. Symmetrical
sand waves are expected to be found under almost symmetrical tidal flow and are not
expected to migrate substantially. The asymmetrical sand waves are associated with a net
sediment transport rate in the direction of the largest velocity. The sand wave length is in the
range of 3 to 20 times the water depth. The sand wave height is in the range of 0.1 to 0.3
times the water depth. Migration velocities are in the range of 0 to 5 m per year in
(asymmetrical) tidal currents. Often mega-ripples are migrating over the sand waves. Wave-
induced ripples will be present on the mega-ripples during rough weather conditions.
The transient behaviour of sand waves due tidal variations (neap-spring cycle) and storm
events (high waves) is also of interest. The effect of a storm has been observed to cause a
significant decrease of the sand wave height (30%reduction, Langhorne, 1982). The effects
of tidal variations seem to be confined to the crest region of the sand waves (Langhorne,
1982), showing oscillating crest movements (over 2 m) in the neap-spring cycle.
Tobias (1989) analyzed echo sounding data collected in 10 areas near the approach channel
(Eurogeul, North Sea) to the harbour of Rotterdam (period 1975-1985). The water depths (to
mean bed level) were in the range of 20 to 35 m. The peak current velocities of the springtide
were 0.8 m/s (flood) and 0.7 m/s (ebb), which implies an: asymmetry of Ai = 0.1 m/s in
the flood direction. The monthly-mean significant wave height in the area was about H, =
1.4 m during the winter period (November-February) and H, = 1.0 m during the summer
period (May-September). The monthly-maximum significant wave heights were resp. 4 m and
2.8 m. The bed material sizes were in the range of 250 to 500 pm. Analysis of the echo
soundings shows the presence of transverse sand waves with mega-ripples migrating over the
back of the sand waves (see Fig. 5.5.3). The basic data of the sand waves (A > h) and
mega-ripples ( A < h) are given in the following Tables 5.3 and 5.4.
maga-ripples
Relative height Relative length Number of ripples Number of ripples
Location on stoss-side on lee-side
A,& A/h
2 0.012 0.66 12 4
4 0.03 0.95 7 1
9 0.014 0.37 8 4
Table 5.3 Characterisctics of mega-ripples super-imposed on sand waves in North Sea
sand wave data

Location Relative Relative Asymmetry Relative crest Water Propagation
height length width normal depth h velocity c
to current
AP A,/h (m) (mfyear)
1 0.122 9.4 0.76 82 24 -
2 0.260 12.6 0.42 104 20 2.5 m to NE
3 0.131 9.0 0.49 105 22 -
4 0.226 15.9 0.28 278 24 1 m toNE
5 0.103 10.0 0.76 58 30 -
6 0.153 12.4 0.37 65 28 -
7 0.148 15.6 0.36 52 24 -
8 0.128 5.7 0.71 77 34 1 to SW
? 0.124 7.9 0.88 - 30 -
10 0.181 9.5 0.64 26 35 -
7
Table 5.4 Characteristics of sand waves in North Sea

As can be observed, the megaripples are an order of magnitude smaller than the sand waves.
Analysis of bed material samples shows the presence of smaller sizes in the troughs and
larger sizes on the crests of the bed forms. Probably, the smaller particles are suspended by
relatively large velocities at the crest, after which the particles are deposited in the trough
regions (smaller velocities).
Tobias (1989) computed the ratio O,-/L, with u, = peak orbital velocity related
to the maximum significant wave and u,-, = peak current velocity related to the mean tide
and found a value smaller than unity for the summer period and a value larger than unity for
the winter period. This means a current-dominated tiansport mechanism during the summer
and a wave-dominated transport mechanism during the winter period.
Wright (1992) studied sand waves in the southern North S& over a period of 20 years. Most
of the sand waves appeared simply to oscillate back and forth over distances between zero
and 25 metres per year (between zero and 25 % of the average wave length). The heights of
the sand waves changed by only between 1 and 3.5 metres during the 20 year study period
(between 10 and 30% of the sand wave height).
3. Longitudinal furrows, ribbons, ridges and banks
Longitudinal furrows may develop in tidal seas overlying a rocky flat bottom covered with
gravel, pebbles and cobbles and where there are strong currents (1 to 1.5 mls) and a shortage
of mobile sediments. Observations (Belderson et al., 1988) have shown that variable amounts
of coarse sand and fine gravel are transported along the axes of the furrows. With decreasing
velocities these furrows develop through a transition zone into sand ribbons (see Fig. 5.5.4).
The furrows can have a length of the order of 1000 m, a width of the order of 10 m and a
depth of the order of 1 m. The cross-sectional shape of the furrows may be somewhat
asymmetrical. The plan form is slightly sinuous.
Small-scale ribbons (height = 0.1 m) are aligned (parallel) with the main flow direction and
are believed to be generated by secondary currents superimposed on the main flow yielding
a spiral type of fluid motion.
Sand ridges and banks (North Sea, East China Sea) are the largest sedimentary features in
marine conditions with heights of the order of 10 m, widths of the order of 1 km and lengths
of the order of 10 km. Their parallel spacing is of the order of the width (= 1 km).
Generally, sand banks consists of medium to coarse sand (> 500 pm) and are large sources
of sediment. Sand banks are quite stable features. Mega-ripples and sand waves may migrate
over the banks in regions where the velocities are large enough to initiate particle motion.
Closely related to the strength and direction of the currents, sediments are circulating round
and over the bank. The crest axis of the bank deviates typically about 10" to 20" from the
direction of the peak tidal current, which may be caused by Coriolis effects giving an
anticlockwise rotation on the Northern hemisphere (Stride, 1982).
Houbolt (1968) studied the sand riges of the Well bank in the southern bight of the North
Sea. The sand bed of these ridges consists of rather uniform material in the range of 200 to
300 pm. Based on analysis of spatial grain size distributions and analysis of laminations of
cores, a schematic pattern of the sand particle movement was presented by Houbolt (see Fig.
5.5.5). Sand seems to go round the ridge. The entire ridge seems to move slowly towards the
north-east, perpendicular to its long axis and to the tidal currents.
Large tidal sand ridges have been observed in the East China Sea Shelf by Chang-Shu and
Jia Song (1988). These ridges are mainly present in the submerged palaeovalley of the
Changjiang river and the adjacent area. The sand ridges have a length in the range of 10 to
60 km, a width in the range of 2 to 5 km, a spacing of 8 to 14 km and a height of 5 to
20 m. They mainly consist of well-sorted fine sands with abundant debris of marine shells.
The local water depths vary in the range of 50 to 100 m. The presence of (shallow) marine
fossils suggests that the sand ridges probably are developed during the post-glacial sea level
transgression (relict sand ridges) when the water depths were much shallower and the tidal
currents were stronger. As the sea level rose and the tidal currents became weaker, the sand
ridges gradually ceased growing and became relict sand bodies. In the nearshore area they
were later covered by fine-grained sediments of the late Holocene age (buried sand ridges).
The present-day tidal peak currents (= 0.2 mls) are almost parallel to the long axes of the
ridges.
symmetrical asymmetrical wave-curren t

wave- induced current-induced induced ripples
ripples (2.50) ripples (30) (honeycomb pattern, 30)
Figure 5.5.2 Ripple patterns in combined current and wave conditions
record~ng -- July 1975 - February 1976 - - - - - - A p r ~ l1976
Figure 5.5.3 Transverse sand waves with mega-ripples in North Sea (Tobias, 1989)
80cm / sec
Figure 5.5.4 Longitudinal furrows in North Sea (Belderson et al., 1988)
\ Z
Relation of qraln orl*ntotlon

and for8881 s l o w # fobwrrcd fmm
COWS t- t h 8t8.D s l o w or the rrd981
Figure 5.5.5 Schematic pattern of sand particle movement along sand ridge in North Sea
(Houbolt, 1968)
. A coastal sea has a water depth of h = 20 m, the peak flood-current velocity is&
= 0.6 rnls; the peak ebb-current velocity is La = 0.5 m/s. Irregular waves
perpendicular to the flood and ebb current directions are present. The significant wave
height is H, = 1.5 m; the peak period is T, = 8 s. The bed material characteristics are
d, = 300 pm, d, = 600 pm. Other data are: p = 1025 kg/m3, p, = 2650 kg/m3,
v = 1.10" m2/s.
What type of bed forms are present?
What are the bed form dimensions?
The type of bed forms can be determined from Fig. 5/4.6. To ypply this figure, the grain-
related bed-shear velocities have to be computed (u,,, and u,,J .
wave length : h = 20mandTp=8s,yieldsL,=88m
peak orbital velocity
peak orbital excursion : A, = U,(~/2n) = 0.39

I
wave-related friction factor : fw = exp[-6 + 5.2(A,/3Q4'7 = 0.0161
current-related friction factor : f: = 0.24 f l 0 ~ ( 1 2 h / 3 Q ] - =

~ 0.0091
I
bed-shear velocities : u . , = (f:/4)0.5 U, = 0.00194 m/s (waves)
I
u., = (f;/8)OS ii = 0.0202 m/s (flood)
I
u., = (f;/8)0.% = 0.0169 m/s (ebb)
mobility parameters : 0: = 0.081 (waves)
0: = 0.087 (flood)
0: = 0.062 (ebb)
Bed fonn types (Fig. 5.4.6) : 2D waves-ripples superimposed on 3D current ripples

in honey combpattern (see Fig. 5.4.7). Both types
may be superimposed on large-scale sand waves, as
observed by Tobias (1989).
Bed fonn dimensions
2D wave ripples
Van Rijn : A, = 0.07 m , Eq. (5.4.6)

Xr = 0.44 m , Eq. (5.4.7)
3D current-ripples : Tobias : A = 0.02h = 0.4 m

1 = 0.7h = 14 m
sand waves : Tobias : A, = 0.15h = 3 m
A, = 10h = 200 m
2. The water depth in a coastal sea is h = 5 m. The peak tidal ebb and flood velocities are
0.5 mfs. The significant wave height is H, = 1 m. The peak wave period is T, = 8 s.
The bed material characteristics are dm = 250 pm and d, = 500 pm. Other data are:
p = 1025 kg/m3, p, = 2650 kg/m3, v = 1 m2/s
Solution: Bed form type 3D-wave ripples superimposed on 3D-current mega-

:
ripples.
Bed form dimensions : A, = 0.016 m, A, = 0.3 m
A = 0.1 m, A, = 3.5 m
Allen, J.R.L., 1968. Current Ripples. North-Holland Publishing Company, Amsterdam.
Allen, J.R.L., 1976. Computational Models for Dune Time Lag: General Ideas, Dznculties
and Early Results. Sedimentary Geology, Vol. 16, p. 255-279.
Allen, J.R.L., 1982. Sedimentary Structures, Their Character and Physical Basis. Elseviers
Science Publishers, Amsterdam, The Netherlands.
Amos, C.L. and Collins, M.B., 1978. The Combined Effect of Wave Motion and Tidal
Currents on the Morphology of Intertidal Ripple Marks: The Wash, U.K. Journal of
Sedimentary Petrology, Vol. 48, No. 3, p. 849-856.
Bagnold, R.A., 1946. Motion of Waves in Shallow Water. Proc. Royal Society, A 187,
p. 1-15, London.
Belderson, R.H., Wilson, J.B. and Hohne, N.A., 1988. Direct Observation of Longitudinal
Furrows in Gravel and Their Transition with Sand Ribbons of Strongly Tidal Seas. In:
Tide-Influenced Sedimentary Environments and Facies, Edited by P. de Boer, Reidel
Publishing Company.
Bosman, J., 1981. Bed Behaviour and Sand Concentration under Oscillatory Water Motion.
Report M 1695 I, Delft Hydraulics, Delft, The Netherhnds.
Boyd, R., Forbes, B.L. and Heffler, D.E., 1988. Time-sequence Observations of Wave-
fonned Sand Ripples on an Ocean Shor@ace. Sedimentology, Vol. 35, p. 449-464.
Bucx, T.M.M. and-Tobias, F.E., 1986. Bed Morphology in the Eastern Part of the Wester-
schelde Estuary. Dep. Physical Geography, Univ. of Utrecht, The Netherlands.
Carstens, M.R., Neilson, F.M. and Altinbilek, H.D., 1969. Bed F o m generated in the
Laboratory under an Oscillatory Flow. C.E. R. C., Technical Memo 28, USA.
Chang-Shu, Y. and Jia-Song, S., 1988. Tidal Sand Ridges on the East China Sea Shelf. In:
Tide-Inmenced Sedimentary Environments, Edited by P. de Boer, Reidel Publishing
Company.
Chiew, Y.M., 1991, Bed Features in Non Uniform Sediment. Journal of Hydraulic Enginee-
ring, ASCE, Vol. 117, No. 1, p. 116-120.
Clifton, H.E., 1976. Wme-formed Sedimentary Structures - A Conceptual Model Beach and
Nearshore Sedimentation, edited by Davies and Ethington, Econ. Paleontal. Mineral.
Spec. Publ. 24, p. 126-148.
Colby, B.R. and Scott, C.H., 1965. Effects of Water Temperature on the Discharge of Bed
Material. Prof. Paper 462-G, U.S. Geol. Survey, Washington, USA.
Davies, A.G., 1985. Field Observations of the Threshold of Sediment Motion by Wave
Action. Sedimentology, 32, p. 685-704.
Delft Hydraulics, 1988. Turbulence Measurements above Artzjicial Dunes. Report Q789,
Dingler, J.R., 1975. Wave-fonned Ripples in Nearshore Sands. Thesis, Dept. of Oceanogr.,
Univ; Cal., San Diego, USA.
Dingier, J.R. and Inman, B.L., 1976. Wave-fonnedRipples in Nearshore Sands. Proc. 15th
Coastal Eng. Conf., Honolulu, Hawaii.
Dutoit, C.G. and Sleath, J.F.A., 1981. Velocity Measurements Close to Rippled Beak in
Oscillutory Flow. Journal of Fluid Mechanics, Vol. 112, p. 71-96.
Engelund, F. and Hansen, E., 1967. A Monograph on Sediment Transport. Teknish Forlag,
Engelund, F., 1970. Instability of Erodible Beds. Journal Fluid Mechanics, Vol. 42, p. 225-
244.
Fournier, C.P., 1984. Time Development of Reforming Dunes. Thesis, Queen's University,
Kingston, Canuda.
Fredsde, J., 1979. Unsteady Flow in Straight Alluvial Streams: Modification of Individual
Dunes. Journal of Fluid Mechanics, Vol. 91, Part 3, p. 497-512.
Fredsde, J., 1980. The Formation of Dunes. In?. Symp. on River Sedimentation, Beijing,
China.
Fredsde, J., 1981. Unsteady Flow in Straight Alluvial Streams: Transitionfrom Dunes to
Plane Bed. Journal of Fluid Mechanics, Vol. 102, pp. 431-453.
Fredsde, J., 1982. Shape and Dimensions of Stationary Dunes in Rivers. Journal of the
Hydraulics Division, ASCE, HY. 8.
Garde, R.J. and Albertson, M.L., 1959. Sand Waves and Regimes of Flow in Alluvial
Channels. IMR-Congres, Paper 28, Montreal Canada.
Gee, D.M., 1973. Sediment Transport in Non-Steady Flow. Hydr. Eng. Lab., Report HEL
22-3, Berkeley, Univ. of California, USA.
Gill, M.A., 1971. Height of Sand Dunes in Open Channel Flows. Journal of Hydraulics
Division, ASCE, Vol. 97, HY12, pp. 2067-2074.
Goodwin, P., 1986. Sediment Transport in Non-Steady Flow. Hydr. Eng. Lab., Berkeley,
Univ. of California, USA.
Harms, J.C., 1969. Hydraulic Signijicance of Some Sand Ripples. Geological Society of
America Bulletin, Vol. 80, p. 363-396, USA.
Havinga, F., 1992. Sediment Transport in Non-Breaking Waves under an Angle with a
Current. Dep. of Coastal Eng., Delft University of Technology, Delft, The Netherlands.
Hayakawa, N., Tsujimoto, G., and Hashimoto, H., 1983. Velocity D i ~ t r i b ~ oand
n
Suspended Sediment Concentration over Large Scale Ripples. Coastal Engineering in
Japan, Vol. 26, 1983.
REmRENCm (continued)
Houbolt, J., 1968. Recent Sediments in the Southern Bight of the North Sea. Geologie en
Mijnbouw, Vol. 47 (4), pp. 245-273, Delft, The N e t h e r W .
Horikawa, K., Watanabe, A. and Katori, S., 1982. Sediment Transport under Sheet Flow
Conditions. Proc. Coastal Eng. Conf., Vol. II, p. 1335-1352, Cape Town, S. Africa.
Inman, D.L., 1957. Wave-generated Ripples in Nearhsore Sandr. Tech. Memo 100, Beach
Erosion Board, USA.
Iseya, F., 1984. Experimental study of Dune Development. Environmental Research Center,
Univ. of Tsukuba. Ibaraki, 305 Japan.
Julien, P.Y., 1992. Geometry of Bed Forms in Large Rivers. Report Q1386/Q1387, Delft
Kennedy, J.F., 1963. The Mechanics of Dunes and Am'dunes in Erodible-bed Channels.
Journal Fluid Mechanics, Vol. 16, p. 512-544.
Kroon, A. and Van Rijn, L.C., 1989. Space Averaging Aspects of Sediment Concentrations
in the Surf Zone near Groote Keeten, 1987, 1988, The Netherlad. Geopro 89-03,
Univ. of Utrecht, Dep. of Phys. Geography, The Netherlands.
Langhorne, D.N., 1978. ocffshore Engineering and Navigdonul Problem, Relevance of
Sand Wave Research. Inst. of Oceanographic Sciences, Wonnley, Godalming, U.K.
Langhorne, D.N., 1982a. A Study of Dynamics of a Marine Sand Wave. Sedirnentology, 29,
p. 571-594.
Langhorne, D.N., 1982b. The Stability of the Top Metre of the Sea Bed. Inst. Hydrographic
Review, LlX, p. 79-94.
Levey, R.A., Kjerfve, B. and Getzen, R.T., 1980. Comparison of Bed Form Variance
Spectra within a Meander Bed during Flood and average Discharge. Journal of
Sedimentary Petrology, Vol. 50, No. 1, pp. 149-155.
Liu, H.K., 1957. Mechanics of Sediment-Ripple F o m ' o n . Journal of Hydraulics Division,
ASCE, Vol. 83, HY2, p. 1-21.
Mahmood, K., Mehrdad, M.H., Haque, M.I. and Masud Choudri, A., 1984. Bed Form
Data in Acop Canals, 1977-1980. Civ. Eng. Dep., George Washington Univ.,
Washington, USA.
Miller, M.C. and Komar, P.D., 1980. Oscillation Sand Ripples generated by Woratory
,Apparatus. Journal Sedimentary Petrology, 50, p. 173-182.
Mogridge, G.R. and Kamphuis, J.W., 1972. Experiments on Bed Form Generation by
Wave Action. Proc. 13th Coastal Eng. Conf., Vancouver, Canada.
Nap, E. and Van Kampen, A, 1988. Sediment Transport in case of Irregular Non-breaking
Waves with a Current. Delft Techn. Univ., Civ. Eng. Dept., Delfr, The Netherlands.
Nasner, H., 1974. iiber d m Verhalten von Transportkolpern im Tidegebiet. Mitteilungen
Franzius Institut, Hq? 40, T. U. Hannover, Gennany.
Nasner, H., 1976. Time, k g of Dunes for Unsteady Flow Conditions. Proc. Int. Coastal
Eng. Conf., Honolulu, Hawai.
Nasner, H., 1978. Transport Mechanism in Tidal D u w . Proc. Int. Coastal Eng. Conf.,
Hamburg, Gennuny.
Nasner, H., 1983. Dredging of Tidal Dunes. Int. Harbour Conf., Antwerp, Belgium.
Nielsen, P., 1981. Dynamics and Geometry of Wave-generated Ripples. Journal of Geophy-
sical Research, Vol. 86, No. C7, p. 6467-6472.
Nielsen, P., 1983. Entrainment and Distribution of Dzrerent Sand Sizes under Water Waves.
Joumal of Sedimentary Petrology, Vol. 53, No. 2.
Nieuwjaar, M. and Van der Kaaij, Th., 1987. Sediment Concentratl'on and Transport in
case of irregular Non-Breaking Waves with a Current. Delft Tech. Univ., Civ. Eng.
Dep., DelJi; The Netherlands.
Ranga Raju, K.G. and Soni, J.P., 1976. Geometry of Ripples and Dunes in Alluvial
Channels. Journal of Hydraulic Research, Vol. 14, No. 3, The Netherlands.
Reineck, H.E., and Wunderlich, F., 1968. Sedimentology, Vol. 11, p. 99-104.
Ribberink, J. and Van Rijn, L.C., 1987. Influence of Wave-asymmetry and Wave-irregula-
rity on Time and Bed Averaged Sediment Concentrations. Delft Hydraulics, Report
H186-00.1, Delft, The Netherlands.
Richards, K., 1982. Rivers, Form and Process in Alluvial Channels. Methuen, London.
Sakakiyama, T., Shimizu, T., Kajima, R., Saito, S. and Maruyama, K., 1985. Sand
Ripples Generated by Prototype Waves in a large Wave Flume. Coastal Engineering in
Japan, Vol. 28.
Salsman, G.G., Tolbert, W.H. and Villars, R.G., 1966. Sandridge Migration in St.
Andrews Bay, Florida. Marine Geology, 4, no. 1.
Shipp, R.C., 1984. Bed Fonns and Depositional Sedimentary Stnrctures of a Barred Near-
shore System, Eastern Long Island, New York, Marine Geology, Vol. 60, p. 235-259.
Elseviers Science Publ., Amsterdam, The Netherlands.
Simons, D.B. and Richardson, E.V., 1961. Fonns of Bed Roughness in Alluvial Channels.
Journal of Hydraulics Division, ASCE, Vol. 87, HY3, pp. 87-105.
Simons, D.B. and Richardson, E.V., 1962. The Efect of Bed Roughness on Depth
Discharge Relations in Alluvial Channels. Water Supply Paper 1498-E, US Geol.
Survey, Dept of the Interior, Washington, USA.
Simons, D.B., and Richardson, E.V., 1966. Resistance to Flow in Alluvial Channels.
Geological Survey Prof. Paper 422-1, Wmhington, USA.
Steetzel, H.J., 1984. Sediment Suspension in an Oscillating Flow near the Bed (in Dutch).
Delft Univ. of Techn., Coastal Eng., Delft, The Netherlands.
Stehr, E., 1975. Grennchicht-TheoretischeStudie iiber die Gesetze der Strombank und Ruel-
bildung. Hamburger Kiistenforschung, Vol. 34.
Stride, A.H., 1982. Wshore Tidal Sands, Processes and Deposits. Chapman and Hall,
London.
Tanaka, H. and Shuto, N., 1984. Sand Movement due to Wave-Current Combined Motion.
Fourth Congres, Asian and Pacij?c Div., UHR, Chiang Mai, Thailand.
Termes, P.P., 1986. Dimensions of Bed Forms under Steady Flow at High Sediment Trans-
port Rates (in Dutch). Report M2130, Delft Hydraulics, Delft, l%e Netherlands.
Terwindt, J.H.J., 1970a. Observation on submerged Sand Ripples with Heights rangingfrom
30 to 200 cm occuring in Tidal Channels of South West Netherlands. Geologie en
Mijnbow, Vol. 49, no. 6, Delft, The Netherlands.
Terwindt, J.H.J., 1970b. Geometrical Properties of Sand Waves in the Southern Bight of
the North Sea. Marine Geology, Amsterdam, The Netherlands.
Tobias, C.J., 1989. Morphology of Sand Waves in Relation to Current, Sediment and Wave
Data along the Eurogeul, North Sea. Dep. of Physical Geography, Univ. of Utrecht,
The Netherlands.
Tsubaki, T. and Shinohara, K., 1954. On the Characteristics of Sand Waves Formed upon
the Beds of Open Channels and Rivers. Reports of Research Institute for Applied
Mechanics, Vol. VII, No. 25, 1959.
Tsujimoto, T. and Nakagawa, H., 1983. Time Lag Appearing in Unsteady Flow with Sand
Waves. Journal of Hydroscience and Hydraulic Engineering, JSCE, Vol. 1, No. 1, p.
83-95.
Van den Berg, J.H. and Van Gelder, A,, 1989. Scour and Fill Sequences in Flows over
very Fine Sand and Silt. Proc. 4th Int. Conf. on Fluvial Sedimentology, Barcelona,
Spain.
Van Rijn, L.C., 1982. Prediction of Bed Forms, Alluvial Roughness and Sediment Transport.
Delft Hydraulics, S 487 111, Delft, The Netherlands.
Van Rijn, L.C., 1987. Database Sand Concentration Profiles for Currents and Waves. Delfi
Hydraulics, Delft, l%e Netherlands.
Voogt, L., Van Rijn, L.C., and Van de Berg, J.H., 1991. Transport of Fine Sands at High
Velocities. Journal of Hydraulic Engineering, ASCE, Vol. 117, No. 7.
Wijbenga, J.H.A. and Klaassen, G.J., 1981. Changes in Bed Form Dimensions under
Unsteady Flow Conditions in a Straight Flume. Second Int. Conf. on Fluvial Sediments,
Univ. of Keele, U.K.
Wijbenga, J.H.A. and Klaassen, G.J., 1983. Changes in Bed Fomt Dimensions under
Unsteady Flow Conditions. Spec. Publications Int. Assoc. Sediment 6, p. 35-48.
Wijbenga, J.H.A., 1990. Flow Resistance and Bed-fonn Dimensions for Varying Flow
Conditions; Literature Review. Report Q785, Delft Hydraulics, Delft, The Netherlands.
Wilson, K.C., 1989. Friction of Wave-Induced Sheet Flow. Coastal Engineering, 12.
Winterwerp, J.C., Bakker, W.T.,. Mastbergen, D.R. and Van Ram, H., 1992. Hyper-
concentrated Sand-Water Mrxture Flows over Erodible bed. Journal of Hydraulic
Engineering, ASCE, Vol. 118, No. 11.
Wright, P., 1992. Long Tenn Changes to the Positions and Heights of Sand Waves in the
Southern North Sea. In?. Hydrographic Review, Morurco, LXlX(2).
Ya h , M .S., 1972. Mechanics of Sediments Transport. Pergamon Press.

