1

Introduction

1.1 The nature of turbulent flows

There are many opportunities to observe turbulent flows in our everyday
surroundings, whether it be smoke from a chimney, water in a river or
waterfall, or the buffeting of a strong wind. In observing a waterfall, we
immediately see that the flow is unsteady, irregular, seemingly random and
chaotic, and surely the motion of every eddy or droplet is unpredictable. In
the plume formed by a solid rocket motor (see Fig. 1.1), turbulent motions
of many scales can be observed, from eddies and bulges comparable in size
to the width of the plume, to the smallest scales the camera can resolve. The
features mentioned in these two examples are common to all turbulent flows.
More detailed and careful observations can be made in laboratory exper-
iments. Figure 1.2 shows planar images of a turbulent jet at two different
Reynolds numbers. Again, the concentration fields are irregular, and a large
range of length scales can be observed.
As implied by the above discussion, an essential feature of turbulent flows
is that the fluid velocity field varies significantly and irregularly in both
position and time. The velocity field (which is properly introduced in Section
2.1) is denoted by U (x, t), where x is the position and t is time.
Figure 1.3 shows the time history U1 (t) of the axial component of velocity
measured on the centerline of a turbulent jet (similar to that shown in
Fig. 1.2). The horizontal line (in Fig. 1.3) shows the mean velocity denoted
by U1 , and defined in Section 3.2. It may be observed that the velocity
U1 (t) displays significant fluctuations (about 25% of U1 ), and that, far
from being periodic, the time history exhibits variations on a wide range of
timescales. Very importantly, we observe that U1 (t) and its mean U1  are in
some sense ‘stable’: huge variations in U1 (t) are not observed; neither does
U1 (t) spend long periods of time near values different than U1 .

3
4 1 Introduction

Fig. 1.1. A photograph of the turbulent plume from the ground test of a Titan IV
rocket motor. The nozzle’s exit diameter is 3 m, the estimated plume height is 1,500 m,
and the estimated Reynolds number is 200 × 106 . For more details see Mungal and
Hollingsworth (1989). With permission of San Jose Mercury & News.

Figure 1.4 shows the profile of the mean velocity U1  measured in a
similar turbulent jet as a function of the cross-stream coordinate x2 . In
marked contrast to the velocity U1 , the mean velocity U1  has a smooth
profile, with no fine structure. Indeed, the shape of the profile is little different
than that of a laminar jet.
In engineering applications turbulent flows are prevalent, but less easily
seen. In the processing of liquids or gases with pumps, compressors, pipe
lines, etc., the flows are generally turbulent. Similarly the flows around
vehicles – e.g., airplanes, automobiles, ships, and submarines – are turbulent.
The mixing of fuel and air in engines, boilers, and furnaces, and the mixing
of the reactants in chemical reactors take place in turbulent flows.
 1.1 The nature of turbulent flows 5

Fig. 1.2. Planar images of concentration in a turbulent jet: (a) Re = 5,000 and
(b) Re = 20,000. From Dahm and Dimotakis (1990).

5
U1(t)
(m s –1) 4
3

0
0.0 0.1 0.2 0.3
t (s)
Fig. 1.3. The time history of the axial component of velocity U1 (t) on the centerline
of a turbulent jet. From the experiment of Tong and Warhaft (1995).

An important characteristic of turbulence is its ability to transport and

mix fluid much more effectively than a comparable laminar flow. This is well
demonstrated by an experiment first reported by Osborne Reynolds (1883).
Dye is steadily injected on the centerline of a long pipe in which water is
flowing. As Reynolds (1894) later established, this flow is characterized by a
single non-dimensional parameter, now known as the Reynolds number Re.
In general, it is defined by Re = UL/ν, where U and L are characteristic
6 1 Introduction

1.0

〈U1〉
〈U1〉0

0.0
-0.2 0.0 0.2
x2 / x1
Fig. 1.4. The mean axial velocity profile in a turbulent jet. The mean velocity U1 
is normalized by its value on the centerline, U1 0 ; and the cross-stream (radial)
coordinate x2 is normalized by the distance from the nozzle x1 . The Reynolds number
is 95,500. Adapted from Hussein, Capp, and George (1994).

velocity and length scales of the flow, and ν is the kinematic viscosity of the
fluid. (For pipe flow, U and L are taken to be the area-averaged axial velocity
and the pipe diameter, respectively.) In Reynolds’ pipe-flow experiment, if
Re is less than about 2,300, the flow is laminar – the fluid velocity does not
change with time, and all streamlines are parallel to the axis of the pipe. In
this (laminar) case, the dye injected on the centerline forms a long streak
that increases in diameter only slightly with downstream distance. If, on the
other hand, Re exceeds about 4,000, then the flow is turbulent.1 Close to the
injector, the dye streak is jiggled about by the turbulent motion; it becomes
progressively less distinct with downstream distance; and eventually mixing
with the surrounding water reduces the peak dye concentration to the extent
that it is no longer visible.
(Visualizations from a reproduction of Reynolds’ experiment, and from
other canonical turbulent flows, are contained in Van Dyke (1982). There is
great educational value in studying this collection of photographs.)
The effectiveness of turbulence for transporting and mixing fluids is of
prime importance in many applications. When different fluid streams are
brought together to mix, it is generally desirable for this mixing to take
place as rapidly as possible. This is certainly the case for pollutant streams

1 As the Reynolds number is increased, the transition from laminar to turbulent flow occurs over a
range of Re, and this range depends on the details of the experiment.
 1.2 The study of turbulent flows 7

released into the atmosphere or into bodies of water, and for the mixing of
different reactants in combustion devices and chemical reactors.
Turbulence is also effective at ‘mixing’ the momentum of the fluid. As a
consequence, on aircraft’s wings and ships’ hulls the wall shear stress (and
hence the drag) is much larger than it would be if the flow were laminar.
Similarly, compared with laminar flow, rates of heat and mass transfer at
solid–fluid and liquid–gas interfaces are much enhanced in turbulent flows.
The major motivation for the study of turbulent flows is the combination
of the three preceding observations: the vast majority of flows is turbulent;
the transport and mixing of matter, momentum, and heat in flows is of great
practical importance; and turbulence greatly enhances the rates of these
processes.

