1.

OVERVIEW OF SEDIMENT TRANSPORT AUTUMN 2020

1.1 Introduction

In natural channels and bodies of water the bed is not fixed but
is composed of mobile particles; e.g. gravel, sand or silt. These
may be dislodged and moved by the flow – the process of
sediment transport.

Many large rivers are famed for their sediment-carrying capacity: the Yellow River in China
is coloured by its sediment content, and for centuries Egyptian farmers have relied on the rich
deposits carried by the Nile. In many cases erosion occurs but is not noticed because there is
a dynamic equilibrium established whereby, on average, as much sediment is supplied as is
removed from an area. However, short-term events (such as severe storms) and man-made
structures (such as dams) can severely disrupt
this equilibrium. Chanson’s book contains
salutary details of reservoirs rendered useless
by siltation and bridge failures because of scour
around their foundations. The Nile delta is
eroding because the sediment supply from
upstream is being held up by the Aswan Dam.
Bridge piers are highly susceptible to short-
term or long-term scour around their
foundations (see the CFD simulation right),
whilst road and rail transport are often
disrupted by surface damage in flash floods.

Models exist to address three basic questions.

• Does sediment transport occur? (“Threshold of motion”).
• If it does, then at what rate? (“Sediment load”).
• What effect does an imbalance have on bed morphology? (“Scour vs accretion”).

In general, two modes of transport are recognised:

• bed load: particles sliding, rolling or saltating (making short jumps), but remaining
essentially in contact with the bed;
• suspended load: finer particles carried along in suspension by the turbulent fluid flow.
The combination of these two is called the total load.

To predict sediment transport we need to consider:

• particle properties: diameter, specific gravity, settling velocity, porosity;
• flow properties: bed stress, velocity and turbulence profiles.

Finally, we note that, although our primary interest here is in sediment transport by water,
similar processes can be observed in air (“aeolian” or wind-borne transport). Examples are
the raising of dust clouds and sand-dune movement across deserts.

Hydraulics 3 Sediment Transport - 1 Dr David Apsley

Besides overall transport of material, sediment transport gives rise to some classic bedforms:
• ripples (fine particles; Fr ≪ 1; wavelength depends on particle size, not flow depth);
• dunes (Fr < 1; migrate in the direction of flow);
• standing waves (Fr = 1; bed undulations in phase with free-surface standing waves);
• antidunes (Fr > 1; migrate in the opposite direction to flow).

FLOW
Dunes

erosion deposition

bedform migration

FLOW Standing waves

FLOW

erosion deposition Antidunes

bedform migration

Hydraulics 3 Sediment Transport - 2 Dr David Apsley

1.2 Particle Properties

1.2.1 Diameter, 𝒅

Since natural particles have very irregular shapes the concept of diameter is somewhat
imprecise. Common definitions include:
• sieve diameter – the finest mesh that a particle can pass through;
• sedimentation diameter – diameter of a sphere with the same settling velocity;
• nominal diameter – diameter of a sphere with the same volume.

A typical size classification is given below.

Type Diameter
Boulders > 256 mm
Cobbles 64 mm – 256 mm
Gravel 2 mm – 64 mm
Sand 0.06 mm – 2 mm
Silt 0.002 mm – 0.06 mm
Clay < 0.002 mm (cohesive)

Where there is a range of particle sizes the cumulative percentage is attached to the diameter;
e.g. the median diameter 𝑑50 is that sieve size which passes 50% (by weight) of particulate,
whilst a measure of spread is the geometric standard deviation σ𝑔 = (𝑑84.1 /𝑑15.9 )1/2 . (The
percentiles assume a lognormal size distribution.)

This introductory course (and most models) will simply refer to a diameter 𝑑.

