INTERNATIONAL COURSE IN HYDRAULIC ENGINEERING

LECTURE NOTES ON

SEDIMENT TRANSPORT 1

BY
H.N.C. BREUSERS

DELFT
1983 -1984
INTERNATIONAL COURSE IN HYDRAULIC ENGINEERING

LECTURE NOTES ON

SEDIMENT TRANSPORT 1

BY
H.N.C. BREUSERS

DELFT
1983 -1984
 0.1

INTERNATIONAL COURSE IN HYDRAULIC ENGINEERING

SEDIMENT TRANSPORT I

CONTENTS

1. . INTRODUCTION
2. PROPERTIES OF THE WATER
3. PROPERTIES OF THE TRANSPORT MATERlAL
4. INITIATION OF PARTICLE MOTrON
5. TRANSPORT MECHANISM, BEDFORMS, ALLUVIAL ROUGHNESS
6. BED MATERlAL TRANSPORT
6 .1. BED LOAD
6.2. SUSPENDED LOAD
6.3. TOTAL LOAD
7. STABLE CHANNELS
8. RIVER BED VARlATIONS
9 . LOCAL SCOUR
10. t1EASUREMENTTECHNIQUES
11. SEDIMENT TRANSPORT IN PIPES
 1.1

1. INTRODUCTION

A study of the sediment transport by water ~s of importance in severa1

aspects of hydrau1ic engineering:
- f1uvia1 hydrau1ics: know1edge of sediment transport forms the basis for
the design of river-training works, navigation improvement, f100d control.
- irrigation: design of stab1e channe1s, intakes, settling bassins.
- coastal engineering: prediction of littora1 drift, design of coastal pro-
tection works and harbours.
dredging: the suction, transport and deposition of material has many
aspects related to the transport of sediments.

The ma~n objective of sediment transport hydraulics is to predict

whether an equilibrium condition, erosion (scour) or deposition (silting)
will occur and to determine the quantities involved. The rate of sediment
transport, expressed as mass, weight or volume per unit time can be deter-
mined from measurements or from calculations. Both methods only have a low
degree of accuracy so that the sensitivity of the design to possible varia-
tions in the calculated transport rates has to be considered.
The main reason for the empirical character of sediment transport know-
ledge is the complexity of the transport process. The interaction of a
turbulent flow, the characteristics of which are only known by empirism,
and a boundary consisting of loose sediments cannot be described by simple
equations. Most of our knowledge is based therefore on experiments and
measurements both in the field and in laboratories.
The following subjects will be discussed:
- the flow characteristics of the water
- the characteristics of the sediments
- their mutual interaction:
- initiation of motion,
- transport mechanisms,
- bed forms, roughness,
- stable channels,
- bed material transport - bed load,
- suspended load,
- siltation and scour,
- sediment transport measurements
- applications
 1.2

These lecture notes should be considered as an introduction to the

subject. The following genera 1 references may be used for further studies:

S. LELIAVSKY 1955 An introduction to fluvial hydraulics.

Constabie, London.
T. BLENCH 1957 Regime behaviour of canals and rivers.
Butterworths, London.
F.M. HENDERSON 1966 Open channel flow (Ch. 4 - Sediment transport)
MacMillan, New York.
F. ENGELUND 1967 A monograph on sediment transport 1n alluvial
streams.
Teknisk Forlag, Copenhagen.
W.H. GRAF 1971 Hydraulics of sediment transport.
Mc Graw HilI, New York (very complete treatment).
H.W. SHEN (ed.) 1971 River Mechanics I.
Ft. Collins, Colorado.
J .L. BOGARDI 1974 Sediment transport in alluvial streams.
Akadémiai Kiadó, Budapest.
V.A. VANONI 1975 River Dynamics, Advances in app1ied mechanics
Vol. 15, pp. 1 - 87.
(C.S. Yih, ed.) Academic Press, New York.
V.A. VANONI (ed.) 1975 Sedimentation Engineering.
A.S.C.E. New York.
A.J. RAUDKIVI 1976 Loose boundary hydrau1ics (2nd ed.).
Pergamon Press Oxford.
D.B. SIMONS, 1977 Sediment Transport Techno1ogy.
F. SENTtlRK
Water Res. Pubi., Ft. Collins, Colorado
R.J. GARDE, 1977 Mechanics of Sediment Transportation and
K.G. RANGA RAJU Alluvial Stream Problems.
Wiley Eastern Limited, New Delhi.
P.Ph. JANSEN (ed.) 1979 Princip1es of river engineering.
Pitman, London.
H.W. SHEN, 1980 Application of stochastic processes in sediment
H. KIKKAWA (eds.)
transport (Proc. U.S.-Japan Seminar Hawai 1978).
Water Resources Pubi., Littleton, Colorado.
 2.1

2. PROPERTIES OF THE WATER

Some of the relevant properties of water are:

Property symbol dimension remarks

density P kg.m-:3 - Ps - Pw
relative density fl - (ratio) fl =
under water Pw

dynamic viscosity n kg.m-1.s-1 or N.s.m-2 T =

all
näZ
kinematic viscosit) v m2.s-1 \) = n/pw

surface tension o kg.s-2 or N.m-1

The fo11owing S.l. units are used:

mass (kg) (kilogram)
1ength (m) (meter)
time (s) (second)
force (kgm/s2) or (N) (Newton)
energy (kgm2/S2) or (Nm) or (J) (Joule)
power (kgm2/s:3)or (Nm/s) or (J/s) or (W) (Watt)
pressure, stress (kg/ms2) or (N/m2) or Pa (Pascal)

2.1. Density (kg/m3)

The density of fresh water varies with temperature T:

T: 0 4 12 16 21 32 (oC)

Pw : 999.87 1000.0 999.5 999.0 998.0 995.0 (kg/m3)

 2.2

The variation of the density may be neg1ected in most sediment transport

ca1cu1ations.

kg/m3

Pw fresh water 1000

Pw sea water 1026

2.2. Viscosity

~l~~~!~_~!~~~~!~l
(Ns/m2)
Defined as the factor of proportiona1ity in:

n
au
T =
az
which is va1id for 1aminar flow.
~ /~
oU oZ = ve l'
OC1ty gra d'
1ent (s-l)

~!~~~~~!~_Y!~~~~!!l
(m2/s)
Defined by v = nipw
n and Vare a function of temperature. The inf1uence of temperature is
significant.

T 0 5 10 15 20 25 30 35 40 (oC)

-6
V 1.79 1.52 1.31 1.14 1.01 0.90 0.80 0.72 0.65 • (10 m2/s)

2~3. Surfacè tèIi8Ïon

For the surface water/air: cr = 0.074 N/m at atmospheric pressure.

The variation with temperature can be neg1ected.
 2.3

2.4. Uniform flow in open channe1s

The equation of motion for steady, uniform flow is reduced to:

ar _ dp
äZ - dX T = shear stress p pressure
or T(Z) = P g(h - z).I
w
h = water depth
z = distance from the bed
I hydrau1ic gradient or slope

The difficu1ty is now the relation between shear stress and velocity
distribution which is necessary to predict this distribution.
For 1aminar flow the relation is:

which leads to the parabolic velocity distribution:

U(Z) 2v (h2 - (h - z)2)

= ~

and a mean velocity Ü = ~~ • h2

For turbulent flow Prandt1 gave the following empirica1 mixing~length

expression:

T(Z) = P
W
12 (dU(Z)/dZ)2

Near the bed T(Z) '"T , the bed shear stress:

o
T
o
Pw g h I
and 1= KZ

K = kappa, von Kármán's constant ~ 0.4 (from measurements)

This leads to the logarithmic velocity distribution:
U(z) = K-1 • ln(z/z )
~~g~h---I
o

Define uX .;gh I = shear velocity = I T

o
/p w
and take: K = 0.4
 2.4

then: U{z) = 2.5 In (z/z )

UX
o
z = the point where U = 0 according to the logarithmic profile.
o
U{z) is equal to the mean velocity at z ~ 0.4 h
or U
-= 2.5 UX In (0.4 h/z )
o
or
-
U = 5.75 uX log (0.4 h/z ) (In ~ log gives factor 2.303)
o

Although the logarithmic velocity distribution was derived for the area
near the bed, it appears from measurements that the logarithmic velocity
profile is a good approximation for the full depth of the flow due to a
simultaneous decrease in shear stress and mixing-length with z.
Values of z are found from experiments on smooth and rough boundaries.
o
For smooth boundaries a viscous sublayer exists in which viscous effects
predominate. The approximate thickness of this layer is 6 ~ 10 v/ux (see
below) and z ~ 0.01 6 ~ 0.1 v/ux• For boundaries with uniform roughness
o
Nikuradse has found:
Z 0.03
::: k
o s
in which k was the size of the sand grains used as roughness. This k is
s s
used as a standard roughness for other types of roughness.

Smooth boundary Rough boundary

zo '"0.016 Z 0.03
::: k
o s
U{z) = 5.75 uX log (100z/6) U{z) = 5.75 UX log (33 z/k )
s
- x - x
U = 5.75 u log (40 h/6) U = 5.75 u 10g(12 hIk )
s

Ü = 5.75 UX log (k 12h )

+ 0.36
s
or
- 12h
U = (5.75.;g) .1iI log (k + 0.36)
s
or
lü = lS.1iI log (ksl;ho.3~)I(White- Colebrook)

which is the well-known Chézy equation:

Iü=c~ I
A bed is defined as hydraulically smooth for k < 0.1 <5
s
hydraulically rough for k > 6<5
s
The transition laminar - turbulent flow is generally given as:
Re = Ü.h/V :::600 for open channels.
 2.5

x
The va1ue of u is re1ated to the velocity distribution by:

u
x 1
5.75
. ClU(z)
Cl (logz)
1
5.75
. logz
U(2) -
2-
U(I)
logz 1
but this method gives generally inaccurate resu1ts.

viscous sub1ayer 0
In the viscous sub1ayer viscosity predominates. The velocity
distribution therefore follows from T(Z) = T) ClU(z)/Clz
T(Z) = T
o
x
or U(z) = u Z
x \)
u
Intersection with the logarithmic velocity distribution gives a "theoretica1"
va1ue for 0:
o = 11. 6 v/ux
In fact there is a transition zone from the 1inear to the logarithmic profile
extending from:
Z = (5 to 30)V/ux

./
."Yo~/
.
• ,<-. /
~"Y ., .
tq~/ .
'<-~
y .
11.6 v/u x

Roughness va1ue ks

For uniform sediment k = D.

s
For graded sediment ks = D65 to D90.
For ripp1es ks = (0.5 to l)h.r~pp l'e
Errors in k give the fol1owing errors in C:
s
 2.6

k
actual
1 2 5 10
k
estimated

C
est-act 0 5.5 12.5 18

2.5. Turbulence

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 r.m.s. values, amplitude
distribution, correlations and spectra. The amplitudes are generally normally
distributed so that the root-mean-square deviation gives a good idea of the
fluctuations. cr
u
= ~u - Ü)2' where U = the instantaneous velocity and U the
time-averaged value.
A turbulent 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 and the introduction of a eddy-viscosity
concept for the apparent turbulent shear stress -p u'w"
w
-p ~ = P E au/az (u', w' are velocity fluctuations in horizontal
w w m an d' vert1ca 1 d'1rect10n
')

so that the total shear stress becomes :

T = n. au - p ~ = p (v + E ) au
az w w m äZ
E
m
= eddy viscosity.
The logarithmic velocity distribution:

U(z)/ux = lK In (z/z )
0

and the linear shear stress distribution:

T(Z) = T(O) (h - z)/h
give the following distribution for E (z)
m
(z) = K uXz(l - z/h)
E
m
The average value of E (z) (averaging over the depth) is therefore:
m
 2.7

2.6. Diffusion

The diffusion of scalair quantities (concentration, heat) is described

by analogy with the diffusion of momentum by:
N = (D + E )
c
ac/az
in which:
N = lateral flux of scalair quantity
D = molecular diffusivity
E = turbulent diffusion coefficient
c
C concentration
The value of D depends on the properties of the scalair:
-6 2
heat in water D - 0.2.10 m /s
salt in water D ~ 2.10-9 m2/s
The ratio of c to E m depends also on the properties of the scalair but the
E

value of this ratio is generally of the order one.

2.7. Literature

H. Rouse, 1950 Fundamental principles of flow, Engineering

Hydraulics
V.L. Streeter, 1966 Handbook of Fluid Mechanics, Chapter 9 and 10,
Mc Graw HilI New York
J.O. Hinze, 1976 Turbulence, Mc Graw HilI, New York, 2nd edition
P. Bradshaw, 1971 An introduction to turbulence and its measure-
ment, Pergamom Press, Oxford
H. Tennekes, 1972 A first course in turbulence
J.L. Lumley M.I.T. Press, Cambridge, Massachusets
P. Bradshaw (ed.), 1976 Turbulence, Springer Verlag, Berlin.
 2.8

2.8. Prob1ems

For all prob1ems g \)

Pw

2.1 Given: a wide open channe1 has the fo11owing characteristics:

depth h = 2 m roughness k
s
= 1 mm
slope 1 = 10-5
Question: compute Ü. (Is the bed rough/smooth/transition?).
same for k
s
= 0.05 mm and k
s
= 5 mmo

2.2 Given: wide open channe1:

depth h = 1.2 m ks = 0.5 mm
m3
discharge/m1 q = 0.8 /s.m
Question: compute slope I. Is the bed smooth/rough/transition?

2.3 Given: wide open channel:

k
s
= 5 mm
I 2.10-5
m3
q 1.6 /s.m
Question: compute depth h.

2.4 Given: measurements in a wide open channel gave the following velocity
profile:
U (z) = 0.148 log z/zo (U in mis, z in m.)
Questions: 1) compute UX.
2) compute k
s
if the velocity at z = 0.1 m was equal to
m
0.31 Is.

2.5 Given: Velocity measurements in a wide open alluvial channel gave the
following results:
at z = 0.1 m U(z) = 0.345 mis Sediment size D = 150 ~
m
at z = 0.3 m U(z) = 0.427 Is.
Questions: I) Compute ~ (from the difference in the U(z) values,
assuming the logarithmic velocity
distribution).
2) Compute k •
s
3) Compare k with D. Is the bed plane or are bedforms
s
present?
 3.1

3. PROPERTIES OF THE TRANSPORT MATERlAL

Some of the properties of sediment which are of ten used are:

size
shape
density
faU veloei ty
porosity

3.1. Size

A c1assification of partieles according to size is given in tab1e

3.1. This tab1e eives the c1assification by the American Geophysica1 Union
for c1ay, silt, sand, gravel, cobbles and bou1ders.
Various definitions of "diameter" are possib1e:
sieve diameter D = diameter of square mesh sieve whieh wi11 just pass
the partiele.
sedimentation diameter D = diameter of sphere with same density and
s
same settling velocity in same f1uid at same temperature.
nominal diameter D
n
= diameter of sphere with equa1 volume.
triaxial dimensions a, b, e (a = largest, c = smallest axis)

Size determination
bou1ders, eobbles and gravel: direct measurement
gravel, sand sieving
fine sand, silt sedimentation or microscope analysis

3.1.1. Sieving
Sieving ean be app1ied for partieles down to 44 ~ but gives good
results down to 74 ~m. Sieve sizes (openings) are made in a geometrie series
with every sieve being r:2 larger in size than the preceding. Taking every
other size gives a 12 series. For most sands a 12 series gives suffieient
results but a r:2 series may be necessary for very uniform sands. Some genera 1
rules for sieving ean be given:
1. Do not over1oad sieves to avoid clogging. The fo1lowing maximum
residues on individual 8-ineh sieves are reeommended (after Shergo1d
1946).
 3.2

Table 3.1

Major classification of sediment size

according to H.A. Einstein

Size Designation Remark

D < 0.5l-lm Colloids Always flocculated

0.5l-l < D < Su m Clay Sometimes or partially flocculated
5l-l < D < 64l-lm Silt Nonflocculating-individual crystals
64l-l < D < 2rmn Sand Rock fragments
2rmn < D Gravel, boulders Rock fragments

American Geophysical Vnion (AGV) grade scale for partiele sizes

Size
Class
Millimeters Microns Inches

4,000-2,000 160-80 Very large boulders

2,000-1,000 80-40 Large boulders
1,000-500 40-20 Medium boulders
500-250 20-10 Small boulders
250-130 10-5 Large cobbles
130-64 5-2.5 Smal 1 cobbles

64-32 2.5-1.3 Very coarse gravel

32-16 1.3-0.6 Coarse gravel
16-8 0.6-0.3 Medium gravel
8-4 0.3-0. 16 Fine gravel
4-2 0.16-0.08 Very fine gravel
2.00-1.00 2,000-1,000 Very coarse sand
1.00-0.50 1,000-500 Coarse sand
0.50-0.25 500-250 Medium sand
0.25-0.125 250-125 Fine sand
0.125-0.062 125-62 Very fine sand

0.062-0.031 62-31 Coarse silt

0.031-0.016 31-16 Medium silt
0.016-0.008 16-8 Fine silt
0.008-0.004 8-4 Very fine silt
0.004-0.002 4-2 Coarse clay
0.0020-0.0010 2-1 Medium c1ay
0.0010-0.0005 1-0.5 Fine clay
0.0005-0.00025 0.5-0.24 Very fine c1ay
I

.. "'"
 3.3

Hax'imum residue in grams

sieve opening u.s. Sieve
mm, nr.
2-series /i-series Yl-series

2.4 8 150 75 38
1.2 16 100 50 25
0.6 30 70 35 18
0.295 50 50 25 12
0.15 100 35 18 9
0.076 200 25 12 6

The total sample size should be about 20 - 50 grams for 8"-inch

sieves and fine sand.
2. A sieving time of 10 minutes with a mechanical sieving apparatus
should be used
3. For coarse sands and gravel the following minimum size 1S recotmnended
to obtain a sufficient number of grains in each fraction (see De
Vries 1971).
Sample size (gram) > 20.D853 D85 in tmn.
Sieve types and series are different in various countries, but are generally
based on a Yl-series.

3.1.2. Sedimentation
For fine sand and silt a size distribution can be determined by sedi-
mentation. For partieles < 50 ~m t~e Stokes law for the settling velocity is
valid; for coarser particles empirical relations have to be used. Various
principles are used: sedimentation balances, pipette analysis, visual
accumulation tube (fig. 3.1) (for a review see ASCE 1969). Sedimentation
gives of course no independent size and shape determination.
 3.4

GIG" Sellli"9 Tube .

T' .. kincJC ........

FIG. 3.1 .-SKETCH OF VlSUAL ACCUMULATIONTUBE ANDRECORDING

MECHANISM

3.1.3. Size distribution

By sieving or sedimentation a size distribution ean be obtained whieh
is generally expressed as a "pereent by -weight"vs "grain size" distribution.
The eumu1ative size distribution of most sediments ean be approximated by
a log-norma1 distribution. A log-norma1 distribution wi11 give a straight
1ine if logarithmie probabi1ity paper is used (figure 3.2).

