Anda di halaman 1dari 20

Level 2.

Fluid mechanics

CHAPTER 3.

EXTERNAL INCOMPRESSIBLE FLOW


In the previous chapter we examined internal flows and studied in detail the
effects of wall friction on the behaviour of flow in pipes. This allowed us to
quantify such things as the energy dissipation in a fluid and to derive, either
theoretically or experimentally, the typical flow profiles in our pipes. In this
chapter we shall adopt a similar approach but this time looking at external
flows.

External flow is the flow over a body which is submerged in an unbounded


fluid. Think, for example, of the flow of air over an aerofoil. External flows
share many similarities with the internal flows discussed in the last chapter. In
our study of internal flow we noted that the fluid velocity at the pipe wall was
always zero (the no slip condition). An examination of the velocity profiles
derived previously, both for laminar and turbulent flow, show the velocity
increasing from zero at the wall to a maximum value at the centre of the duct.
This change in velocity is caused by the viscosity of the fluid. The arguments
used when studying internal flows are applicable also to external flows but only
in a region very close to the surface of the submerged body. Again the fluid
sticks to the wall of the body but the velocity then gradually changes from zero
at the wall to a free stream velocity U, which is the bulk velocity of the
surrounding flow, not a centreline velocity. This region of velocity change (or
gradient) is known as the boundary layer, and it has a thickness which is the
distance from the wall (where the velocity is zero) to the free stream (where the
velocity is U). Thus we could equally apply the concept of a boundary layer to
29

Level 2. Fluid mechanics

both internal and external flows, however it is most commonly referred to in the
study of external flows which we shall cover next.

3.1 The Boundary Layer Concept


The concept of a boundary layer was first introduced by Prandtl in 1904. The
concept was a significant breakthrough since, up until his work, theoretical
ideas were still based on equations for inviscid flow. Of course inviscid flow
predictions differed greatly from experimental measurements; the boundary
layer concept provided the missing link between theory and practice.

Prandtl showed that many viscous flow problems may be analysed by dividing
the flow into two regions: a thin region adjacent to the solid boundary in which
the effect of viscosity is important (the boundary layer) and an outer region in
which the effect of viscosity is negligible and the fluid may be treated as
inviscid. The boundary layer concept permitted the theoretical analysis of many
viscous flow problems that previously had been considered impossible - and
thus founded the modern era of fluid mechanics.

The most straightforward example with which to introduce the concept of the
boundary layer is the flow of a fluid over a stationary plate.
U

u(y)
y

u(y)

x
Laminar

Transition

30

Turbulent

Level 2. Fluid mechanics

The flow over the plate thus separates into three distinct regions:

As the fluid impinges on the plate the fluid particles in direct contact with the
wall assume the same velocity as the wall itself - this is the no-slip condition. If
the flat plate is stationary then the velocity of the fluid adjacent to the wall is
zero. This stationary element of fluid will, in turn, retard the adjacent particles
in the bulk fluid above - this is viscosity in action, slowing the fluid down near
the wall. The effect of viscosity sets up a velocity gradient in the fluid and
hence a shear stress according to the equation

du
dy

As the flow moves along the plate the viscous effects in the fluid propagate
further into the bulk fluid and the boundary layer thickness grows. Under most
conditions, the boundary layer is initially laminar, that is the flow behaves as a
series of ordered layers, or laminae, flowing over one another.

Eventually, a point is reached at which instabilities in the boundary layer cause


a breakdown in the laminar structure of the flow and turbulent mixing begins to
take hold. In this region of transition from fully laminar to fully turbulent flow,
a mixture of both laminar and turbulent flow is present and the boundary layer
thickness grows rapidly. Eventually the flow in the boundary layer becomes
fully turbulent and is characterised by a random, chaotic mixing of the fluid.
However, these turbulent effects are limited only to the region of the boundary
layer - we may still assume that the outer flow behaves like an inviscid fluid.

