Fan Engineering

Woods Practical Guide
TO
Fan Engineering
Foreword
We depend increasingly on the movement of air and gases for
our comfort and survival. This is not confined to ventilation and air
conditioning but includes heat transfer, drying, blowing, particle
transportation and filtering, the control of environments and the
extraction of life threatening fumes and hot smoke.
The prime vehicle for moving air and gases is the fan. The
relatively traditional appearance of fans conceals recent advances
in aerodynamic, mechanical and acoustic technology. The demand
for improvements in performance gathers pace with environmental
awareness, safety legislation and the need to satisfy EC Directives.
The knowledge needed to make these advances can be found
within.
This book, published by Woods of Colchester limited, is the third
edition of a work first issued in July 1952 and was originally aimed
at those concerned with practical fan engineering problems.
Distribution of the first edition which was translated in to French,
exceeded 20,000 copies, being surpassed by that of the second
edition, first published in June 1960, which reached about 27,000
copies. This included translations into French, German, Spanish
and Italian. In the years since the second edition was issued the
fan industry has continued to expand both in demand for its
products and in the technological sophistication of customer
applications and requirements. Moreover, in these intervening
years the growth of sales in a world-wide sense has emphasised
the need for a more systematic presentation of information
relating to fan performance and the newer terminology associated
with different national practices.
This revised edition of the guide, like its predecessors, is aimed
at the practical as well as the professional engineer, and contains
many worked examples. It provides a comprehensive compendium
of information relating to fans, fan drives, fan selection and their
environment of application. It commences with two chapters
describing the ventilation requirements of the individual and of
occupied spaces subject to various thermal environments caused
by industrial, human and solar inputs. The next three chapters
continue with a description of methods of heat rejection and
heating, including discussion of such important current issues as
pollution control and explosion hazards. The heart of the book
treats extensively, in Chapters 6 and 7, the details of
determination of system losses, fan performance specification and
fan sizing. There the scope covers many climatic conditions whilst,
for the more mathematically-minded, the basic theory of fan
pressure generation is included.
vi Foreword
An important feature of present-day society is the interest in

efficient use of energy. Often fans are required to operate over
their characteristic rather than merely at the flow corresponding to
fan best efficiency. In such cases both the motor drive and the
method of fan duty control are of primary importance. Chapters 8
and 9 provide clear illustrations of the merits of different electrical
drives and fan duty controllers necessary to prevent unwanted
energy dissipation. The increasing relevant subjects of noise and
vibration are dealt with in Chapter 10 and include methods for
calculation of fan sound power and assessing vibration
acceptability. The problems of fan testing and conventions are
discussed in the next chapter.
Heat transfer now represents one of the largest applications for
fans and this topic is covered in Chapter 13, Heat Exchange and
Drying. The growth in population, urbanisation and transport has
expanded the demand for the ventilation of tunnels, mines and
underground spaces and this is covered in Chapter 14. A range of
tables and charts is provided whose value and usefulness are
enhanced by bringing together data on many topics within a single
accessible reference.
This foreword could not be considered complete without a
sincere and grateful acknowledgement to the personal effort of the
author, the late Mr. B.B. Daly, formerly Technical Director of
Woods of Colchester Limited. This Guide represents the last major
task of Mr. Daly for Woods and, as the reader will appreciate on
each occasion he consults this book, Mr. Daly has brought together
the practical experience of a lifetime of active involvement in the
fan industry.
Those who read, comprehend and properly apply all the
principles contained in this book will truly be well qualified in Fan
Engineering. I hope that the benefit of this knowledge brings you
success.
D. J. PRIEST
Managing Director
Woods of Colchester
Limited.
Author's Preface
This, the third, edition of Woods Practical Guide to Fan

Engineering conforms entirely to the principles and objectives which
governed the first edition in 1952. It is intended to help the user of
fans-and there is hardly a field of industrial or environmental
activity in which they are not used-and the application engineer
responsible for their proper selection and installation.
The text has been completely rewritten, but this must not be
taken to imply any disrespect for the work of the original editors,
W. C. Osborne and C. G. Turner. The continuing demand for the
book shows how well they struck and held the right note in their
treatment of the subject. It is the development of the subject
matter itself over the last 25 years which has necessitated
extensive revision.
Increasing concern for comfort and safety and the demand for
better working conditions call for environmental control in many
new situations. Growing interest in energy conservation is served
by increasing sophistication in process control. Empiricism has often
been superseded by logical design and a more professional
approach fostered by such bodies as the Institution of Heating and
Ventilating Engineers (now the Chartered Institute of Building
Services) and the Heating and Ventilating Research Association
(now the Building Services Information and Research Association).
Great developments have taken place in the number and quantity
of national standards, and more often than not these now have the
advantage of international consensus. This helps the systematic
presentation of data, and where standards are lacking tables of
"typical" values have been freely used in this book. These give the
engineer a quick, clear picture of the range of possibilities open to
him-to be filled in by the detailed and up-to-date information which
can only be furnished by the actual supplier of the material,
equipment or service in question.
As in earlier editions mathematical analysis which would be
beyond the competence of many readers has been avoided. This
does not mean that figures, formulae and calculation methods are
absent; these are the very essence of the engineer's daily work,
and the primary object of the book is to present them in a way that
is simple to use and understand
The book is unlikely to satisfy the professional specialist in one of
the many fields covered, but every practising engineer knows how
often he requires to understand the essentials of someone else's
speciality. In a similar way it is not intended for the fan designer,
but for those engineers -hundreds of times more numerous-whose
combined contribution is
viii Author's Preface
just as essential to the success of each fan installation. It is not,

however, a mere compendium of practical rules since an attempt
has been made throughout to clarify principles for the benefit of the
student.
The author's thanks are due to the directors of Woods of
Colchester Limited for their encouragement and support, to J. P.
Nelson for organising production of the book, and to Messrs. C. L. B.
Noon and W. Woods Ballard for reading and checking the text. The
treatment is not limited to the practice and environment of the
British Isles but is backed by the experience of Woods of Colchester,
a company exporting to over 120 countries, worldwide.
x Contents
Contents xi
xii Contents
Contents xiii
xiv Contents
Contents xv
xvi Contents
Contents xvii
xx Index to figures and tables
Index to figures and tables xxi
xxii Index to figures and tables
Index to figures and tables xxiii
xxiv Index of figures and tables
Introductory notes on
SI and other units
This, third, edition of Woods Practical Guide to fan Engineering

has been written in the International Standard System of Units
called "SI" (Systeme International).
The student will already be familiar with the new system. The
purpose of these introductory notes is to help the engineer more
familiar with the Imperial or technical metric systems.
Throughout the book the basic SI units have been used for all
formulae and calculations. It is the great strength of this form of
the SI system that all such formulae develop logically one from
another without any arbitrary constants. A few П's appear, but they
are there because of the geometry of the application. The basic SI
system has what is called "coherence": every unit is built up
directly from the basic units of length, mass and time. By choosing
the metre, kilogramme and second for these units, the joule and
the watt are brought in for energy and power, and the established
electrical system-volt, ampere, etc. follows.
The metric and general (not basic) SI systems of course use
larger and smaller units as well. The SI system employs only
multiples and submultiples of 103, 106, 109, etc. (except for 10 and
100 as in the decimetre and centimetre which we have avoided). It
is of course natural to think and speak of distances across country
in kilometres or the dimensions of components in millimetres. This
practice has been followed where appropriate in the general text,
but we strongly recommend that, before making any calculations or
using any formulae, each quantity is converted into basic SI units.
The result will be itself a basic SI unit which can then, and only
then, be converted back into conventional form if desired.
This book will be no exception to the rule that all good rules have
their exceptions. "Air changes per hour" appear in an early chapter
for the ratio of the ventilation rate to the volume of the space
ventilated. These form convenient numbers and are in universal
currency the world over. No one talks seriously in terms of
kiloseconds or megaseconds of time, although the millisecond,
microsecond, and nanosecond (10-9 s) are well established in
science and technology. Some remarks follow on the practice
adopted. Basic SI units are in heavy type.
Length, Area and Volume. The metre, the square metre and the
cubic
metre are appropriate to all purposes of fan and ventilating tech-
nology. The kilometre is reserved for travel distances and the
xxvi Introductory notes on SI and other units
millimetre for component dimensions. The micrometre (µm) is

the right size for limits and fits of machined components.
Velocity. The metre per second is a good unit for all air
movement purposes. In talking of vehicle speeds, it is recognised
that the kilometre per hour will prevail.
Volume Flow. All calculations are performed in cubic metres per

second and this is the appropriate unit for general use in
industrial fan technology. The litre per second (10-3m3/s) is likely
to be more popular for small fan ratings but should be translated
for calculations.
Rotational Speed. The revolution per second is the unit used in this
book, although the more familiar revolution per minute is
sometimes added, in brackets, in the text. Strictly the basic SI
unit is the radian per second which, to a mathematician, is the
natural unit of angular velocity. The engineer will feel that the
complete revolution has the stronger grip on reality.
Mass, Mass Flow and Density. The kilogramme, the kilogramme

per second and the kilogramme per cubic metre are suitable
units for all purposes of fan technology. The gramme and
milligramme are more convenient for chemical analysis, and the
tonne (= 1000 kg) for bulk quantities of massive materials, but
the kilogramme is the basic unit.
Force. It is most important that the newton should become

established in our practice, and indeed our thinking, as the unit of
force. The greatest benefit to be derived from the SI system is
the elimination of the confusion caused by the gravitational units,
even when these are properly named the kilogramme force (=
9.81 N) and the poundforce (= 4.45 N). While the cgs dyne (=
10-5 N) is as logical as the newton, it is much too small for
practical use. Earlier attempts to overcome the confusion
between mass (e.g. the slug) and force or weight (e.g. the
poundal) can pass into deserved oblivion.
Pressure and Stress. Throughout this book the pascal (the name
now internationally adopted for the newton per square metre)
is used as the unit of pressure. 1 Pa (equal to about 0.1mm or
0.004in of water) is as small a unit as is ever needed for ordinary
experimental work with air, while the fact that system pressures
often run to a few thousand Pa does not necessitate the
introduction of the kilopascal, a unit which would always need
decimal places in our field.
Barometric pressures at around 105 Pa are admittedly
inconveniently large, and deference is paid to meteorological
practice by putting the millibar equivalent in brackets in the text
(1 mb = 100 Pa). Nevertheless, important formulae contain the
ratio of fan pressure to barometric pressure and it is important to
maintain consistency. Modern barometers are calibrated in
millibars, and the old mm or in of mercury should gradually
disappear.
Stress in a material is the same quantity as pressure, and the
megapascal (MPa) is a suitable unit. Since most stressed
components will be dimensioned in millimetres rather than
metres, it is convenient,
Introductory notes on Sl and other units xxvii
though heretical, to think of unit stress as one "newton per

square millimetre", which equals 1 MPa.
Energy, Power and Heat. The joule (= 1 newton-metre) and the

watt (= 1 joule per second) are the basic SI units of energy and
power respectively. It is an important advantage that the units
are the same whether the energy is mechanical, electrical or in
the form of a quantity of heat. Similarly, the rate of flow of heat
is an energy flow, or power, and is measured in watts. The old
mechanical equivalent of heat (4.18 joules per calorie, or 3412
BTU's per kilowatt-hour) is gone - banished by the definition of
absolute temperature in thermodynamic terms.
The horse-power, cheval-vapeur, or pferdestarke are on the way
out, too. Electric motors are now rated in kW, though here there
is a pitfall to be avoided. Always recognise and state whether it is
(mechanical) power output in kW or (electrical) power input in
kW that is meant.
The convenient relationship that one kilocalorie raised the
temperature of one kilogramme of water by about 1 °C is lost,
but for the fan and ventilating engineer there is a bonus. The
specific heat of air at constant pressure can be taken as 1000
joules per kilogramme per degree C for practical purposes
under atmospheric conditions.
The joule and watt will be used consistently in all formulae in
this book. Of course a result in watts can be converted into
kilowatts at the end, but it remains important to retain the basic
SI calculation principle.
Expressions for heat conduction, conductance and conductivity
will be found slightly less cumbersome with the use of watts per
square metre than they were in the older units.
Viscosity. The basic SI unit of dynamic viscosity is the

pascal-second and of kinematic viscosity the square metre per
second. These are admittedly inconveniently large for gases, and
tabulations commonly remain in centipoise (1 cP = 10-3 Pa/s) and
centistokes (1 cSt = 10-6m2/s). Of course, the basic SI units (to
which no names have yet been allotted) must be used in the
formulae for Reynolds and other dimensionless numbers.
xxviii Introductory notes on Sl and other units
Introductory notes on Sl and other units xxix
xxx Introductory notes on Sl and other units
Introductory notes on Si and other units xxxi
CHAPTER 1
The air and
human
well-being
Man's sense of comfort and his capacity for work deteriorate

quickly in poor air conditions. More seriously, his general health
may be impaired in the long term by living and working in
ill-ventilated buildings or an enervating climate.
The relevant factors are the temperature and relative humidity of
the air, its circulation and movement over the body, and its purity.
It is the function of the Heating and Ventilating Engineer to keep
these within satisfactory limits.
Ventilation of buildings will give complete control over air
movement and purity. Together with heating it will deal adequately
with temperature and humidity and will produce tolerable conditions
even in difficult cases. Ventilation means the supply of fresh outside
air to a building, its circulation within the building, and its discharge
together with the various polluting substances it has picked up.
Air Conditioning adds the functions of cooling and positive control
of humidity. With the complete system a good measure of comfort
can be assured however severe the climate and weather.
Continuous control is a vital factor and since tastes differ, the ideal
system will include some individual control over temperature and
air movement.
1.1 The Air

1.1.1 Composition of the air
The natural atmosphere up to 11,000m altitude (at which point
the troposphere in which we live ends and the stratosphere begins)
consists of gases well mixed in the following proportions:
c 1
2 Woods Practical Guide to Fan Engineering
The reference of percentages to the dry air part of the total mixture is a
useful convention in view of the highly variable nature of the water content. It
also has advantages in calculation which will appear in later chapters. The air
contains small proportions of many other substances, most of which can be
classed as impurities.
1.1.2 Oxygen is of course the vital component for life support, and indeed
man can function effectively when breathing pure oxygen alone. The average
person at rest passes some 500 litres of air per hour through his respiratory
system. This air carries 140 grams of oxygen, of which some 15% is absorbed
into the bloodstream, a consumption rate of 21 gm per hour.
Oxygen deficiency, with its symptoms of weakness and lethargy, may arise
in two ways. Without any change in the proportion of oxygen, the density
may be reduced, as at high altitude, to the point where the respiratory system
cannot handle the additional volume required. More rarely, the proportion of
oxygen may be excessively reduced by consumption, dilution or displacement.
Consumption or dilution occur in enclosed spaces such as submarines or
sewers. The danger of displacement may arise from heavy vapours left at the
bottom of empty tanks or vats. 25% reduction in available oxygen can be
serious for those with respiratory or cardiac weakness. 50% is a severe test
for the fittest, without acclimatisation.
1.1.3. Nitrogen. together with argon, and traces of the other inert gases,
neon, helium, krypton and xenon, has no physiological effects under normal
conditions. At the high pressures encountered by deep-sea divers an
increased proportion of nitrogen is absorbed into the blood. Too rapid a
return to normal pressure causes the painful and dangerous symptoms known
as "the bends", in which the excess nitrogen reappears as bubbles in the
veins and arteries.
1.1.4 Carbon dioxide also has no significant effect on man. Because we

produce it, together with water vapour, in the air we exhale, the concen-
tration rises in crowded rooms or buildings. For this reason it was at one
The air and human well-being 3
time used as an index of the effectiveness of ventilation with limits of a half to one
per cent, but this practice has died out.
1.1.5 Impurities such as methane or ozone occurring in natural fresh air at

levels of less than 1 ppm (parts per million) may be ignored. Industrial processes,
however, and the circumstances of town life, produce a wide variety of polluting
substances which must be dealt with by ventilation.
For toxic gases and their dilution see Section 5.1.
For the extraction of dust and fumes at source see Section 5.4. For the
cleansing of air from smoke and dirt see Section 5.3.
For precautions against fire and explosion hazards see Section 5.2. For fans
to resist corrosion or abrasion see Section 7.14.
1.1.6 Odours may result from the presence of extremely small proportions of
impurities in the air. The staleness perceptible in the air of crowded rooms is due
chiefly to organic substances given off from the occupants' bodies. While these
are not dangerous, and some people may even like "a good fug'", others are more
sensitive. Loss of appetite is a common reaction and some may even experience
nausea. The ventilation requirements to deal with body odour are dealt with in
Section 2.1.2. They are not great and are usually exceeded substantially by other
requirements of the system.
1.1.7 Bacteria, often associated with dust particles or water droplets, may
remain in suspension in the air far long periods. In hospitals a positive programme
of control is necessary. In ordinary life the spread of infection by coughing and
sneezing is a direct function of the number of people congregated together in a
limited space. There is positive evidence that good ventilation reduces this hazard.
1.2 Body Heat Balance

Why does anyone feel limp in hot surroundings? Why is a breeze refreshing 7
These sensations relate very much to the efforts the body has to make to
maintain its temperature. The less the effort the more comfortable we feel. Man is
a warm blooded animal and must maintain the temperature of his vital organs
within a few degrees of 37°C throughout life.
The human body, fuelled by the food we eat, continuously produces heat,
associated with chemical change and muscular activity. This metabolic rate of
heat generation can exceed 1 kW with maximum exertion. Table 1.2 gives an idea
of the range; individuals will vary with age, weight and other personal
characteristics.
The body has notmuch heat storage capacity because we cannot allow the
greater part of our substance to get significantly hotter or cooler without
distress. Apart from minor temporary deviations we must dissipate heat just
as fast as it is generated - i.e. at the metabolic rate. There are three
processes by which the body loses heat: convection; radiation; evaporation.
1.3 Convection and Radiation

A layer of cool air in contact with warm skin or clothing will pick up heat,
and as its temperature rises its density will fall. The lighter air now rises up
away from the body, taking heat with it, and is replaced by fresh. cool air
which continues the process. This is natural convection.
Even the slightest air movement around the body will increase the rate at
which warm air is replaced by cool, and thus increase the heat loss. The laws
governing this heat exchange by forced convection are dealt with in Chapter
12. If the air temperature is on the high side the extra movement will be felt
as a pleasant breeze, and can be increased with advantage by the use of
ceiling fans or circulating fans. If, on the other hand, the temperature is
normal or low, movement will be felt as a draught and should be kept below
the limit of perception-about 0.25m/s-for maximum comfort.
Like all matter, the body transmits heat by radiation and receives heat by
the same path. If all the surfaces surrounding us were at the same
temperature as our skin there would be no nett gain or loss of heat by
radiation. In practice, of course, the walls and most of the surfaces are cooler
and radiation carries part of the necessary heat loss. Radiant heaters have
high temperatures over small areas, contributing a nett heat gain to the body.
As radiant heat, like light, travels in straight lines, only the "illuminated" side
of the body will be heated, but, in the absence of excessive draught, the
blood stream will distribute the heat satisfactorily over the body.
Within the comfort zone of external air conditions, convection and radiation
losses account for about 75% of the metabolic heat at rest, or with mild
activity. When the temperature rises above or falls below the comfort zone an
automatic reaction known as vasomotor regulation comes into play, with the
object of preserving both internal temperature and heat loss unchanged. The
tissue layers under the skin contain a network of veins which can be enlarged
or contracted under the control of the nervous system. In a warm
environment a copious flow of blood is allowed through these vessels,
bringing the skin temperature up towards a maximum of about 35 °C, so as to
maintain its temperature excess over the surroundings. In a cold environment
the veins are constricted, reducing the thermal conductivity of the tissue layer
so that the skin cools. The blood flow along legs and arms is also modified,
allowing further cooling along their length to keep down the heat loss.
Ultimately, hands and feet may be only a few degrees above the air
temperature even to the extent of allowing frost-bite of fingers and toes
rather than loss of temperature at a vital centre.
1.4 Evaporation
At temperatures above about 30°C the system just described is unable to
secure the necessary heat loss, even at rest. Indeed, above 35°C convection
and radiation cease to be losses and become heat gains making the body's
task even harder. Reaction to this situation is known as evaporative
regulation. It consists in the automatic activation of the sweat glands over an
area of skin proportional to the corrective effort required.
The transformation of sweat into water vapour absorbs energy, just as the
boiling of water does. This energy is taken from the wetted skin surface in a
form known as the latent heat of evaporation (see Chapter 12). The cooling
effect is powerful : the worker in a hot industrial environment may easily
produce and evaporate one kilogram of sweat each hour, which will remove
heat at the rate of 680 watts. The water and salt lost by the body have to be
replaced, and it is natural that such workers should be copious drinkers.
Apart from the sweat mechanism, evaporation of the water vapour in the
exhaled breath carries away heat at a minimum rate of around 20 watts: the
proportions at rest are illustrated in Fig. 1.1. Note that the metabolic rate
cannot be reduced, though we may increase it if we are cold by voluntary
activity-walking briskly, stamping the feet, swinging the arms, etc. Shivering
is an automatic reaction with the same purpose.
6 Woods Practical Guide to fan Engineering
In a still atmosphere the air next to the skin and trapped in the clothing
becomes almost saturated, and its capacity to absorb and carry away
moisture is severely limited. The sweat produced stays wet on the skin and
the body's effort to give off heat is retarded. The surplus sweat which drips
off or is wiped away is virtually useless for heat removal.
Currents of air flowing round the body correct this situation. Saturated air is
replaced by fresh and evaporation is helped in exactly the same manner as
heat removal is helped by forced convection. To understand the limitations of
evaporative regulation, and the value of ventilation and air motion, we must
study the subject of air humidity.
1.5 Humidity
The moisture in humid air takes the form of an invisible vapour, mixed with
the other gaseous constituents. If mist or steam can be seen it means that
more moisture is present than the air can hold; the excess has condensed in
the form of liquid water droplets.
Air that carries the maximum possible amount of water in the vapour form
is said to be saturated. The temperature at which air becomes saturated rises
with its moisture content and is known as the dew point. Thus air saturated at
14°C carries about 1% of water vapour by weight. But at 25 °C it could
contain 2%, so that 1 % moisture content is now only half the saturation
value. It will be sufficient, for our present purpose, to say that its relative
humidity is 50%, saturation value at the same temperature being taken as
100°.
Strictly speaking the relative humidity is the ratio of the number of
molecules of water vapour present to the number in the same volume of
saturated air at the same temperature and pressure. The comparison is based
on equal volumes of moist air rather than equal weights of dry air, and the
values are slightly higher except at 0% and 100%. Relative humidity is also
equal to the ratio of the vapour pressure of the water present to the vapour
pressure at saturation.
1.5 Measurement of Humidity

The true temperature of air, dry or humid, is that measured with an
ordinary mercury-in-glass thermometer and is called the "dry-bulb tem-
perature". Precautions should be taken to shield the bulb from strong radiant
sources. If the bulb of a second thermometer is kept wetted, its temperature
will fall by evaporative cooling. Provided the air is kept moving past it at a
velocity usually specified as 2m/s minimum, it will reach, and record, a stable
temperature known as the wet-bulb temperature. This is intermediate
between the dry-bulb temperature and the dewpoint, and is arrived at by a
process of heat balance described in Chapter 12.
The meterological type of wet- and dry-bulb instrument is designed to
make use of the air currents practically always present out-of-doors, and is
not suitable for use in rooms. The sling psychrometer is simple and reliable
for ventilating work. As illustrated in Fig. 1.2 the two thermometers are
mounted side-by-side in a frame which can be whirled round
the handle, providing 3 to 4m/s air movement over the bulbs. The cloth
covering the wet bulb should be clean and freshly soaked in distilled water.
The other bulb must be quite clean and dry.
Air conditions for the human environment are generally described in terms
of the dry-bulb temperature (°C DB) and the relative humidity (RH). There is
no simple formula relating RH to °C DB and °C WB (wet-bulb temperature).
Tabulated values may be used (see Table 14.8) but more information can be
obtained from a Psychrometric Chart. Fig. 1.3. is a simple example suitable
for the present discussion, giving, in addition
to the RH, the dew-point and the moisture content. Many more complicated
psychrometric charts have been drawn, with refinements of accuracy and giving
additional information for the engineer concerned with air conditioning or process
work. See for example Figs. 4.4 and 12.13.
1.7 Comfort Conditions

During the heating season good heating and ventilating systems will be
provided with thermostatic control, both for economy and for comfort Small rooms
and offices should be individually controlled to meet variations in preferred
temperature, which may range from 18°C to 28°C. Spaces accommodating many
people must of course have a compromise setting which is likely to leave at least
10% of the occupants slightly dissatisfied-some too warm, some too cool.
1.7.1 Air movement and humidity

Well distributed air movement in the range 0.1 to 0.25m/s is important to
create a sense of freshness in the air. As well as eliminating hot spots and cold
spots, it limits the temperature gradient from foot level to head
level; this should not exceed 3°C. Draughts must be avoided, and research
has indicated the range which is free from unpleasant sensation as a function
of temperature. The suggested velocity limitation is:
Humidity is unimportant in the comfort zone provided it is in the range 30% to

70% RH. A rise to 90% would call for a reduction of about 1 °C in the dry-bulb
temperature to maintain comfort at rest-more if active. Humidities of 10 to 20%
RH give rise to uncomfortable dryness in the nose and throat, and are likely to
cause cracking in furniture and other timber seasoned at higher humidities, and
electrostatic sparking.
1.7.2 Optimum temperatures

There is no international consensus on the optimum comfort temperature.
Longer experience of central heating, coupled with a tendency to wear light
warm-weather clothing all the year round, probably accounts for the higher
temperatures preferred in North America. The following is an interpretation of
national practice:
The last condition will not be met in poorly insulated buildings with
convective heating. In such cases cold walls will have to be offset by higher
air temperatures. Conversely, when the heating is mainly radiant a lower air
temperature will give maximum comfort. This problem is dealt with in Section
3.3.
The optimum temperatures quoted will not be detrimental to any activity
provided some clothing is discarded when working hard. However, with
moderate and heavy work loads lower temperatures can be tolerated; for
example the British Factory Acts allow 15 5°C minimum where factory work is
done seated. Outdoor work can, of course, be done in freezing, or indeed
Arctic, conditions given the necessary clothing, fitness, activity and
acclimatisation, but rest periods in warm surroundings should be available.
1.8 Heat Stress

The scale introduced by Houghton and Yaglou in 1923 is still regarded as
the best guide to the severity of exposure to high temperature and humidity,
and the cooling effect of air movement. In these respects it is a measure of
the effort required by the evaporative regulation system of the body. The
scale is based on jury judgements of equality of warmth when passing from
enclosures at various humidity levels to a standard enclosure at 100% RH.
The temperature of the latter is termed the effective temperature. The
diagram in Fig. 1.4 is a conversion to metric units of the upper part of the
definitive chart published by ASHRAE. The chart applicable to men stripped to
the waist and doing light physical work has been chosen. Instructions are
given with the chart for determining the effective temperature in °C over a
range of dry-bulb and wet-bulb temperatures and of air velocities.
The following table of maximum recommended exposures for industrial
workers is derived from the publications of the American Conference of
Governmental Industrial Hygenists. It is in terms of effective temperature.
The Tower limit for cold weather is intended to guard against chill when
leaving work which has caused heavy sweating. It will be seen from Fig. 1.4
that a high air velocity can lower the effective temperature by 5 °C or more,
particularly if the air is humid. At the same time the velocity should be limited
if the exposure is long term. Some recommended velocities from the source
just referred to for use in high effective temperatures are:
The stress refers to the combination of heat and work loading. Intermittent
exposure may be at the site of exceptional heat stress, or in relief areas set
aside for recovery after severe exposure.
1.9 Radiant Heat

1.9.1
An effective temperature determined from readings of wet- and dry-bulb
thermometers and air velocity alone is appropriate only to environments
where radiant heat is not a significant factor. In some industries,
e.g. in steel works, it is precisely the radiant sources which cause the heat
stress.
The prediction and calculation of the heat loads from radiant sources is often
a complex and somewhat unrewarding exercise. The ventilation engineer is
more likely to be called in when certain operations in existing installations are
the subject of complaint. Effective action may include shielding of heat sources,
enclosure and ventilation of sensitive spots such as control desks and relief
areas, or even (for very short exposure to intense heat) the use of protective
clothing with an aluminised surface to minimise radiation absorption.
1.9.2 Globe temperature

If, however, improvement by general or local ventilation is contemplated,
measurement becomes important. The usual instrument for assessing radiant
heat is the globe thermometer. This is a copper sphere, 15cm in diameter and
painted matt black inside and out, with a thermometer to measure the internal
temperature. This, after 20 minutes or so to reach equilibrium, is the globe
temperature ( °C GT) and is such that the radiant heat gain equals the
convection heat loss, including the effect of air velocity but disregarding
evaporative effects.*
A number of investigators have proposed ways in which the effective
temperature scale may be modified to take account of radiant heat using the
globe temperature. These do not agree very closely so it has been felt
permissible to introduce a very simple modification to the Effective
Temperature Nomogram, Fig. 1.4. This produces results within 1 °C or so of
the method suggested by Bedford, within the limited chart area drawn.
We can now assess the severity of heat stress by comparing the measured
effective temperature with the recommendations in Section 1.8. Furthermore,
we can estimate the effectiveness of various factors in bringing the ET down,
which will form the subject matter of Chapter 2.
1.9.3 Example 1
Suppose in a particular workshop the measured dry-bulb temperature is
34°C and the measured wet-bulb temperature 25°C. Fig. 1.3 shows that the
relative humidity is 45% and the moisture content 0.016. On Fig. 1.4 the
location of the straight edge is already drawn as line AB, and shows that the
effective temperature is 28°C with natural air movement only (0.15m/s).
This is on the high side for heavy work and could be reduced to 26 °C with
general air movement at about 1m/s or to 24°C with a positive cooling stream
at 4m/s. Alternatively, if the moisture content could be reduced to 1.0%, say
by local extraction of escaping steam, with an accompanying reduction to 32 °C
dry-bulb, Fig. 1.3 shows that we have
reduced the RH to 33% and the wet-bulb to just over 20'C. This gives an
effective temperature of 24°C without considering the enhanced air
circulation.
Example 2
Suppose the same air conditions exist in a foundry, where the workers
cannot readily be shielded from radiant heat. A globe temperature of
52°C is measured with natural air movement of about 1m/s caused by
convection currents only. Fig. 1.4 shows an effective temperature of 34
°C in which only short periods of heavy work are possible.
At first sight it appears that air velocity has negligible effect. However,
it must be remembered that the globe thermometer should be exposed to
the same conditions as the worker. An experiment is tried with the globe
in the work place exposed to a spot-cooling air stream at 4m/s and the
globe temperature falls to 42°C. This gives an effective temperature of
29°C-target value for the work pattern in question.
CHAPTER 2
General ventilation
Natural ventilation, with open windows in summer, may

suffice for the living rooms and bedrooms in our homes,
where there is plenty of space per person and no generation
of steam or cooking fumes. It is, however, unpredictable,
and will fail altogether in unfavourable conditions of wind
and weather. In many areas within a building, therefore,
mechanical ventilation, powered by fans, is a practical (and
often a legal) requirement.
The rate of ventilation, conveniently measured in litres of
air per second must be sufficient to satisfy the following
three requirements:
(a) Sufficient air movement throughout the room or building
to prevent the formation of pockets of stale, stagnant
air.
(b) Sufficient fresh air supply and foul air exhaust to limit
the level of air pollution from all sources in the building,
including humidity.
(c) Reduction of air temperature, within the limits set by the
climate, by the removal of heat generated within the
building or supplied by the sun.
2.1 Ventilation Requirements

2.1.1 Air change ratings
The simplest method of determining the ventilation rate required
is to make use of the accumulated experience of the industry
expressed in a
14
General ventilation 15
table of "air change rates". The volume in cubic metres (m3) of

the space to be ventilated is estimated and multipled by the
number of air changes per hour to give the ventilation rate in m3
per hour. Division by 3.6 converts this to litres per second.
Table 2.1 gives a guide to the number of air changes generally

recommended. Provided air inlets and outlets are properly sited, the
tabulated values will be sufficient for requirement (a), air movement.
They will also meet requirement (b), for pollution and moisture
removal, in typical examples of the accommodation indicated, and
requirement (c), heat removal, in temperate climates without heat
sources.
However, particular installations should always be checked against
the following questions:
Are there specific bye-laws or other legal requirements
applicable in the locality?
Is the space expected to become so crowded that a higher
ventilation rate might be obtained by calculating on a per person
basis?
Are there abnormal sources of heat or fumes?
Is it desirable to provide a higher ventilation rate for summer
cooling?
2.1.2 Ventilation rates per person

As explained in Chapter 1, each human being produces his personal
quota of heat, water vapour, carbon dioxide, and body odour. The
ventilation requirement for the last item will cover the needs of the
other three during the heating season. It is not sufficient merely to
ensure an acceptable average freshness of atmosphere throughout
the room; if people are crowded together by shortage of space the
average odour concentration must be reduced. This means that the
fresh air supply per person must increase with the number of people
in a room, and the air change rate must increase faster still.
Genera/ ventilation 17
These rates should be increased by a third if smoking is permitted

or if there are doubts about standards of personal hygiene. If
minimum legal requirements are specified, they are generally about
a third less. In contrast, as little as a 1 litre/s will suffice to meet
the strictly physiological needs of one person for oxygen supply and
carbon dioxide limitation.
2.1.3 Recirculation
The flow rates just quoted on a per person basis must be supplied
as fresh outside air. So must the air extracted to remove pollution,
moisture or heat. However, the flow rate required to achieve the air
change recommended in Table 2.1 will usually be considerably
greater. The excess can be recirculated through ducts from that
part of the building where the bulk of the air is exhausted, to the
opposite side. This means that the same air may pass several times
through the various rooms in the building, providing the necessary
air circulation to prevent stagnation, and reducing the heating or
cooling load.
Fig. 2.1 is a schematic example of a ducted system showing the
winter and summer arrangements. Fig. 2.2 shows a unit system
suitable for single storey factories, with winter recirculation of the
warm air collected at roof level.
There would be no value in recirculation if the outside air were

clean enough and warm enough to pass, untreated, through the
building. However, in winter the whole of the fresh air supplied has
to be heated from the outside temperature to the internal comfort
temperature. Furthermore, if the outside air is dirty and requires
cleaning, the life of the filter elements will be lengthened as the
volume of air they handle is reduced. Economy is, therefore,
served by minimising the fresh air flow and maximising the
recirculation during the heating season. On the days it is better to
operate with 100% fresh air, thereby maximising the rate at which
the heat generated by the occupants and other sources can be
disposed of.
2.1.4 Ventilation for air conditioning

Full air conditioning systems invariably embody recirculation. The
cost of cooling plant is high and should not be wasted on the
cooling of more than the necessary minimum volume of hot outside
air. British practice for such installations is indicated by Table 2.3
taken from the IHVE Guide, 1970. The outdoor air supply rates
given take account of the likely density of occupation and the type
and amount of smoking. Of the minimum rates quoted per person
and per mz, the greater should be taken.
2.1.5 Example
Ventilation is required for an open office with dimensions 12m x
30m x 3m high. The number of occupants is assumed to be 90,
this being the maximum permitted under U.K. regulations.
We will first select six air changes per hour from the
recommendations in Table 2.1, bearing in mind that the
occupancy is high. Then the ventilation rate equals:
Since the air space per person is 1080/90 = 12M3, Table 2.2
recommends a fresh air supply of 6 litre/s per person.
Supposing smoking to be permitted, we will increase this by a
third to 8 litre/s, giving a total fresh air supply rate of 8 x 90 =
720 litre/s.
While the building is not to be air conditioned, we will note

that Table 2.3 would give:
We may thus adopt a recirculating system,with a wintertime

budget of:
Adjoining lavatory accommodation is planned with six urinals and nine

WC's. At the recommended rate of 24 litre/s per point this will require an
extraction rate of 15 x 24 = 360 litre/s.
To match the 720 litre/s supply rate, another 360 litre/s extraction will
be required. This can conveniently be passed through a service area
which might be used for printing, or for a small kitchen, etc. This is the
system shown schematically in Fig. 2.1 and as a plan view in Fig. 3.3.
2.2 Sources of Heat

2.2.1 Ventilation rates based on heat gain
The third function of a ventilation system is to mitigate discomfort in
hot weather. We are not now considering the positive cooling provided by
air conditioning, which is something of a luxury in temperate climates.
Here we shall discuss the removal of heat by generous rates of fresh air
ventilation which may be well in excess of those recommended in Table
2.1. The first step in this study is to determine the total heat generated
within or transmitted into the enclosure to be ventilated.
2.2.2 Heat from people

The metabolic heat given off by the occupants was discussed in
Chapter 1. Average values sufficient for the present purpose may be
taken as:
2.2.4 Heat from lighting

Heat from lighting apparatus is simply the aggregate rating of the
tubes and bulbs installed, discounting those that will not be in use when
solar heat gain is a major factor. Sometimes part of the ventilation air is
extracted through special recessed lighting fittings. The convected heat
from the lights will then be extracted with the air, and will not form part
of the room heat load.
The radiant heat load remaining will be about 80% of the rating
for tungsten lamps, 60% for fluorescent.
2.2.5 Heat from electric power

With very few exceptions the whole of the energy supplied to
electrical apparatus will be dissipated as heat within the building.
The maximum will be the full load rating of the appliance in kW,
although allowance can be made for the dispersion factor when
known. This factor is the ratio of the kilowatt-hours consumed in
one hour to the kilowatt rating and takes account of both part-time
and part-load working. Sometimes this factor can be obtained
indirectly; for example, a machine shop may have statistics of the
number of kWh of electricity consumed per shift - or, better, the
maximum half-hour kW demand.
2.2.6 Heat from electric motors

In the case of an electric motor the usual rating is the output in
hp or kW, whereas what we require is the electrical input,
including the motor losses. Typical values are:
2.2.7 Heat from fuel

The combustion of fuel occurs in some part of the
manufacturing process in most industrial plants. In many cases
it is a major factor, and general ventilation is necessary for the
relief of heat stress, quite apart from local ventilation and other
measures which are taken at the sources of heat.
The potential heat load is easily obtained from the calorific
value of the fuel and the rate at which it is used:
However, a large part of this heat will not contribute to the

environmental load. The deductions can be classified under three
headings:
(a) Incomplete combustion-unburnt or partially burnt fuel passing

up the flue.
(b) Flue heat loss-the heat carried up the flue by hot gases
comprising the products of combustion and the excess air.
(c) Heat to stock-occurring when the material worked on (solid,

liquid or gaseous) is heated and then passes, still hot, to other
parts of the building.
In important cases estimates for each of these items should be

available as part of the manufacturing records or plans. More
usually, perhaps, the margins of error in the direct calculation of the
residual heat loss remaining within the building are too great.. In
such cases measurement of the actual air conditions on site, as
described in Chapter 1, will be more helpful.
2.3 Solar Heat
2.3.1 Heat from the sun

Solar heat reaches the earth at a fairly constant rate of 1.36
kW/m2. In passing through the atmosphere some of this heat will
be absorbed or scattered into space. The proportion lost will be
determined by the length of the path through the atmosphere and
by the presence of clouds, dust, water vapour and ozone. Fig. 2.3
gives representative values for a clear sky and various heights
above sea level.
The direct intensity, Id varies with the altitude angle of the sun
above the horizon. The effective intensity on any surface is further
reduced by the inclination of the surface to the sun. In addition,
diffuse radiation, If reaches earth's surface after scattering by
atmospheric dust and vertical walls receive reflected radiation from
the ground.
This simple statement gives complete information for calculating

the solar heat reaching any exposed surface on a sunny day.
However, the work involved in allowing for the sun's movements
during the day, and over the year, is time consuming. Such
allowances are necessary in estimating the refrigerating plant
capacity and energy consumption for a full air conditioning system,
and extensive charts and tables to assist in the task will be found in
works of reference.
2.3.2 Maximum solar heat intensity

However, our present aim is merely to estimate the ventilation
rate required to limit the internal temperature in hot weather. For
this purpose we need consider only that time of day and month of
the year which will give the maximum solar heat load. This has
been done in drawing up Tables 2.4 and 2.5, which can be used to
estimate the maximum solar intensity on walls and roofs.
24 Woods Practical Guide to Fan
Generally speaking the maximum intensities on various parts

of the building structure occur at different times of day, and the
corresponding heat loads cannot be simply added together. The
following procedure is suggested as a basis for calculating the
heat load to be removed by ventilation.
(a) Select the wall which may be expected to provide the
greatest solar heat load, and estimate the maximum
intensity on it from Table 2.4 for the direction in which it
faces and the latitude.
(b) Assume that the "out of sun" intensity of Table 2.4 applies
to the remaining walls at right angles and on the other side
of the building.
(c) Assume that the "flat roof" intensity of Table 2.5 applies to
the whole plan area of the building as if the roof were flat,
even if it is actually pitched. Exclude glazed areas and any
sections shaded during the heat of the day.
(d) Take the intensity applied to the actual area of skylights,
north lights, etc., from Table 2.5 according to their
inclination to the horizontal towards or away from the sun
when it faces wall(a).
It should be noted that if the terrain is very bare and the air very
dry, increased ground reflection and atmospheric clarity may increase
tropical solar intensities by 20%. Conversely, Industrial atmospheres
may reduce intensities by 20% to 40% even on the sunniest days.
Altitude corrections can be derived from Fig. 2.3, but for moderate
heights an addition of 1% per 100m above sea level will suffice. The
figures in the tables are for the time of year and day giving the
highest intensity for each entry. The 24-hour average midsummer
intensity averaged over walls equally exposed to all points of the
compass is remarkably constant at about 0.08 kW/m2, with about 0.35
kW/m2 on roofs. The tables may be used for the southern hemisphere
by interchanging N and S.
2.3.3 Calculation of solar heat gain

Having found the intensities of solar radiation falling on the faces of
a building, the next step is to determine the total solar heat gain
entering through windows, walls and roofs. This is, of course,
proportional to the area of each surface, but is reduced by a factor
taking account of the heat excluded by reflection and external
convection. Values of this factor S for typical building features are
given in Table 2.6.
If A is the gross area of the window, wall or roof section under
consideration in square metres, and S the appropriate transmission
factor, the corresponding rate at which the sun's heat enters the
building will be for any surface:
where Î is the effective maximum solar intensity obtained from Tables

2.4 and 2.5 in accordance with the procedure specified in Section
2.3.2 (a), (b), (c) and (d).
Account must, of course, be taken of those times when the windows
are wholly shaded by neighbouring buildings and trees, or partly
shaded by architectural features such as deep eaves or mullions.
External sun blinds are very effective provided they are used, but
internal blinds or curtains much less so. They themselves become
heated, and re-radiate heat to the room.
2.4 Ventilation Rate and Temperature Rise
2.4.1 Heat capacity of air

For the detailed study of the thermal history of a building we would
need to consider the daily cycle of solar and other heat gains. Then we
would have to examine the thermal capacity of the various parts of
the building structure and the rate at which heat is dissipated as a
result of internal temperature rise. For our present purpose, however,
we will ignore the contribution to temperature limitation which the
building makes, supposing it to be of light construction so that
ventilation must dispose of the whole heat gain.
The specific heat of air at constant pressure and normal
temperature is about 1.0 kJ/kg (kilojoules per kilogram) *. It follows
that a heat gain
*See also Table 14.9.

of 1 kW or 1 kJ/s will raise the temperature of air flowing at the

rate of 1 kg/s by 1 °C. Therefore:
Air will enter the building at the outside air temperature. As it

flows through the building its temperature will rise until, at the
point of extraction, it has reached the value given by this
formula. If there is a major heat source in the building it is
clearly wise to extract air close to the source, preferably so that
nobody need work in the space between. We can then discount
the major source in considering the temperature rise of the
general ventilation system.
Since large volumes of air can be moved quite cheaply
through industrial buildings, general ventilation is a potent
means of maintaining reasonable comfort during spells of hot
weather.
2.4.2 Example
A single-storey machine shop has a plan area of 40m x 12m
and a height of 4m to the eaves and 6m to a central ridge. The
40m wall faces 40° west of south and has five windows. A roof
light on the NE side spans 32m and is 1 m wide. Rural situation
in UK at 52° north latitude.
28 Woods Practical Guide to Pan Engineering
If we set a temperature rise target of 5°C we have, for the

required ventilation rate:
This will be met by an installation of five 800mm propeller fans at

4800 litre/s each, 24m3/s total.
This is the temperature rise of the air when finally extracted. The
occupied zone may be expected to be about 3°C above the outside
temperature.
CHAPTER 3
Air distribution and
circulation
The main topic in this chapter will be the siting of air inlets
and outlets and their design in relation to air movement in
the ventilated spaces.
3.1 Distribution System

3.1.1 Extraction systems
These are of greatest significance for industrial buildings. The first
requirement will be the local extraction of fumes or other
contaminants from localised sources, using hoods and ducts as
necessary; this is dealt with in Chapter 4. Usually, however, a much
larger volume flow of air is needed to meet the general air change
or summer time heat removal requirements.
We may take as an example the 40m x 12m machine shop of
Section 2.4.2. For this we specified 24m3/s of general ventilation,
equalling 36 air changes per hour. Fig. 3.1 c shows how this could
be arranged, fitting five 800mm propeller type exhaust fans in one
long wall, with open windows providing air inlets on the opposite
side.
The size of the inlets is important. Obviously, if they are too small
they will restrict the air flow, and full benefit will not be obtained
from the fans fitted. To match a propeller fan 1 ½ to 2 times the fan
area is advisable. The fan area in the present case is 5 x ∏ (0.4)2 =
2.5m2. If each of the five 4m2 windows is allowed 1 m2 opening this
will be ample.
29
Air distribution and circulation 31
Since the inlet velocity at 24 (m3/s)/5 (m2) = 4.8 (m/s) is quite

high it will be best to allow the opening at the top of the windows,
above head level. Air will then sweep across the roof space-where
the hottest air collects-and out through the fans while gentler, but
still generous, air movement will be induced in the working space
below.
For successful ventilation the exhaust points and natural inlets

must be so arranged that fresh air is drawn across all parts of the
space. This is secured in Fig. 3.1 c, but we will now suppose that
the wall on which the fans were mounted is against another building
and not available for extraction. A single gable-end fan as in Fig.
3.1 a would be a bad arrangement, even though it extracted the full
24M3/s. There would be no air movement at the far end of the
shop, and even the fitting of a fan at each end as Fig. 3.1 b would
leave the centre inadequately served.
The difficulty might be met by ducted extraction, preferably with

a fan at each end, which halves the volume to be carried by the
duct and the distance the air must travel. As indicated in Fig. 3.2b
adjustable registers in the duct wall secure uniformly distributed
extraction along the length of the room. However, when a roof is
available, the fan-powered roof units located as in Fig. 3.2a are
likely to provide the most economical solution, in both first cost and
running cost.
3.1.2 Supply and exhaust systems

Air inlets round the walls are insufficient for large industrial
buildings. They would be too far from the extract points and the air
would become excessively heated or contaminated en route. Roof
mounted, fan powered, supply units and extract units provide a
convenient arrangement for single storey buildings. They may be
spaced alternately every 10 or 15 metres in such a way as to
ensure cross flow over the whole area. The supply units should be
ducted down to a level at which all round radial outlets will
distribute the air above head level. Sometimes, however, the need
to clear travelling cranes necessitates high level inlets, which must
then project a jet of air downwards towards floor level. The general
extraction units should take the air from as high a level as possible,
where it will be hottest. Large multi-storey buildings will, of course,
require supply and extraction fans and ducts with air outlets and
inlets distributed along them.
So far we have considered only high-volume general ventilation

systems for summer time cooling. In winter general ventilation will
still be required, at a lower rate. Where there are fan-powered
supply units, they will be fitted with heater batteries. These may
merely "temper" the incoming air, raising its temperature to the
internal comfort level only, the building fabric losses being supplied
by a separate heating system. More usually, perhaps, in an
industrial installation, the air supply will carry the whole of the
heating load. For this purpose it must enter at a temperature
considerably above the comfort level, which complicates the
distribution aspect, as will be discussed later.
3.1.3 Heating with air circulation

The winter time examination of our machine shop example, which
will be found in Section 4.3 shows that Jm3/s will amply supply the
requirements of fresh air ventilation. This is much too little to carry
the heating load, neither could it set up sufficient air movement for
comfort. Fan-powered unit heaters provide an answer at low cost,
and may be mounted as in Fig. 3.2c to maintain a relatively high
velocity in the upper levels, with induced air circulation below.
Better control of air movement, of comfort temperature, and of

heat supply can be maintained with ducted supply and extraction
systems. They lend themselves to winter time recirculation with
minimum fresh
air supply and are commonly arranged with an excess of supply

over extraction air, so that the surplus will escape outwards through
the various leakage paths round windows, doors, etc. Fig. 3.3.
takes as an example of such a scheme the 30m x 12m x 3m office
of Section 2.1.5.
Parallel supply and extraction ducts establish cross-flow rather

than the less desirable longitudinal flow as the main pattern of air
movement while 0-36M3/s of positive extraction is established
through exhaust fans in the lavatories. The remaining 0-36M3/s of
fresh air is allowed to escape through grilles appropriately placed in
the service area and by leakage around the office windows. The
main damper is set when the system is balanced on installation, so
as to allow 1.08m3/s through the return duct and 0.72m3/s fresh air
supply.
In summer the damper is closed against recirculation and the

whole supply allowed to escape, either through the outlet of the
extract duct or through open windows.
3.2 Jets
3.2.1 Jet controlled circulation
While comparatively crude provisions suffice for high-volume
general ventilation systems in industrial buildings, greater
sophistication is needed for optimum comfort in a leisure
assembly area. The points of air supply are critical, both in their
location and in the manner in which the air is introduced.
The air will generally be supplied in the form of a jet in order
to maintain control over a relatively large floor area. This jet will
be of relatively high velocity and must be kept out of the
"occupied zone"-namely the accessible parts of the room not
more than 1.8 metres above floor level. As the jet advances
across the room it slows down and at the same time expands,
dragging into motion or "entraining" an ever-increasing volume
of air from the room. When the peak velocity has fallen below
0.5m/s the jet breaks up into random motion and ceases to
advance. It is mainly the entrained air, circulating back through
the occupied zone to rejoin the jet nearer the supply outlet,
which provides the controlled air movement, ideally in the range
of 0.1 to 0.25m/s, which is required. In industrial situations
velocities up to 0.4m/s would be applicable.
3.2.2 Velocity profiles

The whole pattern of the jet is determined by the jet velocity
and area at the supply outlet. The following are the basic
equations which apply when the entrained flow pattern is fully
established:
These values of K are for Vo = 4m/s and may be 10% lower at

low Vo, 10% higher at high Vo.
Ao is the throat area of a nozzle outlet, the free area of a grille,

the gross area of a duct outlet, or 60% of the area of a sharp edged
orifice outlet.
In the absence of obstructions the velocity profile across the jet is

approximately as Fig. 3.4, y being the radial distance from
thecentreline. A jet discharged close to a wall or ceiling will cling to
the surface, with peak velocity increased by 40% as shown.
Effectively this is one half of
a jet from an outlet with double the area, the entrained volume
flow being the same as it would be from the same outlet into open
space.
The throw of a jet is defined as the distance Xt at which the peak
velocity Vx has fallen to 0.5m/s. It is good practice to set the throw
at about three-quarters of the distance to the wall opposite the
outlet. Fig. 3.5 illustrates the type of air distribution aimed at. It
also shows, for the case of a cooled air conditioning jet, the result
of inadequate throw (draught-making velocities fall into the
occupied zone) and of excessive throw (cold draughts down the far
wall).
Not only the throw, Xt but also the supply air quantity, Qo, must
be right for satisfactory design, and the selection of Qo. is dealt with
in the next section. Here we may note that the basic equation can
be recast to give the required outlet area in terms of Qo, Xt and Vx.
(= 0.5m/s).
3.2.3 Example
A room 12m x 8m x 3m high is to be air conditioned with outlet
slots near ceiling level along the 12m wall. Four air changes are
required.
A throw of ¾ x 8 = 6m is required, at which point the jet spread to

half velocity (see Fig. 3.4) is about 0.2 x 6 = 1.2m horizontally, or
rather more with the slit shape and proximity of the ceiling. It will
be reasonable to divide the room into four sections, each 3m x 8m x
3m and each supplied with 0.08m3/s through a single outlet.
Inserting these values in equation 3.3
Four outlet registers, each 60mm x 600mm, with an effective area

60% of the gross area, would meet this requirement, with an outlet
velocity Vo = 0.08 (m3/s)/0.021 (m2) = 3.7m/s.
Proprietary outlet registers are available for which performance
figures are quoted directly. Preset or adjustable vanes may befitted
which cause the jet to diverge and reduce the throw. Ceiling
diffusers are an example, and are needed to avoid draught at head
level.
3.2.4 Effects of heated and cooled jets

If we examine the results of the last section we shall find that we
can provide the required throw and set in motion the required
volume of entrained air, either with a small quantity of supply air at
high outlet velocity, or with more air at a lower velocity. In fact, the
product of the two (ρQoVo, the momentum flux of the jet) is
constant for a given result.
However, in an air conditioning system, we shall probably want to
use the jet to supply heat in the winter and remove heat (by
supplying chilled air) in summer. A heated jet will tend to rise and a
cooled jet to fall, since
they have respectively lower and higher densities than the air in
the room. As entrained air is mixed with the jet the average
temperature difference from the room air will fall-substantially in
proportion to the fall in peak velocity. However, the jet will be
excessively deflected up or down if we try to load it with too
much heating or cooling duty.
If a single jet is designed to provide satisfactory throw down
the length, L m, of a room which is H m high by B m wide, then
it should still be satisfactory when carrying a heating or cooling
load of W watts, provided the supply volume flow is not less
than*
Strictly speaking, the limiting value is dependent on the shape

of the outlet and its location relative to the room. Recent work
by Jackman at the British Heating and Ventilating Research
Association (now BSIRA) gives a more general treatment.
3.2.5 Example
Substituting the values Qo = 0.08M3/s, B = 3m, H = 3m from
example 3.2.3. in the last equation, we find that the maximum
heating or cooling load, W, is 740 W per outlet or 3 kW for the
whole space. If the requirement were to be, say, 6 kW, then Qo
must be increased, and Vo decreased, in the ratio 3√6/3 = 1.26.
The revised values become:
3.3 Circulating Fans
3.3.1 Relief in hot climates

The simplest and most commonly applied remedy for the
discomfort felt in a hot climate is increased air movement.
Velocities well above the 0.1 to 0.25m/s range appropriate to
optimum comfort temperatures are needed to exploit the
evaporative cooling capacity of the human body when the air is
hot. Circulating fans setting the air already in the room into
relatively high velocity, directed, jet-stream motion provide an
economical and effective system.
*This expression corresponds to a maximum Archimedean Number, Ar, of 104.

