Anda di halaman 1dari 69

UNIVERSITY OF EXTREMADURA

School of Industrial Engineering


Department of Mechanical, Energetic and Material Engineering

FLUID MECHANICS

Jos Mar Montanero


e
a
Area of Fluid Mechanics

Foreword
This subject aims at teaching the basic concepts of uid statics and dynamics, as well as their applications to industrial
engineering. The subject contents are divided into two parts: (i) general fundamentals of uid mechanics, and (ii)
applications of those fundamentals to the study of ows with constant density. In this last part, we shall consider
problems of hydrostatics, aerodynamics, and hydraulics.
General fundamentals.
Any course of uid mechanics must start by explaining the model adopted to study a uid. This involves distinguishing between the microscopic and macroscopic descriptions, clarifying the relationships between them. These ideas
constitute the continuum hypothesis included in chapter 1. In this chapter, we also explain the dierence between
liquids and gases.
In general, a uid-dynamic problem entails the existence of a ow as the response to the action of several types
of forces. Before analyzing the origin of the forces driving the uid movement, it is appropriate to acquire the basic
tools which allow one to describe the ow. This is the purpose of chapter 2. In this chapter, we introduce several
kinematic concepts to visualize the ow, and to extract relevant information contained in the velocity eld which
characterizes it.
We devote chapter 3 to the analysis of the forces acting on uid elements, and to the study of the energy transfer
taking place in liquids and gases. This involves introducing and explaining the following concepts: surface force,
external force, stress tensor, viscosity, heat ux, thermal conductivity, etc. These concepts allow one to derive the
mechanical equations that describe the evolution of a uid system. The understanding of this part of the subject is a
dicult task. It is necessary that students understand the origin of certain statements without spending a long time
on mathematical formalisms.
In chapter 3, we deduced the mechanical laws for a moving uid system. In most cases, one is not interested in the
evolution of systems, but in the local eects that these systems cause when crossing a certain region (a machine, for
instance). The Reynolds transport theorem allows one to translate the equations for a uid system into the integral
relations which describe the evolution of the mechanical quantities in a control volume. Chapter 4 is devoted to get
those relations. The equations for a control volume can be applied to some specic cases to obtain interesting results;
for example, the generalized Bernoulli equation for momentum, or the energy equation for a hydraulic machine. It is
important to consider these applications for the student to notice the relevance and usefulness of the obtained results.
Because of the complexity of most technological uid-dynamic problems, it is unavoidable in many cases to rule out
any theoretical approach. This is so not only in those situations in which the complexity is evident, but also in apparently simple problems. In this scenario, one must often consider information acquired from experiments exclusively.
Experiments must be conducted obtaining the maximum information with the lowest number of experiments. Dimensional analysis (chapter 5) is the set of techniques and methods used to get information on a specic phenomenon,
prior to the experiments, and just from the knowledge of the variables involved in that phenomenon.
Applications.
In chapter 6, we study the behavior of a uid at rest. We shall focus on the calculation of forces and torques
exerted on surfaces and volumes either partially or totally submerged in liquids.
In the rst section of chapter 7, the concept of turbulence is presented, and the consequences of this phenomenon
on the dynamical behavior of uids is explained qualitatively. An important group of uid-dynamic applications is
that constituted by ows taking place next to solid walls for high enough Reynolds numbers (boundary layers). In
this chapter, we provide a qualitative description of this kind of ow. The detachment of boundary layers requires
special attention due to its inuence on the aerodynamic behavior of blunt bodies and the performance of hydraulic
devices. We explain the origin of the boundary layer detachment and its eects on the form drag.
In chapter 8, we pay attention to the analysis of hydraulic applications of crucial importance in industrial engineering. We shall start by showing Moodys diagram, which allows one to calculate the loss of reduced pressure in pipes
working in both laminar and turbulent modes. A practical approach suggests dealing with the problem qualitatively,
avoiding mathematical demonstrations. In the second part of this topic, we analyze the loss of reduced pressure
located in devices frequently connected to multiple-pipe systems. These devices are commonly used as ow meters

because the local loss of reduced pressure is an indirect measurement of the ow rate crossing them. In the third part
of this chapter, we study the loss of reduced pressure taking place in multiple-pipe systems most commonly used in
hydraulics. We classify multiple-pipe systems according to their structure, and provide the students with the basic
rules to calculate ow rates and losses of reduced pressure. In the last section of this chapter, we study the role played
in multiple-pipe systems by pumps connected both in series and in parallel.
Chapter 9 is devoted to the analysis of some ows with free surfaces. Specically, we study the uniform turbulent
ow in channels. Because of the complexity of a rigurous theoretical approach to this problem, we restrict ourselves
to showing Mannings empirical formula to calculate the ow rate transported by a channel with arbitrary shape.
Finally, we present a qualitative description of weirs and sluicegates, giving some useful empiric results.

ii

Table of Contents

Essentials of Fluid Mechanics

1 INTRODUCTION

1 .- GENERAL COMMENTS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.1.- Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1.2.- Continuum hypothesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 KINEMATICS
1 .- INTRODUCTION

5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 .- LAGRAGIAN AND EULERIAN DESCRIPTIONS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.1.- Streamline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2.- Path . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 .- STREAMLINE AND PATH

4 .- TYPES OF FLOWS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.1.- Steady ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.2.- Uniform ows and streams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.3.- Two- and one-dimensional ows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.4.- Axisymmetric ows

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4.5.- Incompressible ows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 .- FLOW RATE AND MASS FLOW RATE

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.1.- Flow rate and average velocity

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.2.- Mass ow rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5.3.- Flux of a scalar or vector quantity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 EQUATIONS FOR A FLUID SYSTEM


1 .- INTRODUCTION

10

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

2.1.- Surface forces versus external forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

2.2.- Stress tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10

2.3.- Stress tensor symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

11

2.4.- Hydrostatic pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12

2.5.- Viscous stress tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12

2.6.- Navier-Poisson law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12

2 .- SURFACE FORCES

iii

TABLE OF CONTENTS

2.7.- Viscosity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

2.8.- Surface forces exerted on a uid system . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3 .- HEAT CONDUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3.1.- Heat conduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3.2.- Heat ux vector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3.3.- Fourier law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

14

3.4.- Thermal conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15

3.5.- Heat ux in a uid system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15

4 .- MECHANICAL EQUATIONS FOR A FLUID SYSTEM . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15

4.1.- Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

15

4.2.- Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

16

4.3.- Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

16

4 EQUATIONS FOR A CONTROL VOLUME

18

1 .- SYSTEMS AND CONTROL VOLUMES

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18

2 .- REYNOLDS TRANSPORT THEOREM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18

3 .- UNIFORM APPROXIMATION FOR THE FLUX TERM . . . . . . . . . . . . . . . . . . . . . . . . . . . .

19

4 .- CONTINUITY EQUATION FOR A CONTROL VOLUME

. . . . . . . . . . . . . . . . . . . . . . . . . . .

20

5 .- MOMENTUM EQUATION FOR A CONTROL VOLUME

. . . . . . . . . . . . . . . . . . . . . . . . . . .

20

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

22

6.1.- Bernoulli and hydrostatics equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

7 .- ENERGY EQUATION FOR A CONTROL VOLUME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

23

8 .- ENERGY EQUATION FOR A FLUID MACHINE

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

24

8.1.- Application to the adiabatic case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

25

9 .- ENERGY EQUATION FOR A HYDRAULIC MACHINE . . . . . . . . . . . . . . . . . . . . . . . . . . . .

25

6 .- BERNOULLI EQUATION

5 DIMENSIONAL ANALYSIS

27

1 .- INTRODUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27

1.1.- Some denitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

27

1.2.- Dimensional homogeneity principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

28

2 .- BUCKINGHAM THEOREM

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29

2.1.- Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29

2.2.- Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

29

3 .- PHYSICAL SIMILARITY

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

30

3.1.- Physical similarity. Dierent types of similarity . . . . . . . . . . . . . . . . . . . . . . . .

30

3.2.- Fluid-dynamic models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

3.3.- Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

3.4.- Partial similarity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

32

iv

TABLE OF CONTENTS

4 .- IMPORTANT DIMENSIONLESS GROUPS

II

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Applications

32

35

6 HYDROSTATICS

37

1 .- INTRODUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37

2 .- REDUCTION OF A SYSTEM OF FORCES IN HYDROSTATICS . . . . . . . . . . . . . . . . . . . . . . . .

37

2.1.- Equivalent system of parallel forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37

2.2.- Application to hydrostatics

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

38

3 .- FORCES AND TORQUES ON A FLAT SURFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

39

3.1.- Force exerted on a at surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

39

3.2.- Central axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

39

4 .- FORCES AND TORQUES ON SUBMERGED BODIES . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

40

4.1.- Forces exerted on a submerged body . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

40

4.2.- Central axis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41

7 ESSENTIALS OF FLUID DYNAMICS


1 .- INTRODUCTION

42

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 .- THE TURBULENCE PHENOMENON


3 .- BOUNDARY LAYER

42

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

42

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

43

4 .- BOUNDARY LAYER SEPARATION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

45

4.1.- Form drag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

45

4.2.- Boundary layer separation and form drag . . . . . . . . . . . . . . . . . . . . . . . . . . .

46

4.3.- Origin of the BL separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

47

8 HYDRAULICS

49

1 .- INTRODUCTION

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

49

2 .- FLOW IN PIPES

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

49

2.1.- Loss of reduced pressure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

49

2.2.- Equivalent roughness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

50

2.3.- The Moody chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51

3 .- LOCAL LOSSES

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51

3.1.- Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51

3.2.- Formulation of the problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

52

3.3.- Dimensional analysis of the problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

52

3.4.- Drop of reduced pressure in pipes with connections . . . . . . . . . . . . . . . . . . . . . .

53

3.5.- Flow meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

53

4 .- MULTIPLE- PIPE SYSTEMS

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
v

54

TABLE OF CONTENTS

4.1.- Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

54

4.2.- Pipes in series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

54

4.3.- Pipes in parallel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

4.4.- Pipe networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

4.5.- Pipeline with uniform draw-o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

55

5 .- MULTIPLE- PIPE SYSTEMS WITH PUMPS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

56

5.1.- Head supplied by the pump . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

56

5.2.- Pumps in series

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

5.3.- Pumps in parallel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

9 OPEN CHANNELS, WEIRS AND SLUICEGATES


1 .- INTRODUCTION

60

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

60

2 .- UNIFORM FLOW IN OPEN CHANNELS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

60

3 .- WEIRS AND SLUICEGATES

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

61

3.1.- Weirs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

61

3.2.- Sluicegates

61

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vi

Part I

Essentials of Fluid Mechanics

Chapter 1

INTRODUCTION

1 .- GENERAL COMMENTS
Fluid mechanics studies the behavior of uids both at rest (hydrostatics) and in motion (hydrodynamics). In particular,
one is interested in the uid-solid interaction occurring when the uid surrounds the solid body (aerodynamics), or it
is inside the solid element (hydraulics).
Two routes can be followed to analyze a uid-mechanical problem: the theoretical and experimental ones. The
theoretical analysis is necessarily restricted to very simple problems, or to those for which global results (resultant
forces, average pressures, ow rates, etc.) are to be obtained. In more complex phenomena, it is necessary to resort
to experimentation in order to get useful information of the system considered. Experiments, guided by dimensional
analysis, provide data for designing uid machinery.
Recently, numerical methods have become a third route to analyze uid-dynamical problems based on the huge
capability of current computers. They can provide reliable results in situations of technological interest. The simulation
of uid-dynamical problems essentially reduces to the numerical integration of the motion equations with certain initial
and boundary conditions. This integration allows one to obtain detailed information on the uid behavior. The use
of this kind of technique is beyond the scope of the present introductory subject.

1.1.- Fluids
Fluid denition
Normal and shear stresses are the forces per unit area exerted on a surface in the directions normal and tangential to
that surface, respectively (see gure 1.1). Fluid is the state of matter which deforms continuously under the action
of shear stresses. In other words, any shear stress causes a dynamical deformation of the uid, which continues until
that stress disappears. Therefore, shear stresses must vanish in hydrostatics. The above denition distinguishes uids
from solids. Solids show a static deformation when shear stresses are applied to them.

Gases y liquids
Fluids are divided into gases and liquids. Gases easily compress/expand due to variations of pressure, while liquids
are almost incompressible. By way of illustration, consider the following example: one must double the air pressure
to double the density (for a constant temperature), while the water pressure must be raised 100 times to increase the
density by just 0.5%.
Density variations associated with temperature changes may be noticeable in some cases. For instance, the water
density decreases by 0.5-1% (for a constant pressure) if the temperature increases about 30 C. In most of the problems
considered in this subject, we shall not observe signicant temperature variations. Therefore, the liquid density will
be assumed to be constant.
3

INTRODUCTION

normal stresses

shear stresses

fluid element

fluid element

fluid element

Figure 1.1: Normal and shear stresses.

1.2.- Continuum hypothesis


Fluids consist of molecules moving in a vacuum. If the number of molecules is high enough at any postion and
r
instant t, then the uid can be described as a continuum medium whose behavior is characterized from its density ,
velocity , pressure p, and temperature T , which depend on the position and time t. In this case, we shall consider
v
r
that the uid is made of elements or particles, which are very small (innitesimal) portions of this medium.
This description is possible in almost all applications. For instance, there are of the order of 107 molecules in
10 mm3 of air under pressure and temperature standard conditions. Only in exceptional situations, the continuum
hypothesis does not verify (for example, in nanouidics of gases or around spacecrafts entering the atmosphere). In
this case, uid mechanics is replaced with other disciplines (kinetic theory, for instance) that take into account the
uid microscopic structure.
9

Chapter 2

KINEMATICS

1 .- INTRODUCTION
The full resolution of a uid-dynamical problem involves the calculation of the uid density, velocity, pressure, and
temperature at any times and for each position. In other words, one has to obtain the scalar and vector elds (, t),
r
(, t), p(, t), and T (, t). In general, the velocity eld (, t) is a three-dimensional vector function that depends
v r
r
r
v r
on both the spatial coordinates and time. It provides the velocity of a uid element (particle) located at a certain
position at an instant t. In Cartesian coordinates, the velocity eld reads:
r
= u(x, y, z; t) + v(x, y, z; t) + w(x, y, z; t) .
v
i
j
k

(2.1)

In this chapter, the function (, t) is assumed to be known, and some tools will be developed to analyze it.
v r

2 .- LAGRAGIAN

AND

EULERIAN DESCRIPTIONS

In the Lagragian description, attention is paid to the evolution of uid particles. One is interested in the variation
of the hydrodynamic quantities (density, presure, . . . ) of a given uid element as a function of time. In the Eulerian
description, one focuses on the temporal evolution of the hydrodynamic quantities at a xed point of the region
occupied by the uid. In this case, one does not monitorize the evolution uid particles which crossed that point.
In order to clarify these denitions, we shall resort to a simile from our everyday experience. Suppose that trac on
a highway can be compared with a uid motion. Cars are the uid particles in this simile. The Lagragian description
would entail to pay attention to a certain car, observing how its velocity changes. On the contrary, the Eulerian
description involves to focus on a given segment of the highway, watching the variation of the number of cars, average
velocity, etc. As will be seen throughout this course, the Eulerian description is the technique most frequently adopted
in uid mechanics.

