You are on page 1of 206

Offshore Engineering

Dredging Engineering

DREDGE PUMPS
AND SLURRY TRANSPORT OE4625

Dr. ir. V. Matoušek September 2004


ii
CONTENTS

Preface
Introduction

Part 1 Principles of Mixture Flow in Pipelines

1. Basic Principles of Flow of Liquid and Particles in a Pipeline


1.1 Liquid Flow
Case Study 1
1.2 Solid Particles in a Carrying Liquid
1.3 References
1.4 Recommended Literature

Intermezzo I : Settling Velocity of Solid Particle in a Liquid


I.1 Terminal Settling Velocity of a Spherical Particle
I.2 Terminal Settling Velocity of a non-Spherical Particle
I.3 Hindered Settling Velocity of a Particle
I.4 Typical Values of Parameters Describing a Settling Process
I.5 References
I.6 Recommended Literature
Case Study I

2. Soil-Water Mixture and Its Phases


2.1 Soil Properties
2.2 Liquid Properties
2.3 Mixture Properties
2.4 References
Case Study 2

3. Flow of Mixture in a Pipeline


3.1 Flow Regimes and Patterns
3.2 Mean Mixture Velocity and Its Important Values
3.3 Production
3.4 Frictional Head Loss
3.5 Specific Energy Consumption
3.6 References

iii
4. Modeling of Stratified Mixture Flows
4.1 Empirical Modeling
4.2 Physical Modeling
4.3 Summary: General Trends for Frictional Head Loss and Deposition-Limit
Velocity under Various Flow Conditions
4.4 References
4.5 Recommended Literature
Case Study 4

5. Modeling of Non-Stratified Mixture Flows


5.1 Newtonian Flow of Aqueous Mixture of Sand or Gravel
5.2 Non-Newtonian Flow of Aqueous Mixture of Silt or Clay
5.3 References
5.4 Recommended Literature
Case Study 5

Intermezzo II : Rheological Parameters of Non-Settling Mixtures


II.1 Definitions
II.2 Rheograms
II.3 Rheological Models
II.4 References

6. Special Flow Conditions in Dredging Pipelines


6.1 Inclined Flows
6.2 Empirical Modeling of Inclined Flows
6.3 Physical Modeling of Inclined Flows
Case Study 6
6.4 Unsteady Solids Flows
6.5 References

Part 2 Operational Principles of Pump-Pipeline Systems


Transporting Mixtures

7. Pump and Pipeline Characteristics


7.1 The Bernoulli Equation
7.2 H-Q Curve of a Centrifugal Pump
7.3 H-Q Curve of a Pipeline
7.4 Working Point of a Pump-Pipeline System
7.5 Working Range of a Pump-Pipeline System
7.6 Operation Under the Condition of Continuously Fluctuating Density of Mixture

iv
7.7 Operation Under the Condition of Fluctuating Density of Mixture and Mean
Particle Size in a Pipeline
7.8 Effect of Impeller Parameters on Working Point of a System
7.9 References
7.10 Recommended Literature
Case Study 7.1
Case Study 7.2

8. Operation Limits of a Pump-Pipeline System


8.1 Determination of Required Manometric Pressure in a Pump-Pipeline System
8.2 The Upper Limit for a System Operation
8.3 The Lower Limit for a System Operation
8.4 Effect of Pipe Diameter on Operation Limits
8.5 Effect of Pump Position on Operation Limits
8.6 Operation Limits on a H-Q Diagram of a Pipeline
8.7 Recommended Literature
Case Study 8.1
Case Study 8.2

9. Production of Solids in a Pump-Pipeline System


9.1 Production Range for a Pump-Pipeline System
9.2 Production Limited by a Pipeline Length
9.3 Effect of Pump Position on Production
9.4 Recommended Literature
Case Study 9
Case Study from the VBKO Course

10. Systems with Pumps in Series


10.1 Characteristics of a System with Pumps in Series
10.2 Operational Rules for Pumps in Series
10.3 Control of a System with Pumps in Series
10.4 Location of Boosters Along a Long Dredging Pipeline
10.5 Recommended Literature

11. Literature on Dredging Processes


11.1 Survey of All Dredging Processes
11.2 Hydraulic Transport (Pumps and Pipelines)
11.3 Periodicals

v
vi
PREFACE

These lecture notes cover two parts of the course OE4625 “Dredge Pumps and Slurry
Transport”:

- the hydraulic transport of solids in pipelines and

- the co-operation of pumps and pipeline in a hydraulic dredging system.

An aim of these lecture notes is not only to review the basic rules and models for
handling mixture transport in dredging installations but also to explain the physical
processes governing the mixture transport and their description in predictive models.

The lecture notes contain all information required to pass an examination on the studied
subjects. Principles of mixture flow through a dredging installation are described using
various models in these lecture notes. For the examination an application of one of the
models (preferably the best one to solve the specific problem) is sufficient.

vii
viii
INTRODUCTION

A dredging cycle of a hydraulic dredging system is composed of the following processes:

- the disintegration of soil at the bottom of a waterway or in a borrowing pit

- the mixture forming in a cutter head or in a suction head of a dredge

- the transport of a solid-liquid mixture through an inclined (or vertical) pipeline to


the board of a dredge

- the loading of a dredge hopper with a mixture or the transport of a mixture in a


mostly horizontal pipeline from the board of a dredge to a deposit site

- the deposition of dredged solids, i.e. the unloading of a hopper or the storing of
solids in a deposit site at the end of a mostly horizontal pipeline.

In the lecture notes attention is focused to the transport of mixture in both an inclined (or
vertical) pipeline connecting the bottom of a waterway or a borrowing pit with the board
of a dredge and a horizontal pipeline between the board of a dredge and a deposit site.

ix
Principles of flow of soil-water mixture in pipelines

The characteristics describing a flow of mixture in a pipeline connected with a dredge are
of a major importance for a safety and an economy of a dredging operation. They indicate
whether the transportation is carried out in a regime that avoids a danger of a pipeline
blockage and they determine the amount of soil that can be transported together with the
energy dissipated in the flow to transport the required amount of soil through a pipeline.
Mechanisms governing the mixture flow in a pipeline and their prediction using models
are discussed in the first part of this these lecture notes. An aim is to describe the pipeline
flow characteristics for various chosen mixture velocities and soil concentrations and this
for soils of various particle sizes and pipelines of various diameters.

Co-operation between pumps and pipeline to transport soil-water mixtures

The energy is provided to a pipeline flow by pumps. A combination of the pipeline and
the pump characteristics determines a behavior of a dredging system and a production of
solids by the system. The production is limited by the values of the mixture flow velocity
and concentration that the system is capable to establish in a pipeline. A co-operation
between pumps and a pipeline of a dredging system is discussed in the second part of
these lecture notes.

x
xi
xii
1.
BASIC PRINCIPLES OF FLOW OF LIQUID
AND PARTICLES IN A PIPELINE

1.1 LIQUID FLOW

The principles of the flow of a substance in a pressurised pipeline are governed by the
basic physical laws of conservation of mass, momentum and energy. The conservation
laws are expressed mathematically by means of balance equations. In the most general
case, these are the differential equations, which describe the flow process in general
conditions in an infinitesimal control volume. Simpler equations may be obtained by
implementing the specific flow conditions characteristic of a chosen control volume.

1.1.1 Conservation of mass

Conservation of mass in a control volume (CV) is written in the form: the rate of mass
input = the rate of mass output + the rate of mass accumulation. Thus
d ( mass )
= ∑ (qoutlet − qinlet )
dt
in which q [kg/s] is the total mass flow rate through all boundaries of the CV.

In the general case of unsteady flow of a compressible substance of density ρ, the


differential equation evaluating mass balance (or continuity) is

∂ρ G G
+ ∇. ( ρV) = 0 (1.1)
∂t
G
in which t denotes time and V velocity vector.

For incompressible (ρ = const.) liquid and steady (∂ρ/∂t = 0) flow the equation is
given in its simplest form

∂v x ∂v y ∂v z
+ + =0 (1.2).
∂x ∂y ∂z

The physical explanation of the equation is that the mass flow rates qm = ρVA [kg/s]
for steady flow at the inlet and outlet of the control volume are equal. Expressed in
terms of the mean values of quantities at the inlet and outlet of the control volume,
given by a pipeline length section, the equation is

1.1
1.2 CHAPTER 1

qm = ρVA = const. (1.3).

Thus

(ρVA)inlet = (ρVA)outlet (1.4)

qm mass flow rate [kg/s]


ρ density of flowing liquid [kg/m3]
V mean velocity in a pipe cross section [m/s]
A area of a pipe cross section [m2].

In practice volumetric flow rate Q is often used in place of mass flow rate q. The
volumetric flow rate Q = q/ρ = VA. For a circular pipeline of two different diameters
D1 and D2 (see Fig.1.1) the mass balance claims V1D12 = V2 D 22 .

V1 V2
D1 D2

Figure 1.1. Application of continuity equation.

1.1.2 Conservation of momentum

A momentum equation is an application of Newton's second law of motion. The


summation of all external forces on a control volume filled with a substance is equal
to the rate of change of momentum of the substance in the control volume. The sum
of the external forces acting on the control volume is counterbalanced by the inertial
force proportional to the momentum flux of the control volume

d (momentum)
= ∑ Fexternal .
dt

The external forces are


- body forces due to external fields (gravity, magnetism, electric potential) which act
upon the entire mass of the matter within the control volume,
- surface forces due to stresses on the surface of the control volume which are
transmitted across the control surface.
Gravity is the only body force relevant to the description of the flow of a substance in
a conduit. Surface forces are represented by the force from the pressure gradient and
by friction forces from stress gradients at the control volume boundary.
In an infinitesimal control volume filled with a substance of density ρ the force
balance between inertial force, on one side, and pressure force, body force, friction
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.3

force, on the other side, is given by a differential linear momentum equation in vector
form

∂ G G G G G G G G
∂t
( ) ( )
ρV + V.∇ ρV = −∇P − ρg∇h − ∇.T (1.5)

G G
where h denotes the elevation above a datum, V the velocity vector and T the stress
tensor.

To apply the momentum equation to pipeline flow it is convenient to replace the


infinitesimal control volume by a macroscopic one given by a straight piece of pipe of
the differential distance dx, measured in the downstream direction (Fig. 1.2). The
momentum equation written for this control volume is simpler because quantities in
the equation are averaged over the pipeline cross section. The momentum equation is
obtained by integrating the differential linear momentum equation over the pipe cross
section. For the one-dimensional liquid flow it has the form (Longwell, 1966 or
Shook & Roco, 1991)

 ∂V ∂V ∂h  ∂P τo
ρ +V +g  + +4 =0 (1.6).
 ∂t ∂x ∂x  ∂x D

ρ density of flowing liquid [kg/m3]


V mean velocity in a pipe cross section [m/s]
g gravitational acceleration [m/s2]
h elevation above a datum [m]
P mean pressure in a pipe cross section [Pa]
τo shear stress at the pipe wall [Pa]
D pipe diameter [m]

The shear stress at the pipe wall, τo, is defined below by Eq. 1.15.

Figure 1.2. Control volume (CV) for analysis of force balance in flow
in a circular pipe.

Additional conditions (incompressible liquid, steady and uniform flow in a horizontal


straight pipe) make it possible to obtain a simple form of the linear momentum
equation for liquid flow. Under the chosen conditions, the momentum flux at the
control volume inlet is equal to that at the control volume outlet and the inertial force
1.4 CHAPTER 1

in the control volume is zero. In this case the integrated form of the linear momentum
equation relates the driving force generated by the pressure gradient over the pipe
distance dx and the cross section area A (and the perimeter O) to the resisting force
due to viscous friction at the flow boundary, which is a pipe wall. The balance is

dP (1.7),
− A = τoO
dx

i.e. for a pipe of a circular cross section and internal diameter D

dP τ
− =4 o (1.8).
dx D

This equation shows that the wall shear stress must be correlated with the flow
conditions to solve the pressure drop due to friction in pipeline flow.

1.1.3 Friction in pipeline flow of liquid

The Eq. 1.8 is not only valid for a pipe flow boundary; it can also be generalized to
flow within each cylinder of radius r coaxial with a cylindrical pipe. It then provides
an equation for shear stress distribution in the pipe cross section (see Fig. 1.3) that is
valid for both laminar and turbulent liquid flow. This is

dP 2
− =τ (1.9).
dx r

Figure 1.3. Stress and velocity distributions in pipe flow of liquid.

Newton's law of liquid viscosity is

 dv 
τ = µf  − x  (1.10),
 dr 

τ local shear stress within liquid stream [Pa]


vx local liquid velocity in the pipe-axis direction [m]
µf dynamic viscosity of liquid [Pa.s]
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.5

where τ and vx are at the position given by the radius r in a pipe cross section.

In laminar flow, the equation for a shear stress distribution (Eq. 1.9) and Newton's law
of liquid viscosity (Eq. 1.10) determine a velocity profile vx(r) of liquid flow. Its
integration over a pipe cross section

1 8 D/2
Vf = ∫∫ v x dA = ∫ v x rdr (1.11)
AA D2 0

provides a relationship between pressure drop dP/dx and mean velocity Vf

D2  dP 
Vf =   (1.12).
32µ f  dx 

Shear stress at the pipe wall is thus determined as

8Vf
τo = µ f (1.13).
D

This procedure cannot be used for turbulent flow because the relation between shear
stress and strain rate in the turbulent flow is not fully described by the Newtonian
viscous law. In a turbulent stream, the local velocity of the liquid fluctuates in
magnitude and direction. This causes a momentum flux between liquid laminae in the
stream. The momentum exchange has the same effect as a shear stress applied to the
flowing liquid. These additional stresses set up by the turbulent mixing process are
called apparent shear stresses or Reynolds stresses. They predominate over the
Newtonian, purely viscous stresses in the turbulent core of the liquid flow. In a fully
developed turbulent flow the turbulent core usually occupies almost the entire pipe
cross section, excepting only the near-wall region. A turbulent flow regime is typical
for pipelines of an industrial scale.

Thus shear stress τ0 for turbulent flow cannot be determined directly from Newton's
law of viscosity and the force balance equation (Eq. 1.9). Instead, it is formulated by
using dimensional analysis. A function

τ0 = fn(ρf, Vf, µf, D, k) (1.14)

τo shear stress at the pipe wall [Pa]


ρf density of liquid [kg/m3]
Vf mean velocity in a pipe cross section [m/s]
µf dynamic viscosity of liquid [Pa.s]
D pipe diameter [m]
k absolute roughness of the pipeline wall [m]

is assumed. This provides a relation between dimensionless groups


1.6 CHAPTER 1

τo  k
= fn Re,  (1.15).
1  D
ρ f Vf2
2

The dimensionless group Re, Reynolds number of the pipeline flow, relates the
inertial and viscous forces in the pipeline flow

Vf Dρ f Vf D
Re = = (1.16).
µf νf

Re Reynolds number of the pipeline flow [-]


νf kinematic viscosity of liquid µf/ρf [m2/s]

The dimensionless parameter on the left side of the equation 1.15 is called the friction
factor. It is the ratio between the wall shear stress and kinetic energy of the liquid in a
control volume in a pipeline

τo
ff = (1.17).
1
ρ V2
2 f f

The parameter f is known as Fanning friction factor. Darcy obtained a friction


coefficient (called sometimes Darcy-Weisbach friction coefficient)

8τ o
λf = (1.18).
ρ f Vf2

Thus the friction coefficient λf = 4ff.


The equation for the Darcy-Weisbach friction coefficient, combined with the
integrated linear momentum equation for pipeline flow (Eq. 1.8), gives the equation
first published by Weisbach in 1850

2
dP λ f ρ f Vf
− = (1.19)
dx D 2

dP ∆P P1 − P2
that is for − written as = (see Fig. 1.4)
dx L L

λ f ρf Vf2
P1 = P2 + L (1.20).
D 2

P1 absolute pressure at beginning of pipe section [Pa]


P2 absolute pressure at end of pipe section [Pa]
λf Darcy-Weisbach friction coefficient [-]
L length of pipe section [m]
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.7

This equation is known as the Darcy-Weisbach equation for the determination of the
frictional head loss If in liquid flow in a pipeline.

P1 P2

t0

V D

t0
L

Figure 1.4. Flow in a straight horizontal pipe of constant diameter.

For laminar flow, an equation for friction coefficient λf (or ff) is calculated
theoretically from the equation for pressure drop (Eq. 1.12) giving

64
λf = (1.21).
Re

In turbulent flow there is no simple expression linking the velocity distribution with
the shear stress (and so with the pressure gradient) in the pipe cross section. Over the
years an empirical approach has provided a number of correlations λf = fn(Re, k/D)
for different pipe flow regimes. The regimes are: hydraulically smooth, transitional
and hydraulically rough (Fig. 1.5). The λf = fn(Re, k/D) correlations have been
derived from empirical expressions for a velocity profile in the turbulent flow in a
pipeline. The λf = fn(Re, k/D) values can be determined also from the Moody
diagram (Fig. 1.6) or its computational version (Churchill, 1977)

1
 8  12  12
λ f = 8  + ( X + Y) −15
.  (1.22)

 Re  
where
 16
 7  0.9 0.27 k  
X = − 2.457 ln   +  (1.23)
  Re  D  

and
 37530  16
Y=  (1.24)
 Re 

λf Darcy-Weisbach friction coefficient [-]


Re Reynold number for liquid flow [-]
1.8 CHAPTER 1

k absolute roughness of pipe wall [m]


D pipe diameter [m].

The value λf = 0.010 - 0.012 is usually appropriate for an initial estimation of


water-flow friction losses in industrial pipelines (Fig. 1.6).

Figure 1.5. Regimes of flow over a pipeline wall. Regimes for λf determination.
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.9

Figure 1.6a. Moody diagram for a determination of Darcy-Weisbach


friction coefficient ff (λf).
1.10 CHAPTER 1

Figure 1.6b. Moody diagram for a determination of Darcy-Weisbach


friction coefficient λf, zoom to most used region.
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.11

CASE STUDY 1

Frictional head loss in flow of water through a straight horizontal pipeline.

Determine the pressure drop due to friction for flow of water at room temperature (for
calculations take water density ρf = 1000 kg/m3 and kinematic viscosity νf = 10-6
m2/s) in a 1000 m long pipeline of the diameter D = 900 mm. The absolute roughness
of a pipeline wall k = 10-5 m. Mean velocity of water in a pipeline cross section V =
4.5 m/s.

Solution:

a. Friction coefficient, λf

The friction coefficient is dependent on the roughness k and Reynolds number Re of


the water flow. For the above given values of parameters Re = (4.5 x 0.9)/10-6 = 4.05
x 106. The ratio k/D = 1.1 x 10-5.

The numerical approximation of the Moody diagram by Churchill (Eqs. 1.22 – 1.24)
gives for these values of Re and k/D the friction coefficient value λf = 0.0099 = 0.010.
The same value of λf should be obtained directly from the Moody diagram (for D/k =
91000 in Fig. 1.6).

b. Frictional head loss, If

The frictional head loss for water flow in a pipeline is determined using the
Darcy-Weisbach equation (Eq. 1.20). This gives a parabolic relationship between the
hydraulic gradient If and mean mixture velocity Vm. For our inputs

0.010 Vm2 0.010 4.502


If = = = 0.01147 .
0.900 19.62 0.900 19.62

Thus the friction loss If = 0.01147 meter water column over 1 meter pipeline length,
i.e. 11.47 meter water column over 1 kilometer pipeline length. This represents the
pressure drop due to friction ∆P = 0.01147 x 1000 x 9.81 = 112.5 Pa/m' or 112.5 kPa
over the one kilometer long pipeline.

Summary of the results:


friction coefficient : λf = 0.010 [-]
frictional head loss per unit meter of a pipeline length : If = 0.01147 [-]
pressure drop due to friction over 1 000-meter long straight pipeline :
∆P = 112.5 [kPa / 1000 m]
1.12 CHAPTER 1

1.2 SOLID PARTICLES IN A CARRYING LIQUID

Forces acting on solid particles submerged in a liquid have their origin either in a
particle-liquid interaction or in a particle-particle interaction. Particles moving in a
conduit may also interact with a conduit boundary. The forces acting on a single
particle in a dilute suspension are the body forces. The particle-liquid body forces are
the buoyancy force, drag force and lift force. When a solid particle is transported in
the turbulent flow of a carrying liquid the turbulent diffusive force from carrier eddies
is an additional particle-liquid force. Forces acting on solid particles due to
particle-particle interaction are transmitted as the interparticle stress via the particle
contacts. Coulombic stresses occur in a granular body occupied by particles in
continuous contact. When a granular body is sheared and interparticle contacts are
only sporadic, Bagnold stresses are transmitted through the granular body.

1.2.1 Gravitational and buoyancy force

The body force due to gravitational acceleration is determined from the solid particle
volume and density. The gravitational force on a spherical solid particle of diameter d
is
πd 3 (1.25).
FGp = ρs g
6

According Archimedes law, a solid particle immersed in a liquid obeys a buoyancy


effect, which reduces its weight in the carrying medium. The submerged weight of the
solid particle is a result of gravitational and buoyancy effects on the solid particle
immersed in the liquid. For a spherical particle the submerged weight is determined
by the expression
πd 3
( )
Fwp = ρs − ρ f g (1.26).
6

FGp gravitational force on a spherical particle [N]


FWp submerged weight of a spherical particle [N]
ρs density of solid particle [kg/m3]
ρf density of liquid [kg/m3]
g gravitational acceleration [m/s2]
d diameter of a particle [m]

1.2.2 Drag force

When the surrounding liquid moves relative to a solid particle, an additional force is
exerted from the liquid onto the submerged particle. The drag force, FD, acts in the
direction of the relative velocity vr = vf - vs between the liquid and the solid particle.
The magnitude of the drag force is expressed in terms of the drag coefficient CD. This
comes from dimensional analysis of the function

FD = fn(ρf, µf, d, vr) (1.27).


BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.13

It provides two dimensionless groups of parameters:


drag coefficient
8FD
CD = (1.28)
πd 2 v r v r ρ f

and particle Reynolds number


ρf v r d
Re p = (1.29)
µf

giving CD = fn(Rep).

A balance of the gravitational, buoyancy and drag forces on the submerged solid
body determines a settling velocity of the body.

Figure 1.7. Force balance on a solid body submerged in a quiescent liquid.

An experimental determination of the drag coefficient is based on measurement of the


terminal settling velocity of a spherical particle, vts, in a quiescent liquid. Measured
vts is the relative velocity vr.

Figure 1.8. Drag coefficient as a function of particle Reynolds number.

Methods for a determination of particle settling velocity are discussed in Intermezzo I.


1.14 CHAPTER 1

1.2.3 Lift force

The lift force, FL, on a single solid particle is a product of simultaneous slip (given by
relative velocity vr = vf - vs) and particle rotation. The force (sometimes called the
Magnus force) acts in a direction normal to both the relative velocity vr and the
particle rotation vector. A particle rotation combined with a slip results in a lower
hydrodynamic pressure in flow above the particle than in that below the particle. Lift
force is due to this pressure gradient.

(a) (b)
Figure 1.9. Lift force on a rotating solid body. (a) lift force on a rotating cylinder,
(b) the Saffman force, i.e. lift force due to shear and slip.

The lift force is most active near a pipeline wall where the velocity gradient is high.
However, the lift forces due to particle spin play a minor role in the majority of
mixture flow regimes compared to the Bagnold and Coulombic forces.

1.2.4 Turbulent diffusive force

Solid particles are also subject to additional liquid-solids interactions when they are
transported in a turbulent stream of the carrying liquid. An intensive exchange of
momentum and random velocity fluctuations in all directions are characteristic of the
turbulent flow of the carrying liquid in a pipeline. Scales of turbulence are attributed
to properties of the turbulent eddies developed within the turbulent stream. According
to Prandtl's picture of turbulence, the length of the turbulent eddy is given as the
distance over which the lump of liquid transports its momentum without losing its
identity, i.e. before the lump is mixed with liquid in a new location. This distance is
called the mixing length and since it is supposed to represent a mean free path of a
pulse of liquid within a structure of turbulent flow it is considered a length scale of
turbulence. A turbulent eddy is responsible for the transfer of momentum and mass in
a liquid flow. The instantaneous velocity of liquid at any point in the flowing liquid
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.15

and in arbitrary direction (x, y or z) is given by v = v + v' where v is the


time-averaged velocity and v' is the instantaneous fluctuation velocity. The turbulent
fluctuating component v' of the liquid velocity v is associated with a turbulent eddy.

It is well known that turbulent eddies are responsible for solid particle suspension.
The intensity of liquid turbulence is a measure of the ability of a carrying liquid to
suspend the particles. The size of the turbulent eddy and the size of the solid particle
are also important to the effectiveness of a suspension mechanism. The characteristic
size of turbulent eddies is assumed to depend on the pipeline diameter.

A low concentration suspension is described by using a classical turbulent diffusion


model of Schmidt and Rouse. The model was constructed as a flux balance per unit area
perpendicular to the vertical direction in a flow balancing the volumetric settling rate
(characterized by settling velocity vt) in a quiescent liquid and the diffusion flux
(characterized by the liquid velocity fluctuation in a vertical direction v'y, associated
with the length of a turbulent eddy, ML) (see Fig. 1.10). A characteristic value of the

turbulent pulsative velocity ~ v' y = v' 2y , i.e. the root mean square of velocity
fluctuations in the y-direction.

v’y + vt
cv-(ML/2).dcv/dy
ML

cv+(ML/2).dcv/dy
v’y - vt

Figure 1.10. Mixing length model of particle exchange by turbulence.

The balance of

the upward flux per unit area =


1
2  (
 ML  dc v  ~
cv +  v' − v
 2  dy  y t ) and

1
the downward flux per unit area = c v − 
2
 (
 ML  dc v  ~
v' + v
 2  dy  y t )
gives an equation

dc v
− εs = vt .cv (1.30)
dy

ML ~
when solids dispersion coefficient ε s = v' y .
2

Integration of Eq. 1.30 with εs considered constant gives an exponential concentration


profile cv(y) as
1.16 CHAPTER 1

 vt 
c v ( y) = C vb .exp −
εs
( y − y b ) (1.31)
 

i.e. an exponential concentration variation with height, y, in a flow above a boundary


characterized by a position yb and a concentration cvb.

cv local concentration at the height y [-]


cvb known local concentration at the position yb [-]
vt terminal settling velocity of a particle [m/s]
εs solids dispersion coefficient [m2/s]
y vertical distance from pipe wall defining
a position in a pipe cross section [m]
yb vertical distance from pipe wall to boundary [m]
ML mixing length [m]
~
v' y turbulent pulsative velocity in the y-direction [m/s]

A turbulent diffusive force exerted on particles by turbulent eddies is obtained by


rewriting the Eq. 1.30 as a force balance between the turbulent diffusive force and the
submerged weight of the particles in a unit volume of slurry in a horizontal pipe. The
submerged weight is ρfg(Ss-1)cv so the turbulent dispersive force

ε s dc v
Ft = −ρ f g(S s − 1) (1.32).
v t dy

How to determine the solids dispersion coefficient, εs, is a major problem connected
with the application of the turbulent diffusive model. The effect of distance from a
boundary and of the presence of solid particles in a turbulent stream on a local value
of εs cannot be neglected. Further, the neighboring particles also affect the particle
settling velocity handled in the model.

1.2.5 Coulombic contact force

Sand/gravel particles are transported in dredging pipelines often in a form of a


granular bed sliding along a pipeline wall at the bottom of a pipeline. A mutual
contact between particles within a bed gives arise to intergranular forces transmitted
throughout a bed and via a bed contact with a pipeline wall also to the wall

Stress distribution in a granular body occupied by non-cohesive particles in


continuous contact is a product of the weight of grains occupying the body. The
intergranular pressure (or stress) from the weight of grains is transmitted within the
granular body via interparticle contacts. The stress has two components: an
intergranular normal stress and an intergranular shear stress. According to Coulomb's
law these two stresses are related by the coefficient of friction. Du Boys (1879)
applied Coulomb's law to sheared riverbeds. He related the intergranular normal
stress, σs, and intergranular shear stress, τs, at the bottom of a flowing bed by a
coefficient
BASIC PRINCIPLES OF FLOW OF LIQUID AND PARTICLES IN A PIPELINE 1.17

τs τs
tan φ = =
σ s ρ f g(S s − 1)C vb H s (1.33)

φ angle of repose of the grains [-]


σs intergranular normal stress [Pa]
τs intergranular shear stress [Pa]
Ss specific gravity of solids, Ss= ρs/ρf [-]
Ysh thickness of the sheared bed [m]
Cvb maximum solids volume fraction of solids
in the granular bed, it is considered to be
valid for the sheared bed [-]
ρf density of liquid [kg/m3]
g gravitational acceleration [m/s2]

The angle of repose, φ, is considered to be the angle at the internal failure of a static
granular body (Fig. 1.11). The value of this internal-friction coefficient is basically
dependent on the nature of the surface over which the grains start to move, i.e.
primarily on a grain size. When the granular bed motion takes place over a pipe wall,
the value of the bed-wall friction coefficient can be determined by a tilting tube test.

Figure 1.11. The angle of repose of a granular material.

1.2.6 Bagnold dispersive force

Sheared-bed particles flowing in the region of high shear rate maintain sporadic,
rather than continuous contact with each other, provided that solids concentration in
the sheared bed is considerably lower than the loose-poured bed concentration Cvb.
1.18 CHAPTER 1

The nature of an interparticle contact influences the relationship between the


intergranular stress components. It is appropriate to relate the particulate shear and
normal stresses in a granular body experiencing the rapid shearing by using a
coefficient of dynamic friction tanφ' instead of its static equivalent tanφ. Bagnold
(1954. 1956) measured and described the normal and tangential stresses in mixture
flows at high shear rates.

Bagnold's dispersive force is a product of intergranular collisions (particle - particle


interactions) in a sheared layer rich in particles. The direction of the force is normal to
the layer boundary on which it is acting. The force increases with increasing solids
concentration and shear rate in the sheared layer.

1.3 REFERENCES

Bagnold, R.A. (1954). Experiments on a gravity-free dispersion of large solid spheres


in a Newtonian liquid under shear, Proceedings Roy. Soc. (London), Ser. A, 225, 49-
63.
Bagnold, R.A. (1956). The flow of cohesionless grains in liquids, Proceedings Roy.
Soc. (London), Ser. A, 249, 235-97.
Churchill S.W. (1977). Friction-factor equation spans all fluid-flow regimes.
Chemical Engineering, 84(24), 91-2.
Du Boys, P. (1879). Étude du règime du Rhône et de l'action exercée par les eaux sur
un lit à fond de graviers indéfiniment affouillable. Annales des Ponts et Chausées,
18(49 pt 2), pp. 141-95.
Longwell, P.A. (1966). Mechanics of Fluid Flow. McGraw-Hill.

1.4 RECOMMENDED LITERATURE

Shook, C.A. & Roco, M.C. (1991). Slurry Flow. Principles and Practice.
Butterworth-Heinemann.
Streeter, V.L. & Wylie, E.B. (1983). Fluid Mechanics. McGraw-Hill.
2.
SOIL-WATER MIXTURE AND ITS PHASES

2.1 SOIL PROPERTIES

Different sorts of soil are subject to dredging. At different locations dredged solids
may differ strongly in density, size, shape and consistency. It is a very different
3
experience to dredge the iron ore of density round 4700 kg/m or the sand of density
3
2650 kg/m . If clay (a typical particle size is in an order of microns) is dredged it is
necessary to know whether it is cohesive or non-cohesive. The non-cohesive clay will
be transported as a pseudo-homogeneous mixture in a pipeline while the cohesive
clay, cuttered in slices from the bottom of a waterway, will form clay balls in a
pipeline. The balls might be of hundreds of millimetres in a diameter. Such balls will
definitely not be suspended in a carrying liquid. They will roll and slide at the bottom
of a dredging pipeline, giving a totally different flow pattern in a pipeline than
cohensionless clay. As a result the flow resistance will be very different. However, a
majority of dredged materials are just different sorts of sand or gravel that are
non-cohesive and of almost identical density.

The properties of a transported soil are of great importance for a determination of


flow conditions in dredging pipelines. A determination of properties of a transported
soil requires:

a. taking samples of soil


- in situ
- in a pipeline
b. analysis of samples using suitable methods.

Particle size distribution, soil density and in situ concentration of soil are the most
important soil properties influencing behaviour of solid particles in a mixture
transported in a dredging installation. If clay is transported the cohesion (described by
liquid limit and plastic limit) is a further important parameter. In the following text,
only the particle size distribution and solids density will be discussed. A
determination of other properties is described e.g. in van den Berg (1998).

2.1.1. Particle size distribution (PSD)

Several systems are defined for identification and a classification of solids according
to their particle size (e.g. American norm ASTM Standard D288). For the dredging
purposes the following system is used (Table 2.1)
2.2 CHAPTER 2

Table 2.1. Identification and classification of soil for dredging purposes

Main type of soil Particle size


Identification size in [mm]
Boulders Granular - > 200
Non-cohesive
Cobbles - 200 – 60
Coarse 60 – 20
Gravel Medium 20 – 6
Fine 6–2
Coarse 2 – 0.6
Sand Medium 0.6 – 0.2
Fine 0.2 – 0.06
Coarse 0.06 – 0.02
Silt Cohesive Medium 0.02 – 0.006
Fine 0.006 – 0.002
Clay - < 0.002

Dredged soil is seldom uniformly graded. Two techniques are used to determine
particle size distribution (PSD) in samples of dredged sand or gravel – the sieving and
the sedimentation tests, finer solids as silt and clay are tested using the hydrometric
method.

2.1.1.1 Methods for PSD determination

A choice of the PSD method is dependent on the range of particle sizes in the tested
sample of soil. Samples containing particles of sand- or gravel size are usually tested
using the screen method (the sieving) or the sedimentation method (the sedimentation
in the water column). For finer particles (d < 0.074 mm) the distribution is determined
by a sedimentation process in a hydrometer.

A. Sieving
A weighted sample of dry solids is sieved through a series of sieves with standard
sieve meshes. The sample fraction remaining in each sieve is weighted. The fractions
by mass are recalculated to obtain a percentage of the mass of the entire sample, pi.
The fraction pi is considered to contain particles of characteristic diameter di
represented by the size of opening of the sieve at which the fraction pi remained.
SOIL-WATER MIXTURE AND ITS PHASE 2.3

Table 2.2. PSD for the narrow-graded medium sand (determined by a sieving analysis
of a one-kilogram dry sample).

1 2 3 4 5
sieve weight percentage of cumulative characteristic
opening fraction total weight, % particle size,
[mm] [g] pi [%] mass di [mm]
0.85 0 0 100 0.85
0.6 6.7 0.67 99.33 0.6
0.5 37.7 3.77 95.56 0.5
0.42 447.8 44.78 50.78 0.42
0.355 366.4 36.64 14.14 0.355
0.3 74.2 7.42 6.72 0.3
0.21 55.1 5.51 1.21 0.21
0.15 10.6 1.06 0.15 0.15
0.00 1.5 0.15 0 0.00
Σ 1000 Σ 100

Legend:
column 1: sieve opening of the sieves used to test the sample
column 2: weight of soil fraction remaining at the sieve of the opening in col. 1
column 3: weight in col. 2 as a percentage of the weight of the total sample (1 kg)
column 4: cumulative percentage, Σpi
column 5: particle diameter representing the soil fraction remaining at the sieve

Table 2.3. PSD for the broad-graded medium sand (determined by a sieving analysis
of a one-kilogram dry sample).

1 2 3 4 5
sieve weight percentage of cumulative Characteristic
opening fraction total weight, % Particle size,
[mm] [g] pi [%] mass di [mm]
0.85 5 0.5 99.50 0.85
0.6 162.4 16.24 83.26 0.6
0.5 162.4 16.24 67.02 0.5
0.42 162.4 16.24 50.78 0.42
0.355 125.7 12.57 38.21 0.355
0.3 125.7 12.57 25.64 0.3
0.21 125.7 12.57 13.07 0.21
0.15 125.7 12.57 0.5 0.15
0.00 5 0.5 0 0.00
Σ 1000 Σ 100

The columns 4 and 5 of the Tables 2.2 and 2.3 are plotted to the cumulative Particle
Size Distribution curve.
2.4 CHAPTER 2

Samples of relatively narrow graded fine sand (Sand a) and medium sand (Sand c) are
processed together with a broad graded sand (Sand b) in Table 2.4 and plotted to the
cumulative PSD curve on Fig. 2.1.