Y h , M.S., 1985. On the Determination of Ripple Geometry. Journal of Hydrulic Enginee-
ring, Vol. 111, No. 8.
6 . EFFECTIVE BED ROUGHNESS
6.1 Introduction
Nikuradse (1932) introduced the concept of an equivalent or effective sand roughness height
(&) to simulate the roughness of arbitrary roughness elements of the bottom boundary. In case
of a movable bed ?misting of sediments the effective bed roughness (kJ main18 consists of
grain roughness &) generated by skin friction forces and of form roughness (k,) generated
by pressure forces acting on the bed forms. The effective bed roughness for a given bed
material size is not constant but depends on the flow conditions. For example, Figure 6.1.1
shows h-values of the Mississippi river determined from measured surface slopes, water
depths and discharges using the ChCzy-equation.
As can be observed, the kg-value strongly decreases for increasing velocities, probably
because the bed forms become more rounded or are washed out at high velocities.
6.2 Current-related bed roughness
6.2.1 Introduction
The fundamental problem of bed roughness prediction is that the bed characteristics (bed
forms) and hence the bed roughness depend on the main flow variables (depth, velocity) and
sediment transport rate (sediment size). These hydraulic variables are, however, in turn
strongly dependent on the bed configuration and its roughness.
Another problem is the almost continuous variation of the discharge during rising and falling
stages. Under these conditions the bed form dimensions and hence the Chezy-coefficient are
not constant but vary with the flow conditions. Basically, the Chtzy-coefficient should be
determined from the full momentum equation in stead of the ChCzy-equation.
6.2.2 Available methods
Assuming an uniform channel with a non-movable plane rough bed, the bed-shear stress is
proportional to the depth-averaged velocity, as follows: r, = m2. When bed forms are
generated, the friction factor and hence the bed-shear stress will increase due to the presence
of form roughness, as shown in Fig. 6.2.1.
Basically, two approaches can be found in the literature to estimate the bed roughness:
methods based on bed-form and grain-related parameters such as bed-form length, height,
steepness and bed-material size,
methods based on integral parameters such as mean depth, mean-velocity and bed-material
size.
The first method is more universal and can also be used to determine the roughness of a
movable bed in non-steady conditions, provided that the bed-form characteristics are known.
6.2.3 Methods based on bed-form parameters

The bed-shear stress (tJ in an alluvial channel can be divided into:
grain-related bed-shear stress (rb,
form-related bed-shear stress (74.
The distribution of the grain-related bed-shear stress (skin friction) along a bed form (dune)
is shown in Figure 6.2.2. The grain-related stress is low in the flow separation zone and high
near the crest of the bed form.
The form-related bed-shear stress is related to the fluid pressure distribution upstream and
downstream of the crest. The flow is accelerating near the crest, requiring a decrease of the
fluid pressure, whereas the flow is decelerating downstream of the crest requiring an increase
of the fluid pressure. Integrated over the bed form length, there is a net horizontal pressure
force (FJ acting on the bed form. The skin-friction force (FJ and the form-related pressure
forces (FJ acting on a bed form are shown schematically in Figure 6.2.3.
The total bed-shear stress (rd can be divided into r: and ,:r as follows:
Total bed-shear stress : zb =

Fsb + Ff a -
Fs+ - m
Ff 1 //
Tb 71,
X X X
+
Grain-related bed-shear stress : r

I
=
Fs
-
X
=
- ( A
A
) r
= - 1p f 1 i 2
A
Form-related bed-shear stress : " - Ff - C, p l u 2 d z = -1 p f f / c 2 (6.2.3)
'b-A-T2 O 8
in which:
F, = skin-friction force
F, = pressure force
X = bed form length
A = bed form height
7, = grain-related shear stress
a = coefficient ( a 5)
C, = drag coefficient
u = local flow velocity
10'
a
6
4
2 x dm: 210-XX) urn, h: 1- 2 m

o dm: 210-300 urn, h : 5 - 10 m
- -- -- - -
I \ 1 n 1 r R - 1
I x 1 - ILL - I I I I I
-E
2
4
1%
z'.
L 4
2
.-J
g.-al ;
\
2
ldb 0.25 0.50 075 1.00 1.25 1.50 1.75 2.00 2.25 2.50
-
depth averaged flow velocity ( m l s )
Figure 6.1.1 Efective bed-roughness of Mississippi river

-
Figure 6.2.1 Bed-shear stress as a function of depth-averaged velocity
14
12
Ressu* :' 6
0a 2
Bed shear
0.1 stress
Nm" 24 T, Nm"
-0
2
-4
0
-6 -0.1
0 0.1 0.2 0.3 04 0.5

metre
Figure 6.2.2 Fluid pressure and shear-stress distribution along a d m , Langhome (1978)
Figure 6.2.3 Skin-frictionforce and fluid pressure force on a dune

From Eqs. (6.2. I), (6.2.2) and (6.2.3) it follows that:
For reasons of simplicity, Van Rijn (1984b, 1989) assumed that the/effectiveroughness height
of Nikuradse (kJ also can be divided in a grain-related part 04;) and a form-related part
&"), as follows:
in which: $ = current-related grain roughness height and qC= current-related form

roughness height.
Assuming hydraulically rough flow, the current-related friction factor is:
fc =
[
8g Cm2= 0.24 log
[:)r
-
The friction factors for the smooth and the transition regime are presented in Chapter 2.
Grain roughness
Grain roughness is the roughness of individual moving or non-moving sediment particles as
present in the toplayers of a natural plane movable or non-movable bed.
Experimental research shows that grain roughness in the lower regime is mainly related to
the largest particles of the top layer of the bed (&). Van Rijn (1982) analyzed about 120 sets
of flume and field data with and without a mobile bed to determine fhe grain roughness. The
median particle sizes were in the range of 130 to 5000 pm. The kcvalues were computed
from the ChCzy-coefficients (Eq. 6.2.7), which were derived from the measured water depths,
depth-averaged velocities and energy gradients. Based on this analysis, the grain roughness
was found to be in the range of 1 to 10 d,of the bed material, as shown in Figure 6.2.4. A
clear influence of the mobility of the particles cannot be observed. Experiments with a
mobile bed do not show a significantly larger grain roughness than those with an immobile
bed. This effect may, however, be masked by the experimental scatter of the data.
Lyn (1991) studied the resistance of flat-bed sediment-laden and clear water flows in a flume.
Bed materials in the range of 150 to 250 pm were used. The mobility parameter (8) was in
the range of 0.5 to 0.7 (lower regime).
Using the experimental results of Lyn (1991), Van Rijn found:
k& = 2 to 3 d, for non-movable (rigid) plane bed,
15, = 3 to 5 d, for movable plane bed.
The k:c-value of a movable plane bed (moving layer of grains) seems to be somewhat larger
than that of a rigid plane'bed.
Aguirre-Pe and Fuentes (1990) found qC-valuesin the range of 1 to 6 d,for mountain rivers
with beds consisting of gravel and cobbles (lower regime).
Similar values for sand and gravel beds in the lower regime were reported by:
Kamphuis (1974) for 540 S d, I 4200 pm: $ = 2.5 d, (6.2.8)

Gladki (1975) for 1000 I d, I20000 pm: = 2.3 $4 (6.2.9)
Hey (1979) for 1000 I d , C 20000pm: $=3.5& (6.2.10)
Mahmood (1971) for 100 S d, I 600 pm: kc= 5.1 $4 (6.2.11)
Wilson (1987, 1988) studied the effective roughness in the upper regime with high
concentrations (8 > 1) by performing experiments in a conduit with a square cross-section
(0.094 x 0.094 m2). Sand particles (d = 700 pm) and nylon particles (d = 3900 pm) were
tested.
Based on analysis of his experimental results, Wilson found: 4' = 5 8 d, with 8 in the
range of 1 to 7. Hydraulic rough flow conditions were assumed to be present by Wilson. This
is not correct. The effect of the sediment particles on the viscosity in the near-bottom region
where the concentrations are high, should be taken into account.
Einstein-Chien (1955) and Winterwerp et al (1990) studied the effect of high concentrations
on the flow velocity profile (in the upper regime). Both research groups did experiments in
a flume with a rigid plane bed coated by a layer of sand particles. Van Rijn used their data
to compute the I5 -value from C = 18 log(12 RJ(< + 3.3 v,,,/u,)). Side wall roughness was
eliminated. The viscosity of the mixture in the near-bed region was,estimated to be 10 times
larger than that of clear water (v, = lo-' m2/s). The ratio pf and d, is plotted as a
function of the mobility parameter 8 in Fig. 6.2.5. The ratio k/d, increases from 3 to about
20 for an increasing 8 -value.
A logical explanation for the increase of the 4'-value in high-concentration flows is the
interaction of the flow with the sediment particles in the near-bed region. The particle
velocities are greatly reduced by collisions with the bed and with each other resulting in
relatively large differences between the local fluid and particle velocity (slip). The associated
fluid drag forces are the driving forces of the particle motions. Conversely, these drag forces
are reducing the fluid velocities, which can be interpreted as a shear effect additional to the
fluid shear.
Based on the available information (lower and upper regime), Van Rijn proposes to use:
k& = 3 d, for 8 < 1 (lower regime) (6.2.12a)
k& = 3 8 d, for 8 2 1 (upper regime) (6.2.12b)
in which:
$ = grain-related effective roughness height of a plane bed
8 = ~ : / ( ( s - l ) ~d& = mobility parameter ( 8 = 8' for plane bed)
u. = go" i./C = bed-shear velocity (u* = u.' for plane bed)
vm = qm/pm = kinematic viscosity coefficient of fluid-sediment mixture (

near-bed region)
- 10" m2/s in
Equation (6.2.12b) can only be sqlved by interaction. First, the 15/-value is estimated, then
C, u, and 0 are computed and $ is computed from Eq. (6.3.12b). This procedure should
be repeated until the computed k, is equal to the estimated k, .
Equation (6.2.12b) shown in Fig. 6.2.5, is in good agreement yith the equation proposed by
Wilson (assuming d, = 2dJ. Equation (6.2.12) predicts &-values which are somewhat
larger than the data points of Einstein-Chien (1955) and Winterwerp et a1 (1990), see Fig.
6.2.5. This conservative estimate is prefered because the data are related to experiments with
rigid beds. A movable bed of loose material will have a somewhat larger roughness because
a perfectly plane bed will not exist in natural conditions. Small irregularities will always be
present resulting in a larger roughness. This was also shown by the experimental results of
Lyn (1991).
The author advises to use a minimum value of = 0.01 m in case of flow in the upper
regime with high concentrations over a plane bed.
Form roughness
The effective form roughness is related to the bed-form height A, the bed-form steepness
(Alh) and the bed-form shape ( y ). The following functional relation-ship is assumed to be
valid:
The most general case is that of a bed consisting of (mega)ripples superimposed on

asymmetrical dunes and symmetrical sand waves (see Chapter 5). Herein it is proposed to
determine the overall form roughness by summation of the individual values, as follows (see
also Van Rijn, 1989):
k; = k,; + gd+
in which:
= overall current-related form roughness
= form-roughness related to ripples
$ = form-roughness related to asymmetrical dunes
kk = form-roughness related to symmetrical sand waves
Ripples are herein (see Chapter 5) defined as bed forms with a length smaller than the water
depth. Based on the analysis of ripple data of Barton-Lin (1955), Ackers (1964), Vanoni-
Brooks (1957), Mahmood et al (1984) from Pakistan irrigation channels, Nieuwjaar-Van der
Kaaij (1987), Nap-Van Kampen (1988) and Havinga (1992); Van Rijn proposed to use the
following relationship (see Fig. 6.2.6):
in which:
A = ripple height
A, = ripple length
y, = ripple presence factor ( y, = 1 for ripples alone, y, = 0.7 for ripples superimposed
on dunes or sand waves).
Figure 6.2.4 Effective grain roughness
Figure 6.2.5 Effectlavegrain roughness in high-concentration

The form roughness G) shows a strong increase for AJk > 0.1, as presented in Figure
6.2.6.
Basically, the ripple presence factor is defined as y, = n A,./&, with )\, = ripple length,
& = dune length (or sand wave length) and n = number of ripples present on one dune or
sand wave.
When the bed is fully covered with ripples, the ?,-parameter is y, = 1. When the ripples are
superimposed on dunes or on sand waves, the region near the crest and the trough of the
dunes usually is free of ripples resulting in a y,-parameter of about 0.7.
Dunes are herein (see Chapter 5) defined as asymmetrical bed forms with a length of about
7 times the water depth. Based on the analysis of dune data, Van Rijn (1984b, 1989)
proposed (Figure 6.2.7):
in which:
A, = dune height
Ad = dune length
y, = form factor (y, = 0.7 for field conditions)
The form factor (y,J expresses the influence of the dune form on the roughness height. Ogink
(1988) performed flume experiments to study the influence of the dune form on the flow
resistance. He found a considerable reduction of the form roughness for mild leeside slopes
because the flow separation effect was less important.
Van %jn (1989) reanalyzed the data of Ogink (1988) to determine the 7,-parameter as a
function of the leeside slope (h,/A).The results are shown in Figure 6.2.8.
Analysis of field data by Ogink (1988) showed that the leeside slopes of river dunes are much
smaller than those of laboratory dunes. The leeside slopes of the river dunes were in the
range of 1:5 to 1:7, yielding y,-values of about 0.7 (see Figure 6.2.8).
Sand waves are herein (see Chapter 5) defined as symmetrical bed forms with a length much
larger than the water depth. The leeside slopes of symmetrical sand waves are relatively mild.
Hence, flow separation will not occur. Therefore, the form roughness of symmetrical sand
waves is assumed to be zero. Thus,
To show the applicability of Eq. (6.2.14), Van Rijn (1989) calculated the overall roughness
of (artificial) dunes superimposed by ripples, as tested by Ogink (1988). Sand particles with
a median size of 780 pm were coated on the bed forms to simulate grain roughness. The
basic bed form qmensions are presented in Fig. 6.2.9. The grain roughness (kJ was
assumed to be & , = 0.001 m. The predicted and computed k,,c-values are given in the
following Table 6.1. The largest deviation is about 60 % for test A, which seems rather large.
However, the predicted and measured ChCzy-coefficient do not differ more than 10%.
x Havinga (1992)
- ripple steepness A , / h ,
100 p m flume
a Nap-Van Kampen (1988) 100 ,urn flume
0 Nieuwjaar-Van der Kaaij (1987) 2 0 0 p m flume
Mahmood (1984) 150-300 p m channel (field)
A Vanoni-Brooks (1 957) 140 pm flume
Barton-Lin (1985) 180 pm flume
0 Ackers (1964) 180 ,urn channel (lab)
Figure 6.2.6 Form-roughness of ripples (y, = I)
-----+ dune steepness, -

A
A
Figure 6.2.7 Form-roughness of dunes (y, = I)
Figure 6.2.8 Form f~ctorof dunes
sand ~ ~ , , 1 0 . 7 8 m
I
Figure 6.2.9 Bed fonn geometry, Ogink (1988)

Test predicted k,c rIleasUred kc
(m) (m)
A 0.018 0.027
B 0.025 0.029
C 0.022 0.025
D 0.024 0.019
E 0.018 0.014
Table 6. I Predicted and measured k,,,-values for tests of Ogink (1988)
Another approach was used by Engelund (1977) for dunes and Vanoni-Hwang (1967) for
ripples and dunes, who related the form friction coefficient (f/3 to the bed form parameters.
Engelund : ffl = 10 - (e-25Ab)

hl
in which:
f = form friction factor of Darcy-Weisbach
A = bed form height
X = bed form length
h = water depth
e = ratio of lee-slope area and total bed-form area
6.2.4 Methods based on integral parameters
Various methods based on integral parameters such as water depth, mean velocity, slope and
bed material characteristics have been proposed:
Einstein and Barbarossa (1953), Engelund and Hansen (1967), Alam and Kennedy (1969),
Smith and McLean (1977), White et al (1979), and Brownlie (1981).
Herein, the methods of Engelund-Hansen (1967) and White et a1 (1979) are presented,
because these two methods gave the best results in an extensive appraisal of available methods
for river flow conditions as reported by White et al (1979).
The method of Brownlie (1981) and Smith and McLean (1977) are also presented, because
of their simplicity.
7t is noted that the depth (h) should be used for wide channels and the hydraulic radius of the
bed (%) for narrow channels (see Appendix C).
Engelund and Hansen (1967)
I N
They defmed r b ,= rbsc with:
+ rbsc
Based on this, it can be derived that:
They assumed that u!,, can be found from the following relationship:
Using ii = c @ I ) ~and
~ u!,, = (gh'I)OS, Eq. (6.2.23) can be expressed as:
Data of movable bed experiments in flumes (FOR Collins data, USA) wefe analyzed by
Engelund and Hansen. Knowing the basic flume data (6, 6, I and d& , the us,$and h' para-
meters can be computed from Eqs. (6.2.23) and (6.2.20) yielding two equations with two
unknown variables. Based on these results, the 0' and 0-parameters were computed and
plotted, as shown in Figure 6.2.10.
Figure 6.2.10 Mobility parameters 0' and 0 according to Engelund and Hansen (19679
Two regimes can be identified, as follows:
Lower regime: 8' = 0.06 + 0.4 e2 for 8 1 0 . 7 (6.2.26)
Upper regime: 8' = 8 for 0.7 < 8 < 1 (6.2.27a)
8' = (0.3 + 0.7 for 8 r 1 (6.2.27b)
Using Figure 6.2.10 or Eqs. (6.2.26), (6.2.27); the ChCzy-coefficient can be determined.
Two cases are possible:
1. compute u,,, and 0

2. determine 8' from Fig. 6.2.10 or Eqs. (6.2.26), (6.2.27)
3. compute h' = (8//8)h
4. compute C from E (6.2.25)
5. compute ii = C(h1)t.5
Case 11: h and ii are known, find C and I
1. guess h' (sftart with h' = 0.33 h)
2. compute u., from Eq. (6.2.23), compute 8'
3. determine 8 from Fig. 6.2.10 or Eqs. (6.2.26), (6.2.27)
4. compute h' = (8//8)h and compare with initial value
5. '
repeat until h = constant
6. compute C from Eq. (6.2.24)
7. compute I from ii = C(hI)O.'
-White et al, 1979
This method is based on the analysis of 1432 sets of flume data and 263 sets of field data of
the lower regime.
Assuming the water depth h, the slope I and the particle diameter d,, are known, the
computational procedure is straightforward (otherwise iterative computations are necessary):
1. Compute particle parameter D,, n, Y,, P
n = 1 - 0.56 log(D,) for 1 r D, < 60
for 1 r D, < 60
n = 0 for D, 2 60
, = 0.17 for D, 1 60
2. Compute bed-shear velocity, u,,,
3. Compute mobility parameter, YQ
4. Compute mobility parameter, Y

,
5. Compute average flow velocity, G from
6. Compute Ch6zy-coefficient, C
Brownlie presented a method to predict the flow depth (h) as a function of the main flow
variables. The method is based on dimensional analysis and data fitting using 344 flume data
and 550 field data in the lower and upper flow regime.
The following equations were proposed:
Lower regime: h = 0.372 ,d Q,0.654 I -om a,0.105 (6.2.36)
Upper regime: h = 0.284 ,d Q,0.625 I -0.a8QS0.08 (6.2.37)
for I 2 0.006 or Fg 2 1.74 I-In
in which:
The Chkzy-coefficient follows from: C = Q
b h15 10J
Smith and McLean (1977)
Based on the analysis of river data, they proposed:
in which:
$,c = effective current-related bed roughness
= tb/((p,-p)g d& = mobility parameter
0= = critical mobility parameter of Shields
:-t = bed-shear stress
d~ = median particle diameter of bed material
Equation (6.2.38) which is only valid in the lower regime (0 < 0.7) with dunes, yields
&,, = d, for 0 s 0, and & , = 400 dm for 0 = 0.6 and 0, = 0.05.
This method is straightforward when h, I and d,are known; otherwise iterative computations
are necessary.
Several deficiences are common to all roughness-predictors:
the effect of the water temperature on the bed configuration is not properly taken into
account;
the effect of the channel pattern on the overall roughness is not explicitly taken into
account,
the effect of the sediment discharge on the overall roughness is not explicitly taken into
account.
Analysis of river data shows that the slope of a particular river does not change substantially
for varying discharges. The slope is mainly imposed by the locaI bed slope. Consequently,
methods which are exclusively based on slope as an input parameter are inherently inferior
to methods which are based on mean velocity as input parameter.
Guidelines that can be provided are:
use several different predictors;
select and analyse data of similar streams (depth, velocity, width, bed material size, slope,
temperature).
6.2.5 Comparison of methods
The methods of Engelund-Hansen, White et al and Van Rijn were compared for a large
amount of flume and field data (Van Rijn, 1984b). The method of Van Rijn was based on the
prediction of the dune height and length. The effective roughness was determined from Eqs.
(6.2.5), (6.2.12) and (6.2.16) neglecting the form roughness reduction parameter (7, = 1).
Ripples were also neglected (A, = 0).
To evaluate the accuracy of the three prediction methods, the percentage of the predicted C-
values in the following error ranges was determined: C- f lo%, C- f 20% and
Cmd f 30%.
The results are presented in the following Table 6.2.
Percentage of predicted C-values within

Flume data (758)
cme-ed f 10% c m df 20% ~~ f 30%
1. Engelund-Hansen 37 % 65 % 75 %
2. Van Rijn 34 % 56 % 71 %
3. White et al 33 % 54 % 66 %
Field data (786)
1. Van Rijn 43 % 74 % 89%
2. White et al 33 % 58 % 79 %
3. Engelund-Hansen 25 % 47 % 62 %
Table 6.2 Comparison of predicted Wzy-co@cients

For flume conditions the method of Engelund-Hansen produces the best results, whereas for
field conditions the method of Van Rijn is superior.
Particularly, the method of Engelund-Hansen produces rather poor results for field conditions,
probably because only flume data were used for calibration. For field data the method of
Engelund-Hansen and White et al yield values which are, on the average, too large.
1. A wide channel with a depth h = 8 m has a bed covered with dunes. Ripples are
superimposed on the dunes. The dune dimensions are Ad = 1 m, )h = 50 m. The ripple
dimensions are 4 = 0.2 m, X, = 3 m. The bed material characteristics are d, = 300
pm, d, = 500 pm.
What is the effective bed roughness and the Ch6zy-coefficient?
Solution:
Grain roughness (lower regime) : = 3 d, = 0.0015 m
Ripple form roughness (take y, = 0.7) : = 20 yr A,(Ar/hJ
Dune form roughness (take yd = 0.7) : k$ = 1.1 y, A ~ -eI-25 Ad&)
Effective bed roughness : k, = + + Cd= 0 . 4 9 m
2. u
A wide channel has a depth of h = 5 m. The depth-averaged velocity is = 1 m/s. The
bed material characteristics are d, = 500 pm, d, = 1000 pm.
Other parameters are: p = 1000 kg/m3, p, = 2650 kg/m3, v = 1 lo6 m2/s.
What is the ChCzy-coefficient and surface slope according to the method of Engelund
and Hansen and the method of Van Rijn?
Solution:
Method of Engelund-Hansen
Guess h' : h' = 0.33 h = 1.67m
Compute u!,, and 0'
Compute 0 from Eq. (6.2.26)

Compute h' : h' = (0//0)h = 1.7m
within 2 % of initial guess,
which is sufficiently accurate
Compute C from Eq. (6.2.24) : C = 44.3 mH/s
Compute surface slope : I = 02/(c2h) = 1.02 10-4
Method of Van Rijn

Compute grain-related ChCzy : C' = 18 log(lZh/3&) = 77.4 mH/s
= pg(ii/C Y = 1.63 N/m2

I
Compute grain-shear stress : r,
Determine critical shear-stress : r, = 0.27 N/m2 (Fig. 4.1.5)

I
Compute T-parameter :T = (rb - T ~ , ~ ) /=T5~ , ~
Compute D,-parameter : D, = dM((s-l)g/v2)1p = 12.6
Type of bed forms : dunes
Compute dune dimensions : A, = 0.11 h(d,/h)0*3(l-e "' (25 -TI
= 0.64 m
A, = 7.3 h = 36.5 m
Compute effective bed roughness : , = 3 &+ 1.1 yd ~ ~ ( 1-2s

- eA d i d )
= 0.178 m (y, = 0.7)
Compute ChCzy-coefficient : C = 18 log(12h&,J = 45.5 mH/s

Compute surface slope : I = 9.7 lo4
Both methods yield similar results.
3. A wide channel has a water depth h = 3 m. The depth-averaged flow velocity is ii =
1 d s . The bed material characteristics are d, = 300 pm, dgO = 500 pm. The fluid
density is p = 1000 kg/m3, the sediment density is p, = 2650 kg/m3. The kinematic
viscosity is v = 1 lo4 m2/s.
What are the bed form dimensions?
What is the effective bed roughness?
Solution: megaripples on dunes
A = 0.08 m, A = 1.5 m
A, = 0.35 m, A, = 21.9 m
k;, = 0.15 m, C = 42.8 mH/s
4. A wide channel has a water depth of h = 2 m. The depth-averaged velocity is = 0.7
mls. The bed material characteristics are d, = 200 pm, dm = 400 pm, p = 1000
kg/m3, p, = 2650 kg/m3, v = 1 lo4 m2/s.
What is the ChCzy-coefficient according to the method of Engelund-Hansen and Van
Rijn?
Solution : Engelund-Hansen : C = 45.6 mH/s
Van Rijn : C = 39.5 mH/s
5. A wide channel has a water depth of h = 2 m. The water surface slope is I = 2 104.
The bed material characteristics are d,, = 400 pm, d, = 500 pm, dgO = 1000 pm.
Other parameters are p = 1000 kg/m3, p, = 2650 kg/m3, v = 1 lo4 m2/s.
What is the Chkzy-coefficient and the depth-averaged velocity according to the methods
of Engelund-Hansen, White et a1 and Smith-McLean?
Solution : Engelund-Hansen : C = 38.4 mH/s, c = 0.77 m/s
White et a1 : C = 39.5 mH/s, c = 0.79 mls
Smith-McLean : C = 39.5 mH/s, c = 0.79 m/s
6. A wide channel has a water depth of h = 2 m. The flow velocity is 3 d s over a plane
bed in the upper regime. The bed material characteristics are d, = 400 pm, duo= 1000
pm. Other data are: p = 1000 kg/m3, p, = 2600 kg/m3, v = 1 lo4 m2/s.
What is the effective bed roughness according to Van Rijn and Wilson?
Solution : Van Rijn : $ == 0.012
Ikc m
Wilson : 0.007 m
6.3 Wave-related bed roughness
6.3.1 Available method

The effective wave-related bed roughness is expressed as:
in, which:
$ = wave-related grain roughness height
kw = wave-related form roughness height
The wave-related friction factor (f,) for rough oscillatory flow is expressed by
Friction factors for the smooth and transition regime are presented in Chapter 2.
Grain roughness
Grain roughness is dominant when the bed is plane or when the peak orbital excursion at the
- bed is smaller than the ripple length (A, < 13.
Based on laboratory experiments with a perfectly flat non-movable bed (d,= 500, 2200,
12200 and 46000 pm), Kamphuis (1975) proposed:
In natural conditions this value may be somewhat larger, because a perfectly arranged flat bed
will not be present.
Grant and Madsen (1982) re-analyzed the movable bed data of Carstens et a1 (1969) and
concluded that the grain roughness for a mobile bed should be related to the thickness of the
moving bed-load layer. Based on data analysis, Grant and Madsen have proposed a function
that yields:
k:w=l~d, for 6'-28,
&:w = 350 d, for 8' = 20 8,
in which:
8' = effective mobility parameter (based on gw= d&
8, = critical mobility parameter
The method of Grant and Madsen yields relatively large &-values, which may be caused
by the fact that ripple regime data have been analyzed by Grant and Madsen. This introduces
the problem of estimating the ripple form roughness. Small errors in this latter parameter may
lead to relatively large errors in the grain roughness.
High grain roughness values were also found by Raudkivi (1988) using the,same data set of
Carstens et al (1969). Raudkivi proposes an expression which yields Ik, = 0.16 m for
= 1 m/s.
According to Van Rijn, the effective grain roughness of a sheet flyw bed h of the order of
the sheet flow layer thickness or the boundary layer thickness (Ig, = 8$. The sheet flow
layer is a high-concentration layer of bed material particles. The moving particles will reduce
the near-bed velocities due to the presence of drag forces (particle velocity smaller than fluid
velocity) which can be interpreted as a shear effect. The apparent or effective roughness of
the sheet flow layer will be of the order of the thickness of the s h ~ e tflow layer. A good
estimate for this is the thickness of the wave boundary layer. Thus kw= 8,. An indication
for this can be obtained from the experimental results of Horikawa et al (1982), who
measured the boundary layer thickness in a sheet flow experiment (8, = 0.02 m for
0, = 1.27 m/s, T = 3.6 s, d, = 200 pm, d, = 300 am). Ysing the boundary layer
thickness equation: 8 - -
~ ,follows that k, 0.015 m. This latter
= 0.072 ( A J I C J ~ it
-
value is somewhat smaller than thy observed boundary layer thickness of 8,
d, 300 pm, it follows that ]lS,w a 50 d, for the Horikawa-experiment.
0.02 m. Since
Wilson (1989) found c-values in the range of 1 to 40 dm for 8 = $,/((p - pJg dM) in the
range of 0.8 to 8. He proposed:
k& = 5 0 dm (6.3.4)
Van Rijn (1989) proposed the following values (see Eq. (6.2.12)):
] 1 S = 3 d, in the ripple regime for 8 < 1
k:W= 3 8 d, in the sheet flow regime for 0 r 1
in which:
8 = Iu**w) = peak mobility parameter

( ~ - l ) gd50
*
u * , ~= 112 f (ob)ZP-(
= wave-related peak bed-shear velocity
I = e x - + 5 . , A,
k,w + 3.3 v,JG*,w
)-1
= friction factor transition regime
0, = near-bed peak orbital velocity
= near-bed peak orbital excursion

= kinematic viscosity of fluid-sediment mixture in near-bed region (v, - 10" m2/s)
The friction factor for the transition regime can also be expressed as:
4
]lS:w + 4.7 v,/(fr 0,)
Equation (6.3.5b) is solved by interation, see Example 1 of section 6i3.2. Equation (6.3.5)
yields a value in the range of 3 d, to 30 d, for 8 = 1 to 10.
A perfectly plane bed will not exist in natural conditions. Small irregu1,arities will always be
present. Therefore, it is recommended to use a minimum value of I g , = 0.01 m for the
sheet flow regime in natural conditions.
Form roughness
Ripples are the dominant bed forms generated by oscillatory flow. Ripples may be present
on a horizontal bed or superimposed on large sand waves. In the latter case the sand waves
have no friction effect on the water waves, because the water waves experience the sand
waves as a gradual bottom topography. When the near-bed orbital excursion is larger than
the ripple length, the ripples are the dominant roughness elements for the oscillatory motion.
Sayao (1982) analy;ed the experimental ripple data of Carstens et al (1969) and Lofquist
(1980) and found \,w-values in the range of 2 to 4 times the ripple height. No influence of
the ripple steepness was noticed (0.1 < A,/& < 0.25). Thus,
Sayao (1982): \yw = 3 A, for A, < 5

2 e -
1r
Equation (6.3.7) is also assumed to be valid for &/Ar 2 5 because eddy generation is
dominant in this range. For &/Ar < 2, Equation (6.3.7) becomes less reliable because of the
reduced importance of the eddy generation mechanism. For &/Ar < 1 the flow along the bed
forms will be non-separating (potential) flow mainly affected by grain roughness (Honji et
al, 1980 and Davies, 1985). Van Rijn (1989) proposed (see Eq. (6.2.15):
in which:
A = ripple height
A, = ripple length
yr = ripple presence factor (y, = 1 for a ripple covered bed, y r = 0.7 foi ripples
superimposed on sand waves)
Equation (6.3.8) yields values in agreement with those of Sayao (1982) for a ripple steepness
in the range of 0.1 to 0.2 (see also Figure 6.2.6).
Other expressions available in the Literature are:
Swart (1976) : cw
=
1r
A
25 A, (2)
A
Grant-Madsen (1984) : = 28 A, (2)
1r
A
Raudkivi (1988) : = 16 Ar (2)
r

1. The bed of a coastal sea consists of sand with dm = 300 pm, dgO= 500 pm. The water
depth is h = 4 m. The significant wave height is H, = 1.6 m. The peak wave period
is T, = 8 s. Other parameters are p = 1025 kg/m3, p, = 2650 kg/m3, v = 1. lo6
in2/s.
What is the effective bed roughness according to the method of Van k j n ?
Solution:
Compute wave length

Compute peak orbital velocity at bed : & = Ir Ha
T sinh(kh)
= 1.14 m/s
Compute peak orbital excursion : A, = (Tp/2x)fi8 = 1.45 m
Compute mobility parameter
Determine type of bed forms : sheet flow

Estimate 15; and v, : gw = 0.005m
v, = 1.10-5m2/s
Compute friction factor, Eq. (6.3.6) : fw = 0.009
Compute Q,, and 0 :

A
u*,w = [O.S fw uap

^2 5
= 0.0764 m/s
0 = Q:,/((s-~)~ d& = 1.25
Compute k:w from Eq. (6.3.5b) : 4:w = 0.0019 m