1.2 The study of turbulent flows

Many different techniques have been used to address many different ques-
tions concerning turbulence and turbulent flows. The first step toward pro-
viding a categorization of these studies is to distinguish between small-scale
turbulence and the large-scale motions in turbulent flows.
As is discussed in detail in Chapter 6, at high Reynolds number there is a
separation of scales. The large-scale motions are strongly influenced by the
geometry of the flow (i.e., by the boundary conditions), and they control the
transport and mixing. The behavior of the small-scale motions, on the other
hand, is determined almost entirely by the rate at which they receive energy
from the large scales, and by the viscosity. Hence these small-scale motions
have a universal character, independent of the flow geometry. It is natural
to ask what the characteristics of the small-scale motions are. Can they be
predicted from the equations governing fluid motion? These are questions
of turbulence theory, which are addressed in the books of Batchelor (1953),
Monin and Yaglom (1975), Panchev (1971), Lesieur (1990), McComb (1990),
and others, and that are touched on only slightly in this book (in Chapter
6).
The focus of this book is on turbulent flows, studies of which can be
divided into three categories.

(i) Discovery: experimental (or simulation) studies aimed at providing

qualitative or quantitative information about particular flows.
(ii) Modelling: theoretical (or modelling) studies, aimed at developing
tractable mathematical models that can accurately predict properties
of turbulent flows.
8 1 Introduction

(iii) Control: studies (usually involving both experimental and theoretical

components) aimed at manipulating or controlling the flow or the
turbulence in a beneficial way – for example, changing the boundary
geometry to enhance mixing; or using active control to reduce drag.

The remainder of Part I of this book is based primarily on studies in the

first category, the objective being to develop in the reader an understanding
for the important characteristics of simple turbulent flows, of the dominant
physical processes, and how they are related to the equations of fluid motion.
The description of turbulent flows contained in Part I is not comprehensive:
additional material can be found in the books of Monin and Yaglom (1971),
Townsend (1976), Hinze (1975), and Schlichting (1979).
For studies in the second category, that aim at developing tractable math-
ematical models, the word ‘tractable’ is crucial. For fluid flows, be they
laminar or turbulent, the governing laws are embodied in the Navier–Stokes
equations, which have been known for over a century. (These equations
are reviewed in Chapter 2.) Considering the diversity and complexity of
fluid flows, it is quite remarkable that the relatively simple Navier–Stokes
equations describe them accurately and in complete detail. However, in the
context of turbulent flows, their power is also their weakness: the equations
describe every detail of the turbulent velocity field from the largest to the
smallest length and time scales. The amount of information contained in the
velocity field is vast, and as a consequence (in general) the direct approach
of solving the Navier–Stokes equations is impossible. So, while the Navier–
Stokes equations accurately describe turbulent flows, they do not provide a
tractable model for them.
The direct approach of solving the Navier–Stokes equations for turbu-
lent flows is called direct numerical simulation (DNS), and is described in
Chapter 9. While DNS is intractable for the high-Reynolds-number flows of
practical interest, it is nevertheless a powerful research tool for investigating
simple turbulent flows at moderate Reynolds numbers. In the description
of wall-bounded flows in Chapter 7, DNS results are used extensively to
investigate the physical processes involved.
For the high-Reynolds-number flows that are prevalent in applications,
the natural alternative is to pursue a statistical approach. That is, to describe
the turbulent flow, not in terms of the velocity U (x, t), but in terms of some
statistics, the simplest being the mean velocity field U (x, t). A model based
on such statistics can lead to a tractable set of equations, because statistical
fields vary smoothly (if at all) in position and time. In Chapter 3 we present
the concepts and techniques used in the statistical representation of turbulent
 1.2 The study of turbulent flows 9

flow fields; while in Part II we describe statistical models that can be used to
calculate the properties of turbulent flows. The approaches described include:
turbulent viscosity models, e.g., the k–ε model (Chapter 10); Reynolds-stress
models (Chapter 11); models based on the probability density function (PDF)
of velocity (Chapter 12); and large-eddy-simulations (LES) (Chapter 13).
The statistical models described in Part II can be used in some studies in
the third category mentioned above – that is, studies aimed at manipulating
or controlling the flow or the turbulence. However, such studies are not
explicitly discussed here.
A broad range of mathematical techniques is used to describe and model
turbulent flows. Appendices on several of these techniques are provided to
serve as brief tutorials and summaries. The first of these is on Cartesian
tensors, which are used extensively. The reader may wish to review this
material (Appendix A) before proceeding. There are exercises throughout
the book, which provide the reader with the opportunity to practice the
mathematical techniques employed. Most of these exercises also contain
additional results and observations. A list of nomenclature and abbreviations
is provided on page xxi.