There is no fixed distance at which a laminar boundary layer changes into a


turbulent boundary layer, in fact under some conditions the laminar boundary
layer may disappear extremely quickly and be virtually undetectable. The onset
31

Level 2. Fluid mechanics

of transition depends on the ratio between viscous and inertial forces in the
boundary layer and so, unsurprisingly, the Reynolds number is significant when
characterising boundary layer flow. For experiments on flat plates, and under a
zero pressure gradient, values of Re 3900 (the Reynolds number here is
based on the boundary layer thickness ) are common for the transition point.
However this value should be treated only as a very rough guide. Generally
speaking, there is no unique value for Re at which transition from laminar to
turbulent flow will occur - among the factors that effect the onset of turbulence
are:
(i) Pressure gradient
(ii) Surface roughness
(iii) Heat transfer
(iv) Body forces
(v) Free stream disturbances
One of the most important factors above is the effect of the pressure gradient.
So far we have looked at an ideal flat plate with a zero pressure gradient,
however for most practical engineering problems, such as flow over a bluff
body or an aerofoil, pressure gradients greatly affect the nature of the boundary
layer.

3.2 The effect of a pressure gradient

The arguments introduced in the previous section assumed a zero pressure


gradient. However this is not normally the case. To introduce the concept of a
pressure gradient we shall look at a simple duct of varying cross-section:

32

Level 2. Fluid mechanics

Region 1

Region 2

Region 3

<0

=0

>0

U
Backflow

Region 1. p x < 0 : Favourable Pressure Gradient


In this region, the pressure behind a particle lying in the boundary layer is
greater than the pressure opposing the motion, i.e. the pressure gradient is
aiding the movement of the particle and the particle is, in effect, "sliding down a
pressure hill" without any danger of being slowed to zero velocity at a point
away from the wall. This is known as a favourable pressure gradient.
Region 2. p x = 0
We have already examined a simple flat plate for which the pressure gradient is
zero. Under these circumstances it can be shown that the velocity of the fluid in
the vicinity of the solid surface cannot be brought to zero - just like in the
favourable pressure gradient above. Therefore for a zero pressure gradient there
is no danger of flow reversal and subsequent separation.

33

Level 2. Fluid mechanics

Region 3. p x > 0 : Adverse Pressure Gradient


In region 3 we encounter what is known as an adverse pressure gradient. Here,
the pressure gradient retards fluid flow in the boundary layer and the fluid must,
effectively, "climb a pressure hill".

The adverse pressure gradient may

therefore bring to rest a fluid in the vicinity of the wall. On bringing a fluid to
rest we reach what is known as the point of separation. Just downstream from
the point of separation the flow nearest to the wall is reversed. Here, the low
energy fluid in the separated region is forced back upstream by the increased
pressure downstream.

Separation occurs only when p x > 0 .

However

separation does not always necessarily occur if p x > 0 , this depends on local
flow conditions.
We may now apply these concepts to a simple problem such as the flow over a
sphere:

Wake
Boundary
Layer

Point of
separation

For the sphere, a favourable pressure gradient exists over the front section and
so the boundary layer "sticks" to the wall of the sphere. As we progress over
the top of the sphere the pressure gradient changes to an adverse pressure
gradient.

Eventually, the flow separates from the sphere and a viscous,

turbulent, wake is formed. It is this turbulent wake which contributes to the


majority of the drag force on the sphere and so the point of separation is vitally
important in determining the behaviour of the flow around the sphere.
Similar arguments may be applied to any body immersed in an external flow
field and we shall look at this further in the study of drag forces towards the end
of this chapter.
34

Level 2. Fluid mechanics

3.3 Boundary Layer Thickness and Profile


The previous discussion developed the concept of a boundary layer and, in the y
direction, a defined beginning and end to the boundary layer was assumed (if
you like, the viscous area of influence of the wall). In this section we shall
quantify exactly what we mean by the thickness of a boundary layer. Typically
there are three different definitions for boundary layer thickness in common use,
these are
(i) The boundary layer disturbance thickness
(ii) The boundary layer displacement thickness
(iii) The boundary layer momentum thickness

Boundary layer disturbance thickness


The is also known simply as the boundary layer thickness (and was the
thickness alluded to in the previous section) and is defined as the distance from
the plate to a point in the fluid where the velocity is 0.99 times the free stream
velocity, i.e.
U

u(y)=0.99U
y

u(y)
x

i.e.

y =

when

u ( y ) = 0.99U

Obviously this definition is rather arbitrary - why choose 99%? - also this
boundary layer thickness is very difficult to measure accurately. To remove
these problems, the concept of the boundary layer displacement thickness was
introduced.

