Airflow
by
Emeritus Professor E. Markland
(with minor additions by TecQuipment)
TecQuipment Ltd 2009
Do not reproduce or transmit this document in any form
or by any means, electronic or mechanical, including
photocopy, recording or any information storage and
retrieval system without the express permission of
TecQuipment Limited.
TecQuipment has taken care to make the contents of this
manual accurate and up to date. However, if you find any
errors, please let us know so we can rectify the problem.
TecQuipment supply a Packing Contents List (PCL) with
the equipment. Carefully check the contents of the
package(s) against the list. If any items are missing or
damaged, contact TecQuipment or the local agent.
EM/PE/djb/0409
CONTENTS
AUTHORS PREFACE
i
NOTE ON UNITS
iii
CHAPTER 1. A Brief Introduction to Airflow
1
CHAPTER 2.
The AF10 Airflow Bench
11
CHAPTER 3.
The AF10A Multitube Manometer 15
CHAPTER 4.
Bernoullis Equation applied to a ConvergentDivergent
Passage (AF11)
19
CHAPTER 5. Drag Measurement on Cylindrical Bodies (AF12)
31
CHAPTER 6. The Round Turbulent Jet (AF13)
53
CHAPTER 7. Boundary Layers (AF14)
69
CHAPTER 8. Flow Around a Bend in a Duct (AF15)
89
CHAPTER 9. Jet Attachment (AF16)
99
CHAPTER 10. Flow Visualisation (AF17)
109
CHAPTER 11. Aerofoil with Tappings (AF18)
117
APPENDIX A. Determination of the Area Beneath a Curve
A1
APPENDIX B. Installation and Operating Instructions for the Flow
Visualisation Apparatus
A7
AUTHORS PREFACE
It is now over forty years since the set of experiments which I devised for a first year
hydraulics laboratory were put together with a laboratory manual to form an
instructional package. The continuing demand for this equipment is gratifying
evidence of the need for simple experiments to illustrate principles of fluid motion
and to allow students to become familiar with elementary laboratory techniques in
hydraulics.
In many educational establishments the subject is taught as fluid mechanics,
encompassing flow of gases as well as liquids, and a complementary set of
experiments in air flow is now presented to meet laboratory needs which are
generated by this more general approach. Again the equipment has been made
simple, with the intention of facilitating experiments which emphasise the basic
principles. In many cases there is nevertheless scope for extending the tests reported
here into project work, if so desired. I have also found the experiments useful for
classroom demonstration during lectures in a first course in fluid mechanics.
Again I have adopted the style of the laboratory report, usually introduced by some
remarks on the usefulness and significance of the subject, to describe the experiments.
Some teachers will find the guidance far too detailed, and others will find alternative
ways of using the equipment which will suit their purposes better. I hope that,
whatever the mode of use, the equipment will meet the expectations of those who
work with it.
Emeritus Professor E. Markland
D.Sc. C.Eng. F.I.C.E. F.I.Mech.E. Member ASME
i
ii
NOTE ON UNITS
Throughout this book we use the International System of Units (SI) in which the units
of mass, length and time are:
mass kilogram (kg)
length metre (m)
time second (s)
In this system the unit of force, defined as that force which when applied to a unit
mass of 1 kg produces an acceleration of 1 m/s
2
, is the
newton (N)
The newton may be expressed in terms of unit of mass, length and time by the
equation
1 N = 1 kg m/s
2
(1)
which follows from the definition given above.
The density of a fluid at a point is defined as the ratio of mass to volume of an
element surrounding the point, so the SI units of density are
density kg/m
3
Pressure p at a point in a fluid is defined in terms of the normal force acting on an
element of a plane surface through the point; the pressure is the ratio of force to area.
The units are therefore
pressure p N/m
2
and, expressing N in terms of kg, m and s from Equation (1) we find
1 N/m
2
= 1 kg/m s
2
iii
Pressure may thus be written in terms if unit mass, length and time, as
pressure p kg/m s
2
The term V
2
, the socalled dynamic pressure, occurs frequently. If is expressed
in units of kg/m
3
and V in units of m/s, then V
2
appears in units of kg/ms
2
, i.e. in
the same units as those of p.
The SI system provides for the use of particular names for certain derived units, and
the pascal, represented by the symbol Pa, is used to represent a pressure of 1 N/m
2
,
i.e.
1 Pa = 1 N/m
2
For many purposes this unit is rather small, and another unit, the bar, is used. This is
defined by:
1 bar = 10
5
Pa = 10
5
N/m
2
It so happens that 1 bar is roughly equal to the pressure of the atmosphere at sea level,
so for meteorological practice, the millibar, written mbar (or simply mb), is a further
convenient unit. So we have:
1 mbar = 10
3
bar = 10
2
Pa = 10
2
N/m
2
It is recommended, that students become familiar with the derived units of pressure
mentioned. However, in this book the N/m
2
is adopted as the preferred unit,
requiring no effort of memory to relate it to the fundamental units of mass, length and
time.
Some readers may be more familiar with the Imperial system of units in which the
fundamental units of mass, length and time are:
pound lb
foot ft
second s
iv
v
vi
These are related to SI units by the conversion factors
1 lb = 0.453 592 37 kg exactly
or 0.4536 kg for most purposes
1 ft = 0.3048 m exactly
The unit of force, defined as that force which when applied to a unit of mass 1 lb
produces an acceleration of 1 ft/s
2
is the poundal, pdl.
The conversion factor for poundal to newton may be calculated from the foregoing
conversions for mass and length units as follows:
1 pdl = 1 lb ft/s
2
= 0.4536 kg 0.3048 m/s
2
= 0.1383 kg m/s
2
= 0.1383 N
Frequently the alternative unit of force, the poundforce, abbreviated to lbf, is used; it
is defined as that force which when applied to a unit mass of 1 lb produces the
standard acceleration of gravity, viz. 32.174 ft/s
2
. The pound force is therefore larger
than the poundal, the conversion factor being
1 lbf = 32.174 0 pdl
and the conversion from lbf to N is
1 lbf = 32.1740 pdl
= 32.1740 0.1383 N
= 4.448 N
Pressures in the Imperial system are frequently measured in units of pdl/ft
2
, lbf/ft
2
or
lbf/in
2
, and the reader should verify the following conversions
1 pdl/ft
2
= 1.488 N/m
2
1 lbf/ft
2
= 47.88 N/m
2
1 lbf/in
2
= 6895 N/m
2
1. A BRIEF INTRODUCTION TO AIRFLOW
The concept of continuity and Bernoullis equation are fundamental to an elementary
understanding of airflow, and these matters are dealt with in every textbook on fluid
mechanics, hydraulics or aerodynamics. For the sake of completeness and to
introduce some of the terms used in subsequent chapters, a brief outline is given here.
The fluid is considered as a continuous medium, so that if we fix attention at one
point of the space in which a fluid is in motion, we can observe fluid streaming
continuously through that point, and the fluid velocity at the point may conveniently
be represented in magnitude and direction by a vector. This is conveniently called the
velocity vector, and the whole motion is described by the set of velocity vectors at all
points of that space by the velocity field. If we now draw a line in the fluid so that at
every point of the line the tangent is in the direction of the velocity vector, we have a
streamline of the fluid, as illustrated in Figure 1.1.
Figure 1.1 Definition of a Fluid Streamline
Bernoullis equation deals with the case where the velocity field is steady, by which
we mean that the velocity at each point of the motion does not vary with time.
Consider now a bundle of streamlines as shown in Figure 1.2. Since the direction of
the fluid velocity at each point in the surface defined by the streamlines lies along the
surface, no fluid crosses it and the fluid contained inside the bundle may be thought of
as flowing in a tube, a socalled streamtube.
1
Figure 1.2 A Streamtube
Consider a section across a streamtube of infinitesimal area A. If the velocity along
the tube at this section is u, then in unit time the volume of fluid which passes across
the section is uA. The mass of this fluid is uA. This is the rate of mass flow
across the section, and since no fluid crosses the walls of the streamtube,
& m
& m = uA = constant along the infinitesimal streamtube
(11)
This is one form of the equation of continuity of flow. Various other forms may be
derived as shown below. If the streamtube has a finite crosssectional area, then the
rate of mass flow is:
& m = = constant along the streamtube udA
A
(12)
where the integration is taken over the area A of the crosssection. For the particular
case in which and u are both constant over the sectional area,
& m = uA = constant along the streamtube
(13)
2
It is sometimes convenient to deal in terms of volume flow rate. For the infinitesimal
tube the volume Q crossing the elementary area A per unit time is:
Q = uA
(14)
and this is constant along the streamtube only if is constant along it. For a
streamtube of finite crosssectional area A the volume flow rate is:
Q = udA
A
(15)
and for the particular case of constant velocity over the section
Q = uA
(16)
Again, Q is constant along the streamtube only if is constant along it.
Figure 1.3 Forces Acting on an Element of a Streamtube
3
Consider now steady motion of a fluid along an elementary streamtube. Figure 1.3
shows an element of length s and the forces acting on it.
The pressure rises from p to p
dp
ds
s + along the element, and the crosssectional area
increases from A to A . The forces due to pressure on the element are:
dA
ds
s +
On the section at s: pA in the sdirection
On the section at s + s: +
p
dp
ds
s A
dA
ds
s in the sdirection
On the wall of the streamtube: the pressure varies from p to p
dp
ds
s + ,
and the projected area of the wall on a plane normal to the sdirection is
A
dA
ds
s A +
, viz.
dA
ds
s . So the component of force in the sdirection lies
between:
p
dA
ds
s and p
dp
ds
s
dA
ds
s +
To first order of infinitesimal quantities this is:
p
dA
ds
s
The net force in the sdirection due to pressure is thus:
pA p
dp
ds
s A
dA
ds
s p
dA
ds
s +
+
which reduces simply to
A
dp
ds
s
4
to first order of infinitesimals. The force due to the weight of the fluid is:
gAs
acting downwards, and if the zdirection is taken vertically upwards, this is
gA s
in the zdirection. The component of this in the sdirection is
gA
dz
ds
s
The mass of fluid within the element is As. The scomponent of fluid acceleration
along the streamline may be derived by considering the velocity change u over the
length s of the element, which is
u =
du
ds
s
The time t in which this velocity change takes place is the time required for fluid to
travel the distance from one end, where the speed is u, to the other, where the speed is
u
du
ds
s +
So, to first order of small quantities, the time t is
t =
1
u
s
The scomponent of acceleration a
s
is therefore
a
s
=
u
t
du
ds
s
u
s
=
1
5
or
a
s
= u
du
ds
Equating the massacceleration of the fluid to net force, the equation of motion in the
sdirection is therefore
A s u
du
ds
A
dp
ds
s gA
dz
ds
s =
which leads to the result
u
du
ds
dp
ds
g
dz
ds
+ + =
1
0
(17)
Provided that is constant, this equation may be integrated to give
1
2
2
u
p
gz E + + =
(18)
where E is a constant. Multiplication by gives
1
2
2
u p gz P + + =
(19)
and division by g gives
u
g
p
g
z H
2
2
+ + =
(110)
both of which are forms of Bernoullis equation. The constant P is called the total
pressure and H the total head. It should be noted that the result has been derived for
steady motion of a fluid under pressure and gravity forces  shear force due to viscous
action on the wall of the streamtube has been neglected and that the density has been
assumed constant, so the fluid has been assumed incompressible. Note also that the
6
integration has been with respect to s, that is, in the sdirection along the streamline.
The result may thus be stated in words:
The total pressure (or total head) is constant along a streamline in steady motion
of an inviscid, incompressible fluid.
The equation says nothing about the way that the total pressure changes from one
streamline to another.
The first term in Equation (19) is dependent on fluid velocity u, so we refer to:
u
2
as the velocity pressure or dynamic pressure.
The remaining terms depend on pressure p and elevation z, and we refer to:
p + gz as the piezometric pressure.
When the working fluid is air, the static pressure p is much more important than gz.
We shall see that changes in p in the experiments described later are typically about
1000 N/m
2
. For a change of elevation of 1 m, the corresponding change in gz is
about 10 N/m
2
. So in practice, Bernoullis equation for air is frequently written as:
1
2
2
u p P + =
(111)
since the contribution of gz is usually negligible. The equation is strictly valid of
course if p is now taken to indicate piezometric pressure.
We have seen that Bernoullis equation applies only to a fluid of constant density, and
it has been mentioned in the previous paragraph that it may be applied to air, for
which the density clearly changes with temperature and pressure. Is it possible to
make the constant density assumption for air? By an analysis which is beyond our
present scope, it may be shown that P generally exceeds
1
2
2
u p + by an amount
which increases as the velocity increases. The governing parameter in the Mach
number Ma, defined as the ratio of velocity u to the local velocity of sound a, so that:
Ma =
u
a
7
Some numerical values, calculated for airflow at 15C, for which the value of a is
340 m/s, are shown in Table 1.1. Now according to Equation (111), the value of the
quantity tabulated in the last column of the table should be exactly 1, so the amount
by which the numbers in this column exceed unity may be taken as an indication of
the error in Equation (111) due to the compressibility of the air. In the experiments
described in the following chapters, the air velocity rarely exceeds 50 m/s, so the
compressibility error is no more than 0.5%, and the flow may justifiably be regarded
as incompressible.
u (m/s) Ma = u/a (P p) /u
2
50
100
200
300
340
0.147
0.294
0.588
0.882
1
1.005
1.022
1.089
1.210
1.276
Table 1.1 Calculated Difference between Total Pressure and Static
Pressure as a Function of Mach Number
The air density may be calculated from the barometric pressure p and temperature T
from the gas equation
p
= RT
(113)
in which the value of the gas constant R for dry air is
R = 287.2 J/kg K
(114)
or
R = 287.2 Nm/kg K
(114a)
J indicates the SI unit of energy, the joule (which is identical with the newton metre or
Nm) and K indicates the unit of temperature, the kelvin. If t represents temperature
in C, then
8
T = t + 273.15
(115)
and from Equations (113), (114) and (115)
=
p
t 287 2 27315 . ( . ) +
kg/m
3
(116)
In this equation, p is expressed in N/m
2
and t in C. Some typical values of are
given in Table 1.2.
Pressure p 10
5
(N/m
2
)
Temperature t
(C)
Density
(kg/m
3
)
0.95
10
15
20
25
1.168
1.148
1.128
1.109
1.00
10
15
20
25
1.230
1.208
1.188
1.168
1.01325
10
15
20
25
1.246
1.224
1.203
1.183
1.05
10
15
20
25
1.291
1.269
1.247
1.226
Table 1.2 Density of Dry Air
Note that the pressure 1.01325 10
5
N/m
2
is the pressure of a standard atmosphere.
In the event of the barometric pressure being given in mm of mercury, this may be
converted to units of N/m
2
by the hydrostatic relationship
p gh =
(117)
9
which gives the pressure p due to a column of height h of liquid of density , acted on
by gravity g. Using the value 13590 kg/m
3
for the density of mercury, and the value
9.807 m/s
2
for the standard acceleration of gravity, then 1 mm of mercury produces a
pressure p given by
p = 13590 9.807 (1/1000)
or
p = 133.3 N/m
2
So,
1 mm Hg = 133.3 N/m
2
(118)
The experiments described in this book use a manometer that registers air pressures in
terms of millimetres of water gauge. It is therefore useful to establish the relationship
between the velocity pressure u
2
of an airstream, and the corresponding water
gauge reading h mm. For air of standard density = 1.224 kg/m
3
,
1.224 u
2
= 1000 9.807 (h/1000)
or
u = 4.00 h m/s
(119)
10
2. THE AF10 AIRFLOW BENCH
AF10 Airflow Bench (shown with the AF10A)
11
Many readers will know that much of the research and development work in
aerodynamics is done in wind tunnels, which provide the controlled flow of air
required for tests. Some of these tunnels are large and complex, representing major
engineering achievements in their own right, and require considerable power to
operate. The AF10 Airflow Bench is in the nature of a simple miniature wind tunnel;
it provides a controlled airstream for experiments which use matching test equipment.
The bench and associated experiments are intended for use in conjunction with lecture
courses in fluid mechanics or aerodynamics, particularly in the early stages. The
experiments have generally been devised so that they may be set up for classroom
demonstration to illustrate a lecture topic. In many cases a simple qualitative
demonstration of an effect may be all that is needed; in other cases the lecturer may
take one or two specimen results and use the data as the basis for a worked classroom
example. Alternatively the experiments may be built into a formal laboratory
programme of prescribed work in the subject. The expositions presented in this book
are in the style of formal reports, as these provide a convenient means of developing
the relevant theory and of describing the test apparatus, its range and its capabilities.
Those lecturers who wish to use the equipment for project work should have little
difficulty in formulating suitable projects; certain questions and suggestions are
included which will be helpful in this regard.
The airflow bench and its associated equipment is in many ways complementary to
the hydraulics benches which have been manufactured for many years by
TecQuipment. As the basic concepts of hydraulics and of incompressible airflow are
identical, some of the experiments on the airflow bench bear marked similarities to
their counterparts on the hydraulics bench, although the emphasis might be somewhat
different. For example, the Bernoulli theorem experiment in airflow corresponds to
the Venturi meter in the range of hydraulics equipment. On the other hand, an
experiment on boundary layers is probably more relevant in a first course on
aerodynamics than in a first course in hydraulics.
The AF10 Airflow Bench comprises a fan which draws air from the atmosphere and
delivers it along a pipe to an airbox which is above the test area. In the pipe is a valve
which may be used to regulate the discharge from the fan.
There is a rectangular slot in the underside of the airbox to which various contraction
sections may be fitted. The air accelerates as it flows from the box along the
contracting passage, and any unsteadiness or unevenness of the flow at the entry
12
becomes proportionately reduced as the streaming velocity increases towards the test
section, which is fitted at the exit of the contraction. Discharge from the test section
is in most cases directed towards the bench top, in which a circular hole is provided to
collect the air so that it may be led through a duct to the rear of the bench. If
necessary, the exhaust can be taken right out of the laboratory (for example, if, use is
to be made of smoke traces) with the exhaust duct extended as necessary and an
extractor fan fitted at the downstream end if required.
The bench is mounted on wheels with jacking screws so that it may be moved without
difficulty. It requires an earthed, AC singlephase electrical supply.
13
14
3. THE AF10A INCLINABLE MULTITUBE MANOMETER
AF10A Inclinable Multitube Manometer
15
The AF10A is a 14 limb multitude manometer, as shown in Figure 3.1.
Figure 3.1 AF10A Inclinable Multitube Manometer
The reservoir for the manometer liquid is mounted on a vertical rod so that it may be
set to a convenient height. It is recommended that the manometer tubes at the two
sides, marked A in Figure 3.1, and the reservoir connection, be normally left open to
atmospheric pressure. Pressures p
1
, p
2
, p
3
... in tubes 1, 2, 3 are then gauge pressures,
measured relative to an atmospheric datum. (Pressures relative to some other datum
may be obtained by connecting the reservoir and the manometer tubes marked A to
the required datum).
The usual manometer liquid is water, although in some instances a paraffinbased
liquid of low specific gravity is used. To aid visibility, the water may be coloured by a
dye which is supplied with the equipment. The specific gravity of the water is not
significantly altered by addition of the dye. To fill, the reservoir is positioned about
halfway up the bar, and the fitting at the top is unscrewed. Using the funnel provided,
manometer liquid is poured in until the level is halfway up the scale. Any air bubbles
from the manometer tubes are then removed by tapping the inlet pipe, or by blowing
into the tops of the tubes.
The manometer scale is usually graduated in mm. Pressure readings taken in terms of
mm of water may be converted to units of millibar (mb) from the relationship:
1 mm water = 0.0981 mb
16
The manometer may be used vertically, or, for increased sensitivity, inclined at some
suitable angle to the vertical. Two predetermined settings are provided:
0, giving a scale magnification of 1.0 (reading 1);
60, giving a scale magnification of 2.00 (reading 0.5);
78, giving a scale magnification of 5.00 (reading 0.2).
Scale readings are therefore to be divided by factors of 2.00 and 5.00 respectively to
obtain equivalent readings on a vertical scale. The manometer is sufficiently accurate
and sensitive for all of the experiments described in this book.
The manometer must be levelled before taking readings. This can be done by using
the adjustable feet, while observing the spirit level and the manometer liquid levels
across all of the tubes under static conditions.
It is possible that, as the air speed is increased, liquid may be driven out of the tops of
the manometer tubes, or drawn down into the manifold at the base. The connection
between tapping points and the manometer would then have to be cleared, or the
reservoir may need to be refilled. It is therefore advisable, before starting a test, to
guard against these eventualities by adopting the following settingup procedure. With
the fan at rest and the bench valve closed, the manometer should be set to the vertical
position, with the liquid level at about midheight. The fan should then be started, and
the air speed raised gradually by carefully opening the bench valve, while observing
the levels in the manometer tubes. As the pressures in the various tubes change, the
reservoir level should be moved up or down, as found to be necessary to keep all the
liquid levels within the bounds of the scale. A good setting would use most of the
scale at full airspeed. If, however, only a small proportion of the scale is used, the
procedure should be repeated with the manometer inclined to the vertical.
To fill the manometer position the reservoir approximately halfway up the side bar.
Unscrew the fitting on top of the reservoir and, using the funnel provided, pour in a
quantity of water (and dye if required). Continue until the water level is halfway up
the manometer scale. Check the system for air bubbles, and remove by tapping the
inlet pipe, or by gently blowing into the manometer tube at the top.
Having decided on a suitable manometer setting, a final height adjustment of the
reservoir should then be made to bring the datum reading at tubes A to some
convenient scale graduation  such as, for example, 120 mm. This is the value which
17
18
has to be subtracted from the scale readings of the pressures p
1
, p
2
... shown in Figure
3.1 to obtain gauge pressures. It is much easier to perform the subtraction with a
datum that has been conveniently chosen.
4. BERNOULLIS EQUATION APPLIED TO A CONVERGENT
DIVERGENT PASSAGE
AF11 Bernoullis Equation Apparatus
19
Introduction
This experiment demonstrates the use of a Pitotstatic tube, and investigates the
application of Bernoullis theorem to flow along a convergentdivergent passage.
Description of Apparatus
Figure 4.1 Arrangement of the Apparatus for an Experiment on
Bernoullis Equation
A duct of rectangular section is fitted to the exit of the contraction which leads from
the airbox, and liners placed along the inside wall of the duct produce a passage
which contracts to a parallel throat and then expands to the original width. The shape
20
of this convergentdivergent passage is indicated in Figure 4.1, from which it may be
noted that the convergent portion is shorter than the divergent portion. Air is blown
through the passage, and a probe may be traversed along the centre line to measure
the distribution of total pressure P and static pressure p. This probe is a Pitotstatic
probe. Pressure tappings are connected from the airbox and from the Pitotstatic
probe to a multitube manometer.
Theory
The aim of the experiment is to measure the distribution of total pressure P and static
pressure p along the duct and to compare these with the predictions of Bernoullis
equation. Consider how the equation is applied to the present case. Figure 4.2 shows
the duct as a stream tube.
Figure 4.2 Measurement of Total and Static Pressure
21
According to Bernoullis equation the total pressure P, defined by
P u = +
1
2
2
p
A
(41)
should be constant along this tube, provided the flow is steady and that the air is
incompressible and inviscid. If P
o
denotes the total pressure in the airbox, then we
should expect the measured value of P along the passage to be the same everywhere
as P
o
, if Bernoullis theorem is valid for this motion.
Now the total pressure P is measured with comparative ease by an open ended tube
facing the flow. Figure 4.2 shows a streamline starting from the airbox, passing the
duct, and arriving at the mouth of the Pitot tube. The motion is arrested at this point,
so that in Equation (41) the local value of u is zero.
The pressure recorded by the Pitot tube is therefore the local value of total pressure P.
If Bernoullis equation applies along the whole length of the streamline from the
airbox, then P should be the same everywhere as the initial total pressure P
o
. The
value of P
o
may be found easily from a pressure tapping in the wall, since the air
velocity in the box is so slight as to make the difference between total pressure and
static pressure quite negligible.
The variation of static pressure p may be measured by the static pressure tube.