1.2.2 Specific Gravity, 𝒔

The specific gravity (or relative density) s is the ratio of particle density (ρ𝑠 ) to that of the
fluid (ρ):
ρ𝑠
𝑠= (1)
ρ
Quartz-like minerals have density  2650 kg m–3, so the relative density (in water) is 2.65.
drag

1.2.3 Settling Velocity, 𝒘𝒔

The settling velocity is the terminal velocity in still fluid and may be found
buoyancy, mw g
by balancing drag against submerged weight; i.e. ws

𝑑𝑟𝑎𝑔 = 𝑤𝑒𝑖𝑔ℎ𝑡 − 𝑏𝑢𝑜𝑦𝑎𝑛𝑐𝑦

1 π𝑑 2 π𝑑 3
𝑐𝐷 ( ρ𝑤𝑠2 )( ) = (ρ𝑠 − ρ)𝑔
2 4 6
weight, mg
where 𝑐𝐷 is the drag coefficient. This rearranges to give
1/2
4 (𝑠 − 1)𝑔𝑑
𝑤𝑠 = [ ] (2)
3 𝑐𝐷

Hydraulics 3 Sediment Transport - 3 Dr David Apsley

Stokes’ law for the force on spherical particles at small Reynolds numbers gives
24 𝑤𝑠 𝑑
𝑐𝐷 = Re ≡ <1
Re ν
whence (after some algebra):
1 (𝑠 − 1)𝑔𝑑 2
𝑤𝑠 = (3)
18 ν

or, in non-dimensional form:

1/3
𝑤𝑠 𝑑 1 (𝑠 − 1)𝑔𝑑 3 1 ∗3 ∗
(𝑠 − 1)𝑔
= = 𝑑 where 𝑑 = 𝑑[ ]
ν 18 ν2 18 ν2

However, this is valid only for very small, spherical particles (diameter < 0.1 mm in water).
For larger grains (and natural shapes) a useful empirical formula is that of Cheng (1997):
𝑤𝑠 𝑑
= [ (25 + 1.2𝑑 ∗ 2 )1/2 − 5] 3/2 (4)
ν

One of the major uses of the settling velocity is in determining whether suspended load
occurs, and the concentration profile in the water column that results (see Section 4).

1.2.4 Porosity, 𝑷

The porosity 𝑃 is the ratio of voids to total volume of material; i.e. in a volume 𝑉 of space
there will actually be a volume (1– 𝑃)𝑉 of sediment and volume 𝑃𝑉 of fluid.

Porosity is important in modelling changes to bed morphology and the leaching of pollutants
through the bed. For natural uncompacted sediment 𝑃 is typically about 0.4.

1.2.5 Angle of Repose, 𝝓

The angle of repose is the maximum angle (to the horizontal) which a pile of sediment may
adopt before it begins to avalanche. It is easily measured in the laboratory.

The resistance to incipient motion may be quantified by an R

F
effective coefficient of friction μ𝑓 . Consider the basic mechanics
problem of a particle on a slope. Incipient motion occurs when the bed
downslope component of weight equals the maximum friction
force (μ𝑓 × 𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛). Then mg

𝑚𝑔 sin ϕ = μ𝑓 (𝑚𝑔 cos ϕ)
or
μ𝑓 = tan ϕ

Hydraulics 3 Sediment Transport - 4 Dr David Apsley

Although the mechanism for causing motion is not the same, μ𝑓 can then be used to estimate
the effect of gravitational assistance on sloping beds, where both fluid drag and downslope
component of weight act on the particles of the bed (see Section 2).

1.3 Flow Properties.

1.3.1 Friction Velocity, 𝒖𝝉

The bed shear stress τ𝑏 is the drag (per unit area) of the flow on the granular bed. It is
responsible for setting the sediment in motion.