Repeat procedure : 15; = 0.0025 m
2. Same problem as I., but H, = 2 rn.
What is the effective bed roughness according to Wilson?
Solution: 4
; = 0.0044 m
3. The bed of a coastal sea consists of sand with d, = 250 pm and d, = 500 pm. The
water depth is h = 5 m. The significant wave height is H, = 1 m. The peak wave
period is T, = 8 s. Other parameters are: p, = 2650 kg/m3, p = 1025 kg/m3, v =
1 10" m2/s.
What are the bed form dimensions according to the method of Van Rijn?
What is the effective bed roughness according to the methods of Sayao, Swart, Grant-
Madsen, Raudkivi and Van Rijn?
Solution : Bed forms = ripples
Dimensions : A = 0.016 m, X, = 0.3 m
Bed roughness : &,, = 0.05 m, Sayao
k,, = 0.023 m, Swart
k,, = 0.025 m, Grand-Madsen
k,, = 0.015 m, Raudkivi
6.4 Bed roughness in combined currents and waves
6.4.1 Available method
The most important bed form regimes are:
ripples in case of weak (tidal) currents and low waves,
sand waves with ripples in case of (tidal) currents and low waves,
plane bed with sheet flow in case of strong (tidal) currents and high waves (surf zone),
sand waves with sheet flow in case of strong (tidal) currents and high waves (outside surf
zone).
The k-parameter can be determined by summation of grain roughness and form roughness
(4 = k: + q).
- Grain roughness
Grain roughness is dominant for both the wave-related and current-related friction when the
bed is plane.
When bed forms are present and the peak orbital excursion at the bed is smaller than the bed
form length (A, < 13,the grain roughness is also dominant for the wave-related friction.
In that case the bed forms act as topographic features for the waves.
Equation (6.3.5) is assumed to be valid for the wave motion. Equation (6.2.12) is assumed
to be valid for the current motion.
s Form roughness
When the bed is covered with ripples, the ripple roughness is dominant for the current-related
friction. Ripple roughness is also dominant for the wave-related friction when the peak value
of the orbital excursion at the bed is larger than the ripple length (A, > 13. Nap and Van
Kampen (1988) performed flume experiments with combined currents and waves over a
sediment bed of 100 pm. Ripple type bed forms (0.1 < 4/A, < 0.2) were generated during
the experiments. The effective roughness of the ripples was studied by generating a current
(without waves) over the ripples and measuring the vertical distribution of the vel~itiesand
the water surface slopes. Based on the measured surface slopesl, it was found t h a t b = 3 A ,
whereas analysis of the velocity profile data resulted in k;, = 7Ar. The former result
probably is the most realistic result because it is based on the measured surface slope which
is an integral parameter over the flume length, whereas velocity profile data represent local
data at a particular cross-section.
A clear influence of the fjpple steepness (in the range of 0.1 to 0.2) was not observed.
Havinga (1992) found b-values in the range of 0.5 to 2 4 .
Herein, it is proposed to use Eq. (6.2.15) to determine the ripple form roughness.
Thus:
When sand waves with or without (mega or mini) ripples are present, the large-scale sand
waves act as topographic features for the wave motion because the sand waves have a length
(A) much larger than the orbital excursion at the bed (A&. Thus, the wave-related friction
factor is not determined by the large-scale sand wave dimensions, but by the small-scale
ripples (if present) on the back of the sand waves. When the orbital velocities near the bed
become relatively large (in storm periods), the mega and mini ripples will be washed out
resulting in a plane sloping bed with sheet flow on the back of the sand waves. In this latter
case the wave-related friction factor is determined by grain roughness (Eq. 6.3.5b). The
current-related friction factor is determined by a combination of grain roughness and ripple-
form roughness (if present). The sand waves do not contribute to the current-related
roughness, because they act as topographic features for the current.
Apparent roughness
Basically, the &-parameter expresses the physical roughness height of the bed forms.
The overall roughness experienced by the current (in the presence of waves) may be
considerably larger than the physical bed roughness (k,3,see Eq. (2.4.29).
This roughness increase can be represented as an apparent roughness (Q, which is related
to the effect of the wave boundary layer on the current. The apparent roughness strongly
depends on the relative strength of the wave and current motion. In case of a decreasing wave
- -
-
height and a constant currentpelocity the apparent roughness will approach to the effective
bed roughness k, k ,for U, 0). For an increasing wave height the apparent roughness
will be considerably larger than the physical bed roughness (factor 10 to 100).
Lambrakos et al (1988) performed velocity measurements above a flat (rough) sea bed in the
Strait of Juan de Fuca near Vancouver Island (USA). The water depth was 18 m. The bed
was almost flat with rock and gravel sediments. Velocity measurements were made at 0.24,
0.61, 1.22, 1.83 and 3.6 m above the bed.
Analysis of the velocity data showed a &,,-value of about 0.001 to 0.005 m for current alone
conditions 6 = 0.1 to 0.5 mls). For combined current and wave conditions, the apparent
roughness k, was found to be 0.1 to 0.5 m for 4 = 2 which means an increase of a
factor 100. (u- = root mean square value of the oscillatory velocity).
1. The water depth of a coastal sea is h = 20 m. The peak velocities of the tidal flood and
ebb currents are 0.5 mls. The significant wave height is H, = 3 m. The peak wave
period is T, = 10 s. The angle between the wave and current direction is 90". The bed
is covered with large symmetrical sand waves with A,, = 3 m, b, = 200 m. The crests
of the sand waves are perpendicular to the tidal velocities. The sand waves are fully
covered with three-dimensional wave-ripples, 4 = 0.02 m and X, = 0.3 m. The bed
material characteristics are d, = 300 pm and d, = 1000 pm.
What is the physical current-related bed roughness?
What is the physical wave-related bed roughness?
What is the apparent roughness experienced by the current?
Solution:
Grain roughness (lower regime) : :k = 3 & = 0.003 m
Form roughness of wave ripples : :k = k: = 20 Y, A, (A,/Q
Form roughness of sand waves : k:= 0

Physical current-related bed roughness : , = 0.003 + 0.027 = 0.03 m
Physical wave-related bed roughness : , = 0.003 + 0.027 = 0.03 m
=-=a
Compute peak orbital velocity at bed : 0, = = 0.76 m/s
T* &(kh)
Compute apparent roughness, Eq. (2.4.29) : 4 = kSce 1.6 O&i = 0.34 m
2. Same problem as 1.
The bed is covered with large symmetrical sand waves with A,, = 3 m, &, = 200 m.
The sand waves are covered with mega-ripples, Ch, = 0.5 m, = 15 m. Ten mega-
ripples are present on each sand wave. The width of the mega-ripples is 5 m. The crests
of the sand waves and the mega-ripples are perpendicular to the tidal velocities. The
mega-ripples are fully covered with three-dimensional wave-induced ripples, 4 = 0.02
m, & = 0.3 m.
What is the physical current-related bed roughness?
What is the physical wave-related bed roughness?
What is the apparent roughness experienced by the current?
Solution: &, = 0.28 m
&, = 0.03 m (A, < 5 m)
k, = 3.2 m
REFERENCES
Ackers P., 1964. Experiments on Small Streams in Alluvium. Journal of the Hydraulics
Division, ASCE, Vol. 90, No. HY4.
Aguirre-Pe, J. and Fuentes, R., 1990. Resistance to Flow in Steep Rough Streams. Joumal
of Hydraulic Engineering, ASCE, Vol. 116, N. 11.
Alam, A.M.Z. and K e ~ e d yJ.F.,
, 1969. Friction-facors for Flow in Sand-bed Channels.
Journal of the Hydraulics Division, ASCE, HY6.
Barton, J.R., and Lin, P.N., 1985. A Study of the Sediment Transport in Alluvial Streams.
Civ. Eng. Dep., Colorado College, Rep. No. 55JRB2, Fort Collins, USA.
Brownlie, W.R., 1981. Flow Depth in Sand-Bed Channels. Techn. Memo 81-3, W.H. Keck
Lab., California Institute of Pasadena, USA.
Carstens, M.R., Neilson, F.M. and Altinbilek, H.D., 1969. Bed F o m Generated in the
Laboratory under an Oscillatory Flow. C.E.R. C., Technical Memo 28, USA.
Davies, A.G., 1978. Field Observations of the Threshold of Sediment Motion by Wave
Action. Sedimentology, Vol. 32.
Einstein, H.A. and Barbarossa, N.L., 1953. River Channel Roughness. American Society
of Civil Engineers Transactions, Paper 2528.
Einstein, H.A. and Chien, N., 1955. Effects of Heavy Sediment Concentration near the Bed
on Velocity and Sediment Distribution. Inst. of Eng. Research, Univ. of California,
Berkely, California, USA.
Engelund, F. and Hansen, E., 1967. A Monograph on Sediment Transport. Technisk Forlag,
Engelund, F., 1977. Hydraulic Resistance for Flow over Dunes. Progress Report 44, p. 19-
20, Inst. Hydr. and Hydr. Eng., Techn. Univ. of Denmark, Denmark.
Gladki, H., 1975. Discussions of Determination of Sand Roughness for Fixed Beds. Journal
of Hydraulic Research, Vol. 13, No. 2.
Grant, W.D. and Madsen O.S., 1982. Movable Bed Roughness in U n s t e e Oscillatory
Flow. Journal of Geophysical Research, Vol. 87, No. Cl, p. 469-481.
Havinga, F., 1992. Sediment Concentrations and Transport in case of Irregular Non-breaking
Waves with a Current. Coastal Eng. Dep., Delft Univ. of Techn., Delft, The
Netherlands.
Hey, R.D., 1979. Flow Resistance in Gravel-bed Rivers. Journal of Hydraulic Division,
ASCE, No. HY 4.
Horikawa, K., Watanabe, A. and Katori, S., 1982. Sediment Transport under Sheet Flow
Conditions. Coastal Eng. Conf., Vol. 2, Cape Town, South Africa.
Hoqji, K., Kaneko, A. and Matsunaga, N., 1980. Flow Above Oscillatory Ripples. Sedi-
mentology, Vol. 27.
Kamphuis, J.W., 1974. Detem'nution of Sand Roughness for Fixed Beak. Journal of
Hydraulic Research, Vol. 12, No. 2.
Kamphuis, J. W., 1975. Frictr'on Factor under Oscillatory Waves. Journal of the Waterway,
Port, Coastal and Ocean Div., ASCE, Vol. 101, No. WW2, p. 135-144.
Lambrakos, K.F., Myrhaug, D. and Slaattelid, O.H., 1988. Seabed Current Boundary
Layers in Wave-Plus-Current. Journal of Waterway, Port, Coastal and Ocean
Lofquist, K.E.B., 1980. Measurements of Oscillatory Drag on Sand Ripples. Proc. 17th
Coastal Eng. Conf., Vol. 111, p. 3087-3106, Sydney, Australia.
Lyn, D.A., 1991. Resistance in Flut-Bed Sediment-Laden Flows. Journal of Hydraulic
Engineering, ASCE, Vol. 117, No. I .
Mahmood, K., 1971. Flow in Sand Bed Channels. WaterManagement Technical Report No.
11, Colorado State University, Fort Collins, Colorada, USA.
Mahmood, K., Mehrdad, M.H. and Haque, M.I., 1984. Bed form Data in ACOP-canals
1979-1980. George Washington Univ., Civ. Mech. and Environmental Eng. Dep., U U .
Nap, E. and Van Kampen, A., 1988, Sediment Transport in Irregular Non-Breaking Waves
with a Current. Coastal Eng. Dep., Delfr Univ. of Technology, Delfr, The Netherlunds.
Nikuradse, J., 1932. Gesetzmdssigkeiten der Turbulente Stromung in GIatten Rohren. Ver.
Deut. Ing. Forschungsh& 356.
Ogink, H., 1988. Hydraulic Roughness of Bed F o m . Delfr Hydraulics, Report M2017,
De@, The Netherlunds.
Raudkivi, A.J., 1988. The Roughness Height under Waves. Journal of Hydraulic Research,
Vol. 26, No. 5.
Sayao, O.D.S. F.J,, 1982. Beach Profiles and Littoral Sand Transport. Queen 's University,
Kingston, Ontario, Canada.
Smith, J.D. and McLean, S.R., 1977. Spatialty-averaged Flow over a Wavy Su$zce.
Journal Geophysical Research, Vol. 82, p. 1735-1746.
Swart, D. H., 1976. Predictive Equations regarding Coastal Transports. Coastal Engineering
Congress, Honolulu, Hawaii.
Vanoni, V. and Hwang, L., 1967. Relation between Bed F o m and Friction in Streams.
Journal of the Hydraulics Division, ASCE, HY3.
Vanoni, V.A. and Brooks, N.H., 1957. Laboratory Studies of the Roughness and Suspended
Load of Alluvial Streams. Sedimentation Laboratory, California Ins?. of Tech., Report
E-68, USA.
Van Rijn, L.C., 1982. Equivalent Roughness of Alluvial Bed. Journal of the Hydraulics
Division, ASCE, Vol. 108, No. HYIO.
Van Rijn, L.C., 1984a. Sediment Transport, Part I: Bed Load Transport. Journal of
Van Rijn, L.C., 198413. Sediment Transport, Part III: Alluvial Roughness. Journal of
Hydraulic Engineering, Vol. 110, No. 12.
White, W., Paris, E. and Bettess, R., 1979. A New General Method for Predicting the
Frictional Characteristics of Alluvial Streams. H. R. S. Wallingford, Report No. IT 187,
England.
Wilson, K.C., 1988. Frictr'onal Behaviour of Sheet Flow. Progress Report 67, pp. 11-12,
Tech. Univ. Denmark.
Wilson, K.C., 1989. Friction of Wave-Induced Sheet Flow. Coastal Engineering, 12.
Winterwerp, J.C., De Groot, M.B., Mastbergen, D.R. and Verwoert, H., 1990. Hyper-
concentrated Sand- Water Mixture Flows over Flat Bed. Journal of Hydraulic Enginee-
ring, ASCE, Vol. 116, No. 1.
7. BED MATERIAL SUSPENSION AND TRANSPORT IN STEADY
UNIFORM CURRENTS
7.1 Introduction
The transport of bed material particles by a flow of water can be in the form of bed-load and
suspended load, depending on the size of the bed material particles and the flow conditions.
The suspended load may also contain some wash load (usually, particles smaller than 50 pm),
which is generally defined as that portion of the suspended load which is governed by the
upstream supply rate and not by the composition and properties of the bed material. The wash
load is mainly determined by land surface erosion (rainfall, no vegetation) and not by channel
bed erosion. Although in natural conditions there is no sharp division between the bed-load
transport and the suspended load transport, it is necessary to define a layer with bed-load
transport for mathematical representation. The wash load will not be discussed in this chapter.
The sediment transport in a steady uniform current is assumed to be equal to the transport
capacity defined as the quantity of sediment that can be carried by the flow without net
erosion or deposition, given sufficient availability of bed material (no armour layer).
Usually, three modes of particle motion are distinguished: (1) rolling and sliding motion or
both; (2) saltation motion; and (3) suspended particle motion.
When the value of the bed-shear velocity just exceeds the critical value for initiation of
motion, the particles will be rolling and sliding or both, in continuous contact with the bed.
For increasing values of the bed-shear velocity, the particles will be moving along the bed
by more or less regular jumps, which are called saltations. When the value of the bed-shear
velocity exceeds the fall velocity of the particles, the sediment particles can be lifted to a
level at which the upward turbulent forces will be comparable with or of higher order than
the submerged weight of the particles and as result the particles may go in suspension.
Various types of formulae are available to predict the bed-load and the suspended load
transport. The formulae can be divided in five main groups as defined by the relevant
hydraulic parameter :
fluid velocity, qb = (G: -
bed shear stress, qb E (7 - t , 3
probabilistic particle movement
bed form celerity -
energetics (stream power), qb=tu
The n-coefficient is of the order of 1.5. The m-coefficient is in the range of 3 to 5.
Flume and field data show that the sand transport rate is most strongly related to the depth-
averaged velocity, see Figure 7.1.1. The power of velocity is approximately 3.2 for this
location in the Rio Grande river.
In Chapter 6 it has been shown that a certain value of the overall bed-shear stress (7) can
occur at two values of the mean velocity (6) depending on the presence of bed forms or a
flat bed. This means that the sand transport rate is not a unique function of the bed-shear
stress: Sand transport formulae given in terms of the overall bed-shear stress are therefore
less reliable.
In Section 7.2 the motion and transport of the bed-load particles are presented.
In Section 7.3 the suspended-load transport is described.
1 2 3 4 5 6 7 8 9 1 0
Mean velocity. In feet pcr second
Figure 7.1.1 Bed sediment transport in Rio Grade river near Bemalillo (Nordin, 1964)
7.2 Bed-load transport

7.2.1 Introduction and defmitions
Usually, the transport of particles by rolling, sliding and saltating is called the bed-load
transport. For example, Bagnold (1956) defines the bed-load transport as that in which the
successive contacts of the particles with the bed are strictly limited by the effect of gravity,
while the suspended-load transport is defined as that in which the excess weight of the
particles is supported by random successions of upward impulses imported by turbulent
eddies. Einstein (1950), however, has a somewhat different approach. Einstein defines the
bed-load transport as the transport of sediment particles in a thin layer of 2 particle diameters
thick just above the bed by sliding, rolling and sometimes by making jumps with a
longitudinal distance of a few particle diameters. The bed layer is considered as a layer in
which the mixing due to the turbulence is so small that it cannot influence the sediment
particles, and therefore suspension of particles is impossible in the bed-load layer. Further,
Einstein assumes that the average distance travelled by any bed-load particle (as a series of
successive movements) is a constant distance of 100 particle diameters, independent of the
flow condition, the transport rate and the bed composition. In the view of Einstein, the
saltating particles belong to the suspension mode of transport, because the jump lengths of
saltating particles are considerably larger than a few grain diameters.
Many formulae to predict the bed-load transport rate are described in the Literature. The
earliest formula is that of Du Boys in 1879, who assumed that the sediment particles are
moving along the bottom in layers of progressively decreasing velocities in vertical downward
direction.
The first empirical formula was presented by Meyer-Peter and Mueller (1948). They
performed flume experiments with uniform particles and with particle mixtures. Based on data
analysis, a relatively simple formula was obtained, which is frequently used.
Kalinske (1947) and Einstein (1950) introduced statistical methods to represent the turbulent
behaviour of the flow. Kalinske assumed a normal distribution of the instantaneous fluid
velocity at grain level. Einstein gave a detailed but complicated statistical description of the
particle motion in which the exchange probability of a particle is related to the hydrodynamic
lift force and particle weight. Einstein proposed the d:, as the effective diameter for particle
mixtures and ths &, as the effective diameter for gram roughness.
Frijlink (1952) had a very practical approach and made a simple fit of the formulae of Meyer-
Peter-Mueller and that of Einstein.
Bagnold (1966) introduced an energy concept and related the sediment transport rate to the
work done by the fluid.
Van Rijn (1984) solved the equations of motions of an individual bed-load particle and
computed the saltation characteristics and the particle velocity as a function of the flow
conditions and the particle diameter for plane bed conditions.
Basically, the bed load can be defined as the product of particle concentration, particle
velocity and layer thickness, as follows:
in which:
q, = volumetric bed-load transport rate (m2/s)
c, = volumetric concentration (-)
u, = particle velocity (m/s)
6, = thickness of bed load layer
The bed-load transport can also be defined as the product of the number of moving particles
per unit area, the particle volume and the particle velocity, as follows:
in which:
N, = number of moving particle per unit area (mm2)
V, = particle volume (m3)
u, = particle velocity (mls)
Defining the particle velocity as the ratio of the saltation length and the saltation period
u, = AlT, it follows that:
in which:
E = D = N, Vb/T = eroded or deposited volume of particles per unit area and time (mts).
7.2.2 Characteristics of moving bed-load transport
1. Saltation characteristics
A shown by Bagnold (1954), the most typical motion of the bed-load particles is the saltation
motion (or jumping motion).
From detailed experiments of Francis (1973) and Abbott and Francis (1977), the general
characteristics of particle saltations canbe infered. The saltation mode of transport is confined
to a layer with a maximum thickness of about 10 particle diameters, in which the particle
motion is dominated by gravitational forces, although thezparticlemotion may be initiated by
instantaneous turbulent impulses during upward bursts of fluid or just by the effect of shear
in the sense that a body in sheared flow experiences a lift force due to the velocity gradient
near the bed. The particles receive their momentum directly from the flow pressure and
viscous skin friction. On the rising part of the trajectory, both the vertical component of the
fluid drag force and the gravitational force are directed downwards. During the falling part
of the trajectory, the vertical component of the fluid drag force opposes the gravitational
force. The lift force is always directed upwards as long as the particle velocity lags behind
the fluid velocity.
When a particle strikes the bed, it may either impact into the surface or rebound off the
surface particles. During the impact of a particle with the bed, most of its momentum is
dissipated by the particles of the bed in a sequence of more or less horizontal impulses which
may initiate the rolling mode of transport known as surface creep.
A particle will alternate between periods of successive saltations interspersed with periods of
rest on the bed. Saltation may cease when a particle is captured in an occasional bed
depression.
2. Equations of motion
The trajectory of a saltating grain can be described by solving the equations of motion for a
plane bed.
The models of Sekine and I;ikkawa (1984), Anderson and Haff (1988) and Wiberg and Smith
(1987, 1989) are based on a Lagrangian description of particle trajectory, which is obtained
from a consideration of the dominant fluid forces acting on a grain. The effect of turbulence
is neglected, so that the particle trajectory is deterministic once the initial conditions for each
saltation are specified. These models all resort to a stochastic description of collision with the
bed and subsequent rebound.
The model of Sekine and Kikkawa (1992) is also deterministic in the determination of the
particle trajectory, but probabilistic in terms of,bed collision which arises from the random
structure of the bed.
Van Rijn (1984) used a deterministic model based on drag and lift forces for given initial
particle velocities. Herein, the approach of Van Rijn (1984) is presented.
The forces acting on a saltating particle are a downward force due to its submerged weight
(F,) and hydrodynamic fluid forces, which can be resolved into a lift force (FJ, a drag force
(F,), as shown in Fig. 7.2.1. The direction of the drag force is opposite to the direction of
the particle velocity (v3 relative to the flow, while the lift force component is in the normal
direction. Assuming that the particles are spherical and of uniform density, and that the forces
due to fluid accelerations are of a second order, the equations of motions can be represented
as:
in which:
m = particle mass and added fluid mass
vS = [(u-x)~ + (i)2P-( = particle velocity relative to the flow
u = local flow velocity
x and z = longitudinal and vertical particle velocities
2 and Z = longitudinal and vertical particle accelerations
The total mass of the sphere can be represented by:
in which:
a, = added mass coefficient
Assuming potential flow, the added mass of a perfect sphere is exactly equal to half the mass
of the fluid displaced by the sphere. When the flow is separated from the solid sphere, the
added mass may be different.
The submerged particle weight is given by:
The drag force, which is caused by pressure and viscous skin friction forces, can be
expressed as:
in which:
CD = drag coefficient (see Morsi and Alexander, 1972)
A = ?A a d2 = cross-sectional area of the sphere
The lift force in a shear flow is caused by the velocity gradient present in the flow (shear
effect) and by the spinning motion of the particle (Magnus effect). For a sphere moving in
a viscous flow (small Reynolds numbers), Saffman (1965) derived the following expression:
in which:
L = lift coefficient (= 1.6 for viscous flow)
du/dz = velocity gradient
Another expression of the lift force was given by Taylor (1917), who studied the motion of
a solid body in a rotating flow system. The flow system and the solid body are defined to
have the same angular velocity. This type of motion is not identical to that of a solid body
in a uniform flow. The expression of Taylor reads as:
in which:
V = particle volume
C, = lift coefficient ( 0.5)
The lift force due to the spinning motion in a viscous flow was determined by Rubinow and
Keller (196 1):
in which:
a, = lift coefficient (= 0.4 for viscous flow)
o = angular velocity of the particle
Saffman (1965) showed theoretically that for a viscous flow the lift force due to the particle
rotation (spin) is less by an order of magnitude than that due to the shear effect and may
therefore be neglected.
To solve Eqs. (7.2.4) and (7.2.5), the following boundary conditions must be known:
bed level position,
initial velocities,
flow velocity distribution.
The bed level is assumed to be at a distance of 0.25d below the top of the particles, as shown
in Fig. 7.2.2. In its initial position a particle is supposed to be resting on a bed surface of
close-packed identical particles. The most stable position will be that of a particle resting
above one of the interstices formed by the top layer of the particles of the bed surface, which
yields an initial position of about 0.6d above the bed level. It is evident that this
schematization can not represent the movements of all the bed-load particles. The particles
in the compacted bed can only be moved by the highest fluid velocities of the spectrum
resulting in somewhat larger saltations. However, the majority of the particles is supposed
to be moving over the surface of close-packed particles.
Measurements of particles in a water stream by Abboa and Francis (1977) indicate an average
initial longitudinal and vertical velocity of approximately 2&,,. White and Schultz (1977)
analyzed high-speed motion-picture films of saltating particles in air and observed a lift-off
velocity varying from u,,, to 2.5u,,, and a lift-off angle varying from 30" to 70".
The flow velocity distribution over the depth is described by (see Fig. 7.2.2):
in which:
U*,, = bed-shear velocity
K = constant of Von Karman (= 0.4)
z, = O.ll(~/u,,~)
+ 0.03 4, = zero-velocity level above the bed
k.c = equivalent roughness height of Nikuradse
An experiment of Fernandez Luque and Van Beek (1976) was selected for calibration of the
saltation model,
Fernandez Luque and Van Beek carried out flume experiments on bed-load transport (plane
bed). Advanced film techniques were used to measure the average particle velocity, the
average saltation length and the average number of particles deposited per unit area and time
(in water) as a function of the temporal mean shear stress. Fernandez Luque and Van Beek
used four different bed materials: sand, gravel, magnetite and walnut grains. The
experimental conditions were restricted to relatively low transport stages without bed forms
(plane bed). The data of the experiment with gravel particles (d = 1800 pm) and a bed-shear
velocity (b,J of about 0.04 m/s were used. The ratio of the bed-shear velocity and the
particle fall velocity for this experiment was about 0.25. Under these conditions the sediment
particles can only be transported as bed-load. Therefore, the trajectories measured by
Fernandez Luque and Van Beek are considered to be trajectories of saltating particles,
although the trajectories are somewhat wavy at certain locations (Fig. 7.2.3).
Two parameters were used to calibrate the model: the lift coefficient ( a 3 of Eq. (7.2.9) and
the equivalent roughness of Nikuradse (kJ, see Eq. (7.2.12). As input data the following
experimental results were used: d = particle diameter = 1800 pm (= 1.8 mm), p, =
density of sediment = 2650 kg/m3, u,,, = bed-shear velocity = 0.04 m/s, v = kinematic
viscosity coefficient = 1.1W mVs. The initial longitudinal and vertical particle velocities
were assumed to be equal to 2 ~ , = , 0.08 mls.
Figure 7.2.3 shows measured and computed particle trajectories for various lift coefficients
and equivalent roughness heights.
As can be observed, both calibration parameters have a strong influence on the computed
trajectories. A reduced lift coefficient results in a reduction of the saltation length. Increasing
the roughness height also reduces the saltation length considerably, due to its direct influence
on the local flow velocity and thus on the lift and drag forces. As regards the average particle
velocity, the "best" agreement between measured and computed values is obtained for a, =
20 and &/d = 2 to 3, which is a realistic value for plane bed conditions with active sediment
transport (see Table 7.1).
Computed (a, = 20)

Particle characteristics Measured .
kJd = 1 k/d =2 k/d = 3
Particle velocity q,/u,,, 5.4 8.1 6.1 4.9
Saltation length &,Id 2 1-24 23 14.8 10.2
Saltation height &Id 2-3.5 3.5 2.5 1.9
Table 7.1 Measured and computed saltation characteristics (Van Rijn, 1984)
The measured saltation height and the measured particle velocity are "best" represented for
a, = 20 and &Id = 2 to 3. For these values, however, a relatively large discrepancy
between the measured and computed saltation lengths can be observed. Probably, the
measured particle trajectories are influenced by turbulent motions resulting in a relatively
large saltation length. Particularly, the wavy pattern of the measured trajectories indicates the
influence of the upward fluid forces by turbulent action.
FL=lift force
v, f
Figure 7.2.1 Definition sketch of saltating particle
Figure 7.2.2 Initial particle position and velocity profile
8
I 1 axwrimtal r a w 11, -
20 30 40 50
- distance, x * lo3 (m)
Figure 7.2.3 Measured and computed particle trajectories
Finally, some remarks are made with respect to the lift coefficient (aJ. The value a, = 20,
which is needed to represent the particle trajectories measured by-Fernandez Luque, is rather
large compared with the value a, = 1.6 for laminar flow (Eq. (7.2.9)). As the lift coefficient
(a3 is used as a calibration parameter, it reflects all influences. For example, the fluctuating
turbulent motions, additional pressure forces in the proximity of the wail and additional forces
due to local fluid accelerations which are not taken into account by the model.
Applying the mathematical model, the saltation characteristics were computed for a range of
conditions (G,, = 0.04 to 0.14 mls) and particle diameters (d = 100 to 2000 pm) for a plane
bed.
The input parameters were:
j = 2d,
x, = z, = 2 u,,, z, = 0.6 d,
p, = 2650 kg/m3, v = l.104m2/s, a, = 0.5, K = 0.4, Re, = u,d/v
a, = 1.6 for Re, I 5;

a, = 20 for Re, 1 70;
a, = 1.6 to 20 (linear) for 5 < Re, < 70.
3. Saltation height
The computed saltation heights were related to a dimensionless bed-shear parameter T and
a dimensionless particle parameter D, as shown in Fig. 7.2.4. These curves can be
approximated (inaccuracy = 10%)by the following expression:
in which:
6, = saltation height
d = median particle diameter
D, = d((s- l)g/ v 2)1'3 = dimensionless particle parameter
T = ( s b , - T ~ , ~ ) / T ~ ,=
= dimensionless bed-sheaf stress parameters
b c = effective current-related bed-shear stress ( s , , in case of bed forms)
t b,a
= critical bed-shear stress according to Shields
s = p i p = relative density
Experimental results have been presented by Williams (1970).