From the eumu1ative size distribution the mean diameter ean be defined:
i i
D or D = ~ p.D./~ p.
m I.I. I.
I.nwhieh p. : fraction with diameter D .•
I. I.
D.I.is the geometrie mean of the size fraetion 1imits.
A1so the notation D is used whieh denotes the diameter in a mixture
p
of whieh p% is smaller than Dp' D50 is a1so ea11ed the median diameter
For a given distribution we ean define the geometrie mean diameter
Dg = (D84 • DI6)! (whieh is equa1 to 050 for a 10g-norma1 distribution)
and the geometrie standard deviation:

0g = ID84/D16I!
In geo10giea1 1iterature a1so I/)-units are used:
2
I/)
= - log D (D in mm)
I/)(1 mm) = 0, I/)(0.5 mm) = 1 ete.
og beeomes in I/)-units: - 1/)84).
0l/)= !(1/)16
 3.5

SIEVE Sf ZE: (mln)

Fi,_ 3.2 Example of cumulative diatribution of .ieve diam.eter on
!o,arithmic probabiUty paper

3.2. Shape

Beside of the grain-diameter also the shape is of importance. A flat

particle will have a smaller fall velocity and will be more difficult to
transport as a rounded particle as bed load.
Several definitions may be used to characterise the shape:
Sphericity ratio of the,surface area of a sphere and surface area of
the particle at equal volume
Roundness = ratio of the average radius of curvature of the edges and
the radius of circle inscribed in the maximum projected
area of the particle
Shape factor = s.f = c/I.älbin which a, band care three mutually perpendicular
axes, from which a is major, b is intermediate and c is minor
axis.
For spheres s.f = 1, for natural sands s.f ~ 0.7
Roundness and sphericity are not suited for practice whereas the shape factor
gives sufficient results for practical application.
 3.6

3.3. Density

Most sediments originate from disintegration or decomposition of rock.

c1ay fragments of fe1dspars and micas
si1t si1icas
sand quartz
gravel and bou1ders: fragments of origina1 rock
The density of most sediment partieles « 4 mm) varies between narrow
1imits. Since quartz is predominant in natura1 sediments the average density
can be assumed to be 2650 kg/m3 (specific gravity 2.65). Sometimes heavy
minerals are present which can be segregrated during ripp1e formation or
other modes of transport. C1ay minerals range from 2500 - 2700 kg/m3•

3.4. Fa11 velocity

The fa11 velocity of a sediment is an important parameter in studies

on suspension and sedimentation of sediments. The fa11 velocity is defined
by the equation giving equilibrium between gravity force and flow resistance:

gravity resistance
in which CD = drag coefficient
W = fa11 velocity
From this re1ation fo11ows:

W = ( j . ~ . Ö)!
D
in which Ö = (p - p )/p
s w w

Va1ues of CD dep end on a Reyno1d's number W.D/V and the shape of the partiele
(expressed by s.f = c/l.S'b)

For spherica1 partieles and 10w Reyno1ds number (Re < 1), CD can he
given by CD = 24/Re so that:

(Stokes 1aw)

For large Reyno1ds numhers CD hecomes a constant so that W varies as:

(ÖgD)!
Therefore W varies with D~ to 2
 3.7

Relations between CD' Re and s.f are given by Albertson (1953)

(see Figure 3.3). For natural sands s.f ~ 0.7. From these relations graphs
for W as a function of grain size, shape and temperature can be obtained
(see Figure 3.4).
10 -T I
8
6
,!,- I I
... I I, I
4 r-~ Shapefactor = 7foi
~ ~ t:::r- I 0.;.~J_
--" ~t.
2
<, r- ----- r05
---
, N -----rQ}r-- --
'"
U 0.8
r-.
<,
I
0.6
0.4

0.2
i
I
I
<, ~o

- -
I
I

o. I I
4 6 810 2 4 € 8 102 2 4 6 8 103 2 4 6 8 104 2 3
Re.~
'"
Fil.3.3 Drag coefficient vs. Reynolds number for different shape factors. [Af ter
A1..BERTSON (1953).1

The presence of large number of other particles will decrease the fall
velocity of a single particie. A cluster of particles will have a greater
velocity however. Therefore care must be taken with experiments on the
fall velocity to avoid currents in the fluid that will influence the fall
velocity of the particle and the influence of concentration should be con-
sidered.
There are many expressions giving the influence of concentration on the
fall velocity. Based on systematic experiments, Richardson and Zaki (1954)
g1ve a useful expression:
W(c)/W(o) = (1 - c)a o~ c < 0.3
W(c) is the fall velocity of a grain in a suspension with eoneentration
by volume e
W(o) is the fall velocity for a single grain
a is a funetion of Reynolds number W.D/v

Re'< 0.2 a = 4.65

0.2 < Re < 1 a = 4.35.Re-0•03
1 < Re < 200 a = 4.45.Re-0•l
Re > 500 a = 2.39
The coeffieient is slightly dependent on particle shape but this can
be neglected. For fine sediments this means that a concentration of 1% gives
a reduction in fall velocity of 5%.
The f~ll velocity of a partiele in a turbulent fluid ean be different
from that in a quiescent fluid (see chapter 6.2).
 3.8

0
en
w
0 ...J
10
U
......
a::
~
N
.....
a::
<l
:::>
Q 0

Z
a::
0
en ~
~
ti z
UJ
~ ~ ~
(I) ...J )(
<l UJ
~ a::
:::> UJ
u ..... ~
0 <l z
0 Z IL.
Q z
a::
E-t 0 IL.
u
u u, 0
."

"
0
.
-til

13
t.)
Ê'
u
.~

->.
>-
t:
ct:
UJ
I-
U Cf
0 ~
4-l
til
C"')

L/")

11
..
u
0

>
...J

>
0
W UJ

...J
..J
..J
::;: ëi ~
U. ...J (I)
til <l
~ Q) u, 0
..;:t .~
:>
co
0
C ~
0 Z
Z UJ
.~til
,..c:
<l 0
(I)
~ E-t a:: UJ
W :>
Q)
J- a
.-I W
0- ~ Z
~ <l
:< C UJ
w Z
0
LU ..J
> ~
W Cl)

è
en z
::J
LL.. ..J
0
~
z
0
.....
<l
...J
W
a::
'IIt
eD
M
.
Cl)

0 Cl) eDlO ~ rt') N 0 Cl)

..
eDlO ~ rt')
"! 0 q C)
u,
ww UI 'JI,aWOlp iAilS
 3.9

3.5. Bulk density and porosity

In estimating the life of a reservoir and similar cases the calculated

wei3ht of the sediment transported to the reservoir has to be converted into
volume. For this the dry mass per unit volume of sediment in place, bulk
density, Pb' has to be estirnated.
For instance for air-dried fine sediments 1200-2000 kg/m3 applies. The
same material deposited under continuously submerged conditions may range
from 300 - 1000 kg/m3• The density will also depend on the grainsize and
silt content.

Bulk density, Pb = the mass of dry sedimentary material within a unit

of volume (kg/m3). The volume
taken by the sediment depends on the conditions of settling and may be a
function of time due to consolidation. An empirical relation is presented by
Lane and Koelzer (1953) for estimating the bulk density of deposits in
reservoirs:
Pb + Blog T (1 - E)p
S
1
relative pore volume (porosity)
..time in years
= initial bulk density taken to be the value af ter one year of
consolidation
B = consolidation coefficient

sand silt clay

Reservoir operations Pb Pb Pb
B B B
1 1 1

sediment always submerged

or nearly submerged 1500 0 1050 90 500 250

normally a moderate
reservoir drawdown 1500 0 1185 45 750 170

norma11y considerahle
1500 0 1275 15 950 100
reservoir drawdown

reservoir norma11y empty 1500 0 1320 0 1250 0

 3.10

Lane and Koe1zer a1so gave the simp1e re1ation Pb = 817(P + 2)0.13
1
in which P = percentage of sand.
Lara and Pemberton (1963) ana1ysed 1316 samples and gave somewhat
different va1ues of Pb (in kg /m3). The fo1lowing size c1assification was used:
1
clay: material < 4~m
si1t: material 4 to 62.5 ~m
sand: material > 62.5 ~m

Pb
Type Reservoir operation 1
clay silt sand
I Sediment a1ways submerged
or nearly submerged 420 1120 1550
II Norma11y moderate to considerab1e
reservoir drawdown 560 1135 1550

III Reservoir normally empty 640 1150 1550

IV River-bed sediments 960 1170 1550

The r.m.s. deviation for the correlation was 200 kg/m3 which means
that considerable deviations are possible.
Examp1e: A sediment in a type I reservoir contains 20% c1ay, 45% silt and
35% sand. The density of the sediment wi1l then be
Pb = 0.20 x 420 + 0.45 x 1120 + 0.35 x 1550 = 1130 kg/m3
1

Murthy and Banerjee (1976) analysed 832 samples from Indian reservoirs
with type 11 operation. The fo11owing va1ues of Pb1 were obtained:
sand:1506 kg/m3 si1t: 866 kg/m3 c1ay: 561 kg/m3
The results cannot be compared directly with Lara and Pemberton because the
division between sand and si1t was taken at 20 ~.

3.6. Literature

F.A. Shergold, 1946 The effect of sieve loading on the results of

sieve analysis of natural sands.
J. Soc. of Chem. Ind. London, 65, p. 245 - 249

M.L. Albertson, 1953 Effect of shape on the fall velocity of gravel

partieles.
Proc. 5th Hydr. Conf. Univ. of Iowa, Bull. no. 34.
 3.11

E.W. Lane, 1953 Density of sediments deposited in reservoirs.

V.A. Koelzer Rep. no. 9 of a study of methods used in
measurements and analysis of sediment loads
in streams.
Univ. of lowa, lowa.

J.F. Richardson, 1954 Sedimentation and fluidisation.

W. Zaki Trans. lnst. Chem. Eng. 32, p. 35 - 53

W. Batel, 1960 Korngrössen Messtechnik.

Springer Verlag, Berlin

V.A. Vanoni, 1961 Lecture notes on sediment transport and channel

stability.
Calif. lnst. of Technology, Rep. KR-R.I.

J .M. Lara, 1963 lnitial Unit weight of deposited sediments.

E.L. Pemberton Proc. Federal lnteragency Sedimentation Conf.
p. 818 - 845, Mise. Publ. no. 970, US Dept. of
Agriculture.

N.A. Fuchs, 1964 The mechanics of aerosols.

Pergamon Oxford.

T. Allen, 1968 Partiele size measurement.

Chapman and Hall London.

H.P. Guy, 1969 Laboratory theory and methods for sediment

analysis.
US Geol.Survey. Techniques of Water Research
lnvestigations, Book 5, Chapter Cl.

ASCE, 1969 Task Comm. on Preparation of Sedimentation Manual

Sediment'measurement techniques.
Proc. ASCE 95 (HY5), pp. 1515 - 1543.
 3.12

M.de Vries, 1970 On the accuracy of bed-materia1 sampling.

J. of Hydr. Research 8 (4), pp. 523 - 533.

R.J. Gibbs, 1972 The accuracy of partic1e-size ana1ysis uti1izing

settling tubes.
J. Sed. Petr. 42 (1) pp. 141 - 145.

B.N. Hurthy, 1976 Initial unit weight of deposited sediments in

B.K. Banerjee reservoirs with considerab1e draw-down.
Proc. Symp. on Mode1ing Techniques in
Hydrau1ic Engineering, Poon a, Nov. 1976 l,
paper A 6.
 4.1

4. INITIATION OF PARTICLE MOT ION

4.1. Introduction

The equilibrium of a particle on the bed of a stream is disturbed if

the resultant effect of the disturbing forces (drag force, lift force,
viscous forces on the particle surface) becomes greater than the stabilising
forces as gravity and cohesion. Cohesion is only important for sediments in
the clay and silt range or fine sands with an appreciable silt content. The
acting forces have to be expressed in known quantities such as velocities
or bottom shear stress. They will have a strongly fluctuating character so-
that the initiation of motion also has a statistical aspect.
Theoretical work on the initiation of a motion has started with work
by Brahms (1753) who gave a sixth power relation between flow velocity and
the necessary weight of a stone and by Dubuat (1779, 1736) who introduced
the concept of bottom shear stress and did some experiments on particle move-
ment. Host of the older relations have the farm:

u bo ttom, . =
er i,t
(4 - 5) /D (D in m, U in mIs)

As the "bottom" ~s not weIl defined the use of this type of formula is
limited.

4.2. Theory

White (1940) gave a thorough discussion on the equilibrium of a grain

on the bed of a stream.

The disturbing force F (resultant of drag

and lift farces) will be proportional to
the bottom shear stress tand the particle
o
surface area (D2).
The stabilizing gravity force is proportio-
nal to (p - Pw)gD3. Taking the moment
s
G
with respect to the turning point S gives
the equation:
alT .D2 ~ a2 (p - p )gD3
o S W

or: T ~ C(p - P )gD

o s w
 4.2

The factor C will depend on the flow condition near the bed, partiele
shape, the position of the partiele relative to other partieles etc.
The flow condition near the bed can be described by the ratio of grainsize
to thickness of the viscous sublayer which ratio is proportional to
UX D/V = Rex, a Reynoldsnumber based on grainsize and shear velocity.
All other theoretical considerations based for example on drag force
due to velocity will give the same result that:

lJJcr = UX2cr /f..gD=

4.3. Experiments

The relation:

T
cr X
lJJ cr = "(-p ---P~)-g~D = f (Re )
s w
has been investigated by many authors especially by Shields (1936) who did
systematic tests and compared his results with results from other investi-
gations (see figure 4.1). The difficulty in all tests is the definition of
"initiation" of motion. It is the movement of the first partiele or of a
large number of grains? Shields correlated the rate of sediment transport
with
o and defined Tcr by extrapolating to zero material transport.
T
X
For large Re Crough bed) it can be seen that UX varies with 1:5
cr
- X
(figure 4.2). For equal values of h/D and therefore equal values of U/U
it follows that Ü cr - ID and that the critical velocity of a stone is pro-
portional to the 1/6 power of the weight of the stone (or stone weight pro-
portional to Ü6).

4.4. Influence of various factors

4.4.1. Effect of criterion

It is clear that the critical value of T will depend on the criterion

o
for initiation of motion. To get an objective criterion Neill (1968, 1969)
proposed the dimensionless parameter :

in which n is the number of grains displaced per unit area and unit time.
-6
Shields graph corresponds roughly with a N-value of 15.10 for coarse
material. For designs of bottom protections etc. a much lower criterion
 4.3

0 o
o
Q
(X)

til
o c- til
0
-~- ~
0
c
E """:
.til9
~
~ 0- • c:
z .~ Q)
c~1 "" E
-20 ~ N
~
lOm - 0
U>c:t 1.0"':"
m
NNNN~
~~ o
\ Q
T
-_ f------- r!-+-+-- \
(X)

e
U> - --
c Q) Q)
E
Q) 10 0
c ~ ~
....> 0 L; L;
~
I • u ::::::-3: 3:

.~~
<J lil §III ~~~
c"O -- ....+-
1& ~ c c L;-ö 0 L;
2:~ ~_!: lil

N "O~"O"OQj

'J l
e ,
§E §
(/)(.!)(/)(/)
g~ (/)

Q •, 0 0 <J

....,0
~.,. Q.
(X)
I"-
W
7 10
--
..vj ....
E
u WI"- 1010101010
~

..
<, ONI"-NWWU>W
y N
C>
- "':"':N<:tNNNN
~ ~
~~
_/
_L -_ o
-
lil
"0 -
Qi -(/)-

/
/
L; -'-w+-
>. QJi • '-
(/) Q)E3:Q)
/ ---
C
.Q ~ lil C . ..0
o ....~:=
./ ~Q. Q) u::.::::::::>(.!)
.~ "'!!.~
u Q).- C ~ "0
Q,)----
7"
~.e
"0 "0 "0
--- c:t Q)
lil ..occ'::cccc
o E.~ .... coc c c
/ ~
Cl <X...J(.!)(X)(/)(/)(/)(/)

,I Ö E
N
>.
(/)
0
•eee
o

~
-
o
o (X)

o 00
1.010 ~
0
~
0
N
o -
0
0
0 0
(X) 1.010 c:t
00 0
00 0
o
0
N
0
0
0
0
a~d'i7
.IJ
.i
, "'
 4,4

...
C) "l
52"''' ':" 10 '" ..

1'IIIi"", !\ -e- (")

- Cl) """ ",.

~
lil..
,
\'0
1l. "
" ~
i\-j
--
~
E

~'"
~"~~
, .......\,) C)
\()
I -- e- ..,
o ~
11. \
C\j

I~I
IJ

~ cr
"
-
'"S
eo
~
1\
1'11...
~
-
Cl

IQ

o
IJ')
<o
"-
...." \.
'; ~
'" "" -- ~.
N

a.
C/J
-,r", ~

r-,
..,

"
,_.~ b
4-1
H
o ~
'"
' ~~'" E
..
" l.
,
""
. J"O'.>~
-E

'",
....... (..
o ~ -d'i
IQ

\~
~

~
::s:
C
\
\'
, ~
.~

,
.
..,

'"
L-
~ l
. -
---
-
c::;
Cl

10
"0
Q)
:>
-.-I
H
.. _ ..
Q)
"0 -- ..,
Q)
H
cd -t--- f-- 1------ -
C/J '"
Q)
:>
U
H
;::l

-
ti"'''
I
I
i
.. .., .. ..,
-
C)
o
'" "10 It)
'" C5'" Cl " IQ '"
'"
c::;

...
I,
 4.S

should be used (for instanee N = 10-6). Also Paintal (1971) has measured
very low rates of transport with coarse material down to ~ = 0.02, thus weIl
below the Shields value (see Figure 4.3).

Shields experiments were done with several types of material and syste-
matic influence of shape could not be observed. Tests at the Delft Hydraulics
Laboratory with coarse material showed that the critical value of ~ is the
same for various shapes (spheres, cubes, broken stones etc.) if the nominal
diameter D is used for comparison.
n

It will be clear that a wide gradation will have an influence on T •

cr
In practice however the gradation has an influence for D9S/DS > 5 only
(Knoroz, 1971), because the larger grains are more exposed and smaller grains
are shielded by the larger ones. Therefore D50 is a good measure for most
samples. For the effect of a gradation also see Eguisaroff (1965).
For a wide partiele gradation the effect of armoring will occur which
means that fine partieles are eroded and an armor layer of coarse partieles
is formed, which prevents the bed from further scour. This effect is very
important in degradation downstream of dams (Livesey, 1963, Gessier 1970).
In that case DS5 to D95 can be taken as a representative value for the
mixture.

For small values of h/D (waterdepth/particle diameter) a deviation

from Shields graph is possible because T is not representative in that case
o
for the turbulent flow structure. The turbulence structure ne ar the bed in
an infinite fluid is completely defined by bed shear stress (T ) and rough-
o
ness (ks ) but for small values of h/D also the waterdepth gives a limitation
on size and frequency of the large eddies. Also the ratio of eddy duration
and the time necessary to accelerate aparticle becomes small sothat an
influence of h/D may be expected (more stability with smaller h/D). Experi-
ments have indeed shown that ~ increases with decreasing h/D (Ashida 1973).
cr

I'.
 4.6

~1 01
-
_ ... -• ---
-~
.....A-- • •
- 'I'
-
--
.-.
-: - •
~
q,. cr:·W....16
• I
-
0.01 _
.0 2.5mm
0= 22.2 mm
'I'

- 0= 7.95 mm
0.001
10 9 10 8 10-7 10 6 10-5 10-4 10-3 10-2
---~q"

Variation of bed load transport at low shear Débit de charriage à tension de frottement
values. faible.

qs = sed.tr·/m.s

~1 -
rJr
i.dli ~
~ iJVV'9'"
~~:4'.... v IV
....~ ... v v

r q,' cc '" 2.5

I~
0.1

_.- ,-'
... I.IA!
,.
• ..
1r.6 ~

--..
qs.cx: .~16
• I •
I .... _ll!!:. ...!:!I1
4~ •
IJ

6.USWES
• Casey
v Gilbert
0.01
10-4 10-3 10-2 10 1 10
----q,.
Variation of bed load transport at high shear Débit de charriage à tension de frottement
values. élevée.

·0.07
0.06
~.
--
•
'"
1
0.05
0.04 .".
0.03
0= 7.95 mm
0.02
o 10 20 30 40 sa 60 70 80
--_ q, (Ib/ft/hr)
0.07

'" 1 _t~____
0.06
0.05
/----
1--
_tL
-s •
0.04 •
lf - 0.03
0= 2.5mm
0.02
o 10 20 30 40 sa 60 70
--_ q, (Ib/ft/hr)

Determination of critical shear stress. Détermination de la tension de frottement

critique.

Figure 4.3 Measurements by Paintal

 4.7

For a partic1e on a slope the va1ue of T wi11

cr
be reduced. For a horizontal bed the re1ation
F{o) = G tancp
is va1id, in which cp is an ang1e characteristic
for the partic1e stabi1ity.

