POST - GRADUATE COURSE

IN

SEDIMENT TRANSPORT TECHNOLOGY

JUNE 4 – 29, 2007

GENERAL DIRECTORATE OF STATE HYDRAULIC WORKS

TECHNICAL RESEARCH AND QUALITY CONTROL DEPARTMENT

ANKARA - TURKEY
 POST - GRADUATE COURSE
IN

SEDIMENT TRANSPORT TECHNOLOGY

JUNE 4 – 29, 2007

GENERAL DIRECTORATE OF STATE HYDRAULIC WORKS

TECHNICAL RESEARCH AND QUALITY CONTROL DEPARTMENT

ANKARA - TURKEY
Sediment Transport Technology

TABLE OF CONTENTS
Uniform Flow in Open Channels ...................................................................................... 1
Prof. Dr. Halil ÖNDER

Sediment Problems and Properties ................................................................................. 27

Prof. Dr. Melih YANMAZ & Dr. Ş. Yurdagül KUMCU

Erosion by Water and Sediment Yield of a Watershed.................................................... 39

Prof. Dr. İbrahim GÜRER

Measurement of Sediment Load ...................................................................................... 65

Dr. Ş. Yurdagül KUMCU & Assoc. Prof. Dr. M. Ali KÖKPINAR

Flow Resistance in Streams............................................................................................. 89

Dr. Fuat ŞENTÜRK (edited by: Assoc. Prof. Dr. M. Ali KÖKPINAR)

Initiation of Particle Motion ......................................................................................... 143

Prof. Dr. Mustafa GÖĞÜŞ

Water Wave Mechanics ................................................................................................. 159

Assoc. Prof. Dr. Can Elmar BALAS

Mechanism of Sediment Transport ................................................................................ 167

Assoc. Prof. Dr. Şevket ÇOKGÖR

Sediment Transport Rate ............................................................................................... 179

Assoc. Prof. Dr. Şevket ÇOKGÖR

Probabilistic Approach to Sediment Transport ............................................................ 209

Prof. Dr. Beyhan YEĞEN

Numerical Modelling..................................................................................................... 217

Assoc. Prof. Dr. H. Kerem CIĞIZOĞLU

Estimation and Forecasting of Daily Suspended Sediment Data by Multi

Layer Perceptrons ......................................................................................................... 241
Assoc. Prof. Dr. H. Kerem CIĞIZOĞLU
 Table of Contents

Coastal Hydrodynamics and Sediment Transport ........................................................ 257

Prof. Dr. Yalçın YÜKSEL & Prof. Dr. Esin ÇEVİK

Exercises on Uniform Flow in Open Channel, Bed Load, Suspended

Load and Total Load..................................................................................................... 295
Dr. Vehbi ÖZAYDIN

Reservoir Sedimentation ............................................................................................... 353

Prof. Dr. Mustafa GÖĞÜŞ

Design of Stable Channels ............................................................................................ 379

Prof. Dr. A. Melih YANMAZ

Scour at Bridge Sites..................................................................................................... 395

Prof. Dr. A. Melih YANMAZ

Sediment Control Structures for Headworks and Intakes ............................................ 415

Prof. Dr. Nevzat YILDIRIM

Design of Intakes and Settling Basins........................................................................... 467

Prof. Dr. Nevzat YILDIRIM

River Training Structures – Applications ..................................................................... 479

Prof. Dr. İbrahim GÜRER

Hydraulic Transport of Solids in Pipes ........................................................................ 515

Assoc. Prof. Dr. M. Ali KÖKPINAR
 UNIFORM FLOW IN OPEN CHANNELS

Prof. Dr. Halil Önder

Department of Civil Engineering
Middle East Technical University
06531 Ankara -Turkey

1. Introduction
Open channel flow is one of the two modes of transporting flowing water from one
point to another. It is a flow that has a free surface on which the pressure is atmospheric.
The main driving force is the gravity force. The open channel flow is more difficult to
deal with, as compared to pipe flow, due to the presence of free surface, various
configurations of the channel sections, and the changing position of the water surface
with respect to time and space.

1.1. Channel types

An open channel is a conduit in which water flows with a free surface. The channels are
classified under two main groups as natural and artificial channels. The stream and
rivers are examples of natural channels. The artificial channels include canals, flume,
chute, culvert, tunnel, ditch, partly full pipe, or an aqueduct of any shape. The channels
are also categorized as erodible channels and nonerodible channels.

1.2. Channel geometry

A channel having unvarying cross section and constant bed slope throughout its length
is known as prismatic channel. Otherwise the channel is non-prismatic.

The cross section of a channel taken normal to the direction of the flow is referred to as
channel section. A vertical channel section, however, is the vertical section passing
through the lowest point of the channel section.

Natural channel sections are, in general, very irregular in shape. Artificial channels, on
the other hand, are usually designed with section of regular geometric shapes.

1.3. Geometric elements of channel sections

The definitions of various geometric elements are given below:
1. Depth of flow, y: vertical distance from the channel bottom to the free surface
2. Depth of flow section, d: depth of flow normal to the direction of flow; d = ycosθ,
but the terms d and y are used interchangeably
3. Top width, T: width at the free surface
4. Flow area, A: cross-sectional area of the flow normal to the direction of flow; A= by
for a rectangle

1
Sediment Transport Technology

5. Wetted perimeter, P: across a channel section, the length of the channel surface in
contact with water; P = b+2y for a rectangle
6. Hydraulic radius, Rh : ratio of the flow area to the wetted perimeter; Rh = A/P
7. Hydraulic depth, Dh : ratio of the flow area to the top width; Dh = A/T
8. Section factor for uniform flow, Z0 = AR2/3

1.4. Classification of open channel flows

1. Depth Criteria
a) Time Criterion
⎛ ∂y ⎞
i) Steady Flow ⎜ = 0 ⎟
⎝ ∂t ⎠
⎛ ∂y ⎞
ii) Unsteady Flow ⎜ ≠ 0 ⎟
⎝ ∂t ⎠
b) Space Criterion
⎛ ∂y ⎞
i) Uniform Flow ⎜ = 0 ⎟ (depth of flow same at every section practically steady)
⎝ ∂x ⎠
⎛ ∂y ⎞
ii) Non Uniform (Varied) Flow ⎜ ≠ 0 ⎟
⎝ ∂x ⎠
⎛ ∂y ⎞
(1) Gradually Varied Flow ⎜ → small ⎟
⎝ ∂x ⎠
⎛ ∂y ⎞
(2) Rapidly Varied (local phenomenon) Flow ⎜ → large ⎟
⎝ ∂x ⎠
2. Discharge
a) Space criterion
⎛ ∂Q ⎞
i) Continuous Steady Flow ⎜ = 0⎟
⎝ ∂x ⎠
⎛ ∂Q ⎞
ii) Discontinuous (Spatially Varied) Flow (discharge is non uniform) ⎜ ≠ 0⎟
⎝ ∂x ⎠
3. Reynolds Number
⎛ 4R V ⎞
a) Laminar Flow ⎜ Re = h 〈 2000 ⎟
⎝ ν ⎠
⎛ 4 Rh L ⎞
b) Turbulent Flow ⎜ Re = 〉 ∼ 4000 ⎟
⎝ ν ⎠
4. Froude Number
(
a) Subcritical (Tranquil) Flow Fr = V / gDh 〈1 )
(
b) Critical Flow Fr = V / gDh = 1 )
(
c) Supercritical (Rapid) Flow Fr = V / gDh 〉1 )
5. Boundary Criterion
a) Flow in Channels of Fixed Boundary (Artificial)
b) Flow in Channels of Movable Boundary (Natural)

2
3
Figure 1. Geometric Elements of Channel Sections (Chow, 1959)
Uniform Flow in Open Channels
Sediment Transport Technology

2. Review of Governing Equations

Governing equations of open channel flow are obtained by applying the three basic laws
of mechanics. These laws are the law of conservation of mass, Newton’s second law of
motion, and the first law of thermodynamics. These physical laws are normally applied
to a discrete mass of substance, which is often referred to as a system. When this system
is water as a fluid, which is in continuous motion and deformation, the application
becomes not so easy. The alternative approach is to consider a control volume, which is
nondeformable. In this case Reynolds Transport Theorem (RTT) is used to apply the
laws to a fluid flowing continuously through a control volume, and it is given by

DB ∂
Dt ∂t ∫∫∫
= bρ ⋅ d∀ + ∫∫bρ V . d A (1)
∼ ∼
c.v c.s

1 2 3
where B is an extensive property and b is corresponding intensive property. In words,
the Reynolds transport theorem states that the total rate of change of an extensive
property of a fluid (1) is equal to the rate of change of extensive property stored in the
control volume (2), plus the net outflow of extensive property through the control
surface (3). The inflows are always considered negative and outflows positive.

In the application of Reynolds Transport Theorem (RTT), to derive continuity,

momentum, and energy equations, the control volume may be finite or infinitesimally
small. The resulting governing equation, then, will be either integral equations or
differential equations.

2.1. Governing equations using integral approach

2.1.1. Continuity equation
The law of conservation of mass states that the mass, m, of a system remains constant.
The extensive property considered in RTT is the mass, then

B
B = m, b = =1
m

The law of conservation of mass becomes

DB Dm
= =0
Dt Dt

Substituting these values in RTT gives

∂
∂t ∫∫∫
ρ ⋅ d∀ + ∫∫ ρ V . d A = 0 (2)
∼ ∼
c .v c. s

4
 Uniform Flow in Open Channels

2.1.2. Momentum equation

Newton’s second law of motion states that, the time rate of change of linear momentum
is equal to the net force applied in a given direction. That is

D B / Dt = D (mV ) / Dt = ∑ F

The extensive property is B = mV, and the intensive property is b = B / m = V .

Substituting into the Reynolds transport theorem results in

∂
∑ F = ∂t ∫∫∫V ρ ⋅ d∀ + ∫∫V ρ V dA
∼ ∼ ∼ ∼ ∼
(3)
c .v c.s

2.1.3. Energy equation

The first law of thermodynamic states that the rate of heat added to the system plus the
rate of work done on the system is equal to the rate of increase of energy stored in the
system. This law may be written as

δQ δW DE
+ = (4)
dt dt Dt

where Q is the heat added to the system, W is the work done on the system, and E is the
stored energy in the system. In this equation δ( ) donates an increment of a non-
property and D( ) denotes an increment of a property. A property is any observable
characteristics of the system. According to this definition energy is a property.

The extensive property in the above equation is B = E and the corresponding intensive
property is b = e. The Reynolds Transport Theorem will give

δQ δWshaft ⎛p ⎞ ∂
+ = ∫∫ ⎜⎜ + e ⎟⎟ ρVdA + ∫∫∫ eρd∀ (5)
dt dt c.s ⎝
ρ ⎠ ∂t c.v

where, by a suitable choice of the control-volume boundaries, the term for work of shear
stress is eliminated, and the work of normal stress is combined with the stored energy.

2.2. Governing equation using differential approach

2.2.1. Continuity equation
Recall Eq. 2:

∂
∂t ∫∫∫
ρ ⋅ d∀ + ∫∫ ρ V . d A = 0
∼ ∼
c .v c. s

Green-Gauss theorem states that, for any volume in a vector field F bounded by a
~
closed surface A, the normal component ( F . n ) integrated over A is equal to the
~ ~
divergence ( ∇ . F ) integrated over ∀.
~ ~

5
Sediment Transport Technology

∫∫ (F . n)dA = ∫∫∫ ∇ . F d∀
c.s
~ ~
c.v
~ ~

Using this theorem, the surface integral in Eq. 2 is converted to a volume integral as

∫∫ ρ V . dA = ∫∫∫ ∇.ρ V d∀
c.s
∼ ∼
c .v
∼ ∼
(6)

Since the control volume is independent of the time, the order of differentiation and the
order of integration are interchangeable. The first term in Eq. 2 may then be written as

∂ ∂
∂t ∫∫∫ ρdV = ∫∫∫ ∂t (ρd∀)
c .v c .v
(7)

Inserting Eqs. 6 and 7 into Eq. 2 yields

⎡∂ ⎤
∫∫∫ ⎢⎣ ∂t (ρ ) + ∇⋅ ( ρ V )⎥⎦ d∀ = 0
c .v
∼ ∼
(8)

Since the volume is arbitrary, the integrand must vanish for all volumes, for the
equation to be valid.

∂ρ
+ ∇⋅ ρ V = 0
∂t ∼ ∼

or in indicial notation

∂ρ ∂ (ρVi )
+ = 0, i = 1,2,3 (10)
∂t ∂xi

2.2.2. Linear momentum equation

Linear momentum equation (Eq.3) in ith direction can also be written as

∂
∂t ∫∫∫
Vi ρd∀ + ∫∫ Vi (ρ V ⋅ n )dA = ∫∫∫ f bi d∀ + ∫∫ f si .dA (11)
∼ ∼
c.v c.s c.v c.s

where Vi, fbi, and fsi are respectively the components of velocity vector, body force vector
per unit volume and the surface force per unit area acting on the control surface. Thus
the right-hand side is the net force applied. Noting that fs = n·σ, where σ is the total
stress (Alexandrou, 2001) and using Green’s theorem, the last term in Eq. 11 may be
expressed as a volume integral as

∫∫ f si . dA = ∫∫ (n ⋅ σ ) i . dA = ∫∫∫ ∇ σ d∀
∼ ∼
( )
∼ i
(12)
c.s c.s

The first term in Eq. 11 may be written as, (considering that the order of integration and
the order of differentiation are interchangeable, since ∀ is independent of time)

6
 Uniform Flow in Open Channels

∂ ⎡∂ ⎤
∫∫∫ Vi ρd∀ = ∫∫∫ ⎢ (Vi ρ )⎥ d∀ (13)
∂t c .v ⎣
∂t ⎦

Inserting Eq. 12 and 13 into Eq. 11 yields

∂ρVi
+ ∇(Vi ρ V ) = f bi + (∇ .σ ) i (14)
∂t ∼ ∼ ∼ ∼

Eq. 14 is known as Cauchy stress equation. To relate the stress tensor to velocity,
constitutive equations of the fluid, as auxiliary equations, are used

⎡ ∂Vi ∂V j ⎤
σ i , j = − pδ i , j + µ ⎢ + ⎥ (15)
⎣⎢ ∂x j ∂xi ⎦⎥

When Eq. 15 is inserted into Eq. 14, and conservation of mass equation (Eq. 10) is used
the following is obtained.

DV
ρ ∼
= f b − ∇ p + µ∇ 2 V (16)
Dt ∼ ~
∼

Eq. 16 is well known Navier-Stokes Equation for an incompressible flow. Unlike the
Eq. 5, this equation is a partial differential equation, and it must be associated with
appropriate initial and boundary conditions.

2.2.3. Energy equation

The total energy of a parcel of fluid traveling at a constant speed on a streamline is
equal to the sum of the elevation of the parcel above a datum, the pressure head, and the
velocity head. The one-dimensional equation which quantifies this statement is known
as the Bernoulli energy equation or

u A2
H = z A + d A cos θ + (17a)
2g

where subscript A refers to a point on streamline in open-channel flow, zA the elevation

of point A above an arbitrary datum, dA the depth of flow section, θ the slope angle of
channel, and uA the velocity at point A.

Assuming that the velocity profile on the cross section is uniform, for small values of θ
Eq. (17a) reduces to

V2
H = z+ y+ (17a)
2g

where V; spatially averaged velocity of flow, and y; depth of flow.

7
Sediment Transport Technology

3. Steady Uniform Flow in Open Channels

A flow in an open channel is qualified as uniform flow when:
1. The depth, flow area, velocity, and discharge at every section of the channel reach
are constant (Fig.2)
2. The energy grade line, water surface, and channel bottom are all parallel, that is
their slopes are equal, or, Sf = Sw = S0, where Sf = slope of the energy grade line, Sw
= slope of the water surface, and S0 = slope of the channel bed.

For practical reasons, the requirement of constant velocity is often interpreted as

constant mean velocity.

A stable uniform flow can occur when boundary layer is fully developed. Boundary
layer development in open channels is shown in Fig. 3.

3.1. Uniform flow establishment

Establishment of uniform flow in a long channel is presented in Fig. 4.

3.1.1. Continuity equation

For a steady flow of an incompressible fluid, Eq. 2 becomes

Q1 = A1V1 = Q2 = A2V2 (18)

3.1.2. Energy equation

The energy equation is written for sections 1 and 2 as (Chow et al. ,1988):

V12 V22
z1 + y1 + = z2 + y2 + + hf (19a)
2g 2g

Figure 2. Steady uniform flow in an open channel flow

8
 Uniform Flow in Open Channels

Figure 3. Boundary layer development in open channels

Figure 4. Establishment of uniform flow in a long channel

9
Sediment Transport Technology

with V1 = V2, and y1 = y2, Energy Equation (Eq.19a) reduces to

hf = z1-z2 (19b)

Dividing both sides by L, the length of the channel reach, the following is obtained.

hf z1 − z 2
= (20)
L L

z1 − z 2 hf
If we introduce the bed slope S 0 = sin θ = and the friction slope S f = ,
L L
then the Eq. 20 becomes

S0 = S f (21)

Note that, in the above equation, S0 represents the rate of decrease of stored energy,
which consists, in this case, in only potential energy, while Sf represents the energy loss
resulting from the friction between the flowing fluid and the channel wall. In other
words there is a balance between the loss of potential energy and the energy dissipation
through boundary-friction resistance.

3.1.3. Momentum equation

The forces acting on the fluid contained within the control volume are gravity, pressure
and friction forces. The first of these three forces is a body force, while the other two
are surface forces. For a uniform flow the pressure forces at the two ends of the channel
reach are equal and cancel each other (since y1=y2). Again because of uniform flow the
momentum influx and outflux are equal and cancel each other. Then momentum
equation becomes simply

Fg - Ff = 0 (22)

The gravity force, which is the weight of fluid in the control volume, is γALsinθ, where
γ is the specific weight of the fluid, A is the cross-sectional area of the channel section,
and L is length of the reach. The friction force Ff is equal to the product of the wall
shear stress τ0 and the area over which it acts P·L, where P is the wetted perimeter of the
cross section. Inserting this information into Eq. 22 yields

γAL sin θ − τ o PL = 0 (23)

Rearranging results in

γALS o
τo = = γR H S 0 (24)
PL

where RH = A/P is the hydraulic radius. Since for a steady uniform flow S0 = Sf

10
 Uniform Flow in Open Channels

τ 0 = γRH S f (25)

It is to be noted that Eq. 25 expresses a linkage between the momentum and energy
principles. From the momentum viewpoint the effects of friction are represented as the
wall shear stress τ0, while from the energy viewpoint as a rate of energy dissipation Sf.

3.2. Head loss equations

There are three equations widely used for the relationship between velocity and the head
loss:

3.2.1. Darcy-Weisbach equation

In the case of non circular conduits, the Darcy-Weisbach Equation may be written as

L V2
hf = f (25)
4 Rh 2 g

where f; friction factor, Rh; hydraulic radius, L; channel length, V; average velocity, and,
hf; head loss.

The friction factor, f, is a function of Reynolds Number, Re, and relative roughness ε,
(ε = k s 4 Rh ), where ks is the roughness of the channel wall, and Rh is the hydraulic
radius.

Using the definition of the friction slope Sf = hf/L, Eq. 25 can be solved for V

8g
V = Rh S f (26)
f

3.2.2. Chezy’s equation

Chezy’s Equation is commonly written as

V = C Rh S f (27)

where C is called Chezy’s coefficient.

3.2.3. Manning’s equation

1 2 / 3 1/ 2
V= Rh S f (28)
n

When Eq. 28 is written in a similar form as Eq. 26 and Eq. 27

RH1 / 6
V= Rh S f (29)
n

11
Sediment Transport Technology

3.2.4. Comparison of head loss equations

Comparison of Eqs. 26, 27, and 29 gives

8g Rh1 / 6
=C = (30)
f n

From which

f 1/ 6
n= Rh (31)
8g
8 g n2
f = (32)
Rh1/ 3

3.3. Resistance coefficient estimation

In using any of the three equations, namely, Chezy, Manning or Darcy-Weisbach
equations in practice, the fundamental difficulty is the accurate estimation of an
appropriate value of the resistance coefficients C, n, and f. It is often hypothesized that
C and n behave in a manner analogous to the Darcy Weisbach friction factor, f, used to
assess flow resistance in pipe flow. In that case, C and n are not constants, but depends
on the flow through Reynolds number. In this case a roughness coefficient f may be
estimated in a similar manner to that used for pipes, then C and n values may be
determined.

3.3.1. Factors affecting the Manning roughness coefficient

In most practical simulations the value of n is a function of many variables.
Consequently a qualitative knowledge of the factors on which n depends is necessary.
These factors may be listed as follows.

1. Surface roughness
2. Vegetation
3. Channel irregularity
4. Obstruction
5. Channel alignment
6. Sedimentation and scouring
7. Stage and discharge

3.3.2. Methods of estimation of n

1. Soil conservation service method
2. Table method
3. Photographic method
4. Determination from velocity measurement
5. Empirical methods

12
 Uniform Flow in Open Channels

4. Computation of Uniform Flow

4.1. Use of Manning equation
For a given geometry, if three of the four variables in Eq. 18 and 28 are known, then the
fourth variable can be determined. In general the determination of Q, n, S does not
present any difficulty. However, for a given geometry, when Q, S, and n are known, the
determination of the uniform depth, y0, is difficult and cumbersome. For this type of
problem an explicit solution of Eq. 28 is not possible. The following methods are used:

1. Trial and error method

2. Design charts
3. Numerical method

4.1.1. Trial and error method

For a given cross sectional geometry G, if Q, S, and n are known Eqs. 17 and 28 may be
rearranged as

Z 0 = AR 2 / 3 (33)
nQ
Z0 = (34)
S

Z0 in Eq. 33 is the section factor for uniform flow and it is a function of the channel
shape and the depth of flow. The trial and error solution involves the following
procedure. A table as shown below is constructed. The values of y0 are assumed, and
corresponding values of Z0 are computed. When the computed values of Z0 = AR2/3
matches the values of:

nQ
Z0 =
S

computed from known information, the correct value of y0 has been found.

nQ
Trial y (m) A (m2) P (m) R (m) Z0
S
y 00 Z 00
y 10 Z 01
nQ
y 0i Z 0i ←
S
y 0k Z 0k

A graph of the section factor, Z0, versus the depth of flow, y0 is constructed (Fig. 5). The
depth corresponding to Z 0 = nQ S is obtained from this graph.

13
Sediment Transport Technology

y0

y0

Z0=AR2/3
Qn
S

Figure 5. Variation of uniform flow depth, y0, with section factor, Z0

4.1.2. Design charts

In order to simplify the computation, dimensionless curves showing the relation
between depth and section factor have been prepared for rectangular, trapezoidal, and
circular channel sections (Fig. 6).

Figure 6. Dimensionless curves for determining the normal depth

14
 Uniform Flow in Open Channels

The procedure is as follows:

Qn Qn
1. Compute or
Sb 8/3
Sd 8/3
2. On the design chart, enter this value, and determine corresponding point on the
design curve.
3. The ordinate of this point gives y0/b, or y0/d from which y0 can be obtained.

4.1.3. Numerical methods

When a large number of uniform flow depth computations are required, the use of
numerical trial and error procedure is the best approach. The method makes use of a
computer. Such a method is described by French (1986). He gives an interactive
computer code that solves the uniform flow depth problem for rectangular, trapezoidal,
triangular, circular, and natural channels. A logic diagram is shown in Fig. 7.

In this interactive program, the user defines the type of channel and specifies the value
of section factor. The program has subroutines RECT, TRAP, CIRC and NATDAT.
These subroutines create and search an array of depths of flow and their corresponding
values of section factor until an interval, which includes the specified section factor, is
found. Then, this interval is subdivided and a cubic spline is fitted to the interval in
subroutine SPCOEF. The value of y0 corresponding to the user-specified value of the
section factor is estimated in subroutine SPLINE. The subroutines SPCDEF and
SPLINE are standard cubic spline interpolation subroutines.

In this program, triangular channels are treated as a special case of trapezoidal channels.
For natural channels, the cross-section is specified in terms of the distance from one
side of the channel and the height of the channel perimeter above some arbitrary datum.

4.2. Use of Darcy-Weisbach equation and Moody diagram

The validity of Manning’s equation requires the flow to be fully turbulent. If Darcy-
Weisbach equation is to be used, then, the friction factor f is independent of the
Reynolds number Re. Henderson (1966) gives the following criterion for fully turbulent
flow.

n 6 RH S f ≥ 1.1x10 −13 (35)

When the condition given in Eq. 35 is satisfied Manning Equation may be used.When
the flow is not fully turbulent, the flow velocity may be computed with the Darcy-
Weisbach equation. The friction factor f should be determined from Moody diagram
constructed by modifying the one for pipe flow. The diameter D is replaced by 4RH. The
Reynolds number is given by

4 RhV
Re = (36)
ν

The relative roughness ε is defined by

15
Sediment Transport Technology

ks
ε= (37)
4 Rh

where ks is the size of sand grains resulting in a surface resistance equivalent to that
observed in the channel.

Figure 7. Flow chart for an interactive computer solution of the normal depth

16
 Uniform Flow in Open Channels

The equations used in the construction of modified Moody diagram are given by Chow
(1959) and Henderson (1966). In the following, this equation will be summarized.

For laminar flow, when Reynolds number is less than 2000, the friction factor is

CL
f = (38)
Re

Comparing this equation with Eq. B7, it is seen that CL = 96, for a smooth-surfaced
channel of infinite width.

For turbulent flows, we distinguish three types of flow. hydraulically smooth turbulent
flow refers to a flow in which the perimeter roughness elements are completely covered
by the viscous sublayer. Fully rough flow occurs when the perimeter roughness
elements project through the laminar sublayer and dominate the flow behavior. In this
case the friction factor is independent of Reynolds number, Re. Between these two
extremes, there is a transitional region. The three types of turbulent flow are
differentiated from each other by a Reynolds number based on ks and u*.

This Reynolds number is defined by

u* k s
Re* = (39)
υ

in which ν is kinematic viscosity and u* is the shear velocity defined by

τ0
u* = (40)
ρ

The transition region is defined by the limits

4 ≤ Re* ≤ 100 (41)

1. Hydraulically smooth turbulent flows:

if Re < 105, Blasius formula applies

0.316
f = (42)
Re0.25

if Re > 105

1 ⎛R f ⎞
= 2.0 log ⎜ e ⎟ (43)
f ⎜ 2.51 ⎟
⎝ ⎠

17
Sediment Transport Technology

2. Transition region between hydraulically smooth and rough flows: A modified

Colebrook formula is used

1 ⎡ k 2.5 ⎤
= −2.0 log ⎢ s + ⎥ (44)
f ⎣⎢12 Rh Re f ⎥⎦

3. Fully turbulent flows

1 ⎛ k ⎞
= −2.0 log ⎜⎜ s ⎟⎟ (45)
f ⎝ 12 Rh ⎠

The friction factor f can be eliminated between Eq. 31 and Eq. 45, to solve for the
relative roughness ε as a function of Manning’s n and hydraulic radius Rh.

1 / 6
− R /( 4 n 2 g )
ε = 3 x 10 h
(46)

The Moody Diagram for open channel flow is given in Fig. 8. To use Moody Diagram,
an iterative process is utilized. For a given value of Manning’s roughness n, initial
values for RH and V are assumed and ε is calculated by using Eq. 46. Then the Reynolds
number is computed using Eq. 36 and the corresponding value of f read from Fig. 8. An
estimate of V is obtained from Eq. 26. This process is repeated iteratively until the
values for V converge.

Limitations of Moody diagram:

1. It accounts for resistance due to friction elements randomly distributed on the
channel wall, but it does not account for form drag associated with nonuniformities
of the channel.
2. It is valid only for fixed bed channels, not for erodible ones.

Because of these limitations, the Moody diagram should be used only for lined channels
with uniform cross sections.

4.3. Channel section with composite roughness

In most natural channels and in many artificial channels, the roughness along the
perimeter of the channel may vary from part to part. In such cases, an equivalent
roughness coefficient neq is calculated. Then the Manning equation is utilized using this
neq. There are a number of different methods of estimating the equivalent roughness
coefficient.

For the determination of the equivalent roughness in a natural channel, the water area is
divided into N parts of which the wetted perimeters p1, p2, …..pN and the coefficients of
roughness n1, n2… nN are known. In these methods the wetted perimeters, pi’s, do not
include the imaginary boundaries between the subsections.

18
19
Figure 8. Moody chart for open channel flow
Uniform Flow in Open Channels
Sediment Transport Technology

4.3.1. Horton and Einstein method

In this method it is assumed that the mean velocities in the individual subsections, as
well as the mean velocity in the whole section are same (V1 = V2 = ------ = VN = V)

( )
2/3
⎡ N ⎤
⎢ ∑ Pi ni
3/ 2
⎥
n = ⎢ i =1 ⎥
⎢ P ⎥
⎢⎣ ⎥⎦

4.3.2. Pavlovskii, Mühlhofer, and Einstein and Banks method

It is assumed that the total force resisting the flow is equal to the sum of the forces
resisting the flow in subsections

( )⎤⎥
1/ 2
⎡ N
⎢ ∑ Pi ni
2

n = ⎢ i =1 ⎥
⎢ P ⎥
⎢⎣ ⎥⎦

4.3.3. Lotter method

The assumption is that the total discharge of the flow is equal to the sum of the
discharges of the subsections

PR 5 / 3
ne = 5/3
N
Pi Ri
∑
i =1 ni

In the following methods, the boundary angles are bisected and the subdivision is
composed of the channel perimeter, the boundary angle bisectors, and the water surface.
These methods are used mostly for designed channels or laboratory flumes.

4.3.4. Cox method

N

∑n A i i
n= i =1

4.3.5. Colerbatch method

2/3
⎛ N ⎞
⎜ ∑ Ai ni3 / 2 ⎟
ne = ⎜ i =1 ⎟
⎜ A ⎟
⎜ ⎟
⎝ ⎠

20
 Uniform Flow in Open Channels

References

1. Alexandrou, A., “Principles of Fluid Mechanics”, Prentice Hall, New Jersey,

(2001).
2. French, R. H., “Open Channel Hydraulics”, McGraw-Hill, NewYork, (1986).
3. Chow, V.T., “Open-Channel Hydraulics”, McGraw-Hill, NewYork, (1959).
4. Chow, V.T., Maidment, D.R and Mays, L.W., “Applied Hydrology”, McGraw-Hill
New York, (1988).
5. Henderson, F. M., “Open Channel Flow”, Macmillan, New York, (1966).

Appendix A. Derivation of Head Loss Equations

Derivation of Darcy-Weisbach Equation

Eq. (24) may be rearranged as:

hf τ0
Sf = = (A1)
L γRh

Dimensional analysis gives:

τ 0 = f ′ρV 2 (A2)

Combining the two equations gives

fρ
′ V 2L
hf = (A3)
γRh

Further rearrangement of Eq. A3 gives

L V2
hf = 8 f ′ (A4)
4 Rh 2 g

Finally, if we let f = 8 f ′ , Eq. A4 becomes

L V2
hf = f (A5)
4 Rh 2 g

This equation is known as Darcy-Weisbach Equation.

Derivation of Chezy’s Equation

Eq. 24 and Eq. A2 are combined as

γRh S f = f ′ρV 2 (A6)

21
Sediment Transport Technology

When solved for V

γ
V = Rh S f (A7)
ρf ′

with f = 8 f ′ , and γ = ρg , Eq. A7 becomes

g
V= Rh S f (A8)
f′

g
Letting C = ,
f′

V = C Rh S f (A9)

Appendix B. Uniform Laminar Flow

A definition sketch for uniform flow is shown in Fig. 2. Momentum equation may be
applied by considering a section of width b normal to the flow; the flow cross section is
thus a rectangle of depth y0 and width b. In this case the wetted perimeter (i.e. the
boundary offering resistance to the flow) is only the channel bottom, of width b.
Therefore

A by 0
Rh = = = y0
P b

Inserting this in Eq. 23

τ 0 = γy 0 S 0 (B1)

Stress distribution may be expressed as:

⎛ y ⎞ ⎛ y ⎞
τ = τ 0 ⎜⎜1 − ⎟⎟ = γy 0 S 0 ⎜⎜1 − ⎟⎟ = γS 0 ( y 0 − y ) (B2)
⎝ y0 ⎠ ⎝ y0 ⎠

In the laminar flow τ is also given by:

dv
τ =µ (B3)
dy

Inserting Eq. B3 into Eq. B2 gives

dv
µ = γS 0 ( y 0 − y )
dy

22
 Uniform Flow in Open Channels

γS 0 ⎛ y2 ⎞
v= ⎜⎜ y 0 y − ⎟+c
µ ⎝ 2 ⎟⎠

y = 0, v = 0, c = 0

γS ⎛ y2 ⎞
v= 0 ⎜⎜ y 0 y − ⎟⎟ (B4)
µ ⎝ 2 ⎠

Average velocity may be calculated as:

y0
∫
y0
vdy γS y0 ⎛ y2 ⎞ γS ⎡ y 2 y 3 ⎤
V = = 0 ∫ ⎜⎜ y 0 y − ⎟⎟dy = 0 ⎢ y 0 − ⎥
0

yo µy 0 0
⎝ 2 ⎠ µy 0 ⎣ 2 6 ⎦0
γS 0 y 0
V = (B5)
3µ

Replacing S0 by hf/L gives

γy 0 h f
V =
3µL
3µL 3µL 32 V
hf = V= V
γy 0 ρgy 0 32 V
96µL V 2 96 L V2
hf = =
16 ρy 0V 2 g ρV (4 y 0 ) 4 y 0 2 g
96 L V 2
hf = (B6)
Re D 2 g

This last equation shows that the friction factor, f, for a laminar flow is

96
f = (B7)
Re

Appendix C. Energy and Momentum Coefficients

In many open-channel problems it is both convenient and appropriate to use the
continuity, momentum, and energy equation in a one-dimensional form in which the
velocity is assumed to be uniformly distributed on the cross section. In such a case, the
laws of conservation are:

1. Conservation of mass (continuity)

Q = VA

23
Sediment Transport Technology

2. Conservation of momentum

∑ F = ρQ(V 2 − V1 )

3. Conservation of energy

H1 = H2 + Hl

where:

V2
H = z+ y+
2g

Of course, no real flow is one dimensional; therefore, the true transfer of momentum
through a cross section

∫ ∫ ρu dA
2

is not necessarily equal to the spatially averaged transfer ρQV

Thus, in situations where the velocity profile varies significantly in the vertical and/or
transverse directions, it may be necessary to define a momentum correction coefficient,
or

βρ QV = ∫ ∫ ρu 2 dA
A

and

∫ ∫ ρu dA ∫ ∫ ρu dA
2 2

β= A
= A
(C1)
ρQV ρV 2 A

where β is the momentum correction coefficient. In an analogous fashion, the kinetic

energy coefficient is

⎛ V3 ⎞ u3
α ⎜⎜ γ ⎟⎟ = ∫ ∫ γ dA
⎝ 2g ⎠ A
2g

and solving for α gives:

∫ ∫ γu dA
3

α A
(C2)
γV 3 A

24
 Uniform Flow in Open Channels

where α is the kinetic energy correction coefficient. The following properties of β and α
are noted:
1. They are both equal to unity when the flow is uniform. In all other case β and α
must be greater than unity.
2. A comparison of Eqs. C1 and C2 demonstrates that for a given channel section and
velocity distribution α is much more sensitive to the variation in velocity than β.
3. In open-channel hydraulics β and α are generally used only when the channel
consists of a main channel with subchannels and/or berms and floodplains. In such
cases, the large variation in velocity from section to section effectively masks all
gradual variations in velocity, and it is appropriate to consider the velocity in each
of the subsections as constant. In channels of compound section, the value of α may
exceed 2.

Appendix D. Turbulent Velocity Distributions

Within a turbulent boundary layer, Prandtl (Schlichting, 1968) demonstrated that the
vertical velocity profile is approximately logarithmic. Prandtl claimed that the shear
stress at any point in the flow is given by.
2
⎛ du ⎞
τ = ρl ⎜⎜ ⎟⎟
2
(D1)
⎝ dy ⎠

or

τ
du = dy (D2)
ρl 2

where l is termed the mixing length. It is usually assumed that

l = k·y (D3)

where k was originally called von Karman’s turbulence constant; however, it is perhaps
more accurate to term k a coefficient since there is some evidence that k may vary over
a range of values as a function of the Reynolds number. If it is assumed that k can be
approximated as 0.4, then substitution of Eq. D3 in Eq.D2 yields, after integration,

τ0 y
u = 2.5 ln (D4)
ρ y0

or

y
u = 2.5u * ln (D5)
y0

where it is assumed that τ = τo; the shear stress on the bottom boundary, u; turbulent
average velocity at a distance y above the bottom.

25
26
 SEDIMENT PROBLEMS AND PROPERTIES

Prof. Dr. Melih Yanmaz

Civil Engineering Department
Middle East Technical University
Ankara-Turkey

Dr. Ş. Yurdagül KUMCU

Hydraulic Model Laboratory
Technical Research and Quality Control Department
State Hydraulic Works, Ankara-Turkey

1. General
Mechanical weathering of the land surface by water, wind, ice, and gravity causes the
formation of disintegrated particles, called sediment or alluvium. The process of
erosion, transportation and accumulation of these particles is called sediment
transportation. Although the sediment transportation comprise the movement of non-
cohesive or granular, cohesive, and rocky material, the term sediment is commonly used
for aggregates varying from silt to fine gravel for most of the civil engineering
applications. Computation of sediment transportation is often required in the design and
operation of various hydrosystems. Therefore, the understanding and formulation of the
sediment transport phenomenon is an important issue within the fields of hydraulic
engineering. This chapter deals with the interpretation of common problems
encountered in this phenomenon and discusses the properties of sediment.

2. Problems Related to Sediment Transportation

Engineers who are in charge of design and operation of various hydrosystems can face
with problems related to sediment transportation. Some of the major problems are listed
as:
1. Problems of erosion
2. Sediment load computations
3. Reservoir sedimentation
4. Aggradation, degradation, and local scour
5. Morphological computations

2.1. Problems of erosion

The sediment load carried by streams comes from various sources. In the hilly areas the
streams pick up coarse material. Land slides also contribute to the load carried by
streams. However, the major portion of the sediment load carried by streams comes
from the erosion of material in the drainage basin. The sediment load transported by
streams varies from very coarse material to very fine material. The amount of the
sediment load carried depends on the size of material, discharge, slope and channel
characteristic (Garde and Ranga Raju, 1977).

27
Sediment Transport Technology

Erosion of the land surface basically may be divided into two groups. Geologic erosion
is defined as the erosion of the surface of the earth under natural conditions, such as
flowing water, wind, ice, and gravity. Geologic erosion is a long-term event and is
normally of uncontrollable nature. Accelerated erosion is the increased rate of erosion
due to human activity, such as plowing the soil, cutting the protective vegetative cover,
and changing the flow conditions from land surfaces. Accelerated erosion takes place in
much shorter time as compared to the geologic erosion. It is estimated that the total rate
of annual sediment transport in Turkey is 5*109 kN that is approximately 18 times the
total erosion from whole Europe. Other factors that increase the rate of erosion are
various urbanizational activities, highway construction, mining operations, etc.
Implementation of effective soil and water conservation practices tend to reduce the
accelerated erosion.

2.2. Sediment load computations

Flowing water over loose boundaries, which are composed of mainly non-cohesive
material, has erosive effect along the channel perimeter. Much research has been carried
out to express the movement of the sediment load. However, almost all the semi-
empirical and empirical equations are based on laboratory experiments performed under
uniform flow conditions with uniform bed material, which in fact is contrary to the
prototype conditions. Besides, the results of the equations reported in the literature are
quite different from each other when they are applied to a specific case. Actually, each
equation is assumed to be valid under its derivational conditions. Application of any
equation beyond the validity of its experimental conditions may produce appreciable
errors, sometimes more than 100%. A civil engineer is then confronted with the
selection of a suitable equation for his/her study. Suitability of any specific equation
needs to be verified with precise field observations.

2.3. Reservoir sedimentation

Rivers carry some suspended sediment and move the larger solids along the streambed
in the form of bed load. The velocity and turbulence intensity are greatly reduced in a
reservoir. Therefore, the larger suspended particles and most of the bed load are
deposited as a delta at the head of the reservoir. Smaller particles may remain in
suspension for a long time and even be discharged downstream through sluiceways.

Since reservoir sedimentation directly affects the lifetime of a dam, extensive surveys
have to be conducted in planning studies. Mechanics of reservoir sedimentation is
highly complicated due to random changes in sediment-laden inflow, variation of
particle size distribution of sediment, operational modes of reservoirs, and the geometric
characteristics of reservoirs. The percentage of the inflowing sediment, which is
retained in a reservoir (trap efficiency), is a function of the ratio of the reservoir
capacity to total inflow. The rate of sediment accumulation in a reservoir is based on the
intensity of turbulence, which may change according to the size of the reservoir. The
trap efficiency of a reservoir decreases with age as the reservoir capacity is reduced by
sediment accumulation. Thus, complete filling of the reservoir may require a very long
time, but actually the useful life of the reservoir is terminated when the capacity
occupied by sediment accumulation is sufficient to prevent the reservoir from serving
its intended purpose.

28
 Sediment Problems and Properties

It is very difficult to predict the sediment accumulation as it fluctuates from almost zero
during low flow conditions to extremely large quantities of sediment transportation
during major floods. A continuous hydrologic simulation model can best perform
estimation of reservoir sedimentation. For that purpose, daily sediment samples should
be collected for two or three years to provide data to calibrate the simulation model.
Empirical methods, such as the area reduction method or the area increment method
(USBR, 1987) are used to estimate the sediment distribution in a reservoir. The
dimensionless profiles of reservoirs may provide valuable information about the
location of a delta and the distribution of sediment in the reservoir. Use of such
information may also aid the functional design and operation of outlet facilities.
Formation of a delta may cause the raising of backwater elevations in the upstream,
which increases the flooding potential (Yanmaz, 2001).

2.4. Aggradation, degradation, and local scour

If a certain length of an alluvial river is said to be in regime or in dynamic equilibrium,
it means that the amounts of sediment entering and leaving this reach are equal to each
other. Hence, the riverbed elevation will not change over a long period of time.
However, if entering and leaving sediment loads are different, the bed level must either
rise or fall. A rise in bed level is known as aggradation, while a fall in bed level is
termed as degradation or retrogression. Degradation around any hydraulic structure due
to the accelerations in flow is known as local scour around that structure, e.g.,
downstream of an outlet structure, around bridge piers, and abutments, etc.

2.5. Morphological computations

Morphologic predictions comprise determination of long-term variation of plan form
and bed profile of rivers. Within this context, civil engineers and geological engineers
are involved in morphological computations. As the design and operation of all types of
river training facilities and hydrosystems are the main concerns of civil engineers, the
morphological changes during the physical lifetime of these structures should be
estimated. Therefore, it is clear that the time interval required for morphological
computations of hydrosystems are much smaller than required foe geologic predictions.
Factors affecting river behavior can be listed as river discharge, sediment load,
longitudinal slope, bank and bed resistance, vegetation, geology, and human activities.
Morphological modeling of a river is extremely complicated since these factors are of
random character. An overall solution to a river engineering problem is not possible
because of the complexity of the whole process. However, partial solution can be
achieved with some simplification. For the calibration of models dealing with
morphological predictions, continuous monitoring and measurement are required.

3. Properties of Sediment
Sediment transport computations are based on the identification of various properties of
sediment. Within this context, these properties can be categorized according to the
features of a solitary particle or particles in a group. The common properties of
individual particles are size, shape, fall velocity, mineral composition, surface texture
and orientation, whereas the bulk properties comprise particle size distribution, specific
weight, porosity and angle of repose. In the followings, a brief discussion is introduced
for these properties.

29
Sediment Transport Technology

3.1. Properties of individual particles

3.1.1. Size
Particles forming beds and banks of natural rivers vary in size from colloids to boulders.
Sediment size, D, is an important factor involved in sediment transport rate
computations. The grade scale classification proposed by the American Geophysical
Union is presented in Table 1. Since sediment particles have numerous shapes, various
ways are used to express their sizes. Following definitions can be used to define particle
size:
a) Nominal diameter is the diameter of a sphere of the same volume as the given
particle.
b) Sieve diameter is the length of the side of a square sieve opening through which
the given particle will just pass.
c) Sedimentation diameter is the diameter of a sphere of the same specific weight
and the same terminal settling velocity in the same sedimentation fluid at the
same temperature as the given particle.
d) Fall diameter is the diameter of a sphere having the same specific gravity and
standard fall velocity as that of the particle.
e) Triaxial size is the arithmetic mean of three lengths measured along three
mutually perpendicular axes whose origin is located at the centroid of the
particle:

a+b+c
D= (1)
3

where a, b, and c are the lengths of the particle measured along the major, intermediate,
and minor axes, respectively. Sieve size is generally used to characterize the size of
granular particles for the sake of practical ease. The size of silts and clays is generally
expressed as a sedimentation diameter and determined by sedimentation methods
because of convenience. Although the sedimentation diameter is actually a fictitious
size, its physical significance is based on the fact that it enables the calculation of the
settling velocity. The nominal diameter is of minor importance in sediment transport
rate computations, but it may be useful in discussing the nature of sedimentary deposits.

Table 1. Sediment grade scale according to the American Geophysical Union (Lane, 1947)
Class D (mm) Class D (mm)
Very large boulders 4000-2000 Coarse sand 1-0.5
Large boulders 2000-1000 Medium sand 0.5-0.25
Medium boulders 1000-500 Fine sand 0.25-0.125
Small boulders 500-250 Very fine sand 0.125-0.062
Large cobbles 250-130 Coarse silt 0.062-0.031
Small cobbles 130-64 Medium silt 0.031-0.016
Very coarse gravel 64-32 Fine silt 0.016-0.008
Coarse gravel 32-16 Very fine silt 0.008-0.004
Medium gravel 16-8 Coarse clay 0.004-0.002
Fine gravel 8-4 Medium clay 0.002-0.001
Very fine gravel 4-2 Fine clay 0.001-0.0005
Very coarse sand 2-1 Very fine clay 0.0005-0.00024

30
 Sediment Problems and Properties

3.1.2. Shape
The shape of particles influences the bed resistance and the mean velocity close to the
bed. Particle shape can be expressed according to the classification given in Table 2 in
which Vp is the volume of the particle, Ap is the surface area of the particle, D is the size
of the particle, Vc is the volume of circumscribing sphere, r1 is the smallest radius of
curvature in the projected area, Apr is the projected area of the particle, and Aa is the area
of circle of diameter a.

Sphericity is related to the shape of the particle as compared to the sphere. Roundness is
related to the sharpness of various corners and edges of the particle. The maximum
values of the shape factor, modulus of flatness, and modulus of length are unity. Shape
factor of a spherical particle is unity, whereas the natural quartz sand has a shape factor
of approximately 0.7. Flatness ratio is equal to or greater than unity and varies from
1.05 to approximately 10 for natural sediment.

Selection of a representative parameter describing the particle shape is based on the

purpose of the study. In hydraulic engineering applications, such as fall velocity
computations, shape factor is the most suitable parameter (Garde and Ranga Raju,
1978).

Table 2. Particle shape definitions (Durand, 1951)

Coefficient based on Parameter Expression No

Vp
Volume constant VC = (2)
D3
Ap
Volume Surface constant SC = (3)
D2
1/ 3
⎛Vp ⎞
Sphericity S = ⎜⎜ ⎟⎟ (4)
⎝ Vc ⎠
Coefficient of r
Cr = 3 1 (5)
roundness abc
Projected area
Coefficient of Apr
Cc = (6)
circularity Aa
c
Shape factor s. f . = (7)
ab
a+b
Flatness ratio F= (8)
2c
Three axes
c
Modulus of flatness Mf = (9)
b
a
Modulus of length Ml = (10)
b

31
Sediment Transport Technology

3.1.3. Fall velocity

When a sediment grain drops into calm water with no initial velocity, its velocity
increases with depth and reaches a maximum or terminal constant value that is known
as the fall velocity, Wf. It takes place when the submerged weight is in equilibrium with
the drag force exerted by water on the particle. The fall velocity of a spherical particle
can be determined from the force equality as

4 ∆gD
Wf = (11)
3 CD

in which ∆=(ρs-ρ)/ρ, ρs is the sediment density, ρ is the density of water, D is the

diameter of the spherical particle, and CD is the drag coefficient. The particle Reynolds
number is defined as Re=WfD/ν, where ν is the kinematic viscosity of water. Solution of
Equation (11) is subject to a trial and error procedure since CD is a function of Re, which
is expressed in terms of unknown Wf. For small Reynolds numbers, i.e. Re<0.1, laminar
flow conditions prevail around the particle during falling. For large Reynolds numbers,
i.e. Re>2.1, turbulent flow develops around the descending particle. Transition flow
occurs between these limiting values of the Reynolds number. For laminar flow, also
known as the Stoke range, the value of the drag coefficient is approximated to
CD=24/Re. Drag coefficient tends to a constant value of approximately 0.40 for fully
developed turbulent flow conditions. For the transitional regime,

24 3
CD = + + 0.34 (12)
Re Re

Fall velocity of quartz sand in water at 20°C can be computed from (Breusers and
Raudkivi, 1991)

W f = 663D 2 for D<0.15 mm (13)

W f = 134.5 D for D>1.5 mm (14)

where Wf is in mm/s and D is in mm. Variation of fall velocity for the range 0.15
mm≤D≤1.5 mm is presented in Table 3.

The fall velocity of a particle depends on its diameter, shape, orientation during falling,
boundary proximity, turbulence intensity, temperature, and sediment concentration. In
the following, the effects of these factors will be explained briefly.

Table 3. Fall velocities of quartz sand for 0.15 mm≤D≤1.5 mm at 20°C

D (mm) 0.15 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.20 1.50

Wf (mm/s) 14.8 21.1 36.1 50.0 64.0 76.4 88.6 99.0 110 121 137.3 166

32
 Sediment Problems and Properties

For a given Reynolds number, the drag coefficient decreases with increasing shape
factor, i.e. a spherical particle has the smallest drag coefficient as compared to natural
grains. Therefore, the fall velocity of a granular particle having a high shape factor is
greater than that of an irregular particle. When a particle falls in a fluid, part of the fluid
around the particle moves in the opposite direction. When the diameter of the container,
d, is small, it exhibits more retarding effect on the falling particle. The actual fall
velocity, Wfa, influenced by side boundaries is obtained from (Garde and Ranga Raju,
1978).

Wf
W fa = for D/d<0.3 (15)
K1
Wf
W fa = for D/d≥0.3 (16)
K2
2
⎛ 9D ⎞ ⎛ 9D ⎞
K1 = 1 + ⎜ ⎟+⎜ ⎟ (17)
⎝ 4 d ⎠ ⎝ 4d ⎠
−2.5
⎛ D⎞
K 2 = 0.53π ⎜1 − ⎟ (18)
⎝ d⎠

There is a mutual interaction among the particles falling in a group. The presence of
suspended particles reduces the fall velocity of an isolated grain. The fall velocity of a
particle in suspension, Wfp, can be computed from

W fp = W f 0 (1 − C )
β
(19)

where C is the volumetric concentration of the suspended material, e.g. lt/lt. For
common natural grains with s.f.≈0.7, the values of β are 4.67 and 2.35 for D*<40 and
D*>≈800, respectively. For 40≤D*≤800,

β = 7.48 D*−0.129 (20)

where D* is a dimensionless particle size proposed by Van Rijn (1984) as

1/ 3
⎛ ∆gD 3 ⎞
D* = ⎜⎜ 2 ⎟⎟ (21)
⎝ ν ⎠

Since the kinematic viscosity of water is indirectly proportional with the temperature,
the fall velocity decreases with increasing temperature. Turbulence also reduces the fall
velocity. The actual fall velocity in turbulent flow, Wft, in m/s is

u
W ft = W f − 0.132 (22)
y

where u is the mean flow velocity in m/s, and y is the flow depth in m.

33
Sediment Transport Technology

Recently, Göğüş et at. (2001) studied the effect of particle shape on fall velocity of
angular particles. A new shape factor SF was defined from a transformed form of the
original particle (Equation. (23)). Furthermore, they verified that the fall velocity of any
irregular angular particle could be estimated using graphical curves obtained the
analysis of regular angular particles.

⎛ a + b ⎞⎛ a b c ⎞
SF = ⎜⎜ 1 1 ⎟⎟⎜ 1 1 1 ⎟ (23)
⎝ 2c1 ⎠⎝ ∀ ⎠

where a1, b1, and c1 are the maximum, intermediate and minimum dimensions of box-
shaped prism circumscribing a particle, respectively, and ∀ is volume of particle.

3.2. Bulk Properties of Particles

3.2.1. Particle size distribution
Particle size distribution of a granular material can be obtained from a sieve analysis.
For the fine material having sizes smaller than approximately 1/16 mm, sedimentation
methods are used, which are out of the scope of this text. A typical particle size
distribution is shown in Fig. 1. There are various ways to express a characteristic
particle size. Let pi be the fraction of the weight of each class divided by the total
weight of the sample. The mean diameter, Di, of each class, e.g. sand, can be found
from

Di = Di max Di min (24)

where Dimax and Dimin are the extreme values for each class. The following sizes can be
defined to express the gradation:

100
90
80
70
60
% Finer

50
40
30
20
10
0
0,01 0,1 1 10
D (mm)

Figure 1. Particle size distribution of a well graded material

34
 Sediment Problems and Properties

a) Arithmetic mean diameter, Dm:

n
Dm = 0.01∑ Di pi (25)
i =1

where n is the number of classes and pi is expressed as percentage.

b) Geometric mean diameter, Dg:

n
log D g = 0.01∑ pi log Di (26)
i =1

c) Median size, D50, is the particle size that corresponds to 50% finer in the particle size
distribution curve or cumulative frequency histogram of the distribution. The median
size is different from the mean size. They can only be equal to each other for a
symmetrical distribution.

d) Standard deviation, σ:
1/ 2
⎛ n ⎞
σ = ⎜ ∑ (Di − Dm )2 f i ⎟ (27)
⎝ i =1 ⎠

where fi is the relative frequency of occurrence of size Di in a population. For a large

number of classes, pi tends to fi. For data following normal distribution, σ can be
expressed in the form

σ = (D84.1 − D50 ) = (D50 − D15.9 ) (28)

If the data follow log-normal distribution, the geometric standard deviation, σg, is given
by

D84.1 D84.1 D
σg = = = 50 (29)
D15.9 D50 D15.9

The mean diameter for a log-normally distributed data is

Dm = D50 exp ⎢ (ln σ g ) ⎥

⎡1 2⎤
(30)
⎣2 ⎦

The geometric standard deviation, σg, is mostly used in hydraulic applications. For a
typical gravel river, the value of σg is about 3.5-4.0. The coefficient of uniformity, Cu, is
a key parameter to classify the particle size distribution as either uniform or well
graded. It is defined as

D60
Cu = (31)
D10

35
Sediment Transport Technology

For Cu>3.0, the particle size distribution is termed as well graded.

3.2.2. Porosity
Porosity, n, is defined as the ratio of the volume of voids, Vv, to the total volume, V, of
the sample

Vv
n= (32)
V

It is affected by the particle size, shape, and degree of compaction. Porosity of fine
aggregates is greater than that of coarse aggregates. The porosity of various types of
soils is given in Table 4.

Komura (1961) has obtained the following relationship for the porosity of
unconsolidated saturated material:

n = 0.245 + 0.0864 D50−0.21 (33)

in which D50 is in cm. Void ratio, e, is defined as the ratio of the volume of voids, Vv, to
the volume of solids, Vs

Vv
e= (34)
Vs

The relationship between porosity and void ratio is given by

e
n= (35)
1+ e

3.2.3. Specific weight

Dry specific weight or unit dry weight, γ, is the ratio of the dry weight, Wdry, of the
sample to the volume of sample, V, including empty spaces

Wdry
γ= (36)
V

Table 4. Porosity of various soils

Type of soil %n

Coarse sand 39-41

Medium sand 41-48
Fine sand 44-49
Fine sandy loam 50-54

36
 Sediment Problems and Properties

Specific weight of submerged material is obviously smaller than that of the dry
material. It increases with increasing particle size as presented in Table 5 in which γsub
is the submerged specific weight of the sediment. Specific weight of sediment
accumulated in a reservoir increases with time because of consolidation. This increase is
more pronounced for fine material as compared to the case of coarse material.

3.2.4. Angle of repose

Angle of repose, φ, is an important parameter, which is required in hydraulic design
concerning slope stability, e.g. stable channel design. It increases with increasing
particle size (See Table 6).

Table 5. Specific weight of various types of soils (Garde and Ranga Raju, 1978)

γsub γdry
Type of soil
(kN/m3) (kN/m3)

clay 6.4-9.6 9.6-12.8

silt 8.8-12 12-13.6
clay-silt mixture 6.4-10.4 10.4-13.6
sand-silt mixture 12-15.2 15.2-17.6
clay-silt-sand mixture 8-12.8 12.8-16
sand 13.6-16 13.6-16
gravel 13.6-20 13.6-20
poorly sorted sand and gravel 15.2-20.8 15.2-20.8

Table 6. Values of angle of repose in degrees (Garde and Ranga Raju, 1978)

D50
Crushed ledge rock Very angular Very rounded
(mm)

0.25 32.0 31.4 29.2

1.25 34.5 32.9 29.5
2.50 36.6 33.8 29.9
12.7 40.0 37.5 32.5
25.4 40.8 39.1 34.8
127 42.0 41.2 38.3
254 42.2 41.5 39.2

References

1. Breusers, H.N.C., and Raudkivi, A.J., “Scouring - hydraulic structures design

manual 2”, International Association for Hydraulic Research, A. A. Bakema,
Rotterdam/Brookfield, (1991).
2. Durand, R., “Differntes Definitions Techniques de la Forme des Galets”, La Houille
Blanche, Sp. No. A (in French), (1951).

37
Sediment Transport Technology

3. Garde, R.J., and Ranga Raju, K.G., “Mechanics of Sediment Transportation and
Alluvial Stream Problems”,Wiley Eastern Limited, New Delhi, (1978).
4. Göğüş, M., İpekçi, O.N., and Kökpınar, M.A., “Effect of particle shape on fall
velocity of angular particles”, ASCE, J. of Hydr. Eng., Vol:127, No:10, pp. 860-
869.
5. Komura, S., “Bulk Properties of River bed Sediments and its Applications to
Sediment Hydraulics”, Proc. Jap. Nat. Cong. For Applied Mechanics, (1961).
6. Lane, E.W., “Report on Subcommittee on Sediment Transport Terminology”,
Trans. AGU, Vol. 28, No. 6, (1947).
7. USBR., “Design of Small Dams”, Third Edition, Water Resources Technical
Publication, Washington, (1987).
8. Van Rijn, L. C., “Sediment Transport, Part 1: Bed Load Transport”, J. Hydraulic
Engineering, ASCE, Vol. 110, No.10, pp. 1430-1457, (1984).
9. Yanmaz, A.M., “Applied Water Resources Engineering”, Second Edition, METU
Press Publishing Company, ISBN: 975-7064-02-5, Ankara, (2001).

38
 EROSION BY WATER AND SEDIMENT YIELD OF A
WATERSHED

Prof. Dr. İbrahim GÜRER

Department of Civil Engineering
Gazi University
06570 Ankara-Turkey

1. Introduction
A watershed is all the land which drains into a given body of water such as a creek,
pond, river or ocean. In general, the larger the body of water, the larger its watershed is.
For example the Kızılırmak river watershed, covers 78 180 km2 and drains about 10%
of Turkey.

For a better understanding of the sediment yield of a watershed from both hydrological
and hydraulics points of views, it may be rather useful to define the following terms,
some of which may be indirectly related to the sediment yield problem of watersheds.
In water resources projects, most of the time, the terms of Erosion and Sediment are
used together. In general, the term erosion is defined as the process of carrying away the
surface soil of a watershed by rain and /or wind. Sediment is the deposition of the soil
produced by erosion process.

Soil Conservation is conserving the various properties of the soil, which eventually
means to maintain and retain various properties of the soil and thus protect the soil in its
best possible form. Geomorphology deals with the arrangement and differentiation of
landforms and the processes that have been or are shaping them. Also time dependent
models of landscape evolution are produced by Geomorphology. Pedology is concerned
with processes involved in soil formations. Dokuchaev in 1898 suggested the following
soil forming equation:

S = f(cl, o, p) to (1)

where S is soil, cl is climate, o is organisms, p is parent material, and to is relative age

such as youthfulness, maturity, senility [Gerard, 1992].

In a watershed, the local environmental conditions, such as strong winds, heavy rains,
droughts, dissected relief, and especially the man made factors such as wrong irrigation
water application methods, misguided land use, ploughing of steep slopes, overgrazing
of pastures cause different types of erosion.

Due to heavy rainfall on the areas of dissected relief and bedrock soils that are not
resistant to washout, water erosion occurs. Forest destruction (Deforestation) is the
main contributor to the water erosion processes. According to expert estimates, 40-50%
of soils are subject to water erosion [USDA 2001]. Also the wrong methods of

39
Sediment Transport Technology

application of irrigation water and abundant irrigation cause erosion as a result of

surface washout of soils. Over 80% of all irrigated lands, mainly newly irrigated lands,
as well as rocky and sandy soils, are subject to this type of erosion.

The basins which are experiencing moderate to extreme levels of water-related soil
erosion (loss of top soil and terrain deformation) in the world are shown in Fig. 1. The
water erosion indicator is based on the Global Assessment of Soil Degradation
(GLASOD), developed by the International Soil Reference and Information Centre
(ISRIC) for the United Nations Environment Program (UNEP). The basins with the
highest levels of water erosion are found mainly in China, Southeast Asia, India, and
Madagascar. Other basins, such as the Orange and Volta basins in Africa, the Danube,
Dnieper, and the Guadalquivir basins in Europe, and the Balsas and Uruguay basins in
Central and South America, have more than 15 percent of their land area exposed to
moderate to extreme water erosion [UNEP–ISRIC, 1990].

Wind erosion is the phenomenon of wearing away, conveying and accumulating the
surface soils. The transportation of worn-away soils occurs mainly in regions where
wind velocity is more than 10-15 m/sec. Strong winds destroy soil structure and carry
away the humus, causing the decrease of agricultural productivity of the land. The most
severe wind erosion in Turkey occurs at Konya, Karaman plains and Istanbul, Lake
Terkoz locations.

Weathering is the process altering the rock forming materials until they come into
equilibrium with the surface or near surface environment. The rate of land surface
denudation is described by the removal of the weathered rock materials by rivers as
sediment discharge.

Afforestation is planting and covering the soil surface by vegetable covers and mainly
with trees. The term reforestation is also used. Torrents are fast flowing mountain
creeks, or rivers.

In case of Europe (Fig. 2), the soil erosion is also mainly caused by water and is most
serious in the Mediterranean region [USDA 2001]

Figure 1. Areas affected by water erosion [earthtrends.wri.org]

40
 Erosion by Water and Sediment Yield of a Watershed

Figure 2. Water erosion vulnerability: Europe[www.grida.no]

In Turkey, the average elevation is 1132 m above mean sea level, 45.7 % of the land is
between 1000 to 2000 m elevation belt and 62.5 % of the land has slope steeper than 15
%, the average annual precipitation is 643 mm, with a maximum of 2270 mm at Rize
and a minimum of 231 mm at Kars, Aralık. The soil in Anatolian peninsula is very
susceptible to erosion. With such a big variation in precipitation, a rather high mean
elevation and slope, more than 72 % of the country experiences severe erosion. Another
important parameter is deforestation. Although in antique period there was about 72 %
forest cover in Anatolia, now it is barely 26 %, and 35 % of the country is defined as
steppes.

2. Water Erosion
Raindrops, the most common from of precipitation (Fig. 3.), can be very destructive
when they strike bare soil. With impacts of over 40 km/h, raindrops splash grains of soil
into the air and wash out soil and the seeds. The heavy rainfall at the upper reaches of
the watershed is the main source of the erosion and this starts with splash erosion. The
raindrop effect on the surface when it strikes is shown in Fig. 4. Previously it was
thought that sheet flow (Fig. 5.), the flow of water over the soil, was the most
destructive. It is known that the kinetic energy of a moving object is equal to half its
mass multiplied by its speed squared:

Figure 3. The raindrop effect on the surface Figure 4. An example of the effect of
of the earth raindrop impact

41
Sediment Transport Technology

Splash out of Original surface of

soil particles the ground

Earth surface

Erosion limit

Figure 5. The splash of a raindrop and the ignition and the limit of sheet erosion

e = m·v2/2 (2)

As water droplets grow in size, both their speed and mass increase. The mass of a 5 mm
raindrop is 5x5x5=125 times that of a 1 mm drop and its terminal speed doubles,
resulting in a destructive energy 500 times larger. Thus the destructive power of rain
increases dramatically as the rainstorm produces larger drops, which is relatively rare.
But when it occurs, its effect is profoundly destructive. In the past ten years, starting
around 1987, rains have become heavier everywhere in the world, and with it, erosion
from raindrop impact.

The amount of the material produced by splash erosion on bare and un-vegetated and
moderate slopes is about 5 kg.m-2.yr-1. The aggregate analysis of soil material contained
in raindrop splash is given in Table 1.

Water erosion is one of the most significant agricultural problems throughout the world.
This erosion occurs due to the energy of rain drop as it falls toward the earth and flows
over the surface. Almost any area where crops are grown and food is produced has to
deal with this problem. For example it was estimated in the early 1970s that as much as
4 billion Mg (metric tons) of soil were lost in the United States to the water erosion
processes of detachment, transport and deposition each year [WEPP, 2003]. That
number is somewhat less today, however, due laws and regulations and government
conversation programs.

Table 1. Aggregate analysis of material contained in raindrop splash

Raindrop Height Size of grain in mm

Velocity
size of fall >2 1-2 0.5-1 0.25-0.5 0.105-0.25 > 0.105
fps
mm ft Percent of materials
5.1 6.7 19.2 2.40 3.89 5.07 5.24 11.38 71.98
5.1 3.9 15.0 1.41 3.68 4.98 5.12 11.53 73.25
5.1 2.3 12.0 1.04 2.87 5.74 6.56 12.11 71.65
3.5 6.7 18.0 0.74 2.87 5.24 6.65 10.83 73.76
3.5 3.9 14.5 0.94 2.81 5.66 5.85 10.81 73.91
3.5 2.3 12.0 o.44 2.31 6.25 8.22 13.86 68.89

42
 Erosion by Water and Sediment Yield of a Watershed

3. Erosion Process
The main steps of erosion by water are as follows:
1. Detachment of the soil particle by the impact of rain drop which is 0.5-6 mm in
diameter and may have a velocity of 2-9 m/sec (Fig. 4. and Fig. 5.)
2. Splash erosion (Fig. 6.)
3. Overland flow (Fig. 6.)
4. Sheet erosion (Fig. 6.)
5. Gully erosion (Fig. 7.)
6. Mass movement or land slides (Fig. 8).

Figure 6. The formation of different gullies

[Sara H. DeVita and Jed A. Long,
OMAF, 2004 ]
Figure 7. Examples of rills and gullies
on cut slopes of land
(http://www.co.cayuga.ny.us/)

Figure 8. Land slide near to GR-BG Figure 9. Rill and gully erosion on
border Kızılcahamam Road cut section

43
Sediment Transport Technology

Sheet and interrill erosion are mainly caused by rainfall. However, some of the more
severe erosion problems such as rill erosion, channel erosion and gully erosion all
result from concentrated overland flow (Fig. 9.) Other types of erosion by water include
landslides.

3.1. Rate of erosion

The rate of erosion is the function of the following parameters:
a) -rainfall regime
b) -vegetal cover, which:
i) -absorbs energy of the drop
ii) -reduces the drop size
iii) -provides mechanical protection against gully erosion
iv) -increases the infiltration capacity
c) -soil type ,which can be exemplified as:
i) -less splash erosion in well cemented soil
ii) -more splash erosion in loose soil
iii) -less splash erosion in aggregated soil
iv) -more splash erosion in loose grained soil
v) -less splash erosion if there is more water-stable
vi) aggregates in the soil
vii) -more splash erosion if the soil has more sand
d) -land slope, which has two fold effects (gradient and length)
i) -greater rate of erosion in steep slopes
ii) -more affective splash erosion in steep slopes
iii) -greater overland flow on steep slopes
iv) -more mass movement in steep slopes
v) and easy to reach the channel if the length of slope is shorter
vi) -higher overland flow velocity and bigger overland flow rate if the length of
slope is longer

Several models exist to predict water erosion rates (e.g. the Universal Soil Loss
Equation or USLE). However, relying on such empirical tools is often risky since most
of them apply only to particular agro-ecological conditions and require specific sets of
preconditions which are often not met in the dry areas.

3.1.1. Universal Soil-Loss Equation (USLE)

Using 10.000 plot-years of data from various research facilities, mainly in the eastern
half of the United States, Wischmeier and Smith developed the USLE in 1960. The
USLE is perhaps the most widely used and accepted model for planning erosion-control
practices in the United States. The principal application of the USLE has been to
provide specific and reliable guides for selecting appropriate erosion-control practices
for farmlands and construction areas. Other useful applications have been in
determining upland erosion for reservoir sedimentation and stream loading, control of
pollution from cropland, and alternative land use and treatment combinations.

Following Wischmeier and Smith [1978], USLE can be written as [Singh, 1992];

A = R K LS C P (3)

44
 Erosion by Water and Sediment Yield of a Watershed

where, A; average annual soil loss: the predicted erosion or tolerance factor in ton/ha, R;
rainfall erosivity factor: a factor dependent on climate and likelihood of extreme events.
R can be evaluated for each storm (summed over hours) from;

R = 0.01∑ E I (4)

where the summation is performed over the time increments of the storm and E is given
by;

E = (916+331 log I) (5)

in which E is the kinetic energy in foot-tons per acre-inch and I is the rainfall intensity
in inches per hour. Soil erosion losses from single storms strongly correlate with the
maximum 30-min rainfall intensity.

Soil erodibility factor, K: An estimate made from soil properties. It depends on the
particle sizes and proportions of sand, silt and clay, organic matter, granularity and
profile permeability to water. K describes the inherent erodibility of the soil expressed
in the same units as the annual erosion losses.

Slope length factor, L: The slope length is the length of the field in a down-slope
direction. The larger slope length, the more water accumulates at the bottom of the field,
increasing erosion. It also depends on the land's slope.

Slope steepness factor, S: Calculated from the slope of the land in %.

The cropping-management factor, C: For bare soils is taken is a standard value equal to
unity. The factor C accounts for soils under different cropping and management
combinations such as different vegetation, canopy during growth stage, before and after
harvesting, crop residues, mulching, fertilizing and crop sequence.

Supporting practice factor, P: Reflects the use of contours, strip cropping and terracing.
The conservation practise factor P equals 1 for downslope rows and typical values for
contouring, strip cropping and terracing [Maidment, 1993].

A tolerance factor of 4-11 ton /ha/yr is aimed for, which is still much larger than natural
soil formation, but which does not appear to affect productivity. If the tolerance factor
exceeds sustainability of productivity, different cover management practices or
supporting practices are selected and enforced.

3.1.2. Modified Universal Soil Loss Equation (MUSLE)

The MUSLE, can be expressed as;

A = B (VQ·QP) 0.56 K LS C P (6)

where, VQ; the volume of runoff in m3, QP; the peak flow rate in m3/sn, B is a constant,
and other terms have the same meaning as in the USLE. Based on the data from small
watersheds near, Riesel, Texas and Hastings, Nebraska, B was found to be 95. If VQ is

45
Sediment Transport Technology

in m3, QP in m3/sn, then B would be 11.8. Eq. 6 explained %92 of the variation in
sediment yield of these watersheds [Singh, 1992].

3.2. Relation between erosion and precipitation

Fig. 10 shows the relationships of water erosion (continuous lines) and wind erosion
(broken lines) with increasing mean annual precipitation. The curves for water erosion
indicate the relationships with mean annual precipitation for (A) areas of natural
vegetation cover and (B) bare ground (after Schumm, 1969). The curves for wind
erosion indicate the relationships with mean annual precipitation for (C) areas of natural
vegetation cover and (D) bare ground. These curves are based on what would be
expected from the relationship of wind erosion to vegetation cover and to moist soil
(from Marshall, 1973).

3.3. Soil erosibility

The erosibility of soils is defined as their susceptibility to erosion and it depends on the
following factors (most of which are actually interrelated):
a) Structure
b) Stratification
c) Permeability
d) Moisture content
e) Texture
f) Mechanical composition
g) Type and extent of vegetal cover
h) Land slope

3.3.1. Relation between land slope and erosion

Fig. 11 shows how quickly erosion accelerates. Many soil and slope processes such as
overland flow and rill development show extreme spatial and temporal variability.

Figure 10. The relationships of water erosion (continuous lines) and wind erosion (broken
lines) with mean annual precipitation. Reproduced from Drought, edited by J. V.
Lovett by permission of Angus & Robertson Publishers, Sydney, Australia

46
 Erosion by Water and Sediment Yield of a Watershed

Figure 11. The erosion process as function of slope [Anthoni, 2000]

Relationships between the types of soils, the slope position and the slope gradient are
general changes to soil profiles and sediment producing events. Geomorphological
processes acting on the slopes are largely governed by the slope angle, the length of the
slope and the curvature and therefore they should also be related to position in the
drainage basins, as measured by the stream order [Anthoni, 2000].

As far as sustainability is concerned, any land steeper than 5% should not be ploughed,
but converted to perennial crops like viticulture, horticulture or grassland. Slopes above
10% are left to be for trees. If it is accepted that only land sloping less than 5 degrees
can be cropped sustainably, then over 70% of arable cropland appears unsustainable.

3.4. Relative erosion potential

The International Geosphere-Biosphere Program (IGBP) uses Relative Erosion Rates
(RER) as given in Table 2 for different types of land use. By taking land slope and
rainfall into account, the following equation is used to estimate the Relative Erosion
Potential (REP):

REP = J1.5·RER·P/1000 (7)

where, J; slope, RER; Relative Erosion Rates, P; the rainfall precipitation is the mean
monthly rainfall for the peak rainfall month, in mm. This formula does not take evapo-
transpiration from the vegetated land into account, but assumes that erosion occurs
mainly when the soil is saturated and rainfall is maximal, with heaviest rains occurring
in one month. As could be seen from Table 2, the RER for croplands and urban areas is
over ten times that of an evergreen forest. Watersheds with REP less than 5 are under
low threat; between 5 and 45 medium and those above 45 high. A maximum REP of
1400 is possible for large rivers with poor soil management [www.icarda.org. 2003].

47
Sediment Transport Technology

Table 2. The Relative Erosion Rates (RER) for different

types of land use [www.icarda.org].

Land category RER

water body, lake 0.5

evergreen broadleaf forest 1.0
evergreen needle leaf forest 1.5
deciduous needle leaf forest 2.0
closed shrub land 4.0
open shrub land 5.0
woody savanna 6.0
savannas 8.0
permanent wetlands 8.0
croplands, natural mix 12.0
grasslands 12.5
croplands 21
urban & built-up 21
barren or sparse vegetation 21

3.5. Water erosion maps

The global positioning system (GPS) can be used for watershed erosion maps. By using
GPS, it is possible to map the temporal and spatial variations of the erosion in a given
area and also identify the causes of soil erosion at that particular site. For example (Fig.
12) at two benchmark sites at the northern Syria, researchers found that water erosion
was caused mainly by (i) overgrazing of the hillsides (ii) up-and-down tillage, and (iii)
uncontrolled run-on of surface water from roads, tracks, and (animal) paths [ICARDA
Annual Report 2002].

Figure 12. A water erosion map at Yakhour (Northwest Syria)

developed with simple GPS and GIS tools
(ICARDA Annual Report 2002)

48
 Erosion by Water and Sediment Yield of a Watershed

3.6. Watershed erosion control projects

In order to have an erosion control project in a watershed to be realized it is necessary to
analyze the following parameters:
a) Topography
b) Climate (precipitation, temperature)
c) Geology
d) Soil characteristics
e) Existing land utilization
f) Social and economic level of local people
g) Existing level of erosion

Erosion control projects are realized starting at upper parts of watersheds of the
reservoirs, at land slide slopes, and at slopes damaged by forest fires. At Gelibolu and
Marmaris Peninsulas after the forest fires, at Bursa Dogancı, Artvin Deriner dam (Fig.
13) drainage basins for slope stability and erosion control and at Burdur, Malatya,
Southeastern Anatolia (Fig. 14) and Çakıt sub basin of Seyhan drainage area, a series of
erosion control projects have been realized by DSI and Reforestation Department of
Ministry of Environment and Forestry.

4. Sediment Sources
Watersheds behave as open systems, where inputs are precipitation and energy and
outputs are water and material which is termed as sediment yield. Sediments may be
released to stream by above mentioned types of surface erosion on hill slopes and gully
erosion or erosion of the channel banks. For most of the rivers, the largest part of the
sediment load is derived from the headwater areas of high relief that comprise only a
small portion of the drainage basin. For example, in Amazon basin 82% of the
suspended sediment is supplied from 12% of the drainage area on Andes Mountains. In
global scale about 80% of the sediment load comes from the 5-10 % of drainage areas
in mountains.

Figure 13. Slope erosion control at Artvin Figure 14. Erosion control at Malatya,
Deriner Dam by DSI Southeastern Anatolia by
Reforestation department

49
Sediment Transport Technology

Especially in developing countries, the soils on alluvial landforms of the world major
rivers, cause many environmental problems. Such problems range from flooding and
water logging to salinization and loss of civil works. As an example, the catastrophic
flood of 1980 spring originating from Çakıt stream can be mentioned. The flood
produced by the melting of snow and rain, damaged the E-5 highway and Ankara-
Adana railway and endangered the Adana city itself. The real cause of the damage is
nonexistence of any protective vegetal cover as the result of forest destruction at the
uphill of the catchment.

4.1. Annual sediment yield

The average annual sediment yield of a watershed is the function of:
1. Climate
2. Soil type
3. Land use
4. Topography
5. Presence of reservoirs in the watershed

When the rain drops on the earth surface, it is either retained where it falls (Surface
Retention or Surface Storage or Initial Detention) and then infiltration process takes
place and at the same time direct evaporation occurs or through surface channel system
it becomes part of the surface runoff.

From the hydrological point of view the retaining of the rain drop can be done by:
a) vegetal cover which is called interception and it has a rate about 25% of average
annual precipitation in dense forests
b) surface puddles which is called depression storage
c) soil surface as a film of water which is called surface detention

In this process, the soil is continually being moved from the land surface of the earth
and transported downhill in the river basins until it is deposited in reservoirs, lakes,
estuaries, and finally in the oceans. Water is both erosion and transportation agent of the
process. The hydrologist is interested with the following items in this process:
i) rates of deposition in the reservoirs, harbors and estuaries
ii) methods of erosion control if possible at its origin
iii) conservation of soil in place
iv) minimize the accumulation at reservoirs and harbors

Once the tributary or river water, loaded with sediment and nutrients, reaches main river
(Fig. 15) or to the sea (Fig. 16), it forms a mud plume, the size of which increases
gradually as the river flow increases. The graph given in Fig. 17 shows plume length
(km) against river flow in thousands of m3·s-1, assuming that river plumes dissolve into
the sea water continuously and assuming there are no coastal currents to displace these
over huge areas. In the graph, the plume is not expressed as surface area , but only as a
distance extending out from the river's mouth, hence the curve's flattening out. The
invisible plume of released nutrients is estimated to be far larger and longer lasting
[www. icarda.org. 2003].

50
 Erosion by Water and Sediment Yield of a Watershed

Figure 15. The sediment plume of a tributary Figure 16. Sakarya river reach Black Sea
when reached the Filyos [www.geo-motion.net/.../bilder/blacksea1.png]

Figure 17. The relation between the river plume length (km) with river flow (m3·s-1)

The sediments transported by rivers are derived from the rocks, soils and superficial
deposits within the drainage basins. As individual grains, the characteristics of a
sediment grain can be described by three important factors:
1. Size (clay, silt, sand, gravel etc.)
2. Shape (spheres, rods, discs, blades)
3. Degree of roundness (angular, sub-angular, sub-rounded, rounded)

Climate predominates on the rate of sediment yield on the global scale because climate
controls weathering and runoff rates, and the character of the vegetation cover. However
rivers with high sediment yields are associated also with areas of high relief and on a
smaller scale with human activities, particularly in vegetation clearance and land
cultivation.

51
Sediment Transport Technology

The sediment load of rivers is the sum of the products of the weathering processes.
Physical weathering produces mechanical breakdown of rocks into smaller fragments,
where as chemical weathering produces chemical decomposition of rock-forming
minerals into different materials. The solute load, which was defined as the load
transported by river in solution, are contributed by precipitation and dry fallout, but it is
rather difficult to separate them, they are considered as bulk precipitation. At global
scale the chemical composition of surface water has been related to climate
(atmospheric precipitation and evaporation) and to Geology [Gibbs, 1970].

Compared to the streams draining resisting sand stones, from the high relief areas
composed of mud stones about 10 times higher sediment yield is produced. Active
gullies on slopes ranging from 35o to 55o and sediment traps located at their foot,
yielded erosion rates of about 50 kg.m-2.yr-1, in England [Petts, 1983]

Overland flow, or surface runoff, then carries away the detached soil, and may detach
additional soils and then sediments can be deposited at the entrance of the reservoirs
(Fig. 18) and /or at the delta of the rivers (Fig. 19).

There is a close relationship between the size of the particles eroded from the earth
surface and the current velocities for erosion, sedimentation and transport as shown in
Fig. 20. Fig. 20 shows how particles are transported. The left curve gives the speed at
which particles settle out; the rightmost curve the speed to erode cohesive material; the
middle curve to erode loose material. In Fig. 20, the abscissa shows the water speed
from 0.01cm/s to 1000cm/s (36 km/hr or 20 knots), and the ordinate is the grain size
from 1 micron (clay) to 10mm (pebbles). The colored bands show where clay becomes
silt, then sand, then granules, and pebbles. The transportation gap between erosion and
sedimentation, widens as the particles become smaller. Thus silt and clay particles
reluctantly settle out. It takes clay weeks in fresh water. With this knowledge, one can
understand that particles settle out in different places, as a river slows down in its lower
reaches. First the pebbles will settle and then further down clean sand and perhaps near
the sea the silt. But the clay travels all the way to the sea [www.seafriends.org.nz/
enviro/soil/erosion.htm].

Figure 18. Sediment Deposition at the entrance Figure 19. Sediment deposition at the delta
of a DSI reservoir at Burdur region of a river

52
 Erosion by Water and Sediment Yield of a Watershed

Figure 20. The relationship between the size of Figure 21. The variation of sedimentation
the particles eroded from the earth rates in fresh and salty water
surface and the current velocities for
erosion, sedimentation and transport

There is also difference between the rates of settlements of particles in salty and fresh
water. The sedimentation rates in fresh and salt water are given in Fig. 21. In salt water
fine particles and organic matter settle out considerably faster than in fresh water (4-200
times, depending on particle size). It is thought that the salt water encourages particles
to join together, thus making them larger and sinking faster, but this goes against
Bernouilli's law and the sedimentation/transport diagram above [www.seafriends.org.nz
/enviro/soil/erosion.htm].

4.2. Magnitudes of sediment yields of watersheds

Sediment yield is the rate at which sediment is moved away and it is expressed as ton.
km -2. year-1 and it is also termed as sediment discharge. The solute load is the material
transported by rivers in solution and it is also called dissolved load.

In order to emphasize the magnitude of the sediment yield problem of the watersheds,
the sediment yields of some rivers are given in Table 3. The annual flow volume, size of
the drainage basin, annual sediment discharges are given in the same table. As could be
seen in the table, the annual sediment yield differs from the annual water volume and
the size of the drainage area. This is because of the percentage and the location of the
sediment source section of the basin besides the type and the intensity of precipitation.

The long term mean value of sediment discharge in Britain is 33 ton.km-2.year -1, with a
range of maximum of 488.0 and minimum of 0.8 ton.km-2.year-1. For the earth as a
whole, the average sediment discharge has been estimated about 125 ton.km-2.year-1,
which corresponds a lowering of the land surface 30 mm in 1000 year [Petts, 1983].

In Anatolia, the erosion and resulting sediment transport problem started in III. Century
B.C. and most of the antique harbors were filled with the sediment carried from the
upper reaches of the watersheds. For example, due to the filling of the estuaries of the
rivers of Büyük Menderes (Figs. 22 and 23), and Küçük Menderes and Mersin harbor,
the ruins of antique cities of Ephesus. At present Milet and Salinin antique towns are
located many kilometers inland.

53
Sediment Transport Technology

Table 3. Water and sediment discharge in some large rivers of the world [McLennan, 1993]
Water Drainage Sediment Sediment
River discharge area (106 discharge yield
(km3/yr) km2) (106 t/yr) (t/km2/yr)
Amazon 6300 6.15 1200 195
Colorado 20 0.64 0.01 0.02
Columbia 251 0.67 10 15
Congo (Zaire) 1250 3.72 43 12
Danube 206 0.81 67 83
Ganges/Brahmaputra 971 1.48 1060 716
Huang He 49 0.75 1050 1400
Indus 238 0.97 59 61
Mackenzie 306 1.81 42 23
Mekong 470 0.79 160 202
Mississippi 580 3.27 210 64
Niger 192 1.21 40 33
Nile 30 3.03 0 0
Orinoco 1100 0.99 150 152
St. Lawrence 447 1.03 4 4

Figure 22. Buyuk Menderes Estuary Figure 23. Buyuk Menderes delta

During the reign of Ottomans, the erosion and sediment problem was less damaging
because the folk was not very much dependent on land. During Turkish Republic, as the
population increased from 13 Million in 1923 to 68 million in 2000, more agricultural
land was needed, this caused forest damage and consequently erosion became a problem
and sediment discharge of the watersheds increased. This is the most important reason
of sudden decrease in active volumes of most of the reservoirs. Although 13 small water
supply dams built during Ottoman period such as Topuzlu are still active. Some of the
dams in operation which have sediment problem are listed with their hydraulic
characteristics in Table 4, where the dead volumes are given till 1990.

If the sediment load observations are not available when the planning of a reservoir is
done, the dead volume of the reservoir for economic life of 50 years, is computed with
the relation given below:

54
 Erosion by Water and Sediment Yield of a Watershed

Table 4. Sediment yield ratio of 16 dams in Turkey

Reservoir
Original Original
Drainage Reservoir Transported operation
Name of the Geographic dead sediment yield
area volume sediment started
dam region volume ratio
(km2) (hm³) (m³/year) in
(106 m3) (m3/year/km2)

Altinapa Central Anat. 589 15.7 4.0 377417 1965 13.582

Bayindir Central Anat. 70 7.0 1.0 102500 1965 *250.00
Buldan Agean 180 46.0 5.0 984000 1967 555.56
Caygören Aegean 1510 137.6 15.0 915231 1971 198.68
Cip East Anatolia 286 7.0 1.7 105800 1965 109.58
Cubuk I Central Anat. 840 12.5 0.04 166333 1935 0.95
Demirköprü Central Anat. 6590 1320.0 165.0 8563800 1960 500.76
Hirfanli Central Anta 26170 5980 880.0 98400000 1959 672.53
Karamanli Mediterranean 164 24.0 2.1 379000 1972 *250.00
Kartalkaya Mediterranean 1130 195.0 15.0 2684263 1972 265.49
Kemer Aegean 2500 544.0 12.0 4612955 1957 *380.00
Kesikköprü Central Anat. 360 97.4 37.9 624200 1966 2105.56
Selevir Aegean 722 74.6 9.9 995857 1965 274.09
Seyhan Mediterranean 19254 1200.0 300.0 11450000 1955 311.62
Sürgü East Anatolia 275 72.2 10.0 931625 1969 727.27
Yalvaç Mediterranean 133 8.9 1.6 225700 1973 240.60
(*) recomputed according to last depth-volume maps of the reservoirs

Dead Volume = (0.001-0.0024) x 50 yrs x Qmean (8)

The coefficient above, given as varying between 0.001 and 0.0024, is determined
according to vegetation cover of the watershed [Değirmenci, 1990].

At present from the river Kızılırmak 65 million ton, from the Euphrates 108 million ton
soil is carried to sea. Every year 450-500 million ton top soil goes to dams and lakes of
Turkey. This corresponds to 1 mm top soil loss in every two years, but it is also known
that 1 cm of top soil can be generated in 1000 years, i.e. the soil layer created in 50 yr
period is lost in one year. This very high rate of erosion in Turkey, is 6 times more than
of North America, 17 times of Europe and 22 times of Africa.

5. Sediment Yield Estimates

The average annual sediment yield of a watershed is affected by various parameters
listed in former paragraphs. Long term observations of adequate data for a complete
analysis of all these factors are difficult to obtain. There are some general conceptual
approaches, for example in Fig. 24, sediment yield as a function of mean annual
precipitation is given by Langbein and Schumm [1958] The curve shown in figure is
based on the observations in the United States, but it is rather approximate.

Fleming [1969] set up a relation between mean annual sediment load and mean annual
discharge by using the data obtained from 250 basins, considering various vegetation
types but the expected error is about +50% and the equation is:

Qs = a Qn (9)

55
Sediment Transport Technology

Annual sediment yield, metric tons / km2

0 50 100 150 200 250
60 1500

Mean annual precipitation, mm

50 1250
Mean annual precipitation, in. 40 1000
Forest

30 750

20 500
Grassland

10 250

0 Desert shrub
0

0 200 400 600 800

Annual sediment yield, tons / ml2
Figure 24. Sediment yield as a function of mean annual precipitation

The a and n values are given in Table 5 for different types of vegetal covers. A
combined approach is proposed by Özbek [1994] which is based on USLE, synthetic
rainfall-runoff relations given by Snyder and Mockus, and Total load transport concepts
developed according to stream sediment transport capacities, but with the available
sediment data at present, the application may need modifications.

5.1. Methods of predicting sediment yields from watersheds

There are three methods in general use where no existing basic data are available for the
watershed above a proposed dam. These are use of existing data from comparative
watersheds, installation of sediment-load measuring stations estimating watershed
erosion and delivery ratio of sediment.

Table 5. Values of a and n in Fleming’s equation

a
Vegetation type n For Qs in metric tons
Q in m3/sec

Mixed broadleaf and coniferous 1.02 4,000

Coniferous forest and tall grassland 0.82 59,000
Short grassland and scrub 0.65 177,000
Desert and scrub 0.72 446,000

56
 Erosion by Water and Sediment Yield of a Watershed

5.1.1. Predicting sediment yields from comparative watersheds

The expected sediment yield to a proposed reservoir may sometimes be predicted by
assuming that the rate equal to that measured on a nearby similar watershed. This
method is acceptable if both watersheds are basically same in respect to soils,
topography, size, cover, climate and other conditions influencing erosion and movement
of sediment, listed above. The method is best for small watersheds with homogeneous
surface cover conditions such as forest or rangeland.

It is possible to come up with considerable error in estimating when using data from one
watershed if the two watersheds involved have mixed cover-slope complexes or have
other characteristics which differ in respect to influence on erosion or sediment-delivery
ratio. When such conditions exist, data for estimating must be expanded to include
ranges in factors as a basis for estimating the yield from a watershed without
counterpart. Referring to Gottschalk, Chow [1967] mentioned a fairly reliable estimate
of the rate of the sedimentation of small stock ponds and reservoirs in South Dakota.
The relationship derived from regression analyses of measured sediment accumulation
in 18 ponds and reservoirs is

S = 0.0522C+0.0027A+0.2681T-1.7974 (10)

where, S; total sediment accumulation in acre-ft, C; capacity of the pond or reservoir in

acre-ft, A; net drainage area in acres, and T; age in years.

Variations in rates of sediment accumulation still existed depending upon the relation
between watershed area and capacity and time. In other studies of groups of watersheds,
investigators have found different parameters, which better define conditions that
influence erosion or movement of sediment for the specific problem area involved.
These include such parameters as gross erosion, density of incised channels, density of
non-incised channels, mean slope of third order streams, hydrologic events and others.

5.1.2. Installation of sediment - load measuring stations

Installation of a stream-gauging station and periodic sampling of stream flow prior to
the development of design may obtain at or near the site of a proposed structure the
sediment yield from a watershed. This method is used for major structures on large
watersheds where investigations prior to design of the structure, permit an adequate
period of sampling to obtain measurements representative of expected long-term
discharges. The results of such measurements, adjusted for expected long-term sediment
yield, are used directly for design purposes.

5.1.3. Estimating watershed erosion and the delivery ratio of sediment

In most of the developing countries sufficient data on sediment yield are not available to
evaluate the relative effects of factors influencing sediment yield. Also, sufficient time
may not be adequate to conduct sediment-load measurements prior to design of a water
impoundment structure. In such instances the sediment yield may be predicted by
estimating the gross erosion in the watershed and adjusting for the expected delivery
ratio of the sediment to the structure site. This method is applicable to smaller
watersheds where surface runoff is controlled by small earth dams and the collected

57
Sediment Transport Technology

water is used for irrigation and/or animal water seriously and particularly useful where
abnormal changes in land use and agricultural practices are anticipated in the watershed
during the design life of the structure.

The estimated amount of sheet erosion is calculated by standard methods. Basic

information on land use, slopes, and other factors needed to estimate sheet erosion can
be obtained for some regions from village affairs department of governor` office of the
city (former KHGM) in Turkey. The amount of sediment derived from channel and
other types of erosion in the watershed must be added to the amount derived for sheet
erosion to estimate gross erosion. It is possible to estimate the rates of channel erosion
in the field, and the usual method consisting of comparison of current channel
conditions with those old ones if available and shown on older aerial photographs. The
delivery ratio of eroded material is then determined on the basis of the size of the
watershed, the topography, and the degree of canalization. It is used to estimate the
amount of eroded material delivered to the structure.

5.2. Sediment simulation

Nagev [1967] proposed a simulation program including almost all the parameters
affecting sediment yield of a watershed but this model also did not have the chance of
application due to lack of bed-load data required for developing the algorithm. In order
to estimate the total suspended load the following set of equations will be used:

Soil splash (R): R = K1 ij (11)

k
Splash Residue (S): S = K2 Rs Qo (12)
Sediment washed from impervious areas (E): E = K3 R (13)
Total wash load (W): W = R+S+E (14)
m
Gully erosion (G): G = K4 qo (15)

The part of gully erosion having the substantially same size as the bed material load in
the stream

Material load in the stream (B): B = (1-K5) G (16)

Interload (I): I = G-B = K5 G (17)
Total suspended load (qs): qs = K6 Is q n+K7 qr (18)

In the above set of equations the terms used can be explained as follows:
i : intensity of rain
Rs : residue in storage on the ground surface
qo : rate of overland flow
Is : quantity of interload in storage
q : mean daily discharge

Also there are assumptions in this simulation model such as:

1. If there is no overland flow the only source of load is wash load
2. Nagev (1967) divided the gully sediment into two parts B and I

58
 Erosion by Water and Sediment Yield of a Watershed

3. Interload represents 95% of the bed load material

4. The quantity of interload in storage in the streambed is calculated by maintaining a
running balance of the input from erosion and the outflow of suspended interload
material
5. All the coefficients from K1 to K7 are defined by calibration
6. K5 is estimated from particle size data
7. The overland flow rate is calculated with a flow simulation model
8. j = 3, k = 2.5, m = 2.5(assumed values)
9. Nagev [1967] assumes that n = r, slope of the sediment rating curve of the station
under consideration

Long sediment records are needed for better estimate of sediment yields of the
watersheds because of the year to year variability of the sediment transported.
Occasional, infrequent sediment samples do not provide a good basis for calibration of
the model. A program of daily sediment samples or continuous monitoring is important
in improving the ability of the scientist to deal with sediment problem.

A number of conceptual models have been devised for the simulation of discharge,
water quality and sediment load etc. for the same watershed and they may be helpful in
estimating the parameters of the simulation model.

5.3. Models for sediment yield estimates

5.3.1. Estimation of sediment yield using integrated watershed-channel models
CCHE1DST calculates non-equilibrium; non-uniform sediment transport in channel
networks under unsteady flow conditions. It simulates bed aggradation and degradation
processes and the resulting morphologic changes. The model computes variations of the
bed material gradation by dividing the bed into several layers, simulating hydraulic
sorting and armoring. The model also simulates channel-widening processes through
bank erosion and stability analysis algorithms [www.ncche.olemiss.edu].

The CCHE1D channel network model is designed to integrate with watershed models
such as AGNPS or SWAT (Fig. 25). The integrated modeling system can be used as a
tool to evaluate the effectiveness of erosion control and channel remediation measures
on sediment yield, and to study the influence of land use changes and agricultural
management practices on erosion and sedimentation in channels and upland areas. The
landscape analysis tool TOPAZ is used to define the subwatersheds and extract the
channel network from the digital elevation model (DEM) for watershed and channel
modeling. A graphical user interface has been developed based on ArcView GIS to
handle data preparation, simulation and post-processing of modeling results
[www.ncche.olemiss.edu/sedimentgroup/estimation_ofsedimentyield.htm].

6. Controlling Excessive Sediment Yield

The purpose is to prevent erosion and consequent sediment yield of a watershed. This
can be done by minimizing the sheet flow and lowering it to non-damaging level. In
application in uphill and in lowland valleys some differences are observed [Silvio,
1993]. Because the characteristics of the torrents (Mountain rivers) with respect to
lowland rivers are different in the following respects:

59
Sediment Transport Technology

Figure 25. Procedures for the integration of CCHE1D channel network model with watershed
model AGNPS or SWAT

1. Irregular geometry such as width and slope

2. Non-uniform grain size such as silt, boulders etc.
3. Unsteady hydrology such as flush floods
4. Direct sediment input such as mass movement, land slide

Due to these reasons

a) water flow is basically controlled by section geometry
b) sediment transport is basically controlled by sediment input

In the long run afforestation (reforestation) is the best solution for controlling excessive
sediment transport because:
1. It provides cushion to falling rain drops and reduce impact and consequent damage
2. The roots of the plants go deep into the soil, and stabilize it against water erosion.
3. It creates obstacle to the path of flowing water and reduces the velocity and
minimizes the scour of surface runoff.

Although, the basic solution to minimize the sediment load input to reservoirs is
afforestation or reforestation, starting from the upper reaches of the watersheds, but it is
a long term and time consuming solution. At the drainage areas of Atatürk dam 1740 ha,
of Keban dam 4080 ha, and of Kurtboğazı dam 1047 ha were already afforested.

Since it takes rather long time for the forest to grow, usually it is recommended to built
small drop structures and sediment traps at the upper reaches along the tributaries of the
water courses.

In upper reaches where debris flow takes place Full Check dams made of concrete,
masonry or timber are preferred to decrease bottom slope to reduce the debris flow
occurrence and to stabilize valley slopes. It is also possible to use sediment storage
dams to stabilize the material in the uphills.

60
 Erosion by Water and Sediment Yield of a Watershed

In lower reaches where suspended and bed load transport takes place, permeable check
dams are used to intercept material during exceptional events.

There are some cheaper methods such as construction of the Rough Ramps using huge
rocks with a bucket below the ramp in shape of natural scour, small vertical barrages by
little cataracts, Field Bunds and Contour Bunds to reduce the velocity. It is also possible
to use Mud Terracing in which mud is mixed with certain stabilizing agents to cover the
land surface. In Turkey Ministry of Forestry General Directorate of Erosion Control and
Afforestation is responsible to coordinate these works. They use different types of
terraces, and pasturing, and the controlled use of existing pastures [OGM, 1993].

The quick solution is to built sediment traps, regulate the sediment carrying tributaries
and build reservoirs. Sediment output variation of a watershed when a reservoir is built
is shown in Fig. 26. As could be seen from the Fig 26, with the completion of reservoir,
the sediment discharge decreases, but the stream discharge flow continues as before. A
series of erosion and sediment control methods and their applications is given in Fig. 27.

7. Erosion and Sediment Control Law and Regulations

Most of the countries and the states have laws or regulations for implementing their
erosion and sediment control programs. The basic goal is to control soil erosion,
sedimentation, and nonagricultural runoff from regulated "land-disturbing activities" to
prevent degradation of property and natural resources. The regulations specify Minimum
Standards, which include criteria, techniques and policies that must be followed on all
regulated activities in the state or country. These statutes delineate the rights and
responsibilities of decision makers and those of property owners who must comply. As
an example Virginia Erosion and Sediment Control Law, Regulations, and Certification
Regulations, is codified at Title 10.1, Chapter 5, Article 4 of the Code of Virginia,

Figure 26. The effect of a reservoir built on river to the sediment discharge of the river
[http://ia.water.usgs.gov/nawqa/factsheets/fs]

61
Sediment Transport Technology

(a) (b)

(c) (d)

(e) (f)

Figure 27. Different types of erosion and sediment control applications at different parts of the
world and in Turkey control (a), (b), (c), and, (d) Rize, Eastern Black Sea Region,
(e) Muğla, Kara Muğla (f) Keban Dam
[www.dsi.gov.tr, www.oired.vt.edu/amarew/impact/gully, http://www.osmre.gov].

62
 Erosion by Water and Sediment Yield of a Watershed

regulations are found at Section 4VAC30-50, and certification regulations are found at
Section 4VAC50-50 of the Virginia Administrative Code. In Turkey DSI and
Department of Reforestation of Ministry of Environment and Forestry are the
responsible organizations for watershed management.

References

1. Anthoni J, F., “Soil: erosion and conservation”, www.seafriends.org.nz/enviro/soil

/erosion.htm, (2000).
2. Chow, V.T., “Handbook of Hydrology”, Prentice Hall, USA, (1967).
3. Değirmenci, I., “Örneklerle Su ve Toprak Kaynaklarının Planlanması”, Yeni Çağ
Basın Yayın Ltd., Ankara, (1990).
4. Fleming, G., “Design Curves for Suspended load Estimation”, Proc. Inst. Civ. Eng.,
Vol. 43, pp1-9, (1969).
5. Gerard, J., “Soil Geomorphology”, London, UK, (1992).
6. Gibbs, R., “Mechanisms Controlling World Water Chemistry”.
7. Langbein, W. B., and Schumm, S. A., “Yield of Sediment in Relation to Mean
Annual Precipitation”, Trans. Am. Geophys. Union, Vol. 39 pp. 1076-1084, Dec.
1958 Science, 170, pp. 1088-1090, (1958).
8. Maidment, D. R., “Handbook of Hydrology”, Mc Graw-Hill, USA, (1993).
9. McLennan, S. M., “Weathering and global denudation”, J. Geol., 101, 295–303,
(1993).
10. Nagev, M., “A Sediment Model on a Digital Computers”, Stanford University,
Department of Civil Engineering, Technical Report 76, March, (1967).
11. Özbek, T., “Havza Sediment Verimi ile İlgili Yaklaşımlar”, Su ve Toprak
Kaynaklarının Geliştirilmesi Konferansı, DSİ, Ankara, (1994).
12. Özkahraman, İ., “Türkiye Erezyon Sorunu”, T.C. Orman Bakanlığı, Ankara,
(1993).
13. Özkahraman, İ., “Adana Çakıt Çayı Erozyon Kontrolü Projesi”, Orman Genel
Müdürlüğü, Büyük Uygulama Projeleri, No:2, Tarım Orman ve Köyişleri Bakanlığı
Yayınları.
14. Petts, G., “Sources and Methods in Geography-Rivers”, London, (1983).
15. Revenga, C., S., Murray, J., Abramovitz, and A. Hammond, “Watersheds of the
World: Ecological Value and Vulnerability”, World Resources Institute,
Washington, DC, (1998).
16. Silvio, Di, “Hydraulique des riviéres torrentielles”, Conseil de l’Europe Réseau
Européen Monde Alpin, Universite Europeene D’ete sur les Risques Naturels,
Espagne, (1993).
17. Singh, V. P., “Elementary Hydrology”, Prentice-Hall, USA, (1992).
18. UNEP – ISRIC, “Global Assessment of Soil Degradation (GLASOD)”,
International Soil Reference and Information Centre, Wageningen, Netherlands,
(1990).
19. Virginia Erosion and Sediment Control Law, “Regulations, and Certification
Regulations”.
20. http://ia.water.usgs.gov/nawqa/factsheets/fs
21. http://www.oired.vt.edu/amarew/impact/gully3
22. http://www.osmre.gov
23. http://www.ncche.olemiss.edu/

63
Sediment Transport Technology

24. http://www.earthtrends.wri.org
25. http://www.dsi.gov.tr

64
 MEASUREMENT OF SEDIMENT LOAD

Dr. Ş. Yurdagül KUMCU

Assoc. Prof. Dr. M. Ali KÖKPINAR
Hydraulic Model Laboratory,
Technical Research and Quality Control Department,
State Hydraulic Works, Ankara-Turkey

1. Introduction
Sediment load is defined as the rate of sediment transported by a running water stream.
This transportation comes if the flow velocity is enough. Alternations in the flow
velocities, either caused by nature or by human interference in the natural conditions
will influence the magnitude of the sediment transport (Boiten, 2003). It is essential to
have an idea about the rate of sediment transported for following purposes:

Irrigation: The design of stable channels, of sand traps and of intake structures requires
sufficient insight into the sediment transport phenomena.

River Engineering: The design of rivers works, e.g. navigation, flood control, reservoir
sedimentation, requires knowledge of sediment transport.

The information of sediment transport rate can be obtained either by semi-empirical

relations or by direct measurements. It is evident that the existing theories and empirical
equations for the computation of sediment load give only approximate values as the
interaction between the water movement and the sediment transport formulae is very
complex and it is difficult to describe mathematically. Although computed sediment
transports are rather inaccurate, in most cases theoretical computation is only the load
that it can be used. But the engineers would rather prefer to use the data obtained from
the measurements of transport rate wherever possible, such measurements not only
provide reliable data for design but also enable checking or modification of the existing
methods of computation of sediment load. Therefore, one can be said that measurement
of sediment load is vitally important for transport problems.

The particles move by rolling or sliding over, hopping near and by sweeping away from
the bed depending on:
1. The composition of bed material of an alluvial channel (size, specific weight, ...etc.),
2. The flow conditions (velocity, depth…etc.), and
3. The fluid properties (density and viscosity).