The Archimedean Number is a measure of the ratio of buoyancy to dynamic
forces. If g is the acceleration of gravity (9.81 m/s2) Dh the hydraulic diameter of
the room cross section, 2 BH/(B + H), and ∆T the initial difference in absolute
temperature:
3.3.2 Table and ceiling fans

Table fans, for individual use, and ceiling fans, for places of public
assembly, are well known and their sizes and speeds have been
standardised by the I.E.C. (International Electrotechnical
Commission).
Both types may be used, seated, with the head directly in the air
stream. They are therefore tested by exploring the air velocity
distribution at appropriate distances directly in front of the fans.
These distances are taken as three impeller diameters in front of
table fans, or 1.5 metres below a ceiling fan placed 3m above the
floor. Typical performances are shown in Fig. 3.6, the air volume
flows being integrated to include all the air entrained at the test
plane.
Speed control is usual since the highest air velocities will be

needed only on the hottest days. Oscillation mechanisms to swing
the jet stream through an arc of 60° to 120° are a refinement. The
advantage here is that air motion rising periodically to a peak is
found more pleasant than a steady velocity of the same average
value.
3.3.3 Pedestal fans

Pedestal fans are similar to table fans but of larger size and
longer throw. The oscillating jet-stream is generally directed above
the heads of the occupants to promote vigorous entrained air
movement over a considerable floor area. A typical plot of
constant-velocity contours in the jet stream is shown in Fig. 3.7 and
I.E.C. standard sizes are:
It will be seen that ceiling and table fans are capable of producing
air movement in the range 1 to 4m/s over the upper part of the
body. Velocities at the higher end of this range are welcome when
the temperature is very high. At more moderate temperatures the
fan speed may be lowered or table fans may be moved to moderate
the air movement.
Pedestal fans are perhaps most useful for occasional hot spells in
temperate climates when air movement is otherwise inadequate.
Being generally designed with a view to attractive appearance and
quietness they are most often found in restaurants and similar
surroundings.
3.3.4 Mancoolers
The portable cooling fan or mancooler is a much more powerful
jetblast device. It is used for the relief of industrial heat stress (see
Section 1.8) rather than comfort cooling. A constant-velocity
contour plot of the
jet from a typical mancooler is shown in Fig. 3.8. It will be seen

that the worker can be provided with a local air stream at velocities
from 4 to 20m/s according to the distance the fan is from him. With
this assistance he can perform short tasks in a spot at which the
heat would otherwise be insupportable.
CHAPTER 4
Heating and cooling

of buildings
We have seen that a part, or indeed the whole, of the heating

and cooling requirements of a building may be carried by the
ventilating air. Brief attention will, therefore, be given to the
principles governing the estimation of these loads.
A significant part of the load is often provided by the unintended
infiltration of outside air caused by the wind. A discussion of wind
effects on buildings is therefore included in this chapter.
4.1 The Heating Load

4.1.1. Design temperatures
Heat flows through the wall of a building in proportion to the tem-
perature difference between the space inside, tc, and the air
outside, ta. The flow, Hf, is further proportional to the wall area, A,
and to a constant, U, called the thermal transmittance and
dependent on the wall structure.
to will be the same as the comfort index temperature discussed

in Chapter 2. to is usually taken as the "external design
temperature" established by convention in different parts of the
world. This is not the extreme
40
Heating and cooling of buildings 41
minimum temperature on record, but is a value that can be expected on

several days in an average year. Here are some typical figures-though
the local convention should always be ascertained. See also Table 14.16.
4.1.2 Thermal transmittance

Some typical values for thermal transmittance, U, are given in Table
4.1. These are for illustration only and the heating engineer will need
to consult much more comprehensive data for the particular materials
and construction at his disposal. Good examples are those in the IHVE
and ASHRAE Guides. ,
For each value of U the corresponding gross area of wall, window,
roof, etc., will be calculated. Partitions, floors and ceilings should not be
forgotten, as they may not always be exposed to the same tempera-
ture, tc, on the far side. Floors on the ground are listed in Table 4.1
with the inclusion of ground transmittance through to to. Heat will flow
to the nearest exposed wall and a way of dividing the floor into
lengthwise strips and corner sections with different U values is indicated
in the key diagram.
4.1.3. Total heat load

The contributions of the sections are now added together to obtain
the total heat load, Hf, to maintain the comfort temperature, tc, with
outside temperature to, considering losses through the building fabric
only.
There will bean additional heat load, Hv required to raise the

temperature, of the fresh air brought in, Qv, from to to ta (not
necessarily the same as tc - see Section 4.2).
Solar and other heat gains discussed in Chapter 2 will form the major
part of the heat load on a cooling system. If the building is air condi-
tioned there will be an additional load due to the heat flowing into the
building from outside temperature to to the lower internal comfort
temperature, tc. This will be Hf + Hv calculated as before.
4.2 Relative Influence of Radiation and Convection

Comfort temperatures are assessed on the assumption that the mean
radiant temperature and the air temperature are the same. Heating the
room by warming the air only (100% convection) will mean that the
42 Woods Practical Cruide to Fan Engineering
walls will be relatively cold. This will increase the heat which the
body disposes of by the radiation route. To maintain constant heat
loss, and therefore comfort, the body must secure an equivalent
reduction in the heat disposed of by the convection route. This can
be done if the air temperature is raised, though the rise required is
less if the air movement is substantial. At an air movement rate
around 01m/s the rise in air temperature needed is about 2°C for
every 1 °C fall in the mean radiant temperature. At 0.2 to 0.4m/s it
will be nearer to 1 °C for every 1 °C.
Precisely the opposite effect occurs if the heat is supplied mainly
by radiation. The body's radiant heat loss will fall, its convective
loss must rise, and the air temperature for comfort will be below
the comfort index temperature by 2°C per 1 °C radiant excess at
0.1m/s or 1 °C per 1 °C at the higher air velocity.
Since the whole approximation process is fairly crude it is
sufficient to class heating systems as mainly convective or mainly
radiant. The following is a general classification by type, with typical
proportions:
Fortunately the influence of the thermal insulation of the room or

space under consideration can be estimated quite simply. The
heating design will have determined the total fabric loss, Hf, to
maintain the desired comfort temperature tc with design external
temperature to. Calculate the total area At of walls, floor and ceiling.
Then, for optimum comfort in the case of a mainly convective
heating system:
Imagine the case of a 100% radiant system. The radiant heat will
travel directly to the walls without heating the intervening air. If
there is no ventilation the air will ultimately reach the same
temperature tr as the walls. However, there must actually be fresh
air introduced at a rate Qv m3/s requiring Hv watts to reach tc °C.
This air can, however, be preheated to a lower temperature ta which
will enable a new balance to be found maintaining comfort
temperature tc. In the case of a mainly radiative heating system:
An example will show that a mainly radiative system will consume

rather less heat than a mainly convective one:
4.3 Example of Heating System

The machine shop of example 2.4.2 is to be heated to a comfort
temperature of 18°C with an external design temperature of -1 °C.
Minimum winter ventilation for 60 persons at 8 litre/s each = 480

litre/s. Installation of five window fans at 200 litre/s each is
planned, extracting from small supervisors' and shop clerical offices,
gauge room, lavatories, etc., along NE wall. Total extraction 1000
litre/s, equalling 1 .5 air changes per hour, and balanced by natural
infiltration without planned intakes.
Clearly this winter heat load will be supplied in substantial part by

the casual heat gains in the building (see example 2.4.2). However,
60 kW of heating plant will be installed for early morning and other
slack period heating, assuming ventilation loss at the natural
infiltration rate of about 10 kW only for these periods.
Consider first convection heating with four unit heaters of 15 kW
rating each and 500 litre/s air throughput. The distribution has been
illustrated in Fig. 3.2c and it is clear that, with entrained air
circulation, there will be substantial air movement. We shall,
therefore, take the factor in the air temperature formula as 0.15.
As a radiant heating alternative, consider twelve 5 kW overhead

gas heated panels. Air movement will now be low, and we will take
the air temperature factor as 0.2.
This is a saving of 9 kW, but the lack of air movement could be

unsatisfactory and there may be difficulty in siting the heaters to
avoid excessive local intensity.
4.4 Wind Effects

The wind provides the principal motive power both for the natural
ventilation of buildings and for the unwanted infiltration of cold or
hot air. Wind speed is the customary measure of this power and the
meteorological wind speed quoted in weather reports and statistics
has a specific meaning: in the UK it is the average air velocity 10
metres above the ground in open, level country. This is
substantially below the velocity reached at high altitude owing to
the drag forces exerted by the ground. In towns the buildings cause
a further reduction in average velocity, though eddies between and
around the buildings account for local peaking.
4.4.1 Wind speed statistics
Fig. 4.1 shows the standard variation of wind speed with altitudes
for a meteorological wind speed of 9m/s. Typical lower values
experienced in suburban areas and in a modern city centre with
many tall buildings are also plotted for the same meteorological
conditions,. 9m/s is recommended by the IHVE as representative of
severe winter weather in the UK for the design of heating systems.
Higher values may be considered for exposed locations, reaching a
maximum of perhaps 13 to 14m/s on a hill top or cliff edge exposed
to strong prevailing winds. For meteorological wind speeds other
than 9m/s all three curves of Fig. 4.1. should be adjusted in
proportion.
To be of significance for heat loading, both directly at the building

surface, and indirectly by cold air infiltration, the wind speed must be
sustained. Observations at Heathrow Airport, London, show that tem-
peratures below 2.2°C are combined with wind speeds as follows in
an average year:
Such cold winds tend to come from a predominant direction-in this

case north-east. In fact 50% of the above cases are within t 45° of NE
and 30°.6 within ± 22.5°.
The maximum wind speed likely to be reached during the lifetime of
a building is the important factor for structural design-including the
mechanical design of cowls and other exposed components of air
conditioning systems. A map of the UK published by the Building
Research Station gives contours of maximum gust speeds (at the
10m, open ground, level) which are likely to be exceeded once only in
50 years. They range from 38m/s in the London area to 55m/s in the
Outer Hebrides.
4.4.2 Natural ventilation involves open windows, doors and

ventilators of substantial area deliberately provided to promote
air change by wind and stack effects. The volume flow passing
through an opening in a wall exposed to the full impact of the
wind would theoretically equal the free area of the opening
multiplied by 65% of the unobstructed wind speed. Half this
would be a more likely value in an actual building, mainly
because of the obstructions to flow through the building and out
on the leeward side.
A fairly conservative estimate of the volume flow through a
building with no internal partitions is:
Here Vw (m/s) is the free wind velocity at roof level and Ae is

derived from the total free area of inlet openings on the wall
facing the wind, A1m2, and the total free area of outlet openings
on the remaining faces, A2M2:
4.4.3 Stack effect arises from the buoyancy of the warm air
inside a building which will be at a lower density than the cold
air outside. It is most effective in high, single-storey, factory
buildings with ridge mounted or other ventilators in the roof. An
approximate formula for the volume flow is:
Ae is calculated from the formula in the preceding section, with

A2M2 the free area of the roof outlets and A1m2 the free area of
the inlets which should be not far from ground level. H1m is the
height of the outlets above the inlets, to°C the outside
temperature, and to°C the average inside temperature over the
height H.
Example
A factory bay 40m x 14m x 15m high to the ridge, volume
7500m3, has ridge-mounted natural ventilators of 16m2 free
area, and bucket type window ventilators 2m above the ground
along each wall of 20m2 total free area. If the average air
temperature is 25 °C inside and 20°C outside what is the
ventilation rate?
If the temperature reached at the ridge outlet is 30°C, the

ventilation will remove:
equals 200 kW of process heat, 1.2, and 1.0 being the density
and specific heat of the air, respectively.
4.4.4 Infiltration
When windows are closed, inward and outward leakages still take
place round the edges, and have a significant effect on the heating
requirements. The leakage depends on the pressure across the
window and its sealing qualities. As a guide to estimation three
standard leakage coefficients, Aw have been established for the
infiltration rate per metre of joint which are representative of the
upper limits to be expected from modern, metal-framed windows:
λw
Weather-stripped and pivotted 0.05 litre/s per metre at 1 Pa
Weather-stripped and sliding 0.125 litre/s per metre at 1 Pa
Not weather-stripped 0.25 litre/s per metre at 1 Pa
When the length of window joint is Le metres and the pressure

difference ∆p Pa, the leakage rate of standard air expected is:
Q = λw . Le . (∆p)0.63 litre/s
The distribution of pressure over a building of the rectangular
office block type has been studied with wind-tunnel models, and
typical results reported by Jackman (HVRA report 53) are
summarised in Fig. 4.2. This expresses the positive and negative
pressures over the windward and leeward faces of the building as
multiples of ½ ρ v2w where vw m/s is the unobstructed wind speed
at the height of the roof of the building.
As a general conclusion the gross pressure difference providing
the motive force for the inward and the necessarily equal outward
leakage flows is taken to be 1.2 x ½ ρ v2w. Half of this is assumed
to act across the inflow side.
To evaluate the effective length of window joint on the inflow
faces, Le, consider a rectangular building facing north, south, east
and west with total lengths of window joint LN, Ls, LE and Lw on the
respective faces. La would be LN for a north wind, but is increased
to to take account of the worst wind direction in the
north-east
quadrant. Fig. 4.3 is plotted to give the leakage rate directly from
the wind speed at roof level derived from Fig. 4.1 in accordance
with:
Example
An open plan office block in a suburban area is 30m x 15m x
50m high and has 200 windows each on the north and south
faces, 100 windows each on the east and west. The windows
are not weatherstripped and each has a periphery of 5m. What
winter infiltration rate should be allowed?
4.4.5 Infiltration-air change basis

Buildings with irregular fenestration and internal subdivision
into rooms and corridors do not lend themselves to leakage
calculations. Empirical values have been published by the IHVE
for use in rough estimates of winter heat loading. The rates in
Table 4.2 have been extracted from their tabulation.
4.5 Cooling Systems

A full air conditioning system with refrigerating and dehumidifying
plant is of course capable of securing complete comfort in the
hottest climate. The component parts of such a system can be
assembled in many different ways, and the subject is outside the
scope of this book.
4.5.1 Evaporative cooling

One simple cooling system, known as evaporative cooling, which
requires a fan as its only moving part, may be described here.
It is relatively inexpensive but is practicable only in a hot dry
climate (with relative humidity usually less than, say, 50%) where
nevertheless a supply of running water is available.
It was pointed out in Chapter 1 that a considerable amount of
heat is necessary to convert liquid water into water vapour. This is
known as the latent heat of evaporation and amounts to more than
2400 kJ/kg (kilojoules per kilogramme) at atmospheric
temperatures.
In an evaporative cooling installation hot air from outside the
building is drawn through and over layers of fibrous material kept
continually wetted by gravity flow or capilliary action, or through
sprays. This promotes evaporation, and the only sources from
which the corresponding latent heat can be supplied are the air and
water flows. Almost all the latent heat must come from the air,
heating or cooling the water to wetbulb temperature only
accounting for 1 or 2% of the total.
The additional moisture content of the air leaving the evaporator
carries with it a corresponding quota of latent heat. This equals the
heat extracted from the dry air part of the air water vapour mixture
so that the total heat of the mixture (called the enthalpy) remains
the same before and after the evaporation. The heat content of the
dry air part is called the sensible heat. When this is reduced the
dry-bulb temperature
falls at the rate of about 1 °C for each 1000 joules of sensible heat
lost per kilogramme of dry air. Two very simple approximate
relationships follow:
A cooling rate of 1 kW (= 1 kJ/s) applied to an air flow of 1 kg/s
will reduce the dry-bulb temperature by 1 °C.
Latent heat at the rate of 1 kW is extracted by the evaporation of
water at the rate of:
Combining these, 1.5 Mt litre per hour of water must be

evaporated to cool M kg/s of air by t°C dry-bulb.
It must be realised that the dry-bulb temperature of the air has
been reduced at the expense of increasing its relative (and
absolute) humidity. This reduces the cooling effect available, which
however still remains
substantial provided the air was not too humid to start with. The
process can be followed on a portion of the psychrometric chart
selected in Fig. 4.4 to suit hot climate applications. On this chart
lines of constant
effective temperature have been drawn-the broken lines sloping

steeply down. These have been plotted from a chart similar to Fig.
1.3 but drawn for people with customary indoor clothing, sitting,
with light activity. Also for an air velocity of 1 m/s as the lowest
value appropriate where cooling air streams from table and ceiling
fans are available.
4.5.2 Example of evaporative

cooling
Suppose the air outside is in the condition represented by A on
the diagram (44°C, 30% RH). As it is drawn across the wetted
surfaces its dry-bulb temperature will fall, along a constant wet-bulb
temperature line, towards saturation at C where wet-bulb and
dry-bulb would be equal at 28°C. In practice it will only cool a
certain fraction of the way towards saturation, emerging from the
cooler in condition B shown as 34°C dry-bulb-62% of the way
towards saturation.. This saturation efficiency of 62% is typical of
simple wetted fabric coolers. Spray type coolers may reach 90%,
but will use water many times in excess of the minimum necessary
amount, unless the water is recirculated.
It will be seen that we have reduced the effective temperature
from nearly 32°C ET at A to 29°C ET at B, a useful amelioration of
an unpleasantly hot environment. This is achieved with a simple
construction in which panels of chemically treated fabric, glass fibre
or wood-straw are fed with a controlled quantity of water calculated
to match the required cooling rate. Given a volume flow of 4M3/s,
mass flow 4.5 kg/s at 1 .12 kg/m3 density, the evaporation rate is:
4.5.3 Comparison with refrigeration methods

If the same outside air were passed through a cooler battery of
pipes fed with chilled water, the air condition would move along a
line such as ADE. E is the dew-point maintained on the outside
surface of the pipes at 18°C. Supposing the saturation efficiency to
be 65%, the average condition of the air leaving the battery would
be D. 17° drybulb and more than 8° effective temperature
reduction have been achieved, and the moisture content has been
reduced by condensation on the pipes, though the relative humidity
is up to 63%.
With full air conditioning more heat and moisture would be
extracted at the cooling stage, by the use of refrigerant in the
cooler battery in the place of chilled water. This cools the air still
further, but at the expense of a rather high relative humidity, say at
point G. A reheat stage follows at constant moisture content (i.e.
horizontally on the chart) to H at the target dry-bulb temperature of
26°C with a satisfactory relative humidity. It is the overcooling and
reheating stages which give the complete control over the final air
condition characteristic of full air conditioning.
CHAPTER 5
Pollution control
The ventilating system has the major role in the control of

pollution within buildings. The techniques required to deal
with toxic and explosive gases, vapours, fumes and dust
form the subject of this chapter.
5.1 Toxic Gases and Vapours

Complete exclusion of poisonous material from the human body is
impracticable. Traces of virtually any substance may from time to
time be found in the food we eat or the air we wreathe. However,
the body can successfully eject or absorb such foreign invasions,
provided the quantities involved are not too great.
Medical science has identified and studied the action of a great
many harmful substances. Many of these appear as gases or
vapours in industrial atmospheres and experience has shown what
concentrations are tolerable. Table 5.1 lists some common gases
with the limitations usually recommended but it is intended only to
illustrate the application of ventilation techniques. Threshold limits
are officially specified in many countries and must be observed.
National legislation administered by the Factory Inspectorate or
Public Health Authority must always be studied and respected. The
53
regulations deal in detail with many more chemical formulations than

could be listed here and are kept abreast of current medical opinion. The
maximum allowable concentration, known as the "threshold limit value" is
such that lifetime exposure during the working day should not damage
the health. Nevertheless, some individuals may be particularly susceptible
to certain toxic agents; medical tests can often identify such people, and
prevent their employment in areas harmful to them.
Some poisons are cumulative in their effect, and the allowable
concentrations are then far below levels that would be quite safe for an
isolated exposure. Others cause immediate harm, and these can be
identified in Table 5.1 by noting that the short period maximum is the
same as the daily average.
5.1.1 Dilution ventilation for toxic hazards

General ventilation will deal with many of the toxic hazards arising in
industry. The manufacturing process must be studied to determine the
rate at which the pollutant is generated. If this is expressed in milligrams
per second* of gas or vapour, with density ρkg/m3 at atmospheric
temperature and pressure, then the ventilation rate required for dilution
is:
In this formula both M and ppm (parts per million by volume) may be
either daily average or short-period values, whichever gives the higher
value of Q. SF is a safety factor, commonly chosen between 3 and 6,
according to the uncertainty of the data and to possible defects in the
ventilation distribution. Good distribution was discussed in Chapter 3; the
danger is of local gas concentration in places where workers may spend a
significant part of their time.
5.1.2 Example
Acetone is used as a solvent in a workroom of 50M3 volume. The
maximum recorded consumption is 5 litres in an eight hour shift.
The density of liquid acetone is 790 kg/m3, that is 0.79 kg/litre, so that
the vapour generation rate is about 4 kg in eight hours, ½ kg/hr or 140
mg/s.
The maximum daily average concentration of acetone vapour from
Table 5.1. is 1000 ppm and the vapour density is 2.44 kg/m3. Taking the
factor of safety as 5, the required general ventilation rate is:
Pollution control 55
5.1.3 Fume cupboards

General ventilation is not recommended for the more highly toxic gases
(such as those listed at less than 50 ppm). Nor will it be economic if an
excessive volume flow of air is called for, requiring to be heated in winter.
Local extraction is the answer, the sources being preferably enclosed within
individual hoods on the lines discussed in Section 5.4 on fume removal.
For general purpose laboratory use the fume cupboard is usual, and Fig.
5.1. illustrates the principles of construction. With the sash type door fully
open for access, room air should enter at an average velocity of 0.25 to
0.75m/s according to the toxicity of the fumes handled. The sash should not
close fully, 25mm opening being left at the bottom to maintain airflow,
through the cupboard.
Extraction is through top and bottom slits in a baffle leading to the exhaust
duct. This ensures the extraction of heavy gases collecting towards the floor
of the cupboard, as well as light gases at the top.
When dangerous micro-organisms or highly radio-active materials have to
be handled, protection from contact must be as complete as possible.
Glove-boxes with air-locked entry arrangements for specimens and
high-efficiency filters monitored for pressure drop in the exhaust ducts are
used, see Fig. 5.2.
5.1.4 Dust hazards

Certain dusts encountered in mining and quarrying cause pneumoconiosis,
a serious group of diseases in which the lungs are progressively damaged,
over years of exposure, by the formation of scar tissue. Some
of these are listed m Table 5.2, with recommended maximum concentrations

either on a weight, or on a number of particles, basis. Many chemical and
metallurgical dusts and fumes are highly toxic, and will be subject to
precautions well known in the industries concerned. Ordinary dust derived
from the soil and from combustion is, however, relatively harmless, but
should be kept within the industrial nuisance limits quoted in the interests of
comfort and safety (e.g. good visibility).
Dust clouds can also explode, but only in concentrations exceeding the
10mg/m3 quoted as the maximum for industrial atmospheres. Aluminium and
magnesium powders cause particularly violent explosions, while organic dusts
such as starch are easily ignited. Good housekeeping is important, with
accumulations of dust in odd corners rigorously excluded. Otherwise a small
local explosion may raise a dense dust cloud resulting in a massively
destructive blast.
General ventilation is not effective for the control of industrial dust. Local
extraction at source will be required, followed by dust collectors to prevent
dust nuisance outside the plant. These are discussed later in this chapter.
5.2 Explosion Hazards
5.2.1 Dilution ventilation for explosive gases

Most explosive gases and vapours are toxic. To form an explosive mixture
with air they must be present in a proportion rarely less than 1 % by volume,
that is 10,000 ppm. It follows that workshop atmospheres which are
satisfactory for health are unlikely to present an explosion risk.
However, local extraction systems will produce much higher concentrations
of pollutant in the exhaust ducts. Sufficient air must always be extracted
along with the gas or vapour removed to dilute the latter to a safe
concentration. Table 5.3 lists some dangerous substances giving the lower
limit of the range of flammable (and therefore explosive) concentrations.
There is also an upper limit, but it is never safe to operate an extraction
system on the assumption that high non-explosive concentrations can be
maintained.
The ventilation rate for dilution can be calculated by formula 51, the same
as for toxic gases. It is good practice to ensure that the peak local
concentration of pollutant does not exceed 25% of the lower flammable limit.
This means that the factor of safety, SF, based on total airflow and gross gas
or vapour production rates, must substantially exceed 4. For example, SF =
10 is the lowest value recommended for batch ovens, even when precautions
are taken to ensure air circulation round the heated stock.
5.2.2. Example
A batch oven dries painted sheet steel products at a temperature of
180°C. Xylol solvent is driven off at an estimated rate of 1.5 kg per hour (=
420 mg/s). The lower explosive limit for xylol is 10,000 ppm (Table 5.3) and
the density of its vapour is 3.8 kg/m3 at room temperature.
As both the air and the vapour will leave the oven at 180°C, the flow rate
to be produced in the extract duct (proportional to absolute temperature, i.e.
273 + t °C) will be:
5.2.3 Properties of flammable gases and vapours

When supplied at a limited rate, as through a burner, and ignited, gases
and vapours will continue to burn at the point where they are diluted with
sufficient air to enter the flammable concentration range. The danger of fire
can be judged by the characteristic properties given in Table 5.3, thus:
(a) The upper and lower explosive limits, or limits of flammability with air. A
wide range, e.g. the 4% to 74% of hydrogen, indicates particular danger
of ignition and explosive propagation. At the other extreme, methyl
bromide with 13.5% to 14.5% is practically noninflammable, and is, in
fact, successfully used as a fire-extinguishing medium.
(b) The ignition temperature. This is the minimum temperature at which a
flammable mixture of air and gas or vapour will ignite. It is chiefly
important when it is abnormally low, as in the case of carbon disulphide.
This ignites at about 120°C, and may be fired by a steam pipe, or even
by the hot surface of an electric motor.
(c) The flash point applies to liquids. It is the lowest temperature at which
sufficient vapour is given off to cause explosive burning when exposed to
a means of ignition. It is a quantity which is dependent to a considerable
extent on the method of test; a widely-used rating based on the closed
cup method is listed in Table 5.3. Liquids with closed-cup flash points
below 21 °C should be regarded as highly inflammable.
5.2.4 Control of explosive atmospheres

Many industrial processes, including such common activities as
cellulose-spraying and petrol-dispensing, are carried on in spite of the
increased fire hazard associated with the materials handled. Precautions must
then be taken to limit the chance of fire or explosion to a very low level.
This is best achieved when the probability of disaster is the product of two
probabilities, each in itself small. Thus we may ensure:
(a) That sources of ignition are excluded from the hazardous area, and can
arise only by accident or human error.
(b) That flammable concentrations of the hazardous gas are excluded by
design, and can arise only in abnormal circumstances, e.g. failure of the
ventilation system.
Thus if the chance of a flammable concentration is 1 in 1,000 and the
accidental ignition hazards five per annum, then a fire will occur once in
about 200 years - or longer since the two abnormal happenings must
coincide in space as well as time.
In keeping with this philosophy hazardous atmospheres have been

classified into three types or Zones. In Zone 0 the probability of a flammable
mixture being present approaches 100%, in Zone 2 it is very small; and in
Zone 1 intermediate between these extremes.
5.2.5 Electrical apparatus for explosive atmospheres

The International Electrotechnical Commission has published as IEC 79
guidance for the safe use of electric motors, control gear, lighting equipment,
etc. in flammable atmospheres. This is based on the experience and research
of various national bodies, and in most countries corresponding specifications
exist - in the UK BS 4683. These specifications are generally followed by the
responsible authorities in approving particular industrial installations and
apparatus, though judgement and discretion remain necessary. Both the
apparatus and the environment require study, and an outline of the
classifications adopted appears on the next page.
In Zone 0 areas it is preferable to exclude all electrical apparatus except
light current intrinsically safe equipment classed Ex ia. This cannot draw
enough power from the electrical supply to overheat or produce an
incendive spark (i.e. one with more than 0001 joule energy). When motors or
other electrical equipment must be used in a continuously explosive
atmosphere, Ex p, pressurised, machines can be considered, though, with all
the necessary precautions, they tend to be excessively costly except in the
largest sizes.
Ventilation may provide a solution by reducing a Zone 0 to a Zone 1 or 2

danger level. A plantroom or cubicle may be so treated, reducing the level of
machine construction required to Ex d or Ex e. The approach for Ex e
certification is to make sure that the machine is so designed and controlled
that failure causing sparking or overheating is very unlikely. With Ex d
protection on the other hand it is assumed that the most explosive mixture of
the gases in question has penetrated the enclosure and been ignited. It is
then ensured (by type test) that the enclosure will withstand the resulting
pressure, which may reach ten atmospheres; also that the hot gases
escaping through the specified spigot joints, gaps around shafts, etc. will be
sufficiently cooled not to ignite the explosive mixture outside.
5.2.6 Fans for hazardous atmospheres

Fans may be driven by Ex d, Ex a or Ex n electric motors and controllers. If
the motor is outside the fan casing, as with centrifugal or bifurcated axial
types, Zone 1 conditions will certainly be met. To secure the more relaxed
Zone 2 rating it is necessary to ensure that the pressure within the fan casing
is below atmospheric pressure; consideration must also be given to the
possibility of flammable gases coming into contact with the motor by leakage
or otherwise when the fan is stopped and restarted.
Axial fans with Ex d flameproof motors in the air stream have been widely
used in the past, but the authorities now look at such installations more
critically. There is no doubt that Zone 1 conditions will be met with proper
dilution of the hazardous fumes. In such applications as cellulose spray
extraction, however, there is concern that overspray deposits on the motor
carcase might cause overheating and consequent ignition. The fourth hazard
classification-maximum surface temperature - is relevant here, although at
the time of writing it is not backed by much data on the ignition temperatures
of different gas or vapour mixtures. Motor carcase temperatures will
generally be within 100°C by design - Class T5. Therefore, if adequate
precautions are taken against overheating, only gases marked F of those in
Table 5.3 should be suspect.
Another factor to be considered is the danger of incendive sparking if
metals or stones are struck together in an explosive atmosphere. A a number
of explosions in coal mines have been traced to this cause and equipment in
steel or cast iron is now specified. Magnesium, and to a lesser extent some
aluminium, alloys have been found to produce more energetic sparks, more
likely to ignite the methane in the atmosphere of fiery mines.
5.3 Air Cleaning

The atmosphere out-of-doors carries in suspension a wide range of solid
particles classified as dusts, fumes or smokes, as well as liquid droplets in the
form of mist or rain. Typical quantities of solid matter lie within the following
ranges:
Rural and suburban areas 0.05 to 0-5mg/m3

Metropolitan districts 0.1 to 1 mg/m3
Industrial districts 0.2 to 5 mg/m3
Temporary local concentrations of 100 or 1000mg/m3 are common in

mining, quarrying or dusty factory operations, but they should not be allowed
to persist. "Smokeless zone" legislation has had considerable success in
cleansing the air of cities.
Even in clean atmospheres the air supply to public buildings is generally

filtered. This makes only a minor contribution to health and comfort, since
our nasal passages provide excellent filtration for the air we breathe. The
value is economic, since the cost saved in less frequent cleaning and
redecorating far outweighs the cost of servicing the filters. Much higher
standards of filtration can be provided for the clean rooms and sterile rooms
needed for sensitive industrial and laboratory operations.
Attention must also be paid to the possibility of nuisance when

contaminated air is discharged from a factory. A high chimney may suffice,
relying on the wind to spread the contaminant over many square kilometres
of countryside. Nevertheless, anti-pollution legislation is always tightening
standards, and when planning new plant it is wise at least to allow space for
the insertion of air cleaning equipment at a later date, if the slightest doubt
exists.
5.3.1 Classification of dust

The upper part of Fig. 5.3 gives a general impression of the size range of
the terms commonly used to describe particulate matter. Fumes are solid
particles formed when the vapours of substances normally solid condense in
the atmosphere. Size is the main factor determining the behaviour of the
particles in air, and the three linear charts of Fig. 5.4 result from exact
calculations for spherical particles. Of course, actual particles are not so
conveniently shaped, nor are they graded into species of constant size.
Nevertheless the extremely rapid change of each quantity with size gives
significance to even the roughest estimation.
Dust concentrations can be estimated by weighing filter papers through

which measured volumes of the air have been passed. Numbers and sizes of
particles are estimated by examining samples captured on microscope slides.
Statistical analysis can then give the research worker separate distributions of
number, weight or surface area by size. Each may then be expressed in terms
of equivalent spherical particles.
The terminal velocity reached by particles in free-fall through the

atmosphere determines the method to be used for their removal. In steady
fall the atmospheric drag force holding the particle back will equal the
downward force of gravity on it. Several distinct flow regimes can be
distinguished.
Turbulent flow applies to particles larger than 600µm or so, when the drag
force increases with the square of the velocity.
Streamline flow is predominant in the 0.1 to 100 µm range. The drag force
is viscous and proportional to velocity.
Transitional flow occurs between the turbulent and streamline forms, the
drag force being higher than in either.
Brownian movement exists below 0.1µm. Such small particles partake of

the random motion of the air molecules around them, with no net tendency
to move downwards.
The air itself is always in motion, and when its velocity is well above the
terminal velocity of a particle, that particle will be carried with the air. Thus
particles in the 1 to 100µm range are classed as temporary impurities - quite
mild air movement will keep them in suspension for a substantial period.
Smaller particles move so slowly under gravity that they are likely to remain
as permanent impurities. Larger particles will settle more or less rapidly, but
can be transported, and thereby removed, by deliberate air currents of
sufficient magnitude.
5.3.2 Performance of air cleaners

There is little standardisation in the design of air filters and dust collectors
so that manufacturers' data should be studied and their advice sought in case
of doubt. The efficiency of an air cleaner is that percentage (by weight) of the
dust in the incoming air which is captured by the cleaner, the remainder
passing through to the clean air outlet.
As the efficiency is very much dependent on the size of the particles, rating
tests are carried out with certain standardised dusts; the size distributions of
some of these are illustrated in Fig. 5.5. Efficiencies from 70% or 80% to
99.99% are available, but an unnecessarily high performance attracts
disproportionately high maintenance and first costs.
Air cleaners capture particles by a variety of mechanisms which may be
summarised as follows:
Gravity and centrifugal force effective in transporting large heavy particles

to a collecting point.
Screening of particles which are too large to pass through the air passages
in a filter medium.
Adhesion to sticky or wetted surfaces off which the larger particles would
otherwise rebound.
Impingement on dry surfaces by particles small enough to be retained by

the surface forces associated with intimate contact.
Adsorption of gas or vapour molecules which occurs, selectively, by

condensation on the extended internal surface of certain microporous
substances.
5.3.3 Classification of air cleaners

Fig. 5.3 shows the range of particle size for which each of the main types is
best adapted.
Settling chambers. While settlement is relied on to dispose of dust clouds in
the natural surroundings of mines and quarries, settling chambers are rarely
used. The size of chamber is generally too great in relation to the volume flow
to be handled.
The cyclone is a dynamic settling chamber in which a stable vortex motion

is established by a high velocity jet of dusty air entering tangentially.
Centrifugal force drives the heavier particles to the wall. There they travel
round and downwards into a hopper, as illustrated in Fig. 5.6, while the
cleaned air emerges upwards at the centre.
Fans should not be connected to a cyclone outlet without consulting the
fan makers. The rapidly swirling air stream from the cyclone will greatly
increase the power loading of a fan rotating in the opposite direction.
Conversely the performance of a fan rotating in the same direction will be
greatly reduced.
Cyclones are used in the conveying and handling of grain and flour, in
collecting the dust from grinding, buffing, sanding or crushing processes, and
for recovering re-usable industrial dust. Pressure drop ranges from 200 to
500 Pa.
High-efficiency cyclones. The centrifugal force on a particle of mass m
moving in a circle of radius r is m v2/r. Air cleaners comprising a multiplicity of
small cyclones operating in parallel exploit this fact. By reducing r the force is
increased, and much smaller particles can be extracted. Finer grades of
chemical, pharmaceutical and food products are handled, and pulverised fuel
fly-ash can be extracted from high temperature flue gases. Pressure drop
ranges from 500 to 1500 Pa.
Wet collectors may take the form of cyclones in which the collecting
surface is wetted with a continuous water spray. These will be more efficient
than the corresponding dry cyclones, but power must be supplied to pump
the water, and the water itself will usually need to be filtered before
discharge.
Spray chambers or towers containing wetted surfaces designed to deflect

the air repeatedly will extract dust. They are widely used for the same
industrial purposes as cyclones - particularly in the ceramics industry. They
may also reduce fire and explosion risk.
The washers and scrubbers used in air conditioning plants are primarily for
humidity control, and should not be relied on to remove fine atmospheric
dust.
Viscous filters contain panels of fabric or fibrous mat, often in plastic, glass
or metal strands, and coated with an appropriate oily liquid. The oil must
have good wetting power and be odourless, non-volatile and noninflammable.
The tortuous path of the air through the panel causes dust particles to
impinge on the strands where they adhere to the oily surface.
The strands of viscous filters are generally relatively coarse and open and
most are at their best with particles in the4to 10Nmrange. The panels are
usually washable and can be re-oiled and re-used repeatedly. Automatic filters
can, therefore, be constructed from endless bands of filter material
continuously circulating through a bath which serves for both washing and
re-oiling.
Viscous filters find their major application in the cleansing of the air supply
to buildings. They are particularly useful in cities with badly polluted
atmospheres, as they can deal economically with heavy dust loads. Pressure
drop is low - 100 to 150 Pa.
Dry filters are also made in panels or rolls of fabric or fibrous material,
natural or synthetic.
For air conditioning supply systems they have finer, more closely packed
strands and will remove most of the carbonaceous dusts in the 0.5 to 5m
range responsible for dirtying fabrics and decorations. The material is
generally thrown away when dirty, but can be cleaned in special solutions.
To maintain a moderate pressure drop-25 to 60 Pa when new, 100 to 250

Pa when soiled - the material is generally folded into a V-formation, see Fig.
5.7. In this way the customary face velocity for all air conditioning
applications – 1.25 to 2.5m/s - can be maintained with a much lower air
velocity through the dense material.
Fabric arresters are dry filters made in bag or stocking form for the capture
of industrial dust. The purpose may be recovery of valuable material or
prevention of dust nuisance. The fabric is chosen to suit the dust and, unlike
cyclones, can be effective on particles of less than 1 µm.
The material is removed periodically by reverse blowing or shaking most of it
builds up as a dust layer on the upstream side of the fabric. The method is
used for textile fibres, chemicals, ceramics, grain, cement and woodworking
dust. To limit the space occupied pressure drops are high - 500 to 1500 Pa at
0.01 to 0.05m/s through the fabric.
Electrostatic-precipitators as used in air conditioning are two-stage devices.

The incoming air is first passed through an ioniser. The assymetric electric
field between earthed rods and fine wires held at a positive potential of up to
13,000 volts gives to each dirt particle a positive charge. In the second stage
the air with its load of charged particles passes between plates alternately
earthed and held at + 6000 volts.
Each particle is repelled by the positive and attracted to the (earthed)
negative plate, to which it becomes attached. The plates are periodically
washed to remove the dirt. This principle ensures that the smallest particles
will be removed, but large particles can cause electrical breakdown and those
above 5µm should have been first removed by a viscous filter or otherwise.
The first cost of electrostatic precipitators is high although they are cheap
to maintain. They come into their own in very large sizes to control industrial
pollution. Such applications include the cleansing of flue gas from power
stations and blast furnaces, or of contaminated air
discharged from large chemical plants. For these installations ionisation and
collection are combined in one stage with wires at 30,000 to 75,000 volts
(negative) between earthed plates. The pressure drop of all electrostatic
precipitators is very low, and the expense is associated with the power pack
to produce the high DC voltages and the safety precautions these voltages
necessitate.
Absolute filters are made from a dry filter medium akin to paper or felt.
They are made possible by the very small diameters in which glass and
asbestos fibre can now be produced. The pores are so fine that the diffusion
or Brownian movements of particles as small as 0.01 µm will cause
impingement and retention. The medium can be quite thin so that, with deep
corrugations, maximum face velocities of 1.25 to 2.5m/s can be reached.
Pressure drop is likely to reach 250 to 500 Pa, and soiled filter medium must
be thrown away. If the filter is used to capture bacteria only it can be
sterilised and re-used several times.
Such filters are used in the air supply to clean rooms or sterile rooms of the
highest standard. Operating theatres, pharmaceutical, photographic and
biological laboratories and the trapping of radio-active dust may be
mentioned. If necessary an efficiency of 99.995% can be reached with
methylene blue test dust. The cost is high and pre-filters are necessary to
remove coarse dust and lengthen the life of the absolute medium.
Activated carbon filters can be used for the adsorption of gases and
vapours. The carbon is produced in granulated form, packed into panels
through which the room air is circulated and recirculated. The granules need
not be tightly packed, and many of the large molecules characteristic of
odours are captured in an active way reminiscent of the capillarity of liquids.
Each carbon granule is a structure with a large surface area formed on a
multitude of internal pores. The medium has a long life since it can absorb up
to half its own weight of gas or vapour, which can then be driven off by
heating and the carbon used again. 95% efficiency is obtainable and the
pressure drop can be constant in clean air at about 60 Pa. These filters are
perhaps most commonly found in cooker hoods for domestic kitchens.
5.4 Dust and Fume Collection

The importance of local extraction at source has been mentioned several
times already in this chapter. The requirement is to see that the great
majority of grit, dust or fume particles are transported from the point where
they are generated into a duct along which they are conveyed to a discharge
or collection point.
The critical phase is the capture of the particle by an air stream sufficiently
strong to direct and transport the particle towards the duct. The air direction
must be proof against excessive deflection by draughts; 0.5m/s is usually
enough to secure this. 0.5m/s is also sufficiently above the terminal velocity
given in Fig. 5.4 for the capture of most dusts. Industrial grit, however, is
usually too heavy for air capture; hoods must be placed in line with the
expected flight from grinding wheels and
similar sources. Similarly the natural flow of air and vapour upwards from
boiling pans or other heated sources must be respected in the siting and
sizing of a hood above.
The velocity of the air towards a suction point falls rapidly with the distance
away-in fact with the square of the distance when this is more than about
twice the diameter of the suction opening. The distribution of velocity close to
circular suction openings is shown in Fig. 5.8, the figures being air velocities
in per cent of average duct velocity. Wide flanges will improve the inward
velocities in their neighbourhood, but will make little difference to the
approach velocities along the centre line.
5.4.1 Hoods and canopies

The first rule for design is that the hood should enclose the source of fumes
or grit just as completely and closely as is possible without impeding
production operations. Maintenance operations should be allowed for only by
making the hood, or part of it, removable. Not only will such a hood intercept
and capture the inevitable stray particles: it will also need less volume flow to
maintain the required approach and capture velocities.
A number of examples are shown in Figs. 5.9 to 5.13. In each case a
nominal inlet area is identified together with the average air velocity that
should be maintained over this area. Multiplying the two together gives the
m3/s to be extracted Where a range of velocities is indicated, the higher
values are for higher densities of grit, high cutting velocities, and greater
rates of particle production. Over boiling pans, kettles or tanks the air volume
flow must exceed the highest possible rate of steam or vapour production.
Very large canopies must be subdivided with separate exhaust outlets to

make sure that one end is not starved. Alternatively, a baffle plate as Fig. 5.9
may be used to ensure even intake distribution with a single exhaust duct.
Fumes from cold tanks can be removed by a full length extract slot as Fig.
5.10a. Volume flow ventilation and heat loss may be reduced by a supply slot
providing an air curtain across the tank surface as Fig. 5.10b, while heat loss
from the tank is often limited by a layer of plastic balls floating on the
surface.
5.4.2 Dust transport

When the dust-laden air has been collected transport must be assured by
an adequate air velocity in the exhaust duct. Table 5.4 gives some
recommended conveying velocities.
CHAPTER 6
Air flow and pressure in
ducted systems
The air flow rates and velocities required for adequate

ventilation have been discussed in the first five chapters.
This chapter is concerned with the fan pressure needed to
circulate the flow.
6.1 Definitions of Pressure

The opening discussion of the meaning of pressure may seem too
detailed and too elementary to many. However, the exact meaning
of the terms used is vital, and experience has shown that
misunderstanding can easily arise and persist. A summary of the
facts will be found in Table 6.1.
6.1.1 Pressure, absolute and atmospheric
It will be accepted that atmospheric air experiences a pressure,

caused by the weight of the air above. This is the barometric
pressure po and is quite substantial, amounting to about 100,000
pascals. In the units used by meterologists 100,000 Pa is 1,000
millibars, since 1 mb = 100 Pa exactly; older barometers will still
read the gravitational equivalent - 750mm of mercury.
If air is blown into a child's balloon it will be commonly described
as "under pressure" - a pressure applied by the tension in the
rubber skin of
75
the balloon. The air inside the balloon does experience a greater
pressure than the atmospheric air outside, though the difference
will not be very great-perhaps 105,000 Pa compared with 100,000
Pa in the atmosphere.
The term "pressure" without qualification describes both po =
100,000 Pa outside the balloon and pa = 105,000 Pa inside - the air
being supposed to be still in both cases. If there is any danger of
misunderstanding, this form of pressure can be particularised as
absolute pressure.
612S i
For the purposes of fan engineering it is the difference between

the absolute pressure at the point under consideration and the
atmospheric pressure which is of importance. This is called the
static pressure, and in the above example the static pressure of the
air in the balloon would be 105,000-100,000 = 5,000 Pa. Static
pressure is reckoned positive when the absolute pressure is above
atmospheric pressure, negative when below.
The main advantage of working with static pressures rather than
absolute pressures is that the natural variations in atmospheric
pressure are compensated. Thus if the barometric pressure in our
example had been po = 960 mb instead of 1,000 mb we should
have found that the same effort would result in blowing up the
balloon to pa = 101,000 Pa, when the tension in the skin would have
also been the same. This corresponds to an unchanged static
pressure of 101,000-96,000 = 5,000 Pa.
Calculations in terms of static pressure will be independent of
atmospheric pressure only so long as the static pressure is
relatively small.
The air can then be treated as an incompressible fluid (except for
density determination) in which case the laws of motion are
relatively simple. Errors due to compressibility are just perceptible
at 2,000 Pa static pressure and become significant above 5,000 Pa.
In this book we shall deal with the influence of compressibility on
fan testing where we want precise results. Otherwise we shall
assume that the air is incompressible.
6.1.3 Velocity pressure, pv
It is common experience that the wind exerts a force on any

object standing in its path. This is mainly because the absolute or
static pressures on the windward side are greater, on average, than
those on the leeward side. The wind is not, of course, stopped in
any final sense - it flows round the body. There is, however, always
one point on the surface at which the air is actually brought to rest,
without flowing either to right or left, as illustrated in Fig. 6.1. At
this point, called the stagnation point, the pressure at S (both
absolute and static) will have risen by an amount pv above the
pressure in the undisturbed air stream. pv in Pa is exactly related to
the air density, ρ kg/m3, and velocity, v m/s, in the undisturbed
stream by the relation pv = ½ρv2.
Air flow and pressure in ducted systems 77
pv is called the velocity pressure and is an important quantity to which all

the pressure and drag effects of a moving airstream can be related. It is the
velocity of the body relative to the undisturbed air which counts. It does not
matter whether moving air is slowed down from v to zero at the stagnation
point on a stationary body, or whether still air is accelerated from zero to v at
the stagnation point on a moving body. In both cases the change in velocity
is v (strictly -v or +v) and in both cases the rise in pressure is the velocity
pressure, ½ρv2, which is always positive.
6.1.4`Total pressure, p,
At any point in the air the sum of the static pressure and the velocity
pressure is called the total pressure.
We must now observe that in order to raise the pressure (absolute or
static) of a sample of air we must do work on it; in other words the air must
absorb energy. In our example we must work with the lungs to raise the
pressure in the balloon. Conversely, if the air pressure falls it must give out
energy. Limiting ourselves to moderate static pressure changes for which the
air can be considered incompressible we can neglect changes in density and
temperature and forget about the corresponding complications of
thermodynamics. It is then found that the change in static pressure is equal
to the energy absorbed or given out by unit volume of air. Dimensionally we
can easily see, thanks to SI, that this is right.
Next we remember that the ordinary formula for the kinetic energy of a
mass m kg moving with a velocity v m/s is ½mv2 joules. Therefore the kinetic
energy of air which has a mass of ρ kilogrammes per cubic metre is ½ρv2
joules per cubic metre when it is moving with velocity v. This is also its
velocity pressure in Pa and, dimensionally:
We can now see that the formula at the end of this section in terms of
pressures has a corresponding expression in terms of energies:
Here total pressure and total energy both have the restricted, relative,
meaning of their definitions. They are not "total" in an absolute sense since
the basic pressure and energy of the atmosphere are excluded.
6.1.5 Pressures at a duct cross-section
The pressures considered in earlier sections have been defined as applying

to a point in an air stream. In practice average values are required which can
be applied to the whole volume flow passing through successive
cross-sections of a duct or system of airways.
These are not easy to measure, or even define, in the immediate
neighbourhood of an airway element such as a fan or a bend, liable to create
a chaotic pattern of air velocity, both in magnitude and direction. However,
after traveling a short distance along a straight duct, the departures from the
axial direction will have been largely removed and the static pressure will
have become uniform over the whole cross section (unless the flow is rotating
as a whole). A good average value of static pressure pr can then be measured
at a simple hole in the duct wall.
The equations at the head of this section show how values of V pv and pt
may be defined in terms of the cross-sectional area A and volume flow Q. If
the velocity were uniform and axial over the whole area these would be true
average values, and the corresponding kinetic energy of the flow would be
Q½ρV2. In fact the air velocity, even if axial, is far from uniform, rising from
zero at the duct wall to a maximum substantially in excess of V. It follows
that the true kinetic energy of flow must exceed Q½ρV2 since the local
energy flux is proportional to qv2, or v3, and the average of a number of cubes
must exceed the cube of their average.
Air flow and pressure inducted systems 79
However this excess kinetic energy, and additional excess associated

with any rotation of the flow, are by convention, ignored. This simplifi-
cation is justified because the excess velocity pressure components are
unlikely to contribute to the useful work of circulating air round the
system.
6.1.6 Total pressure drop and fan total pressure