Figure 4.2 shows a further streamline emanating from the airbox and flowing close to
the surface of the probe. Provided that the holes in the surface of the probe are placed
far enough from the tip of the tube as to be unaffected by the disturbance in this
locality (which means in practice about 6 tube diameters away from the tip) then the
flow is undisturbed by the holes, which measure the undisturbed pressure, which is
the static pressure, p. To compare the measured values of p with the results of
calculations we must use the continuity equation as well as the Bernoulli equation.
Taking the flow as onedimensional, that is assuming the velocity over any chosen
crosssection to be uniform over that section, then the continuity equation for
incompressible flow gives the volume flow rate as:
Q uA u
t t
= =
(42)
22
(The suffix t indicates conditions at the throat). The velocity distribution along the
duct may thus be written in the form of the ratio:
u
u
A
A
t
t
=
(43)
and since the depth of the duct is constant, and the crosssectional area is proportional
to the width, then
u
u
B
B
t
t
=
(44)
The velocity ratio following from continuity may therefore be calculated simply from
the dimensions of the convergentdivergent passage. This now may be compared
with the velocity ratio inferred from pressure distribution using Bernoullis theorem.
For Equation (41) this gives the local velocity as:
u
P p
=
2( )
(45)
and in particular the velocity u
t
at the throat is
u
P p
t
t t
=
2( )
(46)
so from Equations (45) and (46)
u
u
P p
P p
t t
=
t
(47)
The righthand side of this equation may be evaluated from the measured pressure
distribution and compared with the values from Equation (44).
23
Results
Air temperature 22C = 295 K
Barometric pressure 1028 mb = 1.028 10
5
N/m
2
The profile of the convergentdivergent passage is shown in Figure 4.3.
Figure 4.3 Dimensions of a ConvergentDivergent Passage
In Table 4.1, measurements of P
o
, P and p are recorded as the probe traverses along
the duct. These pressures are gauge pressures, which are measured relative to
atmospheric pressure. Note that the readings of P and p in a single line of the table do
not represent the same physical position of the probe, because the static pressure holes
lie 25 mm downwind of the tip. By measuring pressures at longitudinal spacings of
12.5 mm and 25 mm, P and p are obtained at identical stations but at different probe
settings. The initial value, x = 4 mm, was a convenient starting point with the
particular equipment under test and may vary somewhat from one test rig to another.
24
x
(mm)
P
o
(N/m
2
)
P
(N/m
2
)
p
(N/m
2
)
B
B
t
P p
P p
t t
4
16.5
29
41.5
54
66.5
79
91.5
104
129
154
179
204
229
254
279
304
800
800
800
800
800
800
800
800
800
800
800
800
800
800
800
800
800
795
790
785
790
790
785
790
790
790
790
785
785
780
775
780
780
780
195
35
130
315
540
805
945
980
980
845
630
480
355
255
170
95
25
0.593
0.643
0.701
0.772
0.857
0.965
1.000
1.000
1.000
0.946
0.867
0.801
0.744
0.694
0.651
0.613
0.440
0.582
0.653
0.719
0.790
0.867
0.948
0.990
1.000
1.000
0.961
0.894
0.845
0.801
0.763
0.733
0.703
0.653
Table 4.1 Total and Static Pressure Distributions
Any convenient starting value may be chosen, the subsequent calculations being
changed accordingly. The values of B
t
/B are calculated from the known dimensions
of the contraction. For example, in the converging section, when x = 29 mm
B = = 76 76 44
29
70
62 7 ( ) .
So
B
B
t
= =
44
62 7
0 701
.
.
and in diverging section, when x = 204 mm
B = = 76 76 44
100
190
59 2 ( ) .
So
B
B
t
= =
44
59 2
0 744
.
.
25
Discussion of Results
Figure 4.4 and Figure 4.5 show the results in graphical form. The total pressure P is
seen to remain very close to the airbox pressure P
o
over the whole length of the duct,
despite the considerable fluctuation of static pressure p. Bernoullis equation has
therefore been verified for the streamline along the centre of the duct, along which
significant velocity changes take place. The distribution of velocity, measured by the
Pitotstatic probe, is compared in Figure 4.5 with the velocity distribution inferred
from the continuity equation. In the converging section the results are almost
identical, but in the diverging section downstream of the throat a steadily increasing
discrepancy arises. The airstream is apparently decelerating less quickly than the
geometrical shape of the passage would indicate.
Figure 4.4
It will be seen in a later experiment that a boundary layer forms adjacent to any fixed
surface along which air flows, and in this layer the velocity reduces from the free
stream value down to zero at the surface. The thickness of the layer increases in the
direction of flow, and it is found experimentally that the growth in thickness is more
26
rapid in regions of rising pressure (i.e. where the main stream is decelerating) than in
regions where the pressure is constant. The converse is true; where the pressure falls
in the direction of flow the growth of boundary layer thickness is retarded.
Figure 4.5
The results presented in Figure 4.5 are consistent with this concept. In the converging
section and the throat, the measured pressures agree closely with those calculated
from the variation in duct width so the boundary layer has scarcely any effect. In the
diverging section, however, thickening of the boundary layer would give the
27
appearance of the crosssection of the duct enlarging less rapidly than it actually does,
the retarded air in the thickening boundary layer presenting a partial blockage to the
flow.
We may therefore conclude that the experiment as a whole has demonstrated that
Bernoullis equation is sensibly valid along the central streamline of the convergent
divergent duct, since the total pressure has been shown to be virtually constant along
its length. The calculated pressure distribution, which depends on the concept of
continuity as well as constant total pressure, shows a significant discrepancy from the
measured results in the divergent portion, and this may be explained by the growth of
boundary layers on the walls of this portion.
Questions for Further Discussion
1. What boundary layer thickness do your results lead you to expect. Can you
infer this from the graph of Figure 4.5?
2. What is the Mach number at the throat of the duct? For approximate
calculation, you may assume that the static pressure and temperature there are
approximately the same as in the airbox. The air velocity at the throat may be
found from the Pitotstatic reading, and the acoustic velocity may be estimated
from the equation
a R = T
(47)
in which,
is the ratio of specific heats = 1.4 for air
R is the gas constant = 287.2 J/kg K
T is the absolute temperature in K
For the results given here:
= =
=
p
RT
kg m
1028 10
287 2 295
1213
5
3
.
.
.
28
P p N m
t t
= + = 790 980 1770
2
1
2
2
1213 1770 = . u
t
u m
t
s = 54 0 .
Also
a
t
= 14 287 2 295 . .
a m
t
s = 344
Ma
u
a
t
t
t
= = =
54 0
344
0157
.
.
3. What difference to the results would you expect if the flow direction were
reversed? You may check your prediction by reversing the liners.
4. What suggestions have you for improving the experiment?
5. How might you check whether there is in fact a boundary layer of significant
thickness at exit from the duct? A possible project would be to devise and
construct a suitable simple traversing gear from a Pitot tube which would
measure the velocity distribution. Would it be necessary to traverse along
more than one axis?
29
30
5. DRAG MEASUREMENT ON CYLINDRICAL BODIES
AF12 Drag Force Apparatus
31
Introduction
The resistance of a body as it moves through a fluid is of great technical importance
in hydrodynamics and aerodynamics. In this experiment we place a circular cylinder
in an airstream and measure its resistance, or drag, by three methods. We start by
introducing the ideas which underline these methods.
Consider the cylindrical body shown in crosssection in Figure 5.1. The reader may
be unfamiliar with the idea of a noncircular cylinder. In the present context the
word cylinder is used to describe a body which is generated by a straight line
moving round a plane closed curve, its direction being always normal to the plane of
the curve. For example, a pencil of hexagonal crosssection is by this definition a
cylinder.
Figure 5.1
The curve shown in Figure 5.1 represents a section of an oval cylinder. An essential
property of a cylinder is that its geometry is twodimensional; each crosssection is
exactly the same as every other crosssection, so that its shape may be described
without reference to the dimension along the cylinder axis. We shall use the term
circular cylinder to denote the particular and important case of the cylinder of
circular cross section. Motion of the cylinder through stationary fluid produces
actions on its surface which give rise to a resultant force. It is usually convenient to
analyse these actions from the point of view of an observer moving with the cylinder,
to whom the fluid appears to be approaching as a uniform stream. At any chosen point
32
A of the surface of the cylinder, the effect of the fluid may conveniently be resolved
into two components, pressure p normal to the surface and shear stress along the
surface. It is convenient to refer absolute pressure p
a
to the datum of static pressure p
o
in the oncoming stream; p is then a gauge pressure, that is:
p p
a o
p =
Let U denote the uniform speed of the motion and the density of the fluid, then the
dynamic pressure in the undisturbed stream, U
2
, is
1
2
2
U P
o o
= p
(51)
where P
o
is the total pressure in the oncoming stream. This pressure is a useful
quantity by which the gauge pressure p and shear stress may be non
dimensionalised, and the following dimensionless terms are defined:
Pressure coefficient, c
p
U
p
=
1
2
2
(52)
Skin friction coefficient, c
U
f
=
1
2
2
(53)
The combined effect of pressure and shear stress (sometimes called skin friction)
gives rise to resultant force on the cylinder. This resultant may conveniently be
resolved into the following components acting at any chosen origin C of the section as
shown in Figure 5.1:
1. A component in the direction of U, called the drag force, of intensity D per
unit length of cylinder.
2. A component normal to the direction of U, called the lift force, of intensity L
per unit length of cylinder.
3. A moment about the origin C, called the pitching moment, of intensity M per
unit length of cylinder.
33
These components may be expressed in dimensionless terms by definition of drag,
lift, and pitching moment coefficients as follows:
Drag coefficient, C
D
U d
D
=
1
2
2
(54)
Lift coefficient, C
L
U d
L
=
1
2
2
(55)
Pitching moment coefficient, C
M
U d
M
=
1
2
2 2
(56)
in which d denotes a suitable dimension which characterises the size of the cylinder.
In Figure 5.1 this is shown as the width measured across the cylinder, normal to U,
which is the usual convention. (An important exception is the aerofoil, where the
length in the direction of flow or chord of the section is used instead). The
coefficients C
D
, C
L
and C
M
are of prime importance since they are invariably used for
correlating aerodynamic force measurements.
We may see how pressure and skin friction coefficients are related to lift and drag
coefficients. Consider an element of length s of the surface, at a point where the
normal is inclined at angle to the direction of U, as shown in Figure 5.1. The
element of drag D per unit cylinder length due to p and is
( ) D p = + cos sin s
and integrating this round the whole perimeter yields
( ) D p = +
cos sin ds
This may now be cast in dimensionless form:
D
U d d
p
U U
ds
1
2
2
1
2
2
1
2
2
1
= +
cos sin
34
or
( )
C
d
c c
D p f
= +
1
cos sin ds
(57)
Similarly
( )
C
d
c c
L p f
= +
1
sin cos ds
(58)
These results show that the drag of a cylinder may be found by measuring p and
over the surface and calculating the drag coefficient by Equation (57). Now it is easy
to measure the distribution of p over a cylinder merely by drilling fine holes into its
surface, but measurement of is a much more difficult task. For the case of the
circular cylinder, however, the contribution to drag from shear stress (the skin
friction drag) is found to be very much smaller than from pressure (the pressure
drag) and may be neglected. Making this assumption and writing
s R
d
= =
2
(59)
for the circular cylinder of Figure 5.2 simplifies Equation (57) to:
C
d
c
d
d
D p
=
1
2
cos
=
C c
D p
1
2
0
2
cos
d
d
From consideration of symmetry, we may therefore write
C c
D p
=
cos
0
2
(510)
C C
L M
= = 0
(511)
35
for the circular cylinder. Equation (510) allows us to calculate C
D
from the measured
pressure distribution over the cylinder surface.
Figure 5.2 The Circular Cylinder
At the point marked S in Figure 5.2, the oncoming airstream is brought to rest. S is
called the stagnation point, and the streamline arriving at S is the dividing
streamline. Moving around the cylinder from S, we expect the velocity over the
surface to increase from zero at S, and so according to Bernoullis equation, we might
expect the pressure and therefore the pressure coefficient to fall. By an analysis which
is beyond our scope, the velocity u over the surface is given in terms of by the
simple equation
u
U
Sin = 2
(512)
provided that the fluid is incompressible and nonviscous.
36
Writing p
a
as the absolute static pressure at A, Bernoullis equation is:
P p U p
o o a
= + = +
1
2
2
1
2
2
u
(513)
The gauge pressure p at A is thus
p p p U
a o
= =
1
2
2
1
2
2
u
From Equation (512)
( )
p U Si =
1
2
2 2
1 4 n
so the pressure coefficient, c
p
, is
c
p
U
Sin
p
= =
1
2
2
2
1 4
(514)
This is the theoretical result for an incompressible, inviscid fluid, and forms the basis
of comparison with experimental results.
Figure 5.3 Application of the Momentum Equation to Flow along a Duct past a
Cylindrical Body
There is an entirely different way of finding the drag on a cylinder which depends on
the application of the momentum equation to the airflow. This equation is derived in
textbooks on the subject, but again for completeness a brief exposition is given here.
37
Consider the flow of a fluid along a duct of width 2h past a cylindrical body which
spans the duct, so that the motion is twodimensional as indicated in Figure 5.3. The
velocity is U and the pressure is p
o
in the oncoming flow. Downstream of the cylinder
the velocity is no longer uniform; let the velocity be u at distance y from the duct
centreline. The pressure across the downstream section is assumed to be uniform and
has the value p
e
. It is convenient to refer to the space bounded by the upstream
section, downstream section and duct walls as the control volume and the surface
formed by these boundaries as the control surface.
The forces in the xdirection acting on the fluid in the control volume are, per unit
length of cylinder:
At the upstream section 2h p
o
At the downstream section 2h p
e
At the cylinder D
Note the minus sign for the force exerted by the cylinder on the fluid, which is equal
and opposite to the force exerted by the fluid on the cylinder. Forces due to shear
stress on the wall of the duct and due to the fluid weight are neglected. The
momentum flux per unit width over the downstream section is:
u dy
h
h
2
and over the upstream section is
U dy
h
h
2
Equating the net force in the xdirection to the momentum flux out of the control
volume
2 2
2 2
h p h p D u dy U dy
o e
h
h
h
h
=
(515)
38
Rearranging and making nondimensional gives the result:
C
D
U d
h
d
p p
U d
u
U
dy
D
o e
h
h
= =
+
1
2
2
1
2
2
2
2
2 2
1
(516)
The integral may also be made nondimensional by the substitution
y h =
(517)
so that
1 1
2
2
2
2
1
1
u
U
dy h
u
U
d
h
h
and the final result is
C
h
d
p p
U
h
d
u
U
d
D
o e
=
+
2 2
1
1
2
2
2
2
1
1
(518)
Equation (518) provides a means to calculate C
D
from the pressure drop along the
duct and the velocity distribution in the wake. Note that the derivation does not
restrict the result to pressure drag only; the contributions of both pressure and skin
friction forces are contained in the force D which comes into the momentum equation.
The skin friction drag on the walls also contributes to the momentum change and is
also included in D. It is also worth mentioning that Equation (518) applies only to the
case of flow along a duct where the flow is confined between parallel walls.
The foregoing analysis shows how drag force may be found from the pressure
distribution over the surface of the cylinder and by measurements in the wake. The
results obtained from both of these methods may be compared with the drag measured
by direct weighing, and this is described in the next section.
39
Description of Apparatus and Test Procedure
Figure 5.4 shows the three main experiment arrangements for the AF12. Figure 5.4a
shows the main pressure tapping points and their notation. These arrangements allow
study of the drag of various bodies, the pressure distribution around a cylinder and the
wake of different bodies. The balance arm is attached to the side of the main unit and
holds the models in place inside the main unit. It is used with adjustable weights in
the drag force experiments and as a fixed support in the wake traverse experiments. It
is not used in the pressure distribution experiment. For the pressure distribution
experiment, a protractor and cylinder model is fitted in the main body. The cylinder is
connected to the separate AF10a manometer to display the pressure distribution as the
cylinder is rotated. For the wake traverse experiments, a pitot tube assembly is
attached to the main unit.
Test model
AF12 Main unit
Pitot tube assembly
Quick release coupling
Clear plastic tube
Stainless steel tube
To manometer
AF12
AF10 Contraction
section
(Outlet Duct)
Thumb Nut
Balance Arm Locking Screw
Cylinder
Quick release coupling
Protractor (located at back
of main unit)
To manometer
AF12 Main
unit
AF12
Balance
Arm
AF10 Contraction
section
(Outlet Duct)
AF12
Test model
AF12 Main unit
Quick release coupling
AF10 Contraction
section
(Outlet Duct)
10 g Weight
100 g Weight
Retaining Screw
Test Model Clamp Screws (x2)
Test Model Thumbnuts (x2)
Drag Force Pressure Distribution
Wake Traverse
Figure 5.4 Diagram of Apparatus
The apparatus is supplied with the cylinder and protractor model for pressure
distribution experiment and three other models for the wake traverse and drag force
experiments. The models include: a circular cylinder, a flat plate (inverted prism) and
a symmetrical aerofoil section.
40
P
o
p
o
p
P
e
p = 0, atmospheric datum
e
Pitot tube for
traverse in exit plane
AIR FLOW
BENCH AF10
Figure 5.4a Pressure Tappings
The Drag Force Experiments  Procedure
Fit the circular cylinder model in position and adjust the weights to achieve
equilibrium. Record the value of the weights. Every 1 mm on the front scale (100 g
weight) represents 1 g and every 1 mm on the back scale (10 g weight) represents 0.1
g (use for fine adjustment). Switch on the fan and adjust the wind speed to a low level
and readjust the weights to achieve equilibrium. Record the value of the weights and
the total pressure (Po) and the static pressure (po) at the inlet. Increase the wind speed
in increments up to its maximum level, each time readjust the weights to achieve
equilibrium and record the pressures and the value of the weights. Subtract the initial
value of the weights (with zero wind speed) from all the readings to give an actual
value of equilibrium at each wind speed. Repeat the procedure with the other two
models.
41
The Pressure Distribution Experiments  Procedure
Fit the circular cylinder and protractor model into the AF12. Connect the model to the
manometer. Set the wind speed to maximum and the protractor to zero angle. Record
the surface pressure (p) and the static pressure p
o
. Rotate the protractor in 5 intervals
for readings over the front half (0 to 90) and 10 intervals for readings over the rear
half (90 to 180). Monitor the total pressure (P
o
) and static pressure (p
o
) at the inlet to
ensure that the wind speed is kept constant throughout the experiment.
The Wake Traverse Experiments  Procedure
Fit the circular cylinder model into the AF12. Fit the pitot traverse assembly. Set up a
constant wind speed (a pressure difference of approximately 450 N/m
2
is
recommended). Traverse the Pitot tube across the section, in increments of 2 mm near
the centre, increasing to 5 mm or 10 mm in regions where the total pressure is seen to
be substantially constant. Record the values of P
e
and p
e
The readings in the wake are
extremely unsteady because of the high level of turbulence. Some users may wish to
damp the oscillations by placing a tubing clip on the flexible connection to the
manometer. If this method is used, take care to avoid excessive damping that can
cause error. It is better to use too little than too much. Repeat the experiment with the
other two models.
42
Results
Table 5.1 gives the drag force measured in units of gramforce by direct weighing at
various air speeds. The drag force is written as the product of the drag D per unit
length and the length l of the cylinder.
Drag Force
Dl (gmf)
P
o
(mm H
2
O)
p
o
(mm H
2
O)
P
o
p
o
= U
2
(N/m
2
)
39.2
38
34
27.6
19.3
13.5
4.9
190
188
181
170
155
145
128
128
128
126
125
124
123
121
608.22
588.6
539.55
441.45
304.11
215.82
68.67
Table 5.1 Drag Force on the cylinder measured by Direct Weighing
Air temperature 24C = 297 K
Barometric pressure 1040 mb = 1.04 10
5
N/m
2
Air density =
p
RT
= 219 . 1
297 2 . 287
10 04 . 1
5
=
kg/m
3
Diameter of cylinder d = 12.5 mm = 0.0125 m
Length of cylinder l = 48 mm = 0.048 m
Half width of working section h = 50 mm = 0.050 m
h
d
= 4
In Figure 5.5 the force is plotted against dynamic pressure and a good linear
relationship is established with a slope
600
38
2
2
1
=
U
Dl
3
10 81 . 9
600
38
m
2
= m
4
10 213 . 6
2
(Note that 1 gmf = 981 dyn = 9.81 10
3
N)
43
C
D
may now be found by substitution in Equation (54)
C
D
U d
D
U dl
D
= =
1
2
2
1
2
2
1
03 . 1
048 . 0 0125 . 0
10 213 . 6
4
=
=
Value of drag coefficient by direct weighing C
D
= 1.03
0
5
10
15
20
25
30
35
40
0 100 200 300 400 500 600 700
Dynamic Pressure 1/2U
2
(N/m
2
)
D
r
a
g
F
o
r
c
e
D
1
(
g
m
f
)
Figure 5.5 Measured Drag Force on a Circular Cylinder
The measured pressure distribution round the cylinder is given in Table 5.2, and
graphs of c
p
and c
p
cos as functions of angle from the front are presented in
Figures 5.6 and 5.7. The velocity pressure at inlet during this test was:
P
o
p
o
= 588.6 (x2), 598.41, 598.4 (x7) : Mean 596.448 N/m
2
U
2
= 596.448 N/m
2
44
The pressure is seen to be relatively symmetrical about the line = 0. The pressure
coefficient falls over the front portion to a minimum at = 70, and thereafter rises a
little. Over the rear half of the cylinder the pressure is reasonably uniform. The
distribution of c
p
for a cylinder in inviscid, incompressible fluid is shown for
comparison, calculated from Equation (514).
(degree)
p p
o
(N/m
2
)
C
p
=
p p
o
U
2
1
2
C
p
Cos
(degree)
p p
o
(N/m
2
)
C
p
=
p p
o
U
2
1
2
C
p
Cos
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
100
110
120
130
140
150
160
170
180
559.17
549.36
529.74
461.07
372.78
245.25
117.72
19.62
186.39
353.16
480.69
598.41
696.51
755.37
765.18
716.13
657.27
627.84
608.22
618.03
608.22
608.22
618.03
618.03
618.03
618.03
618.03
608.22
0.94
0.92
0.89
0.77
0.63
0.41
0.20
0.03
0.31
0.59
0.81
1.00
1.17
1.27
1.28
1.20
1.10
1.05
1.02
1.04
1.02
1.02
1.04
1.04
1.04
1.04
1.04
1.02
0.94
0.92
0.87
0.75
0.59
0.37
0.17
0.03
0.24
0.42
0.52
0.58
0.58
0.54
0.44
0.31
0.19
0.09
0.00
0.18
0.35
0.51
0.67
0.79
0.90
0.97
1.02
1.02
0
5
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
100
110
120
130
140
150
160
170
180
549.36
549.36
519.93
470.88
392.4
284.49
147.15
19.62
137.34
313.92
451.26
559.17
676.89
765.18
794.61
774.99
735.75
667.08
637.65
637.65
627.84
627.84
627.84
627.84
637.65
627.84
608.22
608.22
0.92
0.92
0.87
0.79
0.66
0.48
0.25
0.03
0.23
0.53
0.76
0.94
1.13
1.28
1.33
1.30
1.23
1.12
1.07
1.07
1.05
1.05
1.05
1.05
1.07
1.05
1.02
1.02
0.92
0.92
0.86
0.76
0.62
0.43
0.21
0.03
0.18
0.37
0.49
0.54
0.57
0.54
0.46
0.34
0.21
0.10
0.00
0.19
0.36
0.53
0.68
0.81
0.93
0.99
1.00
1.02
Table 5.2 Pressure Distribution around a Cylinder
45
3.00
2.00
1.00
0.00
1.00
0 20 40 60 80 100 120 140 160 180
(Degrees)
C
p
0 to 180
0 to 180
Ideal Fluid Theory
Calculated from equation
514
Figure 5.6 Pressure Distribution around a Circular Cylinder
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1.2
0 20 40 60 80 100 120 140 160 180
(Degrees)
C
p
c
o
s
0 to180
0 to 180
Figure 5.7 Distribution of c
p
cos around a Cylinder
46
C
D
may be obtained from Figure 5.7 from the area beneath the curve. The area A
beneath the mean curve is
A c
p
o
= cos
d
which, from Equation (510), we recognise as the drag coefficient C
D
. This area can
be evaluated in various ways; Appendix A describes how it is done by use of
Simpsons rule. The result is
A = 1.07
The value of drag coefficient obtained by plotting C
D
= 1.07
y
(mm)
d
y
e e
p P
(mm
H
2
O)
U
2
1
2
1
1
2
U
U
U
U
y
(mm)
d
y
e e
p P
(mm
H
2
O)
U
2
1
2
1
1
2
U
U
U
U
0
1
3
5
7
9
11
13
15
17
19
22
25
35
40
45
0
0.08
0.24
0.4
0.56
0.72
0.88
1.04
1.2
1.36
1.52
1.76
2
2.8
3.2
3.6
19
20
21
22
27
33
37
41
47
50
53
55
57
58
58
48
5.53
5.67
5.81
5.95
6.59
7.29
7.72
8.12
8.70
8.97
9.24
9.41
9.58
9.66
9.66
8.79
0.24
0.24
0.24
0.24
0.21
0.18
0.16
0.13
0.08
0.06
0.03
0.02
0.00
0.01
0.01
0.07
0
1
3
5
7
9
11
13
15
17
19
22
25
35
40
45
49
0
0.08
0.24
0.4
0.56
0.72
0.88
1.04
1.2
1.36
1.52
1.76
2
2.8
3.2
3.6
3.92
19
20
21
23
25
31
35
40
47
50
53
55
55
56
57
57
43
5.53
5.67
5.81
6.08
6.34
7.06
7.50
8.02
8.70
8.97
9.24
9.41
9.41
9.49
9.58
9.58
8.32
0.24
0.24
0.24
0.23
0.22
0.19
0.17
0.14
0.08
0.06
0.03
0.02
0.02
0.01
0.00
0.00
0.11
Table 5.3 Results of Velocity Traverse in the Wake
47
In table 5.3 the value of U
2
is calculated from the experiment results:
) ( 2
2
e e
p P g
U
=
To obtain a value of drag coefficient from the Wake Traverse method, the effects of
the duct must be taken into consideration. Because of this the value of free stream
velocity (U
1
) is calculated using the free stream flow of the duct (the area outside the
wake of the cylinder. Looking at the results of Pe & pe in table 5.3, taken during the
experiment, the free stream flow region can be easily identified as the stable region.
In this case the values obtained between 25 mm & 40 mm.
The average values of Pe & pe are:
P
e
= 16.7
P
e
= 11
Using these values, U
1
is calculated using;
) ( 2
1
e e
p P g
U
=
22 . 1
) 11 7 . 16 ( 81 . 9 2
1
= U
U = 9.57 m/s
0.05
0
0.05
0.1
0.15
0.2
0.25
0.3
4 3 2 1 0 1 2 3 4
y/d
(
U
2
/
U
1
)
*
(
1

