lOMoARcPSD|3826304

Fluid Mechanics
Fluid Mechanics
This course will be, for almost everyone, a first course in Fluid Mechanics. The fact that
Mechanical, Civil and Chemical Engineering students all take this course gives an immediate
indication of how much of a core topic this is, and how wide are the applications of Fluid
Mechanics in Engineering. Detailed course aims and “learning outcomes” are given in the
“Course Descriptor” but, in summary, by the end of the course, you should have learned about
fluid properties, many of the key qualitative flow phenomena, and have begun to learn the
building blocks (and limitations) of quantitative flow modelling.

1. INTRODUCTION
1.1 Properties of Fluids

1.1.1 Definition of a Fluid

Liquids and gases lack the ability of solids to offer a permanent resistance to a deforming force
– they deform continuously for as long as the force is applied. Deformation is caused by
shearing forces which act tangentially to the surfaces to which they are applied.

Figure 1.1: Deformation of a notional element of fluid under the action of a shearing force.

A fluid is a substance which deforms continuously under

the action of shearing forces, however small

Corollary: for a fluid at rest ⇒ must no shearing forces

⇒ all forces must be perpendicular to planes upon which they act

1.1.2 Viscosity

Figure 1.2: Deformation (shear strain) of a notional element of fluid under the action of a shear
stress (i.e. shear force per unit area).

shear force per unit area = F/A = shear stress, τ φ = shear strain

For a solid, a given shear stress τ results in a given, fixed shear strain φ , i.e. φ ∝ τ (SM2!)

1
 lOMoARcPSD|3826304

For a liquid, however, it turns out that it is the rate of change of strain that is proportional to the
applied shear stress;
∂ϕ
∝τ
∂t
From Figure 1.2, when φ is small, φ ≈ x / y

∂ϕ ∂  x  1 ∂x u
⇒ =  = =
∂t ∂t  y  y ∂t y

where u is the velocity in the x direction.

∂φ u u
⇒ τ ∝ ∝ ⇒ τ = const ×
∂t y y
du
⇒ τ =µ NEWTON’S LAW OF VISCOSITY
dy

shear stress dynamic viscosity

(Nm-2) (kgm-1s-1)

Thus, viscosity is a property of a fluid which tells us how easy (or difficult) it is to shear the fluid,
i.e. how easily will it flow under given shearing forces.

Fluids obeying Newton’s Law of Viscosity are known as Newtonian Fluids.

Non-Newtonian fluids include “plastic fluids” such as sewage sludge and toothpaste and “visco-
elastic fluids” such as custard…

Shearing forces within the body of a fluid arise due to adjacent “layers” of fluid moving past each
other, e.g. near a wall;

Figure 1.3: Fluid flow near a wall / boundary. Note zero velocity at the boundary –
the no slip condition

Viscosity is a measure of the fluid friction between adjacent (notional) layers of the flowing fluid.

Intuitively, µgases << µliquids

For gases, friction (viscosity) is due to inter-molecular interactions / collisions;

µgas ↑ as temperature T ↑

In liquids, friction (viscosity) is due to intermolecular cohesion, so

µliquid ↓ as temperature T ↑

2
 lOMoARcPSD|3826304

E.g. the viscosity of water decreases by a factor of 3 between temperatures of 5 0C and

55 0C.

Note: the combination µ / ρ occurs sufficiently often in Fluid Mechanics equations that it is given
its own symbol (ν, “nu”) and name (kinematic viscosity). FM2 will use dynamic viscosity, µ as
this is the more fundamental and physically “tangible” measure, but you will see ν elsewhere, so
you need to know about it!

1.1.3 Density

density = mass per unit volume, i.e. ρ = m / V

specific density or specific gravity ≡ ρ / ρwater (with ρwater taken as 1000 kgm-3).

Flows in which variations in ρ are negligible are termed incompressible

Flows in which variations in ρ are non-negligible are termed compressible

A working definition of negligible might be < 5%.

• for our purposes, all liquid flows may be regarded as incompressible

• gas flows in which the flow speed u « sound speed cs show only small variations in
density and may also be regarded as incompressible.

u / cs ≡ M the Mach number

• for M ≤ c. 0.3, ∆ρ / ρ ≤ c. 5%

1.1.4 Surface Tension (liquids only)

Within the body of the liquid, inter-molecular attractive

forces on average cancel out.

At an interface, upward and downward attractions may

be unbalanced. The liquid surface behaves like an
elastic membrane under tension.

• surface tension σ = force in liquid surface

perpendicular to a line of unit length in the
surface.
• σ is constant over the surface of separation (at
a given temperature)
• as T ↑ , σ ↓

The effect of surface tension is to reduce the free surface area of a body of liquid to a minimum,
e.g. droplets.