Williams carried out flume experiments with bed-load transport (d,= 1350 pm) in channels
of different widths and depths. According to visual estimation, the bed-load particles moved
within a zone of no more than about 8 particle diameters high at the largest transport stage
(plane bed). The height of the bed-load layer was independent of the flow depth, but
increased as the transport stage increased. Using the data of Williams (G,, = 0.09 mls, u,,,
= 0.03 mls), Eq. (7.2.13) predicts a saltation height of about 10 particle diameters, which
is remarkably close to the observed value.
Finally, it is remarked that Eq. (7.2.13) may predict relatively high values of the saltation
height, because it is based on computations for individual particles, neglecting the influence
of adjacent particles. In the case of collective motion of the particles, the actual saltation
characteristics as well as the particle velocity, will be reduced by particle collisions in the
bed-load layer.
4. Saltation height
The computed saltation lenghts are shown in Fig. 7.2.5. At the same T-parameter a small
particle performs a shorter saltation than a large particle, because the bed-shear velocity in
the small particle case is much smaller.
The curves can be approximated (inaccuracy = 50%) by:
Experimental support for the computed saltation lengths can be obtained from experiments
concerning the sampling efficiency of bed-load samplers carried out by Poreh et al (1970).
They observed saltation lengths in the range of 5d, to 40d, (d,, = 1900 pm, u*,, = 0.04 to
0.05 mls). For these conditions, the mathematical model predicts a range from 10d, to 30d,
(Fig. 7.2.5).
5. Particle velocity
The particle velocity on a (downward) sloping plane bottom can be derived from a force
balance consisting of the drag force F,, the gravity force Fg and the friction force F, exerted
by the bottom on the particle during contact.
Thus,
in which:
FD = 112 P CD A (I.+ - q2
Fw = (P, - P) g V ma cosp
Fg = (PS - P) g V sinp
A = l/4 51 d2 = particle area
V = 116 x d3 = particle volume
4 = angle of repose
p = angle of local bottom slope
u, = particle velocity
u, = fluid velocity
The force balance yields:
4(s - l)g d tan4 sin(+- P)

%=%-( 3 CD sin,
Assuming u, = a,
4 -
u , , and - tan4 = a, it follows that
(- ,Ii"
3 CD
-Ub- - a , - a,
'
sin(, -
uc,~ 8 sin,
in which:
UePc = bed-shear velocity (u!,~in case of bed forms)
88*'O = critical mobility parameter of Shields on a horizontal bottom
= mobility parameter
a,, a2 = coefficients
The term sin($-p)/sin@ expresses the bottom slope effect on the particle velocity. Figure
7.2.6 shows computed particle velocities on a horizontal bottom according to the saltation
model. The curves of Van Rijn can be approximated (inaccuracy = 10%) by:
For a particle mobility parameter (8) approaching the critical value, the particle velocity is
supposed to approach zero. Figure 7.2.6 shows some data of Fernandez Luque and Francis.
As regards the experiments of Fernandez Luque, only the sand (d = 900 pm) and the gravel
(d = 1800 pm) experiments for a flat bed-surface slope ( P = 0) were used. From the
experiments of Francis only the gravel data were used. As can be observed, the (scarce) data
do not confirm the influence of the D,-parameter as expressed by the mathematical model.
The computed velocities appear to be too high for D, > 15.
Equation (7.2.18) yields values in the range of 3 to 11 G,,.A reasonable average value is
about u, = 7 u,,,.
The computational results can also be approximated (50 % inaccuracy) by the following simple
expression:
Ub = 1.5
[(s - l)g d]0.5
Equation (7.2.19) yields smaller values for large particles compared to Eq. (7.2.18).
Using an expression similar to Eq. (7.2.17) and data fitting, Engelund and Fredsnre (1976)
derived:

The influence of particle shape was investigated by Francis (1973). His experiments show that
angular particles travel slower than spherical particles. The spherical particles made violent
rebounds from the bed and were lifted to higher levels where they experienced higher flow
velocities than the angular particles, resulting in a relatively large average particle velocity.
---+tmnsport stago pammter. T
Figure 7.2.4 Computed saltation heights (Van Rijn, 1984)

-------+ tmnsport s t a g pammeter, T
Figure 7.2.5 Computed saltation lengths (Van Rijn, 1984)
l5
14
measured (sand)
-
---
o Fernandez ~ u q u e4; 19.5--
computed
van Rij"
Engo~und-
e Fernandez Luaw &=39
13
12
11
10
0'
-
0.l 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Q9
mobility parameter, (8,"

Figure 7.2.6 Computed and measured particle velocities (Van Rijn, 1984)
1.0
7.2.3 Particle pick-up from the bed
1. Definitions
Usually, the pick-up rate of bed material particles is defined in terms of the number of
particles (N,)picked up from the bed per unit area and time, as:
in which:
rl = fraction of susceptible particles per unit area exposed to the flow (-)
n = I/(&, d2) = number of particles (at rest) per unit area (-)
d = particle diameter (m)
&I = shape constant (= ?A a for a sphere), (-)
ps = number of pick-ups per grain per unit time (s-')
Various researchers have studied the pick-up of bed material particles. It is, however, not
always clear what is ment by "pick-up", i.e. which events contribute to the pick-up rate and
which do not.
Einstein (1950) assumes that a particle can be picked up only after a period of rest, which
is large compared to the pick-up time scale. In this approach the total travel distance between
two successive periods of rest may consists of several jumps or saltations.
Yalin (1977) assumes that a particle is picked-up whenever it leaves the bed surface to
perform a jump. Hence, each particle jump involves a pick-up and a deposit event. It must
be stated that the pick-up rate according to the definition of Yalin may be about 10 times as
large as according to the denifition of Einstein (assuming an average travel distance equal to
about 10 jump lengths). Herein, the definition of Yalin is adopted. A detailed review was
given by Van Rijn (1983).
2. Pick-up functions
Einstein (1950)
Based on experiments, Einstein assumed that the characteristic time scale of the pick-up
process is proportional to the period necessary for a particle to settle over a distance equal
to its diameter. According to Einstein, a particle will be eroded when the instantaneous lift
force exceeds the submerged particle weight. The lift force is assumed to have a Gaussian
distribution.
Based on this stochastic approach, Einstein found:
in which:
E = pick-up rate in mass per unit area and time
P = fraction of time during which a sediment particle is picked-up by the flow (=
probability of the instantaneous lift force to exceed the submerged particle weight),
a = coefficient
d = median particle diameter
Yalin (1977)
Yalin assumed that the pick-up time scale is proportional to the ratio of the particle diameter
and the bed-shear velocity. Using a stochastic approach, Yalin proposed:
E = a pSu,P (7.2.23)
in which:
u, = bed-shear velocity
P = pick-upvelocity, see Fig. 7.2.7
De Ruiter (1982, 1983)
According to De Ruiter, the pick-up time scale is equal to the time period during which the
particle is moved from rest over a distance of half its diameter. De Ruiter has shown that this
time scale is much smaller than the time period during which the instantaneous bed-shear
stress exceeds the critical shear stress. Based on a stochastic approach, De Ruiter found:
in which:
F = pick-up probability function (see Van Rijn, 1983)
o = standard deviation of instantaneous bed-shear stress
p = critical instantaneous bed-shear stress at horizontal bed
= angle of repose
a = coefficient (= 0.016)
Nagakawa and Tsujimoto (1980)
Based on experimental data (8 = 0.03 to 0.2 and d, = 3 to 13.5 mm), they proposed a
deterministic expression:
in which:
8 = particle mobility parameter
The applied pick-up definition of the authors is not quite clear.

Fernandez Luque (1974)
~ a s e don experimental data (8 = 0.05 to 0.11, d, = 0.9 to 1.8 mm), he proposed a
deterministic function.
in which:
8, = critical mobility parameter according to Shields
3. Experimental results
Van Rijn (1983, 1984, 1986) performed experiments to determine the pick-up rate of particles
in the range of 130 to 1500 pm. A sediment lift was designed by which the sediment particles
were moved upwards at a constant rate through a circular opening in the flume bottom, as
The particles were pushed upwards by use of a piston connected to an electromotor. The
upward piston speed could be set at different (constant) values by means of a mechanical
drive system.
The sediment lift was installed in the center line of the test section of a small flume (length
= 30 m, width = 0.5 m and depth = 0.7 m). To establish a uniform flow, the flume was
equipped with a false floor supported by a jack system which was given (by trial and error)
a pre-set slope equal to the expected water surface slope in each experiment. The false floor
was covered with prefabricated wooden plates coated with sediment particles of a size equal
to those on the sediment lift. In all, five types of almost uniform sand material were used.
For each type of sand, a series of tests was performed with mean flow velocities in the range
of 0.5 to 1.0 mls. The flow depth was kept constant at a value of 0.25 m.
Prior to each run the sediment lift was filled with sediment particles by using a (perspex) tube
with a length larger than the flow depth and placed in the opening of the sediment lift. Using
this method, the lift could be filled without emptying the flume. The total mass to be poured
into the lift tube was computed from the volume of the lift tube, assuming a porosity factor
of 0.4. Some mechanical stirring was necessary to fill the tube with all available material.
At the beginning of a run, the opening of the sediment lift was covered to prevent initial pick-
up of sediment particles. The (constant) piston speed of the lift was set at a value slightly
higher than the expected pick-up rate of the flow (trial and error). Then the flow (constant
discharge) was started. After an adjustment period of about 10 min to establish equilibrium
flow conditions, the cover plate was removed and the sediment lift was started. When the
surface of the movable particles was observed (visually) to rise above the surrounding surface
of fixed particles, the lift was stopped for a short period and restarted when the over-height
of particles had disappeared. The flow was stopped when all particles were removed. To
reduce the subjective element in this procedure, different observers were used to operate the
sediment lift. The experiments with a relatively large flow velocity, and therefore a relatively
large pick-up rate, were repeated several times to increase the accuracy of the results. The
maximum variation between similar tests using different observers was about 20%.
The pick-up rate was determined as:
in which:
E = pick-up rate in mass per unit area and time (kg/sm2)
M = total sediment mass (kg)
A = area of movable surface (m2)
AT = measuring period (s)
Analysis of the experimental data yielded the following empirical pick-up function:
in which:
D, = dm [ ( ~ - l ) g / v ~ ] ~ ~
= (rb - tb,cr)/rb,~
-* o'= 7;
( Ps -p)gd
Figure 7.2.7 Pick-up probability according to Einstein (1950) and Yalin (1977)
2.10'~
Z sediment lift
lo-'
8
6
n
0
73
4
0
w=.
I
UJ
=: 2
,.
V)
Q
II
en 10-3
8
I
2
-
1u4-
lo0 2 4
-
6 8 10'
D:.~
Figure 7.2.8 Empirical pick-up function of Van Rijn (1984d)
2 4 6 810'
Figure 7.2.9 Pick-up rate according to Mastbergen (1987)
FERNANDEZ LUQUE
'--l
'.S
lo-=
lo-' 2 4 6 810° 2
Figure 7.2.10 Measured and computed pick-up rates for d,, = 130 pm, 190 pm and 360 pm
10'~
8
LUQUE
6
lo-=
8
4, I I
NAGAKAWA
lo'* 2 4 6 8 lo1 2 3.10"

2
- @ = u*
(s-1 )gd
Figure 7.2.11 Measured and computed pick-up rates for d,, = 790 pm and 1500
Equation (7.2.28) is shown in Fig. 7.2.8. The relative standard error is about 30 % (dashed
lines).
Mastbergen-(1987)used the data of Van Rijn and data from high-concentration6- = 0.4)
flows to determine the pick-up rate for a large range of mobility parameters (8). The results
are shown in Fig. 7.2.9 (tan$ = bottom slope, tan@ = 0.6).
4. Evaluation of pick-functions
The existing pick-up functions were compared to available experimental rsults by fitting of
the a-coefficients. The best-fitted a-values and the originally proposed a-values are shown
in Table 7.2. Einstein and Yalin did not propose an a-constant because their pick-up functions
were an integral part of a bed-load transport theory.
Einstein Yalin De Ruiter Nagakawa Fernandez Luque

Tsuj imoto
Sediment
Best Proposed Bes Proposed Best Proposed Best Proposed Best Proposed
fitted fitted fitted fitted fitted
130 p 0.0077 0.0088 0.0092 0.0150 0.0121

190p 0.0077 0.0118 0.0089 0.0331 0.0224
360 p 0.0094 0.0167 0.0141 0.016 0.0642 0.02 0.0436 0.02
790 p 0.0067 0.0157 0.0182 0.144 0.0854
1500 p 0.0050 0.0125 0.020 0.228 0.138
Table 7.2 a-values of pick-up functions
The "best-fitted" a-coefficients of Einstein and Yalin are not very much dependent on the
particle size. The "best-fitted" a-coefficient of De Ruiter is weakly dependent on the sediment
size, and is in good agreement with the proposed a-coefficient, particularly for the larger
particle sizes.
The "best-fitted" a-coefficients of Nagakawa-Tsujimoto and Fernandez Luque are strongly

dependent on the particle size. As regards the largest particle sizes, the "best-fitted" a-
coefficients are about 10 times too large. For the smaller particle sizes the "best-fitted" a-
coefficients are in relatively good agreement with the proposed values, which is a remarkably
good result because the proposed values are based on experiments with rather coarse bed
material particles.
Figures 7.2.10 and 7.2.1 1 show that the variation (trend) of the measured pick-up rates with
the dimensionless particle mobility parameter (8) is best represented by the curves of
Fernandez Luque and Nagakawa-Tsujimoto, while also the curve of De Ruiter shows good
results for the larger particle sizes (790 and 1500 pm).
7.2.4 Deterministic bed load transport formulas
1. Meyer-Peter Mueller (1948)

Extensive experimental work has carried out by Meyer-Peter and Mueller at the
"Eidgenossische Technische Hochschule" (ETH) in Switserland. The experiments were
performed in a laboratory flume with a cross-section of 2 x 2 m2 and a length of 50 m.
Uniform bed material as well as particle mixtures were used in the experiments (d = 0.4 to
29 mm, slope I = 0.0004 to 0.02, depth = 0.1 to 1.2 m). The bed-load transport rate is
expressed (see Fig. 7.2.12):
10 '
8
6
5
4
3
10 O
8
6
5
4
Y!
r
0 3
0"
Y 2 2
D"
-L 10-1
B
w
U
8
i :
6
10 -*
8
6
5
4
3
1o -3
10-2 2 3 4 5 6 8 10-1 2 3 4 5 6 8 100 2 3 4 5 6 8101
gTb,c
__j
(ps- P ) ~ ~ S O
Figure 7.2.12 Bed load transport rates

in which:
qb,c
4b =
(s- 1)O.S g05
= dimensionless bed-load transport rate
= dimensionless particle mobility parameter
= volumetric bed load transport rate (m2/s)

= current-related bed-shear stress (N/m2)
= mean particle diameter (m)
= bed-form factor or efficiency factor
= overall ChCzy-coefficient (mX/s)
= grain-related ChCzy-coefficient (mS/s)
= water depth (m)
= energy gradient (-)
= effective bed roughness (m)
= relative density (-)
The 0.047-value of Eq. (7.2.29) can be interpreted as the critical mobility parameter (8,).
Since, the formula is related to coarse material, the authors preferred to use a constant value
of 0.047.
Meyer-Peter Mueller proposed to use the mean particle diameter 4, as the characteristic
particle diameter, defined as: dm = C pi di. The &-parameter is about 1.1 to 1.3 times
larger than the d,-parameter for nearly uniform material.
Equation (7.2.29), however, is only weakly dependent on the particle diameter. Therefore,
the median particle diameter dm may also be used. The grain-roughness is related to the dm-
diameter.
Influence of particle diameter
The effect of the particle diameter on the bed-load transport is shown by a computation
example.
Given : bed-shear stress tb = 5 N/m2, bed form factor p = 0.5.
Compute : bed load transport for & = 600, 700, 800, 900 and 1 0 pm.
Solution : see Table 7.3.
Table 7.3 Bed-load transport rates according to Meyer-Peter-Mueller

Thus, a 25%-variation of the particle diameter (d,= 800 f 200 pm) results in a 10%-
variation of the transport rate (8, = 4.1 f 0.4 lo-' m2/s).
2. Fn'jlink (1952)
The formula of Frijlink essentially is an approximation of the formula of Meyer-Peter and

Mueller (1948) and that of Einstein (1950), as shown in Fig. 7.2.12.
The formula, which is given here because of its simplicity, reads as:
The parameters are identical to those of Meyer-Peter-Mueller (1948).
The forces per unit area moving the bed-load particles are (see Fig. 7.2.13):
the applied bed-shear stress: rb = p g h I
the gravity component: r, =
("L~)
- g m sine
in which:
m = p, V, = mass of bed-load particles per unit area (kg/m2)
$ = slope angle
V,, = solid volume of bed-load particles per unit area (m)
p, = sediment density (kg/m3)
The moving grains exert a normal stress on the bed,
The tangential stress at the bed resisting the moving bed-load grains is:
in which:
tan+ = dynamic friction coefficient
The applied bed-shear stress (at the base of moving bed-load layer) consists of a dispersive
grain-shear stress (t,J and an intergranular fluid bed-shear stress (rbf).
The dispersive stress (rb> is a shear stress which is transferred by grain to grain interaction.
Since, the moving grains have a velocity smaller than the local fluid velocity, the grains
receive their momentum from the fluid motion. Thus, the grains by receiving momentum
resist the fluid motion.
The ry-parameter is the residual fluid shear stress exerted directly on the bed by drag and
skin friction. The rbJ-parameter is negligible small in the near-bed layer where the grain
concentration is high.
According to Bagnold, grains are set in motion only by the fluid-shear stress (rbf). As many
layers of bed-load particles will be eroded as necessary to develop a dispersive grain shear
stress which just keeps the fluid-shear stress at the immobile bed below the critical bed-shear
stress for initiation of motion.
Thus, the bed-load layer acts as a protective layer to obtain a stable bed. When there is no
bed-load layer, each successive layer of bed particles will be eroded and suspended because
the fluid bed-shear stress remains always larger than the critical shear stress.
Equilibrium of shear stress at the bed requires (see Fig. 7.2.13):
The work (W,) required to be done by the grain-shear stress in moving the bed-load particles
is:
Defining the volumetric bed-load transport (m2/s) as q,,,, = V, u,, it follows that:
The available fluid energy per unit area and time is:
Bagnold assumed that: Wr = e, W,, yielding:
in which:
¶b,c = volumetric bed-load transport (m2/s)
r
,b = p g h I = overall bed-shear stress (N/m2)
I = energy gradient (-)
% = efficiency factor (0.1-0.2)
tan4 = 0.6 = dynamic friction coefficient (-)
tanP = Ib . = bed slope (-)
h = water depth (m)
g = gravity acceleration (m/s2)
Vb = volume of bed-load per unit area (m3/m2)
It should be noted that:

the bed-load transport is independent of the particle diameter (d),
the transport rate is related to the overall bed-shear stress (rJ and not to the effective
stress (r,,); the effect of the bed forms is not taken into account.
4. Van Rijn (1984~)
Four different methods were used by Van Rijn to compute the bed-load transport rate.
A. Van Rijn followed the approach of Bagnold (1954) assuming that the motion of the bed-
load particles is dominated by particle saltations (jumps) under the influence of
hydrodynamic fluid forces and gravity forces. The saltation characteristics have been
determined by solving the equations of motions for an individual bed-load particle. The
bed-load transport rate (qb,J is defined as the product of the particle velocity (h),the
saltation height (6,) and bed-load concentration (q,)resulting in q,,, = ub 6, q,.
The bed-load concentration (c,) was calculated from measured bed-load transport rates
(q,,,) using Eqs. (7.2.13) and (7.2.18).
In all, 130 flume experiments with particle diameters (&) ranging from 200 to 2000 pm,
water depths larger than 0.1 m and a Froude number smaller than 0.9 were selected
from the Literature. The influence of side-wall roughness was eliminated by using the
method of Vanoni and Brooks (1957). The form roughness was eliminated by using a
bed form factor p = (c/c')~.
Analysis of the results showed that the bed-load concentration can be represented by (see
Fig. 7.2.14):
in which:
q, = volumetric bed-load concentration (-)
c, = maximum volumetric concentration = 0.65
Using Eqs. (7.2.13), (7.2.19) and (7.2.43), the bed-load transport rate for particles in
the range of 200 to 2000 pm can be derived from g,,= 6, u, q,, resulting in:
Equation (7.2.44a) was found to overpredict (factor 2) the transport rates for T 2 3.
Therefore, a modified expression is proposed for this range:
for
Figure 7.2.13 Stresses at base of bed load
----+ -
flow particle parameter, -
T
D*
Figure 7.2.14 Bed load concentrations according to Vim Rijn (1984a)
in which:
qb.c = volumetric bed load transport rate (m2/s)
/
T = (tb,C - tb,cr)/~b,cr = dimensionless bed-shear parameter
~6 = ~grnC1~ = effective bed-shear stress (N/m2)

C' = 181og(12h/3d& = grain-related ChCzy-coefficient (m"/s)
h = water depth (m)
d50, = particle diameters (m)
-
u
d90
= depth-averaged velocity (m/s)

= critical bed-shear stress according to Shields (N/m2)
= dimensionless particle parameter (-)

= relative density (-)
= sediment density (kgIm3)
= fluid density (kg/m3)
= kinematic viscosity coefficient (m2/s)
= acceleration of gravity (m/s2)
The effective bed-shear stress (r:,~ can also be expressed as rb, = p ip, with
p = (CIC')~ = bed form factor, c = overall ~hezy-coefficient,rb,c= pg u/~' =
overall bed-shear stress.
B. A more s p p l e formula can be obtained from Eqs. (5.2.13), (7.2.43) and assuming
ub = 7 yielding:
in which:
I
uer = ( r k l P Y = go5 G/c ' = effective bed-shear velocity (m/s)
C. Equation (7.2.44) can be -
- approximated by the following formula based on the
independent variables u, u, h and d,:
in
- which:
u = depth-averaged flow velocity (mfs)
- = critical depth-averaged flow velocity based on Shields (mls)
u~
h = water depth (m)
- = 0.19(d&0.110g(12h/3Q for 0.0001SdroSO.0005 (7.2.47)
-
uct = 8.50 (d&Oe6 log(l2h/3Q for 0.0005 5 d, r 0.002
d,, d, = particle diameters of bed material (in metres)
D. The bed-load transport can also be calculated by multiplication of the pick-up rate (E)
and the saltation length (b)resulting in g,,= E A,, as shown by Van Rijn (1986), see
also Eq. (7.2.3).
Influence of particle diameter
The effect of the particle diameter on the bed-load transport is shown by means of a
computation example.
Given' bed-shear stress r, = 5 N/m2, bed form factor p = 0.5, v = 1. lo4 m2/s, p,
:
= 2600 kg/m3, p = 1000 kg/m3
Compute : bed load transport rate for d, = 600, 700, 800, 900 and 1000 pm using
EQ. (7.2.44) and (7.2.45)
Solution : r, = p r, = 2.5 N/m2, u, = ( T L / ~ ) ~=" 0.05 mls, results see Table 7.4.
dso D* =b,* T qb (m2/s)
be> (-) (NIm2) (-1 Eq. (7.2.44) Eq. (7.2.45)
600 15.2 0.30 7.3 5.2 6.5 lo4
700 17.7 0.35 6.1 4.7 10" 5.6
BOO 20.2 0.40 5.3 4.5 lom5 5.0 10-5
900 22.8 0.47 4.3 3.8 3.9
1000 25.3 0.55 3.5 3.2 lo-' 3.1 10"
Table 7.4 Bed-load transport rates according to Van Rijn
According to Equation (7.2.44), a 25 %-variation of the particle diameter (d,= 800 f 200
pm) results in a 25%-variation in the transport rate (B, = 4.3 f 1.1 l o 5 m2/s). Equation
(7.2.45) yields a 35 %-variation of the transport rates. The computed transport rates are in
reasonable agreement with those according to the Meyer-Peter-Mueller method (see Table
7.3).
Information of the influence of the particle diameter on measured bed-load transport rates can
be obtained from the data of Guy et al (1966). Figure 7.2.15 shows measured bed-load
transport rates as a function of bed-shear stress for two grain sizes. As can be observed, the
transport rate is inversely proportional to the grain size diameter. A decrease of the grain size
by a factor 2 yields an increase of the transport rate by a factor 2. This value is similar to
that according to the method of Van Rijn (see Table 7.4), but larger than that according to
the method of Meyer-Peter-Mueller (see Table 7.3).
x measured Guy e t al (1966)
- bed-sheer stress, -rb (N/m 2)

Figure 7.2.15 Influence of grain size on bed-load transport rate
7.2.5 Bed load transport at low shear stress
Extensive measurements of low bed-load transport rates were performed and analyzed by
Paintal (1971). Based on the experimental results, Paintal proposed:
in which:
a = 6.56 lo1*
p = 16
Ob = qb/[(s-l)0 5 g 0.5 g]= dimensionless bed-load transport
0 = ~ ; / [ ( s - l )d~d = bed-shear stress parameter
Equation (7.2.48) which is valid for 8 < 0.06 and dm = 1 to 25 mm is shown in Figure
7.2.16.
Application of Eq. (7.2.48) will lead to large errors in the transport rate as a result of small
errors in the bed-shear velocity.
Figure 7.2.16 Bed-load transport at low bed shear stress

7.2.6 Bed-load transport at steep slopes
1. Longitudinal bed slopes
Most bed-load transport formulae were derived for nearly horizontal beds (bed slope = tan P
5 lQ3).Experimental data of steep channels (Smart, 1984) show a strong increase of the
transport rates at steep bed slopes.
It appears that the bed slope may affect the transport rate in three ways, as follows:
the bed slope will influence the local near-bed flow velocity;
the bed slope will change the threshold conditions;
the bed slope may change the transport rate once the sediment is in motion.
The first effect is a matter of hydrodynamics and will not be discussed herein. The second
effect (see Section 4.1.2) can be taken into account properly by the Schoklitsch-factor (Eq.
(4.1.20)).
The third effect was studied by Bagnold (1966). Based on his work (see Eq. (7.2.42)), a
relationship between the transport rate on a sloping bed and on a horizontal bed can be
derived, which reads as:
in which:
qb,slope = bed-load transport on a sloping bed
qb.0 = bed-load transport on a horizontal bed
a, = -4 = Bagnold slope factor (see Fig. 7.2.17),
cosp (tanQ*tanp) (+ for upsloping flow, - for downsloping flow)
= angle of repose (tan4 = 0.6)-

= local bed slope (longitudinal direction)
The Bagnold slope factor was studied by Hardisty and Whitehouse (1988) for aeolian sand
transport. The Bagnold factor was found to give a significant underprediction of the measured
results.
Smart (1984) measured bed-load transport rates in steep channels (bed slopes tanP = 0.03
to 0.2, d = 0.002 to 0.0105 m, depths h = 0.01 to 0.09 m, plane bed). Based on his data
and the old data of Meyer-Peter-Mueller, he proposed the following formula for plane bed
(d r 400 pm, I, = 0.0004 to 0.2):
in which:
dimensionless bed-load transport rate
dimensionless mobility parameter
volumetric bed-load transport rate (m2/s)
longitudinal bed slope
water depth
relative sediment density
ChCzy-coefficient
acceleration of gravity
u = (&/Go.' = sediment gradation coefficient
4 = x pi di = mean particle diameter

= characteristic bed material sizes
= critical mobility parameter of Shields corrected for the slope
effect (8, = kg 8 see Section 4.1.2)
Comparing Eq. (7.2.50) to the original formula of Meyer-Peter-Mueller (1948) for plane bed
represented as Qb = 8(8-8a)1.5, yields an expression for the slope factor:
in which:
a, = slope factor ( a , 2 1)
tb = bed-shear stress
t b,= = critical bed-shear stress of Shields corrected for the slope effect (t, = k, t,,,).
Equation (7.2.51) which is only valid for downsloping flow, is shown in Fig. 7.2.17 for
tb/rbp = 1.5,2.5, 5 and -, C = 35 and 70 m"/s, &/dm = 4 and plane bed. Figure 7.2.17
expresses that the Meyer-Peter-Mueller formula considerably underpredicts the transport rates
at steep slopes (factor 2 to 10). The largest effect does occur for t, close to t,, . The effect
is smallest for r,/r, - m. The slope factor becomes important ( a , > 1) for a slope of
about 0.01. For steep slopes the increase of the bed-load transport rate may be a factor 2 to
10, depending on the C-value. This effect is additional to the reduction of the critical bed-
shear stress effect taken into account by the k p -factor (see Chapter 4). The slope factor of
Bagnold appears to be much too small (see Fig. 7.2.17).
Rickermann (1991) studied the influence of fine sediments (wash load with volume
concentrations upto 22%) on the bed-load transport of coarse materials at steep slopes. The
wash load had no influence on the bed roughness in the turbulent flow regime. The water and
sediment mixture behaved as a slurry with a different rheological behaviour than that of a
clear water flow.
For a given bed slope and flow discharge, the clear-water bed-load transport rate showed a
considerable increase (factor 2 to 3) for increasing slurry concentration (increasing fluid
density), particularly at steeper slopes. This change did occur at almost the same depth and
velocity implying that flow resistance did not significantly change. The increase of the bed-
load transport was found to be mainly related to the decreasing relative density (s = p,/p,, p,
= sediment density, p, = mixture density).
At slurry (clay) concentrations above 17% (p, < 1280 kg/m3), a decrease of the bed-load
transport rate was observed with further increase of the slurry concentration. This decrease
was found to occur at a particle Reynolds number smaller than about 10, which means a
viscous sublayer thickness larger than the grain diameter.
H
mild slopes
-
W
steep slopes
H
-..-... Bagnold
.-- -- C= 35 ml/2/s
C= 70 m1/2/s
d90 /d 30 =
plane bed
- bed slope, Ib=tan p

Figure 7.2.17 Ratio of bed-load transport rate over sloping bed (q,,,J
bed (q,,J
and over horizontal
u b , X direction ' /
Figure 7.2.18 D@nition sketch for a particle moving in a curved channel with transverse
slope y
10-2 2 3 4 5 6 810-1 2 3 4 5 6 8100

------b transverse slope, tan y
Figure 7.2.19 Ratio of bed-load transport in transverse direction (q,,,) and in longitudinal
direction (qb,J
2. Transverse bed slopes
Bed-load transport on transverse slopes was studied by Engelund (1974), and by Ikeda (1982,
1988).
Herein, the approach of Ikeda is followed. Based on analysis of the equations of motion in
transverse (n) and longitudinal direction (s), the particle velocity in n-direction (see Fig.
7.2.18) was found to be:
in which:
v,,, = particle velocity in n-direction
v, = particle velocity vector
-
Ub,n - near-bed fluid velocity in n-direction
ub = near-bed fluid velocity vector
-
Ub,m - critical near-bed fluid velocity (threshold)
y = transverse slope angle (tany = - %/ity)
E = (1 + a p/p) = coefficient
a = ratio of lift and drag coefficient
p = dynamic coefficient of Coulomb friction (0.5-0.8)
Assuming: vp -- vp,s* ub = %,s' =
It follows that:
in which:
q = bed-load transport in s-direction
T, = bed-shear stress in s-direction
tan6 = I+,,/%, (tan6 = 0 for transverse slopes in straight channels)
Based on measured qb,,-values for transverse slopes along a straight channel in a wind-tunnel
(d,, = 260 pm and 420 pm), the E -factor was found to be about E = 1.5.
The ratio of qb,, and qbV8
is shown ir, Figure 7.2.19 for tan6 = 0, E = 1.5 and + b / + b ,=
1.5, 2, 5 and 10.
7.2;7 Bed-load transport of non-uniform material
I . Bed load transport
The bed material in natural conditions consists of non-uniform sediment particles. The effect
of the non-uniformity of the bed material will result in selective transport processes which
was studied by Einstein (1950), Egiazaroff (1965), Samaga et al(1986), Ribberink (1987) and
others.
Generally, the approach is to divide the bed material in a number of size fractions and to
apply an existing formula for each size fraction with a correction factor ( t i ) to account for
the non-uniformity effects.
The correction is necessary because the coarser particles are more exposed to the flow than
the finer particles which are somewhat sheltered by the coarse particles. This correction can
be effectuated by increasing the critical shear-stress of the finer particles and decreasing the
critical value of the coarser particles.
Using the formula of Meyer-Peter and Mueller (1948), the bed-load transport rate integrated
over the fractions reads as:
in which:
4b = bed-load transport rate integrated over N size classes (fractions)
Pi = percentage of size class i of the bed material
4 = particle diameter of size class i
s = specific density (= 2.65)
IJ = bed form factor
6i = mobility parameter of size class i
6i = correction factor of size class i
8, = critical mobility parameter based on the average diameter d,
d, = C pi di = average diameter of bed material
N = number of size fractions
Equation (7.2.56) should be used when d,/d,, 1 5.
Generally, the correction factor Si is assumed to be equal to unity for 4 = d,.Thus