35

Level 2. Fluid mechanics

Boundary layer displacement thickness


This measure of boundary layer thickness seeks to compare the boundary layer
profile with the profile of an equivalent flow but with no viscosity present, i.e.
what is the effect of the viscosity of the fluid, over and above, that which would
be present if the flow was inviscid.

U
Equal
area

u(y)

U-u(y)

Inviscid (uniform)
flow

Boundary
layer flow

Thus the boundary layer displacement thickness is a balance between deficit in


the mass flow rate of the actual boundary layer flow and that of an equivalent
inviscid flow. And so if we equate mass flow rates we find that

U = (U u )dy
0

For incompressible flow

= 1

u
dy

Practically, one should truncate this integral at a suitable limit. Normally this is
chosen to be , and so

= 1

u
u
dy 1 dy

0
U
U

The displacement thickness represents the additional outward displacement of


the streamlines caused by the viscous effects of the plate.
36

Level 2. Fluid mechanics

Boundary layer momentum thickness


The boundary layer momentum thickness is often used when determining the
drag on an object. Flow retardation (caused by the object) within the boundary
layer results in a reduction in momentum within the boundary layer when
compared to inviscid flow. Thus the momentum thickness is defined as the
thickness of a layer of fluid, of velocity U, for which the momentum is equal to
the deficit of momentum through the boundary layer. This gives

u = u (U u )dy
2

For incompressible flow

u u
1 dy

0U
U

The boundary layer displacement and momentum thickness may appear rather
abstract. They are however far easier to evaluate accurately from experimental
data when compared to the boundary layer thickness . Therefore and are
far more commonly used when studying boundary layer properties.

The concept of displacement and momentum thickness may be extended


further, with the aid of the continuity and momentum equations, to the study of
drag on an object. As in the case of internal flows, exact solutions exist for
laminar boundary layer drag.

Analysis may also be extended to turbulent

boundary layers by use of the velocity power law discussed in Section 2.2.3.
We shall look briefly in the next section at how we compute velocity profiles
within the boundary layer region, the theoretical analysis of drag is however
rather complex and beyond the scope of this course. Instead, when studying
drag, we shall focus on empirical methods.

37

Level 2. Fluid mechanics

3.3.1 Velocity profiles in the boundary layer


In general, the shape of a velocity profile will depend upon the nature of the
pressure gradient present. For adverse pressure gradients, the velocity profile
may take on many different forms. However for a zero pressure gradient, the
typical velocity profile for both laminar and turbulent flow is, as you would
expect, similar to that seen for internal flow. If we look first at laminar flow,
then based on the analysis in Chapter 2, the most obvious velocity profile to
choose is

u=

Uy

or

where is the boundary layer thickness. This assumes that the theory for two
parallel plates, one of which is moving with a velocity U is valid here also. The
equation above does of course predict a linear velocity profile however this is
not always a good approximation for actual laminar boundary layer profiles.
Instead cubic, parabolic, or even sinusoidal velocity profiles have been
suggested as a better fit for an external laminar boundary layer.

For turbulent boundary layers the power law velocity relation discussed in
Chapter 2 provides an adequate correlation:
1

y n
=
U
u

and so this equation is equally applicable to external boundary layers.

comparison of (non-dimensional) velocity profiles for a laminar and turbulent


boundary layer are shown, for a zero pressure gradient, below:

38

Level 2. Fluid mechanics

1
Laminar (linear)

Laminar (parabolic)

Turbulent n=7
0

uU

Note here that both linear and parabolic profiles are shown for laminar flow (the
parabolic profile generally provides a better fit for actual data). It is evident
here that, as seen in the previous chapter, the turbulent profile is much fuller
(more blunt) than the laminar profile.