(
U
2
/
U
1
)
)
0 to 45 mm
0 to 45 mm
Figure 5.8 Velocity Traverse in the Wake of a Cylinder
48
The drag coefficient may be obtained by use of the curve of
1
2
1
2
1
U
U
U
U
in Figure
5.8. The area beneath the curve is found using the trapezoidal rule (see Appendix A)
to be:
525 . 0 1
/
/
1
2
1
2
=
d
y
d
U
U
U
U
d y
d y
Using this Equation gives
d
y
d
U
U
U
U
C
d y
d y
D
/
/
1
2
1
2
1 2
C
D
= 2 x 0.525 = 1.05
Value of drag coefficient from wake traverse C
D
= 1.05
Discussion
The values obtained for drag coefficient of the circular cylinder are as follows
By direct weighing C
D
= 1.03
By pressure plotting C
D
= 1.07
By wake traverse C
D
= 1.05
Taking the first of these as being the most reliable, we see that pressure plotting yields
a result within the probable limits of accuracy.
There have been very many measurements of drag of circular cylinders, and the
variation of C
D
with Reynolds number Re, where
Re =
Ud
is well established. In this experiment, the absolute viscosity is, at 24C,
49
= 2.40 10
5
kg/m s
so the kinematic viscosity is
=
=
216 . 1
10 40 . 2
5
= 1.97 10
5
m
2
/s
The velocity U at the typical value of velocity pressure
U
2
= 500 N/m
2
is
6 . 26
22 . 1
500 2
2
1
=
= U m/s
So a typical value of Re for these tests is:
5
10 97 . 1
0125 . 0 6 . 28
Re
= =
Ud
Re = 2.4 10
4
It is well established that in the range of Re from 10
4
to 10
5
, C
D
is almost constant, the
value usually quoted for a cylinder being
C
D
= 1.20
This is for the case of a long cylinder, for which
l
d
is so great that the three
dimensional flows at the ends have no significant effect on the result. Our
experiments are made for a cylinder with
l
d
= 3.9 for which end effects must be
significant. Moreover, the cylinder presents an appreciable blockage to the cross
section of the flow  its projected area is about 1/8th of the cross sectional area of the
working section, so the results cannot be compared directly with those obtained from
cylinders in unconfined flows.
50
Questions for Further Discussion
1. A yawmeter is an instrument for finding the direction of a fluid stream. One
type of yawmeter consists of a circular cylinder with two surface holes at
different positions in the same diametral plane. It is placed in the stream and
rotated slowly about its axis until a balance is obtained between the observed
pressures at the holes. Suggest a suitable angular spacing between the holes to
give good sensitivity.
2. Measure the drag coefficients of a flat plate and an aerofoil section by direct
weighing, and comment on the results obtained. Could pressure plotting be
used to establish the drag coefficients of these sections?
(C
D
= 2.0 for plate, C
D
= 0.04 for aerofoil)
3. The circular cylinder presents a substantial blockage to the flow along the
working section. Suppose instead of using the approach velocity U, the
velocity U
1
past the cylinder was used as the velocity on which the reference
value of velocity head is based. Show that
U
l
=
8
U
7
for the results presented here, and find the corresponding value C
D1
from
C
D
U d
D1
1
2 1
2
=
(0.76)
51
52
6. THE ROUND TURBULENT JET
AF13 Round Turbulent Jet Apparatus
53
Introduction
The behaviour of a jet as it mixes into the fluid which surrounds it has importance in
many engineering applications. The exhaust from a gas turbine is an obvious
example. In this experiment we establish the shape of an air jet as it mixes in a
turbulent manner with the surrounding air. It is convenient to refer to such a jet as a
submerged jet to distinguish it from the case of the free jet where no mixing with
the surrounding medium takes place, as is the case when a smooth water jet passes
through the atmosphere.
If the Reynolds number of a submerged jet (based on the initial velocity and diameter
of the jet) is sufficiently small, the jet remains laminar for some length  perhaps 100
diameters or more. In this case the mixing with the surrounding fluid is very slight,
and the jet retains its identity. Laminar jets are important in certain fluidic
applications, where a typical diameter may be 1 mm, but the vast majority of
engineering applications occur in the range of Re where turbulent jets are produced.
Figure 6.1 Schematic Representation of a Round Turbulent Jet
54
The essential features of a round turbulent jet are illustrated in Figure 6.1. The jet
starts where fluid emerges uniformly at speed U from the end of a thinwalled tube, of
crosssectional radius R, placed in the body of a large volume of surrounding fluid.
The sharp velocity discontinuity at the edge of the tube gives rise to an annular shear
layer which almost immediately becomes turbulent.
The width of the layer increases in the downstream direction as shown in the diagram.
For a short distance from the end of the tube the layer does not extend right across the
jet, so that at section 1 there is a core of fluid moving with the undisturbed velocity U,
the velocity in the shear layer rising from zero at the outside to U at the inside.
Further downstream the shear layer extends right across the jet and the velocity u
o
on
the jet axis starts to fall as the mixing continues until ultimately the motion is
completely dissipated.
There is entrainment from the fluid surrounding the jet by the turbulent mixing
process so that the mass flux in the jet increases in the downstream direction. The
static pressure is assumed to be constant throughout, so there is no force in the
direction of the jet. The momentum of the jet is therefore conserved. The kinetic
energy of the jet decreases in the downstream direction due to the turbulent
dissipation. It should be emphasised that the velocity profiles indicated in Figure 6.1
are mean velocity distributions, and that the very severe turbulence in the jet will
cause instantaneous velocity profiles to vary considerably from these mean ones.
Velocity Distribution and Momentum Flux
Consider the jet of Figure 6.1. If we assume that the flow pattern is independent of
Reynolds number, then we might expect the velocity on the jet axis to depend on
position in the dimensionless form
u
U
f
x
R
o
=