As stress has dimensions of [𝑑𝑒𝑛𝑠𝑖𝑡𝑦] × [𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦]2 it is possible to define from τ𝑏 an

important stress-related velocity scale called the friction velocity (or shear velocity) 𝑢τ such
that:
τ𝑏 = ρ𝑢τ2 or 𝑢τ = √τ𝑏 /ρ (5)

1.3.2 Mean-Velocity Profile

It may be shown (elsewhere!) that a fully-developed turbulent boundary layer adopts a

logarithmic mean-velocity profile. For a rough boundary this is of the form
𝑢τ 𝑧
𝑈(𝑧) = ln(33 ) (6)
κ 𝑘𝑠
where 𝑢𝜏 is the friction velocity, κ is von Kármán’s constant (a famous number with a value
of about 0.41) and 𝑘𝑠 is the roughness height (typically 1 – 2.5 times particle diameter). 𝑧 is
the distance from the bed.

1.3.3 Eddy-Viscosity Profile

A classical model for the effective shear stress 𝜏 in a turbulent flow is to assume, by analogy
with laminar flow, that it is proportional to the mean-velocity gradient:
d𝑈 d𝑈
τ = μ𝑡 or τ = ρν𝑡 (7)
d𝑧 d𝑧
μ𝑡 is the eddy viscosity. (𝜈𝑡 = μ𝑡 /ρ is the corresponding kinematic eddy viscosity). μ𝑡 is not
a true viscosity, but a means of modelling the effect of turbulent motion on momentum
transport. Such models are called eddy-viscosity models and they are widely used in fluid
mechanics. In a fully-turbulent flow μ𝑡 is many times larger than the molecular viscosity μ.

At the bed, the shear stress is τ = τ𝑏 ≡ ρ𝑢τ2. At the free surface (𝑧 = ℎ), in the absence of
significant wind stress, τ = 0. In fully-developed flow it may be shown that the shear stress
varies linearly across the channel; hence, to meet these boundary conditions it is given by
𝑧
τ = ρ𝑢𝜏2 (1 − ) (8)
ℎ
Differentiating the mean-velocity profile (6) we find:

Hydraulics 3 Sediment Transport - 5 Dr David Apsley

 d𝑈 𝑢τ
= (9)
d𝑧 κ𝑧
Substituting these expressions for stress and mean-velocity gradient into (7) gives
𝑢τ
ρ𝑢τ2 (1 − 𝑧/ℎ) = ρν𝑡
κ𝑧
whence:
ν𝑡 = κ𝑢τ 𝑧(1 − 𝑧/ℎ) (10)
The kinematic eddy viscosity 𝜈𝑡 therefore has a parabolic profile (in channel flow).

z z z

U tb nt

1.3.4 Formulae For Bed Shear Stress

Many sediment-transport formulae rely on knowledge of the bed shear stress τ𝑏 , which is
what sets the particles in motion and determines the bed load.

Recall that in normal flow there is a balance between the downslope component of gravity
and bed friction, leading to:
τ𝑏 = ρ𝑔𝑅ℎ 𝑆 (𝑅ℎ is the hydraulic radius; 𝑆 is the slope) (11)

Alternatively, by definition of the (skin-)friction coefficient 𝑐𝑓 :

1
τ𝑏 = 𝑐𝑓 ( ρ𝑉 2 ) (𝑉 is the channel-average velocity) (12)
2
If you are lucky, 𝑐𝑓 may be given; (probably the only possibility in air). Otherwise one could
adopt one of the following approaches. Both assume fully-developed flow (although this
assumption is regularly stretched.)

In normal flow, if the discharge is known then Manning’s equation:

1 2/3 1/2
𝑉= 𝑅 𝑆 (13)
𝑛 ℎ
may be used to determine the hydraulic radius and thence, from (11), the bed shear stress. A
useful correlation when the bed consists of granular material is (the dimensionally-
inconsistent) Strickler’s equation:
𝑑1/6
𝑛= (14)
21.1
where 𝑑 is the particle diameter in m. This gives 𝑛 = 0.015 m–1/3 s for a grain size of 1 mm.