Table 2.4. Processed PSD for three sorts of sand.

(%) za nd a za nd b za nd c
(µm ) (µm ) (µm )
10 250 1300 850
20 180 500 620
30 150 270 500
40 140 170 400
50 120 120 350
60 90 75 290
70 75 55 240
80 60 40 180
90 40 25 130

Tota a l 1105 2555 3560


dmf 123 284 396

KLEI SILT ZAND GRIND


FIJN GROF FIJN GROF FIJN
g ew ic htsp erc enta g e o p d e zeef d roo g

10

20
a b c
30

40

50

60

70

80

90

100

0,001 0,002 0,005 0,01 0,02 0,05 0,1 0,2 0,5 1 2 5


korreld ia m eter in m m

Figure 2.1. Cumulative curves of particle size distribution for three sorts of sand
according to Table 2.4.

b. Sedimentation column
The method of sedimentation a long sedimentation column is used as alternative to
the sieving method for solid particles of sand (and gravel) size. This method has the
advantage of direct provision of the settling velocity of the particles, which is the
parameter characterising the solids impact on the slurry flow behaviour rather than the
SOIL-WATER MIXTURE AND ITS PHASE 2.5

particle size. The particle size distribution can be obtained from the settling velocity
distribution using appropriate settling velocity equation.
A weighted sample of solids is collected in a cup at the top of a sedimentation column
filled with water. The water level in the sedimentation column touches the bottom of
the cup. The cup is opened and the time is measured which solid particles of the
sample need to reach a plate at the bottom of the sedimentation column. The distance
between the cup and plate is known. The measuring principle is the sensing of the
progression of the deflection in time of a thin metal membrane connected with both
the cup and the plate. The membrane senses an impulse determining the time at which
the cup is opened and the time-dependent increase in the weight of the plate as the
particles settle on its surface. This signal is converted and plotted as a summation
curve of % solids mass passed versus particle settling velocity.

The effect of the concentration of solids in a settling cloud on the particle settling
velocity in a sedimentation column is negligible because of small solids samples and
thus the low concentration of solids in the settling cloud. The disadvantage of this
method is that the length of the sedimentation column is too short to test very coarse
particles, which need only a few seconds to reach the plate at the bottom of the
sedimentation column.

c. Hydrometer tests
This method determines a particle size distribution for soils too fine to be sieved, i.e.
finer than 0.074 mm. A hydrometer is a small glass bottle with a calibrated volume. In
a hydrometer a small sample of soil is suspended in distilled water. At certain position
density of suspension is measured. Due to slow sedimentation of particles in
quiescent water in a hydrometer the suspension density at the position gradually
decreases. The change of density with time is sensed. This can be interpreted as a
development of settling velocity of particles in a soil sample and thus according to
Stokes law of settling as a particle size distribution. As the sieving analysis the
hydrometer analysis provides the cumulative PSD curve.

2.1.1.2. Characteristic diameters of particle

The mass-median diameter d50 is the diameter for which 50% (by mass) of the
particles in a soil sample are finer. In other words one half of all particles in a sample
is larger and one half of all particles is smaller than is the size d50. The d85 size gives
the diameter for which 85% (by mass) of the particles in a soil sample are finer.

A comparison of PSD for these two sand samples demonstrates clearly that the
mass-median diameter d50 of particles does not represent fully the particle size in a
soil sample. The d50 value is identical for both samples (d50 = 0.42 mm) but the
deviation from the median size is much larger in a broadly graded sample. A
deviation from the median size has to be incorporated in the parameter(s) representing
the solids size in mixture flow models. Usually such a parameter is sufficiently
represented by mean diameter or decisive particle diameter defined below, but for
specific cases as is that simulated in Table 2.3 is this not enough. Therefore the flow
models employing at least two characteristic particle diameters (e.g. d50 and d85) are
more suitable.
2.6 CHAPTER 2

The mean diameter of particles in a type of solid with a certain degree of grading is
determined as

∑ d i pi
ds = i (2.1)
∑ pi
i

where ∑ p i = 1.00.
i

In a dredging practice the Fuhrboter classification of the characteristic particle size is


often used. This so-called decisive particle diameter (in Dutch "maatgevende
korreldiameter") is defined as

d + d 20 + .............. + d80 + d 90
d mf = 10 (2.2).
9

Table 2.5. Comparison of characteristic particle sizes for medium sand samples in
Table 2.2 and Table 2.3.

Narrow-graded Broadly
sand graded sand
(Table 2.1) (Table 2.2)
d50 0.42 0.42
d85 0.47 0.62
d15 0.36 0.22
ds 0.38 0.38
dmf 0.42 0.42

2.1.1.3. Form factor of a particle

The form (shape) of the particle influences the settling process of a particle in a
carrying liquid, but also the wear of components in contact with flowing mixture and
packing of solids in granular beds. The shape factor is determined experimentally by
measuring the settling velocity of non-spherical particles. The shape-factor for
v
non-spherical particles is ξ = t (as already discussed in Intermezzo I) and its value
v ts
drops from one for spherical particles to the lowest values for the particles that have
one dimension much smaller than the other two.
SOIL-WATER MIXTURE AND ITS PHASE 2.7

2.1.2. Density of solid grains and granular body

Density of solids is a property independent of temperature and pressure of the


surrounding matter or other parameters that might vary during the solids
transportation in a mixture.

Density is determined by a pycnometric method. A pycnometer is a glass bottle with a


calibrated volume gauge. This is used to determine a volume of dry solids sample of
known weight. The dry sample is dropped to the pycnometer containing a known
volume of water. The change in a volume is read out as the change in the position of
the water level at the pycnometer gauge. This gives a total volume of particles of a
sample. The ratio of the weight of a sample and its volume gives density of solid
grains, ρs, in a sample.

For dredging purposes, however, an another parameter associated with the mass and
the volume of transported soil is important. The amount of cubic meters of
transported material decisive for dredging work payments is not based on the density
of solid particles, ρs, but rather on the in situ density of the soil body, ρsi. If a known
volume of a soil (i.e. a granular body) submerged in water is weighted, the ratio of the
weight and body volume gives the in situ density of a soil, ρsi. This is lower than the
density of solids occupying a granular body since it includes the volume of water
(lighter than solids) present in the pores between particles in the granular body.

A measure of the difference between the density of solid particles and density of soil
occupied by the solid particles is given by the porosity of a soil, n, that is obtained as

ρ − ρsi
n= s (2.3).
ρs − ρ f

n porosity of soil [-]


ρs density of solid grains [kg/m3]
ρsi density of soil in situ [kg/m3]
ρf density of liquid [kg/m3]

The typical value of the in situ density is 2000 kg/m3 for a submerged sand bed. This
indicates that 60 % of the sand-bed volume is occupied by sand particles having
density 2650 kg/m3 and the interstitial water of density 1000 kg/m occupies 40 % of
the total volume3. The porosity n = 0.40.
2.8 CHAPTER 2

Table 2.6. Typical values of density and porosity of various soils

Density of Density of soil Porosity


solids in situ (wet)
ρs [kg/m3] ρsi [kg/m3] n [%]
silt 2650 1100 - 1400 80 - 90
loose clay 2650 1400 - 1600 60 - 80
packed clay 2650 1800 - 2000 35 - 50
sand with clay 2650 1800 - 2000 40 - 50
sand 2650 1900 - 2000 35 - 45
coarse sand 2650 2050 - 2200 28-36
with gravel
clay boulders 2650 2320 20

The ratio ρs/ρw gives the relative density (also "specific gravity" of solid particle),
Ss. The parameter ρw is the density of water with the typical value 1000 kg/m3. More
precise values are discussed in the following paragraph.

2.2 LIQUID PROPERTIES

Density and viscosity are the most important properties of a carrying liquid from the
point of view of the mixture flow in a dredging pipeline. Both parameters are
sensitive to temperature. Sensitivity to pressure can be neglected over the range of
conditions encountered in the hydraulic transport, the liquids are considered
incompressible.

Viscosity that relates the shear stress and the shear rate according to Newton's law of
viscosity (see Chapter 1) is called dynamic viscosity, µ, and it has a unit Pa.s. The
ratio of the dynamic viscosity of liquid and density of liquid is known as kinematic
viscosity, ν = µ/ρ, having a unit m2/s.

Table 2.7. Properties of water

Temperature Density, Dynamic Kinematic Vapour


viscosity, viscosity, pressure,
T [oC] 3
ρw [kg/m ] µw [Pa.s] νw [m2/s] pv [Pa]
0 999.8 1.781 x 10-3 1.785 x 10-6 0.61 x 103
5 1000.0 1.518 x 10-3 1.519 x 10-6 0.87 x 103
10 999.7 1.307 x 10-3 1.306 x 10-6 1.23 x 103
15 999.1 1.139 x 10-3 1.139 x 10-6 1.70 x 103
20 998.2 1.002 x 10-3 1.003 x 10-6 2.34 x 103
25 997.0 0.890 x 10-3 0.893 x 10-6 3.17 x 103
30 995.7 0.798 x 10-3 0.800 x 10-6 4.24 x 103
SOIL-WATER MIXTURE AND ITS PHASE 2.9

Density of water, ρw [kg/m3], can be approximated within a temperature range 5 < T


< 100 0C using

ρw = 999.7 – 0.10512(T – 10) – 0.005121(T – 10)2 + 0.00001329(T – 10)3 (2.4).

Dynamic viscosity of water, µw [Pa.s], can be approximated using

0.10
µw = (2.5),
2.1482([T − 8.435] + 8078.4 + [T − 8.435]2 ) − 120

in which T is temperature in 0C.

Density of seawater is slightly higher than that of river water due to the contents of
dissolved salt. The typical value is ρf = 1025 kg/m3, i.e. relative density Sf = ρf/ρw =
1.03.

2.3 MIXTURE PROPERTIES

2.3.1 Mixture density and concentration of solids

Mixture is composed of two phases: solids and liquid. The density of the mixture, ρm,
is influenced by the fraction of solids present in a carrying liquid. The fraction is
determined by the parameter called concentration. The volumetric concentration, Cv,
determines the fraction of the mixture volume that is occupied by solids. The Cv =
0.60 means that 60 % of the total volume of mixture is occupied by solid particles.
The fraction of solids in mixture can be also expressed in weight giving the weight
concentration of solids, Cw. The Cw = 0.60 means that 60 % of the total weight of
mixture is exerted by solid particles present in the mixture.
A relationship between the mixture density and the volumetric concentration of solids
in a mixture proportion is derived from the general formula

MASSmixture = MASSliquid + MASSsolids (2.6)

considering MASS = density (ρ) x volume (U) and Um = Uf + Us

ρmUm = ρfUf + ρsUs = ρf(Um-Us) + ρsUs (2.7)

dividing by Um and considering Cv = Us/Um gives finally

ρm = ρf(1-Cv) + ρsCv (2.8).

Rearranging gives the equation for the volumetric concentration

ρ − ρ f S m − Sf
Cv = m = (2.9).
ρs − ρ f Ss − Sf
The corresponding weight concentration
2.10 CHAPTER 2

ms ρ U ρ
Cw = = s s = s Cv (2.10).
mm ρm U m ρm

Cv volumetric concentration of solid grains [-]


Cw weight concentration of solid grains [-]
ρs density of solid grains [kg/m3]
ρm density of mixture [kg/m3]
ρf density of liquid [kg/m3]
Ss relative density of solid grains [-]
Sm relative density of mixture [-]
Sf relative density of liquid [-]
ms mass of solid grains [kg]
mm mass of mixture [kg]
Us volume of solid grains [m3]
Um volume of mixture [m3]

Volumetric concentration is a parameter more often used than weight concentration in


a dredging practice. This is because the payment of dredging works is based on the
amount of cubic meters of material either dredged from the bottom of a waterway or
dumped at the deposit site. The prize is determined for 1 m3 of the in situ soil
dredged.

The volumetric concentration of in situ soil is determined as

ρ − ρf
C vsi = m (2.11).
ρsi − ρ f

The volume of in situ soil body, Usi, dredged is composed of the volume of grains,
Us, and the volume of water, Uf, captured in voids (pores) between grains in a
submerged granular body, i.e. Usi = Us + Uf. The volume of water in the in situ
granular body is expressed by the parameter called porosity n = Uf/Usi. Then Usi(1-n)
= Us and the volumetric concentration of in situ solids is

U si Us C
C vsi = = = v (2.12).
U m (1 − n )U m 1 − n

The typical value of porosity for loose-poured bed of sand grains is n = 0.4. For this
bed the concentration Cv = 0.60. Remember that Cv = 0.60 means that 60 % of the
total volume of mixture is occupied by solid particles and 40 % by water. If the
porosity n = 0.4 (i.e. 40 %) the Cvsi = 1.00. This says that for a dredging operation a
creation of the loose-poured bed in a deposit site is a reference level considered as
unity.

The volumetric concentration of solids, Cv, is an important parameter in mixture flow


modeling, the volumetric concentration of in situ soil, Cvsi, is an important parameter
in calculating production and thus economical costs of a dredging operation.
SOIL-WATER MIXTURE AND ITS PHASE 2.11

For the FLOWING MIXTURES the physical meaning of the solids concentration
must be further specified. A different fraction of solids is determined in a mixture
flowing in a pipeline when it is measured at two different places:
- in a horizontal length section of a pipeline (e.g. by weighting of the pipe section)
and
- at the outlet of a pipeline (e.g. by collecting the discharged mixture to the tank).
The volume fraction of solids resident in an isolated length section of a horizontal
pipe somewhere within the pipeline determines the spatial (resident) volumetric
concentration of solids, Cvi, in a pipe. The volume fraction of solids in the mixture
discharged to the collecting tank, i.e. delivered from a pipeline, determines the
delivered volumetric concentration of solids, Cvd.

The spatial volumetric concentration gives the fraction of solids actually resident in a
slurry pipeline and it is calculated as the ratio between solids and slurry volumes in a
pipeline section

Us
C vi = (2.13).
Um

The delivered volumetric concentration gives a fraction of solids delivered from a


slurry pipeline (Fig. 2.2) and it is calculated as the ratio between solids and slurry
flow rates

Qs ∆U s ∆t
C vd = = (2.14)
Qm ∆t ∆U m

Cvi spatial volumetric concentration of solids [-]


Cvd delivered volumetric concentration of solids [-]
Us volume fraction of solids in mixture [m3]
Um total volume of mixture, Um=Us+Uf [m3]
Qs volumetric flow rate of solids [m3/s]
Qm volumetric flow rate of mixture, Qm=Qs+Qf [m3/s]
∆Us volume of solids delivered by a pipe during
time period ∆t [m3]
∆Um volume of mixture delivered by a pipe during
time period ∆t [m3].

Qm

Qs

Um

Us

Figure 2.2. Definition of volumetric delivered concentration.


2.12 CHAPTER 2

Table 2.8. Maximum average attainable concentration Cvdsi by various dredges


(typical values of Cvdsi according van den Berg, 1998)

Cvdsi Density of Cvd


in situ mixture solid grains
[-] ρm [kg/m3] [-]
CSD without SP 0.25 1250 0.15
CSD with SP 0.30 1300 0.18
Bucket wheel dredge 0.50 1500 0.30
with SP
Plain suction dredge 0.40-0.60 1400-1600 0.24 - 0.36
Modern THSD during 1300-1600 0.18 - 0.36
hopper loading 0.30-0.40
(suction process)
Modern THSD during
hopper unloading 0.70-0.80 1700-1800 0.42 - 0.48
(pumping to shore)

Legend:
CSD cutter suction dredge
SP submerged pump
THSD trailing hopper suction dredge

The difference between the spatial concentration and the delivered concentration can
be illustrated on a hypothetical case in a dredging pipeline delivering the sand-water
mixture. Imagine that the carrying liquid velocity in a dredging pipeline drops
suddenly to the value that is not sufficient to carry the sand particles. All sand
particles will settle down to the bottom of a pipeline and form a stationary bed. The
carrying water will flow above this stationary bed. Measurements of concentration in
a horizontal pipe section will indicate a fraction of solids in a pipe section but
measurements at the pipeline outlet will not register any solids in discharging carrier
(Fig. 2.3). The spatial concentration will reach a reasonably high value but the
delivered concentration will be just zero.

Qm

Q s= 0

Um

U s= 0

Figure 2.3. Difference between spatial and delivered concentrations.


SOIL-WATER MIXTURE AND ITS PHASE 2.13

This extreme case is unlikely to occur during a dredging operation but an operation
during which a certain portion of transported sand forms a slowly sliding bed or even
a stationary bed is not quite unusual. During such operation the different values of
Cvi and Cvd also occur in a pipeline.

The difference between the spatial and the delivered concentration indicates slip
(hold-up) within the mixture flow caused by the different velocities of the carrying
liquid and that of the solid phase within a mixture stream. Govier & Aziz (1972)
described this phenomenon as follows:
"When the phases of two-phase flow differ in density and/or viscosity, one of
them - usually the less dense phase - tends to flow at a higher in situ average velocity
than does the other. This gives rise to an all-important characteristic of two-phase
flow, the existence of "slip" of one phase past the other, or "holdup" of one phase
relative to the other."
Govier and Aziz also summarised the factors influencing slip:
- the existence of a velocity profile across the pipeline cross section
- the existence of a concentration profile across the pipeline cross section
- the local relative velocity between phases (vs-vf) caused by gravitational effects.

The slip between two phases in a cross section of slurry pipeline can be quantified by
the mean slip velocity in a pipeline cross section, Vs-Vf, or by the ratio of the mean
velocities of solids and mixture in a pipeline cross section, Vs/Vm. This latter
parameter, called the slip ratio (or “transport factor” in a dredging practice), is
exceptionally suitable for the evaluation of slip in a pipeline. This ratio is also equal
to the ratio of mean concentrations in pipeline cross sections Cvd/Cvi since

Qs VA VC A
C vd = = s s = s vi (2.15),
Q m Vm A Vm A

in which As is the part of the cross sectional area of the pipeline occupied by solids,
thus

Vs C
= vd (2.16).
Vm C vi

Cvd delivered volumetric concentration of solids [-]


Cvi spatial volumetric concentration of solids [-]
Vs mean velocity of solids in pipeline cross section [m/s]
Vm mean velocity of mixture in pipeline cross section [m/s]

The slip phenomenon may influence the accuracy of the determination of the solids
flow rate from the measurements on a dredging pipeline. Solids flow rate through a
slurry pipeline connected with a dredge is often determined from measurements of the
mean solids concentration and the mean slurry velocity in a horizontal pipeline
section. The radiometric measurement of the concentration in a horizontal pipeline
gives the value of the spatial concentration (not the delivered concentration that might
be lower) so that Qs calculated as CviVmA overestimates the real solids flow rate
2.14 CHAPTER 2

unless the slip in a pipeline is negligible. Generally, it is desirable to take the slip into
account during slurry flow calculations.

Table 2.9. Transport factor Cvd/Cvi for various solids and solids concentrations in a
mixture flow of velocity range Vdl < Vm < 2Vdl (typical Cvd/Cvi values
according to Matousek, 1997).

At low At high
Cvd Cvd
Silt and finer solids 1.00 1.00
Fine to medium sand 0.80-1.00 0.90-1.00
Medium to coarse sand 0.70-0.90 0.85-1.00
Coarse sand 0.65-0.85 0.75-0.95
Fine gravel 0.65-0.85 0.75-0.90
Boulders 0.40-0.65 0.40-0.65

2.4 REFERENCES

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
Govier, G.W. & Aziz, K. (1972). The Flow of Complex Mixtures in Pipes. Van
Nostrand Reinhold Company.
Matousek, V. (1997). Flow Mechanism of Sand-Water Mixtures in Pipelines. Delft
University Press.
SOIL-WATER MIXTURE AND ITS PHASE 2.15

CASE STUDY 2

Consider that sand particles occupy 27 per cent of the total volume of a dredging
pipeline. The rest is occupied by carrying water. The sand-water mixture is
discharged from a dredging pipeline at a deposit site. The porosity of sand in a
deposition n = 0.4.

Determine the density ρm of sand-water mixture in the pipeline and the weight
concentration Cw of solids in the mixture. Further, determine the in situ density ρsi
and the spatial volumetric concentration of in situ sand, Cvsi, in a deposition and in a
pipeline.

Inputs:

The spatial volumetric concentration of solids in mixture flow: Cv = 0.27


The density of sand particles: ρs = 2650 kg/m3
The density of carrying water: ρf = 1000 kg/m3
Porosity of sand body in a deposition: n = 0.4.

Solution:

a. Density of mixture (ρm) & weight concentration of solids in mixture (Cw)

Eq. 2.9 determines the density of mixture

ρm = ρf(1-Cv) + ρsCv = 1000(1-0.27)+2650x0.27 = 1445.5 kg/m3.

Eq. 2.10 determines the weight concentration

ρs 2650
Cw = Cv = 0.27 = 0.495, i.e. 49.5 %.
ρm 1445.5

b. In-situ density (ρsi) & in-situ volumetric concentration of soil (Cvsi) in a


deposition

ρ − ρsi
According to Eq. 2.3 the porosity n = s and thus
ρs − ρ f

ρsi = ρs − (ρs − ρ f )n = 2650 – 1650 x 0.4 = 1990 kg/m3 .

In a deposition the density of the sand-water mixture is equal to ρsi and thus Cvsi =1
(see Eq. 2.11).
2.16 CHAPTER 2

c. The in-situ volumetric concentration in a pipeline

Eq. 2.12 determines the volumetric concentration of in situ material in a pipeline as

Cv 0.27
C vsi = = = 0.45, i.e. 45 %.
1 − n 1 − 0.4
3.
FLOW OF MIXTURE IN A PIPELINE

3.1 FLOW REGIMES AND PATTERNS

3.1.1 Flow regimes

Generally, a carrying liquid may flow either in a LAMINAR or a TURBULENT


regime in a pipeline. A laminar flow is composed of thin layers (lamina) that move
over each other at different velocities forming a typical parabolic velocity profile in a
pipeline cross section. There is no exchange of mass and momentum between
neighboring layers. Thus each liquid particle has zero velocity components in
directions other than is that of the flow and this is given by an axis of a conduit. A
stability of a laminar flow is given by Reynolds number of the flow and its value 2300
is experimentally determined as a threshold for the maintaining of a laminar flow
regime in a conduit. From this it is clear that a laminar regime can hardly occur in a
dredging pipeline when the flowing carrying liquid is water. This has the value of
kinematic viscosity of about 10-6 m2/s and considering the diameter of a dredging
pipeline of a typical value 1 meter, the velocity of a carrier should be maximally 2.3
mm/s to maintain a laminar regime of flow. However, a laminar flow is maintained to
higher velocities when viscosity of the carrier is higher than that of water. This is the
case when non-settling or very-slowly-settling solid particles are transported at high
concentration with water in a pipeline. Water and very fine particles form together a
carrier of high density and viscosity. As a result the laminar flow might occur in a
dredging pipeline if highly dense non-settling mixtures are transported. In practice,
however, the operational velocity is often higher than the threshold velocity for a
laminar flow even for these high concentrated non-settling mixtures. A turbulent flow
is a result of disturbances occurring at the interface between neighboring layers if the
difference in their velocities becomes higher than is acceptable for maintenance of the
laminar regime. Turbulent eddies are developed as a result of the disturbances. The
turbulent eddies are responsible for an intensive random transfer of mass and
momentum in all directions within a liquid stream. This is sensed as a continuous
fluctuation of velocity of fluid particles in time and space within a stream. The
turbulent flow regime is typical for dredging pipelines. The flow eddies due to
turbulence produce energy dissipation additional to that due to friction in a laminar
flow. Turbulent flows dissipate much more mechanical energy than laminar flows.

3.1.2 Flow patterns

A tendency of a solid particle to settle in a flowing carrying liquid and a tendency of a


flowing carrier to suspend solid particles are the most important indicators of a pattern
of a flow of solid-liquid mixture in a pipeline. The settling tendency of a solid particle
to settle is given by the particle settling velocity and the tendency of a carrying stream
to suspend the solid particles is given an intensity of turbulence, i.e. basically by mean

3.1
3.2 CHAPTER 3

velocity of a stream in a pipeline. The mixture flow is considered FULLY


STRATIFIED if intensity of turbulence of a carrier flow is not sufficient to suspend
any solid particle in a pipeline. Then all solid particles occupy a granular bed that is
either stationary or slides over the bottom of a pipeline. The opposite extreme to the
fully-stratified flow is a FULLY-SUSPENDED flow in which all solid particles are
suspended within a stream of a carrying liquid. No granular bed occurs in a pipeline.
The fully-suspended flow may be considered pseudo-homogeneous if a distribution of
solid particles across a cross section of a stream is almost uniform. This is usually the
case if solid particles of silt or clay size are transported in a pipeline. Fully-suspended
flow exhibiting a certain concentration gradient across a stream is typical for fine to
medium sand mixtures flowing at high velocities. An intermediate flow pattern – the
PARTIALLY-STRATIFIED flow – is most usual during dredging operations. A
mixture flow exhibits a considerable concentration gradient across a pipeline cross
section indicating an accumulation of a portion of solids near the bottom of a pipeline
and a non-uniform distribution of the rest of solids across the rest of a pipeline
cross-sectional area. This pattern is also known as a heterogeneous flow.

The following flow patterns occur in dredging pipelines:

Homogeneous flow of non-Newtonian mixtures


– flow of clay and silt mixtures at high concentrations
Pseudo-homogeneous flow of Newtonian mixtures
- coarse silt or fine sand mixtures (in case of fine sand the velocities must be
considerably higher than is the deposition-limit velocity)
Slightly-stratified heterogeneous flow (partially-stratified flow without a stationary
deposit)
- medium or medium to coarse sand mixture in which a majority of solid particles is
suspended and only a small portion of particles travels within a granular bed
Very-stratified heterogeneous flow
- medium to coarse sand, coarse sand or fine gravel mixture in which a majority of
solid particles travels within a granular bed and only a small portion of particles is
suspended (in case of fine gravel the velocities must be considerably higher than
is the deposition-limit velocity)
Fully-stratified flow with an eroded top of the bed
- fine to medium or medium gravel mixture in which a great majority of solid
particles travels within a granular bed and only a small portion of particles is
either sheared or moves by jumping and rolling (the process called “saltation”)
over the top of the sliding granular bed
Fully-stratified flow
- medium to coarse or coarse gravel mixture or mixture containing cobbles and
boulders, in this flow all solid particles travel within a granular bed

The particle settling velocity and the mean mixture velocity (indicating a measure of
an intensity of turbulence in a mixture stream) are the most important parameters to
determine the flow pattern in a dredging pipeline. A pipeline size is an additional
parameter influencing the pattern. This might play an important role if flows are
scaled from laboratory pipes (typically of a diameter between 50 mm and 150 mm) to
dredging pipelines (a diameter between 500 mm and 1200 mm).
FLOW OF MIXTURE IN A PIPELINE 3.3

If the velocity in a dredging pipeline is too low the carrier stream might be incapable
to keep all solid particles (of sand size and coarser) in motion. The bed composed of
solid particles settled at the bottom of a pipeline will not slide and the flow pattern
with a stationary bed develops. As discussed later, a pipeline operation with such
flow pattern is inefficient and potentially dangerous.

I. Pseudo-homogeneous flow

II. Heterogeneous flow without a granular bed at the bottom of a pipe

III. Heterogeneous flow with a low developed bed

IV. Heterogeneous flow with an en bloc sliding bed

V. Stratified flow with a stationary bed

Figure 3.1. Various patterns of mixture flow with distributions of mixture velocity
(vm), spatial concentration (cvi) and delivered concentration (cvd).

Knowledge of a flow pattern in a dredging pipeline is important for the design and
prediction of operational parameters of a dredging pipeline. Important parameters for
the design and operation of a dredging pipeline are those which provide information
about the safety and economy of dredging pipeline operation. These are:
- the mean mixture velocity and its important values
- the production of solids
- the frictional head loss
- the specific energy consumption.
3.4 CHAPTER 3

3.2 MEAN MIXTURE VELOCITY AND ITS IMPORTANT


VALUES

3.2.1 Definition

Mean velocity in a pipeline is a basic parameter characterizing pipeline flow. It is


defined as the bulk velocity, V, of a matter (liquid, solids, mixture) obtained from the
volumetric flow rate, Q, of a matter passing a pipeline cross section of the area, A.
The equation V = Q/A is for a circular pipe of an inner diameter D written as

4Q
V= (3.1).
πD 2

V bulk velocity, i.e. mean velocity of a matter in


a pipe cross section [m/s]
Q volumetric flow rate of a matter in a pipe [m3/s]
D pipe diameter [m]

The determination of an appropriate mean mixture velocity, Vm, is crucial to safe and
low-cost pipeline operation.

3.2.2 Deposition-limit velocity

Solid particles of sand/gravel size and density tend to settle in a flowing carrier.
Usually, these solid particles are distributed non-uniformly in a pipeline flow. If the
carrier velocity is too low for the carrier lift forces to suspend all solid particles, a
portion of the particles forms a bed at a bottom of a slurry pipeline. With extremely
large particles and/or extremely low mean velocities in a pipeline all particles occupy
a bed. The threshold velocity at the initiation of turbulent suspension, Vtt, is the mean
velocity of the mixture in a pipeline cross section at which the first solid particles
leave the bed, being supported by the diffusive effect of carrier turbulent eddies. This
velocity is used in the evaluation of a measure of a flow stratification, but for practical
pipeline operation the threshold velocity at which solid particles occupying a bed at
the bottom of a pipeline stop their sliding and start to form a stationary deposit, i.e. a
stationary bed, is more important. Operation below this threshold velocity might be
inefficient and potentially dangerous. Under certain circumstances a stationary
deposit may be transformed into a solid plug which blocks the pipeline. The mean
slurry velocity at the limit of stationary deposition is called the deposition-limit
velocity, Vdl, or, less accurately, the critical velocity. Slurry flow at velocities above
the deposition-limit value is free of a stationary deposition.

3.2.3 Minimum velocity

The mean slurry velocity at which the least energy is dissipated in slurry flow is
called the minimum velocity, Vmin. The minimum slurry velocity determines the
velocity at which the slurry flow is most economical of energy. This is the optimal
transport velocity for slurry of a given slurry density. It is well known, however, that
FLOW OF MIXTURE IN A PIPELINE 3.5

the minimum velocity is not a viable operation velocity in a conveying system. In


practice the operation velocity, a result of an interaction between a pipeline and a
pump, is taken as Vm > Vmin. This avoids an unstable transport regime in a
conveying system experiencing a variation in Vm.

At the minimum velocity the hydraulic gradient (see below) is minimal (Fig. 3.3).
Therefore the derivative of the hydraulic-gradient correlation Im = fn(Vm, etc.) (see
Chapter 4) determines a relationship between the minimum velocity and additional
parameters of mixture flow: dIm/dVm = 0 at Vm = Vmin.

3.2.4 Relation between deposition-limit velocity and minimum velocity

Sometimes the deposition-limit velocity is considered equal to the minimum velocity


(Fig. 3.2). In majority of slurry pipeline conditions, however, the deposition-limit
velocity and the minimum velocity differ. To estimate Vdl as equal to Vmin might be
acceptable for low-concentration flows but fails for high concentrated flows. Actually,
the trends seem to be opposite in a development of Vdl and Vmin when solids
concentration grows in a pipeline. The deposition-limit velocity tends to drop while the
minimum velocity tends to grow (see a 0.20-0.50 mm sand in Fig. 3.5).

Some authors have tried to relate the minimum and the critical velocity by using
coefficient of proportionality ξ as

Vmin = ξVdl (3.2)

(e.g. ξ = 0.64/ 6 D according to Jufin & Lopatin, 1966).

3.3 PRODUCTION

3.3.1 Definition

The flow rate of solids transported through a dredging pipeline is termed production
in the dredging practice. This is an important parameter from the economic point of
view. It gives the amount of dry solids (in volume or mass) delivered at the pipeline
outlet over a certain time period. It is defined as the flow rate (either volumetric, Qs,
in m3/s or mass in kg/s) of solids at the outlet of a slurry pipeline.
3.6 CHAPTER 3

3.3.2 Production of solids (on the dry-weight basis)

In the dredging practice the volumetric flow rate is handled rather than the mass flow
rate and the production Qs = QmCvd is calculated as

π 2  m3 
Qs = D Vm C vd 3600   (3.3).
4  hour 

Qs volumetric flow rate of solids, i.e. production of


solids [m3/s]
D pipe diameter [m]
Vm mean mixture velocity [m/s]
Cvd volumetric delivered concentration of solids [-].

During a dredging operation the parameters Vm and Cvd are usually measured in a
pipeline of known D so that the production of solids given by a solids flow rate can be
determined.

3.3.3 Production of in-situ soil

For the payment of a dredging work, however, the production based on in-situ volume
of transported soil is decisive. As discussed in Chapter 2 the in-situ volume of soil is
composed of the volume of solid particles and of the volume of pores in an in-situ soil
C
body. The delivered concentration of the in situ soil C vdsi = vd so that the
1− n
production of in-situ soil can be calculated as

π 2 Q  m3 
Qsi = D Vm C vdsi 3600 = s   (3.4).
4 1− n  hour 

Qsi production of in-situ soil [m3/s]


Qs production of solids [m3/s]
n porosity for in-situ soil [-]
D pipe diameter [m]
Vm mean mixture velocity [m/s]
Cvdsi volumetric delivered concentration of in-situ soil [-].

Since the porosity gives a value lower than one (typically n=0.4 for a loose-poured
sand), the production of in situ soil is higher than the production of the solid particles.
FLOW OF MIXTURE IN A PIPELINE 3.7

3.4 FRICTIONAL HEAD LOSS (HYDRAULIC GRADIENT)

3.4.1 Definition

Flow resistance is given by the amount of mechanical energy dissipated in a slurry


flow when flowing through a pipeline. The mechanical energy balance along a
pipeline section - expressed by the Bernoulli equation - shows that energy dissipation
in a steady slurry flow is characterized by the pressure difference along a horizontal
pipeline section of constant diameter. The resistance is evaluated as

- the pressure drop ∆P = P1-P2 (differential pressure over a pipeline section


confined at the inlet by the pipeline cross section 1 and at the outlet by the
pipeline cross section 2) [Pa],

∆P
- the pressure gradient (pressure drop over a pipeline section divided by the
L
length L of a pipeline section between cross sections 1 and 2) [Pa/m] or

∆P
- the hydraulic gradient due to friction, also termed the frictional head loss
ρf gL
(Im) (head that is lost owing to friction is divided by the length of a pipeline
section, L), which is dimensionless and expresses the pressure gradient by the
ratio of meter liquid column and meter pipeline length.

∆P
Head, , is a measure of the mechanical energy of a flowing liquid per unit mass.
ρf g
It is expressed as the height of the fluid column exerting the pressure, which is
equivalent to the pressure differential over a pipeline section.

The sum of the hydraulic gradient due to friction in straight pipeline sections, the
hydraulic gradient due to minor losses in fittings (see Chapter 7) and the geodetic
gradient (the potential energy gain or loss in a mixture from different geodetic heights
at an inlet and an outlet of a pipeline, see Chapters 6 and 7) determine the amount of
energy which has to be fed by pumps to mixture flow in a pipeline.

3.4.2 Pipeline-resistance curve

A relation between the mechanical dissipation due to mixture flow through a pipeline
and the mean mixture velocity is expressed in a pipeline-resistance curve. This relates
∆P
the head (in meter water column) lost due to friction in a straight pipeline with
ρf g
the mean mixture velocity Vm (see Fig. 3.2).
3.8 CHAPTER 3

Figure 3.2. A typical course of a pipeline-resistance curve for water flow


and mixture flow.

.12
=0
30
Head lost due to friction [m.w.c.]

d
Cv

2
0.1
37 r
e,

25 mm mi x tu
0.

v =
,C
low

0
v =

20
r f ous f
,C
e
low
W ogen

15
a te
m
Ho

10
(a 0.18 mm mixture)
5
Developed bed Low bed No bed

0
0 1 2 3
Mixture velocity [m/s]

6
4
Minimum velocity
2 for mixture 0.18 mm

0
0 1 2 3
Mixture flow rate [m3/s]
Figure 3.3. Typical courses of a resistance curve for various sand mixtures
(in a 800 mm pipeline).
FLOW OF MIXTURE IN A PIPELINE 3.9

3.4.3 Course of a pipeline-resistance curve

3.4.3.1 Theoretical considerations

A course of a resistance curve within a wide velocity range indicates different flow
patterns that occur in a pipeline. A resistance curve for water flow in a pipeline is a
parabolic curve. The pseudo-homogeneous flow of fine solids is represented by a
parabolic curve also, only the gradient of a curve is higher (see Fig. 3.3).