In Section 6.1.3 it was stated that air experienced a static pressure rise
of ½ρv2 when slowing down to rest at a stagnation point. The velocity will
have fallen from v to zero and the velocity pressure from ½ρv2 to zero,
so it follows that the total pressure, which is the sum of the static and
velocity pressures remains the same. The total energy per unit volume is
therefore unchanged, which means that the process was loss-free, the
whole of the kinetic energy being transformed into static energy. More
generally, velocity changes will not be loss-free. Some energy will be lost
(or rather turned into heat which is not significant when the fluid is
incompressible) and the total pressure will fall by the total pressure drop:
K depends on the geometry of the system and ½ρV12 is the con-

ventional (see 6.1.5) velocity pressure at the beginning of the section
considered. ½ρV22may be chosen if it is the larger, but which must be
made clear when defining K. Total pressure falls in the direction of flow
from 1 to 2, corresponding to the loss of energy which always occurs
whether v falls, rises or remains the same. If for example, a duct is
expanded to double its original area through a cone of 30° included
angle, 20% of the initial kinetic energy will be lost (see Fig. 14.7 b). For
this process, therefore, K is 0.2 and the total pressure drop from inlet (1)
to outlet (2) of the conical expander will be:
With simple incompressible flow the only exception to the rule that
total pressure falls in the direction of flow will be a fan. A fan does work
on the air and the total pressure will rise from fan inlet to fan outlet by an
amount called the fan total pressure, which will be considered in more
detail later.
6.1.7 Static pressure difference

The pressure difference between two cross-sections of a duct system,
1 followed by 2 in the direction of air flow is ps2 - ps1 positive if the
pressure is rising, negative if it is falling. Note particularly that the
pressure drop ps1 - ps2 is always opposite in sign to the pressure
difference. From the relation of Section 6.1.4 we may write the static
pressure difference as:
The second part of this expression is the total pressure drop which will
always make a negative contribution to the static pressure difference.
The first part is the change in conventional velocity pressure, with a
negative contribution to static pressure difference if the air speeds up

between 1 and 2, a positive contribution if it slows down.
Thus the static pressure can, unlike the total pressure, rise in the
direction of airflow. In the example of the conical expander (Section
6.1.6) V2 = ½V1, and K = 0.2, giving a net rise in static pressure - a
process known as diffusion.
6.1.6 Relative nature of ps pv and pt

The absolute pressure pa and the density ρ are the only true physical
properties possessed by the air. All the rest are abstractions convenient
for calculation but variable according to the frame of reference adopted
for the problem in question.
In the great majority of cases the frame of reference will be the still
atmosphere. Then velocities will have their ordinary meaning relative to
stationary objects, and static and total pressures will be measured above
or below atmospheric pressure.
Sometimes, however, a different choice is to be preferred. Suppose, for

instance, that the ventilation system for a railway carriage is to be studied. The
atmosphere traveling inside with the carriage will be the frame of reference for
the small velocities and pressures of internal air movement. The outside
atmosphere then forms a relative wind at train speed. The resulting external
pressure distribution can be studied on a model in a wind tunnel to help in
locating inlet and outlet openings and considering leakage - which will jointly
set the carriage pressure in relation to the outside atmosphere.
In this way the system pressures of the order of 100 Pa which generate
internal air circulation can be isolated for calculation from external pressure
differences which may exceed 1500 Pa.
6.2 Pressures in a Ducted Fan System

Fig. 6.2 illustrates most of the definitions of pressure which have just been
discussed. The heavy lines S and T follow the changes of static pressure and
total pressure respectively from inlet to outlet. po is the atmospheric pressure
and p the absolute pressure in the duct, both measured from an absolute
zero far below the bottom of the page. The other quantities can be identified
by the arrows on the diagram as:
Following the lettered stages through the system with reference to the fig.
numbers where numerical data will be found
a. Air velocity increases from zero in the free atmosphere towards the duct
entry. There is no loss, so p, is constant at its free atmosphere zero
value. As V increases ps falls with ½ρV2.
b. Small drop in pt corresponding to entry loss.
c, e. pt falls gradually at friction gradient.

pv is constant and ps = pt – pv.
d. pt rises by the fan total pressure. In the diagram ps rises by the same
amount, but this is only because inlet and outlet areas are equal in the
case illustrated. Note that the change in ps is never equal to the "fan
static pressure", for the definition of which see Chapter 7.
f. When the duct area is suddenly enlarged, pt will fall corresponding to the
loss of energy. The nominal fall is sharp, as shown by the full line. The
broken line illustrates the lengthwise distribution of energy loss which in
fact occurs. ps will rise sharply, both nominally and actually as measured
by the wall pressure. This, "static pressure regain" arises because the
whole of the drop in kinetic energy from e to g is not lost.
h. The drop in pt here represents "useful work" done in overcoming the

flow resistance of a necessary element in the system-for example a
heater battery. ps drops by the same amount because Vg has the same,
low, value both approaching and leaving the battery.
j. The air is accelerated with little loss of pt but with a large fall in ps,
corresponding to the large rise in pv to ½ρVk2.
k. pt and ps fall with pressure gradient corresponding to the velocity in duct

k - which is greater than that in duct c.
I. The gradual increase of duct area is accompanied by a moderate energy

loss and drop in pt together with a substantial "static pressure regain" in
ps associated with the large drop in V. This is an example of a diffuser
deliberately inserted to improve the efficiency of the system.
m. At any system outlet to the free atmosphere, the whole of the kinetic
energy of the flow will be lost. Thus, in contradistinction to the inlet a,
where ps fell and pt remained zero, at an outlet pt will fall from its value
just before the outlet to zero while ps will reach zero before discharge at
m, and remain zero.
n. The broken line again distinguishes the actual loss of energy along the
outlet jet from the sharp fall in nominal pt at m.
6.3 Design of a Ducted System

An orderly procedure is necessary and is best discussed in relation to an example.
A building ventilation system will be chosen for this purpose.
(i) The volume flows to be supplied to and extracted from each room in
the building will first be determined, as discussed in the preceding
chapters.
(ii) The building site will next be considered to settle the best areas for intake from
and discharge to the atmosphere, avoiding the dangers of recirculation and
excessive wind and weather exposure.
(iii) The choice between unitary or central systems will be made, and whether one
or more plant rooms are needed according to the requirements for
accommodation, maintenance and noise control of the equipment. The plant
will be sited as centrally as possible to minimise the distance air has to be
moved.
(iv) The building plans will be studied, and the building itself if it exists, to find the
best location for room inlets and outlets and the duct runs joining them to the
fans and plant. These must meet the requirements for good room air
distribution, while keeping the duct layout as simple as possible with room for
gradual changes of duct area, shape and direction.
(v) Having settled the route and flow rate for each section of main and branch
duct, the size and shape of each grille, duct, bend and section change must be
determined, which is the subject matter of this chapter. One objective is to find
the right balance between running costs-represented by the total pressure
drop, assuming that maintenance snags have been avoided-and first cost,
which tends to be higher the lower the loss of each system element.
(vi) There are several ways of approaching duct system design. The simplest is the
velocity method, which involves selecting main and branch air velocities on the
basis of experience. This is probably worth doing in any case as a first step,
before moving to more refined methods. Trial and error is an almost inevitable
feature of design, and while this can be systematised, and handled by
computer programme, it remains important to understand the successive
steps.
Two factors influence velocity selection. Firstly velocities must fall as the size of
duct is reduced to avoid increasing pressure gradients. Secondly, noise generation
increases rapidly with velocity at grilles, bends, and other fittings where the flow
separates from the walls, leaving turbulent eddies in its wake. Table 6.2 gives some
typical recommendations. High velocity systems require noise control by means of
sound absorbent units between the duct system and room outlets or inlets. They
are mainly to save space in city-centre buildings, where every available cubic metre
commands a high rental, offsetting the additional running cost.
The higher values apply to the larger ducts, to areas less sensitive to noise, and
to locations farther from the occupants. Inlet and outlet grille velocities may be
influenced by a noise control survey.
6.4 Estimation of Total Pressure Drop

While the resistance of a ducted system can be estimated in terms of static
pressure changes, the total pressure method is adopted here. This avoids the
use of the somewhat confusing "static regain" concept, and is logical in that
total pressure drop is a true measure of energy loss. When using total
pressure it is important to remember that the loss of total pressure at system
outlet must always be included. This applies whether the air leaves through
an outlet grille into a room, leaves through weather louvres to the
atmosphere or is directly discharged by the fan itself. In each case there is an
item of total pressure drop equal to the velocity pressure corresponding to the
average velocity at outlet.
The loss in each element of the system (bend, duct length, etc.) is
dependent on the average velocity through it, which is taken as:
From this velocity, and the air density, ρ(kg/m3) the conventional velocity
pressure is determined:
This is then multiplied by a factor, K, and the result is the total pressure
drop from the inlet (1) to the outlet (2) of the element. Since we are calling it
a pressure drop, we can ignore the negative sign which it would have as a
pressure change (from high to low in the direction of flow).
Typical values of K applying to V1 or V2 as specified have been established

for the commoner system elements, and a number are given in Sections 6.6.1
to 6.6.5. Different authorities do not always agree on the values, but
exactitude is not to be expected. The fact is that the velocity distribution
across the inlet cross-section affects the loss in each element, which in turn
affects the element next following. In a practical duct system an overall
accuracy of ± 10% in the total pressure drop estimation would be very good
indeed, and would only cause some
± 2 or 3% variation in the volume flow handled by the fan. The elements of total
pressure drop are added together round the system, and may be classified as
follows
Losses at entry to the system from atmosphere. Losses due to

friction in duct lengths.
Losses at changes of duct area or shape.
Losses at bends and changes of direction.
Losses at division of flow into branches.
Losses caused by obstructions, grilles and louvres. Losses in
filters, heaters, and other "useful" elements. Losses at
discharge from the system to atmosphere. Change in
atmospheric pressure from inlet to outlet.
The last item brings out the fact that a ducted system should be regarded as a
complete circuit. Of course the air entering the system is not (or should not be)
the same air that leaves the system outlets. But it is flowing at exactly the same
volume (strictly mass) rate and can be looked upon as ultimately circulating
round at almost zero velocity through the almost infinite area offered by the free
atmosphere.
The data in section 6.6 are derived from a number of sources, the most
important being the research results and survey published by BHRA (Internal
Flow - a guide to losses in pipe and duct systems. D. S. Miller, British
Hydromechanics Research Ass. Bedford). Slight liberties have been taken in
smoothing and extrapolating the data in the interests of simplicity and clarity, but
this is a field where differences of 20 to 25% between one authority and another
are by no means unusual.
The K values given apply to medium to large volume flows at medium to high
velocities-these being, of course, the conditions under which the resulting energy
losses will be most significant. In smaller systems, as judged by the Reynolds
number criterion (see 13.5.2), K will be greater. The factor by which it should be
increased is roughly as in the following table for low-loss bends and similar
components; the increase is less for sharp bends, and in other cases where the
flow separates from the walls.
The table shows the effective diameter, Demm, corresponding to each

Reynolds number with atmospheric air at 5 and 15 m/s velocity.
6.5 Discussion of System Resistance Elements

6.5.1 inlets. A plain intake should always be avoided unless the velocity is very
low - as for a filter. Flanging will nearly halve the loss, but a 60° cone D/6
long is best for the usual installation. Curved bell-mouth
intakes are the optimum, but require tooling and are mostly limited to
proprietary equipment, such as fans.
6.5.2Outlets. Always remember to include one velocity pressure for the final
discharge to atmosphere - whether from a duct or a fan. This velocity
pressure is already included in Fig. 6.8 and the tables for outlet louvres and
cowls to show the overall effect of outlet area selection in these cases.
6.5.3Sudden expansion cannot be improved upon for area ratios up to 1.3.

At higher ratios diffusers should be considered.
6.5.4 Sudden contraction does not produce much loss at area ratios down
to 0.8 but does generate separated flow and some noise. A 60° or less cone
contraction eliminates these defects, and is particularly necessary at the inlet
to a fan.
6.5.5 Orifices produce substantial loss, and

will only be used for a specific purpose such
as flow measurement or flow control. As
sketched here the air stream contracts after
passing through the hole, raising the velocity
up to 1 -66 times, the increased velocity
pressure being substantially lost. Rounding
the inlet side would reduce the loss, but in
fact the pressure drop is usually the whole
purpose of inserting the diaphragm, damper
or valve. Much the same drop will be
produced with single holes of any shape
having the same area ratio.
6.5.6 Guards must have mesh sizes meeting the safety requirements and the
wire or bar of which they are made must be sufficiently robust. Once these
needs are met any further reduction in free area will cause unnecessary loss.
Locate the guard in a low velocity section if possible. Wire mesh or perforated
plate designed to make K = 2 is used in test airways to produce a velocity
distribution as uniform as possible over the cross section.
6.5.7 Obstructions. Every effort should be made to avoid passing pipes or

struts through air ducts. If it cannot be avoided streamlining may be
considered. A simple rounded nose and 20° wedge-shaped tail may be
sufficient.
If a conflict cannot be avoided (a failure of overall planning) it may be
better to bypass the obstruction provided gradual expansions or limited
changes of direction can be achieved.
6.5.6 Filters, heat exchangers and the like should have pressure loss data
provided by the supplier. Filters in particular show a K value decreasing as the
air velocity rises, since the pressure drop does not increase as fast as the
velocity squared - an effect of the low Reynolds number in the material.
6.5.9 Diffusers joining two ducts are usually required to expand through a
specified area ratio while the length is at the designer's choice. The loss falls
with increasing length to a minimum (K = 0.06) with an included angle of 7°
or less. A 7° diffuser is very often too long in practice, bearing in mind that 2
to 4 diameters of uniform duct are needed after the diffuser for full recovery.
If the limited length (L/De) available (or economical) places the design in the
region of Fig. 6.7 above the broken line, it is better to truncate the diffuser as
illustrated in the example. Note that diffusers with an included angle more
than 25° are rarely worth while - sudden expansion is nearly as good.
While the total pressure falls, the static pressure rises from inlet to outlet
of a diffuser. This rise is called the static pressure regain.
6.5.10 Outlet diffusers save energy by reducing the velocity at which the
air is finally discharged. If they expand at too steep an angle the extra gain in
velocity pressure will be more than offset by extra diffusion losses. Generally
speaking the designer will need to limit the length of the diffuser for cost
reasons or because space is limited - a diffuser outlet should be at least 1½
times its small diameter from a wall. He can, however, choose the outlet area,
and Fig. 6.8 shows the area ratio for minimum loss as a function of length.
The figure also gives the corresponding loss and included angle.
6.5.11 Bends are a prolific source of unnecessary loss - and unwanted

noise. As shown by the high K values a sharp throat is the worst feature - it is
good practice never to make the throat radius less than half the width of the
duct. If there is no room for a decent throat radius the loss may be reduced
by fitting various types of vane to organise the direction change of air. The
high loss of a sharp bend is caused by separation of flow at the throat, and
unless a length of duct follows the bend, none of the high velocity pressure
thus generated will be recovered - resulting in a spectacularly large outlet
loss.
6.5.12 Interaction between bends which are close to one another affects the
loss in the second bend. See 6.6.6. Provided the bends have a throat radius
of at least half the width the results are favourable - a reduction in total loss.
Sharp bends on the other hand must be spaced at least two diameters apart,
or else the losses will be greatly increased for the reasons discussed in the
last paragraph. To insure against under estimating the losses, diffusers and
expanders should be spaced three diameters from bends.
6.5.13 Branches. Where flows combine or divide at branch junctions losses

occur affecting both main and branch flows. A study of Figs. 6.11 to 6.14
shows the advantage of entering or leaving the main flow with a 45°
direction change only - or, if a right angle branch is necessary, by a well
radiused or chamfered junction. Under these circumstances combining flow
causes little overall loss, there being rather a transfer of energy from branch
to main or main to branch according to which has the higher velocity. In a
dividing branch the loss accompanies
velocity reduction, and it is best to design for roughly equal velocities. The
same applies to Y-type flow divisions.
For minimum dividing branch loss it is best to use a rectangular duct in
which branches are taken off with half-throat-radius bends of constant height,
while the main duct area is proportionately reduced at each branch. Velocity
reduction by more than 10% should be avoided both in branch and main,
though acceleration by 30% or more is quite safe.
6.5.14 Wind pressure must be taken account in system design in exposed

locations-which include high buildings. Air taken in on the leeward side and
forced out on the windward side will encounter an adverse pressure
difference which should be included as an item of total pressure drop.
Locations can be found on the roofs of some buildings where the air pressure
is always much the same as it is on the leeward side. These are clearly
favourable for system outlets since the wind pressure will make a zero or
negative contribution to the system pressure drop, and may therefore be
neglected. See Section 4.4 for wind pressure estimates.
6.5.15 Straight lengths. A short length, L metres, of straight duct may be

roughly estimated for total pressure drop Pf by the use of the following
approximate values of the loss factor K in the basic formula.
However, when the duct is long enough to make a significant contribution

to the total pressure drop of the system, account must be taken of the
variability of the friction factor, f, in the formulae which should really be
written.
The friction factor f varies in quite a complex manner with the gas velocity,
the properties of the gas, the diameter of the duct, and the effective
roughness of its walls. The general case, applicable to long airways in mining,
tunnelling and industrial process work, is discussed in Chapter 13. For
ventilating, and most air handling work it is adequate, and much simpler, to
use a chart which takes into account the variations of f with duct diameter
and velocity when atmospheric air is the gas flowing.
Fig. 6.17 is such a chart, drawn up to give the pressure gradient directly
with air of standard density, = 1.2 kg/m3, and with wall roughness, ε µm
(micrometres) representative of typical ventilating duct
construction. Full details of the basis and use of this chart are given, with an
example, on pages 98 and 99. Corrections are given for density - exact when
the gas is predominantly air - and wall roughness. The roughness correction is
approximate only, but then so, usually, is the evaluation of e. The chart is
plotted from the formula:
6.5.16 Equivalent diameters

In order to use Chart 6.17 for ducts of other than cylindrical crosssection
an appropriate value must be found for D. This is not simply the value giving
the same area. It can be shown that a dimension Dh called the hydraulic
diameter, should be used in the basic equation, and will cover cross-sections
of any shape.
This is still not the value to use in the chart. What is wanted is a genuinely
equivalent cylindrical duct which, if substituted for the rectangular or other
duct, would pass the same volume flow, Q, under the same pressure gradient.
The area and velocity for such a duct would be different, and so therefore
would the friction factor f. Allowance for this can be made (see Chapter 13) by
the somewhat awkward formula below under the same conditions of standard
air and wall roughness. Use of the formula can be avoided by referring to
Chart 6.18 which gives, for a rectangular duct a x b mm, an effective diameter
Dq mm transferrable directly across to chart 6.17.
6.6 Losses in Duct System Elements

For each system element the drop in total pressure Pf equals
V (m/s) is the average velocity in the section indicated in each diagram.
K includes allowances for the influence of the component on duct losses

downstream, and for the duct length occupied by the component.
The system loss is the sum of all the K½ρV2 items.

6.6.3 Losses in diffusers

Fig 6.7 gives the loss factors for expanding diffusers joining two ducts of
areas A1 and A2. De is the effective diameter of the inlet duct, defined as
follows for various expander shapes:
The downstream duct should be at least 4D2 long, for full pressure
recovery.
Truncated Diffusers. If the length available for a diffuser is limited, it is

better to use the whole length L to
expand to the area Ax given by the area ratio
at the intersection of the broken line Kx with
the appropriate L/De ratio, and then to expand
abruptly to the downstream area A2 assumed
greater than Ax. The overall loss factor can
then be taken as:
Example. A length L = 2.4m is available to expand from 1.6m diameter (A1

= 2.02M2) to 2.7m (A2 = 5.71m2)
From Fig 6.7 (point A) K = 0.25 if the diffuser expands to 2.7m.
If a truncated cone is adopted the intersection X shows that Kx = 0.12 and

the area ratio is 2.2. Therefore Ax = 2.2 x 2.02 = 4.44M2 and Dx = 2.4m.
Open outlet diffusers. For a given length L there is an optimum expansion

ratio which will minimise loss and maximise static pressure regain. This area
ratio and the corresponding loss factor K are given by the full line curves in
Fig 6.8 for normal duct approach flow.
An exceptionally uniform approach velocity (as in a venturi-nozzle flow
meter) allows the more rapid expansion and lower loss given by the broken
lines.
For an optimum diffuser at fan outlet the manufacturer should be
consulted.
6.6.4 Losses in bends

The bend loss Pf = K.½ρV2 is to be
added to the duct friction loss calculated
up to bend entry L1 and on from bend
exit L2.
The ratios R/D and R/A plotted relate
to the Throat radius R shown here. Some
authorities measure radius to the centre
line of the bend, quoting a radius ratio
equal to:
For a rectangular duct A x B, A is the

dimension in the same plane as R.
Losses depend on the aspect ratio B/A
and are higher when A is the larger
dimension.
An outlet duct length at least equal to
D or A is required for pressure recovery
after the bend. If this is omitted, i.e. if
the bend is open outlet, the value of K
may be multiplied up to two and a half
times.
When space does not permit the use
of a generous throat radius, the
following vane constructions may be
considered.
Turning vanes of thin sheet
circular arc form, spanning a mitre
bend in a round or rectangular duct.
Splitter vanes in a rectangular
bend of small throat radius R.
One splitter has a radius =
Two splitters have
and
respectively.
Aerofoil vanes of solid or hollow

construction in a rectangular bend. (n-1)
vanes will form n passages each with a
throat to width ratio of 1.0. The throat
radius of the main bend must be A/n.
Each vane has an inside radius of 2A/n
and an outside radius of A/n plus straight
extensions.
K for bends turning through angles other than 90° can be expressed
approximately as a percentage of the 90° value.
6.6.5 Losses at branches

These loss factors apply to round or rectangular ducts with constant area
main. They depend on the branch area ratio A1/A3 and the branch velocity
ratio V1/V3 - therefore on the volume flow ratio Q1/Q3 = A1V1/A3V3. Figs. 6.12
and 6.14 apply to the 45° branch case, but right angle branches with 0.5 or
more throat radius or chamfer will be nearly as efficient. Y- and T- flow
division is dealt with in Fig. 6.15.
6.6.6 Influence of one bend on another

The overall loss of two closely spaced, well radiused, bends is less than the
sum of their separate losses, unlike most inter-action effects which are
unfavourable as are those for sharp bends in arrangements (a) and (b).
Multiply the sum of the losses in the two bends and the duct between them by
C from Fig. 6.76. Correction for angles other than 90° can be used for type (a),
but not for type (b) or (c) configurations.
6.7 Duct Sizing in Branched Systems

The equal friction method is often used for ventilating systems in which the
ducting is extensively branched. The basic idea is that branches with only a
fraction of the main duct flow should be operated at lower velocity so that the
pressure gradient would be everywhere the same. A pressure gradient of 1
Pa/m makes a good first trial for an ordinary low velocity system. The method
will be readily followed from the following example, making direct use of Fig.
6.17.
The calculated diameters would, of course, be rounded off or converted into

reasonably equivalent rectangles. For example, to keep main and branch ducts
of common height we might choose, using Fig 6.18:
with the main ducts tapering in width as the m3/s fall. The ratio of branch
velocity to main velocity is, however, low for efficient junction and it would be
better to bring them into closer agreement. See Fig.6.13.
In a branched system the total pressure drop is usually calculated for the run
to the farthest branch outlet or inlet. The total pressure drop to a nearer branch
would be less, and as equal volume flow at all outlets or inlets is probably the
design objective a damper would be needed in each branch to absorb the
excess pressure. The system would be "balanced" after installation, the damper
in the farthest branch being set fully open and the remainder closed by trial
until equal volume flow is achieved at each.
A static regain method is sometimes recommended for long supply ducts with
many take-off branches. The idea is to maintain constant static pressure at all
take-off points in the main duct thereby minimising the damper losses. The
main duct is resized for a lower velocity after each take-off and it is assumed
that about 25% of the fall in velocity pressure will be lost at each junction, the
remaining 75% being "regained" as a rise in static pressure. The velocity
reduction is chosen so that the regain just balances the loss of total pressure at
the junction plus the duct friction of the section.
Fuller data on branch and junction losses is now available, see Figs. 6.11 to
6.15, and we may judge by how much (if at all) the main velocity may exceed
the branch velocity without incurring excessive junction loss. See the last
paragraph of Section 6.9 for an example of satisfactory balance.
6.8 Influence of Density on Total Pressure Drop

It is a good plan always to calculate the total pressure drop in an airway
system, Pr first on the assumption that it is handling air at 1.20 kg/m3
standard density. This will give the correct volume flow rate irrespective of
the actual density (or gas) when used in conjunction with the standard fan
characteristic as described in the next chapter. The reason is that both fan
total pressure and total pressure drop are equally proportional to density at
constant volume flow, and therefore remain in balance as density changes.
At any density, ρkg/m3, the total pressure drop is:
Bearing in mind that Pi estimation to plus or minus 10% is good, the

influence of density can be ignored in the majority of systems at low and
medium fan pressure which are open to the atmosphere. Correction is required
with high pressure fans, at high altitude where the barometric pressure is low,
and in closed circuits containing air or other gas, probably at non-standard
temperature and pressure. In these cases the density can be calculated from
one of the following formulae:
In these formulae it is sufficient to take the temperature t°C and the absolute
pressure pa pascals at the average of their values round the system. Remember
that pa = po + ps for circuits open to the atmosphere, po being the barometric
pressure in pascals ( = 100 x millibars). Also that the pressure in a closed circuit
may be quoted as a "gauge pressure", in which case the barometric pressure
must be added to obtain pa.
For gases other than air ps is the value at standard temperature and pressure
relative to air as given in Table 14.11. The standards are conventionally po =
101,325 Pa (760mm of mercury) and 20°C. For ease of calculation they are
here taken as po = 100,000 Pa and t = 16°C, at which all densities remain
standard. For precise density calculation, taking into account humidity, see
Chapter 10.
Example 1: An air conditioning plant in Johannesburg (altitude 1700m)
circulates air at 25°C through a high velocity supply system with a total
pressure drop of 2,000 Pa, calculated at ρ = 1.2 kg/m3.
The normal atmospheric pressure at 1700m is given in Table 14.5 as 82,500
Pa. The average static pressure in the system can be taken as:
Further correction for average ps = 810 Pa and pa = 83,310 Pa is not worth

making.
Example 2: A process plant circulates carbon dioxide at a gauge pressure
quoted as 1.5 kgf/cm2 at 120°C. The total pressure drop as calculated at 1.2
kg/m3 is 3,000 Pa. What is the pressure drop under working conditions?
1.5 kgf/cm2 = 15,000 kgf/m2 = 147,000 Pa gauge pressure. Taking po =
100,000 Pa barometric pressure gives pa = 247,000 Pa absolute pressure,
which will be assumed to be the average round the system.
Table 1411 gives the density, ρs of carbon dioxide as 1.53 relative to air at
standard temperature and pressure.
6.9 Example of System Pressure Drop Calculation

Fig. 6.19 illustrates a ventilation system to supply fresh warm air to a
factory building. A total volume flow of 16m3/s is to be supplied at 2m3/s
each through eight overhead outlets in line of which four are shown. The
fan proposed is a 1250mm 720 rev/min adjustable pitch model. What fan
total pressure is required?
The results of the successive stages of calculation are tabulated below the
diagram. At each section the following are determined: the area A - the
intended volume flow Q - the mean air velocity V - the corresponding
velocity pressure pv - the length L of the section, where applicable. The loss
factor K or gradient g is then estimated from the information in Section 6.6
and multiplied by pv or L to give the contribution of the section to the overall
pressure drop. The steps in this process are commented on, line by line,
below.
(a) Proprietary weatherproof louvres are proposed with a rated loss of
50% of pv in the duct.
(b) Fig. 6.7 applies. A2/A1 = 3.60/1.96 = 1.84.
De = 2 x 1.4 x 1.4/(1.4 + 1.4) = 1.40m. Therefore L/De = 0.54 and K =
0.15.
(c) From the rules for a truncated diffuser in Section 6.6.3 the additional
loss factor will be (1.96/3.60)2 - (1.96/6.25)2 = 0.20.
(d) The pressure drop across the battery is quoted by the makers as 57
Pa for a face velocity of 2.56m/s.
Air 1/ow and pressure in ducted systems 103
(e) The fairing is that provided for an open inlet fan and any loss is
allowed for in the fan performance.
(f) K would only be 0.01 for a 2½% sudden contraction without change
of shape (Fig. 6.3). Allow a nominal K = 0.05 for the transition from round
to rectangular.
(g) Fig. 6.18 gives Dq = 1200mm as the equivalent diameter of a 1000 x
1200mm section. Fig. 6.17 gives g = 1.4 Pa/m at Q = 16 and D = 1200.
1.4 x 20m = 28 Pa.
(h) Fig 6.10 applies. R/A = 0.6/1.2 = 0.5. B/A = 1.0/1.2 = 0.83.
Therefore K = 0.26.
(i) As item (g) g = 1.4 Pa/m and 1.4 x 5m = 7 Pa.
(j) The data in Section 6.6.5 for low loss right angle tees applies. V1/V3 =
12.3/13.3 exceeding the minimum specified as 0.9. r/b = 0.3/0.65 is
slightly better than the 0.5 specified. At Q1 /Q3 = 8/16 = 0.5, K31 = 0.2.
(k) From Fig. 6.18, 650 x 1000 gives 880mm Dg. From Fig. 6.17, Q = 8
and D = 880 gives g = 1.8 Pa/m and multiplication by 4m gives 7Pa.
Similarly:
(o) From Fig. 6.10 at A = 0.3m. B = 0.65m, taking the 0.15m

chamfered bend as equivalent to 0.5 R/A with B/A = 2.2 gives K =0.20.
(q) A proprietary adjustable pressure loss outlet diffuser is to be
used with a minimum drop equal to 50% of the approach velocity
pressure. We can check that this is suitable for equalisation of branch
volume flow with the aid of Section 6.6.5. Fig. 6.14 applies to the
present case with the following results for successive junctions, the Pv
to which K31 applies being 63 Pa in all cases.
We may note in passing that we were correct to ignore junction losses in

the main duct, K32 being negligible for all cases in Fig. 6.14. This conclusion
is reinforced by the successive reductions in main duct area, which reduce
the decelerations which would be the cause of any loss.
A table can now be constructed to determine the pressure drop to the
outlet of each branch. The volume flow must remain the same as assumed.
Pressure differences are taken up by setting the outlet diffuser for increased
drop, the minimum of 0.50 pv1 = 32 Pa occurring at junction 2.
CHAPTER 7
Fans and fan

performance
Fans are built in widely differing types and designs, with

performances just as various. Yet, in theory at any rate, one
and the same duty could be met by fans to any of these
designs, given the right size and speed. This chapter deals
with those features of performance and construction which
direct the choice to a limited number of alternatives.
While it is not the purpose of this book to enter the
province of the fan designer, an elementary treatment of
the mechanism of pressure generation is given at the end of
the chapter. It can be passed over by those with a distaste
for even a little mathematics, and it is to be hoped that they
will be satisfied by the purely descriptive explanation in
Clause 7.5.
7.1 What is a Fan?

A fan is defined as a rotary, bladed, machine maintaining a
continuous flow of air. Continuous because the air flows steadily
into, through, and out of the fan, a feature distinguishing it from
positive displacement (piston, vane or lobe) machines, which
generally produce a pulsating flow. Afan has a rotating impeller
invariably carrying blades of some kind. These blades exert force
on the air, thereby maintaining the flow and raising the (total)
pressure.
105
The pressure rise is low or moderate in a fan, distinguishing it from a

compressor, where the pressure rise is comparatively great. There is no hard
and fast dividing line, though fans do not normally raise the absolute pressure
more than 30% (30,000 Pa fan total pressure with standard air at inlet).
Compressors and high-pressure fans are sometimes called blowers if they are
used to take in atmospheric air and raise its pressure, exhausters if they suck in
at low pressure, discharging at atmospheric pressure, but this distinction does
not correspond to any difference in the fans themselves.
A distinct subclass is described by the term circulating fan. This includes
ceiling fans, table fans and pedestal fans, and is distinguished by the absence of
a fan casing or of any barrier between the inlet and outlet sides. They, therefore,
generate no usable pressure, their function being to deliver a jet of air along the
discharge axis. This jet entrains a much larger volume of surrounding air which
circulates around the room in which the fan is placed. See Chapter 3 for a
discussion of jet action.
7.2 The Fan Characteristic

Fan performance cannot be adequately described by single values of pressure
rise or volume flow. Both quantities are flexible, but they have a fixed
relationship with one another, and this relationship is best defined graphically in
the form of a fan characteristic. Fig. 7.1 is an example; the volume flow rate is
always plotted along the base, with the fan total pressure (and other
performance quantities) as ordinates.
Fans and fan performance 107
One point on the characteristic can always be found at which the efficiency of
the fan is a maximum. This is called the best efficiency point-d on the diagram.
As well as providing the lowest power consumption for a given duty, operation at
this point usually secures the lowest noise level for that particular fan.
However, the fan can be successfully operated at other points on the

characteristic. The region of satisfactory operation should be defined by the
manufacturer as the guaranteed performance range, and such a range is
indicated on Fig. 7.1 by the heavy line part of the characteristic. Outside this
range the performance may be uncertain and is likely to be commercially
unacceptable for efficiency, noise or cost. Part of the characteristic may also be
positively forbidden because operation there would cause overloading or
undercooling of the electric motor fitted, or vibration associated with unsteady
flow.
7.3 The System Characteristic and Operating Point

Chapter 6 showed how to find the total pressure drop caused by the
resistance of a system of airways to the required volume flow rate. These two
quantities can be plotted on the same graph as the fan characteristic to show
the desired duty point A. In general the fan characteristic will not pass through
A-though it can be made to come very close when selecting from an adjustable
pitch axial or adjustable speed centrifugal series. What will happen if this fan is
installed in this system?
To solve this problem we assume that the loss factors K for the various items
in the system remain constant when the flow rate is varied. The losses of total
pressure are then proportional to the various velocity pressures, and each of
these, and therefore their sum when they are added together, Pr, is proportional
to the square of the volume flow rate. This relationship can be shown on the
graph by plotting the line marked "system characteristic" which passes through
the duty required at A and follows the "square law"
At the point B where the system characteristic intersects the fan characteristic
the operating conditions for both the fan and the system are met. B is therefore
the operating point.
7.3.1 Example
Consider the airway system of section 6.15 in the last chapter. The desired
duty point is at 16.Om3/s and 316 Pa, as plotted at A in Fig. 7.1. The square law
system characteristic is then drawn. The fan characteristic in this figure is for a
1250mm diameter adjustable pitch axial fan, running at 12 revolutions per
second and set at a pitch angle allowing 10% margin over the estimated total
pressure drop at 16M3/s. If the
system resistance is exactly as estimated, the operating point will be at B where:
It should be noticed that the mutual accommodation of the fan characteristic

and the system characteristic to one another has a strong stabilising effect on
the performance. Thus suppose that the total pressure drop in the example had
exceeded the estimate by 25%, becoming 395 Pa at 16.0m3/s - point C in place
of A. The dotted system characteristic through C cuts the fan characteristic at D
where the volume flow becomes 15.5m3/s at 370 Pa. This is a loss of only 7%
from 16.6m3/s, resulting from a system resistance error of 25%.
7.4 Fan Static Pressure and Fan Velocity Pressure

In Fig. 7.2 four more curves have been added to give a set of complete fan
characteristics. They are called complete when they run from free flow (zero Ps)
to no flow (zero Q). The parts outside the guaranteed performance range of Q
are given for information only, plotted from test results of progressively
increasing uncertainty.
The fan total pressure is divided into two parts: the fan velocity pressure,
which is a conventional measure (see 6.6) of the kinetic energy at fan outlet,
and the fan static pressure, which represents the work done in compression. The
fan static pressure is fully available to the user, but the fan velocity pressure
suffers from the handicap that some of it will inevitably be lost, although this
loss appears as part of the system pressure drop when calculated by the method
in the last chapter. At an open fan outlet, or at the end of a duct system kept at
fan outlet area all the way, the loss is 100% of the fan velocity pressure. If the
duct area is gradually expanded, or a diffuser (conical expander) fitted to the fan
outlet, the loss can be limited by careful design to 25% of the fan velocity
pressure. Most practical cases lie between these extremes.
In order that the user may know what he is getting, it is usual in the
English-speaking world to quote or plot both the fan total pressure and the fan
static pressure; the fan velocity pressure can then be found by subtraction.
Alternatively, one of the pressures may be given, together with the outlet
velocity or area from which the others can be calculated. In some European
countries, however, only one "fan pressure" is recognised. This is usually equal
to our "fan static pressure" for an open outlet fan or to our "fan total pressure"
for a ducted outlet fan. It seems to the writer illogical that 100% of the fan
velocity pressure should be credited to a fan connected to a duct, where some is
bound to be lost, and none to an open outlet fan when some can be recovered,
and no more need be lost.
Q1 is the volume flow rate at inlet density.

Pt is the fan total pressure, which is the rise in total pressure from inlet to
outlet, as measured in a standardised test circuit.
Pv is the fan velocity pressure which is calculated from the average forward
velocity over the gross fan outlet area.
Ps is the fan static pressure which is equal to the fan total pressure less the
fan velocity pressure.
W; is the power supplied to the fan impeller by the driving motor.

ήt is the fan total efficiency, which is the ratio of work done to work supplied,
calculated as follows:
multiplied by 100 if required as a percentage. At high pressures the

expression needs modifying for the effects of compression - see Chapter 11.
7.5 The Main Types of Fan

Fans can be broadly classified according to the path taken by the air
through the impeller and the resulting mechanism of pressure genera-
tion. All fans work by aerodynamic action in so far as this implies the action
of forces exerted by the impeller blades on the air.
The total pressure rise takes place wholly within the volume swept by the
rotating blades. Neglecting losses the total pressure is constant up to the
blade entry, and constant again downstream of the blade exit.
7.5.1 Axial flow. In a pure axial fan the

effective progress of the air is straight
through the impeller at a constant
distance from the axis. The primary
component of blade force on the air is
directed axially from inlet to outlet, and
thus provides the pressure rise by a
process that may be called direct blade
action. The blade force necessarily has an
additional component in the tangential
direction, providing the reaction to the
driving torque: this sets the air spinning
about the axis independently of its
forward motion.
7.5.2 Radial flow. In a pure centrifugal fan

the air enters the impeller axially, turns
through 90°, and progresses radially
outwards through the blading. The blade
force is mainly tangential, causing the air
to spin with the blades. The centrifugal
force resulting from the spin is now in line
with the outward progress of the air and
is the main cause of the rise of pressure.
There may or may not be an outward
blade force component adding pressure
by blade action.
7.5.3 Mixed flow. In a mixed flow fan the

air enters axially and turns outwards
through an angle which may range from
30° to 90°. An essential feature is that
the impeller blading extends over the
curved part of the flow path. In this
region the blade force has a component
in the direction of the arrow on the
diagram as well as a tangential
component and the pressure rise is
developed partly by blade action and
partly by centrifugal action.
7.5.4 Orifice flow. This is characteristic of a

propeller fan operating at zero or low fan
static pressure. The air enters the impeller
inwardly from all inlet-side directions, and
curves round into a parallel stream at the
vena contracta on the outlet side. The fan
pressure is wholly developed by blade
action. The total pressure rise required is
the velocity pressure in the vena contracts
plus any static pressure difference
between the inlet and outlet sides.
7.5.5. Cross flow. The action of cross flow

fans is radically different. A vortex is
formed and maintained by the blade
forces and has its axis parallel to the shaft
and near to a point on the impeller
circumference. The outer part of this
vortex is peeled off and discharged
through an outlet passage. An equal
volume of entering air joins the inward
flowing side of the vortex.
7.5.6 Outlet flow. The air leaving the

impeller carries with it kinetic energy
proportional to its velocity pressure. The
component of this in the spin direction is
unlikely to be effective when the fan is
installed. It can be recovered for use by
redirecting the velocity towards the fan
outlet. A volute casing does this for
centrifugal fans, and guide vanes for axial
fans. The casing may also be shaped to
allow a gradual enlargement of area as
the outlet is approached. This diffusion
does not increase fan total pressure, but
it does convert some fan velocity pressure
into fan static pressure.
7.6 Axial Flow Fans-General-Purpose

See Fig. 7.22 for performance comparisons with centrifugal fans.
7.6.1 The general-purpose range
Axial flow fans are well established for air conditioning and industrial process
applications of all kinds. Manufacturers offer standard ranges of general-purpose
fans commonly running from 300 or 400mm to 1800 or 2000mm diameter and
displacing 0.5 to 50m3/s at pressures up to 800 Pa or so. Standard units can be
operated in parallel or series to extend the range of volume flow or pressure
almost indefinitely.
Impellers usually have blades cast in aluminium alloy with crosssections of

aerofoil shape. As compared with curved sheet blades aerofoils can apply
greater force to the air, thereby increasing maximum pressure, and can
maintain better efficiency over a wider range of volume flow. Also by increasing
the thickness and curvature of the inner sections the blades can be made stiffer;
this limits flutter and allows the impellers to be run at higher speeds.
7.6.2 Pitch adjustment
The blades of general-purpose fans are often attached to the hubs by a
system which allows the pitch angle (see Fig. 7.3) to be set at any desired
value on assembly. This adjustable pitch feature enables the fan to be supplied
pre-set for exactly the volume flow required, eliminating the waste of power
inseparable from a range of fixed steps from which the next larger must be
chosen. Variable pitch or controllable pitch fans are special models in which the
blade angle can be continuously varied while the fan is running. They are
naturally a great deal more costly, and their use is discussed in Chapter 9.4. Fig.
7.4 shows the range of pressure and volume flow covered by a typical
adjustable pitch fan.
7.6.3 Tip clearance

Axial flow fan impellers are generally mounted on the shaft extension of an
electric driving motor. This is supported in a cylindrical casing surrounding the
impeller with the minimum practical running clearance, a typical value for this
(radial) clearance being 0-25% of the impeller diameter. Departures from the
designed tip clearance will affect mainly the peak pressure development, and,
therefore, the flow rate when the fan is operating close to peak pressure. The
following table is an illustration for a particular fan, and should not be regarded
as general.
At a normal operating point-say 85% of peak pressure - the loss of volume

flow would only be about 2% at double the standard tip clearance. It is unwise to
design to operate steep pitch axial fans much closer to peak pressure, as little
margin would be left for under-estimation of the system resistance.
7.6.4 Variants on the standard range

General-purpose axial fans are particularly compact units with straight through
flow and no wasted space. Accessories can be made available to modify the
performance or adapt to particular installations. This practice can be illustrated
from the Woods Aerofoil range.
(a) The standard casing normally has a flange

for inlet duct connection. When operating
without an inlet duct an inlet flare should be
fitted to guide the air smoothly in without
breaking away from the casing (see Fig.
6.31). To be effective the inlet needs a
length of at least 15% and a radial depth at
least 10% of the fan diameter.
(b) Guide vanes downstream of the impeller are

sometimes incorporated in the fan as
standard, sometimes as an optional extra.
They will certainly improve the pressure
development and efficiency (see 7.5.6)
though no general rule can be given - it
depends on the fan and on what is
connected to the outlet. Reduction of outlet
swirl is particularly important when discharg-
ing into along high velocity duct. Swirl
persists for great distances in such a case,
and can increase the pressure drop several
times.
(c) Upstream guide vanes cause the incoming air to swirl in the opposite
direction to the impeller rotation, serving to cancel out the swirl introduced
by the impeller. They are not often used since they improve efficiency less
than downstream guide vanes, and have an unfavourable effect on noise
level. However, they do cause a substantial increase in peak pressure and
will sometimes overcome a pressure limit set by a fixed motor speed.
(d) Diffusers will convert some of the fan velocity pressure into fan static
pressure, which is the only useful form at an open outlet. They will reduce
the system loss when used to connect the fan outlet to a larger duct or
chamber. Their performance in these roles is very much dependent on the
flow pattern at the fan outlet, and for this reason manufacturers will
generally quote the combined performance of the fan and a diffuser of
recommended dimensions.
(d) If two identical guide vane fans are run in series the fan total pressure
would be approximately doubled at a given volume flow, apart from
losses due to mutual interference, the fan velocity pressure remaining the
same. Two non-guide-vane fans will only produce about 50% more
pressure than one, since the second stage will be unloaded by the swirl
produced by the first. The effect may be looked upon as a reduction in
the rotational speed of the second impeller relative to the air entering it,
which is spinning in the same direction.
(f) Contra-rotating fans provide a better system of series connection. Each

impeller is driven by its own motor and they rotate alternately in opposite
directions. Thus each impeller cancels out the swirl of the one before and
guide vanes are unnecessary. Furthermore, the second, and all
even-numbered, stage impellers receive air which is swirling in the
opposite direction. This is equivalent to an increase in relative speed of
rotation and the fan total pressure of a contra-
rotating pair will be about two-and-a-half times that of a single stage.

The blades of the even-numbered and odd-numbered stages differ
fundamentally from each other since one must be designed as a
right-handed and the other a left-handed screw. Pitch angles are
generally adjusted so that impeller powers are equal around the best
efficiency point. This automatically secures an output flow free from
swirl.
(g) Reversal of the direction of rotation of an axial fan reverses the

direction in which the air flows. The performance of guide vane fans in
reverse is extremely poor, but non-guide-vane and contrarotating fans
will deliver 60% to 70% of the forward volume flow when reversed on a
given system. The reduction is due to the fact that the aerofoil sections
are operating tail-first and have their camber (curvature) in the wrong
direction. A truly reversible impeller can be built from standard parts by
rotating every other. blade through 180°. Half will then be running
nose-first and half tail-first, the volume flow being about 85% of normal
in each direction.
(h) Fractional solidity impellers can be assembled on a standard hub by

omitting some of the blades. Mechanical balance must, of course, be
preserved, but there is no need for the blades to be evenly spaced.
Peak pressure is reduced and the best efficiency point moves to a lower
pressure and volume so that the speed must be increased for a given
duty. This can be an advantage when the impellers are directly driven
by electric induction motors. Such motors have better efficiency and
lower cost at higher speeds-a point which can be particularly significant
with large low speed fans. Fig. 7.6 shows the performance range of fans
with 12 left- or right-handed adjustable pitch blades, which could be
assembled with 10, 9, 8, 6, 4, 3 or 2 blades, and multi-staged.
7.7 Axial Flow Fans-More specialised designs

7.7.1 High pressure axial fans such as that in Fig. 7.7 are designed with hub
diameters between 50% and 70% of the impeller diameter, compared with
30% to 40% for a general-purpose range of competitive cost.
Aerodynamically this reduces the pressure limitation set by the slowmoving
roots of the blades. Mechanically the short blades can be made far stiffer so
that the impeller can be run at higher tip speeds without danger of flutter.
The ratio of the annular flow area to the total blade area decreases, making
guide vanes or contra-rotation essential to recover the increased swirl
energy.
7.7.2 High efficiency justifies axial fan designs of relatively high first cost
when the power absorbed is measured in hundreds of kilowatts. Among
features distinguishing such designs from the general-purpose type are:
(a) Hub diameters of 50% or more to improve the aerodynamic balance of

the design from blade root to tip.
(b) Blade form designed specifically for the required duty. When dieforming is
not justified, this entails increased labour to provide a good surface finish.
(c) Aerofoil-section guide vanes, again designed specifically for the required
duty.
(e) Careful streamlining of the annulus passage, and fairing of bearing supports
or other obstructions. This may entail moving the driving motor right out of
the casing, introducing the necessary transmission elements to the impeller.
(e) Space for a long tail-fairing following the impeller hub and guide vanes to
maximise fan total pressure by conversion of annulus velocity pressure.
(f) Space for a long, gradually expanding diffuser to minimise outlet velocity
pressure, and maximise fan static pressure.
These measures may raise the peak fan total efficiency to 90% compared
with 80% for a good general-purpose model at optimum duty.
7.7.3 Small axial fans continue the general-purpose range downwards, but
are more often manufactured with plastic impellers and sometimes casings also.
In the 100 to 200mm diameter range the high pressure conformation is often
used to counter the 50 or 60 rev/s speed limitation set by induction driving
motors. In the aircraft field small axial fans of exceptionally high power-weight
ratio are made, using the high induction motor speeds possible on a 400 Hz
supply.
7.7.4 Low pressure axial fans are met with particularly in large sizes with
volume flows from 50M3/s upwards at fan static pressure from 100 to 200 Pa.
As an example they may be applied singly, discharging from the top of
evaporative cooling towers, or in multiple, circulating air across extensive banks
of heat exchange tubes.
Hubs are small and the blades long, slender, and few in number three, four
or six. Blades were at one time of timber, but now hollow glass-reinforced
polyester, or similar, mouldings, or hollow aerofoil sections from steel or
aluminium sheet are more usual. Guide vanes are unnecessary.
7.8 Propeller Fans

These may be regarded as a special type of axial fan designed to operate
without a casing, the impeller being situated in a hole in a wall or partition. The
fans are simple low cost units with broad-bladed impellers usually formed from
sheet metal. The blades are shaped to co-operate with an orifice flow pattern,
deflecting the air with the minimum flow separation or vortex formation. Design
may perhaps be considered as much an art as a science, though flow
visualisation with stroboscopically viewed smoke trails can be helpful.
The blade form is usually optimised for pressure differences across the
partition from zero to about 100 Pa. Above the designed pressure the flow
pattern changes drastically. The outlet jet assumes an expanding conical form
with reverse circulation at its core as sketched. Towards zero volume flow,
discharge is radially outwards, and the centrifugal mechanism is now responsible
for pressure development.
Propeller fans are quiet and effective for ventilation purposes, both supply and
exhaust. They are also used for unit heaters and similar applications where some
resistance is encountered. For these an experimental matching of the fan and
the unit is important since the pressure development and the flow pattern over
the heat exchanger are very dependent on the blade and orifice plate positions.
As more of the impeller projects on the outlet side of the orifice, the free flow
volume falls, because the inlet orifice flow no longer covers all the blade. At the
same time the pressure at low flow rises, because more blade is exposed on the
outlet side for centrifugal action. The free flow can be substantially increased by
rounding the orifice edge or fitting a rounded inlet ring. This is because the vena
contracta is expanded (see section 7.5.4) and less velocity pressure is required
for a given volume flow. Moderate pressure performance is also helped, but high
pressure development is impaired. If the rounding is enlarged into a true
7 22 Woods Practical Guide to fan Engineering
bell mouth and a short tunnel formed around the impeller, the fan becomes in
effect an axial fan, and is better served by an aerofoil section impeller.
The impellers of propeller fans are almost invariably mounted on the shaft of
the driving motor. The air flow cools the motor, which can be totally enclosed to
keep out dust. The impeller power rises rather sharply if the volume flow is
drastically restricted, and the motor could be overheated, particularly if on the
downstream side, where centrifugal flow starves it of cooling air. However,
propeller fans are not often used in systems where such excessive resistance
could arise.
Table and pedestal fans may be fitted with very similar impellers to propeller
fans. The orifice flow pattern is the same, though the zero fan static pressure
condition only has to be considered and no plate orifice or ring is needed.
7.9 Centrifugal Fans-General-Purpose

The general-purpose centrifugal fan has a volute or scroll-shaped casing as
illustrated in Fig. 7.12. The air enters through a circular inlet adapted for use
with either open inlet from the atmosphere or duct connection. The outlet is
rectangular, and intended for duct or diffuser connection only (without such a
connection the performance would suffer). Three or four alternative impellers
are often available, such as those described in 7.9.1, 7.9.2 and 7.9.3 with
differing performance characteristics. The impeller may be direct driven by an
external motor, but is more commonly belt driven for flexibility in speed
selection.
See Fig. 7.22 for performance comparisons with axial fans.
7.9.1 Forward-curved multi-vane type

For upwards of three-quarters of a century the multi-vane impeller has been
used for the established general-purpose type of centrifugal fan in small and
medium sizes. It provides the most compact and probably the quietest form, and
the most competitive in first cost. The chief disadvantage is a limited efficiency -
60% to 70% at most - and a steeply rising power characteristic towards free
flow, necessitating a relatively high-powered driving motor.
Small impellers in the 50 to 250mm diameter range are commonly made from
punched strip rolled into cylindrical form. Better performance is obtained in
larger sizes with individual blades which can be fully formed to conform to the
aerodynamic requirements. The blade crosssection is commonly a circular arc of
90° or so, the outside, trailing, edge
making an angle with the radius from 45° upwards, forward in the direction of
rotation. The air velocity relative to the impeller changes from backwards at
blade inlet to forwards at blade outlet with little change in magnitude. However,
the air leaves the impeller with a very high tangential absolute velocity - higher
than the blade velocity. This means that a good deal of the pressure rise must
come from velocity conversion in the volute casing.
With the usual proportions - impeller width about 50% and blade depth about
8% of the diameter - mechanical considerations limit the tip speed so that the
maximum pressure is about 900 Pa. This can be
raised to 1500 Pa by strengthening the blades with a mid-span stiffening ring

and spoke-type ties to the inlet ring. Alternatively, specially shaped
forward-curved blades may be designed, typically of increased depth at the back
plate end.
The double inlet fan comprises two impellers back to back in a double width
casing and will generally deliver rather less than double the volume flow at the
same pressure. This is best suited to free inlet, ducted outlet installations. In
small direct motor-driven sizes, narrow width impellers serve to fill the
performance gaps between diameter steps. Larger sizes are often belt-driven so
that duty adjustment is by pulley and speed change.
7.9.2 Backward-curved impellers