\

.

(61)
In the core of the jet, we have already observed that
u
U
o
= 1
55
Far downstream, when the length of the core ceases to have influence, there is some
theoretical justification (supported by experiment) for expecting the centreline
velocity to decay inversely as x, i.e.
u
U
c
x
o
=
(62)
where c is a constant.
The velocity u at any position (r, x) in the jet may also be written in the dimensionless
form
u
u
g
x
R
r
x
o
=

\

.

,
(63)
Consider now the velocity distribution over a section far downstream, i.e. where
x
R
is
large. We might reasonably expect that the velocity distribution across the section
would not depend appreciably on the precise detail of the flow near the tube exit, so
we might ignore the dependence upon
x
R
and simply write
u
u
g
r
x
o
=

\

.

(64)
far downstream. Velocity profiles of this type, in which the velocity ratio depends on
a parameter, are frequently called similar, in the sense that a single expression is
used to characterise the velocity distribution at any number of chosen sections. Using
certain assumptions about the nature of the turbulent processes, it is possible to show
that Equation (64) should take the form
u
u
r
x
o
=
+

\

.

1
1 0 25
2
2
.
(65)
where is a constant which is to be determined by experiment.
56
Values of u/u
o
computed from this expression are presented in Table 6.1. The value
r/x = 1.287 is included, as this makes u/u
o
= 0.5. When comparing with
experimental results it is useful to have this value, since the radius at which u/u
o
= 0.5
is easily identified on the velocity profile.
r/x u/u
o
0
0.2
0.4
0.6
0.8
1.0
1.287
1.4
1.6
1.8
2.0
2.25
2.5
3.0
4.0
1.000
0.980
0.925
0.842
0.743
0.640
0.500
0.450
0.372
0.305
0.250
0.195
0.152
0.095
0.040
Table 6.1 Calculated Velocity Profile of Round Jet
Figure 6.2 Annular Element of a Round Jet
57
Coming now to mass, momentum and energy flux, we see in Figure 6.2 an annular
element of the jet through which fluid of density is flowing with velocity u. The
area of the element is
A = 2r r
So the mass flux through it is & m
& m = 2ur r
The total mass flux through the section of the jet is & m
& m u =
2
0
r dr
r dr
(66)
The momentum flux J through the section is similarly found to be
J u
o
=
2
2
(67)
and the kinetic energy flux E to be
E u
o
=
2
1
2
3
r dr
R
(68)
It is convenient in many instances to relate these to the corresponding fluxes at the
tube exit, as follows:
& m U
o
=
2
J = U R
o
2 2
E = U R
o
3 2
1
2
with the results
58
&
&
m
m
u
U
r
R
d
r
R
o
=
2
0
(69)
J
J
u
U
r
R
d
r
R
o
=
2
0
2
(610)
E
E
u
U
r
R
d
r
R
o
=
2
3
0
(611)
Description of Apparatus and Procedure
The round jet is produced by discharging air from the airbox through a short tube as
indicated in Figure 6.3. The inlet of the tube is rounded to prevent separation so that
a substantially uniform velocity distribution is produced at the tube exit.
A traversing mechanism is supported on the tube so that a Pitot tube may be brought
to any desired position in the jet. Measurements are normally made in one plane, but
if it is necessary to check on the symmetry of the jet about the axis, the traversing
mechanism may be rotated as a whole to any position.
The Pitot tube is first brought into the plane of the exit of the jet tube and the scale
readings are noted for which the axial position x and the radial position r are zero.
The latter may be obtained by taking the average of the readings when the tube is set
in line with one side and then the other side of the tube. The pressure P
o
in the airbox
is then brought to a convenient value and traverses are made at various axial stations
along the length of the jet. The readings of total pressure P fluctuate violently
because of the turbulence and some damping is required; however, excessive damping
must not be used. It is recommended that graphs of total pressure P against radius r
be plotted as the experiment proceeds to ensure that the profile is wellestablished by
a sufficient number of readings in the critical regions.
59
P in air box
o
Traversing gear
mounted on tube
Scale for x
Scale for r
P
R
U
x
r
Figure 6.3 Arrangement of Jet Apparatus
Results and Calculations
Diameter D of jet tube 51.6 mm
Radius R 25.8 mm
Pressure P
o
in airbox 900 N/m
2
Air temperature 22
o
C = 295 K
Barometric pressure 1025 mb = 1.025 10
5
N/m
2
Air density
1025 10
287 2 295
5
.
.
= 1.210 kg/m
3
Coefficient of viscosity, u = 1.82 10
5
kg/ms
60
Coefficient of kinematic viscosity, =
u
=
150 10
5 2
. m s
Velocity U at tube exit, U
2
= 870 N/m
2
U =
2 870
1210
=
.
37.9 m/s
Reynolds number Re at tube exit Re =
UD
=
37 9 0 0516
150
10
5
. .
.
Re = 1.30 10
5
The velocity along the axis of the jet was first found by traversing axially, the results
being presented in Table 6.2 and Figure 6.4.
x
(mm)
P
(N/m
2
)
u
U
o
0
50
75
100
125
150
175
200
225
250
300
350
400
870
860
845
835
830
810
775
730
675
620
505
430
340
1.00
0.99
0.99
0.98
0.98
0.96
0.94
0.92
0.88
0.84
0.76
0.70
0.63
Table 6.2 Velocity Distribution along Jet Axis
61
100 200 300 400 500
1.0
0.8
0.6
0.4
0.2
0
5
10 15 x/R
u
/
U
o
x (mm)
Figure 6.4 Centreline Velocity along Jet
For the initial portion the centreline velocity u
o
is seen to be almost constant, and
further downstream it starts to fall more rapidly as the shear layer extends to the
centre. Extrapolating the falling curve backwards to the line u
o
/U = 1 shows the
length of the core to be
x
c
= 175 mm or x
c
/R = 6.8
The results of radial traverses made at various values of x are shown in Table 6.3 and
in Figures 6.5(a) to 6.5(d). It may be noted that for x = 300 mm a check was made to
find whether the velocity distribution was symmetrical about the axis, and this
established that there was no appreciable departure from roundness. The profile at
x = 75 mm shows a distinct region of constant velocity in the core, and at x = 150 mm
there is still some evidence of a flat top to the profile. Further downstream, however,
this has disappeared. In Figure 6.6 a dimensionless comparison of the profiles is
made by dividing the radius by the radius at which the velocity ratio is 0.5. If you
take a set of readings further downstream than x = 300 mm and plot the velocity
profile, it will be very similar to the x/R = 11.6 curve. This shows a similar profile.
The transition from the squaretopped profile at the tube exit to the similarity profile
is clearly demonstrated. The curve calculated from Equation (65) as shown in Table
6.1 is also plotted. There is good agreement with the similarity profile near the centre
of the jet, but Equation (65) over estimates u/u
o
at the outer edge.
62
r x = 75 mm x = 150 mm x = 300 mm
(mm)
P
(N/m
2
)
u/u
o
P
(N/m
2
)
u/u
o
P
(N/m
2
)
u/u
o
0
5
10
15
17.5
20
22.5
25
27.5
30
32.5
35
40
45
50
55
60
65
70
75
80
85
90
840
840
840
835
810
755
630
465
275
160
50
5
0
1.00
1.00
1.00
1.00
0.98
0.95
0.87
0.74
0.57
0.44
0.24
0.08
0
810
800
770
720
590
430
250
135
70
25
10
5
0
1.00
0.99
0.97
0.94
0.85
0.73
0.56
0.41
0.29
0.18
0.11
0.08
0
520
520
495
450
375
330
245
185
155
110
75
45
30
20
15
5
0
1.00
1.00
0.98
0.93
0.85
0.80
0.69
0.60
0.55
0.46
0.38
0.29
0.24
0.20
0.17
0.10
0
Table 6.3 Velocity Distribution at Various Sections of the Jet
63
Figure 6.5(a)(c) Velocity Profiles in Jet at Various Distances Downstream
1.0
0.8
0.6
0.4
0.2
0
u
/
u
o
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
r/r
0.5
x/R = 2.91
x/R = 5.81
x/R = 11.6
Equation (55)
Figure 6.6 Dimensionless Velocity Profiles in the Jet
64
0 2 4
0.6
0.4
0.2
0
(
u
/
U
)
r
/
R
2
x = 75 mm
x = 150 mm
x = 300 mm
r/R
Figure 6.7 Momentum Flux in the Jet
A check on momentum conservation may be made by application of Equation (610).
On Figure 6.7 the curves of
u
U
r
R

\

.