Heterogeneous flows are characterized by resistance curves that exhibit the minimum
at so called “minimum velocity” (see Figs. 3.3 & 3.4). A descending curve section at
velocities below the minimum velocity indicates a developed stationary or sliding
bed.

Figure 3.4. Resistance curves and stationary-bed zone.

A threshold velocity between a stationary bed and a sliding bed is illustrated on Fig.
3.4 (the threshold velocity Vdl is sensitive on the concentration of solids in a
pipeline). If the velocity increases above the minimum velocity value the bed
gradually dissolves (particles gradually loose their mutual contacts and they become
suspended by a carrying liquid).

3.4.3.2 Experimental observations

The Figure 3.5 shows resistance curves (representing here the hydraulic gradient due
to friction in a 1-meter horizontal pipe versus mean mixture velocity) for flows of
different soil sorts in a 150-mm pipeline of a test loop in Laboratory of Dredging
Technology of Delft University of Technology. Shapes of measured resistance curves
are very different for various sand and gravel sorts. Furthermore, shapes of measured
resistance curves are very sensitive to concentration of solids in tested mixture flows.
3.10 CHAPTER 3

Generally, flow friction increases with both the particle size and the concentration of
solids in a pipeline. A shape of a pipeline-resistance curve is very different for fine
solids and coarse solids:

- A resistance curve of a fine sand (0.10-0.12 mm sand) indicates a slightly


stratified heterogeneous flow within a velocity range 1.5 – 4.0 m/s. At velocities
higher than 4.0 m/s the flow may be considered pseudo-homogeneous.

- The curves for a medium sand (0.20-0.50 mm sand) indicate a considerable flow
stratification near the deposition-limit velocity and a gradual bed dissolution
within a velocity range 3 – 5 m/s. The pseudo-homogeneous flow regime is
reached at velocities higher than 6 m/s.

- The resistance curves for coarse sand (0.50-1.00 mm sand) indicate that an
interaction between a developed granular bed and suspension flow above the bed
governs a flow behavior at velocities not far above the deposition-limit threshold.
The bed is sheared due to a fast current of suspension above the bed within a
range of mean velocity values near the deposition-limit velocity. Bed shearing
results to a relatively low friction loss. Bed shearing is gradually dumped if
velocity increases (3 – 4 m/s) and the compact bed is restored – friction increases
rapidly. The bed is disintegrated and a portion of solids resuspended if the
velocity grows further (4 – 7 m/s).

- The resistance curves for a gravel flow (3.0-5.0 mm gravel) show that the
frictional head loss drops when the steady sliding of a granular bed is reached by
increasing velocity above the velocity range in which an unstable sliding-stopping
bed was observed (2.2 – 3.5 m/s). A further increase in the mean velocity does not
provide a decrease of the difference between the frictional loss values for mixture
and for water. This is an indication that the sliding bed is not disintegrated even at
high mean mixture velocities. The flow remains stratified.

The Figure 3.5 shows also that the values of deposition-limit velocity, Vdl, vary with
the concentration and that the deposition-limit velocity differs from the minimum
velocity in flows of high concentrations of solids. The value of the deposition-limit
velocity coincides with the minimal velocity (Vdl = Vmin) only if the initial sliding of
the bed and the initial disintegration of the bed occur at the same velocity. Vdl <
Vmin if the bed is thick enough to slide en bloc before it starts to be initially
disintegrated.
FLOW OF MIXTURE IN A PIPELINE 3.11

Figure 3.5. Experimentally determined resistance curves and deposition-limit


velocities for mixture flows of various materials and various mixture densities.
(Data: Laboratory of Dredging Technology, Delft University of Technology)
3.12 CHAPTER 3

3.5 SPECIFIC ENERGY CONSUMPTION

3.5.1 Definition

The efficiency of a slurry pipeline is evaluated by means of a parameter called


specific energy consumption (SEC). The SEC is an appropriate optimization
parameter because it contains both a measure of energy dissipation and of solids load
in a pipe flow. The SEC determines the energy required to move a given quantity of
solids over a given distance in a pipeline. It is defined as a ratio between the power
consumption per meter of pipe and the (dry) solids throughput in a pipe.
Power consumption per meter of a pipe is given by Im.ρf.g.Qm and
solids throughput (mass flow rate of solids) is ρs.Cvd.Qm.
Then
I g  J 
SEC = m in units   (3.5)
SsC vd  kg. m 

or in more practical physical units


I  kWh 
SEC = 2.7 m (3.6).
Ss .C vd  tonne. km 

SEC specific energy consumption [kWh/tonne-km]


Im frictional head loss [mH2O/m']
g gravitational acceleration m/s2]
Ss relative density of solids [-]
Cvd volumetric delivered concentration of solids [-].

3.5.2 Specific power consumption

A similar parameter - specific power consumption (SPC) - can be derived for flow in a
vertical pipe section of a length L that is connected with a pump outlet. The input
power to the pipe section is given by the output power of the pump ρf.g.Qm.H. The
output power of a pipe section is given by the work done to transport the solids (of
mass ms) over a pipe length L in a unit time period ∆t. The output power is actually
the power supplied to solids and it is expressed as ms.g.L/∆t, i.e. ρs.g.Qs.L or ρ
s.g.Cvd.Qm.L. The efficiency of the system represented by a pipe section is then

ρ .g.Q m .C vd .L Ss .C vd .L W
ηs = s =  W  , i.e. [-] (3.7).
ρf .g.Q m .H H

ηs efficiency of the pump-pipe system [-]


L length of a pipe [m]
H pump head [m]
Ss relative density of solids [-]
Cvd volumetric delivered concentration of solids[-].
FLOW OF MIXTURE IN A PIPELINE 3.13

The SPC is the inverse of the efficiency. For a horizontal pipe this derivation loses its
physical meaning because no potential energy is added to the solids during the flow in
a horizontal pipe, thus no work is done on solids. However, the economic significance
of the parameter remains valid. The ratio H/L is now the frictional head loss Im and
I
SPC = m [-] (3.8)
Ss C vd
A comparison of equations gives SEC = g.SPC.

3.5.3 SEC – Production diagram

Figure 3.6. Relationship between SEC and production for various sorts of sand and
various mixture densities in a 150 mm pipeline.
(Data: Laboratory of Dredging Technology, Delft University of Technology)
3.14 CHAPTER 3

The SEC is plotted against solids throughput (the production of solids) in Fig. 3.6 for
flows of various solids tested in a 150 mm pipeline of the test loop in the Laboratory
of Dredging Technology of Delft University of Technology. The lowest values of the
specific energy consumption were found for mixtures of the volumetric
concentrations of solids higher than 25%, flowing at mean velocities equal/similar to
the minimum velocity, Vmin.

According to Fig. 3.6 the specific energy consumed to transport medium sand is
approximately twice that needed to transport fine sand if transport velocity
approaches the minimum velocity in the 150 mm pipeline. The specific energy
consumed to transport coarse sand is approximately twice that needed to transport
medium sand.

3.6 REFERENCES

Jufin, A.P. & Lopatin, N.A., (1966). O projekte TUiN na gidrotransport zernistych
materialov po stalnym truboprovodam. Gidrotechniceskoe Strojitelstvo, 9, 49-52.
4.
MODELING OF STRATIFIED MIXTURE
FLOWS
(Heterogeneous flows)

Almost 50 years ago, substantial progress in the exploitation of the hydraulic transport
of solids in pipelines initiated systematic investigations in this field. With the design
of new industrial pipelines, some of which were of considerable length, the demand
for the reliable models capable of predicting slurry flow behaviour grew. Over the
years, as experimental, theoretical and computational techniques have progressed, the
predictive models have been gradually improved.

The first predictive tools were developed in the 1950's and 1960's using methods of
empirical modeling. The tools were empirical correlations constructed to predict the
basic slurry pipeline characteristics - the frictional head loss and the deposition-limit
velocity - for various slurry flow conditions in a pipeline. The correlations were based
on the experimental measurements of integral parameters of slurry flow in pipelines.
Usually these parameters were mean slurry velocity, volumetric delivered
concentration and pressure in flows of slurry containing particles of certain diameter.
Some of the models have become popular and are still used in practice (e.g. Durand &
Condolios, 1952; Führböter, 1961; Jufin & Lopatin, 1966). They are simple to use and
easy to modify to the user's own data. Recently, a semi-empirical model for a
heterogeneous flow in slurry pipelines which is calibrated by using the
integral-parameter data has been introduced by Wilson et al. (1992-6).

Since the mid 1980's attempts have been made to construct a general model for
solid-liquid flow in pipelines by using a microscopic approach. A microscopic model
defines the laws governing a slurry flow for an infinitesimal control volume of slurry.
A slurry flow mechanism is described by using a set of differential equations for
conservation of mass, momentum and energy in the solid-liquid flow. A microscopic
model provides a numerical solution to the equations in local positions of a pipeline
cross section. As a result, the model predicts the concentration and velocity profiles in
a pipeline cross section, together with the pressure drop over a pipeline length section.
Despite progress in the development of sophisticated experimental techniques which
enable reasonably accurate measurements of the internal structure of the flow
(concentration and velocity profiles) in a slurry pipeline, there is still not enough
information on the slurry flow mechanism at microscopic level.

A suitable compromise between the microscopic and empirical approaches is an


approach using the principles of macroscopic modeling (called also physical
modeling). This approach applies the balance (conservation) equations to a larger
control volume of slurry given, for instance, by a pipeline cross sectional area of
approximately uniform concentration of solids in a unit length of a pipeline. In a

4.1
4.2 CHAPTER 4

chosen control volume, the balance equations are formulated by using mean
quantities, i.e. quantities averaged in the control volume.
Newitt et al. (1955) were the first to apply the balance formulations to a macroscopic
control volume to obtain the friction loss equations for different slurry flow regimes
in a slurry pipeline. Wilson (1970) introduced the concept of a mechanistic force-
balance model to predict the velocity at the limit of stationary deposition in a fully-
stratified flow. Wilson (1976) developed the model further to provide a unified
predictive tool, called a two-layer model, to predict both the limiting deposition
velocity and the frictional head losses in fully-stratified and partially-stratified flows
in a horizontal slurry pipeline.

4.1 EMPIRICAL MODELING

Different approaches to empirical modeling of slurry flow in pipelines are described


by using several widely used models:
- the model of Laboratoire Dauphinois d'Hydraulique (called the Durand model)
- the Führböter model
- the Jufin - Lopatin model
- the Wilson - GIW model.

4.1.1 Model of Laboratoire Dauphinois d'Hydraulique


(Durand model)

FRICTIONAL HEAD LOSS

The empirical model to predict the pressure drop due to friction in the pipeline flow of
slurry was constructed by using techniques for dimensional analysis. Durand and his
co-workers sought an empirical relationship among the dimensionless groups of
quantities anticipated to be of major importance for a description of slurry flow in a
pipeline.

A. Experimental observations:

Experimental data were collected for a reasonably wide range of slurry flow
conditions including several pipeline sizes and sorts of sand and gravel. Based on the
experimental results (for low concentration slurries with delivered concentration Cvd
up to 22%), the following issues were proposed for the pressure loss, represented by
the hydraulic gradient Im, in the heterogeneous slurry flow characterised by constant
particle size d and pipeline size D:

- the solids effect Im-If decreases gradually with increasing mean slurry velocity Vm
in flow of constant delivered volumetric concentration of solids Cvd

- the solids effect Im-If increases approximately linearly with increasing Cvd at
constant Vm.
MODELING OF STRATIFIED MIXTURE FLOW 4.3

B. Construction of the Durand correlation and its ultimate version:

The latter condition was written as Im - If = const.Cvd. To eliminate the direct


influence of the properties specific to one experiment (such as the pipe roughness and
the slurry temperature) the ratio of the solids effect and the hydraulic gradient for
liquid flow, If, was introduced in place of the solids effect alone. Then the condition
was described by the dimensionless group, marked Φ,

I −I
Φ = m f = const. (4.1).
I f C vd

Im hydraulic gradient for mixture flow [-]


If hydraulic gradient for liquid flow [-]
Cvd delivered volumetric solids concentration [-]

The flow coefficient Φ is not constant for slurry flows of different pipeline size D,
solids size d, or slurry flow rate VmA.

An effect of these parameters was introduced by correlating Φ with the Froude


V2
number for mixture flow Fr 2 = m and the Froude number for a solid particle
gD

2 v 2t
Frvt = . The Froude number is a criterion of dynamic similarity for flows with a
gd
dominant effect of inertia and gravity in different flow conditions.
The new dimensionless group was marked Ψ:

2
−1 = Vm gd
Ψ = Fr 2 Frvt (4.2).
gD v t

Fr2 Froude number for mixture flow [-]


Frvt-1 particle Froude number [-]
Vm mean mixture velocity in a pipe cross section [m/s]
g gravitational acceleration [m/s2]
D pipe diameter [m]
d particle diameter [m]
vt terminal settling velocity of a particle [m/s]

A general empirical relationship was established for a resistance due to friction in a


heterogeneous slurry flow as

Φ = KΨn (4.3)

with the empirical coefficients K and n. The Φ - Ψ relationship might be determined


using the hyperbolic curve in the Φ - Ψ plot (Fig. 4.1) proposed by Durand &
Condolios or using the curve approximation giving K = 180 and n = -1.5. A plot Φ
4.4 CHAPTER 4

versus Ψ was proposed as a unified pattern for the evaluation of experimental data for
solids of d = 0.18 - 22.5 mm and pipelines of D = 40 - 580 mm.

The ultimate correlation obtained by Durand et al. is


−1.5
I m − If  V 2 gd 
= 180 m  (4.4)
I f C vd  gD v t 
 

Figure 4.1. The Φ = KΨn relationship of the Durand model


(Φ on the vertical axis and Ψ on the horizontal axis).
4,0

H2 Ca teg o riën va n
3,0
za nd - en g rind so orten d n (mm ) C 'x

H2 0,20 3,42
L3 0,27 1,96
L4 0,37 1,34
L3 L5 0,58 1,06
2,0 L6 0,89 0,88
L7 1,33 0,80
L8 2,05 0,72
A9 2,80 0,67
1,5 L4 A10 4,20 0,62

L5
1,0
L6
0,9
L7
0,8
C 'x

L8
A9
0,7
A10
0,6
d ia meter va n d e korrel d n in mm
0,5
0,2 0,3 0,4 0,5 0,6 0,7 1 1,5 2 3 4 5 6

d n (mm )

Figure 4.2. Modified particle Froude number Frvt −1 = gd , here marked C'x ,
vt
determined experimentally for various sorts of sand and gravel
by Durand et al.
MODELING OF STRATIFIED MIXTURE FLOW 4.5

gd
The plot on Fig. 4.2 can be used to estimate the value for sand or gravel
vt
particles.

According to van den Berg (1998) the relationship between Frvt-1 and d can be
obtained also from the "best fit curve" correlation

−1 = gd = 0.69 − 0.01d1.5 + 0.161


Frvt (4.5).
vt d1.5

An effect of broad particle size distribution is taken into account by determining an


average value of the particle Froude number from values of Froude number for soil
fractions pi of different sizes di

−1 gd i
∑ Frvt,i .pi ∑ .p i
−1 gd i i v t ,i
Frvt = = = (4.6).
vt 100 100

Over the years the Durand type of correlation has been tested by using different
experimental data bases and a considerable number of values has been proposed for K
and n by various investigators (see survey in Kazanskij, 1978).

C. Discussion on the applicability of the Durand correlation:

An advantage of the Durand method is that it provides a simple relationship that


covers a wide range of slurry flow conditions and correlates all basic parameters
influencing the behaviour of slurry flow in a pipeline.

A major disadvantage is the inaccuracy of the determination of Φ for the extreme


values of the coefficient Ψ (see Fig. 4.1). The Φ - Ψ curve is very steep at low Ψ. A
small difference in Ψ may create a big difference in Φ (so predicted Im may differ by
from ten per cent to several hundred per cent). At high Ψ values the coefficient Φ
decreases very slowly with increasing Ψ. The regions of the extreme Ψ values
represent a transition from heterogeneous flow to the extreme slurry flow patterns:
fully-stratified for the lowest values of Ψ and fully-suspended (pseudo-homogeneous)
for the highest values of Ψ. The insensitivity of the correlation at its extremes reveals
the fact that the model does not reflect different slurry flow patterns.

Doubts about the applicability of the correlation to a wide range of slurry flow
conditions have been confirmed by a number of tests during the years. A large
discrepancy between Durand’s prediction and experimental data has been experienced
specifically for a coarse slurry flow exhibiting considerable stratification. The
correlation might, however, provide a satisfactory prediction for medium and medium
to coarse sand mixtures at flows falling within the approximate range 4 < Ψ < 15.
4.6 CHAPTER 4

DEPOSITION-LIMIT VELOCITY (CRITICAL VELOCITY)

The deposition-limit velocity Vdl is given by an empirical correlation based on visual


observations of the initial formation of a stationary bed in pipelines for different mixture
flow conditions.
V2
Durand experiments showed that the Froude number Fr 2 = e remained constant for
gR h
pipeline flow when a stationary bed was formed and gradually became thicker under
decreasing Vm. The Froude number for flow above the stationary bed was based on the
velocity above a stationary bed, Ve, and on the hydraulic radius, Rh, of discharging area
above the stationary bed. The constant value for Fr2 was experienced in flows of
constant solids density, particle diameter and delivered concentration. For flow
conditions at the beginning of the stationary bed (Ve = Vm = Vdl and D = 4Rh) this
condition was written as
2
Vdl
2
Fr = = const. (4.7)
gD

Figure 4.3. Modified Froude number FL for a determination of the deposition-limit


velocity according to the Durand et al. model.
MODELING OF STRATIFIED MIXTURE FLOW 4.7

An effect of various particle diameters and delivered concentrations Cvd on the value of
the Vdl was expressed in the empirical relationship FL = f(d, Cvd) presented as a graph
(Fig. 4.3).

The correlation for deposition-limit velocity by Durand et al. is

Vdl = FL 2g(Ss − 1)D (4.8)

Vdl deposition-limit velocity


(critical velocity) [m/s]
FL empirical coefficient; graph FL = f(d, Cvd) [-]
g gravitational acceleration [m/s2]
Ss relative density of solids [-]
D pipe diameter [m].

4.1.2 Führböter model

FRICTIONAL HEAD LOSS

A. Experimental observations:

Experimental data were collected for slurry flow conditions in a 300 mm laboratory
pipeline for sand and gravel of particle size range between 0.15 mm and 1.8 mm.
Based on the experimental results, the following issues were proposed for the
hydraulic gradient Im in the heterogeneous slurry flow characterised by constant
particle size d and pipeline size D.

B. Construction of the Führböter correlation and its ultimate version:

The correlation was found

C vi
I m − I f = Sk (4.9)
Vm

in which Sk was the empirical coefficient dependent on solids properties.


Practical calculations are done for Cvd instead of Cvi, thus the slip effect was
incorporated to obtain Cvd in the above equation. The constant value of the slip ratio
Cvd/Cvi = 0.65 was considered to hold for all mixture flow conditions. The transport
factor Skt was obtained by Skt = Sk.Cvi/Cvd.

The values of the factor Skt were empirically determined:

Skt = 2.59 dm - 0.037 for 0.2 < dm < 1.1 mm


Skt from graph on Fig. 4.4 for 1.1 < dm < 3.0 mm
Skt is approximately 3.3 for dm > 3.0 mm.
4.8 CHAPTER 4

3,5

3,0

2,5

2,0

1,5
Skt (m / s)

1,0

0,5

0
0,2 0,6 1,0 1,4 1,8 2,2 2,6 3,0
0 0,4 0,8 1,2 1,6 2,0 2,4 2,8

d mf

Figure 4.4. Transport factor Skt for the Führböter correlation.

The Führböter correlation is

C vd
I m − I f = S kt (4.10)
Vm

Im hydraulic gradient for mixture flow [-]


If hydraulic gradient for liquid flow [-]
Skt transport factor [m/s]
Cvd delivered volumetric solids concentration [-]
Vm mean mixture velocity in a pipeline [m/s]

C. Discussion on the applicability of the Führböter correlation:

Advantage:
The model is very easy to use and to calibrate by own data (only one coefficient has to be
determined).

Disadvantage:
The transport factor Skt must cover all effects of particle settling process (the settling
velocity is not handled by the model) and effects of various soil and liquid densities
(these are not explicitly handled by the model) on energy dissipation in a mixture flow.
Thus the Skt factor value, determined experimentally for certain flow conditions, can
hardly be considered applicable to any different conditions.
Furthermore, the assumption of a constant slip ratio value is unacceptable for mixture
flows of different particle sizes, mean mixture velocities and solids concentrations.
The model offers no possibility to introduce an effect of a broad PSD on flow resistance.
MODELING OF STRATIFIED MIXTURE FLOW 4.9

4.1.3 Jufin - Lopatin model

This model was constructed as a proposal for the Soviet technical norm in 1966. The
authors did not submit a new model but selected the best combination of correlations for
the frictional head loss and the critical velocity from four models submitted by different
Soviet research institutes.

FRICTIONAL HEAD LOSS

A. Experimental observations:

Four submitted models were tested by a large experimental database collected by a


number of researchers. The database contained data from both laboratory and field
measurements (including data from dredging installations). The data covered a wide
range of pipeline sizes (103 – 800 mm) and particle sizes (sand and gravel, 0.25 - 11
mm).

B. Construction of the Jufin - Lopatin correlation and its ultimate version:

The correlation was based on the empirical experience suggesting that the hydraulic
gradient Im at the minimum velocity Vmin was independent of the mixture flow
properties and it was three times higher than the hydraulic gradient of water flow at the
same velocity in a pipeline. Thus Im = 3If at Vmin. This was experienced also in the
American dredging industry (see Turner, 1996).

The frictional-head-loss correlation by Jufin & Lopatin (in the revised version by
Kobernik, 1968) is

  Vmin  
3

I m = I f 1 + 2    (4.11).
  Vm  
 

The minimum velocity Vmin is given by the empirical correlation


1
 ∗ 
Vmin = 5.3 C vd .ψ .D  6 (4.12)
 
in which the parameter ψ* = f(d) is determined either using a table by Jufin & Lopatin
(see Tab. 4.1) or calculated as modified Froude number of a solid particle, ψ* = Frvt1.5.

Im hydraulic gradient for mixture flow [-]


If hydraulic gradient for liquid flow [-]
Vm mean mixture velocity in a pipeline [m/s]
Vmin minimum velocity [m/s]
Cvd delivered volumetric solids concentration [-]
ψ* particle settling parameter [-]
D pipeline diameter [m]
4.10 CHAPTER 4

Table 4.1. Particle settling parameter for the Jufin-Lopation model.

particle settling
particle settling
size fraction of solids, parameter, ψ*
parameter, ψ*
d [mm] Jufin & Lopatin
Jufin (1971)
(1966)
0.05 - 0.10 0.0204 0.02
0.10 - 0.25 0.093 0.2
0.25 - 0.50 0.404 0.4
0.50 - 1.00 0.755 0.8
1.0 - 2.0 1.155 1.2
2.0 - 3.0 1.50 1.5
3.0 - 5.0 1.77 1.8
5 - 10 1.94 1.9
10 - 20 1.97 2.0
20 - 40 1.80 2.0
40 - 60 1.68 2.0
> 60 1.68 2.0

An effect of broad particle size distribution is taken into account by determining an


average value of the modified particle Froude number from values of modified Froude
number for soil fraction pi of different size di

1.5 *
∑ Frvt,i .pi ∑ ψ (d i ).pi
ψ* = Frvt
1.5 = i = i (4.13).
100 100

C. Discussion on the applicability of the Jufin-Lopatin correlation:

Advantage:
The model was based on experiments carried out on large pipelines and thus it is
considered suitable for pipeline-flow predictions in dredging.

DEPOSITION-LIMIT VELOCITY (CRITICAL VELOCITY)

Jufin and Lopatin proposed the following correlation for the deposition-limit velocity

1 1
Vdl = 8.3D 3  C vd .ψ∗  6 (4.14).
 
MODELING OF STRATIFIED MIXTURE FLOW 4.11

4.1.4 Wilson - GIW model

The semi-empirical Wilson - Georgia Iron Works model for heterogeneous flow in
slurry pipelines is based on considering the heterogeneous flow as a transition
between two extreme flows governed by different mechanisms for support of a solid
particle in the stream of the carrying liquid: the fully-stratified flow (all particles are
transported as a contact load) and the fully-suspended flow (all particles are
transported as a suspended load).

FRICTIONAL HEAD LOSS

A. Experimental observations:

Circuit tests in the experimental laboratory of the GIW Inc. provided a data base for a
verification of the friction-loss correlation. The circuits are of the pipeline size 200
mm, 440 mm respectively. Data for medium to coarse sands in mixtures of delivered
concentrations up to 0.16 were used.

B. Construction of the correlation and its ultimate version:

The model for partially-stratified (heterogeneous) flow operates with a parameter V50
expressing the mean slurry velocity at which one half of the transported solid particles
contribute to a suspended load and one half to a contact load. An equation for this
velocity expresses the influence of suspension mechanisms from the carrier turbulent
diffusion and the hydrodynamic lift acting on particles larger than the sub-layer
thickness in the near-wall region.

The energy dissipation due to the presence of solid particles in a carrier flow is
predominantly due to mechanical friction between contact-load particles and a
pipeline wall. Basically, a resisting force of the contact bed against the carrier flow is
related to the submerged weight of the bed via the coefficient of mechanical friction.
Thus at velocity Vm = V50 the pressure loss due to presence of solids (∆Pm - ∆Pf) is
due to the submerged weight of the moving bed containing one half of the total solid
fraction [0.5Cvd(ρs - ρf)g] times the friction coefficient (µs). Rewritten for head
losses this basic balance is

Im - If = 0.5µsCvd(Ss-1) (4.15)

for the condition Vm = V50.


Experimental data, plotted in log-log coordinates, showed a linear relationship
I m − If
between the ratio (called the relative solids effect) and the mean mixture
C vd (Ss − 1)
velocity Vm. The relationship was found the same for flows of different
concentrations in pipes of different sizes (Fig. 4.5). A slope of the line in the plot was
considered sensitive only on the particle size distribution of transported solids.
4.12 CHAPTER 4

Figure 4.5. Relationship between relative solids effect and mean slurry velocity for
masonry sand mixture (d50 = 0.42 mm), after Clift et al. (1982).

This led to the following form of the Wilson - GIW correlation

−M −M
I m − If V  V 
= 0.5µs  m  = 0.22 m  (4.16)
C vd (Ss − 1)  V50   V50 

Im hydraulic gradient for mixture flow [-]


If hydraulic gradient for liquid flow [-]
Cvd delivered volumetric solids concentration [-]
Ss relative density of solids [-]
Vm mean mixture velocity in a pipeline [m/s]
V50 value of Vm at which one half of solids is
suspended in a carrier flow [m/s]
µs coefficient of mechanical friction between
solids and the pipeline wall (µs = 0.44) [-]
M empirical exponent sensitive on PSD [-]

V50 should be obtained experimentally or estimated roughly by the approximation


MODELING OF STRATIFIED MIXTURE FLOW 4.13

0.45
 S − 1
V50 ≈ 3.93(d 50 ) 0.35  s  (4.17).
 1.65 

in which the particle diameter d50 is in mm and the resulting V50 in m/s. The
exponent M is given by the approximation

−1
  d 
M ≈ ln 85  (4.18).
  d 50 

M should not exceed 1.7, the value for narrow-graded solids, nor fall below 0.25.

C. Discussion on the correlation:

The Wilson - GIW model gives a scale-up relationship for friction loss in slurry
pipelines of different sizes transporting solids of different sizes at different
concentrations. It is based on the assumption that there is a power-law relationship
between the relative solids effect and the mean slurry velocity that is valid in all slurry
flow conditions. The exponent M of this relationship is assumed to be dependent on
the particle size distribution only.
4.14 CHAPTER 4

4.1.5 MTI Holland model

DEPOSITION-LIMIT VELOCITY (CRITICAL VELOCITY)

In MTI Holland the correlation has been developed (see e.g. van den Berg, 1998) for
the threshold velocity between the "fully suspended heterogeneous flow" regime and
the regime of "flow with the first particles settling to the bottom" of a pipeline. This
velocity was considered as the lowest acceptable velocity for a economic and safe
operation of a dredging pipeline and was therefore also called the critical velocity

1
 
Vcrit = 1.7 5 −
1  D  C vd  6 Ss − 1 (4.19)
 d mf   C + 0.1  1.65
   vd 

In Eq. 4.19 the particle diameter dmf is in millimetres and the pipe diameter D in
meters.

The correlation has an advantage of being based on data including those from various
dredging pipelines. MTI recommends the correlation for grains of sand and gravel
size and pipelines larger than 200 mm.

Figure 4.6. Critical velocity according to the MTI model (Eq. 4.19). The nomograph
does not include the effect of deliverde concentration Cvd.
MODELING OF STRATIFIED MIXTURE FLOW 4.15

4.2 PHYSICAL MODELING

4.2.1 Prediction of fully-stratified flow of mixture using


NOMOGRAPHS and/or approximations based on outputs of a
physical two-layer model

FRICTIONAL HEAD LOSS

The frictional head loss in the fully-stratified flow is can be predicted successfully
by using a two-layer model. It can be computed in its original shape (a set of mass and
force balance equations) by iteration. To avoid these computations the nomograph
was constructed as an interpolation of typical outputs from the original two - layer
model.

A nomograph on Fig. 4.7 provides the values of Im in the fully stratified flow for
various combinations of input d, D, Ss, Cvd and Vm. The nomograph is composed of
a locus curve, determining the boundary of the stationary deposit zone, and of a set of
fit-function curves relating

I −I I m − If
- the relative excess pressure gradient m f = with
I pg 2µs (Ss − Sf )C vb

Vm
- the relative velocity Vr = for different
Vsm

C vd
- relative concentrations C r = .
C vb

Im hydraulic gradient for mixture flow [-]


If hydraulic gradient for liquid flow [-]
Ipg hydraulic gradient for plug flow [-]
µs coefficient of mechanical friction between
solids and the pipeline wall [-]
Ss relative density of solids [-]
Sf relative density of carrying liquid [-]
Cvd delivered volumetric solids concentration [-]
Cvb loose-poured bed concentration,
typically Cvb = 0.60 [-]
Vm mean mixture velocity in a pipeline [m/s]
Vsm maximum value of Vdl for various solids
concentrations in flowing mixture [m/s]
4.16 CHAPTER 4

Figure 4.7. Curves of relative excess pressure gradient, from Wilson et al. (1992).

The nomograph curves can be approximated by the simple expression

− 0.25
I m − If  Vm 
=   (4.20)
C vd (Ss − 1)  0.55Vsm 

in which Vsm is the maximum value of deposition-limit velocity for different solids
concentrations in slurry flow of certain Ss, d and D. The Vsm is determined from the
"demi-McDonald" nomographic chart (see further below) or by its approximation

0.55
 µ (S − Sf ) 
8.8 s s  D 0.7d150
.75
 0.66 
Vsm = (4.21)
d 50 + 0.11D 0.7
2

D pipeline diameter [m]


d50 mass-median particle diameter [mm]

in which d50 is in millimetres and D in metres.


MODELING OF STRATIFIED MIXTURE FLOW 4.17

DEPOSITION-LIMIT VELOCITY

The computation of the force balance at incipient motion of the bed in fully-stratified
flow gives a locus curve - the curve relating the deposition-limit velocity, Vdl, with
the position of an interface between the layers in the pipeline cross section. The locus
curve has a maximum (see Fig. 4.7) which determines the maximum velocity at the
limit of stationary deposition Vsm.

Maximum velocity at the limit of stationary deposition, Vsm:

Wilson (1979) processed Vsm values obtained as the model outputs for a variety of
values of input parameters (d50, D, Ss) to the nomographic chart (Fig. 4.8),
sometimes called the demi-McDonald.

The demi-McDonald curve has a turbulent branch (for small particles of diameter less
than approximately 0.5 mm) and a fully-stratified branch (for particles larger than
approximately 0.5 mm). The threshold particle size delimits two branches at the peak
of the nomographic curve.

The fully-stratified branch of the demi-McDonald curve was constructed from outputs
of the model for fully-stratified flow. According to this part of the demi-McDonald
curve, Vsm decreases with increasing particle size in a pipeline of a certain D. Thus a
lower Vm is required to initiate motion in a coarser bed. An explanation of this
phenomenon requires knowledge of physical principles mixture motion in a stratified
flow expressed using a force balance in a two-layer model. This will be discussed
later in paragraph 4.2.2. Basically, the faster flowing of the coarser bed, when
compared to the finer beds, is caused by a fact that the top of the coarser bed is
rougher (the roughness is related to the size of particles occupying the bed top). Via
the rougher interface the higher driving force is transmitted to the bed from the faster
flowing carrier above the bed.
Vsm in turbulent branch is affected by a variable thickness of bed at an incipient
motion under different flow conditions. The bed thickness diminishes owing to a
turbulent suspension process that picks up the particles from the bed surface and
suspend them in flow above the bed. Thus the mixtures containing fine particles
create thinner bed than mixture of coarse particles. Lower mean flow velocity is
required to put the thinner bed into a motion. Therefore the maximum deposition-limit
velocity drops with a size of particles transported in a pipeline.

The entire demi-McDonald nomograph can be approximated by the fit function ( Eq.
4.21)
0.55
 µ (S − Sf ) 
8.8 s s  D 0.7d150
.75
 0 .66 
Vsm = .
d 50 + 0.11D 0.7
2
4.18 CHAPTER 4

snelheid b ij limiet va n sc huivend d ep ot v (m/ s) voor korrels met ρk=2650 kg/ m 3

0,10 10 1,0
8,0
7,0
0,11 6,0
5,0 1,1 1,1
0,12
4,0
0,13 1,2 1,2
3,5
1,3
0,14
3,0 1,4 1,3
v (m/ s) voor korrels met ρk=2650 kg/ m 3

0,15 1,5
0,5
0,4

0,16 0,3 2,5 1,6 1,4


1,0
1,7 rela tieve
0,17 1,8 d ic htheid 1,5
2
0,18
3 0,2 2,0 ρk
0,19 4 2,2 voor korrels 1,6
0,20 5 2,0 2,4 1,7
5
65

1,9 2,6,8 1,8


10
k =2,

0,15 1,8 2 ,0
3 1,9
1,7
et )

20
b uis d ia m eter D (m )

0,25
ls m m

2,0
4,0
rre ( m

30 1,6
5,0
ko r d

0,3 1,5
6,0
or te

2,5
1,4 7,0
vo a m e

8,0
0,4
1,3 3,0
di

0,5 3,5
1,2
20 4,0

1,1 5,0
6,0
1,0 7,0
8,0
1,5 1,0 10

Figure 4.8. Nomographic chart for maximum velocity at limit of stationary deposition
Vsm after Wilson (1979).
MODELING OF STRATIFIED MIXTURE FLOW 4.19

The incorporation of a shear layer to the pattern of a two-layer model has recently led
to a modification of the demi-McDonald nomograph (Wilson, 1992). If the sharp
interface between a bed and the carrier flow above the bed is replaced by the shear
layer the driving force transmitted from the upper-layer flow to the bed is no longer
dependent on the roughness of the top of a bed. Therefore the particle size does not
directly influence the maximum deposition-velocity Vsm. Wilson (1992) proposed
that the Vsm for fully-stratified flow with the shear layer (marked Vsm, max) should
be determined by an approximation

Vsm, max 0.13


 0.018 
=  (4.22).
2gD(Ss − 1)  λ f 

Vsm,max value of Vsm for fully-stratified flow


with the shear layer [m/s]
g gravitational acceleration [m/s2]
Ss relative density of solids [-]
D pipeline diameter [m]
λf Darcy-Weisbach friction coefficient for
liquid flow (from Moody diagram) [-].

The Vsm,max by the approximation is considered the Vsm value if it is lower than the
Vsm value from the nomograph (i.e. also from the Eq. 4.21).