Hydraulics 3 Sediment Transport - 6 Dr David Apsley

Alternatively, one can find an analytical expression for depth-averaged velocity 𝑉 by
integrating the velocity profile (6) to get (exercise: do it!):
1 ℎ 𝑢τ 12ℎ
𝑉= ∫ 𝑈 (𝑧) d𝑧 = ln( ) (15)
ℎ 0 κ 𝑘𝑠
The skin friction coefficient 𝑐𝑓 is then, from its definition:
τ𝑏 ρ𝑢τ2 𝑢τ 2
𝑐𝑓 ≡ = = 2( )
1 2 1 2 𝑉
2 ρ𝑉 2 ρ𝑉
whence
0.34
𝑐𝑓 = (16)
[ln(12ℎ/𝑘𝑠 )]2
Typical values of 𝑐𝑓 are in the range 0.003 – 0.01.

Hydraulics 3 Sediment Transport - 7 Dr David Apsley

2. THRESHOLD OF MOTION

2.1 Shields Parameter

In general, a granular bed will remain still until the flow is sufficient to move it. This point is
called the threshold of motion and the bed stress that initiates it the critical stress, τ𝑐𝑟𝑖𝑡 .

According to a simple frictional model, on a flat bed:

critical stress × representative area = friction coefficient × normal reaction
π𝑑 2 π𝑑 3
τ𝑐𝑟𝑖𝑡 × 𝑐 = μ𝑓𝑟𝑖𝑐𝑡 × (ρ𝑠 − ρ)𝑔
4 6
where 𝑐 is a constant of order unity. Hence,
τ𝑐𝑟𝑖𝑡 2μ𝑓𝑟𝑖𝑐𝑡
=
(ρ𝑠 − ρ)𝑔𝑑 3𝑐
i.e.
τ𝑐𝑟𝑖𝑡
= dimensionless function of particle size and shape (17)
(ρ𝑠 − ρ)𝑔𝑑
In practice, the dependence on shape is not very significant.

The non-dimensional stress

τ𝑏
τ∗ = (18)
(ρ𝑠 − ρ)𝑔𝑑
is called the Shields parameter (in the sediment-transport literature often denoted θ) after
American engineer A.F. Shields, who, in 1936, plotted his results on the initiation of
∗
sediment motion in the form of a graph of 𝜏𝑐𝑟𝑖𝑡 against the particle Reynolds number Re𝑝 :
τ∗𝑐𝑟𝑖𝑡 = 𝑓(Re𝑝 ) (19)
in what became known as a Shields diagram. Here,
𝑢τ 𝑑
Re𝑝 = (20)
ν

In practice, the choice of dimensionless groups in the original Shields diagram is not
convenient for predicting the threshold of motion because the bed stress τ𝑏 effectively
appears on both sides of the equation. (Remember that the friction velocity in Re𝑝 is given by
𝑢τ = √τ𝑏 /ρ.) You will recall from dimensional analysis in Hydraulics 2 that one is at liberty
to replace either of the dimensionless Π groups by another independent combination. In this
case we can eliminate the dependence on bed stress (or 𝑢τ ) by forming
Re2𝑝 (𝑠 − 1)𝑔𝑑 3
=
τ∗ ν2
where 𝑠 = ρ𝑠 ⁄ρ is the relative density. Taking the cube root in order to give a dimensionless
parameter proportional to diameter:
1/3
∗
(𝑠 − 1)𝑔 (21)
𝑑 = 𝑑[ ]
ν2

Hydraulics 3 Sediment Transport - 8 Dr David Apsley

Shields’ threshold line may then be replotted as a function of τ∗𝑐𝑟𝑖𝑡 against 𝑑 ∗ :
τ∗crit = 𝑓(𝑑 ∗ ) (22)

0.15

0.10
t*crit

0.05

0.00
1 10 100 1000
d*

A convenient curve fit to experimental data is provided by Soulsby (1997)1:

0.30
τ∗𝑐𝑟𝑖𝑡 = + 0.055 [1 − exp(−0.020𝑑 ∗ )] (23)
1 + 1.2𝑑 ∗

For large diameters the critical Shields parameter tends to a constant value; this is 0.055 for
Soulsby’s formula, although 0.056 is actually a more popular figure in the literature.