2
are drawn as functions of
r
R
for each of the sets
of radial traverses. The areas under these curves represent the integrals
u
U
r
R
d
r
R
0
2
and so are a measure of momentum flux. The measured areas lead to the results of
Table 6.4.
The values do not remain constant at 1.0 as expected, but rise significantly as the jet
develops. There can be no doubt that the momentum flux does not increase since
there is no force acting in the direction of the jet, so the apparent rise must be due to
65
experimental error. The most likely source is turbulence which could have the effect
of giving an excessive mean velocity pressure.
x
(mm)
x/R
2
u
U
r
R
d
r
R
2
75
150
300
2.91
5.81
11.6
1.01
1.10
1.15
Table 6.4 Momentum Flux in Jet
Conclusion
The diffusion of a turbulent air jet into the surrounding atmosphere has been
measured by velocity traverses along the centreline and along several radii. The first
part of the jet is found to have a central core of almost constant velocity which
extends for a length x
c
= 6.8R along the axis. Thereafter the centreline velocity
reduces and the velocity profile rapidly tends to similarity, that is to a profile which
may be characterised by the single parameter r/x. The momentum flux in the jet,
which must be constant in a constantpressure atmosphere, appears to rise by
approximately 15% along its length. The discrepancy is attributed to measurement
error due to turbulence.
66
Suggestions for Further Experiments
1. Obtain the angle at which the jet spreads by establishing the trajectory along
which u/u
o
= 0.5.
2. Compare the variation of centreline velocity with Equation (62).
3. Investigate the effect of initial turbulence in the jet by placing a wire gauze
over the exit of the tube and comparing the results with those obtained with a
plain exit.
67
68
7. BOUNDARY LAYERS
AF14 Boundary Layer Apparatus
69
Introduction
It is a fact wellestablished by experiment that when a fluid flows over a solid surface
there is no slip at the surface. The fluid in immediate contact with a surface moves
with it, and the relative velocity increases from zero at the surface to the velocity in
the free stream through a layer of fluid which is called the boundary layer.
Consider steady flow over a flat smooth plate as shown in Figure 7.1, where the
streaming velocity U is constant over the length of the plate. It is found that the
thickness of the boundary layer grows along the length of the plate as indicated in the
diagram. The motion in the boundary layer is laminar at the beginning, but if the plate
is sufficiently long, a transition to turbulence is observed. This transition is produced
by small disturbances which, beyond a certain distance, grow rapidly and merge to
produce the apparently random fluctuations of velocity which are characteristic of
turbulent motion. The parameter which characterises the position of the transition is
the Reynolds number Re
x
based on distance x from the leading edge:
Re
x
Ux
=
(71)
Figure 7.1 General Characteristics of the Boundary Layer over a Flat Plate
70
The nature of the process of transition renders it prone to factors such as turbulence in
the free stream and surface roughness of the boundary, so it is impossible to give a
single value of Re
x
at which transition will occur, but it is usually found in the range
1 10
5
to 5 10
5
.
Definitions of Thickness
A little consideration will show that the boundary layer thickness , shown in Figure
7.1 as the thickness where the velocity reaches the free stream value, is not an entirely
satisfactory concept. The velocity in the boundary layer increases towards U in an
asymptotic manner, so the distance y at which we might consider the velocity to have
reached U will depend on the accuracy of measurement.
Figure 7.2 Velocity Distribution and Displacement Thickness of Boundary Layer
A much more useful concept of thickness by which fluid outside the layer is displaced
away from the boundary by the existence of the layer, as indicated schematically in
Figure 7.2, by the approaching streamline. In Figure 7.2 the curve 0A shows the
distribution of velocity u within the layer as a function of distance y from the
boundary. If there were no boundary layer, the free stream velocity U would persist
right down to the boundary as shown by the line CA. The reduction in volume flow
rate (per unit width normal to the diagram) due to the reduction of velocity in the
layer is therefore
Q U u
h
= ( )
0
dy
(72)
71
which corresponds to the shaded area 0AC in the diagram, the dimension h being
chosen so that u = U for any value of y greater than h. If the volume flow rate is now
considered to be restored by displacement of the streamline at AA away from the
surface to a position BB through a distance *, the volume flow rate between AA
and BB is also Q, and this is seen to be
Q U = *
(73)
Equating the results of Equation (72) and (73) gives
( )
* =
1
0
U
U u d
h
y
or
* =

\

.

1
1
0
u
U
dy
h
Now h is any arbitrary value which satisfies the condition
u = U
or
1 0 =
u
U
for all values of y greater than h. The value of h may therefore be increased
indefinitely without affecting the value of the integral, so we allow h to increase
towards infinity:
h
and obtain the result
* =
{
\
]
)
]
1
0
u
U
dy
(74)
We shall see that in the practical measurement of * from a measured velocity
distribution the infinite upper limit presents no difficulty.
72
A further definition is required when momentum effects within the boundary layer are
considered. Consider a control volume of length x, height h (greater than the
boundary layer thickness ) and unit thickness normal to the plane of the diagrams
shown in Figure 7.3. The rate of mass inflow is at the lefthand end, and the rate of
mass outflow at the righthand end is
& m
&
&
m
dm
dx
+ x . Consideration of continuity then
shows the outflow through the upper surface to be
dm
dx
x
&
. The momentum equation
may now be derived as follows.
Figure 7.3 Mass and Momentum Flux in Boundary Layer
The net rate of efflux of xcomponent of momentum from the control volume is
the sum of
&
M
&
&
M
dM
dx
x + at the righthand end
&
M at the lefthand end
and U
dm
dx
x
&
at the upper surface
Note: Over the upper surface, the xcomponent of velocity is U, and the mass
outflow rate is
dm
dx
x
&
.
If the surface shear stress is
w
acting in the direction shown in the diagram, the
momentum equation is then
73
= +
w
x M
dM
dx
x M U
dm
dx
x
&
&
&
&
which simplifies to
w
U
dm
dx
dM
dx
=
&
&
or
( )
w
d
dx
Um M = &
&
(75)
Now
& m u
h
=
0
dy
dy
(76)
and
&
M u
h
=
2
0
(77)
so substituting these results into Equation (75) gives
( )
w
h
d
dx
Uu u dy =
2
0
or
w
h
U
d
dx
u
U
u
U
dy =

\

.

2
0
1
Since u = U for all values of y greater than h, the arbitrary upper limit may be
replaced by infinity, giving
w
U
d
dx
u
U
u
U
dy =

\

.

2
0
1
(78)
It is convenient to express
w
in the dimensionless form of a local skin friction
coefficient
74
c
U
f
w
=
1
2
2
(79)
and if this is done, Equation (78) becomes
c
d
dx
u
U
u
U
dy
f
=
{
\
]
)
]
]
]
]
]
]
]
]
]
]
]
]
2 1
0
(710)
The writing of this result is simplified if we now define
=
]
{
\
]
)
]
u
U
u
U
dy
0
1
(711)
where is known as the momentum thickness of the boundary layer, and Equation
(710) becomes
c
d
dx
f
= 2
(712)
The total skin friction force per unit width on a plate of length L is
D d
f w
L
=
0
x
(713)
Writing
w
in terms of c
f
from Equation (79)
D U c
f f
o
L
=
1
2
2
dx
and from Equation (712)
75
D U
d
dx
dx
f
L
=
1
1
2
2
0
2
When x = 0, = 0, and writing
L
for the momentum thickness at distance L from
the leading edge,
D U
f L
=
1
2
2
2
(714)
The skin friction force D
f
is now written in terms of a dimensionless overall skin
friction coefficient C
F
where
C
D
U L
F
f
=
1
2
2
and substituting D
f
from Equation (714) gives
C
L
F
L
=
2
(715)
This equation gives the overall skin friction coefficient on a flat plate very simply in
terms of the momentum thickness at the trailing edge and the length of the plate.
It is frequently useful to refer to the ratio of displacement thickness * to momentum
thickness , and this is called the shape factor H:
H =
*
(716)
The calculation of the velocity profiles and thickness of boundary layers is beyond the
scope of this manual, but for reference and for comparison with results of experiments
a few results are presented here. For a laminar boundary layer along a flat plate
with uniform free stream velocity, the velocity profile has been calculated and some
numerical results are presented in Table 7.1
76
y Re x
x
0 0.5 1.0 1.5 2.0 3.0 4.0 5.0 6.0
u/U 0 0.1659 0.3298 0.4868 0.6298 0.8461 0.9555 0.9916 0.9990
Table 7.1 Velocity distribution in laminar boundary layer along flat plate
Note the dimensionless parameter y used in the table, which generalises the
results to any value of distance x along the plate. For this laminar layer, the
displacement thickness * and the momentum thickness are given by
x
x
Re
*
.
Re
=
1721x
x
(717)
and
=
0 664 .
Re
x
x
(718)
from which it may be noted that the thickness along the plate grows in proportion to
x . The shape factor is
H = 2.59
(719)
For a turbulent boundary layer along a smooth flat plate there are no
corresponding calculated results. Frequently the velocity distribution is expressed in
the form
u
U
y n
=

\

.

1
(720)
where n is an index which varies from about 5 to 8 as the value of Re
x
increases in the
range of 10
5
to 10
9
, although there are many alternative expressions. The
displacement and momentum thickness are frequently quoted as
*
.
Re
.
=
0 046
0 2
x
x
(721)
77
=
0 036
0 2
.
Re
.
x
x
(722)
with the shape factor
H = 1.29
(723)
The Effect of Pressure Gradient
The preceding discussion has related to boundary layer development along a smooth
plate with uniform flow in the free stream  in conditions of zero pressure gradient
along the plate. If the stream is accelerating or decelerating, substantial changes take
place in the boundary layer development. For an accelerating free stream, the pressure
falls in the direction of flow, the pressure gradient being given by differentiating
Bernoullis equation in the free stream as
dp
dx
U
dU
dx
=
(724)
The boundary layer grows less rapidly than in zero pressure gradient and transition to
turbulence is inhibited. For a decelerating free stream, the reverse effects are
observed. The boundary layer grows more rapidly and the shape factor increases in
the downstream direction. The pressure rises in the direction of flow, and this
pressure rise tends to retard the fluid in the boundary layer more severely than that in
the main stream since it is moving slower. Energy diffuses from the free stream
through the outer part of the boundary layer down towards the surface to maintain the
forward movement against the rising pressure. However, if the pressure gradient is
sufficiently steep, this diffusion will be insufficient to sustain the forward movement,
and the flow along the surface will reverse, forcing the main stream to separate. It is
this separation, or stall as it is sometimes called, which leads to the main component
of drag on bluff bodies and to the collapse of the lift force on an aerofoil when the
angle of incidence is excessive.
78
Description of Apparatus
Figure 7.4 shows the arrangement of the test section attached to the outlet of the
contraction of the airflow bench. A flat plate is placed at mid height in the section,
with a sharpened edge facing the oncoming flow. One side of the plate is smooth and
the other is rough so that by turning the plate over, results may be obtained on both
types of surface.
Figure 7.4 Arrangement of Test Section
A fine Pitot tube may be traversed through the boundary layer at a section near the
downstream edge of the plate. This tube is a delicate instrument which must be
79
handled with extreme care to avoid damage. The end of the tube is flattened so that it
presents a narrow slit opening to the flow. The traversing mechanism is spring loaded
to prevent backlash and a micrometer reading is used to indicate the displacement of
the Pitot tube.
Liners may be placed on the walls of the working section so that either a generally
accelerating or generally decelerating free stream may be produced along the length
of the plate, depending on which way round they are fitted. With the liners removed,
uniform freestream flow conditions are obtained over the plate length.
To obtain a boundary layer velocity profile, the Pitot tube is set approximately 10 mm
from the surface and the desired wind speed is established by bringing the pressure P
o
in the airbox to the required value. Readings of total pressure P measured by the Pitot
tube are then recorded over a range of settings of the micrometer as the tube is
traversed towards the plate. At first the readings should be constant, indicating that
the traverse has started in the free stream; if this is not the case, go back and start with
an initial setting further from the plate. As the Pitot tube reading begins to fall, the
step length of the traverse should be reduced so that at least 10 readings are obtained
over the range of reducing readings. The reading does not fall to zero as the tube
touches the wall because of its finite thickness, so the traverse is stopped as soon as
contact is indicated either by the electrical circuit or by the readings becoming
constant as the micrometer is advanced towards the surface.
Readings obtained in turbulent boundary layers are subject to unsteadiness which
leads to difficulty in obtaining average readings on the manometer. Damping may be
provided by squeezing the connecting plastic tube, but care should be taken that the
restriction is not too severe, which can lead to false readings.
Results and Calculations
(a) Turbulent Boundary Layers on Smooth and Rough Surfaces
The plate was installed in the test section without the liners fitted, and measurements
were made in the boundary layer formed on the smooth surface and then on the rough
surface.
Air temperature 19C = 292 K
80
Barometric pressure 1010 mb = 1.010 10
5
N/m
2
Air density =
1010 10
287 2 292
5
.
.
= 1.204 kg/m
3
Coefficient of viscosity u = 1.80 10
5
kg/m s
Coefficient of kinematic viscosity =
u
= 1.49 10
5
m
2
/s
Length of plate from leading edge to traverse section, L = 0.265 m
Thickness of Pitot tube at tip, 2t = 0.40 mm
Displacement of tube centre from surface when in contact, t = 0.20 mm
Pressure in airbox: 640 N/m
2
Readings of Pitot pressure P are tabulated in Table 7.2 and Table 7.3. Values of y
shown in the tables are obtained from the micrometer reading at which the tube just
touched the surface, making allowance for the initial displacement t due to the
thickness of the Pitot tube. Values of u/U are found from
u
U
P
P
o
=
where P
o
is the Pitot tube reading in the free stream.
The free stream velocity U is obtained from:
1
2
2 2
550 U N = m
=
= U m
2 550
1204
30 2
.
. s
Re
. .
.
. = =
=
UL
30 2 0 265 10
149
537 10
5
5
81
Micrometer
Reading (mm)
y
(mm)
P
(N/m
2
)
u/U
21.0
20.0
19.0
18.0
17.0
16.5
16.0
15.8
15.6
15.4
15.2
15.14
6.06
5.06
4.06
3.06
2.06
1.56
1.06
0.86
0.66
0.46
0.26
0.20
550
555
550
530
495
460
415
380
360
320
250
180
1.00
1.00
1.00
0.98
0.95
0.91
0.87
0.83
0.81
0.76
0.67
0.57
Table 7.2 Velocity Distribution in Boundary Layer
on Smooth Flat Plate, Re = 5.37 10
5
Micrometer
Reading (mm)
y
(mm)
P
(N/m
2
)
u/U
25.0
24.0
23.0
22.0
21.0
20.0
19.0
18.5
18.0
17.5
17.0
16.5
16.3
16.10
9.10
8.10
7.10
6.10
5.10
4.10
3.10
2.60
2.10
1.60
1.10
0.60
0.40
0.20
540
540
525
515
500
470
420
375
335
275
215
150
125
100
1.00
1.00
0.99
0.98
0.96
0.93
0.88
0.83
0.79
0.71
0.63
0.53
0.48
0.43
Table 7.3 Velocity Distribution in Boundary Layer
on Rough Flat Plate, Re = 5.37 10
5
82
Figure 7.5 shows the velocity distributions plotted for both smooth and flat plates.
Also shown are the curves of u/U(1 u/U), which are easily deduced from the curves
of u/U by reference to Table 7.4:
Figure 7.5 Velocity Distribution in Turbulent Boundary Layer on Smooth Plate
Re = 5.37 10
5
83
u/U 0.5 0.6 0.7 0.8 0.9 1.0
u
U
1
u
U

\

.