Deposition-limit velocity Vdl - the effect of solids concentration the Vsm value:

Vsm gives the maximum value on the locus curve delimiting the stationary deposit
zone (see Fig. 4.7). The position on the curve is given by a solids concentration. A
locus curve is a product of a two-layer model, thus a physical explanation of the curve
is given by the principles of force balance in a two-layer flow pattern. A shape of a
locus curve is dependent on the several parameters from which the size of a particle
and a pipeline are the most important ones.
Nomographs were developed to determine the critical velocity Vdl from Vsm without
a necessity to compute an original two-layer model. The nomograph curves were also
approximated by fit functions. A process of Vdl determination goes in following
steps:

1. Vsm using demi-McDonald nomograph (Fig. 4.8) or its approximating fit function
(Eq. 4.21)

2. Crm, the concentration at which Vsm occurs, using the nomograph Fig. 4.9 or its
approximating fit function (Eq. 4.23)

3. Vdl/Vsm, the relative deposit velocity, using the nomograph Fig. 4.10 or its
approximating fit functions (Eqs. 4.24 & 4.25).
4.20 CHAPTER 4

ad 2. The relative solids concentration Crm

Figure 4.9. Computer output for relative solids concentration Crm at maximum
deposit velocity, from Wilson (1986).

Fit function:
− 0.17
 S − Sf 
C rm = 0.16D 0.40d − 0.84  s  (4.23)
 1.65 

in which d [mm] and D [m].

Vdl
ad 3. The relative deposit velocity
Vsm

Figure 4.10. Plot of relative deposit velocity Vdl/Vsm versus Cr = Cvd/Cvb for
various values of Crm, from Wilson (1986).
MODELING OF STRATIFIED MIXTURE FLOW 4.21

Fit functions:
for Crm ≤ 0.33
2
ln (0.333)  ln (0.333) 
Vdl ln C rm  ln C rm 
= 6.75C r 1 − C r  (4.24)
Vsm  
 
and for Crm > 0.33
ln (0.666)  ln (0.666) 
2 
Vdl ln (1− C rm ) ln (1− C rm ) 
= 6.75(1 − C r ) 1 − (1 − C r )  (4.25).
Vsm  
 

4.2.2 TWO-LAYER MODEL: principles and mathematical


formulation

When solids such as sand or gravel are transported in a slurry pipeline some degree of
slurry flow stratification usually occurs. This is the effect of the tendency of solid
particles in the carrying liquid to settle. Stratified slurry flow forms a particle-rich
zone and a particle-lean zone in the pipeline cross section. According to the shape of
its concentration profile, the slurry flow may be considered fully-stratified or
partially-stratified.

4.2.2.1 Principles of the model

Figure 4.11. Definition sketch for two-layer model of stratified flow,


after Wilson et al. (1992).

The two-layer model takes into account the slurry flow stratification and transforms a
real concentration profile in a pipeline cross section into a simplified two-layer
pattern. When the slurry flow is fully stratified solid particles transported in the
carrying liquid are all accumulated in a granular bed sliding at the bottom of the
4.22 CHAPTER 4

pipeline (Fig. 4.11). All particles in this lower layer are in mutual contact. The
volumetric concentration of solids in the lower layer of the fully-stratified flow
approaches the concentration value of a loose-poured bed. The stream of the carrying
liquid above the granular bed is particle-free. The position of an interface between
two layers is determined by the angle β.

In a partially-stratified flow a considerable fraction of the total transported solids


mass is suspended in the carrier stream. Suspended particles are assumed not to be
in contact with other particles and the flow boundaries. Velocity distribution - as
well as the concentration distribution - is idealised as being uniform within both the
upper and the lower layers (Fig. 4.12). The distribution of the suspended particles
within an idealised two-layer pattern has been subjected to investigation. Early
versions of the model anticipated a suspension only in the upper layer. For an
idealised flow pattern the recent modification of the model assumes a uniform
distribution of suspended particles along the entire pipeline cross section.

Fully - stratified flow Partially - stratified flow

V1 C1=0 A1 C1 V1

V2 C2 A2 C2 V2

Particle-free layer Suspended layer Contact layer

Figure 4.12. Schematic cross section for two-layer model.

The model is based on the assumption that there are two physical mechanisms for
solid particle support in a pipeline - interparticle contact and particle suspension
in a carrying liquid. Thus solids are transported as both suspended and contact loads.
According to Bagnold (1956), the suspended particles transfer their submerged weight
directly to the carrier, while the submerged weight of the non-suspended particles is
transferred via interparticle contacts to the pipeline wall.

According to the model the behaviour of the flow is governed by the principle of
force balance between driving and resisting forces in the flow in two layers. The
driving force in the flow in a pressurised horizontal pipeline is produced by the
pressure gradient over a pipeline length section. The resisting force is represented by
shear stress exerted by flowing matter at a flow boundary. The same formulation of
the force balance between the driving and resisting forces, combined with a friction
coefficient equation, gives the Darcy-Weisbach equation (Eq. 1.20) for friction losses
in a water pipeline. The Darcy-Weisbach equation is obtained from a two-layer model
for the limiting case in which the particle-free upper layer occupies the whole pipeline
cross section.
MODELING OF STRATIFIED MIXTURE FLOW 4.23

4.2.2.2 Mathematical formulation of the model

The model is composed of a set of equations expressing the conservation of mass and
momentum in a mixture flow in both layers in the pipe section. A set of conservation
equations is computed by iteration. The layer occupying a pipe length L is considered
to be a control volume. Flow in the control volume is steady and uniform. The
quantities describing the properties of the layer are given by values averaged in the
control volume. This can be seen in Fig. 4.12 where V1 and V2 denote the mean
velocity of mixture in the upper (lower respectively) layer. The same is valid for mean
volumetric concentrations C1 and C2 in the layers. Slip between solid phase and
liquid phase is considered negligible within both the suspension flow and the flow of
contact particles. The model-equation parameters defining the geometry of the
schematic cross section for a two-layer model are described in Fig. 4.13.

Fully-stratified flow Heterogeneous flow

O1

A1

O12

A2
β β
O2

Figure 4.13. Geometry of schematic cross-section for two-layer model.

The following equations (Eq. 4.26, 4.29, 4.30, 4.31, 4.32, 4.34, 4.38, 4.39, 4.40, 4.42,
4.43, 4.44, 4.45) form the two-layer model:

Mass balance for flow in two layers

The application of the mass conservation law to a two-layer pattern gives the balances

for slurry flow rate: Qm = Qm1 + Qm2


VmA = V1A1 + V2A2 (4.26),

for solids flow rate: Qs = Qs1 + Qs2


AsVs = CviAVs = C1A1V1 + C2A2V2 (4.27)

and for liquid flow rate:


Qf = Qf1 + Qf2
AfVf = (1-Cvi))AVf = (1-C1)A1V1 + (1-C2)A2V2 (4.28).
4.24 CHAPTER 4

The solids volume balance is written as


CviA = C1A1 + C2A2 (4.29).

Since Cvd = Qs / Qm the Eq. 4.27 for solids flow rate can be written as
CvdAVm = C1A1V1 + C2A2V2 (4.30).

Momentum balance for flow in two layers

A law of conservation of momentum is formulated as force balance between driving


and resisting forces acting on the flow boundaries of each layer in a horizontal
pipeline of the length L:

DRIVING FORCES = RESISTING FORCES.

The force balance for the upper layer is written as


∆PA1 = τ1O1L + τ12 O12 L (4.31)

and for the lower layer as


∆PA 2 + τ12O12 L = (τ 2f + τ 2s )O 2 L (4.32).

A summation of these two equations gives a force balance in the whole pipeline
section

∆PA = τ1O1L + (τ 2f + τ 2s )O 2 L (4.33).

Resistance forces against flow are due to viscous or mechanical friction at flow
boundaries.

4.2.2.3 Friction mechanisms by two-layer model

Mechanical friction between solids and pipeline wall

Solid particles in contact with each other and with the pipeline wall transmit their
submerged weight to the pipeline wall. This is the source of the resisting force exerted
by the contact load solids against the flow driving forces. The force is due to the
solids stress acting at the pipeline wall. In a horizontal pipeline the stress σs between
the solids grains and the pipeline wall acts in a radial direction in the pipeline cross
section, so that it is normal to the pipeline wall. The normal stress σs produces the
(Coulombic) intergranular shear stress at the pipeline wall τs = µsσs. In this
relationship µs is the coefficient of mechanical friction between solid particles and the
pipeline-wall material. The total resistance force exerted by the sliding granular bed is

µsFN = τ2sO2.
MODELING OF STRATIFIED MIXTURE FLOW 4.25

The total normal force, FN, exerted by the normal intergranular stress against the
pipeline wall is obtained by integrating the normal stress over the pipeline perimeter
O2. The result of integrating is

D2
FN = g(ρs − ρf )C vb (sin β − β cos β) (4.34).
2

The force FN differs from Fw, which is the submerged weight of the granular bed.
The force Fw, which represents the gravitational effect on a granular body, is
integrated from the intergranular stress component σw. Only this component can act
to support the bed weight. At each local pipeline-wall position, given by angle α, the
stress σw = σscosα. By integrating over the perimeter O2 of the interface between a
bed and a pipeline wall
D2
Fw = g(ρs − ρ f )C vb (β − sin β cos β) (4.35)
4
D2
where (β − sin β cos β) = A2 and therefore
4
FW = g(ρs − ρ f )C vb A 2 (4.36).

For a dense-phase flow (called also the plug flow) FN = 2FW. Force balance at an
initial motion of a plug flow is written as

∆P ∆P
A = µs 2 FW that is A = 2µs g(ρs − ρ f )C vb A .
L L

Hydraulic gradient required to initiate a plug sliding is then

∆P 1
I pg = = 2µ s (Ss − 1)C vb (4.37).
L ρf g
The shear stress, τ2s, due to mechanical friction between granular bed forming a
contact layer and pipeline wall is velocity-independent. It is determined from σs, the
normal intergranular stress at the pipeline wall. A resisting force due to mechanical
friction between a contact layer and a pipeline wall is perpendicular to the normal
intergranular force FN exerted against the pipeline wall and it is related with FN by
µsFN.

Viscous friction at flow boundaries

Viscous friction between the flowing carrying liquid and the flow boundary is a
velocity-dependent process described by the boundary shear stress (τ1, τ12, τ2f).
Shear stress is related to the velocity gradient between the flowing carrier and the
flow boundary by a friction coefficient expressing flow conditions at the boundary
(see Chapter 2). The conditions are given by the flow regime and the boundary
roughness. The Darcy-Weisbach friction coefficient (Eq. 1.18) is related to the
Reynolds number of the flow, Re, and/or the boundary roughness, k. The friction
4.26 CHAPTER 4

coefficient for water flow in a pipeline is obtained from the Moody diagram or its
computational version (Churchill, 1977). The Reynolds number characterising the
flow in the layer is calculated from the hydraulic diameter, Dh, of the layer (Dh1 =
VD hρf
4A1/O1, Dh2 = 4A2/O2) by Re = .
µf
Friction coefficients λ1, for flow in the upper layer over the pipeline wall of
perimeter, O1, is
V D ρ 
λ1 = fn  1 h1 f , k  (4.38)
 µf 
(λ1 determined using the Moody diagram) and λ2, for flow in the lower layer over the
pipeline wall of perimeter, O2,
V D ρ 
λ 2 = fn  2 h 2 f , k  (4.39)
 µf 
(λ2 determined using the Moody diagram).

The coefficient λ12 for flow in the upper layer over the interface between two layers
(the perimeter O12) differs according to the conditions at the interface. When the
interface is represented by a clearly identifiable flat surface of a contact bed it can be
considered to have a roughness proportional to the diameter of the particles occupying
the bed surface. The interfacial friction law is given by a formula for turbulent liquid
flow over a fully-rough boundary, e.g.

 D 
4 log  + 3.36
8  d12 
= (4.40)
λ12 0 .5 + X
d 
in which X = 5 + 1.86 log 12  for d12/D > 0.002 and X = 0 otherwise. In the
 D 
equation, d12 is the diameter of particle at the interface. This is determined by
assuming that all particles larger than the particle of the d12 size are below the
interface.
The condition of a flat and sharp interface is fulfilled more likely in pipeline flow
containing very coarse particles. In flow containing finer particles the top of a contact
bed is usually sheared off and a sharp interface is replaced by a transition zone, called
shear layer, with concentration and velocity gradient. Thus an interface becomes
virtual rather than real. For the virtual interface the particle-size roughness is no
longer a parameter determining interfacial friction. The Wilson analysis of a flow at
high shear stress above a stationary granular bed revealed that the thickness of the
shear layer is a crucial parameter determining the interfacial friction. This is related to
the hydraulic gradient in the total flow so that the interfacial friction coefficient can
be determined as

0.78
 I 
λ12 = 0.87 m  (4.41).
 Ss − 1 
MODELING OF STRATIFIED MIXTURE FLOW 4.27

The boundary liquid-like shear stresses are written


for the pipe wall in the upper layer

λ
τ1 = 1 ρ f V12 (4.42),
8

for the pipe wall in the lower layer

λ
τ 2f = 2 ρ f V22 (4.43),
8

and for the interface between the upper layer and the lower layer

λ
τ12 = 12 ρf (V1 − V2 )2 (4.44).
8

4.2.2.4 Model coefficients

The two-layer model for fully-stratified flow has the following coefficients:
- the coefficient of mechanical friction, µs, between a granular bed and a pipe wall,
- the viscous friction coefficient, λ12, for liquid flow at the interface between two
layers,
- the viscous friction coefficient, λ1, for liquid flow at the boundary between a liquid
flow in an upper layer and a pipe wall,
- the viscous friction coefficient, λ2, for liquid flow at the boundary between a
granular bed and a pipe wall,
- the volumetric spatial concentration, Cvb, in the contact layer.
These have to be either prepared as inputs to the model (µs, Cvb) or determined in the
model (λ1, λ2, λ12).
4.2.2.5 Model computation: inputs and outputs

To determine the frictional head loss for a certain value of the mean mixture velocity,
Vm, in a pipeline the following input parameters are required.

Input parameters: Liquid: Sf, νf


Solids: d, Ss, µs
Slurry flow: Vm, Cvd, Cvb
Pipe: L, k, D

The model provides the following output parameters, characterising the friction, slip
and simplified concentration and velocity distribution in a fully-stratified flow.

Output parameters: ∆P (Im respectively)


Y12/D
Cvi
V1, V2.
4.28 CHAPTER 4

To determine deposition-limit velocity, Vdl, various values of Vm are used as inputs to


the model and the force balance is sought for Vm value at which V2 = 0.

4.2.2.6 Model adaptation to partially-stratified flow

The two-layer model can be used as a predictive tool for the partially-stratified
(heterogeneous) flow also. The only condition is that the mixture flow is sufficiently
stratified, i.e. it contains a granular bed that is of any significance to the mixture flow
behavior. It seems that this condition is fulfilled for flows of medium sand (if
travelled at velocities near the deposition-limit value) and coarser. In such flows only
one part of solids occupies a granular bed and the rest is suspended in the carrying
liquid. The particle suspension is predominantly due to the dispersive effect of liquid
turbulent eddies, at certain flow conditions, however, the transported particles might
be also suspended due to the shearing of a top of a granular bed. In the sheared layer
the particles have sporadic rather than permanent contact so that at each moment a
portion of solid particles within the sheared layer might be considered suspended. The
shear layer is a transition region between the granular bed where all particles are in
permanent mutual contact and the upperst layer in which particles are either not
present or they are present but have no mutual contact with each other and with a pipe
wall – they are suspended in a carrier stream.

A reliable application of the two-layer model to partially-stratified flow is still subject


to investigation. It requires experimental experience based on the measurements of the
concentration (and velocity) profiles in mixture flow under various conditions. To
date only few such experiments have been carried out. Most recently the experiments
were carried out in the Laboratory of Dredging Technology and Bulk Transport of
Delft University of Technology. This gives us an opportunity to discuss this subject
on a basis of certain experimental experience lacking when earlier model assumptions
had been made.

- the threshold between fully-stratified and partially-stratified flow


An analysis of the interaction between settling particles and turbulent carrier flow
gives a condition for the initiation of particle suspension: the length scale of liquid
turbulence (represented by the mixing length, discussed in Chapter 2) has to be larger
than the particle size. Only particles smaller than a certain portion of the mixing
length could be supported by the eddies, otherwise the turbulent dispersive
mechanism is not effective in suspending transported particles. The turbulent length
scale is considered to be dependent on the local position within a pipeline flow and
thus the average mixing length depended on the pipeline diameter. This gives to rise
to rather complex relationships between the mean mixture velocity at the beginning of
turbulent suspension and particle/pipeline size. However, a rough estimation of a
threshold between the fully-stratified flow and the partially-stratified flow can be
satisfied by a simple d/D ratio value. The flow would be fully stratified for d/D >
0.018 according to Wilson, our data from a 150 mm pipeline suggest rather higher
value of the d/D ratio, 0.03 approximately.
MODELING OF STRATIFIED MIXTURE FLOW 4.29

- the additional model coefficient: the stratification ratio


For fully-stratified flow, the two-layer model considers the upper layer as particle-free
and the lower layer as occupied by particles, all of which are in continuous contact. In
a partially-stratified flow the solids are transported in a carrying liquid both as a
contact load and as a suspended load. The amount of solids occupying a slurry
pipeline is given by the volumetric spatial concentration Cvi that is the sum of a solids
fraction in suspension, Cs, and a solids fraction in contact, Cc. A suitable method
must be used to predict the amounts of suspended solids Cs or of solids in contact Cc.
Assuming a two-layer pattern according to Fig. 4.14, the Cc determines the
concentration of solids in contact within the lower layer, C2c, by recalculating of Cc
from the cross-sectional area of the entire pipeline, A, to the cross-sectional area of
the lower layer, A2, using C2c = Cc.A/A2.

Figure 4.14. Two-layer pattern for the model for partially-stratified flow.

An additional coefficient has to be introduced to the two-layer model to predict the


partially-stratified flow. This coefficient is called the stratification ratio Cc/Cvi.
Although subject to further investigation the stratification-ratio correlation for overall
mixture flow conditions can be written as

Cc  V 
= exp − X m  (4.45).
C vi  vt 

A value of the empirical coefficient X was found equal to 0.018 according to Gillies
et al. (1990) data and 0.024 from tests in our laboratory.

- the buoyancy effect on the bed submerged weight


Modification of the two-layer flow pattern for the partially-stratified flow (see Fig.
4.14) required modification of the method used to determine the normal intergranular
force against the pipeline wall FN. The buoyancy effect associated with the presence
of suspended coarse particles (d > 0.074 mm) and fine particles (d < 0.074 mm) in the
lower layer is included to the equation for the normal solids stress at the pipeline wall,
σs. In the lower layer the suspended coarse particles, the fine particles smaller than
0.074 mm and the liquid form a mixture of density ρ2f determined as
4.30 CHAPTER 4

ρ2f(1-C2c) = ρsC1 + ρfines(1-C2c-C1), so that

ρ
ρ 2f = fines
(1 − C 2 ) + ρsC1 (4.46).
1 − C 2 + C1

in which ρfines is density of a mixture composed of the liquid and fine particles
smaller than 0.074 mm and C2 = C1 + C2c. The normal force FN is integrated from
Eq. 4.46 as

D2
FN = g(ρs − ρf )C vb (sin β − β cos β) (4.47).
2

4.3 SUMMARY: GENERAL TRENDS FOR FRICTIONAL HEAD


LOSS AND DEPOSITION-LIMIT VELOCITY UNDER
VARIOUS FLOW CONDITIONS

4.3.1 Pressure drop due to friction

The total frictional pressure drop in mixture flow is composed of the frictional
pressure drop in a carrying liquid and an additional frictional pressure drop, called the
solids effect, due to a presence of solid particles in a mixture. The solids effect
extends the frictional loss of a carrier alone if mixture flows at velocity round the
deposition-limit value. However, at higher velocities the water friction creates a major
part of a total frictional loss in a mixture flow. This is particularly valid for low
concentrated flows.

Pressure drop in flow of carrying water

Frictional loss in flow of carrying water (If) is particularly sensitive to flow velocity
and pipeline diameter. Higher throughputs (flow rates) in a pipeline of a certain
diameter are always paid in higher pressure losses due to friction. Less energy is
dissipated due to friction in flow through a larger pipeline than through a smaller
pipeline at the same velocity. Additionally, the roughness of a pipeline wall affects
the losses. The rougher wall the higher frictional losses.

Solids effect in a mixture flow

The solids effect (Im-If) on the total frictional losses is predominantly due to
mechanical friction between transported particles and a pipeline wall. Thus a
thickness of a granular bed is a major indicator of the solids effect for flow under
certain conditions.

The solids effect is sensitive to flow velocity, particularly for flows with a
considerable change in a degree of flow stratification within a operational range of
MODELING OF STRATIFIED MIXTURE FLOW 4.31

mixture velocities. A granular bed diminishes in flow with an increasing mixture


velocity and so diminishes the solids effect.

A coarse material tends to form a thicker bed than a fine material in a flow of certain
velocity. Thus the solids effect (Im-If) increases with the size of particles in mixture
flow. For very coarse particles, however, all solids are transported as a bed load and a
settling tendency of particles is not of importance for flow friction. The solids effect
of fully-stratified flow is virtually independent of a particle size. The solid effect is
further sensitive to the particle size distribution. A broad graded material might cause
lower friction losses than a narrow graded material of the same mass-median size
(d50) and the same concentration in mixture flow. This is so if a broad graded
material contains a considerable portion of fine particles that can be easily suspended
in carrying liquid. A granular bed is thinner when compared to flow with a narrow
graded material and thus the solids effect is smaller.

The solids effect (Im-If) grows with the concentration of solids in a pipeline. The
relationship can be estimated as linear, at least according to various experiments for
flows of Cvd between 0.05 and 0.25 approximately.

Scaling of mixture flow parameters obtained in one pipeline to pipelines of different


sizes (diameters) is a source of uncertainty. A lack of data of appropriate range and
quality from large pipelines prevents to evaluate the pipeline-size effect on flow
mechanisms. There is no agreement among the predictive models with regard to the
influence of a pipeline size on the solids effect (Im-If). The Wilson model and the
Führböter model do not predict any difference in the solids effect if flow of certain
parameters is scaled to pipelines of different diameters. The Durand model predicts an
increase in the solids effect with the pipeline diameter. Jufin and Lopatin predict the
opposite trend - the solids effect should be smaller in a larger pipeline.

4.3.2 Deposition-limit velocity

A physical description of a force balance between two layers gives an appropriate


basis for an explanation of the trends in the velocity of initial sliding of a granular bed
in a stratified mixture flow.

Deposition-limit velocity tends to increase with a particle size for flow of fine and
fine to medium sands. This can be explained by a fact that larger particles form a
thicker bed and higher velocity is required to put the thicker bed to motion. For
coarser materials (coarser than approximately 0.4-0.5 mm), however, the critical
velocity does not grow further. This is because a bed at its initial sliding has
approximately the same thickness for different particle sizes in a flow. Coarser
particles are not suspended and tend to increase a thickness of a bed but at the same
time a top of a bed is sheared off so that the effective thickness of a stationary bed
does not change. For very coarse particles the deposition-limit velocity even drops for
still coarser particles since the bed composed of these particles is subjected to
increasing driving force from the flow above the bed. This force acts at the top of a
bed as a viscous shear stress related to the roughness of the bed surface. The
4.32 CHAPTER 4

roughness is given by a size of particles occupying a bed surface. The coarser particle,
the rougher bed surface and the higher shear stress acting to the bed.

The value of deposition-limit velocity is higher in a larger pipe than in a smaller one
for flow under identical other conditions. In a smaller pipe a higher pressure drop is
built up over a pipe section than in a larger pipe (see the relationship between the
hydraulic gradient and the pipeline diameter in the Darcy-Weisbach equation). The
pressure drop is a source of a major driving force acting to put a granular bed to
motion in a pipe.
For concentrations usually handled in dredging pipelines (Cvd > 0.10) the value of
the deposition-limit velocity tends to drop in more concentrated flows. This is
basically because more concentrated suspension flowing above the bed exerts higher
driving force to the bed than the flow of low concentrated suspension or particle-free
carrying liquid. This has been detected in both the laboratory pipe and a field
dredging pipeline (Matousek, 1997). In a dredging practice the effect of Vdl reduction
in concentrated mixtures is not taken into account for a flow prediction. The
concentration of solids is difficult to control during a dredging operation and it may
vary within a rather wide range.

4.4 REFERENCES

Bagnold, R.A. (1956). The flow of cohesionless grains in liquids, Proceedings Roy.
Soc. (London), Ser. A, 249, 235-97.
van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
Churchill S.W. (1977). Friction-factor equation spans all fluid-flow regimes.
Chemical Engineering, 84(24), 91-2.
Clift, R., Wilson, K.C., Addie G.R. & Carstens, M.R. (1982). A
mechanistically-based method for scaling pipeline tests for settling slurries. Proc.
Hydrotransport 8, BHRA, Cranfield, UK, pp. 91-101.
Durand, R. & Condolios, E. (1952). Transport hydraulique et decantation des
materiaux solides. Deuxiemes Journees de l'Hydraulique, 27-55.
Durand, R. (1953). Basic relationships of the transportation of solids in
pipes - experimental research. Proc. Minnesota International Hydraulics Convention,
Minneapolis, Minnesota, pp. 89-103.
Führböter, A. (1961). Über die Förderung von Sand-Wasser-Gemischen in
Rohrleitungen. Mitteilungen des Franzius-Instituts, H. 19.
Gillies, R. G., Shook, C. A. & Wilson, K. C. (1991). An improved two layer model
for horizontal slurry pipeline flow. Can. J. of Chem. Engng., 69, 173-178.
Jufin, A.P. & Lopatin, N.A., (1966). O projekte TUiN na gidrotransport zernistych
materialov po stalnym truboprovodam. Gidrotechniceskoe Strojitelstvo, 9, 49-52.
Kazanskij, I. (1978). Scale-up effects in hydraulic transport theory and practice.
Proc. Hydrotransport 5, BHRA Fluid Engineering, Cranfield, UK, pp. B3-47-B3-74.
Matousek, V. (1997). Flow Mechanism of Sand-Water Mixtures in Pipelines. Delft
University Press.
Newitt, D.M., Richardson, J.F., Abbott, M. & Turtle, R.B. (1955). Hydraulic
conveying of solids in horizontal pipes. Trans. Inst. Chem. Eng., 33, 93-113.
MODELING OF STRATIFIED MIXTURE FLOW 4.33

Turner, T. (1996). Fundamentals of Hydraulic Dredging. ASCE New York.


Wilson, K.C. (1976). A unified physically-based analysis of solid-liquid pipeline
flow, Proceedings Hydrotransport 4, BHRA, Cranfield, UK, pp. 1-16.
Wilson, K.C. (1979). Deposition-limit nomograms for particles of various densities in
pipeline flow, Proceedings Hydrotransport 6, BHRA, Cranfield, UK, pp. 1-12.
Wilson, K.C. (1986). Effect of solids concentration on deposit velocity, Journal of
Pipelines, 5, 251-7.
Wilson, K.C. (1992). Influence of particle properties on solids effect, Proc.10. Int.
Kol. Massenguttransport durch Rohrleitungen, Univ. GH Paderborn, Meschede,
Germany, A.
Wilson, K.C. (1996). Heterogeneous slurries. Chapter 4.0 of the Proceedings
Hydrotransport Seminar. Back to Basics - With the Experts. BHRG, Cranfield, UK.
Wilson, K.C., Addie G.R., Sellgren, A. & Clift, R. (1997). Slurry Transport Using
Centrifugal Pumps. Blackie Academic & Professional.

4.5 RECOMMENDED LITERATURE

Gibert, R. (1960). Transport hydraulique et refoulement des mixtures en conduites.


Annales des Ponts et Chausees, 130(3), 307-74, 130(4), 437-94 (Dutch translation:
Het persen van mengsels door leidingen. TU Delft).
Kazanskij, I. (1978). Scale-up effects in hydraulic transport theory and practice.
Proc. Hydrotransport 5, BHRA Fluid Engineering, Cranfield, UK, pp. B3-47-B3-74.
Matousek, V. (1997). Flow Mechanism of Sand-Water Mixtures in Pipelines. Delft
University Press.
Shook, C.A. & Roco, M.C. (1991). Slurry Flow. Principles and Practice.
Butterworth-Heinemann.
Wilson, K.C., Addie G.R., Sellgren, A. & Clift, R. (1997). Slurry Transport Using
Centrifugal Pumps. Blackie Academic & Professional.
4.34 CHAPTER 4

CASE STUDY 4

Mixture flow in a horizontal pipeline

An aqueous mixture of fine sand or medium gravel (see previous Case studies) is
transported from a dredge to a deposit site through a dredging pipeline that is 1.5
kilometer long and has an internal diameter of 900 millimeter.

Propose a suitable transport velocity for mixture in a pipeline and determine the
energy lost due to friction, the specific energy consumption and the production for
mixture transport at the chosen velocity. The absolute roughness of a pipeline wall is
20 microns.

Remark: Consider a horizontal pipeline and no boosters. Consider 1.1Vdl


(deposition-limit velocity) a suitable transport velocity of mixture in the pipeline. For
a simplification consider a narrow graded soil characterised by the median diameter
only.

Inputs:

d50 = 0.120 mm of d50 = 6.0 mm


ρs = 2650 kg/m3
ρf = 1000 kg/m3
νf = 0.000001 m2/s
Cvd = 0.27
L = 1500 m
D = 900 mm
k = 0.00002 m

Solution:

a. The deposition-limit velocity

Fine sand (d = 0.120 mm)

Durand: Vdl = FL 2g(Ss − 1)D


FL = 1.05 (see a nomograph in Fig. 4.3)
Vdl = 1.05 2x 9.81(2.65 − 1)0.9 = 5.67 m/s.

Wilson: Vsm = 1.72 m/s (zie Eq. 4.21 for µs = 0.4)


Crm = 0.66 (see a nomograph in Fig. 4.9)
Cr = Cvd/Cvb = 0.27/0.60 = 0.45
Vdl = 0.86* Vsm = 1.48 m/s.

MTI: Vdl = 3.23 m/s (see a nomograph in Fig. 4.6 or Eq. 4.19).
MODELING OF STRATIFIED MIXTURE FLOW 4.35

Medium gravel (d = 6.0 mm)

Durand: Vdl = FL 2g(Ss − 1)D


FL = 1.35 (see a nomograph in Fig. 4.3)
Vdl = 1.35 2x 9.81(2.65 − 1)0.9 = 7.29 m/s.

Wilson: Vsm = 5.21 m/s (zie a nomograph in Fig. 4.8 or Eq. 4.21 for µs = 0.4)
(Vsm,max = 5.83 m/s according to Eq. 4.22)
Crm = 0.05 (see a nomograph in Fig. 4.9)
Cr = Cvd/Cvb = 0.27/0.60 = 0.45
Vdl = 0.33* Vsm = 1.72 m/s.

MTI: Vdl = 7.03 m/s (see a nomograph in Fig. 4.6 or Eq. 4.19).

The MTI results are taken as predicted values of Vdl.


The suitable transport velocity for sand-water mixture: Vm = 1.1Vdl = 3.60 m/s.
The suitable transport velocity for gravel-water mixture: Vm = 1.1Vdl = 7.70 m/s.

b. Energy loss due to friction

Fine sand (d = 0.120 mm)

Water flow:
Re = 3.6*0.9/0.000001 = 3.24 x 106
k/D = 0.00002/0.9 = 2.2 x 10-5 (D/k = 45000)
λf = 0.0107 (see Moody diagram, Fig. 1.6)
Friction head loss from the Darcy-Weisbach equation (Eq. 1.20)
λ V 2 0.0107 3.62
If = f m = = 0.00785 [-].
D 2g 0.900 19.62

Mixture flow:

Durand model: (Eq. 4.4, for the vt value see Case study I)
−1.5 −1.5
I m − If  V 2 gd   3.62 9.81x 0.00012 
= 180 m  = 180 = 14.68
I f C vd  gD v t   9.81x 0.9 0.00947 
   
Im = 0.00785 + 14.68 x 0.27 x 0.00785 = 0.0390 [-].

Wilson model for heterogeneous flow: (Eqs. 4.16 and Eq. 4.17)
0.45
0.35  Ss − 1 
V50 ≈ 3.93(d 50 )   = 3.93(0.12) 0.351 = 1.87 m/s.
 1.65 
−M −1.7
I m − If V   3.60 
= 0.22 m  = 0.22  = 0.07225
C vd (Ss − 1)  V50   1.87 
Im = 0.00785 + 0.07225 x 0.27 x 1.65 = 0.0400 [-].
4.36 CHAPTER 4

Energy head, H [meter water column, mwc], lost over a pipeline length L = 1500
metre:
H = Im x L = 0.0400 x 1500 = 60.0 mwc.

Medium gravel (d = 6.0 mm)

Water flow:
Re = 7.7*0.9/0.000001 = 6.93 x 106
k/D = 0.00002/0.9 = 2.2 x 10-5 (D/k = 45000)
λf = 0.010 (see Moody diagram, Fig. 1.6)
0.010 7.7 2
If = = 0.03358 [-] (see Eq. 1.20, i.e. Darcy-Weisbach eq.)
0.900 19.62

Mixture flow:

Durand model: (Eq. 4.4, for the vt value see Case study I)
−1.5 −1.5
I m − If  V 2 gd   7.7 2 9.81x 0.006 
= 180 m  = 180  = 12.40
I f C vd  gD v t   9.81x 0.9 0.27374 
   
Im = 0.03358 + 12.40 x 0.27 x 0.03358 = 0.1460 [-].

Wilson model for fully-stratified flow: (Eq. 4.20)


Vsm = 5.21 m/s (zie a nomograph in Fig. 4.8 or Eq. 4.21 for µs = 0.4).
− 0.25 − 0.25
I m − If  Vm   7.70 
=  =  = 0.7810
C vd (Ss − 1)  0.55Vsm   0.55x 5.21 
Im = 0.03358 + 0.7810 x 0.27 x 1.65 = 0.3815 [-].
The ratio d/D = 6/900 = 0.0067 < 0.018, i.e. the flow is not considered fully
stratified.

Wilson model for heterogeneous flow: (Eq. 4.16 and Eq. 4.17).
0.45
 S − 1
V50 ≈ 3.93(d 50 ) 0.35  s  = 3.93(6.0) 0.351 = 7.36 m/s
 1.65 
−M −1.7
I m − If  Vm   7.70 
= 0.22  = 0.22  = 0.20374
C vd (Ss − 1)  V50   7.36 
Im = 0.03358 + 0.20374 x 0.27 x 1.65 = 0.1243 [-].

Energy head, H [meter water column, mwc], lost over a pipeline length L = 1500
metre:
H = Im x L = 0.1243 x 1500 = 186.5 mwc.
MODELING OF STRATIFIED MIXTURE FLOW 4.37

c. Specific energy consumption (Eq. 3.6)

Fine sand (d = 0.120 mm)


I 0.0400
SEC = 2.7 m = 2.7 = 0.151 [kWh/(tonne.km)].
Ss .C vd 2.65x 0.27

Medium gravel (d = 6.0 mm)


I 0.1243
SEC = 2.7 m = 2.7 = 0.469 [kWh/(tonne.km)] .
Ss .C vd 2.65x 0.27

d. Production

Fine sand (d = 0.120 mm)

Production of solids: (Eq. 3.3)


π π
Qs = D 2 Vm C vd 3600 = 0.9 23.6x 0.27x 3600 = 2226.1 [m3/hour].
4 4

Production of in situ soil: (for porosity n = 0.4) (Eq. 3.4)


π Q
Qsi = D 2 Vm C vdsi 3600 = s = 3710.2 [m3/hour].
4 1− n

Medium gravel (d = 6.0 mm)

Production of solids: (Eq. 3.3)


π π
Qs = D 2 Vm C vd 3600 = 0.9 27.7x 0.27x 3600 = 4761.4 [m3/hour].
4 4

Production of in situ soil: (for porosity n = 0.4) (Eq. 3.4)


π Q
Qsi = D 2 Vm C vdsi 3600 = s = 7935.6 [m3/hour].
4 1− n

Summary of the results:

Fine sand (d = 0.12 mm):


suitable transport velocity: Vm = 3.60 m/s
frictional head loss: Im = 0.0400 [-]
head lost over the pipeline 1500 m long: H = 60.0 mwc
specific energy consumption: SEC = 0.151 kWh/(tonne.km)
production of in situ soil: Qsi = 3710.2 m3/hour
Medium gravel (d = 6.00 mm):
suitable transport velocity: Vm = 7.70 m/s
frictional head loss: Im = 0.1243 [-]
head lost over the pipeline 1500 m long: H = 186.5 mwc
specific energy consumption: SEC = 0.469 kWh/(tonne.km)
production of in situ soil: Qsi = 7935.6 m3/hour
4.38 CHAPTER 4
5.
MODELING OF NON-STRATIFIED
MIXTURE FLOWS
(Pseudo-homogeneous flows)

Uniform (or almost uniform) distribution of transported solids across a pipeline cross
section is characteristic of pseudo-homogeneous mixture flow. This flow occurs if the
settling tendency of particles transported in a flowing liquid is weak in comparison
with the tendency of a carrying liquid to keep particles suspended. Very fine particles
are practically non-settling and the pseudo-homogeneous character of the mixture
flow is maintained at all operational velocities in a pipeline. Coarser particles (as fine
and medium sand) may form a fully-suspended mixture if the intensity of turbulence
in a flow of a carrying liquid does not allow solid particles to settle. This is the case at
high operational velocities in a pipeline.