3 .- STREAMLINE

AND

PATH

3.1.- Streamline
Streamlines are those tangential to the velocity eld at each point of the uid domain for a given instant. In other
words, the lines that verify the dierential equations
dx1
dx2
dx3
=
=
,
v1 ()
r
v2 ()
r
v3 ()
r

(2.2)

where v1 , v2 , and v3 are the components of the velocity vector along the orthogonal axes x1 , x2 , and x3 , respectively1 .
v
The integration of equations (2.2) provides two relationships which dene the streamline passing through a given point.
1 Hereafter, the axes x , x , and x will refer to any (Cartesian, cylindrical, or spherical) orthogonal system of coordinates. The unit
1
2
3
vectors 1 , 2 , and 3 indicate the corresponding directions.
e e
e

KINEMATICS

In Cartesian coordinates, equation (2.2) reads


dx
dy
dz
=
=
,
u()
r
v()
r
w()
r

(2.3)

where u, v, and w are the components of the velocity vector along the axes x, y, and z, respectively. In cylindrical
v
coordinates, the equations dening the streamline are:
dr
d
dz
=r
=
,
vr
v
vz

(2.4)

where vr , v , and vz are the components of the velocity vector along the axes r, , and z, respectively. Finally, they
v
are written in spherical coordinates as
dr
d
d
=r
= r sen
,
(2.5)
vr
v
v
where vr , v , and v are the components of the velocity vector along the axes r, , and , respectively.
v
Two streamlines can pass across each other only at singular points where the uid velocity is zero (stagnation
point) or innite (sink or source). Otherwise, the velocity vector would be tangential to two lines simultaneously.
A streamtube is a surface in the shape of a tube which is made up by streamlines (see gure 2.1). A common
example is any imaginary cylinder coaxially located inside a pipe. No matter which ow is, the uid contained inside
a streamtube at a given instant moves without getting out of it at that instant. Otherwise, two streamlines would
cross at the point where the uid gets out of the streamtube.

Figure 2.1: Streamtube.

3.2.- Path
A path is the trajectory followed by a uid particle over time. To get its parameter (as a function of the parameter
t) equations, one has just to integrate the equations of motion, i.e.,
dx1
= v1 (x1 , x2 , x3 ; t) ,
dt

dx2
= v2 (x1 , x2 , x3 ; t) ,
dt

dx3
= v3 (x1 , x2 , x3 ; t) .
dt

(2.6)

In gure 2.2, we sketch two streamlines at an instant t, and the path traced by a uid particle over time.

4 .- TYPES

OF

FLOWS

4.1.- Steady ow
In a steady ow, the velocity eld does not depend on time at any point, i.e., = (). It can be readily seen that
v v r
streamlines and paths coincide in steady ows. In unsteady ows, they can coincide or not. In addition, in a steady
ow the uid contained in a streamtube moves without getting out of it over time.
6

4 .- TYPES

OF

FLOWS

t=t0

t=t3

Streamlines

t=t1

t=t2
Paths

Figure 2.2: Stream line and path.

It must be noted that uid particles can undergo accelerations in steady ows. In fact, if a uid particle moves
according to an inhomogeneous velocity eld, its velocity will change (acceleration) despite the fact that the velocity
vector remains constant at each point (steady ow). For instance, in the steady ow taking place in a converging nozzle
(see gure 2.3), the uid particle speeds up as approaches the nozzle outlet because the magnitude of the velocity eld
is larger in that region.

t=t1

t=t2
x

Figure 2.3: Acceleration of a uid particle in a nozzle.

4.2.- Uniform ows and streams


In a uniform ow, the velocity eld does not depend on the spatial coordinates, i.e.,, = (t). If the ow is steady
v v
too, it is called uniform stream.

4.3.- Two- and one-dimensional ows


A two-dimensional ow is referred to as that whose velocity eld has only two non-zero Cartesian components, and
those components depend on their corresponding coordinates (and time) exclusively; for instance,
= u(x, y; t) + v(x, y; t) .
v
i
j

(2.7)

In a one-dimensional ow, these conditions are met for one of the velocity components.

4.4.- Axisymmetric ows


An axisymmetric ow is a three-dimensional ow which possesses a symmetry axis, so that one angular coordinate
and the corresponding velocity component do not take part in the problem.
In cylindrical coordinates, the velocity eld of an axisymmetric ow can be written as
= vr (r, z; t) r + vz (r, z; t) z ,
v
u
u

(2.8)

where vr and vz are the radial and axial velocity components, respectively (v = 0 is the angular component), and z
is the symmetry axis. In spherical coordinates, the velocity eld is given by the expression
= vr (r, ; t) r + v (r, ; t) ,
v
u
u
7

(2.9)

KINEMATICS

where vr and v are the radial and polar velocity components, respectively (v = 0 is the azimuthal component). In
this case, = 0 is the symmetry axis of the problem.
The classical examples of axisymmetric ows are the uniform streams that strike solid bodies with an axis of
revolution which coincides with that of the uniform stream. To describe this class of ows, either cylindrical or
spherical systems of coordinates are used depending on the solid body shape. For instance, a cylindrical system of
coordinates is used to describe the ow around a cylinder immersed in a uniform stream (see gure 2.4 (left)), while
the ow around a sphere is analyzed in terms of a spherical system of coordinates (see gure 2.4 (right)2 ).

symmetry axis

symmetry axis

u
Figure 2.4: Axisymmetric ow around a cylinder (left) and a sphere (right).

4.5.- Incompressible ows


In an incompressible ow, the density of any uid particle remains constant over time. Any liquid motion, as well as
those gas ows driven by relatively low-pressure gradients, can be regarded as incompressible.

5 .- FLOW RATE

AND

MASS FLOW

RATE

5.1.- Flow rate and average velocity


The ow rate Q is the volume of uid crossing a surface S per unit time. To calculate the ow rate from the velocity
eld, we shall consider a dierential surface element of area dS, and the volume dV built from it as shown in gure
2.5. The volume of uid that passes through dS within a period of time dt equals dV . This last quantity can be
calculated as
dV = dS cos v dt ,
(2.10)
where is the angle formed by the unit vector normal to the surface element and the velocity vector at that point.
n
Also, v is the magnitude of that velocity vector. Then, one can obtain the ow rate Q crossing S from the integration
of (2.10):

dV
Q=
=
dS =
v n
dS .
v
(2.11)
S dt
S
S

n
v

dV

dS

v dt
Figure 2.5: Flow rate that crosses a surface element.

In addition, the average velocity vm at which the uid crosses the surface S is dened as:
vm =
2 In

the gures, vectors are represented in bold letters.

Q
.
S

(2.12)

5 .- FLOW RATE

AND

MASS FLOW

RATE

5.2.- Mass ow rate


The mass ow rate m is the mass of uid that passes through a surface S per unit time. The mass ow rate is
calculated analogously to the ow rate. Let us consider, for instance, the uid element in gure 2.5. The mass dm of
this element is obtained from the expression
dm = dV = dS cos v dt ,

(2.13)

where the meaning of these variables is that established in the previous section. The volume dV of this uid element
equals the volume of uid that crosses dS within the time interval dt. Therefore, the mass transported across dS per
unit time is
dm
= dS .
v n
(2.14)
dt
The mass ow rate m that crosses the entire surface S is calculated from the integral

m =
dS .
v n
(2.15)
S

If the uid is incompressible (liquid), the density is constant and can be brought out of the integral. In this case,
the following relation also holds:
m = Q .
(2.16)
5.3.- Flux of a scalar or vector quantity
Let B be any scalar or vector quantity used to describe the ow. Let dB() be its value for a uid element located
r
at the position . Suppose that dB() is proportional to the mass dm of the uid element. The quantity () is the
r
r
r
amount of B per unit mass, i.e., = dB/dm.
The quantity dB is transported by the uid element owing to its motion. Consider, for instance, the uid element
shown in gure 2.5. The value dB corresponding to this element can be obtained from the equation
dB = dm = dV = dS cos v dt ,

(2.17)

where the meaning of these variables is that established in previous sections. The volume dV of this uid element is
the same as the volume of uid that crosses dS within the time interval dt. Therefore, the amount of B transported
across dS per unit time is
dB
= ( ) dS .
v n
(2.18)
dt
The ux of the quantity B across the surface S is calculated from the integral

=
( ) dS .
v n
(2.19)
S

The ux represents the amount of B transported by the uid across S per unit time. The expressions for the
ow rate Q and the mass ow rate m can be regarded as two particular cases of equation (2.19). Thus, = Q when
B = V (uid volume) or = 1/ (specic volume), and = m when B = M (uid mass) or = 1.
If the surface S encloses a volume V and the vector is dened outwards, then > 0 implies that the amount
n
of B contained in V decreases over time; while < 0 implies the opposite. We will come back to this point when the
Reynolds transport theorem is presented.
To close this chapter, we will introduce Einsteins notation for repeated subindexes, which we will resort to later
on. Two repeated subindexes indicate a sum over those subindexes. For example, aj bj = a1 b1 + a2 b2 + a3 b3 , or
ai aj
xj xi

=
+

a1 a1
a1 a2
a1 a3
a2 a1
a2 a2
a2 a3
+
+
+
+
+
x1 x1
x2 x1
x3 x1
x1 x2
x2 x2
x3 x2
a3 a2
a3 a3
a3 a1
+
+
.
x1 x3
x2 x3
x3 x3

(2.20)

As can be seen, Einsteins notation allows one to simplify enormously equations involving vectors and/or matrixes.
9

Chapter 3

EQUATIONS FOR A FLUID SYSTEM

1 .- INTRODUCTION
A uid system is a given mass (portion) of uid. A uid system is delimited by a uid surface, which separates it
from the surroundings. The uid surface consists of elements of uid surface that move with the uid velocity at the
point they are.
The momentum and energy of a uid system can vary over time. This variation has two origins. On one side,
the exchange of momentum and energy with the surroundings across the uid surface. On the other side, the external sources (external forces and sources of energy). In the course of this chapter, we shall calculate the variations
corresponding to the rst case.

System

Surroundings
Figure 3.1: Fluid-surroundings interaction.

2 .- SURFACE FORCES
2.1.- Surface forces versus external forces
As mentioned above, the forces acting on a uid system have two origins. First, the external forces exerted on the
uid system. Gravity and forces exerted by solid elements in contact with the system are the most frequent cases.
Second, the surroundings exerts through the uid surface forces on the system. These forces are generically called
surface forces. The surface forces are calculated from the stress tensor, which we shall focus on immediately.

2.2.- Stress tensor


Consider a uid surface element located at a point 0, and that separates two uid regions [(1) and (2)] (see gure 3.2).

This element is characterized by the vector dS0 , normal to the surface, with a magnitude equal to the element area,
and pointing from (2) towards (1). It is possible to demonstrate (we will not do it here) that the component i of the

force dF0 exerted by (1) on (2) through the surface element is given by the expression
dF0i = ij (0 )dS0j ,
r
10

(3.1)

2 .- SURFACE FORCES

where ij is the stress tensor, which depends only on the position 0 where the surface element is located. Einsteins
r
notation1 was introduced in equation (3.1). This equation can also be written as


dF01
11
dF02 = 21
dF03
31

12
22
32

13
dS01
23 dS02 .
33
dS03

(2)

r0

(3.2)

(1)
(2)
(1)

Figure 3.2: Surface forces.

2.3.- Stress tensor symmetry


As shown above, surface forces are calculated in terms of the stress tensor. In this section, we shall demonstrate that
this tensor is symmetric. For this purpose, let us consider the uid element shown in gure 3.3. The torque dM0 with
respect to the axis 0 exerted by the surface forces is
dM0 = 2(21 x2 x3 )

x1
x2
2(12 x1 x3 )
.
2
2

(3.3)

By denition, dM0 is positive if produces a rotation as indicated in the gure, and negative otherwise. The rst
addend on the right side of equation (3.3) corresponds to the contribution to dM0 of the front and back faces of the
uid element, while the second reects the contribution of the two lateral faces. The external forces must be applied
in the uid element center, and thus they do not contribute to the torque dM0 . Therefore,
I

d
= dM0 ,
dt

(3.4)

where I is the moment of inertia of the uid element.


The uid element volume scales as 3 , being a characteristic length of the uid element (for instance, x1 ). The
distance of an element portion to the axis 0 scales as , and thus the moment of inertia I scales as 4 . However, the
torque dM0 scales as 3 [see equation (3.3)]. Therefore, equation (3.4) can not be satised for an innitesimal uid
element ( 0) unless
dM0 = 0 .
(3.5)
From equations (3.3) and (3.5), one gets
12 = 21 .

(3.6)

It is easy to verify that if one had chosen the axis 0 as that parallel to 1 or 2 , then one would have concluded that
e
e
23 = 32

or

13 = 31 ,

respectively. In this way, it is demonstrated that the stress tensor is symmetric.


1

Let as remember that two repeated subindexes indicate a sum over those subindexes. For instance, aj bj = a1 b1 + a2 b2 + a3 b3

11

(3.7)

EQUATIONS FOR A FLUID SYSTEM

e3

0
e2

e1

x3
x1
x2

Figure 3.3: Torque exerted by the surface forces with respect to an axis.

2.4.- Hydrostatic pressure

We have just shown that the component i of the force dF0 exerted by a uid region on another across the surface
element dS0 that separates them (see gure 3.2a) is
dF0i = ij (0 )dS0j ,
r

(3.8)

where ij (0 ) is the stress tensor that depends only on the position 0 where the surface element is located.
r
r
As explained in chapter 1, the dening feature of uids is the fact that they deform continuously under the action
of shear stresses. Therefore, shear stresses must vanish in uids at rest. In other words, surface forces must be
perpendicular to the surfaces in hydrostatics. In this case,

dF0 = p(0 ) dS0 ,


r

(3.9)

where the scalar quantity p(0 ) is called the hydrostatic pressure2 . This equation is equivalent to
r
ij (0 ) = p(0 ) ij ,
r
r

(3.10)

ij being the identity tensor.


The hydrostatics equation (3.9) is referred to as Pascals principle. Pascals principle establishes that forces in
hydrostatics are perpendicular to the surface elements, and their magnitudes are independent of the orientation of
those surface elements. We will show in chapter 6 how to calculate the hydrostatic pressure eld p(0 ).
r
2.5.- Viscous stress tensor
Equation (3.10) provides the value of the stress tensor for a uid at rest. It is natural to assume that its value in a
ow is given by (3.10) plus a contribution ij associated with the uid motion:
ij () = p()ij + ij () .
r
r
r

(3.11)

This contribution is due to viscous friction caused by the displacement of some uid layers with respect to others. For
this reason, ij is called the viscous stress tensor. For uids at rest or inviscid uids, ij = 0, and ij reduces to the
stress tensor (3.10).
2.6.- Navier-Poisson law
In this section, we shall introduce an empirical law which allows one to calculate the viscous stress tensor ij from
the velocity eld. To simplify the analysis, we will focus on the particular (and most common) case of incompressible
ow. This condition holds for liquids always, and for gases as long as pressure (density) variations are small enough.
2 Taking into account how the vector dS is dened (see gure 3.2a), it is obvious that the negative sign in equation (3.9) ensures that
0
p always takes a positive value.