Si = 1 for 4 = d,.
Egiazaroff (1965) proposed:
Equation (7.2.57) yields Si = 5 for 4/d, = 0.2 and Si = 0.4 for &Id, = 5.
The particle diameter of the transported bed-load particles can be expressed as:
Ribberink has carried out sensitivity computations applying the formula of Meyer-Peter and
Mueller. Figure 7.2.20 shows the ratio qb,,/qb as a function of the number of size fractions
N for 3 different gradation coefficients a, = 1.23, 1.96 and 2.45. The bed material was
assumed to have a log-normal distribution. It can be observed that approximately 5 fractions
are necessary to obtain a constant ratio qb,,/qb. Furthermore, it appears that the transport rate
is hardly affected by using the size-fraction method (maximum variation = 10%).
Klaassen (1991) performed flume experiments with uniform and non-uniform sand materials
(d,, = 770 pm, a, = 1.06 and 5,, = 660 pm, ug = 2.34).
The measured dune heights for non-uniform material were approximately 20 % larger than for
uniform material. The dune lengths were approximately 50% larger than for uniform
material. The ChCzycoefficient was found to be larger for non-uniform material because of
the decrease of the dune steepness (reduced form roughness). The bed-load transport rates for
non-uniform material were approximately 50% larger than for uniform material.
Selective transport processes are most typical for gravel-bed rivers, because of the rather wide
size distribution of the bed material-(1 to 100..mm). Gravel-bed rivers are generally found
near the head of a river system (upland regions) near the primary source of sediments.
Sediment supply events tend to be episodic related to the presence of high flows.
The most typical feature of a gravel bed is a surface cover that is relatively coarse in
comparison with the bulk sediment mixture beneath it (hYd, byer = &,ad,hyer). The
surface layer (thickness = &) is termed pavement or armour layer. The lower layer is
termed subpavement, substrate or subsurface layer.
During low flows the finer particles are winnowed from the surface leaving a surface of
coarser particles; which will act as an armour layer. During high flows the surface layer may
become mobile depending on the magnitude of the applied bed-shear stress compared to the
critical bed-shear stress. Even in high flows particle motion is sporadic; only a small fraction
of the surface grains of a given size will be in motion. Fine grains may be eroded from the
subsurface layer by eddies generated in the holes created by coarse grains just eroded by the
flow. Laboratory experiments show that grains of all size classes are in motion at high flows
(Parker and Klingemann, 1982; Parker et al, 1982; Thorne et al, 1987).
Parker and Klingemann (1982) proposed a transport formula (based on d, of subsurface
layer) for gravel-bed rivers.
Ashiq et a1 (1992) found bad agreement between measured and computed transport rates for
a gravel-bed river using the d, as characteristic diameter in the formulae of Meyer-Peter-
Mueller (1948) and Parker et al (1982). The computed values were much too small.
Figure 7.2.20 Influence of size fractions on bed-load transport (Ribberink, 1987)

2. Vertical sorting
Vertical sorting takes place in the upper layer of the bed which generally is covered with bed
forms. The particles that are (occasionally) taking part in the transportation process are
present in a layer with a thickness equal to the largest bed form height.
Flume and field observations show a vertical sorting effect with finer particles dominating in
the upper layers and coarser particles dominating in the lower layers at the trough level of
the bed forms. This sorting effect can be. explained by the transportation process. After the
particles have passed the bed form crest, they roll and ~lide~down the leeside slope of the bed
form. As a result of gravity the coarsest particles have thC! largest probability to reach the
deepest part of the bed form trough. Armouring of the bed occurs when an immobile fraction
is present in the top layer of the bed. The finer particles will be carried away by the flow and
the immobile particles will eventually form an armour layer.
3. Horizontal sorting (armour layer)
In non-uniform bed material an armour layer will be formed when the coarser particles of bed
material have critical stresses which are larger than the mean bed-shear stress (zd, whereas
the finer particles have critical stresses smaller than the mean bed-shear stress. In that case
the finer particles will be eroded and transported in downstream direction, but the coarser
particles remain stable (immobile) and form an armour layer against erosion of sediment
material of the lower layers. If the flow velocity increases to a point where the smaller
particles of the armour layer can be disturbed, some of the finer particles become exposed
and will be carried downstream over and around the (stable) coarser particles of the armour
layer. Whenever, the protective armour particles are moved and carried downstream, the
process of armouring with respect to the particles of the lower layer is repeated.
Coarsening of the bed was observed downstream of the Lake Mead reservoir in the USA
(Lane et al, 1949). Livesey (1965) observed a flat armour layer consisting of one layer of
gravel downstream of Fort Randall Dam in the Missouri river in the USA. The median size
of the bed sediment before closure of the dam was about 170 pm. This material was still
present below the armour layer.
Armouring may occur when there is no supply of sediment from upstream (sediments trapped
in reservoirs) or when the supply from upstream consists of finer material than that of the
local bed.
Gessler (1970, 1973) proposed a method to determine the composition of the armour layer
and the eroded material.
The fraction @,Jof particles of size class i of the armour layer is given by:
The fraction @,Jof particles of size class i of the eroded material is given by:
in which:
po,i = fraction of size class i of original bed material
ri = probability that a particle from size class i will remain stable
The probability ri was derived from experimental results and is shown in Figure 7.2.21 as a
function of the ratio rcrj/rbwith ra = critical bed-shear stress of particle with diameter d,
from size class i, and rb = mean bed-shear stress (= p ghI).
Measured and computed particle size distributions show good agreement (see Fig. 7.2.22).
The d, of the armour layer was found to be approximately equal to the d, of the original bed
mater~al.The erosion depth above the armour layer was found to-be about 1 to 2 d,,.
Figure 7.2.21 Probability that a grain will be stable (Gessler, 1970)
Measured
- Cornpuled
r b = 1 . 3 N/m2
Figure 7.2.22 Particle size distributions (Gessler, 1970)

7.2.8 Comparison of bed-load transport formulae
A large amount (566) of bed-load transport rates (d,in the range of 110 to 1500 pm)
measured in flume and field conditions have been selected from the Literature to verify the
formulae of Van Rijn, Meyer-Peter and Mueller, and Frijlink (see Van Rijn, 1986). The
results are presented in terms of a discrepancy ratio (r) defined as the ratio of the computed
and measured bed load transport rate. The percentage of r-values of all data falling in the
ranges 0.75 Ir I 1.5; 0.5 Ir 5 2 and 0.33 5 r I3 are given in TabIe 7.5.
Percentage of r-values in range

Formula
0.75 I r 5 1 . 5 0.5 5 r 5 2 0.33 5 r 5 3
Van Rijn 34 % 63 % 85 %
Meyer-Peter-Mueller 31 % 60 % 81 %
Frijlink 31 % 62 % 89 %
Table 7.5 Comparison of measured and computed transport rates

The results of the three methods are about the same. The formulae of Engelund-Hansen and
Ackers-White (Section 7.3) produce similar values for large particle sizes (d, 2 500 pm).
Finally the accuracy of the measured transport rates is discussed. A comparison of flume
experiments performed under similar flow conditions (equal depth, velocity, particle size,
temperature) by different research workers shows deviations of the transport rates upto a
factor 2. Thus, even under controlled flume conditions, the accuracy of the measured values
is rather low, which may be caused by the influence of the applied width-depth ratio, the
applied adjustment period to establish uniform flow conditions and the applied experimental
method (sand feed or recirculating flume).
Based on this, it is stated that it is hardly possible to predict the transport rate with an
inaccuracy less than factor 2.
The formula of Meyer-Peter and Mueller (1948) and that of Frijlink (1952) are also shown
in Figure 7.2.12 for a specific case (C = 63 m"/s, h/d, = 1000 and p = 0.76).
7.2.9 Stochastic bed-load transport formulae
I . Introduction
Physically, the stochastic approach is more realistic than the deterministic approach because
the stochastic nature (turbulence) of the fluid forces on the grains is taken into account. The
proposed equations are, however, not easy applicable because of the statistics involved.
Furthermore, for uniform flow the predicted transport rates based on stochastic theories are
not more accurate than those based on deterministic theories. Both approaches need
experjmental data for calibration.
give realistic results because the mean bed-shear stress will approach zero (T, -
In strongly decelerating flow (adverse pressure gradient) the deterministic formulae do not
0) in case
of flow separation and reattachment. Hence, the computed transport rate will also approch
zero. Physically, this is not realistic because at the location where t, approaches zero, the
turbulent fluctuations usually are large and may generate bed-load transport. These effects can
be represented by applying a stochastic approach, as introduced by Kalinske (1974), Einstein
(1950), Yalin (1977) and by Van Rijn (1987). The stochastic approach may also give more
accurate results for the computation of transport rates close to initiation of motion.
He assumed that the grains are transported in a layer with thickness equal to the particle
diameter (d).
The instantaneous particle velocity is taken as:
in which:
U = instantaneous fluid velocity at particle level
U, = critical fluid velocity (threshold velocity)
The instantaneous fluid velocity is assumed to have a normal distribution with average value
(u) and standard deviation (a) , as follows:
The mean particle velocity is given by:
The volumetric bed load transport is defined as:
in which:
N = number of moving grains per unit area (= p/(%rd2))
V = volume of a grain
p = fraction of moving solid material per unit area (p = 0.35)
This leads to:
Equation (7.2.65) is shown in Figure 7.2.12 for a specific case (C = 63 m"/s, h/d, = 1000,
p = 0.76).
3. Einstein (1950)
The most detailed theory of particle motion and transport was presented by Einstein (1950).
His approach is based on the assumption that in steady uniform flow there is equilibrium
between the number of particles eroded and deposited per unit area and time (E = D).
The erosion or pick-up rate is given by (see Eq. (7.2.22) and Figure 7.2.7):
E = a, [(s-l)g dl0" p
in which:
p = probability of erosion
a , = coefficient
The deposition rate is defined as (see Eq. (7.2.3)):
in which:
X = average travel distance of a particle
Using Eqs. (7.2.66), (7.2.67) and E = D, it follows that:
The average travel distance of a particle was found to be:
in which:
1-P = probability of deposition
a, = coefficient (= 100)
Using Eqs. (7.2.69) in Eq. (7.2.68), yields:
in which:
4 = dimensionless bed-load transport
A, = empirical coefficient (= 43.5)
P = probability of erosion
The probability of erosion (P) was expressed as a function of the particle mobility parameter
8' :
in which:
P = probability of erosion (see Fig. 7.2.7 for uniform material)
0' = ~:/[(p,-~)gdl = mobility parameter
I
t = pgh' I = effective bed-shear stress
hb' = effective water depth (see Einstein and Barbarossa, 1952)
I = energy gradient
a = EY(P/P J2 = coefficient
For graded bed material the transport is obtained by summation of the transport rate per
fraction (4 = 4i).The a -coefficient represents various effects related to the behaviour
of particles in a graded mixture such as the hiding of smaller particles between the larger
particles ( E ) , the variation of lift coefficient (y) and the variation of particle positions
(PIP *I2.
For almost uniform bed material it was proposed to use the 4, as the characteristic particle
diameter related to particle mobility and the 4, as the effective grain roughness diameter. The
a -coefficient is unity ( a = 1) for uniform material. Figure 7.2.7 shows the probability of
erosion (P) for uniform material. Equation (7.2.70) is shown in Figure 7.2.12 for a specific
case.
Einstein (1950) described a very detailed theory taking many physical phenomena of the
particle motions into account. Despite all physical details, the scatter of the data around the
proposed function is not significantly less than that of other more simple methods. The
justification of the Einstein approach is given by its detailed representation of physical
phenomena involved, more than by its accuracy.
Given : d,, = 800 pm, t, = 1.3 N/m2, plane bed
Compute : @ and qb
Solution : 0 = rb/[(p,-a)g dJ = 0.1

P = 0.8, see Figure 7.2.7
@ = (1143.5) (0.810.2) = 0.092
qb = 0.092 (1.65 9.81)0.5(0.0008)1.5= 8.4 lo4 m2/s
4. Van Rijn (1987)
It is assumed that the instantaneous bed-load transport rate is related to the instantaneous T-
parameter, as follows (see Eq. (7.2.44)):
in which:
I
Tm = ('b - 5b,cr)/5b,cr = instantaneous shear stress parameter
= instantaneous effective bed-shear stress
= instantaneous critical bed-shear stress
= efficiency factor
The effective bed-shear stress r; is assumed to have a normal distribution (mean value ?;,
standard deviation o ), as follows:
I
The probability that r; > r , , , , (positive in flow direction) or -rb < rbSa2(negative against
flow direction) is:
which can be represented as the shaded areas in Figure 7.2.23.
Applying Equation (7.2.74), the time-averaged shear stress parameter (T,,,)~.' can be derived,
as follows:
in which:
0
I, = + / (x12.' e - ' ~ ' - ~dx,) ~ see Fig. 7.2.23

I, = - / (x)~.'e-lh('-p)' dx, see Fig. 7.2.23
0
The I,-integral (I, = -I,) is shown in Figure 7.2.24. It is noted that the I,-integral has a
negative sign, representing the transport of bed-load particles against the flow direction due
to negative velocity fluctuations (against flow direction). Because the transport of material is
related to a particular direction, the bed-load transport against the flow direction has to be
subtracted from the value in the flow direction.
Based on the work of De Ruiter (1983), the instantaneous critical bed-shear stress was found
to be:
The standard deviation of the effecive bed-shear stress in uniform flow was found to be @e
Ruiter, 1983):
Using Eqs. (7.2.73, (7.2.76) and (7.2.77), Equation (7.2.72) was recalibrated for uniform
flow, yielding:
To apply Eq. (7.2.78), the values of a, tb,cr,l and tb,a2must be known.

Example
Given :
bed material d5, = 800 pm, D, = 20, plane bed, bed-shear stress r, = 1.3
N/m2, critical bed-shear stress = 0.4 N/m2
Compute : bed-load transport using stochastic and deterministic method
Solution :
stochastic method, Eq. (7.2.78)
0 = 0.4 5, = (0.4) (1.3) = 0.52 N/m2
r = (1.3 - 0.6)/0.52 =. 1.35 - I, = 7 (Fig. 7.2.23)

P = (-1.3 - 0.6)/0.52 =, -3.65 - & = 10" (Fig. 7.2.23)
Deterministic method, Eq. (7.2.44)
T = p b - ~b,=)l~b.= = (1.3 - 0.4)/0.4 = 2.25
7.2;6 Examples and problems

1. A wide river has a deyth of h = 3 m. The depth-averaged velocity is i = 1.2 m/s. The
energy slope is I = 1.67 lo". The mean particle size of the bed material is d, = 2000
pm, d,, = 1800 pm, 9,
= 3000 pm.
Compute the bed-load transport according to Meyer-Peter-Mueller, Bagnold and Van

Rijn?
Solution:
Meyer-Peter-Mueller : C = ~/(N)O.~ = 53.6 mH/s
C' = 18 log(l2Wh) = 73.4 mIhls
P = ( C / C ' ) * .=
~ 0.623
Bagnold
tanp = I = 1.67 lo4
cosp = 1
Van Rijn (Eq. (7.2.44)) : C' = 18 log(12W3 dgO) = 64.8 m"/s
b,a = 1.2 N/m2 (Shields)
2. A wide river has the following characteristics: h = 2m, u

= 1.4 m/s, I = 8 lo4, d,
= 800 pm, d,, = 700 pm, d, = 1500 pm, v = 1 lo4 m2/s.
Compute the bed-load transport rate according to Meyer-Peter-Mueller, Van Rijn and
Bagnold?
Solution: Meyer-Peter-Mueller : qb = 1.4 lo4 m2/s

Bagnold : qb = 2.3 10' m2/s
Van Rijn, Eq. (7.2.44): qb = 1.1 lo4 m2/s
3. A wide river has a depth of h = 3 m. The depth-averaged velocity is 6 = 1.2 m/s. The
energy gradient is unknown. The bed material characteristics are: d,, = 1800 pm, Go
= 3000 pm . The bed is plane (no bed forms). Other data are: v = 1.1O4 m2/s, p, =
2650 kg/m3.
Figure 7.2.23 Gaussian shear stress distribun'on
Compute the saltation height and length of the particles, the mean particle velocity
according to Engelund-Freds~eand Van Rijn, and the fluid velocity at the saltation
height?
Solution:
Saltation height (VanRijn) : C' = C = 18log(12h/3&) = 64.8m"/s
D. = 45.5 (see example 1)

T = 1.8 (see example 1)
Saltation length (Van Rijn) : A, = 3 d, D? = 0.0906 m

Particle velocity (Engelund) : u. = u, = (~O.'/C)5 = 0.058 mls
, = 0.041 (Shields)
8
Particle velocity (Van Rijn) : ub = 1.5 [(s-l)g d,)"" 'OT = 0.37 m/s
Fluid velocity : z = 6, = 0.0105 m, u, = 0.058 m/s
k, = 3 d, = 0.009 m, u&/ v = 520 (rough flow)
4. Same question as 3. Use h = 3 m, ii = 1.5 m/s, I = 5 lo', d,, = 1800 pm, 4, =

3000 pm, dunes are present.
Compute saltation characteristics, particle velocity and fluid velocity at the saltation
level?
Solution: use u! to estimate grain roughness

saltation height : 0.0144 m
saltation length : 0.16 m
particle velocity : 0.48 m/s (Engelund)
: 0.53 m/s (Van Rijn)
fluid velocity : 0.70 m/s
5. The bed of a river consists of non-uniform material with grain sues between 100 and
. 6000 pm. The size fractions are given in Table 7.6. The d,, of the bed material is 980
pm; the 4, is 3600 pm. The water depth is h = 5 m. The bed slope is I, = lo4. The
ChCzy-coefficient is C = 50 mIh/s.Other data are: p, = 2650 kg/m3, p = 1000 kg/m3,
v = 1 l o 6 m2/s.
Compute the bed-load transport according to Meyer-Peter-Mueiler, using the mean

particle diameter and the size-fraction method?
Solution:
a. based on mean particle diameter

mean grain diameter : d, = pi di = 1460 pm
grain roughness : C' = 18 log(l2hlb) = 76 mHls
bed-form factor : p = (C/C')15 = 0.533
bed-shear stress : rb = p g h I = 4 . 9 N / m *
mobility parameter : 8 = r,/[(pg-p)g dm] = 0.207
pe = 0.11
bed load transport : = 8(p8-0.047)1.5= 0.127
qb+ = 0.127(~-1)~" = 2.8 10" m21s
b. based on size-fraction method
= 37
critical mobility parameter : D. = d,[(s-l)g~v~]~~
8, = 0.037 (Shields)
The bed load transport rate per fraction is given in Table 7.6.
size class 4 - pi ti pBi A = pi(P8i-ti0a)1.5 ~ ~ = 8 ( ~05

. -g10.5)41" A
Ocm) Ocm) (-1 (-1 (-) (-) (m2/s)
100-500 300 0.30 4.67 0.538 0.0662 11.0 106
500-1000 750 0.22 1.67 0.215 0.0132 8.7 1 0
'
1000-2000 1500 0.23 0.98 0.108 0.0044 8.2 106
2000-3000 2500 0.10 0.71 0.065 0.0008 3.1 lo'
3000-4000 3500 0.08 0.59 0.046 0.0003 2.0 '01
4000-5000 4500 0.05 0.52 0.036 0.0001 1.1 '01
5000-6000 5500 0.02 0.48 0.029 0.0002 0.3 106
Total 34.4 lo4 m2/s
Table 7.6 Bed-load transport based on size fraction method

The size-fraction method yields a value which is only 20% larger than that based on the
mean particle diameter.
6. The bed material of a river consists of non-uniform material with 20% between 100 and
500 pm, 30% between 500 and 1000 pm, 20% between 1000 and 2000 pm, 20%
between 2000 and 3000 pm and 10% between 3000 and 4000 pm. Other data see
example 5.
Compute bed-load transport using mean particle size and size fraction method (Meyer-
Peter-Mueller formula)?
Solution: mean particle size q,,, = 2.7 l o 5 m2/s
size fraction method q,,, = 3.3 m2/s
7. A reservoir has been built in a river. The bed of the river section downstream of the
reservoir consists of non-uniform material with a size distribution according to Table
7.7.
What is the composition of the armour layer which will be formed downstream of the
reservoir, when the flow passing this section exerts a bed-shear stress of sb = 2.5
N/m2?
Solution: use method of Gessler, see Table 7.7
'b,ad according to Shields
Ti according to Gessler
size class di P0.i 'a$

-
'
a$
ri ri P , i Pri 2 P4
Tb
OLm) OLm) - (N/m2) (-) (-1 (-1 (-1 (-1

100- 500 300 0.30 0.18 0.07 0.05 0.015 0.08 0.08
500-1000 750 0.25 0.36 0.14 0.07 0.018 0.09 0.17
1000-2000 1500 0.20 0.85 0.34 0.15 0.030 0.16 0.33
2000-3000 2500 0.10 1.7 0.68 0.30 0.030 0.16 0.49
3000-4000 3500 0.08 2.8 1.12 0.55 0.044 0.23 0.72
4000-5000 4500 0.05 3.6 1.44 0.75 0.038 0.19 0.91
5000-6000 5500 0.02 4.6 1.84 0.92 0.018 0.09 1.0
0.193
Table 7.7 A m u r layer composition

Based on the di and C p .-columns of Table 7.7, the characteristic diameters of the
armour layer are: d, = 2%0 pm, d, = 4500 pm.
8. Same question as that of example 7.
Use the bed material composition of example 6.
The applied bed-shear stress downstream of the reservoir is s, = 2 N/m2.
Solution: armour layer d, = 2600 pm, d90 = 3700 pm
p i = 0.05,0.10,0.16,0.36,0.33
9. A wide mountain river has a bed slope of I = 10'.
The water depth is h = 0.1 m. The bed is plane. The bed material characteristics are
d, = 0.02 m, d, = 0.012 m, d, = 0.03 m. Other data are p, = 2650 kg/m3, p =
1000 kg/m3, angle of repose 4 = 30".
What is the dimensionless bed-load transport according to Meyer-Peter-Mueller and
according to Smart?
Solution: Meyer-Peter-Mueller 4 = 1.1
Smart 4 = 1.6
7.3 Suspended load transport
7.3.1 Introduction
Usually, these modes of motions of the bed material particles are distinguished:
rolling and sliding;
saltation, and
suspension.
When the value of the bed-shear velocity just exceeds the critical value for initiation of
motion, the particles will start rolling and sliding in continuous contact with the bed. For
increasing values of the bed-shear stress the particles will perform more or less regular jumps
which are called saltations.
When the value of the bed-shear velocity exceeds the particle fall velocity, the particles can
be lifted to a level at which the upward turbulent forces will be comparable to or higher than
the submerged particle weight and as a result of the particle with the bed in the suspension
mode is occasional and random. The particle velocity in longitudinal direction is almost equal
to the fluid velocity. Usually, the behaviour of the suspended sediment particles is described
in terms of the sediment concentration, which is the solid volume (m3)per unit fluid volume
(m3) or the solid mass (kg) per unit fluid volume (m3).
The hydraulic conditions at initiation of suspension are described in Section 4.4. According
to Bagnold (1966): suspension will occur for u, ;? w,. According to Van Rijn (1984b)
suspension will start at considerably smaller bed-shear velocities.
observations show that the suspended sediment concentrations decrease with distance up from
the bed. The rate of decrease depends on the ratio of the fall velocity and the bed-shear
velocity (w,/u,) .
The depth-integrated suspended-load transport (q,3 is herein defined as the integration of the
product of velocity (u) and concentration (c) from' the edge of the bed-load layer (z=a) to the
water surface (z =h) , yielding (see Figure 7.3.1) :
with
in which:
%,c = volumetric suspended load transport (m2/s)
u = fluid velocity at height z above bed -
c- = sediment concentration (volume) at height
- - z above bed
u = depth-averaged fluid velocity
c, = reference concentration at height z=a above bed
h = water depth
F = dimensionless shape factor (see Van Rijn, 1984b, Dyer and Soulsby, 1988 and
McLean , 199 1b)
Application of Eq. (7.3.1) requires information of the velocity profile, concentration profile
and a known concentration (cb close to the bed (z=a). These latter parameters are refered
to as the r@erence concentration and the reference level, c, at z=a.
Sometimes, the suspended load transport is given as a mean volumetric concentration defined
as the ratio of the volumetric suspended load transport (= sediment discharge) and the flow
discharge:
The mean concentration (c,J is approximately equal to the depth-averaged concentration@

for fine sediments.
The concentration can be expressed as a weight concentration (c,) in kg/m3 or as a volume
concentration (c,) in m3/m3. Both are related through: c, = p, c., Sometimes the volume
concentration is expressed as a volume percentage after multiplying with 100%.
In the American literature the sediment concentration often is expressed as a concentration
(%) in parts per million by weight. In this latter case the sediment mass in a sample is
multiplied by lo6and divided by the total mass (sediment and fluid) of the same sample. The
c,, c, and cv-values are related by:
with s = pJp = specific density = 2.65.

For low concentrations c, < 10 kg/m3 it follows that c, = lo6 s c, or c, = (lo6/p) c,.
Some rivers carry very high concentrations of fine sediments (particles < 50 pm), usually
refered to as the wash load.
Experience shows that the presence of fines enhances the suspended sand transport rate
because the fluid viscosity and density are increased by the fine sediments. As a result the
fall velocity of the suspended sand particles will be reduced with respect to that in clear water
and hence the suspended sand transport capacity of the flow will increase.
A drop in temperature will have the same effect, Lane et al (1949) reported an increase of
the measured sand transport rate by a factor 2 from summer to winter (temperature drop of
about 15°C) at the same discharge.
Many analytic models are available to describe the sediment concentration distribution over
the water depth.
These models can be divided into:
diffusion models,
energy models (see Bogardi, 1974),
stochastic models (see Ashida-Fujita, 1986).
Herein, only diffusion models and energy models (Bagnold, 1966) are discussed.
Figure 7.3.1 Definition sketch suspended load transport
surface
concen tro tion profile
bed
\VAY/AVAVX\VX\VA~
Figure 7.3.2 Definition sketch vertical sediment diflvsion
7.3.2 Mass balance equation suspended sediment (diffusion model)
A complete derivation of the 3D-mass balance for the suspended sediment particles is given
in section 12.4.2. Here, a simple derivation for suspended sediment particles in a steady and
uniform turbulent flow is presented. The particles are assumed to have a uniform size (d) and
fall velocity (w,).
The instantaneous sediment concentration at a certain level 1 is assumed to be c+c I .
The instantaneous sediment concentration at level 2 is c-c I (see Figure 7.3.2), with c =
time-averaged concentration and c' = concentration fluctuation (by definition c1 = 0 in
steady uniform flow). .
The instantaneous upward sediment velocity at level 1 is w1-w, , with w = vertical velocity
fluctuation (by definition w = 0 in steady uniform flow).
The instantaneous downward sediment velocity at level 2 is w1+wS.
The time-averaged upward transport is : qn = 3

(w1-w$(c+c
r,
The time-averaged downward transport is : 4, = (w'+w$(c-c 9
In steady uniform flow the time-averaged upward and downward transport rates are equal,
giving (q,=q,J:
-
The cw, term represents the downward transport of sediment by gravity. The clw'term
represents the upward transport by the turbulent velocity fluctuations.
Adopting the diffusion model (also known as the law of Fick), the vertical transport of
sediment by turbulence (cIw )' is proportional to the vertical concentration gradient (dcldz).
The proportionality coefficient is called the diffusion or mixing coefficient (E,) . Thus:
Combining Eqs. (7.3.8) and (7.3.9), yields:
in which:
c = time-averaged sediment concentration at height z above bed
w, = particle fall velocity (constant)
E,
= mixing coefficient at height z above bed
Equation (7.3.10) is valid for low concentrations because the fall velocity is assumed to be
constant. In mixtures of high concentrations the fall velocity is not constant but dependent on
the concentration because the particles are hindering each other during settling. The falling
particles are affected by the return flow of the displaced fluid but also by additional effects
such as particle collisions, particle-induced turbulence, modified drag coefficients and group
effects. The overall effect can be represented by (see Section 3.2.5):
in which:
w,,, = particle fall velocity in a mixture
w, = particle fall velocity in clear still fluid
7 = coefficient (4 to 5 for particles of 50 to 500 pm)
7.3.3 Fluid and sediment mixing coeffcient
Various distributions of the fluid mixing coefficient (e,) can be found in the Literature.
Herein, the following distributions are given (see Figure 7.3.3):
1
constant : E ~ = - K u , ~ (7.3.12)
a1
linear
parabolic
parabolic-constant : E, = K u, h -z (I--)z for -z < 0.5

h h h
z
ef = 0.25 K u, h for - 2 0.5
h
in which:
ut = current-related bed-shear velocity
h = waterdepth
z = vertical coordinate
K = Von Karman constant (0.4)
a,,a, = coefficient
The parabolic distribution is most satisfactory in a physical sense because it is based on a

linear shear stress distribution and a logarithmic velocity profile. A disadvantage of the
parabolic distribution is that it yields a zero-concentration at the water surface. Mixing
coefficients based on the analysis of measured concentrationprofiles (Coleman, 1970) indicate
a parabolic-constant distribution rather than a parabolic one, as shown in Figure 7.3.4.
Usually, the mixing or diffusion of the sediment particles is related to the fluid mixing
coefficient for a clear fluid (e,) , as follows:
--no- constant Ef (a,=6) parabolic Ef
-.I- linear E~ (a2=3) --- parabolic -constant E f
Figure 7.3.3 Fluid mixing coflcients
- height, -
hz
Figure 7.3.4 Sediment mixing comcient according to Coleman (1970)

The P -factor describes the difference in the diffusion of a fluid "particle" (or small coherent
fluid structure) and a discrete sediment particle. Herein, the P -factor is assumed to be
constant over the flow depth.
Carstens (1952) concluded that P < 1 because the sediment particles cannot fully respond
to the turbulent velocity fluctuations, which were assumed to be one-dimensional. Singamsetti
(1966) found P > 1 for two-dimensional eddy motion because of the presence of centrifugal
forces acting on the particles (higher density) causing the particles to be thrown to the outside
of the eddies with a consequent increase of the effective mixing length.
Information of the p -values can also be obtained from the experimental results of Coleman
(1970) who measured concentration profiles and fall velocities in flume and field conditions.
Based on Eq. (7.3. lo), Coleman computed the sediment mixing coefficient from the measured
variables: c, dc/dz and w,, see Figure 7.3.4. Van Rijn (1984b) used the results of Coleman
to determine the p -factor, defined as P = e&efm. The &,--value was computed from
Equation (7.3.14) for z/h = 0.5. The e- -value was determined as the average value of the
&,-values in the upper half of the flow (as given by Coleman, Figure 7.3.4) where the
concentrations are relatively small. The p -factors can be represented by the following
function:
Equation (7.3.17) specifies a value larger than unity indicating a dominant influence of the
centrifugal forces which cause the particles to be thrown to the outside of the eddies with a
consequent increase of the effective mixing length. Given the limited knowledge of the
physical processes involved, it is not advisable to use a P -factor larger than 2.
The +-factor expresses the influence of the sediment particles on the turbulence structure of
the fluid. This effect is extremely important in the upper regime with high concentrations
(c > 10 kg/m3) because this results in stratification and hence damping of turbulence (see
Section 7.3.9).
7.3.4 Concentration profiles
Expressions for the sediment concentration profiles in the lower regime (low concentrations
c < 10 kg/m3) can be obtained by integration of the convection-diffusion equation (Eq.
7.3.10) using relevant expressions for the mixing coefficient. As bed-boundary condition a
reference concentration (c,) at a small distance (a) above the mean bed is applied (integration
constant).
Hindered settling effects and turbulence damping effects can be neglected in the lower regime
( w , , ~= w, and+ = 1).
Integration of Eq. (7.3.10) yields:
constant
linear
parabolic
parabolic-constant rf : for -hz < 0.5
w81(B -4(w,/(p xu,))(z/h-0.5)