Summary of Calculation Formulae For Threshold of Motion

τ∗𝑐𝑟𝑖𝑡 = 𝑓(𝑑 ∗)
e.g. graphically, or
0.30
τ∗𝑐𝑟𝑖𝑡 = + 0.055 [1 − exp(−0.020𝑑 ∗ )]
1 + 1.2𝑑 ∗
where
τ𝑏
τ∗ = (Shields parameter)
(ρ𝑠 − ρ)𝑔𝑑
1/3
∗
(𝑠 − 1)𝑔
𝑑 = 𝑑[ ] (𝑠 = ρ𝑠 /ρ)
ν2

1
Soulsby, R., 1997, “Dynamics of Marine Sands”, Thomas Telford.

Hydraulics 3 Sediment Transport - 9 Dr David Apsley

Example. (Exam 2007 – part)

An undershot sluice is placed in a channel with a horizontal bed covered by gravel with a
median diameter of 5 cm and density 2650 kg m–3. The flow rate is 4 m3 s–1 per metre width
∗
and initially the depth below the sluice is 0.5 m. Assuming a critical Shields parameter 𝜏𝑐𝑟𝑖𝑡
of 0.06 and friction coefficient 𝑐𝑓 of 0.01:

(a) find the depth just upstream of the sluice and show that the bed there is stationary;

(b) show that the bed below the sluice will erode and determine the depth of scour.

2.2 Inception of Motion in Normal Flow

In the large-diameter limit, particles will move if

τ𝑏
> 0.056
(ρ𝑠 − ρ)𝑔𝑑
But for normal flow, with hydraulic radius 𝑅ℎ and slope 𝑆:
τ𝑏 = ρ𝑔𝑅ℎ 𝑆
Putting these together, and noting that ρ𝑠 ⁄ρ = 2.65 for sand in water shows that, for large
particles in normal flow, the bed will be mobile if
𝑑 < 10.8𝑅ℎ 𝑆 (24)
For example, a uniform flow of depth 1 m on a slope of 10–4 will move sediment of diameter
about 1 mm or less.

Note that this is an estimate, applying for large particles of a particular density in water and
assuming normal flow. If possible it is better to compare τ with τ𝑐𝑟𝑖𝑡 (or τ∗ with τ∗𝑐𝑟𝑖𝑡 ) to
establish whether the bed is mobile.

2.3 Effect of Slopes

The main effect of slopes is to add (vectorially) a downslope component of gravity to the
fluid stress acting on the particles. For a downward slope the gravitational component will
assist in the initiation of motion, whereas for an adverse slope it will oppose it.
flow
flow

 mg mg

Let τ𝑐𝑟𝑖𝑡 be the critical stress on a slope and τ𝑐𝑟𝑖𝑡,0 be the critical stress on a flat bed. Let the
slope angle be β and the angle of repose 𝜙.

Hydraulics 3 Sediment Transport - 10 Dr David Apsley

For slopes aligned with the flow:
sin(ϕ + β)
τ𝑐𝑟𝑖𝑡 = τ𝑐𝑟𝑖𝑡,0 (where β is positive for upslope flow)
sinϕ

For slopes at right angles to the flow:

tan2 β
τ𝑐𝑟𝑖𝑡 = cos β √1 − τ
tan2 ϕ 𝑐𝑟𝑖𝑡,0

For arbitary alignment see, e.g., Soulsby (1997) 1 or Apsley and Stansby (2008)2.

2
Apsley, D.D. and Stansby, P.K., 2008, Bed-load sediment transport with large slopes, model formulation and
implementation within a RANS flow solver, Journal of Hydraulic Engineering, 134, 1440-1451.