Therefore, usually three modes for the movement of sediment particles can be defined:
a) Contact load is defined as that part of the sedimentary load of the stream which
is moving sliding, being rolled or pushed along on the streambed by the tractive
force of the moving water. The particles are in contact with the bed practically
all of the time. The movement of particles has usually a discontinuity. The

65
Sediment Transport Technology

particles may roll or slide for some time, remain stationary for a while and again
start rolling or sliding. The mechanisms by which it moves can be varied and
complex.
b) Saltation load, which is the sediment bounced along the streambed by the flow
or by the impact of other particles. The initial impetus which launches a particle
into the flow may be due to the striking of one particle by another, the rolling of
one particle up over another, or the flowing of water other curved surface of a
particle thus producing a negative pressure.
c) Suspended load, which is the sediment that is supported by vertical components
of the velocities in turbulent flow while being carried forward in the stream by
horizontal components of these velocities for appreciable lengths of time. All
particles those are small enough to be in suspension shift up and down in the
flow and presumably move readily into and out of the bed layer.

In spite of such a classification, sediment load can be divided into two main categories
for convenience of the analysis and also for field measurements, namely, bed load and
suspended load. Contact and saltation loads are grouped together and called bed load.
Sum of bed load and suspended load gives the total load.

The rate of sediment transport can be determined either directly or indirectly by taking
into account the amount of sediment passing a section somewhere along the stream. The
direct method involves the determination of weight of sediment passing at a section in a
particular time. Whereas the indirect methods require the measurement of sediment
concentration, wetted area of cross-section and the velocity of the particles in motion. In
general, the direct and indirect methods are convenient for bed load and suspended load,
respectively.

2. Measurement of Suspended Sediment

The measurement of suspended sediment in a stream can be realized by using the
indirect method. In this method, the concentration and velocity measurements are
combined to obtain the suspended sediment. Therefore, samples taken by any sampler
should truly represent the sediment concentration of streams at the measurement points.
In order to fulfill this requirement, the velocity within the cross-section of the entrance
tip of a sampler should be equal to the stream velocity. For this purpose, the entrance
should be pointed against approaching flow.

2.1. Definitions of sediment concentration

Sediment concentrations are usually expressed in parts per million (ppm), i.e. ratio of
dry weight to total weight of the water sediment mixture.

2.1.1. Vertical and horizontal distribution of suspended sediment

Solid particles are generally heavier than the water. In spite of this fact, solid particles
can be transported in suspension by the water having vertical component of turbulent
flow. Concentration of these suspended materials in water has its minimum and
maximum values on water surface and near streambed, respectively. Therefore,
accuracy of the measurements depends not only on the type of samplers but also to the
determination of concentration and flow velocity. For this reason, suspended load

66
 Measurement of Sediment Load

measurement leads to determination of the vertical and horizontal concentration

distributions (Fig. 1). Vertical distribution of suspended sediment is given by,

2.5 ( ws / u* )
C ⎛h−z a ⎞
=⎜ ⎟ (1)
Ca ⎝ z h − a ⎠

where, h is the water depth, z is the vertical distance from streambed, a is equal to 2dp,
C is the sediment concentration at any distance z from streambed (PPM), Ca is the
sediment concentration at the distance a, and ws is the settling velocity of suspended
sediment particle.

2.1.2. Solid particle concentration

Concentration of solid particles in water can be expressed in three ways:
G
i. C ′ = s (2)
GT

Vertical depth (z)

Suspended sediment concentration (PPM)

(a) Vertical

iso-concentration lines

(b) Horizontal

Figure 1. Vertical and horizontal distribution of suspended sediment concentration

67
Sediment Transport Technology

Gs
ii. C ′′ = (3)
V
Gs
iii. C ′′′ = (4)
G

where, Gs is the dry weight of solid particles, GT is the total weight of solid-water
mixture, G is the water weight in mixture, V is the total volume of mixture.

2.1.3. Static concentration

This concentration is expressed by the relation given in Eq. (3) and can be used in three
different ways:
1. Static concentration at a point: This is a point concentration and expresses the
concentration given by Eq. (3) in a very small volume covering measuring volume.
Every arrow in Fig. 2a shows this concentration.
2. Static concentration in vertical: This is the solid particle concentration in a prism
volume having unit width and length along water depth (Fig. 2b) and can be
expressed by the following relation:

⎛ h ⎞ ⎛ h + h3 ⎞ ⎛ h + h4 ⎞ ⎛ h + h5 ⎞ ⎛h ⎞
C1 ⎜ h1 + 2 ⎟ + C 2 ⎜ 2 ⎟ + C3 ⎜ 3 ⎟ + C5 ⎜ 4 ⎟ + C5 ⎜ 5 + h6 ⎟
2⎠ ⎝ 2 ⎠ ⎝ 2 ⎠ ⎝ 2 ⎠ ⎝2
C= ⎝ ⎠ (5)
H

3. Static concentration along cross-section: This is defined as a concentration in a

prism having unit length with wetted area (Fig. 3), and given by,

C1 F1 + C 2 F2 + C3 F3
C= (6)
F
F = F1 + F2 + F3 (7)

h1
C1
h2
C2
h3
H C
C3
h4
C4
h5
C5
h6
(a) (b)
Figure 2. Distribution of vertical static concentration

68
 Measurement of Sediment Load

L1 L2 L3 L4
L2/2 L2/2 L3/2 L3/2

C1 C2 C3

F1 F2 F3

Figure 3. Static concentration over cross-section

where, F is denoted as area.

2.1.4. Average concentration against water discharge

This is the ratio of the amount of solid particle to water discharge. Average
concentration against discharge can also be defined in three ways as given in the
previous section; i.e. point, vertical, and cross-section.

The following relation can express averaged concentration over a cross-section if there
exist some concentration values for vertical:

∑C Q i i
CQ = i =1
n
(8)
∑Q
i =1
i

On the other hand, the following relationship can be written between the static and
averaged concentration:

Cs u
= (9)
CQ ws

where, Cs is static concentration, CQ is averaged concentration against discharge, u is

flow velocity, and ws is settling velocity of suspended material.

2.2. Nomination of sampling section

Sampling verticals can be nominated as expressed below depending on sampling
purpose.

2.2.1. Various aspects of suspended load sampling

Three aspects should be considered in order to realize the measurement of suspended
load of as explained below:
a) Location of verticals: Suspended load could be obtained by multiplying the
mean concentration and the discharge in the cross-section. Therefore, sampling
verticals in the cross-section should be nominated carefully in order to obtain

69
Sediment Transport Technology

correct measurement data. Sampling verticals are usually chosen by one of the
following methods:
i) Single vertical along cross-section: This method is used only for small
stream or in the case of the samples along the vertical can represent those of
overall cross-section. If there is information about thalweg, sampling vertical
is taken as the vertical passing through thalweg, otherwise mid-width of the
cross-section is considered as sampling verticals, Fig. 4 and Fig. 5.
ii) Three verticals over cross-section: This method is used when the sediment
distribution across the width is taken into account. The sampling verticals
are fixed at 1/4, 1/2, and 3/4 or 1/6, 1/2, and 5/6 of width as shown in Fig. 6
and Fig. 7.
iii) Four or more verticals with equal distance: The verticals are chosen with
equal distances.
iv) Verticals at middles of section of equal discharge: In this case, cross-section
is divided into three or more segment having equal discharge and the
verticals are placed their centroids (Fig. 8). Therefore, this method requires
knowledge of the distribution of discharge in the cross-section. Usually, this
led to prepare the relationships similar to the curves given in Fig. 9 between
water level and distance from the bank and relative discharge.

B B/2 B/2

Hmax

Figure 4. Single vertical along the maximum Figure 5. Single vertical along half-width
water depth of the cross-section

3B/4 5B/6
B/2
B/2

B/4 B/6

Figure 6. Three verticals along the cross- Figure 7. Three verticals along the cross-
section for B/4, B/2, and 3B/4 section for B/6, B/2, and 5B/6

70
 Measurement of Sediment Load

Q/3 Q/3 Q/3

Figure 8. Equal discharge verticals

10
Cumulative discharge as a percentage of total discharge

90

80

70

60

50

40

30
Gage
20 420 120 40
reading

10
Distance from the
6 12 24 30 36 42 48 54 bank (m)

Figure 9. Discharge variation with gage height

Both method (iii) and (iv) are used when the sediment investigations are important.

b) Methods of sampling: There are two main sampling methods, namely, depth
integration method and point integration method used in suspended sediment
measurements.
i) Depth integrating method: In this method, sampler is moved first down and
then up with a selected velocity related to water flow velocity that during
one cycle movement bottle of the sampler has to be fulfilled. Averaged
concentration along the verticals is obtained with this method of
measurement.

If water depth is less than 4-5 m, sampler can be moved with a constant
velocity from water surface to streambed and vice-versa. It is not necessary
to take the velocities equal moving down and up, but keeping it constant
along one way is recommended.

If the water depth is in between 4.5 m and 9-m, sampling can only be done
along one direction.

71
Sediment Transport Technology

When water depth is greater than 9 m, verticals are divided into some parts
and samplers are taken for each part separately.

Sampling velocities of the sampler, which will be used in depth integrating

method, are limited. The manufacturers give these velocities for each
sampler depending on the bottle volume, size of nozzle and flow velocity.
Permissible vertical velocities of samplers are given in Fig. 10 for different
three nozzle diameters.

Samplers taken by depth integrating method represent averaged

concentration against discharge. In this method, there is a region in which it
is not possible to sample because of sampler geometry as shown in Fig. 11.
That is why some corrections are needed.

ii) Point integrating method: In this method, samplers are taken at the points
along verticals. These sampling points can be denoted as below:
• One point on the surface
• One point at 60% of water depth
• Two points; one from surface and other from the bottom. In this case,
mean concentration should be obtained by,

Cbottom + C surface
C mean = (10)
2

• Three points; one from surface, second from mid-depth and third
from bottom.

C surface + Cbottom + C mid −depth

C mean = (11)
3

150

125
Filling time (s)

100

75
1/4"
50 3/16"
1/8"

25

0
0,00 0,50 1,00 1,50 2,00 2,50 3,00

Flow velocity (m/s)

Figure 10. Filling time for DH-49 sampler (Julien, 1995)

72
 Measurement of Sediment Load

Water surface

Sampled region
Flow d

Unsampled
region sampler River bed

(a)

Sampling
verticals
Water surface

Sampled
zone

Unsampled
zone
(b)
Figure 11. Measured and unmeasured sediment load (White, 1982)

• Two points; one from 80% and other from 20% of water depth,
Straub [2].

3 5
C mean = C1 + C 2 (12)
8 8

where, C1 and C2 are the concentration at 80% and 20% of water depth,
respectively.

c) Frequency of sampling: Variations in the sediment load of a stream occur due to

the variations in,
i) Storm characteristics,
ii) Size, shape, geological and topographical features of drainage area,
iii) Stream characteristics.

Therefore, it is not easy to specify a general sampling interval, which will be

satisfactory from the point of view of accuracy of data. However, regarding to the
present practices, the sampling interval can be fixed by depending on the purpose of

73
Sediment Transport Technology

collection of data and the nature of stream. Sediment load should be measured twice
every week; however the sampling intervals can also be varied from day to a week
during low and mean water or daily during floods. During the rising stage of a flood-
sampling interval should be varied from 30 minutes to 12 hours. On falling stage, the
interval can vary 2 hours to 24 hours.

2.3. Suspended load samplers

Suspended load samplers are designed and manufactured in such a form and geometry
that they do not disturb water flow during sampling. These samplers can be divided into
two main groups depending on their type of operations:
a) Instantaneous suspended load samplers: They have many limitations for field
use, therefore are now obsolete.
b) Time integrating suspended sediment samplers: These samplers can be
subdivided into two groups:

i) Point-integrating samplers are designed to collect through time a sample at a

given point in the stream vertical. Specially designed an air compression
chamber is added in the point-integrating samples to prevent initial sudden
inrush. In this design, the air chamber and the sample container are
interconnected and the air chamber has a permanent opening at the bottom.
The rotary valve can close or open the passage between the air chamber and
the sample container and also closes and opens the intake nozzle and the air
exhaust port. During the measurement, as the sampler is lowered to the
sampling depth, the intake nozzle and the air exhaust port are kept closed
and the pressure equalizing port open. As the water fills the sample
container, the air pressure in the chamber and the bottle are raised, and the
bottle is raised to the hydrostatic pressure. Then, the pressure equalizing part
is closed and the air exhaust parts opened to collect the sample with the
intake velocity equal to the undisturbed flow velocity. At the end of the
sampling time all the three parts are closed and the sampler lifted up. The
dried sample measures the time-averaged sediment concentration.

Point-integrating samplers such as the P-61, P-63, and P-72 are commonly
used to obtain information on sediment distribution along a vertical.

• US P-61 Sampler: The US P-61 is a point-integrating suspended

sediment sampler that has an electrically operated valve for starting and
stopping the collection of a sample. The sampler is lowered to the
desired depth in the water column, the sample collected by remotely
opening and closing the valve, and then the sampler is raised to the
surface for removal of the sample container. To eliminate a sudden
inrush of sample due to hydrostatic pressure, the diving bell principle is
used to balance the air pressure in the bottle with the hydrostatic pressure
at the nuzzle prior to opening the valve at the start of sampling. The
“diving belly” in the sampler is the body cavity behind and surrounding
the sample costumer. It is connected by ports through the valve system to
the surrounding stream and to the sample bottle (USGS, 2005).

74
 Measurement of Sediment Load

• US P-63 Sampler: This sampler can be used in stream velocities ranging

from 0.45 to 4.5 m/s and to depth of 54 m with a pint container.
Although the measurement principle of this sampler looks like to US-61
sampler, as it is used in deeper depths it is heavier than the previous one.
• US P-72 Sampler: This sampler uses 0.48 cm internal diameter nozzle
and can be used in stream velocities ranging from 0.45 to1.6 m/s. It can
be used to a depth of 22 m with a pint container.

Figure 12. US P-61 type point integrating sampler

Figure 13. US P-63 type point integrating sampler

Figure 14. US P-72 type point integrating sampler

75
Sediment Transport Technology

ii) Depth-integrating samplers: The depth-integrating sampler traverses down

and up with a selected constant velocity along the complete depth of the
stream and collects a sample having the average concentration in the
vertical. Therefore, a special attention has to be paid to the sampler velocity
in the vertical. The minimum and the maximum rates of traversing should be
determined for a sampler with different sized nozzles. Fig. 15 can be used to
obtain a suitable transit rate for a depth-integrating sampler with a one-pint
and a standard milk bottle as the container for the sediment water mixture.
The time used in traversing the stream vertical need not be the same in both
directions of travel. To keep the rate constant in any one direction is enough.
A control condition for the integrating sampler is that the sampler should be
filled to less than 90 percent capacity. The advantages of depth-integration
sampling are that a single sample provides a discharge-weighted
concentration for a complete sampling vertical- or in a very deep stream, for
a major fraction of a sampling vertical. The disadvantages of depth-
integrating sampling are the unsampled zone, which may require special
computation of unmeasured suspended sediment.

Depth integrating samplers are classified with their use in stream depths.
These depths are:
(1) Wadeable
(2) Depths to 5 m
(3) Depths to 10 m
(4) Depths to 35 m
(5) Depths to 70 m

Herein, a few example samplers are given.

• US DH-48 Sampler: This sampler is used in shallow flows (wadeable)

when the unit discharge does not exceed 1 m2/s (Fig. 16). The weight and
length of the sampler are 2 kg and 33 cm, respectively, and the container
has 400cc volume filled by 0.65 cm nozzle.
• US DH-49 and DH-59 samplers: They are designed for use in streams
not more than about 5 m in depth (depths to 5 m). These samplers are
suited for use with a cable suspension system. The samplers have
streamlined bodies weighing about 28 and 11 kg, respectively. Inside of
the body is suitable to accommodate round one-pint bottle sample
container. The heads of the samplers are drilled and tapped to receive the
intake nozzles. Three brass nozzles with 0.65, 0.48 and 0.32 cm diameter
bare can be supplied by hard to the head of the samplers.
• D-99 samplers: The US D-99 is a collapsible-bag suspended
sediment/water-quality sampler capable of collecting a 6-liter sample. It
is 39 in long, weighs 125 kg, and has a hollow cavity inside the sampler
body. It is fabricated from a bronze casting with a high-density
polyethylene tail. It has a hinged head that closes horizontally and holds
the nozzle holder and nozzle in place. All metal parts are plastic coated.
The bag is attached to the nozzle holder with a hook-and-loop strap. The
US D-99 sampler will collect acceptable flow weighted samples in

76
 Measurement of Sediment Load

stream velocities ranging from 0.9 to 4.5 m/sec. Extreme care should be
practiced when deploying the sampler at stream velocities above 3 m/sec.
The US D-99 sampler is capable of sampling to a depth of 2 m with a 0.8
cm internal diameter.

Figure 15. Maximum and minimum transit velocities for a depth-integrating sampler with a
one-point container

77
Sediment Transport Technology

Figure 16. US DH-49 type depth-integrating sampler

Figure 17. US D-99 type depth-integrating sampler

2.4. Determination of sediment concentration

After the sediment sample is taken, the volume and weight of the sample are determined
and the sample is filtered and dried in an oven. The concentration of suspension is
expressed in terms of weight of dry sediment per unit volume of suspension
(grams/liter, or kg/m3), or in percentage by weight, which is equal to the weight of dry
sediment divided by the weight of the suspension and multiplied by 100. The relation of
these two expressions is given below:

78
 Measurement of Sediment Load

1000
C1 = (13)
1 100 1
+ −
γ s γ 0C2 γ 0
1000
C2 = (14)
1000γ 0 γ 0
1+ −
C1 γs

where, Ws is weight of dry sediment in grams, W is weight of the suspension in grams

(sediment and water), V is volume of the suspension in liters (sediment and water), γs is
unit weight of sediment in g/cm3, and γ0 is unit weight of water=1 g/cm3. C1 (weight of
sediment per unit volume of suspension),

Ws (Grammes)
C1 =
V (litres)
Ws
C2 (% concentration of suspension by weight) = 100x (%)
W

If the variation of sediment concentration of any measurement points within the cross-
section is less than 20% of their arithmetic mean, the arithmetic mean will be adopted as
the average. Otherwise, the arithmetic mean of each vertical is computed and multiplied
by the corresponding discharge of the segment. The sum of these products, when
divided by the total discharge of the cross-section gives the average value of
concentration:

C=
∑C Q
i i
(15)
∑Q i

2.5. Determination of suspended sediment load

In general graphical integration method is used to evaluate suspended sediment load
from field data. For this purpose, distribution of water flow velocity and concentration
along the verticals has to be determined by field measurements.

For a given cross-section concentration and velocity as given in Fig. 18 and Fig. 19
suspended load can be computed as explained below:

I II III

U1 U2 U3

C1 C2 C3

Figure 18. Mean values of flow velocity and concentration in the case of depth-integration
method

79
Sediment Transport Technology

(Qs)Total

I II III
q a i = C1 U 1 h 1 q a 2 = C2 U 2 h 2 q a 3 = C3 U 3 h 3

Figure 19. Graphical integration for depth-integrating sampling

In the case of depth integration measurements, Fig. 18, unit suspended sediment load
along verticals can be obtained by,

qi = CiU i (16)

where, Ci and Ui are mean concentrations and velocities for each vertical, respectively.
Total suspended sediment load can be found by graphical integration along the channel
width.

In the case of point integrating measurements, Fig. 20, unit suspended sediment load
along the verticals can be computed by integrating (CiUi) terms over water depth.

h
qi = ∫ C U dy
y =a
i i (17)

where Ci and Ui are the point values of concentrations and velocities, respectively. This
integration can be evaluated by using either analytical method or graphical method.

For analytical method, variation of concentration and velocity along the water depth has
to be expressed in a mathematical form. As it is seen from Fig. 20 that both the
concentration and velocity are functions of water depth and to express these functions
mathematically is a hard work. Therefore, in general, graphical integration is used

I II III
U= f(y)

C= f(y)

Figure 20. Distribution of flow velocity and concentration in the case of point integrating
method

80
 Measurement of Sediment Load

instead of analytical integration. For this purpose, product of (CiUi) is obtained from the
values Ci and Ui values given in Fig. 20. So variations of (CiUi) along the verticals can
be drawn. The areas between vertical (CiUi) curves give unit suspended load transported
in unit width along the verticals. Total suspended load can be found by re-integrating
these unit discharges along channel width.

3. Measurement of Bed Load

Bed load may be defined as that part of sediment load, which is sliding along, rolling or
hopping on or near the riverbed. From this bottom layer of flow within which sediment
load has considered as bed load, the rate of sediment transportation may be determined
by placing a sampler on the stream bed which will trap all solid particles moving the
river bed during a specified period. The thickness of this bottom layer, which
determines the height of the opening of the sampler, cannot be very well defined.
Generally, the minimum thickness of the bottom layer is estimated as twice the
maximum grain size of bed material. In addition, maximum thickness is taken to be
minimum distance from the riverbed at which a suspended load can be operated.

Bed load is difficult to measure on account of the fluctuation in bed-load movement, as

riverbed material usually moves in the form of ripples, sandbanks or dunes. Also there
is considerable difficulty in constructing an apparatus that will lie flat on the bottom
without disturbing the riverbed and the movement of bed load. So far no satisfactory
equipment has been developed to meet all practical requirements, and in the following
statement, the various methods in use are only briefly discussed.

3.1. Selection of bed load sampler

Bed load sampler may be grouped into a few distinct types according to their
construction and their principle of operation as follows:
a) The basket type: Probably the oldest and the most common type of bed load
samplers are the box or basket type samplers. Fundamentally, this type consists
merely of a box or basket, made of meshed material, which is lowered to rest on
streambed with the upstream end opened to catch a sample of the moving
material within a certain time interval. The introduction of the sampler on the
streambed creates a considerable backpressure, which reduces the entrance
velocity and causes some of the coarse material to drop out before entering the
basket. At the same time, fine materials creeping along the bed is intensively
deflected as a result of deflection of the slow-moving current around the
sampler. Therefore, the rate of sediment caught by the sampler is smaller than
that actually moving on the riverbed. On the other hand, if there is tendency of
forward movement or digging of the sampler, the catch may be much more than
those resulting from bed load movement only. Therefore, that trap efficiency of
each sampler must be determined beforehand to obtain reliable results.

The box or basket type sampler is considered the most suitable for use in
mountainous streams with coarse gravel as bed load material. It is the smallest
type for entrance dimensions and therefore the least cumbersome. Fig. 21 shows
an example of a basket type bed load sampler used by Swiss Federal Authority
for Water Utilization.

81
Sediment Transport Technology

Figure 21. Basket type bed load sampler

b) The tray or pan type: This type consists of a flat pan or a tray shaped device with
baffles or slots to check the moving materials. Since this type also causes
obstruction to the stream and consequently reduction of entrance velocity and
movement of material, it must be calibrated to determine its efficiency.

The pan type samplers seem to be the best suited for cases with low rates of
movement on comparatively smooth sand bed, when the whole of the
transportation is concentrated at the bottom layer. Fig. 22 gives an example of
this type of bed load samplers sometimes called Polyakov Sampler.

Figure 22. Tray type bed load sampler (Polyakov sampler)

82
 Measurement of Sediment Load

c) Pressure difference type: This type is designed to overcome the objection of

decreased velocity and bed load movement at the entrance to the sampler. A
rational solution to the problem lays in the formation of a pressure drop the exit
of the apparatus just sufficient to overcome the energy losses, thus giving the
same entrance velocity as that in the undisturbed stream. This is accomplished
by designing the apparatus with a section diverging in a downstream direction,
which will cause suction at the entrance. With such a diverging section the
velocity towards the downstream end of the sampler is decreased and some of
the material is deposited. Thus if this section is of sufficient length the necessity
of a collecting screen at the exit end may be eliminated. Some of the samplers of
this type have a collecting screen, while others have only baffles to check the
movement of the material.

For sandy beds the pressure difference type of sampler seems to be the most
satisfactory, especially when the entrance section is small and the frame flexible
enough to guarantee a snug fit against the irregularities of the bed. The Arnhem
or Dutch Sampler designed on this principle is shown in Figs. 23, 24, and 25
show such a sampler manufactured by DSI Technical Research and Quality
Control Department.

Figure 23. Pressure difference type bed load sampler (Arnhem sampler)

83
Sediment Transport Technology

Figure 24. Bed load sampler manufactured at TRQC (perspective view)

Figure 25. Bed load sampler manufactured at TRQC (top view)

Halley and Smith (1971) introduced a pressure difference bed load sampler that
is structurally modified version of the Arnhem Sampler. The Halley-Smith bed
load sampler has an expanding nozzle, sample bag and frame. The sampler was
designed to use in flows with mean velocities up to 3 m/s and sediment sizes
from 2 to 10 mm. The sampler has a square 7.62-cm entrance nozzle and a 46-
cm long sample bag constructed from 0.25-mm mesh.

d) The slot type: Another system, which has been used in special cases, is the
construction of slots in the canal or streambed, allowing the moving material to
drop into them. In one type the material is piped to bank from the slots while in
another type horizontal sliding doors are provided. Such special arrangements
are considered suitable only for small stream or canals.

3.2. Calibration of sampler

The various types of bed load samplers mentioned above require careful calibration
before they can be used to measure actual bed load samples in natural streams or canals.
For this purpose they are generally calibrated in laboratory flume where the bed load is
directly measured in a sump at the end of the flume or in small channels as mentioned
above.

84
 Measurement of Sediment Load

Hydraulic efficiency of a bed load sampler is defined as the ratio of the mean velocity
of water discharge through the sampler to mean velocity of water discharge which
would have occurred through the area occupied by the opening in the sampler nozzle
had the sampler not been there.

Hydraulic efficiencies of a basket type and tray type are 0.45 and 0.70, respectively.
Hydraulic efficiency of the pressure difference type sampler manufactured by DSI
Technical Research and Quality Control Department was found by laboratory studies.
In the laboratory studies, velocity profiles were measured in the sampler entrance and
the same location the sampler not been there. Hydraulic efficiency of the sampler has
the values in the range 1.20–1.50 for the velocity between 0.40–0.60 m/s. Hydraulic
efficiency of the Halley-Smith bed load sampler is approximately 1.54 (Garde, 1977).

The sampling efficiency of a bed load sampler is defined as the ratio of the weight of
bed load collected during a sampling time to the weight of bed load that would have
passed through the sampler within the same time, had the sampler not been there.
Ideally, this ratio should be 1.0. The manufacturers give the sampling efficiency for
each sampler.

3.3. Methods and frequency of bed load measurements

The bed load sampler is usually held in position by a rod if the stream is shallow, or by
a cable from boat, from a trolley or a pulley running a cable spanned across the river, or
from bridge. A winch also provided for lifting and lowering the sampler into position.

Measurement of bed load is taken at sites selected for the measurements of discharge
and the condition specified for the selection of site for the letter is also considered
applicable to the former.

As in the case of measurement of discharge, it is desirable to have the bed load

measured at various stages of flow on which a rating curve showing the relation
between discharge and bed load may be prepared. However, measurement of bed load
during higher stages of flow during flood period is not only difficult but also often
impossible, in particular on large rivers, on account of difficulties encountered in
keeping in keeping a boat as well as the sampler in position. No standards can be given
be given be given for the frequency of measurement, except to say "as frequently as
conditions and available resources permit".

The number of measuring points in a cross-section should generally be from 3 to 10. As

the rate of bed load movement fluctuates at any given point even if the river flow is
steady, it is advisable to keep the sampler on the river bed for as long a period as
possible, until approximately one third of the total capacity of the collecting basket or
box is filled with sediment.

The bed load collected in the sampler is dried and weighed. The dry weight, when
divided by the time taken for measurement and the width of the sampler, gives the rate
of bed load movement per unit width of riverbed per unit time at the point of
measurement. With the rate of bed load movement as zero at the river bank, a curve
showing the rate of movement across the river bed can be constructed from which the
total rate of bed load movement over the entire cross section can be computed.

85
Sediment Transport Technology

3.4. Selection of sampling section

The followings are the factors should be considered for the selection of sampling cross-
sections:
a) Sampling section has to be such a place where flow velocity measurement could
be done easily.
b) Water flow should be uniform around the sampling section. There should not be
any bend at the upstream of the sampling section in a safe range.
c) There should not be any secondary stream before sampling section.
d) Sampling section has to be a place where water discharge should be constant
during sampling.

3.5. Matters for sampling

The following matters have to be considered during sampling in order to obtain a
reliable result.
a) Bed load measurement might be performed at different 3-10 points for each
cross-section.
b) Sampler should be placed on the streambed in such a way that the bed forms will
not be disturbed.
c) At least two different samples should be taken. Measured bed load data should
be obtained by averaging from these data.
d) Sampler might be kept on streambed as a long time as possible. Sampling bed
material volume should never be greater than 1/3 of the total sampler volume.
e) Samples should be taken at the different water stages and by a short time
interval.
f) Samples have to be transported to the laboratory without loss of contents.
g) Samples have to be labeled by including the following information.

• For suspended load:

1. Name of the river
2. Name of the sampling section
3. Number of sampling vertical
4. Sampling depth (only for point-integrating)
5. Water depth at the sampling vertical
6. Water flow velocity at the sampling point (only for point-integrating)
7. Mean water flow velocity at the sampling vertical (only for point-integrating)
8. Water stages if there exist a gauge station, otherwise water discharge during
sampling
9. Water temperature
10. Name of the person conducting sampling test
11. Date and time of sampling

• For bed load:

1. Name of the river
2. Name of the sampling section
3. Sampling vertical
4. Water depth and sampling verticals
5. Water stages if there exist a gauge station, otherwise water discharge

86
 Measurement of Sediment Load

6. Mean water flow at the sampling vertical

7. Resting time of the sampler on the bed
8. Date and sampling time
9. Name of the person conducted the test

In addition to these information the following knowledge are also be collected:

1. Slope of water surface
2. Granulometric analysis of bed material
3. Two or three cross-section taken from upstream and downstream side each and also
sampling section

References

1. A.A. Kalinske, “Criteria for determining Sand-Transport by surface creep and

saltation”, Trans. Amer. Geophys Union, Pt. 2, pp 639-643.
2. Acaroğlu, E.R. and Demiröz E., “A comparative study of total sediment load
theories in natural waterways”, ODTÜ İnş. Müh. Böl., Rapor no: 70-03, (1971).
3. Boiten, W., “Hydometry”, IHE Helft Lecture Note Series, A.A. Balkema
Publishers, The Netherlands, (2003).
4. Corps. of Eng. US dept. of Army, “Equipment used for sampling bed load and bed
material” Report No. 2, St. Paul district, Minn., (1940).
5. Çekmeceli, A.Ö., “Akarsuların Rusubat Yükünün Ölçümü ve Analizi”, DSİ
Yayınları, No: 617, (1969).
6. Emmet, W.W., “Measurement of bed load in rivers”, Symposium on erosion and
sediment transport measurement, 22-26 June, Florence, Italy, pp. 3, (1981).
7. Corps. of Eng. US dept. of Army, “Field practice and equipment used in sampling
suspended sediment”, Report No. 1, St. Paul district, Minn.
8. Garde, R.R. and Ranga Raju, K.G., “Mechanics of sediment transportation”, Wiley
Eastern Ltd., New Delhi, India, (1977).
9. Einstein, H.A., “The bed load function for sediment transportation in open channel
flow”.
10. Hubbell D.W., “Apparatus and techniques for measuring bed load”, US Geological
Survey, Water Supply Paper 1748, (1964).
11. Raudkivi, A.J., “Loose Boundary Hydraulics”, Pergamon press pub., (1967).
12. Kahl Scientific Instrument Corp., “Depth-Integrating Suspended Sediment
Sampler”, Instruction Manual.
13. Şentürk, F., “Nehir Hidroliği”, DSİ yayınları, (1968).
14. Task Committee on Preparation of Sediment Manual, “Sediment Measurement
Techniques”, Proc. of ASCE, pp. 1477-1511, Sept. (1969).
15. TCP, “Sediment Measurement Techniques”, Fluvial Sediment, JHD, Proc., ASCE,
Vol. 95, No. HY-5, Sept. (1969).
16. USGS, “A Guide to the Proper Selection and Use of Federally Approved Sediment
and Water Quality Samplers”, Virginia, (2005).
17. Vanoni, V.A., “Transportation of suspended sediment by water”, Trans. Amer. Soc.
of Civil Eng., 111: pp. 67-133, (1946).

87
88
 FLOW RESISTANCE IN STREAMS

Dr. Fuat ŞENTÜRK

Retired Director of
DSİ Technical Research and Quality Control Department
06100, Ankara, Turkey

(edited by: Assoc. Prof. Dr. Mehmet Ali KÖKPINAR)

e-mail: mkokpinar@superonline.com

1. Introduction
In the flow of a fluid energy is being continually dissipated. This occurs because the
fluid has to do work against resistance originating in fluid viscosity. Whether the flow is
laminar or turbulent, the basic resistance mechanism is the shear stress by which a slow-
moving layer of fluid exerts a retarding force on an adjacent layer of faster moving
fluid. The laws of flow resistance are essentially similar in open channels as in closed
pipes running full. The resistance equation can be derived by balancing the retarding
shear force at the boundary against propulsive force acting in the direction of flow. This
force, in an open channel, is essentially supplied by the weight of the flowing water
resolved down a slope, whereas in the pipe flow, this propulsive force is supplied by a
pressure gradient.

However, the boundary condition in pipes with circular section the shear stress
restraining the fluid motion is uniformly distributed around the boundary of the cross
sections. In an open channel, two factors which tend to make the boundary shear
nonuniform: the existence of a free surface on which the shear is negligibly small, and
the wide variety of possible cross section shapes, each with its own distribution of shear
stress around the solid boundary.

The studies (Chow, 1959; Henderson, 1966) showed that in a cross section of a channel,
secondary currents moving from the banks towards the center of the stream effect the
velocity distribution. Because of these currents, weight component is not unbalanced
and fluid is accelerated by low velocities at the bank and high velocities at the center of
the channel. One can describe this exchange of momentum as the equivalent of a shear
resistance, which balances the weight component. Because of these secondary currents,
the shear-stress distribution on the boundaries of an open channel may be very
complicated. However, the resistance equation in terms of mean shear stress can be
written even though the shear-stress distribution is unknown. As described above, the
factors governing flow in alluvial channels are numerous and complex and their
interactions are not well known.

2. Theoretical Method
Major variables affecting the flow are (Simons and Şentürk, 1992):

89
Sediment Transport Technology

φ (U , d , S , ρ , µ , g , D,σ , ρ s , S p , S R , S c , f s ) = 0 (1)

where, U is the flow velocity, d is the depth, S is the slope of the energy grade line, ρ is
the density of water, µ is the dynamic viscosity of water, g is the gravitational
acceleration, D is the representative diameter of bed material, σ is the standard
deviation of the size distribution of bed material, ρs is the density of sediment, Sp is the
shape factor of the sediment, SR is the shape factor of the channel reach, Sc is the shape
factor of the channel cross section, and fs is the seepage force in the bed and banks of
the channel.

The influences of some of these factors on resistance to flow are not known but are
difficult consider because of their complex interactions with other variable. With the
knowledge of these circumstances, Eq. 1 can be transformed into (Simons and Şentürk,
1992, Şentürk, 1994),

U ⎛ D 1 τ ⎞
= φ ⎜⎜ Re , , , ' ⎟⎟ (2)
U∗ ⎝ d R w γ sD ⎠

where, U/U* is the measure of resistance, U / gRS , Re is the Reynolds number, UR/ν,
D/d is the relative roughness, Rw is the Reynolds number of the particle (related to fall
velocity, w), wD/ν, and τ γ s' D is the Froude number related to particle.

With a rigid boundary, γ s' represents the specific weight of the boundary particle, and its
fall velocity w. If these two parameters are assumed to be infinity, then τ γ s' D and 1/Rw
are both zero and Eq. 2 can be simplified as,

U ⎛ D⎞
= φ ⎜ Re , ⎟ (3)
U∗ ⎝ d⎠

Then U/U* can be transformed to

U U 1 F
= = r (4)
U∗ gR S S

where Fr = U gR is the Froude number of the flow. Substituting the value of U/U*
from Eq. 3 and rearranging yields:

⎛ D⎞
S = Fr2 f 1 ⎜ Re , ⎟ (5)
⎝ d⎠

or

n
⎛D⎞
S = F f 2 (Re )⎜ ⎟
r
2
(6)
⎝d⎠

90
 Flow Resistance in Streams

Many resistance formulas are of this form.

3. Formation and Magnitude of Resistance to Flow in Rigid and Loose Boundary

Channels
3.1. Surface drag and form drag
In the case of a wide channel with a plane bed subjected to turbulent flow, the resistance
to flow reduces to a function of D/d or to skin friction or surface drag. When bed forms
occur, a form drag must be added to the surface drag. To express surface drag, two
general approaches may be utilized.

1. A certain fraction of energy is consumed overcoming the resistance due to surface

drag, and remainder is used to overcome the form drag. This idea was generally
developed by Meyer-Peter and Müller (1948). If S ′ corresponds to the fraction of
energy slope required to overcome surface drag and S ′′ corresponds to the fraction
required to offset form drag, it is possible to write,

S = S ′ + S ′′ (7)

2. Considering a prism of flow over the bed of a wide river (Fig. 1), it is possible to
define the hydraulic radius as,

bh
R= (8)
b

If the numerator and denominator of this ratio are multiplied by l in such a way that
b·l=A0=1 and b·h·l = V0;

bhl V0 V0
R= = = (9)
bl A0 1

where b·l = A0 is the unit area and V0 is the volume superimposed on this unit area. Then
the hydraulic radius represents the volume of a rectangular prism with a height equal to
the depth of water. This concept of R was first given by Einstein.

Figure 1. Visualization of hydraulic radius

91
Sediment Transport Technology

Here h is defined as the part of this volume required to dump the turbulence generated
by the boundary. Then if V′ corresponds to the necessary volume for dumping the
turbulence of surface friction, then V ′′ can be defined as required volume to dump
turbulence attributed to bed forms. With this definition,

V = V ′ + V ′′ (10)

or from Fig. 1b,

R = R ′ + R ′′ (11)

It can be concluded from this definition that, as the boundary resistance increases then R
increases. The increase in S caused by increase in roughness is more difficult to
understand because S = f(various factors) and it can be kept constant by changing R.
Therefore, Eq. 11 is considered basic in various relations developed.

From Eq. 11, the shear stress equation can be developed:,

τ 0' = γR ′S (12)
τ = γR ′′S
''
0 (13)

where, τ 0' is the fraction of the total shear taken by the skin friction or surface drag and,
τ 0'' is the fraction of the total shear taken by the bed forms and called as the form drag.
The total shear can be defined as

τ 0 = τ 0' + τ 0'' (14)

It is known from turbulence theory that

τ0
U∗ = (15)
ρ

where U* is the friction velocity and,

U ∗2 = u ′ν ′ (16)

The value of U* represents a variation of the fluctuation of turbulence around a point

near the boundary of a channel. From Eq. 15:

U ∗ = gRS (17)
U ∗' = gR ′S (18)
U ∗'' = gR ′′S (19)
U ∗2 = U ∗'2 + U ∗''2 (20)

92
 Flow Resistance in Streams

For a channel with a flat bed, R = R´ because R ′′ = 0 and

U ∗ = U ∗' (21)

3.2. Common parameters for describing resistance to flow

Three most common parameters for describing the resistance to steady uniform flow
are:
1. The Darcy-Weisbach friction factor f
2. The Chezy resistance factor C
3. The Manning roughness coefficient n

The Darcy-Weisbach formula primarily developed for flows in pipes, states that:

L U2
hf = f (22)
D 2g

where hf is the friction loss, f is the friction factor, L is the length of the pipe, D is the
diameter of the pipe, U is the mean velocity of flow, and g is the acceleration of
gravity.

Since D = 4R and energy gradient SE= hf/L, Eq. 22 can be written in terms of friction
factor as

8 gRS E
f = (23)
U2

where R is the hydraulic radius. Since U ∗2 = gRS E , Eq. 23 yields,

U 8
= (24)
U∗ f

Eq. 24 can be applied to flow in open channels, and sometimes presented as

Fr 8
= (25)
SE f

or

8S E
f = (26)
Fr2

The Chezy formula is expressed as

U = C RS E (27)

93
Sediment Transport Technology

where C is a resistance factor. The energy slope SE can be computed from

S E = S w − Fr2 ( S w − S b ) (28)

where Sw is the slope of the water surface Fr2 = U 2 gR , U is the mean velocity, R is the
hydraulic radius associated with SE, and Sb is the slope of the channel bottom. The use
of SE is recommended because it reflects more accurately the state of flow at a particular
location.

The dimension of Chezy’s resistance factor shown in Eq. 29,

L1 / 2
C= (29)
T

Hence,

C = C′ g (30)

Thus,

U = C ′ g RS f = C ′U ∗ (31)

in Eqs. 30 and 31, C´ is dimensionless. Substituting the value of U from Eq. 27 in Eq.
24 yields,

C 8
= (32)
g f

or

8
C′ = (33)
f

In general, C g is used to describe the resistance to flow in open channels.

Manning formula in metric system is expressed as

1 2 / 3 1/ 2
U= R SE (34)
n

and in the English system,

1.486 2 / 3 1 / 2
U= R SE (35)
n

94
 Flow Resistance in Streams

where n is the Manning roughness coefficient. The Chezy resistance coefficient C can
be expressed as a function of n. That is,

1.486 1 / 6
C= R (36a)
n

in English system or

R1 / 6
C= (36b)
n

in metric system.

A similar expression between C and f can be obtained by using Eq. 32.

C 1
= (37)
8g f

Taylor and Brooks (1962) suggested that f can be divided into two parts, that due to the
grain roughness

8 gR ′S E
f′= (38)
U2

and that due to form roughness

8 gR ′′S E
f ′′ = (39)
U2

In this case, the total f

8 gRS E
f = = f ′ + f ′′ (40)
U2

Using the same reasoning, Bajorunas (1952) assumed that the n is composed of two
parts

n = n ′ + n ′′ (41)

where, n´ is due to the grain roughness, and n ′′ is due to form roughness. From Eqs. 34
and 14, and τ = γS (R ′ + R ′′) , the relation

3/ 2
⎛ n′′ ⎞
R = R ′⎜1 + ⎟ (42)
⎝ n′ ⎠

or

95
Sediment Transport Technology

3/ 2
⎛n⎞
R = R ′⎜ ⎟ (43)
⎝ n′ ⎠

• The Darcy-Weisbach’s friction parameter f is used for both open channel and
pressure flow.
• The Manning’s n gives good results for fully rough and smooth conditions in rigid
boundary channels. However, it is less satisfactory in the transition range and for
alluvial boundary flow. Its value is highly dependent upon the form of bed
roughness.

4. Resistance in Open Channels-Flat Bed and Rigid Boundary

The simplest way to solve the resistance problem is by assuming the form drag and the
wall effect equal to zero. This condition is analogous to very wide bed a rigid, plane
bottom.

Number of formulas have been developed for this case and summarized by (Simon and
Şentürk, 1992; Şentürk, 1994) and given below.

4.1. The semi-logarithmic resistance relations

The semi-logarithmic functions of resistance to flow were developed based on Prandtl’s
(1926) mixing length theory.

⎛ y ⎞
u = ⎜⎜ 8.5 + 5.75 log ⎟⎟U ∗ (44)
⎝ ks ⎠

and

⎛ yU ∗ ⎞
u = ⎜ 5.5 + 5.75 log ⎟U ∗ (45)
⎝ ν ⎠

where ks is the equivalent sand roughness, y is the distance from the wall where U is
measured. Eq. 44 was developed for the resistance to flow of a pipe with rough wall in
turbulent regime and Eq. 45 was developed for the resistance to flow of a pipe with
smooth wall in turbulent regime. These equations were integrated for various shapes of
cross sections (Simon and Şentürk, 1992; Şentürk, 1994) and given below.

• Circular cross section, smooth wall:

U RU ∗
= 3.5 + 5.75 log (46)
U∗ ν

• Very large rectangular cross section, smooth wall:

U RU ∗
= 3.0 + 5.75 log (47)
U∗ ν

96
 Flow Resistance in Streams

• Circular cross section, rough wall:

U R
= 6.5 + 5.75 log (48)
U∗ ks

• Very large rectangular cross section, rough wall:

U R
= 6.0 + 5.75 log (49)
U∗ ks

• Rectangular cross section, rough wall and free surface:

U R
= 6.25 + 5.75 log (50)
U∗ ks

These concepts were applied to open channel flow by Zegzhda (1939) and others and
they studied the Darcy-Weisbach parameter f to apply these concepts to open channel
flow. The parameter f was expressed in three categories.

1. The resistance to flow in open channels with fully developed roughness:

1 ⎛ R ⎞ k sU ∗
= c log⎜⎜ a ⎟⎟ where > 70 (51)
f ⎝ ks ⎠ ν

The resistance in hydraulically smooth channels can be expressed as

1 ⎛ f ⎞ k sU ∗
= c log⎜ Re ⎟ where ≤5 (52)
f ⎜ b ⎟ ν
⎝ ⎠

2. For natural rough (non-uniform grains) boundaries in the transition zone between
hydraulically smooth boundaries and boundaries with fully developed roughness:

1 ⎛ R f ⎞
= c log⎜ a + Re ⎟ (53)
f ⎜ ks b ⎟
⎝ ⎠

In general,

f = f1(ks, Re, Sc, a, b, c) (54)

For open channels:

c = 2.0 and 2.03

a = f2(c, Sc)
b = f3(c, Sc)

97
Sediment Transport Technology

Table 1 shows the values of a and b for given values of c and cross sectional shape. In
Tables 2 and 3 Zegzhda (1939) and Ackers (1958) suggested ks values for rigid
boundary conduits. For sand bed Einstein (1950) suggested ks ≅ D65, Meyer-Peter
(1948) suggested ks ≅ D90, Simon and Richardson (1966) ks ≅ D85. These differences
could be related to particle size distribution.

Table 1. Coefficients for determining f

Channel Shape c a b Investigators
Wide 2.03 11.09 3.41 Keulegan (1938)
Wide 2.00 12.62 2.98 Keulegan (1938)
Wide 2.03 12.20 3.033 Thijsse 81949)
Wide 2.14 ------- 7.17 Sayre and Albertson (1961)
Wide 2.00 12.59 0.69 Graf (1964)
Rectangular 2.00 11.55 ------ Zegzhda (1938)
Rectangular 2.00 ------- 3.86 Tracey and Lester (1961)
Rectangular 2.03 ------- 4.37 Tracey and Lester (1961)
Trapezoidal 2.03 12.27 3.09 Keulegan (1938)
Trapezoidal 2.00 13.99 2.27 Keulegan (1938)
Circular 2.03 14.80 2.50 Thijsse 81949)
Circular 2.00 14.83 2.52 Colebrook (1938) and Bretting (1948)

Table 2. Values of ks in feet (after Zegzhda, 1938)

0.001- Exceptionally smooth concrete surface faced with cement plaster and
reinforced, expansions joints and other seams worked flush with the
surface
0.0016- Concrete cast in lubricated steel molds, with carefully smoothed or
pointed seams and joins
0.0016 to 0.0049- Smooth –finished gunite
0.0038- Ordinary concrete faced with cement plaster and rubbed down, or
concrete cast in wooden forms, not faced or rough-faced
0.0082- Short lengths of concrete pipe of small diameter without special facing
of butt joints
0.014- Roughly made concrete conduits
0.01or 0.033- Untreated gunite

Table 3. Values of ks in feet (after Ackers, 1958)

0.0005- Concrete (monolithic construction against oiled steel forms with no
surface irregularities)
0.002- Wood-stave pipes, planed-plank flumes, and concrete (monolithic
construction against steel forms, wet-mixe, or spun-precast pipe).
Smooth troweled surfaces. Glazed sewer pipe
0.005- (Monolithic construction against rough forms or cement-gun surface).
Glazed brickwork
0.01- Precast pipes with mortar squeeze at joints.
0.02- Rubble masonry
0.1- Straight, uniform earth channels

98
 Flow Resistance in Streams

4.2. Manning’s formula

The average velocity in open channels is expressed

1 2 / 3 1/ 2
U = R SE (55)
n

This equation defines the resistance to flow. Eq. 55 can be transformed to

U R1 / 6
= (56)
U∗ n g

The dimension of n is,

n = TL − 1 / 3 (57)

According to Eq. 57, n = f(t). To overcome this difficulty, it is generally assumed that
n = n ′ / g and substitute this into Eq. 57 yields,

n ′ = L1 / 6 (58)

if n´ is called n, then Eq. 58 transforms to

n = L1 / 6 (59)

Similar transformation can be written for Eq. 50 as

U ⎛R ⎞
= 5.75 log⎜⎜ 12.21⎟⎟ (60)
U∗ ⎝ ks ⎠

Combining Eq. 60 and Eq. 56 yields

( R / ks )
1/ 6

n g = n′ = k s1 / 6 (61)
5.75 log 12.21(R / k s )

or

1/ 6
n′ = n = L (62)

Also Eq. 56 can be written as

1/ 3
gn 2 ⎛D⎞
S = Fr2 ⎜ ⎟ (63)
D1 / 3 ⎝R⎠

Comparing with Eq. 6, Eq. 63 shows that

99
Sediment Transport Technology

n = f ( Re ) (64)

However, n not only changes with Reynolds number but is also function of:
1. Depth of flow,
2. Sediment transport (bed load, suspended load),
3. Floating debris,
4. Cross section geometry,
5. Geometric pattern of scouring and silting in the channel,
6. Size of bed material,
7. Type and extent of vegetation on the banks,
8. Water temperature, and
9. Wind direction and magnitude.

Table 4 shows the detailed list of suggested n values.

4.3. The Chezy formula

The drag force in terms of the dynamic variables of resistance, and velocity may be
expressed as,

f
τ0 = ρU 2 (65)
8

Substituting this relationship to Eqs. 17 and 15 and applying to flow in open channels
by evaluating the resistance coefficient f for various flow characteristics and channel
shapes,

8γ
U = RS E (66)
fρ

or

U = C RS E (67)

where

8γ
C= (68)
fρ

The term C is called Chezy resistance coefficient and has dimensions,

L1 / 2
C= (69)
T

Scaling this relationship by C´

100
 Flow Resistance in Streams

Table 4. Values of the roughness coefficient n (After Chow, 1959)

Type of Channel and Description Minimum Normal Maximum

CLOSED CONDUITS FLOWING PARTLY FULL

METALS
Brass, smooth 0.009 0.010 0.013
Steel
Lockbar and welded 0.010 0.012 0.014
Riveted and spiral 0.013 0.016 0.017

Cast iron
Coated 0.010 0.013 0.014
Uncoated 0.011 0.014 0.016

Wrought iron
Black 0.012 0.014 0.015
Galvanized 0.013 0.016 0.017
Corrugated metal
Subdrain 0.017 0.019 0.021
Storm drain 0.021 0.024 0.030
NONMETALS
Lucite 0.008 0.009 0.010
Glass 0.009 0.010 0.013
Cement
Neat surface 0.010 0.011 0.013
Mortar 0.011 0.013 0.015
Concrete
Culvert, straight and free of debris 0.010 0.011 0.013
Culvert with bends, connections, and some debris 0.011 0.013 0.014
Finished 0.011 0.012 0.014
Sewer with manholes, inlet, etc., straight 0.013 0.015 0.017
Unfinished, steel form 0.012 0.013 0.014
Unfınished, smooth wood form 0.012 0.014 0.016
Unfinished, rough wood form 0.015 0.017 0.020
Wood
Stave 0.010 0.012 0.014
Laminated, treated 0.015 0.017 0.020
Clay
Common drainage Lue 0.011 0.013 0.017
Vitrified sewer 0.011 0.014 0.017
Vitrifıed sewer with manholes, inlet, etc. 0.013 0.015 0.017
Vitrified subdrain with open joint 0.011 0.016 0.018
Brickwork
Glazed 0.011 0.013 0.015
Lined with cement mortar 0.012 0.015 0.017

101
Sediment Transport Technology

Table 4. Values of the roughness coefficient n (After Chow, 1959) (Continued)

Type of Channel and Description Minimum Normal Maximum

Sanitary sewers coated with sewage slimes with
0.012 0.0 13 0.016
bends and connections
Paved invert, sewer, smooth bottom 0.016 0.019 0.020
Rubble masonry, cemented 0.018 0.025 0.030
LINED OR BUILT-UP CHANNELS

METALS
Smooth steel surface
Unpainted 0.011 0.012 0.014
Painted 0.012 0.013 0.017
Corrugated 0.021 0.025 0.030
NONMETALS
Cement
Neat surface 0.010 0.011 0.013
Mortar 0.011 0.013 0.015
Wood
Planed, untreated 0.010 0.012 0.014
Planed, creosoted 0.011 0.012 0.015
Unplaned 0.011 0.013 0.015
Plank with battens 0.012 0.015 0.018
Lined with roofing paper 0.010 0.014 0.017
Concrete
Trowel finish 0.011 0.013 0.015
Float finish 0.013 0.015 0.016
Finished, with gravel on bottom 0.015 0.017 0.020
Unfinished 0.014 0.017 0.020
Gunite, good section 0.016 0.019 0.023
Gunite, wavy section 0.018 0.022 0.025
On good excavated rock 0.017 0.020 -.---
On irregular excavated rock 0.022 0.027 -.---
Concrete bottom float finished with sides of
Dressed stone in mortar 0.015 0.017 0.020
Random stone in mortar 0.017 0.020 0.024
Cement rubble masonry, plastered 0.0 16 0.020 0.024
Cement rubble masonry 0.020 0.025 0.030
Dry rubble or riprap 0.020 0.030 0.035
Gravel bottom with sides of
Formed concrete 0.017 0.020 0.025
Random stone in mortar 0.020 0.023 0.026
Dry rubble or riprap 0.023 0.033 0.036
Brick
Glazed 0.011 0.013 0.015
In cement mortar 0.012 0.015 0.018

102
 Flow Resistance in Streams

Table 4. Values of the roughness coefficient n (After Chow, 1959) (Continued)

Type of Channel and Description Minimum Normal Maximum

Masonry
Cemented rubble 0.017 0.025 0.030
Dry rubble 0.023 0.032 0.035
Dressed ashlar 0.0 13 0.0 15 0.017
Asphalt
Smooth 0.013 0.013
Rough 0.016 0.016
Vegetal lining 0.030 --- 0.500

EXCAVATED OR DREDGED

Earth, straight and uniform

Clean, recently completed 0.016 0.018 0.020
Clean, after weathering 0.018 0.022 0.025
Gravel, uniform section, clean 0.022 0.025 0.030
With short grass, few weeds 0.022 0.027 0.033
Earth, winding and sluggish
No vegetation 0.023 0.025 0.030
Grass, some weeds 0.025 0.030 0.033
Dense weeds or aquatic plants in deep channels 0.030 0.035 0.040
Earth bottom and rubble sides 0.028 0.030 0.035
Stony bottom and weedy banks 0.025 0.035 0.040
Cobble bottom and clean sides 0.030 0.040 0.050
Dragline-excavated or dredged
No vegetation 0.025 0.028 0.033
Light brush on hanks 0.035 0.050 0.060
Rock cuts
Smooth and uniform 0.025 0.035 0.040
Jagged and irregular 0.035 0.040 0.050
Channels not maintained, weeds and brush uncut
Dense weeds, high as flow depth 0.050 0.080 0.120
Clean bottom, brush on sides 0.040 0.050 0.080
Same, highest stage of flow 0.045 0.070 0.110
Dense brush, high stage 0.080 0.100 0.140

NATURAL STREAMS2

MINOR STREAMS (TOP WIDTH AT FLOOD STAGE <100 FT)

Streams on plain
Clean, straight, full stage, no rifts or deep pools 0.025 0.030 0.033
Same as above, but more stones and weeds 0.030 0.035 0.040
Clean, winding, some pools and shoals 0.033 0.040 0.045
Same as above, but some weeds and stones 0.035 0.045 0.050

103
Sediment Transport Technology

Table 4. Values of the roughness coefficient n (After Chow, 1959) (Continued)

Type of Channel and Description Minimum Normal Maximum

Streams on plain, Continued
Clean winding, some pools and shoals, but some weeds
and stones, Iower stages, less effective slopes and cross
section 0.040 0.048 0.055
Clean winding, some pools and shoals, weeds and more
stones 0.045 0.050 0.060
Sluggish reaches, weedy deep pools 0.050 0.070 0.080
Very weedy reaches, deep pools, or floodways with
heavy stand of timber and underbrush 0.075 0.100 0.150
Mountain streams, no vegetation in channel, banks
usually steep, trees and brush along banks submerged at
high stages
Bottom: gravels, cobbles and few boulders 0.030 0.040 0.050
Bottom: cobbles with large boulders 0.040 0.050 0.070
FLOODPLAINS
Pasture, no brush
Short grass 0.025 0.030 0.035
High grass 0.030 0.035 0.050
Cultivated areas
No crop 0.020 0.030 0.040
Mature row crops 0.025 0.035 0.045
Mature field crops 0.030 0.040 0.050
Brush
Scattered brush, heavy weeds 0.035 0.050 0.070
Light brush and trees in winter 0.035 0.050 0.060
Light brush and trees in summer 0.040 0.060 0.080
Medium to dense brush in winter 0.045 0.070 0.110
Medium to dense brush in summer 0.070 0.100 0.160
Trees
Dense willows, summer, straight 0.110 0.050 0.200
Cleared land with tree stumps, no sprouts 0.030 0.040 0.050
Same as above, but with heavy growth of sprouts 0.050 0.060 0.080
Heavy stand of timber, a few down trees,
little undergrowth, flood stage below branches 0.080 0.100 0.120
Same as above, but with flood stage reaching branches 0.100 0.120 0.160
MAJOR STREAMS (TOP WIDTH AT FLOOD STAGE> 100 FT). THE n VALUE IS LESS
THAN THAT FOR MINOR STREAMS OF SIMİL4R DESCRIPTION BECAUSE BANKS
OFFER LESS EFFECTIVE RESISTANCE.
Regular section with no boulders or brush 0.025 --- 0.060
Irregular and rough section 0.035 --- 0.100

C = C′ g (70)

where C´ is dimensionless.

104
 Flow Resistance in Streams

Combining Eqs. 25, 32 and 50 yields,

C R ⎛ R ⎞
= 6.25 + 5.75 log + 5.75 log⎜⎜12.21 ⎟⎟ (71)
g ks ⎝ ks ⎠

Substituting g = 9.81m/s2 into 71 yields

⎛ R ⎞
C = 18 log⎜⎜12.21 ⎟⎟ (72a)
⎝ ks ⎠

or

⎡ RU ∗ ν ⎤
C = 18 log ⎢12.21 (72b)
⎣ ν k sU ∗ ⎥⎦

Einstein added a correction factor χ to Eq. 72 and modified it

⎛ RU ∗ ⎞ kU
C = 18 log⎜12.27 ⎟ − 18 log s ∗ + 18 log χ (73)
⎝ ν ⎠ ν

where χ = f(ksU*/ν) . Variation of Chezy coefficient is shown in Fig. 2.

Figure 2. Variation of Chezy coefficient as a function of RU*/ν or ksU*/ν k(ks/δ)

105
Sediment Transport Technology

Evaluation of Fig. 2 reveals that:

1. For values of RU*/ν > 70, the radius of curvature of the curves increases very
rapidly and for RU*/ν > 500 it is practically equal to infinity.
2. For values of RU*/ν < 3.63, the bed is considered hydraulically smooth; the
representative curves terminate when ks/δ = 0.3129.
3. The representative curves cross the axis of abscissa at different points according to
different values of RU*/ν. However, C = 0 is a condition that can not be realized.
4. It can be shown that for a surface to be hydraulically smooth it is sufficient that

U ∗ks 5ν
≤ 5 or k s ≤ (74)
ν U∗

This condition leads to

5Cν
ks =
gU

Schlichting (1955) gives

ν
k s = 100 (75)
U

for the average condition of a hydraulically smooth surface. Using this,

U ν 1
= 100 (76a)
U∗ U ∗ ks

or

C 1
= 100 (76b)
g (k sU ∗ /ν )

Substituting the value of RU*/ν from Eq. 64 into Eq. 66b yields C=113.5 in English unit
and C=62.64 in metric units. This value of C corresponds to RU*/ν ≅1000.

Example 1.

Show on the f vs. Re diagram, given in Fig. 3, the variation of Chezy coefficient (Eq. 73)
and discuss the results.

Solution
Combined with Eqs. 32 and 24, Eq. 73 can be written as,

8g ⎛ RU U ∗ ⎞ kU ⎛k ⎞
= 18 log⎜12.27 ⎟ − 18 log s ∗ + 18 log f ⎜ s ⎟ (77)
f ⎝ ν U ⎠ ν ⎝δ ⎠

106
 Flow Resistance in Streams

and

1 RU ks ⎛k ⎞
− 1.015 log f = −0.87 + 2.03 log − 2.03 log + 2.03 log f ⎜ s ⎟ (78)
f ν δ ⎝δ ⎠

Two critical values of ks/δ are discussed:

1. ks/δ=1/3 represents the boundary of hydraulically smooth bed
2. ks/δ = 6 represents a boundary with completely developed roughness.

If these values are introduced to Eq. 78,

1 RU ks 1
− 1.015 log f = 2.03 log + 0.183 for = (79)
f ν δ 3

and

1 RU ks
− 1.015 log f = 2.03 log − 2.37 for =6 (80)
f ν δ

The variations of f vs. Re using Eqs. 79 and 80 are also shown in Fig. 3. The
representative lines define quite well the limits of the hydraulically smooth bed and
completely rough bed.

4.4. The Strickler, Manning-Strickler, and Lane formulas

Strickler (1923) defined Manning’s n as a function of particle size,

D1 / 6
n= (81)
21.1

where D is the diameter of uniform sand (in mm) used in performing the study. The
experiments were conducted in a small flume, on the bottom and wall of which the
experimental sand was pasted. Therefore, this experimental equation is not valid for
movable bed.

Meyer-Peter and Müller (1948), transformed Strickler’s formula by using sand mixture,
1/ 6
D90
n= (81a)
26

In Eq. 81a, D is in meter. This equation can be used to approximate Manning’s n if the
bed is not paved with cobbles.

Different forms of Strickler’s formulas,

1/ 6
D50
n= = 0.032 D50
1/ 6
(D50 in ft) (81b)
31.3

107
Sediment Transport Technology

Figure 3. Variation of coefficient f for ks/δ=1/3 and ks/δ=6 as a function of Re

1/ 6
D65
n= = 0.0132 D65
1/ 6
(D65 in mm) (81c)
75.75

Keulegan (1938)

1/ 6
D50
n= (D50 in ft) (81d)
46.9

in (1949) he suggested

1/ 6
D90
n= (D90 in ft) (81e)
49

108
 Flow Resistance in Streams

and

1/ 6
D65
n= (D65 in ft) (81f)
29.3

Table 5 compares the results obtained by application of Eq. 81a with corresponding
measured values. From the result of the field study in San Luis Valley, Lane and
Carlson (1953) suggested that

Table 5. Comparison of computed values of n determined from Eq. 81a with

corresponding measured values

Name of River D90 mm** n Computed Measured

Clark Fork at St. Regis, Montana 0.030 0.028

Clark Fork above Missoula, Montana 0.032 0.030
Blackfoot River near Ovando, Montana 280 0.031 0.031
Coeur d’Alene River near Prichard, Idaho 650 0.036 0.032
West Fork Bitterroot River near Conner, 265 0.031 0.036
Montana
Wenatchee River at Plaine, Washington 320 0.032 0.037
Spokane River at Spokane, Washington 360 0.032 0.038
Middle Fork Fiathead River near Essex, 285 0.03 1 0.041
Montana
Grande Ronde River at la Grande, Oregon 157 0.028 0.043
Provo Rivernear Hailstone, Utah 310 0.032 0.045
0.073
South Fork Clearwater River near Grangeville, 440 0.034 0.051
Idaho
East Branch Au Sable River at Au Sable Forks, 1500 0.041 0.055
New York
Rock Creek Canal near Darby, Montana 375 0.033 0.060
Merced River at Happy Isles Bridge near 550 0.035 0.065
Yosemite, California
Boundary Creek near Porthill, Idaho 375 0.033 0.073
Rock Creck near Darby, Montana 415 0.033 0.075

* Geological Survey Water-Supply Paper 1849 “Roughness Characteristics of Natural

Channels.”
** D is assumed equal to D (the value used in the original paper) and the D value is
introduced in Eq. 71a; D90 is in mm.

109
Sediment Transport Technology

1/ 6
D75
n= (D75 in inch) (82)
39

In this study the beds of the canals were paved with cobbles naturally. Substituting the
value of n in Eq. 81a into Manning’s equation yields,

1/ 6
26 ⎛ R ⎞
U = 1 / 2 R1 / 2 ⎜⎜ ⎟⎟ S 1/ 2 g 1/ 2 (83)
g ⎝ D90 ⎠

or

1/ 6 1/ 6
U 26 ⎛ R ⎞ ⎛ R ⎞
= 1/ 2 ⎜⎜ ⎟⎟ = 8.30⎜⎜ ⎟⎟ (84)
U∗ g ⎝ D90 ⎠ ⎝ D90 ⎠

Eq. 84 is known as Manning-Strickler equation and its general form is,

C (1 / 2+ m ) 1 / 2
U= R S (85)
Dm

where D is representative diameter of the bed material.

4.5. Monomial-type formulas describing resistance to flow

The monomial–type formula has the form

1/ m
U ⎛ R ⎞
= k ⎜⎜ ⎟⎟ (86)
U∗ ⎝ D90 ⎠

The coefficients k and m may be determined for each particular case. For k = 8.30 and
m=6, Eq. 76 reduces to Manning-Strickler equation. Combining Eqs. 86 and 51 and
eliminating U/U* yields,

1/ m
⎛ R ⎞ R
k ⎜⎜ ⎟⎟ = 6.25 + 5.75 log
⎝ D90 ⎠ D90

For m = 6 the mean value of k derived from Fig. 4, then Eq. 86 becomes

1/ m
U ⎛R ⎞
= 8.12⎜⎜ ⎟⎟ (87)
U∗ ⎝ ks ⎠

Eq. 87 can be used to estimate the resistance to flow for natural channels. However, it is
difficult to estimate ks accurately. Fig. 4 shows the variation of Eq. 86 if it is expressed
as:

110
 Flow Resistance in Streams

Figure 4. Variation of resistance to flow expressed by the monomial-type equation

U 1 ⎛ R ⎞
log = log k + log⎜⎜ ⎟⎟ (88)
U∗ m ⎝ D90 ⎠

For k = 8.12 and m = 6, Eq. 88 is plotted on Fig. 4 yielding straight line 1-1.

If channel is wide and bottom is flat with uniform roughness, flow characteristics can be
expresses by Eq. 48. This case can be seen on Fig. 4 by curve 3-3.

There is a difference between the straight line 1-1 and the curve 3-3 due to the variation
in the value of the parameter m. In addition, the straight line 2-2 represents Eq. 89.

1/ 7
U ⎛R ⎞
= 9.0⎜⎜ ⎟⎟ (89)
U∗ ⎝ ks ⎠

There is an apparent overlapping of these three representative relations when

R
15 < < 500 (90)
ks

Hence, it is possible to use Eq. 89 or 87 to estimate the resistance to flow in natural

channels. An experimental method can be derived based on these relationships. If k and
m in Eq. 86 are determined directly in situ, a monomial-type formula for particular river

111
Sediment Transport Technology

can be obtained to define the variation of resistance. This type of relationship is very
useful in physical and mathematical model studies. Using field measurements and
observations, model studies can be conducted easily and accurately if the resistance to
flow can be approximated by a monomial-type formula.

Example 2.
Extensive measurements were made on the Göksu River, Turkey, to define the variation
of resistance to flow. Using these data, determine the monomial-type formula
representing this variation and compare the results obtained with results when using
different resistance formulas.

Data: The river data obtained by direct measurements are given in Tables 6 and 7.

Solution:
Values of U/U* and R/D90 are computed and listed in Table 8. Values of U/U* and R/D90
in Table 8 are plotted on Fig. 5. The straight line representing the variation of the
resistance of the Göksu River in this reach is then the straight line A-A shown in Fig. 5.
The monomial-type formula corresponding to this line is

1 / 5.64
U ⎛ R ⎞
= 4.6⎜⎜ ⎟⎟ (91)
U∗ ⎝ D90 ⎠

Table 6. Flow characteristics of the Göksu River, Turkey

U (m/sec) R (m) S Q (m3/sec)

0.37 0.94 0.0001117 24
0.45 1.08 0.0001125 42
0.70 1.37 0.0001800 83
1.00 1.64 0.0003000 148
1.44 2.09 0.0004280 270

Table7. Sediment characteristics of the Göksu River, Turkey

D35 = 0.0015 m D65 = 0.0027 m D90 = 0.003 m γ s' =1.7 t/m3

Table 8.Values of U/U* and R/D90 for the Göksu River, Turkey

U (m/sec) R (m) S U* (m/sec) U/U* R/D90

0.37 0.94 0.000112 0.03214 11.51. 313
0.45 1.08 0.000113 0.03460 13.01 360
0.70 1.37 0.000180 0.04918 14.23 457
1.00 1.64 0.000300 0.06947 14.39 547
1.44 2.09 0.000428 0.09368 15.37 697

112
 Flow Resistance in Streams

Figure 5. Variation of the resistance to flow for the Göksu River, Turkey

Variation of resistance characteristics computed by Manning and Chezy formulas are

tabulated in Table 9 below.

Considering Strickler’s formula

1/ 6
D90
n= = 0.015 (92)
26

Using the semi-logarithmic resistance formula

⎛ RU ∗ ⎞ D U
C = 18 log⎜12.21 ⎟ − 18 log 65 ∗ + 18 log χ (93)
⎝ ν ⎠ ν

Substituting the values of RU ∗ /ν , D65U ∗ /ν and χ , based on the data of Example 2

yields the results in Table 10. Table 11 summarizes the results obtained using the
different methods.

Table 9. Variation of coefficients n and C

U (m/sec) C n
0.37 36.66 0.027
0.45 42.20 0.024
0.70 44.66 0.0236
1.00 45.25 0.024
1.44 49.16 0.023

113
Sediment Transport Technology

Table 10. Variation of Chezy’s and Manning’s roughness coefficient computed from the semi-
logarithmic relation, ν=10-6 m2/sec
R U*
S RU*/ν D65U*/ν χ C n
(m) (m/sec)
0.94 0.000112 0.03214 30,212 87 1 65.3 0.01516
1.08 0.000113 0.03460 37,368 93 1 66.4 0.01525
1.37 0.000180 0.04918 67,377 133 1 68.3 0.01543
1.64 0.000300 0.06947 113,931 188 1 69.7 0.01558
2.09 0.000428 0.09368 195,791 253 1 71.6 0.01579

Table 11. Comparison of results obtained by application of different resistance formulas with
measured values U/U*
R Monomial Manning-Strickler Logarithmic
Measured Value
(m) Formula Formula Formula
0.94 12.74 (11)* 21.63 (88)* 20.85 (81)* 11.51
1.08 13.06 (0)* 22.14 (70)* 21.20 (63)* 13.01
1.37 13.62 (4)* 23.03 (62)* 21.18 (53)* 14.23
1.64 14.07 (2)* 23.73 (65)* 22.25 (55)* 14.39
2.09 14.68 (4)* 24.71 (61)* 22.78 (48)* 15.37
* Values between parenthesis show the discrepancies from the measured values in percent

The monomial-type formula approximates the variation of resistance to flow in Göksu

River accurately. Manning-Strickler and Semi-logarithmic formulas do not do not
approximate the variation of resistance to flow in the Göksu river accurately. The
difference between the computed values by Manning-Strickler and semi-logarithmic
formulas and measured values of n may be attributed to the existence of bed forms and
bars in the channel.

5. Resistance in Open Channels with Movable Boundaries

In general assuming a flat bed and only skin friction does prediction of resistance to
flow for rigid boundary. The bed forms are simulated by the selection of equivalent
sand roughness ks. The equivalent sand roughness is equal to the roughness of sand that
produces the resistance to the flow equal to the resistance resulting from the particles
and the bed forms.

Equations given in the previous section are developed for rigid boundary channels and
are valid for plane beds only. Thus, R´ replaces R and f´ replaces f. However, Lovera
and Kennedy (1969) stated that using friction-factor predictors developed one-phase
flows in rigid-boundary conduits to estimate f´ for alluvial-channel flows is tenuous
procedure. Because f = g(Re, ks) may not exhibit the same form for rigid and movable
boundary channels. Thus, they analyzed the available data on friction factors for flat-
bed flow in both natural streams and laboratory flumes and developed Fig. 6. This
figure can be used for predicting depth discharge relations in the flat-bed regime and
approximating surface drag friction factor f´ for flow in other regimes. In this figure r is
the total hydraulic radius.

114
 Flow Resistance in Streams

Figure 6. Friction-factors for flat-bed flows in alluvial channels, the number by each point is
r/d50x10-2 (after Lovera and Kennedy, 1969)

For alluvial boundary channels, the resistance due to form drag must be added to the
resistance due to skin friction. The evaluation of the resistance due to form drag is
complicated. It is possible that different resistance laws correspond to different bed
forms. Thus, the transition between particular bed forms may require special relations
and many attempts have been made by hydraulic engineers to solve this problem.

5.1. Einstein’s method for determining resistance to flow

Einstein (1950) presented two relations for solving resistance to flow with movable
boundaries and bed forms. The first relation is resistance due to grain roughness. The
second relation utilizes graphics to define the resistance to flow caused by bed forms.
The first expression is similar to Eq. 48 and is written

U ⎛ R' ⎞
'
= 5.75 log⎜⎜12.27 χ ⎟⎟ (94)
U∗ ⎝ ks ⎠

where χ is a correction factor that compensates for conditions where the channel bed is
not fully rough. Here χ = f(ks/δ) and δ is the thickness of the laminar sublayer and equal
to 11.6ν / U ∗' . Due to suspended sediment Eq. 94 may require modification. When U is
known then Eq. 94 yields the value of R´.

The second component of the resistance is due to the effect of the bed forms. This form
roughness or form drag is a function of flow, thus any change in form of the bed

115
Sediment Transport Technology

roughness will change bed roughness and resistance to flow. Einstein (1950) stated that
the form drag was related to the total sediment transport and suggested

U
= φ (ψ ′) (95)
U ∗′′

where ψ′ is the intensity of shear on the particle and expressed by,

γ ′ − γ D35 γ s′ D35
ψ′= = (96)
γ SR ′ γ SR ′

Also, ψ′ is the inverse of the Froude number related to the grain size. The
U / U ∗′′ = f (ψ ′) relationship, as developed from the field data is shown in Fig. 7.

Fig. 7 shows the resistance of different bed forms to flow as a single curve. Bishop,
Simons and Richardson (1965) showed that the relation is not valid for upper regime
flows in sand channels, and Şentürk (1975) pointed out that this curve is inadequate
when ψ′<10. However, Einstein and Barbarossa (1952) interpreted this figure
differently i.e. The factors governing the resistance are reduced to (γR ′S / γ s′ D35 ) and
R ′ / D65 . Here (γR ′S / γ s′ D35 ) is the Froude number and R ′ / D65 is the relative
roughness. Comparing Eq. 94 and Fig.7 with Eq. 6, it can be seen that parameters
defining resistance to flow are used by Einstein in their general form. Re replaced by Re
of the particle, Fr replaced by Fr expressed in terms of the friction velocity.

Figure 7. Friction loss due to channel irregularities, as a function of sediment transport (after
Einstein and Barbarossa, 1952)

116
 Flow Resistance in Streams

Inspection of this method shows the importance of the variation of U ∗′′ on the resistance
to flow. For example as discharge increases in the meandering channel the sediment
transport along the bed increases or sinuosity decreases because the meanders
progressively straighten. The value of U ∗′′ experiences a similar decrease.

Einstein method was checked by Harrison and Melema (1967) with data from Missouri
and Mississippi Rivers, found good agreement between computed and measured results.
However, Şentürk (1945) showed that application of this method to natural water course
with small slopes that transport fine sand large discrepancies occur between computed
and measured results.

The procedure suggested by Einstein and Barbarossa (1952) in Fig. 7 follows: After
surveying the selected reach, determine S, D65, D35 and the dimensions of the average
cross section of the channel. Then, using Einstein’s method can solve two kinds of
problem.
1. Given Q, determine R.
2. Given R, determine Q.

To solve the first problem, the following procedure is suggested.

Step 1: Assume a reasonable R′.
Step 2: Apply Eq. 81 and DuBoys model of bed load transport to determine U.
Step 3: Compute ψ′ and then read U / U ∗′′ from Fig. 7.
Step 4: Compute U ∗′′ and then R ′′ .
Step 5: Compute R = R ′ + R ′′ and determine the corresponding values of stage y and
cross sectional area A.
Step 6: Determine Q from Q = A·U. If this value of Q matches the given value of the
discharge, the problem is solved. If an appreciable difference exists between
computed and known value of Q, choose another value of R′ and repeat Steps 1
through 6 until agreement is achieved.

To solve the second problem:

Step 1: Assume a reasonable value of R′.
Step 2: Apply Eq. 81 and DuBoys model of bed load transport to determine U.
Step 3: Compute ψ′ and then read U / U ∗′′ from Fig. 7.
Step 4: Compute U ∗′′ and then R ′′ .
Step 5: Compute R = R ′ + R ′′ . If this value of R agrees with the given discharge, the
problem is solved. Otherwise repeat the procedure until a solution is found.
Then determine the corresponding values of stage y and cross sectional area A.
Step 6: Determine Q from Q = A·U.

5.1.1. Einstein’s method as interpreted by V. Vanoni

Using the Einstein procedure given above to solve the Resistance to flow is laborious.
Thus, Vanoni and Brooks (1957) simplified the solution by using the two parameters
( U 3 / gνS ) and ( U / gk s S ) (see Fig. 8a) and an extension of this analysis, Fig. 8b
developed by Simons. This figure can be used to determine U ∗′ without trial. The
Einstein method of computation is given in Examples 3 and 4.

117
Sediment Transport Technology

(a)

(b)

Figure 8. Curves for the graphical solution of the Einstein-Barbarossa equations for
determination of R′.

118
 Flow Resistance in Streams

Example 3.
Determine the characteristics of a stream using Einstein’s method and the following
data.

S = 0.0004, D35 = 0.0006 m, D65 = 0.001 m,

Q = 149.00 m3/sec ν = 0.106x10-5 m2/sec A = f(d) (see Fig. 9)
γ s′ = 1.68 t/m3 g = 9.81 m/sec2

Solution:
The solution follows the procedure given in previous section. The results are
summarized in Table 12.

Example 4.
Solve for the discharge Q using the data of Example 3 assuming R=1.374m.

Solution:
The problem can be solved in the same manner as Example 3. However, instead of
comparing the values of Q in column 12 of Table 12 with known Q, the values of R
listed in column 10 must be compared with the known R. Then Q can be computed
using the procedure summarized in columns 11 and 12.

Figure 9. A = f(R) curve in Example 3 and Example 5

119
Sediment Transport Technology

Table 12. Solution of Example 3

S = 0.0004 D35 = 0.0006m D65 = 0.001m ν = 1.06x10-6 m2/sec Q = 149 m3/sec

1 2 3 4 5 6 7 8 9 10 11 12
R′ U ∗′ U U ∗′′ R″ R A Q
D65/δ χ ψ′ U / U ∗′′
m m/s m/s m/s m m m2 m3/s
0.10 0.0198 1.61 1.51 0.372 25.20 6.0 0.062 0.98 1.08 135 50.22
0.15 0.0242 1.97 1.38 0.474 16.80 7.0 0.067 1.14 1.29 235 111.39
0.17 0.0254 2.07 1.37 0.503 15.27 7.3 0.069 1.21 1.37 295 148.39

Column 1: Choose a value of R′. Column 7: U / U ∗′′ = f (ψ ′) from Fig. 7.

Column 2: U ∗′ = gSR′ Column 8: U ∗′′ = U / (U / U ∗′′ )
Column 3: D65 / δ = (1 / 11.6)(U ∗′ D65 /ν ) (
Column 9: R′′ = U ∗′′2 / gS )
Column 4: χ = f ( D65 / δ ) from DuBoys Column 10: R = R′ + R′′
Column 5: U = U ∗′ [5.75 log12.27(R′ / D65 )χ ] Column 11: Determine A from Fig. 9.
Column 6: ψ ′ = 1.68(D35 / R′S ) Column 12: Using Q = AU determined Q equals the given
Q; if not choose a new R′.

Example 5.
The mean velocity of a stream is measured and U = 1.72 m/sec. Determine the discharge
of this stream for D35 = 1cm, D65 = 1.5cm, S = 0.001, A = f(R), as given in Fig. 9, and
ν=9.8x10-6 m2/sec.

Solution:
Step 1: Compute the two parameters

U3 1.72 3
= = 5.3 x10 7
gνS 9.81 ⋅ 9.8 x10 ⋅ 0.001
−6

U 1.72
= = 141.79 ≈ 142
gD65 S 9.81 ⋅ 0.015 ⋅ 0.001

From Fig. 8, U U ∗′ = 17, U ∗′ = 1.72/17 = 0.101, and R ′ = 1.04m.

Step 2: Compute ψ′

0.01
ψ ′ = 1.68 = 16.15
1.04 ⋅ 0.001

Step 3: Determine, using Fig. 7, U / U ∗′′ , U ∗′′ and R ′′ .

U
=7
U ∗′′
1.72
U ∗′′ = = 0.246m / sec
7
R ′′ = 6.17 m

120
 Flow Resistance in Streams

Step 4: Compute R

R = R ′ + R ′′ = 7.21 m

Step 5: Using Fig. 9, A = 400 m2

Step 6: Compute the discharge

Q = A ⋅ U = 400 ⋅ 1.72 = 688 m 3 /sec

5.2. Shen’s Method for determining resistance to flow

Resistance defined by Einstein is for plain bed. The curve of Fig. 7 is based on data for
flow with different bed configurations. ψ′ does not sufficiently characterize and predict
bed forms. For example, The fall velocity of the bed material significantly influences
the bed configurations. Shen (1962) defined a parameter λ that varies with the particle
Reynolds number wD/ν (Fig. 10).

For fine material wD / ν < 100 the variation of U ∗′′ / U is given by

U ∗′′ ψ′
= 0.03 + 0.11 log (97)
U λ

For coarse material with wD / ν > 100 . Eq. (97) takes the form

U ∗′′ ψ′
= 0.064 − 0.0909 log (98)
U 7.12

Figure 10. Variation of λ as a function of wD/ν

121
Sediment Transport Technology

for ψ′>10, where

γR ′S
ψ′ =
γ s′ D50

For ψ′<10, the suggested relationship is

U ∗′′ ψ′
= 0.03 − 0.11log (99)
U 7.12

Fig. 11 shows the variation of the resistance given by Eqs. 97 and 98.

5.3. Engelund and Hansen’s method

Engelund (1966) developed a method based on the friction loss due to the bed forms
using expansion-loss equation.

⎡ ⎤
∆H ′′ q2 ⎢ 1 1 ⎥ U 2 ⎛ Am ⎞ 2
S ′′ = = ⎢ − ⎥≈ ⎜ ⎟ (100)
L 2 gL ⎢ 1 1 ⎥ 2 gL ⎝ d ⎠
d − Am d + Am
⎣⎢ 2 2 ⎦⎥

where ∆H″ is the friction due to bed forms with a wave length L, q is the unit discharge,
d is the mean depth, and Am is the amplitude of the sand waves. Substituting the values
of τ 0′ and τ 0′′ from Eqs. 12 and 13 into Eq. 14 yields:

Figure 11. Variation of U*/U as a function of ψ′⁄λ (after H. W. Shen)

122
 Flow Resistance in Streams

τ 0 = γS (R ′ + R ′′) (101)

Adopting the idea of Meyer-Peter,

τ 0 = γR(S ′ + S ′′) (102)

and Eq. 89 yields,

τ 0 τ 0′
= + S ′′ (103)
γR γR

Substituting Eq. 100 for S″ into Eq. 103, and assuming R ≈ d for a wide channel

2
τ 0 τ 0′ U 2 ⎛ Am ⎞
= + ⎜ ⎟ (104)
γd γd 2 gL ⎝ d ⎠

assume

dS
θ= (105)
[(ρ s ρ ) − 1]D
d ′S 1 Am2
θ′ = and θ ′′ = Fr2 (106)
[(ρ s ρ ) − 1]D 2 [(ρ s ρ ) − 1]DL
then Eq. 104 becomes

θ = θ ′ + θ ′′ (107)

This relationship proposed by Engelund and Hansen (1967). Fig. 12 shows a diagram

Figure 12. Flow resistance (after Engelund et al., 1967)

123
Sediment Transport Technology

obtained by plotting the flume data gathered by Simons and Richardson (1966). In Fig.
12, in the upper regime, the curve in which there are antidunes, standing waves, flat bed
no form drag is associated with the flow, thus, θ = θ′ for these conditions.

Fig. 12 can be applied to determine a stage-discharge relationship by using the

following procedure:
1. Find S, d and geometric properties of the cross section from field surveys,
2. Compute θ from Eq. 105,
3. Find θ′ from Fig. 12 for the value of θ,
4. Compute d′ from Eq. 106
5. Compute U from Eq. 94, and
6. Find the cross-sectional area from the geometric properties of the X-section and
determine Q = A·U.

5.4. Alan and Kennedy’s method

Alan and Kennedy (1969) investigated the flume data by Vanoni and Brooks (1957)
Brooks (1958), Laursen (1958) and Guy, Simons and Richardson (1966). They also
investigated field data from the Missouri and Mississippi Rivers (Einstein and
Barbarossa, 1952), the Elkorn River (Beckman and Furness, 1962) and West Pakistan
Canals and the Rio Grande (Nordin, 1964) and suggested the functional relationship.

⎛ U D ⎞
f ′′ = φ ⎜ , 50 ⎟ (108)
⎜ gD R ⎟⎠
⎝ 50

and plotted the data to obtain Fig. 13.

Figure 13. Graphical predictor of the bed form friction factor (after Alan and Kennedy, 1969)

124
 Flow Resistance in Streams

By applying this figure and Fig. 6 for f, which is equivalent to f′, they calculated depth-
discharge relations for the Niobrara and the Colorado River. The computed results fit
the observed values better than Einstein and Barbarossa’s (1952) method.

5.5. Bajorunas’ method

Bajorunas (1952) investigated the effect of bed forms on the value of Manning’s n. He
assumed that n = n ′ + n ′′ . The functional relation he developed Eq. 109 which is similar
to Einstein’s (1950) Eq. 95.

n ′′ = θ (ψ ′) (109)

Based on actual river data, he developed Fig. 14. This figure and Fig. 7 have the same
accuracy for predicting the resistance.

5.6. Simons and Richardson’s method

The relations in the previous methods that describe resistance to flow consider the
resistance as a whole. The nature of bed configurations is not taken into account. The
resistance formulas suggested by Simons and Richardson (1966) and Şentürk (1975)
can be classified in a different category. They considered the resistance of bed forms
separately and suggested particular formulas for each bed form.

Richardson and Simons (1967) suggested the following resistance equations:

1. For a plane bed without sediment transport
C d
= 5.9 log + 5.44 (110)
g D85

2. For a plane bed with sediment transport

C d
= 7.4 log (111)
g D85

Figure14. Flow resistance due to bed forms (after Bajorunas, 1952)

125
Sediment Transport Technology

3. For ripples
C ⎛ 0.3 ⎞ 0.13
= ⎜⎜ 7.66 − ⎟⎟ log d + + 11 (112)
g ⎝ U∗ ⎠ U∗

4. For dunes and antidunes

C d ∆RS E
= 7.4 log 1− (113)
g D85 RS E

where D85, RSE, ∆RSE and d are in feet. The term ∆RSE is an adjustment for RSE to
compensate for the form roughness. For dunes, ∆RS E = φ (RS E ) . As shown in Fig. 15,
there is one relation for D50<0.5 mm and another for the 0.93 mm sand.

The ∆RSE vs. RSE relation is valid for streams as large as the Mississippi River. For
antidunes, the correction term 1 − (∆RS E / RS E ) is a function of sand size, τ, and d
(Fig. 16).

In determining the average velocity in a sand-bed channel with known d, SE, D50, and
D85, the following procedure is suggested by Richardson and Simons (1967):
1. Assume a bed configuration.
2. Compute C g from appropriate equation (Eqs. 110-113).
3. Compute U from Eq. 27.

Figure 15. Relation between ∆RS and RS for dune-bed configuration.

The symbols are: ∆: 0.19mm, ○ :0.27 and 0.28mm, σ: 0.45 mm,
□: 0.93 mm, ◓: Elkhorn River near Waterloo, Nebraska,
●: Rio Grande near Bernardo, New Mexico,■: Mississippi River
near St. Louis, Missouri (after Richardson and Simons, 1967)

126
 Flow Resistance in Streams

Figure 16. Relation between Chezy coefficient correction term,

shear stress, and depth for antidune-bed configuration.
∆: 0.19 mm, O : 0.27 and 0.28 mm, σ :0.45mm (after
Richardson and Simons, 1967)

4. Compute τ 0U ≈ gdSU .
5. Find the bed form from Fig. 17 for values of τ0U and D50 and check with the
assumed one.
6. If the bed forms are different, use the bed form found and repeat Steps 2-5.

5.7. Sentürk’s method

Sentürk (1965-1967) stated two hypotheses namely:
1. The factor responsible for the formation of bed form is the inertial force exerted by
the flow and the weight of each sediment particle,
2. For each type of bed form, corresponds to a particular trend in the variation of
resistance to flow.

The first hypothesis led to utilization of equilibrium equation given below,

τc
= k tan θ (114)
γ s′ Ds

Sentürk (1967) has shown that k is proportional to (w65 D65 ν ) , the Reynolds number
2

related to fall velocity where w65 is the fall velocity of the D65 particle, and ν is the
kinematic viscosity. Sentürk (1967) developed a parameter f of the form,

τ γ s′ D65
f = (115)
(w65 D65 ν )2

127
Sediment Transport Technology

Figure 17. Relation of bed form to stream power and median fall diameter of bed sediment
(after Simons and Richardson 1966)

and used this parameter to predict bed forms and their resistance to flow. The parameter
f can be written as

f = f ′ + f ′′ (116)

where

τ ′ γ s′ D65
f′= (117)
(w65 D65 ν )2
and

128
 Flow Resistance in Streams

τ ′′ γ s′ D65
f ′′ = (118)
(w65 D65 ν )2
The values of τ′ and τ″ are

τ ′ = γR ′S (119)
τ ′′ = γR ′′S (120)

Eqs. 117 and 118 yield

R ′ 1 γ s′ D35 (w65 D65 ν )

2

= =C (121)
D65 f ′ γ D65 S
R ′′ 1 γ s′ D35 (w65 D65 ν )
2

= =C (122)
D65 f ′′ γ D65 S

If f′ and f″ are governing parameters in the formation of bed configurations, then C can
replace both of them. Sentürk (1969) has also shown that this parameter is a major
factor in the build-up of the resistance to flow on a movable bed and he developed the
formulas predicting not only the bed resistance but also the formation of bed
configurations (Fig. 18).

1. For ripples (in metric unit):

Skin friction
U R′
= 6 + 6.50 log (123)
U ∗′ D65

Drag
U R ′′
= −52 + 2.5 log C + 18 log (123a)
U ∗′′ D65

where

U ∗′ = ( gR ′S )
1/ 2
(124)

U ∗′′ = ( gR ′′S )
1/ 2
(125)
R = R ′ + R ′′ (126)
γ ′ D (w D ν )2
C = s 35 65 65 (127)
γ D65 S

2. For dunes (in metric unit)

Skin friction
U R′
= 6 + 6.5 log (128)
U ∗′ D65

129
Sediment Transport Technology

Figure 18. Variation of the resistance to flow for a given bed material and constant energy
slope (after Sentürk). R′ and R″ are shown separately on the abscissa and U is
reported on the axis of ordinates. The black lines show the variation of the
resistance for S constant and bed material constant

Drag
U R ′′
= 48 − log C + 11 log (123b)
U ∗′′ D65
D γ ′ (w D ν )
2

C = 35 s 65 65 (124)
D65 γ S

3. For plane bed with appreciable sediment transport (in metric unit). If X is a factor
affecting the resistance of the plane bed with appreciable sediment transport to the
flow

X = 2 log C (125)

then

130
 Flow Resistance in Streams

U R′
= 6 − X + 6.5 log (126)
U ∗′ D65

and

U R′
= 6 − 2 log C + 6.5 log (127a)
U ∗′ D65

For example, if the bottom of a water course is covered with ripples; its resistance to
flow can be obtained from Eqs. 123 and 123a. Obtain R ′ from Eq. 110 and R ′′ from
Eq. 110a. Then, R = R ′ + R ′′ .

In the case of large rivers U / U ∗′′ is, in general, greater than the value provided by Eq.
110b. The corresponding bed forms are very large sand bars. The representative point in
Fig. 18 lays on the RA curve but higher than point A. Eqs. 123 and 123a represent
resistance to flow. In such a case, the curve XB′ intersects the curve RA (Sentürk,
1987). He used the following assumptions to obtain the resistance to flow in the
transition regions:
1. The change of the bed pattern is progressive and occurs when prevailing flow
conditions dictate this transformation.
2. The transition is accompanied by a rapid change in resistance to flow.
3. The transition occurs when the resistance to the flow of a different bed pattern is
smaller than for the prevailing flow conditions.
4. The resistance to flow due to drag is always greater than the resistance due to skin
friction.

These assumptions lead to the determination of a group of formulas defining the

resistance to flow in the transition region that also indicates the bed forms expected
(Sentürk 1987).

5.7.1. Resistance to flow in the upper transition region

The transition from dunes to plane bed is called the upper transition. According to
assumptions (1) and (2), the beginning of the transition is located at point B, in Fig. 18.
At this point U is max. accompanying dune pattern. This value of U is derived from
Eqs. 123 and 123b by differentiation

R B′′ = exp10 (3.3936 − 0.091log C − log D65 ) (128)

Substituting this value of RB′′ in Eq. 110b yields

U B = 9.57( gSRB′′ )
1/ 2
(129)

According to assumption (3), the resistance to flow in the transition region may be
defined by a straight line passing through B (Fig. 18) and perpendicular to the line
representing the resistance due to skin friction.

131
Sediment Transport Technology

The equation of straight line predicting the resistance in the upper transition region is

R ′ (R ′ − RB′′ )
2a(d + 6.5 log R ′) R ′ = U B − (130)
a(e + 6.5 log R ′)

where

gS
a=
2
1
d = 6 + 6.5 log (131)
D65
1
e = 11.65 + 6.5 log (132)
D65

Eq. 130 gives R′ corresponding to the intersection of the straight line with the curve
representing the resistance of skin friction.

The velocity corresponding to R′ is called UPL and is defined as

⎛ R′ ⎞
U PL = ⎜⎜ 6 + 6.5 log ⎟ gR ′S (133)
⎝ D65 ⎟⎠

The representative resistance equation is then

R ′ (R ′ − Rt )
U PL = U − (134)
a(e + 6.5 log R ′)

where Rt is the hydraulic radius in the transition region.

5.7.2. Resistance – transition from ripples to dunes

Resistance to flow in the transition region is determined according to assumptions (3)
and (4). This resistance is approximated by the straight line NM on Fig. 18. Point X is at
the intersection of the dune and skin friction curves. The coordinates of this point are:

R x = exp10 (2.4 − 0.057 log C + log D65 ) (135)

and

U x = (21.6 − 0.373 log C ) gSR x (136)

Points N and M are determined as follows:

• The Eqs of a straight line normal to dune curve and passing through X′ (Fig. 18) is
R ′′ (R ′′ − RR′′ )
2a(c ′ + 11 log R ′′) R ′′ = U x − (137)
a(b + 11log R ′′)

132
 Flow Resistance in Streams

where

gS
a=
2
1
b = 48 − log C − 11 log − 9.57
D65
c ′ = b + 9.57

• The RR′′ value is obtained from ripple equation assuming U = Ux

• Substituting R″ from Eq. 137 in the dune equation, the value of UM is:
⎛ R ′′ ⎞
U M = ⎜⎜ 48 − log C − 11 log ⎟ gSR ′′
⎝ D65 ⎟⎠

• To compute UN, the following equation of the straight line passing through point X
and normal to the ripple curve may be solved for R″
R x (R x − R ′′) ⎛ R ′′ ⎞
Ux + = ⎜⎜ − 52 + 2.5 log C + 18 log ⎟⎟( gSR ′′)1 / 2 (138)
a(b − 11 log R x ) ⎝ D65 ⎠

then

⎛ R ′′ ⎞
U N = ⎜⎜ − 52 + 2.5 log C + 18 log ⎟⎟( gSR ′′)1 / 2 (139)
⎝ D 65 ⎠

These equations determine the coordinates of points N and M. The resistance in the
medium transition region is then approximated by,

U N −U M
U MT = (RMT
′′ − RM ) + U M (140)
R N − RM

where UN is the velocity corresponding to point N, RN is the hydraulic radius of the

ripple bed related to UN, UM is the velocity corresponding to point M, RM is the
hydraulic radius of the ripple bed related to UM, UMT is the velocity in the medium
transition region RMT is the hydraulic radius corresponding to UMT, if UMT is given, Eq.
140 can be used to compute RMT or vice verse.

5.7.3. Resistance – transition from plane bed to ripple bed

Observations show that the transition from plane bed w/o sediment movement to ripple
bed occurs rapidly. If Q = constant during this phenomena, d increases and U decreases.

Fig. 18 shows that for S = constant and for a given bed material, the curve T, X, B′
represents the resistance due to skin friction. For U<UBM, UBM is the velocity
corresponding to the beginning of motion, skin friction is the responsible factor for the
resistance. When U = UBM, ripples are formed, and hydraulic radius changes from RBM
to RR. RR represents the hydraulic radius related to ripples with RR > RBM . But U
approaches ULT < UBM.

133
Sediment Transport Technology

• Even though ULT < UBM bed material continues to move because RLT > RBM.
• Thus an intermediary bed pattern establishes.
• Bed forms are similar to ripples but not well formed.
• If Q decreases, bed pattern continues to change, but the plane bed is only reached at
the end of relatively long time period.
• If SE = constant the beginning of motion is mostly function of hydraulic radius.
• Fig. 19 gives the critical value RBM corresponding to threshold conditions as a
function of wD/ν .

Thus, for known w, D, and ν, U*D/ν can be obtained from Fig. 19, D/ν = constant, the
only variable is U* or D. For R > RBM and SE, the solid particles of diameter D are in
motion on the bed. Absolute tranquility is reached only where R=RBM in the lower
transition region. The relation for R is R = R ′ + R ′′ where R′ is the hydraulic radius
related to the surface drag, and R″ is the hydraulic radius related to the form drag. R″min
=R′, thus, just before the absolute tranquility the hydraulic radius of the flow is
R = R′ + R′ .

If this value of R = RBM, the threshold condition is reached, then

RBM
RBM = 2 R ′ and R ′ =
2

For R < R′, the solid particles are static and the boundaries can be considered as rigid. If
Q increases, R increases accordingly, but the particle motion start only when R
approaches its limiting value RBM. The variation in the lower transition region is
approximated by the straight line TR on Fig. 18. A flow with a mean velocity ULT
satisfies the relation U T < U LT < U BM .

Figure 19. Diagram for the computation of critical shear

134
 Flow Resistance in Streams

This relation identifies that the flow takes place in the lower transition region (Fig. 18).
The corresponding R can be obtained from

U LT − U T
RLT = (RBM − RT ) + RT (141)
U BM − U T

where RLT is the hydraulic radius of flow with velocity ULT in the lower transition
region, ULT is the velocity of flow in the lower transition region, UT is the velocity of
flow corresponding to absolute tranquility of solid particles, RT is the hydraulic radius of
flow corresponding to UT, RBM is the hydraulic radius of flow corresponding to the
beginning of motion in case of plane bed, and UBM is the velocity of flow corresponding
to the beginning of motion in case of plane bed.

Example 6:
Determine the hydraulic radius and depth of low which takes place in a rectangular
flume 2.4 m wide using Sentürk’s method when U = 1.012 m/sec.

Data: D35 = 0.37 mm, D50 = 0.45 mm, D65 = 0.53 mm

D85 = 0.70 mm, S = 0.00301, ν = 1.03x10-6m2/sec
w50 = 0.060 m/sec, g = 9.805 m/sec2,
γ s′ =1.68 t/m3, and w65 = 0.076 m/sec

Solution:
Step 1: Determine UBM, UB, Ux, UN, UM, UPL as follows:

(1) Use Fig. 19 to determine UBM

wD50 0.06 x0.00045

= = 26.2
ν 1.03 x10 −6

The corresponding value of U*D50/ν is 7.0 (Fig. 19). Then

7.0 x1.03x10 −6
U∗ = = 0.016 m/s
0.00045
U2
R = ∗ = 0.009 m
gS

Substituting these values into Eq. 129 (Example 3)

⎛ R ⎞
U BM = U ∗ ⎜⎜ 6.25 + 5.75 log χ ⎟ = 0.23 m/sec
⎝ D65 ⎟⎠

(2) Use Eq. 128 to determine RB′′ and Eq. 129 to determine UB

135
Sediment Transport Technology

D (w D /ν ) = 595899.6
C = 35 γ s′ 65 65
2

D65 S
RB′′ = exp(3.4936 − 0.091 log 595899.6 + log D65 ) = 0.4924 m (142)
U B = 9.57( gSRB′′ )
1/ 2
= 1.15 m/sec

(3) Use Eqs. 135 and 136 to determine Ux

U x = (21.6 − 0.373 log 595899.6) gSRx = 0.83 m/sec (143)

where

R x = exp10 (2.4 − 0.057 log 595899.6 + log D65 ) = 0.06237 m (144)

(4) Use Eqs. 137, 138 and 139 to determine UM and UN

UM = 0.89 m/sec
UN = 0.79 m/sec

(5) Use Eqs. 130–133 to determine UPL

2a = gS = 9.805 x0.00301 = 0.17179

1
d = 6 + 6.5 log = 27.137 (145)
0.00053
1
e = 11.65 + log = 32.9422
0.00053

The hydraulic radius R′ related to UPL can be obtained from Eq. 130.

R ′ ( R ′ − 0.49)
0.17179(27.137 + 6.5 log R ′) R ′ = 1.15 −
0.085895(32.9422 + 6.5 log R ′)
R ′ ≅ 0.11 m

Substitute R′ in Eq. 133 to compute UPL

⎛ 0.11 ⎞
U PL = ⎜⎜ 6 + 6.5 log ⎟ 0.11gS
⎝ D65 ⎟⎠
UPL= 1.20 m/sec

Step 2: Predict the most likely bed forms

U M = 0.89 < U = 1.012 < U B = 1.15 m/sec

Hence, the bed is covered with dunes.

136
 Flow Resistance in Streams

Example 7.
Determine the hydraulic radius and depth of a flow in a rectangular flume 2.4 m wide
using Simons and Richardson’s method.

Data: U = 0774 m/sec = 2.54 ft/sec, S = 0.00275

D85 = 1.5 mm, and g = 32.2 ft/sec2

Solution:
Step 1: Assume a dune bed.
Step 2: Assume d = 1.11 ft then R = 0.869 ft.
Step 3: Solve Eq. 100 using Fig. 15 for the determination of ∆RS.

RS = 2.38975x10-3
∆RS = 1.6815x10-3
U 0.3048 1.6815
= 7.4 log × 1.11 1 − = 9.58
U∗ 0.0015 2.3898
U = 2.66 ft/sec
U computed − U mean 2.66 − 2.54
= = 5.0%
U computed 2.66

The error is 5%. If more accuracy is required, one may choose a depth slightly
greater, say 1.11 ft, and repeat the calculations.

Step 4: Compute τ·U

τ ⋅ U = γRSU = 62.4 × 0.869 × 0.00275 × 2.63 = 0.392 ft.lb/sec.ft 2

Step 5: The bed form, from Fig. 18, corresponding to the values of τ·U and D50 is dunes.
Hence, the assumption in Step 1 is true and the d = 1.11 ft or 0.34 m.

5.8. Comparison of different methods for estimating resistance to flow

Seven methods are presented for determining resistance to flow of movable bed. Four
methods were applied to flows in natural and artificial channels and their results were
compared.

Those four methods are

1. Einstein’s method
2. Shen’s method
3. Sentürk’s method
4. Simons’ method

Data used for comparison were obtained from:

1. Colorado State University (CSU)
2. Missouri River
3. USGS, and

137
Sediment Transport Technology

4. Pakistan

Comparison of the results was presented in Table 13.