12

2 .- SURFACE FORCES

As mentioned in the previous section, viscous forces are caused by the displacement of some uid layers with
respect to others. The strain rate tensor
(
)
1 vi
vj
ij =
+
(3.12)
2 xj
xi
is a measure of the displacement magnitude. In this regard, it is natural to hypothesize that ij is proportional to ij .
Symmetry considerations (not included here) allow one to assert that, for incompressible ows,
ij = 2 ij ,

(3.13)

where is the uid viscosity . Its value depends on the uid as well as its pressure and temperature at the considered
point. Equation (3.13) is referred to as the Navier-Poisson law. In a Cartesian system of coordinates, the elements of
the viscous stress tensor (3.13) are:
(
)
u
u v
,
xy =
+
,
xz = zx ,
xx = 2
x
y
x
yx = xy ,
(
zx =

yy

u w
+
z
x

vr
,
r

zy = yz ,
(

r =

(
zr =

vr
vz
+
r
z

zz = 2

1 vr
v
v
+

r
r
r
(

r = r ,

yz =

v w
+
z
y

In a cylindrical system of coordinates, they are:


rr = 2

v
= 2
,
y

= 2

1 v
vr
+
r
r

w
.
z

)
,

rz = zr ,

(
,

z =

v
1 vz
+
z
r

)
,

z = z ,

(3.14)

zz = 2

)
,

(3.15)

vz
.
z

Liquids commonly used in industrial applications (water, oils, . . . ) obey the Navier-Poisson law. However, there
some substances for which the linear relationship between the viscous stress and strain rate tensors is no longer valid.
Among those substances, one can mention plastic, dilatational, or pseudoplastic liquids. Rheology is the discipline
devoted to the analysis of such liquids. Newtonian uids are those which verify the Navier-Poisson (Newton) law,
while non-Newtonian liquids are those whose behavior can not be described by that law. In this course, we shall
consider Newtonian uids exclusively.
2.7.- Viscosity
Viscosity can play an essential role in the dynamics of uids. Its inuence can be measured through the ratio /,
where is the uid density. This ratio indicates the relation between the viscous force (proportional to the viscosity
) and inertia or the uid resistance to acceleration (proportional to the density ). Therefore, is an indicator of the
eects of the viscous force on the uid dynamics. Because its units are kinematics (m2 /s), it is called the kinematic
viscosity. In order to better distinguish the two viscosities, the quantity is often referred to as dynamic viscosity.
As mentioned above, both the dynamic and kinematic viscosities depend on the uid considered. There is a large
range of viscosity values. For instance, for standard pressure and temperature conditions,
glicerine 1400water 80000air

and

glicerine 80air 1400water .

(3.16)

The dynamic viscosity could also depend on the uid pressure and temperature. However, variations of pressure have
negligible inuence, while temperature changes may produce considerable eects. The dependence of the viscosity
upon temperature is very dierent in gases and liquids. In fact, viscosity increases with temperature in gases, while
the opposite occurs in liquids.
13

EQUATIONS FOR A FLUID SYSTEM

2.8.- Surface forces exerted on a uid system


We have introduced the stress tensor to calculate the force exerted by a uid region on another through the uid
surface element that separates them from each other. Consider a uid system S enclosed by the system (uid) surface

SS. The component i of the resultant surface force Fs exerted by the surroundings on the uid system S through SS is

Fsi =
ij dSj =
ij nj dS ,
(3.17)
SS

SS

where (as always) the vectors dS and are dened outwards. If the stress tensor is split into the pressure and
n
viscous contributions, one gets
Fsi = Fpi + Fvi ,
(3.18)

where

Fpi =

p ij nj dS =
SS

p ni dS

(3.19)

SS

and
Fvi =

ij nj dS

(3.20)

SS

are the components i of the pressure and viscosity forces exerted on the system, respectively.

3 .- HEAT CONDUCTION
3.1.- Heat conduction
The energy transfer due to heat can take place by means of three mechanisms: convection, radiation, and conduction.
Convection is the transfer of energy between two regions due to the uid motion from one to the other. It is not an
exchange of energy between the uid system and its surroundings, but the spatial transport of energy associated with
the uid displacement. Therefore, it must not be considered in this chapter. In radiation, energy coming from external
sources is transmitted by electromagnetic waves (photons), crossing the surroundings and reaching the uid system,
where it is partially absorbed. Therefore, it must not be considered here either. Heat conduction is the transfer of
energy between the two sides of a uid surface (the uid system and its surroundings) because they are at dierent
temperatures. This phenomenon does correspond to the scope of this chapter, and will be analyzed in the next section.
3.2.- Heat ux vector
Consider a uid surface element located at a point 0, and that separates two uid regions [(1) and (2)] (see gure

3.4). This element is characterized by the vector dS0 . This vector is normal to the surface, has a magnitude equal to
the element area, and points from (2) towards (1). It is possible to demonstrate (we will not do it here) that the heat

transfer of energy per unit time dq0 between (1) and (2) through the surface element is given by the expression

dq0 = (0 ) dS0 ,
q r

(3.21)

where is heat ux vector, which depends only on the position 0 where the surface element is located. By denition,
q
r

the quantity dq0 is positive when the region (1) transfers energy to the region (2), and negative otherwise. Taking

into account the direction of dS0 (see gure), it is obvious that indicates the direction in which energy ows.
q
3.3.- Fourier law
Heat conduction is caused by the temperature gradient in the uid, so that energy ows from the hotter to the colder
region. It is natural to assume that is a vector proportional to and with the opposite direction of the temperature
q

gradient T ; i.e.,

= T ,
q
(3.22)
14

4 .- MECHANICAL EQUATIONS

FOR A

FLUID SYSTEM

d
(2)

dq'

(1)
(2)
(1)

Figure 3.4: Heat ux.

where is the thermal conductivity, which depends on the uid considered as well as its pressure and temperature.
Equation (3.22) is the Fourier law.
3.4.- Thermal conductivity
As happens with the viscosity value, pressure variations have a slight inuence on the thermal conductivity, while
temperature changes may produce a signicant conductivity variation. The dependence of the conductivity with
respect to the temperature value is similar to that of the viscosity: it increases with temperature in gases, whilst the
opposite occurs in liquids.
3.5.- Heat ux in a uid system
We have introduced the heat ux vector to calculate the heat exchange between two uid regions across the surface
element that separates them. Consider a uid system S delimited by a system (uid) surface SS. The energy per unit
time exchanged by S with the surroundings across SS due to heat conduction is

=
Q =
dS
q
qj nj dS ,
(3.23)
SS

SS

where the vectors dS are point outwards and, therefore, Q > 0 if the system gains energy, and Q < 0 otherwise.
n

4 .- MECHANICAL EQUATIONS

FOR A

FLUID SYSTEM

The momentum and energy of a uid system change over time owing to the action of both external factors and the
surroundings. In this chapter, we have analyzed the role played by the surroundings. We present in this section the
equations for the evolution of the mechanical quantities characterizing a uid system from the results obtained above.
4.1.- Mass
The mass m of a uid system can be calculated as

m=

() dV ,
r

(3.24)

VS

where VS is the volume occupied by the uid system at a given instant. As mentioned previously, this quantity
remains constant over time. Therefore, the conservation equation

d
() dV = 0
r
(3.25)
dt VS
15

EQUATIONS FOR A FLUID SYSTEM

always holds.
4.2.- Momentum

The momentum P of a uid system,

P =

() () dV ,
r v r

(3.26)

VS

can vary over time because of the action of external Fext and surface Fs forces. Taking into account the expressions
(3.18)(3.20), the momentum equation for a uid system reads

() () dV = Fext
r v r
p() dS +
r n
() dS ,
r n
(3.27)
dt VS
SS
SS
where SS is the uid surface that encloses the system. This equation corresponds to Newtons second law for a uid
system.
4.3.- Energy
In uid mechanics, the total energy per unit mass e is dened as the sum of the thermal energy and the mechanical one;
specically, the sum of the internal energy associated with temperature, the kinetic energy due to the uid motion,
and the potential energy owing to the gravitational eld:
e = u + 1 v 2 + gz ,
2

(3.28)

where g is the gravitational acceleration, and z the upward axis.


The total energy E accumulated in a uid system,

E=
() e() dV ,
r
r

(3.29)

VS

evolves because of the following two factors: (a) the work done by external forces, (b) heat transferred by external
sources, (c) work done by surface forces, and (d) heat conduction. The rst two factors correspond to external sources,

while the last two ones origin from the interaction with the surroundings. The symbols Wext and Q stand for the
ext
energy transfer per unit time due to the rst and second factors, respectively.
The uid layer in contact with an impermeable solid wall moves with the same velocity as that of the wall. This
is the so-called non-slip boundary condition, and it is the result of the mechanical equilibrium established almost
instantaneously between the layer and the wall. Any mechanical system exchanges energy due to work only if moves
while exerting or withstanding a force. Therefore, the uid system can only exchange energy due to work with moving
solid elements (blades, propellers, pistons, . . . ) in contact with it. Solid walls at rest (pipes, valves, . . . ) exert and

withstand forces but can not exchange energy with the uid system due to work. On the other side, the term Wext
must not include the action of gravity. The potential energy associated with the gravitational eld is included in the
denition of total energy e. Therefore, gravity does not alter the total energy value, it only transforms a kind of energy
into another.
The internal energy quanties both the kinetic energy associated with the thermal motion of the uid molecules,
and the potential energy accumulated in the intermolecular interactions. It must not be included here the chemical
energy associated with the intramolecular bonds. In an exothermic chemical reaction (combustion), part of the energy
contained in the intramolecular bonds is released and absorbed by a uid system as internal or kinetic energy. In
this regard, combustion modies the total uid energy e, and must be considered as an external source (Q ) in the
ext
energy equation, although it is physically originated in the uid itself.
Taking into the above comments, the evolution of the uid system energy is given by the equation

() e() dV = Q
r
r
() dS
q r
p() () dS
r v r n
ext
dt VS
SS
SS

Wext +
() [ () ] dS .
v r
r n
SS

16

(3.30)

4 .- MECHANICAL EQUATIONS

FOR A

FLUID SYSTEM

Equation (3.30) can be regarded as the generalization of the rst law of thermodynamics for a moving uid system. It
includes both the mechanical terms and viscous energy dissipation. The sign convention used in this equation coincides
with that of the rst law of thermodynamics for the corresponding terms: the heat exchange is positive if the system
gains energy, and negative otherwise; the work done by the surroundings is positive when the uid system compresses

( < 0), and negative otherwise. Because of the minus sign in front of Wext , Wext < 0 if energy is transmitted to
v n

the system and Wext > 0 otherwise.

17

Chapter 4
EQUATIONS FOR A CONTROL VOLUME

1 .- SYSTEMS

AND

CONTROL VOLUMES

In the previous chapter, we obtained the laws of mechanics for a uid system. In most cases, one is not interested
in the evolution of uid systems, but in the local eects that those systems cause when crossing a certain region (for
instance, a machine).
To understand this twofold perspective, we must distinguish uid systems from control volumes. Let us remain
that a uid system is a certain part (or portion) of the uid under consideration. This part is separated from the
surroundings by a uid surface. A uid system can move, rotate, and deform over time. Each element of the uid
surface moves with the same velocity as that of the uid at the point where the element is.
A control volume (CV) is a xed spatial region which a uid system crosses, while the control surface (CS) is the
one delimiting that volume (see gure 4.1). CVs are used to enclose hydraulic machines, mobile solid bodies, hydraulic
devices, etc. Some considerations to appropriately choose CVs are:
Use must be made of a frame of reference solidly moving with the CV, so that this volume remains at rest with
respect to that frame of reference.
Fluid systems must completely ll CVs. Therefore, the CV and CS coincide with the volume and surface,
respectively, of the uid system contained in that CV.
In some applications, CVs contain uid systems made up by two phases (for instance, water and air) (see gures
4.4 or 4.6).
Control
volume

System 1

System 2

Figure 4.1: Systems and CVs.

Equations (3.25), (3.27), and (3.30) describe the evolution of mechanical quantities referred to a uid system. As
will be shown in the next section, the Reynolds transport theorem allows one to get the equations for CVs from their
counterparts for the uid systems contained in those volumes. Here, this theorem will be formulated in mathematical
terms, and applied to uid mechanics. Its demonstration is beyond the scope of this introductory subject.

2 .- REYNOLDS TRANSPORT THEOREM


Let V (t) and V (t) be two arbitrary volumes, and S(t) and S (t) the corresponding limiting surfaces (see gure 4.2).
Assume that these two volumes merge at a given instant t = t0 , i.e., V (t0 ) = V (t0 ) and S(t0 ) = S (t0 ). Let (, t) be
r
18

3 .- UNIFORM APPROXIMATION

FOR THE

FLUX TERM

a scalar or vector function dened over the volume V (t) (or V (t)). The Reynolds transport theorem establishes that,
for t = t0 ,

d
d
(, t) dV =
r
(, t) dV +
r
(, t0 ) [r (, t0 ) ] dS ,
r
v r
n
(4.1)
dt V (t)
dt V (t)
S(t0 )
where r (, t0 ) is the relative velocity of the S (t0 ) element with respect to that of the S(t0 ) element, both located at
v r
the point . In addition, is the outwards unit vector normal to the surface S(t0 ) (or S (t0 )).
r
n
t=t
0

V'

r0
S'
S

Figure 4.2: Reynolds transport theorem.

Equation (4.1) has a clear interpretation as applied to uid mechanics. Let B be a scalar or vector mechanical

quantity whose evolution in a CV we want to calculate (for instance, momentum P ). Let (, t) be the amount of B
r

per unit mass at the point and instant t (for example, (, t)). Suppose that V corresponds to the system volume
r
v r
(SV) while V is the CV which contains it at the instant t = t0 . In this case, S and S are the system surface (SS)
and the CS, respectively. Because CV is a xed volume, the relative velocity r is simply the uid velocity at the
v
v
point under consideration. In addition, one can identify the variable (, t) with the product (, t) (, t). Taking
r
r
r
into account the above considerations, the Reynolds transport theorem for the quantity B leads to

d
d
dV =
dV +
( ) dS .
v n
(4.2)
dt SV
dt CV
CS
The left side of this equation represents the variation per unit time of B referred to the uid system. The rst addend
of the right side of the equation corresponds to the variation per unit time of the amount of B contained in the
CV, while the second is the ux (equation (2.19)) of this quantity through CS. Therefore, the Reynolds transport
theorem establishes that the variation of B in the CV equals the sum of the variation associated with the contained
uid system plus the dierence between the amount of B entering and getting out of the CV (the ux).
The equations for the conservation of mass, momentum, and energy in a CV will be obtained in this chapter.
To this end, we shall consider equation (4.2) as applied to the appropiate values of . Besides, we shall account for
equations (3.25), (3.27), and (3.30) for a uid system.

3 .- UNIFORM APPROXIMATION

FOR THE

FLUX TERM

In many applications, the uid under consideration crosses the CS only through at inlet and outlet sections. In this
case, only those sections contribute to the ux . Very often, the ow can be regarded as uniform over those sections
(see gure 4.3). In this case, the ux can be calculated as:

( ) dS =
v n
k k (k k ) Ak =
v n
k m ,
(4.3)
k
CS

where k , k , and k stand for the values of the corresponding quantities for the section k. Also, k and Ak are the
v
n
unit vector and the area of section k, respectively. In the above equation, m is the magnitude of the mass ow rate
k
going through the surface k. The positive sign applies if the uid gets out of the CV across that surface, and the
negative one otherwise.
If the CS is chosen so that the inlet and outlet sections are perpendicular to the ow (see gure 4.3), then equation
(4.3) can be written as

k k vk Ak ,
(4.4)
( ) dS =
v n
CS

19

EQUATIONS FOR A CONTROL VOLUME

2
3

CS
1

Figure 4.3: Uniform approximation for the ux term.

where the positive and negative signs apply to the outlet and inlet sections, respectively.
Figure 4.3 represents, for instance, the ow inside a hydraulic machine. The CV is chosen so that CS coincides
with the machine walls. As can be seen, the uid crosses CS through three at sections, where the ow is uniform
and perpendicular to CS. In this case, the ux can be calculated as

( ) dS = 1 1 v1 A1 + 2 2 v2 A2 + 3 3 v3 A3 .
v n
(4.5)
CS

4 .- CONTINUITY EQUATION

FOR A

CONTROL VOLUME

In this section, we will obtain the equations which describe the temporal evolution of the uid mass contained in a
CV. Consider equation (4.2) with = 1. Accounting for equation (3.25) for the mass of a uid system, one gets the
expression

d
dV =
( ) dS ,
v n
(4.6)
dt CV
CS
which is referred to as the continuity equation for a CV.
If the mass contained in a CV remains constant over time (steady regime), then

( ) dS = 0 .
v n

(4.7)

CS

By using the approximation of uniform ow at the inlet and outlet sections [equation (4.3)], the above expression can
be written as

k (k k ) Ak =
v n
m = 0 ,
(4.8)
k
k

where m is the (magnitude of the) mass ow rate that crosses section k. The positive sign applies if the uid gets
k
out of CV through section k, and the negative sign otherwise. If the ow is incompressible and there are only two
(the inlet and outlet) sections, then
Qinl = Qout ,
(4.9)
where Qinl and Qout are the (magnitude of the) inlet and outlet ow rates, respectively.