For a ~~~_~!~E~_b~_!~~_É!~~_~bE~~!b~~ with ang1e

a the fo110wing stabi1ity condition ho1ds:
F(a) + G sina = N tancp=
G cosa tancp
F(a) = G cosa tancp- G sina
F(a) G cosa tancp- G sina
F(o) = G tancp
=cosa sinp - sina cosCP
sincp
F(a) sin(cp- a)
k(a) = F(o) = sincp (given by Schok1itsch in 1914!)

For a ~b~~_~!~E~_~!!~_~~g!~_ê
Stabi1ity condition

R h(3) 2 + G2sin213= G cosl3tanep

F (13) IG2cos213 tan2cp- G2sin213' IIcf~~;1
_ F(I3)_ ~cós213 tan2cp - sin213•
kél3) -F('ö)- Á. _
= cosl3
tan213'
(given by Leiner in 1912!)
tan2cp tan2cp
For a combination of longitudina1 and side slope the reduction factor
k(a,l3)becomes k(a,l3)= k(a).k(I3).

It might be expected that an inflow or outflow of water from a sand

bed has an influence on the stability of the sand particles. The pore-water
flow may be caused by a ground-water tab1e lower or higher than the river
water level. It has been shown by Oldenziel and Brink (1974) however, that
the inf1uence is very limited. For hydrau1ic gradients up to + 0.3 on1y
a factor of 2 in the transport rate was observed. In view of the strong
variation of transport rate with ~ near incipient motion this means on1y
a few percent variation in ~ and can be neglected {inflow of water had
cr
 4.8

a somewhat stabilising tendency). There is one exception however. Harrison

and Clayton have shown that a seepage into the bed for a flow carrying fine
silt particles gives an enormous increase in stability due to the formation
of a plastered bed layer.

4.5 Cohesive sediments

A cohesive character of a soil will increase the resistance against

erosion. Empirical data on critical mean velocities are given by Lane 1953.

material loose moderately compact compact

sandy clay 0.45 mis 0.9 mis 1.25 mis

clay 0.35 mis 0.8 mis 1.20 mis
lean clayey soil 0~30 mis 0.7 mis 1.05 mis

Several authors have tried to correlate critical shear stress with mecha-
nical properties of the soil (siltcontent, plasticity index, vane shear
strength) (see Smerdon and Beasly (1959), Carlson and Enger (1960),
Partheniades (1965, 1970). From the data given it appears that for cohesive
soils with D50 = 10 - 100 ~ a critical shear velocity U~r of 3 - 4,5 cm/s
is possible.
There is some tendency for an increase of UX with vane shear strength
cr
and plasticity index.
For very recently deposited sediments (silt in estuaries) Migniot
(1968) and Partheniades (1970) give relations between UX , vane shear strength
cr
and dry weight of the sediments. Minimum values are in the order of ~
cr
1.0 cm/s (consolidation period of some days) to 3.0 cm/s for consolidation
periods of some weeks. For an example see Figure 4.4 taken from Terwindt
and Breusers (1972).
For an exact determination of a critical shear stress of a cohesive
soil a special test for each soil will be necessary. Raudkivi (1974) and
Arulanandan (1975) have shown that the erosion resistance of clay depends
very much on the type of clay mineral and the chemical composition (salts)
of the pore water and the eroding fluid.
 4.9

_ VOL .• t, WATER

100 9~ 90 8!> 80
6

u
"
" .
...
011

E
4 "
u "
u
3 i--- - HQURS

~,"... ...
•
::::>
2

~ t~~
23
.
r o
o
tI
100 200 300 400 !>OO 600

_C (g/I)

SANO THICKNESS
MUD: IN OF MUDLAYER
-t, IN cm

6.LA VILLAINE· (MIGNIOT ,1968 ) e 12

•• MAHURY • (MIGNIOT, 1968) 2 12

--+-- I 37 2
• n 7 20
o m 2

Fig. 4.4 Critica 1 shear velocity (U.c) in relation to mud concentration (c).

The stability of stones on dams or in revetments is discussed by

several authors. Taking a "safe" value for the Shields parameter 1jJ = 0.03
and ks ~ 2 D (in view of the large roughness of stones) the following
relation is obtained:

6h
= 1.0 log D

Isbash (1935) neglects the influence of h/D and gives the empirical relation
for the stability of a stone in a bed:

Ucr = 1.2 12llgD = 1.7 IllgD

For a stone on the top of a dam the critical velocity is reduced~

Ucr = 0.86 12gt.D = 1,2 IllgD

Ooncharov (see Shamov 1959) gives the following relations:

Ucr
y'LSgD = O •75 1og--8.8h for absolute rest of a stone
D

and ;X~~
= 1.07 log 8~8h for the critical condition.
 4.10

Levi (see Shamov 1959) gives the empirical relation:

Ucr I 4 (~)0,2
l~gDi . D

Maynord (1978) gives the empirical expression:

D50
Fr = u
11 Igh

This can be converted into (taking ~ = 1.65):

All relations are compared in Fig. 4.5

The formulas given do not take into account the influence of turbulence
generated by constructions for example dams.
In that case the critical velocity has to be reduced with a factor
1.45
a = 1+3r
in which r is the relative turbulence intensity and a value r = 0.15 has
been assumed in uniform flow over a rough bed.
Just downstream of a hydraulic jump (stilling basin) values of
r in the order of 0.3 to 0.35 can be expected. This gives a value for
a of about
a = 0.7
This agrees with the design graphs given by Cox (1958).
 4.) )

I I
Uer I1
I
!
i
Y4,Z:; I I
!

--/JlllfS H:
_.--.-._ --
itroN6
--4:.....
VN
.... --

. I
I I
11

. I1

()
1 I I
I /0 ~ /00

..2>
;:/~ ~. S <CA;t:.;7'AI. "~~tlC; '7'/~J ;O,e J7D~4./
 4.12

4.7. Literature

A. Brahms, 1753 Anfangsgründe der Deich- und Wasserbaukunst,

Aurich
L.G. Dubuat, 1779 Principes d'Hydrau1ique, Paris
L.G. Dubuat, 1786 Principes d'Hydrau1ique et de Pyrodynamique,
Paris
O. Leiner, 1912 Zur Erforschung der Geschiebe- und Sinkstoff-
bewegung,
Zeitschrift für Bauwesen ~, p. 490 - 515
A. Schok1itsch, 1914 Uber Sch1eppkraft und Geschiebebewegung,
Leipzig und Ber1in, W. Engelman
A. Shie1ds, 1936 Anwendung der Aehn1ichkeitsmechanik der
Turbu1enzforschung auf die Geschiebebewegung
Mitt. der Preuss. Versuchsansta1t für
Wasserbau und Schiffbau Ber1in, Heft 26
C.M. White, 1940 The equilibrium of grains on the bed of
a stream, Proc. Roya1 Society London A 174,
no , 958, p. 332
E.W. Lane, 1953 Progress report on studies on the design of
stab1e channe1s, Proc. ASCE ~, sep. 280
E.W. Lane, 1954 Some observations on the effect of partic1e
E.J. Car1son
shape and the movement of coarse sediment.
T. Am. Geoph. Uno 35 (3), p. 453-462
R.T. Smerdon, 1959 The tractive force theory app1ied to stabi1ity
R.P. Beas1ey
of open channe1s in cohesive soi1s.
Co11. of Agrig. Exp. Station. Res. Bull 715
E.J. Car1son, 1962 Tractive force studies of cohesive soi1s
P.F. Enger
for design of earth cana1s. U.S. Bur. of
Reel. Denver Colorado.
R.H. Livesey, 1963 Channe1 armoring be10w Fort Randall Dam
Proc. Fed. Interagency Sed. Conf. U.S. Dept.
of Agric. Mise. pub1. 970, p. 461-470
E. Partheniades, 1965 Erosion and deposition of cohesive soi1s.
Proc. ASCE 2! (HY1) , p. 105-139
I.V. Eguisaroff, 1965 Ca1cu1ations of non-uniform sediment concentra-
tion. Proc. ASCE 91 (HY4), ju1y, p. 225-247.
 4.13

AS CE , 1966 Initiation of motion. Proc. ASCE 92,

pp. 291 - 314.
ASCE, 1968 Erosion of cohesive sediment Proc. ASCE 94
(HY4).
C.M. Migniot, 1968 Etude des propriétés physiques de different
sediments tres fines et de leur comportement
sous des actions hydrodynamiques. La Houi11e
Blanche 23 (7), p. 591 - 620.
C.R. Nei11 , 1968 A re-examination of the beginning of movement
for coarse granu1ar-bed materiais, H.R.S.
Wa11ingford.
C.R. Nei11, 1969 Quantitative definition of beginning of
bed movement, Proc. ASCE 95 (HY1), p. 585-588
E. Partheniades, 1970 Erodibi1ity of channe1s with cohesive
boundary, Proc. ASCE 96 (HY3), p. 755-771
J. Gessier, 1970 Se1f-stabi1izing tendencies of a11uvia1
channe1s, Proc. ASCE 96 (WW2), May, p. 225-249
S.S. Harrison, 1970 Effects of ground-water seepage on f1uvia1
L. C1ayton
processes. Bull. Geo1. Soc. Am. no. 811,
p. 1217-1225
A.S. Painta1, 1971 Concept of critica1 shear stress in loose
boundary open channe1s. J. Hydr. Res. ~ (1),
p , 91-113
V.S. Knoroz, 1971 Natura1 armouring and its effects on
deformations of channe1 beds formed by materials
non-uniform in size. Proc. 14th IAHR Congress,
Paris, paper ~, p. 35-42.
J.H.J. Terwindt, 1972 Experiments on the origin of f1aser, 1enticu1ar
H.N.C. Breusers
and sand-c1ay a1ternating bedding. Sedimento1ogy
.!2_, p. 85-98.
K. Ashida, 1973 Initiation of motion and roughness of flow
M. Bayazit
in steep channe1s. Proc. IAHR Congress,
Istanbu1, paper A 58.
D.M. 01denziel, 1974 Inf1uence of suction and b10wing on entrainment
W.E. Brink
of sand particles. Proc. ASCE 100 (HY7),
p. 935-949.
A.J. Raudkivi, 1974 Erosion kaolinite c1ay by f10wing water.
D.L. Hutchinson
Proc. Royal Society London A 337, p. 537-554.
 4.14

K. Aru1anandan, 1975 Pore and eroding f1uid inf1uence on surf ace

P. Loganathan, R.B. Krone
erosion of soils. Proc. ASCE 101 (GT1),
p. 51-66.

Stabi1ity of stones

S.V. Isbash, 1935 Construction of dams and other structures

by dumping stones into f10wing water. Trans.
Res. Inst. Hydrot. Leningrad II p 12-66.
See also \
S.V. Isbash, Kh.Khaldre, 1970 Hydraulics of river channel closure.
Butterworths, London.
R.G. Cox, 1958 Velocity forees on submerged rock lJ.S.W.E..
S.
Vicksburg. Mise. Paper No 2-265.
G.I. Shamov, 1959 River Sediments.
Leningrad, Gidrometeorizdat.
S.T. Maynord, 1978 Practical riprap design.
U.S.W.E.S. Vieksburg, Mise. Paper H-78-7.
 4.15

4.8. Problems

Use Shields curve and k D unless otherwise specified.

s

4.1 Given: A wide open channel excavated in uniform material (ps = 2650 kg/m3)
with D = 2 mm has a slope 1 = 0.5.10-3 and a depth of h = 2 m.
Question: Is the channel bed stable?

4.2 Given: A wide open channel has a depth of h = 1.7 m, a mean velocity
Ü= 2.5 mIs.
Question: What is the minimum size of the bed material to obtain a stable
bed? Ps = 2650 kg/m3

4.3 Given: A wide open channel has a slope 1 = 10-5 and bed material
D = 0.2 mmo No bedforms are present.
Question: What is the maximum discharge Iml without movement of bed
material. (Ps = 2650 kg/m3)?

4.4 Given: A wide open channel is excavated in uniform material with

(Ps = 2650 kg/m3) and
D = 3 mm under a slope 1 = 10-4.
Question: What is the permissible discharge/ml ?

4.5 Given: The bottom of a wide open channel with a depth of 4 m is protected
with stones with a mass of 30 kg. Ps = 2800 kg/m3•
Question: What is the critical mean velocity for this bottom protection,
using ~cr = 0.03 and the nominal diameter as the representative
size.

4.6 Given: Experiments are designed to check Shields curve, using a wide
flume (neglect side-wall effects). The waterdepth for the
experiments is 0.6 m.
Question: If uniform flow is required (water surface slope = bed slope),
what is the required slope of the channel bed and discharge/ml
for: a) an experiment with uniform sand ks = D = 200 um;
b) an experiment with uniform gravel ks = D = 4 mm ,

Ps = 2650 kg/m3
 5.1

5 TRANSPORT MECHANISM, BED FORMS, ALLUVIAL ROUGHNESS

5.1 Introduction

For turbulent flow over a rigid bed a description of the flow structure
could be given only by empirical methods. Bottom shear stress, waterdepth
and bed roughness were the most important parameters. Description of particle
motion under the action of the flow is also largely empirical sothat it is
not difficult to understand why there is only a limited theoretical basis for
the relation between flow and sediment transport.
Most of the existing knowledge is obtained from experiments and general
physical arguments. For the initiation of motion a reasonable picture was
obtained in this way. At greater values of the bed-shear stress sediment
transport will increase and deformation of the bed will occur. As the defor-
mation is also time-dependent and nature is always unsteady, an equilibrium
situation will be hardly found in practice.

5.2 Transport mechanism

According to the mechanism of transport two major modes may be dis-

tinguished:
1. Bed load - movement of particles in contact with the bed by
rolling, sliding and jumping
2. Suspended load - movement of particles in the flow. The settling
tendency of the particle is continuously compensated
by the diffusive act ion of the turbulent flow field.
A sharp distinction is not possible. A general criterion for the beginning
of suspended load is a ratio of shear velocity and fall velocity UX/W ~ l.S.
Sometimes also saltation load is mentioned. This is the mode where particles
bounce from one position to another. This is only important for particle move-
ment in air. The maximum particle elevation of a particle moving in water is
in the order of 2-3 times the diameter sothat this mode 'of transport can be
considered as bed load.
According to thé origin of the transported material a distinction is
made as follows:
A. Bed-matérial This transport has its origin in the bed, which
transport
means that the transport is determined by the bed
and flow conditions (can consist of bed load and
suspended load).
 5.2

B. Wash load Transport of particles not or in small quantities in

the bed. The material is supplied by external sources
(erosion) and no direct relationship with the local
conditions exists (can only be transported as suspended
load, generally fine material < 50 ~m). It can have
influence on turbulence and viscosity and therefore
have some influence on the flow.

mec bed material transport

suspended
wash load
Wash load is not important for changes in the bed of a river but only for
sedimentation in reservoirs etc.

5.3 Bed Forms

Much literature exists on the classification and dimensions of bedforms,

mainly in the form of empirical relations. Bed forms are of interest in
practice for several reasons.
Bed forms determine the roughness of a stream. A change in bed
form can give changes in friction factor of 4 and more.
- Navigation is limited by the maximum bed level and depends therefore
on the height of the bed deformation.
- Bed forms and sediment transport have a mutual influence.

A generally accepted classification is the following:

A. Lower flow regime (Froude number Fr = Ü/!gh < 0.4 to 1; no sharp transition).

A.l flat bed At values of the bed shear stress just above the critical~
sediment transport without deformation of the bed is possible. Grains
are transported by rolling and bouncing.
A.2 ripples For sediment sizes < 0.6 mm and and increasing bed shear stress
small regular waves appear with wavelenghts in the order of 5-10 cm
and heights in the order of 1 cm. They become gradually irregular and
three-dimensional in character.
A.3 dunes For all sediment sizes and increasing shear stress dunes
are developed. Dunes are more two-dimensional than ripples and have

Ih· ."~
 5.3

much greater wavelengths and heights. The crests of the waves are per-
pendicular to the flow, the form is more or less triangular with a
gentie slope along whi.ch the particles are transported and a steep
downstream slope where particles are deposited. The angle of this
slope is roughly the angle of repose of the material.