Applications of these formulas to natural and artificial flows yield deviations from the
measured resistance to flow. Normal Distribution Law was used to compare the
deviations. Table 14 presents the results of the analysis. The accuracy of the results, by
using these formulas for estimating the resistance to flow for existing various bed
configurations, were checked with,

Table 13. The range of application of resistance formulas

Plane bed Plane bed
without with Upper
Method Ripples Dunes Transition
sediment sediment regime
motion motion
1. Einstein formula yes yes yes no yes yes
2. W.Shen’s formula yes yes yes no yes no
3. Sentürk’s formula no yes yes yes yes yes
4. Simons and
Richardson’s yes yes yes no yes yes
formula

Table 14. Comparison of flow formulas applied to flow in artificial and natural water courses
% of Occurrences of Errors in Intervals

Statistical Standard
Variance
Formulas Data Mean Deviation 0-5 0-10 0-15 0-20 0-25 0-30 0-35 0-40 0-45 0-50
σ2
Error σ

CSU1
-0.137 0.0851 0.2917 12.18 24.12 35.50 46.12 55.82 64.46 71.95 78.34 83.60 87.85
Pakistan2
-0.034 0.0348 0.1866 2036 40.15 57.05 70.78 81.20 88.60 93.48 96.49 98.23 99.16
Einstein USGS3
0.229 0.1022 0.3196 9.64 19.15 28.47 37.41 45.95 53.90 61.26 67.91 73.87 79.04
Missouri
0.159 0.0102 9.1911 12.11 27.43 46.30 65.69 81.57 91.82 97.04 99.14 99.80 99.96
River4

CSU1
0.241 0.1935 0.4399 7.79 15.50 23.09 30.47 37.57 44.36 50.18 56.81 62.41 67.56
Pakistan2
0.191 0.0955 0.3090 10.64 21.07 31.23 40.84 49.86 58.15 65.62 72.24 77.96 82.85
Shen USGS3
0.013 0.0413 0.2033 19.37 37.60 53.81 67.32 77.98 85.87 91.38 95.02 97.26 98.58
Missouri
0.682 0.0532 0.2306 0.24 0.55 1.04 1.84 3.03 4.92 7.55 11.12 15.77 21.56
River4

CSU1
-0.074 0.0228 0.15 10 23.10 44.34 62.35 76.28 86.18 92.61 96.36 98.36 99.33 99.75
Pakistan2
0.027 0.0224 0.1496 25.75 48.88 67.57 81.12 89.96 95.14 97.86 99.15 99.69 99.90
Sentürk USGS3
0.029 0.0326 0.1806 21.52 41.52 58.74 72.52 82.78 89.89 94.42 97.11 98.60 99.37
Missouri
- 0.034 0.0087 0.0933 38.43 68.49 86.87 95.63 98.85 99.76 99.96 100.0 100.0 100.0
River4

CSU1
-0.023 0.0491 Ö.2215 17.74 34.61 49.89 63.06 73.78 82.14 88.36 92.74 95.64 97.50
Pakistan2
Simons and -0.025 0.0263 01620 23.93 45.76 63.93 77.72 87.26 93.27 96.72 98.53 99.39 99.77
USGS3
Richardson -0.107 0.0520 0.228 1 15.56 30.54 44.42 56.84 67.51 76.33 83.34 88.69 92.61 95.33
Missouri
-0.213 0.0233 0.1526 10.08 21.02 33.21 46.37 59.54 71.59 81.58 89.00 93.98 97.00
River4

Vanoni’s CSU1
-0.15 1 0.0740 0.2721 12.52 24.76 36.44 47.31 57.17 65.92 73.49 79.86 85.07 89.19
approach to Pakistan2
-0.036 0.0348 0.18M 2034 40.10 57.01 70.72 81.14 88.57 93.45 96.46 98.22 99.15
solve USGS3
0.204 0.0822 0.2867 10.79 21.38 31.67 41.51 50.70 59.15 66.80 73.53 79.30 84.18
Einstein’s Missouri
0.155 0.0101 0.1004 12.76 2830 47.94 67.30 82.80 92.56 97.39 99.26 99.83 99.97
formula River4
1 cf. to Guy, Simons and Richardson (1966) 4 Missouri River Hydr. Anal. from Sioux City to mouth
2 cf. to Culbertson, Scott and Bennett (1972) 5 The original work was completed by H.A. Senturk
3 cf. to Schiller, Jr. (1968)

138
 Flow Resistance in Streams

1. CSU,
2. USGS, and
3. Pakistan irrigation canal data

Table 15 summarizes the results of this study.

5.9. Conclusions
• Resistance to flow of a movable bed is a very complex problem.
• In lower regime resistance to flow is large and hence d is large and the U is small for
a given Q. As Q↑, the transition zones are encountered.
• The transition from ripples to dunes or from dunes to plane bed is accompanied by
fluctuations in flow characteristics.
• All of the methods presented consider the importance of bed forms.
• Engelund’s (1965), Simons and Richardson’s (1966), Richardson and Simons’
(1967) and Sentürk’s (1967-1973) methods take explicit care of this effect. The
other methods do implicitly.
• None of the methods is generally acceptable for estimating resistance to flow in
alluvial channels.
• However, Einstein and Barbarossa’s (1952), Simons and Richardson’s (1966),
Richardson and Simons’ (1967) and Sentürk’s (1974) methods have been validated
by laboratory data and some field data.
• A comparison of various formulas has shown that Einstein and Sentürk’s methods
give approximately same standard deviations when applied to streams.
• However, Sentürk’s formula is more precise.
• Sentürk’s method and Simons and Richardson’s method give good results for
artificial sand-bed canals.
• Simons and Richardson’s method is suggested for a channel of gravel and cobble
bed or concrete-lined bed.

Table 15. Efficiency of resistance formulae when applied to different bed configurations

Formulae Lower TR Ripple Middle TR Dunes Upper TR Plane Bed

X =56.92% 38.82% 24.82% 16.83% 20.0% 12.30%

1. Einstein
(N.R) (N.R) (N.R) (R) (R) (R)
X =38.57% 21.97% 27.20% 53.78% 16.07% 28.9%
2. H.W.Shen
(N.R) (R) (N.R) (N.R) (N.R) (R)
X =18.21% 15.40% 11.52% 12.10% 17.50% 14.10%
3. Sentürk
(N.R) (R) (R) (R) (R) (R)
X =15.76% 16.20% 18.94% 20.83% 20.63% No
4. Simons
(N.R) (R) (R) (R) (R) Prediction

X =
Absolute mean value of the errors obtained by comparison of
the computed and measured resistance to flow
N.R. = Not recommended
R = Recommended

139
Sediment Transport Technology

References

1. Ackers, P., “Resistance to Fluids Flowing in Channels and Pipes”, Hydraulic

Research Paper No. 1, H.M.S.O., London, (1958).
2. Alan, M.S., and Kennedy J.F., “Friction Factors for Flow in Sand-Bed Channels”,
J. of the Hydraulics Div., ASCE, Vol.95, No.HY6 Proceedings Paper 6900, pp.
1973-1992, (1969).
3. Bajorunas, L., “Discussion of “River Channel Roughness”, by H.A. Einstein and
N.L. Babarossa. Trans. ASCE, Vol. 117, pp1140-1142, (1952).
4. Beckman, E.W., and Furness L.W., “Flow Characteristics of Elkhorn River near
Waterloo, Nebraska”, U.S Geological Survey, Water-Supply Paper 1498-B, 34 p,
(1962).
5. Bishop, A.A, D.B. Simons and Richardson E.V., “Total Bed Material Transport”, J.
of the Hydraulics Division, ASCE, Vol.91, HY2, March (1965).
6. Brooks, N.H. “Mechanics of Stream with Movable Beds of Fine Sand”, Trans.
ASCE, Vol. 123, (1958).
7. Chow, V.T., “Open-Channel Hydraulics”, McGraw-Hill, New York, (1959).
8. Einstein, H.A., “The Bed Load Function for Sediment Transport in Open Channel
Flows”, Technical Bulletin 1026, U.S. Dept. of Agriculture, Soil Conservation
Serv., Washington, D.C., 70 p, (1950).
9. Einstein, H.A. and Barbarossa, N.L., “River Channel Roughness”, Trans. ASCE,
Vol.117, pp. 1121-1132, (1952).
10. Engelund, F., “Hydraulic Resistance of Alluvial Streams”, J. of the Hydraulics
Division, ASCE, Vol.92 HY2, Proceedings Paper 4739, pp.315-326, March (1966).
11. Engelund, F. and Hansen E., “A Monograph on Sediment Transport in Alluvial
Streams”, Teknisk Forlag, Copenhagen, (1972).
12. Guy, H.P., D.B. Simons,and Richardson E.V., “Summary of Alluvial Channel Data
from Flume Experiments, 1956-1961”, U.S. Geological Survey, Professional Paper
462-I, 96p., (1966).
13. Harrison, A.S., and Mellema W.J., “Movable Bed Model for Alluvial Channel
Studies”, IAHR, 12th Congress, Fort Collins, Colorado, (1967).
14. Keulegan, G.H., “Laws of Turbulent Flows in Open Channels”, J. National Bureau
of Standards, Vol. 21, Research Paper 1151, pp. 707-741, December (1938).
15. Laursen, E.M., “An Investigation of the Total Sediment Load”, J. of the Hydraulics
Division, ASCE, Vol.84 HY1, Proceedings Paper 1530, February (1958).
16. Lovera, F., and Kennedy J.F., “Friction-Factors for Flat-Bed Flows in Sand
Channels”, J. of the Hydraulics Division, ASCE, Vol.95 HY4, Proceedings Paper
6678, pp.1227-1234, July (1969).
17. Meyer-Peter, E., and Müller R., “Formulas for Bed-Load Transport”, Proc. 3rd
Meeting IAHR, Stockholm, pp.39-64, (1948).
18. Nordin, C.F., Jr., “Aspects of Flow Resistance and Sediment Transport: Rio Grande
near Bernalillo, New Mexico”, U.S. Geological Survey, Water-Supply Paper 1498-
H, (1964).
19. Prandtl, L., “Über die Ausgebildete Türbülenz”, ZAMM, 5, pp.136-139, 1925; and
Proc. 2nd Intern. Congr. Appl. Mech., Zürich, pp. 62-75; also Coll. Works II, pp.
736-751 (1926).
20. Richardson, E.V., and Simons D.B., “Resistance to Flow in Sand Channels”, Proc.
12th Congress IAHR, Fort Collins, Colorado, pp.141-150, (1967).

140
 Flow Resistance in Streams

21. Sayre, W.W., and Albertson M.L., “Roughness Spacing in Rigid Open Channels”,
J. of the Hydraulics Division, ASCE, Vol.87 HY3, Proceedings Paper 2823,
pp.121-150, May (1961).
22. Sentürk, F., “La Resistance de Fond des Ecoulements Naturales Etudiés en Fonction
d’un Noveau Parametre (f) Appele Parametre de Friction”, C.R.S.A.S., Paris, Vol
260, p.408, (1965).
23. Sentürk, F., “Determination de laResistance d’un Fond Mobile Sous 1’action d’un
Ecoulement”, IAHR 12th Congress, Fort Collins, Colorado, (1967).
24. Sentürk, F., “Mechanisc of Bed Formulations”, La Houille Blanche, No. 2, (1969).
25. Sentürk, F., “A New Category of Bed Resistance to Flow Formulas”, Technical
Bulletin, No. 26, DSI Publications, Ankara, Turkey, (1976).
26. Sentürk, F., “Sediment Transport Technology”, Symposium, UNESCO-DSI,
Ankara, Turkey, (1987).
27. Sentürk, F., “The Buildup of Resistance to Flow in Natural Streams”, Post
Graduate Course in Sediment Transport Technology, UNESCO-DSI, Vol. 1,
Ankara, Turkey, (1989) and (1994).
28. Shen, H.W., “Development of Bed Roughness in Alluvial Channels”, J. of the
Hydraulics Division, ASCE, Vol.88, No. HY3, (1962).
29. Simons, D.B., and E.V. Richardson, “Resistance to Flow in Alluvial Channels”,
U.S. Geological Survey, Professional Paper 422-J, (1966).
30. Simons, D.B., and Sentürk F., “Sediment Transport Technology, Water and
Sediment Dynamics”, Water Resources Publication, Littleton, Colorado, (1992).
31. Strickler, A., “Some Contributions to the Problems of the Velocity Formulas and
Roughness Factor for Rivers, Canals and Closed Conduits”, Mitteilungen des
Eidgennössischer Amtes fur Wasserwirtschaft, Bern, Switzerland, No. 16 g,
(1923).
32. Taylor, R.H., and Brooks N.H., “Resistance to Flow in Alluvial Channels”,
Discussion by D.B. Simons and E.V. Richardson, Trans. ASCE, Vol. 127, Part I,
pp. 982-992, (1962).
33. Tracey, H.J., and Lester C.M., “Resistance Coefficient and Velocity Distribution-
Smooth Rectangular Channel”, U.S. Geological Survey, Water Supply Paper 1592-
a, 18 p., (1961).
34. Vanoni, V.A., and Brooks N.H., “Laboratory Studies of the Roughness and
Suspended Load of Alluvial Streams”, Report E-68, California Inst. of Tech.,
Pasadena, California, (1957).
35. Zegzhda, A.P., “Theory of Similarity and Methods of Design of Models for
Hydraulic Engineering”, Gosstroiizdat, Leningrad, (1938).

141
142
 INITIATION OF PARTICLE MOTION

Prof. Dr. Mustafa GÖĞÜŞ

Civil Engineering Department
Middle East Technical University
06531 Ankara - Turkey

1. Introduction
The studies on the subject of sediment transportation go back to very old times. The
first advancements were developed in China, Mesopotamia, Egypt and Anatolia about
thousands of years before the start of present era. In the years of renaissance the
Leonardo da Vinci was the first to introduce the experimental method (Simons and
Senturk, 1992).

One of the main questions in sediment transportation has been the beginning of motion
of particles. This boundary state of movement and resting has been named in many
ways. Initiation of motion, inception of motion, incipient motion, threshold of motion...
all refer to this critical condition. The earliest research on the subject has been made by
Brahms (1753), in which he resulted with the equation:

Vd = K ′ ⋅ W 1 / 6 (1)

which is known as the sixth power law. Here, W is the weight of the grain, K′ is an
empirical constant and Vd is the competent velocity.

Starting with Brahms, for more than two centuries many researchers have worked on
the problem of incipient motion. The Task Force Committee on Preparation of the
Sedimentation Manual prepared a general bibliography related to the topic in 1966.
Then Lavelle and Mofjeld (1987) updated the bibliography by including the subsequent
work after 1966. Since then many additional researches have been made that bear on
conditions of initial motion.

Inside the deep and branched literature that the incipient motion has, there lies an
important fact. Like the different names it has, there are also different definitions for the
critical condition. The main divergence in the definitions is due to the methodology
used. There are primarily two methods, which can be used: deterministic approach and
stochastic approach. The deterministic approach tries to relate a single displacing force
to each particle size. Stochastic approach, on the other hand, tries to include the effect
of various measurable and non-measurable factors (Stelczer, 1981).

It is known from the previous experiences that the initiation of particle motion has a
random character in time and space. Because of this statistical aspect, stochastic
methods are welcome. Deterministic approach, which may be regarded as classical, tries
to come up with a single result. Nevertheless, due to the fluctuating character of the

143
Sediment Transport Technology

phenomenon, repeated experiments are performed in this approach and a range of

critical conditions is covered. Consequently the mean of these critical conditions is
adopted as the final result.

In the deterministic approach, a deterministic relationship is established between

displacing forces and resisting forces. If the displacing force is expressed in terms of
velocity of the flow it is based on the impact theory. In this type of approach the particle
of diameter d starts to move by the thrust force applied on its projected area
perpendicular to the flow. Another method is frictional drag theory, in which the
particle of diameter d starts to displace by the friction force of the water column moving
on it.

Another dilemma in the results obtained arises from the technique used in the
determination of critical condition criterion. The criterion can be based on visual
observation or it can be a reference based one (Buffington, 1999). The visual technique
is very subjective, since the critical condition may be taken as the movement of a single
particle or a group of particles, amount of which is up to the observer. In this technique
usually the work of Kramer (1935), is taken as a reference. Kramer classified the types
of movement into three, weak movement, in which only a few particles are in motion,
medium movement, in which, particles having mean diameter are in motion and general
movement, where there is a continuous transport of bed material (Stelczer, 1981).
Reference technique is the one in which a general bed load transport formula is
extrapolated to zero transport value. However, most of the bed-load transport formulae
yield zero transport for zero velocity. For this reason a very small value of transport is
taken as reference. To be illustrative, according to Buffington (1999), Shields (1936),
who left a remarkable trace on incipient motion subject, has used reference technique to
determine incipient motion thresholds for his experiments. Even though many
researchers have studied on the subject of incipient motion, the effect of particle shape
has not been examined explicitly yet.

2. Theoretical Considerations
2.1. Analytical approach and the Brahms equation
Brahms (1753) equated the dynamic force on the particle to the resisting force of the
particle.

πd s2 Vd2 πd s3
ξ⋅ ⋅ρf ⋅ = ⋅ (γ s − γ f ) ⋅ tan φ (2)
4 2 6

in which, ρf and γf are density and specific weight of fluid respectively.

γs : specific weight of the sediment particle,
ds : diameter of the particle,
Vd : competent velocity of the flow,
tanφ : friction coefficient,
φ : angle of repose,
ξ : fraction indicating the part of the total area of the particle exposed to flow

144
 Initiation of Particle Motion

4 ρ tan φ
Vd2 = ⋅ g ⋅ ( s − 1) ⋅ ⋅ ds (3)
3 ρf ξ
Vd2 = K ⋅ d s (4)

where

4 ρ tan φ
K= ⋅ g ⋅ ( s − 1) ⋅
3 ρf ξ
Vd6 = (K ⋅ d s )
3
(5)

where (K ⋅ d s ) ∝ W ,weight of the sediment particle. Then,

3

Vd = K ′ ⋅ W 1 / 6 (6)

at the condition of incipient motion.

2.2. Dimensional analysis and the Shields diagram

Albert Frank Shields is one of the most cited researchers when the topic of incipient
motion is considered. Unfortunately his work has been lost during World War 2
(Buffington, 1999). What Shields (1936) did was a dimensional analysis of the question.

From dimensional analysis point of view, inception of motion is a critical condition,

which involves many measurable and non-measurable constraints. Therefore,
dimensional analysis is a very useful method to obtain a functional relation between the
constraints. The parameters involved in the condition of incipient motion are classified
as:

Flow characteristics:
V : Velocity of the flow,
h : Depth of flow,
B : Cannel width
S : Slope of the flow,
g : Gravitational acceleration,
k : Roughness height of the bed.

Properties of fluid:
ρ : Density of the fluid,
ν : Kinematic viscosity.

Properties of the sediment particles:

ds : Size of sediment particle,
SF : Shape factor of particle,
ρs : Density of particle,
ξ : Particle size distribution.

145
Sediment Transport Technology

The parameters given above are rather generalized and for this single case a few
modifications are possible. In this research solitary particles and group of uniform
particles are used. Therefore there is no effect of particle size distribution, ξ. The
measurements are made under uniform flow conditions. Consequently the slope of the
flow, S, is merely the slope of channel bottom (Fig. 1).

f (V , h, B, S , g , k , ρ ,ν , d s , SF , ρ s ) = 0 (7)

The resulting dimensionless parameters are:

⎛ V V ⋅ ds ds ds ds ρs ⎞
f⎜ , , , , , , SF , S ⎟ = 0 (8)
⎜ g ⋅d ν h B k ρ ⎟
⎝ s ⎠

The term

V
g ⋅ ds

is known as particle Froude number and the term

V ⋅ ds
ν

is the grain Reynolds number. If shear velocity, u*, is used instead of flow velocity, V;
the effect of slope, S, and flow depth, h, is taken into account implicitly.

τ0
u* = and τ0 = γ ⋅h⋅S (9)
ρ

The particle Reynolds and Froude numbers can be rewritten in terms of shear velocity:

⎡ V ⎤ ⎡ u* ⎤ ρ ⋅u 2 =τ 0 ⎡ τ0 ⎤
⎢ ⎥→⎢ ⎥ ⎯⎯*⎯⎯→⎢ ⎥ (10)
⎢⎣ g ⋅ d s ⎥⎦ ⎢⎣ g ⋅ d s ⎥⎦ ⎣ (γ s − γ ) ⋅ d s ⎦

V ρ,ν
h

ds , ρs

S
Figure 1. Parameters related to the condition of incipient motion

146
 Initiation of Particle Motion

⎡V ⋅ d s ⎤ ⎡ u* ⋅d s ⎤
⎢ ν ⎥→⎢ ν ⎥ (11)
⎣ ⎦ ⎣ ⎦

The effect of ratio ρs/ρ of sediment density to fluid density is also taken into account in
the final form of the particle Froude number. For the specific case of B>>ds, the
influence of ds/B can be neglected. However, for coarse particles in shallow flow, the
condition of h>>ds is not satisfied. Therefore it is not possible to drop the term ds/h.
Instead of this term the use of parameter ds/Rb is proposed in order to count for the
effect of the finite width of the channel. Here, Rb is the effective hydraulic radius
corresponding to the bed only. Rb is obtained using sidewall correction procedure,
which is proposed by Johnson (1942) and modified by Vanoni and Brooks (1957).
Moreover if the ratio of ds/k is kept constant then the resulting dimensionless parameters
related to the incipient motion can be represented simply as:

⎛ τ0 u ⋅d d ⎞
f ⎜⎜ , * s , s , SF ⎟⎟ = 0 (12)
⎝ (γ s − γ ) ⋅ d s ν Rb ⎠

The relationship:

u* ⋅ d s ⎛ τ0 d ⎞
= f ⎜⎜ , s , SF ⎟⎟ (13)
ν ⎝ (γ s − γ ) ⋅ d s Rb ⎠

is almost the same relationship with the one that Shields utilized in his experiments.
However, this time, grain Reynolds number is expressed as a function of entrainment
function together with a shape factor and including the effect of relative hydraulic
radius.

Since there are many variables concerned with the problem after the analysis, Shields
simplified the problem for some certain conditions. Assuming that Re>>ds and all the
sediment particles are uniform, he eliminated the terms ds/Rb and SF from Eq. 13 and
defined the parameters effecting the initiation of motion and resulted with the relation
(Yalin, 1977):

τc u ⋅d
= f ( *cr s ) (14)
(γ s − γ ) ⋅ d s υ

τc : critical bed shear stress

γs : specific weight of the particle
γ : specific weight of the fluid
ds : diameter of the particle
u*cr : critical shear velocity
υ : kinematic viscosity of the fluid

The term, τc/[(γs-γ)·ds], on the left hand side is named as entrainment function. It is
equivalent to densimetric Froude number and it is usually denoted as Fr. The term,

147
Sediment Transport Technology

u*cr·ds/υ, on the right hand side is named as particle Reynolds number or grain Reynolds
number and usually referred as Re*.

Shields (1936) was the first researcher to include the effect of boundary layer (Stelczer,
1981). Considering the bed roughness, Shields divided the function that he obtained into
three regions:
1. Hydraulically smooth region,
2. Transitional region,
3. Hydraulically rough (wholly turbulent) region.

According to Shields the bed is hydraulically smooth for grain Reynolds number, Re*,
less than 1.0 (Stelczer, 1981). In this case the critical shear stress independent of the
roughness of the bed. The boundary layer thickness is larger than the representative
grain size, Ds. In case of 1.0 < Re* < 600 the flow is in transitional region and the
critical condition depends on the bed roughness and the viscosity. For Re* > 600 the
boundary condition is completely rough. In this region the boundary layer thickness is
small when compared to the particle size. The entrainment function is constant and
equal to 0.06.

τc
= 0.06 (15)
(γ s − γ ) ⋅ d s

The graph that Shields obtained is given in Fig. 2. It is applicable to fully developed
turbulent flow within the ranges of sediment sizes from 0.4 mm to 3.4 mm. After
Shields, many comments and revisions have been made on his work. Shields proposed a
wide band accounting for the variability in the incipient motion values. The curve
shown in Fig. 2 was fit by Rouse (1939, 1949).

The Task Committee on Preparation of Sedimentation Manual (1966) suggested the use
of a third parameter:

ds ⎛γ ⎞
⋅ 0.1 ⋅ ⎜⎜ s − 1⎟⎟ ⋅ g ⋅ d s (16)
υ ⎝γ ⎠

where, γs is the specific weight of the particle, γ is the specific weight of the fluid, ds is
the diameter of the particle, and υ is the kinematic viscosity of the fluid. Consequently,
from the intersection of correct parallel line with the Shields diagram, corresponding
value of entrainment function is found. Richardson and Simons (1957), Vanoni (1966),
Henderson (1966), Shulits and Hill (1968), Senturk (1969), Gessler (1971), Paintal
(1971), Miller et al. (1977), Yalin and Karahan (1979), contributed to the Shields
diagram with their studies (Simons and Senturk, 1992 and Buffington, 1999).

2.3. Other experimental approaches

The concept of lift force has been studied by Einstein and El-Samni (1949) and
Apperley (1968). Einstein and El-Samni conducted the experiments on two different
sediment beds. One of the beds used was made of hemispheres of 0.225 ft diameter

148
 Initiation of Particle Motion

Figure 2. Shields diagram. (Simons and Senturk, 1992)

cemented to the bed in a hexagonal pattern and the other was consisting of gravel with
mean size approximately equal to 0.225 ft. Einstein and El-Samni has measured the
difference in mean static pressure in sediment beds at level of the bottom of the top
layer of sediment and at the wall of the channel at the level of the top of top layer of
sediment as stated by the ASCE Task Committee (1975).

These experiments has yielded a pressure difference of

1
∆p = 0.178 ⋅ ⋅ ρ ⋅ V02 (17)
2

which is the lift pressure on the grains and where V0 is the velocity at a distance of
0.35d35 above the theoretical bed and d35 is the size of grains for which 35% by weight
of the bed material is finer. For the uniform hemispheres this size is taken as the
diameter, d. This equation is valid only for rough boundaries and the range of Re* in the
experiments were in the range of 50 000.

The theoretical bed is defined as the position of the origin of y for which the measured
velocity profiles satisfy the equation below with ar equal to 8.5. This equation gives the
velocity profile for a two-dimensional, free surface flow over a fixed bed.

149
Sediment Transport Technology

u y
= a r + 5.75 ⋅ log (18)
u* ks

where,
u : velocity at a distance y above the bed,
u* : shear velocity,
ks : characteristic roughness size of sediment,
ar : a function of the boundary Reynolds number, u*·ks/υ, which is obtained
experimentally.

The level of theoretical bed is equal to 0.2 times the sphere diameter for the bed
consisting of hemispheres and 0.2.d67 for the gravel bed. This distance is measured from
below the tops of the uppermost grains and d67 is the size of the sediment for which
67% by weight is finer in case of gravel bed.

Novak and Nalluri (1975) have investigated the condition of incipient motion in circular
conduits and rectangular flumes with fixed smooth beds and free surface flow. They
have used a tilting flume. Novak and Nalluri have worked with sand, gravel, plastic,
anthracite, and lead particles. The sand and gravel sizes ranged from 0.6 mm to 50 mm.
The particle shape factor they have utilized was equal to c a ⋅ b , where c is the minor
axis and a and b are other two axes. The plastic particles had a shape factor varying
between 0.6~0.95 while the shape of anthracite particles were quite comparable to
natural sands. The lead particles were rather spherical with shape factors changing
between 0.77~0.98. Specific gravities of the particles were 1.18, 1.6 and 11.74, for
plastic, anthracite and lead particles respectively.

In the end of their research, Novak and Nalluri have obtained the relations for a range of
10< Re* < 1000:

V0
0.35.d

0.2.d
theoretical bed level

Figure 3. Theoretical bed in case of hemispheres

V0
0.35.d35
0.2.d67
theoretical bed level

Figure 4. Theoretical bed in case of gravels

150
 Initiation of Particle Motion

Fr* = 0.065 ⋅ ( Re* ) −0.52 (19)

−0.61
Fr* = 0.06 ⋅ ( Re* ) (20)

for the rectangular channel and for the channels with circular cross section respectively.
The comparison of their result with the Shields diagram is given in Fig. 5.

In terms of critical velocity the equations are represented as:

Vcc = 0.17 ⋅ ( S s − 1)1 / 2 ⋅ d s0.24 (21)

Vcc = 0.16 ⋅ ( S s − 1)1 / 2 ⋅ d s0.16 (22)

for the rectangular channel and for the channels with circular cross section respectively.
Here, Vcc is the critical cross sectional velocity in m/s, Ss is the specific gravity of the
sediment, ds is the diameter of sediment particle in mm.

In terms of critical shear stress the equations are given as:

τ c = 0.128 ⋅ ( S s − 1) ⋅ d s0.4 (23)

Eq.19
Fr*2

Eq.20

Re*
Figure 5. Entrainment function versus particle Reynolds number (Novak and Nalluri, 1975)

151
Sediment Transport Technology

for rectangular channel, and

τ c = 0.104 ⋅ ( S s − 1) ⋅ d s0.4 (24)

for circular channels. Here tc is obtained in N/m2 and ds is taken in mm.

In order to decrease the effect of channel shape, Novak and Nalluri, proposed to analyze
the results in terms of relative roughness and presented the equation:

−0.27
Vcc ⎛d ⎞
= 0.61 ⋅ (S s − 1)
1/ 2
⋅⎜ s ⎟ (25)
g ⋅ ds ⎝R⎠

where, relative roughness ds/R is the ratio of particle diameter to hydraulic mean radius
of conveyance. The equation is valid for the range of 0.01 < ds/R < 1.

In their next study Novak and Nalluri (1984) have studied on the incipient motion of
single and grouped particles on fixed smooth and also fixed rough beds with roughness
smaller than the particle size. The experiments were done in tilting flumes with circular
and rectangular cross sections. The rectangular channel was artificially roughened. The
smaller roughness sizes (0.3 mm, 0.42 mm) were achieved by gluing waterproofed sand
paper to the bed while the larger roughness sizes (1.44 mm, 2.2 mm, 4.2 mm) were
achieved by sticking coarse gravel and sand particles to the bed. Equivalent diameter
sizes of particles were from 0.6 mm to 50 mm and average relative density was 2.56. In
each set of the experiments particle sizes were always larger than the roughness sizes.
Grouped particles were studied in separate series according to their positioning;
particles spaced along the width of the flume (W-series), particles spaced longitudinally
along the centerline of the flume (L-series), particles touching each other in rows across
the width of the flume (T-series).

For grouped particles on smooth fixed beds they have showed that if the particles touch
each other their stability is in between the stability of a single particle on a smooth bed
fixed and the stability of particles in case of movable beds given by Shields. Fig. 6
demonstrates this condition, n being the number of rows of sediment particles touching
each other resting on a smooth fixed bed. As the number of rows increases there is a
departure from the critical values for isolated single particles towards movable bed
behavior.

On the other hand, if particles are spaced across the width of the channel or along the
channel on a smooth fixed bed; the increasing the spacing of particles has an effect in
favor of stability. This result is represented in the following figure, where with
increasing spacing between the particles the critical velocity of the group, Vsg
approaches to that of a solitary particle, Vss. Here, the subscript ss stands for smooth
boundary, single particle and sg denotes the condition of smooth boundary and grouped
particles. Novak and Nalluri have stated that for both W- and L-series Vsg/Vss = 1 for
r/a ≅ 20 and Vsg/Vss ≅ 0.8 for r/a ≅ 1. Here r is the space between the grouped particles
(lateral and longitudinal) and a is the major axis of the particle i.e. axis perpendicular to
the direction of flow.

152
 Initiation of Particle Motion

Fr*2

Re*
Figure 6. Fr* vs. Re* n row of grouped particles touching each other (Novak and Nalluri, 1984)
Vsg/Vss

r/a
Figure 7. Vsg/Vss against r/a for k = 0 (Novak and Nalluri 1984)

The data from the investigation of incipient motion of discrete particles on beds with
various roughness elements were analyzed and Fig. 8 was prepared as
Vrs g ⋅ d s ⋅ ( S s − 1) versus d s Rb . Here, Rb is the effective hydraulic radius
corresponding to bed only and Vrs is the critical velocity of a single particle on a rough
fixed bed. An average approximate relationship for the entire data suggests the
following equation:

Vrs d s −0.38
= 0.54 ⋅ ( ) (26)
g ⋅ d s ⋅ ( S s − 1) Rb

153
Sediment Transport Technology

Eq.26
Eq.25
V rs
g ⋅ d s ⋅ ( S s − 1)

ds/Rb
Vrs
Figure 8. against ds/Rb (Novak and Nalluri, 1984)
g ⋅ d s ⋅ ( S s − 1)

For comparison Fig. 8 also shows the modified Shields equation (using the Strickler’s
equation for Manning’s n) and Eq. 25 applicable to single particles on a smooth bed. As
can be seen from Fig. 8 and the constants in Eqs. 25 and 26, the difference between
these two cases of single on smooth or rough beds are not large particularly for
relatively high values of ds/Rb, with the critical velocity for a given set of values of ds,
Ss, and Rb always being bigger for the rough than the smooth bed.

In case of rough fixed beds and single particles, ratio of particle diameter to roughness
height plays an important role. Novak and Nalluri obtained the graph below, which is
quite similar to the well-known Nikuradse graph showing the friction coefficient as a
function of Reynolds number. The graph indicates that the value of Fr*2 will be a
function of ds/k for Re* well above 1000.
Fr*2

Re*
Figure 9. Fr* against Re* for single particles and constant ds/k values (Novak and Nalluri 1984)

154
 Initiation of Particle Motion

The effect of ds/k ratio on the critical condition is given in Fig. 10. This effect is given
by the following equation within the range of 75 > ds/k > 2.

Vrs d
= 1 + 1.43( s ) −0.4 (27)
Vss k

Vrs : critical velocity of single particles on rough beds

Vss : critical velocity of single particles on smooth beds.

Fig. 9 indicates that the effect of ds/k diminishes for ds/k > 100.

A more generalized representation of the results together with the grouped particles
with zero spacing and resting on a rough fixed bed is given in Fig. 11. Here it is clearly
seen that the values of Fr*2 against Re* are bounded by the extreme conditions ds/k=1
and the smooth bed conditions.

From the plot of Vrt g ⋅ d s ⋅ ( S s − 1) against d s Rb including other bed roughness sizes
and the smooth bed (k=0) could be in spite of scatter and trend approximated by a single
equation (see Fig. 11):

Vrt d s −0.40
= 0.50 ⋅ ( ) (28)
g ⋅ d s ⋅ ( S s − 1) Rb

where Vrt is the critical velocity of touching particles on rough fixed bed. From Fig. 11
it can be concluded that once we have a sufficient number of touching particles the
effect of the bed roughness on the critical velocity in this representation is very small an
that, furthermore, the results are close to those obtained for a discrete particle resting on
a rough bed.

Figure 10.Vrs/Vss against d/k (Novak and Nalluri 1984)

155
Sediment Transport Technology
Fr*

Figure 11. Fr* versus Re* (Novak and Nalluri 1984)

When Eqs. 25, 26 and 28 are compared it is seen that all the equations are of the type:

Vc ds b
= a⋅( ) (29)
g ⋅ d s ⋅ ( S s − 1) R

where
Vc : critical velocity
Ss : specific gravity of sediment particles
ds : size of sediment particles
R : hydraulic radius of flow

Here, a and b are two empirical constants given in the Table 1.

Table 1. Emprical values of the constants a and b in Eq. 29

a b Eq.
Smooth bed single particles 0.61 -0.27 25
Rough bed single particles 0.54 -0.38 26
Rough and smooth bed touching particles 0.50 -0.40 28
Movable bed (ds=k) 1.7 ~ 1.9 -0.095 ~ -0.167 -
Note: range of experiments 0.01 < ds/R < 0.3, 3.5 < ds/k ≤ ∞

156
 Initiation of Particle Motion

References

1. Apperly, L. W., “Effect of Turbulence on Sediment Entrainment”, thesis presented

to the University of Auckland, Auckland, New Zeland, in partial fulfillment of the
requirements for the degree of Doctor oh Philosophy, (1968).
2. Brahms, A. “Anfangsgrunde der Deich und Wasserbaukunst”, Aurich, (1753).
3. Buffington, John M., “The legend of A: F: Shields”, J. Hydr. Eng., ASCE, 125,
376-387, (1999).
4. Einstein, H. A., and El-Samni, E.S:A:, “Hydrodynamic Forces on a Rough Wall”,
Review of Modern Physics, vol. 21, pp. 520-524, (1949).
5. Gessler, J., “Aggradation and Degradation”, Chap. 8, River Mechanics, H.W. Shen
(ed.)., Water Resources Publications, Littleton, Colorado, (1971).
6. Johnson, J. W., “The importance of Considering Side-Wall Friction in Bed-Load
Investigations”, Civil Engineering, ASCE, Vol. 12, No. 6, p. 329-331, June, (1942).
7. Ippen, A. T.,“Channel transitions and controls”, chap. VIII in Haunter Rouse
(editor) Engineering Hydraulics, John Wiley & Sons, Inc., New York, p. 496-588,
(1950).
8. Kramer, H., “Sand mixtures and sand movement in fluvial models”, Transactions,
ASCE, 100, 798-878, (1935).
9. King, H. W., “Handbook of Hydraulics”, McGraw Hill Book Company, Inc., New
York, (1954).
10. Lavelle, J. W. and Mofjeld, H. O., “Bibliography on sediment threshold velocity”, J.
Hydr. Eng., ASCE, 113: 389-393, (1987).
11. Miller, M. C., McCave, I. N., and Komar, P. D. “Threshold of sediment motion
under unidirectional currents”, Sedimentology, 24(4), 507-527, (1977).
12. Novak, P., and Nalluri C., “Sediment Transport in Smooth Fixed Bed Channels”, J.
Hydr. Res., 101, No. HY9, 1139-1154, (1975).
13. Novak, P., and Nalluri C., “Incipient Motion of Sediment Particles Over Fixed
Beds”, J. Hydr. Res., 22(3), 181-197, (1984).
14. Paintal, A. S., “Concept of critical shear stress in loose boundary open channels.” J.
Hydr. Res., 9(1), 91-113, (1971).
15. Raudkivi, A. J., “Loose Boundary Hydraulics”, Pergamon Press Ltd., (1967).
16. Rouse, H., “An analysis of sediment transportation in light of fluid turbulence”,
Rep. No. SCS-TP-25, Sediment Division, U.S. Dept. of Agr., Soil Conservation
Service, Washington, D.C., (1939).
17. Rouse, H., “Fundamental principles of movement”, Engineering Hydraulics, Proc.,
4th Hydr. Conf., Wiley, New York, 786-804, (1949).
18. Richardson E. V., and Simons D. B., “Discussion of Liu, H. K., 1957 paper”, P. A.
S. C. E. H. D., September, (1957).
19. Senturk, F., “Nehir Hidroligi – River Hydraulics”, DSI publications, No: 66/X/64,
Ankara, Turkey, (1969).
20. Shields, A. F., “Anwendung der Aenlichkeitsmechanikund der Turbulenz-forschung
auf die Geschiebebewegung”, Mitteeilungen der Preusssischen Versuchenstalt für
Wasserbau und Schiffbau, Heft 26, Berlin, (1936).
21. Shulits, S., and Hill R. D., “Bed Load Formulas, Part A. A Selection of Bed Load
Formulas”, Pennsylvania State Univ., College of Eng., University Park,
Pennsylvania, (1968).
22. Simons, D. B., and Senturk, F. “Sediment Transport Technology”, Water Resources
Publications, Littleton, Colo., (1992).

157
Sediment Transport Technology

23. Stelczer, K., “Bed-Load Transport: Theory and Practice”, Water Resources
Publications, Littleton, Colo., (1981).
24. Task Committee on Preparation of Sedimentation Manual, “Sediment transportation
mechanics: Initiation of motion”, P. ASCE, HY2, 4738, (1966).
25. The ASCE Task Committee for the Preparation of the Manual on Sedimentation of
the Sedimentation Committee of the Hydraulics Division, “Sedimentation
Engineering”, ASCE, N.Y., (1975).
26. Vanoni, V. A., “Sediment transportation mechanics: Initiation of motion” J. Hydr.
Div., ASCE, 92(2), 291-314, (1966).
27. Vanoni, V. A., and Brooks, N. H., “Laboratory Studies of the Roughness and
Suspended Load of Alluvial Streams”, Sedimentation Laboratory Report No. E68,
California Institute of Technology, Pasadena, Calif., (1967).
28. Yalin, M. S., and Karahan, E., “Inception of sediment transport.” J. Hydr. Div.,
ASCE, 105(11), 1433-14443, (1979).
29. Yalin, M. S., “Mechanics of Sediment Transport”, Pergamon Press, Oxford; New
York, (1977).

158
 WATER WAVE MECHANICS

Assoc. Prof. Dr. Can Elmar Balas

Civil Engineering Department
Gazi University
Ankara Turkey
e-mail: cbalas@gazi.edu.tr; fax: (+90) 312 231 92 23

1. Introduction
In coastal zones, water waves are important dynamic factors for sediment transport. The
propagations of water waves are affected by variable topography and current, the most
apparent phenomena are the wave shoaling (wave heights increase with the water depth
decrease), wave refraction and diffraction. Both the nature and the frequency of wave
action have impacts on coastal sediment dynamics. Waves allocate sediment in relation
to the distribution and extent of wave forces, where the coarse sediment particles are in
regions characterized by high wave forces and fine-grained deposits in areas subject to
low energy waves.

Sea waves are generated by the wind energy transferred across the water surface and the
size of a wave depends on the generation area and duration of the wind. The turbulence
in the wind produces random pressure fluctuations at the sea surface, which produces
small waves with wavelengths of a few centimeters. Wind blowing over the wave
produces pressure differences along the wave profile causing the wave to grow
exponentially. Finally, the interaction between waves transfers energy from short waves
to waves having higher frequencies in the spectrum.

The primary factor that determines the maximum wave height is the wind speed. But the
duration of the wind (length of time the wind has been blowing at a certain wind speed)
and the fetch, (distance over the water the wind is blowing) also limit the maximum
wave height. When the highest theoretical wave height based on the wind speed cannot
be attained because the wind is not blown for a sufficient period of time, the wave
height is duration limited. When the wave height cannot be attained because the straight
line of which the wind is blowing over the water is too short, the wave height is fetch
limited. Significant wave height (H1/3), is the average of the heights of the highest one-
third of the waves measured over a stated interval of time.

As the waves shoal and are affected by bottom friction, they increase in height and
steepness until they break . The turbulence and currents generated by wave action stir
and suspend sediments resulting in sediment movement along the shoreline by wave
related nearshore currents and tidal currents.

Waves are generally irregular, multi-directional and can be described by the

superposition of infinite numbers of sinusoidal wave components. The wave
components can be obtained by using the Fast Fourier Transform (FFT) to the measured

159
Sediment Transport Technology

wave data. In standard engineering applications, the Linear Wave Theory (LWT) is
utilized and nonlinear wave relations are disregarded in the wave records.

The potential function can be defined to describe the sea waves by assuming that water
is incompressible and flow is irrotational, satisfying the Laplace equation:

∇ 2φ = 0

Solving for the velocity vector:

ur
V = ∇φ

The sea bottom boundary condition can be described by the vertical velocity gradient
equal to zero, if the water depth (h) is uniform and the bottom is impermeable:

∂φ
=0 at z = − h.
∂z

The dynamic and kinematic boundary conditions at the free water surface are described
by the following equations, respectively, if the surface tension is neglected and the
atmospheric pressure is taken as constant:

∂φ 1 2
+ gη + ∇φ = C (t ) at z = η ,
∂t 2
∂φ ∂η
− − ∇ hφ ∇ hη = 0 at z = η ,
∂z ∂t

where;
η (x,y,t) : surface elevation profile,
C (t ) : Bernoulli constant,
g : gravitational acceleration,
∇ and ∇ h : gradient and horizontal gradient, respectively.

Free-surface boundary conditions can be linearized by the expansion of terms at the still
water level ( z = 0 ) assuming diminutive wave steepness, as follows:

∂φ
+ gη = C (t ) at z = 0
∂t
∂φ ∂η
− =0 at z = 0
∂z ∂t

Then the linear solution for the potential function of periodic waves is derived by using
the method of separation of variables for the Laplace equation:

ag cosh[k ( z + h )]
φ= sin θ
σ cosh kx

160
 Water Wave Mechanics

η = a cos θ
C (t ) = 0

where

θ = k (cos χx + sin χy ) − σt + β
r
σ = −∂θ / ∂t is the angular wave frequency, k = ∇ hθ is the wave number vector, χ is
the direction of the wave orthogonal with respect to the x-axis, g is the gravitational
acceleration and β is the phase difference. The wave number and frequency are related
by the following dispersion relation:

σ 2 = gk tanh kh.

2. Non-Linear Waves
Weak non-linear effects can be described mathematically using perturbation series. If
the waves are considered as small (ka << 1) but finite, the wave properties can be
expanded in a power series of ka. For weakly nonlinear waves (ε << ak ) , the potential
function and the surface elevation (η) can be written in the order of ε, as follows:

Φ = Φ (1) + εΦ (2 ) + ε 2Φ (3 ) + L + ε j −1Φ ( j ) + ⋅ ⋅ ⋅
(1) (2 ) (3 )
η = η + εη + ε 2η + Lε j −1Φ ( j ) + L

The third order accuracy solutions in wave steepness can be written as follows:

1 cosh[3k ( z + h)] 2 2
φ (3) = (α 2 − 1)(α 2 + 3)(9α 2 − 13) a k A sin 3θ
64 cosh 3kh
3 3
η (3) = − (α 4 − 3α 2 + 3)a 3 k 2 cosθ + (8α 6 + (α 2 − 1)2 )a 3 k 2 cos 3θ
8 64

⎧ ⎡9 2 ⎤⎫
( )
2
σ 2 = gk tanh( kh) ⎨1 + k 2a 2 ⎢ α − 1 +α 2 ⎥⎬
⎩ ⎣8 ⎦⎭

In the nonlinear solutions, waves have steeper crests and flatter troughs, and the phase
velocity increases with the increase in wave steepness. The maximum amplitude of the
Stokes wave is amax = 0.07L (k·a = 0.44). Water particles in a Stokes wave move along
paths that are nearly circular, but the paths fail to close, and the particles move slowly in
the direction of wave propagation. This is a mass transport, and the phenomenon is
called Stokes drift. But the forward transport near the surface is balanced by an equal
transport in the opposite direction at depth, and there is no net mass flux.

If the perturbation parameter is chosen equal to the ratio of wave amplitude to water
depth then analysis results in the Korteweg de Vries (KdV) equation which includes

161
Sediment Transport Technology

first order non-linear corrections and the Fully Extended Korteweg de Vries (FEKdV)
equation which includes first and second order corrections. The choice of perturbation
parameter can also results in the Extended Korteweg de Vries (EKdV) equation which
also includes first and second order non-linear corrections but which has more general
applicability to large amplitude waves. These equations can be solved in closed form
for special cases and can be integrated for general cases, but require modification to
deal with interacting waves in three dimensions and variable bathymetry.

The EKdV equation of the surface displacement η(x, t):

∂η ∂η ∂ 3η
+ (c + α 0η − α1η 2 ) +β 3 =0
∂t ∂x ∂x

where c, α, α and β are constants, and for α = 0 this reduces to the KdV equation. This
equation has a solitary wave solution of amplitude (a):

sec h 2 [κ ( x − Cmt )]
η ( x, t ) = a
1 − µ tanh 2 [κ ( x − Cmt )]

where κ, µ and Cm are constants. Solitary waves propagate without change of shape, and
two solitons can cross without interaction. The properties of a solitary waves result from
a balance between dispersion which tends to spread the solitary wave into a train of
waves, and non-linear effects which tend to shorten the wave.

3. Shallow Water Waves

The fluid flow in two-dimensional shallow water system can be written by the following
horizontal momentum and the vertically integrated continuity equations, respectively:

∂u ∂u ∂u ∂ ( h + hs )
+u +v = −g
∂t ∂x ∂y ∂x
∂v ∂v ∂v ∂ ( h + hs )
+u +v = −g
∂t ∂x ∂y ∂y
∂h ∂h ∂h ⎛ ∂u ∂v ⎞
+u +v +h ⎜⎜ + ⎟⎟ = 0
∂t ∂x ∂y ⎝ ∂x ∂y ⎠

Substituting the perturbations, the following equations can be obtained:

∂u ' ∂u ' ∂u ' ∂h'

+ (U + u ' ) + (V + v' ) +g =0
∂t ∂x ∂y ∂x
∂v' ∂v' ∂v' ∂h'
+ (U + u ' ) + (V + v' ) +g =0
∂t ∂x ∂y ∂y
∂h' ∂h' ∂h' ⎛ ∂u ' ∂v' ⎞ ∂h ∂h
+ (U + u ' ) + (V + v' ) + ( H + h'− hs ) ⎜⎜ + ⎟⎟ = (U + u ' ) s + (V + v' ) s
∂t ∂x ∂y ⎝ ∂x ∂y ⎠ ∂x ∂y

162
 Water Wave Mechanics

where, H is the undisturbed depth and h' is the vertical displacement from H. For the
linear perturbations in a single layer fluid with a flat bottom:

∂u' ∂u' ∂h'

+U +g =0
∂t ∂x ∂x
∂v ' ∂v ' ∂h'
+U +g =0
∂t ∂x ∂y
∂h' ∂h' ⎛ ∂u' ∂v ' ⎞
+U +H ⎜⎜ + ⎟⎟ = 0.
∂t ∂x ⎝ ∂x ∂y ⎠

These equations can be combined into the wave equation for the free surface
displacement perturbation h’
2
⎛∂ ∂ ⎞ ⎛ ∂ 2h' ∂ 2h' ⎞
⎜ + U ⎟ h' − ( gH ) ⎜⎜ 2 + 2 ⎟⎟ = 0
⎝ ∂t ∂x ⎠ ⎝ ∂x ∂y ⎠

Then, the dispersion relationship for two-dimensional shallow water waves can be
obtained as:

c = U ± gH

References

1. Garrett C., and Munk W., “Internal waves in the ocean”, Ann. Rev. Fluid Mech.,
11: pp. 39-369, (1979).
2. Michallet H., and Barthelemy E., “Experimental study of interfacial solitary waves”,
J. Fluid Mech., 366: pp.159-177, (1998).
3. Grue J., Jensen A., Rusas P., and Sveen J. K., “Properties of large amplitude internal
waves”, J. Fluid Mech., 380, pp. 257-278, (1999).

163
Sediment Transport Technology

Glossary

Beach: A strip of sediment (usually sand or gravel) that extends from the low-water line
inland to a cliff or zone of permanent vegetation.

Foreshore: The distance extending from the lowest tide to the average high-tide level.

Berm: A wave-deposited sediment platform that is flat or slopes slightly landward.

Backshore: A zone extending inland from the berm to the farthest point reached by the
surf.

Wave length: Horizontal distance between two wave crests.

164
 Water Wave Mechanics

Wave height (H): Vertical distance between crest and trough.

Wave amplitude (a): Vertical distance between a wave crest or trough, and the mean
level.

Wave period (T): Interval of time for two wave crests to pass the same position in space.

Wave frequency (f): Number of waves passing a given point in one second (1/T).

Wave steepness: Ratio of wave height to wavelength

Phase velocity (c): Speed at which a particular phase of the wave propagates, for
example, the speed of propagation of the wave crest. In deep water, the phase speed
depends on wave frequency, hence deep-water waves are dispersive. In shallow water,
the phase speed is independent of the wave; it depends on the depth of the water.
Therefore, shallow-water waves are non-dispersive.

Group velocity (cg): Velocity at which a group of waves travels across the sea, i.e. the
propagation velocity of wave energy:

1 g ⎛ 2kH ⎞
cg = tanh (kH ) ⎜⎜1 + ⎟⎟
2 k ⎝ sinh (2kH ) ⎠

Shoaling: The height of a wave varies with the depth of water in which the wave is
traveling.

Wave breaking: As waves become steeper they can break losing energy in turbulent
water motion. A breaker is a wave that has become so steep that the crest of the wave
topples forward, moving faster than the main body of the wave.

Types of breakers

• Spilling breakers: Gentle beach slope allows waves to disperse energy gradually
• Plunging breakers: Moderately steep beach slope gives waves a curling shape that
propels surfers
• Surging breakers: Abrupt beach slope makes waves build up and break rapidly at
the shore

Refraction: As the waves propagate into shallow waters near to the coast, the wave
fronts are bent so that they become more parallel to the depth contours and shoreline.
Refraction takes place at variable bottom depth that controls the phase speed of waves
The phenomenon of the refraction of sea waves is similar to the optical refraction of
light. Curvilinear bottom contours act as positive or negative lenses for the light,
focusing or defocusing the wave rays.

165
Sediment Transport Technology

Diffraction: Waves are bending around and behind barriers. It acts as secondary wave
point source from where the waves propagate in circular wave patterns with smaller
amplitude.

166
 MECHANISM OF SEDIMENT TRANSPORT

Assoc. Prof. Dr. Şevket ÇOKGÖR

Department of Hydraulics
College of Civil Engineering
Istanbul Technical University
Istanbul - Turkey

1. Introduction
The phenomenon of sediment transport is quite complicated, not amenable to a rigorous
theoretical treatment at present. A multitude of formulas and methods have been
proposed for sediment transport calculations. On the other hand, a sound understanding
of the basic mechanism of sediment transport and knowledge of available methods for
computing its magnitude are necessary for almost all the work related to river
engineering.

In this chapter origin of the sediment particles and basic concepts of sediment transport
are introduced. The aim of the chapter is to provide the student with an overall view
rather than to discuss the various methods in detail. Sediment particles are first
classified by origin. Then a distinction is made between bed load and suspended load.
Mechanisms of both modes of motion are described. Some formulas proposed to predict
the transport of bed material are discussed in the next chapter.

2. Origin of Sediment Particles

Sediment (solid particles) that we encounter in rivers are eroded by water from the river
basin and transported to the stream (sheet erosion) or eroded by the streamflow from the
bed of the stream (bed erosion). Sediment grains are transported by the stream until they
are finally deposited.

Sediment particles in motion in a stream are classified as follows according to their

origin:
1. Wash load: particles that have been eroded from the drainage basin of the stream,
2. Bed material load: particles that have been eroded from stream's own bed.

Particles of the wash load class are found in a sample taken from the river bed in an
insignificant amount. An empirical method, which is quite simple and is often used in
engineering practice, is to employ D5 (or D10 in some cases) as the criterion (D5/D10
assumed to be equal to 0.06 mm which is the limit between sand and silt). Sediment
finer than D5, is wash load and coarser than D5 is bed material load, in which D5 is the
grain size for which 5% by weight of the bed material is finer. Strictly speaking, bed
material at a surface can not be sampled by the samplers in current use. Therefore, the
bed material referred to here denotes the material within a certain depth of the bed.
Another practical approach is to overlap the accumulative size distribution curves of

167
Sediment Transport Technology

bed sediment and suspended load, as shown in Fig. 1. The straight line AB is tangent to
and connects the two curves. The intercept of this line with the coordinate axis is at
point O. the diameter corresponding to point O is the critical diameter that distinguishes
bed material load from wash load.

Hydraulically, there is an important difference between wash load and bed material
load. Rate of transport of bed material is a function of the flow characteristics in the
stream, whereas rate of transport of wash load depends only on the hydrological
characteristics of the drainage basin (such as size, soil properties, topography, intensity
and duration of precipitation, vegetation, winds, surface flow, etc.) but not on the
hydraulic conditions of flow. The reason is that for very fine grains that constitute the
wash load, the transporting capacity of the flow is always much larger than their
availability. Hence wash load is transported by the stream as suspended load with
practically no exchange with the bed. Therefore, particles that constitute the wash load
are not found in a sample taken from the bed surface, and wash load transport has no
effect on the river morphology. Conversely, bed material transport is directly related to
the hydraulics of flow, and has an important effect on bed aggradations and degradation.

In most streams greater part of the total load consists of wash load (80-95% in many
streams especially arid or semi arid zones), we shall, however, study only the bed
material transport. The reason for this is that wash load has no effect on the change of
the stream, bed as it can always be transported by the stream (by definition) with no
exchange with the bed. Its transport rate has no relation with the flow hydraulics and
can be estimated only by empirical methods which based on the drainage properties or
by direct measurement. Prediction of the transport rate of wash load is important mainly
in the design of capacity of reservoirs where it will finally be settled.

Although the rate of transport of wash load has no relation to flow characteristics, it is
observed in most rivers that its concentration increases with the flow rate, just as that of
bed material load. The explanation is that high amounts of precipitation results in high
sediment volumes to be eroded from the drainage basin, as well as in high rates of
flow. This should not be confused with the relation of bed material transport to flow
characteristics, which is based on hydromechanical grounds. Another important issue is

Bed sediment
Percentage
finner than
D(%)

Suspended
load

Diameter D (mm)

Figure 1. Critical diameter for distinguishing bed material load and wash load

168
 Mechanism of Sediment Transport

increasing of wash load in the stream by basin usage. Drainage basins almost allover the
worlds are getting urbanize also texture of plantations changing from forest to
agricultural plants. This is caused to increasing finer particles (wash load from drainage
basin) in the streams; therefore, fluctuating components of the velocities (turbulence
intensities) are suppressed. This is a serious problem for water quality because reducing
mass transfer capacity in the stream (such as dissolved oxygen transfer, DO, from air
trough the water).

One can wonder how sediment grains that are being transported as bed material can
have a constant exchange with the bed when the river bed is in equilibrium. In
equilibrium conditions the amount of incoming sediment equals the amount of sediment
leaving a reach of the river within a certain time period. In this process, a number of
grains in motion as bed material are deposited on the bed, and they are replaced by an
equal number of grains that are eroded by the flow from the bed. The aim here to
determine a sediment transport equation that can be used to predict the rate of bed
material transport as a function of flow conditions in the river, as mentioned above, the
amount of bed material load may be small as compared with the wash load; however, it
is important because it shapes the bed, determines the stability of the channel, and the
form of the bed surface.

In the following table (Table 1), main features of bed material and wash load are
summarized by different perspectives.

Table 1. Main features of bed material and wash load

Features Bed material load Wash load

Origin Soil erosion in watershed
Direct origin River bed upstream Sediment yield in watershed
Composition of Main part of the bed material usually not On the bed surface changing with
bed material changing except for heavily sediment- incoming amount of flow intensity
laden rivers
Composition of A small part of the moving sediment Main part of the moving sediment
moving sediment
Patterns of Bed load and suspended load Mainly suspended load
movement
Transport rate Determined from the flow intensity and Determined mainly from the
less correlated with incoming sediment, incoming sediment
but for heavily sediment laden rivers
depending also on the incoming sediment
Relationship Relationship established on basis of Relations based on watershed
between flow and mechanics and may be estimated by characteristics determined by
sediment sediment transport capacity formulas measured or empirical data
transport
Basic laws of For alluvial rivers the laws of movement for bed material load and wash load are
movement identical
Significance Transport rate determines bed stability Transport rate determines rate of
reservoir deposition
Criterion Coarser than D5 (or D10) of the bed Finer than D5 (or D10) of the bed
material material

169
Sediment Transport Technology

3. Mechanism and Modes of Sediment Transport

The force that is exerted by the flow on a sediment grain on the river bed Increases with
the increase of flow velocity (or bed shear stress). Some of the grains are initiated into
motion when critical flow conditions are reached.

When the critical conditions are exceeded that offered historically from Shields (1936)
some of the particles on the bed will start to move by sliding and rolling (Fig. 2b). This
is called bed load motion. Particles moving as bed load are supported directly by the
bed as they are always in contact with the bed. Their time-average velocity is much
lover than the velocity of fluid particles near the bed since sediment grains move in
steps followed by periods of rest (no motion).

At somewhat higher velocities, some of the grains on the bed make small jumps and
then return to bed. Jumps are started either due to the lift force exerted by the flow, or
due to collision of another grain. Once the grain is detached from the bed, gravity and
fluid resistance cause it to return (Fig. 2c). This mode of motion is called saltation (Fig.
2c). It is not an important mode of transport in rivers since water exerts a large
resistance on a grain due to its density (as compared to sediment transport in air). In
rivers height of vertical rise of saltating grains is only a few times the size of the grain,
whereas it can be 1000 times larger in air flow. Therefore sediment transport by
saltation can be neglected in rivers unlike that in air.

Figure 2. Patterns of sediment motions

170
 Mechanism of Sediment Transport

Flows at high velocity are turbulent and have eddies of various sizes. If a particle
jumping from the bed enters such an eddy, it may be carried far away from the bed. For
carrying a particle, the size of the eddy must be much larger than the particle and its
upward velocity component must be higher than the fall velocity of the particle. If an
eddy is of about the same size as a particle, the latter is liable to fall out of the eddy;
hence, the eddy would no longer affect the movement of the particle. On the contrary, lf
an eddy is much greater than a particle; the eddy may carry the particle for a long lime.
And by the time the particle falls out of the eddy, it may already have been carried into
the region of the main flow. Obviously, the transport of suspended particles is mainly
the effect of large-scale eddies.

In the main flow zone, even if a particle falls out of an eddy, it may enter another eddy
and be carried further along. Therefore, the trajectory of a suspended particle is quite
irregular and depends almost completely on the movement of the eddies
surrounding it. If a particle is carried by an eddy close to the bed, it may fall on to the
bed. These particles, carried by eddies and moving downstream at the same velocity as
the flow, are called the suspended load, as shown in Fig. 2d. Suspension of particles
takes a certain amount of energy from the turbulent flow. Hence, on the one hand,
flow turbulence carries sediment particles into suspension, and on the other hand, the
existence of suspended load reduces the turbulence intensity.

As it is very difficult to separate the three modes of motion, saltation is usually included
in'the bed load, as a matter of fact; a certain grain may part of the time move as bed load
and part of the time as suspended load. Thus, there exists an active interchange between
suspension and bed load, and it is difficult to draw a definite limit between the two. In
general only finer grains are brought into suspension. The grain size that separates
suspension and bed load varies with flow conditions. An approximate criterion to
determine the grain size that can be moved in suspension is supplied by the formula
given by Kresser (1964):

V2
= 360 (1)
gd

where V is the flow velocity, g acceleration of gravity. This formula predicts that grains
of size D = 0.3 mm or finer can be moved in suspension when V = 1 m/s whereas the
limiting size is D = 4.5 mm for V = 4 m/s.

Another criterion that is based on the bed shear velocity, U*, is in the form of

W = C ⋅U * (2)

so that particles with settling velocity W less than that given by the formula will be
brought into suspension. C in Eq. 2 varies in the range 0.8-2.

It is important to predict the rate of sediment transport, which is usually given as the
weight of sediment that passes through a cross-section of the stream in unit time. The
rate of transport can be defined for bed load, suspended load, or total load. These should
be determined as functions of flow conditions (discharge) in the river. Unfortunately, no

171
Sediment Transport Technology

exact formula is available even for the simplest case of steady, uniform, two-
dimensional flow over a bed of uniform grains.

It is necessary to know the sediment transport capacity of the river in order to solve
correctly engineering problems related to river engineering. There exists no complete
theory to determine the rate of sediment transport as function of the properties of the
sediment and flow because complexity of the river cross sections, sediment distributions
variation of slopes etc. Therefore we have to use empirical formulas, some of which are
based on a theory for the mechanism of transport, tested in a statistical manner, as the
mechanism of the sediment transport process is not known in enough detail, it is usually
necessary to carry out measurements in nature to find out the rate of sediment transport.

In spite of the large number of formulas to predict the rate of sediment transport
available in the literature, it can only be estimated with low accuracy. Most of the
formulas are of experimental character. Even those which have a certain theoretical
basis use laboratory and/or field measurements to estimate their coefficients. Another
complication arises from the fact that sediment transport measurements in the field are
usually not reliable.

It should be emphasized that actual sediment transport rate in a stream is not always
equal to the transport capacity; we know that this is always the case for wash load. In
the case of bed material load, it may be smaller than the capacity when an armor coat
(pavement layer) exists on the bed. An armor coat is a gravel layer consisting of the
coarser part of the bed material which prevents the exchange of the bed load with the
material. This confusion can be eliminated if bed material is always defined as the
material in the uppermost layer of the bed that is in contact with the flow.

4. Bed Load
Bed load motion is a very complex phenomenon as it is concerned with a two-phase
(solid + liquid) flow near the river bed. For this reason several formulas have been
proposed to calculate the bed load transport rate as function of flow and sediment
characteristics. They lead to different semi-empirical relationships between the non-
dimensional variables of the phenomenon, as somewhat different mechanisms are
assumed in each case. All of these formulas are derived for steady uniform conditions.

It is difficult to compare the various formulas and determine which is best since the
constants in them have to be determined experimentally. Another difficulty arises from
the fact that bed load transport in rivers cannot be measured accurately as measuring
devices always disturb the bed load motion. For this reason formulas are mostly based
on the results of laboratory experiments. It is not clear to what extent they can be
applied in field conditions. Each formula should be applied only to conditions similar to
those used in its derivation.

Derivation of a formula for bed load transport starts with dimensional analysis. Rate of
bed load transport qB is expected to be a function of the following variables;

q B = f ( µ , ρ , D, ρ s , g , τ 0 , h) (3)

172
 Mechanism of Sediment Transport

where, µ; viscosity, and ρ; density, are fluid properties, D; size, and ρs; grain density,
are sediment properties, τ 0 = ρU *2 ; bed shear stress, and h; depth, are flow properties.
In order to reduce the number of variables π-theorem of dimensional analysis is applied
to form non-dimensional variables;

⎛ τ0 ρU * D ρ s D ⎞
φ = f ⎜⎜ , , , ⎟⎟ (4)
⎝ (γ − γ s ) D µ ρ h⎠

where φ is non-dimensional bed load transport parameter, which may have several
forms, such as:

qB q qB
φ= = B3 = (5)
γ sU * D ρU * γ s (gD 3 )1 / 2

τ 0 / (γ s − γ )D is the most important single non-dimensional variable affecting the bed

load transport. It is sometimes called grain Froude number, mobility number of Shields
number. It represents the mobility of a grain as it is the ratio of the moving force to
resisting force for a particle on the bed. Therefore mobility increases with the Increase
of the value of this variable. It appears in all bed load formulas.

Other variables are of much lower significance, ρU * D / µ is the grain Reynolds number
representing the viscous effects on the motion of a grain, and its effect is negligible
when its value is sufficiently high. D/h is the relative roughness, which is not important
for the bed load transport since it takes place near the bed and is not affected directly by
the flow depth. ρs/ρ is the relative density and is not significant (it is already included in
mobility number).

Most bed load formulas can be put into the form of a relation between two
nondimensional variables:

φ = f (Ψ ) (6)

where Ψ is related to τ 0 / (γ s − γ )D . The form of the functional relationship varies from

one formula to another.

As will be seen in the next chapter various formulas have been proposed, none of which
can be expected to give accurate results. If possible, the formula to be used should be
decided on the basis of measured rates, unfortunately it is not easy to measure the actual
bed load in a river accurately.

5. Suspended Load
Solid particles that leave the bed enter the region where turbulent fluctuations
(especially vertical component of these) are strong enough to support them. Particles in
suspension can move for a considerably long time without settling down since their
settling velocities are balanced by the upward vertical component of turbulent velocity

173
Sediment Transport Technology

fluctuations. Such particles follow the motion of the fluid with a relatively small lag.
Thus their velocities can be taken as approximately equal (rarely small) to that of the
fluid. A suspended particle will eventually settle on the bed and be replaced by a new
one entering into suspension.

In order to determine the rate of suspended load transport we must first know the
concentration distribution, c, usually expressed as a non-dimensional ratio, either in
volume or weight, corresponding to the percentage occupied by the solid particles. It is
expected that c will decrease with y, the distance above the bed due to gravity. Thus we
must search for the function c = f(y).

In general c will be a function of the following variables:

c = f (c a , a, y, h, U * , W ) (7)

Sediment properties such as the grain size, shape and density are represented by the fall
(settling) velocity W, which characterizes the tendency of a sediment grain to settle in
water due to gravity. Flow properties are represented by U* and h. The exchange
between the bed load and suspended load is characterized by the reference concentration
ca at y = a. Dimensional analysis leads to:

c ⎛y a W ⎞
= f ⎜⎜ , , ⎟⎟ (8)
ca ⎝ h h U* ⎠

In the following diffusion theory is used to obtain the form of the functional relationship
in the above equation.

Let us consider a horizontal plane of unit area inside the flow medium. In steady
conditions, upward transport rate of solids through the area must be equal to the
downward transport rate. Solid particles are settled down due to gravity, so that the
quantity of sediment passing downward through the unit area in unit time is W·c, where
W is the settling (fall) velocity and c is the time-average concentration in that region
(Fig. 3).

In order to write down an expression for the upward rate of transport, we consider the
diffusion theory. There exists a concentration gradient dc/dy along the vertical since c
will be decreased with the distance from the bed. Transport rate by turbulent diffusion is
proportional to the gradient dc/dy and can be written as -εs dc/dy, where εs is the mass
transfer coefficient (negative sign is due to the fact that dc/dy is a negative quantity).

-εs dc/dy
y c

W·c
Figure 3. Balance of suspended particles settling
due to gravity and rising by diffusion

174
 Mechanism of Sediment Transport

Equating the rates of transport in the upward and downward directions:

dc
W ⋅ c = −ε s (9)
dy

In order to proceed we must find an expression for εs in terms of flow variables. We can
assume that εs will be proportional to εm, coefficient of momentum transfer, since
diffusion of mass in analogous to diffusion of momentum. In turbulent flows shear
stresses produced by momentum transfer (Reynolds stresses) are expressed as:

du
τ t = ρε m (10)
dy

Prandtl's mixing length theory relates τt to the velocity gradient du/dy and mixing length
l in the following way;

2
⎛ du ⎞
τ t = ρl ⎜⎜ ⎟⎟
2
(11)
⎝ dy ⎠

Hence we get the following expression for εm:

2
⎛ du ⎞
ε m = l ⎜⎜ ⎟⎟
2
(12)
⎝ dy ⎠

Assuming logarithmic velocity distribution:

du U *
= (13)
dy ky

where k is Karman's constant (0.40). Shear stress varies linearly in an open channel flow
of depth h:

⎛ y⎞ ⎛ y⎞
τ t = τ 0 ⎜1 − ⎟ = ρU *2 ⎜1 − ⎟ (14)
⎝ h⎠ ⎝ h⎠

Combining these expressions:

⎛ y⎞
ε m = U * ky⎜1 − ⎟ (15)
⎝ h⎠

Assuming that εs is proportional to εm:

⎛ y⎞
ε s = βε m = βU * ky⎜1 − ⎟ (16)
⎝ h⎠

175
Sediment Transport Technology

Introducing this into the equation for dc/dy:

⎛ y ⎞ dc
W ⋅ c + β U * ky⎜1 − ⎟ =0 (17)
⎝ h ⎠ dy

Integrating:

α
c ⎛h− y a ⎞
=⎜ ⎟ (18)
c a ⎜⎝ y h − a ⎟⎠

where, α = W/(βkU*), and ca is the reference concentration at y = a.

This expression (with β = 1) gives the distribution of suspended load concentration

relative to the reference concentration ca as a function of the non-dimensional parameter
α. It is not valid very near the bed, where turbulent fluctuations are not strong enough to
support the suspension (it gives c → ∞ at y = 0).

Comparing concentration distributions for small and large values of α, it is seen that the
distribution is more uniform when α is small (fine grains). For α < 0.01 the distribution
can be assumed to be practically uniform, whereas for larger values of a (coarse grains)
concentration approaches to zero rapidly with the distance from the bed (Fig. 4).

Figure4. Distribution of suspended load concentration along the vertical for

various values of a

176
 Mechanism of Sediment Transport

Measurements show good agreement with the above equation. There are discrepancies
at very high concentrations and also for very coarse particles. Some authors proposed to
use a α value that is based on measurements rather than the expression α = W/(kU*) in
order to achieve a better agreement between theoretical and observed concentrations.
Disagreement between measured and theoretical values is attributed to β ≠ 1 or change
of k due to effect of suspended particles on turbulence. Recent work shows that β = 1
for W/(kU*) < 0.5, whereas β < 1 for W/(kU*) > 0.5. β values are 0.7, 0,5. and 0.35 for
W/(kU*) = l, 2 and 3, respectively, other authors have proposed somewhat different
expressions.

In rivers concentration distribution is usually not uniform in the lateral direction in a

cross-section due to secondary currants. It must also be remembered that the above
expression is for the time-average value of c, and fluctuations as large as 20% of this
value may exist due to turbulence.

References

1. Kresser, W., “Gedanken zur Geschiebe und SchwebstoffClhrung der Gewasser”,

Oesterreichische Wasserwirtschaft, No.1-2, (1964).
2. Raudkivi, A. J., “Loose Boundary Hydraulics”, Pergamon Press, (2nd ed.), (1976).
3. Graf, W. H., “Hydraulics of Sediment Transport”, Mc Graw Hill, (2nd ed. Water
Resources Publications, 1984), (1971).
4. Shen, H. W., “River Mechanics”, Water Resources Publications, 3 Vol., (1971-73).
5. Bogardi, J., “Sediment Transport in Alluvial Streams”, Akademia Kiado, (1974).
6. Yalin, M., “Mechanics of Sediment Transport”, Pergamon Press, (1974).
7. Garde, R. J., Ranga Raju, K.G., “Mechanics of Sediment Transportation and
Alluvial Stream Problems”, Wiley Eastern Ltd., (1977).
8. Simons, D.B., Senturk, F., “Sediment Transport Technology”, Water Resources
Publications, (1977).
9. Jansen, P., et al., “Principles of River Engineering”, Pitman, (1979).
10. Stelczer, K., “Bed-Load Transport, Theory and Practice”, Water Resources
Publications, (1981).
11. Fredsøe, J., Deigaard, R., “Mechanics of Coastal Sediment Transport”, Advanced
Series on Ocean Engineering, Volume 3, Singapore, (1992-93-94).
12. Chien, N., Wan, Z., “Mechanics of Sediment Transport”, ASCE Press, Virginia,
(1999).
13. Bayazit, M., “Mechanism of Sediment Transport”, Post Graduate Course in
Sediment Technology, Proceedings, Ankara, Turkey, (2000).
14. Parker, G., “1D Sediment Transport Morphodynamics with Applications to Rivers
And Turbidity Currents”, e-Book, http://www.ce.umn.edu/~parker, University of
Minnesota, USA, (2004).

177
178
 SEDIMENT TRANSPORT RATE

Assoc. Prof. Dr. Şevket ÇOKGÖR

Department of Hydraulics
College of Civil Engineering
Istanbul Technical University
Istanbul - Turkey

1. Introduction
Mechanism of the sediment transport is discussed in previous chapter. From the point of
view of engineering practice, the sediment transport rate which gives the amount of bed
material transported by the flow is the most significant characteristic of the two-phase
motion. As a consequence of this practical significance, a vast amount of transport
formulae have been proposed by various researchers since Du Boys (1879). Some of
these relations will be discussed in this chapter.

For a particular ratio of densities of the sediment and the fluid, the modes of transport
are now generally believed to depend on the average shear stress on the bed. For
relatively shear stresses, the material is transported on or near the bed and it is called as
bed load. With further increase in shear, a part of the material is transported in a state of
suspension and it is called as suspended load. The total sediment transport rate is the
sum of bed load and suspended load.

In this chapter, firstly bedload formulae will discuss then suspended sediment load
calculation methods evaluate. Last part of the chapter macroscopic methods which are
given bedload and suspended load together will introduce.

2. Bedload Transport
qb denote the volume bedload transport rate per unit width (sliding, rolling, saltating). It
is reasonable to assume that qb increases with a measure of flow strength, such as depth-
averaged flow velocity, U or boundary shear stress τb. A dimensionless Einstein
bedload number qb* can be defined as follows:

qb
qb∗ = (1)
RgD D

A common and useful approach to the quantification of bedload transport is to

empirically relate qb* with either the Shields stress τ* or the excess of the Shields stress
τ* above some appropriately defined critical Shields stress τc*. As pointed out in the
mentioned chapter (initiation of motion), τc* can be defined appropriately so as to a) fit
the data and b) provide a useful demarcation of a range below which the bedload
transport rate is too low to be of interest. The functional relation sought is thus of the
form:

179
Sediment Transport Technology

qb∗ = qb∗ (τ ∗ ) or qb∗ = qb∗ (τ ∗ − τ c∗ ) (2)

2.1. Bedload transport relation of Meyer-Peter and Müller

All the bedload relations in this chapter pertain to a flow condition known as plane-bed
transport, i.e. transport in the absence of significant bedforms. The mother of all modern
bedload transport relations is that due to Meyer-Peter and Müller (1948) (MPM). It
takes the form,

qb∗ = 8(τ ∗ − τ c∗ ) 3 / 2 , τ c∗ = 0.047 (3)

The relation was derived using flume data pertaining to well-sorted sediment in the
gravel sizes. Recently Wong (2003) and Wong and Parker (2004) found an error in the
analysis of MPM. A re-analysis of the all the data pertaining to plane-bed transport used
by MPM resulted in the corrected relation:

qb∗ = 8(τ ∗ − τ c∗ ) 3 / 2 , τ c∗ = 0.047 (4)

If the exponent of 1.5 is retained, the best-fit relation is

qb∗ = 3.97 (τ ∗ − τ c∗ ) 3 / 2 , τ c∗ = 0.0495 (5)

2.1.1. Limitations of Meyer-Peter Müller

The critical Shields stress τc* of either 0.047 or 0.0495 in either the original or
corrected MPM relation(s) must be considered as only a matter of convenience for
correlating the data. This can be demonstrated as follows. Consider bankfull flow in a
river. The bed shear stress at bankfull flow τbbf can be estimated from the depth-slope
product rule of normal flow:

τ bbf = ρgH bf S (6)

Figure 1. Bedload relation in modified MPM

180
 Sediment Transport Rate

The corresponding Shields stress τbf50* at bankfull flow is then estimated as (Fig. 2):

H bf S
τ bf∗ 50 = (7)
RDs 50

where Ds50 denotes a surface median size. For the gravel-bed rivers, however, the
average value of τbf50* was found to be about 0.05. According to MPM, then, these
rivers can barely move sediment of the surface median size Ds50 at bankfull flow. Yet
most such streams do move this size at bankfull flow, and often in significant quantities.
There is nothing intrinsically wrong with MPM. In a dimensionless sense, however, the
flume data used to define it correspond to the very high end of the transport events that
normally occur during floods in alluvial gravel-bed streams. While the relation is
important in a historical sense, it is not the best relation to use with gravel-bed streams.

2.2. Some of bedload transport relations for uniform sediment

Some commonly-quoted bedload transport relations with good data bases are given
below.

Einstein (1950);

1 + ( 0.143 / τ ∗ ) − 2 43.5qb∗
∫
2
1− e −t dt = (8)
π −( 0.143 / τ ∗ ) − 2 1 + 43.5qb∗

Ashida & Michiue (1972);

(
qb∗ = 17(τ ∗ − τ c∗ ) τ ∗ − τ c∗ , τ c∗ = 0.05 ) (9)

Engelund & Fredsoe (1976);

(
qb∗ = 18.74(τ ∗ − τ c∗ ) τ ∗ − 0.7 τ c∗ , τ c∗ = 0.05 ) (10)

1.E+01

1.E+00

τbf∗ 50 1.E-01
Grav Brit
Grav Alta
Sand Mult
1.E-02
Sand Sing
Grav Ida
1.E-03
1.E+02 1.E+04 1.E+06 1.E+08 1.E+10 1.E+12 1.E+14

Q̂
Figure 2. Variation of τbf50* by bed load discharge

181
Sediment Transport Technology

Fernandez Luque & van Beek (1976);

(
qb∗ = 5.7 τ ∗ − τ c∗ )1.5
, τ c∗ = 0.037 ~ 0.0455 (11)

Parker (1979) fit to Einstein (1950);

4.5
⎛ τ c∗ ⎞
q = 11.2 τ
∗
( ) ∗ 1.5
⎜⎜1 − ∗ ⎟⎟ , τ c∗ = 0.03 (12)
⎝ τs ⎠
b

2.2.1. Some notes on the bedload transport relations

The bedload relation of Einstein (1950) contains no critical Shields number. This
reflects this probabilistic philosophy. All of the relations except that of Einstein
correspond to a relation of the form

q ∗b ~ (τ ∗ ) 3 / 2 (13)

in the limit of high Shields number. In dimensioned form this becomes

3/ 2
qb ⎛ u∗2 ⎞ Rgq b
= K ⎜⎜ ⎟⎟ or =K (14)
RgD D ⎝ RgD ⎠ u∗3

where K is a constant; for example in the case of Ashida-Michiue (1972), K = 17. Note
that in this limit the bedload transport rate becomes independent of grain size. Some of
the scatter between the relations is due to the face that τc* should be a function of Rep.
This is reflected in the discussion of the Fernandez Luque-van Beek relation. (Recall
that Re p = RgD D / ν ). Some of the scatter is also due to the fact that several of the
relations have been plotted well outside of the data used to derive them. For example, in
data used to derive Fernandez Luque-van Beek, τ* never exceeded 0.11, whereas the
plot extends to τ* = 1.

1.E+02
1.E+01
1.E+00
1.E-01
1.E-02 E
AM
1.E-03 EF
qb *

1.E-04 FLBSand
P approx E
1.E-05
FLBGrav
1.E-06
E = Einstein
1.E-07 AM = Ashida-Michiue
1.E-08 EF = Engelund-Fredsoe
P approx E = Parker approx of Einstein
1.E-09
FLBSand = Fernandez Luque-van Beek, τc* = 0.038
0.01 0.1 1
FLBGrav = Fernandez Luque-van Beek, τc* = 0.0455
τ*

Figure 3. Plots of bedload transport relations

182
 Sediment Transport Rate

2.2.2. Notes on the relation of Fernandez Luque and van Beek

In the experiments on which the relation is based;
a) Streamwise bed slope angle α varied from near 0 to 22°.
b) The material tested included five grain sizes and three specific gravities, as
given below; also listed are the values of Rep and the critical Shields number
τco* determined empirically at near vanishing bed slope angle (Table 1).
c) It is thus possible to check the effect of Rep and α on the transport relation of
Fernandez Luque and van Beek.

2.2.3. Critical Shields number in the relation of Luque and van Beek
The experimental values of τco* generally track the modified Shields relation, but are
high by a factor ~ 2. This reflects the fact that they correspond to a condition of very
small transport determined in a consistent way (Fig. 4).

The ratio τc*/τco* decreases with streamwise angle α as predicted by the relation of
threshold condition (Fig. 5), but to obtain good agreement θr must be set to the rather
high value of 47° (µc = 1.07).

Table 1. Material effect on bedload parameters

Material D(mm) R Rep τco*

Walnut shells 1.5 0.34 106 0.038
Sand 1 0.9 1.64 108 0.038
Sand 2 1.8 1.64 306 0.037
Gravel 3.3 1.64 760 0.0455
Magnetite 1.8 3.58 453 0.042

0.05

0.045

0.04

0.035

0.03
Data of FLvB
τco*

0.025
Modified Shields Curve
0.02

0.015

0.01 ( −7.7 Rep−0.6 )

τ∗c = 0.5 [0.22 Rep−0.6 + 0.06 ⋅ 10 ]
0.005

0
10 100 1000
Rep

Figure 4. Variation of τco* by Rep

183
Sediment Transport Technology

1.1

0.9
Observed
Predicted
τc*/τco*

0.8

0.7
τ∗c tan α
= cos α(1 − )
0.6
τ∗co µc
0.5

0.4
0 5 10 15 20 25 30

Figure 5. τc*/τco* vs. streamwise angle α

2.3. Sheet flow

Sheet flow occurs in the river while extreme flow conditions such as flood. In this case,
bed material moves such as fluid particles on the bed. For values of τ* < a threshold
value τsheet*, bedload is localized in terms of rolling, sliding and saltating grains that
exchange only with the immediate bed surface. When τs* > τsheet* the bedload layer
devolves into a sliding layer of grains that can be several grains thick. Sheet flows occur
in unidirectional river flows as well as bidirectional flows in the surf zone. Values of
τsheet* have been variously estimated as 0.5~1.5. (Horikawa, 1988, Fredsoe and
Diegaard, 1994, Dohmen-Jannsen, 1999; Gao, 2003). The parameter τsheet* appears to
decrease with increasing Froude number. Wilson (1966) has estimated the bedload
transport rate in the sheet flow regime as obeying a relation of the form:

q ∗b = K (τ ∗ ) 3 / 2 , K = 12 (15)

All the previously presented bedload relations except that of Einstein also devolve to a
relation of the above form for large τ*, with K varying between 3.97 and 18.74.

2.4. Mechanistic derivations of bedload transport relations

A number of mechanistic derivations of bedload transport relations are available. These
are basically of two types, based on two forms for bedload continuity. Let ξbl; the
volume of sediment in bedload transport per unit area, ubl ; the mean velocity of bedload
particles, Ebl; the volume rate of entrainment of bed particles into bedload movement
(rolling, sliding or saltation, not suspension) and Lsbl; the average step length of a
bedload particle (from entrainment to deposition, usually including many saltations).
The following continuity relations then hold:

qb = ξ bl ubl , qb = Ebl Lsbl (16)

In a Bagnoldean approach, separate predictors are developed for ubl and ξbl, the latter
determined from the Bagnold constraint (Bagnold, 1956). Models of this type include

184
 Sediment Transport Rate

the macroscopic models of Ashida and Michiue (1972) and Engelund and Fredsoe
(1976), and the saltation models of Wiberg and Smith (1985, 1989), Sekine and
Kikkawa (1992), and Nino and Garcia (1994a,b). Recently, however, Seminara et al.
(2003) have shown that the Bagnold constraint is not generally correct.

In the Einstein’s approach the goal is to develop predictors for Ebl and Lsbl. It is the
former relation that is particularly difficult to achieve. Models of this type include the
Einstein (1950), Tsujimoto (e.g. 1991) and Parker et al. (2003).

2.5. Calculations with bedload transport relations

To perform calculations with any of the previous bedload transport relations, it is
necessary to specify:
1. the submerged specific gravity R (s-1)of the sediment;
2. a representative grain size exposed on the bed surface, e.g. surface geometric mean
size Dsg or surface median size Ds50, to be used as the characteristic size D in the
relation;
3. and a value for the shear velocity of the flow u* (and thus τb).

Once these parameters are specified, τ* = (u*)2/(RgD) is computed, qb* is calculated

from the bedload transport relation, and the volume bedload transport rate per unit
width is computed as qb = (RgD)1/2Dqb*.

The shear velocity u* is computed from the flow field using the techniques of Chapter
5. For example, in the case of normal flow satisfying the Manning-Strickler resistance
relation,

3 / 10 3 / 10
⎛ k 1/ 3 q 2 ⎞ ⎛ k 1/ 3 q 2 ⎞ S 7 /10
u = ⎜⎜ s 2 w ⎟⎟
2
∗ g 7 / 10
S 7 / 10
τ = ⎜⎜ s 2 w ⎟⎟
∗
(17)
⎝ αr ⎠ ⎝ αr g ⎠ RD

2.6. Alternative dimensionless bedload transport

Again, the case under consideration is plane-bed bedload transport (no bedforms). As a
preliminary, define a dimensionless sediment transport rate W* as

qb∗ Rgq b
W∗ ≡ = 3 (18)
(τ )
∗ 3/ 2
u∗

Now all previously presented bedload transport rates for uniform sediment can be
rewritten in terms of W* as a function of τ*:

a) Einstein (1950);

1 + ( 0.143 / τ ∗ ) − 2 43.5 (τ ∗ ) 3 / 2 W ∗
∫
2
1− e −t dt = (19)
π − ( 0.143 / τ ∗ ) − 2 1 + 43.5 (τ ∗ ) 3 / 2 W ∗

b) Ashida & Michiue (1972);

185
Sediment Transport Technology

⎛ τ ∗ ⎞⎛ τ∗ ⎞
W ∗ = 17⎜⎜1 − c∗ ⎟⎟ ⎜1 − c∗ ⎟ , τ c∗ = 0.05 (20)
⎝ τ ⎠ ⎜⎝ τ ⎟⎠

c) c.) Engelund & Fredsoe (1976);

⎛ τ ∗ ⎞⎛ τ∗ ⎞
W ∗ = 18.74⎜⎜1 − c∗ ⎟⎟ ⎜1 − 0.7 c∗ ⎟ , τ c∗ = 0.05 (21)
⎝ τ ⎠ ⎜⎝ τ ⎟⎠

d) Fernandez Luque & van Beek (1976);

1.5
⎛ τ∗ ⎞
W∗ = 5.7⎜⎜1 − c∗ ⎟⎟ , τ c∗ = 0.037 ~ 0.0455 (22)
⎝ τ ⎠

e) Parker (1979) fit to Einstein (1950);

4.5
⎛ τ c∗ ⎞
W = 11.2 ⎜⎜1 − ∗ ⎟⎟
∗
, τ c∗ = 0.03 (23)
⎝ τs ⎠

2.7. Surface-based bedload transport formulation for mixtures

Consider the bedload transport of a mixture of sizes. The thickness La of the active
(surface) layer of the bed with which bedload particles exchange is given by as

La = na Ds 90 (24)

where Ds90 is the size in the surface (active) layer such that 90 percent of the material is
finer, and na is an order-one dimensionless constant (in the range 1 ~ 2). Divide the bed
material into N grain size ranges, each with characteristic size Di, and let Fi denote the
fraction of material in the surface (active) layer in the ith size range. The volume
bedload transport rate per unit width of sediment in the ith grain size range is denoted as
qbi. The total volume bedload transport rate per unit width is denoted as qbT, and the
fraction of bedload in the ith grain size range is pi, where;
N
qbi
qbT = ∑ qbi , pi = (25)
i =1 qbT

Now in analogy to τ*, q* and W*, define the dimensionless grain size specific Shields
number τi*, grain size specific Einstein number qi* and dimensionless grain size
specific bedload transport rate Wi* as;

τb u∗2 qbi qbi∗ Rgq bi

τ i∗ ≡ = , q ∗bi = , Wi ∗ ≡ = (26)
ρRgD i RgD i RgD i D i Fi (τ i )
∗ 3/ 2
(u∗ ) 3 Fi

It is now assumed that a functional relation exists between qi* (Wi*) and τi*, so that;

186
 Sediment Transport Rate

qbi Rgq bi
qbi∗ = = f q (τ i∗ ) or Wi ∗ = 3
= fW (τ i∗ ) (27)
RgDi Di Fi (u∗ ) Fi

The bedload transport rate of sediment in the ith grain size range is thus given as

u∗3
qbi = Fi RgDi Di f q (τ ) or ∗
i qbi = Fi fW (τ i∗ ) (28)
Rg

According to this formulation, if the grain size range is not represented in the surface
(active) layer, it will not be represented in the bedload transport (Fig. 6).

2.7.1. Bedload relation for mixtures due to Ashida and Michiue (1972)
Mixture bedload transport for various fractional sizes is given;

(
qbi∗ = 17 τ i∗ − τ ci∗ )( τ ∗
i − τ ci∗ ) (29)

Basic transport relation;

⎧ ⎛ Di ⎞
−1

⎪ 0.843⎜ ⎟ for Di ≤ 0.4

⎪ ⎜ ⎟
⎝ Dsg ⎠ Dsg
⎪ 2
τ ci∗ ⎪⎡ ⎤
= ⎨⎢ ⎥ τ scg
∗
= 0.05 (30)
τ scg
∗
⎪⎢ log(19) ⎥ D
⎪⎢ ⎛ ⎥ for i > 0.4
⎪⎢ log⎜19 Di ⎞⎟ ⎥ Dsg
⎪⎢ ⎜ Dsg ⎟ ⎥
⎩⎣ ⎝ ⎠⎦

Effective critical Shields stress for surface geometric mean size is τ scg
∗
= 0.05 . Modified
version of Egiazaroff (1965) hiding function,

N
Dsg = 2ψ s , ψ s = ∑ψ i Fi (31)
i =1

qbi qbi La

η
∆x
z'

Figure 6. Schematized fractional bedload calculation

187
Sediment Transport Technology

This relation has been modified slightly from the original formulation. Here the relation
specifically uses surface fractions Fi, and surface geometric mean size Dsg is specified in
preference to the original arithmetic mean size Dsm = ΣDiFi.

2.7.2. Bedload relation for mixtures due to Parker (1990 a,b)

This relation is appropriate only for the computation of gravel bedload transport rates in
gravel-bed streams. In computing Wi*, Fi must be renormalized so that the sand is
removed, and the remaining gravel fractions sum to unity, ΣFi = 1. The method is based
on surface geometric size Dsg and surface arithmetic standard deviation σs on the ψ
scale, both computed from the renormalized fractions Fi.

Wi ∗ = 0.00218 G (φi )
−0.0951
⎛ Di ⎞ τ sg∗ u∗2
φi = ωφsgo ⎜⎜ ⎟ , φsgo = ∗ , τ sg∗ = , τ ssrg
∗
= 0.0386
D ⎟ τ R gD
⎝ sg ⎠ ssrg sg

⎧ ⎛ 0.853 ⎞
4.5

⎪ 5474⎜⎜1 − ⎟ for φ > 1.59

⎪ ⎝ φ ⎟⎠
⎪
[ ]
G (φ ) = ⎨exp 14.2(φ − 1) − 9.28(φ − 1) 2 for 1 ≤ φ ≤ 1.59 (32)
⎪ φ 14.2 for φ < 1
⎪
⎪⎩
σs
ω = 1+
σ O (φsgo )
[ωO (φsgo ) − 1]

In the above ωO and σO are set functions of φsgo specified in the following stage.

1.6

1.4

1.2

omegaO
ω O, σ O 0.8
sigmaO

0.6

0.4

0.2

0
0.1 1 10 100 1000

φ sgo

Figure 7. Variation of ωO and σO by φsgo

188
 Sediment Transport Rate

2.7.3. Bedload relation for mixtures due to Wilcock and Crowe (2003)
The sand is not excluded in the fractions Fi used to compute Wi*. The method is based
on the surface geometric mean size Dsg and fraction sand in the surface layer Fs.

Wi * = G (φi )
⎧ 0.002φ 7.5 for φ < 1.35
⎪ 4.5
G = ⎨ ⎛ 0.894 ⎞
⎪14⎜⎜1 − φ 0.5 ⎟⎟ for φ ≥ 1.35
⎩ ⎝ ⎠
−b
τ∗ ⎛ Di ⎞
φi = ∗sg ⎜ ⎟ (33)
τ ssrg ⎜D ⎟
⎝ sg ⎠
τ ssrg
∗
= 0.021 + 0.015 exp(−20 Fs )
0.67
b=
1 + exp(1.5 − Di / Dsg )
u∗2
τ sg∗ =
RgDsg

2.8. Effect of Sand Content in the Surface Layer on Gravel Mobility in a Gravel-
Bed Stream
Wilcock and Crowe (2003) have shown that increasing sand content in the bed surface
layer of a gravel-bed stream renders the surface gravel more mobile. This effect is
captured in their relationship between the reference Shields number for the surface
geometric mean size (a surrogate for a critical Shields number) and the fraction sand Fs
in the surface layer (Fig. 8):

τ ssrg
∗
= 0.021 + 0.015 exp(−20 Fs ) (34)

0.04

0.035

0.03

0.025

τ ∗
ssrg 0.02

0.015

0.01

0.005

0
0 0.05 0.1 0.15 0.2 0.25 0.3
Fs
∗
Figure 8. Variation of τ ssrg by Fi

189
Sediment Transport Technology

From Fig. 8, τ ssrg

∗
decreases as Fs increases. The surface layers of gravel-bed streams
rarely contain more than 30% sand; beyond this point the gravel tends to be buried
under pure sand.

2.9. Bedload relation for mixtures due to Powell, Reid and Laronne (2001)
The sand is excluded in the fractions Fi used to compute Wi*. The method is based on
the surface median size Ds50, computed after excluding sand.

4.5
∗ ⎛ 1⎞
Wi = 11.2⎜⎜1 − ⎟⎟
⎝ φ⎠
τ i∗
φ= ∗ (35)
τ sci
− 0.74
τ sci
∗
⎛ D ⎞
= ⎜⎜ i ⎟⎟ , τ sc∗ 50 = 0.03
τ sc 50 ⎝ Ds 50 ⎠
∗

2.10. Computations of annual bedload yield

It is necessary to have a flow duration curve to perform the calculation. The flow
duration curve specifies the fraction of time a given water discharge is exceeded, as a
function of water discharge. This curve is divided into M bins k = 1 to M, such that pQk
specifies the fraction of time the flow is in range k with characteristic discharge Qk. The
value u*,k must be computed for each range. For example, in the case of normal flow
with constant width B, the Manning-Strickler resistance relation,

3 / 10
⎛ k 1 / 3Q 2 ⎞
u2
∗, k = ⎜⎜ s 2 2k ⎟⎟ g 7 /10 S 7 /10 (36)
⎝ B αr ⎠

The grain size distribution of the bed material must be specified; ks can be computed as
2·Ds90 (for a plane bed). Once u*,k is computed, either qb,k (material approximated as
uniform) or qbi,k (mixtures) is computed for each flow range, and the annual sediment
yield qba or total yield qbTa and grain size fractions of the yield pai are given as:

M N
⎛M ⎞ ∑q bi , k pQk
qba = ∑ qb ,k pQk or qbTa = ∑ ⎜ ∑ qbi ,k pQk ⎟ , pai = k =1
(37)
i =1 ⎝ k =1 ⎠
N
⎛M
⎞
k =1
∑ ⎜ ∑ qbi ,k pQk ⎟
i =1 ⎝ k =1 ⎠

Expect the flood flows to contribute disproportionately to the annual sediment yield.

3. Suspended Load Transport

At low values of average shear stress on the bed of an alluvial channel, the material
moves as bed load. However with further increase in shear stress, some of the bed
particles are carried into the main flow. These particles travel with a velocity almost

190
 Sediment Transport Rate

equal to the flow velocity and they constitute the suspended load, so the suspended load
transport can be taken as an advanced stage of bed load transport.

Having an expression for the concentration distribution of suspended load (c), in the
steady flow conditions along the depth of the flow, rate of suspended load transport can
be computed as,

s
q s = γ s ∫ cudy (38)
a

The lower limit of integration (a) must be greater than zero since the expression of
concentration distribution is not valid very near the bed, where turbulent fluctuations
are not strong enough to support suspended grains. Einstein assumed a = 2D.

Considering the following expressions for concentration and logarithmic velocity

distributions, we can get the rate of suspended load transport by the integration of (Eq.
38).

α
c ⎛h− y a ⎞
=⎜ ⎟ (39)
ca ⎜⎝ y h − a ⎟⎠
U ⎛ χh ⎞
= 5.75 log⎜⎜ 30.2 ⎟⎟ (40)
U* ⎝ ks ⎠

and

⎡ ⎛ χh ⎞ ⎤
qs = 11.6γ s csU * a ⎢2.3 log⎜⎜ 30.2 ⎟⎟ I1 + I 2 ⎥ (41)
⎣ ⎝ ks ⎠ ⎦

where χ is correction factor introduced by Einstein and Barbarossa. The two integrals I1
and I2 have been solved numerically by Einstein and are shown in further in this section
as functions of A and α. A is equal to a/h and α is given by:

W
α= (42)
κU *

where, W; is the settling velocity, κ; is the Von Karman constant, U*; is the shear
velocity. The equivalent Nikuradse sand roughness ks. is taken as D65 by Einstein. We
must know the value of reference concentration (ca ) at some distance (a) above the
bed, to get qs from (Eq. 41). There is not yet any accurate method to determine ca.
Einstein assumed that ca is proportional to mean concentration of bed load in a zone of
thickness a = 2D near the bed (bed layer).

qb
ca = A1 (43)
2 DU s γ s

191
Sediment Transport Technology

Assuming that Us is equal to the velocity at the outer edge of the viscous sublayer
(Us=11.6 U*), ca is get as follows:

1 qb
ca = (44)
11.6 2 DU *γ s

In the case of sediment mixtures, one can determine the transport rates of different sizes
of the mixture and then sum them up to obtain the total rate of suspended load transport.
The method proposed by Einstein is as follows; for a particular size d in the mixture,
Eq. 41 can be written as:

is q s = ib q B (PI1 + I 2 ) (45)

since,

11.6γ s caU * a = ib qb (46)

where P shows,

⎛ χh ⎞
P = 2.30 log⎜⎜ 30.2 ⎟ (47)
⎝ d 65 ⎟⎠

is is the fraction of the suspended load of the given size range and ib is the fraction of the
bed load transport of the same size range. The total suspended load transport (qs) is
given as Σisqs.

4. Total Load Transport

A knowledge of the rate of total sediment transport for given flow, fluid and sediment
characteristics is necessary in the study of many aspects of alluvial streams. Alluvial
streams carry not only water but also sediment and that the sediment transport rate is
closely related with the stability of the streams. Alluvial channels must be designed to
carry definite amount of water and sediment discharges. In other words, the rate of total
load transport must be treated as a variable affecting the design of the channel.
Similarly, knowledge of the total sediment transport rate is required for estimating the
rate of silting in reservoirs. Problems of aggradations and degradation, river-training
studies etc. also need knowledge of the total load transport for their solution.

The topics and basic principles of bed load transport, suspended load transport and
wash load have been discussed before. The next logical step is the study of the total load
transport. The total load is obviously the sum of the bed load, suspended load and wash
load.

Washload is made up of grain sizes finer than the size of bed material and thus, is rarely
found in the bed. The determination of wash load is difficult by analytical methods.
Usually the wash load can be related to the available supply of solid particles within the
watershed which enters the watercourse by sheet wash or bank caving. Owing to its

192
 Sediment Transport Rate

small size fractions, it moves readily in suspension and so can be measured and
calculated in suspended bed material load.

Einstein (1950) suggested that the limiting size of wash load and bed material load may
be chosen quite arbitrarily and can be taken as the grain diameter of which 10 percent of
the bed mixture is finer. But this is not certain. Because of this uncertainty of
separation, although the bed material load can be clearly defined by mean velocity of
flow, there is no definite relationships between the discharge and the wash load. Fig. 9
and Fig. 10 illustrate this difference.

Figure 9. Bed material load vs. mean velocity-Niobrara River

Figure 10. Washload vs streamflow-Powder River

193
Sediment Transport Technology

There is an indication from Fig.10 that the washload is small if the streamflow is low. It
appears that, although the bed material load can be fairly well predicted with the
hydraulic knowledge of the stream (Fig. 8), the washload depends on the hydrological
conditions as well. So far, little is known about the washload, but useful information for
the estimation of the sediment erosion rate from the watersheds is collected by
agricultural engineers.

In flume experiments there is no washload and the total load would be the bed material
load, on the other hand, in natural streams the wash load is always present and the total
load is the summation of the bed material load and the wash load. This point is
important during the comparison of laboratory and field data.

The determination of the total load can be made by Indirect and direct methods. Now
let’s discuss the indirect determination (or microscopic methods) first and let’s see the
various approaches for a direct determination (or macroscopic methods). In the first
method the total load is subdivided either into suspended load and bed load. In the
second method the suspension is taken as an advanced stage of traction along the bed;
therefore the total load should be related primarily to the shear parameter and no
distinction needs to be made between bed load and suspended load. The relationships
under this category are based on dimensional analysis, Intuition or complete empirical
considerations.

4.1. Indirect determinations (or microscopic methods)

This approach gives the total load by its fractions, the bed load and the suspended load.
The two commonly used methods under this category are Einstein (1950), and Colby et
al. (1955) procedures. The later is a modification of Einstein’s method so it is frequently
referred as the modified Einstein method. Bagnold (1966) has also made interesting
contributions to this approach.

Before we enter the discussion of the total load determination by direct and indirect
way, a few words of explanation is going to be useful. There are at least two points to
differentiate the bed load from the suspended load. One of them is the difference in the
behavior of the movement of the different loads. Within a rather thin layer, the so-called
bed layer, the particle motion is one of rolling, sliding and sometimes saltating. This
load is defined as bed load. Outside the bed layer the solid particles are continuously
supported by the turbulence and the load is the suspended load. A clear line of division
between the two kinds of loads doesn’t exist. A particle which is a part of bedload, may
be a part of suspended load sooner or later. A general correlation between the bed load
and the suspended load is not possible. This impossibility has been shown with many
field data by different researchers. But generally the quantity of suspended material is
larger than the bedload. In natural rivers (80-90) percent of the total load is suspended
load. The ratio of bed load to total load is lower in lowland streams than in mountain
streams. Second point is the difference in measurement techniques of the bed load and
the suspended load. The bed load is measured with suitable traps, where as the
suspended load is obtained from water samples carrying sediment.

4.1.1. Einstein’s method

In short, Einstein’s method involves the computation of bed load and suspended load
for a given size range and addition of these to get the total load corresponding to that
size range. The process is repeated for other size ranges and the loads are added.

194
 Sediment Transport Rate

In the application of this method, the bed material is classified into different classes
according to their sieves and the load is calculated for each class. It is usually enough to
classify the whole bed material into 6 or 8 classes.

If we combine the Einstein’s relation for bed load and suspended load which have been
discussed before, the total load in a given size range will be given by,

qt = qb + q s
is qs = ib qb ( PI1 + I 2 ) (48)
⎡ ⎛ h⎞ ⎤
it qt = ib qb ⎢2.30 log⎜ 30.2 ⎟ I1 + I 2 ⎥ (49)
⎣ ⎝ ∆⎠ ⎦
it qt = ib qb [1 + PI1 + I 2 ] (50)

and P is given as;

⎛ h⎞
P = 2.30 log⎜ 30.2 ⎟ (51)
⎝ ∆⎠

The second term in Eq. 48 is the fraction of suspended load in total load which was
expressed in terms of bed load. it and ib are the fractions of total load and bedload
respectively and, qt is total unit discharges and qb is bedload unit discharge.

I1 and I2 that appears in the given equations are given as follows;

α
⎡ Aα −1 ⎤ 1 ⎛ 1 ⎞
α ⎥∫ ⎜
I1 = 0.216⎢ − 1⎟ dz (52)
⎣ (1 − A) ⎦ A ⎝ z ⎠
α
⎡ Aα −1 ⎤ 1 ⎛ 1 ⎞
α ⎥∫ ⎜
I 2 = 0.216 ⎢ − 1⎟ ln zdz (53)
⎣ (1 − A) ⎦ A ⎝ z ⎠

In the above equations, ∆ = ks/x , (ks; Nikuradse equivalent sand roughness, x; is the
correction factor given by Einstein and Barbossa), A = a/h (h; flow depth, a; depth from
the bottom that the reference concentration c a is measured), α=W/(κU*) (κ; von
Karman constant, U*; shear velocity).

Hydraulic and sediment load calculations of Einstein’s procedure can be tabulated as in

Table 2 and Table 3.