Surface tension forces are often small in engineering flows, so surface tension is often
neglected (or forgotten). The times when it is important to remember it are when dealing with
fluids problems involving droplets and/or bubbles, where it can be very important in determining
key aspects of the flow (e.g. droplet break-up in a spray).

Beware: surface tension effects can cause problems when using scaled flow models.

3
 lOMoARcPSD|3826304

1.2 Basic Flow Fields and Phenomena

1.2.1 Laminar and Turbulent Flows

Laminar flow
• characterised by smooth fluid motion in “layers” (“laminae”)
• dye injected appears as a single line
• velocity may vary from layer to layer

or

Turbulent flow
• characterised by random, 3-d motion (generally in addition to a mean flow)
• dye injected breaks up into myriad of entangled threads

= +

Most (but certainly not all) flows of engineering interest are turbulent. Understanding turbulence
is one of the most difficult areas of Fluid Mechanics. While many useful (simplifying) models
have been developed over the last 60 years or so, there remain significant unsolved problems
associated with turbulence.

In 1883, Osborne Reynolds studied flows in a tube and found that the regime – laminar or
turbulent – was determined by the parameter

ρuD uD
Re = =
µ ν
Re is the Reynolds number, D is the diameter of the pipe, and ρ, µ and u have their usual
meanings. Note that ν is the kinematic viscosity ≡ µ / ρ (see Section 1.1.2).

For flow in a pipe, we can identify three flow regimes:

Re < c. 2000 laminar flow (if undisturbed)

Re > c. 4000 turbulent flow
2000 < Re < 4000 transitional - could be either laminar or turbulent
(can’t tell from Re alone)

Note that in transitional regime, at any point and time, the flow will be either laminar or turbulent.
Physically, a flow cannot be both laminar and turbulent simultaneously, but it may switch
between regimes, apparently randomly (e.g. due to a small disturbance to flow).

Re is a key parameter in many flow problems (probably the single most-quoted parameter in
Fluid Mechanics).

At low Re, viscous forces dominate

At high Re, viscous forces become unimportant (and can sometimes be neglected)

4
 lOMoARcPSD|3826304

1.2.2 Flow in an Enclosed Duct / Pipe

• no slip at the walls (zero flow velocity relative to the walls at the wall)
• flow slows near to the wall due to viscous friction – region of flow affected by presence of
wall is the boundary layer
• viscosity effects are dominant in the boundary layer
• boundary layer thickness δ ↑ as µ ↑
• “ “ “ δ ↓ as “free stream” flow speed, u ↑

The variation of flow speed (in the down-duct direction) with cross-duct position (e.g. the curve
in the Figure, above) is called a velocity profile.

We will show (in Chapter 3) that the velocity profile for laminar flow in a duct is a parabola.

For turbulent flow, the velocity profile is much “flatter” due to effective mixing across the pipe.

For some purposes it is useful to define an ideal flow which is an incompressible, inviscid flow
(i.e. no viscous effects at all). For an ideal flow, the velocity profile is completely flat. All flow
travels at the same speed. Because viscosity is neglected, there are no boundary layers, and
we have free-slip at the wall.

1.2.3 Flow Along a Flat Plate

• fluid slows near the wall ⇒ shearing taking place between “layers” due to viscous friction
• viscosity dominant in boundary layer
• inviscid may be appropriate for flow outside the boundary layer where viscous effects are
less important and may be negligible (reminder: inviscid + incompressible = “ideal” flow)
• no slip at wall
• boundary layer thickness δ ≡ distance away from wall (usually called “y”) at which
u = 0.99u∞

From what we know already about the difference between laminar and turbulent flow, we might
expect that a boundary layer would grow differently, and take a different shape (or “profile”)
depending upon whether the flow within it is laminar or turbulent (think back to the shapes of the
laminar and turbulent velocity profiles in pipe flow). This is indeed the case, and (as usual) we
need a different model for laminar and turbulent conditions AND a means of predicting which
regime prevails in a particular case.

5
 lOMoARcPSD|3826304

First, the means of predicting whether a boundary layer flow is laminar or turbulent – we use a
Reynolds number again, but this time based not upon a pipe diameter, but on downstream
distance along the plate;
ρ u∞ x
Re x =
µ
Experiment shows that the transition between laminar and turbulent boundary layer takes place
somewhere in the region
105 < Rex < 2 × 105
5
For a laminar boundary layer (Rex < 10 ), classical analytical flow modelling (Mech Eng Fluid
Mechanics 4, or see textbook) gives the following prediction for laminar boundary layer
thickness (supported by experiment)
5x
δ=
Re1x/ 2

⇒ δ ∝ x1/2; δ ∝ u∞-1/2
For a turbulent boundary layer
0.37 x
δ=
Re1x/ 5

⇒ δ ∝ x4/5; δ ∝ u∞-1/5

The presence of the shear stress at the plate ⇒ there must be a force on the plate in the
direction of flow – this is skin friction drag. Experiment and dimensional analysis (see next
Section) give this drag force as
F = 0.5 cf ρ A u∞
2

where cf is the skin friction drag coefficient.

