Michael V.
Lurie
Modeling of Oil Product and
Gas Pipeline Transportation
Related Titles
Olah, G. A., Goeppert, A., Prakash, G. K. S.
Beyond Oil and Gas: The Methanol Economy
2006
ISBN: 9783527312757
Olah, G. A., Prakash, G. K. S. (eds.)
Carbocation Chemistry
2004
ISBN: 9780471284901
Sheldrake, A. L.
Handbook of Electrical Engineering – For Practitioners in the
Oil, Gas and Petrochemical Industry
2003
EBook
ISBN: 9780470864784
Michael V. Lurie
Modeling of Oil Product and
Gas Pipeline Transportation
The Author
Prof. Dr. Michael V. Lurie
Russian State University
of Oil and Gas
Moscow, Russian Federation
Translation
Emmanuil G. Sinaiski
Leipzig, Germany
Cover Picture
TransAlaska Pipeline
All books published by WileyVCH are
carefully produced. Nevertheless, authors,
editors, and publisher do not warrant the
information contained in these books,
including this book, to be free of errors.
Readers are advised to keep in mind that
statements, data, illustrations, procedural
details or other items may inadvertently be
inaccurate.
Library of Congress Card No.: applied for
British Library CataloguinginPublication Data
A catalogue record for this book is available
from the British Library.
Bibliographic information published by
the Deutsche Nationalbibliothek
Die Deutsche Nationalbibliothek lists this
publication in the Deutsche National
bibliograﬁe; detailed bibliographic data are
available in the Internet at
<http://dnb.dnb.de>.
2008 WILEYVCH Verlag GmbH & Co.
KGaA, Weinheim
All rights reserved (including those of
translation into other languages). No part of
this book may be reproduced in any form – by
photoprinting, microﬁlm, or any other
means – nor transmitted or translated into a
machine language without written permission
from the publishers. Registered names,
trademarks, etc. used in this book, even when
not speciﬁcally marked as such, are not to be
considered unprotected by law.
Printed in the Federal Republic of Germany
Printed on acidfree paper
Composition Laserwords, Chennai
Printing Strauss GmbH, M¨ ovlenbach
Bookbinding Litges & Dopf GmbH,
Heppenheim
ISBN: 9783527408337
V
In memory of the Teacher – academician Leonid I. Sedov
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
VII
Foreword
This book is dedicated ﬁrst and foremost to holders of a master’s degree and
postgraduate students of oil and gas institutes who have decided to specialize
in the ﬁeld of theoretical problems in the transportation of oil, oil products and
gas. It contains methods of mathematical modeling of the processes taking
place in pipelines when transporting these media.
By the term mathematical model is understood a system of mathematical
equations in which framework a class of some processes could be studied. The
solution of these equations provides values of parameters without carrying out
model and, especially, full scale experiments.
Physical laws determining the dynamics of ﬂuids and gases in pipes are
presented. It is then shown how these laws are transformed into mathematical
equations that are at the heart of one or another mathematical model. In
the framework of each model, are formulated problems with the aim of
investigating concrete situations. In doing so there are given methods of its
solution.
The book is selfsufﬁcient for studying the subject but the text is outlined in
such a way that it impels the reader to address oneself to closer acquaintance
of considered problem containing in special technical literature.
Professor Michael V. Lurie
Moscow
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
IX
Contents
Dedication Page V
Foreword VII
Preface XIII
List of Symbols XV
1 Fundamentals of Mathematical Modeling of OneDimensional
Flows of Fluid and Gas in Pipelines 1
1.1 Mathematical Models and Mathematical Modeling 1
1.1.1 Governing Factors 3
1.1.2 Schematization of OneDimensional Flows of Fluids and Gases
in Pipelines 4
1.2 Integral Characteristics of Fluid Volume 5
1.3 The Law of Conservation of Transported Medium Mass.
The Continuity Equation 7
1.4 The Law of Change in Momentum. The Equation of Fluid Motion 9
1.5 The Equation of Mechanical Energy Balance 11
1.5.1 Bernoulli Equation 15
1.5.2 Input of External Energy 16
1.6 Equation of Change in Internal Motion Kinetic Energy 17
1.6.1 Hydraulic Losses (of Mechanical Energy) 18
1.6.2 Formulas for Calculation of the Factor λ(Re, ε) 20
1.7 Total Energy Balance Equation 22
1.8 Complete System of Equations for Mathematical Modeling
of OneDimensional Flows in Pipelines 29
2 Models of Transported Media 31
2.1 Model of a Fluid 31
2.2 Models of Ideal and Viscous Fluids 32
2.3 Model of an Incompressible Fluid 34
2.4 Model of Elastic (Slightly Compressible) Fluid 34
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
X Contents
2.5 Model of a Fluid with Heat Expansion 34
2.6 Models of NonNewtonian Fluids 36
2.7 Models of a Gaseous Continuum 38
2.7.1 Model of a Perfect Gas 39
2.7.2 Model of a Real Gas 39
2.8 Model of an Elastic Deformable Pipeline 42
3 Structure of Laminar and Turbulent Flows in a Circular Pipe 45
3.1 Laminar Flow of a Viscous Fluid in a Circular Pipe 45
3.2 Laminar Flow of a NonNewtonian Power Fluid in a Circular Pipe 47
3.3 Laminar Flow of a ViscousPlastic Fluid in a Circular Pipe 49
3.4 Transition of Laminar Flow of a Viscous Fluid to Turbulent Flow 51
3.5 Turbulent Fluid Flow in a Circular Pipe 52
3.6 A Method to Control Hydraulic Resistance by Injection of
AntiTurbulent Additive into the Flow 62
3.7 Gravity Fluid Flow in a Pipe 65
4 Modeling and Calculation of Stationary Operating Regimes of Oil
and Gas Pipelines 73
4.1 A System of Basic Equations for Stationary Flow of an
Incompressible Fluid in a Pipeline 73
4.2 Boundary Conditions. Modeling of the Operation of Pumps and
OilPumping Stations 75
4.2.1 Pumps 75
4.2.2 OilPumping Station 78
4.3 Combined Operation of Linear Pipeline Section and Pumping
Station 81
4.4 Calculations on the Operation of a Pipeline with Intermediate
OilPumping Stations 84
4.5 Calculations on Pipeline Stationary Operating Regimes in
Fluid Pumping with Heating 87
4.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 92
4.6.1 Distribution of Pressure in Stationary Gas Flow in a GasPipeline 94
4.6.2 Pressure Distribution in a GasPipeline with Great Difference
in Elevations 96
4.6.3 Calculation of Stationary Operating Regimes
of a GasPipeline (General Case) 97
4.6.4 Investigation of Thermal Regimes of a GasPipeline Section 98
4.7 Modeling of Blower Operation 100
5 Closed Mathematical Models of OneDimensional NonStationary
Flows of Fluid and Gas in a Pipeline 109
5.1 A Model of NonStationary Isothermal Flow of a Slightly Compressible
Fluid in a Pipeline 109
5.2 A Model of NonStationary Gas Flow in a Pipeline 112
Contents XI
5.3 NonStationary Flow of a Slightly Compressible Fluid
in a Pipeline 113
5.3.1 Wave Equation 113
5.3.2 Propagation of Waves in an Inﬁnite Pipeline 115
5.3.3 Propagation of Waves in a SemiInﬁnite Pipeline 117
5.3.4 Propagation of Waves in a Bounded Pipeline Section 119
5.3.5 Method of Characteristics 121
5.3.6 Initial, Boundary and Conjugation Conditions 124
5.3.7 Hydraulic Shock in Pipes 127
5.3.8 Accounting for Virtual Mass 134
5.3.9 Hydraulic Shock in an Industrial Pipeline Caused by Instantaneous
Closing of the Gate Valve 135
5.4 NonIsothermal Gas Flow in GasPipelines 138
5.5 Gas Outﬂow from a Pipeline in the Case of a Complete Break
of the Pipeline 146
5.6 Mathematical Model of NonStationary Gravity Fluid Flow 149
5.7 NonStationary Fluid Flow with Flow Discontinuities
in a Pipeline 152
6 Dimensional Theory 157
6.1 Dimensional and Dimensionless Quantities 157
6.2 Primary (Basic) and Secondary (Derived) Measurement Units 158
6.3 Dimensionality of Quantities. Dimensional Formula 159
6.4 Proof of Dimensional Formula 161
6.5 Central Theorem of Dimensional Theory 163
6.6 DimensionallyDependent and DimensionallyIndependent
Quantities 164
6.7 Buckingham Theorem 168
7 Physical Modeling of Phenomena 173
7.1 Similarity of Phenomena and the Principle of Modeling 173
7.2 Similarity Criteria 174
7.3 Modeling of Viscous Fluid Flow in a Pipe 175
7.4 Modeling Gravity Fluid Flow 176
7.5 Modeling the Fluid Outﬂow from a Tank 178
7.6 Similarity Criteria for the Operation of Centrifugal Pumps 179
8 Dimensionality and Similarity in Mathematical Modeling
of Processes 183
8.1 Origination of Similarity Criteria in the Equations of a
Mathematical Model 183
8.2 OneDimensional NonStationary Flow of a Slightly Compressible
Fluid in a Pipeline 184
8.3 Gravity Fluid Flow in a Pipeline 186
8.4 Pipeline Transportation of Oil Products. Batching 187
XII Contents
8.4.1 Principle of Oil Product Batching by Direct Contact 188
8.4.2 Modeling of Mixture Formation in Oil Product Batching 189
8.4.3 Equation of Longitudinal Mixing 192
8.4.4 SelfSimilar Solutions 194
References 199
Appendices 201
Author Index 205
Subject Index 207
XIII
Preface
This book presents the fundamentals of the mathematical simulation of
processes of pipeline transportation of oil, oil products and gas. It is shown
how the basic laws of mechanics and thermodynamics governing the ﬂow of
ﬂuids and gases in pipelines are transformed into mathematical equations
which are the essence of a certain mathematical model and, in the framework
of a given physical problem, appropriate mathematical problems are formulated
to analyze concrete situations.
The book is suitable for graduate and postgraduate students of universities
having departments concerned with oil and gas and to engineers and research
workers specializing in pipeline transportation.
Beginners will ﬁnd in this book a consecutive description of the theory and
mathematical simulation methods of stationary and nonstationary processes
occurring in pipelines. Engineers engaged in the design of and calculations on
pipelines will ﬁnd a detailed theoretical and practical textbook on the subject
of their work. Graduate and postgraduate students and research workers will
become acquainted with situations in the theory and methods in order to
generalize and develop them in the future.
The author of the book, Professor Dr. M. Lurie, is a great authority in Russia
in the ﬁeld of the hydromechanics of oil and gas pipeline transportation.
Prof. Emmanuil Sinaiski
Leipzig
‘‘. . . No human investigation could be referred to as true when it is not supported
by mathematical proof’’
Leonardo da Vinci
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
XV
List of Symbols
Symbol Deﬁnition
a radius of the ﬂow core
a dimensionless constant
a parameter of the (Q − H) characteristic
A proportionality factor
A
+
value of parameter A to the left of the discontinuity front
A
−
value of parameter A to the right of the discontinuity front
[A] = A
+
− A
−
jump of parameter A at the discontinuity front
dA
in
elementary work of internal force
dA
ex
elementary work of external force
b parameter of the (Q − H) characteristic
c velocity of wave propagation in a pipeline
c sound velocity in gas
C
2
integration constant
C
f
friction factor
C
p
heat capacity at constant pressure
C
Sh
Chezy factor
C
v
heat capacity at constant volume
cP centipoise, 0.01 P
cSt centistokes, 0.01 St = 10
−6
m
2
s
−1
d pipeline internal diameter
d diameter increment
d
0
nominal internal diameter of pipeline; cylinder internal
diameter
D pipeline external diameter
D velocity of hydraulic shock wave propagation in a pipeline
D velocity of discontinuity front propagation in the positive
direction of the xaxis
D
im
diameter of impeller
D
p
diameter of pump impeller
D
∗
Joule–Thompson factor
e
in
internal energy density; speciﬁc internal energy
e
kin
kinetic energy density; speciﬁc kinetic energy
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
XVI List of Symbols
E elastic modulus in extension and compression,
Young’s modulus
E
in
internal energy
E
kin
kinetic energy
Ei(z) Euler function
˜
f
1
dimensionless factor
f
τ
(Q)− friction force
F restoring force
dF
n
elementary force
Fr Froude number
g acceleration due to gravity
g
0
, g
1
dimensionless constants
h piezometric head
h(S) depth of pipeline crosssection ﬁlling with ﬂuid
h
c
head losses in station communications
h
cr
critical depth
h
π.
head before PLP
h
n
normal depth of gravity ﬂow in the pipe
H head
H differential head
H = F(Q) headdischarge (Q − H) characteristic of a pump
H
1
hydraulic head
˜
H
1
hydraulic head
He Hedstroem number
i hydraulic gradient
i
0
hydraulic gradient
I momentum
I Ilyushin number
I
1
, I
2
Riemann invariants
J gas enthalpy
k factor of string elasticity
k factor of power, Ostwald ﬂuid
k parameter of nonNewtonian ﬂuid
κ factor; heattransfer factor; empirical factor 1/K
κ Karman constant
k dimensionless constant
k kinematic consistency
K heat transfer factor
K elastic modulus of ﬂuid, Pa
K factor of longitudinal mixing of oil product
l
c
length of the mixture region
L length of a pipeline or a pipeline section
˙
M mass ﬂow rate
˙
M
0
initial mass ﬂow rate
OPS oilpumping station
List of Symbols XVII
n factor of power, Ostwald ﬂuid
n exponent in Ostwald rheological law
n exponent
n number of revolutions of centrifugal blower shaft
n unit normal vector
n
0
nominal number of revolutions of blower shaft
n
in
speciﬁc power of internal friction forces
N power consumption, kW
N
mech
power of external mechanical devices
N
us
useful power of mechanical force acting on gas
N/ρ
e
speciﬁc power
p pressure
p difference between internal and external pressures, pressure
drop
p
0
nominal pressure, initial pressure, normal pressure, pressure
at the beginning of the pipeline section
p
en
pressure of gas at the entrance of compressor station
and blower
p
cr
critical pressure
p
ex
external pressure; pressure at initial crosssection
of the pipeline section
p
in
internal pressure
p
L
pressure at the end of the pipeline section
p
l
, p
e
, p
π
pressure at the pressure line of pumps (PLP)
p
r
reduced pressure
p
st
standard pressure, p
st
= 101 325 Pa
p
u
pressure before oilpumping station
p
u
head before station
p
v
saturated vapor tension (pressure)
[p] pressure jump
[p
inc
] incident pressure wave amplitude
[p
reﬂ
] reﬂected pressure wave amplitude
[p
trans
] transmitted pressure wave amplitude
P poise, 0.1 kg m s
−1
P
s
wetted perimeter
Pa pascal (SI unit), kg m
−1
s
−2
Pe Peclet number
q
h
speciﬁc heat ﬂux
q
ex
heat inﬂow (q
ex
> 0) to gas; heat outﬂow (q
ex
< 0) from gas
q
M
speciﬁc mass ﬂow rate
q
n
external heat ﬂux
Q volume ﬂow rate; ﬂuid ﬂow rate
Q
e
ﬂow rate of gas at the entrance to the compressor station
Q
k
commercial ﬂow rate of gas
Q
M
mass ﬂow rate
XVIII List of Symbols
Q
v
volume ﬂow rate of gas at pipeline crosssection
r radial coordinate
r
0
pipeline radius
R gas constant (R = R
0
/µ
g
)
R
0
universal gas constant
R
h
hydraulic radius
R
im
radius of the impeller
R
r
reduced gas constant
Re Reynolds number
Re
cr
critical Reynolds number
Re
∗
generalized Reynolds number
S area of a crosssection; area of pipeline crosssection part ﬁlled
with ﬂuid
S
0
area of pipeline crosssection; nominal (basic) area
St stokes, 10
−4
m
2
s
−1
t time
T absolute temperature
T
0
nominal temperature; initial temperature; temperature of ﬂuid
at normal condition
T
av
average temperature over pipeline section length
T
cr
critical temperature
T
B
temperature of gas at the entrance to the compressor station
T
ex
temperature of external medium
T
L
temperature at the end of pipeline section
T
m
mean temperature
T
r
reduced temperature
T
st
standard absolute temperature
u(y) velocity distribution over crosssection
u
max
maximum value of velocity
u
w
ﬂuid velocity at pipe wall
u
∗
dynamic velocity
v velocity averaged over crosssection
v mean ﬂow rate velocity
v
cr
critical velocity
V volume
[v] ﬂuid velocity jump
w acceleration
x coordinate along the pipeline axis
x
1
coordinate of gravity ﬂow section beginning
x
2
coordinate of gravity ﬂow section end
y coordinate transverse to the pipeline axis; directionof a normal
to the elementary surface dσ
z(x) elevation level of a pipeline crosssection x
(z
1
− z
2
) geometrical height differences of sections 1 and 2
Z overcompressibility factor
List of Symbols XIX
Z
av
average overcompressibility factor
Z = Z
r
reduced gas overcompressibility
α angle of inclination of the pipeline axis to the horizontal
α
κ
, ¯ α
κ
factors
α
v
volume expansion factor
α
T
thermal expansion factor
β compressibility factor
γ adiabatic index
γ ratio between the hydraulic gradient of pipeline section
completely ﬁlled with ﬂuid and the absolute value of the
gravity ﬂow section with slope α
p
to the horizontal
˙ γ shear rate, s
−1
δ pipeline wall thickness
absolute equivalent roughness; roughness of wall surface
ε relative roughness; compression ratio; thickness ratio
ς(t) local resistance factor
η dimensionless radius
η(%) efﬁciency
θ function of temperature; concentration; parameter of over
compressibility factor; parameter of state equation of real gas;
concentration of antiturbulent additive
λ hydraulic resistance factor
λ
eff
effective factor of hydraulic resistance
µ dynamic viscosity factor kg m
−1
s
−1
µ
g
molar mass of gas; molecular weight
µ
t
turbulent dynamic viscosity
˜ µ apparent viscosity of power Ostwald ﬂuid
δ kinematic viscosity factor m
2
s
−1
ν
0
kinematic viscosity factor at temperature T
0
ν
1
kinematic viscosity factor at temperature T
1
ν
P
Poisson ratio
ν
t
turbulent kinematic viscosity
ξ factor of volumetric expansion, K
−1
; selfsimilar coordinate;
dimensionless coordinate
dimensionless parameter; similarity criterion
(x) initial pressure distribution
ρ density
ρ
−
, v
−
, p
−
, S
−
values of parameters before hydraulic shock wave
ρ
+
, v
+
, p
+
, S
+
values of parameters after hydraulic shock wave
ρ
0
nominal density; ﬂuid density at p
0
; density of ﬂuid under
normal conditions
ρ
st
gas density under standard conditions
σ area of suction branch pipe crosssection; hoop stress; degree
of pipe ﬁlling; circumferential stress
XX List of Symbols
dσ elementary surface area; surface element
τ tangential (shear) stress
tangential friction stress
τ
0
critical (limit) shear stress
τ
w
tangential (shear) stress at the pipeline internal surface
υ speciﬁc volume
υ
cr
critical speciﬁc volume
ϕ angle of inclination of a straight line to the abscissa; central
angle
(x) initial ﬂuid velocity distribution
ω frequency of rotor rotation; angular velocity of impeller
rotation
1
1
Fundamentals of Mathematical Modeling of
OneDimensional Flows of Fluid and Gas in Pipelines
1.1
Mathematical Models and Mathematical Modeling
Examination of phenomena is carried out with the help of models. Each model
represents a deﬁnite schematization of the phenomenon taking into account
not all the characteristic factors but some of them governing the phenomena
and characterizing it from some area of interest to the researcher.
For example, to examine the motion of a body the material point model is
often used. In such a model the dimensions of the body are assumed to be
equal to zero and the whole mass to be concentrated at a point. In other words
we ignore a lot of factors associated with body size and shape, the material
from which the body is made and so on. The question is: to what extent would
such a schematization be efﬁcient in examining the phenomenon? As we all
know such a body does not exist in nature. Nevertheless, when examining the
motion of planets around the sun or satellites around the earth, and in many
other cases, the material point model gives brilliant results in the calculation
of the trajectories of a body under consideration.
In the examination of oscillations of a small load on an elastic spring we
meet with greater schematization of the phenomenon. First the load is taken
as a point mass m, that is we use the material point model, ignoring body
size and shape and the physical and chemical properties of the body material.
Secondly, the elastic string is also schematized by replacing it by the socalled
restoring force F = −k · x, where x(t) is the deviation of the material point
modeling the load under consideration from the equilibrium position and k is
the factor characterizing the elasticity of the string. Here we do not take into
account the physicalchemical properties of the string, its construction and
material properties and so on. Further schematization could be done by taking
into account the drag arising from the air ﬂow around the moving load and
the rubbing of the load during its motion along the guide.
The use of the differential equation
m
d
2
x
dt
2
= −k · x, (1.1)
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
2 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
expressing the second Newtonian law is also a schematization of the
phenomenon, since the motion is described in the framework of Euclidian
geometry which is the model of our space without taking into account the
relativistic effects of the relativity theory.
The fact that the load motion can begin from an arbitrary position with an
arbitrary initial velocity may be taken into account in the schematization by
specifying initial conditions at
t = 0 : x = x
0
; v =
_
dx
dt
_
0
= v
0
. (1.2)
Equation (1.1) represents the closed mathematical model of the considered
phenomenon and when the initial conditions are included (1.2) this is the
concrete mathematical model in the framework of this model. In the given case
we have the socalled initial value (Cauchy) problem allowing an exact solution.
This solution permits us to predict the load motion at instants of time t > 0
and by so doing to discover regularities of its motion that were not previously
evident. The latest circumstance contains the whole meaning and purpose of
mathematical models.
It is also possible of course to produce another more general schematization
of the same phenomenon which takes into account a great number of
characteristic factors inherent to this phenomenon, that is, it is possible, in
principle, to have another more general model of the considered phenomenon.
This raises the question, how can one tell about the correctness or
incorrectness of the phenomenon schematization when, from the logical
point of view, both schematizations (models) are consistent? The answer is:
only from results obtained in the framework of these models. For example,
the aboveoutlined model of load oscillation around an equilibrium position
allows one to calculate the motion of the load as
x(t) = x
0
· cos
__
k
m
· t
_
+
_
m
k
·v
0
· sin
__
k
m
· t
_
having undamped periodic oscillations. How can one evaluate the obtained
result? On the one hand there exists a time interval in the course of which
the derived result accords well with the experimental data. Hence the model
is undoubtedly correct and efﬁcient. On the other hand the same experiment
shows that oscillations of the load are gradually damping in time and come
to a stop. This means that the model (1.1) and the problem (1.2) do not take
into account some factors which could be of interest for us, and the accepted
schematization is inadequate.
Including in the number of forces acting on the load additional forces,
namely the forces of dry −f
0
· sign(˙ x) and viscous −f
1
· ˙ x friction (where the
symbol sign(˙ x) denotes the function ˙ x− sign equal to 1, at ˙ x > 0; equal to −1,
at ˙ x < 0 and equal to 0, at ˙ x = 0), that is using the equation
m
d
2
x
dt
2
= −k · x −f
0
· sign(˙ x) −f
1
· ˙ x (1.3)
1.1 Mathematical Models and Mathematical Modeling 3
instead of Eq. (1.1), one makes the schematization (model) more complete.
Therefore it adequately describes the phenomenon.
But even the new model describes only approximately the model under
consideration. In the case when the size and shape of the load strongly affect
its motion, the motion itself is not onedimensional, the forces acting on the
body have a more complex nature and so on. Thus it is necessary to use more
complex schematizations or in another words to exploit more complex models.
Correct schematization frequently represents a challenging task, requiring
from the researcher great experience, intuition and deep insight into the
phenomenon to be studied (Sedov, 1965).
Of special note is the continuum model, which occupies a highly important
place in the following chapters. It is known that all media, including
liquids and gases, comprise a great collection of different atoms and
molecules in permanent heat motion and with complex interactions.
By molecular interactions we mean such properties of real media as
compressibility, viscosity, heat conductivity, elasticity and others. The
complexity of these processes is very high and the governing forces
are not always known. Therefore such seemingly natural investigation
of medium motion through a study of discrete molecules is absolutely
unacceptable.
One of the general schematization methods for ﬂuid, gas and other
deformable media motion is based on the continuum model. Because each
macroscopic volume of the medium under consideration contains a great
number of molecules the medium could be approximately considered as if
it ﬁlls the space continuously. Oil, oil products, gas, water or metals may be
considered as a medium continuously ﬁlling one or another region of the
space. That is why a system of material points continuously ﬁlling a part of space
is called a continuum.
Replacement of a real medium consisting of separate molecules by a
continuum represents of course a schematization. But such a schemati
zation has proved to be very convenient in the use of the mathemat
ical apparatus of continuous functions and, as was shown in practice,
it is quite sufﬁcient for studying the overwhelming majority of observed
phenomena.
1.1.1
Governing Factors
In the examination of different phenomena the researcher is always restricted
by a ﬁnite number of parameters called governing factors (parameters) within
the limits of which the investigation is being studied. This brings up the
question: How to reveal the system of governing parameters?
It could be done for example by formulating the problem mathematically
or, in other words, by building a mathematical model of the considered
4 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
phenomenon as was demonstrated in the abovementioned example. In this
problem the governing parameters are:
x, t, m, k, f
0
, f
1
, x
0
, v
0
.
But, in order to determine the system of governing parameters, there is no
need for mathematical schematizationof the process. It is enough to be guided,
as has already been noted, by experience, intuition and understanding of the
mechanism of the phenomenon.
Let us investigate the decrease in a parachutist’s speed v in the air when
his motion can be taken as steady. Being governed only by intuition it
is an easy matter to assume the speed to be dependent on the mass of
the parachutist m, acceleration due to g, the diameter of the parachute
canopy D, the length L of its shroud and the air density ρ. The viscosity of
the air ﬂowing around the parachute during its descent can be taken into
account or ignored since the force of viscous friction is small compared to
parachute drag. Both cases represent only different schematizations of the
phenomenon.
So the function sought could be assumed to have the following general form
v = f (m, g, D, L, ρ). Then the governing parameters are:
m, g, D, L, ρ.
The use of dimensional theory permits us to rewrite the formulateddependence
in invariant form, that is, independent of the system of measurement units
(see Chapters 6 and 7)
v
_
gD
=
¯
f
_
m
ρD
3
,
L
D
_
, ⇒v =
_
gD ·
¯
f
_
m
ρD
3
,
L
D
_
.
Thus, among ﬁve governing parameters there are only two independent
dimensionless combinations, m/ρD
3
and L/D, deﬁning the soughtfor
dependence.
1.1.2
Schematization of OneDimensional Flows of Fluids and Gases in Pipelines
In problems of oil and gas transportation most often schematization of the
ﬂow process under the following conditions is used:
•
oil, oil product and gas are considered as a continuum continuously ﬁlling
the whole crosssection of the pipeline or its part;
•
the ﬂow is taken as onedimensional, that is all governing parameters
depend only on one space coordinate x measured along the pipeline axis
and, in the general case, on time t;
•
the governing parameters of the ﬂow represent values of the corresponding
physical parameters averaged over the pipeline crosssection;
1.2 Integral Characteristics of Fluid Volume 5
•
the proﬁle of the pipeline is given by the dependence of the height of the
pipeline axis above sea level on the linear coordinate z(x);
•
the area S of the pipeline crosssection depends, in the general case, on x
and t. If the pipeline is assumed to be undeformable, then S = S(x). If the
pipeline has a constant diameter, then S(x) = S
0
= const.;
•
the most important parameters are:
ρ(x, t) – density of medium to be transported, kg m
−3
;
v(x, t) – velocity of the medium, m s
−1
;
p(x, t) – pressure at the pipeline axis, Pa = N m
−2
;
T(x, t) – temperature of the medium to be transported, degrees;
τ(x, t) – shear stress (friction force per unit area of the pipeline internal
surface), Pa = N m
−2
;
Q(x, t) = vS – volume ﬂow rate of the medium, m
3
s
−1
;
˙
M(x, t) = ρvS – mass ﬂow rate of the medium, kg s
−1
and other.
Mathematical models of ﬂuid and gas ﬂows in the pipeline are based
on the fundamental laws of physics (mechanics and thermodynamics) of a
continuum, modeling a real ﬂuid and a real gas.
1.2
Integral Characteristics of Fluid Volume
In what follows one needs the notion of movable ﬂuid volume of the continuum
in the pipeline. Let, at some instant of time, an arbitrary volume of the
medium be transported between crosssections x
1
and x
2
of the pipeline
(Figure 1.1).
If the continuum located between these two crosssections is identiﬁed
with a system of material points and track is kept of its displacement in
time, the boundaries x
1
and x
2
become dependent on time and, together
with the pipeline surface, contain one and the same material points of
the continuum. This volume of the transported medium is called the
movable ﬂuid volume or individual volume. Its special feature is that it
always consists of the same particles of the continuum under consideration.
If, for example, the transported medium is incompressible and the pipeline
is nondeformable, then S = S
0
= const. and the difference between the
demarcation boundaries (x
2
−x
1
) deﬁning the length of the ﬂuid volume
remains constant.
Figure 1.1 Movable ﬂuid volume of the
continuum.
6 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
Exploiting the notion of ﬂuid or individual volume of the transported
medium in the pipeline one can introduce the following integral quantities:
M =
_
x
2
(t)
x
1
(t)
ρ(x, t) · S(x, t) dx −mass of ﬂuid volume (kg);
I =
_
x
2
(t)
x
1
(t)
ρ(x, t) · v(x, t) · S(x, t) dx −momentum of ﬂuid volume
(kg m s
−1
);
E
kin
=
_
x
2
(t)
x
1
(t)
α
k
ρv
2
2
S(x, t) dx −kinetic energy of the ﬂuid volume (J),
where α
k
is the factor;
E
in
=
_
x
2
(t)
x
1
(t)
ρ(x, t) · e
in
(x, t) · S(x, t) dx −internal energy of the ﬂuid
volume, where e
in
is the density of the internal energy (J kg
−1
), that is the
internal energy per unit mass.
These quantities model the mass, momentumand energy of a material point
system.
Since the main laws of physics are often formulated as connections between
physical quantities and the rate of their change in time, we ought to adduce
the rule of integral quantity differentiation with respect to time. The symbol
of differentiation d()/ dt denotes the total derivative with respect to time,
associated with individual particles of a continuum whereas the symbol ∂()/∂t
denotes the local derivative with respect to time, that is the derivative of a
ﬂow parameter with respect to time at a given space point, e.g. x = const. The
local derivative with respect to time gives the rate of ﬂow parameter change at
a given crosssection of the ﬂow while, at two consecutive instances of time,
different particles of the continuum are located in this crosssection.
The total derivative with respect to time is equal to
d
dt
_
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx.
From mathematical analysis it is known how an integral containing a
parameter, in the considered case it is t, is differentiated with respect to
this parameter, when the integrand and limits of integration depend on this
parameter. We have
d
dt
_
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx =
_
x
2
(t)
x
1
(t)
∂
∂t
[A(x, t) · S(x, t)] dx
+A(x, t) · S(x, t)
x
2
(t)
·
dx
2
dt
−A(x, t) · S(x, t)
x
1
(t)
·
dx
1
dt
.
1.3 The Law of Conservation of Transported Medium Mass. The Continuity Equation 7
First, at frozen upper and lower integration limits, the integrand is
differentiated (the derivative being local) and then the integrand calculated at
the upper and lower integration limits is multiplied by the rates of change of
these limits dx
2
/ dt and dx
1
/ dt, the ﬁrst term having been taken with a plus
sign and the second with a minus sign (see Appendix B).
For the case of the ﬂuid volume of the medium the quantities dx
2
/ dt and
dx
1
/ dt are the corresponding velocities v
2
(t) and v
1
(t) of the medium in the
left and right crosssections bounding the considered volume. Hence
d
dt
_
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx =
_
x
2
(t)
x
1
(t)
∂
∂t
[A(x, t) · S(x, t)] dx
+A(x, t) · v(x, t) · S(x, t)
x
2
(t)
−A(x, t) · v(x, t) · S(x, t)
x
1
(t)
.
If, in addition, we take into account the wellknown Newton–Leibniz formula,
according to which
A(x, t) · v(x, t) · S(x, t)
x
2
(t)
−A(x, t) · v(x, t) · S(x, t)
x
1
(t)
=
_
x
2
(t)
x
1
(t)
∂
∂x
[A(x, t) · v(x, t) · S(x, t)] dx,
we obtain
d
dt
_
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx =
_
x
2
(t)
x
1
(t)
_
∂AS
∂t
+
∂ASv
∂x
_
dx. (1.4)
1.3
The Law of Conservation of Transported Medium Mass. The Continuity Equation
The density ρ(x, t), the velocity of the transported medium v(x, t) and the
area of the pipeline crosssection S(x, t) cannot be chosen arbitrarily since
their values deﬁne the enhancement or reduction of the medium mass in
one or another place of the pipeline. Therefore the ﬁrst equation would be
obtained when the transported medium is governed by the mass conservation
law
d
dt
_
x
2
(t)
x
1
(t)
ρ(x, t) · S(x, t) dx = 0, (1.5)
This equation should be obeyed for any ﬂuid particle of the transported
medium, that is for any values x
1
(t) and x
2
(t).
Applying to Eq. (1.4) the rule (1.5) of differentiation of integral quantity with
regard to ﬂuid volume, we obtain
_
x
2
(t)
x
1
(t)
_
∂ρS
∂t
+
∂ρvS
∂x
_
dx = 0.
8 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
Since the last relation holds for arbitrary integration limits we get the following
differential equation
∂ρS
∂t
+
∂ρvS
∂x
= 0, (1.6)
which is called continuity equation of the transported medium in the pipeline.
If the ﬂow is stationary, that is the local derivative with respect to time is
zero (∂()/∂t = 0), the last equation is simpliﬁed to
dρvS
dx
= 0 ⇒
˙
M = ρvS = const. (1.7)
This means that in stationary ﬂow the mass ﬂow rate
˙
M is constant along the
pipeline.
If we ignore the pipeline deformation and take S(x)
∼
= S
0
= const.,
from Eq. (1.7) it follows that ρv = const. From this follow two important
consequences:
1. In the case of a homogeneous incompressible ﬂuid (sometimes oil and
oil product can be considered as such ﬂuids) ρ
∼
= ρ
0
= const. and the
ﬂow velocity v(x) = const. Hence the ﬂow velocity of a homogeneous
incompressible ﬂuid in a pipeline of constant crosssection does not change
along the length of the pipeline.
Example. The volume ﬂow rate of the oil transported by a pipeline with
diameter D = 820 mm and wall thickness δ = 8 mm is 2500 m
3
h
−1
. It is
required to ﬁnd the velocity v of the ﬂow.
Solution. The internal diameter d of the oil pipeline is equal to
d = D −2δ = 0.82 −2 · 0.008 = 0.804 m;
v = 4Q/πd
2
= const.
v = 4 · 2500/(3600 · 3.14 · 0.804
2
)
∼
= 1.37 m s
−1
.
2. In the case of a compressible medium, e.g. a gas, the density ρ(x)
changes along the length of pipeline section under consideration. Since
the density is as a rule connected with pressure, this change represents
a monotonic function decreasing from the beginning of the section to
its end. Then from the condition ρv = const. it follows that the velocity
v(x) of the ﬂow also increases monotonically from the beginning of the
section to its end. Hence the velocity of the gas ﬂow in a pipeline with
constant diameter increases from the beginning of the section between
compressor stations to its end.
Example. The mass ﬂow rate of gas transported along the pipeline
(D = 1020 mm, δ = 10 mm) is 180 kg s
−1
. Find the velocity of the gas ﬂow
1.4 The Law of Change in Momentum. The Equation of Fluid Motion 9
v
1
at the beginning and v
2
at the end of the gaspipeline section, if the
density of the gas at the beginning of the section is 45 kg m
−3
and at the
end is 25 kg m
−3
.
Solution. v
1
=
˙
M/(ρ
1
S) = 4 · 180/(45 · 3.14 · 1
2
)
∼
= 5.1 m s
−1
;
v
2
=
˙
M/(ρ
2
S) = 4 · 180/(25 · 3.14 · 1
2
)
∼
= 9.2 m s
−1
, that is the gas ﬂow
velocity is enhanced by a factor 1.8 towards the end as compared with the
velocity at the beginning.
1.4
The Law of Change in Momentum. The Equation of Fluid Motion
The continuity equation (1.6) contains several unknown functions, hence the
use of only this equation is insufﬁcient to ﬁnd each of them. To get additional
equations we can use, among others, the equation of the change in momentum
of the system of material points comprising the transported medium. This law
expresses properly the second Newton law applied to an arbitrary ﬂuid volume
of transported medium
dI
dt
=
d
dt
_
x
2
(t)
x
1
(t)
v · ρSdx = (p
1
S
1
−p
2
S
2
) +
_
x
2
(t)
x
1
(t)
p
∂S
∂x
dx
−
_
x
2
(t)
x
1
(t)
πd · τ
w
dx −
_
x
2
(t)
x
1
(t)
ρg sinα(x) · Sdx. (1.8)
On the left is the total derivative of the ﬂuid volume momentum of the
transported medium with respect to time and on the right the sum of all
external forces acting on the considered volume.
The ﬁrst term on the righthand side of the equation gives the difference
in pressure forces acting at the ends of the single continuum volume.
The second term represents the axial projection of the reaction force
from the lateral surface of the pipe (this force differs from zero when
S = const.). The third term deﬁnes the friction force at the lateral surface
of the pipe (τ
w
is the shear stress at the pipe walls, that is the friction
force per unit area of the pipeline internal surface, Pa). The fourth term
gives the sliding component of the gravity force (α(x) is the slope of the
pipeline axis to the horizontal, α > 0 for ascending sections of the pipeline;
α < 0 for descending sections of the pipeline; g is the acceleration due to
gravity).
Representing the pressure difference in the form of an integral over the
length of the considered volume
p
1
S
1
−p
2
S
2
= −
_
x
2
(t)
x
1
(t)
∂pS
∂x
dx
10 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
and noting that
−
_
x
2
(t)
x
1
(t)
∂pS
∂x
dx +
_
x
2
(t)
x
1
(t)
p
∂S
∂x
dx = −
_
x
2
(t)
x
1
(t)
S
∂p
∂x
dx,
we obtain the following equation
d
dt
_
x
2
(t)
x
1
(t)
ρvSdx =
_
x
2
(t)
x
1
(t)
_
−S
∂p
∂x
−S ·
4
d
τ
w
−Sρg sinα(x)
_
dx.
Now applying to the lefthand side of this equation the differentiation rule of
ﬂuid volume
_
x
2
(t)
x
1
(t)
_
∂ρvS
∂t
+
∂ρv
2
S
∂x
_
dx
=
_
x
2
(t)
x
1
(t)
_
−S
∂p
∂x
−S ·
4
d
τ
w
−Sρg sinα(x)
_
dx.
As far as the limits of integration in the last relation are arbitrary one can
discard the integral sign and get the differential equation
∂ρvS
∂t
+
∂ρv
2
S
∂x
= S ·
_
−
∂p
∂x
−
4
d
τ
w
−ρg sinα(x)
_
. (1.9)
If we represent the lefthand side of this equation in the form
v
_
∂ρS
∂t
+
∂ρvS
∂x
_
+ρS
_
∂v
∂t
+v
∂v
∂x
_
and take into account that in accordance with the continuity equation (1.6) the
expression in the ﬁrst brackets is equal to zero, the resulting equation may be
written in a more simple form
ρ
_
∂v
∂t
+v
∂v
∂x
_
= −
∂p
∂x
−
4
d
τ
w
−ρg sinα(x). (1.10)
The expression in brackets on the lefthand side of Eq. (1.10) represents the
total derivative with respect to time, that is the particle acceleration
w =
dv
dt
=
∂v
∂t
+v
∂v
∂x
. (1.11)
Now the meaning of Eq. (1.10) becomes clearer: the product of unit volume
mass of transported medium and its acceleration is equal to the sum of all
forces acting on the medium, namely pressure, friction and gravity forces. So
Eq. (1.10) expresses the Newton’s Second Law and can therefore also be called
the ﬂow motion equation.
Remark. about the connection between total and partial derivatives with respect
to time. The acceleration w = dv/ dt is a total derivative with respect to time
(symbol d()/ dt), since we are dealing with the velocity differentiation of one
and the same ﬁxed particle of the transported medium moving from one
1.5 The Equation of Mechanical Energy Balance 11
crosssection of the pipeline to another one, whereas the partial derivative with
respect to time (symbol ∂()/∂t) has the meaning of velocity differentiation at
a given place in space, that is at a constant value of x. Thus such a derivative
gives the change in velocity of different particles of the transported medium
entering a given crosssection of the pipeline.
Let a particle of the medium at the instant of time t be in the crosssection
x of the pipeline and so have velocity v(x, t). In the next instant of time t +t
this particle will transfer to the crosssection x +x and will have velocity
v(x +x, t +t). The acceleration w of this particle is deﬁned as the limit
w =
dv
dt
= lim
t⇒0
v(x +x, t +t) −v(x, t)
t
=
∂v
∂t
¸
¸
¸
¸
x
+
∂v
∂x
¸
¸
¸
¸
t
·
dx
dt
.
Since dx/ dt = v(x, t) is the velocity of the considered particle, from the last
equality it follows that
dv
dt
=
∂v
∂t
+v ·
∂v
∂x
. (1.12)
A similar relation between the total derivative ( d/ dt), or as it is also called
the individual or Lagrangian derivative, and the partial derivative (∂/∂t), or as
it is also called the local or Eulerian derivative, has the form (1.12) no matter
whether the case in point is velocity or any other parameter A(x, t)
dA(x, t)
dt
=
∂A(x, t)
∂t
+v ·
∂A(x, t)
∂x
.
1.5
The Equation of Mechanical Energy Balance
Consider now what leads to the use of the mechanical energy change law as
applied to the system of material points representing a ﬂuid particle of the
transported medium. This law is written as:
dE
kin
dt
=
dA
ex
dt
+
dA
in
dt
(1.13)
that is the change in kinetic energy of a system of material points dE
kin
is equal
to the sum of the work of the external dA
ex
and internal dA
in
forces acting on
the points of this system.
We can calculate separately the terms of this equation but ﬁrst we should
deﬁne more exactly what meant by the kinetic energy E
kin
. If the transported
medium moves in the pipeline as a piston with equal velocity v(x, t) over the
crosssection then the kinetic energy would be expressed as the integral
E
kin
=
_
x
2
(t)
x
1
(t)
ρv
2
2
Sdx.
12 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
But, in practice, such a schematization is too rough because, as experiments
show, the velocity of the separate layers of the transported medium (ﬂuid or
gas) varies over the pipe crosssection. At the center of the pipe it reaches the
greatest value, whereas as the internal surface of the pipe is approached the
velocity decreases and at the wall itself it is equal to zero. Furthermore, if at
a small velocity of the ﬂuid the ﬂow regime is laminar, with an increase in
velocity the laminar ﬂow changes into a turbulent one (pulsating and mixing
ﬂow) and the velocities of the separate particles differ signiﬁcantly from the
average velocity v of the ﬂow. That is why models of the ﬂow are, as a rule,
constructed with regard to the difference in ﬂow velocity from the average
velocity over the crosssection.
The true velocity u of a particle of the transported mediumis givenas the sum
u = v +u of the average velocity over the crosssection v(x, t) and the additive
one (deviation) u representing the difference between the true velocity and
the average one. The average value of this additive u is equal to zero, but
the rootmeansquare (rms) value of the additive (u)
2
is nonvanishing.
The deviation characterizes the kinetic energy of the relative motion of the
continuumparticle in the pipeline crosssection. Then the kinetic energy of the
transported medium unit mass e
kin
may be presented as the sum of two terms
e
kin
=
v
2
2
+
(u)
2
2
namely the kinetic energy of the center of mass of the considered point system
and the kinetic energy of the motion of these points relative to the center of
mass. If the average velocity v = 0, then
ρv
2
2
+
ρ(u)
2
2
=
ρv
2
2
·
_
1 +
(u)
2
v
2
_
= α
k
·
ρv
2
2
where α
k
= 1 +(u)
2
/v
2
> 1. For laminar ﬂow α
k
= 4/3, while for turbulent
ﬂow the value of α
k
lies in the range 1.02–1.05.
Remark. It should be noted that in onedimensional theory, as a rule, the cases
v = 0 and (u)
2
= 0 are not considered.
With regard to the introduced factor the kinetic energy of any movable
volume of transported medium may be represented as
E
kin
=
_
x
2
(t)
x
1
(t)
α
k
·
ρv
2
2
· Sdx.
Let us turn now to the calculation of the terms in the mechanical energy
equation (1.13). Let us calculate ﬁrst the change in kinetic energy
dE
kin
dt
=
d
dt
__
x
2
(t)
x
1
(t)
α
k
·
ρv
2
2
Sdx
_
.
1.5 The Equation of Mechanical Energy Balance 13
Employing the rule of integral quantity integration with reference to the ﬂuid
volume, that is an integral with variable integration limits, we get
dE
kin
dt
=
_
x
2
(t)
x
1
(t)
_
∂
∂t
_
α
k
·
ρv
2
2
S
_
+
∂
∂x
_
α
k
·
ρv
2
2
S · v
__
dx.
The work of the external forces (in this case they are the forces of pressure and
gravity), including also the work of external mechanical devices, e.g. pumps if
such are used, is equal to
dA
ex
dt
= (p
1
Sv
1
−p
2
Sv
2
) −
_
x
2
(t)
x
1
(t)
ρg sinα · v · Sdx +N
mech
= −
_
x
2
(t)
x
1
(t)
∂
∂x
(pSv) dx −
_
x
2
(t)
x
1
(t)
ρg sinα · v · Sdx +N
mech
.
The ﬁrst term on the righthand side of the last expression gives the work
performed in unit time or, more precisely, the power of the pressure force
applied to the initial and end crosssections of the detached volume. The
second term gives the power of the gravity force and the third term N
mech
the
power of the external mechanical devices acting on the transported medium
volume under consideration.
The work of the internal forces (pressure and internal friction) executed in
unit time is given by
dA
in
dt
=
_
x
2
(t)
x
1
(t)
p
∂(Sv)
∂x
dx +
_
x
2
(t)
x
1
(t)
n
in
· ρSdx.
The ﬁrst term on the righthand side gives the work of the pressure force in
unit time, that is the power, for compression of the particles of the medium,
the factor ∂(Sv)/∂x · dx giving the rate of elementary volume change. The
second term represents the power of the internal friction forces, that is the
forces of mutual friction between the internal layers of the medium, n
in
denoting speciﬁc power, that is per unit mass of the transported medium.
In what follows it will be shown that this quantity characterizes the amount
of mechanical energy converting into heat per unit time caused by mutual
internal friction of the transported particles of the medium.
Gathering together all the terms of the mechanical energy equation we get
_
x
2
(t)
x
1
(t)
_
∂
∂t
_
α
k
·
ρv
2
2
S
_
+
∂
∂x
_
α
k
·
v
2
2
ρvS
__
dx
= −
_
x
2
(t)
x
1
(t)
ρSv
__
1
ρ
∂p
∂x
_
+g sinα
_
dx +
_
x
2
(t)
x
1
(t)
n
in
· ρSdx +N
mech
.
If the transported medium is barotropic, that is the pressure in it depends only
on the density p = p(ρ), one can introduce a function P(ρ) of the pressure
14 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
such that dP = dp/ρ, P(ρ) =
_
dp/ρ and
1
ρ
∂p
∂x
=
∂P(ρ)
∂x
. If, moreover, we take
into account the equality sinα(x) = ∂z/∂x, where the function z(x) is referred
to as the pipeline proﬁle, the last equation could be rewritten in the simple
form
_
x
2
(t)
x
1
(t)
_
ρS
∂
∂t
_
α
k
v
2
2
_
+ρvS
∂
∂x
_
α
k
v
2
2
+P(ρ) +gz
__
dx
=
_
x
2
(t)
x
1
(t)
n
in
· ρSdx +N
mech
. (1.14)
If we assume that in the region [x
1
(t), x
2
(t)] external sources of mechanical
energy are absent. Then N
mech
= 0 and we can go from the integral equality
(1.14) to a differential equation using, as before, the condition of arbitrariness
of integration limits x
1
(t) and x
2
(t) in Eq. (1.14). Then the sign of the integral
can be omitted and the corresponding differential equation is
ρS
∂
∂t
_
α
k
v
2
2
_
+ρvS
∂
∂x
_
α
k
v
2
2
+P(ρ) +gz
_
= ρS · n
in
(1.15)
or
∂
∂t
_
α
k
v
2
2
_
+v ·
∂
∂x
_
α
k
v
2
2
+
_
dp
ρ
+gz
_
= n
in
. (1.16)
This is the sought differential equationexpressing the lawof mechanical energy
change. It should be emphasized that this equation is not a consequence of the
motion equation (1.10). It represents an independent equation for modeling
onedimensional ﬂows of a transported medium in the pipeline.
If we divide both parts of Eq. (1.16) by g we get
∂
∂t
_
α
k
v
2
2g
_
+v ·
∂
∂x
_
α
k
v
2
2g
+
_
dp
ρg
+z
_
=
n
in
g
.
The expression
H =
α
k
v
2
2g
+
_
dp
ρg
+z (1.17)
in the derivative on the lefthand side of the last equation has the dimension of
length and is called the total head. The total head at the pipeline crosssection
x consists of the kinetic head (dynamic pressure) α
k
v
2
/2g, the piezometric head
_
dp/ρg and the geometric head z. The concept of head is very important in the
calculation of processes occurring in pipelines.
1.5 The Equation of Mechanical Energy Balance 15
1.5.1
Bernoulli Equation
In the case of stationary ﬂow of a barotropic ﬂuid or gas in the pipeline the
derivative ∂()/∂t = 0, hence the following ordinary differential equations apply
v
d
dx
_
α
k
v
2
2g
+
_
dp
ρg
+z
_
=
n
in
g
or
d
dx
_
α
k
v
2
2g
+
_
dp
ρg
+z
_
=
n
in
gv
= i, (1.18)
where i denotes the dimensional quantity n
in
/gv called the hydraulic gradient
i =
dH
dx
=
n
in
gv
.
Thus the hydraulic gradient, deﬁned as the pressure loss per unit length of
the pipeline, is proportional to the dissipation of mechanical energy into heat
through internal friction between the transported medium layers (i < 0).
In integral form, that is as applied to transported medium located between
two ﬁxed crosssections x
1
and x
2
, Eq. (1.18) takes the following form
_
α
k
v
2
2g
+
_
dp
ρg
+z
_
1
−
_
α
k
v
2
2g
+
_
dp
ρg
+z
_
2
= −
_
x
2
x
1
i dx. (1.19)
This equation is called the Bernoulli equation. It is one of the fundamental
equations used to describe the stationary ﬂow of a barotropic medium in
a pipeline.
For an incompressible homogeneous ﬂuid, which under some conditions can
be water, oil and oil product, ρ = const.,
_
dp/ρg = p/ρg +const. Therefore
the Bernoulli equation becomes
_
α
k
v
2
2g
+
p
ρg
+z
_
1
−
_
α
k
v
2
2g
+
p
ρg
+z
_
2
= −
_
x
2
x
1
i dx.
If in addition we take i = −i
0
= const. (i
0
> 0), then
_
α
κ
v
2
2g
+
p
ρg
+z
_
1
−
_
α
κ
v
2
2g
+
p
ρg
+z
_
2
= i
0
· l
1−2
(1.20)
where l
1−2
is the length of the pipeline between crosssections 1 and 2.
This last equationhas a simple geometric interpretation(see Figure 1.2). This
ﬁgure illustrates a pipeline proﬁle (heavy broken line); the line H(x) denoting
the dependence of the total head H on the coordinate x directed along the
axis of the pipeline (straight line) with constant slope β to the horizontal
(i = dH/ dx = tgβ = const.) and three components of the total head at an
16 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
Figure 1.2 Geometric interpretation of the Bernoulli equation.
arbitrary crosssection of the pipeline: geometric head z(x), piezometric head
p(x)/ρg and kinetic head α
k
v
2
(x)/2g.
The line H(x) representing the dependence of the total head H on the
coordinate x along the pipeline axis is called the line of hydraulic gradient.
It should be noted that if we neglect the dynamic pressure (in oil and
oil product pipelines the value of the dynamic pressure does not exceed the
pipeline diameter, e.g. at v ≈ 2 m s
−1
, α
k
≈ 1.05 then v
2
/2g
∼
= 0.25 m), and
the length of the section between the pipeline proﬁle and the line of hydraulic
gradient multiplied by ρg gives the value of the pressure in the pipeline cross
section x. For example, when the length of the section AA (see Figure 1.2) is
500 m and diesel fuel with density ρ = 840 kg m
−3
is transported along the
pipeline, then
p
840 · 9.81
= 500 ⇒p = 500 · 840 · 9.81 = 4 120 200 (Pa)
or 4.12 MPa (≈42 atm).
1.5.2
Input of External Energy
In ﬂuid ﬂow in the pipeline the mechanical energy is dissipated into heat and
the pressure decreases gradually. Devices providing pressure restoration or
generation are called compressors.
Compressors installed separately or combined in a group form the pumping
plant destined to set the ﬂuid moving from the crosssection with lesser
pressure to the crosssection with greater pressure. To do this it is required to
expend, or deliver from outside to the ﬂuid, energy whose power is denoted by
N
mech
.
Let index 1 in the Bernoulli equation refer to parameters at the crosssection
x
1
of the pump entrance (suction line) and index 2 at the crosssection x
2
of the
1.6 Equation of Change in Internal Motion Kinetic Energy 17
pump exit (discharge line). Since ρvS = const., the Bernoulli equation (1.14)
may be written as:
_
x
2
x
1
d
dx
_
ρvS ·
_
α
k
v
2
2
+
p
ρ
+gz
__
dx =
_
x
2
x
1
n
in
· ρSdx +N
mech
.
Ignoring the difference between the kinetic and geometric heads we get
ρvS ·
p
2
−p
1
ρ
−
_
x
2
x
1
n
in
· ρSdx = N
mech
.
Denoting by H = (p
2
−p
1
)/ρg the differential head produced by the pump or
pumping plant and taking into account that ρvS = ρQ = const. and n
in
= gv · i,
we obtain
N
mech
= ρgQ · H−
_
x
2
x
1
ρgQ · i dx = ρgQ · H ·
_
1 −
_
x
2
x
1
i
H
dx
_
.
The expression in parentheses characterizes the loss of mechanical energy
within the pump. Usually this factor is taken into account by insertion of the
pump efﬁciency η
η =
_
1 −
_
x
2
x
1
i/H dx
_
−1
< 1
so that
N
mech
=
ρgQ · H
η(Q)
. (1.21)
The relation (1.21) is the main formula used to calculate the power of the
pump generating head H in ﬂuid pumping with ﬂow rate Q.
1.6
Equation of Change in Internal Motion Kinetic Energy
At the beginning of the previous section it was noted that the total
kinetic energy of the transported medium consisted of two terms – the
kinetic energy of the center of mass of the particle and the kinetic energy
of the internal motion of the center of mass, so that the total energy
of a particle is equal to α
k
ρv
2
/2, where α
k
> 1. Now we can derive an
equation for the second component of the kinetic energy, namely the kinetic
energy of the internal or relative motion in the ﬂow of the transported
medium.
Multiplication of motion equation (1.10) by the product vS yields
ρS
d
dt
_
v
2
2
_
= −
∂p
∂x
· vS −
4
d
τ
w
· vS −ρgvS · sinα(x).
18 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
Subtracting this equation termbyterm from the Bernoulli equation (1.15),
one obtains
ρS
d
dt
_
(α
k
−1)
v
2
2
_
=
4
d
τ
w
· vS +ρS · n
in
.
Introduction of n
in
= −gv · i
0
gives
ρS
d
dt
_
(α
κ
−1)
v
2
2
_
=
_
4
d
τ
w
· v
_
S −ρgvS · i
0
. (1.22)
This is the sought equation of change in kinetic energy of internal motion of
onedimensional ﬂow of the transported medium. Its sense is obvious: the
power of the external friction forces (4τ
w
· vS/d) in onedimensional ﬂow minus the
power ρgS(v · i
0
) of internal friction forces between the particles causing transition
of mechanical energy into heat is equal to the rate of change of internal motion
kinetic energy in the ﬂow of the transported medium.
For stationary ﬂow ( d/ dt = 0 +v · ∂/∂x) of the transported medium
Eq. (1.22) gives
d
dx
_
(α
k
−1)
v
2
2
_
=
4
d
τ
w
ρ
−g · i
0
. (1.23)
If v
∼
= const., which for the ﬂow of an incompressible medium in a pipeline
with constant diameter is the exact condition, the lefthand part of the equation
vanishes. This means that the tangential friction tension τ
w
at the pipeline
wall and the hydraulic gradient i
0
are connected by
τ
w
=
ρgd
4
· i
0
. (1.24)
It must be emphasized that in the general case, including nonstationary ﬂow,
such a connection between τ
w
and i
0
is absent (see Section 4.1).
1.6.1
Hydraulic Losses (of Mechanical Energy)
The quantity n
in
entering into Eq. (1.16) denotes the speciﬁc power of the
internal friction force, that is per unit mass of transported medium. This
quantity is very important since it characterizes the loss of mechanical energy
converted into heat owing to internal friction between layers of the medium.
In order to derive this quantity theoretically one should know how the layers
of transported medium move at each crosssection of the pipeline but this is
not always possible. In the next chapter it will be shown that in several cases,
in particular for laminar, ﬂow such motion can be calculated and the quantity
n
in
can be found. In other cases, such as for turbulent ﬂows of the transported
1.6 Equation of Change in Internal Motion Kinetic Energy 19
medium, it is not possible to calculate the motion of the layers and other
methods of determining n
in
are needed.
The quantity of speciﬁc mechanical energy dissipation n
in
has the following
dimension (from now onwards dimension will be denoted by the symbol [ ])
[n
in
] =
W
kg
=
J
s kg
=
N m
s kg
=
kg m s
−2
m
s kg
=
m
2
s
3
=
_
v
3
d
_
.
So the dimension of n
in
is the same as the dimension of the quantity v
3
/d,
hence, without disturbance of generality, one can seek n
in
in the form
n
in
= −
λ
2
·
v
3
d
(1.25)
where λ is a dimensional factor (λ > 0), the minus sign shows that n
in
< 0,
that is the mechanical energy decreases thanks to the forces of internal friction.
The factor 1/2 is introduced for the sake of convenience.
The presented formula does not disturb the generality of the consideration
because the unknown dependence of n
in
on the governing parameters of the
ﬂow is accounted for by the factor λ. This dependence is valid for any medium
be it ﬂuid, gas or other medium with complex speciﬁc properties, e.g. waxy
crude oil, suspension or even pulp, that is a mixture of water with large rigid
particles.
For stationary ﬂuid or gas ﬂow one can suppose the factor λ to be dependent
on four main parameters: the ﬂow velocity v (m s
−1
), the kinematic viscosity
of the ﬂow ν (m
2
s
−1
), the internal diameter of the pipeline d (m) and the
mean height of the roughness of its internal surface (mm or m), so that
λ = f (v, ν, d, ). The density of the ﬂuid ρ and the acceleration due to gravity
g are not included here because intuition suggests that the friction between
ﬂuid or gas layers will be dependent on neither their density nor the force of
gravity.
Note that the quantity λ is dimensionless, that is its numerical value
is independent of the system of measurement units, while the parameters
v, ν, d, are dimensional quantities and their numerical values depend on
such a choice. The apparent contradiction is resolved by the wellknown
Buckingham Itheorem, in accordance with which any dimensionless quantity
can depend only on dimensionless combinations of parameters governing this
quantity (Lurie, 2001). In our case there are two such parameters
v · d
ν
= Re and
d
= ε,
the ﬁrst is called the Reynolds number and the second the relative roughness of
the pipeline internal surface. Thus
λ = λ(Re, ε).
20 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
The formula (1.25) acquires the form
n
in
= −λ(Re, ε) ·
1
d
·
v
3
2
. (1.26)
The factor λ in this formula is called the hydraulic resistance factor, one
of the most important parameters of hydraulics and pipeline transportation.
Characteristic values of λ lie in the range 0.01–0.03. More detailed information
about this factor and its dependence on the governing parameters will be
presented below.
Turning to the hydraulic gradient i
0
, one can write
i
0
= −
n
in
gv
= λ ·
1
d
·
v
2
2g
. (1.27)
Characteristic values of the hydraulic slope are 0.00005–0.005.
If we substitute Eq. (1.27) into the Bernoulli equation (1.20), we obtain
_
α
k
v
2
2g
+
p
ρg
+z
_
1
−
_
α
k
v
2
2g
+
p
ρg
+z
_
2
= λ(Re, ε) ·
l
1−2
d
v
2
2g
. (1.28)
The expression h
τ
= λ · l
1–2
/d · v
2
/2g on the righthand side of this equation
is called the loss of head in DarcyVeisbach form.
Using Eq. (1.27) in the case of stationary ﬂow of the transported medium
permits us to get an expression for the tangential friction stress τ
w
at the
pipeline wall. Substitution of Eq. (1.27) into Eq. (1.24), yields
τ
w
=
ρgd
4
· i
0
=
ρgd
4
·
_
λ
1
d
v
2
2g
_
=
λ
4
·
ρv
2
2
= C
f
·
ρv
2
2
, (1.29)
C
f
(Re, ε) =
λ(Re, ε)
4
where the dimensional factor C
f
is called the friction factor of the ﬂuid on the
internal surface of the pipeline or the Funning factor (Leibenson et al., 1934).
1.6.2
Formulas for Calculation of the Factor λ(Re, ε)
Details of methods to ﬁnd and calculate the factor of hydraulic resistance λ in
Eqs. (1.26)–(1.29) and one of the primary factors in hydraulics and pipeline
transportation will be given in Chapter 3. Here are shown several formulas
exploiting the practice.
If the ﬂowof ﬂuid or gas in the pipeline is laminar, that is jetwise or layerwise
(the Reynolds number Re should be less than 2300), then to determine λ the
Stokes formula (see Section 3.1) is used
λ =
64
Re
. (1.30)
1.6 Equation of Change in Internal Motion Kinetic Energy 21
As the Reynolds number increases (Re > 2300) the ﬂow in the pipeline
gradually loses hydrodynamic stability and becomes turbulent, that is vortex
ﬂow with mixing layers. The best known formula to calculate the factor λ in
this case is the Altshuler formula:
λ = 0.11 ·
_
ε +
68
Re
_
1/4
(1.31)
valid over a wide range of Reynolds number from 10
4
up to 10
6
and higher.
If 10
4
< Re < 27/ε
1.143
and Re < 10
5
, the Altshuler formula becomes the
Blasius formula:
λ =
0.3164
4
√
Re
(1.32)
having the same peculiarity as the Stokes formula for laminar ﬂow, which
does not consider the relative roughness of the pipeline internal surface ε.
This means that for the considered range of Reynolds numbers the pipeline
behaves as a pipeline with a smooth surface. Therefore the ﬂuid ﬂow in this
range is ﬂow in a hydraulic smooth pipe. In this case the friction tension τ
w
at
the pipe wall is expressed by formula
τ
w
= −
λ
4
·
ρv
2
2
= −
0.0791
4
_
vd/ν
·
ρv
2
2
≈ v
1.75
signifying that friction resistance is proportional to ﬂuid mean velocity to the
power of 1.75.
If Re > 500/ε, the second term in parentheses in the Altshuler formula can
be neglected compared to the ﬁrst one. Whence it follows that at great ﬂuid
velocities the ﬂuid friction is caused chieﬂy by the smoothness of the pipeline
internal surface, that is by the parameter ε. In such a case one can use the
simpler Shiphrinson formula λ = 0.11 · ε
0.25
. Then
τ
w
= −
λ
4
·
ρv
2
2
= −
0.11 · ε
1/4
4
·
ρv
2
2
≈ v
2
.
From this it transpires that the friction resistance is proportional to the square
of the ﬂuid mean velocity and hence this type of ﬂow is called square ﬂow.
Finally, in the region of ﬂow transition from laminar to turbulent, that
is in the range of Reynolds number from 2320 up to 10
4
one can use the
approximation formula
λ =
64
Re
· (1 −γ
•
) +
0.3164
4
√
Re
· γ
∗
, (1.33)
where γ
∗
= 1 −e
−0.002·(Re−2320)
is the intermittency factor (Ginsburg, 1957). It
is obvious that the form of the last formula assures continuous transfer from
the Stokes formula for laminar ﬂow to the Blasius formula for turbulent ﬂow
in the zone of hydraulic smooth pipes.
22 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
To calculate the hydraulic resistance factor λ of the gas ﬂow in a gas main,
where the Reynolds number Re is very large and this factor depends only on
the condition of the pipeline internal surface, Eq. (1.34) is often used.
λ = 0.067 ·
_
2
d
_
0.2
(1.34)
in which the absolute roughness is equal to 0.03–0.05 mm.
Exercise 1. The oil (ρ = 870 kg m
−3
, ν = 15 s St) ﬂows along the pipeline
(D = 156 mm; δ = 5 mm; = 0.1 mm) with mean velocity v = 0.2 m s
−1
.
Determine through the Reynolds criterion the ﬂow regime; calculate factors λ
and C
f
.
Answer. Laminar; 0.033; 0.0083.
Exercise 2. Benzene (ρ = 750 kg m
−3
, ν = 0.7 s St) ﬂows along the pipeline
(D = 377 mm; δ = 7 mm; = 0.15 mm) with mean velocity v = 1.4 m s
−1
.
Determine through the Reynolds criterion the ﬂow regime; calculate factors λ
and C
f
.
Answer. Turbulent; 0.017; 0.0041.
Exercise 3. Diesel fuel (ρ = 840 kg m
−3
, ν = 6 s St) ﬂows along the pipeline
(D = 530 mm; δ = 8 mm; = 0.25 mm) with mean velocity v = 0.8 m s
−1
.
Determine the ﬂow regime; calculate factors λ and C
f
.
Answer. Turbulent; 0.022; 0.0054.
1.7
Total Energy Balance Equation
Besides the law (1.13) of mechanical energy change of material points,
applied to an arbitrary continuum volume in the pipeline there is one more
fundamental physical law valid for any continuum – the law of total energy
conservation or, as it is also called, the ﬁrst law of thermodynamics. This law
asserts that the energy does not appear from anywhere and does not disappear
to anywhere. It changes in total quantity from one form into another. As
applied to our case this law may be written as follows
d(E
kin
+E
in
)
dt
=
dQ
ex
dt
+
dA
ex
dt
(1.35)
that is the change in total energy (E
kin
+E
in
) of an arbitrary volume of the
transported medium happens only due to the exchange of energy with
surrounding bodies owing to external inﬂow of heat dQ
ex
and the work
of external forces dA
ex
.
1.7 Total Energy Balance Equation 23
In Eq. (1.35) E
in
is the internal energy of the considered mass of transported
medium, unrelated to the kinetic energy, that is the energy of heat motion,
interaction between molecules and atoms and so on. In thermodynamics
reasons are given as to why the internal energy is a function of state, that
is at thermodynamic equilibrium of a body in some state the energy has a
welldeﬁned value regardless of the means (procedure) by which this state
was achieved. At the same time the quantities dQ
ex
/ dt and dA
ex
/ dt are
not generally derivatives with respect to a certain function of state but only
represent the ratio of elementary inﬂows of heat energy (differential dQ
ex
) and
external mechanical energy (differential dA
ex
) to the time dt in which these
inﬂows happened. It should be kept in mind that these quantities depend on
the process going on in the medium.
In addition to function E
in
one more function e
in
is often introduced,
representing the internal energy of a unit mass of the considered body
e
in
= E
in
/m, where m is the mass of the body.
We can write Eq. (1.35) for a movable volume of transported medium
enclosed between crosssections x
1
(t) and x
2
(t). The terms of this equation are
d(E
kin
+E
in
)
dt
=
d
dt
__
x
2
(t)
x
1
(t)
_
α
k
ρv
2
2
+ρe
in
_
Sdx
_
,
dQ
ex
dt
=
_
x
2
(t)
x
1
(t)
πd · q
n
dx,
dA
ex
dt
= −
_
x
2
(t)
x
1
(t)
∂
∂x
(pSv) dx −
_
x
2
(t)
x
1
(t)
ρg sinα · v · Sdx +N
mech
where q
n
is the heat ﬂux going through the unit area of the pipeline surface
per unit time (W m
−2
); πd · dx is an element of pipeline surface area and d is
the pipeline diameter.
Gathering all terms, we obtain
d
dt
__
x
2
(t)
x
1
(t)
_
α
k
·
ρv
2
2
+ρe
in
_
Sdx
_
=
_
x
2
(t)
x
1
(t)
πd · q
n
dx
−
_
x
2
(t)
x
1
(t)
∂
∂x
(pSv) dx −
_
x
2
(t)
x
1
(t)
ρg sinα · v · Sdx +N
mech
.
Differentiation of the lefthand side of this equation gives
_
x
2
(t)
x
1
(t)
_
∂
∂t
__
α
k
v
2
2
+e
in
_
ρS
_
+
∂
∂x
__
α
k
v
2
2
+e
in
_
ρvS
__
dx
=
_
x
2
(t)
x
1
(t)
πd · q
n
dx −
_
x
2
(t)
x
1
(t)
∂
∂x
_
p
ρ
ρvS
_
dx
−
_
x
2
(t)
x
1
(t)
ρvSg
∂z
∂x
dx +N
mech
24 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
or
_
x
2
(t)
x
1
(t)
_
∂
∂t
__
α
k
v
2
2
+e
in
_
ρS
_
+
∂
∂x
__
α
k
v
2
2
+e
in
+
p
ρ
_
ρvS
__
dx
=
_
x
2
(t)
x
1
(t)
πd · q
n
dx −
_
x
2
(t)
x
1
(t)
ρvSg
∂z
∂x
dx +N
mech
. (1.36)
If we assume that inside the region [x
1
(t), x
2
(t)] the external sources of me
chanical energy are absent, that is N
mech
= 0, then it is possible to pass from
integral equality (1.36) to the corresponding differential equation using, as
before, the condition that this equation should be true for any volume of the
transported medium, that is the limits of integration x
1
(t) and x
2
(t) in (1.36)
are to be arbitrarily chosen. Then the sign of the integral can be omitted and
the differential equation is
∂
∂t
__
α
k
v
2
2
+e
in
_
ρS
_
+
∂
∂x
__
α
k
v
2
2
+e
in
+
p
ρ
_
ρvS
_
= πd · q
n
−ρvSg
∂z
∂x
. (1.37)
Excluding from Eq. (1.37) the change in kinetic energy with the help of the
Bernoulli equation with term by term subtraction of Eq. (1.16) from Eq. (1.37)
we get one more energy equation
ρS
∂
∂t
_
α
k
v
2
2
_
+ρvS
∂
∂x
_
α
k
v
2
2
+
_
dp
ρ
+gz
_
= ρvSg · i
called the equation of heat inﬂow.
This equation could be variously written. First, it may be written through
the internal energy e
in
:
∂
∂t
(e
in
· ρS) +
∂
∂x
(e
in
· ρvS) = πd · q
n
−p
∂vS
∂x
−ρvSg · i
or
ρS
_
∂e
in
∂t
+v
∂e
in
∂x
_
= πd · q
n
−p ·
∂vS
∂x
−ρvSg · i. (1.38)
This equation proved to be especially convenient for modeling ﬂows of
incompressible or slightly compressible ﬂuids because the derivative ∂(vS)/∂x
expressing the change in ﬂuid volume in the pipeline crosssectionis extremely
small as is the work p · ∂(vS)/∂x of the pressure forces. With this in mind
Eq. (1.38) may be written in a particularly simple form:
ρ
de
in
dt
∼
=
4
d
· q
n
−ρvg · i. (1.39)
1.7 Total Energy Balance Equation 25
This means that the rate of internal energy change of the transported medium
is determined by the inﬂow of external heat through the pipeline surface and
heat extraction due to conversion of mechanical energy into heat produced by
friction between the continuum layers.
Second, the equation of heat inﬂow can be written using the function
J = e
in
+p/ρ representing one of the basic thermodynamic functions, enthalpy
or heat content, of the transported medium
∂
∂t
(e
in
· ρS) +
∂
∂x
__
e
in
+
p
ρ
_
ρvS
_
= πd · q
n
+ρvSg ·
_
1
ρg
∂p
∂x
−i
_
or
∂
∂t
(e
in
· ρS) +
∂
∂x
[J · ρvS] = πd · q
n
+ρvSg ·
_
1
ρg
∂p
∂x
−i
_
. (1.40)
If we take into account (as will be shown later) that the expression in
parentheses on the righthand side of this equation is close to zero, since for a
relatively light medium, e.g. gas, the hydraulic slope is expressed through the
pressure gradient by the formula i = 1/ρg · ∂p/∂x, the equation of heat inﬂow
can be reduced to a simpler form
∂ρS · e
in
∂t
+
∂ρvS · J
∂x
∼
= πd · q
n
(1.41)
in which the dissipation of mechanical energy appears to be absent.
Temperature Distribution in Stationary Flow
The equation of heat inﬂow in the form (1.39) or (1.41) is convenient to
determine the temperature distribution along the pipeline length in stationary
ﬂow of the transported medium.
1. For an incompressible or slightly compressible medium, e.g. dropping liquid:
water, oil and oil product, this equation has the form
ρv ·
de
in
dx
∼
=
4
d
· q
n
−ρvg · i. (1.42)
The internal energy e
in
depends primarily on the temperature of the ﬂuid
T, the derivative de
in
/ dT giving its speciﬁc heat C
v
(J kg
−1
K
−1
). If we take
C
v
= const. then e
in
= C
v
· T +const.
To model the heat ﬂux q
n
the Newton formula is usually used
q
n
= −κ · (T −T
ex
), (1.43)
by which this ﬂow is proportional to the difference between the temperatures
T and T
ex
in and outside the pipeline, with q
n
< 0 when T > T
ex
and q
n
> 0
when T < T
ex
. The factor κ (W m
−2
K
−1
) in this formula characterizes the
overall heat resistance of the materials through which the heat is transferred
26 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
from the pipe to the surrounding medium (anticorrosive and heat insulation,
ground, the boundary between ground and air and so on) or the reverse. This
factor is called the heattransfer factor.
The hydraulic gradient i can sometimes be considered constant i = −i
0
≈
const., if the dissipation of mechanical energy in the stationary ﬂuid ﬂow in
the pipeline with constant diameter is identical at all crosssections of the
pipeline.
With due regard for all the aforesaid Eq. (1.42) is reduced to the following
ordinary differential equation
ρC
v
v ·
dT
dx
= −
4κ
d
(T −T
ex
) +ρvgi
0
(1.44)
for temperature T = T(x). From this equation in particular it follows that the
heat transfer through the pipeline wall (the ﬁrst term on the righthand side)
lowers the temperature of the transported medium when T(x) > T
ex
or raises
it when T(x) < T
ex
, whereas the dissipation of mechanical energy (the second
term on the righthand side) always implies an increase in the temperature of
the transported medium.
The solution of the differential equation (1.44) with initial condition
T(0) = T
0
yields
T(x) −T
ex
−T
⊗
T
0
−T
ex
−T
⊗
= exp
_
−
πdκ
C
v
˙
M
x
_
. (1.45)
Where T
⊗
= gi
0
˙
M/πdκ is a constant having the dimension of temperature;
˙
M = ρvS is the mass ﬂow rate of the ﬂuid (
˙
M = const.). The formula thus
obtained is called the Shuchov formula.
Figure 1.3 illustrates the distribution of temperature T(x) along the pipeline
length x in accordance with Eq. (1.45).
The ﬁgure shows that when the initial temperature T
0
is greater than
(T
ex
+T
⊗
), the moving medium cools down, while when T
0
is less than
(T
ex
+T
⊗
), the medium gradually heats up. In all cases with increase in the
pipeline length the temperature T →(T
ex
+T
⊗
).
In particular from Eq. (1.44) it follows that if the heat insulation of the
pipeline is chosen such that at the initial crosssection of the pipeline x = 0
Figure 1.3 Temperature
distribution along the pipeline length.
1.7 Total Energy Balance Equation 27
the condition of equality to zero of the righthand side is obeyed
−
4κ
d
(T
0
−T
ex
) +ρvgi
0
= 0
that is the factor κ satisﬁes the condition
κ =
ρvdgi
0
4(T
0
−T
ex
)
=
g
˙
M· i
0
πd · (T
0
−T
ex
)
.
And the temperature of the transported medium would remain constant and
equal to its initial value over the whole pipeline section. In such a case the
heat outgoing from the pipeline would be compensated by the heat extracted
by internal friction between the layers. Such an effect is used, for example,
in oil transportation along the TransAlaska oil pipeline (USA, see the cover
picture). Through good insulation of the pipeline the oil is pumped over
without preheating despite the fact that in winter the temperature of the
environment is very low.
From Eq. (1.45) follows the connection between the initial T
0
and ﬁnal T
L
temperatures of the transported medium. If in this formula we set x = L,
where L is the length of the pipeline section, we obtain
T
L
−T
ex
−T
⊗
T
0
−T
ex
−T
⊗
= exp
_
−
πdκL
C
v
˙
M
_
. (1.46)
Expressing nowfrom(1.46) the argument under the exponent and substituting
the result in Eq. (1.45), we get the expression for the temperature distribution
through the initial and ﬁnal values
T(x) −T
ex
−T
⊗
T
0
−T
ex
−T
⊗
=
_
T
L
−T
ex
−T
⊗
T
0
−T
ex
−T
⊗
_
x/L
. (1.47)
Exercise 1. The initial temperature of crude oil (ρ = 870 kg m
−3
, C
v
=
2000 J kg
−1
K
−1
, Q = 2500 m
3
h
−1
), pumping over a pipeline section (d =
800 mm, L = 120 km, i
0
= 0.002) is 55
◦
C. The temperature of the surrounding
medium is 8
◦
C. The heat insulation of the pipeline is characterized by the
heattransfer factor κ = 2 W m
−2
K
−1
. It is required to ﬁnd the temperature at
the end of the section.
Solution. Calculate ﬁrst the temperature T
⊗
:
T
⊗
=
gi
0
˙
M
πdκ
=
9.81 · 0.002 · 870 · (2500/3600)
3.14 · 0.8 · 2
∼
= 2.36 K.
Using Eq. (1.46) we obtain
T
L
−8 −2.36
55 −8 −2.36
= exp
_
−
3.14 · 0.8 · 2 · 120 · 10
3
2000 · 870 · (2500/3600)
_
,
from which follows T
L
∼
= 37.5
◦
C.
28 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
Exercise 2. By how much would the temperature of the oil (C
v
=
1950 J kg
−1
K
−1
) be raised due to the heat of internal friction when the
oil is transported by an oil pipeline (L = 150 km, d = 500 mm, i
0
= 0.004)
provided with ideal heat insulation (κ = 0)?
Solution. In this case it is impossible to use at once Eq. (1.45) since κ = 0. To
use Eq. (1.47) one should go to the limit at κ →0, therefore it would be better
to use Eq. (1.44)
ρC
v
v ·
dT
dx
= ρvgi
0
or C
v
·
dT
dx
= gi
0
,
from which T = gi
0
L/C
v
= 9.81 · 0.004 · 150 · 10
3
/1950
∼
= 3 K.
Exercise 3. It is required to obtain the temperature of oil pumping over the
pipeline section of length 150 km in crosssections x = 50, 100 and 125 km,
if the temperature at the beginning of the pipeline T
0
= 60
◦
C, that at the
end T
L
= 30
◦
C, and that of the environment T
ex
= 8
◦
C. The extracted heat of
internal friction may be ignored.
Solution. Using Eq. (1.46), one gets
T(x) −8
60 −8
=
_
30 −8
60 −8
_
x/L
and T(x) = 8 +52 · (0.4231)
x/150
.
Substitution in this formula of successive x = 50, 100 and 125 gives
T(50)
∼
= 47
◦
C; T(100)
∼
= 37.3
◦
C; T(125)
∼
= 33.4
◦
C.
2. For stationary ﬂow of a compressible medium, e.g. gas, the equation of heat
inﬂow (1.41) takes the form
ρvS
dJ
dx
= πd · q
n
.
In the general case, the gas enthalpy J is a functionof pressure and temperature
J = J(p, T), but for a perfect gas, that is a gas obeying the Clapeyronlawp = ρRT,
where R is the gas constant, the enthalpy is a function only of temperature
J = C
p
· T +const., where C
p
is the gas speciﬁc heat capacity at constant
pressure (C
p
> C
v
; C
p
−C
v
= R). Regarding C
p
= const. and taking as before
q
n
= −κ · (T −T
ex
), we transform the last equation to
C
p
˙
M
dT
dx
= −πdκ · (T −T
ex
)
or
dT
dx
= −
πdκ
C
p
˙
M
· (T −T
ex
).
1.8 Mathematical Modeling of OneDimensional Flows in Pipelines 29
The solution of this differential equation with initial condition T(0) = T
0
gives
T(x) −T
ex
T
0
−T
ex
= exp
_
−
πdκ
C
p
˙
M
x
_
, (1.48)
which is similar to the solution (1.45) for temperature distribution in an
incompressible ﬂuid. The difference consists only in that instead of heat
capacity C
v
in the solution (1.47) we use heat capacity C
p
and the temperature
T
⊗
taking into account the heat of internal friction is absent (for methane
C
p
∼
= 2230 J kg
−1
K
−1
; C
v
∼
= 1700 J kg
−1
K
−1
).
The temperature T
L
of the gas at the end of the gas pipeline section is found
from
T
L
−T
ex
T
0
−T
ex
= exp
_
−
πdκL
C
p
˙
M
_
(1.49)
with regard to which the distribution (1.47) takes the form
T(x) −T
ex
T
0
−T
ex
=
_
T
L
−T
ex
T
0
−T
_
x/L
(1.50)
allowing us to express the temperature through the initial and ﬁnal
temperatures.
Note that for a real gas the enthalpy J = J(p, T) of the medium depends not
only on temperature but also on pressure, so the equation of heat inﬂow has a
more complex form. By the dependence J(p, T) is explained, in particular, the
JouleThomson effect.
1.8
Complete System of Equations for Mathematical Modeling of OneDimensional
Flows in Pipelines
This system consists of the following equations.
1. Continuity equation (1.6)
∂ρS
∂t
+
∂ρvS
∂x
= 0;
2. Momentum (motion) equation (1.10)
ρ
_
∂v
∂t
+v
∂v
∂x
_
= −
∂p
∂x
−
4
d
τ
w
−ρg sinα(x);
3. Equation of mechanical energy balance (1.15)
∂
∂t
_
α
k
v
2
2
_
+v ·
∂
∂x
_
α
k
v
2
2
+P(ρ) +gz
_
= vg · i;
30 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines
4. Equation of total energy balance (1.37)
∂
∂t
__
α
k
v
2
2
+e
in
_
ρS
_
+
∂
∂x
__
α
k
v
2
2
+J
_
ρvS
_
= πd · q
n
−ρvgS
dz
dx
.
The number of unknownfunctions inthis equationis 10: ρ, v, p, S, e
in
, T, τ
w
,
i, q
n
, α
k
, while the number of equations is 4. Therefore there are needed
additional relations to close the system of equations. As closing relations the
following relations are commonly used:
•
equation of state p = p(ρ, T), characterizing the properties of the
transported medium;
•
equation of pipeline state S = S(p, T) characterizing the deformation ability
of the pipeline;
•
calorimetric dependences e
in
= e(p, T) or J = J(p, T);
•
dependence q
n
= −κ · (T −T
ex
) or more complex dependences
representing heat exchange between the transported medium and the
environment;
•
hydraulic dependence τ
w
= τ
w
(ρ, v, ˙ v, d, ν, . . .);
•
dependences α
k
= f (ρ, v, ν, d, . . .), or i =
˜
f (τ
w
),
characterizing internal structure of medium ﬂow.
To obtain closing relations a more detailed analysis of ﬂow processes is
needed. It is also necessary to consider mathematical relations describing
properties of the transported medium and the pipeline in which the medium
ﬂows.
The division of mechanics in which properties of a transported medium
such as viscosity, elasticity, plasticity and other more complex properties are
studied is called rheology.
31
2
Models of Transported Media
Algebraic relations connecting parameters of the transported medium such as
density, pressure, temperature and so on are called equations of state. Each of
these relations represents of course a certain schematization of the properties
of the considered medium and is only a model of a given medium. Let us
consider some models.
2.1
Model of a Fluid
By ﬂuid is meant a continuuminwhichthe interactionof the contacting interior
parts at rest is reduced only to the pressing force of pressure. If ﬂuid particles
interact along the surface element dσ with the unit normal n (Figure 2.1), the
force dF
n
with which the ﬂuid particles on one side of the element act on the
ﬂuid particles on the other side of the element is proportional to the area dσ,
is directed along the normal n and has a pressing action on them. Then
dF
n
= −pn dσ. (2.1)
The magnitude p of this force does not depend on the surface element
orientation and is called pressure.
Thus, p =  dF
n
/ dσ. The absence of tangential friction forces in the state of
rest models the fact that the ﬂuid takes the shape of the vessel it ﬁlls.
Further classiﬁcation of ﬂuids is dependent on whether or not tangential
friction forces are taken into account on exposure to ﬂuid ﬂow. In accordance
Figure 2.1 A scheme of force interactions in a ﬂuid.
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
32 2 Models of Transported Media
with this there are two models: the model of an ideal ﬂuid and the model of a
viscous ﬂuid.
2.2
Models of Ideal and Viscous Fluids
In the model of an ideal ﬂuid it is assumed that tangential friction forces
between ﬂuid particles separated by an elementary surface are absent, not only
in the state of rest but also in the state of ﬂow. Such a schematization
(or model) of a ﬂuid appears to be very fruitful when the tangential
components of interaction forces, that is friction forces, are far smaller
than their normal components, that is pressure forces. In other cases when
the friction forces are comparable with or even exceed the pressure forces the
model of an ideal ﬂuid has proved to be inapplicable. Hence expression (2.1)
for an ideal ﬂuid is true in the state of rest as well as in the state of
ﬂow.
In the model of a viscous ﬂuid tangential stresses resulting in ﬂuid ﬂow are
takeninto account. Let for example the ﬂuidlayers move as showninFigure 2.2.
Here u(y) is the velocity distribution in the ﬂow and y is the direction of a
normal to the elementary surface dσ.
In the model of a viscous ﬂuid it is accepted that the tangential stress τ
between the layers of the moving ﬂuid is proportional to the velocity difference
of these layers calculated per unit length of the distance between them, namely
to the velocity gradient du/ dy:
τ = µ
du
dy
. (2.2)
The tangential stress τ is deﬁned as the friction force between the ﬂuid layers
divided by the area of the surface separating these layers. Then the dimension
of the stress τ is
[τ] =
force
area
=
M×L/T
2
L
2
=
M
L ×T
2
.
In the SI system of units the stress τ is measured by Pa = kg m
−1
s
−2
.
Figure 2.2 Illustration of the deﬁnition of the viscous
friction law.
2.2 Models of Ideal and Viscous Fluids 33
The proportionality factor µ in the law of viscous friction (2.2) is called the
factor of dynamic viscosity or simply dynamic viscosity. Its dimension is
[µ] = [τ] ×T =
M
L ×T
In SI units µ is measured in kg m
−1
s
−1
and is expressed through poise P,
where 1 P = 0.1 kg m
−1
s
−1
. For example the dynamic viscosity of water is
equal to 0.01 P = 0.001 kg m
−1
s
−1
= 1 cP (centipoise).
The factor of kinematic viscosity or simply kinematic viscosity ν of a ﬂuid is
deﬁned as the ratio µ/ρ, therefore
[ν] =
µ
ρ
=
M/(L ×T)
M/L
3
=
L
2
T
.
In SI units ν is measured in m
2
s
−1
and expressed in stokes St, where
1 St = 10
−4
m
2
s
−1
. For example, the kinematic viscosity of water is equal to
0.01 St = 10
−6
m
2
s
−1
= 1 cSt (centistoke). The kinematic viscosity of benzene
is approximately equal to 0.6 cSt; that of diesel fuel is 4–9 cSt and that of
lowviscous oil 5–15 cSt.
The viscosity of oil and of almost all oil products depends on temperature.
As the temperature increases the viscosity decreases, whereas reduction in
temperature leads to viscosity enhancement. To calculate the dependence of
the kinematic viscosity ν on temperature T it different formulas can be used
including the Reynolds–Filonov formula
ν(T) = ν
0
· e
−κ·(T−T
0
)
(2.3)
in which ν
0
is the kinematic viscosity of a ﬂuid at temperature T
0
and κ (1/ K)
is an empirical factor. Equation (2.3) means that the ﬂuid viscosity varies
exponentially with temperature.
To use Eq. (2.3) it is necessary to know the factor κ or the viscosity ν
1
of
the same ﬂuid at another temperature T
1
. Then this factor is found from the
relation
κ =
ln(ν
0
/ν
1
)
(T
1
−T
0
)
. (2.4)
Exercise. The kinematic viscosity of summer diesel fuel at temperature +20
◦
C
is 5 cSt, whereas at a temperature of 0
◦
C it increases to 8 cSt. Determine the
viscosity of the same fuel at temperature +10
◦
C.
Solution. With Eq. (2.4) the factor κ = ln(5/8)/(0 −20)
∼
= 0.0235 is deter
mined. The viscosity is found fromEq. (2.3) ν = 5 · exp[−0.0235 · (10 −20)]
∼
=
6.3 cSt.
34 2 Models of Transported Media
2.3
Model of an Incompressible Fluid
A ﬂuid is called incompressible if its density does not vary when moving, that
is dρ/ dt = 0. If initially all ﬂuid particles had equal densities, that is the ﬂuid
was homogeneous, it remains homogeneous as before, namely ρ = ρ
0
= const.
while moving.
Of course an incompressible ﬂuid is only a model of a real medium,
since as is known absolutely incompressible media do not exist, but when
the change in ﬂuid density in a certain process can be neglected the model
of an incompressible ﬂuid may be very useful. For example, under normal
conditions water density is 1000 kg m
−3
, benzene ≈ 735–750 kg m
−3
, diesel
fuel ≈ 840 kg m
−3
, oil ≈ 870–900 kg m
−3
etc.
If the ﬂuid density depends only on pressure, that is ρ = ρ(p), the condition
of incompressibility is dρ/ dt = 0 being tantamount to dρ/ dp = 0.
2.4
Model of Elastic (Slightly Compressible) Fluid
There are processes which require that account is taken of even a small
variation in ﬂuid density. In such a case the socalled elastic ﬂuid model is often
used. In this model the ﬂuid density depends on pressure as follows
ρ(p) = ρ
0
[1 +β(p −p
0
)] (2.5)
where β (1/Pa) is the compressibility factor; ρ
0
the ﬂuid density at normal
pressure p
0
. The compressibility factor is the inverse of the elastic modulus
K (Pa), that is K = 1/β. Then Eq. (2.5) reduces to
ρ(p) = ρ
0
1 +
p −p
0
K
. (2.6)
Mean values of the elastic modulus for oil and oil products vary in the range
1400–1500 MPa, so that K ≈ 1.4 −1.5 · 10
9
Pa. It follows that the deviation of
density ρ from the normal ρ
0
: ρ = ρ
0
· (p −p
0
)/K is very small for oil and
oil products. For example for a ﬂuid with ρ
0
= 850 kg m
−3
at p −p
0
= 5 MPa
(≈ 50 atm) the deviation ρ is 2.8 kg m
−3
.
2.5
Model of a Fluid with Heat Expansion
The expansion of different media on heating and subsequent compression on
cooling is taken into account in the ﬂuid model with volume expansion. In the
2.5 Model of a Fluid with Heat Expansion 35
Table 2.1 Factor of volume expansion ξ.
Density ρ, kg m
−3
Factor ξ, K
−1
720–739 0.001183
740–759 0.001118
760–779 0.001054
780–799 0.000995
800–819 0.000937
820–839 0.000882
840–859 0.000831
860–880 0.000782
model to be considered the density ρ is a function of temperature T, so that
ρ = ρ(T)
ρ(T) = ρ
0
[1 +ξ(T
0
−T)] (2.7)
in which ξ (1/K) is the factor of volume expansion, ρ
0
and T
0
are the density and
temperature of the ﬂuid under normal conditions (often T
0
= 293 K (20
◦
C);
ρ
0
= ρ(p
0
, T
0
); p
0
= p
st
= 101325 Pa). The values of the factor ξ for oil and oil
products are given in Table 2.1.
From Eq. (2.6) it follows that on heating, that is at T > T
0
, the ﬂuid expands,
that is ρ < ρ
0
whereas at T < T
0
, ρ > ρ
0
, that is the ﬂuid is compressed.
Exercise. The density ρ
0
of benzene at 20
◦
C is 745 kg m
−3
. Determine the
density of the same benzene at 10
◦
C.
Solution. Using Eq. (2.6) and Table 2.1 we have
ρ(10
◦
C) = 745 · [1 +0.00118 · (20 −10)] = 753.3 kg m
−3
.
Thus the density is increased by 8.3 kg m
−3
.
There is also used a model for ﬂuid expansion with regard to baric and
heat expansion. In such a model the density is a function of pressure and
temperature ρ = ρ(p, T) called the state equation
ρ(p, T) = ρ
0
1 +ξ(T −T
0
) +
p −p
0
K
. (2.8)
Here p
0
, T
0
are the nominal pressure and temperature satisfying the relation
ρ
0
= ρ(p
0
, T
0
).
Exercise. The density of benzene ρ
0
at 20
◦
C and atmospheric pressure
p
at
≈ 0.1 MPa is 745 kg m
−3
. Determine the density of the same benzene at
temperature 10
◦
C and pressure 6.5 MPa.
Solution. Using Eq. (2.7) and Table 2.1 we get ρ(p, T) = 745 · [1 +0.00118 ·
(20 −10) +(6.5 −0.1) · 10
6
/(1.5 · 10
9
)] = 757 kg m
−3
. The density is in
creased by 12 kg m
−3
.
36 2 Models of Transported Media
2.6
Models of NonNewtonian Fluids
Fluids modeled by condition (2.2) of viscous friction are called Newtonian
viscous ﬂuids in accordance with the name of the law (2.2). The quantity
˙ γ = du/ dy having the sense of velocity gradient with dimension s
−1
, is called
the shear rate. The linear dependence (2.2) between the tangential friction
stress τ and the shear rate ˙ γ is shown in Figure 2.3. This dependence states:
‘‘Tangential stresses arising in a medium having been modeled by a Newtonian
viscous ﬂuid are proportional to the shear rate of the ﬂuid layers relative to
each other. When the shear rate vanishes, the tangential friction stresses also
disappear.’’
The dynamic viscosity of ﬂuid µis represented in this model by the slope of a
straight line on the plane ( ˙ γ, τ) : µ = tanϕ, where ϕ is the angle of inclination
of the straight line to the abscissa. Many experiments have shown that the
model of a Newtonian viscous ﬂuid well schematizes processes taking place
in many real ﬂuids.
And yet there exist dependences of τ on ˙ γ (ﬂow curve) that differ signiﬁcantly
from that depicted in Figure 2.3. Such ﬂuids are called nonNewtonian.
As an example of a nonNewtonian ﬂuid is a model of a power Ostwald
ﬂuid (Wilkinson, 1960)
τ = k ·
du
dy
n−1
·
du
dy
(2.9)
in which the relation of the tangential friction stresses between ﬂuid layers has
a power nature. In other words the apparent viscosity ˜ µ of such a ﬂuid does
not remain constant as in the model of Newtonian ﬂuid, but depends on the
characteristics of the ﬂow, namely
˜ µ = k ·
du
dy
n−1
. (2.10)
In this model k and n are factors. Fluids with n < 1 are called pseudoplastic
ﬂuids. These models are applied to describe the behavior of suspension ﬂows,
Figure 2.3 A model of a Newtonian viscous ﬂuid.
2.6 Models of NonNewtonian Fluids 37
Figure 2.4 Models of nonNewtonian
ﬂuids: 1, pseudoplastic ﬂuids; 2, dilatant
ﬂuids.
that is viscous ﬂuids with suspended small particles. Flow curves of such ﬂuids
have the form of curve 1 in Figure 2.4 (Wilkenson, 1960).
Fluids with n > 1 are called dilatant ﬂuids. Starch glue is an example of ﬂuid
whose behavior is described by the dilatant model. Flow curves of these ﬂuids
have the form of curve 2 in Figure 2.4 (Wilkenson, 1960).
Model of viscousplastic medium with limit shear stress; model of Shve
dov–Bingham ﬂuid (Wilkenson, 1960). There are ﬂuids in which the stresses
between the ﬂuid layers are sufﬁciently well described by the following
relations
τ = τ
0
+µ
du
dy
, at τ > τ
0
;
˙ γ =
du
dy
= 0, at τ ≤ τ
0
; (2.11)
τ = −τ
0
+µ
du
dy
, at τ < −τ
0
.
These expressions imply that the ﬂow of such a ﬂuid does not begin as long as
the absolute value of the tangential friction stress τ does not exceed a limiting
value τ
0
, being the characteristic of the given mediumand called the limit shear
stress. In this case ˙ γ = 0. At τ ≥ τ
0
and ˙ γ = 0 the medium ﬂows as a viscous
ﬂuid.
Real media whose properties are satisfactory modeled by the viscousplastic
Bingham ﬂuid are, for example, highparafﬁnaceous and solidifying oils, mud
solutions, lacquers, paints and other media.
Flow curves of viscousplastic media are shown in Figure 2.5.
Figure 2.5 Flow curves of a
viscousplastic ﬂuid.
38 2 Models of Transported Media
2.7
Models of a Gaseous Continuum
We proceed now to the description of the basic models used for gas ﬂow.
First let us consider the properties common to all gases. One such property is
that for all gases in a state of thermodynamic equilibrium there is a relation
between pressure p, absolute temperature T and density ρ (or speciﬁc volume
υ = 1/ρ)
(p, υ, T) = 0 (2.12)
called the equation of state. The physical nature of this fact is discussed in
textbooks on statistical physics and thermodynamics. In most models it is
also assumed that when the motion starts the relation (2.12) remains. This
means that the establishment of thermodynamic equilibrium happens much
faster than the nonequilibrium caused by the resulting ﬂow.
The speciﬁc form of the dependence (2.12) is set in the course of socalled
calorimetric measurements, but for the majority of gases this dependence
has one and the same distinctive features. Geometrical representation of
the dependence (2.12) has the form of a twodimensional surface in a three
dimensional space of variables (p, υ, T). Figure 2.6 shows isotherms of real
gases representing intersections of this surface with planes T = const.
For all gases there exists the socalled critical isotherm, depicted in Figure 2.6
by the heavy line, above and belowwhich the properties of the gas are different.
If T ≥ T
cr
, where T
cr
is the critical temperature for a given gas, the gas at any
elevation of pressure remains in the gaseous state. If T < T
cr
, then for each
temperature T there exists a value of pressure p at which the gas begins to
change into the liquid phase, its speciﬁc volume being increased from υ
to
υ
, after which the resulting medium acquires the properties of a liquid.
The point K is called critical point of a given gas, the quantities (T
cr
, p
cr
, υ
cr
)
accounting for the individual properties of a gas are constants.
For example, for methane CH
4
, which is the major constituent of consists
natural gas, T
cr
= 190.55 K and p
cr
= 4.641 MPa. This means that if the gas
Figure 2.6 Gas isotherms.
2.7 Models of a Gaseous Continuum 39
temperature exceeds 190.55 K, this gas will be changed into the liquid state
without the need for any pressure increase.
2.7.1
Model of a Perfect Gas
If the gas pressure is not too high and the temperature not too low, the
isotherms of all gases are similar to each other (see the righthand part of
Figure 2.6 enclosed in the dotted oval line) and with a high degree of accuracy
can be approximated by hyperbolas representing the fact that the pressure p is
inversely proportional to the speciﬁc volume υ.
Under given conditions the interaction of the molecules of a real gas is
independent of the formof the molecules, that is of the spatial conﬁguration of
the constituent atoms, and is determined only by their total mass. Figuratively
speaking, the molecules behave as balls differing in their mass, therefore the
number of parameters characterizing the gas decreases from three to one,
namely the molecular weight µ
g
.
To characterize the thermodynamic state of gases ina givenrange of pressure
and temperature the model of a perfect gas is used. Equation (2.11) of the gas
state then has the simple form
p =
RT
υ
or p = ρRT (2.13)
where R is the only constant in the equation and is called the gas constant
R = R
0
/µ
g
, where R
0
is the universal gas constant, equal to 8314 J mol
−1
K
−1
.
Thus all gas constants for a perfect gas depend only on the molecular weight.
They are: for methane (µ
g
∼
= 16 kg kmol
−1
) R = 8314/16
∼
= 520 J kg
−1
K
−1
);
for oxygenO
2
(µ
g
∼
= 32 kg kmol
−1
) R = 8314/32
∼
= 260 J kg
−1
K
−1
); for carbon
dioxide CO
2
(µ
g
∼
= 44 kg kmol
−1
) R = 8314/44
∼
= 189 J kg
−1
K
−1
; for air
(µ
g
∼
= 29 kg kmol
−1
) R = 8314/29
∼
= 287 J kg
−1
K
−1
.
Equation (2.12) connecting the density, pressure and temperature of a gas is
called the Clapeyron equation.
The model of a perfect gas operates sufﬁciently well over a range of not too
high pressures and moderate temperatures.
2.7.2
Model of a Real Gas
Despite the fact that the name of this model contains the word ‘‘real’’,
one is dealing only with the next, more general, schematization of a gas
model. From Figure 2.6 it follows that the hyperbolic dependence (2.13)
does not suit observations of gas behavior with increase of pressure and
decrease of temperature. Hence, in processes of gas pipeline transportation
40 2 Models of Transported Media
and underground gas storage, where the pressure may be 5–15 MPa the model
of a perfect gas would give improper results were it to be used in calculations.
There is a model of more general form than the model of a perfect gas. This
is the model of a real gas. One can ﬁnd the details of such a model for example
in Porshakov et al., 2001. Without dwelling on the details of its derivation we
note that the mathematical form of this equation is represented as follows
p =
Z(p
r
, T
r
) · RT
υ
or p = Z(p
r
, T
r
) · ρRT (2.14)
differing from Eq. (2.13) by the insertion of the dimensionless factor Z(p
r
, T
r
)
called the overcompressibility factor, being a function of two parameters: the
reduced pressure p
r
and the reduced temperature T
r
, where
p
r
=
p
p
cr
, T
r
=
T
T
cr
.
Here p
cr
and T
cr
are the critical pressure and temperature.
Hence the model (2.14) takes into account not only the molecular weight
of a gas through the constant R but also its thermodynamic parameters such
as critical pressure and temperature. It is evident that for moderate values
of pressure and temperature Z ≈ 1 and the model (2.14) transforms into the
model of a perfect gas. For a real gas Z < 1.
Graphs of the function Z(p
r
, T
r
) are shown in Figure 2.7.
Exercise. It is required to determine the overcompressibility factor Z of a gas
with p
cr
= 4.6 MPa, T
cr
= 190 K at p = 7.5 MPa and T = 288 K.
Solution. Let us calculate ﬁrst the reduced parameters of state: p
cr
= 1.63;
T
cr
=1.52. From the plot in Figure 2.7 we determine Z
∼
= 0.86.
There are a lot of approximating formulas to calculate the factor Z. In fact the
case is to approximate the equation of state (2.12). However, the properties of a
real gas are so complicated that we do not have universal formulas appropriate
for all gases over the whole range of governing parameters. Therefore in
different cases one should use different approximations. In particular, to
simulate processes in gaspipelines we can use
Z(p
r
, T
r
)
∼
= 1 −0.4273 · p
r
· T
−3.668
r
(2.15)
or
Z(p
r
, T
r
)
∼
= 1 −0.0241 ·
p
r
θ
, (2.16)
where
θ = 1 −1.68T
r
+0.78T
2
r
+0.0107T
3
r
.
It should be particularly emphasized that Eqs. (2.15) and (2.16) are no more
than approximations of the state equation of a real gas.
2.7 Models of a Gaseous Continuum 41
Figure 2.7 Graphs of Z(p
r
, T
r
) for natural gas.
Exercise. Determine the value of the overcompressibility factor of a gas with
p
cr
= 4.6 MPa, T
cr
= 190 K at p = 7.5 MPa and T = 288 K (see the previous
example).
Solution. The reduced parameters of state are:
p
r
=
7.5
4.6
∼
= 1.63; T
r
=
288
190
∼
= 1.52.
With formulas (2.15) we obtain
Z = 1 −0.4273 · 1.63 · 1.52
−3.668
∼
= 0.850.
Similar calculation in accordance with Eq. (2.16) gives:
θ = 1 −1.68 · 1.52 +0.78 · 1.52
2
+0.0107 · 1.52
3
∼
= 0.2861,
Z = 1 −0.0241 ·
1.63
0.2861
∼
= 0.863.
42 2 Models of Transported Media
It is seen that the error in calculation with Eq. (2.15) is ≈ 2.3%, whereas with
Eq. (2.16) it is less ≈ 0.8%. But this does not mean that the Eq. (2.16) is more
precise than Eq. (2.15) because both formulas are only approximations of the
state equation of a real gas.
2.8
Model of an Elastic Deformable Pipeline
In the schematization of ﬂuid and gas ﬂow processes in pipelines models of a
pipeline are also used.
The simplest model of the pipeline is a model of nondeformable pipe, that is of
a cylinder with constant invariable internal diameter d
0
and wall thickness δ.
The external diameter D = d
0
+2δ of the pipeline in this model remains also
constant. The model of a nondeformable pipeline appears to be very useful
when researching many technological processes of oil and gas transportation.
However, in some cases, for example in studying the phenomenon of
hydraulic hammer, the model of a nondeformable pipe proves to be inadequate
to perceive the point of the phenomenon and it is necessary to use the more
complicated model of an elastic deformable pipeline.
Experience shows that the volume of the internal space of the pipeline varies
insigniﬁcantly with changes in temperature and pressure of the transported
medium.
In order to account for the volume expansion of a pipeline when the
temperature T deviates from its nominal value T
0
one can use
V(T) = V
0
[1 +2α
p
· (T −T
0
)], (2.17)
where α
p
is the volume expansion factor of the metal from which the pipeline
is produced (for steel α
p
≈ 3.3 · 10
−5
K
−1
).
Exercise. Howdoes the volume of the internal space of a steel pipeline section
with D = 530 mm, δ = 8 mm, L = 120 km change during even cooling by
5 K?
Solution. Using Eq. (2.17) we get:
V(T) −V
0
=
3.14 ·
0.514
2
4
· 2 · (−5) · 3.3 · 10
−5
· 120 000·
∼
= −8.22,
that is the volume decreases by more than by 8 m
3
.
The volume of the pipeline internal space is changed to a greater extent
with variation in the difference between the internal and external pressures.
The simplest formula to calculate changes originated by this phenomenon has
beensuggested by Joukowsky in his work ‘‘On hydraulic hammer in watersupply
pipes’’ (1899). Its derivation is illustrated in Figure 2.8.
2.8 Model of an Elastic Deformable Pipeline 43
Figure 2.8 Derivation of the formula
for the crosssection area change in an
elastic deformable pipeline.
The equation of equilibrium of the upper half of the pipe shell shown in
Figure 2.8 by a heavy line, under the action of a pressure difference (p −p
0
)
and circumferential stresses σ resulting in the pipe metal has the following
form
(p −p
0
) · d = σ · 2δ. (2.18)
Hooke’s law of elasticity as applied to the deformed middle ﬁlament shown in
Figure 2.8 by a dotted line gives the following relation
σ = E ·
π · (d −d
0
)
π · d
0
(2.19)
where E is the Young’s modulus of the pipe material (for steel E ≈ 2 · 10
5
MPa).
Insertion of σ from Eq. (2.18) into Eq. (2.19) and further replacing factor
d by d
0
due to the smallness of the wall thickness as compared to the pipe
diameter gives the dependence of the pipe diameter increment d = d −d
0
on the difference p = p −p
0
of the internal and external pressures
d =
d
2
0
2Eδ
· p. (2.20)
Here d
0
may be taken as the internal diameter of the pipe.
Remark. When deriving Eq. (2.20) it was assumed that axial stresses in the
pipe are absent. Such a state of the pipe is called a planestressstate. But in many
cases this assumption is not valid, in particular in steel welded pipelines used
in the oil industry there is a planedeformable state in which radial expansion of
the pipe generates axial stresses. In such cases Eq. (2.20) should be replaced
by the more general formula
d =
d
2
0
(1 −ν
2
P
)
2Eδ
· p (2.21)
where ν
P
is the Poisson ratio. However, the correction is insigniﬁcant since
for steel pipelines ν
2
P
≈ 0.078.
44 2 Models of Transported Media
From Eq. (2.20) ensue two useful formulas: one for the increment S of
the area of the pipe crosssection and the other for the increment V of the
volume of a pipeline section with length L
S =
πd
3
0
4Eδ
· p, V =
πd
3
0
L
4Eδ
· p (2.22)
and associated formulas
S =
πd
3
0
1 −ν
2
P
4Eδ
· p, V =
πd
3
0
L ·
1 −ν
2
P
4Eδ
· p. (2.23)
Exercise. It is required to calculate the increase in the diameter and volume
of a section of a steel pipeline with D = 530 mm, δ = 8 mm, L = 120 km after
build up of pressure by 5.0 MPa.
Solution. Using Eq. (2.16) we get
d =
0.514
2
2 · 2 · 10
11
· 0.008
· 5 · 10
6
∼
= 0.0004 m = 0.4 mm.
From Eq. (2.17) follows
V =
3.14 · 0.514
3
· 120000
4 · 2 · 10
11
· 0.008
· 5 · 10
6
∼
= 40 m
3
.
For simultaneous deviation of pressure and temperature from their nominal
values p
0
and T
0
it is allowable to use the formula
V(T) = V
0
1 +2α
p
· (T −T
0
) +
d
0
Eδ
· (p −p
0
)
. (2.24)
There are also more complicated models of pipelines taking into account the
viscousplastic properties of the pipe material. Such models are applicable for
pipelines made from synthetic materials, e.g. from plastics.
45
3
Structure of Laminar and Turbulent Flows in a Circular Pipe
In the ﬁrst chapter the ﬂow of a medium transported in a pipeline was
considered in the framework of a onedimensional model, that is a model in
which the ﬂow is described by characteristics of the ﬂuid velocity, density,
pressure, temperature and others averaged over the pipe crosssection. All
characteristics of the ﬂow depended only on the longitudinal coordinate x
of the crosssection and time t. In such a description additional closure
relations reﬂecting the relations between the average parameters of the ﬂow
were needed. For example, relation (1.26) connects the mechanical energy
dissipation with the average velocity of the ﬂow, relation (1.29) expresses the
tangential friction stress at the internal surface of a pipeline through average
parameters of the ﬂow and so on.
It was noted above that to get a closure relation it is necessary to scrutinize
not only onedimensional but also spatial ﬂows taking place in the ﬂow of
transported medium inside the pipeline. Let us consider in greater detail
such ﬂows with regard to the distribution of the parameters over the pipeline
crosssection.
3.1
Laminar Flow of a Viscous Fluid in a Circular Pipe
First let us consider the laminar ﬂow of ﬂuid in a circular pipe with radius
r
0
(Figure 3.1). Such a ﬂow has only one axial velocity component u(r),
dependent on the radial coordinate r equal to the distance from a point under
consideration to the pipe wall.
Next separate inside the ﬂow region a cylinder of arbitrary radius r (r ≤ r
0
)
and write the balance equation of all the forces acting on the cylinder
(p
1
−p
2
) · πr
2
= τ(r) · 2πrL
where τ(r) is the tangential friction stress at the lateral surface of the separated
cylinder. From here it follows that the absolute value of the tangential stress τ
between the ﬂuid layers is proportional to the cylinder radius r
τ(r) =
1
2
·
p
L
· r. (3.1)
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
46 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Figure 3.1 Calculation of laminar ﬂuid ﬂow.
If we now replace τ by its expression through the velocity gradient du/ dr in
accordance with the law of viscous friction (2.2) and take into account that
τ(r) < 0, we get the differential equation
−µ
du
dr
=
1
2
·
p
L
· r (3.2)
for the velocity u(r). This equation should be solved for the boundary condition
u = 0 at r = r
0
, which is the socalled sticking condition (the velocity at the
internal surface of the pipe vanishes). As a result we obtain
u(r) = u
max
·
_
1 −
r
2
r
2
0
_
; u
max
=
p · r
2
0
4 µL
. (3.3)
It is seen that the ﬂuid velocity distribution over the pipe crosssection has a
parabolic form with maximal value u
max
at the pipe axis, that is at r = 0.
The ﬂuid ﬂow rate Q, that is the ﬂuid volume ﬂowing through a unit
crosssection of the pipe in unit time, is equal to
Q =
_
r
0
0
u(r)2πr dr = 2πu
max
_
r
0
0
r ·
_
1 −
r
2
r
2
0
_
dr =
1
2
πu
max
r
2
0
or with regard to Eq. (3.3)
Q =
r
4
0
πp
8 µL
. (3.4)
The relation (3.4) is called the Poiseuille formula. It gives the connection
between the ﬂow rate of laminar ﬂuid ﬂow in a circular pipe and the pressure
drop causing the ﬂow.
Let us introduce the mean ﬂowratevelocity v which when multiplied by the
pipe crosssection area gives the ﬂuid ﬂow rate, that is v = Q/πr
2
0
. Then one
obtains very useful formulas
v =
r
2
0
p
8 µL
= 2 · u
max
or v =
d
2
p
32 µL
(3.5)
where d = 2r
0
is the internal diameter of the pipe.
3.2 Laminar Flow of a NonNewtonian Power Fluid in a Circular Pipe 47
With the help of these relations it is possible to calculate the tangential stress
τ
w
at the internal surface of the pipeline. From Eq. (3.1) follows
τ
w
 =
1
2
·
p
L
· r
0
.
Substitution of p/L from Eq. (3.5) with regard to Eq. (1.29) yields
τ
w
 =
r
0
2
·
32µv
d
2
= C
f
·
ρv
2
2
=
λ
4
·
ρv
2
2
Fromhere follows Stokes formula (1.30) for the factor λ of hydraulic resistance
in laminar ﬂow of a viscous incompressible ﬂuid in a circular pipe
λ =
d
4
·
32 µv
d
2
·
8
ρv
2
=
64
vd/(µ/ρ)
=
64
Re
. (3.6)
Exercise. Calculate the hydraulic resistance factor of oil λ (ν = 25 cST) in
laminar ﬂow in a circular pipe with diameter 50 mm and ﬂow rate 1 l s
−1
.
Solution. First determine the mean velocity of the ﬂow
v =
Q
πd
2
/4
=
0.001
3.14 · 0.05
2
/4
∼
= 0.51 m s
−1
.
Then calculate the Reynolds number
Re =
vd
ν
=
0.51 · 0.05
25 · 10
−6
= 1020
Since 1020 < Re
cr
= 2300, the ﬂow is laminar. In accordance with Eq. (3.6)
the hydraulic resistance factor is equal to
λ =
64
Re
=
64
1020
∼
= 0.063.
3.2
Laminar Flow of a NonNewtonian Power Fluid in a Circular Pipe
Just as we considered the laminar ﬂow of a Newtonian ﬂuid so we can consider
the laminar ﬂow of a nonNewtonian Ostwald power ﬂuid in a circular pipe (see
Chapter 2). The tangential stress τ for this ﬂuid is related to the shear rate
˙ γ = du/ dr by the dependence (2.9)
τ = k ·
¸
¸
¸
¸
du
dr
¸
¸
¸
¸
n−1
·
du
dr
so that the equilibrium Eq. (3.2) takes the form
−k ·
¸
¸
¸
¸
du
dr
¸
¸
¸
¸
n−1
·
du
dr
=
1
2
·
p
L
· r
48 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Since du/dr < 0, the last relation transforms to
du
dr
= −
_
p
2kL
_
1/n
· r
1
n
.
Integration of this equation with the sticking boundary condition u(r
0
) = 0 at
the pipe wall gives the velocity distribution
u(r) = −
n
n +1
·
_
p
2kL
_
1/n
·
_
r
n+1
n
−r
n+1
n
0
_
. (3.7)
The maximal velocity of the ﬂow u
max
is achieved at the pipe axis r = 0 as in
the case of a viscous ﬂuid and is equal to
u
max
=
n
n +1
·
_
p
2kL
_
1/n
· r
n+1
n
0
. (3.8)
The ﬂow rate of the ﬂuid Q is calculated by the formula
Q = 2π
_
r
0
0
r · u(r) dr
= 2π ·
n
n +1
·
_
p
2kL
_
1/n
·
_
r
0
0
r ·
_
r
n+1
n
0
−r
n+1
n
_
dr
Performing integration one gets after some algebra
Q =
πr
0
3
n
3n +1
·
_
r
0
p/L
2k
_
1
n
. (3.9)
It is reasonable that at n = 1, k = µ Eq. (3.9) converts to Eq. (3.4) known as the
Poiseuille law (Wilkenson, 1960).
Introducing the mean ﬂowratevelocity v as in the previous section and the
generalized Reynolds number Re
∗
according to the relations
v =
Q
πr
0
2
and Re
∗
=
v
2−n
· d
n
0
k/ρ
where d
0
= 2r
0
, Eq. (3.9) could be written in the habitual form of the
Darcy–Weisbach law
p
L
= λ ·
1
d
0
·
ρv
2
2
where the hydraulic resistance factor λ is
λ =
8 ·
_
6n +2
n
_
n
Re
∗
. (3.10)
3.3 Laminar Flow of a ViscousPlastic Fluid in a Circular Pipe 49
In the same way Eq. (3.10) at n = 1, k = µ transforms to the Stokes Eq. (1.30)
given earlier.
To determine the rheological properties of oil and oil products special devices
called viscometers are oftenused. The most widespread are capillary viscometers.
Operation of all capillary viscometers is based on the determination of the time
of the outﬂowof a ﬁxed portion of the ﬂuid under test, fromthe device chamber
through a narrow cylindrical tube (capillary). This time is calculated with the
use of Eqs. (3.4) and (3.9) replacing in them the pressure gradient p/L by ρg,
where g is the acceleration due to gravity. The ﬂow rate Q in the viscometer of
the Ostwald power ﬂuid under consideration takes the following form
Q =
πr
0
3
n
3n +1
·
_
r
0
· ρg
2k
_
1
n
=
πr
0
3
n
3n +1
·
_
r
0
· g
2 · k/ρ
_
1
n
. (3.11)
Exercise. In order to determine the properties of oil experiments are carried
out on a free outﬂow of a portion (100 ml) of oil from the viscometer chamber.
In the ﬁrst experiment the outﬂow ﬂows through a cylindrical capillary with
internal radius 1 mm and in the second experiment through a similar capillary
with internal radius 1.5 mm. In the ﬁrst case the time of ﬂuid outﬂowis 1000 s,
in the second 180 s. The ﬂow of the oil is modeled by the law of a power ﬂuid.
It is required to determine the constants n and k/ρ of the power ﬂuid model.
Solution. From Eq. (3.11) it follows that the ratio of ﬂow rates in both
cases is Q
1
/Q
2
= (r
1
/r
2
)
3+1/n
. Since the ratio is inversely proportional to the
times of the ﬂuid outﬂows, we use the equation to determine n, namely
180/1000 = (2/3)
3+1/n
. The solution of this equation gives n
∼
= 0.81.
Using further the results of the ﬁrst experiment we get
0.0001
1000
=
3.14 · 0.81 · 0.001
3
3 · 0.81 +1
·
_
9.81 · 0.002
4 · k/ρ
_
1/0.81
,
from which we get k/ρ
∼
= 0.92 · 10
−6
m
2
s
−1.19
.
3.3
Laminar Flow of a ViscousPlastic Fluid in a Circular Pipe
Consider the laminar ﬂow of another nonNewtonian ﬂuid – viscousplastic
Bingham ﬂuid (Wilkenson, 1960) (see Chapter 2 and Figure 3.2).
The tangential stress τ in this ﬂuid is related to the shear rate ˙ γ = du/dr by
the dependence
τ = τ
0
+µ
du
dr
, (du/dr > 0, τ ≥ τ
0
) (3.12)
where τ
0
is the limit shear stress. Due to the existence of such stress the ﬂow
does not begin immediately after application of the pressure difference to the
50 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Figure 3.2 A scheme of viscousplastic ﬂuid ﬂow.
pipe ends but only when this difference exceeds the shear strength, namely
when it obeys the following inequality
p · πr
0
2
> 2πr
0
L · τ
0
or
r
0
· p/L
2τ
0
> 1. (3.13)
If the condition (3.13) is fulﬁlled then, near the internal surface of the pipe
(a < r < r
0
), the ﬂow of the ﬂuid starts. On approaching the pipe center where
the shear rate decreases the tangential stresses are also reduced and at some
distance a from the pipe axis, just where τ(a) = τ
0
, the ﬂuid begins to move
as a rigid core. Therefore the region (0 ≤ r ≤ a) of the ﬂow is called the ﬂow
core.
It is evident that the following relations are valid
τ(r) = τ
w
·
r
r
0
at a ≤ r ≤ r
0
,
τ(a) = τ
0
= τ
w
·
a
r
0
and
a
r
0
=
τ
0
τ
w
=
2τ
0
r
0
· p/L
.
The velocity distribution in the annular region a ≤ r ≤ r
0
satisﬁes the equation
µ
du
dr
= τ(r) −τ
0
= τ
0
·
_
r
a
−1
_
.
Integration of this equation over r from r = r
0
to an arbitrary r with regard to
boundary condition u(r
0
) = 0 yields
u(r) =
r
0
τ
0
2µ
_
(r/r
0
)
2
−1
a/r
0
−2 ·
r
r
0
+2
_
. (3.14)
In particular the core velocity u(a) is equal to
u(a) = −
r
0
τ
0
2µ
·
(1 −a/r
0
)
2
a/r
0
. (3.15)
The ﬂow rate Q of the ﬂuid can be obtained using Eqs. (3.14) and (3.15)
−Q = 2π
_
r
0
a
r · u(r) dr +πa
2
· u(a).
3.4 Transition of Laminar Flow of a Viscous Fluid to Turbulent Flow 51
Simple rearrangements lead to the following expression
Q =
πr
0
4
p/L
8µ
_
1 −
4
3
·
_
2τ
0
r
0
· p/L
_
+
1
3
·
_
2τ
0
r
0
· p/L
_
4
_
. (3.16)
As would be expected this relation at τ
0
= 0 becomes the known Poiseuille
formula (3.4) for the ﬂow rate of ﬂuid in a circular pipe (Wilkenson, 1960).
If we introduce two dimensionless parameters Re =
v·2r
0
ν
, the Reynolds
number and I =
τ
0
·2r
0
µv
, the Ilyushin number and take into account the relation
2τ
0
/(r
0
· p/L) = 8I/λRe, the equality (3.16) could be represented as
λ =
64
Re
·
1
1 −4/3 · (8I/λRe) +1/3 · (8I/λRe)
4
. (3.17)
If τ
0
= 0, then I = 0 and consequently λRe = 64, that is Eq. (3.17) converts to
the known Stokes formula (1.30) for laminar ﬂow of a viscous ﬂuid. In the
general case the product λRe depends on the Ilyushin number I. To determine
this product one should resolve Eq. (3.17) with respect to λRe for each value of
the parameter I (Romanova, 1977).
3.4
Transition of Laminar Flow of a Viscous Fluid to Turbulent Flow
With increase in the velocity of viscous ﬂuid ﬂow in a pipe the laminar ﬂow
loses hydrodynamic stability and changes into turbulent ﬂow. This ﬂow is
characterized by ﬂuctuating motions of the ﬂuid, generation and development
of eddies and intensive mixing. A lot of theoretical investigations performed
by many outstanding mathematicians and physicists have been devoted to
the problem of stability of laminar ﬂuid ﬂows and the determination of
the conditions and criteria of the transition from laminar to turbulent ﬂow
regimes.
The criterion of transition of laminar ﬂow into a turbulent one is the
Reynolds number Re, representing a dimensionless parameter formed by the
dimension parameters characterizing the ﬂow and the pipeline: Re = vd/ν, so
that at Re < Re
cr
= 2300 the ﬂow is laminar whereas at Re > Re
cr
= 2300 it is
turbulent.
The latter conclusion is true when the critical velocity v
cr
of the ﬂow is
determined only by ﬂow parameters such as d, ρ and µ(ν = µ/ρ). In fact, the
conditions of transition from laminar to turbulent ﬂow are in many respects
determined also by such parameters of the pipeline as the roughness of its
internal wall surface . If we accept that v
cr
= f (d, µ, ρ, ) then, in accordance
with the Itheorem the condition of laminar ﬂow transition to turbulent ﬂow
takes the form
v
cr
· d
µ/ρ
=
˜
f
_
d
_
or Re
cr
=
˜
f (ε)
52 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
where ε is the relative roughness of the pipeline internal wall surface. Thus
at Re < Re
cr
(ε) the ﬂuid ﬂow is laminar, while at Re > Re
cr
(ε) the ﬂow is
turbulent. The presence of the additional parameter ε besides the Reynolds
number means that there is not a singlevalued determined boundary of the
transition from laminar to turbulent ﬂow, since the critical Reynolds number
depends on the degree of preparation of the pipeline and the ﬂuid for the test
experiments. It is known, for example, that in this manner it was possible to
lengthen the transition of laminar ﬂow to turbulent ﬂow up to a value of the
critical Reynolds number equal to 6000–12 000.
The hydraulic resistance factor λ in the transition region does not have
stable values, it is only characterized by drastic lowering. Some formulas
for λ in this region will be considered further. The factor λ takes a stable
value in establishing the developed turbulent ﬂow in the pipe, that is at
Re > 10 000.
3.5
Turbulent Fluid Flow in a Circular Pipe
Consider turbulent ﬂow of a viscous ﬂuid developed in a circular pipe with
radius r
0
(Figure 3.3). To describe such a ﬂow let us use the average velocity
u(r) representing the true timeaveraged velocity of ﬂuid particles passing
through a considered point. It is assumed that the resulting velocities are
parallel to the pipe axis and independent of the radial coordinate r, that is of
the distance between the considered point of the crosssection and the pipe
axis.
Similarly, average tangential stresses τ(r) between the ﬂuid layers moving
with average velocities u(r) are introduced (Figure 3.3). These stresses are also
called Reynolds stresses, named in honor of the famous English engineer O.
Reynolds who contributed greatly to the development of turbulence theory.
The average stress τ(r) represents the ratio of the friction force acting
between macrolayers of turbulent ﬂow separated by a surface element to
the area of this element. As in laminar ﬂow, turbulent tangential stresses
τ(r) are suggested to be proportional to the gradients du/ dr of average ﬂow
velocities
τ(r) = µ
t
·
du
dr
. (3.18)
However, different from laminar ﬂow, the factor µ
t
in this relation represents
not the intrinsic viscosity of the ﬂuid but the socalled turbulent dynamic
viscosity, dependent on the structure and mixing intensity of the ﬂuid layers
rather than on ﬂuid properties.
The turbulent dynamic viscosity µ
t
and the turbulent kinematic viscosity ν
t
=
µ
t
/ρ of the turbulent ﬂow is caused not by molecular friction of the separated
ﬂuid layers, as in laminar ﬂow, but by largescale ﬂuctuations and momentum
3.5 Turbulent Fluid Flow in a Circular Pipe 53
Figure 3.3 The calculation of ﬂuid turbulent ﬂow.
transport with eddies from one macrolayer to another. The transport of mo
mentum is perceived as a friction force acting between these layers. When the
layers are moving with equal velocities, that is du/ dr = 0, then the passing of
the momentumfromone layer to another is compensated by equal momentum
fromanother layer, therefore τ = 0. If one layer moves faster than another, that
is du/ dr > 0, thenthere appears betweenthe layers a frictionforce accelerating
the backward layer and retarding the leading one. This means that the leading
layer loses more momentum than it obtains from the backward layer. Thus the
intensity of momentum exchange between the layers depends on the regime
of turbulent ﬂow rather than on the viscosity of the ﬂuid characterized by the
factors µ or ν. Hence it follows that the turbulent viscosity µ
t
or ν
t
is not ﬂuid
constant as its molecular analog but depends on parameters of turbulent ﬂow.
Let the law of ﬂow have the following form
1
ρ
· τ = ν
t
·
du
dr
(3.19)
It makes sense to suggest that the turbulent viscosity ν
t
determined by the
structure of turbulent ﬂow be dependent on the parameters of this ﬂow. In
accordance with the brilliant idea of Karman it can be accepted that
ν
t
= f
_
ν,
¸
¸
¸
¸
du
dr
¸
¸
¸
¸
,
¸
¸
¸
¸
d
2
u
dr
2
¸
¸
¸
¸
_
(3.20)
that is, the turbulent viscosity ν
t
of the ﬂow depends on the molecular
viscosity ν of the ﬂuid itself and parameters absolute values of the ﬁrst u
and
second u
derivatives of average velocity with respect to the radial coordinate)
characterizing the ﬂow.
Since among arguments of the function f there are only two dimensional
independent variables ν and u
[ν] =
L
T
2
, [u
] =
1
L · T
the dimension of the third argument u
 could be expressed through the
dimensions of the two others as follows (Lurie and Podoba, 1984):
[u
] = [ν]
1/3
· [u
]
2/3
=
1
T
.
54 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Then owing to the theorem the number of arguments of the function f can
be reduced to one and the dependence (3.20) takes the following dimensional
form
ν
t
/ν =
˜
f
_
u