Table 2. Hydraulic calculations of Einstein’s method

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
2
D65 V ⎛ β ⎞
Rb′ U *′ δ χ ∆ V ′
ψ 35 U *′′ Rb′′ Rb h AT Q X Y β* ⎜⎜ ⎟⎟ P
δ U *′′
⎝ β* ⎠
M m/sec m m m/sec m/sec m m m m2 m3/sec m

195
Sediment Transport Technology

Table 3. Sediment load calculations of Einstein’s method

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

D
D ib Rb′ ψ ε ψ* φ iB.qB A α l1 -l2 1+Pl1+l2 iT.qT iT.QT
x
m m (kgs/sec-m) (kgs/sec-m) (tons/day)

Explanations for Table 2:

1. Assume any reasonable value for Rb′ (hydraulic radius) using the method of Einstein
and Barbossa (This value depends on also grain size)
2. U *′ = ( g.R′b.S )
1/ 2

3. δ = 11.6 ν / U *′
5. Take the x from Fig. 11
6. ∆ = D65 U *′ (ks=D65; sand roughness)
⎛ R′ ⎞
7. V = 5.75U *′ log⎜12.27 b ⎟
⎝ ∆ ⎠
Rb′ ; hydraulic radius corresponding to grain size resistance
γ s − γ gD35 γ s* gD35
′ =
8. ψ 35 =
γ U *′2 U *'2
ψ 35
′ ; Einstein’s shear stress parameter corresponding to grains
9. Take the value of V U *′′ from Fig.12. U *′′ ; shear velocity corresponding to bed
undulations
11. Rb′′ = U *′′ g ⋅ S Rb′′ ; hydraulic radius corresponding to bed undulations
2

12. Rb = Rb′ + Rb′′

13. h=Rb (if the resistance of sidewalls is negligible)
14. AT is the cross-sectional area
15. Q=AT·V

Figure 11. Variation of x by ks/δ Figure 12. The variation of V/U* by ψ′

196
 Sediment Transport Rate

16. X is calculated by the following equations (X is the smallest grain diameter on the
bed that is fully affected by the turbulent flow):

X=0.77∆ if ∆>1.80δ for rough bed

X=1.39δ if ∆<1.80δ for hydraulically smooth bed

17. Take Y from Fig. 13. (Y is the correction for the lift force coefficient in the case of
sediment mixtures)
18. β x = log(10.6 X / ∆ )
19. β = log 10.6 : constant
⎛ h⎞
20. P = 2.30 log⎜ 30.2 ⎟
⎝ ∆⎠

Explanations for Table 3:

(
1. D is the geometric mean size of each range D = (Dlower .Dupper )
1/ 2
)
2. ib is the fraction of each range in the total on the bed
3. Rb′ is taken from Table 1
γ −γ d
4. ψ = s Einstein’s shear parameter
γ Rb′ S
5. Take ε from Fig. 14 (ε is the hiding factor)
2
⎛ β ⎞
7. ψ * = ε ⋅ Y ⋅ ⎜⎜ ⎟⎟ ⋅ψ
⎝ βx ⎠
8. Take φ* from Fig. 15
1/ 2
⎛ρ −ρ⎞
9. i B ⋅ q B = ib ⋅ φ* ⋅ γ s = ⎜⎜ s ⎟⎟ (
⋅ gD 3 )
1/ 2

⎝ ρ ⎠
10. A = 2D/h
11. α = W κU *′
12. Calculate I1 (Eq. 52)
13. Calculate I2 (Eq. 53)

Figure 13 Correction factor for lift force (Y) Figure 14. Hiding factor ε for fine grains

197
Sediment Transport Technology

Figure 15. The relation between the Einstein’s parameter φ*and ψ*

4.2. Direct determinations (macroscopic methods)

Several methods of macroscopic nature have been proposed for determination of the
total bed load in the literature. The features of these investigations and the final
relationships are summarized in this part of the chapter.

4.2.1. Laursen’s method

Laursen considered the following parameters to be important in the study of total
sediment transport; U*/W and D/h, the total load concentration c in percent by weight
for the given size range and the ratio of grain shear stress τ 0′ to the critical shear stress
τ 0c for the given sediment size range. In particular his analysis he gives the following
functional relationships,

7/6
⎛D⎞ ⎛ τ 0′ ⎞ ⎛U ⎞ qt
c =⎜ ⎟ ⎜⎜ ⎟⎟ func.⎜ * ⎟, c= (54)
⎝h⎠ ⎝ τ 0c ⎠ ⎝W ⎠ γq

The functional relation of the U*/W term could be find by using Fig 16.

Figure 16. The functional relation of the

term U*/W

198
 Sediment Transport Rate

Eq. 54 is calculated for each size range and summed up, the total load can be
determined as follows,

7/6
γq ⎛D⎞ ⎛ τ o′ ⎞ ⎛U ⎞
qt = ∑ ib ⎜ ⎟ ⎜⎜ ⎟⎟ func.⎜ * ⎟ (55)
100 ⎝h⎠ ⎝ τ 0c ⎠ ⎝W ⎠

Here q is the discharge of flow per unit width. This method gives reasonable results
especially for fine sediments dominated streams.

4.2.2. Garde and Albertson’s equation

Garde and Albertson suggested that the total load concentration can be expressed as;

⎛U h d ⎞
c = func⎜ * , , η ⎟ (56)
⎝ ν h ⎠

in which η is a function of sediment size and it has been expressed as function of Wd/ν.
Later Garde have postulated that the total load equation can take the same functional
form as the bed load equations and gave the following equation for the total load:

qt
γ sU * d
( )
= func τ * (57)

Fig. 17 shows the variation of qt/(γsU*d) with τ* from which it can be seen that the
relationship can be expressed as:

103

102
qt
γ sU *d
10

1.0

0.1
0.01
0.01 0.1 1.0 10
τ0
Figure 17. Garde’s total load relation

199
Sediment Transport Technology

qt
γ sU *d
( )
= 16 τ *4 (58)

Here, τ* is given as τ0/[(γs-γ)d] dimensionless shear stress (or effective shear stress).

Eq. 58 is based on flume and field data covering a wide range of variables. The median
size of the sediment should be used in this equation (The size of sediment, for which
fifty percent of material by weight is finer, is known as the median size).

4.2.3. Bagnold’s equation

Bagnold considers the flow as a mechanical system which transports the sediment
particles and applies the physical principles to this system to calculate the total load.

If q ′s and qb′ represent the submerged weights of suspended load and bed load per unit
area on the bed, and Us. and Ub represent their mean velocities of transport, we can write
the total of bed load and suspended load work rates as:

⎛W ⎞
qb′ U b tgα + q ′s U s ⎜⎜ ⎟⎟ (59)
⎝Us ⎠

Here tanα is the coefficient of intergranular solid friction of the bed material and w is
the fall velocity of suspended material. This total work rate has been related to the
available stream power per unit area τ0·V. Assuming the available stream power τ0·V as
the only common supply of energy to both transport mechanism. Bagnold related the
total load with the stream power as follows:

γs ⎛ eb V ⎞
qT = qb + q s = τ 0V ⎜ + + es (1 − eb )⎟ (60)
γs −γ ⎝ tan α W ⎠

Here eb and es are the transport efficiencies of bed load and suspended load. Bagnold
has shown that eb decreases with the increase of V as well as that of d. However, its
values lies between 0.15 and 0.11 as d increases from 0.03 mm to 1.0 mm. The
suspension efficiency es is shown to have a constant. Value of 0.015 and so es(l-eb) can
be taken as 0.01 for all practical purposes. The value of tanα varies from 0.375 to 0.75
and is a function of τ* and d. This variation was tabulated by Bagnold. Introducing these
values, Bagnold's equation reduces to:

γs ⎛ eb V ⎞
qT = τ 0V ⎜ + 0.01 ⎟ (61)
γs −γ ⎝ tan α W⎠

eb and tanα can be taken from the Fig. 18 and Fig. 19.

4.2.4. Egiazaroff method

Egiazaroff gives two different equations for the determination of total load for fully
developed turbulent flow and for transitional flow.

200
 Sediment Transport Rate

Figure 18. Variation of the efficiency eb by grain diameter and the average flow velocity

Figure 19. Variation of the friction coefficient; tanα

γγ s ⎛τ ⎞
qT = S 1 / 2 qK ⎜⎜ 0 − 1⎟⎟ (62)
γs −γ ⎝ τ 0c ⎠

and for transitional region,

γγ s ⎛ τV ⎞
qT = S 1/ 2 qK ⎜⎜ 0 − 1⎟⎟ (63)
γs −γ ⎝ τ 0cV0c ⎠

Here K ≅ 0.015. It is assumed that the total load is dependent to the stream power rather
than the bed shear stress for transitional region. When sand waves exist on the bed τ0,
should be changed by τ 0′ in Eq. 62 and Eq. 63.

201
Sediment Transport Technology

4.2.5. Bishop, Slaons and Richardson’s method

Bishop at al. (1965) were conducted an analysis of the problem of total load transport
based on flume and field data. When they applied Einstein’s method to these data, the
agreement between the measured and computed transport rates was found to be poor.

They developed a method which is simpler than Einstein’s and yields satisfactory
results for the data used by them. They assume that the same parameters influence to the
bed load and suspended load movements and there is no need to make a distinction
between them. Their method doesn’t compute the transport rate of different sizes but
gives the transport rate of the entire mixture, when the median size d is used as the
characteristic diameter.

They give a unique relation between φT and ψ ′ as shown in Fig. 20. Where,

qT
φT = 1/ 2
(64)
⎛ρ ⎞
ρ s g 3 / 2 d 3 / 2 ⎜⎜ s − 1⎟⎟
⎝ ρ ⎠

and

ρ s − ρ d 35
ψ′= (65)
Q R′S

There are three different segments in Fig. 20. The lower limb corresponds to the ripple
and dune bed regime, while the upper limb corresponds to the plane bed and antidune
bed regime; the middle segment has points from transition regime.

Figure 20. φT vs. ψ ′ curves of Bishop et al. (1965)

202
 Sediment Transport Rate

4.2.6. Engelund and Hansen’s method

Engelund and Hansen (1967) have proposed a total load equation relating the sediment
transport to the shear stress and the friction factor of the bed. They postulated that in the
case of a dune bed, the moving sediment particles are elevated through height equal to
the dune height h. The energy per unit time and per unit width required for elevating the
sediment load will be,

(γ s − γ ) qT h (66)
γs

This gain in potential energy was equated to the work done by the drag forces on the
moving particles during the same time. Assuming the particle velocity to be
proportional by U* and the shear stresses transferred to the particles to be ( τ 0′ − τ c ), the
following equation can be written,

qT
(γ s − γ )h = K (τ 0′ − τ c )LU * (67)
γs

in which K is a non-dimensional constant and L is the length of the dune. Rearranging

the Eq. 67, it takes the following form by multiplying left hand side by (f/f) and right
hand side by (d/d);

f ⋅ qT ⎛ h ⎞ τ i − τ cx
⎜ ⎟⎟ = K 0 U ⋅d (68)
γ s ⎜⎝ L ⋅ f ⎠ (γ s − γ )d *
τ cr
Replacing = 0.06 from shields diagram;
(γ s − γ )d

ρU *2 (γ s − γ )d
τ* = → U* = τ * ⋅
(γ s − γ )d ρ
f ⋅ qT ⎛ h ⎞
⎜ ⎟⎟ = K (τ * − 0.06 ) τ * (γ s − γ )d 3 / ρ (69)
γ s ⎜⎝ L. f ⎠

is established.

They also showed that h/(L·f) is constant for particular value of τ *′ and

τ *′ = 0.06 − 0.4τ *2 (70)

for dune bed. If the constant h/(L·f) is shown by K1, Eq. 69 takes the following form:

f ⋅ qT
K 1 = K ⋅ 0.4τ *2 τ * (γ s − γ )d 3 / ρ
γs

203
Sediment Transport Technology

or,

qT K
f = 0.4τ *3 / 2 (71)
γs (γ s − γ )d 3
/ρ K1

Here;

qT
γs (γ s − γ )d 3 / ρ

is equal to the total load parameter φT. Indeed;

qT qT qT
= = = φT
(γ s − γ )d
3/ 2
3
g ( ρ s − ρ )d 3 ⎛ρ ⎞
γs gρ s ρ s g 3 / 2 d 3 / 2 ⎜⎜ s − 1⎟⎟
ρ ρ ⎝ ρ ⎠

Hence Eq. 71 can be written as:

f ⋅ φT = const. ⋅ τ *5 / 2 (72)

Flume data in a size range of 0.19 mm ∼ 0.93 mm for dunes, transition, standing waves
and antidunes justified the following relation:

f ⋅ φT = 0.1τ *5 / 2 (73)

Here f depends on the type of bed undulation. It is interesting to note that, although this
relation was derived for the dune bed, it is satisfied by dune bed, transition, standing
wave and antidune data.

4.2.7. Yalin’s method

Yalin(1963) proposed a semi-theoretical approach to the problem of total load transport.
His analysis is restricted only by the plane bed and this has a limited practical utility.
The relation was proposed by Yalin, on the basis of Gilbert’s data.

⎛τ 0 ⎞ 1.575αqT ⎡ ⎛τ ⎞⎤
α ⎜⎜ − 1⎟⎟ = = ln ⎢1 + α ⎜⎜ 0 − 1⎟⎟⎥
⎝τ c ⎠ γ sU * d ⎣ ⎝τ c ⎠⎦

where,

0.4
⎛ ρ ⎞
α = 2.45⎜⎜ ⎟⎟ (τ ) 0.5

⎝ ρs
*c
⎠

and d is the median size of the sediment.

204
 Sediment Transport Rate

4.3. General remarks on sediment transport equations

As can be seen from the previous discussions, the methods of sediment transport
computation are either based on a semi-theoretical analysis or on dimensional
considerations. Furthermore, these equations are usually verified by flume data
collected under idealized conditions or on straight reaches of natural courses. However,
hydrologic conditions such as floods effect to the sediment transport. The various
coefficients are used in these equations and not all of them known accurately. These
coefficients are probably responsible for errors of as much as 100 percent between
different approaches. This is especially so, if the sediment transport equations are used
outside the range and conditions for which it was established. Thus a meaningful
comparison of measured and calculated rate of transport is difficult. This discussion
implies that not all the important factors governing the sediment problem are as yet
fully understood.

References

1. Ashida, K., and Michiue, M., “Study on hydraulic resistance and bedload transport
rate in alluvial streams”, Transactions, Japan Society of Civil Engineering, 206:
59-69 (in Japanese), (1972).
2. Bagnold, R. A., “The flow of cohesionless grains in fluids”, Philos Trans. R. Soc.,
London, 249, 235-297, (1956).
3. Bishop, A.A., Simons, D.B., and Richardson, E.V., “Total Bed Material Transport”,
JHD, Proc. ASCE, Vol. 97, No. HY. 2. March, (1965).
4. Bogardi, J., “Sediment Transport in Alluvial Streams”, Akademia Kiado, (1972).
5. Dohmen-Janssen, M., “Grain size influence on sediment transport in oscillatory
sheet flow”, Ph.D. thesis, Technical University of Delft, the Netherlands, 246 p.,
(1999).
6. Du Boys, Leliavsky, S., “An Introduction to Fluvial Hydraulics”, Constable and Co.
Ltd., pp.70-72, (1954).
7. Einstein, H. A., “Formulas for the Transportation of Bed Load”, Trans. ASCE, Vol.
107, (1942).
8. Einstein, H. A., “The Bed-load Function for Sediment Transportation in Open
Channel Flows”, Technical Bulletin 1026, U.S. Dept. of the Army, Soil
Conservation Service, (1950).
9. Egiazaroff, I. V., “Calculation of nonuniform sediment concentrations”, Journal of
Hydraulic Engineering, 91(4), 225-247, (1965).
10. Engelund, F., and Hansen E., “A Monograph on Sediment Transport in Alluvial
Streams”, Teknisk Forlag, Denmark, (1967).
11. Engelund, F., and Fredsoe, J., “A sediment transport model for straight alluvial
channels”, Nordic Hydrology, 7, 293-306, (1976).
12. Fernandez L. R., and van Beek, R., “Erosion and transport of bedload sediment”,
Journal of Hydraulic Research, 14(2): 127-144, (1976).
13. Fredsoe, J., and Deigaard, R., “Mechanics of Coastal Sediment Transport”, World
Scientific, ISBN 9810208405, 369 p., (1994).
14. Garde, R. J., and Albertson, M.L., “Bed Load Transport in Alluvial Channels”, La
Houille Blanche, No.3., (1961).
15. Garde, R. J., and Range Raju, K.G., “Mechanics of sediment Transportation and
Alluvial Stream Problems”, Wiley Eastern Ltd., (1977).

205
Sediment Transport Technology

16. Gao, P., “Mechanics of bedload transport in the saltation and sheetflow regimes”,
Ph.D. thesis, Department of Geography, University of Buffalo, State University of
New York, (2003).
17. Horikawa, K., “Nearshore Dynamics and Coastal Processes”, University of Tokyo
Press, 522 p., (1988).
18. Graf, W. H., “Hydraulics of Sediment Transport”, Mc Graw Hill, (1971).
19. Kalinske, A. A., “Movement of Sediment as Bed Load in Rivers” Trans. AGU,
Vol.28, No.4, (1947).
20. Meyer-Peter, E., and Müller, R., “Formulas for Bed-Load Transport”, Proceedings,
2nd Congress, International Association of Hydraulic Research, Stockholm: 39-
64, (1948).
21. Nino, Y., and Garcia, M., “Gravel saltation, 1, Experiments”, Water Resour. Res.,
30(6), p. 1907-1914, (1994a).
22. Nino, Y., and Garcia, M., “Gravel saltation, 2, Modelling”, Water Resour. Res.,
30(6), p. 1915-1924, (1994b).
23. Parker, G., “Hydraulic geometry of active gravel rivers”, Journal of Hydraulic
Engineering, 105(9), p. 1185-1201, (1979).
24. Parker, G., “Surface-based bedload transport relation for gravel rivers”, Journal of
Hydraulic Research, 28(4): 417-436, (1990a).
25. Parker, G., Solari, L., and Seminara, G., “Bedload at low Shields stress on arbitrarily
sloping beds: alternative entrainment formulation”, Water Resources Research,
39(7), 1183, doi: 10.1029/ 2001WR001253, (2003).
26. Parker, G., “Transport of gravel and sediment mixtures”, ASCE Manual 54,
Sediment Engineering, ASCE, Chapter 3, (in press).
27. Powell, D. M., Reid, I., and Laronne, J. B., “Evolution of bedload grain-size
distribution with increasing flow strength and the effect of flow duration on the
caliber of bedload sediment yield in ephemeral gravel-bed rivers”, Water Resources
Research, 37(5), 1463-1474, (2001).
28. Raudkivi, A. J., “Hydraulics of Sediment Transport”, Mc Graw Hill, (1971).
29. Sekine, M., and Kikkawa, H., “Mechanics of saltating grains”, J. Hydraul. Eng.,
118(4), 536-558, (1992).
30. Seminara, G., Solari, L., and Parker, G., “Bedload at low Shields stress on arbitrarily
sloping beds: failure of the Bagnold hypothesis”, Water Resources Research,
38(11), 1249, doi:10.1029/2001WR000681, (2002).
31. Shields, A., “Anwendug der Aehnlichkeitsmechanik und der Turbulenzforschung
auf die Geschiebebewegung“, Mitteulingen der Preussisehen Versuchsantalt für
Wasserbau and Schiffbau, Berlin, (1936).
32. Smart, G. M., and Jaeggi M. N. R., „Sediment Transport on Steep Slopes“, Mitteil.
64, Versuchsanstalt für Wasserbau, Hydrologie und Glaziologie, ETH-Zürich,
Switzerland, 191 pp. (1983).
33. Tsujimoto, T., “Mechanics of Sediment Transport of Graded Materials and Fluvial
Sorting”, Report, Faculty of Engineering, Kanazawa University, Japan (in Japanese
and English), (1991).
34. Wiberg, P. L., and Smith, J. D., “A theoretical model for saltating grains in water”,
J. Geophys. Res., 90(C4), 7341-7354, (1985).
35. Wiberg, P. L., and Smith, J. D., “Model for calculating bedload transport of
sediment”, J. Hydraul. Eng., 115(1), 101-123, (1989).
36. Wilcock, P. R., and Crowe, J. C., “Surface-based transport model for mixed-size
sediment”, Journal of Hydraulic Engineering, 129(2), 120-128, (2003).

206
 Sediment Transport Rate

37. Wilson, K. C., “Bed load transport at high shear stresses”, Journal of Hydraulic
Engineering, 92(6), 49-59, (1966).
38. Wong, M., “Does the bedload equation of Meyer-Peter and Müller fit its own
data?”, Proceedings, 30th Congress, International Association of Hydraulic
Research, Thessaloniki, J.F.K., Competition Volume: 73-80, (2003).
39. Wong, M., and Parker, G., “The bedload transport relation of Meyer-Peter and
Müller overpredicts by a factor of two”, Journal of Hydraulic Engineering,
(submitted).
40. Yalin, M. S., “An Expression for Bed Load Transportation, JHD, Proc. ASCE, Vol.
89, No. HV. 3, (1963).
41. Yalin, M. S., “Mechanics of Sediment Transport”, Pergamon Press, (1972).

207
208
 PROBABILISTIC APPROACH TO SEDIMENT TRANSPORT

Prof. Dr. Beyhan YEĞEN

Department of Hydraulics
College of Civil Engineering
Istanbul Technical University
Istanbul - Turkey
e-mail: emineb@itu.edu.tr

1. Basic Concepts of Probability Theory

1.1. Random variable, random event
Probability theory is a branch of mathematics which deals with problems having
uncertainty. This science analyses random variables. A random variable is such a
variable that the value it will take in future cannot be predicted beforehand. For
example, the number which will be cast by dice is a random variable. So is the amount
of sediment a river will transport in one year since we cannot tell this amount exactly
before the transportation event happens. Random variables cannot be analyzed by a
deterministic approach; they should be analyzed by a probabilistic approach.

When a random variable takes a certain value during an observation, this is called a
random event. It is not possible to predict which random event will occur at any
observation, but it is possible to determine the chance that any random event will be
seen.

1.2. Sets
The set concept has an important place in probability theory. A set consists of elements
which are definitely defined. For example the set of vowels in English is V={a,e,i,o,u}.
If all the elements of a set also happen to be elements of another set, the first set is a
subset of the second one. If set B is a subset of set A, we show it as B⊂A.

Venn diagrams are used to represent graphically the relations between a set and its
subsets. In Venn diagrams each set is represented by an area.

In Fig. l, B is a subset of A. If two sets have an area in common, then the two sets have
certain elements in common. The set which consists of these common elements is called
the intersection of the two sets. In Fig. l, the shaded area is the intersection of set A and
set C, and is shown as A∩C. If two sets have no elements in common, they are mutually
exclusive sets, in Fig. 1, set A and set D are mutually exclusive. The intersection of two
mutually exclusive sets is an empty set.

The set which consists of elements in either of two sets is called the union of the two
sets. The dotted area shown in Fig. 1 is the union of set E and set F. and is symbolized
as E∪F.

209
Sediment Transport Technology

Figure l. Representing some sets by Venn diagrams

1.3. Probability concept

In probability theory, the chance that a random event will happen is expressed by a
magnitude called probability. According to the basic axiom of probability theory, each
random event has a probability of which the value is in the range between 0 and l. If the
random variable is shown by X and the value this random variable takes during an
observation by x, the probability of the random event X = xi can be shown as follows:

P[ X = x i ] = p i (1)

The value of pi can vary between 0 and l. The probability being equal to 0 means that
this event will never happen, whereas this probability being equal to 1 means that this
event will be met at all observations. As probability increases from 0 to l. the event will
be met more frequently.

If a random variable is observed N times, and a random event X = xi is met ni times, the
probability of this event is defined as:

ni
pi = lim (2)
N →∞ N

According to the basic axiom of probability theory, the probability of the union of two
mutually exclusive events is equal to the sum of the probabilities of the two events. If A
and B are the two mutually exclusive events, then

P[A ∪ B ] = P[A] + P[B ] (3)

On the other hand, for the union of two events which are not mutually exclusive, the
above expression is not true. If A and B are not mutually exclusive, the following
expression must be used for their union:

P[A ∪ B ] = P[A] + P[B ] − P[A ∩ B ] (4)

The above expression can be easily obtained from the relationship of areas in a Venn
diagram (Fig. 1).

210
 Probabilistic Approach to Sediment Transport

1.4. Two dimensional and conditional sample spaces

When two random variables such as X and Y are considered jointly, a two dimensional
sample space is of subject. If during an observation Y = Yj event is observed for random
variable Y, while X=xi event is observed for random variable X, the pair of observations
(xi, yj) defines a point in two dimensional sample space. A probability value corresponds
to this point in the probability space. In such a case we also can define conditional
sample space. The observed X=xi events with the condition Y=yj event happens
constitute a new (one dimensional) sample space, which is a conditional sample space.
The events here are expressed as (X=xi ⎪Y=yj) where the expression to the right of the
vertical line is a condition; it is not considered as a random variable.

The probability of a conditional event can be found in the following manner:

P[ A ∩ B ]
P = [A B ] = for P[B ] ≠ 0 (5)
P[B ]

The probability that event A with the condition that event B happens is equal to the
probability of event A and B happening jointly, divided by the probability of the
condition.

We can write Eq. 5 in another form:

P[ A ∩ B ] = P[A B ] P[B ] (6)

and similarly:

P[A ∩ B ] = P[B A] P[A] (6a)

If there is no dependence between the occurrence of event A and event B, then the
conditional probability P[A⎪B] should not be different from P[A], since event B will not
effect whether event B occurs or not.

P[A B ] = P[A] (7)

In such a case event A and event B are mutually independent events. For mutually
independent events Eq. 6 can be written as:

P[A ∩ B ] = P[A] P[B ] (8)

As is seen the joint occurrence probability of independent events is equal to the product
of the probabilities of their separate occurrences.

1.5. Distributions of random variables

Most random variables in engineering practice are of continuous type. Since the number
of values a continuous random variable can take is infinite, we cannot consider simple

211
Sediment Transport Technology

events as X=x of which the probabilities approach zero. In continuous random variable
case, composite events defined as [x<X<x+dx] are of subject. Therefore the probability
density function (p.d.f.) f(x) is defined (Fig. 2) as:

f ( x)dx = P[x ≤ X ≤ x + dx ] (9)

More explicitly, the area bounded by the curve f(x), x-axis and the verticals drawn from
points x and x+dx shows the probability that the random variable will take a value in the
[x,x+dx] interval.

The probability of the random variable taking values in a finite interval can be
calculated as:

x2

P[x1 ≤ X ≤ x 2 ] = ∫ f ( x) dx (10)
x1

Since the probability that the random variable will take a value in the interval [-∞, ∞] is
certain, f(x) always will satisfy the following condition:

∫
−∞
f ( x)dx = 1 (11)

Another configuration is the express the probability of the random variable not to
exceed a certain value:

F ( x) = P[ X ≤ x] (12)

This function is called the cumulative distribution function. As will be seen from this
function’s definition, F(x) is an increasing function from 0 to 1.

The following relations between F(x) and f(x) can easily be seen from the above
equalities:

Figure 2. Definition of the probability density function (p.d.f.)

212
 Probabilistic Approach to Sediment Transport

x
F ( x) = P[−∞ < X ≤ x] = ∫ f (u )du
−∞
(13)

d ⎡ ⎤ dF ( x)
x
f ( x) = ⎢ ∫ f (u )du ⎥ = (14)
dx ⎣− ∞ ⎦ dx

The cumulative distribution function will always satisfy the following conditions:

0 ≤ F ( x)1, F (−∞) = 0 , F (∞ ) = 1
F ( x + ε ) ≥ F ( x) for ε > 0
F ( x2 ) − F ( x1 ) = P[x1 ≤ X ≤ x2 ]

The probability distribution function and cumulative distribution function defined

above, contain all information related to the random variable’s behavior.

1.6. Parameters of distribution

The main properties of a random variable’s behavior can be summarized by the help of
a few numbers. These special numbers are called the parameters of the distribution and
can be sufficient for many engineering problems.

Parameters determine properties of the distribution such as:

1. The center of the distribution, i.e. the value around which various values the random
variable takes during different observations will accumulate.
2. The magnitude of dispersion of various values of the random variable around this
center value,
3. The skewness of the distribution,
4. The peakedness of the distribution.

A parameter, which is a measure of one of the above cited properties of the distribution,
can be defined in different manners. Here the statistical moment type parameters will be
described.

1.7. Statistical moments of a random variable

1.7.1. Mean
The most important characteristic of a random variable’s distribution is its central value.
Different definitions can be used for the central value around which the values the
random variable will take during various observations will accumulate. However the
one which is most frequently used is the value called the mean or the expected value.
This value is defined as:

∞
µX = E [X ] = ∫ x f ( x) dx (15)
−∞

As will be seen from the above definition, the mean is the abscissa of the center of
gravity of the area bounded by the x-axis and the p.d.f. In problems related to random

213
Sediment Transport Technology

variables, the mean is the most significant parameter. If the random variable is analyzed
from a deterministic point of view, it will always be assumed to be equal to its mean.

1.7.2. Variance
Though the mean shows the central value of a random variable, it does not give any
information about the dispersion around this value. Variance is the most commonly
used parameter to measure this dispersion.

∞
Var [X ] = ∫ (x − µ )
2
X f ( x) dx (16)
−∞

From the above definition, it is seen that variance is in fact a moment of inertia of the
area bounded by the p.d.f. of the random variable and the x-axis with respect to a
vertical axis passing through its center of gravity (Fig. 3).

The variance possesses a dimension which is the square of the dimension of the random
variable. Since this is usually meaningless, standard deviation, the square root of the
variance, is used instead.

σ X = (Var[ X ])1 / 2 (17)

Standard deviation, carrying the same dimension as the random variable, is much more
meaningful. In many cases, the mean and the standard deviation are sufficient to decide
about the random variable's behavior.

1.7.3. Higher order moments

It was shown that the variance is the moment of inertia of the area under the p.d.f.
around its center of gravity. Similarly, higher order central statistical moments of the
distribution can be defined (Fig. 4). The nth order moment is defined as:

[ ] ∫ (x − µ
∞
µ = E (X − µ X ) = )n f ( x)dx
(n) n
X X (18)
−∞

Figure 3. Definition of variance

214
 Probabilistic Approach to Sediment Transport

Figure 4. Definition of higher order moments

It is seen that for n=1, µ x(1) = 0 , and for n=2, µ x( 2) = Var[X ]. The third order central
moment calculated for n=3 is a measure of the skewness of the distribution.

In case the distribution is symmetrical with respect to the

∞
µ X(3) = ∫ (x − µ )
3
X f ( x) dx (19)
−∞

mean, µ x( 3) = 0 . µ x( 3) has the same dimension as the cube of the random variable. In
order to obtain a dimensionless skewness coefficient, the ratio of µ x(3) to the cube of the
standard deviation is taken:

µ X(3)
C sX = (20)
σ X3

If Csx happens to be positive, the distribution is positively skewed (it has a tail extending
to the right), if it happens to be negative, the distribution is then negatively skewed (it
has a tail extending to the left).

1.8. Normal distribution

Among the many continuous distributions used in the probability theory the normal
distribution is by far the most important. Most random variables in practical
applications fit this distribution which is also called Gauss distribution. The p.d.f. of this
distribution is:

f ( x) =
σ X 2π
1
[
exp − ( x − µ X ) / 2σ X2
2
] −∞ < X < ∞ (21)

The two parameters of this distribution, shown as N(µ, σ2) shortly, are the mean (µx)
and the standard deviation (σx) of the random variable. An important feature of a normal
distribution is that it is completely determined by these two parameters. The normal
distribution is symmetrical therefore its coefficient of skewness is zero.

215
Sediment Transport Technology

The expression of the cumulative distribution function F(x) of the normal distribution
cannot be obtained analytically: it can only be calculated by numerical integration and is
tabulated. In order to prepare a unique table, firstly the standard variable is to be
introduced. This process transforms the variable into a standard variable as:

X − µx
Z= (22)
σx

Here X is the random variable and Z is the standard variable. The standard variable is
dimensionless with a mean of zero and standard deviation of unity. The cumulative
distribution function of the standard normal variable is uniquely tabulated.

The p.d.f. of the normal distribution is a bell-shaped curve around the mean µx
extending in both directions. Fortunately it is seldom necessary to extend the tails of a
normal distribution very far because the area under that part of the curve lying more
than 3 standard deviations away from the mean is for most practical purposes
negligible. The probability that the random variable remains in an interval of one, two
or three standard deviations on either side of the mean is respectively 0.65, 0.955 and
0.9975 (approximately 1) (Fig. 5).

Figure 5. The normal distribution

216
 NUMERICAL MODELLING

Assoc. Prof. Dr. H. Kerem CIĞIZOĞLU

Hydraulics Division, Civil Engineering Faculty
Istanbul Technical University
Istanbul - Turkey
e-mail: cigiz@itu.edu.tr

1. Theoretical Approach
In this study two essential problem surface flow model and transport model are
reviewed. Therefore theoretical equations and parameters are explained in two groups:
• Surface flow model
• Transport model

1.1. Governing equations

Governing equations of hydraulic models are based on conservation laws. Flow model
is consisted on equations of:
• Continuity law
• Momentum law

The equations represent mass conservation and momentum balance for each of the two
coordinate directions.

1.1.1. Flow model equations

The basic equations that govern the flow of a fluid were laid in 1750 by Euler as the
conservation law of momentum for a mass point and by applying it in integrated form to
a fluid continuum. He also generalized d’Alembert’s differential equation of continuity.
Hence, the presently accepted form of these equations originates from him. Friction
within the fluid has been highly debated at that time, but promising results for practical
and tractable use were not given before the 19th century.

Attraction and repulsion between molecules was considered by Navier (1822) in a

purely theoretical treatise about the motion of a fluid. However, even though Navier
gave the complete equation of motion for a fluid with internal friction, he could not
attach a particular physical significance to the essential parameter today known as
viscosity. As it is noted the approach of Navier does not fulfill the continuum
assumption of the fluid.

The problem was successfully re-analyzed by de Saint-Venant (1843), who considered

internal stresses instead of collisions of molecules. His assumptions for the stresses of a
continuum fluid are similar to those given earlier by Cauchy (1828) for an elastic
continuum. The form given to these stresses is the only extension possible of Newton’s
hypotheses of fluids with internal frictional shear.

217
Sediment Transport Technology

The assumption that the stresses are linear functions of the rates of distortion, and not
like the pressure uniform in all directions, allowed Saint-Venant to distinguish
tangential and normal stresses. His equations were not restricted to a specific flow like
laminar flow but are valid for any flow involving stresses of the type he had assumed.
The derivation implied that these stresses are proportional to the local formation and
destruction of eddies i.e. the stress tensor does not change when molecular scales or
even large scales are considered.

Eleven years later the derivation of the fundamental equation was obtained for a second
time by Stokes (1845) who also gave a physical interpretation of the parameter inherent
to his theory as dynamic viscosity. He also confirmed the theory by experiments
(Sander, 1998).

Continuity equation
Conservation of mass can be described by continuity equation. In order to explain the
continuity equation, we assume some considerations that flow is steady and fluid is
incompressible (ρ is constant).

Due to these assumptions we should define the constant volume equation for any
section of flow layers.

In one dimensional space (Fig. 1), the velocity variation in y direction should be
neglected. So the velocity in all points is a constant V averaged sectional velocity value
(Güney, 1995).

The volume of fluid that pass from a section at unit time is named as flow and
formulated in Eq. 1.

Q = V1 A1 = V2 A2 = V3 A3 = ... (1)

In this equation A is the cross sectional area of flow layer in m2, V is the sectional
averaged velocity of fluid in m/s, and Q is the flow quantity in m3/s.

The theory of unsteady and incompressible flow shows that water depth changes with
time as it is seen in Fig. 2. In this figure, u is the water velocity component in x
direction in m/s, H is the water level in m, dx is the distance of longitudinal profile.

According conservation of mass, rate of accumulation is expressed Eq. 2 and with local
time (t) derivation in Eq. 3.

Figure 1. Particle path of fluid on cross sectional area with velocity

218
 Numerical Modelling

⎛ ⎛ ∂H ⎞ ⎞
⎜⎜ H − ⎜ ⎟dx ⎟⎟
⎝ ⎝ ∂x ⎠ ⎠
⎛ ⎛ ∂H ⎞ ⎞
⎜⎜ H + ⎜ ⎟dx ⎟⎟
∂u ⎞ ⎝ ⎝ ∂x ⎠ ⎠
⎛
⎜u − dx ⎟ H, u, v
⎝ ∂x ⎠
⎛ ⎛ ∂u ⎞ ⎞
⎜⎜ u + ⎜ ⎟dx ⎟⎟
2dx ⎝ ⎝ ∂x ⎠ ⎠

Figure 2. Longitudinal profile of unsteady flow

Rate of accumulation = flux incoming- flux outgoing (2)

∂H
Rate of accumulation = 2dx (3)
∂t

The flux incoming and outgoing are expressed in Eq. 4 and Eq. 5.

⎛ ∂u ⎞⎛ ∂H ⎞
Flux incoming = ⎜ u − dx ⎟⎜ H − dx ⎟ (4)
⎝ ∂x ⎠⎝ ∂x ⎠
⎛ ∂u ⎞⎛ ∂H ⎞
Flux outgoing = ⎜ u + dx ⎟⎜ H + dx ⎟ (5)
⎝ ∂x ⎠⎝ ∂x ⎠

The rate of accumulation and net flux with incoming and outgoing fluxes are expressed
in Eq.6.

∂H ⎛ ∂u ⎞⎛ ∂H ⎞ ⎛ ∂u ⎞⎛ ∂H ⎞
2dx= ⎜u − dx⎟⎜ H − dx⎟ −⎜u + dx⎟⎜ H + dx⎟ (6)
∂t ⎝ ∂x ⎠⎝ ∂x ⎠ ⎝ ∂x ⎠⎝ ∂x ⎠

From Eq. 6, one dimensional time dependent continuity equation (Eq. 7) can form.

∂ (uH ) ∂H
+ =0 (7)
∂x ∂t

Eq. 8 shows the relation in change and depth with time and x space. The two
dimensional form of with the water velocity componenet v in direction y is shown in Eq.
8 (Dalziel and others, 1999).

∂ (uH ) ∂ (vH ) ∂H
+ + =0 (8)
∂x ∂y ∂t

If there is an injection between two sections, the injected water will be stated at the right
side of the equation with the term Q in m3/s (Eq. 9).

∂ (uH ) ∂ (vH ) ∂η
+ + =Q (9)
∂x ∂y ∂t

219
Sediment Transport Technology

Momentum equations
Fig. 3 below describes the flow in two dimensional x-y form in unit z dimension. The
mass of the figure below is defined as m = ρdxdy and P is pressure on surface (Güney,
1995).

As the fluid is assumed as ideal, frictional forces do not act on its surfaces and the
forces only act on it are surface pressures.

The forces that act on the square are Fx and Fy pressure forces that are described with
Newton’s Second Law in Eqs. 10, 11a-b, 12, 13, 17, and, 18.

The X and Y terms are the mass forces acting on particle. The derivation of total
acceleration components (ax, ay) are given in Eqs. 14, 15, 16, 19, 20, and, 21.

Fx = a x dm (10)
Fy = a y dm (11a)
⎛ ⎛ ∂P ⎞ ⎞
Fx = Pdy − ⎜⎜ P + ⎜ ⎟dx ⎟⎟dy + Xρdxdy (11b)
⎝ ⎝ ∂x ⎠ ⎠
⎛ 1 ⎞⎛ ∂P ⎞
Fx = ⎜⎜ − ⎟⎟⎜ ⎟dm + Xdm (12)
⎝ ρ ⎠⎝ ∂x ⎠
⎛ 1 ⎞⎛ ∂P ⎞
a x dm = ⎜⎜ − ⎟⎟⎜ ⎟dm + Xdm (13)
⎝ ρ ⎠⎝ ∂x ⎠
du
ax = (14)
dt
∂u ⎛ ∂u ⎞ dx ⎛ ∂u ⎞ dy
ax = +⎜ ⎟ +⎜ ⎟ (15)
∂t ⎝ ∂x ⎠ dt ⎜⎝ ∂y ⎟⎠ dt
∂u ⎛ ∂u ⎞ ⎛ ∂u ⎞
ax = + ⎜ ⎟u + ⎜ ⎟v (16)
∂t ⎝ ∂x ⎠ ⎜⎝ ∂y ⎟⎠

y
⎛ ⎛ ∂P ⎞ ⎞
⎜ P + ⎜⎜ ⎟⎟dy ⎟dx
⎜ ⎟
⎝ ⎝ ∂y ⎠ ⎠

⎛ ⎛ ∂P ⎞ ⎞
Pdy ⎜⎜ P + ⎜ ⎟ dx ⎟⎟ dy
⎝ ⎝ ∂x ⎠ ⎠

Pdx

Figure 3. Motion for ideal fluids in two dimensional form

220
 Numerical Modelling

⎛ ⎛ ∂P ⎞ ⎞
Fy = Pdx − ⎜⎜ P + ⎜⎜ ⎟⎟dy ⎟⎟dx + Yρdxdy (17)
⎝ ⎝ ∂y ⎠ ⎠
⎛ 1 ⎞⎛ ∂P ⎞
Fy = ⎜⎜ − ⎟⎟⎜⎜ ⎟⎟dm + Ydm (18)
⎝ ρ ⎠⎝ ∂y ⎠
dv
ay = (19)
dt
∂v ⎛ ∂v ⎞ dx ⎛ ∂v ⎞ dy
ay = +⎜ ⎟ +⎜ ⎟ (20)
∂t ⎝ ∂x ⎠ dt ⎜⎝ ∂y ⎟⎠ dt
∂v ⎛ ∂v ⎞ ⎛ ∂v ⎞
ay = + ⎜ ⎟u + ⎜ ⎟v (21)
∂t ⎝ ∂x ⎠ ⎜⎝ ∂y ⎟⎠

∂v/∂t and ∂u/∂t are local accelarations as a result of temporal variation of velocity.
u(∂v/∂x); v(∂v/∂y); u(∂u/∂x) and v(∂u/∂y) are general convective accelarations as a result
of spatial variation of velocity.

If the force and the acceleration equations are combined, the conservation of momentum
will be formed. The equations, which are known as Eulers Momentum Equations, are
seen in Eqs. 22 and 23 (Güney, 1995).

644accelerati
474 448 644
ons
4forces
7444 8
∂u ⎛ ∂u ⎞ ⎛ ∂u ⎞ ⎛ 1 ⎞⎛ ∂P ⎞
+ ⎜ ⎟u + ⎜⎜ ⎟⎟v = ⎜⎜ − ⎟⎟⎜ ⎟ + { X (22)
{∂t ⎝ ∂x ⎠ ⎝ ∂y ⎠ ⎝ ρ ⎠⎝ ∂x ⎠ mass
local
1442443 14243
convective pressure
forces
644474448 644
accelerations
47444 8
∂v ⎛ ∂v ⎞ ⎛ ∂v ⎞ ⎛ 1 ⎞⎛ ∂P ⎞
+ ⎜ ⎟u + ⎜⎜ ⎟⎟v = ⎜⎜ − ⎟⎟⎜⎜ ⎟⎟ + Y{ (23)
∂t ⎝ ∂x ⎠ ⎝ ∂y ⎠ ρ ⎠⎝ ∂y ⎠ mass
{
local
1442443 ⎝142 43
convective pressure

The momentum equations of fluid dynamics were originally derived by Navier (1827),
Poisson (1831), Saint Venant (1843), and Stokes (1845). In the older classical literature,
they are called the Navier-Stokes equations. In the recent computational literature, the
term Navier-Stokes equations includes the continuity and energy equations, as well as
the momentum equations (Orkwis and Wolff, 2001).

As it was mentioned before, the frictional forces acting on particle are neglected with
ideal fluid assumption. Navier-Stokes equations (Eqs. 24 and 25) improved Eulers’
equations by adding frictional forces νT(∂2u/∂2x+∂2u/∂2y).

∂u ⎛ ∂u ⎞ ⎛ ∂u ⎞ ⎛ 1 ⎞⎛ ∂P ⎞ ⎛ ∂ 2u ∂ 2 v ⎞
+ ⎜ ⎟u + ⎜⎜ ⎟⎟v = ⎜⎜ − ⎟⎟⎜ ⎟ + X + υ T ⎜⎜ 2 + 2 ⎟⎟ (24)
∂t ⎝ ∂x ⎠ ⎝ ∂y ⎠ ⎝ ρ ⎠⎝ ∂x ⎠ 14 ⎝ ∂4
x ∂y
2443⎠
vis cos ity

∂v ⎛ ∂v ⎞ ⎛ ∂v ⎞ ⎛ 1 ⎞⎛ ∂P ⎞ ⎛ ∂ u ∂ 2v ⎞
2
+ ⎜ ⎟u + ⎜⎜ ⎟⎟v = ⎜⎜ − ⎟⎟⎜⎜ ⎟⎟ + Y + υ T ⎜⎜ 2 + 2 ⎟⎟ (25)
∂t ⎝ ∂x ⎠ ⎝ ∂y ⎠ ⎝ ρ ⎠⎝ ∂y ⎠ 14 ⎝ ∂4
x ∂y
2443⎠
vis cos ity

221
Sediment Transport Technology

Saint Venant equations

Saint Venant equations are obtained from Navier-Stokes equations (Eqs. 7 and 8) by
averaging the horizontal velocities along the depth of the water body. The pressure
variable P is replaced with the water height H using the hydrostatic pressure distribution
assumption. The Saint Venant equations in two dimensional forms with are given in Eq.
26 and Eq. 27.

∂u ⎛ ∂u ⎞ ⎛ ∂u ⎞ ⎛ ∂H ⎞
+ ⎜ ⎟u + ⎜⎜ ⎟⎟v = (− g)⎜ ⎟ + { fv − gu⎜
( )
⎟ + kW W + Q(u − u0 )
⎛ u 2 + v2 ⎞
(26)
{∂t ⎝ ∂x ⎠ ⎝ ∂y ⎠ 142 ⎝ ∂x3⎠ corolis ⎜ HC ⎟ 123 142
2 x
H4 3
local
144244 3 4 ⎝ 42443⎠ wind
14 source/ sin k
convective pressure
friction

∂v ⎛ ∂v ⎞ ⎛ ∂v ⎞ ⎛ ∂H ⎞
+ ⎜ ⎟u + ⎜⎜ ⎟⎟v = (− g )⎜⎜ ⎟⎟ + { fu − gv⎜
( )
⎟ + kW W + Q(v − v0 )
⎛ u2 + v2 ⎞
(27)
∂t ⎝ ∂x ⎠ ⎝ ∂y ⎠
{ ⎝4∂y ⎠ coriolis ⎜ HC 2 ⎟ 1 42y
4
3 142 H4 3
local
144 2 443 1 4 2 3 ⎝
1442443 ⎠ wind
source/ sin k
convective pressure
friction

In these equations the variables are explained below:

u : water velocity components in x direction, m/s
v : water velocity components in y direction, m/s
t : time, s
g : acceleration due to gravity, m2/s
H : total water depth, m
η : change in water level, m
f : coriolis parameter
C : chezy friction coefficient, m1/2/s
k : wind stress parameter
W : horizontal wind velocity, m/s
Wx : wind velocity component in x direction, m/s
Wy : wind velocity component in y direction, m/s
Q : injected water, m3/s
u0 : injected water velocity component in x direction, m/s
v0 : injected water velocity component in y direction, m/s

1.2. Transport model

1.2.1. Convection diffusion equation
The movement of sediments, temperature or a pollutant in water can be described by
convection-diffusion equation. The convection-diffusion equation for steady flows is
expressed in Eq. 28 (Olsen, 2000).

∂c ∂c ∂ ⎛ ∂c ⎞
u + w = ⎜ DX ⎟ (28)
1∂4 ∂z 1
x 243 ∂x42 ∂x ⎠
⎝ 43
convection diffusion

The transport model expressed in Eq. 28 is based on two main process named as
convection and diffusion. Convection is the movement of particle by water. Particles
vertical movement w∂c / ∂z should also be defined as convective term. The term is

222
 Numerical Modelling

similar to the Navier-Stokes equation term and should be described as spatial variation
of transport process.

The second term diffusion is explained as turbulent diffusion of particle due to the
turbulent mixing and concentration gradients. Here D x and D y are the dispersivity
coefficients in m2/s. In two dimensional space the conservation of concentration for
unsteady sediment transport process the equation is expressed in Eq. 29 (Olsen, 2000).

∂c ∂c ∂c ∂c ∂ ⎛ ∂c ⎞ ∂ ⎛ ∂c ⎞
v +u +w + = ⎜ D x ⎟ + ⎜⎜ D y ⎟⎟ (29)
∂y ∂x ∂z ∂t ∂x ⎝ ∂x ⎠ ∂y ⎝ ∂y ⎠

As it was mentioned in continuity equation, the flow model is assumed as unsteady.

Therefore H must be expressed in transport equations. Additionally the injected water
flux difference to the conservation of concentration as S = Q(c0 − c ) in equation form.

Q is the flow quantity of discharge in m3/s, c0 is the concentration of discharge in kg/m3,

and S is the flux term is expressed for pollutant, heat and sediment in different forms
(Vatnaskil, 1998).

1. In pollutant transport process, the flux term is expressed in Eq. 30.

S = λV Hc (30)

In this equation λV is the volumetric decay parameter and is calculated from Eq. 31.

ln 10
λV = (31)
T90

In this, T90 is the time in which 90 percent of the population is no longer detectable.

2. In heat transport process, the flux term is expressed in Eq. 32.

S = λ Ac (32)

In this equation λA is the areal decay parameter and calculated from Eq. 33.

K
λA = (33)
ρC P

In Eq. 33 ρ is the density of water in kg/m3, CP is the specific heat of water in J/kg°C,
and K, which can be calculated from Eq. 34, is the atmospheric cooling factor in J/
m2°Cs.

K = 0.2388(4.6 − 0.09(T + c ) + 4.06 W )exp(0.033(T + ct )) (34)

In equation above, T is the reference temperature in °C, |W| is the wind speed at 2 m
elevation in m/s, and ct is the excess temperature °C.

223
Sediment Transport Technology

3. In sediment transport process, the flux term is expressed in Eq. 35.

S= S{1 + S{2 (35)

deposition erosion

Deposition of sediment on bed is calculated with Eq. 36 or Eq. 37.

⎛ τ ⎞
If τ b < τ d , then S1 = wc⎜⎜1 − b ⎟⎟ (36)
⎝ τd ⎠
If τ b > τ d , then S1 = 0 (37)

w is the settling velocity, τb is the bottom shear stress, and τd is the critical deposition
shear stress.

Erosion of sediment from bed is calculated with Eq. 38 or Eq. 39.

⎛τ ⎞
If τ b > τ e , then S 2 = E ⎜⎜ b − 1⎟⎟ (38)
⎝τ d ⎠
If τ b < τ e , then S 2 = 0 (39)

E is the erosion rate coefficient, and τe is the critical shear stress for erosion. According
to the assumptions the general form of transport equation is seen in Eq. 40 (Vatnaskil,
1998).

∂ (Hc ) ⎛ (Hvc ) ⎞ ⎛ (Huc ) ⎞ ∂ ⎛ ⎛ ∂c ⎞ ⎞ ∂ ⎛ ⎛ ∂c ⎞ ⎞

+ ⎜⎜ ∂ ⎟⎟ + ⎜ ∂ ⎟ = ⎜⎜ HDx⎜ ⎟ ⎟⎟ − ⎜⎜ HDy⎜⎜ ⎟⎟ ⎟⎟ + S − Q(c0 − c) (40)
∂t ⎝ ∂y ⎠ ⎝ ∂x ⎠ ∂x ⎝ ⎝ ∂x ⎠ ⎠ ∂y ⎝ ⎝ ∂y ⎠ ⎠

1.3. Parameters
1.3.1. Flow model parameters
The flow equations include different terms about gravitational, frictional, viscous and
inertial forces that are acting on surface flow system. In order to describe the
relationship between these forces, some parameters are used. The parameters that are
used in equations are described below.

Reynolds number
Turbulence arises from the non linear terms in the momentum equation. Reynolds
number describes the importance of these terms and it is the ratio of non-linear terms to
the viscous terms (Eqs. 41 and 42).

⎛ u∂u ⎞ U
⎜ ⎟ U
Non − Linear − Terms ∂x
Re = = ⎝ 2⎠ ≈ L (41)
Viscous − Terms ⎛ υT ∂ u ⎞ υ U
⎜⎜ ⎟⎟
L2
T
⎝ ∂x ⎠

224
 Numerical Modelling

Re =
H (u 2
+ v2 ) (42)
υT

According to this term, if the speed is small, the flow is smooth. This is called laminar
flow. At higher speeds, the flow becomes irregular and turbulent. The transition
occurred at Re = VD / υ T ≈ 2000 , where V is average speed, D is diameter, and υ T is
the kinematic viscosity fluid. As Reynolds number increases, the flow becomes more
turbulent (Stewart, 1998).

Froude number
The value of Froude number determines the type of the flow. If the Froude number is
smaller than 1, then the flow is determined as subcritical or fluvial flow. If the Froude
number is greater than 1, then the flow is determined as supercritical or torrential flow.
And if the Froude is equal to 1, then the flow is critical. Froude number is expressed in
Eq. 43 (Stewart, 1998).

Fr =
(u 2
+ v2 ) (43)
gH

In this equation u, v are the velocity components of water in m/s, g is gravitational

acceleration in m2/s, and H is the water depth in m.

Molecular viscosity
Molecules in a fluid close to a solid boundary can strike to the boundary and transfer
momentum to the boundary. The transferred momentum is called molecular viscosity.
Molecular viscosity is the ratio the stress Tx tangential to the boundary of a fluid and the
shear of the fluid at the boundary. The stress term expression is given in Eq. 44.

∂u
Tx = ρυ (44)
∂z

Some values for viscosity are seen in Table 1 (Stewart, 1998).

Manning coefficient
Manning coefficient describes the roughness of water bed. It is an empirical measured
effect of roughness between natural surface, wall, bed configuration and fluid. The
roughness creates friction and resistance that causes losses on velocity, elevation,
energy and other related terms.

Table 1. Viscosity values of water

υwater 10-6 m2/s
υtar at 15 °C 106 m2/s
υglacier ice 1010 m2/s

225
Sediment Transport Technology

Manning coefficient should be estimated by the following expression in Eq. 45.

26
n= 1
(45)
6
k s

in this equation ks is roughness coefficient. The relation between the roughness and the
grain size distribution for bed material is defined by Van Rijn (1982) in Eq. 46 (Olsen,
2001).

k s = 3d 90 (46)

In this equation d90 is grain size sieve that 90 % of the material is finer. It shows that
manning coefficient depends on the grain size of bed material. In Table 2 various
categories of rivers and definitions of bottom with manning values are listed.

Chezy coefficient
The Chezy is expressed with Manning coefficient in Eq. 47 (Vatnaskil, 1998).

⎛1⎞ 1
C = ⎜ ⎟H 6 (47)
⎝n⎠

In this equation n is the Manning coefficient, H is the water level in m.

Table 2. Values of coefficient n for different categories of rivers (Hydrosoft, 1999)

River Manning
Category definitions
category coefficient
1 Small mountain flow with very irregular bottom, 10 to 30 m 0.038-0.043
wide
2 Large mountain flow from 30 to 50 m wide, with a gradient 0.034-0.037
exceeding 0.002 and a rough gravel bottom (many stones 10
to 20 cm in diameter)
3 Similar or larger rivers, with a gradient between 0.000 and 0.030-0.033
0.002 and a gravel bottom with pebbles whose maximum
diameters generally do not exceed 8 to 10 cm (Rhine in Basel)
4 Gradient ranging from 0.0006 and 0.0008; pieces of gravel 0.027-0.029
from 4 to 8 cm in diameter (Rhône in Lyons)
5 Same gradient with smaller gravel (Rhône near Scex, Lake 0.025-0.026
Geneva)
6 Gradient less than 0.0006 but over 0.00025, very small stones 0.023-0.024
or sand (Danube in Vienna)
7 Low-turbulence river, gradients between 0.00012 and 0.022-0.023
0.00025, sand and mud bottom (Seine, Saône, Lower Rhine)
8 Very large river, very low gradient (less than 0.00012) and 0.020-0.021
very smooth bottom

226
 Numerical Modelling

Coriolis coefficient
Coriolis force becomes evedent due to Earth rotation and deflects fluid motion to the
right in the Northern Hemisphere. Fig. 4 shows the deflections of motion with the effect
of Coriolis force.

Coriolis coefficient is depends on latitude and earth’s rotational speed which is equal to
7.27 × 10 -5 rad/s. Coriolis parameter is expressed in Eq. 48 (Vatnaskil, 1998).

f = 2ΩSinφ (48)

In Eq. 48 Ω is earth’s rotational speed in 1/s, φ is the latitude in degrees.

Wind stress parameter

It is an empirical coefficient depending on which elevation the wind speed measured.
There are number of different formulas about it. According to Bengtsson the wind stress
coefficient should be taken as 1.1× 10 -3 (Olsen, 2001). Another empirical formula is
seen in Eq. 49 (Vatnaskil, 1998).

k=
(ρ AC D ) (49)
ρ

In expression above, ρ A is density of air in kg/m3, ρ is density of fluid in kg/ m3, and
CD is the wind drag coefficient which is an emprical parameter.

1.3.2. Transport model parameters

In this study the transport model is obtained to solve the salinity and oxygen, therefore
the necessary parameters for appropriate transport process are described below.

Figure 4. Coriolis force effect on fluid motion

227
Sediment Transport Technology

Longitudinal dispersivity
The dispersivity coefficient is effective parameter in transport process. It regulates the
concentration distribution in fluid. There are several empirical formulas about heat and
pollutant transport dispersivity coefficients. According to McQuivey and Keefer (1974)
the longitudinal dispersion coefficient should be found with the following empirical
formula in Eq. 50 (Olsen, 2001).

Q
DX = 0.058 (50)
IB

Q is the water discharge in m3/s, I is slope, B is width in m. As longitudinal dispersion

coefficients are a function of stream velocity, and Fisher developed an empirical
formula for dispersivity coefficient in 1979 as seen in Eq. 51 (Olsen, 2001).

DX = 0.011
(UB )
2
(51)
Hu *

Here, U is the water velocity in x direction in m/s, u * is the shear velocity in m/s.
Typical values of longitudinal dispersivity for small streams are 0.05 to 0.3 m2/s and
large rivers can be > 1000 m2/s.

Transversal dispersivity
In previous section, longitudinal dispersivity coeffient was expressed. The difference
between these two coefficents is the directions. If the point of inmjection source is not
uniform across for instance a single near a bank, contaiminant must travel before it
becomes uniformly mixed in cross section which is called transverse mixing zone.
Transversal dispersivity is used to calculate the transverse mixing zone.

Transverse mixing is responsible for the lateral movement and this only occurs as a
result of turbulence. Transversal dispersivity coefficient can be estimated with Eqs. 52
and 53.

D y = 0.15 Hu * for straight channels (52)

D y = 0.60 Hu *
for natural channels (53)

Here H is the water level in m, and u * is the shear velocity in m/s. Measured transversal
dispersivity cofficient has got a range of 0.0014 to 3.1 m2/s.

Shear velocity
Turbulence is due to shear velocity and is given in Eq. 54.

u* = (gHI ) (54)

In this equation g is the earth’s gravitational acceleration in m2/s, H is water level in m,

and I is the the slope of bottom topography.

228
 Numerical Modelling

1.4. Boundary conditions

These conditions are required on physical boundaries to solve the flow and transport
equations. The most used types of boundary conditions are:
• Zero gradient
• Dirichlet

In the first type of boudary condition, the derivative of the variable at the boundary is
zero. The value at the boundary is the same as the value in the element closest to the
boundary. In iterative solutions, this is straightforward. This boundary condition gives
good results at outflow open section. It can also be used at wall boundaries. This
boundary condition is a type of Neumann boundary condition which is defined as a
given gradient prescribed on the boundary.

Dirihclet boundary conditions are defined as given values of variables at boundaries.

This type of boundary condition is usually used at inflow open sections. For example
the water level (H), the concentration at upstream open section are Dirichlet boundary
contions (Olsen, 2001).

Boundaries are defined at internal and external boundaries. Internal boundaries are the
islads, blocks other structures at inner parts of flow field. These are also known as
physical boundaries.

If a part of physical boundary is an open section which means there is an interaction

with another water body, that part of boundary will be defined as open section with
water level, flow quantity or normal water velocity.

Boundaries should be external or internal and defined as wall or open. The special
conditions for wall and open sections are defined under in flow and transport
conditions.

Additionally the source/sink can be defined at any point with flow quantity and
concentration values.

1.4.1. Flow model

Wall boundaries are also known as solid and closed boundary. In these types of
boundaries, two main conditions can be defined as:
• Free slip condition, Un = 0.
• No slip condition, U = V = 0.

In the first type the normal water velocity is defined as 0 as a result of assumption that
the particle can’t pass through the boundary.

In the second type the water velocity components are defined as 0 without the
assumption mentioned above.

The selection of boundary condition depends on several reasons such as geometry of

study area, bottom topography. Comparing the results with field measurements in
studies shows that usually free slip boundary condition gives better results.

229
Sediment Transport Technology

In open sections, boundary conditions should be defined as:

• Water level
• Flow quantity

Any of these conditions defined at a part of physical boundary can be time dependent
boundary. Each of the time interval values are defined as boundary condition.

1.4.2. Transport model

In open sections, two main conditions can be defined as:
• Specified concentration, c ≠ 0
• Zero concentration gradient, indicating convective transport through the boundary,
c=0

The first type of boundary condition can be time dependent. Each of the time interval
values are defined as boundary condition.

2. Numerical Modeling Methods

By the beginning of twentieth century, scientific studies on numerical methods were
accelerated. Although the computer technology hadn’t begun yet, the numerical
methods by different engineers were developed. In order to solve the environmental
problems especially river and lake pollutions, numerical methods have applied in
different forms. By this way the problems effects should be viewed successfully before
they occur.

After the invention of computer, numerical algorithms were developed and applied in
scientific studies. At hydraulic engineering numerical methods for flow problems
weren’t available as the computers were too slow. Later in 1980’s previously developed
methods were carried out with new integrated computers.

The numerical methods are based on known as boundaries and unknowns and operators
and iteration parameters. The most important point in this structure for analyse is
iteration parameters control the stability and accuracy of iteration. At this point to have
better convergence should not be easy sometimes, therefore the CPU time has an
important role in iteration process. By the evolution of computer technology enough
CPU time with significant enviromental properties have been using in hydraulic
engineering (Olsen, 1999).

The numerical codes that are used today have easy understanding structure. The outputs
of the models can be understood easily with minimum knowledge of the user. The main
problem at using these models is based on two points: a) model setup b) decision about
output files.

Model setup process is dominant for the results. The results that have been modeled is
completely created by the input data and the parameters. The inserted data are the
boundaries restrict the unknown data and the initial values for iteration start up.

The results that are generated by model can easily be viewed by user. The important
point at this process is that the results are expected or not. If the results aren’t realistic,
there must be a problem with model setup or numeric code.

230
 Numerical Modelling

Nowadays numerical methods may not include all the processes occuring at hydraulic
system. It is also possible to neglect some of the processes in numerical method with
personal expect.

To have better approximation the user needs to be aware of numerical methods

parameters such as courant number and peclet number.

The conversion of the continuous partial differential equation to its discrete algebraic
counterpart, the process of discretization can be done in various ways. Well-known
methods are:
• The Finite Difference Method (FDM): This method approximates the partial
differential equations using Taylor series expansions, thus creating a system of
algebraic equations.
• The Finite Volume Method (FVM): This method takes the integral form of the
conservation laws as a starting point and uses discrete approximations to the surface
and boundary integrals appearing in the integral formulation.
• The Finite Element Method (FEM): This method defines in each cell, in this
methodology called elements a set of basis functions and the solution is expanded in
terms of these basis functions. The unknown coefficients in this expansion become
the unknowns in FEM. Usually a weighted residual strategy is employed to set up
the system of algebraic equations (Gerritsma and Koren, 2000).

Before explaining the finite element method, some concepts of grids and mesh
generation for numerical models will be explained in next part.

2.1. Grids and mesh generation

An important part when working in the field of numerical models is concerned with the
generation of meshes. Numerical mesh generation arose from the need to compute
solutions to the partial differential equations of fluid mechanics on physical regions
with complex geometries.

Mesh generation method depends on the numerical method which will be used to solve
the domain. Therefore there are number of different mesh generation methods
appropriate to the using numerical method. For instance in finite difference numerical
method, the structured grid is generally used.

Assuming that the numerical method was chosen for the model, the domain geometry
must be defined with an appropriate grid method. The domain bounded by the object
and the outer boundaries is then covered by a so called mesh or grid (Figure 5).

Figure 5. Examples of triangular and rectangular mesh generation

231
Sediment Transport Technology

Grids can be classified according to several characteristics:

a) Shape
b) Orthogonality
c) Structure
d) Blocks
e) Position of variable
f) Grid movements

a) The shape of grid is usually triangular or quadrilateral (Figure 6).

b) The orthogonolity of element is defined with the angle between the links of
element. If the angle is 90’, the element is defined as orthogonal. Otherwise the
grid is defined as non-orthogonal (Figure 7).
c) The shape of grid should be defined as structured if it is possible to make two
dimensional arrays indexing the grid cells. If it is not possible then the grid will
be named as unstructured (Figure 8).
d) A structured grid is named as block grid. If more than two structured grid
generated, the grid will be named as multiblock grid (Figure 9).
e) If all the variables of cell or element are calculated in the same location such as
on link, or node, then the grid will be named as staggered grid. If the calculation
of variables is occured at different locations, grid will be named as unstaggered
(Figure 10) (Olsen, 1999).

Figure 6. Triangular and quadrilateral elements

Figure 7. Orthogonal and non-orthogonal grids

Figure 8. Structured and unstructured grids

232
 Numerical Modelling

Figure 9. Block and multi-block grids

Figure 10. Staggered and unstaggered grids.

Note that in finite element methods the variables can be calculated at nodes and
center.

f) If the grid moves during the iteration process, the grid will be named as
adaptive. Adaptive grids give better results as the grid regenerated with the
effect of water level, bed level changes.

It is well known that the error of approximations of the hosted equations depends not
only on the derivatives of the solution of the hosted equations and the grid spacing, but
also on the rate of change of grid spacing and the departure of the grid from
orthogonality. For a given grid spacing, smooth, orthogonal grids usually result in the
smallest error in simple problems. Thus another major goal of grid-generation
algorithms is to produce smooth grids that are grids where the spacing varies smoothly
and the angles between the grid lines do not become too small. It turns out that it is not
possible to generate orthogonal grids for a wide class of problems.

If the solution of the hosted equation varies rapidly in some part of the physical region,
then is reasonable to choose a finer grid in that part of the region to reduce the error in
the numerical simulation. Such a grid is called solution adapted; it is important to be
able to generate solution adapted grids.

The main advantage of modern grid generation algorithms is that these algorithms can
be used to efficiently generate large grids, containing tens of thousands to a few million

233
Sediment Transport Technology

points, and the resulting grids allow the reduction of error and simplified treatment of
boundary conditions.

As a summary for accuracy and convergency the quality of grid can be generated with
the following informations:
• Defining the necessary elements for grid. Element numbers is the most important
parameter at iteration process. If it is so high, iteration process will take much more
time.
• Drawing the geometry is important as the external boundaries define over it. The
inflow and outflow areas must be defined carefully. At these areas the flow direction
and the dispertion of concentration generally depend on these definitions.
• Grid lines should follow the stream lines as much as possible.
• Grid aspect ratio should not be greater than 2.
• Grid expansion ratio should not be greater than 1.5.
• Higher densities should be necessary at areas with higher velocities and
concentrations.
• Higher densities should be necessary at areas where water depth fast varies.

2.2. Finite element method

Primary advantage of finite element method is that it allows us to accurately model
irregular geometries.

The finite element method minimizes the error in the solution to the underlying
mathematical equations in a global sense whilst finite volume minimizes it in a local
sense. This means that a finite volume method will always conserve mass at each time
step and throughout a simulation.

A finite element method will only have true mass conservation once the grid is refined
to a level where further refinement makes no further change to the solution. In river
modeling this level is difficult to achieve as refinement of the mesh effectively redefines
the bathymetry and sets up a different problem and in work published so far there is
little evidence that mesh independence has been achieved.

A finite element method often uses an unstructured mesh. This means that the matrix A
is sparse. To solve the system of equations, the matrix needs to be inverted. This is a
fairly time-consuming procedure. A good solver is therefore essential in the finite
element method.

The finite element method gives similar problems with stability as the finite volume
method. Implicit solvers, high-order upstream schemes etc. are more difficult to
formulate with the finite element method than the finite volume method. This especially
applies to an unstructured grid, where it is difficult to derive an implicit high-order
upstream procedure.

Many finite element programs have employed various techniques to try to overcome
these problems, with varying degree of success (Wright, 2001).

234
 Numerical Modelling

2.3. Galerkin finite element method

The finite element method can be formulated in different ways. Often a weighed
residual method is used. Then the basic equations are multiplied with a weighting
function, W, and integrated over the element A.

Galerkin formulation is used if the weighting functions are equal to the interpolation
functions. The result of the integration is a residual, which is minimized. The equation
with the residual is called the residual equation.

A term often encountered in the finite element method is the weak formulation. This
means that when formulating the equation system, there are no restrictions that the
derivative of the unknown need to exist on the surface between elements, as opposed to
the original differential equations. The Galerkin formulation is a weak formulation. The
main effect of the weak formulation is that mass continuity need nnot be satisfied
automatically in the satisfied solution procedure (Olsen, 1999).

2.3.1. Finite element notation

In order to define the finite element notation, we need the triangular mesh element to
describe some procedures (Figure 11).

A, B, C : Nodes of triangular element

Ax, Ay : X, Y coordinates of nodes
∆ : Triangular element
∂∆ : Boundary of triangular element ∆
∂∆ A : Boundary edge BC, opposite link to A. Similarly ∂∆ B , ∂∆ C are the
boundary edges of appropriate nodes.
∆ : The area of the triangular element
n : General outward unit normal vector on boundary edge

A b
n

Figure 11. Triangular hydrodynamic element

235
Sediment Transport Technology

a, b, c : Outward normal vectors on boundary edges. The outward normal vector

between B and C is named as a, and equal to the distance between these
nodes. Similarly b and c are the outward normal vectors of appropriate
nodes.
a = (ax, ay) : Components of the outward normal vector in x, y directions.
a = ((B x − C x ), (B y − C y ))

Similarly the other outward normal vectors components depend on the

coordinates of the appropriate nodes.
z : The unit vector on z direction (Vatnaskil, 1998).

We need to define the element matrices:

1. The element divergence matrix;

⎡a x ay ⎤
⎛ 1 ⎞⎢ ⎥
N = ⎜ ⎟ ⎢bx by ⎥
⎝ 2 ⎠⎢
⎣c x c y ⎥⎦

2. The element matrix;

⎡2 1 1 ⎤
⎛ ∆ ⎞⎢
M = ⎜⎜ ⎟⎟ ⎢1 2 1 ⎥⎥
⎝ 12 ⎠ ⎢1 1 2⎥
⎣ ⎦

3. The lumped element matrix;

⎡1 0 0 ⎤
⎛ ∆ ⎞⎢
M 1 = ⎜⎜ ⎟⎟ ⎢0 1 0⎥⎥
⎝ 3 ⎠ ⎢0 0 1 ⎥
⎣ ⎦

4. The element x-stiffness matrix;

⎡ a x2 a x bx ax cx ⎤
⎛ 1 ⎞⎢ ⎥
K X = ⎜⎜ ⎟ a x bx
⎟⎢ bx2 bx c x ⎥
⎝ 4 ∆ ⎠ ⎢a c
⎣ x x bx c x c x2 ⎥⎦

5. The element y-stiffness matrix;

⎡ a y2 a y by aycy ⎤
⎛ 1 ⎞⎢ ⎥
K y = ⎜⎜ ⎟ a y by
⎟⎢ b 2
by c y ⎥
∆
y
⎝ 4 ⎠ ⎢a c
⎣ y y by c y c y2 ⎥⎦

236
 Numerical Modelling

6. The element stiffness matrix;

⎛1⎞
K = K x + K y = ⎜ ⎟ NN T
⎝∆⎠

7. The element x-derivative matrix;

⎡a x bx cx ⎤
⎛ 1 ⎞⎢
L x = −⎜ ⎟ ⎢a x bx c x ⎥⎥
⎝6⎠
⎢⎣a x bx c x ⎥⎦

8. The element y-derivative matrix;

⎡a y by cy ⎤
⎛ 1 ⎞⎢ ⎥
L y = −⎜ ⎟ ⎢ a y by cy ⎥
⎝ 6 ⎠⎢
⎣a y by c y ⎥⎦

The nodes and elements hydrodynamic properties are defined for unknowns with
averaged values, approximations:
a) Approximate water level at nodes A, B and C: h = (hA, hB, hC)
b) Approximate concentration or temperature at nodes A, B and C: c = (cA, cB, cC)
c) Approximate velocity within the element: V = (u,v)
d) Approximate velocity within the adjacent element that a point into: Va = (ua, va)
e) Average water level within the element: hm = 1/3(hA+hB+hC)
f) H, Hm, h and hm are defined similarly
g) Velocity of injected water: Vo = (uo, vo)
h) Wind stress vector: S = (Sx, Sy)
i) Decay vector s = (sa, sb, sc)

2.3.2. Flow model

Finite element weak formulation of continuity equation
The finite element weak formulation of continuity equation (Eq. 9) is given in Eq. 55. ψi
are the weight functions of nodes A, B, and C and taking the value 1 at node i and the
value 0 on the opposite edge (Vatnaskil, 1998).

∂H
∫ (HV )nψds − ∫∫ (HV )∇ψdxdy + ∫∫
∂∆ ∆ ∆
∂t
ψdxdy = ∫∫ Qψdxdy
∆
(55)

The resulting finite element approximation is given in Eq. 56.

∂H
q + H m NV + M =Q (56)
∂t

Here, Q is injected water and for i = A, B, C expressed in Eq. 57 and Eq. 58. q is the
boundary normal outflow and for i = A, B, C expressed in Eq. 59 and Eq. 60.

237
Sediment Transport Technology

Q = (Q A , QB , QC ) (57)
Qi = ∫∫ Qψ i dxdy (58)
∆

q = (q A , q B , qC ) (59)
qi = ∫ qψ i ds (60)
∂∆

Finite element weak formulations of momentum equations

The finite element weak formulation of momentum equations (Eqs. 26 and 27) are
given in Eq. 61 and Eq. 62 (Vatnaskil, 1998).

⎡ ∂u ∂u ∂u ⎤ ⎡1 ⎤
∫∫ ⎢⎣ ∂t + u ∂x + v ∂y ⎥⎦ψdxdy + ∫ ⎢ 2 (V + VA )a ⎥ (ua − u )ψds =
∆ ∂∆ a ⎣ ⎦
⎡ ∂n
∫∫ ⎢⎣− g ∂x − βu + fv + H S
1
x −
Q
(u − u0 )⎤⎥ψdxdy (61)
∆
H ⎦

⎡ ∂v ∂v ∂v ⎤ ⎡1 ⎤
∫∫∆ ⎣ ∂t ∂x ∂y ⎥⎦ψdxdy + ∂∆∫ ⎢⎣ 2 (V + V A )a ⎥⎦(va − v )ψds =
⎢ + u + v
a

⎡ ∂H 1 ⎤
− βv − fu + S y − (v − v0 )⎥ψdxdy
Q
∫∫ ⎢⎣− g
∆
∂y H H ⎦
(62)

The resulting finite element approximation of momentum equations is given in Eq. 63.

dV ⎡ 1 ⎤ ⎡ 1 ⎤
+ ⎢ (V + Va )a ⎥ (Va − V ) = gN T H − ∆ ⎢ βV + fkxV − S + (V − V0 )⎥
Q
∆ (63)
dt ⎣ 2 ⎦ ⎣ H H ⎦

2.3.3. Transport model

Finite element weak formulation of transport equation
The finite element weak formulation of transport equation (Eq. 40) is given in Eq. 64.
The resulting finite element approximation of trasnport equation is given in Eq. 65
(Vatnaskil, 1998).

⎛ ∂c ∂c ⎞ ⎛ ∂c ∂ψ ∂c ∂ψ ⎞ ∂c
∫ H ⎜⎜⎝ D
∂∆
x
∂x
, Dy ⎟⎟nψds − ∫∫ H ⎜⎜ Dx
∂y ⎠ ∆ ⎝ ∂x ∂x
+ Dy
∂y ∂y ⎠
⎟⎟dxdy − ∫∫ Hu ψdxdy =
∆ ∂x
⎛ ∂c ⎞
∫∫∆ ⎜⎝ H ∂t + s − Q(c0 − c )⎟⎠ψdxdy (64)

− P d − H m (D x K x + D y K y )c − H m uL x c = H m M + Ms − QM (c 0 − c )
dc
(65)
dt

Here Pd is the boundary normal mass outflow and for i = A, B, C expressed in Eq. 66
and Eq. 67.

238
 Numerical Modelling

(
P d = p Ad , p Bd , pCd ) (66)
pid = ∫ p ψ ds (67)
d
i
∂∆

2.3.4. Grid size and iteration time step adjustment

It is well known that the advective-dispersive transport equation is more difficult to
solve numerically. The problems are particularly severe when advection dominates over
dispersion. These numerical oscillations tend to be more severe as advection becomes
more dominant.

Peclet number
Relationship between the advective and diffusive components of solute transport is
expressed by Peclet number as the ratio of the product of the average interstitial velocity
and the characteristic length, divided by the dispersion coefficient (Eq. 68). Peclet
number characterizes the oscillations that may occur in the numerical solution when this
number is greater than 2. Small values indicate diffusion dominates; large values
indicate advection dominates (Vatnaskil, 1998).

V (∆l )
Pe = <2 (68)
D

Courant number
According to Courant criteria, the time step should be smaller that a water particle can
pass from one section to another. If Courant number is greater than 1, the iteration will
be unstable. Therefore the time step can be reviewed and adjusted as smaller time step.
Courant number is expressed in Eq. 69 below (Olsen, 2001).

V (∆t )
CN = <1 (69)
∆l

References

1. Dalziel, S. B., Holford, J. M., and Hunt, G. R., “Environmental Fluid Dynamics”,
Department of Applied Mathematics Theoretical Physics, The University of
Cambridge, Cambridge, England, (1999).
2. Gerritsma, M. I., and Koren, B., “Introduction to Computational Fluid Dynamics”,
(2000).
3. Güney, M. Ş., “Akışkanlar Mekaniği”, İzmir Dokuz Eylül University, Engineering
Faculty Publications No: 204, İzmir, (1995).
4. Hydrosoft, “Aquadyn Software Manual Book”, Hydrosoft Energie, Canada, (1999).
5. Olsen, N. R. B., “CFD Algortihms for Hydraulic Engineering”, Norwegian
University of Science and Technology, Department of Hydraulic and
Environmental Engineering, ISBN 82 7598-044-5, (2000).
6. Olsen, N. R. B., “Hydroinformitics Fluvial Hydraulics and Limnology”, Norwegian
University of Science and Technology, Department of Hydraulic and
Environmental Engineering, ISBN 82 7598-046-1, (2001).

239
Sediment Transport Technology

7. Orkwis, P. D., Volff, J. M., “Numerical Methods in Transport Phenomenon”,

Computational Fluid Dynamics, WSU, (2001).
8. Sander, J., “Dynamical Equations and Turbulent Closures in Geophysics”,
Continuum Mech. Termodyn., 10, 1-28, (1998).
9. Stewart, R. H., “Introduction to Physical Oceanography”, Department of
Oceanography, Texas A& M University, Texas, (1998).
10. Vatnaskil, “Aquasea Manual Book”, Vatnaskil Consulting Engineers, Island,
(1998).
11. Wright, N. G., “Conveyance Implifications for 2-D and 3-D Modelling”, Hr
Wallingford and the Environmental Agency, (2001).

240
 ESTIMATION AND FORECASTING OF DAILY SUSPENDED
SEDIMENT DATA BY MULTI LAYER PERCEPTRONS

Assoc. Prof. Dr. H. Kerem CIĞIZOĞLU

Hydraulics Division, Civil Engineering Faculty
Istanbul Technical University
Istanbul - Turkey
e-mail: cigiz@itu.edu.tr

1. Introduction
Sediment yield is defined as the total sediment outflow from a watershed measurable at
a point of reference during a specified period of time. The sediment outflow from the
watershed is induced by processes of detachment, transportation, and deposition of soil
materials by rainfall and runoff. Estimates of sediment yield are required in a wide
spectrum of problems such as design of reservoirs and dams, transport of sediment and
pollutants in rivers, lakes and estuaries, design of stable channels, dams and debris
basins, undertaking cleanup following floods, protection of fish and wildlife habitats,
determination of the effects of watershed management, and environmental impact
assessment [26]. Fine sediment has long been identified as an important vector for the
transport of nutrients and contaminants such as heavy metals and micro-organics.
Suspended sediment is important in its own right, since its presence or absence exerts an
important control on geomorphological and biological processes in rivers and estuaries.
Sediment yield Y(t) at a given point in space (say, watershed outlet) can be represented
as
_
Y (t ) = Y (t ) + ε (t ) (1)

_
in which Y (t ) is the mean value or deterministic component of Y(t), and ε(t) is the
deviation from or fluctuation around the mean value or stochastic component of Y(t).
The relative contribution Y (t ) or ε(t) to Y(t) depends on the watershed and space –time
scales. Clearly, Y(t) encompasses the full range of variability from being entirely
deterministic to being entirely stochastic. All sediment models are special cases of (1).

The deterministic models can be distinguished as empirical and conceptual. These

models usually require long data records, so that average annual sediment yield can be
determined. The conceptual models combine the mechanics of sediment transport with
empirical relationships. Both the empirical and conceptual models approximate the
physical processes controlling sediment yield [26].

Another way to represent the complex sediment behaviour is to interpret a sequence of

sediment yield measurements as random. If the processes governing sediment yield are
assumed to be stochastic and thus governed by the laws of probability the sediment
yield can be described by a stochastic process and associated probability distributions.

241
Sediment Transport Technology

The application of physics-based distributed process complex computer software

programs is often problematic, due to the use of idealised sedimentation components, or
the need for massive amounts of detailed spatial and temporal environmental data which
is not available.

Many of the available techniques for time series analysis assume linear relationships
among variables. In the real world, however temporal variations in data do not exhibit
simple regularities and are difficult to analyse and predict accurately. It seems necessary
that nonlinear methods such as artificial neural networks (ANNs), which are suited to
complex nonlinear models, be used for the analysis of real world temporal data.

There are numerous applications of ANNs in water resources. The rainfall-runoff

relationship is successfully modelled by ANNs [21, 24, 28]. Neural network
methodology was employed for river runoff forecasting [6, 7, 13]. ANNs were also
considered as a powerful tool to use in various groundwater problems. Other
applications of ANNs include unit hydrograph derivation, regional flood frequency
analysis, estimation of sanitary flows, regional drought analysis, classification of river
basins, agricultural vulnerability assessment of rural private wells, modeling hydraulic
characteristics of severe contraction [10, 15, 18, 19, 22, 23, 27]. In other studies, Auto
Regressive Moving Average (ARMA) models were incorporated into the training and
testing data sets of ANNs [8] and neural networks based streamflow data infilling
procedures were investigated by Khalil et al. [17] and Elshorbagy et al. [12]. The
application of ANNs to suspended sediment data estimation is limited in the literature
[1, 5]. The success with which ANNs have been used to model dynamic systems
suggests that the approach may prove to be an effective and efficient way to simulate
the complex river suspended sediment process in situations where explicit knowledge of
internal sub process is not required.

Within the whole study the term MLPs (Multi layer perceptrons) is preferred over the
general ANN definition because there are multiple ANN algorithms and MLPs are just
one of them. Although MLPs constitute the majority of the ANN applications to the
various water resources data so far, other algorithms such as Radial Basis Functions,
RBFs, [13, 20] and probabilistic neural networks, PNNs, [29] have also been employed
in some studies. It seems that in the future these algorithms will gain more significance.

In this paper, two issues are addressed, firstly, forecasting of suspended sediment
concentration of the downstream station using the past sediment observations. The
forecasting study was carried out in two steps; initially downstream sediment series
were employed as input and then the sediment records on the upstream constituted the
input layer of MLPs. The focus of the second MLP application was the investigation of
river flow-suspended sediment relation. Here the estimation of downstream sediment
data was accomplished with the help of downstream flows and upstream flows
separately. The MLP applications are compared with the conventional AR models and
regression models. In the study the term forecasting (prediction) is used for the model
applications with input and output data belonging to the same river station. The term
estimation is preferred when input and output data corresponds to different stations.

242
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

2. Multi –Layer Perceptrons (MLP) Network

Multi-layer perceptrons are feed-forward networks with one or more hidden layers. The
neural network employed in this study possessed a three-layer learning network
consisting of an input layer, a hidden layer and an output layer. Detailed information
about MLP are available in the literature [11].

Given a training set of input-output data, the most common learning rule for multi-layer
perceptrons is the back-propagation algorithm (BPA).Back propagation involves two
phases: a feed forward phase in which the external input information at the input nodes
is propagated forward to compute the output information signal at the output unit, and a
backward phase in which modifications to the connection strengths are made based on
the differences between the computed and observed information signals at the output
units [11]. It is shown that the incorporation of Marquardt algorithm into back
propagation algorithm speeds up the convergence [14]. In this study this algorithm is
employed during the training of MLPs. Tangent sigmoid function is used as neuron
transfer function.

3. Statistical Analysis of the Data

In this study daily mean flow and suspended sediment data of two gauged stations
belonging to Schuylkill River in Philadelphia are used. The data was downloaded from
the web site of United States Geological Survey (USGS). The data time period covers a
29 years long duration (1October 1952-30 September 1981). Related information for the
flow stations are presented in Table 1. Station 1470500 is upstream of station 1473800.
The daily statistical parameters (i.e. mean, x , standard deviation, sx, skewness
coefficient, csx, coefficient of variation, cv, overall maximum, xmax, and overall
minimum, xmin) of the common observed data are computed using the method of
moments (Table 1). A comparison of cumulative flow and suspended sediment amounts
between two stations does not show a deviation (Fig. 1).

Table 1. Related information for two measurement stations in Schuykill river (overall mean,
x , standard deviation, sx, coefficient of variation, cv, and the skewness coefficient, csx,
first order auto-correlation coefficient, r1, overall minimum, xmin , and overall
maximum, xmax) of the common observed daily flow and suspended sediment record
(1952-1981)

Basin
Station Data x max
River Location area x sx cv csx r1 xmin xmax
No. type x
(mi2)

Flow
1470500 Schuykill Upstream 355 19.67 25.47 1.30 7.61 0.78 1.5 736 647.60
(m3/s)

Flow Down-
1473800 Schuykill 1830 75.15 101.79 1.35 6.46 0.76 0.02 2640 610.90
(m3/s) stream

Sediment
1470500 Schuykill Upstream 355 72.73 941.76 12.95 36.60 0.37 0 47100 37.41
(tons/day)

Sediment Down-
1473800 Schuykill 1830 1064 10627 10.00 36.00 0.31 0 650000 35.13
(tons/day) stream

243
 Sediment Transport Technology
Downstream cumulative flow (10 m )

D owns tr eam cumulative s ediment

3

80000
12000
6

70000 y = 3.7736x + 748.01 y = 15.716x - 717.36

2 10000 2
60000 R = 0.9996 R = 0.9962

(1 0 tonnes )
50000 8000
40000
6000

3
30000
20000
4000
10000 2000
0 0
0

5000

10000

15000

20000

100

200

300

400

500

600

700

800
6 3 3
Upstream cumulative flow (10 m ) Upstream cumulative sediment (10 tonnes)

Figure 1. Comparison of cumulative flows and cumulative suspended sediment amounts

between upstream and downstream stations (1 October 1952-30 September 1981)

The statistics given in Table 1 show the difference between flow and suspended
sediment data clearly. The sediment data show a significantly more skewed distribution
(csx=37 and 36 for upstream and downstream, respectively) compared with flow series
(csx=7.6 and 6.5 for upstream and downstream, respectively). The ratio between
standard deviation and mean, cv, for suspended sediment is nearly 10 times of the
corresponding flow value. Finally, the maximum-mean ratios for sediment series are
quite high (648 and 611 for upstream- and downstream sediment series, respectively; 37
and 35 for upstream- and downstream flow series, respectively). Another interesting
characteristic of the daily suspended sediment series is their relatively lower first order
autocorrelations (r1=0.37 and 0.31) compared with flow records (r1=0.78 and 0.76). The
cross correlations between two different station data are presented in Table 2. The
highest correlation is between upstream flow and downstream flow (r=0.87) and the
lowest between downstream flow and upstream sediment (r=0.41). The geographical
description of the river drainage basins is not presented here because it not the concern
of the study.

4. Application of MLPs to the Data

A MATLAB code was written for the feed forward Marquardt-back propagation
algorithm. The application of the ANNs to time series data consisted of two steps. The
first step was the training of the neural networks. It comprised the presentation of daily
flow and suspended sediment data describing the input and output to the network and
obtaining the inter-connection weights. Once the training stage was completed the
MLPs were applied to the testing data. Determining an appropriate architecture of a
neural network for a particular problem is an important issue since the network topology

Table 2. Cross-correlation values for daily flow and suspended sediment values belonging to
upstream (up) and downstream (down) measurement stations in Schuykill River
(1952-1981)
Data pairs
Flow (up) Flow (down)- Sediment (up)- Flow (up)- Flow (up)- Flow (down) -
-Sediment -Sediment Sediment Flow (down) Sediment Sediment (up)
(up) (down) (down) (down)
Cross-
correlations 0.59 0.66 0.69 0.87 0.60 0.41

244
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

______ observed - - - - MLP forecasted 70000

190000 y = 0.2742x + 796.43
Suspended sediment (tons/day)

170000 2

F o r e c a s te d s e di m e n t (to n s / da y )
60000 R = 0.2386
150000
50000
130000
110000 40000
90000 30000
70000
20000
50000
30000 10000
10000 0
-10000 -10000 0 50000 100000 150000 200000
0 200 400 600 800 1000 1200
Time (days) Observed sediment (tons/day)

Figure 2. Forecasting of downstream suspended sediment series by MLPs using the past
downstream sediment values as input (no periodic component is included in input
layer)

directly affects its computational complexity and its generalization capability. The
number of hidden layers and the number of the nodes in the input and hidden layers
were determined after trying various network structures. The network structure
providing the best training result, i.e. the lowest mean square error, MSE, was also
employed for the testing stage. Before training and testing the data were scaled using
the extremes with the formula (xmax-x)/(xmax-xmin), where xmax and xmin represent the
overall maximum and minimum of the whole data set in training and testing stages. The
outputs were rescaled then with the formula [xmax-y(xmax-xmin)] where y corresponds to
the initial output value after testing.

The performance of MLPs was tested for two different applications: suspended
sediment forecasting on downstream (predictions one-day ahead) using the downstream
and upstream sediment data as input separately (Forecasting I and Forecasting II). The
forecasting of downstream flow series was also presented in order to show the MLP
performance difference between two different types of time series. The second
application was about the downstream sediment estimation using flow data. Here again
the estimation study was carried first using downstream flow and then upstream flow as
input. The time periods of the data selected for training and testing are presented in
Table 3. The ANN prediction and estimation results were compared with those of the
stochastic (AR model) and statistical methods (multi linear regression and sediment
rating curve). The performance evaluation measures were the mean square error (MSE)
and the coefficient of determination (R2) between forecasted (or estimated) and
observed suspended sediment values.

Table 3. Training and testing periods for different MLP applications

Study type

Forecasting I Forecasting II Estimation I Estimation 2 Hystheresis

18.10.1952- 18.10.1952- 18.10.1952- 18.10.1952- 18.10.1952-

Training
21.10.1978 21.10.1978 21.10.1978 21.10.1978 21.10.1978
22.10.1978- 22.10.1978- 22.10.1978- 22.10.1978- 14.10.1979-
Testing
17.09.1981 17.09.1981 17.09.1981 17.09.1981 11-04.1980

245
Sediment Transport Technology

4.1. Suspended sediment forecasting using the downstream data

The first set of experiments was made for one-day-ahead prediction of the sediment
values belonging to downstream station. The downstream sediment data has an
autocorrelation value equal to 0.31. The simulations during the training and testing
stages showed that a network structure with one hidden layer having 5 nodes, and 4
input nodes in the input layer provided lowest MSE and highest R2 to forecast the
unique value in the output layer. The input layer nodes represented the four previous
suspended sediment amounts (times t-1,…, t-4) and the output layer node corresponded
to the sediment at time t.

The predictions for the testing period are compared with the observed sediment values
in the form of sediment hydrograph and scatter plot in Fig.2. The scatter plot exhibits
deviations from observed sediment values (R2=0.24, MSE=35x106 tons2/day2). Neural
network structures with 4 and 8 input nodes, respectively, provided higher MSE values
(38x106 tons2/day2, and, 37x106 tons2/day2, respectively). The observed high sediment
parts are underestimated and the low and medium parts are either over-or under
estimated. In order to see the effect of periodicity an additional input representing the
day of the year for the output sediment value is incorporated into the input layer. This
modification slightly increased the forecasting performance of the network (R2=0.27,
MSE=34x106 tons2/day2, Fig. 3, and Table 4, Column: Forecasting I).

The study was extended by applying stochastic models to the testing data following a
model calibration using the sediment values in the training period of MLPs. These
models were the well known autoregressive models; AR(4) model was preferred
because the employed MLP input layer consisted of four previous sediment
measurements of the same station. The performance of AR (4) model is illustrated in
Fig. 4. The comparison of observed values and AR (4) forecasts in Fig. 4 illustrates a
significantly high deviation from observed sediment hydrograph with a quite low
R2=0.07 and higher MSE=42*106 tons2/day2. It is obvious that the MLP provides a
better performance to the stochastic model in sediment forecasting. Both MLP and
AR(4) model provided a few negative forecasts. The number of the negative forecasts is

90000 y = 0.3519x + 748.06

______ observed - - - - MLP forecasted R2 = 0.2674
190000 80000
Fore caste d se di m e n t (ton s/day)

170000 70000
S u sp e n d e d se d im e n t (to n s/d a y )

150000
60000
130000
50000
110000
90000 40000
70000 30000
50000 20000
30000 10000
10000
0
-10000
0 200 400 600 800 1000 1200 -10000 0 50000 100000 150000 200000
Time (days) Observed sediment (tons/day)

Figure 3. Forecasting of downstream suspended sediment series by MLPs using the past
downstream sediment values as input (periodic component is included in input
layer)

246
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

Table 4. Performance comparison of different methods in terms of mean squared error, MSE,
and coefficient of determination, R2, for the testing period

Study type

Method Forecasting I Forecasting II Estimation I Estimation II

MSE 34x106 22x106 3.9x106

7.3x106
MLP (tons2/day2)
R2 0.27 0.50 0.91 0.84
MSE
- - 36x106 34x106
SRC (tons2/day2)
R2 - - 0.85 0.81
MSE -
26x106 - -
MLR (tons2/day2)
R2 - 0.43 - -
MSE
42x106 - - -
AR(4) (tons2/day2)
R2 0.07 - - -

______ observed - - - - ANN forecasted

200000 70000
Fo r e ca s te d s e di m e n t (to n s /da y )

180000
S u s pe n de d s e di m e n t (to n s /da y )

60000 y = 0.0873x + 181.98

160000
R2 = 0.073
140000 50000
120000
40000
100000
80000 30000
60000 20000
40000
20000 10000
0 0
-20000 0 200 400 600 800 1000 1200
-10000 0 50000 100000 150000 200000
Time (days)
Observed sediment (tons/day)

Figure 4. Forecasting of downstream suspended sediment series by AR(4) models using the
past downstream sediment values

16 for MLP, (lowest value = -1500 tons/day) and 18 for AR(4) model (lowest value =
-3400 tons/day). This might be because the MLP structure is trained using the whole
sediment series in the training data set including extremely low and high sediment
values (xmax/ x = 611 for downstream sediment series) and some underestimations in
low values (negative values) arise as the network faces confusion in transition between
these two extreme zones.

The sediment forecasting performance of MLPs is then compared with river flow
forecasting for the same downstream station. An input layer with 10 previous flows and
6 hidden layer nodes provided the best performance (R2 = 0.51; 5886 m6/s2). It is seen

247
Sediment Transport Technology

that the daily flow forecasting is more successful providing an R2 nearly 100% higher
than the one obtained for the corresponding sediment data. The deviations (under- and
over forecasts) from the observed river flow hydrograph are not as high as in the
sediment case (Fig. 5). This result was expected because the autocorrelation of the daily
downstream data (r1=0.76) is noticeably higher than the sediment series of the same
station (r1=0.31) and it is well known that forecasting performance increases parallel to
an increase in persistence.

4.2. Suspended sediment forecasting using the upstream data

Differing from the low autocorrelations of the upstream and downstream sediment
series (0.37 and 0.31, respectively) the cross-correlation between these two series is
significantly higher (r=0.69). Inspiring from this fact the forecasting study is extended
by using the upstream sediment data in the MLP input structure to forecast the
downstream sediment value in the output layer. An MLP structure with 10 upstream
sediment measurements covering a time interval of 10 days (times: t-9,…., t) plus a
periodicity component, 6 hidden layer nodes and a unique downstream value (time: t) as
output provided highest R2 and lowest MSE values (0.50 and 22*106 tons/day2,
respectively, Table 4, Column: Forecasting II). The forecasting results in Fig. 6 show
that deviations from low and high flows are relatively lower compared with the previous
application. As explained before the main reason for the significantly better
performance of MLP using upstream sediment data is that the cross-correlation between
both sediment records is higher than individual auto-correlations. The other reason is
the incorporation of upstream sediment value corresponding to the same time (time: t)
of the unique output value into the input layer.

Because the input and output data belonged to different stations, the comparison with
MLPs was conducted this time with multi-linear regression (MLR) instead of an ARMA
model. The calibration period of the regression relation was the same of the MLPs and
similarly 10 upstream sediment measurements were the independent variables with the
unique downstream sediment value as dependent variable. Regression performance was
lower both in terms of R2 and MSE (0.43 and 26x106 tons2/day2, respectively, Table 4,
Column: Forecasting II).

2000 ______ observed - - - - MLP forecasted y = 0.5591x + 28.582

1200 2
1800 R = 0.5145
1600 1000
1400
F o r e c a ste d flo w (m /s)

800
1200
3
F lo w (m /s)
3

1000 600
800
600 400
400
200
200
0 0
-200 0 200 400 600 800 1000 1200 0 500 1000 1500 2000
-200
Time (days)
3
Observed flow (m /s)
Figure 5. Forecasting of downstream flow series by MLPs using the past downstream flow
values as input (periodic component is included in input layer)

248
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

190000 70000
y = 0.491x + 859.02
170000 ______ observed - - - - ANN forecasted 2
S u s pe n de d s e di m e n t (to n s /da y )

60000 R = 0.4969
150000

Forecas ted s ediment (tons /day)

130000 50000
110000 40000
90000
30000
70000
50000 20000
30000
10000
10000
-10000 0
0 200 400 600 800 1000 1200 0 50000 100000 150000 200000
-10000
Time (days) Observed sediment (tons/day)

Figure 6. Forecasting of downstream suspended sediment series by MLPs using the upstream
sediment values as input (periodic component is included in input layer)

70000 y = 0.3432x + 438.49

2
R = 0.4332
190000 ______ observed - - - - forecasted by multi-linear regression 60000
170000
S u sp e n d e d se d im e n t ( t o n s/d a y )

150000 50000
Forecasted sediment (tons/day)

130000
40000
110000
90000 30000

70000
20000
50000
30000 10000

10000
0
-10000 0 50000 100000 150000 200000
0 200 400 600 800 1000 1200
-10000
Time (days) Observed sediment (tons/day)

Figure 7. Forecasting of downstream suspended sediment series by MLR using the upstream
sediment values as independent variables

4.3. Suspended sediment estimation using the downstream flow data

The relation between suspended sediment concentration and river flow is represented
generally in the literature with the sediment rating curve (SRC) as mentioned by [9],

C = a Qb + ε (2)

where C and Q represent sediment concentration and river flow, respectively, whereas a
and b are coefficients and ε is the error term. This non linear relationship can be
converted into a regression form taking logarithms of both parts:

x = d + f y + ε* (3)

where x and y represent the logged values of sediment concentration and river flow,
respectively, whereas d and f are coefficients and ε* is the error term.

249
Sediment Transport Technology

The focus of this part of the study was the performance comparison between MLPs and
the conventional sediment rating curve in suspended sediment estimation. This time the
neural network structure consisted of an input layer with downstream flows and an
output layer with a unique downstream sediment value. The best performance criteria
were obtained for 6 input nodes (5 previous flow values, for times t,..., t-4; and a
periodic component) and 4 hidden nodes. The unique output node represented the
downstream suspended sediment value for time t. The results illustrated in Fig. 8 show
that the estimated values are close to the observed ones. In the scatter plot (Fig. 8) we
do not observe a significant deviation from the recorded sediment values in all parts of
the testing series (R2 = 0.91; MSE = 3.9x106 tons2/day2; Table 4; Column: Estimation I).
The sediment rating curve (SRC), in contrast, significantly underestimated the observed
sediment record (Fig. 9). The performance criteria of SRC were poorer compared with
MLPs (R2 = 0.85; MSE = 36x106 tons2/day2). Similar to the forecasting simulations the
MLPs produced some negative estimates. In SRC application the zero values in the
calibration period were initially replaced with positive values close to zero and because
of the logarithmic transformation none of the estimates were negative.

190000 ______ observed - - - - MLP estimated y = 0.9623x + 118.18

190000
2
S u sp e n d e d se d im e n t ( t o n s/d a y )

170000 R = 0.9136
Es ti m a te d s e di m e n t (to n s / da y )

170000
150000 150000
130000 130000
110000 110000
90000 90000
70000 70000
50000 50000
30000 30000
10000 10000
-10000 -10000
0 200 400 600 800 1000 1200 0 50000 100000 150000 200000
Time (days) Observed sediment (tons/day)

Figure 8. Estimation of downstream suspended sediment series by MLPs using the downstream
flow values as input (periodic component is included in input layer)

200000 ______ observed - - - - sediment rating curve 20000

y = 0.1016x + 130.51
S us pended s ediment (tons /day)

180000 18000 2
Es timated s ediment (tons /day)

R = 0.8526
160000 16000
140000 14000
120000 12000
100000 10000
80000 8000
60000 6000
40000 4000
20000 2000
0
0
0 200 400 600 800 1000 1200 0 50000 100000 150000 200000
Time (days) Observed sediment (tons/day)

Figure 9. Estimation of downstream suspended sediment series by SRC using the downstream
flow value of the same day as independent variable

250
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

An additional evaluation criteria, total sediment amount of the whole testing period, was
also considered for comparison since in the reservoir management studies the
accumulated sediment total (especially annual sum) plays an important role. As well
known the sediment accumulation within a water reservoir is one of the major
parameters determining the life of the reservoir. SRC provided a sediment total 74%
lower than the observed series for the considered testing period. The respective MLP
estimation was 10% higher than the observed one. This dominant superiority of MLPs
over the SRC shows that the complex non-linear relation between river flow and
suspended sediment is well captured by MLPs and they should be considered as a viable
alternative to the conventional SRC. This is a significant conclusion since the sediment
data may be absent for some time intervals and a rough estimation would really be of
great importance.

4.4. Suspended sediment estimation using the upstream flow data

It is a frequent procedure in water resources engineering to infill the missing
downstream data using the upstream records. The existence of a strong correlation
between downstream and upstream data can be useful in forecasting the sediment series
in the downstream sections (or vice versa) of the river. The correlation between
upstream flow and downstream suspended sediment was found as 0.60 (Table 2). This
part of the study comprised the estimation of sediment values on the downstream by
using the upstream flow data. A MLP structure with 10 flow values (times t, t-1,…, t-9)
and a periodic component in input layer and 6 hidden layer nodes provided lowest MSE
(7.2x106tons2/day2) and highest R2 (0.84). The majority of the observed sediment peaks
were closely approximated by MLPs (Fig. 10). Some of the local peaks, in contrast,
were either over- or under estimated. The MLP estimate for the maximum sediment
value of the testing period was 12% below the original value. The corresponding SRC
significantly underestimated the observed sediment record (Fig. 11). This fact was
reflected with a quite high MSE (34x106 tons2/day2) and lower R2 (0.81). The SRC
underestimated the maximum sediment value by 86%. The MLP and SRC estimated
sediment totals for the considered testing time interval were 22% higher and 70% lower,
respectively, than the recorded sum.

190000 ______ observed - - - - M LP forecasted

170000
170000 y = 0.817x + 332.32
150000 2
R = 0.8362
150000
130000
Suspended sediment (tons/day)

Estimated sediment (tons/day)

130000
110000
110000
90000
90000
70000
70000
50000
50000
30000
30000

10000 10000

-10000 0 -10000 0 50000 100000 150000 200000

200 400 600 800 1000 1200
Time (days) Observed sediment (tons/day)

Figure 10. Estimation of downstream suspended sediment series by MLPs using the upstream
flow values as input (periodic component is included in input layer)

251
Sediment Transport Technology

200000 ______ observed - - - - sediment rating curve 30000

S u s pe n de d s e dim e n t (ton s /day)

180000

Es timated s ediment (tons /day)

y = 0.1268x + 144.4
25000
160000 2
R = 0.8129
140000 20000
120000
100000 15000
80000
60000 10000
40000
20000 5000
0
0
0 200 400 600 800 1000 1200 0 50000 100000 150000 200000
Time (days) Observed sediment (tons/day)

Figure 11. Estimation of downstream suspended sediment series by SRC using the upstream
flow value of the same day as independent variable

As explained before there is a complex relation between the river flow and suspended
sediment. On the flow versus sediment plots, after some turning points a decrease in the
sediment values is observed with the increasing flows. This characteristic is known as
hystheresis and the MLP performance in capturing this phenomenon was investigated
by plotting MLP and SDC sediment estimates versus the observed flows (Fig. 12). The
estimates in this plot were some of the values presented in Figs. 10 and 11 covering a
time interval of 159 days (Table 3; Column: Hystheresis). This illustration clearly
depicts that the MLP series provided a hystheresis though slightly different than the
observed one. The MLP hystheresis approximated the shape of the historical curve in
general but there are under- and over estimations. In contrast, the SRC was completely
unsuccessful in hystheresis regeneration (Fig. 12). The SRC curve did not provide a
decrease after the turning point and the whole curve underestimated the historical one.