Again (as hopefully we now expect), we have two models for cf ;

• for laminar boundary layer

1 .4
cf =
Re1x/ 2
• for turbulent boundary layer
0.074
cf =
Re1x/ 5

(Note that all these formulae are given on the datasheet, but presented without explanation.
Thus you don’t have to remember them, but you need to know how to select and use them.)

6
 lOMoARcPSD|3826304

1.2.4 Flow Past a Bluff Body (“External Flow”)

There are three main regimes here

• main “free stream” flow, perhaps deflected by the body but unaffected in character
• boundary layer flow immediately adjacent to the surface of the body
• separated region(s) – the wake – where the boundary layer has separated from the
surface of the body (at the “separation point”)

The detailed description of the processes at work determining whether the boundary layer stays
“attached” to the body or separates are beyond the scope of this course, but are not beyond the
understanding of an FM2 student – look up “separation” in (e.g.) any edition of Douglas et al.

A key engineering consequence of flow past a bluff body is form drag. A drag force is a force
on the body in the downstream direction, or opposing the motion of a body moving through a
fluid (e.g. a vehicle travelling at speed). The term “drag” is often used to mean form drag but
beware – there are other important types of drag, e.g. skin friction drag; wave-making drag.

In the wake of the body, there is a lot of recirculation, which results in viscous friction and
energy loss, which in turn causes a pressure drop in these separated, wake regions. In front
(upstream) of the body, the flow is slowed causing some rise in pressure in the region of the
stagnation point. Thus we now have a pressure difference set up across the body, from high-to-
low, and so the body experiences a force acting downstream (from high pressure to low).

The wider the wake, the greater will be the drag force, because this pressure difference is acting
on a greater area. This leads to the concept of streamlining to reduce wake width and thus
form drag.

From experiment (and analytical approaches too – see Section 1.3) we find

1
FD = C D ρAu 2
2

projected frontal area of body

drag coefficient
- depends upon geometry of body
- empirically determined

For long, slender bodies placed across a flow (i.e. when the diagram above represents a cut
through a long section), and when the “free stream” flow is quite steady (not changing quickly
with time), the wake is observed to oscillate (with alternate vortex shedding occurring from top
and bottom of body). This alternate vortex shedding causes lift forces (forces on the body at
right angles to the flow direction) at a fixed frequency. This is a forced vibration which may have

7
 lOMoARcPSD|3826304

significant engineering consequence if the frequency of the vortex shedding coincides with a
natural frequency of the body. Examples include
• “singing” of telephone wires
• Tacoma Narrows bridge collapse

For a circular cylinder of diameter D in a steady flow of speed u, the frequency f of the vortex
shedding can be estimated from an empirical formula;

fD  19.7 
≈ 0.1981 −  ≈ 0 .2
u  Re 
for 250 < Re < 2×105.

The quantity fD/u is known as the Strouhal number after Vincenz Strouhal who first
investigated ringing (1878).

1.2.5 Orifice Flow

Key:
1. stagnant region – little or no flow 2. separation at lip of orifice
3. inward flow just after orifice
Note: it can be observed from 1,2 & 3 that flows don’t turn sharp corners

4. “vena contracta” in jet; cross-sectional area < orifice area

5. recirculation outside jet
6.. pressure drop across orifice (⇒ energy losses – see Chapter 5)

1.2.6 Flow Around a (Sharp) Bend

Key:
1. separation and recirculation at outside corner 2. separation at inside corner
3. vena contracta just downstream of corner 4. recirculation

8
 lOMoARcPSD|3826304

1.3 Dimensional Analysis

1.3.1 Introduction

If an equation or relationship between parameters is to make sense, then the dimensions of all
terms must be the same.

1 clown + 2 donkeys = 3 Hearts players

or
12.5 kg + 25 ms-1 =37.5 furlongs per fortnight

are not dimensionally correct, whereas

1 player + 2 players = 3 players

and
230 Jkg-1 + 58 Jkg-1 = 288 Jkg-1

do make sense.

Thus we see that each term must have the same dimensions, by which we mean each term
must be made up of the same basic physical quantities (and is therefore describable /
measurable by the same units). In mechanics, the three basic dimensions are

mass, M
length, L
time, T

(Aside: beyond mechanics, there are four other fundamental dimensions; current, voltage,
temperature and luminescence)

The value of the use of dimensional analysis to find the simplest form(s) of functional
relationships is perhaps best illustrated by examples. We begin with an easy example, and then
expand this to arrive at a more useful conclusion.