3.4 Drag and Lift

On flowing over a solid body, a viscous fluid will exert a net force on that body.
This force is made up of the wall shear stresses w and the pressure force p (it is
this pressure force that we indirectly discussed when looking at pressure
gradients). For example, if we consider the (gauge) pressure and the shear
stress distribution over an aerofoil we find, typically, the following type of
behaviour:
p<0

Pressure

p>0
Shear Stress
U

39

Level 2. Fluid mechanics

The total drag force acting on the aerofoil is a summation of the pressure force
and the shear stress. Of course, to fully compute both sets of forces we should
need to know in detail both the local pressure and the shear stress distribution at
all points around the aerofoil. We have already seen that in the presence of an
adverse pressure gradient (present here over most of the upper surface of the
aerofoil) flow may separate and very complex flow patterns may develop. Thus
calculating pressure and shear forces theoretically, even for very simple objects,
is usually prohibitive.

Therefore, for most shapes of interest we resort to

experimental measurements and to do this we use lift and drag coefficients.

3.4.1 Drag
The drag force is the component of the force acting on a body which acts
parallel to the direction of motion:
U

Drag Force FD

The drag force consists of pressure and shear forces acting on the surface of the
object. The magnitude of the drag force tends to depend strongly on the shape
of the object. The drag force for any body immersed in an incompressible flow
is given by

1
FD = C D U 2 A
2
where CD is known as the drag coefficient and is obtained from experimental
measurements. Of course we may write also that

CD =

FD
1
2

40

U 2 A

Level 2. Fluid mechanics

The number "1 2 " is inserted to form the familiar dynamic pressure.
Thus to find the drag force acting on a body we need only identify CD for that
particular body. Of course it should be remembered that the nature of the flow
over the immersed body will depend heavily on the Reynolds number and so CD
is not a constant. Instead we should be careful to identify the correct value for
CD for a given Reynolds number. Some drag coefficient for a few objects are
given below:

Shape
Semicircular
shell
Semicircular
cylinder

D
D

Reference Area
A
(b = length)

Rectangle

Reynolds
Number

A = bD

2.3
1.1

Re = 2 10 4

A = bD

2.15
1.15

Re > 104

a
D

Drag Coefficient

A = bD

Sold
Hemisphere

A=

Thin Disk

A=

CD
1.9
2.5
2.9
2.2
1.6
1.3

Re = 105

D2

1.17
0.42

Re > 104

D2

1.17

Re > 103

a/d
0.1
0.5
0.65
1.0
2.0
3.0

Note here that the objects listed above have sharp edges and in most cases the
drag coefficients may be listed as constant for a range of Reynolds numbers.
This is because the separation point is fixed by the geometry of the sharp edge

41

Level 2. Fluid mechanics

and so drag coefficients tend to be independent of the Reynolds number above


values of Re of approximately 1000. This is not the case for blunt objects otherwise known as bluff bodies - for examples objects such as a sphere or a
cylinder. The drag coefficient for flow over a sphere is shown below:

100

CD

10

0.1
-1

10

10

10

10

Re

At very low Reynolds numbers (Re<1) there is no fluid separation from the
sphere; the wake is laminar and the drag is predominantly composed of friction
drag in a laminar boundary layer. Stokes was able to show, theoretically, that in
this region
CD =

24
Re

As Re is increased CD drops continuously, but the relative contribution of the


pressure drag increases until at Re 1000 the pressure drag contributes
approximately 95% of the total drag.

In the region 103 < Re < 2 105 the drag coefficient curve is relatively flat and
the point of separation of the flow is just upstream of the sphere midsection. At
a value of Re 2 105 a sudden reduction in drag occurs. This is caused by the
point of separation suddenly moving downstream of the sphere midsection and
42

Level 2. Fluid mechanics

so the size of the wake decreases. The point of separation moves from just
upstream to just downstream of the sphere midsection because the boundary
layer changes abruptly from being laminar to being turbulent. A turbulent
boundary layer possess more momentum than does a laminar boundary layer
and so is greater able to resist the adverse pressure gradient.

The latter point is an important one: turbulent boundary layers - by virtue of


their increased momentum when compared to laminar boundary layers - are
more desirable on a bluff body because this delays the point of separation and
thus reduces the pressure drag. Thus the drag coefficient with a turbulent
boundary layer may be up to 5 times less than that for a laminar boundary layer
on either side of a critical Reynolds number (for this example Re critical 4 105 ).
A classic example of harnessing this effect is the "dimples" inserted on a golf
ball. The dimples are designed to speed up the onset of turbulence in the
boundary layer. The turbulent boundary layer then resists separation and thus
reduces the wake and minimises pressure drag.