5. Summary and Conclusions

The objective of this paper was to investigate the performance of multi layer perceptron
(MLP) neural network model in river suspended sediment estimation and forecasting. It
was seen that in downstream sediment forecasting, the employment of upstream
Suspended sediment (tons/day)

7000 observed ∆ ANN

sediment rating curve
6000
5000
4000
3000
2000
1000
0
0 10 20 30 40 50 60 70 80
3
Flow (m /s)

Figure 12. The hystheresis regeneration by MLPs and SRC

252
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

sediment series as input instead of past downstream records provided more satisfactory
results (R2 value increased nearly 100% and MSE decreased 35%). This result was
expected because the cross-correlation between upstream and downstream sediment
series (r=0.69) was much higher than the auto-correlation of downstream sediment
record (r1=0.31). The conventional AR and multi linear regression (MLR) models
provided poorer performance criteria underestimating sediment peaks significantly.

The estimation of sediment data with the help of either downstream or upstream river
flow series was the next concern of the study. Here the comparison was made with
classical sediment rating curve (SRC) and MLPs provided nearly 9 and 5 times lower
MSE values for downstream and upstream flow input cases, respectively. This
appealing superiority of MLPs was especially due to their closer estimations for local
sediment peaks. In both MLP performances under- or over estimations (or predictions)
were observed but these differences did not exceed the 10% or 20% error limits.

The superiority of MLPs over conventional methods in the forecasting and estimation
river suspended sediment series can be attributed to the capability of the MLPs to
capture the non-linear dynamics and generalize the structure of the whole data set.
Neurocomputing offers a flexible alternative and standard software can be used to
construct intricate multi-purpose non-linear solutions. The method has no limitations in
the form of fixed assumptions or formal constraints. The neural network has a
distributed processing structure. Each individual processing unit or the weighted
connection between two units is responsible for one small part of the input output
mapping system. Therefore each component has no more than a marginal influence with
respect to the complete solution and that the mechanism will still function and generate
reasonable mappings.

As concluding remark this study showed that MLPs could be successfully employed to
simulate the complex river suspended sediment process in situations where explicit
knowledge of internal sub process is not required. Considering the complex non-linear
nature of the suspended sediment time series, as reflected in their statistics, the
utilization of the MLPs in water resources projects, where sediment estimation and
forecasting is required, would be of great benefit. The application of ANN algorithms
other than MLPs to the sediment records could provide improvement in performance
criteria.

Acknowledgements

The author would like to present his thanks to Dr. Sue White from Cranfield University,
UK, and Dr. Robert Abrahart from Greenwich University, UK, for their valuable
comments and to United States of Geological Survey (USGS) because of their free
published sediment and flow data in internet.

References

1. Abrahart, R. J., White, S. M., “Modelling Sediment Transfer in Malawi: Comparing

Backpropagation Neural Network Solutions against a multiple linear regression
benchmark using small data sets”. Phys. Chem.Earth (B). 26(1), 19-24, (2001).

253
Sediment Transport Technology

2. ASCE Task Committee, “Artificial neural networks in Hydrology I”, Journal of

Hydrologic Engineering, ASCE, 5 (2), 115-123, (2000).
3. ASCE Task Committee, “Artificial neural networks in Hydrology II”, Journal of
Hydrologic Engineering, ASCE, 5(2), 124-132, (2000).
4. Box, G. E. P., Jenkins, G. M., “Time Series Analysis, Forecasting and Control”,
Holden Day Inc., San Francisco, California, 552 pp., (1970).
5. Cigizoglu, H. K., “Filling missing suspended sediment data by artificial neural
networks”, XIV. International Conference on Computational Methods in Water
Resources, Delft, Netherlands, pp.1645-1652, (2002).
6. Cigizoglu, H. K., “Intermittent river flow forecasting by artificial neural networks”,
XIV. International Conference on Computational Methods in Water Resources,
Delft, Netherlands, pp. 1653-1660, (2002).
7. Cigizoglu, H. K., “Estimation, forecasting and extrapolation of flow data by
artificial neural networks”, Hydrological Sciences Journal, 48(3), 349-361,
(2003a).
8. Cigizoglu, H. K., “Incorporation of ARMA models into flow forecasting by
artificial neural networks”, Environmetrics, 14(4), 417-427, (2003b).
9. Clarke, R. T., “Statistical Modelling in Hydrology”, John Wiley & Sons, 412 pp.,
(1994).
10. Djebbar, Y., Alila, Y., “Neural network estimation of sanitary flows”. In:
Hydroinformatics Conference (Poster presentations) (ed. by V. Babovic & C. L.
Larsen) (Proc. Third Int. Conf. on Hydroinformatics, Copenhagen, Denmark),
A.A.Balkema, Rotterdam, The Netherlands, vol. 2, pp.764-770, (1998).
11. Eberhart, R.C., Dobbins, R.W., “Neural network PC tools: a practical Guide”,
Academic Press, San Diego, 414 pp., (1990).
12. Elshorbagy, A., Simonovic, S.P., Panu, U.S., “Estimation of missing streamflow
data using principles of chaos theory”, Journal of Hydrology, 255, 123-133,
(2002).
13. Fernando, D. A. K., Jayawardena, A. W., “Runoff forecasting using RBF networks
with OLS algorithm”, Journal of Hydrologic Engineering, ASCE, 3 (3), 203-209,
(1998).
14. Hagan, M. T., Menhaj, M. B., “Training Feedforward Networks with the
Marquardt Algorithm”, IEEE Transactions on Neural Networks, 5 (6), 989-993,
(1994).
15. Hall, M. J., Minns, A. W., “Regional flood frequency analysis using artificial
neural networks”, In: Hydroinformatics Conference (ed. by V. Babovic & C. L.
Larsen) (Proc. Third Int. Conf. on Hydroinformatics, Copenhagen, Denmark),
A.A.Balkema, Rotterdam, The Netherlands, vol.2, pp. 759–763, (1998).
16. Holder, L., “Multiple Regression in Hydrology”, Institute of Hydrology,
Wallingford. 285 pp., (1985).
17. Khalil, M., Panu, U. S., Lennox, W. C., “Groups and neural networks based
streamflow data infilling procedures”, Journal of Hydrology, 241, 153-176,
(2001).
18. Kheir El-din, K. A., “Neural network application for modelling hydraulic
characteristics of severe contraction”. In: Hydroinformatics Conference (ed. by V.
19. Babovic & C. L. Larsen) (Proc. Third Int. Conf. on Hydroinformatics,
Copenhagen, Denmark), A.A.Balkema, Rotterdam, The Netherlands, vol. 2,
pp.771–775, (1998).

254
 Estimation and Forecasting Daily Suspended Sediment Data By Multi Layer Perceptrons

20. Lange, N., “Advantages of unit hydrograph derivation by neural networks”. In:
Hydroinformatics Conference (ed. by V. Babovic & C. L. Larsen) (Proc. Third
Int. Conf. on Hydroinformatics, Copenhagen, Denmark) A.A.Balkema, Rotterdam,
The Netherlands, vol. 2, pp.783–789, (1998).
21. Mason, J. C., Price, R. K., Tem`me, A., “A neural network model of rainfall-runoff
using radial basis functions”, Journal of Hydraulic Research, 34(4), pp. 537-548,
(1996).
22. Minns, A. W., Hall, M. J., “Artificial neural networks as rainfall runoff models”,
Hydrological Sciences Journal, 41 (3), pp. 399-417, (1996).
23. Ranjithan, S., Eheart, J. W., Garrett, J. H., “Neural network-based screening for
groundwater reclamation under uncertainity”, Water Resources Research, 29(3),
pp. 563-574, (1993).
24. Ray, C., Klindworth, K. K., “Neural networks for agricultural vulnerability
assessment of rural private wells”, Journal of Hydrologic Engineering, ASCE,
5(2), pp. 162-171, (2000).
25. Sajikumar, N., Thandaveswara, B. S., “A non-linear rainfall–runoff model using an
artificial neural network”, Journal of Hydrology, 216, pp.32-55, (1999).
26. Shin, H. S., Salas, J. D., “Regional drought analysis based on neural networks”,
Journal of Hydrologic Engineering, ASCE, 5(2), pp. 145-155, (2000).
27. Singh, V. P., Krstanovic, P. F., Lane, L. J., “Stochastic Models of Sediment Yield,
Modeling Geomorphological Systems”, Chapter 9. Edited by Anderson M.G.,
John Wiley & Sons Ltd, pp. 272-286, (1988).
28. Thandaveswara, B. S., Sajikumar, N., “Clasification of river basins using artificial
neural network”, Journal of Hydrologic Engineering, ASCE, 5(3), pp. 290-298,
(2000).
29. Tokar, A. S., Johnson, P. A., “Rainfall-runoff modelling using artificial neural
networks”, Journal of Hydrologic Engineering, ASCE, 4(3), 232-239, (1999).
30. Torfs, P., Wojcik, R., “Local probabilistic neural networks in hydrology”. Phys.
Chem.Earth (B), 26(1), pp. 9-14, (2001).

Appendix A. Formulation of Some Models

Multiple Linear Regression Model

If y is a dependent variable and x1, x2, …, xi are independent variables, then the basic
model is given by [16]

y = a + b1 x1 + b2 x2 + ....... + bi xi + e

where

a, bi : constants
e : random variable

AR Model
The AR model of order p, i.e., AR(p), is given by [4]

255
Sediment Transport Technology

y t = φ1 y t −1 + φ 2 y t −2 + ....... + φ p yt − p + et

where

yt : variable at time t of the sequence of zero mean and unit variance

yt-1, yt-2, …, yt-p : previous values of the variable yt
et : random number at tth time
φ1, φ2, …, φp : autoregressive parameters

256
 COASTAL HYDRODYNAMICS AND SEDIMENT TRANSPORT

Prof. Dr. Yalçın YÜKSEL and Prof. Dr. Esin ÇEVİK

Department of Civil Engineering
Yıldız Technical University
80750 Yıldız, Istanbul - Turkey

1. Introduction
One of the little known truths about erosion in the world is that much of it is man-made!
We’ve destroyed our own beaches. While there are natural reasons for beaches to erode,
most of the worst problems are not natural (Fig. 1). Scientists have found that an awful
lot of beach erosion is caused by works of man such as ship channels, coastal structures
including jetties and groins, dams and flood control works on the beaches, and mining
of beach sand. Any blocking and intercepting sand moving along a coast causes terrible
beach erosion in the world (Douglass, 2002).

Sediment transport plays very important role in many aspects of coastal, estuarine and
offshore engineering. The movement of sand influences; the constructions of
economically viable harbors, the constructions of coastal power stations and refineries,
coastal flood defense, the loss or growth of amenity beaches, the safety of offshore
platforms and pipelines and many other applications.

Sediment transport may be occurred by currents, or by waves, or commonly by both

currents and waves acting together. Movement takes place as a result of friction exerted
on the seabed by the current and/or waves, with turbulent diffusion possibly carrying

Figure 1. Beach nourishment in Ocean City, New Jersey. The left picture shows erosion in
1986 and the right picture shows the beach after beach nourishment in 2001, these
photos taken from Douglass, 2002

257
Sediment Transport Technology

sediment particles up into suspension.

Before describing the process involved in sediment motion, it is necessary to define the
terms used for example beach, shore, etc. (Velden, 1989), (Fig. 2).

Beach: The zone of unconsolidated material that extends landward from the low water
line to the place where there is a marked change in physiographic form (start of cliff or
dunes). The sea ward limit of a beach is the mean low water (the intersection of any
standard low tide datum plane with the shore).

Coast: A strip of land of indefinite width (may be several kilometers) that extends from
the shoreline inland to the first major change in terrain features.

Coastal area: The land and area bordering the shoreline.

Onshore: A direction landward from the sea.

Offshore:
1. In beach terminology, the comparatively flat zone of variable width, extending
from the breaker zone to the seaward edge of the continental shelf.
2. A direction seaward from the shore.

Shore: The narrow strip of land in immediate contact with the sea, including the zone
between high and low water lines. A shore of unconsolidated material is usually called a
beach.

Sediment transport takes place by grains rolling, hopping and sliding along the bed in
response to the friction, and, in the case of sloping beds, gravity. This known as bed
load transport, and is the dominant mode of transport for slow flows and/ or large
grains. If the flow is fast enough and the grains fine enough, sand will be put into
suspension up to a height of several meters above the bed, and carried by currents. This
mode of transport is known as suspended load and is often much greater than the bed
load transport. In typical marine and estuarine situations, the predominant mode of
transport is bed load for grains coarser than about 2 mm, and suspended load for grains
finer than about 0.2 mm (Soulsby, 1997).

Figure 2. Definition sketch (Velden, 1989)

258
 Coastal Hydrodynamics and Sediment Transport

The sediment transport rate is defined as the amount of sediment per unit time passing
through a vertical plane of unit width perpendicular to the flow direction.

Erosion and deposition may be predicted in a site;

1. Calculate the hydrodynamics distributions of currents and waves, and dependent
quantities such as bed shear stress (or bottom friction)
2. Calculate the resulting distribution of sediment transport rates
3. Calculate the distribution of erosion of or deposition rates.

2. Properties of Water and Sand

2.1. Water properties
The density of seawater decreases with temperature and increases with salinity. The
density of seawater can be determined from the following formula (Van Rijn, 1989):

ρ = 1000 + 1.455CL − 0.0065(Te − 4 + 0.4CL )2 (1)

where CL; chlorinity (‰), Te; temperature(oC), and ρ; sea water density (kg/m3). The
chlorinity follows from;

S = 0.03 + 1.805CL (2)

where S; salinity(‰).

The kinematic viscosity coefficient can be approximated by:

[ ]
ν = 1.14 − 0.031(Te − 15) + 0.00068(Te − 15)2 ⋅ 10 −6 (3)

The viscosity coefficient is influenced by the sediment particles. Einstein found;

ν e = ν ⋅ (1 + 2.5C ) (4)

where, νe; effective kinematic viscosity of water-sediment mixture, ν; kinematic

viscosity of water, C; volumetric sediment concentration, based on experiments with
volumetric concentrations in range of 0.1-0.6.

2.2. Sediment properties

Generally the Wenthworth scale is used for the classification of sediment grains (Fig. 3)
in which the φ-scale defined by:

φ = − log 2 d
d = 2 −φ

The degree of sorting of a sediment is defined by geometric standard deviation;

d 84
σg = (5)
d 16

259
Sediment Transport Technology

Figure 3. Conversion chart from φ units to microns and mm (Wenthwort grain size scale)

• If a sediment sample has d84/d16<2 or d90/d10<2.4, then it is well sorted,

• If it has d84/d16>16 or d90/d10>35, then it is well mixed.

2.3. Sediment and water mixtures

Sediment and water mixtures occur in ratios ranging from that of a settled immobile
sand bed, through slurries and suspensions, to clear water. Some definitions are given as
follows:
C : volume concentration = volume of grains/volume of mixture
CM : mass concentration = mass of grains/volume of mixture
ε : porosity = volume of water/volume of mixture
VR : voids ratio = volume of water/volume of grains
ρB : bulk (suspension) density = mass of mixture/volume of mixture

The conservation formulae these five are given in Table 1.

Another definition is the angle of repose φi (angle of internal friction) is the angle to the
horizontal at which grains start to roll on a flat bed of sediment which is gradually tilted
from the horizontal (for sand is 320). The angle of final repose φr is the angle of the
slope avalanching has ceased (for sand is 280).

260
 Coastal Hydrodynamics and Sediment Transport

Table 1. Conversion between measures for sediment and water (Soulsby, 1997)

C CM ε VR ρB
1 ρB − ρ
C - CM/ρs 1-ε
1 + VR ρs − ρ
ρs ρ s (ρ B − ρ )
CM ρsC - ρs (1-ε)
1 + VR ρs − ρ
VR ρs − ρB
ε 1-C 1-CM/ρs -
1 + VR ρs − ρ
1− C ρ s − CM ε ρs − ρB
VR -
C CM 1− ε ρB − ρ
⎛ ρs − ρ ⎞ ρVR + ρ s
ρB ρ (1-C)+ ρsC ρ + C M ⎜⎜ ⎟⎟ ρ ε+ρs(1-ε) -
⎝ ρs ⎠ 1 + VR

2.4. Permeability and liquefaction

The flow of water through a porous medium, such as a bed of sand, depends on the
permeability of the medium.

For laminar flow (d50<1 mm)

Sand grain size smaller than about 1 mm in general. Darcy’s Law is used.

K p dp
VB = (6)
ν ρ dz
VB = K I J (7)

where VB is bulk velocity, dp/dz is pressure gradient, Kp specific permeability(m2), ρ is

density of water, ν is kinematic viscosity of water, and KI is coefficient of permeability
(ms-1).

The pressure gradient and hydraulic gradient are related through;

dp
= ρgJ (8)
dz

The two alternative measures of permeability are related through;

gK p
KI = (9)
ν

For turbulent flow (d50>1 mm)

The force exerted on the grains is a combination of viscous and turbulent forces.
Darcy’s Law is then replaced by Forchheimer’s equation;

261
Sediment Transport Technology

J = a I VB + bI VB2 (10)

1
where a I =
KI

Alternatively, Forchheimer’s equation can be written in a form analogous to Darcy’s

equation.

dp ρν ρ
= a p 2 VB + b p VB2 (11)
dz d d
d2 ν 1
Kp = , aI = a , bI =
2 p
bp
ap gd gd

The dimensionless coefficients ap and bp are functions of porosity, and grain shape,
packing, orientation, and grading.

Permeability is important in the following applications:

1. Damping of wave energy, due to viscous pumping of water through the bed
2. Entrainment of sediment by waves, because pumping of water in and out of the bed
will respectively increase and reduce the effective weight of grains
3. Beach stabilization by artificial drainage
4. Transfer of pollutants into and out of bed
5. Percolation through gravel beaches
6. Stability of rubble-mound breakwaters and rock armour structures

An upward flow of water through a sand bed exerts an upward drag force on grains,
corresponding to the resistance controlling permeable flow. If this force is larger than
the buoyant weight of the grains, the bed becomes liquefied. The weight of the grains is
no longer supported by resting on other grains, but by fluid forces, and the bed behaves
like a fluid, so that a heavy marine structure placed on the bed sinks through it. The
minimum vertical pressure gradient for liquefaction is

⎛ dp ⎞
⎜ ⎟ = g ( ρ s − ρ )(1 − ε ) (12)
⎝ dz ⎠ mf

The corresponding minimum liquefaction velocity of upward flowing water necessary

to achieve liquefaction is;

Wmf =
ν
d
{[10.36 2
+ 1.049ε 4.7 D*3 ]
1/ 2
}
− 10.36 (13)

where

1/ 3
⎡ g ( s − 1) ⎤
D* = ⎢ ⎥ d 50 (14)
⎣ ν
2
⎦

262
 Coastal Hydrodynamics and Sediment Transport

The velocity formulae are (Soulsby, 1997):

Wen and Yu (1966):

Wmf =
ν
d
{[33.7 2
+ 0.0408D*3 ] 1/ 2
}
− 33.7 (15)

Wallingford (d<0.88 mm):

g d 502
Wmf = 5.75 × 10 − 4 ( s − 1) (16)
ν

3. Coastal Hydrodynamics
Waves generated in the deep water of the ocean travel without change of form for a
considerable distance. As they approach the shore, the decreasing depth begins to have
an effect. The waves undergo a decrease in wavelength and increase in height, and
hence become steeper as they travel towards the shore. Waves at an oblique angle of
approach are refracted (so that the crests are turned to become nearly parallel to the
shoreline) before they break. As the waves break, with the attendant air entrainment,
virtually all of the energy is lost and a swash (the onrush of water towards the beach
followed by the down rush away from the beach) is established. When the line of
breaking is at an angle to the beach, currents parallel to the beach, known as long shore
currents, are established. The swash and the long shore currents are important causes of
sediment motion that shapes the coastline (Novak et al.1996).

Beaches adjust themselves to absorb the energy of the waves; if the wave climate
exhibits large variability, the beach profiles become complex. When the waves break,
the flow is intensely turbulent and large quantities of sediment may be transported in
suspension. The subsequent onrush of water can carry even coarse sediments towards
the beach. The down rush is slow and can transport relatively small sediments offshore.
If the energy of oncoming waves is small, only finer sediments are moved onshore. The
sediment transport over a period of time will depend on the variations in the wave
climate, including the direction of wave attack. In addition to the onrush and down rush
of water on the beach, two important types of currents are set up by the wave motion:
rip currents and long shore currents (Novak et al., 1996).

The rip current is the narrow and strong flow seaward as a result of bed features and the
formation of opposing along-shore currents. For example, when water is trapped behind
obstacles such as sandbars, it forms a current along the coast, which can drain with a
high velocity current towards the sea through an opening in the sandbars (Fig. 4). Thus
the transport of water over the sandbar (e.g. by a large breaker) feeds the rip current to
form a near shore circulation (Novak et al., 1996).

When waves break at an angle to the coastline, a long shore current parallel to the
coastline is established (Fig. 4), responsible for the net transport of sediment along
shore. Knowledge of the sediment transport capacity of the long shore currents is
important in the preventing erosion of the beach or in silting a harbour entrance Novak
et al., 1996).

263
Sediment Transport Technology

Figure 4. Rip and long shore currents in and near the surf zone (van Rijn, 1993)

3.1. Current
Currents in the sea may be caused by tidal motions, wind-stress, atmospheric pressure
gradients, wave-induced forces (long shore and rip currents), river outflows, large-scale
quasi-steady water surface slopes and horizontal density gradients associated with
oceanic circulations.

A current flowing over the seabed experiences friction with the bed, which forms a
turbulent boundary layer typically some meters or tens of meters thick. In shallow water
the boundary layer may occupy the entire depth, whereas in deep water it occupies the
lower part of the water column and is overlain by water relatively unaffected by friction.
The depth averaged current speed is;

h
1
U =
h ∫
U ( z )dz
0
(17)

where h is water depth, z is height above sea bed, U(z) is current speed at height z.

u* ⎛ z ⎞
U ( z) = ln⎜ ⎟⎟ (18)
χ ⎜⎝ z 0 ⎠

where u* is friction velocity and χ is von Karman’s constant. This equation is valid is
from a few cm above the bed to the about 20-30% of the water in shallow water (2-3
m), or 20-30% of the boundary layer thickness in deep water (e.g. 20-30 m).

The bed roughness length:

ks ⎡ ⎛ u* k s ⎞⎤ ν
z0 = ⎢1 − exp⎜⎜ − ⎟⎟⎥ + (19)
30 ⎣ ⎝ 27ν ⎠⎦ g u*

264
 Coastal Hydrodynamics and Sediment Transport

Simplified bed roughness:

ks
z0 = for u* ks/ν > 70 hydraulic rough (20)
30
ν
z0 = for u* ks/ν < 5 hydraulic smooth (21)
g u*
k ν
z0 = s + for transition (22)
30 g u*

Nikuradse roughness: ks=2.5d50

For hydrodynamic rough flows:

d 50
z0 = (23)
12

The tidal current velocity profile throughout the water column is given by Soulsby
(1997):

U ln( z / z 0 )
U ( z) = for z0<z<0.5δ (24)
ln(δ / 2 z 0 ) − δ / 2h
U ln(δ / 2 z 0 )
U ( z) = for 0.5δ<z<h (25)
ln(δ / 2 z 0 ) − δ / 2h

Boundary layer thickness:

U aσ −Ub f
δ = 0.0038 (26)
σ2 − f 2

where σ is the radian tidal frequency, f is Coriolis parameter:

f = 1.4544 ⋅ 10 −4 ⋅ sin (latitude) rads -1

and U a and U b are maximum and minimum values of the depth averaged current
velocities through a tidal cycle. Soulsby (1997) alternatively give the following
empirical formulas:

1/ 7
⎛ z ⎞
U ( z) = ⎜ ⎟ U for 0<z<0.5h (27)
⎝ 0.32h ⎠
U ( z ) = 1.07U for 0.5h<z<h (28)

The total bed shear stress includes friction and bed forms:

τ 0 = τ 0s + τ 0 f (29)

265
Sediment Transport Technology

The bed shear stress:

τ 0 = ρCDU 2 (30)

where

f g gn2
CD = = 2 = 1/ 3 (31)
8 C h
τ0
u* = (32)
ρ

A simple power law may be used:

β
⎛z ⎞
CD = α ⎜ 0 ⎟ (33)
⎝ h⎠

The power law:

1/ 7
u* 1 ⎛ d 50 ⎞
= ⎜ ⎟ (34)
U 7⎝ h ⎠

The friction velocity at depth z:

1/ 7
⎛d ⎞
u* = 0.12⎜ 50 ⎟ U ( z ) (35)
⎝ z ⎠

Alternatively for CD:

2
⎡ χ ⎤
CD = ⎢ ⎥ (36)
⎣ B + ln( z 0 / h) ⎦

Assuming χ = 0.4 and B = 1;

2
⎡ 0.4 ⎤
CD = ⎢ ⎥ (37)
⎣1 + ln( z 0 / h) ⎦

For deep water:

B = (δ / 2h) − ln(δ / 2h) and χ = 0.4

The current velocity at height of 1 m above the bed;

τ 0 = ρ C100U 100
2
(38)

266
 Coastal Hydrodynamics and Sediment Transport

For rippled bed:

U ⎛δ ⎞
= 6 + 2.5ln⎜⎜ i ⎟⎟ (39)
u *s ⎝ ks ⎠
u*2s = g δ i J
τ 0 s = ρ u*2s

where ks = 2.5d50

This method is less useful in tidal flows, because the surface slope J associated with the
tidal wave is often not known, and, apart from in very shallow water. For very shallow
water:

τ 0 = ρ g hJ (for h<5 m) (40)

At very high flow speeds, with intense sheet flow, a third component of roughness is
found, which arises from the momentum extracted by the flow to move the sand grains
z0t. Hence;

z 0 = z 0 s + z 0 f + z 0t (41)
5τ 0 s
z 0t = (42)
30 g ( ρ s − ρ )

3.2. Waves
Waves play a major role in stirring up sediments from the sea bed, as well as giving rise
to steady current motions such as long shore currents, undertow, and mass-transport
velocities, which transport the sediments. The asymmetry of velocities beneath the crest
and the trough of waves is another source of net transport of sediment.

The amplitude of the wave orbital velocity just above the bed may be found by Eq. 43.
Fig. 5 gives a curve marked JONSWAP, hence an approximations may be done for the
wave parameters.

πH
Uw = 43)
T sinh(kh)

According to Pierson Moskowitz:

Tz = 0.71T p
2π
Tp =
ωp

For regular waves:

Tp = T

267
Sediment Transport Technology

Figure 5. Bottom velocity for monochromatic waves and random waves (Soulsby, 1997)

1. Stokes (2nd-5th)

h>0.1gT2

2. Cnoidal wave

0.003gT2 <h<0.016gT2

3. Stream function wave

0.006gT2 <h<0.016gT2

For Stokes’ 2nd -order theory, the orbital velocities at sea bed;

⎡ 3kh H ⎤
U wc = U w ⎢1 + ⎥ (44)
⎣ 8sinh( kh) h ⎦
⎡ 3kh H ⎤
U wt = U w ⎢1 − ⎥ (45)
⎣ 8sinh( kh) h ⎦

Alternatively

U wc = U w (46)
U wt = U w [1 − r2 exp(−r3 h / L0 )] (47)
0.65
⎛H ⎞
r2 = 3.2⎜⎜ 0 ⎟⎟
⎝ L0 ⎠
⎛H ⎞
r3 = −27log10 ⎜⎜ 0 ⎟⎟ − 17
⎝ L0 ⎠

268
 Coastal Hydrodynamics and Sediment Transport

The maximum instantaneous shear stress;

1
τ mak = ρ f w (aω ) 2 (48)
2

where a is wave amplitude and ω is angular frequency. Or using orbital bottom velocity;

1
τ w = ρ f w U w2 (49)
2

where fw is wave friction factor. It depends on Reynolds number and relative roughness,

f w = f (Re w , r )
Uw A
Re =
ν
U wT
A=
2π

in which A is semi-orbital excursion.

For turbulent flow:

2
0.32 ⎧⎪ ⎡ ⎛ Re w f w1 / 2 ⎞⎤ 4.71r ⎫⎪
= ⎨ln (6.36r f w ) − ln ⎢1 − exp⎜⎜ − 0.0262
1/ 2
⎟⎟⎥ + 1/ 2 ⎬
+ 1.64 (50)
fw ⎩⎪ ⎣⎢ ⎝ r ⎠⎦⎥ Re w f w ⎪⎭

For only rough turbulent flow (Swart, 1974):

f wr = 0.3 for r ≤ 1.57 (51)

f wr = 0.00251exp 5.21r ( −0.19
) for r > 1.57 (52)

Nielsen (1992):

f wr = exp(5.5r −0.2 − 6.3) (53)

Soulsby (1997):
−0.52
⎛ A⎞
f wr = 1.39⎜⎜ ⎟⎟ (54)
⎝ z0 ⎠
f wr = 0.237r −0.52 (if z0 = ks/30) (55)

For smooth turbulent flow:

f ws = BRe −wN (56)

269
Sediment Transport Technology

Re w ≤ 5 × 10 5 Laminar flow B=2 N = 0.5

Re w > 5 × 10 Turbulence smooth
5
B = 0.0521 N = 0.187

Fredsoe and Deigaard (1992):

B = 0.035 N = 0.16 (57)

Myrhaug (1995):

B = 0.045 N = 0.175 (58)

Total bed shear stress:

τ 0 = τ ws + τ wf (59)

In most areas in shallow depths, apart from the surf zone, the bed is generally formed
into ripples. These may be current generated or wave generated, but it is assumed that
wave generated in this chapter. It is very important to estimate the rate of sediment
transport in surf zone or in outside the surf zone. The limit wave steepness of wave;

For transition depths:

H ⎛ 2πh ⎞
= 0.142tanh⎜ ⎟ (60)
L ⎝ L ⎠

In deep water:

H = 0.142 L (61)

In shallow water:

H = 0.78h (62)

For irregular waves:

H rms ⎛ 33H 0 ⎞
= 0.5 + 0.4 tanh⎜⎜ ⎟⎟ (63)
h ⎝ L0 ⎠

4. Initiation of Motion
Particle movement will occur when the hydraulic forces on a particle are larger than the
critical hydraulic force for initiation of motion. The disturbing force is the resulting
force of fluid and drag forces acting on the particle. This force is proportional to the bed
shear stress and the particle surface area. The stabilizing force is equal to the submerged
particle weight. In addition, due to oscillatory flow, the horizontal fluid acceleration
causes inertia force as disturbing force.

270
 Coastal Hydrodynamics and Sediment Transport

Figure 6. Forces on a sand grain

1. Gravity force:

πd 3
FG = (γ s − γ w ) (64)
6

2. Hydrodynamic drag force:

a) Form drag:

1 πd 2
FD , F = c F ρ uu (65)
2 4

b) Friction drag

FD , S = c s ρd 2 u u (66)

and resulting force

1 πd 2
FD = c D ρ uu (67)
2 4

3. Hydrodynamic lift force:

u2 3
FL = c L ρ d = c L ρu 2 d 2 (68)
d

4. Added mass:

πd 3 du
FA = c A ρ (69)
6 dt

271
Sediment Transport Technology

5. Pressure force:

πd 3 du
FP = ρ (70)
6 dt

6. Inertia force:

πd 3 du
FM = c M ρ (71)
6 dt

7. Reaction force:

πd 3
FM = c M ρ Aω 2 cos ωt (72)
6

in similar manner, drag force may be written in the form of

ρ πd 2
FD = c D Aω sin ωt Aω sin ωt (73)
2 4

If we write the ratio of drag force and inertia force, d/A will be got which is called the
Keulegan Carpenter number.

π
ρ d 3 c M Aω 2
FM 6 4 cM d
= = (74)
1 π 2
FD
ρ d c D A 2ω 2 3 c D A
2 4

However it is known that the incipient motion depends on the dimensionless parameter;
grain Reynolds number and Froude number under steady flow. Additional dimensional
parameter is now defined; Keulegan Carpenter number under waves.

A simple dimensional measure of the fluid forces on a sediment particle under waves is
mobility number ψ will be defined in the following. For sand particles (0.2 mm) under
waves with typical semi-excursions, A of the order 0.1 m- 2 m, the Keulegan Carpenter
number d/A is very small and hence the drag force will tend to dominate over pressure
force (Nielsen,1992). Hence, the total disturbing force on a sand particle at the bed is
approximately proportional to the square of the velocity amplitude Aω, and the ratio
between this disturbing force and the stabilizing force due to gravity is reasonably
described by the mobility number.

ψ =
( Aω )2 (75)
(s − 1)gd
A measure of the balance between disturbing and stabilizing forces on sand grains at the
bed was suggested by Shields (1936) in a study of the incipient sediment motion in
steady flow:

272
 Coastal Hydrodynamics and Sediment Transport

τ (0)u*2
θ= = (76)
ρ (s − 1)gd (s − 1)gd

However in connection with wave motion, the Shields parameter (corresponding to total
stress) is generally defined in terms of the peak bed shear stress (Nielsen, 1992):

1
f w ( Aω )
∧ 2
τ 1
θ= = 2 = f wψ (77)
ρ (s − 1)gd (s − 1)gd 2
where fw is the wave friction factor.

4.1. Critical velocity

For a steady current, the threshold (or critical) depth-averaged speed is given by van
Rijn (1984):

u kr = 0.19(d 50 ) 0.1 log10 (4h / d 90 ) (m/s) for 100 ≤ d50 ≤ 500 µm (78)
u kr = 8.5(d 50 ) 0.6 log10 (4h / d 90 ) (m/s) for 500 ≤ d50 ≤ 2000 µm (79)

Soulsby (1997) gives another equation for the critical velocity:

u kr = 7(h / d 90 )1 / 7 [g ( s − 1)d 50 f ( D* )]
1/ 2
for D* > 0.1 (80)

where

0.3
f ( D* ) = + 0.055[1 − exp(− 0.020D* )] (81)
1 + 1.2D*
1/ 3
⎡ g ( s − 1) ⎤
D* = ⎢ ⎥ d 50 (82)
⎣ ν
2
⎦
s = ρs ρ

Eq. 80 is given in Fig. 7, same equation also valid under waves in Fig 8.

Komar and Miller (1974) give critical orbital velocity:

u wkr = [0.118 g ( s − 1)]

2/3
d 1 / 3T 1 / 3 for d < 0.5 mm (83)
u wkr = [1.09 g ( s − 1)]
4/7
d 3 / 7T 1 / 7 for d > 0.5 mm (84)

CERC(1984) proposed a graph for the critical depth with maximum orbital velocity.

0.5
⎡ ⎛γ ⎞ ⎤
U maks ( − d ) = ⎢8 ⎜⎜ s − 1⎟⎟ g D50 ⎥ (85)
⎣ ⎝γ ⎠ ⎦

273
Sediment Transport Technology

Figure 7. Threshold current speed for motion of sediment by steady flows (Soulsby, 1997)

Figure 8. Threshold orbital velocity for motion of sediment by waves (Soulsby, 1997)

4.2. Critical bed shear stress

The critical bed shear stress was developed by Shields (1936) in terms of ratio of the
force exerted by the bed shear stress acting to move a grain on the bed, to the
submerged weight of the grain counteracting this. The threshold Shields parameter is
defined as:

τ kr
θ kr = (86)
g ( ρ s − ρ )d

It can be plotted in Fig.10 against the dimensionless grain size given by:

274
 Coastal Hydrodynamics and Sediment Transport

Figure 9. Maximum orbital bed velocity according to small amplitude theory (CERC, 1984)

Figure 10. Threshold of motion of sediments beneath waves and/or currents

275
Sediment Transport Technology

1/ 3
⎡ g ( s − 1) ⎤
D* = ⎢ ⎥ d (87)
⎣ ν
2
⎦

Soulsby and Whitehouse (1997) give the algebraic expression of the Shields curve.

0.24
θ kr = + 0.055[1 − exp(−0.02D* )] (88)
D*

The grain diameter, which is just immobile, is under steady flow:

0.250u 2.8
d kr = for dkr>10 mm (89)
h 0.4 [g ( s − 1)]
1.4

under waves:

97.9u w3.08
d kr = for dkr>10 mm (90)
T 1.08 [g ( s − 1)]
2.08

5. Bed Forms
Different observers have given different names to what are, essentially, the same bed
forms and Sleath (1984) classified the bed forms as shown in Fig. 11 for oscillatory
flows. The type of bed forms depends on the strength and nature of the flow; steady
current, tidal current, waves, or a combination of these (Fig. 11). Bed forms are relief
features initiated by the fluid oscillations generated downstream of small local obstacles
over a bottom consisting of movable sediment materials.

Consider a flat bed consisting of loose, cohesionless solid grains with a slowly
increasing current or wave action. Eventually sediment movement will start. With the
beginning of sediment transport bed forms will appear. These bed forms will have a
considerable influence on the transport mechanism (the bed roughness, the
concentration near the bed).

The geometry of these bed forms depends on various flow parameters. There have been
many attempts to determine the boundaries between the different bed regimes. Fig.13
shows the boundaries between different bed regimes.

5.1. Current ripples and sand waves

Current generated ripples are asymmetrical in cross section with the steeper slope on the
downstream side of the crest. Their wave length and height:

λr = 1000d 50 (91)
λ
∆r = r (92)
7

276
 Coastal Hydrodynamics and Sediment Transport

Figure 11. Bed forms produced by oscillatory flows Sleath (1984)

Figure 12. Types of bed forms (a) current ripples, (b) wave
ripples, (c) sand waves, (d) bar (Soulsby, 1997)

277
Sediment Transport Technology

Figure 13. Bed regime boundaries for waves and for current (Velden, 1989)

5.1.1. Dunes and sand waves

Yalın (1964):

For τ 0 s < τ kr ∆s = 0 (93)

h⎛ τ ⎞
For τ kr < τ 0 s < 17.6τ kr ∆s = ⎜⎜1 − kr ⎟⎟ (94)
6 ⎝τ 0s ⎠
For τ 0 s ≥ 17.6τ kr ∆s = 0 (95)

λ s = 2πh

Van Rijn (1984):

Fo r τ 0 s < τ kr ∆s = 0 (96)
0.3
⎛ d 50 ⎞
For τ kr < τ 0 s < 26τ kr ∆s = 0.11h⎜ −0.5T
⎟ (1 − e s )(25 −T s) (97)
⎝ h ⎠
For τ 0 s ≥ 26τ kr ∆s = 0 (98)

λ s = 7.3h

where ∆s height of sand wave, λs wave length of sand wave, h water depth, τ0s bed shear
stress due to skin friction, τkr critical shear stress, d50 median grain size, and

τ 0 s − τ kr
Ts =
τ kr

278
 Coastal Hydrodynamics and Sediment Transport

5.1.2. Sheet flow

θ s = 0.8

or

τ 0 s > 0.8g ρ ( s − 1)d (99)

5.2. Wave ripples

Wave generated ripples are symmetrical about the crest being relatively sharp.

Grant & Madsen (1982):

For θ ws ≤ θ kr
∆r = λr = 0 (100)

For θ kr < θ ws ≤ θ B

∆r = 0.22(θ ws / θ kr ) −0.16 A (101)

λr = ∆r /[0.16(θ ws / θ kr ) −0.04 ] (102)

θ B = 1.8θ kr ( D*1.5 / 4) 0.6

1/ 3
⎡ g ( s − 1) ⎤
D* = ⎢ ⎥ d
⎣ ν
2
⎦

For θ ws > θ B

∆r = 0.48( D*1.5 / 4) 0.8 (θ ws / θ kr ) −1.5 A (103)

λr = ∆r /[0.28( D*1.5 / 4) 0.6 (θ ws / θ kr ) −1.0 ] (104)

Nielsen (1992) under regular waves:

For θ ws ≤ θ kr
∆r = λr = 0 (105)

For θ ws > θ B and for ψ < 156 or θ ws < 0.831

∆r = (0.275 − 0.022ψ 0.5 A (106)

λr = ∆r /(0.182 − 0.24θ ws
1.5
) (107)

Forψ ≥ 156 or θ ws ≥ 0.831

∆r = λr = 0 (108)

279
Sediment Transport Technology

5.3. Friction due to bed forms

For current related ripples:

∆2r
z0 f = ar (109)
λr

where ar is in the range between 0.3 and 3, typical value is 1.0.

For wave generated ripples:

Grant ve Madsen (1982) for ar = 0.923:

2
⎡⎛ θ ⎞
0.5
⎤
z 0t = 5.33( s + 0.5)d 50 θ kr ⎢⎜⎜ ws ⎟⎟ − 0 .7 ⎥ (110)
⎢⎣⎝ θ kr ⎠ ⎥⎦

Nielsen (1992) for ar = 0.267:

z 0t = 5.67(θ ws − 0.05) 0.5 d 50 (111)

Raudkivi (1988) for ar = 0.533:

z 0t = 0.00533u w2.25 (112)

6. Suspension Load and Transport

For current velocity or wave conditions significantly above the threshold of motion,
sand is entrained off the bed and into suspension, where it is carried at the same speed
as the current.

The variation of the instantaneous sediment concentrations in space and time which is
relevant for the cross shore transport process in nonbreaking waves appears to be a
complicated physical process particularly in the ripple regime. This process seems
hardly predictable due to the random character of the phenomena involved.

Cross shore transport process in nonbreaking waves are caused by various mechanisms,
see Fig.14 (van Rijn, 1989);
1. The asymmetry of the wave motion yielding larger onshore peak velocities under
the wave crests than offshore peak velocities under wave troughs which may result
in a net offshore directed transport in case of a ripple covered bed or in a net
onshore directed transport in case of a plane bed (sheet flow).
2. The generation of net mean (Eulerian) onshore directed velocities in the near bed
region which may result in a net onshore directed transport.
3. The generation of forced long waves due to a mean water surface decrease (trough)
under large amplitude wave groups resulting in secondary offshore directed orbital
velocities under the trough and a net offshore directed transport because the
sediment concentrations are largest under large amplitude waves.

280
 Coastal Hydrodynamics and Sediment Transport

Figure 14. Drift velocities due to nonlinear effects (van Rijn, 1993)

Time averaged sediment concentrations which are relevant for cross shore and
longshore transport process due to breaking waves are less complicated to describe and
have been studied by many researchers. Transport processes in breaking waves are
caused by the following mechanisms (Fig. 15):
1. The generation of net mean offshore directed velocities in the lower layers
(undertow)
2. The generation of large scale horizontal circulation cells with longshore currents and
with offshore directed rip currents

It is known that the wave motion reduces the current velocities near the bed, but
strongly increase the near bed concentrations due to its stirring action. Wave related
sediment transport is defined as the transport of sediment particles by the oscillating
fluid components.

Figure 15. Undertow current due to breaking waves (van Rijn, 1993)

281
Sediment Transport Technology

For grains to remain in suspension, their settling velocity must be smaller than the
upward turbulent component of velocity, which is related to shear velocity. This gives a
very approximate relation

u * = ws (113)

6.1. Settling velocity

There are a number of formulae in still water.

Gibbs et al (1971):

ws =
[qν 2
+ g d 2 ( s − 1)(3.869 ± 10 −3 + 04801d ) ]
1/ 2
− 3ν
(114)
0.011607 + 0.074405d

Hallermeier (1981):

ν D*3
For D ≤ 39
3
* ws = (115)
18d
D 2.1
For 39 < D*3 < 10 4 ws = ν * (116)
6d
1.05ν D*1.5
For 10 4 < D*3 < 3 × 10 6 ws = (117)
d

Van Rijn (1984):

ν D*3
For D*3 ≤ 16.187 ws = (118)
18d
10ν
For 16.187 < D*3 ≤ 16187 ws =
d
[
(1 + 0.01D*3 )1 / 2 − 1 ] (119)

1.1ν D*1.5
For D *3 > 16187 ws = (120)
d

Soulsby (1997):

ν
ws =
d
[(10.36 2
+ 1.049D*3 )1 / 2 − 10.36 ] (121)

6.2. Sediment concentration

In a sand suspension the settling of the grains towards the bed is counterbalanced by
diffusion of sand upwards due to the turbulent water motions near the bed (see Fig. 16).
The equation governing this balance is:

dc
wsc C = − K s (122)
dz

282
 Coastal Hydrodynamics and Sediment Transport

Figure 16. Concentration profile (van Rijn, 1993)

where C; volume concentration of sediment at height z, and Ks; eddy diffusivity of

sediment.

The shape of the profile depends on the ratio:

ws
b= (123)
χ u*

where b; Rouse number or suspension parameter, χ; Von Karman’s constant (χ = 0.4),

and u*; total friction velocity. The entrainment of sediment from the bed depends on the
skin friction τ0s = ρu*2s , whereas the diffusion of the sediment higher into the water
column depends on the total shear stress τ0 = ρu*2 .

6.2.1. Concentration profiles under current (Soulsby, 1997)

Power law profile:

−b
⎛ z ⎞
C ( z ) = C a ⎜⎜ ⎟⎟ (124)
⎝ za ⎠

Rouse profile:

−b
⎛ z (h − z a ) ⎞
C ( z ) = C a ⎜⎜ ⎟⎟ (125)
⎝ za h − z ⎠

Van Rijn profile:

− b′
⎛ z (h − z a ) ⎞
For za<z<h/2 C ( z ) = C a ⎜⎜ ⎟⎟ (126)
⎝ za h − z ⎠
b′
⎛ z ) ⎞ ⎡ z 1 ⎤
For h/2<z<h C ( z ) = C a ⎜⎜ a ⎟⎟ exp ⎢− 4b′⎛⎜ − ⎞⎟⎥ (127)
⎝ h − za ⎠ ⎣ ⎝ h 2 ⎠⎦

283
Sediment Transport Technology

b
b′ = + B2
B1
2
⎛w ⎞
For 0.1<ws/u*<1 B1 = 1 + 2⎜⎜ s ⎟⎟ (128)
⎝ u* ⎠
For ws/u*≥1 B1 = 2 (129)
0.8 0.4
⎛w ⎞ ⎛ Ca ⎞
For 0.01≤ws/u*≤1 B1 = 2.5⎜⎜ s ⎟⎟ ⎜ ⎟ (130)
⎝ u* ⎠ ⎝ 0.65 ⎠
For za>0.1h or ws/u* B2 = 0 (131)

where z; height above seabed, za; a reference height near the seabed, C(z); sediment
concentration at height z, Ca; sediment reference concentration at height za, h; water
depth, and, b; Rouse number.

Smith and Mclean (1977):

0.00156Ts 26.3τ kr Ts d 50
Ca = at height za = + (132)
1 + 0.0024Ts ρ g ( s − 1) 12

Van Rijn(1984):

0.015Ts3 / 2
For z a = ∆s / 2 and z a min = 0.01h Ca = (133)
z z D*0.3

6.2.2. Concentration profiles under waves (Soulsby, 1997)

The boundary layer is thin under waves, so

C ( z) = C0 e − z / l (134)
uw uw
For < 18 l = 0.075 ∆r (135)
ws ws
u
For w ≥ 18 l = 1.4 ∆r (136)
ws
C 0 = 0.005θ r3 (137)
2
f wr u
θr = w
(138)
2( s − 1) g d (1 − π ∆ r / λr ) 2

where uw; wave orbital velocity amplitude, ws; grain settling velocity, l; decay length
scale, ∆r; ripple height, λr; ripple wave length, fwr; rough bed wave friction factor (Eq.
52).

u wT
r= (139)
5 π d 50

284
 Coastal Hydrodynamics and Sediment Transport

6.3. Suspended load

The general principle for the calculation of the suspended sediment transport rate qs is to
make an integral through the water depth of the sediment flux U(z)C(z) at height z,
having first established the vertical profiles of velocity U(z) and concentration C(z) (see
Fig. 16).

h
q s = ∫ U ( z )C ( z )dz (140)
za

where za is the reference height near bed the bed, and h is the water depth.

7. Bed Load and Transport

The principal mechanisms of sediment transport in oscillatory flow are similar to those
in steady flow. Sediment transport as bed load remains more or less continuously in
rolling and sliding contact with the bed (Sleath, 1984). Wave induced transport
processes are related to the velocities generated by high and low frequency wave
phenomena. Net onshore transport is dominant in non-breaking wave conditions,
whereas net offshore transport is dominant in breaking wave conditions. Bed load
transport can occur;
1. over a flat bed at low flows
2. in conjunction with ripples or larger bed forms for stronger flows
3. over a flat bed for very strong flows where ripples are washed out (sheet flow)

7.1. Bed load under currents

Bed load terms are:
qb : volumetric transport rate (m3/s)
Qb = ρ s qb : mass transport rate (kg/ms)
ib = (ρ s − ρ )gqb : immersed weight transport rate (N/ms)
q B = qb (1 − ε ) : volume of settled bed material (including pore water) per unit
time per unit width (m2/s)

For quartz (ρs = 2650 kg/m3) and porosity (ε = 0.4)

qb (m3/s)×3.15×107 = qb (m3/m-year) excluding pore water

qb (m3/s)×5.26×107 = qb (m3/m-year) including pore water
qb (m3/s)×8.36×107 = Qb (ton/m-year) excluding pore water

Most of the formulas my be expressed in the form (Soulsby, 1997);

Φ = f (θ ,θ kr ) (141)

where

qb
Φ= = dimensionless bed load transport rate (142)
[g (s − 1)d ]3 1/ 2

285
Sediment Transport Technology

τ0
θ= = Shields parameter (143)
g ρ ( s − 1) d

and θcr; critical Shields parameter, qb; volumetric bed load transport rate per unit width.

Meyer Peter and Muller (1984):

Φ = 8 (θ − θ kr ) 3 / 2 (144)
θ kr = 0.047 (145)

Bagnold (1963):

Φ = FB θ 1 / 2 (θ − θ kr ) (146)
0.1
FB = 1 / 2 (147)
C D (tan φi + tan β )

where θ; total Shields parameter, CD; total drag coefficient, φi; angle of repose, and, β;
bed slope.

Van Rijn (1984):

Φ = FR θ 1 / 2 (θ 1 / 2 − θ kr1 / 2 ) 2.4 (148)

0.2
0.005 ⎛ d ⎞
FR = 1.7 ⎜ ⎟ (149)
CD ⎝ h ⎠

Yalin (1963):

Φ = FY θ 1 / 2 (θ − θ kr ) (150)
0.635 ⎡ 1 ⎤
FY = ⎢1− ln(1 + aT )⎥ (151)
θ ⎣ aT
kr ⎦
a = 2.45θ kr0.5 s −0.4 (152)
T = (θ − θ kr ) / θ kr (153)

Madsen (1991):

Φ = FM (θ 1 / 2 − 0.7θ kr1 / 2 )(θ − θ kr ) (154)

FM = 8 / tan φi for rolling/sliding grains
FM = 9.5 for saltating sand

Ashida ve Michiue (1972):

Φ = 17 (θ 1 / 2 − θ kr1 / 2 )(θ − θ kr ) (155)

286
 Coastal Hydrodynamics and Sediment Transport

Wilson (1966):

Φ = 12θ 3 / 2 (156)

Nielsen (1992):

Φ = 12θ 1 / 2 (θ − θ kr ) (157)

7.2. Bed load under waves

Some formulas are suggested (Soulsby, 1997).

Madsen ve Grant (1976):

qb1 / 2 = FMG ws dθ w3 (158)

For θw>>θkr FMG = 12.5
For θw→θkr FMG → 0

Sleath (1978):

qb1 / 2 = 47ωd 2 (θ w − θ kr ) 3 / 2 for coarse grains (159)

Soulsby (1997):

[
qb1 / 2 = 5.1 g ( s − 1)d 3 ]
1/ 2
(θ w − θ kr ) 3 / 2 (160)

where θw; amplitude of oscillatory component of θ due to wave, and, ω; angular

frequency.

8. Total Load and Coastal Morphology

The bed load and suspended load contributions to the total sediment transport rate may
be calculated separately and added or alternatively the total load sediment transport rate
may be given by an empirical formula. Total bed load transport rates may be expressed
in the following units (Soulsby, 1997):

qt : volumetric transport rate (m2/s)

Qt = ρ s qt : mass transport rate (kg/ms)
it = (ρ s − ρ )gqt : immersed weight transport rate (N/ms)
qT = qt (1 − ε ) : volume of settled bed material (including pore water) per unit
time per unit width (m2/s)

8.1. Total load under current

Different formula may be followed.

287
Sediment Transport Technology

Engelund-Hansen (1972):

0.04C D3 / 2 U 5
qt = (161)
[g ( s − 1)]2 d 50
Ackers & White (1973):

n m
⎛U ⎞ ⎛ FAW − AAW ⎞
qt = C AW U d ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟ (162)
⎝ u* ⎠ ⎝ AAW ⎠

where

1− n
u*n ⎡ U ⎤
= (163)
1/ 2 ⎢
[g ( s − 1)d ] ⎣ 2.46ln(10h / d ) ⎥⎦
FAW

and if

1/ 3
⎡ g ( s − 1) ⎤
D* = ⎢ ⎥ d (164)
⎣ ν
2
⎦

defined. For fine sediment 1 < D* ≤ 60 :

n = 1 − 0.243ln D*
0.23
AAW = 1 / 2 + 0.14
D*
9.66
m= + 1.34
D*
[
C AW = exp 2.86ln D* − 0.434(ln D* ) − 8.13
2
]
and for coarse sediment D* > 60

n=0
AAW = 0.17
m = 1 .5
C AW = 0.025

where d = d35, and u* = C D1 / 2U .

For 1 ≤ D* ≤ 60 :

6.83
m= + 1.67
D*

288
 Coastal Hydrodynamics and Sediment Transport

[
C AW = exp 2.79ln D* − 0.426(ln D* ) 2 − 7.97 ]
For D* > 60

m = 1.78
C AW = 0.025

Van Rijn (1984):

qt = qb + q s
2.4
⎧⎪ U − U kr ⎫⎪ ⎛ d 50 ⎞1.2
qb = 0.005U h⎨ ⎬ ⎜ ⎟ (165)
⎪⎩ [( s − 1) g d 50 ]1 / 21 ⎪⎭ ⎝ h ⎠
2.4
⎧ U − U kr ⎫ ⎛ d 50 ⎞ − 0.6
q s = 0.012U h⎨ 1 / 21 ⎬ ⎜ ⎟( D* ) (166)
⎩ [( s − 1) g d 50 ] ⎭ ⎝ h ⎠

where

0 .1 ⎛ 4h ⎞
For 0.1 ≤ d 50 < 0.5 mm U kr = 0.19(d 50 ) log⎜⎜ ⎟⎟
⎝ d 90 ⎠
0 .6 ⎛ 4h ⎞
For 0.5 ≤ d 50 ≤ 2 mm U kr = 8.5(d 50 ) log⎜⎜ ⎟⎟
⎝ d 90 ⎠

These equations are valid for h = 1∼20 m, U = 0.5∼5 m/s, d50 = 0.1∼2 mm and 15°C
fresh water, however Van Rijn (1984) equation may be also used for sea water.

8.2. Bed load under waves

Bailard (1981):

qt = qb 0 − qbs + q s 0 − q ss

qb 0 =
Cf εB
g ( s − 1)tan φi
2
u u[ ] (167)

qbs =
C f ε B tan β
g ( s − 1)tan φi
2
3
u i [ ] (168)

qs0 =
Cf εs
g ( s − 1)Ws
3
u u [ ] (169)

q ss =
C f ε s2 tan β
g ( s − 1)W 2
[u ]i
5
(170)
s

where

s = ρs/ ρ

289
Sediment Transport Technology

Cf : friction coefficient ( τ = ρ C f u u )
U : total near bottom velocity due to combined waves and currents
φi :angle of internal friction of sediment (tanφi = 0.63)
tanβ : bed slope
i : unit vector directed upslope
ws : sediment settling velocity
εB : efficiency of bed load transport (εB = 0.10)
εs : sediment settling velocity (=0.02)
[]. : is a time average over many waves
qb0 : bed load on horizontal bed
qbs : slope effect on bed load transport
qs0 : suspended load on horizontal bed
qss : slope effect on suspended load

Van Rijn (1993) suggested a new formula for time averaged:

[
qb = 0.25αd 50 D*−0.3 τ b ,cw / ρ ] [(τ
0.5
b ,cw − τ b ,cr ) / τ b ,cr ]
1.5
(171)

where

D* : dimensionless grain parameter (-)

d 50 : median of grain diameter (m)
α : calibration factor
τ b,cw : instantaneous shear stress under wave-current (N/m2)
τ b,cr : critical shear stress (N/m2)

Sediment transport under wave and current on the seabed may be found using its two
components, which are cross-shore and long shore transport, because their mechanisms
are different.

8.3. Cross shore sediment transport

One of the most important factors determining whether a beach will build up or erode is
the transport of sediment in the onshore-offshore direction. Waves can cause a net sand
transport even when no other currents are present and the net mass transport of the
waves is zero. Indeed, under some conditions a negative sand transport, opposite to the
wave propagation direction, has been observed even when the long-term averaged
velocity has been positive. Of course, if the resulting current in the direction of wave
propagation is sufficiently large, there will be a positive sand transport.

Because the mass transport velocity offshore of the breakers is in the direction of wave
propagation in the immediate vicinity of the bed, at any rate if the bed is not too rough,
this current tends to carry sediment in toward the shore. This tendency is opposed by rip
and turbidity currents and by the component of gravity acting back down the slope. In a
closed system with the wave conditions held constant the bed would eventually reach an
equilibrium profile for which these various factors would exactly balance. In real life,

290
 Coastal Hydrodynamics and Sediment Transport

however, conditions are constantly changing and a beach is rarely in equilibrium with
the existing wave field (Sleath, 1984).

The cross-shore sediment transport can be described using the basic equation of
sediment transport, denoted by:

h +η t ′
1
qt =
t′ ∫ ∫ C ( z, t )u( z, t )dtdz
0 0
(172)

where qt; sediment transport rate expressed in (m3/ms), t ′ ; integration time (s), h; local
water depth (m), η; instantaneous water surface elevation (m), C(z, t); instantaneous
concentration of material (-), u(z, t); instantaneous velocity component (m/s), z;
elevation above the bed level (m), t; time(s).

Unlike the long shore sediment transport models it is not possible to reduce the basic
equation for cross-shore sediment transport because the velocity changes with place and
time.

Dean (1973) defined a parameter to predict not only the direction but also, to some
extent, the magnitude of the onshore-offshore transport.

0.6 H 0 gT
(173)
πL0 ws

According to Dean, onshore transport occurs when this parameter is less than unity and
offshore transport when it is greater. Bailard (1981) applied Bagnold’s total load model
to the problem of onshore-offshore transport.

Van Rijn (1993) derived a full, comprehensive theory of sediment transport under
waves based on a mixture of fundamental physics and empirical results (Eq. 171). The
method applies to total (bed load plus suspended load) sediment transport in combined
waves and currents on horizontal and sloping beds. The methods need an algebraic
solution by computer aid.

8.4. Longshore sediment transport

The product of the instantaneous concentration and the instantaneous velocity can
describe transport. Generally the sediment transport through a plane of unit width and
height equal to the water depth is denoted by Eq. 172. Since velocity is almost
independent of time for long shore transport the time-averaged concentration can be
used instead of the instantaneous concentration. The formula for the long shore
sediment transport can therefore be reduced to a much more convenient equation:

h +η

qt = ∫ u ( z ) ⋅ c( z ) ⋅ dz
0
(174)

291
Sediment Transport Technology

The most popular formula is the CERC formula for long shore sediment transport. The
CERC formula did not tackle sediment transport by predicting the sediment
concentration and sediment velocity, but by relating the transport to the available energy
(Fig. 17).

The CERC formula is associated with a bulk-energy model and describes the total
transport through a unit (beach) width of total breaker zone.

Q = 2.03 ⋅ 10 3 f H o5 / 2 F (α o ) (175)

where, f; wave frequency, Ho; deep-water significant wave height, αo; deep water
incident wave angle between wave crest and shoreline, and, F(αo); the direction term.

[
F (α o ) = (cos α o )
1/ 4
sin 2α 0 ]
The Bijker suggest an alternative formula, in contrast to the CERC formula, a detailed
prediction of the sediment transport and the transport distribution through the breaker
zone.

Net long shore transport rate Qn is defined as the difference between the amounts of
littoral drift transported from the right and to the left past a point on the shoreline in a
given time period. Gross long shore transport rate Qg is the sum of the amounts of
littoral drift transported to the right and to the left, past a point on the shoreline in a
given time period.

8.5. Beach profiles

The shape of a beach (its water depth as a function of distance from offshore) is called
the beach profile (Kamphuis, 2000). An essential feature of a coastline is its profile,
which is the result primarily of the cross-shore sediment transport process. Typical
examples of this wave induced sediment movement are well-known summer (swell
profile) and winter (storm profile) beach profiles (Fig. 18).

Figure 17. Wave approaching energy flux components (Yüksel, 2002)

292
 Coastal Hydrodynamics and Sediment Transport

Figure 18. Summer and winter beach profiles (Velden, 1989)

• In winter the mean wave conditions are more intense than in summer. Material is
moved offshore, the beach slope in the surf zone becomes gentler. As a result one or
more offshore bars may form.
• In summer the mean wave height decreases, the mean wave period usually
increases. Beach recovery begins. Sand is transported onshore and a new berm is
created by the run up. Because of this onshore sediment transport, the slope in the
surf zone becomes steeper.
• The transition between a summer and a winter beach profile in the range of 0.025 to
0.03.

When considering cross-shore sediment transport we refer to sediment transport in a

direction perpendicular to the coast. This transport takes place mainly as a result of
changing wave conditions. If one specific wave field attacks a sandy coast, after a while
an equilibrium beach profile will develop. An often-used description of the equilibrium
beach profile is

h( y ) = Ay 2 3 (176)

where, h is the depth, and, y is the distance from the shore line. A dimension factor
having units of length to the one third power.

A beach profile has a practical seaward limiting depth, where the wave conditions can
no longer change the profile. Sediment will still move back and forth, but there is no
perceptible change in depth. Hallermeier (1981) discusses this critical depth or closure
depth approximates it is as (Kamphuis, 2000):

d cr = 1.6 H s ,12 (177)

where Hs,12 is significant wave height which occurs 12 hrs/yr on average.

8.6. Coastal changes

Both longshore and cross-shore sediment transport contribute the changes of coastal line
and profile shape i.e. coastal morphology. The behavior of the coast (accretion and
erosion) may be predicted using a sediment transport formula, an equation of motion
and a continuity equation. There are some numerical solutions to predict the coastal
morphology such as one or multiple line theories.

293
Sediment Transport Technology

References

1. Bailard, J. A., “An energetics total load sediment transport model for a plane sloping
beach”, J. Geophys. Res., 86 (C11), 10938-10954, (1981).
2. Dean, R. G., “Heuristic models of sand transport in the surf zone”, Proc. Conf. Eng.
Dyn. Surf Zone, Sydney, Ausralia, (1973).
3. Douglass, S. L., “Saving America’s Beaches”, Advanced Series of Ocean
Engineering, Vol. 19, World Scientific, London, (2002).
4. Fredsøe, J., and Deigaard, R., “Mechanics of Coastal Sediment Transport”, World
Scientific, (1992).
5. Gourlay, M. R., and Apelt, C. J., “Coastal hydraulics and sediment transport in a
coastal system”, Univ. of Queensland, Australia, (1985).
6. Kamphuis, J. W., “Introduction to Coastal Engineering and Management”, World
Scientific, (2000).
7. Nielsen, P., “Coastal Bottom Boundary Layers and Sediment Transport”, World
Scientific, (1992).
8. Novak, P., Moffat, A. I. B., Nalluri, C., Narayanan, R., “Hydraulic Structures”,
Spon Press, (1996 and 2001).
9. Silvester R., and Hsu, J. R. C., “Coastal Stabilization”, World Scientific, (1997).
10. Sleath, J. F. A., “Sea Bed Mechanics”, Ocean Engineering Series, Wiley, (1984).
11. Soulsby, R., “Dynamics of marine sands”, H.R. Wallingford, Thomas Telford,
London, (1997).
12. Van Rijn, L. C., “Handbook of sediment transport by current and waves”, Delft
Hydraulics, Delft, the Netherlands, (1989).
13. Van Rijn, L. C., “Principles of sediment transport in rivers, estuaries and coastal
seas”, Aqua Publication, the Netherlands, (1993).
14. Van Rijn, L. C., “Principles of coastal morphology”, Aqua Publication, the
Netherlands, (1998).
15. Velden, E. T. J. M., “Coastal Engineering”, Vol. II, Delft Univ. of Technology,
(1989).
16. Yüksel, Y., “Coastal Sediment Transport”, Lecture Notes (unpublished, in Turkish),
(2002).
17. Yüksel, Y., Çevik, E., and Çelikoğlu, Y., “Coastal and Harbour Engineering”,
Chamber of Civil Engineers in Turkey, Ankara (in Turkish), (1998).

294
 EXERCISES ON UNIFORM FLOW IN OPEN CHANNEL, BED
LOAD, SUSPENDED LOAD AND TOTAL LOAD

Dr. Vehbi ÖZAYDIN

Technical Research and Quality Control Department
State Hydraulic Works (DSI)
Ankara - Turkey
e-mail: ozaydinv@dsi.gov.tr

EXAMPLES
1. Triangular channel with Manning’s roughness coefficient n=0.012 has a discharge
capacity of 10 m3/s. Side slopes of the channel is z=3 (1V:3H). Find the critical
depth, critical velocity and critical slope of the channel.

2. Rectangular flume is carrying a discharge of 1.42 m3/s at a slope of 0.0028.

Determine the dimensions of the channel cross section if width must be equal to
twice the depth (n=0.017).

3. A trapezoidal channel is carrying 75 m3/s water. The maximum allowable velocity

of flow is 3 m/s to avoid scouring. Determine the depth of flow and width of the
channel bottom, if the hydraulic radius of the channel is one half of the water depth
of flow. Also, determine the slope of the channel bottom. Use Manning’s roughness
coefficient n=0.02 and side slope z=2 (1V:2H)

4. A rectangular channel with a width of 3 m and n value of 0.014 is carry water at a

flow rate of 13.4 m3/s. Determine the critical depth, velocity and channel slope.

5. A 3 m wide rectangular open channel carries a discharge of 8.5 m3/s. The bottom
slope is adverse of 0.002 in the direction of flow. If the water depth decreases from
2.1 m to 1.65 m in a 150 m length of channel, determine Manning’s n coefficient.

6. Derive that best hydraulic section of trapezoidal section is half of a hexagonal.

7. A lined trapezoidal canal will be designed for Q=11.5 m3/s. Taking n=0.025,
S=0.0016, and umin=0.5 m/s, determine the optimum section dimensions. Unit cost
of excavation, lining and land purchasing are $18.75/m3, $4.75/m2, and $11.25/m2,
respectively. Side slope of the channel is z=1.5 (1V:1.5H).

8. Design discharge of a lined irrigation canal is 1 m3/s with n=0.016 and S=0.003.
Determine the water depths (y), bottom width (b), wetted perimeter (P), and the flow
areas (A) using best hydraulic section approach for: i) trapezoidal canal, ii)
rectangular canal, iii) semi-circular canal. For freeboard use f = 0.2(1+y).

295
Sediment Transport Technology

9. A trapezoidal canal carrying 9 m3/s is to be built with a non-erodible bed. The

bottom slope is S=0.0016, and Manning’s roughness coefficient is n=0.016. The side
inclination is z=2. Determine the channel dimensions. Also, what will be the
dimensions if best hydraulic section is used. Draw the cross sections.

10. Design an unlined trapezoidal canal to carry 5 m3/s on a non-colloidal alluvial silt.
Take Manning’s roughness coefficient n=0.02. Maximum allowed flow velocity is
umax=0.60 m/s. The bottom slope and side inclination are S=0.0004 and z=1.5,
respectively. Use maximum permissible velocity approach.

11. Design an unlined trapezoidal channel using tractive force approach to carry Q=1.5
m3/s. Channel material is sandy soil where D25 is 10 mm and very angular. Assume
z=3, n=0.02 and S=0.0016.

12. Design an unlined trapezoidal channel using tractive force approach to carry Q=7.5
m3/s. The bed material is mainly clay with a void ratio of 0.4. Side slope, z=2.
Manning’s roughness coefficient n=0.03 and bed slope is 0.0015.

13. A natural wide river carries bed load and total load. Calculate the amount of bed
load and total load by different methods. Characteristics of the bed material and
river are as follows:
Unit weight of the sediment, γs, is 2,650 kg/m3
Unit weight of the water, γ, is 1,000 kg/m3
Angle of repose of the sediment, β, is 33˚
Width of the river, B, is 100 m
Slope of the river, S, is 0.0004
Side inclination of the river, z, is 1.5 (1.5H:1V)
Water depth, h, is 3.5 m
Discharge, Q, is 1,530 m3/s
Temperature of water is 20˚C (68˚F)
Viscosity, υ, 1.1*10-6 m2/s
Fraction of the sediment, ib, is 100 %
Fall velocity for d50=0.001 m, w, is 0.18 m/s

d30 d35 d50 d65 d90 d95

Size (m) 0.0004 0.00065 0.001 0.0015 0.003 0.004

14. Calculate the rating curve of the total load of a river by i) Einstein method, ii)
Laursen method, and iii) Graf et all. method. Use previous example’s data for any
missing data.

Size distribution Range Percent

(mm) retained
3.3
4

ib (%)
3

100
Percent finer (%)

1 0.4 - 0.65 11
1.5

80
2 0.65 - 1 20
1

60
0.65

3 1 - 1.5 25
0.4

40

20 4 1.5 - 3 22
0 5 3 - 3.3 12
0.1 1 10 6 3.3 - 4 8
Sediment diameter (mm) Σ 98

296
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

15. Find the expression for the loss of hL in a channel of length L in terms of the
velocity head and the hydraulic radius R.

16. A channel has vertical walls 1.5 m apart and a 0.5 m semi circular bottom. The bed
slope is 0.0004. What will be the water value of Chezy coefficient C if the discharge
is 1.0 m3/s?

17. Water flows on a wide rectangular channel of concrete (n=0.010) laid on a slope of
0.0015. What will be the depth and rate of flow for critical condition in this
channel?

18. A rectangular concrete channel 5 m wide carries a flow rate of 12 m3/s. What will be
the water depth, if the bottom slope and the Manning roughness coefficient are
0.0002 and 0.011, respectively?

19. A metal pipe of 0.5 m diameter flows half full at a slope of 0.002. What is the flow
rate? Take n=0.026.

20. A rectangular channel has a width of B of 2.4 m and a slope of 1 in 400. What will
be the depth of water y if the rate of flow Q is 8.5 m3/s and the coefficient C in the
Chezy formula is 51 in SI units?

21. For the uniform flow condition (clear water) in a wide channel with a=0.9 m and
n=0.02, Determine:
a) The value of b?
b) The depth below the surface at which the velocity u is equal to the mean
velocity?
c) The average of the velocities at 0 to 1 (y/y) depths?

a
b

7.5 cm

10 cm

5 cm y

22. Slope of the channel shown below is 0.1. What is the discharge Q?

n3 = 0.030
2m
n2 = 0.025
5m
n1 = 0.020

25 m 6m 20 m

297
Sediment Transport Technology

SOLUTIONS
Solution 1.
From Table 1:
T
A = zy2 = 3yc2
P = 2y 1+ z2
1 yc
z R = zy / 2 1 + z 2
T = 2zy = 6yc

For the critical flow:

Q 2T
3
= 1 ⇒ [(102)(6yc)] / [(3yc2)39.81] = 1 ⇒ yc = 1.178 m
A g
uc = Q/A = 10/(3*1.1782) = 2.402 m/s
R = A/P = (3*1.178) / (2* 10 ) = 0.559 m

From Manning equation:

1
u c = R 2 / 3 S c1 / 2 ⇒ Sc = [ (0.012) (2.402) / (0.559)2/3 ]2 = 0.00181
n

Solution 2.
Channel width, b
Channel depth, y = b/2, Insert these values to Manning equation

A = by = b·b/2 = b2/2 and P = 2y+b = 2(b/2)+b = 2b

R = A/P = (b2/2)/(2b) = b/4
1.42 = [(b2/2)/(0.017)][(b/4)2/3] (0.0028)1/2 ⇒ 2.3 = b8/3
b = 1.366 m and y = 0.683 m

Table 1. Geometric elements of channel sections

WETTED HYDRAULIC HYDRAULIC
AREA TOP WIDTH
SECTION PERIMETER RADIUS DEPTH
A T
P R D

by
by b + 2y b y
b + 2y

(b + zy )y (b + zy )y
(b + zy )y b + 2y 1+ z2 b + 2 zy
b + 2y 1+ z 2 b + 2 zy

zy
zy 2 2y 1+ z2 2 zy (1 2)y
2 1+ z2
(sin (1 2)θ )d 0
⎛ sin θ ⎞ ⎛ θ − sin θ ⎞
1 8 (θ − sin θ ) d 02 (θ 2)d 0 1 4 ⎜1 − ⎟d 0 or 1 8 ⎜⎜ ⎟⎟d 0
⎝ θ ⎠ 2 y (d 0 − y ) ⎝ sin θ 2 ⎠

298
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Solution 3.
From Table 1:
A = by+zy2
y
P = b+2y(1+z2)0.5
R = A/P = y/2
b
By substituting these values into hydraulic radius equation:
R = A/P ⇒ (y/2) = (by+2y2) / (b+4.472y) ⇒ by = 0.472 y2 ⇒ b = 0.472y

From continuity equation:

A = Q/u ⇒ by + 2y2 = 75 / 3 = 25 m2

By substituting b = 0.472y to the continuity equation:

0.472y2 + 2y2 = 25 ⇒ y = 3.18 m
b = 0.472 * 3.18 = 1.50 m
A ≅ 25 m2, P = 15.72 m, R = 1.59 m

From Manning equation:

2 2
⎛ u⋅n ⎞ ⎛ 3 * 0.02 ⎞
S = ⎜ 2/3 ⎟ = ⎜ 2/3 ⎟
= 0.0194
⎝R ⎠ ⎝ 1.59 ⎠

Solution 4.
Determination of critical flow depth:

q2
yc = 3 where q is unit discharge:
g

q = Q/B = 13.4/3 = 4.467 m3/s/m

From equation yc = 1.267 m

uc = Q/A = 13.4/[1.267*3] = 3.525 m/s

R = A/P = (1.267*3) / (1.267+3+1.267) = 0.687 m

From Manning’s equation:

Sc = (3.525*0.014) / (0.6872/3)2 = 0.00402

Solution 5.
Continuity equation: Hydraulic radius equation:
u1 = Q/A = 8.5 / (2.1*3) = 1.349 m/s R1 = A/P = (2.1*3)/(2.1+3+2.1) = 0.875 m
u2 = Q/A = 8.5 / (1.65*3) = 1.717 m/s R2 = A/P = (1.65*3)/(1.65+3+1.65) = 0.786 m
uave = (1.349+1.717)/2 = 1.533 m/s Rave = (0.875+0.786)/2 = 0.83 m

299
Sediment Transport Technology

hl
u2/2g egl

u2/2g

2.1 m 1.65 m

2
1
150 m

Energy equation between section 1 and 2:

z1+y1+u12/2g=z2+y2+u22/2g+hl
0+2.1+1.3492/(2*9.81) = 0.002*150+1.65+1.7172/(2*9.81)+hl ⇒ hl = 0.0925 m

Energy slope:
S = hl / L

From Manning’s equation:

R2/3
u ave = ave S
n

1.533 =
(0.83)
23
∗
⎛ 0.0925 ⎞
12

⇒ n = 0.014
⎜ ⎟
n ⎝ 150 ⎠

Solution 6.
A=(b+zy)y (Table 1) (1a)
A
From Eq. 1a b = − zy (1b)
y
P = b + 2y 1+ z2 (Table 1) (2)

Insert b in Eq. 1b into Eq. 2

A
P= − zy + 2 y 1 + z 2 (3)
y
∂P A
= − 2 − z + 2 1+ z2 = 0 (4)
∂y y
∂P 1
= −y + 2y 2z = 0 (5)
∂z 2 1+ z2
1+ z2
From Eq. 5 z= (6)
2
By trial and error z = 0.57735 = 60o

300
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

(b + zy ) y
Insert Eq. 1a to Eq. 3 − 2
− z + 2 1+ z2 = 0 (7)
y
When b is solved b = 2 y( 1 + z 2 − z) (8)

Insert z to Eq. 8 b = 1.15y

This is half of a hexagonal.

Solution 7.
From Table 1:
f
A = by+zy2
y P = b+2y(1+z2)0.5
R = A/P

b f is freeboard, f = 0.2(1+y)

From Manning equation:

nQ 0.025 *11.5 (by + 1.5 y 2 ) 5 / 3
AR 2 / 3 = = = 7.19 =
S 0.0016 (b + 2 y 1 + 1.5 2 ) 2 / 3

Total cost of construction:

ΣC = Cexc+Clin+Cpur
( ) ( )
∑ C = by* + 1.5 y*2 18.75 + b + 2 y* 1 + 1.5 2 4.75 + (b + 2 *1.5 y* )11.25
where y* = y + f

By increasing b from 0.1 to 10 m corresponding y is calculated by the equation and the

total cost. Optimum b = 2.10 m, y = 1.65 m, f = 0.53 m with ΣC = $362.77/m.

A = 7.55 m2
uave = Q/A = 1.52 m/s >0.5 m/s ⇒ no siltation
uave < umax=3 m/s for concrete lining ⇒ √
Required condition for stability y/yc = 1.71 > 1.1 ⇒ √

301
Sediment Transport Technology

Solution 8.
i) Best hydraulic section of a trapezoid is half of a hexagonal
A = 3y 2 P = 2 3y R = y/2 (Table 2)

From Manning equation:

Qn 1 * 0.016
AR 2 / 3 = = = 0.292 = ( 3 y 2 )( y / 2) 2/3
S 0.003
y = 0.61 m, P = 2.11 m, A = 0.64 m2,
f = 0.73 m, b = y/cos30 = 0.70 m

ii) Best hydraulic section of a rectangle is half of a square

A = 2y2 P = 4y R = y/2 (Table 2)

Qn 1 * 0.016
AR 2 / 3 = = = 0.292 = (2 y 2 )( y / 2) 2/3
S 0.003
y = 0.58 m, P = 2.32 m, A = 0.67 m2,
f = 0.70 m, b = 2y =1.16 m

iii) Best hydraulic section of a semi-circle is half of a circle

A = πy2/2 P = πy R = y/2 (Table 2)

Qn 1 * 0.016 π
AR 2 / 3 = = = 0.292 = ( y 2 )( y / 2) 2/3
S 0.003 2
y = 0.63 m, P = 1.98 m A = 0.63 m2,
f = 0.76 m, b = 2y =1.26 m

Solution 9.

A = by+zy2 P = b + 2y 1+ z2 R = A/P T = b+2zy (Table 1)

f = 0.2(1+y) (freeboard)

Table 2. Geometric elements of best hydraulic sections

WETTED HYDRAULIC HYDRAULIC
AREA TOP WIDTH
SECTION PERIMETER RADIUS DEPTH
A T
P R D
Trapezoid, half of
a hexagonal 3y 2 2 3y (1 2)y 4 3
y (3 4)y
3
Rectangle, half of
a square
2y2 4y (1 2)y 2y y

Triangle, half of
a square
y2 2 2y
2
y 2y (1 2)y
4
π π
Semicircle y2 πy (1 2)y 2y y
2 4
Parabola
T = 2 2y
4 2 2
y
8 2
y (1 2)y 2 2y (2 3)y
3 3

302
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 1. Experience curves showing bottom width and water depth of lined channels

Select b or y. It is better to take b. Take b = 2.3 m (from Fig. 1).

From Manning equation:

nQ 0.016 ∗ 9 ( 2 .3 y + 2 y 2 ) 5 / 3
AR 2 / 3 = = = 3 .6 =
S 0.0016 ( 2 .3 + 2 y 1 + 2 2 ) 2 / 3
By trial and error y = 1.064 m

A = 2.3*1.064+2*1.064*1.064 = 4.71 m2
P = 2.3+2*1.064*(5)0.5 = 7.06 m
R = A/P = 0.667 m
u = Q/A = 9/4.71 =1.91 m/s < umax=3 m/s for erosion
> umin= 0.5 m/s for deposition

The critical depth is found from:

Q 2T 9 2 (2.3 + 4 y c )
=1 =1
A3 g (2.3 yc + 2 yc2 )3 9.81
By trial and error, yc = 0.89 m

y/yc = 1.064/0.89 = 1.2 > 1.1 √

f = 0.2(1+y) = 0.41 m

Solution with best hydraulic section

A = 3y 2 P = 2 3y R = y/2 (Table 2)

Qn 9 * 0.016
AR 2 / 3 = = = 3.6 = ( 3 y 2 )( y / 2) 2/3
S 0.0016
y = 1.56 m, P = 5.40 m A = 2.7 m2,
f = 0.51 m, b = y/cos30 = 1.80 m

303
Sediment Transport Technology

0.41 m 0.51 m

z=2 z=0.577
(1V:2H) 1.064 m (1V:0.577H) 1.56 m

2.3 m 1.80 m

Solution 10.
From Manning equation:
u n 0.6 * 0.02
R 2 / 3 = max = = 0.6 ⇒ R = 0.465 m
S 0.0004

A = Q/umax =5/0.6 = 8.33 m2

P = A/R = 8.33/0.465 = 17.91 m

A = by+zy2 = by+1.5y2 8.33 = by+1.5y2 (9)

P = b+2y(1+z2)0.5 = b+3.61y 17.91 = b+3.61y (10)

Take b from Eq. 10 as b = 17.91-3.61y and insert to Eq. 9 and re-arrange:

2.11y2-17.91y+8.33 = 0
y1=8 m b1 = -10.95 m these values are not possible (imaginary)
y2 = 0.494 m b2 = 16.13 m √

Since y ≠ 1, umax should be corrected by a correction factor:

CF = y/3+0.70 0.3 < y < 0.6
CF = y/4+0.75 0.6 < y < 1.0

From Fig. 2 CF = 0.87

umax = 0.6*0.87 = 0.522 m/s

Calculation of CF by iteration:
y CF (umax)new R A P y1 b1 y2 b2
0.494 0.87 0.522 0.377 9.58 25.41 0.39 24.0 11.65 -16.66
0.390 0.83 0.498 0.351 10.04 28.60 0.361 27.3 13.2 -19.06
0.361 0.82 0.492 0.345 10.16 29.46 0.354 28.2 13.6 -19.66
0.354 0.818
Since CFi and CFi-1 is approximately equal then √

The critical depth is found from:

Q 2T 1.5 2 (1.92 + 6 y c )
=1 =1
A3 g (1.92 yc + 3 yc2 )3 9.81
By trial and error yc = 0.147 m

y/yc = 0.354/0.147 = 2.4 > 1.1 √

304
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 2. Velocity correction factor

u = Q/A = 5/10.16 = 0.49 m/s ≅ umin = 0.5 m/s

Take a freeboard of f = 0.2(1+y) = 0.27 m

f=0.27 m
Disadventage of this method:
y=0.35 m z=1.5
it gives a larger bottom value
(1V:1.5H)

b=28.2 m

Solution 11.
Side inclination, φ:
1
tan φ = φ = 18.3o
3

Angle of repose of very angular soil for 10 mm, θ = 34.8o (from Fig. 3)
Permissible unit tractive force for 10 mm sandy soil, τp = 0.74 kgf/m2 (from Fig. 5)

sin φ
K = 1− =0.84 if K > 0.78 bottom is critical
sin θ
if K < 0.78 side is critical

305
Sediment Transport Technology

Figure 3. Angle of repose

Figure 4. Permissible velocities for non-cohesive soils

306
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 5. Unit tractive forces for canals in non-cohesive material

Assume b/y = 4 from Fig. 6:

C1 = 0.77 (sides) C2= 0.96 (bottom) (from Fig. 6)
C1 γ y S ≤ K τp (0.77) (1000) y (0.0016) = (0.84) (0.74) y = 0.51 m (side)
C2 γ y S ≤ τp (0.77) (1000) y (0.0016) = 0.74 y = 0.48 m (bottom)

Select the smaller one:

y = 0.48 m b = 4y = 1.92 m
A = by+zy2 = 1.92*0.48 + 3*0.482 = 1.61 m2
P = b+2y(1+z2)0.5 = 1.92+2*0.48+(1+32)0.5 = 4.96 m

(a) (b)
Figure 6. Maximum unit tractive forces in terms of γ·y·S (a) Bottom, (b) Sides

307
Sediment Transport Technology

R = A/P = 1.61/4.96 = 0.32 m

From Manning formula:

1.61
Q= (0.32)2 / 3 0.0016 ≅ 1.5 m3/s
0.02

The critical depth is found from:

Q 2T 1.5 2 (1.92 + 6 y c )
=1 =1
A3 g ( ) 3
1.92 y c + 3 y c2 9.81
By trial and error yc = 0.33 m

y/yc = 0.48/0.33 = 1.47 > 1.1 √

f = 0.2(1+y) = 0.3 m
u = Q/A =1.5/1.61 = 0.93 m/s > umin = 0.6 m/s no deposition √

f=0.30

y=0.48 z=3
(1V:3H)

b=1.92

Figure 7. Permissible velocities for cohesive soils

308
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Solution 12.
Permissible unit tractive force for clay, τp = 1.75 kgf/m2 (Fig. 8)
Bottom is always critical for clay (i.e. cohesive soil)

Assume b/y = 1.6

C2= 0.855 (on bottom) (Fig. 6)
C2 γ y S ≤ τp (0.855) (1000) y (0.0015) = 1.75 y = 1.36 m b = 1.6y = 2.17 m

A = by+zy2 = 2.17*1.36 + 2*1.362 = 6.65 m2

P = b+2y(1+z2)0.5 = 2.17+2*1.36+(1+22)0.5 = 8.25 m
R = A/P = 6.65/8.25 = 0.81 m

From Manning formula:

6.65
Q= (0.81)2 / 3 0.0015 ≅ 7.5 m 3 /s
0.03

Figure 8. Unit tractive forces for canals in cohesive material

309
Sediment Transport Technology

The critical depth is found from:

Q 2T 7.5 2 (2.17 + 4 y c )
= 1 =1
A3 g ( ) 3
2.17 y c + 2 y c2 9.81
By trial and error yc = 0.83 m

y/yc = 1.36/0.83 = 1.64 > 1.1 √

f = 0.2(1+y) = 0.47 m
u = Q/A =7.5/6.65 = 1.13 m/s > umin = 0.6 m/s no deposition √

f=0.47 m

y=1.36 m z=2
(1V:2H)

b=2.17 m

Solution 13.
Cross-sectional area:
A = Bh+zh2 = 100*3.5+1.5*3.52 = 368.4 m2

Wetted perimeter:
P = B + 2h 1 + z 2 = 100 + 2 * 3.5 * 1 + 1.5 2 = 112.6 m

Hydraulic radius:
R = A/P = 368.4/112.6 = 3.27 m

Unit discharge:
q = Q/B = 1530/100 = 15.3 m3/s/m

Average flow velocity:

uave = Q/A = 1530/368.4 = 4.15 m/s

Shear stress:
τ0 = γRS = 1000*3.27*0.0004 = 1.31 kg/m2

Shear velocity:
u* = gRS = 9.81 * 3.27 * 0.0004 = 0.11 m/s

Average diameter of bed material:

dave = d50 = 0.001 m

Critical shear stress:

ud 0.11* 0.001
Re = * ave = = 100
ν 1.1*10 − 6

310
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

From Shield’s diagram (Fig. 9):

τ cr
= 0.045
(γ s − γ )d ave
τcr = 0.045*(2650-1000)*0.001 ⇒ τcr = 0.074 kg/m2

Figure 9. Shields diagram with White data added

Figure 10. Modified Shields diagram

311
Sediment Transport Technology

Figure 11. Erosion-deposition criteria for uniform particles (after Hjulström, 1935)

Figure 12. Critical shear stress for quartz sediment in water as function of grain size
(Shields, 1936 and Lane, 1955)

312
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 13. Critical shear stress as function of grain diameters (After Lane, 1953)

i) Bed load equations

Du Boys equation (1879)

Applicable to uniform, coarse grain size and high velocity (steep slopes) flows.

dave < 4 mm √

qb = ΨDτ O [τ O − τ cr ]

where, qb is total bed load in terms of kg/s m, ΨD is obtained from Fig. 14 by dave, τcr is
obtained from Fig. 14 by dave.

From Fig. 14, τcr = 0.16 kg/m2 and ΨD = 1.75 m3/(kg s)

kg
qb = 1.75*1.31*(1.31-0.16) = 2.64 = 22,810 tons/day
sm

Einstein equation (1950)

Applicable to:
• uniform grains
• fine and medium size sediments, small velocities
• for the small values of φ, the equation gives less amount of bed load

313
Sediment Transport Technology

Figure 14. Evaluation of parameters of Du Boys for metric units (After Zeller, 1963)

⎛γ −γ ⎞ d ave
Ψ = ⎜⎜ s ⎟⎟
⎝ γ ⎠ RS
qb
φ= 0.5
⎛γs −γ ⎞
⎜⎜ ⎟⎟ γ s gd ave
3

⎝ γ ⎠

where, qb is total bed load in terms of kg/s m, φ is obtained from Fig. 15 by Ψ.

Figure 15. Plot of Einstein functions, Φ* and Ψ* (After Einstein, 1950)

314
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

⎛ 2650 − 1000 ⎞ 0.001

Ψ =⎜ ⎟ = 1.26
⎝ 1000 ⎠ 3.27 * 0.0004

From Fig. 15 φ = 5 by Ψ=1.26

qb
5= 0.5
⎛ 2650 − 1000 ⎞ 3
⎜ ⎟ 2650 9.81 * 0.001
⎝ 1000 ⎠
kg
qb = 1.67 = 14,428 tons/day
sm

Einstein-Brown equation (1950)

Applicable to:
• coarse material and high velocities
• well sorted material

1 τO
=
Ψ (γ s − γ )d ave
qb
φ=
⎛γ ⎞ 3
γ s F1 g ⎜⎜ s − 1⎟⎟d ave
⎝γ ⎠
2 36ν 2 36ν 2
F1 = + −
3 3 ⎛γs ⎞ 3 ⎛γs ⎞
gd ave ⎜⎜ − 1⎟⎟ gd ave ⎜⎜ − 1⎟⎟
⎝γ ⎠ ⎝γ ⎠

where, qb is total bed load in terms of kg/s m, φ is obtained from Fig. 16 by 1/Ψ.
3
⎛1⎞ 1
φ = 40⎜ ⎟ (for > 0.09)
⎝Ψ⎠ Ψ

Figure 16. Function φ = f (1 / Ψ ) for Einstein Brown

315
Sediment Transport Technology

2 36 * 0.00000112 36 * 0.00000112
F1 = + − = 0.82 − 0.05 = 0.77
3 3 ⎛ 2650 ⎞ 3 ⎛ 2650 ⎞
9.81 * 0.001 * ⎜ − 1⎟ 9.81 * 0.001 * ⎜ − 1⎟
⎝ 1000 ⎠ ⎝ 1000 ⎠
1 1.31
= = 0.794
Ψ (2650 − 1000) * 0.001

Since 1/Ψ>0.09 φ = 40*(0.794)3 = 20

qb
20 =
⎛ 2650 ⎞
2650 * 0.77 * 9.81 * ⎜ − 1⎟ * 0.0013
⎝ 1000 ⎠
kg
qb = 5.17 = 44,669 tons/day
sm

Kalinske equation (1947)

qb
φ=
u* d ave γ s
τ cr
Ψ=
τ0

where, qb is total bed load in terms of kg/s m, φ is obtained from Fig. 17 by Ψ, τcr is
obtained from Shields diagram (Fig. 9).

0.074
Ψ = = 0.056
1.31

From Fig. 17 φ = 2.1 by Ψ = 0.056

Figure 17. Kalinske’s bed load equation (after Kalinske, 1947)

316
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

qb
2 .1 =
0.11 * 0.001 * 2650
kg
qb = 0.61 = 5,290 tons/day
sm

Meyer-Peter and Muller equation (1948)

Applicable to:
• dave > 0.4 mm
• 0.04 % < S < 2 %
• coarse size sediments
1.5
⎛k ⎞
0.25ρ G = ⎜⎜ s
1/ 3 2/3
b
⎟⎟ γRS − 0.047(γ s − γ )d ave
⎝ kr ⎠
⎛ γ ⎞
qb = Gb ⎜⎜ s ⎟⎟
⎝γs −γ ⎠
u 1
k s = 2 / 3ave 1 / 2 =
R S n
26
kr = 1/ 6
d 90

where, qb is total bed load in terms of kg/s m, n is Manning roughness coefficient. If no

bed forms occur, then kr = ks.

u 4.15
ks = ave
= = 94.3
R S 3.27 2 / 3 S 1 / 2
2 / 3 1/ 2

26 26
k r = 1/ 6 = = 68.5
d 90 0.0031 / 6
1/ 3 1.5
⎛ 1000 ⎞ ⎛ 94.3 ⎞
⎟ Gb = ⎜ ⎟ * 1000 * 3.27 * 0.0004 − 0.047 * (2650 − 1000) * 0.001
2/3
0.25 * ⎜
⎝ 9.81 ⎠ ⎝ 68.5 ⎠
Gb = 2.3
⎛ 2650 ⎞ kg
qb = 2.3 * ⎜ ⎟ = 3.69 = 31,882 tons/day
⎝ 2650 − 1000 ⎠ sm

Schoklitsch equation (1930)

Applicable to sand bed streams that carry not considerable bed sediment in suspension.

qb = 2500S 3 / 2 (q − qc )
5/3
⎛ γ − γ ⎞ d ave 3/ 2
q c = 0.26⎜⎜ s ⎟⎟
⎝ γ ⎠ S
7/6

1
q = h 5 / 3 S 1 / 2 or q = u ave h
n

317
Sediment Transport Technology

where, qb is total bed load in terms of kg/s m, qc is critical unit discharge (m3/s/m), q is
unit discharge (m3/s/m).

5/3
⎛ 2650 − 1000 ⎞ 0.0013 / 2
q c = 0.26 * ⎜ ⎟ 7/6
= 0.17 m 3 /s/m
⎝ 1000 ⎠0.0004
kg
qb = 2500*0.00043/2*(15.3-0.17) = 0.30 = 2,592 tons/day
sm

Shields formula (1936)

qb γ s τ − τ cr
= 10 0
qSγ (γ s − γ )d ave
where, qb is total bed load in terms of kg/s m, q is unit discharge (m3/s/m), τ cr is
obtained from Shields diagram (Fig. 9).

qb * 2650 1.31 − 0.074

= 10 *
15.3 * 0.0004 *1000 (2650 − 1000) * 0.001
kg
qb = 0.02 = 172 tons/day
sm

Smart equation (1984)

Applicable to:
• dave > 0.4 mm
• 0.04 % < S < 20 %
• d90/d30 < 8.5

⎡⎛ d ⎞
0.2
⎤
φ = 4⎢⎜⎜ 90 ⎟⎟ S 0.6 Cθ 0.5 (θ − θ cr )⎥
⎢⎣⎝ d 30 ⎠ ⎥⎦
qb
φ=
⎛γ ⎞ 3
γ s g ⎜⎜ s − 1⎟⎟d ave
⎝γ ⎠
⎛ tan α ⎞
θ cr = θ 0cr cos α ⎜⎜1 − ⎟
⎝ tan β ⎟⎠
u ave
C=
ghS
hS
θ=
⎛γs ⎞
⎜⎜ − 1⎟⎟d ave
⎝γ ⎠

or

318
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

0.2
qb 4γ s ⎛ d 90 ⎞ ⎛ θ ⎞
= ⎜ ⎟⎟ S 1.6 ⎜1 − cr ⎟
q ⎛γs ⎞ ⎜⎝ d 30 ⎠ ⎝ θ ⎠
⎜⎜ − 1⎟⎟
⎝γ ⎠

where, qb is total bed load in terms of kg/s m, α is angle of slope S in degree, β is angle
of repose of the soil, q is unit discharge in m3/s/m, and, θ 0cr is the value found from
Shield’s diagram (Fig. 9) by:
ud
Re = * ave
ν

d90/d30 = 0.003/0.0004 = 7.5< 8.5 √

u* d ave 0.11 * 0.001

Re = = = 100
ν 1.1 * 10 −6

From Shield’s diagram (Fig. 9) ⇒ θ 0cr = 0.045

S = 0.0004 ⇒ α = 0.023º
⎛ tan 0.023 ⎞
θ cr = 0.045 * (cos 0.023) * ⎜1 − ⎟ ≅ 0.045
⎝ tan 33 ⎠

4.15
C= = 35.4
9.81 * 3.5 * 0.0004
3.5 * 0.0004
θ= = 0.85
⎛ 2650 ⎞
⎜ − 1⎟ * 0.001
⎝ 1000 ⎠

φ = 4 * (7.5)0.2 * (0.0004) 0.6 * 35.4 * (0.85) 0.5 * (0.85 − 0.045) = 1.44

qb
1.44 =
⎛ 2650 ⎞
2650 * 9.81 * ⎜ − 1⎟ * 0.0013
⎝ 1000 ⎠
kg
qb = 0.49 = 4,234 tons/day
sm

or

qb 4 * 2650 ⎛ 0.045 ⎞
* (7.5) * (0.0004)1.6 * ⎜1 −
0.2
= ⎟
15.3 ⎛ 2650 ⎞ ⎝ 0.85 ⎠
⎜ − 1⎟
⎝ 1000 ⎠
kg
qb = 0.51 = 4,406 tons/day
sm

319
Sediment Transport Technology

ii) Total load equations

Bishop et al. formula (1965)

⎛γ −γ ⎞ d 35
Ψ = ⎜⎜ s ⎟⎟
⎝ γ ⎠ RS
qT
φ=
⎛γs −γ ⎞ 3
γ s ⎜⎜ ⎟⎟ gd ave1
⎝ γ ⎠
φ = f (Ψ ) (Fig. 18)

where, qT is total load in terms of kg/s m.

⎛ 2650 − 1000 ⎞ 0.00065

Ψ =⎜ ⎟ = 0.82
⎝ 1000 ⎠ 3.27 * 0.0004

From Fig. 18 φ = 15 by Ψ = 0.82

qT
15 =
⎛ 2650 − 1000 ⎞ 3
2650 * ⎜ ⎟ * 9.81 * 0.001
⎝ 1000 ⎠
kg
qT = 5.06 = 43,718 tons/day
sm

Figure 18. ΦT and Ψ relation for different sand sizes

(after Bishop et al., 1965)

320
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Bagnold formula (1966)

Applicable to fully turbulent flows and its results are best for large transport rates.

⎛ γ ⎞ ⎛ e u ⎞
qT = ⎜⎜ s ⎟⎟τ o u ave ⎜⎜ b + 0.01 ave ⎟⎟
⎝γs −γ ⎠ ⎝ tgα w ⎠

where, qT is total load in terms of kg/s m, tgα is taken from Fig. 19, and, eb is taken
from Fig. 20.

From Fig. 20, eb = 0.1 by uave = 4.15*3.281=13.6 ft/s

Figure 19. The solid-friction coefficient (after Bagnold, 1966)

Figure 20. The bed load efficiency factor (After Bagnold, 1966)

321
Sediment Transport Technology

τ0 1.31
From Fig. 19, tgα = 0.44 by = = 0.794
(γ s − γ )d ave (2650 − 1000) * 0.001

⎛ 2650 ⎞ ⎛ 0.1 4.15 ⎞

qT = ⎜ ⎟ *1.31 * 4.15 * ⎜ + 0.01 * ⎟
⎝ 2650 − 1000 ⎠ ⎝ 0.44 0.17 ⎠
kg
qT = 4.12 = 35,597 tons/day
sm

Egiazaroff formula (1965)

Applicable to fully turbulent flows.

⎛ γγ s ⎞ ⎛τ ⎞
qT = ⎜⎜ ⎟⎟ S qK ⎜⎜ 0 − 1⎟⎟
⎝γs −γ ⎠ ⎝ τ cr ⎠

where, qT is total load in terms of kg/s m, K is a coefficient and equals to 0.015 for fully
turbulent region Shields diagram (Fig. 9).

τ cr
= 0.06 ⇒ τcr = 0.06*(2,650-1,000)*0.001 = 0.1 kg/m2
(γ s − γ )d ave

⎛ 1000 * 2650 ⎞ ⎛ 1.31 ⎞

qT = ⎜ ⎟ * 0.0004 *15.3 * 0.015 * ⎜ − 1⎟
⎝ 2650 − 1000 ⎠ ⎝ 0.1 ⎠
kg
qT = 89.2 = 770,688 tons/day
sm

Einstein equation (1950)

It consists of from two parts:

i) Hydraulic calculations are as follows:

1. Rb′ = 3.8 m
2. u*′ = gRb′ S = 9.81 * 3.8 * 0.0004 = 0.122 m/s
ν 0.0000011
3. δ = 11.6 = 11.6 = 0.000104 m
u*′ 0.122
d 65 0.0015
4. = = 14.35
δ 0.000104
d
5. χ = 1 from Fig. 21 by 65 = 14.35
δ
d 65 0.0015
6. ∆ = = = 0.0015
χ 1
⎛ R′ ⎞ ⎛ 3.8 ⎞
7. u = 5.75u*′ log10 ⎜12.27 b ⎟ = 5.75 * 0.122 * log10 ⎜12.27 ⎟ = 3.15 m/s
⎝ ∆ ⎠ ⎝ 0.0015 ⎠

322
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 21. Factor χ in velocity distribution equation

⎛ γ − γ ⎞ gd 35 ⎛ 2650 − 1000 ⎞ 9.81 * 0.00065

8. ψ 35′ = ⎜⎜ s ⎟⎟ 2 = ⎜ ⎟ = 0.71
⎝ γ ⎠ u*′ ⎝ 1000 ⎠ 0.122 2
⎛ γ − γ ⎞ d 35
ψ 35
′ = ⎜⎜ s ⎟⎟
⎝ γ ⎠ Rb′ S
u
9. = 58 from Fig. 22 by ψ 35
′ = 0.71
u*′′
10. u*′′ = u/58 = 3.15/58 = 0.054 m/s
u*′′ 2 0.054 2
11. Rb′′ = = = 0.75 m
gS 9.81 * 0.0004
12. Rb = Rb′ + Rb′′ = 3.8+0.75 = 4.55 m
13. h = 4.55 m (if wide channel h ≈ Rb)
14. AT = Bh+zh2 = 100*4.55+1.5*4.552 = 486.05 m2
15. Q = AT·u = 486.05*3.15 = 1531 m3/s ≈ 1530 m3/s √ (if not change Rb′ in step 1)
16. Determine X. X = 0.77∆ if ∆>1.8δ
X = 1.31δ if ∆<1.8δ
∆ = 0.0015 > 1.8δ = 0.00019 ⇒ X = 0.77*0.0015 = 0.0012
d
17. Y = 0.53 from Fig. 23 by 65 = 14.35
δ
2 2
⎡ ⎛ X ⎞⎤ ⎡ ⎛ 0.0012 ⎞⎤
18. β x = ⎢log10 ⎜10.6 ⎟⎥ = ⎢log10 ⎜10.6 ⎟⎥ = 0.862
⎣ ⎝ ∆ ⎠⎦ ⎣ ⎝ 0.0015 ⎠⎦

323
Sediment Transport Technology

Figure 22. Einstein-Barbossa bar resistance graph

Figure 23. Factor Y in Einstein’s bed load function (Einstein, 1950)

19. β = log10 10.6 = 1.025

2
⎛ β⎞
20. ⎜⎜ ⎟⎟ = 1.521
⎝ βx⎠
⎛ h⎞ ⎛ 4.55 ⎞
21. P = 2.3 log10 ⎜ 30.2 ⎟ = 2.3 log10 ⎜ 30.2 ⎟ = 11.41
⎝ ∆⎠ ⎝ 0.0015 ⎠

ii) Bed load calculations are as follows:

1. dave = 0.001 m ( d ave = d lower d upper if a range is specified)

324
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

2. ib = 100 %
3. Rb′ = 3.8 m
⎛ γ − γ ⎞ d ave ⎛ 2650 − 1000 ⎞ 0.001
4. ψ = ⎜⎜ s ⎟⎟ =⎜ ⎟ = 1.086
⎝ γ ⎠ R ′b S ⎝ 1000 ⎠ 3.8 * 0.0004
d ave 0.001
5. = = 0.87
X 0.0012
d
6. ξ = 1.3 from Fig. 24 by ave = 0.87
X
2
⎛ β ⎞
7. ψ * = ξY ⎜⎜ ⎟⎟ ψ = 1.3*0.53*1.521*1.086 = 1.138
⎝ βx ⎠
8. φ* = 7 from Fig. 15 by ψ * = 1.138
0.5
⎛γ −γ ⎞
9. ib qb = ibφ*γ s ⎜⎜ s ⎟⎟ 3
gd ave
⎝ γ ⎠
0.5
⎛ 2650 − 1000 ⎞
= 1 * 7 * 2650 * ⎜ ⎟ * 9.81 * 0.0013 = 2.36
⎝ 1000 ⎠
d 0.001
10. A = 2 ave = 2 = 0.00044
h 4.55
w 0.18
11. α = = = 3.685
0.4u*′ 0.4 * 0.122

Figure 24. Factor ξ in Einstein’s bed load function (Einstein, 1950)

325
Sediment Transport Technology

12. I1 = 0.072 from Fig. 25 by A = 0.00044 and α = 3.685

13. I2 = -0.62 from Fig. 26 by A = 0.00044 and α = 3.685
14. (1 + PI 1 + I 2 ) = 1+11.41*0.072-0.62 = 1.2
kg
15. iT qT = ib qb (1 + PI 1 + I 2 ) = 2.36*1.2 = 2.84 = 24,538 tons/day
sm

Figure 25. Function I1 in terms of AE for values of z (after Einstein, 1950)

326
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 26. Function I2 in terms of AE for values of z (after Einstein, 1950)

Engelund-Hansen equation (1966)

Applicable to rivers with fine sand beds.

⎛γ −γ ⎞ d ave
Ψ = ⎜⎜ s ⎟⎟
⎝ γ ⎠ RS

327
Sediment Transport Technology

2
⎛u⎞
⎜⎜ ⎟⎟
u
φ = 0.05 ⎝ *2.⎠5
Ψ
qT
φ= 0.5
⎛γs −γ ⎞
⎜⎜ ⎟⎟ γ s gd ave
3

⎝ γ ⎠

where, qT is total load in terms of kg/s m.