In these the outer, trailing, edges of the blades are inclined backwards
rather than forwards, and the curvature is reversed with respect to the
direction of rotation. Again there is little change in the relative air velocity
within the impeller, but there is also no reversal of direction, so that the air
leaves the impeller with comparatively low velocity. For a given duty the
casing is bulkier and the tip speed higher than the forward curved requires.
However, the low velocities and, therefore, friction loss, together with
avoidance of flow separation, make backward-curved fans much more
efficient, 80% to 85% being attainable in general-purpose models.
The geometry of the blading permits fewer blades of much greater radial
depth to be used, so that pressure development is largely by
centrifugal action. Mechanically much higher tip speeds are possible and
full-width general-purpose models will commonly develop up to 2500 Pa.
Another advantage is their non-overloading power characteristic, by which is
meant a power input which does not peak at either free flow or no flow. These
fans may be regarded as the standard industrial centrifugals in the larger sizes,
where their aerodynamic performance justifies increased first cost.
7.9.3 Aerofoil-bladed impellers

The larger sizes of full-width backward-curved impellers may be manufactured
with blading of aerofoil cross-section, formed from hollow sheet. Specially
refined designs of this type may reach 90% efficiency and there is probably
some advantage in efficiency spread over the characteristic and in noise
generation. It is fair to say, however, that comparable results have been claimed
for constant thickness blading of complex curvature to co-operate with the
optimum streamlines as these change shape from back plate to shroud.
7.9.4 Increased pressure range

If the width of a backward-curved impeller is reduced the tip speed can be
increased without overstressing the blades in bending. As the optimum volume
flow is reduced the inlet eye diameter can be reduced also, increasing the
radial depth of the blades and the pressure development by centrifugal action.
A whole family of impeller designs with progressively increasing pressure rise
and decreasing volume flow can thus be provided. The efficiency will fall away
as the width is reduced, and the backward curvature becomes less necessary
until, in the narrowest widths, radial blades become usual with the advantage
of no bending stresses to limit the tip speed. Fig. 7.16 is an example of such a
performance family.
7.10 Centrifugal Fans-more specialised designs

7.10.1 Radial and paddle-bladed fans
In the textile industry the air handled may contain lint, or waste strands of
various kinds, which would catch in the blading and soon clog a backward- or
forward-curved impeller. Widely spaced radial
blading is much more satisfactory in such cases; the air performance is

intermediate between the backward and forward types, but efficiency generally
inferior to either. In the paddle-bladed impeller backplate and shroud are
omitted also, the flat radial blades being supported on a central spider. These
are still more effectively self-cleaning, and the blades are easily replaced if worn
by abrasive particles in the air stream.
7.10.2 In-line centrifugal fans

Instead of using a volute casing to collect the swirling air from a centrifugal
impeller, it may be allowed to spin forwards into a concentric annular casing.
Guide vanes will then convert the swirl velocity pressure into fan static pressure,
and an outlet duct can be fitted in line with the inlet duct as for axial fans.
Performance tends to be somewhat inferior to the corresponding volute model,
and the chief advantage is avoidance of the transverse bulk and right-angle
direction change associated with a standard centrifugal fan.
7.10.3 Multi-stage centrifugal fans

Since the maximum pressure required of any fan can be provided in a single
centrifugal stage, some special reason is needed for multistaging.
One such reason is economic, as illustrated by the example of the domestic
vacuum cleaner. The need for low cost motors and impellers sets a limit to the
tip speed, and hence the pressure development of one stage. The in-line
centrifugal arrangement is adopted but, instead of putting guide vanes in the
annulus, they are placed in a radial passage returning the air to the eye of the
next impeller in the series.
7.11 Mixed Flow Fans

7.11.1 Mixed flow with axial discharge
This type can be regarded as a modification of the axial fan in which the blades
operate in a conically expanding passage, which is then curved back, through
guide vanes, to an axial outlet. A substantial part of the pressure development is
by centrifugal action and the fan static pressure is high compared with an axial
fan at the same speed. With advanced design, efficiency and noise level
comparable to those of a backward-curved centrifugal fan is obtainable, with a
more compact, in-line, casing. They have not become widely popular however,
perhaps
130 Woods Practical Guide to Pan Engineering
because quantity production manufacturing techniques, necessary for

competitive cost, are difficult to develop, while flexibility in pressure and
volume duty is somewhat limited.
7.11.2 Mixed flow with volute casing

Axial entry mixed flow blades can be extended into a radial flow
centrifugal formation leading to a volute outlet casing. High pressure
development can be combined with high efficiency given careful design.
Performance is superior to that of the normal narrow-bladed centrifugal
blower, giving the fan a place in the high power applications of heavy
industry. In fact, the type is an extension into the fan field of the high
strength, high tip-speed designs used for turbo-compressors and
superchargers.
7.11.3 Mixed flow with radial discharge

The roof ventilator is an example of a free outlet fan in which it is
advantageous to discharge the air radially in all directions under a cowl
excluding rain and snow. Propeller, axial, and centrifugal impellers are all used
in roof units, but the mixed flow design is in many ways the ideal. No
compromises for general use are necessary, the impeller, cowl and
roof-mounting curb being designed as a single-purpose aerodynamic and
mechanical unit. Pressures of 250 Pa or so to deal with long ducts or maintain
duty in adverse winds can be developed more quietly than with axial fans. At the
equally important zero pressure condition efficiency is better than with
centrifugal fans, and the power characteristic can be completely
non-overloading.
7.12 Comparative Performances of

General-Purpose Fans
Dimensionless coefficients and charts, with

such concepts as "specific speed" help the fan
designer in type selection, and are touched on
in Chapter 11. The user, however, is
concerned only with those fans which are
economically available. One limitation this
places on his choice is a maximum tip speed
for each class above which construction costs
tend to soar.
Fig. 7.23 illustrates this with typical

examples from each of the main general-
purpose fan types. To provide a basis of
comparison each is chosen for an output, that
is product of volume flow and fan total
pressure, of 10,000 watts. It is clearly seen
that they cater for very different pressure
duties, though axial and centrifugal ranges
overlap at moderate pressures. The range
could be extended almost indefinitely with
higher pressure centrifugals and higher volume
axials.
The maximum power absorbed at any point

on the constant-speed characteristic is listed to
illustrate the difference between overloading
and non-overloading types. Kilowatts absorbed
at the 10 kW output duty are based on typical
efficiencies for each class.
The fan diagrams indicate graphically, to the

same scale, the relative space occupied, and
all these comparisons hold of course for other
sizes, speeds, and duties within the range of
each type.
Selection of a fan below its top rated speed

will increase space and cost per kilowatt of
output, but has the advantage of quieter
operations.
7.13 Cross-Flow Fans

Improvements in casing design have brought this type into prominence for
use in certain small domestic appliances. Impellers similar to those of multi-vane
centrifugal fans are used, with the difference that they can be extended
indefinitely in width, provided bearings are fitted at intervals to preserve
mechanical stability. It therefore becomes possible to use inlets and outlets of
long, narrow, rectangular shape, which opens up new possibilities of functional
and appearance design for such
appliances as table fans and electric fan heaters. Efficiency is low, but the fans
are fairly quiet for their duty.
To obtain reasonable performance an adequate outlet diffuser is necessary
since pressure mainly comes from conversion of the high velocity pressure
leaving the impeller. It is also necessary to design the air circuit through the
appliance and the impeller as one unit, to ensure that the build-up of the vortex
will not be disturbed. Larger models are occasionally made where the
rectangular flow passage is helpful. An example is the provision of air curtains
over the open doorways to exclude the weather.
7.14 Special-Purpose Fans

A fan maybe aerodynamically suitable for a given duty, but disqualified for other
reasons. Special hazards are raised by polluting substances in the air stream. A
classification from this point of view follows.
7.14.1 General-purpose fans maybe defined as fans suitable for operating

continuously in normal indoor atmospheres at temperatures up to 40°C. Particular
general-purpose fans may meet the milder cases of the hazards below. To ensure
long life, however, special-purpose fans should be considered.
7.14.2 Weather resistance. Higher grades of exposed plastic components and paint
finishes are required for repeated exposure to rain, fog, and the ultra-violet radiation
of the sun. Ingress of rain and snow must be limited to protect moving parts and
delicate components. Severe climates pose special hazards. For example, termite and
fungus attack in tropical environments on insulation and other organic materials;
lubricant stiffness and material embrittlement at Arctic temperatures; marine strength
standards for deck-mounted fan casings.
7.14.3 Moisture resistance. Dripping water and repeated condensation from a

saturated atmosphere are to be regarded as abnormal hazards. Higher grade
materials or finishes may be required to resist corrosion. Water must be kept out of
bearing housings by seals or flingers. Motor frame joints must be sealed or gasketted
to keep out liquid water. Humid air will nevertheless enter and condense when the
motor cools; drain holes on the underside are advisable to prevent accumulation of
condensate.
Special fans to handle air containing condensed liquid are called wet gas fans. The
impellers are constructed from material selected to resist erosion, that is pitting at the
points of impact with the liquid drops.
7.14.4 Corrosion resistance. Heavily polluted industrial atmospheres will be

sufficiently corrosive to shorten the life of most general-purpose fans. Provided the
concentration remains within what might be described as the "breathable" range,
additional coats of protective paints will usually suffice. Chlorinated rubber paints are
commonly used for this purpose.
Heavier concentrations in the air handled by the fan demand special surface
treatments going right back to the basic steel or aluminium. Baked enamels, sprayed
metal powders, or thick fused plastics coats may be cited, each requiring a carefully
planned process schedule.
Fans to handle highly corrosive gases, particularly when wet or hot, may require to
be constructed from special corrosion-resistant metals or plastics. Selectivity is
necessary since no material is uniformly resistant to all reagents, though rigid PVC
has a wide range of application in the chemical industry.
7.14.5 Heat resistance. General-purpose fans with motors out of the air stream will
handle air with temperatures well above 40°C, the first limitation being generally
reached with the bearing lubricant. Standard
lithium-based ball bearing greases will give satisfactory life at 60°C to 80°C
housing temperature according to speed. Pedestal housings can be isolated to
maintain such temperatures with much hotter air in the fan. Oil-lubricated bearing
systems can be designed for any practical condition, with circulation through an
oil cooler, or water cooling for the worst cases.
Electric motors operating in the air stream of a fan are normally rated for a
maximum ambient temperature of 40°C. Specially wound motors insulated with
Class F or Class H materials may be rated to operate in temperatures up to 70°C
and 100°C respectively. At higher temperatures the motor should be moved
outside the air stream as with most centrifugal fans and bifurcated axials.
For hot gas fans operating at really high temperatures, the materials of
construction must be changed. Maximum impeller speeds and stresses may
require progressive limitation at temperatures above 200°C for aluminium alloy
and 400°C for mild steel construction.
At temperatures from 600°C to 1000°C special steels are available to resist the
dimensional changes known as creep, and the increased corrosion rates which
would otherwise occur.
7.14.6 Abrasion resistance. The impellers of general-purpose fans may be worn

away by the passage of too much dust of an abrasive character. If the particles
cannot be sufficiently removed, as by a cyclone, an abrasion resistant fan must be
specified. This may be constructed of steel alloy, harder than the material to be
handled, or alternatively easily replaceable plates may be fitted at the areas
subject to wear. Rubber blades moulded to aerofoil shape on a metal core are
sometimes used for axial fans, and will deal with light particles of any hardness.
7.14.7 Explosion hazards. Any possibility of an inflammable or explosive

concentration in the gases handled by a fan must be taken seriously. The
requirements for electrical equipment were discussed in Chapter 5. A flameproof
fan may be necessary with the appropriate class of electric motor, terminal
arrangements, and cabling.
Sparkproof construction may also be specified for the mechanical parts of the
fan. This involves extra precautions against rubbing contact between moving
parts and the avoidance of materials prone to incendive sparking on impact with
foreign bodies. These include magnesium and (to a lesser extent) aluminium
alloys. Steel and cast iron are relatively safe.
7.14.8 Leakage avoidance. For the conveyance of toxic or radioactive gases a

gastight fan is likely to be specified. The same requirement may apply to an
inflammable or valuable gas, e.g. hydrogen or helium in a closed cooling circuit.
The permissible leakage rate at a given pressure should be laid down, and will
determine the jointing and gasketting system for the casing together with the
type of rubbing seal if the drive is external.
7.15 Elementary Theory-Axial Fans

For those who wish to obtain a clearer understanding of the mechanism of
fan pressure generation, a little mathematics is necessary. Fig. 7.24a is an
isometric view of the path of the air through an axial fan; a small portion of the
total flow is considered, confined between two imaginary cylinders of slightly
differing diameter. Radial forces are assumed to be so balanced that the air
which enters flows wholly between the two cylinders with no tendency to
escape, outwardly or inwardly.
As the air passes through the rotating impeller, forces exerted by the blades
give it a spinning motion so that it leaves the impeller in helical paths as if
guided by a very long-pitch screw thread. The condition of radial balance
enables the imaginary cylinder to be unrolled flat so that the helical paths
become straight lines, easier to handle mathematically.
Fig. 7.24 shows this development. Velocities are represented by vectors

giving the magnitude and direction of the air flow. These are distinguished as
follows:
In the absence of external forces air will enter the fan in the axial direction
v1. To find its velocity w1 relative to the leading edge of the moving blade, we
must subtract the blade velocity vector u1 (that is add – u1) to the air velocity
vector v1. This operation is performed by drawing the vector triangle as in Fig.
7.24b. The impeller will be so designed that, at the "design point" volume
flow and corresponding inlet velocity w, will meet the blade at the most
favourable angle of approach.
The blade is given an aerofoil shape designed to deflect the relative velocity in
the direction of rotation from w1 to w2. Note that the components of w1 and w2 in
the axial direction must be the same since there is the same volume flow in the
same annulus area at inlet and outlet. It follows from the geometry of the flow
(note also the increase in blade passage width from s1 to s2) that w2must be less
than w1. The corresponding fall in the relative velocity pressure, (½ρw12 - ½ρw22)
results in an equal rise in static pressure (ps2 – ps1) within the impeller from blade
inlet to blade outlet.
Now static pressure, as explained in Section 6.1.2, is an absolute property of
the air which does not change with the frame of reference. Therefore the rise of
static pressure within the impeller is also the rise of static pressure external to the
impeller from the inlet to outlet side. A fall in relative kinetic energy has been
transformed into a rise in absolute static energy. The vector diagram on the outlet
side, u2 being the same as u1 shows that the absolute velocity v2 is increased as
well as deflected from v1. With a little rearrangement the two diagrams can be
combined into the composite vector diagram 7.24c, which shows the velocity
changes quite clearly. We can now determine the total pressure rise across the
impeller (pt2 - pt1).
An alternative expression can be found with the aid of the absolute flow angles
a shown, together with the relative flow angles β in Fig. 7.24c. For our particular
case a1 = 90° and u2 = u1 while there is no change in the axial component of v,
ie. V2 sin α2 = v1.
Substituting in equation 74 and remembering that sin2 α2 + cos2 α2 =1
7.16 Elementary Theory-Centrifugal Fans

A portion of the flow through a centrifugal fan is shown in Fig. 7.25a. It is
again confined between two imaginary surfaces, initially cylindrical but then
curving smoothly outwards into two parallel radial planes. The area between the
planes is no longer constant, but they follow the natural streamlines of flow as
determined by the casing and impeller. This includes a loss-free turn to the radial
direction when approaching the blades.
The velocity vectors at blading inlet and outlet have the same meaning as
for axial fans, although u2 is no longer equal to u1. Once again the relative
velocity vector w1 determined by the inlet side vector triangle, meets the
leading edge of the blade at a favourable angle at design volume flow. The
radial component at outlet v2 is fixed by the volume flow and outlet area,
while the assumption that the relative velocity w2 is in the direction of the
trailing edge of the blade completes the outlet velocity triangle.
Transformation of relative kinetic energy within the blade passages will
produce the static pressure rise ½ρw12 - ½ρw22 in the axial case, and the
total pressure rise will again include the change in velocity pressure, ½ρv22 -
½ρv12. There is, however, an additional pressure rise in the centrifugal fan
case, derived as follows.
Consider the element of air shown in Fig. 7.25d. Between radii r and (r +
dr) and between successive blades spaced s circumferentially and of width b,
the mass of this air will be:
This air must have a circumferential velocity component a equal to that of

the impeller, that is 2∏ rn where n is the revolutions per second. Irrespective
of any other relative motion it will experience a centrifugal force:
This force, exerted over an area of sb will produce an increment of

pressure from r to (r + dr)
Integrating this from r1 to r2 the contribution of centrifugal force to the

static pressure rise from impeller inlet to impeller outlet is obtained as:
There are thus three component parts to the total pressure rise across the
impeller:
and the sum of these is the fan total pressure:
This expression is in fact quite general for impellers of any kind, including
axial and mixed flow types. It is also unnecessary for the inlet velocity to be
axial, deflection with and against the impeller rotation being respectively
defined by positive and negative values of α1. Formula 76 still applies, the
alternative formula in terms of flow angles being:
7.17 Elementary Theory-Impeller Power

If the cross-sectional areas of the portions of the total flow illustrated in
Figs. 7.23 and 7.24 are a1 at impeller inlet and a2 at impeller outlet, we can
write down the mass flow per second through the portion considered as:
At outlet this flow is swirling with a circumferential velocity component v2

cos a2 at a radius r2. It therefore carries with it each second angular
momentum to the value of:
Similarly the flow entering the impeller brings with it angular momentum at
the rate qmr1 v1 cos α1. The difference between these quantities is the rate at
which angular momentum is being added to the air stream, which must equal
the torque applied by the impeller:
At a speed of n revolutions per second this torque requires a mechanical

power input of:
But, the power output of the fan is Q1 P1 where Q1 is the whole

volume flow. Similarly, the power output of the portions of flow we
have been considering will be q1 Pt where qm = ρql. Substituting for
P, from equation 77 the element of power output becomes:
qm (u2 v2 cos α2 - u1 v1 cos α1)
It is not surprising that the aerodynamic power output should equal
the mechanical power input of equation 78 since all losses have been ignored.
These losses have the effect of reducing the pressure rise and effective output
together with other factors discussed in the next section. The mechanical input
power obtained by summing the equation 78 over the whole flow through the
fan is, however, correct.
7.18 Practical Impeller Design

This idealised treatment is only a starting point. Further factors the
designer must consider include the following
(a) The air will not leave the impeller at the angle set by the trailing edge of the
blade. There will be a deviation angle tending to reduce input and output
power. This is a function of aerofoil shape and blade loading.
(b) The air will meet the leading edge of the blade at the optimum angle only at
the design volume flow. At other flows the incidence angle will change. The
effect of this change on performance over the whole fan characteristic is
again a function of aerofoil shape and loading.
(c) To keep the portion of flow considered between the imaginary surfaces
defining its intended path requires a balancing transverse pressure gradient.
In the axial case, for example, only the so-called free vortex pressure and
velocity distribution will secure flow at constant radius along concentric
cylinders. This distribution requires a constant value of rv cos α, i.e. a spin
component which is smaller the bigger the radius. This limits the work done
towards the blade tips, giving a fan of poor power-size ratio; to overcome
this weakness forced vortex designs are usual, increasing tip work. To main-
tain balance the streamlines tilt outwards through the impeller, so that r2
and u2 are greater than r1 and u1 making it necessary to use equation 76
rather than 74.
(d) Viscous drag forces at the blade surface and wake effects behind the blade
convert some of the work input into heat instead of useful pressure rise.
(e) Tip clearance effects and boundary layer retarded flow along casing, hub,
backplate and shroud spoil the flow pattern at the ends of the blade and
limit the work done (input and output).
(f) The velocity leaving the blades is usually far from uniform in magnitude or
direction. Since the energy is proportional to the square of the velocity,
more is required by the peak velocities than is saved in the troughs. Thus
excess kinetic energy is supplied which will not all be available when the air
has reached the downstream test plane.
CHAPTER8
Fan drives and speed

control
The vast majority of general-purpose fans are driven by electric

motors. Motors suitable for fan drive are discussed in this chapter,
together with their installation and speed control. Fan duty may be
controlled by speed adjustment or by aerodynamic means and
comparison of the methods available is reserved for Chapter 9.
Of course, large numbers of specialised fans are not electrically
driven. Consider, for example, the cooling of internal combustion
engines, by impellers mounted on a take-off from the main shaft.
Even in the automobile field, however, the use of electric motor
drive is spreading, since it overcomes the incompatibility of power
economy at high engine speed with adequate cooling at idling
speed.
8.1 AC Motors
8.1.1 polyphase induction motors are the normal standard drive
for industrial fans, where the necessary three phase AC (alternating
current) supply is almost always available. The usual squirrel-cage
rotor has the merits of minimum maintenance, robustness and low
cost; the electrical conductors are of aluminium cast in place
through holes in the iron rotating member (rotor). In large sizes
copper bars brazed to copper end rings may be preferred.
142
Fan drives and speed control 143
The stator (stationary iron member) is wound with coils of insulated

wire inserted into slots and connected to the supply. The winding
pattern determines the number of poles or peaks of magnetic flux
round the circumference.
These magnetic poles rotate at a synchronous speed which is also the
speed of the motor on no-load.
When an induction motor is loaded the speed falls, the drop from
synchronous speed being known as the slip. The slip at full load ranges
from 1% of synchronous speed for large motors to 10% for fractionals,
typical actual full load speeds being given in Table 14.24. It is the slip
of the rotor conductors past the synchronously rotating magnetic field
which generates the rotor currents; these in turn react with the field to
produce torque.
The slip is approximately proportional to load up to full load, but
thereafter it increases more rapidly until at pull-out the maximum
torque is reached. The relationship between slip and torque is
dependent on rotor design, and two examples are illustrated in Fig.
8.1. Since the initial load at starting is that due to friction only, high
torque motors are not necessary for fan drives. Standard rotors are
normal practice, though there is a tendency for these to incorporate
special cage features to flatten the torque curve and improve starting.
In small sizes where specialised quantity production is justified, fan
motors may have limited pull-out torque and a suitably tailored
torque-speed curve to improve voltage-speed control and limit starting
current.
While standard industrial motors may be used for direct fan drive
with the impeller mounted on the shaft extension, modifications are
often necessary and specialised fan motor designs are produced.
Before specifying a standard motor to replace one which may or may
not be standard, the following points should be checked:
Will the bearing system take the weight load and aerodynamic
thrust? See 8.6.3, 8.6.4 and 8.6.5.
Is the shaft extension suitable to secure the impeller? A shaft
shoulder and shaft-end clamp washer may be necessary.
Will the fan air stream and the motor cooling flow conflict? Removal
of the motor cooling fan and cowl, with re-rating in the fan air stream,
may be necessary for motors in the air stream.
Is the motor insulation and lubrication system and its rating correct
for the air conditions, running hours, etc.? See 8.6.1 and 8.6.6.
Will the motor size and shape, the necessary mounting arms or plat-
form and such features as projecting terminal boxes obstruct the air
flow to or from the impeller, or set up noise generating air flow
patterns?
Will the motor start a high inertia impeller quickly enough to avoid
overheating or unnecessary tripping of the overload protection?
8.1.2 Split phase induction motors

A stator winding connected to single phase AC-the usual domestic
supply-cannot produce any starting torque. Special auxiliary windings
are necessary which are either of high resistance (split-phase type) or
incorporate a condenser (capacitor-start, induction-run type). In the
standard fractional horse-power motor the auxiliary circuit has a very
short time rating-a minute or so. It is disconnected by a centrifugal or
current-controlled switch at about two-thirds of full speed, when a
somewhat inefficient form of rotating magnetic field has developed.
Such motors have sufficient torque to overcome excessive starting
friction generated by a badly-adjusted belt drive or sealing gland.
8.1.3 Permanent split capacitor motors

When the fan impeller is directly mounted on the motor shaft
extension the permanent-split capacitor design is to be preferred.
Provided the impeller will never absorb less than, say, half full load,
it is possible to design main and auxiliary windings to share the
load equally. A given motor frame will then develop the same
power on single phase as on three phase supply, with equal
efficiency and noise level, and at near unity power factor. This
favourable result is peculiar to fan drive, with its roughly constant
load; if such a motor were allowed to run light the phase split
would become very unbalanced, applying excessive voltage to the
capacitor and overheating the auxiliary winding.
8.1.4 Capacitor-start, capacitor-run motors

At full load outputs exceeding a few kilowatts the permanent
split capacitor motor will not develop sufficient starting torque
or will take too long to accelerate a high inertia impeller. If a
three phase supply is not available, as may happen in isolated
rural installations, a capacitor-start, capacitor-run motor is
suitable. In this type an additional, short-time-rated, capacitor
is switched in at low speeds to provide extra starting torque.
8.1.5 Shaded-pole motors -

For outputs of 20 watts or less the shaded-pole induction motor is the
common type for fan drive. No switch or capacitor is needed, the
auxiliary winding taking the form of a closed loop of solid copper, offset
to one side of the centre line of each main pole. Currents induced in
this
loop by transformer action develop a rotating magnetic field. This is
poorly distributed, resulting in a low efficiency, seldom exceeding 25%,
but the motor is quiet, reliable and cost-effective in the smallest sizes.
The squirrel cage rotor always rotates from the main pole towards the
nearest copper loop, unlike all other induction motors, which can be
reversed by reversing the connections of one stator winding.
8.1.8 External rotor induction motors

A specialised mechanical type of induction motor has an external
rotor carrying a squirrel cage rotating round an internal wound
stator. One arrangement has blades mounted on the rotor which
thus forms the hub of an axial impeller. The rotor may be overhung
on a stub axle running in bearings buried within the stator body.
Alternatively, the rotor may have bearings either side of the stator
on a stationary shaft which is hollow
to carry the leads to the stator. The latter arrangement is usual for
ceiling fans and is also occasionally used for small double inlet
multi-vane centrifugal fans.
8.1.7 Synchronous motors

These are rarely used for fan drive, unless the fan load is
continuous and large, forming a substantial fraction of the total load
in the plant. It is then possible to use the unique ability of a
synchronous motor to draw leading power-factor current from the
supply. This will compensate for the lagging power-factor current
drawn by induction motors in the plant, correcting the overall power
factor and so qualifying for better rates under the supply company's
tariff.
The stator is similar to that of a three phase induction motor, but
the rotor is quite different. It has an insulated winding fed with direct
current, either from a DC commutator-type generator on the same
shaft or through slip rings and a rectifier from the AC mains. This
holds it in exact synchronism with the rotating magnetic field of the
stator. Starting and synchronising on load is difficult and designs
sometimes involve running the motor up light and subsequently
accelerating the fan impeller through a fluid clutch or other slip-type
coupling.
8.2 Speed Control of Induction Motors
8.2.1 Voltage control

Small squirrel cage induction motors driving fans can have their
speed reduced or regulated by reduction of the applied voltage,
provided they are of the shaded-pole, permanent-split capacitor, or
polyphase types. Split phase and other single phase motors
incorporating switches cannot be regulated in this way because the
short-time rated auxiliary windings might be brought into circuit.
Reasonably stable running is possible because of the
proportionality of fan torque to the square of the speed. Fig. 8.6
shows the way in which
the motor torque-speed relationship is reduced with successive reduc-

tions in supply voltage. At full speed and first regulated speed there is
a well defined intersection X with the fan torque-speed curve
identifying stable running speeds. In the last case the intersection is
not so well defined and the running speed will fluctuate somewhat with
motor temperature, bearing friction, and other factors.
Substantial speed reduction is inevitably accompanied by energy loss
in the squirrel cage conductors. The loss reaches a maximum at two-
thirds of synchronous speed, and is then 15% of the power the
impeller would absorb at synchronous speed. This loss can be
accommodated
in motors with full load ratings up to about 350 watts output. At higher
outputs the motor frame must be derated, and a practical maximum
for voltage regulation is about 1 kW output.
Small fan motors are often designed with relatively high resistance
cage rotors because, as explained under wound-rotor motors, this in-
creases stability. Voltage control by autotransformer is best for
stability also, since the output volts are independent of load, but series
chokes or resistors are often used.
The permanent-split capacitor motor is best regulated by control of
the main winding voltage, the auxiliary winding and capacitor
remaining at full voltage. This arrangement maintains a better balance
between the field strength and phase angles of the two windings.
increasing stability and reducing regulator load.
Solid state transistor or thyristor controllers can give continuous

rather than stepped speed control. The losses in the controller are
negligible, although motor losses can be substantially increased if the
motor is not specifically designed for this form of control. Stability is
also dependent on appropriate design, Fig. 8.7 showing as an example
the unfavourable effect of too low a rotor resistance. On three phase it
is best if the supply has a neutral connection available, connected
through the controller to the star point of the motor stator, thus
providing three independently controlled phases.
The principle of thyristor operation is the disconnection of the supply
at each current zero, reconnecting at a controlled time interval later so
that the motor receives power for part of each cycle only. The controller
must incorporate protective devices for the thyristors of proved
adequacy in relation to the motor control duty. Radio interference must
be suppressed.
8.2.2 Wound rotor induction motors

In these machines the squirrel cage is replaced by an insulated
winding connected through slip rings and brushes to an external
variable resistor. When the resistance is set at zero the motor runs at
full speed and load exactly like a squirrel-cage motor. As the resistance
is increased the motor torque-speed characteristic is progressively
changed as shown in Fig. 8.8, enabling the fan to run stably at any
speed below full speed.
Since the energy loss has been removed from the rotor winding there
is no longer any limit to the power output for which the motors may be
built. Unfortunately, the loss remains inevitable. Apart from certain
rather costly systems of regenerative recovery it must be dissipated in
the resistor elements. If the full load fan power is Wi watts at n1
revolutions per second, the synchronous speed being ns and the

regulated speed n, the loss in the resistor and (a small item) the rotor
winding will be:
Such motors may be considered for short-time speed control above

the range covered by de-rated squirrel-cage motors, but some
de-rating is still necessary and a more efficient method should in any
case be considered if the running hours are long or the load heavy.
8.2.3 Frequency control

A three phase squirrel-cage motor will operate efficiently at speeds
proportional to the supply frequency provided the supply voltage is
varied in the same proportion. With its use of standard fan motors and
freedom from energy loss this would be the ideal method if it were not
for one drawback-high first cost. It may, nevertheless, be economic for
large fans, or if a large number of fans are to be simultaneously
controlled from a single frequency converter, or if specially accurate
speed control is wanted.
Solid state devices have taken the place of the older rotating
machine sets for the conversion of the 50 or 60 Hz supply to variable
output frequency. Thyristors are used for the repeated switching of the
motor current from phase to phase of the supply at controlled instants
in the cycle. The cost arises largely from the complex protective
circuits required to guard the thyristors from the danger of shortcircuit
or overvoltage surge; as well as to control the frequency. High
frequency harmonic currents decrease motor efficiency by a few per
cent., and may require limitation to satisfy the supply authority. There
are two main types. The rectifier-inverter system first converts the
alternating supply currents into a unidirectional (DC) current and then
reconstitutes AC at a different frequency controlled by an oscillatory
circuit. The cycloconverter system dispenses with the DC stage,
splitting the supply into pulses at several thousand Hz, which are then
modulated at output frequency. Cyclo-converter outputs are preferably
below 30 Hz, which may be favourable for large low speed fan motors.
since they will be wound with fewer poles.
8.2.4 Multi-speed windings

Three phase squirrel-cage induction motors can be wound with more
than one pole number. By switching from one to another different
synchronous speeds can be selected. For each pole number the motor
speed will be the same as that of the corresponding single speed
machine, though the efficiency and output will be somewhat reduced
owing to the necessary reduction of stator conductor cross-section, or
winding factor.
One method involves putting two, or occasionally three, separate
windings into the same stator slots. Since, on fan load, the power
required falls with the cube of the speed, the low speed can be wound
with a much smaller wire cross-section. The bulk of the slot space
remains available for the high speed winding, which will not usually
need to be derated more than 30% to 35%.
The other method is known as pole amplitude modulation (PAM).

In this a single stator winding is tapped at selected places so that
the coils may be rearranged by a two-position switch to change the
number of poles. The system is flexible, it having been shown that
quite unsymmetrical coil arrangements-designed in accordance with
the Rawcliffe patents-will perform without much loss of output or
efficiency.
Common pole number combinations are 4 and 6, 4 and 8, 6 and

8, but the choice is virtually unlimited.
It is possible to provide:
8.3 Commutator Motors

8.3.1 Direct current motors
Constant speed DC motors will not be used for fan drive if an AC
supply is available. This limits them in practice to transportation
applications, often at 6, 12 or 24 battery voltage, or to isolated
locations. In both cases the power is likely to be low.
When the impeller is mounted directly on the motor shaft a series

wound motor is usual. The single field winding is connected in
series with the armature through the brushes and commutator. A
series wound motor should not be used for indirect fan drive
because, if the load be disconnected, for example through belt
failure, the speed will rise to a dangerous level. The latter
weakness is avoided in the compound wound motor. This has, in
addition to the series field winding, a shunt field winding which is
connected across the supply and will maintain sufficient magnetic
field at light load to prevent excessive speed. Pure shunt motors
without a series winding are unsuited to fan drive.
8.3.2 AC series motors

Small series wound motors with ratings up to about 500 W can
be designed to operate on AC as well as DC. At speeds of 100 to
200 revolutions per second the AC and DC performances are quite
similar, and such machines are called universal motors. However,
with the disappearance of DC public supplies this feature has lost
its importance. The ability to operate far above the 50 or 60 rev/s
maximum of the induction motor is nevertheless valuable for such
applications as domestic suction cleaners. As these are also for
intermittent use, the rather short brush and commutator life
characteristic of AC commutator motors is not a serious
disadvantage. Radio interference suppression is always necessary.
8.3.3. Speed control of DC motors

DC (and AC) series fan motors may be reduced in speed by the series
connection of a variable resistor. At higher powers the resistor may also
act as a starter, but the energy losses are substantial, and for fan
powers in excess of 1 kW or so more efficient methods should be
chosen.
One such method is shunt field control. A compound wound motor

with a relatively weak series winding, and additional series connected
commutating poles to suppress sparking, will be used. The shunt field
can be weakened using a variable resistor or potentiometer whose
losses are quite small. As the field drops the speed rises so that
maximum speed and load occur at the weakest magnetic field. This
reduces the output available from a given motor frame and limits the
speed range available to 1 ½ or 2 to 1 at the outside.
8.3.4 DC and other commutator motors for AC supply

The successive developments of mercury arc, semi-conductor, and
thyristor types of rectifier have enabled large DC motors, with shunt
field speed control, to be operated off an AC supply. Of greater
significance is the phase control or "chopped-wave" system of output
voltage control available in addition to rectification when thyristors are
used. In this system the shunt field of the compound wound DC motor
is maintained at full strength through a small auxiliary rectifier. The
main thyristors supply the armature and series field. In addition to full
wave rectification they have a variably delayed closure in every cycle
which controls the effective voltage applied to the armature. This
reduces the motor speed to any desired value, without energy loss,
apart from a slight increase in motor losses caused by high frequency
pulsations. Although relatively costly this system provides an answer to
any problem requiring wide range, accurate speed control with
minimum loss.
Several types of AC commutator motor are made for three phase AC

supply and are capable of operating over a range of speed. Their use
for fan drive is mainly confined to heavy industry where skilled
maintenance is available to look after the brush gear and commutators.
Also many designs involve physical movement of brushes, induction
regulator cores, etc., and may be more suited to manual than to
automatic control.
8.4 Induction Motor Performance
8.4.1 Efficiency and power factor

In the smaller sizes single phase permanent-split capacitor or
capacitor-start capacitor-run motors can be made with the same
efficiency as standard three phase induction motors and a power factor
generally in excess of 0.95.
On DC the product of the supply voltage and supply current equals

the input power in watts. On AC, however, there is an additional inter-
change of energy each cycle between the magnetic field and the electric
supply, measured in "reactive volt-amps", which does not, on average,
supply any power. The useful energy which is supplied is measured

in watts. These two quantities can be treated as phasors, at right
angles to one another, and their phasor sum is known as the
"volt-amps" input.
The ratio of the watts to the volt-amps is known as the "power

factor" and it may be expressed as "cos Ø" where 0 is the phase
angle between the supply voltage and the supply current. The
relations between the supply volts, V; the supply current, I amps;
the supply power, We watts; the output power, Wm watts and the
efficiency, η, are then:
8.4.2 Power factor correction

The capital cost of transmission and power station electrical plant
is largely dependent on the kVA rather than the kW. Hence the
power supply tariff is often based partly on the kilowatt-hours
energy supplied and partly on the kVA maximum (30 minute)
demand. Fans are likely to be continuously running, and therefore
to enter fully into the maximum demand of a plant. If the aggregate
kVA of all fans is considerable, and particularly if the fan motors are
low speed, and therefore of rather low power factor, power factor
correction is worth consideration. Permanent-split capacitor motors
are inherently corrected, so the issue will arise only with three
phase motors.
Banks of static power-factor correction capacitors can be
purchased for connection across the incoming supply terminals.
They will have a rating kVAC but the power factor is "leading"
rather than "lagging" which implies the opposite sign to the WAR of
the motors. We thus have:
The maximum demand will therefore be a minimum, and equal to

kW, when kVAC = WAR = kW tan Ø. The formula shows that kVAC
can fall substantially short of WAR before the kVA rises appreciably
above the minimum.
8.4.3 Supply voltages and windings

The windings of three phase induction motors up to 2 kW input
are normally connected in star for supplies of 380 to 415 volts. With
each end of each phase brought out to 6 terminals it is possible to
reconnect
in delta for 220 or 240 volt three phase supplies as used in some
countries for low-power installations. This makes stock motors more
widely available.
The practice may be extended to 5 kW unless there is a requirement

for star-delta starting (see Section 8.5.1) on the higher voltage. Such
a requirement necessitates motors with delta-wound stators.
Similar considerations sometimes lead to single phase motors being
wound in two parts for parallel connection on 110 volts and series
connection on 220-240 volts, and even for a compromise design
capable of operating on either 50 Hz or 60 Hz in the shaded-pole
range.
8.4.4 High inertial loads

Particularly heavy impellers can in some cases influence the
selection of motors. It can be shown that the heat dissipated in the
rotor of an induction motor while starting equals the kinetic energy
stored in the rotating parts when full speed is reached. Additional heat
is generated if external work is done during the starting period. The
danger is that the rotor will be overheated, particularly if starts are
frequently repeated, as they may be if the fan operates on a
time-cycle or temperature-cycle. The stored kinetic energy of the
rotating parts is:
where the impeller rotates at n, rev/s and the motor at n1 rev/s -

which may of course be different from n1 for an indirect drive. I1 and
Im are the moments of inertia of the impeller and motor rotor
respectively, found from the formula:
in this formula m is mass in kg of the rotating part considered and rg

its radius of gyration in metres. Rg is likely to be between 25 and 30%
of the diameter for an axial impeller, 30 and 36% for a
backward-curved centrifugal, and 36 and 42% for a forward-curved.
For a cylindrical rotor rg is 35% of the diameter.
The torque of an induction motor directly coupled to an impeller is
plotted in Fig. 8.9, but for the present purpose it is sufficient to
estimate an average value, Mm newton-metres, over the whole
starting speed
range. A torque M1 is needed to drive the impeller at full speed, n1

rev/s, and varies with the square of the speed as plotted, leaving a
variable margin for acceleration. Fig. 8.10 gives the dependence of
two factors, A and B, on the ratio of full-load torque to average
starting torque. Then, for the general case when nm is not equal to n1:
Motor makers usually quote the inertia of the motor rotor, together
with the maximum allowed starting time or kinetic energy of rotating
parts. If these are exceeded an increase of motor frame size is
necessary, though this is only likely to occur when a heavy centrifugal
impeller operates at well below its maximum speed, using a
correspondingly low power driving motor.
8.4.5 Example
A 1250mm heavy-duty, backward-curved impeller, weight 350 kg
and radius of gyration 430mm, absorbs 15 kW when belt driven at 12
rev/s by a 24 rev/s motor.
Can a 15 kW D160 motor be used with 0.12 kgm2 rotor inertia

and an average torque 1 .7 times full load?
This is too close to the recommended maximum starting time of

20 seconds. Considering a D180 frame with average starting torque
1.7 times full load of 22 kW the torque ratio for Fig. 8.10 becomes
15/1.7 x 22 = 0.40 giving A = 1.18, Mm = 250 Nm and acceleration
time = 12 seconds, which is satisfactory.
8.5 Motor Control Gear

8.5.1 Direct-on-line and star-delta starters
Any fan may be started direct-on-line (d.o.l.) provided it is not
outside the range allowed by the supply authority. When some
limitation of starting current is called for the first choice on a three
phase AC supply is usually the star-delta starter. The motor windings
are connected to six terminals as in 8.4.3 and are designed for delta
connection when running. At starting a star connection is made, and
has the effect of reducing both the standstill current and the standstill
torque to approximately one-third of the values they would have for
d.o.l. starting in delta.
One-third of normal starting torque is always sufficient to get the
impeller started, but it may sometimes happen that there is insufficient
torque in star to accelerate beyond about two-thirds of full speed.
Switching to delta will, of course, finish the job, but at the expense (in
some cases) of a current rush only a little less than it would have been
with d.o.l. starting. Modern squirrel-cage motors have "double-cage" or
other specialised rotor designs which enable a good torque to be
maintained over the whole speed range. Nevertheless, a requirement
for star-delta starting on fan load should be specified to the motor
manufacturer, who may recommend derating a particular motor frame,
though rarely more than 10% or 20%.
8.5.2 Other starting methods

Large squirrel cage motors are more likely to require special starting
arrangements if the demands on the supply need to be limited. The
autotranslormer starter connects the motor terminals to reduced supply
voltage during the run-up period. The star-delta starter is equivalent to
an autotransfoimer set at 1 / √ 3 = 58% of supply voltage. This may
well be the first autotransformer step with one or more intermediate
steps inserted during the run up to full voltage. Part-winding motors are
popular in America, for the smaller three phase motors. The stator
winding is in two parallel sections, one of which is disconnected at
starting, increasing the impedance to current flow.
Resistance starters insert successively reduced steps of metal strip
resistors between the motor terminals and the supply. They may be
relatively cheap though requiring selection to match the motor. An in-
teresting variant is the liquid electrolyte starter. As the electrolyte is
heated by the passage of the starting current, its resistance falls,
thereby steadily increasing the effective motor voltage. This is
particularly suitable for fan load with torque proportional to (speed)2
and is inherently automatic, the final step being closure of a switch
short-circuiting the electrolyte.
Wound rotor motors with step-by-step metal, or continuously varied
liquid, resistors in the rotor circuit provide the ideal start with minimum
kVA demand at all speeds from start to full load.
8.5.3 Overload protection

The first stage is usually the HRC (high rupturing capacity) fuse to
protect the motor, starter and installation wiring from damage in the
event of a short-circuit. This is done by breaking the circuit within the

first half cycle (1 /100th of a second) before the very high prospective
current, which would flow if the circuit were not broken, is reached.
Next the standard starter will contain a bi-metal or fusible metal
device which will open the contacts after an interval dependent on the
current in the manner of Fig. 8.11. This should disconnect a faulty
motor and should also protect the windings from overheating if the
high current starting period is unduly prolonged for any reason. Slow
starting on high inertia loads may cause unnecessary tripping if the
starter is not well-matched with the motor (i.e. is too sensitive).
Thermal protection devices inserted in the motor windings

themselves can replace current-operated thermal elements in the
starter with advantage. Unnecessary tripping is avoided since they
automatically match the heating rate of the windings. Protection is also
provided against overheating not associated with excess current. Thus
fan motors in the air stream may be overheated either by excessive
temperature of the air handled or through accumulation of conveyed
material, dirt, etc., impairing the motor cooling, or through excessive
reduction of the air flow.
These devices take two main forms, each small enough to be
inserted between the coils of the stator end windings during
manufacture. The winding thermostat is a temperature operated
switch, commonly taking the form of a bi-metallic disc, snapping over
from concave to convex formation at a pre-set temperature. These can
be used in series with the supply connections so that they will directly
open the motor circuit without the need of a separate contactor on
single phase up to 1 kW or so
motor rating. At higher ratings thermostats in each phase can be used

in series to trip the starter.
The thermistor uses a PTC (positive temperature coefficient)
resistor in the form of a small bead of semi-conductor material. This
has a low heat capacity, and will therefore closely follow a rapidly
heated winding. One is embedded in each phase of the motor winding
and they are connected in series with one another and with an
electronic relay which will in turn operate the overload release of the
starter. The resistance of the PTC bead rises from a few hundred to
several thousand ohms over a range of a few degrees, the trip level of
1000 ohms being passed at a predetermined temperature. The
recommended temperatures to limit premature thermal ageing of the
insulation are:
8.6 Other Motor Features
8.6.1 Class of insulation

Insulating materials for electric motors are classified according to the
maximum temperature at which they will provide a satisfactory life. The
life is not exactly specified, acceptability being judged rather by ex-
perience with traditional materials in practical use over many years.
Accelerated life test methods are adopted by the motor manufacturer to
ensure that new materials at their limiting temperatures will be at least
as satisfactory as the traditional materials are at theirs. To ensure
mutual compatibility these tests are often done on model wire coils
inserted into slots by practical manufacturing methods.
The insulation class is therefore dependent on the whole insulation
system, not just the materials used. Nevertheless it is of interest to
note the following common selections for classes E, B, F and H.
The motors themselves are accepted on the basis of temperature

tests which determine the average temperature of the winding by
reference to its electrical resistance. With copper wire this changes
in the following ratio:
Since a failure anywhere will incapacitate the insulation system it is

obviously the temperature of the hottest spot which is really important.
However, this cannot be measured, and a conventional allowance is
made above the measured average. Standard motor ratings are for an
ambient temperature of 40°C, which covers the effective outdoor tem-
pe,ature over most of the world but makes no allowance for additional
air heating or for exposure to the sun's radiation.
Class E insulation is standard in most of Europe, Class B in the U.S.A.
Class F, which is a little more costly, may be used either to increase the
output rating of a given motor frame (by some 10% to 20%) in
standard ambient conditions or to increase the permissible ambient
temperature to 140°C less the actual temperature rise. Class H is
substantially more costly, and is used primarily for operation in ambient
temperatures above those permissible with Class F. Clearly Classes F
and H are particularly useful for motors operating in the air stream of
fans handling heated gases.
All organic insulating materials deteriorate with time, and the end
of their life is reached when degeneration has sapped most of their
mechanical strength and rendered them permeable to moisture, with

increasing ionisation and current leakage.
Roughly speaking their life is halved for every 8° to 10°C rise in

temperature although no guarantees can be given above the rated
maximum temperatures. The rule implies that a full rated standard
motor would lose a quarter of its working life-and therefore be at some
slight statistical risk of failure-by exposure to the following ambient
temperatures and times (including an allowance for increased motor
losses at higher temperatures).
High altitude will impair motor cooling, since a given volume flow of
air will carry away less heat at a lower density. Standard motor ratings
apply to any altitude up to 1000m. Above this 1% should be deducted
from the allowable temperature rise for each 100m above 1000m. The
motor manufacturer should always be consulted about derated or
specially wound motors since temperature rise is not simply
proportional to output.
8.8.2 Motor cooling and enclosure

A rather complex classification of methods of cooling (designated IC
with a two digit number) and protecting against damage to the
machine and injury to the user (IP) will be found in the publications of
IEC (International Electrotechnical Commission) and the corresponding
British Standard BS 4999 in its many parts.
The great majority of fans in the 1 to 100 kW range driven by

external electric motors, use totally enclosed fan cooled (TEFC)
machines. This form of enclosure protects the internal parts from
build-up of dust and damage, while the cooling fan and ribbed carcase
dissipate heat in most cases as efficiently as the through-air-flow of a
ventilated machine. In larger sizes internal and external air circuits
coupled by built-in heat exchangers come into consideration, together
with open ventilated systems if the surroundings are clean.
Axial fans with the impeller mounted on the shaft extension utilise
totally enclosed, air stream cooled motors. These are often TEFC
motors with the cooling fan and cowl omitted, and the TEFC rating will
be achieved provided the air velocity over the motor is not less than
the values indicated by Table 8.4. In fact the cooling is generally so
good that higher ratings are justified. These should be left to the fan
manufacturer to determine, since a number of factors are involved,
including particularly air distribution.
8.8.3 Bearing loading

A fan impeller may be mounted on the shaft of the driving motor or
on a separate shaft which is coupled to or belt driven by the motor. In
either case the shaft is usually carried in rolling bearings which must
carry both journal lie. radial) and thrust (i.e. axial) loads.
The thrust on an axial fan impeller is in the opposite direction to the
air flow through it. It may be taken as the product of the maximum fan
total pressure and the gross area of the impeller disc:
Strictly speaking there should be deductions for the static pressure

regain in guide vanes and over a downstream hub fairing if stationary
(not if attached to the impeller). Also an addition for total pressure
losses outside the impeller. Such refinements are rarely justified how-
ever, and the simple formula should err on the safe side.
In the case of a single inlet centrifugal impeller the thrust will be in
the opposite direction to the air flow entering the inlet. It may be taken
as the product of the maximum fan static pressure and the area of the
impeller eye, i.e. at the inside diameter of the shroud, De.
Once again this is a safe maximum value, since deductions

should be made for reduced pressure in the air behind the
backplate, most of which rotates at half the impeller speed. A
double inlet impeller develops no thrust.
Fig. 8.12a illustrates the forces acting on a horizontal rotor (4)

driving an impeller (3). The weight force W acting on a mass of m
kilograms is mg newtons where g is the acceleration of gravity (9.81
m/s2).
The radial reaction forces, or journal loads, Fr1, and Fr2 are found by
taking moments; the resulting formulae being given below Fig. 8.12.
Additional allowances are needed for unbalanced magnetic pull on the
rotor and mechanical out-of-balance.
The axial reaction force, or thrust load, of the bearing system F,,
must equal the aerodynamic thrust, T. It cannot, however, be shared
between two bearings spaced apart. Excessive manufacturing precision
would be required, and could not in any case be maintained against
thermal expansion. One bearing must have inner and outer races
axially locked to take the whole thrust.
For moderate thrust, bearing 2 is often the one locked, bearing 1,

which must carry the heavierjournal load, being freed of thrust by
giving a sliding fit to the outer race, or by making it a roller bearing.
For heavy thrust, however, it may be better to make bearing 1
capable of carrying the combined journal and thrust load. Bearing 2
is then a light duty type, and a ball bearing is to be preferred to a
roller bearing; axial spring loading in the housing can slide the outer
race to take up the inevitable clearance between balls and races.
Fig. 8.12b illustrates the case of a shaft inclined at an angle to
the horizontal. Forces W3 and W4 are resolved along and at right
angles to the shaft with the results summarised below the diagram.
The thrust loading is increased for fans blowing upwards (Form AU,
BU) reduced for fans blowing downwards (Form AD, BD).
When the impeller shaft runs in its own bearings its weight, W4
can be neglected. The bearings must not be too close together or
the increase in L1/L2 will make Fr1 excessive. A belt drive will apply a
substantial radial force to a pulley outside bearing 2. Being of
somewhat uncertain value it is generally sufficient to suppose that
the belt pull is wholly counterbalanced by an equal journal load Fr2
on bearing 2. Bearing 1 will then carry only the impeller journal load
while the thrust load will be unaffected.
The belt pull Tb depends on the power transmitted, Wm, the
effective diameter of the smaller pulley, Db, and its rotational speed
n, in accordance with this formula:
Kb is a factor accounting for the excess of the actual belt tension

over the minimum required to transmit the power. An estimate for
an accurately adjusted belt can be derived from the maker's data,
but the following range will cover the majority of cases:
8.8.4 Example of bearing capacity