(e) for -z 2 0.5
h
in which:
c = concentration at height z above the mean bed
c, = reference concentration at height z = a above bed
h = water depth
w, = fall velocity in clear water
The parameter Z = ws/(P KU,) is termed the suspension number.
Figure 7.3.5 shows the above-given concentration profiles for w,/u, = 0.2, P = 1, K = 0.4
giving Z = 0.5.
Equation (7.3.20) also known as the Rouse concentration profile and Eq. (7.3.21) yield the
best agreement with measured concentrations. Hunt (1953) proposed a concentration profile
slightly different from the Rouse concentration profile. Figures 7.3.6 and 7.3.7 show Eq.
(7.3.20) for different values of Z in comparison with measured concentrations (Vanoni,
1946). Equation (7.3.2 1) yields a finite concentration at the water surface, while Eq. (7.3.20)
gives a zero concentration at the water surface which is less realistic.
Chien (1954) applied Eq. (7.3.20) to determine the 2-parameter from measured
concentration profiles. The measured Z-values (Z = wS/PK u,) were plotted against
computed Z-values (Z = W,/K u,) , as shown in Figure 7.3.8. The results show smaller
measured Z-values. This can be interpreted as a P -factor larger than 1 (P > 1).
Based on Figure 7.3.7, the relative importance of the suspended sediment load can be
determined, see Table 7.8.
- u*
= Ws
- (K =0.4, P = 1) Suspended sediment distribution over the depth
P-* ws
5 0.5 < 0.1 h)

suspended sediment in near-bed layer (z
2 1.25 suspended sediment upto mid depth (z < 0.5 h)
1 2.5 suspended sediment upto water depth
0.1 25 suspended sediment almost uniformly distributed
over depth
Table 7.8 Suspended sediment distribution over the depth

-+
--
~-. using constant&,
using linear E1
(a,= 6 )
(ap=
3)
a
--- -
Figure 7.3.5 Concentration profiles
using parabolic E,
using parabolic -constant Ef
- Relative concentration
Ca
Figure 7.3.6 Concentrationprofiles based on parabolic mixing co@cients, Vanoni (1946)

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
&Bottom Relative concentration, F~
C'
Figure 7.3.7 Concentration profiles based on parabolic mixing co@cient (Z = 1/32 to 4)
Flume studies
River measurements
x Missouri, 1952
0 1 2
- 3
Zcomputod
4
= ws
Figure 7.3.8 Measured and computed 2-values according to Chien (1954)

5 6 7
Influence of bed f o m
When bed forms are present, it is not fully clear whether the grain-shear velocity (US or the
overall bed-shear velocity (IL) should be used to compute the Z-parameter.
Einstein (1950), later followed by Engelund and Fredsee (1976), proposed to use the ul-
value. Thus, they neglect the influence of the vortex zone (downstream of the crest) on the
mixing process of the suspended sediments. As shown in Figure 5.2.1 1b, the vortex zone has
a large effect on the turbulence structure over a major part of the dune. According to Van
Rijn (1984b), these effects cannot be neglected and it was proposed to use the overall &-value
to compute the Z-parameter.
Ikeda (1980) and Ikeda-Asaeda (1983) measured concentration profiles in experiments with
and without bed forms. Their experimental results show a large effect of the bed forms on
the concentration profiles. He observed that the concentration profiles measured in
experiments with bed forms were much more uniform than those measured in an experiment
with a plane bed at the same hydraulic conditions (wJu, 0.6). The bed forms seem to
cause a more intensive mixing of suspended particles.
7.3.5 Velocity profiles in lower regime
In the lower transport regime the sediment concentrations are relatively low (c < 10 kg/m3)
and do not influence the turbulence structure of the flow (no stratification effect).
In case of hydraulic rough flow over a flat bed the velocity profile can be described by a
logarithmic distribution, as follows:
in which:
u = current velocity at height z above mean bed level
u, = current-related bed-shear velocity
z, = zero-velocity level (= 0.033 kb)
k, = equivalent or effective roughness height of Nikuradse
K = Von Karman constant (= 0.4)
In case of a dune-covered bed, which is the most common condition in the lower regime, the
flow basically is non-uniform showing acceleration effects upstream of the crests and
deceleration effects downstream of the crests (see Figure 5.2.1 la). Figure 7.3.9 shows the
measured velocity profiles (as presented in Figure 5.2.1 1a) in one plot. The layer influenced
by the dunes has a thickness of about 1.5 A, with respect to the mean bed level.
The bed form also has a considerable effect on the turbulence structure, as presented in
Figure 5.2.1 lb. The turbulence intensities in and around the vortex zone downstream of the
crest show a considerable increase.
A simple method describing the non-uniform behaviour of the velocity profiles is not
available. Detailed representation of these types of disturbed velocity profiles requires the
application of a hydrodynamic model in combination with a higher order turbulence model
(K-Epsilon model, see Rodi, 1980).
Space-averaged velocity profiles can to a certain extent (above the crest level) be represented
by the method of Nikuradse which is based on the application of an equivalent roughness
height (k),see Eq. (7.3.22). Figure 7.3.9 shows Eq. (7.3.22) for k, = 0.063 m which was
derived from the known ChCzy-coefficient (C = 32 mH/s).As can be observed, the deviations
between measured and computed velocities are large in the region below the crest level, but
are considerably smaller above the crest level (maximum relative error of about 20%).
McLean (1991) proposed a two-layer model for spatially-averaged flow over dunes;
logarithmic functions were used to describe the velocity profiles in both layers. This method
also yields inaccurate near-bed velocities because the horizontal non-uniformity is not taken
into account.
0.30 -
',. water surface
r
-
-
-
u.= 0.53 m/s
0.25 -
I h = 0.33 m
I A = 0.08 m
I
n I h = 1.60m
E I
w I
0.20 - I
V)
I distance
FU I profile
I from crest
Q,
I no
>
0 I (4
1) I
0 0.21
0.15 - I 2
I
r I 3 0.37
.-m
Q) I
I
4 0.60
s I 5 0.82
I
6 0.97
8 1.27
10 1.58
0.05 -
---------
I I
' -0.2 0 0.2 0.4 0.6 0.8 1.O 1.2
---+velocity (m/s)
Figure 7.3.9 Velocity profiles in case of a dune-covered bed

7.3.6 Reference concentration and reference level
1. Flat bed
The most logical assumption for the reference level (z=a) of the sediment concentration
profiie is the upper edge of the bed-load layer (z = 6,J.The reference concentration (c3 is
defined to be equal to the bed-load concentration (c,). Thus,
A good estimate of the thickness of the bed-load layer in case of a flat bed can be obtained
from Eq. (7.2.13) which expresses the saltation height of the bed-load particles (Van Rijn,
1984a). Equation (7.2.13) yields a value in the range of 6, = 2 to 10 d,, .
The layer-averaged sediment concentration (q,)in the bed-load layer was found to be (see Eq.
(7.2.43), Van Rijn (1984a)):
in which:
c, = volumetric bed-load concentration
c, = maximum volume concentration (= 0.65)
T = dimensionless bed-shear stress parameter, see Eq. (7.2.44)
D, = dimensionless particle size parameter, see Eq. (7.2.44)
6, = thickness of bed-load layer, see Eq. (7.2.13)
The concentration is given as a volume concentration; multiplying by p, yields the
concentration by weight (in kg/m3).
Einstein (1950) assumed that the thickness of the bed-load layer was equal to 2 d,,. The
average particle velocity was found to be ub = 11.6 u,. Using the definition qb = c, u, 6,,
the bed-load concentration was found to be:
in which:
qb = volumetric bed-load transport rate, Eq. (7.2.70)
,
- = dimensionless bed-load transport rate
Engelund and Fredsere (1976) assumed a bed-load layer thickness equal to 2 &. The
volumetric bed concentration was found to be:
with:
in which:
8 = particle mobility parameter = u ? / ( ( ~ - l )dM)
~
8, = particle mobility parameter at initiation of motion (Shields)
u. = 3 ( 6 + 2.5 h(h/2.5 d d ) = bed-shear velocity
Zyserman and Fredsnre (1993) proposed an empirical function defined at z = a = 2d, (flat
bed 8' = 8):
Smith and McLean (1977) assumed h ~ the t thickness of the bed-load layer was equal to the
zero-velocity level z., They found (15 = 3 d , = grain roughness):
The average sediment concentration (volume) in the bed-load layer was proposed to be:
in which:
c, = maximum (volume) concentration (= 0.6)
S = (8-8,)/8, = transport stage parameter
6, = thickness of bed-load layer, Eq. (7.3.29)
The results of the proposed methods are compared in Figure 7.3.10 for a specific case (plane
bed, d, = 300 pm, D, = 7.6). The methods of Einstein, Engelund-Fredsse and Van Rijn
yield results which are in reasonably good agreement. The method of Engelund and Fredsse
predicts an almost constant bed-load concentration for 8 > 0.6. The method of Smith and
McLean yields considerably smaller bed-load concentration values, because it is defined at
a higher level.
Using the proposed methods, the reference concentration is prescribed at a reference level
close to the bed (a = 2 to 10 &). This may easily lead to relatively large errors of the
concentrations at higher levels, as will be shown by a computation example (see, also Van
Rijn, 1984b).
Equation (7.3.10) was solved numerically applying Eqs. (7.3.11) to represent the fall velocity
and Eq. (7.3.15) to represent the parabolic-constant mixing coefficient. The suspension
number Z = w,/(P K U J was taken as 0.625. Because of errors in w,, P , K and u., the Z-
value also has a certain error. The error of the Z-parameter was assumed to be f 25%.
Thus, Z = 0.625 f 0.175. Figure 7.3.1 1 shows concentration profiles for Z = 0.45,O. 625
and 0.8. As can be observed, an error of about 25% in the Z-parameter leads to a
concentration error of a factor 3 at a level of z = 0.5 h (mid depth) and an error of a factor
2 at a level of z = 0.1 h.
----- Van Rijn (19840)
- mobility parameter,
Figure 7.3.10 Bed load concentrations
O
2 4
-
6 8 2 4 6 8 a-' 2
dimensionless concentrat~on,
Ca
4
Figure 7.3.11 Influence of error in 2-parameter on concentration profile

6 6 lo0
Based on these results, it is evident that the application of a reference level at the upper edge
of the bed-load layer is not very attractive for reasons of accuracy. It is recommended to use
a minimum value of a- = 0.01 h.
In stead of a prescribed concentration at the bed it is also possible to apply a prescribed pick-
up rate as bed-boundary condition (Van Rijn, 1984d). Pick-up rate functions are given in
Section 7.2.3.
2. Bed forms
As shown in Section 7.3.5 and in Figure 7.3.9, the flow over a dune-covered bed is strongly
non-uniform.
The bed-load particles are transported close to the bed by rolling, sliding and saltating at the
upsloping part of the dunes. Arriving at the top of the dune, the majority of the bed-load
particles jumps over the edge and rolls down the leeside slope of the dune towards the dune
trough where they are buried waiting for a new (transport) cycle.
The suspended -load particles are entrained from the bed-load layer developing at the
upsloping part of the dune. The entrainment rate is maximum at the dune top (large
velocities). Above the dune trough the suspended particles are transported further upwards
by turbulence mixing produced in the shear layers of the vortex generated in the dune trough.
A detailed mathematical representation of the afore-mentioned processes requires the

application of a sophisticated hydrodynamic model and a higher order turbulence model such
as the K-E model (Rodi, 1980) in combination with a convection-diffusion model for the
suspended sediment particles. The horizontal grid size should be much smaller than the dune
length (Ax a A ) ; the reference level should be applied at the upper edge of the bed-load
layer (a a A). To get a first understanding of the phenomena involved, a relatively simple
two-dimensional vertical suspended sediment model was used by Van Rijn (1985) to compute
the concentration profiles for the flow over dunes.
Figure 7.3.12 shows computed concentration profiles at the dune top and at the dune trough
representing two extreme profiles; all other profiles are within the indicated variation range.
As can be observed, the near-bed concentrations can vary over a range of nearly a factor 10.
In vertical direction the concentration variation is confined to a near-bed region of about 0.3
h. The concentrations in the upper layers are not noticeably affected. Atkins et a1 (1989) show
measured concentrations along a sand dune in an estuary.
Generally, the application of mathematical models for estuarine and coastal problems requires
a schematization on a macro-scale (limited computer facilities) applying a horizontal grid size
much larger than the bed-form length. Consequently, a detailed representation of the
concentration profiles and also the velocity profiles is not feasible. The effect of the dunes
on the flow field, usually, is represented by introducing an effective roughness parameter (&)
as shown in Figure 7.3.12.
To represent the effect of the dunes on the concentration field, it is herein proposed:
to represent the concentration profiles by a schematized concentration profile being an
estimate of the spatially-averaged (over the dune length) concentration profiles,
to specify an effective concentration (c3 at a level through the top of the dunes (a = %A
or a = k, when the dune dimensions are unknown),
to represent the transport of particles below the reference level as bed-load transport using
a simple formula.
The effective reference concentration at the top of the bed forms was proposed to be (Van
Rijn, 1984b):
in which:
c, = reference concentration (volume)
a = reference level above the mean bed (a = k, or a =
D, = dimensionless particle parameter
T = dimensionless bed-shear parameter
The constant in Eq. (7.3.3 1) is based on calibration using flume and field data with velocities
in the range of 0.4 to 1.6 m/s, depths in the range of 0.1 to 25 m and mean particle size in
the range of 180 to 700 pm (Van Rijn, 1984). Voogt et al (1991) showed that Eq. (7.3.3 1)
is valid for velocities upto 3 m/s. The reference level is assumed to be equal to half the bed
form height (a = lhA) or equal to the effective bed roughness (a = kJ when the bed form
height is unknown. The k,-value can be computed from the ChCzy-coefficient.
Based on Eq. (7.3.31), Van Rijn (1987) presented a stochastic method to compute the
reference concentration for strongly non-uniform flow.
A different approach was used by Einstein (1950), Engelund-Freds~e(1976) and Smith-
McLean (1977). They assume that the bed-load layer develops at the upsloping part of the
bed forms. The reference concentratioqis given by Eqs. (7.3.25), (7.3.26) and (7.3.30) using
the effective mobility parameter 8' = tb/[(p, - p)g dS0] with t b = = grain-related bed-shear
stress.
An appraisal of existing methods to compute the reference concentration was presented by
Garcia and Parker (1991). The methods of McLean and Van Rijn were found to give the best
results.
7.3.7 Suspended sediment size in case of non-uniform bed material
In natural conditions the bed material is non-uniform. Observations in flume and field
conditions have shown that the bed material and the suspended sediment material have
different particle size distributions. Usually, the suspended sediment particles are considerably
smaller than the bed material particles. Hubbel and Matejka (1959) found that the median size
of the total load (measured in a special contracted section) in the Middle Loup river was
about 250 pm, whereas the median size of the bed material was about 350 pm. Basically, it
is possible to compute the suspended load for any known type of bed material and flow
condition by dividing the bed material into a number of size fractions and assuming that the
size fractions do not influence each other (see Section 7.2.7).
Van Rijn (1984) has applied the size fraction method to determine a representative suspended
sediment diameter (d,) which accounts for the non-uniformity effects. Using the size fraction
method (as proposed by Einstein, 1950), the total suspended load transport was computed for
various conditions, after which by trial and error the representative (suspended) particle
diameter (d,) was determined that gave the same value for the suspended load transport as
according to the size-fraction method.
natural situation xhomatizotion
concentrution prof iks

- concentrution (mgll)
Figure 7.3.12 Concentration profiles along a dune
1.0
-
0
0.8
C
E"
6
-0 c
a 0,6
.=
U C
e
t9 62 0.4
.-
c C
34 % 0.2
g2
f 2 0
0 2 4 6 8 10 12 14 16 18 20 22 24 26
t equation (7.14)
. -1
4 - 2
da4 d
---+
=1.5
transport parameter, T
data
dm= 190pm Us= 1.4
o dSo= 2 7 0 ~ m a,
=
--- G = '2 . 5
adSo= 280pm 4 =
1.8
2.0
0 ho= 320pm U, = 1.8
Figure 7.3.13 Representative particle diameter of suspended sediment (Van Run, 1984b)
In all, six computations were performed using two types of bed material with a geometric
standard deviation: a , = lh(d,/d, + d,/dld = 1.5 and 2.5. The d, of the bed material was
equal to 250 pm. The mean flow velocities were 0.5 and 1.5 m/s. The flow depth was
assumed to be 10 m. The concentration profile was computed by applying Eq. (7.3.21) with
p = 1, 4 = 1 and K = 0.4. The reference concentration according to Eq. (7.3.31) was
applied at a level of z = 0.05 h. The flow velocity profile was assumed to be logarithmic.
The computed d,-values were found to be related to a,-parameter and to the dimensionless
shear-stress parameter T, as follows:
for 0 < T < 25
ds = *SO for T 2 25
in which:
d, = representative particle diameter of suspended sediment
d, = median particle diameter of bed material
a, = geometric standard deviation of bed material = % (d,/d, + d,/d,,)
T = dimensionless bed-shear stress parameter, Eq. (7.2.44)
Equation (7.3.32) and experimental data of Guy et al(1966) are shown in Figure 7.3.13. The
scatter of the data is too large to detect a clear influence of the a,-parameter. In an average
sense the agreement between measured and computed values (a, = 2.5) is quite good.
Using this method, a better representation of the suspended sediment in case of graded bed
material can be obtained than by taking a particular particle diameter such as the dl,, d,, or
d, of the bed material.
Morse and Townsend (1989) also adressed the modelling of sediment concentration profiles
in river flows carying mixed sediments. The grain size distribution of the bed material was
divided into several size ranges (fractions), which were then characterized by their respective
geometric means. Numerical errors introduced through this procedure were quantified. Depth-
averaged concentrations can be under-estimated by a factor 10 when the concentration
gradients are large (Z-parameter is large). Estimates were found to be particularly sensitive
when the chosen reference level was small (close to bed) and the fraction width was large.
Fall velocity coefficients were introduced to correct the discrepancy between the actual
representative fall velocity of a fraction and that of the geometric mean size of the fraction.
7.3.8 Suspended load transport rates
Herein, the methods of Einstein (1950), Bagnold (1966), Bijker (1971) and Van Rijn (1984b)
are described.
9. Einstein (1950)
The method of Einstein is based on a parabolic distribution of the fluid mixing coefficient and
a logarithmic distribution of the velocity.
The suspended sediment transport rate can be expressed as:
in which:
= volumetric current-related suspended load transport (m2/s)
G?
u, = current-related bed-shear velocity due to the grains (mls)
c, = reference concentration (volume) = qd(11.6 u: a)
a = reference level (= 2 d), (m)
h = water depth (m)
d = particle diameter (m)
A = alh = dimensionless reference level (-)
X = zlh = dimensionless vertical coordinate (-)
Z = w,/(K u,) = suspension number (-)
e = correction factor (-)
According to Einstein, the suspended load transport is related to the grain-shear velocity(u!)
and not to the overall shear velocity (a).
The reference concentration is determined from the bed-load transport, assuming a bed-load
layer thickness equal to 2 particle diameters.
For almost uniform bed material the $, is taken as the representative particle diameter. For
graded bed material the size fraction method should be used.
The I, and I,-integrals can be determined graphically (see Einstein, 1950) or numerically.
The method of Einstein is not given in full detail here, because of its complexity.
Furthermore, an extensive verification study of White et al (1973) has shown that the
predicting ability of the Einstein method is much less than other (more simple) methods.
2. Bagnold (1966)
The method of Bagnold is based on a energy balance concept relating the suspended load
transport to the work done by the fluid.
The suspended load per unit area is supported by a fluid stress (a) normal to the bed
a = (ps- p)g V, cosp in which: V, = solid volume of suspended load per unit area (m3/m2).
Per unit time the suspended load sinks over a vertical distance equal to the fall velocity w,.
To keep the load in suspension (all particles remain statistically at the same level above the
bed), the normal stress (a) must do work (per unit time and area). The work required is:
The volumetric suspended load transport is defined as $,c = V, giving:
The work done per unit time and width by the longitudinal gravity force component to keep
the fluid in motion is z, ii. Part of this energy is available to transport the suspended load,
yielding:
-
in which:
u = depth-averaged velocity
r, = bed-shear stress
e, = efficiency factor related to suspended load (= 0.01 to 0.02)
% = efficiency factor related to bed load (= 0.1 to 0.2)
Per unit time and width the suspended load is moved in longitudinal direction over a distance
equal to G. The work done by the longitudinal gravity component during the unit time period
is:
This quantity is fully available for transportation of the suspended load.
The suspended load transport rate can now be derived from the energy balance
+
W, = W , , Wa,2,yielding:
in which:
q , = volumetric current-related suspended load transport (m2/s)
-
w,
u
= fall velocity of sediment (mls)
= depth-averaged velocity (mls)
t, = overall current-related bed-shear stress (N/m2)
tanp = bottom slope (-)
pS = sediment density (kg/m3)
g = gravity acceleration (m/s2)
% = efficiency factor of bed load ( = 0.1 to 0.2)
es = efficiency factor of suspended load (= 0.01 to 0.02)
p = bottom angle with horizontal datum
Kachel and Sternberg (1971) have shown that the efficiency factors (e, and %) are not
constant, but are strongly related to the bed-shear stress and the particle diameter. Basically,
these results show that the validity of the Bagnold concept is questionable.
3. Bijker (1971)
Based on the concept of Einstein (1950), Bijker (1971) proposed:

in which:
cg,, = suspended sediment transport rate (m2/s)
qb,, = bed load transport rate (m2/s)
I, = integral according to Eq. (7.3.34)
I, = integral according to Eq. (7.3.35)
a = k, = reference level
The current-related bed-load transport rate (m2/s) is expressed as:
in which:
u+ = overall bed shear velocity
8 = mobility parameter
p = (CIC ')I5 = bed form factor
C = overall ChCzy coefficient
C1 = grain-related Chkzy coefficient = 181og(12h/&)
b = coefficient ( = 1 to 5)
Because the reference level is assumed to be equal to the bed roughness height (a = kJ, the
ratio g,$/qb,can be expressed as a function of ihe parameter Z and kh, as shown in Figure
7.3.14.
4. Van Rijn (1984b)
The suspended load transport can be computed from Eq. (7.3.2), yielding:
Using Eqs. (7.3.21) and (7.3.22), the F-factor is:
Equation (7.3.44) cannot be integrated analytically. An approximate solution accurate to about

25% for 0.3 I Z' I 3 and 0.01 I alh I 0.1 is given by:
The F-factor for a/h = 0.01, 0.05 and 0.1 is shown in Figure 7.3.15.
-WL2 5 1U4 2 5 1U3 2 5 10-2 2 5 ~3-1
d ratio of roughness height and dapth, k,/h
Figure 7.3.14 Ratio of suspended load and bed load transport, Bijker
The complete method reads as (see also Appendix A):
in which:
= reference concentration (-)
F = (a/h)" - (a/h)12 = shape factor (-)

(1 - a/h)" (1.2 - z')
= particle parameter (-)
= bed-shear stress parameter (-)
= current-related effective bed-shear stress (N/m2)
= current-related overall bed-shear velocity (m/s)
.I:;[
C' = 18 log -
-
= grain-related ChCzy coefficient (m0.5/s)
= overall ChCzy coefficient (m0.5/s)
= critical bed-shear stress (N/m2)

according to Shields
= suspension number (-)
= suspension number (-)
= stratification correction, Eq. (7.3.86)
= ratio of sediment and fluid mixing coefficient

(P, = 2)
d, = [I + 0.011 (us-1) (T-25)]dm = representative particle size of suspended sediment (m)

d, = dm for T 2 25
%,c
= volumetric current-related suspended load transport (m2Is)
u = depth-averaged velocity ( d s )
h = water depth (m)
a = reference level (m), a = %A or a = k,
k = overall roughness height (m)
A = bed form height (m)
d, = median particle diameter of bed material (m)
dl,, d,, d, = characteristic diameter of bed material (m)
Wa = fall velocity of suspended sediment (based on d,-value) using Fig. 3.2.6 or
Eqs. (3.2.21), (3.2.22), (m/s)
d8 = representative particle size of suspended sediment (m)
= geometric standard deviation of bed material, o, = 1/2($4/d9 + d5dd$
C0 = maximum concentration (= 0.65)
s = specific density (= p,/p)
PS = sediment density (= 2650 kg/m3)
P = fluid density (kg/m3)
The $ -factor yields an increase of the suspended load transport rate, whereas the *-factor
yields a decrease of the transport rate. In the lower regime both effects cancel out ($ = 1,
qr = 0).
A simplified method was given by Van Rijn (1984b). This method is based on computer
computations (using the detailed method, see above) in combination with a roughness
predictor (see Chapter 6). Using regression analysis, the computational results for a depth
range of 1 to 20 m, a velocity range of 0.5 to 2.5 m/s and a particle range of 100 to 2000
pm were represented by a simple power function, as follows:
in which:
%,c
= volumetric suspended load transport (m2/s)
u, = critical depth-averaged velocity according to Shields, see Eq. (7.2.47)
h- = water depth (m)
Equation (7.3.46) only requires ii, h, d, d, and v as input data and can be used to get a
first estimate of the suspended load transport.
When the bed load transport and the suspended load transport are known, the total load
transport of bed material can be determined by summation (q = q, Q).+
The ratio of the suspended and total load transport can be expressed as:
For reasons of simplicity the bed load transport is here defined as:
in which:
- = reference concentration
c,
u, = effective transport velocity of bed-load particles
a = reference level (= bed-load layer thickness)
Substitution of Eq. (7.3.43) and (7.3.48) in Eq. (7.3.47) yields:
The ratio may be identified as the ratio of the average transport velocity of the bed load
and suspended load particles, which varies from about 0.4 for large, steep bed forms in the
lower regime to about 0.8 for flat bed conditions in the upper flow regime. Figure 7.3.16
--
shows the ratio of the suspended load and the total load as a function of the ratio of the bed-
shear velocity and particle fall velocity for uJu = 0.6, K = 0,4, P = 1 and a/h = 0.05. An
empirical relationship given by Laursen (1958) and some data of Guy et a1 (1966) are also
shown.
- u*,c/ws
Figure 7.3.16 Ratio of suspended and total load transport

7.3.9 Stratification effects in high-concentrationsuspensions
1. StrQapcation and hyperconcentrations
A flow system with a vertical density gradient is called:

stable when the density decreases upwards,
unstable when the density increases upwards,
neutral when the density is constant over the depth.
In sediment suspension above a movable bed the sediment concentrations decrease in upward
direction from the bed. Consequently, the flow is stable. In uniform flow with an equilibrium
concentration profile the solid particles at a certain level statistically remain at that level,
because the turbulence forces produced by the shear action in the flow are continuously doing
work as they try to mix the sediment upwards against the action of gravity. This reduces the
turbulence energy available for mixing the fluid resulting in damping of turbulence.
The stability of the suspension can be quantified by the Richardson number or the Monin-
Obukhov number. This latter parameter (herein used) is defined as the ratio of the energy per
unit volume required at a particular level to keep the sediment in suspension and the available
turbulence kinetic energy supplied by the shear action at the same level, as follows:
in which:
c = volume concentration at height z above bed
WSpm = fall velocity in fluid-sediment mixture
&f = fluid mixing coefficient in clear water (= K u, z)
duldz = fluid velocity gradient at height z (= LL/KZ)
t = shear stress at height z (= p ef duldz)
Using the 5-parameter, the suspension is stable for 4 2 0. The 5-parameter has been used
extensively in studies of the velocity profile in the atmospheric boundary layer (Businger et
al, 1971). The effect of the suspended sediment on the near-bed velocity profile in terms of
the 5-parameter was discussed by Soulsby and Wainwright (1987).
Stratification effects are most important in the high-concentration or hyperconcentration
range.
Volume concentrations in the range of 0.2 to 0.6 (20% to 60 vol %) are herein called
hyperconcentrations. For concentrations larger than 0.2 the hindered settling effect will have
a dominating effect on the suspension characteristics, as shown by Winterwerp et al (1990)
and by Chien-Zhaohui (1986). The sediment load will move more or less as a neutrally
buoyant load with the same behaviour as a pure liquid (pseudo one-phase flow). The flow
resistance is only slightly larger than that of a clear water flow in the turbulent rough regime.
Hyperconcentrations in the range of 500 to 1500 kg/m3 were observed at velocities of 6 to
8 m/s in the middle reach of the Yellow river in China (Zhaoyin, 1990). Because of the high
viscosity, the flow behaviour was almost laminar. In the turbulent regime the velocity profiles
were observed to be logarithmic.
Hyperconcentrations in the Yellow river consist of a range of size classes. About 40% of the
material is smaller than 50 pm.
Based on flume experiments (Zhaoyin, 1990), the maximum hyperconcentration of uniform
particles was found to vary from about 1000 kg/m3 for fine particles (d < 150 pm) to about
200 kg/m3 for coarse particles (d = 5000 pm). Higher values upto 1250 kg/m3 were obtained
when graded. sediments were suspended.
2. Velocity profile and fZo w resistance
Based on the Boussinesq hypothesis, the shear stress at height z in a fluid-sediment mixture
flow is defined, as:
in which:
tz
= fluid shear stress at height z above bed
E~~
= mixing coefficient in fluid-sediment mixture
v, = kinematic viscosity coefficient in fluid-sediment mixture (= q,/pm)
u = fluid velocity at height z above bed
pm = p (1 +(s-1)c) = density of mixture at height z
c = sediment concentration at height z
q m = dynamic viscosity coefficient
The shear-stress at height z and the bed-shear stress can also be expressed as:
in which:
I = energy gradient
h- = flow depth
= p(l+(s-l);,) = mean density of layer above height z to surface
-Pa 1
-pm == mean
1 (s- 1 = mean density over depth
-c,c = mean concentration of layer above height z to surface
concentration over depth
s = specific density (= p,/p)
p, = sediment density
p = fluid density
Equations (7.3.52) and (7.3.53) can be expressed as:
The velocity gradient can be obtained from Eqs. (7.3.5 1) and (7.3.54):
For a clear fluid Eq. (7.3.55) reduces to:
In the fully turbulent logarithmic layer in the near-bed region the viscosity can be neglected
and the fluid mixing coefficient can be expressed as: e, = r u, z(l -z)/h) , yielding:
in which 1, = KZ = mixing length.