A detailed knowledge of the likely flow patterns over a body may therefore
provide clues as how to best minimise drag. Work in this area normally relies
on delaying the onset of separation by streamlining the body shape.

The

objective of streamlining being to reduce the adverse pressure gradient that


occurs behind the point of maximum thickness on the body. Of course one
must be careful not to increase by too large a margin the overall surface area of
the body since skin friction may begin to appear again and outweigh any
reductions achieved for the pressure drag.

43

Level 2. Fluid mechanics

3.4.2 Lift
Lift is the component of force acting perpendicular to the fluid motion. The lift
force is defined in the same way as the drag force, i.e.

1
FL = C L U 2 Ap
2
or
CL =

FL
1

U 2 Ap

2
where Ap is the (planform) area projected at right angles to the flow. For an
aerofoil the projected area Ap depends on the angle of attack , and so the lift
force changes with . An example of the coefficient of lift, plotted for an
aerofoil, is given below:

CL

Here, one can clearly see where the wing stalls and the coefficient of lift drops
off rapidly.

44

Level 2. Fluid mechanics

3.5 Example Problems


Q1.
A rotating mixer is constructed from two circular disks. The mixer is rotated at
60 rpm in a large vessel containing a brine solution (specific gravity = 1.1).
If the drag on the rods and the motion induced in the liquid may be neglected,
estimate
(i) The maximum torque, and
(ii) the power required to drive the mixer.
=60 rpm
d=100 mm

0.6 m

0.6 m

[Take the drag coefficient for each disk to be C D = 1.17 ]


Now the mixer will be working against the drag force imparted by each circular
disk. We saw earlier that
1
FD = C D U 2 A
2
Now

U = r

therefore

U = 0.6

60 2

= 3.77 m/s

60

0.12

FD = 1.17 1.11000 3.77


2
4
FD = 71.83 N

Torque

T = 2 Fr = 2 71.83 0.6 = 86.2 Nm


P = 2 Fu

Power

P = 2 71.83 3.77 = 541.6 W


45

Level 2. Fluid mechanics

Q2.
A dragster weighing 700 kg achieves a speed of 380 km/hr after travelling 0.5
km. Immediately after passing through the timing lights the driver opens the
drag chute. The area of the chute is 2.5 m2 and it has a constant drag coefficient
of C D = 1.2 . If the air and the rolling resistance of the car may be neglected
(i.e. the chute dominates the drag) find the time required for the machine to
decelerate to 150 km/hr.
[Take air = 1.2 kg/m 3 ]
If we draw a free body diagram of the car:
FD

u
y
x

ui = 380 km//hr

u f = 150 km//hr

Applying Newton's second law:


FD = ma = m

du
dt

now
1
FD = C D u 2 A
2
and so
1
du
CD u 2 A = m
2
dt
This expression must now be integrated, which may be accomplished after first
separating the variables, i.e.

46

Level 2. Fluid mechanics

uf

1
C D A dt = m du
2
o
ui u
2
thus
uf

1
C D At = m
2
u ui
1

C D At =

2m

t=

t=

uf

1
ui

2m 1 1

CD A u f ui

1
1

3600
1.2 1.2 2.5 150,000 380,000

2 700

t = 5.65 seconds
Q3.
The laminar and turbulent velocity profiles derived in Chapter 2 may be applied
also to the study of the boundary layer profiles discussed here in Chapter 3.
(i) Taking a simple linear velocity profile for laminar flow, evaluate the ratio

Now for a linear laminar velocity profile we have


u

and the momentum thickness was defined as


u u
= 1 dy

47

Level 2. Fluid mechanics

terminating the integral at , we have

y2
y y
y3

=
= 1 dy =

2 2 2 0 6

thus

= 0.1667

(ii) Evaluate for a 1 7 power law profile (i.e. n=7) used to represent
turbulent flow:
1

u y 7
=
U
1
2

17
y 7 y 7
y
y 7

= 1 dy = 1 2 dy

0
0 7
7

8
9
7 y 87 7 y 97
7 7
7 7 7 7
=
= 1

=
2
1
2
8 7 9 7
8 7 9 7
8 9
0

= 0.097

A smaller value for suggests a "flatter" more uniform velocity profile.

48

Anda mungkin juga menyukai