⎛ 2650 − 1000 ⎞ 0.001

Ψ =⎜ ⎟ = 1.26
⎝ 1000 ⎠ 3.27 * 0.0004
2
⎛ 4.15 ⎞
⎜ ⎟
⎝ 0.11 ⎠
φ = 0.05 = 39.8
1.26 5 / 2
qT
39.8 = 1/ 2
⎛ 2650 − 1000 ⎞ 3
⎜ ⎟ * 2650 * 9.81 * 0.001
⎝ 1000 ⎠
kg
qT = 13.42 = 115,945 tons/day
sm

Garde-Albertson equation
4
qT ⎡ τ0 ⎤
= 16⎢ ⎥
γ s u* d ave ⎣ (γ s − γ )d ave ⎦

where, qT is total load in terms of kg/s m.

4
qT ⎡ 1.31 ⎤
= 16 * ⎢ ⎥
2650 * 0.11 * 0.001 ⎣ (2650 − 1000) * 0.001⎦
kg
qT = 1.85 = 15,984 tons/day
sm

Graf et al. equation (1968)

⎛ γ − γ ⎞ d ave
Ψ = ⎜⎜ s ⎟⎟
⎝ γ ⎠ RS
10.39
φ = 2.52 (Fig. 27)
Ψ
C v u ave R
φ=
⎛γs −γ ⎞ 3
⎜⎜ ⎟⎟ gd ave
⎝ γ ⎠

328
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 27. ΦA vs. ΨA relation, with open-channel data only (After Graf et al, 1968)

qT = C v qγ s

where, qT is total load in terms of kg/s m.

⎛ 2650 − 1000 ⎞ 0.001

Ψ =⎜ ⎟ = 1.26
⎝ 1000 ⎠ 3.27 * 0.0004
10.39
φ= = 5.8
1.26 2.52
C v * 4.15 * 3.27
5.8 =
⎛ 2650 − 1000 ⎞ 3
⎜ ⎟ * 9.81 * 0.001
⎝ 1000 ⎠
Cv = 0.0000544
kg
qT = 0.0000544*15.3*2650 = 2.2 = 19,008 tons/day
sm

Laursen’s formula (1958)

Applicable to natural sediments with γ s = 2650 kg/m3.

7/6
⎛d ⎞ ⎛τO ⎞ ⎛u ⎞
qT = 0.01γq ∑ ib ⎜ ave ⎟ ⎜⎜ − 1⎟⎟ f ⎜ * ⎟
⎝ h ⎠ ⎝ τ cr ⎠ ⎝w⎠
1/ 3
ρu ave
2
⎛ d ave ⎞
τ0 = ⎜ ⎟
58 ⎝ h ⎠

329
Sediment Transport Technology

where, qT is total load in terms of kg/s m, q is unit discharge in m3/s/m, τ cr is critical

shear stress found from Shields diagram (Fig. 9), and, γ is unit weight of water.

⎛ 1000 ⎞ 2
⎜ ⎟ * 4.15 1/ 3
⎝ 9.81 ⎠ ⎛ 0.001 ⎞
τ0 = *⎜ ⎟ = 2 kg/m
2

58 ⎝ 3.5 ⎠
⎛u ⎞ u 0.11
f ⎜ * ⎟ = 11 from Fig. 28 by * = = 0.65
⎝w⎠ w 0.17
7/6
⎛ 0.001 ⎞ ⎛ 2 ⎞
qT = 0.01 * 1000 *15.3 * ∑1 * ⎜ ⎟ *⎜ − 1⎟ *11
⎝ 3.5 ⎠ ⎝ 0.074 ⎠
kg
qT = 3.21 = 27,734 tons/day
sm

Levi Formula
d ave 1
Applicable to < .
h 300

3
⎛d ⎞
0.25
⎛ u ⎞ ⎛ u ave ⎞
qT = 0.002⎜ ave ⎟ u cr d ave ⎜ ave ⎟ ⎜⎜ − 1⎟⎟
⎝ h ⎠ ⎜ gd ⎟ ⎝ u cr ⎠
⎝ ave ⎠
⎛γs −γ ⎞
⎜⎜ ⎟⎟2 gd ave
⎛ 8.8h ⎞
u cr = log10 ⎜⎜ ⎟⎟ ⎝ γ ⎠
⎝ d 95 ⎠ 3.5

where, qT is total load in terms of kg/s m.

Figure 28. Function f(u*/w) for Laursen formula

330
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

d ave 0.001 1
= = 0.00029 < = 0.0033 √
h 3 .5 300
⎛ 2650 − 1000 ⎞
⎜ ⎟ * 2 * 9.81 * 0.001
⎛ 8 .8 * 3 .5 ⎞ ⎝ 1000 ⎠
u cr = log10 ⎜ ⎟ = 0.374 m/s
⎝ 0.004 ⎠ 3 .5
0.25 3
⎛ 0.001 ⎞ ⎛ 4.15 ⎞ ⎛ 4.15 ⎞
qT = 0.002 * ⎜ ⎟ * 0.374 * 0.001 * ⎜⎜ ⎟⎟ * ⎜ − 1⎟
⎝ 3.5 ⎠ ⎝ 9.81 * 0.001 ⎠ ⎝ 0.374 ⎠
kg
qT = 0.072 = 622 tons/day
sm

Summary of the results:

Bed Load Bed Load Total Load Total Load

Equation kg/(s m) ton/day Equation kg/(s m) ton/day
Du Boys 2.64 22,810 Bishop et all. 5.06 43,718
Einstein 1.67 14,429 Bagnold 4.12 35,597
Einstein-Brown 5.17 44,669 Egiazaroff 89.2 770,688
Kalinske 0.61 5,290 Einstein 2.84 24,538
Meyer-Peter and 3.69 31,882 Engelund- 13.42 115,945
Muller Hansen
0.3 2,592 Garde- 1.85 15,984
Schoklitsch
Albertson
Shield 0.02 173 Graf et all. 2.2 19,008
Smart 0.49 4,234 Laursen 3.21 27,734
Levi 0.072 622

Solution 14.
i) Einstein method

1 2 3 4 5 6

Lower d Upper d dave ib Fall velocity, w

(mm) (mm) (mm) (%) (m/s)
1 0.40 0.65 0.51 11 0.079
2 0.65 1.00 0.81 20 0.14
3 1.00 1.50 1.22 25 0.19
4 1.50 3.00 2.12 22 0.31
5 3.00 3.30 3.15 12 0.40
6 3.30 4.00 3.63 8 0.42
Σ 98

Column 4 d ave = d upper ∗ d lower

Column 6 w is taken from Fig. 29 by dave

331
 Sediment Transport Technology

332
Figure 29. Fall velocity of quartz spheres in air and water, (Rouse, 1937)
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Figure 30. Drag coefficient vs. Reynolds number for spheres and disks,
experimental data (After Rouse, 1938)

Einstein’s method (hydraulic calculations):

1 2 3 4 5 6 7 8 9 10
Rb' (m) u*' (m/s) δ (m) d65/δ χ ∆ u (m/s) Ψ'35 u/u*" u*" (m/s)
0.70 0.052 0.00024 6.16 1.022 0.0015 1.14 3.83 13.5 0.084
1.40 0.074 0.00017 8.71 1.005 0.0015 1.73 1.92 21.5 0.081
2.10 0.091 0.00014 10.67 1.000 0.0015 2.21 1.28 30 0.074
2.80 0.105 0.00012 12.32 1.000 0.0015 2.63 0.96 42 0.063
3.50 0.117 0.00011 13.78 1.000 0.0015 3.00 0.77 55 0.055
4.20 0.128 0.00010 15.09 1.000 0.0015 3.35 0.64 69 0.049

11 12 13 14 15 16 17 18 19.i 19.ii 20
Rb" (m) Rb (m) h (m) (h=Rb) A (m2) Q (m3/s) X Y βx β (β/βx)2 P

1.80 2.50 2.50 259.6 294.7 0.0011 0.53 0.831 1.025 1.521 10.84

1.65 3.05 3.05 319.1 552.3 0.0011 0.53 0.831 1.025 1.521 11.02

1.38 3.48 3.48 366.6 810.3 0.0012 0.53 0.831 1.025 1.521 11.15

1.00 3.80 3.80 401.4 1054.8 0.0012 0.53 0.831 1.025 1.521 11.23

0.76 4.26 4.26 453.2 1361.1 0.0012 0.53 0.831 1.025 1.521 11.35

0.60 4.80 4.80 514.6 1723.0 0.0012 0.53 0.831 1.025 1.521 11.47

333
 Einstein’s method (bed load calculations): Sediment Transport Technology

334
Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Einstein’s method (bed load calculations) (continued):

335
Sediment Transport Technology

Rating curve for Einstein Method

6

4
h (m)

0
0 500 1000 1500 2000
3
Q (m /s)

Einstein Method
1.000.000
Q (m3/s)
100.000
294,7
iTQT /ib (tons/day)

10.000 552,3
810,3
1.000 1054,8
1361,1
100 1723,0

10
0,10 1,00 10,00
d ave (mm)

Einstein Method
10.000

d > (mm)

0,40
0,65
Q (m /s)
3

1.000 1,00
1,50
3,00
3,30

100
10 100 1.000 10.000 100.000 1.000.000
Σ i T Q T (tons/day)

336
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Column 1, 5, 6 and 7 are taken from Einstein’s method

ii) Laursen’s method:

337
 Laursen’s method (continued): Sediment Transport Technology

338
Column 1, 5, 6 and 7 are taken from Einstein’s method
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Laursen's method
1.000.000

100.000 3
Q (m /s)

294,7
iTQT /ib (tons/day)

10.000
552,3
810,3
1.000 1054,8
1361,1
1723,0
100

10
0,10 1,00 10,00
d ave (mm)

Laursen's method
10.000

d > (mm)

0,40
0,65
Q (m /s)
3

1.000 1,00
1,50
3,00
3,30

100
10 100 1.000 10.000 100.000 1.000.000

Σ i T Q T (tons/day)

339
 iii) Graf et al.’s method: Sediment Transport Technology

340
Column 3 and 8 are taken from Einstein’s method
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Column 3 and 8 are taken from Einstein’s method

Graf et al.’s method (continued):

341
Sediment Transport Technology

Graf et al. method

1.000.000

100.000
Q (m3/s)

294,7
iTQT/ib (tons/day)

10.000
552,3
810,3
1.000 1054,8
1361,1
1723,0
100

10
0,10 1,00 10,00
d ave (mm)

Graf et al. method

10.000

d > (mm)

0,40
Q (m /s)

0,65
3

1.000
1,00
1,50
3,00
3,30

100
10 100 1.000 10.000 100.000 1.000.000
Σ i T Q T (tons/day)

342
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

Einstein Method
10.000
d > (mm)

0,40
0,65
Q (m /s)
3

1.000 1,00
1,50
3,00
3,30

100
10 100 1.000 10.000 100.000 1.000.000
Σ i T Q T (tons/day)

Laursen's method
10.000
d > (mm)

0,40
0,65
Q (m /s)
3

1.000 1,00
1,50
3,00
3,30

100
10 100 1.000 10.000 100.000 1.000.000

Σ i T Q T (tons/day)

Graf et al. method

10.000
d > (mm)

0,40
0,65
Q (m /s)
3

1.000 1,00
1,50
3,00
3,30

100
10 100 1.000 10.000 100.000 1.000.000
Σ i T Q T (tons/day)

343
Sediment Transport Technology

Solution 15.
Chezy equation:
u = C RS ⇒ u2 = C2 RS

For uniform flow:

S = hL/L

Velocity head:
u 2 C 2 RS u 2 C 2 R hL 2g L u 2
= ⇒ = ⇒ hL = 2
2g 2g 2g 2g L C R 2g

This equation is Darcy formula and is applicable to both pipes and channels.

πD 2
A= , P = πD ,
4
πD 2
D
R= 4 =
πD 4

By inserting this value to the previous equation:

2 g 4L u 2 8g L u 2
hL = 2 ⇒ hL = 2
C D 2g C D 2g

8g L u2
If we denote f = 2 ⇒ hL = f
C D 2g
R1 / 6 8g
Remember that C = =
n f

Solution 16.
1.5 m

0.40 m
R

0.75 m

D = 1.5 m
1 πD 2
A= + By = 0.5*(π*1.52/4)+1.5*0.4 = 0.884+0.6 = 1.484 m2
2 4

344
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

1
P= πD + 2 y = 0.5*π*1.5+2*0.4 = 2.356+0.8 = 3.156 m
2
R = A/P = 1.484/3.156 = 0.47 m

Q = Au = AC RS ⇒ 1 = 1.484*C*(0.47*0.0004)0.5 ⇒ C ≅ 49

R1 / 6
From C = ⇒ n = R1/6/C = 0.471/6/49 ≅ 0.018
n

From Manning equation:

A 1.484
Q = R2/3 S = (0.47 )2 / 3 0.004 = 0.997 m 3 /s ≅ 1.0 m 3 /s
n 0.018

Solution 17.
From Manning equation:
1
u = R2/3 S
n

For a wide river: R≅y

For rectangular cross section:

u c = gy c

Inserting these values to the Manning equation:

1 1
gy c = y c2 / 3 S 1 / 2 ⇒ 9.81y c = y c2 / 3 (0.0015)1 / 2
n 0.010
By trial and error procedure yc = 0.28 m

q = y c vc = y c gy c = 0.28 * 9.81 * 0.28 = 0.464 m 3 /s/m

Solution 18.

B=5m

A = By = 5y
P = B+2y = 5+2y
A 5y
R= =
P 5 + 2y

345
Sediment Transport Technology

From Manning equation:

2/3
A 5y ⎛ 5y ⎞
Q = R2/3 S ⇒ 12 = ⎜ ⎟ 0.0002
n 0.011 ⎜⎝ 5 + 2 y ⎟⎠
2/3
⎛ 5y ⎞
1.867 = y⎜⎜ ⎟⎟
⎝ 5 + 2y ⎠
By trial and error procedure y ≅ 1.81 m

Solution 19.

D=0.5 m

1 πD 2 1 π * 0.5 2
A= = = 0.0982 m 2
2 4 2 4
πD π * 0.5
P= = = 0.785 m
2 2
1 πD 2
D 0.5 0.0982
R= 2 4 = = = = 0.125 m
πD 4 4 0.785
2
From Manning equation:
A 0.0982
Q = R2/3 S = (0.125)2 / 3 0.002 = 0.042 m 3 /s
n 0.026

Solution 20.
u = C RS
A By
Q = Au = A C RS = B y C RS = B y C S = B yC S
P B + 2y

2 .4 y 1 0.006 y
8.5 = 2.4 y51 ⇒ 0.0694 = y ⇒
2.4 + 2 y 400 2 .4 + 2 y
⎛ 0.006 y ⎞
0.00482 = y 2 ⎜⎜ ⎟⎟
⎝ 2 .4 + 2 y ⎠

0.011574 + 0.009645y = 0.006y3

346
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

y3 - 1.61y - 1.93 = 0 ⇒ y = 1.66 m

Solution 21.
Mean velocity:
1
u = R 2 / 3 S 1/ 2
n

Velocity at a point:
gy 0 S ⎡ ⎛ y ⎞⎤
U =u+ ⎢1 + 2.3 log⎜⎜ ⎟⎟⎥
κ ⎣ ⎝ y0 ⎠⎦

κ is Van Karman constant and is 0.4 for clear water.

For wide river R ≅ y = 22.5 cm = 0.225 m

1 2 / 3 1/ 2 1
u= R S ⇒ u= 0.225 2 / 3 S 1 / 2 ⇒ S = 0.00292u2
n 0.02

Ua
a) = 0.9 m ⇒ Ua = 2 g ∗ 0.9 = 4.2 m/s
2g

gy 0 S ⎡ ⎛ y ⎞⎤
U =u+ ⎢1 + 2.3 log⎜⎜ ⎟⎟⎥
κ ⎣ ⎝ y0 ⎠⎦
9.81 * 0.225 * 0.00292u 2 ⎡ ⎛ 0.15 ⎞⎤
4 .2 = u + ⎢1 + 2.3 log⎜ ⎟⎥
0 .4 ⎣ ⎝ 0.225 ⎠⎦
4.2 = u+0.2u·0.595 ⇒ 4.2 = u+0.119u ⇒ u = 3.75 m/s and S = 0.041

9.81 * 0.225 * 0.00292 * 3.75 2 ⎡ ⎛ 0.05 ⎞⎤

U b = 3.75 + ⎢1 + 2.3 log⎜ 0.225 ⎟⎥
0 .4 ⎣ ⎝ ⎠⎦
Ub = 3,75+0,753*(-0,502) = 3.37 m/s

U b2 3.37 2
b= = = 0.58 m
2 g 19.62

b) U = u
⎛ y ⎞
0 = 1 + 2.3 log⎜⎜ ⎟⎟
⎝ y0 ⎠
⎛ y ⎞
− 0.435 = log⎜⎜ ⎟⎟
⎝ y0 ⎠
y y
10 −0.435 = ⇒ = 0.367
y0 y0

347
Sediment Transport Technology

c)

y/y0 0.00 0.01 0.05 0.10 0.15 0.20 0.25 0.30

Ub - 1.04 2.25 2.77 3.08 3.29 3.46 3.60

y/y0 0.35 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Ub 3.71 3.81 3.98 4.12 4.23 4.33 4.42 4.50

4,50
4,42
1,0

4,33
0,9

4,23
0,8

4,12
0,7

3,98
0,6

3,81
y/y0

0,5

3,71
3,60
0,4 3,46
3,29

0,3
3,08
2,77

0,2
2,25
1,04

0,1
0,0
0 1 2 3 4 5
Velocity (m/s)

Solution 22.
S = tan (0.1) = 0.00175

A 2 / 3 1/ 2
Q= R S
n

a)
n3 = 0.030

Q2 Q3 2m
Q1
n2 = 0.025
5m
n1 = 0.020

25 m 6m 20 m

348
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

A1 = 6*7 = 42 m2 P1 = 16 m R1 = A1/P1 = 42/16 = 2.625 m

A2 = 25*2 = 50 m2 P2 = 27 m R2 = A2/P2 = 50/27 = 1.86 m
A3 = 20*2 = 40 m2 P3 = 16 m R3 = A3/P3 = 40/22 = 1.82 m
ΣA = 132 m2 ΣP = 65 m

42 42 2 / 3
Q1 = ( ) (tan 0.1)1 / 2 = 166.9 m 3 /s
0.020 16
50 50 2 / 3
Q2 = ( ) (tan 0.1)1 / 2 = 126.0 m 3 /s
0.025 27
40 40 2 / 3
Q3 = ( ) (tan 0.1)1 / 2 = 83.0 m 3 /s
0.030 22

ΣQ = 375.9 m3/s

b)
n3 = 0.030

Q2 Q3 2m
Q1
n2 = 0.025
5m
n1 = 0.020

25 m 6m 20 m

Inclination is 45°.

Ai = 0.5*2*2 = 4 m2
A1 = 42-4 = 38 m2 P1 = 16 m R1 = A1/P1 = 38/16 = 2.375 m
2
A2 = 50+2 = 52 m P2 = 27 m R2 = A2/P2 = 52/27 = 1.93 m
2
A3 = 40+2 = 42 m P3 = 16 m R3 = A3/P3 = 42/22 = 1.91 m
ΣA = 132 m2 ΣP = 65 m

38 38 2 / 3
Q1 = ( ) (tan 0.1)1 / 2 = 141.3 m 3 /s
0.020 16
52 52 2 / 3
Q2 = ( ) (tan 0.1)1 / 2 = 134.5 m 3 /s
0.025 27
42 42 2 / 3
Q3 = ( ) (tan 0.1)1 / 2 = 90.0 m 3 /s
0.030 22

ΣQ = 365.8 m3/s

349
Sediment Transport Technology

c) Horton and Einstein method:

2/3
⎡ n 3/ 2 ⎤
⎢ ∑ Pi ni ⎥ ⎡16 * 0.020 3 / 2 + 27 * 0.0250 3 / 2 + 22 * 0.030 3 / 2 ⎤
2/3

n = ⎢ i =1 ⎥ =⎢ ⎥ = 0.0256
⎢ P ⎥ ⎣ 65 ⎦
⎢⎣ ⎥⎦

132 132 2 / 3
Q= ( ) (tan 0.1)1 / 2 = 345.4 m 3 /s
0.0256 65

It assumes mean velocity in the individual subsections, as well as the mean velocities in
the whole section are same.

d) Pavlovski, Mühlhafer, Einstein and Banks method:

1/ 2
⎡ n 2 ⎤
⎢ ∑ ( Pi ni ) ⎥ ⎡16 * 0.020 2 + 27 * 0.0250 2 + 22 * 0.030 2 ⎤
1/ 2

n = ⎢ i =1 ⎥ =⎢ ⎥ = 0.0257
⎢ P ⎥ ⎣ 65 ⎦
⎢⎣ ⎥⎦

132 132 2 / 3
Q= ( ) (tan 0.1)1 / 2 = 344.1 m 3 /s
0.0257 65

It assumes that the total force resisting the flow is equal to the sum of the forces
resisting the flow.

e) Lotter method:
132 5 / 3
5/3 65 * ( )
PR 65
neq = = = 0.0235
n
Pi Ri5 / 3 42 50 40
∑ ni
16 * ( ) 5 / 3
16
27 * ( ) 5 / 3 22 * ( ) 5 / 3
27 22
i =1 + +
0.020 0.025 0.030

132 132 2 / 3
Q= ( ) (tan 0.1)1 / 2 = 376.3 m 3 /s
0.0235 65

It assumes that the total discharge of the flow is equal to the sum of the discharges of
the subsections.

References

1. ASCE (American Soc. of Civil Engineers), ”Sedimentation Engineering”, Manuals

and Reports on Engineering Practice, No: 54, ASCE, USA, (1975).
2. Bayazit, M., “Hydraulics of Flowing Beds”, Istanbul Technical University
Publications, No: 835, Turkey (in Turkish), (1971).
3. Chow, V.T., “Open Channel Hydraulics”, McGraw-Hill, Singapore, (1959).

350
 Exercises on Uniform Flow in Open Channel, Bed Load, Suspended Load and Total Load

4. Garde, R. J., and Ranga Raju, K. G., “Mechanics of Sediment Transportation and
Alluvial Stream Problems”, Wiley Eastern Limited, India, (1978).
5. Graf, W. H., “Hydraulics of Sediment Transport”, McGraw-Hill, (1971).
6. Simons, D. B., and Senturk, F., “Sediment Transport Technology, Water and
Sediment Dynamics”, Water Resources Publications, USA, (1992).
7. Yanmaz, A. M., “Applied Water Resources Engineering”, Middle East Technical
University, Ankara, Turkey, (1997).

351
352
 RESERVOIR SEDIMENTATION

Prof.Dr. Mustafa Göğüş

Civil Engineering Department
Middle East Technical University
Ankara - Turkey

1. Introduction
When rivers encounter with a reservoir behind a dam, a natural lake or sea, which are
much deeper than themselves, the velocity of the flow entering the pool will be reduced,
and the major part, or all, of the sediment transported into the reach will be deposited in
the reach of backwater influence and in the reservoir. Experience has shown that there
are two characteristic types of sediment deposits in reservoirs along alluvial rivers:
1. Those occurring generally over the reservoir bottom, mostly composed of the finer
fractions of the river sediment load-the silts and clays,
2. Those occurring in a characteristics delta formation at the head of the reservoir,
including all the coarser fractions of the river sediment load-the sands and gravels-
but which may also include large quantities of silts and clays. This sedimentation
process is described as aggradation.

Upstream aggradation of the river channel could under some circumstances cause the
reservoir backwater effect to progress upstream, increasing flood heights. The growth of
the delta into the reservoir lessens the reservoir capacity, affecting its economic life.

It is very obvious that having reservoirs, which are designed for several purposes, such
as; irrigation, drinking water supply, electricity production, flood control, etc., filled up
with sediment in a much shorter time than that for which they are planned and designed
to be in service, is not a desirable situation. Because of such a reason the damage to be
given to the economy of the country is clearly seen. Therefore, not only amount of the
sediment to be deposited in reservoirs, but also the time period required for that and the
new bottom topography of reservoirs as a function of time should be known.

Since sediment transport is a complicated phenomenon, which involves many

parameters, analysis of sedimentation problems analytically is very difficult. To get
solutions to be used in economical reservoir designs, various simplifications are done in
the theoretical equations and then they are tried to be solved. After that, relating then
with the data obtained from field measurements, general solutions for special situations
are searched for.

In this chapter the aim is, in general, to give overall information about methods used in
the analysis and control of reservoir sedimentation.

2. Analysis of Sedimentation in Reservoirs

In the analysis of sedimentation in reservoirs, in addition to mathematical and physical
laboratory models, measurements and surveys conducted at the reservoirs, where

353
Sediment Transport Technology

sedimentation studies are to be done, are also very widely used. The common part of all
these methods is not to give a general solution, which will be applicable to any kind of
reservoir due to the complicated structure of the problem. However, for some special
cases satisfactory results can be obtained.

In the following the mathematical models used in the analysis of sedimentation in

reservoirs are described.

2.1. Mathematical models

Mathematical models developed for reservoir sedimentation can be analyzed into two
basic groups. In the first group, the common point in all of them is to try to solve the
equations of conservation of mass and motion together for flow and sediment for a
given river-reservoir system. The expectation from the common solution of
aforementioned differential equations is to obtain relations for the variation of
sedimentation in reservoirs with time and location, whereas in the second group of
models, a simpler method is applied. Firstly, the backwater curve which forms behind
the dam is determined independently from the sediment transportation. Then, assuming
that the flow conditions in the river-reservoir system is not changing for a certain time
period, sediment transportation and deposition of the sediment are computed for each
length of successive sections selected along a reach of the system. Amount of the
sediment deposited in each section is determined from the difference of sediment load
passing through the beginning and end sections of the section. If the deposited amount
of the sediment, which is calculated using one of the sediment transportation equations,
is much enough to influence the backwater curve, in this case the computation of the
backwater curve is repeated with respect to the new bottom elevation. Successive
application of these calculations to the river-reservoir system yields the sedimentation in
the reservoir.

The second groups of mathematical models, although they look very much simplified,
give results in good agreement with natural river-reservoir systems in respect to
sedimentation, and form of the sedimentation. The two types of models described above
are given in the following with details.

2.1.1. Mathematical models in which flow and sediment equations are solved together
In this type of models, one-dimensional equations of continuity and motion for flow and
sediment are solved together. Applying several simplifications, the equations to be
solved are put into the form of heat equation, and then since a general analytical
solution is not yet available, a proper, a proper numerical solution is applied. Among
those numerical methods, method of characteristics and finite-difference technique can
be counted as the ones very widely used. No matter which numerical method is used,
the problems of stability and convergence may arise. In addition, numerical solutions
always have some kind of periodic properties. In particular, the number of terms to be
taken into consideration from the series used in the solution and also the time period to
be selected must be carefully adjusted. Otherwise, lots of difficulties in the application
of the method come up. If short time intervals and many terms from the series are used,
the time required to get the solution is automatically increased. Use of fewer amounts of
terms and longer time periods, most of the time, eliminates the possibility of
approaching the solution. Therefore, it seems that searching for the formation of deltas
with details using this method looks not impossible but very difficult.

354
 Reservoir Sedimentation

Mathematical models developed by Tsuchiya and Ishizaki (1969), Adachi and Nakatoh
(1969), Garde and Swamee (1973), Sanoyan (1971), and Chang (1969) can be counted
as the most important ones among others which can be classified into this group of
mathematical modeling methods. In all these studies the five equations stated below
have been used.

a) Continuity equation for the flow: as an example the equation of continuity for
steady flow which can be written per unit width of the channel section as

q = Uy (1)

where q is the unit discharge, y is the flow depth and U is the average flow
velocity for the cross section.

b) Equation of motion for the flow: as an example, the equation of water surface
profile for gradually varying one-dimensional flows,

dy S0 − Se
= (2)
dx 1 − U 2 / 2 g

where y is the water depth, x is the distance in the flow direction, S0 is the slope
of the channel bottom, Se is the slope of the energy line, U is the average flow
velocity for the cross section, and g is the acceleration of gravity.

c) Channel friction equation for one-dimensional flows: as an example, Manning

equation,

1 2 3 12
U= y S0 (3)
n

where n is the Manning coefficient for the channel boundary.

d) Continuity equation for sediment,

∂z 1 ∂q s
+ =0 (4)
∂t (1 − P) ∂x

where z is the bottom elevation of the channel, t is the time, P is the porosity of
the material deposited at the bed, and, qs is the sediment discharge per unit width
of the channel.

e) Equation of sediment discharge,

q s = f1 [To ' (To ) c r ] (5)

where To is the bed shear stress and (To)cr is the critical bed shear stress.

355
Sediment Transport Technology

Because of having very complex properties for flow, sediment and topography of the
river-reservoir systems in nature, unless making several assumptions to simplify the
equations given above, they can not be solved. The following cases can be stated as
some of the basic simplifications which are referred in the solution of equations: flow
discharge, bottom roughness, the size of the bed material, the porosity of the deposited
material and water surface elevation along the dam axis are constant, the flow is one-
dimensional and, therefore, all velocity components in the other directions except the
flow direction are negligible and uniform velocity distribution prevails in each cross
section.

Another important point, which should be kept in mind, is that the selected sediment
discharge equation may not always be used for any type of problem. Since those
equations are mainly valid for regions which are similar to those for which they have
been derived.

Consequently, the form of the differential equation to be obtained after substituting all
the simplifications into those five basic equations will be same as the equation of heat.

∂z ∂2z
=K 2 (6)
∂t ∂x

where K is a coefficient which is obtained from data of sediment surveys. And the other
variables are as defined before. Tsuchiya and Ishızaki (1969), and Adachi and Nakatoh
(1969), in their studies in Japan using the results obtained from surveys of Hongu and
Tenryy dams. Gave different values to K and then solved Eq. 6 graphically. The
agreement between the results of the solution and the ones received from natural
surveys was quite good (Fig. 1).

Garde and Swamee (1973) solved the Eq. 6 using explicit finite difference technique.
Because of the general property of this method, the bed profiles obtained by Garde and
Swamee (1973) had ondulations. In order to eliminate this undesirable situation, Fourier
series were used and the modified profile given in Fig. 2 was achieved.

Chang (1969) used method of characteristics not for a river-reservoir system, but for a
river-sea system also considering waves and tide. It was stated that the results of the
study relatively good.

2.1.2. Mathematical models in which backwater profiles and sediment equations are
solved separately
This type of mathematical models can be considered simpler than the one before in
respect to the solution way of the equations. The application of the model is completed
in two steps. In first step, the sediment transport is ignored and M1-type backwater
profile is calculated with the use of one of the known methods of gradually varied open
channel flow. During these calculations it is assumed that the form of the reservoir
bottom, its roughness, discharge of the river and the maximum water level in the
reservoir do not change. In the second step, assuming the hydraulic conditions for which
all the backwater calculations were done in the first step did not change in the direction
of flow in each reach to be selected, the amount of sediment transportation and the

356
 Reservoir Sedimentation

(a)

(b)
Figure 1. (a) Observed and calculated river bed (after Tsuchiya and Ishizaki, 1969),
(b) River bed slope at dam crest (after Tsuchiya and Ishizaki, 1969)

357
Sediment Transport Technology

Figure 2. Pattern of sedimentation in a reservoir obtained by Garde and Swamee (1973) using a
mathematical model.

portion of it which is deposited are computed. By repeating the calculations done in

both steps successively, the formation of sedimentation in the reservoir and its progress
with time can be obtained.

Harrison (1952-1958), and Yücel and Graf (1973) very extensively applied this method
in their studies. Harrison did not get satisfactory results from the application of the
model to Big Bend Reservoir in America. The reason for that was concluded as having
a great number of undefined quantities to be used in the calculations.

Yücel and Graf (1973) developed this model in the form of a computer model and
applied it to a reservoir of well-defined properties, and then achieved very interesting
results about the growth and advancement of sedimentation in the reservoir. In a typical
river-reservoir system, as seen in Fig. 3, if only transportation of bed material is
considered, the coarser sediment particles will be deposited at the upstream part of the
reservoir and form deltas. The finer fraction of the sediment, carried by the river travel
to a region which is much closer to the dam and then is deposited, forms bottom
sediments.

Figure 3. Sketch of definition

358
 Reservoir Sedimentation

The basic flow chart of the computer model in which backwater curve and sediment
transportation calculations are separately done is given in Fig. 4. The short time
duration used in computations increased the validity of solving basic equations
separately. When the calculations for deposition within each reach of the river-reservoir
system are completed, the bottom configuration is readjusted by the addition of the
thickness of the deposition at every reach point to the existing bed elevation. If the
amount of deposition has significantly altered the bed configuration, a new backwater
profile is then calculated. Otherwise, another series of calculations are carried out for
deposition. In their study, a maximum cumulative deposition thickness exceeds 2% of
the local water depths was considered to be significant and to warrant the calculation of
a new backwater profile.

input
information
N = no. of sediment days

index = 0 index = parameter controlling

requirement of a new
backwater profile
DO (1,N)

index=0 yes Calculate new

? backwater profile

no
Print and plot
Calculate bed load
the new profile
deposition

Print and plot bed

profile after deposition

Significant yes
deposition? index = 0

no

index ≠ 0

Figure 4. Flow chart of the computer program (after Yücel, 1973)

359
Sediment Transport Technology

For the determination of backwater profile, the calculations were started at dam section
where the water depth is maximum and progressed in the upstream direction towards
the river. At the end of these calculations all of the average flow conditions for different
cross sections were determined, but during the calculations it was assumed that there
was no sediment transportation in the reservoir (Fig. 5).

When transportation and deposition of bed load being calculated, the properties of flow
were assumed not to be changed, and utilizing a bed load transportation equation
amount of sediment passing through different sections was determined. As seen in Fig.
6 since in the direction of flow the average flow velocity decreased, the sediment load
entering into a section would be different than that leaving the section and therefore,
that difference would be deposited along the length of the reach. The assumption of
having the deposited material uniformly distributed along the length of the reach could
be verified by selecting short time intervals and reach lengths in the computations.
Consequently, new bottom elevation is determined adding the thickness of the deposited
material to the initial bottom elevation. If this new bottom form requires a new
backwater profile calculation, in other words if the thickness of the deposition is
significant, the backwater curve is re-determined for the new bottom form. Successive
application of the procedure yields the formation of deposition in the reservoir.

Most of the sediment load equations which are used in the determination of sediment

Figure 5. Sketch for the backwater profile calculations

Figure 6. Sketch for the bed load deposition calculations

360
 Reservoir Sedimentation

deposition, such as; Schoklitsch’s, Mayer-Peter Müller’s or Einstein’s bed load

equations, are valid for uniform flow conditions and have been developed not for
deposition but for erosion situations. Whereas the type of flows in reservoirs are mainly
gradually varied and bottom and energy slopes are always different from each other.
Therefore, the decision of selecting the effective slope; bottom slope, energy slope or
another slope, is very important. It has been proposed so far that the use of the energy
slope is more logical. On the other hand, those sediment load equations developed for
erosion cases should also be converted into the ones to be used for deposition situations.
For example, before using an equation which contains critical erosion velocity, this
term should be replaced with critical deposition velocity. Although in open channel
flows sufficient study has not yet been conducted on the topic of critical deposition
velocity, Fig. 7 given by Hjulström (see Graf, 1971) can be used to evaluate the critical
deposition water flow rate qcr = y Ucr.

Yücel and Graf (1973) calculated the sediment deposition in the reservoir with the use
of a Schoklitsch type bed load equation (see Graf, 1971),

qs = -x sk (q – y·ucr) (7)

where qs is the bed load transport rate in volume per unit time and unit width, s is the
energy slope of the flow, q is the water flow rate in volume per unit time and unit width,
y is the flow depth, x and k are empirical sediment coefficients and ucr is the critical
deposition velocity to be obtained from Hjulström’s figure.

The second bed load equation which is used in the model is the one proposed by Mayer-
Peter Müller.

Figure 7. Erosion-deposition criteria for uniform particles (after Graf, 1971)

361
Sediment Transport Technology

S s ⎧ 4d ⎡ γys ⎤
qs = ⎨ 1/ 3 ⎢ − 0.047(γ s − γ )⎥ (8)
( S s − 1) ⎩ ρ ⎣ d ⎦

where γ are the specific weights of sediment material and water respectively, d is the
mean diameter of bed sediment, ρ is the density of water, y is the flow depth and s is the
energy slope of the flow. Although the second term inside the parenthesis is related to
the erosion case, so far no attempt has been given on the replacement of it with a
deposition term.

Finally, Einstein’s (1942) equation is utilized in the model as the third one.

( S s − 1) gd ⎡ 0.391( S s − 1)d ⎤
qs = exp ⎢− ⎥ (9)
0.465 ⎣ ys ⎦

where g is the acceleration of gravity, and the other terms are same as defined before.
Since the Eq. 9 developed on the stability of erosion and deposition cases and not
including any critical variable, any change related to the deposition situation was not
made.

The most important result achieved from the application of this model by Yücel and
Graf (1973) is the one related to the formation of the delta. The process of formation of
the delta follows the same trend when different hydraulic conditions and bed load
equations are applied to the model. As seen in Figs. 8 and 9 fine layers are forming and
then they are taking the shape of a triangle and progressing the direction of flow toward
the dam section. Here the interesting point is that even though the mathematical model
used is not so complicated, the formation and progress of the delta look like very much
those encountered in.

Figure 8. Patterns of sedimentation obtained from computer modeling for

three different sediment load equations. (After Yücel, 1975)

362
 Reservoir Sedimentation

Figure 9. Comparison of the patterns of sedimentation obtained from computer

modeling for the nature.

As an example, the three different sediment load equations, (after Yücel, 1975)
similarity between the delta obtained by means of computer modeling and the one
observed in the reservoir of Hoover Dam (Lake Mead) in America is very clearly seen
in Figs. 10 and 11.

A similar mathematical model for different flow discharges was used by Asada (1973)
in Japan for reservoirs situated in mountainous regions.

Figure 10. The growth and advancement of a delta obtained by computer modeling
(after Yücel, 1975)

363
Sediment Transport Technology

Figure 11. Delta formation in the reservoir of Hoover Dam (after Yücel, 1975)

2.2. Laboratory models

First experimental study on reservoir sedimentation was conducted by Schoklitsch
(1937), and the following conclusions were stated: “deposition of sediment occurs
where the slope of the water surface decreasing due to the effect of backwater and the
general forms of this deposited sediment looks like the deltas form where rivers
encounter with seas, and also delta fronts progress in the flow direction”.

In the near past, Bhamidipaty and Shen (1971) and Sugio (1972) analyzed the
sedimentation problems in reservoirs at the laboratory. They fed a channel of sand
bottom with sand more than the sediment capacity of the channel and formed a delta
which was observed that moving in the direction of flow. The result of the former study
stated above is given in Fig. 12.

Figure 12. Typical bed and water surface profiles during aggradation
(after Bhamidipaty and Shen, 1971)

364
 Reservoir Sedimentation

Sugio (1972) used two difterent bottom slopes (0.01 and 0.00375) and different water
and sediment discharges. Figs. 13 and 14 show the advancement of the deltas formed on
mild and steep flume slopes, Sugio (1972), in addition, by means of dimensional
analysis obtained the following equations.

h ⎡ yc ⎤ 2/7
= 0.56 ⎢ ⎥ (10)
yc ⎣ ( S s − 1)d ⎦
qs D q
= 0.0025( so ) −0.7 (11)
q so q

in which h is the water depth on the front, yc is the critical flow depth, Ss is the specific
gravity of sediment grain, d is the grain size of sediment, qsD is the sediment transport
rate at the front per unit width of the channel (sediment discharge carried by the delta),
qso is the supplied net sediment discharge per unit width of the channel from the sand
feeder, and q is the water discharge per unit width of the channel. Sugio (1972)
observed that, for each experiment, the thickness of the delta during the advancement

Figure 13. Typical bed and water surface profiles during aggradation
for mild flume slope (after Sugio, 1972)

Figure 14. Aggradation above dam for steep flume slope (after Sugio, 1972)

365
Sediment Transport Technology

period was approximately constant and this thickness was always greater than the
critical flow depth. Another equation which was derived by the same investigator for
steep channels gives a relation for the slope of the upstream face of the delta, sD.

q so 0.35
s D = 0.0774( ) (12)
q

Kobanbay (1969), who was another investigator worked on the same topic at the
laboratory, proposed the following relations for the progress speed of the delta, UD, and
sediment discharge carried by the delta, qsD;

2510q s
UD = (13)
HL1.11
2.04q s (1 − P)
q sD = (14)
L0.11

in which qs is the unit sediment discharge entering the channel section, P is the porosity
of the bed material, H is the water depth at the section of dam and L is the length of the
delta. The above equations show that the sediment discharge carried by the delta is
decreasing in the flow direction.

Yücel (1975) in his experimental study analyzed the properties of sediment deposition
for one-dimensional flow case. The main goal of this work was to compare the results
obtained from a computer model developed by Yücel with experimental findings. For
this reason, the patterns of growth and advancement of deltas were investigated in a
laboratory channel of 1.0 m, width, 0.5 m, and, depth 20.0 m. Length and a bottom
slope of 0.005, by making use of four different discharges (40, 60, 80, 100 lt/sec), two
different sizes of natural sand (d = 0.98 mm, 1.85 mm) and different maximum water
depths at the maw section (H = 0.25-0.35 m). Sand given from a sand feeder located at
the entrance region of the channel, first formed thin layers which eventually grew into a
typical delta and consequently this delta started advancing in the flow direction without
much of change in its thickness at the delta front. During the later stages of the growth
and advancement of the delta, it was observed that the angle between the front face of
the delta and the channel bottom was almost constant and about 40o, and then sand
particles were being swept along the upstream face of the delta towards downstream
sliding over the front face and deposited. Therefore, the advancement of the delta
continued in the flow direction. In the mean time, when delta forming, the water surface
profile also changed. It was concluded that the agreement between the delta formation-
advancement and result of computer modeling was very good. The rate of advancement
of the delta seemed to remain approximately constant in all experiments maintaining an
astonishingly smooth and straight delta front in transverse direction to the flow.

Some important results of experimental studies which have been conducted in reservoir
sedimentation have been summarized above. Because of having difficulties in
establishing similarity conditions of natural flow and sedimentation cases to use at the
laboratory, all experimental work carried out can not give quantitative results which can
be applied to any kind of reservoir sedimentation problem. On the other hand,

366
 Reservoir Sedimentation

qualitative results achieved from those studies are very important, and especially the
models in which backwater profile calculations involved give some hope about their
successful applicability to related problems. No doubt that the reliability degree of the
laboratory results will increase when sediment transportation and deposition equations
are derived, as well as the establishment of similarity criterions, for gradually varied
flows.

3. Reservoir Sedimentation Measurements and Surveys

In addition to mathematical and experimental studies, measurements and surveys of
reservoir sedimentation have great importance in getting qualitative and quantitative
information about the properties of sedimentation in reservoirs. The basic reasons why
these measurements and surveys are done can be classified as fallows:
a) investigation of watersheds for their sediment production ability
b) investigation of damages to be caused by deposition or erosion in watersheds
c) development of the methods to protect watersheds from deposition or erosion
d) investigation of the pattern of sedimentation and its amount
e) determination of the service-life of reservoirs
f) giving proper decisions for the optimum operation of reservoirs, which strongly
depends upon the amount of sedimentation, according to the conditions for
which they are planned and designed.

Measurements and surveys to be conducted in a reservoir will not only give important
information which is to enable the reservoir to be operated under optimum conditions,
but also that knowledge can be used in the planning and designing of other similar
reservoirs. Therefore, measurements and surveys should be done systematically and
results to be obtained should be analyzed using scientific methods.

The frequency of surveys depends upon the estimated rate of sediment accumulation in
the reservoir. It follows then that with high depositional rates are surveyed more often
than those with low rates. Financing the cost of the survey is often a controlling factor
in how frequently surveys are made. Considering that the cost of the survey is justified
by the need to continue updating the reservoir capacity, a suggested guide to the
frequency is a 5-yr to 10-yr interval depending upon the run off inflow volume and
magnitude of flows. It will be very useful that following each flood period surveys
should be done regardless of the normal survey frequency. If a new dam is to be built at
the upstream part of the reservoir, right after the construction of that dam measurements
and surveys should be carried out in the reservoir due to the rate of sediment deposition
to be influenced by the presence of the new dam.

Equipment to be used during sedimentation measurements and surveys in reservoirs can

be summarized as follows:
a) Boat or raft and auxiliary equipment: To run a survey of a partially or
completely filled reservoir, a boat or a raft is often necessary. The size of either
vessel depends upon how large the reservoir is. For small reservoirs a shallow
draft boat 14 ft long or longer would normally be adequate. Large reservoirs,
correspondingly, would require larger boats and very often two boats are
desirable, one to carry sounding equipment and the other to transport survey
personnel. Rafts are generally more suitable than single boats for sampling
operations because of their stability.

367
Sediment Transport Technology

b) Sounding equipment: Sonic sounders are currently (1973) used to run most
sediment surveys of both small and large reservoirs. Manually operated
sounding lines and sounding poles used in the past to survey small reservoirs are
virtually outdated and replaced by modern sonic sounders. A sonic sounder is a
portable recording instrument designed to measure water depths by projecting a
high energy acoustical signal downward from just below the water surface and
receiving the signal reflected off the reservoir bottom. Measuring the depth of
water depends on the speed of sound through the water.
c) Distance measurement: The location of the boat as it traverses the sounding path
can be determined by standard survey methods or mechanical or electronical
distance measuring instruments. All methods require fixed reference points
along the reservoir shoreline. Mechanical methods generally involve
measurement of a wire length or the towing of a current meter. The highly
sophisticated electronic methods rely on either light or radio wave propagation.
d) Communications: The ability to communicate rapidly between members of a
reservoir survey party is very important to a successful operation. Each shore
and boat party requires a separate radio.
e) Sediment sampling equipment: Sediment sampling equipment is required to
obtain, whenever possible, undisturbed samples of reservoir deposits for
determining the specific weight and grain-size distribution. Special equipment
and samplers required obtaining undisturbed samples from deposits that are
underwater, such as the gravity and piston core types.

When the sediment deposits are above water surface, they are frequently composed of
coarse-grained materials in the sand range that cover sediment sizes 0.062 mm-2 mm by
the American Geophysical Union Classification. These are usually sampled with a
coring-type sampler.

4. Methods of Computation for Sediment Accumulation

Methods which are used in practice for computation of sediment accumulation are: trap
efficiency, area-increment, empirical area-reduction, and elevation of sediment
accumulation. Although all these methods, to some extent look like each other, for
reservoirs of different characteristics they do not give similar results. Therefore, for a
given reservoir the method to be referred should be selected carefully.

4.1. Trap efficiency

The trap efficiency of a reservoir is defined as the ratio of the quantity of deposited
sediment to the total sediment inflow, and is dependent primarily upon a number of
factors. Among these are the ratio between the storage capacity and inflow, age of
reservoir, shape of the reservoir basin, the type of outlets and method of operation, size
grading of the sediment, and the behavior of the finer sediment fractions under various
conditions.

In large reservoirs, possibly those with 10,000 acre-ft or more of storage capacity, it
may be assumed that the trap efficiency will be 100%, i.e., all the sediment entering the
reservoir will remain there. There may be some sediment moved through the pool in
density flows or during periods of very high discharge.

368
 Reservoir Sedimentation

In small reservoirs, sometimes most of the inflowing sediment may be transported

through the pool. This may also occur during high inflow periods when a reservoir
discharges over the spillway and there is an appreciable velocity of flow through the
reservoir. The proportion of sediment passing through the reservoir will depend
primarily on the factors: the average velocity of flow through the pool and the character
of the sediment. In respect to the later, fine sediments (the silt and clay sizes) may
remain in suspension long enough to pass through the reservoir. Sand sizes will not.

A number of attempts have been made to correlate trap efficiency with one or more of
these factors. One of the earlier studies of trap efficiency was made by Brune and Allen
(see Moore, Wood and Renfro, 1960), who developed a curve relating percentage of
eroded soil caught in a reservoir with capacity per square mile of drainage area. The trap
efficiency values given by this method, however, are necessarily low because they are
based upon rates of erosion which are higher than rates of sediment production to the
reservoir.

C.B. Brown (see Moore, Wood and Renfro, 1960) first developed a curve relating the
reservoir capacity-watershed area ratio, C/W, and true trap efficiency. The curve shown
in Fig. 15 with some additional records is represented by:

CT = 100 [1 − 1 /(1 + 0.1C / W )] (15)

where CT is the reservoir trap efficiency in percentage, and C/W is the reservoir capacity
in acre feet per square mile of drainage area. There is a considerable spread in points as
may be noticed. The main reason for the spread of points in the curve is that reservoirs
having the same C/W ratio may have very different capacity-in flow ratios.

Churchill (1948), (see Barland, 1971) presented a relation based on Tennessee Valley
Authority reservoirs. His method relates the percentage of incoming sediment passing
through a reservoir and the sediment index of the reservoir, i.e., the period of retention
(capacity, in cubic feet, at mean operating pool level divided by average daily inflow

Figure 15. Brown’s trap efficiency curve (after Moore, Wood and Kenfro, 1960)

369
Sediment Transport Technology

rate, in cubic feet per second) divided by velocity (mean velocity, in feet per second,
obtained by dividing average cross sectional area, in square feet, into the inflow). The
average cross sectional area in this case is compute by dividing capacity, in cubic feet,
by length, in feet. Churchill’s curve is shown in Fig. 16.

Borland (1971) verified Churchill’s method with reasonable accuracy by applying

known data from a number of sources including reservoirs with a capacity of several
hundred thousand acre-feet, and concluded that Churchill’s method gave better results
than the Brune’s curve.

Brune (1953), (see Borland, 1971), presented an empirical relationship based on the
records of 44 normally pounded reservoirs. His curves, relating trap efficiency and the
ratio between reservoir capacity and mean annual water inflow both in acre-feet, are
shown in Fig. 17.

Figure 16. Churchill’s trap efficiency curve (after Borland, 1971)

Figure 17. Brune’s (1953) trap efficiency curve (after Borland, 1971)

370
 Reservoir Sedimentation

4.2. Area – increment method

The area-increment procedure, which was developed by Borland and Miller (1958), is
based on the assumption that the sediment deposition in a reservoir can be approximated
by reducing the reservoir area at each reservoir elevation by a fixed amount. The area
increment method can be expressed by the following equation,

Vs = A0(H – h0) + V0 (16)

in which Vs represents the sediment volume in acre-feet to be distributed in the

reservoir, A0 is the area correction factor in acres which is the original reservoir area at
the new zero elevation at the dam, H is the reservoir depth at the dam in feet from the
stream bed to the maximum normal water surface, h0 is equal to the depth in feet to
which the reservoir is completely filled with sediment (new zero elevation), and V0 is
the sediment volume in acre-feet below the new zero elevation. Eq. 16 states
mathematically that the total volume of sediment, Vs, consists of the portion that is
uniformly distributed vertically over the height (H-h0) plus the portion below the new
zero elevation of the reservoir. Fig. 18 gives an idea about the related parameters and
the method.

The area-increment method is only an approximation of the probable location of the

sediment deposits. It is known that sediment does not generally reduce the surface area
as by the same amount at every elevation. However, comparisons have been made
between the results obtained by this method and the actual resurvey results, and good
agreement has been achieved in most cases. It should be noted that the applicability of
the method decreases with an increase in the ratio of volume of sediment to capacity.
Also, the accuracy decreases with any deviation of the reservoir in question from a more
or less standard type. Usually the arbitrary stipulation is that if the 100-yr sediment
accumulation exceeds 15% of the original capacity, a more exacting method of
distributing the sediment should be applied. In any case, the area-increment method is a
good first-check procedure for determining the probable sediment distribution.

Figure 18. Sketch of the area increment method

371
Sediment Transport Technology

4.3. Empirical area - reduction method

This method was developed from analysis of sediment distribution data obtained from
resurvey of 30 reservoirs. A series of curves defining type of sediment deposit for
different types of reservoirs, gerge hill, flood plain, foothill, and lake was developed by
Borland and Milller (1958) to determine the constants needed for mathematical
computation of a design curve. The equation used for this purpose is

Ap = Cpm (1 – p)n (17)

where Ap represents a dimensionless relative area related to deposition at a relative

distance p above the stream bed, and C, m, and n are dimensionless constants
determined by the type of reservoir.

The method permits the classification of a proposed reservoir into a standard type that
immediately gives an indication of the manner in which the sediment will distribute.
And also it gives a good approximation of the height of sediment at the dam at various
time intervals.

4.4. Elevation of sediment - accumulation method

Regression analyses of sediment accumulation in small retarding-type reservoirs in the
deep loess area of western Iowa (see Chow, 1964) indicated that the minimum elevation
of sediment accumulation immediately upstream from the dam could be expressed by

Y = 22.6 + 0.886D – 81.2n – 0.175C + 0.494W (18)

where Y is the original reservoir depth filled with sediment in percent, D is the total
original storage depletion at the end of the design period in percent, n is the original n
value or the slope on log-log paper of depth vs. capacity, C is the total storage capacity
remaining at the end of the design period in acre-ft, and, W is the specific weight of
sediment in lb/ft3.

In another statistical study conducted by Borland (1971), upon the survey of 27

reservoirs, the variation of the slope of upstream faces of deltas compared to the initial
reservoir bottom slope is analyzed and results are given in Fig. 19. From the figure it
can be concluded that the upstream face slopes of deltas are about half of the initial
reservoir bottom slope. From Borland’s studies it was also shown that the slope of front
faces of deltas is about 6.5 times of the slopes of delta upstream faces.

Validity and usefulness of all the methods described above, without any doubt, will
depend on the reliability of related measurements and surveys. When a method is to be
selected to use for a reservoir, instead of having an idea to compare available methods
with each other to decide which one is the best, the applicability of them to different
reservoirs should be analyzed. If characteristics of a reservoir, on which one of the
methods have been applied, are similar to that for which a method is being searched,
then that method can be used with confidence. As a conclusion we can say that
equations or relations which have been derived for a reservoir should not be directly
used for other reservoirs unless necessary modifications are made. Quantitative results

372
 Reservoir Sedimentation

Figure 19. Variation of delta slopes with original stream bed slopes obtained by
Borland from surveys of 27 reservoirs (after Borland, 1971)

of these methods, which are based on measurements and surveys, can be valid only for
very special cases. From all these explanations it is very clearly seen that in new
reservoirs measurements and surveys should be done very seriously and carefully.

5. Control of Reservoir Sedimentation

The sediment yield from the watershed, the trap efficiency of the reservoir, and the
ultimate specific weight of deposited sediment fixed the volume of sediment which
must be considered in the design of a structure to avoid premature loss of storage
dedicated to a specific purpose for a given period of time. If economic or physical
limitations are such that storage space can not be incorporated in the reservoir, other
means of reducing sediment inflow or periodic removal of sediment must be considered
to assure proper functioning of the reservoir for the period for which it is designed. This
chapter briefly outlines provisions, which might be considered in the design and
maintenance of structures to assure functioning.

5.1. Design of reservoir

The basic considerations in developing a reservoir are hydrologic conditions and the
service requirements for the reservoir for a given period of time. To prevent premature
loss of storage and the services dependent upon it, an additional storage volume is
usually incorporated in the reservoir for sediment accumulation. This volume of storage,
generally referred as the sediment pool, is equivalent to the volume of sediment
expected to deposit in the reservoir during the design life of the structure.

373
Sediment Transport Technology

The sediment pool of many reservoirs built in the past was often allocated in the lower
elevations of the reservoir, and in some instances the dead storage served as the
sediment pool. More recent studies indicate this procedure to be inadequate for
reservoirs where sediment will be distributed at higher elevations. In reservoirs the
allocation of storage space by elevation is important in determining the elevation of the
outlet sill.

5.2. Venting sediment

Discharge of water flowing over a spillway, in general, is clear and more turbid water,
from which the sediment eventually deposited, remains in the reservoir. Occasionally,
appreciable amounts of sediment may be vented from reservoirs by use of gated outlets
or special curtains to encourage movement of high concentration of sediment in
suspension, through or over the dam.

The placing of outlets in the lower portion of the dam for the purpose of drawing of the
lower strata of water having higher concentrations of sediment has been put to actual
practice, but records of the effectiveness of these outlets, in terms of percent of total
sediment yield discharged, are lacking.

5.3. Sediment removal

Deposited sediment may be removed from reservoirs periodically by hydraulic or
mechanical means. This method has been used successfully to restore the capacity of
many small reservoirs.

The type of equipment used for most economical removal of sediment depends upon the
nature of sediment and location of the disposal area. Hydraulic dredging is the most
economical for fine-grained submerged materials where nearby disposal areas are
available. Often disposal sites can be created by leveeing areas for sediment storage
adjacent to the reservoir.

Coarse-grained materials require mechanical equipment for removal. A fronted loader

or a power shovel may be used for removal of deposits if not submerged.

Off-site disposal of sediment deposits which requires rehandling of materials and

trucking is costly, unless the sediment has commercial value such as for topsoil, land fill
or concrete aggregate.

5.4. Reduction of sediment yield

Sediment-storage requirements for a reservoir may be reduced if sediment can be
retained in the watershed by one means or another. Such retention may be affected by
development of vegetative screens in the reservoir area and by installation of structures
and land-treatment measures in the watershed.

5.4.1. Vegetative screens

Vegetation normally seeds on exposed delta deposits of reservoirs to form natural
vegetative screens which reduce inflow velocities and increase the roughness
coefficients, encouraging deposition above the reservoir crest elevation within the

374
 Reservoir Sedimentation

reservoir basin or immediately upstream. Often the natural seeding processes can be
accelerated to develop more effective screens. This may be brought about by the
introduction of faster-growing and more effective species of plants, by extending
plantings to unseeded areas, and by increasing the plant density of natural screens. The
deposition of sediment above crest can often be enhanced by construction of diversions
in the delta areas to develop better distribution of incoming flood waters and deposition
in backwater areas.

The effectiveness of vegetative screens in reducing sediment inflow to active storage

space depends upon the character of the sediment and the sediment-storage space
available at the head of the reservoir. Generally, the additional amount which could be
stored above crest level within a reservoir area by this method will not exceed 10% of
the total incoming load. However, under certain conditions, such materials may initiate
upstream aggradation, resulting in serious seepage and floodwater damages upstream.

5.4.2. Watershed structures

Several types of structures may be built in a watershed for the specific purpose of
reducing sediment yield to a reservoir. These include such structures as sedimentation
basins to trap sediment below eroding areas and erosion control structures to halt the
production of sediment.

Sedimentation basins are essentially reservoirs, generally located in canyons, designed

to catch the, coarse sediments and prevent their being carried downstream onto
residential or other critical areas. They are not effective means of reducing movement of
fine-grained materials. These types of structures facilitate removal and disposal of
sediment, which might not be possible in the reservoir area itself.

The reduction of channel erosion normally requires a structure. In a watershed, where

the primary source of sediment is derived from channel erosion, installation of control
structures may have an appreciable effect in reducing sediment yield to a reservoir.
Such structures include drop inlets and chutes for reduction of gully erosion, stream-
bank revetment to reduce stream-bank erosion, and sill or drop structures for streambed
stabilization.

5.4.3. Watershed land - treatment measures

Land treatment measures provide an effective and economical means of reducing
erosion and sediment yield where the primary source of sediment is sheet erosion.
These measures include soil improvement, proper tillage methods, strip cropping,
terracing, and crop rotations and others.

The improvement of soil by addition of organic residues increases the infiltration rate
and reduces sheet runoff and sheet erosion. The infiltration rate may also be increased
by tillage methods such as basin listing and contour plowing. The installation of terraces
not only promotes infiltration, but also reduces the effective length of slope, with
correspoading reduction of erosive velocities.

375
Sediment Transport Technology

5.4.4. Control of wind erosion

Wind erosion is a serious problem in areas of low precipitation, frequent drought, and
where temperatures, evaporation, and wind speeds are high. Basic principles of wind-
erosion control are as follows:
a) Produce, or bring to the surface, aggregates or clods.
b) Roughen the land surface to reduce the wind velocity and trap drifting soil.
c) Reduce field widths along the prevailing wind direction by establishing wind
barriers.
d) Establish and maintain vegetation or vegetative residues to protect soil.

5.5. Evaluation of selected control measures

In most cases, it will be cheaper to incorporate storage in the reservoir than to provide
other means of control. This is particularly true where large storage capacities are
developed per unit of drainage. In such instances the additional capacity needed for
sediment accumulation is only a small fraction of the total and requires a dam which
may be only a matter of inches higher than that needed solely for hydrologic reasons or
service requirements.

Where high sediment yields prevail, or where the storage capacity per unit of drainage
is low, the additional storage and height of dam needed to provide for sediment control
may be appreciable. In such instances, additional sites should be investigated, if
available, to determine whether watersheds with lower sediment yields prevail in the
area.

If it is not possible physically feasible to construct a dam of suitable height to provide

for sediment storage in addition to hydrologic or service requirements, alternative
methods of sediment control need to be considered to assure proper functioning of the
reservoir. These include the possibility of venting sediment, periodic removal of
sediment from the reservoir, development of vegetative screens, and installation of
structural or land-treatment measures in the watershed. Each method needs to be
evaluated separately to determine its feasibility, effectiveness, and cost, in selecting the
most practical method.

References

1. Adachi, S., and Nakatoh, T., “Changes of Top-Set-Bed in a Silted Reservoir”, Proc.
IAHR, 13th Congress, Vol.5-1, Kyoto, Japan, (1969).
2. Asada, II., “Some Examples of Bed Profile Calculation of Sedimentation in
Reservoirs in Mountainous Regions”, Proc. IAHR, 15th Congress, Vol. I, Istanbul,
Turkey, (1973).
3. Bhamidipaty, S., and Shen, H. V., “Laboratory Study of Degradation and
Aggradation”, Proc. ASCE, Vol. 97, No. WW4, (1971).
4. Borland, W. M., “Reservoir Sedimentation”, in River Mechanics, (editor: H. Shen),
Colorado State Univ., Water Resources Publication, Fort Collins, Colo, (1971).
5. Borland, W. M., and Miller, C. R., “Distribution of Sediment in Large Reservoirs”,
Trans. ASCE, Vol. 125, (1960).

376
 Reservoir Sedimentation

6. Brune, G. M., “Trap Efficiency of Reservoirs”, Trans. Amer. Geophys. Union, Vol.
34, No. 3, (1953).
7. Chang, F. F. M. “Computer Simulation of Riverbed Degradation and Aggradation
by the Method of Characteristics”, Proc. IAHR, Vol. 1, 13th Congress, Kyoto,
Japan, (1969).
8. Chow, V. T., “Handbook of Applied Hydrology”, Mc Graw-Hill Book Co. Inc.,
New York, (1964).
9. Garde, R. J., and Swamee, P. K., “Analysis of Aggradation Upstream of a Dam”,
Proc. IAHR, Vol. 1, Int. Symposium on River Mechanics, Bangkok, Thailand,
(1973).
10. Graw, W. H. “Hydraulics of Sediment Transport”, Mc Graw-Hill Book Co. Inc.,
New York, (1971).
11. Harrison, A. S., “Deposition at the Head of Reservoirs”, Proc. of 5th Hydraulics
Conference, Univ. of Iowa City, (1952).
12. Harrison, A. S. “Computation of Aggradation Trends in the Headwaters of Big
Bend Reservoir”, U.S. Crops of Engineers, Memo., Omaha, Nebraska, (1958).
13. Kobanbay, Ö., “Non-Steady Sediment Transport Upstream from a Weir”, M.Sc.
Thesis, Hydraulics Lab., Department of Civil Eng., Middle East Technical
University, Ankara, Turkey, (1969).
14. Moore, C. M., Wood, W. J., and Renfro, G. W., “Trap Efficiency of Reservoirs,
Debris Basins, and Debris Dams”, Proc. ASCE, Vol. 86, No. HY2, (1960).
15. Sanoyan, V. G., “Calculation of the process of Sedimentation and Hydraulic
Washout of River Reservoir”, Proc. IAHR, Vol. 5, 14th Congress, Paris, France,
(1971).
16. Sugio, S., “Laboratory Study of Degradation and Aggradation-A Discussion”, Proc.
ASCE, Vol. 98, No. WW4, (1972).
17. Tsuchiya, A., and Ishizaki, K., “Estimation of River Bed Aggradation due to a
Dam”, Proc. IAHR, Vol. 1, 13th Congress, Kyoto, Japan, (1969).
18. Vanoni, V. A., Editor, “Sedimentation Engineering”, By the Task Committes for
the Preparation of the Manual on Sedimentation Committee of the Hydraulics
Division, ASCE M and R, No. 54, New York, (1975).
19. Yücel, Ö., and Graf, W. H., “Bed Load Deposition in Reservoirs”, Proc. of IAHR,
Vol. 1, 15th Congress, Istanbul, Turkey, (1973).
20. Yücel, Ö., “Baraj Göllerinde Çökelme Olayının Model Üzerinde Araştırılması”,
T.B.T.A.K. Araştırma Projesi, MAG-376, Hidrolik Lab., Ankara, (1975).

377
378
 DESIGN OF STABLE CHANNELS

Prof. Dr. A. Melih Yanmaz

Middle East Technical University
Civil Engineering Department
Ankara - Turkey

1. General
An open channel, either natural or artificial, is a conduit in which water flows with a
free surface. Natural channels include all watercourses that exist naturally on the earth,
such as brooks, creeks, streams, and tidal estuaries. Artificial channels, which are either
lined or unlined, are those constructed to perform various project requirements e.g.
irrigation canals and flumes, drainage canals, floodways, navigation channels, roadside
gutters, and laboratory flumes. The hydraulic properties of open channels must be
designed to meet the given requirements with reasonable cost. Although the type of
channel is selected on the basis of cost analysis, drainage canals must be specifically
built as earth channels to fulfill their functions properly. This chapter discusses the
design methods for erodible channels.

The proper cross-section for an erodible channel must have adequate hydraulic capacity
for a minimum cost of construction and maintenance. As a general guide, the cross-
sections of unlined channels are recommended as trapezoidal in shape with side slopes
depending mainly on the type of foundation material. Table 1 gives preliminary values
for the suitable side slopes of various kinds of material. Fig. 1 shows a typical cross-
section of an erodible channel having side slopes of 1V:zH, where z is the horizontal
value of the side inclination. For a final decision, the side slope stability is to be
checked using a suitable approach, such as the tractive force analysis.

A trapezoidal unlined channel will eventually be stabilized as shown in Fig. 2. Because

of greater shear stresses and effect of gravity, initial disturbance is experienced at the
toe of side slopes. Various stages of the stabilization process are illustrated in Fig. 2.

Table 1. Recommended side slopes for unlined channels [Chow, 1959]

Material Side Slope (H:V)

Rock Nearly vertical

Muck and peat soils 1/4: 1
Stiff clay or earth with concrete lining 1/2: 1 to 1:1
Earth with stone lining or earth for large channels 1:1
Firm clay or earth for small ditches 1.5:1
Loose sandy earth 2:1
Sandy loam or porous clay 3:1

379
Sediment Transport Technology

Figure 1. Definition sketch for a trapezoidal erodible channel

Initial water level Final water level

1 3 4
2
5

Stage 1: Man-made trapezoidal channel

Stage 2: Initial erosion
Stage 3: Advanced bank erosion
Stage 4: Sides sliding into channel
Stage 5: Final shape of the section
Figure 2. Stabilization process of an erodible channel [Simons and Şentürk, 1992]

An erodible channel may experience the following modes through the stabilization
process. A channel is said to be in regime if it has a dynamic equilibrium in long term.
Depending on the sediment transport rate, which is affected by flow and sediment
characteristics, a channel may be subject to scouring, accretion, or both along various
reaches. A stabilized channel, which attains its equilibrium in long term, may be
considered as a channel having fixed boundaries. The behavior of an erodible channel
can be assessed with reference to the concept of degree of freedom. A lined prismatic
channel has one degree of freedom, i.e. there is one flow depth, y, for the given steady
flow rate, Q. Two degrees of freedom takes place in a lined channel, which carries
sediment load. The depth, y, and bed slope, S0 are adjusted for the given discharge. In
case of an erodible channel, there exists three degrees of freedom. For the given steady
discharge, the channel adjusts its channel depth, y, surface width, B, and the bed slope,
S0. Lacey (1930) found that natural channels have a tendency to experience a stabilized
section having a semi-elliptical shape. The coarser the sediment, the flatter is the semi-
ellipse and greater is the width at the water surface. Finer material would lead to the
development of almost a semi-circular shape as shown in Fig. 3.

Coarse
Medium

Fine

Figure 3. Effect of sediment size on channel section [Garde and Ranga Raju, 1978]

380
 Design of Stable Channels

Based on investigations of several researchers, it was observed that the depth at a

section increases with the discharge faster than the increase in surface width.

2. Stable Hydraulic Section

The section of an erodible channel, which is not subject to erosion at a minimum flow
area for a given discharge, is called the stable hydraulic section. The U.S. Bureau of
Reclamation has employed the principle of tractive force to develop a theoretically
stable section for erodible channels carrying clear water in non-cohesive materials. In a
stable hydraulic section for maximum efficiency, it is necessary to satisfy the condition
that impending motion shall prevail along the channel perimeter. For a given angle of
repose and discharge, this optimal section will provide not only the channel of
minimum flow area, but also the channel of minimum top width, maximum mean
velocity, and minimum excavation. Typical stable section is shown in Fig. 4.

The critical shear stress acting on differential length dl in Fig. 4 is

γyS 0 dx
τs = = γyS 0 cos β (1)
(dx) 2 + (dy ) 2

On the other hand, the maximum shear stress at the sides is

tan 2 β
τ s = Kτ b = γy 0 S 0 cos β 1 − (2)
tan 2 θ

where K is the tractive force ratio. Combining Eqs. 1 and 2, the water depth, y, can be
obtained as follows:

tan 2 β y
y = y0 1 − = 0 tan 2 θ − tan 2 β (3)
tan θ
2
tan θ

Substituting tanβ = (dy)/(dx) into Eq. 3, one obtains

2 2
⎛ dy ⎞ ⎛⎜ y ⎞⎟
⎜ ⎟ + ⎜ ⎟ tan θ − tan θ = 0
2 2
(4)
⎝ dx ⎠ ⎝ y 0 ⎠

B
θ
x
yo y
dl
dy

y
dx β

Figure 4. Stable hydraulic section

381
Sediment Transport Technology

This is the differential equation of the channel profile. By taking y = y0 at x = 0, the

coordinates of the stable hydraulic section is determined from

⎛ tan θ ⎞
y = y 0 cos⎜⎜ x ⎟⎟ (5)
⎝ y0 ⎠

where y0 is the maximum water depth as shown in Fig. 4 which can be approximately
determined from τp/(0.97γS0), where τp is the permissible unit tractive force.

The Manning equation is expressed for the stable hydraulic section as follows:

ky 08 / 3 S 01 / 2
Q′ = (6)
n
where Q ′ is the discharge that passes through the stable hydraulic section and k is a
parameter, which is a function of θ. With the known particle characteristics of soil, the
value of k, and hence the theoretical discharge, Q ′ , is computed from Eq. 6. If the
theoretical discharge is greater than the given design discharge, Q, the section
dimensions are reduced by subtracting the following horizontal distance, B ′ from the
top width, B:

⎛ Q′ ⎞
B ′ = 0.96⎜⎜1 − ⎟B
⎟ (7)
⎝ Q ⎠

If Q ′ < Q, the section dimensions should be increased by adding a rectangular section

having a width of B ′′ to the midsection:

n(Q − Q ′′)
B ′′ = (8)
y 05 / 3 S 0

Although a stable hydraulic section has a minimum excavation, it is not practical in

engineering applications due to difficulty and higher cost of construction of elliptic or
parabolic sections in the field.

3. Stable Channel Design

Once the formative discharge, Q, and the bed material characteristics are given, the
channel dimensions, i.e. B, y, and S0, which contribute the stable section, can be
determined. To obtain these variables, three equations are needed. This text discusses
the regime theory, and Altunin’s (1972) method for stable channel design. Copeland’s
(1994) method, which is based on empirical equations for sediment discharge and flow
depth computations for trapezoidal channels, will not be covered in this chapter.

3.1. The regime theory

The regime theory, although does not incorporate physical explanations for the
phenomenon, is an empirical approach that has been developed by observing the

382
 Design of Stable Channels

characteristics of stable channels. Initial investigations have been carried out in India,
Pakistan, and Egypt. Extended investigations on this approach have been performed in
the USA and Canada. The regime theory sizes an erodible channel such that its sections
and slope are in equilibrium with the discharges. Therefore, increments or decrements in
Q, modify B and y. That is why after the annual periods, sections and slopes remain
practically constant. This method may be applicable for the design of earth channels
having cohesive or sandy materials. Although there has been several regime equations
reported in the literature, only those, which recapitulate all the previous methods and
consider the characteristics of bed and bank materials, will be included in this text.
Kennedy (1895) proposed the first regime equation. He observed and measured the
dimensions of a set of stable channels. Based on his intensive research, he realized that
the mean velocity was a function of the flow depth as

u = 0.548 y 0.64 (9)

where u is the mean flow (non-silting) velocity (m/s), and y is the mean flow depth (m).
Based on new observations, Eq. 9 has been generalized as

u = cy m (10)

where c varies from 0.67 to 0.95 (Lacey); 0.25 to 1.20 (others), and m is the exponent
that varies between 0.52 to 0.64 (Lacey); 0.61 to 0.73 (others).

Lindley (1919) used the term regime for the first time. His equations are given below.

u = 0.95 y 0.57 (11)

u = 0.59 B 0.355 (12)

Combining Eqs. 11 and 12, one can obtain

B = 3.0 y 1.61 (13)

where y and B are in ft and u is in fps. Lacey (1929-1949) [See Lacey, 1966] defined the
silt factor, f, for the first time. His equations are as follows:

P = 4.838 Q (14)
1/ 3
⎛Q⎞
R = 0.4725⎜⎜ ⎟⎟ (15)
⎝f ⎠
f 5/3
S = 0.000302 1 / 6 (16)
Q
f = 55.66 Dm (17)

where P is wetted perimeter (m); R is hydraulic radius (m); S is slope; Q is design

discharge (m3/s), and Dm is mean diameter of the bed material (m). Lacey proposed the
following equation for the mean flow velocity in m/s.

383
Sediment Transport Technology

R 2 / 3 S 1/ 2
u= (18)
na

where

Dm1 / 8
na = 0.0225 f 0.25 = (19)
16.27

Lacey’s equations can be reduced to:

u = 2Q 5 / 36 Dm13 / 72 (20)

Lacey’s design procedure can be applied as follows:

1. With the known P and R, section dimensions are determined for a trapezoidal
section:

P = b + 2y 1+ z2 (21)
by + zy 2
R= (22)
b + 2y 1+ z2

Note that for a very wide channel, i.e. b>40y, the section can be treated as
rectangular:

P = b + 2y (23)
by
R= (24)
b + 2y

from which the following expressions can be written:

b = P − 2y (25)
2 y 2 − Py + PR = 0 (26)
A = PR (27)

2. For given Dm and Q, the mean flow velocity is computed from Eq. 20.
3. The flow area is computed from A = Q/u.
4. It is checked whether or not A = P·R. If it holds true, the values of b and y are
acceptable. Otherwise, the values of na and f should be modified.
5. For a trapezoidal channel, the similar procedure is followed. The only additional
information is the side slope value, z, which can be taken from Table 1.

Blench (1941-1961) [See Blench, 1957] separated the silt factor into two components as
bed factor, fb, and side factor, fs, which are defined as:

f b = 60.1 Dm (1 + 0.012C ) (28)

384
 Design of Stable Channels

in which C is the suspended sediment concentration (ppm). Eq. 28 is applicable to

sandy beds with dunes. For fine sand, i.e. Dm<0.0005 m, fb≈0.8, for medium and coarse
sand, i.e. Dm ≥0.0005 m, fb ≈1.2. The side factor is:

60.1 Dms
fs = (29)
8

where Dms is the mean size of sand at the bank (m). Typical side factors are:
fs ≈0.1 for material with less cohesion
fs ≈0.2 for moderately cohesive material
fs ≈0.3 for very cohesive material (clay)

Blench’s basic equations are listed as:

u2
f b = 3.28 (30)
y
u3
f s = 10.76 (31)
bm
1/ 4
u2 ⎛ C ⎞⎛ ubm ⎞
= 3.63⎜1 + ⎟⎜ ⎟ (32)
gyS ⎝ 2330 ⎠⎝ ν ⎠

where ν is kinematic viscosity (m2/s), and bm is mean width of the section (m). The
design procedure is outlined below.

1. The design equations are obtained from basic equations as follows:

1/ 2
⎛ f Q⎞
bm = 1.81⎜⎜ b ⎟⎟ (33)
⎝ fs ⎠
1/ 3
⎛ f Q⎞
y = ⎜⎜ s 2 ⎟⎟ (34)
⎝ fb ⎠
f b5 / 6 f s1 / 12
S= (35)
116.8Q 1 / 6 ⎛ C ⎞
⎜1 + ⎟
ν 1/ 4
⎝ 2330 ⎠

The relation between bm and y is obtained from

1.49 f b y
bm2 / 3 = (36)
f s2 / 3

2. For a trapezoidal section, a suitable value is selected for the side inclination, z. The
section dimensions are determined from:

B = bm + zy (37)

385
Sediment Transport Technology

b = bm − zy (38)

Simons and Albertson (1963) introduced five types of channel whose characteristics are
given in Table 2.

Equations proposed by Simons and Albertson (1963) are listed below.

P = K 1Q 0.512 (39)
bm = 0.9 P (40)
R = K 2 Q 0.361 (41)
A
≅ y = 1.21K 2 Q 0.361 for R ≤ 2.13 m (42)
bm
y = 0.61 + 0.93K 2 Q 0.361 for R > 2.13 m (43)
0.37
C2 u2 ⎛ ub ⎞
= = K4⎜ m ⎟ (44)
g gyS ⎝ ν ⎠

where C is the Chezy coefficient. The coefficients K1 through K4 are determined from
Table 3.

The design procedure proposed by Simons and Albertson (1963) is outlined below:
1. With the given channel type and Q, P is determined from Eq. 39.
2. Values of bm and R are obtained from Eqs. 40 and 41, respectively.
3. The flow area is determined from A = PR.
4. The flow depth, y, is determined from Eqs. 42 or 43.
5. The surface and bottom widths are determined from B = bm+zy and b = bm-zy.

Table 2. Channel types [Simons and Albertson, 1963]

Channel type Characteristics of material

1 sand bed and banks
2 sand bed and cohesive banks
3 cohesive bed and banks
4 coarse noncohesive material
5 same as 2 but heavy sediment load (2000-8000 ppm)

Table 3. Coefficients of Simons and Albertson’s method.

Channel type K1 K2 K4
1 6.30 0.41
2 4.74 0.47 0.324
3 3.96 0.56 0.525
4 3.16 0.27 0.87
5 3.09 0.36

386
 Design of Stable Channels

Altunin (1972) classified the sections of river reaches as follows:

Type a: Banks are very resistant to erosion, which may consist of very cohesive
materials
Type b: Banks with little resistance to erosion like alluvial material without
cohesion.

Altunin’s (1972) equations are:

B m = Ky h (45)

where B is surface width of channel (m), yh is hydraulic depth (m) which is (A/B), K is
shape coefficient whose values range between 8-12 for alluvial channels (≈10), 3-4
(type a), and 16-20 (type b). The exponent m in Eq. 45 is given as:

0.1
⎛ ∆Dm ⎞
m = 0.72⎜⎜ ⎟⎟ (46)
⎝ yS ⎠

in which ∆ is relative density. Another relation for m is given by:

us − u
m = 0 .5 + (47)
u

where us is surface velocity (m/s), and u is mean flow velocity (m/s). Altunin (1972)
gives another relationship between B and Q as follows:

1 5 / 3 1/ 2
Q= By h S (48)
n

Combining Eqs. 45 and 48, one obtains:

3 3
⎛ nQK 5 / 3 ⎞ 3+5 m ⎛ Q ⎞ 3+ 5 m
B = ⎜⎜ 1/ 2
⎟⎟ = E⎜ 1/ 2 ⎟ (49)
⎝ S ⎠ ⎝S ⎠

where

( )
3
E = nK 5 / 3 3+ 5 m (50)

Altunin (1972) proposed the following equations for three degrees of freedom:

0.00192 E 15 / 23 (auφ )
75 / 23

S= (51)
Q15 / 46
3.494Q13 / 23 E 20 / 23
B= (52)
(auφ )15 / 23

387
Sediment Transport Technology

0.353Q 25 / 69
yh = (53)
(auφ )20 / 69 E 50 / 69
where the value of a is 1.0 for mountainous and plateau zones, and 1.5 for plains. In the
above equations, uφ is mean threshold velocity (m/s) which is expressed as:

uφ = 113.6 Dm + 0.466 for 0.0003 m ≤ Dm ≤ 0.00263 m (54)

1
uφ = for 0.00263 m < Dm ≤ 0.0303 m (55)
1.341 − 12.5Dm
0.366
uφ = 2.248 − for 0.0303 m < Dm ≤ 0.0865 m (56)
Dm
Dm
uφ = for Dm > 0.0865 m (57)
0.259 Dm + 0.0247

The following concluding remarks can be made for the stable channel design. Kennedy
and Lindley have not provided explicit equations for y, i.e. y is a function of u and B,
which are also design variables. They have not considered bed and bank resistance in
terms of a silt factor. Lacey considered only the bed resistance with f factor, whereas
Blench considered both bed and bank resistances. Blench, and Simons and Albertson
proposed similar approaches with almost equal computational efforts. Altunin proposed
an elaborate analysis. However, no explicit expressions for S, B, and y have been
provided since m = f(y, S).

4. Example
An earth channel will be designed to carry Q=10 m3/s using the regime theory and
Altunin's method. The bed material is composed of uniform coarse sand with Dm=2 mm.
The banks are of moderately cohesive clay. The channel is expected to subject to almost
clear water conditions with C≈50 ppm. The following values are accepted in the design:
z = 2.0, ν = 10-6 m2/s, ∆ = 1.65, and n = 0.025.

Solution
a) Lacey's method:

u = 2(10 ) (0.002)13 / 72 ≅ 0.9 m/s

5 / 36
(Eq. 20)

Q 10
A= = = 11.1 m2
u 0.9

P = 4.838 10 = 15.3 m (Eq. 14)

f = 55.66 0.002 = 2.49 (Eq. 17)

1/ 3
⎛ 10 ⎞
R = 0.4725⎜ ⎟ = 0.75 m (Eq. 15)
⎝ 2.49 ⎠

388
 Design of Stable Channels

A = PR = 15.3*0.75 = 11.48 m2 ≈ 11.1 m2 O.K.

S = 0.000302
(2.49)5 / 3 = 0.00094 (Eq. 16)
(10)1 / 6
Let us adopt a trapezoidal section with z = 2.0.

P = b + 2 y 1 + 2 2 = b + 4.47 y = 15.3 or

b = 15.3-4.47y (58)

by + 2 y 2
R= = 0.75 (59)
b + 4.47 y

Simultaneous solutions of Eqs. 58 and 59 yields:

y 2 − 6.19 y + 4.65 = 0

y1 = 5.32 m, which is not acceptable since it gives b = - 8.48 m

y2 = 0.88 m O.K. Using Eq. 58,

b = 15.3-4.47*0.88 = 11.37 m
B = 11.37+2.0*2.0*0.88 = 14.89 m.

Section dimensions of the channel are shown in Fig. 5.

b) Blench's method:

for Dm = 0.002 m → fb ≈ 1.2. Side factor can be taken as fs=0.2.

1/ 2
⎛ 1.2 *10 ⎞
bm = 1.81⎜ ⎟ = 14 m (Eq. 33)
⎝ 0.2 ⎠

1/ 3
⎛ 0.2 *10 ⎞
y = ⎜⎜ ⎟
2 ⎟
= 1.12 m (Eq. 34)
⎝ (1.2) ⎠

B=14.89 m

y=0.88m 1
S=0.00094 2

b=11.37 m

Figure 5. Channel dimensions based on Lacey’s method

389
Sediment Transport Technology

(1.2) 5 / 6 (0.2)1 / 12
S= = 0.000184 (Eq. 35)
116.8(10)1 / 6 ⎛ 50 ⎞
⎜1 + ⎟
(10 −6 )1 / 4 ⎝ 2330 ⎠

B = 14+2*1.12 = 16.24 m (Eq. 37)

b = 14-2*1.12 = 11.76 m (Eq. 38)

Section dimensions of the channel are shown in Fig. 6.

c) Simons and Albertson’s method:

Channel type 2 with K1 = 4.74, K2 = 0.47, and K4 = 0.525

P = 4.74(10) 0.512 = 15.4 m (Eq. 39)

bm = 0.9 *15.4 = 13.9 m (Eq. 40)

R = 0.47(10) 0.361 = 1.08 m (Eq. 41)

Since R ≤ 2.13 m, Eq. 42 is used.

y = 1.21*1.08 = 1.31 m

A = PR = 15.4*1.08 = 16.6 m2

Q 10
u= = = 0.6 m/s
A 16.6

S is determined from Eq. 44.

(0.6) 2
S= 0.37
= 0.000146
⎛ 0.6 *13.9 ⎞
9.81 * 1.31 * 0.525⎜ −6 ⎟
⎝ 10 ⎠

B and b values are determined from Eqs. 37 and 38 as follows:

B=16.24 m

y=1.12m 1
S=0.000184 2

b=11.76 m

Figure 6. Channel dimensions based on Blench’s method

390
 Design of Stable Channels

B = 13.9+2*1.31 = 16.52 m and b = 13.9-2*1.31 = 11.28 m.

Section dimensions of the channel are shown in Fig. 7.

d) Altunin's (1972) method:

Since S and y values are unknowns at the beginning of the computations, the value of m
cannot be determined. However, since the results of Blench, and Simons and Albertson
approaches are close to each other, the average values of S and y, obtained from these
two methods, may be used for determining Altunin’s m value.

0.000184 + 0.000146 1.12 + 1.31

S≅ = 0.000165 y≅ = 1.215 m
2 2

0.1
⎛ 1.65 * 0.002 ⎞
m = 0.72⎜ ⎟ = 0.95 (Eq. 46)
⎝ 1.215 * 0.000165 ⎠

Let us take K = 10 and a = 1.5. Therefore,

( )
3
E = 0.025(10) 5 / 3 3+ 5*0.95 = 1.06 (Eq. 50)

For Dm = 0.002 m, Eq. 54 is used to determine the value of uφ.

uφ = 113.6 * 0.002 + 0.466 = 0.69 m/s

0.00192(1.06)15 / 25 (1.5 * 0.69) 75 / 23

S= = 0.00105 (Eq. 51)
(10)15 / 46

3.494(10)13 / 23 (1.06) 20 / 23
B= = 13.20 m (Eq. 52)
(1.5 * 0.69)15 / 23

0.353(10) 25 / 69
yh = = 0.77 m (Eq. 53)
(1.5 * 0.69) 20 / 69 (1.06) 50 / 69

Now let us use S = 0.00105 and y = 0.77 m and repeat the procedure iteratively. With

B=16.52 m

y=1.31m 1
S=0.000146 2

b=11.28 m

Figure 7. Channel dimensions based on Simons and Albertson’s method

391
Sediment Transport Technology

these values very small changes are observed in the parameters which finally converged
to S = 0.001056, B = 13.26 m, and yh = 0.77 m. Let us also check the section dimensions
by assigning a trapezoidal geometry.

10
A= = 14.49 m2,
0.69

B = b + 4 y = 13.26

A = by + 2 y 2 = 14.49

Simultaneous solutions of the above equations give y = 1.38 m and b = 7.73 m. Section
dimensions of the channel are shown in Fig. 8. The bed slope and section dimensions,
obtained from Altunin’s method, are relatively different than the results of Blench, and
Simons and Albertson approaches. This example clearly illustrates that unlined channel
design requires special care. Dimensions may be checked using additional elaborate
approaches and perhaps, physical models.

The results of the example are summarized in Table 4.

As a conclusion, it can be stated that the results obtained from Blench, and Simons and
Albertson are close to each other. In practice, the final design may be achieved by
considering the local conditions.

B=13.26 m

y=1.38 m 1
S=0.001056 2

b=7.73 m

Figure 8. Channel dimensions based on Altunin’s method

Table 4. Summary of the results

b B y
Method S
(m) (m) (m)

Lacey 11.37 14.89 0.88 0.00094

Blench 11.76 16.24 1.12 0.000184
Simons and Albertson 11.28 16.52 1.31 0.000146
Altunin 7.73 13.26 1.38 0.001056

392
 Design of Stable Channels

References

1. Altunin, V. S., “Channel deformations”, Moscow, (1972).

2. Blench, T., “Regime behavior of canals and rivers”, Butherworths Scientific
Publications, London, (1957).
3. Chow, V. T., “Open Channel Hydraulics”, McGraw Hill, New York, (1959).
4. Copeland, R., “Application of channel stability methods-case studies”, Report No:
TR-HL-94-11., U.S. Army Corps of Engineers, Waterways Experiment Station,
Vicksburg, MS, (1994).
5. Garde, R. J., and Ranga Raju, K. G., “Mechanics of Sediment Transportation and
Alluvial Stream Problems”, Wiley Eastern Limited, New Delhi, (1978).
6. Lacey, G., “Stable channels in alluvium”, Min. Proc. ICE, Vol. 229, Paper No:
4736, London, (1930).
7. Lacey, G., “Sediment as a factor in the design of unlined irrigation canals”, 6th
Congress of ICID, Q20, R.12, Delhi, (1966).
8. Lindley, E. S., “Regime channels,” Proc. Punjab Engrg. Congress, Vol. 7, (1919).
9. Kennedy, R. G., “The provision of silting in irrigation canals”, Proc. ICE, Vol. 119,
Proc. No: 2826, London, (1895).
10. Simons, D. B., and Albertson, M. L., “Uniform water conveyance channels in
alluvial material”, Trans. ASCE, Vol. 128, Part 1, (1963).
11. Simons, D. B., and Şentürk, F., “Sediment Transport Technology”, Water
Resources Publications, Littleton, Colorado, (1992).

393
394
 SCOUR AT BRIDGE SITES

Prof. Dr. A. Melih Yanmaz

Civil Engineering Department
Middle East Technical University
Ankara - Turkey

1. General
Many bridges, crossing wide rivers, fail due to development of adverse effects some of
which are induced by external impacts e.g. floods, chemical spills, and earthquakes,
whereas the other factors may result from the improper design, construction,
maintenance, deterioration in time, fatigue, and operational deficiencies. This chapter is
specifically devoted to the discussion on bridge scouring. The overall scouring
mechanism at bridge sites is relatively complex because of the combined effects of
three-dimensional riverbed degradation, localized scour due to the channel constriction
at the bridge opening, local scour around bridge piers and abutments due to the
accelerated flow, and possible human interference, such as channel mining upstream of
a bridge site (Yanmaz and Çiçekdağ, 2000). Most of the parameters characterizing the
overall phenomenon are of probabilistic nature. Although the aforementioned processes
are normally interdependent to a certain extent, they are usually treated as independent
events because of the difficulty in the simulation of the overall phenomenon.

Lack of assessment of hydraulic and structural interactions can lead to the generation of
undesirable hydraulic conditions during floods, such as formation of a considerable
backwater, increased scouring potential around bridge piers and abutments, debris
accumulation, or formation of a hydraulic jump through the bridge opening (Yanmaz
and Kürkçüoğlu, 2000). Lack of relevant information and simplicity of the deterministic
models would lead to a considerable uncertainty in hydraulic design of bridges
(Yanmaz and Çiçekdağ, 2001; Yanmaz and Üstün, 2001; Yanmaz, 2001). Depending on
the severity of flow conditions and degree of erosion in close vicinity of the bridge
opening, an underdesigned bridge may be subject to a recoverable damage or complete
failure during a flood (Yanmaz and Coşkun, 1995).

Bridges might have defects due to the development of various adverse effects. In one of
the studies (Wardhana and Hadipriono, 2003), with the observation of over 500 failures
of bridges in the United States between 1989 and 2000, the most frequent causes of
bridge failures were attributed to floods and collisions. The overwhelming number of
external events, representing 83% of all principal causes triggered the bridges to fail.
Natural external events include floods, earthquakes, fires, ice, and hurricanes. Floods
represent 53% of all failures, whereas human-induced external events that constitute
20% of all failures include bridge overloads and lateral impact of land and marine
vehicles on bridges (Wardhana and Hadipriono, 2003). In a survey of 823 bridge
failures in the USA since 1950, Shirhole and Holt (1991) found that 60% of the failures
were associated with the effects of hydraulics, including both channel bed scour around

395
Sediment Transport Technology

bridge foundations and channel instability. Melville and Coleman (2000) discussed
several bridge failure events induced by hydraulic deficiencies. The current conditions
of the bridges can be evaluated with respect to safety and serviceability by performing
periodic inspection practices. With the implementation of necessary structural
modifications according to the information gathered through such studies, the existing
safety levels of the bridges would be greatly improved (Yanmaz et al., 2007).

2. The Mechanism of Scouring

Scour occurs at alluvial beds whenever the rate of sediment transport is accelerated due
to the contraction of the flow area, such as at channel constrictions, around bridge piers
and abutments, or due to impinging effect of jets leaving outlet facilities. The sediment
continuity equation for the riverbed is:

dV
= Qsi − Qso (1)
dt

in which V is the volume of the control element at the loose bed, t is the time, Qsi and
Qso are the rates of sediment transport into and out of the control volume. A localized
contraction, such as a bridge opening, would lead to an increase in the local sediment
transport capacity due to flow acceleration. Since the upstream transport rate is
normally smaller than that of the contracted section, Qso will be greater than Qsi that
would lead to a decrease in the volume of the control element. It means that the bed is
lowering with respect to time. This phenomenon is named as scouring. If it takes place
around an object like a pier or abutment, it is known as local scour around that obstacle.
If scouring takes place at a constriction having an alluvial bed, the phenomenon is
termed as the localized or contraction scour. The general or degradation scour is that
phenomenon, which occurs irrespective of the presence of any human interference, such
as implementation of various types of hydraulic structures, use of rivers, etc. The
general scour, which may take place over a geologic time scale, is influenced by the
fluvial characteristics of the river. Due to the slow process of lowering of the entire
riverbed elevation, the general scour is unlikely to be of significance to the engineer
assessing a bridge site with respect to total potential scour. Therefore, only the
mechanism of local and contraction scours will be discussed in this text.

Two types of scour develop at loose boundaries, i.e. clear water scour and live-bed
scour, depending on whether or not there is bed load transportation at the upstream. In
case of clear water scour, the flow inertia is relatively low that the bed shear stress is
smaller than the threshold value defined for the given bed material. Scour occurs in the
contracted section due to the local acceleration of the flow and continues until the scour
hole becomes stable. Rate of change of scour is a function of the erosive capacity of the
flow in the scour hole. Local scour at the downstream of a dam due to the impinging
effect of the jet is a typical example for clear water scour. When the bed shear stress
exceeds the threshold value, there is bed load transportation at the upstream, which
interferes with the local sediment transport rate in the scour hole. The rate of scour is a
function of the relative transport capacity of the approach flow and the flow in the scour
hole that is influenced by the flow intensity as well as the geometric characteristics of
the pier. For live bed scour, sediment is transported through the scour hole by means of
an excess shear stress. However, in case of clear water conditions, particles on the

396
 Scour at Bridge Sites

surface of the scour hole may be occasionally moved but are not carried away. Types of
scour are illustrated in Fig. 1. It shows the scour development around a bridge pier
against time, t, and velocity, u. In Figure 1, uc is the mean threshold velocity. Under live
bed condition, scour depth, ds, increases rapidly with time and then fluctuates about a
mean value known as equilibrium scour depth, dse, because of the continuous changes of
the bed resistance under turbulent flow conditions and random supply of sediment into
the scour hole.

2.1. Contraction scour at bridge sites

A bridge crossing a wide river is subject to two different flow conditions; namely the
flow in the main channel and in the floodplains which are only occupied by the flow
during floods. Under low flow conditions, floodplains are normally utilized for
agricultural activities. Therefore, floodplains may sometimes be spanned by long
embankments to reduce the total length of the bridge. Such an application is shown in
Fig. 2 for which the level of hydraulic and structural interaction depends on the degree
of constriction and the geometric features of the structural elements of the bridge
opening. The upstream of the bridge opening is subject to strong curvature of the flow
lines towards the main channel, where flow accelerates. Towards the edges of the
floodplains, slowly moving eddy creates ineffective flow conditions. Details of the flow
conditions in contraction and expansion zones shown in Fig. 2 are out of the scope of
this text.

The degree of contraction may be dictated by a contraction ratio, β=W1/W2, where W1

and W2 are the widths of the approach and contracted sections, respectively. A
definition sketch is given for contraction scour in Fig. 3 in which y1 and y2 are the flow
depths at the approach and contracted sections, respectively. A number of contraction
scour equations have been proposed in the literature. Two of them having practical ease
are given below. Komura (1966) proposed Eqs. 2 and 3 for clear water and live-bed
scours, respectively.

y2
= 1.6 Fr01.2 β 0.67σ g− 0.5 (2)
y1
y2
= 1.45 Fr01.2 β 0.67σ g− 0.5 (3)
y1

ds
ds
live-bed scour clear-water scour

dse dse

live bed scour

Clear water scour
t u
uc
Figure 1. Variation of scour depth with time and velocity

397
Sediment Transport Technology

Nonuniform velocity
distribution

Contraction Reversed
length, Lc eddy
Lateral
flow

L*
Expansion Embankment
length, Le Expansion
zone

Edge of main Edge of

channel floodplain

Figure 2. Flow conditions around a bridge in a wide river (Yanmaz and Caner, 2006)

y1
W1 W2 y2

dsc
loose bed
plan profile

Figure 3. Definition sketch for contraction scour

where σg is the geometric standard deviation of the bed material distribution and Fr1 is
the Froude number in the approach channel. Ignoring the velocity heads, the depth of
contraction scour, dsc, can be computed as dsc =y2-y1. Gill’s (1981) relationships for clear
water and live-bed scours are given in Eqs. 4 and 5, respectively.

−3 / 7
y2 ⎛τ ⎞
= β 6 / 7 ⎜⎜ c ⎟⎟ (4)
y1 ⎝ τ1 ⎠
−3 / 7
y2 ⎡ ⎛ τ ⎞ τc ⎤
= β 6 / 7 ⎢ β 1 / m ⎜⎜1 − c ⎟⎟ + ⎥ (5)
y1 ⎣ ⎝ τ1 ⎠ τ1 ⎦

in which τc is the critical bed shear stress, τ1 is the bed shear stress in the approach
reach, and m is the exponent, which is equal to 3 according to Einstein-Brown’s
[Brown, 1950] bed load transport equation.

398
 Scour at Bridge Sites

2.2. Local scour around bridge piers

With the presence of a bridge pier in a flow section, the net flow area reduces, which
accelerates the local flow field. Water surface elevation at the upstream face of the pier
increases due to an abrupt decrease of velocity and formation of a stagnation pressure
plane. The degree of rising of water level depends on the velocity of the approach flow,
u0, and geometric characteristics of the bridge pier. Due to a strong pressure increase at
the upstream face of the pier, the three-dimensional turbulent boundary layer separates.
Since the velocity u0 decreases from the free surface downwards, the stagnation
pressures, ρu 02 / 2 , also decrease from the surface downwards, producing a downward
pressure gradient, and hence a downward velocity component. This velocity component
interferes with the approach flow and creates the so-called horseshoe vortices at the bed
level as shown in Fig. 4. The downflow acts like a vertical jet eroding a groove in front
of the pier and the eroded material is carried around the pier by the combined action of
accelerating flow and the spiral motion of the horseshoe vortex. Breusers and Raudkivi
(1991) state that the horse-shoe vortex is a consequence of scour, not the cause of it,
although it becomes effective in transporting material away from the scour hole due to
the increased local sediment transport capacity of the flow. The horseshoe vortex
extends downstream, past of the sides of the pier, for a few pier diameters before
loosing its identity and becoming part of general turbulence. The strength of these
vortices depends on the flow Reynolds number and the geometry of the bridge pier.

Wake vortices develop as a result of shear stress gradients in the separated negative-
pressure zone around the bridge pier, which occurs at the sides and back of the pier as
shown in Fig. 4. Eroded particles from the bed by the wake vortices are transported
downstream in relation to the flow intensity and the geometric characteristics of the
pier. As the wake region behind the pier may extend far downstream from the pier, the

Pier
Bow wave

Surface roller

Downflow

Wake vortices

Scour hole

Loose bed

Horse-shoe vortices

Figure 4. Formation of vortex systems around a bridge pier (Yanmaz, 2002)

399
Sediment Transport Technology

eroded particles are carried by wake vortices in the downstream direction until the
effects of these vortices diminish. The horseshoe vortices are stronger than wake
vortices. Therefore, the maximum scour depths are normally observed at the upstream
side of the pier. A dimensional analysis is presented below to identify relevant variables
involved in the scouring phenomenon. The effects of governing variables on this
mechanism are discussed in the following sections.

2.2.1. Dimensional analysis of scouring parameters

Estimation of the scour depth at a bridge pier in an alluvial river is required for the safe
design of the bridge. Several investigators have studied the problem of local scour
around bridge piers and abutments extensively. However, no single analytically derived
equation is available because of the difficulties of the problem, such as combined
effects of complex turbulent boundary layer, time-dependent flow pattern, and sediment
transport mechanism in the scour hole [Yanmaz and Altınbilek, 1991]. A dimensional
analysis can be performed to examine the interrelationship among the dimensionless
terms, which are effective on the development of the scour hole around bridge piers and
abutments. The following functional relationship can be proposed for a vertically
mounted single smooth pier in a wide river composed of non-cohesive bed material:

f (d s , ρ ,ν , g , d 0 , u , α , u c , S 0 , ρ s , D50 , σ g , b, K s , t ) = 0 (6)

where ρ; density of water, ν; kinematic viscosity of water, g; gravitational acceleration,

d0; depth of approach flow, u; mean velocity of approach flow, α; angle of approach
flow with the pier axis, uc; mean threshold velocity, S0; bed slope, ρs; density of
sediment, D50; median sediment size, σg; geometric standard deviation of particle size
distribution, b; characteristic size of pier, Ks; factor indicating the effect of pier shape,
and t; duration of flow. For the sake of simplicity the river is considered to be prismatic
and its bed is composed of non-cohesive sediment with quartz sand. Therefore, both S0
and ∆ values are constant. The dimensionless terms can be determined by using
Buckingham’s π theorem [Shames, 1992]. After rearranging the dimensionless terms,
the following expression is obtained:

ds ⎛ u uD d u ut b ⎞
= f1 ⎜ , 50 , 0 , , , ,α ,σ g , K s ⎟ (7)
b ⎜ gd ν b u c b D50 ⎟
⎝ 0 ⎠

where u / gd 0 is the Froude number (Fr), uD/ν is the particle Reynolds number (Re),
∆=(ρs-ρ)/ρ is the relative submerged density. The term u/uc is said to be flow intensity,
which reflects the type of bed regime. For fully-developed turbulent flow with live bed
characteristics, effect of Reynolds number may be ignored. Therefore,

ds ⎛ d u ut b ⎞
= f 2 ⎜⎜ Fr , 0 , , , , α, σ g , K s ⎟⎟ (8)
b ⎝ b u c b D 50 ⎠

Effects of the governing parameters given in Eq. 8 on the scouring phenomenon are
discussed in subsequent sections.

400
 Scour at Bridge Sites

Effect of Froude number

The scale effect for pier experiments has been discussed by Ettema et al. (1998). The
Froude number may be re-defined to specifically account for the effect local energy
gradients for flow around piers. The stagnation pressure head at the upstream face of a
pier, u2/2g, can be made dimensionless by dividing it by the pier size to obtain the form
of the square of the pier Froude number, u2/gb. The study carried out by Ettema et al.
(1998) indicated that the scour depth at piers did not scale linearly with pier width
unless there was complete geometric similitude of pier, flow, and sediment particles
forming the bed. Similitude of scour depth data from the flume to the field requires the
constancy of Fr, u/uc, and d0/b. Flume experiments, however, are normally forced to use
bed materials, which have comparable sizes as the prototype since very fine particles in
a laboratory medium would be subject to suspension. Therefore, to maintain similar
modes of sediment transportation, the term reflecting the effect of sediment coarseness,
b/D50, cannot satisfy the length scale selected according to the other geometric
variables, such as pier width, flow depth, etc. So, the constancy of b/D50 cannot be
maintained.

To attain a desired particle mobility level, the value of u/uc should be the same in the
laboratory and the field. This implies that greater values of u than required in Froude
number simulation are needed in a laboratory medium. Hence, the Froude number used
in laboratory experiments may be greater than that for the corresponding field
conditions. On the other hand, as can be seen in Fig. 5, relative scour depth, ds/b,
increases with increasing Froude number in the range of the available data. However, to

2.0

1.8

1.6

1.4

1.2
d s /b

1.0

0.8

0.6

0.4
Chabert and Engeldinger (1956)
Laursen (1963)
0.2 Hancu (1971)
Basak et al. (1977)

0.0
0.00 0.20 0.40 0.60 0.80 1.00
Fr

Figure 5. Variation of relative scour depth against pier Froude number

401
Sediment Transport Technology

formulate the effect of the Froude number, a wider range of data is needed. Therefore,
clear water conditions in a laboratory medium can cause deeper scour holes, relative to
the pier width, than any likely to occur at bridge piers in rivers. As a concluding remark,
scour-prediction equations derived using laboratory data overestimate the scour depths,
and hence can be assumed to give conservative values for the design.

Effect of approach flow depth

Implementation of a pier into a flow section would lead to development of a surface
roller, which has an opposite sense of rotation as the horse-shoe vortex at the bed level.
For shallow flows, the surface roller tends to retard the downflow. Therefore, it results
in reduction of scour depth. Melville and Coleman (2000) classify the piers as narrow,
intermediate width, and wide with respect to d0/b ratio using the following limiting
values. For deep flows and narrow piers (d0/b>1.43), the scour depth is independent of
flow depth but depends on b. Conversely, for shallow flows with wider piers (d0/b≤0.2),
the scour depth is independent of b but varies linearly with d0. For intermediate width
piers, i.e. 0.2<d0/b≤1.43, the scour depth is proportional to the square root of the
product of d0 and b. When the available clear water data are plotted as relative scour
depth, ds/b, versus relative approach flow depth, d0/b, it is observed that the design
curves suggested by Melville and Coleman (2000) bound the available clear water scour
data (see Fig. 6). Moreover, the studies of Melville and Coleman (2000) show that the
live bed data are also bounded by these limiting curves.