3
ν · u

2
_
⇒ν
t
= ν ·
˜
f
_
u
3
ν · u

2
_
. (3.21)
The formula (3.21) shows that the turbulent viscosity ν
t
is equal to the
molecular viscosity ν of the ﬂuid multiplied by the dimensionless factor
˜
f dependent only on one dimensionless parameter η = u

3
/(ν · u

2
).
Experiments testify, however, that in most of the pipe crosssection, that
is in the ﬂow core, except for the narrow wall layer the turbulent viscosity is
practically independent of the molecular viscosity of the ﬂuid. This is because
the turbulent viscosity is determined by exchange of momentumbetween ﬂuid
layers due to largescale eddies rather than by molecular friction. Therefore
Eq. (3.21) should have such a structure that the molecular viscosity would
have no impact over almost all the crosssection of the ﬂow. The last could be
achieved by assuming the function
˜
f to be linear, in other words, if
˜
f (η) = κ · η,
where κ is a constant called the Karman constant. With regard to the accepted
assumption Eq. (3.21) for turbulent viscosity can be rewritten as
ν
t
= ν ·
_
κ
2
u

3
ν · u

2
_
= κ
2
·
u

3
(u
)
2
.
The law of turbulent friction (3.19) is expressed by
1
ρ
· τ = κ
2
u

3
(u
)
2
· u
(3.22)
called the Karman formula (Loizyanskiy, 1987). The last equation represents
the basic equation of the phenomenological, that is resulting from abstract
reasoning though adequate for the phenomena under consideration, Karman
theory. The constant κ in Eq. (3.22) has been shown by a plethora of experi
ments to be approximately equal to 0.4 and is a universal constant of the model
in the sense that it is the same for all regimes of turbulent ﬂow of ﬂuids in
pipes. The relation (3.22) differs signiﬁcantly from the analogous relation (2.2)
for laminar ﬂows.
Let us determine now the distribution u(r) of average velocities of ﬂuid
turbulent ﬂow in a circular pipe. First note that tangential stresses τ(r) as in
laminar ﬂow are linearly distributed over the radius
τ(r) = −
1
2
·
p
L
· r
since the balance of forces acting on an arbitrarily separated ﬂuid cylinder (see
Figures 3.2 and 3.3) is independent of the ﬂow regime (laminar or turbulent).
Introduce into the consideration the tangential stress τ
w
caused by the friction
3.5 Turbulent Fluid Flow in a Circular Pipe 55
at the pipe walls
τ
w
= τ(r
0
) = −
1
2
·
p
L
· r
0
.
Then τ(r) could be expressed through τ
w
τ(r) = τ
w
·
r
r
0
. (3.23)
Substitution instead of τ(r) from Eq. (3.22) leads to the differential equation
κ
2
u

3
u
u
2
=
1
ρ
τ
w
·
r
r
0
(3.24)
for the velocity distribution u(r) over the radius.
The quantity (τ
w
/ρ)
1/2
has the dimensions of velocity. It is commonly called
the dynamic velocity and speciﬁed by u
∗
. In fact it has the sense of a friction
stress at the pipe wall since ρu
2
∗
= τ
w
. In addition the dynamic velocity is
related to the above introduced factor λ of hydraulic resistance
u
2
∗
=
τ
w

ρ
=
1
2
C
f
· v
2
=
λ
8
· v
2
⇒
u
∗
v
=
_
λ
8
where v is the ﬂuid velocity averaged over the crosssection. Since the factor
λ is small (λ ≈ 0.01–0.03), the dynamic velocity u
∗
∼
= 0.05 · v, from which
follows that the dynamic velocity is 20–25 times smaller than the average ﬂow
velocity.
In terms of dynamic viscosity the basic equation (3.24) attains a more
compact form
κ
2
u

3
u
u
2
= −u
2
∗
·
r
r
0
, (τ
w
< 0). (3.25)
Solve this equation for a ﬂow in a circular pipe (Figure 3.3) u
< 0 and u
< 0,
thus
κ
2
u
3
u
u
2
= −κ
2
u
4
u
2
= −u
2
∗
·
r
r
0
.
From here it ensues that
−
u
u
2
=
κ
u
2
•
·
_
r
0
r
or
d
dr
_
1
u
_
=
κ
u
2
∗
·
_
r
0
r
.
Repeated integration gives
u(r) =
u
∗
κ
·
__
r
r
0
+C
1
+C
2
· ln
¸
¸
¸
¸
_
r
r
0
−C
2
¸
¸
¸
¸
_
(3.26)
where C
1
and C
2
are constants of integration.
56 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
To determine C
1
and C
2
one needs two boundary conditions at the internal
pipe wall (Lurie and Podoba, 1984). The ﬁrst condition is evident. It is the
sticking condition in accordance with which the velocity u
w
of ﬂuid particles
should vanish at the pipe walls
u
w
= u(r
0
) = 0. (3.27)
Substituting r = r
0
in Eq. (3.26) and taking u = 0 we obtain
0 =
u
∗
κ
· [1 +C
1
+C
2
· ln1 −C
2
].
Subtracting further the resulting equality termbyterm from Eq. (3.26) we get
u(r) =
u
∗
κ
·
__
r
r
0
−1 +C
2
· ln
¸
¸
¸
¸
√
r/r
0
−C
2
1 −C
2
¸
¸
¸
¸
_
. (3.28)
The second condition is more complicated and was not met in the model
of laminar ﬂow. The problem is that the differential equation (3.25) for the
turbulent model has higher order thanthe analogous differential equation (3.2)
for turbulent ﬂow. Therefore to solve it one needs an additional boundary
condition reﬂecting the interaction of turbulent ﬂow with the pipe walls. Since
this condition models the ﬂuid ﬂow in a narrow wall layer it has to connect
parameters u
w
and u
w
of the turbulent ﬂow at the pipe wall with the molecular
viscosity ν, whose inﬂuence is strong in this layer, and the smoothness of the
internal surface of the pipe characterized by its absolute roughness . In other
words the missing boundary condition should be expressed by the relation
G(u
w
, u
w
, ν, ) = 0 (3.29)
in which the subscript ‘‘w’’ indicates that the corresponding derivatives are
calculated at points on the internal pipe surface. The dimension analysis as
applied to Eq. (3.29) allows us to rewrite it in dimensionless form
˜
G
_
ν · u
2
w
u
w

3
,
· u
w

u
w

_
= 0.
Having resolved the last expression with respect to the ﬁrst argument, we get
ν · u
2
w
u
w

3
= ˜ g
_
· u
w

u
w

_
.
For small values of the roughness the righthand part of this equation could
be expanded into a Taylor series leaving in it only the ﬁrst two terms
ν · u
2
w
u
w

3
= g
0
+g
1
·
· u
w

u
w

(3.30)
3.5 Turbulent Fluid Flow in a Circular Pipe 57
where g
0
and g
1
are dimensionless constants. If we now take into account that
at r = r
0
in accordance with Eq. (3.25) the relation
κ
2
·
u
4
w
u
2
w
= u
2
∗
takes place, meaning that, at the pipe walls, u
w
 = κ · u
2
w
/u
∗
, the condi
tion (3.30) simpliﬁes to
ν
u
w

3
·
κ
2
· u
4
w
u
2
w
= g
0
+g
1
·
u
w

·
κ · u
2
w
u
∗
and takes the ﬁnal form (u
w
< 0):
ν · u
w
= −
k · u
2
∗
1 +a · (u
∗
· /ν)
(3.31)
where k = g
1
/κ
2
and a = −g
2
/κ are dimensionless constants. These constants
are universal like the Karmanconstant κ, that is, they do not depend onconcrete
ﬂow and are given once and for all. In other words they are phenomenological
constants of the model. A large body of calculation results for different types
of ﬂows correlated to each other gives k = 28 and a = 0.31 (Lurie and Podoba,
1984). Hence, the theory of ﬂuid turbulent ﬂow in a circular pipe is based on
three phenomenological constants κ
∼
= 0.4; k
∼
= 28 and a
∼
= 0.31.
In the particular case of an ideal smooth internal surface of the pipe ( ≈ 0),
the boundary condition (3.31) reduces to
νu
w
= −k · u
2
∗
. (3.32)
Thus the boundary condition (3.31) permits us to obtain the second integration
constant C
2
in expression (3.28) for the velocity u(r). As a result we have
u
=
u
∗
κ
·
_
1
2
√
r · r
0
+
C
2
√
r/r
0
−C
2
·
1
2
√
r · r
0
_
,
u
w
= u
(r
0
) =
u
∗
2r
0
· κ
·
1
1 −C
2
.
Insertion of the calculated derivative into condition (3.31) yields the equation
−
ν · u
∗
2r
0
· κ
·
1
1 −C
2
=
k · u
2
∗
1 +a · u
∗
/ν
,
which gives
C
2
= 1 +
1 +a · u
∗
· /ν
2kκ · r
0
u
∗
/ν
58 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
or, in dimensionless form
C
2
= 1 +
1 +a · ε · Re · u
∗
/v
kκ · Re · u
∗
/v
(3.33)
where ε = /d is the relative roughness; Re = vd/ν the Reynolds number; v
the mean ﬂow velocity; d = 2r
0
the pipeline diameter. At the same time note
that for turbulent ﬂows with Re > 10
4
, ε < 10
−3
, u
∗
/v ≈ 10
−2
the constant C
2
is close to one,
C
2
≈ 1 +
1 +0.31 · 10
−3
· 10
4
· 10
−2
28 · 0.4 · 10
4
· 10
−2
∼
= 1.001. (3.34)
Hence, Eq. (3.28) with constant (3.33) gives the distribution of the average
velocity of turbulent ﬂow in a circular pipe under the condition that the
dynamic velocity u
∗
is known. The latter in its turn can be expressed through
the average ﬂow velocity v and the hydraulic resistance factor λ with Eq. (3.24)
as u
∗
= v ·
√
λ/8.
Figure 3.4 shows dimensionless turbulent velocity proﬁles u(r)/u
max
related
to the maximal value of the ﬂuid velocity at the pipe axis. The lower curve
corresponds to Re = 23 000, the top curve to Re = 3 200 000, the middle one to
intermediate values of the Reynolds number (Loitzyanskiy, 1987). The dotted
curve depicts the parabola (3.3) giving the velocity distribution in laminar
ﬂow regime. The comparison of laminar and turbulent velocity distributions
shows that the turbulent velocity proﬁle has more plane form and the greater the
Reynolds number the more the plane become curves. For the laminar ﬂow
regime in accordance with Eq. (3.5) u
max
/v = 2. For the turbulent regime this
ratio is far less. In general it depends on the numbers Re, ε and the average
ratio is equal to u
max
/v = 1.15 −1.25.
The integration constant C
2
in Eq. (3.33) can be represented as
C
2
= 1 +
1 +a · ε · Re ·
√
λ/8
kκ · Re ·
√
λ/8
. (3.35)
Figure 3.4 Dimensionless average
velocity proﬁles in turbulent
ﬂows (Loitzyanskiy, 1987).
3.5 Turbulent Fluid Flow in a Circular Pipe 59
On the other hand the average velocity is by deﬁnition equal to
v =
Q
π · r
2
0
=
1
π · r
2
0
·
_
r
0
0
2πr · u(r) dr =
2
r
2
0
·
_
r
0
0
r · u(r) dr.
Substituting here the distribution (3.28) and taking into account the
formula (3.35) for C
2
we get the dependence of the hydraulic resistance
factor λ on the Reynolds number, that is in fact on the average velocity v, and
on the relative roughness ε of the internal surface of the pipe
v =
2u
∗
κr
2
0
_
r
0
0
r
__
r
r
0
−1 +C
2
· ln
C
2
−
√
r/r
0
C
2
−1
_
dr
=
2u
∗
κ
·
_
1
0
η
_
√
η −1 +C
2
· ln
C
2
−
√
η
C
2
−1
_
dη (3.36)
where η = r/r
0
. Calculating the integral (3.36) with regard to remark (3.34),
we obtain
_
8
λ
=
1
κ
·
_
ln
kκ · Re ·
√
λ/8
1 +a · ε · Re ·
√
λ/8
−
137
60
_
. (3.37)
Inserting in Eq. (3.37) the numerical values of the constants k, κ and a we get
the equation for the dependence of λ on the numbers Re and ε
1
√
λ
= 0.88 · ln
Re ·
√
λ
1 +0.11 · ε · Re ·
√
λ
−0.8 (3.38)
called the universal resistance law.
1. If we ignore the effect of roughness, that is we assume
0.11 · ε · Re ·
√
λ 1, e.g. 0.11 · ε · Re ·
√
λ < 0.11 or Re ·
√
λ < ε
−1
,
then the equation for λ becomes
1
√
λ
= 0.88 · ln(Re ·
√
λ) −0.8. (3.39)
Its approximate solution takes the form
λ =
0.3164
4
√
Re
(3.40)
mentioned above and called the Blasius formula. The estimation of
allowable Reynolds numbers can be conducted as follows
Re ·
√
λ
∼
= 0.56 · Re
7/8
<
1
ε
⇒Re < 1.93 · ε
−8/7
≈ 2 · ε
−1.143
,
therefore Eq> (3.39) is true when Re < 2 · ε
−1.143
.
2. At very large Reynolds numbers (Re > 10
5
) the equation for λ acquires
the form
1
√
λ
= 0.88 · ln
1
0.11 · ε
−0.8
60 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
where λ is independent of Re. We have
1
√
λ
=
_
0.88 · ln
1
0.11
+0.88 · ln
1
ε
−0.8
_
−2
⇒
λ
∼
=
_
0.88 · ln
1
ε
+1.14
_
−2
. (3.41)
Exercise 1. It is required to calculate the hydraulic resistance factor λ in the
ﬂowof diesel fuel withν = 4 cSt ina pipeline withd = 500 mm, = 0.25 mm.
The ﬂow rate of the ﬂuid is Q = 1000 m
3
h
−1
.
Solution. Determine the average velocity v of the ﬂow
v =
4Q
πd
2
=
4 · 1000/3600
3.14 · 0.5
2
= 1.415 m s
−1
the Reynolds number
Re =
vd
ν
=
1.415 · 0.5
4 · 10
−6
= 176 875
and the relative roughness
ε =
d
=
0.25
500
= 0.0005
As a result we get the transcendental equation for λ
1
√
λ
= 0.88 · ln
176 875 ·
√
λ
1 +0.11 · 0.0005 · 176 875 ·
√
λ
−0.8
We look for the solution of this equation by the method of successive
approximations. Consider the function
(λ) =
1
√
λ
−0.88 · ln
176 878
√
λ
1 +9.728 ·
√
λ
+0.8
representing the difference in both parts of the resulting equation. We have
λ = 0.02, (0.02) = −0.279;
λ = 0.019, (0.019) = −0.086;
λ = 0.018, (0.018) = 0.123;
λ = 0.0185, (0.0185) = 0.016;
λ = 0.0186, (0.0186) = −0.004.
It is seen that λ
∼
= 0.0186.
Exercise 2. It is required to calculate the hydraulic resistance factor λ in the
ﬂow of benzene with ν = 0.6 cSt in a pipeline with d = 361 mm, = 0.2 mm.
The ﬂow rate of the ﬂuid is Q = 500 m
3
h
−1
.
3.5 Turbulent Fluid Flow in a Circular Pipe 61
Solution. Determine the average velocity of the ﬂow
v =
4Q
πd
2
=
4 · 500/3600
3.14 · 0.361
2
= 1.358 m s
−1
,
the Reynolds number
Re =
vd
ν
=
1.358 · 0.361
0.6 · 10
−6
= 817 063.
and the relative roughness
ε =
d
=
0.2
361
= 0.00055.
As a result we get the transcendental equation for λ
1
√
λ
= 0.88 · ln
817 063 ·
√
λ
1 +49.4 ·
√
λ
−0.8.
We look for the solution of this equation by the method of successive
approximations. Consider the function
(λ) =
1
√
λ
−0.88 · ln
817 063
√
λ
1 +49.4 ·
√
λ
+0.8,
representing the difference between both parts of the resulting equation. We
have
λ = 0.02, (0.02) = −1.36;
λ = 0.018, (0.018) = −0.17;
λ = 0.016, (0.016) = 0.29;
λ = 0.017, (0.017) = 0.049;
λ = 0.0172, (0.0172) = 0.003.
It is seen that λ
∼
= 0.0172.
Turbulent Flows of NonNewtonian Fluids
For the turbulent ﬂowof a power ﬂuid(see Section 2.6) the following expression
for the universal resistance lawhas been obtained by Dodge and Metzner, 1959
1
√
λ
=
0.88
n
3/4
ln
_
Re ·
_
λ
4
_
1−n/2
_
−
0.4
n
1.2
(3.42)
where n is the exponent in the Ostwald rheological law. This equation may
be obtained on the basis of the Karman model (3.25), if in the boundary
condition (3.32) we take the factor k to be dependent on the exponent n, that is
k = k(n). As is known k is equal to 28 for a Newtonian viscous ﬂuid.
62 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Romanova (1989) gave the resistance law for a power ﬂuid in the form
1
√
λ
=
0.88
n
ln
_
k(n)Re ·
_
λ
8
_
1−n/2
_
−2.83,
where k(n) = 2
1−n/2
· exp
_
n · (2.83 −0.2/n
1.2
)/0.88
_
. In the same publication
the following formulas were suggested as approximate solutions of the above
equation
λRe
1/3
= 0.698 · n −1.94 · 10
−2
at 0.2 ≤ n ≤ 0.5,
λRe
1/4
= 0.353 · n −3.80 · 10
−2
at 0.5 < n ≤ 1.25,
λRe
1/5
= 0.234 · n −5.13 · 10
−2
at 1.25 < n ≤ 2.0.
In all these formulas the generalized Reynolds number is deﬁned as
Re = v
2−n
· d
n
/k, where k is the kinematic consistency.
The universal resistance law for a viscousplastic ﬂuid was suggested
by Potapov (1975)
1
√
λ
=
_
1 −
8He
λ Re
2
_
·
_
0.88ln(Re
√
λ) −0.8
_
+2.76 ·
8He
λ Re
2
(3.43)
where Re = vd/ν is the Reynolds number and He = τ
0
d
2
/ρν
2
is the Hedstroem
number.
3.6
A Method to Control Hydraulic Resistance by Injection of AntiTurbulent Additive
into the Flow
Friction losses are the key reason for electric energy expenditure on ﬂuid and
gas pumping along pipelines. They are caused by the forces of internal friction
between the layers of the moving ﬂuid. In laminar and turbulent ﬂows there
occurs the socalled dissipation of mechanical energy of ordered motion and
its transition into the energy of chaotic motion of the ﬂuid particles, in other
words into heat. For turbulent ﬂows this transition has a multistage character.
The mechanical energy of average motion is transformed ﬁrst into the energy
of largescale eddies of the turbulent medium, then into the energy of the
ﬂuctuation motion of smallscale eddies and ﬁnally, due to viscosity forces,
into the heat energy of the ﬂuid. Therefore engineers and scientists involved
in the problem of pipelines have for a long time been interested in methods
of governing the turbulent ﬂow structure with the aim of reducing energy
losses.
One such method discovered by the English scientist Toms in the late 1940s
consists in injection into the turbulent ﬂow of special highmolecular weight
additives to lower the hydraulic resistance. This effect is called by the name of
the discoverer, the Toms effect.
3.6 Controlling Hydraulic Resistance by Injection of AntiTurbulent Additive into the Flow 63
The mechanism of operation of all varieties of antiturbulent additives is
based on damping turbulent ﬂuctuations near the internal surface of the
pipeline by interaction of the longlength molecules of the additive with
turbulent eddies generated near the pipeline wall. As a rule this effect is
achieved with very small concentrations of additives, measured commonly
in ppm (parts per million of the ﬂuid volume to which the additive is
added).
Owing to the damping of the nearwall turbulence there is a reduction in
the ﬂow hydraulic resistance caused by the pipeline wall. Hence, increased
pumping efﬁciency is achieved with conservation of the pressure drop or
pressure lowering at pumping stations. The decrease in hydraulic resistance
reduces expenditure on electric power by 20–60%.
The best known antiturbulent additives for oil products are CDR produced
by the American company DuponConoco and NECCAD547 produced by the
Finnish company Neste. These products are based on hydrocarbons. The ﬁrst
is equally suitable for benzene and diesel fuel pumping while the second is
recommended primarily for diesel fuels. Both antiturbulent additives were
put through production tests on pipelines in Russia.
The use of antiturbulent additives has some speciﬁc restriction: during
prolonged action of the additives in turbulent ﬂow they become degraded;
their degradation is especially great when passing through pumps. Therefore,
when using additives it is necessary to inject a portion of fresh additives into
the ﬂowafter each pumping station. It is rational to use antiturbulent additives
to build up the carrying capacity of certain pipeline sections, ﬁrst the limiting
ones.
All antiturbulent additives reduce the hydraulic resistance factor λ. To
calculate this factor the universal resistance law (3.37) containing the constant
k related to the turbulent ﬂow interaction with pipeline walls was used (see
boundary conditions (3.31) and (3.32)). It was found that k
∼
= 28.
The effect of the antiturbulent additive is that it changes the intensity of the wall
turbulence, that is the additive acts on the magnitude of the constant k. Therefore it
is reasonable to accept as a model of turbulent ﬂowwith antiturbulent additive
a model with variable constant k (Ishmuchamedov et al., 1999) dependent on
the concentration of antiturbulent additive θ. Thus the universal quantity k
which was taken earlier as constant would be, in the presence of antiturbulent
additive, a function of θ, that is k = k(θ). In the absence of antiturbulent
additive (θ = 0) then k(0) = 28.
The results of tests of antiturbulent additive CDRhave giventhe dependence
k(θ) shown in Table 3.1.
Table 3.1 Dependence k(θ) for CDR.
θ, ppm 20 30 40 50 60 70 80 90
k(θ) 61.4 95.1 143 187 249 276 340 380
64 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Table 3.2 Dependence k(θ) for NECCAD547.
θ, ppm 40 60 100 180
k(θ) 50 75 150 340
The dependence k(θ) for antiturbulent additive NECCAD547 is shown in
Table 3.2.
Remark. The data cited in Tables 3.1 and 3.2 could be improved by changing
the antiturbulent additive composition.
Exercise 1. The pumping of diesel fuel with antiturbulent additive CDR with
θ = 40 ppmis conducted at Reynolds number 40 000. It is required to calculate
the hydraulic resistance factor λ.
Solution. The formula (3.39) at Re = 40 000 and k(40) = 143 taken from
Table 3.1 gives for λ the following transcendental equation
1
√
λ
= 0.88 · ln(143 · 40 000 ·
√
λ) −3.745.
Its solution, found by the method of successive approximations, yields
λ = 0.0153. This value is signiﬁcantly less than 0.0224 corresponding to
the value of λ in the case of antiturbulent additive absence in the ﬂow of oil
product at the same Reynolds number. The effect is about 31.7%.
Exercise 2. The pumping of diesel fuel with antiturbulent additive NECCAD
547 with θ = 180 ppm is conducted at Reynolds number 40 000. It is required
to calculate the hydraulic resistance factor λ.
Solution. The formula (3.39) at Re = 40 000 and k(180) = 340 taken from
Table 3.2 gives for λ the following transcendental equation
1
√
λ
= 0.88 · ln(340 · 40 000 ·
√
λ) −3.745.
Its solution, found by method of successive approximations, yields λ = 0.0129.
This value is signiﬁcantly less than 0.0224 corresponding to the value of λ in
the case of antiturbulent additive absence in the ﬂow of oil product at the
same Reynolds number. The effect is about 42.4%.
In order to select the necessary concentration θ of antiturbulent additive
one can proceed as follows (Ishmuchamedov et al., 1999). From Eq. (3.39) for
k(θ) follows the relation
k(θ) =
1
Re ·
√
λ
· e
1+3.745·
√
λ
0.88·
√
λ
(3.44)
which permits the determination of k(θ) for given λ. Then with the help of
Tables 3.1 and 3.2 we obtain the concentration θ of the antiturbulent additive
3.7 Gravity Fluid Flow in a Pipe 65
in the oil product. Multiplying the latter by the total volume of the ﬂuid to be
pumped we get the required amount of the antiturbulent additive.
Exercise 3. It is required at a given pressure resource to increase by 30% the
carrying capacity of an oilpipeline with D = 377 mm, δ = 8 mm pumping
diesel fuel with ν
d
= 9 cSt and ﬂow rate 450 m
3
h
−1
. Determine the amount
of antiturbulent additive CDR needed to do this.
Solution. Let us calculate the initial values of the pumping rate v
0
, Reynolds
number Re and the factor of hydraulic resistance λ
0
:
v
0
= 4Q/S = 4 · 450/(3600 · 3.14 · 0.361
2
) = 1.221 m s
−1
,
Re
0
= v
0
d/ν
d
= 1.221 · 0.361/(9 · 10
−6
) = 48 976,
λ
0
= 0.0213.
Since the carrying capacity must be increased by 30%, the new values of the
pumping rate v and the Reynolds number Re will be
v = 1.3 · v
0
∼
= 1.587 m s
−1
, ⇒Re = 1.3 · Re
0
= 63 669.
Due to the invariability of the pressure resource it should be
λ
0
(Re
0
, 0) · v
2
0
= λ(Re, θ) · v
2
.
This relation gives a new value of λ
λ = λ
0
· (v
0
/U)
2
= 0.0213 · (1/1.3)
2
= 0.0126.
The constant k(θ) is determined by Eq. (3.44)
k(θ) =
1
63 669 ·
√
0.0126
· e
1+3.745
√
0.0126
0.88·
√
0.0126
≈ 246.
From Table 3.1 we ﬁnd that this value of k corresponds to θ = 60 ppm of the
antiturbulent additive.
Exercise 4. It is required to increase by 25% at a given pressure resource
the carrying capacity of an oilpipeline section with D = 530 mm, δ = 8 mm
pumping diesel fuel with ν
d
= 9 cSt and ﬂow rate 950 m
3
h
−1
. Determine the
amount of antiturbulent additive NECCAD547 needed to do this.
Answer. About 340 ppm.
3.7
Gravity Fluid Flow in a Pipe
The above considered ﬂows of ﬂuid are pertinent to the class of the socalled
enforced (pumped) ﬂows, since the motion of the ﬂow was forced, that is to
66 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
overcome the friction force a pressure gradient was needed. However, there
are ﬂows in which the primary driving force could be a component of the
gravity force. Such ﬂows are called gravity ﬂows. A variety of gravity ﬂows
are gravity stratiﬁed (divided into layers) ﬂows. In these ﬂows the ﬂuid moves
without completely ﬁlling the crosssection, the ﬂuid ﬂows over the lower part
of the pipe whereas the upper part of the pipe is ﬁlled with vapor and gases
evolved from this ﬂuid.
Considering the stratiﬁed gravity ﬂow of a viscous incompressible ﬂuid
along a section of the pipeline with diameter d and roughness of the
internal wall surface inclined at an angle α to the horizontal (Figure 3.5).
Our interest is the dependence of the ﬂow rate Q = v · S on the governing
parameters, that is the form of the function Q = f (S, ρ, µ, g, sinα, d, ).
We can write the sought dependence in dimensionless form guided by
dimensional theory. Among six arguments there are three dimensionally
independent ones, for example d, ρ, µ. Hence the number of independent
arguments in dimensional V would be reduced from six to three (Lurie, 2001)
and the function Q = f (S, ρ, µ, g, sinα, d, ) takes the form
Q/S
_
g sinα · d
=
˜
f
_
S
d
2
,
g · sinα
ν
2
/d
3
, ε
_
(3.45)
where ν = µ/ρ is the kinematic viscosity and ε = /d is the relative roughness
of the internal surface.
In the theory of gravity ﬂuid ﬂows a parameter R
h
, called the hydraulic
radius (Leibensone et al., 1934), is often introduced. The hydraulic radius is
deﬁned as the ratio between the area S of a part of pipe crosssection ﬁlled
with ﬂuid and the wetted perimeter P
s
(Figure 3.6).
Figure 3.5 A scheme of gravity ﬂuid ﬂow in a pipe.
Figure 3.6 Deﬁnition of the hydraulic radius.
3.7 Gravity Fluid Flow in a Pipe 67
S =
1
2
· r
0
2
φ −
1
2
· r
0
2
sinφ =
1
2
· r
0
2
(φ −sinφ);
S
r
0
2
=
1
2
(φ −sinφ).
Since
˘
AB = r
0
· φ and P
s
= r
0
· φ
then
R
h
=
S
P
S
=
1/2 · r
0
2
(φ −sinφ)
r
0
· φ
=
r
0
2
·
_
1 −
sinφ
φ
_
(3.46)
where ϕ is the central angle where is seen the part of crosssection ﬁlled with
ﬂuid and r
0
= d/2 the pipe radius. If the pipe is completely ﬁlled with ﬂuid
(ϕ = 2π), then R
h
= r
0
/2 = d/4.
Since S/d
2
and R
h
/d depend only on the degree of the section ﬁlling
with ﬂuid, that is on the angle ϕ, the dependence (3.45) without disturbing
generality could be written in equivalent form
Q/S
_
g sinα · R
h
=
˜
f
1
_
S
d
2
,
g · sinα
ν
2
/d
3
, ε
_
or
Q = S · C
Sh
·
_
R
h
· sinα (3.47)
where C
Sh
=
√
g ·
˜
f
1
(S/d
2
,
_
gd sinα · d/ν, ε) is the socalled dimensionless
Chezy factor;
˜
f
1
is a dimensionless factor dependent on the parameters of the
ﬂow regime
_
gd sinα · d/ν, degree of ﬂuid ﬁlling S/d
2
and the smoothness
parameter of the internal surface of the pipe ε (Archangelskiy, 1947 and
Christianovitch, 1938).
The formula (3.47) could be represented in a form analogous to the
Darcy–Weisbach relation for enforced ﬂow
ρg sinα =
2 · (d/R
h
)
˜
f
2
1
·
1
d
·
ρv
2
2
(3.48)
when in the last expression we replace the pressure gradient p/L
responsible for enforced ﬂow with the rolling down component of the
gravity force ρg sinα causing motion in the case of gravity ﬂow. Comparison
of Eq. (3.48) with the Darcy–Weisbach relation shows that the hydraulic
resistance factor λ in stratiﬁed gravity ﬂow is related to the factor
˜
f
1
by the
equality
λ =
2 · (d/R
h
)
˜
f
2
1
68 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
from which follows the expression for the Chezy factor
C
Sh
=
_
2g · (d/R
h
)
λ
. (3.49)
There are many empirical formulas for the Chezy factor for pipes with circular
as well as noncircular crosssections (Leibensonet al., 1934). Avoiding detailed
treatment of these formulas let us only say that in a ﬁrst approximation it
is possible to use Eq. (3.49) replacing d by 4R
h
, to give C
Sh
=
_
8g/λ, where
λ = λ(Re, ε); Re = 4vR
h
/ν and R
h
the hydraulic radius related to the degree of
ﬁlling of the pipe crosssection by formulas (3.46).
Pipeline Sections of Gravity Flow
When the pressure in the pipeline section is equal to the saturated vapor
tension of the transported ﬂuid, then inside the section continuously appear
cavities ﬁlled with ﬂuid vapor. In this case the ﬂow could be stratiﬁed or could
have a more complicated structure in which portions of ﬂuid alternate with
vaporgas cavities (bubbles). The latter ﬂow regime is called slug ﬂow.
The ﬂowina section[x
1
, x
2
] of the pipeline inwhichit moves under the action
of the gravity force partially (incompletely) ﬁlling the pipeline crosssection
while the remainder is ﬁlled with vapor of this ﬂuid, is called gravity ﬂow. The
pressure inside the vaporgas cavity remains practically invariable and equal
to the saturated vapor tension (pressure) p
v
. In spite of this the difference in
pressures between the sections x
1
(the beginning of the gravity ﬂow section)
and x
2
(the end of the gravity ﬂowsection) nevertheless exists, it is merely equal
to the difference in geometrical heights (z
1
−z
2
) of these sections (Figure 3.7).
Stationary gravity ﬂow can exist only on descending sections of the pipeline.
The beginning of the gravity ﬂow section x
1
is called the transfer section. The
transfer section always coincides with the region of the pipeline proﬁle peak.
Figure 3.7 A scheme of pipeline gravity ﬂow.
3.7 Gravity Fluid Flow in a Pipe 69
The line of hydraulic gradient of the gravity ﬂow section passes parallel to
the pipeline proﬁle at the distance p
v
/ρg over it. Consequently the hydraulic
gradient i of the gravity ﬂow section is equal to the slope of the pipeline proﬁle
to the horizontal i = tanα
p
.
The ﬂuid ﬂow rate in the gravity ﬂow section in the stationary ﬂow regime
is equal to the ﬂow rate Q of ﬂuid in the ﬁlled sections of the pipeline
Q = v
0
S
0
= v · S (3.50)
from which may be concluded that the velocity of ﬂuid ﬂow v in the gravity
ﬂow section exceeds the velocity v
0
of the ﬂuid in the pipeline sections ﬁlled by
the ﬂuid, since the area S of the crosssection part of each gravity ﬂow section
ﬁlled with the ﬂuid is less than the area S
0
of the complete crosssection of the
pipeline, that is v = v
0
· S
0
/S > v
0
.
If the ﬂow of ﬂuid in the gravity ﬂow section is stratiﬁed, the degree σ = S/S
0
of pipeline ﬁlling with the ﬂuid depends on the ratio γ = i
0
/ tanα
p
 between
the hydraulic gradient i
0
= λ
0
· 1/d · v
2
0
/ρg of the pipeline sections completely
ﬁlled with ﬂuid and the absolute value of the gravity ﬂow section slope
α
p
to the horizontal. This dependence can be obtained from Eqs. (3.47) or
(3.48) solving them with respect to S. To calculate the degree σ of pipeline
section ﬁlling with ﬂuid it the following approximation formulas have been
suggested (Ishmuchamedov et al., 1999):
1. σ = 1 at γ = i
0
/ tanα
p
 ≥ 1. In this case the pipeline crosssection is
completely ﬁlled with ﬂuid;
2. σ = 1 −2.98 · 10
−2
·
_
2
λ
0
· (1 −
√
γ); at 32.32 · λ
0
≤ γ < 1;
3. σ = 9.39 · 10
−2
·
_
2γ
λ
0
+0.113; at 4.87 · λ
0
≤ γ < 32.32 · λ
0
; (3.51)
4. σ = 0.1825 ·
_
2γ
λ
0
_
0.356
; at γ < 4.87 · λ
0
.
Exercise. The oil ﬂow rate (ν = 8.6 cSt) in the gravity ﬂow section of an oil
pipeline (D = 720 mm, δ = 10 mm, = 0.2 mm) is 900 m
3
h
−1
. The proﬁle
of the section is inclined to the horizontal at an angle α
p
= −1
◦
. It is required
to ﬁnd the degree of pipeline crosssection ﬁlling with oil in this section.
Solution. Calculate the pumping rate v
0
, the Reynolds number Re, the factor
of hydraulic resistance λ
0
and the hydraulic gradient i
0
inthe pumping sections
of the pipeline:
v
0
= 4 · 900/(3600 · 3.14 · 0.700
2
) = 0.650 m s
−1
; tan1
◦
= 0.0175;
Re = 0.65 · 0.7/(8.6 · 10
−6
)
∼
= 52907; λ
0
= 0.0219;
i
0
= λ
0
· 1/d · v
2
0
/(2 · g) = 0.0219 · 1/0.7 · 0.65
2
/(2 · 9.81)
∼
= 0.0007.
Determine the parameter γ:
γ = i
0
/ tanα
p
 = 0.0007/0.0175 = 0.04.
70 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
Since γ = 0.04 < 4.87 · λ
0
= 0.1067, in formulas (3.51) using the fourth case:
σ = 0.1825 · (2γ/λ
0
)
0.356
= 0.1825 · (2 · 0.04/0.0219)
0.356
∼
= 0.29,
that is, the considered section of the pipeline is approximately 29% ﬁlled.
How to determine whether or not there are gravity ﬂow sections in pipeline
sections under consideration. To answer this question we need to build a
combined picture of the pipeline proﬁle and the hydraulic gradient line. If the
line of the hydraulic gradient is everywhere over the pipeline proﬁle and the
amount by which it exceeds it is the quantity p
v
/ρg, where p
v
is the saturated
vapor pressure of the ﬂuid, then gravity ﬂowsections in the pipeline are absent.
If the line of hydraulic gradient at any point approaches the pipeline proﬁle
closer than p
v
/ρg or even intersects it, then there exists one or several gravity
ﬂow sections in the pipeline (Ishmuchamedov et al., 1999).
Now let us consider Figure 3.8 where a section of pipeline OO
1
is depicted.
Let us begin to build the line BK
2
2
K
1
1
Aof the pipeline hydraulic gradient
from the end O
1
of the pipeline section under consideration. To do this it is
sufﬁcient to know the pressure and the hydraulic gradient at the end of the
section. The line of the hydraulic gradient in the segment BK
2
lies signiﬁcantly
over the pipeline proﬁle, therefore crosssections of the pipeline section are
completely ﬁlled. However, the line of the hydraulic gradient at the point
K
2
approaches the pipeline proﬁle up to a distance p
v
/ρg, thus the point K
2
represents the end of the ﬁrst gravity ﬂow section. Hence, one of the gravity
ﬂow sections is found. The line of hydraulic gradient K
2
2
at this section is
parallel to the pipeline proﬁle.
Continue to build the line of the hydraulic gradient. It leaves the point
2
at
an angle whose tangent is equal to the hydraulic gradient, that is in fact parallel
to the segment BK
2
. It turns out that this line at the point K
1
approaches
the pipeline proﬁle at the distance p
v
/ρg for the second time. Consequently,
the pressure inside the pipeline again becomes equal to the pressure of the
saturated vapor and in the pipeline there should exist vaporgas cavities. The
point K
1
represents the end of the second gravity ﬂow section. Its beginning
at the point
1
is a transfer section. It is called a transfer section because it is
Figure 3.8 A scheme to determine the location of a pipeline gravity ﬂow section.
3.7 Gravity Fluid Flow in a Pipe 71
sufﬁcient to deliver the transported ﬂuid to the point
1
so that it reaches then
the end O
1
of the section by itself with the help of the gravity ﬂow. Hence,
the second gravity ﬂow section K
1
1
is found. The line K
1
1
of the hydraulic
gradient at this section passes parallel to the pipeline proﬁle at the distance
p
v
/ρg from it.
At the section
1
Athe line of the hydraulic gradient is parallel to its segments
BK
2
and
2
K
1
having been built for completely ﬁlled pipeline segments.
From Figure 3.8 it follows that the presence of gravity pipeline sections leads to
enhancement of the initial hydraulic head H
1
(and consequently the pressure p
1
)
at the station and therefore requires higher expenditures of energy for pumping
as compared with a pipeline in which such sections are absent. If the line of
hydraulic gradient beginning from the point K
2
were to be lengthened up to
the initial crosssection of the considered pipeline section, it would be possible
to determine the hydraulic head
˜
H
1
needed to pump ﬂuid with the same ﬂow
rate in a pipeline of the same length and with the same diameter but without
gravity ﬂow sections. It is evident that H
1
≥
˜
H
1
.
Exercise. Oil (ρ = 870 kg m
−3
, ν = 8.5 cSt, p
v
= 0.02 MPa) with ﬂow rate
400 m
3
h
−1
is pumped along an oilpipeline (L = 140 km; D = 530 mm,
δ = 8 mm, = 0.2 mm). The proﬁle of the section has the form represented
in Table 3.3. The pressure at the end of the section is 0.2 MPa. It is required
to determine the pressure at the beginning of the section.
Solution. Calculate ﬁrst the hydraulic gradient.
v
0
= 4 · 400/(3600 · 3.14 · 0.514
2
)
∼
= 0.536 m s
−1
;
Re = 0.536 · 0.514/(8.5 · 10
−6
) = 32 412;
λ
0
= 0.11 · (0.2/514 +68/32 412)
0.25
∼
= 0.025;
i
0
= 0.025 · 1/0.514 · 0.536
2
/(2 · 9.81)
∼
= 0.00071.
Then determine the head losses in the pipeline section between 120 and
140 km. They are h
120–140
= i
0
· 20 000 = 14.2 m. Therefore the head at
the end of the slope, that is at the crosssection x = 120 km is equal to
0.2 · 10
6
/870 · 9.81 +14.2
∼
= 14.43 m.
Since p
v
/ρ · g = 20 000/(870 · 9.81)
∼
= 2.34 m, the pipeline at the cross
section x = 120 km is still ﬁlled with oil. However, the difference in height
at the descending section is 100 m (see the proﬁle of the pipeline), thus it
is evident that at some crosssection the pressure of oil will be equal to the
saturated vapor pressure of oil p
v
, so that a part of the descending pipeline
Table 3.3
x, km 0 80 120 140
z, m 100 100 0 0
72 3 Structure of Laminar and Turbulent Flows in a Circular Pipe
section will inevitably become a gravity ﬂow section. It is evident that the
beginning of this section coincides with the beginning of the descent at
x = 80 km.
The hydraulic gradient at the plain (completely ﬁlled) pipeline segment
between the section beginning and the 80th kilometer is equal to the
hydraulic gradient at the completely ﬁlled pipeline segment between 120
and 140 km, that is 0.71 m km
−1
, so that the loss of hydraulic pressure is
h
0–80
∼
= 56.8 m. Therefore the pressure p
1
at the beginning of the section is
equal to 870 · 9.81 · 56.8
∼
= 484 771 Pa or ≈4.95 atm.
73
4
Modeling and Calculation of Stationary Operating Regimes
of Oil and Gas Pipelines
In this chapter we consider the calculation of stationary operating regimes of
pipelines for transportation of oil, oil products and gas. The equations obtained
in the ﬁrst chapter are used as a basis.
4.1
A System of Basic Equations for Stationary Flow of an Incompressible Fluid in a
Pipeline
In stationary ﬂow all parameters of the transported ﬂuid at each crosssection
of the pipeline remain constant, that is independent of time. Therefore partial
derivatives with respect to time ∂()/∂t in the equations of Section 1.8 should
be taken equal to zero. Consider successively the basic equations describing
these ﬂows:
1. Continuity equation (1.6) leads to the equation
d
dx
(ρvS) = 0
which means that the mass ﬂow rate
˙
M of the transported ﬂuid remains
constant
˙
M = ρvS = const.
If the ﬂuid is incompressible ( dρ/ dt = 0) and homogeneous (ρ = const.)
and the pipeline has invariable diameter (S = const.), the velocity of the
ﬂuid would be the same at each crosssection of the pipeline
v = const. (4.1)
2. The momentum equation (1.10) gives
ρv
dv
dx
= −
dp
dx
−
4
d
τ
w
−ρg · sinα(x).
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
74 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
With regard to the condition v = const. the momentum equation is
simpliﬁed and yields
dp
dx
= −
4
d
τ
w
−ρg · sinα(x).
If we take in the last equation
τ
w
=
λ(Re, ε)
4
·
ρvv
2
and sinα(x) =
dz
dx
,
the momentum equation is transformed into
d
dx
_
p
ρg
+z
_
= −λ(Re, ε) ·
1
d
·
vv
2g
. (4.2)
Note that the Bernoulli equation (1.19) leads also to Eq. (4.2) if we replace
in it the hydraulic gradient i
0
by 4τ
w
/ρgd = λ(Re, ε) · v
2
/2gd in
accordance with Eq. (1.23).
3. The equation of total energy balance (1.36) for stationary ﬂow has the
form
ρvS ·
d
dx
_
e
in
+
p
ρ
+gz
_
= πd · q
n
or
ρv
de
in
dx
=
4
d
· q
n
−ρv ·
d
dx
_
p
ρ
+gz
_
.
If in this equation we take e
in
= C
v
· T +const.,
d
dx
_
p
ρ
+gz
_
= −λ(Re, ε) ·
1
d
·
vv
2
and q
n
= −K · (T −T
ex
)
the equation of total energy balance takes the form
ρvC
v
dT
dx
= −
4K
d
(T −T
ex
) +λ(Re, ε)
1
d
·
ρv
3
2
. (4.3)
The system of equations (4.1)–(4.3) with the addition of relations for the
hydraulic resistance factor λ(Re, ε) and heat transfer factor K serve as the
basis for calculation of the stationary operating regimes of pipelines tran
sporting incompressible ﬂuids, to which class belong oil and oil products.
4.2 Boundary Conditions. Modeling of the Operation of Pumps and OilPumping Stations 75
4.2
Boundary Conditions. Modeling of the Operation of Pumps and OilPumping
Stations
The Bernoulli equation in algebraic form results from the continuity
equation (4.1) and the differential momentum equation (4.2)
_
p
ρg
+z
_
x=0
−
_
p
ρg
+z
_
x=L
= λ(Re, ε) ·
L
d
·
v
2
2g
(4.4)
In this relation x = 0 and x = L denote, respectively, the initial and terminal
crosssections of the pipeline section with length L. Thus we have one
algebraic equation relating three parameters of the ﬂow – the pressure p
0
at the beginning of the pipeline section, the pressure p
L
at the end of the
pipeline section and the velocity v of the ﬂuid ﬂow. To determine the velocity
v or, what is the same, the ﬂow rate of pumping, additional information
on the pressures at both ends of the pipeline is needed. This information
reﬂecting the interaction of the considered pipeline section with the rest of
the pipeline is introduced into the mathematical model through boundary
conditions.
In some cases the pressure p
L
at the end of the pipeline, that is at x = L
could be taken as given, determined e.g. by the conditions required for ﬂuid
pumped into reservoirs through a system of intrabase pipelines. Hence, one
of the boundary conditions can be a simple condition p(L) = p
L
.
Another boundary condition models the operation of the oil pumping
station (OPS) located at the beginning of the pipeline section, that is at
x = 0. In ﬂuid ﬂow in the pipeline the pressure gradually decreases because
mechanical energy is spent overcoming the force of viscous friction between
the ﬂuid layers and is then turned into heat. That is why in a pipeline special
equipment producing pressure is needed. In general such equipment is called
a pump.
4.2.1
Pumps
Pumps represent equipment for compulsory ﬂuid movement from a cross
section with lesser head (line of suction) to a crosssection with greater head
(line of discharge). Since the elevations of the pump entrance and exit are, as
a rule, identical, one can say that these pumps are equipment for forced ﬂuid
movement from a crosssection with lesser head (line of suction) to a crosssection
with greater head (line of discharge).
The simplest mathematical model of a pump can be represented as an
algebraic equation of the form
H =
p
ex
−p
in
ρg
= F(Q) (4.5)
76 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
characterizing the dependence of the differential head H produced by the
pump on the ﬂuid ﬂow rate Q. For every actual pump the differential head
H appears to be dependent on the ﬂuid ﬂow rate Q called in this case the
feed. The greater the head produced by a pump, the less, as a rule, is its feed.
The dependence H = F(Q) deﬁnes the socalled headdischarge (Q −H)
characteristic of the pump.
In order to understand the physical basis of this model, let us closely
consider the operating principle of one of the commonly encountered pumps,
namely the centrifugal pump. In centrifugal pumps, used for pumping oil and
oil products, the ﬂuid moves from the crosssection with lesser pressure to
that with greater pressure under the action of the centrifugal force produced
by the rotation of an impeller with proﬁle blades.
Figure 4.1 shows a scheme of a pump impeller with proﬁle blades.
Considering the frame of reference related to the rotating impeller, the
impeller is believed to be immovable whereas the centrifugal force of inertia
ρω
2
r, where ρ is the ﬂuid density, ω the angular velocity of the impeller
rotation and r the distance of the ﬂuid particle from the rotation axis, acts on
the ﬂuid ﬁlling the pump.
The centrifugal force causes the ﬂuid to move along the impeller blade from
its center to the periphery. This force is capable of overcoming the pressure
drop p = p
ex
−p
in
, equal to the pressure difference between the pumping
pressure p
ex
at the impeller periphery and the suction pressure p
in
at the
center of the impeller, that is to force the ﬂuid to move from the region of low
pressure to the region of high pressure. It is selfevident that to produce such
forced motion one needs to spend energy for impeller rotation.
For simplicity let us consider an impeller with radially located blades. The
balance equation of forces acting on the ﬂuid moving along the impeller radius
from its center to the periphery can be written as:
ρω
2
r −
dp
dr
= ρ · f
τ
(Q)
where dp/ dr is the radial gradient of pressure opposing the motion and f
τ
(Q)
the friction force depending on the discharge Q and increasing with Q.
Integration of the force balance equation over the radius from 0 to R
im
,
where R
im
is the radius of the impeller, gives
ρω
2
R
im
2
2
−p = R
im
ρ · f
τ
(Q) or p =
ρω
2
R
im
2
2
−R
im
ρ · f
τ
(Q).
Division of both sides of this equation by ρg yields
H =
ω
2
R
im
2
2g
−
R
im
g
· f
τ
(Q). (4.6)
Thus the rotation of the impeller with angular velocity ω can force the ﬂuid
to move against the pressure drop p between the periphery and the central
4.2 Boundary Conditions. Modeling of the Operation of Pumps and OilPumping Stations 77
Figure 4.1 Operating principle of a
centrifugal pump.
part of the impeller. The maximal value of the pressure drop which the
centrifugal force is capable of overcoming is equal to ρω
2
R
im
2
/2. This value of
p is achieved at Q = 0 in the absence of a friction force. At Q > 0 Eq. (4.6)
determining the (Q −H) characteristic of the pump, where H = p/ρg,
is obeyed. The pump feed Q decreases with increase in p and, conversely,
the smaller the pressure drop which the blower has to overcome, the greater
the pump feed.
Exercise. It is required to determine the maximal differential head developed
by a centrifugal pump with radial located impeller blades having radius 0.25 m
and rotating at 3000 rpm.
Solution. 3000 rpm corresponds to the angular velocity ω = 2π · 3000/60 =
2π · 50 s
−1
. Then, in accordance with Eq. (4.6) we get
(H)
max
=
ω
2
R
b
2
2g
=
4π
2
· 50
2
· 0.25
2
2 · 9.81
∼
= 314.4 m.
The (Q −H) characteristics of centrifugal pumps operating in stationary
regimes are often approximated by the twoterm dependence
H = a −b · Q
2
(4.7)
where the differential head H is measured in (m) and the ﬂow rate Q in
(m
3
h
−1
), therefore the dimension of the factor a is (m) and the factor b
is (m/(m
3
h
−1
)
2
). For example, the main pump HM 1250260 produced in
Russia rated at a nominal feed of 1250 m
3
h
−1
and nominal head 260 m,
has the (Q −H) characteristic H = 331 −0.451 · 10
−4
· Q
2
, main pump
HM 2500230 with impeller diameter D
im
= 430 mm rated at nominal feed
2500 m
3
h
−1
and nominal head 230 m has the (Q −H) characteristic
H = 280 −0.792 · 10
−5
· Q
2
(H in m, Q in m
3
h
−1
) and so on (Vasil’ev
et al., 2002).
Figure 4.2 shows the (Q −H) characteristic of the centrifugal pump HM
2500230.
The two upper curves of this ﬁgure represent the (Q −H) characteristics
of a pump with accessory impellers (385 and 430 mm, the middle curve
shows the dependence of the power consumption N (kW) on the ﬂow rate
78 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
Figure 4.2 Characteristics of the centrifugal pump HM 2500230.
Q and the bottom curve illustrates the dependence of efﬁciency η(%) on
the ﬂow rate of the transported ﬂuid. In the same ﬁgure is also marked the
operating range of the pump, that is the range of ﬂow rates Q of the pump. In
this range (1800 < Q < 3000 m
3
h
−1
) the efﬁciency η ≈ 85% and the power
N ≈ 1600 kW of the pump have maximal values.
4.2.2
OilPumping Station
Pumps connected in series or parallel provide the basis of oilpumping stations
intended to produce driving pressure.
The (Q −H) characteristics of pumps connected in series (Fig. 4.3) are
summarized, the ﬂuid ﬂow rates of the pumps are identical Q
1
= Q
2
= Q and
the differential heads are given by H = H
1
+H
2
.
If H
1
= a
1
−b
1
· Q
2
is the characteristic of the ﬁrst pump and H
2
=
a
2
−b
2
· Q
2
the characteristic of the second pump, the characteristic of a
system of two pumps connected in series is equal to
H = (a
1
+a
2
) −(b
1
+b
2
) · Q
2
. (4.8)
4.2 Boundary Conditions. Modeling of the Operation of Pumps and OilPumping Stations 79
Figure 4.3 Series connection
of pumps.
In parallel connection of pumps (Fig. 4.4) their (Q −H) characteristics are
different. Fluid discharges in pumps are given by Q = Q
1
+Q
2
but the heads
produced by each pump are identical H = H
1
= H
2
.
If H = a
1
−b
1
· Q
2
is the characteristic of the ﬁrst centrifugal pump and
H = a
2
−b
2
· Q
2
that of the second one, the characteristic of the system of
two pumps connected in parallel is
_
(a
1
−H)
b
1
+
_
(a
2
−H)
b
2
= Q. (4.9)
Exercise 1. The (Q −H) characteristic of a centrifugal pump with impeller
diameter 440 mmis H = 331 −0.451 · 10
−4
· Q
2
. Another pump of the same
type but with impeller diameter 465 mm has the (Q −H) characteristic
H = 374 −0.451 · 10
−4
· Q
2
, (H in m, Q in m
3
h
−1
). What characteristic
has a system of two pumps connected in series?
Solution. In accordance with Eq. (4.8) we obtain H = (331 +374) −2 ·
0.451 · 10
−4
· Q
2
= 705 −0.902 · 10
−4
· Q
2
.
Exercise 2. The (Q −H) characteristic of a centrifugal pump with
impeller diameter 440 mm is H = 331 −0.451 · 10
−4
· Q
2
. Another pump
of the same type but with impeller diameter 465 mm has the (Q −H)
characteristic H = 374 −0.451 · 10
−4
· Q
2
, (H in m, Q in m
3
h
−1
). It is
required to ﬁnd the characteristic of a system of two pumps connected in
parallel?
Figure 4.4 Parallel connection of pumps.
80 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
Solution. In accordance with Eq. (4.9) we have
_
(331 −H)
0.451 · 10
−4
+
_
(374 −H)
0.451 · 10
−4
= Q
or
√
331 −H +
√
374 −H = 6.716 · 10
−3
· Q, where H < 331 m.
The (Q −H) characteristic of a pumping station is the total (Q −H)
characteristic of all pumps operating in the station (connected in series or
parallel) minus the (Q −H) characteristics of the supply communications.
The latter is taken as an element connected in series with the pumps of the
station.
Exercise 3. At a pumping station two pumps operate in series with
characteristics H = F
1
(Q) = 331 −0.451 · 10
−4
· Q
2
and H = F
2
(Q) =
374 −0.385 · 10
−4
· Q
2
. It is also known that the head losses h
c
in the station
communications, that is the pipeline system of the station, are represented by
the dependence
h
c
= 25 −0.036 · 10
−4
· Q
2
(H, h
c
in m, Q in m
3
h
−1
). It is required to ﬁnd the characteristic of the
pumping station?
Solution. The characteristic of the pumping station H = F(Q) is repre
sented by the sum of the characteristics of the pump systemminus head losses
in the supply communication F
1
(Q) +F
2
(Q) −h
c
(Q):
H = F(Q) = 680 −0.800 · 10
−4
· Q
2
.
Hence, if (Q −H) the characteristic of the pumping station H = F(Q) is
known, the boundary condition at the initial crosssection x = 0 of the pipeline
section can be the following condition given by this characteristic
p
ex
−p
in
ρg
= F(Q) or
p
0
ρg
=
p
u
ρg
+·
˜
F(v) (4.10)
that is, a condition similar to condition (4.5), where p
0
= p
ex
is the pressure
at the initial crosssection of the pipeline section, p
u
= p
in
the pressure
before the oilpumping station, called the head before pumping station,
F(Q) ≡ F(vS) ≡
˜
F(v). When using the twoterm dependence of the station
differential head H on the ﬂow rate Q the boundary condition (4.10) at the
beginning of the pipeline section takes the form
p
0
ρg
=
p
u
ρg
+a −b · S
2
(3600)
2
· v
2
(4.11)
where the velocity v is measured in (m s
−1
).
4.3 Combined Operation of Linear Pipeline Section and Pumping Station 81
4.3
Combined Operation of Linear Pipeline Section and Pumping Station
To calculate the combined operation of a linear pipeline section and the
pumping station located at the beginning of the pipeline section the Bernoulli
equation (4.4) is used, in which the pressure p
0
= p(0) at the initial cross
section of the pipeline section is excluded with the help of boundary
condition (4.11)
_
¸
¸
_
¸
¸
_
_
p
0
ρg
+z
0
_
−
_
p
L
ρg
+z
L
_
= λ(Re, ε) ·
L
d
·
v
2
2g
,
p
0
ρg
=
p
u
ρg
+(a −1.296 · 10
7
S
2
b · v
2
)
where a and b are the approximation factors of the pumping station (Q −H)
characteristic, the velocity v is measured in (m s
−1
). After eliminating p
0
from
these equations we obtain
p
u
ρg
−
p
L
ρg
+(z
0
−z
L
) +a −1.296 · 10
7
S
2
b · v
2
= λ(Re, ε) ·
L
d
·
v
2
2g
.
(4.12)
This equation is called the head balance equation. At given values of the head
before the pumping station p
u
and pressure at the pipeline section end p
L
Eq. (4.12) serves to determine the unknown velocity v of the ﬂuid ﬂow in the
pipeline.
To solve Eq. (4.12) it is convenient to rearrange all the terms containing
the unknown velocity v on the righthand side of the equation leaving on the
lefthand side only the given quantities
p
u
ρg
−
p
L
ρg
+(z
0
−z
L
) +a =
_
λ ·
L
d
·
1
2g
+1.296 · 10
7
S
2
b
_
· v
2
.
This equation could be solved by the method of successive approximation
(iteration method). We can demonstrate it with exercises.
Exercise 1. Two identical pumps connected in series and having iden
tical (Q −H) characteristics H = 331 −0.451 · 10
−4
· Q
2
, (H in m,
Q in m
3
h
−1
) are pumping diesel fuel (ρ = 840 kg m
−3
, ν = 9 cSt)
along the pipeline section (D = 530 ×8 mm, L = 120 km, = 0.2 mm,
z
0
= 50 m, z
L
= 100 m). It is required to ﬁnd the ﬂow rate and pres
sure at the beginning of the section when the pressure p
L
at the end
of the section is 0.3 MPa, the head before the pumping station h
u
is
30 m and it is known that sections of gravity ﬂows are absent in the
pipeline.
82 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
Solution. Write Eq. (4.11) of the head balance
p
u
ρg
−
p
L
ρg
+(z
0
−z
L
) +2a = (λ ·
L
d
·
1
2g
+1.296 · 10
7
S
2
· 2b)v
2
.
Insertion of the given data yields
30 −
0.3 · 10
6
840 · 9.81
+50 −100 +2 · 331 =
_
λ ·
120 000
0.530 −2 · 0.008
·
1
2 · 9.81
+1.296 · 10
7
_
3.14 · 0.514
2
4
_
2
· 2 · 0.451 · 10
−4
_
· v
2
and
605.6 = v
2
· (11 899.2 · λ +50.3).
If as a ﬁrst approximation it is accepted that λ
(1)
= 0.02 then this equation
gives v = 1.449 m s
−1
. We need to verify whether or not the factor λ is correctly
taken. To do this let us determine the ﬁrst Reynolds number
Re
(1)
=
1.449 · 0.514
_
9 · 10
−6
_ = 82 754.
Then with formula (1.31) we get
λ = 0.11 ·
_
0.2
514
+
68
82 754
_
0.25
∼
= 0.0205 > λ
(1)
= 0.02.
It is seen that the obtained value of the hydraulic resistance factor should be
enhanced.
As a second approximation we take λ
(2)
= 0.0205. Then the given equation
yields v = 1.434 m s
−1
, after which it is necessary to verify whether the factor
λ is correctly taken. We have
Re
(2)
=
1.434 · 0.514
(9 · 10
−6
)
∼
= 81 897;
λ = 0.11 ·
_
0.2
514
+
68
81 897
_
0.25
∼
= 0.0206 ≈ λ
(2)
= 0.0205.
Thus there is good coincidence between the taken and received factor λ.
Hence, v
∼
= 1.434 m s
−1
and
Q =
3.14 · 0.514
2
4
· 1.434
∼
= 0.2974 m
3
s
−1
or
Q = 0.2974 · 3600
∼
= 1071 m
3
h
−1
.
The pressure p
0
at the beginning of the pipeline section is determined with
the formula p
0
= ρg · [h
u
+F(Q)]. As a result we have
4.4 Calculations on the Operation of a Pipeline with Intermediate OilPumping Stations 83
p
0
= 840 · 9.81 ·
_
30 +2 ·
_
331 −0.451 · 10
−4
· 1071
2
__
∼
= 4.85 · 10
6
Pa
or 4.85 MPa.
Answer. 1071 m
3
h
−1
; 4.85 MPa.
Exercise 2. The pumping of crude oil (ρ = 870 kg m
−3
, ν = 25 cSt) is be
ing conducted by two pumps: HM 2500–230 with characteristic H =
251 −0.812 · 10
−5
· Q
2
and HM 3600230 with H = 273 −0.125 · 10
−4
· Q
2
connected in series and rated at feed 1800 m
3
h
−1
. It is known that the
(Q −H) characteristic of the supply communication of the oilpumping
station has the form H = 0.15 · 10
−4
· Q
2
(here and above H is in m
and Q in m
3
h
−1
). It is required to determine the pumping ﬂow rate under
the condition that the oilpipeline section (D = 820 ×10 mm, L = 150 km,
z
0
= 80 m, z
L
= 120 m, h
u
= 70 m, h
L
= 40 m) has an approximately ﬂat
character and that sections of gravity ﬂow are absent. Besides it is known that
head losses due to local resistances comprise ≈ 2% of the head losses due to
friction.
Solution. The equation of the head balance is
_
80 +70 +
_
251 −0.812 · 10
−5
· Q
2
_
+
_
273 −0.125 · 10
−4
· Q
2
_
−0.15 · 10
−4
Q
2
_
−[120 +40] = 1.02 · λ ·
150 000
0.800
·
v
2
2 · 9.81
.
After simpliﬁcation this equation takes the form
514 = v
2
· (9748 · λ +116.6).
This equation is solved by the iteration method.
As a ﬁrst approximation we take λ
(1)
= 0.02. Then the equation gives
v = 1.284 m s
−1
. Now verify whether λ has been chosen correctly.
Re =
1.284 · 0.8
(25 · 10
−6
)
= 41 088,
λ =
0.3164
4
√
41 088
∼
= 0.0222 > λ
(1)
= 0.02.
As the second approximation we take λ
(2)
= 0.0222. Then the equation yields
v = 1.242 m s
−1
. Now verify whether λ has been chosen correctly
Re =
1.242 · 0.8
(25 · 10
−6
)
= 39 744;
λ =
0.3164
4
√
39744
∼
= 0.0224 ≈ λ
(2)
= 0.0222.
Hence v = 1.242 m s
−1
which is equivalent to Q
∼
= 2246 m
3
h
−1
.
Answer. 2246 m
3
h
−1
.
84 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
4.4
Calculations on the Operation of a Pipeline with Intermediate OilPumping
Stations
Consider a pipeline consisting of n successive sections separated by oil
pumping stations (OPS). The transportation of ﬂuid is performed in the so
called pumptopump regime. When intermediate ﬂuid dumping and pumping
are absent we can write the Bernoulli equation for each section
_
¸
¸
_
¸
¸
_
[z
1
+h
n1
+F
1
(Q)] −[z
2
+h
n2
] = h
1–2
(Q),
[z
2
+h
n2
+F
2
(Q)] −[z
3
+h
n3
] = h
2–3
(Q),
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
[z
n
+h
n,n
+F
n
(Q)] −[z
L
+h
L
] = h
n−(n−1)
(Q),
(4.13)
where H = F
1
(Q), H = F
2
(Q), . . . , H = F
n
(Q) are the hydraulic (Q −
H) characteristics of oilpumping stations; h
j−(j−1)
(Q) the head losses in the
sections betweenthe oilpumping stations dependent onthe pumping ﬂowrate
Q; z
1
, z
2
, . . . , z
n
the elevations of the oilpumping stations; h
u,1
, h
u,2
, . . . , h
u,n
the heads before the oilpumping stations equal to h
u,i
= p
u,i
/(ρg); z
L
, h
L
=
p
L
/(ρg) the elevation and piezometric head at the pipeline end (x = L),
respectively.
Equations (4.13) represent a system of n algebraic equations (according to
the number of sections) with n unknown quantities: ﬂow rate Q and (n −1)
heads h
u,j
before the intermediate oilpumping stations.
First consequence (equation of head balance):
Termbyterm summation of the equations of the system (4.13) yields the
equation called the balance equation of the heads for the whole pipeline
(h
u,1
−h
L
) +
j=n
j=1
F
j
(Q) = (z
L
−z
1
) +
j=n
j=1
h
j−(j−1)
(Q) (4.14)
This equation serves to determine the ﬂowrate Q of the ﬂuid (carrying capacity
of the pipeline), with all unknown heads h
u,j
before the intermediate pumping
stations being excluded.
It should be taken into account that the ﬂow rate Q found from Eq. (4.14)
can be realized in the considered pipeline only when the heads h
u,j
of all the
intermediate stations are greater than the minimum allowed value assuring
pump operation without cavitation and the pressure in all crosssections of the
pipeline is less than the permissible value deﬁned by the pipeline strength.
Second consequence (equation for heads before the oilpumping stations):
Termbyterm summation of only the ﬁrst s (s < n) equations of the system
(4.13) yields the equation for the heads h
u,s
before the sth intermediate
pumping station
h
u,s
= h
u,1
+(z
1
−z
L
) +
j=s
j=1
_
F
j
(Q) −h
j−(j−1)
(Q)
_
. (4.15)
4.4 Calculations on the Operation of a Pipeline with Intermediate OilPumping Stations 85
The ﬂow rate Q in this equation is assumed to be given, since it can be
obtained from Eq. (4.14).
In determining the head losses in the pipeline sections it is necessary to
account for the possibility of existing transfer points and segments of gravity
ﬂow in these sections (see Section 3.7). Let us illustrate the aforesaid with an
exercise.
Exercise. An oilpipeline with length L = 450 km consists of three linear
sections the data for which are given in the table below. The head h
u.1
before
the leading oilpumping station is 50 m and the head h
L
at the end of the
pipeline is 30 m.
At the beginning of each section there is an oilpumping station with two
identical pumps connected in series the characteristics of which are given in
the following table
It is required to determine the carrying capacity of the oilpipeline when
pumping oil (ρ = 900 kg m
−3
, ν = 30 cSt) through it and the heads of the
intermediate oil pumping stations.
Solution. The balance equations of the heads for the pipeline sections are
_
50 +50 +2 · (251 −0.812 · 10
−5
· Q
2
)
_
−[60 +h
u,2
]
= λ ·
150 000
0.704
·
v
2
2 · 9.81
,
_
60 +h
u,2
+2 · (285 −0.640 · 10
−5
Q
2
)
_
−[70 +h
u,3
]
= λ ·
180 000
0.704
·
v
2
2 · 9.81
,
_
70 +h
u3
+2 · (236 −0.480 · 10
−5
Q
2
)
_
−[180 +30]
= λ ·
120 000
0.704
·
v
2
2 · 9.81
.
No. Length, km D, mm δ, mm z
0
, m z
L
, m
1. 150 720 8 50 60
2. 180 720 8 60 70
3. 120 720 8 70 180
No. Type of pump (Q−H) characteristic Positive suction head, m
1. II 2500230 H = 251 −0.812 · 10
−5
Q
2
40
2. II 3600230 H = 285 −0.640 · 10
−5
Q
2
40
3. II 5000210 H = 236 −0.480 · 10
−5
Q
2
40
86 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
Here we assume that, due to the invariability of the pipeline diameter, the
rate of pumping and the hydraulic resistance factors are identical when
passing from one section to another; h
u,2
, h
u,3
are the unknown heads of the
intermediate stations that are to be determined.
Termbyterm summation of the above cited equations gives
1434 −3.864 · 10
−5
· Q
2
∼
= 32 579 · λv
2
or
1434 = v
2
· (32 579 · λ +75.8).
This equation(balance of heads for the whole pipeline) is solved by the iteration
method. We take ﬁrst λ
(1)
= 0.02. Then from the above equation we obtain
v
(1)
= 1.404 m s
−1
. Next we verify the correctness of the factor λ:
Re =
1.404 · 0.704
(30 · 10
−6
)
∼
= 32947,
λ =
0.3164
4
√
32947
∼
= 0.0234 > 0.02.
As the second approximation we take λ
(2)
= 0.0234. Then the equation yields
v = 1.308 m s
−1
. We verify again whether the factor λ is correctly chosen:
Re =
1.308 · 0.704
(30 · 10
−6
)
∼
= 30 694;
λ =
0.3164
4
√
30694
∼
= 0.0239 ≈ 0.0234.
Thus v = 1.308 m s
−1
or Q
∼
= 1832 m
3
h
−1
.
From the ﬁrst balance equation of heads we determine h
u,2
[50 +50 +2 · (251 −0.812 · 10
−5
· 1832
2
)] −[60 +h
u,2
]
= 0.0234 ·
150 000
0.704
·
1.308
2
2 · 9.81
and h
u,2
∼
= 52.7 m.
The second balance equation gives h
u,3
[60 +52.7 +2 · (285 −0.640 · 10
−5
· 1832
2
)] −[70 +h
u,3
]
= 0.0234 ·
180000
0.704
·
1.308
2
2 · 9.81
and h
u,3
∼
= 48.0 m.
Both the values for the heads of the intermediate oilpumping stations comply
with the requirement of positive suction head and, therefore, the obtained
pumping regime is realizable.
4.5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 87
4.5
Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with
Heating
To calculate the stationary operating regime of pipelines performing ﬂuid
pumping with heating (hightemperature pumping) Eqs. (4.1)–(4.3) are used.
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
d
dx
(ρvS) = 0
d
dx
_
p
ρg
+z
_
= −λ(Re, ε) ·
1
d
·
vv
2g
ρvC
v
dT
dx
= −
4K
d
(T −T
H
) +λ(Re, ε) ·
1
d
·
v
3
2g
(4.16)
The ﬁrst equation gives ρvS = const., meaning that the mass ﬂow rate is
constant in the stationary operating regime. If we take the pipeline diameter
to be invariable d = d
0
= const., that is we neglect heat expansion of the pipe,
and the ﬂuid density to be insigniﬁcantly variable (ρ ≈ const.), that is we also
ignore heat expansion of the ﬂuid, from the constancy of mass ﬂow rate the
condition of pumping rate constancy v ≈ v
0
= const. follows.
The latter condition v ≈ v
0
= const. allows us to rewrite the system of
differential equations (4.16) as
_
¸
¸
_
¸
¸
_
d
dx
_
p
ρg
+z
_
= −λ(Re, ε) ·
1
d
0
·
v
0
v
0

2g
ρv
0
C
v
dT
dx
= −
4K
d
0
(T −T
H
) +λ(Re, ε) ·
1
d
0
·
v
0

3
2g
(4.17)
As initial conditions, that is conditions at the initial crosssection x = 0 of the
pipeline, we can take
p(0) = p
0
; T(0) = T
0
(4.18)
signifying that the pressure and temperature at the beginning of the pipeline
are known. If it is needed to model the oilpumping station with given
(Q −H) characteristic H = a −b · Q
2
located at the beginning of the
pipeline, the initial conditions at x = 0 should be taken as
p
0
ρg
=
p
u
ρg
+a −b · S
2
0
(3600)
2
· v
0
2
; T(0) = T
0
. (4.19)
Here, in accordance with Eq. (4.11) a and b are approximated factors the of
(Q −H) characteristics of OPS; p
u
the head before the station; S
0
= πd
2
0
/4.
In the general case the system of equations (4.16) should be integrated
numerically or solved using one or other simpliﬁcation suggestions. One
consists in ignoring the heat released in the ﬂuid due to the work of internal
88 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
friction forces as compared to external heat exchange
λ(Re, ε)
1
d
0
·
v
0