B. Upperflow regime (Fr> 0.4 to 1.0)

B.l plane bed As the velocity is further increased, the dunes are
flattened, gradually disappear and the bed becomes flat. Sediment
transport rates are high.
B.2 antidunes A further increase 1n velocity to Froude numbers around 1.0
causes the water surf ace to become instabie. Interaction of surface
waves and the bed (sediment transport is maximum under the troughs of
the surface waves) gives a bed form called antidunes.
They can travel upstream and occur in trains of 4 to 20. Antidunes and
surface waves grow in amplitude and of ten break in a way similar to
ocean waves.
B.3 chute and pools At still higher velocities chutes and pools are formed.
For an illustration of the bed forms see figure 5.1 (Simons and
Richardson 1968).

~~~T-:::"~>:')~:i ';•..:.\.~
:::::;::.::." '.
..~.: ... C
..:/.,....
:.:.:.r.,.:;;.
:.:.::.:::::::.• : ...•.• :::.::::;:: ... i-.

(0) Typical ripple pattern (,,) Plane bed

Weak boil

(c) Dunes (g) Ant idune breaking wave

_r@,îr\~\~ît~"i%ffi,~Ê";W"i'~l~I;!1
(d) Washed-out dunes or transition

FI,.5.1 ldealizedbedforms in alluvial channels. [AfIer Sn,tONS el al. (/961).]

 5.4

5.4 Classification Criteria

Several authors have tried to develop theoretical explanations for the

origin of ripples and dunes (see for example Exner (1925) who discusses the
growth of an initial instability on a sand bed.)
Other authors have assumed potential flow to predict the reaction of
the main-flow on variations in bed level (Kennedy, 1963). The result of
Kennedy's work is arelation between the wavelength L of the bed deformation
and the Froude number (see figure 5.2). U
Fr = 7ëh
2.8 I
I
Fr
I
NF:(NF)m

2.4
I
:~/:o
I
·
..
•
Borton
Broaks
Brooxs
a Lin 0.I8-mm sand
0.086 - mm sand
0.145 - mm sand
2.0 • Kennedy 0.157- 0.46- mm sand

.
I
I
•
•
Kennedy
Kennedy
0.233 - mm sand
0.549 - mm sand

..,
• Laursen O.I-mm sand
'I Plate 0.253 - mm sand
1.6 \i Plate 0.350-mm sand
.. Plate 0.448 - mm sand
~
·
•
Plate 0.545-mm
Simons et ol. 0.45-mm
sand
sand

1.2 ~
.•kNF:(NF)q
·• Tison 0.04-1.75-mm
" Tsubaki et oi. I .03-mm sand
Tsubaki et cl. 1 .26-mm so'nd
.. Tsubaki et ol, 1.46 - mm sand
sand

O.
8
v,.~

•
\,~.. .. ... .
.;.. ~.

D ~ •
~ Tsubaki et ol. 2 .26-mm sand
• Dunes
o Antidunes

.,".~....,..., "il'f.:.
ft", , --- •-- . -.
-;---
- ,.l~ - f!..---- !L -" ........
...
O.4
.... -. l. _._
-r, -
O.0
o 2 4 6 Q 10 12 14 16 18
2n h/L
FII·5.2 Comparison of predicted and observed bedform regions. [Af ter KENNEDY
(1963).]

Results of the theoretical models are not very convincing ..

sothat we have to
rely again on empirical correlations. The first classification was given by
Liu (1957) who proposed Ux/W vs U~/v as a criterion for ripple formation.
This diagram was extended by Simons (1966) for other bed forms (see fig. 5.3).
 5.5

0.1 1.0 10 100

keynoldlo criteria ..!!_
rtjOOO 10.000 100.000

FII.5.3 Criteria for bedforms. [Aft" SIMONS., al. (/96la).) "

Simons et al 1963 gave a diagram based on grainsize and streampower (T .Ü), see
o
figure 5.4
LeQend
",·Ounes
• Plant bed
100 25
o Antidunes
Upper

...
20

.
....
'
,. .
I. •
...
Antidunes
~ Lower
';:'
Ounes regime
~
•
IS
.'
..
0

~ 1
•a.
0
10
E

..~
0

VI

0.1 No ripples when S

dso >0.60 mm

o 0.2 0.4 0.6 0.8 1.0 1.2

Medion foll diometer,mm OL_------~------~------~~------~
o 0.5 1.0 1.5 2.0
Fr
Fil. ~.4 Relation of bedforms to stream
'1,.5.5 Stabilitydiagram; tbe bedforms are indicatcd. [After
power and grain diameter. [Af ter SIMONS ENOELUND et al. (1966).]
et al. (/963a).]

The Froude number has to be considered as an important parameter as weIl.

This was done by several authors (Garde-Albertson, figure 5.6) and by
Engelund (fïgure 5.5). UXt is the value of ~ computed from Ü by assuming
the grainsize as bed roughness .and taking ks '"I 2.5 D50 and h = h t
 5.6

8.0 Q. I~
,
6.0
Q.
0

4.0

.. e
Q.
Q

f..J0
-
."

~
I
~
0
0
b
~
c:
~ Q.
Q.

4
•• •
0 0
b
~
- 2.0 •c:
lil q. .....:- _.. .

$
.l'
....
1+
')
Q.
0
ct
0
Q.
~ ~
:.
• • $
'J1.
~

IJ
0
;:1
P q.ö._
~~
-ç-
cf
~

·1-...... ~.
,
.J+
1.0
0
1*
~:t:(
op 0
~~
cf9_: jll
..-
L+ + ·ti~~
0.8
o 9 v ~
~t,.......+ tt-+ ~ ~-~r--
0.6 :t:( ~90 P

9
0
f
v

o
6
o
6
g
~ç
ÇO
0
O:ç
9
9-
I.,....
+ ++ ....
~ ~
1..
-4
~- ... _±••
~
t--.
~.
~,...
0.4

9
-v 0-'0
v

6
Q
Q
.. ~
4

,
IT

0.2
9 0
6
(r

I .. I lil

"
4- 1 c: 1 -
c:
~
:p
'0
9
) ~4- -
.~
.;;
c:
~
c:

,_~
'0
~
~
J
C>
C>
.. ...
0.10 ~ ~'O
~ / <r ..
0.08
...
C>

0.06
/
/
0 Ripples and dunes

• Transition
Antidunes
8
0.04

/ (SM Table 1)

0.02
0.08 0.1 0.2 0.4 0.6 0.8 1.0 2.0
CRITERIA FOR SEDIMENT REGIMES IN ALLUVIAL CHANNELS
. ü
acc. to Garde - Albertson (1959)
---- - Fr=--
..
FIGURE 5.6 ViP.
 5.7

It must be borne in mind that the transition in practical conditions

from one bed form to another may show an important phase lag with changes in
flow condition.
Raudkivi (1967) has measured the shear stress distribution on a dune
profile. The maximum shear stress on the upper part of the ripple had
about the same value as for a horizontal bed with the same mean velocity
and grain roughness. Behind the steep downstream face of the dune an
eddy develops. Around the reattachment point the flow 1S very turbulent
sothat particles are transported in bursts.

5.5 Alluvial roughness

The bed forms discussed in par. 5.3 all have their specific roughness.
For a flat bed without transport it can be assumed that the roughness is in
the order of the grainsize (for example D65 or D90). For flows over ripples
and dunes the total resistance consists of two parts:
the roughness of the grains and the form drag of the bed forms. The roughness
of a dune bed is much greater than that of a flat bed and the corresponding
friction factor is also much larger. Dunes generally give the maximum rough-
ness of a flow.
A flat bed with sediment transport (B.l) can have a friction factor
slightly different from that of a flat bed without transport. The presence
of antidunes does not appreciably change the magnitude of the effective
roughness of the bed if compared with a flat bed. If the waves break however,
the friction factor will be increasèd due to the energy dissipation in wave
breaking.
It cannot be expected in general that the friction factor of an allu-
via 1 channel is constant. Experiments have shown that the friction factor can
vary by a factor 5 or more. This is demonstrated in figure 5.8 and 5.9 where
changes in bed form give a great difference in bed roughness.
Figure 5.9 shows that the same value of T can occur for different
"0
values of U (take for example T
o
= 0.1 lbs/fta). Due to phase lags between
.
bed form (and roughness) and flow condition rivers very often exhibit
hysteresis effects in discharge-stage relations (not to be confused with
the hysteresis during a flood wave).
Prediction methods for the roughness of an alluvial stream generally
divide the total shear T or friction factor (C or À) into two parts, one
o
for the grain roughness (surface drag) denoted by T ' or C' or À' and one
o
for the form drag (T ", C" or À").
o
 5.8

10
0.16

/ 'XI(
/
oStondino woyes Of ontidunes
0.14
./ Y'lt \ "'" f versus u. !//
0.12
1/ fi'"-' )/
5.0
....
1 1..1 • ~".I~ /
.o~) ~ 0.10
I J "\ ~, 177
I~.II 1 \ ""') //
:: f versus ü-x.
1(, P
~A .....
~ 0.08
11

, i7'
0
j ) 1
0
e

I ~roÎ:z~ '-. 0.06

l
I
Î

Ijl.
1
I( "0 versus Ü 7
I(
j \
~Yx
If'
/

~-
I(
0.04

(dU7 V
._t-J 11· ~
-.
Lower rt9ime

)
[, ,,,~.",,, ,~,~
stondi"9 waveS,ontidunes)
0.02

0
__ 1
r:.
~
100 ....
fot-I( -.

2 3 s; fps 4 5
1.0
I 0 0.1 0.2 0.3 u., t ps 0.4 0.5
• j.
0/
I Flow resistanee due to bcdfonns. (A.fter RAUDICIVI (l967).J

0.5
1.0 5.0 10
Velocity. Ips

Fi,. 5.8 Relarlon of hydraulic radius 10 velocity for Rio Grande near
Bernalillo, (After NORDIN (1964).)

100
90
. le90nd f- 70
f-
\. -,
80 o Missowi Rivet near Fort Rondall. S.O.
70 ti MinourÎ Rive, near Pi'tr., S.O. f-
_. 50

. ,.
60 .. Missouri Ri"., neor Omaha, Nebr. f-
50 - ~ 6 Elkhorn Rive, n.or Woterloo,Nebr. f- 40
'\ <,
+ Bi9 5ÎOUll River ntor AIIoron.10_0
40
• Plotte River near Ash'ond. Hebr, f- kCurve 2
o ~ n ""
. 30
NiobrorQ River neG' But", Hebr.
f-
<,
iL 20
• 0

~ .._:;. ~.
v 5olinO$ Rlver at $on Lucas. Colif.
o Nocimiento Rive, near Junction. Calif .
Solinos River at Pa.o RIobIn
f-
f- 20
(o~
0

l~~ _., r-, . • '" e

\7~ ••
u; or--. ~ • (heovily veve'ated , c u \7 .\
(
ti
0
P'~0
No
~ ~+
Ü
u; 0
\7 •
r-, ~
~ J .~
,
cP ~ 0<;; ~o
10
9
8 . '10
• ~ ~ 'iJ" ~~ q
7
6
"!.
...... \7\7 ~
5 7
4

3 5
Curve ,../'r--.........
Q4MQ6 Q8 10 3 4 5 6 78910 20 30 40 50
1335
FIl· 5.10 Flow resisIanc:e duo to bedlOl'llll. (Afte- EINnmI n 11/. (19S1).)
2
0.50.7 2 20 30

FI,.5.11 Flow resistancedue to bedforms;curve I-river

data, curve2-ftume data. [Af ter SIMONS et al. (/966).]

t. "".
 5.9

By definition:

T = T ' + T " À = À' + À"

000

1 U2
À is defined by I = slope = À • 4R • 2g

Several procedures are given in literature.

1. Einstein-Barbarossa (1952)
E.B. divide the hydraulic radius R in two parts: R' and R", where
R' + R" = Rand T'
o /T".
0 R.'/R"
UX' is compu ted btk'
= y a 1ng k s = D 65 1n
.
the Chezy relation and ~35 is computed from:

~35 =
- X"
With the diagram given in figure 5.10 the value of U/U is found by trial
and error. For larger values of ~j5 (> 7) deviations are observed for river
data (see figure 5.11).

Procedure

a) If land hare given and U has to be known: guess h', compute 8'35' Ux' and
U and with fig. 5.10:U/UH". Compute hIt from UH" and h = h' + hIt. If h
is not correct, estimate a new value for h' and repeat untill h = h' + hIt.
Then use the last value of U.
b) If q and hare given and I or C has to be computed: estimate h' and
compute
,
and U/U ". From U
835
,
and U " a new value of h' can be obtained.
x x x
Repeat untill U ' remains constant. Then compute
K
,2 + U ,,2)!
U
M
= (U
x X '
land C.

2. Engelund and Hansen (1967)

E. and H. give an expression for f" of the form:
f" = aH2 /h. L H = dune height L = dune length h = water depth
where f = T/(~PU2) = 2g/C2 = iÀ
and introduce the dimensionless parameters:

™ = T/pg~D50 T
K'
= T'/pg~D
50
T' = pU
H' 2
Ü/U*' = 5.75 log 4.8h'/D50

Engelund concludes that TK is a function of TH' only (see figure 5.12).

 5. JO

I
8 h' T* TI
6 h T* - T I
It

4
XJ ,
XX ,,
I ,,,-
2 K

. />""
X '
Antidunes

T * 1.0
0.8
0.6
0.4
X x
x,
'l?/ ~
-<,.\C>
. ~
<",,0
,/
Standin9 waves and
Flat bed

,~ x X'

XI A
7!" Dunes
X2
T T

0.2 T 0.06 i+ 0.4 T

0.2 0.40.6 1.0 2 4 6

Tifr.l

Fi•• 5.12 Flow resistance [After ENGELUNO]

et al. (1967).]

Procedure

a) If land hare known and U has to be computed:

Compute T, TH and with fig. 5.12 T*'. This gives T', UX' and h'.
Then compute U.
b) If hand q are given and I or C has to be computed:
Guess h', compute U I, T ' and with fig. 5.12 T* and then T. Compute h'
from h'/h = T'/T
* *
; if'different from first value, repeat calculation
untill h' is constant. Then compute U , land C.
*
3. White, Paris, Bettess (1980)
t-lPBgive an empirical relation between:
U
=_.......:.:*~ , D
gr
= D • (~~/13
v
(6gD) I

and F = ---
U
*
n -U }I-n (a)
gr (6gD)~ {
.5.64 log (1~/D)
where the characteristic diameter is D = D35.

The relation is given by (see Fig. 5.13):

F - A
gr
="'"-....,.. = 1.0 - 0.7 6 (b)
Ffg- A

where A and nare functions of D

gr
n = 0 and A = 0.17 for Dgr ~ 60
 5. ] ]

n ].0 - 0.56 log Dgr }

] ~ D < 60
A = 0.23 D -l + 0.]4 gr
gr

(see a1so Fig. 6.8)

Procedure:

a) If hand I are given and U has to be computed:

Compute U ,D ,n, A and Ffg. Compute F using Fig. 5.]3 or formu1a (b).
* gr gr
Compute U from expression (a).
b) If U and hare given and I or C has to be known:
Estimate a va1ue for C, compute U and fo110w procedure a).
If the computed va1ue of U is not *correct, repeat the procedure using the
new va1ue of C.

4. Paris (]980)
Paris gives an empirica1 re1ation between cIc' and X = ~/~cr:

C
cr = ].0 - 0.47 log X + 0.]2 (log X)
2

c' = ]7.7 log (]Ohcr/D35)

h
cr
= X-1.h

The expression is va1id for X ?]. If X ~] then C = C'.

~cr is computed using the expressions given by White (~cr = A2) where ~cr is a
function of Dgr (computed from D35).

~cr = 0.029 for D gr ~ 60

D -l + 0.]4)2 for < 60

'l'cr= (0.23 gr
"1 ]~ D
gr

Procedure

a) If hand I are given and U has to be known:

Compute ~, ~cr' X, hcr' C' and C. Then compute U using the Chezy re1ation.
b) If Ü and hare given and I or C has to be known:
Estimate a va1ue for C, compute I and fo110w the procedure under a).
If C is not correct, repeat the procedure using the new va1ue of C.
 5.12

Comparison of various prediction methods

White c.s. and Paris have compared various methods with data (1432 flume tests
and 263 river data). The result was for a = C 1 Ic b
ca c. 0 serve
d

Method 0.7 < a < 1.4 0.89 < a < I. 11

Einstein 21% -
Engelund 83% 38%
White c.s. 89% 48%
Paris 89% 46%

The last three relations seem to give a reasonable result and can be used
to give a first estimate. If more accurate data are required, observations
in the field for the specific situation are necessary. Observations on
dune-bed rivers have shown that dune height and resistance increase with
increasing water temperature (Vanoni, 1975).
 5.13
I I I I

- -
~
--
- _Jo -

.!.
I
- .
............,._
...
;e
,..............
- ...

~~..
'" IL
~
r-- --.J

• -§ "'0
f-
f-
-
- -e•
~
o

-
f-
•
-
I
• ä
..... -., i
ö
e
f-
• • - ë
• •~
• •• Cl.
..... -

• ;;.
C
-•
~
'15
lil
o
- -N
e
o

• •
•
.. • ••
•

.\
2
- • • • -
-
- ~ -
• • -
f- • -on
•
f- • • -
• • 'M
r;...

-
,•
f-

•
•
I- -N

•
I
..I I I
...-
..
:!:
{
y-
~
I~}
• :
-I
~ 0
 5.14

5.6 Literature

F.M. Exner, 1925 Ober die Wechse1wirkung zwischen Wasser und

Geschiebe in F1üssen. Sitzber. Akad.
Wiss. Wien pt Ia Bd ]34.
H.A. Einstein, 1952 River channe1 roughness. Trans. ASCE,
N.L. Barbarossa
117, p. 1121-1132.
H.K. Liu, 1957 Mechanics of sediment - ripp1e format ion,
Proc. ASCE, 83 (HY2), p. 1-2]
R.J. Garde, 1959 Sand waves and regimes of flow in a11uvia1
M.L. A1bertson
channe1s. Proc. IAHR, Montrea1 ~, paper 28.
D.B. Simons, 1961 Forms of bed roughness in a11uvia1 channe1s.
E.V. Richardson
Proc. ASCE, 87 (HY3), p. 87-105.
D.B. Simons, 1963 Uniform water conveyance in a11uvia1
M.L. A1bertson
material. Trans. ASCE, 128 (1)
J.F. Kennedy, 1963 The mechanics of dunes and anti-dunes in
erodib1e bed channe1s. J. F1uid Mech.,
~ (4), p. 52]-544.
C.F. Nordin, 1964 Aspects of flow resistance and sediment
transport. Rio Grande near Berna1i110, New
Mexico. U.S. Geo1. Survey Water Supp1y,
paper 1498-H.
D.B. Simons, 1966 Resistance to flow in a11uvia1 channe1s.
E.V. Richardson
U.S. Geo1. Survey-Prof. paper 422-J.
F. Enge1und, 1967 A monograph on sediment transport in a11uvial
E. Hansen
stream. Teknisk For1ag Copenhagen.
A.J. Raudkivi, 1967 Loose boundary hydrau1ics. Pergamon, Oxford.
E.V. Richardson, 1967 Resistance to flow in sand channe1s. IAHR
D.B. Simons
Ft. Co11ins paper.
W.R. White, 1980 The frictiona1 charatteristics of a11uvia1
E. Paris,
R. Bettess streams: a new approach.
Proc. lnst •.Civ. Eng. ~, Sept. p. 737-750.
E. Paris, ]980 Un metodo per i1 ca1colo del coefficiente
di scabrezza in a1vei mobi1i.
l'Energia E1ettrica (no. ]2) p. 577-581.
E. P aris, ]9 80 New criteria for predicting the frictiona1
characteristics in a11uvia1 streams.
Proc. Int. Symp. River Eng., Belgrade,
paper DS.
 5. IS

5.7 Problems v s.f. = 0.7

5. I Given: depth h = 2 m grainsize D = ISO jlm (uniform)

1
T = 20° C chezy C 63 m2/S
mean vel. U
- 0.6 »t s
Question: What bedform can be expected according to:
a) Simons - Liu (Fig. 5.3)
b) Simons - stream power (Fig. 5.4)
c) Engelund (Fig. 5.5) (take h' = 0,5h)
d) Garde - Albertson (Fig. 5.6)

5.2 Same questions as in 5.1

for h 0.5 m D mm C 42 m~/s
-
U = 1.2 mis T

5.3 Given: depth h = 2 m, grainsize D = 0.5 mm (uniform)

Slope 1 = 2.10-4 T = 200 C

Question: Compute Ü using the methods of Engelund-Hansen, White c.s.

and Paris.
For the Engelund-Hansen method assume that dunes are present.

5.4 Same question for slope I

5.5 Given: depth h = 2 m, grainsize .D 0,2 mm (uniform)

U=0.7m/s T 200 C

Question: Compute C using the methods of Engelund-Hansen and Paris.

 6.1

6 BED MATERlAL TRANSPORT

Bed material transport can be divided in bed load and suspended load.
Both modes of transport have an influence on processes of eros ion and
deposition. Many relations between sediment transport and flow conditions
are based on the bed shear stress. It has been shown that the bed-shear
stress may be divided in a form drag and a grain roughness. It will be
clear that the form drag does not contribute to the transport but that
only the grain roughness will be of importance. Measurements of water depth
and slope give the total bed shear stress, so that most transport relations
require a reduction of the total bed shear stress to a value which is rele-
vant for the transport.
This reduction factor is cal led the ripple factor ~. Theoretically one
À' C 2
should expect: ~ = ~ = (~) .
Many authors use ~ as a closing term, however, so that various expressions
are given. This manipulation with the bed shear stress has led several
authors to use the mean velocity Ü instead of T as the important factor
o
for the sediment transport. The problem then is that the same value of Ü
in different water depths will give different sediment transport rates, so
that again some correction is necessary.

6.1 Bed load

Because several authors use some type of a physical model to predict

a sediment transport relation it is not surprising that most formulas may
be expressed as relations between dimensionless groups. The most common are
a group related to the transport:

~= SJ'[D3/2(g~)1/2]

S = transport in m3/ms transport = volume of grains

For convers ion to total volume, S has to be divided by (1 - E)
in which E = porosity. As a first estimate, take E = 0.4
~ = (ps - pw)/pw
D = grainsize
and a group related to the flow:
~ = Ux2 /~gD ~'=~.~ = effective value of ~
(the parameter used by Shie1ds for the initiation of motion).
Some of the relations given in literature are the f01lowing:
 /
6.2

1. Du Bois (1879)

Du Boys gave a simple model in which layers of sediment move relative

to each other. The number of layers was proportional to T IT
o cr
The resulting expression is of the form:
S = const. T (T - T )
cr o 0
Although the physical model is not very convincing, it has been found that
the fo nn of the relation can be used to describe eJIPeriments in areasonabie
way.

2. Kalinske (1947)

Kalinske assumed that grains are transported in a layer with thickness

D with an instantaneous grain velocity U equal to:
g
ug = b(U 0 - U cr )
Uo = instantaneous fluid velocity at grain level
U cr= critical fluid velocity to start grain movement.
For U o anormal distribution is assumed:

f (U ) = --,, 1
o am
__ expo

cr = r.m.s. value of velocity fluctuations.

Taking the number of grains per unit area p/(~/4D2) and using U then the
g
mean rate of partiele movement, by dry weight per unit width and time is:

p= 0.35
- 00
where U
g
= b J (U - U )f(U )dU
U 0 cr 0 0
b= 1.0
c
The resulting expression may be made dimensionless with the parameters
~ and ~ with the result:

~ = 2.5~~{ffl expo [- 2~2 (/0~12'_ 1)2] - (/0~12\_ 1) 2k erf l

[~ (/0~12 _ I)]} (see figure 6.1)

in which r = al Üo r= 0.17
Kalinske did not reduce the bed shear stress, so the relation is valid
for plane beds only, so ~ = ~'
 6.3

~ ,
10

7
- !.o. ·il.
• • ,. I._!, ...
. fr I.f:
.).-.; .Ij
._. j I!' i . i ·1··.. JII/I/ ·1/'
1 6
5
·1· . • r' I

- ~.
.
• ~ 1- •
.
" .. I ...
-,
I/Af . ~'II
:,··r/ lIL.- V' ~
i

4
3 • t ••••
, t··

; ,. 1

2 1:',..71,,'1/'" ... I
.::j 1111 111!ttll: .• ;'0'
: VI: tj; ~j.:": ~ : ; ;
I
; :! 0 ill; '; 1 '1:: i : 1 I:: ~ !::: I

, . I'
; -! • ~
,
•
•~
lt··
.. 0' ,;; L
. :: ..:-~ t
.;
.i
~-{. 1irl·,~ .. <t
~-+ t ~ "1 .•• t- '""'t~

. - . +: .f.- .! • - 'i' 0 .FT ... rl 1-+ h t ..H tl.!ti l+ • "'.

8
7
8
5
4
3

0.1 I t , •.
'" =_::t
iT ... t- - , .+
8 1+ . 1 -+ - ti t
7 . ..
6
"~

ft
J
5
Kalinske 4

_____ Meyèr-Petèr 3
1-- -

Müller 2
-------- Einstein
....................... FrUlink 0.01

_._._ snlnobara 8
7
Tsubaki 6
5
_"--".- Rottner 4
3
-+-+- Egu;asaroff
_._._._._._. Garde
Albertson

i 0 I I'
0.01 2 3 4 5 6 7890.1 2 3 4 587891

--~ ..... tjJ'

COMPARISONOF BED-LOAD TRANSPORTEqUATIONS
FIGURE 6.1
 6.4

3. Meyer - Peter and Müller

M.P.M. have performed a large number of experiments in a wide flume

with coarse sands. The resulting empirical expression may be written in
~ and ~' units as:

~ = (4~' - 0.188)3/2 (figure 6.1).

By comparison of results with flat beds and dune beds the ripple factor
is found:

~ = (C/C,)3/2 (theoretical exponent 2). Cl = 18 log ~~~

For a mixture M.P.M. take:

Dm = D = E p.D/Ep as the relevant parameter for the value of ~' and ~

D DgO for the grain roughness.

4. Einstein (1950)

Einstein gave a complicated statistical description of the grain

transport process in which the exchange probability of a grain is related
to flow conditions. The resulting expression is given in figure 6.1 in a
graphical form.
For the determination of the ripple factor ~ a graphical procedure is
.
g1ven by E' . x)He use d D
1nste1n. = h
D35 as tere 1evant parameter f or the
transport and D = D65 for the roughness. The correlation is not valid for
large rates of transport because there the transport varies with the first
power of velocity ij only.

X)See par. 5.5

The relations given were for bed-lead. In most conditions a predominant

contribution of suspended load will be present. The final accuracy of the
bed-material discharge will depend therefore mostlyon the accuracy of the
suspended load determination.

6.2 Suspéndéd lead

Suspended load can be determined from measurements of U(z)_and C (z)

and integration of:

S = JhC(z) U(z) dz
o
 6.5

.v.c(z)

In most cases estimates based on theoretical expressions will be

necessary. The basic equation describing the concentration distribution
in uniform steady flow is:

W .C + E:
a C
o
S ·äZ

The first term W • C (W fall velocity; C = volume concentration of

sediments) represents the settling tendency of the flow. The second term
represents the diffusive action of the turbulence. E: is the turbulent
s
diffusion coëfficient. An explanation for this term is the following. Water
packets moving upward carry a larger amount of grains than packets moving
downward because there is a concentration gradient. Although there 1S

no net transport of water there will be a net vertical transport due to

this exchange of water packets, which will be proportional to the local
value of the concentration gradient.
If it is assumed that the diffusion coëfficient for sediment is
equal to the coëfficient to the exchange of momentum, then:

E:
s
= E:
m
= K.U* z (1 - z/h)

The resulting equation may be integrated and gives:

C(z)_ ( h - z a )a
ë('ä)- z h-a

with a = W/KUx• a is a reference level where C C(a).

For a graphical presentation see figure 6.2
 6.6

• u
c
.,
~ 0.8
-'

0.9
'lr-

I'-.......
- ~
I
t---.<>
I'-
t'I> ~
- N ...... K
...
~9 ~
l\
'I>_
-,
\
~ f",_~ ~ ~ [\ [\ .\. kt ~ 0.34
~ 0.7
_!!
., Cl: 146'/
"< Cl ~1.03 '\ ï\ E\ \~
~ 0.6 ~
a. :
" .89
K~ I.---- ~.OrO.52
" -. < ~:0.6~ \
~I~~0.5
N~~

I
':\ \<>1\ \ 1\ \
f\.~ 9\
~ 0.4 Curve Symbol z
c:
c
.~ 0.3
A
B " lA~1
1.03
'0,r-, ~~ ~
-c "
o
§ 0.2
C
0
E 9• ' 11
-e
0.66 r>-", ~~ ~ r\ \
f>\ K~
c- 0.52 r-,q ."'-,
u
o F 0.34 ~d
~·0.1 ~
~~0.05
r-,~ ~ ~
~O' 0.002 0.004 0.0' 0.02 0.04 0.' 0.2 0.4 10
• "":__Bottom
Relotive concentration f-a D ..depth
Fig. 6.2 Distribution of suspended sediment; comparison of experimental data with
Eq. (8.35). [Af ter VANONI (J946).]

From this figure and the analytical expression the fo11owing rough criteria
may be given:
x
W/KU UX/W description

1.6 1.5 some suspension

0.8 3 concentration at surface > 0
0.25 10 fu11y deve10ped suspension
0.06 40 a1most uniform concentration

The last criterion shows that particles < 50 ~m (W < 0.2 cm/s) are uniform1y
distributed for UX > 8 cm/s or Ü > 1 - 1.5 mis.
A1though the basic equation is very simpie, some critica1 remarks have
to be made:
1. The term W.C. shou1d be (1 - C).C.W to account for the presence of the
particles (see Hunt 1954). This correct ion is not important for C~ 1.
2. The fal1 velocity is changed by the presence of other particles (see
chapter 3) and by the turbulent movements of the water. Symmetric vertical
velocity fluctuations give a-symmetrical drag force for non-Stokes
particles. Therefore, although the mean value of the vertical velocity
is zero, there wil1 be a resultant vertical force which will reduce the
settling velocity.
 6.7

3. The expression for E

s
gives E
s
=0 for z= 0 or ac/az 00 at z = 0
which is not very realo
4. The value of C = C(a) is not given. Several assumptions are made in the
literature. Einstein (1950) divides the computed bed-load by a layer
with thickness 2 D and by the velocity in this layer. (11.6 U ')
H
The value of C(a) is one of the problems to be solved in sediment
transport.
5. The velocity distribution is influenced by the presence of particles.
The weight of the particles suppresses the vertical velocity fluctuations
and gives a decrease in the momentum diffusion coëfficient. This is
similar to a decrease in the value of K.ln fact several expressions have
been given in which K decreases with the power to keep the sediment in
suspension: C.W.~/U.I (see figure 6.3).
Velocity profiles become less "full" by this effect. Care should be