5 .- MOMENTUM EQUATION

FOR A

CONTROL VOLUME

In this section, we will obtain the equations which describe the temporal evolution of the momentum contained in a
CV. Consider equation (4.2) with = . Assuming the second Newtons law (3.27) for a uid system, one gets the
v
vector equation

d
ext
dV +
v
( ) dS ,
v v n
(4.10)
F
p dS +
n
dS =
n
dt CV
CS
CS
CS
20

5 .- MOMENTUM EQUATION

FOR A

CONTROL VOLUME

where the fact that SS and CS coincide has been taken into account. The term associated with viscous stresses can
be neglected in most practical applications because those stresses are much smaller than the hydrostatic pressure over
the inlet and outlet sections of CS. It must be noted that viscous friction between solid walls and the uid system do
not correspond to that term because those walls are not inlet or outlet sections. On the contrary, that viscous friction

must be included in the term Fext .


If the momentum contained in CV remains constant over time (steady regime), then

Fext + Fp =
( ) dS ,
v v n

(4.11)

CS

where Fp denotes the force exerted by the surroundings on the uid system due to pressure.
By using the approximation of uniform ow at the inlet and outlet sections [equation (4.3)], the above expression
can be written as

(4.12)
Fext + Fp =
m k ,
kv
k

where the positive sign applies if the uid gets out of CV through section k, and the negative sign otherwise. In this
equation, m denotes the (magnitude of the) mass ow rate crossing section k. If the ow is incompressible,
k

Qk k ,
v
(4.13)
Fext + Fp =
k

Qk being the (magnitude of the) ow rate crossing section k.


In spite of the important approximations leading to equation (4.13), this expression is used in many practical
applications. If there are only two CS sections crossed by the uid (the inlet and outlet), then

Fext + Fp = Q(out inl ) ,


v
v

(4.14)

where Q is the (magnitude of the) ow rate crossing CV, and out and inl are the uid velocities at the outlet and
v
v
inlet sections, respectively.
External forces are typically gravity and those exerted by solid elements in contact with the uid system contained

in CV. The latter are the unknowns in most problems. In addition, if the pressure is constant over CS, then Fp =
0.
The demonstration of this assertion is easy:

Fp =
p dS = p
n
dS = .
n
0
(4.15)
CS

CS

In the problem sketched in Fig. 4.4, a water jet strikes the screen located in a moving cart. One wants to calculate
the force F driving the cart motion. The CV used to analyze this problem is that represented in the gure, and the
uid system is made up by both the water and air contained in that volume. In this way, CV and the system volume
coincide. The frame of reference must move solidly with the cart, so that CV remains still for that frame of reference.
The equation for the horizontal component of momentum allows one to calculate the force exerted by the screen on the
uid system (water and air) in the CV. That force, with the opposite sign, is the reaction exerted by the uid system
(water and air) on the screen, which drives the cart motion. In this example, both the water jet and the surrounding

air are at the (constant) atmospheric pressure, and thus Fp =


0.
CS

p=pat

Figure 4.4: Force exerted by a water jet on a screen.

21

EQUATIONS FOR A CONTROL VOLUME

6 .- BERNOULLI EQUATION
Consider a streamtube element of innitesimal length ds in a steady ow. As shown in chapter 2, the uid moves
within the streamtube without getting out of it (see gure 4.5). Suppose that the ow is parallel to the direction s
perpendicular to the inlet and outlet sections, and uniform over those sections. We will use the symbol A to denote
the inlet section area, and , v, p, and m to refer to the density, velocity, pressure, and mass ow rate at that section,
respectively. These quantities take the values A + dA, + d, v + dv, p + dp, and m + dm at the outlet section,
respectively. The quantities dA, d, dv, dp, and dm are innitesimals because the distance between the inlet and
outlet section is innitesimal as well.
A+dA

p+dp/2

+d

p+dp v+dv m'+dm'

p v m'
z

ds

Figure 4.5: Innitesimal streamtube element.

To analyze this ow, one chooses a CV which matches the streamtube element, and neglects second-order innitesimals. In this case, the continuity equation is
dm = 0 ,
(4.16)
and the projection along the axis s of the momentum equation becomes
dFext + dFp + dFv = m dv + vdm .

(4.17)

In this equation, dFext and dFp are the components along the s axis of the external and pressure forces, respectively.
Also, dFv is the s-component of the force associated with the viscous stresses applied onto CS.
In absence of solid elements, dFext is the result of the gravitational eld:
dFext = Adsgs = Adzgz = Adzg .

(4.18)

To calculate the pressure force, one has to consider its action on the inlet and outlet sections, and on the lateral
surface. The pressure on the lateral surface can be calculated as the average of those of the inlet and outlet sections.
1
The projection along the axis s of the pressure force on the lateral surface is (p + 2 dp)dA. Therefore, the resultant
pressure force is
(4.19)
dFp = pA + (p + 1 dp)dA (p + dp)(A + dA) = dpA .
2
Taking into account (4.18)(4.19), the momentum equation (4.17) yields
dp
dFv
+ v dv + g dz
ds = 0 ,

dm

(4.20)

where dm = A ds is the mass of the streamtube element.


Any streamtube is made up by an innite number of elements as that sketched in gure 4.5. One can relate the
properties of the streamtube beginning (section 1) to those of the end (section 2) by integrating the two sides of
equation (4.20) along the axis s. The result is

2
1

dp 1 2
2
+ 2 (v2 v1 ) + g(z2 z1 ) + wv = 0 ,

22

(4.21)

7 .- ENERGY EQUATION

FOR A

CONTROL VOLUME

where wv is the work per unit mass done by the viscous force dFv along the tube, i.e.,
2
dFv
ds .
wv =
1 dm

(4.22)

Because the viscosity force opposes the uid motion, dFv < 0 and wv > 0 always. The energy wv dissipated by viscous
stresses increases with the distance between 1 and 2.
In deducing equation (4.21), use has been made of the uniform ow approximation at the spanwise sections. This
approximation always holds for a streamtube innitely thin (streamline). Equation (4.21) is usually referred to as the
generalized Bernoulli equation, and is valid for any streamline of a steady ow. However, its usefulness is very limited
due to the diculties associated with the calculation of the viscous term (4.22).
In a steady ow, the generalized Bernoulli equation for a streamline (path) describes the evolution of the mechanical
energy associated with a uid particle over its motion. This mechanical energy is the sum of the pressure, potential,
and kinetic energies. The viscous dissipation term wv represents the loss of mechanical energy between 1 and 2 due
to the work done by the friction force exerted by the surroundings on the uid particle.

6.1.- Bernoulli and hydrostatics equations


If the ow is incompressible, equation (4.21) reduces to
p1
v2
p2
v2
+ 1 + z1 =
+ 2 + z2 + hp ,
g 2g
g 2g

(4.23)

where all the addends were divided by g for them to have length dimension. The sum p/g + v 2 /2g + z is called the
head. The term hp = wv /g represents the head loss between 1 and 2 due to viscous friction.
If the uid is not very viscous, hp 0, equation (4.23) becomes
p1
v2
p2
v2
+ 1 + z1 =
+ 2 + z2 ,
g 2g
g 2g

(4.24)

known as the Bernoulli equation.


In hydrostatics, the kinetic energy terms can be neglected, which leads to the fundamental equation of hydrostatics
p1 + gz1 = p2 + gz2 ,

(4.25)

applicable to whichever the points be.

7 .- ENERGY EQUATION

FOR A

CONTROL VOLUME

Consider equation (4.2) with = e. Taking into account the energy conservation equation (3.30) for a uid system,
one gets the expression

e dV +
e ( ) dS ,
v n
(4.26)
Q
dS
q n
p dS Wext +
v n
( ) dS =
v
n
ext
dt CV
CS
CS
CS
CS
where we have accounted for that CV and SV coincide.
If the energy contained in CV is constant, then

Q
p dS Wsol Wv =
v n
CS

e ( ) dS .
v n

(4.27)

CS

In this equation, Q stands for the total energy exchanged due to heat (conduction plus external sources) per unit

time; Wsol is the power transferred through the moving solid elements in contact with the uid system; and Wv is the
23

EQUATIONS FOR A CONTROL VOLUME

power associated with viscous stresses exerted on CS. One must remember that Q > 0 means an increase of energy

in CV, while W > 0 means the opposite. The work done by gravity must not be included in Wext because it does not
alter the total energy value (it only transforms potential energy into other types of energy or vice versa).

It is important to point out that only moving solid elements contribute to Wsol . The uid layer in contact with
a solid surface remains attached to that surface (non-slip boundary condition), and thus the solid and uid velocities
coincide over that surface. Therefore, if the solid wall remains at rest, the uid layer will be still as well, and exchange

of energy due to work will not take place. The viscous term Wv can be neglected in many applications. The reason
lies in the fact that viscous stresses exerted on CS are much smaller than pressure.

To simplify equation (4.27), the term due to the pressure force is moved to the right-side of that equation. Taking
into account that h = u + p/, equation (4.27) becomes

)
(

v n
(4.28)
Q Wsol =
h + 1 v 2 + gz ( ) dS .
2
CS

This equation can be interpreted in the following way. The energy exchanged with the external sources (plus conduction) is the dierence between the hydraulic energy transported by the uid at the inlet and outlet sections of the CV.
This hydraulic energy is the sum of the total energy e and that associated with the hydrostatic pressure p/. In this
way, one eliminates from the analysis the work done by the surroundings.
By considering the uniform ow approximation at the inlet and outlet sections [equation (4.3)], and neglecting the
dierence between the heights of those two sections, the above equation can be written as
(
)

m hk + 1 vk + gzk ,
(4.29)
Q Wsol =
k
2
k

where the positive sign applies if the uid gets out of CV through section k, and the negative sign otherwise. In this
equation, m , hk , vk , and zk are the mass ow rate, enthalpy, velocity, and height of section k, respectively.
k
If the ow is incompressible,

Q Wsol =

(
)
1 2
Qk hk + 2 vk + gzk ,

(4.30)

Qk being the (magnitude of) ow rate crossing section k.

8 .- ENERGY EQUATION

FOR A

FLUID MACHINE

In this section, we will examine the energy balance taking place in a machine which processes a uid. One can
distinguish two working regimes. At the initial stage, there is always a transient regime through which the machine
adapts to the working conditions. This rst stage is usually very short, and gives rise to a steady regime in which the
amount ofmass, momentum, and energy contained in the machine remains constant on time. In this last stage, the
term d/dt CV in the CV equations vanishes. In this section, this condition will be assumed.
Let us give an example to illustrate the above comment. When a pump is turned on, the propeller starts rotating,
and the water inside the machine begings moving. The momentum and energy contained inside the pump increase
sharply. After a short transient regime, the propeller rotation speed reaches a constant value, and thus the momentum
and energy of the water inside the pump no longer vary.
A CV enclosing the machine is used to study the energy balance in that device. Suppose that uid gets in and
gets out of CV across two surfaces normal to the ow (see gure 4.6), i.e., the sections of the inlet and outlet pipes.
The ow is assumed to be uniform over these sections. The surroundings exerts viscous stresses on the uid system
only through the inlet and outlet sections. As mentioned previously, and apart from exceptional situations, this eect

is negligible, and thus Wv can be neglected in the energy equation.


Under the above conditions, the continuity equation establishes that the mass ow rate m at the inlet and outlet
sections coincide. The energy equation reads
(
)
2
2

(4.31)
Q Wsol = m ho hi + 1 vo 1 vi + gzo gzi ,
2
2
24

9 .- ENERGY EQUATION

FOR A

HYDRAULIC MACHINE

machine
CV
Figure 4.6: Sketch of a uid machine.

where the subindexes i and o refer to the inlet and outlet values, respectively.
If all the terms of equation (4.31) are divided by m , one gets the expression
2
2
hi + 1 vi + gzi = ho + 1 vo + gzo + wsol q ,
2
2

(4.32)

where wsol = Wsol /m is the work done by the moving solid elements, and q = Q /m is the total heat exchanged by
the uid system, both per unit mass.
1
The sum h+ 2 v 2 +gz is the stagnation (total) enthalpy h0 per unit mass, and thus equation (4.32) can be written as

h0i = h0o + wsol q .

(4.33)

It must be pointed out that equations (4.32) and (4.33) are valid only if the machine is working in the steady
mode, the ow is uniform over the inlet and outlet sections, and the power associated with viscous stresses on those
sections is negligible. Despite these restrictions, equations (4.32) and (4.33) are valid in the majority of applications,
regardless the shape, size, or working principle of the considered machine. This degree of generality confers a great
relevance on this equation.

8.1.- Application to the adiabatic case


Most of the ows taking place inside uid machines are so fast that heat exchange mechanisms can be neglected
(adiabatic ow, q = 0), and thus equation (4.33) reads
h0 = wsol ,

(4.34)

where h0 is the variation of stagnation enthalpy taking place between the inlet and outlet sections. In pumps and
fans, wsol < 0 and h0 > 0, while the opposite occurs in turbines. The usefulness of the above equation is enormous
in industrial engineering.

9 .- ENERGY EQUATION

FOR A

HYDRAULIC MACHINE

The energy equation (4.32) can be re-written as


pi
po
2
2
+ 1 vi + gzi =
+ 1 vo + gzo + wsol + uo ui q ,

2
2
i
o

(4.35)

where use has been made of the relationship between internal energy and enthalpy h = u + p/. By denition, the

ow in a hydraulic machine is incompressible (i = o = ), and equation (4.35) reduces to


v2
po
v2
pi
+ i + zi + HB =
+ o + zo ,
g 2g
g 2g
where

(4.36)

1
HB = (wsol + uo ui q) .

(4.37)

25

EQUATIONS FOR A CONTROL VOLUME

It must be mentioned again that wsol and q are the energies per unit mass that the system exchanges with external
factors in the shape of work and heat, respectively. The quantity HB is the head added to or subtracted from the uid
by the hydraulic machine. In pumps and fans, wsol < 0 and HB > 0, while the opposite occurs in turbines.
Suppose now that the ow inside a hydraulic machine is almost inviscid. In this case, energy dissipation can be
neglected, and thus there is no internal heat production mechanism. Therefore, all the energy absorbed (or released)
by the uid system in the shape of heat is included in the quantity q (energy exchange with outside). The rst law of
thermodynamics establishes for an incompressible ow that the increase (or decrease) of the internal energy of a uid
particle is the same as the heat absorbed (or released) by that uid particle. Therefore, uo ui q = 0 in the inviscid

and incompressible case, and hence


W
W
wsol
= sol = sol .
(4.38)
HB =
g
m g
gQ
In other words, the energy wsol transmitted (or absorbed) by a machine in almost inviscid and incompressible ows
translates into an increase (or decrease) of head. On the contrary, if viscous eects were not negligible, there would
be an energy dissipation, us ue q, which would cause a decrease of the head supplied to the uid. This decrease

is quantied by the hydraulic performance h = 1 (s ue q)/wsol .


u

26

Chapter 5

DIMENSIONAL ANALYSIS

1 .- INTRODUCTION
Due to the complexity of most uid-dynamic problems of technological interest, one has to rule out in many occasions
any theoretical approach. This is so not only in applications whose complexity is apparent, but also in those looking
simple. Therefore, experimentation constitutes most of the time the only route to get useful information. Experimentation is generally very expensive, and thus it must be conducted in a rationale manner, obtaining the maximum
information possible from the minimum number of experiments.
In a broad sense, dimensional analysis is the set of techniques, rules, and procedures used to get information on
a certain phenomenon just from the knowledge of the quantities (variables) involved in that phenomenon. The steps
typically followed in a dimensional analysis can be arranged as:
1. Following physical arguments, one proposes the dimensional quantities involved in the problem. Contrarily to
what one may think, this rst step may become really dicult, because requires having a certain background in
the discipline the problem belongs to.
2. Following a certain procedure, the relevant dimensionless numbers are constructed.
3. Taking into account the relevant dimensionless numbers, one draws up an experiment with a prototype or a
model.
In this chapter, we shall explain in detail these ideas.