3
2g
≤
4K
d
0
· T −T
ex
.
Then the second equation of system (4.16) can be easily integrated and the
solution of this equation yields (see Eq. (1.44))
T(x) = T
ex
+(T
0
−T
ex
) · exp
_
−
πd
0
· K
C
v
˙
M
0
x
_
(4.20)
where
˙
M
0
= ρv
0
S
0
is the mass ﬂow rate; T
ex
the temperature of the external
medium, taken as constant. It should be noted that the ﬂuid velocity v
0
is
constant but unknown.
The ﬁrst equation of the system (4.16) can be represented as
_
p
0
ρg
+z
0
_
−
_
p
L
ρg
+z
L
_
=
_
L
0
λ(Re, ε) dx ·
1
d
0
·
v
0
2
2g
This equation differs from the standard form of the Bernoulli equation in that
it takes into account variability of the factor λ with pipeline length. Really, if
T = T(x), thenT = const., the kinematic viscosity of the ﬂuid ν is not constant,
it depends on temperature, therefore the Reynolds number Re = v
0
d
0
/ν(T) =
Re(x) and the factor of hydraulic resistance λ are functions of x. This circum
stance is taken into account in the Bernoulli equation obtained by integration
of λ over the pipeline section length. With regard to Eq. (4.19) we have
p
u
ρg
−
p
L
ρg
+(z
0
−z
L
) +a −1.296 · 10
7
S
0
2
b · v
0
2
=
_
L
0
λ(Re, ε) dx ·
1
d
0
·
v
0
2
2g
The system of equations
_
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
_
p
u
ρg
−
p
L
ρg
+(z
0
−z
L
) +a −1.296 · 10
7
S
0
2
b · v
0
2
=
_
L
0
λ(Re, ε) dx ·
1
d
0
·
v
0
2
2g
,
T(x) = T
ex
+(T
0
−T
ex
) · exp
_
−
πd
0
· K
C
v
˙
M
0
x
_
(4.21)
provides a basis to determine the unknown velocity v
0
and consequently the
ﬂow rate of ﬂuid pumping with heating. The kinematic viscosity as a function
of temperature ν(T) is assumed to be given, see for example Eq. (2.3).
Analytical Solution
The system of equations (4.21) has an analytical solution when
ν(T) = ν
ex
· e
−κ(T−T
ex
)
,
4.5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 89
where ν
ex
is the kinematic viscosity of the ﬂuid at the temperature T
ex
of the
external media; κ the dependence factor Eq. (2.3);
λ =
0.3164
4
√
Re
that is the ﬂow regime of the ﬂuid corresponds to the ﬂow region of the
socalled hydraulic smooth pipes (Blasius zone). In this case the integral on
the righthand side of the ﬁrst equation of the system could be calculated in
quadratures
_
L
0
λdx =
0.3164
4
√
v
0
d
0
·
_
L
0
ν
1/4
dx =
0.3164
4
√
v
0
d
0
·
_
L
0
ν
1/4
ex
e
−
κ
4
(T−T
ex
)
dx
= λ
ex
·
_
L
0
e
−
κ
4
(T−T
ex
)
dx, where λ
ex
=
0.3164
4
_
v
0
d
0
/ν
ex
(4.22)
If we convert to the dimensionless coordinate ξ = x/L. Then
_
L
0
λdx = L ·
_
1
0
λdξ = L ·
_
T
L
T
0
λ
dξ
dT
dT.
If now in Eq. (1.44) we neglect the heat release, that is we take T
⊗
= 0, then
Eq. (1.44) gives
ρv
0
C
v
dT
dξ
= −
4KL
d
0
· (T −T
ex
) ⇒
dξ
dT
= −
ρv
0
C
v
d
0
4KL
·
1
T −T
ex
.
Thus
_
L
0
λdx = L ·
_
T
L
T
0
λ
dξ
dT
dT = −
ρv
0
C
v
d
0
4KL
· L ·
_
T
L
T
0
λ
T −T
ex
dT.
With regard to Eq. (4.22) we have
_
L
0
λdx = −
ρC
v
v
0
d
0
4KL
· L · λ
ex
_
T
L
T
0
e
−
κ
4
(T−T
ex
)
T −T
ex
dT.
The integral on the righthand side can be transformed to
_
T
L
T
0
e
−
κ
4
(T−T
ex
)
T −T
ex
dT =
_
−
κ
4
(T
L
−T
ex
)
−
κ
4
(T
0
−T
ex
)
e
η
η
dη.
Now gathering together all the results we obtain the following expression
for head losses in a nonisothermal ﬂuid ﬂow
_
L
0
λdx ·
1
d
·
v
0
2
2g
= λ
eff
·
L
d
·
v
0
2
2g
(4.23)
90 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
where λ
eff
is the effective factor of hydraulic resistance determined by
λ
eff
= λ
ex
·
1
m
· [Ei(−k) −Ei(−ke
−m
)]; λ
ex
=
0.3164
4
_
v
0
d
0
/ν
ex
;
k =
κ
4
· (T
0
−T
ex
); m =
4KL
ρC
v
v
0
d
0
=
πKd
0
· L
C
v
˙
M
0
. (4.24)
Here we take into account the equality T
L
−T
ex
= (T
0
−T
ex
) · exp(−m)
following from the basic formula (1.44) and as Ei(z) it is denoted the Euler
function
Ei(z) =
_
z
−∞
e
η
η
dη, (4.25)
which is widely encountered in technical applications and for which there are
special tables. Some values of this function are listed in Table 4.1.
The basic equation (4.21) to determine the velocity v
0
of nonisothermal ﬂuid
ﬂow with regard to Eqs. (4.22)–(4.24) takes the form
p
u
ρg
−
p
L
ρg
+(z
0
−z
L
) +a −1.296 · 10
7
S
0
2
b · v
0
2
= λ
eff
·
L
d
0
·
v
0
2
2g
(4.24)
Exercise. Along a practically horizontal oilpipeline section (D = 720 ×
10 mm, L = 120 km) is pumped oil (ρ = 870 kg m
−3
, C
v
= 2000 J kg
−1
K
−1
),
ν
1
= 5 cSt at T
1
= 50
◦
Cand ν
2
= 40 cSt at T
2
= 20
◦
C) with heating. The initial
temperature T
0
of the oil is 50
◦
C, the temperature T
ex
of the environment
is 10
◦
C. The heattransfer factor K averaged over the pipeline section is
3.5 W m
−2
K
−1
). The pumping is carried out with two pumps connected in
series. The characteristic of each pump is H = 273 −0.125 · 10
−4
Q
2
(H in
m, Q in m
3
h
−1
, h
L
= h
u
). It is required to ﬁnd the ﬂow rate of pumping and
the temperature of the oil at the section end.
Solution. First determine with Eq. (2.30) the dependence of the oil viscosity
on temperature ν(T). We have
ν(T) = 5 · e
−κ·(T−50)
.
Here we take ν(50) = 5 cSt. The second condition ν(20) = 40 cSt yields the
equation for the factor κ
40 = 5 · e
−κ·(20–50)
from which κ
∼
= 0.0693 K
−1
.
Table 4.1
z −1.0 −0.8 −0.6 −0.4 −0.2 −0.1 −0.05
Ei(z) −0.22 −0.31 −0.45 −0.70 −1.22 −1.82 −2.47
4.5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 91
Writing Eq. (4.24) for the balance of heads
2 · [273 −0.125 · 10
−4
Q
2
] = λ
ef f
·
120 000
0.7
·
v
0
2
2 · 9.81
.
The effective factor λ
eff
of the hydraulic resistance taking into account its
variability with pipeline section length in accordance with Eq. (4.23) is
λ
eff
=
0.3164
4
_
v
0
· d
0
/ν
ex
·
1
m
·
_
Ei(−k) −Ei(−ke
−m
)
_
where
k =
κ
4
· (T
0
−T
ex
), m =
πKd
0
· L
C
v
˙
M
0
=
4K · L
ρC
v
v
0
d
0
.
Substituting in the balance equation of heads the expression for the ﬂow rate
Q using the velocity v
0
of the ﬂuid ﬂow
Q =
3.14 · 0.7
2
4
· v
0
· 3600
and taking into account other conditions we get
546 = v
0
2
· (8737.4 · λ
eff
+47.94) (4.26)
which can be solved by the iteration method.
First approximation. Let us take λ
eff
(1)
= 0.02. Then from Eq. (4.26) we ﬁnd
the velocity of ﬂuid ﬂow v
0
(1)
= 1.566 m s
−1
and verify the correctness of the
obtained value. We have
ν
ex
= 5 · exp[−0.0693 · (10 −50)] = 79.95 cSt;
λ
ex
=
0.3164
4
_
1.566 · 0.7/(79.95 · 10
−6
)
∼
= 0.029;
k = 1/4 · 0.0693 · (50 −10) = 0.693;
m =
4 · 3.5 · 120 000
1.566 · 0.7 · 870 · 2000
∼
= 0.881;
k · exp(−m) = 0.693 · exp(−0.881)
∼
= 0.287;
λ
eff
= 0.029 ·
1
0.881
· [Ei(−0.693) −Ei(−0.287)]
= 0.029 ·
1
0.881
· [−0.379 −(−0.939)]
∼
= 0.0186 < λ
eff
= 0.02.
Since there is a difference between the taken and calculated values of λ
eff
we
make a second approximation.
92 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
Second approximation. Let us take λ
eff
(2)
= 0.0186. From Eq. (4.25) we get
the new velocity of ﬂuid ﬂow v
0
(2)
= 1.611 m s
−1
and verify its correctness. We
have
κ = 0.0693 K
−1
; ν
ex
= 79.95 cSt;
λ
ex
=
0.3164
4
_
1.611 · 0.7/(79.95 · 10
−6
)
∼
= 0.029;
k = 1/4 · 0.0693 · (50 −10) = 0.693;
m =
4 · 3.5 · 120 000
1.611 · 0.7 · 870 · 2000
∼
= 0.856;
k · exp(−m) = 0.693 · exp(−0.856)
∼
= 0.294;
λ
eff
= 0.029 ·
1
0.856
· [Ei(−0.693) −Ei(−0.294)]
= 0.029 ·
1
0.856
· [−0.379 −(−0.921)]
∼
= 0.0184 ≈ 0.0186
= λ
eff
(2)
.
Since the taken and calculated values of the factor λ
eff
show good coincidence
the process of successive approximations ends. Hence, v
0
∼
= 1.611 m s
−1
and
Q = 2231 m
3
h
−1
. Hence, the temperature of the oil at the pipeline section end
in accordance with Eq. (4.21) is T
L
= 10 +(50 −10) · exp(−0.856)
∼
= 27
◦
C.
Answer. 2231 m
3
h
−1
, 27
◦
C.
4.6
Modeling of Stationary Operating Regimes of GasPipeline Sections
For modeling the stationary ﬂow of a compressible gas in a gaspipeline the
following equations are used:
•
continuity equation (1.6):
d
dx
(ρvS) = 0 ⇒
˙
M = ρvS = const. (4.27)
Since the gas density ρ decreases with pressure drop, from Eq. (4.27) it
follows that the gas velocity v increases from the beginning of the pipeline
section to its end;
•
momentum equation (1.10):
ρv
dv
dx
= −
dp
dx
−
4
d
τ
w
−ρg sinα(x).
If the gas velocity v increases, the acceleration v · dv/ dx of the gas is distinct
from zero. It is evident that the estimation
4.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 93
ρv
dv
dx
=
d
dx
(ρv
2
)
d
dx
(p)
is valid when the gas velocity v is small compared to the velocity of sound
in a gas c, c =
_
γRT ≈
√
1.31 · 500 · 300
∼
= 440 m s
−1
, where γ = C
p
/C
v
is
the adiabatic index (for methane γ = 1.31). So, for example, when the
velocity of compressed gas with density 50 kg m
−3
varies over v the
quantity ρv · v is 50 · 10 · v = 500 · v at gas velocity 10 m s
−1
, whereas
the pressure variation p calculated with the Joukowski formula p = ρcv
is 50 · 440 · v = 22 000 · v, that is about 45 times greater. Thus, the
acceleration of gas and often the gravity component ρg sinα in the momentum
equation may, as a rule, be neglected. Then the momentumequation of the gas
expresses in essence the equality of the driving forces: pressure and friction
dp
dx
= −
4
d
τ
w
= −
4
d
· C
f
ρv
2
2
= −λ
1
d
·
ρv
2
2
; (4.28)
•
balance equation of total energy
ρvS ·
dJ
dx
= πd · q
n
This equation ignores the work due to the force of gravity. If in this equation
we use the dependence of enthalpy J on pressure p and temperature T, that is
we take J = J(p, T), and the external heat exchange in the form of Newton’s
law (1.42), then
ρv
_
_
∂J
∂T
_
p
·
dT
dx
+
_
∂J
∂p
_
T
dp
dx
_
= −
4K
d
(T −T
ex
).
Denoting (∂J/∂T)
p
= C
p
as the speciﬁc heat of a gas at constant pressure and
(∂J/∂p)
T
= −D
∗
C
p
, where D
∗
is the JouleThompson factor, the last equation
takes the form
ρvC
p
dT
dx
−ρvC
p
D
∗
dp
dx
= −
4K
d
(T −T
ex
). (4.29)
Using the expression for gas enthalpy J involving internal energy and other
parameters of state
J = e
in
+
p
ρ
= C
v
T +Z(p, T) · RT = [C
v
+Z(p, T)R] · T.
The factors C
p
and D
∗
can be expressed through the factors C
v
and Z(p, T) as
follows
C
p
=
_
∂J
∂T
_
p
= C
v
+R ·
_
Z +RT
_
∂Z
∂T
_
p
_
;
D
∗
= −
1
C
p
_
∂J
∂p
_
T
= −
RT
C
p
·
_
∂Z
∂p
_
T
. (4.30)
94 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
From this it is seen that the JouleThompson effect (cooling of heatinsulated
gas in a gaspipeline through a pressure drop, (∂Z/∂p)
T
< 0, see Fig. 2.7)
manifests itself only for a real gas when Z = 1 (as a rule the factor
D
∗
≈ 3–5 K MPa
−1
).
If we neglect the JouleThompson effect, the energy balance equation is
simpliﬁed to
ρvC
p
dT
dx
= −
4K
d
(T −T
ex
) (4.31)
and its solution takes the form of Eq. (1.46):
T(x) −T
ex
T
0
−T
ex
= exp
_
−
πd · K
C
p
˙
M
x
_
. (4.32)
If we deﬁne the average temperature T
av
of the gas over pipeline section length
by the expression
T
av
=
1
L
·
_
L
0
T(x) dx
and substitute into it the distribution T(x) from Eq. (4.32), we get
T
av
= T
ex
+
T
0
−T
L
ln
_
T
0
−T
ex
T
L
−T
ex
_ (4.33)
where the temperature T
L
at the end of pipeline section is represented by the
expression
T
L
−T
ex
= (T
0
−T
ex
) · exp
_
−
πd · K · L
C
p
˙
M
_
following from Eq. (4.32).
4.6.1
Distribution of Pressure in Stationary Gas Flow in a GasPipeline
This distribution is obtained from Eqs. (4.27) and (4.28)
_
_
_
ρvS =
˙
M = const.,
dp
dx
= −λ
1
d
·
ρv
2
2
(4.34)
and the equation of the gas state p = ZρRT. Transformation of the second
equation (4.32) yields
dp
dx
= −λ
1
d
·
ρv
2
2
= −
1
2
· λ ·
ρ
2
v
2
S
2
d · ρS
2
= −
1
2
· λ ·
˙
M
2
(p/ZRT) · (π
2
d
5
/16)
.
4.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 95
If we take
•
λ = const., see e.g. Eq. (1.34);
•
T ≈ T
av
= const.;
•
Z(p, T) ≈ Z
av
= const.
the resulting differential equation gives after integration the distribution of
p(x) along the pipeline section length
p
2
·
dp
dx
= −
16
π
2
·
λ · Z
cp
RT
cp
·
˙
M
2
d
5
or
p
2
(x) = p
0
2
−
16λ · Z
av
RT
av
˙
M
2
π
2
d
5
x. (4.35)
Here we used the initial condition p = p
0
at x = 0. Equation (4.33) means that
p
2
(x) decreases linearly with the pipeline section length
p
2
(x) = p
0
2
−(p
0
2
−p
L
2
) ·
x
L
(4.36)
where p
L
is the pressure at the section end.
If we deﬁne the average pressure p
av
over the pipeline section length as
p
av
=
1
L
·
_
L
0
p(x) dx
and we insert it into the distribution of p(x) from Eq. (4.36), we get
p
av
=
2
3
_
p
0
+
p
L
2
p
0
+p
L
_
. (4.37)
From Eq. (4.36) one gets in particular
•
pressure p
L
(in Pa) at the end of the pipeline section with length L for given
mass ﬂow rate
˙
M
p
L
2
= p
0
2
−
16λ · Z
av
RT
av
· L
π
2
d
5
·
˙
M
2
(4.38)
or
p
L
=
_
p
0
2
−
16λ · Z
av
RT
av
· L
π
2
d
5
·
˙
M
2
; (4.39)
•
mass ﬂow rate
˙
M (in kg s
−1
) for given pressures p
0
and p
L
at the beginning
and end of the pipeline section, respectively
˙
M =
π
4
_
p
0
2
−p
L
2
λ · Z
av
RT
av
· L
· d
5
. (4.40)
96 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
If we take λ = 0.067(2/d)
0.2
in accordance with Eq. (1.34), the mass ﬂow
rate is found to be proportional to d
2.6
˙
M = A ·
_
p
0
2
−p
L
2
L
· d
2,6
where A is the proportionality factor.
The mass ﬂowrate
˙
Mof gas (kg s
−1
) canalso be measured by the volume ﬂow
rate taken under standard conditions, i.e. at p
st
= 101 325 Pa; T
st
= 293.15 K.
To do this it is sufﬁcient to divide
˙
M by the density ρ
st
of gas under standard
conditions
Q
k
=
˙
M
ρ
st
=
ρ
ρ
st
· vS =
ρ
ρ
st
· Q
v
(4.41)
Here ρ and Q
v
= vS are the density and volume ﬂow rate of gas at the pipeline
crosssection, respectively. The quantity Q
k
(m
3
s
−1
) is the socalled commercial
ﬂow rate of gas. In fact the commercial ﬂow rate of gas is the mass ﬂow rate
expressed in volume units under standard conditions, in other words the mass
ﬂow rate in volume calculus. From Eq. (4.41) in particular follows that the
commercial ﬂow rate Q
k
of gas is ρ/ρ
st
times greater than the volume ﬂow
rate Q
v
. The key advantage is that in contrast to the volume ﬂow rate that
varies from one crosssection to another, the commercial ﬂow rate remains
invariable with the length of the gaspipeline in stationary ﬂows.
4.6.2
Pressure Distribution in a GasPipeline with Great Difference in Elevations
In this case one has to use the following equations instead of Eq. (4.34)
_
_
_
ρvS =
˙
M = const.,
dp
dx
= −λ
1
d
·
ρv
2
2
−ρg · sinα, α =
dz
dx
(4.42)
containing the pipeline proﬁle z(x). If we take T ≈ T
av
= const. and
Z ≈ Z
av
= const., we get
dp(x)
2
dx
= −
2g
Z
av
RT
av
dz
dx
· p(x)
2
−
λ · Z
av
RT
av
·
˙
M
2
(πd
2
/4)
2
d
(4.43)
to ﬁnd the function p(x).
The solution of this equation with initial condition p(0) = p
0
is
p(x)
2
= −
λ · Z
av
RT
av
·
˙
M
2
(πd
2
/4)
2
d
·
_
x
0
exp
_
2g
Z
av
RT
av
(z(ς) −z(x))
_
dς
+p
0
2
×exp
_
2g
Z
av
RT
av
(z
0
−z(x))
_
. (4.44)
4.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 97
Assuming p(L) = p
L
in Eq. (4.44), the formula for the mass ﬂow rate of gas
takes a form similar to Eq. (4.40)
˙
M =
π
4
_
p
0∗
2
−p
L
2
λ · Z
av
RT
av
· L
∗
· d
5
(4.45)
in which the initial pressure and the length of the pipeline section are
changed to
_
¸
¸
¸
_
¸
¸
¸
_
p
0∗
= p
0
·
_
exp
_
2g
Z
av
RT
av
(z
0
−z
L
)
_
,
L
∗
= L ·
_
1
L
·
_
L
0
exp
_
2g
Z
av
RT
av
(z (ς) −z
L
)
_
dς
_
(4.46)
It should be noted that the ratio 2g(z −z
L
)/(Z
av
RT
av
) is usually small when the
difference in pipeline elevations is 100–1000 m (Z
av
RT
av
≈ 150 000 m
2
s
−1
).
Therefore Eq. (4.46) could be simpliﬁed by expansion of the exponential
function in the Taylor series. Accurate to the second term we obtain
_
¸
¸
_
¸
¸
_
p
0∗
= p
0
·
_
1 +
g
Z
av
RT
av
(z
0
−z
L
)
_
,
L
∗
= L ·
_
1 +
2g
Z
av
RT
av
(z
av
−z
L
)
_ (4.47)
Here z
av
is the average elevation of the gaspipeline section
z
av
=
1
L
·
_
L
0
z(ς) dς.
Equations (4.47) shows that, even in the case when elevations of the beginning
and the end of the pipeline section coincide, that is at z
0
= z
L
, the length
of the section has to be changed. At z
av
> z
L
, it should be enhanced (gas
pipelines running through a mountain pass), whereas at z
av
< z
L
it should be
diminished (gaspipelines running along the sea bottom).
4.6.3
Calculation of Stationary Operating Regimes of a GasPipeline (General Case)
To calculate the stationary operating regimes of a gaspipeline the following
system of ordinary differential equations is used
_
¸
¸
_
¸
¸
_
dp
dx
= −λ
1
d
·
ρv
2
2
−ρg · sinα(x), sinα =
dz
dx
, (v > 0),
ρvC
p
dT
dx
−ρvC
p
D
∗
dp
dx
= −
4K
d
(T −T
ex
)
98 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
When taking into account that ρg dz dp we have a system
_
¸
¸
_
¸
¸
_
dp
dx
= −λ
1
d
·
ρv
2
2
,
ρvC
p
dT
dx
= −
4K
d
(T −T
ex
) −ρvC
p
D
∗
· λ
1
d
·
ρv
2
2
(4.48)
This system should be supplemented with the following equations
ρvS =
˙
M, ρ =
p
(ZRT)
, Z = Z(p, T);
D
∗
=
−RT(∂Z/∂p)
T
C
p
; λ = 0.067 ·
_
2
d
_
0.2
as well as with initial or initial and boundary conditions.
The following problems can be solved:
•
the pressure p
0
and temperature T
0
at the initial crosssection of the
gaspipeline section as well as the mass ﬂow rate
˙
M of gas are known. It is
required to ﬁnd the pressure p
L
and temperature T
L
at the end of the
gaspipeline section;
•
the pressure p
0
and temperature T
0
at the initial crosssection of the
gaspipeline section as well as the pressure p
L
at the end of the gaspipeline
section are known. It is required to ﬁnd the mass ﬂow rate
˙
M and the
temperature T
L
at the end of the gaspipeline section;
•
the pressure p
0
and temperature T
0
at the initial crosssection of the
gaspipeline section and the pressure p
L
at the end of the gaspipeline
section as well as mass ﬂow rate
˙
M of gas are known. It is required to ﬁnd
the diameter d of the gaspipeline providing this ﬂow rate;
•
at the initial crosssection of the gaspipeline section there is a compressor
station. The pressure p
en
, ﬂow rate Q
en
and temperature T
en
of gas at the
entrance to the compressor station and the pressure p
L
at the end of the
gaspipeline section are known. It is required to choose the number and
type of gaspumping aggregates, compression ratio ε = p
0
/p
en
and the
number of revolutions n of the centrifugal blower shafts.
4.6.4
Investigation of Thermal Regimes of a GasPipeline Section
To consider the thermal regimes of gas transportation it is convenient to pass
into the phase plane of variables (p, O). Dividing the second equation of the
system (4.48) by the ﬁrst equation of the same system, we get the differential
equation
dT
dp
= D
∗
(p, T) +
8
q
3
M
·
K
λ
·
ρ(T −T
ex
)
C
p
(p, T)
(4.49)
4.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 99
containing only one unknown function T(p). This equation should be solved
at the segment p
L
≤ p ≤ p
0
under the condition T(p
0
) = T
0
, where p
0
and T
0
are the pressure and temperature, respectively, of the gas at the beginning of
the pipeline section; p
L
the pressure at the section end being unknown and to
be determined; q
M
= Q
M
S is the speciﬁc mass ﬂow rate.
If we neglect the Joule–Thomson effect (D
∗
≈ 0), the righthand side of
Eq. (4.49) would be positive at T > T
ex
and negative at T < T
ex
. Hence, in the
ﬁrst case the temperature of gas monotone increases with pressure whereas in
the second case it monotone decreases. Mathematically speaking, the straight
line T = T
ex
parallel to the abscissa axis is called the separatrix of Eq. (4.49)
because it separates solutions of different types. When the initial temperature
of the gas exceeds the temperature of the environment, the gas cools when
moving from higher to lower pressure, whereas in the opposite case it heats.
The inclusion of the Joule–Thomson effect changes the pattern of solution
of Eq. (4.49). On the plane (p, T) exists as before a separatrix separating the
different types of solutions of Eq. (4.49). However, this separatrix is no longer
a straight line. It always lies below the straight line T = T
ex
, i.e. it appears to be
in the temperature region below the temperature of the surrounding medium
(see Fig. 4.5, curve 4).
If the temperature of the environment is greater than T
ex
, then dT/ dp > 0.
This means that when the pressure falls from p
0
at the beginning of the
pipeline section to p
L
at the end of the section, it may become less than the
Figure 4.5 Solutions of differential equation (4.49):
1 – monotonic solutions; 2 – solutions with a maximum point;
3 – curve dT/ dp = 0 of the temperature maxima; 4 – separatrix of
solutions of Eq. (4.49).
100 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
temperature of the environment, since if T = T
ex
the derivative dT/ dp > 0 is
independent of the value of the heattransfer factor (see Fig. 4.5 curve 1).
If the initial parameters of the gas are such that the point (p
0
, T
0
) is located
below the separatrix 4 of Eq. (4.49), the derivative dT/ dp could change sign
(see Fig. 4.5, curves 2). The latter happens at points on line 3 determined by
the condition of vanishing of the righthand side of Eq. (4.49), namely
D
∗
(p, T) +
8
q
3
M
·
K
λ
·
ρ(T −T
ex
)
C
p
(p, T)
= 0.
At points (p, T) of this curve dT/ dp = 0, therefore the temperature of the gas at
these points reaches an extremum, namely a maximum. The gas temperature
ﬁrst increases withdecreasing pressure, at the point of intersectionwithcurve 3
it reaches a maximum(T
max
< T
ex
) and thenbegins to diminishmonotonically,
remaining as before lower than the temperature of the environment.
The presence of maximum points on the curve T(p) giving temperature
distribution is related to processes having diverse actions: heating of the gas
due to its mass exchange with the surrounding medium and cooling owing to
the Joule–Thomson effect.
The phenomena under consideration may be used in exploitation of a
gas pipeline section located in permafrost earth. If we provide an initial
temperature of the gas T
0
such that the point (p
0
, T
0
) is located below the
separatrix 4 of the solutions of Eq. (4.49), the temperature of the gas will
remain below the temperature of the permafrost earth all the way along the
pipeline section. This excludes the possibility of warming of the surrounding
ground and imparts greater stability to the pipeline in the earth.
4.7
Modeling of Blower Operation
The motion of gas in a gaspipeline is determined by compressor stations
(GCS) located at the beginning of each pipeline section, or more precisely by
blowers accomplishing gas compression. The main purpose of the blowers
and of GCS as a whole is to force gas to move from a region of lower pressure
(suction region at the GCS entrance) into a region of greater pressure (region
of pumping at the GCS exit). The gas by itself cannot ﬂow against pressure,
therefore it is necessary to spend energy for forced ﬂow in this direction.
Such forced ﬂow of gas against a pressure force is performed by gaspumping
aggregates (GPA) consisting of a drive (gasturbine, electric, gasmotor and so
on) producing rotation of an impeller shaft in centrifugal blowers (CFP) or
the reciprocating motion of a piston in piston engines (PP), and by the blower
itself. Displacement of gas from a region of lower pressure into a region of
higher pressure, in other words gas compression, is accomplished in the blower.
It is quite clear that gas in the gaspipeline moves from a crosssection with
4.7 Modeling of Blower Operation 101
greater pressure to a crosssection with lower pressure overcoming friction
forces.
As a rule the compression station consists of separate plants equipped
with several GPA with blowers connected in parallel in the case of single
step compression or in series in the case of multistep compression. On
main gaspipelines centrifugal blowers are predominantly used following the
same pattern as centrifugal blowers for ﬂuid, see Fig. 4.1. Gas is sucked into
the center of the impeller and thrown by centrifugal rotational force to the
periphery of the impeller in the discharge line. The rotational rate of the
impellers of a large CFP is 4000–15 000 rpm. The gas compression ratio
depends on the type of blower, the rotational speed of its impeller (modern
GPA have commonly controlled rotational speed), pressure and temperature
at the discharge line and above all on gas ﬂow rate.
A mathematical model of the centrifugal blower operating in the stationary
regime involves algebraic dependences of the gas compression ratio ε and the
developed speciﬁc power N/ρ
e
on the gas parameters at the blower suction
line indicated by the subscript e and the number of impeller revolutions n
_
ε =
p
0
p
B
= ε(ρ
B
, p
B
, T
B
, Q
B
, n, . . .),
N = N(ρ
B
, p
B
, T
B
, Q
B
, n, . . .)
(4.50)
where the dots denote parameters related to structural features of the blower.
From dimensional theory (see Chapter 6) it follows that these dependences
may be represented as
_
_
_
ε =
p
0
p
B
= f
_
p
B
/ρ
B
n
2
D
2
im
,
Q
B
/σ
nD
im
, . . .
_
,
N
ρ
B
=
_
n
n
0
_
3
· φ
_
p
B
/ρ
B
n
2
D
2
im
,
Q
B
/σ
nD
im
, . . .
_ (4.51)
where D
im
, σ, n
0
are the diameter of the impeller, the area of the suction
branch pipe crosssection, the nominal number of revolutions of the blower
shaft, respectively. If the structural parameters of the blower are taken as
invariable dependences, Eq. (4.50) takes the following form
_
_
_
ε =
p
0
p
B
= f
_
p
B
/ρ
B
n
2
,
Q
B
n
, . . .
_
,
N
ρ
B
=
_
n
n
0
_
3
· φ
_
p
B
/ρ
B
n
2
,
Q
B
n
, . . .
_ or
_
_
_
ε =
p
0
p
B
= f
_
Z
B
RT
B
n
2
,
Q
B
n
, . . .
_
,
N
ρ
B
=
_
n
n
0
_
3
· φ
_
Z
B
RT
B
n
2
,
Q
B
n
, . . .
_
To determine the universal characteristic of the centrifugal blower the socalled
reduced conditions, denoted by subscripts r, are used. The number of blower
102 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
shaft revolutions in such cases is taken equal to a nominal value n
0
, the
properties of the gas and the conditions at the blower entrance are ﬁxed:
R = R
r
; T = T
r
; Z = Z
r
. Tests of blowers conducted under these conditions
permit the determination of the functions
_
_
_
ε = f
_
Z
r
R
r
T
r
n
0
2
,
Q
B
n
0
, . . .
_
,
_
N
ρ
_
r
= φ
_
Z
r
R
r
T
r
n
0
2
,
Q
B
n
0
, . . .
_
(4.52)
It is evident that the operating characteristics of a blower observed under
arbitrary conditions but not under reduced conditions could be found from the
universal characteristics written as follows
_
_
_
_
_
_
_
_
ε = f
_
Z
B
R
B
T
B
n
2
,
Q
B
n
_
= f
_
Z
r
R
r
T
r
n
0
2
·
n
0
2
n
2
Z
B
R
B
T
B
Z
r
R
r
T
r
,
Q
B
· n
0
/n
n
0
_
,
N
ρ
B
=
_
n
n
0
_
3
·
_
N
ρ
B
_
r
=
_
n
n
0
_
3
· φ
_
Z
r
R
r
T
r
n
0
2
·
n
0
2
n
2
Z
B
R
B
T
B
Z
r
R
r
T
r
,
Q
B
· n
0
/n
n
0
_
It follows that the operation characteristics of each blower under arbitrary
conditions are obtained from some universal characteristics of the same
blower, called reduced, through divisionof the arguments of these characteristic
dependences governing the parameters by
_
n
n
0
_
2
·
Z
r
R
r
T
r
Z
B
RT
B
and
n
n
0
,
respectively. This conclusion has a simple geometrical interpretation. If in the
space to build the graphics of the dependences
ε =
˜
f [(n/n
0
)
r
, (Q
B
)
r
] and (N/ρ
B
)
r
=
˜
φ[(n/n
0
)
r
, (Q
B
)
r
],
the characteristics of a centrifugal blower operating under arbitrary entrance
conditions and at a number of revolutions n distinct from the nominal value
n
0
are determined from these graphics by lengthening of the argument axes
by factors n/n
0
·
_
Z
r
R
r
T
r
/Z
B
RT
B
and n/n
0
, respectively. At this the graphic of
the second dependence is also stretched by the factor (n/n
0
)
3
in the direction
of the function axis.
If we introduce
_
n
n
0
_
r
=
n
n
0
·
_
Z
r
R
r
T
r
Z
B
RT
B
and (Q
B
)
r
= Q
B
·
n
0
n
. (4.53)
4.7 Modeling of Blower Operation 103
The characteristics of the centrifugal blower take the universal form
_
_
_
ε =
˜
f
__
n
n
0
_
r
, (Q
e
)
r
_
,
_
N
ρ
B
_
r
=
˜
φ
__
n
n
0
_
r
, (Q
e
)
r
_
(4.54)
where
N
ρ
B
=
_
n
n
0
_
3
·
_
N
ρ
e
_
r
.
In Fig. 4.6 are depicted reduced (universal) characteristics of one of the
centrifugal blowers 370181 produced in Russia. The parameters of this
blower are: n
0
= 4800 rpm; T
r
= 288 K; Z
r
= 0.9; R
r
= 490 J kg
−1
K
−1
.
Figure 4.6 Reduced characteristics of the blower 370181: T
r
= 288 K;
Z
r
= 0.9; R
r
= 490 J kg
−1
K
−1
.
104 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
Exercise. It is required to determine the rotational speed (number of
revolutions per minute, rpm) of the centrifugal blower shaft 370181 needed
to provide transportation of natural gas (µ = 17.95 kg kmol
−1
, p
cr
= 4.7 MPa,
T
cr
= 194 K) with commercial ﬂow rate 22 million m
3
day
−1
and compression
ratio ε = 1, 25. It is known that the pressure and temperature of the gas at the
suction line of the blower are 3.8 MPa and +15
◦
C, respectively.
Solution. Calculate ﬁrst the parameters of the transported gas
R =
8314
17.95
∼
= 463.1 J kg
−1
K
−1
),
with Eq. (2.14) we obtain:
Z
B
= 1 −0.4273 ·
3.8
4.7
·
_
288
194
_
−3.668
∼
= 0.919
ρ
st
=
p
st
RT
st
=
101 300
463.1 · 293
∼
= 0.746 kg m
−3
ρ
B
=
p
B
Z
B
RT
B
=
3.8 · 10
6
0.919 · 463.1 · 288
∼
= 31.002 kg m
−3
Q
B
= Q
k
· ρ
st
/ρ
B
=
22 · 10
6
24 · 60
·
0.746
31.002
∼
= 367.6 m
3
min
−1
.
Then determine the reduced parameters of the operating regime of the
centrifugal blower
_
n
n
0
_
r
=
n
n
0
_
Z
r
R
r
T
r
Z
B
RT
B
=
n
n
0
_
0.90 · 490 · 288
0.919 · 463.1 · 288
∼
= 1.018 ·
n
n
0
;
(Q
B
)
r
= Q
B
n
0
n
= 367.6 ·
n
0
n
m
3
min
−1
.
Since the compression ratio ε is known and is equal to 1.25, it is necessary,
using the reduced characteristics of the blower 370181 (see Fig. 4.6), to select
a value of n/n
0
such that the point with coordinates Q
B
= 367.6/(n/n
0
) and
ε = 1.25 would lie on the characteristic (n/n
0
)
r
= 1.018 · n/n
0
. The solution is
sought by the iteration method.
1. Let (n/n
0
)
r
= 0.9 ⇒n/n
0
∼
= 0.916;
(Q
B
)
r
= 367.6/0.916
∼
= 401 m
3
min
−1
⇒ε
∼
= 1.2 < 1.25 (see Fig. 4.6),
consequently (n/n
0
)
r
should be increased.
2. Let (n/n
0
)
r
= 1.0 ⇒n/n
0
= 1.0/1.018
∼
= 0.982;
(Q
B
)
r
= 367.6/0.982
∼
= 374 m
3
min
−1
⇒ε
∼
= 1.25 (see Fig. 4.6),
consequently the solution can be taken as correct.
Hence n = 0.982 · n
0
= 0.982 · 4800
∼
= 4714 rpm.
Answer. 4714. rpm.
4.7 Modeling of Blower Operation 105
Useful Power of a Blower
Let us show now how we can estimate the useful power needed for gas
compression from pressure p
B
at the blower entrance to pressure p
0
at the
blower exit.
From the total energy balance equation (1.35) written for a mass of gas going
between the entrance crosssection x
1
and the exit crosssection x
2
of a blower
in the case of stationary ﬂow it follows that
__
α
k
v
2
2
+e
in
+
p
ρ
_
· ρvS
_
x
2
−
__
α
k
v
2
2
+e
in
+
p
ρ
_
· ρvS
_
x
1
= q
ex
+N
us
where q
ex
is the external heat inﬂow (q
ex
> 0) to the gas or heat outﬂow from
the gas (q
ex
< 0); N
us
is the useful power of mechanical forces acting on the
gas, i.e. the useful power of the blower. Taking into account that the mass ﬂow
rate of gas
˙
M = ρvS through all gaspipeline crosssections in stationary ﬂow
remains constant, we have
__
α
K
v
0
2
2
−
α
K
v
B
2
2
+(J
0
−J
B
)
__
·
˙
M = q
ex
+N
us
where J = e
in
+p/ρ is the gas enthalpy. Neglecting the difference (α
K
v
0
2
−
α
K
v
B
2
)/2 in the kinetic energies of the gas before and after compression
and assuming the process of gas compression in the blower to be adiabatic
(q
ex
= 0), we get
N
us
∼
=
˙
M· (J
0
−J
B
) = ρ
B
Q
B
· (J
0
−J
B
) (4.55)
where Q = vS is the volume ﬂow rate of gas (ρ
B
Q
B
= ρ
0
Q
0
= const.).
For a perfect gas J = C
p
T +const. and the following relations are valid
N
us
= ρ
B
Q
B
· (J
0
−J
B
) = ρ
B
C
p
Q
B
· (T
0
−T
B
) =
C
p
R
p
B
Q
B
T
B
(T
0
−T
B
)
or
N
us
=
C
p
C
p
−C
v
· p
B
Q
B
_
T
0
T
B
−1
_
=
γ
γ −1
· p
B
Q
B
_
T
0
T
B
−1
_
where γ = C
p
/C
v
is the adiabatic index (for methane γ ≈ 1.31).
Taking into account that in an adiabatic process the temperature T varies in
accordance with the power law T/T
e
= (p/p
e
)
γ−1
γ
, we ﬁnally obtain
N
us
=
γ
γ −1
· p
B
Q
B
_
ε
γ−1
γ
−1
_
. (4.56)
Here ε = p
0
/p
B
is the compression ratio.
106 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
A similar formula written as a function of the gas parameters at the exit of
the blower has the form
N
us
=
γ
γ −1
· p
0
Q
0
_
1 −ε
1−γ
γ
_
. (4.57)
Here the subscript 0 denotes that the relevant quantity is taken at the exit of
the blower.
Exercise. The pressure p
B
before the blower is 3.5 MPa, the compression ratio
of a gas with γ = 1.31 is 1.4, the volume ﬂow rate at the entrance Q
B
of the
blower is 500 m
3
min
−1
. It is required to determine the useful power N
us
spent
by the blower for gas compression.
Solution. The useful power is obtained from Eq. (4.48)
N
us
=
1.31
1.31 −1
· 3.5 · 10
6
·
500
60
·
_
1.4
1.31−1
1.31
−1
_
∼
= 10.2 · 10
6
W.
Hence the useful power is 10.2 MW.
Answer. 10.2 MW.
Finally in this chapter let us consider an exercise on the calculation of a
gaspipeline section in combination with a compressor station.
Exercise. Natural gas (R = 18.82 kg kmol
−1
, p
cr
= 4.75 MPa, T
cr
= 195 K) is
being transported along a 105km gaspipeline section (D = 1220 ×12 mm,
= 0.03 mm) with the help of two identical GPA equipped with blowers
370181 connected in parallel. It is required to determine the compression
ratio of the gas ε and the number of revolutions n of the blower rotors needed
to ensure a commercial ﬂow rate of 21 billion m
3
year
−1
in the gaspipeline
(the number of working days in a year is taken to be 350). It is given that
the pressure at the end of the pipeline section is 3.8 MPa and at the blower
suction line is 4.7 MPa; the temperature of the gas at the suction line is 12
◦
C;
the temperature of the gas after compression is expected to be 30
◦
C; the
temperature of the surrounding ground is 8
◦
C.
Solution. Taking in Eq. (4.33) x = L and Q
k
=
˙
M/ρ
st
, we have
p
0
2
= p
L
2
+
16λ · Z
av
RT
av
ρ
st
2
· L
π
2
d
5
· Q
k
2
.
Now calculate successively the quantities in this relation
Q
k
=
21 · 10
9
350 · 24 · 3600
∼
= 694.4 m
3
s
−1
;
R =
R
0
µ
=
8314
18.82
∼
= 441.7 J kg
−1
K
−1
;
4.7 Modeling of Blower Operation 107
ρ
st
=
p
st
RT
et
=
10 125
441.7 · 293
∼
= 0.783 kg m
−3
;
λ = 0.067 ·
_
2
d
_
0.2
= 0.067 ·
_
2 · 0.03
1196
_
0.2
∼
= 0.0093.
Calculate the average gas temperature T
av
in the pipeline section
T
av
= T
ex
+
T
0
−T
L
ln
_
T
0.
−T
ex
T
L
−T
ex
_ = 8 +
30 −12
ln
_
30 −8
12 −8
_
∼
= 18.6
◦
C = 291.6 K.
The gas overcompressibility factor Z is calculated with Eq. (2.16) taking the
average pressure p
av
in a ﬁrst approximation to be equal to the pressure at
the end of the pipeline section and the temperature to be the temperature
averaged over the section
Z = 1 −0.4273 · (3.8/4.75) · (291.6/195)
−3.668
∼
= 0.922.
After this we calculate the pressure p
0
at the beginning of the pipeline section.
We begin with the factor A
A =
16 · λ · Z
av
RT
av
ρ
st
2
· L
π
2
d
5
· Q
k
2
=
16 · 0.0093 · 0.922 · 441.7 · 291.6 · 0.783
2
· 105 000
3.14
2
· 1.196
5
· 694.4
2
∼
= 22.73 · 10
12
and then calculate the pressure p
0
p
0
=
_
(3.8 · 10
6
)
2
+22.73 · 10
12 ∼
= 6.1 · 10
6
Pa or 6.1 MPa.
The obtained value shows that the average pressure p
av
inthe pipeline sectionis
equal to 2/3 · (4.5 +6.1
2
/10.6)
∼
= 5.34 MPa, which is greater than the expected
3.8 MPa. Hence, the calculation should be corrected.
Performing the second approximation for pressure p = p
av
= 5.34 MPa, we
get
Z = 1 −0.4273 · (5.34/4.75) · (291.6/195)
−3.668
∼
= 0.890;
A
∼
= 21.94 · 10
12
Pa
2
;
p
0
=
_
(3.8 · 10
6
)
2
+21.94 · 10
12 ∼
= 6.0 · 10
6
Pa or 6.0 MPa.
We see that the obtained value of p
0
is practically unchanged. Thus the
compression ratio ε, which should be provided by the blowers 370181, is
6.0/4.7
∼
= 1.28.
Once the required compressionratio has been obtained, it is time to calculate
the gas parameters at the suction line of each of the blowers connected
108 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines
inparallel. We have
Z
B
= 1 −0.4273 ·
4.7
4.75
·
_
285
195
_
−3.668
∼
= 0.895;
ρ
B
=
p
B
Z
B
RT
B
=
4.7 · 10
6
0.895 · 441.7 · 285
∼
= 41.716 kg m
−3
.
In a parallel connection of identical blowers the ﬂow rate is distributed equally
between them, therefore it is
Q
B
= Q
k
· ρ
st
/ρ
B
=
[(21 000/2)/350] · 10
6
24 · 60
·
0.783
41.716
∼
= 391 m
3
min
−1
.
Determine the reduced parameters of the operating regime of the centrifugal
blower:
_
n
n
0
_
r
=
n
n
0
_
Z
r
R
r
T
r
Z
B
RT
B
=
n
n
0
_
0.90 · 490 · 288
0.895 · 441.7 · 285
∼
= 1.062 ·
n
n
0
;
(Q
B
)
r
= Q
B
n
0
n
= 391 ·
n
0
n
m
3
min
−1
.
As the compression ratio ε has already been found to be equal to 1.28, it is
necessary, using the characteristics of the blower 370181 depicted in Fig. 4.6,
to select n/n
0
in such a way that a point with coordinates (Q
B
)
r
= 391/(n/n
0
)
and ε = 1.28 would lie on the characteristic (n/n
0
)
r
= 1.062 · n/n
0
. The
selection is performed by the iteration method.
1. We take (n/n
0
)
r
= 1.0 ⇒n/n
0
= 1.0/1.062
∼
= 0.942;
(Q
B
)
r
= 391/0.942
∼
= 415 m
3
min
−1
⇒ε
∼
= 1.25 (see Fig. 4.6), which is
less than the required value 1.28. Therefore (n/n
0
)
r
should be increased.
2. Let now (n/n
0
)
r
= 1.05 ⇒n/n
0
= 1.05/1.062
∼
= 0.989;
(Q
B
)
r
= 391/0.989
∼
= 395 m
3
min
−1
⇒ε
∼
= 1.28 (see Fig. 4.6). Hence,
the solution is found.
As a result we have n = 0.989 · n
0
= 0.989 · 4800
∼
= 4750 rpm.
109
5
Closed Mathematical Models of OneDimensional
NonStationary Flows of Fluid and Gas in a Pipeline
In this chapter are considered the most important models of onedimensional
nonstationary ﬂows of ﬂuid and gas in pipelines. The equations obtained in
Chapter 1 are used as a starting point.
5.1
A Model of NonStationary Isothermal Flow of a Slightly Compressible Fluid
in a Pipeline
At the basis of this model lie the following assumptions:
•
the variation of ﬂuid density ρ is much less than its nominal value ρ
0
, that
is ρ ρ
0
, where ρ = ρ
0
×(p −p
0
)/K in accordance with Eq. (2.6). For
example, at ρ
0
= 1000 kg m
−3
, p −p
0
= 1.0 MPa (10
6
Pa ≈ 10 atm),
K = 10
3
MPa (10
9
Pa), the variation of ﬂuid density ρ is only 1 kg m
−3
;
•
the variation of pipeline crosssection area S is much less than its
nominal value S
0
, that is S S
0
, where S = πd
0
3
/4Eδ · (p −p
0
) or
S = S
0
d
0
/Eδ · (p −p
0
). For example, at d
0
= 500 mm, δ = 10 mm,
ρ
0
= 10
3
kg m
−3
, p −p
0
= 10
7
Pa (≈100 atm), E = 2 · 10
11
Pa (pipe steel),
the variation of the pipeline diameter d is 0.06 mm, the variation of the
pipeline crosssection area S is ≈0.5 cm
2
, whereas S
0
∼
= 1960 cm
2
;
•
the tangential friction stress τ
w
 at the pipeline walls in accordance with
Eq. (1.28) is determined by the formula τ
w
 = λ(Re, ε) · ρv
2
/8 with the
factor λ dependent on the governing parameters Re = vd/ν and ε = /d
given in the same form as in stationary ﬂow. Such an assumption is called
the hypothesis of quasistationarity. For example, for λ = 0.02, v = 1.5 m s
−1
,
ρ
0
= 1000 kg m
−3
, then τ
w

∼
= 5.6 Pa.
The ﬁrst equation of the model is continuity equation
∂ρS
∂t
+
∂ρvS
∂x
= 0.
With regard to the accepted assumptions this equation could be transformed
to the following form
∂ρS
∂t
= ρ
∂S
∂t
+S
∂ρ
∂t
≈ ρ
0
dS
dp
∂p
∂t
+S
0
dρ
dp
∂p
∂t
=
_
ρ
0
S
0
d
0
Eδ
+
ρ
0
S
0
K
_
·
∂p
∂t
;
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
110 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
∂ρvS
∂x
≈ ρ
0
S
0
·
∂v
∂x
.
As a result the following equation is obtained
_
ρ
0
K
+
ρ
0
d
0
Eδ
_
·
∂p
∂t
+ρ
0
∂v
∂x
= 0.
The factor in parentheses before the derivative of pressure with respect to time
has dimension inverse to the square of the velocity, therefore it can be denoted
as 1/c
2
, where the parameter c, given by
c =
1
_
ρ
0
K
+
ρ
0
d
0
Eδ
(5.1)
is called the speed of wave propagation in the pipeline (c ≈ 1000 m s
−1
). If
ρ
0
= 1000 kg m
−3
, K = 10
9
Pa, d
0
= 500 mm, δ = 10 mm, E = 2 · 10
11
Pa,
then
c =
1
_
10
3
10
9
+
10
3
· 0.5
2 · 10
11
· 0.01
∼
= 895 m s
−1
.
With regard to the introduced designation the ﬁrst equation of the model takes
the form
∂p
∂t
+ρ
0
c
2
·
∂v
∂x
= 0. (5.2)
The second equation of the model is the momentum equation (1.10)
ρ
_
∂v
∂t
+v
∂v
∂x
_
= −
∂p
∂x
−
4
d
θ
τ
w
−ρg sinα(x)
Replacement of τ
w
with the expression containing the average velocity of the
ﬂow v yields
ρ
_
∂v
∂t
+v
∂v
∂x
_
= −
∂p
∂x
−λ
1
d
0
ρvv
2
−ρg sinα(x). (5.3)
For slightly compressible ﬂuids, among which are water, oil and oil products,
the following simplifying assumptions can be made
∂v
∂t
≈ ρ
0
∂v
∂t
,
ρv
∂v
∂x
∼
= ρ
0
v
∂v
∂x
=
∂
∂x
_
ρ
0
v
2
2
_
,
∂
∂x
_
p +
ρ
0
v
2
2
_
≈
∂p
∂x
.
5.1 A Model of NonStationary Isothermal Flow of a Slightly Compressible Fluid in a Pipeline 111
The last approximation is valid because it is easy to verify that (ρ
0
v
2
/2)
p. Really, at ρ
0
≈ 1000 kg m
−3
and v ≈ 1–2 m s
−1
, (ρ
0
v
2
/2) ≤ 2000 Pa
(≈0.04 atm), whereas p is measured in atmospheres or even tens of
atmospheres. In the general case (ρ
0
v
2
/2) ≈ ρ
0
v v while p ≈ ρ
0
c v,
hence (ρ
0
v
2
/2)/p ≈ v/c. Since v c in pipelines, the ratio (ρ
0
v
2
/2)/p
is negligibly small.
With regard to the given assumptions the momentum equation takes the
following form
ρ
0
∂v
∂t
+
∂p
∂x
= −λ(Re, ε) ·
1
d
0
ρ
0
vv
2
−ρ
0
g sinα(x)
This equation is the second equation of the model.
Hence, the mathematical model of slightly compressible ﬂuids is represented
by a system of two differential equations
_
¸
¸
_
¸
¸
_
∂p
∂t
+ρ
0
c
2
·
∂v
∂x
= 0,
ρ
0
∂v
∂t
+
∂p
∂x
= −λ(Re, ε)
1
d
0
ρ
0
vv
2
−ρ
0
g sinα(x)
(5.4)
to determine the two unknown functions p(x, t) and v(x, t) dependent on the
coordinate x and time t.
The system of differential equations (5.4) requires for its solution initial and
boundary conditions, being also components of the model under consideration.
We will deal with these conditions below.
Virtual Mass
The hypothesis of quasistationarity in accordance with which the tangential
stress τ
w
at the internal surface of the pipe is represented by the equality τ
w
=
λ(Re, ε) · ρ
0
vv/8, asserts in particular that it depends on the instantaneous
value of the average ﬂow velocity v(x, t) but not on derivatives of the velocities
with respect to time and coordinate. In fact the following equation holds
4
d
τ
w
· vS = ρ
0
gvS · i
0
+ρ
0
S
d
dt
_
(α
k
−1)
v
2
2
_
(5.5)
(see Eq. (1.23)), reﬂecting the transformation of the work of friction forces
(4τ
w
· vS/d), and in onedimensional ﬂow appearing as external forces, into
kinetic energy ρ
0
S
d
dt
[(α
k
−1)
v
2
2
] of the intrinsic motion of the ﬂuid layers
relative to its center of mass and into heat ρ
0
gS(v · i
0
) due to the work of
internal friction forces. The values of the factor α
k
in this equation vary from
4/3 for laminar ﬂow up to 1.02–1.05 for turbulent ﬂow. If we take for α
k
the
mean value of this factor, from Eq. (5.5) we get
4
d
τ
w
= ρ
0
(α
k
−1) ·
dv
dt
+ρ
0
g · i
0
(5.6)
112 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
indicating that the tangential stress τ
w
contains the term ρ
0
(α
k
−1) · dv/dt
proportional to the ﬂuid particle acceleration. The physical nature of this term
is hidden in the origin of the additional resistance force ρ
0
(α
k
−1) · ˙ v caused
by realignment of the internal structure of the ﬂow taking place even when
there is dissipation of mechanical energy into heat owing to the work of the
viscous friction force, that is the term ρ
0
g · i
0
, could be neglected. It is evident
that if v = const. the ﬁrst term vanishes.
Substitution of Eq. (5.6) into the momentum equation (5.2) yields
ρ
0
dv
dt
= −
∂p
∂x
−ρ
0
(α
k
−1)
dv
dt
−ρ
0
gi
0
−ρ
0
g sinα(x)
or
α
k
ρ
0
dv
dt
= −
∂p
∂x
−λ(Re, ε) ·
1
d
0
ρ
0
vv
2
−ρ
0
g sinα(x) (5.7)
Equation (5.7) differs from Eq. (5.3) only in that in the lefthand side enters
not the true ﬂuid density ρ
0
but a quantity α
k
ρ
0
differing from ρ
0
by the
factor α
k
. The quantity ρ
0
(α
k
−1) may be called the virtual (additional)
mass of the ﬂuid. Hence, the inertial properties of a ﬂuid in nonstationary
processes are characterized by changing density, to the latter is added a
certain quantity dependent on the ﬂow regime. In developed turbulent
ﬂows this change is slight (α
k
≈ 1.03), but for laminar ﬂow it is greater
(α
k
≈ 4/3).
5.2
A Model of NonStationary Gas Flow in a Pipeline
At the basis of this model lie the following assumptions:
•
the transported media (gas) is compressible, i.e. ρ = ρ(p, T);
•
the variation of gaspipeline crosssection area S can be ignored compared
with the area itself S
0
, i.e. S S
0
. Therefore S
∼
= S
0
= πd
0
2
/4 = const.;
•
the internal energy of the gas is e
in
= C
v
T +const.
The system of basic equations is
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
∂ρ
∂t
+
∂ρv
∂x
= 0,
∂ρv
∂t
+
∂
∂x
(p +ρv
2
) = −
4τ
w
d
0
−ρg sinα,
ρ
_
∂e
in
∂t
+v
∂e
in
∂x
_
=
4q
n
d
0
−p
∂v
∂x
−ρ · n
in
(5.8)
The ﬁrst equation of this system is the continuity equation reﬂecting the law of
gas mass conservation in each pipeline crosssection.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 113
The second equation of the system is the momentum equation expressing
Newton’s second law.
The third equation of the system is the equation of heat inﬂow following
from the laws of total energy conservation of the ﬂow and the variation of
mechanical energy of the transported media.
To the system of equations (5.8) should be added the socalled closing
relations, e.g.
τ
w
=
λ(Re, ε)
8
· ρv
2
, n
in
= −λ(Re, ε) ·
1
d
0
ρv
3
2
,
ρ =
p
Z(p, T) · RT
, e
in
(T) = C
v
· T +const., q
n
= −κ · (T −T
ex
).
Hence, if λ(Re, ε) and Z(p, T) are known as functions of their arguments, the
systemof equations (5.8) represents a closed systemof three partial differential
equations for three unknown functions p(x, t), v(x, t) and T(x, t) dependent
on the coordinate x and time t.
5.3
NonStationary Flow of a Slightly Compressible Fluid in a Pipeline
Consider the nonstationary ﬂow of a slightly compressible ﬂuid in a pipeline.
The basic equations of such ﬂow are represented by the system of
equations (5.4).
5.3.1
Wave Equation
Let us consider ﬁrst the nonstationary ﬂow of a slightly compressible ﬂuid in
a horizontal pipeline (α = 0) neglecting for a while terms accounting for the
friction force. Such an assumption is quite allowable for short pipelines and
ﬂuids with not too high viscosity. In such a case the system of equations (5.4)
takes the form
_
¸
¸
_
¸
¸
_
∂p
∂t
+ρ
0
c
2
·
∂v
∂x
= 0,
ρ
0
∂v
∂t
+
∂p
∂x
= 0
or
_
¸
¸
_
¸
¸
_
∂v
∂x
= −
1
ρ
0
c
2
·
∂p
∂t
,
∂v
∂t
= −
1
ρ
0
·
∂p
∂x
.
(5.9)
114 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Differentiation of the ﬁrst equation with respect to t and the second with
respect to x subject to the condition v
x,t
= v
t,x
yields the equation for p(x, t)
∂
2
p
∂t
2
= c
2
∂
2
p
∂x
2
. (5.10)
This equation is called the wave equation since it describes the propagation of
waves and is encountered in different ﬁelds of physics.
A similar equation can be obtained for ﬂuid velocity v(x, t)
∂
2
v
∂t
2
= c
2
∂
2
v
∂x
2
. (5.11)
Equations (5.10) and (5.11) represent partial differential equations whose
general solution is expressed by two arbitrary functions
p(x, t) = f
1
(x −ct) +f
2
(x +ct) (5.12)
the ﬁrst of these functions is dependent only on the argument ξ = x −ct and
the second on η = x +ct.
Let us show that Eq. (5.12) gives a solution of Eq. (5.10). We have
∂p
∂t
= −cf
1ξ
+cf
2η
,
∂
2
p
∂t
2
= c
2
f
1ξξ
+c
2
f
2ηη
= c
2
(f
1ξξ
+f
2ηη
),
∂p
∂x
= f
1ξ
+f
2η
;
∂
2
p
∂x
2
= f
1ξξ
+f
2ηη
.
c
2
∂
2
p
∂x
2
=
∂
2
p
∂t
2
.
Hence, Eq. (5.12) is a solution of Eq. (5.10) for arbitrary functions f
1
and f
2
.
The function f
1
(x −ct) represents a traveling wave in the positive direction
of the xaxis whereas the function f
2
(x +ct) represents a traveling wave in the
negative direction of this axis. The magnitude of the velocity propagation of
both waves, i.e. the propagation velocity of a certain value of function f
1
or f
2
is identical and equal to c.
The form of each of the functions f
1
and f
2
is determined by the initial
conditions for pressure and velocity distributions in the pipeline, that is by
p(x, 0) and v(x, 0), as well as by the boundary conditions at the pipeline ends.
The velocity of ﬂuid v(x, t) is determined by the formula
v(x, t) = g
1
(x −ct) +g
2
(x +ct).
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 115
Functions g
1
(ξ) and g
2
(η) are expressed through functions f
1
(ξ) and f
2
(η).
From Eq. (5.9) follows
_
¸
¸
_
¸
¸
_
g
1ξ
+g
2η
=
1
ρ
0
c
· [f
1ξ
−f
2η
],
−c · g
1ξ
+c · g
2η
= −
1
ρ
0
[f
1ξ
+f
2η
],
or
_
ρ
0
cg
1ξ
= f
1ξ
,
ρ
0
cg
2η
= −f
2η
.
Integration of the ﬁrst equation with respect to ξ and the second with respect
to η, taking into account that g
1
and f
1
depend only on ξ and g
2
and f
2
only on
η, yields
_
ρ
0
cg
1
= f
1
+const.,
ρ
0
cg
2
= −f
2
+const.
From here follows
ρ
0
c · v(x, t) = f
1
(x −ct) −f
2
(x +ct) = const. (5.13)
With the help of Eqs. (5.12) and (5.13) one can ﬁnd solutions to different
problems. We will consider some of them.
5.3.2
Propagation of Waves in an Inﬁnite Pipeline
A pipeline is called inﬁnite when it runs in the direction of the xaxis from−∞
to +∞. Of course it is only a model of a real pipeline, but it is very useful in
the case of a very long pipeline when boundary effects can be ignored, that is
when waves reﬂected from the beginning and the end of a pipeline could be
neglected. Suppose also that friction forces are absent.
Addition and subtraction of Eqs. (5.12) and (5.13) yield
_
p +ρ
0
c · v = 2 · f
1
(x −ct) +const.,
p −ρ
0
c · v = 2 · f
2
(x +ct) +const.
From these expressions it is seen that at those points of the plane (x, t) where
(x −ct) remains constant the expression I
1
= p +ρ
0
c · v is also constant and at
those points of the plane (x, t) where (x +ct) remains constant the expression
I
2
= p −ρ
0
c · v is also constant (see Fig. 5.1).
The lines x −ct = const. and x +ct = const. are called characteristics of
the wave equation and the quantities I
1
= p(x, t) +ρ
0
c · v(x, t) and I
2
=
p(x, t) −ρ
0
c · v(x, t) are called the Riemann invariants.
Hence, at each characteristic x −ct = const. with positive slope, dx/dt = +c,
the ﬁrst Riemann invariant I
1
is conserved whereas at each characteristic
116 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Figure 5.1 Characteristics on
(x, t)plane.
x +ct = const. with negative slope, dx/dt = −c, the second Riemanninvariant
I
2
is conserved:
at lines x = ct +const. : I
1
= p +ρ
0
cv = const.;
at lines x = −ct +const. : I
2
= p −ρ
0
cv = const.
Problem. Let at an initial instance of time t = 0 in an inﬁnite pipeline
(−∞< x < +∞) there be distributions of pressure p(x, 0) = (x) and ﬂuid
velocity v(x, 0) = (x). It is required to determine, what motion appears
in the pipeline at t > 0, i.e. it is required to ﬁnd the functions p(x, t) and
v(x, t) satisfying Eqs. (5.10) and (5.11) and initial conditions p(x, 0) = (x),
v(x, 0) = (x).
Solution. Consider a plane (x, t) depicted inFig. 5.1. Let M(x, t) be anarbitrary
chosen point of this plane at t > 0. Draw from the point M(x, t) straight lines
(characteristics) MA: x −ct = x
1
and MB: x +ct = x
2
.
Since at the characteristic x −ct = x
1
the ﬁrst Riemann invariant I
1
is
constant, we can write
p
M
(x, t) +ρ
0
c · v
M
(x, t) = p
A
(x
1
, 0) +ρ
0
c · v
A
(x
1
, 0).
At the characteristic x +ct = x
2
the second Riemann invariant I
2
is constant
p
M
(x, t) −ρ
0
c · v
M
(x, t) = p
A
(x
2
, 0) −ρ
0
c · v
A
(x
2
, 0).
Resolving the system of obtained equations relative to p
M
(x, t) and v
M
(x, t),
we get
_
¸
¸
_
¸
¸
_
p
M
(x, t) =
p(x
1
, 0) +p(x
2
, 0)
2
+ρ
0
c ·
v(x
1
, 0) −v(x
2
, 0)
2
,
v
M
(x, t) =
p(x
1
, 0) −p(x
2
, 0)
2ρ
0
c
+
v(x
1
, 0) +v(x
2
, 0)
2
or
_
¸
¸
_
¸
¸
_
p
M
(x, t) =
(x
1
) +(x
2
)
2
+ρ
0
c ·
(x
1
) −(x
2
)
2
,
v
M
(x, t) =
(x
1
) −(x
2
)
2ρ
0
c
+
(x
1
) +(x
2
)
2
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 117
Substituting in these relations instead of x
1
and x
2
their expression through x
and t, we receive the solution of the problem
_
¸
¸
_
¸
¸
_
p
M
(x, t) =
(x −ct) +(x +ct)
2
+ρ
0
c ·
(x −ct) −(x +ct)
2
,
v
M
(x, t) =
(x −ct) −(x +ct)
2ρ
0
c
+
(x −ct) +(x +ct)
2
(5.14)
Since functions (x) and (x) are known, the problem is completely solved.
Equalities (5.14) are called the d’Alambert formulas.
5.3.3
Propagation of Waves in a SemiInﬁnite Pipeline
A pipeline is called semiinﬁnite when it has the initial crosssection (x = 0)
and runs from it in the positive direction of the xaxis (x > 0) to inﬁnity. It
is a model of a pipeline in which the conditions at one of the ends (left end)
are taken into account whereas the inﬂuence of another end (right end) is
neglected.
Problem. Let, at the initial instant of time t = 0 the ﬂuid in a semiinﬁnite
(0 < x < +∞) pipeline be quiescent v(x, 0) = 0 and the pressure constant
p(x, 0) = p
0
. The ﬂuid velocity in the initial crosssection x = 0 at t > 0
suddenly begins to change with a law v(0, t) = (t). It is required to determine
the pattern of ﬂuid ﬂow in the pipeline at t > 0.
Solution. It is required to determine what kind of velocity and pressure waves
begin to propagate in a semiinﬁnite pipeline caused by perturbations created
at the left end of the pipeline.
Consider a plane (x, t) at t > 0, x ≥ 0 as depicted in Fig. 5.2. Draw through
the origin of the coordinates in this plane the characteristic x = ct with positive
slope. At points N of this plane located below the straight line x = ct the
d’Alambert formulas (5.14) give the following equations
_
¸
¸
_
¸
¸
_
p
N
(x, t) =
p
0
+p
0
2
+ρ
0
c ·
0 −0
2
= p
0
,
v
N
(x, t) =
p
0
−p
0
2ρ
0
c
+
0 +0
2
= 0
(5.15)
This result has a simple physical meaning. In the region x > ct of the pipeline,
to which the instant of time t has not yet come, a perturbation (signal) is
propagated with velocity c from the initial pipeline crosssection, the ﬂuid is at
rest as before, that is its velocity is 0 and the pressure is p
0
.
Consider now the region x < ct of the pipeline to which at time t have come
perturbations from the initial crosssection of the pipeline. Draw through a
118 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Figure 5.2 Diagram illustrating
the problem on traveling waves.
point M(x, t) of this region two characteristics MC and MB and from the point
C, located at initial crosssection of the pipeline (time t
0
), the characteristic CA
(Fig. 5.2).
Withthe helpof the conditionat the characteristic CAconsisting inconstancy
of the second Riemann invariant, and relevant boundary condition we ﬁnd the
pressure in the initial pipeline crosssection at time t
0
p
C
(0, t
0
) −ρ
0
c · v
C
(0, t
0
) = p
A
(x
1
, 0) −ρ
0
c · v
C
(x
1
, 0) = p
0
−0 = p
0
.
From this follows
p
C
(0, t
0
) = p
0
+ρ
0
c · v
C
(0, t
0
).
Hence the pressure p
C
at the initial pipeline crosssection, having been initially
equal to p
0
, has varied over ρ
0
c · v
C
(0, t
0
) as a result of the increase in velocity
v
C
(0, t
0
) in this crosssection.
Since the pressure and velocity of ﬂuid at points C and B are now known,
one can ﬁnd the pressure at the point M(x, t)
_
p
M
(x, t) +ρ
0
c · v
M
(x, t) = p
C
+ρ
0
c · v
C
= (p
0
+ρ
0
c · v
0
) +ρ
0
c · v
0
,
p
M
(x, t) −ρ
0
c · v
M
(x, t) = p
B
−ρ
0
c · v
B
= p
0
−ρ
0
c · 0 = p
0
From this follows
_
p
M
(x, t) = p
0
+ρ
0
c · v
C
(0, t
0
),
v
M
(x, t) = v
C
(0, t
0
)
(5.16)
Note that t
0
is determined through the coordinates (x, t) of the point M by
the formula t
0
= t −x/c (the characteristic equation for MC is x = c · (t −t
0
)).
Therefore relations (5.16) have the ﬁnal form
_
¸
_
¸
_
p
M
(x, t) = p
0
+ρ
0
c · v
C
_
0, t −
x
c
_
= p
0
+ρ
0
c ·
_
t −
x
c
_
,
v
M
(x, t) =
_
t −
x
c
_
,
(5.17)
where is a function determining the velocity change at the initial (x = 0)
pipeline crosssection not at the instant of time t but at a later time t −x/c.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 119
It is clear that the quantity x/c is equal to the time in which the perturbation
(signal) reaches the considered crosssection x from the initial one.
Formulas (5.17) show that any velocity change at the initial pipeline cross
section propagates rightwards along the pipeline as a traveling wave giving
rise to a pressure traveling wave exceeding the initial value p
0
by ρ
0
˜ n · v
C
.
Formulas (5.15) and (5.17) give the complete solution of the considered
problem.
Exercise. The pumping in quiescent diesel fuel (ρ
0
= 840 kg m
−3
, c =
1060 m s
−1
) in a semiinﬁnite pipeline has begun with constant ﬂow rate
so that v
0
= 1.5 m s
−1
. It is required to determine by how much the pressure
has increased at the pumping crosssection.
Solution. The ﬁrst formula (5.17) gives p(0, t) −p
0
= ρ
0
c · v(0, t) = ρ
0
cv
0
=
840 · 1060 · 1.5
∼
= 1.34 · 10
6
MPa or approximately 13.6 atm.
Answer. 1.34 MPa.
5.3.4
Propagation of Waves in a Bounded Pipeline Section
Consider the problem on wave interaction in a limited pipeline section
(0 ≤ x ≤ L) between the initial crosssection x = 0 and the ﬁnal crosssection
x = L assuming the absence of friction. In such a problem we need, besides
the initial conditions p(x, 0) and v(x, 0), boundary conditions reﬂecting the
interaction of the pipeline section under consideration with equipment located
at the pipeline ends, that is at x = 0, t > 0 and x = L, t > 0.
Now we should say something about the number of these conditions.
Through each point of the left pipeline section boundary x = 0, t > 0 passes
only one characteristic of the wave equation, namely the characteristic of
negative slope x = −ct +const. Along it the condition p −ρ
0
c · v = const.
should be obeyed. Thus at points on this boundary there is always one
algebraic relation between the quantities p and v. To determine a unique
solution for p and we need an additional relation, that is one more boundary
condition.
Analogously, through each point of the right boundary x = L, t > 0 of the
pipeline section goes only one characteristic with positive slope x = ct +const.
Along this characteristic the compatibility condition p +ρ
0
c · v = const. is
obeyed. Thus at these points there should be an additional boundary
condition.
The form of the boundary conditions could be of a great variety, dependent
on the type of equipment set at the end crosssections of the pipeline section
under consideration. For example, at the left end of the pipeline can be placed
a piston pump providing constant delivery of ﬂuid into the pipeline. Then
120 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
the boundary condition at x = 0 would be v(0, t) = v
0
= const. at all t > 0. If
the right end of the pipeline is open, the boundary condition at the right end
would be p(0, t) = p
0
= const. at t > 0. Of course there are also many other
possible boundary conditions.
Problem. Let a ﬂuid in the pipeline section 0 ≤ x ≤ L be initially (at t = 0)
quiescent (v(x, 0) = 0) and the pressure constant (p(x, 0) = p
0
). At t > 0
the ﬂuid starts to be delivered into the pipeline by the law v(0, t) = (t).
The end crosssection x = L of the pipeline is open to the atmosphere,
so that the pressure at the crosssection is held constant p(L, t) = p
0
. It
is required to determine the motion generated in the pipeline at t > 0
(Fig. 5.3).
Solution. Consider on the plane (x, t) a strip 0 ≤ x ≤ L, t > 0 corresponding
to the variability domain of the problem (Fig. 5.3).
1. The solution in the region 1 enclosed by the triangle OCL, i.e. the
region which the perturbation has not yet reached, is as follows:
p(x, t) = p
0
; v(x, t) = 0.
2. Find now the solution in the region 2 restricted by the triangle OCD
(2L/c is the time of the wave double path lengthwise in the pipeline
section).
Figure 5.3 Interaction of waves in the pipeline section.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 121
Determine ﬁrst the pressure p
S
at an arbitrary point S of the left
boundary
p
S
−ρ
0
c · v
S
= p
Q
−ρ
0
c · v
Q
= p
0
−0,
p
S
(0, t) = p
0
+ρ
0
c · v
S
(t) = p
0
+ρ
0
c · (t).
Then obtain the pressure and velocity at an arbitrary point M(x, t) of the
region under consideration
_
p
M
+ρ
0
c · v
M
= p
S
+ρ
0
c · v
S
= p
0
+2ρ
0
c · (t
S
),
p
M
−ρ
0
c · v
M
= p
0
−ρ
0
c · 0 = p
0
from which follows
_
_
_
p
M
(x, t) = p
0
+ρ
0
c ·
_
t −
x
c
_
,
v
M
(x, t) =
_
t −
x
c
_
3. Find the solution in the region 3 bounded by the triangle CDG (3L/c is
the time of the wave triple path lengthwise in the pipeline section).
Determine ﬁrst the ﬂuid velocity v
F
at an arbitrary point F of the right
boundary
p
F
+ρ
0
c · v
F
= p
S
+ρ
0
c · v
S
= [p
0
+ρ
0
c · (t
S
)] +ρ
0
c · (t
S
)
Since p
F
= p
0
and t
S
= t −L/c, then v
F
(L, t) = 2 · (t −L/c).
After this we get the pressure and velocity at the arbitrary point N(x, t) of
the considered region
_
¸
_
¸
_
p
N
+ρ
0
c · v
N
= p
R
+ρ
0
c · v
R
= p
0
+2ρ
0
c ·
_
t −
x
c
_
,
p
N
−ρ
0
c · v
N
= p
F
−ρ
0
c · v
F
= p
0
−2ρ
0
c ·
_
t −
x
c
_
Hence, p
N
= p
0
; v
N
= 2(t −x/c).
In a similar manner by the method of characteristics the solution in
regions 4, 5 and other regions of the strip under consideration could be
found.
5.3.5
Method of Characteristics
Let us return to the system of equations (5.4) describing nonstationary ﬂow
of a slightly compressible ﬂuid with regard to viscous friction forces
_
¸
¸
_
¸
¸
_
∂p
∂t
+ρ
0
c
2
·
∂v
∂x
= 0,
ρ
0
∂v
∂t
+
∂p
∂x
= −λ(Re, ε)
1
d
0
ρ
0
v
2
2
−ρ
0
g sin α(x).
122 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Multiplication of the second equation by c and addition of the result to the ﬁrst
equation yields
_
∂p
∂t
+c
∂p
∂x
_
+ρ
0
c ·
_
∂v
∂t
+c
∂v
∂x
_
= −λ
ρ
0
cvv
2d
−cρ
0
g sinα.
In a similar manner after subtraction of the second equation multiplied by c
from the ﬁrst one we obtain
_
∂p
∂t
−c
∂p
∂x
_
−ρ
0
c ·
_
∂v
∂t
−c
∂v
∂x
_
= λ
ρ
0
cvv
2d
+cρ
0
g sinα.
If at the plane (x, t) we consider straight lines determined by the equations
1.
dx
dt
= c, x −ct = ξ = const.,
2.
dx
dt
= −c, x +ct = η = const.,
which for the wave equation are called characteristics, it is easy to reveal that
for any parameter A(x, t)
∂A
∂t
+c ·
∂A
∂x
=
_
dA
dt
_
ξ=const.
is valid.
This means that the expression on the lefthand side is the derivative of the
function A(x, t) in the direction of the ﬁrst characteristic (ξ = const.). Similarly
it is true that
∂A
∂t
−c ·
∂A
∂x
=
_
dA
dt
_
η=const.
i.e. the expression on the lefthand side is the derivative of the function A(x, t)
in the direction of the second characteristic (η = const.).
Using now the notion of a directional derivative, one can write the above
obtained equations as follows
_
dp
dt
_
ξ=const.
+ρ
0
c ·
_
dv
dt
_
ξ=const.
= −λ
ρ
0
cvv
2d
−ρ
0
cg sin α,
_
dp
dt
_
η=const.
−ρ
0
c ·
_
dv
dt
_
η=const.
= λ
ρ
0
cvv
2d
+ρ
0
cg sin α,
or
_
¸
¸
_
¸
¸
_
d
dt
(p +ρ
0
c · v)
ξ=const.
= −λ
ρ
0
cvv
2d
−ρ
0
cg sin α,
d
dt
(p −ρ
0
c · v)
η=const.
= λ
ρ
0
cvv
2d
+ρ
0
cg sin α.
(5.18)
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 123
At λ = 0 and α = 0 the righthand sides of Eqs. (5.18) vanish. This means
that along the characteristics of positive slope is conserved the quantity
I
1
= p
0
+ρ
0
c · v, whereas the quantity I
2
= p
0
−ρ
0
c · v is conserved along the
characteristic of negative slope. This conclusion is consistent with the results
obtained above for the wave equation.
If λ = 0, the quantities I
1
and I
2
are not constants at relevant characteristics.
Nevertheless, Eqs. (5.18) may be used to calculate various nonstationary ﬂows
in a pipeline, especially when numerical methods are used.
Let for example at the instant of time t
m−1
(in particular at t = 0), the
distributions of pressure p = p(x, t
m−1
) and velocity v = v(x, t
m−1
) in the
pipeline be known. Show how the values of these parameters at the next
instant of time t
m
= t
m−1
+t could be calculated.
Consider on the plane (x, t) a rectangular grid with coordinate step x
and time step t = x/c (Fig. 5.4). Through the nodes of the resulting grid
let us draw characteristics x = ct +const. and x = −ct +const. of positive
and negative slope, respectively. Continuous distribution of the sought
functions p(x, t) and v(x, t) is replaced with discrete values p
k,m
= p(x
k
, t
m
)
and v
k,m
= v(x
k
, t
m
) of the grid functions at the grid nodes. Suppose that all
values p
k,m−1
and v
k,m−1
are known and it is required to ﬁnd the values p
k,m
and v
k,m
of grid functions at t = t
m
.
Let M(x
k
, t
m
) be anarbitrary point onthe plane (x, t). Replacing the directional
derivatives in Eqs. (5.18) with ﬁnite differences along the characteristics AM
and BM we obtain
_
¸
¸
¸
_
¸
¸
¸
_
(p +ρ
0
c · v)
t
¸
¸
¸
¸
ξ=const.
= −c · φ
A
,
(p −ρ
0
c · v)
t
¸
¸
¸
¸
η=const.
= c · φ
B
,
Figure 5.4 Design diagram of the characteristic method.
124 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
where φ = λ
ρ
0
vv
2d
+ρ
0
g · sin α. In addition
(p +ρ
0
c · v)
¸
¸
ξ=const.
= (p
M
+ρ
0
c · v
M
) −(p
A
+ρ
0
c · v
A
)
(p −ρ
0
c · v)
¸
¸
η=const.
= (p
M
−ρ
0
c · v
M
) −(p
B
−ρ
0
c · v
B
).
From this follows a system of equations to determine the pressure p
M
and
velocity v
M
of the ﬂuid at point M through known values of these parameters
at points A and B
_
p
M
+ρ
0
c · v
M
= p
A
+ρ
0
c · v
A
−t · cφ
A
p
M
−ρ
0
c · v
M
= p
B
−ρ
0
c · v
B
+t · cφ
B
.
or
_
p
k,m
+ρ
0
c · v
k,m
= p
k−1,m−1
+ρ
0
c · v
k−1,m−1
−x · φ
k−1,m−1
p
k,m
−ρ
0
c · v
k,m
= p
k+1,m−1
−ρ
0
c · v
k+1,m−1
+x · φ
k+1,m−1
.
From this system of linear equations we get the pressure p
k,m
and the velocity
v
k,m
of the ﬂow
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
p
k,m
=
p
k−1,m−1
+p
k+1,m−1
2
+ρ
0
c ·
v
k−1,m−1
−v
k+1,m−1
2
+
x
2
· (φ
k+1,m−1
−φ
k−1,m−1
)
v
k,m
=
p
k−1,m−1
−p
k+1,m−1
2ρ
0
c
+
v
k−1,m−1
+v
k+1,m−1
2
−
x
2ρ
0
c
· (φ
k−1,m−1
+φ
k+1,m−1
)
(5.19)
Hence, the recurrent formulas (5.19) give the solution of the formulated
problem, because they allow one to calculate the pressure and velocity of the
ﬂow at the following instant of time t
m
from known values of these parameters
at the preceding instant of time t
m−1
. Since at the ﬁrst instant of time we can
take the initial values of the pressure and velocity of the ﬂow at t = 0, then
calculating successively the pressure and velocity with formulas (5.19), we can
get ﬂow parameters at an arbitrary instant of time t > 0.
5.3.6
Initial, Boundary and Conjugation Conditions
Let us investigate nonstationary ﬂuid ﬂow at a pipeline section 0 ≤ x ≤ L
starting from a certain instant of time t = 0 taken as the initial time. In order
to know how the nonstationary process progresses it is necessary to have
information about the initial and boundary conditions, that is to know the state
of the ﬂow before starting and what is happening at the edges of the pipeline
sections, e.g. at the crosssection x = 0 and x = L. The ﬁrst information is
called the initial conditions and the second the boundary conditions.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 125
Initial Conditions
The state of the pipeline section at the initial instant of time can be arbitrary,
but often as the initial state is taken the stationary ﬂuid ﬂow existing in the
pipeline at the initial instant of time.
Let, for example, for stationary ﬂow, the known ﬂuid ﬂow rate be
Q = Q(x, 0) = Q
0
and the distribution of head be H(x, 0) = H
0
−i
0
· x,
where H
0
= H(0, 0) is the head at the beginning of the pipeline section;
i
0
= (H
0
−H
k
)/L, where H
k
= H(L, 0) is the head at the end of the pipeline
section. Then as initial conditions (m = 1, t
m−1
= t
0
= 0) may be accepted
v(x, 0) =
Q
0
S
0
= const.; p(x, 0) = ρ
0
g · [H
0
−i
0
· x −z(x)]
or
v
k,1
=
Q
0
S
0
= const.; p
k,1
= ρ
0
g(H
0
−i
0
· x
k
−z
k
)
where k = 1, 2, . . . , N +1, x
k
= (k −1) · x, x
1
= 0, x
N+1
= L, z
k
= z(x
k
),
x = L/N (N being the number of parts into which the pipeline section is
divided).
Boundary Conditions
The formulas (5.19) permit one to ﬁnd p and v at any point on the strip
0 < x < L, t > 0 determining the pipeline section except its edges – beginning
(x = 0) and ending (x = L).
Only one characteristic of the negative slope dx/dt = −c comes from the
integration domain to a point M(x = 0) of the left pipeline section boundary
(Fig. 5.5a). It gives one condition for two unknown quantities p
1,m
and v
1,m
p
1,m
−ρ
0
c · v
1,m
= p
2,m−1
−ρ
0
c · v
2,m−1
+x · φ
2,m−1
therefore an additional condition is needed. Such a condition can be an
algebraic equation F(p, v) = 0 expressing the relation between pressure
p
M
(0, t) and velocity v
M
(0, t) at the initial crosssection of the pipeline. As
a rule this condition models the operation of a pumping station and is
represented by its (Q −H) characteristic. Thus, boundary conditions at
points on the left pipeline section boundary may be represented as a system of
equations
x = 0, t > 0 :
_
p
1,m
−ρ
0
c · v
1,m
= p
2,m−1
−ρ
0
c · v
2,m−1
+x · ϕ
2,m−1
F(p
1,m
, v
1,m
) = 0.
Similarly, only one characteristic of the positive slope dx/dt = +c comes
from the domain of integration to a point E(x = L) of the right pipeline
section boundary (Fig. 5.5b), Therefore the boundary condition at points on
126 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Figure 5.5 Calculation of p and v at boundary crosssections.
the right pipeline section boundary may be represented as a system of two
equations:
x = L, t > 0 :
_
p
N+1,m
+ρ
0
c · v
N+1,m
= p
N,m−1
+ρ
0
c · v
N,m−1
−x · ϕ
N,m−1
G(p
N+1,m
, v
N+1,m
) = 0.
The dependence G(p, v) = 0 expresses the relation between the pressure
p
N
(L, t) and velocity v
N
(L, t) at the end of the pipeline section. It should be noted
that there are also possibilities for more complicated boundary conditions.
Conjugation Conditions
If the equipment responsible for process nonstationarity is located inside
the pipeline section, e.g. at the crosssection x
∗
, 0 < x
∗
< L, then at this
crosssection there can exist a discontinuity of hydraulic parameters. Such a
discontinuity requires additional conditions called conjugation conditions.
Let take place, for example, at a pipeline crosssection x
∗
ejection or injection
of ﬂuid with ﬂow rate q (q < 0 ejection; q > 0 injection). Then such a
crosssection is characterized by continuity of pressure and discontinuity
of ﬂow rate or velocity. Let us denote the values of the parameters before
ejection or injection with superscript (−) and after ejection or injection with
superscript (+). Then at the crosssection x
∗
the following conditions should
be obeyed:
p
+
(x
∗
, t) = p
−
(x
∗
, t); v
+
(x
∗
, t) −v
−
(x
∗
, t) =
q
S
.
Then to calculate the three unknown parameters p
k,m
, v
+
k,m
, v
−
k,m
of non
stationary ﬂow at point x
∗
= x
k
we use the following system of three linear
equations
_
¸
¸
_
¸
¸
_
p
k,m
+ρ
0
c · v
−
k,m
= p
k−1,m−1
+ρ
0
c · v
k−1,m−1
−x · φ
k−1,m−1
p
k,m
−ρ
0
c · v
+
k,m
= p
k+1,m−1
−ρ
0
c · v
k+1,m−1
+x · φ
k+1,m−1
v
+
k,m
−v
−
k,m
=
q
S
0
with p
+
k,m
= p
−
k,m
= p
k,m
.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 127
In the case when a gate valve is located at the crosssection x
∗
= x
k
, its
operation is modeled by conjugation conditions
_
¸
_
¸
_
v
+
(x
∗
, t) = v
−
(x
∗
, t)
p
−
(x
∗
, t) −p
+
(x
∗
, t) = ς(t) ·
ρ
0
v
2
(x
∗
, t)
2
where ς(t) is the local resistance factor that varies during valve closing or
opening. The ﬁrst condition means the continuity of the ﬂow rate, whereas
the second condition signiﬁes pressure discontinuity at different sides of the
valve. Thus, the model of the valve is represented by the following system of
three equations
_
¸
_
¸
_
p
−
k,m
+ρ
0
c · v
k,m
= p
k−1,m−1
+ρ
0
c · v
k−1,m−1
−x · φ
k−1,m−1
p
+
k,m
−ρ
0
c · v
k,m
= p
k+1,m−1
−ρ
0
c · v
k+1,m−1
+x · φ
k+1,m−1
p
+
k,m
−p
−
k,m
= ς(t
m
) · ρ
0
v
2
k,m
/2
for the three parameters p
+
k,m
, p
−
k,m
, v
k,m
with v
+
k,m
= v
−
k,m
= v
k,m
.
5.3.7
Hydraulic Shock in Pipes
In all the above stated it was assumed by default that the functions
ρ(x, t), p(x, t) and v(x, t) in the differential equations are differentiable with
respect to time and coordinate and in any case are certainly continuous.
Nevertheless, in engineering there are processes in which these functions
vary very quickly with time and in space. An example of such a process
is hydraulic shock. The essence of hydraulic shock is that the stationary
ﬂow of ﬂuid in a pipeline is disturbed by the abrupt closing or opening
of a gate valve, the switching on or switching off of a pump and so on,
resulting in hard braking or acceleration of the ﬂuid and shock compression of
the ﬂuid particles. The front at which the variation of the hydrodynamic
parameters of the ﬂuid takes place has a relatively small extent and
propagates in the form of a pressure wave downstream and upstream of
the ﬂuid.
Similar phenomena occur in the pipeline in other cases when the velocity
(ﬂow rate) of the ﬂuid varies in a stepwise manner. The possibility of hydraulic
shock should be taken into account in the exploitation of pipelines, since shock
pressure can far exceed permissible standards, leading to pipe breakage and
an emergency situation.
The explanation of hydraulic shock was given by Joukovski in his
article ‘‘On hydraulic shock in watersupply pipes’’ (1899). He connected
the magnitude of the pressure jump [p] with the properties of ﬂuid
128 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
compressibility and the elasticity of the pipe and obtained the following
formula
[p] = ρ
0
D· [v] (5.20)
where D is the velocity of shock wave propagation in the pipeline (see Eq. (5.2))
and [v] the magnitude of the stepwise change in the ﬂuid.
It should be noted that the introduction of stepwise variations (jumps)
of hydrodynamic ﬂow parameters is nothing more than a model of the
phenomenon under consideration. In fact each such discontinuity has a
transition region, though very narrow, from the value of parameter A
+
to the left of the discontinuity front up to the value A
−
of the same
parameter to the right of the front. The quantity [A] = A
+
−A
−
is called
the jump of parameter A at the discontinuity front. To describe the
structure of this transition zone needs as a rule a more complicated
model than the given one. Nevertheless, introduction of the system of
equations (5.4) into consideration of discontinuity solutions has proved to
be very fruitful.
Conditions at Discontinuities (Jumps) of Hydrodynamic Parameters
Now let us obtain conditions which should satisfy the ﬂow parameters at
the discontinuity front. Let the discontinuity front of the ﬂow parameters
propagate with velocity D = dx
ˆ o
/dt in the positive direction of the xaxis
(Fig. 5.6).
The wave of the hydraulic shock is characterized by the parameters of ﬂow
(and pipeline crosssection area) that are subjected to discontinuities, or jumps,
at their mobile front. However the limiting values of these parameters before
ρ
−
, v
−
, p
−
, S
−
and after ρ
+
, v
+
, p
+
, S
+
the wave front cannot be prescribed
arbitrarily. They have to obey the conditions of ﬂuid mass and momentum
conservation.
In the time interval t all ﬂuid particles present at the beginning of this
interval at a distance (D −v
−
) · t before the front pass through the front.
Therefore the mass of ﬂuid ﬂowing in the wave in time t is determined by
the expression ρ
−
S
−
(D −v
−
) · t.
In the same time interval a ﬂuid mass equal to ρ
+
S
+
(D −v
+
) · t ﬂows out
from the wave.
It is evident that the mass of ﬂuid ﬂowing into and out of the wave front
should be equal
Figure 5.6 Hydraulic shock in
a pipeline.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 129
ρ
−
S
−
(D −v
−
) = ρ
+
S
+
(D −v
+
)
or
ρ
+
v
+
S
+
−Dρ
+
S
+
= ρ
−
v
−
S
−
−Dρ
−
S.
If we denote a jump of any parameter by [ ] then the last relation maybe
rewritten as
[ρvS −ρD · S] = 0. (5.21)
Now we can use the theorem that the momentum variation of the ﬂuid mass
that has passed through the wave front is equal to the impulse of the pressure
force
ρ
+
S
+
(D −v
+
)t
. ,, .
m
·v
+
−ρ
−
S
−
(D −v
−
)t
. ,, .
m
·v
−
= (p
−
−p
+
)S
−
· t.
In the righthand side of this equation the projection of the reaction force of
the pipeline lateral surface onto the xaxis is taken into account.
We can write the last equation in the following form
(ρ
+
S
+
v
+2
−ρ
+
S
+
v
+
D) −(ρ
−
S
−
v
−2
−ρ
−
S
−
v
−
D) = −S
−
p
or
[v · (ρvS −Dρ · S)] = −S
−
· [p].
Then, in accordance with equality (5.21), [ρvS −ρD · S] = 0 the obtained
equation may be simpliﬁed to
(ρ
−
v
−
S
−
−Dρ
−
S
−
) · [v] = −S
−
· [p]
or
_
v
−
D
−1
_
· ρ
−
D · [v] = −[p].
With regard to ρ
−
≈ ρ
0
and the smallness of the ratio v
−
/D we get the
Joukovski formula
[p] = ρ
0
D· [v]. (5.22)
We can show that the velocity of the hydraulic shock wave D for a slightly
compressible ﬂuid coincides with the velocity of the perturbation propagation
in a pipeline with elastic walls. In order to do this we can simplify the
relation (5.21) as follows:
ρ
0
S
0
[v] −DS
0
[ρ] −Dρ
0
[S] = 0
130 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
where it is assumed that owing to the smallness of the variations in ρ and S
variations the following approximations can be made
[ρvS] ≈ ρ
0
S
0
[v]; [ρS] ≈ ρ
0
[S] +S
0
[ρ].
The use of Eqs. (5.22), (2.22) and (2.23)
[v] = −
1
ρ
0
D
· [p]; [ρ] =
ρ
0
K
· [p]; [S] =
π · d
0
3
4δ · E
· [p],
yields
S
0
D
· [p] −DS
0
ρ
0
K
· [p] −ρ
0
D
π · d
0
3
4δ · E
· [p] = 0.
Since [p] = 0, we have
1 = D
2
·
_
ρ
0
K
+
ρ
0
d
0
δ · E
_
.
From here it follows that the absolute value of the hydraulic shock wave
velocity D is equal to the velocity of the perturbation propagation in the
pipeline (Eq. (5.1))
D =
1
_
ρ
0
K
+
ρ
0
d
0
δ · E
= c. (5.23)
Exercise 1. Oil (ρ
0
= 870 kg m
−3
, K = 1.5 · 10
9
Pa) ﬂows with velocity
v = 1.0 m s
−1
in a steel pipeline (d
0
= 0.8 m, δ = 10 mm, E = 2 · 10
11
Pa).
It is required to determine how much the pressure rises at the pipeline
crosssection before the gate valve suddenly closes.
Solution. Calculate ﬁrst the velocity D of the hydraulic shock wave
D = c =
1
_
870
1.5 · 10
9
+
870 · 0.8
0.01 · 2 · 10
11
∼
= 1038 m s
−1
.
Then determine with Eq. (5.22) the pressure jump [p] = ρ
0
D· [v] = 870 · 1038 ·
1.0 = 903 060 Pa or approximately 9.21 atm.
Answer. 0.903 MPa.
Exercise 2. It is required to determine how much the pressure rises
at the initial crosssection of a steel pipeline (d
0
= 0.516 m, δ = 8 mm,
E = 2 · 10
11
Pa) on abrupt switching on of the pumps providing the benzene
feed (ρ
0
= 750 kg m
−3
, K = 1.3 · 10
9
Pa) with velocity v = 1.5 m s
−1
.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 131
Solution. Calculate ﬁrst the velocity D of the hydraulic shock wave generated
by the benzene pumping into the pipeline
D = c =
1
_
750
1.3 · 10
9
+
750 · 0.516
0.008 · 2 · 10
11
∼
= 1105 m s
−1
.
Then determine the pressure jump with Eq. (5.22) [p] = ρ
0
D· [v] = 750 · 1105 ·
1.5 = 1 243 125 Pa or approximately 12.7 atm.
Answer. 1.243 MPa.
Remark on carrying out calculations. Notice one interesting circumstance
following from Eqs. (5.22) and (5.23). Let, for example, D > 0, i.e. the wave
of discontinuity propagates in the positive direction of the xaxis. Then the
characteristic of negative slope x +ct = const. intersects the discontinuity at
point I (Fig. 5.7).
Ignoring friction, we have
1. p
− −ρ
0
c · v
− = p
B
−ρ
0
c · v
B
– the condition at the linear segment
B of the characteristic BM;
2. p
+ −p
− = ρ
0
c · (v
+ −v
−) – the condition (5.22) at point I of the
discontinuity;
3. p
M
−ρ
0
c · v
M
= p
+ −ρ
0
c · v
+ – the condition at the linear segment
+
M of the characteristic BM.
Subtraction of the ﬁrst relation from the last one with regard to the second
relation yields
p
M
−ρ
0
c · v
M
= p
B
−ρ
0
c · v
B
.
Hence the Riemann invariants (in the considered case I
2
= p −ρ
0
c · v) are
conserved at the relevant characteristics, even when these characteristics
intersect the shock front.
Problem 1. On the phases of direct hydraulic shock. It is required to study the
nonstationary ﬂowgenerated in a pipeline section 0 ≤ x ≤ L by abrupt closing
of the gate valve at the righthand end (x = L) of the pipeline (direct hydraulic
shock). It is assumed that the ﬂuid before closing the gate (t = 0) was ﬂowing
Figure 5.7 Interaction of a
discontinuity with a small perturbation.
132 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
with constant velocity v = v
0
and friction was absent so that the pressure at
all pipeline crosssections was constant p = p
0
. It is assumed also that at the
initial crosssection (x = 0) the pressure is constant and equal to p
0
.
Solution. The solution is shown in Fig. 5.8. By convention the solution may
be divided into four phases.
1. The phase of direct shock continues for a time t = L/c. When the ﬂuid
ﬂow is stopped, the hydraulic shock wave appears and travels from the
end of the pipeline section to its beginning. The wave front brings the
ﬂuid to an abrupt stop and the pressure after the jump is raised by
ρ
0
c · v
0
.
2. The phase of the reﬂected wave runs from t
1
= L/c to t
2
= 2L/c equal to
double the time taken by the wave path along the pipeline section. In
this phase the shock wave reﬂects at the initial crosssection and begins
to move in the opposite direction. The ﬂuid then ﬂows out of the
pipeline (v = −v
0
) and the pressure falls to its initial value p
0
.
3. In the third phase, lasting from the instant of time t
2
= 2L/c to
t
3
= 3L/c, the ﬂuid continues to ﬂow out of the pipeline with velocity
v = −v
0
whereas the pressure in the wave reﬂected from the
crosssection x = L is lowered stepwise by ρ
0
c · v
0
becoming lower than
its original value. If the pressure is lowered to the saturated vapor
Figure 5.8 Phases of direct hydraulic shock.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 133
pressure of the ﬂuid, the latter would begin to boil, otherwise at
(p
0
−ρ
0
c ×v
0
) > p
s
the ﬂuid does not vaporize and the pipeline section
remains completely ﬁlled with ﬂuid. The velocity behind the shock
front propagating in the negative direction vanishes.
4. In the fourth phase of the process at 3L/c < t < 4L/c the ﬂuid again
begins to ﬂow into the pipeline (v = v
0
) and the pressure becomes equal
to p = p
0
. At the instant of time t
4
= 4L/c the situation returns to the
initial one, after which the process is periodically repeated.
Problem 2. On the damping of the pressure bow shock at the hydraulic shock wave
front (Charniy, 1975). In the presence of friction the hydraulic shock wave is
gradually damped in the pipeline, in particular the magnitude of the pressure
jump at the shock front decreases. It is required to determine the intensity of
such a fall.
Solution. Let the front of the hydraulic shock travel from the end of the
pipeline crosssection x = L to its beginning. Consider two negative slope
characteristics parallel to the discontinuity along its left and righthand sides.
Then in accordance with Eq. (5.18) we have
d
dt
(p
+
−ρ
0
c · v
+
)
¸
¸
¸
¸
˙ x=−c
= λ
+
·
1
d
0
ρ
0
c · v
+
v
+

2
+ρ
0
cg sinα,
d
dt
(p
−
−ρ
0
c · v
−
)
¸
¸
¸
¸
˙ x=−c
= λ
−
·
1
d
0
ρ
0
c · v
−
v
−

2
+ρ
0
cg sinα,
p
+
−p
−
= −ρ
0
c · (v
+
−v
−
).
Subtracting termbyterm from the second equation from the ﬁrst one and
taking into account the third equation, we get
d
dt
(p
+
−p
−
)
¸
¸
¸
¸
˙ x=−c
=
1
2
·
1
d
0
ρ
0
c
2
·
_
λ
−
· v
−
v
−
 −λ
+
· v
+
v
+