The capacity of a ball bearing under combined radial and axial
loading is best explained by means of an example. Fig. 8.13 relates
to a 1600mm axial fan operating at 16.2 rev/s (975 rev/min) on a
D250 electric motor. The mass of the rotor is 153 kg and of the
impeller 92 kg and the fan is to be capable of working with the
spindle at any angle to the horizontal. The drive end bearing is a
70mm medium ball journal to carry the thrust load, and a life of
20,000 hours is required. Applying formulae 85 and 87:
The elliptical plot at the bottom of Fig. 8.13 shows the relationship
between Fr and Fa with T = 0 for all angles between + 90° (blowing
upwards) and - 90° (blowing downwards). Unbalance forces are
neglected.
The straight lines on Fig. 8.13 show the combined journal and
thrust capacity of the 70mm bearing at 975 rev/min for a life of
20,000 hours. A radial, journal, load of 7500 N can be carried with
no influence from thrust until the latter reaches 2100 N. From that
point on the radial capacity is steadily reduced until it has vanished
altogether when the thrust has reached 4900 N.
Reverting to the example, with horizontal spindle, point a, the
radial load is 1880 N and the thrust capacity ab = 4200 N. With
vertical spindle, blowing upwards, fg becomes the weight load of
2400 N and the remaining capacity for thrust, gh, is 4900 - 2400 =
2500 N. At an intermediate inclination the weights of the
components apply a radial load fc and an axial load cd to the drive
end bearing. This leaves a thrust capacity de for aerodynamic
loading, and de passes through a minimum of 2420 N when α =
65°, obtained by applying the condition tan α = (1880-2400) tan β
The gross area of a 1600mm diameter impeller is 2.01 m2. Therefore

the maximum fan total pressure which can be maintained at any angle
to the horizontal is 2420/2.01 = 1200 N/m2 or Pa for a bearing life of
20,000 hours.
8.8.5 Bearing ratings

The internationally accepted method of rating ball bearings for com-
bined journal and thrust loads is given in ISO 8281, and is used in
bearing makers' catalogues. It is a little complicated and Table 14.22
and Figs. 14.3 and 14.4 in Chapter 14 have been devised to simplify
estimation.
The table gives the pure radial, or journal, load that can be carried
for 20,000 hours at the speed quoted. Actually the life is not
determined in terms of hours, but of the total number of revolutions
made. Thus the life in hours will be shorter the higher the speed as
indicated by the nomogram of Fig. 14.3. This figure gives the factor KB
for ball or KR for roller bearings by which the basic capacity, PB, PA or PR
should be multiplied to take account of speed and life requirements.
Further multiplication by KJ and KT give the journal and thrust capacities
for any combination of the two. It will be seen that KT for a ball bearing
is dependent on the speed-life factor KB, but this is not the case for an
angular contact bearing.
The base speed no rev/min is so chosen that the static capacity Co in
the makers' tables is equalled by the dynamic capacity for 2000 no
revolutions. 2000 is an arbitrary multiple chosen to give reasonable
values of no, but by making it a constant one uniform diagram, Fig.
14.4, is obtained for all sizes and speeds of bearing.
It should be mentioned that, while identical externally, different
makes of bearing have differing internal ball and track designs. Basic
capacities therefore differ and the highest or lowest values are by no
means always peculiar to one maker. For Table 14.22 the lower ratings
have been selected from two major manufacturers' catalogues. For
precision, preliminary selections thus made should be checked against
the catalogue of the maker used.
A life of 20,000 hours is reasonable for ordinary general-purpose use.
It equals 10 years each of 50 weeks of five days of eight hours. Other
lives may be selected on the following lines:
Bearing life is a statistical concept only, and loads are rarely

maintained sufficiently accurately to warrant much precision of
estimation. The nominal life is the figure attained or exceeded by 90%
of a large batch on test. About 96% should survive for half the life and
50% may be expected to last five times the nominal value. Of course,
badly fitted, poorly lubricated, or dirty bearings can fail very quickly.
The comparative journal capacities of roller bearings are included in
Table 14.22. These, of course, have no thrust capacity. The speed
dependence of roller bearings is different from that of ball bearings, the
load for equal life being proportional to:
Angular contact ball bearings are very useful for increasing thrust
capacity. They are made with differing angles of contact between balls
and races, and the angle with the highest capacity usually available 40°
(from the radial direction) is selected for listing in Table 14.22. They
should be used in pairs or as a single double-row assembly and will then
take thrust in either direction as well as radial loads. The shape of the
combined load diagrams is no longer dependent on life, which leads to
simpler formulae.
8.6.8 Bearing lubrication

Ball and roller bearings require lubrication for three main purposes:
To prevent corrosion. While the balls and races are made from
corrosion resistant chromium steels, their highly finished rolling surfaces
would soon lose their perfection if subject to moisture attack.
To lubricate the rubbing surfaces of the cage against the balls or
rollers.
To preserve an oil film, even if only of molecular dimensions, to
separate the metal surfaces in spite of the high pressures to which they
are subjected. If this film is lost, local welding will occur, minute
particles will be torn from the surfaces, and failure will result.
Greases are commonly used, and should be of a premium grade
approved as a result of rig tests by the bearing manufacturers. Lithium
base greases with a rust-inhibiting additive have the most general
application. General-purpose grades may be used down to - 10°C and
up to 100°C grease temperature, which may correspond to 70° to 80°
ambient temperature with averagely well-cooled bearing housings.
Special greases containing diester or silicone oils may extend the range
down to - 40°C or up to 150'C. Standard bearings should not, however,
be subjected to more than about 125°C temperature, although specially
heat treated grades are available up to 350°C.
Bearing makers recommend that not more than 30% to 50% of the
available space in a bearing housing should be filled with grease. This is
particularly important at high speeds where the churning action of
the balls forcing their way through would raise the temperature and
cause rapid loss of oil. Where lubricators are fitted this means that an
escape path for excess grease must be provided-either through a
clearance around the shaft or by means of a relief valve. Dirt is fatal
to all bearings, and great care must be taken to prevent its entrance
along with fresh grease.
Bearings and bearing housings cannot be "sealed for life" except
for very. intermittent use or low speed. The grease will deteriorate
and must be periodically replaced, either through lubricators or by
dismantling, washing and repacking the bearings with fresh grease.
The latter procedure is to be preferred if performed with due care to
avoid damage, but re-lubrication through nipples or "Stauffers" is
serviceable enough provided it is done with care and discretion. There
should be a regular maintenance schedule, and the lubricators must
be operated neither too often nor at too long intervals.
The re-lubrication intervals recommended by one bearing maker
are summarised in Fig. 8.14. The chart applies to ball bearings at
grease temperatures not exceeding 70°C, which may be taken as
corresponding to 50°C ambient at low or 40°C ambient at high
speeds. The following factors may be applied to the re-lubrication
interval at other temperatures, and roller or angular contact bearings.
A grease is a honeycomb structure of tiny cells containing oil, the

walls being thin membranes formed by emulsification with sodium,
calcium or lithium soaps. The function of the cellular structure is to
hold the oil in place in bearing housings from which free oil would
escape. In contact with the moving elements of the bearing the cells
are ruptured and oil is released. A grease that is too "soft" will lose all
its oil too soon. A "hard" grease will not release it quickly enough for
the application.
It is dangerous to re-lubricate with a different grease from the one
already in the housing. If they are incompatible the resulting mixture
will be unduly soft. On the other hand, grease may harden simply
through standing idle in the housing. The grease should be examined
if a new machine has stood for a year or more before entering service.
The lubricating properties of a grease are entirely due to the oil

content, and the oil can be used perfectly well by itself for rolling bear-
ings. The difficulties are solely those of a practical nature. Efficient seals
are necessary to prevent oil from escaping, and this makes horizontal
spindle operation simpler than vertical. An oil bath should reach the
centre of the lowest ball or roller and is suitable for low speeds only. At
higher speeds the oil should be circulated through a filter, and changed
as oxidation develops. The highest speeds necessitate an oil mist
directed at the bearing from an atomiser fed with dry compressed air.
High temperatures can be dealt with by passing the circulating oil
through an oil cooler, or by water cooling the housing.
Good quality mineral oils are used for the oil lubrication of rolling
bearings, and it is important that the oil viscosity should have the right
value at the working temperature. This value is dependent on the linear
speed of the balls or rollers, and on the load. Too low a viscosity would
violate the hydrodynamic conditions for establishing an oil film at the
rolling surfaces. Too high a viscosity would cause excessive friction. Fig.
8.15 gives one maker's recommendation of optimum viscosity at
working temperature taking the loading corresponding to 20,000 hours'
life. For well-cooled bearings the working temperature and correspond-
ing viscosity might be taken as 40°C. Oil makers will quote the viscosity
at various temperatures, but SI units may not be used. A comparison
with Redwood and other scales is given in Table 14.15, Chapter 14.
Plain oil-lubricated sleeve bearings are still used for some large low
speed fans, or for very small domestic fans in the interests of silence.
Their engineering aspects are outside the scope of this book.
CHAPTER 9
Fan duty control and

energy saving
Speed control of the fan driving motor was discussed in the last
chapter. Other methods of controlling the performance may well be
more favourable in particular circumstances, and a comparative
study is the subject of this chapter.
Often the full volume and pressure for which the fan is selected
are not required continuously and to leave the fan operating as if
they were is very wasteful of energy. Half or more of the
kilowatt-hours supplied to the motor can commonly be saved by
adjusting the fan performance automatically to match requirements
and fuel economies resulting from control of the air duty may
multiply the savings. A significant UK government exercise was
reported in 1974. 306 office and public buildings were fitted with
advanced heating control systems at an average cost of £1,500:
the resulting annual saving in fuel costs per building was £860 (that
is 30% to 50% of the original annual cost). This is good business
under any accounting system.
Variable volume flow is a feature of several of the best air
conditioning control systems currently available and its inclusion in
new plant will enable advantage to be taken of new ideas in a
rapidly developing field.
In the chemical, food processing, and many other industries
control is necessary to maintain consistent quality. Variations in
throughput, raw material analysis, ambient temperature, etc. must
be compensated, and if this can be done by adjusting the flow of air
or gas there will be accompanying savings in circulating energy.
171
9.1 Step-by-Step Systems

9.1.1 Multi-speed motors, described under 8.2.4, page 150 are probably
the lowest in first cost of all control means. They are also, at the speeds
available and in applications with constant system resistance, the most
efficient. This is because the fan operates at a constant point on the
non-dimensional system characteristic so that both static and total
efficiencies remain constant, while the motor efficiency is very little
below the optimum for each duty. They are not suitable for constant
pressure applications since, if the pressure can be developed at low
speed, the fan will be operating well below optimum pressure at high-
speed.
The chief drawback of multi-speed motors for fans is the size of the
step to the first regulated speed. This may prejudice their consideration
except for simple applications such as ventilation where the criteria for
successful operation are somewhat vague. The maximum ratio of the
first regulated speed to full speed depends on the pole number of the
motor at full speed (see Table 8.1) in the following manner:
In the case of belt- or gear-driven fans it may be noted that some

choice is possible by selecting the top motor speed.
9.1.2 Fans in parallel can obviously be used for flow control by switching
off one or more of the units. This system is mainly used with axial fans
for relatively large volume flows. A line of parallel fans may also best
suit the building or plant layout.
It is essential to block the passage through the fan casing against

reverse flow when the fan is not running. This should be done by fitting
air- or motor-operated automatic shutters at inlet or outlet to close
while the fan is slowing down.
It is best also to pass the combined air flow at relatively low velocity
through a plenum chamber between the fans and the system and to
choose fans having a fairly low ratio of outlet velocity pressure to fan
static pressure-secured if necessary by fitting outlet diffusers. Otherwise
the mismatch, inevitable at some duties, between the fan outlet velocity
and the system velocity will waste energy.
The characteristics of fans in parallel are illustrated in Fig. 9.1. for the
case of three identical units. Volume flows at constant pressure are of
course proportional to the number of fans running. The dotted system
characteristics show that constant pressure is ideal for fans in parallel,
fan efficiency being maintained at peak value. In such a system fans
would be switched in to maintain the plenum pressure applied to a
series of resistive outlets as fresh outlets were opened up. On the other
hand constant resistance systems can only be operated efficiently over
a fairly small control range.
Fan duty control and energy saving 173
A stable characteristic is important for fans in parallel. Fig. 9.2 illustrates the
danger of using a fan with a marked drop in pressure at restricted volume
flow. The dotted system resistance line shows the volume flow per fan on the
assumption that two fans share the volume flow equally between them,
operating at A. If one fan is shut down the other will operate at point B,
carrying the whole volume flow in place of the half share plotted at E. If the
shut down fan is now re-started it will be unable to move out of the stalled
region finishing up at C, while the other operates at D. The average volume
flow per fan is now at F well short of the intended volume and pressure point
A.
9.1.3Fens in series can provide for duty control if of the contra-rotating axial type
(see 7.6.4 (F) ). When the motor driving one of the impellers is disconnected from
the supply, it will idle at about two-thirds of full speed, absorbing around 50% of
fan velocity pressure in aerodynamic loss. Centrifugal fans and axial fans with
guide vanes have much greater idling losses and are not used for this form of duty
control.
As indicated by Fig. 9.3 fans in series are best suited to pressure control at fairly
constant flow rate (constant volume system) ; not a great deal of fan efficiency is
lost on a constant resistance system, however. Since the pressure rise across each
driven impeller depends on the swirl
preceding it, the overall pressure is not simply proportional to the number of
impellers driven. The switching order is important to avoid excessive counter-swirl
which would overload one of the motors, and the makers' advice should be sought
on this point.
Contra-rotating fans in series are best suited to applications requiring pressures
of 1000 Pa upwards. Simple standard impellers can be used, not the more
specialised mechanical and aerodynamic designs needed to develop such
pressures in a single stage. Motor costs are not much affected over 10 kW rating
(two 50 kW motors may well be slightly cheaper than one for 100 kW) and rather
less noise is generated.
The added facility of step-by-step duty control is, therefore, obtained very
reasonably.
9.1.4 energy saving. Throughout this chapter a uniform system will be adopted
for assessing the efficiency of a control system. The reference standard will be the
energy consumed in the absence of control - that is to say the power input to the
driving motor at full fan duty multiplied by the time per day, week or year during
which any air flow is required.
As a rule the volume or mass flow is the quantity actually required by the user.
This requirement is supposed to vary over a control range expressed as a
percentage of the maximum flow, at full fan duty. On the rarer constant volume
systems the fan total pressure is the variable. For comparison purposes it will then
be assumed that any volume flow within the control range is equally likely - if the
range is split into a
number of equal intervals of volume flow, an equal time is spent in each. The
case where the distribution of volume flow with time is known is touched on in
9.6.
A constant system resistance case is sketched in Fig. 9.4. The pressure
varies with the square, and the power input with the cube of the volume flow,
and the latter is plotted along 1 2 3. A two-step control is considered, without
significant variations in fan or motor efficiency. The fan operates at full power
until the requirement has fallen to point 2, from which point the reduced
power suffices until the end of the control range at 3. The energy saved by
the two-step control is represented by area D (power reduction x proportionate
time at reduced power). The reference power is the whole area A + B + C +
D and the percentage ratio giving the control saving is 100D/(A + B + C + D).
For any control range 1 to 3 there is an optimum volume flow 2 at which to
make the change from full to reduced power for maximum
control saving. This can be calculated and its location is shown by the broken
line on Fig. 9.5, as a function of the control range. As well as the two-step
case with optimum location of change point 2, the optimum three-step case
has been calculated, and also the most efficient control possible, continuous
control, in which the only energy used is that for the pressure and volume flow
actually required from moment to moment.
Finally, the control savings for two-speed 4-6 pole (2:3) and 6-8 pole (3:4)
motors are shown. While ineffective for a limited volume reduction it will be
seen that these are practically as good as the optimum two-stage change point
for control ranges exceeding 60% and 40% respectively. This would not be
the case, of course, if more than proportionate time were spent at the higher
volume flows.
9.1.5 Example of stage control. The three methods of duty control described in
9.1.1, 9.1.2 and 9.1.3 are combined in this example. The fans are used to
ventilate a 1.5 km vehicle tunnel with a capacity of
2,000 cars per hour joining Kowloon and Shatin at Hong Kong. There are supply and
exhaust units at each end of the tunnel and each of the four plants consists of 8
stages of 1500mm axial fans in four parallel sets of contra-rotating pairs. Each
impeller is driven by a 40 kW 970 - 485 rev/min two-speed induction motor.
The fans are connected to supply and exhaust ducting on the transverse principle
(see Chapter 13), and the carbon monoxide content of the air is monitored at
several sampling points in the tunnel. By switching the fans in different
combinations the volume flow per unit can be varied from 15 to 140M3/s in eight
steps according to the pollution level, the top rate being for emergency use only in
case of a vehicle fire. Since all the stages are not normally required, spares for
maintenance or breakdown are minimised.
Fig. 9.6 shows the fan characteristics for the combinations usually employed, and
their working intersections A to H with the system
characteristic of the installation. The corresponding data for each operating point
are listed in Table 9.1. It will be seen that the average daily energy consumption
of 3200 kWh for the four plants compares with 103 x 24 x 4 = 9900 kWh to
maintain the full normal duty of 90m3/s, a control saving of (9900 - 3200)/9900
= 68%.
9.2 Continuous Speed Control

9.2.1 A variety of fan drives by electric motors capable of stepless speed variation
were discussed in Chapter 8. It is also possible for a constant speed motor to drive
a variable speed fan through some form of steplessly variable transmission. Many
different principles have been exploited and examples are to be found driving fans
in test and research laboratories and in heavy process equipment where
specialised mechanical design is justified.
However, they are not often found in general-purpose installations. The
standard fan and its electric motor drive are commonly capable of 20,000 to
50,000 hours' life with the bare minimum of maintenanceeven sometimes with
none at all. Variable speed transmissions are seldom designed with that kind of
reliability in mind and this may be the reason for their comparative rarity. It is also
the case that nearly all variable duty fan systems are designed for automatic, or at
any rate remote, control. A variable ratio fan transmission should therefore be
capable of operation by pneumatic or electric servo motors or thrusters.
9.2.2 Examples of variable speed transmission

Some of the commonest systems are briefly listed here:
V-belt drives are made with pulleys of variable effective diameter secured by
varying the width between the inclined flanges. Ratings up to about 10 kW are
available, and this is the design most commonly found in test laboratories with
manual control-although units with motorised control are also available.
Fluid couplings can be operated with continuous slip if fitted with scoops or
other means of controlling the internal fluid circulation.
Losses are typical of slipping devices and they are most useful to ease the
motor starting load with a capacity for occasional short time regulation.
Torque converters are similar in effect to fluid couplings with other than 1 : 1
speed ratio. Away from the speed reduction for which they are designed they
entail losses of the same order as those of slipping devices.
Eddy-current couplings have a solid iron driven element carrying a magnetic
field generated by the driving element. Currents in the iron similar to those in a
squirrel cage rotor arise as the slip increases and produce torque which is capable
of close control by varying the relatively small magnetising current. The fan speed
changes to balance the applied torque.
A fluid-power system has two elements which can be located a little distance
from one another. One is a positive displacement pump driven by a constant speed
motor but producing nevertheless a variable flow of oil through a variable stroke
piston or other means. The oil is passed through high pressure piping to a
constant displacement hydraulic motor driving the fan, whose speed will be
proportional to the rate of oil flow.
Variable ratio gearboxes are made, usually, with a rolling element under
pressure to provide a friction drive at an adjustable radius between driving and
driven surfaces.
To these must be added, for heavy industrial use, steam- or gasturbines,
geared down to fan speed, and controlled by throttle and governor systems.
9.2.3 Power consumption with continuous speed control

Of course, every control system has its own particular sources of energy loss,
dependent also on size, speed, and speed ratio. Nevertheless, two broad
categories can be recognised each of which has a basic minimum power
consumption over and above which the losses incidental to the particular detailed
design must be added.
Variable ratio systems:
Variable V-belt pulleys; variable ratio gear; fluid power drives; variable
frequency motor supply; DC and AC commutator motors.
Variable slip systems:
Fluid couplings; torque converters; eddy-current couplings; wound rotor
induction motors.
For a constant resistance system the fan efficiency will remain constant over
the speed range. The basic power input will fall in proportion to the cube of the
speed ratio for variable ratio systems as plotted in Fig. 9.7. In the case of
variable slip systems the full fan torque is borne by the driving motor without any
"gearing" ratio" reduction; this means that the basic input power is proportional
to the square of the speed ratio.
The losses additional to the basic power input will be those in the driving
motor and in the transmission system which can be determined from the overall
efficiency of the assembly at full fan speed. In the absence of detailed knowledge
it is a safe assumption that these additional losses will remain constant over the
speed range. In fact most of them will fall with speed and load, but not by any
means in proportion.
9.3 Damper Control of Fan Duty

9.3.1 Most air conditioning systems and many industrial installations employ
dampers to control the air flow, and in consequence to vary the fan duty. While
these are effective for their purpose and indeed necessary
when division between branches is to be adjusted, they quite fail to realise the
potential savings of energy unless accompanied by control of the fan itself.
All dampers serve simply to throttle the flow in the airway in which they are
placed. Back pressure is built up by accelerating the air to high velocity and then
allowing its kinetic energy to be dissipated in turbulence downstream. They do not
have an "efficiency" in energy terms but their characteristics in respect of noise
generation, stability, and smoothness of control are important.
While most dampers are merely placed across the airway in which the airflow is
to be reduced, an alternative exists which may or may not be preferable according
to the fan characteristic and other circumstances. This is the by-pass duct
containing a damper which, when opened, permits some air to flow back from the
fan outlet to the inlet, thereby reducing the flow in the main system. The same
effect can be secured by simply opening a damper in the airway wall downstream
of the fan to "bleed off" by-pass air, or alternatively to open an entry port for
by-pass air in a suction airway upstream of the fan. In both the latter cases
possible disturbance to conditions outside the duct must be considered.
9.3.2 Influence of the fan power characteristic

In the absence of any other form of fan duty control, variation of volume flow
by the closure of a damper will move the operating point to a new position on the
fan characteristic. This will involve a change in fan efficiency (generally for the
worse) and a change in the fan power input. Classification into fans with rising,
falling, or non-overloading power characteristics is important in this respect.
Forward-curved fans have rising characteristics, particularly so in the multi-vane
type which may have an impeller power at full open flow six or eight times that at
no flow. High pressure blowers, paddle bladed fans and others with radial outlet
blading have a rising characteristic which is less marked, four or five being more
typical ratios of maximum to minimum power.
Falling characteristics are peculiar to the propeller fan which, if allowed to
develop centrifugal outlet side flow and corresponding pressure in a diaphragm
mounting, may take at no flow 1½ to 2½ times the free flow power.
Non-overloading characteristics are usual in backward-curved centrifugal and
axial designs, particularly those with high efficiency. The power taken at the best
efficiency point on the characteristic is quite close to the maximum, and then falls
away both towards free flow and no flow. Typical power characteristics for these
classes are compared in Fig. 9.8, taking the volume flow and power input at the
best efficiency point as 100% in each case.
Assuming that fans have been chosen to operate near the best efficiency point
at full duty, it will be clear that series dampers should be used for fans with rising
power characteristics, so that fan flow and system flow fall together and the
power input is reduced. On the other
hand series dampers, or indeed any undue flow restrictions, are bad for a
propeller fan; if duty control is necessary the by-pass or bleed-off method is
always better for this type-if it can be arranged.
The backward-curved fan is more flexible. There is little prospect of power
saving however, what there is being usually better with series than with by-pass
damper control. Axial fans depend very much on the
blade pitch angle; low volume, shallow pitched blading places them in the same
class as backward-curved fans. High volume, steep pitched impellers on the other
hand should not be subjected to much volume reduction by series dampers or the
unstable, stalled flow region may be entered. By-pass control is better, and may
permit power saving of 20% to 40%.
9.3.3 Performance with damper and by-pass control
Figs. 9.9 and 9.10 illustrate the application of damper and by-pass control
respectively to a constant resistance, "square-law" system
characteristic. The most suitable fan designs for power saving are selected for
the illustrations, and it will be seen that a reduction of impeller power slightly
greater than the reduction in volume flow is achieved in both cases. The
electrical losses in the driving motor are not likely to fall quite so fast, and it
must also be noted that a forward-curved fan for series damper control is
unlikely to be as efficient at full duty as an axial fan for by-pass control.
It is interesting to draw a comparison between the best constant speed

fan-damper or fan by-pass combinations and variable ratio or variable slip
speed control. On constant resistance systems the ratio of input power on
regulation to full duty input power will be approximately equal to:
If the system requires constant volume flow with a fluctuating system

pressure, a series control damper may be used with the fan remaining at full
duty and power input throughout. However, power can be saved by
adopting by-pass control and an axial or other fan with a falling power
characteristic. The by-pass is adjusted to dump the surplus air when the pressure
required is low.
To maintain constant minimum pressure in an enclosure (e.g. a dustfree
workshop) with a varying air change requirement the control damper should be
placed in the outlet from the room. By-pass control of the supply is then
functionally possible, but it would be simpler and more economical to choose a
backward-curved or other fan with a
fairly flat fan static pressure characteristic and a rising power characteristic over
the volume flow range required. The outlet damper would then suffice to control
volume flow and efficiency could remain quite high.
Many designs of damper, when partially closed, deflect the air strongly over to
one side of the duct in which they are placed (or to both sides in the butterfly
case). Placed less than, say, five duct diameters upstream of a fan inlet these will
badly affect its performance; if closer spacing is necessary the damper should be
of the opposed vane or similar symmetrical, non-swirl-generating type. Similar
considerations apply to the flow from a by-pass duct entering the suction duct
close to a fan inlet.
Where a by-pass is installed to accept part of the flow velocity inequalities

approaching the by-pass will introduce substantial branching losses (see Section
6.7).
9.4 Variable Pitch Fans 9.4.1 General

A principal factor determining the performance of all fans-axial, mixed flow, or
centrifugal-is the angular setting of the blades, particularly the angle made by the
outlet edge with the tangent to the direction of rotation (α2 in Fig. 7.24). As this
angle is increased the forward component of air velocity on which the volume flow
depends is increased while the swirl component on which the pressure depends
remains substantially constant. It follows that the volume flow could be controlled
if the angular setting of the blades could be altered while the impeller was
rotating.
The variable pitch aircraft propeller and the Kaplan runner for hydraulic turbines
show that pitch adjustment in the axial configuration can be developed to the
highest levels of reliability. Fans with variable pitch axial impellers have been in
successful service for many years, though in an unnecessarily limited range of
application. With increasing emphasis on flexibility of control and energy
economy, both in air conditioning and process work, their use is spreading.
Centrifugal impellers have been produced with variable pitch trailing edges to the
blades, rather after the pattern of aircraft wing-flaps. However, the mechanical
design problems of this configuration are severe and it is seldom encountered.
Variable pitch axial fans can be used with constant resistance, constant volume
flow, or constant pressure systems-in fact with any form of system characteristic.
The overall energy savings are among the highest available with any form of duty
control. A unique feature is the ability to control volume flow down to zero even at
constant pressure, and to produce reverse flow if required. Standard pneumatic or
electric continuous control system can be employed, and the speed of response
tailored to requirements. Noise level falls with reduction of volume flow, whereas
it tends to rise with damper or vane control; the fall is not nearly so rapid as it is
with speed control, but the use of a multi-speed motor with a variable pitch
impeller largely bridges this gap.
9.4.2 Performance of typical variable pitch fan

The aerodynamic design of a variable pitch axial fan can be exactly the same as
that of the general purpose adjustable pitch range described in Section 7.6. Fig.
9.11 is an example: a 1600mm fan with downstream guide vanes and a diffuser
expanding to a 2000mm diameter outlet duct in which the fan performance is
measured. The fan total pressure and inlet volume flow are plotted for a pitch
angle range from 32° (at the blade tips) to - 8°; this range could be extended if
negative flow was required. Contours of constant fan total efficiency are also
plotted from the peak of about 86% down to 50%.
The broken lines A, B, C and D are examples of the range of system
characteristic which can be dealt with by this fan.
A is a constant resistance characteristic for which the required total

pressure drop equals:
This is the commonest system, and applies when the required volume flow
is delivered (or exhausted) through a system of airways, filters, heater
batteries, etc. with the specified total pressure drop varying with the square
of the volume flow. In the example the volume flow can be
continuously and stably adjusted from 50m3/s to zero - or to reverse flow if

required. The fan efficiency is between 75% and 85% from 50 down to 31
m3/s, that is from 66 down to 16 kW impeller power input.
B is a part constant pressure, part constant resistance characteristic with a
total pressure drop equal to:
Characteristics of this type are met with in air conditioning systems requiring a
fixed pressure (here 300 Pa) to operate the final room outlet distribution and
control units. The rest of the pressure drop arises in the airways and equipment
supplying the outlet units. The example shows a volume flow adjustable from
46M3/s down to zero with 75% to 85% efficiency from 46 to 21 m3/s - 60 to 13
kW input.
C is a constant pressure system maintaining a constant static pressure of 600 Pa

in the 2000mm duct. It is assumed that the velocity pressure at the outlet of this
duct (which equals 0.06 Q2) is all lost (though some could well be recovered) which
leads to the following total pressure requirement:
Characteristics of this type arise when the fan supplies a plenum chamber
feeding a number of outlets, only some of which are in service at any one time.
These may be process supply points, each requiring a fixed volume flow and
pressure, or the plenum may be on the suction side of the fan, drawing from a
varying number of fume cupboards, for example. In air conditioning the supply
might be to a number of floors in a building, each with its own conditioning unit,
and each operating for different periods according to occupancy. The fan pitch
would be automatically adjusted to maintain constant pressure or suction in the
plenum. In the example the volume flow may be adjusted from 60m3/S to zero and
75% to 85% efficiency is maintained over the wide range of 58m3/s to 18m3/s - 62
to 15 kW input.
D is a constant volume system maintaining a volume flow of 40m3/S over a

range of total pressure drop from 1000 Pa down to zero (or negative pressure if
required). 75% to 85% efficiency is maintained over a pressure range from 1000
to 400 Pa with corresponding power input from 53 to 22 kW. This type of
characteristic is required when a specified volume or mass flow is to be maintained
against a variable flow resistance, as might arise for example in the fuel bed of a
combustion process, in a flotation chamber or pneumatic transport system for
powered material, or across a filter as it becomes fouled. The fan pitch would be
automatically adjusted to maintain constant mass flow irrespective of pressure, and
the set flow rate could be altered from time to time as required.
9.4.3 Applications of variable pitch fans

The versatility of the variable pitch axial fan will be evident from the above
example. It will be seen that over 75% efficiency is maintained over a power input
range of 100% to 25% in each variable volume case; efficiency is still 50% at less
than 10% of full power. It is not necessary for the whole duty to be dealt with by
one fan. Large volumes or pressures can be dealt with by using one or more
variable pitch fans in parallel or series with a number of similar fixed pitch units,
which are switched in and out to provide coarse control steps, the fine control
between steps being provided by the variable pitch unit.
Among successful applications of variable pitch axial fans may be cited:

Air conditioning systems in a variety of designs for commercial and public
buildings.
Vehicle tunnels and garages where control of carbon monoxide and smoke
is required.
Underground railways for station ventilation and heat removal.
Power stations for forced and induced draught and gas circulation.
Heavy industry, chemical and metallurgical, for process control, drying and
cooling.
Textile industry for humidity control.
Oil industry for large scale "fin-fan" heat exchange.
Laboratories for fume cupboards, wind tunnels, etc.
Mines for main ventilation.
Factories for ventilation and pollution control.
9.4.4 Mechanical features of variable pitch fans
Centrifugal forces are predominant in the mechanical design. The
centrifugal force on a mass of m kg revolving at n rev/sec at a radius of r
metres is:
For the 1600mm 975 rev/min fan of 9.4.2 2∏n = 102 radians/sec. Taking M,
the mass of one blade, as 3 kg, and its radius of gyration, rg, as 550mm, the
centrifugal force per blade will be:
This is nearly 600 times the weight of the blade, which is 3 x 9.8 = 29.4
newtons and the situation is sometimes spoken of as equivalent to operating
in a gravitational field of 600g. g = 9.8m2/s is the acceleration due to gravity
at the earth's surface.
Such a force, and indeed forces several times as great, are within the static
capacity of ordinary ball-thrust or needle roller thrust bearings, and these are
commonly used to carry the centrifugal force at the blade root while
permitting blade rotation with minimum friction. The taper-roller thrust
bearings used for aircraft propellers are normally unnecessary for fans.
Levers at the base of each blade translate the common pitch angle
adjustment into axial movement of a sliding member within the hub. This may
be actuated in four ways:
(a) Automatically, by the expansion of a pneumatic bellows of reinforced
rubber within the hub against a spring. The bellows is fed with
compressed air through a rotary air seal on the shaft extension.
(b) Automatically, by an external pneumatic or electric thruster through a
lever system which applies pressure to the stationary race of a ball thrust
bearing the revolving race of which is coupled to the sliding actuator
within the hub.
(c) Automatically, by an external pneumatic or electric thruster which applies

actuating force as in (b) while the reaction force is transmitted from the
hub to the body of the thruster through a second thrust bearing. This
arrangement relieves the main fan bearing of control thrust load.
(d) Manually, by means of a screw jack when the fan is at rest.
Fig. 9.12 shows the cross-section of the hub of a variable pitch axial fan
with inbuilt pneumatic control of type (a). The air pressure in the bellows is
adjustable by the control system to any value between 20 and 100 kPa above
atmospheric pressure. For each value there is a corresponding compression of
the spring, position of the sliding actuator, and pitch angle of each blade.
When the fan is running forces must be applied to each fan blade to keep it at
the required pitch angle. Left to itself it would rotate to a position near zero
pitch angle where the centrifugal forces on it were in balance. Weights W are
attached to the blade root in such a position (at right-angles to the blade) that
they apply a counterbalancing turning moment and minimise the actuating
force required.
9.5 Vane Control

9.5.1 Principles of vane control
A set of movable vanes is used, placed in the airway at or near the inlet to the
fan. They are usually pivotted on radial axes arranged like the spokes of a wheel
and are linked together so that they can be moved in unison. Unlike a damper
they do not merely throttle the airflow but are designed to produce a controllable
degree of swirl in the air entering the impeller. This has the effect of altering the
fan performance - moving the fan characteristic across the pressure-volume flow
chart in fact. They are thus more efficient than plain dampers, and with some
types of fan they are capable of raising as well as reducing the normal fan duty.
In Section 7.15 it was shown that the total pressure rise across any fan
impeller-axial, mixed flow or centrifugal - was given (ignoring losses) by:
Vane control operates on the inlet swirl angle α1 With the vanes in the neutral
position the inlet flow is axial, cos α1, and the second term are zero. Swirl in the
same direction as the impeller rotation (a, less than 90° and cos α1 positive) will
reduce the pressure rise. Swirl in the opposite direction (a, more than 90° and cos
α1 negative) may raise the fan pressure, unless the impeller blading is incapable of
carrying the increased deflection so that α2 is not maintained.
The impeller power will be decreased or increased in proportion to the
pressure. In the first case some of the torque needed to produce the outlet swirl
α2 will have been transferred from the impeller to the vanes; in the second case
the impeller has to apply additional torque to counteract that applied by the
vanes. Since the guide vanes are not rotating they can apply torque without
consuming power, unlike the impeller.
9.5.2 Vane control for centrifugal fans

Vane control is the standard aerodynamic method of varying the performance
of a fixed speed centrifugal fan. The radial vane system just described is the most
usual, although other arrangements are possible. With flat plate vanes the
resistance when set in the neutral position, 90°, is small so that the normal fan
characteristic is hardly affected. Vane control is most effective with high efficiency,
backward-curved, centrifugal fans, and a typical set of characteristics for such a
combination is shown in Fig. 9.14.
Hardly any increase on normal performance is obtainable and angles greater
than 90° are not used in this case. The duty may be controlled down to zero with
constant resistance, constant pressure, or constant volume flow system provided
the vanes are capable of being set for complete closure as dampers; reverse flow
is not obtainable. The method is best suited to rather small ranges of duty control.
On a constant resistance system, for example, the efficiency has dropped from
85% to 75% with only 10% reduction in volume flow and 30% in impeller
power. The equivalent variable pitch axial fan of Fig. 9.11 covered 38% volume
reduction and 75% power reduction for the same change in efficiency.
9.5.3 Vane control for axial fans

Vane control is seldom used for axial fans, being inferior in efficiency and range
to variable-pitch control. However, it is lower in first cost and could provide a
useful method of regulation for the smaller, simpler units. Some 10% to 20%
increase above the normal fan pressure or volume flow is usually available at
negative settings, though at reducing efficiencies and increasing noise level.
The method is best suited to constant volume flow operation with reasonably
efficient control of pressure down to zero. On constant system resistance a typical
unit could be regulated to 60% of normal volume flow at 36% pressure with 30%
to 40% reduction in power consumption. At lower flows the power taken rises
towards equality with the value for damper control when the vanes are completely
closed. At constant pressure vane control is likely to encounter instability problems
unless the impeller pitch angle is very small.
9.6 Comparative Energy Savings
9.6.1 Example of control at constant resistance

Fig 9.15 gives a direct comparison of most of the methods previously discussed.
The maximum duty is supposed to be 50m3/s at 1000 Pa fan total pressure as in
Figs. 9.11 and 9.14. The maximum available control of volume flow is required on
a system in which the pressure varies with the square of the volume flow.
The "ideal control" curve represents an unattainable ideal solution in which a
fan with an efficiency of 85% maintained at all control volumes is driven without
either motor or control losses. At 100% volume flow the power input of 59 kW is
taken as 100% and reduces with the cube of the volume flow.
The band marked SP covers the probable range of electrical input power with
the more efficient forms of speed control discussed in Section 8.2, these include
AC commutator (Schrage) motors, induction motors with thyristor control of
frequency, or DC motors with thyristor control of voltage. The fan will retain its full
efficiency of 85%, but the electrical losses will be greater than those of a constant
speed 60 kW induction motor, which would have an efficiency of about 92%.
The band marked SL covers the probable range of electrical input power of
variable slip systems. These include wound-rotor induction motors, and standard
induction motors driving fluid couplings, torque converters and eddy-current
couplings, and the power input is seen to be considerably greater. Neither SL nor
SP have been taken below 30% volume flow because of the probability of
instability with most methods at lower speeds.
The damper control curves for backward- and forward-curved fans follow the
respective power characteristic examples of Fig. 9.8, with motor losses added
assuming motor efficiencies to be:
In the forward-curved case a probable maximum fan efficiency of 70% at full

duty has been taken, compared with 85% for the backward-curved and axial
cases. By-pass damper control would be used in place of series damper control for
the axial fan with its falling power characteristic, and this limits the volume flow
reduction to 40% in the example of Fig. 9.10. Vane control for a backward-curved
centrifugal fan (as Fig. 9.14) will introduce a little extra fully open loss, but is
clearly better than any type of damper control. At zero flow vane and damper
control coincide, merely serving to close off the system.
The variable pitch axial fan shows to great advantage over almost the whole
control range; the example of Fig. 9.11 has been taken. The power input at
maximum duty is up a little, corresponding to an efficiency of 76% at full duty.
This is because the ability to increase performance by raising the pitch angle from
the optimum 20° to 32° has been used, thereby maintaining efficiency over a
100% control range. For a
narrower range, e.g. 50% volume flow, the 20° pitch angle could be used for
maximum duty, giving 85% efficiency there, the fan diameter being increased
from 1600mm to 1800mm at the same speed.
In Fig. 9.15 "ideal control" represents the mechanical input to a fan which
maintains a constant efficiency of 85% over the whole control range. All other
curves include the electrical and other losses in the driving motor and control
means.
8.6.2 Example of control at constant pressure

In examples summarised in Fig. 9.16 a constant total pressure requirement has
been assumed to avoid the complication of allowing for the recovery of any
particular proportion of the fan velocity pressure. The
symbols have the same meaning as in Fig. 9.15, IC being the power absorbed by
a fan of constant efficiency (85%) with no allowance for motor or other driving
losses, conditions which are of course unattainable in practice. At constant total
pressure, power is then clearly proportional to volume flow.
Of damper control, systems the by-pass (curve BP) is effective for control of
volume flow over a range up to 25% reduction; it is applicable to any fan, which
continues to operate at full duty and efficiency, the unwanted volume flow being
simply dumped. There is no control saving, however, and for a large control range
series dampers will be more efficient for centrifugal fans in which there is a
substantial fall of power with volume flow. Indeed, if a forward-curved centrifugal
fan is used, series damper control is best for all ranges (curve DP FC noting that
70% rather than 85% fan efficiency is assumed).
For a backward-curved centrifugal fan vane control (curve VC, BC) shows to best
advantage on a constant pressure system, the example of Fig. 9.14 at 900 Pa
coming within 5% of the optimum energy saving for control ranges up to 50%
volume flow reduction. The variable pitch axial fan is even better, however,
maintaining a good efficiency over a 100% control range. It should be noted that
curve VP AX in Fig. 9.16 (corresponding to the example of Fig. 9.11 at 800 Pa) is
chosen to suit a 100% control range. For a narrower range the increased pitch angle
of 32°, which accounts for the increased power consumption at 100% volume flow,
need not be used. A slightly larger fan could be operated over the 80% to 40%
range of Fig. 9.16, giving within 3% of the optimum saving over a 50% volume flow
range.
9.6.3 Examples of control at constant volume flow

For an application in which a constant volume flow is required, the ideal,
unattainable power consumption will clearly be proportional to the total pressure
required to maintain the flow (curve IC in Fig. 9.17 which is again the impeller power
at 85% constant fan efficiency).
As the pressure required to maintain flow falls the excess developed by the fan
can be absorbed in a series damper (curve DP) but there will then be no energy
saving. An axial fan with a falling power characteristic will gain somewhat from
by-pass damper control (curve BP AX from Fig. 9.10) and a backward-curved
centrifugal fan from vane control (curve VC BC from Fig. 9.14 at 50m3/s). In neither
case is more than about half the available energy saving secured.
Speed control is suitable for constant volume flow (though not for constant
pressure). Curve SP assumes 83% full load efficiency for the variable speed motor
and control circuitry combined, a figure which will in practice depend on the system
adopted. At lower speeds and pressures the electrical losses will fall, but the fan will
move to points of lower efficiency on the fan characteristic as the free flow point is
approached (10% total pressure in Fig. 9.17). Curve SL allows for the additional
losses involved in speed control with electrical or mechanical slip.
Once again the example of variable pitch axial fan control gives the best energy
saving over a wide range of volume control. (Curve VP AX from Fig. 9.11 at 40m3/s.)
Again selection of optimum pitch angles will bring energy saving within 3% of the
ideal value for a more limited control range - 40% in this case.
9.6.4 Energy savings

Figs. 9.15, 9.16 and 9.17 showed typical values for the (normally electrical)
input power at all points in a control cycle. The significant factor is the total
energy consumption in kilowatt-hours* over a representative period of time. To
find this we must know the proportion of running time occupied at each fraction of
full duty, and sum up the total kWh. This will now be done for the cases
summarised in the last three sections.
'Gigajoules is the approoriate SI Unit. 1 GJ = 278 kWh.

For distribution in time the simplest assumption is that all volume flows in a
constant resistance or pressure system and all pressures in a constant volume
system are equally likely. With such a control regime the energy consumption over
a representative time period will be simply proportional to the area under the
appropriate curve in the figures cited, over the control range. This assumption has
been made in Table 9.2.
As a standard of comparison the electrical energy which would be consumed

over the same period if there were no reduction in fan duty will be taken as 100%.
This will be calculated for a fan efficiency of 85% and a motor efficiency of 92%,
both appropriate to the duty of 50m3/s at 1000 Pa total pressure for which the
examples have been selected.
The "theoretical minimum average powers" in Table 9.2 assume that 85%
aerodynamic efficiency and 92% motor and drive efficiency could be maintained
over the whole control range. The more realistic values of fan, motor and drive
losses indicated by Figs. 9.11, 9.14, 9.15, etc. have been used to compute the
"actual average powers" in the Table.
9.6.5 Example with constant system resistance

An air conditioning system is expected to operate for 1,250 hours per annum at
a duty to be controlled over the following range, all volume flows being considered
equally likely.
This is a constant resistance system and Table 9.2 may be applied to compare
the annual energy consumption of the fan with various control means. The
reference power for the Table will be the product of maximum volume flow and
pressure increased by the losses of an 85% efficient fan and 92% motor.
Annual energy consumption is 64 kW x 1250 hours = 80,000 kWh at 100%

reference level. Table 9.2 for a minimum volume flow 40% of maximum gives:
9.6.6 Example of constant volume flow control

An industrial process requires a flow of 50m3/S to be maintained against total
pressure having a most probable value of 850 Pa varying between extreme limits of
700 Pa and 1000 Pa. On the assumption of "normal statistical distribution" the
hours spent in each 50 Pa band of pressure in a total of 1,000 hours will be as in
the table below. Fig. 9.17 applies and the reference power for this figure ignores
motor losses.
100% power for Figs. 9.15, 9.16 and 9.17
Multiplying this reference power by the percentage in Fig. 9.17 for each pressure
band and by the number of hours spent in that band, and then adding the results
will lead to the total energy consumption per
1,000 hours.
CHAPTER 10
Noise,vibration
and fatigue
The ideal fan would no doubt perform its function silently,

smoothly and without attention for as long as it was needed.
In the practical world these qualities are unattainable, and it
is the joint responsibility of the fan designer and the
engineer responsible for selection and installation to see
that no nuisance is caused, and that life expectancy is
satisfactory.
If the fan duty is substantial and particularly if the
pressure is high, both noise and vibration are likely to need
control measures to ensure that no annoyance is caused to
people in the vicinity. That subject is thoroughly treated in
our companion volume "Woods Practical Guide to Noise
Control", and this chapter deals only with the generation
mechanism and the practical assessment of noise, vibration
and balance quality.
10.1 Rating System for Fan Noise

10.1.1 Sound power
The noise control engineer needs to know "how much" noise the
fan will generate when installed. This data should be provided by
the manufacturer, and there is now general agreement on the form
it should take, expressed in such standards as BS 848: Part II in
Great Britain, or AMCA Standard 300 in North America. Sound
pressure and sound level
200
Noise, vibration and fatigue 201
are deprecated as units of fan rating because their value is entirely

dependent on the location of the test microphone. Instead the rating should
be the Sound Power Level (SWL) of the fan as defined, expressed in decibels
(dBW) above a reference level of 10-12 watts. This provides a fixed starting
point for noise control calculations dealing with the effects of the
environment.
The word "decibel" is here used in its basic sense-the logarithmic

expression of the ratio of two powers. It must not be confused with the
decibel (dB) of sound pressure or the dBA of sound level on the A weighting
scale. See Woods Practical Guide to Noise Control.
The fundamental sound power is the rate at which sound energy leaves the
fan and travels along a duct attached to the inlet or outlet. Strictly speaking
there is one value for the inlet and another for the outlet, but with many fans
the two are sufficiently close for a single common value to be quoted.
If the duct is removed the quantity radiated into the surrounding
atmosphere is called the "open inlet (or outlet) sound power". At low
frequencies this is less than the ducted sound power, some of the sound
energy generated being reflected back into the fan casing. Alternatively, by
analogy with an electric generator, we may say that the higher "acoustic
impedance" presented by the open atmosphere to the fan outlet "terminal"
reduces the output power.
10.1.2 Frequency spectrum

The fan sound power level is incomplete as a performance rating without
additional information as to the frequency distribution of the sound energy.
This is not only because of the differing sensitivity of the human ear to
different frequencies. It is also because the sound transmission and
absorption of air distribution and noise control systems are also very
frequency dependent. Generally speaking the higher frequencies are more
readily absorbed than the lower frequencies.
A weighting system such as the A scale of sound level is quite unsuitable
for sound power rating. The spectrum shape at the point where the sound is
heard, and therefore the appropriate weighting, will change with every
change in sound absorption and listening point. An octave band frequency
spectrum of sound power level gives suitable evidence of fan performance for
noise control purposes. The standard centre frequencies of the bands are:
Of these the 63 and 8000 Hz bands are subject to greater uncertainty of

measurement (particularly 63 Hz) and may generally be omitted without
detriment to the achievement of noise control. The 125 and 250 Hz bands (for
centrifugal fans) or the 250 and 500 Hz bands (for axial fans) are often the
critical ones for noise control purposes, but it is unwise to ignore the
remaining frequencies.
Fan noise tests are often made with analysing equipment which automatically
plots the results in third octave bands. These may serve to locate and roughly
assess the magnitude of prominent pitched tones in the noise. The 24 bands are
too numerous for practical noise control work, however, and they should be
reduced to octave form for quotation.
10.1.3 The sound power scale
A sense of the magnitude of sound power levels (SWL) will grow with familiarity,
but there is an unfortunate initial tendency towards confusion with sound pressure
levels (SPL). The two would be numerically the same in the somewhat imaginary
circumstances of a very small sound source, radiating equally in all directions, and
at a distance of about 30cm from the ear. At 3m distance the sound power level
would be unchanged but the sound pressure level would be 20 d8 less (SWL =
SPL3 + 20).
Noise control, even ordinary transmission through ducts, will make the
difference greater and the inexperienced user must not be intimidated by the
high-sounding values of fan sound power levels. Table 10.1 gives an idea of the
whole gamut of sound power, together with air duties at which the fans are likely
to feed similar sound powers into a ducted system.
10.2 Typical Fan Sound Ratings
10.2.1 Sound power levels

Fig. 10.1 is a chart drawn up to illustrate the range of sound power levels
normally produced by fans of standard type. The fan is supposed to be operating
near the best efficiency point on its characteristic with
air of standard density, and SWL is the level of sound power which would enter
and travel along a duct connected to either the fan inlet or the fan outlet. Lp is
derived from the fan total pressure and Lq from the volume flow.
The air duty has been used rather than the size, speed, or mechanical power
input sometimes employed in similar surveys; this makes for a fairer comparison
between fans of differing type and mechanical efficiency. Any range of sizes and
speeds all to one particular fan design will have Lp values close to a straight line
sloping upwards at an angle between 45° and 50°.
A manufacturers noise ratings for a standard fan series will be based on a
standardised test procedure (see Chapter 11) leading to such a design line or its
equivalent. These ratings should always be used they are as much a function of
the particular product as is the fan efficiency. Nevertheless, some general
observations can be made on the values to be expected in typical cases.
The shaded area in Fig. 10.1 covers the range within which Lp may be expected
to lie. Standard axial and forward-curved multi-vane centrifugal ranges will be
located about the middle of the band. Backward-curved centrifugal and mixed flow
designs should be found in the lower half of the area, while the lowest Lp values
may be approached by high efficiency aerofoil bladed backward-curved centrifugal
fans, but are unlikely to be improved upon by any design. The bottom limit of the
band has been sloped steeply upwards at the higher pressure because designs for
these pressures are likely to approach the radial tipped form, which is not so quiet.
In general, fans of standard types should only be found with ratings near the
top of the band if they are poorly selected-in particular for pressures well below
the best efficiency point. On the other hand such levels may well be reached or
exceeded in practice if the installation has poor arrangements for controlling the
inlet flow. Fig. 10.2 has been reproduced from Woods Practical Guide to Noise
Control to illustrate this important point. Axial fans are illustrated, but similar
considerations apply to all types. Any sharp change of direction or area, and any
obstruction to the flow should be kept well away from the fan inlet.
10.2.2 The fan laws for sound power

The use of Lq and Lp with any straight line on Fig. 10.1 implies the following
proportionality for sound power:
SW is proportional to Q.(PFt)a
Now the volume flow Q is proportional to the fan speed n rev/sec and the fan
total pressure PFt to n2. Therefore, for a given fan
SW is proportional to n(2a + 1)
SW is often taken as proportional to ns (i.e. a = 2.0) a value which applies

reasonably closely to some centrifugal fans, but for accurate
prediction an experimental value should be used. In the case of the Aerofoil axial
range SW ∝ n5.5 (i.e. a = 2.25) and the index for a pure dipole source (see
10.3.1) would be 6 (a = 2.5).
Sound power (SW, watts) has the same units as air output power (Q x PFt
m3/s X Pa = watts) and it is instructive to express the first as a fraction of the
second. This fraction is known as the sound power ratio (SWR) on a duty
basis.
A representative value of SWR is 10-6 at 250 Pa fan total pressure. Taking

Q as 1 m3/s the air output at this pressure is 1 x 250 = 250 watts and the
sound power 250 x 10-6 = 250 microwatts. The corresponding sound power
level is:
Since SW is proportional to Q.(PFt)a it follows that SWR is proportional to