1.1.- Some denitions


Before starting, we must make sure that some denitions are correctly understood.
Fundamental dimensions: they are the basic dimensions from which the quantities describing a phenomenon are
dened. The number of fundamental dimensions is two (length [L] and time [T ]) if the problem is kinematic,
three (length, time, and mass [M ]) if it is dynamical, and four (length, time, mass, and temperature [T ]) if it
is thermo-dynamical. Systems of units (for instance, the International Unit System) are used to measure the
fundamental dimensions and their combinations.
Dimensional variables: they are variables measured in terms of fundamental dimensions (density, pressure,
viscosity, . . . ). Therefore, constant or dimensionless parameters such as angles, for instance, are not included
in this category. Naturally, the value taken by a dimensional variable depends on the system of units used to
quantify the fundamental dimensions; for example, water = 103 kg/m3 and 1 gr/cm3 .
Dimensionless group: they are dimensionless monomials formed by dimensional variables. A monomial is the
product of several variables raised to certain powers. The Reynolds number Re= vL/ is a very important
example. Because dimensionless groups have no dimension, their values do not depend on the system of units
used to to quantify the fundamental dimensions.
27

DIMENSIONAL ANALYSIS

1.2.- Dimensional homogeneity principle


The dimensional homogeneity principle establishes that any equation relating dimensional variables must be dimensionally homogeneous; i.e., all the addends involved in such relation must have the same fundamental dimensions. If
an equation is dimensionally homogenous, then it is invariant under a change of the system of units, which constitutes
a necessary condition for that equation to have physical meaning. By way of illustration, let us consider the Bernoulli
equation (4.25):
p1
v2
p2
v2
+ 1 + z1 =
+ 2 + z2 .
(5.1)
g 2g
g 2g
It can be readily veried that all the addends of this equation have dimensions of length.
Based on the dimensional homogeneity principle, one can perform an elemental dimensional analysis of simple
problems. Consider the problem aiming at measuring the pendulum period for small-amplitude oscillations. Firstly,
assume that the dimensional variables intervening in this phenomenon are the period , mass m, and length of the
pendulum, as well as the gravitational acceleration g. We have allowed ourselves to make an error here, because
does not depend on m. Secondly, one can expect that these variables are related through an equation of the form
= C ma b g c ,

(5.2)

where the factor C and the exponents a, b, and c are unknown dimensionless constants. The dimensional homogeneity
principle forces the two sides of the equation to have the same dimensions. It can be easily veried that this condition
is equivalent to the following system of equations for the exponents:
a =
b+c =
2c =

0
0
1,

(5.3)

whose solution is a = 0, b = 1/2, and c = 1/2. The rst, second, and third equalities must hold for the dimensions
of mass, length, and time, respectively, to be the same on both sides of equation (5.2). Taking into account (5.2) and
the solution of (5.3), one concludes that obeys the expression

=C
.
(5.4)
g
As can be seen, this elemental analysis, simply based on the dimensional homogeneity principle, shows that the mass
m must be ruled out of the pendulum formula, thus correcting our intentional error1 . Finally, it is necessary to perform
experiments to conrm hypothesis (5.2), and to determine the value of C. In gure 5.1, the values of the dimensionless
1
group /(/g) 2 are plotted for several experiments. The experimental results conrm hypothesis (5.2) and allows one
to calculate the C value.

8
1/2

/(l/g)

C=2

4
tests ( ,m,l,g)

Figure 5.1: Oscillation period of a pendulum for dierent experiments.

The kind of analysis presented here is applicable only to simple problems, for which a law of type (5.2) is valid. For
the rest of cases, it would be desirable to have a systematic and general method at ones disposal. As will be shown
immediately, the Buckingham theorem provides the appropriate procedure.
1 In fact, a dimensional quantity can intervene in a phenomenon only if its fundamental dimensions appear in at least another quantity
of the problem.

28

2 .- BUCKINGHAM THEOREM

2 .- BUCKINGHAM THEOREM
2.1.- Formulation
Suppose that a certain phenomenon can be described in terms of the n dimensional variables a1 , a2 , . . . , an :
f (a1 , a2 , . . . , an ) = 0 .

(5.5)

Assume that these variables involve the m fundamental dimensions d1 , d2 , . . . , dm , and that the dimensions [a1 ],
[a2 ], . . .,[an ] of the n dimensional variables are given by2
[a1 ] =
[a2 ] =
.
.
.
[an ] =

d11 d12 . . . d1m


m
1
2
d21 d22 . . . d2m
m
1
2
dn1 dn2 . . . dnm .
m
1
2

(5.6)

The Buckingham theorem establishes that it is possible to form n j, and not more than n j, independent
dimensionless groups3 from the n dimensional variables, j being the rank of the matrix

11 12 . . . 1m
21 22 . . . 2m

.
(5.7)

n1 n2 . . . nm
In addition, the theorem establishes that equation (5.5) is equivalent to
g(1 , 2 , . . . , nj ) = 0 ,

(5.8)

where 1 , 2 , . . . , nj are the dimensionless groups mentioned above. The natural number j is the reduction of the
problem and, obviously, is equal to or smaller than the number m of fundamental dimensions (columns of (5.7)).

2.2.- Application
The application of the Buckingham theorem to any problem consists of the following steps:
1. To propose the n dimensional variables involved in a phenomenon
2. To form the matrix (5.7) from the fundamental dimensions of those variables
3. To calculate the rank j of that matrix
4. To obtain the n j independent dimensionless numbers. There are always alternatives to obtain these groups.
It is convenient to choose those which can be easily recognized by other experimenters. We shall mention some
of them in section 4.
The equation which describes the considered problem involves these dimensionless groups.
For the sake of illustration, let us consider again the measurement of the pendulum oscillation period. Suppose
again that the dimensional quantities involved in this problem are the period , mass m, and length of the pendulum,
instance, if a1 = , [a1 ] = d1 d3 , d1 and d2 being the mass [M ] and length [L] fundamental dimensions, respectively. In this case,
2
11 = 1, 12 = 3, and 13 = 14 = ... = 1m = 0.
3 Independent dimensionless groups are those which can not be written as a function of the others.
2 For

29

DIMENSIONAL ANALYSIS

as well as the gravitational acceleration g. The involved fundamental dimensions are [M ], [L], and [T ]. The matrix of
type (5.7) for this problem is

[L]

0
1

1
0

[M ]
1
0
0
0

[T ]

0
0
.
2
1

(5.9)

The determinant of the submatrix formed by the rst three rows and columns of (5.9) is dierent from zero. Therefore,
the matrix rank is j = 3. The only dimensionless group of the problem is

.
1
/g

(5.10)

Consequently, the theorem establishes that the pendulum period veries

=C,
/g

(5.11)
1

C being a dimensionless constant. Again, gure 5.1 can represent the values of the dimensionless group /(/g) 2 for
dierent experiments. In this case, the experimental results allow one to determine the C value: C = 2.
Contrarily to what occurred in the previous analysis, here we did not formulate any (5.2)-type hypothesis; we only
had to propose the dimensional variables involved in the phenomenon. This simple example also shows how the
theorem allows one to discard variables which do not intervene in a certain physical process. Hovever, the validity
of the hypothesis made in step 1 determines the validity of the nal result most of the time. The experimenter will
propose correctly the relevant dimensional variables only if he/she possesses sucient knowledge on the topic.

3 .- PHYSICAL SIMILARITY
As mentioned in the introduction of this chapter, the last step in the dimensional analysis of a phenomenon is to
design an appropriate experiment with a prototype (real conditions) or by using a physically similar model. In this
section, concepts such as similarity and model will be dened, and their usefulness in experimentation will be claried.

3.1.- Physical similarity. Dierent types of similarity


Consider two phenomena [(1) and (2)] which involve the same dimensional quantities. The theorem allows one to
obtain the dimensionless groups in both cases. Because the dimensional variables are the same in both phenomena,
so are the dimensionless groups. Let as denote by
(1)

(1)

(1)

1 , 2 , . . . , k

and

(2)

(2)

(2)

1 , 2 , . . . , k

(5.12)

the values taken by the dimensionless numbers in the phenomena (1) and (2), respectively. Phenomena (1) and (2)
are physically similar if
(1)
(2)
i = i
for i = 1, 2, . . . , k ;
(5.13)
i.e., the dimensionless groups take the same values in both phenomena.
The geometrical conguration necessarily intervenes in the phenomenon considered. Therefore, all the relevant
distances of the problem must be included in the set of governing dimensional variables. Thus, the sets of dimensionless
numbers (5.12) involve all the independent ratios which can be obtained from those relevant distances. If all those
dimensionless groups take the same values in (1) and (2), those two phenomena are geometrically similar. In other
words, two phenomena are geometrically similar if their geometries dier just in a global scale. Geometrical similarity
is a necessary but not sucient condition for physical similarity.
30

3 .- PHYSICAL SIMILARITY

L /L =l /l
3

L /L =l /l

L /L =l /l
4

l
l

L /L =l /l
5

Figure 5.2: Geometrical similarity.

By way of illustration, gure 5.2 shows two problems geometrically similar. A uid stream crosses the duct formed
between two horizontal walls whose width (the dimension perpendicular to the paper) is much larger than any other
distance of the problem. A sphere is located inside that duct. This geometrical conguration can be characterized by
ve distances. The four independent ratios of those distances take the same values in the two problems.
One frequently assumes that the geometrical conguration of a certain phenomenon is xed, i.e., it can not be
varied in the experiments. In this case, two phenomena are physically similar if, provided that they are geometrically
similar, condition (5.13) is veried for the purely physical dimensionless groups, i.e., excluding the geometrical ones.

3.2.- Fluid-dynamic models


Consider two physical phenomena [(1) and (2)] which include the same dimensional variables. Suppose that those two
phenomena are geometrically similar. Let (5.12) be the two sets of values taken by the purely physical dimensionless
groups,. The theorem allows one to write any of them, 1 , as a function of the rest of k 1 groups:
1 = h(2 , 3 , . . . , k ) .

(5.14)

Assume that the function h is the same for the two phenomena. In this case, if the k 1 arguments of the function h
took the same values at (1) and (2), then
(2)
(1)
(5.15)
1 = 1 ,
and the two phenomena would be physically similar. This is the result which experimentation with models is based
on.
Suppose that we aim at measuring a variable of a certain phenomenon [phenomenon (1)]. Assume that, given the
experimental conditions, that measurement is not feasible, and thus we design a viable experiment [phenomenon (2)]
geometrically similar to the original one. Phenomenon (1) is known as the prototype, while phenomenon (2) is referred
to as the model. Let us suppose that the variable whose value we want to measure belongs to the dimensionless group
1 . The model must be designed so that the k 1 resting dimensionless groups take the same values in the prototype
and its model. In this way, equation (5.15) veries, and the unknown value can be inferred.

3.3.- Example
Consider the incompressible ow around a smooth sphere of diameter D immersed in an innite uniform stream of
density , viscosity , and upstream speed V . Simple symmetry arguments allow one to conclude that the uid
exerts a force on the sphere along the stream direction. Let FD be the magnitude of this drag force. Our objective
is to measure FD as a function of the above mentioned variables. It must be noted that, because of the inniteness
of the uid stream and the symmetry of the sphere, the only geometrical variable intervening in the problem is D.
The fundamental dimensions involved in this problem are the length [L], mass [M ], and time [T ]. The matrix (5.7)
corresponding to this case is

31

DIMENSIONAL ANALYSIS

V
D

FD

[L]

[M ]

[T ]

1
1
3
1
1

0
0
1
1
1

1
0
0
1
2

(5.16)

The determinant of the submatrix formed by the three rst rows and columns of (5.16) is dierent from zero. Therefore,
the rank of the matrix is j = 3. The two relevant dimensionless groups of the problem are
1

FD
V 2 D2

and

V D
.

(5.17)

The group 1 is frequently replaced with FD /(1/2 V 2 A), where A = D2 /4 is the sphere cross section area4 . The
group 2 is the Reynolds number Re, which appears in all problems where viscosity plays a signicant role.
The theorem establishes that the dimensional variables of this problem can be related as
FD
1
V 2 A
2

= CD (Re) ,

(5.18)

where CD is the drag (aerodynamic) coecient, which is to be determined experimentally.


Suppose now that the experimental conditions are not appropriate to measure the drag force FD . To overcome this
obstacle, we measure the drag force exerted on a model of diameter D by a uniform stream of density , viscosity

, and moving at a speed V . Let us denote by FD the result of this measurement. As explained above, if the model

Reynolds number Re coincides with the corresponding value Re in the prototype, then the model and the prototype
are physically similar, and thus equation (5.18) yields
FD
= CD (Re) = CD (Re ) =
1
2
2 V A
which leads to

FD = FD

FD
,
1
2
2 V A

V 2 D2
.
V 2 D2

(5.19)

(5.20)

Figure 5.3 shows the drag coecient measured for a wide range of Reynolds numbers. The symbols correspond to
data obtained from dierent models. As can be observed, the function CD (Re) is the same for all the models.
3.4.- Partial similarity
A numerous set of dimensionless groups comes up in studying problems with many dimensional variables. In this case,
it may become technically impossible to equalize the values of all the dimensionless numbers characterizing the model
and the prototype. Then, it is indispensable to neglect the inuence of some of them. In this case, one says that
there is partial similarity between the model and the prototype, and the results for the former must be extrapolated
with caution to the latter. Partial similarity is common in technological problems; for instance, in the experimental
analysis of the ow inside hydraulic machines.

4 .- IMPORTANT DIMENSIONLESS GROUPS


There is a set of dimensionless groups which appear frequently in uid mechanics. Here, we just mention some of
them:
4 The

cross section area is the area of the body viewed along the stream direction.

32

4 .- IMPORTANT DIMENSIONLESS GROUPS

CD=24/Re

Figure 5.3: Sphere drag coecient as a function of the Reynolds number.

Reynolds number:
Re =

U L
,

(5.21)

where and are the uid density and viscosity, respectively, while U and L are characteristic velocity and
length of the ow, respectively. This group appears whenever the viscous friction be relevant.
Euler number:
Eu =

p
,
U 2

(5.22)

where is the uid density, and p and U are characteristic pressure and velocity of the ow, respectively. This
group appears in Eulerian ows, where the pressure contribution to the surface force is more important than the
viscous one.
Froude number:
Fr =

U2
,
gL

(5.23)

where U and L are characteristic velocity and length of the ow, respectively, and g is the gravitational acceleration. This number intervenes in ows driven by gravity (open channels, weirs, and sluicegates).
Mach number:

U
,
(5.24)
a
where U is the characteristic velocity of the ow, and a is the sound velocity. This parameter appears continuously
in compressible ows, and indicates the inuence of the uid compressibility on the ow behavior.
M=

Prandtl number:

cp
,
(5.25)

where , cp , and are the uid viscosity, specic heat constant, and thermal conductivity, respectively. This
group intervenes in ows where both viscous and conduction mechanisms are important.
Pr =

33

Part II

Applications

Chapter 6

HYDROSTATICS

1 .- INTRODUCTION
We will analyze in this chapter the behavior of a liquid at rest. In this case, the problem reduces to calculating the
pressure distribution p() in the liquid bath. In this way, one can obtain the forces and torques exerted on submerged
r
solid bodies. The pressure distribution is calculated from the fundamental equation of hydrostatics (4.25):
p1 + gz1 = p2 + gz2 ,

(6.1)

where 1 and 2 are two arbitrary points of the liquid domain. Because the velocity eld vanishes, so do viscous stresses.