0.25 0.24 0.21 0.16 0.09 0
Table 7.4 Values of u/U (1  u/U)
The appropriate areas under the curves, evaluated by use of Simpsons rule, as
described in Appendix A, are:
Smooth plate: *
=
1

\

.

1
u
U
dy
= 0.53 mm
=
u
U
u
U
dy
1

\

.

1
= 0.40 mm
H
=
*
= 1.32
Rough plate: *
=
1

\

.

1
u
U
dy
= 1.50 mm
=
u
U
u
U
dy
1

\

.

1
= 0.98 mm
H
=
*
= 1.53
In Figure 7.5 a 1/7th power law is shown, corresponding to
u
U
y
=

\

.

3
1
7
and this is seen to compare reasonably well with the experimental results on the
smooth plate. For the rough plate, however, the velocity distribution does not fall
towards zero at y = 0. This is because the origin of the traverse has been taken from
the highest points of the rough surface. An examination of the structure of the
roughness would be required to establish the position of the mean surface from which
y should be measured.
84
The values of * and calculated for a turbulent boundary layer along a smooth
surface from Equations (721) and (722) are, with the length L = 0.265 m inserted.
( )
*
. .
.
. .
.
=
= =
0 046 0 265
537 10
085 10 087
5
0 2
3
m mm
and
( )
=
= =
0 036 0 265
537 10
0 68 10 0 68
5
0 2
3
. .
.
. .
.
m mm
The experimental results for the smooth plate are noticeably lower than these values,
indicating that over part of the length of the surface the boundary layer is laminar,
yielding an overall skin friction less than if the whole length of the layer was
turbulent.
For the rough plate the boundary layer thickness is more than twice that of the smooth
plate. Also it is more than the calculated values for a smooth surface, showing that the
roughness has produced a significant increase in skin friction drag to a value higher
than could be obtained on a smooth plate even if the whole length of the boundary
layer was turbulent.
(b) Effect of Pressure Gradient
The test was repeated with the liners fitted to give a generally decelerating flow over
the plate length. The Reynolds number based on the main stream velocity at the exit
and the length of the plate was
Re = 4.90 10
5
which is not sufficiently different from the previous value to affect the results. The
procedure is the same as before, so full details of the working are not presented.
Figure 7.6 shows the measured velocity profile in comparison with the previous
results, from which it is clear that the layer has grown appreciably thicker in the rising
pressure which is produced by the decelerating flow.
85
Figure 7.6 Effect of Pressure Gradient on Boundary Layer on Smooth Plate
Re = 4.90 10
5
86
The thickness and shape factor are:
* . =

\

.

1
= 1 0
u
U
dy 86
=

\

.

1
=
u
U
u
U
dy mm 1 0. 60
H = 1.43
Summary and Conclusions
Velocity traverses in turbulent boundary layers have been made with a specially
shaped fine Pitot tube traversed close to the surface. The resulting velocity profiles
have shown that roughness of the plate surface, and a rising pressure gradient, both
serve to increase the rate of growth of the boundary layer.
Suggestions for Further Experiments
1. Investigate the effect of a falling pressure gradient on the boundary layer by
repeating the tests with the liners reversed.
2. It has been suggested that the velocity distribution in a turbulent boundary
layer may be approximated by a logarithmic profile
u
U
A B
y
= +

\

.