Hydraulics 3 Sediment Transport - 11 Dr David Apsley

3. BED LOAD

3.1 Dimensionless Groups

Bed-load transport of sediment is quantified by the bed-load flux 𝑞𝑏 , the volume of non-
suspended sediment crossing unit width of bed per unit time. The tendency of particles to
move is determined by the drag imposed by the fluid flow and countered by the submerged
particle weight, so that 𝑞𝑏 may be taken as a function of bed shear stress τ𝑏 , reduced gravity
(𝑠– 1)𝑔 (where 𝑠 = ρ𝑠 /ρ), particle diameter 𝑑, and the fluid density ρ and kinematic
viscosity 𝜈. A formal dimensional analysis (6 variables, 3 independent dimensions) dictates
that there is a relationship between three non-dimensional groups, conveniently taken as
𝑞𝑏
𝑞∗ = (dimensionless bed-load flux) (25)
√(𝑠 − 1)𝑔𝑑 3
τ𝑏
τ∗ = (dimensionless bed stress, or Shields parameter) (26)
ρ(𝑠 − 1)𝑔𝑑

(𝑠 − 1)𝑔 1/3
𝑑∗ = 𝑑 [ ] (dimensionless particle diameter) (27)
ν2

3.2 Bed-Load Transport Models

Some of the more common bed-load transport formulae are listed in the table below. (For
those containing τ∗𝑐𝑟𝑖𝑡 , bed-load sediment transport is zero until τ∗ exceeds this.)

Reference Formula Comments

Meyer-Peter and
𝑞 ∗ = 8(τ∗ − τ∗𝑐𝑟𝑖𝑡 )3/2
Müller (1948)

Nielsen (1992) 𝑞 ∗ = 12(τ∗ − τ∗𝑐𝑟𝑖𝑡 )√τ∗

0.053 τ∗
Van Rijn (1984) 𝑞∗ = (∗ − 1)2.1
𝑑 ∗ 0.3 τ𝑐𝑟𝑖𝑡

Einstein3-Brown 𝐾exp(−0.391/τ∗ ) ∗
τ < 0.182 2 36 36
𝑞∗ = { 0.465 𝐾 = √ + ∗3 − √ ∗3
(Brown, 1950) 3 𝑑 𝑑
40𝐾τ∗ 3 τ∗ ≥ 0.182

1 τ∗
∗
√τ𝑐𝑟𝑖𝑡
Yalin (1963) 𝑞 ∗ = 0.635𝑟√τ∗ [1 − ln(1 + σ𝑟)] 𝑟 = τ∗ − 1, σ = 2.45
σ𝑟 𝑐𝑟𝑖𝑡 𝑠 0.4

3
Prof. H.A. Einstein was a noted hydraulic engineer – he was the son of Albert Einstein!

Hydraulics 3 Sediment Transport - 12 Dr David Apsley

References

Brown, C.B. (1950). “Sediment Transport”, in Engineering Hydraulics, Ch. 12, Rouse, H.
(ed.), Wiley.

Meyer-Peter, E. and Müller, R. (1948). “Formulas for bed-load transport.” Rept 2nd Meeting
Int. Assoc. Hydraul. Struct. Res., Stockholm, 39-64.

Nielsen, P., (1992). Coastal Bottom Boundary Layers and Sediment Transport, World
Scientific.

Van Rijn, L.C., (1984). “Sediment transport. Part I: bed load transport.” ASCE J. Hydraulic
Eng., 110, 1431-1456.

Yalin, M.S., (1963). “An expression for bed-load transportation”, Proc. ASCE, 89, 221-250.

Note.
(i) Models have been written here in a common notation, which may differ a lot from
that in the original papers.

(ii) The actual derivations of the original models vary considerably. Some are mechanical
models of particle movement; others are probabilistic.