PFt(a-1). On Fig. 10.1 three straight lines have been drawn through 84 dBW,
250 Pa at slopes a = 2.0, 2.25 and 2.5 corresponding to SW ∝ n5 n5.5 and n6.
The equations of these lines are:
The corresponding relationships between the sound power ratio, SWR and
the fan total pressure are plotted in Fig. 10.3.
10.2.3 Change of sound power with duty point
So far we have considered only the fan sound power level at the best
efficiency point on the fan characteristic. This is generally the lowest value, and
the typical examples in Fig. 10.4 show the way in which the sound power
increases away from this point. Variations of 5 to 10 dB are common and may
be quite sharp if the characteristic contains a marked stall point. The shapes
shown are representative of the fan types quoted.
When the characteristic of the fan itself is capable of modification as by

pitch control of an axial or vane control of a centrifugal fan other methods of
representation are needed. Fig. 10.5 illustrates the use of contours of constant
sound power level laid over the volume flow-fan total pressure chart. The fan
is the same variable pitch axial model whose performance with constant
efficiency contours was plotted in Fig. 9.11. An example will show the use of
this chart:
Example: The 1600mm 975 rev/min variable pitch axial fan of Figs. 9.11 and
10.5 operates in an air conditioning installation along system characteristic B.
The following table can be drawn up, reading SWL values to 1 dBW from the
nearest contour on Fig. 10.5.
The last line in this table is the experimental value of LP corresponding to the
observed value of SWL. Comparing these results with the Lp band on Fig. 10.1,
the result for 16° lies on the mid line. The 24° and 32° results are better than
average, and steep pitched axial fans working near peak pressure generally do
give the most favourable noise levels for a given duty. The LP values at 0° and 8°
are not so good, partly because
of the relatively low usage of available pressure, and partly because the blade is
aerodynamically "off-design" for these angles, being designed for optimum
performance around 24°.
It is interesting to note the strong correlation between the sound power and the
aerodynamic loss of a fan. Reasons for this correlation will appear in Section 10.4,
and to demonstrate it Fig. 10.6 has been drawn to the same scale as Fig. 10.5.
The aerodynamic losses in watts can be
expressed on the same decibel scale as the sound power and over the whole
of the volume-pressure chart the difference is between 48 and 52 dB. At 50
dB the sound power would be one hundred-thousandth of the aerodynamic
loss. For fans of the same design, but different speeds and diameters this
ratio would vary with (tip speed)2.5.
10.2.4 Typical fan sound spectra

The total sound power level quoted for a fan should exclude any
measurements in bands below 63 Hz. Their inclusion might rdise the total by
several dBW, the measurement would be very unreliable, and the sensitivity of
the ear to sound in this region is less than onethousandth of that at the middle
frequencies. The remaining power should be split between the eight standard
octave bands giving the octave band sound power level in each.
A convenient method adopted in fan catalogues is to quote from a number
of sound spectra one which is applicable to the fan in question. Each sound
spectrum will give for each octave band the dB deduction
to be made from the total sound power level in dBW to obtain the band level.
Some typical examples are quoted in Table 10.2, but it is important to use the
manufacturers data since variations are substantial, e.g. from - 2 dB to - 12 d8 in
the 63 Hz band, and - 15 dB to - 35 d8 in the 8000 Hz band.
The generally higher frequencies of the axial fan sound output make it more
audible when heard directly, but on the other hand it is more readily attenuated by
noise control. The spectra quoted apply to fans for around 10m3/s at 250 Pa. The
sound powers will move towards lower frequencies at lower pressures or larger
volumes, towards higher frequencies at higher pressures and smaller volumes.
These values apply to the sound power flow along a duct. If there is no duct,
less power will be radiated into surrounding space from the open outlet or inlet.
A rough idea of the difference may be obtained by assuming plane wave
reflection at the opening. The d8 deduction to be made to determine the octave
band sound power level radiated depends on the area of the inlet or outlet
opening and the band frequency as shown in Fig. 10.7.
The measured sound power of a fan will be predominantly broad band or
random frequency noise. However, it is liable to include some discrete frequency
sound particularly at a frequency equal to the rate at which the fan blades pass a
fixed point. This becomes identifiable to the ear when it raises the appropriate
third-octave band 5 dB above its neighbours (making only 2 dB contribution to
the octave band) and can be objectionable if it is much greater.
Example: The 1600mm 975 rev/min variable pitch axial fan of the last
example has an overall sound power level of 110 dBW at 24° pitch angle. What is
the sound power spectrum entering the plant room from an open fan inlet? The
manufacturers spectrum correction, the reflection correction from Fig. 10.7 and
the resultant octave band sound power levels are as follows:
10.3 The Mechanism of Aerodynamic Sound Generation

Aerodynamic noise is caused by the fluctuating forces generated by the flow of
air or gas. Regular variations of force produce pitched tones of recognisable
frequency. Randomly fluctuating forces produce broad band noise. A steady force
in a fixed location does not generate sound, thus streamline or laminar flow is
silent. In practical installations, however, the flow is always turbulent, in part at
least, and noise is produced by one of the mechanisms described below.
10.3.1 Lift and drag forces

Examples are noise from: fan blades, turning vanes, grilles, dampers, bends,
sudden contractions and enlargements in ducts and obstacles to flow.
In all these cases there are forces between the air and a solid body. The forces
are unsteady owing to turbulence in the approaching air,
turbulent boundary layers, flow separation, and wake shedding. These are the
principal sources of noise in fans and air distribution systems. They are best
limited by good aerodynamic design and by keeping down
the flow velocity. The forces are random and produce broad band noise, being known
acoustically as dipole sources. If V is the relative velocity of the air and the surface, the
sound power will vary with V6.
10.3.2 Interaction noise

This arises at a fan and is caused by guide vanes, bearing supports, or poorly
designed approach airways.
The essential feature is that some regular pattern, repeating each revolution of the
fan, is imposed on the otherwise random force fluctuations between the fan blades and
the air.
Upstream guide vanes are a potent source, but downstream guide vanes should be
satisfactory if more than 1½ blade chords from the impeller to smooth out the
fluctuations in the wake of the fan blades. Inevitable upstream obstructions should be
shaped to minimise the turbulent wake. Fan inlets are carefully shaped to provide
smooth, uniform flow into the impeller. The installation engineer must see that these
precautions are not ruined by defects such as those illustrated in Fig. 10.2 presenting
the blade tips with a boundary layer which thickens locally at particular places in each
revolution. This will cause interaction noise.
Interaction noise is an obtrusive pitched tone at blade passage frequency and its
harmonics. It is a dipole source with sound power varying as V6.
10.3.3 Jet noise

Examples are noise from : air blast; steam exhaust; aircraft jet engines.
This noise is generated within the air in the mixing region at the edge of the jet. As
shown by Lighthill in 1952 it is caused by the interaction of turbulence with the high
shear stress caused by the steep velocity gradient. It diminishes rapidly with air speed,
and is negligible in duct systems except perhaps at nearly closed dampers. In quiet
surroundings however, it limits the velocity of jet-throw distribution systems. The
remedy is to introduce more air at lower velocity for the same momentum.
The source elements are known acoustically as "lateral quadrupoles". The noise is
broad band and the sound power proportional to V8.
10.3.4 Propeller noise

This is characteristic only of aircraft (or hovercraft) airscrews.
It is caused by the main propulsive force at each airscrew blade, which is steady in
time, but is moving to and fro at very high speed, relative to an observer away from the
axis, as the airscrew revolves. With only 3 or 4 blades, moving close to the speed of
sound, powerful propeller noise is heard. The lower speeds and more numerous blades
make this source entirely negligible in fan impellers.
The source is dipole, but the mechanism of propagation makes the sound power in
the transonic velocity range vary as V9 upwards.
10.3.5 Pulsating flow

Examples are positive displacement compressors, reciprocating internal combustion
engines and sirens.
The noise source is a regular pulsation in the mass-flow-rate of the air or gas
stream, entering or leaving the atmosphere. To limit the noise the pulsation must
be reduced, as by a reservoir, a cavity resonator if the speed is fixed, or a diesel
engine type silencer.
A pitched tone is produced at pulsation frequency with many harmonics. A
powerful source, particularly when the flow change has a steep wave front.
Acoustically a monopole source with sound power varying as V4.
10.3.6 Induced vibration

Examples of noise from this source include the drumming of duct walls, the
flapping of canvas or fabric, and the rattle of loose joints or fixtures.
The fluctuating forces between a solid body and the air flowing over it tend to
produce vibration. The structure must be stiff enough, or damped sufficiently, to
limit the vibration. Curved sheets have inherent stiffness, but flat sheets require
stiffening ribs, particularly if they are of large area, and therefore, efficient
generators of airborne sound.
With random frequency aerodynamic forces the structure will vibrate at its own
resonant frequencies.
10.3.7 Other aerodynamic sources

Whistles, rumbles and rattles are sometimes heard in air systems which, when
traced to their source, prove to have elements in common with musical
instruments. Organ pipe resonance, cavity resonance, reed vibration, string
vibration, surface drumming and edge tones may all be encountered. Experiments
with stiffening or deadening the parts, closing leaks or diverting the local flow will
usually provide the remedy.
10.4 Theory of the Fan Laws for Sound

Dimensional analysis indicates that the sound power ratios SWR = SW/W of a
geometrically similar series of fans should be dependent only on Mach number,
Ma = V/Va, (the ratio of a typical fan velocity to the velocity of sound) and
Reynolds number, Re = DVρ/µ which is the ratio of the dynamic to viscous forces
involved,
Taking the rotational speed N, diameter D, and output power W of the impeller
as the typical quantities:
Acoustical theory, considering only the Mach number effect, yields the
following relations for the source types mentioned in 10.3
Fan noise is principally dipole in origin and, for the simplest cases of boundary
layer separation, there is reason to believe that the Reynolds number exponent y
should be about – 0.4, leading to:
Axial fans seem to follow this relationship quite closely (e.g. N5.5 D7.5), but other
types may well introduce more complex generation mechanism. For example,
centrifugal fans are usually considered to follow the law:
Geometrical similarity as it affects Reynolds number is not well maintained in

practice from size to size (thickness, clearances, even number of blades). It is wise
to keep the index of D 2 greater than the experimental index of N and to treat
small changes with size as a scale effect on the constant. The influence of
atmospheric conditions is small -generally within 1 dB.
10.5 Vibration
A fan, like any other rotating machine, inevitably vibrates to some extent, and is
a source of vibration in the structure on which it is mounted. Unless the fan is
small and light, or the foundation heavy and solid it is advisable to mount the fan
on vibration isolators, and this subject is treated in Woods Practical Guide to Noise
Control. The fan manufacturer will normally advise on the choice of isolators, and
sometimes provides them built-in between the driving motor and the mounting
points.
This section deals with standards that have been established for assessing the
severity of vibration, and with the limitation of the commonest source in fans - out
of balance.
10.5.1 Measuring units for vibration

There are three things about a vibrating surface which can be measured, and
each has its function.
Displacement is the natural factor to consider if one is concerned about
reducing running clearances or causing rubs. It is also the most obvious to the
eye.
Velocity is generally agreed to be the best general measure of vibration
severity, particularly in relation to its disturbing effects on people and other
equipment.
Acceleration is the factor which gives rise to forces and stresses within a
machine, and between a machine and its foundation.
These three quantities have a simple and important relationship in the particular
case of sine-wave vibration. This is a vibration in which the displacement is
proportional to sin ωt where ωt is an angle which advances from 0° to 360° in one
complete cycle of the vibration. ω is the angular velocity and equals 2∏n where n
is the cycles per second, or Hertz.
In such a vibration the velocity and acceleration are sine waves also, advanced in
time by one quarter (90°) and one half (180°) of a cycle respectively.
The velocity is best represented by its rms (root-mean-square) value
which is 1 /√2 times the peak value for a sine wave. All these quantities are
illustrated and tabulated in terms of v with reference to the sine wave in Fig. 10.8.
Vibration which has its origin principally in out-of-balance is sine-wave in character,
but vibrations arising from such sources as rough bearings or air turbulence are not.
The relations of Fig. 10.8 then break down, and acceleration in particular may be
much greater than is indicated in relation to displacement.
10.5.2 Vibration severity

For practical assessment the smallest significant step in vibration severity is
considered to cover the ratio 1 to 1.6. A scale of quality judgement is proposed in
BS 4675: 1971 with four grades A, B, C, D each covering two such steps, i.e. a
ratio of 1 to 2.5. Table 10.3 summarises part of this scheme. The vibration to be
considered is that on the external surface of the complete machine, with particular
reference to feet or other fixing points and to bearing housings. So far no specific
recommendations have been established for particular products, but it will be seen
that the grading increases in severity towards Class I which is for small units likely
to be fitted in more delicate situations. It is intended that, as experience is gained,
the grades should become quoted in specifications.
The grading applies only to frequencies from 10 Hz to 100 Hz. Vibration severity is
given as v, rms.
10.5.3 Static out of balance

Two forms of balance defect can be recognised: Static unbalance and couple
unbalance. Complications arise with flexible rotors, but fan impellers can always
be taken as rigid from the balancing point of view.
Static unbalance simply implies that the rotor has a heavy side which can be
counterbalanced by a single mass m added at a radius r exactly opposite. If the
shaft were free from all constraint at its bearings it would vibrate keeping always
parallel to itself - more precisely the shaft axis would rotate in a circular path (end
view) with a radius e. The centre of this circle would be the centre of gravity (c.g.)
of the whole rotating assembly which would thus become stationary - a state of
natural running balance, like a top.
In practice, of course, the bearings must be supported, and if they are rigidly
held they will experience the full centrifugal force due to the whole mass M at a
radius e - since the shaft axis is now stationary instead of the c.g. The product
Me is called the unbalance and the rotor will be statically balanced by
counterweight m if:
suitable units are gm.mm for mr and kg.µm for Me.

The counterweight m will not be perfect, and there will be a residual
unbalance ∆m.r = ∆e.M. With a completely soft bearing mounting, speed n rev/s
= 2∏ω, and residual c.g. eccentricity ∆e = ∆m.r/M the vibration will have:
10.5.4 Balance quality grading

Fig. 10.10 shows the relationship of permissible eccentricity a with speed n
for a number of balance quality grades, G1 , G2.5, etc. specified in ISO 1940.
The international committee responsible for this specification has tentatively
(1973) proposed quality classifications for particular products, extracts from
which are given in Table 10.4. It will be seen that G 6.3 is the grade
considered appropriate for fan impellers.
Note that the product eω defining the grade is a peak velocity. Since
out-of-balance vibrations are generally sine wave it follows that the rms
velocity = 0.71 eω. This is the vibration severity measure of Table 10.3 and it
will be seen that quality grades A B C and D correspond to balance grades G1,
G2.5, G6.3 and G16 + respectively for Class I small machines and components.
This equivalence of vibration and out-of-balance grades applies only to the
motion of soft mounted rotor bearings. The vibration will be reduced in
proportion to any added mass of bearing housing, and to the further
constraints of semi-rigid or rigid mounting. The significant vibration severity
with isolation mounts is that on the far side of the mount-vibration of the
machine itself is then relatively harmless.
10.5.5 Couple (or dynamic) unbalance

A rotor may be in perfect static balance and yet produce vibration at its
bearings by the rocking motion indicated in Fig. 10.9. Equal and opposite
counterweights m at radius r in correction planes spaced a apart are said to
correct a couple unbalance equal to m.r.a.
Any residual couple unbalance ∆mra will act, unchanged, at the bearings,
and if these are I apart the unbalance at each bearing will be ∆mra/I. This is
the same unbalance as would be produced by a static unbalance half-way
between the bearings of:
Clearly the bearings must be well spaced (I not too small) if a small couple
unbalance is not to produce large effects. With a rotating mass M the
effective value of ∆e for balance quality grading is:
This eccentricity need not be added to the residual static unbalance

eccentricity ∆e to give the overall balance quality grading of the whole
machine, though each must be within limits. They may be additive for
internal bearing loading, however, and for loads on the isolation mounts.
10.5.6 Balancing
Narrow impellers (Fig. 10.9) can be corrected by static balancing,
particularly if produced by a dimensionally consistent method such as
diecasting. Close tolerances on the shaft-hub fit are essential, and must not
be spoiled by careless fitting. Dynamic unbalance will be produced by a small
run-out such as might result from forcing the impeller on against a bruised
shaft or shoulder.
Impellers with a breadth exceeding some 20% of the diameter are likely to
require dynamic balancing. This involves driving the rotor in the bearings of a
balancing machine which are soft mounted in a horizontal direction. The
resulting bearing vibrations are picked-up, analysed and displayed to show
the positions and values of counterweights to be attached in two selected
correcting planes.
Built up impellers may have their component parts separately balanced
subject to a final check on the complete assembly. For example, an axial
impeller with a broad heavy hub might have this component dynamically
balanced, followed by a static balance when a set of comparatively light and
narrow blades had been attached. This
final stage could be more accurate than a dynamic operation since the sensitivity
of the latter may be impaired by the masking effect of the fluctuating
aerodynamic forces on the moving blades.
10.5.7 Dynamic balancing examples

The way in which the various unbalances may be combined and corrected is
best shown by examples
Example 1. A 1000mm, 1800 maximum rev/min axial fan impeller weighing 25

kg is to be statically balanced by adding weight at 180mm radius. Within what
limits should the weights be adjusted ?
The G6.3 balance quality grade is appropriate. At 30 rev/s, Fig. 10.4 gives ∆e
= 32µm as maximum c.g. eccentricity. The corresponding maximum
out-of-balance weight is:
Example 2. Fig. 10.11 (a) illustrates the operation of dynamically balancing a

250mm diameter, 3600 maximum rev/min (60 rev/s) multivane impeller
weighing 2.5 kg. Two counterbalance weights are to be attached at 110mm
radius in correcting planes 150mm apart. Initial static unbalance: 240gm.mm,
50mm from one plane. Initial couple unbalance: 120gm.mm x 150mm at
right-angles to the static unbalance.
The 240gm.mm static counterbalance components will be divided between the
correcting planes so as to give equal and opposite couples
The couple counterbalance components of 120 gm mm in each correcting plane

will directly oppose the couple unbalance moments.
Combining the counterbalance vectors in each plane gives 200 and 144 gm
mm respectively in the directions shown. These determine the following
counterbalance weights:
Example 3. Fig. 10.11 (b) illustrates the impeller of the last example overhung
from a housing weighing 1.5 kg with bearings 80mm apart. Suppose the
impeller to have been balanced just to the G6.3 limits, viz ∆e = 16 Nm at 60
rev/s both statically and dynamically.
The static unbalance:
The couple unbalance:
Note that these are in each case one-sixth of the initial unbalance.
The static unbalance will produce a reaction unbalance force of 40 gm.mm
which will be located half-way between the bearings. These
two opposite forces, 140mm apart, will produce a couple unbalance of 40 x

140 = 5600 gm, mm2 to be resisted by the bearings.
The couple unbalance of the impeller will be transferred unchanged, giving

3000 gm.mm2 at right-angles to the 5600 gm.mm2. Combining these the total
couple at the bearings is:
On the total mass of 2.5 + 1.5 = 4.0 kg this will produce with soft
mounted bearings:
This is much too large a vibration for a Class I application as Table 10.3 will
show. The solution, however, is not to attempt to improve the balance
quality of the impeller but rather to stiffen the bearing housing against
rocking motion. This may be done either by attaching the housing rigidly to a
base of substantial mass (increasing M) or by fitting the housing with rigid
arms of substantial span attached for example, to the fan casing (increasing
the effective I).
10.6 Critical Speeds

10.6.1 Resonant vibration
There exists for any rotating shaft, rotor and bearing combination, a certain
speed at which resonant vibration develops. The slightest unbalance will cause the
shaft to bend, increasing the unbalance and bending without limit except for
damping effects and loss of exact resonance. This is a dangerous condition and
causes intense vibration so that the critical speed must never be held for any length
of time. Apart from the first, and lowest, critical speed there are others corres-
ponding to more complex modes of vibration.
While high-speed turbines are designed to pass through the first, and often
other, critical speeds during run up, this is never the case for fans. Fan shafts are
designed with sufficient stiffness to keep below the first critical speed at all times,
although a resiliently mounted fan will pass through low speed resonance as a
complete assembly. In this case there is no bending of the shaft.
10.6.2 Extended fan shafts with overhung impellers

It is sometimes desirable to mount an impeller at a substantial distance from the
nearest bearing. The fan may be handling gases which are so hot that oil
circulation or water cooling would be needed if the bearing housing were immersed
in the gas. The impeller may be running inside an oven or a sealed and insulated
stove with the running gear outside for servicing.
Fig. 10.12 (a) illustrates such a case. The impeller, of mass M, will not be
perfectly balanced. Suppose its centre of gravity is distant a from the centre line of
the shaft which is straight when unloaded. When bent through a deflection y, there
will be a proportionate reaction force F = ky. This must equal the centrifugal force
produced by M kg rotating in a circle of radius (y + e) with angular velocity ω =
2∏n. That is:
The deflection thus becomes indefinitely large when Mω2 equals k, which
determines the critical speed:
While fans should never be allowed to reach the first critical speed, it is of
theoretical interest to note that, if they do, y changes over to a negative value. This
means that the impeller centre of gravity CG now rotates inside, rather than
outside, the shaft centre as shown in Fig. 10.12 (c). As the speed rises the c.g. will
move nearer and nearer the
centre line through the bearings, as a smaller and smaller eccentricity suffices to
supply the force required to bend the shaft. The deflection y ultimately
approaches e.
At all speeds these shaft forces will be balanced by reactions at the bearings.
The bearing housing must be rigidly mounted so as not to yield; if it does y will
increase while k and the critical speed will fall. The alternative of soft mounting is
not usually satisfactory because of excessive impeller off-centre when passing
through low speed resonance.
10.8.3 Calculation of critical speed

The formulae for bending of shafts yield the following expression for the end
deflection y m under an end load F newtons. Y1 is due to the bending in length L
and y2 to that in I.
Taking E = 2 x 1011 N/m2 for steel, and converting dimensions L, I, D and d

from m to mm for convenience, we find the critical speed nc with an effective
mass M kg at the end of the shaft to be:
Example. Consider the extended shaft assembly shown with mm dimensions in

Fig. 10.12 (a). Can a 630mm impeller weighing 10 kg be operated safely at the
end of such a shaft with a speed of 24 rev/s (1440 rev/min) ?
To allow for the distributed mass of the shaft extension I about 40% of its
mass of 3.8 kg should be added to the impeller mass making M = 11.5 kg.
The operating speed is thus 56% of the first critical speed, which is
satisfactory - 60% to 65% is a reasonable maximum if the rigidity of the bearing
housing can be relied on.
10.7 Structural Fatigue

10.7.1 Limits of alternating stress
In common with most machinery, fans operate with mechanical stresses far
below the ultimate capacity of their materials. In part the reason is uncertainty
of loading, though the working loads are more predictable than usual, the worst
hazards being often the shocks and blows of transit and installation. The fan
designer has a particular problem, however-the vibrations produced by
fluctuating aerodynamic forces.
It is well known that a load many times applied and removed will cause a part
to break when it could indefinitely withstand two or three times that load if
steadily applied. The relationship between the maximum alternating stress Se
and the number of cycles (complete reversals) to failure is indicated in Fig.
10.13. These are typical values only, the actual fatigue performance depending
on the alloy, and the heat treatment and other processes used in its
manufacture, even for standard test specimens.
When a steady stress is applied in addition the alternating part must be reduced
for the same life although the permissible peak stress rises. The straight line
relationship of Fig. 10.14 is a fairly safe assumption for most structural materials. For
any desired life applying to both Sa and Se:
10.7.2 Alternating stress in axial fan blading
The fan designer must take into account the alternating stresses
experienced by an impeller. As an example consider one blade of an axial fan.
A steady mean lift force is applied to the blade corresponding to the pressure
rise, but this has a randomly varying component in addition as discussed in
10.3.1. Both forces produce a bending moment and corresponding stresses
which reach a maximum near the blade root, and add to the centrifugal
stresses. This is a mixed stress condition similar to that of Fig. 10.14.
These stresses can be examined experimentally. The first step is to coat the
blade with a test layer of a special brittle lacquer. When the blade surface is
stretched by tensile stress, cracks appear in the lacquer. These cracks run at
right-angles to the direction of the stress, and are closer together the greater
the value of the stress. The points on the surface where the tensile stress is a
maximum can thus be identified, and strain gauges are attached at these
points, one on each side of the blade. Completing an electrical bridge network
and bringing out the four
corners of the bridge through small precious metal slip rings it is possible to
display the alternating component of the bending stress.
Fig. 10.15 is a record of such a display. A consistent vibration frequency is
immediately obvious, and this proves to be equal to the natural frequency of
vibration of a stationary blade, if struck. On the other hand it is clear that the
vibrations do not build up indefinitely as would a resonant vibration. The
random nature of the air forces has a damping as well as an exciting action.
Of course, resonance with any fixed pattern imposed on the random forces
must be avoided. Such a pattern would be caused by the interaction force of
10.3.2, and it is wise to design for a natural blade frequencyfive orsix times
maximum rotation frequency to be well above any likely excitation.
As the examples show the maximum vibration amplitude occurs when the
fan pressure is a maximum. The more irregular forces of separated flow in
the stalled region are not so effective. This facilitates a development
programme for determining the maximum permissible tip speed of a specific
design. Maximum operating stress amplitudes can be related to static life
tests on actual production wings.
CHAPTER 11
Fan testing and

performance
prediction
Air is a difficult subject for the tester. It cannot be seen and

measured: nor can it be caught in a container and weighed.
Furthermore, its motion is three-dimensional and turbulent. Thus a
great many measurements of velocity and direction would be
needed to integrate a total volume flow, unless the air were first
collected, steadied, and organised in a suitable test airway.
11.1 Types of Test

Such standardised airways are used to provide a reliable test of
fan performance, and they are discussed later in this chapter.
Often, however, we shall want to measure the airflow in an actual,
working installation. If the installation is a ducted one we shall look
for the place, often quite remote from the fan, where the air
velocity is most nearly steady and parallel, for our measurements.
This is a site test for volume flow and will check whether the
system resistance has been correctly estimated and the fan is
giving its standard performance.
If the volume flow does not match requirements we shall want to
find out why. A measurement of the pressure difference across the
fan will help to decide whether the fan or the system is at fault.
However, site
227
conditions are unlikely to be such that we shall get a good

measurement of fan total pressure. If fan performance is in dispute it
can only be properly checked by a test in a standardised airway. If a
very large fan is to be tested, standardised airways may be considered
too costly, or the power of available motors may be insufficient. In such
cases an acceptance test may be conducted on site, by agreement
between the parties, but such an agreement presupposes an airway
layout which is not too different from one of the standardised
arrangements as regards pressure measurement.
A model test may well give more reliable performance data for a large
fan than a site test with an unfavourable airway layout. The model is
made geometrically similar to the large fan, with an impeller diameter
usually between 500mm and 1000mm, and is tested with air in a
standardised airway, preferably at about the same tip speed (adjusted if
a different gas is to be used on site). The efficiency of the large fan is
likely to be slightly greater than that measured on the model, but such
a scale effect improvement may only be claimed by agreement between
the parties, based on evidence for the fan type in question. For the
corresponding reason a model test is not allowed for the acceptance of
a fan smaller than the model.
The fan laws are rules for calculating the performance of a fan at a
speed or gas density different from that at which it was tested, and also
for converting the results of a model test to full scale. They hold with
precision for relatively small changes of speed or density, but, if the
pressure rise is large, the effects of change in compression ratio must
be considered; see 11.4.3.
Methods of test for fan sound power are dealt with in our companion
volume - Woods Practical Guide to Noise Control, Chapter 3. Reference
should also be made to Sections 10.2.4 and 10.4 in this book, where
the fan laws for noise are discussed.
11.2 Methods of Measurement
11.2.1 Pressure
The only two quantities that can always be measured at a point in an
air stream are the temperature and the pressure. The pressure will be
measured separately, in two parts:
Firstly, absolute or barometric reference pressure of the still
atmosphere close to the test airways.
Secondly, the gauge pressure, or pressure difference between the
reference atmosphere and the still air within a small hole exposed at
the required point in the air stream.
If the air is flowing parallel to the surface in which the hole is made,
the gauge pressure will be the static pressure of the air stream. If, how-
ever, the hole is the open end of a tube facing the airflow, the air
stream will be brought to rest locally, and the gauge pressure measured
will be the total pressure of the air stream; see Section 6.1.3.
Fan testing and performance prediction 229
The instruments used to measure these pressures are nearly always of

the liquid column type, and may be roughly classified according to the
pressure to be measured, as follows:
For less than 100 Pa: a micromanometer;
For 100 to 1000 Pa: an inclined-tube manometer;
For 1000 to 15,000 Pa: a vertical, or U-tube manometer;
For 15,000 to 110,000 Pa: a mercury manometer or barometer.
The liquid level in a micromanometer is determined with the aid of a
micrometric, magnifying, eyepiece. In the Betz type the scale is etched
on a glass stem carried on a float in the suction column which rises
centrally above a large liquid reservoir. The small movement of the
reservoir surface is allowed for in the scale. This instrument can be read
to 0-2 Pa, and can maintain ± 0.25% accuracy up to several thousand
Pa, so that it may also be used as a reference standard for calibrating the
inclined tube instruments.
Inclined tube manometers are the general-purpose instruments in
most common use. The multiplying factor is the sine of the inclination to
the horizontal and may be up to twenty to one while, in some models,
the tube may be reset vertically for direct reading. The scale may be
drawn to allow for the angular setting, the movement of the reservoir
surface, and the density of the fluid used and it may be calibrated in mm
of water or Pa or kPa. The former is, of course, the gravitational unit of
pressure actually observed and, if the vertical height of the liquid column
is h,
Strictly speaking g should be the local acceleration of gravity and p the

actual density of the fluid, less the density of the air or gas over the
reservoir. However, except in precision research, g may be taken as 9.8
and ρ as the average room temperature density of the liquid alone. For a
water column, therefore, with ρ close to 1000 kg/m3
Inclined tube manometers must be carefully levelled and, for

acceptance tests on fans, calibrated in situ. The gauge fluid must con-
sistently wet the glass to preserve a meniscus of constant shape-water
needs a wetting agent. Pressures fluctuate continuously in turbulent flow,
and may cause the meniscus to dance up and down in unreadable
fashion. Damping must then be applied by restricting either the liquid
connection from tube to reservoir, or the air tubing, until the average
level can be read to, say, ± 1 %. The damping must be symmetrical to air
or liquid flow in both directions, otherwise a false level will be obtained,
high in the direction of easy flow.
11.2.2 Velocity by Pitot-static tube

The standard instrument for measuring the air velocity at a point within
a duct is the Pitot-static tube. A variety of designs has been developed, all
with a strong family resemblance, and with individual advantages which
tend to appear at the research level only. Those illustrated in
Fig. 11.1 may be used without correction in ordinary work provided the
following conditions are met:
(a) The tube must be aligned with the direction of airflow within 5°
(0.5% error) or 10° (1% error).
(b) The tube diameter must not exceed one twenty-fifth times the duct
diameter or one twentieth times the smaller side of a rectangular
duct. It must also not exceed 15mm.
(c) The velocity should be between 3m/s and 70m/s for standard air.
(d) The pressure difference ∆ ρ = pt - ps is measured, together with the
air density ρ, and the velocity calculated from:
Condition (a) is usually met by aligning the tube parallel with the duct
axis and checking that there is nothing tending to produce swirl in the
air stream upstream of the tube. If there is, straightening vanes must
be inserted to ensure reasonably parallel flow. Contractions, bends and
other disturbing features should preferably be at least 10 duct
diameters upstream of the measurement.
Condition (b) usually rules out velocity measurement by Pitot-static
tube in ducts smaller than about 200mm diameter. Measurements of
rather less accuracy can still be made by means of a small "facing tube"
or Pitot tube within the maximum diameter limits to determine pt.ps is
measured at holes drilled in the duct wall level with the open end of the
tube. It is vital that all static and total pressure openings should be
clear, square and free from burrs and deformities of all kinds.
11.2.3 Determination of volume flow

The velocity will vary from point to point over the cross-section of a
duct, usually, but by no means always, peaking near the middle. We
must, therefore, measure the velocity at a number of points, so
distributed as to give an unbiased average. A suitable method for
circular ducts is to divide the area into three or four concentric parts of
equal area, and to find the velocity in each by averaging six velocities
taken at 60° intervals round a circle - 18 or 24 points in all. The radii at
which measurements are taken have been selected by the "log-linear
rule", being off-centre by an amount which takes account of the most
probable velocity distribution across the element area. Fig. 11.2a shows
the location of these measuring points, and all the calculated velocities
can be averaged together. Do not average the velocity pressures, which
would overweight the higher velocities.
A rectangular duct should be divided into at least 25 equal
rectangular areas (5 divisions each side). For long thin rectangles better
accuracy is secured by dividing the long side into 6 (30 points) or 7 (35
points) parts. The points for velocity measurement are determined in
this case by the "log-Tchebycheff rule" in the locations shown in Fig.
11.2b. Again a simple average of all velocities, multiplied by the total
duct area, gives the volume flow rate.
Other distributions of measuring points appear in various national and
international flow-measurement and fan-test codes. Differences will not
be great, though the accuracy is likely to be slightly better when a large
number of points are taken.
11.2.4 Volume flow by anemometer traverse

The vane anemometer is a convenient instrument for volume flow
measurement in the "walk-in" airways of mines and tunnels. The
velocity may be low for the Pitot-static tube and, while the anemometer
is not so precise an instrument, it is quite good for averaging a velocity
which is fluctuating both in magnitude and direction. It will measure
velocities as low as 0.5m/s.
The usual form of instrument has a windmill-type rotor, driven round
at a speed which is proportional to the air velocity except at low speeds
where friction forces are significant. The number of revolutions made is

indicated on a dial or may be sensed electronically and compared with a
time base to give rotational speed directly. A calibration curve is neces-
sary, and should be frequently checked.
The anemometer axis should not be closer than 1 ½ rotor diameters
to the airway wall, which means that it is most suitable for rectangular
airways. It should be held successively for not less than one minute in
each of the measuring positions of Fig. 11.2. The operator should be at
least 1.5m on the downstream side to avoid disturbing the flow past the
anemometer, which he will usually hold and control remotely by rods. It
is better to time a suitable number of revolutions with a stop-watch
rather than repeatedly to start and stop the rotor.
An alternative method is to start both rotor and stop-watch together
at the first measurement point, and allow each to run continuously to
the last point, the anemometer being held for at least 20 seconds at
each point before moving to the next. This process is repeated until
three velocity measurements are obtained which do not differ by more
than 2%, the average of these three being multiplied by the airway area
to obtain the volume flow.
11.2.5 Location of site test measurements

The test plane for flow measurement must be substantially free from
swirl, a condition which can be checked by observing the direction taken
by small "flags" of thread or wool moved over the section on the end of
a stick. A place should also be sought where the flow is as steady and
symmetrical as possible. Both these conditions are best met, as a
general rule, on the inlet side of the fan, and in a place where the duct
is straight and uniform for a length equal to several duct diameters. The
flow test plane should not be closer than 1 ½ duct diameters upstream
of the fan inlet, or 4 diameters from the fan outlet if it has to be on that
side.
Measurements of the pressure difference across the fan cannot, of
course, be made too far away from the inlet and outlet, and may
therefore have to be in less favourable flow conditions. A Pitot-static
tube should be traversed over the measuring points indicated in Fig.
11.2 and the readings of static pressure taken and averaged. The less
uniform these measurements are across the section, the less reliable
the average will be. If in particular the static pressure at the duct centre
is markedly less than that at the walls it shows that the airflow is
swirling; individual static pressure measurements are then likely to be
in error, and their average does not represent the true effective
pressure.
Suppose the average static pressure at the test plane upstream of the
fan is ps1, and that on the downstream side ps2. The volume flow rate q
may have been measured at one of these test planes, or it may have
been measured elsewhere. In either case we can calculate the average
velocities v1 and v2 at the test planes from their areas together with the
volume flows q,, q, correcting if necessary for density changes. Then
the conventional total pressure rise between the test planes is:
This quantity should equal the calculated total pressure drop for a
volume flow q round the system of airways outside the test planes.
Any differences must be accounted for either by errors of estimation or
by errors of test, or by system faults such as leaks or badly-matched
flanges. Pt2 - Pt1 plus the calculated pressure drop in any duct sections
between the test planes and the fan, should equal the rated fan total
pressure PFt. Any differences may be due to a faulty fan or fan
installation, or to site test errors, or to departures from the
standardised airway layout and provisions for pressure measurement.
11.3 Fan Tests with Standardised Airways

11.3.1 Installation types
A recent survey of the national fan performance test codes of eleven
industrialised countries showed 65 distinct test methods in use. No
doubt the majority of tests carried out in accordance with these codes
yield comparable results, but serious discrepancies can arise. A
committee of the International Organisation for Standardisation (ISO)
has been charged with the preparation of a uniform system and,
although agreement is as yet by no means complete, considerable
progress has been made.
The performance of a fan is affected by the connections made to its
inlet and outlet. From the infinite variety of possible connections, two
have been selected as representative of the range of performance:
In one the fan inlet or outlet is open directly to the unobstructed free
atmosphere.
In the other the fan inlet or outlet is connected to a long straight
duct of the same area as the inlet or outlet.
These two alternative connections may be combined in four possible
installation types as follows:
Type A: Free inlet, free outlet;

Type B: Free inlet, ducted outlet;
Type C: Ducted inlet, free outlet; and
Type D: Ducted inlet, ducted outlet.
Each of these installation types has, in principle, a different per-

formance characteristic. In practice the Type B and Type D
characteristics will coincide closely provided the fan is supplied, in its
free inlet form, with a properly-shaped entry cone or bell-mouth. With
the same proviso Type A performance will coincide with Type C. The
essential difference remaining is that between free outlet and ducted
outlet performance, which is significant for fans of all kinds though it
diminishes as the ratio of fan velocity pressure to fan total pressure
falls.
11.3.2 Examples of standardised airways

Fig. 11.3 gives, for each type of installation, one example out of
many standardised airways capable of realising the principles which
have international acceptance.
The space marked "fan" represents a cased fan under test, which may
be of any type-centrifugal, axial or mixed flow-having an inlet and
outlet, one or both of which may be designed for duct connection. Air-
flow is in the direction of the arrow.
No definitive standards have as yet been agreed, but in the

meantime many national standard methods, including those listed
below from the British and American codes, will provide a fair basis of
comparison The differences between the individual tests relate mainly
to methods o1 flow measurement and flow control. Some are best for a
permanent general-purpose installation, others for an airway
purpose-built for a large fan, others for small fans, and so on.
Methods broadly conforming to international principles are:
11.3.3 Fan pressure conventions

Satisfactory measurements of pressure cannot be made at the fan
inlet or outlet. It is necessary to establish pressure test stations some
distance upstream and downstream where the flow can be normalised.
The quantity measured at these stations is the static pressure, to
which is added a conventional velocity pressure to obtain the effective
total pressure.
When the fan draws air directly from the test enclosure as Type B,
Fig. 11.3, or effectively directly as Type D, and also when air is dis-
charged directly to the test enclosure as Types A and C, the test
station is any point in the surrounding atmosphere where the air is still
and unaffected by the flow from the fan. At this test station:
When the test station is in a duct or chamber the static pressure

may be measured by side tappings (Types A and C Fig. 11.3) or by
averaging the static pressures of a Pitot-static traverse (Types B and
D of Fig. 11.3). In these cases Ps is as measured relative to the
ambient atmosphere.
The fan total pressure is equai to the effective total pressure at the
outlet side test station minus the effective total pressure at the inlet
side test station plus an allowance for pressure loss between the fan
and the test stations. In a ducted outlet Type B or D test the added
allowance represents the friction of the ducts between the fan and the
test stations. In an open outlet Type A or C test, however, the
allowance also includes the conventional fan outlet velocity pressure.
This is because the whole
of the outlet kinetic energy is in fact dissipated within the outlet jet
as the air slows to near zero velocity at the test station where the
test enclosure ambient pressure is ps4 = pt4 = 0
The fan static pressure is by definition the fan total pressure
minus the conventional fan outlet velocity pressure. In the case of
open outlet Type A or C tests this deduction exactly cancels out the
added velocity pressure mentioned in the last paragraph, so that
the fan static pressure equals zero minus the effective total
pressure on the inlet side, e.g. for a Type A test:
The above are algebraic quantities, and sign must be carefully

considered. All velocity pressures are positive, but static pressures
are positive if above the surrounding atmosphere, negative if
below. On the inlet side static pressures are generally negative.
Thus if in a Type C test
The friction losses Pf in the test airways are taken as those that
would arise if the flow were axial, with the distribution finally
reached in a long straight duct. This is the "handbook value" which
a user would employ
in estimating the resistance of his system. The actual friction losses

are likely to be greater owing to the disturbed flow close to the fan,
but this excess is, by convention, ignored - that is it is not claimed
as part of the fan output although the fan does have to produce it.
The loss due to an average velocity v in a duct diameter D and
length L is:
In most current codes f is taken as 0.02, and L is increased to
include an allowance for the flow straightener. The code convention
must, of course, be followed in reporting fan performance under any
particular code. In practice f depends to a substantial extent on
Reynold's number, a proposed relationship appropriate to the normal
range of surface roughness found in test airways being:
11.3.4 Type A-free inlet, free outlet-standardised airways
In the Type A layout of Fig. 11.3 the inlet side test chamber has only
one essential requirement: that the flow pattern at the fan inlet should
not differ significantly from what it would be in a completely unobstruc-
ted free atmosphere. An area ratio chamber: fan inlet of at least 6: 1
secures this, together with smoothing screens to provide reasonably
uniform, axial, approach flow. An auxiliary booster fan is shown to over-
come the resistance of the test airways, enabling the fan under test to
be operated through to zero or even negative fan static pressures. This
might be a variable pitch axial booster fan, variable to negative pitch
angles to provide control resistance, and mounted between anti-swirl
units to protect the test fan and the flow measurement nozzle. Nozzles
are discussed in 11.3.7.
An inlet side test chamber is a versatile device, which may be per-
manently constructed and instrumented to produce, rapidly and
accurately, performance data for fans in a very wide range of sizes and
types. The test fan may be fitted with short, auxiliary airways to provide
Type B, C or D characteristics as well as Type A, as indicated in Fig.
11.4, b, c and d. Actual inlet side installation conditions may be
reproduced as for the propeller type wall fan of Fig. 11.4a. An
awkwardly-shaped unit such as a double inlet, free outlet, centrifugal
fan (Fig. 11.4e) or a crossflow fan may have casing and inlets wholly
immersed in the chamber. Fig. 11.4f shows an adaptation for
fan-powered roof ventilators which need to be tested with a vertical axis
because they have gravity-operated shutters.
11.3.5 Type B-free inlet, ducted outlet-standardised airways

The Type B diagram in Fig. 11.3 shows a Pitot-static tube traverse for
volume flow measurement, the differential pressure, Ps4 between the
static holes and atmosphere being also measured and averaged. Flow
control is by damper, although an auxiliary exhaust fan might also be
used for control. Pitot-static traverses are time-consuming, and are
mainly used for one-off acceptance tests on large fans to save the cost
of building more complicated airway systems.
Three other methods of test for volume flow, each requiring only a
single pressure-difference measurement to integrate the whole flow
rate, are shown in Fig. 11.5. By fitting a diffuser of 15 ° included angle
to the venturi nozzle, expanding to rather more than the original duct
diameter, almost all the fan characteristic can be covered, though an
auxiliary exhauster could be used if preferred. As the diagram
demonstrates, the immersed orifice plate is best suited to small fan
tests because of the length involved; an auxiliary exhaust fan would be
needed to cover the whole characteristic. The orifice at system outlet,
used both for flow measurement and flow control (by changing the
orifice plate) is the flow measurement principle of the French standard
NFX 10 (AFNOR, formerly CETIAT).
It will be noticed that all the airway layouts in Fig. 11.5 show a
common duct design from the fan outlet to plane AA, some five duct
diameters downstream, containing a standardised flow straightener in a
standard location. International standardisation of this "common

part" would ensure that different standardised airways would lead
to the same values of fan total pressure for Type B and Type D
installation tests.
The differing values of fan total pressure that arise without such
standardisation stem from the character of the airflow leaving the
fan. Even when free from swirl the velocity distribution at outlet
can, as illustrated in Fig. 11.6, be far from uniform. This results in
an excess of kinetic energy over the conventional allowance of ½
2
ρ2v2 , caused by the
proportionality of kinetic energy to the local value of pv3 (mass

flow x velocity pressure) so that the excess where v is high
exceeds the deficit where v is low.
Now the non-uniformity of the axial velocity components
diminishes as the flow proceeds down the duct and the excess
energy reaches a minimum of a few per cent. of ½ ρv2 within a
length equal to two to three duct diameters. Part of the original
excess is lost as indicated by the broken line in Fig. 11.6 but part
is converted into additional static pressure, as indicated by the full
lines, the conventional velocity pressure remaining constant. This
addition to the fan static pressure and the conventional fan total
pressure is available for overcoming external
resistance, and in order to credit it to the fan, as it should be for Type

B and Type D installations, the test station for outlet side pressure
measurement should be more than three duct diameters from the
outlet.
On the other hand the swirl energy at fan outlet is never recovered
in a straight uniform duct, and only decays over long distances-more
than 100 diameters. In the presence of swirl simple measurements of
effective pressure or volume flow are impossible, and it must therefore
be removed when tests are to be taken in a duct on the outlet side for
Type B or D performance. An effective flow straightener will do this, and
if it removed just the swirl energy and no more the minimum energy
convention would be satisfied. Unfortunately, the energy actually
removed is very dependent on the combination of swirl pattern and
straightener design, hence the need to standardise the latter.
In practice a fan with pronounced outlet swirl ought not to be
selected for use with a long straight outlet side duct, because the
friction loss in the latter will be substantially increased. Guide vanes
should be fitted which will remove and recover (instead of removing and
destroying) the
swirl energy. The flow straightener will then merely ensure that the test
conditions are satisfactory in the downstream duct; the relatively small
outlet swirl components from centrifugal, guide-vane axial or contra-
rotating fans will be removed without significant disturbance to the
performance.
11.3.6 Type C-ducted inlet, free outlet-standardised airways

These difficulties of definition and test do not arise with a Type C
installation. Satisfactory conditions for measuring volume flow and
(negative) effective total pressure are easily achieved in an inlet side
duct. Subtracting this total pressure from the static pressure at fan
outlet, which is simply taken as zero, gives the fan static pressure. The
2
fan velocity pressure ½ ρ2v2 is added to give the fan total pressure for
2
a meaningful statement of fan efficiency. This ½ ρ2v2 will be taken
away again as the outlet loss part of the system resistance (see Chapter
6) leaving the fan static pressure as the effective pressure for
overcoming the upstream part of the system resistance when the fan
has free outlet.
A Type C test installation can be extended to provide Type D per-
formance by fitting a "common part" outlet duct as illustrated in item 2
Fig. 11.7, here shown with an open outlet in the position occupied by
the static tappings of Fig. 11.5, giving virtually the same result. This is
valid for a fan with limited outlet swirl, but unreal differences can arise
between Type C and Type D performances as Fig. 11.7 illustrates for an
extreme case - an axial fan without guide vanes having a high propor-
tion of fan velocity pressure. When swirling air is discharged into the
free atmosphere it forms an expanding vortex which does convert part
of the swirl energy into useful pressure, so that curve 1 is the true Type
C performance characteristic. When swirl is removed in a straightener,
curve 2, this conversion does not take place, leading to increasing loss
of nominal Type D fan pressure as the resistance rises and the fan
outlet swirl correspondingly increases.
As explained in the preceding section airway 2 does not give a
practical characteristic for a long duct application when there is swirl,
since it includes a non-existent flow straightener. Only a test with the
actual long duct could determine the true performance, and it is in any
case bad practice to use a fan with pronounced swirl in such a case.
With a short outlet duct, however, the swirl energy conversion will still
take place at the duct outlet, and will more than compensate for the
increased friction in the duct. The UK and USA national test codes have
for many years included methods with an outlet duct two or three
diameters long to represent ducted outlet. For fans without outlet swirl
this practice will provide a true Type D characteristic. When the fan has
pronounced swirl the characteristic, of which curve 3 is an example, will
include both axial excess energy and swirl energy recovery. This will be
a fair characteristic for short ducts, say up to five diameters long, but
cannot be termed a Type D performance.
Even better recovery of swirl energy is obtainable with an outlet
diffuser to control the vortex expansion. Steeper diffuser angles are
possible without flow separation, but it is essential to determine the
characteristic for the fan and diffuser in combination as one unit. The
fan velocity pressure is now the lower value at diffuser outlet, and
the broken line 4 shows that there is a loss of total pressure in the
combination. The fan static pressure is then a better standard of
comparison, giving the pressure available to overcome the upstream
system resistance, without need to specify or allow for the area
selected for the fan or diffuser outlet.
The performance of a fan with outlet swirl can be presented by

plotting fan static pressure, fan velocity pressure and fan (total)
efficiency against volume flow for the Type C, open outlet,
configuration. A factor based on test results can then be quoted by
which the fan velocity pressure should be multiplied to give the
additional static pressure available when a short outlet duct, a
diffuser of specified dimensions, or any other outlet side component is
fitted.
11.3.7 Flow measurement by nozzle and orifice

Volume flow in standardised airways may be determined by Fitot-
static tube traverse exactly as described in paragraphs 11.2.2 and 3.
With careful observance of the requirements of a national test code it is
possible to limit the error to ± 2%.
Nozzles and orifices depend on the measurement of the pressure drop
across a restriction through which the whole flow is passed and several
varieties are illustrated in Fig. 11.8. The orifice is in one sense a simple
hole in a plate, but it is also an instrument, and must be made and
maintained with careful attention to code requirements - particularly the
sharpness of the upstream edge of the hole. A nozzle is a smoothly-
formed contraction to a parallel throat. Dimensional details are fairly
complicated, and should be taken from a standardised test code. A

nozzle becomes a venturi-nozzle when the throat is followed by a
gradual expansion designed to recover most of the pressure drop, which
is now measured by a static tapping in the throat. A classical venturi-
tube is the forerunner of the venturi-nozzle, formed with a 21 ° cone in
place of the curved contraction.
General flow measurement codes such as BS 1042, DIN 1952, ISO
8541 and ISO 8781 are limited to the use of orifices and nozzles
inserted in a length of duct or pipe. However, several fan test codes
give coefficients which have been established when the same devices
are fitted at the inlet to a duct from free space or at the outlet from a
duct into space.
11.3.8 Calculation of flow rate

The flow rate can be calculated either in terms of volume flow:
or of mass flow:
d is the throat diameter of nozzle or orifice in m.
D is the duct diameter at the upstream pressure tapping in m.
∆p is the upstream static pressure minus the downstream or throat
static pressure in Pa.
ρ 1 is the density at the upstream tapping in kg/m3.
Qv is measured in m3/s at the upstream density, ρ 1
Qm is constant right through the airway system in kg/m3.
α is the flow coefficient, as given in the appropriate code.