_
Taking v
+
= v
0
; v
−
= v
+
−(p
+
−p
−
)/ρ
0
c = v
0
−[p]/ρ
0
c; and [p] = p
+
−p
−
,
one obtains the ordinary differential equation for the variation of the pressure
jump [p] at the wave front
d[p]
dt
¸
¸
¸
¸
˙ x=−c
= −
ρ
0
c
4d
0
· (λ
0
v
0
2
−λ
−
v
−2
) (5.24)
This equation should be integrated under the initial condition [p] = −ρ
0
c · v
0
at t = 0. If λ
−
(Re
−
, ε) = λ
−
(v
−
) is known, the solution of the problem can be
obtained without difﬁculty.
At λ
0
v
0
≈ λ
−
v
−
= 2a > 0, where a is a constant, the solution has a particu
larly simple form (for laminar ﬂow this result gives an exact solution). We have
d[p]
dt
¸
¸
¸
¸
˙ x=−c
= −
aρ
0
c
2d
0
· (v
0
−v
−
) = −
a
2d
0
· [p] (5.25)
134 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
and
[p] = −ρ
0
cv
0
· exp(−a/2d
0
· t), (5.26)
that is, the pressure jump at the hydraulic shock wave front decays
exponentially.
From the obtained formula follows, in particular, that in pipelines of large
diameter the value of the bow pressure jump at the hydraulic shock wave front
decays more slowly than in pipelines of smaller diameter.
Problem 3. At the junction of two pipes with different internal diameters
d
1
and d
2
comes, as viewed from the ﬁrst pipe, the pressure shock wave
(incident wave) with pressure amplitude of magnitude [p
inc
]. There are then
generated two pressure waves: one is the reﬂected wave, which travels in the
opposite direction with amplitude [p
reﬂ
]; the other is the transmitted wave
with amplitude [p
trans
], which travels through the second pipe. It is required
to express the amplitudes of the reﬂected and transmitted waves through the
amplitude of the incident wave, if the velocities of wave propagation in the ﬁrst
and second pipes are c
1
and c
2
respectively.
Answer. [p
reﬂ
] = [p
inc
] ·
c
1
d
2
2
−c
2
d
2
1
c
1
d
2
2
+c
2
d
2
1
; [p
trans
] = [p
inc
] ·
2c
2
d
2
1
c
1
d
2
2
+c
2
d
2
1
.
Remark. From the solution of the problem it follows that at c
1
d
2
2
= c
2
d
2
1
or
d
2
= d
1
·
√
c
2
/c
1
, [p
ref
] = 0, that is, the reﬂected wave is not generated at the
junction of the pipes.
Problem 4. It is required to prove that the amplitude of the hydraulic shock
wave falling at the closed pipeline end will be doubled when reﬂecting from
the pipeline end.
Hint. Use the results of the previous problem.
5.3.8
Accounting for Virtual Mass
In Section 5.1 it was indicated that in nonstationary processes the inertial
properties of ﬂuid in a pipeline are characterized by variable density, hence an
additional quantity called virtual mass should be added. The main equations of
nonstationary ﬂow of a slightly compressible ﬂuid in a pipeline with regard to
virtual mass take the form
_
¸
¸
_
¸
¸
_
∂p
∂t
+ρ
0
c
2
·
∂v
∂x
= 0
α
k
ρ
0
dv
dt
= −
∂p
∂x
−λ(Re, ε) ·
1
d
ρ
0
vv
2
−ρ
0
g sinα(x)
(5.27)
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 135
If we denote ρ
∗
= α
k
ρ
0
and –c
∗
= –c/
√
α
k
, the system of equations (5.27)
transforms into
_
¸
¸
_
¸
¸
_
∂p
∂t
+ρ
∗
c
∗
2
·
∂v
∂x
= 0
ρ
∗
dv
dt
= −
∂p
∂x
−λ(Re, ε) ·
1
d
ρ
0
vv
2
−ρ
0
g sinα(x)
(5.28)
equivalent to the system of equations (5.4). From this follows that the
Joukowski formula (5.20) for the amplitude [p] of the hydraulic shock varies; it
will be enhanced by the factor
√
α
κ
[p] = ρ
∗
c
∗
· [v] = α
k
ρ
0
·
c
√
α
k
· [v] = ρ
0
c
√
α
κ
· [v] (5.29)
The velocity D of hydraulic shock wave propagation will also vary, it will be
reduced by a factor
√
α
κ
D =
1
_
ρ
∗
K
+
ρ
∗
d
0
δ · E
= c
∗
=
c
√
α
κ
(5.30)
If we take into account that for turbulent ﬂow α
k
≈ 1.03, corrections to
Joukovski formula will be small (one of them (5.29) was ﬁrst obtained
by Leibenson et al., 1934). But for laminar ﬂowα
k
≈ 1.33 and these corrections
could be signiﬁcant.
5.3.9
Hydraulic Shock in an Industrial Pipeline Caused by Instantaneous Closing of the
Gate Valve
In Fig. 5.9 is shown the development of the hydraulic shock in a pipeline
section with diameter D = 1020 mm (δ = 10 mm) and length L = 100 km
upon instantaneous closing of the gate valve at the righthand edge of
the pipeline section. The pipeline is transporting crude oil with density
ρ
0
= 880 kg m
−3
and viscosity ν = 20 cSt. At the initial instant of time t = 0
the ﬂow in the pipeline is stationary with velocity v(x, 0) = v
0
≈ 1.5 m s
−1
;
the velocity of the hydraulic shock wave propagation c is equal to 870 m s
−1
.
The abrupt closing of the gate valve at the righthand edge of the pipeline is
modeled by the condition v(L, t) = 0 at t > 0.
Figures 5.9a and 5.9b demonstrate the distributions of the head (z +p/ρg)
at 2 and 40 s, respectively, after closing the gate valve.
The initial value of the pressure jump p
f
at the wave front is related to the
initial velocity v
0
by
p
f
= ρ
0
cv
0
= 880 · 870 · 1.5
∼
= 1.15 MPa or ≈ 11.7 atm.
Figures 5.10a and 5.10b showthe following stages of pressure wave upstream
propagation.
136 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Figure 5.9 Pressure wave propagation: (a) 2 s and (b) 40 s after appearance of the wave.
The velocity v(x, t) of the ﬂuid ﬂow behind the pressure wave front is small,
therefore head losses in this region are also small. That is why the head, and
consequently the pressure before the closed gate valve, are always raised. The
head H
L
(t) = z
L
+p
L
(t)/ρg at the end of the pipeline section, i.e. before the
gate valve, can be taken as approximately equal to the head after the pressure
wave front, that is to
H
L
(t) = z
L
+
p
L
(t)
ρ
0
g
H
f
+
[p]
ρ
0
g
,
where z
L
is the elevation of the section end and H
f
the head at the pipeline
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 137
Figure 5.10 Pressure wave propagations: (a) 80 s and (b) 100 s after appearance of the
wave.
crosssection which the hydraulic wave shock has reached. From this it follows
that for an approximate estimation of the pressure at the pipeline section end
one can use the formula
p
L
(t) ρ
0
g(H
f
−z
L
) +[p]
or
p
L
(t) ≈ ρ
0
gi
0
· ct +[p], (5.31)
where i
0
is the hydraulic gradient of the ﬂow in the undisturbed region and t
is the time elapsed after closure of the gate valve.
138 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
It should be noted that owing to viscous friction the ﬂuid behind the wave
front would not come to a halt immediately but gradually, therefore the
pressure shock amplitude [p] reduces monotonically (see formula (5.26)).
Protection of Pipelines from Hydraulic Shocks
The necessity to prevent the destructive force of hydraulic shock in pipelines
transporting heavy dropping liquids (oil, oil products, water and other) is
manifested by the fact that such pipelines, as distinct from gaspipelines, are
not equipped with cocks that close the pipeline crosssection too rapidly, but
are equipped with valves, gates and slowly closing cocks. All of these should
ensure safe ﬂuid braking in the pipeline.
Pumping stations are sometimes equipped with special devices intended
to protect pipelines from hydraulic shock waves. In the suction lines of
pumping stations are, for example, ﬂow dampers of hydraulic shock – special
safety valves or systems of pressure wave smoothing in case of a sudden pumping
station switchoff when the pressure before the station begins to build up.
These devices operate on the principle of emergency discharge of part of the
ﬂuid from the pipeline into a special reservoir to decrease the magnitude
of the pressure and its rate of increase. The safety valves open the ﬂuid
discharge when the pressure exceeds a certain value, the systems of pressure
wave smoothing are switched on when the rate of pressure buildup in
the suction line of the pumping station is greater than the permissible
magnitude.
5.4
NonIsothermal Gas Flow in GasPipelines
Consider now nonstationary and nonisothermal gas ﬂow in gaspipelines.
The main distinction of such ﬂows from the ﬂows considered above is that the
gas represents a signiﬁcantly compressible medium with density dependent not
only on pressure but also on temperature. Thus, to describe these ﬂows it is
necessary not only to use the laws of mass and momentum conservation
but also the law of energy transformation. In other words, as well as
continuity and momentum equations, the equation of heat inﬂow should
be invoked.
Basic equations for the calculation of nonstationary nonisothermal ﬂows in
a gaspipeline are represented by the system (5.8) giving
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
∂ρ
∂t
+
∂ρv
∂x
= 0
∂ρv
∂t
+
∂
∂x
(p +ρv
2
) = −λ(Re, ε)
1
d
0
ρv · v
2
−ρg sinα
ρ
_
∂e
in
∂t
+v
∂e
in
∂x
_
= −
4κ
d
0
(T −T
ex
) −p
∂v
∂x
+λ(Re, ε)
1
d
0
ρv
3
2
(5.32)
5.4 NonIsothermal Gas Flow in GasPipelines 139
in which ρ = p/ZRT is the equation of the gas state and λ(Re, ε) is the
hydraulic resistance factor.
The system of equations (5.32) represents partial differential equations for
three unknown functions p(x, t), v(x, t) and T(x, t) dependent on x and t.
These nonlinear equations have a complicated structure, therefore the
question arises: how to solve them? To answer this question it is necessary
to examine the structure of the system (5.32) treating it as well as was done
when investigating equations (5.4) in the model of nonstationary ﬂow of a
slightly compressible ﬂuid. In the course of examining this model we have
seen that on the plane of variables (x, t) there are certain lines (characteristics)
along which the system of partial differential equations (5.4) transforms into
an ordinary differential equation providing a relation between unknown
functions. Through each point M(x, t) of the plane (x, t) go just two (by the
number of equations) such lines, or strictly speaking
d
dt
(p +ρ
0
cv) = I
1
(p, v) along the line ˙ x = c (or x = ct +const.),
d
dt
(p −ρ
0
cv) = I
2
(p, v) along the line ˙ x = c (or x = −ct +const.).
The presence of two characteristics is the distinctive property of the considered
system of equations and allows us to assign the system to a class of hyperbolic
differential equations and give a constructive method of its solution.
Let us show that the system of equations (5.32) is hyperbolic (the number of
its characteristics is equal to the number of equations, that is three). We look
for characteristics on the plane (x, t), i.e. lines x = x(t) such that along them
Eqs. (5.32) give certain differential relations between unknown functions.
Let x = x(t) be a line on the plane (x, t) on which the values of the functions
p(x, t), v(x, t), T(x, t) are known. Then it is possible to write three equations
relating the derivatives
∂p
∂t
,
∂p
∂x
,
∂v
∂t
,
∂v
∂x
and
∂T
∂t
,
∂T
∂x
of these functions along the
line x = x(t). To do this let us differentiate p, v, T along x = x(t)
dp
dt
¸
¸
¸
¸
x(t)
=
∂p
∂t
+
∂p
∂x
·
dx
dt
dv
dt
¸
¸
¸
¸
x(t)
=
∂v
∂t
+
∂v
∂x
·
dx
dt
(5.33)
dT
dt
¸
¸
¸
¸
x(t)
=
∂T
∂t
+
∂T
∂x
·
dx
dt
where dx/dt is the slope of this line (we shall call it a characteristic) to
the taxis. The slope can be taken as given since the function x = x(t) is
known.
Since the lefthand sides of Eqs. (5.33) are known as well as the slope dx/dt
of the line x(t), these equations could be considered as three linear equations to
determine six partial derivatives p, v, T with respect to time and coordinate. If
140 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
to these equations we add three equations of the system (5.32), which reduces
also to linear equations with respect to the same derivatives, we get six linear
equations for six partial derivatives
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
_
∂ρ
∂p
_
T
∂p
∂t
+v
_
∂ρ
∂p
_
T
∂p
∂x
+ρ
∂v
∂x
+
_
∂ρ
∂T
_
p
∂T
∂t
+v
_
∂ρ
∂T
_
p
∂T
∂x
= 0
v
_
∂ρ
∂p
_
T
∂p
∂t
+
_
1 +v
2
_
∂ρ
∂p
_
T
_
∂p
∂x
+ρ
∂v
∂t
+2ρv
∂v
∂x
+v
_
∂ρ
∂T
_
p
∂T
∂t
+v
2
_
∂ρ
∂T
_
p
∂T
∂x
= J
2
p
∂v
∂x
+ρC
v
∂T
∂t
+ρvC
v
∂T
∂x
= J
3
∂p
∂t
+
dx
dt
·
∂p
∂x
= J
4
∂v
∂t
+
dx
dt
·
∂v
∂x
= J
5
∂T
∂t
+
dx
dt
·
∂T
∂x
= J
6
where J
2
= −λρv · v/2d
0
−ρg sinα; J
3
= −4κ(T −T
ex
)/d
0
+λρv
3
/2d
0
; J
4
=
(dp/dt)
x(t)
; J
5
= (dv/dt)
x(t)
; J
6
= (dT/dt)
x(t)
. The total derivatives of the
functions p, v and T are taken along the line x(t), on which their values
are known.
If the principal determinant of the system is nonvanishing, all six partial
derivatives as well as the functions p, v, T along the curve x(t) can be
determined uniquely and independently from each other. If this determinant
vanishes and the system of linear equations is compatible, the dependence
between the values of p, v, T on the curve x(t) exists.
Let us equate the principal determinant of the system of six linear equations
to zero
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
∂ρ
∂p
_
T
v
_
∂ρ
∂p
_
T
0 ρ
_
∂ρ
∂T
_
p
v
_
∂ρ
∂T
_
p
v
_
∂ρ
∂p
_
T
1 +v
2
_
∂ρ
∂p
_
T
ρ 2ρv v
_
∂ρ
∂T
_
p
v
2
_
∂ρ
∂T
_
p
0 0 0 p ρC
v
ρvC
v
1 dx/dt 0 0 0 0
0 0 1 dx/dt 0 0
0 0 0 0 1 dx/dt
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
= 0
and calculate it. To do this, let us multiply the ﬁrst, third and ﬁfth columns of
this determinant by dx/dt and then subtract the resulting products from the
5.4 NonIsothermal Gas Flow in GasPipelines 141
second, fourth and sixth columns of the same determinant
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
∂ρ
∂p
_
T
(v − ˙ x)
_
∂ρ
∂p
_
T
0 ρ
_
∂ρ
∂T
_
p
_
∂ρ
∂T
_
p
(v − ˙ x)
v
_
∂ρ
∂p
_
T
1 +v(v − ˙ x)
_
∂ρ
∂p
_
T
ρ ρ(2v − ˙ x) v
_
∂ρ
∂T
_
p
_
∂ρ
∂T
_
p
v(v − ˙ x)
0 0 0 p ρC
v
ρC
v
(v − ˙ x)
1 0 0 0 0 0
0 0 1 0 0 0
0 0 0 0 1 0
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
The obtained determinant can be calculated by the method of determinant
decomposition in terms of the thirdcolumn elements. As a result we get the
cubic equation
ρC
v
_
dx
dt
−v
_
3
_
∂ρ
∂p
_
T
+
_
p
ρ
_
∂ρ
∂T
_
p
−ρC
v
_
_
dx
dt
−v
_
= 0 (5.34)
with respect to the difference (dx/dt −v). Roots of this equation are evident
1.
dx
dt
−v = 0 ⇒
dx
dt
= v;
2.
dx
dt
−v = ±
_
C
v
−p/ρ
2
· (∂ρ/∂T)
p
C
v
_
∂p
∂ρ
_
T
.
Since C
v
=
∂e
in
∂T
and −
p
ρ
2
_
∂ρ
∂T
_
p
=
∂
∂T
_
p
ρ
_
p
, then
C
v
−
p
ρ
2
_
∂ρ
∂T
_
p
=
∂
∂T
_
e
in
+
p
ρ
_
p
=
_
∂J
∂T
_
p
= C
p
where J(p, T) is enthalpy. The expression under the square root is simpliﬁed to
dx
dt
−v = ±
_
C
p
C
v
_
∂p
∂ρ
_
T
= ±c
where
c
2
=
C
p
C
v
·
_
∂p
∂ρ
_
T
= γ ·
_
∂p
∂ρ
_
T
and γ(p, T) = C
p
/C
v
is the adiabatic index.
The quantity c having the dimension of velocity is called the adiabatic velocity
of sound in gas. For a perfect gas (∂p/∂ρ)
T
= RT, γ = const., c =
_
γRT. For
example, at γ = 1.31; R = 450 J kg
−1
K
−1
; T = 273 K the velocity of sound is
equal to c =
√
1.31 · 450 · 273
∼
= 400 m s
−1
.
142 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
The equation (5.34) provides three families of characteristics, two of them
having velocities (slopes) (dx/dt)
1,2
= v ±c and the third dx/dt = v. Since
the gas velocity v in gaspipelines as a rule does not exceed 10 m s
−1
and
c ≈ 400 m s
−1
, then sometimes we could take dx/dt
∼
= ±c. We shall see
below that the slope of characteristics to the taxis is simply the velocity
of propagation of small perturbations in gas (sound velocity). Therefore the
assumption dx/dt
∼
= ±c means that this velocity of gas moving in a pipe is
approximately equal to the sound velocity in quiescent gas. In the general case
it is of course not so, especially for gas ﬂowing with high velocity, for example
in propulsive nozzles or for gas ﬂowing from openings, oriﬁces, nozzles and
so on. In such cases the difference (dx/dt −c) cannot be ignored.
Since the determinant of the system of the six linear equations under
consideration vanishes at ˙ x = v and ˙ x = v ±c, for compatibility of this system
the determinant
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
_
∂ρ
∂p
_
T
v
_
∂ρ
∂p
_
T
0 ρ
_
∂ρ
∂T
_
p
0
v
_
∂ρ
∂p
_
T
1 +v
2
_
∂ρ
∂p
_
T
ρ 2ρv v
_
∂ρ
∂T
_
p
J
2
0 0 0 p ρC
v
J
3
1 ˙ x 0 0 0 J
4
0 0 1 ˙ x 0 J
5
0 0 0 0 1 J
6
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
¸
obtained from the principle determinant of the system by replacement of the
last column on free terms of the system of equations, should also vanish. In
the theory of linear equations the condition of the latter determinant vanishing
is called the compatibility condition. This condition as applied to our case may
be called the compatibility condition at characteristics.
Characteristic form of equations. Omitting cumbersome calculation of this
determinant, we give the ﬁnal result
_
dx
dt
−v
_
2
_
∂ρ
∂p
_
T
_
J
3
ρ
−C
v
J
6
_
−
_
dx
dt
−v
_
p
ρ
2
_
∂ρ
∂p
_
T
(ρJ
5
−J
2
)
−
_
J
3
ρ
−C
p
J
6
+
p
ρ
2
_
∂ρ
∂p
_
T
J
4
_
= 0. (5.35)
Substitution successively in Eq. (5.35) of the values of dx/dt for each of the
characteristic families yields conditions for the functions p, v, T to be obeyed
at these characteristics:
1.
dp
dt
+ρc
dv
dt
= +c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
(5.36)
at dx/dt −v = +c or dx/dt = v +c, where J
2
= −λρv · v/2d
0
−ρg sinα,
J
3
= −4κ(T −T
)/d
0
+λρv
3
/2d
0
and derivatives with respect to time are
taken in the direction of the characteristic dx/dt = v +c.
5.4 NonIsothermal Gas Flow in GasPipelines 143
2.
dp
dt
−ρc
dv
dt
= −c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
(5.37)
at dx/dt −v = −c or dx/dt = v −c, where derivatives with respect to time
are taken in the direction of the characteristic dx/dt = v −c.
3. C
p
dT
dt
−
p
ρ
2
_
∂ρ
∂p
_
T
dp
dt
=
J
3
ρ
(5.38)
at dx/dt −v = 0 or dx/dt = v, where derivatives with respect to time are
taken in the direction of the characteristic dx/dt = v.
Method of Characteristics
Formulas (5.36)–(5.38) give a constructive way to solve the system of
equations (5.41). The method by which this solution is obtained is called
the method of characteristics.
Basic equations for this method are Eqs. (5.36)–(5.38). The idea of this
method is that at each point M(x
k
, t
m
) of the plane (x, t) relevant to the
time t
m
‘‘in the past’’, that is at t < t
m
, there are just three characteristics
dx/dt = v ±c and dx/dt = v, at which should be satisﬁed the compatibility
conditions (5.36)–(5.38). Each of these compatibility conditions represents an
ordinary differential equation which could be integrated over the direction
of the respective characteristic. Thus, at the point M(x
k
, t
m
), being the
intersection point of these characteristics, appear three equations to determine
three unknown quantities p(x
k
, t
m
), v(x
k
, t
m
) and T(x
k
, t
m
). For numerical
realization of the method of characteristics various schemes may be used.
Consider one of them.
Let it be required to get the solution of the system of equations (5.33)
in the region 0 < x < L, t > 0 of the plane (x, t). Divide this region by
a rectangular mesh with straight lines x
ˆ e
= x · (k −1), t
m
= t · (m−1),
where 1 ≤ κ ≤ N +1, N = [L/x]. The time step t is chosen so that
t = x/(v +c)
max
, where v
max
and c
max
are the maximum possible values
of the gas and sound velocities respectively (c
max
=
_
γRT
max
; T
max
is the
maximum possible value of the gas temperature), m = 1, 2, 3, . . ..
Neglecting the work of the gravity force, the basic system of differential
equations (5.33) may be written in the socalled divergent form (see Eq. (1.36)
at α
k
∼
= 1)
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
∂ρ
∂t
+
∂ρv
∂x
= 0
∂ρv
∂t
+
∂
∂x
(p +ρv
2
) = −λρv · v/2d
0
∂
∂t
_
ρ
_
v
2
2
+e
in
__
+
∂
∂x
_
ρv ·
_
v
2
2
+e
in
+
p
ρ
__
= −
4κ
d
0
(T −T
ex
)
(5.39)
The lefthand sides of Eqs. (5.39) are represented by differential operators of
the form
144 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
div{A, B} =
∂A
∂t
+
∂B
∂x
expressing the divergence of a vector with coordinates {A, B} in the space of
variables (x, t).
Let us ﬁrst integrate the system of equations (5.39) over the area of the mesh
cell ABCD (x
k
≤ x ≤ x
k+1
; t
m−1
≤ t ≤ t
m
) with sides x and t (Fig. 5.11)
and transform the integrals over the mesh area into integrals over the mesh
contour using the formula
_ _
ABCD
_
∂A
∂t
+
∂B
∂x
_
dxdt =
_
ABCD
[Acos(nt) +Bcos(nx)]dσ
=
_
ABCD
(Adx +Bdt) = (A
k+1/2,m
−A
k+1/2,m−1
)x +(B
k+1
−B
k
)t.
Here the quantities with fraction subscripts denote mean values of the
corresponding parameters at the horizontal sides (BC and AD) of the mesh
cell, whereas the quantities with integer subscripts denote mean values of
parameters at the vertical sides (AB and CD) of the same mesh cell.
Application of this transformation to each equation of the system (5.39)
yields the following system of ﬁnite difference equations
ρ
k+1/2,m
= ρ
k+1/2,m−1
−[(ρv)
k+1
−(ρv)
k
] ·
t
x
,
ρ
k+1/2,m
v
k+1/2,m
= ρ
k+1/2,m−1
v
k+1/2,m−1
−
_
(p +ρv
2
)
k+1
−(p +ρv
2
)
k
_
·
t
x
−(λρv · v/2d
0
)
k+1/2,m−1
· t,
ρ
k+1/2,m
_
v
2
2
+C
v
T
_
k+1/2,m
= ρ
k+1/2,m
_
v
2
2
+C
v
T
_
k+1/2,m−1
−
__
∂v ·
_
v
2
2
+C
v
T +
ρ
∂
__
k+1
Figure 5.11 Mesh cell in the plane of
variables (x, t).
5.4 NonIsothermal Gas Flow in GasPipelines 145
−
_
∂v ·
_
v
2
2
+C
v
T +
ρ
∂
__
k
_
×
t
x
−
4κ
d
0
(T −T
ex
)
k+1/2,m−1
· t.
(5.40)
Here ρ
k+1/2,m
; v
k+1/2,m
and T
k+1/2,m
are, respectively, the mean values of the
density, velocity and temperature of the gas in segment (x
k
, x
k+1
) of the
gaspipeline at the instant of time t = t
m
; ρ
k+1/2,m−1
, v
k+1/2,m−1
and T
k+1/2,m−1
are the mean values of the same functions at the previous moment of time
t
m−1
= t
m
−t.
The physical meaning of the obtained relations is clear: each of these
equations represents the integral balance of one or other parameter in
the mesh cell. For example, the ﬁrst relation (5.40) reﬂects the fact that
the gas mass ρ
k+1/2,m
· x in the segment (x
k
, x
k+1
) of the pipeline at the
instant of time t
m
is equal to the gas mass ρ
k+1/2,m−1
· x in the same
segment at the previous instant of time added to the mass difference of
the gas [(ρv)
k+1
−(ρv)
k
] · t ﬂowing in time t from the kth segment
into the (k +1)th segment and from the (k −1)th segment into the kth
one. The two other relations are interpreted in the same manner with
the only difference being that they deal with momentum and total energy,
respectively.
Nevertheless, Eqs. (5.40) are not closed since they include unknown
quantities denoted by integer subscripts and representing the transfer of
mass, momentum and energy from one mesh cell into another. The essence
of the method is that the values of these quantities are found from compatible
conditions (5.36)–(5.38) at characteristics.
Let us represent the conditions (5.32)–(5.37) in the form of ﬁnite difference
equations
p
k,m
−p
k−1/2,m−1
t
+(ρc)
k−1/2,m−1
v
k,m
−v
k−1/2,m−1
t
=
_
c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
_
k−1/2,m−1
,
p
k,m
−p
k+1/2,m−1
t
−(ρc)
k+1/2,m−1
v
k,m
−v
k+1/2,m−1
t
(5.41)
=
_
−c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
_
k+1/2,m−1
,
C
p
T
k,m
−T
k−1/2,m−1
t
−
_
p
ρ
2
_
∂ρ
∂p
_
T
_
k−1/2,m−1
·
p
k,m
−p
k−1/2,m−1
t
=
_
J
3
ρ
_
k−1/2,m−1
.
146 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
As a result we have three linear equations for three unknown quantities
p
k,m
, v
k,m
, T
k,m
. First we ﬁnd quantities p
k,m
and v
k,m
from the ﬁrst
two equations, then from the third equation we obtain T
k,m
and from
the relation ρ
k,m
= p
k,m
/RT
k,m
we calculate the gas density. Substituting
ρ
k,m
, p
k,m
, v
k,m
, T
k,m
into Eq. (5.40), we get average values of the gasdynamic
parameters in the considered mesh cell at the instant of time t
m
.
To Eqs. (5.37) and (5.38) should be added initial and boundary conditions.
Boundary conditions vary depending on the concrete problem. As an example,
they could be taken in the form of two relations between the pressure, ﬂow rate
and temperature of the gas at one edge (left) of the pipeline section, reﬂecting
conditions of compressor station operation and one condition at the right edge
of the pipeline section, for example, with a given pressure.
5.5
Gas Outﬂow from a Pipeline in the Case of a Complete Break of the Pipeline
Let us illustrate the use of the method of characteristics as applied to non
stationary processes in a gaspipeline with an example of the calculation of
gas outﬂow from a pipeline in the case of its complete break. The dynamics
of gas outﬂow happen as a rule in two regimes. At ﬁrst at the crosssection,
through which occurs gas outﬂow, appears a critical regime with the velocity
of gas outﬂow equal to the local velocity of sound (≈380–400 m s
−1
). After
the pressure in the gaspipeline is lowered by a certain value (for natural gas
by a factor of 1.8–1.9 greater than atmospheric pressure) the outﬂow regime
becomes subsonic and the gas velocity gradually decreases fromsound velocity
to zero.
The process of gas outﬂow is not isothermal. The gas temperature, owing
to adiabatic expansion and the JouleThomson effect falls signiﬁcantly at the
break in the crosssection as well as far from it. For example, with a complete
break of a gaspipeline the gas temperature can be reduced by 80–100 K. Only
at the ﬁnal stage of the process is there a gradual restoration of temperature
due to external heat inﬂow.
Numerical calculations are carried out on the basis of recurrent formu
las (5.38). The lefthand end of the pipeline (x = 0) is closed, whereas
the righthand end (x = L) is suddenly opened and remains open to the
atmosphere.
Hence, the boundary conditions are as follows.
1. Crosssection x = 0. The ﬁrst boundary condition is v
1,m
(0, t
m
) = 0, since
the lefthand end of the pipeline is taken to be closed. Hence it follows
that at v = 0 the left integration boundary coincides with one of the
characteristics of the differential equations (5.36), therefore the second
boundary condition at the crosssection x = 0 (k = 1) is formulated as a
condition at the characteristic dx/dt = v = 0
5.5 Gas Outﬂow from a Pipeline in the Case of a Complete Break of the Pipeline 147
C
p
T
1,m
−T
1,m−1
t
−
_
p
ρ
2
_
∂ρ
∂p
_
T
_
1,m−1
·
p
1,m
−p
1,m−1
t
=
_
J
3
ρ
_
1,m−1
.
The compatibility condition at the characteristic dx/dt = v −c coming
from the integration domain to the initial pipeline crosssection (in our
case at the characteristic dx/dt = 0 −c = −c), provides the third boundary
condition
p
1,m
−p
3/2,m−1
t
−(ρc)
3/2,m−1
v
1,m
−v
3/2,m−1
t
=
_
−c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
_
3/2,m−1
.
2. Crosssection x = L. Two characteristics of the system (5.39) come to this
crosssection from the integration domain. These characteristics are
dx/dt = v +c and dx/dt = v with positive slope. Thus, the ﬁrst two
boundary conditions at x = L (k = N +1) are
p
N+1,m
−p
N+1/2,m−1
t
+(ρc)
N+1/2,m−1
v
N+1,m
−v
N+1/2,m−1
t
=
_
c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
_
N+1/2,m−1
,
C
p
T
N+1,m
−T
N+1/2,m−1
t
−
_
p
ρ
2
_
∂ρ
∂p
_
T
_
N+1/2,m−1
·
p
N+1,m
−p
N+1/2,m−1
t
=
_
J
3
ρ
_
N+1/2,m−1
.
The third boundary condition at x = L has a different form depending on
whether the gas outﬂow is subsonic v < c(p, T) or sonic v = c(p, T). If the
gas outﬂow happens with local sound velocity, then dx/dt = v −c = 0 and
the right boundary of the integration domain coincides with the characteri
stic of the system (5.39). Hence, the third boundary condition is nothing
but a condition at this characteristic
p
N+1,m
−p
N+1,m−1
t
−(ρc)
N+1,m−1
v
N+1,m
−v
N+1,m−1
t
=
_
−c · J
2
+
(γ −1)
p/ρ · (∂ρ/∂p)
T
· J
3
_
N+1,m−1
.
In subsonic ﬂow v < c(p, T) only two characteristics of the system (5.39) come
from the integration domain to the boundary points x = L, therefore one more
condition should be given. Such a condition is p
N+1,m
= p
aTM
meaning that
the pressure at the opened pipeline end is equal to the external atmospheric
pressure.
Calculations show that the gas is signiﬁcantly cooled during outﬂow.
Figure 5.12 shows graphs of the gas temperature distribution over a pipeline
148 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
L.km
Figure 5.12 Distribution of gas temperature over a pipeline section length: 1, t = 0; 2,
t = 5 s; 3, t = 60 s; 4, t = 120 s; 5, t = 200 s; 6, t = 420 s.
section length (D = 1220 mm, L = 5 km) at different instants of time. It is
seen that the initial temperature of the gas, equal to 0
◦
C decreases by more
than 100 K. This is explained by gas expansion due to the high outﬂow velocity
and the Joule–Thomson effect. Gas cooling in the pipeline happens because
the abovementioned effects are too fast to be compensated by heat inﬂow
from the surrounding medium.
Computer modeling shows the effect of gas suction into the pipeline in the
ﬁnal stage of the outﬂow process. Figure 5.13 represents the dependence of
the gas velocity on time at the crosssection where the pipeline undergoes a
break (D = 1220 mm, L = 1 km) with initial pressure p(x, 0) = 5.5 MPa. The
oscillation process is seen to appear in the ﬁnal stage of gas outﬂow and lasts
for about 2 min.
The section AB of the graph characterizes sonic outﬂow of gas lasting about
20 s. The velocity of the outﬂow gradually decreases as a consequence of
reduction in the pressure and temperature of the gas at the crosssection of
the break. The section BF of the graph characterizes subsonic outﬂow of gas.
About after a 30th of a second from the outﬂow beginning in the pipeline an
oscillation process occurs in which the gas periodically changes its direction
of motion. Once the velocity of the gas becomes equal to zero it continues to
decrease and at the section CD of the graph remains negative, testifying that
atmospheric air is being sucking into the pipeline. The same behavior of gas is
observed in the following instants of time. Calculations show that the maximal
velocity of air suction in the oscillation process exceeds 50 m s
−1
.
Since the mixture of natural gas and air could achieve an explosive
concentration, the discovered phenomena appears to be a serious hazard
to attendants. In particular, it is strongly recommended to take the utmost care
in repairreconditioning operations.
5.6 Mathematical Model of NonStationary Gravity Fluid Flow 149
n
L
, ms
1
t.s
Figure 5.13 Dependence of gas outﬂow velocity on time.
5.6
Mathematical Model of NonStationary Gravity Fluid Flow
Such a ﬂow has already been discussed in Section 3.7 but then a stationary
gravity ﬂow was considered, that is a ﬂow in which all the hydrodynamic
parameters remain constant at each crosssection of the pipeline. In other
words these parameters were independent of time. Consider now a one
dimensional mathematical model of nonstationary gravity ﬂuid ﬂow. In this
model there are two parameters governing the ﬂow: S(x, t) the area of the
pipeline crosssection and v(x, t) the ﬂow velocity (Fig. 5.14).
The differential equations of the model under consideration are
_
¸
¸
_
¸
¸
_
∂ρS
∂t
+
∂ρvS
∂x
= 0
∂ρvS
∂t
+
∂ρv
2
S
∂x
= −ρgScos α ·
∂h
∂x
−ρgSsinα −
ρgScos α
C
Sh
2
R
h
· vv.
Figure 5.14 Gravity ﬂuid ﬂow.
150 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
There R
h
(S) is the hydraulic radius of the ﬂow (see Section 3.7); h(S) is
the depth of the pipeline crosssection ﬁlling with ﬂuid; α is the angle
of inclination of the pipeline axis to the horizontal; C
Sh
is the Chezy
factor.
The ﬁrst of these equations (continuity equation) expresses the law of
mass conservation in the ﬂuid ﬂow in a pipeline with crosssection partially
ﬁlled by ﬂuid. The second equation (momentum equation) is the law of
momentum change, that is Newton’s Second law: on the left is the derivative
of momentum with respect to time, the quantity ρv
2
S = ρvS · v representing
the ﬂux of momentum; the term ρgScos α ·
∂h
∂x
on the right is the Boussinesq
force acting on the ﬂuid due to its free surface being nonparallel to the pipeline
axis, i.e. due to excess of the ﬂuid level at one crosssection of the pipeline as
compared to the ﬂuid level at another crosssection; the term−ρgSsinα is the
component of the gravity force and the term −ρgScos α · vv/(C
Sh
2
R
h
) is the
force of the ﬂow resistance due to ﬂuid friction against the pipeline walls.
The system of equations may be rewritten in an equivalent form if we take
into account that
ρg cos α ·
_
S
0
Sdh = ρg cos α ·
_
S
0
S
dh(S)
dS
dS
As a result we get
_
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
_
∂ρS
∂t
+
∂ρvS
∂x
= 0
∂ρvS
∂t
+
∂
∂x
_
ρv
2
S +ρg cos α ·
_
S
0
Sdh
_
= −ρgSsinα −
ρgScos α
C
Sh
2
R
h
· vv
(5.42)
The system of equations (5.42) belongs to the class of quasilinear (that is linear
with respect to derivatives) differential equations of the hyperbolic type. Solution of
these equations can be obtained by specially elaborated methods, which are
beyond the scope of this book. We advise those who are interested in these
methods to consult the book by Rozhdestvenski and Yanenko (1977), which
among other things contains voluminous literature devoted to this problem.
For gravity stationary ﬂuid ﬂow (∂/∂t = 0) from the ﬁrst equation of
the system (5.42) it follows that ρvS = const. If we take into account that
ρ
∼
= ρ
0
= const., then v · S = const. If we assume in addition that S = const.,
which is true everywhere except in small regions close to the transfer sections
where gravity ﬂows are formed, from Eq. (5.42) it follows that
0 = −ρgS · sinα −
ρgScos α
C
Sh
2
R
h
· vv ⇒v = C
Sh
_
R
h
· tanα (5.43)
5.6 Mathematical Model of NonStationary Gravity Fluid Flow 151
p
sat
Figure 5.15 Gravity section in pipeline.
which after some transformation allows one to get the degree of ﬁlling with
ﬂuid of the gravity ﬂow section, σ, depending on the hydraulic gradient tan α
(see Eq. (3.51)).
The point at which the ﬁrst gravity section in the pipeline begins is called
the transfer section. Since crosssections after the transfer section are only
partially ﬁlled by ﬂuid, the pressure in this section is constant and equal to the
saturated vapor pressure p
sat
of this ﬂuid. Figure 5.15 illustrates the behavior
of the hydraulic gradient in gravity ﬂow. It is seen from this ﬁgure that in this
section the line of the hydraulic gradient goes parallel to the pipeline axis at a
distance p
sat
/ρg from it owing to the pressure constancy in the gas cavity over
the ﬂuidfree surface.
An ordinary differential equation serves to calculate the depth h(x) of ﬂuid
in the pipe.
dh
dx
= −
tanα +Q
2
/(C
Sh
2
R
h
S
2
)
1 −(Q
2
dh/dS)/(g cos α · S
3
)
(5.44)
following from Eq. (5.42) for the case of stationary ﬂow (∂/∂t = 0) with
h(x) = h[S(x)]; R
h
= R
h
(S); Q is the ﬂuid ﬂow rate.
The depth h
n
of the ﬂuid in the pipe at which the numerator of the fraction
on the righthand side of Eq. (5.44) vanishes is called the normal depth of
gravity ﬂow in the pipe. Fluid ﬂow with a normal depth of ﬂow happens under
the condition h
n
= const. In this case the relation (5.43) holds.
The depth h
cr
, at which the nominator of the fraction on the righthand side
of Eq. (5.44) vanishes is called the critical depth. In crosssections with such a
depth the derivative dh/dx tends to ∞ and the ﬂuid ﬂow varies the level of
pipe ﬁlling abruptly. Depending on the relation between the depths h
n
and
h
cr
different ﬂow regimes are possible. Investigations of these regimes are
described in special monographs (Archangelskiy, 1947; Leibenson et al., 1934;
Christianovitch, 1938).
152 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
5.7
NonStationary Fluid Flow with Flow Discontinuities in a Pipeline
The preceding classical models of nonstationary ﬂows of a slightly
compressible ﬂuid (see Sections 5.1–5.3) contained an essential restriction
on the absence of phase transition in the ﬂuid. It was tacitly supposed that the
ﬂuid under no circumstances passes into the vaporgas phase, even when the
pressure drops to the value of the saturated vapor pressure. However, in
the propagation of a rarefaction wave in the pipeline this condition can be
violated at many pipeline crosssections and ﬁrst and foremost at the tops
of the pipeline proﬁle. When the pressure in the rarefaction wave reduces to
the saturated vapor pressure the ﬂuid boils, the vapor column breaks and the
pipeline crosssection becomes partially ﬁlled with vapor. From this point on
all further results predicted by classical theory appear to be wrong.
For example, at the disconnection of a pumping station or an aggregate
of this station a rarefaction wave is propagated downstream in the pipeline.
The pressure in such a wave reduces leading to the formation of voids at the
tops of the pipeline proﬁle. These voids are capable of growing and turning
into stationary gravity ﬂow sections or, on the contrary, contracting and even
disappearing altogether. To perform calculation of such processes on the base
of classical theory is of course impossible.
We can give another example: closing the gate valve generates a compression
wave propagating upstream with a rise in the pressure in the wave. This wave
having beenreﬂected fromthe opensurface of the reservoir or froma vaporgas
cavity inside the pipeline initiates a rarefaction wave traveling in the opposite
direction and reduces the pressure in the ﬂuid. This brings into existence
temporal transfer points at some tops of the pipe proﬁle and cavities. This
leads sometimes to ﬂuid ﬂows with a partially ﬁlled pipeline crosssection. If
the pressure supply in the pipeline is not high, the reduction in pressure leads
to ﬂow discontinuity and the generation of vaporgas cavities. So, for example,
in laboratory installations it could be observed that the ﬂuid before the gate
valve literally boiled owing to the sharp reduction in pressure. Calculation in
these cases on the basis of classical theory is also impossible.
One more example can be given: Connection of a lateral tap from the
oilpipeline to an intermediate oil tank leads to the propagation of rarefaction
waves up and down stream from the place of tap cutting up and downstream.
These waves are able to break the ﬂuid column at many crosssections of the
pipeline proﬁle and turn the enforced ﬂow into a gravity one characterized by
the presence of vaporgas cavities and gravity ﬂow sections. Such cases also
defy calculation in the framework of classical theory.
All the aforesaid is true also for pipelines transporting the socalled
unstable ﬂuids, among which are gascondensate and a wide fraction of
light hydrocarbons with saturated vapor pressure from 3 to 30 atm. Any
sharp pressure reduction gives rise to a plethora of vaporgas cavities the
disappearance of which in the pipe lead to powerful hydraulic shocks.
5.7 NonStationary Fluid Flow with Flow Discontinuities in a Pipeline 153
Proﬁle Hydraulic Shock
It should be noted that the appearance or disappearance of voids in the
pipeline is unsafe and is rather dangerous for pipeline integrity. In particular,
in Lurie and Polyanskaya (2000), the discovery and investigation of the origin
of powerful hydraulic waves in the pipeline called proﬁle hydraulic shock are
described. For example, each time the gate valve, located before a section
of pipeline with signiﬁcant slope, was closed the rarefaction resulting in the
region close to the valve led to a sequence of powerful hydraulic shocks. These
shocks were gradually damped with time and the ﬂow in the pipeline was
stabilized. Similar phenomena were observed in the pipeline on disconnection
of some pumps or the pumping station as a whole.
The nature of the proﬁle hydraulic shock is as follows. Whenthe gate valve (at
x = 0, Fig. 5.16) closes the pressure in the sloping section of pipeline falls and
the ﬂuid column, initially supported by this pressure, begins little by little to slip
down and acquire inverse ﬂow (Figs. 5.17–5.19). The ﬂuid in this inverse ﬂow
is accelerated and whenvaporgas voids, having originated before the gate valve,
are taken out, there is an abrupt stop of ﬂuid ﬂow and, as a consequence, hy
draulic shock (Fig. 5.18). The power of the shock is then especially great, since
Figure 5.16 Initial stage of the process.
Figure 5.17 Formation of gravity ﬂow in the upward sloping pipeline section.
154 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Figure 5.18 Formation of reverse ﬂuid ﬂow in the upward sloping pipeline section.
Figure 5.19 Proﬁle hydraulic shock.
the inverse ﬂow of the ﬂuid runs against the closed gate valve and, as follows
from the Joukovski formulas, the amplitude of the hydraulic shock redoubles.
The hydraulic shock wave reﬂected from the gate valve and accompanied
by ﬂuid stop, propagates downstream. Reaching a temporary transfer section,
at which a vaporgas void has been formed, the wave is reﬂected from this
void and now, in the form of a rarefaction wave, heads back to the gate valve,
the ﬂuid column in the ascending section of the pipeline again becomes
weightless and begins to slip down to the closed gate valve. Then a secondary
hydraulic shock occurs, after which the process is repeated again and again
with decreasing intensity.
Generalized theory, (Lurie and Polyanskaya, 2000). In accordance with the
classical theory of nonstationary processes the wave processes generating in a
completely ﬁlled pipeline at its startup or stopping, opening or closing of the
gate valve or lateral tap and so on, are described by differential equations (5.4)
for pressure p(x, t) and velocity v(x, t)
_
¸
¸
_
¸
¸
_
∂p
∂t
+ρ
0
c
2
·
∂v
∂x
= 0
ρ
0
∂v
∂t
+
∂p
∂x
= −λ(Re, ε)
1
d
0
ρ
0
vv
2
−ρ
0
g sinα(x)
5.7 NonStationary Fluid Flow with Flow Discontinuities in a Pipeline 155
These equations represent the laws of mass conservation and variation of
momentum of ﬂuid particles moving in a pipeline. In the generalized theory it
is assumed that in the pipeline there are completely ﬁlled enforced (pumped)
sections as well as only partially ﬁlled gravity sections in which the pressure
is equal to the saturated vapor pressure p
sat
. The ﬂuid ﬂow in the sections
of enforced ﬂow is described by the system of equations (5.4) whereas in the
sections of gravity ﬂow it is described by Eqs. (5.42).
Method of calculation. For numerical calculation of oil product ﬂow in com
pletely ﬁlled as well as in partially ﬁlled pipeline sections there is an elaborate
scheme of endtoend calculation based on the ideas of Godunov (Samarskiy,
1977). This scheme involves consideration of the socalled problems on the
disintegration of arbitrary discontinuity in the system of hyperbolic equations.
Results of calculation. Figures 5.16–5.23 demonstrate the results of calcula
tions on successive stages of unloading wave propagation in a 10km pipeline
with internal diameter d = 516 mm at an abrupt drop in the pumping delivery
of benzene (ρ = 750 kg m
−3
, p
sat
= 0.7 atm) from 1500 to 200 m
3
h
−1
. It is
seen that, as distinct from existing theory, the line of hydraulic gradient at
Figure 5.20 Decay of hydraulic shock wave.
Figure 5.21 Formation of the secondary reverse ﬂow in the upward sloping pipeline
section.
156 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline
Figure 5.22 The second top of the proﬁle hydraulic shock.
Figure 5.23 Formation of new stationary ﬂow.
no instant of time intersects the pipeline proﬁle, that is, the pressure in the
pipeline would never be less than the ﬂuid saturated vapor pressure.
The ﬁgures also show how the wave of pressure decrease runs over the
pipeline proﬁle top (lower polygonal line of Fig. 5.16), and how, at this top, a
temporary transfer point and further gravity ﬂow section with inverse ﬂow of
ﬂuid is being formed (Fig. 5.17 and Fig. 5.18).
Since the ﬂuid ﬂows in the opposite direction (as if it would run on a closed
gate valve), in 24 s after the aggregate disconnection a powerful hydraulic
shock is formed in the pipe (Fig. 5.19), in which the pressure is nearly twice
the initial pressure at the station. The amplitude of the hydraulic shock wave
gradually decreases (Fig. 5.20) and in a further 10 s in the upward sloping
section of the pipeline slope the inverse ﬂow appears again (Fig. 5.21). After
46 s the secondary hydraulic shock occurs (Fig. 5.22), but already with lesser
amplitude. In the given calculation there were six such hydraulic shocks. Only
4 min after the pumping regime change the stationary regime in the pipeline
is achieved (Fig. 5.23), where there exists a single gravity ﬂow section of length
500 000 m in the pipeline downward sloping section.
157
6
Dimensional Theory
Dimensional theory contains fundamental propositions for representing
equations of mathematical regularities modeling different phenomena in
invariant form, that is, in a form independent of the choice of the units of the
measurements. Such representations permit one to compile classes of similar
phenomena and to model them on experimental installations.
When setting out the fundamentals of dimensional and similarity theory
and the modeling of phenomena we have followed the methodology of Sedov
(1965).
6.1
Dimensional and Dimensionless Quantities
A quantitative description of various physical phenomena, including the
transport of ﬂuids and gases in pipelines, is connected with measurements of
the characteristics of these phenomena, whose numerical values depend on the
choice of measurement units. For example, pipeline diameter can be expressed
through the numbers 1; 10; 100; 1000; 3.28; 39.4 and so on depending on that
what unit of measurement is taken: meter, decimeter, centimeter, millimeter,
foot or inch (1 foot
∼
= 0.3048 m; 1 inch
∼
= 0.0254 m); the length of a pipeline
can be expressed through the numbers 150 000; 150 or 93.21 and so on,
depending on what is used as the measurement unit of the length: meter,
kilometer or mile (1 mile
∼
= 1.6093 km). The same is true for many other
physical quantities. For example, the volumetric ﬂuid ﬂow rate in a pipeline
may be measured by the numbers 1000, 277.8 or 73.4 depending on whether
the units of volume and time are taken as cubic meter and hour; liter and
second; gallon (1 gallon in USA
∼
= 3.785 liter) and second. The pressure can
also be measured with different numbers 64, 6.28 or 0.0433 depending on the
measurement units: technical atmosphere (kgf s
−1
m
−2
), millions of pascals
(megapascals, MPa) or psi (pound per square inch: 1 force pound
∼
= 4.448 N;
1 mass pound
∼
= 453.6 g, 1 psi
∼
= 0.006896 MPa
∼
= 0.0703 atm).
It is important to note that the choice of measurement units depends on the
researcher and therefore is, to a great extent, arbitrary. Two different researchers
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
158 6 Dimensional Theory
describing one and the same phenomenon but at the same time using different
measurement units can obtain diverse numerical values of one and the same
parameter.
And yet there exist parameters whose numerical values do not depend on the
researcher, that is, do not depend on measurement units. These parameters
are said to be invariant relative to the choice of measurement units. For
example, the ratio of the pipeline length to the pipeline diameter L/d, the
ratio of the pressure at a pumping station discharge line to the pressure at a
pumping station suction line p
dl
/p
sl
or a more complex combination such as
the Reynolds number Re = vd/ν (v is the average ﬂuid velocity in the pipeline,
d is the internal diameter of the pipeline, ν is the ﬂuid kinematic viscosity)
are independent of the choice of measurement units. This means that the
numerical values of L/d, p
dl
/p
sl
and v · d/ν would be unchanged for any choice
of length, pressure, velocity and viscosity units.
Hence, it is possible to give the following deﬁnition: quantities whose
numerical values depend on the choice of measurement units are called dimensional
quantities; quantities whose numerical values do not depend on the choice of
measurement units are called dimensionless quantities.
6.2
Primary (Basic) and Secondary (Derived) Measurement Units
Measurement units, having been introduced empirically by arbitrary condi
tions and propositions, are called primary or basic units. Among these are, in
particular, the units of length, time and mass. In the international SI system
they are deﬁned as follows.
Meter is a unit of length measurement. In accordance with the deﬁnition
taken at the 11th General Conference on Weights and Measures (1960),
1 meter is a length equal to 1650763.73 lengths of the wave emitted by the
krypton (Kr) atom in vacuum at its energylevel transition. The international
standard of the meter before 1960 was a bar of platinum–iridium alloy marked
on one side of its planes. This bar is kept on deposit at the International
Bureau of Weights and Measures in Sevre near Paris. At ﬁrst the meter was
deﬁned as 10
−7
part of a quarter of the Earth’s meridian.
Second is the measurement unit of time. There are recognized the atomic
(standard) second reproduced by cesium standards of frequency and time and
the ephemeris second equal to 1/31556925.9747 part of the tropical year.
Kilogram is a unit of mass measurement. The standard kilogram is equal
to the mass of the international prototype kept at the International Bureau
of Weights and Measures. The prototype of the kilogram is made from
platinum–iridium alloy in the form of a cylindrical weight. The relative error
of standard copies compared with the original does not exceed 2 · 10
−9
.
There are also other basic units of measurement, such as coulomb (C) – a
unit of electricity quantity (electrical charge), volt (V) – a unit of electrical stress
6.3 Dimensionality of Quantities. Dimensional Formula 159
(voltage, potential); degrees (
◦
C,
◦
F, K and other) – a unit of temperature and
so on.
Measurement units of other quantities are by deﬁnition introduced through
the basic measurement units. Such units are called secondary or derived units.
Among them are the following.
Velocity which is deﬁned as the ratio of length to time, therefore velocity
units can be m s
−1
, km h
−1
, mile h
−1
and so on.
Acceleration which is deﬁned as the ratio between the velocity increment and
time, hence acceleration units can be m s
−2
, foot s
−2
, km s
−2
, mile h
−2
and
so on.
Force (weight) which is deﬁned as the product of mass and acceleration,
thus force units can be dyne (1 dyn = 1 g cm s
−2
); newton (1 N = 1 kg m s
−2
);
pound (
∼
=4.448 kg m s
−2
) and so on.
Density which is deﬁned as the mass of the medium unit volume, therefore
its units can be kg m
−3
, g cm
−3
, t m
−3
and so on.
Speciﬁc weight which is deﬁned as the weight of the medium unit volume,
hence its measurement units can be N m
−3
, dyn cm
−3
and so on.
Pressure which is deﬁned as the ratio of force to unit area, thus force units
can be: pascal (1 Pa = 1 N m
−2
= 1 kg m
−1
s
−2
), pound inch
−2
and so on.
Fluid ﬂow rate (e.g. volumetric) which is deﬁned as the ﬂuid volume crossing
the pipeline crosssection area in a unit time, therefore its units can be m
3
s
−1
,
m
3
h
−1
, m
3
min
−1
, l s
−1
and so on.
Current intensity which is deﬁned as a charge transmitted in a unit time,
hence the unit of measurement can be the ampere (1 A = 1 C s
−1
).
All these and analogous measurement units are derived fromthe basic units.
6.3
Dimensionality of Quantities. Dimensional Formula
Let there be a physical parameter A. Expression of its measurement unit
through basic units is called the dimension of a given parameter and is denoted
by the symbol [A]. This expression written as a formula is called the dimensional
formula.
Denote through L the measurement unit of length, through T the
measurement unit of time and through M the measurement unit of mass.
Then the expression of the measurement units of many other quantities can
be written as the following formulas:
(A = v) – velocity: [v] =
L
T
= L · T
−1
= M
0
· L · T
−1
;
(A = a) – acceleration: [a] =
L
T
2
= L · T
−2
= M
0
· L · T
−2
;
(A = F) – force: [F] = M· [a] = M· L · T
−2
;
(A = ρ) – density: [ρ] =
M
L
3
= M· L
−3
· T
0
;
160 6 Dimensional Theory
(A = γ) – speciﬁc weight [γ] =
M· L · T
−2
L
3
= M· L
−2
· T
−2
;
(A = p) – pressure: [p] = [F]/L
2
=
M· L · T
−2
L
2
= M· L
−1
· T
−2
;
(A = Q) – volumetric ﬂow rate: [Q] =
L
3
T
= M
0
· L
3
· T
−1
;
(A = G) – mass ﬂow rate: [G] =
M
T
= M
1
· L
0
· T
−1
;
(A = B) – dimensionless parameter [B] = M
0
· L
0
· T
0
.
Thus, particular examples show that in all cases the dimensional formula of
parameter A has form of a power monomial
[A] = M
m
1
· L
m
2
· T
m
3
(6.1)
where m
1
, m
2
, m
3
are certain real positive or negative numbers.
What does the dimensional formula mean? It allows one to determine very
simply by how manyfold the numerical value of parameter A would be changed, on
going from one system of basic measurement units to another one differing from the
ﬁrst system only by the scales of the basic units. For example, if the transition from
the new system of basic measurement units to the old one is accomplished by
variation of the mass unit k
1
fold, of the length unit k
2
fold, of the time unit
k
3
fold, the numerical value of the parameter A would vary k
m
1
1
· k
m
2
2
· k
m
3
3
fold,
that is the new value
`
A
of this parameter would be determined by the formula
`
A
= k
m
1
1
· k
m
2
2
· k
m
3
3
·
`
A. (6.2)
Let us explain the afore said by examples.
Example 1. The velocity v in the SI system of basic measurement units (m, s,
kg) is 1 m s
−1
. For a new system of basic measurement units (cm, s, kg) the
transition from this system to the SI system is performed by increasing one of
the basic units (length) by a factor of 100, the value of the velocity will also be
enhanced 100 times
v
= 1
0
· 100
1
· 1
−1
= 100 · v = 100 cm s
−1
.
Example 2. The pressure p in the SI system of basic units (m, s, kg)
is 6 MPa, that is 6 000 000 kg m
−1
s
−2
. For a new system of units (inch,
s, pound) the transition from this system to the SI system is carried
out by changing the length scales 100/2.54 = 39.37 times and the mass
by 1/0.4536 = 2.205 times, hence the value of the pressure will vary
by a factor 2.205
1
· 39.37
−1
· 1
−2
= 0.056 and will be 0.056 · 6 000 000 =
336 043 (psi).
In Eq. (6.1), obtained empirically, there are only three basic units of
measurement because we have considered examples solely from mechanics.
However, it is easy to verify that even if the number of physical parameters
6.4 Proof of Dimensional Formula 161
is increased by other parameters from electrical, heat, chemical and
other phenomena and the number of basic units grows and exceeds
three, the dimensional formula would not be radically changed. Only the
number of factors will increase. Hence, it could be assumed, that the
dimensional formula in the general case would have the form of a power
monomial
[A] = [a
1
]
m
1
.
[a
2
]
m
2
.
[a
3
]
m
3
.
. . . .
.
[a
n
]
m
n
(6.3)
where A is a physical parameter, whose dimension is derived from the
dimensions of basic quantities denoted by a
1
, a
2
, . . . ., a
n
.
6.4
Proof of Dimensional Formula
Let us prove the validity of the dimensional formula (6.3) in the general case.
This formula means that if we change the scale of a basic measurement unit
a
1
by k
1
times, of a
2
by k
2
times, . . . ., of a
n
by k
n
times, the numerical value
of the parameter A would be changed by k
m
1
1
· k
m
2
2
· . . . · k
m
n
n
times, that is the
new value of this parameter A
would be equal to
A
= k
m
1
1
· k
m
2
2
· . . . · k
m
n
n
· A.
Let us use this circumstance.
Let there be three researchers B, C and D measuring one and the same
physical parameter A but using different systems of basic measurement
units, namely
B: {a
1
, a
2
, . . . , a
n
};
C:
_
a
1
, a
2
, . . . , a
n
_
, so that its basic units are related to the basic units of
researcher B by the formulas
a
1
= α
1
· a
1
a
2
= α
2
· a
2
. . . . . . . . . . . .
a
n
= α
n
· a
n
where k
1
= α
1
, k
2
= α
2
, . . . , k
n
= α
n
;
D:
_
a
1
, a
2
, . . . , a
n
_
, so that its basic units are related to the basic units of
researcher C by the formulas
a
1
= β
1
· a
1
a
2
= β
2
· a
2
. . . . . . . . . . . .
a
n
= β
n
· a
n
162 6 Dimensional Theory
where k
1
= β
1
, k
2
= β
2
, . . . , k
n
= β
n
, and to the basic units of researcher B by
the formulas
a
1
= β
1
α
1
· a
1
a
2
= β
2
α
2
· a
2
. . . . . . . . . . . .
a
n
= β
n
α
n
· a
n
where k
1
= β
1
α
1
, k
2
= β
2
α
2
, . . . , k
n
= β
n
α
n
.
It is reasonable to suppose that the values of the parameter A measured by
the three researchers would be different: A measured by B,
`
A
measured by C
and
`
A
measured by D.
Let a function F(k
1
, k
2
, . . . , k
n
) show how manyfold the numerical value
of parameter A in one of the system of basic units would be changed when
passing to another system differing by k
1
, k
2
, . . . , k
n
times from the ﬁrst one
in the scales of the basic units, respectively. Then if we go from the units of
researcher B to the units of researcher C, we obtain
A
= F(α
1
, α
2
, . . . , α
n
) · A.
If we then go from the units of researcher C to the units of researcher D,
we get
`
A
= F(β
1
, β
2
, . . . , β
n
) · A
or
`
A
= F(β
1
, β
2
, . . . , β
n
) · F(α
1
, α
2
, . . . , α
n
) · A. (6.4)
On the other hand, if we go at once from the units of researcher B to the units
of the researcher D, bypassing researcher C, we get
`
A
= F(α
1
β
1
, α
2
β
2
, . . . , α
n
β
n
) · A (6.5)
The result should of course be independent of the transition route from
one researcher to another one, thus it must identically satisfy the following
functional equation
F(α
1
β
1
, α
2
β
2
, . . . , α
n
β
n
) = F(α
1
, α
2
, . . . , α
n
) · F(β
1
, β
2
, . . . , β
n
) (6.6)
which ought to be valid for any values of the factors α
1
, α
2
, . . . , α
n
and
β
1
, β
2
, . . . , β
n
.
Find the solution of this equation.
Differentiation of both parts of the identity (6.6) with respect to β
i
, where i
may be equal to 1, 2, 3, . . . , n, yields
α
i
∂F
∂ξ
i
=
∂F
∂β
i
· F(α
1
, α
2
, . . . , α
n
), where ξ
i
= α
i
β
i
.
6.5 Central Theorem of Dimensional Theory 163
Since this equality also represents an identity, then β
1
= 1, β
2
= 1, . . . , β
n
= 1.
As a result the following differential equation is obtained
α
i
·
∂F(α
1
, α
2
, . . . , α
n
)
∂α
i
= m
i
· F(α
1
, α
2
, . . . , α
n
) (6.7)
in which m
i
= ∂F/∂β
i
at β
1
= 1, β
2
= 1, . . . , β
n
= 1.
The solution of differential equation (6.7) gives the dependence of the sought
function F on parameter α
i
. Really, the separation of variables provides
∂
∂α
i
(lnF) =
m
i
α
i
⇒lnF = m
i
· lnα
i
+c ⇒F = K · α
m
1
i
where the integration constant K (K = e
C
) is a function of the remaining
parameters α
j
. Since α
i
was any one of the arguments of the function F, the
latter should have the following form
F = K
0
· α
m
1
1
· α
m
2
2
· . . . · α
m
n
n
where K
0
= const. If F(1, 1, . . . , 1) = 1, because the value of the parameter A
is not changed by variation of the basic measurement unit scales, K
0
= 1 and
the function will be
F = α
m
1
1
· α
m
2
2
· . . . · α
m
n
n
or redesignating the variables as k
1
, k
2
, . . . , k
n
, one gets
F = k
m
1
1
· k
m
2
2
· . . . ·, k
m
n
n
. (6.8)
Hence, when changing the basic unity a
1
by k
1
times, the numerical value of
the parameter A would vary by k
1
m
1
times, when changing the basic unity a
2
by k
2
times the numerical value of the parameter A would vary by k
2
m
2
and so
on. Thus, the parameter A has the dimension
[A] = [a
1
]
m
1
.
[a
2
]
m
2
.
[a
3
]
m
3
.
. . . .
.
[a
n
]
m
n
which proves the dimensional formula (6.3).
6.5
Central Theorem of Dimensional Theory
It is appropriate now to interrupt the description of the theory and to formulate
a question concerning an apparent contradiction due to the use of dimensional
quantities.
Let the mathematical formof a certainphysical phenomenonbe expressed by
the dependence of a parameter A on other parameters a
1
, a
2
, . . . , a
n
governing
this phenomenon as follows
A = f (a
1
, a
2
, . . . , a
n
). (6.9)
164 6 Dimensional Theory
For example, the enhancement of pipeline crosssection area S when
producing a positive pressure P depends on the magnitude of this pressure as
well as on the pipeline diameter D, the wall thickness δ and the elastic modulus
(Young’s modulus) E of the steel from which the pipeline is made. Hence we
have A = S, a
1
= P, a
2
= D, a
3
= δ, a
4
= E, so that S = f (P, D, δ, E).
It is evident that the dependence under consideration exists objectively and
should not depend either on the researcher performing the investigation or
on the choice of measurement units being used to calculate the values of the
function, i.e. parameter A, and its arguments (a
1
, a
2
, . . . , a
n
).
On the other hand all quantities entering into the dependence (6.9) are
dimensional quantities whose numerical values depend on the choice of
measurement units and consequently on the researcher.
The question arises as to at which point the dependence (6.9) reﬂects the
objectively existing physical regularity when numerical values of the function and its
arguments depend on the researcher.
The answer to this question gives the central theorem of dimensional theory
called the Buckingham theorem. This theorem states: ‘‘Each dependence
between dimensional quantities reﬂecting an objectively existing physical regularity
could be rewritten in invariant form independent of the choice of measurement units,
namely in the form of a dependence between dimensional complexes composed from
governing parameters’’.
Now we will prove this theorem, removing the contradiction between the
objective character of any physical regularity and the subjective character
of the choice of measurement units. However, before we do this, we
need ﬁrst to deﬁne dimensionallydependent and dimensionallyindependent
quantities.
6.6
DimensionallyDependent and DimensionallyIndependent Quantities
It is said that a quantity ‘‘a’’ is dimensionallydependent on the quantities
a
1
, a
2
, . . . , a
n
when its dimension [a] is expressed through the dimensions
[a
1
], [a
2
], . . . , [a
n
] by the formula
[a] = [a
1
]
m
1
[a
2
]
m
2
. . . [a
n
]
m
n
(6.10)
namely there exist such numbers m
1
, m
2
, . . . , m
n
that the equality (6.10)
is obeyed. If such numbers do not exist, it is said that the quantity a is
dimensionallydependent on the quantities a
1
, a
2
, . . . , a
n
.
It is evident that parameters with dimensions of time t, length l and mass
m are dimensionallyindependent of each other. Parameters with dimensions
of velocity v and density ρ are dimensionallyindependent. On the other hand,
a parameter with the dimension of pressure p is dimensionallydependent on
parameters with dimensions of density ρ and velocity v. It is easy to verify that
6.6 DimensionallyDependent and DimensionallyIndependent Quantities 165
[p] = [ρ][v]
2
[p] = M
1
L
−1
T
−2
; [ρ] = M
1
L
−3
T
0
; [v
2
] = M
0
L
2
T
−2
⇒
M
1
L
−1
T
−2
= (M
1
L
−3
T
0
)(M
0
L
2
T
−2
).
Hence, the ratio p/(ρv
2
) is dimensionless since the dimensions of the
numerator and denominator coincide.
There is a general algorithm capable of determining whether one or another
parameter is dimensionallydependent or dimensionallyindependent of other
given parameters. For example, let us consider mechanics, in which there
are three basic measurement units: mass (M), length (L), time (T) and it is
required to clarify whether the parameter P with the dimension of pressure
is dimensionallydependent on parameters µ with the dimension of dynamic
viscosity, g with the dimension of acceleration and D with the dimension of
length. Let us show how to do it.
Write the dimensions of all the parameters under consideration
[p] = M
1
L
−1
T
−2
,
[µ] = ML
−1
T
−1
,
[g] = LT
−2
,
[D] = L.
Now let us look for the numbers m
1
, m
2
and m
3
that would obey the equality
[p] = [µ]
m
1
[g]
m
2
[D]
m
3
.
Insertion of the considered dimensions in this equality yields
M
1
L
−1
T
−2
= (ML
−1
T
−1
)
m
1
(LT
−2
)
m
2
(L)
m
3
.
Equating the exponents of identical basic measurement units in the left and
righthand sides of the last relation, we get a system of three linear equations
1 = m
1
,
−1 = −m
1
+m
2
+m
3
,
−2 = −m
1
−2m
2
to determine the three unknown quantities m
1
, m
2
, m
3
. This system has a
single solution
m
1
= 1, m
2
=
1
2
, m
3
= −
1
2
.
Thus, we obtain
[p] = [µ][g]
1/2
[D]
−1/2
(6.11)
which proves that pressure is a dimensionallydependent parameter in the
system of three parameters – viscosity, acceleration, length. From Eq. (6.11)
166 6 Dimensional Theory
it follows that the ratio p/(µg
1/2
D
−1/2
) is a dimensionless quantity since the
dimensions of the numerator and denominator coincide.
Ina similar manner many examples of the same type could be considered. Let
us focus our attention on the question of the maximal number of dimensionally
dependent quantities for a given set of dimensional parameters {a
1
, a
2
, . . . , a
n
}.
If this set contains n elements, it is always possible to separate from it a
subset containing the maximum possible number k ≤ n of dimensionally
independent parameters.
This assertion follows from the known theorem of algebra that states that
from any system of linear equations can be separated the maximum possible
number of linearly independent equations, this number being called the rank
of the system of equations. If we write dimensional formulas for all the
parameters a
1
, a
2
, . . . , a
n
,
[a
1
] = [O
1
]
x
1
[O
2
]
x
2
. . . .[O
k
]
x
k
,
[a
2
] = [O
1
]
y
1
[O
2
]
y
2
. . . .[O
k
]
y
k
,
. . . . . . . . . . . . . . . . . . . . . . . . ..,
[a
n
] = [O
1
]
z
1
[O
2
]
z
2
. . . .[O
k
]
z
k
where O
1
, O
2
, . . . , O
k
are symbols of basic measurement units in the given
system of units and x
1
, x
2
, . . . , x
k
; y
1
, y
2
, . . . , y
k
; z
1
, z
2
, . . . , z
k
are exponents of
this formula, the question of the dimensional dependence (or independence)
of each of themon the other as well as the question of the maximumnumber of
dimensionallyindependent quantities is reduced to solutions of the system of
linear equations
m
1
x
1
+m
2
y
1
+. . . +m
n
z
1
= 0,
m
1
x
2
+m
2
y
2
+. . . +m
n
z
2
= 0,
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
m
1
x
k
+m
2
y
k
+. . . +m
n
z
k
= 0
(k equations with n unknowns) with respect to the unknowns m
1
, m
2
, . . . , m
n
and to determine the rank of the system matrix.
In order to separate a subset containing the maximum number of
dimensionallyindependent quantities of the set {a
1
, a
2
, . . . , a
n
} we proceed
as follows. Let us take a quantity a
1
. If it is a dimensional quantity, the next
quantity a
2
is added to it. If a
2
has dimension different from the dimension of
a
1
, the system{a
1
, a
2
} would consist of dimensionallyindependent quantities.
Next a
3
is added to the system of quantities {a
1
, a
2
} a
3
. If the dimension
of a
3
is expressed through {a
1
, a
2
} by the formula (6.10), this quantity is
rejected and instead we take a
4
, which in its turn is tested for independence
from the quantities {a
1
, a
2
}. If the dimension of a
3
is not expressed through
the dimensions of {a
1
, a
2
}, the system {a
1
, a
2
, a
3
} represents a system of
dimensionallyindependent quantities. In such a way all quantities in the set
6.6 DimensionallyDependent and DimensionallyIndependent Quantities 167
{a
1
, a
2
, . . . , a
n
} are considered and gradually the subset containing the number
of dimensionallyindependent quantities is separated.
Note that for one and the same set several different subsets containing
the maximum number of dimensionallyindependent quantities could be
separated. Such subsets by themselves can be different but the number of
elements in them would be equal.
In particular, when the dimensions of all quantities in the set {a
1
, a
2
, . . . , a
n
}
are expressed through the dimensions M, L, T, this set can have no more than
three dimensionallyindependent parameters.
Exercise. It is required to separate the maximum number of dimensionally
independent quantities among a set of the following parameters: p – pressure;
ρ – density; v – velocity; Q – volumetric ﬂow rate; ν – kinematic viscosity;
D – diameter; g – acceleration due to gravity.
Solution. Write the dimensional formulas for all given parameters using the
SI system of basic measurement units
[p] =
kg
m×s
2
; [ρ] =
kg
m
3
; [v] =
m
s
; [Q] =
m
3
s
;
[ν] =
m
2
s
; [g] =
m
s
2
; [D] = m.
Since all the parameters are expressed through mass, length and time, the max
imumnumber of dimensionallyindependent quantities is less than or equal to
three. As the ﬁrst quantity, we take D. As the second quantity we take g, because
its dimension contains time and consequently it cannot be expressed through
the dimensions of D. Finally, as the third dimensionallyindependent quantity
we take ρ. It is evident that the dimension of ρ cannot be expressed through
the dimensions of D and g, since it contains mass. Hence, the set {D, g, ρ}
consisting of three parameters represents a basis of the maximum number of
dimensionallyindependent quantities in the given set of parameters.
There are of course other possible subsets of the given set consisting also
of a maximum number of dimensionallyindependent quantities. It is easy
to verify that, for example, the subsets {p, v, D}, {ρ, Q, D} and so on consist
of dimensionallyindependent quantities and the number of dimensionally
independent quantities in each of these subsets is a maximum and equal to
three.
Exercises.
1. It is required to separate a basis consisting of the maximum number of
dimensionallyindependent quantities among a set of the following
parameters: ρ – density; ω – frequency of revolutions; D – wheel
diameter; Q – volumetric ﬂow rate; g – acceleration due to gravity.
Answer. For example {ρ, ω, D}.
168 6 Dimensional Theory
2. It is required to separate a basis consisting of a maximum number of
dimensionallyindependent quantities among a set of the following
parameters: v – velocity; S – area of the crosssection occupied by ﬂuid;
g – acceleration due to gravity; D – diameter of the pipeline; α – angle of
inclination of the pipeline axis to the horizontal (dimensional quantity);
ν – kinematic viscosity, m
2
s
−1
.
Answer. For example {D, v}.
3. It is required to separate a basis consisting of a maximum number of
dimensionallyindependent quantities among a set of the following
parameters: p – pressure; v – velocity; θ – temperature; ρ – density;
λ – thermal diffusivity factor, W m
−1
K
−1
; c – heat capacity, J kg
−1
K
−1
).
Answer. For example {ρ, v, c, θ}.
6.7
Buckingham Theorem
Now we go to the proof of the central theorem of dimensional theory, the
Buckingham theorem, which was partially formulated in Section 6.5.
Let a physical regularity be represented by a function
A = f (a
1
, a
2
, . . . , a
k
, a
k+1
, . . . , a
n
) (6.12)
of arguments a
1
, a
2
, . . . , a
n
among which can be dimensional as well as
dimensionless parameters.
Let the maximum number of dimensionallyindependent arguments in the
set a
1
, a
2
, . . . , a
n
be k and without disturbing the generality it can be taken that
the ﬁrst k arguments are a
1
, a
2
, . . . , a
k
. Then the remaining (n −k) arguments
of this function a
k+1
, a
k+2
, . . . , a
n
would be dimensionallydependent on the
ﬁrst k arguments, that is
[a
k+1
] = [a
1
]
m
1
[a
2
]
m
2
. . . [a
k
]
m
k
,
[a
k+2
] = [a
1
]
n
1
[a
2
]
n
2
. . . [a
k
]
n
k
,
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
[a
n
] = [a
1
]
p
1
[a
2
]
p
2
. . . [a
k
]
p
k
where m
i
, n
i
, . . . , p
i
are real numbers. Thus, the relations
1
=
a
k+1
a
m
1
1
a
m
2
2
· · · a
m
k
k
,
2
=
a
k+2
a
n
1
1
a
n
2
2
· · · a
n
k
k
, (6.13)
n−k
=
a
n
a
p
1
1
a
p
2
2
· · · a
p
k
k
6.7 Buckingham Theorem 169
are dimensionless parameters since the dimensions of the quantities in the
numerators and denominators of these fractions are identical.
The dimension of the quantity A should also be expressed through the
dimensions of the arguments a
1
, a
2
, . . . , a
k
. If it is not expressed through
these dimensions, it would also not be expressed through the dimensions of
all the quantities a
1
, a
2
, . . . , a
n
. Hence, there should exist a relation
[A] = [a
1
]
q
1
[a
2
]
q
2
. . . [a
k
]
q
k
meaning that the ratio
=
A
a
q
1
1
a
q
2
2
· · · a
q
k
k
(6.14)
has to be dimensionless.
Let us revert to the dependence (6.12). This dependence may be rewritten as
A
a
q
1
1
a
q
2
2
· · · a
q
k
k
=
¯
f
_
a
1
, a
2
, . . . , a
k
,
a
k+1
a
m
1
1
a
m
2
2
· · · a
m
k
k
, . . . ,
a
n
a
p
i
1
a
p
2
2
· · · a
p
k
k
_
where
¯
f represents a function resulting from f after redeﬁnition of its
arguments. This dependence with (6.13) and (6.14) can be represented as
follows
=
¯
f (a
1
, a
2
, . . . , a
k
,
1
,
2
, . . . ,
n−k
). (6.15)
If we now arbitrarily and independently from each other vary the numerical
values of the arguments a
1
, a
2
, . . . , a
k
by going from one system of basic
measurement units to another one, the numerical values of the parameters
and
1
,
2
, . . . ,
n−k
would not be changed because they are dimensionless
quantities. From here it follows that the function cannot depend on its
ﬁrst k arguments a
1
, a
2
, . . . , a
k
, since the dependence (6.15) would have the
following simple form
=
¯
f (
1
,
2
, . . . ,
n−k
). (6.16)
Thus, we have shown that any physical dependence between dimensional
quantities of the type (6.12) will be invariant, that is, independent of the
choice of measurement units, of the form (6.16) between dimensionless
complexes made up from arguments of the dependence under consideration.
The number of these complexes would be less than the number of arguments
of the initial dependence by a number equal to the maximum number of
dimensionallyindependent quantities among these arguments.
Exercise 1. The augmentation of the crosssection area S of a steel pipeline
when setting up in it excess pressure P depends on the value of this pressure,
the diameter D of the pipeline, the thickness δ of the pipeline wall and the
elastic modulus (Young’s modulus) E of the steel fromwhich the pipe is made.
170 6 Dimensional Theory
Using the theorem, it is required to write this dependence in dimensionless
form and to clarify how many dimensionless parameters determine it.
Solution. The dependence to be sought can be written in a general form
as S = f (P, D, δ, E). The dimensions of the arguments of this dependence
in the SI system are: [P] = kg m
−1
s
−2
; [D] = m; [δ] = m; [E] = kg m
−1
s
−2
.
Among these there are only two dimensionallyindependent quantities, for
example D and E. Consequently, the number of arguments in the dependence
under consideration may be reduced to two. Thus we have:
= S/D
2
;
1
= P/E;
2
= δ/D;
=
¯
f (
1
,
2
) or S = D
2
·
¯
f (P/E, δ/D).
Hence, the dimensional analysis has shown that in the considered dependence
there are only two dimensionless complexes P/E and δ/D instead of four
dimensional arguments.
Moreover, if additionally we invoke the reasoning that the variation of
pipeline crosssection area S should be proportional to P/E (the function,
¯
f ,
could be expanded in a Taylor series in the vicinity of the point
1
= 0 leaving
in the expansion only the ﬁrst term, because S = f (0,
2
) = 0 at P = 0 and
the ratio P/E is very small: P ≈ 2–7 · 10
6
a; E
∼
= 2 · 10
11
a, and inversely
proportional to δ/D(δ ≈ 5–10 mm; D ≈ 300–100 mm), the dependence under
study could be written as
S = D
2
·
P
E
·
1
δ/D
·
¯
f
0
=
¯
f
0
·
D
3
· P
δ · E
where
¯
f
0
is a certain constant. Hence, in the dependence under study there is
only one constant left to be obtained. Theoretical investigation shows that the
constant
¯
f
0
is equal to π/4
∼
= 0.785.
Exercise 2. It is known that the laminar ﬂow of a viscous incompressible
ﬂuid in a circular pipe loses stability and becomes turbulent. It is required
to investigate the dependence of the critical ﬂow velocity v
cr
at which this
transitionhappens, taking the critical velocity as a functionof three parameters:
pipeline diameter d, dynamic viscosity µ and density ρ of the ﬂuid. In other
words it is required to investigate the dependence v
cr
= f (d, µ, ρ).
Solution. In the given case A = v
cr
, a
1
= d; a
2
= µ; a
3
= ρ; n = 3. The di
mensions of the parameters in the SI system are: [v
cr
] = m s
−1
; [d] = m;
[µ] = kg m
−1
s
−1
; [ρ] = kg m
−3
. From this it follows that all three arguments
of the function are dimensionallyindependent, i.e. k = n = 3, and the number
of arguments in the dimensionless writing of the function under study may
be reduced by three, i.e. to 0.
6.7 Buckingham Theorem 171
The single dimensionless complex can be written as = v
cr
· d/(µ/ρ).
The sought dependence then takes the especially simple form
cr
=
v
cr
· d
(µ/ρ)
= const. (6.17)
The ratio µ/ρ is called the kinematic viscosity of the ﬂuid and is denoted
by ν ([ν] = L
2
/T). In the SI system the unit of kinematic viscosity is
stokes (St); 1 St = 10
−4
m
2
s
−1
; the viscosity of water is ≈0.01 St = 1 cSt
(centistokes) = 10
−6
m
2
s
−1
.
The dimensionless parameter
cr
determining the transition of laminar
ﬂow to turbulent ﬂow is called the critical Reynolds number and is denoted by
Re
cr
. The theory and experiments have shown that Re
cr
∼
= 2300. At Re < Re
cr
the ﬂow is laminar whereas at Re > Re
cr
it is turbulent.
Exercise 3. A ball of mass m and diameter D is dropped in a viscous ﬂuid with
density ρ and viscosity ν under the action of gravity (gravity acceleration – g)
with constant velocity v. The dependence of this velocity on the governing
parameters: v = f (m, D, ρ, ν, g) is to be investigated. Using the theorem it
is required to write the sought dependence in dimensional form.
Solution. Dimensions of the parameters (n = 5), in this problem(A = v, a
1
=
m, a
2
= D, a
3
= ρ, a
4
= ν, a
5
= g), in the SI system are [v] = m s
−1
; [m] = kg;
[D] = m; [ρ] = kg m
−3
; [ν] = m
2
s
−1
; [g] = m s
−2
. The maximum number of
dimensionallyindependent parameters among the arguments is equal to
three (as such parameters can be taken for example ρ, D and g), the number
of arguments in the sought dimensionless writing may be reduced to two
(5 −3 = 2).
Setting up dimensionless complexes ,
1
and
2
=
v
g
1/2
D
1/2
,
1
=
m
ρ · D
3
,
2
=
ν
g
1/2
D
3/2
,
the sought dependence can be represented as follows
= f (
1
,
2
)
or
v =
_
gD · f
_
m/ρD
3
, ν/
_
gD
3
_
.
It is seen that the dependence contains in fact not ﬁve dimensional arguments,
but only two dimensionless complexes.
Exercises.
1. Using the theorem, it is required to write in dimensional form the
dependence of the resistance force F experienced by a submarine
172 6 Dimensional Theory
moving in water (density ρ, kinematic viscosity ν) with velocity v, if we
accept that this force depends also on the diameter of the submarine
crosssection D and the submarine length L, that is F = f (v, ρ, ν, D, L).
Answer. F/(ρv
2
D
2
) =
¯
f (vD/ν, D/L).
2. Using the theorem, it is required to write in dimensional form the
dependence of the oscillation period T of a mathematical pendulum
(massive point on nonstretchable line), if the mass of the latter is m,
the length of the line is L and the acceleration due to gravity is g, that is
T = f (m, L, g).
Answer. T/
_
L/g = const.
3. Using the theorem, it is required to write in dimensional form the
dependence of the time T of complete outﬂow of ﬂuid (density ρ and
viscosity ν) from a tank car with diameter D and length L. The outﬂow
happens under gravity (acceleration due to gravity g) through the drain
system with ﬂow area s located at the tank bottom. The sought function
is T = f (ρ, ν, g, D, L, s).
Answer. T/
_
D/g =
¯
f (
√
gD
3/2
/ν, L/D, s/D
2
).
4. Using the theorem, it is required to write in dimensional form the
same dependence as in the previous exercise with the single distinction
that the outﬂow happens not only under the action of gravity but also
under the action of positive pressure p created inside the tank. The
sought function is T = f (p, ρ, ν, g, D, L, s).
Answer. T/
_
D/g =
¯
f (
√
gD
3/2
/ν, p/(ρgD), L/D, s/D
2
).
5. Using the theorem, it is required to write in dimensional form the
dependence of the volumetric ﬂow rate Q of a ﬂuid with density ρ and
viscosity ν in an inclined pipeline (inclination angle α) having
crosssection area S
0
, if this ﬂow represents voluntary (gravity) ﬂow.
The ﬂow happens under the action of the gravity force projection
g sinα, the area of the pipeline crosssection ﬁlled with ﬂuid is S < S
0
.
The sought function is Q = f (g sinα, S, S
0
, ρ, ν).
Answer. Q/[S
5/4
0
· (g sinα)
1/2
] =
¯
f (S/S
0
, (g sinα)
1/2
· S
3/4
0
/ν).
173
7
Physical Modeling of Phenomena
The main advantage of the dimensional theory is that it opens up possibilities
to use the similarity laws of physical phenomena and allows modeling of these
phenomena through replacing them in nature by similar phenomena on a
reduced or enlarged scale under experimental conditions.
7.1
Similarity of Phenomena and the Principle of Modeling
In order to elucidate the essence of modeling, let us consider several examples.
Assume that it is required to calculate the radius of a circle inscribed in a
triangle whose sides are very large, for example 1, 2 and 3 km. This problem
may be easily solved by simple algebraic calculation. However, if it was
necessary to perform this measurement experimentally, one could proceed
as follows: on a sheet of paper draw a triangle on a reduced scale similar to
the given triangle, for example a triangle with sides 10, 20 and 30 cm. The
similarity factor of this triangle to the fullscale one would be equal to 10 000.
Inscribing a circle in the depicted triangle, it is easy to measure its radius. Then
the obtained number multiplied by the similarity factor, i.e. 10 000, yields the
sought radius of the circle inscribed in the fullscale triangle. Of course in the
case under consideration we are dealing with a simple geometric similarity and
geometric modeling but it clariﬁes the essence of modeling in the general case.
Let it be required to determine experimentally the dump time of a tank of
complex geometrical form and very large size. In order to solve this problem it
is decided to make a copy of the tank on a reduced scale, ﬁll it with some model
liquid and then measure the dump time. The question arises as to whether
it is enough to provide merely geometric similarity of fullscale and model
tanks and to use in experiments the same liquid or is it necessary to replace
the liquid with another one with specially selected properties. Moreover, is it
necessary to determine the ratio between the measured dump time and that
actually taking place in the fullscale object. The answers to these questions
should give the theory of simulation (modeling).
Finally, one more example from engineering practice. It is required to clarify
whether the river dam will withstand the dynamic head of ﬂooding water.
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
174 7 Physical Modeling of Phenomena
For this purpose a reduced size copy of the dam is made, mounted in an
experimental channel in a hydrological laboratory. It is evident that if the dam
were made from the same material as in the natural conditions, i.e. from
ferroconcrete, the dam would withstand any water head. It is necessary to
determine the velocity of water in the experimental channel in order to model
the river head. When reducing the linear sizes of the considered phenomenon,
it is insufﬁcient to provide geometric similarity. It is necessary especially to
change the scales of many other parameters of the phenomenon.
Similar problems take place in different ﬁelds of engineering: hydraulic
engineering, aviation, transport, the storage of oil and gas and so on.
Deﬁnition. Two phenomena are called similar when, from the given parame
ters of one phenomenon, analogous parameters of another phenomenon are
determined by simple recalculation of the same kind as for transition fromone
system of measurement units to another. Each of such phenomena is called a
model of another phenomenon from this set of phenomena.
7.2
Similarity Criteria
Let us determine the necessary and sufﬁcient conditions of two phenomena
to be similar. Such conditions are called similarity criteria.
Let a phenomenon be such that a certain physical quantity A is determined
by a set of physical parameters a
1
, a
2
, . . . , a
n
, so that
A = f (a
1
, a
2
, . . . , a
n
). (7.1)
The model under consideration consists of the dependence of an analogous
physical quantity A
on the same physical parameters the numerical values of
which a
1
, a
2
, . . . , a
n
differ from those determining the quantity A. Thus, we
have
A
= f
_
a
1
, a
2
, . . . , a
n
_
. (7.2)
In accordance with the Itheorem both dependences (7.1) and (7.2) may be
rewritten in dimensionless form as follows
= f
_
1
,
2
, . . . ,
n−k
_
,
(7.3)
= f
_
1
,
2
, . . . ,
n−k
_
,
where k is the number of dimensionallyindependent parameters among the
quantities a
1
, a
2
, . . . , a
n
.
Relations (7.3) show that if the parameters a
1
, a
2
, . . . , a
n
are chosen such
that the following conditions are obeyed
1
=
1
,
2
=
2
, . . . ,
n
=
n
. (7.4)
7.3 Modeling of Viscous Fluid Flow in a Pipe 175
Then the following condition would also be satisﬁed
= . (7.5)
The value of the parameter A could be found by simple recalculation of the
parameter A
through
A = A
·
a
m
1
1
a
m
2
2
· · · a
m
k
k
a
m
1
1
a
m
2
2
· · · a
m
k
k
(7.6)
and the considered phenomena would, by deﬁnition, be similar.
Thus, necessary and sufﬁcient conditions of two phenomena to be
similar are equalities of the dimensionless complexes determining these
phenomena, namely conditions (7.4). Hence, the dimensionless parameters
1
,
2
, . . . ,
n−k
are the sought similarity criteria.
7.3
Modeling of Viscous Fluid Flow in a Pipe
As an example of two similar phenomena, consider the modeling of the
stationary ﬂow of a viscous incompressible ﬂuid in a model pipe with reduced
size as compared to the fullscale one.
In accordance with the results given in Section 1.6 this dependence is
p
L
d
·
ρv
2
2
= λ(Re, ε).
Denote through ρ
, v
, d
, µ
,
, L
, p
the values of the hydrodynamic
parameters relating to the ﬂow in the model pipe. The same parameters
without the superscript ‘‘prime’’ refer to the phenomena to be modeled. Then
p
L
d
·
ρ
v
2
2
= λ(Re
, ε
).
The parameters of the model pipe and the ﬂow regime in it are chosen to obey
the relations
Re = Re
, ε = ε
(7.7)
or
vd
ν
=
v
d
ν
,
d
=
d
.
In so doing we ensure the equality
p
L
d
·
ρ
v
2
2
=
p
L
d
·
ρv
2
2
176 7 Physical Modeling of Phenomena
or
p = p
·
L
L
·
d
d
·
ρ
ρ
·
_
ν
ν
_
2
. (7.8)
If we now take the ratio d
/d, showing how many times smaller are the sizes
of the model pipe than the sizes of the fullscale pipe (factor of geometric
similarity), and the ratio ν
/ν of the viscosities of ﬂuids ﬂowing in fullscale
and model pipes, respectively, we could determine the ﬂuid ﬂow velocity and
roughness of the pipe walls of the experimental (model) installation needed to
achieve the similarity
v
= v ·
d
d
·
ν
ν
,
= ·
d
d
. (7.9)
Formulas (7.9) state that the similarity in the case under consideration is
afforded by fulﬁlling two conditions: equality of the Reynolds numbers and
equality of the wall relative roughness. Consequently, the Reynolds number
and relative roughness serve as similarity criteria for the problem on the ﬂow
of a viscous ﬂuid in a pipe.
7.4
Modeling Gravity Fluid Flow
This type of ﬂowhas already beenconsidered inSection 3.7. We nowset a ques
tiononthe physical modeling of this process. Let the ﬂowrate Q of the ﬂuid ina
pipe inclined to the horizontal at anangle α be given. It is required to determine
the depth h of the ﬂuid ﬁlling the pipe crosssection, that is, the function
h = f (Q, d, g sinα, ν, ).
Rewrite the dependence to be sought in dimensionless form using the 
theorem. Among ﬁve arguments of this function there are two dimensionally
independent, for example ν, d, therefore in dimensionless form the number
of independent arguments will be reduced from ﬁve to three.
h
d
=
¯
f
_
(Q/d
2
) · d
ν
,
_
gd sinα · d
ν
,
d
_
.
It is more convenient in this dependence to use the ratio of the ﬁrst argument
to the second one instead of the second argument. As a result we have
h
d
=
¯
f
1
_
(Q/d
2
) · d
ν
,
(Q/d
2
)
_
gd sinα
,
d
_
.
In the ﬁrst argument it is easy to recognize the Reynolds number Re of the ﬂow
calculated from the velocity v = Q/d
2
and the kinematic viscosity ν = µ/ρ,
so that the ﬁrst similarity criteria would be
1
= Re. The second argument
7.4 Modeling Gravity Fluid Flow 177
is called the Froude number, Fr. The Froude number is, in general, equal to
v
2
/gd, thus in our case we are dealing with a Froude number calculated from
the velocity Q/d
2
and the component of the acceleration due togravity g sinα.
Nevertheless, the second similarity criterion can be taken as
2
= Fr. Finally,
the third argument of this dependence
3
= ε is the relative roughness of the
pipe’s internal wall surface. Hence, the dependence under study is
h
d
=
ˆ
f (Re, Fr, ε). (7.10)
To model this phenomenon it is necessary to provide the following similarity
conditions
Re
= Re ⇒
Q
ν
d
=
Q
νd
⇒ Q
= Q ·
_
ν
ν
_
·
_
d
d
_
,
Fr
= Fr ⇒
Q
2
g sinα
· d
5
=
Q
2
g sinα · d
5
,
⇒ Q
= Q ·
_
sinα
sinα
·
_
d
d
_
5/2
, (7.11)
ε
= ε ⇒
d
=
d
⇒
= ·
d
d
.
If these conditions are obeyed, the depth h to which the pipe is ﬁlled can be
calculated using the formula
h
d
=
h
d
⇒ h = h
·
d
d
. (7.12)
In order for the ﬁrst two conditions to be consistent, the following equalities
should be true
_
ν
ν
_
·
_
d
d
_
=
_
sinα
sinα
·
_
d
d
_
5/2
or
sinα
= sinα ·
_
ν
ν
_
2
·
_
d
d
_
3
. (7.13)
If in the model the same liquid as in the fullscale pipe is used, then a different
slope of the pipe should be used in the model, it must be chosen in accordance
with relation (7.13). If α
= α, then the viscosity of the ﬂuid used in the model
should be different, namely ν
= ν · (d
/d)
3/2
.
For example, let the linear size of the experimental pipe be 1/5th that
of the fullscale pipe, that is, d
/d = 1/5, and the viscosity ν
of the model
liquid be ﬁve times less than the viscosity of the natural liquid. Then
the conditions for the experiment should be: Q
= 1/5 · 1/5 · Q = 0.04 · Q;
sinα
= 1/25 · 5
3
· sinα = 5 · sinα;
= 0.2 · , that is, the ﬂuid ﬂow rate in
the experiment has to be reduced by a factor of 25, the internal surface of the
178 7 Physical Modeling of Phenomena
pipe should be so polished that the absolute roughness is reduced by ﬁve times
as compared to the fullscale pipe and the sine of the inclination angle of the
model pipe should be increased by ﬁve times. Thus, the depth h of the ﬂow in
the fullscale pipe has to be ﬁve times greater than the model depth h
, that is,
h = 5 · h
.
In experiments it is not always possible to obey all the required similarity
conditions, therefore only the main conditions are satisﬁed. If, for example, in
a given case a ﬂow of diesel fuel with viscosity 3 cSt is to be investigated, then
benzene with viscosity
∼
=0.6 cSt could be taken for the model. Reduction of
the ﬂow rate and enhancement of the pipe slope (at small values of α) present
no special problems. It is far more complicated to fulﬁll the last similarity
condition and so sometimes it is neglected.
7.5
Modeling the Fluid Outﬂow from a Tank
Let there be a railway tank with a boiler diameter D and length L or, in general,
a tank of arbitrary geometric form intended to transport oil with density ρ and
kinematic viscosity ν. The tank is provided with bottom drain equipment with
the area of the ﬂowing crosssection S. It is required to design an experimental
installation to model the process and to investigate the dependence of the time
T of the oil outﬂow on the parameters of the liquid and tank.
The dependence under study T = f (D, L, ρ, ν, g, S) in dimensionless form
is written as
T
_
D/g
=
¯
f
_
L
D
,
S
D
2
,
_
gD · D
ν
_
.
Therefore the similarity criteria are:
1
=
L
D
,
2
=
S
D
2
the criteria of geometric similarity;
3
=
g
1/2
· D
3/2
ν
the criterion of dynamic similarity.
At the fulﬁllment of the ﬁrst two criteria at which the model tank should
be similar to the fullscale one, the phenomena would be similar under the
condition
g
1/2
· D
3/2
ν
=
g
1/2
· D
3/2
ν
,
where the superscripts ‘‘prime’’ refer to the model parameters. The last relation
shows that the sufﬁcient condition for the dynamic similarity of the model
7.6 Similarity Criteria for the Operation of Centrifugal Pumps 179
tank to the fullscale one is the condition
ν
= ν ·
_
D
D
_
3/2
.
If this condition is fulﬁlled, then T
/
_
D
/g = T/
_
D/g or
T = T
·
_
D
D
. (7.14)
From this follows, in particular, that for modeling of oil outﬂow on the
installation with sizes distinct from the fullscale tank, it is necessary to
use a liquid with velocity also differing from the oil under study. Let the
experimental tank be a cylindrical tank 10 times smaller than the fullscale
one, that is D/D
= 10. Then ν
∼
= 0.0316 · ν, i.e. the kinematic viscosity of the
liquid used in the experiments, should be 32 times lower than the viscosity
of the oil being modeled. If this oil has sufﬁciently high viscosity then it is
easy to select such a liquid, otherwise the problem is complicated. The time
T of oil outﬂow is determined by the time T
measured in the experiment. In
accordance with the formula (7.14) it is T
∼
= 3.162 · T
.
7.6
Similarity Criteria for the Operation of Centrifugal Pumps
As already described in Section 4.2.1 pumps are equipment to make liquids
ﬂow against a pressure force, that is, in the direction from lesser pressure p
suc
at the suction line to the greater pressure p
d
at the line of discharge under
pressure. Of course, it is possible only by the work of external energy sources
(mechanical, thermal, electrical and so on).
Centrifugal pumps represent a variety of pump in which the centrifugal
force, acting on ﬂuid particles rotating in the impeller, makes the liquid ﬂow
against the pressure. The propulsive device of the impeller or, as it is called,
the pump drive, can be an internal combustion engine, vaporgas turbine or
any other source of rotational moment (see Fig. 4.1).
The volumetric ﬂuid ﬂow rate Q (or, as it is called, the feed) depends on
the pressure drop p = p
d
− p
suc
which the liquid should overcome at its ﬂow
from the impeller center to the periphery. The greater the pressure drop the
lesser is the ﬂow rate of the liquid.
If the area σ of the outlet branch pipe at the discharge line is given by the
pump design, the ﬂuid velocity Q/σ is a function of the impeller diameter
D
im
(centrifugal pumps may have accessory impellers), the angular velocity ω
of the impeller rotation (centrifugal pumps may be supplied with equipment
to change ω), the density ρ and viscosity ν of the ﬂuid to be pumped, so
that
Q
σ
= f (p, D
im
, ω, ρ, ν). (7.15)
180 7 Physical Modeling of Phenomena
The dependence (7.15) is called the (Q − p) characteristic of the cen
trifugal pump and is to a large extent determined by its constructive
peculiarities.
Among the arguments of the function (7.15) there are three dimensionally
independent parameters, for example D
im
, ω, ρ, therefore its dimensionless
form could be reduced to
Q/σ
ωD
im
=
˜
f
_
p
ρω
2
D
im
2
,
ν
ωD
im
· D
im
_
.
The last argument represents none other than a quantity inversely proportional
to the Reynolds number Re. This parameter reﬂects the effect of viscosity on
the characteristic of the pump operation.
If we introduce into consideration the quantity H = p/ρg (the differential
head of the pump) and solve Eq. (7.15) with respect to this head, we get the
socalled (Q − H) characteristic of the pump
H = ω
2
D
im
2
· F
_
Q
ωD
im
,
ωD
im
2
ν
_
. (7.16)
Here the constants σ and g are taken into account by the formof the function F.
It is interesting to note that the density ρ of the liquid to be pumped does
not enter into the dependence (7.16), i.e. the form of the latter is true for the
pump operated by liquid of any density. As for the inﬂuence of the viscosity
ν on the form of (Q − H) characteristic of centrifugal pumps, it is small.
Reynolds numbers ωD
im
2
/ν have rather large values (≈10
6
) due to the high
rotationvelocity of the pumpimpeller (ω ≈ 300 s
−1
, ω = 2π · n; n ≈ 3000 rpm;
D
im
≈ 0.2–0.7 m; v ≈ 1–10 · 10
−6
m
2
s
−1
), therefore in practice their variation
only slightly affects the (Q − H) characteristic of the centrifugal pump and
thus the inﬂuence of the pumping ﬂuid viscosity is also small. The latter
conclusion is of course true only up to certain limits.
If we ignore the inﬂuence of pumping ﬂuid viscosity on the (Q − H)
characteristic of the centrifugal pump, then it can be represented in the simple
form
H = ω
2
D
im
2
· F
_
Q
ωD
κ
_
. (7.17)
From the derived formulas some practically important conclusions follow:
•
If a centrifugal pump operating with angular velocity ω
0
or revolutions per
minute n
0
has a characteristic H = F
∗
(Q), the same pump working with
varied rotation frequency ω or revolutions per minute n, has the
characteristic
H =
_
ω
ω
0
_
2
· F
∗
_
ω
0
ω
· Q
_
; (7.18)
•
If a centrifugal pump, operating with an impeller of diameter D
κ0
has the
characteristic H = F
∗
(Q), the same pump operating with another
7.6 Similarity Criteria for the Operation of Centrifugal Pumps 181
impeller of diameter D
im
has the characteristic
H =
_
D
im
D
im0
_
2
· F
∗
_
D
im0
D
im
· Q
_
; (7.19)
•
If a centrifugal pump, operating with rotational velocity ω
0
and impeller
diameter D
im0
has the characteristic H = F
∗
(Q), the same pump
operating with varied rotational frequency ω and impeller diameter D
im
has
the characteristic
H =
_
ωD
im
ω
0
D
im0
_
2
· F
∗
_
ω
0
D
im0
ωD
im
· Q
_
. (7.20)
Rules (7.18)–(7.20) allow us to change the (Q − H) characteristic of
centrifugal pumps by changing the rotation speed or/and the impeller
diameter.
In many cases, as has been said, the (Q − H) characteristics of centrifugal
pumps are represented in the form of a parabola
H = F
∗
(Q) = a − b · Q
2
(7.21)
where a and b are approximation factors. If we now change the impeller
diameter from D
im0
to D
im
and the rotation frequency from ω
0
to ω, the
(Q − H) characteristic of the same pump takes the form
H = a ·
ω
2
D
im
2
ω
2
0
D
2
im0
− b · Q
2
, (7.22)
that is the parabola graphic in the plane (Q, H) undergoes a displacement
along the Haxis by
a ·
_
1 −
ω
2
D
im
2
ω
2
0
D
im0
2
_
.
Exercise. The diameter of a centrifugal pump impeller is 490 mm and the
impeller velocity is 3200 rpm. The pump has the following characteristic
H = 331 − 0.451 · 10
−4
· Q
2
(H in m, Q in m
3
h
−1
). It is required to determine the characteristic of the
same pump if we reduce the impeller diameter to 480 mm and the number of
revolutions per minute to 3000.
Solution. In accordance with Eq. (7.22) the sought characteristic has the
following form
H = 331 ·
_
3000 · 480
3200 · 490
_
2
− 0.451 · 10
−4
· Q
2
= 279 − 0.451 · 10
−4
· Q
2
.
182 7 Physical Modeling of Phenomena
Exercises.
1. How does the (Q − H) characteristic of a centrifugal pump vary if the
frequency speed of the impeller is increased from 3000 to 3200 rpm?
The given characteristic of the pump is: H = 360 − 0.42 · 10
−4
· Q
2
,
where H is measured in m, Q in m
3
h
−1
.
Answer. H = 410 − 0.42 · 10
−4
· Q
2
.
2. How does the (Q − H) characteristic of a centrifugal pump vary if the
frequency speed of the impeller is increased from 3000 to 3200 rpm?
The given characteristic of the pump is: H = 360 − 0.42 · 10
−4
· Q
1.75
,
where H is measured in m, Q in m
3
h
−1
.
Answer. H = 410 − 0.375 · 10
−4
· Q
1.75
.
3. How does the (Q − H) characteristic of a centrifugal pump vary if the
impeller diameter is reduced from 480 to 470 mm. The given
characteristic of the pump is: H = 360 − 0.42 · 10
−4
· Q
1.75
, where
H is measured in m, Q in m
3
h
−1
.
Answer. H = 345 − 0.436 · 10
−4
· Q
1.75
.
183
8
Dimensionality and Similarity in Mathematical Modeling
of Processes
In the previous chapter we have seen that dimensionless similarity criteria
1
,
2
, . . . ,
n−k
appear in the considered problems after ﬁxation of a
set of governing parameters a
1
, a
2
, . . . , a
n
by way of heuristic reasoning.
Hence, it is by no means necessary to know which equations satisfy these
parameters and with which physical laws they are connected. It is enough
to know only the dimensions of these parameters in order to set up
dimensionless similarity criteria by which the class of considered phenomena
is characterized.
8.1
Origination of Similarity Criteria in the Equations of a Mathematical Model
A somewhat different approach takes place in the mathematical modeling of
phenomena in construction systems of algebraic or differential equations with
initial and boundary conditions in which we would like to see an adequate
model of the considered phenomena or process. Dimensionless similarity
parameters originate in these models in a strictly speciﬁed way of bringing
model equations to a dimensionless form.
Let us illustrate the afore said by a simple example. It is known that
onedimensional oscillations of a pont weight with mass m on an elastic
spring around an equilibrium position (x = 0) are described by an ordinary
differential equation
m
d
2
x
dt
2
= −k · x (8.1)
where x(t) is the linear dependence of the weight coordinate on time and k is
the restoring force factor. The initial, at t = 0, position and velocity ˙ x(0) = v
0
of the weight are also given, that is the mathematical model of the process.
Let us bring the model equation to dimensionless form. First we introduce
dimensionless variables
¯ x =
x
L
and
¯
˙ x =
˙ x
v
0
, (L = 0, v
0
= 0).
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
184 8 Dimensionality and Similarity in Mathematical Modeling of Processes
Then the dimensionless time
¯
t = t/(L/v
0
) is determined. In new variables the
differential equation transforms to
m·
L · d
2
¯ x
(L/v
0
)
2
· d
¯
t
2
= −k · (L · ¯ x)
or
d
2
¯ x
d
¯
t
2
= −
kL
2
mv
2
0
· ¯ x. (8.2)
The initial conditions take the form ¯ x(0) = 1;
¯
˙ x = 1.
Hence, the oscillation of different weights with different mass on springs
with different elasticity caused by different initial conditions is in fact
described by a mathematical model containing only one dimensionless parameter
= kL
2
/mv
2
0
. If in two situations this parameter appears to be identical, the
solutions of Eq. (8.2) are indistinguishable and consequently we have to deal
with similar situations.
It is easy to verify that the solution of the Eq. (8.2) is
¯ x(
¯
t) = cos
_
√
¯
t
_
+
1
√
· sin
_
√
¯
t
_
.
Returning to dimensional quantities, we obtain
x(t) = L · cos
__
k
m
· t
_
+
_
m
k
· v
0
· cos
__
k
m
· t
_
.
As is known
_
k/m = ω is the frequency of harmonic oscillations of the weight
and the amplitude h is the square root of the sum of the squared factors of the
sine and cosine
h =
_
L
2
+
mv
2
0
k
= L ·
_
1 +
1
.
8.2
OneDimensional NonStationary Flow of a Slightly Compressible Fluid
in a Pipeline
The theory of onedimensional nonstationary ﬂows of a slightly compressible
ﬂuid in a pipeline was considered in detail in Sections 5.1–5.3. The main
equations modeling such ﬂows are
_
¸
¸
_
¸
¸
_
∂p
∂t
+c
2
∂(ρv)
∂x
= 0
ρ
_
∂v
∂t
+v
∂v
∂x
_
= −
∂p
∂x
−
λ(Re, ε)
d
·
ρv
2
2
(8.3)
8.2 OneDimensional NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 185
in which the quantity
c =
1
_
ρ
0
K
+
ρ
0
d
δ · E
(8.4)
is called the velocity of the pressure wave propagations in pipeline.
Consider a certain problem on the calculation of nonstationary ﬂow of a
slightly compressible ﬂuid in pipeline taking Eqs. (8.3) as the mathematical
model of this ﬂow.
Problem. Let there be at the pipeline section 0 ≤ x ≤ L stationary ﬂuid
ﬂow with velocity v
0
. However, starting from some time a valve located
at the end crosssection of the pipeline x = L begins to vary its opening
level with frequency ω. It is required to reveal the similarity criteria of this
phenomenon.
Solution. Introduce the following dimensionless variables marked by the
horizontal bar at the top
t =
1
ω
·
¯
t, x = L · ¯ x, v = v
0
· ¯ v, ρ = ρ
0
· ¯ ρ, p = p
0
· ¯ p.
Equations (8.3) in the new variables take the form
_
¸
¸
_
¸
¸
_
p
0
1/ω
·
∂ ¯ p
∂
¯
t
+c
2
·
ρ
0
v
0
L
·
∂(ρv)
∂ ¯ x
= 0
ρ
0
¯ ρ
_
v
0
1/ω
·
∂¯ v
∂
¯
t
+
v
2
0
L
· ¯ v
∂v
∂ ¯ x
_
= −
p
0
L
·
∂ ¯ p
∂ ¯ x
−
λ(Re, ε)
d
· ρ
0
v
2
0
·
ρv
2
2
If now we take p
0
= ρ
0
v
0
c, the system of equations simpliﬁes to
_
¸
¸
_
¸
¸
_
ωL
c
·
∂ ¯ p
∂
¯
t
+
∂(ρv)
∂ ¯ x
= 0
ωL
c
·
∂¯ v
∂
¯
t
+
v
0
c
· ¯ v
∂¯ v
∂ ¯ x
= −
∂ ¯ p
∂ ¯ x
−
λ(Re, ε) · L
d
·
v
0
c
·
ρv
2
2
(8.5)
From Eqs. (8.5) it is seen that there are three dimensionless criteria governing
the class of problems under consideration and differing from each other only
by the numerical values of the parameters entering in these equations
1
=
v
0
c
;
2
=
ωL
c
;
3
=
λL
d
.
The ﬁrst of these criteria is called the Mach number. The ratio of the ﬁrst
criterion to the second one
1
/
2
= v
0
/(ωL) is called the Strouchal number
St = v
0
/(ωL).
The Mach number of the ﬂuid or gas ﬂow in a main pipeline is, as a rule,
very small (for example, for ﬂuid ﬂow M ≈ 0.001; for gas ﬂow M ≈ 0.03),
186 8 Dimensionality and Similarity in Mathematical Modeling of Processes
therefore it often turns out that the second term in the lefthand side of
the last equation (8.5) can be neglected in comparison with the ﬁrst one.
If the ﬂow velocity v
0
is comparable to the sound velocity c, as it may be,
for example, at the gas outﬂow from a pipeline through a short nozzle,
the second term would become almost the primary term and could not be
ignored.
The Strouchal number characterizes the degree of process nonstationarity.
If this number is great (St = v
0
/ωL 1, that is, the characteristic time of
the process 1/ω is large compared to L/v
0
the time of ﬂuid particle passage
through the pipeline), the nonstationarity degree is small and the process is
close to stationary. If the Strouchal number is small (that is the frequency
of the process ω is large compared to v
0
/L) then the nonstationarity of the
process cannot be neglected.
The third criterion λL/d does not have a special name. It characterizes the
magnitude of the resistance to the ﬂuid friction in the pipeline.
8.3
Gravity Fluid Flow in a Pipeline
Consider nowproblems connected with thegravity ﬂowof incompressible ﬂuid
in a pipeline (see Sections 3.7 and 7.3). Earlier we were dealing with stationary
gravity ﬂuid ﬂow, that is with ﬂow in which all the hydrodynamic parameters
at each pipeline crosssection remained constant, now we will consider the
general case of nonstationary gravity ﬂow in a descending pipeline section
characterized by the slope angle to the horizontal α (α < 0). The differential
equations describing such ﬂow are
_
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
_
∂ρS
∂t
+
∂ρvS
∂x
= 0
∂ρvS
∂t
+
∂
∂x
_
ρv
2
S +ρg cos α ·
_
S
0
Sdh
_
= −ρgSsinα −
ρgScos α
C
Sh
2
R
h
· vv
Here S(x, t) is the area of the pipeline crosssection ﬁlled by ﬂuid; R
h
(S) the
hydraulic radius of the ﬂow; h(S) the depthof ﬂuid inthe pipeline crosssection;
C
Sh
the Chezy factor (see Section 3.7).
Let v
0
be the ﬂuid velocity at the inlet to the descending section of the
pipeline and ρ
∼
= ρ
0
= const. Let us introduce the following dimensionless
parameters
¯ x =
x
d
;
¯
t = t ·
v
0
d
; ¯ v =
v
v
0
;
¯
S =
S
S
0
=
S
(πd
2
/4)
;
¯
R
h
=
R
h
d
= 0.25 · (1 −sinφ/φ) (see Eq. (3.46));
¯
C
Sh
=
C
Sh
√
g
.
8.4 Pipeline Transportation of Oil Products. Batching 187
With new variables the system of equations describing gravity ﬂuid ﬂow
transforms to
_
¸
¸
¸
¸
¸
¸
¸
¸
_
¸
¸
¸
¸
¸
¸
¸
¸
_
ρ
0
S
0
v
0
d
·
_
∂
¯
S
∂
¯
t
+
∂¯ v
¯
S
∂ ¯ x
_
= 0
ρ
0
v
0
2
S
0
d
·
_
∂¯ v
¯
S
∂
¯
t
+
∂
∂ ¯ x
_
¯ v
2
¯
S +
gd cos α
v
0
2
_
¯
S
0
¯
Sd
¯
h
__
= −ρ
0
gS
0
¯
Ssinα−
ρ
0
g cos α · S
0
v
0
2
gd
·
¯
S · ¯ v¯ v
¯
R
h
¯
C
Sh
or, after simpliﬁcation
_
¸
¸
¸
_
¸
¸
¸
_
∂
¯
S
∂
¯
t
+
∂¯ v
¯
S
∂ ¯ x
= 0
∂¯ v
¯
S
∂
¯
t
+
∂
∂ ¯ x
_
¯ v
2
¯
S +
gd cos α
v
0
2
_
¯
S
0
¯
Sd
¯
h
_
= −
gd sinα
v
0
2
¯
S · −
¯ v¯ v ·
¯
Scos α
¯
R
h
¯
C
Sh
.
(8.6)
It is seen that, in the system of equations (8.6), besides the dimensionless
angle α (
1
= α) there is one more dimensionless complex
2
= gd/v
0
2
or
v
0
/
_
gd. We already met the latter one in Section 7.4; it is the socalled Froude
number
Fr =
v
0
_
gd
. (8.7)
Thus, the Froude similarity criterion (Froude number) is obtained in the
mathematical problem on gravity ﬂuid ﬂow in a pipeline, which was
earlier derived on the basis of general reasoning on the similarity of such
ﬂows.
8.4
Pipeline Transportation of Oil Products. Batching
At the present time light oil products (benzenes, kerosenes, diesel fuels and
others) are pumped by the batching method (Ishmuchamedov et al., 1999).
In oil treatment the factory plants simultaneously produce a plethora of oil
products, mostly the socalled light oil products and, above all, motor oils.
As a result of further compounding (mixing of two or several oil products to
prepare fuels with given properties) there are obtained different sorts of oil
products ready for use. It is evident that the building of a separate pipeline
for each of produced oil products would be unproﬁtable, therefore most of the
oil products are transported by one and the same pipeline, pumping them in
series one after another (batching).
188 8 Dimensionality and Similarity in Mathematical Modeling of Processes
8.4.1
Principle of Oil Product Batching by Direct Contact
The essence of batching by direct contact consists in the different oil products
being combined in separate batches, each of several thousand or even ten
thousand tons, which are pumped into the pipeline in series, one after
another, and transported to the user. In this way each batch displaces the
previous one and is, in turn, displaced by the following batch. It is as if the
oilpipeline along its full length were ﬁlled by different oil products arranged
in a chain and contacting with each other at the places where one batch comes
to an end and another begins. Thus, the key advantage in the batching of oil
products is that different sorts of oil products are pumped not along different
pipes but along one and the same pipe.
However, despite all the advantages of batching it has one signiﬁcant
disadvantage consisting in the formation of a mixture of different oil products
by their mutual displacement in the pipeline. Although the mixing of similar
oil products, for example benzenes of different sorts or diesel fuels of different
sorts, threatens little the quality of the resulting oil product, because oil
products relating to one group of fuels are more compatible than oil products
relating to different groups, the mixing of dissimilar oil products, for example
benzenes, kerosenes and diesel fuels threatens signiﬁcantly the quality of the
oil products. Nevertheless, batching with direct contact of the oil products
has received wide recognition because the quantity of mixture forming in the
contact zones of batches moving in series is relativey small compared to the
large volume of transported fuels and the whole mixture can be decomposed
into the initial oil products, preserving the quality of the latter.
The mixture formed in the contact zone of oil products is caused by physical
processes inherent to ﬂuid ﬂow in the pipeline and to displacment of one oil
product by another. If contacting oil products displaced each other as rigid
bars with plane interface boundaries, mixing in the contacting zone would
of course be absent. However, ﬂuid oil products are not rigid bodies and
mutual displacement happens nonuniformly over the pipeline crosssection.
The velocities of ﬂuid particles at different points of the pipeline crosssection
are distinct. At the pipeline wall they vanish whereas at the pipeline axis they
achieve a maximumvalue. Thus the displacement of one oil product by another
occurs more at the pipeline center than at the pipeline wall. At each instance
of time the wedge behind the moving oil product becomes as if penetrated
into the leading ﬂuid, the penetration happening more when the proﬁle of the
average velocity is more stretched along the pipeline axis. There takes place
socalled convection (or convective diffusion) of the impurity of one oil product
into another one owing to, and together with, ﬂuid layers transferring relative
to each other.
Nonuniformity of the ﬂuid average velocity distribution at the pipeline cross
section is not the only reason for mixture formation of oil products in the
zone of their contact. Light oil products are, as a rule, pumped in a turbulent
8.4 Pipeline Transportation of Oil Products. Batching 189
ﬂow regime in which the ﬂuid particles do not move parallel the pipeline walls
but execute chaotic turbulent motion, as can be seen in smoke jets gushing
from heat plants. In turbulent ﬂows there exists intensive mixing of different
particles over the pipeline crosssection caused by velocity ﬂuctuations and the
chaotic motion of particles. This process is called turbulent diffusion. It mixes
over the pipeline crosssection the edge of the displacing ﬂuid as well as the
rest of the ﬂuid to be displaced, causing their more or less uniformdistribution
in each pipeline crosssection.
Hence, the mixing process of oil products displacing and to be displaced
happens in accordance with the following scheme: the edge of the oil product
moving behind penetrates the oil product moving in front while processes of
turbulent diffusionmix the penetrated impurity over the pipeline crosssection.
Since the concentration of the displacing oil product at the pipeline axis is
greater than at the wall, transport of the displacing oil product into the region
occupied by oil product to be displaced occurs. Conversely, back transport of
the oil product to be displaced into the region of the displacing oil product
also occurs. These processes are inseparable. They operate permanently and
simultaneously over the course of the displacement time, determining thus
the intensity of the longitudinal mixing, the volume and length of the resulting
mixture.
8.4.2
Modeling of Mixture Formation in Oil Product Batching
Light oil products possess the following property: if a volume V
1
of the ﬁrst oil
product is mixed with a volume V
2
of the second oil product, the volume V
c
of
the resulting mixture is, to a high degree of accuracy, equal to the sum of the
volumes of the components V
c
= V
1
+V
2
. Therefore, the additivity property
of ﬂuid volume on mixing of its components is taken as a main assumption in
the construction of the model of mixture formation in a pipeline.
If we denote through ρ
1
and ρ
2
the densities of the contacting oil products,
the volume concentrations θ
1
and θ
2
of the oil products may be expressed
through these quantities and the density ρ
c
of the mixture. In accordance with
the mass conservation law we have
ρ
1
V
1
+ρ
2
V
2
= ρ
c
V
c
.
From this it follows that
ρ
1
V
1
V
c
+ρ
2
V
2
V
c
= ρ
c
⇒ρ
1
θ
1
+ρ
2
θ
2
= ρ
c
.
Taking further that θ
2
= 1 −θ
1
or θ
1
= 1 −θ
2
we obtainthe following formulas
θ
1
=
ρ
c
−ρ
2
ρ
1
−ρ
2
, θ
2
=
ρ
c
−ρ
1
ρ
2
−ρ
1
. (8.8)
190 8 Dimensionality and Similarity in Mathematical Modeling of Processes
We now develop the model of mixture formation in the ﬂuid ﬂow in pipeline.
The mixture of oil products in the contact zone can be characterized by the
concentration θ(x, t) of one of the oil products, for example the displacing one,
in onedimensional ﬂuid ﬂow in a pipeline. The cases θ = 1 or θ = 0 have a
direct relationship to the mixture. In the mixture region 0 < θ < 1. The case
θ = 1 corresponds to the region of displacing oil product, whereas the case
θ = 0 corresponds to the oil product ahead that is to be displaced.
In order to derive a mathematical model of mixture formation it is necessary
to reveal the mass exchange regularities of oil products in the mixture region,
that is to specify the relation between the volumetric ﬂow rate q(x, t) of
the displacing oil product (the volumetric ﬂow rate of the oil product to be
displaced would clearly be equal to v
0
S −q(x, t)) and the parameters of the
concentration distribution θ(x, t) in the ﬂow.
Figure 8.1 represents a scheme of mass exchange inanarbitrary crosssection
of the mixture region. The total ﬂuid ﬂow rate through the crosssection x
in the moving frame of reference is equal to zero, but the transfers of ﬂuid
(mixture of oil products) from left to right and from right to left are nonzero;
they are equal in magnitude but opposite in sign. The ﬂow of the mixture
through the crosssection x from left to right with ﬂow rate w
1
happens mainly
in the central part of the pipeline, while the ﬂow of the mixture in the reverse
direction from right to left with ﬂow rate w
2
= −w
1
occurs chieﬂy close to the
internal surface of the pipeline.
The ﬂow rate w = w
1
= −w
2
is determined by the proﬁle of the velocity ˆ u(r)
averaged over the interval 0 ≤ r ≤ r
∗
w = 2π ·
_
r
∗
0
r · [ ˆ u(r) −v
0
] dr. (8.9)
If we take the velocity proﬁle ˆ u(r) equal to the logarithmic proﬁle in turbulent
ﬂow (Loitzyanskiy, 1987)
ˆ u(r) −u
max
u
∗
=
1
κ
· ln
_
1 −
r
r
0
_
(8.10)
from Eqs. (8.9) and (8.10) ensue the following relations
v
0
= u
max
−4.08 · u
∗
, r
∗
= 0.805 · r
0
, w = 1.26 · u
∗
· S. (8.11)
Here κ
∼
= 0.4 is the Karman constant and u
∗
is the dynamic velocity, see
Section 3.5. The latter is expressed through the tangential frictional stress τ
w