In the viscous sublayer adjacent to the (p1ane)bed the velocity gradient is given by:
In a fluid-sediment mixture the turbulence structure and hence the velocity profile and the
flow resistance are modified.
Experimental work was done by Vanoni (1946), Einstein-Chien (1955), Ippen (197I),
Coleman (1980), Wang (198 1) and Winterwerp et al (1990). Most experiments were done in
the concentration range below s 0.1. Wang and Winterwerp et al did experiments with
c
concentrations upto = 0.4.
Analysis of measured velocity profiles in sediment suspensions shows a decrease of the near-
bed velocities and an increase of the near-surface velocities, as shown in Figure 7.3.17 (Lau,
1982 and Woo et al, 1988). Clearly, an increase of the velocity gradient in the outer region
(zlh > 0.05) can be observed.
The effect of the suspended sediment on the flow resistance is not very clear. A decrease of
the flow resistance (Itakura-Kishi, 1980) as well as an increase of the flow resistance
(Winterwerp et al, 1990) has been reported.
Defining the flow resistance as,
in which:
r, = bed-shear stress
f, = friction factor of fluid-sediment flow
m! = 'mixture density = p(1 + (s- 1) 9
c = depth-averaged concentration
p = fluid density
J
a Etnstatn and Ctwn Doto
S-13
s-a s-12
Figure 7.3.17 Measured velocity profiles in high-concentration flows of

Einstein- Chien (1955)
Figure 7.3.18 Friction factor

it follows that both the mixture density and the friction factor are important. The mixture
density increases for increasing concentration resulting in an increase of flow resistance. The
behaviour of the friction factor is presented schematically in Figure 7.3.18, showing:
no change of friction factor in fully rough regime,
small decrease of friction factor entering the transition regime from the rough regime
(decreasing Re-number) at the same relative roughness (&ID),
increase of friction factor for decreasing Re-number in smooth regime,
small decrease of friction factor by transition from smooth to laminar regime,
increase of friction factor for decreasing Re-number in laminar regime.
Analysis of the high-concentration mixture flows of Einstein-Chien (1955) and Winterwerp

et al (1990) shows u,-values in the range of 0.1 to 0.2 mls, h-values in the range of 0.001
to 0.005 m (see also Section 6.2.3). The viscosity of the fluid-sediment mixture in the near-
bed region with concentrations of about 0.4 will be about 10' m2/s. Based on these values,
the Re, = u, k,/v, number will be in the range of 10 to 100, which means flow in the
transition regime (Re, = 5 to 70) or in the rough regime (Re, > 70), depending on the
hydraulic conditions.
The stratification phenomena affecting the velocity profile and the flow resistance can be
described as:
damping of turbulence because turbulence energy is consumed in keeping the particles in

suspension (transfer of momentum from fluid to particles). This effect which is related to
the concentration gradient (maximum in near-bed region) can be interpreted as a (slight)
decrease of the Reynolds number. This will not influence the friction factor in the rough
regime; the friction factor in the transition regime may be slightly reduced, whereas it will
be increased in the smooth regime.
damping of turbulence by increased viscous dissipation. This effect is directly related to

the value of the concentration. In high-concentration suspension (c = 0.2 to 0.4) the
effective viscosity increases by a factor 5 to 10 resulting in an increase of the viscous
sublayer thickness. In the smooth regime this effect can be interpreted as an increase of the
effective roughness (z, = 3.3 v,/u,) and hence as an increase of the flow resistance.
reduction of mean flow energy by transfer of momentum from the fluid to the particles (in
the longitudinal direction) to keep the solid mass in motion. This effect is related to the
concentration value results in an increase of the flow resistance (increase of f-factor).
The overall effect probably is a slight decrease of the flow resistance at low concentrations
(c < 0.1) near the transition regime (Re, = 50 to loo), as observed by many resarchers
(Vanoni, 1946) and an increase of the flow resistance at high concentrations (c > 0.1) near
the smooth regime (Re, = 5 to 50) as observed by Winterwerp et al (1990).
A very informative discussion about the effect of suspended sediment on the flow resistance
has been given by Chien and Zhaohui (1986), based on Chinese research related to
hyperconcentrations.
In the fully rough regime almost no influence was found of hyperconcentrations on flow
resistance. A slight decrease of flow resistance was found when hyperconcentrated flow
entered the transition regime (Re = 2000 to 4000) from smooth to laminar flow due to an
increase of the concentration and hence viscosity resulting in a sudden damping of turbulence
(Chien-Zhaohui, 1986; Zenghai, 1988). In the fully laminar regime (Re > 2000) the friction
factor increased with (hyper)concentration due to increase of viscosity and hence decrease of
Re-number. These observed phenomena can be explained from the behaviour of the friction
factor (see Fig. 7.3.18).
The turbulence damping effect was clearly observed by Xingkui and Ning (1989). Based on
turbulence measurements obtained with a pressure probe, they found a consirable reduction
(factor 2) of the- near-bed turbulence intensity from = 2u, to lu, for a volume
concentration increasing from 0 to 8 % .
Many researchers have tried to model the effect of the suspended sediment on the velocity
profile.
Roughly, four types of approaches can be found in the literature:
modification of the mixing length scale (lJ or the fluid mixing coefficient (E, = 1, u,)
by reducing the K,-value of the fluid-sediment mixture; K, is either constant over the
depth or varies with z, Einstein-Chien (1955), Ippen (197 l), Yalin-Finlayson (1972), Wang
(1981), Van Rijn (1984b), Winterwerp et al (1990),
modification of the mixing length scale),,I( by introducing the Monin-Obukhov length scale
(Lo), Itakura-Kishi (1980), McLean (1991a,b),
K Z
,1 = and K = constant
l+yz/L,o
modification of the mixing length scale (u

by introducing the Coles wake function ( a =
wake strength parameter), Coleman (1981), Valiani (1988)
K
lm = and u = constant
-'
z +(n a/h) sh(xz/h)
modification of the mixing coefficient (E,) by using a higher order turbulence model
(k-Epsilon model with equations for the turbulent energy and its dissipation rate),
DeVantier and Larock (1983).
in which: urn = Von Karman coefficient in fluid-sediment mixture, K = Von Karman

coefficient in clear fluid; y = coefficient (5 to lo), a = wake strength parameter (0.1
to 1).
In most methods the L-value or the &,,-value is reduced by the suspended sediment resulting
in an increase of the velocity gradient (dutdz = u,/ld.
Einstein and Chien (1955) found that the K,-value of the fluid-sediment mixture is related
to the ratio of the energy required to keep the sediments in suspension and the total available
energy. This can be expressed as:
-
in which:
-
c = depth-averaged volume concentration
w, = fall velocity of suspended sediment
s = specific density (= pJp)
I = energy gradient
Equation (7.3.63) shown in Fig. 7.3.19, yields values from 0.4 to 0.2 for increasing
concentrations.
Ippen (1971) considered both the effect of turbulence damping by viscous dissipation and
mixing against gravitation forces. He found:
in which:
c, = bed-load concentration (volume),
Equation (7.3.64) yields values from 0.4 to 0.2 for increasing &-values upto 0.4.
Based on energy considerations and analysis of velocity profiles measured in the Yellow
river, Wang (1981) found:
-ws,which:
in
= fall velocity of suspended sediment in a fluid-sediment mixture (based on q.
Wang explicitly took into account the transfer of potential energy into kinetic energy
- - with the longitudinal flow of the solid mass. This is expressed by the term
associated
w,, - u I, yielding
a decrease of xm for increasing concentrations, if GS, - ii I > 0
an increase of tern for increasing concentrations, if Gsm - ii I < 0
- fall velocity is greatly reduced by the
At very high concentrations (hyperconcentrations) the
.hindered settling effect. This may lead to Es, - u I < 0, implying transport of solids
without energy supplied by bottom-related turbulence. The energy is now supplied by the loss
of potential energy of the solid mass during transportation. This stage was termed auto-
suspension by Bagnold (1962).
Wang found ,K -values between 0.4 and 0.1 for volume concentrations upto 0.4.
Winterwerp et al (1990) observed that the measured velocity profiles had a logarithmic
behaviour in the outer region (zlh > 0.1). The velocity profiles were fitted by (see Fig.
7.3.28):
d K
-
Figure 7.3.19 K, -values according to Einstein- Chien (1955)
l I I
-
5 I
10 100 1000 10.000 10 20 XI U)
Relati= Verticai Position %+ C w c e n t r a h ? (%)
Figure 7.3.20 Measured velocity and concentration profiles, Wintenuelp et al (1990)

Figure 7.3.21 K,-values according to Wintenverp et a1 (1990)
The slope of the velocity profiles (Fig. 7.3.20) first decreases with increasing concentrations
upto c = 0.2 indicating an increase of turbulence damping (decrease of %-value). A further
increase of the concentration (E = 0.2 to 0.4) results in a steepening of the slope of the
velocity profile upto the clear water value again (increase of %-value). Figure 7.3.2 1 shows
the behaviour of the K~-valuesas a function of E. A shift of the logarithmic profiles with
increasing concentrations was also observed indicating an increase of the flow resistance. This
is caused by an increase of the thickness of the viscous sublayer related to an increase of the
effective viscosity and by additional friction related momentum transfer from the fluid to the
solid particles (particles are accelerated by the fluid and retarded by grain-grain collisions and
grain-bed collisions).
Van Rijn used a logarithmic velocity distribution for the transitional flow regime to compute
the velocity distribution of Test 57 (mean concentration c = 0.11) and Test 83 (c = O M ) ,
experiments of Winterwerp et al(1990) and Run S-15 (E = 0.030), experiments of Einstein-
Chien (1955). The velocity distribution is given by (Van Rijn, 1990):
-u which:
in
= depth-averaged velocity
h = waterdepth
z, +
= 0.033 k, 0.11 vJu, = zero-velocity level
18
k. = 3 d, = effective grain roughness, Eq. (6.2.12b)
8 = u,/((s-l)g d& = mobility parameter
v, = kinematic viscosity coefficient of fluid-sediment mixture near bed ( = m2/s)
u, = (g h = bed-shear velocity
Measured and computed velocity distributions are shown in Fig. 7.3.22. Reasonably good
agreement can be observed. The maximum relative difference is about 25 % .
Based on this, it is concluded that the velocity profiles in high-conce~trationflov~scan be
described reasonably well by generally-accepted expressions for turbulent flow.
/
Einstein-Chien 1955 I
/
Run 5-15 I
.
- G = 2.2 m/s / ,
h
1
=0.124 m
= 0.0168
I'
/
/
..
- dS0= 280pm r-
.
dgO = 380 p m W'
/
E = 0.03
,.&
'
-
)
,---#
1
0 0.5 1 .O 1.5 2.0 2.5
flow velocity, u (m/s)
Figure 7.3.22 Measured and computed velocity profiles in high-concentration flows

Itakura-Kishi (1980) and McLean (1991a,b) assumed a constant K-value equal to the clear
fluid value (K, = K = 0.4). The effect of the suspended sediment on the velocity profile
was modeled by introduction of the Monin-Obukhov length scale (Lo), as follows:
-c which:
in
= depth-averaged concentration
w, = fall velocity in clear fluid
K = Von Karman coefficient in clear fluid (= 0.4)
The velocity gradient is given by:
Based on analysis of measured velocity profiles in suspensions (5 < 0.05), the y -coefficient
was found to be y = 7.
The velocity profiles are expressed as:
for -
u*ks > 70
"m
in which:
v, = viscosity coefficient of mixture = v (1 +2.5 cb)
cb = bed-load concentration
The depth-averaged velocity can be obtained by integration over the depth:
for u*4< 5
-
--6+-+ln[e)
u
-
u - 7h
2d
1 for' - > 70
* '
"m
(7.3.73)
The extra term (7h12~ L) in Eqs. (7.3.72) and (7.3.73) results in an increase of the mean
velocity' in a sediment-laden flow with respect to a clear flow at the same %-value. This
means a decrease of the friction factor. The method is valid for mixtures, with depth-averaged
concentrations upto 0.05.
To apply Eqs. (7.3.72) and (7.3.73), the depth-averaged concentration z and the bed-load
concentration cb must be known. Generally, the ;-value is not known because the velocity
profile will influence the concentration profile. This effect was also studied by Itakura-Kishi
(1980), see next Section.
An iterative solution method is required to compute both the velocity and concentration
profile.
McLean (1991a, b) also used the Monin-Obukhov concept based on a local value of the
concentration in stead of the depth-averaged concentration. The effect of the suspended
sediment on the viscosity was also taken into account.
The fluid and sediment mixing coefficients in a mixture were modelled as:
in which:
If = fluid mixing coefficient (clear water)
y ,P,o = coefficients
5 = stratification parameter based on the Monin-Obukhov concept
A numerical iterative method is required to compute the velocity and concentration profile.
Coleman (1981) analysed his own data (maximum concentration = 0.005) and re-examined
the data from earlier experiments. He found that the change in K-valueas observed by others
was caused by the incorrect application of the logarithmic velocity distribution in the outer
flow region where the logarithmic profile is not really valid. By applying the logarithmic
profile to the near-bed region (where it is valid), Coleman found that the presence of the
suspended sediments (c < 0.005) had no effect on the K -value (thus, K, = K = 0.4). The
velocity profiles in the outer region were found to be affected by the wake-effect. The
velocity distribution valid for the whole depth is given by:
in which:
a = wake strength parameter
B = coefficient depending on hydraulic conditions
K = constant of Von Karman ( = 0.4)
h = water depth
The effect of the suspended sediment on the fluid viscosity was taken into account. The B-
coefficient represents the effect of bottom roughness and the effect of the solid particles on
the flow. Close to the bottom where the sine function is insignificant, the wake function does
not affect the velocity profile. Equation (7.3.76) was fitted to measured velocity profiles.
Coleman found that K was not affected by the suspended sediment and equal to its clear water
value. The only effect of the suspended sediment is to increase the wake strength parameter
( a ) . The a -value varied between a = 0.19 for clear water and a = 0.86 for a suspension
with c = 0.005. Valiani (1988) reanalysed the data of Coleman using the same velocity
profile, Eq. (7.3.76). Valiani used an (objective) analytical-numerical fitting method, whereas
Coleman used a graphical fitting method. Valiani found that the K,-value decreased for
increasing concentrations (K, = 0.4 to 0.2). The B-coefficient was also found to be related
to the suspended sediment concentration. The wake strength parameter a was only poorly
affected by the sediment concentrations. The results of Valiani (using the same data set) are
different from those of Coleman. Valiani stressed the importance of accurate measurements
in the near-bed region.
Umeyama and Gerritsen (1992) modified the mixing length scale, as follows:
in which:
K = Von Karman constant in clear water (0.4)
a = 112 + 112 P (c1c.J = coefficient (c = 0 yields a = 112.)
c = concentration
c, = reference concentration
p = coefficient
The P -coefficient was determined by fitting with measured velocity profiles and found to
increase with the concentration. A general relationship of the P -coefficient was not given.
A similar method was proposed to compute the concentration profiles.
3. Concentration profiles and transport rates
In the upper regime with high or hyperconcentrations the effects of the suspended sediments
on the density, viscosity, on the (hindered) fall velocity and on the turbulence structure
(damping) cannot be neglected.
Assuming that the hindered settling effect can be represented as w,,, = (1-c)'w, and that
damping factor @ is constant over the depth (see Eq. (7.3.16)), an analytic solution of the
diffusion equation can be obtained for a parabolic mixing coefficient, as follows:
inwhich:
c = concentration at height z
c, = reference concentration
a = reference level
h = water depth
Z = w,/(P 4 K u,) = suspension number
4 = turbulence damping factor (constant over depth)
p = ratio of sediment and fluid mixing (constant over depth)
K = Von Karman constant (= 0.4)
w, = fall velocity of suspended sediment in clear water
A numerical solution method is required when the turbulence damping effect varies with the
concentration or concentration gradient ( 4 not constant). An iterative solution method is
required when the damping effect is related to the velocity gradient; both the concentration
profile and the velocity profile must be computed simultaneously.
The effect of the suspended sediments on the concentration profile was studied by Einstein-
Chien (1955), Ippen (1971), Itakura-Kishi (1980), Van Rijn (1984b), DeVantier-Larock
(1983), Woo et a1 (1988), Winterwerp et a1 (1990) and McLean (1991a, b).
Einstein and Chien (1955) used the following set of equations to compute the concentration
profile:
The velocity gradient and the depth-averaged concentration (c) must be known. Hence, an
iterative solution method is required.
Ippen (1971) used a modified velocity gradient to compute the mixing coefficient distribution.
The velocity gradient was modeled as:
The cp -parameter is an empirical parameter which was found to be dependent on the K,-
value (according to Eq. (7.3.64)) and the Z-parameter.
Good agreement between computed and measured (data Einstein-Chien, 1955) concentration
profiles was obtained.
Itakura and Kishi (1980) introduced the Monin-Obukhov length scale and proposed:
in which:
c, = reference concentration at height z=a above bed
4 = correction factor = 7 h / h o
L;, = Monin-Obukhov length scale, Eq. (7.3.68)
Z = suspension number = w,/(K u,)
K = Von Karman constant in clear water (=0.4)
w, = fall velocity in clear water
h = water depth
Iterative computations are required because h0depends on the mean concentration c .

McLean (1991a, b) also proposed a (numerial) method based on the Monin-Obukhov concept.
DeVantier and Larock (1983) studied high-concentration effects by introducing a higher-order

turbulence model (K-Epsilon model) to model the effect of the suspended sediment on the
mixing coefficient. This sophisticated numerical approach offers the most promissing
representation of the physical phenomena involved.
Woo et a1 (1988) modeled the effect of the sediment concentrations on the fall velocity
(hindered settling), on the density and on the viscosity. The velocity gradient was determined
from measured velocity profiles (curve fitting). The computed concentration profiles showed
good agreement with the measured profiles of Einstein-Chien (1955). The method of Woo et
a1 (1988) is not predictive because the velocity profiles must be known to determine the
velocity gradient.
Winterwerp et al (1990) found that the concentration profiles in suspensions with mean
concentrations in the range of c = 0.2 to 0.4 can be represented by the Rouse profile u s i n g ~
= 0.4 and w,,, = w, ( l - ~ )as
~ ,shown in Figure 7.3.20. The Rouse concentration profile was
solved by iteration as w,,, varies with concentration. The turbulence damping effect as
expressed by the modified K,-value was not taken into account.
Van Rijn (1984b) assumed that:
K, = @ K and E, = p4 E, (7.3.83)
in which @ = damping factor dependent on concentration and P = ratio of sediment and

fluid mixing, see Eq. (7.3.16).
The fluid mixing coefficient (E,) in clear water was modeled by a parabolic-constant
distribution according to Equation (7.3.15).
Using w,,, = (1-c)' w,, Equation (7.3.10) was solved numerically. The @-factor was
determined by data fitting using the experimental results of Einstein-Chien (1955), Barton-Lin
(1955) and Vanoni-Brooks (1957), yielding:
for c>0.0010rc>2.5kg/m3
in which c, = maximum volume concentration (= 0.65).
- Equation (7.3.84) is shown in Figure 7.3.23. Values reported by Yalin-Finlayson (1972) are
also reported. Computed and measured (Run S-15, Einstein-Chien, 1955) concentration
profiles are shown in Fig. 7.3-24. The reference concentration was assumed to be equal to
the concentraion measured in the lowest sampling point (c, = 625,000 ppm at a = 0.005 m).
The Rouse concentration profile with @ = 1 (no damping effect) yields concentrations which
are much too large.
The proposed method is not very suitable for practical use because it requires numerical
computations. Therefore, a simplified method based on Eq. (7.3.21) in combination with a
modified suspension number Z' was introduced by Van Rijn (1984b). The modified 2'-value
is defined as:
- dlrnens~onloss concentration,
Figure 7.3.23 $-factor according to Van Rijn (1984b)

A
co
6.16' measured
NIT ---- Rouse concentration profile ( + = I )

+-
.c
0 16'
- computed using Eq. 7.3.86
.c
-.- computed using Eq. 7.3.84
Ica=625003ppm
105 106
-----+concentration, c (pprn)
Figure 7.3.24 Measured and computed concentration profile for Run S-15 of
Einstein-Chien (1955)
110-I8
6
5
-..-
- Engelund-Hansen
Van R i j l
4
3
2
10-21 I I I I I I I I
0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
depth-averaged flow velocity (rn/s)
Figure 7.3.25 Measured and computed sand transport rates at velocities up to 3 m/s
in which:
Z = w,/ QKU,
$ = stratification correction parameter
w, = fall velocity suspended sediment transport in clear water
K = Von Karrnan constant (=0.4)
Q = ratio of sediment and fluid mixing, Eq. (7.3.17)
The $ -parameter represents the turbulence damping effect and the hindered settling effect.
The $-parameter was determined by a trial and error method which implied the numerical
computation of the concentration profiles for a range of hydraulic conditions and the
determination of the $-value giving the same concentration profile as that acording to the
numerical computation. Based on analysis of the available $-values, the following
relationship was obtained (for c > 0.001 or c > 2.5 kglm3):
w* s 1
for 0.01 s - and 0.01 s a,/h s 0.1
u*
in which:
c, = reference volume concentration, use Eq. (7.3.3 1)
c, = maximum volume concentration ( = 0.65)
Figure 7.3.24 shows an example of the concentration profile according to the simplified
method for Run S-15 of the Einstein-Chien data.
The simplified method is valid for depth-averaged concentrations upto c = 0.1.
Voogt et al (1991) have shown that the method of Van Rijn can be used to compute the sand
transport rates for velocities upto 3 mls, as shown in Figure 7.3.25. The methods of Ackers-
White (1973) and Engelund-Hansen (1967) also yield good results. The Van Rijn method
without the stratification corection ( q =0) is also shown in Figure 7.3.25. The suspended
load transport without the $ -correction is a factor 1.8 higher for ii = 1 m/s and a factor 2.8
higher for ii = 2.5 m/s.
7.4 Total load transport
7.4.1 Prediction methods

Herein, the following methods are described:
Einstein (1950)
Bagnold (1966)
- Engelund-Hansen (1967)
Bijker (1971)
Ackers-White (1973, 1990)
Yang (1973)
Van Rijn (1984b)
The methods of Engelund-Hansen, Ackers-White and Yang donot make a distinction between
bed load and suspended-load transport, but directly give the total load transport rate.
Computed transport rates according to the methods of Engelund-Hansen and Van Rijn for
specific cases are given in Appendix A.
1. Einstein (1950), Bagnold (1966), Bijker (1971), Van Rijn (1984b)
The total load transport of bed material particles can be obtained by summation of the bed
load and suspended load transport, as follows:
Qc = qb,c + %,c (7.4.1)
A numerical program-TRANSPOR of the method of Van Rijn is available, see Appendix A.
2. Engelund-Hansen (1967) ,'
The method of Engelund-Hansen is based on a energy-balance concept.

The work (per unit time and width) required to elevate the sediment load over a height equal
to the bed form height A is:
The work (per unit time and width) done by the fluid on moving the particles over a length
equal to the bed form length X is
The energy balance W, = W, yields:
in which:
= volumetric total load transport (m2/s)
q,,
t, = effective bed-shear stress (N/m2)
; b,a
= critical bed-shear stress according to Shields (N/m2)
= particle diameter (m)
U* = overall bed-shear velocity (m/s)
f = friction coefficient = 2g/C2
C = ChCzy-coefficient
X,A = bed-form length and height (m)
Based on data analysis, it was found that Xf/A is approximately constant, giving:
The dimensionless transport rate at = ((s - 1)O.' O5 d le5) is introduced, yielding:
Using Eq. (6.2.26) and assuming 0, = 0.06, it follows that 8'-0, = 0.4 0' for the lower
regime. Thus:
The a,-coefficient was determined by data fitting (approx. 100 flume data), giving a, =
0.1. Thus
in which:
Equation (7.4.8) does not account for the critical bed-shear stress and is, therefore, not
-accurate close to initiation of motion. Computed values are given in Appendix A.
Equation (7.4.8) can be rearranged to:
in which:
gee - volumetric current-related total load transport (m2/s)
C = ChCzy-coefficient (m1'2/s)
s = specific density (= pJ p )
dm = median particle size of bed material (m)
3. Yang (1973)
Yang (1973) assumed that the sediment transport is related to the unit stream power, defined
as: ii I . The total sediment concentration (4,defined as the ratio of the sediment and fluid
discharge per unit width, (c, = qJq), was expressed as:
Analysis of flume and field data resulted in:

in which:
- == total
c, load concentration in parts per million by weight (ppm)
-
u
=
depth-averaged
depth-averaged
velocity (mls)
velocity at initiation of motion (mts)
u
,
I = energy gradient
d, = median particle diameter of bed material (m)
w, = fall velocity (based on d, of bed material) (mls)
u, = bed-shear velocity (m/s)
2.5
for 1.2 < -< 70
log(u,ds/v) - 0.06 v
The total load transport rate (in kglsm) is given by:
in which:
'It,= -
- total load transport rate (kglsm)
h = water depth (m)
4. Ackers-White (1973)
Based on analysis of 925 sets of flume and field data, the following empirical formula was
proposed:
in which:
$C
= total load transport (m2/s)
u = depth-averaged velocity (m/s)
u+ = bed-shear velocity (m/s)
Y = particle mobility parameter (-)
ya = critical particle mobility parameter (-)
n,m,K = coefficients (-)
s = specific density (-)
d35 = representative diameter of bed material (m)
for 1 < D, < 60
n = 1-0.561og(D,) for 1 < D, < 60
for 1 < D, < 60
for 1 < D, < 60
for
The K and m coefficients were later (HR Wallingford, 1990) revised:
for 1 < D, < 60
K = 0.025 for D, 2 60
for 1 < D, < 60
The form of the Y-parameter ensures that the fine sediment transport depends upon the total
bed shear stress and the coarse transport rate upon the effective grain shear stress.
Revision of the original coefficients was necessary because the original formula predicted
transport rates which were considerably too large for relatively fine sediments (d,< 200
pm) and relatively coarse sediments.
7.4.2 Comparison of methods
Computed transport rates according to the methods of Engelund-Hansen (1967) and Van Rijn
(1984b) are shown in Appendix A.
Van Rijn (1984b) used 486 sets of river data to verify the methods of Engelund-Hansen
(1967, Ackers-White (1973), Yang (1973) and Van Rijn (1984). Bed material sizes were in
the range of 100 to 400 pm. Flow velocities were in the range of 0.4 to 2.4 mls. The results
have been expressed in terms of of a discrepancy ratio (r) defined as the ratio of the predicted
and measured transport rate. Table 7.9 shows the percentage of r-values of all data falling
in the range of 0.5 Ir 5 2. The method of Van Rijn yields the best results for field data
with 76% of the predicted -transport rates within a factor 2 of the measured values.
The method of Yang yields excellent results for flume data and small-scale river data, but
very poor results for large-s@e river data (depth > 1 m). This cannot be attributed to the
quality of the field data because the other other methods produce reasonable results for these
data sets. Therefore, the method of Yang must have serious systematic errors for large
depths. On the average, the predicted values were much too small.
Voogt et al (1989) carried out large-scale flume experiments with bed material of 200 pm and
velocities in the range of 1 to 3 m/s. Comparison of predicted and measured transport rates
showed good results for all three methods. Voogt et al also compared predicted rates with 120
sets of measured estuary data. Bed material sizes were in the range of 200 to 300 pm. Flow
velocities were in the range of 1 to 2 mls. The results are given in the Table 7.9. The method
of Van Rijn yields the best results with about 80% of the predicted transport rates within a
factor 2 of measured values. The results of the other two methods are less good, yielding a
considerable overprediction, particularly the method of Ackers-White.
Method 297 486 120

Flumedata River data Estuary data
Engelund-Hansen (1967) 68 % 64 % 33 %
Ackers-White (1973) 68 % 63 % 26 %
yang (1973) 89 % 39 % -
Van Rijn (1984) 76 % 76 % 89 %
Table 7.9 Percentage of predicted values within factor 2 of measured values
White et al(1973) examined various transport formulae using about 1000 flume data and 260
field data. Froude numbers greater than 0.8 were excluded. The results of the formulae of
Ackers-White (1973), Engelund-Hansen (1967),Einstein (1950) and Bagnold (1966) are given
in Table 7.10.
Method Flume and field data (1260)

Ackers-White (1973) 68 %
Engelund-Hansen (1967) 63 %
Einstein (1950) 46 %
Bagnold (1966) 22 %
Table 7.10 Percentage of predicted values within factor 2 of measured values
The method of Ackers-White yields the best results with 68 % of the predicted transport rages
within a factor 2 of measures values. The method of Bagnold yields poor results with a score
of 22%.
Yang and Molinas (1982) compared the Yang-formula with 1093 laboratory data and 166
river data; yielding a value of 95 % of the predicted transport rates within a factor 2 of
measured transport rates. Large-scale river data were not considered.
Usually, the available sediment transport formulae require as input data: mean velocity, water
depth, energy slope and particle characteristics. Van Rijn (1984~)tried to predict the transport
rate in combination with a roughness predictor using as input data: the mean velocity, the
water depth and the particle characteristics. The methods of Engelund-Hansen (1967) and
Ackers-White (1973) were used in a similar way. The predicted transport rates were
compared with measured values from USA rivers (266 data sets). The results are presented
in Table 7.11. The relatively poor results of the methods of Engelund-Hansen and Ackers-
White stem from the relatively strong influence of the predicted bed roughness on the
transport rate.
Method Field data (266)

Van Rijn
Engelund-Hansen
Ackers-White 44 %
Table 7.11 Percentage of predicted values within a factor 2 of measured values
7.96
Finally, it is noted that it is very difficult to predict the sediment transport rate with an
accuracy smaller than a factor 2 because of errors in the measured transport rates. Figure
7.4.1 shows mean concentrations 6 = q&) for two size ranges as measured by different
researchers. Deviations of more than a factor 2 can be observed for "similar" conditions.
x Laursen ( b =O.91m
Oxford (b=1.22m )
moan flow velocity, Ci (mls)
Os 51.3
x Guy o t al ( b = 2 . 4 4 m )
a Barton-Lin(b=1.22m)
0 0.2 0.4 0.6 0.8 1.O 1.2

,-• mean flow v ~ l o c i t yi, ( m l s )
Figure 7.4.1 Mean sediment concentration measured by different researchers under