Effect of flow intensity

For clear water conditions, the depth of scour increases considerably with increasing
velocity up to the threshold conditions at which a maximum value is attained. Review of

10.00
Chabert and Engeldinger (1956)
Tarapore (1962)
Laursen (1963)
Shen et al. (1966)
Hancu (1971)
White (1975)
Başak et al. (1977)
Jain and Fischer (1979)
Melville (1984)
Yanmaz (1989)
Kothyari et al. (1992)
Ettema (1980)
Wide pier
Intermediate width pier
ds/b

1.00 Narrow pier

0.10
0 0 1 10
d0/b

Figure 6. Variation of relative scour depth against relative approach flow depth (Yanmaz, 2002)

402
 Scour at Bridge Sites

the related literature reveals that the scour depth decreases from its maximum value at
the threshold to a minimum at about u/uc ≈ 1.5-2.0, where uc is the critical mean velocity
for threshold conditions. At this value of u/uc, the bed generally has steepest features.
Depth of scour increases again with further increase in the u/uc ratio up to a new peak
value at the transition flat bed condition. Over the transition flat bed, the form drag
component of energy loss is absent and a great amount of flow energy is used for
sediment transport and scouring. At still higher relative velocities under upper flow
regime, bed forms appear again and dissipate some of the flow energy. The scour depth
appears to decrease slightly. However, there are limited data for this range.

Effect of sediment grading

Laboratory experiments performed by Raudkivi and Ettema (1983) showed that
sediment grading has a strong influence on the equilibrium depth of clear water scour.
As the geometric standard deviation of the grain size distribution increases, the larger
grains form and exert an armoring effect to the underlying sediment, and hence the
depth of scour decreases. An adjustment coefficient, Kσ, relative to uniform sediment is
introduced by Raudkivi (1986). The author converted Raudkivi’s (1986) graphical
relationship to the following best-fit equation:

K σ = 0.013462σ g5 − 0.2232σ g4 + 1.4066σ g3 − 4.0864σ g2

+ 5.007σ g − 1.1179 (1.0 ≤ σg ≤ 5.0) (9)

For a typical gravel-river with σg=3.5, the expected scour depth is only 20% of the
scour depth that is likely to occur under uniform sediment case. The effect of sediment
grading on the depth of live-bed scour is more complex. However, there is very limited
information on this topic and is of interest for further research.

Effect of pier and sediment sizes

Effect of relative sizes of pier width and sediment size, b/D50, on the development of
scour has been examined by Ettema (1980). His results indicate that the grains are large
compared to the groove excavated by the downflow for the range of b/D50<25, and the
erosion process is impeded. For coarse sediment, the grains are large relative to the
width of the groove excavated by the downflow and the bed is fairly porous. A
significant fraction of the downflow penetrates the coarse bed material and dissipates its
energy [Breusers and Raudkivi, 1991]. The relative scour depth is observed to be
independent of relative pier to sediment size for b/D50≥25. An adjustment coefficient,
Kd, is introduced by Melville and Coleman (2000)

⎛ b ⎞
K d = 0.57 log⎜⎜ 2.24 ⎟⎟ (for b/D50<25) (10)
⎝ D50 ⎠

Effect of time
The equilibrium scour depth under clear water conditions is reached in quite long flow
duration due to the fact that a case should exist during which the combined effects of
the temporal mean shear stress, the weight component, and the turbulent agitation are in

403
Sediment Transport Technology

equilibrium in the scour hole. According to Raudkivi (1986) and Melville and Coleman
(2000), the duration required to reach equilibrium conditions may be of the order of
several days in a laboratory medium. The corresponding flood duration in the prototype
may consequently be beyond reasonable limits. Therefore, the design of bridge pier
foundations on the basis of equilibrium scour depths under clear water conditions may
yield considerably greater values than may occur if the flow is of short duration. For a
known time to peak value of the design flood hydrograph, smaller scour depths may be
obtained, which may reduce the total cost of construction (Yanmaz, 2006). For long
flow durations under live-bed conditions, effect of time can be ignored (see Fig. 1).

Effect of Pier Shape and Alignment

Piers having smooth geometric shapes should be placed in water flow in such a way that
the pier body is almost aligned with the flow direction to reduce the projected width of
the pier. This practice would lead reduction of the effective area for vortex generation
and also reduction in the drag force exerted by the flow. For piers aligned with the flow,
the shape adjustment factor, Ks, relative to the cylindrical pier is presented in Table 1
for commonly used pier geometries. In this table, Lb is the length of the pier and b is the
width of the pier normal to the flow direction. As can be seen from Table 1, a square
pier causes 22% greater scour depth compared to a cylindrical pier because of a
pronounced separation at the upstream which would lead to generation of severe
vortices, and hence greater depth of scour. The scour depth at the pier is also dependent
on the slope of the upstream face of the pier to the vertical. The scour depth increases
for an upwards-broadening pier and decreases for an upwards-narrowing pier. The rate
of scour is further influenced by the foundation details, i.e. the presence of a footing, a
caisson, or capping blocks on piles. Details of this topic will not be covered in this text.
Further discussion on this topic can be found in Yanmaz (2002).

The depth of local scour for all shapes, except a cylindrical pier, is influenced by the
alignment of the pier relative to the flow direction. The depth of scour is a function of
the projected width of the pier, i.e. the width normal to the flow. A long pier, which is
mounted obliquely to the flow direction, is definitely subject to a greater scour since a
larger area is subject to vortex development. If the angle between the pier axis and the
approach flow is α as shown in Fig. 7, a projected pier width, bt = bcosα+Lpsinα, should
be used in scour computations, or the pier width b is multiplied by an alignment
adjustment factor, Kα, which is given below (Richardson and Davis, 2001):

Table 1. Shape adjustment factor for vertically mounted piers

(Breusers and Raudkivi, 1991)

Pier shape b/Lp Ks

Cylindrical - 1.0
Square 1:1 1.22
Rectangular 1:3 1.08
1:5 0.99
Rectangular with semi-circular nose 1:3 0.90
Semi-circular nose with wedge shaped tail 1:5 0.86

404
 Scour at Bridge Sites

Lp b

bt
Flow
α

Figure 7. Definition sketch for incoming flow approaching the pier axis with an angle

0.65
⎛ Lp ⎞
K α = ⎜⎜ b cos α + sin α ⎟⎟ (11)
⎝ b ⎠

Solution of Eq. 11 is illustrated in Fig. 8. As an example, it is observed from Fig. 7 that

the depth of scour around a pier having Lp = 6b with α=30° is more than twice of the
value that could be obtained under the case of α=0°. That is why some training facilities
should be implemented in the close upstream of bridges, especially at the bends, to
provide an approach of the flow with approximately zero degree with the pier axis.

Effect of pile groups

In the literature, there is limited information on the effect of pile groups on the scouring.
Breusers and Raudkivi (1991) interpreted the experimental results of Hannah (1978)
who performed experiments under clear water conditions with D50=0.75 mm and
σg=1.32. For two piles, which were aligned with the flow direction (α=0°), the scour
depth at the front pile was the same as for a single pile ds under a/b=1 where a is the
spacing between the center lines of the piles and b is the common pile diameter. With
increasing separation (a/b>1), a maximum value was experienced at a/b=2.5 due to the

Lp/b
6
12
Angle adjustment factor

5 10

8
4
6

3
4

2
2

1
0 15 30 45 60 75 90
α (°)

Figure 8. Variation of angle adjustment factor, Kα, with respect to α.

405
Sediment Transport Technology

reinforcing effect, which is the increase in the scour depth due to the interference of
horseshoe and wake vortices. This causes the overlapping of scour holes around closely
spaced piles and results in a lowering of the bed behind the front pile. Hence it leads to
transportation of more material to the downstream. The reinforcing effect was
maintained until about a/b=11 for the aforementioned experimental conditions. For
larger spacing, the scour depth was the same as for a single pile and separate scour holes
formed around bridge piers.

2.3. Scour-prediction öethods for bridge piers

Many scour prediction equations have been reported in the literature, which can be
found in Yanmaz (2002). However, the results of the proposed methods, which were
mainly derived from laboratory investigations, differ widely from each other. Each
method is normally assumed to be valid for the range of its experimental conditions.
Field data of scour are very limited because of the observational difficulties during
floods. This leads to the problem of model calibration. That is why there is no single
method, which is applicable to universal design conditions. The following methods will
only be given because of their practical importance. In these equations d0 and Fr are the
approach flow depth and flow Froude number directly upstream of the pier. The results
obtained from the following equations can be modified by multiplying them with the
adjustment factors Kσ, Kd, Ks, and Kα whenever required.

2.3.1. Johnson (1992) method

Johnson (1992) proposed Eq. 12 for live bed scour around a single cylindrical pier:

0.02
ds ⎛d ⎞
= 2.02⎜ 0 ⎟ Fr0.21σ g−0.24 (12)
b ⎝ b ⎠

2.3.2. Richardson and Davis (2001) method

This method is widely used in the USA and is also referred to as the Colorado State
University equation or HEC-18 procedure. The maximum depth of scour around a
single pier is given by:

0.35
ds ⎛d ⎞
= 2.0 K s K α K b K z ⎜ 0 ⎟ Fr0.43 (13)
b ⎝ b ⎠

where Kb is a factor to account for bed condition, which can be taken as 1.1 for clear
water scour, plane bed and small dunes, and 1.2 and 1.3 for medium and large dunes,
respectively. The adjustment factor Kz in Eq. 13 accounts for the armoring effect of the
bed material. For D50<2 mm or D95<20 mm, the value of Kz can be taken as unity. For
coarser materials, i.e. D50≥2 mm and D95≥20 mm, the value of Kz attain smaller values
than unity with its minimum value of 0.4.

2.3.3. Melville and Coleman (2000) method

Melville and Coleman (2000) propose the following equation composed of several
multiplicative adjustment factors for general scour at single bridge piers:

406
 Scour at Bridge Sites

d s = K yb K I K d K s K α K t (14)

where Kyb is a factor accounting for the combined effects of flow depth and pier size, KI
is the flow intensity factor, which is unity for live bed conditions and is u/uc for clear
water scour in uniform bed, Kd is an adjustment factor for the relative effect of pier and
sediment sizes, which can be determined from Eq. 10 for b/D50<25, Ks is pier shape
factor (1.0 for cylindrical piers and rectangular piers with rounded noses, and 1.1 for
square piers), and Kα is a factor to account for the effect of angle of approach flow
which can be determined from Eq. 11 for non-cylindrical piers. The adjustment factors
Kyb can be determined from 2.4b, 2(d0b)0.5 and 4.5d0 for d0/b>1.43, 0.2≤d0/b≤1.43, and
d0/b<0.2, respectively. Melville and Coleman (2000) propose adjustment factor, Kt, for
time effect relative to the time achieved during equilibrium scouring situations.
However, this factor is subject to an unknown level of uncertainty since it was not
verified by other researchers.

2.3.4. Oliveto and Hager (2002) method

Oliveto and Hager (2002) proposed the following equation for clear water scour at
cylindrical bridge piers:

ds
= 0.068σ g−0.5 Fd1.5 log Td (15)
LR

where LR = b 2 / 3 d 01 / 3 , Fd = u/(∆gD50)0.5, Td = t/tR, and tR = LR/(∆gD50)0.5. This equation is

independent of equilibrium scouring parameters. Therefore, it is of practical importance.

2.3.5. Yanmaz’s (2006) semi-empirical model

Yanmaz (2006) modified his previous semi-empirical model presented in Yanmaz and
Altınbilek (2001). This model is based on the application of the sediment continuity
equation to the scour hole around a single cylindrical pier under clear water conditions.
This equation is:

dV
= Qso − Qsi (16)
dt

where V; volume of scour hole around the pier, Qso; volumetric rate of sediment
transport from the scour hole, and Qsi; volumetric rate of sediment transport into the
scour hole. For clear water conditions, Qsi = 0. Based on measurements of the scour
contours, Yanmaz (1989) proposed the following expression for the volume of the scour
hole around a single cylindrical pier:

π ⎛ d s3 3d s2 b ⎞
V= ⎜ + ⎟ (17)
3 tan φ ⎜⎝ tan φ 2 ⎟⎠

where φ is the angle of repose of the bed material. Yanmaz (2006) proposed the
following form to formulate the rate of sediment transport from the scour hole:

407
Sediment Transport Technology

EA
Qso = λ (18)
∆ρ s

where λ; coefficient of proportionality to account for the scour hole geometry, sediment
and flow properties, E; sediment pickup rate in mass per unit time and area, A; unit area
from which sediment is picked up, and ∆; relative density. A recent sediment pickup
function valid for sloping beds proposed by Dey and Debnath (2001) is used in the
formulation of Qso. This equation is as follows:

E = 0.0006TD*0.24σ 1g.9 ρ s ∆gD50 (19)

where T; transport-stage parameter due to scouring and D* = D50(∆g/υ2)1/3. The unit area
for sediment pickup is taken as A = 2dscotφ+b. Using experimental scour depths and
dimensional analysis of relevant parameters:

−0.95 0.37
⎛u t ⎞ ⎛ ds ⎞
λ = 2000⎜ * ⎟ ⎜⎜ ⎟⎟ (20)
⎝ b ⎠ ⎝ b tan φ ⎠

Combining Eqs. 16, 17, 18, and 20, one obtains:

−0.95 0.37
⎛u t ⎞ ⎛ ds ⎞ ⎛ 2d s ⎞
1.2⎜ * ⎟ ⎜⎜ ⎟⎟ TD*0.24σ 1g.9 D50 ∆gD50 ⎜⎜ + b ⎟⎟
d (d s )
=
⎝ b ⎠ ⎝ b tan φ ⎠ ⎝ tan φ ⎠ (21)
dt π∆ ⎛ d s 2
⎞
⎜ + bd ⎟
tan φ ⎜⎝ tan φ ⎟
s
⎠

which is a first order, nonlinear differential equation. Numerical solution of this

equation agreed well with the experimental data. This equation needs to converted to
dimensionless form for practical ease by taking the following dimensionless terms
S=ds/b and Ts = tD50(∆gD50)0.5/b2. Therefore, one obtains:

dS αS 0.37 (2S cot φ + 1)

= (22)
dTs Ts0.95 ( S 2 cot φ + S )

where

−0.95
⎛ u*b ⎞
α = 0.231(tan φ ) 0.63 ⎜ ⎟ TD*0.24σ 1g.9 (23)
⎜ D ∆gD ⎟
⎝ 50 50 ⎠

For a particular combination of flow, pier, and sediment properties, the value of α is
constant. Variation of S against Ts is obtained by numerical solution of Eq. 22. Fig. 9
shows this information for the experimental range of 0.036 ≤ α ≤ 0.146. Using the input
data, the value of α is computed. With the known time to peak value of the design flood,
Ts is computed and the corresponding S is determined from Fig. 9.

408
 Scour at Bridge Sites

2.0

1.5

1.0
S

α=0.06
0.5 α=0.08
α=0.10
α=0.036
α=0.146

0.0
0 200 400 600 800 1000 1200
Ts

Figure 9. Variation of S with respect to Ts in Yanmaz (2006) model

3. Local Scour around Abutments

The protrusion of a bridge abutment into the channel creates an obstruction to the flow
in an alluvial river and causes acceleration and separation of flow at the upstream face
of the abutment. Common shapes of abutments can be classified as vertical wall, wing
wall, and spill through abutments. Selection of a particular type is governed by local
geotechnical conditions as well as structural requirements. However, a spill through
abutment is preferable from hydraulics point of view since it produces almost half the
scour depth that occurs around a vertical wall abutment under the same hydraulic
conditions and bed material characteristics. At bridge abutments along a continuous
wall, the downflow is weakened because of retarded flow adjacent to the wall aligned
with the flow direction. The maximum spiraling vortex motion, also called as principal
vortex as shown in Fig. 10, forms at the upstream face of the abutment as a result of
interaction of the downflow and the approach flow. Intercepting overland flow, which is
forced into the channel at the abutment, can also pronounce scour at abutments.

Scouring phenomenon at bridge abutments is very similar to that of pier scouring. That
is why details of this topic are omitted in this text. Only Froehlich’s (1989) live-bed
equation is given.

0.57
d sa ⎛ yf ⎞
= 2.27 K s K θ ⎜⎜ ⎟⎟ Fr0.61 + 1.0 (24)
La ⎝ La ⎠

where d sa ; maximum depth of scour around the abutment, L a ; length of abutment

409
Sediment Transport Technology

Surface roller

Downflow Wake vortices

Scour hole

Loose bed Principal vortex

Figure 10. Flow pattern around a wing-wall abutment (Yanmaz, 2002)

perpendicular to the flow direction, yf and Fr are the average depth and Froude number
of the approach flow in the floodplain, Ks is abutment shape factor (1.0 for vertical wall
abutment, 0.82 for wing wall abutment, and 0.55 for spill-through abutment), Kθ;
adjustment factor for the effect of attacking angle, θ, with the abutment axis
(Kθ=(θ/90)0.13, θ in degrees).

4. Scour Countermeasures
Methods to protect bridges from scour and channel instability are termed as scour
countermeasures that are normally installed during the construction of bridges.
However, countermeasures can also be implemented at existing crossings to increase
the safety. There exist various types of countermeasures, which are mainly of hydraulic,
structural, and non-structural types.

Hydraulic countermeasures comprise river training structures and armoring

countermeasures. Structural countermeasures are composed of foundation
strengthening, pier, and abutment geometry modification. Non-structural measures
mostly refer to scour monitoring that can be performed by fixed or portable
instrumentation and visual monitoring. Design guidelines for these countermeasures are
given in HEC-23 by Lagasse et al. (1997). A brief description of various types of
hydraulic and structural countermeasures is presented in Table 2. Details of this topic
can be found in Yanmaz (2002). Guidelines for the selection of a proper river bridge
site can be found in Yanmaz (1994) and Yanmaz (2002).

410
 Scour at Bridge Sites

Table 2. Description of various bridge scour countermeasures [Melville and Coleman, 2000]

Scour type Countermeasure Example Purpose

Armoring devices Riprap, gabions, cable- Prevention of erosion to the
tied blocks, tetrapods, channel bank in the vicinity
precast concrete blocks, of the bridge; stabilization of
used tires, vegetation the channel alignment
planting
Retards Timber piles, sheet piles, Reduction of flow velocity
Lateral Jack or tetrahedron fields, near channel bank and
erosion vegetation planting inducement of deposition of
sediment
Groynes Groynes, spurs Reduction of flow velocity
near channel bank and
inducement of deposition of
sediment; stabilization of the
channel alignment
Check dams --- Control of channel grade
Channel lining Concrete or bituminous Control of channel
Degradation concrete pavement degradation
Bridge modification Increase of bridge To retard contraction scour
opening width
Channel improvement Dredging, clearing of Increased sediment transport
channel, formation of a to reduce sediment deposition
cut-off at bridge crossing
Aggradation Controlled mining Sediment withdrawal Reduction in sediment input
from channel bed at bridge site
Debris basin Reduction in sediment input
at bridge site
Armoring devices Riprap, gabions, cable- Reduced local scour
tied blocks, etc.
Flow altering devices Sacrificial piles, deflector Reduced local scour at piers
vanes by implementing piles or thin
plates placed upstream of
bridge piers which deflect
flow away from the piers and
induce deposition in the scour
hole at the pier
Local scour Underpinning of bridge --- Increased safety against local
piers scour at piers by extending
the length of piers to a lower
foundation level
Guide banks --- Improved flow alignment at
bridge crossing by extending
side embankments to the flow
section; reduction in local
scour at abutments

Rereferences

1. Başak, V., Başlamışlı, Y., and Ergun, O., “Local scour depths around circular pier
groups aligned with the flow”, State Hydraulic Works, Report No. 641,Ankara,
Turkey (in Turkish), (1977).

411
Sediment Transport Technology

2. Breusers, H. N. C., and Raudkivi, A. J., “Scouring-Hydraulic Structures Design

Manual 2”, International Association for Hydraulic Research, A. A. Bakema,
Rotterdam/Brookfield, (1991).
3. Brown, C. B., “Sediment Transportation”, In: H. Rouse (Ed.) Engineering
Hydraulics, Chapter XII, John Wiley, New York, 769-857, (1950).
4. Chabert, J., and Engeldinger, P., “Etude des affouillements autour des piles des
ponts”, Laboratoire National d’Hydraulique, Chatou, France, (in French), (1956).
5. Dey, S., and Debnath, K., “Sediment pickup on streamwise sloping beds”, Journal
of Irrigation and Drainage Engineering, ASCE, 127(1), 39-43, (2001).
6. Ettema, R. B., Melville W., and Barkdoll, B., “Scale effect in pier-scour
experiments”, Journal of Hydraulic Engineering, ASCE, Vol. 124, No. 6, pp. 639-
642, (1998).
7. Ettema, R., “Scour at bridge piers”, Ph.D. Thesis, University of Auckland, New
Zealand, (1980).
8. Froehlich, D.C., “Local Scour at Bridge Abutments”, Proceedings of ASCE
National Hydraulics Conference, Colorado Springs, Colorado, USA, 13-18,
(1989).
9. Gill, M. A., “Bed Erosion in Rectangular Long Contraction”, Journal of the
Hydraulics Division, ASCE, 107(HY3), 273-294, (1981).
10. Hancu, S., “Sur le calcul des affouillements locaux dans la zone des piles de ponts”,
Proceedings 14th IAHR Congress, 3, 299-313 (in French), (1971).
11. Hannah, C. R., “Scour at pile groups,” Report No: 78-3, Department of Civil
Engineering, University of Canterbury, Christchurch, New Zealand, (1978).
12. Jain, S. C., and Fischer, E. E., “Scour around bridge piers at high Froude numbers”,
Report Number F.H.W.A.R.D.:79-104, Federal Highway Administration,
Washington D.C., (1979).
13. Johnson, P. A., “Reliability-based pier scour engineering,” Journal of Hydraulic
Engineering, ASCE, 118(10), 1344-1358, (1992).
14. Komura, S., “Equilibrium depth of scour in long constrictions”, Journal of the
Hydraulics Division, ASCE, 92(HY5), (1966).
15. Kothyari, U. C., Garde, R. J., and Ranga Raju, K. G., “Live bed scour around
cylindrical bridge piers”, Journal of Hydraulic Research, 30(5), 701-716, (1992).
16. Lagasse, P. F., Byars, M. S., Zevenbergen, L. W., and Clopper, P. E., “Bridge scour
and stream instability countermeasures-experience, selection, and design guidance,”
Hydraulic Engineering Circular 23, FHWA HI 97-030, Washington, D.C., (1997).
17. Laursen, E. M., “An analysis of relief bridge scour”, Journal of Hydraulics
Division, ASCE, 89(HY3), 93-118, (1963).
18. Melville, B. W., “Live bed scour at bridge piers”, Journal of Hydraulic
Engineering, ASCE, 110(9), 1234-1247, (1984).
19. Melville, B. W., and Coleman, S. E., “Bridge Scour”, Water Resources
Publications, Highlands Ranch, Colorado, (2000).
20. Oliveto, G., and Hager, W. H., “Temporal evolution of clear water pier and
abutment scour”, Journal of Hydraulic Engineering, ASCE, 128(9):811-820,
(2002).
21. Raudkivi, A. J., and Ettema, R., “Clear-water scour at cylindrical piers”, Journal of
Hydraulic Engineering, ASCE, 109(HY3), 338-350, (1983).
22. Raudkivi, A. J., “Functional trends of scour at bridge piers”, Journal of Hydraulic
Engineering, ASCE, 112(1), 1-13, (1986).

412
 Scour at Bridge Sites

23. Richardson, E. V., and Davis, S. R., “Evaluating scour at bridges.” Hydraulic
Engineering Circular No:18, Report No: FHWA NHI 01-001, FHWA, US. Dept.
of Transportation, Washington, D.C., (2001).
24. Shames, I. H., “Mechanics of fluids”, McGraw Hill, Singapore, (1992).
25. Shen, H. W., Schneider, V. R., and Karaki, S., “Mechanics of local scour, data
supplement”, Colorado State University, CER 66-67 HWS-VRS-SK-27, prepared
for U.S. Department of Commerce, Bureau of Public Roads, Office of Research and
Development, Structures and Applied Mechanics Division under Contract No. CPR
11-8022, (1966).
26. Tarapore, Z. S., “A theoretical and experimental determination of the erosion
patterns caused by obstructions in an alluvial channel with particular reference to a
vertical cylindrical pier”, Ph.D. Thesis, University of Minnesota, (1962).
27. Shirhole, A. M., and Holt, R. C., “Planning for a Comprehensive Bridge Safety
Program”, Transportation Research Record 1290, Vol. 1, Transportation Research
Board, National Research Council, Washington, D.C., 39-50, (1991).
28. Wardhana, K., and Hadipriono, F. C., “Analysis of Recent Bridge Failures in the
United States”, J. Perf. of Const. Fac., ASCE, 17 (3), 144-150, (2003).
29. White, W. R., “Scour around bridge piers in steep streams”, Proceedings 16th
IAHR Congress, Sao Paulo, 2, 279-284, (1975).
30. Yanmaz, A. M., “Time-dependent analysis of clear water scour around bridge
piers”, Ph.D. thesis, Civil Engineering Department, Middle East Technical
University, Ankara, Turkey, (1989).
31. Yanmaz, A. M., and Altınbilek, H. D., “Study of time-dependent local scour around
bridge piers”, Journal of Hydraulic Engineering, ASCE, 117(10), 1247-1268,
(1991).
32. Yanmaz, A. M., “Flood interaction with river crossings: a case study”, Proc. NATO
ASI Series, Series E, Applied Sciences, 257, 565-569, (1994).
33. Yanmaz, A. M., and Coşkun, F., “Hydrological aspects of bridge design: case
study”, Journal of Irrigation and Drainage Engineering, ASCE, 121(6), 411-418,
(1995).
34. Yanmaz, A. M., and Çiçekdağ, Ö., “Channel Mining Induced Stream Bed
Instability Around Bridges”, CD-ROM Proceedings of Watershed Management
Conference, ASCE, (2000).
35. Yanmaz, A. M., and Kürkçüoğlu, S., “Assessment of Hydraulic and Structural
Interactions for Bridges,” CD-ROM Proc. 4th International Conference on
Hydroscience and Engineering, Seoul, Korea, (2000).
36. Yanmaz, A. M., “Uncertainty of local scouring parameters around bridge piers,”
Turkish Journal of Engineering and Environmental Sciences, 25(2), 127-137,
(2001).
37. Yanmaz, A. M. and Çiçekdağ, Ö., “Composite reliability model for local scour
around cylindrical bridge piers”, Canadian Journal of Civil Engineering, Vol. 28,
No. 3, 520-535, (2001).
38. Yanmaz, A. M., and Üstün, İ., “Generalized Reliability Model for Local Scour
Around Bridge Piers of Various Shapes”, Turkish Journal of Engineering and
Environmental Sciences, Turkish Scientific and Technical Research Council, Vol.
25, No:6, 687-698, (2001).
39. Yanmaz, A. M., “Köprü Hidroliği”, METU Press Publishing Company, Ankara,
(in Turkish), (2002).

413
Sediment Transport Technology

40. Yanmaz, A. M., and Caner, A., “Hydraulic-structure interaction for bridges with
special emphasis to effect of type of infrastructural elements”, CD-ROM
Proceedings of 7th International Congress on Advances in Civil Engineering,
Istanbul, (2006).
41. Yanmaz, A. M., “Temporal variation of clear water scour at cylindrical bridge
piers”, Canadian Journal of Civil Engineering, Vol. 33, No: 8, 1098-1102, (2006).
42. Yanmaz, A. M., Caner, A., and Berk, A., “Renovation of a Safety-Inspection
Methodology for River Bridges”, Journal of Performance of Constructed
Facilities, ASCE, (in press), (2007).

414
 SEDIMENT CONTROL STRUCTURES
FOR HEADWORKS AND INTAKES

Prof. Dr. Nevzat YILDIRIM

Department of Civil Engineering
Gazi University
06570 Maltepe Ankara - Turkey

1. Introduction
Water is taken from streams for different usage purposes i.e. water supply (domestic
usage), irrigation and energy. In order to divert and take water from a stream, it is
generally necessary to construct a water diverting structure which is called the intake.
As in the cases of other hydraulic structures, an intake structure also undergoes the
following three basic design stages:
a) Hydrological design
b) Hydraulic design
c) Structural design

The structural design is out of the scope of this course.

The main hydrological design criterion requires that the water source (the stream flow)
must guarantee and comfortably meet the demand (diverted water) during the time
period of diversion. Also, the information related to the maximum and minimum stages
and maximum flood characteristics of the stream is needed.

This course mainly deals with the hydraulic design of the intakes. The hydraulic design
means that the intake and the affiliated structures must be designed in such a way that
they behave hydraulically efficient. The hydraulic efficiency of an intake in a stream
along with other functions, basically involves the diverted water discharge capacity and
sediment content in the diverted water from both the amount and size point of views.
Since the design criterion of the diverted water discharge capacity is the concern of the
well-known classical hydraulic engineering, it is not the theme of this course. The main
theme of this course is the design principles of an intake from the sediment content of
the diverted water point of view.

Generally, sediment in a stream is transported in the following three positions (Fig. 1):
a) Sediment transported in suspension (suspended load)
b) Sediment transported at the bed by rolling and sliding (bed load)
c) Sediment transported in saltation moving by jumping at and close to the bed.
Since the saltation occurs across a small height close to the bed, the sediment
transported in saltation is merged into the bed load from the intake design point
of view in this course. In other words, the sediment transported in a stream is
considered to be consisting of suspended load and bed load.

415
Sediment Transport Technology

Figure 1. Sediment transport in a stream

As shown in Fig. 1, small size and light sediment particles make the suspended load.
The sediment particles go into suspension due to turbulence and buoyancy forces. When
the vertical velocity component of the turbulent stream flow is sufficiently large
(strongly turbulent), it introduces a vertical lift force along with buoyancy force to the
particle. Thus, the sediment particle goes into suspension. The sediment particles
moving at the bed (bed load particles) are larger and heavier than those of the suspended
load.

The sediment transported by a stream comes from the eroded bed of the stream and the
land soil of the watershed (drainage-area) of the stream. The characteristics of the
sediment transported by a stream depend on several factors i.e. type of the soil,
topography, vegetal cover, agricultural activities, regime of the stream flow, type and
duration of precipitation etc. Generally, the suspended load is about 80% and the bed
load is about 20% of the total sediment load of a common stream. Depending on the
characteristics of the stream and its drainage area, the percents of these loads may
change. For example; for a mountain stream (high slope and high velocity) both the
amount and size of the bed load sediment can increase and even be larger than the
suspended load. During a large flood, a stream may carry sediment much larger than
that it carries in a whole year with its normal regime. As it is known, the sediment
transport of a stream highly depends on its flow depth and slope. Since the plain stream
has a small slope, most of the sediment is transported as suspended fine sediment load
(80-90%) whose particle size distribution is close to the uniform. It is well known that
larger the amount of sediment content, (especially suspended sediment), higher is the
viscosity of the stream-flow. The viscosity of water having sediment concentration is
given as [Raudkivi, 1967];

µ susp = µ (1 + k1 C ) (1)

in which; µ; viscosity of sediment free (clear) water, k1; a constant, C; sediment

concentration (volume based), and µsusp; viscosity of fluid with sediment concentration.
The sediment concentration also affects the fall (settling) velocity of the sediment
particle. The fall velocity is given as [Raudkivi, 1967]

w = w0 (1 - C ) β (2)

416
 Sediment Control Structures for Headworks and Intakes

in which w; fall velocity of the particle in water with sediment concentration, w0; fall
velocity of the particle in clear-still water and β is a coefficient. It is clear that higher the
concentration, lower is the fall velocity. No doubt that sediment concentration (hence
velocity) increases the shear force hence the drag force acting on the sediment particle.
Along with other factors, this is one of the main reasons why during a flood a stream
caries so much sediment.

Beverage and Culbertson, 1964 have defined the flows with concentrations higher than
20, 40, 60 and 80 % as high, extreme and hyper-concentrated and mud flows
accordingly. As the concentration of the sediment (especially the cohesive sediment)
considerably increases, the stream fluid flow loses its Newtonian-identity and it behaves
as non-Newtonian fluid which is called the rheological (mud) flow. It is difficult to deal
with the analytical analysis of this type rheological (mud) flow.

Drainage lands with high slopes, less vegetal cover and soils of fine particles (i.e. clay)
supply much more suspended sediment with high concentration. A stream flow with
high concentration is able to erode and make changes in its bed forms. In the cases of
high slope streams, the relationship between the stream flow discharge and sediment
load vary from flood to flood.

The reason for this is that during the flood, the high-slope streams (i.e. mountain
streams) carry large boulders, stones and gravels which cause highly non-uniform flow.
As the flow recedes, they settle at the bed, form a kind of armoring and prevent the
erosion of the finer bed material. During the next larger flood, these gravels, boulders
and stones are again displaced and moved. Thus as the armor coat is removed, the finer
bed material is again eroded.

During this inevitable cycle, the friction factor, critical shear stress and drag forces on
the sediment particles and the transported sediment size distribution abruptly change.
This is the reason why the commonly know approximate sediment transport and
hydraulic formulae don’t give satisfactory results for the mountain streams while they
give reasonable results for the plain streams. The slope of the stream plays a very
important role on the bed and bank stability of the stream and the characteristics of the
sediment it carries. Çeçen (1995), has classified the rivers according to their slopes as
following;

1. Rivers with stable beds:

a) Slope is constant for a long period of time
b) Very little change of cross-section, practically no change in the bed elevation
c) Size distribution of bed material does not vary in time
2. Rivers in aggradations:
a) Sand bars in the cross-section
b) Trend of development of new bends
c) Aggradation on banks
d) Levees to be heightened in short intervals
e) Increase in ground water level
3. Rivers in degradation:
a) Sand bars moving downstream

417
Sediment Transport Technology

b) Erosion of banks
c) Continuous protection of levees necessary
d) Scour around bridge piers
e) Decrease in ground water level
4. Rivers in latent erosion:
a) River bed armored by coarse gravel and rocks
b) Slope and bed profile constant for a long period of time
c) No sand bars, fine sediment in motion
d) Large amount of sediment transport of various sizes during floods

2. How to Take (Divert) Water From Sediment Carrying Streams?

In hydraulic engineering it is well known that the sediment is the biggest enemy of the
hydraulic structures constructed on the sediment carrying streams. The sediment
problem must be resolved in large extend at the very point where the water is diverted
from the stream. For a given stream flow condition, the amount and size of the sediment
in the diverted water highly depend on the orientation of the intake and method of the
diversion.

A good diversion method and a well designed intake can considerably eliminate the
sediment-related problems at the very beginning of the supply line and affiliated
structures; otherwise the operation and maintenance of these systems will be very
expensive and uneconomical (whole water supply system can be out of service). As
other hydraulic structures, the intake and affiliated water supply systems are also
irreversible structures. Once they are constructed, there is no an economical way of
alteration or replacement. Therefore, before the construction, the designer must study
the problem in very detail and carefully design the intake structure.

It must be known that, it is uneconomical and even impossible to divert water without
any sediment content (completely sediment-free water) through the intake structure.
Inevitably some sediment will be present within the diverted water. The following three
criteria must essentially be considered along with others (such as, economical
construction, maintenance and operation):
a) The intake must be able to divert the demand water discharge.
b) The amount of the sediment must be the least in the diverted water.
c) The maximum size of the sediment particle allowed to be present in the diverted
water must be as small as possible the flow conditions permit.

The stream flow may carry chemical, biological and physical (solid material) pollutants.
Depending on the purpose of usage of the diverted water, one has to be aware of the
pollutants in choosing the method of diversion, the location and type of the intake when
there are alternative locations for the intake structure along the stream. As an example,
consider the stream in Fig.2.

If the diverted water is to be used for the purpose of drinking (domestic usage), the
section A in Fig. 2 is prefer to B or C as the location of the intake provided that the
stream section at A comfortably supplies the demand discharge. The stream section A
carries much less sediment and chemical pollutants than B and C. The stream flow at C
carries a lot of chemical and physical (sediment) pollutants. The treatment of the water

418
 Sediment Control Structures for Headworks and Intakes

Figure 2. Location of the intake on the alignment of the stream

taken at C is very expensive. If the diverted water is used for irrigation purposes, after
the section A, the next location of the intake can be the section B provided that the
sediment content (clay soil) carried by the diverted water is not harmful for the
agriculture (plant growth) when it is given to the field and it covers or penetrates
through the soil.

In the world there are streams carrying useful soils and they help in improving the soil
quality when their waters are given to the agriculture field and barren soil by means of
diverted irrigation water as in the case of the Nile River (Egypt) and the Yellow River
(China). On the other hand, there are rivers carrying soils such as limestone, salt, etc.
which are harmful for the agricultural fields. When their waters are given to the
agricultural field of good quality (productive) soil by means of diverted irrigation water,
they can spoil the agricultural field soil. Due to this adverse effect, several civilizations
in the history have been ended.

Before going into details and types of intakes, it is better to explain the following three
important topics which are common principle concepts for most of the intakes and
diversion methods to be utilized at a section of a stream. They are as follows:
a) Sediment concentration distribution across the depth of the stream flow.
b) Effect of the bend (curvature) on the sediment transport in the stream.
c) Effect of branching of a stream on the sediment transport.

2.1. Sediment concentration distribution

It can be proved that in two-dimensional steady uniform stream flow carrying
suspended sediment, the suspended sediment concentration distribution across the depth
of the stream flow is as follows (ASCE, 1975);

z
C ⎛h-y a ⎞
=⎜ ⎟ (3)
C a ⎜⎝ y h - a ⎟⎠

419
Sediment Transport Technology

in which the non-dimensional parameter z is;

w
z= (4)
β k u*

where, C; volume based sediment concentration at y, Ca; the reference concentration at

y = a, k; Von Karman constant (generally taken as equal to 0.4), u*; bed shear velocity,
and β; a constant which is about 1 (β ≈ 1 taken). In the above equality y is measured
vertically upward from the stream bed. Eq. 3 is diffusion based and also available in all
books dealing with sediment transport. The graph of Eq. 3 is as shown in Fig. 3. In this
figure; d; total flow depth of the stream, and a = 0.05d taken.

In Fig. 3 it is seen that as the size of the sediment in suspension gets smaller, the
concentration distribution approaches the uniform (concentration changes very slowly
with y). This is the case for plain alluvial streams where slope is very less and sediment
particles are very fine. As the particle size gets larger (coarser) the concentration
distribution rapidly decreases as y increases or it rapidly increases as y becomes small.
As y approaches zero; the concentration distribution near the bed becomes infinitely
large. This case mostly corresponds to mountain streams (high slope streams).

In practice, it is relatively easier to deal with the fine sediment in comparison to the
large size sediment (coarse gravel and stone etc.). Therefore fine sediment particles are
preferred to the large size sediment particles to be present in the diverted water. In Fig.
3, the sediment concentration distribution indicates that the intake must take water from
regions as close to water surface as possible so as to have the least amount and small
size particles of sediment in the diverted water. As the percent of water taken from the
regions close the bottom increases, both the amount and particle size of the sediment
content of the diverted water increase which is no good. In cases where the intake

Figure 3. Concentration distribution [ASCE, 1975]

420
 Sediment Control Structures for Headworks and Intakes

inevitably placed close to the bottom (i.e. depth of the stream flow is small) special
precautions and measures must be taken (i.e. fortified screening) in order to reduce the
sediment entrainment.

2.2. Effect of the bend (curvature) of the stream

Let us consider a bending (meandering) section of a stream flow as shown in Fig. 4. The
curvature (bend) of the stream is in the horizontal plane.

Figure 4. Effect of the secondary flow on the scouring and deposition in a bend

421
Sediment Transport Technology

The distribution of the actual velocity of the main flow across a vertical depth (i.e. EF)
of the stream in the bend is shown in Fig. 4d.

V0 is the average velocity across the vertical depth (i.e. EF). In theory it is generally
assumed that there is no energy loss in the bend (no friction). Therefore, total energy is
the same across the width of the bend as shown in Fig. 4b. Since the resistance (energy
loss) is assumed to be negligible, all fluid particles across the width of the bend in Fig.
4b have same total energy of H0. Therefore;

H0 = V02/(2g) + p/γ + z = constant (5)

in which; p; pressure, z; geometric elevation, γ = ρg; specific weight of the fluid, g;

gravity acceleration, and ρ; density of the fluid.

Let us consider an inviscid (ideal fluid, no shear stress) fluid element of a rectangular-
arch prism as shown Figs. 4b and 4c. The pressure at its center is denoted as p; in
general p = (r, z, θ) where r; radius of curvature (drawn in horizontal plane) of streamlines
at any vertical line, i.e. EF in Figs. 4b and 4c, and θ; the angle as shown in Fig. 4c and
Fig. 5.

One can write;

∂p ∂p ∂p
dp = dθ + dr + dz (6)
∂θ ∂r ∂z

As shown in Fig. 5, due to the tangential velocity (main flow velocity) the fluid element
is subjected to the centrifugal acceleration of V02/r and thus the centrifugal force of
(ρ.r.dθ.dz.dr) (V02/r). The fluid element has no vertical and radial velocity components.
It has velocity of V0 in θ direction only. The variations of the flow characteristics in θ
direction are negligible. The equation of motion (Navier-Stokes equation) of the invicid
fluid element in θ direction yields;

Figure 5. Forces on a curve fluid element

422
 Sediment Control Structures for Headworks and Intakes

∂p
≈0 (7)
∂θ

The equation of motion in r horizontal-radial direction gives;

1 ∂p V02 ∂p V02
0=- + or =ρ (8)
ρ ∂r r ∂r r

The equation of motion in z-vertical direction results in;

1 ∂p ∂p
0=- - g or =- ρ g (9)
ρ ∂z ∂z

If one substitutes Eqs. 7, 8 and 9 into Eq. 6, it yields;

V02 V2
dp =ρ dr − ρgdz or d ( p + γz ) = ρ 0 dr (10)
r r

Let us define;

p
+ z = h (piezometric level) or ( p + γz ) =γh ( piezometric pressure) (11)
γ

Hence, Eq. 5 can be rewritten as;

V02 V2
H0 = + h = constant or h = const - 0 (12)
2g 2g

One should keep in mind that V0 varies with r only; V0 = V0(r). The derivative of Eq.(12)
with respect to r gives;

dh V dV0
=- 0 (13)
dr g dr

From Eqs. 10 and 11;

dh V2
=ρ 0 (14)
dr r

Eqs. 13 and 14 yield;

dV0 dr dV0 V
=- or =- 0 (15)
V0 r dr r

After integration;

423
Sediment Transport Technology

ln (V0 r ) = C1 = a constant

C0
V0 = (16)
r

in which, C0 is a constant. This result indicates that the flow in the bend, under the
aforementioned assumptions, is also a circular potential flow. Eq. 16 is valid for flows
in the bends of the closed conduits and free surface flow.

In reality, there is friction loss (resistance to the flow) along the boundary of the flow.
Due to this resistance, the velocity distribution given in Fig. 4c may change. In
conjunction with this change in the velocity distribution, especially due to the effects of
the bend, the point of the maximum velocity (max.V) well approaches towards to inner
bank. The maximum velocity (max.V) doesn’t occur just at the inner bank where the
depth of flow is minimum, but it occurs at some small distance from the inner bank as
in Fig. 4c which also shows how the path of the maximum velocity moves towards the
outer bank immediately just after the bend. In some cases, starting just at and about the
beginning of the fourth-quarter of the bend, the path of the maximum velocity may
considerably oscillate between the inner and outer banks of the bend.

This maximum velocity and the oscillation in its path can cause damages and changes in
the beds of the canals artificially constructed in erodible soils.

Since V02/r and V0/r are always positive, Eqs. 14 and 15 prove that as r increases, V0
decreases and the piezometric level h increases from inner bank to the outer bank. This
means, flow depth at the outer bank is larger than that at the inner bank. The flow
velocity V0 at the inner bank is larger than that at the outer bank. Let us consider a
vertical fluid element of mass of dm in Fig. 6. The radial pressure forces acting on right
and left sides of the element are denoted as F1 and F2 in Fig. 6. The radial centrifugal
force acting on the element is dmV02/r. The equation of motion of the element in r
direction yields;

Figure 6. Velocity distribution and the centrifugal force

424
 Sediment Control Structures for Headworks and Intakes

V02 V02
F1 + dm - F2 = 0 or F2 - F1 = dm (17)
r r

Since dmV02/r is always positive, Eq. 17 infers that F2 is larger than F1. Their difference
is balanced by the centrifugal force. The balancing force dmV02/r is same for all
particles in the same vertical (i.e. EF). It is clear that pressure force (or pressure)
increases as one goes from the inner bank to the outer bank. The actual velocity
distribution across the vertical depth (i.e. EF) is not uniform (Fig. 6 or Fig. 4d).

The actual velocity distribution across a vertical depth (i.e. EF) of the stream flow
shows that the velocities in the upper regions are much larger the velocities in the
regions close to the bottom. This is due to the viscosity effect (friction) of the bottom
boundary. As presented in Fig. 4d and Fig. 6, above the level of B, the actual velocity is
larger than the average velocity V0 (V>V0) and below B, the actual velocity is less than
the average velocity (V<V0). At B, V = V0.

Since the centrifugal acceleration is proportional to V2, the particles having large
streamwise velocities are subjected to large centrifugal forces and the particles having
small velocities are subjected to small centrifugal forces. Therefore the particles above
B (Fig. 6), especially the fluid particles close to surface, will receive centrifugal forces
much larger the average centrifugal force based on V0 (or dmV02/r) and the particles
below B (Fig. 6), especially those close to the bottom will receive centrifugal forces
much less than the average centrifugal force based on V0 (or dmV02/r).

Since they receive large centrifugal force, the fluid particles above the level of B
(especially those close to the water surface) move outwards (towards the outer bank)
from the center (vertical line). On the other hand, consider a vertical column of fluid
element of CTRE as shown in Fig. 7. In this column, take two fluid elements with the
same mass of dm and cross-section area of dA normal to the figure. One of the fluid
elements is taken in the region close to the free surface and called the upper element.
The other fluid element is taken in the region close to the bottom boundary and called
the lower element as shown in Fig. 7a. In this figure a, b, c, d, e, f, j and s are the
pressure values. From Eq. 12 it is seen that the pressure distribution across the vertical
depth is almost hydrostatic. V1 and V2 are the velocities for the upper and lower
elements respectively (Fig. 7a).

The pressure on the fluid element close to the free surface is relatively small. But the
velocities of the fluid elements close to the free surface (i.e. above the point B) are very
large. The centrifugal force dmV12/r acting on the upper element in r direction is much
larger than the net pressure force (pavup·dA) acting on the upper element in opposite
direction of r. Here, pavup is the average net pressure on the left side surface of the upper
element. Writing the equation of motion of the upper element in r direction gives (Fig.
7b);

⎛ V2 ⎞
⎜ dm 1 - p avup dA ⎟ = net force is in r direction
⎜ r ⎟
⎝ ⎠
+ (net force is towards the outer bank) (18)

425
Sediment Transport Technology

F Vmax outer bank

C
free surface
a upper element c
inner bank dm V1
actual velocity
r b G K d B
Vo
center of curvature e M N j pressure distribution
dm V2
V1 > Vo > V2
f lower element s
pressure distribution V1 >> V2

T R

dA Pavup
upper element upper element upper element
a c c-a

b G
dm
K d
dm V 1
2
/r
= dm
d-b
dm V1 / r
2

= dm

( dm V1 / r - Pavup dA )
2
( dm V1 / r > Pavup dA )
2

lower element lower element Pavl

e j j-e lower element

f
dm
s
dm V 2
2
/r
= dm
s-f
dm V2 / r
2

= dm

( Pavl dA - dm V2 / r )
2
( dm V2 / r < Pavl dA )
2

inner bank inner bank

outer bank
outer bank
center of curvature
= secondary current
=
Figure 7. Development of the secondary flow in a bend

Since dmV12/r >> pavup·dA, the upper element will be moved and pushed towards the
outer bank by the force difference (net force) acting on it in the direction of r (towards
the outer bank).

Let us do the same analysis for the lower element below the level of B.

Velocities of fluid elements close to the bottom (i.e. below B) are very small but the
pressure is very high. Therefore, the net pressure force pavl·dA >> dmV22/r. Here, pavl is
the average net pressure on the right side surface of the lower element. Writing the
equation of motion of the lower element yields (Fig. 7c);

⎛ V2 ⎞
⎜ dm 2 - p avl dA ⎟ = net force is in the opposite r direction
⎜ r ⎟
⎝ ⎠
- (net force is towards the inner bank) (19)

Since dmV22/r << pavl·dA , the net force acting on the lower element (pavl·dA - dmV22/r)
will be in the direction of the larger force (pavl·dA) which is acting towards the inner

426
 Sediment Control Structures for Headworks and Intakes

bank (opposite to r direction). Therefore, the lower element is pushed and moved
towards the inner bank.

From the aforementioned explanations, it is well understood that the fluid particles in
the upper layers are moving towards the outer bank and the fluid particles in the lower
layers close to the bottom are moving towards the inner bank. As a result of the opposite
movements of the upper and the lower fluid particles, a circulatory cross-current in the
transverse cross-section of the flow in the bend develops which is called the secondary
flow or secondary current as shown in Fig. 7d and Fig. 4b. The secondary flow close to
the bottom occurs from the outer bank to the inner bank.

In the above equations, it is seen that the strength of the secondary flow highly depends
on the flow velocity (especially in the upper layers) and the radius of curvature of the
bend. Smaller the radius of curvature, larger is the centrifugal force and hence, stronger
is the secondary flow. Similarly, larger the flow velocity (especially in the upper layers)
stronger is the secondary flow. The centrifugal force is directly proportional to square of
the velocity (V2) and inversely proportional to r. The strength of the secondary flow
parabolicly increases with V. This means that the dependency or sensitivity of the
strength of the secondary flow on flow velocity (V) is more than that of the radius of
curvature r. The effect of V on the strength of the secondary flow is more than that of r.

In natural rivers and man made canals with erodible soil banks, the secondary flow
erodes (scours) the outer bank of the bend and deposits the sediment in the inner bank
of the bend and causes the cross-section of the bend to take the general shape of the one
shown in Fig. 7e.

As it is very often mentioned later, sediment movement about the intakes is highly
under the influence of the secondary flow that has very important effects on controlling
the sediment entrainment and sediment content of the diverted water through the intake.
The positive or negative effects of the secondary current on the sediment transport in
the stream, on the design of the intakes and on the stability of the river bed and its
alignment etc. must never be forgotten and underestimated.

The secondary current scours the outer bank, carries the sediment from the outer bank
and deposits it in the inner bank as shown in Figs. 4b and 4c. In this figure, it is obvious
that if the intake is constructed at the outer bank, the diverted water contains less
sediment. If one constructs the intake at the inner bank, the sediment content of the
diverted water will be several times more than that of the intake constructed at the outer
bank. From the aforementioned explanations it is clearly understood that the
construction of an intake at the inner bank is a very big mistake. The intake must be
placed and constructed at the outer bank of the bend. Experiences have proven that the
proper location for the intake is a place within the second-half of the bend as shown in
Fig. 4c.

The velocity of the secondary flow in the bend with small radius of curvature can be
compatible with the main stream flow. It is very clear that the secondary flow and its
effects are always present and the intake must be placed on the outer side of the
curvature (never on inner side of the curvature) wherever the streamlines gain
curvatures. This is a very important principle for the sediment entrainment at intakes
and diversion canals in connection with the streams carrying sediment.

427
Sediment Transport Technology

Let us consider the fluid elements in the upper layer close to the free surface and lower
layer close to the bottom boundary (Fig. 8).

A fluid element has two velocity-components. One is due to the secondary flow and the
other is due to the main stream flow. The total resultant velocity of the fluid element is
the resultant of these two velocities. The fluid element moves in the direction of the
total velocity vector as shown in Figs. 8b and 8c.

The velocity of the secondary flow Vs in the region close to the free surface is towards
the outer bank and compatible with the velocity of the main stream (V). The total
velocity vector Vr is considerably inclined towards the outer bank (Fig. 8b).

Since the fluid particles close to the free surface move in the direction of their resultant
velocity vectors Vr; their paths are considerably inclined towards the outer bank (Fig.
8d). Hence, the upper layer fluid particles move towards the outer bankwise.

Figure 8. Development of the spiral flow in a bend

428
 Sediment Control Structures for Headworks and Intakes

On the other hand, in the case of lower layer fluid elements close to the bottom, since
the secondary flow velocity vector Vs is towards the inner bank and compatible with the
main flow velocity V close to the bottom, the resultant velocity vector Vr is considerably
inclined towards the inner bank. Therefore, the fluid particles close to the bottom
boundary flow towards the inner bankwise (Fig. 8d).

In reality, there is friction between the upper and lower layer fluid flows shown in Fig.
8d. Therefore, these flows affect each other and determine to flow in an agreed
combined resultant pattern which is in the form of a spiral-flow advancing in the main
flow direction. Thus, the real flow pattern in the bend section of a stream is going to be
the spiral-flow as approximately presented in Fig. 8e and Fig. 4c. Any interference to
this spiral flow affects the sediment motion about the intake or in any place wherever
there is a curvature.

3. Effects of Some Elements on Sediment Movement

3.1. Spurs
As it is explained later, in practice, in order to eliminate or reduce the adverse effects of
the curvature flow or to create an artificial curvature flow so as to utilize its effects in a
useful way, several types of structures or arrangements are used such as spurs, baffle-
walls, guide-vanes, etc.

If there is no bend in the stream flow, an artificial curvature flow can be created by
means of spurs, guide vanes, baffle walls and submerged plates, etc. These structures
can be fully or partially submerged. Especially the spurs are very often used. A spur is
an artificial structure of some length projecting from the bank into the stream flow. The
type of the spur structure and its construction material depend on the local conditions
and economy. Generally it is constructed with local natural material (i.e. gravel- stone
filled body’s surface is covered with a reinforced concrete-shield). The spurs are also
useful for the protection of stream banks against scouring. Depending on the stream
flow and local conditions, several spurs in series with required intervals can be
constructed. A spur and its effect on the stream flow are sketched in Fig. 9.

3.2. Screens
Sand screens filter the unwanted coarse sediment and direct the bed load away from the
entrance of the diversion canal (or the intake) back into the stream. For the screens to be
effective, there must be a sufficiently high-velocity stream flow passing from the point
of the diversion to carry the bed load away. Otherwise, sediment detained by the screen
deposits and accumulates and the efficiency of the screen decreases.

Figure 9. Effect of a spur

429
Sediment Transport Technology

3.3. Guide-vanes (submerged vanes)

The guide vanes are used to develop a helicoidal (spiral) flow similar to those created
by the bend of a stream. The guide vanes are two types as explained below.

a) Submerged bottom guide-vanes: These guide vanes are placed at upstream of the
intake near the stream bed in order to direct the stream flow close to the bottom
and hence direct the bed sediment away from the intake as shown in Fig. 10a.
b) Surface vanes: These vanes are projected into the water for some sufficient
depth from the water surface in order to direct the surface flow towards the
intake. This introduces a transverse flow to the bottom by means of which the
bed sediment is directed away from the intake (Fig. 10b).

3.4. Submerged dividing-plate

In some cases, in order to eliminate the movement of the bed sediment towards the
intake by means of the secondary flow close to the bottom caused by curvature of the
stream flow entering the diversion canal (or the diverging branch of the stream) and
make the diverted water through the intake consists of the upper layer (surface water)
only so as to make the diversion water as free of bed load as possible, a submerged plate
of sufficient length and width is placed just at the diversion point between the surface
flow entering the intake and the bottom flow not entering the intake.

If the stream flow depth is sufficiently large, the plate can be placed as shown in Fig.
11b. If the stream flow depth isn’t sufficiently large, the bottom of the stream about the
intake and to some length downstream of the intake can be dropped for a sufficient
depth as shown in Figs. 11c, 11d and 11e.

(a) (b)
Figure 10. Effect of guide vanes, (a) Bottom guide-vanes; (b) Surface guide-vanes
[ASCE, 1975]

430
 Sediment Control Structures for Headworks and Intakes

(e) [Çeçen, 1995]

Figure 11. Effect of a dividing plate on sediment movement at a diverging-branching section

4. Effect of Branching on the Location of the Intake

Type of branching is determined by the flow directions in branches. Branching can be
two types:
a) Diverging branch
b) Joining branch

431
Sediment Transport Technology

4.1. Diverging branch

Consider a diverging-branching section of a main stream flow as shown in Fig. 12. The
branches are named as branch I and branch II. The axis of the branch I is the same as
that of the main stream. The branch I flow doesn’t make an angle with the direction of
the main stream flow. On the other hand, the flow of branch II makes an angle of ϕ with
the direction of the main stream flow.

(d)
Figure 12. Effect of diverging-branch on the sediment movement
(a), (b), (c) [Bayazıt, 1971],
(d) [Bulle, 1926 reported by Çeçen, 1995]

432
 Sediment Control Structures for Headworks and Intakes

As water enters the branch II, the flow lines in the main stream about the entrance of the
branch II has to bend. As a result of bending of the streamlines, there develops a
secondary flow close to bottom which sends a lot of sediment from the main stream into
the branch II. The sediment load in branch II is very sensitive to the angle of ϕ which
determines the amount of curvatures of the streamlines. Hence, a small increment in ϕ
causes so much increment in the amount of the sediment entering the branch II. The
sediment load in the branch II is much more than the sediment load in the branch I as
shown in Fig. 12d which is obtained from the experimental results [Bulle, 1926,
reported by Çeçen, 1995]. For example, for φ = 90˚, if both branches I and II carry 50%
of the stream flow-discharge, as shown in Fig. 12b, 95% of the sediment load of the
main stream enters the branch II because the branch II takes place on the inner side of
the curved flow. The secondary current near the bed carries and forces the sediment
close to the bottom to enter the branch II. Only 5 % of the sediment load of the main
stream enters the branch I since it remains on the outer side of the curved flow. When a
submerged sill (vane or wall) is placed as shown in Fig. 12c [Bayazıt, 1971], the surface
flow entering the branch I is almost a straight flow, but the bottom flow entering the
branch II is curved so that the branch II takes place on the outer side of the curved
bottom flow. Hence, the occurring secondary flow carries and forces the bottom
sediment to enter the branch I rather than branch II. Therefore, due to the submerged sill
positioned as shown in Fig. 12c, the branch I carries 95% of the sediment load of the
main stream and remaining 5% of the sediment is carried by the branch II.

4.2. Joining branch

The stream flow having large specific-force (or momentum function) component normal
to the direction of the main stream flow where it just enters the confluence is called the
dominant stream which repels and pushes the flow of the stream having smaller
specific-force component normal to the main stream flow aside as shown in Fig. 13.

Figure 13. Sediment movement at the confluence of two streams

433
Sediment Transport Technology

Therefore, the stream flow of large specific-force component separates from the
adjacent bank of the main stream and makes a turn (curvature) before it completely
mixes with the other stream flow at some distance downstream from the confluence.
The interface of the confluent streams is clearly observable by an abrupt change in
color. The secondary flow occurring due to the curvature of the flow of the dominant
tributary in the confluence transports the sediment close to the bottom from the interface
to the adjacent bank of the main stream at the confluence and deposits the sediment
there as shown in Fig. 13. The bank section of the main stream where the sediment
deposition occurs is no good location for an intake as indicated in Fig. 13.

5. Types of Intakes
The selection of the type of the intake is very important because of the sediment content
of the diverted water. The type of intake to be chosen depends on several factors such as
the amount of the diverted water discharge (demand), the purpose of usage of the
diverted water, sediment load, sediment size, sediment concentration distribution in the
main stream in all its regimes, flow velocity, flow depth, direction of the stream flow,
topography and geometry of the stream flow boundaries, whether or not the stream bed,
banks and its alignment are stable, pollution, accessibility, costs of construction,
operation and maintenance and other local conditions. Not to forget that the stream flow
section where the intake is constructed must guarantee the supply of water larger than
the demand at least during the period of diversion and the intake must comfortably
divert and pass the demanded water in all conditions with the least amount of sediment
content.

As the amount of the diverted water increase, the size of the structure and costs of
construction, operation and maintenance of the intake considerably increase. Therefore
it may be better to classify the intakes as large scale and small scale intakes from the
amount of diverted water point of view.

5.1. Large scale intakes

These intakes are supposed to provide considerably large amount of diverted water.
Therefore, their sizes and cost of construction operation and maintenance are large.
They mostly require water ponding or water surface elevating structures with or without
gate systems which cause an additional cost and other concerns. The most common
large scale intakes are as follows:
i) Bottom type intakes
ii) Lateral type intakes
iii) Frontal type intakes
iv) Other types of intakes

Let us explain the aforementioned intake types one by one.

5.1.1. Bottom (Tyrol) type intake

Bottom type intake is also known as Tyrol type intake. As in the case of mountainous
streams, when the stream flow depth is relatively small and slope is large and the stream
carries a lot of coarse (gravel, stone, rock) bed load sediment, the bottom type intake is
a good choice for water diversion as shown in Fig. 14 and Fig. 15. A kind of weir is

434
 Sediment Control Structures for Headworks and Intakes

Figure 14. Bottom type intake, Kavraz-Turkey [Çeçen, 1995]

Figure 15. Different bottom type of Intake [Danelia, 1975]

constructed across the width of the stream (if necessary, the width of the stream is
artificially and smoothly narrowed down to a desired value by guiding-walls or levee-
structures, but must be able to pass the maximum flood discharge and large stones and
trees etc. washed out from the banks and carried by the flood water.

The diversion canal takes place within the body of the weir. There are bottom racks or
screens placed over the diversion canal. As the stream flow passes over the weir, the

435
Sediment Transport Technology

required amount of the stream flow enters the diversion canal through the rack or screen
and large size bed sediment (gravel, stone etc.) and debris (such as trees etc.) pass over
the diversion canal. Thus, the diverted water in the diversion canal is free of unwanted
coarse sediment. The coarse sediment (gravel, stone, rock and debris) passed over the
racks and sent to downstream of the weir (or intake) are swept and carried away by
either the remaining stream flow or later by the flood water. Depending on the
characteristics of the stream flow and the sediment carried by the stream, the sizes of
the rack or screen openings are determined by the designer. There are design examples
in the literature i.e. Fig. 16. The small size sediment content can be separated from the
diverted water by suitable means.

5.1.2. Lateral type of intake

The lateral type intake is very commonly used. As it is indicated by its name, water is
diverted laterally from the stream by means of an intake structure placed at the banks of
the stream section. Due to the way it diverts the water, one has to very properly design
and choose the location of the lateral intake structure. Especially, the effects of the
curvature of the stream flow on the sediment entrainment are very highly pronounced
for the lateral intakes in comparison to other types of intakes. Lateral intakes are
capable of diverting large amount of water. Generally (if the depth of the stream is not
sufficient and its variation is large) a weir with and without gates is constructed across
the stream width in order to elevate and keep the stream water surface at a desired level
as shown in Figs. 17, 18, and 19.

There are several types of lateral intakes successfully utilized in practice. There is
always a rack-screen just at and about the intake in order to prevent the entry of the
coarse bed material and debris. It is very essential that the intake must always be placed
on the outer side of the curved stream flow. Any necessary measure must be taken to
make the lateral intake to stay on outer side of the curved flow. For this purpose, as
explained before, artificial structures such as spurs, baffle-walls, submerged-vanes etc.
are successfully used in practice in order to create a secondary flow which sends the bed
sediment away and keep it off from the intake. Otherwise it would enter the intake. If
the lateral intake flow makes no angle with the stream flow direction there will not be a
secondary flow developed (Fig. 20). In this case, although no additional bed sediment is
sent to the intake, the normal bed material of the stream enters the intake. Even in these

Figure 16. Bottom type of intake, İkizdere-Turkey [Çeçen and Garbrecht, 1957]

436
 Sediment Control Structures for Headworks and Intakes

Figure 17. Types of lateral intakes [Çeçen and Garbrecht, 1957]

Figure 18. Lateral intakes [Müller, 1955 reported by Çeçen, 1995]

437
Sediment Transport Technology

Figure 19. Common elements of a lateral intake [Çeçen,1995]

Figure 20. Spur type lateral intakes [Grishin, 1987]

situations, to keep the normal bottom sediment away from the intake, an artificial
structure such as a spur or a baffle-wall etc. which creates a curved flow (in such a way
that the intake stays on the outer side of the curve) is still advised. For this aim, the
spurs generally are placed on the opposite side bank of the lateral intake (Figs. 17 and
18).

In spite of all these precautions, some coarse bed material, by any reason may still try to
enter the intake. To eliminate their entry, a gravel passage with a regulating gate is
constructed at just downstream of the intake (between the weir and the intake as shown
in Fig. 19). Although the gravel passage structure is part of the weir structure, the
bottom level of the gravel passage is well below the bottom level of the intake structure.
The main purpose of the gravel passage is to remove and flush the coarse bed sediment
away otherwise it accumulates and enters the intake.

Depending on the amount of water in the stream, the gravel passage can continuously be
opened (flushing) or its gate can be opened for flushing from time to time. The flushing
period is determined according to local conditions (i.e. sediment sampling of the
diverted water is done). When the unwanted coarse sediment starts to be present in the
diverted water, the gravel passage gate is opened and flushing is conducted). The weir
structure makes the flushing through the gravel-passage more possible and effective. By
elevating the stream water surface, a weir also provides the necessary energy level to
make the flushing of the settling-basin (generally constructed at some downstream
distance from the intake) possible in the cases of plain streams. Müller, 1955 [reported

438
 Sediment Control Structures for Headworks and Intakes

by Çeçen, 1995] has conducted several experimental works and concluded that in rivers
carrying large amount of sediment, only half of the stream flow discharge can be
diverted without entering at the same time sediment material.

5.1.3. Frontal type of intake

The frontal type of intake is relatively new in comparison to lateral and bottom types of
intakes. This intake type is developed by Çeçen [Camp, 1946, Çeçen, 1973, 1975, 1977,
Çeçen and Garbrecht, 1957, Çeçen and Bayazıt, 1975]. The frontal type intake is much
more effective than the lateral and bottom intakes from the coarse sediment entrainment
point of view. It admits the water almost opposite to the stream flow direction. The
stream flow almost directly (but not very necessarily) enters the intake. The frontal
intake can successfully be used for both relatively high slope and plain streams.

The intake entrance structure is projected into the stream flow for some proper distance.
A typical frontal intake structure is shown in Fig. 21. There are also two-storey frontal
intakes (Fig. 22).

As it is seen in Fig. 21, a rack-screen, a weir structure, a spillway, a gravel passage with
a regulating sluice gate just underneath the bottom of the frontal intake and approach
flow smoothing walls are present. Depending on the size of the intake, the entrance of
the intake can be divided into more than one division by means of partition walls. The
sluice gate is always partially opened in such away that the large bed sediment particles
can pass through the opening of the gate (opening larger than this isn’t recommended
due to loss of excessive water). The flushing through the gravel-passage is continuous.
It is recommended that the bottom slope of the gravel passage must be larger than 0.02
but it shouldn’t be very large either. This intake has two basic advantages in comparison
to the lateral and bottom intakes. They are as follows:
a) It eliminates the development of the artificial curve flow and hence the effects of
the secondary current.
b) It is a kind of skims the stream flow close to the surface (the percentage of the
stream flow close to the bottom is very less)

Figure 21. The frontal intake developed in the Figure 22. Two-storey frontal intake
Technical University of Istanbul [Grishin, 1987]
[Çeçen, 1995]

439
Sediment Transport Technology

Due to these two important reasons, the coarse sediment can’t enter the frontal intake if
it is properly operated. The coarse sediment content of the diverted water through the
frontal intake is much more less than that of the diverted waters through the lateral and
bottom intakes. Today, in practical application, there are several good examples of the
frontal intakes proved to be working properly after several years in service. One
example is given in Figs. 23 and Fig. 24.

The comparison of a properly designed frontal intake with the lateral intake under the
same flow conditions is provided in Fig. 25 which is obtained from experimental
results. Its application has shown that the diverted water through a properly designed
and operated frontal intake can be as much as 90% free of the sediment of the stream

Figure 23. Picture of the prototype frontal intake and affiliated structures (settling-
basins are seen in foreground), Göksu-Turkey [Çeçen, 1995]

Figure 24. Project details of the frontal intake constructed at Göksu-Turkey [Çeçen, 1995]

440
 Sediment Control Structures for Headworks and Intakes

Figure 25. Comparison of the lateral and frontal intakes (having

identical flow conditions) from the sediment
entrainment point of view [Çeçen, 1995]

flow. The secondary currents and high turbulence must be dampened and minimized.
Because, these types of currents lift the bottom sediment up and entrain to the intake.
Also there must be a dividing wall of sufficient length towards the upstream between
the gravel passage and the weir. It is also advised that the flow smoothening curtain
walls of some sufficient length have to be placed at the upstream of the gravel passage.

In order to avoid the clogging of the gravel passage and prevent the entry of the coarse
sediment, during the flood, the sluice gate opening in the gravel passage and hence the
flushing discharge are increased. The distance between the sluice gate and the beginning
point of the intake effects choking [Çeçen, 1995]. During the flood or when the sluice
gate opening is increased, the strength of the turbulence of the flow at the entrance of
the intake increases. Due to the turbulence, the coarse bed sediment may be lifted up
and entrained the intake. This is the reason why the proper operation of the sluice gate is
required.

5.1.4. Suggestions on choosing the intake type

The main principle is that the designer must evaluate the data of the sediment size,
sediment concentration distribution, discharge, depth, velocity, slope stability of the
banks and alignment of the stream flow, construction, operation and maintenance of the
intake, topography, accessibility, the amount and usage purposes of the diverted water
etc. and determine the type of the intake. Basing on the experience of the researches,
some suggestions on how to choose the type of the intake are mentioned below.

Lateral intake
• The lateral intake is generally more economical in comparison to others.
• The lateral intake is good for the stream flows carrying fine sediment.

441
Sediment Transport Technology

• Lateral intake is good when the discharge of the diverted water is large.
• Lateral intake is good for the streams that haven’t excessive slopes.

Bottom intake
• Bottom intake is good for the mountain streams that have slopes greater than or
about 0.04.
• Bottom intake is good if the coarse bed sediment is much more than the finer
sediment in the stream.
• If the finer sediment is much more than the coarse bed sediment in a stream that has
higher slope and supercritical regime with high Froude number, the bottom intake
isn’t recommended. Because a lot of fine sediment enters the intake (or diverted
water). In such cases, one can use other types of intakes or multi-story bottom
intakes [Danelia, 1975] with rack at different levels.

Frontal intake
• The frontal intake can be used for both the mountain streams (high slope) and plain
streams (low slope) provided that sufficient stream water depth is available or
created by means of a weir structure and flushing operation is properly conducted.
• The frontal type intake is also good for a stream where up to 50% of the stream
discharge is diverted.

5.1.5. Other types of intakes

There can be several typical arrangements which work as intakes. The typical intakes
are generally determined from the case-studies. Therefore, they are not widely used in
practice. The scale of the diverted water through these typical intakes is relatively
smaller than that diverted through the lateral, bottom and frontal intakes. The sizes of
these structures aren’t that large. Some typical intake types documented by several
researchers reported by Çeçen, 1995 are presented in the following figures with related
explanations.

Figure 26. Aksoy’s artificial island [Reported by Çeçen, 1995]

442
 Sediment Control Structures for Headworks and Intakes

Curvature is given to both side flows by constructing an arch type wall (generally made
of concrete) and a circular artificial island in the stream in order to keep the lateral
intakes to stay on outer sides of the corresponding curved flows.

Figure 27. A special type lateral intake [Scheuerlein, 1984]

Figure 28. A special type intake utilized in the River Linth-Switzerland [Scheuerlein, 1984]

Figure 29. Pre-settling basin

443
Sediment Transport Technology

Frontal type intake: The diverted water is taken to a pre-settling basin. From time to
time, the settled sediment in the settling basin is removed by any means (i.e. flushing,
hand or mechanical removal).

Diverted water is taken from the surface layers of the stream by skimming or using fine
screens.

Figure 30. Skimming of the surface flow of the stream

Figure 31. Horizontal rack screens are placed on the weir piers [Samarin et al., 1961]

Figure 32. Vertical rack screens are placed on the weir piers [Samarin et al., 1961]

444
 Sediment Control Structures for Headworks and Intakes

Figure 33. Vertical screens are placed at the upstream front of a pier [Midgly, 1977]

Figure 34. Vertical screens are placed at the upper front of the weir piers
[Midgley, 1977, and Samarin et al., 1961]

Figure 35. Screens are placed at the ceiling of the closed conduit [Samarin et al., 1961]

445
Sediment Transport Technology

Figure 36. Water is taken from an artificial filtering-well constructed by fortified wire mesh
filled with graded gravel and filtering material

5.2. Small scale intakes

In the cases of relatively small amount of diverted water (i.e. say about <500 lt/s), no
need to construct expensive diversion structures. The required amount of water can
directly be diverted from a water resource by means of a simple pipe or duct intake as
presented in Fig. 37.

These intakes are very simple, small in size and their construction, operation and
maintenance costs are very low. The most commonly used small intake in practice is
either a pipe or a rectangular (duct) intake.

As for small intakes, two main problems are encountered which cause deficiencies in
the capacity, efficiency and maintenance of the intake and the affiliated water
conveying structures after the intake. They are;
a) Air-entrainment
b) Sediment entrainment

Type II

Type III
Type IV
Type I

Type V

(a) (b)
Figure 37. (a) Different types of small scale intakes; (b) Color-visualization of flow entering
intake type II (Adapted from Kocabaş and Yıldırım, 2002)

446
 Sediment Control Structures for Headworks and Intakes

The main purpose is to have the least amount, least size of sediment and no air-
entrainment in the diverted water which highly depend on the location, position and
configuration of the intake in the stream flow.

Considering the aforementioned explanations (also see the explanations little later), it is
clearly understood that the vertically flowing downward intake should be preferred
(intake Type II in Fig. 37) and the intake should be as close to the free surface as
possible from the least amount and size of the sediment content of the diverted water
point of view.

In Fig. 38a, it is seen that if the topography, property rights, and construction site
conditions permit and the banks and bed of the stream consist of sufficiently permeable
soil, the diverted water taken by means of drainage collector pipes of sufficient
dimensions placed and buried at a depth (i.e. well below the minimum stage of the
stream ) in the bank soil of the stream will be almost free of sediment due to the
filtering effects of the bank-soil and there is no problem of air-entrainment (Fig. 38).
Therefore, the drainage-collector pipe type intake (Type V in Fig. 37) is preferred
wherever it is possible although its initial construction cost may be a little higher than
that of others.

Figure 38. Diverting water by means of a drainage-collector pipe

447
Sediment Transport Technology

If the flow depth in the stream is very low, water flows around the stones in the stream
bed, stream bed is impervious and no water-level rising (damming) structure is used, to
take (divert) water from such a stream; one can bury a drainage pipe into a sufficiently
deep and wide trench opened across the width of the stream bed and fill out with
filtering gravel (which should withstand the flood flow conditions) as shown in the Fig.
38b.

If other factors are suitable but only the stream bank-soil is not sufficiently permeable,
artificially constructed stone-gravel drainage connections of sufficient number and sizes
between the drainage collector pipe and the stream flow boundary are constructed as
shown in the Fig. 39. If the construction of the drainage collector pipe is not possible,
other types of intakes presented in Fig. 37 are to be used.

Let us consider the sediment concentration across the depth of a stream as shown in Fig.
37 which indicates that the orientation of the intake has effects as important as that of
the location of the intake on both the amount and size of the sediment to be present in
the diverted water. Fig. 37 infers that if the diverted water is taken from the region close
to the free surface of the stream, both the amount and size of the sediment in the
diverted water will be the least. Hence, the most favorable location for the intake
(entrance of the intake) is the region close to the free surface of the stream. The stream
flow zones under the boundary effects (sheltering effects) of the intake pipe and solid
boundaries supply relatively small percentage of the intake flow. Most of the diverted
water comes from the stream flow zones where the boundary effects are least.

The vertically flowing downward intake type (which is the next most favorable intake
after the drainage type intake) is to be used due to the reasons explained before.

The vertical distance of the intake to the free surface is called the submergence of the
intake and generally denoted by S.

As the intake gets closer to the free surface, a well known free-vortex develops and, air
enters the intake by means of an air-core vortex tube. The air entering the intake reduces

Figure 39. Diverting water by means of stone-gravel drainage-collectors

448
 Sediment Control Structures for Headworks and Intakes

Figure 40. Vertically flowing downward intake

the discharge capacity of the intake. Researchers have found that 1% air-entrainment
reduces the discharge of the intake for about 15% [Denny, 1956]. If the arrangement in
Fig. 40b is used, one should make sure that the intake does not take place within the
downstream (wake-vortices region) of the floats so that wake vortices can not stimulates
air-entraining vortices at the intake.

To avoid the air-entrainment the submergence (S) of the intake should be larger than its
critical value which is called the critical submergence (S = Sc). Studies of Yıldırım et
al., 2000 have proven that the critical submergence Sc for a pipe intake is equal to the
radius of a spherical sink surface (which is called the critical spherical sink surface
(CSSS)) that has the same center and the discharge as the intake.

In general, the critical submergence Sc is computed from the continuity equation.

⎛ πD 2 ⎞
Qi = Ac ⋅ Vs = Vi ⎜⎜ i ⎟
⎟
(20)
⎝ 4 ⎠

in which Qi; intake discharge, Vi; average velocity at the intake, Di; internal diameter of
the intake, Ac; net working area of the critical spherical sink surface (CSSS) and Vs;
radial velocity at the CSSS. Vs is constant for a given the flow and geometrical
conditions. For example, for a pipe intake in a uniform flow of Vs ≈ U∞ /2, where U∞ is
the velocity of the uniform flow at upstream of the intake. Depending on the flow and

449
Sediment Transport Technology

geometry, the CSSS can be a complete or incomplete sphere. For a complete CSSS,
Ac = 4 π S c2 .

The computation details of the critical submergence Sc are given in the studies of
Yıldırım and Kocabaş (2000, 2002) (for circular-cross section intake) and for the
rectangular cross-section intake [Yıldırım, 2004 and 2007, Eroğlu and Bahadırlı, 2007].

As the intake gets closer to the free surface, the amount and the size of the sediment
decrease but the amount of air entrainment increases as indicated in Fig. 40.
Experiments have shown that the critical submergence Sc for a pipe intake in a common
stream flow is about Sc/Di ≈ 1-2.

In order to avoid the entrainments of the floating debris and larger sediment particles, a
sufficiently strong rack and fortified fine-screen of stainless metal against the
destructive impacts and flood flow forces can be placed at and about the intake as
sketched in Fig. 41.

If no filtering or so mechanism is used, the complete avoidance of the presence of the

sediment in the intake flow is uneconomic and almost impossible. It is possible to make
both the amount and size of the sediment entrainment least as much as the flow and
physical conditions permit. Therefore, depending on the purpose of usage of the
diverted water through the intake, the sediment particles smaller than or equal to the
permissible sediment particle are allowed to enter the intake and be present in the
diverted water. To satisfy this condition, if necessary, a settling pool can be constructed
just after the intake line. In the design of the intake and affiliated diverted water
conveying structures, the size and other characteristics of the permissible particle are
essential.

Due to the drag and lift forces originating from the effects of intake flow, not only the
sediment particles at and above the intake level but also the sediment particles traveling
in the stream flow below the intake level are also elevated and entrained the intake. It is
clear that about the intake, the slope and the curvature of the stream lines get larger
(Fig. 42).

Figure 41. Taking diverted water with a pipe intake housed in a fortified fine screen

450
 Sediment Control Structures for Headworks and Intakes

Figure 42. Entrainment of sediment by wake vortices

Therefore, the drag force on the particle traveling towards the intake considerably
increases and the particles are entrained the intake (Fig. 42). In addition, due to the
presence of the rising intake pipe section, the stream flow separates and wake vortices
occur along the rising intake pipe section (Fig. 42). These wake vortices have strong
three dimensional velocity components. Due to the strong upward velocity component,
the sediment particles well below the intake level - which can not be dragged and
entrained the intake under normal flow conditions - are easily lifted up, dragged and
entrained the intake (Fig. 42). Experiments of Yıldırım (1985) have shown that both the
amount and size of sediment entrained the intake by means of side and wake vortices of
the rising intake pipe section are several times more than those entrained from other
sections of the intake. Especially, in the cases where the intake level is close to the
stream bed a lot of sediment picked up from the bed and lower layers of the stream and
carried into the intake by means of wake vortices.

The sediment entrainment by means of separation vortices must be eliminated and also
the slope of stream lines about the intake should be decreased so as to reduce the drag
and lift forces on the sediment particles and decrease the sediment entrainment from the
layers below the intake level. Experiments of Yıldırım (1985) have shown that the
following simple structures are very effective on the reduction of the sediment
entrainment:
i) a disk
ii) a blunt body (wake structure)

These two simple structures (Fig. 43) are advised to be used together for an intake in a
stream flow. The main function of the disk is to obscure and cover the separation
vortices at their tops and avoid the entry of the highly sediment laden vortices and also
decrease the slope of the streamlines (hence, reduce the drag and lift forces on the

451
Sediment Transport Technology

Figure 43. The disk and the blunt-body

sediment particles) coming from the lower layers (Fig. 43). The disk is affixed to the
intake at its top. The disk is made of sufficiently strong stainless steel plate and it is
very useful for intakes in still water reservoirs and stream flows. The diameter of the
disk is determined later.

The blunt body (wake structure) is used for the intake in a stream flow (not needed for
an intake in still water reservoir). The main function of the blunt body (Fig. 43) is to
eliminate or shift the separation vortices further downstream so that the sediment
particles otherwise lifted up by the separation vortices can’t be entrained to the intake.
Hence, the entrainment of the highly sediment laden flow of the lower layers is avoided
and sediment content of the diverted water is considerably reduced. The blunt body is
affixed (i.e. welded) across the vertical height of the intake pipe (Fig. 43) and aligned in
the stream flow direction. The blunt body can be empty or solid, and preferably its top
and bottom are closed. The blunt body must start before the flow separation point on the
pipe and must be non-erodible and sufficiently strong against the impact forces under
the water. For simplicity, the blunt body can be made of sufficiently strong stainless
steel sheet.

5.2.1. Multiple intakes

Experimental studies of Yıldırım (1985) have shown that if the intakes are placed within
the influence region of one another’s wake flow, the wake vortices of the adjacent
intake structure displace and lift the sediment particles from the bed and lower layers to
the upper layers and eventually a lot of sediment (much more than that of a single intake
under normal conditions) is entrained to the intake subjected to the effects of wake
vortices. Similarly, the intakes shouldn’t be placed at downstream of the structures
causing vortices (i.e. bridge piers). The distance between the intakes must be kept
sufficiently large so that the wake vortices and their effects are completely dampened
out before they reach out the other intake. Therefore, in a good design, the intakes must
be placed out of wakes of one another in such a way that one intake isn’t influenced by
the other intake’s wake flow (Fig. 44).

452
 Sediment Control Structures for Headworks and Intakes

very

Figure 44. Positions of multiple pipe intakes

5.2.2. Dimensions of the disk and the blunt-body

Since the sediment entrainment by means of separation vortices occurring in
downstream -half of the intake is much more than those entering from the upstream-
half of the intake, the flow entering from the downstream of the intake is essential in the
computation of the dimensions of both the disk and the blunt structure. Let Df diameter
of the disk and L length of the blunt body (Fig. 45).

As an approximate method, the potential solution is used. The flow towards a pipe
intake is assumed to be a point sink. The stream flow about the intake is approximated
as a uniform flow. The combination of a point sink and a uniform flow cause a dividing
stream surface (or capturing surface) which is well known as the Rankine-half-body of
revolution or the Rankine ovoid [Yuan , 1967].

The flow within the boundaries of the Rankine ovoid enters the intake. The flow outside
the Rankine ovoid can not enter the intake. In Fig. 45, the flow at downstream of the
Rankine stagnation point can’t enter the intake. This suggests that the length of the blunt
body, L, and the diameter of the disk, Df, must be greater than or equal to the distance of
the Rankine stagnation point, as, to the centre of intake which is given as [Yuan, 1967]:

Figure 45. Dimensioning the disk and the blunt-body

453
Sediment Transport Technology

1/ 2 1/ 2
Qi D ⎛V ⎞ a s 1 ⎛ Vi ⎞
as = = i ⎜⎜ i ⎟⎟ or = ⎜ ⎟ (21)
4πU ∞ 4 ⎝U∞ ⎠ Di 4 ⎜⎝ U ∞ ⎟⎠

in which Vi is the average intake flow velocity, and Di is the inner diameter of the intake.
1/ 2 1/ 2
Df a Df 1⎛ V ⎞ L a L 1 ⎛ Vi ⎞
≥ s or ≥ ⎜⎜ i ⎟⎟ and ≥ s or ≥ ⎜ ⎟⎟ (22)
Di Di Di 4 ⎝ U ∞ ⎠ Di Di Di 4 ⎜⎝ U ∞ ⎠

One should keep in mind that the above expressions are for a perfect point sink and
uniform flow combination. In reality, a pipe intake isn’t a perfect sink and the viscosity
effect always exists. Potential flow assumption underestimates as, Df and L. Therefore,
at least Di/2 must be added to as in Eqs. 21 and 22. Also, they should be multiplied by a
correction factor larger than unity. Especially the length of the blunt body should be
increased since it is supposed to reduce and shift the vortices further downstream.
Laboratory experiments of Yıldırım (1985) have shown that the correction factor of 2 is
sufficient. Hence, the recommended sizes of the disk and the blunt body are calculated
from the following formula.

L Df ⎡1 ⎛ V ⎞
1/ 2
⎤
= ≥ 2 ⎢ ⎜⎜ i ⎟⎟ + 0.5⎥ (23)
Di Di ⎢⎣ 4 ⎝ U ∞ ⎠ ⎥⎦

If the intake has a bell, the diameter of the bell should be used in lieu of Di in Eq. 23.
For common plain streams U∞ is about 0.1 – 0.8 m/s. For a very slow stream flow (U∞ is
very small) as, Df and L in Eqs. 21 and 23 become very large which is not practical. In
reality, the streams having very small velocities (i.e. say < 0.1 m/s) can’t have
considerable sediment concentration (especially non-cohesive sediment). Although the
intakes in these streams do not need a disk or a blunt body, it is advised that a disk or a
blunt body of Df/Di = L/Di ≈2 should be used.

In the case of still-water reservoirs (i.e. lakes) since there is no stream flow velocity
(U∞= 0), a blunt body isn’t needed. Experiments [Yıldırım, 1985] have shown that a
three-dimensional 360°-strongly rotating fluid flow occurs all around the intake across
the vertical distance of the intake level to the bottom of the reservoir (clearance of the
intake). If the clearance of the intake isn’t sufficiently large, this rotating water body or
vortex dislocates the bottom sediment, elevates and carries it to the intake by means of
its high vertical velocity component (Fig. 46).

This suggests that the circulation of the flow about the intake must be eliminated or at
least be weakened. For this purpose several measures can be taken (Fig. 46).

As it was mentioned for the intakes in very slow stream flows (i.e. U∞ ≤ 0.1 m/s), not to
have very large disks, the diameter of the disk for the intake in a still-water is suggested
to be about;

Df
≥ 2−4 (24)
Di

454
 Sediment Control Structures for Headworks and Intakes

Figure 46. Pipe intake in a still-water reservoir

6. Settling Basins
It is important to remove the sediment by the most inexpensive method available.
Despite the all measures taken to avoid the sediment entrainment to the intake, still
some unwanted sediment will be present in the diverted water. As it was mentioned
earlier, sediment content of the diverted water must be resolved at and about the very
beginning of the diversion canal (supply line). Otherwise, it is very difficult and
uneconomic to tangle with the sediment problems along the diversion canal and at the
place where the diverted water is utilized. Since the type of the intake is carefully
chosen and its operation is well conducted, the diverted water sediment content is
already kept very low in comparison to the sediment load of the stream flow. Although
the diverted water contains very less sediment (say about 10%-20%) it is still very large
amount to deal with it in practice. Therefore, very large percentage of the sediment load
must be removed from the diverted water right after the intake before its further journey
along the diversion (supply) line. The suspended sediment from the diverted water is
generally given back to the stream at downstream of the intake by flushing or other
means. In order to remove most of the sediment from the diverted water, generally a
stilling structure which is called the settling basin is constructing right after some
downstream distance from the intake as shown in Fig. 23.

The diverted water coming from the intake is directly given and passed through the
settling basin in which the velocity of the diverted water decreases and most of the
sediment particles larger than the permissible particle are settled. The permissible
particle is the largest particle allowed to be present in the diverted water leaving the
settling basin. The size of the permissible particle depends on the purpose of usage of

455
Sediment Transport Technology

the diverted water. For example, the permissible particle size for irrigation water can be
larger than that for the water of domestic usage. After dropping very large percentage of
its sediment load (say about 85%) in the settling basin, the diverted water leaves the
settling basin and continues on its way in the diversion system towards the place where
it is used or further treated. The maximum size of the sediment particle present in the
diverted water leaving the settling basin can’t be larger than the size of the permissible
particle. For example, for power production purpose, if the diverted water is passed
through the turbines, the sediment particles larger than about 0.25 mm can wear the
turbine off and make it out of the service: This means, the sediment particles larger than
the permissible particle (here it is given as 0.25 mm) must be settled and kept in the
settling basin before the diverted water is sent to the turbine. It is important to note that
there should be a spillway of sufficient dimensions and capacity provided on the side of
the settling basin as close to the main stream as possible in order to avoid the sudden
overflowing and flooding which may somehow occur during the operation of the whole
diversion system. More details about the settling basins are given in the next chapter.

If the sediment particles are so fine (colloidal) which can’t settle due to very small
turbulence and buoyancy forces, some chemicals increasing the cohesion forces are
added to the water in the settling basin and some agitation is introduced by mechanical
means or so. As a result of collision, the fine sediment particles get together. This event
is called the coagulation due to which the weight of the coagulated material increases
and the settlement of the fine material are made possible. There is no single rule or
criterion for the classification of the settling basins. Some researchers [Çeçen, 1995]
have classified the settling basins by considering their geometry (rectangular, circular),
the method of flushing (continuous or intermitted flushing or mechanical removal) and
the flushing direction (in the same or opposite directions of the flow, transverse flushing
and flushing by funnels located at the bottom etc.).

6.1. Location of the settling basin

In order to eliminate the sediment related problems at the very beginning of diversion
line in an economical way, the settling basin is generally constructed at a suitable place
sufficiently and safely close to the stream bank at some downstream distance from the
intake. To make the flushing and removal of the settled sediment in the settling basin
more possible and effective, the designer must be aware of the danger of the flooding,
erosion of the stream bank, stability of the hill side and the bank (land slide), rock falls,
mud and material washed out from the hill sides. The accessibility, construction
operation and maintenance conditions have to be considered.

6.2. Flushing of settling basins

The settling basins are basically designed for the settlement of the fine sediment. A
properly designed settling basin can be flushed in a short period of time with reasonably
small amount of water. As for the high slope streams (mountain streams), the diverted
water may carry coarse sediment, where flushing may not be that easy once they settle
at the beginning of the settling basin. Their flushing may require a lot of loss of water.
In such a case, it is better to construct either a pre-flushing system such as a vortex tube
settling basin [Çeçen, 1977] or some other means for the removal of the coarse sediment
between the intake and the settling basin.

456
 Sediment Control Structures for Headworks and Intakes

Flushing systems are located at the bottom of the settling basins. The most common
flushing methods used for the flushing of the settling basins are as follows [Çeçen,
1995]:
i) Flushing in the flow direction (Dufour system) (Fig. 47)
ii) Flushing by funnels located at the bottom (Fig. 48)
iii) Others (transverse, bottom-end etc. Figs. 49, 50 and 51)

Figure 47. Dufour system settling basin

Figure 48. Flushing the bottom of settling basin [ICID, 1975]

Figure 49. Flushing against flow direction [Samarin, et al., 1961]

457
Sediment Transport Technology

Figure 50. Bottom flushing [Samarin et al., 1961]

Figure 51. Flushing from the very bottom-end of the settling basin

Details of the design of a settling basin are explained in the next chapter.

6.3. Pre-flushing of coarse sediment entering the settling basin

In some cases, considerable coarse bed sediment unavoidably enters the intake and be
present in the diverted water. If the diverted water is given to the settling basin, the
coarse sediment particles clog the racks, increase the turbulence and settle at the
entrance region of the settling basin which makes the flushing difficult, causes loss of
excessive water for flushing and reduces the efficiency of the settling basin. Therefore,
it is necessary to separate the unwanted coarse sediment from the diverted water before
it enters the settling basin. For this purpose, a flushing arrangement at or just after the
intake before the settling basin which is called the pre-flushing is made. The pre-
flushing can be intermittent or continuous. The most common pre-flushing arrangements
used in practice are the frontal flushing, the floor flushing, the vortex tube and the bath
tube sediment extractors which are explained below.

458
 Sediment Control Structures for Headworks and Intakes

6.3.1. Frontal flushing

The coarse sediment at the bottom is flushed by diverting the bottom flow through an
opening with or without a gate. Important thing is that the measures (from both
structural and operation point of views) must be taken in order to eliminate or reduce
the turbulence which can otherwise lift the coarse bed sediment and be present in the
diverted water on its way to the settling basin. If the turbulence is not dampened, the
function of the pre-flushing is nullified. The flushed sediment is given back to the
stream. A good prototype example of the frontal intake utilized in Maden-Turkey
[Çeçen, 1977] which is in service for several years is presented in Fig. 52.

6.3.2. Floor flushing

In this method of flushing, the coarse bed sediment is sucked through an opening(s) of
sufficient dimension placed at the bottom floor. In order to separate the high-
concentration (coarse sediment) bottom flow from the upper flow of relatively low-
concentration, a dividing plate or slab of sufficient dimensions is placed between the
bottom and the upper flow layers. The flushed sediment is given back to the stream after
some downstream distance from the flushing structure. A prototype example of the
bottom flushing utilized at the Dere plant- Turkey is given in Fig. 53.

6.3.3. Vortex-tube ejector

This type of coarse sediment ejector basically consists of an open-top tube or channel as
shown in Fig. 54. The open-top tube is installed at an angle of 45º to the axis of the flow
in the headwork channel [ASCE, 1975]. The edge or lip of the tube is set level with the
bottom grade of the canal whether it is lined or unlined. Water flowing the over the
opening induces a spiraling flow in the tube throughout its length. The spiral flow picks
up the bed load and moves it along the tube to an outlet at the downstream end of the
structure [ASCE, 1975]. The 10%-15% of the stream water is used for sluicing through

Figure 52. Project details of the frontal intake constructed at Maden-Turkey [Çeçen, 1995]

459
Sediment Transport Technology

Figure 53. Floor flushing at Dere Plant (Turkey) [Çeçen, 1995]

Figure 54. Vortex tube sediment ejector [ASCE, 1975]

the vortex tube. The velocity of the flow within the vortex tube must be sufficiently
high in order to prevent the deposition of the sediment in it. For the bed load of small
particles, the size of the vortex tube slot should be narrower to prevent excessive waste
of water.

The field experiments of Ahmad (1960, 1963) [reported by ASCE, 1975] have proven
that the particle size of the bed load has an appreciable influence on the slot width.

460
 Sediment Control Structures for Headworks and Intakes

6.3.4. Bath-tube vortex ejector

The bath-tube vortex extractor is developed by Çeçen in the Technical University of
Istanbul-Turkey [Çeçen, 1973, 1975]. The main working principle of the bath-tube
vortex extractor is the same as flow in a bend of a stream flow. A circulation is
introduced the stream flow in a circular basin which has an orifice of sufficient
dimension at its centre (Fig.55).

Due to the viscosity (boundary) effect, the flow velocity close to the bottom boundary is
very small in comparison to the velocity close to the water surface in the basin. Due to
the imposed circulation on the stream flow in the basin, water elements are subjected to
the centrifugal forces. Since the water particles close to the free surface have larger
tangential velocities in comparison to the fluid particles in the regions close to the
bottom, they receive higher centrifugal forces. Exactly due to the same reasons
explained for the curve-flow in the bends, the stream flow depth decreases towards the
centre and increases outwards. As a result, because of the pressure difference, a
secondary flow occurs from outside towards the centre of the basin. The secondary flow
sends the coarse bed sediment from the outer boundary towards the orifice at the very
centre of the circular basin. Hence, the coarse bed sediment is flushed out through the
orifice opening. The flushing must be continuous for the bath-tube extractor. Since
higher the tangential velocity, stronger is the centrifugal force, higher is the depth
difference between the water surfaces at the periphery and the centre, stronger is the
secondary flow and more effective are the movement of the coarse bed material towards
the orifice and the flushing. It is seen that higher tangential velocity of the flow within
the basin is essential for the effective removal and flushing of the bed sediment. The
maximum tangential velocity is provided to the flow within the basin by the main
stream flow almost tangentially entering the basin through a suitable bottom-level
entrance of a sufficient dimension at the side periphery-wall of the basin as shown in

Figure 55. Project details of prototype bath-tube sediment ejector constructed at Sızır-Turkey
[Çeçen, 1977]

461
Sediment Transport Technology

Fig. 55. The stream flow tangentially entering the intake introduces a considerable
circulation on the flow field within the bath-tube basin.

Experiments [Çeçen, 1973, 1975] have shown that a solid particle entering the basin
follows a helicoidal path as it advances towards the orifice (Fig. 56). The length of the
helicoidal path of the sediment particle is much more than the periphery-diameter of the
basin. Experiments of Kocabaş and Yıldırım (2002) have also shown that the path of a
fluid particle traveling towards the intake (i.e. an orifice) in a still-water is also
helicoidally. The helicoidally paths of both the sediment and the fluid particles are due
to the radial velocity towards the centre (orifice), the effects of the Coriolis forces
(Earth’s rotation), imposed circulation and other irregularities present within the flow
field of the bath-tub basin.

If the depth of the flow over the orifice (the vertical distance of the orifice to the fluid
surface, which is called the submergence as explained earlier in the topic of “Small
Scale Intakes”, is less than or equal to its critical value (critical submergence), so called
the air-core vortex or air-entraining free vortex develops above the orifice by means of
which air enters the orifice. Kocabaş and Yıldırım (2002) have proven that, the
circulation imposed on the flow causes the air-entraining vortex to occur faster, stronger
and more air-entraining and the critical submergence becomes much smaller than that in
the case of no-circulation imposed. In order to avoid the loss of the excessive flushing
water, the occurrence of a strong air-entraining free vortex-core is on purpose required
to be always present in the bath-tub basin. Because, the air-core and the entrained-air
reduces the working cross-section of the orifice, decreases its discharge capacity and
hence causes less water for flushing. As it is understood, the circulation imposed on the
flow in the bath-tub basin has two useful factors. Firstly, it causes the development of a
strong secondary flow which carries the coarse bed sediment towards the orifice.

Figure 56. Bottom sediment movement in a laboratory model of bath-tube sediment ejector
[Çeçen, 1995]

462
 Sediment Control Structures for Headworks and Intakes

Secondly, it reduces the amount of water utilized for flushing by means of a strong air-
entraining vortex-core. The diverted water is taken from the layer close to free surface
just above the place of entry of the stream flow by means of a horizontal dividing plain
plate of sufficient dimensions and proper orientation installed between the stream flow
entering and leaving the bath-tube basin at the same section (Figs. 55 and 56). The
installed horizontal dividing plate a kind of skims the upper layer flow which has
completed a 360o-revolution. An interesting diversion is also developed by Salakov,
1975 as shown in Fig. 57, in which the height of the sidewall all around the basin is kept
lower than its necessary value. The upper layer of the flow which is free of coarse bed
sediment spills out and collected in a circular open canal constructed all around the
basin and then diverted. In order to make use of the weight components of coarse
particles in an effective flushing, about 2% slope is given to the bottom of the basin
towards the center (orifice).

The bath-tube vortex extractor is very economical in comparison to others and it is very
suitable as a pre-flushing device for the power generation and irrigation intakes. Its
coarse sediment separation (removal) ratio and diverted water ratio are as high as about

Figure 57. Bath-tube sediment ejector types [Paul et al., 1991]

463
Sediment Transport Technology

97%. It spends only about 3-10% of the stream flow entering the basin for flushing the
sediment through the orifice. As it is seen the bath-tub basin is very efficient. To this
date, several prototype bath-tub extractors are realized and successful in service for
several years through out the world as presented in the Fig. 57. Paul et al. (1989) have
experimentally shown that the optimum peripheral-diameter of the bath-tube basin is 5
times the width of the stream flow canal tangentially entering the bath-tube basin.

The prototype bath-tube extractor built in the Sızır hydropower plant (Fig. 58) (Sivas-
Turkey) is a preliminary flushing device to remove the coarse sediment before the main
settling-basin. The characteristic prototype data are as follows (also see Fig. 55). The
diameter D = 15 m, discharge Q = 5.25 m3/s, the width of the inlet canal = 1.75 m,
bottom slope = 0.02, and the diameter of the orifice = 60 cm. It utilizes only 3% of the
total discharge for preliminary flushing. The plant works very well.

Figure 58. Picture of the prototype bath-tub sediment ejector at Sızır-Turkey [Çeçen, 1995]

References

1. Ahmad, M., Ali, M., and Kaliq, A., “Sediment Exclusion Methods and Devices at
the Intake of Canals”, Proceedings of the West Pakistan Engineering Congress,
Paper No. 34, Vol.44, Part I, (1960).
2. Ahmad, M., “Design of Silt Excluders and Ejectors”, West Pakistan Engineering
Congress (Golden Jubilee Publication), (1963).
3. ASCE, Task Committee, “Sedimentation Engineering”, American Society of Civil
Engineers, New York, (1975).
4. Bayazıt, M., “Hareketli Tabanlı Akımların Hidroliği”, İstanbul Teknik Üniversitesi
İnşaat Fakültesi Matbaası, İstanbul, (1971).

464
 Sediment Control Structures for Headworks and Intakes

5. Beverege, J. P., and Culbertson, J. K., “Hyperconcentrations of Suspended

Sediment”, Journal of the Hydraulic Division, ASCE, Vol.90, No.HY6, Proc.
Paper, 117-128, (1964).
6. Bulle, H., “Untersuchungen Über Die Geschiebeableitung bei der Spaltung von
Wasserlaufen.”, Diss. T.H., Karlsruhe, (1926).
7. Camp, T. R., “Sedimentation and the Design of Settling Tanks.” Trans. ASCE, 111,
(1946).
8. Çeçen, K., “Water Intake Problems in Torrents”, IAHR, Bangkok, (1973).
9. Çeçen, K., “Frontal Intake Structures to Control the Sediment Entrainment”, ICID,
Moscow, (1975).
10. Çeçen, K., and Garbrecht, G., “Über Die Wasserfassung aus Geschiebeführenden
Flüssen“, Die Wasserwirtchaft, (1957).
11. Çeçen, K., “Die Wasserfassung bei Gebirgflüssen“, Die Wasserwirtchaft, (1967).
12. Çeçen, K., “Die Verhinderung Des Geschiebeeinlaufes zu Wasserfassungsanlagen“,
Versuchsanstalt für Wasserbau, Bericht 35, T.U., München, (1977).
13. Çeçen, K., “Hydraulic Criteria of Settling Basins for Water Treatment, Hydropower
and Irrigation”, Proc. 17th Congress of the Int. Assoc. of Hydr. Res., Baden-Baden,
West Germany, 275-294, (1977).
14. Çeçen, K., “Sediment Control Facilities for Headwork and Intakes”, Post-Graduate
Course in Sediment Transport Technology, Proceedings Vol.I, DSI, Ankara,
(1995).
15. Çeçen, K., and Bayazit, M., “Some Laboratory Studies of Sediment Controlling
Structures”, Proc. of 9th Congress of ICID, Moscow, Soviet Union, 107-111,
(1975).
16. Danelia, N. F., and Kutateladze, N. G., “Sediment Control in Water Intakes of
Piedmont and Mountain Areas”, ICID, Moscow, (1975).
17. Denny, D. F., “An experimental Study of Air-Entraining Vortcies in Pump Sumps”,
Proc., Inst. of Mech. Engrs., London, 170(2), 106-116, (1956).
18. Eroğlu, N., and Bahadırlı, T., “Prediction of Critical Submergence for a Rectangular
Intake”, Journal of Energy Engineering, ASCE, Vol.133 (2), (2007).
19. Ferrese, R., and Pemberton, E. L., “Design Consideration to Control Diversion of
Sediment at Intake Structures for Main Canals”, ICID, R. 28, Moscow, (1975).
20. Grishin, M. M., “Hydraulic Structures”, Vol.2, Mir Publishers, Moscow, (1987).
21. Kocabaş, F., and Yıldırım, N., “Effect of Circulation on Critical Submergence of an
Intake Pipe”, Journal of Hydraulic Research, IAHR, Vol.40, No.6, 741-752,
(2002).
22. Li, R. M., “Solid Particle Settlement in Open-Channel Flow”, Journal of the
Hydraulic Division, ASCE, Vol.101, No.HY7, (1975).
23. Mashauri, D. A., “Modelling of Vortex Settling Basin for Primary Clarification of
Water”, thesis presented at the Tampere University of Technology, at Tampere,
Finland, in partial fulfillment of the requirements for the degree of Doctor of
Technology, (1986).
24. Melone, A. M., Richardson, E. V., Simson, D. B., “Exclusion and Ejection of
Sediment from Canals”, Dept. Civil Engrg., Colorado State Univ., Fort Collins,
Colorado, (1975).
25. Midgley, D. C., “Abstraction of Water from Sediment Laden Rivers in Southern
Africa”, Versuchsanstalt für Wasserbau, Bericht, 35, T.U. München, (1977).
26. Müller, R., “Wasserfassungen in Geschiebeführenden Flüssen”, Wasser-und
Energiewirtschraft, Nr. 9, (1955).