The pseudo-homogeneous flow composed of very fine particles behaves differently


than the pseudo-homogeneous flow composed of sand-size particles. This is due to
rather different mechanisms of internal friction in these flows. The fluid-like (not
mechanic) friction in the flowing matter is described the law of viscosity.
Pseudo-homogeneous mixtures that obey Newton’s law of viscosity (see Chapter 1)
are called Newtonian mixtures. Pseudo-homogeneous mixtures obeying a more
complex relationship between shear stress and strain rate than is that given by
Newton’s law of viscosity are called non-Newtonian mixtures.

Pseudo-homogeneous mixture flows experience no (or at least very weak)


accumulation of solid particles near the bottom of a pipeline, thus the deposition-limit
velocity is an irrelevant parameter to predict. A slip between phases plays also no
role. Thus the attention is focused to predicting frictional head losses in pipelines.

5.1 NEWTONIAN FLOW OF AQUEOUS MIXTURE OF SAND OR


GRAVEL

5.1.1 General model

Generally, the frictional head loss in the fully-suspended (pseudo-homogeneous) flow


is predicted by Clift et al. (1982) as

I m − If I −I
= m f = A ′I f (5.1)
C vd (Ss − 1) S m − 1
5.2 CHAPTER 5

Im hydraulic gradient for pseudo-homogeneous


mixture flow [-]
If hydraulic gradient for liquid (water) flow [-]
Cvd delivered volumetric solids concentration [-]
Ss relative density of solids [-]
Sm relative density of mixture [-]
A’ empirical coefficient [-].

An increase in the frictional head loss due to the presence of solid particles in a
carrying liquid forming a Newtonian fully-suspended mixture is attributed to
increased carrier friction at the pipeline wall. A measure of the slurry density effect on
friction process is given by an empirical coefficient A'.

5.1.2 Equivalent-liquid model

The Eq. 5.1 gives the "equivalent liquid" model

Im = SmIf for A' = 1 (5.2).

According to this model the pseudo-homogeneous slurry flow behaves as a flow of a


single-phase liquid having the density of the slurry. The "equivalent liquid" has the
density of the mixture but other properties (as viscosity) remain the same as in the
liquid (water) alone. This model suggests that all suspended particles contribute to an
increase in the mixture density. The increase in a mixture density is responsible for a
proportional increase in the shear stress resisting the flow at the pipeline wall. The
above equation is obtained in the same way as the Darcy-Weisbach equation (see
Chapter 1) with the only one exception: the density of mixture is considered instead of
dP τo 8τo dP λ f ρm Vm2
the liquid density. Thus − =4 and λ f = provide − = .
dx D ρm Vm2 dx D 2
dP ρm λ f Vm2
Rearrangements give I m = − = = Sm If .
dxρf g ρf D 2g

This model may be successful to predict flows of relatively fine particles (fine sand,
coarse silt), particularly if solids concentration is relatively low so that the viscosity is
not affected.

5.1.3 Liquid model

If A' = 0 in the Eq. 5.1, it is assumed that solids present in a flowing liquid do not
affect a flow friction at all,

Im = If for A' = 0 (5.3).

Such a behavior in fast-flowing horizontal fully-suspended sand mixtures was


reported by Carstens & Addie (1981). This behavior can be explained by an
assumption that relatively coarse particles suspended in fast-flowing mixture are
MODELING OF NON-STRATIFIED MIXTURE FLOW 5.3

repelled from the pipeline wall due to large velocity gradient near the pipeline wall (a
possible effect of lift forces discussed in Chapter 1). Since solid particles are not
present in the region nearest the pipeline wall they also do not affect the wall shear
stress that is decisive for the friction process.

5.1.4 Experimental observations

Pipeline tests in the Laboratory of Dredging Technology have revealed that the liquid
model is not applicable to fast-flowing fully-suspended mixtures. The
equivalent-liquid model tends to overestimate slightly the frictional losses. The
parameter A’ of the general model seems to be dependent on flow conditions.

5.2 NON-NEWTONIAN FLOW OF AQUEOUS MIXTURE OF


SILT OR CLAY

Very fine particles (of silt size and finer, d < 40 µm approximately) are practically
non-settling in a flow of carrying liquid. They interfere with the carrying liquid to
increase its density and viscosity. In a mixture flow containing these fine particles the
viscosity of the pseudo-homogeneous mixture grows with the increasing fraction of
solids in the mixture.

The suspension does not obey Newton’s law of viscosity and its constitutive
rheological equation (the equation relating the shear stress, τ, with the shear rate,
dvx/dy) has to be determined experimentally to give a basis for friction-loss predictive
models.

Modeling of non-Newtonian mixtures is even more complex than the modeling of


Newtonian mixtures in pipelines. A constitutive rheological equation of a
non-Newtonian mixture is rather sensitive to too many factors. Actually, each
particular mixture obeys its own law of viscosity. Therefore it is necessary to know a
rheogram [a relationship between shear stress, τ, and strain rate, dvx/dy] for each
particular mixture handled in a pipeline. A tube viscometer is a preferable instrument
to determine a rheogram of a mixture. Conditions within the tube viscometer are
geometrically similar to those in prototype pipes, assuring the similarity in the stress
distributions. Data from a tube viscometer can be successfully used to scale up the
frictional head loss to larger pipes or to determine a mixture rheological model.

A general procedure for a determination of the frictional head losses in a pipeline


flow of a non-Newtonian mixture is composed of the following steps:

1. the rheological parameters of a mixture


2. the mixture flow regime (laminar or turbulent)
3. the losses using a scale-up method or an appropriate friction model.
5.4 CHAPTER 5

5.2.1 Rheological parameters of a mixture

The rheology of a mixture is determined from laminar-flow data obtained from


measurements in either a tube viscometer or a rotational viscometer. The rheological
constants are determined from measured values of parameters Vm, ∆P/L, D etc.
according to a method discussed in Intermezzo II.

5.2.2 Mixture flow regime (laminar or turbulent)

An information whether pipe flow is laminar or turbulent is important because it


determines a method for the friction-loss prediction. A laminar regime in
non-Newtonian mixtures holds to higher velocities than for Newtonian mixtures in a
pipeline of the same diameter. The laminar regime may occur in a dredging pipeline if
highly viscous mixtures are transported.

The most accurate method of a determination of a transitional velocity, VT, between


the laminar and the turbulent regime is to find experimentally (in a laboratory pipe) an
intercept between Im-Vm curves for laminar and turbulent regimes of the mixture
flow. Then the transition can be scaled up with the resistance curves to pipes of larger
sizes.

Theoretical models for a regime transition are also available. These are often the
by-products of the flow models for laminar and turbulent flows of non-Newtonians.
Thomas (1963) proposed for Bingham plastic flow the following equation that is often
used in practice

2100ηB  τyD 
VT = 1 +  (5.5).
ρm D  6ηBVT 

The simpler equation for the transition velocity in a Bingham plastic flow is obtained
if the Bingham Reynolds number

ρ m Vm D
Re B = (5.6)
 τyD 
ηB 1 + 

 6ηBVm 

VT value of Vm at the transition between laminar and


turbulent regime of non-Newtonian mixture flow [m/s]
Vm mean mixture velocity in a pipe [m/s]
D pipe diameter [m]
ρm density of mixture [kg/m3]
ηB tangential viscosity of Bingham plastic mixture [Pa.s]
τy yield stress of Bingham plastic mixture [Pa]
ReB Reynolds number of flow of Bingham plastic mixture [-]
MODELING OF NON-STRATIFIED MIXTURE FLOW 5.5

is taken as equal to 2100. This threshold value of Reynolds number is identical with
τyD
that for a Newtonian flow. Since >> 1 the equation ReB = 2100 gives for
6ηB Vm
flow of Bingham plastic mixture

τy
VT ≈ 19 (5.7)
ρm

5.2.3 Friction losses using a scale-up method or an appropriate friction model

Friction-loss predictions based on the flow modeling are generally less accurate than
those based on the scaling up of tube viscometer data.

5.2.3.1 Scale-up methods

The Im-Vm results from tube viscometers can be scaled up to prototype pipes without
an intermediary of rheological model. The principle of the scaling-up technique is that
in non-Newtonian flows the wall shear stress is unchanged in pipes of different sizes
D. The wall shear stress fully determines the stress distribution within the pipe.

Scale-up techniques are different in laminar and turbulent flow. Scaling up between
two different pipeline sizes [e.g. between a tube viscometer (index 1) and a prototype
pipeline (index 2)] is carried out as follows.

Laminar flow:
the Rabinowitsch-Mooney transformation applies:

D.∆P 8Vm
τ0 = = idem , = idem ;
4L D

8Vm
this says that if a τo versus relationship for a laminar flow of a certain mixture
D
is determined (experimentally) in one pipe it is valid also for pipes of all different
sizes.

Thus
D1
I m2 = I m1 (5.8)
D2

D2
Vm2 = Vm1 (5.9).
D1
5.6 CHAPTER 5

Turbulent flow:

D.∆P 8Vm
τ0 = = idem but ≠ idem
4L D

because the near-wall velocity gradient is not described by 8Vm/D (see Chapter 1, p.
1.5); instead the friction-law is sought relating friction coefficient λf with mean
velocity Vm.

Thus the turbulent-flow data Im, Vm from pipeline (1) can be scaled to pipeline (2)
using
D
I m2 = I m1 1 (5.10)
D2

 λ  D 
Vm2 = Vm1 1 + 2.5 f ln 2  (5.11)
 8  D1 

Im hydraulic gradient for pseudo-homogeneous


mixture flow [-]
Vm mean mixture velocity in a pipe [m/s]
D pipe diameter [m]
λf Darcy-Weisbach friction coefficient for fluid flow [-].

In Eq. 5.11 the final term within the brackets determines an effect of the equivalent
turbulent-flow viscosity on wall shear stress. This term is usually not greatly different
from zero.

Figure 5.1. Scaling to a larger pipeline.


MODELING OF NON-STRATIFIED MIXTURE FLOW 5.7

5.2.3.2 Flow-friction models

The scale-up technique is certainly a preferable predictive method. However,


sometimes the tube test data are not available because a test instrument or a mixture
sample are not available. Test data for turbulent flow regime might be unavailable
even if tube viscometer tests are carried out. This might be, for instance, due to a
small diameter of a viscometer tube that causes that the turbulent flow regime is not
reached even at the highest velocities in the tube. If data, and thus a rheogram, are not
available the rheological parameters employed in theoretical rheological models
(constitutive equations) have to be used to derive a I-V relationship for mixture flow.

Laminar flow:

For a laminar flow a chosen constitutive equation is integrated over a pipe cross
section and hence velocity distribution obtained. This gives a relation between
pressure gradient and mean velocity (the same procedure as described in Chapter 1 for
Newtonian liquid flow) in a homogeneous flow of mixture.
Integrating of the yield pseudo-plastic rheological model
n
 dv 
τ = τ y + K x 
 dy 
πD 2
over a pipe cross section A = gives
4
8Vm
=
4
( )
τo − τ y n 
( ) +
( )
1+ n  τ o − τ y 2 2 τ y τ o − τ y
+
τ 2y 
 (5.12)
D 1  1 + 3n 1 + 2n 1+ n
 
K n τ 3o
D.∆P
where the wall shear stress τ o = .
4L

For a Bingham mixture integrating of a constitutive equation provides


 4τ 4 
8Vm τ y τy 
= o 1 − + (5.13).
D ηB  3τ o 3τ 4 
 o
Eqs. 5.12 and 5.13 give a relationship between a frictional head loss and mean
mixture velocity in a pipeline as a function of rheological parameters of a mixture.
This relationship can be rewritten to the standard friction-loss equation

∆P 2
λ nN Vm
Im = = (5.14)
Lρ f g D 2g

in which, for a laminar flow,

64
λ nN = (5.15)
Re nN
5.8 CHAPTER 5

if the equation for Reynolds number RenN for a non-Newtonian flow gets a modified
form given by a rheological type of a flowing mixture. For the Bingham plastic
mixture (n = 1, K = ηB) combining of Eq. 5.13 and Eq. 5.14 provides

−1
 4 
64ηB  4τ y τ y  64
λB = 1− + = (5.16)
DVmρ m  3τ o 3τ 4  Re B
 o

and thus the modified Reynolds number, by neglecting the fourth-power term in the
above equation

ρ m Vm D
Re B = (5.6).
 τyD 
ηB 1 + 

 6 η B Vm 

Im hydraulic gradient for mixture flow [-]


Vm mean mixture velocity in a pipe [m/s]
D pipe diameter [m]
λnN Darcy-Weisbach friction coefficient for
non-Newtonian flow [-]
g gravitational acceleration [m/s2]
ReB Reynolds number of flow of Bingham plastic mixture [-]
ρm density of mixture [kg/m3]
ηB tangential viscosity of Bingham plastic mixture [Pa.s]
τy yield stress of Bingham plastic mixture [Pa]

A solution of the integral equation for a Bingham plastic flow (Eq. 5.13) can be
accomplished using the Hedström nomograph. The nomograph (Fig. 5.2) gives the
friction coefficient λB as a function of two dimensionless groups:

Hedström number, He,

τ y D 2ρ m
He = (5.17)
η2B

and Reynolds number, Reb,

ρ V D
Re b = m m (5.18).
ηB

Steep lines for different He values give the friction coefficient for a laminar flow
regime. The less steep line valid for all He values gives the friction coefficient in a
MODELING OF NON-STRATIFIED MIXTURE FLOW 5.9

turbulent regime of mixture flow. It is suggested that there is no effect of the yield
stress on pipeline friction if the flow is turbulent.

Figure 5.2. Friction coefficient λB as a function of He number (Eq. 5.17)


and Reb number (Eq. 5.18).

Turbulent flow:

In a turbulent flow the rheological models are again the basis for friction models.
However, an integration of a rheological model, and thus a direct determination of
velocity profile, is not possible in turbulent flow (see Chapter 1). A friction law is
required that relates the friction coefficient, λnN, with the flow Reynolds number, Re,
and the pipe-wall roughness factor.

The Slatter model (Slatter, 1995) was tested by data from a number of non-Newtonian
mixtures (kaolin etc.) in various viscometric tubes. It suggests the following equations to
describe a friction law for a yield pseudo-plastic mixture in different turbulent flow
regions delimited by a value of the roughness Reynolds number

8ρ m V 2
Re r = * (5.19),
n
 8V 
τ y + K  * 
 d85 
5.10 CHAPTER 5

- smooth wall turbulent flow (Rer ≤ 3.32)

8  D 
= 2.5 ln  + 2.5 ln Re r + 1.75 (5.20)
λ nN  2d 85 

- fully developed rough wall turbulent flow (Rer > 3.32)

8  D 
= 2.5 ln  + 4.75 (5.21)
λ nN  2d 85 

Rer roughness Reynolds number for non-Newtonian flow [-]


V* shear velocity, V* = Vm(λnN/8)0.5 [m/s]
d85 characteristic particle size [m]

The frictional head loss is again given by Eq. 5.14

∆P λ V2
Im = = nN m .
Lρ f g D 2g

5.2.4 Prediction of frictional losses during transportation of silt mixture in a


dredging pipeline

The figures 5.3-5.5 show measured rheological characteristics of the aqueous mixture of
silt dredged from Caland Kanaal in the Europort entrance (taken from v.d. Berg, 1998).
The mixture behaves like Bingham plastic liquid.

Figure 5.3. Rheogram of the “Caland”silt mixture measured for different mixture
densities (measurements: rotoviscometer).
MODELING OF NON-STRATIFIED MIXTURE FLOW 5.11

Figure 5.4. Figure 5.5.

Figure 5.4. Relationship between viscosity and density of the silt mixture.

Figure 5.5. Relationship between yield stress and density of the silt mixture.

Figure 5.6. Prediction of pipeline resistance of the silt mixture using a friction model
(lines for various mixture densities in a 700 mm pipeline)
(from v.d. Berg, 1998).
5.12 CHAPTER 5

Fig. 5.6 shows the resistance curves for the Caland silt mixture flow in a 700 mm
dredging pipeline predicted by the Hedström method. The intercepts between laminar
curves and turbulent curves for a certain chosen mixture density determines the transition
velocity VT. A laminar regime holds to the mean mixture velocity 1.7 m/s if silt-water
mixture has density of 1300 kg/m3.

5.3 REFERENCES

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
Carstens, M.R. & Addie, G.R. (1981). A sand-water slurry experiment, Journal of
the Hydraulic Division, ASCE, 107(HY4), 501-7.
Clift, R., Wilson, K.C., Addie G.R. & Carstens, M.R. (1982). A
mechanistically-based method for scaling pipeline tests for settling slurries. Proc.
Hydrotransport 8, BHRA, Cranfield, UK, pp. 91-101.
Slatter, P. (1995). The turbulent flow of non-Newtonian slurries in pipes, Proc. 8th
Int. Conf. on Transport and Sedimentation of Solid Particles, CTU Prague.
Thomas, D.G. (1963). Ind. Eng. Chem., 55, p. 27.

5.4 RECOMMENDED LITERATURE

Govier, G.W. & Aziz, K. (1972). The Flow of Complex Mixtures in Pipes. Van
Nostrand Reinhold Company.
Wilson, K.C., Addie G.R., Sellgren, A. & Clift, R. (1997). Slurry Transport Using
Centrifugal Pumps. Blackie Academic & Professional.
MODELING OF NON-STRATIFIED MIXTURE FLOW 5.13

CASE STUDY 5

Frictional head loss in flow of non-Newtonian mixture through a horizontal


pipeline

The silt dredged from Calandkanaal of the Rotterdam harbour is transported


hydraulically at the mixture density 1250 kg/m3 through a 500-metre long horizontal
pipeline of the diameter 700 mm. The rheometrical test of the mixture sample in a
viscometer has shown that the silt mixture behaves as a Bingham liquid with the yield
stress 33 Pa and the plastic viscosity 36 mPa.s.

Determine the pressure drop due to friction over the entire pipeline length for two
mixture flow rates: 1.0 m3/s and 2.0 m3/s.

Inputs:

ρm = 1250 kg/m3
τy = 33 Pa
ηB = 0.036 Pa.s
L = 500 m
D = 700 mm
Qm = 1.0 and 2.0 m3/s

Solution:

a. Mean mixture velocity and laminar-turbulent threshold

Vm = 4Qm/(πD2), i.e. 2.60 m/s for Qm=1.0 m3/s and 5.20 m/s for Qm=2.0 m3/s.
VT = 3.09 m/s (Eq. 5.7).
Thus the flow is laminar for Qm=1.0 m3/s and turbulent for Qm=2.0 m3/s.

b. Frictional pressure drop in the laminar flow

ReB = 1500 (Eq. 5.6) for Vm = 2.60 m/s


λB = 64/ ReB = 0.043
Im = 0.021 (Eq. 5.14), thus ∆P = 0.021 x 500 x 9810 = 103 814 Pa.

The total pressure drop due to friction is 104 kPa (approximately 1 bar) at the flow
rate 1.0 m3/s of the silt mixture through a 700-mm pipeline that is 500 meter long.

c. Frictional pressure drop in the turbulent flow

Reb = 126 389 ≈ 1.3 x 105 (Eq. 5.18) for Vm = 5.20 m/s
λB ≈ 0.021 (Fig. 5.2) for Reb ≈ 1.3 x 105 (He number value is not important)
Im = 0.041 (Eq. 5.14), thus ∆P = 0.041 x 500 x 9810 = 202 800 Pa.

The total pressure drop due to friction is 203 kPa (approximately 2 bar) at the flow
rate 2.0 m3/s of the silt mixture through a 700-mm pipeline that is 500 meter long.
5.14 CHAPTER 5
6.
SPECIAL FLOW CONDITIONS IN
DREDGING PIPELINES

6.1 INCLINED FLOWS

Solids, dredged from a bottom of a waterway or in a borrowing pit, are transported


hydraulically to the board of a dredge by using an inclined pipeline. A pipeline slope
varies from rather flat to almost vertical according to the depth from which the
material is dredged. The largest modern THSD's are capable of dredging from water
deeper than 100 meter. This gives an imagination of how long inclined pipelines
might be. Mixture flows, particularly that exerting a certain degree of stratification,
are very sensitive to changes in a pipeline inclination. In this chapter the effects are
discussed of the pipeline inclination on the hydraulic gradient and the deposition-limit
velocity in mixture flow. Further the effects of inclination on a flow pattern are
demonstrated and a physical description is given of these inclination effects.

6.1.1 Static head and friction head

Consider a horizontal pipeline of a certain length occupied by a flowing mixture. A


total pressure drop in mixture flow over the pipeline section is equal to the pressure
drop due to internal friction in flowing mixture if there are no additional minor losses
from the local sources of energy dissipation as are pipe joints, bends etc. The
frictional head loss in the mixture flow is due to both the frictional losses in a carrying
liquid and the additional losses due to a presence of solids in a carrying liquid as
described in details in Chapters 4 and 5.

If the pipeline section is inclined the total pressure drop over the section changes
considerably. The differential pressure transmitter measuring the pressure difference
over the pipeline section senses a considerable increase (in case of the ascending
pipeline section) or a considerable decrease (in case of the descending pipeline
section) of the total pressure differential. This is due to a hydrostatic pressure
differential emerging as a result of a change of a geodetic position of the end of the
pipeline section regarding to its begin. A change of a pipeline elevation gives to arise
to the static pressure differential caused by a pressure exerted by a mixture column of
a height given a vertical distance between the begin and the end of a pipeline section.
The hydrostatic column increases pressure at the beginning of a pipeline section in the
case of an ascending pipeline or at the end of a pipeline section in the case of a
descending pipeline. Considering a pipeline section of a length L inclined to an angle
ω from a horizontal position, the height of a vertical column is Hpipe = L.sinω.

6.1
6.2 CHAPTER 6

Thus the total pressure gradient (P1-P2)/L = ∆P/L over a pipeline section of the
length L (see Fig. 6.1) is composed of
- the static pressure gradient (ρgHpipe)/L, giving the potentially reversible effect of
elevation change on the total pressure gradient in a mixture flow of the density ρ
gaining the height Hpipe and
- the pressure gradient due to friction (P1-P2)/L - (ρgHpipe)/L that is the
irrecoverable energy loss due to friction in inclined mixture flow over the pipe
length L.

P1 −ω

concentration
profile L H pipe

two layer
pattern
ρf g H pipe

P = P1 - P2
total pressure differential

P2
manometer
hose filled with fluid (water)

P + Hpipeρf g
manometric pressure differential

Figure 6.1. Schematic length-section of inclined pipe.

The pressure differential between two pipeline cross sections 1 and 2 separated from
each other by the pipeline length L is measured usually as a manometric pressure
differential P1 + Hpipe.ρf.g - P2, i.e. ∆P + Hpipe.ρf.g (Fig. 6.1). The total pressure
differential ∆P is obtained by eliminating the static pressure differential due to the
water column in the hoses of the differential pressure transmitter (manometer). The
pressure differential due to friction is obtained from the total pressure differential ∆P
by subtracting (or adding) the static pressure differential Hpipe.ρ.g due to a mixture
column in the measuring pipe section from a pipe elevation.

The static pressure gradient Hpipe.ρ.g/L is produced by a mixture column of the


height Hpipe in the pipeline section L. The density of the column, ρ, is determined
from the concentration of solid particles in the section L which contribute to the
weight of the slurry column. Empirical models for the prediction of friction losses in
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.3

inclined pipelines assume that the slurry of the column has the density ρm = ρf +
Cvd(ρs - ρf).

6.1.2 Deposition-limit velocity

Deposition-limit velocity tends to increase with a pipe slope and reaches its maximum
at an angle of about 25 - 35 deg in an ascending pipeline. A further increase in an
angle of an ascending pipe inclination causes a gradual decrease of the Vdl value until
zero at the limit inclination angle 90 deg. At angles higher than approximately 45 deg
a bed is gradually disintegrated owing to a continuously diminishing cross-pipe
component of solid particle weight - the force component usually responsible for the
formation of a bed. In a descending pipeline the deposition-limit velocity gradually
decreases when the pipeline is inclined gradually from 0 deg to -90 deg.

6.2 EMPIRICAL MODELING OF INCLINED FLOWS

Typical models for inclined flows are extensions of models for flows at limit
inclinations: in horizontal and vertical pipes.

6.2.1 Vertical-flow model

Uniform distribution of solids across a pipeline cross section is characteristic of


mixture flow in a vertical pipeline. The homogeneous character of mixture makes
prediction of vertical flows easier than prediction of horizontal and inclined flows.
Coulson et al. (1996) summarized the simplest conclusions for the prediction of
frictional head loss in a vertical mixture pipeline as follows:

- for non-settling suspensions the standard equation for a single phase fluid is used
with the physical properties of the suspension in place of those of the liquid (i.e.
transported particles do not affect the friction process in coarse-particle mixture flow
in a vertical pipeline)

- for a suspension of coarse particles the value calculated for the carrying fluid alone,
flowing at the mixture velocity, is used.

It should be stressed, however, that the above-formulated rules are considered only for
Newtonian mixtures. The non-Newtonian mixtures exert in vertical pipelines frictional
head losses equal to that in horizontal pipelines.
6.4 CHAPTER 6

6.2.2 Inclined-flow model by Worster & Denny

Worster & Denny (1955) suggested a simple equation for the energy loss in settling
slurries flowing in inclined pipelines

Imhω = If + (Im - If )cosω + Cvd (Ss -1)sinω (6.1)

Imhω manometric gradient in mixture flow in inclined pipe [-]


Im hydraulic gradient in the same mixture flow in horizontal pipe [-]
If hydraulic gradient in liquid flow [-]
ω pipe inclination angle [deg]
Cvd delivered volumetric solids concentration [-]
Ss relative density of solids [-]

The angle ω is considered to have positive values in an ascending pipeline and the
negative values in a descending pipeline.
The head loss due to the potential energy change registered by a differential pressure
transmitter is represented by the last term in Eq. 6.1. This is the hydrostatic effect on
the pressure differential measured over a section of an inclined mixture pipe.

Im = If + (Im − If )

Figure 6.2. Pressure drops in inclined pipelines, after Worster & Denny (1955).
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.5

The ratio between solids effect on the frictional head loss in the inclined pipeline and
solids effect on the frictional head loss in the horizontal pipeline for the same mixture
flow parameters is then given as

I mω − I f
= cos ω (6.2).
I m − If

Imω hydraulic gradient due to friction in mixture flow in inclined pipe [-]
Imω - If solids effect in inclined pipe [-]
Im - If solids effect in horizontal pipe [-]

6.2.3 Inclined-flow model by Gibert

Gibert (1960) adapted the Durand & Condolios correlation (Eq. 4.4) to inclined
pipelines by using a simple assumption that only the gravitational acceleration
component perpendicular to an inclined-pipeline axis (g.cosω) influences the solids
effect on the frictional head loss

−1.5
I mω − I f  V2 gd 
= K m  (6.3).
I f C vd  gD v t cos ω 
 

This gives

I mω − I f
= (cos ω)1.5 (6.4).
I m − If

6.2.4 Inclined-flow model by Wilson

FRICTIONAL HEAD LOSS

Wilson et al. (1997) proposed the following modification of the Worster & Denny
formula

Imhω = If + (Im - If )cosω(1+Mγ) + Cvd (Ss -1)sinω (6.5)

thus
= (cos ω)(1+ Mγ )
I mω − I f
(6.6).
I m − If

The power γ has a lower limit of 0.333 for very fine particles and, hypothetically, an
upper limit of unity for very coarse particles. The power M is PSD-dependent and it
gains a value 1.7 for a uniform PSD. Lower values of M are obtained for a well-
graded PSD according to Eq. 4.18.
6.6 CHAPTER 6

DEPOSITION-LIMIT VELOCITY

The application of maximum deposition-limit velocity Vsm by demi-McDonald can


be extended to inclined pipelines by using the dimensionless deposition parameter ∆D
sensitive to an angle of a pipeline inclination. The deposition-limit velocity in an
inclined pipeline Vsmω is given as

Vsmω = Vsm + ∆ D 2gD(Ss − 1) (6.10).

Vsmω maximum deposition-limit velocity in inclined pipe [m/s]


Vsm maximum deposition-limit velocity in horizontal pipe [m/s]
∆D deposition parameter, ∆D = fn(ω) from a graph on Fig. 6.3 [-]
ω pipeline inclination angle [deg]
D pipeline diameter [m]
Ss relative density of solids [-]

Figure 6.3. Effect of angle of pipe inclination on Durand deposition parameter,


after Wilson & Tse (1984).

6.2.5 Discussion of the empirical models

According to models of Worster & Denny, Gibert or Wilson, the solids effect is
always lower in inclined pipelines (for both the negative and positive slopes) than in
horizontal pipelines. Furthermore, the friction loss is the same in pipe sections of the
negative and the positive slope when the pipe inclination angle and flow parameters
Vm, Cvd, d are identical.
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.7

The above prediction is based on an assumption that the frictional head loss due to a
presence of solids is caused predominantly by solids settling tendency in a direction
perpendicular to an inclined pipeline wall. A measure of a settling tendency
represented by a particle settling velocity decreases with an increasing pipeline
inclination angle because the component of the particle settling velocity in a direction
perpendicular to a pipeline wall decreases and so decreases the possibility that the
particles form a granular bed.

A physical consideration of a friction process in a stratified flow learns that the above
assumption is not generally acceptable. A simplification of the flow process assumed
in models described above might be acceptable for slightly stratified flows in which
only a small portion of solids tends to form a bed.

The above-discussed models assume that all particles occupying the pipeline section
attribute to the mixture column that produces a static head. Static pressure difference
caused by pipe elevation is considered to be produced by a mixture column of
concentration Cvd. Accepting the fact that all solids present in inclined pipe
contribute to mixture column weight, the actual spatial concentration Cvi should
determine the solids concentration in the mixture column. Correct determination of Im
ω from measured manometric pressure differential demands understanding of the
difference between Cvd and Cvi in a measuring pipe section.

6.3 PHYSICAL MODELING OF INCLINED FLOWS

6.3.1 Two-layer model

Modeling of inclined flows might be successfully carried out using the empirical
models if flow is only slightly stratified. Flows exerting significant flow stratification
obey physical rules that are not considered in the empirical models. A two-layer
model for inclined flows considering the two-layer flow pattern shown on Fig. 6.1
should handle such flows. For this model the force balance equations are formulated
as follows (Matousek, 1997):
in the upper layer

(∆P + H pipeρ1g )A1 = τ1O1L + τ12O12 L (6.7)

and in the lower layer

(∆P + H pipeρ1g )A 2 − FW sin ω = τ2f O2 L + µs FN cos ω − τ12O12 L (6.8).

In the whole pipe section the balance is then

(∆P + H pipeρ1g )A − FW sin ω = τ1O1L + τ2f O2 L + µs FN cos ω (6.9).


6.8 CHAPTER 6

6.3.2 Discussion of the physical model

The model of this configuration for the inclined flows takes the inclination effects of
different solids-support mechanisms into account. Basically, this model distinguishes
between the frictional pressure differential and the static pressure differential on a
basis of the physical picture of a friction process in inclined flows.

6.3.3 Observations verifying a validity of the physical model principles

Experiments carried out in Laboratory of Dredging Technology of Delft University of


Technology in 1995-1996 (Matousek, 1997) provided the following conclusion:
The mixture flow behaves differently in an ascending pipe and in a descending pipe if
flow tends to be stratified, i.e. if there is a bed at the bottom of the pipe. The
difference was detected in measured frictional head loss, flow stratification, slip
between phases in the flow and velocity of the bed. The difference diminishes at
pipe-inclination angles above approximately 45 deg where the bed starts to be
disintegrated.

Figure 6.4. Concentration profiles in a 150 mm pipe cross section for flow of an
aqueous mixture of a 1.4 – 2.0 mm sand at velocity 3.5 m/s.
(Data from Laboratory of Dredging Technology, Delft University of Technology).

a. Flow stratification
Different degrees of flow stratification have been observed in the ascending pipe and
the descending pipe for the same slurry flow conditions (Vm, Cvd). The difference is
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.9

large for the coarse slurry flow and small for the relatively fine slurry flow. Thus the
effect occurs in slurry flow where the turbulent suspension mechanism plays a minor
role (or it is not effective at all) and the majority of particles occupy a granular bed.
A sharp flow stratification in the descending pipe and a gradual concentration
change across the pipe cross section in the ascending pipe for coarse slurries (see Fig.
6.4) suggest that the small concentration gradient in the ascending pipe is the product
of dispersive forces acting within the shear layer rather than of the turbulent mixing
process in the liquid flow. In the ascending pipe the bed-submerged-weight
component exerted against the flow direction has a resisting effect on the sliding bed
and, owing to the steep velocity gradient between the sliding bed and the flow above
it, a thick shear layer is developed. In the descending pipe, owing to the propelling
effect of the submerged weight component in the flow direction, the velocity of the
moving bed is sufficient to prevent the formation of a shear layer. Liquid turbulence
alone is not able to suspend the coarse particles. Finer slurry flow at the same pipe
inclinations (+ω, -ω) demonstrates a considerably smaller difference between the
shapes of concentration profiles, suggesting that here the carrier turbulence is the
prevailing suspension mechanism and the shear layer is not well developed. It should
be remembered that these effects are of importance primarily in flows inclined to
angles not far above 35 deg. At these angles the cross-pipe component of the
submerged bed weight is still important and the pipe slope is not the main cause of
bed disintegration.

b. Slip ratio (transport factor) and bed velocity


The variation in the slip ratio is primarily due to a variation in the shift between
layers in a stratified flow. Slip ratio is found to be strongly dependent on the shape of
the concentration profile (see Fig. 6.4). The slip ratio value tends to approach unity
when the flow becomes less stratified. If FWsinω exceeds µsFNcosω in a descending
pipe the slip ratio reaches a value higher than unity. This is caused by a fact that a bed
moves faster than the suspension layer above the bed under the above given condition
in a descending pipeline. In an ascending pipeline the slip ratio is always lower than
unity, i.e. the bed moves always slower than the suspension layer above the bed.

c. Static head
According to Bagnold's concept for the solids support in a mixture flow, the
contact-load particles transfer their submerged weight to the pipe wall via the
interparticle contacts. The particles are supported by the interparticle contacts. Solid
particles with no interparticle contacts (suspended particles) transfer their weight to
the carrying liquid and increase the density of the suspension. Thus only the solid
particles whose submerged weight is not transmitted to the pipeline wall contribute to
the slurry column which exerts the static pressure differential over an inclined
pipeline section.
The density of the slurry column is the density of the mixture of the carrying liquid
and suspended particles in an inclined pipeline section. The spatial concentration Cvi
in a pipeline section can be used to calculate slurry column density only when all
particles are suspended. The delivered concentration Cvd determines the slurry
column density only when all particles are suspended and, furthermore, the slip
between phases in a pipeline section is negligible.
6.10 CHAPTER 6

6.3.4 Comparison of empirical and physical approaches to the inclined flow


modeling

Consider as an example an inclined flow of fully-stratified mixture in a descending


pipeline inclined to -35 degrees. In such a flow the total pressure differential is
experimentally detected as almost equal to that of carrying liquid alone. According to
an empirical model this is because the frictional pressure drop produced by a presence
of solids in a carrying liquid flow is reduced by a pressure gain due to the static
pressure from a mixture column (containing all solid particles) in a pipeline section.
According to the two-layer model, however, no pressure gradient is required to push a
bed because the bed moves gravitationally (it might even move faster than the
carrying liquid in the descending pipeline). Thus no extra frictional pressure
differential is developed due to the presence of solids in a pipeline. In the same time,
no static pressure differential occurs due to presence of solids because all solid
particles occupy the bed and thus do not contribute to the column exerting a static
pressure differential. Thus solid particles do not affect the total pressure differential
over a descending pipeline section. The predicted sum of the frictional and the static
pressure differential is similar from both the empirical and physical models.
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.11

CASE STUDY 6

Mixture flow in an inclined pipeline

An aqueous mixture of fine sand or medium gravel (see previous Case studies) is
pumped from a borrowing pit to a hopper on board of a trailing suction hopper
dredge. The dredging depth is 50 meter. A suction pipeline of an internal diameter
900 millimeter is inclined under the angle 45 deg.