ε is the expansibility factor, as given in the appropriate code.
For testing with atmospheric air ε may be taken as 1.00
provided ∆p does not exceed 1000 Pa.
α is sometimes, but less conveniently, replaced by CE where
C is the coefficient of discharge and
E is the velocity of approach factor =
Both a and C are dependent on
β= d/D the diameter ratio and
ReD or Red the duct or throat Reynold's number which is either

based on the average velocity and air conditions in the
approach duct D, or those in the nozzle throat or orifice
opening d.
The Reynold's number need only be known quite roughly since the
changes it produces in α or C are small. When testing with atmospheric
air it is sufficient to use the approximate formulae below, using either Qv
the expected volume flow, or ∆p the observed pressure difference.
Preliminary values of α and a may be taken from Fig. 11.9. Having

found Re a precise value of α can be looked up in the charts or tables
in the code, which will also give precise data for ε, appropriate to the
particular device.
Any flow measurement device uses up some of the fan total

pressure, and representative pressure losses are given in Fig. 11.10.
It is assumed that they are placed in an outlet side test duct, and the
velocity pressure lost at final outlet from the duct or device is
included. The loss is plotted against the useful pressure, ∆p, both in
terms of the velocity pressure upstream of the device. The lower the
loss the closer we can carry the fan characteristic towards completely
free flow without needing an auxiliary booster fan.
11.3.9 Calculation of air density

The first step is to calculate the density of the ambient air in the test
room. For ordinary purposes measurement of the barometric pressure,
po Pa, and the dry-bulb temperature, to °C, is sufficient, the density
being:
The above formula is correct for air with 1% moisture content by

weight and holds within ± 0.5% for any condition between dry air
at any temperature and saturated air up to 25°C.
For precision tests, and in air of tropical heat and humidity, it is
necessary to measure the wet-bulb temperature, tw °C, preferably
with a sling or Assmann hygrometer. The density may then be
found from the following formula, which is correct within ± 0.1%,
or from tabulated data or psychrometric charts. See also Table
14.10 and Fig. 14.1 in Chapter 14.
The factor A depends on tw only and may be obtained from the

following linear chart:
In a standardised airway test the density, ρx, is required at the

section of pressure measurement upstream of the fan and also
upstream of the flow measurement device; in each case the static
pressure, psx Pa, and the dry-bulb temperature, tx °C, will be
observed. Then noting that psx may be positive or negative:
Example. Upstream of the fan in a Type C standardised airway the

pressure ps = -760 Pa and the temperature 28°C. The reference
barometric pressure is 101,300 Pa, dry-bulb 24°C and wet-bulb
12°C. What is the fan inlet density ρx?
Ignoring humidity:
Using the more precise formula, note that at tw = 12°C, A = 0.0054.
11.3.10 Calculation of efficiency

The impeller power measures the work done to drive the impeller
round against the aerodynamic forces. The shaft power is often the
quantity actually measured. It includes the bearing losses, which
make a negligible difference except in very small fans, provided the
bearings are of ball or roller type. Losses in transmission elements
such as belt drives must be estimated separately, and deducted
before calculating the fan total efficiency. An overall efficiency is
sometimes quoted to include the losses in an electric driving motor, or
other source of power.
The power input is best measured by a swinging frame
dynamometer if a suitable one is available. Usually, however, the
quantity measured will be the electrical power input at the driving
motor. This may have an efficiency calibration or the efficiency may be
calculated by the summation of losses method. In both cases it is
important to make sure that the working temperature of the motor
windings is about the same as that on which the efficiency estimate is
based. If:
The above simple formula holds for moderate fan pressures-say up

to 2500 Pa-when the air can be considered incompressible. At higher
pressures it is necessary to take account of the rise in air density
through the impeller: the air discharged will be compressed to a
volume flow rate Qv2 smaller than Qv1 Both the work done and the
power input are reduced by these changes, and, to avoid
overestimating the efficiency, 0.1 must be reduced to a mean value Kp
Qv1where Kp is called the compressibility factor.
The change in density is not a simple function of the pressure rise

because the air is also heated, both by the compression and by the
losses in the fan. At first sight there is a difficulty here because we do
not know the losses until we know the efficiency, and we do not know
the
efficiency until we know the losses! Fortunately, trial and error

methods can be avoided by applying the thermodynamic laws of
polytropic compression which yield:
Calculation of Kp can be avoided by the use of Fig. 11.12, which is

drawn for direct use with air, i.e. with y = 1.4.
There is not a big difference, but, for a machine approaching the

limiting pressure rise for a fan, r = 1.3 or 30,000 Pa with
atmospheric pressure at inlet, the drop would be from 86% apparent
to 79% true efficiency.
11.4 The Fan Laws
11.4.1 Change of speed and density

A standardised airway test will be carried out at a speed, N rev/s,
and inlet density, ρ, kg/m3, which are unlikely to be exactly those
specified. To find the performance at the required speed, N1, and
density, ρ1, (which will normally be 1.20 kg/m3) the observed values
of inlet volume
flow, Qv1, fan total pressure, PFt etc. should be converted to Q1v1PFt
etc. by the use of the following relationships:
where 10a ranges from 50 for some centrifugal fans to 55 for

some axial fans and p1 is the absolute inlet pressure (see Section
10.4). The coefficients 22 and 12 are approximations to the small
corrections required.
Note that the efficiency does not change. The fan velocity
pressure and fan static pressure change in the same ratio as the
fan total pressure. a, the speed exponent for sound power, will be
between 5.0 and 5.6 according to the type and design of fan. All
the formulae are sufficiently accurate for use at fan total pressures
up to 2500 Pa in atmospheric air. Above that pressure account
should be taken of compressibility - see 11.4.3.
11.4.2 Change of size-model tests

When the performance of a model fan of diameter D m is to be
used to predict the performance of a larger, geometrically similar, fan
of diameter D' the rules for speed and density change are extended as
follows:
In the formula for sound power level a has the same

significance as in Section 11.4.1, but account has been taken of
the possibility of change in inlet absolute pressure p1 and in the
gas, y. These do not affect the other formulae.
The provisos of the previous section still hold. The fan tested
must always be the smaller of the two, and if the size difference is
great, the
larger fan may be expected to develop slightly more than the fan total
pressure given by the above formula, with corresponding slight
improvement in efficiency. This is the so-called scale effect, and
adjustments to the simple formula can only be made if both parties to
the test agree that there is satisfactory experimental evidence for a
scaling formula applying to the type and design of fan in question.
If the performance of a whole range of fans is to be predicted it is
best to take thorough performance tests at diameter intervals of not
more than 2: 1. Not only is there a Reynold's number effect, but there
are inevitably some significant departures from true geometric
similarity, notably in surface roughness (which should increase with
size, but does not), blade thickness and impeller to casing clearances.
11.4.3 Influence of compression on the fan laws

Suppose that a high-pressure fan is tested at half speed, and that it
raises the air density by 2½% in passing from inlet to outlet of the
impeller. If the speed is now doubled to full speed the compression will
become 10% which means that, while the peripheral speed at impeller
outlet has increased by 100%, the axial air velocity has only increased
by 85%. Clearly this alters the flow pattern through the impeller, and
we cannot expect the full speed performance to be accurately predicted
by applying the simple speed change law to the half speed test. While
geometrical similarity of impeller and casing are preserved geometrical
similarity of air flow has been lost.
However, the prediction is found to be much better if, instead of
applying the simple laws to the inlet volume flow and density Qv1 and
ρ1, we apply them to a mean air condition, Qvm and ρm, which can be
visualised as existing half-way through the impeller. Equal and opposite
distortions of the flow patterns at inlet and outlet will minimise the
errors. A suitable mean density is defined by the use of the compressi-
bility factor Kp (see 11.3.10).
The mean volume flow is not directly applicable to performance ratings

or practical application, so the mean flow form of the laws is best
converted by reference to the inlet condition thus:
The compressibility coefficient K'P applying to the predicted per-

formance is not known in advance. It can, however, be calculated
from the following formula, in terms of quantities which are all
known from the test.
The fan laws in this form need not be applied below 2500 Pa, nor
for Kp changes of less than 0.01. They should hold for values of K'P
within the range (Kp ± 0.03). For larger changes in compression
ratio they will give the best available prediction, when
supplemented by any necessary scale effect adjustments for size
change. If there is a change of gas with isentropic exponent y' in
place of y and a computer is available the exact value of Kp/ K'P
could be calculated from:
11.4.4 Examples of the application of the fan laws
Conversion from test to standard conditions

A 1000mm fan is tested at 1380 rev/min (23.0 rev/s) in a Type B
standardised airway with the following results:
What is the performance at 1470 rev/min (24.5 rev/s) and 1.20

kg/m3 inlet density?
Performance prediction with change of size

What is the expected performance of a 1250mm fan at 1770
rev/min (29.5 rev/s), geometrically similar to that above?
Correction for compressibility

At the 4020 Pa of the last example we are in the range where
compressibility should be considered. In the original, 1000mm,
1380 rev/ min, test:
Exact calculation using the computer formula gives 1.0077.
The changes in this case are insignificant, justifying ignoring

compressibility below 3000 Pa.
11.4.5 Non-dimensional coefficients

Some readers will be surprised that no reference has been
made to the non-dimensional volume flow, fan pressure, and
power coefficients Φ, Ψ and λ. Confusion can be caused by the
lack of international agreement on the reference quantities for
these, leading to differing numerical values, and practical
requirements are met by the above treatment. They are, in fact,
of more use to fan designers than fan users, and each designer
will have his own personal definitions.
11.5 Accuracy
11.5.1 Uncertainty of measurement

Standard fan test codes do not always state the accuracy levels
appropriate to the measurements to be made, and Table 11.1
contains some suggested values for use in default of specific code
instructions. The first column is a normal standard, suitable for
most tests in standardised airways. A need may be felt in some
cases for greater precision, particularly when testing large, high
efficiency, fans which contribute significantly to the running costs
and performance of a plant. For this reason a second column has
been included suggesting the accuracy likely to be attainable with
first-class equipment and trained staff, together with the additional
calculation steps which then become justified.
There are basic uncertainties in the values of the flow coefficient

a for the standardised nozzles and orifices. These range from ±
0.5% to 1.5% and combine with the errors of measurement to
produce a probable uncertainty of ± 2.0% to 2.5% for flow
measurement in a standard test. In a precision test readings should
be repeated until the desired level of statistical consistency is
reached, which should secure a flow measurement uncertainty of ±
1.5% if the fan produces reasonably smooth and steady flow.
The uncertainty in determining fan pressures depends on the

proportion of fan velocity pressure to fan total pressure and may be
obtained from the following expression in which a = 1.5% to 2.0%
and b = 5% to 10% :
To summarise the lowest uncertainties that can be expected from

standard test procedures will occur around the best efficiency point
and are likely to be:
The percentages quoted here are all 95% confidence limits.
This is a statistical term and means that:

11.5.2 Tolerance on fan performance

It would be futile to guarantee fan performance to closer limits than
the uncertainties of measurement, and to those above must be added
errors in measuring speed and air density. In addition there will be the
manufacturing tolerances - all fans to a given design cannot be
expected to be identical. Fortunately, the combination of the fan
supplied with the user's system greatly reduces the deviation from the
specified volume flow due to departures from the specified fan
performance, and normally this is the significant issue.
Suppose that a fan has been tested, and a smooth fan characteristic,
the full line in Fig. 11.13, drawn through the test points. At any point
on the fan characteristic we can plot the tolerance range of volume

flow horizontally and fan pressure vertically. Now the extreme
positive value of q occurs once in 40 tests only and so does the
extreme positive value of p. So they are extremely unlikely (once in
40 x 40 = 1,600 tests) to occur simultaneously. In fact the mutual
dependence of the two errors produces an elliptical area within
which the true test point has a 95% probability of lying. Strictly
speaking the way in which the fan pressure is calculated gives the
pressure error a slight dependence on
the volume flow error, which is allowed for by canting the ellipse at
high volumes.
It will be seen that the broken lines touching the outer edges of
all the error ellipses will be two fan characteristics between which
the true characteristic will be in 95% of cases. Crossing these has
been drawn a system resistance line-here a "square law" line, but
see Chapter 9 for other possibilities-representing exactly the user's
estimate. The intersections with the broken lines now give the
range of volume flow within which the fan performance will be.
Often this will be one half or less of the basic uncertainty range of
volume flow.
CHAPTER 12
Use of air in heat

exchange and drying
In many industries the product requires to be heated, cooled or

dried at some stage in the manufacturing process. The techniques
involved are numerous and varied, and it would be quite outside
the scope of this book to attempt to cover such a broad field.
However, atmospheric air is often a partner with the product in an
exchange process in which heat or moisture is transferred from one
to the other. The effectiveness and economy of this process is
greatly dependent on the way in which the airflow is organised and
there are plenty of examples of gross waste of fan power. This
chapter deals in a very general way with the principles of heat and
moisture transfer.
12.1 Heat Transfer

12.1.1 Heat capacity of air
For ordinary heat transfer processes, which take place without
significant compression or expansion of the air, the appropriate
constant is cp the specific heat at constant pressure. This is the heat
in joules (or more conveniently kilojoules, kJ) required to raise the
temperature of 1 kg of air by 1 °C. It is practically independent of
pressure and varies only slightly with temperature, thus:
Temperature, °C 20 100 200 300 400 500
cp - kJ/kg, °C 1.006 1.01 1.02 1.04 1.06 1.09
For water vapour cp is about 2.01 kJ/kg °C, which means an
effective value for cp of approximately (1.01 + m) in the case of
humid atmospheric air containing m kg of moisture per kg mixture
- a correction that
259
can usually be neglected. m is greater in hot air drying processes, but the
much greater heats of evaporation or condensation are then the predominant
factors.
Example. Ambient air at a maximum temperature of 40°C is passed at a rate

of 0.60m3/S through an electrical machine to remove losses estimated at 6.0
kW. What is the average temperature of this air as it leaves the machine?
Assuming a barometric pressure of 100,000 Pa the density at inlet is:
The airflow is thus 0.66 kg/s. Adding heat at the rate of 6.0 kW = 6.0 kJ/s
produces a temperature rise of:
12.1.2 Heat transfer by convection

Convection implies the continuous movement of successive masses of air to
and from a hot surface, extracting and taking away heat as they pass (strictly
speaking the first stage of heat transfer takes place by conduction across a
thin stagnant film of gas which never leaves the surface). The movement may
be due simply to the tendency of the heated air, being less dense, to rise and
make room for fresh, cold air. This is natural convection as found in many
familiar space heating appliances. The converse passage of heat from the air
to a colder surface follows similar laws, but for simplicity we shall speak as if
the air were always the colder.
Natural convection is too slow and uncertain for most industrial processes,
and is replaced by forced convection. In this the surface from which heat is to
be extracted is placed in a relatively high velocity air stream generated by a
fan. The air stream is preferably confined by a casing which ensures that
none is wasted, passing too far away to take part in the heat extraction.
Calculation for both natural and forced convection is made from the
relationship:
The difference is that, in forced convection, hc is dependent on the air

velocity and the pattern of forced air flow over the surface, whereas in natural
convection ho is dependent on the temperature difference and the geometry
of the surface which together determine the pattern of air movement.
Use of air in heat exchange and drying units 261
12.1.3 Forced convection

For a high rate of heat transfer, turbulent flow must be established over most
of the hot surface. This is necessary for the thorough mixing required to carry
away the heat-in laminar, streamline, flow only the layers next to the surface will
be heated. In fact there is a strong parallelism between the friction factor K of
Chapter 6 and hc - high values of one tend to accompany high values of the
other.
With turbulent flow parallel to smooth surfaces as in Fig. 12.1, hc is found to be
proportional to (mass flow)0.8. With separated flow, however, as when air flows at
right-angles to a bank of tubes, hc is more nearly proportional to (mass flow)0.6.
The constant of proportionality can only be guessed at in a complex assembly
such as a finned tube battery, for which the maker's heat transfer data should in
any case be consulted.
Estimates for simple cases with flow parallel to smooth surfaces can sometimes
be built up from the established elements plotted in Fig. 12.1. The graph applies
to air at 16°C and 100,000 Pa, and corrections are given allowing for density and
viscosity changes at other temperatures and pressures. Surface roughness will
increase ho, by a factor of about 2 at 500 µ m for example.
Example. In an accelerated paint drying enclosure wet sheets each 1m x 2m are

suspended at 100mm spacing. Air at 80°C dry-bulb, 38°C wet-bulb flows between
them at 8m/s. At what rate is heat supplied to each sheet?
The 4m2 surface (both sides) will be at the wet-bulb temperature of 38°C so
that ∆t = 80 - 38 = 42°C. From Fig. 12.1 the appropriate diagram for T = 100mm
and v = 8m/s gives ho = 25 W/m2 °C. At 38°C, Kt = 0.96.
12.1.4 Natural convection

As in the case of forced convection, the natural heat loss from objects of
complex form can only be found by trial. However, for certain simple shapes, the
data are well established, and some of these are summarised in Fig. 12.2. The
figures should be used with caution, noting the following limitations on their
reliability.
The hot surface should be sufficiently isolated in space for air to flow freely to
the base and away at the top under the small gravitational forces generated by
changes in density. Thus the vertical surface curve would not be accurate for a
wall heated all over, still less for a ceiling where the inability of the heated air to
get away would make nonsense of the assumptions. On the other hand both
faces of a domestic "radiator' are effective natural convection heat sources, even
if one is spaced only a few centimetres from the wall.
A draught in the 0-. to 1 m/s range can increase the natural convection heat
transfer of a small source by 50% to 100%, or more if the temperature
difference is low. On the other hand a large, hot, surface creates its own
draught pattern, which is much less sensitive to casual air movement.
12.1.5 Radiation
All bodies radiate heat at a rate which is proportional to the fourth power of
their absolute temperature - (273 + t°C). On the other hand all the surfaces
around radiate heat back, and if they are at the same temperature there is
precise balance and no heat is transferred. The appropriate temperature to take
for the surroundings is the mean radiant temperature, tr°C (see 1.9) and if the
hot surface is at t°C the net heat emission will be:
The factor ε is called the emissivity and ranges from near 1.0 for a matt black
surface to nearly zero for a brightly polished mirror-like one, which not only
reflects back most of the radiant heat that reaches it, but also retains precisely
the same fraction (1 - ε) of the radiant heat it could emit as a black body,
thereby preserving balance in surroundings of equal temperature.
Fig. 12.3 is based on the above formula, with tr at room temperature level,
and Table 12.1 records the current recommendations of the IHVE for the
emissivity of surfaces found in heating practice. The radiant heat
transfer is additional to, and completely independent of, forced or natural

convection transfers. Fig. 12.3 applies only when the walls of the room or
enclosure are at a relatively uniform temperature. They then become
effectively one "black body" (ε = 1) since they absorb all the radiation in the
room, some after repeated reflection. When there is a strong radiant heat
source, or a sink such as a large window, more detailed analysis is
necessary.
The heat transfer coefficient hr W/m2 °C is directly comparable with hc for
convection, and in the following formula for the heat transfer rate H, the
area A m2 is that of all exposed surfaces. In re-entrant pockets of surface,
however, only the visible, projected area across the mouth of the recess
should be counted.
H = hr A (t - tr) watts (122)
Example. A white-painted panel in a hot water heating system is 2.Om long

by 0.5m high and has a surface temperature of 75°C. What heat output
might be expected in a room with an air temperature of 25°C and mean
radiant temperature 20 °C ?
From curve B in Fig. 12.2 and a temperature difference of 75 - 25 = 50°C,
hc = 5.0 while the surface area of both sides of the panel is 2.0 m2. The heat
transfer by natural convection is thus:
From Fig. 12.3 with a temperature difference of 75 - 20 = 55 °C, and an

emissivity assessed at 0.95, hc = 6.5, and the radiant heat transfer rate is:
This gives an estimate of just over 1200 watts, but it must be emphasised that
an actual radiator will be rated on the basis of tests under controlled conditions
of installation, and in terms of the entering water temperature and flow rate
which set the surface temperature distribution over the whole unit.
12.2 Heat Exchangers

12.2.1 Temperature distribution
Consider a heat exchange battery designed to transfer heat from hot water
flowing through a bank of tubes to an air stream flowing over them. The water
will be cooled and the air heated progressively as each passes through the
battery. The temperature at which each stream leaves the battery can be
calculated from the quantity of heat transferred and the heat capacity of each
stream as in the following example:
Heat transferred: 100 kW
Entering air flow: 1.20m3/s at 0°C: 1.53 kg/s

cp = 1 -01 kJ/kg 'C temperature rise = 100/(1 -53 x 1 -01) = 65 'C
Entering water flow: 1.20 kg/s at 100°C

cp = 4.20 kJ/kg °C temperature fall = 100/(1.20 x 4.20) = 20°C
Clearly the local temperature difference across the tube wall, on which the
local heat transfer depends, will vary widely from point to point through the
battery. It might range from as much as 100°C if the entry points of air and
water coincide to as little as (100 - 20) - 65 = 15°C if the exit points coincide.
A simple average of the entry difference and the exit difference will always
overestimate the effective value. The best estimate is usually the quantity
known as the logarithmic mean temperature difference ∆tlm. If, of the air
entry and air exit differences, the larger is ∆tmx and the smaller ∆tmn
Fig. 12.4 also illustrates the case we have just calculated in the diagram
marked "parallel flow". In this it is assumed that air and water enter together
at one end of the battery, travel along its length, and emerge together.
Clearly, in this case:
If in this same battery we were to reverse the flow of water so that hot
water enters where hot air leaves, we should have a "counter flow" system,
and would find that the temperatures became those shown in the second
diagram. Then:
We see that the log mean temperature rise has gone up by about 10%,
and so have the temperature rise of the air and the temperature fall of the
water. This equality of percentage changes is in fact the condition that
determines the new values, and it means also that the heat exchange rate of
the battery has gone up 10% to 110 kW. Indeed a fully counterflow system
quite generally makes the most effective use of a given battery if it can be
achieved.
12.2.2 Effectiveness of heat exchangers

In most heat exchangers involving two fluids such as air and water the
objective is to change the temperature of one of them, the primary fluid,
only. The other, secondary, fluid is there merely as a means to an end. From
one point of view we might suppose that the exchanger was
100% effective if it brought the primary fluid right to the entering tem-
perature of the secondary. Not only is this strictly impossible, however, but it
would require a grossly oversized exchanger and excessive secondary flow to
get close to it. Nevertheless, the heat exchange effectiveness defined as:
is a valuable index of performance. It also has the property of remaining the

same for geometrically similar heat exchangers whatever the temperatures of
the two fluids, provided the flow rates are kept in the same proportions.
If we have a flow of MP kg/s of the primary fluid with specific heat cp

kJ/kg °C it will have a heat capacity of GP = MP cp kW/°C, this being the heat
transfer rate needed to change its temperature by 1 °C. The heat exchange
effectiveness E now depends on two factors:
GH/GP and Gs/GP
where GH is the heat exchange capacity of the exchanger in kW per °C log

mean temperature difference and Gs is the heat capacity of the flow of
secondary fluid. Gs = Mscs, kW/°C. The curve marked m is used for steam.
Because steam supplies heat by condensing at substantially a
constant temperature throughout the battery tubing it is equivalent to an almost

infinite flow of hot water which could give up heat without any drop in
temperature.
The general nature of the dependence of E on these two factors is shown in Fig.
12.5. The cross-flow chart applies to the form of heat exchanger typified by the air
heater battery in which a narrow bank of tubes spans a duct through which the air
flows. This particular chart is enlarged in Fig. 12.6 to demonstrate its use in
predicting the result of changes in temperatures and flow rates when the
performance of a heat exchanger is known for one set of conditions only. Changes
in flow rate will change Gs or GP with resulting change in GH. For a cross-flow
battery with most of the temperature difference on the air side GH will be roughly
proportional to (Gs)0.6 - or (GP)0.6 if air is the primary fluid. The linear chart in Fig.
12.7 follows this relationship.
12.2.3 Example-Use of effectiveness chart

A process liquid circulating at 110°C is to be cooled to 50°C before discharge, to
limit flammability. This is done in a cross-flow heat exchange battery using
atmospheric air as a secondary fluid. The battery is designed to deal with 4 kg/sec
of liquid having a specific heat of 1.80 kJ/kg °C with an air flow of 32m3/s at 35°C.
Combining this with Gs / GP = 36/7.2 = 5.0 we see from Fig. 12.6 that GH / GP
should be 2.0. Thus the temperature difference across the exchanger should be
∆tH = AtP x (GP / GH) = 30°C. This checks with a log mean temperature
difference derived on the assumption that the
air temperature is constant at its mean value of 41 °C - fair enough for a

cross-flow exchanger.
Suppose now that a variable pitch fan is used, set to control the volume flow
over the range 32 to 12m3/s in order to maintain the liquid discharge at 50°C
irrespective of air temperature. At 12m3/s Gs will be reduced to 37% of its
original value.
New Gs / GP = 0.37 x 5.0 = 1.9

From the linear chart in Fig. 12.7 a reduction to 37% in Gs produces a

reduction to 55% in GH.
New GH / GP = 0.55 x 2.0 = 1.10
Entering these values on Fig. 12.6 we obtain a new effectiveness E = 0.57. In

order to cool the liquid from 110°C to 50°C we now need a temperature
difference at entry of (110 - 50)/0.57 = 105 °C, i.e. an entering air temperature
of 110 - 105 = 5°C.
We can therefore conclude that the variable pitch fan can be used to control
the liquid discharge temperature at 50°C while the atmospheric temperature
varies over the range 5 °C to 35 °C. The potential for energy saving can be seen
from completion of the exercise:
12.3 Industrial Drying

12.3.1 General principles
Metals apart, almost all the materials used in industry and commerce contain
significant amounts of free moisture - that is water which is not in chemical
combination. The quantity of this hygroscopic moisture varies throughout the life
of the material, being in equilibrium with the relative humidity of the atmosphere
at levels indicated in Fig. 12.8 for a few natural organic materials. A number of
important properties of the material are dependent on the moisture content,
notably mechanical strength, handling properties, and electrical and thermal
resistance. Also life may be greatly affected through the biological attack of
fungus and mould, when damp.
In service, paints, varnishes and impregnants will slow down, though they
cannot altogether stop, deterioration, and sensitive materials must be protected
from too damp or too dry an environment - particularly one which swings from
one extreme to the other. When first produced, however, many natural and some
synthetic products contain much too much free moisture. These must be dried to
a moisture content which
experience has shown to be satisfactory for transport, storage, further

processing or marketing, but which may substantially exceed the equilibrium
moisture content.
Quality is often greatly influenced by the drying process, and slow,
traditional, sun-and-air, methods have tended to set the standards. In
engineering applications drying implies the acceleration and control of the
natural process to secure a cheaper and more consistent product.
Drying times will vary from minutes to weeks, and rates must be carefully
scheduled to suit the product. Too rapid drying causes damage in several
ways:
Parts already dried may be overheated.
Over-rapid evaporation may injure the fibres as water forces its way out
of the interior.
If the surface layers are allowed to dry out while the interior is still wet,
they will shrink and may crack or warp the whole piece.
12.3.2 Moisture content

The weighing of a sample before and after the drying process will clearly
give the moisture removed as a percentage of the initial wet weight. To find
the dry weight it will be necessary to raise the temperature and allow time
for all the hygroscopic moisture to be expelled (but
not that which is chemically combined, as in crystalline hydrates). As this

complete desiccation may damage the substance, it is likely to be carried out
on a small specimen only. In drying calculations it is usually more convenient
to express the moisture content on the basis of dry weight thus:
It is important always to distinguish this quantity from the moisture content on

the basis of wet weight which is:
The relationship between the two is given in Fig. 12.9.
12.3.3 Drying processes

Bulk liquid and large amounts of surface water are most economically
removed by mechanical means before the stage of drying proper begins.
Sedimentation in settling tanks, filtration, drainage by gravity on mesh
support, or by centrifuging which is accelerated draining, and mechanical
pressing between rolls or otherwise are common methods.
Hygroscopic moisture on the other hand can be removed only by

evaporation, which is the process we are concerned with in this chapter.
Two things are necessary to promote evaporative drying:
Heat must be supplied-almost as much heat is needed to evaporate a

given weight of moisture at room temperature as would be needed to boil it.
The air immediately in contact with the surface is quickly saturated, and
must be continually replaced if evaporation is to continue.
Industrial driers may be classified according to the way in which the air is
brought into contact with the moist surface and replaced as it saturates.
Drying sheds or lofts are naturally ventilated buildings in which the product
is stacked, hung or festooned for comparatively long periods. Air movement
may be promoted by circulating fans.
Batch driers are heated ovens in which the product is spread on trays or racks.
Sufficient air change must be ensured to prevent saturation, and an internal air
circulating system may be incorporated. Air temperature and humidity may be
adjusted to a programme covering the drying period.
Conveyer driers are similar in principle to batch driers, but provided with a
continuously or intermittently moving conveyer on which wet stock enters at one
end and dried stock leaves at the other. Air conditions will be steady with time,
but may be adjusted along the length of the drier, particularly by a counter flow
system.
Through flow driers are used for a granular or porous product supported on a
perforated base so that air may be passed directly through it. Hop kilns and
in-sack grain driers are examples.
Rotary driers are rotating drums through which the stock is cascaded or
tumbled so that it comes into intimate contact with air flowing through.
Spray driers inject a solution or slurry through nozzles forming a fine spray
which evaporates rapidly, leaving the product in powder form.
Heat may be supplied wholly through the air stream, which is preheated before
entering the drier. If the drying temperature is high economy depends on a close
approach to saturation in the air leaving the drier. At low drying temperatures,
although drying times are longer, heating may be dispensed with altogether,
particularly in a hot, dry climate. The ambient air itself is then the source of heat,
which will flow to the stock by virtue of the low, wet -bulb, temperature of the
surface.
Alternatively, the heat may be applied directly to the stock, generally by
radiation which is particularly suitable for sheets and similar forms with a high
ratio of exposed surface to mass. In special cases heat may be generated within
the body of the product, either by resistance heating from an electric current
flowing through, or by electro-magnetic induction or dielectric heating in a high
frequency field.
12.3.4 The physical mechanism of drying

In the evaporative drying of any product, four main phases may be
distinguished.
1. The rising rate period during which the heat supplied is used to raise the
temperature of the wet stock to the designed maximum, with a
correspondingly rising evaporation rate.
2. The constant rate period during which temperature is steady, the surface of
the stock remains wet, and the evaporation rate is constant, and at its
maximum.
3. The first falling rate period, which commences at the critical moisture content
for the material, when dry patches begin to appear on the surface and the
evaporation rate falls.
4. The second falling rate period, which is governed by the rate of evaporation,
migration and diffusion of moisture within the stock. The surface appears dry,
and air humidity no longer influences the rate.
From the point of view of the fan engineer the constant rate period is the
significant one. The air flow must remove the moisture generated at a
planned humidity level, and the heat required for the evaporation must be
supplied, both being at their maximum in this period. Control to lower levels
of air flow and heat supply will be appropriate to the falling rate periods.
12.4 Drying Calculations

12.4.1 Heat required for evaporation
During the constant rate period the rate of heat supply to the stock must
balance the latent heat of evaporation multiplied by the rate at which water is
evaporated. The latent heat in kilojoules (kJ) per kg of water is a function of
temperature only and this temperature will be the wet-bulb temperature of
the air in contact with the surface. Fig. 12.10 gives this relationship.
Example. At what rate must heat be supplied to evaporate 100 kg per hour
of water from a wet surface at 80°C ?
It does not matter whether this heat is applied via the air or directly to the
stock. There will, however, be additional heat required in practice to cover:
Heat lost through the walls of the drier.
Heat wasted in so far as the whole of the air does not reach saturation.
Heat needed during the rising rate period to raise the temperature of the
stock and its moisture content to 80 °C; this will not have been available for
evaporation.
12.4.2 Evaporation rate

The convective heat transfer rate h. W/m2 °C discussed earlier in this
chapter is directly linked with the evaporation rate. It has already been
pointed out that the steady state, constant-rate, temperature of the wet
surface is equal to the wet-bulb temperature of the air. The temperature
difference available for heat transfer is the difference between this and the
dry-bulb temperature, that is the wet-bulb depression ∆tw - provided the air
is well mixed by turbulent flow. The heat supplied to A m2 of wet surface is
therefore:
This is now equated to the heat required to evaporate M kg/s of vapour of

latent heat L kJ/kg:
It follows that the evaporation rate in kg per sec per m2
Example. A bed of wet sand 2m x 2m in area is dried by an air stream of

4m/s with a temperature of 90°C dry-bulb, 35°C wet-bulb. What is the
maximum evaporation rate to be expected ?
From Fig. 12.1, with L = 2000mm and v = 4m/s ho = hc = 16 W/m2 °C. But
this is for a smooth surface; a substantial multiple should be used for a rough
sand surface - say 2.5. We therefore take hc = 40 W/m2 °C.
Working in the more practical units kg/hr, the evaporation rate from the
expression just quoted is:
12.4.3 Evaporation rate with direct heating

When additional heat is fed directly to the wetted surface by radiation or
conduction, the surface temperature will rise above the wet-bulb temperature
of the air. This will reduce the rate at which heat is drawn from the air, but
will increase the rate of evaporation.
The new rate is dependent on the partial pressure of the water vapour,
which is the independent contribution made by the water molecules to the
total pressure of the air-vapour mixture.
Two values of vapour pressure are significant, and may be obtained from
the relationship with temperature in Fig. 12.11
VPws (Pa) the saturation vapour pressure at the temperature of

the wet surface.
VPDP (Pa) the vapour pressure in the main air stream,
which is derived from its dew point, that is the
temperature at which it would saturate, if
cooled.
The difference measured in Pa between these two vapour pressures is the

motive power which drives the water vapour out into the main stream,
making room for fresh evaporation to maintain saturation at the wet surface.
The situation is analogous to the convective transfer of heat due to the
temperature difference between a hot surface and the temperature of the air
stream. The same constant, hcW/m2 °C, applies, in a new expression for the
evaporation rate, in kg per hour:
Example. What would be the evaporation rate in the last example if the wet
sand surface were independently heated to 50°C in the same air stream ?
From the psychrometric chart, Fig. 12.13, the dew point of air at 90°C
dry-bulb and 35°C wet-bulb is 19°C, giving VPDP = 2200 Pa from Fig. 12.11.
At 50°C VPws = 12,300 Pa from Fig. 12.11.
As before h. will be taken as 40 W/m2 °C - roughness allowance is too

uncertain to justify minor correction for Kt.
12.4.4 Estimation of actual wet surface temperature

The, last calculation assumes a surface temperature of 50°C, but generally
speaking this will not be known in advance. Organised trial and error is needed
as the following treatment indicates.
Suppose the wet sand bed of the previous examples to be supported on a
plate heated by steam at 110°C. The heat transfer rate between the steam and
the wet surface of the sand is estimated at 50 W/m2 °C.
The heat required per square metre to evaporate the quantity of moisture
just calculated for a surface temperature of 50°C is:
at other surface temperatures the evaporation heat rate will be proportional to

(VPws - 2200) as plotted in Fig. 12.12. Also plotted are the heat transfer rates to
the surface from:
the air, which will be zero when the surface is at dry-bulb temperature and
increase by hc = 40 W/m2 for each °C below;
the steam, which will be zero when the surface is at steam temperature,
and increase by 50 W/m2 for each °C below.
Point P where the total heat transfer equals the evaporation heat rate, will be
the probable surface temperature: 47°C.
At P the evaporation heat rate is 4.8 kW per m2 giving a total of 19 kW over

the 4M2. This corresponds to an evaporation rate of:
12.5 Psychrometric Chart for Drying

Many Psychrometric charts contain a great deal of valuable data, but are liable
to confuse those unfamiliar with them just because of their rather intimidating
completeness. Fig. 12.13 has been deliberately drawn to include only those
quantities which are needed for the solution of drying problems.
The key diagram, Fig. 12.14, should be self-explanatory. At each point, A, B, C
or M, in a conveyer drying system the first line or two of text gives the information
available, the last line or two the information obtained from the chart. All
quantities are per kilogram of dry air entering and leaving the drier thus:
H = kJ of total heat per kg of dry air

m = kg of moisture per kg of dry air
they may also be used as flow rates however:
H = kJ/s or kW per kg/s of dry air

m = kg/s evaporation rate per kg/s of dry air
m = kg/hour evaporation rate per kg/hour of dry air
Note that each point defines an air condition at some stage in the drying
process, and that:
The condition point moves horizontally to the right as heat is added to the air
without change of moisture content.
The condition point moves diagonally up one of the constant H lines as
moisture is absorbed by the air with no addition or subtraction of heat. This is
very nearly a line of constant wet-bulb temperature.
The evaporative efficiency E is comparable in value to the heat exchange
effectiveness of a system of similar geometry. It is called an "efficiency" rather
than an "effectiveness" because it really does give the fraction of the heat
supplied to the air which is used in evaporating moisture, the rest being wasted.
At least that would be the situation if the surface temperature of the stock
equalled the wet-bulb temperature of the incoming air; and all the heat were
delivered to the air with none lost. Heat delivered directly to the stock is used at
100% efficiency.
In an air heated drier of economical size the air velocities required to maintain
a good drying rate may be so high that an excessive quantity of air flows through
the drier. This means a low moisture content in the air leaving the drier with a low
evaporative efficiency and consequent loss of heat in the excess air.
This difficulty may be overcome by recirculating a substantial proportion of the

air flowing over the stock back to a point upstream of the heater. The bigger the
ratio R of recirculated air to fresh air the higher the efficiency, though to go too
far would reduce the evaporation rate because the circulating air would be
saturated.
Similar results arise from simple accelerated circulation within a batch oven.
CHAPTER 13
Ventilation of tunnels
and mines
Effective ventilation is an essential requirement for life support

in underground workings of all kinds, both in operation and during
construction. Toxic and inflammable gases must be diluted to a
safe level; dust and smoke must be dispersed; excessive heat
should be relieved; and escape routes from fire or other
emergency must be kept clear of smoke and flame.
Clearly these vital matters must be the responsibility of a
professional consultant in the interests of public safety. Here we
can only touch on the salient points of tunnel and mine ventilation
technology, but the treatment does serve to introduce the jet fan,
together with aspects of system design which are of importance to
the fan engineer.
13.1 Pollutants in Road Tunnels

13.1.1 Carbon monoxide
Both the time of exposure and the concentration are significant,
the progressive effects being loss of alertness, headache and
unconsciousness. In view of the short time the driver is exposed in
normal traffic quite high concentrations are tolerated, but there
should be provision for
282
Ventilation of tunnels and mines 283
a "switch off engines" instruction in the event of a hold up. Typical

recommendations in parts of CO per million by volume are:
250 ppm: a common maximum for extreme traffic conditions.
150 ppm: suitable for normal peak traffic flow.
100 ppm :control level for ventilation rate reduction.
50 ppm: maximum average exposure for full-time tunnel staff.
13.1.2 Other products of car exhausts

Toxic oxides of nitrogen and sulphur are also produced by spark ignition
engines, the recommended maximum 8-hour day exposures being: nitric oxide, 25
ppm NO; nitrogen dioxide, 5 ppm NO2; sulphur dioxide, 5 PPM S02. With present
day engine designs these limits will not be reached until the CO content is as high
as 500 ppm so that ventilation design may be based on carbon monoxide dilution.
Unburnt hydrocarbons in the exhaust are responsible for bad smells and eye
irritation, but once again these should be negligible provided the carbon monoxide
is adequately controlled.
13.1.3 Diesel engine exhaust smoke

Compression ignition engines produce much less carbon monoxide and other
toxic gases, but a successful tunnel ventilation system must limit the unpleasant
smoke and smell characteristic of heavy vehicle exhausts. The standard test is of
visibility which can be monitored by photo-electric devices calibrated in Dmg/m3 -
milligrams of standard diesel smoke particles per cubic metre of air. This measure
is less than the total solid content (mostly unburnt carbon) in the air since large
heavy particles play relatively little part in the obscuring effect of the smoke haze.
90m visibility is satisfactory for 60 km/hr driving speed. 2 and 4 cd/m2 = are
measures of the luminance of the surface seen.
13.1.4 Estimation of pollution load

The total load is clearly equal to the total numbers of cars in the tunnel at any
one time multiplied by the average emission per car. The letter is dependent on
the speed and gradient and, of course, on the car and its condition. Various
national transport authorities have published estimates of average vehicle
performance, those deduced by the ASF from actual trials in Swiss road tunnels
being summarised in Figs. 13.1 and 13.2. These charts are for sea level, and more
pollutant is generated in mountain tunnels where the engine must produce the
same powerudren supplied with air of reduced density. Table 13.1 gives the factor
by which the sea level value should be multiplied for various altitudes.
Traffic statistics will indicate the normal number of cars and lorries to be
dealt with, but on trunk roads there will always be occasions when each lane is
used to its maximum capacity. Many studies have shown that there is a
minimum spacing which the average driver considers safe, and the way in
which this varies with speed is indicated in Fig. 13.2 for
the case where the traffic is wholly private cars of European size (4m long). It is
interesting to note that there is a maximum capacity of the order of 1800 cars per
hour per lane which occurs around 60 km/hour with reduction at both lower and
higher speeds.
Multiplying together the number of cars per km (Fig. 13.2) by the length of the
tunnel the number of traffic lanes and the pollution rate per vehicle gives the total
to be dealt with by the ventilation system. This
reaches a maximum with congested, crawling traffic in the 10 to 30 km/ hour

speed range, and maxima derived from Figs. 13.1 and 13.2 are given in Table
13.2. The importance of avoiding steep gradients in long tunnels is obvious -
particularly at high altitudes. For the Diesel smoke load traffic statistics must
be relied on to indicate the proportion of heavy lorry or coach vehicles and
their average loaded weight.
78E Woods Practical Guide to fan Engineering
13.2 Ventilation of Road Tunnels

13.2.1 Ventilation rate
In a good ventilation system the maximum pollution concentration anywhere
in the tunnel is likely to equal the total pollutant generated, divided by the total
volume of air exhausted. Since 1 m3 = 106cm3 the concentration of carbon
monoxide in parts per million will be:
and of diesel smoke:
To determine the ventilation rate required it will be necessary to set Inrpet

pollution levels for various traffic conditions which might be, for example
250 ppm CO: congested, crawling, flow; all private cars.
100-150 ppm CO: expected maximum flow at 40-60 km/hr; all private cars.
2-3 Dmg/m3: congested flow with maximum forecast proportion of

heavy goods vehicles and buses.
1-171 Dmg/m3: expected maximum tonnage flow of diesel vehicles at

40-60 km/hr.
Each of these targets will set a separate value for the ventilation rate, and the
largest of these will be the m3/s peak capacity to be installed, plus an allowance
for plant out of service for maintenance, etc.
13.2.2 Sectionalising of tunnel length

Ventilation cost is greatly influenced by the section length between access
points at which fresh air may be supplied and polluted air exhausted. Thus, for a
given traffic flow and cross-sectional area of tunnel and ducts:
Pollution rate is proportional to: (length)
Ventilation rate is proportional to: (length)
Pressure gradient is proportional to: (length)2
Pressure drop is proportional to: (length)3
Power required is proportional to: (length)4
Doubling the section length thus multiplies the fan power required by
1111. Most long tunnels have ventilation sections of the order of 1 km long
where they can be conveniently provided, but with tunnels under wide rivers,
straits or high mountains this is difficult and costly. If such
tunnels are very long it is usual to limit the maximum traffic density by control at
entry. Minimum speed and vehicle spacing may also be controlled.
Even with strict capacity limitation large powers may be involved Thus, the 16
km long tunnel now under construction to replace the St. Gotthard Pass between
Switzerland and Italy has one 5.5 km long ventilating section and the installed fan
power will be over 20,000kw. Duty control to relate the ventilation rate to the
pollution level produced by the number of vehicles actually in the tunnel can reduce
the average power to a fraction of the installed capacity - see the example of the
Lion Rock tunnel at Hong Kong described in Section 9.1.5, with an average
measured power about one-eighth of capacity.
13.2.3 Ventilation systems

Fig. 13.3 illustrates a number of ways in which air can be supplied and
exhausted in a road tunnel. To facilitate comparison of air flow and pollution levels
each system, a to f, is applied to a common ventilation problem
Tunnel: length 1 km; area 40m2; two-lane; one-way.
Target pollution level: 100 ppm CO.
Cars in tunnel: 80 at 40 km/hr or 300 stationary.
CO generation: 80 cars at 300Cm3/s or 300 cars at 80cm3/s, either producing

24,OOO cm3/s CO and requiring 240m3/s total air supply and extraction for
100 ppm CO.
The diagrams under the heading "arrangement" indicate the tunnel with traffic
flow from left to right, together, where appropriate, with air supply and exhaust
ducts. These are branched at frequent intervals to supply and exhaust ports
situated at road and roof level respectively in the tunnel walls. Arrows coded
according to volume flow indentify the points where air enters and leaves the
system.
The "tunnel velocity" curves show the longitudinal air velocity averaged over the
40M2 cross-section, the full line applying to the stationary car case. When the cars
are moving they will induce air to flow with them along the tunnel. The broken line
shows the velocity resulting from an increase from 240m3/s to 320m3/s in system
(a) or an induced flow of 160m3/s in the remainder, where there is no net
fan-induced flow.
Finally, the variation of "pollution level" along the tunnel is shown again in full
line for the stationary and broken line for the moving cg cases.
13.2.4 Longitudinal ventilation

Longitudinal ventilation is a system without ducts in which the whale required
volume flow moves through the tunnel at constant velocity from one end to the
other. In tunnels up to 200m or 250m long natural ventilation is often relied on, the
air movement being induced by natural wind and thermal forces supplemented
by traffic drag. These forces may
always become momentarily equal and opposite resulting in ventilation failure,

but if there is much traffic (which must be one-way) and therefore appreciable
pollution the drag forces are likely to prevail.
In tunnels of greater length, or subject to rush-hour congestion, axial jet fans
are used to provide positive longitudinal ventilation. The principle of the jet fan is
discussed in Section 13.6; in practice they are mounted in the space between the
maximum traffic height gauge and the tunnel roof, repeating at intervals of 100
metres or more. They blow in the same direction as the traffic (normally
one-way) though they are capable of reversal in emergency. They are particularly
well suited to cut-and-cover tunnels with flat roofs; a line of axial fans in parallel
across the whole width of the tunnel will occupy minimum headroom and provide
the necessary thrust with minimum power.
The fact that the tunnel itself is used as the airway gives the longitudinal
system notable economy both in first cost and energy consumption. However, the
volume flow and air velocity must increase in proportion to the length. which is
therefore limited to about 1 km for roads with heavy traffic, corresponding to an
average air velocity around 8m/s. As Fig. 13.3 shows, the average pollution level
is low, and the brisk turbulent flow in the tunnel eliminates high local
concentrations of exhaust fumes.
13.2.5 Transverse ventilation

Transverse ventilation is a fully ducted system with a fresh air supply duct and
a polluted air extract duct each running the full length of the tunnel. Closely
spaced inlet jets (e.g. every 5m) connect the supply duct with the tunnel, usually
at roadway level to limit the formation of stagnant pockets with high exhaust gas
content if the traffic is stalled. Similarly spaced outlets connect to the exhaust
duct at ceiling level, so that there is a continuous slow upward drift of fresh air
across the tunnel.
Both supply and exhaust ducts are generally fitted with fans at both ends of
each ventilating section with obvious saving of duct cost and energy consumption
- half the volume flow to be conveyed half the distance. Transverse systems are
likely to be highest in both first and
running cost, though they are virtually essential for very long tunnels, since they
can be readily sectionalised through supply and exhaust access shafts. Some
consulting engineers also consider them best at restricting the spread of a fire,
and extracting the smoke. This may imply supply fans reversible to extract in an
emergency, and capable of continuing to function for 15 to 30 minutes in very hot
gases.
Large, high efficiency, fans are likely to be needed, installed in plant rooms with
the duty and layout generally favouring the axial type. Variable pitch or other
control means are important to save energy by matching ventilating rate to need.
In urban surroundings towers may be needed, both to obtain fresh unpolluted air,
and to dissipate the exhaust gases by chimney action so that they will not
constitute a nuisance to neighbouring buildings.
13.2.8 Semi-transverse ventilation

Semi-transverse ventilation uses the same supply duct as a transverse system,
but the outgoing air travels along the tunnel as in the longitudinal system. If the
traffic is not moving or if it is two-way, the outflow will be equally towards each
end of the tunnel, and in the middle there will be no longitudinal flow. At this
point-indicated by a break in the 100 ppm CO line in Fig. 13.3c - the supply jets
must be relied on to stir up any exhaust gas pockets, which will then drift,
adequately and uniformly diluted, towards the ends of the tunnel.
The corresponding pollution level diagram Fig. 13.3d shows the unsatisfactory
result of operating a semi-transverse system with an exhaust, rather than a
supply, duct. There is again a point of no longitudinal flow, at the centre with
stationary traffic, and displaced towards the exit end as the traffic speeds up. This
time, however, instead of fresh air flowing away from the null point, polluted air
will be flowing towards it, resulting in a very high local level. Furthermore, the
exhaust ports produce no jets so that local mixing is not promoted.
Semi-transverse systems are intermediate in cost and are most frequently found
for tunnels in the 1 to 3 km range. Where the tunnel has a circular cross-section
the supply duct can be conveniently accommodated under the roadway. Mixed
systems are sometimes used with a supply duct along one half of the tunnel
length and an exhaust duct along the other. As Fig. 13.3e shows it is better to
have the supply duct at the exit end if there is one-way traffic. Both
semi-transverse and mixed systems are often sectionalised with reversible fans so
that it is possible to direct smoke and flames from an accident away from traffic
trapped on one side.
13.3 Ventilation of Rail Tunnels

13.3.1 Main line tunnels
Main line tunnels hardly ever require mechanical ventilation. Whereas the drag
of one-way traffic induces air velocities in the 4 to 8m/s range in road tunnels, the
frontal area of such vehicles is usually 5% to 10% of the tunnel cross-section.
Railway rolling stock, on the other hand, runs from 20% to 30% (double track) to
50% to 60% (single track) of
tunnel area; this introduces a "piston effect" which further increases the
propulsive force, providing adequate ventilation by displacing air from entry to
exit portal.
The tunnel should nevertheless be sectionalised by means of intermediate
vent shafts. By limiting the volume of air to be moved these reduce the pressure
required and the corresponding extra tractive effort. It is also important to limit
the sudden rise or fall of pressure, within the train, proportionate to the drag
force, which occurs when the train enters or leaves the tunnel or passes a vent
shaft. A change of 3% in absolute pressure ( ± 3000 Pa) is unpleasant to the
passengers, and may harm the ear of someone whose pressure equalising
mechanism (the Eustachian tube) is partially blocked, or who has some other
special physiological sensitivity.
13.3.2 Underground railways

Underground railways forming part of an urban "rapid transit" system involve
long tunnels, have a far higher density of traffic, and have frequent underground
station platforms. The comfort of the passengers requires better air conditions
on the platforms than are necessary in the tunnels, and excessive air velocities in
the access subways should be avoided. The heat generated underground will
equal practically the whole of the (normally electrical) energy supplied to the
rolling stock and services. Bearing in mind the repeated acceleration and braking
this is very large, and cannot be absorbed by the earth around the tunnel
without building up an excessive temperature.
Consequently a sufficient volume of air must pass through all the underground
workings to carry away the whole of the heat generated with a limited rise of
temperature. The air drawn in will be at the atmospheric temperature on the
surface, so that climate and season have a strong influence on the volume flow
required. Taking the specific heat of air as 1 kJ/kg °C
Fans will be used both to supplement the air movement produced by the
trains, and to control the distribution of supply and exhaust between the
stations, the tunnels, and the plant areas. Extraction at the centres of tunnel
sections is usual, with supply and pressure relief vents near the ends,
controllable so that cooling air can be drawn or blown into the stations in hot
weather, limiting the platform temperature to, say, 3°C above the outside level.
Higher exhaust temperatures may be permissible elsewhere; for example there
may be fan-powered extraction points under station platforms to remove heat
stored in the brakes with exhaust air well above platform temperature.
The fan powers are large so that energy economy calls for a sophisticated
control system. This will operate through damper and fan duty control, secured
by on-off switching, variable-pitch or speed control as discussed in Chapter 9.
Reliable estimates of flow resistance are also
necessary, obtained by the methods of Chapter 6, supplemented by the

treatment of friction drop in long airways to be found later in this chapter.
13.4 Mine Ventilation
13.4.1 Ventilation requirements

The first responsibility of the mine ventilation engineer is to maintain at all
times, including emergencies, the air needed for life-support underground. With
this requirement satisfied he will want to improve the quality of the air in the
workings to limit the stress imposed on the miners by an inevitably hostile
environment. Dust suppression is important because of the danger of respiratory
diseases grouped under the general heading of pneumoconiosis. Finally, some
mines, notably British coal mines, are "fiery", introducing the need to limit the
concentration of gases such as methane below the flammable or explosive level.
In most countries these requirements are specified as precise statutory
obligations.
In deep mine workings heat and humidity are the principal enemies of good
working conditions. Even in dry strata humidity is high when water is used to lay
the dust at the working face. Heat will arise from the energy supplied to the
mining machinery and also from two natural sources:
Heat generated in the earth's core by radioactive processes flows to all parts
of the surface where it is dissipated by radiation. There is, therefore, a
geothermic temperature gradient, determined by geological factors, which
makes the deep rock temperature higher than the ground temperature at the
surface. Three examples are given in Fig. 13.5; the British coal-bearing strata
have a particularly low thermal conductivity (unless they are water-bearing)
leading to the high geothermic gradient shown.
As air flows down a mine shaft it is gradually compressed by the increasing
weight of atmosphere above it. As we saw in Section 11.3.10 compression heats
the air, and the adiabatic temperature gradient isolates this effect by ignoring
any other heat given up or acquired by the air during its descent. The broken
line in Fig. 13.5 shows this gradient, starting from a typical UK surface
temperature. When the air returns to the surface its adiabatic temperature will
follow this line in reverse, there being in fact an average adiabatic temperature
appropriate to each depth below sea level.
When the rock underground is wet, or when water is used for dust
suppression, there will be evaporation which will substantially lower the dry-bulb
temperature of the air. This will not improve the working conconditions much,
however, the heat stress being primarily dependent on the wet-bulb
temperature. In Europe 28°C WB is regarded as the maximum for effective
working, but in the deep mines of South Africa selected and acclimatised miners
can operate at over 30°C WB. Air movement at 1 to 2m/s is helpful if the
dry-bulb temperature is not too high, as shown in Section 1.8.
As an example of heat balance an investigation may be quoted from the

Nottinghamshire coalfield, U.K. The mine was 850m deep with workings
extending 5 km from the main shafts and the air extracted was 150m3/s. Heat
flow rates are averaged over a 24 hour period.
This was balanced by the following heat absorbed by means of the ventilation
system:
13.4.2 Main mine fans