at the pipeline wall
τ
w
 = ρ · u
2
∗
, u
∗
=
_
τ
w

ρ
.
8.4 Pipeline Transportation of Oil Products. Batching 191
Figure 8.1 Scheme of mass exchange in the mixture region.
Since τ
w
 = λ/8 · ρv
0
2
, where λ is the hydraulic resistance factor, the dynamic
velocity is related to this factor by
u
∗
=
_
λ
8
· v
0
.
Insertion of the expression for u
∗
in Eq. (8.11) yields the connection between
the mass exchange ﬂow rate w and the pumping ﬂow rate Q = v
0
S
w = 1.26 ·
_
λ
8
· v
0
S = 0.446 ·
√
λ · Q. (8.12)
From this formula it follows that the quantity exchange ﬂow rates are relatively
not large. So, for example, at λ = 0.022 the quantity w = 0.066 v
0
S, which is
only 6.6% of the pumping ﬂow rate.
Counter ﬂows of ﬂuid transfer both the ﬁrst and second oil product
through the crosssection x of the moving frame of reference, but the average
concentrations θ
and θ
in the transfer ﬂows are different. Thus, the ﬂow rate
q(x, t) of the displacing oil product through the crosssection x is given by
q(x, t) = w · θ
−w · θ
= w · (θ
−θ
).
In the ﬁrst ﬂux (from left to right) the concentration θ
is equal to the
concentration of the displacing oil product averaged over the crosssection at
some distance l
1
behind the crosssection x. In the second ﬂux (from right
to left) the concentration θ
is equal to the concentration of the displacing oil
product at some distance l
2
ahead of the crosssection x. Lengths l
1
and l
2
can
be called mixing lengths, since they are equal to the lengths over which the
turbulent diffusionmixes the invading impurity over the pipeline crosssection.
Accurate to small quantities of the highest order it could be written as
_
¸
¸
_
¸
¸
_
θ
= θ(x −l
1
, t)
∼
= θ(x, t) −l
1
·
∂θ
∂x
+. . . ,
θ
= θ(x +l
2
, t)
∼
= θ(x, t) +l
2
·
∂θ
∂x
+. . .
(8.13)
192 8 Dimensionality and Similarity in Mathematical Modeling of Processes
and
q(x, t) = w · θ
−w · θ
= w · (θ
−θ
) = −w · (l
1
+l
2
) ·
∂θ
∂x
.
Substitution of w from (8.12) in the latter relation yields
q(x, t) = −0.446 ·
√
λ · (l
1
+l
2
) ·
∂θ
∂x
· v
0
S
or
q(x, t) = −K ·
∂θ
∂x
· S (8.14)
where
K = 0.446 ·
√
λ · (l
1
+l
2
) · v
0
. (8.15)
The relation (8.14) expressing the proportionality of the volumetric ﬂow rate
q(x, t) of the displacing oil product gradient ∂θ/∂x to its concentration is
called the law of longitudinal mixing and the factor K (m
2
s
−1
) the effective factor
of the longitudinal mixing. The minus sign in Eq. (8.14) shows that the ﬂux
of each oil product is directed from the higher concentration to the lower
one, that is opposite to the concentration gradient of the oil product under
consideration.
There are many theoretical and experimental formulas for the factor K of
longitudinal mixing (Ishmuchamedov et al., 1999). We consider one of them,
namely the formula derived by Taylor when investigating the dispersion of an
impurity in turbulent ﬂuid ﬂow in a pipe
K = 1.785 ·
√
λ · v
0
d. (8.16)
Comparing this formula with Eq. (8.15), we ﬁnd for the sum of the mixing
lengths (l
1
+l
2
) the value 4d. This result appears to be true for Reynolds
numbers higher than 3 · 10
4
.
8.4.3
Equation of Longitudinal Mixing
To derive a mathematical model describing the process of mixture formation
in a pipeline for displacement of one ﬂuid by another one we use the
volume balance equation for each component. Since the sum of the volume
concentrations θ
1
and θ
2
of oil products is equal to 1, it is enough to write only
one balance equation for one of the components, for example the displacing
component, taking θ
1
= θ(x, t), θ
2
= 1 −θ(x, t).
8.4 Pipeline Transportation of Oil Products. Batching 193
Figure 8.2 Derivation of the volume
balance equation.
Consider a ﬂuid volume (see Section 1.2) enclosed between crosssections
x
1
(t) and x
2
(t), Fig. 8.2.
The ﬁrst (displacing) oil product occupies only a part of this volume V
1
which can be written as
V
1
=
_
x
2
(t)
x
1
(t)
θ(x, t) · Sdx.
If the exchange of oil products through the crosssections x
1
(t) and x
2
(t)
were absent, the quantity V
1
would be constant, but in reality it is not. The
quantity V
1
varies owing to mutual penetration of the oil products into each
other and this variation is determined by the difference in transfer ﬂow rates
q(x
1
, t) −q(x
2
, t). Thus
dV
1
dt
=
d
dt
__
x
2
(t)
x
1
(t)
θ(x, t) · Sdx
_
= q(x
1
, t) −q(x
2
, t). (8.17)
If we take into account the following identities:
d
dt
__
x
2
(t)
x
1
(t)
θ(x, t) · Sdx
_
=
_
x
2
(t)
x
1
(t)
_
∂θS
∂t
+
∂v
0
θS
∂x
_
dx, see Eq. (1.4),
q(x
1
, t) −q(x
2
, t) = −
_
x
2
(t)
x
1
(t)
∂q
∂x
dx,
the relation (8.17) may be written as
_
x
2
(t)
x
1
(t)
_
∂θS
∂t
+
∂v
0
θS
∂x
_
dx = −
_
x
2
(t)
x
1
(t)
∂q
∂x
dx.
If, in addition, we recall that the ﬂuid volume under consideration is arbitrarily
chosen, that is the integration limits x
1
(t) and x
2
(t) are arbitrarily chosen, then
from the latter integral equation follows the differential equation
∂θS
∂t
+
∂v
0
θS
∂x
= −
∂q
∂x
expressing the volume balance of the ﬁrst oil product in the mixture.
Insertion instead of q, using its expression through the concentration
gradient q(x, t) = −K · S · ∂θ/∂x (see Eq. (8.14)) with regard to the conditions
S = const. and v
0
= const. yields the differential equation for the concentration
194 8 Dimensionality and Similarity in Mathematical Modeling of Processes
θ(x, t) of the displacing oil product in the mixture
∂θ
∂t
+v
0
∂θ
∂x
= −K ·
∂
2
θ
∂x
2
(8.18)
Equation (8.18) represents the differential equation of longitudinal mixing
referring to the class of heat conduction equations (Ishmuchamedov et al., 1999).
If we introduce the dimensionless variables ¯ x = x/L and
¯
t = t/(L/v
0
),
Eq. (8.18) can be rewritten in dimensionless form
1
(L/v
0
)
·
∂θ
∂
¯
t
+
v
0
L
·
∂θ
∂ ¯ x
=
K
L
2
·
∂
2
θ
∂ ¯ x
2
or
∂θ
∂
¯
t
+
∂θ
∂ ¯ x
= Pe
−1
·
∂
2
θ
∂ ¯ x
2
(8.19)
where the dimensionless parameter Pe = v
0
L/K called the Peclet number is the
main characteristic of quantity of mixture forming in the course of oil product
pumping and the similarity criterion in problems on longitudinal mixing of
ﬂuids by their displacement in the pipe.
8.4.4
SelfSimilar Solutions
The reasoning of dimensional theory can sometimes bring very important
results, being capable of ﬁnding solutions of differential equations reﬂecting
the most important peculiarities of processes and being in some sense limiting
for solutions, taking into account the minor peculiarities and details of
processes. To such solutions belong the socalled selfsimilar solutions. Let us
demonstrate these with the example of Eq. (8.18).
In problems on the displacement of one oil product by another we can
consider the dimensional parameters x, t, K, v
0
, L. However, the number of
these parameters could be reduced without serious consequences for the
physics of the processes. If for example we change to a frame of reference
x
∗
= x −v
0
t moving with average velocity v
0
of the pumping ﬂuid, then
Eq. (8.18) would be simpliﬁed to
∂θ
∂t
= K ·
∂
2
θ
∂x
∗
2
(8.20)
If the boundary conditions at the pipeline section edges have become, after
frame of reference transformation, movable x
∗
= −v
0
t and x
∗
= L −v
0
t, but
far from the region occupied by the mixture they can be replaced by conditions
at ‘‘inﬁnitiy’’ (±∞), that is θ →0 when x →+∞and θ →1 when x →−∞
8.4 Pipeline Transportation of Oil Products. Batching 195
and the initial conditions at t = 0 we can take as instantaneous changing of oil
products, namely θ(x
∗
, 0) = 1 at x
∗
< 0 and θ(x
∗
, 0) = 0 at x
∗
> 0 or
θ(x
∗
, 0) =
_
1, x
∗
< 0,
0, x
∗
> 0.
(8.21)
Then, in the problem under consideration, there remain only three
dimensional quantities x, t, K. Since the solution of the problem is the
dimensional function θ, depending on three dimensionallyindependent
parameters from which may be built only one dimensionless combination then
ξ =
x
∗
√
Kt
(8.22)
in accordance with the theorem the solution should depend on one variable.
Thus, the solution of the problem (8.20) and (8.21) has to be sought in the
form θ(x
∗
, t) = θ(ξ).
Such a solution is called selfsimilar because it is as if it transforms itself
in space in a similar way, namely the solution of the problem at an arbitrary
instant of time t may be obtained from the solution of the problem at any
previous instant of time t
1
by stretching the distribution θ(x
∗
, t
1
) along the
x
∗
axis with factor
√
K · t/t
1
, because it is well known that the multiplication of
the function argument by a certain number brings extension or compression
of the function graph along the abscissa axis.
Substitution of θ(x
∗
, t) = θ(x
∗
/
√
Kt) in Eq. (8.20) leads to an ordinary
differential equation of the second order
−
ξ
2
·
dθ
dξ
=
d
2
θ
dξ
2
from which follows
θ(ξ) = A ·
_
ξ
0
e
−α
2/4
dα +B
where A and B are constants of integration.
Using the boundary conditions θ →0 at ξ →+∞ and θ →1 at ξ →−∞
gives A = 1/(2
√
π), B = 1/2. In deriving the latter we use the known equality
_
∞
0
e
−α
2
dα =
√
π
2
.
The solution of the problem is
θ(ξ) =
1
2
_
1 −
2
√
π
·
_
ξ/2
0
e
−α
2
dα
_
196 8 Dimensionality and Similarity in Mathematical Modeling of Processes
or in dimensional variables
θ(x
∗
, t) =
1
2
_
1 −
2
√
π
_
x
∗
/
√
4Kt
0
e
−α
2
dα
_
. (8.23)
Graphs of the concentration distribution θ(x
•
, t) in the moving frame of
reference are shown in Fig. 8.3. The heavy line depicts the initial distribution
whereas other lines show the concentration distribution at successively
increased instants of time.
A remark on the imperfection of the model. It can be seen at once that the
obtained solution has a defect. This is that the mixture, which initially was
absent, at just the next instant of time would propagate through the whole
pipeline length. In fact such a case could not occur in practice and the obtained
paradox is the result of imperfectness of the model. The dispersion model of
longitudinal mixing, as in general all models of this kind, represents only a
certain schematization, in the given case it is the formation of a mixture in
the contact zone of the displacing ﬂuid and the ﬂuid to be displaced. But the
result of such a schematization does not give a particularly bad result. The
function θ(x, t) tends very quickly to zero at x →∞and to one at x →−∞.
In the main domain of variability this function approximates the concentra
tion distribution in the mixture zone well. Therefore, the dispersion model
of longitudinal mixing based on Eq. (8.20) has received wide application in
calculations of mixtures forming in the batching of oil products.
Mixture volume. The volume V
c
of a mixture of pumping oil products forming
in a pipeline at the instant of time t calculated by Eq. (8.23) in the range of
concentrations 0.01 < θ < 0.99 is determined by the expression
V
c
∼
= 6.58 ·
πd
2
4
·
√
K · t. (8.24)
A remark on the quantity of mixture volume. Taylor obtained Eq. (8.16) for the
factor K of longitudinal mixing under the assumption that the densities
and viscosities of contacting ﬂuids are close to each other. Therefore,
strictly speaking, this equation is not suitable for the case when the
densities and viscosities of contacting ﬂuids differ signiﬁcantly from each
other. Such situations occur in the pumping of oil products, for example
Figure 8.3 Selfsimilar distribution of concentration.
8.4 Pipeline Transportation of Oil Products. Batching 197
when pumping benzenes (ρ ≈ 730–750 kg m
−3
, ν ≈ 0.6 cSt) with diesel fuels
(ρ ≈ 830–850 kg m
−3
, ν ≈ 4–9 cSt). It is evident that for these ﬂuids the
factors of hydraulic resistance λ are also different.
The formula for the mixture volume V
c
as applied to the case under
consideration can be improved if we calculate the mixture volume as the
arithmetical mean of two volumes: the ﬁrst calculated by Eq. (8.24) with
the factor K
1
= 1.785 ·
√
λ
1
· v
0
d and the second by Eq. (8.24) with the factor
K
2
= 1.785 ·
√
λ
2
· v
0
d. It is equivalent to formula (8.24) if we take
K = 0.446 ·
_
4
_
λ
1
+
4
_
λ
2
_
2
· v
0
d. (8.25)
Exercise. It is required to calculate the length and the volume of the
mixture region in a symmetric concentration range 0.01 < θ < 0.99 when
batching benzene (ν
B
= 0.6 cSt) and diesel fuel (ν
D
= 6 cSt) in an oilpipeline
(D = 530 ×8 mm, = 0.15 mm, L = 700 km) when the transportation of oil
products occurs with ﬂow rate Q = 1000 m
3
h
−1
.
Solution. Determine the average transportation velocity v
0
of the oil products
v
0
=
4Q
πd
2
=
4 · 1000/3600
3.14 · (0.530 −2 · 0.008)
2
∼
= 1.34 m s
−1
.
Calculate the Reynolds numbers
Re
B
=
v
0
d
ν
B
=
1.34 · 0.514
0.6 · 10
−6
∼
= 1 147 933,
Re
D
=
v
0
d
ν
D
=
1.34 · 0.514
6 · 10
−6
∼
= 114 793.
Calculate the hydraulic resistance factors λ
B
and λ
D
λ
B
= 0.11 ·
_
0.15
514
+
68
1 147 933
_
0.25
∼
= 0.015,
λ
D
= 0.11 ·
_
0.15
514
+
68
114 793
_
0.25
∼
= 0.019.
Calculate with Eq. (8.25) the factor K of longitudinal mixing
K = 0.446 ·
_
4
√
0.015 +
4
√
0.019
_
2
· 1.34 · 0.514
∼
= 0.222 m
2
s
−1
.
With Eq. (8.24) calculate the mixture volume V
c
V
c
∼
= 6.58 ·
πd
2
4
·
√
K · t = 6.58 ·
πd
2
4
·
_
K ·
L
v
0
= 6.58 ·
3.14 · 0.514
2
4
·
_
0.222 ·
700 000
1.34
∼
= 465 m
3
.
198 8 Dimensionality and Similarity in Mathematical Modeling of Processes
The length of the mixture region l
c
is
l
c
=
V
c
πd
2
/4
=
465
3.14 · 0.514
2
/4
∼
= 2242 m or 2.242 km.
Answer: 2242 m; 465 m
3
.
It should be noted that nonselfsimilar solutions of the considered problem,
that is solutions taking into account the ﬁnite extent of the pipeline and the
conditions at its edges, differ slightly fromthe selfsimilar solution. The facility
to formulate the problem, making it selfsimilar and yet without changing its
most important features, characterizes the high skill of the scientist. That is
why selfsimilar solutions play such an important role in the different ﬁelds of
science (Sedov, 1965).
199
References
Archangelskiy V.A., (1947) Calculation of
NonStationary Flow in Open Water
Courses, Academy of Sciences USSR,
Moscow (in Russian).
Charniy I.A., (1975) NonStationary Motion
of Real Fluid in Pipes, 2nd edn., Nedra,
Moscow (in Russian).
Christianowitch S.A., (1938)
NonStationary Flow in Channels and
Rivers, Collected articles on Some Problems
in Continuum Mechanics, Academy of
Sciences USSR, Moscow (in Russian).
Dodge D.W., Metzner A.B., (1958)
Turbulent Flow of NonNewtonian
Systems, AIChE J., 2, 189–204 .
Ginsburg I.P., (1958) Applied
HydroGasDynamics, LGU, Leningrad
(in Russian).
Ishmuchamedov I.T., Isaev S.L., Lurie
M.V., Makarov C.P., (1999) Pipeline
Transportation of Oil Products, Oil and
Gas, Moscow (in Russian).
Leibenson L.S., Vilker D.S., Shumilov P.P.,
Yablonskiy V.S., (1934) Hydraulics, 2nd
edn., Gosgorgeolnephteizdat, Moscow,
Leningrad (in Russian).
Loitzyanskiy L.G., (1987) Mechanics of Fluid
and Gas, Nauka, Moscow (in Russian).
Lurie M.V., (2001) Technique of Scientiﬁc
Researches. Dimensionality, Similarity
and Simulation of Phenomena in
Problems of Oil Transportation and Oil
and GasStorage, Oil and Gas, Moscow
(in Russian).
Lurie M.V., Polyanskaya L.V., (2000) About
One Dangerous Source of Hydraulic
Shock Waves in Oil and Oil Products, Oil
Facilities, No. 8 (in Russian).
Lurie M.V., Podoba N.A., (1984)
Modiﬁcation of Karman Theory for
Turbulent Shear Flows, Papers of the
Academy of Science of the USSR, 279(3),
(in Russian).
Vasil’ev G.G., Korobkov G.E., Lurie M.V.
et al., (2002) Pipeline Transportation of
Oil, Vol.1, S.M. Veinstock, Nedra (in
Russian).
Polyanscaya L.V., (1965) Investigation of
nonstationary processes in changing
operation regime with centrifugal
pumps, Kand. Sci. Thesis, Gubkin Oil &
Gas Institute, Moscow (in Russian).
Porshakov Yu.P., Kosachenko A.N.,
Nikishin V.I., (2001) Power of Pipeline
Gas Transportation, Oil and Gas, Moscow
(in Russian).
Potapov A.G., (1975) Hydraulic Resistance
Factor in Turbulent Flow of ViscousPlastic
Fluids, Volgogradnipinepht, Volgograd,
No. 23 (in Russian).
Rozhdestvenskiy B.L., Yanenko N.N.,
(1977) Systems of QuasiLinear Differential
Equations, Nauka, Moscow (in Russian).
Romanova N.A., (1985) Laminar and
Turbulent Flows in Pipes and Channels
with Moving Walls, Kand. Sci. Thesis,
Gubkin Oil & Gas Institute, Moscow (in
Russian).
Samarskiy A.A., (1977) Introduction to the
Theory of Difference Schemes, Nauka,
Moscow (in Russian).
Sedov L.I., (1965) Methods of Similarity and
Dimensionality in Mechanics, Nauka,
Moscow (in Russian).
Tihonov A.N., Samarskiy A.A., (1966)
Equations of Mathematical Physics,
Nauka, Moscow (in Russian).
Wilkenson U.L., (1960) NonNewtonian
Fluids, Pergamon Press, London,
NewYork.
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
201
Appendices
Appendix A. Increment and Differential of a Function
In some sections of this book the formula for the increment f of the
differentiable f (x)
f (x +x) −f (x) ≈
df
dx
x
· x
where x is the increment of the argument. Let us explain this formula.
Figure A.1 shows a graph of a differentiable and, consequently, contin
uous function y = f (x). The difference f (x +x) −f (x) of the values of
this function at two points x and (x +x) is called the increment of the
function f .
If at the point M on the abscissa x we draw a tangent to the plot of the
function, the increment of the function can be interpreted as the sum of the
two segments AB and BC, Fig. A2.
Since
df
dx
= lim
x→0
f (x +x) −f (x)
x
this means that
df
dx
x
−
f (x +x) −f (x)
x
= ε →0 at x →0
that is, ε is an inﬁnitesimal quantity. From this it follows that the increment
f = (x +x) −f (x) of a differentiable function f (x) can be represented as
f (x +x) −f (x) =
df
dx
x
· x
AB
+ ε · x
BC
.
Since the value of the derivative at the point x is equal to the slope of the
function plot at this point to the horizontal, it is evident that the ﬁrst term on
the righthand side of the last formula is nothing but the length of the segment
AB. If df / dx = 0, it appears to be a quantity of the same order as x.
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
202 Appendices
Figure A.1 Increment f and differential
df of a function.
The second term (ε · x) is depicted in Fig. A.1 by segment BC. This term is
a smaller quantity than the ﬁrst term, because ε →0 at x →0. This means
that the smaller x then the greater the accuracy when neglecting the second
term. Thus, there is an approximation formula
f (x +x) −f (x) ≈
df
dx
x
· x.
It is commonly said that this formula is true up to an inﬁnitesimal quantity of
higher order.
The principal linear (in x ≡ dx) part
df
dx
x
· x of the function increment
f is called the function differential df
df =
df
dx
x
· dx; f = df +ε · dx.
The differential of function df is plotted by segment AA, Fig. A.1.
Appendix B. Proof of the Formula
d
dt
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx =
x
2
(t)
x
1
(t)
∂
∂t
[A(x, t) · S(x, t)] dx
+A(x, t) · S(x, t)
x
2
(t)
·
dx
2
dt
−A(x, t) · S(x, t)
x
1
(t)
·
dx
1
dt
to calculate the total derivative with respect to time of an integral with integration
limits determining the motion of a ﬂuid (individual) volume of a continuum in the
pipeline.
Proof. Consider two successive locations at the time instances t and t +t
of one and the same ﬂuid volume in a pipeline, Fig. B.1.
In accordance with general deﬁnition of a derivative as a limit of the ratio of
a function increment to an argument increment when the latter tends to zero
Appendix B. Proof of the Formula 203
Figure B.1 Derivation of the formula of time differentiation of an integral with time
dependent limits.
we have
d
dt
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx
= lim
t→0
1
t
x
2
(t+t)
x
1
(t+t)
A(x, t +t) · S(x, t +t) dx
−
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx
.
The integrals in square brackets can be represented as follows (see Fig. B.1)
x
2
(t+t)
x
1
(t+t)
A(x, t +t) · S(x, t +t) dx
=
x
2
(t)
x
1
(t+t)
A(x, t +t) · S(x, t +t) dx
+
x
2
(t+t)
x
2
(t)
A(x, t +t) · S(x, t +t) dx
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx
=
x
1
(t+t)
x
1
(t)
A(x, t) · S(x, t) dx +
x
2
(t+t)
x
1
(t+t)
A(x, t) · S(x, t) dx.
Consequently
x
2
(t+t)
x
1
(t+t)
A(x, t +t) · S(x, t +t) dx −
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx
=
x
2
(t)
x
1
(t+t)
[A(x, t +t) · S(x, t +t) −A(x, t) · S(x, t)] dx
+
x
2
(t+t)
x
2
(t)
A(x, t +t) · S(x, t +t) dx
−
x
1
(t+t)
x
1
(t)
A(x, t) · S(x, t) dx.
204 Appendices
Up to inﬁnitesimal quantities of higher order this can be written as
x
2
(t)
x
1
(t+t)
[A(x, t +t) · S(x, t +t) −A(x, t) · S(x, t)] dx
∼
=
x
2
(t)
x
1
(t+t)
∂(A · S)
∂t
· t · dx
x
2
(t+t)
x
2
(t)
A(x, t +t) · S(x, t +t) dx
∼
= A(x
2
, t) · S(x
2
, t) ·
dx
2
dt
· t,
x
1
(t+t)
x
1
(t)
A(x, t) · S(x, t) dx
∼
= A(x
1
, t) · S(x
1
, t) ·
dx
1
dt
· t.
Dividing the sum of the latter three expressions by t and passing to the limit
at t →0 we get the formula
d
dt
x
2
(t)
x
1
(t)
A(x, t) · S(x, t) dx =
x
2
(t)
x
1
(t)
∂
∂t
[A(x, t) · S(x, t)] dx
+A(x, t) · S(x, t)
x
2
(t)
·
dx
2
dt
−A(x, t) · S(x, t)
x
1
(t)
·
dx
1
dt
.
When going to the limit we took into account the continuity prop
erty of functions x
1
(t) and x
2
(t), namely lim
t→0
x
1
(t +t) = x
1
(t) and
lim
t→0
x
2
(t +t) = x
2
(t).
205
Author Index
a
Archangelskiy V.A. 67, 151
c
Charniy I.A. 133
Christianowitch S.A. 67, 151
d
Dodge D.W. 62
g
Ginsburg I.P. 21
i
Ishmuchamedov I.T. 63, 65, 187,
192, 194
j
Joukowski N.E. 42, 93
l
Leibenson L.S. 20, 66, 135, 151
Loitzianskiy L.G. 58, 59, 190
Lurie M.V. 19, 54, 56, 57, 66, 153,
154,
m
Metzner A.B. 61
p
Podoba N.A. 53, 56, 57
Polyanskaya L.V. 153, 154
Porshakov B.P. 40
Potapov A.G. 62
r
Romanova N.A. 51, 62
Rozhdestvenskiy B.L. 150
s
Samarskiy A.A. 114, 155
Sedov L.I. 3, 157, 198
t
Tichonov A.N. 114
v
Vasil’ev G.G. 77
w
Wilkinson U.L. 36, 48, 49, 51
y
Yanenko N.N. 150
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
207
Subject Index
a
absolute roughness 19
acceleration 4
adiabatic
–expansion of gas 146
–index 93
–velocity of sound in gas 141
air density 4
Altshuler formula 21
analytical solution 88
angular velocity of impeller rotation
77
antiturbulent additive 62
apparent viscosity 36
area of pipeline crosssection 7
average
–physical parameter 4
–tangential stress 52
–velocity 52
b
balance equation of forces 76
barotropic
–gas 15
–ﬂuid 15
–medium 13
basic measurement units, see
primary measurement units
batching 187
Bernoulli equation 15
Blasius formula 21
boundary condition 46
Boussinesq force 150
bringing of model equations to
dimensionless form 183
Buckingham theorem 164
c
calorimetric dependences 30
capillary viscometers 49
Cauchy problem, see initial value problem
central theorem of dimensional theory
163
centrifugal
–blower 98
–force 76
–pump 76
change of total energy 22
characteristic equation 118
characteristics of wave equation 119
Chezy factor 67
circular pipe 45
Clapeyron law 28
Clapeyron equation 39
closed mathematical model 2
closed mathematical model of
onedimensional
–nonstationary ﬂows of ﬂuid and
gas in pipelines 109
closing relations 30
combined operation of linear pipeline
section and
–pumping station 81
commercial ﬂow rate of gas 96
compatibility condition 119
–at characteristics 142
compressibility 3
–factor 34
compressible medium 28
compression wave 152
concentration of antiturbulent additive
63
compressor 16
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie
Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim
ISBN: 9783527408337
208 Subject Index
concrete mathematical model 2
conditions at discontinuities (jumps) of
hydrodynamic
–parameters 128
conditions of transition from laminar to
turbulent ﬂow 51
conjugation conditions 124
conservation law of transported medium
mass 7
continuity equation 7
continuum 3
convective diffusion 188
Coriolis factor 6
correctness of the model 2
criteria of
–dynamic similarity 178
–geometric similarity 178
–transition from laminar to turbulent
ﬂow 51
critical
–depth 151
–isotherm 38
–point of gas 38
–pressure 40
–Reynolds number 52
–temperature 40
–velocity 51
d
D’alambert formulas 117
damping of pressure bow shock at
wave front of hydraulic shock 133
DarcyWeisbach
–law 48
–relation 67
density of
–internal energy 6
–medium 5
–transported 7
dependences characterizing internal
structure of medium ﬂow 30
derived measurement units,
see secondary measurement units
descending sections of pipeline 68
developed turbulent ﬂow 52
differential
–equation of longitudinal mixing
194
–head 17
–of pump 180
differentiation of integral quantity
regarding ﬂuid volume 7
dilatant ﬂuid 37
dimensional
–formula 159
–quantities 158
dimensionality of quantities 159
dimensionless quantities 157
dimensional theory 157
dimensionaldependent quantities 164
dimensionalindependent quantities 164
direct hydraulic shock 131
discharge line 17
discontinuity of
–ﬂow rate 126
–velocity 126
dissipation of mechanical energy 15
divergence of vector 144
divergent form of differential equations
143
drag force 4
dynamic pressure 8
dynamic
–velocity 55
–viscosity 33
e
effective factor of longitudinal mixing
192
efﬁciency of the model 2
elastic
–modulus 34
–spring 1
elasticity 3
elementary surface 32
energy equation 24
energy of material point system 6
enforced ﬂow 67
enthalpy 25
equation for
–head before oilpumping stations
84
–variation of pressure jump 133
equation of
–ﬂuid motion 9
–heat inﬂow 24
–internal motion kinetic energy change
17
–mechanical energy balance 11
–medium state 29
–pipeline state 30
–state 30
–total energy 29
Euclidian geometry 2
Euler function 90
Eulerian derivative, see partial derivative
Subject Index 209
external
–force 9
–inﬂow of heat 22
–sources of mechanical energy
24
f
factor of
–dynamic viscosity 33
–kinematic viscosity 33
–volume expansion 35
feed 76
ﬁlling degree of gravity ﬂow 151
ﬁrst law of thermodynamics 22
ﬂow
–in hydraulic smooth pipe 21
–core 50
–curve 36
ﬂuid particle of transported medium
7
ﬂuid volume 193
force of dry friction 2
force of viscous friction 2
formation of oil product mixtures in
contact zones 188
friction
–factor 20
–force 13
Froude number 177
fundamental laws of continuum
physics 4
Funning factor 20
g
gas
–compression 100
–compressor station (GCS) 100
–constant 28
–enthalpy 105
–isotherm 38
–outﬂow from a pipeline 146
–speciﬁc capacity at constant
pressure 28
gaspumping aggregates (GPA) 98
gate valve 127
–closing 153
generalized
–Reynolds number 48
–theory of hydraulic shock 154
geometric
–head 14
–modeling 173
–similarity 174
governing factors 3
gravity
–acceleration 4
–ﬂow 66
–of incompressible ﬂuid in a pipeline
186
–force 9
–stratiﬁed ﬂow 66
h
head 75
–balance equation 81
–before pump 81
–before pumping station 80
headdischarge (Q −H) characteristic of
pump 76
heat
–conduction equation 194
–conductivity 3
–energy inﬂow 23
–exchange between transported medium
and environment 30
–ﬂux 23
heattransfer factor 26
height of pipeline axis above sea level 5
Hedstroem number 62
homogeneous ﬂuid 15
homogeneous incompressible ﬂuid 15
Hooke law of elasticity 43
hydraulic
–dependence 30
–gradient 15
–hammer 42
–losses 18
–radius 66
–resistance factor 20
–resistance in laminar ﬂow of viscous
incompressible
–ﬂuid in circular pipe 47
–shock 127
–in industrial pipeline caused by
instantaneous
–closing of gate valve 135
–wave velocity 130
–smooth pipe 21
–smooth surface 21
hydrodynamic stability 51
hyperbolic differential equations 139
hypothesis of quasistationarity 109
i
Ilyushin number 51
impeller 76
210 Subject Index
incident wave 134
incompressible
–ﬂuid 8
–medium 18
independent dimensionless
combinations 4
individual
–derivative 11
–particle of continuum 6
inertial properties of ﬂuid in non
stationary processes 112
initial and boundary conditions
124
initial value (Cauchy) problem 2
input of external energy 16
integral characteristics of ﬂuid
volume 5
integral with variable integration
limits 13
interaction of waves in pipeline
section 120
intermittency factor 21
internal energy 23
–of ﬂuid volume 6
international system of
measurement units SI
158
invariant form 4
–of equations 153
isotherm of real gas 38
iteration method 81
j
Joukowski formula 93
Joule–Thompson
–effect 94
–factor 93
jump of parameter at discontinuity
front 128
k
Karman
–constant 54
–formula 54
–model 61
kinematic
–consistency 62
–viscosity 19
kinetic energy 11
–of ﬂuid volume 6
–of internal motion relative mass
centre 12
–of particle mass centre 12
kinetic head 14
l
Lagrangian derivative, see individual
derivative
laminar ﬂow
–of nonNewtonian Ostwald power ﬂuid
in circular pipe 47
–of viscous ﬂuid 45
–of viscousplastic ﬂuid in circular pipe
49
–regime 12
law of
–longitudinal mixing 192
–momentum change 9
–total energy conservation 22
limit shear stress 37
line of hydraulic gradient 16
local derivative with respect to time 6
longitudinal mixing 189
loss of head in DarcyVeisbach form 20
m
Mach number 185
mass 4
–conservation law 7
mass ﬂow rate 5
–of ﬂuid volume 5
–of material point system 5
mathematical model 1
–of centrifugal blower operating in
stationary regime 101
–of mixture formation 190
–of nonstationary gravity ﬂuid ﬂow
149
–of pump 75
–of slightly compressible ﬂuid 109
mathematical modeling 1
mean asperity height of pipeline internal
surface 19
mean ﬂow rate velocity 46
measurement units 157
mechanical energy
–balance equation 11
–energy change law 14
medium macroscopic volume 3
mesh cell 144
method of
–characteristics 121
–successive approximation, see iteration
method
mixture volume 146
model 1
model of
–ﬂuid baric and heat expansion 34
Subject Index 211
–continuum 3
–elastic deformable pipeline 42
–elastic ﬂuid 34
–ﬂuid 5
–with heat expansion 34
–gaseous continuum 31
–ideal ﬂuid 32
–homogeneous ﬂuid 34
–incompressible ﬂuid 34
–material point 1
–medium 31
–nondeformable pipeline 42
–nonNewtonian ﬂuid 36
–nonstationary gas ﬂow in pipeline
112
–nonstationary isothermal ﬂow of
slightly compressible
–ﬂuid in pipeline 109
–perfect gas 39
–power Ostwald ﬂuid 36
–real ﬂuid 5
–real gas 5
–ShvedovBingham ﬂuid 37
–valve 127
–viscous ﬂuid 32
modeling of
–blower operation 100
–ﬂuid outﬂow from tank 178
–gravity ﬂuid ﬂow 176
–mixture formation in oil product
batching 188
–stationary ﬂow of compressible gas
in gaspipeline 92
–stationary ﬂow of viscous
incompressible ﬂuid in a pipe 175
molecular weight 39
momentum
–equation 29
–of ﬂuid volume 6
–of material point system 6
motion equation 29
movable
–ﬂuid volume of continuum 5
–individual ﬂuid volume 5
multistep compression 101
mutual displacement of oil products
188
n
nearwall turbulence 63
negative slope characteristic 131
Newton formula 25
NewtonLeibniz formula 7
Newtonian viscous ﬂuid 36
nonisothermal ﬂuid ﬂow 89
nonNewtonian ﬂuid 36
nonstationary
–ﬂuid ﬂow with ﬂow discontinuities
152
–nonisothermal gas ﬂow in
gaspipelines 138
normal depth 151
o
oil
–product batching by direct contact
188
–pumping station 75
onedimensional
–ﬂow 1
–model 45
–nonstationary ﬂow of slightly
compressible ﬂuid
–in pipeline 184
–theory 12
operation of pipeline with intermediate
oilpumping stations 84
origination of similarity criteria in
equations of mathematical
–model 183
oscillation of small load 1
Ostwald rheological law 61
overcompressibility factor 40
–for natural gas 40
p
parabolic (Q −H)characteristic of
centrifugal pump 181
parallel connection of pumps 79
parameter of pipeline internal surface
smoothness 21
partial derivative 10
Peclet number 194
perfect gas 28
phase of
–direct hydraulic shock 131
–reﬂected wave 132
phenomenological Karman theory 54
physical modeling 173
piezometric head 14
pipeline
–proﬁle 14
–transportation of oil products 187
piston 100
–engine 100
212 Subject Index
planedeformable state 43
planestressstate 43
point mass 1
Poiseuille
–formula 46
–law 48
Poisson ratio 43
positive slope characteristic 115
power of
–external mechanical devices 13
–gravity force 13
–internal friction forces 13
–pressure force 13
pressure 5
–jump 128
–wave 127
primary measurement units 158
problem on
–disintegration of arbitrary
discontinuity 155
–oil and gas transportation 4
–traveling waves 118
–wave interaction in limited pipeline
section 119
proﬁle
–blades 76
–hydraulic shock 153
propagation of waves in
–bounded pipeline section 119
–inﬁnite pipeline 115
–semiinﬁnite pipeline 117
protection of pipeline from hydraulic
shock 138
pseudoplastic ﬂuid 36
pump 16
–efﬁciency 17
–power 17
pumptopump regime 84
pumped ﬂow, see enforced ﬂow
pumping
–plant 16
–pressure 76
pumps connected in series 78
q
quasilinear differential equations of
hyperbolic type 150
r
rarefaction wave 152
rate of
–elementary volume change 13
–internal energy change 25
reaction force 9
rectangular mesh 143
reduced
–pressure 40
–temperature 40
–universal characteristic of centrifugal
blower 103
reﬂected wave at the joint of two pipe
sections 134
relative roughness 66
–of pipeline inner surface 19
relativistic effects of the relativity theory 2
restoring force 1
Reynolds
–number 19
–stresses 52
ReynoldsFilonov formula 33
rheological properties 49
rheology 30
Riemann invariants 115
rootmeansquare (rms) value 12
rule of integral quantity differentiation with
respect to time 6
s
safety valve 138
schematization of
–initial conditions 2
–onedimensional ﬂows of ﬂuids and
gases in pipelines 4
–phenomenon 2
second Newtonian law 2
secondary measurement units 158
separatrix 99
shear rate 36
shear stress 5
shock front 131
Shuchov formula 26
similarity 173
–criteria 174
–of operation of centrifugal pumps
179
–factor 173
–theory 157
slightly compressible ﬂuid 24
slug ﬂow 68
saturated vapor tension (pressure) 68
selfsimilar
–distribution of concentration 196
–solutions 194
small perturbations 142
space coordinate measured along
pipeline 4
Subject Index 213
speciﬁc
–mechanical energy dissipation 19
–volume 38
speed of wave propagation in pipeline
110
square ﬂow 21
stability of laminar ﬂuid ﬂow 51
stationary
–ﬂow 8
–of barotropic medium 15
–operation regime of
–hightemperature pumping
87
–gaspipeline together with
compressor station 98
steady motion 4
stepwise variations (jumps) of
hydrodynamic ﬂow parameter
128
sticking condition 46
Stokes formula 20
string elasticity factor 1
Strouchal number 185
suction
–line 16
–pressure 76
surface element 31
system matrix rank 166
system of
–ﬁnite difference equations 144
–measurement units 4
systems of pressure wave smoothing
138
t
tangential
–friction tension 18
–stress 32
temperature 5
thermodynamic equilibrium 23
time 4
Toms effect 62
total
–derivative with respect to time
6
–energy balance equation 22
–head 14
–kinetic energy 17
transfer section 68
transition of laminar ﬂow into
turbulent ﬂow 52
transmitted wave at the joint of two
pipe sections 134
traveling wave 114
turbulent
–diffusion 189
–dynamic viscosity 52
–ﬂow
–in circular pipe 52
–kinematic viscosity 52
–of nonNewtonian ﬂuid 61
–of power ﬂuid 61
–regime 12
–velocity proﬁle 58
twoterm (Q −H)characteristic of
centrifugal pump 78
u
undamped periodic oscillations 2
undeformable pipeline. 5
unit normal 31
universal
–characteristic of centrifugal blower
103
–gas constant 39
–resistance law 59
unstable ﬂuid 152
useful power
–of blower 105
–needed for gas 105
v
vaporgas
–cavities 68
–phase 152
velocity
–gradient 32
–of medium 5
–of perturbation propagation in
pipeline 129
–of propagation of small
perturbations in gas (sound velocity)
142
–of shock wave propagation in
pipeline 128
–of transported medium 8
–stepwise change 128
virtual mass of ﬂuid 112
viscosity 3
viscousplastic Bingham ﬂuid 37
void 152
volume
–balance equation 192
–expansion factor of metal pipeline
42
volumetric ﬂow rate 157
214 Subject Index
w
wave equation 113
wetted perimeter 66
work of
–external forces 9
–internal forces 13
–pressure and internal friction, see work
of internal forces
y
Young modulus 43
Michael V. Lurie Modeling of Oil Product and Gas Pipeline Transportation
Related Titles
Olah, G. A., Goeppert, A., Prakash, G. K. S.
Beyond Oil and Gas: The Methanol Economy
2006
ISBN: 9783527312757
Olah, G. A., Prakash, G. K. S. (eds.)
Carbocation Chemistry
2004
ISBN: 9780471284901
Sheldrake, A. L.
Handbook of Electrical Engineering – For Practitioners in the Oil, Gas and Petrochemical Industry
2003 EBook
ISBN: 9780470864784
Michael V. Lurie
Modeling of Oil Product and Gas Pipeline Transportation
The Author Prof. Dr. Michael V. Lurie Russian State University of Oil and Gas Moscow, Russian Federation Translation Emmanuil G. Sinaiski Leipzig, Germany
All books published by WileyVCH are carefully produced. Nevertheless, authors, editors, and publisher do not warrant the information contained in these books, including this book, to be free of errors. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate. Library of Congress Card No.: applied for British Library CataloguinginPublication Data A catalogue record for this book is available from the British Library. Bibliographic information published by the Deutsche Nationalbibliothek Die Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliograﬁe; detailed bibliographic data are available in the Internet at <http://dnb.dnb.de>. 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim All rights reserved (including those of translation into other languages). No part of this book may be reproduced in any form – by photoprinting, microﬁlm, or any other means – nor transmitted or translated into a machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not speciﬁcally marked as such, are not to be considered unprotected by law. Printed in the Federal Republic of Germany Printed on acidfree paper Composition Laserwords, Chennai Printing Strauss GmbH, M¨ vlenbach o Bookbinding Litges & Dopf GmbH, Heppenheim ISBN: 9783527408337
Cover Picture TransAlaska Pipeline
V
In memory of the Teacher – academician Leonid I. Sedov
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 9783527408337
VII
Foreword
This book is dedicated ﬁrst and foremost to holders of a master’s degree and postgraduate students of oil and gas institutes who have decided to specialize in the ﬁeld of theoretical problems in the transportation of oil, oil products and gas. It contains methods of mathematical modeling of the processes taking place in pipelines when transporting these media. By the term mathematical model is understood a system of mathematical equations in which framework a class of some processes could be studied. The solution of these equations provides values of parameters without carrying out model and, especially, full scale experiments. Physical laws determining the dynamics of ﬂuids and gases in pipes are presented. It is then shown how these laws are transformed into mathematical equations that are at the heart of one or another mathematical model. In the framework of each model, are formulated problems with the aim of investigating concrete situations. In doing so there are given methods of its solution. The book is selfsufﬁcient for studying the subject but the text is outlined in such a way that it impels the reader to address oneself to closer acquaintance of considered problem containing in special technical literature.
Professor Michael V. Lurie Moscow
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 9783527408337
IX
Contents
Dedication Page V Foreword Preface
VII XIII XV
List of Symbols 1
1.1 1.1.1 1.1.2 1.2 1.3 1.4 1.5 1.5.1 1.5.2 1.6 1.6.1 1.6.2 1.7 1.8
Fundamentals of Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines 1 Mathematical Models and Mathematical Modeling 1 Governing Factors 3 Schematization of OneDimensional Flows of Fluids and Gases in Pipelines 4 Integral Characteristics of Fluid Volume 5 The Law of Conservation of Transported Medium Mass. The Continuity Equation 7 The Law of Change in Momentum. The Equation of Fluid Motion The Equation of Mechanical Energy Balance 11 Bernoulli Equation 15 Input of External Energy 16 Equation of Change in Internal Motion Kinetic Energy 17 Hydraulic Losses (of Mechanical Energy) 18 Formulas for Calculation of the Factor λ(Re, ε) 20 Total Energy Balance Equation 22 Complete System of Equations for Mathematical Modeling of OneDimensional Flows in Pipelines 29 Models of Transported Media 31 Model of a Fluid 31 Models of Ideal and Viscous Fluids 32 Model of an Incompressible Fluid 34 Model of Elastic (Slightly Compressible) Fluid
9
2 2.1 2.2 2.3 2.4
34
Modeling of Oil Product and Gas Pipeline Transportation. Michael V. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. KGaA, Weinheim ISBN: 9783527408337
5 3.1 5.6 2.7.5 2.3 4.5 4.2.7.6.4 4.6.3 3.2 4.4 4.6 4.2 4.2.7 5 5.2 4.4 3.1 3.7 2.2 .8 3 3.6.1 4. Modeling of the Operation of Pumps and OilPumping Stations 75 Pumps 75 OilPumping Station 78 Combined Operation of Linear Pipeline Section and Pumping Station 81 Calculations on the Operation of a Pipeline with Intermediate OilPumping Stations 84 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 87 Modeling of Stationary Operating Regimes of GasPipeline Sections 92 Distribution of Pressure in Stationary Gas Flow in a GasPipeline 94 Pressure Distribution in a GasPipeline with Great Difference in Elevations 96 Calculation of Stationary Operating Regimes of a GasPipeline (General Case) 97 Investigation of Thermal Regimes of a GasPipeline Section 98 Modeling of Blower Operation 100 Closed Mathematical Models of OneDimensional NonStationary Flows of Fluid and Gas in a Pipeline 109 A Model of NonStationary Isothermal Flow of a Slightly Compressible Fluid in a Pipeline 109 A Model of NonStationary Gas Flow in a Pipeline 112 3.1 4.2 3.6 Model of a Fluid with Heat Expansion 34 Models of NonNewtonian Fluids 36 Models of a Gaseous Continuum 38 Model of a Perfect Gas 39 Model of a Real Gas 39 Model of an Elastic Deformable Pipeline 42 Structure of Laminar and Turbulent Flows in a Circular Pipe 45 Laminar Flow of a Viscous Fluid in a Circular Pipe 45 Laminar Flow of a NonNewtonian Power Fluid in a Circular Pipe 47 Laminar Flow of a ViscousPlastic Fluid in a Circular Pipe 49 Transition of Laminar Flow of a Viscous Fluid to Turbulent Flow 51 Turbulent Fluid Flow in a Circular Pipe 52 A Method to Control Hydraulic Resistance by Injection of AntiTurbulent Additive into the Flow 62 Gravity Fluid Flow in a Pipe 65 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines 73 A System of Basic Equations for Stationary Flow of an Incompressible Fluid in a Pipeline 73 Boundary Conditions.7 4 4.X Contents 2.2 2.1 4.1 2.6.3 4.
3 5.6 5.3.2 6.6 158 6.Contents XI 5.3.6 5.1 8.7 7 7.4 6.7 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 113 Wave Equation 113 Propagation of Waves in an Inﬁnite Pipeline 115 Propagation of Waves in a SemiInﬁnite Pipeline 117 Propagation of Waves in a Bounded Pipeline Section 119 Method of Characteristics 121 Initial. Boundary and Conjugation Conditions 124 Hydraulic Shock in Pipes 127 Accounting for Virtual Mass 134 Hydraulic Shock in an Industrial Pipeline Caused by Instantaneous Closing of the Gate Valve 135 NonIsothermal Gas Flow in GasPipelines 138 Gas Outﬂow from a Pipeline in the Case of a Complete Break of the Pipeline 146 Mathematical Model of NonStationary Gravity Fluid Flow 149 NonStationary Fluid Flow with Flow Discontinuities in a Pipeline 152 Dimensional Theory 157 Dimensional and Dimensionless Quantities 157 Primary (Basic) and Secondary (Derived) Measurement Units Dimensionality of Quantities.3.3.3 5.3 8. Batching 187 8.5 5.5 5.1 7.4 .3.2 5. Dimensional Formula 159 Proof of Dimensional Formula 161 Central Theorem of Dimensional Theory 163 DimensionallyDependent and DimensionallyIndependent Quantities 164 Buckingham Theorem 168 6 6.8 5.9 5.2 8.3.3 6.3.7 5.5 6.5 7.3 7.3.3.6 8 Physical Modeling of Phenomena 173 Similarity of Phenomena and the Principle of Modeling 173 Similarity Criteria 174 Modeling of Viscous Fluid Flow in a Pipe 175 Modeling Gravity Fluid Flow 176 Modeling the Fluid Outﬂow from a Tank 178 Similarity Criteria for the Operation of Centrifugal Pumps 179 Dimensionality and Similarity in Mathematical Modeling of Processes 183 Origination of Similarity Criteria in the Equations of a Mathematical Model 183 OneDimensional NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 184 Gravity Fluid Flow in a Pipeline 186 Pipeline Transportation of Oil Products.2 7.4 5.4 5.1 5.1 6.4 7.
4.1 8.4 Principle of Oil Product Batching by Direct Contact 188 Modeling of Mixture Formation in Oil Product Batching 189 Equation of Longitudinal Mixing 192 SelfSimilar Solutions 194 References 199 Appendices 201 Author Index Subject Index 205 207 .4.2 8.3 8.XII Contents 8.4.4.
Emmanuil Sinaiski Leipzig ‘‘.XIII Preface This book presents the fundamentals of the mathematical simulation of processes of pipeline transportation of oil. Weinheim ISBN: 9783527408337 . in the framework of a given physical problem. . Engineers engaged in the design of and calculations on pipelines will ﬁnd a detailed theoretical and practical textbook on the subject of their work. M. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. . Lurie. The author of the book. Beginners will ﬁnd in this book a consecutive description of the theory and mathematical simulation methods of stationary and nonstationary processes occurring in pipelines. No human investigation could be referred to as true when it is not supported by mathematical proof’’ Leonardo da Vinci Modeling of Oil Product and Gas Pipeline Transportation. oil products and gas. Graduate and postgraduate students and research workers will become acquainted with situations in the theory and methods in order to generalize and develop them in the future. Professor Dr. is a great authority in Russia in the ﬁeld of the hydromechanics of oil and gas pipeline transportation. It is shown how the basic laws of mechanics and thermodynamics governing the ﬂow of ﬂuids and gases in pipelines are transformed into mathematical equations which are the essence of a certain mathematical model and. Prof. The book is suitable for graduate and postgraduate students of universities having departments concerned with oil and gas and to engineers and research workers specializing in pipeline transportation. Michael V. appropriate mathematical problems are formulated to analyze concrete situations. KGaA.
speciﬁc kinetic energy Modeling of Oil Product and Gas Pipeline Transportation.01 P centistokes. 0.01 St = 10−6 m2 s−1 pipeline internal diameter diameter increment nominal internal diameter of pipeline. cylinder internal diameter pipeline external diameter velocity of hydraulic shock wave propagation in a pipeline velocity of discontinuity front propagation in the positive direction of the xaxis diameter of impeller diameter of pump impeller Joule–Thompson factor internal energy density. Weinheim ISBN: 9783527408337 . Michael V. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. 0. speciﬁc internal energy kinetic energy density. KGaA.XV List of Symbols Symbol a a a A A+ A− [A] = A+ − A− dAin dAex b c c C2 Cf Cp CSh Cv cP cSt d d d0 D D D Dim Dp D∗ ein ekin Deﬁnition radius of the ﬂow core dimensionless constant parameter of the (Q − H) characteristic proportionality factor value of parameter A to the left of the discontinuity front value of parameter A to the right of the discontinuity front jump of parameter A at the discontinuity front elementary work of internal force elementary work of external force parameter of the (Q − H) characteristic velocity of wave propagation in a pipeline sound velocity in gas integration constant friction factor heat capacity at constant pressure Chezy factor heat capacity at constant volume centipoise.
XVI List of Symbols E Ein Ekin Ei(z) f˜1 fτ (Q)− F dFn Fr g g0 . hn H H H = F(Q) H1 ˜ H1 He i i0 I I I1 . heattransfer factor. Pa factor of longitudinal mixing of oil product length of the mixture region length of a pipeline or a pipeline section mass ﬂow rate initial mass ﬂow rate oilpumping station . g1 h h(S) hc hcr hπ. I2 J k k k κ κ k k K K K lc L ˙ M ˙0 M OPS elastic modulus in extension and compression. Ostwald ﬂuid parameter of nonNewtonian ﬂuid factor. empirical factor 1/K Karman constant dimensionless constant kinematic consistency heat transfer factor elastic modulus of ﬂuid. Young’s modulus internal energy kinetic energy Euler function dimensionless factor friction force restoring force elementary force Froude number acceleration due to gravity dimensionless constants piezometric head depth of pipeline crosssection ﬁlling with ﬂuid head losses in station communications critical depth head before PLP normal depth of gravity ﬂow in the pipe head differential head headdischarge (Q − H) characteristic of a pump hydraulic head hydraulic head Hedstroem number hydraulic gradient hydraulic gradient momentum Ilyushin number Riemann invariants gas enthalpy factor of string elasticity factor of power.
0. heat outﬂow (qex < 0) from gas speciﬁc mass ﬂow rate external heat ﬂux volume ﬂow rate. ﬂuid ﬂow rate ﬂow rate of gas at the entrance to the compressor station commercial ﬂow rate of gas mass ﬂow rate . pe . pst = 101 325 Pa pressure before oilpumping station head before station saturated vapor tension (pressure) pressure jump incident pressure wave amplitude reﬂected pressure wave amplitude transmitted pressure wave amplitude poise. pπ pr pst pu pu pv [p] [pinc ] [preﬂ ] [ptrans ] P Ps Pa Pe qh qex qM qn Q Qe Qk QM factor of power. pressure drop nominal pressure. initial pressure. pressure at the beginning of the pipeline section pressure of gas at the entrance of compressor station and blower critical pressure external pressure.1 kg m s−1 wetted perimeter pascal (SI unit). kg m−1 s−2 Peclet number speciﬁc heat ﬂux heat inﬂow (qex > 0) to gas. kW power of external mechanical devices useful power of mechanical force acting on gas speciﬁc power pressure difference between internal and external pressures.List of Symbols XVII n n n n n n0 nin N Nmech Nus N/ρe p p p0 pen pcr pex pin pL pl . normal pressure. pressure at initial crosssection of the pipeline section internal pressure pressure at the end of the pipeline section pressure at the pressure line of pumps (PLP) reduced pressure standard pressure. Ostwald ﬂuid exponent in Ostwald rheological law exponent number of revolutions of centrifugal blower shaft unit normal vector nominal number of revolutions of blower shaft speciﬁc power of internal friction forces power consumption.
area of pipeline crosssection part ﬁlled with ﬂuid area of pipeline crosssection. 10−4 m2 s−1 time absolute temperature nominal temperature. initial temperature. temperature of ﬂuid at normal condition average temperature over pipeline section length critical temperature temperature of gas at the entrance to the compressor station temperature of external medium temperature at the end of pipeline section mean temperature reduced temperature standard absolute temperature velocity distribution over crosssection maximum value of velocity ﬂuid velocity at pipe wall dynamic velocity velocity averaged over crosssection mean ﬂow rate velocity critical velocity volume ﬂuid velocity jump acceleration coordinate along the pipeline axis coordinate of gravity ﬂow section beginning coordinate of gravity ﬂow section end coordinate transverse to the pipeline axis. direction of a normal to the elementary surface dσ elevation level of a pipeline crosssection x geometrical height differences of sections 1 and 2 overcompressibility factor .XVIII List of Symbols Qv r r0 R R0 Rh Rim Rr Re Recr Re∗ S S0 St t T T0 Tav Tcr TB Tex TL Tm Tr Tst u(y) umax uw u∗ v v vcr V [v] w x x1 x2 y z(x) (z1 − z2 ) Z volume ﬂow rate of gas at pipeline crosssection radial coordinate pipeline radius gas constant (R = R0 /µg ) universal gas constant hydraulic radius radius of the impeller reduced gas constant Reynolds number critical Reynolds number generalized Reynolds number area of a crosssection. nominal (basic) area stokes.
compression ratio. dimensionless coordinate dimensionless parameter. S+ ρ0 ρst σ average overcompressibility factor reduced gas overcompressibility angle of inclination of the pipeline axis to the horizontal factors volume expansion factor thermal expansion factor compressibility factor adiabatic index ratio between the hydraulic gradient of pipeline section completely ﬁlled with ﬂuid and the absolute value of the gravity ﬂow section with slope αp to the horizontal shear rate. density of ﬂuid under normal conditions gas density under standard conditions area of suction branch pipe crosssection. hoop stress.List of Symbols XIX Zav Z = Zr α ¯ ακ . similarity criterion initial pressure distribution density values of parameters before hydraulic shock wave values of parameters after hydraulic shock wave nominal density. ﬂuid density at p0 . selfsimilar coordinate. p+ . molecular weight turbulent dynamic viscosity apparent viscosity of power Ostwald ﬂuid kinematic viscosity factor m2 s−1 kinematic viscosity factor at temperature T0 kinematic viscosity factor at temperature T1 Poisson ratio turbulent kinematic viscosity factor of volumetric expansion. v− . roughness of wall surface relative roughness. thickness ratio local resistance factor dimensionless radius efﬁciency function of temperature. K−1 . circumferential stress . S− ρ+ . degree of pipe ﬁlling. p− . s−1 pipeline wall thickness absolute equivalent roughness. ακ αv αT β γ γ γ ˙ δ ε ς(t) η η(%) θ λ λeff µ µg µt µ ˜ δ ν0 ν1 νP νt ξ (x) ρ ρ− . parameter of overcompressibility factor. concentration of antiturbulent additive hydraulic resistance factor effective factor of hydraulic resistance dynamic viscosity factor kg m−1 s−1 molar mass of gas. v+ . parameter of state equation of real gas. concentration.
angular velocity of impeller rotation . surface element tangential (shear) stress tangential friction stress critical (limit) shear stress tangential (shear) stress at the pipeline internal surface speciﬁc volume critical speciﬁc volume angle of inclination of a straight line to the abscissa. central angle initial ﬂuid velocity distribution frequency of rotor rotation.XX List of Symbols dσ τ τ0 τw υ υcr ϕ ω (x) elementary surface area.
For example. Here we do not take into account the physicalchemical properties of the string. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. Each model represents a deﬁnite schematization of the phenomenon taking into account not all the characteristic factors but some of them governing the phenomena and characterizing it from some area of interest to the researcher. Further schematization could be done by taking into account the drag arising from the air ﬂow around the moving load and the rubbing of the load during its motion along the guide. Secondly. In the examination of oscillations of a small load on an elastic spring we meet with greater schematization of the phenomenon. the material from which the body is made and so on. to examine the motion of a body the material point model is often used. Michael V. its construction and material properties and so on. Nevertheless. dt2 (1. First the load is taken as a point mass m.1) Modeling of Oil Product and Gas Pipeline Transportation.1 1 Fundamentals of Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines 1. that is we use the material point model.1 Mathematical Models and Mathematical Modeling Examination of phenomena is carried out with the help of models. In such a model the dimensions of the body are assumed to be equal to zero and the whole mass to be concentrated at a point. when examining the motion of planets around the sun or satellites around the earth. Weinheim ISBN: 9783527408337 . the elastic string is also schematized by replacing it by the socalled restoring force F = −k · x. the material point model gives brilliant results in the calculation of the trajectories of a body under consideration. The use of the differential equation m d2 x = −k · x. ignoring body size and shape and the physical and chemical properties of the body material. In other words we ignore a lot of factors associated with body size and shape. where x(t) is the deviation of the material point modeling the load under consideration from the equilibrium position and k is the factor characterizing the elasticity of the string. The question is: to what extent would such a schematization be efﬁcient in examining the phenomenon? As we all know such a body does not exist in nature. KGaA. and in many other cases.
1) and the problem (1. it is possible. from the logical point of view. On the other hand the same experiment shows that oscillations of the load are gradually damping in time and come to a stop. This solution permits us to predict the load motion at instants of time t > 0 and by so doing to discover regularities of its motion that were not previously evident. equal to −1. in principle. This means that the model (1. In the given case we have the socalled initial value (Cauchy) problem allowing an exact solution. that is. ˙ ˙ at x < 0 and equal to 0. Hence the model is undoubtedly correct and efﬁcient. This raises the question. both schematizations (models) are consistent? The answer is: only from results obtained in the framework of these models. 0 (1. the aboveoutlined model of load oscillation around an equilibrium position allows one to calculate the motion of the load as x(t) = x0 · cos k ·t + m m ·v0 · sin k k ·t m having undamped periodic oscillations. that is using the equation m d2 x ˙ ˙ = −k · x − f0 · sign(x) − f1 · x dt2 (1. how can one tell about the correctness or incorrectness of the phenomenon schematization when. It is also possible of course to produce another more general schematization of the same phenomenon which takes into account a great number of characteristic factors inherent to this phenomenon. at x = 0).1) represents the closed mathematical model of the considered phenomenon and when the initial conditions are included (1.2) this is the concrete mathematical model in the framework of this model. ˙ ˙ namely the forces of dry −f0 · sign(x) and viscous −f1 · x friction (where the ˙ ˙ ˙ symbol sign(x) denotes the function x− sign equal to 1. The fact that the load motion can begin from an arbitrary position with an arbitrary initial velocity may be taken into account in the schematization by specifying initial conditions at t = 0 : x = x0 . at x > 0. Including in the number of forces acting on the load additional forces.2 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines expressing the second Newtonian law is also a schematization of the phenomenon.2) do not take into account some factors which could be of interest for us.2) Equation (1. and the accepted schematization is inadequate. For example. The latest circumstance contains the whole meaning and purpose of mathematical models. How can one evaluate the obtained result? On the one hand there exists a time interval in the course of which the derived result accords well with the experimental data.3) . since the motion is described in the framework of Euclidian geometry which is the model of our space without taking into account the relativistic effects of the relativity theory. v = dx dt = v0 . to have another more general model of the considered phenomenon.
oil products. elasticity and others. the motion itself is not onedimensional. gas and other deformable media motion is based on the continuum model. But such a schematization has proved to be very convenient in the use of the mathematical apparatus of continuous functions and. which occupies a highly important place in the following chapters. Therefore such seemingly natural investigation of medium motion through a study of discrete molecules is absolutely unacceptable. Thus it is necessary to use more complex schematizations or in another words to exploit more complex models.1 Governing Factors In the examination of different phenomena the researcher is always restricted by a ﬁnite number of parameters called governing factors (parameters) within the limits of which the investigation is being studied. requiring from the researcher great experience. But even the new model describes only approximately the model under consideration.1. (1. One of the general schematization methods for ﬂuid. 1965). as was shown in practice. viscosity. gas.1). The complexity of these processes is very high and the governing forces are not always known. In the case when the size and shape of the load strongly affect its motion. Oil. That is why a system of material points continuously ﬁlling a part of space is called a continuum. This brings up the question: How to reveal the system of governing parameters? It could be done for example by formulating the problem mathematically or. 1. Correct schematization frequently represents a challenging task. Replacement of a real medium consisting of separate molecules by a continuum represents of course a schematization. water or metals may be considered as a medium continuously ﬁlling one or another region of the space. it is quite sufﬁcient for studying the overwhelming majority of observed phenomena. by building a mathematical model of the considered . comprise a great collection of different atoms and molecules in permanent heat motion and with complex interactions. Therefore it adequately describes the phenomenon. the forces acting on the body have a more complex nature and so on.1.1 Mathematical Models and Mathematical Modeling 3 instead of Eq. including liquids and gases. one makes the schematization (model) more complete. heat conductivity. in other words. intuition and deep insight into the phenomenon to be studied (Sedov. By molecular interactions we mean such properties of real media as compressibility. It is known that all media. Because each macroscopic volume of the medium under consideration contains a great number of molecules the medium could be approximately considered as if it ﬁlls the space continuously. Of special note is the continuum model.
oil product and gas are considered as a continuum continuously ﬁlling the whole crosssection of the pipeline or its part. The use of dimensional theory permits us to rewrite the formulated dependence in invariant form. in the general case.4 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines phenomenon as was demonstrated in the abovementioned example. Let us investigate the decrease in a parachutist’s speed v in the air when his motion can be taken as steady. m. g. Being governed only by intuition it is an easy matter to assume the speed to be dependent on the mass of the parachutist m. there is no need for mathematical schematization of the process. that is. deﬁning the soughtfor dependence. x0 . g. that is all governing parameters depend only on one space coordinate x measured along the pipeline axis and. among ﬁve governing parameters there are only two independent dimensionless combinations. t. It is enough to be guided. The viscosity of the air ﬂowing around the parachute during its descent can be taken into account or ignored since the force of viscous friction is small compared to parachute drag. So the function sought could be assumed to have the following general form v = f (m. m/ρD3 and L/D.2 Schematization of OneDimensional Flows of Fluids and Gases in Pipelines In problems of oil and gas transportation most often schematization of the ﬂow process under the following conditions is used: • oil. ρD3 D Thus. • the governing parameters of the ﬂow represent values of the corresponding physical parameters averaged over the pipeline crosssection. independent of the system of measurement units (see Chapters 6 and 7) v gD =f m L . k. In this problem the governing parameters are: x. L. the length L of its shroud and the air density ρ. But. Both cases represent only different schematizations of the phenomenon. ρ). . acceleration due to g. the diameter of the parachute canopy D. D. intuition and understanding of the mechanism of the phenomenon. • the ﬂow is taken as onedimensional. ρ. . D. f1 . Then the governing parameters are: m.⇒ v = ρD3 D gD · f m L . f0 . as has already been noted. by experience.1. . 1. on time t. in order to determine the system of governing parameters. v0 . L.
then S(x) = S0 = const. Let. This volume of the transported medium is called the movable ﬂuid volume or individual volume. Figure 1. and the difference between the demarcation boundaries (x2 − x1 ) deﬁning the length of the ﬂuid volume remains constant. t) – shear stress (friction force per unit area of the pipeline internal surface). If the pipeline has a constant diameter. t) – pressure at the pipeline axis. Its special feature is that it always consists of the same particles of the continuum under consideration. t) = ρvS – mass ﬂow rate of the medium. kg s−1 and other. the area S of the pipeline crosssection depends. t) = vS – volume ﬂow rate of the medium.1).1. modeling a real ﬂuid and a real gas. τ(x. in the general case. together with the pipeline surface. If the continuum located between these two crosssections is identiﬁed with a system of material points and track is kept of its displacement in time. for example. p(x. the boundaries x1 and x2 become dependent on time and. degrees. m s−1 . contain one and the same material points of the continuum. t) – density of medium to be transported.1 Movable ﬂuid volume of the continuum. Pa = N m−2 . . an arbitrary volume of the medium be transported between crosssections x1 and x2 of the pipeline (Figure 1. m3 s−1 . Pa = N m−2 . then S = S0 = const.2 Integral Characteristics of Fluid Volume • 5 • • the proﬁle of the pipeline is given by the dependence of the height of the pipeline axis above sea level on the linear coordinate z(x). t) – velocity of the medium. the transported medium is incompressible and the pipeline is nondeformable. If the pipeline is assumed to be undeformable. the most important parameters are: ρ(x. If. t) – temperature of the medium to be transported. T(x. Mathematical models of ﬂuid and gas ﬂows in the pipeline are based on the fundamental laws of physics (mechanics and thermodynamics) of a continuum. on x and t. Q(x. v(x..2 Integral Characteristics of Fluid Volume In what follows one needs the notion of movable ﬂuid volume of the continuum in the pipeline. 1. kg m−3 . then S = S(x). ˙ M(x. at some instant of time.
t) dx. that is the derivative of a ﬂow parameter with respect to time at a given space point. t) dx − kinetic energy of the ﬂuid volume (J). Since the main laws of physics are often formulated as connections between physical quantities and the rate of their change in time. t) dx − momentum of ﬂuid volume (kg m s−1 ). t) dx − internal energy of the ﬂuid volume. t) · S(x. x = const. From mathematical analysis it is known how an integral containing a parameter. The total derivative with respect to time is equal to d dt x2 (t) x1 (t) A(x. t) dx = ∂ [A(x. different particles of the continuum are located in this crosssection. t)] dx ∂t dx2 dx1 + A(x. momentum and energy of a material point system. t) · S(x. Ekin = x2 (t) x1 (t) αk ρv2 S(x. where ein is the density of the internal energy (J kg−1 ). when the integrand and limits of integration depend on this parameter.g. 2 where αk is the factor. is differentiated with respect to this parameter. The symbol of differentiation d()/ dt denotes the total derivative with respect to time. Ein = x2 (t) x1 (t) ρ(x. dt dt . t) · S(x. These quantities model the mass.6 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines Exploiting the notion of ﬂuid or individual volume of the transported medium in the pipeline one can introduce the following integral quantities: M= I= x2 (t) x1 (t) x2 (t) x1 (t) ρ(x. t)x1 (t) · . The local derivative with respect to time gives the rate of ﬂow parameter change at a given crosssection of the ﬂow while. t) · S(x. We have d dt x2 (t) x1 (t) x2 (t) x1 (t) A(x. at two consecutive instances of time. t) · S(x. that is the internal energy per unit mass. t)x2 (t) · − A(x. t) · ein (x. t) · S(x. e. in the considered case it is t. t) dx − mass of ﬂuid volume (kg). associated with individual particles of a continuum whereas the symbol ∂()/∂t denotes the local derivative with respect to time. t) · S(x. t) · v(x. ρ(x. t) · S(x. we ought to adduce the rule of integral quantity differentiation with respect to time.
∂x x2 (t) x1 (t) A(x.5) This equation should be obeyed for any ﬂuid particle of the transported medium. t) · S(x. The Continuity Equation 7 First. t) · v(x. t)] dx. Therefore the ﬁrst equation would be obtained when the transported medium is governed by the mass conservation law d dt x2 (t) x1 (t) ρ(x. t) · S(x. t) · S(x. t)x1 (t) = we obtain d dt x2 (t) x1 (t) x2 (t) x1 (t) ∂ [A(x. . t)x2 (t) − A(x. If. t) dx = x2 (t) x1 (t) ∂ [A(x. t) · S(x. Applying to Eq. Hence d dt x2 (t) x1 (t) A(x. t)x1 (t) . (1. t) · v(x. the velocity of the transported medium v(x. t) · v(x.4) 1. t) · S(x. (1. t). t) · v(x. we take into account the wellknown Newton–Leibniz formula. t) · S(x. the ﬁrst term having been taken with a plus sign and the second with a minus sign (see Appendix B). t) · S(x. t) and the area of the pipeline crosssection S(x. according to which A(x. t) · v(x. t) · S(x. t)] dx ∂t + A(x.3 The Law of Conservation of Transported Medium Mass. t) dx = ∂AS ∂ASv + ∂t ∂x dx. t)x2 (t) − A(x. t) dx = 0. For the case of the ﬂuid volume of the medium the quantities dx2 / dt and dx1 / dt are the corresponding velocities v2 (t) and v1 (t) of the medium in the left and right crosssections bounding the considered volume.4) the rule (1. we obtain x2 (t) x1 (t) ∂ρS ∂ρvS + ∂t ∂x dx = 0. at frozen upper and lower integration limits.5) of differentiation of integral quantity with regard to ﬂuid volume. The Continuity Equation The density ρ(x.1. the integrand is differentiated (the derivative being local) and then the integrand calculated at the upper and lower integration limits is multiplied by the rates of change of these limits dx2 / dt and dx1 / dt. t) · S(x. that is for any values x1 (t) and x2 (t). (1. t) cannot be chosen arbitrarily since their values deﬁne the enhancement or reduction of the medium mass in one or another place of the pipeline.3 The Law of Conservation of Transported Medium Mass. in addition.
∂t ∂x (1. Solution. In the case of a compressible medium. Example. the density ρ(x) changes along the length of pipeline section under consideration.8 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines Since the last relation holds for arbitrary integration limits we get the following differential equation ∂ρS ∂ρvS + = 0. the last equation is simpliﬁed to dρvS ˙ = 0 ⇒ M = ρvS = const. v = 4Q/πd2 = const. Since the density is as a rule connected with pressure.008 = 0. v = 4 · 2500/(3600 · 3.7) it follows that ρv = const. Then from the condition ρv = const. If the ﬂow is stationary. The internal diameter d of the oil pipeline is equal to d = D − 2δ = 0.804 m. Hence the ﬂow velocity of a homogeneous incompressible ﬂuid in a pipeline of constant crosssection does not change along the length of the pipeline. a gas. Example. Find the velocity of the gas ﬂow . = 2. = from Eq.7) ˙ This means that in stationary ﬂow the mass ﬂow rate M is constant along the pipeline. δ = 10 mm) is 180 kg s−1 . this change represents a monotonic function decreasing from the beginning of the section to its end. In the case of a homogeneous incompressible ﬂuid (sometimes oil and oil product can be considered as such ﬂuids) ρ ∼ ρ0 = const. (1. e. From this follow two important consequences: 1. and the = ﬂow velocity v(x) = const.g.8042 ) ∼ 1.. that is the local derivative with respect to time is zero (∂()/∂t = 0). dx (1.82 − 2 · 0. The volume ﬂow rate of the oil transported by a pipeline with diameter D = 820 mm and wall thickness δ = 8 mm is 2500 m3 h−1 .6) which is called continuity equation of the transported medium in the pipeline. It is required to ﬁnd the velocity v of the ﬂow.14 · 0. If we ignore the pipeline deformation and take S(x) ∼ S0 = const.37 m s−1 . The mass ﬂow rate of gas transported along the pipeline (D = 1020 mm. it follows that the velocity v(x) of the ﬂow also increases monotonically from the beginning of the section to its end. Hence the velocity of the gas ﬂow in a pipeline with constant diameter increases from the beginning of the section between compressor stations to its end.
v1 = M/(ρ1 S) = 4 · 180/(45 · 3.2 m s−1 . Pa).6) contains several unknown functions. Representing the pressure difference in the form of an integral over the length of the considered volume p1 S1 − p2 S2 = − x2 (t) x1 (t) ∂pS dx ∂x .8 towards the end as compared with the velocity at the beginning. = ˙ v2 = M/(ρ2 S) = 4 · 180/(25 · 3. The Equation of Fluid Motion 9 v1 at the beginning and v2 at the end of the gaspipeline section. if the density of the gas at the beginning of the section is 45 kg m−3 and at the end is 25 kg m−3 . ˙ Solution.14 · 12 ) ∼ 9. among others. α > 0 for ascending sections of the pipeline. On the left is the total derivative of the ﬂuid volume momentum of the transported medium with respect to time and on the right the sum of all external forces acting on the considered volume. To get additional equations we can use.1. the equation of the change in momentum of the system of material points comprising the transported medium.4 The Law of Change in Momentum.14 · 12 ) ∼ 5.). The fourth term gives the sliding component of the gravity force (α(x) is the slope of the pipeline axis to the horizontal. 1. The Equation of Fluid Motion The continuity equation (1. The third term deﬁnes the friction force at the lateral surface of the pipe (τw is the shear stress at the pipe walls. g is the acceleration due to gravity).8) x2 (t) x1 (t) ρg sin α(x) · S dx. that is the gas ﬂow = velocity is enhanced by a factor 1. The ﬁrst term on the righthand side of the equation gives the difference in pressure forces acting at the ends of the single continuum volume. The second term represents the axial projection of the reaction force from the lateral surface of the pipe (this force differs from zero when S = const. that is the friction force per unit area of the pipeline internal surface. hence the use of only this equation is insufﬁcient to ﬁnd each of them.4 The Law of Change in Momentum.1 m s−1 . This law expresses properly the second Newton law applied to an arbitrary ﬂuid volume of transported medium d dI = dt dt − x2 (t) x1 (t) v · ρS dx = (p1 S1 − p2 S2 ) + πd · τw dx − x2 (t) x1 (t) x2 (t) p x1 (t) ∂S dx ∂x (1. α < 0 for descending sections of the pipeline.
Now applying to the lefthand side of this equation the differentiation rule of ﬂuid volume x2 (t) x1 (t) ∂ρvS ∂ρv2 S + ∂t ∂x x2 (t) x1 (t) dx dx. Remark. friction and gravity forces. (1. since we are dealing with the velocity differentiation of one and the same ﬁxed particle of the transported medium moving from one . namely pressure.11) Now the meaning of Eq.9) and take into account that in accordance with the continuity equation (1. dt ∂t ∂x (1.10) The expression in brackets on the lefthand side of Eq. = −S ∂p 4 − S · τw − Sρg sin α(x) ∂x d As far as the limits of integration in the last relation are arbitrary one can discard the integral sign and get the differential equation ∂p 4 ∂ρvS ∂ρv2 S + =S· − − τw − ρg sin α(x) . ∂x we obtain the following equation d dt x2 (t) x1 (t) ρvS dx = x2 (t) x1 (t) −S ∂p 4 − S · τw − Sρg sin α(x) ∂x d dx. ∂x d (1. (1.10) represents the total derivative with respect to time. The acceleration w = dv/ dt is a total derivative with respect to time (symbol d()/ dt). ∂t ∂x ∂x d If we represent the lefthand side of this equation in the form v ∂ρS ∂ρvS + ∂t ∂x + ρS ∂v ∂v +v ∂t ∂x (1.10 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines and noting that − x2 (t) x1 (t) ∂pS dx + ∂x x2 (t) x1 (t) p ∂S dx = − ∂x x2 (t) x1 (t) S ∂p dx. So Eq. the resulting equation may be written in a more simple form ρ ∂v ∂v +v ∂t ∂x =− ∂p 4 − τw − ρg sin α(x). about the connection between total and partial derivatives with respect to time. (1.10) expresses the Newton’s Second Law and can therefore also be called the ﬂow motion equation.6) the expression in the ﬁrst brackets is equal to zero. that is the particle acceleration w= dv ∂v ∂v = +v .10) becomes clearer: the product of unit volume mass of transported medium and its acceleration is equal to the sum of all forces acting on the medium.
t + t). t) ∂A(x. or as it is also called the individual or Lagrangian derivative. that is at a constant value of x.5 The Equation of Mechanical Energy Balance 11 crosssection of the pipeline to another one. dt ∂t ∂x (1.5 The Equation of Mechanical Energy Balance Consider now what leads to the use of the mechanical energy change law as applied to the system of material points representing a ﬂuid particle of the transported medium. The acceleration w of this particle is deﬁned as the limit w= v(x + dv = lim t⇒0 dt x. t) = +v· . t + t t) − v(x. Let a particle of the medium at the instant of time t be in the crosssection x of the pipeline and so have velocity v(x.12) no matter whether the case in point is velocity or any other parameter A(x. This law is written as: dEkin dAex dAin = + dt dt dt (1. t) over the crosssection then the kinetic energy would be expressed as the integral Ekin = x2 (t) x1 (t) ρv2 S dx. or as it is also called the local or Eulerian derivative. t) is the velocity of the considered particle. t) ∂A(x.1.13) that is the change in kinetic energy of a system of material points dEkin is equal to the sum of the work of the external dAex and internal dAin forces acting on the points of this system. Thus such a derivative gives the change in velocity of different particles of the transported medium entering a given crosssection of the pipeline. 2 . dt ∂t ∂x 1. has the form (1. In the next instant of time t + t this particle will transfer to the crosssection x + x and will have velocity v(x + x. We can calculate separately the terms of this equation but ﬁrst we should deﬁne more exactly what meant by the kinetic energy Ekin . whereas the partial derivative with respect to time (symbol ∂()/∂t) has the meaning of velocity differentiation at a given place in space. and the partial derivative (∂/∂t).12) A similar relation between the total derivative ( d/ dt). t). from the last equality it follows that ∂v ∂v dv = +v· . t) dA(x. If the transported medium moves in the pipeline as a piston with equal velocity v(x. t) = ∂v ∂t + x ∂v dx . · ∂x t dt Since dx/ dt = v(x.
the cases v = 0 and ( u)2 = 0 are not considered. but the rootmeansquare (rms) value of the additive ( u)2 is nonvanishing. If the average velocity v = 0.02–1. then ( u)2 ρ( u)2 ρv2 ρv2 + = · 1+ 2 2 2 v2 = αk · ρv2 2 where αk = 1 + ( u)2 /v2 > 1. with an increase in velocity the laminar ﬂow changes into a turbulent one (pulsating and mixing ﬂow) and the velocities of the separate particles differ signiﬁcantly from the average velocity v of the ﬂow. 2 . The average value of this additive u is equal to zero. At the center of the pipe it reaches the greatest value. as a rule. 2 Let us turn now to the calculation of the terms in the mechanical energy equation (1. The deviation characterizes the kinetic energy of the relative motion of the continuum particle in the pipeline crosssection. such a schematization is too rough because. in practice. if at a small velocity of the ﬂuid the ﬂow regime is laminar. That is why models of the ﬂow are. t) and the additive one (deviation) u representing the difference between the true velocity and the average one. With regard to the introduced factor the kinetic energy of any movable volume of transported medium may be represented as Ekin = x2 (t) x1 (t) αk · ρv2 · S dx. the velocity of the separate layers of the transported medium (ﬂuid or gas) varies over the pipe crosssection. whereas as the internal surface of the pipe is approached the velocity decreases and at the wall itself it is equal to zero. constructed with regard to the difference in ﬂow velocity from the average velocity over the crosssection. It should be noted that in onedimensional theory.05.12 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines But. Furthermore. Remark. while for turbulent ﬂow the value of αk lies in the range 1. Let us calculate ﬁrst the change in kinetic energy d dEkin = dt dt x2 (t) x1 (t) αk · ρv2 S dx . as experiments show. For laminar ﬂow αk = 4/3. Then the kinetic energy of the transported medium unit mass ekin may be presented as the sum of two terms ekin = v2 ( u)2 + 2 2 namely the kinetic energy of the center of mass of the considered point system and the kinetic energy of the motion of these points relative to the center of mass. The true velocity u of a particle of the transported medium is given as the sum u = v + u of the average velocity over the crosssection v(x. as a rule.13).
one can introduce a function P(ρ) of the pressure .g. e. The second term gives the power of the gravity force and the third term Nmech the power of the external mechanical devices acting on the transported medium volume under consideration. that is the forces of mutual friction between the internal layers of the medium.1. The work of the internal forces (pressure and internal friction) executed in unit time is given by dAin = dt x2 (t) x1 (t) p ∂(Sv) dx + ∂x x2 (t) x1 (t) nin · ρS dx. The second term represents the power of the internal friction forces. we get dEkin = dt x2 (t) x1 (t) ρv2 ρv2 ∂ ∂ αk · S + αk · S·v ∂t 2 ∂x 2 dx. In what follows it will be shown that this quantity characterizes the amount of mechanical energy converting into heat per unit time caused by mutual internal friction of the transported particles of the medium. that is the pressure in it depends only on the density p = p(ρ). for compression of the particles of the medium. is equal to dAex = (p1 Sv1 − p2 Sv2 ) − dt =− x2 (t) x1 (t) x2 (t) x1 (t) ρg sin α · v · S dx + Nmech x2 (t) ∂ (pSv) dx − ∂x x1 (t) ρg sin α · v · S dx + Nmech . the factor ∂(Sv)/∂x · dx giving the rate of elementary volume change. The ﬁrst term on the righthand side gives the work of the pressure force in unit time. Gathering together all the terms of the mechanical energy equation we get x2 (t) x1 (t) ρv2 v2 ∂ ∂ αk · αk · ρvS S + ∂t 2 ∂x 2 x2 (t) x1 (t) dx x2 (t) x1 (t) =− ρSv 1 ∂p ρ ∂x + g sin α dx + nin · ρS dx + Nmech . that is the power. the power of the pressure force applied to the initial and end crosssections of the detached volume. including also the work of external mechanical devices. that is an integral with variable integration limits.5 The Equation of Mechanical Energy Balance 13 Employing the rule of integral quantity integration with reference to the ﬂuid volume. more precisely. The ﬁrst term on the righthand side of the last expression gives the work performed in unit time or. pumps if such are used. The work of the external forces (in this case they are the forces of pressure and gravity). that is per unit mass of the transported medium. If the transported medium is barotropic. nin denoting speciﬁc power.
16) ∂ ∂t αk v 2 2 + ρvS ∂ ∂x αk v 2 + P(ρ) + gz = ρS · nin 2 (1. The concept of head is very important in the calculation of processes occurring in pipelines. where the function z(x) is referred to as the pipeline proﬁle. the last equation could be rewritten in the simple form x2 (t) x1 (t) ρS = ∂ ∂t x1 (t) αk v 2 2 + ρvS ∂ ∂x αk v 2 + P(ρ) + gz 2 dx (1. The total head at the pipeline crosssection x consists of the kinetic head (dynamic pressure) αk v2 /2g.15) This is the sought differential equation expressing the law of mechanical energy change.14) x2 (t) nin · ρS dx + Nmech . the condition of arbitrariness of integration limits x1 (t) and x2 (t) in Eq. .14 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines ∂p such that dP = dp/ρ. Then the sign of the integral can be omitted and the corresponding differential equation is ρS or ∂ ∂t αk v 2 2 +v· ∂ ∂x αk v 2 + 2 dp + gz = nin . the piezometric head dp/ρg and the geometric head z. Then Nmech = 0 and we can go from the integral equality (1. (1. we take ρ ∂x into account the equality sin α(x) = ∂z/∂x. ρg g The expression H= αk v 2 + 2g dp +z ρg (1. It represents an independent equation for modeling onedimensional ﬂows of a transported medium in the pipeline. If.14).14) to a differential equation using. x2 (t)] external sources of mechanical energy are absent. It should be emphasized that this equation is not a consequence of the motion equation (1.17) in the derivative on the lefthand side of the last equation has the dimension of length and is called the total head. as before. moreover. P(ρ) = dp/ρ and 1 ∂x = ∂P(ρ) . If we assume that in the region [x1 (t). If we divide both parts of Eq.16) by g we get ∂ ∂t αk v 2 2g +v· ∂ ∂x αk v 2 + 2g dp nin +z = . ρ (1.10). (1.
In integral form.1 Bernoulli Equation In the case of stationary ﬂow of a barotropic ﬂuid or gas in the pipeline the derivative ∂()/∂t = 0. is proportional to the dissipation of mechanical energy into heat through internal friction between the transported medium layers (i < 0). deﬁned as the pressure loss per unit length of the pipeline. x1 If in addition we take i = −i0 = const. that is as applied to transported medium located between two ﬁxed crosssections x1 and x2 . oil and oil product. (1. It is one of the fundamental equations used to describe the stationary ﬂow of a barotropic medium in a pipeline. This last equation has a simple geometric interpretation (see Figure 1..2). This ﬁgure illustrates a pipeline proﬁle (heavy broken line).5 The Equation of Mechanical Energy Balance 15 1. x1 (1.20) where l1−2 is the length of the pipeline between crosssections 1 and 2. For an incompressible homogeneous ﬂuid. the line H(x) denoting the dependence of the total head H on the coordinate x directed along the axis of the pipeline (straight line) with constant slope β to the horizontal (i = dH/ dx = tgβ = const. ρ = const.5. ρg gv (1. then ακ v 2 p + +z 2g ρg − 1 ακ v 2 p + +z 2g ρg = i0 · l1−2 2 (1. dx gv Thus the hydraulic gradient. (i0 > 0). Eq. dp/ρg = p/ρg + const. hence the following ordinary differential equations apply v or d dx αk v 2 + 2g dp nin +z = = i.18) takes the following form αk v 2 + 2g dp +z ρg − 1 αk v 2 + 2g dp +z ρg =− 2 x2 i dx.) and three components of the total head at an . Therefore the Bernoulli equation becomes p αk v 2 + +z 2g ρg − 1 p αk v 2 + +z 2g ρg =− 2 x2 i dx.1.19) This equation is called the Bernoulli equation.18) d dx αk v 2 + 2g dp nin +z = ρg g where i denotes the dimensional quantity nin /gv called the hydraulic gradient i= nin dH = . which under some conditions can be water.
e. Compressors installed separately or combined in a group form the pumping plant destined to set the ﬂuid moving from the crosssection with lesser pressure to the crosssection with greater pressure.25 m).2 Input of External Energy In ﬂuid ﬂow in the pipeline the mechanical energy is dissipated into heat and the pressure decreases gradually. For example. piezometric head p(x)/ρg and kinetic head αk v2 (x)/2g. It should be noted that if we neglect the dynamic pressure (in oil and oil product pipelines the value of the dynamic pressure does not exceed the pipeline diameter.2) is 500 m and diesel fuel with density ρ = 840 kg m−3 is transported along the pipeline. and = the length of the section between the pipeline proﬁle and the line of hydraulic gradient multiplied by ρg gives the value of the pressure in the pipeline crosssection x. when the length of the section AA (see Figure 1. Devices providing pressure restoration or generation are called compressors.12 MPa (≈42 atm).05 then v2 /2g ∼ 0. Let index 1 in the Bernoulli equation refer to parameters at the crosssection x1 of the pump entrance (suction line) and index 2 at the crosssection x2 of the .16 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines Figure 1. The line H(x) representing the dependence of the total head H on the coordinate x along the pipeline axis is called the line of hydraulic gradient.g. or deliver from outside to the ﬂuid. then p = 500 ⇒ p = 500 · 840 · 9. arbitrary crosssection of the pipeline: geometric head z(x). energy whose power is denoted by Nmech .81 = 4 120 200 (Pa) 840 · 9.2 Geometric interpretation of the Bernoulli equation. αk ≈ 1.81 or 4. To do this it is required to expend. at v ≈ 2 m s−1 . 1.5.
21) is the main formula used to calculate the power of the pump generating head H in ﬂuid pumping with ﬂow rate Q. Since ρvS = const. 1. so that the total energy of a particle is equal to αk ρv2 /2. Usually this factor is taken into account by insertion of the pump efﬁciency η η= 1− so that Nmech = ρgQ · H .14) may be written as: x2 x1 d ρvS · dx αk v 2 p + + gz 2 ρ dx = x2 x1 nin · ρS dx + Nmech .. ∂x d .21) x2 −1 i/ H dx x1 <1 The relation (1. namely the kinetic energy of the internal or relative motion in the ﬂow of the transported medium. and nin = gv · i. we obtain Nmech = ρgQ · H− x2 x1 ρgQ · i dx = ρgQ · H· 1− x2 x1 i dx . Ignoring the difference between the kinetic and geometric heads we get ρvS · p2 − p1 − ρ x2 x1 nin · ρS dx = Nmech . Multiplication of motion equation (1. Denoting by H = (p2 − p1 )/ρg the differential head produced by the pump or pumping plant and taking into account that ρvS = ρQ = const. H The expression in parentheses characterizes the loss of mechanical energy within the pump. η(Q) (1. the Bernoulli equation (1.1.10) by the product vS yields ρS d dt v2 2 =− ∂p 4 · vS − τw · vS − ρgvS · sin α(x).6 Equation of Change in Internal Motion Kinetic Energy At the beginning of the previous section it was noted that the total kinetic energy of the transported medium consisted of two terms – the kinetic energy of the center of mass of the particle and the kinetic energy of the internal motion of the center of mass. Now we can derive an equation for the second component of the kinetic energy.6 Equation of Change in Internal Motion Kinetic Energy 17 pump exit (discharge line). where αk > 1.
that is per unit mass of transported medium. ﬂow such motion can be calculated and the quantity nin can be found. (1. In order to derive this quantity theoretically one should know how the layers of transported medium move at each crosssection of the pipeline but this is not always possible. In the next chapter it will be shown that in several cases. d (1. dt 2 d Introduction of nin = −gv · i0 gives ρS d v2 (ακ − 1) = dt 2 4 τw · v S − ρgvS · i0 .22) This is the sought equation of change in kinetic energy of internal motion of onedimensional ﬂow of the transported medium. one obtains ρS v2 4 d (αk − 1) = τw · vS + ρS · nin . Its sense is obvious: the power of the external friction forces (4τw · vS/d) in onedimensional ﬂow minus the power ρgS(v · i0 ) of internal friction forces between the particles causing transition of mechanical energy into heat is equal to the rate of change of internal motion kinetic energy in the ﬂow of the transported medium. which for the ﬂow of an incompressible medium in a pipeline = with constant diameter is the exact condition. such as for turbulent ﬂows of the transported . (1.22) gives v2 d 4 τw − g · i0 . 1. This quantity is very important since it characterizes the loss of mechanical energy converted into heat owing to internal friction between layers of the medium.16) denotes the speciﬁc power of the internal friction force.1).24) It must be emphasized that in the general case.15).18 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines Subtracting this equation termbyterm from the Bernoulli equation (1.1 Hydraulic Losses (of Mechanical Energy) The quantity nin entering into Eq.. including nonstationary ﬂow. 4 (1. such a connection between τw and i0 is absent (see Section 4.23) If v ∼ const. in particular for laminar. This means that the tangential friction tension τw at the pipeline wall and the hydraulic gradient i0 are connected by τw = ρgd · i0 . (αk − 1) = dx 2 d ρ (1. the lefthand part of the equation vanishes.6. In other cases. For stationary ﬂow ( d/ dt = 0 + v · ∂/∂x) of the transported medium Eq.
d the ﬁrst is called the Reynolds number and the second the relative roughness of the pipeline internal surface. . hence. it is not possible to calculate the motion of the layers and other methods of determining nin are needed. one can seek nin in the form nin = − λ v3 · 2 d (1. are dimensional quantities and their numerical values depend on such a choice. ε). The density of the ﬂuid ρ and the acceleration due to gravity g are not included here because intuition suggests that the friction between ﬂuid or gas layers will be dependent on neither their density nor the force of gravity.25) where λ is a dimensional factor (λ > 0). the kinematic viscosity of the ﬂow ν (m2 s−1 ). gas or other medium with complex speciﬁc properties. ν. The apparent contradiction is resolved by the wellknown Buckingham Itheorem.1. The presented formula does not disturb the generality of the consideration because the unknown dependence of nin on the governing parameters of the ﬂow is accounted for by the factor λ. In our case there are two such parameters v·d = Re ν and = ε. d. The quantity of speciﬁc mechanical energy dissipation nin has the following dimension (from now onwards dimension will be denoted by the symbol [ ]) [nin ] = J Nm kg m s−2 m m2 W v3 = = = = 3 = . that is a mixture of water with large rigid particles.g. in accordance with which any dimensionless quantity can depend only on dimensionless combinations of parameters governing this quantity (Lurie. waxy crude oil. The factor 1/2 is introduced for the sake of convenience. without disturbance of generality. Note that the quantity λ is dimensionless. ν. that is its numerical value is independent of the system of measurement units. d. so that λ = f (v. This dependence is valid for any medium be it ﬂuid. For stationary ﬂuid or gas ﬂow one can suppose the factor λ to be dependent on four main parameters: the ﬂow velocity v (m s−1 ). that is the mechanical energy decreases thanks to the forces of internal friction. Thus λ = λ(Re.6 Equation of Change in Internal Motion Kinetic Energy 19 medium. while the parameters v. 2001). ). e. the minus sign shows that nin < 0. suspension or even pulp. the internal diameter of the pipeline d (m) and the mean height of the roughness of its internal surface (mm or m). kg s kg s kg s kg s d So the dimension of nin is the same as the dimension of the quantity v3 /d.
.28) The expression hτ = λ · l1 – 2 /d · v2 /2g on the righthand side of this equation is called the loss of head in DarcyVeisbach form. Here are shown several formulas exploiting the practice. ε) 4 = ρv2 λ ρv2 · = Cf · . (1.6.005.27) Characteristic values of the hydraulic slope are 0.25) acquires the form nin = −λ(Re. 4 2 2 (1.20 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines The formula (1. (1.27) in the case of stationary ﬂow of the transported medium permits us to get an expression for the tangential friction stress τw at the pipeline wall. yields τw = 1 v2 ρgd ρgd · i0 = · λ 4 4 d 2g λ(Re.30) . gv d 2g (1. More detailed information about this factor and its dependence on the governing parameters will be presented below. ε) Details of methods to ﬁnd and calculate the factor of hydraulic resistance λ in Eqs. (1. that is jetwise or layerwise (the Reynolds number Re should be less than 2300).26)–(1. Using Eq. (1. If we substitute Eq. we obtain p αk v 2 + +z 2g ρg − 1 αk v 2 p + +z 2g ρg = λ(Re. If the ﬂow of ﬂuid or gas in the pipeline is laminar. Substitution of Eq.20).24).01–0.26) The factor λ in this formula is called the hydraulic resistance factor.00005–0. then to determine λ the Stokes formula (see Section 3. Re (1. d 2 (1. Turning to the hydraulic gradient i0 .27) into Eq. Characteristic values of λ lie in the range 0. one of the most important parameters of hydraulics and pipeline transportation.03. ε) · 2 l1−2 v2 . 1.29) Cf (Re.2 Formulas for Calculation of the Factor λ(Re.27) into the Bernoulli equation (1. 1934). ε) = where the dimensional factor Cf is called the friction factor of the ﬂuid on the internal surface of the pipeline or the Funning factor (Leibenson et al. ε) · 1 v3 · . one can write i0 = − nin 1 v2 =λ· · .29) and one of the primary factors in hydraulics and pipeline transportation will be given in Chapter 3. d 2g (1.1) is used λ= 64 . (1.
that is vortex ﬂow with mixing layers.3164 λ= √ 4 Re (1. If 104 < Re < 27/ε1. In such a case one can use the simpler Shiphrinson formula λ = 0. in the region of ﬂow transition from laminar to turbulent. that is by the parameter ε. Therefore the ﬂuid ﬂow in this range is ﬂow in a hydraulic smooth pipe.3164 · γ∗ .002·(Re−2320) is the intermittency factor (Ginsburg. 4 2 4 2 From this it transpires that the friction resistance is proportional to the square of the ﬂuid mean velocity and hence this type of ﬂow is called square ﬂow. · (1 − γ• ) + √ 4 Re Re (1. that is in the range of Reynolds number from 2320 up to 104 one can use the approximation formula λ= 64 0. 1957). which does not consider the relative roughness of the pipeline internal surface ε. The best known formula to calculate the factor λ in this case is the Altshuler formula: λ = 0.75 4 2 vd/ν 2 signifying that friction resistance is proportional to ﬂuid mean velocity to the power of 1.31) valid over a wide range of Reynolds number from 104 up to 106 and higher.32) having the same peculiarity as the Stokes formula for laminar ﬂow.33) where γ∗ = 1 − e−0. It is obvious that the form of the last formula assures continuous transfer from the Stokes formula for laminar ﬂow to the Blasius formula for turbulent ﬂow in the zone of hydraulic smooth pipes.25 .11 · ε0. the Altshuler formula becomes the Blasius formula: 0. Then τw = − λ ρv2 0. In this case the friction tension τw at the pipe wall is expressed by formula τw = − λ ρv2 0.6 Equation of Change in Internal Motion Kinetic Energy 21 As the Reynolds number increases (Re > 2300) the ﬂow in the pipeline gradually loses hydrodynamic stability and becomes turbulent.11 · ε + 68 Re 1/4 (1.143 and Re < 105 . Finally.11 · ε1/4 ρv2 · =− · ≈ v2 .0791 ρv2 · · =− 4 ≈ v1. If Re > 500/ε. the second term in parentheses in the Altshuler formula can be neglected compared to the ﬁrst one. . This means that for the considered range of Reynolds numbers the pipeline behaves as a pipeline with a smooth surface.1.75. Whence it follows that at great ﬂuid velocities the ﬂuid friction is caused chieﬂy by the smoothness of the pipeline internal surface.
03–0.35) that is the change in total energy (Ekin + Ein ) of an arbitrary volume of the transported medium happens only due to the exchange of energy with surrounding bodies owing to external inﬂow of heat dQ ex and the work of external forces dAex . 0. δ = 8 mm. Turbulent. 0. 0. .34) is equal to 0. δ = 5 mm.022. Exercise 2. Determine the ﬂow regime.0054. where the Reynolds number Re is very large and this factor depends only on the condition of the pipeline internal surface. = 0.05 mm.2 m s−1 . 0. This law asserts that the energy does not appear from anywhere and does not disappear to anywhere. calculate factors λ and Cf . applied to an arbitrary continuum volume in the pipeline there is one more fundamental physical law valid for any continuum – the law of total energy conservation or. = 0.34) is often used. Turbulent.033.4 m s−1 . Exercise 3.0083.7 Total Energy Balance Equation Besides the law (1. ν = 6 s St) ﬂows along the pipeline (D = 530 mm. ν = 15 s St) ﬂows along the pipeline (D = 156 mm. Benzene (ρ = 750 kg m−3 . Answer. ν = 0. Laminar. Diesel fuel (ρ = 840 kg m−3 . 0. The oil (ρ = 870 kg m−3 . Answer. calculate factors λ and Cf . Determine through the Reynolds criterion the ﬂow regime. As applied to our case this law may be written as follows d(Ekin + Ein ) dQ ex dAex = + dt dt dt (1.017. Answer. (1.22 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines To calculate the hydraulic resistance factor λ of the gas ﬂow in a gas main. 0. δ = 7 mm. Determine through the Reynolds criterion the ﬂow regime. = 0.7 s St) ﬂows along the pipeline (D = 377 mm. λ = 0.067 · 2 d 0. the ﬁrst law of thermodynamics.2 (1.1 mm) with mean velocity v = 0. Eq. as it is also called.15 mm) with mean velocity v = 1. 1.13) of mechanical energy change of material points. in which the absolute roughness Exercise 1.25 mm) with mean velocity v = 0. It changes in total quantity from one form into another. calculate factors λ and Cf .8 m s−1 .0041.