taken in the application of this correlation because the determination
of K from velocity profiles or concentration profiles is not very
accurate. For literature see Einstein and Ning Chien (1954) and Ippen
(1971)•

<, ....
.....

IX
0
dl
0.10

I':'
0
_,
""b..[
~ 0

'":\
CwA 0
~ + )(

UI ~ 0

):(

0-t~ 0
0
o X Xo ~~~
0.01
0 .
x
• V"
-\
0
It .. ' •
• tf \
~ , o. •
x C.
x 0
x xx
\: ~

0.001
Flume studies
•
x
'b

x
\,
x x,o
0.10mm}
0 0.16 mm Ismoil
0

• Vononi
x
C. Kolinskeond Hsio
)(0
•
1.55 mm
*
JI
+
0.94 mm
0.27 mm
} Chien
I
x
x )(

0.0001 r Rinr mj!osurs:mj!nts )(

x x
0 t952 Missouri River
)( Atchofoloyo River
I I x
0.15 0.20 0.25 0.30 0.35 0.40

Fil. 6.3 Effect ofsuspended laad on the k value. [After EINSTEINet al.
(1954).]
 6.8

6. The assumption E
s
= Em has also some objections. It is not necessary
that the diffusion of particles is equal to that of momentum.
Measurements by Coleman (1970) show indeed that E -values derived
s
from concentration profiles give some differences with values of
E obtained from:
s
E = E = KUxz(l - z/h).
s m
It is areasonabie assumption, however. Differences between E
and E
s m
are generally put in K which is of ten used as a c10sing factor. If
U(z) and C(z) are known, integration wi11 give the suspended load. The
integration cannot be performed ana1ytica11y. Graphs are presented by
Einstein (see Graf 1971, p. 189-195).

6.3 Tota1 bed-materia1 load

The tota1 bed-materia1 load of a stream can be determined by adding

the bed-1oad and the suspended load. This is done in the Einstein (1950)
procedure. This procedure was modified by Co1by (1955, 1961). Also
Toffa1etti (1969) gives a procedure which is especia11y adapted for
computer programming.
Besides these "adding" procedures severa1 direct empirica1 re1ations
are proposed in 1iterature:
1. Shinohara and Tsubaki (1959) gave an empirica1 relation:
~ = 25(~~1.3 (~' - 0.038) (see figure 6.1)
The corresponding ripple factor ~ = (CiC').
2. Garde and Albertson (1961) gave a graphica1 relation of ~II~
with u/~ as the third variab1e. The resu1ting ~ - ~ relation is
a1most identica1 with Shinohara (figure 6.1).
3. Colby (1964) has given a graphical re1ation between totalload, mean
velocity U, flow depth and grain-size with correct ion factors for
temperature and silt content (see figures 6.4 and 6.5).
4. Engélund and Hansen (1967) gave an empirical relation of the form:
~ = 0.lf-l~2.5 with f = T/(~pU2)= 2g/c2
The formu1a is based on measurements with D50 < 1 mm ànd gave good
results in comparison with sediment transport measurements in rivers.
At all values of W, the sediment rate increases with the fifth power
of the velocity.
 6.9

5. Ackers and White (1973) define the parameters:

F Uxn . (U* I) l-n UH

I -
U
gr
(~gD)J 5.64 log(IO h/D)

D D. (~g)1/3 (dimensionless grain size)

gr \)2

S *
G
gr =
UD
. (~ )n (transport parameters)
U
The relation between the transport parameter G and the sediment mobility
gr
number F is given as:
gr
F
G C ( gr _ I)m
gr A

in which C, A, mand nare functions of the dimensionless grain S1ze

D (see figure 6.8).
gr
I

For coarse materials (D > 60) n = 0 and U* = U* sothat the parameters F

gr gr
and Gare
gr reduced to a more simple form.

For a modest range of particle sizes (D84/D16 < 5) Ackers and lfhite suggest to
take D = D 35. For a wider gradation a fraction by fraction computation is
suggested, using a corrected value for A:

Di -0.2
AI = A • (-D ) 1n which
50
Di = average size of the fraction (Ackers and White, 1980)

NOTE
It must be noted that due to the strong variation of sediment transport with
velocity, predictions of total sediment load will not be very accurate.
Differences of a factor 10 between various formulas or between computations and
measurements are no exception (see figure 6.6 and 6.7).

6.4 Comparison of relations

White, Milli and Crabbe (1975) have made a comparison of 8 of the most
widely used transport relations (a.o. Meyer-Peter Müller, Einstein, Engelund and
Hansen and Ackers and White) with 840 flume data and 260 field experiments in
natural water courses. If the percentage of all data with a ratio R of calcu-
lated to observed transport in the range
~ < R < 2 is taken then the following result is obtained:
Ackers and White 68%
Engelund and Hansen 63%
Einstein 46%
 6.10
It is not surprising that Ackers and White give relatively good results
in view of the large number of tuning parameters (C, A, mand nare arl
functions of grain size). It is surprising however that the far more simple
formula of Engelund and Hansen gives such a good result.
Application of formulas remains a matter of experience. For each
situation a comparison with field measurements and an adjustment of the
formulas remains necessary for reliable results.

10.000
I I
Of" COItItt:OTlDllI'.lGTO. -..POIt
.. .....-, .- .. e
EXPLANA.TION
_TCItTEil,."."","

e- ..
0
......................
---
...- - •
4 ii
--- - • é: I UI\~\d ~
%
b D<"" D<"" -- 0(""" •
I uw l\
DO""
Ijlll I~~\
l~ ~\
.
0.1" 1.0ft 100"
i.ooo '\" 4. 0 • • 4
1000
:!;
._
0
+ï.f-- - • I1 I ~\'-
"-
Cl:
~
g
. 4
• 3 4"',
iyr-- r- -
'/,f
- • I r
11I

~
t\-
~ l\

Cl:
~ 1IIIl I\'i- :\ f~~ ~~---

:: .
., 100 _. 111 11 I Il
---
=-
100 -- I I -- --- ,
Z •F=
- I -I-
• - -11 .
1----1-
_.
-- -- .\
fJ I I I
ö ~
'/~ ril ill I f IJ \I}
15

-1- ,.- _.-

~I I
10
I -I .-
\1\ ~I\ \;- \
--
i~
_I -/1
I
4
: ~
J 1 UI I • I
'" :t !I/ 1 I
·
ij
-
-,- l-

----
P

'rt. -~!i-/ 0

11 11
/.,1 n 4

0.'
mi /1"
;11' OJ
• • I 4 •• 10 I I 4 •• I I I 4 •• 10 I I 4 ••
10 .0
IIIEAN VELOCITY,IN FT PER SEC

FIG. 6.4 -COLBY'S (2-H.14) RELATIONSHIP FOR DISCHARGEOF BANDSIN

TERMS OF MEANVELOCITY FOR 6 MEDIANSIZES OF BED BANDS,4 DEPTHS
OF FLOW, ANDWATER TEMPERATURE OF 60' F

2.000
I I/~+'/
2_000
.....,..,: f7I
4!/v:
o.}>I.1TTI COL8'
N
• OBSERVED
.. I ~C/N
1.000
0.800
1/ 1.000
csoo /1/ a •
0_600 / ~P'~I I.#:J
OAIOO JI;+-)} fL
OAOO
/
i! /} • t '" ~r; I/tri OAOO
j ,:JU 1/;. V
,
::
I~ • "') L iIY
~
-
/ /~71 /
v: ".-jJ 'i
i 0.200
/, liJ / I~
10.200
I 111, fY IJ/'/{ ~
I I '" /. :lr/'l .s:
~
IW
~
0.100
0.080
().()6()
~
11', I
'tiVI §iH'
..;
/ !0.100
~ 0.080
_l!! 0.060 /
dj
~
/ ,., VI
'}
/
Q040 L~"Jl~ ~ ~V IQ040 L -J' .!f.I 1'1', IJ /
~
._
Z
lol .v"~ II
~ r--~
,-+-~
I
'--~
~'1:;
/ lIIFfEII PETEII-IllULLI •• _-
1 11 11
f--

,
Q020
.....-N
I If-r-VÎ vr- COLORADORIVER
Q020

0.010
~r-.
~!/ IA'
• OBSERVED

0.010 IIOIIRARA RIVER

O_OOS IJ, 1;~f-- AT llIYLOR" FEIIIIY o.ooe
1#1 S.QOOI29 ft/ft
0.006 f--

0.0<>4
r-
-11, S '0.000211 111ft
d,.Q.32.0 ""ft
o.ooe
o.oD4 J !LIL. d,.O.21!...",
0",-1.6
'!!II, 0;'1.44
f-- T.60·F
T.6O'Ir
0.002 7V 0.002
0.4
~
o.e o.a I 2 4 6 810 20 40
2 4 6 8 10 20 40 60 80 100 200
WATER DlSCHARGf;. cu, It, po, _.,.. It. WATER DtSCHARGE, eu. ft. ,., He. per 't.

FIG. 6.6 _ .-SEDIMENT DISCHARGE AS FUNCTION OF WATER DISCHARGE FOR FIG.6.7 -SEDIMENT DISCHARGE AS FUNCTION OF WATER DISCHARGEFOR
COLORADO RIVER AT TAYLOR'S FERRY OBTAINED FROM OBSERVATIONS AND NIOBRARA RIVER NEAR CODY, NEB. OBTAINED FROM OBSERVATIONSANDCAL-
C,\i.CUL<\ nONS BY SEVERAL FORMULAS CULATIONS BY SEVERAL FORMULAS
 6.11

' ...
, ..... I
I
.....
~

- I

\~\'
I
r-- ~ ... E I ...
I

!--
...
~
~ ... r-- ~
... I
...
~ I I

j ...5! !
~
~
. 10
.... r--
~ ~
..... ..,
~ \
-
Do. 0

I
Q
:t «111. ~
r-- ó
-
I-- ...~
•
_;d,~~~ ... .dil j c
c
~IJ

I:
~ I~ o
R)Q
1.. ....
5!J- ,.. b~
oe. :/
hDI'J
<l
Üc
.t
,
")i- ~~ 0
... ...
:;;
...
VI
~
VI
i- .
~/
... __..A ~ -, ~
a-
8.
VI
....
Y' ~ .F
lI"\..
~
...;.
• - ..... •c
.. I.;":-;; ~
'V
~ ~~r oL' • .~
~~ i : }'i"! !
I............ I-
" ... e" 'i J.l~ t~ ...
~c:I
== v)'" ~ ~
~ 0 o ...; G
... ... E
è i
..
Q
~ ~
Q
Ó
Q
....
~ 1
Cl)
I
UI • JN3NOdU

"!G
~
I
•
..
I
I
..... C
G
o
- c
I

I
.. ~ U
...:t
I
I .-
C

- ...~ : ... I

-s
I
!!! I
....
-I
ct I
... U
i4:
I

I
~
,.:.,
l I
... ... 8 I
.......
I

-
-~
0 ~ I
!
!
....
~
I
I Q~
3-
2Q~
~...
- ~i °
I 0
....~
,.
10- ... :tI
Q'...." • J q<l
~ / ...
I
:!
..,
11\
0:sl 'J)
"" r ~ ...Q'" <l e- 'î.
~
I
... ~
ó
8
I
~ ... 0 ° 10
'0
.....
... ~
...:
c": .. o~Po ~ ~~D ""~
5!1
<, .A
~

°
h
<'J
..
~

oe
0 ~.n.
Ca ,a: ::>

è;~ ~
° /
v"
.. 6
..
~o ,/ 'I' t '" /
~
1'\
... [IJ'
~
...
~
(~
... -....
V .....! ~
c V -
~ ~ ~ ~ ...
~ - Q ~
Ó 0 0 Q Ó ti ti ó 0 ti
u •#/3J3""YIJYd NO/J/SNY#/J I I

fig.S.S'
 6.12

6.4 Literature

M.P. Duboys, 1879 Le Rhone et les Rivières a lit affoui11able.

Mem. Doc. Ann. Ponts et Chaussees, ser. 5,
vol. 18
A.A. Kalinske, 1947 Movement of sediment as bed-load in rivers.
Trans. Am. Geoph. Un., 28 (4), p. 615 - 620
E. Meyer-Peter, 1948 Formulas for bed-load transport.
R. Müller
Int. Ass. Hydr. Res. 2nd Meeting, Stockholm
H.A. Einstein, 1950 The bed-load function for sediment trans-
portation in open channel flows.
u.S. Dept. of Agric. Soil Cons. Service,
TB no. 1026
I
J.N. Hunt, 1954 The turbulent transport of suspended sediments
in open channels.
Proc. Roy. Soc. London 224 A, p. 322-335
H.A. Einstein, 1954 Second approximation to the solution of the
Ning Chien
suspended load theory.
Unif. of Ca1if. lnst. Eng. Res., Publ. no. 3
H.A. Einstein, 1955 Effects of heavy sediment concentration near
Ning Chien
the bed on velocity and sediment distribution.
Unif. Calif. lnst. Eng. Res., Publ. no. 8
B.R. Co1by, 1955 Computations of total sediment discharge.
C.H. Hembree
u.S. Geol. Survey. Water Supply Paper 1357.
K. Shinohara, 1959 On the characteristics of sand waves formed
T. Tsubaki
upon the beds of the open channels and rivers.
Rep. Res. lnst. App1. Mech., Kuyushu Univ.
(25), p , 15-45
B.R. Colby, 1961 Simplified method for computing total sediment
D.W. Hubbell
discharge with the modified Einstein procedure.
u.S. Geol. Survey Water Supp1y paper 1593.
R.J. Garde, 1961 Bed-load transportation in a1luvia1 channe1s.
M.L. A1bertson
La Houil1e Blanche ~ (3), p. 274-286
ASCE, 1963 Suspension of sediment. Proc. ASCE 89 HY5, p.
p. 45-76.
B.R. Colby, 1964 Practical computations of bed-materia1
discharge. Proc. ASCE 90 HY2, p. 217.
 6.13

F. Enge1und, 1967 A monograph on sediment transport.

E. Hansen
Teknisk For1ag Copenhagen.
F.B. Toffa1etti, 1969 Definitive computations on sand discharge in
rivers. Proc. ASCE 95 HYl, p. 225-248
N.L. Co1eman, 1970 F1ume studies of the sediment transfer
coefficient. lo1aterRe s, R ~ (3), p , 801-809.
W.H.Graf, 1971 Hydrau1ics of sediment transport.
McGraw HilI, New York.
AS CE , 1971 Fundamentals of sediment transportation.
Proc. ASCE 2l, HYI2, p. 1979-2023.
ASCE, 1971 Sediment discharge formu1as. Proc. ASCE,
2l HY4, p. 523-567.
A.T. Ippen, 1971 A new look at sedimentation in turbulent
streams. Journ. of the Boston Soc. of Civ.
Eng. 58 (3), p , 131-163.
P. Ackers, 1973 Sediment transport: New approach and ana1ysis.
W.R. White
Proc. ASCE 99 HY1l, p. 2041-2060
W.R. White, 1975, Sediment transport theories: a review.
H. Milli,
A.D. Crabbe Proc. Inst. Civ. Eng. (London), part 2,
June, ~, p. 265 - 292.
P. Acke rs , 1980, Bed-materia1 transport: A theory for tota1
W.R. White
load and its verification.
Int. Symp. on River Sedimentation, Beijing,
Hareh. Paper BlO.
 6.14

6.5 Problems

6.1 Given: A wide river with h = 3 m, U I .2 m/s, I 1/6000

D90 3 nnn.
Question: Compute the bed load with the Heyer-Peter-MÜl1er method.

6.2 Given: A wide river with depth h = 2 m, width B = 80 m, U = 1.1 mis,

-
-4 ,Dm
I = 8.10 = 0.6 mm , D90 = 1.5 nnn,' E = 0.4.
Question: Compute the annual bulk transport using the M.P.M. method.

6.3 Given: A wide river has the following characteristics:

depth h = 3.2 m
Slope 1=0.5.10
-4
bed mat.: D = 0.5 nnn D 90 = I nnn
m
Temp.: T = 20° C
Questions: I) Determine the critical shear stress T of the bed material
cr
and the bottom shear stress T . Is there transport?
o
2) Which type of bot tom configuration is present according to
Simons-Liu (Fig. 5.3)? Use D to obtain the fall velocity.
m
3) If the mean velocity U 0.66 mis, what is the bed
roughness k ?
s
Is the bed hydraulically smooth or rough?
4) Compute the ripple factor ~ according to M.P.M.
5) Compute T' = ~T and the bed load/m' according to M.P.M.
o
6) Will there be transport in suspension?

6.4 Given: In a wide river a trench is made for

a pipe line crossing. The depth of the
trench is 3 m. Because it takes some
time to lay the pipe and the river
transports bed load)some storage has to
be provided.
River data: U = 1.2 mis h = 4 m C
Dm = 1 nnn D90 = 2 nnn E = 0.4.
The relation of M.P.M. is valid.
Question: How large should L be to provide sufficient storage for 2 days?

6.5 Given: A wide open channel, h = 3 m, U = 1.2 mis, I = 10-4•

Transported material D = 150 ~m (uniform)
The concentration at z = 0.5 m is 250 mg/l.
Question: Compute the concentration at z = 0.25 mand z = 2 m.
 6.15

6.6 Given: A wide river with h = 2 m 1=1.5.10

-4 -
U
sediment uniform D = 0.2 mm E = 0.4.
Question: What is the bulk transport/m' .day using the Engelund-Hansen
method.

6.7 Same question for:

h = 3 m I = 10-4 U 0.8 mis D 0.15 mm (uniform)

6.8 Use the data of 6.6 and compute the total-Ioad with the method of
Ackers-White (as bulk load/m.day).

6.9 Use the data of 6.1 and the method of Ackers-White to compute the
transport. D35 = 1.5 mmo

6.10 Given: Sediment size D = ISO ~m (uniform) and shear velocity U* = 0.05 mis.
Question: Compute fall velocity (Fig. 3.4), critical shear stress
(Fig. 4.2) bedform according to Simons (Fig. 5.3) and the
degree of suspension (tabie page 6.6).

6.11 Same question for D = 2 mmo and U* = 0.05 mis.

 7.1

7. STABLE CHANNELS

7.1 Introduction

For the design of stabie channels two approaches can be distinguished:

1. The tractive force theory (Lane)
2. The regime theory (originating in India. Deve10ped by Kennedy, Lind1ey,
Lacey, Ing1is,Blench).
Both methods are used to design stabie channels. A stab1e channe1 in a1luvia1
material is one in which scour of banks and changes in alignment do not
occur. Deposition on or scour of the bed 1S not objectionable in general,
providmthere is equilibrium over a long period.
The stability of a channel depends on the properties of the excavated
material (grain-size, cohesion),of the flow (discharge, si1t content, trans-
ported material) and the design variables such as profile, shape, slope.
Sediments introduced in the channel must be conveyed in view of the definition
given above.

7.2 Tractive force theory

This approach is specially suited if the flow transports very litt1e

or no sediment. The design is then based on a limiting velocity or critica1
shear stress of the bed material.
For uniform cohesionless material, Shields'graph may be used to compute
Tcr • In practice materials wil1 have a wide gradation and wi1l have some
cohesion due to the silt content. For these materiais, Lane's (1953) design
curves are recommended (see figure 7.1).
It must be noted, however, that the large values of T as compared with
cr
Shie1ds values are due to the fact that T -values are based on actual
cr
channel roughness, inc1uding irregularities, bedforms, whereas Shields'
graph is based on a flat bed. When there is some bed load, the problem is
more complicated and calculations shou1d be made to check the transport
capacity of the channe1s.
Bank stability wi1l depend on the characteristics of the bed material
and the side slopes of the channel. From experiments and calculations it
appeared that for trapezoidal channels with side slopes 1:1 to 1:2 the
following values can be given for the shear stress:
T , horizontal part of the profile ~ pghI
o
T , side slopes ~ !pghI
o
 7.2

For non-cohesive materials the reduction of T due to the side slope was
cr
given in chapter 4:

T (S)
= cos S
__;_cr~-:-
I 1 _ (tgS) 2 (see figure 7.2)
T (0) tgt
cr
20.9 4
1 I I1 I I 1 I I 1 1 I"" I 3
Recommendedvolue for conols Line representing relotions Of~
10.0. with high content of fine \. troctive farces Ibs/U2, 0..5 2
80. Sjdiieît lin ~helwoter I \ diom.,in.. Troctive force ~
kq/m2, diom.,cm I
60.
40. I-
Fortier 8 Scobey - Recommended
for conols in fine sond with
water contoining colloids
rk-
(opproximotely )
I I
l.l U.S.5R.conols
~ i2
i)I~
NK~ ~
1.0.

N
E
30.
I 1.111 IJ.
U.S.S.R.-Conols w,th _
Ä with 0..1%
colloids in~
water
~V 0..6

z....
_u
20.
2.5% colloids in wote~\
I I II I I 'INIK .&
0..4
0..3
..
0 Schoklitsch- recommended 1\ 17'/ Recornmended
10. f--l- for conpis in scnd-; \ volve for conols t- 0..2
ai 8 "] 1-:10"[7"" / with low content t-
u
~
..... 6 I\~UERNBERG ) 1~ ~~ ~ i'~1:P" V of fine sedimentt--
in the woter 0..1
Cl>
>
KULTURAMT.NK ~ ~ ~ i-" . cu
+= 4 , Recommendedvolues for _ >
u '> /N~fZ v- conols in coorse noncohesive
6
.i'
~2 0..0. u

0
3
NK 10-""
A.VV r'
moteriol size 25 % ond ,-
lorger 1 1 +-
0..0.4 0
e
.~ 2 ~ ~ ecommend,:~ volues for conols
·cu ./ v- .... I---" wlth cleor water I I DO. 3 ~
/"
i-" ~ Stroub volues of critica I troctive force ~
X 0..0.2 u
10. r'
.80.
,60.
..- ..-
~ z- ~
r-, ./
~
U.S.S.R.-co:nolsI r: c~o~woter
Fortier 8 Scobey-Recommended for conols
./
V in fine sond ond cleor water
DO.
.40.
.30.
V I®0.6
.20. 0..0.0.4
0..0.0.3
.10
0.1 0..2 0..3 ~ 0..6 0..81.0. 23456810. 20. 30. 40. 60. 80 10.0.
Meon diometer,mm

Fil. 7.1 Critical shear stress as function of grain diameter. [Af ter LANE (1953).1

Assuming .d /Th·or1.zonta1 = ~ and t = 30 - 400 it can be seen that a

S1.e T

side slope of 1:2 to 1:3 is necessary. For practical va1ues of t see

figure 7.3
A theoretical stabie profile for which at all points the same critical
conditions occur is found in the following way. Assume that the local value
of the shear stress T(y) is proportional to the local water depth h(y) and
-1
acts on a 1ength of 1/cos8{y). Then T(y) T
max
.h(y).h
max
cosB(y) =
in which h is the maximum depth with correspondi~g T •
max . max
With the reduction formula for T (B) the following theoretica1 S1.nus-
cr
profile results:
h(y)/h
max
= sin(y/h
max
.tgt). y = horizontal coordinate
 7.3

,110:,-
40
-r--- ... ~

- --..,
- ~
~
I I
.... fO 0_ fIU&'_

• - &".L~0' .,
• - UI"
• - COl.•

0' 110,1
'lO~1
I
V ,_ t..·t
,oa, Lf_.
TaiiT'i'"""

L.
.. ~ ~
~ ~

~"'~
I~:.-
-f ..__
:.
~
....
3:'1 r---
r-
: ••. 1- 10

'!:"':--
..
0
~I;I-
co
~
•
r---~ ~
I! .. "'- ~
..- '""-_ <, ......
1.14:.-::
r---
N
-
:.-:c
j....
-
lol

0
.. ~~ r-. \ ~
'"~'.1-
...
0

'"
lP

-r--r---... r-,
~ ~ ~
~ .:1- •• i"'-o..
-........
lil

- ~
10 ~ ~
<,
K· CRITICAL TRACTIVE FORCEON SIDES IN FRACTION
~
• OF VALUE FOR LEVEL BOTTOM
I
I FOII NON-COHES'VE MATERlAL
I I I 1 '\ t
1'1' Ot 0j' Ol· _ï o.t Dj'
ï 0j'
.0 !!!
1
- K = Critica! tractive force on aide. in fradion o( value (or
level hottom (or Don-cobe.ive material.
Acc. to Lau ("55) FIG. 7.2

Mean Diameter in MiJl i meters

0.3 1.0 20 4.0 100 20,0 4060 IC)O 2:'>0 400 ~
I- en., --I-t- ~.
!
45 .,
Ot
0
•
I0
40 Go .64t ~.. -\
• .- .

36
-.,
a::
0

Ot
~t~
(j<\).
-;J-t
",Cl. fl.t\
~\)\G(
....~I._I-
.:.;..c: . G
~ "t<"t
~o\)
~t6

C - _o._ .. ___
ct ~

o Round.d
I-
• Roundtd a Angular
• Angular
25

20
Mean Diameter, Inches
0.01 0.1 1.0 10 20
ANGLE OF RE POSE OF NON-COHESIVE MATERlAL
Ace.. to SIIrIoA. aAd Albertson (1110 I FIG.7.3
 7.4

The width of the channe1 is then B

max
= n.h
Itg~ but may be extended of
course with a horizontal section. It is c1ear that the sinusioda1 profile
is difficu1t to construct and wi11 be approximated by a trapezoida1 profile
in practice.
Lane (1953) gives some reduction factors to account for the sinuosity
of channels.

Channe1 type T
cr
IT
cr
straight U
cr
Iu cr straight
straight 1.00 1.00
slight1y sinuous 0.90 0.95
moderate1y sinuous 0.75 0.87
very sinuous 0.60 0.78

A1so 10ca1 effects of contractions, bridges etc. should be considered.

In app1ying Shie1ds' graph for coarse material (~ z 0.06) for the
cr
stabi1ity of stones on revetments and banks care shou1d be taken with the
criterion (general movements). For a safe design without movement of the
stones a va1ue ~
cr
= 0.03 is recommended.

7.3 Regime theory

With the tractive force theory designs of channe1s can be made. Another
approach to the prob1em is to study succesful1 al1uvial channe1s. Numerous
studies of man-made and natura1 a11uvial channe1s have given empirica1
re1ations between depth, width, velocity, discharge, sediment transport and
material characteristics. These techniques are referred to as "regime theory".
Usua11y three equations are presented: (1) a flow formu1a which gives the
required slope (2,3) formu1as for channe1 depth and width. Regime theory
originated in India where extensive cana1 systems were built. One of the
disadvantages of the regime theory is that resu1ts are re1ated to a specific
area, so that application to other areas can give errors.
"Regime" can be defined as a situation in which a channe1 will not
change on a long-term average. Short term changes wi11 occur with changes in
discharge or sediment transport.
Important contributions were given by Kennedy, Lind1ey, Lacey and B1ench
(1957). Some of the resu1ts of B1ench are given here. B1ench gave three
equations:
 7.5

(1) Fb U2/h Fb bedfactor (ft.s-units)

(2) F U3 • h/A F = side factor A = area of cross section

s s
F~/6 F!/12 ,}/4
(3) I =
3.63Ql!6.g(l + c/2330)
I = slope
v = viscosity of water-sand mixture
c = sediment concentration in p.p.m. by weight

From these relations the following equations are derived:

B = A/h = /(Fb/Fs).Q

h = f(F »,2)Q ., h = r q2 /Fb q Q/B

-
U VFb·Fs·Q

Blench suggests: Fb = 1.9ID(1+0.012 c)

with D in mm, sand range only and F = 0.1, 0.2, 0.3 for loam with very
s
slight, medium and high cohesion. For practical applications tests in
similar channels seem necessary.

Simons and Albertson (1960) have analysed a large number of Indian and
American canals. The results presented are valid for sediment concentrations
< 500 p.p.m. and grain sizes 0.1 < D < 7.5 mmo From figure 7.4 to 7.8 width
and depth can be selected. Curves E should be used for channels with high
sediment load only. From the graphs A = P.R can be computed and also
U= Q/A. Values of depth, average width and top width can be adjusted as
required to maintain these values of hydraulic radius and wetted perimeter.
If the bank is non-cohesive, the side slope must not exceed the value for the
angle of repose given in figure 7.3. For a good design values 5-100 lower
than the angles given should be taken. Figures 7.9 to 7.11 can be used to
estimate three values of the slope S (depth = D in Simons notation, W = width,
R = hydraulic radius).The designer must now "invoke his engineering judgement,
guided by these slopes to arrive at the design slope".
It will be clear that with the regime theory only rough estimates. of
channel dimensions will be obtained. Experience in a specific area will be
of equal importance.
 7.6

1000 !DO

" So"" .... """

bonh 210

,.; a ,__o""
co.... lv. bon". ft.2.!UOU1'
IL
.!: C 00.... 1•• bH ..:
too IL
onel bank.
.,., ~ ~~
..•
CL
0 Coo, .. ftOft-co·
... "'" _or lal
E ..... rlol Dato
./
F.-.
~
V
j IIko a
UlO 31:
• Sln4 Cano', G
r
10
t
•
...
CL

•
31:
/.
V vit
e. D
~ -
1".
o SIIIton,
• '''''job
• Sind
a...,
Cono'.
COMI,
100

ao
J
1--
'1-1-1- -
1",,.,10' Cano".

• U8.8.R COfMIII
• h,,...,601 Conale
DI,charo., Q In cf, Wt".d Perimeter, Ft. 1-
10 110 1000 10000 Dó IDD
"'" I<UU
Ft•. 7.4 Variation ol wetted perimeter with dhcharge Fig. 7. 5 Variation of average 'Widthwith wetted perimeter
{or regime chaanele. lor regime channeh.

40 TTlil
I 1111111 1 Ui j_ II
2Cl IA Co ... .," bed and ""ka ~_-LUI
240

ROO
..
IL
a:
.;
IC ::11
...
B
C
0
E
Sabel Mct anel COhlll..1 bonlIl
Soncl bed ond "nb
COGf.. noft-Cohel"" mot ......
Irnperkll 0010 'Ik. B
R.O.43 Qo.HI
~
.!: 0 r-
a:
~ f-- ~ I-
Ie( .ti! .2
-; .- ,...,
-!. p ~ ..
-
V o Si-.onl ......

f;>r:f-' ,__
0 R.O.24700."1
i • PUft!."CaM', J." --- f-~ ~~
12(
•
l
:z;
.V ~
.
t;...- · Slmonl • a"ulII'

10
1
lO
o.er -::;z.
-- P
:-:.- Iz: -_--
~-- . - ·· Pun)ab Conota
o.s
··
1-,-,- $Ind canall
" --.-- ---j- ... 1----t-- -- - U.S. B.R. Cana"
40
0. -~
I- Imp.rlGlCana"

"'
Tap Wldth WT,ln Ft. 0.: Dl.charoe, Q
I ..... .... 110 220 210 uo 1000
100 I000O
Fig. 7.6 VariatloD of top wldth wlth averaie wldth for "
Fig. 7.7 Variation of hydraulic radius with dtec ha ege {or
regime channele. reginle channeh.

12

".1

1•
,V
~ v'•
8
~
.;
... ,
0
a:
.. .!! I
....1..1'"
0 Shnonl • Ie"', 0

:; SI'" c.a'I.
...!! !/r '" I.,.rlo' _'"

..
I

~ :z;'"
"')"

2
~ I-e""
V"'"
./
Avtraoe B.d D.pth, Ft.
~ 4 ti 8 10 1 15
Fig. 7.8 Varlatlon of average bed depth wlth hydraulic
radius for regime channeh.
 7.7

I IJ -l---- T 1_ I

!-
eoar.. non-coh..... ,-
f---- f-

I
111* . i

..-..
Y-IT.9(ftZS,o.-.1 •

"
• .....•
oE
1

•
~ ~~ ]:
u,,;, --)-/

-/e
LI. a

r> 0

---:~rtrvi
l
>
.....
'ü
"i
>
0

.... _
:
0
0
s~... ';";.
Y.18.0(Ras)t_

V~
OM
I)

~ ~
~.

0"
--
•
""""
.91....
. __

0
. 0*
....A y·I3.88(Ras)
o'Vc Sond b.d,
I!
! I"-

ond bonk.
.l-
o sI_a......,
I· P..,jall Cana.
• 9ind Conol.
• u.s.a. R. Conolt
• I",p.,iol C_I •
o SI",on. 8 Bend...
• Punjab Canal.
.. Sind Canaio
• U.S. B.R. Canai.
IJ~
.....
.. ~,.. ;:,7
<7,
,ji
... -
I'"
-
eo.o

~A ...... ,. line
Cl ~~ RlS
I .aaal .DOI .DI I
1 I 1I I0.0
E
Fig.7.9 Variation of mean velocity with RZS for
regime channels. (Lacey type slope relation. )
...
6
~ .. oE
--
...;" 0
...
rr .J>" el

Valu••
of O\_
~"
dt-.
.,...... 'ö
2
"0

-...
el
CD I.o
rtff
I
I
~- -- 1-
.. _- IIII
st' ~
1-. -
I TI
0.525(VW"pI'
I1III --

~ .. ~ •
T

~
-.. ./
"ti'
,'"

1
& c;
TI'
I1 llJ
0

:?
el
~ .-
.~
1-' ~I' E
+-ft=~ . --- -n_ I 11 ~n"" 0
u.,----r I ~ 1-- , •
1\":''' ",~d>- 0
- - po --

" t- -
pr;. c
a-
f+--

a ~bf-.~v~
~~~
,..... ~~
~ ~~~
I"';!I..I" v\;-'~~~IIIIII 0
ou