log
where A and B are constants. Check whether your results fit this expression.
3. Observe the growth of boundary layer along a plate with a constant main
stream velocity by making traverses at successive stations along the plate
length. (The plate may be withdrawn from the working sections to a variety of
positions allowing traverses to be made at different stations along it). Plot the
growth of along the plate and consider what information this gives about the
skin friction coefficient c
f
.
87
88
4. Consider the possibility of making measurements of laminar boundary layers
in this apparatus. If the minimum main stream velocity for which reasonably
accurate velocity traverses may be obtained is 10 m/s, and the laminar layer
along the plate persists up to Re = 1 10
5
, show that the layer will extend
about 0.15 m from the leading edge. Show also that the displacement
thickness at this section would be about 0.8 mm according to Equation (717).
8. FLOW AROUND A BEND IN A DUCT
AF15 Flow Around a Bend Apparatus
89
Introduction
The engineer is frequently presented with problems of flow contained within tubes
and ducts. Such flows may be classified as internal flows to distinguish them from
flows over bodies such as aerofoils, called external flows. It is sometimes necessary
to shape a duct in such a way that particular requirements are met. For example, it
may be necessary to change the shape of crosssection from square to rectangular with
a small loss of total pressure, or it may be required to form a bend in such a way that
the distribution of velocity at the exit is as nearly uniform as it can be made.
Due to the presence of boundary layers along the duct walls, the fluid mechanics of
such flows are sometimes extremely complicated. Separation may be produced where
the pressure rises in the direction of flow, as illustrated in Figure 8.1(a).
Figure 8.1(a) Schematic Representation of Separating and Reattaching Flow
in a Duct
This shows a duct of increasing crosssectional area in which the flow decelerates
with an accompanying rise of pressure. Separation of flow from one wall is shown,
followed by a region of severe turbulence in which there is mixing between the main
flow and the region of recirculating flow (often called the separation bubble). The
turbulent mixing leads to loss of total pressure, the size of this loss depending on the
90
extent of the separation. It should be emphasised that the flow shown in the figure is
schematic only.
The separation line is rarely steady. The size of the separated zone often fluctuates
violently, and in some cases the separation is intermittent. Separation might occur
over more than one surface and would not normally take place uniformly over one
side as shown for illustrative purposes in the diagram. A further complication arises
from secondary flow which is again due to boundary layer effects.
Figure 8.1(b) Formation of Secondary Flow in a Bend of a Duct
Figure 8.1(b) shows one example of the formation of a secondary flow in a gently
curving duct of rectangular crosssection. The curvature of the flow is accompanied
by a pressure gradient which rises across the section from the inner to the outer wall.
The pressure gradient extends over the whole section, so that the boundary layers on
the upper and lower walls are subjected to the same pressure gradient as the main
flow. But because the streaming velocity in the boundary layer is less than in the
main part of the flow, the curvature of the streamlines in the boundary layer is more
severe, as indicated. This gives rise to a net inwarddirected flow adjacent to the
upper and lower walls, which sets up a secondary flow in the form of a double
rotation, superimposed on the main stream. The motion emerging from the curve in
the duct is therefore a pair of contrarotating spirals, the strength of which depends on
the amount of curvature and on the thickness of the boundary layer.
91
Simple Theory of Flow in a Bend
Figure 8.2 Assumed Velocity Distribution in a Bend
In this experiment we investigate the flow around a 90 bend in a duct of rectangular
section, using pressure tappings along the walls to establish the pressure distributions.
Figure 8.2 indicates flow approaching a bend with a uniform velocity U. Within the
bend we shall assume a free vortex distribution of velocity, given by
u
C
r
=
(81)
where u is the streaming velocity at radius r from the centre of curvature of the bend.
Separation and secondary flow will be neglected. The constant C may be found by
applying the equation of continuity as follows:
( ) Q Ub r r b u
r
dr
r
= =
2 1
1
2
(82)
where b is the width of the section of the duct. Substituting for u from Equation (81)
and performing the integration leads to the result
92
C U
r r
r r
=
2 1
2 1
ln ( / )
(83)
so the velocity distribution is, in dimensionless form,
u
U
r r
r r r
=
2 1
2 1
ln ( )
(84)
The corresponding pressure distribution may be found by assuming that Bernoullis
equation may be applied between the upstream section and a section within the bend
as follows:
p U p
o
+ = +
1
2
2
1
2
2
u
(85)
where p
o
is the static pressure upstream and p is the pressure at radius r in the bend. It
is convenient to express p in the form of a dimensionless pressure coefficient c
p
where
c
p p
U
p
o
=
1
2
2
(86)
From Equation (85) this may be written
c
u
U
p
= 1
2
2
(87)
which may be evaluated for any radius r by substituting the appropriate value of u/U
obtained from Equation (84). A comparison with measured values of c
p
may be
made as shown in Table 8.2.
Description of Apparatus
Figure 8.3 shows the dimensions of the bend and the positions of the pressure
tappings. There is a reference pressure tapping 0 on the side face near the entry, and
three sets of tappings; one set of 10 along the outer curved wall, one set of 10 along
93
the inner curved wall and a set of 9 along a radius of the bend. Air from the
contraction section is blown along the duct and is exhausted to atmosphere.
Figure 8.3 Dimensions of the Bend and the Positions of Pressure Tappings
Experimental Procedure and Results
The pressure tappings along the outer wall, the reference tapping 0 and the pressure
tapping in the airbox are all connected to the manometer. The air speed is adjusted to
a value slightly below the maximum, as indicated by the airbox pressure, and the
pressures are recorded. (The setting of air speed slightly below the maximum is to
ensure that the same setting may be repeated in later tests). The tappings on the inner
wall are then connected in place of the ones on the outer wall. The airbox pressure is
adjusted to the previous value and a further set of readings are recorded. Finally the
procedure is repeated with the third set of pressure tappings. In Table 8.1 the
pressures p are recorded relative to an atmospheric datum and the pressure
coefficients c
p
are calculated from Equation (86).
94
Airbox pressure = P = 630 N/m
2
Reference tapping pressure = p
o
= 80 N/m
2
Velocity pressure of uniform flow along duct P p
o
=
1
2
2
U = 550 N/m
2
Tapping Outer Wall Inner Wall Radial Section
No. p (N/m
2
) c
p
p (N/m
2
) c
p
p (N/m
2
) c
p
1
2
3
4
5
6
7
8
9
10
90
145
205
325
330
320
240
85
35
0
0.02
0.12
0.23
0.45
0.45
0.44
0.29
0.01
0.08
0.15
70
40
240
415
440
395
255
25
10
0
0.02
0.07
0.58
0.90
0.95
0.86
0.61
0.19
0.13
0.15
265
150
40
50
115
170
215
260
295
0.63
0.42
0.22
0.05
0.06
0.16
0.25
0.33
0.39
Table 8.1 Measured Pressures and Pressure Coefficients
From Figure 8.3 the inner and outer surfaces of the bend have radii
r
1
= 50 mm
r
2
= 100 mm
From Equation (84) the velocity distribution across the section according to the free
vortex assumption is therefore
u
U r
= =
50
2
721
ln( )
.
r
where r is expressed in mm. In Table 8.2 we calculate this ratio and the
corresponding value of c
p
from Equation (87) for a number of values of r.
95
r
(mm)
u
U
c
p
50
55
60
65
70
75
80
85
90
95
100
1.443
1.312
1.202
1.110
1.030
0.962
0.902
0.849
0.801
0.759
0.721
1.081
0.720
0.445
0.232
0.062
0.075
0.187
0.280
0.358
0.423
0.480
Table 8.2 Calculated Pressure Coefficients
Figure 8.4 Distribution of Pressure Coefficient c
p
over Walls
Figure 8.4 shows the distribution of measured pressure coefficient over the curved
walls and compares the measured and calculated values across the radial section. It
may be seen that the pressure across the inlet section is nearly uniform. As the flow
approaches the bend, the pressure on the inner wall falls rapidly and on the outer wall
96
rises rapidly to values which remain substantially constant round most of the curve.
This indicates that the curvature of the flow is also likely to be reasonably constant.
The distribution of c
p
over the radial section follows the calculated curve quite
closely, indicating that the assumption of a free vortex velocity distribution made in
Equation (81), together with the assumption that Bernoullis equation applies to the
flow, give a fairly accurate distribution of the pressure field. The measured pressure
distribution varies rather less steeply than calculated, indicating a vortex strength C
somewhat less than that given by Equation (83).
Downstream of the bend, the wall pressures readjust until at the duct exit the pressure
is constant across the section. It is, however, a little lower than the reference pressure
at the inlet, and this difference represents a pressure loss round the bend. It is
convenient to express this loss p in terms of the velocity pressure U
2
in the
uniform approaching flow by the expression
K
p
U
=
1
2
2
(88)
where K is the dimensionless loss coefficient. In this case we find, from the change
in c
p
from the inlet to the outlet sections, the value:
K = 0.15
(89)
Conclusion
The distribution of pressure over the curved walls of a 90 bend of rectangular section
has been established by pressure plotting. The pressure coefficient is negative and
almost constant round the inner wall, and positive and almost constant round the outer
wall. Across the 45 crosssection the pressure distribution may be predicted with
reasonable accuracy by assuming freevortex velocity distribution over the section.
The value of loss coefficient K is 0.15 for this bend.
97
Questions for Further Discussion
1. Do you consider that there is likely to be any separation of flow anywhere in
the bend, and can you suggest any way by which this might be investigated?
2. Do you consider that there might be any secondary flow in the stream,
downstream of the bend, and can you suggest how this might be investigated?
3. It has been proposed to measure flow rate Q in a duct system by placing
pressure tappings on the inner and outer walls at the 45 section of any
convenient 90 bend which occurs in the line of the duct, and measuring the
differential pressure p between the tappings. Using Equations (82) to (85)
show that Q is given by
4.
( )
Q
b r r
r r
r
r
p
=
1 2
2
2
1
2
2
1
2
ln
where b, r
1
and r
2
are defined in Figure 8.2
Noting that the measured pressures do not quite agree with the theoretical
values, this equation may be modified to
( )
Q C
b r r
r r
r
r
p
d
=
1 2
2
2
1
2
2
1
2
ln
in which C
d
is a discharge coefficient. Show that C
d
is given by
( )
[ ]
( )
[ ]
( ) ( )
C
r r r r
r r r r
U
p
d
=
2 1 2 1
2
1
2
2 1 2 1
1
2
2 1 1
ln
and hence find C
d
from the experimental results (C
d
= 1.06)
98
9. JET ATTACHMENT
AF16 Jet Attachment Apparatus
99
Introduction
In a previous experiment on the flow round a circular cylinder, the phenomenon of
separation of flow from a surface had been observed. The present experiment deals
with an effect which in some respects is the reverse, namely, the tendency for a plane
jet to attach itself to an adjacent wall and to flow along it. Figure 9.1(a) shows a
typical configuration. A plane jet emerges from the slit which discharges into the
atmosphere alongside a wall. It is found to deflect sideways and to attach to the wall.
If the wall curves as shown, it will follow the curve and so may suffer a considerable
change of direction. Exploitation of this phenomenon was proposed in the 1930s by
Henri Coanda, who made several inventions which used this form of jet deflection;
the phenomenon is therefore sometimes referred to as the Coanda Effect.
Figure 9.1 Mechanics of Jet Attachment to a Wall
A descriptive explanation of why a jet should exhibit this behaviour can be made
along the following lines. It is known that a jet emerging from a tube or slit will
entrain fluid from the surroundings as it mixes into the ambient fluid; in most cases of
engineering importance the mixing is turbulent and entrainment is much more intense
than if the process were laminar.
Consider now entrainment into the jet of Figure 9.1(a). In the first moment after
starting, the jet is straight and entrainment takes place equally on both sides. The
inflow from the surroundings into the jet is, however, restricted on one side due to the
100
presence of the wall and this restriction results in a reduction of pressure which bends
the jet towards the wall.
This in turn further restricts the supply of fluid to this side of the jet, causing further
reduction of pressure and further jet curvature until very soon the jet moves over to
the wall as shown in the diagram. In the final condition shown, there is a zone of
separated flow in which recirculation takes place. The rate at which fluid is entrained
into the jet is balanced by the return flow into the separation zone from the region of
jet attachment. The pressure in the separation zone is approximately constant, and is
lower than in the atmosphere on the opposite side of the jet. Figure 9.1(b) shows the
effect of opening a hole from the atmosphere into the separation zone. Fluid flowing
through the hole is entrained into the jet; the entrainment rate now balances the sum
of recirculation and inflow through the hole. The pressure in the separation zone rises
somewhat, the jet curvature is reduced and the separation zone lengthens. If the
inflow rate from the atmosphere is sufficiently high, the entrainment rate may be
insufficient to maintain the balance, and the jet will detach from the wall.
Returning to Figure 9.1(a), we see that having attached to the wall, the flow tends to
stick to it because separation would require a supply of fluid to the space between the
surface and the separating jet. Separation is therefore unlikely to occur unless the
curvature of the surface is unduly severe or if an adverse pressure gradient is
encountered.
The previous explanations have assumed twodimensional flow. This assumption will
be valid if the jet is very wide in the direction normal to the plane of the diagram in
comparison to its width. If this is not so, however, flow from the atmosphere round
the ends of the jet into the separation zone will have considerable effect. Because of
the importance of these end effects, the ratio of jet width (in the direction normal to
the diagram) to jet thickness has acquired a specific description, known as the aspect
ratio.
The principle of wall jet attachment has recently found an application in the
technology of fluidics. Figure 9.2 shows a typical fluidic switch in which the supply
jet S is directed to either outlet 01 or 02 depending on to which of the two walls the
jet attaches. Suppose the outlet 01 is active. By introduction of fluid at control hole
C1, the jet may be switched across to outlet 02, and it will remain there after the
inflow at C1 has ceased. Further action at C1 will not have any effect; to switch the
jet back to 01 requires a signal at C2.
101
The switch is a fluidic counterpart of an electronic flipflop and is called a fluidic
flipflop. Other switches or gates, such as AND and OR gates, may be constructed,
and fluidic logic circuits may be developed by interconnection.
Figure 9.2 A Fluidic FlipFlop
In the experiments which follow, the attachment of a jet to a single adjacent wall is
first studied, and the behaviour of a flipflop is then observed.
Description of Apparatus and Procedure
The essential features of the arrangement are shown in Figure 9.3. The equipment,
which fixes to the outlet flange of the contraction section of the airflow bench,
consists of a nozzle plate which houses a rectangular supply nozzle. The jet which
emerges from this nozzle is contained between side plates that may be moved laterally
so that the offset between the nozzle and the attachment wall be varied. The aspect
ratio of the nozzle may be altered by removing one of the nozzle blocks and fitting a
different sized block.
102
Figure 9.3 Details of Apparatus and Notation
For the singlewall tests, the lefthand attachment wall is fitted between the side
plates. It is mounted on a spindle which terminates at a control used to rotate the
attachment to any desired angle. For tests on the flipflop, a further attachment wall
and a splitter block are added. The space between the nozzle block and the
attachment wall may be left open to atmosphere or may be sealed by closing a flexible
seal as indicated.
The lefthand attachment wall is first fitted and the flexible seal attached. The
second, larger control is used to lock the wall at any desired angle. The righthand
seal should be in its open position. The offset, dimensioned y in Figure 9.3, and the
wall angle are both set to zero. The wind speed is then brought up to a convenient
value close to the maximum and is then held constant by maintaining a constant
airbox pressure, throughout the tests. The wall angle is now slowly increased until
separation is observed at angle
s
. This is most easily detected simply by holding the
hand in the jet, some 150 mm downstream of the trailing edge of the wall so as not to
interfere with the flow along it, and noting when the flow pattern suddenly changes.
The change is usually audible. The process should be repeated once or twice to
ensure that the value of
s
is established to within about 1. Then, starting with a
detached jet and reducing the wall angle, the value at which reattachment occurs
r
, is
established. The same procedure is repeated at several different values of offset y in
the range from approximately 4 mm to 20 mm. Intervals of 2 mm are recommended,
but smaller steps may be required where large changes in
s
or
r
are seen to take
place. The whole test may then be repeated with the flexible seal removed.
103
Proceed now to construct a flipflop by inserting the second attachment wall, and
fixing the flexible seals to both walls. Centralise the assembly on the centreline of the
nozzle. Starting with parallel walls, both set at zero wall angle, slowly increase the
wall angles until the jet is clearly attached to one or the other of the walls. Then lock
the walls in position. Try to switch the jet by prising open the seal on the attached
side. If this does not produce the desired switching, increase the angles in small steps
until it does. It should then be possible to demonstrate flipflop action, switching the
jet back and forth at will by briefly prising open the seals. Further increase the angles
to discover the upper limit at which a satisfactory flipflop action is possible.
The central splitter may now be added. It will be observed that this enhances the
bistability, as the range of wall angles which give satisfactory switching is
considerably increased.
Results
Nozzle width w = 10 mm
Nozzle breadth b = 100 mm
Aspect ratio = 10
Attachment wall length L = 100 mm
Attachment wall radius R = 9.5 mm
Length from nozzle to attachment wall x = 12.5 mm
Flexible Seal Closed Flexible Seal Open
y mm
s
()
r
() y mm
s
()
r
()
4
2
0
2
4
6
8
10
12
14
16
18
20
53
58
63
68
72
76
82
86
90
96
100
110
73
37
39
44
47
52
58
65
44
37
35
32
29
22
4
2
0
2
4
6
8
10
12
14
16
18
20
45
52
63
51
46
38
34
31
28
26
25
24
23
28
36
44
36
32
29
26
25
23
22
22
21
21
Table 9.1 Separation and Reattachment Angles at a Single Wall
104
Table 9.1 gives results obtained for separation and reattachment of the jet at a single
wall, and the results are plotted on Figure 9.4.
The following observations were made when two walls were used.
Offset y to each wall = 13.5 mm
Without splitter fitted:
= 15 Jet attaches to one wall and will not separate when seal is lifted.
= 20  30 Jet attaches to one wall and may be switched back and forth as a
flipflop.
= 35 Jet attaches to one wall but when seal is fitted it moves to centre
position instead of attaching to opposite wall.
With splitter fitted
= 15  80 Jet attaches to one wall and emerges from the passage between the
wall and splitter without spilling into the opposite passage.
Switches as a flipflop.
Discussion
Figure 9.4 exhibits the way in which the jet may be deflected through very
considerable angles, exceeding 90, by the Coanda effect. When the wall projects
into the jet, i.e. when y is negative, the behaviour is much the same with a seal fitted
as when it is removed. There is some 17
hysteresis between detachment and
reattachment in this range of y. As soon as the wall is moved out of the jet, however,
the two conditions behave entirely differently. With the seal fitted, the separation
angle and the reattachment angle both continue to grow until, at a value of y about 8
mm, the hysteresis range is suddenly increased by a sharp drop in the reattachment
angle. There is some 50 of hysteresis at this condition. The separation angle
continues to grow as y increases to 20 mm, reaching a maximum of about 110.
When the wall is unsealed, as y increases from zero, both separation and reattachment
angles decrease steadily and the hysteresis also decreases, diminishing to
approximately 3 when y reaches 20 mm.
105
Figure 9.4 Detachment and Reattachment for a Single Wall
The behaviour with two attachment walls is in fair agreement with predictions that
may be made from Figure 9.4. Consider the separation condition. At y = 13.5 mm
(which is the value used in the experiment with two walls), Figure 9.4 shows that the
jet will separate from the unsealed wall at
s
= 26. The experimental result observed
when two walls were used is that for
s
greater than 20, the jet detaches when the
seal is lifted. So separation occurs at a somewhat smaller angle when there is a
further sealed wall on the other side of the jet. Again, at y = 13.5 mm, Figure 9.4
shows that the jet will reattach to a sealed wall at
r
= 35. This agrees exactly with
the observed behaviour of flipflop action up to 30, but for = 35 or more, the jet
moves to the centre position when the seal is lifted instead of attaching to the opposite
wall.
106
The presence of a splitter improves the bistability. It generates a recirculation, as
indicated in Figure 9.2, which strengthens the attachment very considerably, thereby
increasing the range of successful flipflop action.
Conclusion
Tests on a singlewall configuration have shown that the wall attachment effect, or
Coanda effect, may be used to divert a plane jet along a plane wall through angles up
to 90 or more. Considerable hysteresis is found between the condition for jet
separation and that for jet attachment. The effect of offset between the jet and the
wall, and venting the space between the nozzle and the wall, have been investigated.
A fluidic flipflop which exploits the phenomenon of wall attachment has been
constructed and satisfactory switching has been observed.
Suggestions for Further Experiments
1. Investigate the effect of changing the aspect ratio of the jet on the singlewall
detachment and reattachment characteristics. Change the nozzle width to, say,
8 mm and 5 mm in turn.
2. Determine the significance of the side plates on the singlewall results. Blank
off, 20 mm at each end of the breadth of the nozzle with adhesive tape, leaving
40 mm clear in the middle, so that the jet is not in contact with the side plates
when it emerges from the nozzle, and repeat the tests.
3. Consider the possibility of an asymmetric fluidic switch. Suppose it were
desired to produce a switch which stayed on until a control signal switched it
off. Could this be done and if so, can such a switch be made from the parts
described here?
107
108
10. FLOW VISUALISATION
AF17 Flow Visualisation Apparatus
109
Introduction
In experiments with fluid mechanics, it is often necessary to study the nature of the
motion by direct observation of part or of the whole of the flow pattern. Important
features may be observed, such as regions of steady or unsteady flow, thickening
boundary layers and separation, secondary flows and so on. Flow visualisation can
show characteristics of the motion, which might previously have been totally
unexpected. It frequently explains phenomena which may otherwise defy explanation.
It sometimes provides the starting point for new theoretical or analytical studies. A
good historical example is provided by Osborne Reynolds visualisation of laminar
and turbulent flows in glass tubes, which was the starting point for a rational
understanding of the resistance to flow experienced by a fluid as it moves along a
pipe.
There are many techniques for visualisation. In water, dye filaments as used by
Osborne Reynolds are still used. A more recent innovation is the use of tiny bubbles
of hydrogen produced by an electrode which sheds a sheet of bubbles, or which may
be arranged to shed discrete streams of bubbles which behave like dye filaments. In
air, the most common practice is to use smoke injected through a tube or a row of
tubes, usually called a rake. The smoke must be of nearly neutral density so that it
does not rise or fall through the flow due to the effect of gravity and needs to be quite
dense if the traces are to be observed for more than a short length downstream of the
injector tube.
Description of Apparatus
Figure 10.1 shows the main parts of the flow visualisation module. Smoke from the
generator passes through flexible tubing into a streamlined manifold that spans the
duct at the inlet to the working section. This duct contracts from a settling chamber,
which contains honeycomb and gauzes to reduce turbulence, to the working section.
Smoke emitted from the rake of tubes (smoke comb) accelerates with the airflow
along the inlet duct into the working section and the filaments may be observed
through the clear plastic front surface against a matt black background. A strong,
diffused light from either side can help to give good contrast. The apparatus exhausts
through the outlet in the bench top; a flexible air hose should be fitted to conduct the
exhaust to a suitable ventilator, otherwise the smoke will easily fill a laboratory in
minutes. The smoke is harmless, but you must ventilate it away for safety and
common sense.
110
Figure 10.1 Flow Visualisation Apparatus
The individual filaments retain their separate laminar identities for the whole length
of the working section for speeds up to approximately 1 m/s. It is important to adjust
the injection rate so that the velocity of injection matches the air velocity past the
injection tubes, otherwise premature turbulence in the smoke filaments is likely to
occur. It is recommended that students gain a little experience in setting the air
velocity and smoke injection velocity with the working section clear of any models,
before proceeding to visualisation of flow around various bodies.
Operation
Important make sure that you use this equipment in a wellventilated area and the
wind tunnel outlet is directed to a suitable air extractor.
Make sure the wind tunnel fan is off. Fit your chosen model into the working section.
Put the smoke generator onto the wind tunnel bench top as shown in the picture at the
start of this section. Connect the outlet of the smoke generator to the smoke comb
connector. Refer to Appendix B for full details on operation of the equipment.
111
Typical Results
Various objects may be placed in the working section and the flow pattern observed.
Three examples of the models supplied for use with the apparatus are described
briefly here. Figure 10.2 shows the flow around a circular cylinder. The motion over
the front part of the cylinder is steady as indicated by the almost unwavering smoke
filaments. Separation occurs at around 80 from the front of the cylinder. A wake
forms, which is shown to be unsteady by the mixing of the smoke. The unsteadiness
is transmitted to the flow outside the wake. It is interesting to note that the pressure
in the separated flow, as indicated by the surface pressure readings recorded in Figure
5.6, is almost constant.
Figure 10.2 Flow around a Circular Cylinder
112
Figure 10.3a Flow over an Aerofoil 
Small Incidence
Figure 10.3b Flow over an Aerofoil 
Large Incidence
Figure 10.3a shows flow over an aerofoil at a small incidence. The flow remains
attached to the surface over almost the whole chord; this represents the normal or
unstalled condition, at which useful lift is generated, while the drag is comparatively
small. The lift is due to the difference in pressures on the upper and lower surfaces of
the aerofoil. Over the upper surface, convergence of the smoke traces indicate an
acceleration of the flow, particularly over the first quarter of the chord, and this is
accompanied by a fall of pressure which contributes much of the total lift on the
aerofoil.
113
Figure 10.3b shows what happens if the angle of incidence is increased too much.
The flow no longer sticks to the upper surface but separates, causing stall. The lift is
reduced considerably, and the drag increases as a consequence of the wider wake
which results from the separation.
Figure 10.4 Flow through a SharpEdged Slit
Figure 10.4. shows an example of flow through a sharpedged slit. The contraction of
a vena contracta at approximately one halfslit width downstream of the edges can
readily be seen.
114
Further Observations
1. Extend the observations to flow around a flat plate placed across the stream,
flow along a long straight, flow through a roundedged slit, flow through a
convergentdivergent pipe, and other cases which come to mind. Sketch the
flow patterns and identify unsteady zones, separation points and so on.
2. The photographs of Figures 10.2 to 10.4 were taken with the smoke filaments
running in the mid plane between the front and rear walls of the working
section. Repeat these tests and observe what happens when the smoke is
directed close to the front wall. If different patterns are now observed, what
inferences may be drawn? Do you consider that secondary flows are present,
and if so, how do they arise?
115
116
11. AEROFOIL WITH TAPPINGS (AF18)
Note: Professor Markland did not create this section of the manual. TecQuipment
created this section to complement Professor Marklands work.
The Aerofoil with Tappings (AF18)
117
Introduction
The Aerofoil with Tappings fits onto the AF10 and shows the pressure distribution
around a symmetrical NACA0020 aerofoil and the characteristics of lift.
It is a smallscale model wing, but the results from its tests can be scaled up to
compare with larger aerofoils.
Description
A clearsided duct contains the aerofoil in closed ends arrangement, so airflow is
only across the curved surfaces of the aerofoil, and not around its ends (or wingtips).
This stops any wingtip vortices or drag caused by the wing tips. Therefore, the
aerofoil shows twodimensional flow. This arrangement also gives the aerofoil a
theoretical unlimited span or infinite span. The duct holds the aerofoil in a vertical
position, relative to the ground.
Pressure tappings are along the top and bottom surfaces of the aerofoil. They connect
to a set of numbered, small pipe connectors on a plate next to the aerofoil. A set of
larger bore pipes (supplied) connects the numbered pipe connectors to the AF10A
Inclinable Multitube Manometer or other suitable manometer. Just above the aerofoil
at the inlet to the duct is an extra pressure tapping for measurement of the static
pressure upstream of the aerofoil, and for use with the tapping on the AF10 to
calculate air velocity upstream of the aerofoil.
A control on the front of the unit allows the user to adjust the aerofoils angle of
incidence (also known as the angle of attack). A scale shows the angle, relative to
the airflow from the AF10.
Figure 11a. Angle of Incidence
118
Tapping Number Tapping Position
From the leading edge (mm)
1 2.2
2 3.9
3 6.1
4 8.7
5 11.8
6 14.8
7 20.0
8 25.6
9 31.4
10 37.3
11 43.4
12 49.5
Table 11a. Tapping Positions
Table 11a shows the positions of the tappings along the surface of the aerofoil. Note
that tappings 1, 3, 5, 7, 9 and 11 are all on one side of the aerofoil. Tappings 2, 4, 6, 8,
10 and 12 are all on the other. Table 11b shows the details of the aerofoil.
Figure 11b. The Aerofoil
Aerofoil Type NACA0020 (symmetrical)
Aerofoil Chord 63 mm
Aerofoil Wing Span 49 mm
Effective surface area 0.0031 m
2
Table 11b. Aerofoil Details
119
Theory
Figure 11c. Pressure around a symmetrical Aerofoil
120
In aerodynamics and fluid mechanics, an object in a flow produces lift when the
pressure on one side of the object is greater than the pressure on the other.
A symmetrical aerofoil in an airflow produces no lift when its angle of incidence is
zero, as the pressures above and below the aerofoil are equal.
As the angle of incidence increases positively from 0, the aerofoil starts to produce
lift. Pressure decreases above the main part of the aerofoil as the incidence angle
increases. The aerofoil shape and the airflow velocity determine the pressure
distribution above and below the aerofoil.
When the incidence angle reaches a certain point (the stall angle), the airflow over the
upper surfaces separate from the wing. The pressure distribution changes and the
centre of lift moves its position. The aerodynamic centre of the aerofoil moves
backwards. These factors stall the aerofoil, so the lift reduces and drag increases.
Engineers use lift curves to show the performance of an aerofoil (see Figure 11d).
They test the aerofoils lift at a range of incidence angles for any given air velocity,
until it stalls. This shows them the operating area for the aerofoil at that velocity.
The angle (slope) of an ideal lift curve from theoretical calculation is 2 C
L
/radian,
but real results are never be greater than the absolute maximum ideal value of 5.7
C
L
/radian.
Figure 11d. Lift against Incidence Angle for a Symmetrical Aerofoil
You can measure the performance of a wing in two ways:
1. By direct measurement  with force sensors on the aerofoil.
2. By measurement of the pressures on the surface of the aerofoil, relative to
local air pressure.
The AF18 allows students to measure the pressures on the aerofoil surface. The
results can be shown as readings of pressures against and tapping positions. For
comparison with other aerofoils of different sizes you must convert your readings into
dimensionless values. These are the pressure coefficient and the chord ratio.
121
2
2
1
=
V
p p
C
p
Where:
C
p
= Pressure Coefficient (dimensionless)
p = Pressure at a given point (tapping) (Pa or N.m
2
)
p
= Static Pressure (Pa or N.m
2
)
= Density (kg.m
3
)
V
) ( 2
eff airbox
p p
V
=
For each tapping point on the aerofoil, find the pressure coefficient with the equation:
2
2
1
=
V
p p
C
eff n
p
Where p
n
= pressure at the tapping.
The aerofoil is in a vertical position, but to understand the results more easily, you
must consider pressure tappings 1, 3, 5, 7, 9 and 11 to be on the lower surface of the
aerofoil with respect to the airflow, and tappings 2, 4, 6, 8, 10 and 12 to be on the
upper surface of the aerofoil.
For each angle, plot the pressure coefficient against the chord ratio (mentioned in the
theory). Draw bestfit lines for the pressures on the upper and lower surfaces. To
compare your results with theory, extend your curves to the zero pressure coefficient
line at a chord ratio of zero and 1. You should find that the pressure coefficient near
to the leading edge is zero, because the aerofoil forces the air to stop moving at this
(stagnation) point. The exact position of this stagnation point changes with incident
angle.
The area between your curves is the coefficient of lift. Use the trapezium rule or other
suitable method to find this area for each angle.
Plot the coefficient of lift and compare it to the lift curve shown in the theory.
From your curve, find:
The gradient of the lift curve
The maximum lift coefficient
The stall angle
What do you notice about the pressure distribution at angles greater then the stall
angle?
125
Typical Results
Pressure Profile at 0 degrees
4
3
2
1
0
1
0.0 0.2 0.4 0.6 0.8 1.0
X / C
C
o
e
f
f
i
c
i
e
n
t
o
f
P
r
e
s
s
u
r
e
C
P
Pressure Profile at 10 degrees
4
3
2
1
0
1
0.0 0.2 0.4 0.6 0.8 1.0
X / C
C
o
e
f
f
i
c
i
e
n
t
o
f
P
r
e
s
s
u
r
e
C
P
Pressure Profile at 17.5 degrees
4
3
2
1
0
1
0.0 0.2 0.4 0.6 0.8 1.0
X / C
C
o
e
f
f
i
c
i
e
n
t
o
f
P
r
e
s
s
u
r
e
C
P
126
Pressure Profile at 20 degrees
4
3
2
1
0
1
0.0 0.2 0.4 0.6 0.8 1.0
X / C
C
o
e
f
f
i
c
i
e
n
t
o
f
P
r
e
s
s
u
r
e
C
P
Pressure Profile at 22.5 degrees
4
3
2
1
0
1
0.0 0.2 0.4 0.6 0.8 1.0
X / C
C
o
e
f
f
i
c
i
e
n
t
o
f
P
r
e
s
s
u
r
e
C
P
Pressure Profile at 25 degrees
4
3
2
1
0
1
0.0 0.2 0.4 0.6 0.8 1.0
X / C
C
o
e
f
f
i
c
i
e
n
t
o
f
P
r
e
s
s
u
r
e
C
P
127
Lift against Incidence Angle
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 5 10 15 20 25
Incidence Angle (Degrees)
C
o
e
f
f
i
c
i
e
n
t
o
f
l
i
f
t
(
C
L
)
Conclusions
The results for the aerofoil show a stall angle of approximately 20 degrees and a
maximum lift coefficient of approximately 1.25. TecQuipment used the trapezium
rule to measure the area between the curves, which gave a dimensionless
measurement of lift coefficient. At angles greater than stall, the pressure peak near
the front of the aerofoil drops, as the air flow separates from the upper surface.
The slope of the curve is approximately 0.08 C
L
/degree or 4.56 C
L
/radian.
128
APPENDIX A: DETERMINATION OF THE AREA BENEATH A CURVE
We sometimes need to evaluate the area enclosed below a curve that has been
established by experiment. Suppose that such a curve, as shown in Figure A.1, has
ordinates y
0
, y
1
, y
2
.y
n
, spaced at equal intervals h over the range from zero to L in
the xdirection. The true area under the curve is
A = ydx
o
L
We aim to approximate to this result in terms of the ordinates y
0
, y
1
, y
2
, ... y
n
, as
measured from the curve.
Figure A.1
The Trapezoidal Rule provides the simplest approximation. Imagine the curve to be
replaced by the straight lines shown dashed in Figure A.1. The area of the set of
trapeziums produced by this replacement is
A
t
= ( ) ( ) ( )
1
2
1
1
2
1 2
1
2
1
y y h y y h y y
o n
+ + + + +
..... h
n
which reduces to
A
t
=
( )
 