where m is the mass of the body. ∂ (pSv) dx − ∂x x2 (t) x1 (t) dAex =− dt x2 (t) x1 (t) ρg sin α · v · S dx + Nmech where qn is the heat ﬂux going through the unit area of the pipeline surface per unit time (W m−2 ). At the same time the quantities dQ ex / dt and dAex / dt are not generally derivatives with respect to a certain function of state but only represent the ratio of elementary inﬂows of heat energy (differential dQ ex ) and external mechanical energy (differential dAex ) to the time dt in which these inﬂows happened.35) for a movable volume of transported medium enclosed between crosssections x1 (t) and x2 (t). (1.7 Total Energy Balance Equation 23 In Eq.35) Ein is the internal energy of the considered mass of transported medium. It should be kept in mind that these quantities depend on the process going on in the medium. (1. Differentiation of the lefthand side of this equation gives x2 (t) x1 (t) ∂ ∂t = ∂ αk v 2 + ein ρS + 2 ∂x x2 (t) x1 (t) αk v 2 + ein ρvS 2 p ρvS ρ dx dx πd · qn dx − x2 (t) x1 (t) x2 (t) x1 (t) ∂ ∂x − ρvSg ∂z dx + Nmech ∂x .1. representing the internal energy of a unit mass of the considered body ein = Ein /m. 2 = x2 (t) x1 (t) πd · qn dx. In thermodynamics reasons are given as to why the internal energy is a function of state. that is the energy of heat motion. We can write Eq. that is at thermodynamic equilibrium of a body in some state the energy has a welldeﬁned value regardless of the means (procedure) by which this state was achieved. we obtain d dt x2 (t) x1 (t) αk · ρv2 + ρein S dx = 2 x2 (t) x1 (t) x2 (t) x1 (t) πd · qn dx − x2 (t) x1 (t) ∂ (pSv) dx − ∂x ρg sin α · v · S dx + Nmech . unrelated to the kinetic energy. Gathering all terms. In addition to function Ein one more function ein is often introduced. interaction between molecules and atoms and so on. The terms of this equation are d d(Ekin + Ein ) = dt dt dQ dt ex x2 (t) x1 (t) αk ρv2 + ρein S dx . πd · dx is an element of pipeline surface area and d is the pipeline diameter.
37) we get one more energy equation ρS ∂ ∂t αk v 2 2 + ρvS ∂ ∂x αk v 2 + 2 dp + gz = ρvSg · i ρ called the equation of heat inﬂow. that is the limits of integration x1 (t) and x2 (t) in (1.36) are to be arbitrarily chosen.36) to the corresponding differential equation using.38) may be written in a particularly simple form: ρ dein ∼ 4 = · qn − ρvg · i. Then the sign of the integral can be omitted and the differential equation is ∂ ∂t ∂ αk v 2 + ein ρS + 2 ∂x p αk v 2 + ein + ρvS 2 ρ ∂z . the condition that this equation should be true for any volume of the transported medium. With this in mind Eq. ∂x dx (1. (1.38) This equation proved to be especially convenient for modeling ﬂows of incompressible or slightly compressible ﬂuids because the derivative ∂(vS)/∂x expressing the change in ﬂuid volume in the pipeline crosssection is extremely small as is the work p · ∂(vS)/∂x of the pressure forces. (1. then it is possible to pass from integral equality (1. (1.39) . that is Nmech = 0. ∂x (1.24 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines or x2 (t) x1 (t) ∂ ∂t = ∂ αk v 2 + ein ρS + 2 ∂x x2 (t) x1 (t) p αk v 2 + ein + ρvS 2 ρ ∂z dx + Nmech . (1. First. dt d (1. as before.37) = πd · qn − ρvSg Excluding from Eq. This equation could be variously written.37) the change in kinetic energy with the help of the Bernoulli equation with term by term subtraction of Eq. it may be written through the internal energy ein : ∂ ∂ ∂vS (ein · ρS) + (ein · ρvS) = πd · qn − p − ρvSg · i ∂t ∂x ∂x or ρS ∂ein ∂ein +v ∂t ∂x = πd · qn − p · ∂vS − ρvSg · i.16) from Eq.36) πd · qn dx − x2 (t) x1 (t) ρvSg If we assume that inside the region [x1 (t). x2 (t)] the external sources of mechanical energy are absent. ∂x (1.
For an incompressible or slightly compressible medium.40) ein + p ρvS = πd · qn + ρvSg · ρ 1 ∂p −i ρg ∂x If we take into account (as will be shown later) that the expression in parentheses on the righthand side of this equation is close to zero. oil and oil product. the hydraulic slope is expressed through the pressure gradient by the formula i = 1/ρg · ∂p/∂x.42) The internal energy ein depends primarily on the temperature of the ﬂuid T.43) by which this ﬂow is proportional to the difference between the temperatures T and Tex in and outside the pipeline. 1. dx d (1.1. e. (1. dropping liquid: water.g. e.41) is convenient to determine the temperature distribution along the pipeline length in stationary ﬂow of the transported medium. gas. then ein = Cv · T + const.41) Temperature Distribution in Stationary Flow The equation of heat inﬂow in the form (1. The factor κ (W m−2 K−1 ) in this formula characterizes the overall heat resistance of the materials through which the heat is transferred . (1. If we take Cv = const.39) or (1. ρg ∂x (1. with qn < 0 when T > Tex and qn > 0 when T < Tex . Second. To model the heat ﬂux qn the Newton formula is usually used qn = −κ · (T − Tex ). the equation of heat inﬂow can be written using the function J = ein + p/ρ representing one of the basic thermodynamic functions. of the transported medium ∂ ∂ (ein · ρS) + ∂t ∂x or ∂ ∂ (ein · ρS) + [J · ρvS] = πd · qn + ρvSg · ∂t ∂x 1 ∂p −i .g. this equation has the form ρv · dein ∼ 4 = · qn − ρvg · i. the derivative dein / dT giving its speciﬁc heat Cv (J kg−1 K−1 ). the equation of heat inﬂow can be reduced to a simpler form ∂ρvS · J ∼ ∂ρS · ein + = πd · qn ∂t ∂x in which the dissipation of mechanical energy appears to be absent. since for a relatively light medium.7 Total Energy Balance Equation 25 This means that the rate of internal energy change of the transported medium is determined by the inﬂow of external heat through the pipeline surface and heat extraction due to conversion of mechanical energy into heat produced by friction between the continuum layers. enthalpy or heat content.
44) for temperature T = T(x). (1.). (1. the medium gradually heats up. . the boundary between ground and air and so on) or the reverse. The ﬁgure shows that when the initial temperature T0 is greater than (Tex + T⊗ ).3 illustrates the distribution of temperature T(x) along the pipeline length x in accordance with Eq.26 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines from the pipe to the surrounding medium (anticorrosive and heat insulation. ˙ ˙ M = ρvS is the mass ﬂow rate of the ﬂuid (M = const. (1. Figure 1. In particular from Eq. ground. if the dissipation of mechanical energy in the stationary ﬂuid ﬂow in the pipeline with constant diameter is identical at all crosssections of the pipeline. The formula thus obtained is called the Shuchov formula. From this equation in particular it follows that the heat transfer through the pipeline wall (the ﬁrst term on the righthand side) lowers the temperature of the transported medium when T(x) > Tex or raises it when T(x) < Tex . the moving medium cools down. while when T0 is less than (Tex + T⊗ ).44) with initial condition T(0) = T0 yields πdκ T(x) − Tex − T⊗ = exp − x . With due regard for all the aforesaid Eq.44) it follows that if the heat insulation of the pipeline is chosen such that at the initial crosssection of the pipeline x = 0 Figure 1.3 Temperature distribution along the pipeline length. ˙ T0 − Tex − T⊗ Cv M (1.45). In all cases with increase in the pipeline length the temperature T → (Tex + T⊗ ). The hydraulic gradient i can sometimes be considered constant i = −i0 ≈ const. whereas the dissipation of mechanical energy (the second term on the righthand side) always implies an increase in the temperature of the transported medium..42) is reduced to the following ordinary differential equation ρCv v · 4κ dT = − (T − Tex ) + ρvgi0 dx d (1. This factor is called the heattransfer factor.45) ˙ Where T⊗ = gi0 M/πdκ is a constant having the dimension of temperature. The solution of the differential equation (1.
Through good insulation of the pipeline the oil is pumped over without preheating despite the fact that in winter the temperature of the environment is very low.36 3.1. see the cover picture). ˙ T0 − Tex − T⊗ Cv M (1. The heat insulation of the pipeline is characterized by the heattransfer factor κ = 2 W m−2 K−1 . It is required to ﬁnd the temperature at the end of the section. we obtain TL − Tex − T⊗ πdκL = exp − . The temperature of the surrounding medium is 8 ◦ C.36 K. L = 120 km.5 ◦ C. Q = 2500 m3 h−1 ).45).81 · 0. The initial temperature of crude oil (ρ = 870 kg m−3 . Cv = 2000 J kg−1 K−1 . in oil transportation along the TransAlaska oil pipeline (USA. (1. From Eq.8 · 2 Using Eq. In such a case the heat outgoing from the pipeline would be compensated by the heat extracted by internal friction between the layers. = .46) we obtain TL − 8 − 2. 55 − 8 − 2. we get the expression for the temperature distribution through the initial and ﬁnal values T(x) − Tex − T⊗ = T0 − Tex − T⊗ TL − Tex − T⊗ T0 − Tex − T⊗ x/L .46) the argument under the exponent and substituting the result in Eq. (1.45) follows the connection between the initial T0 and ﬁnal TL temperatures of the transported medium.36 2000 · 870 · (2500/3600) from which follows TL ∼ 37.7 Total Energy Balance Equation 27 the condition of equality to zero of the righthand side is obeyed − 4κ (T0 − Tex ) + ρvgi0 = 0 d ˙ ρvdgi0 g M · i0 = . where L is the length of the pipeline section. Solution. If in this formula we set x = L. (1.47) Exercise 1. i0 = 0.14 · 0. Such an effect is used. 4(T0 − Tex ) πd · (T0 − Tex ) that is the factor κ satisﬁes the condition κ= And the temperature of the transported medium would remain constant and equal to its initial value over the whole pipeline section.002) is 55 ◦ C. πdκ 3.14 · 0. pumping over a pipeline section (d = 800 mm. (1.8 · 2 · 120 · 103 = exp − . for example. Calculate ﬁrst the temperature T⊗ : T⊗ = ˙ gi0 M 9.002 · 870 · (2500/3600) ∼ = = 2.46) Expressing now from (1.
where R is the gas constant. To use Eq. (1. the equation of heat inﬂow (1.. but for a perfect gas. The extracted heat of internal friction may be ignored. T(100) ∼ 37.81 · 0. = Exercise 3. T(125) ∼ 33. (1. if the temperature at the beginning of the pipeline T0 = 60 ◦ C. For stationary ﬂow of a compressible medium.004) provided with ideal heat insulation (κ = 0)? Solution. (1.4 ◦ C. dx In the general case. e. Solution. =− ˙ dx Cp M . 100 and 125 km. Using Eq. It is required to obtain the temperature of oil pumping over the pipeline section of length 150 km in crosssections x = 50. therefore it would be better to use Eq. Regarding Cp = const.47) one should go to the limit at κ → 0. the enthalpy is a function only of temperature J = Cp · T + const. one gets T(x) − 8 = 60 − 8 30 − 8 60 − 8 x/L and T(x) = 8 + 52 · (0. T).4231)x/150 . By how much would the temperature of the oil (Cv = 1950 J kg−1 K−1 ) be raised due to the heat of internal friction when the oil is transported by an oil pipeline (L = 150 km.44) ρCv v · from which dT = ρvgi0 dx or Cv · dT = gi0 . Cp − Cv = R). where Cp is the gas speciﬁc heat capacity at constant pressure (Cp > Cv . gas. (1. and taking as before qn = −κ · (T − Tex ). dx T = gi0 L/Cv = 9.004 · 150 · 103 /1950 ∼ 3 K.3 ◦ C. d = 500 mm. In this case it is impossible to use at once Eq. that is a gas obeying the Clapeyron law p = ρRT.41) takes the form ρvS dJ = πd · qn . Substitution in this formula of successive x = 50.g. i0 = 0.28 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines Exercise 2. 100 and 125 gives T(50) ∼ 47 ◦ C. and that of the environment Tex = 8 ◦ C. = = = 2.45) since κ = 0. we transform the last equation to ˙ dT = −πdκ · (T − Tex ) Cp M dx or dT πdκ · (T − Tex ). the gas enthalpy J is a function of pressure and temperature J = J(p.46). that at the end TL = 30 ◦ C.
Cv ∼ 1700 J kg−1 K−1 ). Equation of mechanical energy balance (1.48) which is similar to the solution (1. ∂x d 3. Momentum (motion) equation (1. 1. The difference consists only in that instead of heat capacity Cv in the solution (1. = = The temperature TL of the gas at the end of the gas pipeline section is found from πdκL TL − Tex = exp − ˙ T0 − Tex Cp M with regard to which the distribution (1.8 Complete System of Equations for Mathematical Modeling of OneDimensional Flows in Pipelines This system consists of the following equations. Continuity equation (1. = exp − ˙ T0 − Tex Cp M (1. 1.1.45) for temperature distribution in an incompressible ﬂuid. T) of the medium depends not only on temperature but also on pressure. T) is explained. in particular. 2 .10) ρ ∂v ∂v +v ∂t ∂x =− ∂p 4 − τw − ρg sin α(x).15) ∂ ∂t αk v 2 2 +v· ∂ ∂x αk v 2 + P(ρ) + gz = vg · i. By the dependence J(p.50) allowing us to express the temperature through the initial and ﬁnal temperatures. so the equation of heat inﬂow has a more complex form.47) takes the form T(x) − Tex = T0 − Tex TL − Tex T0 − T x/L (1. the JouleThomson effect. ∂t ∂x 2. Note that for a real gas the enthalpy J = J(p.6) ∂ρS ∂ρvS + = 0.8 Mathematical Modeling of OneDimensional Flows in Pipelines 29 The solution of this differential equation with initial condition T(0) = T0 gives πdκ T(x) − Tex x .47) we use heat capacity Cp and the temperature T⊗ taking into account the heat of internal friction is absent (for methane Cp ∼ 2230 J kg−1 K−1 .49) (1.
p. while the number of equations is 4. S. ν. . T). Equation of total energy balance (1. v. .37) ∂ ∂t αk v 2 ∂ + ein ρS + 2 ∂x = πd · qn − ρvgS dz . qn . ˙ w w ˜ w • dependences α = f (ρ. • dependence q = −κ · (T − T ) or more complex dependences n ex representing heat exchange between the transported medium and the environment. ν. characterizing the properties of the transported medium. . Therefore there are needed additional relations to close the system of equations. i. As closing relations the following relations are commonly used: • equation of state p = p(ρ. • calorimetric dependences e in = e(p. αk . or i = f (τ ). d. elasticity. v. . ein . T. The division of mechanics in which properties of a transported medium such as viscosity. • equation of pipeline state S = S(p. T) characterizing the deformation ability of the pipeline.30 1 Mathematical Modeling of OneDimensional Flows of Fluid and Gas in Pipelines 4. v. k characterizing internal structure of medium ﬂow. T). d. T) or J = J(p. It is also necessary to consider mathematical relations describing properties of the transported medium and the pipeline in which the medium ﬂows. • hydraulic dependence τ = τ (ρ.).). . τw . v. To obtain closing relations a more detailed analysis of ﬂow processes is needed. dx αk v 2 + J ρvS 2 The number of unknown functions in this equation is 10: ρ. plasticity and other more complex properties are studied is called rheology. . .
Modeling of Oil Product and Gas Pipeline Transportation. p =  dFn / dσ. (2.31 2 Models of Transported Media Algebraic relations connecting parameters of the transported medium such as density. is directed along the normal n and has a pressing action on them. Let us consider some models. KGaA. In accordance Figure 2.1 A scheme of force interactions in a ﬂuid. The absence of tangential friction forces in the state of rest models the fact that the ﬂuid takes the shape of the vessel it ﬁlls. Thus. 2. Further classiﬁcation of ﬂuids is dependent on whether or not tangential friction forces are taken into account on exposure to ﬂuid ﬂow. the force dFn with which the ﬂuid particles on one side of the element act on the ﬂuid particles on the other side of the element is proportional to the area dσ.1) The magnitude p of this force does not depend on the surface element orientation and is called pressure.1). Each of these relations represents of course a certain schematization of the properties of the considered medium and is only a model of a given medium. If ﬂuid particles interact along the surface element dσ with the unit normal n (Figure 2. Weinheim ISBN: 9783527408337 . Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. Michael V. temperature and so on are called equations of state. pressure.1 Model of a Fluid By ﬂuid is meant a continuum in which the interaction of the contacting interior parts at rest is reduced only to the pressing force of pressure. Then dFn = −pn dσ.
= area L2 L × T2 In the SI system of units the stress τ is measured by Pa = kg m−1 s−2 . 2.1) for an ideal ﬂuid is true in the state of rest as well as in the state of ﬂow. Let for example the ﬂuid layers move as shown in Figure 2. Figure 2.2 Models of Ideal and Viscous Fluids In the model of an ideal ﬂuid it is assumed that tangential friction forces between ﬂuid particles separated by an elementary surface are absent. Then the dimension of the stress τ is [τ] = M M × L/T2 force = . In the model of a viscous ﬂuid tangential stresses resulting in ﬂuid ﬂow are taken into account. . In the model of a viscous ﬂuid it is accepted that the tangential stress τ between the layers of the moving ﬂuid is proportional to the velocity difference of these layers calculated per unit length of the distance between them. Such a schematization (or model) of a ﬂuid appears to be very fruitful when the tangential components of interaction forces.2 Illustration of the deﬁnition of the viscous friction law. that is pressure forces.2. are far smaller than their normal components. Hence expression (2. Here u(y) is the velocity distribution in the ﬂow and y is the direction of a normal to the elementary surface dσ. dy (2. not only in the state of rest but also in the state of ﬂow. In other cases when the friction forces are comparable with or even exceed the pressure forces the model of an ideal ﬂuid has proved to be inapplicable.32 2 Models of Transported Media with this there are two models: the model of an ideal ﬂuid and the model of a viscous ﬂuid. namely to the velocity gradient du/ dy: τ=µ du . that is friction forces.2) The tangential stress τ is deﬁned as the friction force between the ﬂuid layers divided by the area of the surface separating these layers.
For example.3) ν = 5 · exp[−0.4) the factor κ = ln(5/8)/(0 − 20) ∼ 0.1 kg m−1 s−1 .3) in which ν0 is the kinematic viscosity of a ﬂuid at temperature T0 and κ (1/ K) is an empirical factor. The kinematic viscosity of benzene is approximately equal to 0. (T1 − T0 ) (2. (2.2 Models of Ideal and Viscous Fluids 33 The proportionality factor µ in the law of viscous friction (2. The viscosity is found from Eq. (2.0235 · (10 − 20)] ∼ = 6. As the temperature increases the viscosity decreases.0235 is deter= mined. Then this factor is found from the relation κ= ln(ν0 /ν1 ) .4) Exercise. Solution. To use Eq.001 kg m−1 s−1 = 1 cP (centipoise). the kinematic viscosity of water is equal to 0. For example the dynamic viscosity of water is equal to 0.2.01 P = 0. therefore [ν] = µ L2 M/(L × T) = .6 cSt. Its dimension is [µ] = [τ] × T = M L×T In SI units µ is measured in kg m−1 s−1 and is expressed through poise P.3) means that the ﬂuid viscosity varies exponentially with temperature.3 cSt. The viscosity of oil and of almost all oil products depends on temperature.3) it is necessary to know the factor κ or the viscosity ν1 of the same ﬂuid at another temperature T1 . Equation (2. The factor of kinematic viscosity or simply kinematic viscosity ν of a ﬂuid is deﬁned as the ratio µ/ρ. = 3 ρ T M/L In SI units ν is measured in m2 s−1 and expressed in stokes St. The kinematic viscosity of summer diesel fuel at temperature +20 ◦ C is 5 cSt. Determine the viscosity of the same fuel at temperature +10 ◦ C. where 1 St = 10−4 m2 s−1 . .2) is called the factor of dynamic viscosity or simply dynamic viscosity. (2. To calculate the dependence of the kinematic viscosity ν on temperature T it different formulas can be used including the Reynolds–Filonov formula ν(T) = ν0 · e−κ·(T−T0 ) (2. where 1 P = 0. whereas at a temperature of 0 ◦ C it increases to 8 cSt. whereas reduction in temperature leads to viscosity enhancement. that of diesel fuel is 4–9 cSt and that of lowviscous oil 5–15 cSt.01 St = 10−6 m2 s−1 = 1 cSt (centistoke). With Eq.
34 2 Models of Transported Media 2. Of course an incompressible ﬂuid is only a model of a real medium. 2. under normal conditions water density is 1000 kg m−3 . 2. the condition of incompressibility is dρ/ dt = 0 being tantamount to dρ/ dp = 0.5 Model of a Fluid with Heat Expansion The expansion of different media on heating and subsequent compression on cooling is taken into account in the ﬂuid model with volume expansion. The compressibility factor is the inverse of the elastic modulus K (Pa). If the ﬂuid density depends only on pressure. For example for a ﬂuid with ρ0 = 850 kg m−3 at p − p0 = 5 MPa (≈ 50 atm) the deviation ρ is 2. It follows that the deviation of density ρ from the normal ρ0 : ρ = ρ0 · (p − p0 )/K is very small for oil and oil products. In this model the ﬂuid density depends on pressure as follows ρ(p) = ρ0 [1 + β(p − p0 )] (2. diesel fuel ≈ 840 kg m−3 .5 · 109 Pa. so that K ≈ 1. oil ≈ 870–900 kg m−3 etc. (2. since as is known absolutely incompressible media do not exist. that is the ﬂuid was homogeneous. namely ρ = ρ0 = const.4 Model of Elastic (Slightly Compressible) Fluid There are processes which require that account is taken of even a small variation in ﬂuid density.5) where β (1/Pa) is the compressibility factor. ρ0 the ﬂuid density at normal pressure p0 . that is ρ = ρ(p). benzene ≈ 735–750 kg m−3 . In such a case the socalled elastic ﬂuid model is often used.6) Mean values of the elastic modulus for oil and oil products vary in the range 1400–1500 MPa.8 kg m−3 . In the . Then Eq. it remains homogeneous as before.4 − 1. that is K = 1/β. but when the change in ﬂuid density in a certain process can be neglected the model of an incompressible ﬂuid may be very useful. while moving. (2. that is dρ/ dt = 0. For example. If initially all ﬂuid particles had equal densities.5) reduces to ρ(p) = ρ0 1 + p − p0 K .3 Model of an Incompressible Fluid A ﬂuid is called incompressible if its density does not vary when moving.
2.5 Model of a Fluid with Heat Expansion Table 2.1 Factor of volume expansion ξ. Density ρ, kg m−3 720–739 740–759 760–779 780–799 800–819 820–839 840–859 860–880 Factor ξ, K−1 0.001183 0.001118 0.001054 0.000995 0.000937 0.000882 0.000831 0.000782
35
model to be considered the density ρ is a function of temperature T, so that ρ = ρ(T) ρ(T) = ρ0 [1 + ξ(T0 − T)] (2.7)
in which ξ (1/K) is the factor of volume expansion, ρ0 and T0 are the density and temperature of the ﬂuid under normal conditions (often T0 = 293 K (20 ◦ C); ρ0 = ρ(p0 , T0 ); p0 = pst = 101325 Pa). The values of the factor ξ for oil and oil products are given in Table 2.1. From Eq. (2.6) it follows that on heating, that is at T > T0 , the ﬂuid expands, that is ρ < ρ0 whereas at T < T0 , ρ > ρ0 , that is the ﬂuid is compressed.
Exercise. The density ρ0 of benzene at 20 ◦ C is 745 kg m−3 . Determine the density of the same benzene at 10 ◦ C. Solution. Using Eq. (2.6) and Table 2.1 we have ρ(10 ◦ C) = 745 · [1 + 0.00118 · (20 − 10)] = 753.3 kg m−3 . Thus the density is increased by 8.3 kg m−3 . There is also used a model for ﬂuid expansion with regard to baric and heat expansion. In such a model the density is a function of pressure and temperature ρ = ρ(p, T) called the state equation
ρ(p, T) = ρ0 1 + ξ(T − T0 ) +
p − p0 . K
(2.8)
Here p0 , T0 are the nominal pressure and temperature satisfying the relation ρ0 = ρ(p0 , T0 ).
Exercise. The density of benzene ρ0 at 20 ◦ C and atmospheric pressure pat ≈ 0.1 MPa is 745 kg m−3 . Determine the density of the same benzene at temperature 10 ◦ C and pressure 6.5 MPa. Solution. Using Eq. (2.7) and Table 2.1 we get ρ(p, T) = 745 · [1 + 0.00118 · (20 − 10) + (6.5 − 0.1) · 106 /(1.5 · 109 )] = 757 kg m−3 . The density is increased by 12 kg m−3 .
36
2 Models of Transported Media
2.6 Models of NonNewtonian Fluids
Fluids modeled by condition (2.2) of viscous friction are called Newtonian viscous ﬂuids in accordance with the name of the law (2.2). The quantity γ = du/ dy having the sense of velocity gradient with dimension s−1 , is called ˙ the shear rate. The linear dependence (2.2) between the tangential friction stress τ and the shear rate γ is shown in Figure 2.3. This dependence states: ˙ ‘‘Tangential stresses arising in a medium having been modeled by a Newtonian viscous ﬂuid are proportional to the shear rate of the ﬂuid layers relative to each other. When the shear rate vanishes, the tangential friction stresses also disappear.’’ The dynamic viscosity of ﬂuid µ is represented in this model by the slope of a straight line on the plane (˙ , τ) : µ = tan ϕ, where ϕ is the angle of inclination γ of the straight line to the abscissa. Many experiments have shown that the model of a Newtonian viscous ﬂuid well schematizes processes taking place in many real ﬂuids. And yet there exist dependences of τ on γ (ﬂow curve) that differ signiﬁcantly ˙ from that depicted in Figure 2.3. Such ﬂuids are called nonNewtonian. As an example of a nonNewtonian ﬂuid is a model of a power Ostwald ﬂuid (Wilkinson, 1960) τ=k· du dy
n−1
·
du dy
(2.9)
in which the relation of the tangential friction stresses between ﬂuid layers has a power nature. In other words the apparent viscosity µ of such a ﬂuid does ˜ not remain constant as in the model of Newtonian ﬂuid, but depends on the characteristics of the ﬂow, namely µ=k· ˜ du dy
n−1
.
(2.10)
In this model k and n are factors. Fluids with n < 1 are called pseudoplastic ﬂuids. These models are applied to describe the behavior of suspension ﬂows,
Figure 2.3
A model of a Newtonian viscous ﬂuid.
2.6 Models of NonNewtonian Fluids Figure 2.4 Models of nonNewtonian ﬂuids: 1, pseudoplastic ﬂuids; 2, dilatant ﬂuids.
37
that is viscous ﬂuids with suspended small particles. Flow curves of such ﬂuids have the form of curve 1 in Figure 2.4 (Wilkenson, 1960). Fluids with n > 1 are called dilatant ﬂuids. Starch glue is an example of ﬂuid whose behavior is described by the dilatant model. Flow curves of these ﬂuids have the form of curve 2 in Figure 2.4 (Wilkenson, 1960). Model of viscousplastic medium with limit shear stress; model of Shvedov–Bingham ﬂuid (Wilkenson, 1960). There are ﬂuids in which the stresses between the ﬂuid layers are sufﬁciently well described by the following relations τ = τ0 + µ γ= ˙ du , at τ > τ0 ; dy (2.11)
du = 0, at τ ≤ τ0 ; dy du , at τ < −τ0 . dy
τ = −τ0 + µ
These expressions imply that the ﬂow of such a ﬂuid does not begin as long as the absolute value of the tangential friction stress τ does not exceed a limiting value τ0 , being the characteristic of the given medium and called the limit shear stress. In this case γ = 0. At τ ≥ τ0 and γ = 0 the medium ﬂows as a viscous ˙ ˙ ﬂuid. Real media whose properties are satisfactory modeled by the viscousplastic Bingham ﬂuid are, for example, highparafﬁnaceous and solidifying oils, mud solutions, lacquers, paints and other media. Flow curves of viscousplastic media are shown in Figure 2.5.
Figure 2.5 Flow curves of a viscousplastic ﬂuid.
38
2 Models of Transported Media
2.7 Models of a Gaseous Continuum
We proceed now to the description of the basic models used for gas ﬂow. First let us consider the properties common to all gases. One such property is that for all gases in a state of thermodynamic equilibrium there is a relation between pressure p, absolute temperature T and density ρ (or speciﬁc volume υ = 1/ρ) (p, υ, T) = 0 (2.12)
called the equation of state. The physical nature of this fact is discussed in textbooks on statistical physics and thermodynamics. In most models it is also assumed that when the motion starts the relation (2.12) remains. This means that the establishment of thermodynamic equilibrium happens much faster than the nonequilibrium caused by the resulting ﬂow. The speciﬁc form of the dependence (2.12) is set in the course of socalled calorimetric measurements, but for the majority of gases this dependence has one and the same distinctive features. Geometrical representation of the dependence (2.12) has the form of a twodimensional surface in a threedimensional space of variables (p, υ, T). Figure 2.6 shows isotherms of real gases representing intersections of this surface with planes T = const. For all gases there exists the socalled critical isotherm, depicted in Figure 2.6 by the heavy line, above and below which the properties of the gas are different. If T ≥ Tcr , where Tcr is the critical temperature for a given gas, the gas at any elevation of pressure remains in the gaseous state. If T < Tcr , then for each temperature T there exists a value of pressure p at which the gas begins to change into the liquid phase, its speciﬁc volume being increased from υ to υ , after which the resulting medium acquires the properties of a liquid. The point K is called critical point of a given gas, the quantities (Tcr , pcr , υcr ) accounting for the individual properties of a gas are constants. For example, for methane CH4 , which is the major constituent of consists natural gas, Tcr = 190.55 K and pcr = 4.641 MPa. This means that if the gas
Figure 2.6
Gas isotherms.
more general. pressure and temperature of a gas is called the Clapeyron equation. equal to 8314 J mol−1 K−1 .11) of the gas state then has the simple form p= RT υ or p = ρRT (2. Equation (2. The model of a perfect gas operates sufﬁciently well over a range of not too high pressures and moderate temperatures. that is of the spatial conﬁguration of the constituent atoms.13) where R is the only constant in the equation and is called the gas constant R = R0 /µg . the molecules behave as balls differing in their mass. namely the molecular weight µg .13) does not suit observations of gas behavior with increase of pressure and decrease of temperature.55 K. Thus all gas constants for a perfect gas depend only on the molecular weight. Figuratively speaking. and is determined only by their total mass. the isotherms of all gases are similar to each other (see the righthand part of Figure 2. in processes of gas pipeline transportation . = = for oxygen O2 (µg ∼ 32 kg kmol−1 ) R = 8314/32 ∼ 260 J kg−1 K−1 ). Hence.12) connecting the density.2 Model of a Real Gas Despite the fact that the name of this model contains the word ‘‘real’’. where R0 is the universal gas constant. therefore the number of parameters characterizing the gas decreases from three to one. schematization of a gas model.7. Under given conditions the interaction of the molecules of a real gas is independent of the form of the molecules. They are: for methane (µg ∼ 16 kg kmol−1 ) R = 8314/16 ∼ 520 J kg−1 K−1 ).6 it follows that the hyperbolic dependence (2. 2.2. 2. for carbon = = dioxide CO2 (µg ∼ 44 kg kmol−1 ) R = 8314/44 ∼ 189 J kg−1 K−1 . = = Equation (2.1 Model of a Perfect Gas If the gas pressure is not too high and the temperature not too low.7 Models of a Gaseous Continuum 39 temperature exceeds 190.7. From Figure 2.6 enclosed in the dotted oval line) and with a high degree of accuracy can be approximated by hyperbolas representing the fact that the pressure p is inversely proportional to the speciﬁc volume υ. this gas will be changed into the liquid state without the need for any pressure increase. To characterize the thermodynamic state of gases in a given range of pressure and temperature the model of a perfect gas is used. one is dealing only with the next. for air = = (µg ∼ 29 kg kmol−1 ) R = 8314/29 ∼ 287 J kg−1 K−1 .
One can ﬁnd the details of such a model for example in Porshakov et al. For a real gas Z < 1. It is evident that for moderate values of pressure and temperature Z ≈ 1 and the model (2. pcr Tr = T .86. Tcr = 190 K at p = 7.7 we determine Z ∼ 0.78Tr2 + 0. In particular. Graphs of the function Z(pr .14) takes into account not only the molecular weight of a gas through the constant R but also its thermodynamic parameters such as critical pressure and temperature. 2001.15) and (2. where pr = p .7. Tcr =1. In fact the case is to approximate the equation of state (2. Hence the model (2. Solution.15) = or pr Z(pr . This is the model of a real gas.16) are no more than approximations of the state equation of a real gas.0241 · . Tr ) ∼ 1 − 0. Tr ) are shown in Figure 2.52.4273 · pr · Tr−3. Without dwelling on the details of its derivation we note that the mathematical form of this equation is represented as follows p= Z(pr .12). (2. Tr ) ∼ 1 − 0. = θ where θ = 1 − 1. Tcr Here pcr and Tcr are the critical pressure and temperature.5 MPa and T = 288 K. Let us calculate ﬁrst the reduced parameters of state: pcr = 1. However.668 (2. It should be particularly emphasized that Eqs.14) transforms into the model of a perfect gas. Tr ) · ρRT (2. Exercise.16) . being a function of two parameters: the reduced pressure pr and the reduced temperature Tr . where the pressure may be 5–15 MPa the model of a perfect gas would give improper results were it to be used in calculations.. to simulate processes in gaspipelines we can use Z(pr . the properties of a real gas are so complicated that we do not have universal formulas appropriate for all gases over the whole range of governing parameters. There is a model of more general form than the model of a perfect gas.40 2 Models of Transported Media and underground gas storage.6 MPa. Tr ) called the overcompressibility factor. (2.14) differing from Eq. (2.63.68Tr + 0. = There are a lot of approximating formulas to calculate the factor Z. It is required to determine the overcompressibility factor Z of a gas with pcr = 4. Tr ) · RT υ or p = Z(pr . Therefore in different cases one should use different approximations.0107Tr3 . From the plot in Figure 2.13) by the insertion of the dimensionless factor Z(pr .
190 With formulas (2.7 Models of a Gaseous Continuum 41 Figure 2.7 Graphs of Z(pr .6 Tr = 288 ∼ = 1.4273 · 1.6 MPa.63 ∼ Z = 1 − 0.2861. Tcr = 190 K at p = 7.850.863. 4.15) we obtain Z = 1 − 0.668 ∼ 0. Exercise. 0.78 · 1.522 + 0.63 · 1.523 ∼ 0.52−3. (2.0107 · 1.2861 . Tr ) for natural gas. Solution.52 + 0.68 · 1. The reduced parameters of state are: pr = 7.0241 · = 0.16) gives: θ = 1 − 1. = 1.52. Determine the value of the overcompressibility factor of a gas with pcr = 4.5 MPa and T = 288 K (see the previous example).2. = Similar calculation in accordance with Eq.5 ∼ = 1.63.
Using Eq. How does the volume of the internal space of a steel pipeline section with D = 530 mm. The model of a nondeformable pipeline appears to be very useful when researching many technological processes of oil and gas transportation. The simplest model of the pipeline is a model of nondeformable pipe. In order to account for the volume expansion of a pipeline when the temperature T deviates from its nominal value T0 one can use V(T) = V0 [1 + 2αp · (T − T0 )].17) where αp is the volume expansion factor of the metal from which the pipeline is produced (for steel αp ≈ 3.5142 4 · 2 · (−5) · 3. whereas with Eq. 2. However.3 · 10−5 K−1 ). in some cases. that is of a cylinder with constant invariable internal diameter d0 and wall thickness δ. . (2. (2.16) it is less ≈ 0.3%. Experience shows that the volume of the internal space of the pipeline varies insigniﬁcantly with changes in temperature and pressure of the transported medium. The simplest formula to calculate changes originated by this phenomenon has been suggested by Joukowsky in his work ‘‘On hydraulic hammer in watersupply pipes’’ (1899).42 2 Models of Transported Media It is seen that the error in calculation with Eq.3 · 10−5 · 120 000· ∼ −8.8.15) is ≈ 2. L = 120 km change during even cooling by 5 K? Solution. Its derivation is illustrated in Figure 2. (2. the model of a nondeformable pipe proves to be inadequate to perceive the point of the phenomenon and it is necessary to use the more complicated model of an elastic deformable pipeline.8%. (2.22. The external diameter D = d0 + 2δ of the pipeline in this model remains also constant. (2.14 · 0. = that is the volume decreases by more than by 8 m3 .16) is more precise than Eq. for example in studying the phenomenon of hydraulic hammer.17) we get: V(T) − V0 = 3. The volume of the pipeline internal space is changed to a greater extent with variation in the difference between the internal and external pressures. δ = 8 mm. Exercise. (2.15) because both formulas are only approximations of the state equation of a real gas.8 Model of an Elastic Deformable Pipeline In the schematization of ﬂuid and gas ﬂow processes in pipelines models of a pipeline are also used. But this does not mean that the Eq.
Such a state of the pipe is called a planestressstate.8 by a dotted line gives the following relation σ=E· π · (d − d0 ) π · d0 (2. In such cases Eq. under the action of a pressure difference (p − p0 ) and circumferential stresses σ resulting in the pipe metal has the following form (p − p0 ) · d = σ · 2δ.19) and further replacing factor d by d0 due to the smallness of the wall thickness as compared to the pipe diameter gives the dependence of the pipe diameter increment d = d − d0 on the difference p = p − p0 of the internal and external pressures d= 2 d0 · 2Eδ p.078. (2. (2. 43 The equation of equilibrium of the upper half of the pipe shell shown in Figure 2.8 Derivation of the formula for the crosssection area change in an elastic deformable pipeline.18) into Eq. (2.21) where νP is the Poisson ratio.2.20) Here d0 may be taken as the internal diameter of the pipe. the correction is insigniﬁcant since for steel pipelines ν2 ≈ 0. P .20) it was assumed that axial stresses in the pipe are absent. Remark.19) where E is the Young’s modulus of the pipe material (for steel E ≈ 2 · 105 MPa). in particular in steel welded pipelines used in the oil industry there is a planedeformable state in which radial expansion of the pipe generates axial stresses. (2.8 Model of an Elastic Deformable Pipeline Figure 2. (2.20) should be replaced by the more general formula d= 2 d0 (1 − ν2 ) P · 2Eδ p (2. (2. Insertion of σ from Eq. However.18) Hooke’s law of elasticity as applied to the deformed middle ﬁlament shown in Figure 2. But in many cases this assumption is not valid.8 by a heavy line. When deriving Eq.
= 2 · 2 · 1011 · 0.44 2 Models of Transported Media From Eq.5143 · 120000 · 5 · 106 ∼ 40 m3 .4 mm.20) ensue two useful formulas: one for the increment S of the area of the pipe crosssection and the other for the increment V of the volume of a pipeline section with length L S= 3 πd0 · 4Eδ p. V= 3 πd0 L · 1 − ν2 P · 4Eδ p. = 4 · 2 · 1011 · 0. L = 120 km after build up of pressure by 5. It is required to calculate the increase in the diameter and volume of a section of a steel pipeline with D = 530 mm. .008 From Eq. Using Eq. (2.g. Eδ (2.5142 · 5 · 106 ∼ 0. δ = 8 mm.24) There are also more complicated models of pipelines taking into account the viscousplastic properties of the pipe material. from plastics. e.22) and associated formulas S= 3 πd0 1 − ν2 P · 4Eδ p.008 3. (2. Solution.17) follows V= For simultaneous deviation of pressure and temperature from their nominal values p0 and T0 it is allowable to use the formula V(T) = V0 1 + 2αp · (T − T0 ) + d0 · (p − p0 ) .0 MPa. Such models are applicable for pipelines made from synthetic materials.14 · 0.0004 m = 0.16) we get d= 0. V= 3 πd0 L · 4Eδ p (2.23) Exercise. (2. (2.
relation (1.1). From here it follows that the absolute value of the tangential stress τ between the ﬂuid layers is proportional to the cylinder radius r τ(r) = 1 p · · r. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. density. It was noted above that to get a closure relation it is necessary to scrutinize not only onedimensional but also spatial ﬂows taking place in the ﬂow of transported medium inside the pipeline. For example.29) expresses the tangential friction stress at the internal surface of a pipeline through average parameters of the ﬂow and so on. KGaA. 3. Weinheim ISBN: 9783527408337 . Such a ﬂow has only one axial velocity component u(r). Let us consider in greater detail such ﬂows with regard to the distribution of the parameters over the pipeline crosssection.45 3 Structure of Laminar and Turbulent Flows in a Circular Pipe In the ﬁrst chapter the ﬂow of a medium transported in a pipeline was considered in the framework of a onedimensional model. 2 L (3. Next separate inside the ﬂow region a cylinder of arbitrary radius r (r ≤ r0 ) and write the balance equation of all the forces acting on the cylinder (p1 − p2 ) · πr 2 = τ(r) · 2πrL where τ(r) is the tangential friction stress at the lateral surface of the separated cylinder.1) Modeling of Oil Product and Gas Pipeline Transportation. Michael V. that is a model in which the ﬂow is described by characteristics of the ﬂuid velocity. In such a description additional closure relations reﬂecting the relations between the average parameters of the ﬂow were needed.1 Laminar Flow of a Viscous Fluid in a Circular Pipe First let us consider the laminar ﬂow of ﬂuid in a circular pipe with radius r0 (Figure 3. relation (1. temperature and others averaged over the pipe crosssection. dependent on the radial coordinate r equal to the distance from a point under consideration to the pipe wall. pressure.26) connects the mechanical energy dissipation with the average velocity of the ﬂow. All characteristics of the ﬂow depended only on the longitudinal coordinate x of the crosssection and time t.
that is at r = 0.4) The relation (3. umax = 2 p · r0 . The ﬂuid ﬂow rate Q.46 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Figure 3. is equal to Q= 0 r0 u(r)2πr dr = 2πumax 0 r0 r· 1− r2 2 r0 dr = 1 2 πumax r0 2 or with regard to Eq.4) is called the Poiseuille formula.5) where d = 2r0 is the internal diameter of the pipe.1 Calculation of laminar ﬂuid ﬂow. that is the ﬂuid volume ﬂowing through a unit crosssection of the pipe in unit time. 4 µL (3. . 8 µL (3.2) for the velocity u(r). (3. It gives the connection between the ﬂow rate of laminar ﬂuid ﬂow in a circular pipe and the pressure drop causing the ﬂow. This equation should be solved for the boundary condition u = 0 at r = r0 . As a result we obtain u(r) = umax · 1 − r2 2 r0 . that is v = Q/πr0 . Let us introduce the mean ﬂowratevelocity v which when multiplied by the 2 pipe crosssection area gives the ﬂuid ﬂow rate. If we now replace τ by its expression through the velocity gradient du/ dr in accordance with the law of viscous friction (2. we get the differential equation −µ du 1 p = · ·r dr 2 L (3. Then one obtains very useful formulas v= 2 r0 p = 2 · umax 8 µL or v= d2 p 32 µL (3.3) Q= 4 r0 π p . which is the socalled sticking condition (the velocity at the internal surface of the pipe vanishes).3) It is seen that the ﬂuid velocity distribution over the pipe crosssection has a parabolic form with maximal value umax at the pipe axis.2) and take into account that τ(r) < 0.
(3.5) with regard to Eq. 2 4 d ρv vd/(µ/ρ) Re (3.052 /4 vd 0. (3. The tangential stress τ for this ﬂuid is related to the shear rate γ = du/ dr by the dependence (2. Calculate the hydraulic resistance factor of oil λ (ν = 25 cST) in laminar ﬂow in a circular pipe with diameter 50 mm and ﬂow rate 1 l s−1 . From Eq.51 m s−1 .6) Exercise.3.063.2 Laminar Flow of a NonNewtonian Power Fluid in a Circular Pipe Just as we considered the laminar ﬂow of a Newtonian ﬂuid so we can consider the laminar ﬂow of a nonNewtonian Ostwald power ﬂuid in a circular pipe (see Chapter 2). Re 1020 3. 2 L p/L from Eq.05 = 1020 = ν 25 · 10−6 Then calculate the Reynolds number Re = Since 1020 < Recr = 2300. = = πd2 /4 3.9) ˙ τ=k· du dr n−1 · du dr so that the equilibrium Eq. (1.30) for the factor λ of hydraulic resistance in laminar ﬂow of a viscous incompressible ﬂuid in a circular pipe λ= d 32 µv 8 64 64 · 2 = · = .29) yields λ ρv2 r0 32µv ρv2 · 2 = Cf · = · 2 d 2 4 2 Substitution of τw  = From here follows Stokes formula (1.001 ∼ 0. the ﬂow is laminar.2 Laminar Flow of a NonNewtonian Power Fluid in a Circular Pipe 47 With the help of these relations it is possible to calculate the tangential stress τw at the internal surface of the pipeline.51 · 0.6) the hydraulic resistance factor is equal to λ= 64 64 ∼ = = 0. First determine the mean velocity of the ﬂow v= Q 0. In accordance with Eq. Solution. (3.2) takes the form −k · du dr n−1 · du 1 p = · ·r dr 2 L . (3.1) follows τw  = p 1 · · r0 .14 · 0.
(3. Eq. (3. Introducing the mean ﬂowratevelocity v as in the previous section and the generalized Reynolds number Re∗ according to the relations v= Q πr0 2 and Re∗ = n v2−n · d0 k/ρ where d0 = 2r0 .48 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Since du/dr < 0. (3.7) The maximal velocity of the ﬂow umax is achieved at the pipe axis r = 0 as in the case of a viscous ﬂuid and is equal to umax = n · n+1 p 2kL 1/n n+1 · r0 n . 1 Integration of this equation with the sticking boundary condition u(r0 ) = 0 at the pipe wall gives the velocity distribution u(r) = − n · n+1 p 2kL 1/n · r n+1 n n+1 − r0 n .9) could be written in the habitual form of the Darcy–Weisbach law 1 ρv2 p · =λ· L d0 2 where the hydraulic resistance factor λ is 8· λ= 6n + 2 n Re∗ n . (3. k = µ Eq. 1960). the last relation transforms to p du =− dr 2kL 1/n ·rn.4) known as the Poiseuille law (Wilkenson.8) The ﬂow rate of the ﬂuid Q is calculated by the formula Q = 2π 0 r0 r · u(r) dr p 2kL 1/n = 2π · n · n+1 · 0 r0 n+1 r · r0 n − r n+1 n dr Performing integration one gets after some algebra Q= πr0 3 n · 3n + 1 r0 p/L 2k 1 n . (3. (3. (3.9) It is reasonable that at n = 1.9) converts to Eq.10) .
11) Exercise. The solution of this equation gives n ∼ 0. (3.4) and (3.10) at n = 1. (3.19 . 1960) (see Chapter 2 and Figure 3. The ﬂow rate Q in the viscometer of the Ostwald power ﬂuid under consideration takes the following form πr0 3 n r0 · ρg Q= · 3n + 1 2k 1 n πr0 3 n = · 3n + 1 r0 · g 2 · k/ρ 1 n .12) where τ0 is the limit shear stress. k = µ transforms to the Stokes Eq. Due to the existence of such stress the ﬂow does not begin immediately after application of the pressure difference to the .81 · 0. dr (du/dr > 0.81 + 1 9. namely 180/1000 = (2/3)3+1/n . in the second 180 s. from the device chamber through a narrow cylindrical tube (capillary).3 Laminar Flow of a ViscousPlastic Fluid in a Circular Pipe 49 In the same way Eq. This time is calculated with the use of Eqs. Since the ratio is inversely proportional to the times of the ﬂuid outﬂows. In the ﬁrst experiment the outﬂow ﬂows through a cylindrical capillary with internal radius 1 mm and in the second experiment through a similar capillary with internal radius 1. from which we get k/ρ ∼ 0.5 mm.2). From Eq.81 · 0. The most widespread are capillary viscometers. τ ≥ τ0 ) (3.9) replacing in them the pressure gradient p/L by ρg.002 4 · k/ρ 1/0. = 3. In the ﬁrst case the time of ﬂuid outﬂow is 1000 s.0013 = · 1000 3 · 0. = Using further the results of the ﬁrst experiment we get 0.11) it follows that the ratio of ﬂow rates in both cases is Q1 /Q2 = (r1 /r2 )3+1/n .30) given earlier.3. It is required to determine the constants n and k/ρ of the power ﬂuid model.81 . To determine the rheological properties of oil and oil products special devices called viscometers are often used.92 · 10−6 m2 s−1. The ﬂow of the oil is modeled by the law of a power ﬂuid.0001 3. we use the equation to determine n.3 Laminar Flow of a ViscousPlastic Fluid in a Circular Pipe Consider the laminar ﬂow of another nonNewtonian ﬂuid – viscousplastic Bingham ﬂuid (Wilkenson. (3. The tangential stress τ in this ﬂuid is related to the shear rate γ = du/dr by ˙ the dependence τ = τ0 + µ du . (3. (1. where g is the acceleration due to gravity. In order to determine the properties of oil experiments are carried out on a free outﬂow of a portion (100 ml) of oil from the viscometer chamber. Solution. Operation of all capillary viscometers is based on the determination of the time of the outﬂow of a ﬁxed portion of the ﬂuid under test.81.14 · 0.
a/r0 r0 (3. near the internal surface of the pipe (a < r < r0 ). It is evident that the following relations are valid r at a ≤ r ≤ r0 . dr a Integration of this equation over r from r = r0 to an arbitrary r with regard to boundary condition u(r0 ) = 0 yields u(r) = r0 τ0 2µ r (r/r0 )2 − 1 −2· +2 . the ﬂow of the ﬂuid starts. namely when it obeys the following inequality p · πr0 2 > 2πr0 L · τ0 or r0 · p/L > 1.14) and (3. the ﬂuid begins to move as a rigid core. just where τ(a) = τ0 .13) If the condition (3. Therefore the region (0 ≤ r ≤ a) of the ﬂow is called the ﬂow core.50 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Figure 3. On approaching the pipe center where the shear rate decreases the tangential stresses are also reduced and at some distance a from the pipe axis.13) is fulﬁlled then. · 2µ a/r0 (3. r0 a a τ0 2τ0 and = = τ(a) = τ0 = τw · .15) The ﬂow rate Q of the ﬂuid can be obtained using Eqs. pipe ends but only when this difference exceeds the shear strength.15) −Q = 2π a r0 r · u(r) dr + πa2 · u(a).2 A scheme of viscousplastic ﬂuid ﬂow. r0 r0 τw r0 · p/L τ(r) = τw · The velocity distribution in the annular region a ≤ r ≤ r0 satisﬁes the equation µ du r = τ(r) − τ0 = τ0 · −1 . (3. .14) In particular the core velocity u(a) is equal to u(a) = − r0 τ0 (1 − a/r0 )2 . 2τ0 (3.
If we accept that vcr = f (d.4) for the ﬂow rate of ﬂuid in a circular pipe (Wilkenson. If we introduce two dimensionless parameters Re = v·2r0 . 3. To determine this product one should resolve Eq. then I = 0 and consequently λRe = 64.16) could be represented as λ= 64 1 .17) with respect to λRe for each value of the parameter I (Romanova. so that at Re < Recr = 2300 the ﬂow is laminar whereas at Re > Recr = 2300 it is turbulent.30) for laminar ﬂow of a viscous ﬂuid. the conditions of transition from laminar to turbulent ﬂow are in many respects determined also by such parameters of the pipeline as the roughness of its internal wall surface . The latter conclusion is true when the critical velocity vcr of the ﬂow is determined only by ﬂow parameters such as d. This ﬂow is characterized by ﬂuctuating motions of the ﬂuid. ρ. The criterion of transition of laminar ﬂow into a turbulent one is the Reynolds number Re. generation and development of eddies and intensive mixing.16) As would be expected this relation at τ0 = 0 becomes the known Poiseuille formula (3. representing a dimensionless parameter formed by the dimension parameters characterizing the ﬂow and the pipeline: Re = vd/ν. ) then.17) converts to the known Stokes formula (1. In fact. (3. In the general case the product λRe depends on the Ilyushin number I.4 Transition of Laminar Flow of a Viscous Fluid to Turbulent Flow 51 Simple rearrangements lead to the following expression Q= 4 πr0 4 p/L 1− · 8µ 3 2τ0 r0 · p/L + 1 · 3 2τ0 r0 · p/L 4 . the Reynolds ν ·2r number and I = τ0µv 0 . µ. A lot of theoretical investigations performed by many outstanding mathematicians and physicists have been devoted to the problem of stability of laminar ﬂuid ﬂows and the determination of the conditions and criteria of the transition from laminar to turbulent ﬂow regimes. the Ilyushin number and take into account the relation 2τ0 /(r0 · p/L) = 8I/λRe. · Re 1 − 4/3 · (8I/λRe) + 1/3 · (8I/λRe)4 (3. that is Eq.3. (3. in accordance with the Itheorem the condition of laminar ﬂow transition to turbulent ﬂow takes the form vcr · d = f˜ µ/ρ d or Recr = f˜ (ε) . (3.4 Transition of Laminar Flow of a Viscous Fluid to Turbulent Flow With increase in the velocity of viscous ﬂuid ﬂow in a pipe the laminar ﬂow loses hydrodynamic stability and changes into turbulent ﬂow. the equality (3. ρ and µ(ν = µ/ρ).17) If τ0 = 0. 1977). 1960).
that is at Re > 10 000. Thus at Re < Recr (ε) the ﬂuid ﬂow is laminar. These stresses are also called Reynolds stresses. as in laminar ﬂow. Some formulas for λ in this region will be considered further. 3. Similarly. dependent on the structure and mixing intensity of the ﬂuid layers rather than on ﬂuid properties. that is of the distance between the considered point of the crosssection and the pipe axis. the factor µt in this relation represents not the intrinsic viscosity of the ﬂuid but the socalled turbulent dynamic viscosity. To describe such a ﬂow let us use the average velocity u(r) representing the true timeaveraged velocity of ﬂuid particles passing through a considered point. The presence of the additional parameter ε besides the Reynolds number means that there is not a singlevalued determined boundary of the transition from laminar to turbulent ﬂow. for example. As in laminar ﬂow. but by largescale ﬂuctuations and momentum . dr (3. The factor λ takes a stable value in establishing the developed turbulent ﬂow in the pipe. The average stress τ(r) represents the ratio of the friction force acting between macrolayers of turbulent ﬂow separated by a surface element to the area of this element. The hydraulic resistance factor λ in the transition region does not have stable values. turbulent tangential stresses τ(r) are suggested to be proportional to the gradients du/ dr of average ﬂow velocities τ(r) = µt · du . since the critical Reynolds number depends on the degree of preparation of the pipeline and the ﬂuid for the test experiments. that in this manner it was possible to lengthen the transition of laminar ﬂow to turbulent ﬂow up to a value of the critical Reynolds number equal to 6000–12 000. The turbulent dynamic viscosity µt and the turbulent kinematic viscosity νt = µt /ρ of the turbulent ﬂow is caused not by molecular friction of the separated ﬂuid layers. Reynolds who contributed greatly to the development of turbulence theory. it is only characterized by drastic lowering. named in honor of the famous English engineer O. average tangential stresses τ(r) between the ﬂuid layers moving with average velocities u(r) are introduced (Figure 3. while at Re > Recr (ε) the ﬂow is turbulent. It is known.3).3).5 Turbulent Fluid Flow in a Circular Pipe Consider turbulent ﬂow of a viscous ﬂuid developed in a circular pipe with radius r0 (Figure 3.52 3 Structure of Laminar and Turbulent Flows in a Circular Pipe where ε is the relative roughness of the pipeline internal wall surface.18) However. It is assumed that the resulting velocities are parallel to the pipe axis and independent of the radial coordinate r. different from laminar ﬂow.
then the passing of the momentum from one layer to another is compensated by equal momentum from another layer. When the layers are moving with equal velocities. dr dr 2 (3. Let the law of ﬂow have the following form du 1 · τ = νt · ρ dr (3.3 The calculation of ﬂuid turbulent ﬂow. the turbulent viscosity νt of the ﬂow depends on the molecular viscosity ν of the ﬂuid itself and parameters absolute values of the ﬁrst u and second u derivatives of average velocity with respect to the radial coordinate) characterizing the ﬂow. therefore τ = 0. In accordance with the brilliant idea of Karman it can be accepted that νt = f ν. T2 [u ] = 1 L·T the dimension of the third argument u  could be expressed through the dimensions of the two others as follows (Lurie and Podoba. Thus the intensity of momentum exchange between the layers depends on the regime of turbulent ﬂow rather than on the viscosity of the ﬂuid characterized by the factors µ or ν. then there appears between the layers a friction force accelerating the backward layer and retarding the leading one. du d2 u . Since among arguments of the function f there are only two dimensional independent variables ν and u [ν] = L .20) that is. that is du/ dr > 0. transport with eddies from one macrolayer to another. that is du/ dr = 0.19) It makes sense to suggest that the turbulent viscosity νt determined by the structure of turbulent ﬂow be dependent on the parameters of this ﬂow. If one layer moves faster than another.5 Turbulent Fluid Flow in a Circular Pipe 53 Figure 3. T . 1984): [u ] = [ν]1/3 · [u ]2/3 = 1 . Hence it follows that the turbulent viscosity µt or νt is not ﬂuid constant as its molecular analog but depends on parameters of turbulent ﬂow. The transport of momentum is perceived as a friction force acting between these layers.3. This means that the leading layer loses more momentum than it obtains from the backward layer.
if f˜ (η) = κ · η.3) is independent of the ﬂow regime (laminar or turbulent). that in most of the pipe crosssection. (3. The relation (3. (u )2 The law of turbulent friction (3.21) for turbulent viscosity can be rewritten as νt = ν · κ2 u 3 ν · u 2 = κ2 · u 3 . Therefore Eq. (3.21) The formula (3. Karman theory. (3. 1987). The last could be achieved by assuming the function f˜ to be linear.22) called the Karman formula (Loizyanskiy. however.21) shows that the turbulent viscosity νt is equal to the molecular viscosity ν of the ﬂuid multiplied by the dimensionless factor f˜ dependent only on one dimensionless parameter η = u 3 /(ν · u 2 ). (3.4 and is a universal constant of the model in the sense that it is the same for all regimes of turbulent ﬂow of ﬂuids in pipes. except for the narrow wall layer the turbulent viscosity is practically independent of the molecular viscosity of the ﬂuid.21) should have such a structure that the molecular viscosity would have no impact over almost all the crosssection of the ﬂow. The last equation represents the basic equation of the phenomenological. Experiments testify. in other words. This is because the turbulent viscosity is determined by exchange of momentum between ﬂuid layers due to largescale eddies rather than by molecular friction. With regard to the accepted assumption Eq.22) has been shown by a plethora of experiments to be approximately equal to 0.2 and 3.19) is expressed by 1 u 3 · τ = κ2 ·u ρ (u )2 (3. Let us determine now the distribution u(r) of average velocities of ﬂuid turbulent ﬂow in a circular pipe.54 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Then owing to the theorem the number of arguments of the function f can be reduced to one and the dependence (3. where κ is a constant called the Karman constant.20) takes the following dimensional form νt /ν = f˜ u 3 ν · u 2 ⇒ νt = ν · f˜ u3 ν · u 2 .2) for laminar ﬂows. First note that tangential stresses τ(r) as in laminar ﬂow are linearly distributed over the radius p 1 ·r τ(r) = − · 2 L since the balance of forces acting on an arbitrarily separated ﬂuid cylinder (see Figures 3. that is resulting from abstract reasoning though adequate for the phenomena under consideration. that is in the ﬂow core. The constant κ in Eq. Introduce into the consideration the tangential stress τw caused by the friction .22) differs signiﬁcantly from the analogous relation (2.
2 u r0 (τw < 0). u2 u r0 u κ = 2· 2 u u• κ · 2 u∗ From here it ensues that − r0 r or d dr 1 u = r0 .3) u < 0 and u < 0. thus κ2 r u 3 u u4 2 = −κ2 2 = −u∗ · .25) Solve this equation for a ﬂow in a circular pipe (Figure 3. the dynamic velocity u∗ ∼ 0. Since the factor λ is small (λ ≈ 0. The quantity (τw /ρ)1/2 has the dimensions of velocity. .03).24) for the velocity distribution u(r) over the radius. In fact it has the sense of a friction 2 stress at the pipe wall since ρu∗ = τw . In terms of dynamic viscosity the basic equation (3.05 · v. r Repeated integration gives u(r) = u∗ · κ r + C1 + C2 · ln r0 r − C2 r0 (3.3. τw = τ(r0 ) = − · 2 L Then τ(r) could be expressed through τw τ(r) = τw · r . (3.5 Turbulent Fluid Flow in a Circular Pipe 55 at the pipe walls p 1 · r0 .23) Substitution instead of τ(r) from Eq. It is commonly called the dynamic velocity and speciﬁed by u∗ .24) attains a more compact form κ2 r u 3 u 2 = −u∗ · . In addition the dynamic velocity is related to the above introduced factor λ of hydraulic resistance 2 u∗ = τw  λ u∗ 1 = Cf · v 2 = · v 2 ⇒ = ρ 2 8 v λ 8 where v is the ﬂuid velocity averaged over the crosssection.22) leads to the differential equation κ2 1 r u 3 u = τw · 2 u ρ r0 (3. r0 (3.26) where C1 and C2 are constants of integration. (3.01–0. from which = follows that the dynamic velocity is 20–25 times smaller than the average ﬂow velocity.
29) in which the subscript ‘‘w’’ indicates that the corresponding derivatives are calculated at points on the internal pipe surface.26) and taking u = 0 we obtain 0= u∗ · [1 + C1 + C2 · ln 1 − C2 ]. The ﬁrst condition is evident. The dimension analysis as applied to Eq. It is the sticking condition in accordance with which the velocity uw of ﬂuid particles should vanish at the pipe walls uw = u(r0 ) = 0. Substituting r = r0 in Eq.25) for the turbulent model has higher order than the analogous differential equation (3. uw  For small values of the roughness the righthand part of this equation could be expanded into a Taylor series leaving in it only the ﬁrst two terms ν · uw2 = g0 + g1 · uw 3 · uw  uw  (3. (3. (3. Since this condition models the ﬂuid ﬂow in a narrow wall layer it has to connect parameters uw and uw of the turbulent ﬂow at the pipe wall with the molecular viscosity ν. The problem is that the differential equation (3. and the smoothness of the internal surface of the pipe characterized by its absolute roughness . In other words the missing boundary condition should be expressed by the relation G(uw . G uw 3 · uw  uw  = 0.29) allows us to rewrite it in dimensionless form 2 ˜ ν · uw . κ (3. whose inﬂuence is strong in this layer. ν. uw .26) we get √ u∗ r r/r0 − C2 u(r) = · − 1 + C2 · ln . we get ν · uw2 ˜ =g uw 3 · uw  . Therefore to solve it one needs an additional boundary condition reﬂecting the interaction of turbulent ﬂow with the pipe walls. 1984).30) . (3.28) κ r0 1 − C2 The second condition is more complicated and was not met in the model of laminar ﬂow.27) Subtracting further the resulting equality termbyterm from Eq. (3.2) for turbulent ﬂow.56 3 Structure of Laminar and Turbulent Flows in a Circular Pipe To determine C1 and C2 one needs two boundary conditions at the internal pipe wall (Lurie and Podoba. ) = 0 (3. Having resolved the last expression with respect to the ﬁrst argument.
+√ · √ · √ κ 2 r · r0 r/r0 − C2 2 r · r0 u∗ 1 . These constants are universal like the Karman constant κ. (3. Hence.31) where k = g1 /κ2 and a = −g2 /κ are dimensionless constants. = = = In the particular case of an ideal smooth internal surface of the pipe ( ≈ 0). (3. they do not depend on concrete ﬂow and are given once and for all.31) reduces to 2 νuw = −k · u∗ . k ∼ 28 and a ∼ 0.5 Turbulent Fluid Flow in a Circular Pipe 57 where g0 and g1 are dimensionless constants. · 2r0 · κ 1 − C2 uw = u (r0 ) = Insertion of the calculated derivative into condition (3.31) yields the equation − 2 ν · u∗ k · u∗ 1 = · . uw  = κ · uw /u∗ .31) permits us to obtain the second integration constant C2 in expression (3.30) simpliﬁes to 4 2 κ · uw ν κ2 · uw · · = g0 + g1 · 3 2 uw  uw uw  u∗ and takes the ﬁnal form (uw < 0): ν · uw = − 2 k · u∗ 1 + a · (u∗ · /ν) (3.4. If we now take into account that at r = r0 in accordance with Eq. the condition (3.25) the relation κ2 · 4 uw 2 = u∗ uw2 2 takes place. A large body of calculation results for different types of ﬂows correlated to each other gives k = 28 and a = 0. 1984). meaning that. 2r0 · κ 1 − C2 1 + a · u∗ /ν which gives C2 = 1 + 1 + a · u∗ · /ν 2kκ · r0 u∗ /ν . at the pipe walls.32) Thus the boundary condition (3.31 (Lurie and Podoba. As a result we have u = u∗ C2 1 1 .28) for the velocity u(r).31.3. the theory of ﬂuid turbulent ﬂow in a circular pipe is based on three phenomenological constants κ ∼ 0. the boundary condition (3. In other words they are phenomenological constants of the model. that is.
(3.35) √ kκ · Re · λ/8 Figure 3. the top curve to Re = 3 200 000. 1987). The dotted curve depicts the parabola (3.4 · 104 · 10−2 (3. the middle one to intermediate values of the Reynolds number (Loitzyanskiy. The lower curve corresponds to Re = 23 000.31 · 10−3 · 104 · 10−2 ∼ = 1. Eq. (3. The integration constant C2 in Eq. Figure 3. 28 · 0.24) √ as u∗ = v · λ/8.4 Dimensionless average velocity proﬁles in turbulent ﬂows (Loitzyanskiy. The comparison of laminar and turbulent velocity distributions shows that the turbulent velocity proﬁle has more plane form and the greater the Reynolds number the more the plane become curves. 1987).33) can be represented as √ 1 + a · ε · Re · λ/8 C2 = 1 + . .33) gives the distribution of the average velocity of turbulent ﬂow in a circular pipe under the condition that the dynamic velocity u∗ is known. in dimensionless form C2 = 1 + 1 + a · ε · Re · u∗ /v kκ · Re · u∗ /v (3. (3. For the laminar ﬂow regime in accordance with Eq.34) Hence. The latter in its turn can be expressed through the average ﬂow velocity v and the hydraulic resistance factor λ with Eq.4 shows dimensionless turbulent velocity proﬁles u(r)/umax related to the maximal value of the ﬂuid velocity at the pipe axis. At the same time note that for turbulent ﬂows with Re > 104 .5) umax /v = 2. C2 ≈ 1 + 1 + 0. Re = vd/ν the Reynolds number. (3.33) where ε = /d is the relative roughness.28) with constant (3.58 3 Structure of Laminar and Turbulent Flows in a Circular Pipe or.001. (3. For the turbulent regime this ratio is far less. u∗ /v ≈ 10−2 the constant C2 is close to one.3) giving the velocity distribution in laminar ﬂow regime. v the mean ﬂow velocity.25. d = 2r0 the pipeline diameter. ε < 10−3 . In general it depends on the numbers Re. ε and the average ratio is equal to umax /v = 1.15 − 1.
5 Turbulent Fluid Flow in a Circular Pipe 59 On the other hand the average velocity is by deﬁnition equal to v= Q 1 = · 2 2 π · r0 π · r0 r0 0 2πr · u(r) dr = 2 · 2 r0 r0 0 r · u(r) dr.38) √ = 0. 2.3164 λ= √ 4 Re mentioned above and called the Blasius formula. (3. (3. 1.39) λ Its approximate solution takes the form 0.37) √ λ κ 60 1 + a · ε · Re · λ/8 Inserting in Eq.8 √ = 0. that is in fact on the average velocity v. Substituting here the distribution (3.143 .36) η η − 1 + C2 · ln = · κ C2 − 1 0 where η = r/r0 .37) the numerical values of the constants k. At very large Reynolds numbers (Re > 105 ) the equation for λ acquires the form 1 1 − 0.35) for C2 we get the dependence of the hydraulic resistance factor λ on the Reynolds number.88 · ln(Re · λ) − 0. (3.88 · ln √ − 0.g.28) and taking into account the formula (3.143 .88 · ln 0.11 · ε · Re · λ called the universal resistance law. e.39) is true when Re < 2 · ε−1. and on the relative roughness ε of the internal surface of the pipe √ 2u∗ r0 C2 − r/r0 r v= 2 r − 1 + C2 · ln dr r0 C2 − 1 κr0 0 √ 1 C2 − η 2u∗ √ dη (3.36) with regard to remark (3.3.8.34).93 · ε−8/7 ≈ 2 · ε−1. then the equation for λ becomes √ 1 √ = 0. we obtain √ 1 kκ · Re · λ/8 137 8 = · ln − . that is we assume √ √ √ 0.8 λ 1 + 0.11 · ε λ (3.11 · ε · Re · λ 1. The estimation of allowable Reynolds numbers can be conducted as follows √ 1 Re · λ ∼ 0.40) . If we ignore the effect of roughness.11 or Re · λ < ε−1 . = ε therefore Eq> (3. κ and a we get the equation for the dependence of λ on the numbers Re and ε √ Re · λ 1 (3.56 · Re7/8 < ⇒ Re < 1. 0. Calculating the integral (3.11 · ε · Re · λ < 0.
Solution. (0. (0.019.415 m s−1 πd2 3. = 0. Exercise 1.8 √ = 0.02) = −0.52 vd 1.415 · 0. (0.279.02. = 0.88 · ln − 0. λ = 0.123.016. (0.004. The ﬂow rate of the ﬂuid is Q = 500 m3 h−1 .0186) = −0.5 = 176 875 = ν 4 · 10−6 0. Consider the function √ 176 878 λ 1 (λ) = √ − 0. We have λ = 0.60 3 Structure of Laminar and Turbulent Flows in a Circular Pipe where λ is independent of Re. .0185.0005 500 the Reynolds number Re = and the relative roughness ε= d = As a result we get the transcendental equation for λ √ 176 875 · λ 1 √ = 0.11 ε λ λ ∼ 0.0005 · 176 875 · λ We look for the solution of this equation by the method of successive approximations. It is required to calculate the hydraulic resistance factor λ in the ﬂow of benzene with ν = 0.0186. The ﬂow rate of the ﬂuid is Q = 1000 m3 h−1 . (0.2 mm.8 λ 1 + 9.88 · ln = 1 + 1.086.0185) = 0. We have 1 1 1 + 0.41) .018. λ = 0.0186.88 · ln √ + 0.14 · 0.14 ε −2 −2 ⇒ (3. Determine the average velocity v of the ﬂow v= 4Q 4 · 1000/3600 = = 1. = Exercise 2.728 · λ representing the difference in both parts of the resulting equation. It is required to calculate the hydraulic resistance factor λ in the ﬂow of diesel fuel with ν = 4 cSt in a pipeline with d = 500 mm.25 mm. λ = 0.88 · ln √ − 0.019) = −0.8 λ 1 + 0.25 = 0.88 · ln 0. λ = 0.11 · 0.018) = 0.6 cSt in a pipeline with d = 361 mm. It is seen that λ ∼ 0.
8.2 = = 0.8.0172. = ν 0.02) = −1.88 · ln √ + 0. 1959 0.36.018) = −0. if in the boundary condition (3. (0.003.049.017) = 0.016) = 0. This equation may be obtained on the basis of the Karman model (3. λ = 0. (0.358 · 0.88 1 √ = 3/4 ln Re · n λ λ 4 1−n/2 − 0. (0.4 · λ representing the difference between both parts of the resulting equation.0172.017.4 · λ ε= We look for the solution of this equation by the method of successive approximations.88 · ln √ − 0. d 361 As a result we get the transcendental equation for λ √ 1 817 063 · λ √ = 0. πd2 3. Consider the function √ 817 063 λ 1 (λ) = √ − 0.3. Determine the average velocity of the ﬂow 4 · 500/3600 4Q = = 1.29.016.6) the following expression for the universal resistance law has been obtained by Dodge and Metzner.14 · 0. As is known k is equal to 28 for a Newtonian viscous ﬂuid. λ 1 + 49. It is seen that λ ∼ 0.018. λ 1 + 49. (0.361 = 817 063. .2 (3.4 n1.02. (0.32) we take the factor k to be dependent on the exponent n. λ = 0.6 · 10−6 and the relative roughness Re = 0. We have λ = 0. = Turbulent Flows of NonNewtonian Fluids For the turbulent ﬂow of a power ﬂuid (see Section 2.25). λ = 0.5 Turbulent Fluid Flow in a Circular Pipe 61 Solution.0172) = 0.42) where n is the exponent in the Ostwald rheological law. that is k = k(n). λ = 0.358 m s−1 .17.00055.3612 the Reynolds number v= vd 1.
due to viscosity forces. This effect is called by the name of the discoverer. The universal resistance law for a viscousplastic ﬂuid was suggested by Potapov (1975) 1 8He √ = 1− λ Re2 λ √ 8He · 0.234 · n − 5. at 0.83. where k(n) = 21−n/2 · exp n · (2.88ln(Re λ) − 0.62 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Romanova (1989) gave the resistance law for a power ﬂuid in the form 0. They are caused by the forces of internal friction between the layers of the moving ﬂuid.2/n1.6 A Method to Control Hydraulic Resistance by Injection of AntiTurbulent Additive into the Flow Friction losses are the key reason for electric energy expenditure on ﬂuid and gas pumping along pipelines. 3.76 · λ Re2 (3.2 )/0.0. in other words into heat.5 < n ≤ 1.698 · n − 1. .8 + 2. where k is the kinematic consistency.80 · 10−2 λRe 1/5 at 0. The mechanical energy of average motion is transformed ﬁrst into the energy of largescale eddies of the turbulent medium. into the heat energy of the ﬂuid. In laminar and turbulent ﬂows there occurs the socalled dissipation of mechanical energy of ordered motion and its transition into the energy of chaotic motion of the ﬂuid particles.94 · 10−2 λRe1/4 = 0. One such method discovered by the English scientist Toms in the late 1940s consists in injection into the turbulent ﬂow of special highmolecular weight additives to lower the hydraulic resistance.43) where Re = vd/ν is the Reynolds number and He = τ0 d2 /ρν2 is the Hedstroem number. For turbulent ﬂows this transition has a multistage character.2 ≤ n ≤ 0. then into the energy of the ﬂuctuation motion of smallscale eddies and ﬁnally.25. = 0. Therefore engineers and scientists involved in the problem of pipelines have for a long time been interested in methods of governing the turbulent ﬂow structure with the aim of reducing energy losses.83 − 0.13 · 10 −2 In all these formulas the generalized Reynolds number is deﬁned as Re = v2−n · dn /k.5.25 < n ≤ 2. at 1.353 · n − 3.88 1 ln k(n)Re · √ = n λ λ 8 1−n/2 − 2. In the same publication the following formulas were suggested as approximate solutions of the above equation λRe1/3 = 0.88 . the Toms effect.
Thus the universal quantity k which was taken earlier as constant would be. It was found that k ∼ 28. In the absence of antiturbulent additive (θ = 0) then k(0) = 28. Hence..6 Controlling Hydraulic Resistance by Injection of AntiTurbulent Additive into the Flow 63 The mechanism of operation of all varieties of antiturbulent additives is based on damping turbulent ﬂuctuations near the internal surface of the pipeline by interaction of the longlength molecules of the additive with turbulent eddies generated near the pipeline wall. The decrease in hydraulic resistance reduces expenditure on electric power by 20–60%. The best known antiturbulent additives for oil products are CDR produced by the American company DuponConoco and NECCAD547 produced by the Finnish company Neste. As a rule this effect is achieved with very small concentrations of additives.31) and (3.1 40 143 50 187 60 249 70 276 80 340 90 380 . These products are based on hydrocarbons. ppm k(θ) 20 61. 1999) dependent on the concentration of antiturbulent additive θ. Table 3. Owing to the damping of the nearwall turbulence there is a reduction in the ﬂow hydraulic resistance caused by the pipeline wall. a function of θ. = The effect of the antiturbulent additive is that it changes the intensity of the wall turbulence. in the presence of antiturbulent additive. Both antiturbulent additives were put through production tests on pipelines in Russia. measured commonly in ppm (parts per million of the ﬂuid volume to which the additive is added). that is k = k(θ). that is the additive acts on the magnitude of the constant k. Therefore. increased pumping efﬁciency is achieved with conservation of the pressure drop or pressure lowering at pumping stations. The results of tests of antiturbulent additive CDR have given the dependence k(θ) shown in Table 3.37) containing the constant k related to the turbulent ﬂow interaction with pipeline walls was used (see boundary conditions (3. The ﬁrst is equally suitable for benzene and diesel fuel pumping while the second is recommended primarily for diesel fuels.4 30 95. Therefore it is reasonable to accept as a model of turbulent ﬂow with antiturbulent additive a model with variable constant k (Ishmuchamedov et al. their degradation is especially great when passing through pumps. θ. It is rational to use antiturbulent additives to build up the carrying capacity of certain pipeline sections. ﬁrst the limiting ones. The use of antiturbulent additives has some speciﬁc restriction: during prolonged action of the additives in turbulent ﬂow they become degraded.3. All antiturbulent additives reduce the hydraulic resistance factor λ.1.1 Dependence k(θ) for CDR. To calculate this factor the universal resistance law (3.32)). when using additives it is necessary to inject a portion of fresh additives into the ﬂow after each pumping station.
7%. This value is signiﬁcantly less than 0.39) at Re = 40 000 and k(40) = 143 taken from Table 3.88· λ Re · λ √ (3. found by method of successive approximations. 1999).0153. The data cited in Tables 3.2 could be improved by changing the antiturbulent additive composition.88 · ln(340 · 40 000 · λ) − 3. The formula (3.. From Eq. The pumping of diesel fuel with antiturbulent additive CDR with θ = 40 ppm is conducted at Reynolds number 40 000. found by the method of successive approximations.745· λ √ 1 k(θ) = √ · e 0. θ. This value is signiﬁcantly less than 0.1 and 3.745. λ Its solution.2 we obtain the concentration θ of the antiturbulent additive .44) which permits the determination of k(θ) for given λ.64 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Table 3.2.1 gives for λ the following transcendental equation √ 1 √ = 0. λ Its solution. The effect is about 31. yields λ = 0.0129. Exercise 2. Solution. It is required to calculate the hydraulic resistance factor λ. Then with the help of Tables 3. yields λ = 0.2 gives for λ the following transcendental equation √ 1 √ = 0.39) at Re = 40 000 and k(180) = 340 taken from Table 3. It is required to calculate the hydraulic resistance factor λ. (3.0224 corresponding to the value of λ in the case of antiturbulent additive absence in the ﬂow of oil product at the same Reynolds number.745.0224 corresponding to the value of λ in the case of antiturbulent additive absence in the ﬂow of oil product at the same Reynolds number. The formula (3. ppm k(θ) 40 50 60 75 100 150 180 340 The dependence k(θ) for antiturbulent additive NECCAD547 is shown in Table 3.39) for k(θ) follows the relation 1+3. Exercise 1.1 and 3. Solution.88 · ln(143 · 40 000 · λ) − 3. In order to select the necessary concentration θ of antiturbulent additive one can proceed as follows (Ishmuchamedov et al. Remark. The pumping of diesel fuel with antiturbulent additive NECCAD547 with θ = 180 ppm is conducted at Reynolds number 40 000. The effect is about 42.2 Dependence k(θ) for NECCAD547.4%.
Since the carrying capacity must be increased by 30%.3.0126.1 we ﬁnd that this value of k corresponds to θ = 60 ppm of the antiturbulent additive.7 Gravity Fluid Flow in a Pipe The above considered ﬂows of ﬂuid are pertinent to the class of the socalled enforced (pumped) ﬂows.7 Gravity Fluid Flow in a Pipe 65 in the oil product. since the motion of the ﬂow was forced.14 · 0. The constant k(θ) is determined by Eq. θ) · v2 . Let us calculate the initial values of the pumping rate v0 . Answer. From Table 3.0126 ≈ 246. Reynolds number Re and the factor of hydraulic resistance λ0 : v0 = 4Q/S = 4 · 450/(3600 · 3. Exercise 3.3612 ) = 1. 3.44) k(θ) = 63 669 · 1 √ · √ 1+3.745 0. 0) · v0 = λ(Re. (3. Exercise 4. Multiplying the latter by the total volume of the ﬂuid to be pumped we get the required amount of the antiturbulent additive.361/(9 · 10−6 ) = 48 976. This relation gives a new value of λ λ = λ0 · (v0 /U)2 = 0. that is to .3)2 = 0. = Due to the invariability of the pressure resource it should be 2 λ0 (Re0 . the new values of the pumping rate v and the Reynolds number Re will be v = 1. Determine the amount of antiturbulent additive NECCAD547 needed to do this. λ0 = 0.587 m s−1 . It is required to increase by 25% at a given pressure resource the carrying capacity of an oilpipeline section with D = 530 mm. Re0 = v0 d/νd = 1.0126 √ 0. About 340 ppm.0213.88· 0.0213 · (1/1.0126 e 0.221 m s−1 . It is required at a given pressure resource to increase by 30% the carrying capacity of an oilpipeline with D = 377 mm. ⇒ Re = 1.3 · Re0 = 63 669. Solution. δ = 8 mm pumping diesel fuel with νd = 9 cSt and ﬂow rate 450 m3 h−1 .3 · v0 ∼ 1. δ = 8 mm pumping diesel fuel with νd = 9 cSt and ﬂow rate 950 m3 h−1 .221 · 0. Determine the amount of antiturbulent additive CDR needed to do this.
for example d. µ. sin α. The hydraulic radius is deﬁned as the ratio between the area S of a part of pipe crosssection ﬁlled with ﬂuid and the wetted perimeter Ps (Figure 3. ρ. is often introduced. Among six arguments there are three dimensionally independent ones. However. 1934). Our interest is the dependence of the ﬂow rate Q = v · S on the governing parameters.45) where ν = µ/ρ is the kinematic viscosity and ε = /d is the relative roughness of the internal surface. . ρ. that is the form of the function Q = f (S.5 A scheme of gravity ﬂuid ﬂow in a pipe.6).66 3 Structure of Laminar and Turbulent Flows in a Circular Pipe overcome the friction force a pressure gradient was needed. In the theory of gravity ﬂuid ﬂows a parameter Rh . Considering the stratiﬁed gravity ﬂow of a viscous incompressible ﬂuid along a section of the pipeline with diameter d and roughness of the internal wall surface inclined at an angle α to the horizontal (Figure 3. d. ) takes the form Q/S g sin α · d = f˜ S g · sin α . there are ﬂows in which the primary driving force could be a component of the gravity force. g. the ﬂuid ﬂows over the lower part of the pipe whereas the upper part of the pipe is ﬁlled with vapor and gases evolved from this ﬂuid. ρ. µ. ). g.6 Deﬁnition of the hydraulic radius. sin α. We can write the sought dependence in dimensionless form guided by dimensional theory. Figure 3. µ. 2001) and the function Q = f (S.ε d2 ν2 /d3 (3. d. . called the hydraulic radius (Leibensone et al. In these ﬂows the ﬂuid moves without completely ﬁlling the crosssection. A variety of gravity ﬂows are gravity stratiﬁed (divided into layers) ﬂows. Hence the number of independent arguments in dimensional V would be reduced from six to three (Lurie. Such ﬂows are called gravity ﬂows. Figure 3..5).
46) and Ps = r0 · φ where ϕ is the central angle where is seen the part of crosssection ﬁlled with ﬂuid and r0 = d/2 the pipe radius.7 Gravity Fluid Flow in a Pipe 67 1 1 1 · r0 2 φ − · r0 2 sin φ = · r0 2 (φ − sin φ). 1947 and Christianovitch.45) without disturbing generality could be written in equivalent form Q/S g sin α · Rh or Q = S · CSh · Rh · sin α (3.3.ε d2 ν2 /d3 when in the last expression we replace the pressure gradient p/L responsible for enforced ﬂow with the rolling down component of the gravity force ρg sin α causing motion in the case of gravity ﬂow. . f˜1 is a dimensionless factor dependent on the parameters of the ﬂow regime gd sin α · d/ν. 1938). The formula (3. ε) is the socalled dimensionless Chezy factor. that is on the angle ϕ.48) with the Darcy–Weisbach relation shows that the hydraulic resistance factor λ in stratiﬁed gravity ﬂow is related to the factor f˜1 by the equality λ= 2 · (d/Rh ) f˜ 2 1 . (3. Comparison of Eq.48) = f˜1 S g · sin α .47) could be represented in a form analogous to the Darcy–Weisbach relation for enforced ﬂow ρg sin α = 2 · (d/Rh ) 1 ρv2 · · d 2 f˜12 (3. 2 2 2 S 1 = (φ − sin φ). the dependence (3. degree of ﬂuid ﬁlling S/d2 and the smoothness parameter of the internal surface of the pipe ε (Archangelskiy.47) √ where CSh = g · f˜1 (S/d2 . gd sin α · d/ν. r0 2 2 S= Since ˘ AB = r0 · φ then Rh = S 1/2 · r0 2 (φ − sin φ) r0 sin φ = = · 1− PS r0 · φ 2 φ (3. Since S/d2 and Rh /d depend only on the degree of the section ﬁlling with ﬂuid. If the pipe is completely ﬁlled with ﬂuid (ϕ = 2π). then Rh = r0 /2 = d/4.
. Re = 4vRh /ν and Rh the hydraulic radius related to the degree of ﬁlling of the pipe crosssection by formulas (3. The pressure inside the vaporgas cavity remains practically invariable and equal to the saturated vapor tension (pressure) pv . Avoiding detailed treatment of these formulas let us only say that in a ﬁrst approximation it is possible to use Eq. then inside the section continuously appear cavities ﬁlled with ﬂuid vapor. λ (3.7 A scheme of pipeline gravity ﬂow. is called gravity ﬂow.7). 1934).46). Pipeline Sections of Gravity Flow When the pressure in the pipeline section is equal to the saturated vapor tension of the transported ﬂuid. where λ = λ(Re. ε). x2 ] of the pipeline in which it moves under the action of the gravity force partially (incompletely) ﬁlling the pipeline crosssection while the remainder is ﬁlled with vapor of this ﬂuid. to give CSh = 8g/λ. Figure 3.49) replacing d by 4Rh . The beginning of the gravity ﬂow section x1 is called the transfer section. The transfer section always coincides with the region of the pipeline proﬁle peak. The latter ﬂow regime is called slug ﬂow.68 3 Structure of Laminar and Turbulent Flows in a Circular Pipe from which follows the expression for the Chezy factor CSh = 2g · (d/Rh ) . In spite of this the difference in pressures between the sections x1 (the beginning of the gravity ﬂow section) and x2 (the end of the gravity ﬂow section) nevertheless exists.49) There are many empirical formulas for the Chezy factor for pipes with circular as well as noncircular crosssections (Leibenson et al.. In this case the ﬂow could be stratiﬁed or could have a more complicated structure in which portions of ﬂuid alternate with vaporgas cavities (bubbles). (3. it is merely equal to the difference in geometrical heights (z1 − z2 ) of these sections (Figure 3. Stationary gravity ﬂow can exist only on descending sections of the pipeline. The ﬂow in a section [x1 .
87 · λ0 .65 · 0. This dependence can be obtained from Eqs.7 Gravity Fluid Flow in a Pipe 69 The line of hydraulic gradient of the gravity ﬂow section passes parallel to the pipeline proﬁle at the distance pv /ρg over it.04. (3.7002 ) = 0. The proﬁle of the section is inclined to the horizontal at an angle αp = −1◦ . (3.0219 · 1/0. λ0 = 0.113. √ 2 2.650 m s−1 . since the area S of the crosssection part of each gravity ﬂow section ﬁlled with the ﬂuid is less than the area S0 of the complete crosssection of the pipeline. The oil ﬂow rate (ν = 8. Exercise.3. = 0.6 cSt) in the gravity ﬂow section of an oilpipeline (D = 720 mm.6 · 10−6 ) ∼ 52907.0175. = ◦ Determine the parameter γ: γ = i0 / tan αp  = 0. σ = 9.48) solving them with respect to S.7/(8. at γ < 4.356 0.2 mm) is 900 m3 h−1 . the Reynolds number Re. In this case the pipeline crosssection is completely ﬁlled with ﬂuid. 1999): 1.7 · 0. the factor of hydraulic resistance λ0 and the hydraulic gradient i0 in the pumping sections of the pipeline: v0 = 4 · 900/(3600 · 3. 3. .81) ∼ 0.0007. The ﬂuid ﬂow rate in the gravity ﬂow section in the stationary ﬂow regime is equal to the ﬂow rate Q of ﬂuid in the ﬁlled sections of the pipeline Q = v0 S0 = v · S (3.51) .39 · 10−2 · 4. = 2 i0 = λ0 · 1/d · v0 /(2 · g) = 0.0219. the degree σ = S/S0 of pipeline ﬁlling with the ﬂuid depends on the ratio γ = i0 / tan αp  between 2 the hydraulic gradient i0 = λ0 · 1/d · v0 /ρg of the pipeline sections completely ﬁlled with ﬂuid and the absolute value of the gravity ﬂow section slope αp to the horizontal.87 · λ0 ≤ γ < 32. It is required to ﬁnd the degree of pipeline crosssection ﬁlling with oil in this section.. tan 1 = 0. To calculate the degree σ of pipeline section ﬁlling with ﬂuid it the following approximation formulas have been suggested (Ishmuchamedov et al.50) from which may be concluded that the velocity of ﬂuid ﬂow v in the gravity ﬂow section exceeds the velocity v0 of the ﬂuid in the pipeline sections ﬁlled by the ﬂuid.32 · λ0 ≤ γ < 1.14 · 0.0007/0. σ = 1 at γ = i0 / tan αp  ≥ 1. If the ﬂow of ﬂuid in the gravity ﬂow section is stratiﬁed.47) or (3.0175 = 0. Solution. at 4. that is v = v0 · S0 /S > v0 .1825 · 2γ λ0 2γ + λ0 0. δ = 10 mm.98 · 10−2 · λ0 · (1 − γ). σ = 0. Re = 0. σ = 1 − 2. at 32. Consequently the hydraulic gradient i of the gravity ﬂow section is equal to the slope of the pipeline proﬁle to the horizontal i = tan αp .32 · λ0 .652 /(2 · 9. Calculate the pumping rate v0 .
. Consequently. Continue to build the line of the hydraulic gradient.04 < 4. Its beginning at the point 1 is a transfer section.356 = 0. The point K1 represents the end of the second gravity ﬂow section. It turns out that this line at the point K1 approaches the pipeline proﬁle at the distance pv /ρg for the second time. in formulas (3. 1999). To answer this question we need to build a combined picture of the pipeline proﬁle and the hydraulic gradient line. the pressure inside the pipeline again becomes equal to the pressure of the saturated vapor and in the pipeline there should exist vaporgas cavities.70 3 Structure of Laminar and Turbulent Flows in a Circular Pipe Since γ = 0. that is in fact parallel to the segment BK2 . The line of hydraulic gradient K2 2 at this section is parallel to the pipeline proﬁle. If the line of the hydraulic gradient is everywhere over the pipeline proﬁle and the amount by which it exceeds it is the quantity pv /ρg. the considered section of the pipeline is approximately 29% ﬁlled. therefore crosssections of the pipeline section are completely ﬁlled. It is called a transfer section because it is Figure 3. Hence. thus the point K2 represents the end of the ﬁrst gravity ﬂow section. If the line of hydraulic gradient at any point approaches the pipeline proﬁle closer than pv /ρg or even intersects it. To do this it is sufﬁcient to know the pressure and the hydraulic gradient at the end of the section.0219)0. = that is.1825 · (2γ/λ0 )0.04/0. .29.1825 · (2 · 0.8 where a section of pipeline OO1 is depicted.87 · λ0 = 0.51) using the fourth case: σ = 0. How to determine whether or not there are gravity ﬂow sections in pipeline sections under consideration. Now let us consider Figure 3.356 ∼ 0. However. where pv is the saturated vapor pressure of the ﬂuid. one of the gravity ﬂow sections is found. The line of the hydraulic gradient in the segment B K2 lies signiﬁcantly over the pipeline proﬁle. the line of the hydraulic gradient at the point K2 approaches the pipeline proﬁle up to a distance pv /ρg.8 A scheme to determine the location of a pipeline gravity ﬂow section. Let us begin to build the line BK2 2 K1 1 A of the pipeline hydraulic gradient from the end O1 of the pipeline section under consideration. then there exists one or several gravity ﬂow sections in the pipeline (Ishmuchamedov et al.1067. It leaves the point 2 at an angle whose tangent is equal to the hydraulic gradient. then gravity ﬂow sections in the pipeline are absent.
ν = 8. Hence.2 ∼ 14. = Then determine the head losses in the pipeline section between 120 and 140 km. However.00071. thus it is evident that at some crosssection the pressure of oil will be equal to the saturated vapor pressure of oil pv . Exercise.5 cSt. They are h120 – 140 = i0 · 20 000 = 14.3.3 x.8 it follows that the presence of gravity pipeline sections leads to enhancement of the initial hydraulic head H1 (and consequently the pressure p1 ) at the station and therefore requires higher expenditures of energy for pumping as compared with a pipeline in which such sections are absent. the second gravity ﬂow section K1 1 is found.5142 ) ∼ 0.7 Gravity Fluid Flow in a Pipe 71 sufﬁcient to deliver the transported ﬂuid to the point 1 so that it reaches then the end O1 of the section by itself with the help of the gravity ﬂow. It is required to determine the pressure at the beginning of the section.02 MPa) with ﬂow rate 400 m3 h−1 is pumped along an oilpipeline (L = 140 km.14 · 0. it would be possible ˜ to determine the hydraulic head H1 needed to pump ﬂuid with the same ﬂow rate in a pipeline of the same length and with the same diameter but without ˜ gravity ﬂow sections.536 m s−1 . the difference in height at the descending section is 100 m (see the proﬁle of the pipeline).514 · 0. Therefore the head at the end of the slope.43 m. Solution. m 0 100 80 100 120 0 140 0 .34 m. = 0.025 · 1/0.2 m. λ0 = 0. the pipeline at the crosssection x = 120 km is still ﬁlled with oil.3.81) = 2. = Re = 0. pv = 0.025. δ = 8 mm. km z. = ∼ Since pv /ρ · g = 20 000/(870 · 9. It is evident that H1 ≥ H1 .2 mm). From Figure 3.2 MPa.5 · 10−6 ) = 32 412. If the line of hydraulic gradient beginning from the point K2 were to be lengthened up to the initial crosssection of the considered pipeline section.81) ∼ 0.536 · 0. that is at the crosssection x = 120 km is equal to 0. Calculate ﬁrst the hydraulic gradient. The line K1 1 of the hydraulic gradient at this section passes parallel to the pipeline proﬁle at the distance pv /ρg from it.81 + 14.2/514 + 68/32 412)0. At the section 1 A the line of the hydraulic gradient is parallel to its segments BK2 and 2 K1 having been built for completely ﬁlled pipeline segments. v0 = 4 · 400/(3600 · 3.5362 /(2 · 9.25 ∼ 0.2 · 106 /870 · 9.514/(8. so that a part of the descending pipeline Table 3. The pressure at the end of the section is 0. = i0 = 0. The proﬁle of the section has the form represented in Table 3.11 · (0. Oil (ρ = 870 kg m−3 . D = 530 mm.
= .8 ∼ 484 771 Pa or ≈4. so that the loss of hydraulic pressure is h0 – 80 ∼ 56.8 m.72 3 Structure of Laminar and Turbulent Flows in a Circular Pipe section will inevitably become a gravity ﬂow section. Therefore the pressure p1 at the beginning of the section is = equal to 870 · 9.95 atm. The hydraulic gradient at the plain (completely ﬁlled) pipeline segment between the section beginning and the 80th kilometer is equal to the hydraulic gradient at the completely ﬁlled pipeline segment between 120 and 140 km. that is 0.71 m km−1 .81 · 56. It is evident that the beginning of this section coincides with the beginning of the descent at x = 80 km.
) and the pipeline has invariable diameter (S = const. dx dx d (4. Therefore partial derivatives with respect to time ∂()/∂t in the equations of Section 1. Michael V.1 A System of Basic Equations for Stationary Flow of an Incompressible Fluid in a Pipeline In stationary ﬂow all parameters of the transported ﬂuid at each crosssection of the pipeline remain constant. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co.73 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines In this chapter we consider the calculation of stationary operating regimes of pipelines for transportation of oil.). the velocity of the ﬂuid would be the same at each crosssection of the pipeline v = const. oil products and gas. If the ﬂuid is incompressible ( dρ/ dt = 0) and homogeneous (ρ = const. Consider successively the basic equations describing these ﬂows: 1. 2. The momentum equation (1.6) leads to the equation d (ρvS) = 0 dx ˙ which means that the mass ﬂow rate M of the transported ﬂuid remains constant ˙ M = ρvS = const. 4.8 should be taken equal to zero. Continuity equation (1. The equations obtained in the ﬁrst chapter are used as a basis. KGaA.1) Modeling of Oil Product and Gas Pipeline Transportation. Weinheim ISBN: 9783527408337 .10) gives ρv dv dp 4 =− − τw − ρg · sin α(x). that is independent of time.
dx d If we take in the last equation τw = λ(Re. ρ p d ein + + gz = πd · qn dx ρ If in this equation we take ein = Cv · T + const. ε) · .36) for stationary ﬂow has the form ρvS · or ρv 4 d dein = · qn − ρv · dx d dx p + gz .23). ε) · · . the momentum equation is simpliﬁed and yields 4 dp = − τw − ρg · sin α(x).1)–(4..2) if we replace in it the hydraulic gradient i0 by 4τw /ρgd = λ(Re. dx d d 2 (4. ε) · v2 /2gd in accordance with Eq. to which class belong oil and oil products. The equation of total energy balance (1.2) Note that the Bernoulli equation (1. (4. 4 2 dx the momentum equation is transformed into d dx 1 vv p + z = −λ(Re. (1. ε) and heat transfer factor K serve as the basis for calculation of the stationary operating regimes of pipelines transporting incompressible ﬂuids. ρg d 2g (4. .19) leads also to Eq. 3.74 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines With regard to the condition v = const. ε) · · and qn = −K · (T − Tex ) ρ d 2 the equation of total energy balance takes the form ρvCv 1 ρv3 dT 4K = − (T − Tex ) + λ(Re.3) with the addition of relations for the hydraulic resistance factor λ(Re. ε) ρvv dz · and sin α(x) = .3) The system of equations (4. d dx p 1 vv + gz = −λ(Re.
respectively.2) p +z ρg − x=0 p +z ρg = λ(Re. This information reﬂecting the interaction of the considered pipeline section with the rest of the pipeline is introduced into the mathematical model through boundary conditions. Another boundary condition models the operation of the oil pumping station (OPS) located at the beginning of the pipeline section. identical. Thus we have one algebraic equation relating three parameters of the ﬂow – the pressure p0 at the beginning of the pipeline section. In general such equipment is called a pump.1) and the differential momentum equation (4. additional information on the pressures at both ends of the pipeline is needed. To determine the velocity v or. one can say that these pumps are equipment for forced ﬂuid movement from a crosssection with lesser head (line of suction) to a crosssection with greater head (line of discharge).5) . Modeling of the Operation of Pumps and OilPumping Stations 75 4. by the conditions required for ﬂuid pumped into reservoirs through a system of intrabase pipelines. the ﬂow rate of pumping.2 Boundary Conditions. Modeling of the Operation of Pumps and OilPumping Stations The Bernoulli equation in algebraic form results from the continuity equation (4. 4.4.g. the pressure pL at the end of the pipeline section and the velocity v of the ﬂuid ﬂow. what is the same.2 Boundary Conditions. one of the boundary conditions can be a simple condition p(L) = pL . Hence. that is at x = L could be taken as given. determined e. The simplest mathematical model of a pump can be represented as an algebraic equation of the form H= pex − pin = F(Q) ρg (4. Since the elevations of the pump entrance and exit are. That is why in a pipeline special equipment producing pressure is needed.1 Pumps Pumps represent equipment for compulsory ﬂuid movement from a crosssection with lesser head (line of suction) to a crosssection with greater head (line of discharge). ε) · x=L L v2 · d 2g (4.2. In ﬂuid ﬂow in the pipeline the pressure gradually decreases because mechanical energy is spent overcoming the force of viscous friction between the ﬂuid layers and is then turned into heat. as a rule. that is at x = 0. the initial and terminal crosssections of the pipeline section with length L.4) In this relation x = 0 and x = L denote. In some cases the pressure pL at the end of the pipeline.
In centrifugal pumps. namely the centrifugal pump. the impeller is believed to be immovable whereas the centrifugal force of inertia ρω2 r. where ρ is the ﬂuid density. For every actual pump the differential head H appears to be dependent on the ﬂuid ﬂow rate Q called in this case the feed. The centrifugal force causes the ﬂuid to move along the impeller blade from its center to the periphery. is its feed. where Rim is the radius of the impeller. The greater the head produced by a pump.6) Thus the rotation of the impeller with angular velocity ω can force the ﬂuid to move against the pressure drop p between the periphery and the central . For simplicity let us consider an impeller with radially located blades. equal to the pressure difference between the pumping pressure pex at the impeller periphery and the suction pressure pin at the center of the impeller. gives ρω2 Rim 2 − 2 p = Rim ρ · fτ (Q) or p= ρω2 Rim 2 − Rim ρ · fτ (Q). Integration of the force balance equation over the radius from 0 to Rim . This force is capable of overcoming the pressure drop p = pex − pin . Considering the frame of reference related to the rotating impeller. the ﬂuid moves from the crosssection with lesser pressure to that with greater pressure under the action of the centrifugal force produced by the rotation of an impeller with proﬁle blades.76 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines characterizing the dependence of the differential head H produced by the pump on the ﬂuid ﬂow rate Q. It is selfevident that to produce such forced motion one needs to spend energy for impeller rotation. 2g g (4. ω the angular velocity of the impeller rotation and r the distance of the ﬂuid particle from the rotation axis. that is to force the ﬂuid to move from the region of low pressure to the region of high pressure. Figure 4. as a rule. The dependence H = F(Q) deﬁnes the socalled headdischarge (Q − H) characteristic of the pump. acts on the ﬂuid ﬁlling the pump. 2 Division of both sides of this equation by ρg yields H= ω2 Rim 2 Rim − · fτ (Q). In order to understand the physical basis of this model. The balance equation of forces acting on the ﬂuid moving along the impeller radius from its center to the periphery can be written as: ρω2 r − dp = ρ · fτ (Q) dr where dp/ dr is the radial gradient of pressure opposing the motion and fτ (Q) the friction force depending on the discharge Q and increasing with Q. the less.1 shows a scheme of a pump impeller with proﬁle blades. used for pumping oil and oil products. let us closely consider the operating principle of one of the commonly encountered pumps.
main pump HM 2500230 with impeller diameter Dim = 430 mm rated at nominal feed 2500 m3 h−1 and nominal head 230 m has the (Q − H) characteristic H = 280 − 0. Solution. the smaller the pressure drop which the blower has to overcome.792 · 10−5 · Q 2 ( H in m. The two upper curves of this ﬁgure represent the (Q − H) characteristics of a pump with accessory impellers (385 and 430 mm..1 Operating principle of a centrifugal pump.451 · 10−4 · Q 2 . the greater the pump feed. Then. The pump feed Q decreases with increase in p and. Modeling of the Operation of Pumps and OilPumping Stations Figure 4. in accordance with Eq. is obeyed.252 ∼ ω2 Rb 2 = = 314. 77 part of the impeller. conversely. Figure 4. For example. At Q > 0 Eq. (4.81 The (Q − H) characteristics of centrifugal pumps operating in stationary regimes are often approximated by the twoterm dependence H = a − b · Q2 (4. 2002).4 m.2 shows the (Q − H) characteristic of the centrifugal pump HM 2500230. (4. 3000 rpm corresponds to the angular velocity ω = 2π · 3000/60 = 2π · 50 s−1 . 2g 2 · 9. the main pump HM 1250260 produced in Russia rated at a nominal feed of 1250 m3 h−1 and nominal head 260 m. therefore the dimension of the factor a is (m) and the factor b is (m/(m3 h−1 )2 ). has the (Q − H) characteristic H = 331 − 0. This value of p is achieved at Q = 0 in the absence of a friction force.7) where the differential head H is measured in (m) and the ﬂow rate Q in (m3 h−1 ). Exercise.6) we get ( H)max = 4π2 · 502 · 0.6) determining the (Q − H) characteristic of the pump. It is required to determine the maximal differential head developed by a centrifugal pump with radial located impeller blades having radius 0.25 m and rotating at 3000 rpm. the middle curve shows the dependence of the power consumption N (kW) on the ﬂow rate .2 Boundary Conditions. The maximal value of the pressure drop which the centrifugal force is capable of overcoming is equal to ρω2 Rim 2 /2. where H = p/ρg. Q in m3 h−1 ) and so on (Vasil’ev et al.4.