Propose a suitable transport velocity for mixture in a pipeline and determine the
required manometric head of the pump, the energy lost due to friction in a suction
pipeline, the specific energy consumption and the production for mixture transport at
the chosen transport velocity. The absolute roughness of a pipeline wall is 20 microns.

Remark: Consider 1.1Vdl (deposition-limit velocity) a suitable transport velocity of


mixture in the inclined pipeline. For a simplification consider a narrow graded soil
characterized by the median diameter only.

Inputs:

d50 = 0.120 mm of d50 = 6.0 mm


ρs = 2650 kg/m3
ρf = 1000 kg/m3
νf = 0.000001 m2/s
Cvd = 0.27
∆hdepth = 50 m
ω = 45 deg
D = 900 mm
k = 0.00002 m

Solution:

a. The deposition-limit velocity

Fine sand (d = 0.120 mm)

In Case study 4 the deposition-limit velocity was determined for a horizontal flow by
the MTI correlation: Vdl = 3.23 m/s (Eq. 4.19).
The deposition-velocity correction for inclined flow is carried out using the Wilson
method. The deposition parameter ∆D is found for ω = 45 deg from the nomograph on
Fig. 6.3: ∆D = 0.33.
The Eq. 6.10 gives
Vdlω = Vdl + ∆ D 2gD(Ss − 1) = 3.23 + 0.33 2x 9.81x 0.9x1.65 = 5.01 m/s.

Medium gravel (d = 6.0 mm)

In Case study 4 the deposition-limit velocity was determined for a horizontal flow by
the MTI correlation: Vdl = 7.03 m/s (Eq. 4.19).
6.12 CHAPTER 6

The deposition-velocity correction for inclined flow is carried out using the Wilson
method. The deposition parameter ∆D is found for ω = 45 deg from the nomograph on
Fig. 6.3: ∆D = 0.33.
The Eq. 6.10 gives
Vdlω = Vdl + ∆ D 2gD(Ss − 1) = 7.03 + 0.33 2x 9.81x 0.9x1.65 = 8.81 m/s.

The suitable transport velocity for sand-water mixture: Vm = 1.1Vdlω = 5.50 m/s.
The suitable transport velocity for gravel-water mixture: Vm = 1.1Vdlω = 9.70 m/s.

b. Energy loss due to friction & required manometric head of the pump

Fine sand (d = 0.120 mm)

Water flow:
Re = 5.5*0.9/0.000001 = 4.95 x 106
k/D = 0.00002/0.9 = 2.2 x 10-5 (D/k = 45000)
λf = 0.0103 (see Moody diagram, Fig. 1.6)
Friction head loss from the Darcy-Weisbach equation (Eq. 1.20)
λ V 2 0.0103 5.52
If = f m = = 0.0176 [-].
D 2g 0.900 19.62
Mixture flow: Wilson model for heterogeneous flow
Horizontal pipeline (Eq. 4.16 and Eq. 4.17)
0.45
 S − 1
V50 ≈ 3.93(d 50 ) 0.35  s  = 3.93(0.12) 0.351 = 1.87 m/s.
 1.65 
−M −1.7
I m − If V   5.50 
= 0.22 m  = 0.22  = 0.03515
C vd (Ss − 1) V
 50   1 . 87 
Im = 0.0176 + 0.0352 x 0.27 x 1.65 = 0.0333 [-].

Inclined pipeline
Frictional head loss (Eq. 6.6) for M = 1.7 and γ = 0.4 (estimated):
= (cos ω)(1+ Mγ ) = (cos 45)(1+1.7x 0.4 ) = 0.5586
I mω − I f
I m − If
Imω = 0.0176 + 0.5586(0.0333 - 0.0176) = 0.0264 [-].

Manometric gradient (Eq. 6.5):


Imhω = 0.0264 + 0.27 x 1.65 x sin(45) = 0.3414 [-].

The head that must be delivered by a dredge pump to lift the mixture from a
borrowing pit to a hopper is the head required to overcome the friction and the
difference in a geodetic position of the pit and the hopper. If the position of the pump
and the hopper inlet is considered equal to the water-level position and the local
losses in a suction pipeline are neglected, the required head, Hman [meter water
column, mwc], is:

Hman = Imhω x Linc = Imhω x ∆hdepth/sin(ω) = 0.3414 x 50/sin(45) = 24.1 mwc.


PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.13

Medium gravel (d = 6.0 mm)

Water flow:
Re = 9.7*0.9/0.000001 = 8.73 x 106
k/D = 0.00002/0.9 = 2.2 x 10-5 (D/k = 45000)
λf = 0.010 (see Moody diagram, Fig. 1.6)
Friction head loss from the Darcy-Weisbach equation (Eq. 1.20)
λ V 2 0.010 9.7 2
If = f m = = 0.0533 [-].
D 2g 0.900 19.62
Mixture flow: Wilson model for heterogeneous flow
Horizontal pipeline (Eq. 4.16 and Eq. 4.17)
0.45
0.35  Ss − 1 
V50 ≈ 3.93(d 50 )   = 3.93(6.0) 0.351 = 7.36 m/s
 1.65 
−M −1.7
I m − If  Vm   9.70 
= 0.22  = 0.22  = 0.1376
C vd (Ss − 1)  V50   7.36 
Im = 0.0533 + 0.1376 x 0.27 x 1.65 = 0.1146 [-].
Inclined pipeline
Frictional head loss (Eq. 6.6) for M = 1.7 and γ = 0.9 (estimated):
= (cos ω)(1+ Mγ ) = (cos 45)(1+1.7x 0.9 ) = 0.4161
I mω − I f
I m − If
Imω = 0.0533 + 0.4161(0.1146 - 0.0533) = 0.0788 [-].

Manometric gradient (Eq. 6.5):


Imhω = 0.0788 + 0.27 x 1.65 x sin(45) = 0.3938 [-].

The head that must be delivered by a dredge pump to lift the mixture from a
borrowing pit to a hopper is the head required to overcome the friction and the
difference in a geodetic position of the pit and the hopper. If the position of the pump
and the hopper inlet is considered equal to the water-level position and the local
losses in a suction pipeline are neglected, the required head, Hman [meter water
column, mwc], is:

Hman = Imhω x Linc = Imhω x ∆hdepth/sin(ω) = 0.3938 x 50/sin(45) = 27.9 mwc.

c. Specific energy consumption (Eq. 3.6)

Fine sand (d = 0.120 mm)


I 0.3414
SEC = 2.7 mωh = 2.7 = 1.288 [kWh/(tonne.km)].
Ss .C vd 2.65x 0.27

Medium gravel (d = 6.0 mm)


I 0.3938
SEC = 2.7 mhω = 2.7 = 1.486 [kWh/(tonne.km)].
Ss .C vd 2.65x 0.27
6.14 CHAPTER 6

d. Production

Fine sand (d = 0.120 mm)

Production of solids: (Eq. 3.3)


π π
Qs = D 2 Vm C vd 3600 = 0.9 25.5x 0.27x 3600 = 3401.0 [m3/hour].
4 4

Production of in situ soil: (for porosity n = 0.4) (Eq. 3.4)


π Q
Qsi = D 2 Vm C vdsi 3600 = s = 5668.3 [m3/hour].
4 1− n

Medium gravel (d = 6.0 mm)

Production of solids: (Eq. 3.3)


π π
Qs = D 2 Vm C vd 3600 = 0.9 29.7x 0.27x 3600 = 5998.1 [m3/hour].
4 4

Production of in situ soil: (for porosity n = 0.4) (Eq. 3.4)


π Q
Qsi = D 2 Vm C vdsi 3600 = s = 9996.8 [m3/hour].
4 1− n

Summary of the results:

Fine sand (d = 0.12 mm):


suitable transport velocity: Vm = 5.50 m/s
frictional head loss: Imω = 0.0264 [-]
required manometric head: Hman = 24.1 mwc
specific energy consumption: SEC = 1.288 kWh/(tonne.km)
production of in situ soil: Qsi = 5668.3 m3/hour

Medium gravel (d = 6.00 mm):


suitable transport velocity: Vm = 9.70 m/s
frictional head loss: Imω = 0.0788 [-]
required manometric head: Hman = 27.9 mwc
specific energy consumption: SEC = 1.486 kWh/(tonne.km)
production of in situ soil: Qsi = 9996.8 m3/hour
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.15

6.4 UNSTEADY SOLIDS FLOW

During dredging operations slurry density varies in time and space along the entire
long pipeline of a conveying system. The solids flow is unsteady (Qs ≠ const) even if
controlled global operational parameters of the system (slurry flow rate, Qm, through
the conveying system and pump speed) are assumed to be maintained at an
approximately constant level during the entire operational period of the system.

Figure 6.5. Process of solids aggregation along a long dredging pipeline.


6.16 CHAPTER 6

6.4.1 Solids aggregation along a long dredging pipeline

A fluctuating density, generated at the inlet of the system, moves through a pipeline.
Field measurements (see Fig. 6.5) on a dredging installation with a pipeline that is
approximately 10 km long and which has three booster stations in series, show that
density fluctuations in the flow of slurry containing rather broad-graded medium sand
are not flattened. Whilst passing along the pipeline with pumps in series, they are
transformed into long density waves with a high amplitude. The transformation of
density fluctuations indicates a solids aggregation process. The influence of a pump
performance on density waves transformation is negligible. An aggregation
mechanism is active in the pipeline.

6.4.2 Description of the solids-aggregation process

When unsteady solids flow in a long pipeline is modeled by means of basic


hydrodynamic equations, including transport and turbulent dispersion effects, the
fluctuating slurry density entering the system is assumed to be gradually flattened and
become almost constant in time and space along the long pipeline. This mechanism is
effective in a short time and length scale and causes a flattening of short-time density
fluctuations behind a dredge pump (compare Gr and Ja density signals of Fig. 6.5).
Over a longer time and length scale (more suitable for a description of the process in a
pipeline which is more than 10 km long, in which each particle needs almost one hour
to reach its destination from the bottom of a lake) a different mechanism may be
prevailing.

With respect to the specific flow conditions in a long slurry pipeline connected with a
dredge, it is believed that a process of material aggregation is caused by the
hydrodynamic interaction between the bed layer and the suspension layer in a
partially-stratified flow of mixture. This interaction leads to the mass exchange
between the bed layer and the suspension layer (the non-equilibrium between the
settling flux and erosion flux across the interface between layers), to the variation in
the bed velocity and produces variable slip in an unsteady solids flow along the long
pipeline.

The top of the granular bed is subjected to the highest shear stress if the densest
suspension passes the bed. At this situation the the bed velocity and the erosion flux
are the highest. When the measured signal for local solids velocity at the bottom of
the pipeline cross section is compared with the measured signal for mean slurry
density just passing the pipeline cross section, the reaction of bed velocity to the
fluctuating slurry density is clearly seen (Fig. 6.6a, b). An exact description of the
aggregation process requires an analysis of the hindered settling and hindered erosion
in high concentrated mixture.

The aggregation of solids to high density waves occurs at low average velocities
round and below the deposition-limit value. The interaction between layers becomes
weak if mean mixture velocity grows far above the deposition-limit threshold.
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.17

Figure 6.6a. Local velocity of solids at the bottom of a pipeline for flow of medium
to coarse sand mixture under fluctuating density.

Figure 6.6b. Local velocity of solids at the bottom of a pipeline for flow of medium
to coarse sand mixture under fluctuating density.
6.18 CHAPTER 6

6.4.3 Practical consequences of the solids-aggregation process

An aggregation process may have an influence on the efficiency and safety of the
operation of the system. Consideration of the effects of the aggregation process on
mechanical energy dissipation and granular deposit formation in a slurry pipeline may
lead to a more effective control of a conveying system.

6.4.3.1 Consequences for pipeline operation

a. fine-to-medium sand with a small proportion of coarse sand and silt

The solids aggregation phenomenon observed in a long slurry pipeline connected with
a dredge is not dangerous for pipeline operation. The formation of high density waves
does not produce moving dunes or a stationary deposit at the bottom of a pipeline, nor
does it increase the friction loss in the slurry flow in the pipeline.

Frictional head loss in unsteady solids flow does not increase significantly with the
mean slurry density in the pipeline cross section if the transported solids are relatively
broadly graded (see Fig. 6.7). As a result the specific energy consumption (SEC) in
the dredging pipeline decreases rapidly when the density of transported mixture
increases.

Figure 6.7. Pressure drop due to friction over a long pipeline section compared with
changing mixture density in the pipeline section.
Measured hydraulic gradient compared with the Durand model.
PIPELINE FLOW CONDITIONS TYPICAL FOR DREDGING 6.19

b. coarse sand or gravel

Different phenomena may occur when coarse solids are pumped. In this case there is
no impelling effect caused by the denser suspended layer, since the majority of
particles occupy the bed in the flow of a mean slurry velocity not far above the
deposition-limit value. The unsteady state of the solids flow causes that the thickness
of the bed varies significantly along the pipeline. Further instabilities may occur
owing to shear stress variation at the top of a bed of variable thickness. Instabilities
may lead to the gradual development of dunes, their mutual separation and their
transformation into plugs along the pipeline if the mixture flows at velocity near the
deposition-limit threshold. Such plugs may block the pipeline.

6.4.3.2 Consequences for pump-pipeline operation

The formation of density waves in a dredging pipeline has a considerable impact on


the operation of slurry pumps and drives incorporated into a conveying system.
Density waves passing through the slurry pumps cause the working point of a
pump-pipeline system to vary in time during the operation of the system (see Chapter
7). The situation is more complex in a system composed of a pipeline and a set of
pumps. Analysis of the pump-pipeline interactions and of the impact of slurry density
fluctuation on the efficiency of a conveying system is an interesting subject for further
research.

6.4.3.3 Production measurement on board of a dredge

Slip occurs between solid and liquid phases in slurry pipelines as a result of a flow
stratification. The slip must be taken into account when the solids throughput in a
dredging pipeline is being determined. In horizontal pipelines occupied by the slurry
exhibiting a considerable slip the solids concentration Cvi (a fraction of solids
actually present in the a pipeline section) is higher than the delivered concentration
Cvd. During a dredging operation the solids throughput is usually determined on-line
and displayed on the dredgemaster's control board. The solids throughput Qs is
calculated as Qs = CvVπD2/4 from on-line signals of the measured mean liquid
velocity (Vf, by a magnetic flow meter) and mean spatial concentration (Cvi, by a
radiometric density meter) in a pipeline cross section. The measuring instruments are
often installed in a horizontal pipeline section at some distance behind a dredge pump.
If flow stratification resulting in slip occurs in this pipeline section, the values of Qs
obtained when using CviVfπD2/4 may be too high. Using CvdVmπD2/4 would give
the correct values. Thus the monitoring system for a dredging installation may
overestimate the solids throughput in a pipeline connected with a dredge.
6.20 CHAPTER 6

6.5 REFERENCES

Coulson, J.M., Richardson, J.F., Backhurst, J.R. & Harker, J.H. (1996).
Chemical Engineering. Vol. 1: Fluid Flow, Heat Transfer and Mass Transfer.
Butterworth-Heinemann.
Gibert, R. (1960). Transport hydraulique et refoulement des mixtures en conduites.
Annales des Ponts et Chausees, 130(3), 307-74, 130(4), 437-94 (Dutch translation:
Het persen van mengsels door leidingen. TU Delft).
Matousek, V. (1997). Flow Mechanism of Sand-Water Mixtures in Pipelines. Delft
University Press.
Wilson, K.C. & Tse, J.K.P. (1984). Deposition limit for coarse-particle transport in
inclined pipes. Proc. Hydrotransport 9, BHRA Fluid Engineering, Cranfield, UK, pp.
149-61.
Wilson, K.C., Addie G.R., Sellgren, A. & Clift, R. (1997). Slurry Transport Using
Centrifugal Pumps. Blackie Academic & Professional.
Worster, R.C. & Denny, D.F. (1955). Hydraulic transport of solid materials in
pipelines. Proc. Instn Mech. Engrs, (London), 169, 563-86.
7.
PUMP AND PIPELINE CHARACTERISTICS

In the previous chapters the flow of mixture in pipelines was discussed in details. The
following chapters will deal with the effects of an interaction between a pipeline and
pumps on flow of mixture through a dredging installation. Study of these chapters
assumes knowledge of a basic theory of slurry pipeline flows (handled in Chapters 1
to 6) and of a basic theory of centrifugal dredge pumps and pump drives as given in
professor Vlasblom’s course.

7.1 THE BERNOULLI EQUATION

The amount of mechanical energy available in a pipeline flow of fluid is quantified in


the Bernoulli equation. This equation is obtained by integrating the Euler’s equation
of motion along a streamline. If flow is steady, frictionless and incompressible then at
an arbitrary location along a streamline

p v 2f
h+ + = const. (7.1)
ρf g 2g

h geodetic position of a location,


elevation above datum [m]
p pressure at a location on a stream line [Pa]
vf velocity of fluid at a location on a streamline [m/s].
ρf density of flowing fluid [kg/m3]
g gravitational acceleration [m/s2].

Each term of the above equation represents a head (Dutch: opvoerhoogte) with unit
[m]. The head can be interpreted as energy per unit gravity force (see Chapter 3). The
first term is potential energy of fluid control volume per unit gravity force, the second
term flow energy (or flow work) and the third term kinetic energy (see Fig. 7.1). Thus
for two fluid control volumes at two different locations along a streamline the sum of
these three energy terms is constant. However, the proportion of the values of the
particular energy terms changes when flow conditions change during a motion of a
fluid particle from one location to another.

7.1
7.2 CHAPTER 7

P Vm2
h+ + = Level of Mech. Energy
ρ f g 2g

Figure 7.1. Different forms of head in Bernoulli equation.

Consider a water flow through a pipe section. Flow is steady (flow rate is constant)
and incompressible (density is constant). If the pipe section is horizontal and a pipe
diameter at the beginning of a pipe section is smaller than at the end of the pipe
section than the pressure at the section begin is lower than that at the section end (see
Fig. 7.2: a horizontal pipe section in front of a pump). This is because the velocity at
the inlet is higher than at the outlet. A portion of kinetic energy is transformed to flow
energy in a pipe section. If a pipe remains of a constant diameter but the pipe section
is inclined the pressure at the top of the pipe section is smaller than that at the bottom
of the pipe section (see Fig. 7.2: an inclined suction pipe). Work had to be done (flow
energy lost) to lift water particles from the bottom to the top of a pipe section. Lifted
particles gained potential energy.

In practice, transported media of our interest (water or mixture) are considered


incompressible but the flow of these media can not be considered frictionless. If
flowing through a pipe, water or mixture dissipates a portion of their mechanical
energy. They transform a portion of their mechanical energy into thermal energy
(heat). The mechanical energy loss along a pipe section (between cross sections 1 and
2) must be incorporated to the Bernoulli equation so that
PUMP AND PIPELINE CHARACTERISTICS 7.3

2 V22
P1 V1 P2
h1 + + = h2 + + + H totalloss (7.2)
ρ f g 2g ρ f g 2g

Htotalloss total head loss due to mechanical energy dissipation


between pipe cross sections 1 and 2 [m]
P mean absolute pressure in a pipe cross section [Pa]
h geodetic height of a pipe cross section [m]
V mean velocity in a pipe cross section [m/s]
ρf density of fluid [kg/m3]
g gravitational acceleration [m/s2].

pump-suction pressure
pressure in suction pipe

pump-discharge pressure
pipe-inlet pressure

reference level (zero-pressure line)

water level

pressure meter

waterway bottom

Figure 7.2. Pressure variation along a pipeline connected with a pump (schematic).

7.2 H - Q CURVE OF A CENTRIFUGAL PUMP

A rotating impeller of a centrifugal pump adds mechanical energy to the medium


flowing through a pump. As a result of an energy addition a pressure differential
occurs in the pumped medium between the inlet and the outlet of a pump (see the
difference between the pump-suction pressure and the pump-discharge pressure in
Fig. 7.2). The pressure, or the energy head, added to the medium depends on the
7.4 CHAPTER 7

speed (revolutions per minute, r.p.m.) of an impeller and on the flow rate of medium
through a pump. A relationship between the head, H, the flow rate (called also
capacity), Q, and the revolutions per minute of the impeller, n, is given by a set of
H-Q curves (Fig. 7.3). A course of these curves is specific for each particular pump.
The course of the curves is sensitive to the geometry of a pump housing and of an
impeller and thus to flow conditions within a pump. The curves are determined by a
pump test. Usually, a pump manufacturer delivers the curves (called pump
characteristics) with a pump. For dredge pumps the pump characteristics may change
in time because the flow conditions within a pump are influenced by a wear of an
impeller and pump housing.

As discussed in Chapter 3, the head H is a measure of the mechanical energy of a


flowing liquid per unit gravity force. It is expressed as the height of the column of
∆P
liquid of ρf exerting the pressure differential ∆P, so that H = .
ρf g

Figure 7.3. Schematic Q-H curves of a centrifugal pump.

The head due to pressure differential generated by a pump is called the manometric
head, Hman, and it has a unit meter water column [mwc] (see Fig. 7.3).

The Hman-Q curve of a pump gives an amount of energy that a pump provides to a
pump-pipeline system for a certain r.p.m. (speed) of a pump impeller and a flow rate
(Q) through a pump.

The manometric head that is delivered by a pump to medium is determined by a


parameter called the manometric pressure and it is given (see also Fig. 7.4) as

ρm ( Vp2 − Vs2 )
Pman = Pp − Ps + ρm g ( h p + h s ) + (7.3)
2
PUMP AND PIPELINE CHARACTERISTICS 7.5

Pman absolute manometric pressure [Pa]


Pp absolute pressure at the discharge outlet of a pump [Pa]
Ps absolute pressure at the suction inlet of a pump [Pa]
hp vertical distance between the pump axis and
the discharge outlet of a pump [m]
hs vertical distance between the pump axis and
the suction inlet of a pump [m]
Vp mean mixture velocity at the discharge outlet of a pump [m/s]
Vs mean mixture velocity at the suction inlet of a pump [m/s]
ρm density of pumped medium [kg/m3]
g gravitational acceleration [m/s2].

Vp
Pp

hp
hs Ps

Vs
Figure 7.4. Conditions at inlet and outlet of a pump.

7.2.1 Affinity laws for pump characteristics

Pump characteristic curves give Hman-Q, Wout -Q (Wout is pump output power) and
η-Q (η is pump efficiency) relationships for certain constant speed n [rpm].

The affinity laws


Q m, n1 n H man, n1  n1 
2 Wout, n1  n1  3 ηf , n1
= 1, =  , =  , =1
Q m, n 2 n 2 H man, n 2  n 2  Wout, n 2  n 2  ηf , n 2

enable to produce the pump curves for different constant speeds, n, of the pump (for
more details see Vlasblom’s lecture notes referred in Chapter 11). An application of
the affinity laws is shown in Case study 7 at the end of this chapter (Fig. C7.2).
7.6 CHAPTER 7

7.2.2 Different regions of pump operation (an interaction between pump and
drive)

A pump-drive system operates in a region of


- the constant speed or
- the constant torque
(for details see Vlasblom’s lecture notes referred in Chapter 11). This affects a shape
of an Hman-Q curve (see Fig. 7.5).

Figure 7.5. Operation regions of a pump-drive system.


PUMP AND PIPELINE CHARACTERISTICS 7.7

7.2.3 Effect of solids on pump performance

Figure 7.6. Effect of mixture density on pump characteristics. Tests of a


0.5-m-impeller pump connected with a 162 kW MAN diesel engine.
Speed: 1000 rpm. Pumped material: 0.2 – 0.5 mm sand.
(Data from Laboratory of Dredging Technology, TU Delft).
7.8 CHAPTER 7

Solid particles of a pumped mixture diminish the efficiency of a dredge pump (see
Fig. 7.6). The ratio of pump efficiencies when pumping mixture or water fc = ηm/ηf
is also a measure of manometric pressure reduction and output power reduction
Pman, m ρ m Win, m ρ m Wout, m ρ m ηm
= fc , = , = fc , = fc (7.4).
Pman,f ρf Win,f ρf Wout,f ρf ηf

Figure 7.7. Parameter fc giving reduction of the pump performance due to presence
of sand/gravel particles in pumped liquid according to Eq. 7.5a.

The parameter fc for sand and gravel mixtures is according to Stepanoff (1965)
related to the particle size, d50, and delivered concentration, Cvd, of solids in
transported mixture by

f c = 1 − C vd (0.8 + 0.6 log d 50 ) (7.5a).

In this equation d50 is in [mm] and Cvd in [-]. Reduction of pump efficiency and
manometric head increases with a particle size and solids concentration. The
reduction is relatively small for fine sand but it is very significant if mixtures of
coarse sand or gravel are pumped (see Fig. 7.7).
The original Stepanoff equation (Eq. 7.5a) does not consider the effect of an impeller
size. However, this effect may be of significant importance. The revised Stepanoff
equation (e.g. Miedema, 1999) including the impeller diameter Dimpel is

C vd (0.466 + 0.4 log d 50 )


fc = 1 − (7.5b).
D impel
In a pump-pipeline system the manometric pressure (or manometric head) of a dredge
pump is required to overcome the total head loss in mixture transported in a pipeline
connected to a dredge pump. The total head loss is composed of
- the major and minor losses due to flow friction in a suction pipeline,
- the loss due to the change in elevation of a suction pipeline,
- the major and minor losses due to flow friction in a discharge pipeline,
- the loss due to the change in elevation of a discharge pipeline,
- the losses due to mixture acceleration in a pipeline.
PUMP AND PIPELINE CHARACTERISTICS 7.9

7.3 H-Q CURVE OF A PIPELINE

The Hman-Q curve of a pipeline gives an amount of energy that the pipeline requires
maintaining a certain flow rate in a pump-pipeline system.

The required amount of mechanical energy is equal to the sum of the energy
dissipated due to friction in a flow of mixture through a pipeline and the potential
energy delivered to (or lost in) mixture to reach a pipeline outlet if this is at a higher
(or lower) geodetic level than a pipeline inlet.
A head lost due to flow of mixture in a pipeline is regarded as sum of major losses
due to internal friction in flow of mixture through straight pipeline sections and minor
losses due to flow friction caused by pipeline fittings.

7.3.1 Head loss in straight pipelines (major loss)

A determination of the frictional head loss for flow of water or mixture in straight
pipelines was a subject to discussion in Chapters 1, 4 and 5. The frictional head loss
in water flow is determined using the Darcy-Weisbach equation (Chapter 1). This
gives a parabolic H-Q curve (called a pipeline-resistance curve) described by the
equation

2 2
λ L Vf λ L Qf
H major, f = f = f (7.6).
D 2g D 2 gA2

Hmajor,f head loss due to friction of water in a straight pipe [Pa]


λf flow friction coefficient [-]
L length of a pipe [m]
D diameter of a pipe [m]
Vf mean mixture velocity in a pipe [m/s]
Qf mixture flow rate through a pipe [m3/s]
A area of a pipe cross section [m2]
g gravitational acceleration [m/s2].

The course of a pipeline-resistance curve is more complicated for mixture flows.


However, they are models available that are capable of predicting the resistance
curves for a various mixtures flowing in a pipeline (see Chapters 4 and 5). The
models predict the hydraulic gradient Im and this is interpreted as the head lost along
a pipeline of a length L using

Hmajor,m = ImL (7.7)

Hmajor,m head loss due to friction of mixture in a straight pipe [Pa]


Im hydraulic gradient in mixture flow according to
a suitable model [-]
L length of a pipe [m].
7.10 CHAPTER 7

7.3.2 Head loss in flow through fittings (minor loss)

Fittings as bends, joint balls, expansions and contractions of a discharge area, valves
and measuring instruments act as obstructions to the flow. The pipeline inlet and
outlet are also sources of local losses. Obstructions cause flow separation and an
induced mixing process in the separated zones dissipates mechanical energy. This
energy dissipation is additional to that in flow through straight pipeline sections. A
portion of energy dissipated due to a presence of fittings is usually considerably
smaller than frictional losses in straight pipes. In long dredging pipelines behind a
dredge the minor losses might be even considered negligible in comparison with
straight-pipe losses.

The minor losses for water flow obey a quadratic relationship between local head loss
and mean velocity through a fitting

Vf2
H min or ,f = ξ (7.8)
2g

Hminor,f head loss due to friction of water flow in fittings [Pa]


ξ minor loss coefficient [-].

Remark: ξ value for the pipeline outlet is 1.0 (ξoutlet=1.0).

Figure 7.8. Coefficient of minor losses for some fittings.


PUMP AND PIPELINE CHARACTERISTICS 7.11

Values of the coefficient ξ vary between zero and one for different fittings (e.g. Fig.
7.8). Experimentally determined values for ξ are available for water flow through
various fittings and various fitting configurations in the literature.

A correct determination of minor losses for mixture flows is more complicated,


particularly for stratified flows. In a dredging practice a simple assumption is often
applied that mixture density alone sufficiently represents an effect of solids on the
minor loss so that

Vf2 ρm
H min or ,m = ξ (7.9).
2 g ρf

This might be a suitable approach for fully-suspended flows in which slurry density
might directly influence frictional losses through wall shear stresses. In stratified
flows, however, an induced local turbulence of a carrying liquid might take a portion
of particles from a bed to suspension and reduce frictional losses in a pipeline section
of a certain length behind a fitting. This effect must be taken into account. If no extra
suspension is assumed due to flow disturbances (in flow of coarse particles), an
energy dissipation takes place through small turbulent eddies of a carrying liquid that
decay to the viscosity of the carrier. The carrier viscosity is not affected by a presence
of coarse solid particles. Therefore the value of a minor head should not be influenced
either. Thus Eq. 7.9 might overestimate minor losses in stratified flows. However,
very little is known about an effect of solids on the minor losses in pipelines yet. It is
an interesting subject to further investigation.

7.3.3 Total frictional head loss

The total frictional head loss is a sum of head losses due to friction in a straight
pipeline sections and in fittings mounted to a pipeline

Htotalloss = Hmajor + Hminor (7.10).

A total frictional pressure drop over a pipeline of the length L is given for a water
flow by the equation

 L 1
∆Ptotalloss,f =  λ f + ξ  ρ f Vf2 (7.11)
 D 2

and for a mixture flow by the equation

1 2
∆Ptotalloss, m = I mρf gL + ξ ρmVm (7.12).
2

A development of the total frictional pressure drop under the changing mean velocity
in a pipeline is given schematically in Fig. 7.9.
7.12 CHAPTER 7

Figure 7.9. Schematic Htotalloss-Q curve of a horizontal pipeline.

7.3.4 Geodetic head

A geodetic head (called also static head) was discussed in detail in Chapter 6
concerning inclined flows,

Hstatic = Sm∆h (7.13)

Hstatic geodetic head [m]


∆h elevation change over a pipeline; difference in
geodetic height between pipeline inlet and outlet [m]
Sm relative density of mixture [-].

7.4 WORKING POINT OF A PUMP-PIPELINE SYSTEM

Velocity of transported medium in a pump-pipeline system is determined by a cross


point of a pump H-Q curve and a pipeline H-Q curve (Fig. 7.10). The cross point
gives the velocity at which a balance is found between the energy provided to a
system by a pump and the energy required to overcome a flow resistance in a pipeline
and a change in a geodetic height between the pipeline inlet and outlet.

Practically this means that if water is pumped through a pipeline of certain geometry
(given by diameter, length, elevation and a number of fittings) the rpm installed on a
pump determines directly the velocity of water in a pipeline. An increase in the rpm
(i.e. a step to an another pump H-Q curve of constant rpm) increases the water
velocity because a new working point (Dutch: werkpunt) is found on a pipeline
resistance curve.
PUMP AND PIPELINE CHARACTERISTICS 7.13

Figure 7.10. Working points of a pump-pipeline system when water is pumped


(schematic).

The same rules are valid for mixture pumping if flow conditions are steady, i.e.
mixture density and size of transported solids do not change in time.

For a dredging operation, however, a fluctuation of mixture density is typical. Even if


other parameters (as pump rpm and size of transported solids) are constant, this
density fluctuation might produce a fluctuation of the mean mixture velocity in a
pipeline. The increasing density of mixture in a pipeline is a source of increasing flow
resistance. Thus a balance between a provided energy head (that is constant if rpm of
a pump does not change) and a required energy head (that increases with mixture
density in a pipeline) is found at lower velocity. The velocity increases again if mean
mixture density gradually drops in a pipeline due to lower density of mixture
generated in a pipeline inlet.

7.5 WORKING RANGE OF A PUMP-PIPELINE SYSTEM

If a pipeline resistance is given by just one H-Q curve of a pipeline (a pipeline lay out,
properties of soil and density of transported mixture are constant in time) there is just
one working point at which an installation of a pipeline and a pump at a constant
speed operates. If pipeline resistance changes (usually due to fluctuating mixture
density in a pipeline) an installation operates within a working range (Dutch:
werkgebied) instead of at a working point (Fig. 7.11).
7.14 CHAPTER 7

Figure 7.11. Working points and working range of a pump-pipeline system


(schematic).

A mixture density fluctuates with a high frequency and amplitude within a dredging
pipeline, particularly if a discharge pipeline is connected with a cutter suction dredge
(see Fig. 7.12).

Figure 7.12. Mixture density fluctuations (here referred as Cvsi fluctuations) in


a 500 meter long discharge pipeline connected with a trailing
suction hopper dredge (TSHD) and a cutter suction dredge (CSD)
(after v.d. Berg, 1998).

This causes fluctuation in a manometric pressure provided by a pump. However, a


position of a working point of a pump-pipeline installation is influenced by the mean
mixture density in an entire pipeline rather than by local density fluctuation in a
pump.
PUMP AND PIPELINE CHARACTERISTICS 7.15

Let’s consider an installation composed of a pump, a suction pipe and a discharge


pipe that is considerably longer than a suction pipe. A dredging operation is
monitored from the beginning to the end of a transportation cycle. The short-time
fluctuations of the density of transported mixture can be neglected. The following
stages of a pump-pipeline operation are of importance:

1. the beginning of a cycle: only water flows through the suction pipe and the
discharge pipe
2. the beginning of a soil excavation process: the suction pipe and the pump are
filled with mixture, the discharge pipe is still filled with water only
3. the mixture transportation: both the suction and the discharge pipes are filled with
mixture
4. the end of a cycle: the suction pipe and the pump are filled with water, the
discharge pipe is filled with mixture.

7.5.1 Pump operation within a range of the constant speed

If a pump operates within a range of the constant speed during an entire transportation
cycle a position of a working point varies for the four different stages (described
above) in a way displayed on Fig. 7.13.

Figure 7.13. Working range within a constant speed region of a dredge pump.

An area defined by points 1,2,3 and 4 on Fig. 7.13 gives the working range.
7.16 CHAPTER 7

7.5.2 Pump operation within a range of the constant torque

Shifting of a working point during a cycle described by steps 1 to 4 is different from


that for a pump operating in a constant-speed regime (compare Figs. 7.13 and 7.14).

Figure 7.14. Working range within a constant torque region of a dredge pump.

7.5.3 Pump operation within a range around the nominal torque point

Figure 7.15. Working range round a nominal torque point.


PUMP AND PIPELINE CHARACTERISTICS 7.17

Shifting of a working point during a cycle described by steps 1 to 4 is different from


that for pumps operating either in the constant-speed regime or in the constant-torque
regime (Fig. 7.15).

Let’s compare operations of a certain installation (pumping only water for this
particular example) if a pipeline length changes and other parameters remain constant
(Fig. 7.16). The resistance curve R2 is for a pipeline of an original length, R1 for a
longer pipeline and R3 for a shorter pipeline.

The Fig. 7.16 shows that pumping through a longer pipeline (R1) is associated with
the drop in the output power than pumping through a pipeline of an original length
(R2) if operation realizes within a constant-speed region of a pump. However, a
shortening of the pipeline (R3) can lead to similar drop in the output power the
working point shifts to the constant torque line. In the constant-torque regime the
engine speed decreases in order to avoid overloading of the motor.