1
2
2
0 1 2 1
n
y y y y y
n n
+ + + + +
...... L
A1
The trapezoidal rule introduces error, which is obviously the sum of the differences
between the areas enclosed by the curve and by the trapezoids. A much smaller error
is given by Simpsons Rule. This gives the area enclosed under the curve between
three successive points such as y
0
, y
1
, and y
2
as
*
A
s
= ( )
1
3
4
1 2
y y y
o
+ + h
Applying this rule repeatedly over the whole set of ordinates from y
o
to y
n
gives the
result
A
s
= ( )
1
3
4
1
3
4
1
3
4
1 2 2 3 4 2 1
( ) .... ( ) y y y h y y y h y y y
o n
+ + + + + + + +
h
n n
which reduces to
A
s
= ( ) ( )
 
1
3
4 2
1 3 1 2 4 2
n
y y y y y y y y
o n n
+ + + + + + +
.... ....
L
This repeated application of Simpsons rule obviously involves an even number of
intervals, i.e. n will be an even number. If, for some reason, it is necessary to divide
the length L into an odd number of intervals, then Simpsons rule may be used up to
the penultimate interval, and the trapezoidal rule then used for the remaining last step.
For example, when analysing Figure 5.7, we need the area
A = c d
p
o
cos
The mean of the two curves of Figure 5.7 is reproduced in Figure A.2, with the
required area shown shaded. Note that part of the area, from about 35 to 90, is to be
reckoned negative. The following ordinates have been measured from the curve:
*
Simpsons rule is exact for any cubic curve of the form
y = a + bx + cx
2
+ dx
3
So error arises only from term of 4th and higher orders of x. This is very much better than the
trapezoidal rule, which is exact only to 1st order of x.
A2
() 0 20 40 60 80 100 120 140 160 180
c
p
cos 0.98 0.62 0.18 0.52 0.17 0.16 0.47 0.72 0.88 0.92
Figure A.2
Since the number of steps is odd, Simpsons rule is used in the range 0 to 160, and
the trapezoidal rule for the last step from 160 to 180.
For 0 to 160:
Simpsons rule with n = 8, L = 160 = 160 (/180) radians:
( ) ( )
 
A = 0.98 +4 0.62 0.52 +0.16 +0.72 0.18 0.17 +0.47 +0.88 160
s
1
3 8
2
180
+

\

.

A
s
= 0.70
(Note the negative ordinates, and the conversion factor (/180) from degrees to
radians)
For 160
o
to 180
o
:
Trapezoidal rule with n = 1, L = 20 = 20(/180) radians:
A3
 
A = 0.88 + 0.92 20
180
t
1
2 1
0 31

\

.

=
.
Total area A from 0 to 180 : A = A
s
+ A
t
= 1.01
So
c
p
o
cos d
= 1.01
Figure A.3
As a further example, consider Figure 7.5, reproduced as Figure A.3, from which it is
required to evaluate the integral:
* = ( ) 1
u U dy
o
A4
This is represented by the shaded area. Note that (1 u/U) is dimensionless, and y has
dimensions of length, being measured in units of mm. So the shaded area will also
appear in units of mm.
Values of (1 u/U) read from the curve are as follows:
y (mm) 0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 >5.0
(1u/U) 1.00 0.25 0.15 0.10 0.06 0.04 0.03 0.02 0.01 0.01 0.00 0.00
Simpsons rule with n = 10, L = 5.0 mm:
( )
( )
A =
1.00 +4 0.25 +0.10 +0.04 +0.02 +0.01
0.15 +0.06 +0.03 +0.01 +0.00
s
1
3 10 2
5 0
+
. mm
A = 0.53 mm
So
* = ( ) ( ) 1 1
5
= =
u U dy u U dy mm
o o
/ / 053 .
A5
A6
APPENDIX B: INSTALLATION AND OPERATING INSTRUCTIONS FOR
THE AF17 FLOW VISUALISATION APPARATUS
Installation
NOTE: For safety and local transport regulations, the gas cylinder is supplied
empty. You must fill it with compressed carbon dioxide before use.
1. Use the Airflow Bench in a wellventilated area.
2. Remove the standard contraction piece from the AF10 Airflow Bench
3. Put the bottom (outlet ducting) of the smoke tunnel into the hole in the
Airflow Bench.
4. Use the four clips of the airbox to fix the smoke tunnel contraction piece and
visualisation section in place.
5. Attach the outlet ducting to the bottom of the visualisation section using the
two clips.
6. Attach the exhaust pipe to the bench outlet and connect it to a suitable
laboratory ventilation system, or out to atmosphere. Do not vent directly to
atmosphere in windy conditions, as this can seriously affect the airflow.
7. Put the smoke generator in position on the bench top. Put the gas bottle in its
holder to the back of the Airflow bench. Connect the regulated output of the
gas bottle outlet to the smoke generator. Fit the metal rake adaptor to the
flexible tube (supplied). Fit the other end of the flexible tube to the outlet of
the smoke generator. Connect the rake adaptor to the smoke rake connection.
8. Make sure that the smoke generator switch is off, and connect its mains lead to
one of the power sockets on the Airflow Bench.
A7
The Smoke Generator
Principle of Operation
The smoke generator works by heating nontoxic foodgrade oil and forcing it out of
the smoke generator by means of compressed carbon dioxide gas. The oil vapour
condenses and becomes smoke when it meets normal cold air at its outlet. The
smoke is actually a suspension of oil drops in the air, rather than a product of
combustion.
Operating the Smoke Generator
Read the smoke generator manufacturers instructions and make sure that the smoke
generator has enough oil in its reservoir.
Isolator
Valve
Delivery
Pressure
Control Valve
Figure B.1
There are two pressure gauges on the compressed gas bottle. One shows the pressure
available from the bottle, the other shows the delivery pressure after the control valve.
There are two valves at the bottle. The isolator valve is on the top of the bottle to
completely isolate the gas supply. The pressure regulator control valve is between the
gauges and sets the outlet (delivery) pressure. Shut the control valve (turn clockwise).
Fully open the isolator valve.
Switch on the smoke generator. Its display will show numbers, one is the heater
temperature and the other is the temperature that it needs to reach so that it works
(approximately 314 degrees Celsius). Switch on the smoke button. Slowly open the
control valve until the delivery pressure is at 1 bar. Switch off the smoke button.
When the numbers in the display become the same, the ready lamp will go on.
A8
A9
Operating the AF17
The experiment is supplied with four models:
Aerofoil
Flat plate
Cylinder
Sharpedged orifice
1. Remove the back panel of the visualisation section and attach a model.
Replace the back panel.
2. Start the smoke generator as described earlier and press its smoke button.
3. Open the valve on the airflow bench to approximately one quarter of fully
open. If the air speed is too high, the smoke trails will be poor and the smoke
may be forced back out of the smoke generator and into the room.
4. After a few minutes smoke will begin to flow into the visualisation section.
Controlling the Smoke
1. The airflow bench control sets the quality of the smoke trails. You cannot
control the amount of smoke from the smoke generator. A low air flow rate
gives the best results, but you must experiment to find the best setting for your
model.
2. The sharpedged orifice obstructs the flow the most, so you must take more
care to adjust the flow correctly for this model. Too much or too little air flow
will force smoke to come out of the smoke generator outlet.
3. If condensation happens inside the smoke rake, its holes may become blocked.
To unblock the holes, switch off the smoke generator, remove the back of the
visualisation section and use a clean cloth or tissue to wipe the smoke rake and
absorb the oil.