the characteristic of a system of two pumps connected in series is equal to H = (a1 + a2 ) − (b1 + b2 ) · Q2 . that is the range of ﬂow rates Q of the pump. If H1 = a1 − b1 · Q2 is the characteristic of the ﬁrst pump and H2 = a2 − b2 · Q 2 the characteristic of the second pump. Q and the bottom curve illustrates the dependence of efﬁciency η(%) on the ﬂow rate of the transported ﬂuid.8) .2 OilPumping Station Pumps connected in series or parallel provide the basis of oilpumping stations intended to produce driving pressure. The (Q − H) characteristics of pumps connected in series (Fig. 4. 4.3) are summarized. (4. In this range (1800 < Q < 3000 m3 h−1 ) the efﬁciency η ≈ 85% and the power N ≈ 1600 kW of the pump have maximal values. In the same ﬁgure is also marked the operating range of the pump.78 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines Figure 4.2.2 Characteristics of the centrifugal pump HM 2500230. the ﬂuid ﬂow rates of the pumps are identical Q1 = Q2 = Q and the differential heads are given by H = H1 + H2 .
Modeling of the Operation of Pumps and OilPumping Stations Figure 4.451 · 10−4 · Q 2 . It is required to ﬁnd the characteristic of a system of two pumps connected in parallel? Figure 4.451 · 10−4 · Q 2 . What characteristic has a system of two pumps connected in series? Solution. Another pump of the same type but with impeller diameter 465 mm has the (Q − H) characteristic H = 374 − 0. Q in m3 h−1 ).9) Exercise 1.451 · 10−4 · Q 2 . (4. ( H in m.2 Boundary Conditions. 4.902 · 10−4 · Q 2 .4.451 · 10−4 · Q 2 . If H = a1 − b1 · Q 2 is the characteristic of the ﬁrst centrifugal pump and H = a2 − b2 · Q 2 that of the second one. ( H in m. Another pump of the same type but with impeller diameter 465 mm has the (Q − H) characteristic H = 374 − 0. b2 (4. The (Q − H) characteristic of a centrifugal pump with impeller diameter 440 mm is H = 331 − 0.3 Series connection of pumps. In accordance with Eq. the characteristic of the system of two pumps connected in parallel is (a1 − H) + b1 (a2 − H) = Q.8) we obtain 0. Fluid discharges in pumps are given by Q = Q1 + Q2 but the heads produced by each pump are identical H = H1 = H2 . Q in m3 h−1 ).4 Parallel connection of pumps. H = (331 + 374) − 2 · Exercise 2. .4) their (Q − H) characteristics are different. 79 In parallel connection of pumps (Fig. The (Q − H) characteristic of a centrifugal pump with impeller diameter 440 mm is H = 331 − 0.451 · 10−4 · Q 2 = 705 − 0.
called the head before pumping station. that is the pipeline system of the station.716 · 10−3 · Q.11) .036 · 10−4 · Q 2 ( H. When using the twoterm dependence of the station differential head H on the ﬂow rate Q the boundary condition (4. It is also known that the head losses hc in the station communications. are represented by the dependence hc = 25 − 0. the boundary condition at the initial crosssection x = 0 of the pipeline section can be the following condition given by this characteristic pex − pin = F(Q) ρg or p0 pu ˜ = + ·F(v) ρg ρg (4. (4. It is required to ﬁnd the characteristic of the pumping station? Solution.9) we have (331 − H) + 0. where H < 331 m. The characteristic of the pumping station H = F(Q) is represented by the sum of the characteristics of the pump system minus head losses in the supply communication F1 (Q) + F2 (Q) − hc (Q): H = F(Q) = 680 − 0.10) at the beginning of the pipeline section takes the form p0 pu = + a − b · S2 (3600)2 · v2 ρg ρg where the velocity v is measured in (m s−1 ). hc in m. ˜ F(Q) ≡ F(vS) ≡ F(v). if (Q − H) the characteristic of the pumping station H = F(Q) is known. The (Q − H) characteristic of a pumping station is the total (Q − H) characteristic of all pumps operating in the station (connected in series or parallel) minus the (Q − H) characteristics of the supply communications. Hence. pu = pin the pressure before the oilpumping station. At a pumping station two pumps operate in series with characteristics H = F1 (Q) = 331 − 0.451 · 10−4 or √ 331 − H+ √ (374 − H) =Q 0. a condition similar to condition (4.451 · 10−4 374 − H = 6. In accordance with Eq.5). Q in m3 h−1 ).385 · 10−4 · Q 2 . The latter is taken as an element connected in series with the pumps of the station.10) that is.80 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines Solution.800 · 10−4 · Q 2 .451 · 10−4 · Q 2 and H = F2 (Q) = 374 − 0. Exercise 3. where p0 = pex is the pressure at the initial crosssection of the pipeline section. (4.
in which the pressure p0 = p(0) at the initial crosssection of the pipeline section is excluded with the help of boundary condition (4.4.11) pL L v2 p0 + z0 − + zL = λ(Re. (4.3 MPa. z0 = 50 m. the velocity v is measured in (m s−1 ). L = 120 km. zL = 100 m). Two identical pumps connected in series and having identical (Q − H) characteristics H = 331 − 0. the head before the pumping station hu is 30 m and it is known that sections of gravity ﬂows are absent in the pipeline. = 0.2 mm. .296 · 10 S b · v ) ρg ρg where a and b are the approximation factors of the pumping station (Q − H) characteristic. ε) · · .12) serves to determine the unknown velocity v of the ﬂuid ﬂow in the pipeline. ε) · · . After eliminating p0 from these equations we obtain L v2 pu pL − + (z0 − zL ) + a − 1.451 · 10−4 · Q 2 . ν = 9 cSt) along the pipeline section (D = 530 × 8 mm.4) is used.12) it is convenient to rearrange all the terms containing the unknown velocity v on the righthand side of the equation leaving on the lefthand side only the given quantities pu L 1 pL − + (z0 − zL ) + a = λ · · + 1. ρg ρg d 2g (4.296 · 107 S2 b · v2 = λ(Re. ρg ρg d 2g pu p0 7 2 2 = + (a − 1. Exercise 1. It is required to ﬁnd the ﬂow rate and pressure at the beginning of the section when the pressure pL at the end of the section is 0. To solve Eq.3 Combined Operation of Linear Pipeline Section and Pumping Station 81 4.12) This equation is called the head balance equation. ρg ρg d 2g This equation could be solved by the method of successive approximation (iteration method). 3 −1 Q in m h ) are pumping diesel fuel (ρ = 840 kg m−3 . ( H in m. At given values of the head before the pumping station pu and pressure at the pipeline section end pL Eq.296 · 107 S2 b · v2 . We can demonstrate it with exercises. (4.3 Combined Operation of Linear Pipeline Section and Pumping Station To calculate the combined operation of a linear pipeline section and the pumping station located at the beginning of the pipeline section the Bernoulli equation (4.
451 · 10−4 · v2 Then with formula (1.14 · 0. If as a ﬁrst approximation it is accepted that λ(1) = 0. v ∼ 1.514 ∼ = 81 897.0205.02.434 m s−1 .0205 > λ(1) = 0.02 then this equation gives v = 1. = It is seen that the obtained value of the hydraulic resistance factor should be enhanced.81 + 1.2 68 + 514 82 754 0.0206 ≈ λ(2) = 0.296 · 107 and 605. Hence. Write Eq.11 · 0.14 · 0.514 = 82 754.449 · 0. ρg ρg d 2g Insertion of the given data yields 30 − 120 000 1 0.296 · 107 S2 · 2b)v2 .25 ∼ 0.449 m s−1 .11) of the head balance pL L 1 pu − + (z0 − zL ) + 2a = (λ · · + 1.0205.2974 m3 s−1 or = 4 Q = 0. Then the given equation yields v = 1.2 68 + 514 81 897 0.434 ∼ 0.81 0. 9 · 10−6 3.11 · ∼ 0.31) we get λ = 0.5142 4 2 · 2 · 0. We have Re(2) = 1.3). As a result we have .25 λ = 0.434 · 0.008 2 · 9.2 · λ + 50. = Thus there is good coincidence between the taken and received factor λ. = The pressure p0 at the beginning of the pipeline section is determined with the formula p0 = ρg · [hu + F(Q)].530 − 2 · 0.82 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines Solution.6 = v2 · (11 899.2974 · 3600 ∼ 1071 m3 h−1 .5142 · 1. (9 · 10−6 ) 0. (4. We need to verify whether or not the factor λ is correctly taken. after which it is necessary to verify whether the factor λ is correctly taken. As a second approximation we take λ(2) = 0.434 m s−1 and = Q= 3. To do this let us determine the ﬁrst Reynolds number Re(1) = 1.3 · 106 + 50 − 100 + 2 · 331 = λ · · 840 · 9.
4 41 088 As the second approximation we take λ(2) = 0.0222. It is known that the (Q − H) characteristic of the supply communication of the oilpumping station has the form H = 0.4 Calculations on the Operation of a Pipeline with Intermediate OilPumping Stations 83 p0 = 840 · 9. ν = 25 cSt) is being conducted by two pumps: HM 2500–230 with characteristic H = 251 − 0. hu = 70 m.0222.0222 > λ(1) = 0.02. This equation is solved by the iteration method.8 = 41 088.0224 ≈ λ(2) = 0. zL = 120 m. 0. As a ﬁrst approximation we take λ(1) = 0. hL = 40 m) has an approximately ﬂat character and that sections of gravity ﬂow are absent. (25 · 10−6 ) 0.85 · 106 Pa = 1071 m3 h−1 . L = 150 km. The equation of the head balance is 80 + 70 + 251 − 0.125 · 10−4 · Q 2 connected in series and rated at feed 1800 m3 h−1 .812 · 10−5 · Q 2 + 273 − 0.4.8 = 39 744. Answer.3164 ∼ λ= √ = 0. 4 39744 Hence v = 1. Solution. Now verify whether λ has been chosen correctly. Then the equation gives v = 1.81 · 30 + 2 · 331 − 0.81 0.85 MPa.812 · 10−5 · Q 2 and HM 3600230 with H = 273 − 0. Besides it is known that head losses due to local resistances comprise ≈ 2% of the head losses due to friction. = Answer.284 · 0. 4. Re = 1.451 · 10−4 · 10712 or 4. Exercise 2.6).15 · 10−4 · Q 2 (here and above H is in m and Q in m3 h−1 ). (25 · 10−6 ) 150 000 v2 · . 2246 m3 h−1 .125 · 10−4 · Q 2 −0.02.02 · λ · After simpliﬁcation this equation takes the form 514 = v2 · (9748 · λ + 116. Now verify whether λ has been chosen correctly Re = 1. ∼ 4.242 · 0. It is required to determine the pumping ﬂow rate under the condition that the oilpipeline section (D = 820 × 10 mm.3164 ∼ λ= √ = 0.15 · 10−4 Q 2 − [120 + 40] = 1.85 MPa.242 m s−1 which is equivalent to Q ∼ 2246 m3 h−1 . z0 = 80 m.800 2 · 9.284 m s−1 . . The pumping of crude oil (ρ = 870 kg m−3 .242 m s−1 . Then the equation yields v = 1.
j before the intermediate oilpumping stations... (4..15) ..... H = F2 (Q). . z2 . hu... (4.. where H = F1 (Q). . . Equations (4. . When intermediate ﬂuid dumping and pumping are absent we can write the Bernoulli equation for each section [z1 + hn1 + F1 (Q)] − [z2 + hn2 ] = h1 – 2 (Q).2 .s before the sth intermediate pumping station j=s hu..... . z1 .4 Calculations on the Operation of a Pipeline with Intermediate OilPumping Stations Consider a pipeline consisting of n successive sections separated by oilpumping stations (OPS).14) can be realized in the considered pipeline only when the heads hu.1 ...s = hu.. hu. ... Second consequence (equation for heads before the oilpumping stations): Termbyterm summation of only the ﬁrst s (s < n) equations of the system (4..13) .i = pu. with all unknown heads hu.j of all the intermediate stations are greater than the minimum allowed value assuring pump operation without cavitation and the pressure in all crosssections of the pipeline is less than the permissible value deﬁned by the pipeline strength.1 − hL ) + j=1 Fj (Q) = (zL − z1 ) + j=1 hj−(j−1) (Q) (4.1 + (z1 − zL ) + j=1 Fj (Q) − hj−(j−1) (Q) .n + Fn (Q)] − [zL + hL ] = hn−(n−1) (Q).. It should be taken into account that the ﬂow rate Q found from Eq.. The transportation of ﬂuid is performed in the socalled pumptopump regime...j before the intermediate pumping stations being excluded.84 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines 4. hu. hj−(j−1) (Q) the head losses in the sections between the oilpumping stations dependent on the pumping ﬂow rate Q. H = Fn (Q) are the hydraulic (Q − H) characteristics of oilpumping stations. [z2 + hn2 + F2 (Q)] − [z3 + hn3 ] = h2 – 3 (Q). ... (4..13) yields the equation for the heads hu.14) This equation serves to determine the ﬂow rate Q of the ﬂuid (carrying capacity of the pipeline).13) yields the equation called the balance equation of the heads for the whole pipeline j=n j=n (hu. ..n the heads before the oilpumping stations equal to hu. .. .. [zn + hn... respectively.. zn the elevations of the oilpumping stations..i /(ρg). . zL .13) represent a system of n algebraic equations (according to the number of sections) with n unknown quantities: ﬂow rate Q and (n − 1) heads hu.. hL = pL /(ρg) the elevation and piezometric head at the pipeline end (x = L). First consequence (equation of head balance): Termbyterm summation of the equations of the system (4. .
2 + 2 · (285 − 0. 0. Let us illustrate the aforesaid with an exercise. The head hu. 3. Solution.704 2 · 9.1 before the leading oilpumping station is 50 m and the head hL at the end of the pipeline is 30 m. (4. km 150 180 120 D.81 120 000 v2 · .4 Calculations on the Operation of a Pipeline with Intermediate OilPumping Stations 85 The ﬂow rate Q in this equation is assumed to be given. 1. 1.81 60 + hu.812 · 10−5 Q 2 H = 285 − 0. 3. 2.812 · 10−5 · Q 2 ) − [60 + hu.81 180 000 v2 · . Type of pump II 2500230 II 3600230 II 5000210 (Q − H) characteristic Positive suction head. since it can be obtained from Eq. In determining the head losses in the pipeline sections it is necessary to account for the possibility of existing transfer points and segments of gravity ﬂow in these sections (see Section 3. 2. mm 720 720 720 δ.2 ] =λ· v2 150 000 · . mm 8 8 8 z0 . m 40 40 40 H = 251 − 0. m 50 60 70 zL .4. m 60 70 180 No. At the beginning of each section there is an oilpumping station with two identical pumps connected in series the characteristics of which are given in the following table It is required to determine the carrying capacity of the oilpipeline when pumping oil (ρ = 900 kg m−3 . 0.480 · 10−5 Q 2 .640 · 10−5 Q 2 ) − [70 + hu. Length.7).3 ] =λ· 70 + hu3 + 2 · (236 − 0. The balance equations of the heads for the pipeline sections are 50 + 50 + 2 · (251 − 0. An oilpipeline with length L = 450 km consists of three linear sections the data for which are given in the table below.14).704 2 · 9.640 · 10−5 Q 2 H = 236 − 0.704 2 · 9. ν = 30 cSt) through it and the heads of the intermediate oil pumping stations. Exercise. 0.480 · 10−5 Q 2 ) − [180 + 30] =λ· No.
812 · 10−5 · 18322 )] − [60 + hu. = 0.0234 > 0.3164 ∼ λ= √ = 0.3164 ∼ λ= √ = 0.704 ∼ = 30 694. the rate of pumping and the hydraulic resistance factors are identical when passing from one section to another. = 0.7 + 2 · (285 − 0.3082 · and hu. (30 · 10−6 ) 0.81 The second balance equation gives hu.3 ∼ 48.3082 · and hu.0 m. hu.2 [50 + 50 + 2 · (251 − 0. Then the equation yields v = 1. Then from the above equation we obtain v(1) = 1. We take ﬁrst λ(1) = 0.7 m. (30 · 10−6 ) 0.308 m s−1 or Q ∼ 1832 m3 h−1 . therefore. Next we verify the correctness of the factor λ: Re = 1.704 2 · 9.8).704 ∼ = 32947.02. the obtained pumping regime is realizable.2 ] = 0. hu.704 2 · 9.86 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines Here we assume that.0239 ≈ 0. Termbyterm summation of the above cited equations gives 1434 − 3. .404 m s−1 .308 m s−1 .864 · 10−5 · Q 2 ∼ 32 579 · λv2 or = 1434 = v2 · (32 579 · λ + 75.3 ] = 0.640 · 10−5 · 18322 )] − [70 + hu.2 .404 · 0. = From the ﬁrst balance equation of heads we determine hu. We verify again whether the factor λ is correctly chosen: Re = 1. due to the invariability of the pipeline diameter.0234 · 150 000 1. 4 32947 As the second approximation we take λ(2) = 0.81 Both the values for the heads of the intermediate oilpumping stations comply with the requirement of positive suction head and. 4 30694 Thus v = 1.3 [60 + 52.02.0234.0234 · 180000 1.2 ∼ 52.3 are the unknown heads of the intermediate stations that are to be determined.308 · 0.0234. This equation (balance of heads for the whole pipeline) is solved by the iteration method.
the initial conditions at x = 0 should be taken as p0 pu = + a − b · S2 (3600)2 · v0 2 . The latter condition v ≈ v0 = const. we can take p(0) = p0 .16) should be integrated numerically or solved using one or other simpliﬁcation suggestions. that is we also ignore heat expansion of the ﬂuid.. pu the head before the station. ε) · · (4.18) signifying that the pressure and temperature at the beginning of the pipeline are known. In the general case the system of equations (4. One consists in ignoring the heat released in the ﬂuid due to the work of internal .. (4. allows us to rewrite the system of differential equations (4.4.11) a and b are approximated factors the of 2 (Q − H) characteristics of OPS.19) Here.5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating To calculate the stationary operating regime of pipelines performing ﬂuid pumping with heating (hightemperature pumping) Eqs. ε) · 1 · v v H dx d d 2g The ﬁrst equation gives ρvS = const.3) are used. follows. meaning that the mass ﬂow rate is constant in the stationary operating regime. that is we neglect heat expansion of the pipe.). in accordance with Eq.16) as p 1 v0 v0  d dx ρg + z = −λ(Re. d (ρvS) = 0 dx d p 1 vv + z = −λ(Re.5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 87 4. T(0) = T0 (4. (4. (4. S0 = πd0 /4. that is conditions at the initial crosssection x = 0 of the pipeline. from the constancy of mass ﬂow rate the condition of pumping rate constancy v ≈ v0 = const. ε) · d · 2g 0 3 ρv C dT = − 4K (T − T ) + λ(Re. ε) · 1 · v0  0 v H dx d0 d0 2g (4. If we take the pipeline diameter to be invariable d = d0 = const. If it is needed to model the oilpumping station with given (Q − H) characteristic H = a − b · Q 2 located at the beginning of the pipeline. and the ﬂuid density to be insigniﬁcantly variable (ρ ≈ const.1)–(4.16) dx ρg d 2g 3 ρvC dT = − 4K (T − T ) + λ(Re. 0 ρg ρg T(0) = T0 .17) As initial conditions.
21) provides a basis to determine the unknown velocity v0 and consequently the ﬂow rate of ﬂuid pumping with heating.16) can be easily integrated and the solution of this equation yields (see Eq. Tex the temperature of the external medium.. therefore the Reynolds number Re = v0 d0 /ν(T) = Re(x) and the factor of hydraulic resistance λ are functions of x. the kinematic viscosity of the ﬂuid ν is not constant. The kinematic viscosity as a function of temperature ν(T) is assumed to be given.20) ˙ where M0 = ρv0 S0 is the mass ﬂow rate. With regard to Eq. ε) dx · 1 v0 2 · d0 2g The system of equations p u − pL + (z0 − zL ) + a − 1. Really.19) we have pu pL − + (z0 − zL ) + a − 1. This circumstance is taken into account in the Bernoulli equation obtained by integration of λ over the pipeline section length.88 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines friction forces as compared to external heat exchange λ(Re.296 · 107 S0 2 b · v0 2 ρg ρg = 0 L λ(Re. it depends on temperature. then T = const. d0 2g d0 Then the second equation of system (4. ε) 4K 1 v0 3 ≤ · · T − Tex .44)) T(x) = Tex + (T0 − Tex ) · exp − πd0 · K x ˙ Cv M 0 (4. taken as constant. . if T = T(x).21) has an analytical solution when ν(T) = νex · e−κ(T−Tex ) . The ﬁrst equation of the system (4.296 · 107 S0 2 b · v0 2 ρg ρg L 1 v0 2 = λ(Re. see for example Eq. (1. d0 2g 0 T(x) = Tex + (T0 − Tex ) · exp − πd0 · K x ˙ Cv M 0 (4. It should be noted that the ﬂuid velocity v0 is constant but unknown. (4. ε) dx · 1 v0 2 · d0 2g This equation differs from the standard form of the Bernoulli equation in that it takes into account variability of the factor λ with pipeline length.16) can be represented as p0 + z0 − ρg pL + zL = ρg L 0 λ(Re. ε) dx · · . Analytical Solution The system of equations (4. (2.3).
5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 89 where νex is the kinematic viscosity of the ﬂuid at the temperature Tex of the external media.3164 λ= √ 4 Re that is the ﬂow regime of the ﬂuid corresponds to the ﬂow region of the socalled hydraulic smooth pipes (Blasius zone). T − Tex With regard to Eq.44) gives ρv0 Cv Thus L 0 dξ dT 4KL ρv0 Cv d0 1 · (T − Tex ) ⇒ . η Now gathering together all the results we obtain the following expression for head losses in a nonisothermal ﬂuid ﬂow L 0 λ dx · 1 v0 2 L v0 2 · = λeff · · d 2g d 2g (4. (2. Then L 0 λ dx = L · 0 1 λ dξ = L · TL T0 λ dξ dT.4.22) If we convert to the dimensionless coordinate ξ = x/L. that is we take T⊗ = 0. In this case the integral on the righthand side of the ﬁrst equation of the system could be calculated in quadratures L 0 0.44) we neglect the heat release.3164 λ dx = √ · 4 v0 d 0 = λex · 0 L L 0 0.23) . dT If now in Eq.3164 ν1/4 dx = √ · 4 v0 d 0 dx. (1. κ the dependence factor Eq.22) we have L 0 λ dx = − ρCv v0 d0 · L · λex 4KL TL T0 e− 4 (T−Tex ) dT. where L 0 1/4 νex e− 4 (T−Tex ) dx κ κ e− 4 (T−Tex ) λex = 0.3164 4 v0 d0 /νex (4. then Eq. (4. =− =− · dξ d0 dT 4KL T − Tex λ dx = L · TL T0 λ dξ ρv0 Cv d0 dT = − ·L· dT 4KL TL T0 λ dT.3). 0. (1. T − Tex κ The integral on the righthand side can be transformed to TL T0 e− 4 (T−Tex ) dT = T − Tex κ κ − (TL −Tex ) 4 κ − (T0 −Tex ) 4 eη dη.
22 −0. ν1 = 5 cSt at T1 = 50 ◦ C and ν2 = 40 cSt at T2 = 20 ◦ C) with heating. We have ν(T) = 5 · e−κ·(T−50) .30) the dependence of the oil viscosity on temperature ν(T).8 −0.82 −0.0693 K−1 .22 −0.70 −0.24) takes the form pL L v0 2 pu − + (z0 − zL ) + a − 1. = Table 4. Some values of this function are listed in Table 4. Q in m3 h−1 . m λex = 0.21) to determine the velocity v0 of nonisothermal ﬂuid ﬂow with regard to Eqs. The initial temperature T0 of the oil is 50 ◦ C.05 −2. Solution.31 −0.1.22)–(4.24) κ 4KL πKd0 · L · (T0 − Tex ).3164 4 v0 d0 /νex .296 · 107 S0 2 b · v0 2 = λeff · (4. Along a practically horizontal oilpipeline section (D = 720 × 10 mm.0 −0.45 −0. The characteristic of each pump is H = 273 − 0. The second condition ν(20) = 40 cSt yields the equation for the factor κ 40 = 5 · e−κ·(20 – 50) from which κ ∼ 0.6 −0. First determine with Eq.44) and as Ei(z) it is denoted the Euler function Ei(z) = z −∞ eη dη. (2. It is required to ﬁnd the ﬂow rate of pumping and the temperature of the oil at the section end.1 −1. L = 120 km) is pumped oil (ρ = 870 kg m−3 . η (4. The basic equation (4. The pumping is carried out with two pumps connected in series. (4. The heattransfer factor K averaged over the pipeline section is 3.5 W m−2 K−1 ). m = = .4 −0.24) · ρg ρg d0 2g Exercise. the temperature Tex of the environment is 10 ◦ C. Here we take ν(50) = 5 cSt. Cv = 2000 J kg−1 K−1 ). (4.1 z Ei(z) −1.90 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines where λeff is the effective factor of hydraulic resistance determined by λeff = λex · k= 1 · [Ei(−k) − Ei(−ke−m )]. ˙ 4 ρCv v0 d0 Cv M 0 Here we take into account the equality TL − Tex = (T0 − Tex ) · exp(−m) following from the basic formula (1.25) which is widely encountered in technical applications and for which there are special tables.47 .125 · 10−4 Q 2 ( H in m. hL = hu ).2 −1.
.02.029 · 1 · [Ei(−0.72 · v0 · 3600 4 and taking into account other conditions we get 546 = v0 2 · (8737. 0.95 cSt.881.23) is λeff = where k= πKd0 · L 4K · L κ = .24) for the balance of heads 2 · [273 − 0.939)] ∼ 0.029 · · [−0.7/(79. Let us take λeff (1) = 0.3164 4 1. 4 · 3.0693 · (50 − 10) = 0.693 · exp(−0.287)] 0.566 m s−1 and verify the correctness of the obtained value. (4.7 2 · 9. (4. We have νex = 5 · exp[−0.94) (4.4 · λeff + 47.379 − (−0. = k = 1/4 · 0.566 · 0.566 · 0.287.4.26) we ﬁnd the velocity of ﬂuid ﬂow v0 (1) = 1. λex = 0.0693 · (10 − 50)] = 79.95 · 10−6 ) ∼ 0.881) ∼ 0. = 1. First approximation.81 The effective factor λeff of the hydraulic resistance taking into account its variability with pipeline section length in accordance with Eq.3164 4 v0 · d0 /νex · 1 · Ei(−k) − Ei(−ke−m ) m Substituting in the balance equation of heads the expression for the ﬂow rate Q using the velocity v0 of the ﬂuid ﬂow Q= 3. Since there is a difference between the taken and calculated values of λeff we make a second approximation.02. m = ˙0 4 ρCv v0 d0 Cv M 0.26) which can be solved by the iteration method.881 = 0.881 1 = 0.029.693) − Ei(−0.693. (4.5 Calculations on Pipeline Stationary Operating Regimes in Fluid Pumping with Heating 91 Writing Eq. · (T0 − Tex ).7 · 870 · 2000 k · exp(−m) = 0.0186 < λeff = 0.14 · 0.5 · 120 000 ∼ 0. Then from Eq. = m= λeff = 0.125 · 10−4 Q 2 ] = λef f · v0 2 120 000 · .
from Eq.27) Since the gas density ρ decreases with pressure drop. = Answer. We have κ = 0. (4.95 · 10−6 ) k = 1/4 · 0. λex = 4 = 1.7/(79. dx dx d If the gas velocity v increases.95 cSt. Let us take λeff (2) = 0.0693 K−1 .5 · 120 000 ∼ 0. νex = 79.6): d ˙ (ρvS) = 0 ⇒ M = ρvS = const.21) is TL = 10 + (50 − 10) · exp(−0.611 · 0.611 m s−1 and verify its correctness.029.611 m s−1 and = Q = 2231 m3 h−1 . = 1.693 · exp(−0.029 · · [−0. • momentum equation (1. 27 ◦ C. 4.0184 ≈ 0.856. 2231 m3 h−1 . It is evident that the estimation .921)] ∼ 0.0186.611 · 0.27) it follows that the gas velocity v increases from the beginning of the pipeline section to its end. the acceleration v · dv/ dx of the gas is distinct from zero. = m= λeff = 0. the temperature of the oil at the pipeline section end in accordance with Eq.856) ∼ 0.92 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines Second approximation.379 − (−0.029 · 1 · [Ei(−0.0186 = 0.7 · 870 · 2000 k · exp(−m) = 0.294.693) − Ei(−0. 0. From Eq.856 = λeff (2) . dx (4. 4 · 3. v0 ∼ 1.856 1 = 0.856) ∼ 27 ◦ C. (4.693. (4.3164 ∼ 0. Since the taken and calculated values of the factor λeff show good coincidence the process of successive approximations ends. Hence.6 Modeling of Stationary Operating Regimes of GasPipeline Sections For modeling the stationary ﬂow of a compressible gas in a gaspipeline the following equations are used: • continuity equation (1.0693 · (50 − 10) = 0. Hence.10): ρv dv dp 4 =− − τw − ρg sin α(x).25) we get the new velocity of ﬂuid ﬂow v0 (2) = 1.294)] 0.
Then the momentum equation of the gas expresses in essence the equality of the driving forces: pressure and friction dp 4 ρv2 4 1 ρv2 = − τ w = − · Cf = −λ · . for example. that is about 45 times greater.29) Using the expression for gas enthalpy J involving internal energy and other parameters of state J = ein + p = Cv T + Z(p. c = γRT ≈ 1.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 93 ρv d dv = (ρv2 ) dx dx d (p) dx is valid when the gas velocity v is small compared to the velocity of sound √ in a gas c. whereas the pressure variation p calculated with the Joukowski formula p = ρc v is 50 · 440 · v = 22 000 · v. where D∗ is the JouleThompson factor.31 · 500 · 300 ∼ 440 m s−1 . that is we take J = J(p. when the velocity of compressed gas with density 50 kg m−3 varies over v the quantity ρv · v is 50 · 10 · v = 500 · v at gas velocity 10 m s−1 . T). d Denoting (∂J/∂T)p = Cp as the speciﬁc heat of a gas at constant pressure and (∂J/∂p)T = −D∗ Cp .30) .28) balance equation of total energy ρvS · dJ = πd · qn dx This equation ignores the work due to the force of gravity. as a rule. and the external heat exchange in the form of Newton’s law (1. the acceleration of gas and often the gravity component ρg sin α in the momentum equation may. p ∂J ∂p =− T RT · Cp ∂Z ∂p (4. be neglected.42). the last equation takes the form ρvCp dT dp 4K − ρvCp D∗ = − (T − Tex ). So.4. If in this equation we use the dependence of enthalpy J on pressure p and temperature T. ρ The factors Cp and D∗ can be expressed through the factors Cv and Z(p. dx dx d (4. T) as follows Cp = D∗ = − ∂J ∂T 1 Cp = Cv + R · Z + RT p ∂Z ∂T . dx d d 2 d 2 • (4. T . T)R] · T. T) · RT = [Cv + Z(p. Thus. then ρv ∂J ∂T · p dT + dx ∂J ∂p T dp dx =− 4K (T − Tex ).31). where γ = Cp /Cv is = the adiabatic index (for methane γ = 1.
1 Distribution of Pressure in Stationary Gas Flow in a GasPipeline πd · K · L ˙ Cp M This distribution is obtained from Eqs.94 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines From this it is seen that the JouleThompson effect (cooling of heatinsulated gas in a gaspipeline through a pressure drop.27) and (4. If we neglect the JouleThompson effect. 4. (4.6. 1 ρv2 dp = −λ · dx d 2 (4. 2 dx d 2 2 d · ρS 2 (p/ZRT) · (π2 d5 /16) . 2. see Fig.33) where the temperature TL at the end of pipeline section is represented by the expression TL − Tex = (T0 − Tex ) · exp − following from Eq.28) ˙ ρvS = M = const. (4.31) and its solution takes the form of Eq.32).32). (1.34) and the equation of the gas state p = ZρRT. Transformation of the second equation (4. (∂Z/∂p)T < 0.32) yields ˙ 1 dp 1 ρv2 1 ρ2 v2 S2 M2 =− ·λ· = −λ · =− ·λ· . ˙ T0 − Tex Cp M (4.32) If we deﬁne the average temperature Tav of the gas over pipeline section length by the expression Tav = 1 · L L T(x) dx 0 and substitute into it the distribution T(x) from Eq. (4.. the energy balance equation is simpliﬁed to ρvCp dT 4K = − (T − Tex ) dx d (4.7) manifests itself only for a real gas when Z = 1 (as a rule the factor D∗ ≈ 3–5 K MPa−1 ).46): πd · K T(x) − Tex = exp − x . we get Tav = Tex + T0 − TL T0 − Tex ln TL − Tex (4.
π2 d 5 (4. T) ≈ Z av = const. respectively π ˙ M= 4 p0 2 − pL 2 · d5 . (4. • Z(p..37) • From Eq. Equation (4.38) or pL = • p0 2 − 16λ · Zav RTav · L ˙ 2 ·M . (4.g. see e.36) one gets in particular pressure pL (in Pa) at the end of the pipeline section with length L for given ˙ mass ﬂow rate M pL 2 = p0 2 − 16λ · Zav RTav · L ˙ 2 ·M π2 d 5 (4. λ · Zav RTav · L (4.34). we get pav = 2 pL 2 p0 + 3 p0 + pL . the resulting differential equation gives after integration the distribution of p(x) along the pipeline section length ˙ 16 λ · Zcp RTcp · M2 p dp · =− 2 · 5 2 dx π d or p2 (x) = p0 2 − ˙ 16λ · Zav RTav M2 x. • T ≈ T av = const.40) . (4.33) means that p2 (x) decreases linearly with the pipeline section length p2 (x) = p0 2 − (p0 2 − pL 2 ) · x L (4. If we deﬁne the average pressure pav over the pipeline section length as pav = 1 · L L p(x) dx 0 and we insert it into the distribution of p(x) from Eq.4.39) ˙ mass ﬂow rate M (in kg s−1 ) for given pressures p0 and pL at the beginning and end of the pipeline section.36). Eq.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 95 If we take • λ = const.. (1.36) where pL is the pressure at the section end.35) Here we used the initial condition p = p0 at x = 0. π2 d 5 (4.
2 in accordance with Eq. From Eq. ˙ To do this it is sufﬁcient to divide M by the density ρst of gas under standard conditions Qk = ˙ M ρ ρ = · vS = · Qv ρst ρst ρst (4. 4.44) + p0 2 × exp 2g (z0 − z(x)) . In fact the commercial ﬂow rate of gas is the mass ﬂow rate expressed in volume units under standard conditions. at pst = 101 325 Pa. respectively. (4. α = dx d 2 dx containing the pipeline proﬁle z(x). Zav RTav .34). (4. The quantity Qk (m3 s−1 ) is the socalled commercial ﬂow rate of gas.15 K. in other words the mass ﬂow rate in volume calculus.6 ˙ M =A· p0 2 − pL 2 2. If we take T ≈ Tav = const.2 Pressure Distribution in a GasPipeline with Great Difference in Elevations In this case one has to use the following equations instead of Eq. The solution of this equation with initial condition p(0) = p0 is p(x)2 = − ˙ λ · Zav RTav · M2 · 2 /4)2 d (πd x (4.42) 1 ρv2 dz dp = −λ · − ρg · sin α. the commercial ﬂow rate remains invariable with the length of the gaspipeline in stationary ﬂows. The key advantage is that in contrast to the volume ﬂow rate that varies from one crosssection to another. i.43) exp 0 2g (z(ς) − z(x)) dς Zav RTav (4.34) ˙ ρvS = M = const. (1...6 ·d L where A is the proportionality factor. the mass ﬂow rate is found to be proportional to d2.e.96 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines If we take λ = 0. and Z ≈ Zav = const. Tst = 293. ˙ The mass ﬂow rate M of gas (kg s−1 ) can also be measured by the volume ﬂow rate taken under standard conditions.6. (4.067(2 /d)0.41) Here ρ and Qv = vS are the density and volume ﬂow rate of gas at the pipeline crosssection. we get ˙ λ · Zav RTav · M2 dp(x)2 dz 2g =− · p(x)2 − 2 /4)2 d dx Zav RTav dx (πd to ﬁnd the function p(x).41) in particular follows that the commercial ﬂow rate Qk of gas is ρ/ρst times greater than the volume ﬂow rate Qv .
even in the case when elevations of the beginning and the end of the pipeline section coincide. (4. it should be enhanced (gaspipelines running through a mountain pass). (4.47) L = L · 1 + 2g (z − z ) ∗ av L Zav RTav Here zav is the average elevation of the gaspipeline section zav = 1 · L L z(ς) dς. 4. (4. dx d 2 dx ρvCp dT − ρvCp D∗ dp = − 4K (T − Tex ) dx dx d . that is at z0 = zL . (v > 0).46) L 2g L = L · 1 · ∗ exp (z (ς) − zL ) dς L 0 Zav RTav It should be noted that the ratio 2g(z − zL )/(Zav RTav ) is usually small when the difference in pipeline elevations is 100–1000 m (Zav RTav ≈ 150 000 m2 s−1 ).45) in which the initial pressure and the length of the pipeline section are changed to 2g (z0 − zL ) . Accurate to the second term we obtain g p = p · 1 + 0∗ (z0 − zL ) . At zav > zL .3 Calculation of Stationary Operating Regimes of a GasPipeline (General Case) To calculate the stationary operating regimes of a gaspipeline the following system of ordinary differential equations is used 1 ρv2 dz dp = −λ · − ρg · sin α(x). the formula for the mass ﬂow rate of gas takes a form similar to Eq.6 Modeling of Stationary Operating Regimes of GasPipeline Sections 97 Assuming p(L) = pL in Eq.6. 0 Equations (4.46) could be simpliﬁed by expansion of the exponential function in the Taylor series. the length of the section has to be changed.47) shows that. Therefore Eq.40) π ˙ M= 4 p0∗ 2 − pL 2 · d5 λ · Zav RTav · L∗ (4. whereas at zav < zL it should be diminished (gaspipelines running along the sea bottom).4.44). p0∗ = p0 · exp Zav RTav (4. 0 Zav RTav (4. sin α = .
(ZRT) 2 d 0.6. dx d 2 2 ρvCp dT = − 4K (T − Tex ) − ρvCp D∗ · λ 1 · ρv dx d d 2 This system should be supplemented with the following equations ˙ ρvS = M.067 · Cp as well as with initial or initial and boundary conditions.2 (4. T) + 3 · · dp Cp (p.98 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines When taking into account that ρg dz dp we have a system 2 dp = −λ 1 · ρv . It is required to ﬁnd the pressure pL and temperature TL at the end of the gaspipeline section. ρ = D∗ = p . Z = Z(p.49) . The following problems can be solved: • the pressure p and temperature T at the initial crosssection of the 0 0 ˙ gaspipeline section as well as the mass ﬂow rate M of gas are known. O). T) qM λ (4.48) by the ﬁrst equation of the same system. • the pressure p and temperature T at the initial crosssection of the 0 0 gaspipeline section as well as the pressure pL at the end of the gaspipeline ˙ section are known. λ = 0. • at the initial crosssection of the gaspipeline section there is a compressor station. It is required to choose the number and type of gaspumping aggregates. compression ratio ε = p0 /pen and the number of revolutions n of the centrifugal blower shafts. Dividing the second equation of the system (4. T).48) −RT(∂Z/∂p)T . • the pressure p and temperature T at the initial crosssection of the 0 0 gaspipeline section and the pressure pL at the end of the gaspipeline ˙ section as well as mass ﬂow rate M of gas are known. 4.4 Investigation of Thermal Regimes of a GasPipeline Section To consider the thermal regimes of gas transportation it is convenient to pass into the phase plane of variables (p. ﬂow rate Qen and temperature Ten of gas at the entrance to the compressor station and the pressure pL at the end of the gaspipeline section are known. It is required to ﬁnd the mass ﬂow rate M and the temperature TL at the end of the gaspipeline section. It is required to ﬁnd the diameter d of the gaspipeline providing this ﬂow rate. The pressure pen . we get the differential equation 8 K ρ(T − Tex ) dT = D∗ (p.
49): 1 – monotonic solutions.49). the straight line T = Tex parallel to the abscissa axis is called the separatrix of Eq. If we neglect the Joule–Thomson effect (D∗ ≈ 0). then dT/ dp > 0. This equation should be solved at the segment pL ≤ p ≤ p0 under the condition T(p0 ) = T0 .6 Modeling of Stationary Operating Regimes of GasPipeline Sections 99 containing only one unknown function T(p).4. 2 – solutions with a maximum point. When the initial temperature of the gas exceeds the temperature of the environment. (4.e. (4. 3 – curve dT/ dp = 0 of the temperature maxima. of the gas at the beginning of the pipeline section.49) would be positive at T > Tex and negative at T < Tex . The inclusion of the Joule–Thomson effect changes the pattern of solution of Eq. respectively. Mathematically speaking. this separatrix is no longer a straight line. (4. . (4. qM = QM S is the speciﬁc mass ﬂow rate. If the temperature of the environment is greater than Tex . it may become less than the Figure 4.49). (4. whereas in the opposite case it heats. pL the pressure at the section end being unknown and to be determined. Hence.5 Solutions of differential equation (4. However. curve 4). it appears to be in the temperature region below the temperature of the surrounding medium (see Fig. the righthand side of Eq. On the plane (p. It always lies below the straight line T = Tex . where p0 and T0 are the pressure and temperature. 4 – separatrix of solutions of Eq. 4. i. the gas cools when moving from higher to lower pressure. T) exists as before a separatrix separating the different types of solutions of Eq. This means that when the pressure falls from p0 at the beginning of the pipeline section to pL at the end of the section.49) because it separates solutions of different types.5. in the ﬁrst case the temperature of gas monotone increases with pressure whereas in the second case it monotone decreases.49).
gasmotor and so on) producing rotation of an impeller shaft in centrifugal blowers (CFP) or the reciprocating motion of a piston in piston engines (PP). namely D∗ (p. T) + 8 K ρ(T − Tex ) = 0. curves 2). therefore it is necessary to spend energy for forced ﬂow in this direction. T0 ) is located below the separatrix 4 of Eq.49). at the point of intersection with curve 3 it reaches a maximum (Tmax < Tex ) and then begins to diminish monotonically. T) of this curve dT/ dp = 0. T0 ) is located below the separatrix 4 of the solutions of Eq.49). (4.7 Modeling of Blower Operation The motion of gas in a gaspipeline is determined by compressor stations (GCS) located at the beginning of each pipeline section. If we provide an initial temperature of the gas T0 such that the point (p0 . the temperature of the gas will remain below the temperature of the permafrost earth all the way along the pipeline section. The main purpose of the blowers and of GCS as a whole is to force gas to move from a region of lower pressure (suction region at the GCS entrance) into a region of greater pressure (region of pumping at the GCS exit). in other words gas compression.100 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines temperature of the environment. 4. the derivative dT/ dp could change sign (see Fig. If the initial parameters of the gas are such that the point (p0 . remaining as before lower than the temperature of the environment. The gas by itself cannot ﬂow against pressure. T) q3 λ M At points (p. electric. The presence of maximum points on the curve T(p) giving temperature distribution is related to processes having diverse actions: heating of the gas due to its mass exchange with the surrounding medium and cooling owing to the Joule–Thomson effect. · · Cp (p. or more precisely by blowers accomplishing gas compression. 4. since if T = Tex the derivative dT/ dp > 0 is independent of the value of the heattransfer factor (see Fig. The gas temperature ﬁrst increases with decreasing pressure. Such forced ﬂow of gas against a pressure force is performed by gaspumping aggregates (GPA) consisting of a drive (gasturbine. It is quite clear that gas in the gaspipeline moves from a crosssection with . This excludes the possibility of warming of the surrounding ground and imparts greater stability to the pipeline in the earth. therefore the temperature of the gas at these points reaches an extremum. (4.5. is accomplished in the blower. Displacement of gas from a region of lower pressure into a region of higher pressure. (4.49). The latter happens at points on line 3 determined by the condition of vanishing of the righthand side of Eq. The phenomena under consideration may be used in exploitation of a gas pipeline section located in permafrost earth. namely a maximum. 4. and by the blower itself.5 curve 1).
..). pB N = N(ρB . pB .. .. the nominal number of revolutions of the blower shaft. As a rule the compression station consists of separate plants equipped with several GPA with blowers connected in parallel in the case of singlestep compression or in series in the case of multistep compression. n.. . A mathematical model of the centrifugal blower operating in the stationary regime involves algebraic dependences of the gas compression ratio ε and the developed speciﬁc power N/ρe on the gas parameters at the blower suction line indicated by the subscript e and the number of impeller revolutions n ε= p0 = ε(ρB .. .. n. see Fig. .. Eq. The gas compression ratio depends on the type of blower.7 Modeling of Blower Operation 101 greater pressure to a crosssection with lower pressure overcoming friction forces. .. QB .51) N n 3 pB /ρB QB /σ = ·φ .. . . . respectively. 4.. . The number of blower . pB n2 n or N n 3 pB /ρB QB . TB .) (4. the rotational speed of its impeller (modern GPA have commonly controlled rotational speed). QB . .. On main gaspipelines centrifugal blowers are predominantly used following the same pattern as centrifugal blowers for ﬂuid.50) takes the following form pB /ρB QB p0 =f . n0 are the diameter of the impeller.. = ·φ . the area of the suction branch pipe crosssection. .50) where the dots denote parameters related to structural features of the blower.1.4. ε= pB n2 D2 im nDim (4. TB . Gas is sucked into the center of the impeller and thrown by centrifugal rotational force to the periphery of the impeller in the discharge line. pressure and temperature at the discharge line and above all on gas ﬂow rate. The rotational rate of the impellers of a large CFP is 4000–15 000 rpm. . ε= .. pB . n2 n 3 ZB RTB QB ·φ . ρB n0 n2 D2 im nDim where Dim . σ... are used. From dimensional theory (see Chapter 6) it follows that these dependences may be represented as pB /ρB QB /σ p0 =f . If the structural parameters of the blower are taken as invariable dependences. n2 n To determine the universal characteristic of the centrifugal blower the socalled reduced conditions. (4. ... ρB n0 n2 n ε= p0 =f pB N n = ρB n0 ZB RTB QB . denoted by subscripts r.
.52) r It is evident that the operating characteristics of a blower observed under arbitrary conditions but not under reduced conditions could be found from the universal characteristics written as follows Zr Rr Tr n0 2 ZB RB TB QB · n0 /n ZB RB TB QB =f . This conclusion has a simple geometrical interpretation. called reduced. respectively.. n2 n n0 2 n Zr Rr Tr n0 N n 3 N · ρ = n ρB r 0 B n 3 Zr Rr Tr n0 2 ZB RB TB QB · n0 /n = ·φ · 2 . n (4. . At this the graphic of the second dependence is also stretched by the factor (n/n0 )3 in the direction of the function axis..102 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines shaft revolutions in such cases is taken equal to a nominal value n0 ... . n0 respectively. . the properties of the gas and the conditions at the blower entrance are ﬁxed: R = Rr . n0 2 n0 Zr Rr Tr QB =φ . n0 n0 2 n Zr Rr Tr n0 ε=f It follows that the operation characteristics of each blower under arbitrary conditions are obtained from some universal characteristics of the same blower. If we introduce n n0 = r n · n0 Zr Rr Tr ZB RTB and (QB )r = QB · n0 . If in the space to build the graphics of the dependences ε = f˜ [(n/n0 )r .53) . through division of the arguments of these characteristic dependences governing the parameters by n n0 2 · Zr Rr Tr ZB RTB and n . the characteristics of a centrifugal blower operating under arbitrary entrance conditions and at a number of revolutions n distinct from the nominal value n0 are determined from these graphics by lengthening of the argument axes by factors n/n0 · Zr Rr Tr /ZB RTB and n/n0 . T = Tr . · 2 .. (QB )r ] and ˜ (N/ρB )r = φ[(n/n0 )r . (QB )r ]. Tests of blowers conducted under these conditions permit the determination of the functions N ρ ε=f Zr Rr Tr QB . . Z = Zr . n0 2 n0 (4.
7 Modeling of Blower Operation 103 The characteristics of the centrifugal blower take the universal form n ε = f˜ . (Qe )r . The parameters of this blower are: n0 = 4800 rpm. n0 r N n ˜ =φ . Rr = 490 J kg−1 K−1 .9. 4. Rr = 490 J kg−1 K−1 .54) · N ρe . (Qe )r ρB r n0 r where N = ρB n n0 3 (4. Tr = 288 K. .4. Zr = 0. r In Fig. Zr = 0.6 are depicted reduced (universal) characteristics of one of the centrifugal blowers 370181 produced in Russia.6 Reduced characteristics of the blower 370181: Tr = 288 K.9. Figure 4.
Let (n/n0 )r = 1. 4. . = (QB )r = 367. respectively. Tcr = 194 K) with commercial ﬂow rate 22 million m3 day−1 and compression ratio ε = 1.982 · n0 = 0. = = consequently the solution can be taken as correct. 0.2 < 1.90 · 490 · 288 ∼ = 1.7 MPa. 24 · 60 31. 25. 2.1 · 288 22 · 106 0.6/0.6). Solution.916.919 · 463.8 · 106 ∼ 31.6/(n/n0 ) and ε = 1.668 ∼ 0. It is required to determine the rotational speed (number of revolutions per minute. Hence n = 0. 1.018 · .916 ∼ 401 m3 min−1 ⇒ ε ∼ 1.0 ⇒ n/n0 = 1.746 kg m RTst 463.002 QB = Qk · ρst /ρB = Then determine the reduced parameters of the operating regime of the centrifugal blower n n0 = r n n0 n Zr Rr Tr = ZB RTB n0 n 0. rpm) of the centrifugal blower shaft 370181 needed to provide transportation of natural gas (µ = 17.982 ∼ 374 m3 min−1 ⇒ ε ∼ 1. to select a value of n/n0 such that the point with coordinates QB = 367.746 ∼ · = 367.982.018 ∼ 0.95 with Eq.4273 · ρst = ρB = 3. using the reduced characteristics of the blower 370181 (see Fig. Let (n/n0 )r = 0.8 · 4.7 288 194 −3. 17.002 kg m−3 = = ZB RTB 0. rpm.6). It is known that the pressure and temperature of the gas at the suction line of the blower are 3. pcr = 4.104 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines Exercise. 4.25 would lie on the characteristic (n/n0 )r = 1. = (QB )r = 367. 4714.25. = Answer. n n Since the compression ratio ε is known and is equal to 1.1 · 288 n0 (QB )r = QB n0 n0 = 367.6 m3 min−1 .982 · 4800 ∼ 4714 rpm. = = consequently (n/n0 )r should be increased.6/0.919 · 463.1 · 293 pB 3.25 (see Fig. The solution is sought by the iteration method.8 MPa and +15 ◦ C. it is necessary. (2.6 · m3 min−1 .1 J kg K−1 ). 4.25 (see Fig. Calculate ﬁrst the parameters of the transported gas R= 8314 ∼ −1 = 463.6).0/1.9 ⇒ n/n0 ∼ 0.14) we obtain: ZB = 1 − 0.919 = pst 101 300 ∼ −3 = = 0.95 kg kmol−1 .018 · n/n0 .
Taking into account that in an adiabatic process the temperature T varies in accordance with the power law T/Te = (p/pe ) Nus = γ−1 γ · pB QB ε γ − 1 . . Neglecting the difference (αK v0 2 − αK vB 2 )/2 in the kinetic energies of the gas before and after compression and assuming the process of gas compression in the blower to be adiabatic (qex = 0).55) where Q = vS is the volume ﬂow rate of gas (ρB QB = ρ0 Q0 = const. we have α K vB 2 α K v0 2 − + (J0 − JB ) 2 2 ˙ · M = qex + Nus where J = ein + p/ρ is the gas enthalpy. i. γ−1 γ−1 γ . Taking into account that the mass ﬂow ˙ rate of gas M = ρvS through all gaspipeline crosssections in stationary ﬂow remains constant.e.31).).4. From the total energy balance equation (1.35) written for a mass of gas going between the entrance crosssection x1 and the exit crosssection x2 of a blower in the case of stationary ﬂow it follows that p αk v 2 + ein + 2 ρ · ρvS x2 − αk v 2 p + ein + 2 ρ · ρvS x1 = qex + Nus where qex is the external heat inﬂow (qex > 0) to the gas or heat outﬂow from the gas (qex < 0). the useful power of the blower. Nus is the useful power of mechanical forces acting on the gas. we get Nus ∼ M · (J0 − JB ) = ρB QB · (J0 − JB ) = ˙ (4. For a perfect gas J = Cp T + const. and the following relations are valid Nus = ρB QB · (J0 − JB ) = ρB Cp QB · (T0 − TB ) = or Nus = Cp · pB QB C p − Cv γ T0 −1 = · pB QB TB γ−1 T0 −1 TB Cp pB QB (T0 − TB ) R TB where γ = Cp /Cv is the adiabatic index (for methane γ ≈ 1.7 Modeling of Blower Operation 105 Useful Power of a Blower Let us show now how we can estimate the useful power needed for gas compression from pressure pB at the blower entrance to pressure p0 at the blower exit.56) Here ε = p0 /pB is the compression ratio. we ﬁnally obtain (4.
Taking in Eq. Tcr = 195 K) is being transported along a 105km gaspipeline section (D = 1220 × 12 mm.31 − 1 ∼ 10. (4.4 m3 s−1 . Answer. It is given that the pressure at the end of the pipeline section is 3.5 · 106 · · 1. The pressure pB before the blower is 3. Natural gas (R = 18. It is required to determine the useful power Nus spent by the blower for gas compression. Exercise. the temperature of the gas at the suction line is 12 ◦ C.31 − 1 60 Hence the useful power is 10.7 J kg K−1 . It is required to determine the compression ratio of the gas ε and the number of revolutions n of the blower rotors needed to ensure a commercial ﬂow rate of 21 billion m3 year−1 in the gaspipeline (the number of working days in a year is taken to be 350). the compression ratio of a gas with γ = 1.8 MPa and at the blower suction line is 4.31 500 · 3. the temperature of the gas after compression is expected to be 30 ◦ C. the volume ﬂow rate at the entrance QB of the blower is 500 m3 min−1 . Exercise.4 1.2 MW.2 · 106 W.82 kg kmol−1 . µ 18.2 MW. (4.75 MPa.4.33) x = L and Qk = M/ρst . (4.31 is 1.31−1 1.48) Nus = 1. = 1. Solution.82 . Finally in this chapter let us consider an exercise on the calculation of a gaspipeline section in combination with a compressor station.106 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines A similar formula written as a function of the gas parameters at the exit of the blower has the form Nus = 1−γ γ · p0 Q0 1 − ε γ γ−1 . the temperature of the surrounding ground is 8 ◦ C. = 0. = 350 · 24 · 3600 8314 ∼ R0 −1 = R= = 441. π2 d 5 Now calculate successively the quantities in this relation Qk = 21 · 109 ∼ 694. 10. we have p0 2 = pL 2 + 16λ · Zav RTav ρst 2 · L · Qk 2 .5 MPa. The useful power is obtained from Eq.03 mm) with the help of two identical GPA equipped with blowers 370181 connected in parallel.7 MPa. ˙ Solution. pcr = 4.57) Here the subscript 0 denotes that the relevant quantity is taken at the exit of the blower.
= A ∼ 21.0 MPa.75) · (291.6 · 0.03 1196 0. = Once the required compression ratio has been obtained.8 · 106 )2 + 21. = The gas overcompressibility factor Z is calculated with Eq.668 ∼ 0. Hence.94 · 1012 Pa2 . = After this we calculate the pressure p0 at the beginning of the pipeline section. = The obtained value shows that the average pressure pav in the pipeline section is equal to 2/3 · (4. = We see that the obtained value of p0 is practically unchanged.94 · 1012 ∼ 6. which should be provided by the blowers 370181. it is time to calculate the gas parameters at the suction line of each of the blowers connected .0093.6 K.890.0093 · 0.42 3. = Calculate the average gas temperature Tav in the pipeline section Tav = Tex + T0 − TL T0.922.73 · 1012 ∼ 6. We begin with the factor A A= = 16 · λ · Zav RTav ρst 2 · L · Qk 2 π2 d 5 16 · 0.6 ◦ C = 291.0 · 106 Pa or 6.0/4. Thus the compression ratio ε. = p0 = (3.922 · 441. RTet 441. (2.5 + 6.34 MPa.34/4.067 · 2 · 0.7 Modeling of Blower Operation 107 ρst = pst 10 125 ∼ −3 = = 0.8/4. Performing the second approximation for pressure p = pav = 5.28.12 /10.75) · (291.7 ∼ 1.7 · 293 2 d 0.7 · 291.668 ∼ 0. the calculation should be corrected.6/195)−3.2 λ = 0.8 MPa.783 kg m .6) ∼ 5. which is greater than the expected = 3.16) taking the average pressure pav in a ﬁrst approximation to be equal to the pressure at the end of the pipeline section and the temperature to be the temperature averaged over the section Z = 1 − 0.8 · 106 )2 + 22.4273 · (5. − Tex ln TL − Tex =8+ 30 − 12 30 − 8 ln 12 − 8 ∼ 18.34 MPa. we get Z = 1 − 0.1 · 106 Pa or 6. is 6.7832 · 105 000 · 694.142 · 1.1 MPa.1965 ∼ 22.4273 · (3.6/195)−3.2 ∼ 0.067 · = 0.73 · 1012 = and then calculate the pressure p0 p0 = (3.4.
which is = = less than the required value 1.28. = = ZB RTB 0. 2.90 · 490 · 288 ∼ = 1. As a result we have n = 0. We have ZB = 1 − 0. n n As the compression ratio ε has already been found to be equal to 1.7 · 285 n0 (QB )r = QB n0 n0 3 = 391 · m min−1 . to select n/n0 in such a way that a point with coordinates (QB )r = 391/(n/n0 ) and ε = 1.05 ⇒ n/n0 = 1.6. = pB 4. = (QB )r = 391/0. Let now (n/n0 )r = 1.062 · n/n0 . = (QB )r = 391/0. Therefore (n/n0 )r should be increased.7 · 285 In a parallel connection of identical blowers the ﬂow rate is distributed equally between them.75 285 195 −3.0/1.716 kg m−3 .716 Determine the reduced parameters of the operating regime of the centrifugal blower: n n0 = r n n0 n Zr Rr Tr = ZB RTB n0 n 0.28 (see Fig.062 ∼ 0.05/1. 4.0 ⇒ n/n0 = 1. = = the solution is found. = . We take (n/n0 )r = 1.895 · 441.25 (see Fig. 1.942 ∼ 415 m3 min−1 ⇒ ε ∼ 1. 0.062 ∼ 0. 4.989. The selection is performed by the iteration method. therefore it is QB = Qk · ρst /ρB = [(21 000/2)/350] · 106 0.895 · 441.989 · n0 = 0. it is necessary.108 4 Modeling and Calculation of Stationary Operating Regimes of Oil and Gas Pipelines inparallel.28 would lie on the characteristic (n/n0 )r = 1.6).942.062 · . 24 · 60 41.7 · 4.895.989 · 4800 ∼ 4750 rpm. Hence. using the characteristics of the blower 370181 depicted in Fig.668 ∼ 0.6).783 ∼ · = 391 m3 min−1 . 4.28.4273 · ρB = 4.989 ∼ 395 m3 min−1 ⇒ ε ∼ 1.7 · 106 ∼ 41.
5.109 5 Closed Mathematical Models of OneDimensional NonStationary Flows of Fluid and Gas in a Pipeline In this chapter are considered the most important models of onedimensional nonstationary ﬂows of ﬂuid and gas in pipelines. For example. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. Weinheim ISBN: 9783527408337 ρ0 S0 d0 ρ0 S0 + Eδ K · ∂p . The equations obtained in Chapter 1 are used as a starting point. E = 2 · 1011 Pa (pipe steel). p − p0 = 1. the variation of ﬂuid density ρ is only 1 kg m−3 .6 Pa. Michael V. for λ = 0. = • the tangential friction stress τ  at the pipeline walls in accordance with w Eq. KGaA.6). ∂t ∂x With regard to the accepted assumptions this equation could be transformed to the following form dS ∂p dρ ∂p ∂ρS ∂S ∂ρ =ρ +S ≈ ρ0 + S0 = ∂t ∂t ∂t dp ∂t dp ∂t Modeling of Oil Product and Gas Pipeline Transportation. v = 1. = The ﬁrst equation of the model is continuity equation ∂ρS ∂ρvS + = 0. p − p0 = 107 Pa (≈100 atm).28) is determined by the formula τw  = λ(Re. Such an assumption is called the hypothesis of quasistationarity.02. ρ0 = 103 kg m−3 . at d0 = 500 mm. whereas S0 ∼ 1960 cm2 . that is ρ ρ0 . at ρ0 = 1000 kg m−3 . then τw  ∼ 5. For example. where S = πd0 3 /4Eδ · (p − p0 ) or nominal value S0 . that is S S = S0 d0 /Eδ · (p − p0 ). δ = 10 mm. K = 103 MPa (109 Pa). For example. (2. where ρ = ρ0 × (p − p0 )/K in accordance with Eq. (1.1 A Model of NonStationary Isothermal Flow of a Slightly Compressible Fluid in a Pipeline At the basis of this model lie the following assumptions: • the variation of ﬂuid density ρ is much less than its nominal value ρ0 . the variation of the pipeline crosssection area S is ≈0.0 MPa (106 Pa ≈ 10 atm). ρ0 = 1000 kg m−3 .06 mm. ε) · ρv2 /8 with the factor λ dependent on the governing parameters Re = vd/ν and ε = /d given in the same form as in stationary ﬂow. ∂t .5 cm2 .5 m s−1 . • the variation of pipeline crosssection area S is much less than its S0 . the variation of the pipeline diameter d is 0.
d0 = 500 mm. among which are water.1) is called the speed of wave propagation in the pipeline (c ≈ 1000 m s−1 ).3) For slightly compressible ﬂuids. E = 2 · 1011 Pa. δ = 10 mm. ∂t ∂x The second equation of the model is the momentum equation (1. = With regard to the introduced designation the ﬁrst equation of the model takes the form ∂p ∂v + ρ0 c2 · = 0.5 10 + 109 2 · 1011 · 0. ∂x ρ0 v2 2 . ∂t ∂t ρv ∂v ∂ ∂v ∼ = = ρ0 v ∂x ∂x ∂x ≈ ∂p .01 3 ∼ 895 m s−1 . If ρ0 = 1000 kg m−3 .2) Replacement of τw with the expression containing the average velocity of the ﬂow v yields ρ ∂v ∂v +v ∂t ∂x =− ∂p 1 ρvv −λ − ρg sin α(x). therefore it can be denoted as 1/c2 . oil and oil products. the following simplifying assumptions can be made ∂v ∂v ≈ ρ0 . ∂x ∂x As a result the following equation is obtained ρ0 d0 ρ0 + K Eδ · ∂p ∂v + ρ0 = 0. then c= 1 103 · 0.110 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline ∂v ∂ρvS ≈ ρ0 S0 · . K = 109 Pa. ∂x d0 2 (5. given by c= 1 ρ0 ρ0 d0 + K Eδ (5.10) ρ ∂v ∂v +v ∂t ∂x =− ∂p 4 − τw − ρg sin α(x) ∂x dθ (5. ρ0 v2 ∂ p+ ∂x 2 . where the parameter c. ∂t ∂x The factor in parentheses before the derivative of pressure with respect to time has dimension inverse to the square of the velocity.
5) (see Eq.4) ∂p 1 ρ0 vv ∂v ρ 0 + = −λ(Re.5. ε) · − ρ0 g sin α(x) ∂t ∂x d0 2 This equation is the second equation of the model. 0 ∂t ∂x (5. The system of differential equations (5. asserts in particular that it depends on the instantaneous value of the average ﬂow velocity v(x. ε) − ρ0 g sin α(x) ∂t ∂x d0 2 to determine the two unknown functions p(x. into 2 kinetic energy ρ0 S d [(αk − 1) v2 ] of the intrinsic motion of the ﬂuid layers dt relative to its center of mass and into heat ρ0 gS(v · i0 ) due to the work of internal friction forces. t) dependent on the coordinate x and time t.05 for turbulent ﬂow. hence (ρ0 v2 /2)/ p ≈ v/c.4) requires for its solution initial and boundary conditions. Hence.6) . ε) · ρ0 vv/8. (1. (5. the ratio (ρ0 v2 /2)/ p is negligibly small. from Eq.5) we get 4 dv τw = ρ0 (αk − 1) · + ρ0 g · i0 d dt (5. t) and v(x.02–1. at ρ0 ≈ 1000 kg m−3 and v ≈ 1–2 m s−1 . t) but not on derivatives of the velocities with respect to time and coordinate. and in onedimensional ﬂow appearing as external forces. In the general case (ρ0 v2 /2) ≈ ρ0 v v while p ≈ ρ0 c v. Since v c in pipelines. being also components of the model under consideration. If we take for αk the mean value of this factor.1 A Model of NonStationary Isothermal Flow of a Slightly Compressible Fluid in a Pipeline 111 The last approximation is valid because it is easy to verify that (ρ0 v2 /2) p. (ρ0 v2 /2) ≤ 2000 Pa (≈0. the mathematical model of slightly compressible ﬂuids is represented by a system of two differential equations ∂p + ρ c2 · ∂v = 0. With regard to the given assumptions the momentum equation takes the following form ρ0 ∂p 1 ρ0 vv ∂v + = −λ(Re. In fact the following equation holds 4 d v2 (αk − 1) τw · vS = ρ0 gvS · i0 + ρ0 S d dt 2 (5. Really. The values of the factor αk in this equation vary from 4/3 for laminar ﬂow up to 1. Virtual Mass The hypothesis of quasistationarity in accordance with which the tangential stress τw at the internal surface of the pipe is represented by the equality τw = λ(Re. reﬂecting the transformation of the work of friction forces (4τw · vS/d). whereas p is measured in atmospheres or even tens of atmospheres.04 atm).23)). We will deal with these conditions below.
03). ∂t ∂x ∂ρv 4τw ∂ − ρg sin α. (5. 5.e. • the variation of gaspipeline crosssection area S can be ignored compared S0 . could be neglected.3) only in that in the lefthand side enters not the true ﬂuid density ρ0 but a quantity αk ρ0 differing from ρ0 by the factor αk . i. Substitution of Eq.2) yields ρ0 or αk ρ0 dv ∂p 1 ρ0 vv =− − λ(Re. but for laminar ﬂow it is greater (αk ≈ 4/3).2 A Model of NonStationary Gas Flow in a Pipeline At the basis of this model lie the following assumptions: • the transported media (gas) is compressible. ε) · − ρ0 g sin α(x) dt ∂x d0 2 (5. (5.112 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline indicating that the tangential stress τw contains the term ρ0 (αk − 1) · dv/dt proportional to the ﬂuid particle acceleration. ρ = ρ(p.7) differs from Eq. the inertial properties of a ﬂuid in nonstationary processes are characterized by changing density. i. with the area itself S0 . S = • the internal energy of the gas is e in = Cv T + const. Therefore S ∼ S0 = πd0 2 /4 = const. the ﬁrst term vanishes. It is evident that if v = const. Hence. that is the term ρ0 g · i0 . . to the latter is added a certain quantity dependent on the ﬂow regime.. In developed turbulent ﬂows this change is slight (αk ≈ 1. The physical nature of this term ˙ is hidden in the origin of the additional resistance force ρ0 (αk − 1) · v caused by realignment of the internal structure of the ﬂow taking place even when there is dissipation of mechanical energy into heat owing to the work of the viscous friction force. The system of basic equations is ∂ρ + ∂ρv = 0. The quantity ρ0 (αk − 1) may be called the virtual (additional) mass of the ﬂuid.6) into the momentum equation (5. T).8) The ﬁrst equation of this system is the continuity equation reﬂecting the law of gas mass conservation in each pipeline crosssection.e. + (p + ρv2 ) = − ∂x d0 ∂t ρ ∂ein + v ∂ein = 4qn − p ∂v − ρ · nin ∂t ∂x d0 ∂x (5.7) ∂p dv dv =− − ρ0 (αk − 1) − ρ0 gi0 − ρ0 g sin α(x) dt ∂x dt Equation (5.
1 Wave Equation Let us consider ﬁrst the nonstationary ﬂow of a slightly compressible ﬂuid in a horizontal pipeline (α = 0) neglecting for a while terms accounting for the friction force.5. nin = −λ(Re.3.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline Consider the nonstationary ﬂow of a slightly compressible ﬂuid in a pipeline. t) and T(x.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 113 The second equation of the system is the momentum equation expressing Newton’s second law. ∂t ρ0 ∂x . the system of equations (5. ε) and Z(p. 8 d0 2 p .g..8) should be added the socalled closing relations. T) are known as functions of their arguments. 0 ∂t ∂x ∂v ∂p ρ 0 + =0 ∂t ∂x or ∂v = − 1 · ∂p . Such an assumption is quite allowable for short pipelines and ﬂuids with not too high viscosity. t) dependent on the coordinate x and time t. ein (T) = Cv · T + const. ρ= Z(p. The basic equations of such ﬂow are represented by the system of equations (5.4) takes the form ∂p + ρ c2 · ∂v = 0. ε) · . 5. τw = λ(Re.8) represents a closed system of three partial differential equations for three unknown functions p(x. ε) 1 ρv3 · ρv2 . In such a case the system of equations (5. ∂x ρ0 c2 ∂t (5.9) 1 ∂p ∂v =− · . qn = −κ · (T − Tex ). To the system of equations (5. The third equation of the system is the equation of heat inﬂow following from the laws of total energy conservation of the ﬂow and the variation of mechanical energy of the transported media.4). T) · RT Hence. e. 5. v(x. if λ(Re. t).
12) is a solution of Eq. the propagation velocity of a certain value of function f1 or f2 is identical and equal to c.11) Equations (5. ∂x2 ∂t ∂ 2p = f1ξξ + f2ηη . Let us show that Eq.e.12) the ﬁrst of these functions is dependent only on the argument ξ = x − ct and the second on η = x + ct. The velocity of ﬂuid v(x.12) gives a solution of Eq. ∂x2 Hence. The function f1 (x − ct) represents a traveling wave in the positive direction of the xaxis whereas the function f2 (x + ct) represents a traveling wave in the negative direction of this axis. The form of each of the functions f1 and f2 is determined by the initial conditions for pressure and velocity distributions in the pipeline. t) ∂ 2v ∂ 2v = c2 2 . A similar equation can be obtained for ﬂuid velocity v(x. (5. t) = g1 (x − ct) + g2 (x + ct). The magnitude of the velocity propagation of both waves. 0).10). ∂t2 ∂p = f1ξ + f2η . ∂x c2 ∂ 2p ∂ 2p = 2. t) = f1 (x − ct) + f2 (x + ct) (5. 0) and v(x. .10) This equation is called the wave equation since it describes the propagation of waves and is encountered in different ﬁelds of physics. Eq. as well as by the boundary conditions at the pipeline ends.114 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Differentiation of the ﬁrst equation with respect to t and the second with respect to x subject to the condition vx. 2 ∂t ∂x (5. 2 ∂t ∂x (5. ∂t ∂ 2p = c2 f1ξξ + c2 f2ηη = c2 (f1ξξ + f2ηη ). that is by p(x. We have ∂p = −cf1ξ + cf2η .t = vt. (5.10) and (5. i.x yields the equation for p(x. t) ∂ 2p ∂ 2p = c2 2 . (5.10) for arbitrary functions f1 and f2 . t) is determined by the formula v(x.11) represent partial differential equations whose general solution is expressed by two arbitrary functions p(x. (5.
3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 115 Functions g1 (ξ) and g2 (η) are expressed through functions f1 (ξ) and f2 (η). that is when waves reﬂected from the beginning and the end of a pipeline could be neglected. and x + ct = const. taking into account that g1 and f1 depend only on ξ and g2 and f2 only on η. (5.3. t) + ρ0 c · v(x. 5.13) With the help of Eqs. 5. at each characteristic x − ct = const. p − ρ0 c · v = 2 · f2 (x + ct) + const. ρ0 cg2 = −f2 + const. Addition and subtraction of Eqs.13) one can ﬁnd solutions to different problems.. t) = f1 (x − ct) − f2 (x + ct) = const.12) and (5. We will consider some of them.9) follows g + g = 1 · [f − f ]. with positive slope. From these expressions it is seen that at those points of the plane (x. t) and I2 = p(x.1). but it is very useful in the case of a very long pipeline when boundary effects can be ignored. (5. ρ0 or ρ0 cg1ξ = f1ξ . The lines x − ct = const.. 1ξ 2η 2η ρ0 c 1ξ 1 −c · g1ξ + c · g2η = − [f1ξ + f2η ]. dx/dt = +c. the ﬁrst Riemann invariant I1 is conserved whereas at each characteristic . yields ρ0 cg1 = f1 + const. t) are called the Riemann invariants. From Eq. (5. (5. t) where (x − ct) remains constant the expression I1 = p + ρ0 c · v is also constant and at those points of the plane (x.13) yield p + ρ0 c · v = 2 · f1 (x − ct) + const. ρ0 cg2η = −f2η .5. Integration of the ﬁrst equation with respect to ξ and the second with respect to η. Suppose also that friction forces are absent. t) where (x + ct) remains constant the expression I2 = p − ρ0 c · v is also constant (see Fig. t) − ρ0 c · v(x. Of course it is only a model of a real pipeline. are called characteristics of the wave equation and the quantities I1 = p(x. From here follows ρ0 c · v(x.12) and (5. Hence.2 Propagation of Waves in an Inﬁnite Pipeline A pipeline is called inﬁnite when it runs in the direction of the xaxis from −∞ to +∞.
Consider a plane (x. Since at the characteristic x − ct = x1 the ﬁrst Riemann invariant I1 is constant. 5. t) = pA (x2 . 0) = (x). t) + ρ0 c · vM (x. 0) v (x. with negative slope. t) = + 2ρ0 c 2 . at lines x = −ct + const. i. At the characteristic x + ct = x2 the second Riemann invariant I2 is constant pM (x. 0) = (x). : I1 = p + ρ0 cv = const. 0) + ρ0 c · vA (x1 . t) = M + 2ρ0 c 2 or p (x. dx/dt = −c. t)plane. Resolving the system of obtained equations relative to pM (x. 0). t) = p(x1 . 0) + ρ c · v(x1 . Let at an initial instance of time t = 0 in an inﬁnite pipeline (−∞ < x < +∞) there be distributions of pressure p(x. we get p (x. what motion appears in the pipeline at t > 0.116 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Figure 5. t) and vM (x. 0) − ρ0 c · vA (x2 . t) be an arbitrary chosen point of this plane at t > 0. 0). it is required to ﬁnd the functions p(x. t) = (x1 ) + (x2 ) + ρ c · (x1 ) − (x2 ) . 0) − v(x2 . Let M(x.1. 0) − p(x2 . 0) v(x1 . 0) = (x) and ﬂuid velocity v(x. t) and v(x.10) and (5. the second Riemann invariant I2 is conserved: at lines x = ct + const. t) satisfying Eqs.e. M 0 2 2 p(x1 . 0) . : I2 = p − ρ0 cv = const. x + ct = const. 0) = (x). v(x.1 Characteristics on (x.. 0) + v(x2 .11) and initial conditions p(x. t) depicted in Fig. Solution. we can write pM (x. 0) + p(x2 . t) − ρ0 c · vM (x. It is required to determine. Problem. t) = pA (x1 . M 0 2 2 (x1 ) − (x2 ) (x1 ) + (x2 ) vM (x. t). t) straight lines (characteristics) MA: x − ct = x1 and MB: x + ct = x2 . Draw from the point M(x. (5.
a perturbation (signal) is propagated with velocity c from the initial pipeline crosssection.14) are called the d’Alambert formulas. 0) = 0 and the pressure constant p(x. Solution. 5. Consider now the region x < ct of the pipeline to which at time t have come perturbations from the initial crosssection of the pipeline. 5. t) = (t). It is required to determine the pattern of ﬂuid ﬂow in the pipeline at t > 0. 0 M 2 2 v (x. that is its velocity is 0 and the pressure is p0 .2. t) = p0 − p0 + 0 + 0 = 0 N 2ρ0 c 2 This result has a simple physical meaning.14) give the following equations p (x. Problem. t) = (x − ct) + (x + ct) + ρ c · (x − ct) − (x + ct) .3.5.14) Since functions (x) and (x) are known. Let. In the region x > ct of the pipeline. 0) = p0 . 0 0 N 2 2 (5. Consider a plane (x. the problem is completely solved. to which the instant of time t has not yet come. the ﬂuid is at rest as before.3 Propagation of Waves in a SemiInﬁnite Pipeline A pipeline is called semiinﬁnite when it has the initial crosssection (x = 0) and runs from it in the positive direction of the xaxis (x > 0) to inﬁnity. Draw through a . It is required to determine what kind of velocity and pressure waves begin to propagate in a semiinﬁnite pipeline caused by perturbations created at the left end of the pipeline. we receive the solution of the problem p (x. At points N of this plane located below the straight line x = ct the d’Alambert formulas (5. It is a model of a pipeline in which the conditions at one of the ends (left end) are taken into account whereas the inﬂuence of another end (right end) is neglected. t) = (x − ct) − (x + ct) + (x − ct) + (x + ct) M 2ρ0 c 2 (5. at the initial instant of time t = 0 the ﬂuid in a semiinﬁnite (0 < x < +∞) pipeline be quiescent v(x. x ≥ 0 as depicted in Fig. Equalities (5. The ﬂuid velocity in the initial crosssection x = 0 at t > 0 suddenly begins to change with a law v(0.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 117 Substituting in these relations instead of x1 and x2 their expression through x and t. t) at t > 0. Draw through the origin of the coordinates in this plane the characteristic x = ct with positive slope.15) v (x. t) = p0 + p0 + ρ c · 0 − 0 = p .
t0 ). c c (5. pM (x. having been initially equal to p0 . t) − ρ0 c · vM (x. t0 ) as a result of the increase in velocity vC (0. t0 ). 0) − ρ0 c · vC (x1 . t0 ) − ρ0 c · vC (0. 0) = p0 − 0 = p0 .17) x vM (x. the characteristic CA (Fig. Therefore relations (5. t) = vC (0.16) Note that t0 is determined through the coordinates (x. t) = pB − ρ0 c · vB = p0 − ρ0 c · 0 = p0 From this follows pM (x. t0 ) in this crosssection. With the help of the condition at the characteristic CA consisting in constancy of the second Riemann invariant. t) + ρ0 c · vM (x. located at initial crosssection of the pipeline (time t0 ). has varied over ρ0 c · vC (0. . t − t− = p0 + ρ0 c · . 5. t) of the point M by the formula t0 = t − x/c (the characteristic equation for MC is x = c · (t − t0 )). t) = p0 + ρ0 c · vC (0. Since the pressure and velocity of ﬂuid at points C and B are now known.118 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Figure 5. point M(x. t) pM (x. one can ﬁnd the pressure at the point M(x. From this follows pC (0. Hence the pressure pC at the initial pipeline crosssection. and relevant boundary condition we ﬁnd the pressure in the initial pipeline crosssection at time t0 pC (0. t0 ) = p0 + ρ0 c · vC (0.16) have the ﬁnal form x x pM (x. t) of this region two characteristics MC and MB and from the point C. t) = p0 + ρ0 c · vC 0. c where is a function determining the velocity change at the initial (x = 0) pipeline crosssection not at the instant of time t but at a later time t − x/c. t0 ) (5. t0 ) = pA (x1 . t) = t− .2). vM (x. t) = pC + ρ0 c · vC = (p0 + ρ0 c · v0 ) + ρ0 c · v0 .2 Diagram illustrating the problem on traveling waves.
To determine a unique solution for p and we need an additional relation. at the left end of the pipeline can be placed a piston pump providing constant delivery of ﬂuid into the pipeline. dependent on the type of equipment set at the end crosssections of the pipeline section under consideration.6 atm. Along it the condition p − ρ0 c · v = const. should be obeyed. that is at x = 0. = Answer. The pumping in quiescent diesel fuel (ρ0 = 840 kg m−3 . t > 0 and x = L. that is one more boundary condition. Solution. For example. c = 1060 m s−1 ) in a semiinﬁnite pipeline has begun with constant ﬂow rate so that v0 = 1. besides the initial conditions p(x. 0) and v(x. 1.3. Formulas (5. Along this characteristic the compatibility condition p + ρ0 c · v = const.5.17) gives p(0. boundary conditions reﬂecting the interaction of the pipeline section under consideration with equipment located at the pipeline ends. Analogously.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 119 It is clear that the quantity x/c is equal to the time in which the perturbation (signal) reaches the considered crosssection x from the initial one. 5. t) − p0 = ρ0 c · v(0.15) and (5.5 m s−1 . 0).5 ∼ 1. t) = ρ0 cv0 = 840 · 1060 · 1. Thus at points on this boundary there is always one algebraic relation between the quantities p and v. The ﬁrst formula (5.34 · 106 MPa or approximately 13.17) show that any velocity change at the initial pipeline crosssection propagates rightwards along the pipeline as a traveling wave giving ˜ rise to a pressure traveling wave exceeding the initial value p0 by ρ0 n · vC . Through each point of the left pipeline section boundary x = 0. The form of the boundary conditions could be of a great variety.34 MPa. t > 0 of the pipeline section goes only one characteristic with positive slope x = ct + const. Thus at these points there should be an additional boundary condition. Formulas (5.17) give the complete solution of the considered problem. through each point of the right boundary x = L. Now we should say something about the number of these conditions. Exercise. It is required to determine by how much the pressure has increased at the pumping crosssection. t > 0 passes only one characteristic of the wave equation. is obeyed.4 Propagation of Waves in a Bounded Pipeline Section Consider the problem on wave interaction in a limited pipeline section (0 ≤ x ≤ L) between the initial crosssection x = 0 and the ﬁnal crosssection x = L assuming the absence of friction. In such a problem we need. t > 0. Then . namely the characteristic of negative slope x = −ct + const.
at t > 0. It is required to determine the motion generated in the pipeline at t > 0 (Fig. The end crosssection x = L of the pipeline is open to the atmosphere. If the right end of the pipeline is open. so that the pressure at the crosssection is held constant p(L. t) = p0 . 0) = 0) and the pressure constant (p(x.3). t) = p0 = const. Figure 5. Find now the solution in the region 2 restricted by the triangle OCD (2L/c is the time of the wave double path lengthwise in the pipeline section). t) = (t). t > 0 corresponding to the variability domain of the problem (Fig.3 Interaction of waves in the pipeline section.120 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline the boundary condition at x = 0 would be v(0. At t > 0 the ﬂuid starts to be delivered into the pipeline by the law v(0.3). 1. Let a ﬂuid in the pipeline section 0 ≤ x ≤ L be initially (at t = 0) quiescent (v(x. at all t > 0. Consider on the plane (x. 0) = p0 ). i. the boundary condition at the right end would be p(0. is as follows: p(x. Solution. The solution in the region 1 enclosed by the triangle OCL. t) = v0 = const. . t) = 0. v(x. t) = p0 . 2. the region which the perturbation has not yet reached. 5.e. 5. Of course there are also many other possible boundary conditions. t) a strip 0 ≤ x ≤ L. Problem.
pN = p0 . then vF (L. t) = p0 + ρ0 c · vS (t) = p0 + ρ0 c · (t).4) describing nonstationary ﬂow of a slightly compressible ﬂuid with regard to viscous friction forces ∂p + ρ0 c2 · ∂v = 0.5.3. ε) 1 ρ0 v − ρ g sin α(x). t) = 2 · (t − L/c). Determine ﬁrst the ﬂuid velocity vF at an arbitrary point F of the right boundary pF + ρ0 c · vF = pS + ρ0 c · vS = [p0 + ρ0 c · (tS )] + ρ0 c · (tS ) Since pF = p0 and tS = t − L/c.5 Method of Characteristics Let us return to the system of equations (5. t) of the region under consideration pM + ρ0 c · vM = pS + ρ0 c · vS = p0 + 2ρ0 c · pM − ρ0 c · vM = p0 − ρ0 c · 0 = p0 from which follows pM (x. 5. 5 and other regions of the strip under consideration could be found. t− 3. ∂t ∂x 2 ρ ∂v + ∂p = −λ(Re. pS (0. c (tS ). c x pN − ρ0 c · vN = pF − ρ0 c · vF = p0 − 2ρ0 c · t− c Hence. In a similar manner by the method of characteristics the solution in regions 4. vN = 2 (t − x/c). 0 0 ∂t ∂x d0 2 . t) of the considered region x pN + ρ0 c · vN = pR + ρ0 c · vR = p0 + 2ρ0 c · t− . t) = p0 + ρ0 c · x vM (x. After this we get the pressure and velocity at the arbitrary point N(x. Find the solution in the region 3 bounded by the triangle CDG (3L/c is the time of the wave triple path lengthwise in the pipeline section). Then obtain the pressure and velocity at an arbitrary point M(x.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 121 Determine ﬁrst the pressure pS at an arbitrary point S of the left boundary pS − ρ0 c · vS = pQ − ρ0 c · vQ = p0 − 0. t) = t− c x .
= −λ dt 2d d (p − ρ0 c · v)η=const. 2d If at the plane (x. ξ=const. Similarly it is true that ∂A ∂A −c· = ∂t ∂x dA dt η=const. the expression on the lefthand side is the derivative of the function A(x. which for the wave equation are called characteristics. dt 2d (5. is valid. 2d ξ=const. = −c. t) in the direction of the ﬁrst characteristic (ξ = const. ρ0 cvv + ρ0 cg sin α. 0 ξ=const.. η=const. one can write the above obtained equations as follows dp dt dp dt or + ρ0 c · − ρ0 c · dv dt dv dt = −λ =λ ρ0 cvv − ρ0 cg sin α. Using now the notion of a directional derivative. i. x − ct = ξ = const. it is easy to reveal that for any parameter A(x. 2d ρ0 cvv d (p + ρ c · v) − ρ0 cg sin α.e. t) ∂A ∂A +c· = ∂t ∂x dA dt ξ=const. dt dx 2.. t) in the direction of the second characteristic (η = const. x + ct = η = const. t) we consider straight lines determined by the equations dx = c. 2d In a similar manner after subtraction of the second equation multiplied by c from the ﬁrst one we obtain ∂p ∂p −c ∂t ∂x − ρ0 c · ∂v ∂v −c ∂t ∂x =λ ρ0 cvv + cρ0 g sin α.18) .). This means that the expression on the lefthand side is the derivative of the function A(x. η=const.122 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Multiplication of the second equation by c and addition of the result to the ﬁrst equation yields ∂p ∂p +c ∂t ∂x + ρ0 c · ∂v ∂v +c ∂t ∂x = −λ ρ0 cvv − cρ0 g sin α.). = λ ρ0 cvv + ρ0 cg sin α. dt 1.
5. t) a rectangular grid with coordinate step x and time step t = x/c (Fig.m = v(xk . t). tm ) of the grid functions at the grid nodes. Figure 5. especially when numerical methods are used.m of grid functions at t = tm . tm ) and vk. t) is replaced with discrete values pk. tm−1 ) in the pipeline be known. t η=const. (5. the quantities I1 and I2 are not constants at relevant characteristics. t) and v(x.18) with ﬁnite differences along the characteristics AM and BM we obtain (p + ρ0 c · v) = −c · φA . Continuous distribution of the sought functions p(x. .18) vanish. (p − ρ0 c · v) = c · φB . Nevertheless. This means that along the characteristics of positive slope is conserved the quantity I1 = p0 + ρ0 c · v.m = p(xk .3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 123 At λ = 0 and α = 0 the righthand sides of Eqs. Let for example at the instant of time tm−1 (in particular at t = 0). t ξ=const. This conclusion is consistent with the results obtained above for the wave equation. If λ = 0. Replacing the directional derivatives in Eqs. whereas the quantity I2 = p0 − ρ0 c · v is conserved along the characteristic of negative slope. tm ) be an arbitrary point on the plane (x. the distributions of pressure p = p(x. Through the nodes of the resulting grid let us draw characteristics x = ct + const. Suppose that all values pk.18) may be used to calculate various nonstationary ﬂows in a pipeline. and x = −ct + const. tm−1 ) and velocity v = v(x.4).5. (5. Consider on the plane (x.m and vk.m−1 and vk.4 Design diagram of the characteristic method. Let M(xk . (5. Eqs. Show how the values of these parameters at the next instant of time tm = tm−1 + t could be calculated.m−1 are known and it is required to ﬁnd the values pk. respectively. of positive and negative slope.
m−1 − vk+1.m−1 + pk+1.m−1 pk.m−1 + x · φk−1.m − ρ0 c · vk.6 Initial. η=const.m−1 + vk+1.124 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline vv where φ = λ ρ02d + ρ0 g · sin α. In order to know how the nonstationary process progresses it is necessary to have information about the initial and boundary conditions. Boundary and Conjugation Conditions Let us investigate nonstationary ﬂuid ﬂow at a pipeline section 0 ≤ x ≤ L starting from a certain instant of time t = 0 taken as the initial time.m−1 + φk+1. The ﬁrst information is called the initial conditions and the second the boundary conditions. = (pM + ρ0 c · vM ) − (pA + ρ0 c · vA ) = (pM − ρ0 c · vM ) − (pB − ρ0 c · vB ). at the crosssection x = 0 and x = L. In addition (p + ρ0 c · v) (p − ρ0 c · v) ξ=const. From this system of linear equations we get the pressure pk.m + ρ0 c · vk. the recurrent formulas (5.m−1 + vk−1.g.m = pk−1. e.m−1 ) 2 (5.m−1 2ρ0 c 2 x · (φk−1.19) vk.m of the ﬂow p vk−1.m−1 + ρ0 c · 2 2 x + · (φk+1.19). 5. that is to know the state of the ﬂow before starting and what is happening at the edges of the pipeline sections.m−1 + ρ0 c · vk−1.19) give the solution of the formulated problem.m−1 ) − 2ρ0 c Hence.m−1 .m−1 − pk.m = pk+1.m = pk−1. .m−1 x · φk+1.m−1 − pk+1.m−1 − ρ0 c · vk+1. Since at the ﬁrst instant of time we can take the initial values of the pressure and velocity of the ﬂow at t = 0.m−1 − φk−1.3. or pk. because they allow one to calculate the pressure and velocity of the ﬂow at the following instant of time tm from known values of these parameters at the preceding instant of time tm−1 .m and the velocity vk. From this follows a system of equations to determine the pressure pM and velocity vM of the ﬂuid at point M through known values of these parameters at points A and B pM + ρ0 c · vM = pA + ρ0 c · vA − t · cφA pM − ρ0 c · vM = pB − ρ0 c · vB + t · cφB . we can get ﬂow parameters at an arbitrary instant of time t > 0. then calculating successively the pressure and velocity with formulas (5.m = k−1.
5. 0) = Q0 and the distribution of head be H(x. .m ) = 0. i0 = (H0 − Hk )/L.m p1. tm−1 = t0 = 0) may be accepted v(x. It gives one condition for two unknown quantities p1. for example. t) and velocity vM (0. v) = 0 expressing the relation between pressure pM (0. t > 0 determining the pipeline section except its edges – beginning (x = 0) and ending (x = L). zk = z(xk ). 0) = or vk.5a).m−1 − ρ0 c · v2.m−1 − ρ0 c · v2. . . Boundary Conditions The formulas (5. boundary conditions at points on the left pipeline section boundary may be represented as a system of equations x = 0.m − ρ0 c · v1.1 = Q0 = const. only one characteristic of the positive slope dx/dt = +c comes from the domain of integration to a point E(x = L) of the right pipeline section boundary (Fig. Therefore the boundary condition at points on . xk = (k − 1) · x. 2. . but often as the initial state is taken the stationary ﬂuid ﬂow existing in the pipeline at the initial instant of time.. x = L/N (N being the number of parts into which the pipeline section is divided). N + 1. 0) = H0 − i0 · x. 0) = ρ0 g · [H0 − i0 · x − z(x)] S0 Q0 = const. the known ﬂuid ﬂow rate be Q = Q(x. x · ϕ2. 0) is the head at the beginning of the pipeline section.m = p2. As a rule this condition models the operation of a pumping station and is represented by its (Q − H) characteristic. x1 = 0. Then as initial conditions (m = 1.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 125 Initial Conditions The state of the pipeline section at the initial instant of time can be arbitrary. xN+1 = L. v1.5b). t > 0 : p1.m . t) at the initial crosssection of the pipeline. Such a condition can be an algebraic equation F(p.m−1 + F(p1. Thus. for stationary ﬂow.19) permit one to ﬁnd p and v at any point on the strip 0 < x < L. Only one characteristic of the negative slope dx/dt = −c comes from the integration domain to a point M(x = 0) of the left pipeline section boundary (Fig. 0) is the head at the end of the pipeline section. where Hk = H(L. Let.m and v1.m−1 Similarly.m = p2. where H0 = H(0.m − ρ0 c · v1. p(x. pk.m−1 + x · φ2.. 5.m−1 therefore an additional condition is needed.1 = ρ0 g(H0 − i0 · xk − zk ) S0 where k = 1.5.
vN+1.m .m + ρ0 c · v k. vk.m−1 + ρ0 c · vN. at a pipeline crosssection x∗ ejection or injection of ﬂuid with ﬂow rate q (q < 0 ejection.g. Let take place. S + − Then to calculate the three unknown parameters pk. Such a discontinuity requires additional conditions called conjugation conditions.m = S0 with p+ k. e.m of nonstationary ﬂow at point x∗ = xk we use the following system of three linear equations − pk. t) and velocity vN (L. t) at the end of the pipeline section. at the crosssection x∗ . Then at the crosssection x∗ the following conditions should be obeyed: p+ (x∗ .m = pN.m−1 + x · φk+1.126 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Figure 5.m .m−1 pk. Let us denote the values of the parameters before ejection or injection with superscript (−) and after ejection or injection with superscript (+). .m + ρ0 c · vN+1.m−1 Conjugation Conditions If the equipment responsible for process nonstationarity is located inside the pipeline section.m . v+ (x∗ . t).m = pk−1. vk. t) − v− (x∗ .m = pk. x · ϕN.m−1 − x · φk−1. then at this crosssection there can exist a discontinuity of hydraulic parameters.m . Then such a crosssection is characterized by continuity of pressure and discontinuity of ﬂow rate or velocity. 0 < x∗ < L.5 Calculation of p and v at boundary crosssections.m−1 + q v k.m − v− k.m−1 − ρ0 c · vk+1. q > 0 injection). v) = 0 expresses the relation between the pressure pN (L. for example. the right pipeline section boundary may be represented as a system of two equations: x = L.m − ρ0 c · v+ k. It should be noted that there are also possibilities for more complicated boundary conditions.m−1 + ρ0 c · vk−1. t > 0 : pN+1.m−1 − G(pN+1.m = pk+1. t) = p− (x∗ . The dependence G(p. t) = q .m ) = 0.m = p− k.
m−1 + ρ0 c · vk−1.m = ς(tm ) · ρ0 v2 k.m .m − p− k.m = v− k. Thus.7 Hydraulic Shock in Pipes In all the above stated it was assumed by default that the functions ρ(x. vk.5. The essence of hydraulic shock is that the stationary ﬂow of ﬂuid in a pipeline is disturbed by the abrupt closing or opening of a gate valve. The explanation of hydraulic shock was given by Joukovski in his article ‘‘On hydraulic shock in watersupply pipes’’ (1899). since shock pressure can far exceed permissible standards.m with v+ k.m−1 − ρ0 c · vk+1. Similar phenomena occur in the pipeline in other cases when the velocity (ﬂow rate) of the ﬂuid varies in a stepwise manner. t) = v (x∗ .m = pk−1.m .m−1 − p+ − ρ0 c · vk.m /2 x · φk−1. p− k.m + ρ0 c · vk. the model of the valve is represented by the following system of three equations − p k.m = vk.m−1 for the three parameters p+ k. t). The front at which the variation of the hydrodynamic parameters of the ﬂuid takes place has a relatively small extent and propagates in the form of a pressure wave downstream and upstream of the ﬂuid. in engineering there are processes in which these functions vary very quickly with time and in space. t) − p+ (x∗ . t) = ς(t) · ρ0 v (x∗ . t) 2 − p (x∗ .3. Nevertheless. He connected the magnitude of the pressure jump [p] with the properties of ﬂuid .3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 127 In the case when a gate valve is located at the crosssection x∗ = xk . The possibility of hydraulic shock should be taken into account in the exploitation of pipelines. 5.m = pk+1.m−1 + + k. t) 2 where ς(t) is the local resistance factor that varies during valve closing or opening. t) in the differential equations are differentiable with respect to time and coordinate and in any case are certainly continuous.m−1 x · φk+1. leading to pipe breakage and an emergency situation. its operation is modeled by conjugation conditions + − v (x∗ .m p k. The ﬁrst condition means the continuity of the ﬂow rate.m . p(x. whereas the second condition signiﬁes pressure discontinuity at different sides of the valve. t) and v(x. resulting in hard braking or acceleration of the ﬂuid and shock compression of the ﬂuid particles. An example of such a process is hydraulic shock. the switching on or switching off of a pump and so on.
6). (5. though very narrow. It is evident that the mass of ﬂuid ﬂowing into and out of the wave front should be equal Figure 5. introduction of the system of equations (5. To describe the structure of this transition zone needs as a rule a more complicated model than the given one. They have to obey the conditions of ﬂuid mass and momentum conservation. S+ the wave front cannot be prescribed arbitrarily.2)) and [v] the magnitude of the stepwise change in the ﬂuid. p+ . Let the discontinuity front of the ﬂow parameters propagate with velocity D = dxo /dt in the positive direction of the xaxis ˆ (Fig. Therefore the mass of ﬂuid ﬂowing in the wave in time t is determined by the expression ρ− S− (D − v− ) · t. p− . The quantity [A] = A+ − A− is called the jump of parameter A at the discontinuity front. In the same time interval a ﬂuid mass equal to ρ+ S+ (D − v+ ) · t ﬂows out from the wave. 5. v− . from the value of parameter A+ to the left of the discontinuity front up to the value A− of the same parameter to the right of the front. It should be noted that the introduction of stepwise variations (jumps) of hydrodynamic ﬂow parameters is nothing more than a model of the phenomenon under consideration. at their mobile front. . Nevertheless. However the limiting values of these parameters before ρ− .4) into consideration of discontinuity solutions has proved to be very fruitful. or jumps. The wave of the hydraulic shock is characterized by the parameters of ﬂow (and pipeline crosssection area) that are subjected to discontinuities. Conditions at Discontinuities (Jumps) of Hydrodynamic Parameters Now let us obtain conditions which should satisfy the ﬂow parameters at the discontinuity front. In fact each such discontinuity has a transition region.6 Hydraulic shock in a pipeline. S− and after ρ+ .20) where D is the velocity of shock wave propagation in the pipeline (see Eq.128 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline compressibility and the elasticity of the pipe and obtained the following formula [p] = ρ0 D · [v] (5. In the time interval t all ﬂuid particles present at the beginning of this interval at a distance (D − v− ) · t before the front pass through the front. v+ .
Then.5. (5. (5. We can write the last equation in the following form (ρ+ S+ v+2 − ρ+ S+ v+ D) − (ρ− S− v−2 − ρ− S− v− D) = −S− p or [v · (ρvS − Dρ · S)] = −S− · [p].22) We can show that the velocity of the hydraulic shock wave D for a slightly compressible ﬂuid coincides with the velocity of the perturbation propagation in a pipeline with elastic walls. In order to do this we can simplify the relation (5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 129 ρ− S− (D − v− ) = ρ+ S+ (D − v+ ) or ρ+ v+ S+ − Dρ+ S+ = ρ− v− S− − Dρ− S.21) Now we can use the theorem that the momentum variation of the ﬂuid mass that has passed through the wave front is equal to the impulse of the pressure force ρ+ S+ (D − v+ ) t ·v+ − ρ− S− (D − v− ) t ·v− = (p− − p+ )S− · m m t. in accordance with equality (5. D With regard to ρ− ≈ ρ0 and the smallness of the ratio v− /D we get the Joukovski formula [p] = ρ0 D · [v]. If we denote a jump of any parameter by [ ] then the last relation maybe rewritten as [ρvS − ρD · S] = 0.21).21) as follows: ρ0 S0 [v] − DS0 [ρ] − Dρ0 [S] = 0 . In the righthand side of this equation the projection of the reaction force of the pipeline lateral surface onto the xaxis is taken into account. [ρvS − ρD · S] = 0 the obtained equation may be simpliﬁed to (ρ− v− S− − Dρ− S− ) · [v] = −S− · [p] or v− − 1 · ρ− D · [v] = −[p].
(5. [ρS] ≈ ρ0 [S] + S0 [ρ]. It is required to determine how much the pressure rises at the initial crosssection of a steel pipeline (d0 = 0. 0. δ = 10 mm.0 = 903 060 Pa or approximately 9.5 m s−1 .5 · 109 Pa) ﬂows with velocity v = 1.22) and (2.8 m. Solution. ρ0 D [ρ] = ρ0 · [p].1)) D = 1 ρ0 ρ0 d0 + K δ·E = c.3 · 109 Pa) with velocity v = 1. D K 4δ · E Since [p] = 0.22). (5. (5. . K = 1.21 atm.23) Exercise 1.23) [v] = − yields S0 π · d0 3 ρ0 · [p] − DS0 · [p] − ρ0 D · [p] = 0. K [S] = π · d0 3 · [p]. (5. (2.903 MPa. 1 · [p]. K = 1.0 m s−1 in a steel pipeline (d0 = 0. Oil (ρ0 = 870 kg m−3 . δ = 8 mm. Answer.22) the pressure jump [p] = ρ0 D · [v] = 870 · 1038 · 1.130 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline where it is assumed that owing to the smallness of the variations in ρ and S variations the following approximations can be made [ρvS] ≈ ρ0 S0 [v].01 · 2 · 1011 ∼ 1038 m s−1 .516 m.8 870 + 1. It is required to determine how much the pressure rises at the pipeline crosssection before the gate valve suddenly closes. E = 2 · 1011 Pa). 4δ · E From here it follows that the absolute value of the hydraulic shock wave velocity D is equal to the velocity of the perturbation propagation in the pipeline (Eq. The use of Eqs. E = 2 · 1011 Pa) on abrupt switching on of the pumps providing the benzene feed (ρ0 = 750 kg m−3 . Calculate ﬁrst the velocity D of the hydraulic shock wave D=c= 1 870 · 0. we have 1 = D2 · ρ0 ρ0 d0 + K δ·E .5 · 109 0. Exercise 2. = Then determine with Eq.
22) and (5. pM − ρ0 c · vM = p + − ρ0 c · v + – the condition at the linear segment + M of the characteristic BM.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 131 Solution.7). D > 0.008 · 2 · 1011 ∼ 1105 m s−1 . It is required to study the nonstationary ﬂow generated in a pipeline section 0 ≤ x ≤ L by abrupt closing of the gate valve at the righthand end (x = L) of the pipeline (direct hydraulic shock). = Then determine the pressure jump with Eq. It is assumed that the ﬂuid before closing the gate (t = 0) was ﬂowing Figure 5. 1. intersects the discontinuity at point I (Fig.7 atm. Hence the Riemann invariants (in the considered case I2 = p − ρ0 c · v) are conserved at the relevant characteristics. Ignoring friction. i.7 Interaction of a discontinuity with a small perturbation.3 · 10 0. .22) [p] = ρ0 D · [v] = 750 · 1105 · 1. (5.243 MPa. Problem 1. Answer. we have 1. p + − p − = ρ0 c · (v + − v − ) – the condition (5. even when these characteristics intersect the shock front. Calculate ﬁrst the velocity D of the hydraulic shock wave generated by the benzene pumping into the pipeline D=c= 1 750 · 0.e. the wave of discontinuity propagates in the positive direction of the xaxis. for example.22) at point I of the discontinuity.5 = 1 243 125 Pa or approximately 12. Remark on carrying out calculations. (5.5. p − − ρ0 c · v − = pB − ρ0 c · vB – the condition at the linear segment B of the characteristic BM. 2. On the phases of direct hydraulic shock. Subtraction of the ﬁrst relation from the last one with regard to the second relation yields pM − ρ0 c · vM = pB − ρ0 c · vB . 3.516 750 + 9 1.23). Then the characteristic of negative slope x + ct = const. 5. Notice one interesting circumstance following from Eqs. Let.
By convention the solution may be divided into four phases.132 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline with constant velocity v = v0 and friction was absent so that the pressure at all pipeline crosssections was constant p = p0 . The wave front brings the ﬂuid to an abrupt stop and the pressure after the jump is raised by ρ0 c · v0 . Solution. If the pressure is lowered to the saturated vapor Figure 5. When the ﬂuid ﬂow is stopped. The phase of the reﬂected wave runs from t1 = L/c to t2 = 2L/c equal to double the time taken by the wave path along the pipeline section. the ﬂuid continues to ﬂow out of the pipeline with velocity v = −v0 whereas the pressure in the wave reﬂected from the crosssection x = L is lowered stepwise by ρ0 c · v0 becoming lower than its original value. 5. In this phase the shock wave reﬂects at the initial crosssection and begins to move in the opposite direction. 2.8. The solution is shown in Fig.8 Phases of direct hydraulic shock. The phase of direct shock continues for a time t = L/c. lasting from the instant of time t2 = 2L/c to t3 = 3L/c. 3. the hydraulic shock wave appears and travels from the end of the pipeline section to its beginning. In the third phase. 1. It is assumed also that at the initial crosssection (x = 0) the pressure is constant and equal to p0 . The ﬂuid then ﬂows out of the pipeline (v = −v0 ) and the pressure falls to its initial value p0 . .
d0 2 ˙ x=−c p+ − p− = −ρ0 c · (v+ − v− ). We have d[p] dt =− aρ0 c a · (v0 − v− ) = − · [p] 2d0 2d0 (5. one obtains the ordinary differential equation for the variation of the pressure jump [p] at the wave front d[p] dt =− ρ0 c · (λ0 v0 2 − λ− v−2 ) 4d0 (5. Let the front of the hydraulic shock travel from the end of the pipeline crosssection x = L to its beginning. after which the process is periodically repeated. In the fourth phase of the process at 3L/c < t < 4L/c the ﬂuid again begins to ﬂow into the pipeline (v = v0 ) and the pressure becomes equal to p = p0 . otherwise at (p0 − ρ0 c × v0 ) > ps the ﬂuid does not vaporize and the pipeline section remains completely ﬁlled with ﬂuid.25) ˙ x=−c . the latter would begin to boil. the solution of the problem can be obtained without difﬁculty.and righthand sides. The velocity behind the shock front propagating in the negative direction vanishes. At λ0 v0 ≈ λ− v− = 2a > 0. and [p] = p+ − p− . At the instant of time t4 = 4L/c the situation returns to the initial one. Solution. ε) = λ− (v− ) is known. Problem 2. Subtracting termbyterm from the second equation from the ﬁrst one and taking into account the third equation. Then in accordance with Eq. On the damping of the pressure bow shock at the hydraulic shock wave front (Charniy. the solution has a particularly simple form (for laminar ﬂow this result gives an exact solution). If λ− (Re− . where a is a constant. It is required to determine the intensity of such a fall. d0 2 1 ρ0 c · v− v−  + ρ0 cg sin α. (5. Consider two negative slope characteristics parallel to the discontinuity along its left. In the presence of friction the hydraulic shock wave is gradually damped in the pipeline. we get d + (p − p− ) dt + − ˙ x=−c = 1 1 ρ0 c · · λ− · v− v−  − λ+ · v+ v+  2 d0 2 Taking v = v0 . 4. in particular the magnitude of the pressure jump at the shock front decreases.18) we have d + (p − ρ0 c · v+ ) dt d − (p − ρ0 c · v− ) dt ˙ x=−c = λ+ · = λ− · 1 ρ0 c · v+ v+  + ρ0 cg sin α. v = v+ − (p+ − p− )/ρ0 c = v0 − [p]/ρ0 c.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 133 pressure of the ﬂuid.24) ˙ x=−c This equation should be integrated under the initial condition [p] = −ρ0 c · v0 at t = 0.5. 1975).