Therefore, the surface force dF is

dF = p()dS .
r
(6.2)
The forces and torques exerted on fully or partially submerged solid bodies (damps, sluicegates, oating bodies, etc.)
will be calculated from equations (6.1) and (6.2).
p

p(z)

H
z

Figure 6.1: Pressure forces.

2 .- REDUCTION

OF A

SYSTEM

OF

FORCES

IN

HYDROSTATICS

2.1.- Equivalent system of parallel forces


Analyzing the eect of a system of many forces exerted at dierent points may become fairly complicated. To simplify
this analysis, one frequently resorts to an equivalent system of forces. An equivalent system of forces is that exerting
the same resultant and torque (with respect to any point) as those of the original one. Therefore, the equivalent and
original systems produce exactly the same mechanical eect on any object.
Let us now consider the particular case of parallel forces. In this case, one must distinguish between systems with
non-zero and zero resultant.

1. System with non-zero resultant. The torque MA exerted by the system of forces with respect to a certain point
A is
r

MA =
(A i ) Fi = A
r
r
r
Fi
i Fi = A F
r
r
i Fi ,
r
(6.3)
i

37

HYDROSTATICS

where {Fi } and {i } are the forces and their application points, respectively. All the vectors {Fi } have the same
r

direction. Therefore, Fi = Fin , where Fin is the component of the force Fi in that direction, and is the
n
n

corresponding unit vector. Then, the torque MA is given by the expression


(
)

A = A F

M
r
Fini .
r
n
(6.4)
i

Consider the point C dened as


C =
r

1
Fini .
r
F i

(6.5)

Taking into account this denition1 , MA is given by the expression

MA = A F F C = A F C F = (A C ) F .
r
r
n r
r
r

(6.6)

Therefore, the torque MA exerted by the system equals that exerted by the resultant F at the central point
a passing through C have the same
(6.5). It is obvious that all the points belonging to the line parallel to F
r
characteristic feature as that of the central point. This line is referred to as the central axis.
To summarize, a system equivalent to a set of parallel forces with non-zero resultant is that resultant applied at
any point of the central axis.

2. System with zero resultant. Consider the results (6.3) obtained in the previous case. Because now F = the
0,
A is given by the expression
torque M
(
)

MA =
Fini .
r
n
(6.7)
i

As can be seen, MA does not depend on the point A considered. Equation (6.7) provides the value of an intrinsic
0 of the system of forces. This intrinsic torque can be obtained with a couple. Therefore, a system of
torque M
parallel forces with zero resultant reduces to a couple with the same intrinsic torque.
2.2.- Application to hydrostatics
Consider a Cartesian system of coordinates dened by the axes x1 , x2 , and x3 . In hydrostatics, it is frequent to treat
each component separately; i.e., as if those components were independent systems of parallel forces. In this way, it is
possible to apply the above results to each component.
Consider the surface S submerged in a liquid of density . {dFj } are the components j (j = 1, 2 or 3) of the

pressure forces {dF } exerted by the liquid on the surface S. We shall denote by Fj the corresponding resultant:

Fj =
dFj = p dSj = [pa + g(H h)] dSj .
(6.8)
S

Here, the fundamental equation of hydrostatics (6.1) has been considered, h is the height of the surface element dS,
and pa is the reference pressure at the height h = H. One frequently selects as the reference pressure the ambient
(atmospheric) pressure at the free surface (see gure 6.1). In this case, H h is the depth of the considered point.
Now, we must distinguish the two cases mentioned above:

1. If Fj = 0, the equivalent system is formed by the resultant Fj = Fj j applied on the corresponding central axis.
e
Naturally, this axis is parallel to the direction j . Taking into account (6.5), the coordinates xkcp which dene
e
the position of this axis obey the expression

xk [pa + g(H h)] dSj


1
xk dFj = S
xkcp =
,
(6.9)
Fj S
[p + g(H h)] dSj
S a
where k represents any of the components dierent from j. As mentioned above, the resultant can be applied
on any point of the central axis. The selected point is called the center of pressure.
1 As can be seen, it is essential that the resultant of the system of forces does not vanish. Otherwise, the denition (6.5) does not make
any sense.

38

3 .- FORCES

AND

TORQUES

ON A

FLAT SURFACE

2. If Fj = 0, the equivalent system of forces is a couple which produces an intrinsic torque M0j . The value of this
torque is [see equation (6.7)]
(
)
(
)
0j =
M
dFj j = j
r
e
e
[pa + g(H h)] dSj .
r
(6.10)
S

If M0j = the action of the system of forces j is null.


0,
Once the three components have been reduced to their corresponding equivalent systems, the total one is obtained

by adding those three systems. We will show in the next sections that Fj and M0j vanish for two components in most
applications. In this case, pressure forces reduce to a resultant applied on the corresponding central axis.

3 .- FORCES

AND

TORQUES

ON A

FLAT SURFACE

3.1.- Force exerted on a at surface


Consider the pressure forces acting on one of the two faces of the at surface sketched in gure 6.2. They constitute a
system of parallel forces normal to that surface. We aim at calculating the resultant of this system and its application
point. For this purpose, we shall use a system of coordinates whose origin is located at the surface center of gravity.
The axes x and y are located on that surface, while z is the perpendicular axis pointing towards the pressure direction.
As can be seen in the gure, the axis y forms an angle with the horizontal axis. Taking into account (6.8), the
resultant is

Fz = [pa + gl] dS = pa A + g
l dS ,
(6.11)
S

where l is the depthindicated in the gure, and A stands for the surface area. Because the depth of the center of
gravity is lcg = 1/A S l dS, the result (6.11) can be re-written as:
Fz = (pa + glcg )A = pcg A ,

(6.12)

where pcg represents the pressure at the center of gravity.


pa
l

lcg

y
cg
x

Figure 6.2: Force exerted on a at surface.

3.2.- Central axis


The central axis can be obtained from equation (6.9). The component xcp of this axis is

1
1
xcp =
x (pa + gl) dS =
x[pa + g(lcg y sen )] dS
Fz S
Fz S
[
]

1
g sen
=
(pa + glcg ) x dS g sen xy dS =
Ixy ,
Fz
pcg
S
S

where Ixy is the moment of inertia Ixy = 1/A S yx dS. Analogously,


ycp =

g sen
Ixx ,
pcg
39

(6.13)

(6.14)

HYDROSTATICS

where Ixx represents the moment of inertia Ixx = 1/A

y 2 dS.

The results obtained in this section are very important from a practical viewpoint. They allow one to calculate
forces and torques exerted on any kind of weir, obstacle, dam, etc. The results are valid regardless of the shape of the
considered surface, its orientation, or whether it is fully or partially submerged. In this latter case, only the submerged
part must be considered.

4 .- FORCES

AND

TORQUES

ON

SUBMERGED BODIES

4.1.- Forces exerted on a submerged body


Consider the pressure forces acting on a surface S which delimits a solid body as that sketched in gure 6.3. We aim
at calculating the equivalent system of this distribution of forces. Let us rst consider the horizontal components x
and y. Taking into account (6.8), the resultants Fx and Fy obey the equations

Fx = [pa + g(H z)] dSx


and
Fy = [pa + g(H z)] dSy ,
(6.15)
S

where H is the depth of the origin of the system of coordinates. The integrand of the rst integral in (6.15) depends
only on the coordinate z. Therefore, the contribution to that integral of any surface element is canceled by the opposite
element along the x axis. For instance, dFx1 + dFx2 = 0 in gure 6.3. The same argument can be applied to the
component y. Therefore, Fx = Fy = 0.
pa

dFx2

dFx1
y
x

Figure 6.3: Horizontal forces exerted on a submerged solid body.

Now one has to calculate the intrinsic torques associated with the two horizontal components. The torque M0x
associated with the component x is
(
)
0x =
M
i
[pa + g(H z)] dSx = (z y a + g(H z)] dSx = .
r
j
k)[p
0
(6.16)
S

In the last equality, use has been made of the argument represented in gure 6.3. Analogously,
(
)

M0y =
j
[pa + g(H z)] dSy = (x z a + g(H z)] dSy = .
r
k
i)[p
0
S

(6.17)

Because both the resultants and intrinsic moments of the horizontal components are zero, the action of these components is null.
The resultant Fz obeys the expression

Fz = [pa + g(H z)] dSz = {[pa + g(H z1 )]dSz1 } + {[pa + g(H z2 )]dSz2 }
S

= g (z1 z2 ) |dSz | = gV ,

(6.18)

V being the submerged volume. To obtain this result, we have considered simultaneously an element of the upper
part and its vertical projection on the lower part (see gure 6.4).
40

4 .- FORCES

AND

TORQUES

ON

SUBMERGED BODIES

Equation (6.18) is referred to as Archimedes principle. It can be formulated this way: any body submerged in a
liquid experiences a buoyancy (upward force) equal to the weight of the amount of liquid the body displaces. It must
be noted that for Archimedes principle to be applied the body must be (partially or totally) submerged; i.e., it must
be surrounded by liquid over the entire surface S which delimits the volume V .

4.2.- Central axis


The central axis can be obtained from equation (6.9). The coordinate xcp of this axis is

g
1
x (z1 z2 ) |dSz | =
x dV .
xcp =
gV S
V V
Analogously,
ycp =

1
V

(6.19)

y dV .

(6.20)

As can be seen, the central axis is the vertical axis passing through the center of gravity of the submerged volume
(the center of pressure).

pa
dS 1
H

z1-z2

y
x
dS 2

|dSz|

Figure 6.4: Vertical force exerted on a submerged body.

If a solid body remains still under the action of the pressure and gravitational forces exclusively, the weight and
buoyancy have the same magnitude, and the center of gravity and pressure are located on the same vertical axis.

41

Chapter 7

ESSENTIALS OF FLUID DYNAMICS

1 .- INTRODUCTION
In the previous chapter, we have studied problems of hydrostatics where liquids remain at rest. However, most of
uid-mechanical processes in industrial engineering involve moving uids. The Navier-Stokes equations constitute the
mathematical tool which allows one to analyze rigorously and systematically uid-dynamical problems. Because of
their complexity, we will not consider here those equations, and will restrict ourselves to a qualitative description of
two key concepts: turbulence and boundary layer.

2 .- THE TURBULENCE PHENOMENON


In any ow, whether it be steady or time-dependent, one can observe uctuations of the hydrodynamic quantities
characterizing that ow. We here mean by uctuations the rapid and chaotic variations of those quantities. These
uctuations are initially caused by external factors, which inevitably intervene in the ow evolution modifying it
slightly.
Under certain conditions, the uctuations disappear as soon as the external factors vanish, and the ow goes back
to its original state. In this case, the uctuations must not be regarded as an essential part of the ow, but the result
of our incapacity of controlling the conditions under which the ow develops.
Under other circumstances, the uctuations do not disappear and remain as an intrinsic feature of the ow. These
uctuations may become so intense that the ow acquires a chaotic character. In this second case, uctuations are no
longer associated with external factors, but they become an inseparable part of the uid motion.
Taking into account the above experimental facts, one can distinguish two ow types. In the rst case mentioned
above, the ow is laminar, while in the second case it is turbulent. There is a blurred border between these two regimes
because the magnitude of the turbulent uctuations increases gradually as the ow undergoes a transition from the
laminar to the turbulent regimes. The convention frequently adopted is that a ow is laminar if the magnitude of
the uctuations are below 2% of the average values, while it is turbulent otherwise. The transformation of a laminar
ow into a turbulent motion usually takes place through a transient regime where outbreaks of turbulence appear and
disappear intermittently.
Figure 7.1 represents the temporal evolution of the value taken by any hydrodynamic quantity A at a xed point
of a steady [case (a)] and unsteady [case (b)] ow in both the laminar and turbulent regimes. The turbulent regime
can be clearly distinguished from the laminar one due to the existence of uctuations which remain over time. These
uctuations present a chaotic behavior which can be hardly predicted. Naturally, one can only talk about steady ow
in the turbulent regime if one refers to the average values of the hydrodynamic quantities, and not in a strict sense.
In an unsteady ow, the average value A changes over time, but the time interval within which uctuations occur
is much shorter than the characteristic evolution time of A. In this sense, turbulent uctuations can be regarded as
fast.
Laminar and turbulent ows are distinguished not only by the uctuations but also by the evolution of the average
values. The uctuations of the velocity and temperature modify substantially the transport of momentum and energy
in the uid. This factor makes the temporal and spatial variations of the average values very dierent in the two
42

3 .- BOUNDARY LAYER

laminar

laminar

turbulent
A

turbulent
A
t

(a)

(b)

Figure 7.1: Evolution of the hydrodynamic quantity A in the laminar and turbulent regimes for steady (a) and unsteady
(b) ow.

regimes. Figure 7.2 shows an important example: the ow in a cylindrical pipe. In the laminar regime, the velocity
prole is parabolic, while in the turbulent one it becomes quasi-uniform.
turbulent

laminar

Figure 7.2: Velocity proles in a pipe for the laminar and turbulent regimes.

As mentioned above, the existence of the laminar or turbulent regimes depends on the conditions under which the
ow develops. In 1883, Osborne Reynolds showed empirically that the dimensionless parameter which determines the
appearance of turbulence is the Reynolds number
Re =

LU
.

(7.1)

Let us remind here that and are the uid density and viscosity, respectively, while L and U stand for a ow
characterisitc length and velocity, respectively. These characteristic quantities must be appropriately dened in each
specic situation. The values of Re for which the transition from the laminar to the turbulent regime takes place
depends on the considered problem. A general rule might be:

laminar
0 < Re < 103
103 < Re < 104
transition
(7.2)
4
10 < Re
turbulent
In many applications, one can not dene an only Reynolds number to characterize the ow behavior. This occurs
when the characteristic length and/or velocity depend on the ow region considered. In this case, the ow can undergo
a transition from the laminar to the turbulent regime as the uid moves from one to another region. The ow over a
plate immersed in a uniform stream is a good example of this type of application.
By way of illustration, gure 7.3 shows velocity measurements in a jet steadily emitted. These measurements show
the temporal evolution of the axial velocity component for r/a = 0.67 at x/a = 0.67 [gure (a)], x/a = 1.7 [gure (b)],
x/a = 3.7 [gure (c)], x/a = 5.6 [gure (d)], and x/a = 8.0 [gure (e)]. The symbol a denotes the injector radius, while
r and x are, respectively, the radial and axial distances to the injector center. As can be observed, the magnitude of
the uctuations increases as the considered point separates from the injector. The ow next to the orice is laminar,
while it becomes turbulent for longer distances.

3 .- BOUNDARY LAYER
If the Reynolds number takes values high enough, viscous eects can be neglected at relatively short distances from
the solid walls which delimit the ow. However, there is always a region next to those walls where viscosity forces are
43

ESSENTIALS OF FLUID DYNAMICS

Figure 7.3: Fluctuations of the velocity axial component in a jet steadily emitted.

relevant. The size of that region decreases as the Reynolds number increases. In most cases, this region is conned to
a thin layer attached to the solid walls. This layer is called the boundary layer (BL), while the rest of the uid domain
is referred to as the external stream.
Figure 7.4 sketches the ow surrounding an obstacle immersed in a uniform stream. The uid in contact with the
wall remains still (non-slip boundary condition). The velocity value approaches that of the uniform stream as the
distance from the wall increases. The velocity gradient in the external stream is much lower than in the vicinity of
the solid wall. Therefore, the uid domain can be split into two regions: the BL and the external stream. The uid
viscosity plays a fundamental role in the BL, while its action can be neglected in the external stream.
x)
U(

External stream

BL

V
y

Figure 7.4: Boundary layer attached to a solid wall.