465
Sediment Transport Technology

27. Ogihara, K., and Sakaguchi, S., “New Systems to Separate the Sediments from the
Water Flow by Using the Rotating Flow”, Proc. Fourth Congress of the Asian and
Pasific Division, Int. Assoc. of Hydr. Res., Chiang Mai, Thailand, 753-766, (1984).
28. Paul, T. C., Sayal, S. K., Sakhuja, V. S., and Dhillon, G. S., “Vortex-Settling Basin
Design Considerations”, Journal of Hydraulic Engineering, ASCE, Vol.117, No.2,
172-189, (1991).
29. Raudkivi, A. J., “Loose Boundary Hydraulics”, Pergamon Press Publication,
(1967).
30. Salakhov, F. S., “Rotational Designs and Methods of Hydraulic Calculation of
Load-Controlling Water Intake Structures for Mountain Rivers”, Proc. of 9th
Congress of the ICID, Moscow, Soviet Union, 151-161, (1975).
31. Samarin, E. A., Popow, K. W. U., Fandejew, W. W., “Wasserbau”, VEB Verlag für
Bauwesen, Berlin, (1961).
32. Scheuerlein, H., “Die Wasserentnahme aus Geschiebeführenden Flüssen”, Ernst
und Sohn, Berlin, (1984).
33. Yıldırım, N., Kocabaş, F., and Gülcan, S., “Flow-Boundary Effects on Critical
Submergence of an Intake Pipe”, Journal of Hydraulic Engineering, ASCE, Vol.
126, No.4, 288-297, (2000).
34. Yıldırım, N., Kocabaş, F., and Gülcan, S., “Errata on “Flow-Boundary Effects on
Critical Submergence of an Intake Pipe”, Journal of Hydraulic Engineering,
ASCE, Vol.133 (4), (2007).
35. Yıldırım, N., “Critical Submergence for a Rectangular Intake”, Journal of
Engineering Mechanics, ASCE, Vol. 130, No.10, 1195-1210. 288-297, (2004).
36. Yıldırım, N., “Nehirlerdeki Bağlamasız Küçük Su Alma Ağzının Hava ve Katı
Madde Bakımından Tahkiki (Sakarya Su Alma Ağız Tipi)”, DSİ Teknik Bülteni,
Sayı, 57, (1985).
37. Yuan, S. W., “Foundation of fluid mechanics”, Prentice-Hall Inc., Englewood
Cliffs, N.J., (1967).

466
 DESIGN OF INTAKES AND SETTLING BASINS

Prof. Dr. Nevzat YILDIRIM

Department of Civil Engineering
Gazi University
06570 Maltepe Ankara - Turkey

1. Introduction
As it was explained in the previous chapter it is almost impossible to divert sediment-
free water from a stream carrying sediment. To have the least amount and size of
sediment in the diverted water, a properly designed and operated intake and other
affiliated structures are constructed. Depending on the stream flow and local conditions,
the type of the intake is chosen. The type and properties of an intake used in plain
streams are different than those used in the mountain streams. The types of intakes were
explained in the previous chapter. Here, the settling basins are mainly considered.

In this course, the settling basins utilized in hydropower and irrigation water diversion
systems are studied only. The settling basins used in the domestic water diversion
systems are out of the scope of this course since they require special attention and
processes such as the removal of the colloidal material by means of coagulation etc.

The sediment particles carried by the diverted water can erode and wear off the
penstock and steel parts of the machines such as the runner vanes of the turbines or
pumps, lining or the construction material of the boundaries of the canals and tunnels
through which the diverted water passes. Hence, the sediment particles can very shorten
the useful life of these expensive structures.

The erosion or the wear-off of the materials just mentioned is also known as the
abrasion. The experiences in practice have shown that;
• The abrasion increases with increasing particle size.
• The abrasion rapidly increases when the particle hardness exceeds that of the surface
material being abraded.
• The angular or sharp-edged particles cause greater abrasion than that of rounded
particles.
• The abrasion increases with concentration.
• The flow parallel to the solid boundary causes less abrasion.
• The particles having high velocity (kinetic energy) or head cause greater abrasion
than the slow particles.

In practice, it has been witnessed that the abrasion caused by sharp-edged coarse
sediment on a Pelton-wheel is about 6-7 mm [Mosonyi, 1965]. Due to the abrasion, the
efficiency of a turbine dramatically drops. As an example, the abrasion on the Francis-
turbine of the Doğankent hydropower plant (Turkey) operating under the head of 192 m
after the 8000 hours –operation is presented in Fig. 1 [Avcı, 1995].

1-467
Sediment Transport Technology

Figure 1. The eroded parts of a Francis turbine in the Doğankent Hydro-Power Plant- Turkey
[Avcı, 1995]

There are several examples of ruined turbines or pumps due to the sediment-particles in
the diverted water [ICID, 1975]. The importance of the removal of the sediment
particles within the diverted water is now more clear for two significant reasons besides
others. These are:
i) Reduction of the abrasion of the expensive parts of the machinery and
diversion canal material.
ii) Reduction of the accumulation of sediment at the place of destination of
diverted water.

As it was mentioned in the previous chapter, the main function of a settling-basin is to

reduce the velocity of the diverted water and let the sediment particles larger than or
equal to the permissible particle to settle within the basin.

The practical experiences have shown that the settling basins are necessary when the
sediment content exceeds 0.5 kg/m3 or when the amount of fractions of the hazardous
sediment isn’t less than 0.2 kg/m3. The quartz particles larger than 0.25 mm and softer
particles larger than 0.4 mm in size are considered to be hazardous. The approximate
size of the permissible particles varies with the pressure head for the hydropower plant
systems as shown in Table 1.

In some cases, if the head is several 100 m, the sizes of the permissible sharp-edged
quartz particles not to have abrasion of the mechanical parts may be as small as between
0.02-0.002 mm. In such cases the design of settling basin becomes a very difficult
problem [Mosonyi, 1965, Camp, 1946, Avcı, 1995]. The characteristics of a typical
settling basin are shown in Fig. 2.

Table 1. [Çeçen, 1976 reported by Avcı, 1995]

Pressure head at the plant (m) Permissible particle size (mm)

80-100 Settling may be necessary
100-200 0.6
200-300 0.5
300-500 0.3
500-1000 0.1

1-468
 Design of Intakes and Settling Basins

Figure 2. A settling basin constructed in form of multiple cells [Avcı, 1995]

As it is indicated in Fig. 2, a settling basin generally consists of three basic sections

which are as follows:
i) The inlet section
ii) Transition section
iii) The settling basin section

The inlet section consists of a sill, coarse rack and the inlet canal. The transition section
includes the entrance control gate and smooth transition from the inlet canal cross-
section to that of the settling basin.

The settling basin has the following parts:

• Screens
• Energy breaking rods for stilling
• Main settling zone
• Flushing canal, sluice gates.

The screens hold the debris and larger particles somehow present in the diverted water.
The energy breaking rods are placed alternately in order to break the excessive energy

1-469
Sediment Transport Technology

of the flow so as to make the flow as free of turbulence as possible for the effective
settlement of the particles in the settling-basin. The screen and the energy breaking rods
take place at the beginning of the main settling zone. The main settling zone is the
useful part of the basin in which particles settle. The computation of the dimensions of
the main settling zone is presented later. In order to make the flushing of the settled
sediment effective, both longitudinal and transverse slopes are given to the bottom floor
of the basin and also a longitudinal small canal (cunette) along the centerline of the
bottom floor is constructed (Fig. 2) to make the flushing possible when the depth of
water in the basin is very small. At the very end of the main settling zone, flushing canal
and sluice-gate(s) takes place. The flushing can be continuous or intermitted. In some
cases the cleaning of the settling sediment can be done by mechanical means (scrapers,
excavating equipments etc.) As it was mentioned before, the flushing period is
determined during the operation. The flushed sediment is given back to the main stream.

2. Design Considerations for Settling Basins

Generally, the settling basins of rectangular cross-sections are constructed. The settling
basin is constructed in the form of longitudinal multiple cells (or compartments) of
sufficient dimensions and number as shown in Fig. 3.

For the sake of the continuous operation and uninterrupted service, the cells are
constructed in double number of the needed. While the half number of the cells work
the other half are being flushed or cleaned. The other advantage of the basin constructed
in number of cells is to create a uniform flow free of rotations and turbulence in each
cell and increase the efficiency of the settling basin. Both flows approaching to and
within the basin must be straight without any considerable curvature in order not to
have secondary flow and turbulence (Fig. 4). The approaching flow canals shouldn’t
also have curvature.

The flow within the settling-basin should be as free of secondary flow and turbulence as
possible. The flow shouldn't enter the basin in the form of a jet. In such a case, a high
velocity flow and vortices occur about the centerline and dead-zones, respectively. This
phenomenon is known as the hydraulically short circuit which reduces the efficiency of
the basin. Flow should uniformly enter the basin as shown in Fig. 5.

Figure 3. The main elements of a settling basin (Avcı, 1995)

2-470
 Design of Intakes and Settling Basins

Figure 4. Hydraulically incorrect design of a settling basin

[Mosonyi, 1965, reported by Avcı, 1995]

Figure 5. Design of a settling basin [Linsly and Farazini, 1972]

In a shallow basin, the deposition of the sediment occurs in a short distance and fills up
the basin in a short period of time. The depth and the width of the basin should be
reasonable. The practical experiences have shown that for economical designs the flow
velocity within the basin is about 0.4-0.6m/sec.

Outlet weirs and sluice gates must properly be operated to keep the flow depth constant
within the basin. The flow in outlet flushing canal shouldn't enter the basin when the
water level in the basin is minimum (i.e. zero).

Studies of Çeçen (1977) have proven that the higher weirs shouldn't be used. Because,
they cause undesirable vertical and horizontal vortices that reduce the effective length of
the basin.

In the hydraulic design of the settling basins the following points are considered [Avcı,
1995]:

2-471
Sediment Transport Technology

1. The characteristics of the sediment carried by the stream (concentration distribution,

size distributions and specific weights of suspended and bed sediment) should be
known in details. Bed load sediment is generally avoided by the well designed
intake, sills and flushing canals. The abrasion (wear off) of the mechanical devices
is generally due to the impingements of the suspended sediment particles.
Therefore, the suspended sediment is more concerned. The suspended sediment
characteristics highly depend on the regime of the stream (discharges), the section
and course of the stream. In mountain streams, the concentration of the suspended
sediment varies between 2 to 10 kg/m3, during the floods it may be about 50-60
kg/m3. In the plain streams the suspended sediment concentration is about 0.1-1
kg/m3, during the floods it may be about 5-10kg/m3. In some rivers these values can
be much greater. For example, in the Yellow River- China, sediment concentration
varies between 30 kg/m3 and 500 kg/m3 [Avcı, 1995].
2. The sediment load removal ratio or sediment removal efficiency of the settling
basin has to be determined. The sediment removal efficiency is defined as the ratio
of concentrations after and before settling as presented below:

CP
× 100 (percent) (1)
C

in which, C is the concentration of the diverted water entering the settling basin
(concentration of the inflow) and CP is the concentration of the diverted water
leaving the settling basin (concentration of the outflow). It is obvious that the
removal ratio depends on the flow and geometrical conditions and operation
schedule of the settling basin. Since the particles larger than the permissible particle
(Table 1) settle in the basin, if the gradation curve of the suspended sediment is
known, the removal ratio can directly be read off from the gradation curve for the
determined permissible particle size (some times it is also known as limit particle
size).
3. The settling velocity (fall velocity) of the permissible particle allowed to be present
in the diverted water leaving the settling basin is determined either by computations
or by tests. It is assumed that the size of the particle doesn't change as it goes
through the settling basin.

3. Computation of Dimensions of the Settling Basin

For a simple solution, generally a rectangular settling basin of horizontal bottom is
considered. Let B the total net width of the basin (sum of the net widths of the cells), H
the depth of the basin, L the useful length of the main settling zone, Q total discharge of
the diverted water passing through the settling basin, V uniform flow velocity in the
basin, w settling (fall) velocity of the permissible sediment particle in still-water, t time,
x horizontal axis and y vertical downward axis as shown in Fig. 6.

Q =V ⋅B⋅H (2)
Q q Q
V = = ; q= (3)
B⋅H H B

3-472
 Design of Intakes and Settling Basins

Figure 6. Determination of length of the settling basin

It is assumed that the horizontal velocity of the sediment particle is the same as that of
the flow. The particle vertically drops down under its weight with its settling (fall)
velocity of w. During the same time of t, the particle takes both the horizontal distance
of x=V·t and the vertical distance of y=w·t. The path of the sediment particle within the
basin will be the one as shown in Fig. 6. Elimination of t between x and y gives,

x V
= (4)
y w

As x takes the value of L, y becomes H. Hence, x = L, y = H. Thus, Eq. 4 can be

rewritten as:

L V V
= or L≥ H (5)
H w w

Notice that in the derivation of the above equations, the vertical velocity of the sediment
particle is taken as equal to its fall velocity (w) in a still-water. In reality, although all
aforementioned precautions are taken to eliminate the unwanted turbulence within the
settling basin, still there is some inevitable turbulence in the flow within the settling
basin (think about the magnitude of the Reynolds number due to large dimensions of the
settling basin). The vertical upward velocity component of the turbulent flow causes a
vertical upward force on the settling particle. This means, the sediment particle can’t
settle down with w. It settles down with a velocity smaller than w. The reduction in w is

3-473
Sediment Transport Technology

due to the vertical upward velocity component w′ of the turbulent flow. Hence the
particle settles down with the net vertical velocity of (w- w′ ). Therefore, for the real
case, w must be replaced with (w- w′ ) in the above equalities. Eq. 5 can be rewritten as;

V
L≥ H (6)
( w − w′)

There are several experimental studies for the prediction of w′ . For example, Mosonyi,
1965 gives;

0.132
w′ = V (7)
H

in which the flow depth H is in terms of m. From Eqs. 6 and 7, one finds the net useful
length of the main settling basin zone as,

VH 3 / 2
L≥ (8)
wH 1 / 2 − 0.132V

The settling velocities (w) of the spherical quartz sediment particles are given in Fig. 7.
The uniform flow velocity V in the settling basin shouldn’t be larger than the maximum
permissible velocity (or critical velocity) for a given sediment. Otherwise the settled bed
sediment can be moved and become suspended again which is no good. The maximum
permissible velocities for loose granular bed sediment (quartz) have been determined by
Mosonyi, 1965 and presented in Table 2.

Figure 7. Fall velocity of quartz sediment particle in a silty still-water of different density
[Mosonyi, 1965, Reported by Avcı, 1995]

3-474
 Design of Intakes and Settling Basins

Table 2. Maximum permissible mean velocities for (V in the settling basin) loose granular bed
material [Mosonyi, 1965, reported by Avcı, 1995]

Diameter of Maximum mean velocity in

particle case of h=1 m
Material
d V1
(mm) (m/s)
Very coarse gravel 200-150 3.9-3.3
150-100 3.3-2.7
Coarse gravel 100-75 2.7-2.1
75-50 2.4-1.9
50-25 1.9-1.4
25-15 1.4-1.2
15-10 1.2-1.0
10-5 1.0-0.8
Cobble 5-2 0.8-0.6
Coarse sand 2-0.5 0.6-0.4
Fine sand 0.5-0.1 0.4-0.25
Very fine sand (Mo) 0.1-0.02 0.25-0.20
Silt 0.02-0.002 0.20-0.15

Correction coefficients to formula

V = α·V1

Depth h Correction coefficient

(m) α
0.3 0.80
0.6 0.90
1.0 1.00
1.5 1.10
2.0 1.15
2.5 1.20
3.0 1.25

By conducting experiments, Camp (1946) has found the empirical relation between the
diameter d of the sediment particle and the maximum permissible velocity V (in the
settling pool) which is as follows:

V =a d (9)

in this equation d is in terms of mm and V is cm/sec. The values of the coefficient a are
as given below:

For d > 1 mm; a = 36 is taken.

For 0.1mm < d < 1mm; a = 44 is taken. (10)
For d < 0.1mm; a = 51 is taken.

3-475
Sediment Transport Technology

Generally, increasing the width of the basin is economical than increasing the depth of
the flow in the basin. In the above equations it is clear that the number of unknowns is
more than the number of equations. This indicates that some of the quantities must
reasonably be chosen at the beginning of the solution.

Practical applications have proven that the net depth of each cell or comportment of a
settling basin is between 1.5 m and 4 m, the net width of the cell is between 2 m and 6
m. As it is seen in Eq. 8, if 0.132V>wH1/2, the denominator of the Eq. 8 becomes
negative (or L becomes negative). This means that for the conditions satisfying
0.132V>wH1/2, the particle doesn’t settle within the settling basin. Therefore, flow
conditions satisfying 0.132V<wH1/2 must be searched until the solution is found.
Examine the solved example given below.

Example
A settling basin for a hydraulic power unit is to be designed. The diverted water
discharge is 10 m3/sec. The quartz sediment particles larger than 0.5 mm is not wanted
to be present in the diverted water at the outlet of the settling basin. The density of the
stream water is about 1.064 tons/m3.

Solution
The permissible particle diameter is given as d = 0.5 mm. Since 0.1 mm < d = 0.5 mm <
1 mm, from Eq. 10 a = 44 is taken. From Eq. 9, the permissible uniform flow velocity in
the settling basin is V = a d = 44 0.5 mm = 31.13 cm/sec. Let us choose the depth of
the flow within the settling basin as H = 3 m (experiences indicate that H is to be chosen
between 1.5 m and 4 m).

The total net width of the cells of the settling basin can be computed from continuity
equation (Eq. 2),

Q = B ⋅ H ⋅V
10 m 3 /sec
B= = 10.71 m
3 m × 0.3113

If the number of the cells is n and the net width of a cell is b, then, B = n ⋅ b . In practice
the net width of a cell is advised to be between 2 m and 6 m. Let us choose b = 4 m.
Thus,

B 10.71
n= = = 2.68 cells ≈ 3 cells
b 4

The new value of B = 3 cells x 4 m = 12 m (not 10.71 m). For a continuous operation
and proper maintenance, the number of the cells to be constructed is twice the computed
value. As the 3 cells work other 3 cells are being flushed and cleaned. Hence, the
number of cells to be constructed = 2n = 2 x 3 = 6 cells, each has a net width of 4 m and
the depth of 3 m. Therefore, the total net width of the settling-basin to be constructed is
BT = 4 m x 6cells = 24 m.

3-476
 Design of Intakes and Settling Basins

The new velocity of the uniform flow in the settling basin is,

Q 10
V = = = 0.277 m/sec < 31.13 cm/sec
BH 12 * 3 m

From Fig. 7, the fall (settling) velocity of the quartz particle of d = 0.5 mm. (in water of
density of 1.064 tons/m3) is read as w = 6 cm/sec = 0.06 m/sec.

From Eq. 8, the net useful length of the main settling-zone is as follows:

33 / 2 (0.277 )
L= = 21.37 m
(0.06)31 / 2 − 0.132(0.277 )
L 21.37
The settling time = = = 77.14 sec
V 0.277

Also, there are statistical methods for the prediction of the dimensions of the settling
basin [Velikanow, Reported by Mosonyi, 1965, Sümer, 1977]. Since the statistical
methods generally well over estimate the dimensions of the settling basin, they are not
explained here. If one is interested in, he should look it up in the references provided at
the end of this chapter.

4. Hydraulic Modeling of Settling Basins

For important projects, experimental studies on the hydraulic models of the settling
basins are strongly advised. The Froude similarity is utilized for both the flow velocity
and settling (fall) velocity of the sediment [Çeçen, 1977]. The model scale should be as
large as possible for a better simulation. The distorted models are not advised for the
settling basins [Avcı, 1995].

References

1. ASCE, Task Committee, “Sedimentation Engineering”, American Society of Civil

Engineers, New York, (1975).
2. Avcı, I., “Design of Intakes and Settling Basin”, Post-Graduate Course in
Sediment Transport Technology, Proceedings Vol. I, DSI, Ankara, (1995).
3. Bayazıt, M., “Hareketli Tabanlı Akımların Hidroliği”, İstanbul Teknik Üniversitesi
İnşaat Fakültesi Matbaası, İstanbul, (1971).
4. Camp, T. R., “Sedimentation and the Design of Settling Tanks”, Trans. ASCE, 111,
(1946).
5. Çeçen, K., “Frontal Intake Structures to Control the Sediment Entrainment”, ICID,
Moscow, (1975).
6. Çeçen, K., “Hydraulic Criteria of Settling Basins for Water Treatment, Hydropower
and Irrigation”, Proc. 17th Congress of the Int. Assoc. of Hydr. Res., Baden-Baden,
West Germany, 275-294, (1977).
7. Çeçen, K., “Su Kuvvetleri (Hydropower)”, İstanbul Teknik Üniversitesi İnşaat
Fakültesi Matbaası, İstanbul, (1976).

4-477
Sediment Transport Technology

8. Çeçen, K., “Sediment Control Facilities for Headwork and Intakes”, Post-Graduate
Course in Sediment Transport Technology, Proceedings Vol. I, DSI, Ankara,
(1995).
9. ICID, “Question 30”, Transaction of the Congress in Moscow, Vol. II, (1975).
10. Linsley, R. K., and Faranzini, J. B., “Water-Resources Engineering”, McGraw-Hill,
Inc., (1972).
11. Melone, A. M., Richardson, E. V., and Simson, D. B., “Exclusion and Ejection of
Sediment from Canals”, Dept. Civil Engrg., Colorado State Univ., Fort Collins,
Colorado, (1975).
12. Mosonyi, E., “Water Power Development II”, House of the Hungarian Academy
of Science, Budapest, (1965).
13. Sumer, B. M., “Settlement of Solid Particles in Open Channel Flow”, Journal of
Hydraulic Division, ASCE, Vol. 103, No. HY 11, (1977).

4-478
 RIVER TRAINING STRUCTURES - APPLICATIONS

Prof. Dr. İbrahim GÜRER

Department of Civil Engineering
Faculty of Engineering and Architecture
Gazi University, Ankara - Turkey
e-mail: gurer@gazi.edu.tr

1. Introduction
In its natural state, a river flows in a channel which is consistent with the characteristics
of water and channel (surroundings). Scour or build up in a river continues over a
period of time until they reach a stable equilibrium.

Along a river course there are four distinct river regions; they can be listed as:
1. Mountaineous,
2. Sub-mountaineous,
3. The alluvial, and
4. Deltaic

and the slope of river decreases along its length.

The river training structures (Photos 1-6) which will be used along the river course can
be classified into three main groups:
5. Channel constructions structures: Transverse structures and longitudinal structures
6. Bed and bank protection structures: Channel lining
7. Flow guiding structures and measures: Spurs and groynes

Photo 1. A DSI Concrete regulator at Photo 2. A drop structure during flood at

Korkuteli, Antalya Black Sea Region

479
Sediment Transport Technology

Transverse structures are used to train and improve the river beds, regulate transport of
bed material, form reservoirs, and divert water for irrigation. Depending on the
constructional features and their hydraulic functions, they may be subdivided as:
• Weirs
• Weirs with counterweirs
• Sloped and Ogee weirs
• Sills
• Dams

Weirs are low dam structures and built generally at upper and middle reaches of rivers.
They can be designed as vertical, stepped or sloping downstream face depending on soil
mechanics and hydraulic considerations.

Weirs with counterweirs built at downstream of the primary structure and forming a
stilling basin between two structures, are preferred for reducing the erosion caused by
the flow over the crest.

When the soil at the weir site is of poor quality, sloped and Ogee weir; a structure with
a sloping downstream face provides a wide base and is more efficient from static and
hydraulic points of view.

Sills are normally used in the middle and lower reaches of the rivers to stabilize the river
bed when erosion occurs.

Longitudinal structures consist of bank protection structures in water courses and

channels and are divided basically into two groups; massive structures like retaining
walls stepped on the external face, and light structures forming true bank revetments. A
training wall or revetment on the concave bank and a series of spurs or groynes on the
convex bank. The training wall on the concave bank protects it from erosion.

Groynes are used to centralize and correct the direction of flow if it threatens the
stability of bank erosion. Gabion groynes are rather efficient because they are
permeable and flexible. Groynes must meet any variation of the hydraulic regimen and
of river profile. They may be:

Photo 3. A mortared riprap weir at Photo 4. A gabion stepped weir at Rize,

Macedonia Eastern Black sea region of Turkey

480
 River Training Structures - Applications

• Straight groyne
• T-shaped groyne
• Bayonet groyne

The groynes which usually extend to (a) guiding or deflecting the axis of the flow, (b)
establishing normal channel width, (c) promoting scour and deposition of sediment
where desired, and (d) trapping sediment load to build up new river banks.

Besides the above listed structures, for minimizing the river bed and bank erosion, it
may be necessary to make channel lining (Photo 6) for both embankment and river bed.
For lining different types of materials can be used. The bed and banks can be stabilized
by providing stone pitching, lining, concrete revetment, asphaltic pavements, plastic
films, a series of stone dikes and stone rip-rap, etc.

Blanket revetment is constructed of concrete, asphalt or other materials placed to form a

protective cover to the stream banks. Pervious revetments that will cause deposition.

Solid fence is constructed useful is steep slope streams. A problem associated with
rockfill bank protection is the scouring of the foundation material by high velocity
currents. It is recommended to excavate a trench at the toe of the bank and fill this with
rock the avoid undermining.

2. Weir Applications for River Training

2.1. Introduction
In these two exercises a variety of gabion weirs for river training and retaining walls for
soil erosion control and land reclamation are presented.

In the first exercise, erosion of the streams is checked by lowering the velocity of the
river flow. This is done by reducing the gradient to obtain a stable velocity (Fig. 1). The
uniform flow formula is used. The stable slope is given by:

Photo 5. A bank protection with gabion Photo 6. A bank protection with reno
retaining wall and groyne mattress at Mengen, Bolu-Turkey
(Maccaferri, 1988)

481
Sediment Transport Technology

Figure 1. The original (natural) and the trained (stable) slope of river

(u * U 1 )10 / 3 * B 4 / 3 * n 2
ie =
Q4/3

where, ie; stable slope, u; ratio of mean velocity of water to the corresponding velocity
at the river bottom; and nearly equals to 1.3-1.5, U1; maximum permissible velocity
(m/sec), B; wetted perimeter (m), n; Manning roughness coefficient (1/C), Q; design
discharge (for small structures use Tr = 20-30 years, for larger and important works use
50 to 100 years) (m3/sec).

In the natural slope of river is i, the stable slope after training is ie, and the weirs are
located at equidistant points of distance l, and height difference of H (Fig. 2), the
number of weirs for the whole stretch of river, L is given by:

L L(i − ie )
n= =
l H

Gabion weirs are classified:

• Vertical weirs
• Stepped weirs
• Sloped weirs

Figure 2. The weight and spacing of weirs as check dams for stabilization

482
 River Training Structures - Applications

To protect the crest from abrasion, timber, steel or concrete capping with joints at 2
meters are used (Fig. 3).

In the design of vertical weirs, the kinetic energy of the flow is dissipated at the toe of
the structure. In some cases, the drop of water is allowed to scour the bed and form a
pool where the energy is dissipated in the cushion of water and in the formation of a
hydraulic jump (Fig. 4). A secondary weir is placed at the downstream (d/s) end of the
pool to control the jump. This is necessary to ensure the stability of the main weir.

Figure 3. Structural details of concrete capping

Figure 4. Vertical weir with a drop to dissipate energy in a pool

483
Sediment Transport Technology

In case of erodible soils, the river bed in the stilling pool is protected by an apron and
the control of jump is done by either broad crested counterweir or abrupt rise.

2.2. Exercise on weirs for river training

Here, the first example is presented step by first giving the formula used then the
numerical values of each parameter and finally the drawings.

The object is to design and construct a river training structure to obtain a bed slope of
1.0-1.5% and a mean velocity of 2 m/sec so to reach the stable conditions.

The solution is a vertical weir, with a stilling pool unlined and there is no counterweir.
Location is at Bologna, Italy (Agostiniy, R. et al., 1985).

Design data
The original slope is i = 3.5%
The width of the river l = 20 m
The design discharge Q = 20 m3/sec
The unit discharge is 3.5 m3/sec per km2
The roughness coefficient C = (1/n) = 30 m1/3/sec

Computation
By using a modified form of Manning uniform flow equation, the uniform flow depth is
computed as:

y = (Q/(C*l*i1/2))3/5 = (20/30*20*0.0351/2))3/5 = 0.335 m

which is (Z3-f3) in Fig. 6. The mean velocity;

U = Q/(l*y) = 20/(20*0.335) = 2.815 m/sec

The critical velocity;

Ucr = g ⋅ y = 9.81 ∗ 0.355 = 1.87 m/sec

Since U>Ucr so the mean velocity is sufficient to move the material on the bed.

By studying the cross section of the possible weir location from topographical map, and
keeping in mind the main objective of the design the weir shown in Fig. 5 and Fig. 6 is
chosen. The structure is backfilled with compacted soil.

a) Crest design
Type of the crest selected: Trapezoidal and the top of the crest protected by concrete.
Base width = 8 m, Side slopes: 3:4, Flow of crest: Qc, Depth of flow of the crest = (Zg-fg)

By using the following equation for critical discharge

Qcr = A*(g*A/b)1/2

484
 River Training Structures - Applications

Figure 5. The chosen dimensions of the weir

Figure 6. The structural details of the weir

where, Qcr; critical discharge (m3/sec), A; area of flow section (m2), b; top width of
water surface (m).

The flow values corresponding for different flow depths over the crest are computed as;
(Zg-fg) 0.2 0.4 0.6 0.8 1.0 (m)
Qc 2.28 6.56 12.3 19.2 27.3 (m3/sec)

By interpolation it is that found; if Q = 20 m3/sec the corresponding depth of flow over

crest is (Zg-fg) = 0.82 m.

At this depth of flow the corresponding specific head is computed by using the
following formula;

485
Sediment Transport Technology

(Zg-fg)+(A/2b) = 0.82+(7.457/2*10.19) = 1.19 m

When the weir is completed, the velocity in the upstream (u/s) of weir will be very
slow, so the specific head will be same to the head of water (Z0-fg) , but immediately u/s
of the weir flow will be subcritical and transition from subcritical to supercritical will be
formed by hydraulic jump.

When the bed slope of 1% (as objective of the project) is obtained, then the uniform
flow depth will be (for Q = 20 m3/sec design discharge):

Yu = (Q/C*i1/2)3/5 = (20/30*20*0.011/2)3/5 = 0.52 m

and the uniform flow velocity;

Uu = Q/l*Yu = 20/20*0.52 = 1.93 m/sec

Since the critical velocity;

Ucr = g ⋅ Yu = 9.81 ∗ 0.52 = 2.26 m/sec

So the velocity is less than Ucr, and the specific head is

Hu = Yu+Uu2/(2g) = 0.71 m

Immediately above the weir, the river is nearly rectangular and has about 25 m width.
When it is filled by deposited material to the level of crest, the specific head is nearly
equal to the critical head as computed 1.19 m., to which a depth of (Z0-f0) = (Z0-fg) =
1.16 m corresponds (Fig. 6). Backwater curve represents the transition from 1.16 to
1.19m.

b) Stilling pool design

Stilling pool is not paved and shown in Fig. 7. The depth of water y = Z3-f3 = 0.355 m
and the maximum scour depth at the toe of the weir is computed by Schocklitch
formula:

( Z 0 − Z 3 ) 0.2 g 0.57
Z 3 − f b = 4.75
d t0.32

where, Z0-Z3 = 5.25+1.19-0.36 = 6.08 m, q; unit discharge in (m3/sec) per meter width
and computed as

20
q= = 2.2 m3/sec/m
8 + (4 / 3) * 0.38
dt = 160 mm (Aperture diameter in mm of sieve which passes 90% in weight of bed
material)
f 3 − f b = 2.11 − 0.36 = 1.75 m

486
 River Training Structures - Applications

Figure 7. Unpaved stilling in vertical weir without a counter weir

So 2.0 m deep foundation and sheet piling (Fig. 6) at the toe protects the structure from
undermining.

The distance of x of the free fall from the crest is computed using the formula of

x = 2(Z g − f g ) + (Z g − f 3 )

where (Zg-fg) and (Zg-f3) depths are as shown in Fig. 6. So in this example

x = 2 ∗ 0.82 + (0.82 + 5.25) = 3.16 m

In order to simplify the computation, it is possible to compute x from the monogram

given in Fig. 8 according to the depth variations of (Zg-f3).

The level Zv (Fig. 6) can be computed assuming the total head is constant between a
section at Zv and Z3. Therefore

( )
Z v − f 3 = (Z 3 − f 3 ) + U 32 2 g = 0.36 + 0.4 = 0.76 m

c) Check stability and resistance at crest

As shown in Fig. 6 there are seven layers of gabion of total 7 m height and it is crest-
lined with a beam concrete, the gabion foundation was reinforced sheet piling 0.5 m
thick. The stability against overturning of the part of the structure above section A-B
has to be verified.

The input data

Density of gabion filling stones τs = 2600 kg/m3
Porosity of gabion n = 0.30

487
Sediment Transport Technology

Figure 8. Monogram for free fall distance from toe of a weir, as a function of (ZG-f3) depth

Density of gabion τg = τs(1-n) = 1820 kg/m3

Density of gabion submerged τgw = (τs-τw)*(1-n) = 1120 kg/m3
Density of water τw = 1000 kg/m3
Density of soil behind gabion τs = 2600 kg/m3
Porosity of soil n = 0.35
Degree of saturation U = 0 (When soil is dry)
Density of dry soil τto = τs(1-n) = 1690 kg/m3
Density of wet soil τtw = (τs-τw)*(1-n) = 1040 kg/m3

Referring to Fig. 6 for the worst conditions, the values of forces and their corresponding
moments taken about the hinge point A are computed as tabulated below.

Forces Moments
Weight of gabion structure 30820 kg 50880 kgm
Weight of water on the crest and on the u/s backfill 3690 kg 8300 kgm
Weight of u/s backfill 9100 kg 31280 kgm
Pressure of water u/s (Hwm) 25140 kg 48800 kgm
Pressure of water d/s (Hwv) 290 kg 70 kgm
Thrust of soil upstream (Htm) 6240 kg 12480 kgm

488
 River Training Structures - Applications

The forces listed in above table are shown in Fig. 9. In computing these forces the basic
formula used are given as follows:

[
H wm = H w1 = 1 2τ w (h1 + h2 + h3 ) − h12
2
]
H wv = H w 2 + H w3 = 1 2τ w [h4 + h5 ]
2

H tm = 1 2τ w (h2 + h3 ) δ α
2

H tv = 1 2τ w ⋅ h5 ⋅ δ α

where Hw2; hydrostatic pressure acting on that part of structure above soil level, Hw1 and
Hw3; act on that part of structure below soil due to presence of seepage, Hwm; hydrostatic
pressure on u/s face, Hwv; hydrostatic pressure on d/s face, Htm; u/s soil pressure, Htv; d/s
soil pressure, and,

⎛π Φ ⎞
δ α = tan 2 ⎜− ⎟
⎝4 2⎠

where Φ is the angle of internal friction of soil for the elevations refer to Fig. 9.

According to the forces and moments listed in the table the coefficient of the stability
against overturning is:

50880 + 8300 + 31280 + 70

Sr = = 1.5
48800 + 12480

which is bigger than 1.0, so moment of restoring of forces; Ms is bigger than the
moment of overturning forces Mw. Finally the pressure distribution for section A-B is
thus studied (Fig. 9). The maximum pressure at a point A does not exceed the safe
bearing capacity of the soil. The resultant of forces passes outside the middle third of
the base section as in Fig. 10a and Fig. 10b, at a distance XA, from A, and XA is
computed as;

Figure 9. Forces acting on a vertical weir

489
Sediment Transport Technology

Figure 10. Centre of pressure distribution (a) within the middle third,
(b) outside the middle third

ΣM V − ΣM H (50880 + 8300 + 31290) − (48800 + 12480 − 70)

XA = =
ΣV 30820 − 3690 + 9100
XA = 0.67 m

Assuming that the gabion structure will not resist tensile stress, δA is computed as;

2 ∗ ΣV 2 ∗ (30820 + 3690 + 9100)

δA = = = 4.34 kg/cm2
3* X A 3 * 67 *100

2.2.1. Different types of weirs in river training

Besides the example given above section, it is very common to have weirs with stilling
pool with lined floor, so the hydraulic jump is controlled by broad crested weir (Fig.
11). This type is preferred when the river bed consists mainly of loose material of
limited size. The river bed is prevented from abrasion by lined floor. The protection is

Figure 11. Vertical weir with lined floor and hydraulic jump controlled broad crested weir

490
 River Training Structures - Applications

made by any material. In this type of structure (fc-fb) is called the height of counterweir
above the apron.

The basic criteria in a correct design of the pool is to have critical flow occur on the
counterweir and the flow of water in the pool will not be influenced by the flow
conditions d/s of the counterweir.

By using the equation of hydraulic jump, the depth of subcritical flow; (Z1-fb) is
computed as;

Q
(Z1 − f b ) =
lb 2 * g * (Z 0 − f b )

where; lb is the width of the pool in meter. The relative sequent depth of water (Z2-fb) is
computed by:

(Z1 − f b ) 2Q 2 (Z − f b ) 2

(Z 2 − f b ) = − + + 1
2 glb (Z 1 − f b )
2
4

and it is obtained by means of broad crested counterweir at the end of the stilling pool.

The dimensions of the counterweir can be obtained by using the equation of:

Q = µ (Z 2 − f c ) ∗ l c ∗ 2 g (Z 2 − f c )

where, fc; elevation of the counterweir (m), lc; width of the counter (m), (Z2-fc); depth of
water above counterweir, and µ; coefficient of discharge (varies 0.4-0.6).

The critical depth of water (Z1-fb) is more easily computed by using the monogram
given in Fig. 12, where (Z1-fb) is plotted against (Z0-fb) and Q/lb and (Z2-fb) is plotted
against (Z1-fb) and (Q/lb).The elevation of the water ZV adjacent to the d/s face of the
weir is computed by,

0.22
⎡ ⎤
(Z v − f b ) = ( f g − f b )⎢ 2 Q
2
⎥
⎢⎣ glb ( f g − f b ) ⎥⎦
3

The length of stilling pool Lb is computed by,

Lb = L g1 + L12

where, Lg1; the distance of the weir from the position where the supercritical flow of
depth (Z1-fb) is formed, and L12; the portion of the basin in which the jump occurs.
These are computed by using the following formulas:

491
Sediment Transport Technology

Figure 12. Computation (Z1-fb) in broad crested weir, with a stilling pool

(Z g + f g − 2 fb ) Z g − f g
L g1 =
Z g + f g − 2Z v
L12 = 6.9(Z 2 − Z 1 )

A common approach is to compute the stilling pool dimension as a function of drop

number, developed by experimental investigations as;

q2
D=
g( f g − fb )
3

where q is the unit discharge flowing over the crest (Q/lg).

All the necessary stilling pool dimensions are obtained from a ready monogram as
shown in Fig. 13, where D has been plotted against (fg-fb) and q, the specific discharge.

2.2.2. Stilling pool with lined floor and control of jump by abrupt rise
The control of jump can be achieved by means of abrupt rise. In this case tail water
influences the conditions of flow in the pool (Fig. 14). The flow in the pool may be
described as follows:

492
 River Training Structures - Applications

Figure 13. Dimensions of stilling pool as function of D

Q = µ ⋅ l g ⋅ (Z 0 − f g ) ⋅ 2 g (Z 0 − f g ) (over the crest)

Between section (0) and (1):

Q2 Q2
(Z 0 − f b ) + = ( Z − f ) +
2 g (Z 1 − f b ) * l b2
1 b
2 gA 2 2

Between section (1) and (2), where hydraulic jump occurs:

(Z 1 − f b ) 2Q 2 (Z − f b ) 2

(Z 2 − f b ) = − + + 1
2 g l b (Z 1 − f b )
2
4

Between section (2) and (3):

Q2 Q2
(Z 3 − f b ) + ≥ ( Z − f ) +
2 g lb2 (Z 2 − f b )
2 b
2 gA 2 2

493
Sediment Transport Technology

Figure 14. Gabion vertical weir with lined stilling pool, hydraulic jump is controlled by abrupt rise

The symbols used in these equations are already given in figures before.

2.2.3. Stepped weirs

Gabion weir may also be used in stepped form (Fig. 15). Compared to vertical weirs
stepped weirs have the following advantages;
a) better stability
b) the dissipation of some energy on each step
c) the pool itself can be shorlined or neglected

It is not very easy to compute the energy dissipated at each step; therefore the d/s slope
is expressed as average. The design criteria are based on the model studies in hydraulic
laboratories.

494
 River Training Structures - Applications

Figure 15. Correct design of stepped gabion weirs

2.2.4. Sloped weirs

Weirs with a sloped downstream profile are used for bed stabilization in river training
(Fig. 16). Also, it is possible to use them as storage reservoir by creating impermeable
surface on u/s side. This type of reservoirs has 5-15 meter height. The weirs having u/s
and d/s faces sloped, give improved stability. The design criteria do not differ very
much from vertical weirs. Since there is no free fall, only crest and stilling pool design
and hydraulic computations for sloped weirs will differ for vertical weirs, because the
slope is generally 1:2 or 2:3 which is sufficient to guarantee that the flow adheres to the

Figure 16. Sloped weir

495
Sediment Transport Technology

surface. It is not necessary to check this type on weir for stability against overturning;
also from pressure distribution point of view no need to verify if the soil is acceptable or
not, but safety against sliding should be verified.

3. Retaining Walls for Bank Protection

Retaining structure, defined as structure capable of resisting soil pressure, is also used
for bank protection in river training works (Photo 7). Especially, valuable agricultural
lands adjacent to fast flowing rivers are protected by using retaining walls. In design of
retaining structures, morphology of the soil surface, surface water runoff and level of
the local water table are main factors considered. The projects must therefore, be
examined within its climatic and hydrogeologic setting, keeping in mind the adverse
effects that can result to the natural environment.

(a) (b)

(c) (d)
Photo 7. Different types of material used for retaining walls used for bank
protection, (a), (b), (c) Uzungol, Trabzon, Eastern Black Sea Region,
(d) Niigata, Japan

496
 River Training Structures - Applications

Retaining walls for bank protection are permanent structures and usually built at the toe
of the slope, or they may contain backfill. Permeability, uniform or differential
settlement, backfill behind the wall, and depth of freezing are important factors in
design. Retaining walls may be gravity walls, reinforced concrete walls, walls made of
pre-fabricated elements, or gabion walls.

Gabion walls are cellular structures, made of steel wire mesh, filled with stone of
proper size and mechanical characteristics. In unconsolidated soils with modest
mechanical strength and those subject to settlement, the gabions structures are suitable
because of their flexibility, permeability and reinforcement which provide the
possibility to function under tension, and to absorb unforeseen stresses.

Example
In this example, a typical retaining wall made of gabions is presented. First the
necessary dimensions and then the loads transmitted by soil will be estimated. The next
step will be check for stability against sliding, overturning and finally overall stability
will be carried out. Wall and foundation designs will also be given. All will be carried
out for the gabion retaining wall shown in Fig. 17 (Agostini et al., 1987).

Data given
Density of the gabion γg = 1.65 t/m3
Density of the fill behind the wall γt = 1.8 t/m3
Their initial angle of friction ϕ = 30o
Cohesion factor C = 1 t/m3
Total height H = 8.00 m (Fig. 19)
Buried part of the structure = 1.00 m
The structure is battered for α = 6o

Figure 17. Gabion retaining wall for resistance against sliding

497
Sediment Transport Technology

Surcharge load p0=1.8 t/m2 (additional to thrust of soil behind the wall)
Top of the back fill is horizontal ε = 0

Gabions used are made of hexagonal mesh 8x10 with wire diameter of φ = 3.00 mm.
The surcharge load of 1.8 t/m2 corresponds to 1 m thick extra soil layer on the back fill.
Step by step the design computations are as follows:

a) For a unit length of wall

Effective length of the base B = FA = 4.50 m
Pressure acts against the vertical plane passing through the point D
Area of gabion Am=16 m2
Centre of gravity of gabion Gm: Coordinates of Gm with respect to X and Y axis parallel
to the faces of gabion and passing through the point O. Xm = 2.00m, Ym = 1.94 m
Area of the soil in QA′DN section: At = 13 m2
Centre of gravity of the soil is Gt: Its coordinates are Xt = 3.50 m, Yt = 4.04 m

The coordinates of the points Gm, Gt, Q, N, D, F, A in the coordinate system Oxy which
has its origin at O and vertical and horizontal axis rotated 6o with respect to Oxy
coordinate system used in above computations, are computed by using the relations of:

x = X cos α + Y sin α y = − X sin α + Y cos α (1)

The results are;

xm = 2.19 m ym = 1.72 m xt = 3.90 m yt = 3.65 m
xQ = 2.72 m yQ = 6.75 m xN = 5.21 m yN = 6.49 m
xE = xD = 4.48 m yD = -0.47 m xF = - 0.10 m yF = -0.99 m
xA = 4.37 m yA = -1.46 m

b) The areas and the static moments about the hinge point F of the triangles QEP
and PND
AQEP = (x E − xQ ) (tan α + tan ε ) 2 = 0.162 m2
2

M QEP = AQEP (x E − x F − (x E − xQ ) 3) = 0.646 m3

( 2
)
APND = DN ⋅ tan α 2 = 2.58 m2
M PND = APND ( x E − x F + ( x N − x E ) 3) = 12.42 m3

The weight and moments about the hinge F of gabion wall and soil confined at the back
of the gabion and BND plane are;

Wm = Am γ g = 26.4 t/m Wt = (At + AQEP − APND )γ t = 19.06 t/m

M m = Wm ( x m − x F ) = 60.62 tm/m
[ ]
M t = At ( xt − x F ) + M QEP − M PND γ t = 72.58 tm/m

The resultant weight and moment of the surcharge above QE surface;

WC = p0 (x E − xQ ) = 3.16 t/m M C = WC (xQ − x F + (x E − xQ ) 2 ) = 11.69 tm/m

498
 River Training Structures - Applications

Let Wf be the weight of the soil between FA and RA surfaces;

W f = (x A − x F ) ⋅ γ t ⋅ tan α 2 = 1.89 t/m

2

c) Active pressure computations

Referring to Fig. 18, it is possible to write down the related angles as δ = ϕ = 30o, β =
90o and ω = δ+(90-β) = 300, when the β is the angle between the plane of earth thrust
and the horizontal. If the active thrust acts at a height Ha, the coefficient of the active
pressure Ka is;

H a = y Q + (x E − xQ ) ⋅ tan ε − y A = 8.22 m
sin 2 (β + ϕ )
Ka = = 0.297
sin (ϕ + δ )⋅ sin (ϕ − ε ) ⎞
2
⎛
sin 2 β ⋅sin (β − δ )⎜⎜1 + ⎟
⎝ sin (β − δ )⋅ sin (β + ε ) ⎟⎠

The active earth pressure Sa, its horizontal and vertical components Sh and Sv, and its
moment about the hinge point Ma are;

S a = γ t H a2 K a 2 + p 0 H a K a = 22.46 t/m
S v = S a sin ω = 11.23 t/m S h = S a cos ω = 19.45 t/m
⎛ H H + 3 p0 / γ t ⎞
M a = S h ⎜⎜ a a − ( y F − y A )⎟⎟ − S v ( x D − x F ) = −2.09 tm/m
⎝ 3 H a + 2 p0 / γ t ⎠

The negative sign indicates that the thrust intersects the base of the wall.

Figure 18. The related angles, weights and forces used in the computation

499
Sediment Transport Technology

d) Passive pressure computations

The related angles: β = 90o δp = ϕ = 30o ε=0

Passive earth pressure coefficient Kp is:

sin 2 (β − ϕ )
Kp = = 10.10
sin (ϕ + δ p ) ⋅ sin (ϕ + ε ) ⎞
2
⎛
sin 2 β ⋅ sin (β + δ p )⎜1 − ⎟
⎜
⎝ sin (β + δ p )⋅ sin ( β + ε ) ⎟
⎠

The passive pressure Sp is divided into two components:

S ′p - part acting up to the height of the hinge point
S ′p′ - part acting up to the level of the plane of sliding

Also, these are divided into their horizontal and vertical components, and if the moment
of Sp is Mp, then the results are:

S ′p = γ t y F2 K p 2 = 8.99 t/m
S ′ph = S ′p cos δ p = 7.78 t/m S ′pv = S ′p sin δ p = 4.49 t/m
M ′p = S ′ph y F 3 = 2.58 tm/m
S ′p′ = γ t y A2 K p 2 = 19.50 t/m
S ′ph′ = S ′p′ cos δ p = 16.89 t/m S ′pv′ = S ′p′ sin δ p = 9.75 t/m

Check the resistance to sliding along the horizontal plane RA passing through the
lowest point of the foundation. Let Fi be the sum of the forces causing sliding, Fs be the
sum of the forces resisting sliding, and, ηs be the safety factor, then:

Fi = S h = 19.45 t/m
Fs = S ′ph′ + C ⋅ B ⋅ cos α + (Wm + Wt + W f + WC + S v − S ′pv′ ) tan ϕ = 51.39 t/m
η s = Fs Fi = 2.64

Let Mi be the sum of the overturning moment, and Ms be the sum of the restoring
moment. The safety factor against overturning ηr = Ms/Mi is not computed because as
already noted Mi is negative:

M i = M a = −2.09 tm/m
M s = M m + M t + M C + M ′p = 147.47 tm/m

Let N be the normal face acting on the base section, FA, e the eccentricity of the
resultant, and σ1 and σ2 the soil pressures at points F and A:

N = (Wm + Wt + WC ) cos α + S a sin (ω + α ) + S ′p sin (δ p + α ) = 56.27 t/m

500
 River Training Structures - Applications

B (M s − M i )
e= − = −0.408 m
2 N
⎛ 6e ⎞ N ⎛ 6e ⎞ N
σ 1 = ⎜1 + ⎟ = 5.7 t/m2 σ 2 = ⎜1 − ⎟ = 19.3 t/m
2

⎝ B⎠B ⎝ B⎠B

Since the computed values of the soil pressure are significantly higher at the heel then at
the toe, it can be assumed that these results are on the safe side and that in practice the
soil pressures will be more uniform.

e) Computation of the allowable soil bearing pressure

Let T be the shear force acting at section FA, let Plim be the limiting pressure, and δam
the allowable soil bearing pressure, Nc, Nq, Nγ, iq, iγ, dc, dq, dγ the coefficients to be used
in Hansen’s formula:

T = −(Wm + Wt + WC )sin α + S a cos(ω + α ) − S ′p cos(δ p + α ) = 5.82 t/m

N q = e π ⋅tan ϕ ⋅ tan 2 (45 + ϕ 2 ) = 18.40
iq = 1 − 0.5 T N = 0.948
d q = 1 + 0.35 y F B = 1.077
N c = (N q − 1)cot ϕ = 30.14 d c = d q = 1.077
N γ = 1.8(N q − 1) tan ϕ = 18.08 iγ = iq2 = 0.899 dγ = 1
Plim = CN c d c + y F γ t N q d q iq + (1 2 )γ t BN γ d γ iγ = 131.8 t/m
σ am = Plim 3 = 43.9 t/m2 > σmax

f) Computation of the safety factor for overall stability

Assuming the failure surfaces to be planes and δ = ϕ. Let xB, yB be the coordinates of the
point of intersection B of the top the fill with the line NA:

[
QN = (x N − xQ ) + ( y N − y Q )
2
]
2 12
= 2 .5 m

AB = [( x − xB ) + ( y A − yB )]
2 12
= 8.26 m
2
A

x B = x N + QN sin (α + ε )sin α = 5.24 m

y B = y N + QN sin (α + ε ) cos α = 6.75 m

W1 = QN tan (α + ε )γ t 2 = 0.591 t/m

2

Varying ω the angle at which the failure plane passing through point A makes with
horizontal, it is found that the safety factor is least for ω = 43.58o. With this value of ω,
let C be the point of intersection the failure plane with the soil surface, let T be the
intersection point of the failure plane with the vertical through F, W2 the weight of the
soil section ABC, W3 the weight of the soil section AFT, W4 resultant of the surcharge
acting on QC:

501
Sediment Transport Technology

1 2 cos(α + ε ) cos(α + ω )
W2 = AB γ t = 57.48 t/m
2 sin (ω − ε )
W3 = ( x A − x F ) (tan α + tan ω )γ t 2 = 19.05 t/m
2

BC = AB cos(α + ω )sin (ω − ε ) = 7.77 m

[ ]
W4 = (x B − xQ ) + BC cos ε p 0 = 18.51 t/m

Let W be the sum of the destabilizing weights and S the thrust on the plane OFT, which
closes the force triangle with side S, W and the reaction along TC:

W = W1 + W2 + W3 + W4 + At γ t + Wm = 145.43 t/m
S = W sin (ω − ϕ ) cos(2ϕ − ω ) = 35.60 t/m

Let Rp be the passive thrust acting on the plane OFT:

yT = y F − (x A − x F )(tan α + tan ω ) = −5.72 m

R p = yT2 K p γ t 2 = 297.7 t/m

Let ηi be the safety factor for overall stability:

η i = R p S = 8.361

The search for minimum value of ηi was carried out by using the following values:
ω = 45o ηi = 8.399
ω = 43.4o ηi = 8.362
ω = 43.58 o
ηi = 8.361

Computation of tension at section F′A′ at the bottom of the stem, assuming the load is
acting along the line QA′. The weight, the moment arm about hinge point F′ and the
moment exerted on the wall section are:

Ws = 18.89 t/m b = 1.20 m Ms = 22.76 tm/m

Let β be the included angle between the plane on which the active earth pressure acts
and the horizontal, let δ, assumed equal to ϕ, be the angle between the plane on which
the passive earth pressure acts and the horizontal, ω the angle between the line of the
active earth pressure and the horizontal, Ha the height at which the active earth pressure
acts which is the vertical distance between the points Q and A′, Ka the coefficient of
active pressure, Sa the active pressure, Ma the moment of Sa about hinge point F′. The
computations are carried out in a manner similar to those given above, thus:

( )
β = α + arctan ZF′ F′A ′ − ZQ = 83.91o
δ = 30 o ω = δ + (90 − β ) = 36.09 o
Sa = 20.32 t/m Ha = 7.12 m Ma = 12.55 tm/m Ka = 0.348

502
 River Training Structures - Applications

Let M be the moment about the hinge point F′, N the normal force and T the tangential
force acting at section F′A′, e the eccentricity, σmax the maximum tension on the section,
and σam the allowable tension:

M = M s − M a = 10.21 tm/m N = Ws cos α + S a cos(ω + α ) = 32.49 t/m

T = −Ws sin α + S a cos(ω + α ) = 13.10 t/m B = F′A ′ = 2.5 m
e = B 2 − M N = 0.936 m x = (B 2 − e ) 0.4 = 0.786 m
σ max = N (0.8 x ) = 51.69 t/m2 σ am = 40 + 50(γ g − 1.4) = 52.50 t/m2 > σmax

Assuming the contribution of the weight of the wire mesh on the lower side of the
foundation to be Pu = 10.3 kg/m3, then the cohesion of the gabion surface is:

C g = 0.03Pu − 0.05 = 0.26 kg/cm2 = 2.6 t/m2

Let ϕ* be the fictitious angle of friction of the stone filling in the gabions; τ the average
shear stress on the section, τam the allowable stress:

ϕ * = 25γ g − 10 = 31.25 o τ = T B = 5.24 t/m2

τ am = N tan ϕ* B + C g = 10.49 t/m2 > τ

g) Computation of the deflection ∆ of point Z at the top of the wall

Assuming F′ to be fixed, the structure is divided in elements of thickness ∆H = 0.50 m
and a table is drafted. The surface on which the earth pressure acts is assumed to be
QA′, as computed above, with Ha = 7.12 m. Let h be the distance from Z measured
parallel to the face of the gabions to the centre of gravity of the element, let S be thrust,
W the weight of the structure above, and T the shear at the height of the centre of gravity
of the element, τ the average shear stress and b the width of the element, γ the angle of
motion and ∆ξ the contribution of the element to the movement in the direction ZQ.
W = Aγ g = 1.65 A T = S cos(ω + α ) − W sin α = 0.742 S − 0.1045W
( ) ( )
S = (1 2)K a γ t H a2 h ZF′ + K a p 0 H a h ZF′ = 0.3238h 2 + 0.637h
2

τ =T b γ = τ G = τ 25 ∆ξ = γ ⋅ ∆H = 0.5γ

Let also σ be the average normal tension, ητ the ratio between the allowable tangential
tension and the average tangential tension on the element.

σ = (S sin (ω + α ) + W cos α ) b = (0.6704 S + 0.9945W ) b

ητ = (σ tan ϕ* + C g ) τ = (0.6068σ + 1.5) τ

The total movement of point Z in the direction ZQ is obtained from Table 1, and finally
∆, the horizontal deflection of Z is computed:

Σ(∆ξ) = 0.685 m ∆ = Σ(∆ξ)cosα = 0.680 m

503
Sediment Transport Technology

Table 1. The total movement of point Z in the direction ZQ

H(m) b(m) A(m2) T(t/m) S(t/m) τ(t/m2) γ(-) ∆ξ(m) σ(t/m2) ητ(-)
0.25 1.0 0.250 0.090 0.180 0.090 0.0036 0.0018 0.53 20.22
0.75 1.0 0.750 0.360 0.660 0.360 0.0144 0.0072 1.67 6.98
1.25 1.0 1.250 0.751 1.302 0.751 0.0300 0.0150 2.92 4.36
1.75 1.0 1.750 1.261 2.106 1.261 0.0504 0.0252 4.28 3.25
2.25 1.5 2.375 1.870 3.073 1.247 0.0499 0.0249 3.97 3.14
2.75 1.5 3.125 2.578 4.200 1.719 0.0687 0.0344 5.30 2.74
3.25 1.5 3.875 3.406 5.190 2.270 0.0908 0.0454 6.69 2.45
3.75 1.5 4.625 4.354 6.942 2.902 0.1161 0.0580 8.16 2.22
4.25 2.0 5.500 5.400 8.556 2.700 0.1080 0.0540 7.38 2.21
4.75 2.0 6.500 6.545 10.331 3.273 0.1309 0.0655 8.80 2.09
5.25 2.0 7.500 7.810 12.269 3.905 0.1562 0.0781 10.26 1.98
5.75 2.0 8.500 9.196 14.368 4.598 0.1839 0.0920 11.79 1.88
6.25 2.5 9.625 10.680 16.630 4.272 0.1709 0.0854 10.78 1.88
6.75 2.5 10.875 12.262 19.053 4.905 0.1962 0.0981 12.24 1.82
Σ=0.685

The computed ∆ is the sum of the movements due to the earth pressure and that due to
the surcharge load; it is worthwhile to separate the two. In fact the structures eventually
superimposed on the backfill will be affected by the ∆′ due to surcharge, while the
permanent deformation ∆′′ of the structure due to the soil pressure.

Assuming a constant value for the tangential modulus of elasticity the results are:

⎛ 1 ⎞ h
∆ ′ = ∑⎜ K a p0 H a (h / ZF ′) cos(ω + α )∆H cos α ⎟ = 0.0094 ∑ = 0.256 m
⎝ bG ⎠ b

∆ ′′ = ∆ − ∆ ′ = 0.424 m

The effect of the surcharge is to increase the horizontal deflection by 0.26 m

remembering that the wall is inclined at 6o, a point on top of the wall is 0.31 m behind a
vertical line passing through the corresponding point at the bottom of the wall, when it
is only loaded by earth pressure. When the surcharge is in position this distance
becomes 0.05 m.

4. Longitudinal Structures for River Training

4.1. Introduction
Longitudinal structure is defined as the river training structure built parallel to river
flow. They are used for erosion protection control of river meandering, provision of
normal flow channel in urban areas and flood protection (Photo 8). They can be used
alone or in combination with other structures, such as groynes etc.

504
 River Training Structures - Applications

(a) (b)

(c) (d)
Photo 8. Different types of materials used for longitudinal protection of watercourses and shores,
(a) Rize, Black Sea; (b) Konya, Ermenek; (c) Lake Ohrid; (d) Rize, Black Sea region

4.2. Design steps

4.2.1. Morphology of water courses
To define the morphology of the water course it is necessary to examine the stability
both in plan and in profile view.

Although flood cycles will cause some local variations in hydraulic characteristics such
as gradient, flow depth, channel width and grading of the river bed material, in long
term these parameters are assumed to remain substantially constant.

The term stability index will be used to identify the functional relationship among the
hydraulic parameters mentioned above. Stability index has a functional relationship
such as:

∆ = f(Q, I, d, w, δs) = constant

the symbols are given in Fig. 19 and δs is density of the solid portion of the river flow.

505
Sediment Transport Technology

Figure 19. Geometric characteristics of channel in profile

The relation above varies according to the nature of the river bed and also on the basis
of the equilibrium assumptions. The river beds are classified as gravel-river beds and,
clay-river beds. It is also necessary to distinguish the critical equilibrium and dynamic
equilibrium conditions in river flow from solid particles point of view. For different
river bed and equilibrium conditions, different relations among the hydraulic parameters
exist.

4.2.2. Behavior in the horizontal plane

To understand the behavior of a river in a horizontal plane it necessary to analyze the
variations and instabilities in river planimetry. Radius or Curvature for the center line of
a river bend is given by Langhein and Leopold, with the equation:

Lm K 1.5
Rc =
13( K − 1) 0.5

where the symbols are given in Fig. 20 and sinuosity K is defined as the ratio between
the length along the curve center line and the corresponding cord distance. Empirical
equations for river meanders are given in Table 2.

Figure 20. Geometric characteristics of channel in plan

506
 River Training Structures - Applications

Table 2. Empirical equations for river meanders

Standard Number
Eq. deviation (%) Correlation of cases Field of application
Equation
no coefficient
+ - observed

Relation between the characteristics of the meander

2 Lm = 1.25Lb 32 24 0.99 102 5.5 ≤ Lb ≤ 13.300 m
3 Lm = 1.63B 31 24 0.99 155 3.7 ≤ B ≤ 13.700 m
4 Lm = 4.53Rc 21 17 0,99 78 2.6 ≤ Rc ≤ 3.600 m
5 Lb = 0.80Lm 32 24 0.99 102 8 ≤ Lm ≤ 16.500 m
6 Lb = 1.29B 13 24 0.99 102 3.7 ≤ B ≤ 10.000 m
7 Lb = 3.77Rc 35 26 0.98 78 2.6 ≤ Rc ≤ 3.600 m
8 B = 0.61Lm 31 24 0.99 155 8 ≤ Lm ≤ 23.200 m
9 B = 0.78Lb 31 24 0.99 102 5.5 ≤ Lb ≤ 13.300 m
10 B = 2.88Rc 42 29 0.98 78 2.6 ≤ Rc ≤ 3.600 m
11 Rc = 0.22Zm 21 17 0.99 78 10 ≤ Lm ≤ 16.500 m
12 Rc = 0.26Lb 35 26 0.98 78 6.8 ≤ Lb ≤ 13.300m
13 Rc = 0.35 42 29 0.98 78 5 ≤ B ≤ 10.000m
Relation between the cross section geometry and the characteristics meander of the river
14 A = 0.0054Lm1.53 103 51 0.96 66 10 ≤ Lm ≤ 23.200 m
15 A = 0.0085Lb1.53 140 58 0.95 41 6 ≤ Lb ≤ 13.300 m
16 A = 0.012B1.53 97 49 0.97 63 5 ≤ B ≤ 11.600 m
17 A = 0.061Rc1.53 138 58 0.97 28 2 ≤ Rc ≤ 3.600 m
18 W = 0.17Lm0.89 56 36 0.96 191 8 ≤ Lm ≤ 23.200 m
19 W = 0.23Lb0.89 56 36 0.97 102 5 ≤ Lb ≤ 13.300 m
20 W = 0.27B0.89 63 39 0.96 153 3 ≤ B ≤ 13.700 m
21 W = 0.71Rc0.89 48 32 0.97 79 2.6 ≤ Rc ≤ 3.600 m
22 D = 0.027Lm0.66 79 44 0.86 66 10 ≤ Lm ≤ 23.200 m
23 D = 0.036Lb0.66 72 42 0.90 41 7 ≤ Lb ≤ 13.300 m
24 D = 0.037B0.66 66 40 0.90 63 5 ≤ B ≤ 11.600 m
25 D = 0.085Rc0.66 90 47 0.90 28 2.6 ≤ Rc ≤ 3.600 m
Relation between the characteristics of the river meander and the geometry cross section
26 Lm = 30A0.63 59 37 0.96 66 0.04 ≤ A ≤ 20.900 m2
27 Lb = 22A0.63 77 43 0.95 41 0.04 ≤ A ≤ 20.900 m2
28 B = 18A0.63 56 36 0.97 63 0.04 ≤ A ≤ 20.900 m2
29 R = 5.8A0.63 76 43 0.97 28 0.04 ≤ A ≤ 20.900 m2
30 Lm = 7.5W1.12 65 39 0.96 191 1.50 ≤ W ≤ 4.000 m
31 Lb = 5.1W1.12 65 39 0.97 102 1.50 ≤ W ≤ 4.000 m
32 B = 4.3W1.12 74 42 0.96 153 1.50 ≤ W ≤ 4.000 m
33 Rc = 1.5W1.12 55 35 0.97 79 1.50 ≤ W ≤ 4.000 m
34 Lm = 240y1.52 142 59 0.86 66 0.03 ≤ y ≤ 18 m
35 Lb = 160y1.52 128 56 0.90 41 0.03 ≤ y ≤ 17.6 m
36 B = 148y1.52 115 53 0.90 63 0.03 ≤ y ≤ 18 m
37 Rc = 42y1.52 165 62 0.90 28 0.03 ≤ y ≤ 17.6 m

507
Sediment Transport Technology

Table 2. Empirical equations for river meanders (continued)

Standard Number
Eq. deviation (%) Correlation of cases Field of application
Equation
no coefficient
+ - observed

Relation between bankfull width W, depth y and the water course sinuosity
38 W = 21.3y1.45 160 62 0.81 67 0.03 ≤ y ≤ 18 m
39 y = 0.12W0.69 94 48 0.81 67 1.50 ≤ W ≤ 4.000 m
40 W = 96y1.23K-2.35 121 55 0.87 66 0.03 ≤ y ≤ 18 m
1.20 ≤ K ≤ 2.60
41 y = 0.09W0.39K1.46 73 42 0.86 66 1.50 ≤ W ≤ 4.000 m
1.20 ≤ K ≤ 2
A : Area of bankfull Lm : Meander length Rc : Radius of
W : Bankfull water surface width Lb : Length along the center curvature
y : Bankfull average water depth line of the meander K : Sinuosity of the
B : Meander deflection water course

4.2.3. Behavior in the vertical plane

The analysis of river bends is important to verify the stability of the bank protection
itself. To prevent undermining from causing instability it is essential to plan a safe
foundation level for the structures, and therefore in wertex areas it is reasonable to
deepen the foundation or to extend the length of the anti-scour apron.

4.3. Analysis of the forces involved

Basically two distinct types of problems can be identified. Problems are due to:
• Infrastructures and industrial development in the vicinity of water course
• Characteristics of the river bed and bank materials

The velocity of water depends on:

• Geometry of the cross-section
• Roughness of the wetted perimeter
• Gradient of the reach under consideration
• Asymmetric distribution of velocity of river flow and variation of river bed profile
and river bend (Fig. 21).

The velocity increment (∆V) and velocity distribution in a river bend are obtained from
the monogram given in (Fig. 22a and 22b), as a function of the parameter θ ′ , which is
approximated equation:

y max g
θ ′ = 0.42θ
B x

where x represents Chazy's resistance coefficient, θ is the angle of deviation of the

curve measured in degrees.

508
 River Training Structures - Applications

Figure 21. Analysis of deposition and erosion during normal, maximum and
minimum water level, at successive bends along a meandering river

Figure 22. (a) Diagram for estimating the velocity distribution in a river bend,
(b) Erosion velocity, transport and sedimentation as function of the
diameter of the bottom materials

Erosion at bend is function of the following factors:

• Increase in velocity at bend compared to straight channel velocity (K1)
• Decrease in depth in the bend lee side (K2)
• Decrease in stability of river bed material due to slope of the river band and effect of
cross- section (K3)

K1, K2 and K3 are called coefficients of decrease of erosion velocity, and they are
computed as function of the following parameters:

509
Sediment Transport Technology

• Increment in velocity at bend, ∆Vmax

• Distance from the river bank, x
• Depth of flow corresponding to maximum velocity vector in vertical profile, y
• Bank slope, m
• Friction coefficient for the angle repose of the material
• Angle included between the direction of flow and the line formed by the intersection
of the horizontal plane and the bank
• Radius of curvature of bend, Rc

The complete coefficient of reduction of the critical velocity of erosion is expressed,

K = K1 ⋅ K 2 ⋅ K3 < 1

and then it necessary to check the stability of the entire section for the critical conditions
developing close to the bank. For this purpose, the distribution of average velocity over
the width is computed

⎡ ⎛ 2 y ⎞2 ⎤
Vcp = 0.4Vmax ⎢1 − ⎜ ⎟ ⎥
⎣⎢ ⎝ B ⎠ ⎦⎥

where Vmax = Vaverage/0.945, and the final check is made for:

Vcp ≤ K ⋅ Vcritical (m, m0 )

The analysis of stability of the river bed materials must take into account both to
dynamic action of the current as well as the equilibrium conditions from geotechnical
point of view. The river current exerts a shear stress on the particles whose resistance
depends on the granulometry and on the nature of the formation. The maximum
permissible velocities for noncohesive soils (Fig. 23) and cohesive soils (Fig. 24) and
maximum permissible unit shear stress for channel banks in noncohesive soils (Fig. 25)
are presented to give the limiting situations in river training. For various types of soils,
the unlined bank slopes are given in Table 3.

Table 3. Unlined bank slopes various soils

Group Soil type Bank slope

1 Highly compacted clay (hard, pick, required) 1:1-1:1.25

2 Medium compact sandy clay soils 1:1.25-1:1.5

Slightly compact silty clay or sandy clay (soft,

3 1:1.50-1:1.75
spade required)

Noncohesive fine sandy soil (loose, shoved

4 1:1.75-1:3
required) soils with humus or peat content

510
 River Training Structures - Applications

Figure 23. Values of the maximum permissible velocities for noncohesive soils

Figure 24. Maximum permissible velocities

for noncohesive soils

Figure 25. Maximum permissible unit

shear in stress for channel banks
in noncohesive soils

511
Sediment Transport Technology

4.4. Selection and design of the protection works

The basic requirements for a protection work are stability, flexibility, durability, easy
maintenance, safety, environmental acceptance and reasonable cost. In the construction
of the river bank protection; rigid, semirigid, flexible, loose dumped materials and earth
levees are used. Box gabions, geotextile containers, reno mattresses, georigid linings
and synthetic products are listed in flexible structures group.

4.5. Ready computer programme

A ready computer program written in Quick Basic is available for fast computation
(Agostini, et all., 1988). Initially a deterministic design is tried, when the solution is
achieved probabilistic verification of the solution realized. The basic steps in the
computation and the data used as follows:

4.5.1. Inputs
• Plan and profile of the river section to be trained
• The characteristics and grading of materials on river bed and banks (gradient,
section width, free board of stage)
• Q2 for design purpose, Q100 for controlling the design

4.5.2. Calculations
Analysis of the conditions in the vertical plane: For each critical section along the river
reach to be trained, compute the geometric characteristics, and uniform flow velocity
and discharge values. Also determine the stability index for various sectors according to
the river bed and bank materials existing. If possible, decide on a constant stability
index value.

Analysis of the conditions in the horizontal plane: According to the geometric alignment
chosen for the river to be trained, for each section compute the radius of curvature
average depth of flow, width and degree of shift (Rc/W), because maximum tendency
toward lateral migration of the curve, i.e. the curve erosion is obtained at a value of
Rc/W = 3. Decide on the method to be employed, either change the plan layout to obtain
an increase of the radius of curvature or install bank protection structures. If, new layout
is planned, according new river width and radius of curvature values for each segment
are computed. Then, the required bank protection will be less extensive and lighter on
account of the reduced bank erosion conditions. Because the closeness of adjacent
structures, if the change of radius of curvature is not possible, then in addition to
appropriate protective measures check the average depth expected.

4.5.3. Selection and design of the protective structures

Possible structures are protective revetment, massive bank protection, lowering the river
bottom, drop structures. When a tighter curve is proposed, the hydraulic computations
and coefficient of critical velocity reductions should be repeated.

512
 River Training Structures - Applications

References

1. Agostini, R., Cesario, L., Conte, A., Masetti, M., and Papetti, A. “Flexible Gabion
Structures in Earth Retaining Works”, Officine Maccaferri S. P. A., Bologna, Italy,
(1987).
2. Agostini, R., Cesario, L., Ferraiolo, F., and Papetti, A., “Flexible Gabion and Reno
Mattress Structures in River and Stream Training Works”, Section 1. Weirs, Section
2: Two Longitudinal Structures, Officine Maccaferri S. P. A., Bologna, Italy,
(1988).
3. Chow, V. T., “Open Channel Hydraulics”, McGraw Hill, New York, (1959).

513
514
 HYDRAULIC TRANSPORT OF SOLIDS IN PIPES

Assoc. Prof. Dr. M. Ali KÖKPINAR

Hydraulics Laboratory
Technical Research and Quality Control Department
State Hydraulic Works, Ankara-Turkey

1. Introduction
The flow of mixtures of solids and liquids in pipes differs from the flow of
homogeneous liquids in a number of ways. With liquids the complete range of
velocities is possible, and the nature of the flow (i.e. laminar, transition, or turbulent)
can be characterized from a knowledge of the physical properties of the fluid and the
pipe system. Characterization of slurry flow is not as simple as for liquid flow for two
reasons. Firstly, there are, superimposed on the properties of the liquid, the properties of
solid particles to be accounted for and also the effect of the particles on the mixture
properties. Secondly, a range of slurry behavior is possible depending on the particular
conditions (Wasp et al. 1979).

The first successful application of hydraulic conveyance of solid material in closed

conduits dates back to 19th century, in the gold mines of California, USA. However, the
spectacular growth of slurry pipeline as a mode of transportation has been witnessed
during the last two decades. The driving force behind the surge of interest in slurry
pipelines is undoubtedly economics. Mineral deposits which were considered
economically inaccessible due to high transportation costs have been brought within the
economic reach of minerals industry by the slurry pipeline technology.

Table 1 is the collection of the important slurry pipelines in operation throughout the
world (Yücel et al. 1978). In all these commercial slurry pipelines the carrier liquid is
water. Fig. 1 gives an approximate means of estimating the water required annually to

Figure 1. Water requirements for slurry pipelines

515
Sediment Transport Technology

Table 1. Selected Commercial Slurry Pipelines (Yücel et al. 1978)

Length Pipe Diameter Capacity Operation

Material Location
(km) (cm) (mil.tons/yr) Year
Coal USA 173 27.3 1.3 1957
Coal USA 440 45 4.8 1970
Coal Russia 61 30 1.8 1957
Coal USA 1668 96.5 22.7 1979
Iron Tasmaia 82 22.7 2.3 1967
Iron Canada 130 40 9 1974
Iron Brazil 400 51.8 12 1977
Iron Germany 210 70 36 1980
Copper Japan 65 20 1 1972
Copper Indonesia 110 5 0.3 1972
Copper Turkey 63 12.7 0.3 1970
Copper USA 19 10 0.4 1974
Limestone UK 92 30 1.7 1964
Limestone Colombia 27 19.4 0.4 1971
Gilsonite USA 116 15 0.6 1957
Phosphate Brazil 113 24.4 1.8 1978
Sulphur Canada 1300 20.4 2.5 1978

transport a given annual tonnage of solids.

Fig. 2 can be used for a quick estimate of the pipe diameter given the annual throughput
requirement in tons per year. Physical properties of some of commercially transported
slurries and the selected design velocities are shown in Table 2.

Figure 2. Pipe diameter for slurry pipelines

516
 Hydraulic Transport of Solids in Pipes

Table 2. Slurry Characteristics (Yücel et al. 1978)

Specific Max. particle Concentration by Operating velocity

Mineral
weight size (mm) weight (%) (m/s)
Coal 1.40 2.38 45-55 1.50
Copper 4.30 0.210 60-65 1.50
Iron 4.90 0.149 60-65 1.85
Limestone 2.70 0.297 60-65 1.10

The economy of slurry pipelines as compared to the conventional modes of

transportation is illustrated in Fig. 3.

Figure 3. Transportation costs (a) coal (b) iron, copper, limestone

517
Sediment Transport Technology

2. Hydraulics of Solid-Liquid Flow in Pipelines

2.1. Classification of solid-liquid flows
Solid-liquid flow in pipes can be broadly divided into two categories as homogeneous
and heterogeneous flows.

Homogeneous flow is the term given to systems in which the solids are uniformly
distributed throughout the liquid medium. However, homogeneous flow is encountered
in slurries of high solid concentrations and fine particle sizes. The particles are kept in
suspension by the turbulence of the carrier fluid. The presence of solids can have a
significant effect on the system properties, usually resulting in a sharp increase in
viscosity as compared to that of the carrier fluid. Often these systems exhibit a non-
Newtonian rheology. Typical examples of suspensions which behave more or less
homogeneously at normal pipeline velocities are sewage sludge, and drilling muds.

Heterogeneous flow is the term given to systems in which pronounced concentration

gradients exist along the vertical axis of the pipe. Particle inertial effects are significant,
i.e. the fluid and solid phases to a large extent retain their separate identities, and the
viscosity of the carrier fluid is little affected. Heterogeneous slurries tend to be of lower
solids concentration and have larger particles sizes than homogeneous slurries. Typical
examples are phosphate rock, coal, and iron ore slurries.

Heterogeneous flow may further be divided into two regimes:

i) Non-deposit regime: where the slurry is transported totally in suspension, and
ii) Deposit regime: where the mean velocity in the pipe is low enough for some
particles to settle out and hence the solids may form a uniform layer on the bottom
of the pipe. Slurry is transported in suspension and a bed load by saltation.

The condition of flow is governed to a great extent by the relative magnitude of the
viscous-turbulence driving forces of the carrier fluid and the internal forces of the solid
particles.

If the former mechanism is represented by the mean velocity of the flow and the latter
by the settling velocity of the solid particles, then one may expect a flow-classification
as shown in Fig. 4.

Consider a horizontal pipe; the bottom of the pipe is covered with a plane, stationary
bed of loose, cohesionless solid particles of uniform size with a certain settling velocity.
As soon as liquid starts to flow hydrodynamic forces are exerted on the solid particles of
the bed. If the flow velocity is increased some bed particles start to move (saltation)
with a further increase in flow velocity some particles start to move in suspension while
some continues to move by saltation (heterogeneous-deposit regime). If the flow
velocity reaches to the so called critical velocity, Vc, all the particles are dislodged from
the bed and carried in suspension, although in greater concentrations close to the bed
(heterogeneous flow-non deposit regime). If the velocity is further increased to the so
called transition velocity, VH, the concentration gradient disappears and the material is
carried with almost uniform concentration throughout the pipe (homogeneous flow).

518
 Hydraulic Transport of Solids in Pipes

Figure 4. Classification of solid-liquid flow

2.2. Transition velocity (homogeneous-heterogeneous flow)

Various expressions have been proposed for the determination of transition velocity
between homogeneous and heterogeneous flow conditions for solid-liquid flows in
horizontal pipes.

According to Newitt (1955):

V H = 1800 gDW (1)

According to Wasp (1971):

1/ 7
⎛ WDρ m ⎞
VH = 294⎜⎜ ⎟⎟ (2)
⎝ µ m ⎠

and according to Spells (1955):

0.775
⎛ Dρ m ⎞
V 1.225
= 0.075 gd ⎜⎜ ⎟⎟ (s − 1) (3)
⎝ µm
H
⎠

in these expressions, VH; transition velocity, g; gravitational acceleration, D; pipe

diameter, W; settling velocity, d; particle size, ρm; density of solid-liquid mixture, and
µm viscosity of solid-liquid mixture.

519
Sediment Transport Technology

In addition to the above expressions, some general guidelines have been proposed for
distinguishing between homogeneous and heterogeneous flow conditions.

According to Durand (1953), for solid particles with settling velocities less than 0.60
mm/sec, homogeneous flow, and with settling velocities, greater than 1.50 mm/sec
heterogeneous flow conditions prevail. According to Condolius and Chapus (1976)
solids of less than 50 microns form homogeneous slurries, where as solids of greater
than 200 microns form heterogeneous slurries. Aude (1971) classifies the types of flow
according to the size and specific weight of the solid particles as shown in Fig. 5.

Behavior of the slurry in the transition region (i.e. between 0.60-1.50 mm/sec settling
velocity, or 50-200 microns particle size, or between the two curves in Fig. 5) depends
mainly on the solid concentration. Higher concentrations force the flow toward
homogeneous and lower concentrations towards heterogeneous conditions.

2.3. Critical velocity (nondeposit-deposit regime in heterogeneous flow)

The problem of predicting critical flow velocity in solid-liquid carrying pipelines with a
sufficient accuracy is of great interest to many hydraulic engineers. The interest is the
minimum cost of slurry transportation at this velocity. In addition, for the velocities
below the critical velocity, there may be deposition of solid particles in the pipeline,
causing excessive pipe abrasion, head loss, and plunging.

To define the critical flow velocity in a solid-liquid carrying pipe, a stationary layer of
noncohesive coarse solid particles is considered initially. The noncohesive solid
particles usually do not alter the rheological properties of the carrier fluid employed,

Figure 5. Homogeneous-heterogeneous flow regimes

520
 Hydraulic Transport of Solids in Pipes

which is usually water. Therefore, solid and liquid phases behave somewhat separately.
When the flow velocity increases gradually, because of shear exerted on the particles,
some particles move at the uppermost layer tend to move along the flow direction by
jumping, rolling, or sliding. This situation follows an appearance of a number of bed
formations. A further increase in flow velocity results in the initiation of motion of
particles at the lower layers and eventually, at a critical condition, all the particles at the
pipe wall move without any deposition. In this condition, solid particles pass from a
saltation regime to a heterogeneous regime. Within the transition zone between
heterogeneous and saltation regimes, there is a unique velocity, corresponding to
minimum head loss in the pipeline, below which deposits will occur but above which no
deposit in the pipeline will be encountered. This velocity is defined as the critical flow
velocity, Vc, and, for any design problem related to the topic, information on this
velocity is strongly required. Solid-liquid carrying pipeline systems may be operated
economically if there is adequate information to define a critical velocity, Vc, of the
mixture (Graf 1971; ASCE 1975; Wasp et al. 1979; Swamee 1995; Matousek 1996).
However, it is advisable to operate at a velocity somewhat higher than the predicted
critical velocity.

There have been many attempts to present a generalized expression for critical velocity,
perhaps the best known being that due to Durand (1952);

Vc = FL 2 gD(s − 1) (4)

coefficient FL is a function of particle size and concentration as shown in Fig. 6.

For particle sizes up to 1 mm, both concentration and particle diameter have an effect
on the value of FL. For particles greater than 1 mm, the dependence becomes weaker,
until, for large particles the value of FL is constant regardless of system properties. Thus
for a given system of particles larger than say 2 mm, the critical velocity as given by Eq.
4, is a function of pipe diameter alone, varying as a square root of D.

Wasp et al. (1970) have presented a review of the published data for critical velocities.
Fig. 6b shows a plot of F L versus solids concentration for sand-water slurries and

Figure 6. Variation of Durand’s coefficient FL

521
Sediment Transport Technology

includes data from Durand (1952), Sinclair (1962), Yotsukura (1961) and Wicks (1968),
together with the correlation line given by Eq. 4 and Fig. 6a. The Durand correlation
gives excellent prediction of the data with the exception of Wicks, which are for low
solids concentrations (one percent by volume) and a varying fluid density and viscosity.

Although the Durand expression shows excellent agreement for sand-water slurries, it is
inadequate for suspensions of coal and iron particles as reported by Sinclair (1962). It
must be concluded, therefore, that Eq. 4 does not fully describe the function of relative
density as it affects critical velocity.

Several other empirical relations for critical velocity have been proposed, some of
which are:

Spells (1955):

0.775
⎛ Dρ m ⎞
V 1.225
= 0.025 gd ⎜⎜ ⎟⎟ (s − 1) (5)
⎝ µm ⎠
c

Zandi and Gavatos (1967):

Dg (s − 1)
Vc2 = 40C v (6)
CD

where CD is the drag coefficient of the particle settling in clear water.

Hungmark (1961)

Vc
= f (C v (s − 1)FD ) (7)
gD

where the function is plotted in Fig. 7, and the coefficient FD which is dependent on
particle size is shown.

Figure 7. Hughmark’s function f

522
 Hydraulic Transport of Solids in Pipes

Predicted values of the critical velocity, by the above expressions, show a wide
variation. While all these studies are basically experimental in nature, Thomas (1964)
attempted a theoretical development followed by an experimental verification and
proposed the following expression for critical velocity.

1/ 3
Vc* ⎛W ⎞
*
= 1 + 2.8⎜ * ⎟ Cv1 / 2 (8)
V ⎝V ⎠

where Vc* is friction velocity at critical condition for the slurry at concentration Cv, and
V* is friction velocity for the flow of clear water.

The use of Eq. 8 requires prediction of Vc* as a function of velocity. Thus the procedure
will first involve determination of velocity and then the establishment of the value of
critical velocity by using Eq. 8.

In the derivation of above expressions either uniform sized slurry has been used in the
experiments, or a single representative size such as d50 or d85 is used in concentrations.
Uniform sized particles are seldom encountered; when particles of mixed sizes are
encountered, mechanism of deposition becomes very complicated.

Kökpınar and Göğüş (2001) recently investigated the critical flow velocity for both
uniform and non-uniform solid-liquid mixtures. They proposed an empirical equation
(Eq. 9) including the effect of settling velocity of a solid particle in a concentrated
medium. The equation gives quite good predictions for both uniform and non-uniform
materials.

−0.60
Vc ⎛d ⎞ ⎛ ρ f wm d s ⎞
Cv0.27 (s − 1) ⎜ ⎟
0.07
= 0.055⎜ s ⎟ (9)
⎝D⎠ ⎜ µ ⎟
gD ⎝ f ⎠

where ds is particle diameter (i.e. d50 for non-uniform and dn (nominal diameter) for
uniform materials) and wm is the particle settling velocity at concentration Cv.

2.4. Friction loss in solid-liquid flow

Friction loss for clear water flowing through a pipe can be expressed by the Darcy-
Weisbach equation:

L V2
hf = f (10)
D 2g

where hf is friction loss (in height of liquid column), f is friction factor, L is pipe length,
D is pipe diameter, and V is mean flow velocity.

Variation of friction factor with Reynolds number and/or relative roughness is well
established and provided in all hydraulic texts in the form of Moody Diagram.

523
Sediment Transport Technology

Mechanism of friction loss in solid-liquid flows, however, is much more complicated. A

typical friction loss-velocity response for a solid-liquid flow is shown in Fig. 8. For very
high velocities (point 1) solids are carried as homogeneous slurry with almost uniform
concentration throughout the pipe, and the friction loss-velocity response tends towards
a position parallel to the response of clear liquid. As the velocity is decreased (point 2)
the distribution of solids becomes more and more pronounced and non-deposit-
heterogeneous flow conditions prevail until the velocity is reduced to the critical value
(point 3). If the velocity is further reduced below the critical value, deposition starts, a
thin stationary or sliding bed appears on the bottom of the pipe, and the volume
concentration of the transported solids decrease. These deposit heterogeneous flow
conditions (points 4 and 5) prevail until the velocity is reduced to such a low value that
all the solids deposit and form a stationary bed. During the deposit-heterogeneous
regime even some bed forms (ripples, dunes) may be created. In this particular
illustration, the amount of rate of solids injected to the pipeline has been kept constant.
At high velocities all the injected solids were carried (Cv = constant), but as the velocity
is reduced below critical value more and more solids started to deposit, thereby
reducing the volume concentration of the solids being transported. If one wants to keep
the volume concentration of the solids in transport constants (same in deposit and
nondeposit regimes) then a typical friction loss-velocity response would be like the one
shown in Fig. 9.

In homogeneous flow conditions friction loss-velocity response is parallel to the

response of clear water. For the same velocity, friction loss increases indirect proportion
to the volume concentration of solids.

Figure 8. Friction losses in solid-liquid flows

524
 Hydraulic Transport of Solids in Pipes

Figure 9. Friction loss with constant concentration

For heterogeneous flow conditions, minimum friction loss occurs at critical velocity, i.e.
at the start of deposition. Therefore, in practice the solid-liquid pipelines are designed to
operate at a velocity slightly above the critical velocity.

2.4.1. Friction loss-homogeneous flow

When very fine particles (i.e. less than 50 microns) are placed in suspension in water
they generally form above a minimum concentration, plastic and non-Newtonian slurry.
If the flow is in the laminar regime, flow conditions are governed to a great extent on
the rheological behavior of the slurry, (Fig. 10), and hence a very thorough rheological
study is necessary. However, correct operation of pipeline with homogeneous slurry
requires that the solids be transported under turbulent flow conditions to avoid settling
out in the pipe. For in the absence of turbulence there would be no force to suspend
even trace amounts of heterogeneous particles. Consequently, it becomes important to
determine the minimum velocity above which turbulent flow occurs. An expression for
this minimum velocity, Vmin is proposed by Durand for the Bingham fluids as follows:

Vmin =
1000
D
[
1 + 1 + D 2τ y0 / 3000η ] (11)

where, η is the plasticity, τy is the yield stress (Fig. 10), ρ is the density of the slurry, D
is the pipe diameter.

Above this velocity (i.e. in turbulent flow), the friction losses for homogeneous slurry
can be expressed as a height of mixture, hm, by the formula;

525
Sediment Transport Technology

Figure 10. Rheology of plastic slurries

L V2
hm = f (12)
D 2g

This is an experimental result from which it can be seen that the friction losses
expressed as a height of mixture are the same as with clear water (see Eq. 9). This is
generally known as the mixture density rule.

A typical experimental result is shown in Fig. 11, where the friction loss in terms of
height of mixture is plotted as a function of velocity. In this logarithmic plot, the clear-
water friction losses appear as a straight line at a slope of 2:1, and coincide with the
slurry friction losses for turbulent flow. Friction losses for laminar flows of the slurry
are noticeably greater than those for clear water and their dependence on velocity is less
pronounced.

Figure 11. Head loss in homogeneous flow

526
 Hydraulic Transport of Solids in Pipes

2.4.2. Friction losses for heterogeneous flows

In practice, heterogeneous slurries are carried in nondeposit regime, i.e. with a velocity
higher than the critical velocity. This flow is always turbulent since the carrier liquid is
invariably water which is never in laminar flow under the conditions of pipe diameter
and velocity encountered in commercial applications. Most of the fundamental work in
heterogeneous flow is understandably experimental in nature.

The work of Durand (1952) is perhaps the most comprehensive one. Durand’s
experiments covered a wide range of conditions with pipe diameters varying from 2 to
70 cm, sand and coal particle sizes from 100 microns to 2.5 cm, pipe velocities from 0.6
to 6 m/s, and volume concentrations from 2 to 23 percent. The final correlation for
horizontal heterogeneous flows in non-deposit regime can be expressed as (Wasp 1978):

⎡ ⎡ gD(s − 1) ⎤
3/ 2
⎤
⎢
f m = f1 1 + K ⎢ 2 ⎥ Cv ⎥ (13)
⎢ ⎣⎢ V C D ⎦ ⎥ ⎥
⎣ ⎦

where, fm; friction factor for mixture, f1; friction factor for liquid, Cv; volume concentration
of solids, K; constant (80 to 150), D; pipe diameter, V; velocity of flow, s; relative
density of solids, and CD; particle drag coefficient.

By regarding Eq. 13, the increase in friction loss of the slurry over that of the carrier
fluid flowing under the same conditions, expressed as a fraction of the carrier fluid
friction loss is;

−3 / 2
f m − f1 ⎡V 2 1 ⎤
= 82C v ⎢ CD ⎥ (14)
f1 ⎣ gD ( s − 1) ⎦

This incremental loss is seen to be a function of the Froude number and volume
concentration, relative density and drag coefficient of the solids.

Durand’s work was carried out using closely graded size particles and apart from some
disagreement on the numerical value ascribed to the constant K in the equation is
generally accepted.

For the more realistic situation of a heterogeneous flow in which the particle size varies
over a range, Condolius and Chapus (1963) proposed to calculate a weighted mean
value of the drag coefficient for the solids distribution, and to employ this value of CD
in Eq. 14. With this modification Eq. 13, has become to known as Durand-Condolius
Equation.

3. Operating Solid-Liquid Pipelines

3.1. Types and choice of pumps
In solid-liquid pipelines, centrifugal, positive displacement, jet pump or some special
pumps may be employed. Following is a brief summary of their general operating
characteristics:

527
Sediment Transport Technology

3.1.1. Centrifugal pumps

Centrifugal pumps are used typically for short distances, such as in-plant transportation
of slurries in the mining, cement or other industries. On long distance (several
kilometers) lines, they generally serve as charge pumps providing suction pressure
required for the main reciprocating pumps. In order to provide wear-resistance, a rubber
lining (for fine particles) or a special wear-resistant alloy (for coarse particles) may be
used in the mechanical design of centrifugal pumps.

The head versus flow characteristics for a centrifugal slurry pump is flat. Thus if flow
rate drops near to the critical velocity where the system head loss curve hooks the pump
characteristics could cause the flow area to drop below critical velocity. This will result
in deposition which could develop into plunged pipeline. This problem is overcome by
installing one or two units with variable speed drives.

The major limitations of centrifugal are their low discharge heads and low efficiencies
compared to positive displacement pumps (Table 3). They provide, however, lower
initial cost, smoother and more flexible operation and capability to handle coarser
particles while requiring less maintenance compared to positive displacement pumps.

3.1.2. Positive displacement pumps

This type of pump has been used for pumping fine slurries since the early twentieth
century. It was first applied for high-pressure pumping in oil industry and more recently
has been used in solid-liquid pipelines in minerals industry.

Figure 12. Slurry pumps

528
 Hydraulic Transport of Solids in Pipes

Table 3. Characteristics of slurry pumps

Max.
Max. pressure Max. flow Effciency
Type particle size Note
(m) (m3/s) (%)
(mm)

Centrifugal 500-4000 3.20 40-75 150 Series installation

Low capital cost

Piston 1750-2100 0.17 85-90 2.5

Plunger 2450-2800 0.06 85-90 2.5 Used for abrasive slurries

There are three basic types of positive displacement pumps (plunger, piston and
diaphragm) all of which provide high efficiency at any desired flow rate. Selection of a
particular type of a positive displacement pump is generally governed by the degree of
abrasiveness of the slurry.

Piston pumps may be used for low-abrasion slurries such as coal and mud, with solid
sizes limited to 2.5 mm or lower. They are more suitable for high-volume flows and
pressure up to 2000 m. Plunger pumps are considered for high-abrasion fine mixtures
such as magnetite, silica and sand slurries. This type is usually considered for pressures
above 2000 m.

Diaphragm pump is a variation of Piston pump, where all working parts (with the
exception of valves) are isolated from contact with the abrading slurries, and hence it is
less likely to wear. Diaphragm pumps are particularly suited to the transportation of fine
slurries, in low capacities and pressures.

Valves are the most important element in positive displacement pumps, for no matter
what type is adopted, whether piston, plunger, or diaphragm, the valves can not be
isolated from slurry. Consequently these pumps are limited in application because of
valve wear and the ability of the valves to pass large solids.

3.1.3. Jet pump

Jet pump is a device which is operated by a pump producing a driving jet of clean liquid
through a nozzle which will entrain the solid-liquid mixture (Fig. 12). The momentum
change between the driving jet and entrained mixture occurs through a mixing tube, and
the high kinetic energy is then converted into pressure energy through a diffuser. Jet
pump does not contain any moving parts and hence it is less affected by abrasion.
Further, jet pumps are capable of high suction lifts, making them suitable for deep see
dredging.

3.1.4. Special pumps

Some unique types of pumps have been developed for special applications in slurry
pipelines. These are; Mars, Moyno, Slug and Lockhopper pumps (Wasp 1978). The
latter two, however, are still in development and improvement stage.

529
Sediment Transport Technology

Mars pumps are basically a piston pump which utilizes an oil barrier in the piston-liner
portion of the pump. The oil prevents contact between the pistons and cylinders and the
abrasive slurry thus greatly reducing wear.

Moyno pump is a variation of the screw principle that utilizes a single helical rotor
within a cylindrical stator. As the rotor turns cavities in which the material is trapped
advance toward the discharge end of the unit. Moyno pump has been used for sand and
cement grout and centrifuge cakes.

3.2. Operational reliability

Two major concerns in the operation of solid-liquid pipelines are the dangers of
mechanical blockage and hydraulic instability.

In order to prevent blockage of the pipeline the velocity has to be equal to or preferably
larger than the critical velocity. The pipe diameter is recommended to be equal to or
greater than, three times the diameter of the largest solid particles. On the other hand, in
order to prevent blockage of the pump, the maximum particle diameter must be smaller
than one-half the size of the smallest passage in the pump (Condolius, 1978).

Hydraulic stability of operation is determined by the comparison of the pump

characteristic (head-discharge) curve. For a stable operation, pump characteristics must
be above the system characteristics, for all discharges below the operating point. The
larger the angle between the two characteristic curves, the greater will be the stability
around this operating point. It will, therefore, be possible to handle safety accidental
variations of operating conditions-for example an accidental increase in concentration
or a decrease in pump power.

Fig. 13 shows two system characteristic curves corresponding to solid discharges Qs1
and Qs2, for a horizontal pipeline. Assuming solids discharge will fluctuate between

Figure 13. Characteristic curves

530
 Hydraulic Transport of Solids in Pipes

these two values, the cross-hatched area will denote the possible range of operating
conditions for the pipeline. The shaded line VL indicates the border between deposit and
nondeposit regimes of transport, and the curve Je shows the head losses for clear water.
On the same figure, head-discharge curves for:

I- Gravity installation
II- Constant speed centrifugal pump
III- Piston pump or variable-speed centrifugal pump with constant discharge

are plotted.

The gravity installation is fairly stable at the lower concentration Qs1 but if the solids
discharge approaches maximum, Qs2 the pipeline would begin to deposit material and
become blocked along its entire length.

In the case of constant speed centrifugal pump (curve III), the system is very stable for
Qs1, but barely stable for Qs2. The least increase in solid discharge over Qs2 would
initiate total blockage of the pipeline.

In the case of piston or variable-speed centrifugal pump (curve III) stability of operation
is safely ensured for all solid discharges.

Finally, for a reliable operation of solid-liquid pipeline start-up and shut-down

procedures must be well established. A system featuring a single pump delivering to a
horizontal or near-horizontal pipe is fairly easy to start-up and shot-down. Both
operations should, however, be carried out with clear or almost clear water and by
increasing or decreasing the pump speed very gradually to avoid water hammer. A plant
with pumps in series is trickier to start up, and this should be done, in accordance with
the results of surge analysis in a predetermined order.

A plant with vertical or steeply inclined pipes can not be shut down while there is steel
solid material in the pipeline, as this is likely to lead to blockage. Suitable emptying
arrangements have to be provided for use in a sudden shut down.

3.3. Concepts of pipeline installations

Over the years numerous solid-liquid pipelines have been designed, built and operated.
Condolius (1963) have summarized all the different concepts of such installations with
typical diagrams; these diagrams are reproduced in Fig. 14. Condolius (1978) described
their installation as follows:

3.3.1. Gravity sesign (Types A)

Gravity-plant operations costs are very low, often less than for any other type
transportation. A gravity plant may operate either under a constant head A1 or at a
constant rate of flow A2, with delivery ranges depending on the head available and the
physical properties of the material to be conveyed. Under favorable topographical
conditions such installations can be made to deliver over long distances.

531
Sediment Transport Technology

Nomenclature
Qs Solids Qf Liquid before overflow
Q Liquid to be discharged q Overflow or additional liquid

Figure 14. Hydraulic stability of solid-liquid pipeline

3.3.2. Designs with solids-handling pumps (Types B)

These are the most common types of system for hydraulic transportation over any
distance. Frequently they are integrated with the other circuits in the plant or mine,
connecting one or more stations along a production line and satisfactorily replacing
conventional handling equipment, particularly when circuits are elaborate and
complicated.

When constant-speed centrifugal pumps are used, the rate of flow is controlled by a
constant-level tank (Type B1).

In other types of installation with pumps driven by electric motor or diesel engine flow
control is affected by varying the pump speed.

532
 Hydraulic Transport of Solids in Pipes

Long-distance lines, i.e. over several dozen miles, may feature several pumping
stations, with one or more pumps to a station, which may be either constant- or variable-
speed type. Economic conditions need very close consideration for this type of line,
especially if it is to convey coarse material or waste.

3.3.3. Designs with clear water pump and lock chambers (Types C)
For special applications, solids can be fed into the line after the pump. The pump may
then be of a conventional type for water, e.g., single or multi-stage centrifugal,
reciprocating or membrane type. If the material is fine and is to be conveyed over only a
short distance, ejectors can be employed to feed the solids into the line (Types C1 and
C2). If the material has larger dimensions or is very abrasive, and if transport distance is
long or vertical, then it is advisable to introduce the material by means of a lock system.
This system may be of the single or multiple type, depending on whether solids
transport rate can be intermittent or not (Types C3 and C4).

3.3.4. Dredging and back-filling installations (Types D, E and F)

These involve suction of the solids from rivers, sea, dam reservoirs, settling basins, etc.
The pumping station is usually mobile, whether it is installed on a floating pontoon, on
a dredge, or on land-bound rigs, as on a truck or on rails, on an overhead traveling
crane, etc.

Type F permits the dredging of deposits located in deep water.

References

1. ASCE, “Sedimentation Engineering”, Manuals and Reports on Engineering

Practice, editor: V.A. Vanoni, American Society of Civil Engineering, No.54,
(1975).
2. Aude, T. L., Cowder, N. T., Thomson, T. L., Wasp, E. J., “Slurry Piping Systems:
Trends, Design Methods, Guidelines”, Chemical Eng., June, (1971).
3. Bain, A. G., Bonnington, S. T., “The Hydraulic Transport of Solids by Pipeline”,
Pergamon Press, London, (1970).
4. Çeçen, K., Beyazıt, M., Sumer, M., Avcı, İ., “Slurry Pipelines Technical,
Operational Characteristics, Construction, Operation Costs”, Istanbul Technical
University, (in Turkish), (1976).
5. Condolios, E., Chapus, E. E., “Operating Solids Pipelines”, Chemical Eng., June-
July, (1963).
6. Condolios. E., Chapus, E. E., “Transporting Solid Materials in Pipelines”, Chemical
Eng., June-July, (1963).
7. Condolios, E., Chapus, E. E., “Hydraulic Transport of Solids”, Pump Handbook,
editor: Karassik, I. J. et al., McGraw Hill, (1976).
8. Durand, R., Condolios, E., “Experimental Investigations on the Transport of Solids
in Pipes”, Le Journel d’Hydraulique, Societe Hydrotechnique de France, Grenoble,
(1952).
9. Durand, R., “Basic Relationships of the Transportation of Solids in Pipes-
Experimental Research”, Proc. 5th Congress of International Association of
Hydraulic Research, Minneapolis, Minn., (1953).

533
Sediment Transport Technology

10. Hungmark, G. A., “Aqueous Transport of Settling Solids”, Industrial and

Engineering Chemistry, Vol. 53, No. 5, (1961).
11. Graf, W., “Hydraulics of Sediment Transport”, Series in Water Resources and
Environmental Engineering, McGraw-Hill, New York, (1971).
12. Kökpınar, M. A., Göğüş, M., “Critical Flow Velocity in Slurry Transporting
Horizontal Pipelines”, ASCE Journal of Hydraulic Research, Vol. 127, No:9,
pp.763-771, (2001).
13. Matousek, V., “Solid Transportation in a Long Pipeline Connected with a Dredge”,
Terra et Aqua, Vol. 62, pp.3-11, (1996).
14. Newitt, D. M., Richardson, J. F., Abbitt, M., Turtle, R. B., “Hydraulic Conveying of
Solids in Horizontal Pipes”, Transactions, Inst. of Chem. Eng., Vol. 33, (1955).
15. Sinclair, C. G., “The Limit Deposit Velocity of Heterogeneous Suspensions”,
Chemical Eng., London, (1962).
16. Spells, K. E., “Correlation for Use in Transport of Axueous Suspensions of Fine
Solids through Pipes”, Transactions, Inst. of Chem. Eng., Vol. 33, (1955).
17. Swamee, P. K., “Design of Sediment Transporting Pipeline”, ASCE Journal of
Hydraulic Engineering, Vol. 121, No.1, 72-76, (1995).
18. Thomas, D. G., “Transport Characteristics of Suspensions”, American Inst. of
Chem. Eng. Journal, Vol. 10, May, (1964).
19. Wasp, E. J., Thompson, T. L., Aude, T. C., “Process Basis and Economics of
Pipeline Systems”, The Canadian Mining and Metallurgial Bulletin, Dec., (1970).
20. Wasp, E. J., Thompson, T. L., Aude, T. C., “Initial Economic Evaluation of Slurry
Pipeline Systems”, Proc. Amer. Soc. of Civil Eng., Vol. 97, May, (1971).
21. Wasp, E. J., Kenny, P. J., Gandhi, R. L., “Solid-Liquid Flow Slurry Pipeline
Transportation”. Gulf Pub. Com., Houston, (1979).
22. Wicks, M., “Transportation of solids at Low Concentrations in Horizontal Pipes”,
Amer. Soc. of Civil Eng. Symposium on Solid-Liquid Flow in Pipes, University of
Pennsylvania, March, (1968).
23. Yotsukura, N., “Some Effects of Bentonite Suspensions on Sand Transport”, Ph.D.
Dissertation, Colorado State University, (1961).
24. Yücel, Ö., Sevük, S., Ger, M., “Technical and Financial Feasibility for Hasançelebi-
Iskenderun Iron Ore Slurry Pipeline”, Middle East Technical University, (in
Turkish), February, (1978).
25. Zandi, I., Gavatos, G., “Heterogeneous Flow of Solids in Pipeline”, Proceedings,
Amer. Soc. of Civil Eng. Hydraulic Division, May, (1967).

534