Figure 7.16. Working points for systems of different pipeline lengths.


(Legend: R1 longer pipe, R2 original length, R3 shorter pipe)
7.18 CHAPTER 7

7.6 OPERATION UNDER THE CONDITION OF


CONTINUOUSLY FLUCTUATING DENSITY OF MIXTURE

A pump of a dredging installation reacts on short-time fluctuations of density of


mixture passing through the pump (Figs. 7.17, 7.18). Thus the manometric head
provided by the pump fluctuates in time as fluctuates the mixture density. The
variation in the manometric head should lead to variation in mixture velocity in a
pipeline connected with a pump. However, an effect of the discharge pressure
fluctuation on the flow conditions in a discharge pipeline depends on a length of the
pipeline. At each moment a working point of a pump-pipeline installation is
determined by the mean flow conditions (average mixture density, see Fig. 7.17) in an
entire pipeline rather than by local flow conditions in a pump.

Figure. 7.17. Immediate slurry density in a pump (Ja), immediate pressure at the
beginning (Ja) and at the end (Du) of a pipeline section (Ja-Du).
Average mixture density and pressure drop in the pipeline section (Ja-Du) [++++].

Figure. 7.18. Effect of mixture density fluctuation on torque and speed of a booster
pump (Du) (after Matousek, 1997).
PUMP AND PIPELINE CHARACTERISTICS 7.19

This contradiction between momentarily large density fluctuations in a pump and


small changes in a mean density of mixture over an entire pipeline length (see Fig.
7.18, 7.19 or 7.12) makes a prediction and a control of processes in a dredging
installation insecure. This is particularly the case if more pumps are installed in series
in a transportation system.

7.7 OPERATION UNDER THE CONDITION OF FLUCTUATING


DENSITY OF MIXTURE AND MEAN PARTICLE SIZE IN A
PIPELINE

Fig. 7.19 shows a variation of a working-point position in a pump-pipeline system


under the condition of changing mean mixture density (marked here as γgem [that is
ρm]) and particle size (normaal materiaal/grover materiaal). The system operates
within a region of a constant torque of a pump drive. The working-point variation is
plotted in Hman [mwc, meter water column] versus Q [litre/second] co-ordinates.

Figure 7.19. Variation of working point under various mixture flow conditions in a
pump-pipeline system.

The flow rate through a system drops if mixture density or particle size of transported
solids increases in a system.
7.20 CHAPTER 7

7.8 EFFECT OF IMPELLER PARAMETERS ON WORKING


POINT OF A SYSTEM

If a pipeline of a pump-pipeline system becomes very short the working point of the
system reaches the smoke limit of a diesel drive and the drive collapses. This can be
avoided by replacing the impeller of a pump if a pipeline becomes short. The use of a
smaller impeller or of an impeller with fewer blades causes a shift (from A to B in
Figs. 7.21 and 7.21) of a working point to a position far above the smoke-limit point.

Figure 7.20. Effect of a use of smaller impeller in a system with a shorter pipeline.

Figure 7.21. Effect of a use of impeller with fewer blades in a system with
a shorter pipeline.
PUMP AND PIPELINE CHARACTERISTICS 7.21

7.9 REFERENCES

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
Matousek, V. (1997). Flow Mechanism of Sand-Water Mixtures in Pipelines. Delft
University Press.
Miedema, S.A. (1999). Considerations on limits of dredging processes. Proc. WEDA
19th Technical Conference and 31st Texas A&M Dredging Seminar, Louisville,
Kentucky, pp. 233-54.
Stepanoff, A.J. (1965). Pumps and Blowers, Two-Phase Flow: Selected Advanced
Topics. J.Wiley & Sons, Inc.

7.10 RECOMMENDED LITERATURE

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
de Bree, S.E.M. (1977). Centrifugaal Baggerpompen. IHC Holland.
7.22 CHAPTER 7

CASE STUDY 7.1

A deep dredge has a centrifugal pump on board. The heart of the pump is on the same
geodetic height as the water level. The suction and the discharge pipes are mounted to
the pump at the pump-heard level. The suction pipe of the dredge is vertical and the
discharge pipe is horizontal. Both pipes have a diameter 500 mm. The dredge pump
pumps the 0.2-mm sand from the bottom of the waterway that is 7 meter below the
water level (thus the dredging depth is 7 meter). The density of a pumped sand-water
mixture is 1400 kg/m3. The discharge pipe is 750 meter long. The pump-pipeline
installation is supposed to keep the production at 700 cubic meter of sand per hour.

1. Determine the manometric pressure (manometric head) that the pump must deliver
to ensure the required production of the sand for mixture of the density 1400
kg/m3.

2. What is the equivalent manometric pressure of the pump for water service at the
same flow rate? This enables to place the working point to the eventually available
pump characteristic H-Q for water service.

3. Assume that the plotting of the working point into the H-Q nomograph of the
pump revealed that the working point corresponds with the pump speed 400 rpm
(the point lays on the H-Q curve for 400 rpm). The maximum speed with which an
engine can provide the pump is 450 rpm. What would be the flow rate and the
manometric pressure if the maximum speed would be installed?

For the calculation consider the friction coefficient of the suction/discharge pipes λ =
0.011. The following minor losses must be considered:
- the inlet to the suction pipe: ξ = 0.5,
- the 90-deg bend in suction pipe: ξ = 0.1,
- several flanges in suction/discharge pipes: ξ = 0.3,
- the outlet from the discharge pipe: ξ = 1.0.
Additional inputs:
ρf = 1000 kg/m3
ρs = 2650 kg/m3

Inputs:

∆hdepth = 7 m
Lhor = 750 m
D = 500 mm
d50 = 0.20 mm
ρs = 2650 kg/m3, ρf = 1000 kg/m3, ρm = 1400 kg/m3
λf = 0.011, Σξ = 1.9
Qs = 700 m3/hour = 0.194 m3/s

Remark: To make a calculation simpler the effect of a pipeline roughness on frictional


losses in a pipeline is considered to be represented by a constant value of the
frictional coefficient λf, i.e. independent of variation of mean mixture
velocity.
PUMP AND PIPELINE CHARACTERISTICS 7.23

1. Determination of the manometric pressure for mixture service

Mean velocity of mixture in a pipeline, Vm:

ρ − ρf 1400 − 1000
C vd = m = = 0.2424 [-],
ρs − ρf 2650 − 1000

Qs 0.1944
Qm = = = 0.802 m3/s,
C vd 0.2424

4Qm 4 x 0.802
Vm = = = 4.085 m/s.
πD2 3.1416 x 0.52

Manometric pressure of the pump, Pman:

There is a balance between


the pressure difference among the inlet and the outlet of the installation:
Pinlet – Poutlet and the pressure drop over the total length of a pipeline:
∆Pstatic + ∆Ptotloss,m - Pman.

∆Pstatic is the static pressure differential between the inlet and the outlet;
∆Ptotloss,m is the total pressure loss (both major and minor) over the length of a pipe;
Pman is the manometric pressure of the pump.

The balance reads


Pinlet - Poutlet = ∆Pstatic + ∆Ptotloss,m - Pman,

in which Pinlet = Patm + ∆hdepth.ρf.g,


∆Pstatic = ∆hdepth.ρm.g,
Poutlet = Patm.

Thus Pman = ∆hdepth(ρm - ρf)g + ∆Ptotloss,m.

The pressure drop due to losses:

∆Ptotloss,m = ∆Pmajor,m + ∆Pminor,m.

2
Vm 4.092
Minor loss: ∆Pminor,m = Σξ ρm = 1.9 1400 = 22.2 kPa.
2 2

Major loss:

vertical pipe: the equivalent-liquid model: ∆Pvert,m = ∆Pvert,f ρm [Pa].


7.24 CHAPTER 7

The Darcy-Weisbach equation:


∆h depth Vm2 7 4.092
∆Pvert,f = λ f ρf = 0.011 1000 = 12.9 kPa
D 2 0.5 2

∆Pvert,m = ∆Pvert,f ρm = 12880 x 1400 = 18.0 kPa.

horizontal pipe: the Wilson model: ∆Phor,m = fn(d, D, Cvd, ρs, ρf, ∆Ppipe,f) [Pa].

The Wilson model:


0.45
 Ss − 1 
V50 ≈ 3.93(d 50 ) 0.35
  = 3.93(0.20)0.351 = 2.24 m/s.
 1.65 
−1.7 −1.7
Im − If V  Im − If  4.09 
= 0.22  m  => = 0.22   = 0.079 [-].
C vd ( Ss − 1)  V50  C vd ( Ss − 1)  2.24 

∆Phor,m = 0.079xCvd (Ss -1)gρf Lhor + ∆Phor,f [kPa].

Lhor Vm 2 750 4.092


∆Phor,f = λ f ρf = 0.011 1000 = 138.0 kPa
D 2 0.5 2
∆Phor,m = 0.079 x 0.2424 x (1.65-1) x 9.81x1000 x 750 + 138 000 = 229.6 kPa.

∆Ptotloss,m = ∆Phor,m+ ∆Pvert,m+ ∆Pminor,m = 229.6 + 18.0 + 22.2 = 269.8 kPa

Thus
Pman = ∆hdepth(ρm - ρf)g+∆Ptotloss,m = 7(1400-1000)9.81+269800 = 297.3 kPa.

The manometric pressure that the pump must deliver to maintain the required
production is 297 kPa, i.e. 3 bar. The pump provides this manometric pressure at the
flow rate of mixture 0.802 m3/s.

2. Determination of the manometric pressure for water service

Equivalent manometric pressure for water service, Pman,f :

Pman,m 297
Pman,f = = ,
ρm 1400
1 − C vd ( 0.8 + 0.6 log d 50 )  1 − 0.2424 ( 0.8 + 0.6 log 0.2 ) 
ρf  1000 
Pman,f = 233.7 kPa.

The working point for water service is:


Pman,f = 233.7 kPa at Qm = 0.802 m3/s.

Further, it is assumed that this working point holds for pump operation at 400 rpm.
PUMP AND PIPELINE CHARACTERISTICS 7.25

3. Determination of the working point for another speed of the pump

2
n1 Pman,n1  n1 
Qm,n1
The affinity laws: = , =  .
Qm,n2 n 2 Pman,n2  n 2 

The working point at 450 rpm:

0.802 400 3
= , thus Qm,450 = 0.902 m /s.
Qm,450 450

2
233.7  400 
=  , thus Pman,f,450 = 295.7 kPa.
Pman,f ,n2  450 

The working point for water service at 450 rpm is:


Pman,f = 295.7 kPa at Qm = 0.902 m3/s.

ρm
Pman,m,450 = Pman,f ,450 1 − C vd ( 0.8 + 0.6 log d 50 )  =
ρf 
1400
= 295.7 1 − 0.24 ( 0.8 + 0.6 log 0.2 )  = 375.8 kPa.
1000 

The working point for mixture service at 450 rpm is:


Pman,m = 375.8 kPa at Qm = 0.902 m3/s.
7.26 CHAPTER 7

CASE STUDY 7.2

A dredging installation transports a mixture of narrow-graded medium sand (d50 =


0.30 mm, ρs = 2650 kg/m3) and lake water (ρf = 1000 kg/m3) from a dredging pit to a
construction site. The installation is composed of an on-board pump (a centrifugal
pump IHC 125-27.5-50, see pump characteristics in Tab. C7.1) and a pipeline of the
diameter 500 millimetre. Fig. C7.0 shows a lay-out of the installation. The dredging
depth is 15 meter. A suction pipeline is inclined under the angle 45 deg (the pipe
length 21 meter) and horizontal (a 2 meter long section in front of a pump suction
mouth). The centre of the pump is at the water-level position. A discharge pipeline is
horizontal and its geodetic position is considered identical with a water level along its
entire length (an elevation of a discharge pipeline is zero). The discharge pipeline is
composed of a 200 meter long floating pipeline and of an on-shore pipeline of a
variable length. During a dredging operation the average density of pumped mixture
is 1412.5 kg/ m3.

The dredged material has to be delivered to a construction site of a quite large area.
Therefore a length of a discharge pipeline will vary during an operation. Determine
the maximum length of a pipeline attainable when pumping mixture of the above
required density. What will be the flow rate through a pipeline of a maximum length?

2m 200 m variable
15 m

45
Figure C7.0. Schematic lay-out of a pump-pipeline system.

INPUTS:

Lay-out of a pump-pipeline system (see also Fig. C7.0):

∆hdepth = 15 m
ω = 45 deg
Lhoriz,suction = 2 m
Lhoriz,floating = 200 m
Lhoriz,shore = variable
D = 500 mm
PUMP AND PIPELINE CHARACTERISTICS 7.27

Mixture flow characteristics:

d50 = 0.30 mm
ρs = 2650 kg/m3, ρf = 1000 kg/m3, ρm = 1412.5 kg/m3 (i.e. Cvd = 0.25)
λf = 0.011

Remark: To make a calculation simpler the effect of a pipeline roughness on frictional


losses in a pipeline is considered to be represented by a constant value of the
frictional coefficient λf , i.e. independent of variation of mean mixture
velocity.

Pump & drive parameters:

The dredging pump: type IHC 125-27.5-50


the 5-blades impeller of a diameter 1250 mm and a breadth 275 mm
the diameter of a pump inlet: 500 mm

Table C7.1. Input to Case study 7


Characteristics of the pump IHC 125-27.5-50 if pumping water at the maximum speed
(nmax = 475 rpm)

Qm,nmax Pman,f, ηf,nmax


[m3/s] nmax [kPa] [%]
0.45 679.9 55.7
0.50 676.7 58.8
0.55 673.4 61.6
0.60 670.1 64.0
0.65 666.7 66.1
0.70 663.2 67.9
0.75 659.7 69.5
0.80 656.1 70.9
0.85 652.5 72.1
0.90 648.8 73.1
0.95 645.0 74.0
1.00 641.0 74.8
1.05 637.0 75.4
1.10 632.9 76.0
1.15 628.7 76.5
1.20 624.4 76.9
1.25 619.9 77.2
1.30 615.3 77.5
1.35 610.6 77.7
1.40 605.7 77.9
1.45 600.7 78.0
1.50 595.6 78.1

Drive parameter:
The maximum power available at the pump shaft is 1000 kW (at the speed 475 rpm).
7.28 CHAPTER 7

CALCULATION:

a. Pump characteristics

The Pman,f,nmax–Qm,nmax curve of the IHC pump pumping water at the maximum
speed 475 rpm is a curve fitting the points given in Tab. C7.1. This constant-speed
curve can be approximated (correlation coefficient Rxy = 1.00, Fig. C7.1) by the
equation

Pman,f , n max = 702.5 − 42.44Q m, n max − 19.06Q 2m, n max [kPa] (C7.1).

The efficiency curve (ηf,nmax–Qm,nmax) of the pump operating at the maximum


speed is a curve fitting the points given in Tab. C7.1. This curve can be approximated
(correlation coefficient Rxy = 1.00, Fig. C7.1) by the equation

ηf , n max = 1.953Q m, n max − 2.0Q 2m, n max + 0.989Q3m , n max − 0.195Q 4m, n max [-]
(C7.2).

Remark: Theoretically, the 3rd order polynomial is sufficient to relate the pump
efficiency with the flow rate.

Figure C7.1. Characteristics of the dredging pump IHC 125-27.5-50


at the maximum speed 475 rpm.
PUMP AND PIPELINE CHARACTERISTICS 7.29

The affinity laws


Q m, n max n Pman,f , n max  n max  2 η f , n max
= max , =  , =1
Qm n Pman,f  n  ηf
enable to produce the constant-speed curves (Pman,f–Qm) and efficiency curves
(ηf–Qm) for different constant speeds, n, of the pump (see Fig. C7.2 for speeds 475,
450, 425, 400 and 375 rpm).

Figure C7.2. Characteristics of the dredging pump pumping water at


different constant values of pump speed.
(An application of affinity laws).
7.30 CHAPTER 7

The power delivered by a drive to the pump is limited by a value Win,max = 1000
kW at the maximum speed nmax = 475 rpm. This maximum-power value is reached
at the flow rate Qm = 1.244 m3/s (the Qm value is calculated using ηf =
PmanQm/Win,max combined with Eqs. C7.1 & C7.2). For higher flow rates the
diesel engine, that drives the pump, can not maintain the constant speed and the
engine operates at the constant torque. Thus the revolutions of a shaft drop if the flow
rate grows above Qm = 1.244 m3/s. The constant torque has a value equal to
60Win,max/(2πnmax) = 20 104 Nm.

The constant-torque curve of a pump-drive set is obtained from Eqs. C7.1 & C7.2 for
the condition

Q m 60 60 n2
Pman,f = Win, max where Pman,f = Pman,f , n max ,
ηf 2πn 2πn max 2
n max
Q m, n max n max η f , n max
= and =1
Qm n ηf

The condition is fulfilled for a set of [Pman,f, Qm] points (see Tab. C7.2 and Fig.
C7.3) that can be approximated (correlation coefficient Rxy = 0.9976) by the equation

Pman,f = 20950 − 43120Q m + 30690Q 2m − 7365Q 3m [kPa] (C7.3).

Figure C7.3. Characteristics of the dredging pump IHC 125-27.5-50


when pumping water, mixture of Sm = 1.4125 respectively,
in the range of maximum speed of the pump and in
the range of constant torque of a drive.
PUMP AND PIPELINE CHARACTERISTICS 7.31

Table C7.2:
Constant torque points for the performance of the pump-drive set pumping water at
different pump speeds:

n Qm Pman,f ηf
[rpm] [m3/s] [kPa] [-]
475 1.244 620.21 0.771
460 1.317 571.64 0.777
450 1.368 539.83 0.780
440 1.420 508.53 0.779
430 1.469 478.13 0.776
420 1.514 448.73 0.768
410 1.551 420.72 0.756
400 1.578 394.31 0.739
375 1.606 334.88 0.681

Pump characteristics if mixture is pumped instead of water:


The constant-speed curve is obtained (see Eqs. 7.4 & 7.5) from
 S −1 
Pman, m = Pman,f S m 1 − m (0.8 + 0.6 log d50 ) [kPa],
 Ss − 1 
thus for d50 = 0.30 mm and Ss = 2.65
Pman, m = Pman,f Sm [1 − 0.2947(S m − 1)] [kPa] (C7.4)
and for Sm = 1.4125
Pman, m = 1.2408Pman,f [kPa].

The constant-torque curve fulfils the condition

Q m 60 60 n2
Pman,f S m = Win, max where Pman,f = Pman,f , n max ,
ηf 2πn 2πn max n 2max

Q m, n max n ηf , n max
= max and = 1.
Qm n ηf

The condition is fulfilled for a set of [Pman,m, Qm] points that can be approximated
(correlation coefficient Rxy = 1.00) for Sm = 1.4125 (see Tab. C7.3 and Fig. C7.3)
by the equation
Pman, m = 2408 − 3457Q m + 2205Q 2m − 575.4Q 3m [kPa] (C7.5)

valid for flow rate values Qm > 0.743 m3/s.


7.32 CHAPTER 7

Table C7.3:
Constant torque points for the performance of the pump-drive set pumping mixture of
Sm = 1.4125 at different pump speeds:

n Qm Pman,m ηm
[rpm] [m3/s] [kPa] [-]
475 0.743 819.47 0.609
460 0.798 761.72 0.628
450 0.836 724.08 0.639
440 0.874 687.24 0.649
430 0.914 651.00 0.657
420 0.955 615.45 0.665
410 0.999 580.41 0.672
400 1.045 545.96 0.677
375 1.170 462.26 0.685
350 1.287 384.14 0.671

Remark:

The variable speed in the constant-torque regime of the engine can be approximated
(correlation coefficient Rxy = 0.9998) for Sm = 1.4125 (see Tab. C7.3) by the
equation
n = 723 − 396.1Q m + 83.08Q 2m [rpm]

valid for flow rate values Qm > 0.743 m3/s.

b. Pipeline characteristics

b.1 Major losses in a pipeline

Water: Darcy-Weisbach equation (Eq. 1.20):


λ V 2 0.011 Vm 2
2
If = f m = = 0.00112Vm [-],
D 2g 0.5 19.62
Q m 4Q m 4Q m
For Vm = =
2
=
2
= 5.09296Q m ,
A πD 3.1416x 0.5
thus
If = 0.02905 Q 2m [-] (C7.6).

Mixture: Wilson model for heterogeneous flow in horizontal pipe (Eqs. 4.16 - 17):
M = 1.7
0.45
0.35  Ss − 1 
V50 ≈ 3.93(d 50 )   = 3.93(0.30) 0.351 = 2.58 m/s.
 1.65 
PUMP AND PIPELINE CHARACTERISTICS 7.33

−M −1.7
I m − If V  I −I V 
= 0.22 m  => m f = 0.22 m 
(S m − 1)  V50  (Sm − 1)  2.58 

Im = If + 1.10199(Sm - 1) Vm−1.7 [-],


Im = If + 0.06924(Sm - 1) Q −
m
1.7 [-] (C7.7),

for Sm = 1.4125
−1.7
Im = If + 0.02856 Q m [-].

Mixture: Wilson model for heterogeneous flow in inclined pipe (Eq. 6.6):
M = 1.7
γ = 0.5
ω = 45 deg

= (cos ω)(1+ Mγ ) = (cos 45)1.85 = 0.52668


I mω − I f
I m − If
Imω = If + 0.52668(Im - If) [-] (C7.8).

Pressure drop, ∆pmajor,m [Pa], lost in a pipeline section of the length Lsection:
∆pmajor,m = Imρf gLsection.

Horizontal pipeline:
∆pmajor,hor,m = (Eq. C7.6) + (Eq. C7.7)
∆pmajor,hor,m = 0.02905 Q 2m gLhoriz + 0.06924(Sm - 1) Q −1.7
m gLhoriz
[kPa],
∆pmajor,hor,m = 0.28498 Q 2m Lhoriz + 0.67924(Sm - 1) Q −1.7
m Lhoriz [kPa]
(C7.9),
for Sm = 1.4125
∆pmajor,hor,m = 0.28498 Q 2m Lhoriz + 0.28019 Q −1.7
m Lhoriz [kPa],

∆h depth
Inclined pipeline: Lincl =
sin ω
∆pmajor,incl,m = (Eq. C7.6) + (Eq. C7.8)
∆pmajor,incl,m = 0.02905 Q 2m gLincl + 0.52668[0.06924(Sm - 1) Q − 1.7
m ]gLincl
[kPa],
∆ h depth −1.7 ∆ h depth
∆pmajor,incl,m = 0.28498 Q 2m + 0.35774(Sm - 1) Q m
sin ω sin ω
[kPa] (C7.10),
7.34 CHAPTER 7

for Sm = 1.4125
∆h depth 1.7 ∆h depth [kPa].
∆pmajor,incl,m = 0.28498 Q 2m + 0.14756 Q −
m
sin ω sin ω

b.2 Minor losses in a pipeline

Values of minor-loss coefficients for different pipeline sections:


Suction pipeline: pipe entrance: ξ = 0.4
all bends, joints etc.: ξ = 0.3
Floating pipeline: all bends, joints etc.: ξ = 0.8
Shore pipeline: all bends, joints etc.: ξ = 1.5
Total value: Σξ = 3.0

Remark:
The coefficient of minor losses for a shore pipeline is further considered constant if
the length of a pipeline varies.

Head loss due to friction in fittings (Eq. 7.9):


V2 ρ V2
H min or, m = Σξ m m [mwc], i.e. ∆p min or, m = Σξ m ρ m [Pa].
2g ρ f 2
V2
∆pminor,m = Σξ m S m = 12.97 Σξ Sm Q 2m [kPa] (C7.11),
2

for Sm = 1.4125 and Σξ = 3.0


∆pminor,m = 2.12 Vm2 = 54.96 Q 2 [kPa].
m

b3. Static head in a pipeline

The static head that must be overcome by a pump is

Hstatic = Sm ∆hdepth - Sf ∆hdepth [mwc], i.e.


∆pstatic,m = ρmg ∆hdepth - ρfg ∆hdepth [Pa],

∆pstatic,m = (Sm - 1) x 9.81 x ∆hdepth [kPa] (C7.12),

for Sm = 1.4125
∆pstatic,m = 0.4125 x 9.81 x ∆hdepth = 4.05 ∆hdepth [kPa].

b4. Total loss in an entire pipeline

∆ptotalpipe,m = (Eq. C7.9) + (Eq. C7.10) + (Eq. C7.11)+ (Eq. C7.12)


∆ptotalpipe,m = ∆pmajor,hor,m + ∆pmajor,incl,m + ∆pminor,m + ∆pstatic,m [kPa]
PUMP AND PIPELINE CHARACTERISTICS 7.35

∆ptotalpipe,m = 0.28498 Q 2m (Lhoriz+Lincl) + Q −1.7


m (Sm - 1)(0.67924Lhoriz +
0.35774Lincl) + 12.97 Σξ Sm Q 2m + 9.81 (Sm - 1)∆hdepth [kPa]

∆h depth
∆ptotalpipe,m = 0.28498 Q 2m (Lhoriz+ ) + Q−1.7
m (Sm - 1)(0.67924Lhoriz +
sin ω
∆h depth
0.35774 ) + 12.97 Σξ Sm Q 2m + 9.81 (Sm - 1)∆hdepth [kPa]
sin ω
(C7.13).

c. Working point of a pump-pipeline system

Balance:
Pman,m = ∆ptotalpipe,m
(Eq. C7.4) (Eq. C7.13)
or
(Eq. C7.5)

OUTPUTS:

The maximum attainable length of a pipeline:


For Sm = 1.4125: Lmax = 1150 m at Qm = 0.756 m3/s.

The maximum pipeline length in a stable operation regime:


For Sm = 1.4125: Lmax = 975 m at Qm = 0.897 m3/s.

Table C7.4:
Flow rates at different lengths of an entire pipeline
(see also Fig. C7.4):

Sm L Qm n
[-] [m] [m3/s] [rpm]
1.4125 400 1.245 359
1.4125 500 1.181 371
1.4125 600 1.121 383
1.4125 700 1.062 396
1.4125 800 1.004 409
1.4125 900 0.944 423
1.4125 950 0.913 431
1.4125 975 0.897 435
1.4125 1000 0.881 439
1.4125 1150 0.756 471
7.36 CHAPTER 7

Figure C7.4. Working points of a pump-pipeline system for different lengths of a


pipeline. Pump at max. speed (475 rpm) or max. torque.
Pumped mixture of constant density 1412.5 kg/m3.
Dredging depth: 15 m, pipeline diameter: 500 mm.
8.
OPERATION LIMITS OF
A PUMP-PIPELINE SYSTEM

8.1 DETERMINATION OF A REQUIRED MANOMETRIC


PRESSURE IN A PUMP-PIPELINE SYSTEM

A lay out of a dredging pipeline, properties of transported solids and required mixture
flow conditions (mixture velocity and density in a pipeline) determine a manometric
pressure that must be produced by a dredge pump. The manometric pressure required
to overcome the dredging-pipeline resistance is a pressure differential over a dredge
pump, i.e. a differential between the pressure at the pump outlet to a discharge pipe
and the pressure at the inlet to a pump connected with a suction pipe. If no geodetic
height is assumed between the pump inlet and outlet
ρ m  Vp2 − Vs2 
Pman = Pp − Ps +   (8.1).
2

For flow of mixture of density ρm the absolute suction pressure at a pump inlet (Fig.
8.1)
ρf Vs2
Ps = Patm + ρfghs,pipe - ρmg(hs,pipe – hs,pump) - ρfgHtotloss,s,m - (8.2).
2

Ps absolute suction pressure at a pump inlet [Pa]


Patm absolute atmospheric pressure [Pa]
hs,pipe depth of a suction pipe inlet below a water level [m]
hs,pump depth of a pump inlet below a water level [m]
Htotloss,s,m total head lost due to friction in a suction pipe [m]
Vs mean velocity of mixture in a suction pipe [m/s]

and the absolute discharge pressure at a pump outlet (Fig. 8.1)


ρf Vp2
Pp = ρmg(hd,pipe+hd,pump) + ρfgHtotloss,d,m + Patm - (8.3)
2

Pp absolute discharge pressure at a pump outlet [Pa]


hd,pipe vertical distance between a water level and
a discharge pipe outlet [m]
hd,pump depth of a pump outlet below a water level [m]
Htotloss,d,m total head lost due to friction in a discharge pipe [m]
Patm absolute atmospheric pressure [Pa].
Vp mean velocity of mixture in a discharge pipe [m/s]

8.1
8.2 CHAPTER 8

Patm hd,pipe

hs,pump
Ps hd,pump

hs,pipe

Pump-pipeline system:
mean velocity V
mixture density rm

Figure 8.1. Lay-out of a pump-pipeline system.

The Eqs. 8.1 – 8.3 give a relationship between the manometric pressure delivered by a
pump to mixture and the velocity of mixture in a pipeline connected to the pump. This
relationship is further dependent on solids size and concentration in a pipeline and to
a pipeline lay-out. The relationship is used to optimise the production and the energy
consumption of a pump-pipeline system during a dredging operation. A suitable range
of a system operation is confined by limits arising from processes occurring in a
dredging pipeline. An entire system does not work successfully if a dredge pump
operates outside the operational limits.

In a pump-pipeline system the flow rate of mixture must be controlled to remain


within a certain range suitable for a safe and economic operation. The flow-rate range
has a lower limit given by the deposition-limit velocity and an upper limit given by
the velocity at which pump starts to cavitate.

8.2 THE UPPER LIMIT FOR A SYSTEM OPERATION:


VELOCITY AT THE INITIAL CAVITATION OF A PUMP

A cavitation phenomenon is associated with low absolute pressure in a liquid. A


cavitation is a condition in a liquid in which the local pressure drops below the vapour
pressure and vapour bubbles (cavities) are produced. Cavitation decreases
considerably a pump efficiency and might be a reason of a damage of pump
components (pitting and corrosion). A cavitating pump provides lower manometric
head and thus the lower production of solids by a dredging pipeline. The pump
cavitation must be avoided during a pump-pipeline system operation.
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.3

8.2.1 Criterion for non-cavitational operation of a system

A pump begins to cavitate, i.e. cavitation occurs at the suction inlet to a pump
impeller, if the Net Positive Suction Head (NPSH) available to prevent pump
cavitation is smaller than NPSH required by a pump to avoid cavitation.

The no cavitation condition for a certain pump-suction pipe combination is:


(NPSH)r < (NPSH)a

in which:

The available (NPSH)a is a total available energy head over the vapour pressure at
the suction inlet to the pump during an operation at velocity Vm in a suction pipe of a
certain geometry and configuration.

Ps − Pvapour Vm2 Patm


( NPSH )a = + = + h s,pipe − H totloss,s,m − Sm ( h s,pipe − h s,pump )
ρf g 2g ρf g
Pvapour

ρf g
(8.4)

(NPSH)a Net Positive Suction Head Available [m]


Pvapour vapour pressure [Pa].

The vapour pressure of a pumped medium limits the minimum absolute pressure that
can be theoretically reached at the suction side of a pump. At this pressure the liquid
(water) is transformed into steam. The steam bubbles develop in a water flow, they
enter the pump and deteriorate its efficiency. The vapour pressure is dependent on the
temperature of a medium. For water the typical values are:

Temperature T [oC]: Vapour pressure Pvapour [kPa]:


10 1.18
20 2.27

The lay-out of a suction pipe and flow conditions in a pipe determine the absolute
suction pressure available at the pump inlet.
8.4 CHAPTER 8

Figure 8.2. Net Positive Suction Head Available on a suction inlet of a pump.

The required (NPSH)r is a minimum energy head a certain pump requires to prevent
cavitation at its inlet. This is a head value at the incipient cavitation. The (NPSH)r-Q
curve is a characteristic specific for each pump and it must be determined by tests. A
design (dimensions, shape) and an operation (specific speed) of a pump decide the
absolute suction pressure at the initial cavitation.

Ps, min − Pvapour Vm2


(NPSH )r = + (8.5)
ρf g 2g

(NPSH)r Net Positive Suction Head Required [m]


Ps,min minimum absolute suction pressure without cavitation [Pa]
Pvapour vapour pressure [Pa].

At the incipient cavitation the absolute suction pressure Ps,min at the pump inlet is
equal to the difference between the atmospheric pressure Patm and the so-called
“decisive vacuum” (Dutch: maatgevend vacuum) (Vac)d, i.e.

(Vac)d = Patm – Ps,min (8.6).

The decisive vacuum is the relative suction pressure that represents a threshold
criterion for a non-cavitational operation of a certain pump.

If a pump starts to cavitate it looses its manometric head. The (Vac)d is defined as the
vacuum at the flow rate for which the manometric head is 95 per cent of the
non-cavitational manometric head at the same pump speed (r.p.m.). The (Vac)d is
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.5

related with the flow rate in a “decisive-vacuum curve” in a H-Q plot (see Fig. 8.3).
The decisive-vacuum curve is determined by a cavitation test.

300

5%
manometrische druk (kPa)

250

200

vernauwde
zuigleidingen
150

vacuüm
maatgevend vacuüm

75
manometrische druk (kPa)

50

vernauwde onvernauwde
25
zuigleidingen zuigleiding

0 1 2 3
debiet (m3/s)
Figure 8.3. Decisive vacuum (Dutch: Maatgevend vacuum) curve of a pump.

A substitution of Eq. (8.5) to Eq. (8.6) and rearranging gives a relationship between
the (NPSH)r and the decisive vacuum (Vac)d

( Vac) d Patm Pvapour Vm2


= −( NPSH ) r + − + (8.7).
ρf g ρf g ρf g 2g

As follows from the relationship between the (HPSH)r and the decisive vacuum
(Vac)d a cavitation test gives also the (NPSH)r-Q curve, i.e. the minimum NPSH as a
function of capacity Q.
8.6 CHAPTER 8

An upper limit for the working range of a pump-pipeline system is given by points of
intersection of a pump decisive vacuum curve and a set of vacuum curves of a suction
pipe for various mixture densities. The vacuum curve of a suction pipe summarises
the friction, geodetic and acceleration heads over an entire length of the suction pipe
to the total vacuum head and relates this head with a pump capacity (see Fig. 8.4a,
8.4b and 8.4c). The total vacuum head, Vac/ρfg, is a difference between the total
absolute suction pressure head and the atmospheric pressure head

V2
Vac Patm − Ps
ρf g
=
ρf g
( )
= S m h s, pipe − h s, pump − h s, pipe + H totloss,s, m + s
2g
(8.8)

Vac vacuum; the pressure relative to atmospheric Patm [Pa]


ρf density of liquid [kg/m3]
g gravitational acceleration [m/s2]
Ps absolute suction pressure at a pump inlet [Pa]
Patm absolute atmospheric pressure [Pa]
Sm relative density of mixture (ρm/ρf) [-]
hs,pipe depth of a suction pipe inlet below a water level [m]
hs,pump depth of a pump inlet below a water level [m]
Htotloss,s,m total head lost due to friction in a suction pipe [m].
Vs mean velocity of mixture in a suction pipe [m/s]

Figure 8.3a. Decisive vacuum curve and vacuum curves of a suction pipe for flow of
mixture of various densities (schematic).
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.7

Figure 8.4b. Figure 8.4c.

Figure 8.3b. Decisive vacuum curve and vacuum curves of a suction pipe transporting
mixture of various densities from the depth 9 meter (after v.d.Berg,
1998).
Figure 8.3c. Decisive vacuum curve and vacuum curves of a suction pipe transporting
mixture of various densities from the depth 18 meter
(after v.d. Berg, 1998).

Table 8.1. Points of intersection between decisive vacuum curve and vacuum curves for
different mixture densities; the intersection points determine the maximum
production of solids attainable for given mixture density in a pump-pipeline
system lifting mixture from a certain depth (see Fig. 9.1 in Chapter 9).
8.8 CHAPTER 8

8.2.2 How to avoid cavitation

Basically, cavitation is avoided if the absolute suction pressure of a pump is


maintained above a certain critical value. An analysis of the above explained
cavitational criterion leads to the following proposals:
- to reduce the static head that the pump must overcome, i.e. to put the pump as low
as possible (see par. 8.5)
- to reduce the head lost due to flow friction, i.e. to minimise local losses and a
suction pipe length
- to increase pressure by using a larger pipe at the suction inlet of a pump (see par.
8.4).

During an operation (if the position of a pump and a geometry of a suction pipeline
can not be changed) friction losses can be reduced
- either by diminishing the mean mixture velocity in a pipeline
- or by reducing the mixture density in a suction pipeline.