'b
-I--- ~ ••i.. b.d,
-.-I--+---+-+-.........,..,.. .- ~ .. -~
1-----

_.- t--~
""
ii21'FÇ; .. - -..----t--+
2
0.10
•
IJ)
.4~f;1J & i!jl" • ~~D.li 1 ,..--JO
'" ban'"
0 Sand bed. a cah•• Iv. 1IonU.'i.- . Sond bede a bant. 1---- !-- . .. con.be4'
IOC ~
•
_.,IL ".:~
> - ---+- - ._-
W ~F\. I I Im
1--'-
1-/
/ o Slmon. a Bend., ,...... !roctive Force,Y' RS, In Ibs/ft I
02
Q"Of 0.10 1.0-
• Punjab Cana"
'-_. -- .- • Sind Canala Flg.7.11 Variation of tractive force with mean alze of bed
material for regime ehannel.. (Tractive-force
type slope relation.)
VW
11 .~
T 111

VZ
I(l1O

VW
10
• s = slope
Fig. 7.10 Variation of gOS with ~ for regime channels.
iBlench-King type slope relation. ) W = average width
D = average depth
R = hydraulic radius
 7.8

7.5 Literature

R. Koech1in, 1924 Mécanisme de l'eau. Beranger, Paris.

E.W. Lane, 1953 Progress report on studies on the design
of stabie channe1s. Proc. ASCE~, Sep. 280
E.W. Lane, 1955 Design of stabie channe1s. Trans ASCE 120,
p , 1234-1 279
T. B1ench, 1957 Regime behaviour of cana1s and rivers.
Butterworth, London
L.B. Leopo1d, 1957 River-channe1 patterns: Braided, meandering
M.G. Wolman
and straight. u.s. Geo1. Survey, prof.
Paper 282 B.
D.B. Simons e.a., 1960 Uniform water-conveyance channe1s in a11uvia1
M.L. A1bertson
material. Proc. ASCE 86 (HY5), p.31-71
F.M. Henderson, 1966 Open channe1 flow. Madfi11an, New York
 7.9

7.6 Problems

7. 1 Given: A channel is excavated in

very rounded material with
D = 5 cm. Design depth is
~
I 4 m ---.fl-
l ~~
I-:-:--_..... 4 m

2 m, side slopes 1 vert:2 hor.