"similar"conditions
1. Sand concentration measurements were carried out in the Mississippi river at station
1250, 20 April 1961 by Scott and Stephens (1966). The results are presented in Table
7.9. The measured suspended load transport was 2.4 kg/sm. The bed-load transport was
not measured.
The hydraulic parameters are: water depth h = 12.05 m, energy gradient I = 0.000076,
mean velocity i = 1.55 m/s. The bed material characteristics are: d,, = 350 pm,
d, = 400 pm, d, = 1500 pm, a, = 2.5. Other data are: p, = 2650 kg/m3,
p = 1000 kg/m3, v = lo6 m2!s.
a. Use the Rouse concentration profile and reference concentration of Van Rijn and
Smith-McLean to compute the sediment concentrations at z = 0.55, 1.59, 3.63,
6.62, 9.39 and 10.95 m above the bed?
b. Compute the suspended load transport according to Bagnold and Van Rijn?
c. Compute the total load transport according to Bagnold, Engelund-Hansen, Ackers-
White, Yang and Van Rijn?
Solution:
a. Concentration profile
Bed-shear stress : tb = pgh I = 9 N/m2
Bed-shear velocity : u+ = 0.095 m/s
Effective bed roughness : k, = 12h10-C'18=0.21m
Particle size parameter : D, = 10.1
Critical bed-shear stress : Tb,~ = 0.23 N/m2 (Fig. 4.1.5)
Van Rijn
Grain ChCzy-coefficient : C' = 181og(12h/3&) = 81 mS/s

I
Grain bed-shear stress : rb = pg(6/~')2 = 3.6 N/mZ
Reference concentration Van Rijn : c, = 0.015

ds -
- T" = 0.0008
(reference level a = k, = 0.21 m) ks
Suspended sediment size : d, = (1+0.011(aS-1)(T-25))dm = 330 ,urn
Fall velocity : w, = 0.046 m/s (Fig. 3.2.6)

Suspension number
= 0.4, co = 0.65)
(IC
2' = 0.82 + 0.1

I
= 0.92
Concentration profile : c =cay- hfa)l', see Table 7.12

Smith-McLean
Reference level
Reference concentration,
Eq. (7.3.30)
Suspension number : = w,/(Ku,) = 1.2

- h-z a z
Concentration profile : C , see Table 7.12
- (Tb-a)
Concentrations (kg/m3)
Height above
bed z Computed
Measured .
(m) Van Rijn McLean
0.55 0.41 0.84 0.65
1.59 0.24 0.29 0.16
3.63 0.14 0.11 0.05
6.62 0.10 0.04 0.01
9.39 0.08 0.02 0.004
10.95 0.05 0.006 0.004
Table 7.12 Measured and computed concentrationsfor Mississippi river
7.99
Comparison of measured and computed concentrations shows good agreement in the
lower layer (z C 4 m), but less good agreement in the upper layer where the computed
values are much too small. This is caused by the gradation of the bed material. The finer
fractions will be suspended in the upper layers and the coarser fractions in the near-bed
region. This cannot be represented by using one representative fall velocity.
b. Suspended load transport (see Table 7.13)
Bagnold : tb = = 9 ~ / r n ~
-
u = 1.55mIs
w, = 0.055 m/s (d,, = 400 pm, Fig. 3.2.6)
Van Rijn : a = k,=0.21m
c. Total load transport
Bagnold
qt = qb + q, = 1.2 + 0.4 = 1.6 kglsm

Van Rijn : qb = ~.l((s-l)~)O~"; DP.~TI.' = 9 lo-' m2/s = 0.24 kglsm
qt = qb + q, = 0.2 + 2.3 = 2.5 kglsm
Engelund-Hansen : C = 5 1.2 mXls
Yang
= 58 ppm
9t = 10" ct ii h = 1.1 kglsm

Ackers- White : D, = &((S-~)~/V~)"
= 8.9
K = 0.0202
= 2.44
= 0.468
= 0.217
= 0.892
qt = KG ) )Y-Y, = 6.5 104m2/s
Computed Measured suspended

Method load transport
Bed load Suspended load Total load Org/=)
transport m
r
t transport
(kg/=@ (kg/=l) (kg/=)
Bagnold 0.4 1.2 1.6
Engelund-Hansen 2.6
yaw - 1.1 2.4
Ackers-White - 1.7
Van Rijn 0.2 2.3 2.5
Table 7.13 Measured and computed transport rates for Mississippi river
2. A wide river has a water depth of h = 3 m, energy gradient I = 2.104, depth-averaged
flow velocity ii = 1.5 mls. The bed material characteristics are d3, = 250 pm, & =
300 pm, dgO= 600 pm, a, = 1.8. The water temperature is Te = 20°C, v = 1. lo4
m2/s. Other data are p, = 2650 kg/m3, p = 1000 kg/m3.
a. Use the Rouse concentration profile and reference concentration of Van Rijn and
McLean to compute sediment concentrations at z = 0.03, 0.1, 0.5, 1, 2 and 3 m
above bed?
b. Compute suspended load transport according to Bagnold and Van Rijn?
c. Compute total load transport according to Bagnold, Engelund-Hansen, Ackers-
White, Yang and Van Rijn?
Solution:
Computed concentrations are given in Table 7.14.
Height Concentration
above bed (kg/m3)
z
(m) Van Rijn McLean
0.03 15.4 12.4

0.1 3.8 1.8
0.5 0.53 0.12
1 0.19 0.03
2 0.04 0.003
3 0 0
Table 7.14 Computed concentrations

Suspended load transport : Bagnold q, = 1.04 kg/sm
Van Rijn q, = 1.72 kg/sm
Total load transport : Bagnold qt = 1.3 kg/sm

Engelund-Hansen qt = 1.7 kg/sm
Ackers-White qt = 2.1.kgIsm
yang q, = 0.9 kglsm
Van Rijn q, = 2.0 kglsm
3. A wide river has a water depth of h = 3 m, mean flow velocity of u = 1.2 m/s,
energy gradient I = 2. lo4, bed material characteristics d,, = 120 pm, dso = 150 pm,
d, = 300 pm, a, = 1.5. Other data are Te = 20°C, v = 1 lo4 m2/s, p, = 2650
kg/m3, p = 1000 kg/m3.
a. Compute representative size of suspended sediment?
b. Compute reference concentration according to Van Rijn?
c. Compute sediment concentration (Rouse profile) at z = 0.1, 0.3, 0.5, 1 and 2 m
above the bed?
d. Compute suspended load transport according to Bagnold and Van Rijn?
e. Compute ratio of suspended load transport according to Bagnold, Van Rijn and
Laursen?
f. Compute total load trmport according to Engelund-Hansen, Ackers-White,
Bagnold and Van Rijn?
Solution:
d. Bagnold q, = 1.7 kg/m3
Van Rijn q, = 1.6kg/m3
e. Bagnold g/qt = 0.90

Van Rijn q,/q, = 0.97
Laursen q,/qt = 0.92
f. Engelund-Hansen qt = 2.2 kg/sm
Ackers-White qt = 0.8 kg/sm
Bagnold qt = 1.9 kglsm
Van Rijn qt = 1.7 kglsm
4. A wide river has a depth of h = 3 m; the energy gradient is I = 5 104. The bed
material characteristics are d, = 150 pm, d, = 300 pm. The fall velocity is w, =
0.0 17 m/s. The bed is plane. The depth-averaged concentration is c = 5 kg/m3. Other
data are: v = 1 10" m2/s, p, = 2650 kg/m3, p = 1000 kg/m3.
What is the depth-averaged velocity with and without stratification according to Itakura
and Kishi?
What are the velocities at z = 0.05, 0.1, 0.5, 1, 2 and 3 m above the bed?
Solution:
Assume near-bed viscosity : v, = 10v = 10" m2/s
Bed-shear velocity : u, = (ghI)O.' = 0.121 m/s
Mobility parameter : 8 = u : / ( ( ~ - l ) ) ~d, = 6.1
Effective bed roughness, : k:= = 3 8 dm,= 0.0055 m

Eq. (6.2.12b) Take = 0.01 m
Reynolds number : u, k/v, = 121 - rough regime
Monin-Obukhov length scale : cvob

- = 5/p, = 512650 = 0.0019
Depth-averaged velocity : u
- = u,[6 + 7h/(2xL) + 2.5 h(h/ks)]
Depth-averaged velocity
without stratification : u
- = u*[6 + 2.5 h(h/4)]
Velocity profile
Uo.05 = 1.53 m/s

Uo. 1 = 1.75 m/s
U0.s = 2.34 m/s
U1 = 2.67 m/s
U2 = 3.13 m/s
U3 = 3.50 m/s
5. A wide river has a discharge of q = 10 m2/s. The energy gradient is I = 5.104. The
total sediment discharge is q , = 60 kg/sm. The bed material characteristics are d, =
150 pm, d, = 300 pm. The fall velocity is w, = 0.017 m/s. The bed roughness is k,
= 0.01 m. Other data: v = 1 10" m2/s, p, = 2650 kg/m3, p = 1000 kg/m3.
What is the mean (volume) sediment concentration?

What is the flow depth according to the method of Itakura and Kishi (assume rough
flow)?
Solution: c, = 0.0023
h = 3.3 m (by trial and error)
8.5 References
Abbot, J.E. and Francis, J.R.D., 1977. Saltation and Suspension Trajectories of Solid
Grains in a Water Stream. Proc. Royal Soc., Vol. 284, A 1321.
Ackers, P. and White, W.R., 1973. Sediment Transport: New Approach and Analysis.
Journal of the Hydraulics Division, ASCE, No. HY 1I , USA.
Anderson, R.S. and Haff, P.K., 1988. Simulation of Eolian Saltation. Science, 241, pp.
820-823.
I.
Ashida, K. and Fujita, M., 1986. Stochastic Model for Particle Suspension in Open Chan-
nels. Journal of Hydroscience and Hydraulic Engineering, ISCE, Vol. 4, p. 21-46.
Atkins, R., Soulsby, R.L., Waters, C.B. and Oliver, N., 1989. Field Measurements of
Sediment Suspension above Bed Forms in a S a m Estuary. Report SR 203, Hydrau-lics
Research Wallingford, England.
Bagnold, R.A., 1956. The Flow of Cohesionless Grains in Fluids. Proc. Royal Soc. Philos.
Trans., London, Vol. 249.
Bagnold, R.A., 1962. Auto-Suspension of Transported Sediment. Proc. Royal Society,
Vol. A, No. 1322, England.
Bagnold, R.A., 1966. An Approach to the Sediment Transport Problem from General Phy-
sics. Geological Survey Prof. Paper 422-1, Washington.
Barton, J.R. and LIN, P.N., 1955. A Study of the Sediment Transport in Alluvial Streams.
Rep. No. 55 JR 82, Civ. Eng. Dep., Colorado College, Fort Collins, USA.
Bijker, E. W., 1971. Longshore Transport Computations. Journal of the Waterways, Har-
bours and Coastal Engineering Division, Vol. 97, No. WW4.
Bijker, E. W., 1978. Lecture Notes Coastal Engineering, Dept. Coastal Engineering. Delft
Univ. of Technology, Delft, The Netherlands.
Bogardi, J.L., 1974. Sediment Transport in Alluvial Streams. Akademiaci Kiado, Budapest.
Businger, J.A., Wyngaard, J.C., Izumi, Y. and Bradley, F., 1971. Flux-Profile Relation-
ships in the Atmospheric Sulface Layer. Journal Atmos. Science, Vol. 28, pp.
181-189.
Carstens, M.R., 1952. Accelerated Motion of a Spherical Particle. Trans. Amer. Geoph.,
Union, Vol. 33, No. 5, pp. 713-720.
Chien, N., 1954. The Present Status of Research on Sediment Transport. Proc. ASCE,
voi. 80.
Chien, N. and Zhaohui, W., 1986. A Critical Review of the Research on the Hyperconcen-
trated Flow in China. Publication IRTCES, Intern. Res. and Train. Centre on Eros. and
Sedim., Beijing, China.
Coleman, N.L., 1970. Flume Studies of the Sediment Transfer Comcient Water Resources
Research, Vol. 6, No. 3, USA.
Coleman, N.L., 1980. Velocity Profiles with Suspended Sediment. Journal of Hydraulic
Research, Vol. 19, No. 3, The Netherlands.
DeVantier, B. A. and Larock, B.E., 1983.Sediment Transport in Stratified Turbulent Flow.
Journal of Hydraulic Engineering, Vol. 109, No. 12.
Dyer, K.R. and Soulsby, R.L., 1988. Sand Transport on the Continental Shelf. Annual
Review Fluid Mechanics, Vol. 20, pp. 295-324.
Egiazaroff, P.I., 1965. Calculation of Non-Unifonn Sediment Concentration. Journal of
Hydraulics Division, ASCE, No. 44.
Einstein, H.A., 1950.The Bed-Load Function for Sediment Transportation in Open Channel
Flow. Technical Bulletin No. 1026, U.S. Dep. of Agriculture, Washington, D. C.
Einstein, H.A. and Abdel-aal, F.M., 1972.Einstein Bed Load Function at High Sediment
Rate. Journal of Hydraulics Division, ASCE, Vol. 98, No. HYl, USA.
Einstein, H. A. and Barbarossa, N. L., 1952.River Channel Roughness. Trans. ASCE,
Vol. 117, pp. 1121-1132.
Einstein, H.A. and Chien, N., 1955.Ecffects of Heavy Sediment Concentrations near the Bed
on Velocity and Sediment Distribution. Inst. of Eng. Research, Univ. of Berkely,
California, USA.
Engelund, F. and Hansen, E., 1967. A Monograph on Sediment Transport in Alluvial
Stream. Teknisk Forlag, Copenhagen, Denmark.
Engelund, F., 1974.Flow and Bed Topography in Channel Bends. Journal of Hydraulics
Division, ASCE, Vol. 100, No. 11.
Engelund, F. and Fredsee, J., 1976.A Sediment Transport Model for Straight Alluvial
Channels. Nordic Hydrology, 7, pp. 293-306.
Fernandez-Luque, R., 1974. Erosion and Transport of Bed-Load Sediment. Dissertation,
Krips Repro B. V., Meppel, The Netherlands.
Fernandez-Luque, R. and Van Beek, R., 1976.Erosion and Transport of Bed-Load Sedi-
ment. Journal of Hydraulic Research, Vol. 14, No. 2, The Netherlands.
Francis, J.R.D., 1973.Experiments on the Motion of Solitary Grains along the Bed of a
Water-Stream. Proc. Royal Soc. A., 332.
Frij.link, H.C., 1952.Discussion of Bed Load Movement Formulas. Report No. X2344/LV,
Delj? Hydraulics, Delft, The Netherlands.
Garcia, M. and Parker, G., 1991.Estrainment of Bed Sediment into Suspension. Journal
of Hydraulic Engineering, ASCE, Vol. 11 7, No. 4.
Gessler, J., 1970.Self-stabilizing Tendencies ofAlluvial Channels. Journal of Waterways and
Harbours Division, ASCE, Vol. 96, No. WW2
Gessler, J., 1973. Behaviour of Sediment Mixtures in Rivers. L M R Int. Symp. in River
Mechanics, Bangkok, pp. 395-485.
Guy, H.P., Simons, D.B. and Richardson, E.V., 1966. Summary of Alluvial Channel Data
from Flume Experiments, 1956-1961. Geol. Survey Prof. Paper, 462-1, Washington.
Hardisty, J. and Whitehouse, R.J.S., 1988. Evidence for a New Sand Transport Process
from Experiments on Saharan Dunes. Nature, Vol. 332, No. 6164, pp. 532-534.
Hr Wallingford, 1990. Sediment Transport, The Ackers and White Theory Revised. Report
SR 237, Wallingford, England.
Hubbell, D.W. and Matejka, D.Q., 1959. Investigations of Sediment Transportatz'on, Middle
Loup River. Water-Supply Paper No. 1476, U.S. Geol. Survey; Washington, USA.
Hunt, J.N., 1954. The Turbulent Transport of Suspended Sediment in Open Channels. Proc.
Royal Society, London, Vol. 224A, pp. 322-335.
Ikeda, S., 1980. Suspended Sediment on Sand Ripples. Third Int. Symp. on Stoch. Hydr.,
Tokyo, Japan.
Ikeda, S. and Asaeda, T., 1983. Sediment Suspension with Rippled Bed. Journal of
Ikeda, S., 1982. Lateral Bed-Load Transport on Side Slopes. Journal Hydraulics Division,
ASCE, Vol. 108, No. 11.
Ikeda, S., 1988. Lateral Bed Load Transport on Side Slopes. In: Civil Engineering Practice
2, Technomic Publishing Company, USA.
Ippen, A.F., 1971. A New Look at Sedimentation in Turbulent Streams. Journal of the
Boston Soc. of Civ. Engrs., Vol. 58, No. 3, USA.
Itakura, T. and Kishi, T., 1980. Open Channel Flow with Suspended Sediments. Journal
of Hydraulics Division, ASCE, Vol. 106, No. HY8.
Kachel, N.B. and Sternberg, R. W., 1974. Transport of Bed Load as Ripples during an Ebb
Current. Marine Geology, 19, p. 229-244.
Kalinske, A.A., 1947. Movement of Sediment as Bed Load in Rivers. Trans. American Geo-
physical, Vol. 28, No. 4.
Klaassen, G.J., 1991. Experiments on the Eflect of Gradation and Vertical Sorting on
Sediment Transport Phenomena in the Dune Phase. Grain Sorting Seminar, Ascona,
Switzerland.
Lane, E.W., Carlsen, E.J. and Hansen, D.S., 1949. Low Temperature Increases Sediment
Transportation in Colorado River. Civil Engineering, ASCE, Vol. 19, No. 9, USA.
Lau, Y.L.A, 1982. Suspended Sediment Effect on Flow Resistance. Journal of Hydraulic
Laursen, E.M., 1958. The Total Sediment Load of Streams. Journal of the Hydraulics
Division, ASCE, No. HY 1.
Livesey, R.H., 1965. Channel Annouring below Fort Randall Dam. Publ. 1970, Agric.
Research Service, USA Dep. of Agriculture, Washington, USA.
Mastbergen, D., 1987. Sand- Water Mixture Flows. Report 246-02, Delft Hydraulics, Delft,
The Netherlands.
McLean, S.R., 1991a. Stratrpcation Eflects due to Suspended Sediments. Int. Symp.
Transport of Suspended Sediments, Florence, Italy.
McLean, S.R., 1991b. Depth-Integrated Suspended Load Calculations. Journal of Hydraulic
Engineering, ASCE, Vol. 117, No. I l .
Meyer-Peter, E. and Mueller, R., 1948. F o m l a s for Bed-Load Transport. Sec. Int. IAHR
congress, Stock-holm, Sweden.
Morse, B. and Townsend, R.D., 1989. Modelling Mixed Sediment Suspended Load Profiles.
Journal of Hydraulic Engineering, ASCE, Vol. 115, No. 6.
Morsi, S.A. and Alexander, A.J., 1972. An Investigation of Particle Trajectories in Two-
Phase Flow Systems. Journal of Fluid Mechanics, Vol. 55, Part 2.
Nagakawa, H. and Tsujimoto, T., 1980. Sand Bed Instability due to Bed-Load Motion.
Journal of Hydraulics Division, Vol. 106, No. HY 12, USA.
Nordin, C.F., 1964. Aspects of Flow Resistance and Sediment Transport: Rio Grande near
Bernalillo, New Mexico. Water-Supply Paper 1498-H, U.S. Geol. Survey, Washington,
USA.
Paintal, A.S., 1971. Concept of Critical Shear Stress in Loose Boundary Open Channels.
Journal of Hydr. Research, Vol. 9, No. 1.
Parker, G. and Klingeman, P.C., 1982. On Why Gravel Bed Streams are Paved. Water
Resources Research, Vol. 18, No. 5, pp. 1409-1428.
Parker, G., Klingeman, P.C. and McLean, D.G., 1982. Bed Load and Size Distribution
in Paved Gravel Bed Streams. Journal of the Hydr. Div., ASCE, Vol. I08, No. HY 4.
Poreh, M., Sagiv, A. and Seginer, J., 1970. Sediment Sampling Eflciency of Slots. Journal
of the Hydr. Div. ASCE, HY 10.
Ribberink, J.S., 1987. Mathematical Modelling of One-Dimensional Morphological Changes

in Rivers with Non-Uniform Sediment. Thesis, Univ. of Technology Delft, Delft, The
Netherlands.
Rickermann, D., 1991. Hyperconcentrated Flow and Sediment Transport at Steep Slopes.
Rodi, W., 1980. Turbulence Modeki and their Application in Hydraulics. Institut fir Hydro-
mechanik, Univ. of Karlsruhe, G e m n y .
Rubinow, S.I. and Keller, J.B., 1961. The Transverse Force on a Spinning Sphere Moving
in a Viscous Fluid. Journal of Fluid Mechanics, Vol. 11.
Ruiter, J.C .C. de, 1982. The Mechanism of Sediment Transport on Bed Fonns. Proceedings
Euromech Conference 156, Techn. Univ. of Instanbul, Turkey.
Ruiter, J.C.C. de, 1983. Incipient Motion and Pick-up of Sediment. Report R 65743, De@
Saffman, P.G., 1968. The Lift on a Small Sphere in a Slow Shear Flow. Journal of Fluid
Mechanics, Vol. 22, 1965, Vol. 31.
Samaga, B.R., Ranga Raju, K.G. and Garde, R.J., 1986, Bed Load Transport of Sediment
Mixtures. Journal of Hydraulic Engineering, ASCE, Vol. 112, No. 11.
Scott, C.H. and Stephens, H.D., 1966. Special Sediment Investigation, Mississippi River
at St. Louis, Missouri, 1961-1963. Geol. Survey Water-Supply Paper 1819-J,
Washington, USA.
Sekine, M. and Kikkawa, H., 1984. On the Characteristics of Bed-Load Transportation.
Proc. Japan SOC.Civ. Eng., 351, pp. 247-250.
Sekine, M. and Kikkawa, H., 1992. Mechanics of Saltating Grains, II. Journal of Hydraulic
Engineering, ASCE, Vol. 118, No. 4, pp. 536-558.
Singamsetti, S.R., 1966. Dimion of Sediment in a Submerged Jet. Journal of the Hydrau-
lics Div., ASCE, Vol. 92, No. HY 2.
Smart, G.M., 1984. Sediment Transport Formula for Steep Channels. Journal of Hydraulic
Smith, J.D. and McLean, S.R., 1977. Spatially Averaged Flow over a Wavy Surface. Jour-
nal of Geophysical Research, Vol. 82, No. 12, pp. 1735-1746.
Soulsby, R.L. and Wainwright, B.L., 1987. A Criteriumfor the Efect of Suspended Sedi-
ment on Near-Bottom Velocity Profiles. Journal of Hydraulic Research, Vol. 25,
No. 3.
Taylor, G.I., 1917. Motion of Solids in Fluids when the Flow is not Irrotational. Proc. Royal
Society London, A93, pp. 99-113.
Thorne, C.R., Bathurst, J.C. and Hey, R.D., 1987. Sediment Transport in Gravel-Bed
Rivers. Wiley, New York.
Umeyama, M. and Gerritsen, F., 1992. Velocity Distribution in Uniform Sediment-Laden
Flow. Journal of Hydraulic Engineering, ASCE, Vol. 118, No. 2.
Valiani, A., 1988. An Open Question regarding Shear Flow with Suspended Sediments.
Meccanica, 23, pp. 36-43, Italy.
Vanoni, V.A., 1946. Transportation of Suspended Sediment by Water. Trans. ASCE,
Vol. I l l .
Vanoni, V.A. and ~rooks,N.H., 1957. Loboratory Studies of the Roughness and Suspended
Load of Alluvial Stream. Sedimentm'on Laboratory, California Institute of Technology,
Pasadena, Report E-68, USA.
Van Rijn, L.C., 1983. Sediment Pick-up Functions. Report M1531, Part N,Delfr Hydrau-
lics, Deljt, The Netherlands.
.
Van Rijn, L.C ,1984a. Sediment Transport, Part 1:Bed Load Transport. Journal of Hydrau-
lic Engineering, ASCE, Vol. 110, No. 10.
Van Rijn, L.C., 1984b. Sediment Transport, Part XI: Suspended Load Transpon. Journal
of Hydraulic Engineering, ASCE, Vol. 110, No. 11.
Van Rijn, L.C., 1984~.Sediment Transport, Part-111: Bed Forms and Alluvial Roughness.
Van Rijn, L.C., 1984d. Sediment Pick-Up Functions. Journal of Hydraulic Engineering,
ASCE, Vol. 110, No. 10.
Van Rijn, L.C., 1985. Mathematical Models for Sediment Concentration Profiles in Steady
Flow. Euromech 192, Transport of Suspended Solids in Open Channels, Munich, F.R.
Germany.
Van Rijn, L.C., 1986. Applications of Sediment Pick-Up Functions. Journal of Hydraulic
Engineering, ASCE, Vol. ,112, No. 9.
Van Rijn, L.C., 1987. Mathematical Modelling of Morphological Processes in the Case of
Suspended Sediment Transport. Thesis, Dep. of Fluid Mech., Delfr Univ. of Techno-
logy, Delfr, The Netherlands.
Van Rijn, L.C., 1990. Principles of Fluid Flow and Suvace Wavesfor Rivers, Estuaries,
Seas and Oceans. Aqua Publications, Amsterdam, The Netherlands.
Voogt, L., Van Rijn, L.C. and Van den Berg, J.R., 1991. Transport of Fine Sands at High
Velocities. Journal of Hydraulic Engineering, ASCE, Vol. 117, No. 17.
Wang, S., 1981. Variation of K a m n Constant in Sediment-Laden Flow. Journal of Hydrau-
lics Div., ASCE, Vol. 107, No. HY 4.
White, B.R. and Schulz, J.C., 1977. Magnus Efect in SaItation. Joumal of Fluid Mecha-
nics, Vol. 81.
White, W.R., Milli, H. and Crabbe, A.D., 1973. Sediment Transport, An Appraisal of
Available Methods. Int. Report 119, Wallingford, England.
Wiberg, P.L. and Smith, J.D., 1987. A Theoretical Model for SaItathg Grains in Water.
Joumal Geophysical Research, Vol. 90, No. 4, pp. 7341- 7354.
Wiberg, P.L. and Smith, J.D., 1987. Model for Calculating Bed-Load Transport of Sedi-
ment. Journal of Hydraulic Engineering, ASCE, Vol. 115, No. 1, pp. 101-123.
Williams, P.G., 1970. Flume Width and Water Depth Eflects in Sediment Transport Fxperi-
ments. -Geol. Survey Prof. Paper 562-H, Washington, D. C.
Winterwerp, J.C., De Groot, M.B., Mastbergen, D.R. and Verwoert, H., 1990. Hyper-
concentrated Sand-Water Mixture Flows over a Flat Bed. Journal of Hydraulic Engi-
neering, ASCE, Vol. 116, No. 1.
Woo, H.S, Julien, P.Y. and Richardson, E.V., 1988. Suspension of Large Concentrations
of Sands. Journal of Hydraulic Engineering, ASCE, Vol. 114, No. 8.
Xingkui, W. and Ning, Q., 1989. Turbulence Characteristics of Sediment-Laden Flow.
Yalin, M.S. and Finlayson, G.D., 1972. On the Velocity Distribution of the Flow-Carrying
Sediment in Suspension. Symp. to honour Prof. H.A. Einstein, USA.
Yalin, M.S., 1977. Mechanics of Sediment Transport (2nd Edition). Pergamon Press,
Oxford, England.
Yang, C.T., 1973. Incipient Motion and Sediment Transport. Journal of the Hydr. Div.,
ASCE, Vol. 99, No. HY 10.
Yang, C.T. and Molinas, A., 1982. Sediment Transport and Unit Stream Power Function.
Journal of the Hydraulics Division, Vol. 108, No. HY 6.
Zenghai, R., 1988. Experimental Study on Resistance of the Flow with Hyperconcentrat'on
in Open Channels and Pipes. Hyperconcentrations of Sediment, Beijing, China.
Zhaoyin, W., 1990. Limit Concentration of Suspended Sediment. Proc. National Conference
Hydr. Div. ASCE, San Diego, USA.
Zyserman, J.A. and F'redsee, J., 1993. Data Analysis of Bed Concentration of Suspended
Sediment, Submitted to Journal of Hydraulic Engineering, ASCE.

River Engineering 1-7

Uploaded by

Copyright:

Available Formats

River Engineering 1-7

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

River Engineering 1-7

Uploaded by

Copyright:

Available Formats

L.C.

van Rijn Lecture Notes

Principles of Sediment Transport in

1 INTRODUCTION ...................................... 1.1

4 . INITIATION OF MOTION ............................... 4.1

4.2 Initiation of motion in waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.23

BEDFORMS ......................................... 5.1

6 EFFECTIVE BED ROUGHNESS ........................... 6.1

7 BED MATERIAL SUSPENSION AND TRANSPORT IN STEADY

8 BED MATERIAL SUSPENSION AND TRANSPORT IN WAVES ....... 8.1

8.5 Computation ofpdiment transport rates . . . . . . . . . . . . . . . . . . . . . 8.58

11 TRANSPORT OF COHESIVE MATERIAIS ................... 11.1

11.5 Deposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11.15

13 MEASURING INSTRUMENTS FOR SEDIMENT TRANSPORT.

APPENDICES A: TRANSPOR-program; Computation of sediment transport in current

Chapter 10 is related to the transport of sediments in non-steady and non-uniform conditions.

Sediment transport cannot be studied without proper knowledge of measuring instruments.

Leo C. van Rijn

Usually, three modes of particle motion are distinguished:

wash load moving as suspended load

The material, the particle sizes of which are found in appreciable

The most important symbols used in this book are:

= reference level (m)

Instantaneous variables (velocity) are represented by Capital letters (U). Time-averaged

= current-related bed-shear stress

1.4 Characteristic parameters

The following dimensionless characterisic parameters are herein used:

The excess bed-shear stress parameter T, is defined as:

The dimensionless transport @ usually is represented as:

Another dimensionless expression is:

2.2.1 Current boundary layer

For z = 0 this yields the bed-shear stress:

By defdtion r, = p(u,,J2, yielding

Hydraulically smooth flow, for 4 -- -

Figure 2.2.1 Shear stress

turbulent logarithmic layer

Figure 2.2.2 Current boundary layer

Figure 2.2.3 Velocity distribution in smooth and rough flow

Figure 2.2.4 Miring length and eddy viscosity distribution

Rough flow regime : z, = 0.033 k, for u*sc4

Transition regime : z, = 0.11

2.2.4 Fluid mixing coefficient

The overall time-averaged bed-shear stress is defined as:

or using K = 0.4 and g = 9.81 m/s2:

Using Eqs. (2.2.9), (2.2.10) and (2.2.11), yields:

The friction factor follows from: f, -- -

3.3.1 Near-bed orbital velocities

with a = 1 + 0. 3(H8/h)and 0, according to linear wave theory.

Figure 2.3.1 Wave boundary layer

Reynolds' number, 06As IV

Figure 2.3.2 Hydraulic regimes in oscillatory flow

In case of turbulent flow Jonsson and Carlsen (1976) proposed):

Equation (2.3.7) is based on theoretical and experimental research. Artificial triangular

2.3.3 Hydraulic regime in waves

Hydraulic regime = F (0;"

in which: d, = median particle size of bed material.

0.0 0.5 1.O 0.0 0.5 1.0

2.3.5 Bed shear stress and bed friction

3. Smooth turbulent flow

Equation (2.3.16) can be approximated by