2 2 c1 d2 + c2 d1 [ptrans ] = [pinc ] · 2 2c2 d1 . There are then generated two pressure waves: one is the reﬂected wave. which travels in the opposite direction with amplitude [preﬂ ]. From the obtained formula follows.134 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline and [p] = −ρ0 cv0 · exp(−a/2d0 · t). 5. which travels through the second pipe. From the solution of the problem it follows that at c1 d2 2 = c2 d1 or √ d2 = d1 · c2 /c1 .1 it was indicated that in nonstationary processes the inertial properties of ﬂuid in a pipeline are characterized by variable density. Use the results of the previous problem. ε) · − ρ0 g sin α(x) dt ∂x d 2 . as viewed from the ﬁrst pipe. It is required to express the amplitudes of the reﬂected and transmitted waves through the amplitude of the incident wave. the reﬂected wave is not generated at the junction of the pipes. the other is the transmitted wave with amplitude [ptrans ]. that is. the pressure shock wave (incident wave) with pressure amplitude of magnitude [pinc ]. It is required to prove that the amplitude of the hydraulic shock wave falling at the closed pipeline end will be doubled when reﬂecting from the pipeline end. in particular. Answer. Problem 4. 2 2 c1 d2 + c2 d1 2 Remark.27) dv ∂p 1 ρ0 vv αk ρ0 =− − λ(Re. that in pipelines of large diameter the value of the bow pressure jump at the hydraulic shock wave front decays more slowly than in pipelines of smaller diameter. Problem 3.8 Accounting for Virtual Mass In Section 5. [preﬂ ] = [pinc ] · 2 2 c1 d2 − c2 d1 . Hint. if the velocities of wave propagation in the ﬁrst and second pipes are c1 and c2 respectively. hence an additional quantity called virtual mass should be added.26) that is. [pref ] = 0.3. At the junction of two pipes with different internal diameters d1 and d2 comes. the pressure jump at the hydraulic shock wave front decays exponentially. (5. The main equations of nonstationary ﬂow of a slightly compressible ﬂuid in a pipeline with regard to virtual mass take the form ∂p + ρ c2 · ∂v = 0 0 ∂t ∂x (5.
27) transforms into ∂p + ρ c 2 · ∂v = 0 ∗ ∗ ∂t ∂x (5.28) dv ∂p 1 ρ0 vv ρ∗ =− − λ(Re.15 MPa or ≈ 11.30) If we take into account that for turbulent ﬂow αk ≈ 1.5 m s−1 . 5.9 is shown the development of the hydraulic shock in a pipeline section with diameter D = 1020 mm (δ = 10 mm) and length L = 100 km upon instantaneous closing of the gate valve at the righthand edge of the pipeline section.03. after closing the gate valve. At the initial instant of time t = 0 the ﬂow in the pipeline is stationary with velocity v(x. 0) = v0 ≈ 1.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 135 √ If we denote ρ∗ = αk ρ0 and –c∗ = –c/ αk .9a and 5. respectively. the system of equations (5.7 atm. t) = 0 at t > 0. The pipeline is transporting crude oil with density ρ0 = 880 kg m−3 and viscosity ν = 20 cSt. = Figures 5. ε) · − ρ0 g sin α(x) dt ∂x d 2 equivalent to the system of equations (5.5. Figures 5.3.5 ∼ 1. 5. From this follows that the Joukowski formula (5. . it will be √ reduced by a factor ακ D = c = c∗ = √ ακ ρ∗ d0 ρ∗ + K δ·E 1 (5.29) The velocity D of hydraulic shock wave propagation will also vary. The abrupt closing of the gate valve at the righthand edge of the pipeline is modeled by the condition v(L..20) for the amplitude [p] of the hydraulic shock varies.10b show the following stages of pressure wave upstream propagation. But for laminar ﬂow αk ≈ 1.4).33 and these corrections could be signiﬁcant. the velocity of the hydraulic shock wave propagation c is equal to 870 m s−1 .9 Hydraulic Shock in an Industrial Pipeline Caused by Instantaneous Closing of the Gate Valve In Fig. corrections to Joukovski formula will be small (one of them (5. The initial value of the pressure jump pf at the wave front is related to the initial velocity v0 by pf = ρ0 cv0 = 880 · 870 · 1.29) was ﬁrst obtained by Leibenson et al. it √ will be enhanced by the factor ακ √ c [p] = ρ∗ c∗ · [v] = αk ρ0 · √ · [v] = ρ0 c ακ · [v] αk (5. 1934).10a and 5.9b demonstrate the distributions of the head (z + p/ρg) at 2 and 40 s.
that is to HL (t) = zL + pL (t) ρ0 g Hf + [p] . i. ρ0 g where zL is the elevation of the section end and Hf the head at the pipeline .e. That is why the head. t) of the ﬂuid ﬂow behind the pressure wave front is small.136 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Figure 5. The velocity v(x. and consequently the pressure before the closed gate valve. are always raised.9 Pressure wave propagation: (a) 2 s and (b) 40 s after appearance of the wave. before the gate valve. can be taken as approximately equal to the head after the pressure wave front. The head HL (t) = zL + pL (t)/ρg at the end of the pipeline section. therefore head losses in this region are also small.
5.3 NonStationary Flow of a Slightly Compressible Fluid in a Pipeline 137 Figure 5.10 Pressure wave propagations: (a) 80 s and (b) 100 s after appearance of the wave. crosssection which the hydraulic wave shock has reached. From this it follows that for an approximate estimation of the pressure at the pipeline section end one can use the formula pL (t) or pL (t) ≈ ρ0 gi0 · ct + [p]. .31) ρ0 g(Hf − zL ) + [p] where i0 is the hydraulic gradient of the ﬂow in the undisturbed region and t is the time elapsed after closure of the gate valve. (5.
gates and slowly closing cocks. the systems of pressure wave smoothing are switched on when the rate of pressure buildup in the suction line of the pumping station is greater than the permissible magnitude. ε) − ρg sin α (5. All of these should ensure safe ﬂuid braking in the pipeline. ε) ρ ∂t ∂x d0 ∂x d0 2 .26)). are not equipped with cocks that close the pipeline crosssection too rapidly. as distinct from gaspipelines. oil products. Pumping stations are sometimes equipped with special devices intended to protect pipelines from hydraulic shock waves. 5. the equation of heat inﬂow should be invoked. Thus. In other words.4 NonIsothermal Gas Flow in GasPipelines Consider now nonstationary and nonisothermal gas ﬂow in gaspipelines. but are equipped with valves. Basic equations for the calculation of nonstationary nonisothermal ﬂows in a gaspipeline are represented by the system (5. to describe these ﬂows it is necessary not only to use the laws of mass and momentum conservation but also the law of energy transformation.32) ∂x d0 2 ∂t ∂v ∂ein ∂ein 4κ 1 ρv3 +v = − (T − Tex ) − p + λ(Re. therefore the pressure shock amplitude [p] reduces monotonically (see formula (5. as well as continuity and momentum equations. The main distinction of such ﬂows from the ﬂows considered above is that the gas represents a signiﬁcantly compressible medium with density dependent not only on pressure but also on temperature. ﬂow dampers of hydraulic shock – special safety valves or systems of pressure wave smoothing in case of a sudden pumping station switchoff when the pressure before the station begins to build up.8) giving ∂ρ + ∂ρv = 0 ∂t ∂x ∂ρv 1 ρv · v ∂ + (p + ρv2 ) = −λ(Re. for example. water and other) is manifested by the fact that such pipelines. The safety valves open the ﬂuid discharge when the pressure exceeds a certain value. Protection of Pipelines from Hydraulic Shocks The necessity to prevent the destructive force of hydraulic shock in pipelines transporting heavy dropping liquids (oil. In the suction lines of pumping stations are. These devices operate on the principle of emergency discharge of part of the ﬂuid from the pipeline into a special reservoir to decrease the magnitude of the pressure and its rate of increase.138 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline It should be noted that owing to viscous friction the ﬂuid behind the wave front would not come to a halt immediately but gradually.
32) represents partial differential equations for three unknown functions p(x. ∂x and ∂T . (5. t) on which the values of the functions p(x. t). v.32) give certain differential relations between unknown functions. ∂v . t) of the plane (x. therefore the question arises: how to solve them? To answer this question it is necessary to examine the structure of the system (5.4 NonIsothermal Gas Flow in GasPipelines 139 in which ρ = p/ZRT is the equation of the gas state and λ(Re. Through each point M(x. t). t). lines x = x(t) such that along them Eqs. v) along the line x = c (or x = ct + const.33) = x(t) = x(t) where dx/dt is the slope of this line (we shall call it a characteristic) to the taxis. t) dependent on x and t. We look for characteristics on the plane (x.e.32) treating it as well as was done when investigating equations (5.). (5. Since the lefthand sides of Eqs. Then it is possible to write three equations ∂p ∂v relating the derivatives ∂p . T with respect to time and coordinate. Let x = x(t) be a line on the plane (x. If . T(x.33) are known as well as the slope dx/dt of the line x(t).32) is hyperbolic (the number of its characteristics is equal to the number of equations. ∂T of these functions along the ∂t ∂t ∂t ∂x line x = x(t). T along x = x(t) dp dt dv dt dT dt = x(t) ∂p ∂p dx + · ∂t ∂x dt ∂v ∂v dx + · ∂t ∂x dt ∂T ∂T dx + · ∂t ∂x dt (5. that is three). these equations could be considered as three linear equations to determine six partial derivatives p. dt The presence of two characteristics is the distinctive property of the considered system of equations and allows us to assign the system to a class of hyperbolic differential equations and give a constructive method of its solution. i.). dt d ˙ (p − ρ0 cv) = I2 (p.4) transforms into an ordinary differential equation providing a relation between unknown functions. or strictly speaking d ˙ (p + ρ0 cv) = I1 (p. ε) is the hydraulic resistance factor.5. ∂x . Let us show that the system of equations (5. The slope can be taken as given since the function x = x(t) is known. These nonlinear equations have a complicated structure.4) in the model of nonstationary ﬂow of a slightly compressible ﬂuid. v) along the line x = c (or x = −ct + const. t) there are certain lines (characteristics) along which the system of partial differential equations (5. In the course of examining this model we have seen that on the plane of variables (x. t) are known. v(x. v. The system of equations (5. t) and T(x. To do this let us differentiate p. t) go just two (by the number of equations) such lines. t). v(x.
J5 = (dv/dt)x(t) .140 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline to these equations we add three equations of the system (5. J6 = (dT/dt)x(t) . v and T are taken along the line x(t). third and ﬁfth columns of this determinant by dx/dt and then subtract the resulting products from the . all six partial derivatives as well as the functions p. v. let us multiply the ﬁrst. on which their values are known. the dependence between the values of p. If this determinant vanishes and the system of linear equations is compatible. The total derivatives of the functions p. T on the curve x(t) exists. T along the curve x(t) can be determined uniquely and independently from each other. we get six linear equations for six partial derivatives ∂p ∂p ∂ρ ∂v ∂ρ ∂T ∂ρ ∂T ∂ρ +v +ρ + +v =0 ∂p ∂p T ∂x ∂x ∂T p ∂t ∂T p ∂x T ∂t ∂v ∂v ∂ρ ∂p ∂T ∂ρ ∂p v ∂ρ + 1 + v2 + ρ + 2ρv +v ∂p T ∂t ∂p T ∂x ∂t ∂x ∂T p ∂t ∂T ∂ρ = J2 + v2 ∂T p ∂x ∂v ∂T ∂T + ρCv + ρvCv = J3 p ∂x ∂t ∂x ∂p dx ∂p ∂t + dt · ∂x = J4 ∂v dx ∂v ∂t + dt · ∂x = J5 ∂T dx ∂T + · = J6 ∂t dt ∂x where J2 = −λρv · v/2d0 − ρg sin α. Let us equate the principal determinant of the system of six linear equations to zero ∂ρ ∂p T ∂ρ v ∂p T 0 1 0 0 v ∂ρ ∂p 2 0 T ρ 2ρv p 0 dx/dt 0 1+v ∂ρ ∂p ρ T 0 dx/dt 0 0 0 0 1 0 ∂ρ ∂T ∂ρ v ∂T ρCv 0 0 1 p p ∂ρ ∂T ∂ρ v2 ∂T ρvCv 0 0 dx/dt v p p =0 and calculate it. J3 = −4κ(T − Tex )/d0 + λρv3 /2d0 . v. To do this. J4 = (dp/dt)x(t) .32). which reduces also to linear equations with respect to the same derivatives. If the principal determinant of the system is nonvanishing.
−v=± dt Cv ∂ein p and − 2 ∂T ρ p ρ2 ∂ρ ∂T = p ∂p ∂ρ p ρ = p .34) with respect to the difference (dx/dt − v). dx dx −v=0⇒ = v. R = 450 J kg−1 K−1 . For example. For a perfect gas (∂p/∂ρ)T = RT. T) = Cp /Cv is the adiabatic index. at γ = 1. T Since Cv = ∂ρ ∂T ∂ ∂T = p ∂ ∂T p ρ . As a result we get the cubic equation ρCv dx −v dt 3 ∂ρ ∂p + T p ρ ∂ρ ∂T − ρCv p dx −v =0 dt (5.. dt dt Cv − p/ρ2 · (∂ρ/∂T)p dx 2. T = 273 K the velocity of sound is √ equal to c = 1. γ = const. The expression under the square root is simpliﬁed to Cp dx −v=± dt Cv where c2 = Cp · Cv ∂p ∂ρ =γ· T ∂p ∂ρ = ±c T ∂p ∂ρ T and γ(p. c = γRT.31. Roots of this equation are evident 1.31 · 450 · 273 ∼ 400 m s−1 .4 NonIsothermal Gas Flow in GasPipelines 141 second. fourth and sixth columns of the same determinant ∂ρ ∂p v 0 1 0 0 ∂ρ ∂p ˙ (v − x) T ∂ρ ∂p 0 T ρ ˙ ρ(2v − x) v p 0 0 0 ∂ρ ∂T ∂ρ ∂T p ∂ρ ∂T ∂ρ ∂T 0 0 0 ˙ (v − x) p ˙ 1 + v(v − x) T ∂ρ ∂p ρ T ˙ v(v − x) p p 0 0 0 0 0 0 1 0 ρCv 0 0 1 ˙ ρCv (v − x) The obtained determinant can be calculated by the method of determinant decomposition in terms of the thirdcolumn elements. = .5. then p Cv − ein + ∂J ∂T = Cp p where J(p. The quantity c having the dimension of velocity is called the adiabatic velocity of sound in gas. T) is enthalpy.
T to be obeyed at these characteristics: dp (γ − 1) dv · J3 (5.36) + ρc = +c · J2 + dt dt p/ρ · (∂ρ/∂p)T at dx/dt − v = +c or dx/dt = v + c. In such cases the difference (dx/dt − c) cannot be ignored. 1. for example in propulsive nozzles or for gas ﬂowing from openings. . Since the gas velocity v in gaspipelines as a rule does not exceed 10 m s−1 and c ≈ 400 m s−1 . Characteristic form of equations. then sometimes we could take dx/dt ∼ ±c.2 = v ± c and the third dx/dt = v. J3 = −4κ(T − T )/d0 + λρv3 /2d0 and derivatives with respect to time are taken in the direction of the characteristic dx/dt = v + c. Therefore the assumption dx/dt ∼ ±c means that this velocity of gas moving in a pipe is = approximately equal to the sound velocity in quiescent gas. (5. This condition as applied to our case may be called the compatibility condition at characteristics. where J2 = −λρv · v/2d0 − ρg sin α.34) provides three families of characteristics. should also vanish. we give the ﬁnal result dx −v dt − 2 ∂ρ ∂p T J3 − Cv J 6 − ρ ∂ρ ∂p dx p −v 2 dt ρ ∂ρ ∂p (ρJ5 − J2 ) T J3 p − Cp J 6 + 2 ρ ρ J4 = 0.142 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline The equation (5. two of them having velocities (slopes) (dx/dt)1. We shall see = below that the slope of characteristics to the taxis is simply the velocity of propagation of small perturbations in gas (sound velocity). especially for gas ﬂowing with high velocity. In the theory of linear equations the condition of the latter determinant vanishing is called the compatibility condition. Since the determinant of the system of the six linear equations under ˙ ˙ consideration vanishes at x = v and x = v ± c.35) Substitution successively in Eq. oriﬁces. v. for compatibility of this system the determinant ∂ρ ∂ρ ∂ρ v 0 ρ 0 ∂p T ∂p T ∂T p v 0 1 0 0 ∂ρ ∂p 1 + v2 T ∂ρ ∂p ρ T 2ρv p 0 ˙ x 0 v ∂ρ ∂T J2 p 0 ˙ x 0 0 0 0 1 0 ρCv 0 0 1 J3 J4 J5 J6 obtained from the principle determinant of the system by replacement of the last column on free terms of the system of equations.35) of the values of dx/dt for each of the characteristic families yields conditions for the functions p. nozzles and so on. T (5. Omitting cumbersome calculation of this determinant. In the general case it is of course not so.
The method by which this solution is obtained is called the method of characteristics. v(xk . appear three equations to determine three unknown quantities p(xk . 3. tm ) of the plane (x.36)–(5. Basic equations for this method are Eqs. tm ) and T(xk . that is at t < tm . tm ). where vmax and cmax are the maximum possible values of the gas and sound velocities respectively (cmax = γRTmax . Method of Characteristics Formulas (5. . The time step t is chosen so that t = x/(v + c)max . Divide this region by a rectangular mesh with straight lines xe = x · (k − 1). p ∂ρ J3 dT dp − 2 = (5.36)–(5.33) in the region 0 < x < L. Thus. (5. For numerical realization of the method of characteristics various schemes may be used. 2. there are just three characteristics dx/dt = v ± c and dx/dt = v. m = 1. The idea of this method is that at each point M(xk . Neglecting the work of the gravity force. Consider one of them. .39) are represented by differential operators of the form .5. the basic system of differential equations (5. Tmax is the maximum possible value of the gas temperature). Let it be required to get the solution of the system of equations (5.. (5.41). (1. t). at the point M(xk . Each of these compatibility conditions represents an ordinary differential equation which could be integrated over the direction of the respective characteristic.39) ∂x ∂t ∂ 4κ ∂ p v2 v2 ρ ρv · = − (T − Tex ) + ein + + ein + ∂t 2 ∂x 2 ρ d0 The lefthand sides of Eqs. at which should be satisﬁed the compatibility conditions (5. tm = t · (m − 1). being the intersection point of these characteristics. . ˆ where 1 ≤ κ ≤ N + 1.38) 3. where derivatives with respect to time are taken in the direction of the characteristic dx/dt = v.4 NonIsothermal Gas Flow in GasPipelines 143 dv dp (γ − 1) − ρc = −c · J2 + · J3 (5. N = [L/ x].38). t > 0 of the plane (x.37) dt dt p/ρ · (∂ρ/∂p)T at dx/dt − v = −c or dx/dt = v − c. 2. Cp dt ρ ∂p T dt ρ at dx/dt − v = 0 or dx/dt = v.36)–(5. where derivatives with respect to time are taken in the direction of the characteristic dx/dt = v − c.33) may be written in the socalled divergent form (see Eq. tm ). t) relevant to the time tm ‘‘in the past’’.38). tm ).38) give a constructive way to solve the system of equations (5.36) at αk ∼ 1) = ∂ρ + ∂ρv = 0 ∂x ∂t ∂ρv ∂ + (p + ρv2 ) = −λρv · v/2d0 (5.
m−1 vk+1/2.m v2 + Cv T 2 t − (λρv · v/2d0 )k+1/2.11 Mesh cell in the plane of variables (x.m−1 − (p + ρv2 )k+1 − (p + ρv2 )k · ρk+1/2.m = ρk+1/2. 5. whereas the quantities with integer subscripts denote mean values of parameters at the vertical sides (AB and CD) of the same mesh cell.m−1 − [(ρv)k+1 − (ρv)k ] · t . Application of this transformation to each equation of the system (5.m k+1/2. Here the quantities with fraction subscripts denote mean values of the corresponding parameters at the horizontal sides (BC and AD) of the mesh cell. t).m t.m = ρk+1/2.m − Ak+1/2.39) over the area of the mesh cell ABCD (xk ≤ x ≤ xk+1 . tm−1 ≤ t ≤ tm ) with sides x and t (Fig.11) and transform the integrals over the mesh area into integrals over the mesh contour using the formula div{A. Let us ﬁrst integrate the system of equations (5. B} = ∂A ∂B + dxdt = ∂t ∂x [A cos(nt) + B cos(nx)]dσ ABCD ABCD = ABCD (Adx + Bdt) = (Ak+1/2. .m−1 − ∂v · v ρ + Cv T + 2 ∂ k+1 Figure 5.m vk+1/2. x ρk+1/2.39) yields the following system of ﬁnite difference equations ρk+1/2.m−1 ) x + (Bk+1 − Bk ) t.m−1 · x = ρk+1/2. B} in the space of variables (x. t). v2 + Cv T 2 2 k+1/2.144 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline ∂A ∂B + ∂t ∂x expressing the divergence of a vector with coordinates {A.
= c · J2 + p/ρ · (∂ρ/∂p)T k−1/2.m − vk+1/2.m−1 pk.40) reﬂects the fact that the gas mass ρk+1/2.m−1 Cp Tk. the ﬁrst relation (5.m−1 t .m − pk−1/2. k−1/2.m−1 pk.5.40) are not closed since they include unknown quantities denoted by integer subscripts and representing the transfer of mass.m−1 + (ρc)k−1/2.m−1 are the mean values of the same functions at the previous moment of time tm−1 = tm − t. For example.m are. ρk+1/2.m − pk−1/2.40) Here ρk+1/2.m − vk−1/2. Eqs. Nevertheless. vk+1/2. The two other relations are interpreted in the same manner with the only difference being that they deal with momentum and total energy.m−1 . = −c · J2 + p/ρ · (∂ρ/∂p)T k+1/2.m .m−1 · x in the same segment at the previous instant of time added to the mass difference of the gas [(ρv)k+1 − (ρv)k ] · t ﬂowing in time t from the kth segment into the (k + 1)th segment and from the (k − 1)th segment into the kth one. The physical meaning of the obtained relations is clear: each of these equations represents the integral balance of one or other parameter in the mesh cell. the mean values of the density.m−1 (5. velocity and temperature of the gas in segment (xk . (5.4 NonIsothermal Gas Flow in GasPipelines 145 − ∂v · × v2 ρ + Cv T + 2 ∂ k 4κ t − (T − Tex )k+1/2. vk+1/2.m−1 t t (γ − 1) · J3 .m−1 p − 2 t ρ = J3 ρ . respectively.m and Tk+1/2. Let us represent the conditions (5.m−1 pk. xk+1 ) of the gaspipeline at the instant of time t = tm .37) in the form of ﬁnite difference equations vk.m − pk+1/2. xk+1 ) of the pipeline at the instant of time tm is equal to the gas mass ρk+1/2. respectively.m−1 − (ρc)k+1/2.m−1 and Tk+1/2.41) ∂ρ ∂p · T k−1/2.m − Tk−1/2. The essence of the method is that the values of these quantities are found from compatible conditions (5. momentum and energy from one mesh cell into another.m−1 t t (γ − 1) · J3 .m · x in the segment (xk .m−1 · x d0 t.38) at characteristics.32)–(5.m−1 vk. (5.36)–(5.
then from the third equation we obtain Tk.m /RTk. therefore the second boundary condition at the crosssection x = 0 (k = 1) is formulated as a condition at the characteristic dx/dt = v = 0 . The lefthand end of the pipeline (x = 0) is closed.40).38) should be added initial and boundary conditions. The dynamics of gas outﬂow happen as a rule in two regimes. To Eqs. through which occurs gas outﬂow. The ﬁrst boundary condition is v1. Substituting ρk. The process of gas outﬂow is not isothermal. they could be taken in the form of two relations between the pressure.m (0. Only at the ﬁnal stage of the process is there a gradual restoration of temperature due to external heat inﬂow.8–1. The gas temperature.146 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline As a result we have three linear equations for three unknown quantities pk. reﬂecting conditions of compressor station operation and one condition at the right edge of the pipeline section. owing to adiabatic expansion and the JouleThomson effect falls signiﬁcantly at the break in the crosssection as well as far from it.m from the ﬁrst two equations.m and from the relation ρk. vk. whereas the righthand end (x = L) is suddenly opened and remains open to the atmosphere. Crosssection x = 0. Boundary conditions vary depending on the concrete problem. (5. vk. with a given pressure. with a complete break of a gaspipeline the gas temperature can be reduced by 80–100 K. we get average values of the gasdynamic parameters in the considered mesh cell at the instant of time tm .37) and (5. pk. Hence it follows that at v = 0 the left integration boundary coincides with one of the characteristics of the differential equations (5. After the pressure in the gaspipeline is lowered by a certain value (for natural gas by a factor of 1.9 greater than atmospheric pressure) the outﬂow regime becomes subsonic and the gas velocity gradually decreases from sound velocity to zero.m . Tk.36).m we calculate the gas density. ﬂow rate and temperature of the gas at one edge (left) of the pipeline section. 1.m . At ﬁrst at the crosssection. for example.m .m . tm ) = 0. For example. since the lefthand end of the pipeline is taken to be closed. Hence.m . the boundary conditions are as follows. 5. Tk. appears a critical regime with the velocity of gas outﬂow equal to the local velocity of sound (≈380–400 m s−1 ).5 Gas Outﬂow from a Pipeline in the Case of a Complete Break of the Pipeline Let us illustrate the use of the method of characteristics as applied to nonstationary processes in a gaspipeline with an example of the calculation of gas outﬂow from a pipeline in the case of its complete break. Numerical calculations are carried out on the basis of recurrent formulas (5.m .38).m into Eq. First we ﬁnd quantities pk. As an example. (5.m and vk.m = pk.
Crosssection x = L.m−1 The third boundary condition at x = L has a different form depending on whether the gas outﬂow is subsonic v < c(p.m−1 + (ρc)N+1/2. These characteristics are dx/dt = v + c and dx/dt = v with positive slope. Calculations show that the gas is signiﬁcantly cooled during outﬂow.12 shows graphs of the gas temperature distribution over a pipeline .5 Gas Outﬂow from a Pipeline in the Case of a Complete Break of the Pipeline 147 Cp T1.39).m − p1.m−1 p − 2 t ρ ∂ρ ∂p · T 1.39) come to this crosssection from the integration domain. Figure 5. = −c · J2 + p/ρ · (∂ρ/∂p)T 3/2.m−1 = t J3 ρ .m − vN+1/2. Hence. 1.m − vN+1. If the gas outﬂow happens with local sound velocity. therefore one more condition should be given.m−1 p − 2 t ρ · ∂ρ ∂p T N+1/2.m−1 In subsonic ﬂow v < c(p.m−1 v1.m−1 − (ρc)3/2. then dx/dt = v − c = 0 and the right boundary of the integration domain coincides with the characteristic of the system (5. Such a condition is pN+1.m−1 The compatibility condition at the characteristic dx/dt = v − c coming from the integration domain to the initial pipeline crosssection (in our case at the characteristic dx/dt = 0 − c = −c).m−1 = t J3 ρ .m − v3/2.m−1 p1.m−1 t t (γ − 1) · J3 .m − pN+1. = c · J2 + p/ρ · (∂ρ/∂p)T N+1/2.m−1 Cp TN+1. = −c · J2 + p/ρ · (∂ρ/∂p)T N+1.m−1 t t (γ − 1) · J3 .m−1 − (ρc)N+1. T) only two characteristics of the system (5.m−1 t t (γ − 1) · J3 . N+1/2.m−1 vN+1.m − TN+1/2.m = paTM meaning that the pressure at the opened pipeline end is equal to the external atmospheric pressure. T) or sonic v = c(p. the third boundary condition is nothing but a condition at this characteristic pN+1. the ﬁrst two boundary conditions at x = L (k = N + 1) are pN+1.m − T1.m−1 pN+1.5.39) come from the integration domain to the boundary points x = L.m − pN+1/2.m−1 2.m−1 vN+1. T).m − p3/2. Two characteristics of the system (5. Thus.m − pN+1/2. provides the third boundary condition p1.
it is strongly recommended to take the utmost care in repairreconditioning operations. Calculations show that the maximal velocity of air suction in the oscillation process exceeds 50 m s−1 . 6. section length (D = 1220 mm. t = 200 s.km Figure 5. Once the velocity of the gas becomes equal to zero it continues to decrease and at the section CD of the graph remains negative. Gas cooling in the pipeline happens because the abovementioned effects are too fast to be compensated by heat inﬂow from the surrounding medium. Since the mixture of natural gas and air could achieve an explosive concentration.13 represents the dependence of the gas velocity on time at the crosssection where the pipeline undergoes a break (D = 1220 mm.5 MPa. L = 1 km) with initial pressure p(x. t = 0. 5. testifying that atmospheric air is being sucking into the pipeline. The oscillation process is seen to appear in the ﬁnal stage of gas outﬂow and lasts for about 2 min.12 Distribution of gas temperature over a pipeline section length: 1. The same behavior of gas is observed in the following instants of time. 0) = 5. About after a 30th of a second from the outﬂow beginning in the pipeline an oscillation process occurs in which the gas periodically changes its direction of motion. t = 60 s. Computer modeling shows the effect of gas suction into the pipeline in the ﬁnal stage of the outﬂow process. t = 120 s. t = 5 s. . L = 5 km) at different instants of time. t = 420 s. It is seen that the initial temperature of the gas. This is explained by gas expansion due to the high outﬂow velocity and the Joule–Thomson effect. The section BF of the graph characterizes subsonic outﬂow of gas. The velocity of the outﬂow gradually decreases as a consequence of reduction in the pressure and temperature of the gas at the crosssection of the break. Figure 5. the discovered phenomena appears to be a serious hazard to attendants. 3. 2. equal to 0 ◦ C decreases by more than 100 K. The section AB of the graph characterizes sonic outﬂow of gas lasting about 20 s.148 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline L. In particular. 4.
5.13 Dependence of gas outﬂow velocity on time.6 Mathematical Model of NonStationary Gravity Fluid Flow Such a ﬂow has already been discussed in Section 3.5.7 but then a stationary gravity ﬂow was considered. ms1 149 t. t) the area of the pipeline crosssection and v(x. that is a ﬂow in which all the hydrodynamic parameters remain constant at each crosssection of the pipeline.14). Consider now a onedimensional mathematical model of nonstationary gravity ﬂuid ﬂow. In other words these parameters were independent of time. 5. In this model there are two parameters governing the ﬂow: S(x. + = −ρgS cos α · − ρgS sin α − ∂t ∂x ∂x CSh 2 Rh Figure 5.6 Mathematical Model of NonStationary Gravity Fluid Flow nL. .14 Gravity ﬂuid ﬂow. The differential equations of the model under consideration are ∂ρS ∂ρvS ∂t + ∂x = 0 ∂ρvS ∂ρv2 S ∂h ρgS cos α · vv. t) the ﬂow velocity (Fig.s Figure 5.
i. which are beyond the scope of this book. We advise those who are interested in these methods to consult the book by Rozhdestvenski and Yanenko (1977). due to excess of the ﬂuid level at one crosssection of the pipeline as compared to the ﬂuid level at another crosssection.150 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline There Rh (S) is the hydraulic radius of the ﬂow (see Section 3. If we take into account that ρ ∼ ρ0 = const.e. The second equation (momentum equation) is the law of momentum change. CSh is the Chezy factor. the term ρgS cos α · ∂x on the right is the Boussinesq force acting on the ﬂuid due to its free surface being nonparallel to the pipeline axis. The ﬁrst of these equations (continuity equation) expresses the law of mass conservation in the ﬂuid ﬂow in a pipeline with crosssection partially ﬁlled by ﬂuid.7).42) The system of equations (5. For gravity stationary ﬂuid ﬂow (∂/∂t = 0) from the ﬁrst equation of the system (5.. that is Newton’s Second law: on the left is the derivative of momentum with respect to time. the quantity ρv2 S = ρvS · v representing ∂h the ﬂux of momentum.42) belongs to the class of quasilinear (that is linear with respect to derivatives) differential equations of the hyperbolic type. The system of equations may be rewritten in an equivalent form if we take into account that ρg cos α · 0 S Sdh = ρg cos α · 0 S S dh(S) dS dS As a result we get ∂ρS ∂ρvS ∂t + ∂x = 0 S ∂ ∂ρvS + Sdh ρv2 S + ρg cos α · ∂t ∂x 0 = −ρgS sin α − ρgS cos α · vv CSh 2 Rh (5. (5. If we assume in addition that S = const. the term −ρgS sin α is the component of the gravity force and the term −ρgS cos α · vv/(CSh 2 Rh ) is the force of the ﬂow resistance due to ﬂuid friction against the pipeline walls.42) it follows that ρvS = const. = which is true everywhere except in small regions close to the transfer sections where gravity ﬂows are formed. which among other things contains voluminous literature devoted to this problem.42) it follows that 0 = −ρgS · sin α − ρgS cos α · vv ⇒ v = CSh Rh · tan α CSh 2 Rh (5.43) . from Eq. h(S) is the depth of the pipeline crosssection ﬁlling with ﬂuid. then v · S = const.. α is the angle of inclination of the pipeline axis to the horizontal. Solution of these equations can be obtained by specially elaborated methods.
(5. (5.44) following from Eq. Leibenson et al.44) vanishes is called the normal depth of gravity ﬂow in the pipe. which after some transformation allows one to get the degree of ﬁlling with ﬂuid of the gravity ﬂow section. In this case the relation (5. tan α + Q 2 /(CSh 2 Rh S2 ) dh =− dx 1 − (Q 2 dh/dS)/(g cos α · S3 ) (5. (5. Depending on the relation between the depths hn and hcr different ﬂow regimes are possible.15 Gravity section in pipeline. 1938). Fluid ﬂow with a normal depth of ﬂow happens under the condition hn = const.6 Mathematical Model of NonStationary Gravity Fluid Flow 151 psat Figure 5. Since crosssections after the transfer section are only partially ﬁlled by ﬂuid. 1934. . Investigations of these regimes are described in special monographs (Archangelskiy.43) holds. An ordinary differential equation serves to calculate the depth h(x) of ﬂuid in the pipe. In crosssections with such a depth the derivative dh/dx tends to ∞ and the ﬂuid ﬂow varies the level of pipe ﬁlling abruptly. The depth hcr . Figure 5..5. σ. Q is the ﬂuid ﬂow rate. It is seen from this ﬁgure that in this section the line of the hydraulic gradient goes parallel to the pipeline axis at a distance psat /ρg from it owing to the pressure constancy in the gas cavity over the ﬂuidfree surface.42) for the case of stationary ﬂow (∂/∂t = 0) with h(x) = h[S(x)].51)). 1947. The depth hn of the ﬂuid in the pipe at which the numerator of the fraction on the righthand side of Eq. the pressure in this section is constant and equal to the saturated vapor pressure psat of this ﬂuid. The point at which the ﬁrst gravity section in the pipeline begins is called the transfer section. Rh = Rh (S).44) vanishes is called the critical depth.15 illustrates the behavior of the hydraulic gradient in gravity ﬂow. (3. at which the nominator of the fraction on the righthand side of Eq. depending on the hydraulic gradient tan α (see Eq. Christianovitch.
These voids are capable of growing and turning into stationary gravity ﬂow sections or.1–5. This brings into existence temporal transfer points at some tops of the pipe proﬁle and cavities. even when the pressure drops to the value of the saturated vapor pressure. This wave having been reﬂected from the open surface of the reservoir or from a vaporgas cavity inside the pipeline initiates a rarefaction wave traveling in the opposite direction and reduces the pressure in the ﬂuid. One more example can be given: Connection of a lateral tap from the oilpipeline to an intermediate oil tank leads to the propagation of rarefaction waves up and down stream from the place of tap cutting up. Calculation in these cases on the basis of classical theory is also impossible. in laboratory installations it could be observed that the ﬂuid before the gate valve literally boiled owing to the sharp reduction in pressure. These waves are able to break the ﬂuid column at many crosssections of the pipeline proﬁle and turn the enforced ﬂow into a gravity one characterized by the presence of vaporgas cavities and gravity ﬂow sections.and downstream. So. on the contrary.3) contained an essential restriction on the absence of phase transition in the ﬂuid. Any sharp pressure reduction gives rise to a plethora of vaporgas cavities the disappearance of which in the pipe lead to powerful hydraulic shocks. We can give another example: closing the gate valve generates a compression wave propagating upstream with a rise in the pressure in the wave. However.7 NonStationary Fluid Flow with Flow Discontinuities in a Pipeline The preceding classical models of nonstationary ﬂows of a slightly compressible ﬂuid (see Sections 5. This leads sometimes to ﬂuid ﬂows with a partially ﬁlled pipeline crosssection. contracting and even disappearing altogether. at the disconnection of a pumping station or an aggregate of this station a rarefaction wave is propagated downstream in the pipeline. The pressure in such a wave reduces leading to the formation of voids at the tops of the pipeline proﬁle. among which are gascondensate and a wide fraction of light hydrocarbons with saturated vapor pressure from 3 to 30 atm. the vapor column breaks and the pipeline crosssection becomes partially ﬁlled with vapor. When the pressure in the rarefaction wave reduces to the saturated vapor pressure the ﬂuid boils. All the aforesaid is true also for pipelines transporting the socalled unstable ﬂuids. To perform calculation of such processes on the base of classical theory is of course impossible. Such cases also defy calculation in the framework of classical theory. If the pressure supply in the pipeline is not high. It was tacitly supposed that the ﬂuid under no circumstances passes into the vaporgas phase. the reduction in pressure leads to ﬂow discontinuity and the generation of vaporgas cavities.152 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline 5. For example. . in the propagation of a rarefaction wave in the pipeline this condition can be violated at many pipeline crosssections and ﬁrst and foremost at the tops of the pipeline proﬁle. From this point on all further results predicted by classical theory appear to be wrong. for example.
initially supported by this pressure. are taken out. These shocks were gradually damped with time and the ﬂow in the pipeline was stabilized. there is an abrupt stop of ﬂuid ﬂow and. located before a section of pipeline with signiﬁcant slope. Fig. The nature of the proﬁle hydraulic shock is as follows. . 5. In particular. in Lurie and Polyanskaya (2000). hydraulic shock (Fig. Similar phenomena were observed in the pipeline on disconnection of some pumps or the pumping station as a whole. each time the gate valve. When the gate valve (at x = 0. 5. the discovery and investigation of the origin of powerful hydraulic waves in the pipeline called proﬁle hydraulic shock are described. The ﬂuid in this inverse ﬂow is accelerated and when vaporgas voids.16 Initial stage of the process.19).7 NonStationary Fluid Flow with Flow Discontinuities in a Pipeline 153 Proﬁle Hydraulic Shock It should be noted that the appearance or disappearance of voids in the pipeline is unsafe and is rather dangerous for pipeline integrity. having originated before the gate valve.16) closes the pressure in the sloping section of pipeline falls and the ﬂuid column. was closed the rarefaction resulting in the region close to the valve led to a sequence of powerful hydraulic shocks. 5. For example. since Figure 5. Figure 5.17–5. begins little by little to slip down and acquire inverse ﬂow (Figs.17 Formation of gravity ﬂow in the upward sloping pipeline section.5. as a consequence. The power of the shock is then especially great.18).
are described by differential equations (5. heads back to the gate valve.19 Proﬁle hydraulic shock.4) for pressure p(x.18 Formation of reverse ﬂuid ﬂow in the upward sloping pipeline section. Then a secondary hydraulic shock occurs. propagates downstream. ε) − ρ0 g sin α(x) ∂t ∂x d0 2 . In accordance with the classical theory of nonstationary processes the wave processes generating in a completely ﬁlled pipeline at its startup or stopping. in the form of a rarefaction wave. 2000). Generalized theory. the wave is reﬂected from this void and now. t) ∂p + ρ c2 · ∂v = 0 0 ∂t ∂x ∂v ∂p 1 ρ0 vv ρ0 + = −λ(Re. Reaching a temporary transfer section. at which a vaporgas void has been formed. The hydraulic shock wave reﬂected from the gate valve and accompanied by ﬂuid stop. t) and velocity v(x. Figure 5.154 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Figure 5. opening or closing of the gate valve or lateral tap and so on. the inverse ﬂow of the ﬂuid runs against the closed gate valve and. after which the process is repeated again and again with decreasing intensity. as follows from the Joukovski formulas. (Lurie and Polyanskaya. the ﬂuid column in the ascending section of the pipeline again becomes weightless and begins to slip down to the closed gate valve. the amplitude of the hydraulic shock redoubles.
psat = 0.21 Formation of the secondary reverse ﬂow in the upward sloping pipeline section. as distinct from existing theory. Method of calculation. It is seen that. the line of hydraulic gradient at Figure 5.7 atm) from 1500 to 200 m3 h−1 . For numerical calculation of oil product ﬂow in completely ﬁlled as well as in partially ﬁlled pipeline sections there is an elaborate scheme of endtoend calculation based on the ideas of Godunov (Samarskiy. 1977).23 demonstrate the results of calculations on successive stages of unloading wave propagation in a 10km pipeline with internal diameter d = 516 mm at an abrupt drop in the pumping delivery of benzene (ρ = 750 kg m−3 .5. The ﬂuid ﬂow in the sections of enforced ﬂow is described by the system of equations (5. This scheme involves consideration of the socalled problems on the disintegration of arbitrary discontinuity in the system of hyperbolic equations. Results of calculation.42). (5. Figures 5.20 Decay of hydraulic shock wave. Figure 5. . In the generalized theory it is assumed that in the pipeline there are completely ﬁlled enforced (pumped) sections as well as only partially ﬁlled gravity sections in which the pressure is equal to the saturated vapor pressure psat .7 NonStationary Fluid Flow with Flow Discontinuities in a Pipeline 155 These equations represent the laws of mass conservation and variation of momentum of ﬂuid particles moving in a pipeline.4) whereas in the sections of gravity ﬂow it is described by Eqs.16–5.
156 5 Mathematical Models of 1D NonStationary Flows of Fluid and Gas in a Pipeline Figure 5. Since the ﬂuid ﬂows in the opposite direction (as if it would run on a closed gate valve).22). The amplitude of the hydraulic shock wave gradually decreases (Fig. 5. no instant of time intersects the pipeline proﬁle. in 24 s after the aggregate disconnection a powerful hydraulic shock is formed in the pipe (Fig. 5. in which the pressure is nearly twice the initial pressure at the station. the pressure in the pipeline would never be less than the ﬂuid saturated vapor pressure. 5. that is.23 Formation of new stationary ﬂow. The ﬁgures also show how the wave of pressure decrease runs over the pipeline proﬁle top (lower polygonal line of Fig. 5.18).21).17 and Fig.19). In the given calculation there were six such hydraulic shocks.20) and in a further 10 s in the upward sloping section of the pipeline slope the inverse ﬂow appears again (Fig. a temporary transfer point and further gravity ﬂow section with inverse ﬂow of ﬂuid is being formed (Fig. at this top. 5. 5.23). and how. . Figure 5. After 46 s the secondary hydraulic shock occurs (Fig. but already with lesser amplitude. 5.22 The second top of the proﬁle hydraulic shock. 5. Only 4 min after the pumping regime change the stationary regime in the pipeline is achieved (Fig.16). where there exists a single gravity ﬂow section of length 500 000 m in the pipeline downward sloping section.
millions of pascals (megapascals. decimeter. whose numerical values depend on the choice of measurement units.28 or 0. Michael V.157 6 Dimensional Theory Dimensional theory contains fundamental propositions for representing equations of mathematical regularities modeling different phenomena in invariant form. pipeline diameter can be expressed through the numbers 1. Two different researchers Modeling of Oil Product and Gas Pipeline Transportation. KGaA.448 N. Weinheim ISBN: 9783527408337 . Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. 3.21 and so on. 1000. gallon (1 gallon in USA ∼ 3. depending on what is used as the measurement unit of the length: meter.6093 km). millimeter.0433 depending on the measurement units: technical atmosphere (kgf s−1 m−2 ). For example. When setting out the fundamentals of dimensional and similarity theory and the modeling of phenomena we have followed the methodology of Sedov (1965). Such representations permit one to compile classes of similar phenomena and to model them on experimental installations. arbitrary.3048 m. foot or inch (1 foot ∼ 0.28. 100.6 g. is connected with measurements of the characteristics of these phenomena. MPa) or psi (pound per square inch: 1 force pound ∼ 4. For example. liter and second. 1 inch ∼ 0.8 or 73.0703 atm). in a form independent of the choice of the units of the measurements. = = = It is important to note that the choice of measurement units depends on the researcher and therefore is. kilometer or mile (1 mile ∼ 1. The pressure can = also be measured with different numbers 64. 6.006896 MPa ∼ 0.4 and so on depending on that what unit of measurement is taken: meter. that is. to a great extent.1 Dimensional and Dimensionless Quantities A quantitative description of various physical phenomena. 6. the length of a pipeline = = can be expressed through the numbers 150 000. the volumetric ﬂuid ﬂow rate in a pipeline may be measured by the numbers 1000. 277.785 liter) and second. centimeter. = 1 mass pound ∼ 453. 1 psi ∼ 0. 150 or 93. including the transport of ﬂuids and gases in pipelines. The same is true for many other = physical quantities. 39. 10.4 depending on whether the units of volume and time are taken as cubic meter and hour.0254 m).
At ﬁrst the meter was deﬁned as 10−7 part of a quarter of the Earth’s meridian. are called primary or basic units.73 lengths of the wave emitted by the krypton (Kr) atom in vacuum at its energylevel transition. volt (V) – a unit of electrical stress . quantities whose numerical values do not depend on the choice of measurement units are called dimensionless quantities. In the international SI system they are deﬁned as follows. The prototype of the kilogram is made from platinum–iridium alloy in the form of a cylindrical weight. 6. This means that the numerical values of L/d. 1 meter is a length equal to 1650763. The relative error of standard copies compared with the original does not exceed 2 · 10−9 . velocity and viscosity units. The standard kilogram is equal to the mass of the international prototype kept at the International Bureau of Weights and Measures. d is the internal diameter of the pipeline. There are also other basic units of measurement. time and mass. Hence. such as coulomb (C) – a unit of electricity quantity (electrical charge). There are recognized the atomic (standard) second reproduced by cesium standards of frequency and time and the ephemeris second equal to 1/31556925. The international standard of the meter before 1960 was a bar of platinum–iridium alloy marked on one side of its planes. Second is the measurement unit of time.9747 part of the tropical year. in particular. that is. This bar is kept on deposit at the International Bureau of Weights and Measures in Sevre near Paris. it is possible to give the following deﬁnition: quantities whose numerical values depend on the choice of measurement units are called dimensional quantities. Among these are. For example. In accordance with the deﬁnition taken at the 11th General Conference on Weights and Measures (1960). the ratio of the pressure at a pumping station discharge line to the pressure at a pumping station suction line pdl /psl or a more complex combination such as the Reynolds number Re = vd/ν (v is the average ﬂuid velocity in the pipeline. do not depend on measurement units. These parameters are said to be invariant relative to the choice of measurement units. the ratio of the pipeline length to the pipeline diameter L/d. having been introduced empirically by arbitrary conditions and propositions.2 Primary (Basic) and Secondary (Derived) Measurement Units Measurement units. Kilogram is a unit of mass measurement. ν is the ﬂuid kinematic viscosity) are independent of the choice of measurement units. And yet there exist parameters whose numerical values do not depend on the researcher. Meter is a unit of length measurement. the units of length. pdl /psl and v · d/ν would be unchanged for any choice of length.158 6 Dimensional Theory describing one and the same phenomenon but at the same time using different measurement units can obtain diverse numerical values of one and the same parameter. pressure.
mile h−2 and so on. Dimensional Formula Let there be a physical parameter A. Fluid ﬂow rate (e. Among them are the following.6. Acceleration which is deﬁned as the ratio between the velocity increment and time. therefore its units can be kg m−3 . 6. T (A = F) – force: [F] = M · [a] = M · L · T−2 .g. Pressure which is deﬁned as the ratio of force to unit area. Denote through L the measurement unit of length. thus force units can be: pascal (1 Pa = 1 N m−2 = 1 kg m−1 s−2 ). through T the measurement unit of time and through M the measurement unit of mass. volumetric) which is deﬁned as the ﬂuid volume crossing the pipeline crosssection area in a unit time.3 Dimensionality of Quantities. mile h−1 and so on. foot s−2 . Then the expression of the measurement units of many other quantities can be written as the following formulas: L = L · T−1 = M0 · L · T−1 . therefore velocity units can be m s−1 . Velocity which is deﬁned as the ratio of length to time. therefore its units can be m3 s−1 . L (A = v) – velocity: [v] = . Dimensional Formula 159 (voltage. degrees ( ◦ C. Speciﬁc weight which is deﬁned as the weight of the medium unit volume. km s−2 . t m−3 and so on. All these and analogous measurement units are derived from the basic units. dyn cm−3 and so on. m3 h−1 . K and other) – a unit of temperature and so on.448 kg m s−2 ) and so on. Such units are called secondary or derived units. Measurement units of other quantities are by deﬁnition introduced through the basic measurement units. Force (weight) which is deﬁned as the product of mass and acceleration. g cm−3 . T L (A = a) – acceleration: [a] = 2 = L · T−2 = M0 · L · T−2 .3 Dimensionality of Quantities. hence the unit of measurement can be the ampere (1 A = 1 C s−1 ). thus force units can be dyne (1 dyn = 1 g cm s−2 ). newton (1 N = 1 kg m s−2 ). l s−1 and so on. Current intensity which is deﬁned as a charge transmitted in a unit time. hence its measurement units can be N m−3 . Density which is deﬁned as the mass of the medium unit volume. hence acceleration units can be m s−2 . m3 min−1 . M (A = ρ) – density: [ρ] = 3 = M · L−3 · T0 . Expression of its measurement unit through basic units is called the dimension of a given parameter and is denoted by the symbol [A]. potential). km h−1 . ◦ F. This expression written as a formula is called the dimensional formula. pound (∼ =4. pound inch−2 and so on.
The velocity v in the SI system of basic measurement units (m.37−1 · 1−2 = 0. pound) the transition from this system to the SI system is carried out by changing the length scales 100/2. that is 6 000 000 kg m−1 s−2 . Example 2. m2 . In Eq. 1 2 3 Let us explain the afore said by examples. the value of the velocity will also be enhanced 100 times (6.54 = 39. it is easy to verify that even if the number of physical parameters . kg) is 1 m s−1 . s. of the length unit k2 fold. s.1). (6.056 and will be 0.056 · 6 000 000 = 336 043 (psi). L3 M · L · T−2 = M · L−1 · T−2 . s.205 times. Example 1. if the transition from the new system of basic measurement units to the old one is accomplished by variation of the mass unit k1 fold. kg) the transition from this system to the SI system is performed by increasing one of the basic units (length) by a factor of 100.37 times and the mass by 1/0. the numerical value of the parameter A would vary km1 · km2 · km3 fold. obtained empirically. For example.4536 = 2.2) v = 10 · 1001 · 1−1 = 100 · v = 100 cm s−1 . (A = Q) – volumetric ﬂow rate: [Q] = T M = M1 · L0 · T−1 . For a new system of basic measurement units (cm. particular examples show that in all cases the dimensional formula of parameter A has form of a power monomial [A] = Mm1 · Lm2 · Tm3 (6. there are only three basic units of measurement because we have considered examples solely from mechanics. hence the value of the pressure will vary by a factor 2. What does the dimensional formula mean? It allows one to determine very simply by how manyfold the numerical value of parameter A would be changed. m3 are certain real positive or negative numbers.160 6 Dimensional Theory M · L · T−2 = M · L−2 · T−2 . For a new system of units (inch. (A = p) – pressure: [p] = [F]/L2 = L2 L3 = M0 · L3 · T−1 . 1 2 3 ` that is the new value A of this parameter would be determined by the formula ` ` A = km1 · km2 · km3 · A. on going from one system of basic measurement units to another one differing from the ﬁrst system only by the scales of the basic units. However.2051 · 39.1) where m1 . s. The pressure p in the SI system of basic units (m. kg) is 6 MPa. (A = G) – mass ﬂow rate: [G] = T (A = B) – dimensionless parameter [B] = M0 · L0 · T0 . of the time unit k3 fold. (A = γ) – speciﬁc weight [γ] = Thus.
of a2 by k2 times. n 1 2 Let us use this circumstance. . . namely B: {a1 . . k2 = α2 . [a2 ]m2 .. .. that the dimensional formula in the general case would have the form of a power monomial [A] = [a1 ]m1 .6. .. an = β n · an ...3) where A is a physical parameter. Only the number of factors will increase... heat... a2 .4 Proof of Dimensional Formula 161 is increased by other parameters from electrical. whose dimension is derived from the dimensions of basic quantities denoted by a1 . 6.. C and D measuring one and the same physical parameter A but using different systems of basic measurement units. · kmn · A. a2 . the numerical value of the parameter A would be changed by km1 · km2 · . .. chemical and other phenomena and the number of basic units grows and exceeds three.. . . .. .3) in the general case. . . .. kn = αn . . Let there be three researchers B. . an }. . the dimensional formula would not be radically changed. a2 . C: a1 . · kmn times. a2 .. [an ]mn (6. Hence.. of an by kn times.. that is the n 1 2 new value of this parameter A would be equal to A = km1 · km2 · . . . . [a3 ]m3 .. . D: a1 . This formula means that if we change the scale of a basic measurement unit a1 by k1 times.. . an . ... .4 Proof of Dimensional Formula Let us prove the validity of the dimensional formula (6. . . . so that its basic units are related to the basic units of researcher C by the formulas a1 = β 1 · a1 a2 = β 2 · a2 .. . . . . an . it could be assumed.... so that its basic units are related to the basic units of researcher B by the formulas a1 = α 1 · a1 a2 = α 2 · a2 . an = α n · an where k1 = α1 . . an . .
αn ) · A. . . βn . αn βn ) = F(α1 . αn βn ) · A (6. . αn ) · A. . . where i may be equal to 1. k2 . . α2 . . we get ` A = F(α1 β1 . . . . bypassing researcher C. . βn ) · A or ` A = F(β1 . k2 = β2 α2 . α2 .. . . . . . Let a function F(k1 . Find the solution of this equation. . . we obtain A = F(α1 . . β2 ..5) The result should of course be independent of the transition route from one researcher to another one. . .. Then if we go from the units of researcher B to the units of researcher C. . . β2 . and to the basic units of researcher B by the formulas a1 = β 1 α 1 · a1 a2 = β 2 α 2 · a2 . . . . . . . . .. k2 . α2 .. thus it must identically satisfy the following functional equation F(α1 β1 . (6.6) which ought to be valid for any values of the factors α1 . . . k2 = β2 . .6) with respect to βi . . . Differentiation of both parts of the identity (6. . respectively. . . where ξi = αi βi . . . kn = βn . . . . α2 β2 . kn times from the ﬁrst one in the scales of the basic units. . . ∂ξi ∂βi . . A measured by C ` and A measured by D. It is reasonable to suppose that the values of the parameter A measured by ` the three researchers would be different: A measured by B. 2. yields αi ∂F ∂F = · F(α1 .162 6 Dimensional Theory where k1 = β1 . βn ) · F(α1 . . . .. . If we then go from the units of researcher C to the units of researcher D. . . . βn ) (6. . an = β n α n · an where k1 = β1 α1 . β2 . kn ) show how manyfold the numerical value of parameter A in one of the system of basic units would be changed when passing to another system differing by k1 . if we go at once from the units of researcher B to the units of the researcher D.. . .. ... . n. αn and β1 . . we get ` A = F(β1 . α2 . 3.. kn = βn αn . .4) On the other hand. . αn ) · F(β1 . β2 . α2 β2 . α2 . . αn ). .
·. K0 = 1 and the function will be F = αm1 · αm2 · . an ). . β2 = 1. one gets F = km1 · km2 · .3). αn ) ∂αi (6. a2 . k2 .8) Hence. . . β2 = 1. . 1. [a2 ]m2 . . As a result the following differential equation is obtained αi · ∂F(α1 . . Thus.5 Central Theorem of Dimensional Theory It is appropriate now to interrupt the description of the theory and to formulate a question concerning an apparent contradiction due to the use of dimensional quantities. because the value of the parameter A is not changed by variation of the basic measurement unit scales.6. βn = 1. [an ]mn which proves the dimensional formula (6. [a3 ]m3 . . . . The solution of differential equation (6. . . . . · αmn n 1 2 or redesignating the variables as k1 . . kn . . If F(1.7) gives the dependence of the sought function F on parameter αi . . . kmn . . . . . . the parameter A has the dimension [A] = [a1 ]m1 . . . Let the mathematical form of a certain physical phenomenon be expressed by the dependence of a parameter A on other parameters a1 . . . Since αi was any one of the arguments of the function F. . βn = 1.5 Central Theorem of Dimensional Theory 163 Since this equality also represents an identity. .. . . . n 1 2 (6. the latter should have the following form F = K0 · αm1 · αm2 · . . . a2 . . . . when changing the basic unity a2 by k2 times the numerical value of the parameter A would vary by k2 m2 and so on. then β1 = 1. the numerical value of the parameter A would vary by k1 m1 times. . . .9) . α2 . . (6. when changing the basic unity a1 by k1 times. αn ) = mi · F(α1 . . · αmn n 1 2 where K0 = const. an governing this phenomenon as follows A = f (a1 . α2 . 1) = 1. the separation of variables provides mi ∂ (ln F) = ⇒ ln F = mi · ln αi + c ⇒ F = K · αm1 i ∂αi αi where the integration constant K (K = eC ) is a function of the remaining parameters αj .7) in which mi = ∂F/∂βi at β1 = 1. 6. Really. .
. length l and mass m are dimensionallyindependent of each other.e. 6. If such numbers do not exist. On the other hand all quantities entering into the dependence (6. . a3 = δ. . the wall thickness δ and the elastic modulus (Young’s modulus) E of the steel from which the pipeline is made. namely in the form of a dependence between dimensional complexes composed from governing parameters’’. and its arguments (a1 . an ). The answer to this question gives the central theorem of dimensional theory called the Buckingham theorem. m2 .164 6 Dimensional Theory For example. so that S = f (P. an . [an ]mn (6. . . a2 . Now we will prove this theorem. This theorem states: ‘‘Each dependence between dimensional quantities reﬂecting an objectively existing physical regularity could be rewritten in invariant form independent of the choice of measurement units. Parameters with dimensions of velocity v and density ρ are dimensionallyindependent. a2 . . . . It is evident that the dependence under consideration exists objectively and should not depend either on the researcher performing the investigation or on the choice of measurement units being used to calculate the values of the function. . D. . removing the contradiction between the objective character of any physical regularity and the subjective character of the choice of measurement units. . a1 = P. It is easy to verify that . .9) are dimensional quantities whose numerical values depend on the choice of measurement units and consequently on the researcher. .9) reﬂects the objectively existing physical regularity when numerical values of the function and its arguments depend on the researcher. [an ] by the formula [a] = [a1 ]m1 [a2 ]m2 . .10) namely there exist such numbers m1 . the enhancement of pipeline crosssection area S when producing a positive pressure P depends on the magnitude of this pressure as well as on the pipeline diameter D. E). Hence we have A = S. a parameter with the dimension of pressure p is dimensionallydependent on parameters with dimensions of density ρ and velocity v. . On the other hand. . a2 = D. The question arises as to at which point the dependence (6. . mn that the equality (6. .6 DimensionallyDependent and DimensionallyIndependent Quantities It is said that a quantity ‘‘a’’ is dimensionallydependent on the quantities a1 . . a4 = E. However. a2 . . . . it is said that the quantity a is dimensionallydependent on the quantities a1 . an when its dimension [a] is expressed through the dimensions [a1 ]. i. before we do this. parameter A. It is evident that parameters with dimensions of time t. we need ﬁrst to deﬁne dimensionallydependent and dimensionallyindependent quantities. [a2 ].10) is obeyed. δ.
and righthand sides of the last relation. −2 = −m1 − 2m2 to determine the three unknown quantities m1 . Insertion of the considered dimensions in this equality yields M1 L−1 T−2 = (ML−1 T−1 )m1 (LT−2 )m2 (L)m3 . Equating the exponents of identical basic measurement units in the left. time (T) and it is required to clarify whether the parameter P with the dimension of pressure is dimensionallydependent on parameters µ with the dimension of dynamic viscosity. From Eq. [g] = LT−2 . Write the dimensions of all the parameters under consideration [p] = M1 L−1 T−2 . Let us show how to do it. the ratio p/(ρv2 ) is dimensionless since the dimensions of the numerator and denominator coincide. [v2 ] = M0 L2 T−2 ⇒ M1 L−1 T−2 = (M1 L−3 T0 )(M0 L2 T−2 ). Hence. length (L). m3 = − . length. −1 = −m1 + m2 + m3 . let us consider mechanics. This system has a single solution m1 = 1. m2 = Thus. acceleration. m2 and m3 that would obey the equality [p] = [µ]m1 [g]m2 [D]m3 .6 DimensionallyDependent and DimensionallyIndependent Quantities 165 [p] = [ρ][v]2 [p] = M1 L−1 T−2 . we obtain [p] = [µ][g]1/2 [D]−1/2 (6. [D] = L. For example. m3 . in which there are three basic measurement units: mass (M). 2 2 which proves that pressure is a dimensionallydependent parameter in the system of three parameters – viscosity. Now let us look for the numbers m1 .11) .6. we get a system of three linear equations 1 = m1 . m2 . [µ] = ML−1 T−1 . There is a general algorithm capable of determining whether one or another parameter is dimensionallydependent or dimensionallyindependent of other given parameters. (6.11) 1 1 . g with the dimension of acceleration and D with the dimension of length. [ρ] = M1 L−3 T0 .
. If it is a dimensional quantity. . . .. an }.. . If the dimension of a3 is expressed through {a1 .. . O2 . [a1 ] = [O1 ]x1 [O2 ]x2 . . . .. an . .. a2 . . a2 }. . . . . . . . m1 xk + m2 yk + . .. this quantity is rejected and instead we take a4 . Let us focus our attention on the question of the maximal number of dimensionallydependent quantities for a given set of dimensional parameters {a1 .. y1 . . zk are exponents of this formula. an } we proceed as follows.[Ok ]zk where O1 . the question of the dimensional dependence (or independence) of each of them on the other as well as the question of the maximum number of dimensionallyindependent quantities is reduced to solutions of the system of linear equations m1 x1 + m2 y1 + . If this set contains n elements. a2 } a3 . . If the dimension of a3 is not expressed through the dimensions of {a1 . + mn z1 = 0. .. If a2 has dimension different from the dimension of a1 . . z1 . the next quantity a2 is added to it. .. .. . yk . . . . . . . . . . Let us take a quantity a1 .. a2 . . . which in its turn is tested for independence from the quantities {a1 . .. ... .. ... a2 . it is always possible to separate from it a subset containing the maximum possible number k ≤ n of dimensionallyindependent parameters. If we write dimensional formulas for all the parameters a1 . x2 . + mn z2 = 0. .. the system {a1 . . . y2 . .[Ok ]xk ... . mn and to determine the rank of the system matrix. . . . This assertion follows from the known theorem of algebra that states that from any system of linear equations can be separated the maximum possible number of linearly independent equations. .166 6 Dimensional Theory it follows that the ratio p/(µg1/2 D−1/2 ) is a dimensionless quantity since the dimensions of the numerator and denominator coincide.10). . Next a3 is added to the system of quantities {a1 .. . . . . . [an ] = [O1 ]z1 [O2 ]z2 ... In such a way all quantities in the set . . . a2 } by the formula (6. . z2 .[Ok ]yk . . this number being called the rank of the system of equations. . . the system {a1 . a2 }. .. In order to separate a subset containing the maximum number of dimensionallyindependent quantities of the set {a1 . a2 } would consist of dimensionallyindependent quantities.. xk . .... . .. m2 . In a similar manner many examples of the same type could be considered. + mn zk = 0 (k equations with n unknowns) with respect to the unknowns m1 . . m1 x2 + m2 y2 + . Ok are symbols of basic measurement units in the given system of units and x1 ... . . a2 . a3 } represents a system of dimensionallyindependent quantities. . . . [a2 ] = [O1 ]y1 [O2 ]y2 . ..
Answer. . As the second quantity we take g. In particular. As the ﬁrst quantity. {ρ. It is evident that the dimension of ρ cannot be expressed through the dimensions of D and g. There are of course other possible subsets of the given set consisting also of a maximum number of dimensionallyindependent quantities. ν – kinematic viscosity. [Q] = m × s2 m s s m2 m . an } are expressed through the dimensions M. . D}. this set can have no more than three dimensionallyindependent parameters. [v] = . It is required to separate the maximum number of dimensionallyindependent quantities among a set of the following parameters: p – pressure. . s s Since all the parameters are expressed through mass. as the third dimensionallyindependent quantity we take ρ. Finally. D – diameter. Exercises. Note that for one and the same set several different subsets containing the maximum number of dimensionallyindependent quantities could be separated. ρ – density. a2 . when the dimensions of all quantities in the set {a1 . 1. ω. ρ} consisting of three parameters represents a basis of the maximum number of dimensionallyindependent quantities in the given set of parameters. the maximum number of dimensionallyindependent quantities is less than or equal to three. g – acceleration due to gravity. . for example.6 DimensionallyDependent and DimensionallyIndependent Quantities 167 {a1 . It is easy to verify that. D}. For example {ρ. . Such subsets by themselves can be different but the number of elements in them would be equal. [g] = 2 . an } are considered and gradually the subset containing the number of dimensionallyindependent quantities is separated. D} and so on consist of dimensionallyindependent quantities and the number of dimensionallyindependent quantities in each of these subsets is a maximum and equal to three. T. the subsets {p. the set {D. L. Write the dimensional formulas for all given parameters using the SI system of basic measurement units [p] = [ν] = m3 kg m kg . a2 . Q – volumetric ﬂow rate. . we take D. [ρ] = 3 . Q. g. v – velocity. v. [D] = m. . Solution. Hence. D – wheel diameter. . It is required to separate a basis consisting of the maximum number of dimensionallyindependent quantities among a set of the following parameters: ρ – density. Q – volumetric ﬂow rate. because its dimension contains time and consequently it cannot be expressed through the dimensions of D. . . ω – frequency of revolutions.6. g – acceleration due to gravity. length and time. since it contains mass. Exercise.
. Thus. J kg−1 K−1 ).. . . an among which can be dimensional as well as dimensionless parameters. S – area of the crosssection occupied by ﬂuid.. a2 .. .... . ρ – density.. a2 . ak+1 .5. . c – heat capacity. D – diameter of the pipeline. [ak+2 ] = [a1 ]n1 [a2 ]n2 . ni . ak . the Buckingham theorem. [ak ]nk . a2 .. pi are real numbers.. ... Let the maximum number of dimensionallyindependent arguments in the set a1 . It is required to separate a basis consisting of a maximum number of dimensionallyindependent quantities among a set of the following parameters: v – velocity. Answer.. . . . Then the remaining (n − k) arguments of this function ak+1 . .168 6 Dimensional Theory 2. .. v}. For example {D.12) of arguments a1 . [an ] = [a1 ]p1 [a2 ]p2 . . . . . For example {ρ. . .. m2 s−1 . 2 = n1 n2 a1 a2 · · · ank k an n−k = p1 p2 p a1 a2 · · · ak k (6. λ – thermal diffusivity factor.. .. . an ) (6.. c. that is [ak+1 ] = [a1 ]m1 [a2 ]m2 . ak+2 . v. ν – kinematic viscosity..13) . . Let a physical regularity be represented by a function A = f (a1 . . . . [ak ]pk where mi . . 3. It is required to separate a basis consisting of a maximum number of dimensionallyindependent quantities among a set of the following parameters: p – pressure.. . an would be dimensionallydependent on the ﬁrst k arguments. an be k and without disturbing the generality it can be taken that the ﬁrst k arguments are a1 . a2 .7 Buckingham Theorem Now we go to the proof of the central theorem of dimensional theory. . θ}. α – angle of inclination of the pipeline axis to the horizontal (dimensional quantity). . Answer. .. g – acceleration due to gravity.. v – velocity. . [ak ]mk .. θ – temperature. . which was partially formulated in Section 6. the relations ak+1 . 1 = m1 m2 a1 a2 · · · amk k ak+2 .... ak . 6... . . . W m−1 K−1 . .
.12) will be invariant. . . . the numerical values of the parameters and 1 .6. that is. . there should exist a relation [A] = [a1 ]q1 [a2 ]q2 . (6. the diameter D of the pipeline. . the thickness δ of the pipeline wall and the elastic modulus (Young’s modulus) E of the steel from which the pipe is made. a2 . . n−k ). . (6. . .7 Buckingham Theorem 169 are dimensionless parameters since the dimensions of the quantities in the numerators and denominators of these fractions are identical. ak . . . pi p2 p am1 am2 · · · amk a1 a2 · · · ak k 1 2 k where f represents a function resulting from f after redeﬁnition of its arguments. 2. a2 . a2 .16) between dimensionless complexes made up from arguments of the dependence under consideration. From here it follows that the function cannot depend on its ﬁrst k arguments a1 . .12). Hence. . . . ak . . If it is not expressed through these dimensions.14) can be represented as follows = f (a1 .14) has to be dimensionless. . .13) and (6. 1. . . . n−k would not be changed because they are dimensionless quantities. . of the form (6. a2 . an . The augmentation of the crosssection area S of a steel pipeline when setting up in it excess pressure P depends on the value of this pressure. . . [ak ]qk meaning that the ratio = q q a11 a22 A q · · · akk (6. . . . . we have shown that any physical dependence between dimensional quantities of the type (6. The number of these complexes would be less than the number of arguments of the initial dependence by a number equal to the maximum number of dimensionallyindependent quantities among these arguments. Exercise 1. . ak . . . . . This dependence may be rewritten as q q a11 a22 A q =f · · · akk a1 . . . 2. .16) Thus. a2 . . n−k ). This dependence with (6. . ak by going from one system of basic measurement units to another one. since the dependence (6. ak+1 an . . .15) would have the following simple form =f( 1.15) If we now arbitrarily and independently from each other vary the numerical values of the arguments a1 . independent of the choice of measurement units. 2 . ak . a2 . . Let us revert to the dependence (6. . it would also not be expressed through the dimensions of all the quantities a1 . The dimension of the quantity A should also be expressed through the dimensions of the arguments a1 .
the number of arguments in the dependence under consideration may be reduced to two. δ/D). could be expanded in a Taylor series in the vicinity of the point 1 = 0 leaving in the expansion only the ﬁrst term. 1 2) = P/E. a3 = ρ. or 2 = δ/D. Hence. µ. [ρ] = kg m−3 . to 0. the dependence under study could be written as S = D2 · 1 P D3 · P · · f0 = f0 · E δ/D δ·E where f0 is a certain constant. From this it follows that all three arguments of the function are dimensionallyindependent. k = n = 3. 2 ) = 0 at P = 0 and the ratio P/E is very small: P ≈ 2–7 · 106 a.785. taking the critical velocity as a function of three parameters: pipeline diameter d. a1 = d. Hence. E). D. The dimensions of the parameters in the SI system are: [vcr ] = m s−1 . if additionally we invoke the reasoning that the variation of pipeline crosssection area S should be proportional to P/E (the function. δ. i. in the dependence under study there is only one constant left to be obtained. = Exercise 2.e. [D] = m. =f( 1.e. i. In the given case A = vcr . Consequently. a2 = µ. n = 3. Thus we have: = S/D2 . In other words it is required to investigate the dependence vcr = f (d.170 6 Dimensional Theory Using the theorem. the dimensional analysis has shown that in the considered dependence there are only two dimensionless complexes P/E and δ/D instead of four dimensional arguments. Moreover. and inversely = proportional to δ/D (δ ≈ 5–10 mm. [δ] = m. The dependence to be sought can be written in a general form as S = f (P. because S = f (0. it is required to write this dependence in dimensionless form and to clarify how many dimensionless parameters determine it. Among these there are only two dimensionallyindependent quantities. E ∼ 2 · 1011 a. Solution. S = D2 · f (P/E. [d] = m. for example D and E. The dimensions of the arguments of this dependence in the SI system are: [P] = kg m−1 s−2 . ρ). D ≈ 300–100 mm). Solution. dynamic viscosity µ and density ρ of the ﬂuid. . It is required to investigate the dependence of the critical ﬂow velocity vcr at which this transition happens. and the number of arguments in the dimensionless writing of the function under study may be reduced by three. f . [E] = kg m−1 s−2 . Theoretical investigation shows that the constant f0 is equal to π/4 ∼ 0. It is known that the laminar ﬂow of a viscous incompressible ﬂuid in a circular pipe loses stability and becomes turbulent. [µ] = kg m−1 s−1 .
Setting up dimensionless complexes . a5 = g). ρ · D3 2 = ν g 1/2 D3/2 . [g] = m s−2 . The theory and experiments have shown that Recr ∼ 2300. the sought dependence can be represented as follows =f( or v= gD · f m/ρD3 . ρ. a3 = ρ. the viscosity of water is ≈0. 1 St = 10−4 m2 s−1 . D. [ρ] = kg m−3 . 2) It is seen that the dependence contains in fact not ﬁve dimensional arguments. At Re < Recr = the ﬂow is laminar whereas at Re > Recr it is turbulent.7 Buckingham Theorem 171 The single dimensionless complex can be written as = vcr · d/(µ/ρ). [m] = kg. (µ/ρ) (6. [D] = m. The sought dependence then takes the especially simple form cr = vcr · d = const. the number of arguments in the sought dimensionless writing may be reduced to two (5 − 3 = 2). A ball of mass m and diameter D is dropped in a viscous ﬂuid with density ρ and viscosity ν under the action of gravity (gravity acceleration – g) with constant velocity v. but only two dimensionless complexes. The dependence of this velocity on the governing parameters: v = f (m. in the SI system are [v] = m s−1 .17) The ratio µ/ρ is called the kinematic viscosity of the ﬂuid and is denoted by ν ([ν] = L2 /T). 1. 1 = m . Dimensions of the parameters (n = 5). In the SI system the unit of kinematic viscosity is stokes (St). The dimensionless parameter cr determining the transition of laminar ﬂow to turbulent ﬂow is called the critical Reynolds number and is denoted by Recr . a4 = ν. D and g). Using the theorem it is required to write the sought dependence in dimensional form. Exercise 3. in this problem (A = v.6.01 St = 1 cSt (centistokes) = 10−6 m2 s−1 . a1 = m. The maximum number of dimensionallyindependent parameters among the arguments is equal to three (as such parameters can be taken for example ρ. ν. a2 = D. Using the theorem. g) is to be investigated. 1 and 2 = v g 1/2 D1/2 . Exercises. Solution. [ν] = m2 s−1 . 1. it is required to write in dimensional form the dependence of the resistance force F experienced by a submarine . ν/ gD3 .
ρ. ν). g. 5/4 3/4 . S. ρ. D. L). if we accept that this force depends also on the diameter of the submarine crosssection D and the submarine length L. D. s). Using the theorem. The sought function is Q = f (g sin α. D. T/ D/g = f ( gD3/2 /ν. The outﬂow happens under gravity (acceleration due to gravity g) through the drain system with ﬂow area s located at the tank bottom. the length of the line is L and the acceleration due to gravity is g. s/D2 ). Using the theorem. ρ. Using the theorem. 3. T/ L/g = const. √ Answer. L. L. (g sin α)1/2 · S0 /ν). if this ﬂow represents voluntary (gravity) ﬂow. p/(ρgD). it is required to write in dimensional form the dependence of the volumetric ﬂow rate Q of a ﬂuid with density ρ and viscosity ν in an inclined pipeline (inclination angle α) having crosssection area S0 . that is T = f (m. 2. ν. g. T/ D/g = f ( gD3/2 /ν. The ﬂow happens under the action of the gravity force projection g sin α. L/D. if the mass of the latter is m. ν. it is required to write in dimensional form the dependence of the time T of complete outﬂow of ﬂuid (density ρ and viscosity ν) from a tank car with diameter D and length L. it is required to write in dimensional form the dependence of the oscillation period T of a mathematical pendulum (massive point on nonstretchable line). S0 . √ Answer. ν. that is F = f (v. s/D2 ). the area of the pipeline crosssection ﬁlled with ﬂuid is S < S0 . The sought function is T = f ( p. it is required to write in dimensional form the same dependence as in the previous exercise with the single distinction that the outﬂow happens not only under the action of gravity but also under the action of positive pressure p created inside the tank. L/D.172 6 Dimensional Theory moving in water (density ρ. F/(ρv2 D2 ) = f (vD/ν. g). s). L. Answer. 5. The sought function is T = f (ρ. 4. Q/[S0 · (g sin α)1/2 ] = f (S/S0 . kinematic viscosity ν) with velocity v. D/L). Using the theorem. Answer. Answer.
The question arises as to whether it is enough to provide merely geometric similarity of fullscale and model tanks and to use in experiments the same liquid or is it necessary to replace the liquid with another one with specially selected properties.173 7 Physical Modeling of Phenomena The main advantage of the dimensional theory is that it opens up possibilities to use the similarity laws of physical phenomena and allows modeling of these phenomena through replacing them in nature by similar phenomena on a reduced or enlarged scale under experimental conditions.e. In order to solve this problem it is decided to make a copy of the tank on a reduced scale. yields the sought radius of the circle inscribed in the fullscale triangle. The similarity factor of this triangle to the fullscale one would be equal to 10 000. It is required to clarify whether the river dam will withstand the dynamic head of ﬂooding water. However. Michael V. 10 000.1 Similarity of Phenomena and the Principle of Modeling In order to elucidate the essence of modeling. one could proceed as follows: on a sheet of paper draw a triangle on a reduced scale similar to the given triangle. This problem may be easily solved by simple algebraic calculation. ﬁll it with some model liquid and then measure the dump time. Let it be required to determine experimentally the dump time of a tank of complex geometrical form and very large size. Modeling of Oil Product and Gas Pipeline Transportation. for example a triangle with sides 10. KGaA. Assume that it is required to calculate the radius of a circle inscribed in a triangle whose sides are very large. 7. i. Of course in the case under consideration we are dealing with a simple geometric similarity and geometric modeling but it clariﬁes the essence of modeling in the general case. Moreover. it is easy to measure its radius. let us consider several examples. Inscribing a circle in the depicted triangle. Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. Then the obtained number multiplied by the similarity factor. one more example from engineering practice. is it necessary to determine the ratio between the measured dump time and that actually taking place in the fullscale object. The answers to these questions should give the theory of simulation (modeling). Finally. Weinheim ISBN: 9783527408337 . if it was necessary to perform this measurement experimentally. 20 and 30 cm. 2 and 3 km. for example 1.
(7. . n−k n−k .3) show that if the parameters a1 .2) may be rewritten in dimensionless form as follows =f =f 1. 2. . a n . analogous parameters of another phenomenon are determined by simple recalculation of the same kind as for transition from one system of measurement units to another.e. (7. n = n. a2 . . . transport. . (7. 7. 2. . .1) and (7. so that A = f (a1 . i. When reducing the linear sizes of the considered phenomenon. the storage of oil and gas and so on. .2) In accordance with the Itheorem both dependences (7.3) where k is the number of dimensionallyindependent parameters among the quantities a1 . . we have A = f a1 . .2 Similarity Criteria Let us determine the necessary and sufﬁcient conditions of two phenomena to be similar. Similar problems take place in different ﬁelds of engineering: hydraulic engineering. . from the given parameters of one phenomenon. from ferroconcrete. It is necessary especially to change the scales of many other parameters of the phenomenon. aviation. a2 . . Each of such phenomena is called a model of another phenomenon from this set of phenomena. . an ). . Thus. an . . . . . an . Relations (7. a2 . . an differ from those determining the quantity A. it is insufﬁcient to provide geometric similarity. . . . Two phenomena are called similar when. a2 . . . . the dam would withstand any water head. It is evident that if the dam were made from the same material as in the natural conditions. a2 . . . (7. . an are chosen such that the following conditions are obeyed 1 = 1. . Such conditions are called similarity criteria. mounted in an experimental channel in a hydrological laboratory. It is necessary to determine the velocity of water in the experimental channel in order to model the river head.174 7 Physical Modeling of Phenomena For this purpose a reduced size copy of the dam is made. . 1. . . .4) . .1) The model under consideration consists of the dependence of an analogous physical quantity A on the same physical parameters the numerical values of which a1 . 2 = 2. . Deﬁnition. . a 2 . . Let a phenomenon be such that a certain physical quantity A is determined by a set of physical parameters a1 .
namely conditions (7. consider the modeling of the stationary ﬂow of a viscous incompressible ﬂuid in a model pipe with reduced size as compared to the fullscale one. ε).6 this dependence is p = λ(Re. = ν ν d = d . Thus. L ρv2 · d 2 Denote through ρ . v . µ .3 Modeling of Viscous Fluid Flow in a Pipe As an example of two similar phenomena. by deﬁnition.4).3 Modeling of Viscous Fluid Flow in a Pipe 175 Then the following condition would also be satisﬁed = . the dimensionless parameters 1 . L ρv2 · d 2 The parameters of the model pipe and the ﬂow regime in it are chosen to obey the relations Re = Re . 2 .7) In so doing we ensure the equality p p = L ρv2 L ρv2 · · d 2 d 2 . Then p = λ(Re . (7. ε = ε or vd vd . L . be similar. Hence.6) and the considered phenomena would. . p the values of the hydrodynamic parameters relating to the ﬂow in the model pipe. The same parameters without the superscript ‘‘prime’’ refer to the phenomena to be modeled. d . n−k are the sought similarity criteria.7. 7. . . necessary and sufﬁcient conditions of two phenomena to be similar are equalities of the dimensionless complexes determining these phenomena. In accordance with the results given in Section 1. . . ε ). (7.5) The value of the parameter A could be found by simple recalculation of the parameter A through A=A · am1 am2 · · · amk 1 2 k a1m1 a2m2 · · · akmk (7.
ν gd sin α · d .7. respectively. ν d It is more convenient in this dependence to use the ratio of the ﬁrst argument to the second one instead of the second argument.4 Modeling Gravity Fluid Flow This type of ﬂow has already been considered in Section 3. h =f d (Q/d2 ) · d . As a result we have h = f1 d (Q/d2 ) · d (Q/d2 ) . . ν gd sin α d In the ﬁrst argument it is easy to recognize the Reynolds number Re of the ﬂow calculated from the velocity v = Q/d2 and the kinematic viscosity ν = µ/ρ. d (7. ν. we could determine the ﬂuid ﬂow velocity and roughness of the pipe walls of the experimental (model) installation needed to achieve the similarity v =v· d ν · .9) state that the similarity in the case under consideration is afforded by fulﬁlling two conditions: equality of the Reynolds numbers and equality of the wall relative roughness. It is required to determine the depth h of the ﬂuid ﬁlling the pipe crosssection. 7. (7. d. The second argument .9) Formulas (7. ). that is. We now set a question on the physical modeling of this process. .8) If we now take the ratio d /d. therefore in dimensionless form the number of independent arguments will be reduced from ﬁve to three. for example ν. Among ﬁve arguments of this function there are two dimensionallyindependent. Consequently. the function h = f (Q. Rewrite the dependence to be sought in dimensionless form using the theorem. d. so that the ﬁrst similarity criteria would be 1 = Re. and the ratio ν /ν of the viscosities of ﬂuids ﬂowing in fullscale and model pipes. Let the ﬂow rate Q of the ﬂuid in a pipe inclined to the horizontal at an angle α be given. the Reynolds number and relative roughness serve as similarity criteria for the problem on the ﬂow of a viscous ﬂuid in a pipe. g sin α. . showing how many times smaller are the sizes of the model pipe than the sizes of the fullscale pipe (factor of geometric similarity).176 7 Physical Modeling of Phenomena or p= p · L d ρ ν · · · L d ρ ν 2 . d ν = · d .
the following equalities should be true ν ν or sin α = sin α · ν ν 2 · d d = sin α · sin α d d d d 3 5/2 · . (7. d (7.13) If in the model the same liquid as in the fullscale pipe is used. Then the conditions for the experiment should be: Q = 1/5 · 1/5 · Q = 0. ε). The Froude number is. sin α = 1/25 · 53 · sin α = 5 · sin α. and the viscosity ν of the model liquid be ﬁve times less than the viscosity of the natural liquid. that is. let the linear size of the experimental pipe be 1/5th that of the fullscale pipe. d d d (7. the depth h to which the pipe is ﬁlled can be calculated using the formula h d h = ⇒ h=h · . that is.10) To model this phenomenon it is necessary to provide the following similarity conditions Re = Re ⇒ Fr = Fr ⇒ Q Q = ⇒ Q =Q· νd νd Q2 Q2 = . it must be chosen in accordance with relation (7.12) In order for the ﬁrst two conditions to be consistent. = 0. then the viscosity of the ﬂuid used in the model should be different. the internal surface of the . g sin α · d 5 g sin α · d5 sin α · sin α ⇒ = d d · 5/2 ν ν · d d . d (7. equal to v2 /gd.2 · . Fr. Hence.11) d d If these conditions are obeyed.04 · Q. ⇒ Q =Q· ε =ε ⇒ = . If α = α. namely ν = ν · (d /d)3/2 . For example. the third argument of this dependence 3 = ε is the relative roughness of the pipe’s internal wall surface. the dependence under study is h = fˆ (Re. d . in general.7. d /d = 1/5. the ﬂuid ﬂow rate in the experiment has to be reduced by a factor of 25.4 Modeling Gravity Fluid Flow 177 is called the Froude number. the second similarity criterion can be taken as 2 = Fr.13). Nevertheless. then a different slope of the pipe should be used in the model. Fr. Finally. thus in our case we are dealing with a Froude number calculated from the velocity Q/d2 and the component of the acceleration due togravity g sin α.
Thus. The last relation shows that the sufﬁcient condition for the dynamic similarity of the model . Reduction of = the ﬂow rate and enhancement of the pipe slope (at small values of α) present no special problems. If. S) in dimensionless form is written as T D/g =f L S . The tank is provided with bottom drain equipment with the area of the ﬂowing crosssection S. in a given case a ﬂow of diesel fuel with viscosity 3 cSt is to be investigated. The dependence under study T = f (D. D D2 gD · D . 2 = 2 the criteria of geometric similarity.178 7 Physical Modeling of Phenomena pipe should be so polished that the absolute roughness is reduced by ﬁve times as compared to the fullscale pipe and the sine of the inclination angle of the model pipe should be increased by ﬁve times. for example. It is required to design an experimental installation to model the process and to investigate the dependence of the time T of the oil outﬂow on the parameters of the liquid and tank. ν ν where the superscripts ‘‘prime’’ refer to the model parameters. ν. L. ν Therefore the similarity criteria are: L S . h=5·h. ρ. D D 1/2 3/2 g ·D = the criterion of dynamic similarity. therefore only the main conditions are satisﬁed. the depth h of the ﬂow in the fullscale pipe has to be ﬁve times greater than the model depth h . ν = 1 3 At the fulﬁllment of the ﬁrst two criteria at which the model tank should be similar to the fullscale one. .6 cSt could be taken for the model. g. It is far more complicated to fulﬁll the last similarity condition and so sometimes it is neglected. In experiments it is not always possible to obey all the required similarity conditions.5 Modeling the Fluid Outﬂow from a Tank Let there be a railway tank with a boiler diameter D and length L or. that is. then benzene with viscosity ∼0. the phenomena would be similar under the condition g 1/2 · D3/2 g 1/2 · D 3/2 = . 7. in general. a tank of arbitrary geometric form intended to transport oil with density ρ and kinematic viscosity ν.
ρ. The propulsive device of the impeller or.14) From this follows. Let the experimental tank be a cylindrical tank 10 times smaller than the fullscale one. so that Q = f ( p.e. D (7. The greater the pressure drop the lesser is the ﬂow rate of the liquid. the feed) depends on the pressure drop p = pd − psuc which the liquid should overcome at its ﬂow from the impeller center to the periphery. makes the liquid ﬂow against the pressure. the kinematic viscosity of the = liquid used in the experiments. Of course. should be 32 times lower than the viscosity of the oil being modeled. Dim . If this oil has sufﬁciently high viscosity then it is easy to select such a liquid. it is necessary to use a liquid with velocity also differing from the oil under study. = 7. If the area σ of the outlet branch pipe at the discharge line is given by the pump design. that is. electrical and so on). that for modeling of oil outﬂow on the installation with sizes distinct from the fullscale tank. If this condition is fulﬁlled. The time T of oil outﬂow is determined by the time T measured in the experiment. can be an internal combustion engine.6 Similarity Criteria for the Operation of Centrifugal Pumps As already described in Section 4. thermal. that is D/D = 10. the ﬂuid velocity Q/σ is a function of the impeller diameter Dim (centrifugal pumps may have accessory impellers).6 Similarity Criteria for the Operation of Centrifugal Pumps 179 tank to the fullscale one is the condition ν =ν· D D 3/2 . it is possible only by the work of external energy sources (mechanical. the angular velocity ω of the impeller rotation (centrifugal pumps may be supplied with equipment to change ω). in the direction from lesser pressure psuc at the suction line to the greater pressure pd at the line of discharge under pressure. vaporgas turbine or any other source of rotational moment (see Fig.15) . otherwise the problem is complicated. ω.162 · T . in particular. The volumetric ﬂuid ﬂow rate Q (or.2. then T / D /g = T/ D/g or T=T · D . as it is called. the density ρ and viscosity ν of the ﬂuid to be pumped. as it is called. ν).1 pumps are equipment to make liquids ﬂow against a pressure force. σ (7. i.1).14) it is T ∼ 3. Then ν ∼ 0. In accordance with the formula (7.0316 · ν. 4. Centrifugal pumps represent a variety of pump in which the centrifugal force.7. acting on ﬂuid particles rotating in the impeller. the pump drive.
n ≈ 3000 rpm. has the characteristic H= ω ω0 2 · F∗ ω0 ·Q .e. The latter conclusion is of course true only up to certain limits. Dim ≈ 0.15) is called the (Q − p) characteristic of the centrifugal pump and is to a large extent determined by its constructive peculiarities. therefore its dimensionless form could be reduced to Q/σ = f˜ ωDim p ν . i. Among the arguments of the function (7. ω.18) • If a centrifugal pump. the same pump operating with another . . ωDim ν (7. therefore in practice their variation only slightly affects the (Q − H) characteristic of the centrifugal pump and thus the inﬂuence of the pumping ﬂuid viscosity is also small. v ≈ 1–10 · 10−6 m2 s−1 ).2–0. the form of the latter is true for the pump operated by liquid of any density. If we ignore the inﬂuence of pumping ﬂuid viscosity on the (Q − H) characteristic of the centrifugal pump. This parameter reﬂects the effect of viscosity on the characteristic of the pump operation. it is small.180 7 Physical Modeling of Phenomena The dependence (7. . we get the socalled (Q − H) characteristic of the pump H = ω2 Dim 2 · F Q ωDim 2 . As for the inﬂuence of the viscosity ν on the form of (Q − H) characteristic of centrifugal pumps. operating with an impeller of diameter Dκ0 has the characteristic H = F∗ (Q).16) Here the constants σ and g are taken into account by the form of the function F.15) with respect to this head. then it can be represented in the simple form H = ω2 Dim 2 · F Q ωDκ . If we introduce into consideration the quantity H = p/ρg (the differential head of the pump) and solve Eq.15) there are three dimensionallyindependent parameters. It is interesting to note that the density ρ of the liquid to be pumped does not enter into the dependence (7. for example Dim . ω = 2π · n.16). (7. the same pump working with varied rotation frequency ω or revolutions per minute n. ω (7.7 m.17) • From the derived formulas some practically important conclusions follow: If a centrifugal pump operating with angular velocity ω0 or revolutions per minute n0 has a characteristic H = F∗ (Q). (7. ρω2 Dim 2 ωDim · Dim The last argument represents none other than a quantity inversely proportional to the Reynolds number Re. Reynolds numbers ωDim 2 /ν have rather large values (≈106 ) due to the high rotation velocity of the pump impeller (ω ≈ 300 s−1 . ρ.
H) undergoes a displacement along the Haxis by a· 1− ω2 Dim 2 ω2 Dim0 2 0 .21) where a and b are approximation factors.20) allow us to change the (Q − H) characteristic of centrifugal pumps by changing the rotation speed or/and the impeller diameter.18)–(7. as has been said. The pump has the following characteristic H = 331 − 0. If we now change the impeller diameter from Dim0 to Dim and the rotation frequency from ω0 to ω. the same pump operating with varied rotational frequency ω and impeller diameter Dim has the characteristic H= ωDim ω0 Dim0 2 · F∗ ω0 Dim0 ·Q . the (Q − H) characteristics of centrifugal pumps are represented in the form of a parabola H = F∗ (Q) = a − b · Q 2 (7.22) the sought characteristic has the following form H = 331 · 3000 · 480 3200 · 490 2 − 0.451 · 10−4 · Q 2 . It is required to determine the characteristic of the same pump if we reduce the impeller diameter to 480 mm and the number of revolutions per minute to 3000. The diameter of a centrifugal pump impeller is 490 mm and the impeller velocity is 3200 rpm. Dim (7. ωDim (7.20) Rules (7. Solution.6 Similarity Criteria for the Operation of Centrifugal Pumps 181 impeller of diameter Dim has the characteristic H= • Dim Dim0 2 · F∗ Dim0 ·Q .451 · 10−4 · Q 2 = 279 − 0. the (Q − H) characteristic of the same pump takes the form H =a· ω2 Dim 2 − b · Q 2.7. ω2 D2 0 im0 (7. (7.451 · 10−4 · Q 2 ( H in m. In many cases. Q in m3 h−1 ).22) that is the parabola graphic in the plane (Q. operating with rotational velocity ω0 and impeller diameter Dim0 has the characteristic H = F∗ (Q). .19) If a centrifugal pump. In accordance with Eq. Exercise.
Answer. Q in m3 h−1 . H = 410 − 0. 1.182 7 Physical Modeling of Phenomena Exercises.75 .436 · 10−4 · Q 1. 3.75 . Q in m3 h−1 . Answer. How does the (Q − H) characteristic of a centrifugal pump vary if the frequency speed of the impeller is increased from 3000 to 3200 rpm? The given characteristic of the pump is: H = 360 − 0. Q in m3 h−1 .375 · 10−4 · Q 1.75 . H = 410 − 0. where H is measured in m.42 · 10−4 · Q 2 . 2. How does the (Q − H) characteristic of a centrifugal pump vary if the frequency speed of the impeller is increased from 3000 to 3200 rpm? The given characteristic of the pump is: H = 360 − 0. The given characteristic of the pump is: H = 360 − 0. where H is measured in m. Answer. How does the (Q − H) characteristic of a centrifugal pump vary if the impeller diameter is reduced from 480 to 470 mm. H = 345 − 0.42 · 10−4 · Q 1.75 .42 · 10−4 · Q 2 . .42 · 10−4 · Q 1. where H is measured in m.
Let us bring the model equation to dimensionless form. 8.1 Origination of Similarity Criteria in the Equations of a Mathematical Model A somewhat different approach takes place in the mathematical modeling of phenomena in construction systems of algebraic or differential equations with initial and boundary conditions in which we would like to see an adequate model of the considered phenomena or process. Hence. that is the mathematical model of the process.1) where x(t) is the linear dependence of the weight coordinate on time and k is ˙ the restoring force factor. Dimensionless similarity parameters originate in these models in a strictly speciﬁed way of bringing model equations to a dimensionless form. it is by no means necessary to know which equations satisfy these parameters and with which physical laws they are connected. . .183 8 Dimensionality and Similarity in Mathematical Modeling of Processes In the previous chapter we have seen that dimensionless similarity criteria 1 . v0 = 0). position and velocity x(0) = v0 of the weight are also given. . . KGaA. It is enough to know only the dimensions of these parameters in order to set up dimensionless similarity criteria by which the class of considered phenomena is characterized. Weinheim ISBN: 9783527408337 . an by way of heuristic reasoning. First we introduce dimensionless variables ¯ x= x L ˙ x ¯ ˙ and x = . 2 . Let us illustrate the afore said by a simple example. . . . n−k appear in the considered problems after ﬁxation of a set of governing parameters a1 . at t = 0. The initial. Michael V. v0 (L = 0. Modeling of Oil Product and Gas Pipeline Transportation. a2 . Lurie Copyright 2008 WILEYVCH Verlag GmbH & Co. . It is known that onedimensional oscillations of a pont weight with mass m on an elastic spring around an equilibrium position (x = 0) are described by an ordinary differential equation m d2 x = −k · x dt2 (8.
t t Returning to dimensional quantities. It is easy to verify that the solution of the Eq. (8.1–5. If in two situations this parameter appears to be identical.2) is ¯ t x(¯) = cos √ √ 1 ¯ + √ · sin ¯ . The main equations modeling such ﬂows are ∂p 2 ∂(ρv) ∂t + c ∂x = 0 (8. the solutions of Eq. the oscillation of different weights with different mass on springs with different elasticity caused by different initial conditions is in fact described by a mathematical model containing only one dimensionless parameter 2 = kL2 /mv0 . 2 d¯ t mv0 (8. Hence.2) are indistinguishable and consequently we have to deal with similar situations.3.184 8 Dimensionality and Similarity in Mathematical Modeling of Processes Then the dimensionless time ¯ = t/(L/v0 ) is determined.2) ¯ L · d2 x ¯ = −k · (L · x) (L/v0 )2 · d¯2 t ¯ ¯ ˙ The initial conditions take the form x(0) = 1. In new variables the t differential equation transforms to m· or ¯ d2 x kL2 ¯ = − 2 · x.2 OneDimensional NonStationary Flow of a Slightly Compressible Fluid in a Pipeline The theory of onedimensional nonstationary ﬂows of a slightly compressible ﬂuid in a pipeline was considered in detail in Sections 5. 8. m As is known k/m = ω is the frequency of harmonic oscillations of the weight and the amplitude h is the square root of the sum of the squared factors of the sine and cosine h= L2 + 2 mv0 =L· k 1+ 1 . ε) · ρv ∂t ∂x ∂x d 2 . we obtain x(t) = L · cos k ·t + m m · v0 · cos k k ·t . (8. x = 1.3) 2 ρ ∂v + v ∂v = − ∂p − λ(Re.
(8.3) as the mathematical model of this ﬂow.5) it is seen that there are three dimensionless criteria governing the class of problems under consideration and differing from each other only by the numerical values of the parameters entering in these equations 1 = v0 . starting from some time a valve located at the end crosssection of the pipeline x = L begins to vary its opening level with frequency ω. ¯ v = v0 · v. for ﬂuid ﬂow M ≈ 0. ρ = ρ0 · ρ. very small (for example. However. Equations (8. ¯ ¯ p = p0 · p. ε) · L v0 ρv2 ωL ∂ v v0 ¯ · + ·v =− − · · ¯ ¯ c ∂¯ t c ∂x ∂x d c 2 (8. . as a rule.001. Solution.4) is called the velocity of the pressure wave propagations in pipeline. The ratio of the ﬁrst criterion to the second one 1 / 2 = v0 /(ωL) is called the Strouchal number St = v0 /(ωL). c 3 = λL . Consider a certain problem on the calculation of nonstationary ﬂow of a slightly compressible ﬂuid in pipeline taking Eqs. The Mach number of the ﬂuid or gas ﬂow in a main pipeline is. for gas ﬂow M ≈ 0. Problem. Introduce the following dimensionless variables marked by the horizontal bar at the top t= 1 · ¯. Let there be at the pipeline section 0 ≤ x ≤ L stationary ﬂuid ﬂow with velocity v0 .03). ε) · ρ v2 · ρv ρ0 ρ ¯ 0 0 ¯ ¯ 1/ω ∂ ¯ t L ∂x L ∂x d 2 If now we tak