Question: What are the permissible slope and discharge, using Shields
curve for T ? Which part of the channel is critical?
cr
(slopes or horizontal part).

7.2 Given: A channel is designed in coarse very angular stones.

DsO = 2 cm
90
= k s = 4 cm for a discharge of 95 m3• The required
D
depth in the middle section is 3 m. The side slopes are 1:2
(1 vert, 2 hor.) and the width of the horizontal part is 20 m.
Questions: \Vhat is the necessary slope of the channel (uniform flow) to
transport this discharge. Is the channel stable in this
condition if ~
cr
= 0.03 ?

7.3 Given: The side slope of a channel has to be protected with very
rounded gravel. Channel data:
depth (horizontal part) h = 2.5 m
side slope 1:3 (1 vert, 3 hor.)
channel slope 1= 4.10-4
Question: What l.S the minimum stone size assuming uniform material?
Shields diagram can be used for T
cr

7.4 Given: A channel is made in a layer of coarse rounded gravel

3
(DsO = 0.04 m, ks = D90 = 0.08 m) for a discharge of 100 ~ Is.
The slope of the channel in the flow direction is equal to
-4
3,5.10 ,the bottom width is 20 mand the side slopes are 1:2
(I vert, 2 hor.)
Question: What is the water dep th for this discharge and is the gravel
stable both on the horizontal part and on the side slopes?
Shields curve can be used for T
cr
(Compute the discharge for various depth and interpolate to
obtain the design depth. Then check the stability).
~ t _/
~I 2h
~ 20 m . h
i' ./A' 2h
 7.10

7.5 Given: A stabie channel has to be designed for

3
Q = 56 m /s = 2000 cfs. The type is B (sand bed, cohesive
banks).
Questions: What are the dimensions and slope of channel using the Simons-
Albertson method (Figs. 7.4 to 7.10).
 8.1

8. RIVER BED VARIATIONS, AGGRADATION AND DEGRADATION

A natural river will never be in an exact equilibrium condition. Varia-

tions in discharge ean give variations in bed level, roughness etc. Also
changes in the regime of a river may give deviations from an equilibrium
state. If the sediment discharge S entering a river reaeh is greater than
the equilibrium value Se aggradation will occur until a new equilibrium is
approaehed. Some examples:
aggradation upstream from a reservoir,
- tributary channel bringing heavy sediment load to a main channel
giving loeal aggradation,
- river regulations eliminating floods which formerly periodically
eleared the ehannel of accumulated sediment.
Some examples of degradation (S < Se):
- degradation downstream óf dams,
- canals in fine material earrying clear water,
- realigned channels with increased slope.

For calculations on non-steady or non-uniform conditions it is neeessary

to introduee a sediment transport relation, for example of the form s = a.v.b
For the transport relations given in Ch. 6 values of b in the order of 3 - 7
are found (high values for low transport rates), with b = 4 to S for high
sediment rates and fine material. (Engelund b = S, Shinohara b = 4.6).
Applieation is shown in the following example.
What will be the reaction of a river to alocal decrease ~n width?
Suppose Ql = QO (continuity) So = SI Cafter some time continuity of sediment
transport) S s.B B = width.
Cl =
Co (Chézy value, index 1 = new situation; index 0 = old situation).
b l-b b -b -1 -1
From S = B.s = B.av = B .aQ.h Cv = QB h ) it follows that:

From the Chézy formula it follows that:

B h 3/21 1/2 = B h 3/21 1/2
l' I I 00 0
with the relation for hl/hO it follows that:

3
b
 8.2

It can be seen from the equation for hl/hO that for a large va1ue of band
va1ues of BO/B1 not far from 1, that:

hl/hO::::BO/B1

or the decrease in width 1S compensated by an increase 1n depth. The slope

wi11 a1ways decrease.
This rough approximation on1y gives a first estimate. For accurate
va1ues detai1ed computations or model studies wi11 be necessary.

COmputation on river-bed variations

The reaction of a river to a change in its regime (meander cut-off,
dam) can be computed with the equations of motion and continuity of water
and sediment. In most cases non-steadiness of the flow may be neg1ected so
that the fo110wing equations are va1id for constant width:

(1) dU + ah + dZ
U dX u.Iut h waterdepth
~ gdX = -~-
Z = bed level from referénce datum
C = constant
C = Chézy va1ue
(2) U.h = q = constant R = hydrau1ic radius

(3) s = f(u) q = river discharge/m'

s = sediment discharge/m'

(4)

Solving these equations requires numerical techniques (see Vreugdenhi1 and

De Vries 1973, De Vries 1973, De Vries 1975).
As a first approximation it may be assumed that the flow is uniform
and two-dimensional so that equation (1) reduces to:
dZ u2 u3 u03
(5) -=--=-- (q = U.h) Ia =
dX
C2h C2q C2q

or
d2Z
--=
dX2
-3-
U2
. dX
dU
C2q
Combination with (3) and (4) gives (~: = :~ • ~~):

(6)
dZ d2Z
ät-k-=O
dx2

in which k =l C2q.(ds/dU) 1 U.(ds/dU)

3 U2 =3 10

in which I = slope and index 0 refers to the origina1, uniform situation.

 8.3

Af ter linearisation (possible for U/Uo ~ 1)

1 UO(ds/dU)
k ="3 10

b
or for s au k =.!.b
3
s
• I
o
The equation (6) is a parabolic one (diffusion equation) for which solutions
are known. This will be applied to compute the reaction of a river to a
sudden decrease in sediment discharge s which will give a decrease in bed
and water level of Zo after a long time.
Introduce z' = z (x, 0) - z (x, t)
with boundary conditions:
z' (x, 0) 0
z' (0, t) = Zo

The solution of (6) is given by:

z'
-- = erfc (x/21kt)
Zo
in which erfc is the function: erfc (a) =
2
7IT ~ coe- ç2dç
From this solution one can compute for which time tso at x xo' 50% of the
finallowering of the bed (zO) has been reached.

This is the case for erfc (xO/2~0) ~ 0.5

or xo ~ 1.0 • ~ (see tables for erfc (a»

or tso z x~/k

From comparison with the solution of the full equation it appears that this
-1
solution is valid for: xo ~ 2 10 • h.

Suppose s = 104m3/year (river with B = 100 m S

I = 2.10-4
b = 5
h = 3m
Then the solution is valid for x > 30krn

5 104
0.83.108
k = "3 • 2.10-4 =

or ~ 10 years
 8.4

This means that at 30 km from the discontinuity 50% of the expected change
in bed level will be reached in 10 years.
This means that the reaction of this river to change in regime 1S relatively
slow.
For a move complete treatment of these problems see the references
cited above.
 8.5

LITERATURE

M. de Vries, 1973 Riverbed variations, aggradation and degra-

dation.
Int. seminar on hydraulics of alluvial
streams. IAHR New Delhi.
C.B. Vreugdenhil, 1973 Analytical approaches to non-steady bed-load
M de Vries
transport.
Delft Hydraulics Laboratory, Research
Report S 78 - IV
M. de Vries, 1975 A morphological time - scale for riveis.
Proc. l6th IAHR Congress Sao Paulo 1, p. 17-23
 9.1

9. LOCAL SCOUR

9.1 INTRODUCTION

Local scour is caused by local disturbances of the flow and sediment transport
field. Examples are: scour around (bridge)piers and abutments and scour döwn-
stream of dams. In all these cases alocal increase in mean velocity and/or
turbulence intensity gives an increase in local transport capacity. From the
equation of continuity:
ah as
at = ax (h = depth, S = transport)

it follows that scour will occur. The scouring continues until the local depth
has increased so much that the velocities are reduced sufficiently to bring
as to zero.
ax
Scan remain positive of course so that a dynamic equilibrium is obtained, for
example for a pile in a sediment transporting river.

There are too many examples of failure of constructions due to local scour to
neglect the phenomenon. The effects of local scour can be overcome by an increase
in construction depth (bridge piers) or diminished by a bot tom protection.

The following subjects wil1 be discussed:

- scour around (bridge) piers,
- scour downstream of constructions (dams, weirs),
- scour around abutments and spur dikes,
- model investigations,
- protection.
 9.2

9.2 SCOUR AROUND BRIDGE PIERS

Scour around bridge piers is due to a combination of three effects:

- local scour ne ar the bridge pier caused by the disturbance of the flow
field around the pier,
a lowering of the river bed in the cross section of the bridge due to
the contract ion of the river profile at that section,
- a general lowering of the river bed in the river around the bridge site
due to degradation or non-uniform river bed changes during floods.

These last two aspects, together with practical experience for the situation
of a bridge in the flood plain, are discussed in an excellent way by
C.R. Neill (1977).
The local scour near the bridge pier is discussed 1n detail in a review
article by Breusers, Nicollet and Shen (1977).

The dominant feature of the flow near a pier 1S the large-scale eddy structure
which can be composed of the horseshoe-vortex system, the wake-vortex system
and trailing vortices. Vortex filaments, transverse to the flow in a two-
dimensional undisturbed velocity field, are concentrated by the presence of
a blunt-nosed pier to form the horseshoe-vortex system. The mechanism by
which the concentration is accomplished is the pressure field induced by the
pier. If the pressure field is sufficiently strong, it causes a three-
dimensional separation of the boundary layers which, in:turn, rolls up ahead
of the pier to form the horseshoe-vórtex system.
A pier developing such a vortex system is called a blunt-nosed pier (Fig. 9.1).

FLOW 1
OEPTH do

Fig. 9.1
 9.3

Sharp-nosed piers will become "blunt-nosed" under larger angles of attack.

Scouring generally starts at the sides of the pier but as the scourhole
increases in depth also the vortex system increases 1n size and strength.
In a later stage, if the scour hole is sufficiently deep, vortex strensth
and rate of scour decrease again until equilibrium is obtained. Due to the
horeshoe vortex, maximum scour occurs at the ~stream side of a circular
pile.

For an example of the scour around a bridge pier see Fig. 9.2, which shows
data from a prototype case (cylindrical piles with a diameter of 8.5 m in
the Niger River and the corresponding model results). For this low river
stage already scour depths of 8 m were observed, increasing to 12 m for the
flood stage.

_____________ .•• _......

- ..._......
,.:1.........

•••
,
I
,,
!
.....
---. /":.,
I
I
\I
,,
,

----- -- -----
NODIt

Fig. 9.2
 9.4

Scour around piers starts at a velocity equal to about 0.5 Uc (Üc ~s the
critical mean velocity for beginning of motion ot sediment). Scour then
increases with velocity until Ü ~ Uc and remains practically constant
thereafter. For U > Uc variations in scour depth due to approaching bed
forms occur but the average depth is constant due to an equilibrium between
sediment discharged into the scouring hole and the amount of eroded sediment
(see Fig. 9.3, taken from Chabert and Engeldinger).

o 2 3 5 6 o 2 '4 5 6
_fIHOURS) _fl/fOURS)
TESTNo."3 D. 3mm TEST No. 112 D. 3mm
d.. 20cm d.. 20cm
Ü.52 cm/s U.52cm/s

Fiij. 9.3 Scour as a function of time V < Ve• Scour as a function of time V> Ve•

Scour depth ds increases with the initial river water depth do untill the
water depth/pile diameter ratio becomes larger than 2. For larger ratios
scour depth d only depends on the pile diameter b. Grain size of the bed
s
material has a relatively minor influence, but cohesive sediments give a
smaller scour depth. Important factors are the pile shape and the angle
of attack for longer piers (Fig. 9.4, 9.5).

e
ol!
40
DIMENSIONS IN
0,0
cm SHAPE
J's, CIRCULAR
Multiplying Factors
....~
t 30
0o,
x ,
°
~l's
ol» 2 CIRCULAR
ROUND-NOSED
G tor
Angle ot Allack

• CHABERT AND ENGEL-

o, <::::=>J'5 LENTIClLAR DtNGER Lib."

+, C::::>l,s JOUKDWSKI 4 -Ar-,..L..,,.L.+-.,,.....~+---a*-=:::::::::_---1x VARZELIOTISLlb.6

d • .;D.15m KoL
'0 Ü ,aIS5m/s

.,.o:t~2
D , ~1'5 BEVEL-NOSED

THREE PlLES
3

0 7 ts" 30· I-~

..,..-a

30 45 60 75 90
A"9le of A110ck in Ot9feu

Fig. 9.4
Fig. 9.5
 9.5

Observations on railway bridges in India showed that scour de~th could be

related also to the Lacey regime depth

d
s
- d
r,3
= 0 '473 (Q/f)I/3

Q = discharge (m3/s)
f silt factor = 1,76 D50!

For practical application the following design relation ~s given by

Breusers, Nicollet and Shen (1977):

d scour depth (below original river bed)

s
b = width of pier
-
U = mean velocity
Uc critical mean velocity for beginning of motion

d water depth
o
a angle of attack
~ length of pier

U
fl (U) = 0 for !!.....~
U ... 0.5
c c
U U
= 2 (- - 0.5) for 0.5 ~ -~ I.
U U
c c

= 1.0 for !!.....:>-

U "
1.0
c

For most practical situations:U/uc > 1.0 so that fl(U/Üc) 1.0

f2(do/b) = 2.0 tanh (d /b)

0
~n which
x e-x
e ..;.
tanh(x)
x -x
e + e

f3(shape) = 1.0 for circular and round-nosed piers

= 0.75 for stream-lined piers
= 1.3 for rectangular piers

..
For f, (a, ~/b), see Figure 9.5 .
 9.6

If the river bed around the pier is protected with arevetment, then it
should be placed at or below the lowest river bed level to avoid an extra
obstruction. The stone size should be designed for a velocity 2.0 times
the approach velocity Ü to account for the increase in velocities near the
pier. For a preliminary design the following relation can be used:

u = 0.42 12g6D

D stone Sl.ze.

9.3 SCOUR DOWNSTREAM OF CONSTRUCTIONS

The construction of a dam or a weir l.na river changes the transport conditions
and causes local scour. In literature several approaches can be found:

These rel~tions were derived mainly for coarse material (d > 1 mm).
Examples are the relations given by Eggenberger and Müller [1]

overspil1: T
... __ ..._ .... --_.
. -. ... .. ,.i-

---;> cJnr ./,b"//

jl(

undershot: T
L 4'.n~'" .r,,(0,L

Tand H in m, q in.m2/s, D90 in mm

T = sum of downstream waterdepth and scour depth.

or by Kotoulas [2J

T 1.9g-0.35HO.350.7D -0.4
q 95
 9.7

The starting point is the regime depth d for example the Lacey expression:
r

d = 0.473 (Q/f)1/3 (m or ft-units)

r,3
Q = total dis charge (m3/s or ft3/s)

or if the flow is limited in width:

d 2 = 1.34 q2/3.f-l/3 (m-units)

r,
q = discharge per m' (m2/s)
f = siltfactor, sometimes given as 1.76DO• 5 D in mm

The total scoured depth T (sum of original waterdepth and scoured depth)
is than taken as a multiple of the regime depth:

for scour near bridge piers T 2 d

r
for scour at nose of spur dikes and guide banks T = 2 to 2.75 d
r
for flow perpendicular to banks T = 2.25 d
r
downstream of barrages with hydraulic jump on
the stilling-basin floor T 1.75 to 2.25 d
r

For several practical problems, the equilibrium scour depth is not of

interest because the situation in which scour occurs is only of a
temporary character. Examples are closure works in tidal channels in
which scour has to be considered only during the construction phases.

Interpretation of model tests requires in this case the knowledge of the

time scale of the scouring process. The Delft Hydraulics Laboratory developed
relations based on a large number of tests (see Breusers (1967),
Vinjé (1967), van der Meulen and Vinjé (1975».

During the closing of an estuary situations will occur with a greatly reduced
cross section whereas the tidal discharges remain very large. This means
that the mean velocity in the closing gap and the turbulence strongly increase
which gives an increase in scour depth. Especially methods in which an
estuary is closed from the sides (for example with caissons) will have an
enormous scouring potential (see Figures 9.6 and 9.7). The scour depths can
be reduced by making bottom protections on both sides of the closing gap but
scour will always occur. If the sand bed is loosely packed, flow slides
can be triggered by the local scour.
 9.8

parI of dos..-e dam

parI d dosure dam 5

5
'\-1
!
'r-',!
I
I

.
-j/
/""

a.0 1i
.
I --
.........
I
1"-'-
.--...._ 1'-~-»>:
IOLKERAK EEE3 cais!ons
Yeio<:ilydislribulion
. i
I
after Placing of 10 flow
caISSOns '.0 1-.0-11 I
veloeilies befIre
piac ing of caissons x Or·3C1j,

_
eddy ,Ireel
f10wlines
0.'(- - Ij
o "'_!
......._,,_ I
11 r-o
o I
soo
° '00
--.....I
200
IN •
4.00

Fig.9.6 Velocity distribution in a Fig. 9.7 Scouring-pattern after 10 hours model. Dam-heightjwater-
closure eap. depth = 0.6.