1.3.2 Easy example: Drag force (1st pass!).

Experiment and thought suggest

FD = f (u, D, ρ)

i.e. drag force is a function of flow speed, object dimension, and the density of the fluid flowing

The general form of the function relating drag force FD to the three parameters upon which we
think it might depend would be

FD = C1 u a D b ρ c + C2 u d D e ρ f + … etc.

To make physical sense however, the dimensions (i.e. the combination of M, L and T) in each
term on the rhs of the equation (above) must be the same. Using the notation conventional that

[ x ] = “the dimensions of x”

we can thus write

[ FD ] = [ u a D b ρ c ] noting that a constant (e.g. C1) is dimensionless

9
 lOMoARcPSD|3826304

⇒ [ FD ] = [ u ] a [ D ] b [ ρ ]c {equation A}

We now work out the dimensions of each quantity;

[ FD ] = [ force ] = [ mass × acceln ] = [ mass × length per time per time ] = M L T -2

[ u ] = [ speed ] = [ length per unit time ] = L T -1
[ D ] = [ length ] =L
[ ρ ] = [ mass per unit volume ] =ML
-3

Equating dimensions on left- and right-hand sides of equation A (above) gives

M L T -2 = (LT-1) a (L) b (ML-3) c

= L a T -a L b M c L -3c

Now equate powers of individual dimensions M, L and T on left- and right-hand sides;

mass: 1 = c
length: 1 = a + b -3c
time: -2 = -a

Solving these three equations simultaneously (by inspection) gives

a=2
b=2
c=1

⇒ FD ∝ ρ D 2 u 2 ∝ ρ A u 2 as before.

1.3.3 Harder example: Drag force (improved!).

Let’s imagine that we carry out an experiment to determine the drag coefficient cD for a sphere,
getting a value of 0.44. You then use this value in the familiar equation (FD = 0.5 cD ρ A u 2) to
predict drag forces. Predictions and further measurements agree well, except at particularly low
and high flow speeds, or for very small or large sphere diameters. What is going on? How can
we devise a series of experiments to give us cD for all conditions without having to measure for
a huge number of combinations of different ρ , A and u values?.

Let’s rework the above example, but now including the viscosity as one of the parameters upon
which the drag force might depend, i.e. FD = f (u, D, ρ, µ)

[ FD ] = [ u ] [ D ] [ ρ ] [ µ ]
a b c d
⇒

From Newton’s law of viscosity (fundamental definition of µ ), [ µ ] = M L T

-1 -1
(exercise)
-2
⇒ MLT = (LT-1) a (L) b (ML-3) c (M L-1 T -1 )d

= L a T -a L b M c L -3c M d L -d T -d

Equating powers of M, L and T as before;

mass: 1 = c+d {equation M}

length: 1 = a + b -3c - d {equation L}
time: -2 = -a - d {equation T}

10

Downloaded by Suraj Lakshan (surajlanka@gmail.com)

 lOMoARcPSD|3826304

We now have three equations in four unknowns. Thus we cannot fully solve this set of
equations, but we ought to be able to reduce it down to one unknown.

From {M}: c=1-d

From {T}: a = 2 -d
Substitute into {L}; 1 = (2-d) + b - 3(1-d) ⇒ b=2-d

Now d is the only unknown. Substituting back into the original relation;

⇒ FD = f ( u 2-d , D 2-d , ρ 1-d , µ d )

Grouping terms

 ρ u D  −d 
⇒ FD = f   , ρ D 2 u 2 
 µ  

= f { Re, ρ A u 2 )

Comparing this with the (now) familiar FD = 0.5 cD ρ A u shows that the drag coefficient must
2

be a function of the Reynolds number.

Dimensional analysis has thus shown that an experimental programme to determine cD need not
tests the effects of varying all parameters (speed, size, fluid density and viscosity) independently
and in turn. Rather, only a range of Re need be tested - a HUGE simplification and saving in
effort (= money).

Aside: physically, the conclusion that cD = f(Re) tells us that the drag force depends upon the
relative importance of inertial and viscous forces in a given flow problem.

1.3.4 Conclusions on Dimensional Analysis

In conclusion, dimensional analysis offers insight into the physical basis of a problem, but
empirical data is also required to “calibrate” formulae (dimensional analysis can’t help us
establish values of constants etc).

Texts offer systematic (and quite complex) approaches to dimensional analysis (e.g the
“Buckingham π” method) which go beyond what we need here.

The application of these ideas that you will (perhaps) use most frequently is simply the use of
dimensions to check a formula, e.g. if you can’t remember whether p = 0.5 ρ u 2 or p = 0.5 ρ u 3,
you can check dimensions to find out.

11