(iii) Several models contain a factor (τ∗ − τ∗𝑐𝑟𝑖𝑡 ) indicating, as expected, that sediment
does not move until a certain bed stress has been exceed. Formulae for the threshold
of motion:
τ∗𝑐𝑟𝑖𝑡 = 𝑓(𝑑 ∗ )
have already been covered in Section 2.

Hydraulics 3 Sediment Transport - 13 Dr David Apsley

4. SUSPENDED LOAD

4.1 Inception of Suspended Load

Particles cannot be swept up and maintained in the flow until the typical updraft of turbulent
motions exceeds their natural settling velocity 𝑤𝑠 . A typical turbulent velocity fluctuation is
of the order of the friction velocity 𝑢τ . Thus, suspended load will occur if
𝑢τ
>1 (28)
𝑤𝑠

In practice, what is deemed to be suspended load rather than bed load is not precise, and
different authors give slightly different numbers on the RHS.

For much gravelly sediment, suspended load does not occur and bed load is the only mode of
sediment transport.

4.2 Turbulent Diffusion

The sediment concentration 𝐶 is the volume of sediment per total volume of material
(fluid + sediment).

Once sediment has been swept up by the flow it is distributed throughout the fluid depth.
However, since it is always tending to settle out there is a greater concentration near the bed.
Because the concentration is greater near the bed, any upward turbulent velocity fluctuation
will tend to be carrying a larger amount of sediment than the corresponding downward
fluctuation. Thus, where a concentration gradient exists turbulent diffusion will tend to lead
to a net upward flux of material, whereas settling leads to a net downward flux. An
equilibrium distribution is attained when these opposing effects are equal.

dC
C+l
dz
C ws
dC
C-l
dz

Suppose that the average concentration at some level 𝑧 is 𝐶. In a simple model an upward
turbulent velocity 𝑢′ for half the time carries material of concentration (𝐶 − 𝑙 d𝐶/d𝑧), where
𝑙 is a mixing length – a typical size of turbulent eddy. The corresponding downward velocity
for the other half of the time carries material at concentration (𝐶 + 𝑙 d𝐶/d𝑧). The average
upward flux of sediment (volume flux × concentration) through a horizontal area 𝐴 is
1 d𝐶 1 d𝐶 d𝐶
𝑢′𝐴(𝐶 − 𝑙 ) − 𝑢′𝐴(𝐶 + 𝑙 ) = −𝑢′𝑙 𝐴
2 d𝑧 2 d𝑧 d𝑧
The quantity 𝑢′𝑙 is written as a diffusivity 𝐾, so that the net upward flux is
d𝐶 (29)
−𝐾 𝐴
d𝑧

Hydraulics 3 Sediment Transport - 14 Dr David Apsley

This is referred to a gradient diffusion (because it is proportional to a gradient!) or Fick’s law
of diffusion. The minus sign indicates, as expected, that there is a net flux from high
concentration to low.

At the same time there is a net downward flux of material 𝑤𝑠 𝐴𝐶 due to settling. When the
concentration profile has reached equilibrium the upward diffusive flux and downward
settling flux are equal in magnitude; i.e.
d𝐶
−𝐾 𝐴 = 𝑤𝑠 𝐴𝐶
d𝑧
or, dividing by area:
d𝐶
−𝐾 = 𝑤𝑠 𝐶 (30)
d𝑧