The BL size happens to be very small as compared to that of the uid domain1 for suciently high Reynolds
numbers. To characterize the BL size, use is made of the thickness of the BL. The thickness of the BL at a certain
1 As

will be shown, this assertion does not hold when the BL detaches or develops.

44

4 .- BOUNDARY LAYER SEPARATION

Figure 7.5: Separation of the BL on the upper surface of an airfoil in an air uniform stream.

point is the distance from the solid wall for which the velocity magnitude reaches 99% of the magnitude U of the
external stream velocity at that point2 (see gure 7.4).

4 .- BOUNDARY LAYER SEPARATION


Under certain conditions, the uid particles moving over a solid contour detach from that contour and suddenly move
away from the solid body. When this occurs, one can no longer identify a thin layer where viscosity eects are conned,
and the BL concept can not be applied. This phenomenon is called BL separation (detachment). The point at which
the BL detaches is referred to as the separation point. Figure 7.5 shows the separation of the BL on the upper surface
of an airfoil in a uniform stream of air.
The BL separation plays a crucial role in both aerodynamics and hydraulics. If the BL attached to a solid body
separates from it, then the drag force3 experienced by the solid body increases sharply. If the solid body was intended
to provide a lift force, then that force decreases considerably due to the BL detachment. The BL separation inside
ducts causes an extra pressure drop which must be compensated for by supplying to the uid an additional amount
of energy. When this phenomenon takes place inside a pump or fan, this pressure drop lowers the machine hydraulic
performance.

4.1.- Form drag


The force exerted by a uid stream on a solid body immersed in that stream is

Fi =

(pij + ij ) dSj =

ij dSj =
S

p dSi +

ij dSj ,

(7.3)

where S is the surface which delimits the solid body. Consider the projection of this force along the stream direction

(drag force). The contribution S p dSi associated with the pressure is called form drag, while the viscous term

dSj is named skin friction. It must be noted that the form drag is much more important than the skin friction
S ij
for blunt bodies, while the opposite occurs for very thin shapes.
2 Other
3 The

authors dene the BL thickness by using a slightly dierent percentage.


drag force is the component in the stream direction of the force exerted by that stream.

45

ESSENTIALS OF FLUID DYNAMICS

Figure 7.6: Pressure coecient for a sphere in a uniform stream.

4.2.- Boundary layer separation and form drag


As mentioned above, the drag force increases sharply if the BL attached to a solid body separates from it. In order
to understand the link between these two facts, we here consider the axisymmetric ow around a smooth sphere of
diameter D, submerged in a uniform stream of density and viscosity , moving upstream with a velocity V and
pressure p .
Figure 7.6 shows values of the pressure coecient
Cp ()

p() p
1
2
2 V

(7.4)

where p() is the pressure on the sphere surface at a point with a polar angle . If the Reynolds number Re= V D/
takes suciently high values (say, Re 103 ), the results do not depend signicantly on that parameter provided that
the laminar and turbulent regimes are distinguished. The two solid lines of the gure correspond to those two regimes,
while the dashed line is the theoretical prediction obtained from the inviscid theory.
The pressure distribution p() reaches its maximum value at the stagnation point = 0 in both the laminar and
turbulent regimes. It decreases as increases until reaching its lowest value at about = 80 . Once the minimum is
surpassed, the pressure value increases until the BL separates from the sphere contour. The BL detachment occurs
in the laminar regime for a value of smaller than in the turbulent mode. Beyond the separation point, the pressure
takes an almost constant value. In the turbulent regime, this value is greater than in the laminar one.
The form drag can be calculated from the curves shown in gure 7.6. This force is proportional to the balance
between the pressure averaged over the front and back sphere faces. There is a signicant dierence between the
46

4 .- BOUNDARY LAYER SEPARATION

Figure 7.7: Separation of the BL for a low Reynolds number.

front and back average pressures due to the BL separation. Specically, the front average pressure is greater than the
back one in both the laminar and turbulent regimes. This fact implies an increase of the form drag. Because the BL
separation takes place for a smaller value of in the laminar regime, the drag force is greater in that case. This is
why the drag coecient CD sharply decreases for Re 2 105 (see gure 5.3), where the transition from laminar to
turbulent ow takes place.
The behavior described in gure 5.3 for a smooth sphere is common to most fairing bodies (without vertexes
and edges). There are two ranges of the Reynolds number for which the drag coecient takes almost constant (and
dierent) values. Those ranges correspond to the laminar and turbulent regimes. The two values of the drag coecient
corresponding to those two regimes can be found in the literature.
In bodies with vertexes and edges, the separation of the BL occurs at those singular regions, as long as the Reynolds
number is not too small, and independently of whether the ow is laminar or turbulent. This explains, on one side, the
importance of smoothening the solid body contour to improve its aerodynamical performance, and, on the other side,
the fact that the laminar/turbulent transition does not signicantly aect the drag coecient value. This is why the
aerodynamical performance of a solid body with vertexes and edges can be characterized by just one value of the drag
coecient CD . This value constitutes a good approximation for not very small Reynolds numbers (say Re 103 ).
For the sake of illustration, gure 7.7 shows a viscous stream owing over a block. The Reynolds number for this
problem was 0.02. In spite of this low Reynolds number, the stream separates from the back of the solid body, which
produces a recirculation pattern.
4.3.- Origin of the BL separation
The BL separation can be explained in terms of the action of the pressure gradient in the stream direction. One can
distinguish two cases:
1. Favorable pressure gradient. The pressure gradient favors the uid motion. In this case, the velocity prole has
the shape indicated in gure 7.84 , and the BL remains attached to the solid contour.
y'

attached

Figure 7.8: BL with a favorable pressure gradient.

2. Adverse pressure gradient. The pressure gradient opposes the uid motion. In this case, there is an inection
point in the velocity prole4 . There are two possibilities:
4 The

demonstration of this statement is beyond the scope of this subject.

47

ESSENTIALS OF FLUID DYNAMICS

(a) Weak pressure gradient. The inection points locates next to the wall, and thus the velocity prole has the
same shape as that of the favorable pressure gradient (see gure 7.9 (left)).
(b) Strong pressure gradient. The inection point moves away from the solid contour, which produces a backow.
The uid particles moving close to the solid wall ow in the direction opposite to the main stream (see
gure 7.9 (right)), and the BL separates.
y'

y'

i.p.

i.p.

attached

detached

Figure 7.9: BL with adverse pressure gradient.

To summarize, if there is an intense adverse pressure gradient along a solid contour, the uid particle may not have
enough inertia to overcome that opposing force. In this way, the BL is deected and separated from the solid
contour, and thrown away towards the external stream. This occurs on the upper surface of an airfoil (gure 7.5),
or in the diverging part (diuser) of a nozzle (gure 7.10 (left)). In this latter case, there is an extra drop of pressure
downstream.
separation

inviscid bulk
BL

p/ x<0

p/ x>0

Figure 7.10: BL growth in a converging-diverging nozzle (left). Image of the ow inside a nozzle with BL detachment
(right).

Figure 7.10 (right) shows a photograph of the ow inside a nozzle. As can be seen, there is liquid recirculation
in the diverging part due to the BL separation. This eect entails an additional drop of pressure, which reduces the
device hydraulic performance.

48

Chapter 8

HYDRAULICS

1 .- INTRODUCTION
Hydraulics is the set of techniques and procedures for studying incompressible ows inside ducts, pipes, valves, and
other types of devices. We shall devote this chapter to the analysis of this kind of facilities, of great interest in industrial
engineering. In the rst two sections, we shall describe the ow in pipes, both in the laminar and turbulent regimes.
The third section is devoted to the analysis of the loss of pressure taking place in hydraulic devices frequently used to
connect pipes. This loss is an indirect measure of the ow rate crossing a duct, and thus many of these devices can
be used as ow meters. In the fourth section, we will examine the loss of pressure in multiple-pipe systems commonly
used in hydraulics. These systems will be classied according to their structure, and the procedures to calculate the
relevant quantities will be described. In the last section, we shall study the role played by pumps connected to a
multiple-pipe system, focusing our attention on both pumps in series and in parallel. The results presented in this
chapter are valid for incompressible ows, i.e., for gases with relatively small variations of pressure and liquids.

2 .- FLOW

IN

PIPES

2.1.- Loss of reduced pressure


In a pipe entrance, the boundary layer attached to the wall grows as the uid goes into the pipe until that layer occupies
the entire pipe section. When this occurs, the ow is said to be developed. Once the ow develops, the velocity prole
over any pipe cross section becomes parabolic or quasi-uniform depending on whether the ow is laminar or turbulent,
respectively (see gure 8.1). The ow is laminar and turbulent for ReD 2300 and ReD 4000, respectively, where
ReD = vm D/ is the Reynolds number and vm is the mean velocity over a pipe cross section. The intermediate
Reynolds numbers correspond to the laminar/turbulent transition.
If the pipe length L is much greater than its diameter D, then one can assume that the ow is developed over nearly
1/6
the entire pipe. This approximation can be performed if L 0.06ReD in the laminar regime, and if L 4.4ReD D
in the turbulent one. The developed ow inside a pipe is driven by the drop of reduced pressure (see gure 8.2)
P = p + gz

(8.1)

taking place throughout the uid motion. In the above denition, p and z are the mean values of the pressure and
height over a pipe cross section, respectively. The reduced pressure includes the eects of both pressure and gravity.
The uid losses reduced pressure due to the viscous and turbulent dissipation of energy.
The drop of reduced pressure in a pipe can be calculated from the following expression:
2
P
L vm
8f L 2
=f
=
Q ,
g
D 2g
g 2 D5

(8.2)

where P = P1 P2 is the drop of reduced pressure, P1 and P2 are the reduced pressures at the inlet and outlet
pipe sections, respectively, Q = D2 vm /4 is the ow rate crossing the pipe, and f is Darcys friction factor. The
Navier-Stokes equations allow one to demonstrate (we will not do it here) that in the laminar regime
f = 64/ReD ,
49

(8.3)

HYDRAULICS

laminar regime

Figure 8.1: Developed ow in a pipe.

laminar
turbulent

z
Figure 8.2: Drop of reduced pressure in a pipe.

and therefore

128
LQ.
(8.4)
D4
This is the Hagen-Poiseuille law. In turbulent ows, the friction factor also depends on the pipe roughness. This
factor is determined following the procedure described in the next section.
P =

2.2.- Equivalent roughness


When one carefully observes a solid body, certain outbreaks or irregularities can be appreciated over its surface.
This is usually called the material roughness. The shape and size of such irregularities are varied, and thus it is
complicated to choose a set of parameters to characterize them. For our purposes, it is necessary to determine
only the hydraulics eects of the material roughness. In order to simplify the analysis, these hydraulics eects are
characterized by a single parameter called the equivalent roughness. Here, we explain how that parameter is dened.
Nikuradse conducted experiments to analyze the inuence of the material roughness on the drop of reduced pressure
in pipes. To this end, he simulated the natural roughness by uniformly distributing sand grains of almost the same
size over the inner surface of a pipe. Figure 8.3 shows the experimental data obtained by Nikuradse for Darcys friction
factor f for dierent values of the ratio ks /R. Here, ks is the height of the sand grains, and R is the pipe radius. As
can be observed, the friction factor is given by the Hagen-Poiseuille law for low Reynolds numbers (laminar regime).
For high enough values of this parameter, the friction factor depends only upon the ratio ks /R, independently of the
Reynolds number. In this case, the ow is said to be roughness-dominated.
The behavior shown in gure 8.3 is also observed in pipes with natural roughness. In particular, the friction factor
also becomes independent of the Reynolds number for suciently high values of this parameter (ow dominated by
the roughness). This parallelism between pipes with articial and natural roughness allows one to dene a parameter
to characterize the hydraulic eects caused by the natural roughness.
Suppose that we manufacture a pipe of diameter D with a certain rough material. Let f1 be the friction factor of
that pipe for roughness-dominated ow. Suppose that a pipe of the same diameter D is manufactured with a smooth
material, and then sand grains of height ks are uniformly distributed over its surface. The parameter ks is chosen
so that the friction factor for ow dominated by the roughness is f1 as well. If this operation were carried out for
dierent diameters, the value of ks would be essentially the same. In fact, this parameter only depends on the rough
material which the pipe is made of. Therefore, the parameter ks characterizes the hydraulic eects of the material
roughness. It is called is the material equivalent roughness.
To summarize, the equivalent roughness ks of a certain material is the height of the sand grains uniformly distributed
50

3 .- LOCAL LOSSES

Figure 8.3: Darcys friction factor for smooth pipes with sand grains of height ks . The symbol stands for the friction
factor f dened in the main text. Line (1) corresponds to the Hagen-Poiseuille law f = 64/ReD . The other lines
correspond to dierent values of ks .

over a smooth pipe to get the same friction factor for roughness-dominated ow.

2.3.- The Moody chart


Colebrook and White proposed the following empirical formula to calculate the friction factor in the turbulent regime:
(
)
1
ks /D
2.51
= 2.0 log10
+
.
(8.5)
3.7
Ref 1/2
f
The ratio ks /D is called the relative equivalent roughness.
The graphical representation of this formula is the Moody chart (gure 8.14). This chart also includes the values of
the friction factor f for the laminar regime (the Hagen-Poiseuille formula (8.3)). These values are valid for ReD 2300,
while the formula (8.5) can be applied for ReD
4000. The laminar-turbulent transition takes place in the interval
2300 Re 4000. In this interval, neither of the above two formulae provides reliable results.
The friction factor does not depend on the material roughness in the laminar regime. For moderate Reynolds
numbers both viscosity and roughness eects are relevant. If the Reynolds number takes values high enough, the
friction factor becomes independent of that parameter (i.e., the viscosity). In this case, the ow is dominated by the
roughness. The Reynolds number value above which this occurs depends on the relative equivalent roughness: the
greater the relative equivalent roughness the smaller the threshold of the Reynolds number.

3 .- LOCAL LOSSES
3.1.- Introduction
There is a great variety of hydraulic devices which are connected to multiple-pipe systems for dierent purposes (valves,
nozzles, diusers, elbows, etc.). Despite their diversity, they have a similar inuence on the ow evolution. The uid
crossing these devices experiences a drop of reduced pressure which increases with the ow rate. This reduced pressure
drop can become comparable to or even greater than that taking place in the pipe which the device is connected to.
51

HYDRAULICS

In some cases, the local loss of reduced pressure is measured and used to infer the value of the ow rate crossing the
device.

3.2.- Formulation of the problem


Consider a cylindrical pipe which transports a uid of density and viscosity . The diameter D1 of this pipe will be
used as the charaterisitc length of the problem, and indicates somehow the global scale of the problem. Suppose that
there is a hydraulic device which connects that pipe to another with, in general, dierent shape and size. We will refer
to the pipes in front of and behind the hydraulic device as upstream and downstream, respectively (see gure 8.4).
We denote by the symbols vm1 and P1 the upstream average velocity and reduced pressure, respectively. Analogously,
vm2 and P2 are the corresponding values downstream. Suppose that the uid experiences a drop of reduced pressure
P = P1 P2 in the device much larger than that taking place in a straight duct of similar dimensions. We aim at
determining that drop of reduced pressure.
upstream
D

m1

Device

m2

downstream

Figure 8.4: Hydraulic device connected to a pipe.