8.3 THE LOWER LIMIT FOR A SYSTEM OPERATION:


VELOCITY AT THE INITIAL STATIONARY BED IN A
PIPELINE

It was shown in the previous paragraph that high head loss due to too high velocity of
mixture in a dredging installation might cause cavitation in a dredge pump and thus a
considerable reduction of production and even a damage of a pump. On the other
hand too low velocity might cause unnecessarily high head losses due to friction too.
Furthermore the too low velocity might cause a blockage of a pipeline.

8.3.1 Criterion for a deposit free operation of a system

If settling mixtures are transported a portion of solids occupies a granular bed at the
bottom of a pipeline. The part of solids that occupies the bed is strongly dependent on
the mixture velocity in a pipeline. Under the increasing velocity the thickness of the
bed tends to diminish because still more particles tend to be suspended due to
increasing turbulent intensity of a carrying liquid. However, if the velocity is
decreasing instead of increasing the bed becomes thicker and at certain velocity,
called the deposition-limit velocity (or critical velocity), the first particles in the bed
stop their sliding over a pipeline wall. If velocity decreases further the entire bed
stops and, under certain circumstances, dunes might be developed at the top of a
stationary bed. The flow becomes instable and a pipeline might be blocked. This is
more likely to happen in some “critical” parts of a pipeline as are bends, particularly
those to vertical pipe sections. A danger of blockage increases if solids occupy a
considerable part of a total pipeline volume.

Even if a blockage is not likely to happen due to relatively low concentration and/or
an absence of critical pipeline parts during a dredging operation, it is worthwhile to
watch out the deposition-limit value of the mean mixture velocity in a pipeline. A
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.9

presence of a stationary bed means that solids that are actually not transported occupy
a part of a pipeline. A stationary bed reduces a pipeline discharge area and so
tremendously increases the frictional losses. Frictional losses not far below the
deposition-limit velocity might be much higher than losses at even very high mixture
velocities. On the other hand, an operation at velocity only slightly above the
deposition-limit value is economic since the frictional loss at this velocity is usually
considerably lower than at the extremes of a velocity range. The effects of velocity on
the frictional losses and the variation of deposition-limit velocity under the different
mixture flow conditions were discussed to details in earlier chapters.

The deposition-limit velocity is for most dredging operations considered the lower
limit for a range of operational velocity. The boundary given by this velocity can be
plotted to the H-Q (or Im-Vm) plot as a curve connecting deposition-limit velocity
values for different solids concentrations in a mixture flow of certain material in a
pipeline of a certain diameter (see Fig. 8.5).

Figure 8.5. Locus curve giving a velocity at an initial stationary bed.

8.3.2 How to avoid a stationary bed in a pipeline

If the pipeline is composed of sections of different pipe sizes, the mixture flow rate
must be maintained at the level assuring a super-critical regime (Vm > Vdl) in the
largest pipe section (the section of the largest pipe diameter). Consider that in the
largest section the mixture velocity is the lowest (continuity equation) and moreover
the deposition-limit value of the mixture velocity is the highest because Vdl tends to
grow with pipe diameter.
If the solids concentration fluctuates along a pipeline, the mixture flow rate must be
maintained at the level assuring a super-critical regime in the section of an extreme
concentration. For a prediction, use the highest value of the deposition-limit velocity
from the entire range of expected solids concentrations. Vdl is sensitive to solids
8.10 CHAPTER 8

concentration, it is always better to be slightly conservative in a determination of the


appropriate value.

If during a job a dredging pipeline is prolonged, the flow rate supplied by a dredge
pump might become insufficient to assure a super-critical regime in a pipeline. Then
two solutions must be considered:
- to pump mixture at much lower concentration; this will lead to lower frictional
losses and thus higher flow rate that might be high enough to avoid a thick
stationary bed in a pipeline
- to install a booster station; this increases a manometric head provided by pumps
and increase a flow rate.

If coarser solids must be pumped than expected when a dredging installation was laid
out, the flow rate supplied by a pump might become insufficient to assure a
super-critical regime in a pipeline. Then again the above two solutions must be
considered.

8.4 EFFECT OF PIPE DIAMETER ON OPERATION LIMITS

For a certain required flow rate of mixture a larger pipeline means lower mean
velocity in comparison with a smaller pipeline. This means that there is a better
chance to pump a mixture without a danger of pump cavitation if a suction pipe is
larger. Furthermore, a pipe resistance decreases with an increasing pipe diameter.
This has also a positive effect with regard to a pump cavitation limit. On the other
hand a possibility that a stationary bed will be developed in a pipeline increases with
an increasing pipeline diameter.

A suction pipe larger than a discharge pipe is installed in some dredging installations.
The diameter of a suction pipe is chosen to be of about 50 mm larger than that of a
discharge pipe if a system is designed for transportation of fast-settling mixtures
(flows of coarse or heavy particles). An operation at the suction side of a dredging
pipeline is usually limited by a pump cavitation. For a certain mixture flow rate the
velocity in a suction pipe is low and this helps to avoid cavitation. This is more
important than a presence of a stationary bed that may possibly occur in a short
suction pipe. The presence of a stationary bed is more dangerous in a long discharge
pipe and since a cavitation is very unlikely to occur in a discharge pipeline the
deposition-limit velocity limits an operation in a discharge pipeline. It is useful to
choose smaller pipe diameter (when compared to a suction pipe) to avoid the
sub-critical regime of mixture flow. A higher frictional loss and a higher wear of a
pipeline wall of course pay this.
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.11

8.5 EFFECT OF PUMP POSITION ON OPERATION LIMITS

If a pump is placed to a lower position within a pump-pipeline system a suction pipe


becomes shorter. A geodetic height over which a mixture has to be lifted in a suction
pipe becomes smaller. In a shorter suction pipe pressure loss due to flow friction over
a suction pipe length is lower than that in a suction pipe of an original length. A
vacuum curve for a shorter suction pipe shows lower vacuum value at a certain flow
rate for mixture of certain density. Thus a cross point between a decisive vacuum
curve and a vacuum curve for a certain mixture density is reached at higher capacity
Q (compare Fig. 8.4c and Fig. 8.6). Since the total resistance (expressed by a vacuum
curve) of a suction pipe is lower in a shorter suction pipe than in a pipe of an original
length the margin occurs between a net positive suction pressure required and
available at a pump inlet. Consequently, pipe vacuum curves of mixture density
higher than is that for an original pipe still cross the decisive vacuum curve of a
pump. The mixture of density higher than in an original pipe can be pumped before an
upper limit of a pump-pipeline operation is reached (compare Tab. 8.1 and Tab. 8.2).
This means a considerable improvement of production. Therefore a submerged pump
(a pump placed on a inclined pipe below a water level) is often used on dredging
installations.

Figure 8.6. Table 8.2.

Figure 8.6. Decisive vacuum curve and vacuum curves of a suction pipe transporting
mixture of various densities from the depth 18 meter using a pump
positioned 5 meter below the water level (after v.d. Berg, 1998).

Table 8.2. Points of intersection between decisive vacuum curve and vacuum curves for
different mixture densities as shown on Fig. 8.6; the intersection points
determine production of solids at conditions in a pump-pipeline system with
a pump 5 meter below a water level (see Fig. 9.5 in Chapter 9).
8.12 CHAPTER 8

8.6 OPERATION LIMITS ON A H-Q DIAGRAM OF A PIPELINE

Fig. 8.7 shows a working range of a dredge pump that pumps, with a constant pump
speed, a mixture of a constant density through a discharge pipeline of variable length.
The maximum length of the pipeline is limited by the deposition-limit velocity. If the
pipeline would be longer the pressure delivered by the pump would not be enough to
maintain the mean velocity of mixture in the discharge pipeline above the
deposition-limit threshold. The minimum length of the discharge pipeline is limited
by the decisive vacuum of a pump. If the pipeline would be shorter, the high mean
velocity would cause so high frictional pressure losses in a suction pipe that cavitation
would occur in a suction side of the pump.

langste persleiding

constant toerental kortste persleiding


850
manometrische druk (kPa)

750

650

550

werkgebied

450
onderkritisch
bovenkritisch
100
maatgevend vacuüm

75
vacuüm (kPa)

50

zuigleiding-
25
karakteristiek

0 1 Qkritisch 2 3 Qmaatg. vacuüm


debiet (m 3/s)

Figure 8.7. Working range of a dredge pump in a pump-pipeline system. The


deposition-limit (critical) velocity in a pipeline and the decisive vacuum of
the pump limits the working range.
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.13

8.7 RECOMMENDED LITERATURE

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
de Bree, S.E.M. (1977). Centrifugaal Baggerpompen. IHC Holland.
8.14 CHAPTER 8

CASE STUDY 8.1

For this Case study the same dredging installation and the same mixture flow
conditions are considered as in Case study 7.1>

A deep dredge has a centrifugal pump on board. The heart of the pump is on the same
geodetic height as the water level. The suction and the discharge pipes are mounted to
the pump at the pump-heard level. The suction pipe of the dredge is vertical and the
discharge pipe is horizontal. Both pipes have a diameter 500 mm. The dredge pump
pumps the 0.2-mm sand from the bottom of the waterway that is 7 meter below the
water level (thus the dredging depth is 7 meter). The density of a pumped sand-water
mixture is 1400 kg/m3. The discharge pipe is 750 meter long. The pump-pipeline
installation is supposed to keep the production at 700 cubic meter of sand per hour.

1. Determine whether for the above described conditions the mean velocity through
a pipeline high enough is to avoid a stationary deposit in the pipeline.
2. Determine whether for the above described conditions the pressure at the suction
mouth of the pump is high enough to avoid cavitation. The minimum pressure for
the non-cavitational operation is considered 3 x 104 Pa.

For the calculation consider the friction coefficient of the suction/discharge pipes λ =
0.011. The following minor losses must be considered:
- the inlet to the suction pipe: ξ = 0.5,
- the 90-deg bend in suction pipe: ξ = 0.1,
- the flanges in the suction pipe: ξ = 0.05,
- the flanges in the discharge pipe: ξ = 0.25,
- the outlet from the discharge pipe: ξ = 1.0.
Additional inputs:
ρf = 1000 kg/m3
ρs = 2650 kg/m3

Inputs:

∆hdepth = 7 m
Lhor = 750 m
D = 500 mm
d50 = 0.20 mm
ρs = 2650 kg/m3, ρf = 1000 kg/m3, ρm = 1400 kg/m3
λf = 0.011, Σξ = 1.9
Qs = 700 m3/hour = 0.194 m3/s

Remark: To make a calculation simpler the effect of a pipeline roughness on frictional


losses in a pipeline is considered to be represented by a constant value of the
frictional coefficient λf , i.e. independent of variation of mean mixture
velocity.
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.15

1. Comparison of the actual velocity with the deposition-limit velocity

Mean velocity of mixture in a pipeline, Vm:


ρ − ρf 1400 − 1000
C vd = m = = 0.2424 [-],
ρs − ρf 2650 − 1000
Q 0.1944
Qm = s = = 0.802 m3/s,
C vd 0.2424
4Qm 4 x 0.802
Vm = = = 4.085 m/s.
πD 2 3.1416 x 0.5 2

Deposition-limit velocity:

Vsm = 2.9 m/s (the Wilson nomograph, Fig. 4.8)


Vcrit = 3.3 m/s (the MTI nomograph, Fig. 4.6)

The actual average velocity in a horizontal pipeline behind the pump is higher than
the deposition-limit velocity. There will be no stationary deposit at the bottom of the
pipeline for velocity 4.09 m/s.

2. Comparison of the actual suction pressure at the pump with the minimum
pressure for non-cavitational operation

Energy balance for the suction pipe (the Bernoulli equation):


Vm2
Pinlet = Psuct + ∆Pstatic + ∆Ptotloss,m + ρf
2

in which Pinlet = Patm + ∆hdepth.ρf.g,


∆Pstatic = ∆hdepth.ρm.g,
 ∆h depth  Vm2
∆Ptotloss,m =  λf + Σξsuct  ρm .
 D  2

∆Pstatic the static pressure differential between the inlet and the outlet of the
suction pipe;
∆Ptotloss,m the total pressure loss (both major and minor) over the length of a pipe;
Psuct the absolute pressure at the outlet of the suction pipe;
Patm the atmospheric pressure.

 ∆h depth  Vm2 Vm2


Psuct = Patm - ∆hdepth (ρm - ρf) g -  λ f + Σξsuct  ρm - ρf
 D  2 2

 7  4.092 4.092
Psuct = 105–7(1400-1000)9.81-  0.011 + 0.65 1400 - 1000 = 54.7 kPa.
 0.5  2 2

The absolute pressure at the suction mouth of the pump is 54.7 kPa. This is higher
than the minimum non-cavitation pressure 30 kPa. The pump will not cavitate.
8.16 CHAPTER 8

CASE STUDY 8.2

In Case study 7.2 the maximum length was determined of a pipeline connected with a
centrifugal pump operating at its maximum speed if mixture of density 1412.5 kg/m3
composed of water and a 0.3 mm sand is transported. The flow rate of pumped
mixture was determined for a pipeline of a maximum length. It is necessary to check
whether this flow rate is attainable in a system, i.e. whether it lays within an
operational range of a pump-pipeline system.

Determine the limits of an operational range of a pump-pipeline system (the system is


defined in Case study 7.2) and check whether the flow rate for the pipeline of the
maximum length lays within this range. Determine the range of pipeline lengths in
which a pump can operate at the maximum speed (475 rpm) if density of pumped
mixture is 1412.5 kg/m3.

Solution:

A. The upper limit for a system operation:

CALCULATION:

a. Pump characteristics

The decisive-vacuum curve of the IHC pump can be approximated by the equation
(Vac )d = 94.99 − 3.64Q m − 2.43Q 2m [kPa] (C8.1).

b. Suction pipeline characteristics

The vacuum-curve equation (Eq. 8.8) for a suction pipeline of


∆hs,pipe =∆hdepth = 15 m
∆hs,pump = 0 m

gets a form of Eq. C7.13. This equation is solved for the following input values
ω = 45 deg
Lhoriz,suction = 2 m

Minor-loss coefficient: Suction pipeline: pipe entrance: ξ = 0.4


all bends, joints etc.: ξ = 0.3
Total value: Σξ = 0.7

15
Vac = ∆ptotalpipe,m = 0.28498 Q 2m (2+ ) + Q−1.7
m (Sm - 1)(0.67924x2 +
sin(45)
15
0.35774 ) + 12.97x0.7 Sm Q 2m + 9.81 (Sm - 1)x15 [kPa] (C8.2).
sin(45)
For Sm = 1.4125
OPERATION LIMITS OF A PUMP-PIPELINE SYSTEM 8.17

15
Vac = ∆ptotalpipe,m = 0.28498 Q 2m (2+ ) + Q−1.7
m 0.4125(0.67924x2 +
sin(45)
15
0.35774 ) + 12.97x0.7x1.4125 Q 2m + 9.81x0.4125x15 [kPa]
sin(45)
Balance:
(Vac)d = Vac
(Eq. C8.1) (Eq. C8.2)
determines the flow rate value (Qupper) at the beginning of cavitation of a pump. This
flow-rate value is the upper limit of an operational range of a pump-pipeline system
pumping an aqueous mixture of 300-micron sand at mixture density 1412.5 kg/m3.
Only operation at flow rates lower than this threshold value will be cavitation free.

OUTPUT:

For Sm = 1.4125 the upper limit of a pump-pipeline operation is given by


Qupper = 1.115 m3/s.

B. The lower limit for a system operation:

The lower limit is given by the flow rate value (Qlower) at the critical
(deposition-limit) velocity.

CALCULATION:

The MTI correlation (Eq. 4.19) for the critical velocity gives
1
 1   0.25  6 2.65 − 1
Vcrit = 1.7 5 −  0.5   = 3.61 m / s,
 0.3   0.25 + 0.1  1.65

πD 2 πx 0.52
Qlower = Vcrit A = Vcrit = 2.92 = 0.709 m 3 / s.
4 4

OUTPUT:

For Sm = 1.4125 the lower limit of a pump-pipeline operation is given by


Qlower = 0.709 m3/s.

C. The range of lengths of an entire pipeline:


8.18 CHAPTER 8

Balance
Pman,m = ∆ptotalpipe,m
(Eq. C7.5) (Eq. C7.13)

- for the working point at Qupper = 1.115 m3/s gives the length of an entire
pipeline L = 610 meter. This is the minimal length for which the 1412.5 kg/m3
mixture can be pumped. If the pipeline becomes shorter, the flow rate tends to
increase. This would cause cavitation at the inlet of a pump. The density of
transported mixture in a short pipeline must be lowered to avoid cavitation;

- for the working point at Qlower = 0.709 m3/s gives the length of an entire
pipeline L = 1123 meter. This working point lays at the descending part of a
pipeline resistance curve. An operation at this part of the curve should be avoided,
since it is potentially instable and energy costly (see Fig. C7.4). The
recommended minimum flow rate for pumping the 1412.5 kg/m3 mixture at the
maximum speed of the pump is that the maximum length of a pipeline (see Case
study 7): Lmax = 975 meter, i.e. Qminimum = 0.897 m3/s.
9.
PRODUCTION OF SOLIDS IN A
PUMP-PIPELINE SYSTEM

9.1 PRODUCTION RANGE FOR A PUMP-PIPELINE SYSTEM

9.1.1 Maximum production attainable in a system

The maximum production theoretically attainable in a pump-pipeline system is


limited by an upper limit of a working range of a system, i.e. by cavitation of a pump.
For a system of a certain lay-out of a suction pipe the maximum production is
obtained from the points of intersection between vacuum curves for various mixture
densities in a suction pipe and a decisive vacuum curve of a pump. A point of
intersection of a vacuum curve for certain mixture density and a decisive vacuum
curve gives the maximum mixture flow rate, Qm, attainable for mixture flow of the
certain density, ρm. Then the corresponding solids flow rate, i.e. solids production, is
obtained as

ρ −ρ
Qs = Q m m f 3600 [m3/hour] (9.1).
ρs − ρf

Thus all points of intersection for various mixture densities in a suction pipe of a
certain lay-out give a set of ρm, Qm, Qs data. These provide a characteristic curve of
maximum production as a function of mixture flow rate in a pump-pipeline system
(Fig. 9.1). The curve has a maximum at a certain value of mixture density. This
indicates that pumping of mixture at the highest densities does not provide the highest
production. This is because pumping of these high-dense mixtures is possible only at
very low mixture velocities, otherwise cavitation occurs. Theoretically, the curve
approaches zero production at its boundaries. At the lower boundary (Qm = 0) a
pumped mixture is so dense that a cavitation criterion is reached already at the lowest
mixture velocities in a suction pipe. The upper boundary gives the maximal velocity
at which the decisive vacuum is reached. The maximal velocity is reached in flow of
the lowest solids concentration (approaching zero). Naturally, production of such flow
approaches zero too. More concentrated mixture flows provide more resistance and
thus the decisive vacuum is reached already at lower velocities An analysis of the
cavitation criterion shows that a dredging depth is the major factor limiting the
production by a pump-pipeline system (see Fig. 9.1).

9.1
9.2 CHAPTER 9

Figure 9.1. Production limited by decisive vacuum. Qs - Qm diagram (Qs is the


production on the vertical axis and Qm the flow rate of mixture on the horizontal
axis).

9.1.2 Minimum production acceptable in a system

An implementation of the deposition-limit (critical) velocity to a Qs - Qm curve (see


Fig. 9.2) gives an range of production values that may be reached within a working
range of a pump.

Figure 9.2. Production limited by decisive vacuum and deposition-limit velocity.


Qs - Vm diagram (this for a different installation than on Fig. 9.1).
PRODUCTION OF SOLIDS IN A PUMP-PIPELINE SYSTEM 9.3

9.2 PRODUCTION LIMITED BY A PIPELINE LENGTH

Consider a pump-pipeline installation operating at the maximum pump speed and


pumping mixture of certain maximum density attainable under the given dredging
conditions (soil type, soil condition in a pit etc.). The vacuum measured at the pump
inlet is slightly lower than a decisive vacuum of the pump. A working point of the
pump-pipeline installation determines a mixture flow rate through a pipeline and so a
solids production by the installation. A position of the working point on the pump
characteristic curve depends on the course of the pipeline resistance curve, i.e. on the
total resistance of a pipeline. The longer the pipeline the higher the total resistance
and the lower the mixture flow rate through an installation. The pipeline length is a
factor limiting the production.

9.2.1 Cases for which a change of a pipeline length requires a change in a


working range of a pump

Imagine that during the job a discharge pipeline must be considerably lengthened.
The total pipeline resistance increases considerably and a system finds a new working
point at the much lower mixture flow rate. The mixture velocity drops below a
deposition-limit velocity. Thus the new working point falls outside the acceptable
working range of the installation. To prevent the formation of a stationary deposit in a
pipeline the mixture density (i.e. concentration of solids in a pipeline) must be
reduced to diminish a total pipeline resistance and maintain the working point within
the working range of the installation. A considerable decrease in the mixture
density and the mixture velocity leads to a considerable drop in a solids
production due to the prolonging of a discharge pipeline.

Imagine that during the job a discharge pipeline must be shorten. The total pipeline
resistance diminishes and a system finds a new working point at the higher mixture
flow rate. However, an increase in the mixture velocity in a pipeline increases the
vacuum at the pump inlet so that it becomes higher than the decisive vacuum. The
pump starts to cavitate. To prevent the cavitation the mixture velocity must be
lowered by reducing a pump speed. Thus the new installed flow rate, and so
production, is approximately the same as that before a discharge-pipeline shortening.
In installations where the production is limited by a cavitation criterion the
production is virtually independent of a pipeline length.

9.2.2 Cases for which a change of a pipeline length does not require a change of a
working range of a pump

Within the working range of a pump-pipeline installation the production drops with
an increasing pipeline length but the drop is considerably smaller since only mixture
velocity is reduced if mixture is pumped at maximum pump speed. The mixture
density can be maintained at the constant level.

Figure 9.3 shows a relationship between the solids production from a pump-pipeline
installation and the length of a pipeline for various operational regimes of a pump:
9.4 CHAPTER 9

I. TOO SHORT PIPELINE (the pump tends to operate outside a normal working
range, the pump speed must be lowered to avoid cavitation = a pump operation is
limited by a pump cavitation)

II. PIPELINE DISTANCES WITHIN A NORMAL WORKING RANGE OF A


PUMP (a pump operates at maximum speed and mixture of maximum attainable
density is transported)

III. TOO LONG PIPELINE (the pump tends to operate outside a normal working
range, the mixture density must be lowered to maintain velocity above the
deposition limit in a pipeline = a pump operation is limited by the deposition-limit
velocity in a pipeline)

Figure 9.3. Qs - L diagram and corresponding Qm – L and Cvd - L diagram.

The measure of the production drop is different in different working ranges of a


pump. If the working point is found at the constant torque line of a pump the
production drop is usually smaller than for the working point at the constant speed
line of a pump. This is because the reduction in a manometric head provided by a
pump results in bigger flow rate drop for the constant-speed operation than for the
constant-torque operation of a pump.
PRODUCTION OF SOLIDS IN A PUMP-PIPELINE SYSTEM 9.5

Figure 9.4. How to select the right dredge for the job. The Qs - L diagrams for
various types of the IHC cutter suction dredge Beaver if dredging
medium sand (i.e. Soil type B).
9.6 CHAPTER 9

9.3 EFFECT OF PUMP POSITION ON PRODUCTION

It was shown in Chapter 8 (par. 8.5) that a submerged pump allows pumping mixture
of higher density than a pump placed on board of a dredge. Therefore the production
in a system with a submerged pump is limited by decisive vacuum at higher values
than in a system with an on-board pump (Fig. 9.5). The curve on Fig. 9.5 is given by
cross points of the decisive vacuum curve and the vacuum curves for different
mixture densities on Fig. 8.6 (see also Tab. 8.2).

Figure 9.5. Production limited by decisive vacuum. Qs - Qm diagram (Qs is the


production on the vertical axis and Qm the flow rate of mixture on the
horizontal axis).

9.4 SUMMARY

Production of solids by a pump-pipeline installation drops if a pipeline is prolonged.


In some cases, particularly if a discharge pipeline is short, the production is limited by
a cavitation criterion rather than by a total pipeline resistance, i.e. a pipeline length.

9.5 RECOMMENDED LITERATURE

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.
de Bree, S.E.M. (1977). Centrifugaal Baggerpompen. IHC Holland.
PRODUCTION OF SOLIDS IN A PUMP-PIPELINE SYSTEM 9.7

CASE STUDY 9

In Case studies 7.2 & 8.2 the range was determined for lengths of a pipeline
connected with a centrifugal pump that pumps the mixture of density 1412.5 kg/m3
composed of water and a 0.3 mm sand. Furthermore, the flow rate of pumped mixture
was determined for various pipeline lengths (see Tab. C7.4). From these outputs the
production of solids can be determined as a function of a length of a pipeline in a
transportation system.

Determine the production of solids by a pump-pipeline system (the system is defined


in Case study 7.2) for different lengths of a pipeline from the range that admits
pumping mixture of density 1412.5 kg/m3.

CALCULATION:

Production of solids (Eq. 3.3):

S −1 1.4125 − 1
Qs = Q m m 3600 = 3600 Q m = 900Q m [m3/hour] (C9.1).
Ss − 1 2.65 − 1

Production of in situ solids (Eq. 3.4): (for porosity n = 0.4)

Qs
Qsi = = 1500Q m [m3/hour] (C9.2).
1− n

OUTPUT:

Table C9.1:
Production of solids at different lengths of a pipeline (see also Fig. C9.1):

Sm L Speed Qm Qsi
[-] [m] [rpm] [m3/s] [m3/hour]
1.4125 610 385 1.115 1672.5
1.4125 700 396 1.062 1593.0
1.4125 800 409 1.004 1506.0
1.4125 900 423 0.944 1416.0
1.4125 975 435 0.897 1345.5
9.8 CHAPTER 9

Figure C9.1. Qs – L diagram for a system pumping mixture


of constant density Sm = 1.4125.
(rpm as a label in the diagram).
PRODUCTION OF SOLIDS IN A PUMP-PIPELINE SYSTEM 9.9

CASE STUDY from the VBKO course:

Production of solids by a pump-pipeline system pumping mixture of certain


solids.

1. Working points of a system for various densities of pumped mixture. The length of
a pipeline is 2000 meter.

2.
a. Working points of a system for various length of a pipeline. The density of pumped
mixture is 1100 kg/m3.
9.10 CHAPTER 9

b. Working points of a system for various length of a pipeline. The density of pumped
mixture is 1300 kg/m3.

c. Working points of a system for various length of a pipeline. The density of pumped
mixture is 1500 kg/m3.

3. Qs - L diagram composed of working points of a system for various length of a


pipeline and various densities of pumped mixture.
PRODUCTION OF SOLIDS IN A PUMP-PIPELINE SYSTEM 9.11

4. Graphical determination of production Qs (marked as P in Fig.) for various mixture


flow rates and densities and for various pipeline lengths.

5. Operational range for mixture flow rate and production confined by the
stationary-deposition limit and the pipeline-resistance limit for a system with a
pipeline of a length L2.
9.12 CHAPTER 9
10.
SYSTEMS WITH PUMPS IN SERIES

10.1 CHARACTERISTICS OF A SYSTEM WITH PUMPS IN


SERIES

An additional pump (called a booster) is installed to a pump-pipeline transportation


system if an original dredge pump is not sufficient to provide a required production.
This is the case if the manometric head provided by a pump is not high enough to
overcome a pipeline resistance and install an appropriate value of mixture velocity in
a pipeline. Typically, a pipeline resistance grows because a pipeline is lengthened (a
larger dredging depth or a larger distance from a dredge to a deposit site) or a coarser
soil must be transported.

Pumps operating in series should be compatible, i.e.:


- to be designed for the same working range of flow rates
- to have similar shape of the Q-H curve
- to have virtually identical position of the nominal full-torque
- to have passages and connections of similar dimensions (passages of boosters
should be at least of the same size as that of a first pump).

The total manometric head provided by a set of pumps in series is equal to the sum of
manometric heads of particular pumps for a given flow rate. A new H-Q curve for an
installation composed from a set of pumps and a pipeline is a result of this summation
(see Fig. 10.1).

For manometric head, Hman, and manometric pressure, Pman, by a pump at a given
mixture flow rate Qm
n n
H man, total = ∑ H man,i , i.e. Pman, total = ∑ Pman,i (10.1).
i =1 i =1

For input power of a pump, Win,


n
Win, total = ∑ Win,i (10.2)
i =1

and for the pump efficiency, η,


n
Q m ∑ Pman,i
ηtotal = i =1 (10.3).
n
∑ Wi
i =1

10.1
10.2 CHAPTER 10

Figure 10.1. H – Q characteristic for two pumps in series.

The working point of a twin-pump - pipeline installation is given by a point of


intersection between the twin-pump characteristic and the pipeline characteristic On
Fig. 10.2 different working points are found for a set of two pumps and a pipeline at
various delivery distances. On the figure the same is plotted also for an installation
composed of a single pump and a pipeline at various delivery distances.

Figure 10.2. Working points for different pump-pipeline installations.


SYSTEMS WITH PUMPS IN SERIES 10.3

Not only a delivery distance grows if an additional pump is installed, also the
production increases significantly as can be seen on Fig. 10.3.

Figure 10.3. Production as a function of a pipeline length for installations with a


single pump, a set of two pumps respectively.

10.2 OPERATIONAL RULES FOR PUMPS IN SERIES

If a set of pumps in series is installed to a dredging installation the first pump serves
usually as “the suction pump” (Dutch: zuigpomp) and the pumps behind the suction
pump as “the delivery pumps” (Dutch: perspompen). A suction pump is supposed to
handle low suction pressure at its inlet (the pump must have a high decisive vacuum)
and provides only a low delivery pressure. Delivery pumps operate at higher suction
pressure values and provide higher manometric head than a suction pump. Boosters
(i.e. delivery pumps) should operate at suction pressure heads not lower than 10 meter
water column (mwc), i.e. at suction heads not lower than is the atmospheric pressure
head.
A typical example of a set of pumps is an installation with a submerged pump on a
dredge ladder and one or two on-board dredge pumps. The submerged pump acts as a
suction pump and the on-board dredge pump(s) as delivery pump(s) for a long
delivery pipeline.
10.4 CHAPTER 10

10.3 CONTROL OF A SYSTEM WITH PUMPS IN SERIES

An operation of an installation composed of a pipeline and pumps in series should be


controlled to maintain a dredging process as steady as possible. The operation control
must prevent an on-time magnification of velocity fluctuations in a pipeline that
usually develop if mixture is pumped of fluctuating density and solids properties.
Large velocity fluctuations might result to water hammer. Water hammer is
associated with huge pressure gradients over sections of a transport installation that
may cause a total damage of pumps and pipes. To prevent water hammer in
uncontrolled installations it is necessary to avoid a situation that a delivery pump
performs a suction function (i.e. a suction pressure at the inlet to a delivery pump is
lower than the atmospheric pressure). Such a situation may occur, for instance, if a
suction pump would collapse. The suction pressure at the delivery pump would drop,
possibly even below a non-cavitation limit. A very low pressure in front of a delivery
pump increases considerably a risk of water hammer in a system.
Mixture-flow interruptions generated by excavating and suction processes might be a
further reason for a development of water hammer in a pipeline with pumps in series.
A suddenly low supply of mixture to a pipeline in front of a booster can cause that the
booster starts to suck (a result of a low pipeline resistance), i.e. low pressure is
generated at the inlet to a booster pump. This causes an acceleration of mixture
masses in the pipeline in front of the pump and deceleration of mixture masses in a
pipeline behind the pump. If a supply restores again, large masses of mixture are
subjected to acceleration with collisions between mass’s and pump & pipeline
components as results. Further collisions occur between accelerated and decelerated
masses in a pipeline. The collisions cause water hammer effects.

If an installation operation is controlled a delivery pump may perform a suction


function without disastrous effects. A certain control procedure is required also to
start or stop an installation operation. An installation is equipped with various
components, instruments and controllers to prevent water hammer effects.

Usually, speed of a pump controls the mixture flow in a pipeline. A delivery pump is
regulated to a certain velocity set point in order to maintain stable flow. If the mean
velocity in a pipeline grows above a set point value the controller decreases the rpm
of the pump. If the velocity drops below the set point value, the controller increases
the rpm. The speed of pumps is also regulated to prevent cavitation at the suction side
of pimps (the suction pump in particular). If the suction pressure at the inlet of a pump
drops below a non-cavitation limit the controller decreases the pump rpm in order to
decrease the velocity and so increase pressure in a pipeline in front of a pump.

10.4 LOCATION OF BOOSTERS ALONG A LONG DREDGING


PIPELINE

Two most important conditions must be satisfied if a location of a booster is sought:


- a suction pressure at the booster-pump inlet must be sufficiently high (higher than
atmospheric pressure, i.e. approximately 1 bar or 10 mwc)
- a discharge pressure at the booster-pump outlet may not be higher than a
maximum pressure that the pump components can stand.
SYSTEMS WITH PUMPS IN SERIES 10.5

Thus a pressure distribution along a long pipeline is of the greatest importance for a
determination of a suitable booster location.

Figure 10.4. Pressure head distribution along a dredging pipeline for an installation
without a booster station.

Figure 10.5. Pressure head distribution along a dredging pipeline for an installation
with a booster station.

10.5 RECOMMENDED LITERATURE

de Bree, S.E.M. (1977). Centrifugaal Baggerpompen. IHC Holland.


10.6 CHAPTER 10
11.
LITERATURE ON DREDGING PROCESSES

11.1 SURVEY OF ALL DREDGING PROCESSES

Bray, R.N., Bates, A.D. & Land, J.M. (1997). Dredging. A Handbook for Engineers.
Arnold. (ISBN 0 340 54524 0)

Herbich, J.B. (1992). Handbook of Dredging Engineering. McGraw-Hill. (ISBN 0 07


028360 5)

van der Schrieck, G.L.M. (1997). Baggertechniek. College diktaat f14, TU Delft,
Faculteit der Civiele Techniek.

Turner, T. (1996). Fundamentals of Hydraulic Dredging. ASCE New York. (ISBN 0


7844 0147 0)

VBKO course. Voortgezette opleiding uitvoering baggerwerken. Centrale


Baggerbedrijf.

Yell, D. & Riddell, J. (1995). Dredging. Telford. (ISBN 0 7277 2049 X)

11.2 HYDRAULIC TRANSPORT (PUMPS AND PIPELINES)

van den Berg, C.H. (1998). Pipelines as Transportation Systems. European Mining
Course Proceedings, MTI.

de Bree, S.E.M. (1977). Centrifugaal Baggerpompen. IHC Holland.

Brown, N.P. & Heywood, N.I. (1991) Slurry Handling Design of Solid-Liquid
Systems. Elsevier Applied Science. (ISBN 1 85166 645 1)

Jacobs, B.E.A. (1991). Design of Slurry Transport Systems. Elsevier Applied


Science. (ISBN 1 85166 634 6)

Shook, C.A. & Roco, M.C. (1991). Slurry Flow. Principles and Practice.
Butterworth-Heinemann. (ISBN 0 7506 9110 7)

Vlasblom W.J. (1999). Hydraulic Transport of Soil-Water Mixture. Pumps and


Systems. Lecture notes wb3414, TU Delft, Sub-faculty of Mechanical Engineering.

11.1
11.2 CHAPTER 11

Wilson, K.C., Addie G.R., Sellgren, A. & Clift, R. (1997). Slurry Transport Using
Centrifugal Pumps. Blackie Academic & Professional. (ISBN 0 7514 0408 X)

11.3 PERIODICALS

Dredging and Port Construction. International Magazine for Port Engineering.


Published monthly by DMG Business Media Ltd., United Kingdom. (ISSN
0264-4835)

International Dredging Review. Published eight times a year by IDR, USA. (ISSN
0737-8181)

Journal of Dredging Engineering. Published by WEDA (Western Dredging


Association).

Port Engineering Management. International Journal of Dredging, Port


Development & Ocean Technology. Published bi-monthly by Lands Services, United
Kingdom. (ISSN 0264-87833)

Ports and Dredging. Published occasionally by IHC Holland. (ISSN 0166-5766)

Terra et Aqua. International Journal on Public Works & Waterway Developments.


Published quarterly by the IADC, The Netherlands. (ISSN 0376-6411)

World Dredging Mining and Construction. Published monthly by Placer


Management Corp., USA. (ISSN 1045-0343)