The general conclusion of the studies by the Delft Hydraulics Laboratory

on Iocal scour was that for a given flow field, independent of the bed
'material the scour could be expressed as a unique function of time:

h
max
-h-- =
o

h
max
= scour depth (measured from the original bed level)
h
o
= original water depth
= time to reach h = h
max 0

The independenee of bed material and flow velocity is shown in Fig. 9.8.
 9.9

2 3 4 5 6 7 9 10

0.1 1

\ ~t
i
I
1= 0,0211 ••
•
•I
I
__ +
0

0.2 r-- .1 --4- -4---' &~ t~o r

0
material d

h
N ~i-t--(=- _,~_~--f+ •
0
1.50 m
1.50m
1.13 mIs
0.83 mis
sand
sand
220
220
Jlm
Jlm
V 4 0.50m 1.10 mIs sand 220 Jlm
ho 0 0.25m 1.10m/s sand 220 Jlm
,•
-,
0.50m 0.275 mIs pol,st,rene 1500 JI m
~ 0.3
t\v 0.25m 0.20 mis polyst",ene 1500 Jlm
0.56 mis bakelite 1900 Jlm
- ..,
+ 0.25 m
• 0.25 m
0.25 m
0.25m
O.SOm/s
0.72 mIs
0.88m/s
bakel't.
sand
send
600
120
400
Jlm
Jlm
jrm

r.=tJ
V 0.25m 0.88 mIs sanel 1600 jrm

0.4 .. t A-!-_I __'=0,1'1 ~

~--j-~-~
0.5
P und = 2650 kg/m 31
p pol,lt"ene = 1050 kgl m
p bekelit. ~ 1350 kg/m~
3
I r I -T.
Fig. 9.8 Comparison of scour profiles for scour downstream of a rough
horizontal bed.

For two-dimensional scour it was found that:

h
max
-h-- =
o
but for other (three-dimensional) situations other relations apply (see for
example Figure 9.9). These figures also show that the relationships are
independent of bed material and waterdepth for a given geometry.
 9.10

.
rJ.h.:i. " , . .. ' .
...... .
h

:,rI

~
.'lv'"
.
,~v D/hO=o..3
0..2
. . v.- .sand
' bokelite
'/
V • potystyrene 16o.o.}.lm
• polystyrcne 2400).lm
0..1
I I I I I I I
0.0.1 0.0.2 005 0.1 0..2 0.5 2 5 10.
t
-----+ 11
3

·.l;·:
..
: •• ' ft

\ , .. ,;ot
~-r' •

~
_""_'It'·
.
." .e..

.A!"'"
~r
.. D/hO: 0.3
.~.!
0.2 ..
V. . B: lOm ho :0.50m
"
V
-/ .
B= 5m ho =0.25m
0.1 B= 2.5m ho :0.125m
f--
I I I I I 111
0.01 0.0.2 005 0.1 0.2 0.5 2 5 10
-----+ tt1

pig. 9.9 Seour downstream of a dam (sec Fig. 9.7)

Important is the time seale of the proeess; or the seale of ti' For all tests,
both two and three-dimensional, the following relation was valid:

-4.3
n(etÜ - Ü )
cr

n = seale factor (prototype/model)

~ = (ps - pw)/pw

h = waterdepth
et = factor, depending on flow field and turbulenee. For uniform,
two-dimensional flow et = 1.5, whereas for very turbulent
three-dimensional flow situations et ean be as high as 6 - 8
(see Van der Meulen and Vinjé, 1975)
-U mean velocity at the end of the bed protection
'-Uer = eritieal mean velocity for beginning of motion.

This relation has been proven to be valid also for predietions of seour under
prototype eonditions.
The time seale nt for the seouring proeess is different from the hydraulic
I _I
time seale nt = nL . nU
 9.11

All relations given above were for cohesionless materiaIs. In fact only a
limited number of experiments have been performed for cohesive soils. The
scouring resistance of clay is of course larger than for sand. No general
relations can be given however.

9·4. SCOUR AROUND ABUTMENTS AND SPUR DIKES

For these types of constructions no general design rules may be given, except
the general relations of the regime theory. The actual scour depends too much
on the geometry of the construction and the flow field. Some references are
given below:

B.P. DAS. Bed scour at end-dump channel constructions.

Proc. ASCE 99 (HY12), 1973.

C.R. NEILL (ed.). Guide to bridge hydraulics.

Ontario, University of Toronto Press, 1973.

M.A. GILL. Erosion of sand beds around spur dikes.

Proc. ASCE 98 (HY9), p. 1587/1602, 1972.

L. VEIGA DA CUNHA. Erosoes localizadas junto de obstaculos salientes

de margens.
Diss. Lisboa, 1971.

9.5. MODEL INVESTIGATIONS

For model studies on the equilibrium scour depth the following scale laws have
to be considered:
a. undistorted model nL = nh

b. Froude law nÜ = nt in view of the necessary reproduction of the free

surface.
c. nu = nu to obtain a correct reproduction of the equilibrium
x xcr
conditions in the scour hole.
The third law reduces to the simple law:
nD = nL
 9.12

if the bed material in the prototype is so coarse that the model material is
larger than ] mmo If the model material becomes finer, deviátions from this
simple relation occur due to the influence of viscosity (Shields curve).

If the material in the prototype is already fine, one cannot fulfill all
scale relations using sand in the model, sothat materials with a lower
density have to be used. In that case also the time scale of local scour
can be of importance (see par. 9.3.3).

9.6. PROTECTION AGAINST SCOUR

Scour can be reduced by streamlining the construction (bridge piers), making

guide walls (abutments) or by stilling basins (spillways). If the resulting
scour is not acceptable a bottom protection has to be constructed. Except for
the circular bridge piers no general design rules can be given because the
necessary protection depends too much on the actual geometry, the composition
of the bed etc. A minimum requirement is of course that the upper part of
the protection is stabie against the flow and that the filter construction is
sufficient to prevent leaking of sand through the protection. Special care has
to be given to the end of the protection where undermining has to be avoided.

Both stability and filter construction arc discusscd in thc lccturc notes
on "Revetments". See also Par. 4.6.·
 9.13

9.7 LITERATURE

EGGENBERGER , w., 1944 Experimentelle und theoretische Unter-

MtIT.LER,R.
suchungen über das Kolkproblem. Mitt.
Versuchsanstalt für Wasserbau, E.T.H.
Zürich, No. 5.
BLENCH, T. 1957 Regime behaviour of canals and rivers,
Butterworths, London.
KOTOULAS, M. 1967 Das Kolkproblem unter besonderen Berück-
sichtigung der Faktoren Zeit und Geschiebe-
j
m1schung.
Diss. Braunschweig.
BREUSERS, H.N.C. 1967 Two-dimensional local scour 1n loose
sediments.
Delft Hydr. Lab. Publ. No. 64.
VINJé, J.J. 1967 Local scour caused by vortex sheets.
Delft Hydr. Láb. Publ. No. 64.
NEILL, C.R. (ed.) 1973 A guide to bridge hydraulics.
University of Toronto Press.
VAN DER MEULEN, T., 1975 Three-dimensional local scour 1n non-
VINJé, J.J.
cohesive sediments.
Proc. 15th Congress IAHR, Sao Paulo,
Vol. 2, Paper B33.
BREUSERS, H.N.C., 1977 Local scour around cylindrical piers.
NICOLLET, G.,
Journalof Hydraulic Research 15 (3),
SHEN, H.W.
p , 211-252.
 10.1

10. MEASUREMENT TECHNIQUES

10.1 Introduction

Besides computations on sediment transport also measurements are

necessary for verification of theories, good description of a river and of
consequences of changes 1n regime. The existing techniques can be divided
into two groups:
- measurement with samplers
- tracer techniques.

10.2 Measurements with samplers

Due to the difference in mechanism of bed-load and suspended load there

are different samplers for each type of transport.

1. Bed-load samplers
Variations in bed-load transport and the influence of bed forms
will give large variations in results with samplers. For a significant
value of the bed load a large number of observations has to be taken.
Bedforms can give variations in local transport rate of 0 - 2 s (zero
in the trough of a sand wave and 2 s at the crest of the wave). It is
therefore advised to take many samples at various locations instead
of taking long sampling times.
Further problems with bed-load samplers are:
1. It is difficult to give the equipment a correct vertical and hori-
zontal alignment with the bed.
2. The meter should be calibrated to determine the efficiency which is
also a function of the amount of material caught.
3. It should be avoided that the sampler collects bed material during
the lowering of the instrument.
4. A sampler disturbs the flow field. Scour or a decrease of velocity
in the sampler can occur

Most samplers are of the box or basket type and consist of a pervious con-
tainer. Water and sediment enter the sampler; the sediment is caught. As an ,
example figure 10.1 is given. This sampler is used extensively on Dutch
and other rivers for D > 0.4 mmo The Helley-Smith sampler (Benedict 1979) 1S
similar.
 10.2

STEERING FIN

RUBBER
CONNECTION

1.70m
-I

Fig. 10.1 Arnhem sampler (BTMA)

2. Suspended load samplers

The amount of material caught depends on the hydraulic coefficient
(U inside/U outside) and the efficiency (% of material caught ) of the
instrument. Both factors have to be determined by calibration. The following
points should be considered:
1. Suspended load shows large fluctuations So that repeated sampling is
necessary.
2. Suspended sediment sampling also includes part of the wash load.
3. If only concentration is measured also velocity profiles have to be
measured.
There are two types of samplers:
a) concentration samplers: a value of water is sampled at a certain level
or as an average over the depth. (Nansen bottIe, mouse trap, depth
integrating samplers).
 10.3

b) suspended-load samplers: suspended-load (U.C.) 1S measured at a point

or by integration over the depth.
For an example see figure 10.2 (Delft bottle). The bottle acts as a sand
trap so that most of the material is caught. Another type of suspended
load sampler is the US-P61, where a bottle is filled with a velocity
approximately equal to the flow velocity (Fig~ 10.3).

1 STRAIGHT NOZZLE
2 DIFFUSOR CONE
3 FLOW CHANNEL
4 SAMPLING CHAMBERS
5 WATER OUTLET
A
6 HINGE

CROSS SECTION A-A A

Fig. 10.2 Delft bottle

I
I
r-I

3
i
i

-us P-61 POINT-INTEGRATING SUSPENDED-SEDIMENTSAMPLER

Fig. 10.3
 10.4

3. Bed-material samplers

Numerous divices are described in literature to collect bed-samples.

Several types may be distinguished: grabs, corers etc.
For some examples, see Fig. 10.4

@ e @@

SECTION A-A

-us BM-54 BED MATERIAL SAMPLER

.__..-.----e

c....... t .... 1

'CA_DGl" COIIE1I f-------

..... 1
"VAN VEEN'"
GII"8

_.....
II_.,.~-e

Fig. 10.4 Bed-material samplers.

 10.5

10.3 Tracer techniques

For sediment transport measurements also tracers can be used. Grains

are marked so that their transport characteristics are not changed, are
added to the flow in small quantities and their displacements are deter-
mined. From their displacements the transport can be computed.
Several types of tracers are usedi
1. ~!~~!~~~~~!
(luminofores). Marked grains can be detected af ter sampling
under U.V. light. Different types can be used simultaneously.
2. g~2!~~~!!~~.
Natural sand is provided with a coating with radioactive
material. Disadvantage: public health is important because relatively
large quantities are necessary to remain above the back-ground level
of radioactivity. Advantage: detection in situ.
3. ~~!!~~!!~~_~~~!~~!~.
Particles are marked and radiated af ter sampling.
Difficult procedure; applied for silt.

Several techniques for the interpretation are used:

1. ~~~~!~~!_!~i~~!!~~_~~!~~2'
A constant amount of tracer material (rate T)
is distributed over the profile and injected during a long time-interval.
At a downstream cross section samples are collected and concentration
as a function of time is determined. Af ter some time the concentration
becomes constant = CO. Then the rate of transport can be computed from
the relation:

c ~:;;a---c.

2. ~~!~!:!~i~~!!~~_~~!~~2.
At a certain time an amount of tracer material
is injected. At several downstream locations concentration is determined
as a function of distance. From the displacement of the centre of gravity
of the concentration-distance curves the average transport velocity can
be computed. Multiplication with the effective depth of transportation
o gives the rate of transport. The effective depth is of the order of
half the height of the bed forms and can also be determined by sampling
 10.6

in the bed. Both methods are relatively inaccurate.

Problems with these techniques are the length of the measuring interval,
the fact that the external conditions have to be constant and the large
number of observations. Some of these restrictions can be diminished by
applying "dispersion methods". The data are compared with a theoretical
dispersion model (see de Vries, 1966).

For a review of existing measuring techniques see also Jansen (1979).

Development in this field is slow. For same recent developments see
Anon. (1976).
 10.7

10.4 Literature

Anon., 1963 A study of the methods used in measurements

and ana1ysis of sediment loads in streams.
Interagency Comm. on Water Resources, Rep.
no. 14, Minneapolis.
M. de Vries, 1966 App1ications of 1uminophores ~n sand-transport
studies.
Diss. Delft (Delft Hydrau1ics Laboratory
Pub1icatiön no. 39).
H.P. Guy, 1969 Sediment measuring techniques. F1uvia1
sediments.
Proc. ASCE, 95, HY 5, p. 1477-1514.
W.H. Graf, 1971 Hydrau1ics of sediment transport.
McGraw Hi11, New York, eh. 13.
Anon., 1976 Proé 3d Federa1 Interagency Sediment Confe-
rence, Denver Colorado, Sediment Comm.,
Water Resources Counci1.
P.Ph. Jansen eed), 1979 Princip1es of River Engineering
Pitman, London. Part 3 eh. 7.
P.C. Benedict, 1979 Equipment for investigation of f1uvia1
sediment.
Proc. ASCE 105 (HY 3) p. 163-170.
 11.1

11. SEDIMENT TRANSPORT IN PIPES

11.1 Introduction

Sediment transport in pipes is of importance in the field of sewage

transport and dredging. In both cases the purpose of the system is to
transport solids without deposits at minimum head losses. Important aspects
of this way of transport is the prediction of head losses and of minimum
(or critical) velocities to avoid deposition.
In pipe transport several modes of transport can be distinguished:
1. for fine sediments and high velocities a psetido-homogeneous suspension
is formed. The criterion for a fully developed suspension was Ux/W > 10.
- x -
For U/U ~ 20 it follows that U/W > 200. This means in practice
(Ü = 3-4 mis) a value of W < 2 cm/s or D < 150 ~m.
2. heterogeneous suspension. For smaller velocities and coarser material
a heterogeneous suspension is formed with a strongly non-uniform concen-
tration distribution.
3. Sliding bed regime. For very coarse material all sediments will be
transported sliding a10ng the pipe wall. The criterion for beginning
of suspension was Ux/W > 1.5. It follows then for Ü/W > 30 or W > 10 cm/s,
D > 1 mmo
Thus:
D < 150 ~m pseudo-homogeneous suspension
150 ~m < D < 1 mm heterogeneous suspension
D > 1 mm sliding bed
In practice transition zones between the various regimes will be found.
Typical head-loss curves for pipe-line transport of sediments are given
in figure 11.1. For large velocities losses approach the Loss for clear water,
but at the critical velocity where deposition starts head-losses strongly in-
crease. The difference between head-10sses for the mixture and clear water
increases.linearly with concentratlon c.
Critical deposit ';;locity Vc

5.02
3.53

1.5 4 6
V, m/sec

Fil· 11.1 Head loss vs. velocity relationship with equiconcentration

Iines, for sand graded to 0.44 mmo [After CoNDOLIOS et al. (/963).)
 11.2

11.2 Critical velocity

The critical velocity is defined as the velocity where stationary

deposits are formed at the bottom of the pipe, or by deposition from the
suspension or because part of the sliding material comes to rest. This
condition is very critical for the operation of the pipe line because
around the critical velocity head losses are at a minimum and a decrease
in velocity will give an increase in head loss.
A large number of data and formulas are given in literature. The
best known are the results by Durand (1953) (see figure 11 .2 , where

FL = ij . /12gK(2a)' and a = pipe radius).

crI.t
From experiments it fol10ws
that the critical velocity
UcrI.
Ot increases with (2a)~,
slightly increases with
1.2
transportconcentration C (up /
1.0
to volume concentrations of 0.8
fÇ \ \
15%) and increases with grain fi. 0.6
f/ \\ ~C"5%
~C"0%
size D up to D ~ 1 mmo I ~~. 5%
C' 2 %
0.4
The values given by Durand are 0.2

somewhat pessimistic for fine

00 2 3 4
sand in large diameters; here Fig. D.mm
and increase with a smaller 11 • 2 FL value vs. particIe diameter, concentration as parameter.
[After bURAND et al. (/956).]
exponent (1/4 to 1/3 instead
of 1/2) gives better results.
For more information see the
literature.

11.3 Head losses

Most (empirical) relations for the head losses in sand-water mixtures

have the form:
I = I + C.f (U, 2a, grain characteristics)
m w
I = hydraulic gradient for the sand-water-mixture
m
I = hydraulic gradient for c1ear water
w

I =-
À U2
w 2a • 2g
 11.3

For À smooth-pipe values can be used, because the sand polishes the surface
of the pipe. For I various relations are proposed for example:
m
Durand (1953):

1 - 1 3/2
<P = C.l
w
m w = 176
I ~.
Ü 2
~
{go I
(see figure 11.3, v ~ W D ~ 2a d ~ D grainsize)
ss

Führböter (1961):

in which Skt is a parameter depending on grain size

(see figure 11.4, d ~ grainsize, D ~ pipe diameter).

From experience with dredge pipe-lines it appeares that the relation

given by Führböter is suited for fine sand and large pipe diameter and
that Durand's equation is good for smaller pipe diameters and coarse sand.
For further information see literature (par. 11.4).

,..ow Ol' SOLJO.UQUID I1IXTUIW IN I'IPIS 6.0

80
5.0 0=330 mi Sogreah-
I doto
~ Hf--- 0=150 mm} .
..'"
v

<,
4.0
W
0=175 mm sovte+.;
0=200mm doto _

70 E 30 Ll III
+ ~ 1..._ "0=104 m~}-
0=150 mm Sogreah dato
Withouf~t

• Sorttd tond 2.0 .1 I

+ .0 =150 mm SO,?rea~dato
o Miatur.
+ '--r- 0 0= 300 mm Fuhrboter dato
-+\
6 NoturallOnd 1.0
60
1 1 1 1 1
.-
+---Wlth ~it

~-+l o 1 1 1 1 1
o 10 20 30 40 50 60
+ d,mm
+
50
o 2.5 FI,.11.4 Relationship of the sediment constant si, [Af ter
FÜHRBÖTER (196/).]

o 2.5 5 1.5 10 12.5

.t!1!f (D' 4".
IV .. ,.
in dopatition 'oti ... l

,.1J.5Hoed 1<* u. pen by lhe ~ .... tIoa ror oaad ('. - 2.65);
SOOREAH data .... uted. [Aft... 0matT (l1llfO).)
 11.4

11.4 Literature

R. Durand, 1953 Basic re1ations of the transportation of

solids in pipes.
lAHR Congress, Mineapolis, p. 89-103.
R. Gibert, 1960 Transport hydrau1ique et refou1ement des
materiaux, Annales des Ponts et Chausées,
130, p. 307~373, p. 436-486.
A. Führböter, 1961 llber die Förderung von Sand-Nas sar Gemischen
in Rohr1eitungen.
Mitt. Franzius lnst. ~, p. 1-152.
A.G. Bain, 1970 The hydrau1ic transport of solids by pipeline.
G.T. Bonnington
Pergamon Press, Oxford.
l. Zandi (ed.), 1971 Advances in solid-1iquid-f1ow ~n pipes and
its application.
Pergamon Press, Oxford.
W.H. Graf, 1971 Hydrau1ics of sediment transport.
McGraw HilI, New York, ch. 15 and 16.
Anon. Proceedings of conferences on Hydrotransport,
organised by the B.H.R.A.
No. 1, 1970 Conventry, U.K.
No. 2, 1972 Warwick, U.K.
No. 3, 1974 Golden, U.S.A.
No. 4, 1976 A1berta, Canada
No. 5, 1978 Hamburg, BRD.
No. 6, 1980 Canterbury, U.K.
No. 7, 1982 Sendai, Japan
..

"
 ,
I