The main fans are large machines, ranging from a few hundred to as much as
5000 kW. They run for 24 hours a day so that efficiency is important - 85% or so
would be expected. Both axial and backwardcurved centrifugal types come into
consideration with lower pressures favouring the former and higher pressures the
latter. Emergency reversibility is often called for to deal with an underground fire;
reversal of rotation suffices for an axial fan, whereas a centrifugal requires the
installation of by-pass ducts and dampers. Periodic adjustment of fan volume flow
and pressure as the workings are developed and expanded is another
requirement, usually met by pitch-angle adjustment, inlet vane angle adjustment,
or speed change.
Main mine fans are nearly always installed on the surface, for reasons of cost
and ease of access for maintenance and control. The majority are used as
exhaust fans at the top of an upcast shaft so that the whole mine is under
suction. This is safer in a gassy mine since ventilation failure would cause a rise in
underground pressure, inhibiting rather than encouraging the outflow of gas from
the rock face. The shaft for access and mineral winding, for which the fresh air
side is preferred, may then provide the air intake, and need not be air-locked as is
necessary if the supply fan is at the top of the downcast shaft.
13.4.3 Booster fans

The initial plan of a mine embodies a network of intake airways leading from
the downcast shaft to the working faces and return airways leading to the upcast
shaft. The intake airways commonly serve as haulage roadways as well. As the
mine develops, following the coal seams or other strata to be worked new loops
must be added. When there is insufficient pressure to circulate the required air
volume an underground booster fan is added, handling the whole volume in the
loop and topping up the pressure as required. These are normally axial fans, of a
few hundred kW rating, which are naturally "in line" machines and will develop
the necessary pressure in a single stage. When the airway is also a roadway the
booster fan must be in a by-pass passage with an air-lock in the roadway to
prevent short-circuiting.
13.4.4 Auxiliary fans

Working faces may be advanced a considerable distance without a return
airway. A duct of 400 to 800mm diameter is used to supply fresh air to the face,
which then returns along the roadway/airway itself. The fan, invariably of the
axial type, is fitted at the inlet to this duct, adjacent to the fresh air main
roadway. It is portable, from 5 to 50 kW rating, and is slung from the roof of the
roadway. The ducts are also slung from the roof, and must be readily extensible
as the work progresses. When the length is great (100 to 1000m) every
precaution must be taken against leakage and excessive friction, otherwise the
fan power will become prohibitive. A flexible PVC design, with a helical steel
reinforcing wire attached externally, is good in both these respects.
Contra-rotating axial fans of the same diameter as the duct and with each
impeller mounted on the shaft of its own 2-pole electric motor, are
particularly suited to auxiliary fan duty. They may be stocked in 2-stage units, and
built up to 4- or 6-stages as required by the growing length of the airway. In
coal-mining all stages are installed at inlet, building up to a pressure of several
thousand Pa. In some circumstances, however, it may be preferable to string
them at intervals along the duct, thereby limiting the pressure rise above the
airway - though this should never be allowed to become negative, with possible
ingress of contaminated air.
13.4.5 Tunnel drivage fans

A ventilating system similar to the auxiliary mines fan and duct is used during
the construction of road and rail tunnels, and in the initial driving of mine shafts
and roadways. As well as providing heat removal and ventilation for the workers,
the air stream must serve to disperse the dust and fumes arising from the
explosive charges used to split the rock.
Increasing the volume flow will shorten the time lost in shot-firing and may
enable longer shifts to be worked. Axial fans with high tip speeds, e.g. 800mm
diameter at 59 rev/s (3540 rev/min) may therefore be used. Even in the
conditions of a construction site it is advisable to fit such fans with silencers,
which can take the simple form of a metre or two of acoustically lined duct.
Silencers are also increasingly employed for much less noisy booster and auxiliary
fans in mines; mine roadways can be very reverberant, and the value of good
working conditions is becoming more widely recognised.
13.5 Estimation of Pressure Requirements
13.5.1 Items of pressure loss

The high volume flows and substantial power requirements of tunnel ventilation
make it important that each element in the airway system should be designed for
minimum pressure drop compatible with reasonable first cost. The fan duty will be
determined by the summation of items from a list such as the following:
Inlet from atmosphere: loss can be made small by good entry design.
Outlet to atmosphere: loss = ½ρ (outlet velocity m/s)2 - Pa.
Bends, diffusers, screens and louvres: see Chapter 6.
Acoustic treatment and silencers: see proprietary data.
Intake and discharge registers and dampers: a good part of the pressure drop will
arise from the restrictions necessary to equalise flow.
Duct and roadway friction: see Sections 13.5.2. and 13.5.4.
Traffic drag and obstructions: see Section 13.5.5.
Influence of duct leakage: see Section 13.5.6.
Wind pressure: tunnel portals in approach cuttings on level ground will be

insensitive to wind pressure provided they are clear of the pressure
fields surrounding large buildings. In hilly country the wind can have a substantial
influence on the longitudinal flow velocity in a tunnel. Some Japanese tests on a
tunnel piercing a mountain range showed a pressure difference as high as 1.6
times the velocity pressure corresponding to the wind speed quoted in
meteorological reports.
Thermal pressure difference: if the average temperature of the air in a tunnel is

∆t °C above that outside, and if there is a difference of altitude h metres between
the tunnel portals, then there will be an upwardly directed pressure difference of:
where t1 (°C) and ρ1 (kg/m3) are the temperature and density at the lower portal
and g = 9.81 m/s2.
Barometric pressure difference: differences due to changes in altitude along the

tunnel are self-cancelling. Barometric pressure gradients represented by the
closeness of the isobars on a meteorological chart are not significant-hurricane
conditions apart.
13.5.2 Friction drop in long airways

Fig. 6.17 gave the pressure gradient in a duct of normal metal construction
with sufficient accuracy for most purposes in air conditioning and process work.
Airways in tunnels and mines are generally formed in situ by civil engineering
methods. Furthermore, they absorb the greater part of the overall pressure
required, so that the estimation should be as realistic as possible.
Fig. 13.6 is a version of the well-known Moody chart summarising the

researches and formulations of Prandtl, van Karman, Nikuradse and Colebrook. f
is the friction factor relating the pressure drop, ∆p Pa, to the velocity pressure,
½ρv2 Pa, corresponding to the average velocity in the airway, v m/s, and the air
density, ρ kg/m3, and to the length, Lm, and diameter, Dm, of the airway if
cylindrical.
If the airway is not cylindrical, D is replaced by the hydraulic diameter Dh

defined by:
It will be seen that f varies over quite a range according to the values of the
two quantities:
Re is a dimensionless number representing the ratio of the inertial to the

viscous forces. The shear forces in the boundary layers near the walls are
viscous, and cause the drag of the air on the airway walls which results in the
pressure drop. With air (or any gas) of viscosity µPas.
We cannot expect to get closer than f 10% or so in our estimation of f, and

consequently a rough approximation to Re is quite sufficient. When the fluid
flowing is atmospheric air it is good enough to take ρ as 1.2 kg/m3 except
perhaps at high altitude, and µ as 18 x 10-6 Pa s.
Then:
This approximation is used to construct the scale at the top of Fig. 13.6, so
that the chart may be entered directly with the product of D or Dh in metres and
v in m/s.
ε is the equivalent roughness of the airway walls here measured in
micrometres, µm (1000µm = 1 mm). It is not necessarily the same as the
surface roughness of the material used, because account is taken of imperfectly
fitted joints and similar factors representative of normal installation practice. Fig.
13.6 shows that, for any given value of rough-
ness ε, and diameter, D metres, the friction factor falls from a value close to that
for a completely smooth duct (ε = 0), through a transitional curve to a constant
value as v and Re increase.
Values of ε quoted by various authorities as representative of ducts
made from particular materials are quoted in Table 13.3. Also quoted are ranges
corresponding to experimental values of pressure drop reported for a number of
actual road tunnels, mainly in the European Alpine area.
The formulae and the chart are quite general and apply to any duct, pipe or
airway of any (constant) cross-section. Also to any gas or liquid provided the
appropriate values of ρ and µ (see Chapter 14) are used. No practical airflow
absorbing enough power to be worth consideration will fall into the uncertain
area marked critical flow but viscous liquid flows may. At some Re value between
3500 and 2000 the f value for such a flow will fall to the laminar flow line where
turbulence is entirely suppressed and f = 64/Re.
13.5.3 Example of airway friction calculation
A road tunnel 1 km long of the crosssection

illustrated is to carry 300m3/s longitudinal
ventilation. What will the system pressure drop
be?
Outlet loss = ½ρv2 = ½ x 1.2 kg/m3 x (7.4m/S)2 = 33 Pa
System loss (ignoring wind pressure) = 72 + 33 = 105 Pa
That is K = 32 in ∆p = K. ½ρv2
13.5.4 Airways with distributed intake or discharge

In the transverse and semi-transverse ventilation systems the ducts will be
connected to the roadway through a series of ports which are usually equally
spaced and adjusted to admit or extract equal volume flows of air. Both duct and
roadway will thus experience a gradual rise or fall in volume flow and therefore in
velocity and pressure gradient.
If the airway is of constant cross-section throughout it can be shown that the
pressure required to overcome the airway friction will be reduced to approximately
one-third of the pressure that would be required if the whole volume flow travelled
the full length of the airway:
Supply and exhaust ducts are sometimes stepped down in size as they
progress from the fan end to the far end. In any one section with mean velocity
v, at the end nearer the fan and v2 at the other end the pressure loss will be
approximately:
In addition to duct friction there is bound to be some "mixing" loss as the air
passing through each port joins or is extracted from the main stream. This
subject was discussed in Chapter 6, and detail design is important to minimise
loss of energy at these points. Loss of energy is inevitable at the dampers
needed in each port to maintain equality of flow in spite of the greater pressure
difference available at some of them.
Fig. 13.7 illustrates a basic difference in this respect between the supply and
exhaust sides. While the distributions of total pressure will bathe same if mixing
losses are neglected, the static pressure changes far more along the exhaust
duct and relatively large damper losses will be incurred towards the fan end on
this side.
13.5.5 Traffic drag

Any vehicle travelling with a velocity v m/s relative to the air experiences a
drag force - F Newtons. An equal and opposite propulsive force + F is applied to
the air. If the vehicle is in a tunnel of crosssectional area Atm2, the propulsive
force will cause a rise in pressure (F/At) Pa tending to move the air in the same
direction as the vehicle. Conversely, if the vehicle is stationary and the air
flowing along the tunnel, additional pressure (F/At) Pa will be needed to maintain
the air movement against the vehicle drag.
If the vehicle is travelling with velocity a and the air with velocity v in the same
direction their relative velocity is (u - v) m/s. The force F is then proportional to the
relative velocity pressure ½ρ (u - v)2, to the frontal area of the vehicle Av and to a
drag coefficient Cv.
A typical value of Av for a private car is 2.3m2. Cars in convoy benefit from the
"slipstream" effect, and suggested typical values for Cv per car with current models
are:
Vehicles which are large compared with the tunnel add to the drag by "piston
effect'. One model test of double deck buses side by side in a two-lane tunnel
indicated a Cv in the order of 5.
The total drag of a line of stationary vehicles can thus be estimated, and the
corresponding pressure added to that required to maintain longitudinal flow. If they
are moving faster than the air they will add to the other pressures supplied by jet
fans or a semi-transverse system. Trial and error calculations are then needed to
find the tunnel air velocity which will result in balance between the propulsive and
the drag forces.
An example of the results of such a calculation is given in Fig. 13.8 which relates
to a system of natural longitudinal ventilation without forced draught or external
wind or other forces. Two-lane, one-way
traffic is assumed with the maximum traffic density obtained according to

speed from Fig. 13.2. If the pollution data of Section 13.1 for a level roadway
at sea level are applied to this situation it will be found that the CO levels are
satisfactory for speeds above 30 km/hr in a 1000m long tunnel or above 20
km/hr in a 500m long tunnel. At lower traffic densities the situation would be
better. Jet fans or other emergency provision would be necessary, however,
unless the possibility of congested, crawling, traffic can be rigorously excluded.
13.6 Leakage from Long Ducts

When air has to be delivered through ductwork to a distant location great
precautions must be taken to minimise leakage. Even when every care is taken
a surprisingly large proportion of the air supplied fails to arrive at the far end
and some estimate of the loss needs to be made. Ducted air supply during
tunnel construction is one example, and auxiliary mines fan systems another.
Leakage can be measured simply by sealing both ends of a test length and
maintaining pressure through a supply pipe containing a flow measurement
orifice. Tests on ducts in service in mines and tunnel construction show the
effects of installation strains and service damage. Both bolted joints and welded
seams are liable to be less airtight in
service than under laboratory conditions. The following table broadly corresponds
to the recorded experience of the British National Coal Board and other
observers.
13.6.1 Leakage calculations

If a test duct of diameter D and length L is under uniform pressure ∆p, λ can
be found from the measured leakage flow Qλ taken as:
Consider a long duct coupled to a supply fan at the inlet end and intended to
deliver Q2m3/s of fresh air at the far end. If the leakage were zero, the
appropriate friction factor f, and the duct cross-section A m2, there would be a
positive static pressure pf at the fan end equal to:
A greater volume flow Q, will be required at the fan end if there is leakage (Q1
– Q2), in order that Q2 may be maintained. The higher velocities towards the fan
end will increase the pressure drop; requiring a greater pressure, p1.
Mathematical analysis* leads to the practical data of Fig. 13.9; some
approximation is involved and f is taken as 0.015 throughout, but precision is out
of place when only a crude assessment of the leakage coefficient can be made.
13.8.2 Example
An 800mm diameter duct is required to deliver 5m3/s over a distance of 2
km. Welded seam duct lengths are to be used, properly gasketted and
supported at each joint to minimise strain. What fan duty should be specified
?
If extra care in installation secured an initial value of λ = 0.15, Q2/Q1 and pf/p1
would become 0.84. The corresponding fan requirement for Q2 = 5m3/s would be
5/0.84 = 6.0m3/s 2100/0.84 = 2500 Pa 15kW
The fan specified will exceed this duty, moving to a slightly different operating
point on its characteristic. However, the air power is likely to remain fairly constant
at 28 kW, so to estimate new values of Q1 and
Q2 multiply by
If the duct installation were allowed to deteriorate to the poor condition λ = 1.5,
Q2/Q1 = pf/p1 = 0.33:
Clearly there are practical limits to the length over which air may be conveyed
by a single fan. These limits can be extended by employing a number of fans in
series, spaced along the duct length. In fact, if the airway is divided into a number
of sections of equal length, with a fan at the centre of each section the volume
flow discharged will equal that at entry however long the airway.
This is rarely a practical solution, however. Equal leakage volumes will flow,
outwards from the downstream, positive pressure, half of each section and
inwards to the upstream, negative pressure, half. This local recirculation will be
liable to contaminate the fresh air supply in the duct, if that is its purpose. If, on
the other hand, the system is used to extract contaminated air, then the outward
leakage will constitute an escape of contamination.
A more satisfactory method is to locate the fans at the beginning of each
section, perhaps with an additional stage in the first one, to ensure a positive
pressure everywhere in a supply duct, or negative pressure in an extraction duct.
There will then be leakage, but the very high pressure required from a single fan,
and the corresponding excessive leakage, will be avoided. Axial fans are best
suited to such a system, but it is important that they should be guide vane types,
or contra-rotating pairs. This is to avoid excessive outlet swirl, which is maintained
for long distances in a cylindrical duct. Swirl at the inlet of a later fan stage would
reduce its performance, and swirl in the airway increases the pressure drop.
13.7 Jet Fans

13.7.1 Use in tunnel ventilation
Jet fans may provide the whole propulsive force in a longitudinal ventilation
scheme, or they may supplement a semi-transverse or transverse system in
sensitive locations, e.g. at the tunnel entrance or exit. The fan is an axial machine
with open inlet and outlet, designed for
maximum efficiency in this condition rather than when operating against

pressure. The rating is commonly in terms of the thrust applied to the air - with
an equal and opposite reaction force borne by the fan supports. It is mounted
above the traffic suspended from the tunnel roof, but must not be too close
(less than 1 to 1.5 fan diameters) or there will be loss of thrust.
The basic thrust rating is equal to the momentum flux at fan outlet, that is
the product of the mass flow and average velocity:
Acting on the tunnel area, A1 m2, this would produce a pressure rise:
The total pressure rise due to a number of fans, n, is simply the sum of the
individual pressure rises. The fans may be located in parallel groups, but should
be spaced 10 or more tunnel diameters apart lengthwise to ensure complete
mixing of the jet from each.
These formulae are only exact if the mean tunnel velocity is very small
compared with the fan outlet velocity. As much energy can be saved by
employing a lower fan velocity and larger volume flow for the same thrust a
more thorough treatment is justified.
13.7.2 Principles of jet action

Fig. 13.10 has been drawn (exaggerated for clarity) to assist in
understanding the operation. The important factors are the average air
velocities, V1 in the tunnel, V2 at the fan outlet, and V3 in that section of the
tunnel which is by-passed by the flow through the fan. V3 and the
corresponding velocity pressure pv3 will be less than V1 and pv1. As there is very
little energy loss where flow is diverted into the fan, the total pressure in the
by-passed tunnel section will remain virtually unchanged, which means that the
static pressure here must rise-as shown by the heavy lines in the diagram. This
static rise, ∆ps is the first contribution to the effective performance.
The broken line shows the course of the total pressure through the fan and
on through the outlet jet until it rejoins the tunnel total pressure line when the
jet is completely mixed. The fan total pressure, Pt is the total pressure rise after
the impeller, but since the static pressure in the casing must equalise at outlet
with the static pressure in the tunnel by-pass section, it will be seen that the
fan velocity pressure, Pv2, exceeds the
fan total pressure. The difference constitutes a negative fan static pressure,
which will be:
The high outlet velocity of the fan creates a local excess of momentum flux in
the jet. In dropping back to the normal tunnel value as the jet slows down this will
exert a forward force on the air ("force = rate of change of momentum"). The
reaction to this force is provided by a rise in tunnel static pressure, Apj, which is
the second contribution to the
effective performance. The drop in the broken total pressure line represents the
loss of fan outlet energy which takes place in the turbulent decay of the outlet
jet.
The static and total pressure lines have been drawn at a slope which
corresponds to the pressure gradient required to keep the air moving against
tunnel wall friction. The effective pressure changes are measured from the
downstream projections of these gradients, and it will be seen that the effective
boost in total pressure due to the fan:
13.7.3 Calculation and design

The formulae that are needed for the full assessment of a tunnel ventilation
system can be derived from these principles in terms of the quantities:
The jet action efficiency ηj is the ratio of the work done against tunnel friction to
the air power output of the fan Q2Pt.
The overall efficiency will depend on the fan efficiency ηt, also:
These formulae involve residual errors of the order of nP A2/A1 per unit where np
is the number of fans in a parallel group.
The resistance ∆p, to the required volume flow Q1 through the tunnel will be
estimated as the sum of the items listed in 13.5.1. It can be expressed in terms
of the velocity V1= Q1 /A1 as:
The number of fans, n, and total pressure rise per fan, ∆pt must be chosen
so that:
a and ηj are plotted against V2/V1 in Fig. 13.11 showing how the jet action
efficiency can be increased by reducing V2/V1 which, in turn, requires a higher
value of a and hence of nQ2 for a given ventilation load expressed by KQ1.
The designer must, therefore, judge how far to reduce running costs (higher
ηj)at the expense of increase in capital cost (more and larger, lower speed, jet
fans). In any case a substantial reserve jet thrust is desirable to deal with
possible excess in uncertain factors such as wind pressure and traffic drag.
Two-speed operation (e.g. 2/4 pole motors) is worth careful consideration. If
the low speed suffices for normal operation the low value of V2/V1 increases ηj
and saves running cost. For heavy traffic some of the fans can be switched to
top speed, and in emergency more than four times normal thrust is available
when all fans are at full speed.
The selection of jet fans for a given tunnel duty is covered by Fig. 13.11. The
inverse problem can be solved with the aid of the linear chart, Fig. 13.12. This
enables the volume flow generated in the tunnel by a given installation of jet
fans to be determined. The method is best explained by an example.
The following table compares this performance with that of twelve 800mm,
1450-970 rev/min two-speed fans spaced 150m in pairs, calculated in a similar
manner.
Clearly the twelve 800mm fans have the better performance, though their
cost is likely to be about 25% greater - offset by savings in wiring and
installation cost. The low speed performance should meet the stationary
traffic requirement; when in motion the drag would change to a propulsive
force, and Q2 would increase substantially. The power saving is 12% at high
speed, 75% at low, and the low speed fans would be 10 dB quieter.
The margin to deal with the drag of large vehicles and with adverse winds
might be considered inadequate. An installation of eighteen 800mm fans
would produce 150 Pa - a margin of 100 Pa over that required for 150m3/s on
the above basis of calculation.
CHAPTER 14
Useful data
Useful data 313
The factors defined by deca- hecto- deci- and centi- are best
confined to well-established usages such as centimetre or hectare.
The original European definitions of a billion (1012) and a trillion
(1018) appear to be dying out in favour of the current American
usage with a billion equal to 109 and a trillion equal to 1012. A
milliard is unambiguously equal to 109.
The decibel scale

When two powers (measured in W, µW, pW, etc.) differ by x
decibels the ratio of the larger power to the smaller is given by the
following table. The approximate value is sufficient for most of the
purposes for which decibels are used.
Normal statistical distribution

The results of research are often reported with a statement of
the remaining uncertainty in statistical terms. The arithmetic
mean value M of the quantity determined is quoted together with
the standard deviation S. Given the normal distribution of random
errors in the measurements there will then be a y% probability
that the true value lies within the range M - xS to M+ xS.
Engineering tolerances are commonly set at ± 2S (i.e. ± 200

S/M per cent) and are classed as having 95% probability.
Useful data 315
Useful data 317
Useful data 319
Useful data 321
Useful data 323
Useful data 325
Useful data 327
The wet-bulb depressions are for sling-type or other

hygrometers in which the air is in motion over the
thermometer bulbs.
tdb and twb are the dry-bulb and wet-bulb temperatures in

°C psdb and pswb are the saturation vapour pressures in Pa at
tdb and twb respectively. po is the barometric pressure in Pa.
The table is correct for a barometric pressure of about 105
Pa.
Useful data 329
Useful data 331
Intemperate climates the density of atmospheric air maybe

determined from measurements of barometric pressure and dry-bulb
temperature alone, using Table 14.10, provided errors up to ±0.5%
are permissible. The table is calculated for a representative moisture
content of 0.75% by weight except below 10°C where the saturation
content is less than 0.75% and saturation density is used for the
tabulation.
In hot, humid climates (and in any case where greater accuracy is
required) an additional measurement of wet-bulb temperature should
be made, using a sling or other type of forced-convection
thermometer. The correction derived from Fig. 14.1 will then be
applied to the value obtained from Table 14.10, and the result will be
within ± 0.1% of the true value over the tabulated range of pressure.
Example. The measured air condition is 29°C dry-bulb, 986 mb.
From Table 14.10 the density is 1.125 at 98 kPa, 1.137 at 99 kPa.
Density at 98.6 kPa = 1.125 + 0.6 (1.137 – 1.125) = 1.132 ±0.006
kg/m3.
Suppose an additional measurement gives a wet-bulb temperature
of 21 °C. The point marked X on Fig. 14.1 corresponds to 29°C DB,
21 °C WB and gives a correction of – 0.003 kg/m3. The revised air
density is:
Useful data 333
Useful data 335
The VG viscosity grades specified in BS 4231 :1967 and ISO

3448: 1976 are equivalent to the corresponding American ASTM
grades (Saybolt universal seconds) and are particularly suitable for
the classification (with a ± 10% tolerance) of lubricating oils.
The VG grade of a liquid is now established from a centistokes
measurement at 40°C. Formerly the temperature was 100°F
(37.8°C) which is still the basis of the ASTM. Saybolt and Redwood
scales; the difference is only a few per cent. The equivalences in
the table are substantially true at either temperature.
The SAE grades of crankcase and gearbox oils are not directly
comparable; being determined at 0 °F and 210°F.
Useful data 337
Values are for a pressure of one atmosphere except in the case of

saturated steam where the pressure is that of saturated vapour
over water at the temperature quoted (air being absent). See Fig.
12.11.
The full lines for VG grade oils are at a viscosity index (VI) of 50.
The broken lines are for 0 VI (steeper slope) and 95 VI (shallower
slope).
The temperature gradient for most pure gases is similar to that
for atmospheric air.
Table 14.16 Climatic data
The temperatures quoted are those generally accepted for the

design of heating and air conditioning systems. They are not the
recorded extremes and are quite likely to be exceeded on a few
occasions in an average year. The data are mainly extracted from the
IHVE Guide, Book A, which should be consulted for comprehensive
information.
The altitude is that of the meteorological station. Sea-level tempera-
tures in the same area would be expected to be higher by about 1 °C
per 165m difference in altitude. The column headed "% Moist." is the
percentage moisture by weight at the design condition.
Useful data 339
Useful data 341
Useful data 343
Useful data 345
Useful data 347
Useful data 349
Useful data 351
Useful data 353
354 Woods Practical Guide! to Fan
Useful data 355
356 Woods Practical Guide to fan
Useful data 357
The ratings in Table 14.26 give the power supplied by the driving
motor. The power transmitted to the load is commonly taken as
95% of the power supplied. The ratings are for Standard Belts to
BS 1440 and BS 3548 in V-V speed reduction drives. For speed-up
drives take the factor K1 for reversing motors. Premium grade belts
are available to the same dimensions with 30% to 40% higher
rating. Special proprietary belts to different dimensions can
transmit two or more times the power in the same space.
The kW quoted in Table 14.26 must be multiplied by

each of the factors below to obtain the kW rating, W, of
the complete drive with N belts in parallel. Factor K1 is
given below and K2, K3 and K4 on the next page.
Useful data 359
Useful data 361
Uselul data 363
Useful data 365
Bibliography
There are many books on the topics covered by this Practical Guide, most of
them written to meet the needs of the examination student. A few works of
general reference are listed here which themselves contain extensive
bibliographies of books and papers for more detailed study.
I.H.V.E. Guide. The Institution of Heating and Ventilating Engineers, London.
Revised every few years.
Book A. Design Data.
Book B. Installation and Equipment Data.
Book C. Reference Data.
A.S.H.R.A.E. Guide and Data Book. The American Society of Heating,
Refrigerating and Air-Conditioning Engineers, New York. Revised every few
years. Three volumes.
Handbook of Fundamentals.
Systems and Equipment.
Applications.
Air-Conditioning and Ventilation of Buildings. Croome-Galeand Roberts.

Pergamon Press, Oxford, 1975.
Fundamentals of Industrial Ventilation. Baturin.

Heat Transmission. McAdams.

McGraw-Hill, Toronto, 1951.
Formulas for Stress and Strain. Roark.

McGraw-Hill, Toronto, 1965.
Fans. Eck. English translation.

Fans. Osborne.
Index
Figures in italics denote illustrations.
Abrasion resistance, 136 Air flow past an obstacle, 77

Absolute pressure, 76, 80 Air flow in straight ducts for
Absolute filters, 70 ventilating systems, resistance
Acceleration, 214 to, 364
Accuracy, 256-258 Air supply, outdoor, for air conditioned
of individual measurements, 257 spaces, 19
Acetone, 54 by jets, 33
used as a solvent, 54 Air velocities over totally enclosed
AC motors, 142, 151 air stream cooled motors, 162
Activated carbon, 70 Airways, standardised. 227
Adiabatic temperature gradient, 292 distributed intake or discharge
Adjustable pitch axial fan, 113 with, 300
Adjustable pitch blades, 112 friction drop in, 296
Aerodynamic action, 110 Aluminium powders, explosions by, 57
losses. 209 Alternating stress, 224, 225
thrust, 163 in axial fan blading, 226
Aerodynamic sound generation Archimedean number, 36
mechanism of, 211 Areas of circles, 342-344
Aerofoil blading, 126 Auto-transformer starter, 157
vanes, 94 Axial fan, with adjustable pitch blades
Air, The, 1 general-purpose, 112
composition of, 1 with by-pass control, 184
bacteria in, 3 Axial fan blading, alternating
impurities in, 3 stress in, 226
gases in, 2 Axial fans, 63, 110, 113, 115, 120.
heat capacity of, 25 132, 137, 161, 192
odours in, 3 for ducted air supply during
specific heat of, 25, 329 tunnel construction, 118
Air, atmospheric, density of, 330 elementary theory of, 137
density corrections for high pressure, 116
humidity of, 331 low pressure. 119
properties of, 329 small, 119
Air cleaning, 63 specialised designs of, 116
Air cleaners, performance of, 66 with fractional solidity, family of, 117
classification of, 67 Axial flow, 137
Air conditions with altitude,
variations of, 323 Backward-curved impellers, 125-127
Air density, calculation of, 246 132, 192, 207
Air distribution system, 29 Balance quality grading, 217
examples of, 30 Balance quality, tentative classification
good and bad, with wall fans, 30 of, 219
367
Balancing. static, 218 Circulating fans 36, 106

examples of 220, 221 velocity profiles of, 37
Ball bearings, continuous-running, 169 Climate, hot, relief in, 36
oil viscosity at operating Climatic data, 338-341
temperature for, 170 Comfort conditions, 7, 27, 32, 36, 41, 43
Barometric pressure difference, 296 factory acts, British, relating to, 9
Bearing capacity, example of, 164 lavatories, in, 33
Bearing loading, 162 optimum temperatures for, 9
example of, 165 small rooms and offices in, 7
Bearing lubrication, 167 Commutator motors, 151-152
Bearing ratings, 166 for AC supply, 152
Bearing system, forces acting on, 163 Complete fan characteristics, 108, 109
Bearings, ball and roller: load Compressibility factor Kp, 250
capacity, 350
combined speed and life factors of, 357
Compressors, 106
combined thrust and journal loading, Constant minimum pressure, 184
factors for, 351 Constant system resistance,
Beaufort scale of wind, 323 control methods at, 193, 198
Bell mouth inlet ring, 122 Constant volume flow, 183
control, example of, 197, 198
Bends, influence of one on another, 97 Continuous speed control, 178-180
interaction between, 87, 97 passim
right-angle, in ducts of circular basic power input with, 180
section, 95 mption with, 179
right-angle, in rectangular ducts 95 power consuContraction, sudden, 90
Body heat balanBest efficiency point, 107 Contra-rotating fans, 115
ce, 3
Bolts and screws, metric, preferred Control at constant resistance, 192
sizes of, 348 198 passim
Branched systems, duct sizing in, 100 Control saving, 175, 176
Branches 87 Controllable pitch fans, 112
Brownian movement, 66, 70 Convection, 4, 260
By-pass control, performance with, 182 forced, 4, 260-262
natural, 4, 260-262, 263
heat transfer by, 260
Canopies, 71, 72 relative influence of. 41
Carbon monoxide, 282 Conversions, to and from
Ceiling diffusers, 35 SI units. 315-321
Ceiling fans, 37, 106 Fahrenheit to Celsius, 324, 325
velocity profiles of, 37 inches to millimetres, 322
Centrifugal fans, 123 127-129, Cooling systems, 50
132, 138 Corrosion resistance, 135
elementary theory of, 138 Couple unbalance, 216-218
in line. 128 Critical speeds, 222
multi-stage, 129 calculation of, 224
specialised designs in, 127-129 Cross flow fan, 111, 134
Centrifugal force, 110 Cyclone, 67, 68, 136
Characteristic curves, 106
Chopped-wave system, 152 Damper control, performance with, 182
Circles, circumferences and areas of fan duty, 180
of, 342-344 forward-curved fan with series 183
Circular ducts pressure gradient in, 98 DC motors, 151
Index 369
Decibel reduction, 210 Dust, classification of, 64

Decibel scale, the, 313 clouds, 57
Delta connection, 154 collection, 71
AC supply for, 152 hazards, 56
speed control of. 152 particles and filters, classification
Deviation angle, 141 of, 65
Diaphragm mounted propeller fan, 121 quarrying in, 56
Diffusers, 87, 114 size distribution of, 67
open outlet, 92 spherical particles in standard
outlet, 87 air, 65
truncated, 92 102 Dust transport, 74
Diffusion, 80, 111 Duty control, by fans in parallel, 173
Dilution ventilation, for toxic hazards, 54 air velocities for, 74
for explosive gases, 58 by five contra-rotating fans
in series, 174
Direct blade action, 110 examples of energy consumption
Direct current motors, 151 with, 197
Direct-on-line starters, 157
Displacement, 214 Eccentricity for ISO balance quality
Domestic vacuum cleaner, 129 grades, permissible, 219
Downcast shaft, 294 Eddy-current couplings, 179
Effective temperature, 10, 11, 12
Draughts, 9
Electric cables, with copper conductors,
Dry-bulb temperature, 6, 8, 11, 12, current ratings for, 354
50, 328 with aluminium conductors,
Dryers, batch, conveyor, through flow, current ratings for, 355
rotary and spray, 273 Electric motors, approximate full load
Dry filters, 68 currents of. 352
Drying calculations, 274-280 Electrical apparatus, hazard
Drying, industrial, 270-274 classifications for, 62
sheds, 272 Electrostatic precipitators, 69
psychrometric chart for, 278, Emissivity, 263, 265
279, 280 Energy saving, 175, 185, 196
Ducted system, air velocities in comparative, 192
typical, 84 Enthalpy, 50
designs of, 83 Equivalent diameters, 89
Duct sizing in branched systems 100 Evaporation, 5, 272
Duct system elements, losses in, 89 heat required for, 274
Ducts, circular, pressure gradient in, 98 latent heat of. 50
flow lines into, 70 rate of, with direct heating, 275
long, leakage from, 302 Evaporative cooling, 50, 52
pressures at, 78, 81 regulation of, 5
preferred sizes and cross-sectional Exhaust points and inlets, 30, 31
areas of, 345
rectangular, equivalent diameter of, 99 Exhaust smoke, diesel engine, 283
circular equivalents of, 360 Exhaust systems, 31, 72
rough-walled, friction Exhausters, 106
factors for, 360 two-stage centrifugal, 129
sheet metal, construction of, 347 Expansion, sudden, 86, 90
straight, resistance to air flow in, 362 Explosion hazards, 58, 136
uniformly distributed supply and for industrial gases, classification
exhauster ports with, 301 of, 60, 61
Explosive atmospheres, control of, 59 Fans-cons.

electrical apparatus for, 61 pedestal, 38, 106, 122
Explosive limits, 59 propeller, 120
Extended fan shafts with overhung radial flow, 110
impellers, 222 special-purpose, 135
External rotor, 146 table, 37, 106, 122
Extraction systems, 29 variable pitch, 112
wet gas, 135
Fatigue, structural, 224
Fabric arresters, 69 Filters, 86
Fan characteristic, the, 106 Flameproof fan, 136
unsuitable, effect of, 173 Flammability, limits of, 59
with outlet swirl, 243 Flammable gases, properties of, 59
Fan designs at equal output power, 133 Flammable vapours, properties of. 59
Fan duty, damper control of, 180 Flash point, 59
Fan laws, the, 251-255 Flow arrangements, effectiveness
application of, 254 of 268
for sound, theory of, 213, 228 Flow division, Y-type, 97
influence of compression on. 253 Flow entering/leaving main, 96
Fan motor starting, 155 Flow lines into ducts, 71
Fan noise, rating system for, 200 Flow measurement, by nozzle and
Fan power characteristic. influence orifice, 244
of, 181 Flow rate, calculation of, 245
Fan static pressure, 108 Fluid couplings, 179
Fan tests, with standardised Fluid power system, 179
airways, 234 Forced convection, 260, 261, 262
Fan velocity pressure, 108 Forward-curved multi-vane type fans,
Fan, what is a, 105 123-125, 207
Fans, axial flow, 110, 112, 119, 207 Forward-curved fan with series
booster, 294 damper control, 183
ceiling, 37, 106 Fractional solidity impellers, 116
centrifugal, 123 Free vortex pressure, 141
circulating, 36, 106 Frequency control, 150
contra-rotating, 174 Friction calculation, example of
controllable pitch. 112 airway, 299
cross flow, 111, 134 Friction drop in long airways, 296
flameproof, 136 Friction factors in rough airways, 297
forward-curved, 123, 181 Fume collection, 70
multi-vane type, 123-125, 207 Fume cupboards, 56
series damper control with, 183
gastight, 136
guide vane, 115 Gas concentration, recommended
hot gas, 136 maximum, examples of, 55
in-line centrifugal, 128 Gases: argon, carbon dioxide 2;
main types of, 109-111 carbon monoxide, 282; helium,
2, 136; hydrogen, 59, 136;
mine, 119 krypton, neon, nitrogen, oxygen
mixed flow, 110, 129, 131 xenon, 2
multi-vane, 123
properties of, 332
orifice flow, 111 Gastight fan, 136
outlet flow, 111 Gas turbine. turbo-compressor
paddle-bladed, 128 rotor for, 131
Index 371
General-purpose axial flow fans, Impellers, 112, 122-126 passim, 222

112, 114 aerofoil-bladed, 126
General-purpose fans, 135 backward-curved, 125
comparative performances of, 132 critical speed for, 223
Geothermic temperature gradient, 292 paddle-bladed, 128
GH on Gs, dependence of, 270 reversible, 116
Globe temperature. 12 Incidence angle, 141
Globe thermometer, 12 Incompressible fluid, 76
Glove box, 56, 57 Increased pressure range, 127
Guards, 86 Induction motors, speed control of, 147
Guide vanes, 114 efficiency and power factor of, 152
performance of, 152
Hazard classification for electrical three phase typical performance
apparatus, 62 of, 365
Hazardous atmospheres, fans for, 63 typical full load speeds of, 353
Heat balance, 5 Industrial drying, 270-274
Heat, sources of, 20 Industrial gases for explosion hazard
from animals, 20; electric motors, 21 ; classification of, 60-61
electric power, 21 ; fuel, 21 ; Inertia, moments of. 154
lighting, 20; people, 20 Infiltration, 48
solar, see Solar Heat air change basis of, 49
Heat exchangers. 265-270 allowances, typical, 50
effectiveness of. 266, 259 rate, example of. 49
Heat resistance, 135-136 rate from metal framed windows, 48
Heating system, example of, 44 through closed windows. 49
Heating, with air circulation, 32, 44 Inlets, 85
Heat stress, 10, 21 losses at, 90
industrial workers, exposures for. 10 In-line centrifugal fans 128
Heat gain, ventilation rates based on, 20 mixed-flow, 130
Heating load, 40, 41 Insulation, class of 159
Heat transfer, 259-265 Interaction noise 212
High inertial loads 154
Hoods, 71 Jet action. calculation and design
for circular saw, 74 for, 308
for grinding wheel, 73 principles of, 306
Hot climates, relief in, 36 Jet fans, use in tunnel ventilation, 305,
psychrometric chart for, 51 307
Hot gas fans, 136 performance of, 310
Jet noise, 212
Humidifies-per cent relative, 328 Jets, air supply, bY 33-35
Humidity, 6 cooled, effects of, 35
air movement and, 7 heated, effects of, 35
relative, 6, 50 momentum flux of, 35
measurement of, 6 throw of, 35
Hydraulic diameter, 89 velocity profiles of, 33, 34
Ignition temperature, 59 Kinetic energy. 78, 79, 108, 111,
Impeller design, practical, 141 138-139, 141, 155, 181
Impeller power-elementary theory, 140, Kinematic viscosity grades for
249 liquids, 336
Latent heat of evaporation, 50, 274 Motors, AC 142, 151

Leading power factor current, 147 capacitor-start, capacitor run, 145
Leakage, avoidance of, 136 commutator, 151
calculations, 303 direct current, 151
coefficients, 303 induction, 147
from long ducts, 302 multi-speed, 172
influence of on volume flow polyphase induction, 142
and pressure, 304 rotor induction, external, 146
Lift and drag forces, 211 series wound, 151
Limits of flammability 59 shaded-pole, 146
Liquids, properties of, 333 split capacitor, permanent, 145
Losses, bends, in, 94 split-phase induction, 144
branches, at. 96 squirrel cage, 150, 157
diffusers, in, 92 synchronous, 147
dust friction, due to, 99 totally enclosed, 161
guards, screens and obstructions universal, 151
at, 91 wound rotor induction, 149
in-duct diffusers, of, 93 Moisture contents, typical
obstructions to air flow, at, 91 equilibrium, 271
of open-outlet diffusers of optimum expressions, comparison of, 272
proportions, 93 Moisture resistance, 135
sharp-edged orifices, at, 90 Multi-speed motors. 172
tube batteries, across, 91
Multi-speed windings, 150
Magnesium powders, 57 Noise, fan, rating system for, 200
explosions by, 57
jet, 212
Mancoolers, 38-39 propeller, 212
velocity contours of, 39 Noise generating installations,
Measurement, methods of, 228 examples of, 205
Metabolic rate, 2 Nozzle and orifice coefficients for
Methyl bromide, 59 air, 246, 247
Metric bolts and screws, preferred
sizes of, 348-349
Mine fan, 119, 294 Obstructions, 86
Mine ventilation, 292-295 Orifices, 86
Mineral dusts, recommended limits losses at, 90
for. 58 Outlet flow, 111
Mixed flow fan, 110, 129-131 Outlet registers, 35
in-line, 130 Outlet swirl, 114
with axial discharge, 129 Outlets, 86
radial discharge, 131 losses at, 90
volute casing. 130 Overload protection, 157
Mixed-flow roof extract unit, 131 Oxygen deficiency, 2
Moments of inertia, 154
Momentum flux, 306
Motor control gear, 157-159 Paddle-bladed impeller, 128
Motor cooling, 161 Paint-spraying booths, 73
Motor enclosure, 161 Parallel, fans in, 172
Motor windings, temperature duty control by, 173
limits of, 160 Part-winding motors, 157
Index 373
Pedestal fans, 38, 106, 122 Radiation, 4, 263, 264

velocity contours of, 38 relative influence of, 41
Phase control, 152 Radius of gyration, 154
Pitch adjustment, 112 Rail tunnels, ventilation of, 290
Pitch angle, 112, 116 Ratio systems, variable, 179
Pitot-static traverses, 239, 240 Rectifiers, 152
Pitot-static tube, 233 Refrigeration, 50
velocity measurement by. 230 methods, comparison with, 52
Plastics, properties of, 334 Relative humidity, 6
Pole amplitude modulation, 151 Re-lubrication interval, factors for. 168
Pole number, 143 for continuous-running ball
Pollution from car exhausts, 283-285 bearings, 169
Pollution rates at sea level, 284 Residual unbalance, 217
Power and energy consumption with Resonant vibration, 222
two-step control, 175 Reversal of direction of rotation, 116
Power characteristics, typical, 182 Reversible impeller, 116
Power factor correction, 153 Reynolds number, 85 245, 297
Pressure, absolute and atmospheric, 75 Road tunnels, pollutants in, 282-285
definitions of, 75 cross-sections in, examples of, 289
distribution of, 48 sectionalising length of, 286
duct cross-section, at, 78 ventilation in, 288
measurement of, 228 ventilation rate of, 286
over a tall rectangular building, 48 Rotor unbalance, two kinds of, 218
static, 76, 138
total, 77, 79-85 passim
tunnels and mines, requirements Saturated vapour pressure factor-A,248
in, 295 Saturation vapour pressure of water, 276
velocity, 76 Sensible heat, 50
Pressure drop calculation. Series, fans in, 174
101-104, passim
Pressure drop data, summary of, Series-wound motor, 1.51
362-363 Settling chambers, 66
Pressure gradient in circular ducts, 98 Sheet blading, 126
Pressure range. increased, 117 Sheet material, preferred thicknesses
Pressure relationships, 80 and weights of 346
Pressures at a duct cross-section, 78 Shunt field winding, 151
Pressures in a ducted fan system, 81 SI basic and derived units, 312
Propeller fans, 120 multiples and submultiples of, 313
bell mouth inlet ring with, 122 Sine-wave vibration, 215
diaphragm mounted, 121 Site test measurements, location of, 233
Propeller noise, 212 Sling psychrometer, 6, 7
Prospective current, 158 Slip systems, variable, 179
Psychrometer, sling, 6. 7 Slip, the, 143
Psychrometric chart, 8 Solar heat, 22, 23
for hot climates, 51 effective transmission factors for,
Pulsating flow, 212 through building surfaces, 26
gain, calculation of, 25, 41
heat from the sun, 22
Radial blading, 127 intensity, maximum, 22, 23, 24
Radial flow, 139 on roofs, 24
Radial flow fan, 110 on vertical walls, 24
Radiant heat, 10, 12, 264 radiation. intensity of, 23
Sound power, 200-210. Straight lengths, 88

change of, with duty point. 207 Streamline flow, 64
contours of equal, 208 Structural fatigue, 224
fan laws for, 204, 213 Structural materials, properties of, 335
level characteristics, typical
shapes of, 207 Sun, heat from, 22
level of and fan duty, 203 altitude of above horizon, 22
order of magnitude of, 202 altitude, noonday, at L' north or
ratio, variation of, 206 south latitude, 23
scale, the, 202 Supply systems, 31
Sound spectra, typical, 209, 210 Supply ventilation system, 103
Sparkproof construction, 136 Supply voltages and windings, 153
Special-purpose fans, 135 Synchronous speed and pole
Speed control, continuous, 178 number, 143
Speed control, DC motors of, 152 System characteristic and
induction motors, of, 147 operating point, 107
Speed regulation, by voltage control, 147 System pressure drop calculation,
wound rotor motor, of, 149 example of, 102
Speeds, critical, 222 System resistance elements,
Splitter vanes, 94 discussion of 85-89
Spray chambers, 68 Table fans, 37, 106, 122
velocity profiles of, 37
Squirrel-cage motor, 150, 157 Temperature limits of motor
Squirrel-cage rotor, 142 windings, 160
Stack effect, 47 Temperature variation below ground, 293
Stage control, example of, 176, 177 Thermal protection, 158-159
Stagnation point, 76, 79
Standardised airways, 227 Thermal transmittance, 41
examples of, 234, 235 typical, for buildings, 42
fan tests with, 234 Thyriaoa control, of single phase
capacitor motor, 148
free inlet, free outlet, 239, 242
ducted outlet, 239 Thyristor controllers, 149
pressure formulas for. 237 Thyristors, 150
tests with an inlet side chamber, 238 Tip clearance, 113
Star connection, 154 Torque converters, 179
Star-delta starters, 157 Torque curves, 144-146
Starter tripping schedule, 158 Total pressure, 77, 78-84 passim, 301
Starters, auto-transformer, 157 rise, 110
direct-on-line, 157 relative nature of, 80
other, 157 Total praasure drop, estimation of, 84
resistance, 157 influence of density on, 101
star delta, 157 elements of, 85
Starting time and loss coefficients, 156 Total pressure drop and fan total
Static pressure, 76, 114, 138, 301 pressure, 79, 110
difference, 79 Totally enclosed motors, 161
fan, 108 Toxic gases, 53
relative nature of, 80 vapours, 53
Static unbalance, 216 hazards, dilution ventilators for, 54
Step-by-step systems, 172 Traffic drag, 300
Index 375
Transistor controllers, 149 Vane contracts, 111

Transitional flow, 66 Ventilation, 14
Transverse ventilation, 289 air change ratings for, 14
Truncated diffusers. 92, 102 air conditioning for, 18
Tunnel drivage fans, 295 air movement rates for general, 15
Tunnels, road. pollutants in 282-285 heat load to be removed by, 24
rail, ventilation of, 290 lavatory, 20
Turbo-compressor rotor for a gas longitudinal, 287
turbine, 131 mine, 292
Turbulent flow, 64 minimum recommended, 16
Turning vanes, 94 natural, 14, 47, 287
rail tunnels, of, 290
rate and temperature rise, 25, 47
Unbalance, couple, 217-218 rates based on heat gain, 20
rotor, two kinds of, 218 per person, 16
static, 218 re-circulation and, 17, 32
Universal motors, 151 relative nature of, 80
Upcast shaft, 294 road tunnels, of, 290
Upstream guide vanes, 114 semi-transverse, 290
transverse, 290
smoking permitted, when, 16, 18
Vacuum cleaner, 129 tunnel, use of jet fans in, 305
Vane control, principles of, 190 underground railways, of, 291
for axial fans, 191 vehicle tunnel, of, 178
of backward-curved centrifugal Ventilating systems, resistance to air
fan, 191 flow in straight ducts for, 364
for centrifugal fans, 190
Variable pitch fans, 185-189, 192 Venturi nozzle, 239, 240, 244
application of, 187-188 Vibration, 214
mechanical features of, 188 induced, 213
performance of, 185, 186 measuring units for, 214
Vapours, toxic, 53 resonant, 222
flammable, properties of, 59 severity of, 216
quality judgement of, 216
Variable ratio gearboxes, 179 sine-wave, 215
Variable ratio systems, 179 Viscosity grades for liquids,
Variable pitch fans, 112 kinematic, 336
Variable slip systems, 179 Viscous filters, 68
Variable speed transmission, Voltage control, 147
examples of. 179 speed regulation by, 147
Variants on the standard range, 114 Volume flow, determination of, 232
Vasomotor regulation 4 by anemometer traverse, 232
V-belt drives, 179 Volute casing, 111
standard power ratings for, 356-359
Velocity, 214, 229
measurement of 229-231 Walls, heat flow through, 40
Velocity diffusion, 241 Water, properties of, 329
Velocity pressure, 76, 77, 114 Weather resistance, 135
table of, 326-327 Wet-bulb temperature, 6, 11, 328
fan, 108 Wet collectors, 67
Wet-gas fans, 135 Wind speed-cont.

Wet surface, equilibrium temperature stack effect of, 47
of, 277 statistics, 45
Wind effects, 45 in a city centre, 45
at Heathrow Airport, London, 46 Wind velocity profiles for winter
infiltration, and, 48 design, 46
Wind pressure, 88, 295 Winding thermostat, 158
Wind speed, 45, 46
maximum likely, 46
meteorological, 45, 49 y-type flow division, 97

Fan Engineering

Uploaded by

Copyright:

Available Formats

Fan Engineering

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fan Engineering

Uploaded by

Copyright:

Available Formats

Woods Practical Guide

An important feature of present-day society is the interest in

This, the third, edition of Woods Practical Guide to Fan

just as essential to the success of each fan installation. It is not,

This, third, edition of Woods Practical Guide to fan Engineering

millimetre for component dimensions. The micrometre (µm) is

Volume Flow. All calculations are performed in cubic metres per

Mass, Mass Flow and Density. The kilogramme, the kilogramme

Force. It is most important that the newton should become

though heretical, to think of unit stress as one "newton per

Energy, Power and Heat. The joule (= 1 newton-metre) and the

Viscosity. The basic SI unit of dynamic viscosity is the

Man's sense of comfort and his capacity for work deteriorate

1.1 The Air

1.1.4 Carbon dioxide also has no significant effect on man. Because we

1.1.5 Impurities such as methane or ozone occurring in natural fresh air at

1.2 Body Heat Balance

1.3 Convection and Radiation

1.5 Measurement of Humidity

1.7 Comfort Conditions

1.7.1 Air movement and humidity

Humidity is unimportant in the comfort zone provided it is in the range 30% to

1.7.2 Optimum temperatures

1.8 Heat Stress

1.9 Radiant Heat

1.9.2 Globe temperature

Natural ventilation, with open windows in summer, may

2.1 Ventilation Requirements

table of "air change rates". The volume in cubic metres (m3) of

Table 2.1 gives a guide to the number of air changes generally

2.1.2 Ventilation rates per person

These rates should be increased by a third if smoking is permitted

There would be no value in recirculation if the outside air were

2.1.4 Ventilation for air conditioning

While the building is not to be air conditioned, we will note

We may thus adopt a recirculating system,with a wintertime

Adjoining lavatory accommodation is planned with six urinals and nine

2.2 Sources of Heat

2.2.2 Heat from people

2.2.4 Heat from lighting

2.2.5 Heat from electric power

2.2.6 Heat from electric motors

2.2.7 Heat from fuel

However, a large part of this heat will not contribute to the

(a) Incomplete combustion-unburnt or partially burnt fuel passing

(c) Heat to stock-occurring when the material worked on (solid,

In important cases estimates for each of these items should be

2.3 Solar Heat

2.3.1 Heat from the sun

This simple statement gives complete information for calculating

2.3.2 Maximum solar heat intensity

Generally speaking the maximum intensities on various parts

2.3.3 Calculation of solar heat gain

where Î is the effective maximum solar intensity obtained from Tables

2.4 Ventilation Rate and Temperature Rise

2.4.1 Heat capacity of air

*See also Table 14.9.

of 1 kW or 1 kJ/s will raise the temperature of air flowing at the

Air will enter the building at the outside air temperature. As it