3.3.- Dimensional analysis of the problem


If the problem geometry is xed1 , the physical quantities intervening in the calculation of P are the following
ones. Firstly, the density and viscosity of the working uid. Secondly, the upstream pipe diameter D1 , which
characterizes the global size of the problem. Lastly, the upstream average velocity vm1 , which indicates the ow rate
crossing the device.
It must be noted that the downstream average velocity vm2 must not be added to the above set of quantities because
it can be obtained as a function of them and the problem geometry. In fact, the continuity equation establishes that
the ow rate at the inlet must be equal to that of the outlet, and hence
vm2 = vm1

2
D1
2 .
D2

(8.6)

On the other side, the equivalent roughness might be included too. However, it can be regarded as an additional
geometrical variable, and thus its relative value is assumed to be xed as the rest of the geometrical parameters.
The application of the theorem to the present problem allows one to obtain a formal relationship among the
relevant variables:
P
= K(ReD1 ) ,
(8.7)
1
2
2 vm1
where K is the local loss coecient, and ReD1 = vm1 D1 / is the Reynolds number.
The dependency of the local loss coecient K upon the Reynolds number ReD1 is to be determined experimentally.
One can nd in the literature empirical values of K as a function of ReD1 for a great variety of hydraulic devices. The
dependency of K upon ReD1 is similar to that of the friction factor f or the drag coecient CD . For suciently high
values of ReD1 , K becomes independent of that parameter. This means that, if the Reynolds number is high enough,
the drop of reduced pressure P is proportional to the ow rate squared.
1 We already know that xing the problem geometry is equivalent to setting the ratio between any pair of distances, but not the global
scale.

52

3 .- LOCAL LOSSES

Equation (8.7) is usually written as


v2
8K
P
= K m1 =
Q2 .
g
2g
g 2 D4

(8.8)

3.4.- Drop of reduced pressure in pipes with connections


As was explained, the drop P of reduced pressure in pipes is calculated from the equation
2
P
L vm
= fMoody (ReD , ks /D)
,
g
D 2g

(8.9)

where fMoody is the friction factor determined from the Moody chart (gure 8.14). If there are hydraulic devices
connected to the pipe, equation (8.9) must be completed with (8.8) to including the additional local losses associated
with those devices. In other words, the total drop of reduced pressure in the hydraulic facility is
i=N
i=N
2

P
Li vmi
v2
v2
i
=
fMoody
+ Ke me +
Ki mi ,
g
Di 2g
2g
2g
i=1
i=1

(8.10)

i
where fMoody , Li , Di and vmi are the values of the corresponding quantities for the pipe section located in front of
the device i, and Ki the local loss coecient of that device.

In equation (8.10), Ke is the local loss coecient associated with the pipe entrance, and vme is the average velocity
in that region. For suciently high Reynolds numbers, Ke depends exclusively on the entrance geometry. If the
entrance is appropriately rounded, Ke 0. Otherwise, Ke takes values between 0.5 and unity.
3.5.- Flow meters
As mentioned above, the local loss of reduced pressure P taking place in a hydraulic device can be regarded as an
indirect measure of the ow rate crossing that device. Suppose that we aim at measuring the ow rate transported by
a pipe. We connect to that pipe a hydraulic device which causes a local loss of reduced pressure P . This quantity
can be measured with a U-shaped tube as shown in gure 8.5. Let h be the dierence between the two heights
reached by the liquid in the U-shaped tube. It can be easily veried that
P
= (r 1)h ,
g

(8.11)

where r is the density of the liquid in the tube relative to that of the uid circulating through the pipe. From equation
(8.8) one concludes that
)1
(
r 1 2
vm1 =
2gh .
(8.12)
K
The ow rate is trivially calculated from vm1 . It is frequent to re-write (8.12) as

vm1 = C 2gh ,
(8.13)
where

)1
r 1 2
.
(8.14)
K
One can nd in the literature values for the function C(ReD1 ). For high enough Reynolds numbers, this function takes
a constant value.
C(ReD1 ) =

There is a great variety of ow meters which resort to the principle described above to measure the ow rate.
Among them, diaphragms, nozzles, and Venturi tubes are likely to be the most popular. The ow meter for a given
application must be chosen appropriately. If the drop of reduced pressure caused by the ow meter is too small, the
measurement will not be reliable. On the contrary, if P is too large, the ow meter alters signicantly the ow in
the pipe, and thus the measured ow rate is very dierent from the true value (that without the ow meter).
53

HYDRAULICS

upstream
vm1
D
1

downstream
P1

Device

P2
h

Figure 8.5: Flow meter.

Figure 8.6: The diaphragm used in our laboratory.

4 .- MULTIPLE- PIPE SYSTEMS


4.1.- Introduction
A multiple-pipe system is a set of pipes connected to each other to transport and distribute liquids. In this section, we
will describe the main types of multiple-pipe systems, as well as the procedures for calculating the relevant quantities.
Hereafter, we will call junctions the points where two straight pipes are connected. To simplify the problem, we
will assume that the local loss of reduced pressure taking place in those junctions are negligible as compared to that
occurring in the straight pipes. This simplication is valid as long as the pipes are long enough compared to their
diameters.

4.2.- Pipes in series


Pipes in series is a set of pipes in a line (see gure 8.7). The continuity equation establishes that all the pipes transport
the same ow rate; i.e.,
Q = Q1 = Q2 = = Qi = = QN ,

(8.15)

where N is the number of pipes. In addition, the drop of reduced pressure P between the inlet and outlet of the
system is
N
P
8Q2 i
Li
=
fMoody 5 ,
2
g
g i=1
Di

where use has been made of the continuity equation (8.15).


54

(8.16)

4 .- MULTIPLE- PIPE SYSTEMS

L1D1

LiDi

L2D2

LNDN

Figure 8.7: Pipes in series.

4.3.- Pipes in parallel


Figure 8.8 shows an example of N pipes working in parallel. In this case, the continuity equation establishes that the
total ow rate Q crossing the system is the sum of those crossing each pipe; i.e.,
Q=

Qi .

(8.17)

i=1

The uid particles come from and go to the same common points. Therefore, the drop of reduced pressure does not
depend on the pipe considered:
i
8fMoody L 2
Pi
(8.18)
=
5 Qi = const. ,
g
g 2 Di
and hence

1
2
i
N
fMoody L1 Q2
fMoody L2 Q2
fMoody Li Q2
fMoody LN Q2
1
2
i
N
=
= =
= =
.
5
5
5
5
D1
D2
Di
DN

L1D1

(8.19)

L2D2
LiDi

LNDN
Figure 8.8: Pipes in parallel.

4.4.- Pipe networks


Figure 8.9 represents a pipe network. The calculation of the ow rate distribution in a pipe network is not straightforward, and thus one resorts to commercial software to solve this problem.

4.5.- Pipeline with uniform draw-o


Suppose that a liquid is introduced at a ow rate Qi in a pipe of length L and diameter D (see gure 8.10). This
pipe delivers liquid through orices separated from each other by a constant distance. Thus, the ow rate decreases
progressively until it reaches a value Qo at the pipe exit. If the distance between two consecutive orices is very short
compared to L, the loss of ow rate can be assumed to take place in a continuous manner. The symbol q stands for
the (magnitude of the) ow rate delivered by the pipe per unit length. In addition, one assumes that the value of
55

HYDRAULICS

Network I

Network II

Network III

Figure 8.9: Pipe network.


L
Q

Q(x)

dx

x=0
q

Figure 8.10: Pipeline with uniform draw-o.

the friction factor fMoody is not signicantly aected by the drop of ow rate (as happens in ows dominated by the
roughness).
Under all these conditions, the continuity equation establishes that
Q(x) = Qi qx = Qo + q(L x) ,

(8.20)

where x is the pipe axis with x = 0 at the inlet section. The dierential drop of reduced pressure, dP , associated with
a pipe section of length dx is
)
(
dP
8fMoody dx
8fMoody
=
(Qo qx)2 =
dx Q2 + q 2 x2 2Qo qx .
o
g
g 2 D5
g 2 D5

(8.21)

To calculate the total drop of reduced pressure, one has to integrate the above expression:
(
)

( 2
) 8fMoody L
P
8fMoody L
q 2 L2
2
2 2
=
Qo +
Qo qL .
dx Qo + q x 2Qo qx =
g
g 2 D5 0
g 2 D5
3

(8.22)

This result is the same as that given by the expression:


P
8fMoody L 2
=
Qeq ,
g
g 2 D5

(
where

Qeq =

Q2
i

q 2 L2
+
Qi qL
3

)1
2
.

(8.23)

As can be seen, the drop of reduced pressure in the pipeline with uniform draw-o is equivalent to that taking place in
a standard pipe with the same diameter, length, and roughness, but transporting a ow rate Qeq . This last quantity
is called equivalent ow rate, and its value is larger than Qo and smaller than Qi .

5 .- MULTIPLE- PIPE SYSTEMS

WITH

PUMPS

5.1.- Head supplied by the pump


Essentially, the eect produced by a hydraulic pump connected to a multiple-pipe system is to increase the head of
the ow. Equation (4.38) allows one to calculate head supplied by the propeller to the uid as a function of the power

Wsol , neglecting the viscous energy dissipation which takes place inside the machine. The head HB provided by the
pump can be written as
2
Po Pi
v 2 vmi
HB =
+ mo
,
(8.24)
g
2g
56

5 .- MULTIPLE- PIPE SYSTEMS

WITH

PUMPS

where the symbols labeled with the subindexes i and o refer to values at the inlet and outlet sections of the pump,
respectively. The rst addend is the pressure head, while the second is called dynamic head. Due to the mass
conservation (continuity equation), the dynamic head is zero if the inlet and outlet pipes have the same diameter.
Figure 8.11 shows a typical curve of the head supplied by a pump versus the ow rate crossing that pump.

HB

Q
Figure 8.11: Head provided by a pump.

5.2.- Pumps in series


The ow rate Q crossing two pumps in series [see gure 8.12 (left)] is the same. If one neglects the local losses in the
A+B
A
B
connections between the two pumps, the head HB
provided by theem is the sum of the heads HB and HB given
by each one separately [see gure 8.12 (right)].

HB
1
Q

HB

A+B

Q
HB

HB

Q
Figure 8.12: Head (right) provided by two pumps in series (left).

5.3.- Pumps in parallel


The ow rate crossing two pumps in parallel [see gure 8.13 (left)] is shared between them so that the head provided
by each one is the same. In this way, a uid particle starting at the point 1 with a head H1 reaches the point 2 with
a head H2 independently of the way followed to reach that point. If one neglects the local losses in the connections
A+B
between the two pumps, the head curve HB (Q) is obtained by adding the ow rates QA and QB for the same value
of the head [see gure 8.13 (right)].

57

HYDRAULICS

QA

QA
1
Q

A
B

QA
QB

HB

HB

HB

A+B

QB
Q
Figure 8.13: Head (right) provided by two pumps in parallel (left).

58

5 .- MULTIPLE- PIPE SYSTEMS

WITH

PUMPS

Figure 8.14: The Moody chart. It represents the friction factor f as a function of the Reynolds number Re for xed
values of the relative equivalent roughness. These values are indicated on the right side of the graph.

59

Chapter 9

OPEN CHANNELS, WEIRS AND SLUICEGATES

1 .- INTRODUCTION
We shall devote this chapter to study some basic problems related to ows with free surface. Specically, the uniform
turbulent ow in straight channels is analyzed in the rst section. We shall restrict ourselves to present the Manning
formula for calculating the ow rate crossing a channel of arbitrary shape. This empiric equation is used in most
practical situations. In the second section, we will present a qualitative description of the ow trough weirs and
sluicegates.

2 .- UNIFORM FLOW

IN

OPEN CHANNELS

Consider the ow in a straight open channel of arbitrary shape and inclined an angle (see gure 9.1). Let be the
axis in the stream direction located on the free surface. We shall focus our attention on any section perpendicular to
the ow and suciently far away from both the inlet and outlet sections of the channel. The liquid motion through
this spanwise section is similar to that taking place in the developed ow in pipes. In particular, the velocity eld
does not depend on the coordinate which denes that section.
pamb

pamb
pamb

zs
z=0
Figure 9.1: Uniform ow in a open channel.

As mentioned above, the essential feature which distinguishes the ow in channels from those in pipes is the
existence of a free surface which gets the liquid in contact with the atmosphere. Therefore, the pressure on the liquid
side of the free surface can be considered as the ambient pressure pamb . If we neglect the variations of pressure and
height within a spanwise section, then the reduced pressure distribution becomes
P () = pamb + gzs ,

(9.1)

where zs is the free surface height for a given value of (see gure 9.1). As occurs in developed ows in pipes, the
liquid motion is driven by a constant gradient of reduced pressure. However, in open channels this gradient is caused
by the gravity force exclusively. For this reason, the channel must be inclined a certain angle .
Most channels are manufactured to transport low-viscosity liquids (water). They are usually made of rough
materials, and the liquid average velocity is not small. For these reasons, the ow can be regarded as turbulent and
dominated by the roughness in most practical situations. In this case, the formula for cylindrical pipes could provide
60

3 .- WEIRS

AND

SLUICEGATES

good estimates if some considerations were taken into account. Nevertheless, there are other empirical formulae for
open channels with a wider range of validity. Mannings formula constitutes a remarkable example. It reads
vm =

2
1
1
Rh 3 S 2 ,
n

(9.2)

where vm is the average velocity over the channel cross section, n is the Manning coecient which depends on the
material the channel is made of, Rh is the hydraulic radius (dened as the ratio of the cross sectional area to the
wetted perimeter1 ), and S = dzs /ds = tan is the channel slope.
From the Manning formula (9.2), one concludes that, given the channel material and slope, the greater the hydraulic
radius, the greater the ow rate. For regular polygons, the hydraulic radius increases with the number of sides. It can
be veried that the circle is the shape with the greatest hydraulic radius. Therefore, one can assert that a part of a
regular polygon is a good choice from a hydraulic standpoint, while a part of the circle is the optimum (see gure 9.1).

3 .- WEIRS

AND

SLUICEGATES

3.1.- Weirs
A weir is an obstacle located in a channel. Figure 9.2 represents a thick-wall weir placed in a wide channel. The
maximum height of the weir is zv , the ow rate per unit of length is q, and the depth in front of the obstacle is h0 .
q

h0

zv

Figure 9.2: Flow over a thick-wall weir.

These weirs are frequently used as ow meters. The ow rate q can be approximately obtained as
3

q = Cv g 2 (h0 zv ) 2 ,

(9.3)

where Cv is a dimensionless parameter which slightly depends on the weir geometry and the Reynolds number. Its
value is about 0.54.

3.2.- Sluicegates
A sluicegate is a vertical gate which hinders the liquid motion. It has an orice in its lower part for the liquid to pass
(see gure 9.3). We shall denote by hc the orice height, and by h0 and h1 the stream depth in front of and behind
the sluice gate, respectively. In all cases, h1 < h0 , which implies that the liquid stream shrinks downstream (the vena
contracta eect).
Sluicegates are also frequently used to measure the ow rate per unit length q which circulates through a wide
channel. This quantity can be obtained from the empirical formula

q = Cv Cc hc 2g(h0 Cc hc ) ,
(9.4)
where Cv is a dimensionless parameter which depends on the Reynolds number, and Cc = h1 /hc is the coecient of
contraction of the vena contracta, whose value essentially depends on the sluicegate shape. The Cv value lies in the
interval (0.7-1), while Cc takes values around 0.6.

1 The

wetted perimeter is the length of the contact line between the liquid and the solid wall for a given cross section.

61

OPEN CHANNELS, WEIRS AND SLUICEGATES

h0
hc

h1

Figure 9.3: Flow across a sluicegate.

62

Anda mungkin juga menyukai