4.3 Concentration Profile

Because it is the same turbulent eddies that are responsible for transporting both mean
momentum and suspended particulate it is commonly assumed that the diffusivity 𝐾 is the
same as the eddy viscosity 𝜈𝑡 ; i.e. for a wide channel and fully-developed flow:
𝐾 = ν𝑡 = κ𝑢τ 𝑧(1 − 𝑧/ℎ)
where κ is von Kármán’s constant (0.41), 𝑢τ is the friction velocity (= √τ𝑏 /ρ), 𝑧 is the
vertical height above the bed and ℎ is the channel depth. Hence, substituting in (30):
d𝐶
−κ𝑢τ 𝑧(1 − 𝑧/ℎ) = 𝑤𝑠 𝐶
d𝑧
Separating variables and using partial fractions,
d𝐶 𝑤𝑠 1 1
=− ( + ) d𝑧
𝐶 κ𝑢τ 𝑧 ℎ − 𝑧
Integrating between a reference height 𝑧𝑟𝑒𝑓 and general 𝑧 gives, after some algebra:
𝐶 𝑤𝑠 𝑧𝑟𝑒𝑓 (ℎ − 𝑧)
ln = ln
𝐶𝑟𝑒𝑓 κ𝑢τ 𝑧(ℎ − 𝑧𝑟𝑒𝑓 )
whence
𝑤𝑠
𝐶 ℎ/𝑧 − 1 κ𝑢τ
=( ) (31)
𝐶𝑟𝑒𝑓 ℎ/𝑧𝑟𝑒𝑓 − 1

This is called the Rouse profile, and the exponent

𝑤𝑠
(32)
κ𝑢τ
is called the Rouse number after H. Rouse (1937). Typical concentration profiles are shown
below.

Hydraulics 3 Sediment Transport - 15 Dr David Apsley

To be of much predictive use it is necessary to specify 𝐶𝑟𝑒𝑓 at some depth 𝑧𝑟𝑒𝑓 , typically at a
height representative of the bed load. There are many such formulae but one of the simplest is
that of Van Rijn (see, e.g., Chanson’s book):
0.117 τ ∗
𝐶𝑟𝑒𝑓 = min { ( − 1) , 0.65}
𝑑 ∗ τ∗𝑐𝑟𝑖𝑡
1/2
z𝑟𝑒𝑓 τ∗
= 0.3𝑑 ∗ 0.7 ( ∗ − 1) (33)
𝑑 τ𝑐𝑟𝑖𝑡
where 𝑑 ∗ and τ∗ are the dimensionless diameter and stress defined in Section 2.

0.8

0.6
z/h
Rouse number
0.4 0.1
0.2
0.2
0.5
1
0
0 0.2 0.4 0.6 0.8
C/Cref

4.4 Calculation of Suspended Load

Equation (31) describes the concentration profile, but usually the most important quantity is
the total sediment flux.

The volume flux (per unit width) of fluid through a depth d𝑧 in a wide channel is
𝑢 d𝑧
Since concentration 𝐶 is the volume of sediment per volume of fluid, the volume flux of
sediment through the same depth is
𝐶𝑢 d𝑧
Hence, the total suspended flux through the entire depth of the channel is (per unit width):
ℎ
𝑞𝑠 = ∫ 𝐶𝑢 d𝑧 (34)
𝑧𝑟𝑒𝑓

This must be obtained from 𝐶 and 𝑢 profiles by numerical integration – see the Examples.

Hydraulics 3 Sediment Transport - 16 Dr David Apsley

Summary of Suspended-Load Formulae

Incipient suspended load:

𝑢τ
>1
𝑤𝑠

Concentration profile:
𝑤𝑠
𝐶 ℎ/𝑧 − 1 κ𝑢τ
=( )
𝐶𝑟𝑒𝑓 ℎ/𝑧𝑟𝑒𝑓 − 1
where
0.117 τ ∗
𝐶𝑟𝑒𝑓 = min { ( − 1) , 0.65}
𝑑 ∗ τ∗𝑐𝑟𝑖𝑡
1/2
𝑧𝑟𝑒𝑓 τ∗
= 0.3𝑑 ∗ 0.7 ( ∗ − 1)
𝑑 τ𝑐𝑟𝑖𝑡

Velocity profile:
𝑢τ 𝑧
𝑈(𝑧) = ln(33 )
κ 𝑘𝑠

Sediment flux (per unit width):

ℎ
𝑞𝑠 = ∫ 𝐶𝑈 d𝑧
𝑧𝑟𝑒𝑓

Hydraulics 3 Sediment Transport - 17 Dr David Apsley